Peptide Assemblies for Cancer Therapy

Abstract

Supramolecular assemblies made by the self-assembly of peptides are finding an increasing number of applications in various fields. While the early exploration of peptide assemblies centered on tissue engineering or regenerative medicine, the recent development has shown that peptide assemblies can act as supramolecular medicine for cancer therapy. This review covers the progress of applying peptide assemblies for cancer therapy, with the emphasis on the works appeared over the last five years. We start with the introduction of a few seminal works on peptide assemblies, then discuss the combination of peptide assemblies with anticancer drugs. Next, we highlight the use of enzyme-controlled transformation or shapeshifting of peptide assemblies for inhibiting cancer cells and tumors. After that, we provide the outlook for this exciting field that promises new kind of therapeutics for cancer therapy.

Graphical Abstract

This review highlights the recent progress in the application of peptide assemblies for cancer therapy, with an emphasis on the use of enzyme-instructed self-assembly for generating anticancer assemblies in situ. We aim to illustrate the promise of peptide assemblies as a new class of supramolecular medicines for controlling cell behaviors and developing anticancer therapeutics.

Frontispiece:

1. Introduction

Self-assembly of peptides or peptide assemblies has received considerable research attentions since the end of last century. A couple of seminal works^[1] have illustrated the promise peptide assemblies for a wide range of applications. For example, Ghadiri et al. reported on nanotubes based on rationally designed cyclic polypeptides.^[1a] The cyclic peptides self-assemble to form the tubes with the length of hundreds of nanometers and internal diameters of around one nanometer. It was suggested that these open-ended tubes may have possible applications in inclusion chemistry and catalysis. Based on a serendipitous observation, Zhang et al. reported a 16-residue peptide [(AEAEAKAK)₂] that exhibit β-sheet characteristics in water.^[1b] In phosphate-buffered saline (PBS), the peptide self-assembles to form a macroscopic membrane consisting of filaments with diameters 10~20 nm. The peptide assemblies are innocuous to PC12 cells, which is a useful prerequisite for them to serve as biomaterials for maintaining neurite growth^[1e]. By attaching dialkyl chain to peptides, Tirrell et al. demonstrated that peptide amphiphiles can self-assemble at the air–water interface to form stable monolayers.^[1c] Boden et al. presented a generic statistical mechanical model to elucidate the self-assembly of chiral rod-like units, such as β-sheet-forming peptides, into tapes, ribbons, fibrils, and fibers and showed that 8-nm-wide peptide fibrils are extremely robust and sufficiently rigid to form nematic solutions.^[1f] Stupp et al. reported on a rational design of versatile peptide-amphiphiles, which are able to present different functional sequences for applications such as direct mineralization of hydroxyapatite^[1g] or guided differentiation of neural progenitor cells^[2]. Gazit reported on the self-assembly of diphenylalanine to form nanotubes.^[1i] The exceptional self-assembling ability of diphenylalanine has greatly stimulated the exploration of peptide assemblies. Contrasting to the unexpected observation of the hydrogel^[1j] formed by an antibiotic cyclic peptide, vancomycin, Schneider et al. presented an elegant peptide design, intramolecular folding of β-hairpin, to trigger the formation of peptide hydrogels consisting of β-sheet.^[3] The shear thinning nature of the hydrogel makes it to be an excellent vehicle for drug delivery.^[4] These studies have illustrated the exciting promises of peptide assemblies as supramolecular biomaterials.

The early work of peptide assemblies centered on the use of peptide assemblies as scaffolds for tissue engineering, several excellent reviews have been published on the subjects.^[5] The exploration of peptide assemblies for cancer therapy stared about a decade and half ago.^[6] This review will focus on the progress made in last five years. We will limit the discussion to the peptide assemblies that are the major contributors to inhibiting cancer cells, rather than just a medium for drug delivery. In the following sections, we will first discuss the assemblies of the conjugates of peptide and anticancer drugs. Next, we will highlight the use of enzyme-controlled transformation or shapeshifting of peptide assemblies to inhibit cancer cells and tumors. Finally, we will provide the outlook for this exciting field, which promises innovative approaches for cancer therapy. We expect that this brief review will provide a starting point to the molecular and medical scientists who are interested in exploring the potential and relevant molecular and cellular mechanisms of peptide assemblies for cancer therapy.

2. Assemblies of the conjugates of peptide and drug

Peptides conjugated with anticancer drugs are probably the most straightforward approach for employing peptides assemblies in the context of drug delivery or controlled release, as illustrated in some early works^[7] and reviews^[8]. Although the anticancer drugs provide the mechanism of the action for inhibiting the cancer cells, the peptide assemblies usually augment the efficiency of the drugs via enhancing tumor targeting or providing sustained release of the drugs (Figure 1).

Figure 1.

(A) Self-assembly forms assemblies of peptide-drug conjugates. (B) After endocytosis, the assemblies of peptide-drug conjugates release drugs to kill cancer cells.

2.1. Peptides conjugated with anticancer drugs

In a recent study, Parquette et al. reported the co-assemblies of two different peptide drug conjugates. One conjugate (1, Scheme 1) contains two drugs, 5-fluorouracil (5-FU) and camptothecin (CPT), while the other conjugate (2, Scheme 1) bears only CPT. These two conjugates self-assemble to form nanotubes. One unique feature of these peptide nanotubes is the rapid release of 5-FU, followed by slower, sustained production of CPT^[9]. The cytotoxicity of the conjugates is comparable to the cytotoxicity of CPT in cell assays at 96 h. Although the rate of release in cell assays appears to be slow, this type of dual release with different kinetics likely would be useful when the dosing of multiple drugs is necessary in treatment. The authors also proposed that 1 resides at outer shell and 2 at the inner shell of the nanotubes, an interesting assumption that may worth the structural determination by cryo-EM.^[10] In an earlier study by Parquette et al., the authors reported a tetrapeptide (CPT-CCKK, 3, Scheme 1) that self-assembles to form nanotubes for delivering CPT.^[11] Specifically, the self-assembly and oxidative crosslinking of 3 forms intermolecular disulfide bond to control the structure stability and CPT release rate. Incubated in PBS at 20 mM for 3 days, the mature nanotubes exhibited a wide range of diameters (232 ± 60 nm) and lengths of several micrometers measured by transmission electron microscopy (TEM). Furthermore, the hydrolytic release of CPT from the crosslinked nanotubes of 3 was completed within 35 h when exposed to dithiothreitol (DTT) that reduced the disulfide bond. The IC₅₀ of unoxidized nanotubes of 3 against human non-small cell lung cancer (NSCLC) cells A549 and H460 was 0.22 and 0.28 μM, while the crosslinked nanotubes exhibited lower cytotoxicity of 2.43 and 0.21 μM, respectively. This difference was attributed to the higher concentration of glutathione in H460 which was twice of that in A549 cell line and caused faster CPT release.

Scheme 1.

Structures of peptide drug conjugates 1–3.

Based on an earlier work by Cui et al., which reported a peptide-drug amphiphiles qCPT-buSS-Tau (4, Scheme 2) for cancer therapy^[7a], Loverde et al. used molecular dynamics (MD) simulations to investigate the balance of intermolecular forces governing the stability of supramolecular nanotubes formed by 4. Among seven pre-assembled nanotubes composed of varied numbers of 4, the authors selected two of them for computation. Both tubes have a width around 9.5 nm and a hollow channel of 2 to 4 nm, as verified by TEM. The only difference between those two types of nanotubes is the initial conformations of 4, which is the building blocks of the nanotubes. The nanotubes use CPTs as the inner wall while the charged peptide residues are exposed to the outside in both nanotubes. MD simulations suggest that the confined water channel in the self-assembled nanotube with a diameter of 20–25 Å exhibits interesting water dynamics, showing faster diffusion of in-tube water than surface water, with the diffusion coefficients of 3.2 × 10⁻⁹, 2.2 × 10⁻⁹, 0.9 × 10⁻⁹ m²/s for bulk water, in-tube water, and surface water, respectively. Such a relatively small difference in water diffusion may provide insights for understanding the polymorphisms of peptide nanotubes. It would be worthwhile to obtain the atomic structures of the nanotubes of 4 and to use MD simulation to design peptide-drug conjugates.

Scheme 2.

Structures of peptide drug conjugates 4–7.

Wang and Sun et al. conjugated rhein, a tradition Chinese medicine, to diphenylalanine for generating RDP (5, Scheme 2).^[12] Based on coordinated intramolecular π-π stacking and dissipative particle dynamics (DPD) simulation, they reasoned that 5 would co-assemble with both high affinity molecules, CPT(6, Scheme 2), and low affinity molecules, norcantharidin (NCTD, 7, Scheme 2). According to the DPD, the affinity between the drugs and carriers would inhibit the recrystallization of drugs in a high concentration. Indeed, 5 inhibits the crystallization of 6 much more efficiently than 7, and assemblies of 6 have better storage stability than those of 7. The cytotoxicity against mouse breast cancer cells (4T1) for CPT-containing dipeptide-based nanoassemblies 5/6 were higher than free CPT (6) and could efficiently inhibit 4T1 cell proliferation at a concentration of 5 μg/mL. Moreover, the assemblies of 5/6 with the dose of 5 mg/kg exhibit greater antitumor activity than 6 in vivo. This study shows a considerable synergy between 5 and 6, and it is worthwhile to elucidate the origin of such a synergy. It would be interesting to compare the efficacy of the assemblies of the conjugates of 5 and 6 and the co-assemblies of 5 and 6.

Wang et al. reported a potential drug derived from a supramolecular-peptide based on the interactions between cucurbit[7]uril (CB[7], 8) and the Phe-Phe-Val-Leu-Lys-camptothecin conjugates (FFVLK-CPT, 9) (Scheme 3).^[13] FFVLK is a reverse sequence of the KLVFF, the β-sheet-forming peptide domain derived from β-amyloid. The aromatic side chain of FFVLK presumably inserts into 8 to form the desired complex (CPC, 10, Scheme 3). While 9 self-assembles to form nanofibers, 10 self-assembles to form nanoparticles with a hydrodynamic diameter of 164 nm. Because the CB[7]-Phe host-guest pairs can be dissociated by spermine (SPM), the IC₅₀ of 9 in SPM-overexpressed 4T1 cells and SPM-normally expressed AML12 mouse liver cells was 1.3 μM and 135 μM, respectively. Moreover, the tumor growth inhibition test showed significant inhibition of tumor growth in 10-treated mice compared with other groups treated with PBS, CPT, and 9 only. One advantage of this approach is the enhanced biocompatibility of 10. This work illustrates a promising approach for targeting cancer cells. It would be important to confirm the SPM-induced transition of nanoparticles to nanofiber in 4T1 by more rigorous cryo-electron tomography (ET) imaging.

Scheme 3.

Structures of peptide drug conjugates 8–11.

To improve the efficiency of CPT delivery, Wang et al. designed a conjugate of CPT and peptide, CPT-WT-H (11, Scheme 3).^[14] 11 self-assembles into nanoparticles and is a substrate of both prostate-specific membrane antigen (PSMA) and esterases. The authors synthesized 11 by combining hydrophobic CPT and a reported hydrophilic PSMA-responsive pentapeptide,^[15] through a cleavable ester bond. This conjugation led to the formation of nanoparticles with a diameter of approximately 100 nm. Although the cytotoxicity of 11 for PSMA-expressing cancer cells LNCaP-FGC (IC₅₀ = 1.1 μM) was 35 times higher than non-PSMA-expressing cancer cells HepG2 (IC₅₀ = 36.0 μM), it is unclear why the cytotoxicity of 11 for PSMA-expressing PC-3 cells is modest. Moreover, in vivo imaging showed that 11 circulated rapidly in the bloodstream after injection and accumulated in tumor site by the EPR effect and PSMA-mediated active targeting. The improved retention of 11 was further verified by pharmacokinetic study, with a blood circulation half-life t_1/2 to be 10.9 h for 11 and 0.165 h for free CPT. The nanoparticles of 11 also suppress tumor growth without showing obvious systemic toxicity in vivo, as verified by MCF-7 tumor xenograft models and histological analysis results. The responses to dual enzymes are likely important for tumor targeting, as shown in an early study.^[16]

Zhong and Xu et al. investigated the co-assembly of an octapeptide (12, Scheme 4) and HCPT for enhanced therapeutic efficacy.^[17] The peptide 12 contains a self-assembling motif, FKFEY, and an tripeptide (YSV)^[18] that was reported to have antitumor effect. In water, 12 and HCPT co-assemble to form a hydrogel. Cell viability test showed that the mixture of 16 and HCPT had an IC₅₀ of about 0.2 μM, which is approximately 3 time more potent than HCPT against A549 lung cancer cells. The authors also evaluated antitumor efficacy of 12/HCPT in vivo by administrating hydrogels subcutaneous (s.c.) in mice that had A549-xenograft tumors. After 21 days, the average tumor volume in mice receiving the 12/HCPT hydrogel at 6.0 mg/ kg was significantly lower than that of mice receiving injections of either PBS or HCPT solution (at 6.0 mg/kg). This study demonstrated that the combination of anticancer peptide assemblies with potent anticancer drugs produced superior anticancer efficacy. Similarly, Yang et al. described a co-assembly of doxorubicin (Dox) and short peptides Fmoc-FK and Fmoc-FKK. The co-assembled peptide-Dox nanoparticles, with an average size of 50–100 nm and positive charges, exhibit increased cellular uptake compared to free-Dox, causing significant cancer cell death.^[19]

Scheme 4.

Structures of peptide drug conjugates 12–13.

By conjugating HCPT and macrocyclic polyamine cyclen to a self-assembling peptide FFFK, Liu et al. developed a supramolecular hydrogel for nuclear drug delivery.^[20] The conjugate (HCPT-FFFK-cyclen, 13, Scheme 4) self-assembles to form nanofibers that act as the matrices of the hydrogel. The authors reported that the IC₅₀ values of 13 against A549, HeLa, MCF-7 cells were 2.5, 5, and 20 μM, respectively, which were about 17, 10 and 4 times lower than that of HCPT. Moreover, the conjugate also improves nuclear accumulation of HCPT in cancer cells and exhibit enhanced tumor inhibition in vivo. It would be useful to determine the amount and identity of the molecules accumulated in the nuclei of the cancer cells to elucidate the nucleus targeting mechanism.

Wu et al. reported the use of a dodecapeptide ligand (TSFAEYWNLLSP, PMI) that bind to the transcriptional regulators of p53 for conjugating with two anticancer drugs, HCPT and chlorambucil (CRB).^[21] As shown in Scheme 5, they designed and synthesized five conjugates, CRB-K(HCPT)-FFYG-PMI (14), CRB-K(Nap)-FFYG-PMI (15, without HCPT), HCPT-K(Nap)-FFYG-PMI (16, without CRB), CRB-K(HCPT)-FFYG-scrambled PMI (17, with the scrambled PMI peptide), and CRB-K(HCPT)GGGG-PMI (18, without the FFY motif). They showed that all these conjugates self-assemble to form nanoparticles, with diameters ranging from 50 to 170 nm. According to the fluorescent imaging, peptide conjugation increases cell uptake of HCPT, with 14 resulting in the most cellular uptake. The IC₅₀ values of these conjugates against HepG2 cells follow the order of 14 (0.21 μM) < 17 (0.47 μM) < 18 (0.57 μM) < 16 (0.94 μM) < HCPT (10 μM) < 15 (71 μM) < CRB (250 μM). The tumor inhibiting abilities of these conjugates in vivo also follow the same trend. One interesting feature of these conjugates is that peptide conjugation apparently disfavors the nucleus accumulation of HCPT. It is worthwhile to determine the subcellular location of HCPT more precisely for the future optimization.

Scheme 5.

Structures of peptide drug conjugates 14–18.

Wang et al. reported the use of a 5-FU conjugated peptide (FU-NHOH(19), Scheme 6) for the delivery of nucleic acid into nucleus of cells.^[22] in addition to 5-FU, 19 contains a hydroxamic acid group at the C-terminus to inhibit histone deacetylases (HDACs). The authors used 19 to co-assemble with an anticancer aptamer, AS1411, which is a 26-mer guanine-rich oligonucleotide DNA in phase II clinical trials for relapsed or refractory acute myeloid leukemia and renal cell carcinoma^[23]. While 19 inhibits HDAC at IC₅₀ of about 0.3 nM, its cytotoxicity against PA-1, HT-29, or Huh-7 cells is about 10, 75, and 125 μM, respectively, which is rather modest. This large difference indicates other mechanisms likely compensate the inhibition of HDAC. The positively charged 19 mixes with the negatively-charged AS1411 to form nanospheres with the diameters 100 nm. The formation of nanosphere indeed promotes the uptake of AS1411 into cells. However, AS1411 localized less in the nucleus. Comparing to 19 alone, the combination of 19 and AS1411 increases the percentage of apoptosis of PA-1 cells by about 3 times. Preliminary mechanistic investigation shows that 19/AS1411 nanoparticles could significantly decrease the level of WNT5A1 and SOX2, which indicate the stemness of the cancer cells. While this work presented a useful concept for the nuclear delivery of nucleic acid drugs, the efficacy of such nanoparticles needs significant improvement to achieve the same level inhibition at as approved HDAC drugs such as SAHA^[24].

Scheme 6.

Structures of peptide drug conjugates 19–21.

Ding et al. synthesized a peptide conjugate of curcumin as a radiosensitizer for cancer therapy.^[25] The conjugate, Cur-FFE-CS-EE (20, Scheme 6), contains three segments, curcumin, a self-assemble motif, and GSH responsive diglutamic acids. 20 has good water solubility (about 50 mg/mL in PBS). The solution of 20 turns into a hydrogel after GSH addition. Diluting the hydrogel produces a supramolecular nanofiber suspension, in which the nanofibers have diameters about 30 nm and length of micrometers. The nanofibers slowly release curcumin with a rate of about 0.29 μg/ mL per hour for up to 12h. Against HCT-116 cells, the IC₅₀ values of 20 and free curcumin were about 11 μg/mL. While the in vitro cytotoxicity of 20 is comparable to curcumin, 20 plus radiation is more effective than curcumin plus radiation against colon cancer cells (HCT-116 and HT-29). Using an HCT-116 tumor-bearing mouse model, the authors estimated that the nanofibers achieved sensitizer enhancement ratio at 10% cell survival (SER10) value of 2, the highest among currently reported curcumin-based radiosensitizers. This work illustrates the promise of peptide-drug assemblies in radiosensitization, an application likely would enter clinical translation earlier than systemic applications.

To obtain better clinical use of paclitaxel (PTX), Xu et al. designed an iRGD-PTX conjugate (21, Scheme 6).^[26] Using Michael addition, the authors conjugated PTX to iRGD, a cyclic RGD peptide (CRGDKGPDC) reported by Ruoslahti,^[27] for generating 21. TEM shows that 21 self-assembles to form nanoparticles with diameters about 130 nm in PBS. Nanoparticles of 21 exhibits sustained PTX release behaviors in both pH 7.4 and pH 5.0 PBS buffer, with the release 57% and 66% of PTX in pH 7.4 and pH 5.0 PBS buffer, respectively, after 120h. Using Cy5 to conjugate with the nanoparticles, the authors showed enhanced cellular fluorescence intensity inside 4T1 cells, a result that matches with the slightly improved cytotoxic of 21 against integrin positive 4T1 cells. At 48h, the IC₅₀ value of 21 against 4T1 cells appears higher than 80 μg/mL. The biodistribution of 21 in 4T1 bearing mice model shows that, 24h after intravenous injection, 21 mainly accumulates in kidney. Such an unfavorable accumulation may be reduced by carefully engineering the size of the nanoparticles of 21 or by topical application of the iRGD-PTX conjugated, as shown recently by Cui et al. in treating glioblastoma.^[28] In another study for PTX delivery, Wu and Ge et al. reported a pH-responsive peptide hydrogel made of FER-8^[29] for PTX delivery.^[30] While it is encouraging that hydrogels enhanced the efficacy of PTX in a subcutaneous mice mode of H22 cancer cells, the requirement of intratumorally injection would limit the broad clinical applications.

2.2. Peptides containing functional motifs

Hou et al. developed a conjugate (22, Scheme 7) of L-peptide and D-peptide to inhibit HCT116 cancer cells in vitro and in vivo.^[31] The anticancer motif in the conjugate is a dodecameric D-peptide (^DT^DA^DW^DY^DA^DN^DF^DE^DA^DL^DL^DR (D-PMI))^[32], which can activate p53, a tumor suppressor. The L-peptide segment in the conjugate has the sequence of VVVVVHHRGDC (PSP), in which the pentavaline, dihistidine, and RGD are responsible for self-assembling, pH responding, and binding integrin, respectively. At pH 7.4, the conjugate self-assembled to form nanoparticles with diameters about 86 nm, which became nanoparticles with 4 nm diameters at pH 6.5. The authors reported that PSP improves cellular uptake of D-PMI. Impressively, 22 exhibits IC₅₀ to be comparable to nutlin3 in term of molar concentration, in the range of 1–3 μM. In a mice model beading the HCT116 p53+/+ cells, 22 also slowed down the tumor growth. Because cellular transporters are not evolved for D-peptides, it is challenging for D-peptide drug to enter cells. This work illustrates a promising approach to bring D-peptide cargos into cell via self-assembly of heterochiral peptides and pH response.

Scheme 7.

Structures of peptide drug conjugates 22–26.

Hu et al. reported a peptide conjugate that contains and aromatic motif (tetraphenylethylene (TPE)) for delivery of Dox or a porphyrin derivative (P18).^[33] The peptide has the sequence of SKDEEWHKNNFPLSP (STP), which is pH-triggered and VEGFR2-targeting peptide. TPE connects with the N-terminal of the STP segment to form the conjugate (23, Scheme 7), which self-assembles to form nanoparticles. The authors showed that the nanoparticles turned into nanofibers under acidic condition and employed this pH-trigged morphological transition as a mechanism to release drugs (Dox or P18). The authors also suggested that binding to VEFGR2 transformed the nanoparticles to nanofibers. Thus, 23 represents a type of dual-responding peptide conjugates. Murine model studies show that the self-assembled 23 could enhance the in vivo delivery efficiency toward tumor. The photoacoustic feature of P18 loaded in 23 also resulted in a specific high signal even in an early stage of tumor. It would be useful to determine the fate of 23 in cells and in tumors to take full advantage of the physiological stimuli response of the assemblies of the conjugates.

Yang and Liu et al. reported on the use of YSV and self-assembling peptides for cancer inhibition.^[34] While YSV was found to interrupt cell cycle and suppress the activity of histone deacetylase, it only functioned at high concentration.^[18] As shown in Scheme 7, the authors produced two conjugates: Nap-GFFYGYSV(L-YSV, 24) and Nap-GffyGYSV(D-YSV, 25), with 24 being homochiral and 25 heterochiral. The critical micelle concentration (CMC) values of 24 and 25 are 75 and 180 μM, respectively. Interestingly, heating-cooling the PBS solution of 24 and 25 results in a suspension and a hydrogel, respectively. 24 self-assembles to form short nanofibers with diameters about 83±13 nm, while the hydrogel of 25 contains long fibers with diameters about 24±3 nm. Both 24 and 25 were found to be much more potent than YSV for inhibiting BEL-7402, HeLa, and MCF-7 cells, with IC₅₀ values of 103, 54, and 28 μM for 24, and 72, 30, and 22 μM for 25 against BEL-7402, HeLa, and MCF-7 cells respectively. Additionally, both 24 and 25 were more effective than YSV in slowing down the growth of BEL-7402-luciferase tumor in mice, with 25 being statistically more active than 24 for retarding tumor growth. Due to the simplicity and reported activity of YSV, Wang et al. developed a conjugate (26) of YSV and folic acid.^[35] 26 self-assembles into nanoparticles at pH 7.0 and nanofibers at pH 5.0. Although the 26 is able to inhibit cancer cells, the IC₅₀ is still high (about 800 μM against HeLa at 48h) and the tumor inhibition effect is insignificant, likely due to that 26 is too hydrophilic to self-assemble at a low concentration.

As a further development of peptide-drug conjugate, Ryu et al. reported an elaborate conjugate (27, Scheme 8)^[36] that contains a carbonic anhydrase IX (CAIX) ligand and a mitochondria-targeting motif. A surprising feature of 27 is that it forms assemblies in lysosome despite carrying the triphenylphosphonium (TPP) targeting motif for mitochondria. According to the authors, 27 self-assembles into a fibrous aggregate with a negative surface charge in the extracellular environment near CAIX. Then, the surface charges of the nanofibers change to positive at the lysosome to cause lysosomal membrane disruption. The IC₅₀ of 27 against HeLa and MDA-MB-488 cells are about 40 and 14 μM, respectively. Although the cytotoxicity is moderate considering that 27 is a rather large molecule with the molecular weight of 1445 g/mol, the selectivity of 27 towards cancer cells is excellent. It is worthwhile to determine the structural changes of the peptide nanofibers of 27 when pH changes, as such insights would be highly valuable for further developing this molecular architecture.

Scheme 8.

Structures of peptide drug conjugates 27–30

Chen and Yang et al. reported hybrid nanoparticles consisting of fluorenylmethoxycarbonyl-arginine-glycine-aspartate (28) and hemin (29) (Fmoc-RGD/hemin, Scheme 8) as functional artificial enzymes for targeting cancer cells.^[37] The authors reported that the co-assembly of 28 and 29 forms nanoparticles that act as an artificial peroxidase to generate hydrogen peroxide for killing cancer cells. According to the authors, RGD provided a high binding affinity for cancer cells, Fmoc promoted self-assembly to form the nanoparticles and improved the catalytic activity of hemin by preventing dimerization. In addition to behaving like peroxidase, the nanoparticles allow colorimetrically detection cellular hydrogen peroxide. The authors suggested that the nanoparticles could suppress epithelial-to-mesenchymal transition by removing excess reactive oxygen species produced by transforming growth factor-β (TGF-β) in cells.

In another study that also employed RGD, Wang et al. reported a quite potent conjugate (30) against Lewis lung cancer (LLC) cells in vitro and in vivo.^[38] They conjugated pemetrexed (PEM), an FDA-approved multitarget antifolate drug, and RGD at the N-terminal and C-terminal of diphenylalanine to produce the conjugate (PEM-FFRGD (30)). While 30 exists as monomers at pH 7.5, it self-assembles to form nanoparticles at pH 5.0. 30 exhibits IC₅₀ of 0.2 μM against the LLC cells, but about 25% of the LLC cells persist even when the concentration of 30 reaches 50 μM, suggesting a population of cells failing to respond to PEM. Although the dosage required to slow down tumor growth is about 20 mg/kg by i.v. administration, the enhancement of the efficacy of PEM in the mice model is significant.

Stupp et al. reported a peptide conjugate (31, Scheme 9) consisting of DR5-binding peptide and self-assembling peptide amphiphile (PA) for inhibiting cancer cells.^[39] Specifically, the authors incorporated a TRAIL-mimetic peptide, WDCLDNRIGRRQCVKL,^[40] into peptide amphiphile to generate 31. TEM showed that 31 self-assembled into short nanofibers. The co-assembly of 31 and a pegylated PA result in longer nanofibers. The authors showed that 31 can inhibit MDA-MB-231 with IC₅₀ value about 5 μM, and the combination of 31 and pegylated PA exhibited slightly enhanced activity. The assemblies of 31 also exhibit a synergy with PTX. As a DR5 targeting carrier, 31 also showed antitumor activity in vivo when combined with PTX. It would be interesting to examine the in vivo stability of TRAIL-mimetic peptide, which is important for further developing this approach.

Scheme 9.

Structure of peptide drug conjugates 31.

Wang et al. also developed a self-assembling (32, Scheme 10) for inhibiting anaplastic thyroid cancer cells by targeting Akt1 through peptide binding.^[41] Based on the crystal structure of Akt, the authors selected a peptide with the sequence of SHPRSNSGSG, which binds to Akt with an association constant of about 1 μM. To produce 32, the authors attached six arginine residues (RRRRR) and the self-assembling motif (KLVFFAE) from amyloid to the N- and C-terminal of the Akt-binding peptide, respectively. The authors reported that 32 forms a structure of 0.84 nm, which is surprisingly small and requires re-examination. Incubating 32 with 8305C and 8505C cells showed the IC₅₀ values of about 18 and 12 μM, respectively.

Scheme 10.

Structures of peptide drug conjugates 32–35.

Hou and Huang deigned a peptide conjugate (33, Scheme 10) for targeting highly expressed prostate-specific membrane antigen (PSMA) in prostate cancer (PCa)^[42] by following similar design principle. The authors connected a PSMA-targeting small molecule (ACUPA^[43]) and the self-assembling motif (diphenylalanine) to a fluorescent segment DBT to produce the conjugate, DBT-2FFGACUPA (33). They also made two relevant control molecules, DBT-2FFG (34) and DBT-2ACUPA (35). The authors reported that the CMC values follow the trend of 34 (21.8 μM) < 33 (34.8 μM) < 35 (90.55 μM). The self-assembly of 33, 34, or 35 results in homogeneous nanofiber networks, dense wool-like nanofiber networks, and small aggregates, respectively. While 33 exhibits a IC₅₀ of about 200 μM against LnCaP cells, 34 or 35 shows little cytotoxicity even at 400 μM. Using a xenograft 22Rv1 tumor-bearing nude mouse model, the authors showed that 33 at a dosage of 13 mg/kg inhibits tumor growth. This work indicates that the tumor inhibiting effect largely originates from the assemblies of 33.

Maruyama et al. reported a novel hydrogel formed by a self-assembled peptide amphiphile (36, Scheme 11) based on intracellular pH (pHi).^[44] They found that C16-VVAEEE (N-palmitoyl-Val-Val-Ala-Glu-Glu-Glu, 36) forms a hydrogel that is very sensitive to a small pH change around pH 7. While the hydrogel of 36 forms at pH 6.8, it remains as a solution at pH 7.0. TEM revealed branched or entangled long nanofibers with a diameter of a few tens of nanometers in the hydrogel of 36 at pH 6.8. Incubating 36 with cells (MvE, A431, HeLa, and HEK293) that have different pHi shows that the cytotoxicity of 36 correlate with pHi of cells, with the most potent inhibition against HEK293 cells that has a pHi of 6.7. Based on fluorescent imaging, the authors reported that the assemblies of 36 accumulate at endoplasmic reticulum (ER) to cause cell stress and cell death. In vivo experiments revealed that the transcutaneous administration of 36 showed antitumor activity. This study illustrates that intracellular microenvironment can act as a trigger for the intracellular self-assembly of peptides.

Scheme 11.

Structures of peptide derivatives 36–42.

Song and Zhang et al. attached hydrocarbon chains to an antimicrobial peptide to inhibit B16F10 cancer cells both in vitro and in vivo.^[45] As shown in Scheme 11, the antimicrobial peptide CAMEL (CM15, KWKLFKKIGAVLKVL-NH₂) was modified with fatty acids of different chain length to produce four peptide amphiphiles, C4-CAMEL (37), C8-CAMEL (38), C12-CAMEL (39), C16-CAMEL (40). While 39 self-assembles to form nanoparticles, 40 forms nanofibers. Trypsin degraded CAMEL, 37, and 38 within one hour, but over 75% of 39 and 40 remained under the same condition. At the concentration of 10 μM, 37-40 each effectively killed 90% of B16F10 cells. While CAMEL and 39 exhibit strong membrane-lytic activity, 40 is the most potent among these peptides for inhibit the growth of B16F10 tumor in mice models. The in vivo toxicity of 40, however, is still modest, suggesting that a more careful engineering of the peptide sequence is needed.

To mimic biomolecular condensates formation inside cells, Wang et al. reported pH-responsive peptide conjugates (41 and 42) for minimizing drug resistance in cancer therapy.^[46] As shown in Scheme 11, 41 consists of a naphthyl group, a pH-responsive hexapeptide (VEALYL) derived from natural insulin, and a D-glucosamine. Replacing VEALYL in 41 by ^DV^DE^DA^DL^DY^DL produces 42. The CMC values of 41 at pH 7.4 and pH 5.0 are 530 and 102 μM, respectively; the CMC values of 42 at pH 7.4 and pH 5.0 are 191 and 30 μM, respectively. The authors found that 41 or 42 self-assembles to form nanoparticles at physiological pH. The nanoparticles partially turn into nanofibers at pH 5.0. The authors reported that pH decrease resulted in the conformational change of 41 or 42. When incubated with HeLa cells, 41 was less toxic than 42 at the concentration 50 μM, but they exhibited similar IC₅₀ values at about 343 μM (24h). The authors found that 41 or 42 was taken up by HeLa cells mainly through caveolae-dependent endocytosis, causing lysosomal hydrogelation of the peptide derivatives. According to the imaging of the cell uptake of the fluorescent analogs of 41 and 42, 42 accumulated more in the lysosomes due to proteolytic stability of the peptides. The authors also demonstrated that 42 plus Dox inhibited tumor growth in vivo in a murine model.

3. Peptides as enzyme substrates for self-assembly in situ.

Enzymatic formation of molecular nanofibers integrates molecular self-assembly with a wide range of biological processes, thus providing a unique opportunity to explore the intracellular self-assembly of peptides, a subject that was almost unexplored in chemistry and materials a decade and half ago. The ability to form self-assembled nanostructures inside cells, that is, self-assembly in vivo or in situ, offer a new way to examine the functions of peptides at a new level of complexity—supramolecular (higher-order structures with more than one building block) and intracellular (Figure 2). This conceptual advance thus provides an in-situ path to generate nanomedicine for controlling the fate of cells. To form the peptide nanofibers within a cell, a precursor that remains as individual molecules outside cells should turn into an amphiphilic molecule, such as amphiphilic peptides, by an intracellular enzymatic reaction. Then, the amphiphilic peptides self-assemble into the nanofibers or other higher-order structures intracellularly. An earlier reported esterase substrate^[6] meets this simple requirement. That EISA substrate selectively inhibits HeLa cells but is innocuous NIH3T3 cells. Though the concentration of peptide required is about 400 μM, the exceptionally selectivity against cancer cells originates from the difference in enzyme expression and molecular self-assembly, thus establishing the feasibility of enzyme-instructed self-assembly (EISA)^[47] for developing peptide assemblies for cancer therapy and illustrated the use of high-ordered structures of peptide for cancer therapy. Since then, considerable progresses have been made, the following section discusses the works appeared with last five year, roughly according to the chronological order.

Figure 2.

(A) Enzyme-instructed self-assembly (EISA) forming assemblies of peptides. (B) Intracellular EISA kills cancer cells.

3.1. EISA of peptides

While the assemblies of peptide-drug conjugates mainly depend on the activities from the drugs or drug candidates for inhibiting cancer cells, EISA inhibits cancer cells by the dynamic continuum of supramolecular assemblies of small molecules, including peptides, formed by enzymatic reactions. A key question is how to design the small molecules for EISA from the vast molecular space of peptides. Because the cancer inhibitory activity of EISA originates from the assemblies, self-assembling ability of peptides controls the anticancer activity of EISA, as reported in a study^[48] that examines similar substrates of EISA (43-48, Scheme 12). Each of the precursors consists of an N-capped D-tetrapeptide, a phosphotyrosine residue, and a diester or a diamide group. The CMC values of 43, 44, 45, 46, 47, and 48 are 30, 43, 36, 53, 74, 2000 μM, respectively. After being dephosphorylated by alkaline phosphatase (ALP), the corresponding products have the CMC values of 2.7, 4.4, 4.3, 4.9, 6.9, and 46.3 μM. At 24h, the corresponding IC₅₀ values against Saos2 cells are 3.6, 6.0, 7.6, 10, 28, and >200 μM. These results show that, regardless of the stereochemistry and the regiochemistry of the tetrapeptidic backbones, the anticancer activities of these precursors match their self-assembling abilities. Moreover, the resulting peptide assemblies from EISA cause cell death by disrupting cytoskeletons and plasma membranes. This work also indicates the diester or diamide derivatives of the D-tetrapeptides self-assemble intracellularly to cause cell death. Providing an important insight for developing a molecular dynamic continuum for cancer therapy, this work illustrates that CMC is an important molecular parameter to be tailored for cancer therapy by EISA.

Scheme 12.

Structures of EISA substrates 43–48.

The use of checkpoint blockade for cancer immunotherapy has made a considerable progress, but patients’ unresponsiveness to treatment is often due to immunosuppressive adenosine in the tumor microenvironment. One of the mechanisms for generating adenosine to suppress anticancer immunity is that certain cancer cells overexpress ALP, which dephosphorylates adenosine triphosphate to adenosine. However, ALP is considered “undruggable”, despite being identified as a cancer marker over five decades ago. Unlike molecular therapy that relies on enzyme inhibition, EISA catalyzed by ALP selectively inhibits tumors that overexpress ALP, and no ALP inhibition is needed. The advantage of EISA for inhibiting immunoresistant tumor was realized in a murine model recently.^[49] In that study, The phosphate on the tyrosine residue of 43 is cleaved by ALP to trigger self-assembly of the resulting peptides on and inside cancer cells, thus inducing cancer cell death. Bearing a diester at its C-terminal, 43 is also a substrate of CES. CES removes methyl ester to make the peptide hydrophilic again, so 43 exhibits little hepatotoxicity. Because 43 potently inhibits metastatic osteosarcoma (Saos2-lung) cells, the i.v. administration of 43 at the dosage of 15 mg/kg prevents Saos2-lung tumor growth in an orthotopic mice model. This works use EISA alone without the combination of any other drugs. It represents the first example of EISA of peptides to target immunosuppressive tumors in vivo, offering a promising approach to developing novel cancer therapeutics based on EISA.

Being the key requirement for molecular targeted cancer therapy, tight ligand-receptor binding paradoxically is also a major root of drug resistance in cancer chemotherapy. Since EISA kills cancer cells selectively without relying on tight ligand-receptor binding, using the catalysis of carboxylesterases (CES) to generate intracellular nanofibers can selectively inhibit a range of cancer cells that exhibit relatively high CES activities, as shown by a recent study.^[50] The substrates of CES are an enantiomeric pair of N-terminal capped diphenylalanine that conjugates with taurine as the precursors (i.e., 49 and 50, Scheme 13). Particularly, the enantiomer composed by D-amino acids (50) can resist the degradation of endogenous proteases for a long-term function of the peptide assemblies. MTT assays show that 49 and 50 inhibit drug resistant cancer cells (e.g., triple negative breast cancer cells (HCC1937) and platinum-resistant ovarian cells (SKOV3, A2780cis)) with the IC₅₀ values of 28–80 and 25–44 μg/mL, respectively. As expected, 49 and 50 were compatible with normal cells. This work also reported that intracellular EISA causes cell death by multiple modes, including apoptosis and necroptosis, a feature differs from conventional cancer therapy.

Scheme 13.

Structures of EISA substrates 49–53.

The synthesis and folding of numerous proteins, along with the regulation of intracellular calcium, lipid synthesis, and lipid transfer to other organelles, are among the responsibilities of the ER. Despite that the ER is considered as a promising target for cancer therapy, options for selectively targeting cancer cell ERs are limited. A recent study demonstrated that enzymatically created crescent-shaped supramolecular assemblies of short peptides can disrupt cell membranes and selectively induce cancer cell death by targeting the ER.^[51] The EISA precursor (51) consists of D-diphenylalanine as the self-assembling motif, a D-phosphotyrosine to serve as the enzyme trigger, an naphthylacetyl group as N-terminal cap, and a positively charged L-homoarginine. ALP dephosphorylate 51 to form the hydrogel of 52, which forms crescent-shaped assemblies with an average inner diameter of 8 nm and a width of 5 nm. The assemblies interact with synthetic lipid membranes, as evidenced by sedimentation assay. Live cell imaging demonstrates that the assemblies impair membrane integrity, and this is further supported by lactate dehydrogenase (LDH) assays. Incubation of 51 with three ALP expressing cancer cell lines, HeLa, A2780cis, OVSAHO gives an IC₅₀ of 24, 49, and 54 μM, respectively. The structure-activity relationship indicates that the crescent-shaped morphology is crucial for interacting with membranes and controlling cell fate. Additionally, fluorescent imaging shows that the assemblies accumulate in the ER. Time-dependent Western blot and ELISA results indicated that this accumulation causes ER stress and activates the caspase signaling cascade, resulting in cell death. In a related study, a positively charged peptide precursor (53), which is a substrate of typsin-1 (PRSS1), can undergo EISA at the ER of OVSAHO cancer cells to inhibit cell proliferation.^[52] These works show an in-situ form peptide assemblies at a plasma membrane and ER for developing anticancer nanomedicines.

While EISA is emerging as a promising approach for cancer therapy, the kinetics of EISA in the complex environments in or around cells receives little attention. A recent study attempted to link the kinetics and anticancer inhibitory activities.^[53] In that study, three dipeptidic precursors (54, 56, and 50 (Schemes 13 and 14)) were investigated. Upon CES catalyzed hydrolysis these precursors turn into hydrogelators (55, 57, and 58), which self-assemble in water at different rates. The precursors undergo intracellular EISA and selectively kill cancer cells, including high-grade serous ovarian carcinoma (HGSC) cells, with 56 and 50 to exhibit the lowest and the highest activities, respectively. This trend of inhibitory activity inversely correlates with the rates of converting the precursors to the hydrogelators in PBS. The analysis of kinetic modeling in this study yielded a few valuable insights. Firstly, the morphology of the nanostructures formed by hydrogelators and the rate of enzymatic conversion are influenced by the stereochemistry of the precursors. Secondly, reduced extracellular hydrolysis of the precursors promotes intracellular EISA within the cells. Thirdly, the cytotoxicity of intracellular EISA is largely determined by the inherent features, such as self-assembling ability and morphology, of the EISA molecules. This study represents the first kinetic analysis of intracellular EISA and sheds light on how stereochemistry can modulate EISA in the complex intra- and/or extracellular environment to develop molecular processes for anticancer applications. Furthermore, this analysis provides valuable insights into the kinetics and cytotoxicity of aggregates of abnormal proteins or peptides formed inside and outside of cells.

Scheme 14.

Structures of peptide derivatives 54–63.

Lin et al. used naphthalimide (NI) to replace naphthyl or Fmoc for EISA,^[54] which is the first example of using NI as the N-terminal cap for EISA. The authors compared two EISA substrates (Scheme 14), NI-_pY (59) and NI-F_pY (60), and found that 59 was unable to form a hydrogel by EISA, but 60 could. The authors examined the cLogPs of 59 and Fmoc-pY, an EISA substrate that formed a hydrogel upon the addition of ALP,^[55] and suggested that the hydrophilicity of 59 disfavored hydrogelation after EISA of 59. On the other hand, 60 is hydrophobic enough to allow hydrogelation after EISA of 60. One interesting result is that 60 is slightly more potent than Fmoc-F_pY against HeLa and MCF cells, likely because NI-FY self-assembles to form more rigid nanotubes after being generated by EISA of 60.

A promising advantage of EISA is to target immunosuppressive tumors, as discussed earlier in the case of metastatic osteosarcoma.^[49] Castration-resistant prostate cancer (CRPC) is another type of immunosuppressive tumors and remains a challenge for cancer immunotherapy. Because CRPC overexpresses prostate acid phosphatase (PAP), it is feasible to develop EISA for targeting CRPC, as shown in a recent study.^[56] The EISA precursor (61, Scheme 14) is a substrate of PAP. Replacing the phosphate group in 61 with phosphonate produces a control molecule, 62, which resists PAP. The CMC vales of 61 and 62 are comparable, at about 57 μM. PAP dephosphorylates 61 to generate 63, which exhibits the CMC of about 9 μM. The TEM images show that both 61 and 62 self-assemble to form irregular nanoribbon-like structures at pH 5.6. Upon the treatment with PAP, 61 turns into 63, which self-assembles to form uniform nanofibers with a diameter of 6 nm. Under the same PAP addition, 62 exhibits little morphology change. The MTT assay shows that 61 potently inhibits androgen-dependent prostate cancer cells (LNCaP) and CRPC cells (VCaP) with the IC₅₀ values of 58 and 55 μM, respectively. Although the potency of the EISA substrates remains to be improved, this work represents the first example of using PAP-instructed self-assembly of peptides for selective inhibiting CRPC cells.

In a broad sense, cancer stem cells and induced pluripotent stem cells (iPSCs) share distinctive features, such as perpetual proliferation and the overexpression of ALP. In fact, tumorigenic risk of undifferentiated human iPSCs remains a major obstacle for clinical application of iPSCs. This tumor risk associated with residual iPSCs requires effective approaches for selectively eliminating undifferentiated iPSCs without harming iPSC derived cells. EISA based on ALP can achieve this goal, as reported recently.^[57] As shown Scheme 15, an L-phosphopentapeptide (64) contains NBD, four leucine residues, and a tyrosine phosphate. While it is expected that ALP dephosphorylates 64 to form 65, a surprising result is that ALP of iPSC results in intranuclear accumulation of the assemblies of 65 for killing iPSC within two hours. 64 is compatible to normal cells (e.g., HEK293) and iPSC derived cells (e.g., hematopoietic progenitor cell (HPC)) because drastic reduction of ALP expression after iPSC differentiates to somatic cells. The cell lysate from normal cells (e.g., HS-5) can degrade the 64 rapidly before 65 reaches CMC, which explains the cell selectivity. This work represents the first case of intranuclear assemblies of peptides formed by EISA. Replacing the L-amino acid residues by the corresponding D-amino acid residues produces a D-peptide EISA substrate (66).^[58] Upon the dephosphorylation catalyzed by ALP, 66 turns into 67 to form intranuclear nanoribbons in osteosarcoma cells, such as Saos2 and SJSA1. The preliminary mechanistic investigation^[58] indicates that partial dephosphorylation of 66 results in micelles, which enter cells via endocytosis to interact with histone proteins, then form nanoribbons of 67 inside nuclei of Saos2 cells. While this study further establish EISA forming intranuclear peptide assemblies, the detailed mechanism remains to be elucidated.

Scheme 15.

Structures of peptide derivatives 64–67.

To precisely control the location of peptide assemblies inside cells, Wang et al. developed an innovative design of EISA substrate (68, Scheme 16) that employed a protected tyrosine phosphate.^[59] The purpose of protecting phosphate is to avoid undesired dephosphorylation at the cell membrane and cytoplasm, so that acid phosphatase (ACP) catalyzed dephosphorylation would occur at lysosome. As shown by the authors, 68 is stable at pH 7.4, it turns into 69 at pH 5.0, and ACP catalyzed dephosphorylation at pH 5.0 to generate 70, which self-assembles to form nanofibers. TEM showed that 68 hardly forms any observable nanostructures at pH 7.4, but it forms short and discrete nanofibers, which become thin nanofibers of 3.7 nm in diameters. After 24 h (or longer) of the addition of ACP at pH 5.0, thicker nanofibers (6.7 nm in diameters) form. Using fluorescent analogies, the authors confirmed that, at 4 hours of incubation, 68 results in lysosomal localization, while 69 leads to pericellular localization. Although cell assays showed that protecting tyrosine phosphate in 68 reduced cytotoxicity of the EISA precursors, this work underscored the use of reaction for spatiotemporal control of the peptide assemblies. In fact, 68 may serve as an oral EISA-based drug, which remains to be established.

Scheme 16.

Structures of peptide derivatives 68–70.

The Golgi apparatus (GA) is a vital component of intracellular transportation in mammalian cells. It functions as a hub for various signaling pathways that govern the survival and movement of cancer cells. Despite the growing recognition of Golgi as a significant target for cancer therapy, selective targeting Golgi of cancer cells is challenging. The recent exploration of EISA led to a thiophosphopeptide for selectively targeting Golgi of cancer cells.^[60] Specifically, changing an oxygen atom of the phosphoester bond in a phosphopeptide by a sulfur atom produces an EISA precursor (71, Scheme 17). While 71 exhibits CMC of 6.0 μM, 72, generated by dephosphorylation, has the CMC of 2.4 μM. TEM reveals that 71, at 5 μM, exists as micelles, which turn into nanofibers after ALP converts 71 to 72. Although incubating 72 with HeLa cells can slowly accumulate 72 at Golgi of the HeLa cells, the incubation of 71 with HeLa enable instantly targeting Golgi of the HeLa cells. Incubating 71 with several other cancer cell lines (Saos-2, SJSA-1, OVSAHO, HCC1937, HepG2, OVCAR-4, SKOV-3, MCF7) and immortalized normal cell lines (HEK293 and HS-5) indicates that the rates of Golgi-targeting agree with expression levels of ALP in these cells. Using naphthyl group to replace NBD in 71 generates 73, which kills the HeLa cells with the IC₅₀ about 3 μM. This result is noteworthy because no other drug molecules was conjugated to the peptide. Moreover, the introduction of a somewhat “caged” thiol group apparently provides a route for Golgi-targeting, as shown in the subsequent study in which the thiophosphate is replaced by a thioester to generate the substrates (74-75)^[61] of thioesterases. The peptide thioester (74 or 75), above or below its CMC, enters cells mainly via caveolin-mediated endocytosis or macropinocytosis, respectively. As the substrates of Golgi-associated thioesterases, such as PPT1, LYPLA1, or LYPLA2, 74 (or 75) turns into the corresponding thiopeptide 72 (or 77), which can form dimers and possibly react with cysteine-rich proteins for Golgi accumulation. Switching the chirality of diphenylalanine in 74 produces 75, which maintains the GA-targeting ability. The instant Golgi-targeting abilities of 74-75 excludes the possibility that Golgi-targeting is resulted from a specific ligand-receptor interaction. Unlike 71, 74 instantly targets Golgi of a variety of cells, including human, murine, Drosophila cells. Golgi-targeting by 74 disrupts protein palmitoylation at Golgi and redirects the distribution of proteins required palmitoylation for functioning, such GTPase NRas (NRAS). One plausible explanation of Golgi targeting of these thioesters is that the thioester group is a type of high energy bond that promotes peptide assemblies at Golgi. Obviously, the detailed mechanism remains to be elucidated. Nevertheless, these works provide a valuable starting point for selectively targeting Golgi of desired cells by EISA.

Scheme 17.

Structures of peptide derivatives 71–77.

While many reported studies use ALP for EISA, it is rare to employ kinase for EISA. Maruyama et al. reported an elegant design of a kinase substrate (78, Scheme 18) for EISA to inhibit the cancer cells that overexpress kinases.^[62] 78 is a tyrosine-containing peptide amphiphile (C16-EEEEY) that is phosphorylated by tyrosine kinase to form a phosphorylated peptide amphiphile (79, C16-EEEE_pY). The authors reported that, in the presence of calcium ions, 78 self-assembles to form nanosized strings with several hundred nanometers in length, and 79 results in nanofibers over micrometers. When incubated with HeLa, A431, MCF-7, and HepG2 cells, 78, at 0.05 wt %, showed higher cytotoxicity toward A431 cells than to the other cell lines. Since 79 did not show remarkable cytotoxicity toward any cell lines at the same concentration, the authors concluded that it was the in-situ formed 79 (due to kinase phosphorylating 78) caused the inhibition of A431 cells. That is, the self-assembled 79 induced ER stress for apoptosis. The authors also used mice models by peritumorally injection to demonstrate antitumor activity of 78. This work provided a cell selective approach to targeting kinases. Considering many specific and effective kinase inhibitors are already clinical anticancer drugs, the use of enzyme overexpressed in cancer cells for intracellular synthesis of an antitumor self-assembling drug is promising and exciting.

Scheme 18.

Structures of peptide derivatives 78–83.

Shang et al. reported the incorporation of a cell-targeting peptide into an EISA substrate (80, Scheme 18) for inhibiting breast cancer cells.^[63] Based on an earlier study that SKBR-3 breast cancer cells highly internalized a peptide with the sequence of VSSTQDFP,^[64] the authors developed 80, which contained the peptide and an EISA motif. The authors also found that 81, dephosphorylated by ALP, self-assembled to form a hydrogel at 0.4 wt%. The hydrogel contains the uniform nanofibers of 81, with a diameter of 9 nm. Notably, EISA generated 81 appears to adopt α-helix conformation, a phenomenon was first reported by Yang et. al.^[65] Using LC-MS to analyze the uptake of EISA-formed 81 in SKBR-3 and MCF-7 cells, the authors reported that the accumulation of 81 in SKBR-3 and MCF-7 is 2.54 and 0.26 μM, respectively. When using EISA-formed 81 for delivering taxol into SKBR-3 cells, the authors reported that the potency of taxol more than doubled. It would be worthwhile to examine the stability of 81 for the future optimization of this approach

Liang et al. reported an EISA substrate (82, Scheme 18)^[66] containing lysosomotropic morpholine for selectively inducing LMP.^[67] The authors used fluorescent pyrene and 2-morpholinoethan-1-amine to cap the N-terminal and C-terminal, respectively, of the peptide with the sequence of ^DF^DF^DEp^DY to produce 82. TEM revealed that ALP dephosphorylates 82 to form the nanofibers with diameters ranging from 63–85 nm, which are comparable to that of the nanofibers formed by 82 without addition of ALP but are wider that nanofibers formed by 83 (synthesized). At 72 h, 82 exhibited an IC₅₀ value of about 80 μM against HeLa cells. After 2 h incubation with HeLa cells, the fluorescence of 82 or 83 co-localized with the red fluorescence from Lyso-tracker with Pearson’s R values of colocalization 0.74 and 0.73, respectively, agreeing the lysosomotropic property morpholine. Interestingly, the authors reported that 82 also activated autophagy of the HeLa cells. In vivo experiments indicated that both 82 and 83, via i.p. administration, moderately inhibited HeLa tumor growth. To further improve this promising approach, the molecular design, in addition to lysosome targeting, probably needs to be tailored for enhancing LMP.^[68]

3.2. EISA of peptide-drug conjugates

A useful feature of EISA is the modulation of secondary structures of peptide before and after forming the peptide assemblies, especially in the case of short peptides.^[69] Recently, Yang et al. reported that such modulation of secondary structures is also important for the peptide assemblies to inhibit cancer cells in vitro and in vivo.^[65] As shown in Scheme 19, the authors synthesized CRB-G^DF^DF^DY (84) and CRB-G^DF^DF^D_pY (85) , and found that a heating–cooling process produces the suspension of 84 and EISA of 85 at 4 °C generates a hydrogel of 84. TEM images reveal nanoparticles in the suspension and nanofibers in the hydrogel. CD spectra indicates that 84 adopts the α-helix and β-sheet conformation in the suspension and the hydrogel, respectively. The authors reported that the nanofibers are slightly more stable that nanoparticles against proteinase K, and the cellular uptake of the nanofibers is about three times higher than that of nanoparticles. The enhancement of the cellular uptake is roughly proportional to the inhibitory activity of 84 against cancer cells, such as 4T1 and HeLa. It is also impressive that i.v. administrated 84 from nanofibers is even more potent than the parent drug, chlorambucil (CRB). Although the IC₅₀ value of 84 remains relatively high, the use of EISA at 4 °C to control the peptide conformation is an elegant approach for further development, which was realized in a subsequent study by Yang et al.^[70] The authors made a new precursor, CRB-HA-G^DF^DF^D_pY (86, Scheme 19), and used EISA to generate the hydrogel of 87 for co-assembling with an antibody (antiHER2). They found that the nanofibers double the uptake of anti-HER2 antibody in the NCI-N87 cells. The IC₅₀ values for NCI-N87 cells were 61.0, 24.9, and 15.1 μM for nanofibers of 87 without affibody, nanofibers with 10 and 15 wt% antibody, respectively. In a mouse tumor model, the supramolecular nanofibers slow down the HER2+ NCI-N87 tumor growth. This study illustrated the unique feature of EISA for cellular uptake and suggests that EISA may enhance uptake of antibodies, though the details mechanism remains to be elucidated.

Scheme 19.

Structures of peptide derivatives 84–87.

Some of the non-steroidal anti-inflammatory drugs (NSAIDs) that are used clinically are hydrophobic and contain aromatic ring, thus it is feasible to use NSAID as a N-terminal capping group of self-assembling peptide for developing EISA substrates.^[71] Xu et al. designed an EISA substrate, IDM-FF_pYSV (88, Scheme 20),^[72] by combining NSAID, anticancer tripeptide, and a self-assembling motif. Upon the ALP-catalyzed dephosphorylation, 88 turns into 89, which self-assembles in an aqueous solution to afford a hydrogel of 1.5 wt%. TEM revealed the nanofibers of 89 with the diameters of about 8 nm. Using proteinase K to treat the hydrogel of 89, the authors found that the release of IDM-OH was continuous and up to 15% in 72h. The authors reported that IC₅₀ values of 88 and 89 against HeLa cells are 162 and 289 μM, respectively. They used peritumoral injections of Rhodamine B-loaded solution of 88 to HeLa xenografted tumors in mice and found the persistent fluorescence signals. They suggested in-situ formation of the nanofibers of 89. Furthermore, the intravenous administration of the 88 formulation significantly inhibits the tumor growth in a HeLa-xenografted mouse model. This study demonstrated the integration of multiple anticancer elements within an EISA substrate for cancer therapy.

Scheme 20.

Structures of peptide derivatives 88–90.

Butle et al. conjugated osalazine (Olsa), an anti-inflammatory drug, to an enzyme-responsive peptide for theranostic applications.^[73] The peptide contained a furin cleavable motif, RVRR, and a disulfide protected 1,2-aminothiol group, which could be regenerated in the presence of glutathione (GSH), and 2-cyanobenzothiazole (CBT) at the C-terminal of the peptide. The key feature of the conjugate (90, Scheme 20) is the rapid biorthogonal condensation^[74] of CBT and 1,2-aminthiol that can be generated by furin cleaving 90. Another advantage is that the hydroxyl proton on Olsa provides a distinct contrast on chemical exchange saturation transfer magnetic resonance imaging (CEST MRI). Furin-triggered condensation produces a cyclic dimer of Olsa, which self-assembles to form nanoparticles. It was found in vivo that the CEST signal from 90 was 6.5-fold increased and anti-tumor therapeutic effect increased 5.2-fold compared to Osla without conjugating with the peptide. However, the modest tumor inhibition in mice model requires the dose of 90 to be 139 mg/kg, suggesting that more potent CEST molecules are needed. Since furin localizes at Golgi, it would be interested to further determine the subcellular location of the nanoparticles containing Olsa.

Liang et al. reported an ingenious design that use sequential enzyme catalysis for targeting cancer cells by EISA.^[75] The enzyme substrate, Nap-Phe-Phe-Lys(SA-AZD8055)-Tyr(H₂PO₃)-OH (91, Scheme 21), carries AZD8055 (an autophagy inducer) and tyrosine phosphate. Because AZD8055 is linked to lysine side chain via a succinic acid, 91 is a substrate of both ALP and CES. 91 self-assembles to form nanoparticles, which become nanofibers of 92 after being dephosphorylated by ALP. Further release of AZD8055 by CES catalyzed hydrolysis turns the nanofibers to nanofiber bundles of 93. The IC₅₀ value (72h) of 91 against Saos2 cells is about 57 μM. The authors found that 91 enhances the efficacy of Dox against Saos2 more than ten times. Then, they demonstrated that i.v. administration of 1 mg/kg of DOX and 4mg/kg of 91 effectively slow down Saos2 tumor progression in murine models.

Scheme 21.

Structures of peptide derivatives 91–96.

In a subsequent study,^[76] Liang et al. attached two different drugs, CPT and hydroxychloroquine (HCQ), to lysine side chains to generate 94 as the EISA substrates. Unlike AZD8055, HCQ is an autophagy inhibitor. Dephosphorylation of 94 produces 95, and hydrolytic release of CPT and HCQ generates 96. The authors found that 94 self-assembles to form nanobrushes, which become nanoparticles of 95 upon adding ALP, then CES converts the nanoparticles to the nanofibers of 96. After the incubation of 94 with 4T1 cells, the authors used TEM to show the formation nanoparticles and nanofibers inside the cells. 94 exhibited the IC₅₀ value of about 15 μM against 4T1 cells. Using intratumorally injection of 94 at the dosage of 6.7 μmol/kg, the authors demonstrated that 94 effectively inhibit the growth of 4T1 tumors in mouse models. In addition, the authors reported that the intratumoral injection of 94 can enhance anti-PDL1 immune checkpoint blockade therapy against the 4T1 tumors in mouse models. Although Dox is needed to demonstrate in vivo efficacy or intratumorally inject is used, this work underscores that the use of multiple enzymatic reactions to modulate EISA would become more important for cancer therapy.

Wang et al. reported a pathway-dependent supramolecular polymerization of a conjugate of peptide and hydroxycamptothecin (HCPT).^[77] As shown in Scheme 22, the conjugate (97) consists of HCPT, a self-assembling motif (FFY), and a laminin derived sequence (TK: TWYKIAFQRNRK). The purpose of TK is to bind with integrin α₆β₁. The authors phosphorylated the tyrosine in FFY or TK to generation two more conjugates (98 and 99) and made another conjugate (100) that lacks the FFY motif. They compared the efficacies of the assemblies of 97 formed by two different pathways: EISA to generate nanofibers of 97 or direct self-assembly to generate nanoparticles of 97. Incubating HCT116 cells with HCPT and 97-100 provided the IC₅₀ values, which follows the order of 98 (0.5 μM) < 99 < 97 < 100 < HCPT (7 μM). In vivo tumor inhibition in a murine model gives the same trend of inhibition. Fluorescent imaging confirms that HCPT enters cell nucleus in all cases. The authors suggested that EISA of 97 is an efficient pathway for using the TK sequence to mimic the binding of natural laminin to α₆β₁.

Scheme 22.

Structures of peptide derivatives 97–100.

Taking the advantage of EISA at 4 °C for enhancing cellular uptake,^[65] Yang and Li et al. reported the delivery of PMI and HCPT into nucleus of cancer cells.^[78] As shown in Scheme 23, they designed and synthesized two EISA substrates, 101 and 102, that bore NBD and HCPT at the N-terminal of GFF, respectively, and carried PMI at the C-terminal. ALP can dephosphorylate 101 and 102 to produce 103 and 104, respectively. The authors reported a quite interesting observation: the CMC values of 103 depended on the path of its formation. Forming by EISA at 4 °C, the CMC of 103 is about 73 μM, but the CMC value of 103 from heating-cooling cycle or EISA at 37 °C is about 120 μM. While the gel of 103 formed by EISA at 4 °C contains nanofibers with diameters of 20 nm, heating-cooling gives nanofibers with diameters of approximately 15–20 nm, and the hydrogel of 103 formed by EISA at 37 °C mainly contains worm-like micelles. The CD spectra also indicates that the hydrogel of 103 formed by EISA at 4 °C mainly contain α-helix, which favors cellular uptake. A similar trend was observed for 104 made from the three routes. Incubating the three different 104 with HepG2, A549 and U87MG shows that 104 formed by EISA at 4 °C exhibits the IC₅₀ values of 0.22, 0.26 and 0.87 μM, respectively, which are half and a-third of the IC₅₀ values of 104 made by the heating-cooling process and EISA at 37 °C, respectively. Murine model study also showed that 104 from EISA at 4 °C was most effective for inhibiting the growth of the HepG2 tumor.

Scheme 23.

Structures of peptide derivatives 101–104.

To further use EISA for optimizing nanomedicine, Yang et al. used a heptapeptide ATWLPPR (A7R), which specifically binds to NRP-1, to replace PMI, for making a new conjugate (105, Scheme 24).^[79] They also made HCPT-FFpYGRTWLPPA (106) with a scrambled sequence of A7R as a control. ALP dephosphorylation of 105 and 106 produces 107 and 108, respectively. They found that 107 formed using EISA at 4 °C exhibited lowest K_d value in binding with NRP-1 protein, about 8 μM, which were three orders of magnitude lower than that of A7R peptide. 107 formed using EISA at 4 °C exhibited the highest inhibitory capacity on branching junctions and total tube length in a HUVEC tube formation assay. 107 formed by EISA at 4 °C exhibited the lowest IC₅₀ value of 0.13 μM, which translates to that 107 at the dosage of about 16 mg/kg (i.v.) results in the most potent inhibition of HepG2 tumor growth in a xenograft murine model. Demonstrating EISA at 4 °C to enhance cellular uptake and nuclear accumulation of CPT or HCPT, these works underscored the unique advantage of EISA for optimizing anticancer nanomedicines.

Scheme 24.

Structures of peptide derivatives 105–108.

Continuing their ingenious design of tandem self-assembly that combine EISA, GSH reduction, and cell membrane targeting, Yang et al. recently reported the use of peptide assemblies to mimic extracellular vesicles (EVs) for targeting HepG2 tumors.^[80] As shown Scheme 25, the elaborate conjugate (109) consists of HCPT, GFF_pY, disulfide bond, and ERGD. The authors reported that 109, at 100 μM, form aggregates, which, upon dephosphorylation by ALP, turn into nanoparticles of 110 as the artificial EVs. GSH reduces the disulfide bond in 110 to cleave off the ERGD and to give the nanofibers of 111. The authors reported that the dynamic transformation of 110 to 111 inside HepG2 cells allows the efficient delivery of HCPT to the nuclei. The continuous shape change from extracellular nanoparticles (50–100 nm) to intracellular nanofibers (4–9 nm) appears to be important for nuclear delivery. This nuclear targeting of 109 enhanced the inhibitory effect of HCPT to HepG2 cells more than ten times. The authors reported that i.v. injection of 109 (in the dosage of the equivalence of 3 mg/kg HCPT) significantly inhibit HepG2 tumor growth. This work underscores the unique advantage of using enzymatic reactions to modulate the dynamics of supramolecular assemblies for spatiotemporal control, which becomes increasing critical in developing cancer therapy.

Scheme 25.

Structures of peptide derivatives 109–111.

3.3. EISA of peptides containing functional motifs

In another study that uses short peptide for EISA, Zhang et al. combined EISA with photochemistry for cancer cell photoregulation.^[81] The authors synthesized an EISA precursor (112, Scheme 26) modified by the biaryltetrazole with intramolecular photo-click reactivity. ALP catalyzed dephosphorylation of 112 produces 113, resulting in a hydrogel made of nanofibers. Photo irradiation leads to intramolecular photo-click reaction to form 114, which turns the hydrogel to a viscous liquid. The advantage of this EISA substrate is that it provides a way to monitor both EISA and photo-induced disassembly processes on cancer cells. This unique process cell may become useful for mechanistic understanding of the action of EISA in future research.

Scheme 26.

Structures of peptide derivatives 112–116.

Mitochondria is a crucial organelle that plays essential roles in multiple cellular processes. Thus, targeting mitochondria is emerging as an important approach for developing therapeutics. Most of the molecules reported for targeting mitochondria are lipophilic and cationic, and they lack cell selectivity. EISA of a negatively-charged branched peptide (115, Scheme 26)^[82] turns out to be able to target mitochondria of cancer cells. Consisting of a well-established protein tag (i.e., FLAG-tag)^[83] and a self-assembling motif made of D-amino acid residues, 115 forms micelles. Enterokinase (ENTK) cleaves the hydrophilic FLAG motif (DDDDK) from 115 to generate 116, thus turning the micelles to nanofibers. After cancer cells uptake the micelles, intracellular ENTK catalyzed EISA locates the nanofibers mainly at mitochondria. The micelles of 115 can deliver cargos (either small molecules or proteins) into mitochondria within 2h. Such mitochondria delivery relocates doxorubicin (Dox) to mitochondria and enhance the Dox potency eight times against HeLa cells. Moreover, 115 interacts with the nuclear location sequence of H2B to block it from entering the nucleus. Subsequently, EISA of 115 facilitate H2B entering the mitochondria.^[84] In addition, 115 can facilitate gene delivery for mitochondrial protein expression and mitochondrial genome editing in a cancer-specific manner.^[85] These results illustrated EISA as a fundamentally new way to target mitochondria of cancer cells.

In an effort to improve the efficacy of EISA for inhibiting cancer cells, Fan and Ren et al. introduced YSV, the reported HDAC modulator, to self-assembling Nap-G^DF^DF to produce an EISA substrate (Nap-G^DF^DF_pYSV, 117, Scheme 27) for inhibiting cancer cells.^[86] The authors reported that ALP catalyzed dephosphorylation produced hydrogel of 118 at 0.4 wt%. TEM images show long β-sheet nanofibers entangled with each other in the hydrogel. Interestingly, 117 can be dephosphorylated nearly 100% by ALP even at 4 °C. The authors tested the cytotoxicity of 117 and 118 against HeLa, A549, A2780, SKOV3 and L929 cells, in which L929 was a normal cell line. While the IC₅₀ values of 118 against HeLa, A549, A2780, SKOV3 and L929 were 50, 20, 112, 56 and 42 μM, respectively, the IC₅₀ values of 117 against L929, A549, A2780 and SKOV3 cells increased by approximately 10, 20, 4 and 3-times, respectively. But the IC₅₀ value of 117 against HeLa increased little. This result indicates that the inhibitory activity originates from the generation of YSV by dephosphorylation. Subsequently, the same lab used 117 to enhance the sensitivity of cancer cells to ionizing radiation (IR).^[87] These studies, indeed, demonstrated the excellent cell selectivity of EISA toward cancer cells that overexpressed ALP.

Scheme 27.

Structures of peptide derivatives 117–122.

Yang et al. pioneered the introduction of ligand-receptor interactions to EISA substrates to improve the efficacy of EISA. In a recent example, they took advantage of the overexpression of both ALP and cholecystokinin-2 receptor (CCK2R) for inhibiting HeLa and HepG2 cells.^[88] They designed the peptide derivative NBD-GFF_pYG-CCK6 (119, Scheme 27) as an EISA precursor. The CCK-6 peptide (Nle-Gly-Trp-Nle-Asp-Phe) was a specific ligand of CCK2R, which was overexpressed in cancer cell membrane. According to the authors, ALP converts 119 to 120, which self-assembles into short nanofibers with a diameter of about 8 nm. They found that the expression level of extracellular ALP was in the order of HeLa ~ HepG2 > HCT116 > A549 > LO2 and that the expression of membrane CCK2R was in the order of HeLa ~ HepG2 ~ HCT116 > A549 > LO2. They observed yellowish-brown hydrogels at the surfaces of HeLa and HepG2 cells but not around the other three cell lines. They determined that the IC₅₀ values of 119 against HeLa, HepG2, HCT116, A549, and LO2 cells were 42, 32, 75, 154, and 270 μM, respectively. These results underscored the importance of both EISA by ALP and the overexpressed state of CCK2R in the pericellular formation of hydrogels around HeLa and HepG2 cells.

Yang et al. recently reported the first example of the use of EISA for identifying and removing senescent cells.^[89] They took advantage of β-galactosidase (β-Gal) instructed peptide self-assembly^[90] and designed NBD-FFY(gal)G (121, Scheme 27) as a precursor of a hydrogelator, NBD-FFYG (122). While 121, at 0.5 wt%, is a solution containing short nanofibers with diameters of approximately 8 nm and lengths of less than 60 nm, adding β-Gal turns the solution to the hydrogel of 122, with diameters of 5 nm and lengths longer than 400 nm. The CMC values of 121 and 122 are about 446 and 171 μM, respectively. Using cisplatin to induce the senescence of HeLa cells with high expression of galactosidase, the authors incubated the cells with 121 or 122 at 200 μM and found yellow fluorescent dots around the nucleus in the cells treated with 121, but much less fluorescence in the cells treated with 122. For non-senescent normal HeLa cells, there was negligible yellow fluorescent in the cells treated with 121 or 122. The authors reported that the EISA-formed nanofibers alleviate endothelial cell senescence by reducing p53, p21, and p16INK4a expression levels. It would be interesting to develop more effective β-galactosidase substrates for removing senescent cells in vivo.

To evoke immunogenic cell death (ICD) or to convert immunologically cold tumors to hot, Ding et al. developed an elegant EISA substrate (123, Scheme 28)^[91] for induced lysosomal membrane permeabilization (LMP). The conjugate consists of two tyrosine phosphates, a fluorophore that becomes more fluorescent when its intramolecular bond rotation slow down at solid state, and a mitochondrion targeting motif TPP. After the ALP catalyzed dephosphorylation of 123, the resulting 124 self-assembles to form nanoparticles with the diameter of 310 nm. The presence of two tyrosine phosphates appears to override the mitochondria-targeting effect of TPP and drives the assemblies of 124 in lysosomes. The IC₅₀ values of 123 against HeLa cells without and with the light irradiation are about 10 μM, but the light irradiation enhances the cytotoxicity of 123 against 4T1 cells, lowering the IC₅₀ value from above 20 μM to 15 μM. 123 exhibits little cytotoxicity to HEK293 cells without and with light irradiation. By examining the secreted ATP, the authors concluded that EISA plus light enhances ICD. Using the orthotopic 4T1 breast tumor-bearing murine model for intratumorally injection at the dosage of 2 mg/kg, the authors reported that photodynamic therapy using 123 is the most efficacious in suppressing the tumor. This work illustrates the use of EISA-induced LMP cell death for improving cancer immunotherapy in combination with photodynamic therapy.

Scheme 28.

Structures of peptide derivatives 123–126.

Yang et al. also observed the nanofiber to nanoparticle transition triggered by ALP when they conjugated TPP to the side chain of a simpler peptide to make a mitochondria-targeting EISA substrate (125, Scheme 28).^[92] Another unique feature of 125 is that the N-terminal capping group is flurbiprofen (Fbp), an anti-inflammatory drug. TEM revealed that the solution of 125 (600 μM) contains nanofibers with diameters 15–20 nm. The addition of ALP turned the solution opalescent, which contains nanoparticles of 126 with the diameters about 100–150 nm, based on TEM. 125 increased ROS levels of mouse colon carcinoma CT26 cells. According to the Bio-TEM image, the authors suggested the nanoparticles of 126 within the mitochondria of the cells. Using CT26 tumor bearing BALB/c mouse model, accompanied with a combination of immune checkpoint blockade (ICB) therapy using anti-PD-L1 agents, the authors showed that 125, at the dosage of 10 mg/kg, is more effective than anti-PD-L1 at the dosage of 75 μg/mouse. Moreover, the combination of 125 and anti-PD-L1 exhibited significant better tumor suppression. Notably, this work showed that 125 alone can promote dendritic cell recruiting and T-cell infiltration, which worth more detailed mechanism elucidation.

Wang et al. reported the use of cathepsin B (CTSB) to generate peptide assemblies in lysosomes, inducing cell death via LMP.^[93] Based on that CTSB cleave peptides containing the sequence of Arg-Arg-Gly-Lys (RRGK), the authors produced eight peptide derivatives (Schemes 29 and 30): Nap-WYFRRGK (127), Nap-WYFRRGK-Man (128), Nap-wyfttGk (129), Nap-wyfttGk-Man (130), Nap-wyfRRGK (131), Nap-wyfRRGK-Man (132), Nap-wyfGFRARGK (133), and Nap-wyfGFRARGK-Man (134). The authors incubated these peptides with U87MG, HS-5, and HeLa cells, and reported that the IC₅₀ values of 127, 128, 131, 132, and 134 against U87MG cells are 143.2, 58.7, 59.3, 59.2, and 38.5 μM, respectively, while 133 is compatible with U87MG. The IC₅₀ values of these precursors against HS-5 cells are more than two times greater than against U87MG cells. The IC₅₀ values against HeLa cells follows the trend of 134 (35 μM) < 132 (67 μM) < 133 (93 μM) < 131 (199 μM) = 128 (199 μM) < 127 (>500 μM). These results are consistent with that glycosylation at the C-terminal of L-peptides increase proteolytic stability of the peptide derivatives.^[94] The authors also confirmed that CTSB, indeed, results in the peptides with lower CMC values. Although the IC₅₀ values are still relatively high, this work clearly established the peptide assemblies in lysosomes result in cell death.

Scheme 29.

Structures of peptide derivatives 127–130.

Scheme 30.

Structures of peptide derivatives 131–134.

Aiming to degrade PD-L1, Yang et al. developed four EISA substrates (135-138, Scheme 31)^[95] that contain a D-peptide (^DPPA-1) with high binding affinity to PD-L1. These substrates differ in the number of glycine residues at the C-terminal of the peptides, with 135, 136, 137, and 138 containing 4, 3, 2, and 1 glycine residues, respectively. Upon the addition of ALP, the corresponding products are 139, 140, 141, and 142. At the concentration of 0.3 wt%, 139 remained a clear solution, 140 or 141 became viscous, and 142 precipitates significantly. TEM shows that adding ALP turns nanoparticles of 137 to nanofibers of 141, with diameters of about 15 nm. The authors also reported that the materials obtained by treating 137 with ALP showed the highest binding affinity for mPD-L1 and suggested multivalent binding sites on the nanofibers of 141. Using 4T1 cells that show high expression of extracellular ALP and membrane proteins of PD-L1, the authors demonstrated that EISA of 138 resulted in selective formation of nanomaterials around PD-L1 on the cell membrane, which promoted PD-L1 degradation in proteasome. In vivo studies reveal that 138, at the i.v. dosage of 15 mg/kg, is the most effective EISA substrates to inhibit tumor growth in 4T1 tumor-bearing mouse model in this study. Although 138 appears to be less effective than the reported ^DPPA-1, this approach could be more useful for cancer cells express higher ALP than 4T1 does.

Scheme 31.

Structures of peptide derivatives 135–142.

Boron neutron capture therapy (BNCT) is radiotherapeutic modality based on the thermal neutron capture and fission reactions of ¹⁰B for destroying tumor cells. A key requirement of BNCT is to accumulate ¹⁰B in tumors. Gao et al. reported the first example of applying EISA to selective accumulation of boron agents in cancer cells for BNCT.^[96] As shown in Scheme 32, they incorporated borylphenylalanine (BPA) in a phosphorylated self-assembly motif to produce the EISA substrates, in which BPA connected to the C-terminal and the side chain of the peptide to form 143 and 144, respectively. While ALP converts 143 to 145, which forms nanofibers rapidly, 146, made by the dephosphorylation of 144, self-assembles to form nanoparticles, then nanoparticles turn into nanofibers. The authors showed that EISA enables the accumulation of 145 or 146 inside the HeLa cells incubated with 143 or 144, respectively. Among these two EISA substrates, 143 provides faster accumulation of boron inside cells. It would be worthwhile to determine the dephosphorylation rate of the EISA substrates for the further optimizing of EISA for BNCT.

Scheme 32.

Structures of peptide derivatives 143–148.

Cai et al. reported the use of EISA to enhance the monomer–excimer transition of a coumarin (Cou) for reporting ALP activity in vitro and in vivo.^[97] The authors conjugated to a short self-assembling peptide with a hydrophilic ALP-responsive group for producing the EISA substrates (147, Scheme 32). Based on photoluminescence (PL) measurement, the authors reported that the excimer fluorescence from Cou was enhanced after ALP dephosphorylating 147 to form 148, and they attributed that the excimers are from the nanofibers of 148. By incubating 147 with HeLa cells, the authors confirmed that luminescent supramolecular nanofibers of 148 formed on the HeLa cell surfaces. Impressively, the authors showed that intercellular bridges formed by EISA between some HeLa cells. Moreover, the authors used the HepG2 tumor bearing murine model to demonstrate the excimer formation in vivo. It would be worthwhile to develop this approach for in vivo imaging using other imaging modality, such as MRI.

Yang et al. recently engineered three peptide (149-151, Scheme 33)^[98] containing EISA motif and an EGFR binding sequence, YHWYGYTPQNVI^[99]. The three peptides differ in the position of tyrosine phosphorylation. The authors found that 149 was the most effective to bind cell membrane of HeLa cells, which probably because phosphorylation at the leftist tyrosine barely disrupted the EGFR binding and ALP dephosphorylated 149 fastest. TEM showed that ALP converts 149 to 152, which form bundles of nanofibers. Incubating HeLa cells with 149 results in 152 on the cell membrane, but incubating HeLa cells with 150 or 151 results in cell uptake of those two peptides. They also used peptide-protein and peptide-peptide co-assembly strategies to apply two types of antigens, namely ovalbumin (OVA) protein and dinitrophenyl (DNP) hapten respectively, on cancer cell membranes. Although the in vivo stability of the EGFR-binding peptide needs to be improved for further development, this work demonstrated that the use of EISA for spatiotemporal control of peptide assemblies. Such a dynamic control promises a fruitful opportunity for engineering EISA for cancer immunotherapy.

Scheme 33.

Structure of peptide derivatives 149–154

The combination of EISA of peptides with photodynamic therapy (PDT) is a straightforward approach that has received considerable attention. Recently, Liang et al. reported the combination of caspase-catalyzed self-assembly of peptide with PDT for potentially treating oral squamous cell carcinoma (OSCC).^[100] Based on the upregulation of caspases in apoptosis, they reckoned that apoptosis would amplify the assembly of porphyrin nanofibers via EISA for enhanced PDT of OSCC. The water-soluble porphyrin derivative, Ac-Asp-Glu-Val-Asp-Asp-TPP (Ac-DEVDD-TPP, 153, Scheme 33), which contains the DEVDD sequence for cleavage by caspase-3. After being cleaved by caspase-3, 153 becomes D-TPP (154). The CMC of 153 and 154 in PBS were 294.4 and 3.8 μM. 154 self-assemble into porphyrin nanofibers with the diameters of about 20 nm. While cisplatin (20 μM) plus 153 (10 μM) hardly inhibit SCC7 cells at dark state, the addition of laser irradiation was able to kill 90% of the cells. The authors suggested that a possible pyroptosis mechanism, in addition to apoptosis, is responsible for the cell death. This observation is consistent with EISA, which usually lead to multiple cell death modes. It would be worthwhile to use cryo-ET to determine the location of the porphyrin nanofibers for the optimization of this approach.

Perspective and Outlook

Over the last one and half decade, peptide assemblies have emerged as a promising class of molecules for developing anticancer therapeutics.^{[6, 49, 103]} In addition to serve as a component of peptide-drug conjugates and substrates for EISA, peptide assemblies also are being actively explored for immunotherapy,^[104] such as self-adjuvants^[105] for vaccine, antigens for modulating immune responses,^[106] and media for delivering antibodies and cells^[107]. Furthermore, elaborated peptide assemblies, such as modulated disassembly of peptide assemblies^[108], are advancing the controlled drug releases.

To further explore the potential of peptide assemblies for cancer therapy, several limitations remain to be addressed, such as potency, mechanism, and structures. Potency: As shown in Tables 1 and 2, the most potent peptide assemblies are those conjugates containing CPT or HCPT. However, it is likely that the potency of those conjugates originates from the drugs rather than the peptides themselves. The most potent peptide assemblies that do not contain drugs are currently made of 43, with the IC₅₀ of 5 μM and in vivo effective dosage of 15 mg/kg against Saos2 cells or tumors, respectively. It is encouraging that 43 is able to inhibit the growth of osteosarcoma tumors in a murine model. Further molecular engineering will generate EISA precursors with IC₅₀ in nanomolar range without employing known drug candidates. Mechanism: The dynamic nature of peptide assemblies in vivo presents a challenge in elucidating the cell mechanism of their actions, especially in the understanding of the molecular mechanism of adjuvants made of peptide assemblies. Given the similarity of peptide assemblies with β-amyloids, the advancement in the mechanistic investigation of β-amyloids may provide insights for addressing the challenge. On the other hand, developing a general approach to elucidate the interaction of peptide assemblies and proteins^{[103, 109]} may help understand the pathogenesis of β-amyloid. Structures: The atomistic structures of functional peptide assemblies have been an obstacle in understanding the structure-activity relationship. The “resolution revolution” of cryo-EM has greatly advanced the structural elucidation of peptide nanofibers, as recently summarized by Egelman et al.^[10] The use of cryo-EM for determining the atomistic structures of peptide assemblies^[110] will be critical for understanding the molecule interaction of peptide assemblies with other biological molecules in the future, particularly inside cells.^[111]

Table 1.

Summary of the efficacy of the peptide assemblies for inhibiting cancer cells and tumors.

Molecule No.	IC50 (μM)/(μg/mL)	Cell lines	Tumor inhibition (mg/kg)	Administration route	Ref. No.
3	0.22 (0.20)	A549			[11]
	0.28 (0.25)	H460
5/6	(<5)	4T1	2.0	i.v.	[12]
9	49 (53.1)	4T1	27	i.v.	[13]
	109 (118.1)	AML12
10	1.3 (3.1)	4T1	56
	29(65.1)	AML12
11	1.1 (1.3)	LNCaP-FGC		i.v.	[14]
	36.0 (42.3)	HepG2
	13.5 (15.9)	HeLa
	7.2 (8.5)	MCF-7	10
	6.9 (8.1)	DU145
	11.8 (13.9)	PC-3
	> 40.0 (>47.1)	LO2
12/HCPT	0.087[a] (0.032)	A549	6.0	s.c.	[17]
13	< 2.5(<3.2)	A549	6.3	s.c.	[20]
	<5(<6.3)	HeLa
	<20(<25.2)	MCF-7
	<2.5(<3.2)	CEM/C1
14	0.2 (0.6)	HepG2	11.6	i.v.	[21]
15	71 (179.3)		10.4
16	0.94 (2.5)		10.6
17	0.47 (1.3)		11.6
18	0.57 (1.4)		10.4
19	40 (77.6)	PA-1	N/A	N/A	[22]
	120 (233)	Huh-7
	90 (175)	HT-29
	>80 (>155)	293T
19/AS1411	10(19)	PA-1
	80 (156)	Huh-7
	50 (97)	HT-29
	>80 (>155.3)	293T
20	(12)	HCT-116	10	i.p.	[25]
	(<10)	HT-29
21	(>80)	4T1		i.v.	[26]
22	1.3 (3.8)	MCF7			[31]
	1.9 (5.5)	A375
	0.6 (1.7)	HCT116 p53+/+	3	i.p.
	>10 (29.1)	HCT116 p53−/−
23/Dox		MCF-7	5	i.v.	[33]
	<10 (<22.5)	HUVEC
23	>100 (225)	HUVEC
24	103 (114.0)	BEL-7402	10	i.v.	[34]
	54 (59.8)	HeLa
	28 (31.0)	MCF-7
25	72 (79.7)	BEL-7402	10
	30 (33.2)	HeLa
	22 (24.4)	MCF-7
26	>1600 (1677)	NIH-3T3		i.v.	[35]
	~800 (886)	HeLa
		H22	15
27	40 (57.8)	HeLa	N/A	N/A	[36]
	17 (24.6)	HeLa (hypoxia)
	14 (10.2)	MDA-MB-468
	40 (57.8)	4T1
	140 (202.4)	NIH/3T3
	>100 (>144.6)	NIH/3T3 (hypoxia)
	100 (144.6)	HEK293T
28/29	>15 (>18.3)	MCF-7	12.2	s.c.	[37]
30	0.2 (0.2)	LLC	20	i.v.	[38]
	0.2(0.2)	A549
31	5 (25.7)	MDA-MB-231	3.5	i.v.	[39]
	>10 (>51.3)	T74D
32	18.2 (50.2)	8305C			[41]
	12.4 (34.2)	8505C
33	>400 (787.3)	si-PSMA-transfected LNCaP			[42]
	211.8 (416.9)	LNCaP
	~400 (~787.3)	22RV1	13	i.t.
	>400 (>787.3)	PC-3
34	>400 (>546.2)	LNCaP
		22RV1	9	i.t
35	>400 (506.2)	LNCaP
		22RV1	8	i.t
36	640 (>580)	MvE			[44]
	640 (>580)	A431
	510 (~470)	HeLa	14.1	t.d.
	380 (~350)	HEK293
37	8.4 (15.5)	MCF-7			[45]
	7.8 (14.4)	T-47D
	11.7(21.5)	U251
	6.4(11.8)	U87
	6.8 (12.5)	B16-F10	10.5	i.t.
38	7.1(13.5)	MCF-7
	5.4 (10.2)	T-47D
	6.4 (12.1)	U251
	4.5 (8.5)	U87
	6.5 (12.3)	B16-F10	10.5	i.t.
39	6.6 (12.9)	MCF-7
	5.3 (10.3)	T-47D
	6.3 (12.3)	U251
	5.1(9.9)	U87
	5.2 (10.2)	B16-F10	10.5	i.t.
40	5.5 (11.0)	MCF-7
	6.0 (12.1)	T-47D
	7.3 (14.7)	U251
	2.7(5.4)	U87
	7.5 (15.1)	B16-F10	10.5	i.t.
41	~316 (327)	HeLa (24h)	20	i.v.	[101]
	>562 (>582)	HeLa (48h)
	>562 (>582)	HS-5 (24h)
	>562 (>582)	HS-5 (48h)
	~200 (~207)	SK-OV-3 (with presence of 2.5 μM sunitinib)	30	i.v.
42	343 (355)	HeLa (24h)	20	i.v.
	196 (203)	HeLa (48h)
	>562 (>582)	HS-5 (24h)
	>562 (>582)	HS-5 (48h)
		SK-OV-3	30	i.v.

Table 2.

Summary of the efficacy of enzyme-instructed assemblies for inhibiting cancer cells and tumors.

Molecule No.	IC50 (μM)/(μg/mL)	Cell lines	Tumor inhibition (mg/kg)	Administration route	Ref. No.
43	4.4 (3.9)	Saos-2			[48]
	105 (92.4)	MCF-7
	>200 (>176)	T98G
	>200 (>176)	HS-5
	>200 (>176)	HepG2
44	6.0 (5.3)	Saos-2
45	7.6 (6.6)	Saos-2
46	10.2 (9.0)	Saos-2
47	44.3 (38.9)	Saos-2
	53.5 (47)	MCF-7
	~200 (176)	T98G
	~50 (44)	HS-5
	111 (92.4)	HepG2
48	>200 (>151)	Saos-2
43	8.4 (7.4)	Saos-2-luc	15	i.v.	[49]
	3.2 (2.8)	Saos-2-lung	15	i.v.
	>1000 (881)	HepG2
49	40 (29)	HCC1937			[50]
	39 (28)	MCF-7
	67 (49)	A2780
	53 (39)	A2780cis
	63 (46)	SKOV3
	73 (53)	HeLa
	109 (80)	Saos-2
	55 (40)	MES-SA
	441 (322)	MES-SA/Dx5
	129 (94)	A375
	133 (97)	HepG2
	>499 (>365)	U87MG
	>499 (>365)	T98G
	135 (99)	HS-5
	145 (106)	PC12
50	36 (26)	HCC1937
	34 (25)	MCF-7
	51 (37)	A2780
	49 (36)	A2780cis
	42 (31)	SKOV3
	37 (27)	HeLa
	60 (44)	Saos-2
	42 (31)	MES-SA
	52 (38)	MES-SA/Dx5
	120 (88)	A375
	133 (97)	HepG2
	172 (126)	U87MG
	198 (145)	T98G
	68 (50)	HS-5
	85 (62)	PC12
51	24 (21.5)	HeLa	N/A	N/A	[51]
	49 (43.8)	A2780cis
	54 (48.3)	OVSAHO
	>500 (>447)	HS-5
53	380 (708)	OVSAHO	N/A	N/A	[52]
	>400 (>745)	PC3
	>400 (>745)	SKOV3
	>400 (>745)	A2780res
	~400 (~745)	U87MG
	~400 (~745)	K562
	>400 (>745)	HOSE636
53+50 nM BTZ	106 (197.5)	OVSAHO
53+100 nM BTZ	142 (264.6)	OVSAHO
54	68 (~50)	KURAMOCHI	N/A	N/A	[53]
	68 (~50)	OVSAHO
	68 (~50)	JHOS-2
	41(~30)	JHOS-4
	82 (~60)	HeLa
	109 (80)	HS-5
56	109 (~80)	KURAMOCHI
	109 (~80)	OVSAHO
	109 (~80)	JHOS-2
	96 (~70)	JHOS-4
	96 (~70)	HeLa
	160 (117)	HS-5
50	33 (24)	KURAMOCHI
	34 (~25)	OVSAHO
	34 (~25)	JHOS-2
	27 (~20)	JHOS-4
	41 (~30)	HeLa
	79 (58)	HS-5
60	~150 (97)	HeLa			[54]
	~200 (129)	MCF-7
61	58 (44.5)	VCaP			[56]
	55 (42.2)	LNCaP
	107 (82.0)	HS-5
62	>200 (>150)	VCaP
	>200 (>150)	LNCaP
	>200 (>150)	HS-5
64	<200 (<190)	iPSCs			[57]
	>400 (>379)	iPS-derived HPCs
	>400 (>379)	HS-5
	>400 (>379)	HEK293
66	>400 (>379)	iPSC
71	38 (26)	HeLa			[60]
73	2.8 (1.7)	HeLa
	>100 (>62)	HEK293
	>100 (>62)	HS-5
74	15.0 (9.7)	HeLa			[61]
	10.9 (7.1)	HEK293
	9.5 (6.2)	Saos-2
	14.5 (9.4)	MCF-7
	14.9 (9.7)	HS-5
	22.5 (14.6)	SJSA-1
	>100 (>65)	HepG2
75	~20 (~13)	HeLa
78	>534 (>500)	MvE			[76]
	~534 (~500)	HEK293
	~534 (~500)	HeLa
	~167 (~250)	A431	22.5	i.t.
	~534 (~500)	HepG2
	~534 (~500)	MCF-7
79	>492 (>500)	HEK293
	>492 (>500)	HeLa
	>492 (>500)	A431
	>492 (>500)	HepG2
	>492 (>500)	MCF-7
80	0.59 (1.04)	SKBR-3			[63]
	0.88 (1.55)	MCF-7
82	80.1 (90.0)	HeLa	15	i.p.	[66]
83	~200 (~209)	HeLa	15	i.p.
84	<85 (<70)	MCF-7			[65]
	<85 (<70)	4T1	4	i.v.
	<85 (<70)	A549
	<85 (<70)	HeLa
	<85 (<70)	HepG2
85 (4°C)	<50 (<45)	MCF-7
	<50 (<45)	4T1	4.4	i.v.
	<50 (<45)	A549
	<50 (<45)	HeLa
	<50 (<45)	HepG2
87	61 (53)	NCI-N87	3.65	i.v.	[70]
	~80 (70)	SKOV-3
	~100 (88)	BT474
	~70 (61)	PC-3
	>500 (438)	LO2
87+10wt% antibody	24.9 (21.8)	NCI-N87	3.65	i.v.
	~30 (26)	SKOV-3
	~60 (53)	BT474
	~70 (61)	PC-3
	>500 (438)	LO2
87+15wt% antibody	15.1 (13.2)	NCI-N87	3.65	i.v.
	~20 (18)	SKOV-3
	~30	BT474
	~75 (66)	PC-3
	>500 (438)	LO2
88	162.4 (175.6)	HeLa	20	i.v.	[72]
89	289.4 (289.8)	HeLa
90	~250 (347)	HCT116	138.7	i.v.	[73]
	>250 (347)	LoVo
	>250 (347)	CCD-18Co
91	56.8 (84.6)	Saos-2	4	i.v.	[75]
	~200 (~262)	HepG2
	~300 (~394)	HeLa
	~600 (787)	HPAEpic
	~250 (328)	L02
94	15.7 (31.3)	4T1	13.4	i.t.	[76]
97	~1 (2.7)	Caco-2			[77]
	~1 (2.7)	HCT116	20	i.v.
	~25 (66.7)	LO2
98	~0.5 (1.4)	Caco-2
	~0.5 (1.4)	HCT116	21.5	i.v.
	~25 (68.7)	LO2
99	~0.5 (1.4)	Caco-2
	~0.5 (1.4)	HCT116	21.5	i.v.
	~25 (68.7)	LO2
100	~2 (4.4)	Caco-2
	~3 (6.6)	HCT116	17	i.v.
	~50 (110.4)	LO2
102 (EISA,37°C)	0.66 (1.64)	HepG2	10.23	i.v.	[78]
	1.43 (3.55)	A549
	1.96 (4.87)	U87MG
102 (EISA,4°C)	0.22 (0.55)	HepG2	10.23	i.v.
	0.26 (0.65)	A549
	0.87 (2.16)	U87MG
104 (heating-cooling)	0.43 (1.03)	HepG2	9.90	i.v.
	0.55 (1.32)	A549
	1.55 (3.73)	U87MG
107 (EISA,4°C)	0.13 (0.24)	HepG2	15.62	i.v.	[79]
	~50 (~92)	LO2
107 (EISA,37°C)	0.22 (0.41)	HepG2	15.62	i.v.
	~25 (~46)	LO2
106 (EISA, 4°C)	0.40 (0.97)	HepG2	15.62	i.v.
	~2 (4.9)	LO2
107 (heating-cooling)	0.63 (1.16)	HepG2	14.96	i.v.
	~3 (~5.5)	LO2
109	0.08 (0.15)	HepG2	14.0	i.v.	[80]
	~10 (~18)	A549
	~5 (~9)	MCF-7
	~20 (~37)	HeLa
	~15 (~27)	PC-3
	>150 (>275)	LO2
110	0.355 (0.623)	HepG2	13.3	i.v.
111	0.423 (0.504)	HepG2	8.7	i.v.
112	~500 (385)	HeLa	N/A	N/A	[81]
	>500 (>385)	3T3
113	>500 (>345)	HeLa
114	>500 (>331)	HeLa
115	>500 (>912)	HeLa			[102],[84],[85]
	>500 (>912)	U87MG
	>200 (>365)	HEK293
	>200 (>365)	HepG2
	>400 (730)	Saos-2
Dox	3 (1.6)	HeLa
	4.25 (2.31)	MES-SA/DXS
115+Dox	0.4 (0.22)[b]	HeLa
	1.28 (0.70)	MES-SA/DXS
117	48.7 (47.1	HeLa			[86]
	486.5 (470.4)	A549
	436.2 (421.8)	A2780
	178.5 (172.6)	SKOV3
	455.4 (440.4)	L929
118	50.4 (44.7)	HeLa
	20.1 (17.8)	A549
	112.6 (99.9)	A2780
	56.0 (49.7)	SKOV3
	42.0 (37.3)	L929
119	42.3 (69.2)	HeLa			[88]
	31.6 (51.7)	HepG2
	74.9 (122.5)	HCT116
	154 (251.9)	A459
	271 (443.3)	LO2
121	~200 (186)	S-ECs			[89]
	~400 (372)	HeLa
122	>400 (>307)	S-ECs
	>400 (>307)	Senescent HeLa
123	9.7 (16.8)	HeLa (dark)			[91]
	7.3 (12.6)	HeLa (light)
	23.2 (40.1)	4T1 (dark)	2	i.t.
	14.8 (25.6)	4T1 (light)	2	i.t.
	>30 (>52)	HEK-293T
125		CT26	10	i.v.	[92]
127	143.2 (169.0)	U87MG			[93]
	>500 (>590)	HeLa
	>500 (>590)	HS-5
	>500 (>590)	MCF-10A
128	59.3 (79.6)	U87MG
	198.5 (266.3)	HeLa
	181.4 (243.4)	HS-5
	118.6 (159.1)	MCF-10A
131	59.3 (70.0)	U87MG
	198.5 (234.3)	HeLa
	290.4 (342.8)	HS-5
	282.2 (333.1)	MCF-10A
132	59.2 (79.4)	U87MG
	66.5 (89.2)	HeLa
	100.7 (135.1)	HS-5
	119 (159.6)	MCF-10A
133	>500 (>728)	U87MG
	92.9 (135.2)	HeLa
	>500 (>728)	HS-5
	180.3 (262.5)	MCF-10A
134	38.5 (62.2)	U87MG
	35.2 (56.9)	HeLa
	94.4 (152.6)	HS-5
	118 (190.8)	MCF-10A
135	~800 (~1808)	4T1	15	i.v.	[95]
	>1000 (>2260)	LO2
136	~700 (~1645)	4T1	15	i.v.
	>1000 (>2350)	LO2
137	~200 (~448)	4T1	15	i.v.
	>1000 (>2241)	LO2
138	~100 (~253)	4T1	15	i.v.
	>1000 (>2531)	LO2
143	>500 (>448)	HeLa			[96]
	>500 (>448)	HcerEpic
	>500 (>448)	T98G
147	>500 (>656)	HeLa			[97]
	~200 (~262)	HepG2	9.5	i.v.
	>500 (>656)	LO2
	>500 (>656)	H8
153	>12 (>17)	SCC7	14.[c]	i.t.	[100]
	>12 (>17)	Cal-27
153 + Laser	~8 (~11)	SCC7	14.2	i.t.
CDDP + 153 + Laser	~7 (~10)	SCC7	14.2	i.t.

^[a]IC₅₀ is the concentration of HCPT with 12 fixed to be 700 μM.

^[b]115 is fixed to 200 μM, the IC₅₀ is of Dox concentration.

^[c]15 mg/kg CDDP one day before 10 μmol/kg of 153.

A unique advantage of peptide assemblies for cancer therapy is the minimization of acquired drug resistance, as demonstrated in cell assays.^{[46, 112]} The exploration of peptide assemblies for cancer therapy will enter a more fruitful phase of research by addressing above challenges. Moreover, the spatiotemporal control illustrated in EISA of peptide assemblies may lead to new direction of research^[113], such as enabling cancer cell spheroids^[114] or organoids for test peptide assemblies and organelle-targeting^[115]. For example, using EISA of peptides for reporting enzyme activities^[116] is likely to provide more information for understanding the functions of cells, especially when combined with dynamic imaging approaches. Non-peptide small molecules may be another molecular pool for EISA, as shown in a recent study that rig-rod aromatics served as EISA substrates for targeting tumor microenvironment.^[117] Photodynamic and photothermal therapies involving peptide assemblies have garnered significant attention in recent years^[118], combining EISA with more elaborated photoactive motif likely would receive more attentions, as shown by recent studies^[119]. We envision that the research of peptide assemblies for cancer therapy will contribute to a new frontier, cancer supramolecular medicine, at the interface of chemistry and cell biology.

Abbreviations:

5-FU

5-fluorouracil

CPT

camptothecin

HCPT

hydroxycamptothecin

PTX

paclitaxel

CRB

chlorambucil

Olsa

osalazine

HCQ

hydroxychloroquine

Dox

doxorubicin

ALP

alkaline phosphatase

CES

carboxylesterases

CTSB

cathepsin B

ACP

acid phosphatase

ENTK

enterokinase

Biographies

Yuchen Qiao obtained her BS degree from the school of biomedical engineering in Southeast University, China. She is currently in her third year as a graduate student in chemistry supervised by Professor Bing Xu at Brandeis University. Her current research interest lies in designing self-assembling materials for biological applications.

Bing Xu received his BS and MS degrees from Nanjing University in 1987 and 1990, respectively. He obtained his PhD in 1996 from the University of Pennsylvania, under the supervision of Professor Timothy Swager. Prior to beginning his independent research at the Hong Kong University of Science and Technology (HKUST) in 2000, he worked as an NIH postdoctoral fellow in the Whitesides lab at Harvard University. He was tenured as an associate professor in January 2006 and became a full professor in July 2008 at HKUST. Currently, he is a professor at the Department of Chemistry, Brandeis University. His current research focuses on the applications of enzymatic noncovalent synthesis in materials, biology, and medicine.