Poly(peptide): Synthesis, Structure, and Function of Peptide–Polymer Amphiphiles and Protein-like Polymers

CONSPECTUS:

In this Account, we describe the organization of functional peptides as densely arrayed side chains on polymer scaffolds which we introduce as a new class of material called poly(peptide). We describe two general classes of poly(peptide): (1) Peptide–Polymer Amphiphiles (PPAs), which consist of block copolymers with a dense grouping of peptides arrayed as the side chains of the hydrophilic block and connected to a hydrophobic block that drives micelle assembly, and (2) Protein-like Polymers (PLPs), wherein peptide-brush polymers are composed from monomers, each containing a peptide side chain. Peptides organized in this manner imbue polymers or polymeric nanoparticles with a range of functional qualities inherent to their specific sequence. Therefore, polymers or nanoparticles otherwise lacking bioactivity or responsiveness to stimuli, once linked to a peptide of choice, can now bind proteins, enter cells and tissues, have controlled and switchable biodistribution patterns, and be enzyme substrates (e.g., for kinases, phosphatases, proteases). Indeed, where peptide substrates are incorporated, kinetically or thermodynamically driven morphological transitions can be enzymatically induced in the polymeric material. Synergistically, the polymer enforces changes in peptide activity and function by virtue of packing and constraining the peptide. The scaffold can protect peptides from proteolysis, change the pharmacokinetic profile of an intravenously injected peptide, increase the cellular uptake of an otherwise cell impermeable therapeutic peptide, or change peptide substrate activity entirely. Moreover, in addition to the sequence-controlled peptides (generated by solid phase synthesis), the polymer can carry its own sequence-dependent information, especially through living polymerization strategies allowing well-defined blocks and terminal labels (e.g., dyes, contrast agents, charged moieties). Hence, the two elements, peptide and polymer, cooperate to yield materials with unique function and properties quite apart from each alone. Herein, we describe the development of synthetic strategies for accessing these classes of biomolecule polymer conjugates. We discuss the utility of poly(peptide)-based materials in a range of biomedical applications, including imaging of diseased tissues (myocardial infarction and cancer), delivering small molecule drugs to tumors with high specificity, imparting cell permeability to otherwise impermeable peptides, protecting bioactive peptides from proteolysis in harsh conditions (e.g., stomach acid and whole blood), and transporting proteins into traditionally difficult-to-transfect cell types, including stem cells. Poly(peptide) materials offer new properties to both the constituent peptides and to the polymers, which can be tuned by the design of the oligopeptide sequence, degree of polymerization, peptide arrangement on the polymer backbone, and polymer backbone chemistry. These properties establish this approach as valuable for the development of peptides as medicines and materials in a range of settings.

Graphical Abstract

1. INTRODUCTION

Conjugates of biomacromolecules with synthetic polymers are a diverse and broad class of materials. They have been investigated in a range of applications and can take myriad forms. One prominent example comes from the field of protein drug delivery, where polymer–protein conjugates serve to extend the half-life following systemic delivery.^1–4 Generally, biological molecules can be conjugated with polymers for the generation of materials that have the capability of recognizing other biomolecules of interest such as cell-surface receptors,^5–9 or in sensing applications.^10–12 Ideally, these conjugates allow for the generation of materials that can overcome traditional biological barriers, yet maintain biological function.^13–17 Multivalent display of biomolecules conjugated to a polymeric core are increasingly found in the synthetic literature^18–23 and in natural systems. Indeed, glycoprotein structures, containing brushes of oligosaccharides, are an example of naturally occurring brush-type polymer structures where multivalent display imparts control over protein folding, enzymatic recognition, and stability.²⁴ In this Account, we will focus on peptide–polymer bioconjugates of two types, wherein each polymers display multivalent peptides: (1) Peptide–Polymer Amphiphiles (PPAs), which consist of block copolymers with a dense grouping of peptides arrayed as the side chains of a hydrophilic block, connected to a hydrophobic block to yield micelles that preserve specific peptide-directed reactions and interactions, and (2) Protein-like Polymers (PLPs) containing peptide-modified monomers. We describe these materials as “poly(peptide)”, where a multivalent display of peptides is arrayed along a polymer backbone, rather than the classical linear arrangement found in nature’s polypeptides. We will describe synthetic approaches to these peptide–polymer bioconjugates and discuss their resulting properties and utility in biomedical applications from in vivo imaging to therapeutic delivery and tissue targeting.

Peptide–Polymer Amphiphiles (PPAs)

The design of PPAs was inspired by small molecule peptide amphiphiles (PAs) and the remarkable array of functional assemblies they yield, as pioneered by Stupp^25,26 and Tirrell.^27,28 We reasoned that peptides with specific sequences could be used to drive the assembly of micelles, where morphology would be defined by the peptide sequence and how it then interacts with proteins and enzymes, coupled with the nature of the hydrophobic block (Figure 1). The resulting structures would then change their morphology in response to biological recognition events that alter the nature of the amphiphiles, either through binding or through enzymatic action at the surface of the assemblies. The arrangement of peptides in this manner, as the versatile responsive element facilitating many of the desired biological functions of these materials (vide infra), was inspired by the work of Tsien on activatable cell-penetrating peptides (ACPPs) that contain a substrate sequence for inflammation-associated enzymes found in diseased tissues.^29–31 By employing polymers instead of small molecule PA-based surfactants, multiple functional groups can be incorporated into the core of the particles, as well as the termini, including small molecule drugs and/or fluorophores.^32–35 The resulting PPAs allow for the preparation of richly functionalized materials, with high stability and exceptionally low critical aggregation concentrations. Furthermore, the polymer backbone provides a tunable handle on assembly characteristics and morphology transitions by adjusting T_g.³⁶

Figure 1.

Block lengths and chemical structure of amphiphilic block copolymers define the type of morphology they form when assembled in selective solvents. Switching between phases can be caused by stimuli that alter the nature of the polymer amphiphilicity. For peptide polymer amphiphiles (PPAs), stimuli include enzymatic reactions that are peptide sequence specific. (a) For a given hydrophilic block, the volume fraction of the hydrophobic block determines the average mean curvature and therefore, the morphology. (b) Hydrophilic polymer blocks containing peptide sequences (amino acids depicted as colored spheres) interact with specific enzymes according to the nature of the substrate. Here, we show an example of how enzymatic action can cleave the peptide side chain, leading to an increase in the volume fraction of the hydrophobic block and a concomitant change in morphology.⁷⁰ Similarly, other types of enzymatic reactions such as phosphorylation have been used to increase the hydrophilicity of the peptide-block resulting in phase transitions from larger lamellar architectures to higher surface curvature spheres.⁷⁰

We have prepared PPAs via two different synthetic approaches, involving either the conjugation of peptides to a polymer (graft-to) or via the direct incorporation of peptides on a growing polymer chain (graft-through) (Figure 2). We note that there are many elegant examples of peptide–polymer and protein–polymer conjugates wherein polymerizations are initiated from biomolecules in graft-from reactions, or through the usually inefficient conjugation reactions of the two macromolecules.^37–42 This approach produces either 1:1 biomolecule-to-polymer conjugates or biomolecules with multiple polymers attached (vide supra for practical applications of these materials). In this Account, we will focus on systems where peptides are introduced as side chains on a polymer backbone. For the PPA system described here, the peptides comprise the hydrophilic portion of a block copolymer amphiphile that assembles in aqueous solution to yield spherical nanoparticles. These can be accessed in a two-step method, wherein the block copolymers are first dissolved in a common solvent for both blocks, followed by a switch to a selective, aqueous solvent for the peptide block. Alternatively, we have prepared PPA-based micelles by polymerization-induced self-assembly (PISA),^36,43–48 a one-pot method for generating micellar morphologies.⁴³

Figure 2.

Synthetic approaches to generating PPAs. (a) Graft-to strategy: Conjugatable monomers are polymerized and then functionalized with peptides following polymerization. (b) Graft-through strategy: Peptides are directly incorporated as polymerizable monomers.

Protein-like Polymers (PLPs)

The PLP is an analogue of the PPA, wherein there is no hydrophobic block (Figure 3). Initially prepared as control polymers for PPA studies and as a test of our ability to polymerize various peptide-modified monomers,⁴⁹ we soon found unique properties inherent to these materials. Our first clue was that as the degree of polymerization of the peptide block (graft-through synthesis) increased, so too did the resistance of the resulting materials to proteolytic degradation. Indeed, polymerization of PLP structures consisting entirely of peptide substrates optimized for specific enzymes yields materials completely resistant to degradation by those same enzymes.^50–53 Further, PLPs can be prepared with cell penetrating peptides, resulting in proteolytically resistant materials that can access the cytosol of cells in culture⁵⁰ and maintain basic biological activities, including engaging specific organelles.⁵¹ Over time, we recognized that these materials may ultimately serve as a generalizable scaffold for peptide delivery, where high molecular weights are combined with proteolytic resistance of the peptides, while maintaining biological activity specific to those same peptides.

Figure 3.

Graft-to and graft-through strategies for the preparation of PLPs.

2. SYNTHESIS

Bioconjugates of nucleic acids, proteins, and peptides with polymers are generally designed with one of two goals in mind: (1) the use of the polymer to alter the function and behavior of the biomolecule in myriad applications including stabilization and extension of half-life in vivo,^1,8,54–57 or (2) the use of the biomolecule to influence the behavior of the synthetic polymer. The most prevalent examples of the second goal include the use of a biomolecule targeting group to guide a polymer, or polymer-based micro- or nanostructure, to a biological receptor of interest.

In general, bioconjugation reactions of synthetic polymers with biomolecules are achieved via graft-to,^18,58–60 graft-from,^56,61–63 or graft-through polymerizations.^22,64–68 In the peptide–polymer bioconjugates described herein, graft-to or graft-through polymerizations have been the key strategies used (Figure 2). In general, peptide–polymer conjugates prepared by graft-to polymerizations, wherein peptides are reacted with polymers containing conjugatable groups (Figure 2a), yield materials with low to moderate conjugation efficiencies.^{32–34,69–71} Although this approach can, in theory, be optimized and has been shown to be highly versatile in other contexts,⁷² an unavoidable complication arises as there is little ability to control the spacing and number of peptides conjugated per polymer, making characterization difficult and limiting batch-to-batch reproducibility. The alternative is graft-through polymerization, wherein the peptide is first modified with a polymerizable moiety and then directly incorporated into a growing polymer chain (Figure 2b).

Peptide–Polymer Amphiphiles via Graft-to and Graft-through Synthesis

A central goal of our research on peptide–polymer materials is to explore the generality of controlling nanoparticle behavior through the employment of biomolecules covalently bound to polymer backbones as programmable, stimuli-responsive moieties.^{21,32–35,49,50,69–71,73–79} Our early work on PPAs employed graft-to polymerization, wherein diblock copolymers were generated through copolymerization of a hydrophobic norbornenyl monomer together with a hydrophilic monomer containing a conjugatable NHS ester. Various peptides could then be incorporated postpolymerization via reaction with the NHS moiety.⁷⁰

Although graft-to incorporation of peptides into polymers allowed us to generate PPAs with interesting attributes,⁸⁰ we reasoned that precise control over peptide location and density on the polymer backbone would afford greater predictability and control of the overall materials properties. Direct incorporation of peptides on a growing polymer chain provides better control over spacing and number of peptides incorporated per polymer, allowing for the generation of polymers with more reproducible molecular weights, leading to minimal polymer-to-polymer and batch-to-batch variation. In turn, this enables systematic analysis of design parameters of peptide–polymers; namely, how peptide brush density and degree of polymerization impact function. Prior work in the field on the graft-through polymerization of peptides had been limited to aromatic and aliphatic residues, or resulted in polymers with high dispersity and low overall peptide incorporation.^64,65,81,82 For peptides with more complex functionalities, suitable protecting groups had to be employed and subsequently removed post-polymerization.^23,68,83

Living polymerization techniques such as ring opening metathesis polymerization (ROMP), allow for the graft-through generation of peptide–polymers with low dispersity.⁸⁴ The initiators, particularly the Grubbs ruthenium-based initiators,^85–87 display excellent functional group tolerance, permitting the incorporation of a diverse array of functionalized monomers. These include polymerizable peptide-modified monomers (graft-through approach) or bioconjugatable monomers (graft-to approach). A typical synthesis begins with the polymerization of m equivalents of monomer relative to Ru initiator. Upon complete consumption of the monomer, the polymer can either be terminated to produce a homopolymer by addition of a chain transfer agent (CTA) or can be elongated by adding n equivalents of a second monomer to ultimately produce a block copolymer with blocks of lengths m and n repeating units, respectively (Figure 4). In short, the precision and living characteristics of ROMP make it an ideal method to develop well-defined, highly reproducible systems with this type of complex monomer. Furthermore, we have recently demonstrated the synthesis of peptide brush polymers directly in aqueous solution, which provides a route around the need for a monomer to be both soluble in organic solvents for polymerization, and then soluble as a polymer in water where they are applied in biological systems.^36,43

Figure 4.

Ring-opening metathesis polymerization (ROMP) for homo- or copolymer synthesis, incorporating functional termination agents. (i) Polymerization of the first block begins with m equivalents of monomer relative to Ru initiatior. Upon complete consumption of the first monomer, the polymerization can be either extended using n equivalents of a second monomer (ii) to generate block copolymers as in the case of PPAs, which are then terminated using a chain transfer agent (iii). Alternatively, homopolymers can be prepared, again employing a chain transfer agent to terminate, as in the case of PLPs. Diverse functional groups can be incorporated into the chain transfer agent, including fluorescent probes and handles for conjugation.

Our work established ROMP as a viable strategy for the graft-through polymerization of peptides with diverse functionalities, through the analysis of the impact of peptide identity, density, and degree of polymerization on the function of the resulting materials. However, careful consideration of peptide sequence is necessary in graft-through polymerization of complex peptides. Many functional peptides contain amino acid residues capable of reacting with or coordinating to polymerization initiators. In these cases, suitable protecting groups may be used, or the sequence itself can be modified to remove the problematic residue(s) while maintaining the desired function.^23,68,83 Further, the peptide needs to be soluble in the polymerization solvent and remain solubilized throughout the duration of polymerization. An exception is where polymerization-induced self-assembly (PISA) is used to provide targeted nanostructure phases and variable solubility of peptide monomers and comonomers can be used to drive assembly.^36,43 Additionally, it has been demonstrated that many peptides are inherently slow to polymerize by ROMP,⁴⁹ which, coupled with the fact that initiators are oxygen-sensitive, dictates that solvents be rigorously degassed before being added to monomers in an oxygen-free environment. Finally, the most desirable outcome is that the resulting polymers yield to standard characterization strategies, including size-exclusion chromatography and NMR. We note that this is not always the case, with complex amphiphiles and peptide-brush polymers aggregating in solvents used for polymerization (e.g., DMF). Furthermore, polymers may aggregate and/or adsorb to standard separation media during characterization. In these cases, protecting groups can be used to cage problematic residues or a complete redesign or abandonment of a sequence is required. Overall, significant effort is required when designing graft-through peptide–polymer systems and parameters need to be evaluated on a system-by-system basis, often through empirical trial-and-error. This is an expensive process by any metric, thus procedures that utilize high throughput experimental methods⁸⁸ coupled with machine learning for optimizing peptide sequence are highly desirable.⁸⁹ In addition, we recently reported mild conditions for the preparation of peptide-brush polymers using photoinduced reversible-deactivation polymerization,⁹⁰ which addresses some of the limitations of ROMP (e.g., metal-free and less expensive initiators) and builds upon the work by Pun and others on the polymerization of single amino acid residues and simple peptides.^22,91–93 We anticipate new applications for these kinds of materials given the multiple approaches to the preparation of complex peptide–polymer brushes, where control over structural features including backbone molecular weight, biocompatibility, and glass transition temperature are more easily accessible.

Protein-like Polymer Synthesis

As part of our initial studies on PPA formation,⁸⁰ we generated homopolymers of peptides as functional controls (Figure 5). Polymerization of these peptide sequences to high degrees as homopolymers (Figure 5a, b) resulted in solvated systems with intriguing properties of their own (vide infra). This spurred a systematic study of the tolerance of the polymerization technique to peptide amino acid sequence and composition. The resulting polymers, herein termed protein-like peptides (PLPs), form structures akin to globular proteins (Figure 5c, d).⁵⁰ Our exploration of PLPs began as an investigation of the limits of graft-through polymerization of protecting-group-free peptides (Figure 6).⁴⁹ Given the vast sequence space, we narrowed our focus to generating a small library of 31 norbornenyl-modified, protecting-group-free pentapeptides that altogether incorporated all 21 naturally occurring amino acids. In this study, each pentapeptide monomer was polymerized as a set of homopolymers with varying degrees of polymerization via ROMP to afford PLPs (Figure 6a). The rate and degree of polymerization were quantified by a combination of NMR and SEC-MALS (Figure 6b). The only peptide monomers that did not polymerize were those containing cysteine residues, due to deleterious interactions with the initiator. This was mitigated by the addition of protecting groups to the cysteine thiol prior to polymerization. The broad scope for graft-through ROMP of peptide-conjugated norbornene-based monomers demonstrated in this study allowed further investigation of PLPs in biomedical settings where peptides could be protected from degradation, while maintaining bioactivity (vide infra).

Figure 5.

Protein-like polymers (PLPs). (a) General schematic of directly polymerized PLPs. (b) Structure of ROMP polymers directly incorporating peptides into the backbone. (c) Starting state of an in silico model of the PLP, initially with the molecule in an artificial, extended conformation with a straight norbornyl backbone and fully extended peptide side chains. (d) In silico model of the PLP after simulation for 100 ns at 300 K, resulting in structures akin to globular proteins.

Figure 6.

Exploring graft-through polymerization of protecting-group-free peptides. (a) ROMP polymerization scheme of pentapeptides modified with norbornenyl derivatives. Different amino acid side chains were incorporated at positions X₂ and X₅ to altogether encompass all 21 naturally occurring amino acids. (b) Plots of polymerization rates of protecting-group-free and protected amino acids. Adapted with permission from ref 49. Copyright 2013 Royal Society of Chemistry.

3. BIOACTIVITY AND FUNCTION OF PEPTIDE–POLYMER AMPHIPHILE ASSEMBLIES

Imaging in Vivo and ex Vivo

PPAs can be used to interface with complex biological systems and diseased tissue in vivo. In the design of PPAs for interacting with disease specific enzymes, we were inspired by the work of Tsien and co-workers who had shown that enzyme-responsive peptides can be systemically administered intravenously, respond to matrix metalloproteinases (MMPs) at tumor sites, and hence be endocytosed by tumor cells in mouse models via a neutral-to-positive charge switch, inducing cell membrane interactions.^29–31 This strategy has been successful for tissue labeling for imaging and guided surgery, where endosomal escape is not required.^29,94 We reasoned that PPAs designed to change morphology in response to MMP exposure to form large, slowly clearing aggregates could be used to accumulate materials extracellularly at disease sites where MMPs are upregulated in vivo for applications in imaging and drug delivery. Accumulation of these materials within the extracellular matrix of tumors would allow for local drug release of compounds already predisposed to act outside of cells or capable on their own of crossing membranes. We designed enzyme-responsive PPAs with hydrophilic polymer blocks containing peptides with recognition sequences for MMPs. Prior to MMP exposure, the PPAs assembled into spherical nanoparticles < 50 nm in diameter. Upon enzyme exposure, a shift in morphology from discrete nanoparticles to microscale aggregates was observed.^32,71

To date, we have demonstrated that fluorophore-labeled enzyme-responsive PPAs have utility in the imaging of disease states in vivo in animal models of cancer,^32–35 myocardial infarction (MI),³⁴ and peripheral artery disease.⁷⁴ The first generation of disease-imaging enzyme-responsive PPAs was designed based on our early successes with graft-to approaches (Figure 7). In this scheme, a hydrophobic phenyl-based norbornenyl moiety was polymerized via ROMP, followed by the polymerization of a hydrophilic conjugatable NHS-ester (Figure 7a). At this point, the batch of growing polymer was split in half and terminated with either a fluorescein- or rhodamine-based chain transfer agent (FRET donor and acceptor, respectively).⁹⁵ The hydrophilic blocks were further modified with peptides containing MMP recognition sequences postpolymerization, then dialyzed to afford fluorescein-(PPA_F) or rhodamine- (PPA_R) tagged, MMP-responsive nanoparticles (Figure 7b). When assembled within discrete nanoparticles, PPA_F and PPA_R did not produce a Förster resonance energy transfer (FRET) signal, even when coincubated for several months, indicating no interparticle mixing of polymers occurs and that these are kinetically trapped constructs. However, exposure of these micelles to MMP induced a morphology change and resulted in the rapid generation of a FRET signal, indicating a rearrangement of the amphiphiles bringing FRET donor-bearing PPA_F, within the Förster radius of FRET acceptor-bearing PPA_R.

Figure 7.

Fluorescently labeled enzyme-responsive PPA-based micellar nanoparticles for targeting in subcutaneous xenograft models of human cancer in mice. (a) Structure of diblock ROMP copolymers that contain enzyme-responsive peptides in the hydrophobic block and either fluorescein, a FRET donor (left), or rhodamine, a FRET acceptor (right), on the chain ends. (b) Upon transition from organic solvent into aqueous media, 30 nm nanoparticles spontaneously form. Shells are composed of enzyme-responsive peptides and fluorophores. (c) Upon intravenous coinjection and accumulation at the tumor site, tumor-associated enzymes cleave the peptides on PPAs. This causes a rearrangement into microscale aggregates and brings the FRET pair in close proximity generating a FRET signal, visualized in vivo in tumor tissue.

As a proof-of-concept, this motif was studied in HT-1080 fibrosarcoma murine xenografts.⁹⁶ A key design feature of these PPA systems is that analogous, yet enzyme nonresponsive, polymers and nanoparticles can be rapidly generated using peptide sequences composed of d-amino acids. These peptides are not recognized by enzymes as substrates because of their enantiomeric configuration; thus, systems generated with d-amino acids are incapable of undergoing morphology changes when exposed to MMPs and serve as a true negative control for this process. In HT-1080 xenografts, the enzyme-responsive PPA systems can be visualized in tumor tissue as a function of FRET signal via live-animal fluorescence imaging up to 2 days following intravenous (IV) injection, indicating that MMPs are acting upon the nanomaterials and eliciting the morphology change (Figure 7c). In contrast, the tumors remain dark in animals treated with the nonresponsive systems. Together, these results confirm that enzyme-responsive peptide–polymer systems accumulate in tumor tissue only following stimulus response and have utility in imaging tumor tissue following intravenous (IV) systemic administration.

Fluorophore-containing PPAs are also useful in imaging damaged heart tissue following acute myocardial infarction (MI).³⁴ As in cancer, MMPs are upregulated following MI during the left ventricle (LV) remodeling process, thus providing a targeting opportunity unique to these enzyme-responsive PPA-based micelles. In a study of PPA accumulation post-MI, fluorophore-labeled enzyme-responsive PPAs were intravenously injected at 24 h post-MI, and retention of material was evaluated by ex vivo fluorescence at various time points up to 28 days postinjection (Figure 8). Throughout the course of the study, enzyme-responsive PPAs were visualizable in infarcted tissue. Additionally, no accumulation was observed in healthy animals administered the responsive PPAs. Moreover, very little fluorescence signal was observed at 2 days postinjection in infarcted animals administered the nonresponsive controls, indicating that the enzyme-response, not the EPR effect observed acutely post-MI,⁹⁷ is necessary for retention of these materials. These materials are now being investigated for their potential as drug carrier systems to treat MI in the early stages following heart attack. They provide a unique opportunity to address what is a tremendously important health issue the world over, where heart disease and heart attack are major, unsolved and insidious problems. In addition, this approach takes advantage of leaky tissue combined with upregulated MMPs, which are naturally occurring phenomena in humans and in large animal models, making MI an especially attractive target for nanomedicine more broadly.

Figure 8.

Enzyme-responsive PPA-based micellar nanoparticles for targeting to heart tissue postmyocardial infarction (MI). Fluorophore-labeled enzyme-responsive micelles were intravenously injected at 24 h post-MI. The particles can be visualized in the infarct by fluorescence imaging (red, left) and by TEM (red arrows). In healthy animals, no accumulation is observed, whereas microscale aggregates can be visualized in the infarcted tissue up to 28 days post-MI. Adapted with permission from ref 34. Copyright 2015 Wiley.

While these results demonstrate the utility of PPA accumulation and retention in disease states, optimization of surface charge, particle size, and peptide sequence is needed to control circulation half-life, clearance, biodistribution, and interaction with immune cells. Toward this end, we have investigated the macrophage uptake of both PPAs and PLPs as a function of surface charge and morphology.⁴³ A key finding is that zwitterionic assemblies evade macrophage uptake. This suggests that complex polymer architectures may be rendered “stealthy” based on net charge, and that precise control over polymer architecture can impart predictable biological responses, as well as limit off-target accumulation and sequestration by macrophages and other immune cells.

Drug Carriers

Beyond imaging, the enzyme-responsive PPA motif can be used to deliver therapeutic cargo to disease sites at high doses and with minimal toxicity (Figure 9).^{35,69,98–100} A key challenge in nanomedicine is preventing the leakage and burst release of cargo upon injection, as well as low drug loading in the nanocarrier. To address both challenges simultaneously, we generated PPAs via graft-through polymerization of MMP-responsive peptides together with directly polymerized hydrophobic paclitaxel (PTX) moieties (Figure 9a).³⁵ As in PPAs for imaging of disease states, these drug-loaded polymers assembled into spherical nanoparticles, wherein the MMP-responsive peptides formed the nanoparticle shell. However, unlike their predecessors, the core of this new generation of enzyme-responsive nanomaterials contained a payload of chemotherapeutics which only becomes exposed to its environment after a morphology change elicited by MMPs. In this way, “Trojan horse”-like nanomaterials were generated with paclitaxel (PTX) comprising ~60% of the system by weight. PTX was esterified through its 2′-OH group with a norbornenyl moiety to make it amenable to ROMP, rendering it a fully inert prodrug prior to hydrolysis off the polymer backbone. This only occurs after the MMP-induced morphology change. Because of this, the drug is shielded from the body until it reaches its intended target, allowing for the safe delivery of exceptionally high doses of chemotherapeutic intravenously, relative to the clinical formulation of PTX (Figure 9b). This PPA drug delivery system had an estimated maximum tolerated dose (MTD) exceeding 10-fold that of both PTX³⁵ and Abraxane^98,101–107 following single intravenous (IV) injection. In addition, the MMP-induced morphology change is necessary for the PPAs to have a therapeutic effect, as evidenced by the lack of efficacy by the nonresponsive systems when administered intratumorally. Tumor growth was only inhibited by enzyme-responsive systems, whereas no antitumor effect was observed with analogous nonresponsive systems composed of d-amino acids (Figure 9c), following a single intratumoral injection. Further, the responsive PPAs showed efficacy equivalent to the parent compound at equivalent doses, following intravenous administration (Figure 9d).

Figure 9.

Drug-loaded, enzyme-responsive PPAs assemble to yield micellar nanoparticles for IV injectable, targeted cancer therapeutics. (a) Structure of PPAs generated through graft-through polymerization of monomers containing paclitaxel in the hydrophobic block and MMP-responsive peptides in the hydrophilic block. (b) Evaluation of maximum tolerated dose (MTD) reveals that PTX can be safely administered at 16-fold higher doses than in the clinical formulation of PTX. (c) Intratumoral administration of responsive PPAs containing PTX and nonresponsive controls demonstrates that morphology change is necessary to inhibit tumor growth. (d) Responsive PPAs containing PTX show equivalent to the clinical formulation of PTX at equivalent dose. Adapted with permission from ref 35. Copyright 2015 Wiley.

In addition to PTX delivery, PPAs have been employed for the delivery of Pt(II)-based therapeutics to tumor tissue (Figure 10).¹⁰⁰ A central challenge in drug delivery vehicle development and optimization is determining the distribution of both drug and carrier at high resolution in tissues. To address this, we designed a PPA system containing a block composed of a Pt(II) drug copolymerized with an ¹⁵N-labeled norbornenyl monomer, followed by a block of polymerized MMP-responsive peptides and a Cy5.5 near-infrared (NIR) fluorophore (Figure 10a). The design of this system enabled the use of correlated optical and isotopic nanoscopy (COIN)^108,109 to determine the location of the PPA drug carrier (Figure 10b,c) and the Pt(II)-drug (Figure 10d) with high spatial resolution in tumor tissue. As in the PTX-based system, efficacy of the Pt(II)-drug in the PPA paralleled that of the clinically used Pt-based drug, oxaliplatin. As an alternative to polymerization followed by PPA formation, we have explored the capability of one-pot aqueous phase synthesis of Pt(II)-containing nanoparticles using Ring-Opening Metathesis Polymerization-Induced Self-Assembly (ROMPISA).¹¹⁰ This approach could be used to generate materials with exceptionally high drug loadings and access high concentrations of particles in solution.

Figure 10.

Multimodal imaging of enzyme-responsive, drug-loaded PPAs form responsive, targeted micellar nanoparticles carrying platinum-based chemotherapeutic drugs. (a) Structure of PPA containing ¹⁵N-labeled norbornenyl monomers, Pt(II)-drugs, enzyme-responsive peptides, and Cy5.5 NIR fluorophore. (b) Ex vivo Cy5.5 fluorescence in tumor tissue, following intratumoral administration of the enzyme-responsive PPA-based micelles. (c) Hue-saturation intensity (HSI) representation of the ¹²C¹⁵N/¹²C¹⁴N ratio in the tumor tissue, showing areas enriched in ¹⁵N. (d) ¹⁹⁵Pt⁻ ion map reveals the drug distribution in the tumor. Adapted with permission from ref 100. Copyright 2018 American Chemical Society.

4. BIOACTIVITY AND FUNCTION OF PROTEIN-LIKE POLYMERS

In our initial demonstrations of graft-through polymerization of peptide polymer amphiphiles (PPAs), we elucidated that the spatial arrangement and density of peptides allows for the programming of a material’s susceptibility to proteolytic cleavage in vitro, and thus the ability to undergo a change in morphology or function. In this work,⁸⁰ two PPAs were prepared by directly polymerizing an inert hydrophobic block followed by a norbornenyl-modified peptide as the hydrophilic block via ROMP using a modified version of Grubbs’ second generation Ru catalyst. Here, the peptide sequences were designed to be substrates for MMPs. This method generated well-defined PPAs where the density and number of peptides per polymer were known. PPAs formed micellar nanoparticles of uniform size and shape, as confirmed by single-particle reconstruction using cryogenic transmission electron microscopy (cryoTEM).⁷⁰ However, in stark contrast to the previously prepared graft-to systems, minimal or no cleavage of the peptide substrate was observed when these nanoparticles were treated with MMPs in vitro. This could be linked to the degree of polymerization (DP) of the peptide-monomer. Briefly, increasing DP correlates with a decrease in the rate of proteolysis (vide infra). This important finding demonstrates that spatial arrangement and density of incorporation of the peptide on the polymer has a significant and tunable impact. Indeed, this finding led us to the development of PLPs.

Polymerization of peptide-based monomers to generate PLPs imparts novel properties to peptide sequences and on polymer function and architecture. For example, arrangement of peptides as PLPs protects them from proteolysis while maintaining their biological function.^50–53 In addition, it is possible to impart cell-penetrating properties into otherwise impermeable peptides through this technique.⁵⁰ We had reason to examine these functions, as modeling PLPs in silico revealed they generally form condensed protein-like structures akin to globular proteins (Figure 5d).⁵⁰ In these structures, the hydrophobic polymer backbone appears wrapped in peptide side chains, which leads to a protein-like display of the amino acid sequences. A major problem in peptide-based therapeutics is short circulation times in the blood due to cleavage by serum proteases and rapid clearance because of their relatively low molecular weights.¹¹¹ Therefore, packaging peptides as high-density, high molecular weight PLPs could be a strategy to deliver bioactive peptides with extended circulation half-lives.

Many therapeutic and bioactive peptides contain amino acid residues which had never previously been polymerized in a graft-through fashion. Using our knowledge gained from determining substrate scope (see Synthesis section), we generated bioactive peptide-based systems and probed their activity and properties (Figure 11).⁵⁰ As noted above, polymerization of peptides as dense brushes in PLPs greatly inhibits proteolytic susceptibility (Figure 11a), which is a function of the degree of polymerization and depends on the presence or absence of diluent monomers copolymerized. Specifically, we found that the proteolytic susceptibility of PLPs can be tuned by spacing the peptides farther away from one another along the polymer backbone by adding a nonpeptide diluent monomer during polymerization (Figure 11b). Thus, we explored the potential of using high-density brush polymers to protect bioactive peptides from proteolysis. Toward this end, a PLP composed of the cell-penetrating peptide (CPP) Tat was synthesized and analyzed using a fluorescein moiety as model cargo. The proteolytic susceptibility of the PLP was evaluated by exposing the materials to high concentrations of proteases and assaying for peptide cleavage by HPLC and bioactivity by flow cytometry and live-cell confocal microscopy. No evidence of proteolysis was observed for the PLP, whereas the monomeric peptide was completely consumed (Figure 11c). Pretreatment with proteases had no effect on the cell-penetration capability of peptide–polymers; however, the monomeric peptides displayed greater than 90% signal loss (Figure 11d).

Figure 11.

Modulating proteolytic susceptibility and imparting cell-penetrating capabilities on peptides using PLPs as carriers. (a) Polymerization of peptides as PLPs (top) significantly decreases proteolytic susceptibility (bottom) relative to the peptides as monomers. (b) Incorporation of a diluent monomer (top) can be used to modulate proteolytic susceptibility of PLPs (bottom). (c) Exposure of functional peptides to proteolytic enzymes results in complete degradation, whereas very little degradation is observed when the peptide is incorporated into a PLP. (d) Following proteolytic degradation, peptide monomers are incapable of penetrating cells (top), while the PLP retains its functionality (bottom). Adapted with permission from ref 50. Copyright 2014 American Chemical Society.

Next, we investigated the potential of using this PLP motif to enable cellular uptake of otherwise impenetrable peptides (Figure 12).⁵⁰ We hypothesized that the incorporation of positively charged Arg or Lys residues into non-cell-penetrating peptides, followed by polymerization as a high-density brush, would enable cellular uptake of otherwise impermeable peptide sequences. Toward this end, we synthesized a non-CPP and added one or two Arg or Lys residues at either the N- or C-terminus and polymerized it via ROMP, together with a water-soluble OEG block (Figure 12a). We compared these constructs to their monomeric peptide counterparts and found that only the peptide–polymers containing the positively charged amino acid residues entered cells (Figure 12b).

Figure 12.

Imparting cell-penetrating capability to otherwise impermeable peptides using PLPs with function and action on organelles. (a) Structure of PLPs containing peptide sequences incapable of entering cells (GSGSG) with various modifications with cationic amino acids. (b) Only PLPs containing positively charged amino acids enter cells, whereas polymers lacking the positively charged residues and peptide monomers are incapable of cell penetration. (c) Structure of the PLP containing the mitochondrial disrupting KLA peptide. (d) PLPs containing KLA enter cells at all degrees of polymerization tested, whereas the KLA peptide itself is unable to penetrate cells. (e) Incorporation of KLA into the PLP protects the peptide from proteolysis when exposed to trypsin (blue bar) and Pronase (gray bar), whereas the monomeric peptide counterpart is highly susceptible. (f) The cytotoxicity of KLA PLPs is on par with reported CPP-KLA conjugates. (g) KLA PLPs retain their biological function and are capable of mitochondrial disruption. Adapted with permission from ref 51. Copyright 2016 Royal Society of Chemistry.

As many bioactive peptides already contain cationic amino acids, we sought to determine whether the PLP motif could be utilized to deliver therapeutic peptides without appending additional cationic residues.⁵⁰ Further, we wanted to evaluate whether bioactive peptides displayed as PLPs would retain their bioactivity. Thus, we polymerized the KLA peptide, an established mitochondrial disrupting peptide with multiple cationic residues (Figure 12c) whose activity is hindered by extremely low cellular uptake. At all tested degrees of polymerization, KLA as a PLP penetrated cells at levels matching or exceeding that of the positive control, whereas the KLA peptide alone was unable to enter cells (Figure 12d). Further, KLA as a PLP was unaffected by the presence of the proteolytic enzymes trypsin and Pronase in vitro, whereas the KLA peptide was rapidly consumed (Figure 12e). Importantly, the cytotoxicity of KLA as a PLP is on par with reported ranges for CPP-KLA peptide conjugates (Figure 12f) and maintained its biological function (Figure 12g). We note that KLA-peptide brush polymers accessed using photoinduced reversible-deactivation polymerizations similarly lead to enhanced uptake and bioactivity, but without exceptionally enhanced resistance to proteolytic degradation.⁹⁰ This result hints at a potentially unique aspect of the polynorbornene-based backbone in conferring special properties to peptides arrayed as PLPs.

In summary, PLPs are a generalizable platform technology for displaying peptides. We continue to study PLPs for their potential as carriers for bioactive and therapeutic peptides that can resist degradation even in harsh conditions including stomach acid and whole blood. Furthermore, we have shown in recent work that the cell penetrating capability of PLPs can be employed to facilitate the transport of gene editing proteins into stem cells, providing a generalizable and efficient polymer based tool for in vitro chemical biology applications.⁵²

5. OUTLOOK AND SUMMARY

In this Account, we have described the development of peptide-conjugated polymers, in particular those based on graft-through ROMP, although recent work demonstrates analogous structures can be prepared by photoinduced reversible-deactivation radical polymerization.⁹⁰ These new, mild, and metal-free conditions open the door for further development of therapeutic delivery systems and for the incorporation of a broader set of monomers, with increased backbone diversity to control properties, including multivalency and flexibility. The utility of this approach to densely packed peptides arrayed on a polymer scaffold has been illustrated in cell and tissue targeting, in terms of both tuned response to enzymatic reactions^{32–35,69–71,77,100} and in the counter example of enhanced resistance to proteolytic degradation.^{43,49,51,52,112} In addition, the use of PISA to generate PPAs, and hence nanomaterials in situ, is proving to be an excellent approach to high yielding and high drug loaded systems.¹¹⁰ The aqueous phase ROMP initiator employed in these PISA studies also provides a route to preparing PLPs that avoids having to first prepare the polymers in organic solvent prior to transition to aqueous phase for biological applications. This is becoming critical in our studies that necessitate peptide sequences that do not readily disperse in the organic solvents normally used for ROMP.

Critically, in each case discussed here, accessed via several polymerization strategies, the biopolymer endows the synthetic polymer with biological attributes, and conversely, the polymer modifies the biological activity of the biopolymer. They work in concert to give rise to new types of behaviors of utility in a range of applications, from cancer to heart disease. While our focus to date, and in this Account, has largely been on therapeutics and responsive materials for delivery in vivo, we have begun to explore the utility of peptide–polymers more broadly. For example, in another manifestation, we have used responsive, peptide-conjugated PPAs to surface functionalize liquid crystal droplets dispersed in water for detecting enzymes via a propagated optically detectable response in the liquid crystal.¹¹² This type of signal propagation has promise in transducing molecular level sense-response to longer length and time scales. In addition, PLPs have been shown to act as noncovalent carriers of proteins into valuable cell types, including stem cells.⁵² This is timely, as the search continues for efficient and general transporters of proteins into these cell types.^113,114 Furthermore, the retention of bioactivity, coupled with increased stability, could be employed in the development of antifouling surfaces for controlling biofilm formation, antibacterial interfaces, as biocatalysts or in the development of peptide arrays for recognition and detection of analytes.

Overall, the arrangement of peptides as side chains of synthetic polymers provides what is, in essence, a new class of “polypeptide”. Not as nature would have it, consisting of a linear chain of amino acids, but rather as a polymer of peptides, wherein amino acid sequence gives rise to biological response and recognition. Therefore, we introduce this synthetic biopolymer as a “poly(peptide)”. Poly(peptide) possesses new properties connected to density of the display of amino acid sequences, giving rise to multivalency in binding and hindrance to proteolysis, depending on the design. These are properties not necessarily accessible to the individual oligopeptides themselves, with generalizable implications for how we deploy peptides as materials and medicines.

Biographies

Cassandra E. Callmann completed her B.S. in Biochemistry at West Chester University (2012) and both her M.S. (2014) and Ph.D. (2018) in Chemistry at the University of California, San Diego. Under the guidance of Prof. Nathan Gianneschi, she developed targeted delivery systems for cancer therapeutics, with support from the Inamori Foundation, the ARCS Foundation, and the Cancer Researchers in Nanotechnology program (NIH R25T). She is currently the Eden & Steven Romick IIN Postdoctoral Fellow in the lab of Prof. Chad Mirkin at Northwestern University. Cassandra was selected as a 2019 CAS Future Leader.

Matthew P. Thompson received his B. Sc.H. (2000) and Ph.D. (2006) in Chemistry from Queen’s University (Kingston, Canada). In 2008 he joined the Gianneschi Lab where he is currently a Senior Research Associate. His research interests include the synthesis of polymeric materials.

Nathan C. Gianneschi received his B. Sc(Hons) in Chemistry at the University of Adelaide, Australia (1999), his Ph.D at Northwestern University (2005) and conducted his Postdoctoral Fellowship at The Scripps Research Institute. He is currently Jacob & Rosaline Cohn Professor of Chemistry, Materials Science & Engineering, Pharmacology and Biomedical Engineering at Northwestern University. He is interested in multifunctional materials with relevance in biomedical applications, programmed interactions with biomolecules and cells, and basic research into nanoscale materials design, synthesis and characterization.