Enzymatic Noncovalent Synthesis

Abstract

Enzymatic reactions and noncovalent (i.e., supramolecular) interactions are two fundamental non-genetic attributes of life. Enzymatic noncovalent synthesis (ENS) refers to a process where enzymatic reactions control intermolecular noncovalent interactions for spatial organization of higher-order molecular assemblies that exhibit emergent properties and functions. Like enzymatic covalent synthesis (ECS), in which an enzyme catalyzes the formation of covalent bonds to generate individual molecules, ENS is a unifying theme for understanding the functions, morphologies, and locations of molecular ensembles in cellular environments. This review intends to provide a summary of the works of ENS within the last decade and emphasize ENS for functions. After comparing ECS and ENS, we describe a few representative examples where nature uses ENS, as a rule of life, to create the ensembles of biomacromolecules for emergent properties/functions in a myriad of cellular processes. Then, we focus on ENS of man-made (synthetic) molecules in cell-free conditions, classified by the types of enzymes. After that, we introduce the exploration of ENS of man-made molecules in the context of cells by discussing intercellular, peri/intracellular, and subcellular ENS for cell morphogenesis, molecular imaging, cancer therapy, and other applications. Finally, we provide a perspective on the promises of ENS for developing molecular assemblies/processes for functions. This review aims to be an updated introduction for researchers who are interested in exploring noncovalent synthesis for developing molecular science and technologies to address societal needs.

Graphical Abstract

1. Introduction

1.1. Enzymes and Noncovalent Interactions

Enzymes are biomacromolecules that catalyze chemical reactions to make molecules. Enzymes, being categorized into six types (oxidoreductases,¹ transferases,² hydrolases,³ lyases,⁴ isomerases,⁵ and ligases⁶), are highly efficient and specific catalysts for a wide range of reactions that maintain the proliferation, differentiation, migration, and metabolism of cells.⁷ Thus, while enzymes, being proteins, are a genetic feature of cells, enzymatic reaction is a non-genetic attribute of life. Previously, the study of enzymatic reactions has focused on the formation of covalent bonds to generate individual molecules, which consist of the atoms covalently connected. While strong interactions (i.e., covalent bonds) connect the atoms to make the molecules in cells, the intermolecular interactions (e.g., hydrogen bonding, ionic, and hydrophobic interactions) are ubiquitous, weak, and noncovalent (i.e., supramolecular interactions⁸). Therefore, another fundamental non-genetic attribute of life is noncovalent interactions. These noncovalent interactions between biomolecules are unlikely to be completely random; they must be under certain controls. The dynamic characteristics of cells suggest that such controls have to involve reactions. Unavoidably, most of the reactions must be enzymatic reactions. Therefore, enzymatic reactions also control noncovalent interactions in cells. Thus, we refer the formation of covalent bonds catalyzed by enzymatic reactions as enzymatic covalent synthesis (ECS) and the formation of noncovalent interactions controlled by enzymatic reactions as enzymatic noncovalent synthesis (ENS). ECS generates individual molecules, and ENS produces ensembles of molecules.

1.2. Enzymatic Covalent Synthesis

The study of enzymes started with the identification and categorization of enzymes more than 70 years ago.^9–10 The emphasis, then, has focused on the mechanisms of enzyme actions.^11–15 With the rapid advance of structural biology, researches over the last forty years have elucidated the structures of many enzymes, especially their active sites.^16–18 Recently, the advance of activity-based probes has provided a powerful tool to understand the functions of enzymes for drug discovery.¹⁹ Since the discovery of enzymes, enzymatic reactions have carried out industrial biotransformation for making fine chemicals.²⁰ Recently, the design of enzymes by “directed evolution” has achieved exciting success—one can evolve enzymes to carry out new catalytic reactions with high selectivity and efficiency.^21–24 Although the intermolecular interactions between enzymes and their substrates or products are part of the inquiry in understanding the mechanisms of enzyme actions,^11,25 the research activities on enzymes, largely, have focused on enzymatic covalent synthesis (ECS)—that is, enzyme catalyzes the formation of covalent bonds to generate individual molecules.

1.3. Enzymatic Noncovalent Synthesis

Because nearly every major process in a cell is carried out by assemblies of 10 or more protein molecules, Alberts has proposed that a cell is a collection of protein machines.²⁶ Noncovalent interactions are the forces that hold the protein machines together. Recent advances in cell biology have only reinforced such notion. Increasing numbers of studies have shown that biomacromolecules are able to form higher order assemblies, as functional centers (i.e., signalosomes^27–28) or as a form of compartmentalization (i.e., membraneless condensates for biochemical reactions^29–34). These supramolecular assemblies²⁷ are dynamic, heterogeneous, and play a critical role in cell functions,^35–36 but it is still largely unknown how cells control these assemblies for functions. Enzymatic reactions must be one of the controlling mechanisms of life, and probably the most important “rule of life”. Thus, it becomes increasingly necessary to consider and to investigate how enzymatic reactions instruct intermolecular noncovalent interactions to form molecular ensembles for functions. Since the functions of the molecular assemblies usually differ from individual molecules, it is necessary to define enzymatic noncovalent synthesis (ENS)—a process that integrates enzymatic reactions and supramolecular (i.e., noncovalent) interactions for spatial organization of higher-order molecular assemblies to achieve emergent properties or functions. Figure 1 illustrates the relationship among ECS, noncovalent interactions, and ENS. ECS makes one molecule from another by enzyme catalysis; noncovalent interactions results in molecular assemblies from many molecules by self-assembly, but no new molecules are made; ENS converts many molecules (either individually or in ensembles) to assemblies of new molecules by ECS and self-assembly. That is, the target of ECS is one molecule, but the focus of ENS is assemblies of molecules and their emergent properties. By this definition, ENS includes one or more ECS processes, but it has to incorporate intermolecular noncovalent interactions to result in molecular ensembles that exhibit emergent properties. In other words, ENS enables a coherent outcome of the morphological and functional organization of molecules.

Figure 1.

The relationship among self-assembly, ECS, and ENS and the corresponding targets and products.

1.4. Scope and Arrangement

As a unifying theme for understanding functions and locations of molecular ensembles in cellular environments, ENS also represents an emerging research area at the interface of physical science (e.g., chemistry,³⁷ physics,³⁸ and nanotechnology³⁹) and biological sciences (e.g., structural biology,²⁷ biochemistry,²⁹ and cell biology³⁵). The aim of this review is to examine how enzymatic reactions provide spatiotemporal control of noncovalent (or supramolecular) interactions of both natural and man-made molecules to result in emergent properties or functions of molecular assemblies. To provide a relatively comprehensive summary of the works of ENS within the last decade and to put emphasis particularly on the applications of ENS, we start with discussing a few representative examples to show how nature uses ENS to create biomacromolecules ensembles for a myriad of cellular functions and processes. After introducing these facts to support ENS as a rule of life, we focus on ENS of man-made molecules in cell-free conditions, classified by the type of enzymes. Then, we describe ENS of man-made molecules in the context of cells to illustrate the applications of ENS. We discuss intercellular, peri/intracellular, and subcellular ENS for cell morphogenesis, molecular imaging, cancer therapy, etc. Finally, we provide a perspective on the promises of ENS. We hope that this review will serve as an updated introduction and reference for researchers who are interested in exploring noncovalent synthesis for developing molecular technologies for addressing societal needs at various frontiers.

2. Enzymatic Noncovalent Synthesis in Nature

ENS is an integrated behavior of cells because enzymatic reactions and noncovalent interactions are two fundamental non-genetic attributes of cells that act reciprocally. Several well-known biological facts have revealed that enzymatic reactions control the formation of supramolecular assemblies for functions. For example, enzymatic hydrolysis of GTP drives the dynamic of microtubules,⁴⁰ kinases and phosphatases activate inflammasomes,^41–42 proteinases regulate the formation of collagen fibers,⁴³ and aberrant proteolysis or phosphorylation contributes to forming disease-associated amyloids.^44–45 Because enzymes control dynamic posttranslational modifications (PTMs) of polypeptide chains,⁴⁶ and noncovalent interactions drive the formation of the molecular ensembles, cells use ENS as a fundamental biochemical mechanism to control functional assemblies as evidenced by the following representative examples.

2.1. Intracellular ENS

A unique feature of the interiors of all cells is crowding—supramolecular functional structures (Figure 2) like biomacromolecules occupy a significant fraction (typically 20–30%) of the total volume of cells. It has been recognized that biomacromolecular crowding plays a role in all biological processes at the biochemical level,^47–49 and the relevant studies have centered largely on how the crowding (or excluded-volume) promotes colloidal phase separation⁵⁰ or influences the rate of ECS.^51–52 Obviously, crowding is dynamic and associates with noncovalent interactions of the biomacromolecules in the microenvironment. Since local enzymatic reactions change the dynamics of the microenvironment, enzymatic reactions also control biomacromolecular crowding. Such a reciprocal feedback or interdependence between enzymatic reactions and crowding leads to a challenge to parameterize the degree of crowding of the cell interior. Thus, it is necessary to examine enzymatic reactions and noncovalent interactions in a holistic manner. That is, evaluating how enzymatic reactions control noncovalent interactions to lead to emergent properties of molecular ensembles, which is the essence of ENS. The following sections discuss the formation of intracellular supramolecular functional structures from the perspective of ENS. Since it is impossible to numerate all of them, we chose the ensuing representative examples and categorized them following the convention used by the creation of the cell atlas.⁵³

Figure 2.

Major intracellular functional structures. Adapted from Ref. 53. Copyright © 2017 by AAAS.

Actin Filaments.

The formation of actin filaments is an ultimate example of ENS. As one of the most abundant proteins in eukaryotic cells and existing in both monomeric G-actin (G for globular) and polymeric F-actin (F for filament), actin⁵⁴ itself is an enzyme. The substrate of actin is ATP, so actin is an ATPase.⁵⁵ ATP hydrolysis on actin is the key reaction for maintaining the actin filaments.⁵⁶ As shown in Figure 3, in cells, most of the G-actins bind with ATP, and most of the F-actin subunits contain ADP. ATP-actin attaches to the barbed end of F-actin, and ATP gradually hydrolyzes to become ADP. Overtime, ADP-actin dissociates from the pointed end of the filaments. The ADP-actin monomer then exchanges with ATP to form the ATP-actin, which can attach to the barbed end again. An assembly-disassembly process such as this occurs continuously without the presence of actin binding proteins (ABPs). In the presence of ABPs, actin filaments act as a key component of cytoskeletons (Figure 2) for various cellular functions, such as motility,⁵⁷ endocytosis,⁵⁸ and mitochondria dynamics.^59–60 The functions of F-actin are the emergent properties of the assemblies of actins, which rely on the enzymatic hydrolysis of ATP, as well as other protein-protein interactions (PPI). Fitting well with the definition of ENS, the formation of F-actin not only underscores the unique advantage of ENS, but also represents a case of self-organization or assembly away from equilibrium. In addition, the formation of actin filaments, of course, changes crowding of the microenvironment for controlling other enzymatic reactions for various functions of proteins in cells. This kind of feedback regulates observable cellular behaviors, such as focal adhesion.^61–62

Figure 3.

Scheme of formation of an actin filament by ENS. Bold arrows show the net fluxes of pointed-end disassembly and barbed-end growth. The barbed ends, having two configurations, with bound ADP-Pi or ADP with different dissociation rates lead to that ATP hydrolysis affects length fluctuations at the barbed ends.⁵⁶ k^B₊: the association rate constant of ATP-G-actin to barbed ends; k^P₋: the dissociation rate constant of ADP-G-actin from pointed ends. The structures of ATP-actin (PDP ID: 4PKG⁶³) and ADP-actin filament (PDB ID: 6DJO⁶⁴) are adapted from PDB database.

Aggresomes.

As the aggregation of misfolded proteins in eukaryotic cells, aggresomes⁶⁵ are a consequence of multiple enzymatic reactions and multiple proteins. In essence, the presence of misfolded proteins results in ubiquitination, which involves ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2), ubiquitin ligase (E3). When the proteasome is unable to degrade the ubiquitinated proteins, a histone deacetylase (HADC6) further interacts with the ubiquitinated proteins to form the aggresomes.⁶⁶ Since the discovery of aggresomes, by Kopito et al., in the study of undegraded cystic fibrosis transmembrane conductance regulator (CFTR),⁶⁵ considerable efforts have focused on the misfolded proteins related to neurodegenerative diseases. Several other enzymes, such as parkin and ataxin-3,^67–69 are also implicated in aggresome formation. Generally, aggresome formation is a normal cellular response to a large amount of misfolded proteins. Inhibition or loss function of proteasome would result in the activation of aggresome pathways because aggresomes, acting as a holding station of potentially cytotoxic protein aggregates, facilitate their eventual clearance from the cell by lysosomes or autophagy. One interesting feature of aggresomes is that they form around the microtubule-organizing center (MTOC) in eukaryotic cells, adjacent to or enveloping the centrosomes of the cell.⁷⁰ Though it remains to be determined, it is possible that more enzymes and proteins are involved in the formation of aggresomes, which is an ENS process.

Centrosomes.

Centrosomes are the organelles that serve as the MTOCs in animal cells during mitosis. The key components of centrosomes are centroles, pericentriolar material (PCM), and microtubules (Figure 5A).⁷¹ A centrosome^71–72 may consist of hundreds of proteins as the PCM. Since many of the proteins only transiently associate with centrosomes for signaling functions, it is difficult to define the components of centrosomes precisely. The transient nature of PCM implies that enzymatic reactions control the centrosome scaffolds (or assemblies) at the different phases of mitosis. A study of the expansion of the mitotic PCM around the mother centriole in flies, indeed, supports that ENS is a feature of the dynamics of centrosomes. Such an ENS process involves three key molecules: a kinase, Polo, and two proteins, Spd-2, and Cnn. As shown in Figure 5B,⁷³ during interphase, Polo is inactive, Spd-2 and Cnn are not phosphorylated, and there is no fibrous-like scaffold. As cells prepare for mitosis, activated Polo phosphorylates Spd-2. Then, the phosphorylated Spd-2 assembles into a fibrous-like scaffold that spreads away from the centriole and recruits Polo and Cnn. After Polo phosphorylates Cnn, then Cnn assembles into a scaffold, which helps to maintain the Spd-2 scaffold and allow it to expand further and thus recruits more Polo and Cnn. This process creates a positive feedback loop. In this relatively complicated ENS process, Polo, as a master regulator and an enzyme, controls the assembly of phosphorylated Spd-2 and phosphorylated Cnn to form the centrosome scaffolds. This perspective from ENS may help to avoid the difficulty in distinguishing solid-like or liquid phase separations⁷³ in the understanding of centrosomes dynamics.

Figure 5.

(A) Centrosome structure. A pair of centrioles is shown, each with ninefold symmetry owing to the nine triplet microtubules. Each centriole has pericentriolar material (PCM) that nucleates microtubules around the ends closest to one another. Only the maternal centriole has two sets of extra appendages, distal and subdistal; the latter seems to anchor microtubules. A series of interconnecting fibers, different from the PCM, links the closest ends of the two centrioles. Adapted from Ref. 74. (B) The scheme of the putative pathway of scaffold assembly in flies: black arrows indicate recruitment, red arrows indicate phosphorylation, and the orange broken arrow indicates that Cnn helps to maintain the Spd-2 scaffold, but does not recruit Spd-2 into the scaffold. (C) The illustration on how ENS may form a positive feedback loop to drive the expansion of the mitotic PCM around the mother centriole (light green). (i) Interphase, Polo, Spd-2, and Cnn form a torus around the mother centriole: Polo inactive, Spd-2 and Cnn unphosphorylated, no scaffold. (ii) Mitosis, activated Polo phosphorylates Spd-2, which forms the scaffold spreading away from the centriole (indicated by red broken arrows). The phosphorylated Spd-2 scaffold recruits Polo and Cnn, Polo phosphorylates Cnn, and Cnn forms a scaffold. (iii) The Cnn scaffold maintains the Spd-2 scaffold, allowing it to expand and to recruit more Polo and Cnn. (B,C) Adapted from Ref. 73. Copyright © 2019 by Elsevier Inc.

Cleavage Furrow.

At the end of cell division, a cleavage furrow forms and eventually separates the two daughter cells. The key feature of a cleavage furrow is that the assembly and contraction of actomyosin filaments drive a contractile-ring (Figure 6A).⁷⁵ Actin and myosin are ATPases, they catalyze ATP hydrolysis for the formation of actin filament and the movement of myosin. Other enzymatic reactions also control these processes (Figure 6B): GTP-bound RhoA activates two kinases (ROK and CITK) to phosphorylate the myosin’s regulatory light-chain (MRLC) to promote the sliding of myosin heads along actin filaments. GTP-bound RhoA also interacts with formin to promote profilin-mediated actin polymerization. In animal cells, RhoA is a small GTPase, which acts as the key regulator for both the assembly and contractions of the actiomyosine filaments. Like other regulators in cells, the level of RhoA-GTP is controlled by other modulators, such as RhoA-GAP and Rho-GEF.⁷⁶ Thus, the dynamics of contractile-rings is a fine example of a rather sophisticated ENS process evolved by nature.

Figure 6.

(A) The illustration of cleavage furrow and actomyosin contractile-ring. (B) The ENS processes for the dynamics of contractile ring (red represent enzymes and enzymatic reactions): GTP-bound RhoA promotes (i) profilin-mediated actin polymerization through the formin; and (ii) phosphorylation of the myosin regulatory light chain (MRLC) by two kinases, Rho kinase (ROK) and citron kinase (CITK). Together, these events promote the sliding of myosin heads along actin filaments, and therefore the formation and ingression of the cleavage furrow. Adapted from Ref. 75. Copyright © 2009 by The Company of Biologists.

Cytoplasmic Bodies.

Cytoplasmic bodies are the collective term for aggregation of biomacromolecules inside the cytoplasm. Cytoplasmic bodies are poorly defined, except for processing bodies (P-bodies) and stress granules that are known to contain RNA and proteins. Among the cytoplasmic bodies, P-bodies, since its discovery,⁷⁷ received more detailed characterization than other cytoplasmic bodies did. P-bodies, consisting of many enzymes and RNA, function as a center for mRNA turnover, such as degradation of unwanted mRNAs, storing mRNA for translation, and aiding in translational repression,⁷⁸ though some of these functions remain controversial. With the report of a new method to purify P-bodies,⁷⁹ an increased number of proteins in P-bodies have been identified.⁸⁰ Although different P-bodies may have different compositions, many essential protein components in P-bodies carry PTM sites (e.g., serine residues for phosphorylation in the enhancer of mRNA-decapping protein 4 (EDC4),⁸¹ cytoplasmic polyadenylation element-binding protein 1 (CPEB1),⁸² and eukaryotic translation initiation factor 4E transporter (EIF4ENIF1)⁸³), indicating enzymatic control of the assembly and disassembly of P-bodies. Recent reports also suggest that ATPases play a role to maintain P-bodies.^84–85 These observations indicate that enzymatic reactions modulate protein-protein interactions to form P-bodies, which turns out to be true. For example, after the report that JNK kinase phosphorylates DCP1a to control formation of P bodies,⁸⁶ a subsequent study shows a new mode for dynamic P-bodies remodeling by combined enzymatic phosphorylation/ubiquitylation events (Figure 7).⁸⁷

Figure 7.

An example of forming P-bodies by ENS. Trimerization of DCP1a triggers minimal P-body formation. TRAF6 ubiquitylates DCP1a at specific sites for recruiting more DCP1a to be phosphorylated by JNK (red indicates enzymes or enzymatic reaction). This process supports remodeling or “growing” of the P-body structure. Adapted from Ref. 87. Copyright © 2016 by Elsevier Inc.

Cytosol.

Cytosol contains many biomacromolecules. Although it is believed that the biomacromolecular crowding promotes phase separation and oligomerization of colloidal-like biomacromolecules, the reversible and dynamic nature of the assemblies or liquid-liquid phase separation suggests the involvement of ENS. While it is obvious that ENS controls the formation of actin filaments, microtubules, and intermediate filaments, many other biomacromolecular ensembles resulted from ENS either perform normal cellular functions or cause detrimental diseases. Since it is impossible to enumerate all the assemblies or filaments of proteins in cytosol, we arbitrarily chose inflammasomes, Tau filaments, and apoptosomes as the representative examples. Inflammasomes are supramolecular organization centers (SMOC)⁸⁸ for host defense inside cells. The structural determination by Wu et al. has established that three proteins, NLRP3, ASC, and caspase 1, self-assemble to form nucleated filaments as NLRP3 inflammasomes.⁸⁹ As shown in Figure 8A, the adaptor protein ASC bridges the sensor proteins (NLRP3) and caspase-1 to form ternary inflammasome complexes. This study has elucidated the noncovalent interactions between NLPR3 and ASC through pyrin domain (PYD) and between ASC and caspase-1 through caspase activation and recruitment domain (CARD) for the formation of the filaments. Recent advances in the study of NLPR3 inflammasome activation have confirmed that enzymatic reactions posttranslationally modify the proteins for initiating the assembly (Figure 8B).⁴² These processes, again, involve multiple enzymes and multiple proteins for controlling noncovalent interactions among biomacromolecules, which support ENS as a unifying theme in creating multiple protein complexes or biomacromolecular ensembles to exhibit emergent properties.

Figure 8.

(A) The model of the inflammasome assembly constituted by NLPR3, ASC, and Caspase-1. Multiple ASC-CARD molecules cluster to promote caspase-1 filament formation through CARD/CARD interactions. Adapted from Ref. 89. Copyright © 2014 by Elsevier Inc. (B) Schematic structure and PTM sites on different domains of NLRP3 (LRR, NBD, and PYD). BRCC3 and PTPN22 target the LRR domain of NLRP3 to promote NLRP3 inflammasome activation by deubiquitylation and dephosphorylation, respectively. Phosphorylation of NLRP3 by JNK1 and PKD, respectively, are key events for NLRP3 inflammasome activation. Dephosphorylation of NLRP3 PYD domain by PP2A promotes NLRP3–ASC as well as NLRP3 PYD–PYD interactions (red color representing enzymes or enzymatic reactions). Adapted from Ref. 42.

A defining feature of many human neurodegenerative diseases is the filamentous assemblies formed by aberrant Tau proteins.⁹⁰ Recently, cryo-electron microscopy (cryo-EM) has revealed the structures of tau filament from the brain of an individual with Alzheimer’s disease. Besides the determination of the filament cores that comprise residues 306–378 of tau protein and adopt a combined cross-β/β-helix structure, this work also shows that the paired helical filaments (PHF) and straight filaments (SF) are composed of hyperphosphorylated tau.⁹¹ Although the presence of the hyperphosphorylated tau indicates the involvement of kinases, many kinases can phosphorylate tau proteins.⁹² In addition, there are many phosphatases inside cells. Increased numbers of studies indicates that pathogenic tau filaments are resulted from the imbalanced actions of kinases and phosphatases.^93–94 This example highlights the complexity of ENS processes inside cells, which also underscores the importance of ENS in the understanding of pathogenesis of neurodegenerative diseases.

Cell death signaling is an important cytosolic process that utilizes ENS. In other words, ENS, as a multistep process, is an inherent feature of apoptosis. In the intrinsic pathway of apoptosis, enzymatic reaction induces the conformation change of Apaf-1 to allow it to interact with cytochrome c,⁹⁵ which results in subsequent self-assembly that forms the apoptosome.⁹⁶ The formation of the apoptosome further activates cascade events of cell death. During the extrinsically induced cell death,^97–98 the cell death ligands (e.g., TRAIL, TNF, and CD95L), being resulted from enzymatic reactions,^99–101 interact with the cell death receptors (e.g., TRAIL-R1/R2, TNFR1, and CD95) to initiate the self-assembly of the death receptors to form oligomers. Such oligomerization further triggers the down-stream signaling of apoptosis. These ENS features of apoptosis enable the quantitative aspects of signaling transduction, such as location, duration, thresholds, and amplitude. Such multistep ENS processes, avoiding a simple “on or off” (or “live or dead”) switch, achieve targeting with spatiotemporal precision, thus minimizing side effects. In fact, this insight has provided a blue print for developing ENS for selectively targeting cancer cells.¹⁰²

Endoplasmic Reticulum.

The endoplasmic reticulum (ER) is an organelle in eukaryotic cells. The key feature of ER is an interconnected network of flattened, membrane-enclosed sacs or tube-like structures. There are two types of ER, rough (RER) and smooth (SER). The outer face of RER is the site of protein synthesis. The SER is the place for lipid synthesis. The dynamic feature of ER implicates the need of enzymatic reactions to control the formation of the network of ER, which are the assemblies of proteins and phospholipids. A recent study by Rapoport et al. has shown that the addition of GTP in the mixtures of a GTPase (Sey1P), a curvature stabilizing protein (Yop1p), and liposomes results in a tubular network.¹⁰³ Moreover, the same lab also shows that atlastin (ATL),¹⁰⁴ the vertebrate orthologue of Sey1p, forms a GTP-hydrolysis-dependent network on its own (Figure 9). These results show that ENS of a small set of proteins controls the shape of organelles via interacting with lipids.

Figure 9.

(A) S. cerevisiae Sey1p and Yop1p were co-reconstituted into rhodamine-PE-labelled liposomes at protein:lipid ratios of 1:500 and 1:35, respectively, incubated without (left) or with (right) 2 mM GTP, and visualized by fluorescence microscopy. Scale bar 20 mm. Adapted from Ref. 103. Copyright © 2017 by Springer Nature. (B) The mechanism of membrane tethering and fusion is based on two crystal structures (state 1 and state 5) and biochemical experiments. The two ATL molecules dimerize and tether the membranes. Then, GTP hydrolysis causes conformational changes that pull the membranes together. Finally, GDP is released to reset the fusion machinery. Adapted from Ref. 104. Copyright © 2011 by Elsevier Inc.

Endosomes.

Endosomes are the organelles that sort and transport materials to lysosomes for degradation. As a membrane-bound compartment, endosomes form during endocytosis and are composed of lipids, proteins, and cargos. The formation of endosomes is a complicated and transient process, which implicates the involvement of ENS. A recent study¹⁰⁵ of the formation of clathrin-coated pits (CCPs) during clathrin-mediated endocytosis (CME) supports this notion. As shown in Figure 10A–C, during mammalian CME, a reversible phosphorylation on the μ2 subunit of the main endocytic clathrin adaptor, AP2, is crucial during clathrin-coated pit (CCP) initiation and increases throughout CCP lifetime. Such phosphorylation by AAK1 allows binding endocytic NECAP proteins without significantly altering AP2’s cargo affinity. The binding of NECAP recruits SNX9 for the late stages of CCP formation. In another study,¹⁰⁶ AAK1 phosphorylates the μ2 subunit of AP2 to promote CME of LRP6 in WNT signaling (Figure 10D).

Figure 10.

(A, B, C) A schematic model of phosphorylation of AP2 for CCP maturation. Adapted from Ref. 105. Copyright © 2019 by Elsevier Inc. (A) The evolving CCP structure and the phases it passes through. (B) The degree of phosphorylation of μ2T156 of AP2. (C) The individual components and their recruitment. (D) Prolonged exposure to WNT ligand drives AAK1-dependent AP2M1 phosphorylation, CME, and ultimately removal of the WNT receptors from the plasma membrane. Adapted from Ref. 106.

Focal Adhesion Sites.

Focal adhesion (FA) is the signaling center for regulating the mechanical interactions between the extracellular matrix (ECM) and intracellular cytoskeletons.^62,107–108 As dynamic assemblies of multiple proteins, FA undergoes enzyme-controlled assembly and disassembly. For example, during the maturation of FA, integrin stimulation results in the accumulation of phosphorylation proteins (Figure 11A), which is controlled by enzymatic phosphorylation.¹⁰⁹ Because one of the most common interaction domains in FA proteins is SH2 (a tyrosine phosphate binding domain), kinases phosphorylate tyrosines, switch on SH2-mediated interactions, and promote FA assembly. In the reverse process, phosphatases remove the phosphates from tyrosine residues, turn off the SH2-mediated interaction, and lead to FA disassembly. Among these enzymes, a key member for the assembly and disassembly of FA is focal adhesion kinase (FAK), which undergoes integrin-induced autophosphorylation of FAK or phosphorylation by Src family kinases. Carragher¹¹⁰ reported that Src-dependent FAK phosphorylation controls the assembly of a calpain 2/FAK/Src/p42ERK complex, calpain activation, and proteolysis of FAK (Figure 11B), which are critical for adhesion dynamics and cell survival. The studies of FA also highlight the critical role of enzymatic reactions in forming the noncovalent and functional multiple protein assemblies.

Figure 11.

(A) FA signaling is mediated by phosphorylation: the illustration of localization of pTyr in an attached human fibroblast highlights FAs around the cell periphery.¹⁰⁹ (B) Src-induced phosphorylation of FAK modulates adhesion dynamics, cell migration, survival, and anchorage-independent growth. Phosphorylated FAK results in actin stress fiber assembly and focal adhesion via activation of calpain, likely by linking calpain to the ERK/MAPK pathway. The Src-phosphorylated FAK also promotes cell survival and anchorage-independent growth of transformed cells. Adapted from Ref. 110. Copyright © 2004 by American Society for Microbiology.

Golgi Apparatus.

The Golgi apparatus (or Golgi complex) is a stack of flattened membrane-enclosed disks known as cisternae (Figure 12A). Golgi, being dynamically regulated during cell cycles in mammalian cells, acts as the site for posttranslational modifications of proteins (e.g., glycosylation), protein sorting, and protein trafficking (e.g., secretion).¹¹¹ The stack has two sides. The cis face is near the ER, and the other side is trans face. Many enzymes locate and distribute in the Golgi complex, and these enzymes are responsible for the modification of many cellular and secretory proteins, such glycosyltransferases, trimming enzymes, sugar transporters, or enzymes involved in other posttranslational modifications. The dynamic nature of Golgi implicates that proteins mediate the stacking of Golgi cisternae. Barr et al. has identified the proteins involved in the stacking of Golgi cisternae.¹¹² One of the proteins is Golgi reassembly stacking protein with the size of 65 kDa (GRASP65). Enzymatic controlled assembly of GRASP65 is responsible for the stacking and unstacking of the Golgi cisternae (Figure 12B). During interphase, PP2A catalytically dephosphorylates phospho-GRASP65 to generate GRASP65, which oligomerizes to connect the adjacent cisternae to form Golgi stacks. During mitosis, CDK1 and PLK1 phosphorylate multiple sites on GRASP65, resulting in the disassembly of GRASP65 oligomers to homodimers and the subsequent Golgi stack disassembly. This process (Figure 12B) is also a fine example of ENS evolved by nature for controlling the dynamics of subcellular organization.

Figure 12.

(A) The illustration of Golgi apparatus. (B) GRASP oligomerization and Golgi stack formation. During interphase, GRASP65 homodimers self-assemble to form oligomers through the GRASP domain. During mitosis, phosphorylation of GRASP65 results in the disassembly of GRASP65 oligomers and Golgi stack. Adapted from Ref. 113.

Intermediate Filaments.

Intermediate filaments (IFs)¹¹⁴ are one of the three major types of cytoskeletons. Resulting from the assembly of IF proteins, such as lamins, keratins, vimentin, desmin, neurofilaments and glial fibrillary acidic protein, IFs have diameters around 10 nm, which is smaller than microtubules (24 nm), but bigger than actin filaments (7 nm). IFs provide mechanical support to cells, participate in cell adhesion, facilitate intracellular communication, and maintain tissue integrity in cell-specific fashion.¹¹⁵ Since multiple types of PTMs, including phosphorylation, glycosylation, sumoylation, acetylation, and prenylation, regulate the functions of IFs,¹¹⁶ the formation of IFs unavoidably is controlled by enzymatic reactions. For example, Fields et al. suggested that protein phosphatase 1 (PP1) dephosphorylates lamin B for the nuclear lamin reassembly at the completion of mitosis.¹¹⁷ Broday et al. reported that reversible sumoylation of the cytoplasmic intermediate filament (cIF) protein named IFB-1 in C. elegans regulates cIF assembly by maintaining a cytoplasmic pool of nonpolymerized IFB-1.¹¹⁸ A recent report by Boyce and co-workers suggests that site-specific glycosylation of a serine residue in the “head” domain of vimentin promotes the intermolecular interactions of vimentin molecules to favor dimerization.¹¹⁹ This model suggested by Boyce agrees with the unique kinetic of IFs of vimentin, which are resulted from both lateral association and elongation (Figure 13), and indicates ENS in the formation of IFs.

Figure 13.

Formation of IF of vimentin. (A) O-GlcNAc transferase (OGT) glycosylates vimentin monomers to form vimentin dimer (structure with the central α-helical rod domain, green boxes). (B) Lateral assembly step: tetramers associate to form a unit length filament (ULF). (C) Elongation reaction of ULF. Adapted from Ref. 120. Copyright © 2016 by National Academy of Sciences.

Lipid Droplets.

Largely located in the adipose tissues, lipid droplets are lipid-rich cellular organelles for the storage and hydrolysis of neutral lipids, as well as a reservoir for cholesterol and acyl-glycerols. Although lipid droplets were considered just as fat depots initially, the discovery of the proteins in the coat of lipid droplets reveal that they are highly dynamic organelles and play a very important role in inflammatory responses, obesity, cancer, and atherosclerosis. The sizes of lipid droplets range from 20–40 nm to 100 μm. Generally, a phospholipid monolayer surrounds a neutral lipid core (mainly triacylglycerols (TAGs) and cholesteryl esters) to constitute the lipid droplets, which are decorated by a number of proteins involved in the regulation of lipid metabolism (Figure 14A).¹²¹ Although the exact mechanism of formation of lipid droplets remains to be elucidated, increasing numbers of studies suggested that enzymatic reactions control biogenesis of lipid droplets.¹²² The most widely accepted model for lipid droplet formation comprises several discrete steps: nucleation, growth, and budding (Figure 14B).¹²² Several enzymes are involved in this process. For example, DGAT produces neutral lipids for the growth of lipid droplets,¹²³ atlastin for ER shaping,¹⁰⁴ and FITM transfers phosphate-containing headgroups between lipids or in exchange for water.¹²⁴ In this case, multiple enzymatic reactions cooperatively lead to the biogenesis of lipid droplets, and likely more molecular details will emerge, such as the recent structure characterization of the seipin oligomers (Figure 14C).¹²⁵

Figure 14.

(A) The illustration of lipid droplet (LD). Adapted from Ref. 121. Copyright © 2013 by The Association for the Publication of the Journal of Internal Medicine. (B) Model of LD biogenesis. Adapted from Ref. 126. Copyright © 2017 by Elsevier Inc. (C) A Cryo-EM structure of seipin: ribbon diagram side (left) and top (right) views of the luminal domains of seipin. Adapted from Ref. 125. Copyright © 2018 by Sui et al.

Lysosomes.

Lysosomes are spherical, membrane-bound organelles for biomolecular degradation. A lysosome contains more than 60 different enzymes and has over 50 membrane proteins. In addition to receiving their substrates through endocytosis, phagocytosis, or autophagy, lysosomes play roles in other physiological processes, including cholesterol homeostasis, plasma membrane repair, bone and tissue remodeling, pathogen defense, cell death and cell signaling (Figure 15A).¹²⁷ To perform such a diverse range of functions, lysosomes must be dynamic and able to form higher-order assemblies that carry out different functions. For example, damaged lysosomes may cause unwanted cell death, so cells have evolved mechanisms for lysosome repair. A recent study suggested that components of the endosomal sorting complex required for transport (ESCRT)-I, ESCRT-II, and ESCRT-III are recruited for lysosome repair.¹²⁸ Among them, ESCRT-III is the most prominent one because it plays a role in all ESCRT mediated processes.¹²⁹ The higher-order assemblies formed by ESCRT-III proteins are long filaments. The recruitment of ESCRT-III proteins can start with the formation of ESCRT-I, and Pkh1/2 kinase phosphorylates Vps27 to regulate ESCRT-I recruitment.¹³⁰ This mechanism likely involves multiple ENS processes. In a recent study related to lysosome functions, Zoncu et al. reported that Rag GTPase controls the capture of nutrient-induced mTORC1 to the lysosome. As shown in Figure 15B, in the absence of amino acid, Rag heterodimers bind to the Ragulator on the lysosomal surface. In the presence of amino acid, the Rags switch to the active nucleotide state to destabilize the Rag interaction with Ragulator. Then, mTORC1 and Rag GTPases undergo spatial cycling, which regulates the mTORC1 activated by Rheb at the lysosomal surface.¹³¹ This observation, again, confirms that nature ubiquitously uses enzymatic reactions control the assembly and disassembly of higher-order protein structures.

Figure 15.

(A) Function and regulation of lysosomes. Adapted from Ref. 132. Copyright © 2009 by Springer Nature. (B) Rag GTPase controls the capture of nutrient-induced mTORC1 on lysosome for nutrient sensing. Adapted from Ref. 131. Copyright © 2018 by Springer Nature.

Microtubules.

In nature, an enzyme-regulated molecular assembly can play critical roles in many cell processes. The self-organization of tubulins to form microtubules is an example. As the name suggests, microtubules are hollow filaments made of tubulins in all eukaryotic cells. With the diameter of about 25 nm (Figure 16A) and as one type of cytoskeletons, microtubules not only support motility, but also are essential to many cellular functions. Like actins, tubulin itself is an enzyme, a GTPase. The hydrolysis of GTP to GDP catalyzed by tubulin maintains the unique asymmetric growth of microtubules (Figure 16B). All these dynamic behaviors depend on the assembly of microtubules, which, like the formation of actin filaments, is a representative example of an enzymatically controlled dynamic process. That is, microtubules grow (rescue state) and shrink (catastrophe state) at their ends. As a GTPase, β-tubulin binds and hydrolyzes GTP, which provides a conformational switch that helps to drive assembly and disassembly states of the microtubules. Such dynamic states depend on the local concentration of soluble dimers at the microtubule plus end. At low concentrations of tubulin, microtubules shrink, and vice versa.¹³³ Such asymmetric growth interacts with many microtubule-associated proteins (MAPs) to generate movements (Figure 16C). Microtubules also facilitate movement by acting as tracks to pull cargo for intracellular transportation. A recent study¹³⁴ of microtubules by cryo-EM not only provides useful insights into their dynamics, interactions with cellular components, and contributions to cellular functions, but also confirms enzymatic control in these assemblies.

Figure 16.

(A) Electron cryo-micrograph of straight and curved microtubules with a uniform diameter (scale bar = 50 nm). Some microtubules are bundled, flattened or otherwise disrupted. (B) Conformational change of αβ-tubulin accompanying GTP hydrolysis. In the GTP state (β-tubulin in green), α and β tubulin monomer interfaces result in a ‘straight’ tubulin dimer. In the GDP state, the αβ-tubulin dimer interface is curved by 5° (arrow), leading to a ‘bent’ tubulin dimer. During microtubule assembly, the plus ends form a sheet-like group of straight protofilaments. GTP-tubulin dimers (green) assemble on the ends, forming a cap of GTP-tubulin. GTP hydrolysis over time converts GTP-tubulin in the lattice to GDP-tubulin. During the microtubule disassembly phase, GDP-tubulin protofilaments curl and peel off the microtube plus ends. The transitions between growth and shrinkage states are termed catastrophe and rescue. Adapted from Ref. 133. Copyright © 2011 by Elsevier Inc. (C) MAPs and cargo interact with microtubules. (A, C) Adapted from Ref. 134.

Microtubule Ends.

The GTPase activity of β-tubulin, which hydrolyzes GTP to GDP, drives the dynamic instability of microtubules, such as transitions between catastrophe and rescue phases. As shown in Figure 16B, the growth and shrinkage of microtubules are asymmetric and occur at the plus end of microtubules. Since microtubule ends play important roles in cellular functions, considerable amount of research efforts have focused on the elucidation of the molecular details of microtubule ends. Although these studies have revealed that the ends of microtubules are structurally heterogeneous, there still are quite a few unknowns. For example, while microtubules are cylindrical, the ends of microtubule cap during assembly are uneven. This observation excludes the possibility that the ends of microtubule correspond to a single position or tubulin structure. A recent study suggested that curvatures of protofilaments on growing and shrinking microtubules are similar.¹³⁵ The tips of the protofilament are curved, indicating that the bent GTP-tubulin in solution has to be straightened to be incorporated into the microtubule wall. But it remains a challenge to elucidate how it happens from structural analysis alone. It is not unreasonable to speculate that the dynamics of other biochemical states of tubulins may provide clues, such as enzymatic reaction controlled PTM of tubulins.¹³⁶

Microtubule Organizating Centers (MTOCs).

In eukaryotic cells, microtubules emerge from MTOCs. There are several types of MTOCs, such as centrosomes, basal bodies, and spindle pole bodies. In an earlier section we already introduced centrosomes in the context of ENS (Figure 5), here we discuss spindle pole bodies (SPBs), which exist in fungi as the equivalent of centrosomes. SPBs are layered structures embedded in the nuclear envelop of budding yeast (Saccharomyces cerevisiae). Two types of microtubules, nuclear and cytosolic, grow out from SPBs. As one of the best-studied SPBs, yeast SPB consists of at least 18 different proteins, which assemble into a cylindrical organelle with a diameter of 100 nm. The core components of SPBs assemble into three vertical layers, named outer, inner, and central plaques. The outer and the inner plaques face the cytoplasm and the nucleoplasm, respectively, to organize cytoplasmic MTs and the nuclear MTs. Containing a two-dimensional crystal of the protein Spc42, the central plaque embeds in the nuclear membrane to anchor and interconnect the outer and inner plaques.¹³⁷ The kinase Cdc28 directly phosphorylates Spc42 to control its assembly into the SPB.¹³⁸ Moreover, Cdc28 also phosphorylates the Mps1 kinase,^139–141 which is known to regulate the phosphorylation and assembly of Spc42. Such multiple enzymes or multilevel controls of supramolecular assemblies (Figure 17B) support that ENS is a unifying theme in cell biology.

Figure 17.

(A) A schematic representation of the SPB proteins and their location within the organelle. Nuclear and cytoplasmic microtubules are illustrated and the distribution of their plus (+) and minus (–) ends is specified. IL 1, intermediate layer 1; IL 2, intermediate layer 2; cMT, cytoplasmic microtubules; nMT, nuclear microtubules. Adapted from Ref. 137. Copyright © 2013 by Elsevier Inc (B) Cdc28 regulates Spc42 assembly through multiple mechanisms (red color indicating enzymes or enzymatic reactions). Adapted from Ref. 138. Copyright © 2004 by Elsevier Inc.

Midbody.

The midbody is a transient assembly at the intercellular bridge that connects two daughter cells at the end of mitosis (Figure 18A, B).¹⁴² Midbody, controlling the final separation site of the daughter cells (Figure 18C),¹⁴³ plays roles in cell fate, polarity, tissue organization, and cilium and lumen formation. The dynamic nature of midbody implicates that the assembly process is controlled by enzymes. Indeed, a recent study on the interactomes of midbody reveals that MYPT1/PP1β phosphatase regulates microtubule dynamics and dephosphorylates the kinesin component MKLP1/KIF23 of the centralspindlin complex at the end of mitosis.¹⁴³ Specifically, after the completion of furrow ingression, the phosphatase MYPT1/PP1β accumulates at the midbody to dephosphorylate MKLP1. This dephosphorylation antagonizes the actions of kinase aurora B, thus strengthening the association of centralspindlin with PRC1. This result indicates that, during mitosis, the distribution of kinases and counteracting phosphatases controls the assembly and dynamic of cytokinesis proteins. Such control mechanisms indicate that the cellular functions of PP1 family phosphatases would likely involve in other ENS processes, such as the formation of cleavage furrow in early telophase.^75,143

Figure 18.

(A) TEM image of a HeLa cell in cytokinesis. A magnification of the midbody is shown on the right. MM, midbody matrix; MT, microtubules. Scale bar = 1 μm. (B) HeLa cells were fixed and stained to detect CITK, aurora B, tubulin and DNA. The bottom panel shows a 3× magnification of the midbody. (A, B) Adapted from Ref. 142. Copyright © 2016 by Elsevier Inc. (C) PP1β dephosphorylates MKLP1 to strengthen the association of centralspindlin with PRC1. Adapted from Ref. 143.

Mitochondria.

Mitochondria are complex organelles that play a central role in key cellular processes, particularly in acting as the hub for bioenergetic, biosynthetic, and signaling events.^144–150 The advances in mitochondrial biology have revealed that mitochondria, carrying their own DNA (mtDNA)^151–152 and constantly undergoing turnover,¹⁵³ fission,¹⁵⁴ and fusion,¹⁵⁵ are essential for metabolism,¹⁵⁶ stress response,¹⁵⁷ and cell death.¹⁴⁹ Because of the dynamic nature of mitochondria, it is not surprising that ENS plays roles in many mitochondrial processes. One interesting example of mitochondrial ENS is the aggregation of mitochondrial antiviral-signaling protein (MAVS) to form prion-like filaments for activating innate immune response against viruses.^158–160 Without infection, RIG-I bears constitutively phosphorylated serine or threonine residues in their CARDs and C-terminal domains, which represent a signaling-repressed state (i.e., an intramolecular interaction between the helicase domain and the CARDs of RIG-I resulting in an auto-repressed conformation). During infection, RIG-I binds to RNA to undergo an ATPase-dependent conformational change, which releases the CARDs for binding to several regulatory molecules, including phosphatase PP1—PP1α or PP1γ isoforms. PP1 removes the inhibitory phosphorylation marks in their CARDs. Then, the E3 ubiquitin ligases (e.g., TRIM25 or Riplet) attach ubiquitin polymers onto the CARDs and C-terminal domain for the tetramerization of RIG-I. The RIG-I tetramer interacts with the adaptor protein MAVS at the outer membrane of mitochondria to active MAVS. The activated MAVS self-assembles into prion-like filament structures, which further initiate the cascade of immune response.¹⁶⁰ Several enzymatic reactions (e.g., ATPase activity of RIG-I, dephosphorylation by PP1, and ligation of ubiquitin by E3 ligases) participate in the formation of MAVS filaments. Thus, MAVS assembly is a fine example of sophisticated ENS processes.

Mitotic Spindle.

The mitotic spindle is the cytoskeletal structure formed during mitosis of eukaryotic cells for separating chromosomes between the daughter cells.¹⁶² The major components of the spindle are microtubule polymers, therefore, the ENS process for microtubule dynamics plays a role. Besides tubulins acting as GTPases, many other enzymes, obviously, regulate the assembly of the mitotic spindle (Figure 20A).¹⁶³ For example, the attachment of chromosomes to spindle microtubules through kinetochores during mitosis is essential for genome integrity. The dynamic of kinetochore–microtubule (k–MT) attachment is regulated by multiple enzymes (Figure 20B),¹⁶⁴ such as polo-like kinase 1 (PLK1), aurora B kinase (AURKB), cyclin–cyclin-dependent kinases (CDKs), and phosphatases PP1 and PP2A. These enzymes regulate the phosphorylation status of their substrates (e.g., kinesin family member 2B (KIF2B), BUB1-related kinase 1 (BUBR1), bi-orientation of chromosomes in cell division 1 (BOD1), and survivins), thus collectively controlling the k–MT attachment stability.

Figure 20.

(A) The key proteins involved in the assembly of mitotic spindles. Adapted from Ref. 163. Copyright © 2008 by Elsevier Inc. (B) The attachment of kinetochores to microtubules is regulated by enzymes. Adapted from¹⁶⁴. Copyright © 2014 by Springer Nature.

The nucleus is the largest and probably the most important membrane-bound organelle in eukaryotic cells. Being discovered about three centuries ago, the nucleus stores the genes of cells in the form of chromosomes and acts as the control center of the cell. The nucleus consists of several major components, such as the nuclear envelope, the nuclear matrix, nuclear bodies (e.g., nucleoli), and nuclear speckles (Figure 2). The nucleus regulates numerous cell functions, including gene expression, processing of pre-mRNA, through many dynamic processes. These dynamic processes constantly undergo assembly and disassembly of higher-order structures, which are under the control of enzymatic reactions, as shown in the cases of nuclear bodies, nuclear membranes, nuclear speckles, nucleoli, nucleoli fibrillar centers, and the nucleoplasm.

Nuclear Bodies.

Nuclear bodies are membraneless structures inside the cell nucleus (Figure 21). There are several types of nuclear bodies, including Cajal bodies (CBs),^165–166 nucleolus,¹⁶⁷ promyelocytic leukemia protein (PML) nuclear bodies,¹⁶⁸ and ND10s.¹⁶⁹ CBs, being discovered by Cajal in 1903, can vary in size from 200 nm to 2 μm. Containing protein and RNA, CBs dynamically associate with cell cycle or cell stress. Recent advances of in vivo detection and imaging methods have revealed many insights on CBs, and support that multiple enzymatic reactions regulate the dynamic of CBs.^166,170 For example, coilin is the protein marker of Cajal bodies. One interesting feature of the components in CBs is that they also exist at other cellular locations. This observation suggests that enzyme catalyzed PTMs likely regulate these components. Increased evidence supports this notion. Hebert et al. reported that the interactions between coilin and RNA decreased when coilin was hyperphosphorylated,¹⁷¹ and Lazo et al. reported that VRK1 regulates Cajal body dynamics and protects coilin from proteasomal degradation during cell cycle.¹⁷⁰ Such studies have led to a hypothesized model about PTMs on the dynamics of colin and CB formation (Figure 21C), which involves kinases, ubiquitinases, methylases, and probably phosphatases. Future studies on the enzymatic regulation of CBs may shed more light on the detail of the dynamics of CBs. Similar to the formation of CBs, enzymatic reactions also control the formation of PML nuclear bodies.¹⁷² As shown in Figure 21D, oligomerization of non-SUMOylated PML monomers occurs via noncovalent interactions through the RBCC motif and covalent di-sulfide bonds between cysteine residues. A SUMO ligase, UBC9, then promotes (poly-)SUMOylation of the PML moieties for multiple SUMO–SUMO interacting motifs (SIM) interaction to form larger aggregates. Other binding partners carrying SIMs and/or SUMO residues can interact with the preassembled aggregates. These assemblies further self-organize to form a normal PML body.

Figure 21.

(A) Cajal bodies (green) and promyelocytic leukemia protein (PML) bodies (red) in human cells. DNA is stained with DAPI. Scale bar: 2 μm. Adapted from Ref. 173. Copyright © 2005 by Annual Reviews. (B) The dense coiled threads of this CB (arrow) are in direct continuity with the dense fibrillar component at the nucleolar surface. Scale bar = 1 μm. Adapted from Ref. 174. Copyright © 2005 by Wiley Inc. (C) A hypothetical model of the impact of PTMs on the interactions and localizations of SMN and coilin. During mitosis, CBs disassemble and coilin is hyperphosphorylated. CBs reform at G1 phase, in which mitotic coilin is paired in the CB. Less phosphorylated coilin interacts with SMN, Sm proteins, sRNPs and WRAP53. Methylation of coilin (and certain SM proteins) increases it interaction with SMN. Differential phosphorylation results in the differential interaction of coilin with SMN and Sm proteins to allow for subsequent snRNP biogenesis steps. Thus, many non-coding RNAs (curved lines) are enriched in the CB and coilin complex. Adapted from Ref. 166. Copyright © 2017 by Taylor & Francis. (D) Assembly of promyelocytic leukemia (PML) nuclear bodies controlled by SUMO ligases. Adapted from Ref. 172.

Nuclear Membranes.

Nuclear membrane, also referred to as the nuclear envelope (NE), is a double lipid bilayer that encloses the nucleus. Consisting of an outer nuclear membrane (ONM), an inner nuclear membrane (INM)), nuclear pore complexes (NPC), and the nuclear lamina, the NE disassembles and assembles to coordinate with the segregation of sister chromatids in each cell division cycle. Not surprisingly, multiple enzymatic reactions control such dynamic processes. For example, a key feature in the NE is the nuclear pore complexes (NPC),¹⁷⁵ which consist of around 1000 protein subunits (Figure 22A).¹⁷⁶ While it is reported that mitotic disassembly of NPC involves CDK1- and PLK1-mediated phosphorylation of key interconnecting nucleoporins,¹⁷⁷ increasing number of studies suggest that phosphatases play important roles in the NPC re-assembly.¹⁷⁸ Since phosphorylation and hyperphosphorylation of nucleoporins trigger NPC disassembly, dephosphorylation at the end of mitosis would likely promote NPC assembly (Figure 22B). Desai et al. reported the nucleoporin ELYS as a scaffold to recruit PP1.¹⁷⁹ Lamond identified another PP1 binding protein, Repo-Man.¹⁸⁰ The study of Repo-Man during mitotic exit suggests that Repo-man binds stably to PP1 for the accumulation of some NPC components, namely Nup153 and importin β.¹⁸¹ Moreover, the local activation of the phosphatase is able to trigger NPC reformation even in the presence of high CDK1 and PLK1 activity. Another phosphatase, PP2A may dephosphorylate Nup153 for the reformation of NPC. In addition, Nup153 is also a PP1 substrate.¹⁷⁸ More studies likely will reveal the important role of phosphatases for controlling large protein assemblies like NPC.

Figure 22.

(A) Right: overview of the composite structure of the entire NPC, unassigned density is shown in cyan; the nuclear ring is facing up. Right: Zoomed-in view of the inner ring region framed in right. High-resolution structures (colored ribbons) are shown in the context of the tomographic map (transparent isosurface). Adapted from Ref. 176. Copyright © 2017 by AAAS. (B) Mitotic NPC assembly is regulated by Ran and phosphatases. The local generation of Ran-GTP on the chromatin allows release of inhibitory NTRs from nucleoporins (exemplified for MEL28). MEL28 binds to the decondensing chromatin and recruits the Y-complex. Dephosphorylation of nucleoporins enables their interaction and allows NPC assembly. Adapted from Ref. 182. Copyright © 2016 by The Company of Biologists.

Nuclear Speckles.

Nuclear speckles (NSs) or splicing speckles, also called interchromatin granule clusters, are self-organizing membraneless structures for the storage and modification of splicing factors¹⁸³ and may play a general role in RNA metabolism. Recent advances suggest that many enzymes act within NSs to facilitate the regulation of gene expression.^184–185 The best known molecular mechanism of nuclear speckle localization is a phosphorylation/dephosphorylation cycle of the arginine/serine repeat (RS) domain of serine rich (SR) proteins. Although it is generally believed that RS domain phosphorylation drives SR proteins from NSs to the nucleoplasm,¹⁸⁶ a recent study reveals that synergistic interplay between PP1 and two splicing kinases (SRPK1 and CLK1) regulate the location of SR proteins, such as SRSF1.¹⁸⁷ Adams et al. reported that SRSF1 binds to PP1 through the RRM1 domain and represses the catalytic activity of PP1 through an allosteric mechanism. This interaction would allow phosphorylation of hypophosphorylated SRSF1 to act as the substrates of kinases (e.g., SRPK1 and CLK1). The intermediate phosphorylated SRSF1 would reside in the NSs. Further phosphorylation would generate hyperphosphorylated SRSF1 to leave the NSs and to enter the nucleoplasm. The PP1 can dephosphorylate the hyperphosphorylated SRSF1 and bring it back to NSs. Thus, the balanced actions of phosphatase and kinases would result in the NS localization of SR proteins (Figure 23B).¹⁸⁷ Obviously, SR proteins inside the NS would interact with other proteins to form protein assemblies for RNA storage and modification.

Figure 23.

(A) Representative microscopy images of NSs in cells (labeled for SRSF2). Bars = 10 μm (for the fibroblast, SK-MC and SK-N-MC) and 5 μm (for the lymphoblast). Adapted from Ref. 184. (B) Speckle localization is regulated by the balanced actions of kinases (e.g., SRPK1 and CLK1) and phosphatases (e.g., PP1). Adapted from Ref. 187. Copyright © 2017 by Elsevier Inc.

Nucleoli.

As the largest membraneless organelle inside the nucleus, the nucleolus¹⁶⁷ is the site of ribosome biogenesis and a cellular stress sensor. Nucleoli contain three substructures: the fibrillar centers (FCs), dense fibrillar component (DFC), and the granular component (GC). Ribosomes synthesize proteins from amino acids according to the need of cells for new proteins. Nature has evolved elaborated mechanisms to assemble ribosomes in the nucleolus, which, of course, involves enzymatic reactions to regulate assembling processes. For example, one of the proteins found at high levels in the nucleolus is nucleophosmin (NPM), which binds with the proteins containing arginine-motifs (R-motifs). One binding mode is the multimer of NPM interacting with multiple R-motifs of other proteins. Such a binding is dynamic or reversible, and is controlled by enzymatic switch: phosphorylation and dephosphorylation. Kriwacki et. al. reported that the unphosphorylated N-terminal domain of NPM (NPM-N) predominantly populates the pentameric state. Phosphorylation of pentameric NPM-N destabilizes the oligomeric structure, and leads to conformational changes to expose additional, otherwise buried sites for further phosphorylation. The phosphorylation at the buried sites results in monomeric folded or monomeric disordered structures.¹⁸⁸ Thus, enzymes control the degree and state of phosphorylation to regulate the functions of NPM (Figure 24A), which interacts with more than one hundred proteins inside cells. The same lab also proposed that such multivalent interactions with proteins containing R-motifs and ribosomal RNA (rRNA) is one mechanism of nucleolar localization, and involves phase separation of proteins within the nucleolus.¹⁸⁹ This mode of action appears to operate in the context of the tumor suppressor p14arf (Figure 24B).¹⁹⁰

Figure 24.

(A) Pentamer of NPM. Adapt from Ref. 189. (B) The illustration of p14arf‐NPM-N heteropolymerization. (i) p14arf may form homo‐oligomers to interact with different NPM‐N pentamers (blue spheres or ribbons). (ii) Both the N‐terminal end and the central regions of p14arf may interact with different NPM‐N pentamers. (iii) The central region of p14arf interacts with two different NPM-N pentamers at the same time with its two arginine clusters, leaving the N‐terminal ends free to interact with another NPM-N pentamers. (iv) Upon sequential phosphorylation of different exposed and buried Ser/Thr sites played by several kinases, NPM-N monomerizes and unfolds thus releasing active p14arf in the nucleoplasm. Red dots represent phosphorylation sites in NPM-N. Adapted from Ref. 190. Copyright © 2017 by Federation of European Biochemical Societies. (C) Enzymatic control of the conformation and assembly of NPM. Adapted from Ref. 188. Copyright © 2014 by National Academy of Science.

Nucleoli Fibrillar Center.

In the nucleolus, fibrillar centers (FCs) and dense fibrillar components (DFCs) are sites for rDNA transcription and pre-rRNA processing. A recent study by Chen et al. reported the assembly of DFC for sorting nascent pre-rRNA. They found that fibrillarin (FBL) self-assembles to form clusters, which integrate into the DFC surrounding the FC. The nascent pre-rRNA binds FBL. Because FBL diffuses to the DFC, the clusters immobilize FBL-interacting pre-rRNA, thus facilitating pre-rRNA processing and DFC formation. The authors attributed this pre-RNA sorting to liquid-liquid phase separation. The logical question would be what factors control the phase separation? The presence of ubiquitylation sites and phosphorylation sites on FBL implicates that enzymatic control of the PTMs of FBLs likely involves the formation of the clusters of FBLs. This presumption is supported by the report of Komada et al., which reported that USP36 plays a crucial role in regulating the structure and function of nucleoli by deubiquitylating various nucleolar substrate proteins, including fibrillarin.¹⁹¹ Other enzymes, such as Pol I, also play a role in the formation of this dynamic interface of FC and DFC.

Nucleoplasm.

The nucleus contains the nucleolus and nucleoplasm, which are mainly nucleoproteins and nucleic acids. The most important content of the nucleoplasm is chromatin, a complex of DNA and protein. Chromatin serves as the basic constituent of the chromosomes, which is the site of gene storage. There are two types of chromatin materials: heterochromatin and euchromatin. While the former is metabolically and genetically inert, the latter contains metabolically active DNA and is genetically important. Recently, to account for the dynamic features of heterochromatin, Narlikar et al. reported that human HP1α protein (also known as CBX5) displays liquid demixing in vitro, and the HP1α demixing requires N-terminal phosphorylation or DNA binding.¹⁹³ They proposed a model (Figure 26) in which N-terminal extensional phosphorylation of HP1α leads to conformation change, which results in the dimerization of HP1α, which can assemble on the chromatin. Further increase in the concentration of the phosphorylated HP1α promotes the oligomerization, which interacts with chromatin to result in phase separation. This liquid phase would incorporate the macromolecules that interact with HP1α and exclude non-interacting molecules. In a related study, Karpen et al. proposed that the formation of heterochromatin domains was mediated by phase separation by examining the behavior of heterochromatin protein 1 (HP1) in Drosophila cells.¹⁹⁴ But phosphorylation appears missing from the cases of HP1 in Drosophila, implicating that other enzymatic reaction or interaction with other partners may control the phase separation.

Figure 26.

(A) Schematics of HP1α phosphorylation. CD, chromodomain; CSD, chromoshadowdomain; CTE, C-terminal extension; H, hinge; NTE, N-terminal extension. (B) Model for how HP1α switches between a compact and extended state: the N-terminal phosphates interact with basic hinge residues to stabilize inter-dimer contacts in the extended state and promote higher-order oligomerization. Adapted from Ref. 193. Copyright © 2017 by Springer Nature.

Peroxisome.

Peroxisomes, existing in virtually all eukaryotic cells, are single membrane-bound metabolic organelles mainly for degrading fatty acids via β-oxidation and detoxifying hydrogen peroxide. Peroxisomes are highly dynamic and may undergo drastic changes in size, shape, abundance, and protein composition according to the need of cells. To meet the demand of fast dynamics of peroxisomes, cells rely on protein phosphorylation to govern peroxisome biogenesis and functions.¹⁹⁵ For example, in the study of the autophagy in yeast that involves peroxisomes, which is called pexophagy, a Hrr25-mediated phosphorylation of Atg36 is necessary to form the assembly on the surface of peroxisomes for priming the peroxisome for autophagy (Figure 27A).¹⁹⁵ Unlike the case of yeast, which relies more on phosphorylation, mammalian cells employs ubiquitination of the pexophagy. Under stress conditions, ubiquitination of PEX5 and ABCD3 occurs, which leads to the binding of NBR1. This binding allows the tether of peroxisomes to the phagophore.¹⁹⁶ Since the peroxisome is a sophisticated organelle, other peroxisomal membrane proteins involve phosphorylation and ubiquitination to initiate pexophagy. Further studies with more emphasis on enzymatic controls would likely reveal more insights on the biogenesis and functions of peroxisomes.

Figure 27.

Mechanisms of pexophagy in (A) yeast and (B) mammalian cells. (A) Hrr25 phosphorylates Atg30 and Atg36 to allow recruitment of the autophagic scaffold protein Atg11. Recognition of the peroxisome by the autophagic machinery requires Pex14. (B) Several stress conditions (e.g., hypoxia in mammalian cells) leads to ubiquitination of PEX5 and ABCD3 and the recruitment of ubiquitin-binding autophagy receptors NBR1 and SQSTM1 for tethering peroxisomes to the phagophore. Adapted from Ref. 196. Copyright © 2018 by Springer Nature.

Plasma Membrane.

The plasma membrane (also known as the cell membrane), is a biological membrane to protect the cell from its environment. As a lipid bilayer made of mainly cholesterols and phospholipids, the plasma membrane contains many protein molecules as its integral components and is the site for important cellular functions, including signaling and endocytosis. The dynamic compositions and shapes of the plasma membrane, again, imply the enzymatic control of assemblies (or oligomerization).. For example, one of the most studied lipid families is phosphoinositides. Multiple inositol lipid kinases and phosphatases regulate the degree of phosphorylation of the inositol ring in membrane phosphoinositides to control protein assemblies on the plasma membrane for cell signaling.¹⁹⁷ That is, the binding of inactive proteins (E_inactive) and appropriate phosphoinositides at the plasma membrane can activate or recruit additional proteins, and the activated proteins (E_active) can return to the cytosol as effectors for further actions (Figure 28A). In a recent study, Yuan et al. reported that, during the necroptosis mediated by the RIPK1-RIPK3-MLKL signaling, RIPK3 and TAM kinases phosphorylate Ser358 and Tyr376 of MLKL, respectively. These enzymatic reactions lead to the oligomerization of MLKL on the plasma membrane and cause membrane rupture, thus resulting in cell death (Figure 28B).¹⁹⁸ Many receptor-mediated signaling events also take place near the plasma membrane, especially those involving kinases.^199–200

Figure 28.

(A) Enzymatic reaction of phosphoinositides PtdIns(x,y)Pn control several kinds of signaling processes. E_inactive: proteins contain phosphoinositide-recognition domains and assume an inactive conformation. Phosphorylated phosphoinositide (PtdIns(x,y)Pn+1) recruits the protein to the membrane, and to interact with integral (I) or peripheral (P) membrane proteins. The complex can remain active at the membrane and recruit additional proteins. The protein can return to the cytosol in an activated (i.e. phosphorylated) form (E_active). Adapted from Ref. 197. Copyright © 2005 by The Company of Biologists. (B) Diagram of the mechanistic findings for the TAM-MLKL input into the necroptosis pathway, which involves enzymatic phosphorylation and assembly. Adapted from Ref. 198. Copyright © 2019 by Elsevier Inc.

Rods and Rings.

Originally discovered by the use of human autoantibodies, rods and rings (RRs) are filament-like cytoplasmic structures containing proteins involved in the biosynthesis of nucleotides.²⁰¹ The size, shape, and quantity of RRs vary and are context-dependent (e.g., depending on cell types), suggesting that RRs may be transient and are likely resulted from ENS. A recent study on the composition complexity of the RRs formed by inosine monophosphate dehydrogenase (IMPDH)²⁰² supports this notion. In the presence of its inhibitors (e.g., mycophenolic acid (MPA) or ribavirin), IMPDH self-assembles to form polymers that appear as RRs across a wide variety of cell types. Although it was suggested that RRs play a role in the regulation of de novo guanine nucleotide synthesis, the function and regulation of RRs is largely unclear. Kahn et al. show that a regulatory GTPase, ARL2, a subset of its binding partners, and several ER resident proteins localize to RRs (Figure 29A). They found that RRs matured with IMPDH first forming aggregates, followed by ARL2, and only later calnexin, a marker of the ER, suggesting that the formation of RRs was complicated and likely regulated by enzymatic reactions. In fact, inhibiting IMPDH to result in the aggregation of IMPDH confirms that the enzyme activity of IMPDH regulates the assembly of IMPDH. CTP synthase (CTPS) is another enzyme known to form RRs. Recent study by Kollman et al.²⁰³ reveals that human CTP synthase 1 (hCTPS1) polymerizes in the presence of UTP and ATP substrates, but not in the presence of CTP and ADP, which differs from the bacterial CTPS1 (Figure 29B). They also found that hCTPS1 assembled into filaments, which likely locked hCTPS1 in a more active conformation, thus resulting in enhanced activity. Interestingly, the formation of the tetramer from the CTPS requires phosphorylation,²⁰⁴ which further supports the enzymatic regulation of RRs formation.

Figure 29.

(A) ARL2 localizes to IMPDH2-positive structures that are inducible with MPA in HeLa cells. Scale bar in bottom left panel = 10 μm and is the same for each image. Adapted from Ref. 202. (B) The model for the allosteric regulation of CTPS filament assembly. ecCTPS and hCTPS1 undergo a conserved conformation cycle but have opposite determinants for filament assembly. Adapted from Ref. 203. Copyright © 2017 by Springer Nature.

Vesicles.

Vesicles are membrane-bound organelles, consisting of liquid or cytoplasm enclosed by a lipid bilayer, such as endocytic vesicles, secretory granules, peroxisomes, and lipid droplets. About 10% of human proteins localize to vesicles, which define the biological functions of vesicles. One of the most studied types of vesicles are synaptic vesicles (SVs) (Figure 30A), which are the secretory organelles that store and secrete non-peptide neurotransmitters at the synapse. The association of synapsins, which are highly abundant phosphoproteins,²⁰⁵ during the dynamics of SVs (e.g., generation and regeneration) indicates phosphorylation-dependent interactions are likely controlled by enzymatic reactions. For example, synapsins are substrates for several protein kinases. The phosphorylation of synapsins is a response to a wide range of chemical and electrical stimuli. That is, PKA phosphorylated synapsins would detach synapsins from SVs and diffuse in the synaptic bouton and into the axonal compartment. After the stimulation (or in the recovery phase), phosphatase (e.g., PP2A) removes the phosphate groups from synapsins to allow the reclustering of the SVs (Figure 30B).²⁰⁶ This process, of course, is complicated and involves other proteins. Other proteins, such as amphiphysin I and II, dynamin I, and synaptojanin, exhibit a similar phosphorylation/dephosphorylation cycle. Such a dephosphorylation step is required for these proteins to initiate the process of SV endocytosis. Moreover, multiple enzymes (e.g., V-ATPase and trimeric GTPase) present on the surface of SVs, further support that enzymatic reactions regulate the dynamics of SVs.

Figure 30.

(A) The composition of synaptic vesicles. Adapted from Ref. 207. Copyright © 2006 by Elsevier Inc. (B) Synapsin dispersion and recovery in pre-synaptic terminals. That is, synapsin dispersion and recovery in synaptic boutons: phosphorylated synapsin diffusing in the bouton and into the axonal compartment; after the stimulus, synapsin molecules are dephosphorylated and re-cluster on SVs in the bouton. Adapted from Ref. 206. Copyright © 2010 by Elsevier Inc.

2.2. Extracellular ENS

Cells must respond to its surroundings, such as interacting with neighboring cells in tissues or adapting their environment for survival. To carry out these essential functions, cells have evolved extracellular matrices (ECMs) and cell adhesion molecules to transduce signals into the cells and to coordinate with the cytoskeleton dynamics. Increasing evidence suggests that the assembly of these molecules functioning in extracellular space are controlled by enzymatic reactions,^208–209 as illustrated by the following representative cases.

Extracellular Matrices.

Extracellular matrices (ECMs), consisting of the assemblies of proteins and polysaccharides, render distinctive mechanical properties to the tissues and mediate intercellular interactions. To provide a wide range of physical strengths, ECMs are networks of oligomeric assemblies of ECM molecules. To adapt to the environment or to meet the need of intercellular communications, ECMs constantly undergo remodeling, which is regulated by enzymatic reactions. For example, the formation collagen fibers^210–211 is a classic example of ENS of protein assemblies. In animal tissues, collagens exist as very long fibers with a unique axial periodicity (Figure 31A). The fibers act as the major biomechanical scaffold for cell attachment and macromolecular anchorage, thus defining the shape and form of tissues. The collagen molecules self-assemble to form a trimer called procollagen. The procollagen consists of a triple-helical domain that can be as long as 300 nm with non-helical domains at the N-terminal and the C-terminal. Procollagen N-proteinase and procollagen C-proteinase cleave the N-propeptide and C-propeptide regions, respectively, to allow the triple-helical domains to form fibers. The oxidation by lysyl oxidase further crosslinks the fibers to modulate the mechanical strength of the collagen fibers (Figure 31B).

Figure 31.

(A) Negatively stained unipolar collagen fibril isolated from embryonic chick metatarsal tendon. Enlargement shows the axial periodicity. (B) Schematic representation of collagen fibril formation by cleavage of procollagen. Sequential cleavage of the N-propeptides (by procollagen N-proteinase, which are ADAM 2, 3, 14) and the C-propeptides (by procollagen C-proteinase, which are the BMP-1/tolloid family) of procollagen generates collagen that self-assembles into unipolar collagen fibrils. Adapted from Ref. 211.

Cell junctions.

Cell junctions are the regions where cells meet and their plasma membranes closely face each other (Figure 32A).²¹² There are four main types of cell-cell junctions: tight junctions,²¹³ adherens junctions,²¹⁴ desmosomes,²¹⁵ and gap junctions.²¹⁶ There are two types of cell-ECM junctions: focal adhesions and hemidesmosomes. The molecules that constitute the junction typically have ectodomains (exocellular parts) and endodomains (intracellular parts). In epithelial cells, junctional molecular complexes interact with cytoskeletons (e.g., intermediate filaments) to modulate the intercellular communications and to coordinate the morphogenesis and homeostasis of tissues.²¹⁷ These processes require a dynamic continuum of higher-order structures, which usually involve enzymatic reactions and self-assembly. For example, a recent study of the formation of tight junctions, by Honigmann et al.,²¹⁸ reveals that multivalent interactions and enzymatic reactions (de-/phosphorylation) regulate the phase transition of zonula occludens (ZO) proteins to form the supramolecular assembly and tight junctions as continuous networks of adhesion strands between cells.²¹⁸ Because ZO1 and ZO2 are phosphoproteins and their phosphorylation state involves tight-junction formation,^219–221 the authors tested whether the phosphorylation state affected the phase-separation ability of ZO proteins. Using casein kinase-2 (CK2) and lambda phosphatase, respectively, to phosphorylate and to dephosphorylate ZO1, ZO2, and ZO3 in vitro, they found that dephosphorylated ZO1, ZO2, and ZO3 efficiently phase separated into liquid droplets. But all phosphorylated ZOs are unable to phase separate under the tested conditions. Using mass spectrometry, they reported that CK2 phosphorylated ZO1 at 47 residues (Figure 32B). These results indicate that dephosphorylation likely actively triggers phase separation of ZO1 and ZO2 in a locally controlled manner.

Figure 32.

(A) Cell junctions typical for normal epithelial and endothelial cells and the relevant major proteins of junctions. Adapted from Ref. 212. Copyright © 2018 by Elsevier Inc. (B) De-/Phosphorylation controls phase separation of ZO proteins. Scale bar = 5 μm. Adapted from Ref. 218. Copyright © 2019 by Elsevier Inc.

Focal Adhesion Sites.

Focal adhesion¹⁰⁸ is one type of cell junction between the cell and extracellular matrix (ECM) (Figure 32A). At the focal adhesion sites, the transmembrane proteins, such as integrins, interact with their extracellular ligands, and transduce both mechanical and biochemical signals into the cells via the intracellular multipleprotein assemblies that connect to the actin cytoskeleton. While the enzymatic control of the intracellular molecular assemblies at the focal adhesion sites²²² is well-established (Figure 11), the understanding of the extracellular domains of focal adhesion sites focuses more on the ligand-receptor binding. This includes the binding of integrins with the RGD-motif on the ECM proteins. The accumulation of the knowledge on the phosphorylation of ECM proteins²⁰⁸ implies that enzymatic reactions in extracellular spaces also likely play a role in focal adhesion sites, which remain to be elucidated.

3. Enzymatic Noncovalent Synthesis in Cell Free Conditions

Despite significant advancement in noncovalent synthesis and the observations of enzyme-controlled assembly of proteins, little attention was paid to the development of ENS until the report of using enzymatic reaction to trigger self-assembly of small molecules in water to form supramolecular hydrogels.³⁷ Hydrogels formed by three-dimensional elastic networks whose interstitial spaces are filled with water present many useful properties. Although enzymes can catalyze the formation of covalent bonds to crosslink polymers to form hydrogels,²²³ there was no report on using enzymes to control self-assembly of small molecules for forming supramolecular hydrogels (via noncovalent bonds). That is, until the use of alkaline phosphatase (ALP) to catalytically dephosphorylate Fmoc-phosphotyrosine for generating Fmoc-tyrosine, a hydrogelator, to form a hydrogel, was reported.³⁷ This work illustrates a simple and reliable way to explore the emergent properties of small molecules in water, such as self-assembly or phase transition. Because it is easy to observe sol-gel phase transition, the use of enzymes to trigger the formation of supramolecular hydrogels becomes a facile assay to identify the substrates for ENS. Some of those examples^{37,102,224–285} have been summarized in a recent review on supramolecular hydrogels.²⁸⁶ The following section mainly discusses the works published after that review, according to the types of the enzymes.

3.1. Phosphatases

3.1.1. The Early Example and A Protocell Model

Phosphatases remove phosphate group(s) from a substrate. According to the pH for its optimal activity, nonspecific phosphatases fall into two categories: alkaline and acid phosphatases.²⁸⁷ According to its substrates, specific phosphatases are of three kinds: tyrosine,²⁸⁸ serine and threonine,²⁸⁹ and histidine²⁹⁰. Because the broad substrate scope and rapid kinetics of ALP provide a platform for demonstrating many concepts of ENS of man-made molecules, ENS catalyzed by alkaline phosphatases (ALP) has received the most extensive exploration. The first case of ENS of small molecules employs ALP to dephosphorylate a simple substrate N-(fluorenyl-methoxycarbonyl) tyrosine phosphate (Fmoc-_pY, 1) (Figure 33A).³⁷ In that work, 50 mM of 1 at basic pH (9.6) exists as a solution, which turns into a hydrogel (Figure 33B) 30 minutes after the addition of ALP (0.15 U/mL). TEM reveals the nanofibers of Fmoc-Y (2) in the hydrogel (Figure 33C). The formation of a hydrogel or hydrogelation itself turns out to be a facile assay for indicating ENS if the substrate concentration is adequate. Depending on its concentration and pH, it is feasible for 1 to self-assemble to form micelle-like structures, as elucidated from the mechanistic study reported by Ulijn et al.²⁹¹ This result also suggests successive dephosphorylation by ALP, an overlooked subtlety. Nevertheless, this rather simple demonstration³⁷ established the feasibility of ENS of synthetic molecules and led to the researches of ENS of man-made molecules, as summarized by some recent reviews.^{102,292–297}

Figure 33.

(A) ALP catalyzes the dephosphorylation of 1 to form 2. (B) The sol-gel transition after adding ALP to the solution of 1. (C) The TEM image of the hydrogel formed by 2. Adapted from Ref. 37. Copyright © 2004 by Wiley Inc.

Recently, Mann et al. demonstrated the use of ENS catalyzed by ALP for spatiotemporal control of hydrogelation in a model system such as protocells.²⁹⁸ Based on their previous work of protocells,²⁹⁹ the authors prepared ALP-containing water-in-water silica colloidosomes. The addition of 1 to the aqueous continuous phase allows the diffusion of 1 through the semipermeable membrane and results in ALP-catalyzed dephosphorylation of 1. The subsequent production of 2 leads to spontaneous self-assembly to form nanofibers. This simple model system produces a time-dependent sequence of supramolecular assemblies over three stages: First forming an outer wall of 2 on the external surface of the colloidosome; then forming a cytoskeletal-like matrix within the interior of the microcapsules; and finally forming denser assemblies of 2 as an extended matrix of colloidosomes. Although this is a simple model system that could benefit from more quantitative analysis, the simple coupling of ENS and compartmentation is able to result relatively sophisticated colloidosomes comprising spatially delineated regions of hydrogels, further implying the importance of ENS in the context of the origin of life. The same lab also further extended the colloidosome-based protocells to mimic phagocytosis.³⁰⁰ Although phagocytosis is a highly complicated process in living cells, the use of a model system to mimic the essence of biological phenomena is able to reveal useful insights and principles for the design of ENS processes for future applications.

3.1.2. The Broad Substrate Scope of ALP

A unique advantage of ENS catalyzed by ALP is the broad substrate scope of ALP. It even allows otherwise water insoluble molecules to self-assemble in water, as reported by Yang et al.²⁴⁰ In that study, a highly hydrophobic small molecule, Fmoc-Tyr-OMe (4, Figure 34A), being insoluble in water, can’t form a hydrogel by either changing pH or adjusting temperature. But Fmoc-_pTyr-OMe (3), the phosphorylated 4, exhibits relatively good solubility in water. The addition of ALP to the solution of 3 results in the hydrogel of 4 (Figure 34B). This simple demonstration underscores that ENS is a powerful approach for exploring the self-assembly of a wide range of molecules, including hydrophobic molecules that have extremely poor water solubility.²³⁷ One class of such molecules that remain to be explored would be hydrophobic drug candidates known to form aggregates.³⁰¹ Moreover, the broad substrate scope of ALP permits the evaluation of ENS of known self-assembling building blocks by a simple phosphorylation to make a substrate of ALP. For example, it is known that the core segment (GNNQQNY) of yeast prion Sup35 is able to self-assemble to form cross-β spine structures. The attachment of naphthyl (Nap) group to the N-terminal of GNNQQNY produces a hydrogelator Nap-GNNQQNY (6, Figure 34C). The phosphorylation of tyrosine in 6 generates Nap-GNNQQN_pY (5). The addition of ALP to the solution of 5 converts 5 to 6, which results in the hydrogel consisting of the nanofibers of 6 formed by noncovalent interactions (Figure 34C). ALP also is able to dephosphorylate D-phosphopeptide, Nap-gnnqqn_py (7, which is the enantiomer of 5) to form a hydrogel (here, the lower-case letters represent D-amino acids, same for the rest of this review). The hydrogel of 8 is relatively compatible to cells.³⁰² It would be interesting to examine whether the ENS of other cross-β spine forming peptides leads to supramolecular hydrogels and their cell compatibility.

Figure 34.

(A) The structures of 3 and 4. (B) Optical images of the suspension of 4 (1.0 wt%, PBS, pH 8.0), the solution of 3, and the gel formed 10 minutes after adding ALP (16 U/mL) to the solution of 3 (1.0 wt%, PBS, pH 8.0). Adapted from Ref. 240. Copyright © 2009 by American Chemical Society. (C) The structure of 5/6 and 7/8. (D) TEM images of the nanofibers in the gels formed by 6 and 8. Scale bar = 100 nm. Adapted from Ref. 302. Copyright © 2016 by Royal Chemical Society.

Following their work of the ENS of 4 and their pioneering study of using ENS formed hydrogels as vaccine adjuvants, Yang et al. further evaluated the L- and D-peptide hydrogels formed by ENS.³⁰³ They synthesized two phosphotripeptides, Nap-GFF_pY-OMe (9) and Nap-Gff_py-OMe (11) (Figure 35A). After confirming the formation of hydrogels of Nap-GFFY-OMe (10) and Nap-Gffy-OMe (12), they used the hydrogels as vaccine adjuvants and found that both L- and D-peptide hydrogels are more potent than the clinically used alum adjuvant in the increase of the IgG production of OVA. Notably, the OVA encapsulated D-peptide hydrogel exhibits a slightly more efficient accumulation of OVA in the lymph nodes for preventing tumor growth than the L-peptide hydrogel does. This study, indeed, has led to several impressive follow up reports^304–312 on adjuvant studies from the same lab. Markedly, Yang et al. combined tuftsin (TKPR) and Nap-Gffy to develop a novel self-assembling molecule Nap-GffyTKPR, and reported that the nanofibers of Nap-GffyTKPR enhanced the phagocytic activity of macrophages, promoted the maturation of DCs, stimulated the expression of cytokines, and elicited a strong CD8⁺ T immune response. As suggested by the authors, this work³¹³ demonstrated a powerful immune stimulating nanomaterial for cancer immunotherapy. Based on the application of Gffy, Liu et al. also used Gffy to enable the self-assembly of fenofibrate and ketoprofen for treating nonalcoholic fatty liver disease.³¹⁴

Figure 35.

(A) The structures of 9/10 and 11/12. (B) The structures of the hexapeptide (14) in an ITIM, and its enantiomer (16) and the corresponding phosphorylated precursors (13, 15, and 17). (C) The structure of the phosphoserine containing peptides (22, 24, 26, and 28) and their corresponding dephosphorylated products (23, 25, 27, and 29).

One often overlooked factor for enabling the self-assembly of molecules is the concentration of the self-assembling building blocks. Thus, it is rather easy to observe hydrogelation from ENS if the self-assembling blocks have adequate concentrations, even without the attachment of a naphthyl or Fmoc group, as evidenced by the ENS of an immunoreceptor tyrosine based inhibitory motif (ITIM).³¹⁵ As shown in Figure 35B, the hexapeptide, LYYYYL (14), is a ITIM.³¹⁶ ALP catalytically dephosphorylates 13 (at 0.5 wt% and pH 7.4) to form 14, which self-assembles in water to form a hydrogel. ALP is able to catalyze the dephosphorylation of the enantiomer (15) or retro-inverso isomer (17) of 13 to form the same peptide (16), which also self-assembles in water to result in supramolecular hydrogels. Although dephosphorylating 15 or 17 should give the same hydrogel of 16, the hydrogel formed by dephosphorylating 17 is more cell compatible than the gel formed by dephosphorylating 15. For example, when three mammalian cell lines, Saos2, HeLa, and HS-5, which express high, medium, and low levels of ALP, respectively were incubated with 16, all the cells are quite viable. 17 exhibits much lower cytotoxicity to Saos2 cells than 15 does, though the dephosphorylation products from 15 and 17 both are 16. Similarly, the hydrogel formed by ENS from 17 is the most compatible with HeLa cells. This result implies that the process or the history of ENS likely determines the properties of the peptide assemblies, which has profound implications in biomedical applications of ENS. Since immunomodulatory peptides play important roles in biology, further exploration of ENS of other ITIMs or immunoreceptor tyrosine-based activation motifs (ITAMs) may lead to useful soft materials for immunology.

Although it is used extensively to remove the phosphate group from a tyrosine residue, ALP is able to dephosphorylate phosphoserine for self-assembly, as shown by the enzymatic hydrogelation of the peptides containing phosphoserine.³¹⁷ In that study, a series of precursors contain the Nap-FF motif and an L- or D-phosphoserine residue (18 and 20), or both L- or D-phosphoserine and L- or D-phosphotyrosine residues (22–28) (Figure 35C). At the concentration of 0.5 wt%, the addition of ALP into the solutions of these precursors results in the nanofibers formed by the corresponding dephosphorylated peptides (19-29). Hydrogelation, however, only occurs for the peptides that incorporate tyrosine residues (23-29). While it indicates that tyrosine increases the ability of self-assembly of the peptides, this work, again, confirms the broad substrate scope of ALP for exploring ENS.

ALP also catalyzes the dephosphorylation of nanoparticles for self-assembly, as reported in the case of magnetic nanoparticles.²⁷⁸ Recently, Shu and Ding have shown that ALP triggers self-assembly of quantum dots (QDs) by catalytically dephosphorylating the phosphopeptides attached on the QDs.³¹⁸ The authors attached the phosphorylated peptide (GGFF_pY) on the QDs (Figure 36A). ALP hydrolyzes the phosphorylated peptide to form peptide self-assemblies on the QDs surfaces, which also enable the self-assembly of QDs (Figure 36B). According to the authors, this ENS process, coupling with fluorescence resonance energy transfer (FRET) between phosphopeptide modified QDs and dansyl chloride (DNS), has a significant advantage, such as detecting ALP at an exceedingly low detection limit, 0.001 U/L, which is remarkable.

Figure 36.

(A) TEM images of the GGFF_pY decorated QDs in the absence and (B) presence of ALP. Adapted from Ref. 318. Copyright © 2018 by Royal Chemical Society. (C) The structures of 30 and 31. (D) A typical procedure for preparation of the hydrogels via the enzymatic solid-gel transition: 30 dissolves in 7.4 tris-HCl buffer at initial concentration of 0.6 wt% (8.29 mM), the concentration of saturated Ca₃(PO₄)₂ is around 3.87 μM, and [30]₀:[Ca²⁺ or Sr²⁺]₀ is 1:2. Adapted from Ref. 319. Copyright © 2015 by Wiley Inc.

The substrates for ALP-catalyzed ENS also can be other solids (microparticles), as shown by transforming biocomposites of phosphopeptides and calcium (or strontium) ions to supramolecular hydrogels.³¹⁹ Having strong affinity alkaline metal ions (e.g., Ca²⁺ or Sr²⁺), a heterochiral phosphotripeptide (30, Figure 36C) forms precipitates with the addition of calcium chloride. The addition ALP to the suspension of the precipitates results in a hydrogel consisting of the nanofibers of (31) and the microcrystals of calcium phosphates (Figure 36D). As the first example of ENS to assist the dissolution of a solid to form supramolecular nanofibers/hydrogel composites, this approach may find applications in understanding and generating calcified soft tissues.

Based on their pioneering exploration of dipeptides for emulsion,³²⁰ Ulijn et al. used ALP for ENS to form interfacial nanofiber networks for stabilizing emulsions.³²¹ By mixing the Fmoc-_pYL (32, Figure 37A) in water with chloroform and shaking the mixture, they obtained the emulsion, which is relatively weak and phase separates after 1 hour. Shaking after the addition of ALP in the mixture results in the emulsions stabilized by interfacial nanofiber networks of 33. One unique feature of this simple method is that, even after storing the biphasic mixture for weeks, enzyme addition and shaking create the emulsion. It also appears the concentration of 32 should be at a proper range (e.g., 5 mM) to prevent hydrogelation due to the self-assembly of 33. It would be interesting to see how the emulsion behaves when the non-aqueous phase is another organic solvent.

Figure 37.

(A) The structures of 32 and 33. (B) The structures of 34 and 35. (C) In a biphasic system, ALP addition converts a two-phase system with some micelles of 34 at the interface of oil droplets into a more established oil-in-water emulsion by the formation of nanofibers at the interface and surrounding oil droplets, with the emulsifying ability controlled by the specific amount of enzyme used. Adapted from Ref. 322. Copyright © 2017 by American Chemical Society.

Since ALP maintains its activity in heterogeneous conditions,³¹⁹ it also catalyzes the formation of hydrogels in emulsion, as shown by the subsequent study of Ulijn and Tuttle.³²² Without the need of the Fmoc motif, the authors used an unprotected tripeptide, KYF (35), as the self-assembling building block for generating hydrogels by enzymatic dephosphorylation catalyzed by ALP. Particularly, they synthesized the phosphotripeptide, K_pYF (34, Figure 37B), which is a non-gelator. In addition to that ALP converted 34 to 35 to form a hydrogel of KYF (35), the amount of ALP kinetically controls the fiber network formed by KYF (Figure 36C). When the ALP catalyzed self-assembly occurs in biphasic mixtures, nanofibers of 35 form at the oil water interface and throughout the surrounding buffer to stabilize the oil-in-water droplets (Figure 37C). This work illustrates a facile on-demand formation and stabilization of emulsion by ENS, and would probably be very useful for making food or cosmetic products.

3.1.3. Context-Dependent

Differing from self-assembly at thermodynamic equilibrium, ENS is inherently irreversible and context-dependent. Yang et al. reported an example that ENS and heating–cooling processes result in different supramolecular assemblies.³²³ They synthesized Ada-Gff_py (36, Figure 38A), which is a substrate of ALP. Without the heating-cooling process, enzymatic dephosphorylation of 36 to generated the nanoparticles of Ada-Gffy (37), which is a viscous solution. Subjecting the solution to a heating–cooling cycle results in a hydrogel consisting of nanofibers and nanoparticles. Using the nanoparticles and the nanofibers as a protein vaccine adjuvant, the authors found that the nanoparticles resulted in accelerated DC maturation and stronger T-cell cytokine production than the nanofibers. The authors also suggested that ENS gave higher molecular order and conserved the more ordered conformation of the peptides. Although the detailed mechanism remains to be elucidated, this result indicates an entropy driven self-assembly by a heat-cooling process. Moreover, their work demonstrated that ENS is a useful method for preparing supramolecular nanomaterials for enhanced bioactivity. In another related study, Yang et al., reported that ALP catalyzed ENS at different temperatures generated peptide assemblies that had different morphology and exhibited different emergent properties, such as gel or sol.³²⁴ The authors developed a fluorescent phosphopeptide (38, Figure 38B) that, underwent ENS catalyzed by ALP. When 38 is incubated with 0.1 U/mL ALP at 37 °C (or 1 U/mL ALP at 4 °C), the relatively slow rate of dephosphorylation allows the resultant peptide (39) to distribute evenly and self-assemble to form a hydrogel, consisting of nanofibers of 39. When 38 is incubated with 1 U/mL ALP at 37 °C, fast dephosphorylation of 38 creates high local concentration of 39 near the enzymes and results in a solution of nanoparticles (Figure 38C). This work highlights the spatiotemporal control offered by ENS in the context of different temperature and different enzyme concentrations. Interestingly, the authors reported that 3T3 cells uptake about ten times more of the nanofibers than the nanoparticles. It would be useful to elucidate the underlying mechanism, especially by using other types of cells for the cell uptake study.

Figure 38.

(A) The structures of Ada-Gff_py (36) and Ada-Gffy (37). (B) The structures of NBDgff_py (38) and NBDgffy (39). (C) ENS spatiotemporally controlled by temperature and the concentration of ALP. Adapted from Ref. 324. Copyright © 2017 by American Chemical Society. (D) The structures of Nap-YYY (40), Nap-_pYYY (41), Nap-Y_pYY (42), and Nap-YY_pY (43). (E) Schematic illustration of different self-assemble behaviors of three precursors. Adapted from Ref. 325. Copyright © 2018 by Royal Chemical Society.

Besides their works to show that ENS results in different morphologies of the assemblies (of the same peptide) at different temperature, Yang et al. demonstrated that different precursors of the same hydrogelator formed different assemblies after undergoing ENS.³²⁵ They synthesized three phosphorylated peptides (Nap-_pYYY (40), Nap-Y_pYY (41), Nap-YY_pY (42)), which were the precursors of the same peptide (Nap-YYY (43)) (Figure 38D). Upon the addition of ALP, 41 is dephosphorylated to form uniform nanofibers in a stable hydrogel. The dephosphorylation of the other two precursors (40 and 42) by ALP, however, produces metastable hydrogels containing mixed nanostructures. The authors used TEM to show that the preorganization of 41 likely contributes to the formation of the stable hydrogel of 43 (Figure 38E). Another notable feature is that the nanofibers of 43, being made from 42, exhibit quite extensive interfibrillar interactions to form ribbon-like and higher-order structures. This work indicates that the assemblies of the precursors, acting as context-dependent signals, provide a useful strategy to control the outcomes of ENS, such as cellular uptake.³²⁶ It also highlights that the molecular space of the precursors is a key factor for controlling ENS.

Recently, Yang et al. reported that the self-assembly catalyzed by ALP can make a micelle solution that turns into a hydrogel up increasing temperature, as a thermogel.³²⁷ They synthesized two relatively long phosphopeptides (Figure 39A) with the structures of Nap-FFGG_pYGSSSRRAPQT (44) and NBD-GFF_pYGAVPIAQK (46). Adding ALP to dephosphorylate 44 or 46 at 4°C results in 45 or 47, respectively. 45 and 47 both self-assemble to form micelles or short fibers at 4°C, but remain as solutions (Figure 39B). At a physiological temperature of 37°C, the solution of 45 or 47 undergoes sol-to-gel transformations, likely being driven by the increase of entropy. According to the authors, this thermogel may find promising applications such as three-dimensional (3D) cell culture. This work also highlights the opportunity to control the properties of the assemblies of oligopeptides via ENS.

Figure 39.

(A) The structures of Nap-FFGG_pYGSSSRRAPQT (44) and NBD-GFF_pYGAVPIAQK (46). (B) Optical images of the solutions of 45 and 47 formed by ALP catalyzed ENS at 4 °C, and the corresponding hydrogels formed raising the temperature 37 °C. Adapted from Ref.³²⁷. Copyright © 2019 by Royal Chemical Society.

From the discussion in section 2, the most common context-dependent signals used by cells for ENS are ligand-receptor interactions, as evidenced by ENS and ligand–receptor interactions of proteins often occuring together, for example in signalosomes. To understand how ligand–receptor interaction modulates the ENS of small molecules, it would be useful to develop ENS for generating supramolecular assemblies in the presence of a specific and strong ligand-receptor interaction, as shown in a recent study that uses ligand-receptor interaction to modulate the kinetics of ENS (Figure 40).³²⁸ In that study, the ligand-receptor pair is vancomycin (48) and D-Ala-D-Ala, which serves as the basis for vancomycin against Gram-positive bacteria.^329–330 The attachment of D-Ala-D-Ala to the C-terminal of Nap-FFYGG produces a peptide (50) that self-assembles to form nanoribbons. Replacing tyrosine in 50 by phosphotyrosine generates a substrate (Nap-FF_pYGGaa, 49, Figure 40A) of ALP for ENS. Without interacting with 48, enzymatic dephosphorylation of 49 makes 50, which is a hydrogelator and self-assembles to form long nanofibers of 50 (Figure 40B). In the presence of 48 (i.e., the ligand–receptor interaction), enzymatic dephosphorylation of 49 leads to precipitous aggregates containing short nanofibers that are made of 48 and 50 (Figure 40B). An excess amount of 48 can disrupt the nanofibers of 50 and result in the precipitates (of 48 and 50). Directly adding 50 in water results in the nanoribbons of 50, which hardly interact with 48. This result also confirms that the assemblies of 50 exhibit different properties from that of individual molecules of 50. This work represents the first case of using ligand-receptor interactions to control ENS of small molecules. Further exploration of this direction may lead to useful applications of ENS, especially when this type of modulation integrates with cellular functions.^331–332

Figure 40.

(A) The structures of vancomycin (48) and the peptides, 49 and 50. (B) The illustration of ligand–receptor interaction of small molecules dictating the pathways of ENS. Adapted from³²⁸. Copyright © 2016 by American Chemical Society. (C) Schematic representation of sequential (blue and green arrows, Pathways I and II) and competing (brown arrow, Pathway III) biocatalytic pathways from precursors 1 and F-NH₂ for generating assemblies with different morphologies. Adapted from Ref.³³⁶. Copyright © 2017 by Wiley Inc.

Nature also uses the ENS process, as context-dependent signals, to control another ENS process, such as the control of the dynamics of actins. Dynamic self-assembly, indeed, is an emerging concept in supramolecular chemistry.^333–335 Ulijn et al. pioneered the combination of dynamic self-assembly and enzymatic reaction. Taking advantage of the dephosphorylation of 1 by ALP and the amide bond formation catalyzed by thermolysin, Ulijn et al. reported an intriguing self-assembly cascade that utilized those two enzymatic reactions for generating transient (or interconverting) nanostructures.³³⁶ As shown in Figure 40C, ALP dephosphorylates 1 to generate nanofibers of 2; thermolysin catalyzes the (reversible) amide hydrolysis and condensation of 2 and C-terminal capped phenylalanine (F-NH₂). The authors examined the morphology of the resulting assemblies by sequential addition or co-addition of the enzymes. In pathway I, ALP converts the spheres of 1 to the fibers of 2, the subsequent addition of thermolysin and F-NH₂ generates Fmoc-YF-NH₂ (52) which self-assembles to form nanosheets. In pathway II, thermolysin catalyzes the condensation between Fmoc-_pY and F-NH₂ to produce Fmoc-_pYF-NH₂ (51), which self-assembles to form nanotubes/tapes. The subsequent addition of ALP removes phosphate group from tyrosine phosphate to result in the nanosheets of 52. By varying the ratios of the enzymes and co-addition of the enzymes in the mixtures of 1 and F-NH₂ (Pathway III), it is possible to control the morphological transition of the mixtures to make the system evolve more like pathway I or II. This work is an excellent example of how to use multiple ENS processes to control self-assembly, and provides important insights for designing soft nanostructures that exhibit tunable dynamic properties. Recently, Zhang Y. reported an elegant example of a Dual-Enzyme-Instructed Hierarchical Assembly.³³⁷ The next step, besides more rigorous kinetic analysis, is how to translate these concepts and insights into useful applications.

3.1.4. On A Surface

The fast kinetics of ALP catalyzed dephosphorylation allows rapid buildup of the concentration of the products. When the product is a building block for self-assembly, such a rapid ENS process enables localization of the assemblies at the position of the enzymes.²⁶⁷ One way to help increase the local concentrations of the products is to provide a binding site for the products, as shown by the elegant work of Schaaf and Boulmedais.³³⁸ They generated a seed layer composed of poly(acrylic acid) that carried Fmoc-FFC as the pendants. After confirming that PAA-CFF-Fmoc (53) interacted with Fmoc-FF_pY (54) to promote ALP catalyzed enzymatic hydrogelation, they embedded ALP in a polyelectrolyte multilayer, which had 53 as the top layer (Figure 41A). With the top layer acting as the nucleation site for the self-assembly of the Fmoc-FFY peptide (55), a nanofiber network starts to form almost instantaneously when the multilayers come in contact with the solution of 54. The network is able to extend up to several micrometers into the solution after several hours. The authors also demonstrated that changing the peptide density allowed the active seed layer to control the self-assembly kinetics and the geometric features of the fiber network. This model system may provide a useful insight for understanding ENS for spatiotemporal control of molecular assemblies in a more complicated microenvironment, such as cells.²⁶⁷

Figure 41.

(A) The structures of polyelectrolytes (53) and the peptides (54 and 55) and the illustration of the surface-localized ENS. Adapted from Ref.³³⁸. Copyright © 2015 by Wiley Inc. (B) ALP on silica nanoparticles dephosphorylates 54 to form 55, leading to hydrogelation, and the nanofibers of 55 (in blue) self-assembled from NP@ALP by dephosphorylation of 54 leading to a hybrid supramolecular hydrogel undergoing a phase separation over time. (C) Optical images of the gel of 55 before and after phase separation. Adapted from Ref.³³⁹. Copyright © 2019 by American Chemical Society. (D) The structure of Fmoc-GFF_pYGH_pY (56) and its dephosphorylated products. (E) The structures of NBD-FF_pY (60) and NBD-FFY (61).

Instead of attaching phosphopeptides to nanoparticles, Schaaf and Jierry also covalently linked ALP to nanoparticles to catalyze self-assembly in a spatiotemporally controlled manner.³³⁹ In that work, the substrate is 54, and the enzyme is the ALP linked on silica nanoparticles (NPs). As shown in Figure 41B, after the dephosphorylation, the self-assembled peptide nanofibers grow from the NPs, which are able to form a hydrogel that consists of the silica nanoparticles and the nanofibers of 55. The authors reported an unexpected macroscopic shape change of the hydrogel over time, resulted from a phase separation leading to a dense phase (in NPs and nanofibers) in the center of the vial and surrounded by a dilute one, which still contained NPs and peptide self-assemblies (Figure 41C). Although it is similar to a syneresis process, this change only occurs when the ALP is being attached to the NPs. Thus, the authors speculated that the strong interactions of self-assembled peptide nanofibers with the enzymes on the NPs resulted in the dense phase. Since the interaction between the nanofibers and ALP is less explored,²³³ this work may reveal a useful insight for further investigation of the allosteric interaction of the peptides with the ALP. In another related study, Schaaf, Jierry, and Boulmedais made a polymeric hydrogel that incorporated ALP for ENS inside the hydrogel.³⁴⁰ Because of its small size, 54 easily diffuses into the polymeric hydrogel and allows the dephosphorylation of 54. This approach is able to tune the elastic moduli of the hydrogels, but only moderately. Recently, the same group of authors also reported the use of glutathione reductase³⁴¹ for controlling the self-assembly of Fmoc-peptides for gelation that was initiated exclusively from the surface.

Schaaf and Jierry also reported an exciting use of ENS on a surface to generate self-assembled peptide nanofibers for catalysis in the context of flow chemistry.³⁴² To avoid the mechanical fragility of peptide nanofibers in the hydrogels, the authors immobilized ALP on a multilayer film based on a commercial open-cell melamine foam. After packing the foam into a column and passing the solution of Fmoc-GFF_pYGH_pY (56, Figure 41D)³⁴³ through the functionalized column, the hydrogel consisting of the nanofibers of Fmoc-GFFYGHY (59) forms in the pores of the foam. The authors demonstrated that this supramolecular hydrogel was able to catalyze the hydrolysis of an activated ester, 4-nitrophenyl acetate, and the kinetic resolution of racemates. Most impressively, being reusable and stable over months, the resulting hybrid material is robust enough to be a component in continuous flow reactors. This innovative work illustrates ENS as a facile approach to control the emergent functionalities of supramolecular assemblies.

It is difficult to detect enzyme-induced color change in colored medium, such as blood. Yang et al. reported an facile way to detect the activity of ALP in blood via ENS.³⁴⁴ They designed and synthesized NBD-FF_pY (60, Figure 41E) that was able to be converted to NBD-FFY (61) upon dephosphorylation catalyzed by ALP. Taking the advantage of a positively charged surface, they allowed the negatively charged 61, being generated by ALP (in serum), to be attracted to the surface modified with positive charges. This process results in nanofibers/hydrogels on the surface. After removing the serum, the fluorescence turned on at the glass surface reveals activity of the enzyme (ALP). According to the authors, the color depth of hydrogels at the surface is visible even by naked eyes, thus providing a convenient assay for detecting enzymes.

3.1.5. Coupling with Other Controlling Mechanisms

The enzymatic control of the dynamics of self-assembly also is suitable to couple with other controlling mechanisms, such as photochemistry, as shown by Zhang et al.³⁴⁵ Inspired by the dynamic and reciprocal interactions between ECM and cells, the authors developed two building blocks that contained coumarin and acted as self-assembling hydrogelators (Figure 42A). While the peptide derivative (63) self-assembles into nanofibers, the benzoate molecule (64) forms nanobelts. These two molecules exhibit different dynamic profiles and form self-sorted nanostructures. One interesting feature of this two-component mixture is that the growth of the nanofibers of 63 enables the nanobelts of 64 to turn into a layer-by-layer nanosheet (with nine-fold increase in height). This behavior resembles crowding effects to promote protein assemblies. The authors also demonstrated that light and ALP (Figure 42A) were able to control the height and morphology of the self-sorted nanostructures. This kind of dual control appears to have interesting potential applications, but it remains to be explored.

Figure 42.

(A) The structures of the gelators (63 and 64) and the corresponding precursor (62) and photochemical products (65, 66, and 67) and their enzymatic or photochemical conversion. (B) The structures of Nap-gff_py-CHO (68) and Nap-gffy-CHO (69). (C) The illustration of ALP-triggered simultaneous hydrogelation of 55 and chemiluminescence of 72 (D) CL kinetic curve and time sweep of 12.9 mM 54 and 1.25 mM 70 incubated with 2 μM ALP at 25 °C. Insets: time-course CL images of 54 and 70 incubated with ALP at 25 °C. Adapted from Ref.³⁴⁷. Copyright © 2017 by American Chemical Society.

In a different study, Yang and Wang et al. reported the first combination of amine-reactive aldehyde with ALP catalyzed self-assembly.³⁴⁶ As shown in Figure 40B, the phosphopeptide (Nap-gff_py-CHO, 68), in addition to a phosphotyrosine residue, bears an aldehyde group at the C-terminal. 68, despite bearing an aldehyde group, still can be dephosphorylated by ALP. The authors reported that 68, forming a clear solution in PBS buffer, turned into a hydrogel of Nap-gffy-CHO (69) after ALP dephosphorylated 85% of 68 in the solution. Because the drug containing an amino group can form pH-responsive Schiff bases with 69, the hydrogels are useful for encapsulating amine drugs, such as doxorubicin, for controlled release.

To reduce the autofluorescence in biological samples, Liang et al. came up with an original approach that combined ENS and chemiluminescence (CL) catalyzed by ALP.³⁴⁷ The authors used 54 for the ALP catalyzed self-assembly of Fmoc-FFY (55). Using a commercial chemiluminescence agent AMPPD (70, Figure 42C) to incubate with 54 in the presence of ALP, the authors were able to employ CL for characterizing and imaging the simultaneous ENS process of 54. They also found that 54 and 70 competed for ALP, thus 54 delayed the peak of CL and 70 slightly slowed down the gelation (Figure 42D). The further applications of this simple mixture, however, remains to be demonstrated.

3.1.6. Acid Phosphatases

Only a few studies^257,348 have used acid phosphatases (ACP) for ENS after the early example of screening the inhibitors of ACPs based on hydrogelation resulted from ENS.²²⁴ Borner et al. used ACP to dephosphorylate polymer-peptide conjugates for self-assembly.³⁴⁸ As shown in Figure 43, the polymer–peptide conjugates consist of five repeats of alternating threonine and valine diads ((TV)₅) at the N-terminal and a polyethylene glycol (PEO) block at the C-terminal. The (TV)₅ repeats favor the formation of β-sheets in water. Containing three phosphothreonine (_pT) residues, the phosphorylate conjugates (73) dissolve well in acidic water to form a solution, and the phosphorylated peptide segments adopt a coil conformation. The addition of ACP dephosphorylates the phosphothreonine residues on the conjugates and leads to rapid conformation change of the peptides from the coil conformation to β-sheet. However, only after a relatively long time (7 days) incubation with ACP, did the resulting dephosphorylated conjugates self-assemble to form uniform nanofibers. This work illustrates that the concept of ENS is applicable to large synthetic molecules. It remains to be seen if the peptide segment can be phosphorylated by a kinase. If this is true, a more dynamic control of the polymer-peptide conjugates may be achieved.

Figure 43.

(A) The structures of the polymer substrate (73) of acid phosphatase and the dephosphorylated product (74). (B). Microstructures formed by ACP triggered self-assembly of 74 visualized by TEM. Adapted from Ref.³⁴⁸. Copyright © 2009 by Wiley Inc.

3.1.7. Protein Phosphatases

Enzymatic phosphorylation/dephosphorylation is a putative mechanism used by cells to control the conformation of proteins and to modulate the protein functions. Thus, it is reasonable and logical to use enzymatic reaction for modulating the conformation of peptides for self-assembly, as shown by Schneider et al.³⁴⁹ Based on their seminal work on the self-assembly of peptide hairpins,^350–352 the authors designed and synthesized the peptides by having a phosphothreonine at different positions of the hairpins (Figure 44A). The phosphopeptides disfavor the folded and assembled state, thus largely reside in unstructured states (Figure 44B). The addition of lambda protein phosphatase (LPP) almost quantitatively dephosphorylates the peptide (75) within hours. Such dephosphorylation leads to a distinct conformational change from random coil to β-sheet, which assembles into well-defined monomorphic fibrils (~4 nm). The position of the phosphothreonine hardly affects the dephosphorylation kinetics, but modulate the dynamics of self-assembly. The authors also investigated the cytocompatibility of a hydrogel resulting from dephosphorylation of the phosphopeptides (75 & 77) and found that human dermal fibroblasts (HDFs) remain viable on top of the resulting hydrogels. One notable result reported in this work is that 75 appears to be a more suitable substrate to the protein phosphatase than to ALP. Because of the abundance and specificity of protein phosphatases in cells, it is likely that carefully designed ENS substrates of protein phosphatases would lead to more useful applications.

Figure 44.

(A) The sequences of the phosphopeptides and the corresponding peptides. (B) Enzymatic dephosphorylation shifts the conformational equilibrium favoring folded hairpin in the self-assembled state. (B) Enzymatic dephosphorylation shifts the conformational equilibrium to favor folded hairpin in the self-assembled state. (C) The TEM of the nanofibers formed by the self-assembly of 76. Adapted from Ref.³⁴⁹. Copyright © 2018 by Wiley Inc.

3.1.8. Coupling with A Disassembly Process

It is well-established that cells use a pair of enzymes that have counteracting activities to switch the functions of proteins.³⁵³ Such an enzyme-switch is a common feature that associates with ENS, which would result in phase-transition (as shown Figure 32B²¹⁸). In fact, the study of ENS of small molecules has confirmed that it is feasible to use a kinase/phosphatase switch for regulating phase transition.²²⁶ As shown in Figure 45A&B, a hydrogelator Nap-FFGEY (82) forms a gel. The addition of a kinase and ATP to the gel of 82 allows the phosphorylation of 82 to generate Nap-FFGE_pY (81). Because 81 is more hydrophilic than 82, the kinase turns the gel of 82 to the solution of 81. Adding ALP to the resulting solution leads to sol-gel transition because ALP converts 81 back to 82 (Figure 45B). TEM reveals that ALP catalyzed ENS also generates nanofibers with monodispersed diameters (Figure 45C), which are more-ordered than the nanofibers in the gel of 82 formed by changing pH. In vivo study (i.e., subcutaneous injection of 81 in mice) confirms that ENS is able to generate the gel of 82 in vivo.²²⁶ Stupp et al., using a peptide amphiphile (84, Figure 45D) as the substrate of protein kinase A and alkaline phosphatase, also demonstrated the enzymatic switch of the gel of 84.³⁵⁴

Figure 45.

(A) The structures of 81 and 82, which are the substrates of phosphatase and kinase, respectively. (B) The kinase/phosphatase enzyme switch controls phase transition. (C) TEM shows the nanofibers in the hydrogel formed by ENS is more ordered (Right). Adapted from Ref. 226. Copyright © 2006 by American Chemical Society. (D) The structures of a pair of peptide amphiphiles (83/84) that is able to undergo phase transition controlled by protein kinase A (PKA)/ALP. (E) The structures of hydrogelator 86 and the phosphorylate precursor (85). (F) The illustration of the ¹⁹F-fluorinated hydrogelator that undergoes ALP-instructed self-assembly and EGFR-guided disassembly. Hydrogel leads to the “OFF” signal and solution results in the “ON” signal of ¹⁹F NMR/MRI. Adapted from Ref.³⁵⁵. Copyright © 2006 by American Chemical Society. (G) The structures of the molecules, 87, 88, 89, which that exhibit OFF mode and ON mode of NMR by enzymatic reactions.²⁷⁴

Recently, to detect the presence of tyrosine kinase, Liang et al. combined ALP catalyzed self-assembly with disassembly catalyzed by kinase, as shown in Figure 45E.³⁵⁵ The authors synthesized a ¹⁹F-fluorinated hydrogelator (86), as well as its phosphorylated precursor (85). ALP catalyzed dephosphorylation generates the hydrogel of 86, which is silent in F-19 NMR (Figure 45E) because 86 is unable to rotate freely in the nanofibers of 86 in the hydrogel. According to the authors, treating the hydrogel of 86 with tyrosine kinase generates 85 and results in the disassembly of the nanofibers. As the soluble molecule, 85 is F-19 NMR active (Figure 45E), thus the disassembly constitutes a “turn-on” process of the F-19 based MRI. The authors also used the cell lysates of H1975 cells (EGFR overexpression) and LoVo cells (ALP overexpression) to demonstrate the ON/OFF MRI signals. This work, being the first case of using ¹⁹F NMR/MRI to sense the activity of tyrosine kinase and phosphatase, underscores the importance of incorporating ENS into a dynamic process. In related studies,^356–357 Liang et al. also reported the use of liquid-phase scanning tunneling microscopy (L-STM) to directly visualize enzymatic self-assembly/disassembly of nanofibers of an analog, in which F replaces the CF₃ group in 85.

To develop an enzyme probe for reporting tyrosinase, Yang et al. combined ALP catalyzed assembly and tyrosinase catalyzed disassembly, as shown in Figure 45G.²⁷⁴ The authors developed a phosphopeptide (87) containing a fluorinate phenylalanine residue and an adamantane at the N-terminal. After ALP removes the phosphates, the resulting peptide (88) self-assembles to form a nanoparticle suspension, which is silent on F-19 NMR due to the aggregation of 88. The addition of tyrosinase catalyzes the oxidation of 88 to form 89, causing the disassembly of the nanoparticles, thus turning on the F-19 NMR signals. Although the amount of tyrosinases, used in this demonstration, is relatively high, the principle illustrated in this work should be applicable for detecting other enzymes that have high activity and result in disassembly. The above results illustrate ENS for phase transition and promise a versatile way to design and to construct biomaterials for various applications (See Section 4).

3.2. Proteases

In addition to the use of enzymatic dephosphorylation for controlling self-assembly of molecules, enzymatic proteolysis is another type of enzymatic reaction being used to regulate the self-assembly of various substrates. Unlike ALP, protease catalyzed ENS of synthetic molecules has received less exploration, so the following discussion adheres to more of a chronological order than thematic arrangement. Because proteases are evolved for breaking bonds in large molecules, such as proteins,³⁵⁸ it is rather common to cleave peptide segments from a polymer for enabling self-assembly of nanoparticles, as pioneered by Bhatia et al.³⁵⁹ As shown in Figure 46A, the authors generated superparamagnetic Fe₃O₄ nanoparticles coated with biotin or neutravidin that attach to polyethylene glycol (PEG) polymers (10 kDa) via a GPLGVRGC peptide segment. Because GPLGVRGC is a substrate of matrix metalloproteinase-2 (MMP-2), MMP-2 can proteolytically remove the PEG to allow the binding of biotin and neutravidin, thus initiating the self-assembly of the Fe₃O₄ nanoparticles. The authors reported that ten to hundreds of nanoparticles assemble together 3 h after MMP-2 addition. Using MRI, the authors also showed that MMP-2 initiated assembly to amplify the transverse (T2) relaxation of the nanoparticles and suggested that these MMP-2 responsive magnetic nanoparticles may allow MRI detection of tumors. This approach has been applied to the design of polymer nanoparticles for drug delivery that targets cancer cells.³⁶⁰

Figure 46.

(A) The illustration of a proteolytic actuation of self-assembly of neutravidin- and biotin-functionalized superparamagnetic iron oxide nanoparticles by MMP-2 cleaving peptide substrates (GPLGVRGC). Adapted from Ref.³⁵⁹. Copyright © 2006 by Wiley Inc. (B) The structure of 90, a substrate of MMP-9, and the enzymatic hydrolysis process. (C) Optical images of the sol-gel transition and the TEM images of the nanofibers in the gel. Adapted from Ref. 238. Copyright © 2009 by Royal Chemical Society. (D) The structures of 93 and 94. (E) Schematic depicting the enzyme induced stiffening of nanoparticle-hydrogel composites with a structural color. Adapted from Ref.³⁶². Copyright © 2015 by American Chemical Society.

One of the known drawbacks of nanoparticles for imaging is the accumulation of nanoparticles by mononuclear phagocytic and renal systems.³⁶¹ One approach to address this problem is to form the nanoparticles in situ on the tumors by ENS of small molecules. Indeed, it is feasible to use proteases to instruct the self-assembly of small molecules.²³⁸ Because the tetraphenylalanine (FFFF) is an efficient supramolecular hydrogelator with the minimum gelation concentration (mgc) at about 0.4 wt% in water at pH7.4, it is feasible to create a soluble precursor (90) by adding the pentapeptide, CGLDD, at the C-terminal of the tetraphenylalanine. As a substrate of MMP-9, FFFFCGLDD (90), is able to undergo proteolysis catalyzed by MMP9 (Figure 46B). At pH > 6.6, 90 (0.6 wt%) exists as a clear solution. The addition of MMP-9 to the aqueous solution of 90 at room temperature results in the sol-gel transition after 45 minutes. Although only 47% of 90 is converted to 91, TEM shows the nanofibers in the hydrogel (Figure 46C), confirming that MMP-9 catalyzes the proteolysis to initiate the self-assembly of the peptide mixture, which consists of 90, 91, and 92.

To achieving passive monitoring of biological environments by soft materials, Kofinas et al. developed color changing close-packed nanoparticles—hydrogel composites.³⁶² They suspended silica nanoparticles in the solution of 4-arm norbornene-PEG (PEG4NB (93), Figure 46D) and a dicysteine capped peptide (94) for photo crosslinking to make the nanoparticle—hydrogel composites. Proteases, either chymotrypsin or trypsin, catalytically cleave the peptides in hydrogel network, thus initiating self-assembly to form a secondary, physically cross-linked network. This change causes a 1200% increase in storage modulus. The increased packing of the silica particles also results in color change of the composites (Figure 46E). This result is rather counterintuitive because proteolytic cleavage usually results in dissociation of molecular aggregates.³⁶³ The authors suggested that such protease responsive hydrogel nanocomposites may find biomedical applications as degradation-stiffening and collapsing materials.

Based on the work of Ulijn et al.,²²⁵ Huang and Qi investigated the use of α-chymotrypsin for enzymatic hydrogelation.³⁶⁴ In that study, the substrates are Fmoc-F, Fmoc-W, Fmoc-A, Fmoc-S, F-OMe, F-OEt, and Y-OMe. Rapid hydrogelation occurs when the substrates are Fmoc-F and F-OMe (or Y-OMe) and α-chymotrypsin is the enzyme. Without the addition of the enzyme, the substrate exhibits poor solubility and exists as a milky suspension. The authors also suggested the interactions between the substrates and the enzyme promote the self-assembly of the substrates and the product of the enzymatic reaction. This assumption appears to be reasonable since the amount of α-chymotrypsin or trypsin, used in that study, is relatively high (1 mg/mL).³⁶⁴ Ulijn et. al., based on the reverse hydrolysis catalyzed by thermolysin, recently used reversible and irreversibly immobilized enzymes on a surface for ENS of peptide assemblies.³⁶⁵ The authors used bioinspired polydopamine and polyphenol coatings to study the effects of surface anchoring and surface release of enzymes on the self-assembly process. Their studies reveal that the enzymes reversibly bound to the surface catalyze the coupling of the precursors Fmoc-T (95, Figure 47A) and F-NH₂ (96) to result in a bulk gel made of 97. The enzymes, which are irreversibly bound to the surface, catalyze the conversion of the precursors only at the surface, thus restricting self-assembly of the gelator 97 to form nanofibers/hydrogel at the surface (Figure 47B). The work may lead to a useful model system to understand pericellular ENS catalyzed by membrane-anchored enzymes and secreted enzymes, especially when the enzymes are proteases.

Figure 47.

(A) Conversion of the 95 and 96 to the gelator 97, catalyzed by thermolysin. (B) Reversible and irreversible enzyme immobilization on modified surfaces for biocatalytic self-assembly. Adapted from Ref.³⁶⁵. Copyright © 2017 by American Chemical Society. (C) The structure of the branched peptide (98) that is a substrate of ENTK. (D) Optical image of the solution of 98 (2.5 wt%, PBS, pH 7.4) and the hydrogel of 99 formed after adding ENTK to the solution for 24 h and the schematic illustration of ENTK-catalyzed transformation of micelles to nanofibers. Adapted from Ref.³⁶⁸. Copyright © 2017 by American Chemical Society.

While MMPs have received much attention in ENS, transmembrane serine proteases (TMPRSS)³⁶⁶ only are being explored for enzymatic self-assembly recently.^367–368 Enterokinase (ENTK/TMPRSS15) is an enteropeptidase discovered by Pavlov.³⁶⁹ ENTK specifically cleaves the peptide sequence DDDDK from proteins and has led to the development of FLAG-tag (DYKDDDK) for protein purification.³⁷⁰ As shown in Figure 47C, attaching the FLAG-tag as a peptidic branch to a self-assembling motif³⁷¹ affords a negatively charged, soluble, branched peptide (98). 98 self-assembles to form micelles. ENTK catalyzes the cleavage of the hydrophilic FLAG branch to generate a peptide (99) that forms nanofibers, which results in supramolecular hydrogels. This ENTK-based ENS induces both sol-gel and morphological (micelles-to-nanofibers) transitions (Figure 47D). This work is also the first case of using a protease to cut branched peptides for generating supramolecular hydrogels, and opens a new way explore ENS of branched peptides.

To develop oligopeptide hydrogels for localized protein delivery, He and Mo used substrate-selective proteases to catalyze ENS of 103 (Figure 48A) for encapsulating proteins by an innovative design,³⁷² as shown in Figure 48B. In that study, the authors first used polymeric networks to encapsulate a protease (WQ9–2) to prevent the proteolysis of protein cargos, then they mixed the proteins and peptide substrates (101 and 102) of the proteases. Because of the small size of the peptides, it enters the polymeric networks and acts as the substrates of the protease for reverse hydrolysis. This process leads to the buildup of the peptide hydrogelators and results in a supramolecular hydrogel for loading the proteins. The authors used the resulting oligopeptide hydrogels for delivering of an antiangiogenic protein, hirudin, and an apoptosis-inducing cytokine, TRAIL, in a mice model, and observed enhanced synergistic antitumor effects both in vitro and in vivo.³⁷² Although proteins would be the natural substrates for proteases, the combination of protein and protease is much less explored for self-assembly. Recently, Li et al. reported an interesting example of this approach.³⁷³ Using bacillus licheniformis protease (BLP) to catalyze the partial hydrolysis of α-lactalbumin, the authors generated amphiphilic α-lactalbumin peptides that self-assembled to form peptosomes. The authors used the peptosomes to encapsulate a therapeutic peptide and curcumin for drug delivery. According to the authors, this system exhibits enhanced antitumor and antimetastatic activities in vitro and in vivo. This work bears similarity to earlier work on anticancer α-lactalbumin done by Svanborg.³⁷⁴ Further study is needed for identifying the molecular species formed upon partial hydrolysis.

Figure 48.

(A) Reverse hydrolysis that generates 103 from 101 and 102. (B) The illustration of substrate-selective protease-catalyzed self-assembly of the oligopeptide hydrogel for efficient encapsulation of hirudin and TRAIL (Hirudin/TRAIL Gel). Adapted from Ref.³⁷². Copyright © 2017 by American Chemical Society. (C) Enzymatic conversion of the lysine-leucine ehyl ester dipeptides, KLOEt (104), to (KL)_nOEt oligomers (105). (D) Schematic representation of the ENS of 105 formed near the surface adsorbed an α-chymotrypsin layer. Adapted from Ref.³⁷⁵. Copyright © 2017 by American Chemical Society.

Based on their innovative approach of surface immobilized ALP for supramolecular hydrogelation, Schaaf and Boulmedais used proteases for localized enzyme-assisted self-assembly (LEASA).³⁷⁵ As shown in Figure 48C, the enzyme is α-chymotrypsin and the substrate is KL-OEt (104, K: lysine; L: leucine; OEt ethyl ester). After the absorption of α-chymotrypsin and bovine serum albumin (BSA) on a surface made of poly(ethylene imine) and tannic acid, the authors allowed the surface to contact the solution of 104 so that α-chymotrypsin catalytically converted 104 to (KL)_nOEt oligopeptides (105). After reaching a critical concentration near the surface, the formed oligopeptides self-assembled into β-sheets resulting in a fibrillar network localized at the interface. The authors were able to control the lag time (i.e., the time before starting the self-assembly process) by finely tuning the surface density in α-chymotrypsin and the concentration of 104. This model system would be extremely useful if it can mimic the ENS processes that occur on a cell surface. In a related study, Bai et al. devised an interesting application of ENS for making peptide cages.³⁷⁶ Specifically, the authors absorbed thermolysin on porous CaCO₃ microparticles to catalyze the condensation reaction of Fmoc-Y (2) and L-NH₂ in a water/hexane emulsion. This biphasic system limits the self-assembly of Fmoc-YL-NH₂ on CaCO₃ surface. After the removal of the CaCO₃, the authors were able to obtain the peptide cages made by Fmoc-YL-NH₂.

One of advantages of using peptides as the pendant groups of block copolymers is to produce nanoparticles that exhibit narrow size distribution, as shown by the work of Gianneschi et al.³⁷⁷ The authors used the fully water-soluble quaternary amine based Hoveyda–Grubbs type catalyst to initiate the ROMP of the L-amino acid–based norbornene dicarboximide, generating a block copolymer that contains GPLGLAGGWGERDGS as the pendants in the peptide stabilizer block. This type of open-to-air aqueous-phase ring-opening metathesis polymerization induced self-assembly (ROMPISA) forms well-defined peptide polymer nanoparticles. Upon the addition of the protease thermolysin, these nanoparticles further rearrange into larger aggregates. These innovative block copolymers seem to be excellent synthetic substrates to generate a dynamic continuum of particles with different morphologies, as shown by the same lab previously.³⁷⁸

3.3. Esterases

Esterases, as an enzyme commonly used for activating prodrugs, also find applications for the ENS of man-made molecules.^227,249 Ragona and Galeotti reported an innovative approach that indirectly applies acetylcholinesterase (AchE) to trigger the self-assembly of a perylene diimide-theorine derivative (PDI-Thr, 106).³⁷⁹ As shown in Figure 49A, B, myristoylcholine chloride (107) acts as a cationic surfactant, which is able to disperse PDI-Thr. AchE catalyzes the hydrolysis of 107 to result in the self-assembly of 106. One attractive feature accompanying this process is the change of the fluorescence of the assemblies with and without the presence of the AchE. Hydrolysis of esters is a facile way to trigger self-assembly. For example, Zhong et al. reported a simple hydrogelator (111, Figure 49C), which is formed by the hydrolysis of an ester (110).³⁸⁰ The authors used sodium carbonate to trigger the hydrogelation for the encapsulation of 10-hydroxy camptothecin (HCPT) to make the HCPT hydrogel. The authors suggested that the HCPT hydrogel possessed satisfactory stability and recoverability for localized injectable drug delivery. 110 also can act as a substrate of esterase for ENS, as reported earlier.²⁴⁹

Figure 49.

(A) The structure of PDI-Thr (106) and AChE catalyzed hydrolysis of 107. (B) Illustration of AchE catalyzed indirect ENS of the assembly of 106. Adapted from Ref.³⁷⁹. Copyright © 2016 by Royal Chemical Society. (C) The structures of 110 and 111.

3.4. Kinases

Although kinases are an important type of enzyme for cell functions, including endogenous ENS, the use of kinase for controlling the self-assembly of man-made molecules is much less explored. A close related recent example, in fact, is the self-assembly of kinase, as reported by Xiang and Yan.³⁸¹ In their study, they selected an adenylate kinase (AKe) as the building block for self-assembly. Bearing dual substrate-recognition domains for binding adenosine triphosphate (ATP) and adenosine monophosphate (AMP), AKe switches its conformations between closed and open states. This switch in conformation traps these two substrates for enzymatic conversion of an ATP and an AMP to two ADP molecules. The authors also produced a protein amphiphile containing one AKe head and a hydrophobic polypeptide tail. While this protein amphiphile self-assembled to form 1D nanofibers, the addition of a specific ligand, diadenosine-5-pentaphosphate (Ap5A), compresses the AKe conformation to generate 2D nanosheets. This elegant example may provide useful insights for the design of other man-made substrates for ENS catalyzed by kinases.

3.5. Beta-lactamases

As a type of bacterial specific enzyme, β-Lactamases hydrolyze the four-membered β-lactam ring in antibiotics and cause the most widespread antimicrobial drug resistance.³⁸² During the development of a facile approach for detecting the presence of β-lactamases and screening their inhibitors, β-lactamase based ENS was explored.²²⁸ As shown in Figure 50, the substrate (112) of β-Lactamases consists of the cephem nucleus as the linker, a hydrophilic group, and a hydrogelator. The substrate, bearing the hydrophilic group, is too soluble to form a hydrogel. Upon the action of a β-lactamase, the β-lactam ring opens to release the hydrogelator (113). Self-assembling to form nanofibers, 113 results in a hydrogel after reaching adequate concentration. Indeed, the lysates of bacteria (e.g., E. coli) containing different types of β-lactamases are able to catalyze hydrogelation. Besides providing a simple, low-cost strategy to identify β-lactam antibiotic resistant pathogens, this work may provide a way to induce intracellular gel-sol transition in cells with controlled-expression of β-lactamases, an application that remains to be realized. In addition, this study also establishes the feasibility of using microorganism specific enzymes for the development of ENS.

Figure 50.

(A) The structure of the substrate of β-lactamase (Bla) and its degradation catalyzed by Bla. (B) The illustration of the ENS catalyzed by Bla. Adapted from Ref. 228. Copyright © 2007 by American Chemical Society.

3.6. DNAzymes

DNAzymes are able to recognize specific DNA sequences and cut at an RNA base site. Ganneschi et al. pioneered the use of DNAzymes to control the self-assembly of DNA-brush copolymers.³⁸³ The authors synthesized the DNA-brush copolymer amphiphiles containing a RNA base (rA) as an enzymatic cleavage site. The DNA-brush copolymer amphiphiles self-assemble into spherical micelles. The addition of a DNA-based phosphodiesterase (DNAzyme), which recognizes a given DNA sequence and cuts at the RNA base, catalytically generates a truncated ssDNA sequence. This enzymatic reaction also leads to a subsequent sphere-to-cylinder phase transition. Adding a 19-base input DNA sequence can trigger a cylinder-to-sphere morphological transition. This work illustrates a combination of ENS and DNA hybridization for dynamically controlling a phase transition.

4. Biomedical Applications of ENS

The previous two sections have shown the ubiquity of ENS processes in biology and the easiness of using enzymatic reactions to generate/control the noncovalent assemblies (or aggregates) of synthetic molecules in supramolecular chemistry. The obvious logical question would be what are the properties and functions of these assemblies formed by ENS. Since nature uses ENS for functions, it is not surprising that it becomes possible to use ENS for designing/controlling functions of the assemblies of synthetic molecules, as shown in the exploration of potential applications of ENS in biomedicine.³⁸⁴ The following subsections focus on the ENS of synthetic molecules in cellular environment. Because the field is rapidly evolving, we divide the subsections, roughly, according the cellular locations of ENS, such as intercellular, pericellular/intracellular, and subcellular. We choose not to differentiate pericellular and intracellular ENS processes because they often are related. We present the results in each subsection in chronological order (more or less) so that it is easier for the readers to follow the advances of the research.

4.1. Intercellular ENS

Though it is much less explored, intercellular ENS likely promises important applications, such as controlling cell morphogenesis. It is well-established that intercellular interactions in a three-dimensional (3D) environment differ profoundly from a monolayer of cells (2D).³⁸⁵ In addition, increasing numbers of reports suggest that 3D cell organization provides better in vitro models that represent the actual in vivo microenvironment^386–391 more accurately than 2D monolayers do. Such insights have led to the development of 3D in vitro models of human cells in the hopes of bridging the discontinuity between in vitro systems and in vivo animal models for drug discovery, stem cell research, cancer cell biology, and tissue engineering and implantation.^392–394 There is renewed interest in generating cell spheroids as the 3D in vitro models of human cells. Although many efforts have focused on generating cell spheroids, the current procedures are still time-consuming (e.g., > 72 h). Intercellular ENS turns out to be a simple and physiologically-relevant way to generate cell spheroids, as shown in Figure 51.³³¹ The ligand-receptor interaction between 48 and 49 results in a complex (48:49), which is also a supramolecular phosphoglycopeptide (sPGP) because 48 is a glycopeptide and 49 a phosphopeptide (Figure 51A). The sPGP further self-assembles in water to form nanoparticles with diameters of 8±2 nm. ALP dephosphorylates the sPGP to turn the nanoparticles of 48:49 into nanofibers of 48:50 with diameters of 8±2 nm (Figure 51B). Being incubated with fibroblast cells and partially dephosphorylated, the sPGP forms dynamic assemblies in situ on the cell surface (Figure 51C). The dynamic assemblies interact with extracellular matrix molecules and effectively abolish contact inhibition of locomotion (CIL) of the fibroblast cells to result in 3D cell spheroids. This work demonstrates that ENS can generate a dynamic continuum of man-made molecules for regulating cell behavior.

Figure 51.

(A) The illustration of sPGP, formed by non-covalent interactions, as a dynamic continuum in the cell milieu. (B) TEM images of 48:49 (300 μM) before (left) and after (right) being treated by a phosphatase (ALP, 1 U/mL, 24h). Inset structures show CPK models of molecular assemblies. Scale bar is 20 nm. (C) The illustration of forming 3D spheroids from a 2D cell sheet upon the addition of 48:49 and the reversibility of the process. (D) Context-dependent dynamic continuum of molecular assemblies for controlling cell fate. By controlling the activity of the ectophosphatase (e.g., ALPL) on cancer cells with a corresponding enzyme inhibitor, the sPGP (48:49) assemblies can result in a 2D cell sheet, 3D cell spheroids, or cell death. Adapted from Refs.^331,332. Copyright © 2017, 2018 by Wiley Inc.

Context-dependent signaling is a ubiquitous phenomenon in nature. Because one of the unique features of ENS is contextual-dependendency, that is, the resulting assemblies from ENS are highly dependent on the path (or the history of molecular transformation) or local microenvironment, ENS of man-made molecules can mimic the context-dependent signaling in cells. This is shown by a recent study³³² on incubating 48:49 with cancer cells (Saos2) or the co-culture of the cancer cells (Saos2) and stromal cells (HS-5). Depending on the amount of the inhibitor for inhibiting the ALPL overexpressed by Saos2 cells, the assemblies of the 48:49 can either result in cell death or induces 3D cell spheroids (Figure 51D). Similarly, adjusting the ratio of the stromal and cancer cells in a co-culture is able to modulate the expression of ectophosphatase so that ENS of 48:50 enables cell spheroids. The spheroids formed in the co-culture also can mimic the tumor microenvironment for drug screening. As the first example of the self-assembly of small molecules for context-dependent signaling, this work underscores the promises of intercellular ENS as multifunctional processes for controlling cell behaviors.

To assist the formation of functional vascular networks in regenerative tissues, Sun and Li employed ENS to develop an injectable hydrogel of a glycopeptide, which served as a scaffold to release deferoxamine (DFO).³⁹⁵ As shown in Figure 52A, the authors synthesized the glycopeptide precursor (115) that consisted of a naphthyl group, a tetrapeptide segment (Phe-Phe-Asp-Tyr(H₂PO₃)), and a sugar moiety (D-glucosamine) on the side chain of Asp. Using ALP to remove the phosphate from the precursor, the authors produced the glycopeptide hydrogel that incorporates DFO. After subcutaneous injection of the DFO-containing hydrogel in mice, the authors reported the hydrogel acting as an effective matrix to trigger the generation of new blood capillaries in vivo (Figure 52B). This work illustrates a novel approach to generate supramolecular assemblies of a glycopeptide to mimic the glycosylated microenvironment of extracellular matrix.

Figure 52.

(A) The structures of a glycopeptide hydrogelator (116) and its precursor (115). (B) ENS of 116 to for a hydrogel, which encapsulates DFO for inducing angiogenesis in vivo. Adapted from Ref.³⁹⁵. Copyright © 2018 by American Chemical Society.

The intercellular ENS process seems particularly useful for mimicking the protein dynamics in ECM for inducing cell spheroids. A single component molecule precursor (117), which differs from the sPGP (48:49), is able to act as a substrate of ENS to enable 2D to 3D cell morphogenesis.³⁹⁶ As shown in Figure 53A, the biotinylated D-phosphotetrapeptide (117) is a substrate of phosphatases. Biotin serves as a cell compatible ligand of cell surface receptors.³⁹⁷ Before dephosphorylation, 117 self-assembles to form nanoparticles, which turn into nanofibers upon dephosphorylation catalyzed by phosphatases to generate 118. After incubating with HS-5 cells, some of the molecules of 117 become dephosphorylated and result in nanofibers consisting of the mixture of 117 and 118 at the intercellular space (Figure 53B). The formation of the intercellular nanofibers transforms a 2D cell sheet to 3D cell spheroids (Figure 53C). The formation of the 3D cell spheroids depends on dephosphorylation, cell surface binding motif, and proteolytic resistance of the precursor. Similar to the recent report of ENS of proteins for cellular junctions,²¹⁸ this work illustrates ENS as a versatile approach to mimic protein dynamics for modulating intercellular interactions. One of the potential applications that remain to be explored is to apply ENS for modulating supracellular networks^398–400 during morphogenesis for tissue engineering.

Figure 53.

(A) The structure of a single component molecule precursor (117) for intercellular ENS. (B) Intercellular instructed assembly to mimic the essence of the dynamic of an ECM protein. (C) Optical images of HS-5 cells in culture medium (left) and coincubation with 117 (right). Scale bar = 150 μm. Adapted from Ref.³⁹⁶. Copyright © 2019 by American Chemical Society.

4.2. Peri/intracellular ENS

Pericellular and intracellular ENS processes are common features of cells. Although intracellular ENS of man-made molecules was explored in 2007,^227,229 a bona fide pericellular ENS of synthetic molecules was a rather recent event.²⁶⁷ In that study, a proteolytically stable, D-phosphotripeptide (119) turns out to be the substrate of ALP. After being dephosphorylated, 119 becomes 120 (Figure 54A), which self-assembles to form hydrogel/nanofibers in water. This transformation is likely benefited from the promiscuity of ALP to their substrates. That is, ALP is able to catalyze the dephosphorylation of both L- and D-peptide substrates.²⁶³ The addition of 119 to the culture of HeLa cells results in hydrogelation of culture medium. Further investigation reveals that the nanofibers form on the surface of the HeLa cells. The most significant insight is that overexpression of ALP on cancer cells leads to the formation of the pericellular nanofibers (Figure 54B, C), which block cellular mass exchange to induce apoptosis of cancer cells, including multidrug-resistance (MDR) cancer cells, MES-SA/Dx5. Moreover, the substrate is innocuous to normal cells. This discovery is largely due to the use of D-phosphopeptides, which are proteolytically resistant and ALP susceptible. In addition, the pericellular hydrogel/nanonets can entrap secretory proteins, which serves as a medium for enriched secretomes of cancer cells.⁴⁰¹

Figure 54.

(A) The structures of the precursor (119) and the hydrogelator (120) for pericellular ENS. (B) The illustration of ENS of pericellular hydrogel/nanonets to induce cell death. (C) SEM images of freeze-dried HeLa cells treated with 119 (560 μM) for 2 h. Scale bar = 10 mm. (D) Negative stained TEM images of the pericellular hydrogels on the HeLa cells treated by 119 at 280 μM. Adapted from Ref. 267. Copyright © 2014 by Wiley Inc.

To further understand the mechanism on how the pericellular nanofibers formed by ENS selectively kill cancer cells, a more detailed study was carried out.⁴⁰² The elucidation of the cell death mechanism of HeLa cells reveals that the nanofibers of 120, form locally on the surface of the HeLa cells and act as a pericellular nanonet around cancer cells specifically. The fibers are able to present the secreted, different proapoptotic ligands (e.g., TNFα and TRAIL) from cancer cells to bind with different extrinsic cell death receptors (e.g., TNFR1/2 and DR4/5), or directly interact with the death receptors (e.g., CD95) (Figure 55A). These actions lead to the death of cancer cells only. Further investigation on other co-cultures implies that 119 inhibits cancer cells likely via different ENS processes and different mechanisms. Incubated 119 with the co-culture of HeLa and HS-5 confirms that ALP-catalyzed ENS of the nanofibers of 120 selectively kills the cancer cells in the co-culture (Figure 55B). Moreover, the ENS of 120 nanofibers kills cancer cells selectively in different co-cultures.⁴⁰² In addition, inhibiting ALP reduces the dephosphorylation of 119, thus rescuing the cells in the co-culture (Figure 55C); adding extra ALP converts 119 to 120 before 119 reaches the cell surface, also rescues the HeLa cells in the co-culture (Figure 55D). These results confirm that the nanofibers of 120 have to be generated in situ for inhibiting cancer cells, which explains the exceptional selectivity of ENS against the cancer cells. Although the inhibitory concentration of 119 against cancer cells is relatively high in this case, this work indicates that ENS, as a molecular process, increases inhibitory efficacy to cancer cells without increasing toxicity to normal cells. Moreover, the pericellular localization of the nanofibers of 120 appear to be determined by the D-Phe-D-Phe backbone because replacing D-Phe-D-Phe in 119 by D-Trp-D-Trp results in the ENS product localizing in the lysosome and the corresponding ENS process hardly inhibits the HeLa cells.⁴⁰³

Figure 55.

(A) The illustration of pericellular fibrils of 120 formed by ENS to selectively inhibit cancer cells in co-culture via promiscuously activating cell death signaling. (B) The co-cultured cells (HS-5 and HeLa-GFP cells), HeLa-GFP, or HS-5 incubated with 119. (C) The co-cultured HS-5 and HeLa-GFP cells incubated with 119, 119+L-Phe, or L-Phe. (D) The co-cultured HS-5 and HeLa-GFP cells incubated with 119, 119+ALP, or ALP in DMEM for 48 hrs. (B, C,D) The initial number of cells is 1.0×10⁴/well (e.g., 1.0×10⁴ HeLa-GFP or HS-5 cells, or mixture of 5.0×10³ HeLa-GFP cells and 5.0×10³ HS-5 cells) and [119] = 216 μg/mL, [L-Phe] = 1.0 mM, [AP] = 0.1 U/mL. n = 3. Data are shown as mean ± SD. *p < 0.05, **p < 0.01 by Student’s t test. Adapted from Ref.⁴⁰². Copyright © 2017 by Springer Nature.

Because cells lack specific receptors or transporters for D-peptides, the result above implies that the cytotoxicity likely originates from the nanofibers rather than individual molecules. Thus, ENS for inhibiting cancer cells should be independent to the building blocks of the nanofibers, which is also revealed by the work of Pires and Ulijn.²⁷⁵ In their work, they used a simple carbohydrate amphiphile (121) bearing a phosphate group. After enzymatic dephosphorylation by ALP, 121 turns into 122, which self-assembles to form nanofibers. The authors used an osteosarcoma cell line, Saos2, which overexpresses ALP to incubate with 121, and found that hydrogelation and nanofibers formed mainly on the cell surface. By monitoring the metabolic activity of the Saos2 cells, they found that gelation of the pericellular environment reduced metabolic activity at an initial stage (≤7 h), and eventually caused cell death at longer exposure periods (≥24 h). Using prechondrocytes ATDC5 (that exhibits ~15–20 times lower ALP activity compared to Saos2) as a control cell line, they found that ATDC5 was hardly affected by 121 at concentrations ≤1 mM. Since 121 is a simple carbohydrate phosphate derivative unlikely to bind to a receptor with high affinity, this work, along with the example above, confirms that ENS generates the assemblies of small molecules in pericellular space (Figure 56A) for controlling cell fate. In a follow up study, the authors suggested that 121/122 also antagonized the glucose transporter (GLU1) for inhibiting the metabolism of the cancer cells.⁴⁰⁴

Figure 56.

(A) The structures of 121 and 122 and the illustration of enzymatic transformation of 121 to 122 upon enzymatic activity of ALP for in situ self-assembly. Adapted from Ref. 275. Copyright © 2015 by American Chemical Society. (B) The structures of 123, 124, and 125 and the viability assays of cancer cells and normal human cells after incubation with 123 (0.025 wt %) and the MMP-7 concentration in the culture media after culturing the cells. PE represents primary human pancreatic epithelial cells. (C) The structure of 126 and fluorescence recovery after photobleaching of HeLa cells observed using CLSM. Adapted from Ref.⁴⁰⁵. Copyright © 2015 by American Chemical Society.

The use of proteases for ENS^225,238 suggests that it is feasible to use proteases to generate nanofibers in cellular environment, as first demonstrated by Maruyama et. al.⁴⁰⁵ As shown in Figure 56B, the authors designed a peptide amphiphile (123) that was a substrate of matrix metalloproteinase-7 (MMP-7). Being proteolytically cleaved by MMP, 123 becomes a supramolecular hydrogelator (124). Being incubated with cancer cells overexpressing MMP-7, molecules of 123 are hydrolyzed by MMP-7. The cancer cells likely uptake the resulting 124, which is able to self-assemble inside the cancer cells to form nanofibers. The nanofibers disrupt cellular function and thus result in the death of cancer cells. This work also demonstrates that the intracellular self-assembly of the supramolecular gelator selectively kills cancer cells. Although the concentration required for killing cancer cells is still high, the excellent selectivity towards the cancer cells (Figure 56B) promises a fundamentally new approach for developing anticancer drugs. Using a fluorophore (NBD) linked to the peptide to generate the fluorescent analogue of 124, the authors, using fluorescence recovery after photobleaching (FRAP) (Figure 56C), also confirmed the increase of intracellular viscosity by the ENS of those peptide amphiphiles.

Although it is feasible to develop D-peptides that bind to certain cellular targets,^406–410 D-peptides lack interactions with endogenous transporters, thus are usually unable to enter cells efficiently. Taurine, a natural amino acid, is able to boost the cellular uptake of small D-peptides, as reported by recent studies.^411–412 Specifically, the conjugation of taurine at the C-terminal of a D-peptide through an ester bond generates the precursor, 127 (Figure 57A). After entering the cells, intracellular carboxylesterases (CES) catalytically cleaves the taurine group and results in a hydrophobic D-peptide (128), which self-assembles intracellularly to form nanofibers (Figure 57B). Since the nanofibers of 128 hardly diffuse out the cells, 128 accumulates inside the cells (Figure 57C). It is shown that, when the incubation concentrations of the D-peptides are about 200 μM, taurine conjugation, in combination with intracellular ENS, is able to increase the cellular uptake of small D-peptides in mammalian cells by >10-fold, from 118 μM (without conjugating taurine) to >1.6 mM (after conjugating taurine).⁴¹¹ A more carefully mechanistic study⁴¹² reveals that, for dynamin 1, 2, and 3 triple knockout (TKO) mouse fibroblasts, the cells uptake 127 via macropinocytosis and dynamin-dependent endocytosis. Further study using Drosophila larval blood cells derived from endocytic mutants confirms multiple endocytosis pathways contribute to the uptake of 127. Since the uptake is most efficient at 200 μM of 127, it is likely that 127 forms nanoparticles before entering cells, which was confirmed by TEM. These studies indicate that the cellular uptake of negatively charged substrates, including D-peptides, likely results from the aggregation of these relatively hydrophobic molecules.

Figure 57.

(A) The structures of precursor 127 and the corresponding hydrogelators 128 and the esterase catalyzed enzymatic transformation. (B) Taurine conjugation and intracellular ENS boosts cellular uptake of the D-peptide precursors and subsequent accumulation of nanofibers inside cells. (C) Fluorescent confocal microscope images showing the fluorescence emission in HeLa cells with the treatment of 127 at 200 μM concentration in culture medium for 24 h and co-stained with Hoechst 33342 (nuclei). Adapted from⁴¹¹. Copyright © 2015 by American Chemical Society. (D) Schematic illustration of a furin-catalyzed condensation of 131 to form the dimer of 132, followed by self-assembly to form nanoparticles for PET imaging. Adapted from Ref.⁴¹³. (E) The structures of 133 and 134 and the illustration of bacterial ALP-catalytically dephosphorylating 133 to form nanofibers. Adapted from Ref.⁴¹⁴. Copyright © 2016 by American Chemical Society.

For developing a radioactive probe for PET imaging, Liang et al. used the condensation reactions firstly developed by Rao et al.²⁸⁰ for intracellular ENS in tumor cells.⁴¹³ As shown Figure 57D, the authors synthesized a peptide substrate (130), which carried cyanobenzothiazole (CBT) at the C-terminal, a substrate of furin at the N-terminal, and a F-18 radioactive isotope label at the side chain. Intracellular furin catalytically cleaves the N-terminal to generate 131, which exposes the N-terminal of cysteine which condenses with CBT to form a dimer (132). The self-assembly of 132 results in nanoparticles with the F-18 labels. After using the F-19 analog to confirm the condensation reactions, the authors tested the F-18 probes in a tumor grafted murine model. MicroPET imaging of MDA-MB-468 tumor-bearing mice indicates that mice co-injected with 130 and the F-19 analog show higher uptake and longer attenuation of radioactivity in tumors than those mice only injected with same dosage of 130. These results indicate that self-assembly is critical for the retention of the probe and provides a useful approach for developing PET imaging agents based on ENS. In another study of intracellular ENS, Liang et al. also introduced iodine into the substrate of ALP for ENS.⁴¹⁴ They designed an iodinated hydrogelator precursor Nap-F-F(I)-_pY (133, Figure 57E). After being generated by ALP catalyzed dephosphorylation, Nap-F-F(I)-Y (134) self-assembles to form nanofibers, which result in a hydrogel. Notably, the authors applied 133 for direct nano-computed tomography (nano-CT) imaging, and demonstrated the detection of ALP activity in bacteria.⁴¹⁴ This pioneering work promises better nano-CT imaging of ALP activity if high contrast agents can be developed.

To address the problem of drug resistance in cancer therapy, Liang and Zhang used intracellular ENS for generating the nanoparticles of taxol inside cancer cells.⁴¹⁵ They developed a taxol derivative Ac-RVRR-C(StBu)-K(taxol)-CBT (135, Figure 58A) as a substrate for the furin. Furin cleaves 135 to generate CBT-taxol (136), which self-condensates to form dimers that self-assemble.²⁸⁰ According to the authors, 135 exhibits 1.5-fold increase of efficacy for inhibiting the taxol-resistant HCT116 cancer cells in vivo in a murine model, suggesting that 135 selectively targets cancer cells. This work suggests that protease catalyzed intracellular ENS (Figure 58B) may lead to a new strategy for overcoming MDR. This furin triggered condensation reaction is highly versatile for intracellular ENS, especially in developing imaging probes, as reported by Liang et al. to form F-19 MRI probes⁴¹⁶ and by Lin et al. to form PET probes.⁴¹⁷ In a related study using taxol,⁴¹⁸ Liang, taking the advantage of the biphasic effect of taxol on axonal branching, recently reported a taxol derivative Fmoc-FF-K(taxol)-_pY (137, Figure 58C), which was an ENS substrate, for promoting axonal branching. The authors found that ALP catalyzed the dephosphorylation of 137 to form the nanofibers of 138. While 137, at 10 μM, inhibits cell proliferation, at 10 nM, not only does it promote neurite elongation, as taxol does, but it also induces axonal branching. Although the mechanisms of this extremely interesting observation remain to be elucidated, this study shows the application of ENS outside of cancer therapy.

Figure 58.

(A) The structures of 135 and 136. (B) The illustration of intracellular furin controlled self-assembly of 136 for anti-MDR. Adapted from Ref.⁴¹⁵. (C) The structures of 137 and 138.

Rao et al. recently reported the first case of using ENS catalyzed by caspases for generating a tracer for positron emission tomography (PET).^419–420 The authors synthesized a substrate (139, Figure 59A) bearing a F-18 radioisotope. 139 carries the caspase cleavable N-terminal and a protected cysteine. After 139 enters tumor cells, intracellular caspases (i.e., caspases 3/7) catalytically cleave 139, then glutathione reduces the protected cysteine. These two reactions generate 140, which undergoes intramolecular cyclization to form 141. The self-assembly of 141 leads to the accumulation of 141 in tumor cells as nanoaggregates, which can act as a tracer for imaging tumors in vivo in a murine model. It appears, however, that most of the nanoaggregates are still trapped in the liver, which is a likely result from the considerable amount of glutathione in the liver.⁴²¹

Figure 59.

(A) The structures of 139 and its products after caspase cleavage and glutathione reduction. (B) The structures of the precursors (142 and 143) as substrates of carboxylesterase (CES) to the corresponding hydrogelators (144 and 145). (C) The illustration of intracellular ENS for disrupting actin dynamics. (D) The cell viability of A2780 cells and A2780cis cells incubated with the precursors 142 alone, or in combination with cisplatin (CDDP) for 72 h. Adapted from Ref.⁴²². Copyright © 2015 by Wiley Inc. (E) The structures of 147 and 148.

Combining intracellular ENS with anticancer drugs provides a new way to boost the activity of drugs against drug-resistant cancers without increasing systemic toxicity.⁴²² As shown in Figure 59B, the small peptide precursors (Nap-ff-es-taurine, 142, and Nap-FF-es-taurine, 143), being a substrate of CES, undergo CES-catalyzed hydrolysis to lose the hydrophilic taurine group and to form hydrophobic products (Nap-ff-es, 144, and Nap-FF, 145). 144 and 145 self-assembles in water to form nanofibers. After entering the cancer cells that overexpress CES, 142 and 143 turn into the nanofibers of 144 and 145, respectively, inside the cells. The formation of the intracellular peptide assemblies disrupts actin dynamics and results in the death of cancer cells (Figure 59C). At optimal concentration, 142 (or 143), being innocuous to cells, doubles or triples the activity of cisplatin against the drug resistant ovarian cancer cells, such as SKOV3 and A2780cis (Figure 59D). This work illustrates that ENS of small peptides is a useful strategy for combination therapy of cisplatin without increasing systemic burden or other side effects.

In fact, even without cisplatin, intracellular ENS to form peptide nanofibers is effective for selectively killing cancer cells.⁴²³ In the subsequent study of the CES substrates, both the D- and L-peptide precursors, 144 and 145, were tested on a panel of cancer cells. As shown in Figure 59C, intracellular CES catalyzes the cleavage of the ester bond to remove hydrophilic taurine groups, thus generating nanofibers of short peptides for selectively inhibiting cancer cells. The intracellular nanofibers disrupt actin dynamics, thus leading to cell death. This intracellular ENS approach allows selective inhibition of a range of cancer cells that exhibit relatively high CES activities, including drug resistant cancer cells (e.g., triple negative breast cancer (TNBC) cells (HCC1937) and platinum-resistant ovarian cells (SKOV3, A2780cis)). Generally, the D-dipeptide precursors exhibit lower IC₅₀ values (25–44 μg/mL) than those of L-dipeptide precursors (28–80 μg/mL) against the cancer cells. Most importantly, this intracellular ENS of dipeptides is innocuous to normal cells. Such selectivity is also valid in the co-cultures of cancer and normal cells. In addition, these precursors are non-toxic to normal tissues in mice, though the in vivo activity against cancer cells remains to be tested.

To address two shortcomings, that is, high uptake by the reticuloendothelial system and low uptake in tumors, of nanoparticle-based delivery systems, Chen et al. innovatively applied ENS, using Nap-FFK_pY (146, Figure 59E) as the precursor, to co-assemble 147 with indocyanine green (ICG) for in vivo cancer theranostics.⁴²⁴ ICG is a near-infrared (NIR) tricarbocyanine dye, which has been approved by the FDA for clinical imaging. Because ICG has a very short half-life (2–4 min) in vivo before biliary excretion, the uptake of ICG in tumor cells is low. To explore the potential of ICG for anticancer photothermal therapy (PTT), the authors mixed ICG with 146 and found that the mixture formed micelles, likely due to the interaction between the sulfonate group of ICG with the lysine groups of 146. After treating the micelles with ALP, the micelles turn into nanofibers of 147 and ICG adopts a J-aggregate arrangement. J-aggregates of ICG exhibit red-shift of absorbance, which is advantageous for PTT. The authors demonstrated that intravenous injection of the micelles in xenograft tumor mice models, greatly increased the tumor uptake of ICG (25-fold higher than that of the free ICG) and achieved high T/N ratios (>15) that clearly distinguished tumors from the surrounding normal tissue. In addition to being the first case of in vivo formation of tumor-specific ICG-doped nanofibers for PTT theranostics, this work illustrates the versatility and compatibility of ENS with other therapeutic approaches.

Reversible phosphorylation/dephosphorylation is a canonical mechanism for cell signaling. While most of the studies are focus on intracellular phosphorylation and kinases, ectophosphatases, such as ALP, in extracellular domains have received less attention. It has been recently recognized that cancer cells, overexpressing ALP for immunosuppression, resist cancer immunotherapy.⁴²⁵ Thus, profiling the activity of ALP on cancer cells becomes a necessary effort for immunotherapy, but most phosphatase probes are unable to serve as activity probes for imaging tumor specific phosphatases on cancer cells. ENS turns out to be a versatile approach for imaging the activity of ALP on live cancer cells.⁴²⁶ As shown in Figure 60A, a phosphorylated and nitrobenzoxadiazole (NBD) conjugated D-peptide (148) acts as a substrate of ALPs. Being dephosphorylated, 148 forms 149 which self-assembles to generate fluorescent nanofibers. Being non-diffusive, the nanofibers of 149 localize with ALP, thus revealing the activities of ALPs on live cells. This work, indeed, confirmed the significantly higher activities of ALP on cancer cells than on stromal cells in their co-culture (Figure 60B). Further developing this type of ENS-based molecular tool may ultimately lead to a probe for precision medicine that selectively targets cancer cells. Moreover, since ALP is a key enzyme in bacteria, the method shown in Figure 60B is also useful for profiling the activity of ALP in bacteria, as recently reported by Yang et al.⁴²⁷

Figure 60.

(A) The structures of 148 and 149. (B) The illustration of the ENS to form fluorescent pericellular nanofibers of 149 in co-culture and the fluorescent imaging of the fluorescent nanofibers of 149 selectively formed on HeLa cells (the cancer cell, white arrows) and in HS-5 cells (the stromal cell, white arrowheads) in the co-culture of HeLa and HS-5 cells ([148] = 500 mM). Adapted from Ref.⁴²⁶. Copyright © 2016 by Elsevier Inc.

Increasing the therapeutics index for selectively killing cancer cells without harming normal cells remains a holy grail in cancer therapy. ENS, as a multiple step molecular process, is able to enhance the efficiency for killing cancer cells without increasing toxicity to normal cells, as shown in the report of the molecular and cellular validation of ENS for selectively killing cancer cells.⁴²⁸ As shown in Figure 61, nine tetrapeptide derivatives (150–158) share a similar backbone, which consists of D-Phe and D-Tyr. In addition to the slight difference in the sequence of the backbone, they also differ in terms of the state of phosphorylation. The D-tetrapeptide precursors contain one or two phosphotyrosine residues. After enzyme-catalyzed dephosphorylation, the D-phosphotetrapeptides or the D-diphosphotetrapeptides turn into self-assembling tetrapeptides to form nanofibers in water. While the un-phosphorylated D-tetrapeptides are innocuous to several cancer cell lines and one normal cell line, the mono- and diphosphorylated D-tetrapeptides selectively inhibit the cancer cells, but remain nontoxic to the normal cell. Generally, the D-phosphotetrapeptides exhibit higher inhibitory activity against the cancer cells than the D-diphosphotetrapeptides do, confirming the importance of peptide assemblies for cancer inhibition. The cancer cell lines that express higher levels of ALPs are more susceptible to inhibition by the phosphorylated D-tetrapeptides. The cell death modality, either apoptosis or necroptosis, likely also associates with the interactions between nanofibers and the death receptors on different cancer cells. This work is the first comprehensive validation of ENS as a multiple step process for selectively killing multiple cancer cell lines, thus establishing the spatiotemporal defined supramolecular processes/assemblies for developing anticancer therapeutics.

Figure 61.

(A) The structures of the precursors that have one or two phosphotyrosine residues and the corresponding self-assembling D-peptides. (B) ENS of the tetrapeptides to inhibit cancer cells that express different levels of ALPs. (C) TEM images of aggregates/nanofibers in the solutions of different precursors or nanofibers in the hydrogels (inset: optical images) formed by treating the solutions of the precursors with ALP. (D) 48-h cell viability (determined by MTT assay) of HeLa and Saos2 cells incubated with different precursors at the concentrations of 200, 300, and 400 μM in culture medium. Adapted from Ref.⁴²⁸. Copyright © 2016 by American Chemical Society. (E) The illustration of the use of the rate of ENS (controlled by the number of enzymatic sites) to amplify the difference of the expression level of ALPL in different cell lines (HepG2 and Saos2). Adapted from Ref.⁴²⁹. Copyright © 2016 by Wiley Inc.

ENS is able to regulate the rate of the formation of molecular assemblies, which is one of the fundamental features of cells. This feature is useful for selectively targeting cancer cells. The molecular design is rather simple—tailoring the number of phosphates on peptidic substrates of alkaline phosphatases to make 153 and 154, as shown in Figure 61A.⁴²⁹ The pair of cell lines are HepG2 (liver hepatocellular carcinoma) and Saos2 (osteosarcoma), both known to express ALPL, but Saos2 expresses a much higher level of ALPL than HepG2 does. Utilizing different substrates, ENS is able to amplify this difference.. As shown in Figure 61E, the rate for generating the peptide assemblies of 155 should be slower with the diphosphorylated substrates (153) than with the monophosphorylated substrates (154) because the former requires dephosphorylation twice. Therefore, 153 should be more toxic to Saos2 than to HepG2, which is confirmed by cell viability assays: 153 and 154 exhibit similar inhibitory activities against Saos2 cells, but 153 is less cytotoxic than 154 towards HepG2 cells. This result establishes rate regulation of a multiple-step process to inhibit cells selectively, which may provide a way to target cancer cells in a specific organ.

It is facile to combine pericellular ENS with other cellular processes, such as intracellular condensation, as shown by the work of Liang et al.⁴³⁰ The authors developed a phosphopeptide ((SEt)CE_pYFFG-CBT, 159), which carried an ethylthiolester at the N-terminal and CBT at the C-terminal. 159 acts as the precursor of the hydrogelator (SEt)CE_pYFFG-CBT (160), which forms a cyclic dimer (161) via condensation reaction in the presence of intracellular glutathione (GSH). After confirming that both 160 and 161 were able to form hydrogels, the authors used 159 to incubate with HeLa cells. The ALP on the HeLa cells dephosphorylates 159 to form 160, which self-assembles to produce nanofibers of 160. After uptake by the cells via endocytosis, 160 forms the cyclic dimer 161 after the condensation initiated by intracellular GSH. 161 self-assembles into nanofibers that exhibit enhanced mechanical strength. This work illustrates an innovative way to modify peptide assemblies for a hierarchical morphological transition process in cellular environment.

Drug resistance remains the biggest challenge in cancer treatment. Among various types of drug resistance, platinum-resistant ovarian cancer is one of the hardest to treat. Resistance arises from multiple mechanisms, including cholesterol as an inducer.⁴³¹ Intriguingly, ENS of a cholesterol derivative (162, Figure 63A) is able to minimize acquired drug resistance for inhibiting ovarian cancer cells. This unusual feature originates from the multifaceted assemblies of 162/163 which are able to modulate multiple cellular processes.⁴³² Specifically, conjugating cholesterol with a D-phosphotyrosine generates a precursor (162) for ENS catalyzed by ALP. It turns out that 162 inhibits A2780cis cells and is more potent than cisplatin. The IC₅₀ of 162 is 13±1.3 μM (8.7 ±0.8 μg/mL, 48 h), which is about five times lower than the IC₅₀ of cisplatin against A2780cis—71±1.2 μM (21.2±0.4 μg/mL, 48 h). The inhibition of ALP only partially reduces the activity of 162, indicating multiple mechanisms contribute to killing the cancer cells. After the report that ENS, in-situ on or inside the cancer cells, generates the assemblies of the conjugate of cholesterol and D-tyrosine (163) to activate extrinsic and intrinsic cell death signaling simultaneously, a more extensive mechanistic study on the action of 162 further confirms multiple action mechanisms from the assemblies of 163.⁴³³ As shown Figure 63B, the assemblies, acting as polypharmaceuticals, augment lipid rafts, aggregate extrinsic cell death receptors (e.g., DR5, CD95, or TRAILR), modulate the expression of oncoproteins (e.g., Src and Akt), disrupt the dynamics of cytoskeletons (e.g., actin filaments or microtubules), induce endoplasmic reticulum stress, and increase the production of reactive oxygen species, thus resulting in cell death and preventing acquired drug resistance. These results validate that ENS of small molecules is a multifaceted strategy for amplifying the genetic difference between cancer and normal cells and for overcoming drug resistance in cancer therapy. A recent study shows that 162 promotes pro-inflammatory macrophages and induces apoptosis of cancer cells.⁴³⁴ Moreover, ENS enables an exceedingly simple lipid (e.g., the conjugate of phosphotyrosine and dodecyl amine) to target cancer cells selectively.⁴³⁵ These studies imply that ENS, as a molecular process, may lead to a new kind of multitargeting drugs.⁴³⁶

Figure 63.

(A) The structures 162 and 163. (B) ENS forms the assemblies of 162/163 to activate multiple cell death signaling pathways, thus minimizing the acquired drug resistance of the cancer cells. Adapted from Ref.⁴³³. Copyright © 2019 by American Association for Cancer Research.

To improve photoacoustic signal for imaging, one approach is to use ENS to generate the assemblies of the fluorophores, as shown by Wang et al.⁴³⁷ The authors developed a peptide (164, Figure 64A), consisting of a chlorophyll, a substrate of caspase-1, and a cell penetrating peptide (YGRKKRRQRRR). Upon enzymatic cleavage of the YVHDC at the D/C sites, 164 becomes a more hydrophobic molecule (165), which self-assembles to form the nanofibers that produce enhanced photoacoustic signals. According to the authors, this dynamic process allows monitoring of the activity of caspase-1 via ratiometric photoacoustic signals (Figure 64B). The authors suggested that this ENS process could offer a noninvasive method for real-time monitoring of bacterial infection, which associated with the upregulation of caspase-1 in the early stage. This type of protease catalyzed ENS also finds applications in delivering anticancer drugs, as reported by Ulijn et al.⁴³⁸ For example, they utilized MMP-9 overexpressed by cancer cells to enable ENS for enhancing drug specificity against cancer cells. Specifically, the authors synthesized a substrate of MMP-9, Ph-FFAGLDD (166, Figure 64C), which underwent proteolysis in the presence of MMP-9 to form hydrophobic segments, Ph-FFAGL (167) and Ph-FFAG (168). The authors reported that 166 formed micelles, which turned into fibrillar nanostructures upon the addition of MMP-9. The authors demonstrated that this ENS process allowed slow release doxorubicin to cancer cells (Figure 64D) for inhibiting tumor growth in a murine model.

Figure 64.

(A) The structures of 164 and 165. (B) The illustration of enzyme-induced dynamic equilibrium of nano-assemblies for modulating ratiometric photoacoustic signal in living cells. Adapted from Ref.⁴³⁷. Copyright © 2016 by American Chemical Society. (C) The structure of a substrate (166) of MMP-9 and the corresponding proteolysis products (167 and 168). (D) Schematic representation of micelle-to-fiber transition in the presence of cancer cells due to MMP-9 secretion, followed by entrapment of doxorubicin in fibrillar structures, which act as the less mobile depots of the anticancer drug. Adapted from Ref.⁴³⁸. Copyright © 2016 by Elsevier Inc.

Both gain-of-functions (i.e., upregulation) and loss-of-functions (i.e., down-regulation) can lead to cancer. Molecular therapy, based on inhibitory binding, is able to suppress gain-of-function in cancer cells, but it is unable to act on down-regulated targets in cancer cells. ENS of peptides is able to target down-regulation in cancer cells by a proper design, as shown in Figure 65.⁴³⁹ The key feature is to combine enzymatic assembly and disassembly. For example, in order to target down-regulation of carboxylesterases (CES) in OVSAHO, an ovarian cancer cell line, peptidic precursors (169 and 173) act as the substrates of both CES and ALP. The precursors, being dephosphorylated by ALP, turn into self-assembling molecules (170 and 174) to form nanofibers. In the presence of CES, 170 and 174 undergo hydrolysis to cleave the ester bonds and become 172 and 176, respectively. 172 (or 176), being hydrophilic, results in the disassembly of nanofibers of 170 (or 174). Thus, the precursors selectively inhibit OVSAHO, which down-regulates CES. But the same precursors are innocuous to HepG2, a hepatocyte that overexpresses CES. In addition to illustrating ENS as a potential approach for targeting down-regulation (or loss-of-functions) in cancer cells, this work also shows that 173 is more potent than cisplatin for inhibiting OVSAHO cells. The work also illustrates a multiple responsive approach for targeting tumors, which hold great promises, as evidenced by a recent work of Zhong et al. that uses multiple response to design a smart drug release depot for treating prostate cancer.⁴⁴⁰

Figure 65.

(A) The structures of the precursor and its hydrolysis products. (B) The concept of targeting the cells that downregulate CES while expressing ALP. (C) The structure and enzymatic conversion of the precursor 173, and the IC₅₀ values (at 72 h) of 173 against HepG2 or OVSAHO cells. Adapted from Ref.⁴³⁹. Copyright © 2017 by American Chemical Society.

The study of the precursors of ALP and CES also allows evaluating a series of C-terminal capped phosphopeptides (Figure 66), which reveal that self-assembling ability is the key factor for ENS to inhibit cancer cells.⁴⁴¹ Based on the precursor (173) for ENS, five precursors (177, 178, 179, 180, and 181) were made. These precursors, differing from 173 in terms of N-terminal capping, C-terminal modification, stereochemistry, or regiochemistry, aim to elucidate the effect of structural differences (i.e., N-terminal capping, C-terminal modification, stereochemistry, and regiochemistry) to the morphologies of assemblies and the contribution of the self-assembling abilities of either the phosphorylated precursors or the dephosphorylated products. That is, what features determine the efficacy of ENS for inhibiting cancer cells. Examining these ENS precursors reveals that, regardless of the stereochemistry and the regiochemistry of their tetrapeptidic backbones, the anticancer activities of these precursors largely match their self-assembling abilities. As the first case to correlate thermodynamic properties (e.g., self-assembling ability) of small molecules with the efficacy of ENS processes against cancer cells, this work provides an important designing principle for developing ENS for potential cancer therapy.

Figure 66.

Molecular structures of the precursors and the correlation between the ability for self-assembly of small molecules and anticancer activity. Adapted from Ref.⁴⁴¹. Copyright © 2017 by American Chemical Society.

Induced pluripotent stem cells (iPSCs) are another type of cells, besides certain cancer cells, that overexpress ALP. Taking advantage of the difference in ALP expression between iPSCs and normal cells, Saito et al. used ENS for the selective removal of human iPSC cells after using the iPSCs for generating muscle cells.⁴⁴² The authors tested four D-phosphopeptides, which are the substrates of ENS catalyzed by ALP, for inhibiting iPSCs, and found that a D-phospho-tetrapeptide (Nap-fff_py, 182, Figure 67A) is most effective. After dephosphorylation of 182 (at 0.35 mg/mL) by the ALP expressed on the iPSCs, the resulting D-peptide, Nap-fffy (183) self-assembles to kill the iPSCs within 1 hr. But 182 is largely innocuous to non-iPSCs, including primary hepatocytes and iPSC-derived cardiomyocytes. According to the authors, alkaline phosphatase, being highly expressed on iPSCs, induces dephosphorylation of the peptides. The spatial-temporal accumulation of dephosphorylated peptides around iPSCs generates peptide aggregates in situ that trigger cell stress and iPSC death. The absence or low expression of ecto-alkaline phosphatase in other cell types makes them tolerant to 182. The authors reported that two hours of incubation with 182 could efficiently eliminate iPSCs, even in the co-cultures spiked with increasing ratios of iPSCs. Using a mouse tumorigenicity assay, the authors confirmed that 182, being an ENS substrate of ALP, prevented the growth of residual iPSC to form induced teratomas in vivo. These results, again, highlight the excellent cell selectivity of ENS. This innovative work underscores the application of ENS in regenerative medicine based on iPSCs.

Figure 67.

(A) The structures of the phospho-D-peptide derivative. (B) The illustration of the possible mechanism by which phospho-D-peptides induce iPSC-selective cell death. Adapted from Ref.⁴⁴². Copyright © 2017 by Elsevier Inc.

It is a challenge to make short peptides (with <5 amino acids) adopt an α-helical conformation, but Yang et al. reported that ENS may provide a way to assist the self-assembly of peptides into such a conformation.⁴⁴³ The authors synthesized a peptide, chlorambucil-Gffy (CRB-Gffy, 184, Figure 68A), and its corresponding phosphopeptide, CRB-Gff_py (185). After a heating–cooling process, 184 adopts a conformation more similar to a β-sheet and self-assembles into nanoparticles, which result in a suspension. Adding ALP to the solution of 185, at 4 °C, generates 184, which adopts an α-helical conformation and self-assembles into stable nanofibers and hydrogels (Figure 68B). Interestingly, the nanofiber solution exhibits better stability against proteinase K digestion and results in an enhanced cellular uptake. Using a panel of cancer cells and a murine tumor model, the authors reported that the nanofibers of 184 were more effective than the nanoparticles of 184 for inhibiting cancer cells in vitro and in vivo (Figure 68C). Moreover, the authors also reported that 184 was more compatible to white blood cells than CRB. This result is particularly attractive because it may lower the major side effects of CRB. This study shows the unique advantages of ENS to control peptide folding and self-assembly. It would be interesting to know the molecular arrangements in these two types of nanostructures, in addition to the conformational differences of the building blocks.

Figure 68.

(A) The structures of chlorambucil-Gffy (184) and chlorambucil-Gff_py (185) (B) The optical images of the suspension of 184 formed by a heating-cooling process and the hydrogel of 184 formed by ALP enzymatic reaction at 4 °C and the TEM images of the nanostructures in the suspension and the hydrogel (scale bars = 500 nm). (C) the IC₅₀ values of CRB, the nanofibers of 184, and the nanoparticles of 184 against different cancer cells, and the in vivo anti-cancer efficiency of CRB, the nanofibers of 184, and the nanoparticles of 184 against a 4T1 tumor model (the data are shown as mean ± SEM (n = 5); ** represents P < 0.05). Adapted from Ref.⁴⁴³. Copyright © 2017 by Royal Chemical Society.

ALP, being a promiscuous ectoenzyme on the cell membrane, can enable the ENS of a wide range of substrates for many innovative applications, such as the patching of lipid rafts, as reported by Zhang et al.⁴⁴⁴ As shown in Figure 69, the authors used the metal complex Ru(II)(bpy)₃ as a three-dimensional (3D) core to conjugate with the D-phosphotetrapeptide (Nap-ffk_py²⁶³) to generate a metal complex (186) as the substrate of ALP. The authors, after confirming that ALP converted the solution of 186 to a hydrogel, incubated 186 with HeLa, HS-5, Ect1/E6E7, and A375 cells. Ru(II)(bpy)₃, being fluorescent and having a long fluorescent life time, allows the authors to monitor the ENS of 186 on the surface of HeLa cells and to examine how the pericellular assemblies affect the dynamics of cytoskeleton. Based on the imaging results, the authors reported that ALP in the lipid rafts catalyzed the dephosphorylation of 186, thus creating the self-assembled nanofibrils patch on lipid rafts. This patch, inducing lipid raft dysfunction and reinforcing focal adhesion, glues the lipid rafts together to raft-associated receptors to provoke opposing cell migration against focal adhesion. Such contradictory motions eventually cause cell rupture and cell death. This rather comprehensive study underscores the versatility of ENS as a multiple step process for modulating multiple cellular functions.

Figure 69.

The ruthenium complex (186) with octahedral geometry serves as a core building block of peptidic molecule construction for protein-guided molecular self-assembly with enhanced sensitivity. The self-assembled nanofibrils selectively capture the lipid rafts of cervical cancer cells and exert mechanical stimuli on raft-associated receptors, leading to inhibition of cancer cell migration and cellular physical damage. Adapted from Ref.⁴⁴⁴. Copyright © 2017 by Elsevier Inc.

To develop an MRI contrasting agent to report ALP expression, Liang et al. reported an ALP substrate, Nap-FFF_pY-EDA-DOTA(Gd) (187, Figure 70A), which self-assembled into gadolinium nanofibers upon the action of ALP. After confirming the self-assembly of the peptide Nap-FFFY-EDA-DOTA(Gd) (188) by hydrogelation, the authors injected 187 in mice to image a tumor. According to the in vivo T2-weighted MRI at 9.4 T, 187 is able to reveal the HeLa tumor on mice in vivo (Figure 70B). The MRI signal intensity of the HeLa tumor in the mice injected with 187 is higher than the mice injected with Gd-DTPA, suggesting the accumulation of the nanofibers of 188 in the tumor. It remains to be seen if the contrast enhancement will be preserved in a magnetic field with lower strength.

Figure 70.

(A) The structures of 187 and 188. (B) The illustration of ALP-based ENS of gadolinium nanofibers and in vivo T₂-weighted coronal MR images of 187-injected mice (top row) and Gd-DTPA-injected mice (bottom row) at 0 h and 3 h post injection. Adapted from Ref.⁴⁴⁸. Copyright © 2017 by American Chemical Society. (C) The structures of the precursors for ALP-catalyzed assembly. (D) The illustration of intracellular ENS resulted in co-assembly to form nanofibers made of 190 and 192. Adapted from Ref.⁴⁴⁵. Copyright © 2017 by Royal Chemical Society.

To establish a new approach for enhancing the efficacy of dexamethasone (Dex), a steroid for treating inflammation, Liang et al. developed a simple approach that used ENS to co-assemble Dex with a hydrogelator for making hydrogels.⁴⁴⁵ To avoid the formation of Dex precipitates after using ALP to dephosphorylate dexamethasone sodium phosphate (191, Figure 70C), they mixed the hydrogelator precursor Nap-FF_pY (189) with 191. Adding ALP to the solution of a 1:1 (molar ratio) mixture of 189 and 191, they obtained a hydrogel due to co-assembly by ENS of Nap-FFY (190) and Dex (192). According to the authors, intracellular ALP triggered the co-assembly of 190 and Dex and boosted the anti-inflammation efficacy of Dex on two types inflammatory cell models (Figure 70D). This simple approach illustrates a useful application of ENS for intracellular co-assembly, which appears to be a rather general approach⁴⁴⁶ for further development. In fact, Jiang et al. recently reported the use of ENS of 189 to control intermolecular forces for generating β-sheets based on a multi-modal analytical system that satisfied both point-of-care testing (POCT) and laboratory-based testing.⁴⁴⁷

Besides proteases or phosphatases for bond breaking, ligases, such as transglutaminases (TGase),²²³ provide a useful approach for intracellular polymerization and self-assembly, as reported by Wang et al.⁴⁴⁹ They used elastin-based peptide sequences bearing a functional motif (e.g., fluorophore) and one or two pairs of the substrates of the TGases. The TGase-instructed polymerization occurs via formation of an isopeptide bond between the side chains of glutamine and lysine. According to the authors, the substrates enter the cells to undergo intracellular enzyme-catalyzed polymerization, which results in nanoparticles or 3D gel-like structures, depending on the elastin sequences. While the nanoparticles are cell compatible, the 3D gels are cytotoxic. Although more detailed characterization of the 3D gel is warranted, these findings illustrate the versatility of intracellular ENS for biomedical applications.

Autophagy, being an endogenous mechanism of the cell, removes unnecessary or dysfunctional components in cells. Wang et al. recently reported the use of intracellular ENS for monitoring of autophagy.⁴⁵⁰ As shown in Figure 71A, a bis(pyrene) derivative (BP) is connected to a dendrimer core by a peptide linker that is a substrate of an autophagy-specific enzyme, ATG4B, to generate nanoparticles (193). On the nanoparticles, the fluorescence of BP is quenched. Inside cells, ATG4B cleaves the peptide GTFGFSGKG at the G/F site, releases the BP-peptide conjugates (194), which self-assemble to form nanofibers and “turn-on” the fluorescence (Figure 71B). According to the authors, this in situ intracellular self-assembly strategy provides a rapid, effective, real-time, and quantitative method for monitoring autophagy in living cells. Verification of whether the relatively high concentration of nanoparticles also induce autophagy is still needed. Using the same ATG4B substrate, but replacing BP with purpurin-18, the same lab generated another type of nanoparticle (195) as a photoacoustic (PA) nanoprobe for real-time and quantitative detection of autophagy in mice for the first time.⁴⁵¹ That is, ATG4B cleaves the peptide at G/F site to generate 196 for self-assembly. This innovative design, developed earlier by Gianneschi et al.,⁴⁵² which uses enzymes to cleave peptides off polymers or dendrimers to enable the self-assembly of peptides, likely would lead to many other advances in the applications of intracellular ENS. This includes inducing or evading macrophage cell uptake.⁴⁵³

Figure 71.

(A) The structure of dendrimers connected to bis(pyrene) or purpurin 18 by peptide linker. (B) The illustration of the probe for autophagy detection. Adapted from Ref.⁴⁵⁰. Copyright © 2017 by American Chemical Society. (C) Schematic illustration of MMP-2-triggered self-assembly of AuNPs@197/199. Adapted from Ref.⁴⁵⁴. Copyright © 2017 by Royal Chemical Society. (D) The structures of 200 and 201 and the TEM images of 200 treated with ALP for 1 min and for 60 min. Adapted from Ref.⁴⁵⁵. Copyright © 2017 by Wiley Inc.

In fact, it is possible to enable the self-assembly of peptides when the peptides are still attached to the nanoparticles, as shown by the work of Wang and Li et al. that combine ENS with gold nanoparticles.⁴⁵⁴ Peptide modified spherical gold nanoparticles (AuNPs@197/199 in Figure 71C), with a 16:1 ratio of 197 and 199, are able to be monodispersed over a wide range of ionic strengths, pH, and fetal bovine serum (FBS). After 197 is cleaved by matrix metalloproteinase-2 (MMP-2), 199 becomes exposed and interacts with 199 from other gold nanoparticles, which results in the aggregation of the gold nanoparticles. The aggregation leads to a red shift in localized surface plasmon resonance (LSPR) and selective accumulation of the nanoparticles on tumor sites. Taking advantage of the red-shift to the NIR window and enhanced retention in tumor sites, the authors achieved improved photothermal treatment (PTT) against the tumors. One of the remaining issues is the aggregation of the gold nanoparticles in vivo.

To develop a molecular probe to reveal the localization and activity of phosphatases, Liang et al. reported a simple peptide substrate of phosphatase, as shown in Figure 71D.⁴⁵⁵ The peptide (200), bearing a tetraphenylethylene (TPE) at the N-terminal of a phosphotripeptide Y_pYY, undergoes dephosphorylation, catalyzed by ALP, to form TPE-YYY (201). Notably, it appears that ALP converts the fibrillar aggregates of 200 to the nanoparticles of 201. Although it was reported that 200 diffused into cells rapidly, the fluorescence of 201 inside the cells was diffusive and relatively weak. More studies are needed to determine whether TPE binds to other targets inside cells or how stable 201 is inside the cells.

The versatility of ENS allows it to be combined with other stimuli for controlling self-assembly of peptides, as shown by the work of Zhang et al., which integrates ENS with photo-response.⁴⁵⁶ As shown in Figure 72A, the authors developed a phosphopeptide (Tet-Gf_py, 202), bearing a biaryl-substituted tetrazole with an o-allyloxy group on the N-phenyl ring (Tet). 202 turns into Tet-Gfy (203) after the dephosphorylation of 202 catalyzed by ALP, and 203, being a hydrogelator, forms a supramolecular hydrogel. Photo-irradiation transforms the Tet motif to pyrazoline (Pyr), accompanied by fluorescence. Since the self-assembling ability of Pyr-Gfy (204) is lower than that of 203, photo-irradiation also induces gel-sol transition. The authors also demonstrated the combination of ENS and photo-response on the surface of cells, showing that photo-irradiation disassembled the assemblies of 203 formed by pericellular ENS (Figure 72B). This innovative work may lead to a new way to modulate pericellular dynamics for selectively controlling cell death and survival, which may find applications in controlling the fate of iPSCs.⁴⁴²

Figure 72.

(A) The structures and transformations of 202 upon treatment with ALP followed by photoirradiation. (B) The illustration of using 202 for monitoring EISA–PIDA processes and the photo-regulation of the cancer cells. Adapted from Ref.⁴⁵⁶. Copyright © 2017 by Royal Chemical Society. (C) The structures of the C-terminal methylated phosphotetrapeptide (205) and the corresponding tetrapeptide (206). (D) The illustration of ENS in cellular milieu causes cell stress to activate NF-κB signaling for cell survival and the combination of ENS and NF-κB targeting effectively inhibits cancer cells. Adapted from Ref.⁴⁵⁷. Copyright © 2018 by American Chemical Society.

Because ENS is able to selectively target cancer cells, it boosts the efficacy of other cancer therapeutics without increasing side effects, as shown by a recent study in Figure 72C, D.⁴⁵⁷ In that study, ENS exhibits strong synergism with the inhibitors, bortezomib (BTZ)⁴⁵⁸ and BAY 11–7085 (BAY),⁴⁵⁹ both of which target the transcription factor nuclear factor-κB (NF-κB). The substrate, C-terminal methylated phosphotetrapeptide (205), of ALP results in extra- and intracellular assemblies of the C-terminal methylated tetrapeptide (206). The assemblies of 206, though unable to kill cells, causes cell stress, results in inductive expression of tumor necrosis factor receptor 2 (TNFR2),⁴⁶⁰ and decreases the expression of three key proteins (PI3K,^461–462 Akt,⁴⁶³ and MEKK3⁴⁶⁴) at the up-stream of NF-κB signaling in cancer cells. In the presence of the inhibitors targeting NF-κB signaling, 205 drastically decreases cancer cell viability (about an order of magnitude). Further mechanistic study indicates that the combination of the ENS of 206 and NF-κB inhibitors decreases the expressions of those up-stream proteins, which eventually results in cell death via necroptosis. This work implies that ENS, combined with clinical cancer therapeutics, may facilitate the translation of key regulatory circuits into promising targets of cancer therapy.

In cellular environment, it is common for the same enzymes to present at different locations (e.g., extra- and intracellular) in different amounts. When the substrates of the enzymes are used for ENS, it is necessary to analyze the kinetics of ENS to understand the efficacy of the substrates for functions, such as cancer cell inhibition, as shown in Figure 73.⁴⁶⁵ That study has examined three substrates of CES (142, 207, and 209, Figure 77A) for intracellular ENS. In studying hydrolysis catalyzed by CES, these substrates result in hydrogelators, which self-assemble in water to form nanofibers. Because CES exists both extra- and intracellularly, these substrates, being incubated with ovarian cancer cells, undergo intracellular hydrolysis to form intracellular nanofibers in addition to some extracellular hydrolysis. While the three substrates selectively kill cancer cells, 142 and 209 exhibit the highest and the lowest activities, respectively, a trend that inversely correlates with the rates of converting the substrates to the hydrogelators in PBS buffer. Detailed kinetic modeling provides quantification of the key rate constants (Figure 77B) of ENS inside cells and calculates the activity of each substrates for killing the cancer cells (Figure 77C). The kinetic analysis also reveals that (i) substrate stereochemistry determines the rate of enzymatic conversion and the morphology of the assemblies; (ii) less extracellular hydrolysis of the substrates leads to more intracellular ENS; (iii) the self-assembling ability of the ENS molecules also dictate the cytotoxicity of intracellular ENS. This work illustrates that stereochemistry is a useful modulator for developing anticancer ENS in the complex extra- and/or intracellular environment.

Figure 73.

(A) The structures of the precursors and the corresponding hydrogelators. (B) Extracellular hydrolysis and cancer cell death caused by intracellular ENS. (C) IC₅₀ values (against JHOS-4) and t_1/2 values of 209, 207, and 142. Adapted from Ref.⁴⁶⁵. Copyright © 2018 by American Chemical Society.

Figure 77.

(A) The structure of 226. (B) The illustration of MMP-2-triggered transformation of Ppdf-Gd from spherical nanoparticles to nanofibers for amplified MRI and PDT. Adapted from Ref.⁴⁷⁵. Copyright © 2018 by Elsevier Inc. (C) The illustration of ALP-based ENS of the assembly of peptides which inhibit metastatic osteosarcoma in an orthotopic mouse model and the tumor volumes of orthotopic osteosarcoma model established by Saos2-luc and Saos2-lung cells after tail intravenous injection of compound 173 or saline for 4 weeks. Adapted from Ref.⁴⁷⁹. Copyright © 2019 by Elsevier Inc. (D) The structure of Nap-Gff_py (227), the illustration of self-assembly of 227 induced by Ca²⁺, and the optical image of the formed _pY-Gel; (E) Conversion ratios of phosphorylated peptides in PBS, _pY-Gel and Y-Gel. Adapted from Ref. 304. Copyright © 2018 by Royal Chemical Society.

To address the problems of low drug loading and loss of function due to the covalent modification of the antibody in antibody-based medicine, Yang et al. reported an innovative application of ENS.⁴⁶⁶ As shown in Figure 74A, a phosphopeptide (NBD-Gff_py, 38) is mixed with anti-HER2 antibody to form a solution. The addition of ALP to the solution, at 4 °C, produces a clear hydrogel (Figure 74B). This simple process loads >30 wt% of the antibody and significantly improves the stability of the antibody at 37 °C (>15 d in vitro). According to the authors, the nanofibers exhibit high affinity for HER2+ cancer cells and efficiently enters the cells. Using a murine tumor model, the authors demonstrated the shrinkage of the tumors when CRB-HA-Gff_py (185) was mixed with the antibody for making the hydrogel/nanofibers. This study illustrates using ENS to combine antibody and alkylating agents for cancer therapy.

Figure 74.

(A) Optical images of a PBS solution containing NBD-Gff_py (38, 0.2 wt%) and antiHER2 affibody (0.03 wt%) and hydrogel formed by the addition of ALP, and the illustration of the co-assembly process triggered by ALP. Adapted from Ref.⁴⁶⁶. Copyright © 2018 by Wiley Inc. (B) The structures of NBD-GFF_pY-ss-ERGD (211) and the corresponding dephosphorylation and reduction products. (C) Optical images of a PBS solution of 211 (200 μM, 0.03 wt%), PBS solution of 213 formed by adding ALP (1 U/mL) to the solution of 211, and hydrogel formed of 213 by adding GSH (4 equiv.) to the solution of 212, and the proposed mode of the molecular self-assembly extra- and intra-cellular environment of HepG2 and QGY7703 liver cancer cell lines. Adapted from Ref.⁴⁶⁷. Copyright © 2018 by Wiley Inc.

Yang et al. recently developed an innovative tandem molecular self-assembly that is controlled by ENS and an intracellular redox reaction.⁴⁶⁷ As shown in Figure 74C, the peptide (211) consists of two segments, NBD-GFF_pY and ERGD, that are linked by a disulfide motif. 211, upon dephosphorylation catalyzed by ALP, becomes 212, which self-assembles to form a micelle solution. The addition of GSH, reductively cleaving the disulfide bond, generates 213, whose assemblies become nanofibers to form a hydrogel. The authors demonstrated this tandem self-assembly using liver cancer cells that exhibited higher concentrations of both phosphatase and GSH than normal cells. It is also interesting that the morphologies of nanofibers in the two liver cancer cell lines, HepG2 and QGY7703, differ, which may be worth further investigation. This unique utilization of both extracellular and intracellular reactions to trigger tandem molecular self-assembly is exciting and promising for the development of cancer diagnostics and therapy.

Taking the advantage of the long lifetime of (Ru(bpy)₃²⁺) complex,⁴⁶⁸ Liang et al. developed a substrate for intracellular imaging.⁴⁶⁹ The molecule (Cys(StBu)-Lys-(Ru(bpy)₃²⁺)-CBT, 214, Figure 75A) consists of a latent cystine at the N-terminal, Ru(bpy)₃²⁺ at the side chain of lysine of the peptide, and CBT at the C-terminal. As shown in Figure 75B, 214, after entering the cells and being reduced to expose the thiol group in cysteine, undergoes a condensation reaction to form a trimer of 215, which self-assembles to form nanoparticles of 215 with non-quenchable, persistent phosphorescence. The authors also demonstrated the fluorescence from 214 for imaging HepG2 cancer cells in a tumor murine model. It appears, however, that the efficiency of imaging remains to be improved.

Figure 75.

(A) The structure of 214. (B) The illustration of intracellular condensation-controlled transition of 214 to 215-NPs for tumor imaging. Adapted from Ref.⁴⁶⁹. Copyright © 2018 by Royal Chemical Society. (C) The illustration of ALP-catalyzed ENS of 217 to form Dex-nanofibers followed by esterase-controlled slow release of Dex from the nanofiber. Adapted from⁴⁷⁰. Copyright © 2018 by American Chemical Society. (D) The illustration of ALP-catalyzed ENS to form 219-NPs, from 218, for enhanced photoacoustic imaging of tumors. Adapted from Ref.⁴⁷¹. Copyright © 2018 by American Chemical Society.

To develop a strategy for treating hepatic fibrosis, Liang et al. further developed ENS for delivering Dex⁴⁷⁰ after their earlier report that intracellular co-assembly boosted the antiinflammation capacity of dexamethasone.⁴⁴⁵ As shown in Figure 75C, they made a hydrogelator precursor Nap-FFK(Dex)_pY (216) for the slow release of Dex by ENS and enzymatic release. That is, the action of ALP produces Nap-FFK(Dex)Y (217), which self-assembles into nanofibers. Because Dex is linked to lysine via an ester bond, the action of esterase cleaves Dex from the nanofibers for slow release of Dex. This work illustrates the importance of designing ENS not only for assembly, but also for disassembly. More useful applications applying this concept⁴³⁹ will likely emerge.

To develop probes for photoacoustic (PA) imaging, Liang et al. conjugated a NIR dye with a substrate of ALP for tumor imaging.⁴⁷¹ As shown in Figure 75D, they designed and synthesized a NIR probe IR775-FF_pY (218), which, being catalytically dephosphorylated by ALP, became IR775- FFY (219). After ENS forms the nanoparticles of 219, the PA signal is enhanced 6.4-fold. The authors reported that in vivo tumor PA imaging exhibited 2.3 folds increase due to the ENS of 219 catalyzed by ALP. This work represents the first example of an ALP-activatable probe for enhanced PA imaging of tumors.

It is relatively easy to combine ENS with photo activatable molecules for selectively targeting certain cancer cells, as reported by Ding et al.⁴⁷² They designed and synthesized a substrate, 220 (Figure 76A), that consisted of a fluorophore (TPE-Py) and a short peptide with three tyrosine phosphates (_pY). 220, being the first reported triphosphoheptapeptide, exhibits good aqueous solubility, weak fluorescence, and negligible ROS generation ability. After ALP of Saos2 cells, enzymatic action dephosphorylates 220, then, the ALP-catalyzed product (221) self-assembles to form nanoscale aggregates, which are highly fluorescent and are able to produce ROS efficiently during irradiation (Figure 76B). Cellular studies reveal that 220 is specific towards Saos2 cells. It is, however, less clear on how to use photoactivated ROS for treating solid tumors formed by Saos2.

Figure 76.

(A) The structure of TPE-Py-F_pYG_pYG_pY (220). (B) The illustration of the self-assembly of TPE-Py-FYGYGY (221) generated by the action of ALP that is overexpressed on the cancer cell surface, which significantly activates the fluorescence and ROS generation capability. Adapted from Ref.⁴⁷². Copyright © 2018 by Royal Chemical Society. (C) The structures of 222 and 223. (D) Enzymatic transformation of the precursor as a substrate of ALP to the corresponding aggregates for intracellular self-assembly. Adapted from Ref.⁴⁷³. Copyright © 2018 by Royal Chemical Society. (E) The structures of 224 and 225. (F) The scheme of the preparation of the core–shell nanogels by ENS.

To target prostate cancer, Wang et al. combined gold nanoparticles (AuNPs) with a phosphopeptide for ENS, which facilitated photothermal therapy.⁴⁷³ As shown in Figure 79, the authors synthesized a phosphopeptide, CREKA_pYPFFK(Nph) (222, Figure 76C). The sequence CREKA is a pentapeptide structure binding to fibronectin and forming Au–S bonds with AuNPs via the cysteine residue. The conjugation of AuNPs with 222 affords AuNPs@222, which have diameters of 13 nm. Using ALP to treat the AuNPs@222, the authors confirmed the aggregation of the nanoparticles after dephosphorylation converted 222 to CREKAYPFFK(Nph) (223). They incubated PC-3 and MCF-7 cells with AuNPs@222 and found more accumulation of the AuNP@223 in the PC-3 cells than in MCF-7 cells. After AuNPs@222 were injected into mice, the accumulation of AuNPs@223 in tumor sites allowed effective photothermal therapy in vivo (Figure 76D). This work demonstrates, again, that inorganic nanoparticles, with a proper surface modification, can act as the substrates for intracellular ENS.

Figure 79.

(A) The structures of NBD-FFY(gal)G (236) and NBD-FFYG (237). (B) CLSM image of HeLa cells incubated with 236 or 237 (200 μM) for 12 h. Adapted from Ref.⁴⁸². Copyright © 2019 by Royal Chemical Society. (C) The structure of Nap-GGYGSSSRRAPQT (238). (D) Representative laser Doppler perfusion images of mice subjected to limb ischemia at Day 0 and Day 28 after injection of Gel-A, IGF-1, ad Gel-B. Adapted from Ref.⁴⁸⁵. Copyright © 2019 by American Chemical Society.

Using the enzyme attached on the nanoparticles of silica, Wang et al. reported an innovative approach to core–shell nanogels for drug delivery.⁴⁷⁴ As shown in Figure 80, CES, being immobilized on the surface of silicon nanoparticles (SNPs), catalyzes the hydrolysis of a substrate (Nap-FF-es, 224, Figure 76E) to form a hydrogelator (Nap-FF-e, 225). The self-assembly of 225 forms a layer of hydrogel around the silica nanoparticles to generate the core–shell SNP nanogels (Figure 76F). Changing the concentrations of 224 controls the growth of the gel layer. The nanogels can encapsulate drugs, such as doxorubicin (Dox), and release the Dox when proteases cleave the peptides (225) on the surface of the nanoparticles. The Dox-loaded nanogels, being incubated with HeLa cells, undergo endocytosis and release Dox to the cell nuclei within 24 h.

Figure 80.

(A) The structures of the 239 or 240. (B) The illustration of furin-instructed formation of CBT-⁶⁸Ga nanoparticles and microPET images of MDA-MB-468 tumor-bearing mice injected with either CBT-⁶⁸Ga alone or the mixture of CBT-⁶⁸Ga and CBT-Ga. Adapted from Ref.⁴⁸⁶. Copyright © 2019 by American Chemical Society. (C) The illustration of the CTSB-specific bioluminescence “turn-on” after cellular internalization. (D) Time-dependent images of tumors in a murine model. Adapted from Ref.⁴⁸⁸. Copyright © 2019 by American Chemical Society.

To combine MRI and PDT, Han et al. reported a peptide conjugate that consisted of protoporphyrin IX (PpiX), a proteolytic site, and DOTA(Gd), as shown in Figure 77A, B.⁴⁷⁵ This chimeric peptide PpiX-PEG8-SSSPLGLAK(DOTA)-PEG6F4 (denoted as Ppdf-Gd, 226), being a substrate of MMP-2, aims to amplify MRI for guided photodynamic therapy (PDT) of tumors. 226, being amphiphilic, self-assembles to form spherical nanoparticles, which turn into nanofibers upon the action of MMP-2 that hydrolyzes the Pro-Leu-Gly-Leu-Ala (PLGLA) peptide sequence at the site of G/L. One notable feature is that the sphere-to-fiber switch results in a higher relaxation rate of DOTA(Gd). The incorporation of protoporphyrin IX also allows 226 to be an agent for PDT. The in vivo examination of the nanoparticles of the chimeric peptide indicates high accumulation of the particles in the liver, which is a common drawback of nanoparticles.

The study of immunosuppression reveals that immunosuppressive adenosine⁴²⁵ in the tumor microenvironment is a major cause for patients’ unresponsiveness to the treatment⁴⁷⁶ of cancer immunotherapy based on checkpoint blockade.⁴⁷⁷ ALP can rapidly convert ATP to adenosine which causes immunosuppression in the tumor microenvironment. While it is still a challenge to develop suitable inhibitors⁴⁷⁸ for ALP, ENS catalyzed by ALP can target tumors overexpressing ALP, as shown in Figure 77C.⁴⁷⁹ A recent study shows that 173, the substrate of ALP and CES, uses ENS to inhibit osteosarcoma in orthotopic murine models. The key feature of 173 is that, being a substrate of CES, it selectively inhibits osteosarcoma without affecting liver cells. Compared to the control (saline solution), 173 significantly inhibits the progression of Saos2-luc and its metastatic subline Saos2-lung (Figure 77C). Validating the in vivo anticancer efficacy of ENS of peptides for the first time, this work ultimately may lead to a fundamentally new way to target immunosuppressive cancer cells and establish immune normalization for cancer treatment.

To develop approaches for promoting antibodies to recognize phosphorylated proteins, Yang et al. reported an innovative method for the efficient production of antibodies of phosphorylated proteins.³⁰⁴ They used Ca²⁺ to induce the self-assembly of a phosphopeptide, Nap-Gff_py (227, Figure 77D), to form a hydrogel (i.e., Y-Gel). Because the hydrogel, as a medium to encapsulate phosphorylated antigens, reduces the dephosphorylation of the antigen, the ratio of the antibodies for phosphorylated proteins increases (Figure 77E). Although the mechanism remains to be elucidated, this study opens a new way for producing antibodies that recognize specific posttranslational modifications in proteins.

To increase inhibitory capacity and decrease systemic toxicity, Yang et al. combined ligand-receptor binding and pericellular ENS.⁴⁸⁰ As show in Figure 78A, they designed a substrate (229) of ALP containing the CCK-6 peptide (Nle-Gly-Trp-Nle-Asp-Phe), a specific ligand of the human cholecystokinin-2 receptor (CCK2R), at the C-terminal of 60. Using two cancer cell lines (HeLa and HepG2) that overexpressed CCK2R, the authors showed that HeLa cells treated by 229 were covered by large amounts of nanofibers made of 230 (Figure 78B). Using RGD connected to the C-terminal of 60 and 60 as the controls, the authors confirmed the selectivity of 229 to the CCK2R expressed cancer cell lines. This work illustrates a facile approach to enhance the selectivity of pericellular ENS for detecting or inhibiting cancer cells.

Figure 78.

(A) The structures of NBDGFF_pY-G-CCK6 (229); (B) CLSM images of HeLa cells incubated with 229, 230, 231, and 66 (from left to right) for 8 h. Adapted from Ref.⁴⁸⁰. Copyright © 2019 by Royal Chemical Society. (C) The structures of 233, 234, and 235 and the enzymatic reactions from 233 to 234 by ALP and from 234 to 235 by reductase. (D) The illustration of the induction of ER stress by the self-assembly of 233; Adapted from Ref.⁴⁸¹.

To selectively inhibit lung cancer, Yang et al. further developed their tandem self-assembly approach.⁴⁸¹ As shown in Figure 78C, they designed a peptide derivative, NBD-GFF_pYG-N=N-ERGD (233), which was the substrate of both ALP and reductase. According to the authors, ectoenzyme, ALP, converts 233 to NBD-GFFYG-N=N-ERGD (234). 234 self-assembles to form nanoparticles or short nanofibers for efficient endocytosis. After the mitochondrial accumulation of 234, the reductase in the mitochondrial membrane likely converts 234 to NBD-GFFYG-aniline (235), which further self-assembles into nanofibers in the mitochondria to disrupt the mitochondrial membrane. These events could result in cytochrome c (Cyt c) release and reactive oxygen species (ROS) production, which can increase ER stress, activate unfolded protein response (UPR), and lead to selective inhibition of lung cancer cells. Using the A549 lung cancer cell line, the authors demonstrated their concept in cell assays and in a mice model. This innovative study may provide a feasible strategy for developing ENS for treating lung cancer.

Yang et al. reported a rather less explored application of ENS, the removal of senescent cells.⁴⁸² They designed a substrate of β-galactosidase (GLB1), NBD-FFY(gal)G (236, Figure 79A), because GLB1 was overexpressed in senescent cells.^483–484 The authors incubated senescent HeLa cells with 236 and its control (NBD-FFYG, 237) and showed that the nanofibers of 237 formed in the cells treated by 236, not by 237 (Figure 79B). The formation of nanofibers reduces the expression of p53, p21, and p16INK4a in the cells, which is evidence of reducing endothelial cell senescence. According to the authors, the ENS of 236/237 is able to selectively remove senescent endothelial cells by inducing apoptosis. Since GLB1 overexpresses in certain cancer cells, this work illustrates a new way to remove senescent endothelial cells for cancer therapy.

Using ENS to control the conformation of peptides in supramolecular assemblies, Yang et al. demonstrated a simple way to modulate the activities of peptides in the assemblies.⁴⁸⁵ They designed and synthesized a peptide, Nap-FFGGYGSSSRRAPQT (45), consisting of the self-assembling motif (Nap-FF) and a peptide from insulin-like growth factor-I (GYGSSSRRAPQT, IGF-1). Folding into β-sheet conformation, 45, forming a hydrogel (Gel-A), almost preserves its ability to bind with the IGF-1 receptor (IGF-1R) It also has a K_d of 11.5 nM, which is comparable to that of IGF-1 (K_d = 4.3 nM). Without the self-assembling motif, the peptide (Nap-GGYGSSSRRAPQT, 238, Figure 79C) exhibits weak affinities to IGF-1R (K_d = 381 nM). With β-sheet conformation, 45, at 10 nM, efficiently activates IGF-1 downstream signaling and prevents cell apoptosis. In the hydrogel formed by heating-cooling cycle, 45 adopts a β-sheet conformation and is more effective than IGF-1 to improve ischemic hind-limb salvage in vivo. Enzymatic dephosphorylation of the phosphorylated peptide (Nap-FFGG_pYGSSSRRAPQT, 44) results in a gel (Gel-B), in which 45 adopts an α-helix conformation but is less effective than IGF-1 in vivo (Figure 79D). While this study provides a useful strategy to modulate peptide conformations in assemblies for mimicking protein functions, it, again, reveals that ENS differs fundamentally from self-assembly at thermodynamic equilibrium.

PET is commonly used in tumor imaging with high sensitivity. However, tumor-targeted PET probes that can distinguish and image specific tumors are underdeveloped. In a work reported by Liang et al.,⁴⁸⁶ a tumor targeting PET imaging probe CBT-⁶⁸Ga (239, Figure 80A) was designed. It utilizes a furin substrate peptide (RVRR) to target specific tumor cells overexpressing furin. After furin activation, the probe undergoes CBT-Cys condensation and intracellular self-assembly to form nanoparticles with average diameters of 258.3 nm. In vivo microPET imaging suggested the co-injection of CBT-⁶⁸Ga (239) with its cold analogue CBT-Ga (240) would lead to the formation of CBT-⁶⁸Ga/CBT-Ga nanoparticles inside the MDA-MB-468 tumors and have a 9.1-fold increase of the tumor/liver ratio comparing to that of the mice only injected with CBT-⁶⁸Ga (Figure 80B). This work is the first report of a furin-targeted ⁶⁸Ga radiotracer for enhanced tumor microPET imaging. Replacing the furin substrate with other enzyme-specific peptides may give us a hint in developing other sensitive microPET probes.

Cathepsin B (CTSB) is a lysosomal protease that functions in catabolic pathways after protein internalization. Aberrant expression of CTSB is a hallmark of certain cancer, such as esophageal cancer.⁴⁸⁷ In a study reported by Liang et al., a bioluminescence probe Val-Cit-AL (241, Figure 80C) is designed to detect CTSB both in vitro and in vivo.⁴⁸⁸ Upon activation, it has a 67-fold “turn-on” intensity with good linear relationship from 0–40 U/L. It has an excellent lower detection limit of 27 mU/L. Notably, the author reported that the ratio of turnover number (kcat)/Michaelis constant (Km) of CTSB for 241 was about 500-fold higher than that of a Gly-Phe-Leu-Gly (GFLG)-based nanoparticle probe.⁴⁸⁹ Likely benefited from its rapid enzymatic kinetics, this CTSB-specific bioluminescence probe is capable of cancer cell imaging, hardly being interfered by common intracellular substances such as cations, some tumor markers, and proteases. After injecting into MDA-MB-231 tumor-bearing nude mice, the probe can rapidly target tumors within 30 min (Figure 80D). This article is the first example of using bioluminescence to target CTSB, reaching good selectivity both in vitro and in vivo. One of the reasonable questions is the selectivity of 241 for other cathepsins. If 241 is exceptionally specific toward CTSB, it may lead to a useful assay for CTSB.

Liang et al. also applied the concept of “turn-on” nanoparticles for developing NIR nanoprobes based on ENS catalyzed by carboxylesterase (CES).⁴⁹⁰ Specifically, the authors designed H₂N-Cys(StBu)-Lys(Biotin)-Ser(Cy5.5)-CBT (NIR-CBT) (242, Figure 81A), which was a substrate of CES and a ligand of biotin receptors. The NIR-CBT probe is subjected to reduction-controlled condensation and self-assembles to form “OFF” NIR-CBT nanoparticles. After being recognized by biotin receptors overexpressed on tumor cells, intracellular CES catalyzes the cleavage and turns the Cy5.5 fluorescence signal “ON” (Figure 81B). After demonstrating that the nanoparticles of 242 exhibit a 69 fold increase of fluorescence once incubated with CES for 6 h, the authors incubated the nanoparticles of 242 with HepG2 cells. They reported that biotin receptor-assisted endocytosis and intracellular CES hydrolysis turned on the fluorescence. Moreover, these nanoparticles of 243 are able to image the HepG2 tumors, which express biotin receptors and CES, in a murine model.

Figure 81.

(A) The structures of 242 and 243. (B) The illustration of biotinylated NIR-CBT probe for tumor-targeted imaging. Adapted from Ref.⁴⁹⁰. (C) The illustration of GGT-triggered cleavage and GSH-instructed condensation to form 246-Dimers which self-assemble to form 246-NPs with enhanced T₂ MRI contrast. Adapted from Ref.⁴⁹¹. Copyright © 2019 by American Chemical Society.

To develop MRI contrasting agents, Liang et al. designed a “smart” Gd-based probe Glu-Cys(StBu)Lys(DOTA-Gd)-CBT (244, Figure 81C), which underwent ENS catalyzed by γ-glutamyltranspeptidase (GGT).⁴⁹¹ According to the authors, GGT cleaves glutamine to allow the intracellular CBT-Cys condensation reaction to form Gd nanoparticles of 246 for enhancing the T2-weighted MR contrast of a tumor in vivo at 9.4 T. One of the future challenges of this study is to prove that the MRI enhancement is significant when the magnetic field strength is at the clinical range.

To develop an approach for reducing tumor recurrence, Qiao and Wang recently reported the use of intracellular ENS from constructing a long-term drug depot.⁴⁹² The molecule, CPT-LVFFGFLG-PEG-RGD (247, Figure 82A), consists of a hydrophobic anticancer drug (camptothecin, CPT) at the N-terminal and a PEG-RGD at the C-terminal. CTSB recognizes GFLG sequence and catalyzes the cleavage of GF/LG to form CPT-LVFFGF and LG-PEG-RGD. The authors reported that 247 self-assembled to form nanoparticles, which transformed to nanofibers after being treated by CTSB. Based on in vitro and in vivo results, the authors proposed the mechanism of action of 247 (Figure 82B): The nanoparticles of 247 enter cancer cells, CTSB cleaves 247 to form nanofibers of CPT-LVFFGF, the nanofibers slowly dissociate and release CPT intracellularly. The authors reported that, after the resection surgery on mice model, the sustained release of CPT from the nanofiber depot in the remnant cancer cells prevented postsurgical local tumor recurrences. Probably due to the relatively high expression of CTSB in liver cells,⁴⁹³ the liver still accumulates more 247 than tumor sites. Therefore, more optimization of this substrate is required. In a related work, the authors used PEG to synthesize polymer–peptide conjugates (PPCs) that underwent an acid-induced intracellular self-assembly in the tumor microenvironment for enabling drug delivery deep inside the tumor.⁴⁹⁴

Figure 82.

(A) The structure of 247. (B) The illustration of self-assembly behavior of 247 and construction of fibrous drug depots in cells. Adapted from Ref.⁴⁹². Copyright © 2019 by Wiley Inc. (C) The structure of 248 with four motifs. (D) The illustration of CAF instructed nanofibers self-assembly in situ for enhanced tumor imaging. Adapted from Ref.⁴⁹⁵. Copyright © 2019 by Wiley Inc.

A recent study by Wang et al.⁴⁹⁵ has reported a new NIR probe (248, Figure 82C) that is activatable by a prolyl endopeptidase, fibroblast activation protein-α (FAP-α). The peptide (248) consists of four major parts: hydrophilic motifs, FAP substrate, self-assembly motif, and a cyanine dye. Molecule 248 shows no significant cytotoxicity when its concentration reaches 300 μM. Upon FAP cleaving GPA at the P/A site, the resulting peptide increases hydrophobicity and self-assembles to form nanofibers (Figure 82D). In vivo, when the FAP on the surface of cancer associated fibroblasts (CAFs) cleaves 248, the probe selectively accumulates around the tumor (with 4- and 5-fold higher intensity over liver and kidney, respectively). The enzymatic formation of nanofibers not only enables extended retention time (over 48h detectable windows) but also enhances the signal of the probe (5.5-fold signal enhancement). Similar to the earlier work by Chen et al.,⁴²⁴ this study is a successful example that combines ENS with clinically used cyanine dyes for functions.

Fan and Ren reported enhanced anticancer selectivity and efficiency when they combined ENS with a novel inhibitor of histone deacetylase.⁴⁹⁶ They designed a novel peptide, Nap-Gff_pYSV (249, Figure 83A), which is a substrate of ALP for ENS. The authors tested the cytotoxicity of 249 and the dephosphorylated product, Nap-GffYSV (250), on several cancer cell lines and a normal cell line (L929) and measured the secreted ALP by the cells (Figure 83B). One notable difference of this work in contrast to other reports is that 249 is pretreated by ALP prior to being incubated with the cells. Moreover, 249 is more potent than Nap-ff_py (119) for inhibiting HeLa cells, supporting that YSV acts as an inhibitor of histone deacetylase. Although YSV is claimed to act as an anticancer peptide, it hardly inhibits HeLa cells. It is possible that the self-assembly of 250 promotes endocytosis for inhibiting the HeLa cells. Thus, more analogs of 249 or more cell lines need to be tested for elucidating the mechanism of the cytotoxicity of 249 or 250.

Figure 83.

(A) The structures of Nap-G^DF^DFpYSV (249) and Nap-G^DF^DFYSV (250). (B) IC₅₀ values of 249 and the pre-treated 249 by ALP catalysis in vitro against different cells after incubation for 48 h. Adapted from Ref.⁴⁹⁶. Copyright © 2019 by Royal Chemical Society. (C) The structures of 251 and 252 and the ENS process generating supramolecular nanofibers and hydrogels with mannose decoration for cultured fibroblast cells and inhibiting E. coli. Adapted from Ref.⁴⁹⁷. Copyright © 2019 by Wiley Inc. (D) The structures of 253 and 254. (E) Illustration showing an ALP-activatable NIR fluorescence (FL)/MR bimodal probe for in vivo imaging and the biodistribution (% ID/g) of P-CyFF-Gd (red) or P-Cy-Gd (blue) in HeLa tumors and main organs (H: heart, Li: liver including gallbladder, Lu: lung, Sp: spleen, Ki: kidneys, St: stomach, In: intestines, T: tumor) at 4 h after i.p. injection into mice (0.015 mmol kg⁻¹ Gd³⁺). Adapted from Ref.⁴⁹⁸. Copyright © 2019 by American Chemical Society.

To mimic the multivalence exhibited by sugar moieties on glycoproteins, Tian and Li developed a glycopeptide for ENS.⁴⁹⁷ Attaching mannose at the serine, as shown in Figure 83C, the glycopeptide (252), can be phosphorylated at the tyrosine residue to give the phosphoglycopeptide, 251. Using ALP to dephosphorylate 251 generates the hydrogel of 252. According to the authors, the hydrogel of 252, bearing multiple mannose ligands on its self-assembled structure and engaging multivalent carbohydrate–lectin interactions, is able to bind E. coli with high specificity, which results in bacterial adhesion, membrane disruption, and subsequent bacterial death (Figure 83C). The authors also reported that an in vivo wound healing assay (in a full-thickness skin defect mouse model) showed this glycopeptide hydrogel promoting wound healing and preventing E. coli infection. This work illustrates that ENS, combined with specific saccharide–lectin interactions, is suitable for generating multivalent interactions to mimic glycoproteins.

To achieve simultaneously multimodality imaging, Ye et al. developed an impressive bimodal probe that was an ENS substrate for in vivo imaging.⁴⁹⁸ As shown in Figure 83D, the small molecule probe (P-Cy-FF-Gd, 253) consists of an ALP-activatable near-infrared (NIR) fluorophore, a self-assembling dipeptide, and a magnetic resonance (MR) contrasting agent. The authors reported that endogenous ALP, overexpressed on cancer cell membranes, removed the phosphate group on the fluorophore (Cy), resulting the dephosphorylated probe (Cy-FF-Gd, 254) to self-assemble into nanoparticles, as visualized by cryo-SEM. This ENS process, simultaneously enhanced NIR fluorescence over 70-fold and r1 relaxivity about 2-fold, and enabled real-time, high-sensitivity, high-spatial-resolution imaging of the ALP activity of tumor cells in a mice model. One notable result is that, compared to the control molecule (P-Cy-Gd) without the self-assembling motif, 253 significantly enhances the fluorescent and MRI signals from the HeLa tumors in mice. It is also impressive that 253 results in higher MRI signal in the tumor than in the liver of the mice model (Figure 83E). This work illustrates the merit of ENS for designing other activatable multimodal probes to image the activity and locations of enzymes in vivo and in real time.

To develop a self-delivery system consisting of lonidamine (LND) for selective inhibition of cancer cells, Zhong et al. synthesized a substrate (255, Figure 84A) of phosphatase.⁴⁹⁹ 255, consisting of lonidamine and GFF_pY, turns into 256 (or LND-GFFY) after dephosphorylation. After confirming that 255 is a substrate of ENS catalyzed by ALP, the authors showed that cancer cells uptake more 255 than normal cells do. Moreover, 255 inhibits tumor growth in a HeLa xenograft murine model. The authors suggested this approach should be generally applicable to the delivery of other hydrophobic cancer drugs.

Figure 84.

(A) The structures of 255 and 256. (B) The structures of 257 and 258. (C) Self-assembly of Olsa-RVRR (259) into Olsa-NPs (i.e., nanoparticles of 261) through a series of steps. Red line indicates the site of furin cleavage, and the circled hydroxyl group indicates the exchangeable hydroxyl proton that provides OlsaCEST signal at 9.8 ppm from the water frequency. (D) After 259 enters the cytoplasm of high furin-expressing cells (HCT116 cells in this study), it undergoes reduction by GSH and cleavage of the peptide by furin near the Golgi complex where 260 is generated. Amphiphilic oligomers (mostly dimers) are then formed from the click reaction between two 260 molecules, followed by self-assembly into 261-NPs as a result of intermolecular π-π stacking. The intracellular accumulation of 261-NPs then serves as a reservoir of Olsa molecule-enhancing CEST contrast and can inhibit DNA methylation for tumor therapy. Adapted from Ref.⁵⁰². Copyright © 2019 by Springer Nature.

To detect the high expression of furin in live cells, Zhou and Yi developed a relatively simple peptide substrate (RVRRFFF-NBD (257), Figure 84B) of furin.⁵⁰⁰ 257 self-assembles to form a β-sheet nanostructure that is capable of a rapid and specific response to furin in only 5 min in aqueous solution because of the existence of the RVRR motif in 257. The nanostructures of 257 thus can selectively distinguish high furin-expressing cancer cells, like MDA-MB-231 cells, a kind of human breast cancer cell, from normal cells. According to the authors, the nanoparticles of 257 self-assembles and enters cells. The furin in the MDA-MB-231 cells catalytically cleaves the 257 to generate FFF-NBD (258), which is fluorescent. The authors reported that the assemblies 257 was able to stay in live cells for a long time and were capable of durable detection of intracellular furin. Since the intensity of the fluorescence in the cells treated by 257 correlates well with the concentrations of 257, it is likely that 258 also self-assembles inside cells after being cleaved by furin.

The use of furin for intracellular ENS^280,501 has received more attention for developing theranostics, as recently reported by Bulte et al. on chemical exchange saturation transfer magnetic resonance (CEST) imaging.⁵⁰² To enhance the tumor retention of imaging agents for CEST, the authors developed a peptide conjugate (259) that connected olsalazine on the side chain of lysine,⁵⁰³ RVRR at the N-terminal, and 2-cyanobenzothiazole^280,282,284 at the C-terminal. Based on the condensation reaction reported by Rao et al.,²⁸⁰ furin cleaves 259 to generate 260, which forms a dimer of olsalazine (261) in the presence of GSH. 261 self-assembles to form nanoparticles. According to the authors, both 259 and the nanoparticles of 261 are readily detected with CEST due to the exchangeable hydroxyl protons on olsalazine. Using murine xenografts, the authors showed that the CEST signal from olsalazine and anti-tumor therapeutic effect increased 6.5- and 5.2-folds, respectively. Although the concentration of the probe is too high to be clinically useful, this work illustrated that intracellular ENS is an attractive approach for developing MRI based theranostic agents.

The substrates for ENS can be almost any type molecules that self-assemble. Recently, Zhang et al. reported a polyaromatic molecule to selectively alter cell permeability for inducing cell death.⁵⁰⁴ As shown in Figure 85A, the authors, inspired by pore forming toxin, designed and synthesized a substrate (262) of ALP. 262, as nontoxic water-soluble molecular precursors, distributes in extracellular fluid. In the case of cancer cells, ALPs, being overexpressed on the cancer cells, catalytically dephosphorylate 262 to 263. The aromatic motifs in 263, as hydrophobic building blocks, promote the self-assembly of 263, which results in rigid aggregates to insert into the hydrocarbon core of membrane. The insertion of the aggregates of 263 leads to membrane permeability alteration by blocking the transportation of molecules, eventually causing the death of the cancer cells (Figure 85B). This work is the first example that uses the expanded π-conjugation system with an extremely high rigidity and long linear length for ENS-guided membrane insertion. Since there are many rigid molecular motifs known to be able to insert into cell membranes to form pores or channels,⁵⁰⁵ this strategy may eventually lead to cancer cell death without causing drug resistance.

Figure 85.

(A) The structures of 262 and 263. (B) The illustration of ENS-directed membrane insertion of a long and rigid synthetic analogue causing permeability alteration. Adapted from Ref.⁵⁰⁴. Copyright © 2019 by American Chemical Society. (C) The structures of 264 and 266 and their dephosphorylated products. (D) The fluorescence image of HeLa cells incubated with a mixture of 264 and 266 (1 : 1, total concentration 100 μM, 12 h) without and with ALP inhibition, and co-stained with CellMask Red. Adapted from Ref.⁵⁰⁶. Copyright © 2019 by Royal Chemical Society.

While most of studies use one of the enantiomers of the peptide-based ENS substrates to interact with cells, Zhang et al. reported that the co-administration of the two stereoisomers of ALP substrates in cancer cells led to molecular assembly targeting of both the plasma membrane and the lysosomes.⁵⁰⁶ As shown in Figure 85C, they synthesized homochiral-peptide-based boron diketonate complexes (264 and 266) by coupling a boron diketone fluorophore with a pair of homochiral peptides that bear a taurine group at the C-terminals. 264 and 266, being the substrates of ALP, turn into 265 and 267, respectively, upon dephosphorylation. After confirming that 265 or 267 self-assembles to form nanofibers, the authors used it to incubate HeLa cells and found that the assemblies of 265 mainly localized at the plasma membrane and the assemblies 267 mainly in the lysosome. Then the authors showed that co-administration of the two stereoisomers in cancer cells also led to molecular assembly at both organelles, but inhibition of the ALP prevented the assemblies on or in the cells (Figure 85D). The dual-targeted-assembly may generate a synergistic anticancer effect, as suggested by the authors.

Wang et al. recently developed an ingenious application of the ENS of 1 to deliver enzymes into cells for anticancer therapy. Neutrophil-dependent cell inactivation relies neutrophil lysosomes to produce cytotoxic reactive oxygen species (ROS) by a cascade of enzymatic reactions. Based on this fact, he authors chose to load superoxide dismutase (SOD) and chloroperoxidase (CPO) in the hydrogels formed on magnetic nanoparticles (MNPs) as a multifunctional SOD/CPO-loaded nanogel system (SCNG).⁵⁰⁷ As shown in Figure 86A, SOD catalyzes the dismutation of ˙O₂ˉ to form H₂O₂, which serves as a substrate of CPO to produce HOCl. The reaction of HOCl and H₂O₂ generates singlet oxygen (¹O₂) for cell destruction. To build a suitable carrier to entrap CPO and SOD, the authors attached ALP on MNPs and added 1, which produced hydrogel nanoparticles with about several hundred nanometers in diameter. The nanogels, acting as a matrix to protect enzymes from proteolytic deactivation, are suitable for loading SOD and CPO to form the SCNG. The relatively higher level of ROS in tumor microenvironment thus enables the SCNG to generate ¹O₂ to inhibit tumors (Figure 86B). The authors used intratumoral injection of the SCNG to demonstrate the functions of the SCNG. Although the dosage of the SCNG remains high, the use of a simple ENS molecule (1) to enable a sophisticated enzyme cascade is stimulating. The authors also proposed a highly promising concept, enzyme dynamic therapy (EDT), for taking full advantage of redox enzymatic reactions in the tumor microenvironment to treat cancer by ¹O₂. The success of this approach likely depends on the kinetics of ¹O₂ formation by SCNG. In fact, Wang et al. already made progress on enhancing the production of H₂O₂ and ¹O₂ in cancer cells.⁵⁰⁸ Specifically, they combined magnetic hyperthermia with enzyme catalysis by using an alternating magnetic field (AMF) to heat up the MNP@Nanogels for generating H₂O₂ and the MNP-CPO@Nanogels for producing ¹O₂. They named such a construction magnetocaloric–enzymatic tandem therapy (METT). As suggested by the authors, the programmed alternating magnetic field (AMF), similar to the neutrophil activator, elevates H₂O₂ levels in cancer cells, and the CPO within the protective peptide nanolayer converts the H₂O₂ into ¹O₂ in a sustained manner. As a proof of concept, the authors confirmed that both the H₂O₂ and ¹O₂ in cancer cells increase stepwise under a programmed alternating magnetic field (Figure 86C). The authors also reported the effective inhibition of cancer cells in vitro and suppression of tumor growth in animal models. Although the in vivo studies were done via local administrations, this work represents an innovative combination of physical and biochemical approaches for anticancer therapy, which will likely stimulate more research activities along this direction.

Figure 86.

The scheme of the responsive EDT mechanism and preparation of SCNGs. (A) The cascade SOD/CPO-mediated therapy includes the catalysis of ˙O₂ˉ into H₂O₂ by SOD, conversion of both as-obtained H₂O₂ and endogenous H₂O₂ species into final ¹O₂ species with CPO. (B) The fabrication process of SCNGs involves (1) modification of MNPs for attaching the ALP trigger, (2) dephosphorylation of hydrophilic peptide precursors to hydrophobic hydrogelators by the ALP on the MNPs, (3) ALP-triggered self-assembling between the hydrophobic portions and hydrophilic parts of hydrogelators through π–π stacking and electrostatic interactions around MNP, (4) immobilization of cascade SOD and CPO and further a safe and effective EDT in tumor microenvironment. The responsive SCNGs can effectively convert the endogenous ROS (˙O₂ˉ and H₂O₂) into highly reactive ¹O₂ by the cascade reactions of SOD and CPO in tumor region, which subsequently cause cancer cell death. Adapted from Ref.⁵⁰⁷. (C) Relative fluorescence intensity of correlated with the levels of H₂O₂ and ¹O₂ in U87 cells exposed to MNP@Nanogels and MNP-CPO@Nanogels, respectively, with programmed MH. Adapted from Ref.⁵⁰⁸. Copyright © 2019 by Wiley Inc.

Although nuclear accumulation would greatly enhance the efficacy of anticancer drugs, it remains a challenge to enhance nucleus targeting. Yang et al. recently used conformation control by ENS to enhance cellular uptake and nuclear accumulation.⁵⁰⁹ They designed and synthesized four peptides (Figure 87A), NBD-βA-FF_pYGTSFAEYWNLLSP (268) NBDβA-FFYGTSFAEYWNLLSP (269), HCPT-FF_pYGTSFAEYWNLLSP (270), and HCPT-FFYGTSFAEYWNLLSP (271). The sequence, TSFAEYWNLLSP (PMI), is capable of binding with the MDM2 and MDMX in the cell nucleus for activating the p53 gene. The authors tuned the peptide conformations by heating-cooling or ENS. They found that the assemblies formed by ENS at 4 °C showed enhanced cellular uptake and nuclear accumulation (Figure 87B). Impressively, against HepG2, A549 and U87MG cells, the IC₅₀ values of 271 formed by ENS at 37 °C are 0.66, 1.43 and 1.94 μM, respectively, and the IC₅₀ values of 271 formed by ENS at 4 °C are 0.22, 0.26 and 0.87 μM, respectively. Moreover, 271 formed by ENS at 4 °C exhibits the highest in vivo activity. This study, taking advantage of HCPT, a highly potent drug candidate, illustrates a powerful way for modulating the emergent properties of peptide-based supramolecular nanomedicine to enhance efficacy in cancer therapy.

Figure 87.

(A) The structures of 268, 269, 270, and 271. (B) Cellular uptake of the different assemblies into HepG2 cells at 1 h and 4 h (concentration of peptide = 200 μM). Adapted from Ref.⁵⁰⁹. Copyright © 2020 by Elsevier Inc. (C) The structures of 272 and its ENS product 273. Adapted from Ref.⁵¹⁰. Copyright © 2020 by American Chemical Society.

To develop probes for image-guided surgery, Wang et al. recently reported a peptide-based probe for imaging renal cell carcinoma (RCC).⁵¹⁰ As shown in Figure 86C, the peptide, RGDRDDRDDPLGYLGFFC(Cy) (272), consists of a targeting motif, an enzyme cleavable site, and a NIR fluorophore. Specifically, the modular molecular design includes (i) RGD, as a recognition motif, for recognizing the highly expressed αvβ3 integrins in RCC, (ii) PLGYLG, as an enzyme-responsive peptide linker and a substrate to be cleaved by MMP-2/9, (iii) a self-assembly motif (YLGFFC), and (iv) a fluorophore (Cy). According to the design by the authors, the peptide binds to the integrins overexpressed on the cancer cells, and MMP2/9 enzymes overexpressed by the cancer cells cleave the peptide to release the self-assembling peptide attached with the cyanine dye to form fluorescent nanoparticles on the surface of cancer cells. After confirming the in situ enzyme triggered self-assembly of the NIR peptide probes on cancer cells, the authors tested the probes on tumor lesions in a mice model. The authors have shown that the nanofibers formed by the self-assembly of the probes, exhibiting an excretion-retarded effect in the kidney, enabled identifying tiny lesions for complete tumor removal, and significantly reduced the postoperative recurrence of tumors compared with traditional surgery. Moreover, using an ex vivo kidney perfusion model, they also demonstrated the tumor-specific excretion-retarded (TER) effect. Although the detailed enzyme kinetics remain to be elucidated, this work illustrates the promises of the concept of ENS in developing imaging probes.

To target castration-resistant prostate cancer (CRPC) cells, a small D-phosphopeptide (274) has been developed to undergo prostatic acid phosphatase (PAP) catalyzed ENS to inhibit prostate cancer cells.⁵¹¹ As shown in Figure 88A, while dephosphorylating 274 by PAP forms uniform nanofibers that inhibit VCaP, a CRPC cell, a non-hydrolysable phosphate analogue, 276, is ineffective for inhibiting VCap. Although the efficacy of 274 remains to be improved, this work confirms that PAP-catalyzed ENS is critical for selective inhibition of CRPC cells.

Figure 88.

The structures of (A) 274, 275, and 276 and (B) 277 and 278.

Although protein kinases are the most attractive targets in drug discovery, it is rather difficult to use protein kinase to enable ENS for targeting cancer cells. Recently, Gao et al. reported innovative progress on utilizing protein kinase A (PKA) to design PKA-triggered supramolecular assemblies with anticancer activities.⁵¹² They grafted a suitable peptide to PNIPAM to increase the lower critical solution temperature (LCST) of the polymer (277, Figure 88B) to above body temperature. Upon phosphorylation by PKA, the resulting polymer (278) exhibited a critical temperature below body temperature to result in the PKA-triggered supramolecular assembly. They demonstrated that the PKA-triggered assembly occurred selectively in PKA-upregulated MCF-7 cells, which could be used to sensitize tumors for Dox in vivo. This PKA-catalyzed supramolecular assembly would likely lead to a new strategy for combating kinase-upregulated cancer, especially in the case of drug resistance to kinase inhibitors.

Because ENS builds up non-diffusive molecular assemblies, it would increase the local concentration of the desired molecules for further reactions, as shown by the innovative combination of ENS and biorthogonal reactions⁵¹³ demonstrated by Rao et al.⁵¹⁴ To image the activity of enzyme in tissues, the authors further developed target-enabled in situ ligand aggregation, a powerful platform technology that they developed for molecular imaging.⁵¹⁵ As shown in Figure 89A, the key molecule is 279, consists of a caspase cleave site, a protected aminothiol group, an aromatic nitriles, and a trans-cyclooctene (TCO). Such a design allows 279 to undergo two bioorthogonal reactions—the condensation reaction of aromatic nitriles and aminothiols and the inverse-electron demand Diels–Alder reaction between tetrazine and trans-cyclooctene (TCO). Specifically, caspase 3/7 cleaves the DEVD sequence off 279 to expose the N-terminal amine, and GSH deprotects the thiol group. These two reactions lead to intramolecular cyclization to form 280. Being more hydrophobic than 279, 280 self-assembles to form nanoaggregates. Bearing the TCO group, 280 is able to couple tetrazine on a fluorescent substrate (281) via the intermolecular click reaction. After validating this design in vitro, the authors demonstrated the application of this approach in vivo using a mice xenograft with tumors. According to the procedure illustrated in Figure 89B, 24 h after the intratumoral injection of cisplatin to induce cell death, 279 was injected intravenously into the mice, followed by intravenous injection of 281 after 30 min. The authors found significant retention of fluorescent signal in the cisplatin-treated mouse group compared to the untreated group. These results support the versatility of the ENS process for coupling with multiple bioorthogonal reactions in vivo. The concentration of 279 used in the study is still high, probably because the TCO is buried inside the nanoaggregates. Structural elucidation of the nanoaggregates may help address this limitation in the future.

Figure 89.

(A) The structures of the substrate of caspase-3/7 and the oxidant of GSH (279, the cleavage site shown by a red dotted line), the intramolecular cyclization product (280), the tetrazine (281), and the product (282) of the click reaction. (B) The illustration of pre-targeted imaging of enzyme activity based on target-enabled in situ ligand aggregation. Adapted from Ref.⁵¹⁴. Copyright © 2020 by Wiley Inc.

Gao et al. recently devised an innovative indirect ENS process for reducing methicillin-resistant staphylococcus epidermidis (MRSE) infection.⁵¹⁶ Taking the cues from neutrophil extracellular traps (NETs), which stick to bacteria and activate the sudden increase of reactive oxygen species (ROS), the authors developed a molecular mimic of NETs. The authors synthesized a quinazolinone derivative (BQA-GGFF, 283, Figure 90A) with a typical aryl boronate immolative linker,^517–518 which underwent oxidation to yield BQH-GGFF (284). 284 self-assembles to form nanofibers and results in a hydrogel. Upon the oxidation of the hydrogen peroxide produced by glucose oxidase (Gox) and glucose, 283 turns to 284, thus resulting in hydrogelation. After confirming that the nanofibers of 284 within the hydrogel tightly stuck to the bacteria and inhibited bacterial growth in vitro, the authors tested 283 at the bacterial infection site on mice. The authors reported that inflammatory condition (i.e., high level of ROS including hydrogen peroxide) at the infection site induced the formation of fluorescent assemblies of 284. The nanofibers of 284 were able to trap the bacteria, thus lessening bacterial translocation and increasing the overall mice survival ratio (Figure 90B). The use of inflammation to enable ENS for forming in situ supramolecular assemblies would likely have broad applications beyond infection control, especially if this approach is able to reduce inflammation induced tissue injury.

Figure 90.

(A) The structure of 283 and the indirect enzymatic oxidation, by glucose oxidase, to generate the self-assembling molecule, 284. (B) The use of the indirect ENS to form in situ artificial NETs to halt MRSE infection. Adapted from Ref.⁵¹⁶. Copyright © 2020 by Elsevier Inc.

One of the remaining challenges for intracellular ENS is to visualize the formation of supramolecular nanofibers in live cells. Gao and coworkers recently made considerable progress by combining direct stochastic optical reconstruction microscopy (dSTORM) imaging with ALP catalyzed ENS in cancer cells.⁵¹⁹ The authors, aiming to elucidate the dynamic assemblies formed by ALP catalyzed ENS, designed and synthesized two molecules 285 and 287 (Figure 91A). As an assembly precursor, 285 is able to undergo ENS upon ALP catalyzed dephosphorylation to generate 286. Bearing a tetrazine group, 285 reacts with a TCO-modified Cy5 dye to form 287, which is a fluorescent assembly precursor and a substrate of ALP. Using super-resolution fluorescence microscopy, the authors confirmed that ALP catalyzed ENS resulted in the nanofibers of the co-assembly of 286/288 (Figure 91B). After treating HeLa and Saos2 cells with 285 and 287, the authors fixed the cells and used dSTROM to image the nanofibers with the resolution below 50 nm (Figure 91C). Moreover, the authors reported that supramolecular assemblies behaved differently in HeLa and in Saos2 cells. In HeLa cells, the supramolecular nanofibers appear to be transcytosis: they form inside cells, but are excreted from the HeLa cells. In Saos2 cells, the supramolecular assemblies form first on the cell surface with granular morphology, then transform into nanofibers and accumulate in cells, which eventually induces Saos2 cell death via ROS upregulation. Considering that HeLa and Saos2 express different isozymes of ALP and at different levels and locations, these results are reasonable. The dynamics of the D-enantiomers of 285 and 287 also remains to be determined. Nevertheless, this excellent work illustrates the need to develop a method to capture the dynamic morphology of the intracellular molecular assemblies formed by ENS. Furthermore, itshows fluorescent imaging as a methodology to identify the morphology of other classes of ENS assemblies for correlating dynamics of assembly and bioactivities of ENS. Moreover, the observation of exocytosis in HeLa cells, in this work, is very intriguing. Similar phenomena appears in the case of using ENS for imaging lipid rafts.⁵²⁰ These results suggest that ENS may promote exosome secretion, which remains to be confirmed.

Figure 91.

(A) The structures of the substrates (285 and 287) of ALP and the corresponding dephosphorylation products (286 and 288). (B) The illustration of the co-assembly of 286 and 288 to form fluorescent nanofibers after ALP catalyzed ENS. (C) dSTORM images of the assemblies in fixed HeLa and Saos2 cells pre-incubated with 500 μM of 285 and 0.05 μM of 287. Adapted from Ref.⁵¹⁹. Copyright © 2020 by American Chemical Society.

While using the self-assembly⁵²¹ of synthetic molecules to form peptide filaments has been successful in vitro^522–525 or extracellularly,^267,275,426 the generation of artificial intracellular filaments has not been possible. The challenge largely arises from dynamics and crowding of the intracellular environment,⁵²⁶ which makes it difficult to carry out self-assembly, a thermodynamic equilibrium process, of synthetic molecules inside cells. Thus, it is reasonable to use ENS to generate intracellular filaments. Although increased number of reports indicate that it is feasible to use enzymatic reaction to control self-assembly of small molecules inside cells,^{227,253,519,527} the formation of bona fide artificial intracellular filaments is only a recent result (Figure 92).⁵²⁸ The self-assembly of a phosphorylated and trimethylated heterochiral tetrapeptide (289, Figure 92A) forms nanoparticles. Enzymatic dephosphorylation reduces repulsive intermolecular electrostatic interactions and converts the nanoparticles of 289 into two types of filaments (types 1 and 2) of 290. Cryo-EM reveals that type 1 filaments exhibit distinct types of cross-β structures with either C7 or C2 symmetries (Figure 92B), with the hydrophilic C-terminal residues at the periphery of the helix (Figure 92C). After the Saos2 cells uptake the nanoparticles, intracellular ALP converts the nanoparticles of 289 to the filaments of 290. Due to the intracellular crowdedness, the peptide filaments form bundles that extend from the plasma membrane to nuclear membrane (Figure 92D). Electron tomography confirms the intracellular filaments of 290 forming bundles, which hardly interact with endogenous components, including cytoskeletons (Figure 92E). Structural analogs of 289 indicate that stereochemistry and multiple post-translational modification (PTM) of peptides, that is, phosphorylation and trimethylation, are critical for generating the intracellular bundles. This work, as the first case of artificial intracellular filament, illustrates the power of ENS and provides molecular insights for understanding normal and aberrant intracellular filaments. This work also raises other interesting questions, such as the mechanisms of the endocytic pathway and endosomal escape of the nanoparticles, which remain to be elucidated. In addition, intracellular filaments only form from the heterochiral tetrapeptide, which demands more exploration of the molecular space of the precursors so that a general design principle can be formulated.

Figure 92.

(A) The molecular structures of a phosphorylated and trimethylated tetrapeptide (289) and its dephosphorylated product (290). (B) 3D reconstruction of type 1 filaments of 290 from cryo-EM images and the atomic model of the type 1 filaments with cross-β structure. (C) The chemical structures of one layer of the filament at the cross section of (B). (D) CLSM images of Saos2 cells treated with 289 (200 μM) for 24 h. Scale bars = 10 μm. (E) TEM image of a treated Saos2 cell (289, 200 μM, 24h) at two regions of the cells. Scale bar = 500 nm. Adapted from Ref.⁵²⁸.

4.3. Subcellular ENS

The presence of enzymes in subcellular organelles should allow the use of ENS for targeting specific organelles, which is much less explored despite the promising implications for biomedicine. An early report on using ENS for targeting mitochondria⁵²⁹ actually combines ENS and a well-established mitochondria targeting motif (triphenyl phosphonium (TPP)⁵³⁰). As shown in Figure 93A, attaching TPP to a phosphotetrapeptide produces the precursors (291 or 293) that self-assemble form noncovalent oligomers. 291 (or 293), being incubated with Saos2 cells, undergoes dephosphorylation catalyzed by ALP overexpressed on the cells to form 292 (or 294), which further self-assembles to result in large aggregates on the surface of Saos2 cells to facilitate the endocytosis of 292 (or 294). The assemblies of 292 (or 294), escape from the lysosome, and target mitochondria because of TPP. The accumulation of 292 (or 294) on mitochondria induces dysfunction of mitochondria to release cytochrome c, and results in cell death (Figure 93B). One notable result from this work is that Saos2 is unable to evolve acquired drug resistance to 291 (or 293) (Figure 93C), likely due to the assemblies of 292 (or 294) targeting mitochondria. This study, though still relies on pericellular ENS, highlights the feasibility and underscores the importance of using ENS to target subcellular organelles.

Figure 93.

(A) The structures of 291, 293, and their dephosphorylated products. (B) The illustration of ENS for targeting mitochondria and inducing death of cancer cells. (C) Cell viability of unstimulated Saos2 cell line or selected Saos2 cell line (after 5 weeks treatment of the precursors with gradually increasing concentrations) incubated with 291 or 293 at different concentrations for 48 h. Adapted from Ref.⁵²⁹. Copyright © 2016 by American Chemical Society.

In fact, it is possible to target mitochondria by perimitochondrial ENS, as shown in a recent serendipitous observation.³⁶⁷ As shown in Figure 94, a branched peptide (295), consisting of a well-established protein tag (i.e., FLAG-tag: DYKDDDDK)³⁷⁰ as the branch, Nap-ff as the self-assembling motif,^531–533 and NBD as the fluorophore,²⁵³ is a substrate of enterokinase (ENTK).³⁶⁹ 295 self-assembles in water to form micelles, which turn into nanofibers after ENTK cuts off the FLAG motif (DDDDK). After fluorescent microscopy of HeLa cells reveals that the product (296) of ENTK cleavage localizes mainly at mitochondria, TEM of the isolated mitochondria confirms the conversion of the micelles to nanofibers at the mitochondria. This result is the first case of ENS on mitochondria by a synthetic molecule. In addition, the mitochondrial ENS is able to deliver molecular cargos (proteins or small molecules) to mitochondria of cancer cells selectively. This work, indicating ENTK as a perimitochondrial enzyme in cancer cells, illustrates a fundamentally new way for targeting mitochondria by enzymatic reactions. In fact, a recent study reveals that this kind of Flag-branched peptide also enables the trafficking of histone protein (H2B), a nuclear protein, to the mitochondria in cancer cells, likely resulted from the ENTK catalyzed proteolysis of 98.⁵³⁴

Figure 94.

(A) The structures of the branched peptide 295 and its ENTK-cleaved product 296. (B) ENTK cleaving the branch to convert the micelles to nanofibers on mitochondria for targeting mitochondria. (C) TEM images of mitochondria isolated from HeLa cells being incubated with 295 (200 μM) for 24 h. Scale bar equals 100 nm. Adapted from Ref.³⁶⁷. Copyright © 2018 by American Chemical Society.

The perimitochondrial proteolysis catalyzed by enterokinase (ENTK/TMPRSS15) of cancer cells allows the use of ENS for selectively targeting mitochondria of cancer cells, as demonstrated by delivering chloramphenicol (CLRP) into the mitochondria of cancer cells, especially liver cancer cells.⁵³⁵ Liver cells detoxify CLRP via glucuronidation⁵³⁶ catalyzed by a cytosolic enzyme, but mitochondria lack the enzyme for glucuronidation. Because CLRP inhibits protein synthesis, delivering the CLRP into the mitochondria of liver cancer cells should inhibit mitochondrial protein synthesis, thus leading to selective death of cancer cells. Such perimitochondrial ENS has delivered CLRP into the mitochondria of HepG2 and HeLa cells, and reduced protein synthesis and ATP production of the cancer cells.⁵³⁵ As shown in Figure 95, the conjugation of the Flag-tag³⁷⁰ to 2-amino-dibutylhexadecanediamide makes a peptide-lipid conjugate Flag-(C16)₂ (297), which forms micelles to encapsulate CLRP. The micelles selectively enter cancer cells and accumulate at the mitochondria to release CLRP into the mitochondria of cancer cells. As expected, CLRP inside mitochondria inhibits mitochondrial protein synthesis and decreases cancer cell proliferation. Moreover, after the enzymatic cleavage of 297, the resulting lipids (298) also promote mitochondrial outer membrane permeabilization (MOMP) for apoptosis.⁵³⁷ Thus, MOMP and the inhibition of mitochondrial protein synthesis synergistically reduces the viability of cancer cells (Figure 95). Most importantly, this selective targeting of mitochondria increases the efficacy of inhibiting liver cancer cells without increasing the toxicity to normal liver cells.⁵³⁵ This work, as the first example of ENS to repurpose clinically approved ribosome inhibitors, may provide a new way to interrupt the metabolism of cancer cells for cancer treatment.

Figure 95.

The multiple functions of the ENS of 297 in cancer cells for delivering chloroamphenicol (CLRP) into mitochondria to inhibit protein synthesis and for changing permeability of mitochondria membrane: (i) perimitochondrial accumulation, (ii) mitochondria-targeting drug delivery, and (iii) mitochondrial outer membrane permeabilization (MOMP). Adapted from Ref.⁵³⁵. Copyright © 2020 by American Chemical Society.

Endoplasmic reticulum (ER), being the largest subcellular organelle, performs a wide range of functions in eukaryotic cells.^538–539 Although disrupting ER functions is emerging as a new strategy for anticancer therapies,^540–542 selectively targeting the ER of cancer cells is less explored. ENS offers a new strategy for targeting the ER, as reported recently.⁵⁴³ In that study, a substrate of phosphatase, 299, bearing L-homoarginine at the C-terminal, is able to undergo relatively slow dephosphorylation on cancer cells to generate 300 (Figure 96A). The mixture of 299 and 300 self-assembles to form crescent-shaped supramolecular assemblies, which interact with the cell membrane to cause curvatures and disrupt cell membranes. The crescent-shaped assemblies enter cells and accumulate at the ER, a location enriched with protein tyrosine phosphatases (PTP1B²⁵³). The further dephosphorylation at the ER promotes the self-assembly of 300, causes ER stress and subsequently activates the caspase signaling cascade for cell death (Figure 96B). Because the ALP overexpressed by cancer cells results in the formation of the crescent-shaped assemblies, 299 selectively targets the ER of cancer cells to exhibit potent activities for inhibiting cancer cells (e.g., HeLa, A2780cis, and OVSAHO), but not normal cells (e.g., HS-5) (Figure 96C). This work illustrates an approach for in situ generation of membrane binding scaffolds to disrupt the cell membrane and target ER of cancer cells.

Figure 96.

(A) The structures of 299 and 300. (B) The illustration of ENS of peptide assemblies to disrupt the cell membrane and to target the ER to cause cell death. (B) IC₅₀ (24 h) of 299 or 300 against HeLa cells, A2780cis cells, OVSAHO cells, and HS-5 cells. Adapted from Ref.⁵⁴³. Copyright © 2018 by American Chemical Society.

The enrichment of PTP1B at the ER also provides a way to use ENS to generate liquid-like condensates for sequestrating enzymes to the ER, as shown in Figure 97.⁵²⁷ In order to sequestrate enzymes to this target organelle, naproxen (a nonsteroidal anti-inflammatory drug (NSAID) and a ligand of cyclooxygenase-2 (COX-2)) replaces the naphthyl group in 299 to generate a phosphopeptide (301) as a substrate of phosphatases. 301, acting as the precursor of a hydrogelator (302), slowly undergoes dephosphorylation catalyzed by phosphatases to form supramolecular assemblies on the ER. The assemblies, containing 301 and 302 molecules, are able to bind both COX-2 and PTP1B, thus sequestrating them on the ER. The examination of the structural analogs of 301 suggests that the NSAID motif, the phosphotyrosine, and the enzymatic dephosphorylation of 301 are necessary for the enzyme sequestration. This work, for the first time, illustrates the use of subcellular ENS for associating enzymes in cells and provides biochemical insights for understanding intracellular liquid condensates.

Figure 97.

(A) The structures of 301 and 302. (B) The illustration of ENS for intracellular sequestration of PTP1B and COX-2. (C) CLSM images of Saos2 cells treated with 301 (12.5 μM) for 1 h and then stained with antibodies of PTP1B (red) and COX-2 (green) (scale bar = 10 μm). Adapted from Ref.⁵²⁷. Copyright © 2018 by American Chemical Society.

Recently, Zhang reported an innovative way to target the ER of cancer cells by esterase catalyzed ENS.⁵⁴⁴ Based on massive accumulation of alcohol derivatives around the ER to cause ER stress and cell death, the authors designed and synthesized a N-hydroxylethyl peptide (303, Figure 98A) for cytosolic self-assembly. They conjugated Nap-FF with ethanolamine to generate the alcohol derivative (145). Attaching the fluorescent coumarin derivative (304) to the hydroxyl group of 145 produces the esterase substrate 303. They also made two control molecules, 305 and 306. In the presence of CES, 303 was hydrolyzed to form 145 and 304. While 305 hydrolyzes faster than 303, 306 is resistant to hydrolysis catalyzed by esterase. The authors found that HeLa cells cultured with 303 or 305 exhibited strong fluorescence from coumarin in the cytosol, but not in the HeLa cells incubated with 306 (Figure 98B). Notably, cell viability assays on ovarian cancer (SKOV3 and drug resistant OVCAR-3), liver cancer (HepG2), and metastatic pancreatic cancer (PANC-1) reveals potent anticancer activity of 303, especially with the inhibition shown within 24 h (Figure 98C). If the liver cytotoxicity of 303 could be reduced, this approach might lead to an effective prodrug for ER targeting of cancer cells.

Figure 98.

(A) The structures of 303, 304, 305, and 306. (B) The fluorescent images of HeLa cells incubated with 303, 305, or 306 at a concentration of 10 μM for 4 hours and stained with CellMask Green (plasma membrane). Violet represents the emissions of coumarin derivatives. The scale bar represents 20 μm. (C) SKOV-3, OVCAR-3, HepG2 and PANC-1 cell viability under treatment with 303 at various concentrations. Adapted from Ref.⁵⁴⁴. Copyright © 2019 by Royal Chemical Society.

Another type of enzyme for targeting the ER is proteases, as shown by an unexpected observation that trypsin-1 (PRSS1) catalyzes ENS of a branched peptide on the ER of cancer cells.⁵⁴⁵ As shown in Figure 99A, the branched peptide (307), consisting of a D-tetrapeptide backbone and a branch with the sequence of KYDKKKKDG, turns to be a substrate of PRSS1. That is, PRSS1 catalyzes the cleavage of 307 to produce 308 and 309, which self-assemble to form nanofibers (Figure 99B). Based on the cancer cell line encyclopedia (CCLE), the authors incubated OVSAHO, a high-grade serous ovarian cancer cell line, with 307 because OVSAHO overexpresses PRSS1. After using cell assays to confirm that 307 inhibited the proliferation of OVSAHO cells, but not the control cells expressing normal or low level PRSS1, the authors used immunostaining and fluorescent imaging of a fluorescent analog of 307 to reveal the overexpression PRSS1 on the ER of OVSAHO (Figure 99C). This work, illustrating the inhibition of cancer cells by ENS on ER for the first time, offers a new way to target the ER of cancer cells because many enzymes are synthesized at the ER. However, it remains to be determined whether the enzyme activity of PRSS1 at the ER is an aberrant feature of cancer cells.

Figure 99.

(A) The structures of the branched peptide (307), its proteolysis products (308 and 309), and its fluorescent analog (310). (B) TEM images of 307 (200 μM) before and after the addition of PRSS1. Scale bars = 100 nm. (C) Fluorescent images of OVSAHO cells treated with 310 (400 μM) for 2 hours. Cells were stained with ER-Tracker™ red (for ER) and Hoechst 33342 (for nucleus). Scale bars: 10 μm. (B, C) Adapted from Ref.⁵⁴⁵. Copyright © 2020 by Wiley Inc.

One unique advantage of ENS is that it can serve as a context-dependent signal for targeting the ER of cancer cells, as shown in the case of ENS generating the assemblies from the antagonist (AVPI) of inhibitors of apoptotic proteins (IAPs).⁵⁴⁶ Since ALP presents on the surface of certain cancer cells, but not normal cells,⁴²⁶ ALP on the cell surface acts as a context-dependent signal for increasing the efficacy of BTZ⁴⁵⁸ against cancer cells and reducing the known side effect of BTZ (toxicity to bone marrow stromal cells (e.g., HS-5)). As shown in Figure 100A, conjugating AVPI, a binding motif against IAPs,⁵⁴⁷ to the ℇ-NH₂ of lysine in a phosphotyrosine-bearing peptide generates a branched proapoptotic peptide (311), which self-assembles to form micelles. After ALP catalyzed dephosphorylation converts 311 to 312, the micelles turn into nanofibers. When the micelles of 311 (co-incubated with BTZ) enter cancer cells via the caveolin-dependent endocytosis, 311 undergoes dephosphorylation to form the nanofibers of 312 at the ER. Thus, the assemblies of 312, in the cancer cells, sequestrate the IAPs so that the cancer cells become more sensitive to BTZ and commit apoptosis (Figure 100B&C). In the case of HS-5 cells, the micelles of 311 and BTZ enter HS-5 cells via macropinocytosis, but localize in the lysosome of HS-5. Thus, BTZ is unable to reach its target (proteasomes) so the toxicity of BTZ decreases (Figure 100B). Therefore, the phosphopeptide not only forms assemblies to deactivate the IAPs in cancer cells, but also rescues normal cells (e.g. HS-5) from the side effects of BTZ. Incubating 313, the fluorescent analog of 311, with Saos2, HeLa, and HS-5 cells shows that the assemblies of 314, after ENS, indeed localizes at the ER of the cancer cells (i.e., Saos2 and HeLa), but in the lysosomes of the normal cells (i.e., HS-5) (Figure 100D). This work illustrates that ENS, as a context-dependent process, controls endocytosis for selectively targeting the ER of cancer cells. Although it is conceivable that ALP catalyzes the formation of nanofibers to facilitate the endosomal escape of the peptide assemblies, the quantitative role of ENS for endosomal escape remains to be elucidated.

Figure 100.

(A) The molecular structures of the AVPI-bearing phosphopeptides (311 and 313) and the corresponding peptides (312 and 314). (B) The illustration of enzymatically-forming peptide assemblies to sequestrate IAPs for selectively killing cancer cells. (C) Sequestration of AVPI boosting the activity of BTZ for inhibiting cancer cells. (D) Fluorescent images of Saos2, HeLa, and HS-5 cells incubated with 313 (50 μM, 4 h). For Saos2 and HeLa cells, the ER was stained by ER tracker; for HS-5 cells, the lysosome was stained by Lyso tracker. Adapted from Ref.⁵⁴⁶. Copyright © 2020 by Wiley Inc.

5. Conclusion and Outlook

Bioinspiration.

The discussions on the ENS of nature have illustrated that ENS is, at least, one of the universal controlling mechanisms of intermolecular interactions in molecular ensembles for cellular functions. That is, ENS, integrating these two non-genetic features, enzymatic reaction and self-assembly, of cells, represents an overlooked organizing principle of proteins (and other molecules) in cells. Self-assembly, clearly, is a non-genetic feature of cells. Though enzymes are genetically controlled, enzymatic reactions are beyond genetic control. Considering that the cell is a complicated heterogeneous mixture, enzymatic reactions increase the number of molecular components and confer context-dependent specificities. Also, self-assembly provides a facile way for modulating the quantitative aspects of signaling transduction, such as location, duration, thresholds, and amplitudes. Most importantly, self-assembly creates higher-order assemblies, in specific locations, to exhibit emergent properties and functions that are absent in individual molecules. Apparently, the most efficient way to control or organize the higher-order assemblies, being selected by evolution, is enzymatic reaction. From this viewpoint, ENS, arguably, is one of the rules of life. One the other hand, as shown by many examples in nature, enzymes trigger hierarchical self-assembly to create large biological entities like microtubules, biomolecular machines and so on. This feature can likely serve as a general guide for the uniqueness of ENS or supramolecular synthesis. Thus, from the perspective of biomimetics, it is highly attractive to use man-made molecules to explore the principles and applications of ENS. The exploration of ENS of man-made molecules in the last decade has validated the notion that it is feasible to design functions from molecular processes such as ENS. Although the use of enzymes to control the emergent properties of man-made molecules is in its infancy, the promises and opportunities are abundant. The advancement of cell biology is an unlimited source of inspiration for the development of ENS.

Kinetic Analysis.

Compared to molecular self-assembly at thermodynamic equilibrium, which sometimes lead to heterogeneous assemblies (even with one kind of building block), ENS usually produces homogeneous, self-limiting, monodispersed, or more-ordered structures, even when the building blocks or reaction environments are heterogeneous.^57,226 This apparently counterintuitive observation, resembling the formation of Turning patterns by the Belousov–Zhabotinsky reaction,⁵⁴⁸ underscores the pivotal role of kinetics regulated by enzymatic reaction. It would be instructive if such a phenomena could be examined in terms of reaction-diffusion³⁹ by rigorous kinetic and thermodynamic analysis.^{465,549–550} To do that, more detailed enzyme kinetics should be examined in the context of self-assembly, which is still a rather less explored area.⁵⁵¹ One important characteristic of ENS is ultrafast (comparing to crystallization) in which reaction and assembly can complete in minutes or seconds. This feature is particularly useful for functions of molecular assemblies, especially in cellular environment. It is likely that close collaboration between molecular scientists and applied mathematicians would provide more insights for the development of ENS.

Structures, imaging, and modeling.

To understand the emergent properties of the assemblies formed by ENS, it is necessary to elucidate the structures of the assemblies. The rapid advancement of cryo-EM will likely be able to address the needs of structure elucidation.^552–556 However, molecular assemblies in cells are neither static nor crystalline. The understanding of the biological functions of ENS requires further development of new biophysical tools that provide the information of enzyme reaction in real time and in live cells. To extract such dynamic information over a large area and relatively long time, correlated light and electron microscopy and whole cell NMR or a combination of these can probably shine more light for molecular understanding. Obviously, any tools that can provide dynamic information of ENS⁵¹⁹ would be useful. But such kinds of tools remain to be developed. The major challenge is that one has to examine enzymatic and spatiotemporal control of the supramolecular assemblies of proteins or synthetic molecules over multiple length scales (e.g., subcellular, intracellular, or intercellular space) and extended time scales (e.g., milliseconds, seconds, and hours) for elucidating biological functions of ENS. Intriguingly, properly designed ENS-based fluorescent probes, at certain conditions, are able to achieve high resolution imaging over large area and extended time.⁵²⁰ In addition, to fill the experimental gaps, molecular dynamic simulation^{528,557–561} would be highly valuable for providing insights that help understand these dynamic processes,⁵⁶² especially with the increase of computational power and the advancement of computational methods.

Integration with other processes.

While most of the reports on ENS of man-made molecules have focused on the formation of assemblies, the reverse process, disassembly, is equally important for biological functions, as revealed by enzymatically regulated assembly and disassembly of proteins in cellular environment. Combining assembly and disassembly for designing ENS approaches remains challenging, but the success in this direction may lead to new therapeutics that achieve spatial control and precise regulation for treating diseases. For example, the combination of ENS with disassembly has greatly increased the therapeutic windows for targeting tumors.^439,479 With more understanding of biological processes, it is likely that more elaborated, enzyme-controlled assembly and disassembly^563–565 will emerge for a wide range of applications. In fact, regulation of disassembly has already attracted considerable research attention in the field of controlled drug release.^566–567 Moreover, coupling ENS and disassembly would be an effective approach to mimic the cellular signal transduction cascades with feedback loops.⁵⁶⁸ In addition, it would be fruitful to introduce other local changes (e.g., ligand–receptor interactions,^328,569 coordination,^570–571 pH-response,⁴⁹⁴ ionic interaction,⁵⁷² redox reactions,^{518,573–576} bioorthogonal reactions^{513,575,577–578}, and dynamic covalent bonds^579–583) into the substrates of ENS for designing sophisticated molecular systems that can control emergent properties of molecular assemblies and modulate cellular functions.

Molecules and molecular processes.

While its essence is to use reactions to control assemblies, ENS has been explored under other names, such as enzyme-responsive assembly, biocatalytic self-assembly, enzyme-instructed self-assembly, enzyme-assisted self-assembly, or directed assembly, which are familiar to chemists. In fact, some of the studies may unknowingly involve enzyme reactions to generate assemblies, such as the synthetic polymers that use peptides as pendants,⁵⁸⁴ or may be helped by ENS, such as the attempt to carry out enzymatic polymerization⁵⁸⁵ in vivo.⁵⁸⁶ The molecular building blocks used for exploring ENS are largely peptides because the bioactivities of peptides and considerable advances in the self-assembly of peptides.^{384,587–590} Other building blocks, such as small organic molecules,^591–594 lipids,⁵⁹⁵ polymers,⁵⁹⁶ proteins,^27,597–598 nanoparticles,⁵⁹⁹ or drug candidates,^600–601 for self-assembly should be useful as the substrates for ENS as well. The most important advantage of ENS is to confer functions to molecular processes. This feature, functions by kinetics or by rate control,⁴²⁹ is a feature of life. Thus, further advancing ENS research will impact both basic and translational biomedical research, including understanding cell biology and addressing the challenges in oncology, neuroscience, and infectious diseases. Having the ability to make (almost) any sophisticated molecule and the understanding of noncovalent interactions, chemists are well-equipped to develop molecular assemblies with emergent properties and to explore the functions of molecular processes by ENS. Since nature uses ENS for functions, ENS is a universal theme for understanding the functions and locations of molecular ensembles in cellular environments. For example, from the perspective of ENS, it would be easy to understand liquid-liquid phase separation²⁹ in cells and to design approaches for controlling those biochemical phenomena, though they appear to be biophysical. On the other hand, the exploration of enzyme-controlled noncovalent synthesis in the context of cells will open a new paradigm of supramolecular assemblies in cell biology. Thus, providing a necessary path for functions,⁶⁰² such as controlling cell fate, understanding diseases, and developing therapeutics. Certainly, many challenges remain. The ENS of man-made molecules, compared to that of natural molecules, is still relatively primitive. But the path for functions clearly centers on chemistry.^603–604 The interdisciplinary collaboration of chemists, biochemists, molecular biologists, physicists, physiologists, structural biologists, clinicians as well as engineers will likely meet those challenges and make exciting discoveries in the future research of controlling molecular assemblies by reactions.

Figure 4.

Formation of aggresomes by ENS involving multiple enzymes and proteins.⁶⁶ Ub, E1, E2, E3 and HDAC6 are ubiquitin, ubiquitin-activating enzyme, ubiquitin-conjugating enzyme, ubiquitin ligase, and histone deacetylase, respectively.

Figure 19.

(A) The projection image of MAVS filaments calculated from a 2.7 nm slice out of the cryo-ET reconstruction as viewed from the outer surface of the filaments. Yellow arrows point to portions of multiple filaments where apparent helical stripes were resolved. Adapted from Ref. 161. (B) The inactive retinoic acid-inducible gene-I protein (RIG-I) has phosphorylation at its caspase activation and recruitment domains (CARDs) and carboxy-terminal domains (CTDs) and adopts a ‘closed’ auto-inhibited conformation. Following the binding of RNA, an enzymatic reaction cascade results in the self-assembly of MAVS. The MAVS filaments activate TBK1 or IκB kinase-ε (IKKε) as well as the IKKα–IKKβ–IKKγ complex. Adapted from Ref. 160. Copyright © 2016 by Springer Nature.

Figure 25.

Sorting of pre-rRNA at the interface of FC and DFC. Adapted from Ref. 192. Copyright © 2019 by Elsevier Inc.

Figure 62.

(A) The structures of 159 and 160. (B) Proposed mechanism of hydrogelator precursor 159 that undergoes ALP-based ENS to form nanofibers of 160, followed by GSH-associated intracellular condensation and self-assembly to form nanofibers of 161. Adapted from Ref.⁴³⁰. Copyright © 2016 by American Chemical Society.

Abbreviations

3D

three-dimensional

AAK1

AP2-associated protein kinase 1

ABCD3

ATP-binding cassette sub-family D member 3

ABPs

actin binding proteins

AchE

acetylcholinesterase

ACP

acid phosphatase

ADAM

disintegrin and metalloproteinase domain-containing protein

ADP

adenosine diphosphate

Ake

adenylate kinase

Akt

protein kinase B

ALP

alkaline phosphatase

ALPL

tissue non-specific alkaline phosphatase

AMP

adenosine monophosphate

AP2

adaptor protein 2

AP2M1

AP2 complex subunit mu

Ap5A

diadenosine-5-pentaphosphate

ARL2

ADP-ribosylation factor-like protein 2

ASC

the adaptor molecule apoptosis-associated speck-like protein containing a CARD

ATG4B

autophagy-related protein 4 homolog B

Atg11

Autophagy-related protein 11

Atg30

autophagy-related protein 30

Atg36

autophagy-related protein 36

ATL

atlastin

ATP

adenosine triphosphate

AURKB

aurora B kinase

BAY

BAY 11–7085

Bla

β-lactamase

BLP

bacillus licheniformis protease

BMP-1

bone morphogenetic protein 1

BOD1

bi-orientation of chromosomes in cell division 1

BRCC3

Lys-63-specific deubiquitinase BRCC36

BTZ

bortezomib

BUBR1

BUB1-related kinase 1

CAFs

cancer associated fibroblasts

CARD

caspase activation and recruitment domain

CBs

Cajal bodies

CBT

cyanobenzothiazole

CBX5

chromobox protein homolog 5

CCK2R

cholecystokinin-2 receptor

CCP

clathrin-coated pit

Cdc28

Cdk1 protein kinase

CD

chromodomain

CD95

tumor necrosis factor receptor superfamily member 6

CDDP

cisplatin

CDK1

cyclin–cyclin-dependent kinase1

CES

carboxylesterases

CEST

chemical exchange saturation transfer magnetic resonance

CFTR

cystic fibrosis transmembrane conductance regulator

CIL

contact inhibition of locomotion

CITK

citron Rho-interacting kinase

CK

casein kinase-2

CLSM

confocal laser scanning microscopy

CME

clathrin-mediated endocytosis

Cnn

centrosomin

COX-2

cyclooxygenase-2

CPEB1

cytoplasmic polyadenylation element-binding protein 1

CPK

space-filling calotte

CPT

camptothecin

CRB

chlorambucil

CRPC

castration-resistant prostate cancer

CSD

chromoshadowdomain

nano-CT

nano-computed tomography

CTD

carboxy-terminal domain

CTE

C-terminal extension

CTSB

cathepsin B

CTP

cytidine triphosphate

CTPS

cytidine triphosphate synthase

ecCTPS

E. coli cytidine triphosphate synthase

hCTPS1

human cytidine triphosphate synthase 1

Cyt c

cytochrome c

DAPI

4′,6-diamidino-2-phenylindole

DC

dendritic cell

DCP1a

mRNA-decapping enzyme 1A

Dex

dexamethasone

DFC

dense fibrillar component

DFO

deferoxamine

DMEM

DNA

deoxyribonucleic acid

mtDNA

mitochondria DNA

DNS

dansyl chloride

DOTA

1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraacetic acid

Dox

doxorubicin

DR4/5

death receptor 4/5

dSTORM

direct stochastic optical reconstruction microscopy

E1

ubiquitin-activating enzyme

E2

ubiquitin-conjugating enzyme

E3

ubiquitin ligase

ECM

extracellular matrix

ECS

enzymatic covalent synthesis

EDC4

mRNA-decapping protein 4

EGFR

epidermal growth factor receptor

EIF4ENIF1

eukaryotic translation initiation factor 4E transporter

EISA

enzyme-instructed self-assembly

cryo-EM

cryo-electron microscopy

ENS

enzymatic noncovalent synthesis

ENTK

ER

endoplasmic reticulum

ERK

mitogen-activated protein kinase

ESCRT-I

endosomal sorting complex I required for transport

ESCRT-II

endosomal sorting complex II required for transport

ESCRT-III

endosomal sorting complex III required for transport

ELYS

embryonic large molecule derived from yolk sac

F-actin

filamented actin

FA

Focal adhesion

FAK

focal adhesion kinase

FAP-α

fibroblast activation protein-α

FBL

fibrillarin

FBS

fetal bovine serum

FC

fibrillar center

FITM

fat storage-inducing transmembrane protein

Fmoc

N-(fluorenyl-methoxycarbonyl)

FLAG

Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys

FRAP

fluorescence recovery after photobleaching

FRET

fluorescence resonance energy transfer

G-actin

globular actin

GC

granular component

GDP

guanosine diphosphate

GFP

green fluorescent protein

GGT

γ-glutamyltranspeptidase

GLB1

β-galactosidase

GRASP65

Golgi peripheral membrane protein p65

GLU1

glucose transporter

GSH

glutathione

GTP

guanosine triphosphate

H2B

histone protein 2B

HADC6

histone deacetylase

HCPT

10-hydroxy camptothecin

HER2

Receptor tyrosine-protein kinase erbB-2

HP1α

heterochromatin protein 1 homolog alpha

HP1

heterochromatin protein 1

Hrr25

casein kinase I homolog HRR25

ICG

indocyanine green

IF

Intermediate filament

cIF

cytoplasmic intermediate filament

IFB-1

intermediate filament protein B1

IGF-1

insulin-like growth factor-I

IGF-1R

IGF-1 receptor

IKKε

IκB kinase-ε

IL 1

intermediate layer 1

IL 2

intermediate layer 2

IMPDH

inosine-5’-monophosphate dehydrogenase

INM

inner nuclear membrane

ITAM

immunoreceptor tyrosine-based activation motif

ITIM

immunoreceptor tyrosine based inhibitory motif

JNK

c-Jun N-terminal kinase

KIF2B

kinesin family member 2B

LCST

lower critical solution temperature

LD

lipid droplet

LEASA

localized enzyme-assisted self-assembly

LPP

lambda protein phosphatase

LRP6

low-density lipoprotein receptor-related protein 6

LSPR

localized surface plasmon resonance

L-STM

liquid-phase scanning tunneling microscopy

MAP

microtubule-associate protein

MAPK

mitogen-activated protein kinase

MKLP1/KIF23

mitotic kinesin-like protein 1

MAVS

mitochondrial antiviral-signaling protein

MDR

multidrug-resistance

MEKK3

Mitogen-activated protein kinase kinase kinase 3

MEL28

maternal effect lethal protein 28

MM

midbody matrix

MMP-2

matrix metalloproteinase-2

MMP-7

matrix metalloproteinase-7

MMP-9

MPA

mycophenolic acid

MRI

magnetic resonance imaging

MRLC

myosins regulatory light-chain

Mps1

monopolar spindle protein 1

MT

microtubule

cMT

cytoplasmic microtubules

k–MT

kinetochore–microtubule

nMT

nuclear microtubule

MLKL

mixed lineage kinase domain-like protein

MTOC

microtubule-organizing center

MTT

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide

MYPT1

protein phosphatase 1 regulatory subunit 12A

NBD

nitrobenzoxadiazole

NBR1

next to BRCA1 gene 1 protein

ND10

nuclear domain 10

NE

nuclear envelope

NECAP

adaptin ear-binding coat-associated protein 1

NF-κB

nuclear factor kappa-light-chain-enhancer of activated B cells

NIR

near-infrared

NLRP3

NACHT, LRR and PYD domains-containing protein 3

NMR

nuclear magnetic resonance

NPs

nanoparticles

NPC

nuclear pore complexes

NPM

nucleophosmin

NPM-N

N-terminal domain of NPM

NS

nuclear speckles

NSAID

nonsteroidal anti-inflammatory drug

NTE

N-terminal extension

Nup153

nuclear pore complex protein Nup153

OGT

O-GlcNAc transferase

Olsa

olsalazine

ONM

outer nuclear membrane

OVA

ovalbumin

p14arf

p14ARF/p16INK4a fusion protein

p16INK4a

cyclin-dependent kinase 4 inhibitor A

p21

cyclin-dependent kinase inhibitor 1

p42ERK

p42 extracellular signal-regulated kinase

p53

tumor protein p53

PA

photoacoustic

PAP

prostatic acid phosphatase

P-bodies

processing bodies

PBS

phosphate-buffered saline

PCM

pericentriolar material

PDI

perylene diimide

PDT

photodynamic therapy

PE

primary human pancreatic epithelial cells

PEG

polyethylene glycol

PEO

polyethylene glycol

PET

positron emission tomography

PEX5

peroxin-5

Pex14

peroxin-14

sPGP

supramolecular phosphoglycopeptide

PHF

paired helical filament

PI3K

phosphatidylinositol 4,5-bisphosphate 3-kinase

PIDA

photo-induced disassembly

PKA

protein kinase A

PKD

protein kinase D

Pkh1/2

3-phosphoinositide-dependent protein kinase 1/2

PLK1

polo-like kinase 1

PML

promyelocytic leukemia protein

Pol I

POL polyprotein I

Polo

serine/threonine-protein kinase polo

PP1

protein phosphatase 1

PP1α

serine/threonine-protein phosphatase PP1-alpha

PP1γ

serine/threonine-protein phosphatase PP1-gamma

PP2A

serine/threonine-protein phosphatase 2A

PPI

protein-protein interaction

PpiX

protoporphyrin IX

PRC1

protein regulator of cytokinesis 1

iPSCs

induced pluripotent stem cells

PtdIns(x,y)Pn

phosphatidylinositol x,y-n-phosphate

PTM

posttranslational modification

PTP1B

protein tyrosine phosphatase

PTT

photothermal therapy

PYD

pyrin domain

QD

quantum dot

RBCC

N-terminal RING finger/B-box/coiled coil

RER

rough endoplasmic reticulum

RGD

Arg-Gly-Asp

Rheb

Ras homolog enriched in brain

RhoA

Ras homolog family member A

RhoA-GTP

Rho family of GTPases

RhoA-GAP

RhoA GTPase-accelerating proteins

Rho-GEF

Rho guanin exchange factor

RIG-I

retinoic acid-inducible gene 1 protein

RIPK1

receptor-interacting serine/threonine-protein kinase 1

RIPK3

receptor-interacting serine/threonine-protein kinase 1

mRNA

messenger RNA

sRNP

small nuclear ribonucleoprotein

ROK

Rho-associated protein kinase

ROMP

ring-opening metathesis polymerization

ROMPISA

ROMP induced self-assembly

ROS

reactive oxygen species

RR

rods and ring

RS

arginine/serine repeat

SEM

scanning electron microscopy

SER

smooth endoplasmic reticulum

Sey1P

SF

straight filaments

SH2

Src homology 2

SIM

SUMO–SUMO interacting motif

SMN

survival motor neuron protein

SMOC

supramolecular organization centers

SNP

silica nanoparticle

SNX9

sorting nexin-9

Spc42

spindle pole body component SPC42

Spd-2

spindle-defective protein 2

SPB

spindle pole bodies

SQSTM1

sequestosome-1

SR

serine rich

Src

proto-oncogene tyrosine-protein kinase Src

SRPK1

SRSF protein kinase 1

SRSF1

serine/arginine-rich splicing factor 1

SVs

synaptic vesicles

TAG

triacylglycerol

TAM

TAM (Tyro3, Axl, and Mer) kinase

TBK1

TANK-binding kinase 1

TCO

trans-cyclooctene

TEM

transmission electron microscopy

TER

tumor-specific excretion-retarded

TGase

transglutaminase

TKO

triple knockout

TMPRSS

transmembrane serine proteases

TNBC

triple negative breast cancer

TNFα

tumor necrosis factor alpha

TNFR1/2

tumor necrosis factor receptor 1/2

mTORC1

serine/threonine-protein kinase mTOR

TPP

triphenyl phosphinium

TRAIL

tumor necrosis factor related apoptosis inducing ligand

TRAILR

tumor necrosis factor related apoptosis inducing ligand receptor

TRIM25

E3 ubiquitin/ISG15 ligase TRIM25

Ub

ubiquitin

UBC9

SUMO-conjugating enzyme UBC9

ULF

unit length filament

UPR

unfolded protein response

USP36

ubiquitin carboxyl-terminal hydrolase 36

Vps27

vacuolar protein sorting-associated protein 27

VRK1

Serine/threonine-protein kinase VRK1

WRAP53

telomerase Cajal body protein 1

Yop1p

yeast homolog of the polyposis locus protein 1

ZO

zonula occludens

Biographies

Biography of the authors:

Hongjian He obtained his bachelor’s degree from Tongji University, Shanghai, China, in 2015. He joined Brandeis University as a graduate student in 2015. He is currently in his fifth year as a graduate student in the Department of Chemistry with Prof. Bing Xu at Brandeis University. His current works focus on the design of molecular self-assembly for mitochondria targeting.

Weiyi Tan obtained his B.S. degree in the College of Chemistry, Chemical Engineering and Material Science at Soochow University in 2018. He is currently in his second year as a graduate student in chemistry supervised under Prof. Bing Xu at Brandeis University. His current research interest lies in designing self-assembling materials for biological applications.

Jiaqi Guo received her B.S. degree in chemical biology from Xiamen University in 2018. In the same year, she joined the laboratory of Prof. Bing Xu at Brandeis University where she is now a third-year graduate student focusing on peptide assemblies and their emergent properties in cells.

Meihui Yi obtained her B.S. degree in the School of Chemistry and Materials Science at Nanjing Normal University in 2018. During her undergraduate study, she investigated mesoporous silica nanomotor. Now she is a third-year graduate student in Prof. Bing Xu’s lab at Brandeis University. Her research interest focuses on designing peptide assemblies for cancer therapy.

Adrianna N. Shy obtained her B.S. degree in chemistry from Hampton University in 2017. She was admitted to Brandeis University that same year where she joined the laboratory of Prof. Bing Xu. Currently, she is a fourth year PhD candidate where she focuses on the self-assembly of small peptides for various applications.

Bing Xu, after receiving his BS and MS from Nanjing University in 1987 and 1990, respectively, obtained his PhD under the supervision of Prof. Timothy Swager in 1996 from the University of Pennsylvania. Before starting his independent research at the Hong Kong University of Science and Technology (HKUST) on the Aug. 2000, he was an NIH postdoctoral fellow in the Whitesides lab at Harvard University. He was tenured as an associated professor in Jan 2006 and became a full professor in July 2008 at HKUST before moving to Brandeis. Based on his works at HKUST and Brandeis, Bing Xu is identified on the “Highly-Cited Researchers 2014 in Chemistry” list. He has made contributions to metallogels, biofunctional magnetic nanoparticles, self-delivery drugs, enzyme-responsive materials, and supramolecular hydrogels. His current research focuses on enzymatic noncovalent synthesis in materials, biology, and medicine.