Self-assembled peptide nanostructures targeting death receptor 5 and encapsulating paclitaxel as a multifunctional cancer therapy

Abstract

The development of tumor-targeted nanoscale carriers for the delivery of cancer therapeutics offers the ability to increase efficacy while limiting off-target toxicity. In this work we focused on targeting death receptor 5 (DR5), which is highly expressed by cancer cells, and upon binding, triggers programmed cell death. Hence, a nanostructure targeting DR5 would act as a dual targeting and therapeutic agent. We report here on a peptide amphiphile (PA) containing a dimeric, cyclic peptide that self-assembles into cylindrical supramolecular nanofibers and targets DR5. Coassembly of the DR5-targeting PA and a pegylated PA creates a supramolecular nanoscale construct with enhanced binding affinity to DR5 relative to a monomeric targeting PA, and was found to be cytotoxic in vitro. When combined with the chemotherapy paclitaxel, DR5-targeting carriers showed potent antitumor activity in vivo, demonstrating the multifunctional capabilities of peptide-based supramolecular nanostructures.

Graphical Abstract

Introduction

Increasing both the efficacy and precision of existing cancer therapeutics entails the addition of targeting moieties that are specific to cancer cells.^1–2 As one way of achieving this goal, nanotechnology offers advantages in the delivery of cancer therapies through a combination of passive and active targeting of tumors. Passive targeting by nanoscale carriers is the result of the enhanced permeability and retention (EPR) effect,³ and is influenced by shape⁴ and size⁵ of the vehicle. While passive targeting can be used to enhance accumulation into the tumor, the incorporation of targeted functionalities through the use of biologics is designed to actively target tumor cells and improve cellular uptake, typically by binding to extracellular membrane receptors.6^–7

One specific group of receptors, death receptors 4 and 5 (DR4 and DR5), is often highly expressed by cancer cells and represents an intriguing target for antitumor therapies.^8–9 Its ligand, tumor necrosis factor related apoptosis-inducing ligand (TRAIL), is the only known example of an endogenous molecule that initiates apoptosis preferentially in malignant cells.¹⁰ TRAIL binds to DR4 or DR5, which initiates caspase-dependent apoptosis through the Fas-associated death domain as a part of the death inducing signaling complex (DISC). TRAIL is a part of the tumor necrosis factor (TNF) family, and can bind to decoy receptors 1 and 2, which do not induce apoptosis, in contrast to DR4 and DR5. Both recombinant soluble portions of the TRAIL protein and agonistic humanized monoclonal antibodies specific for DR4 and DR5 have shown efficacy in animal models and are currently in clinical trials.^11–13 Recent studies in murine models of metastatic breast cancer suggest that targeting DR5 is more effective at suppressing metastases than targeting DR4.¹⁴ Moreover, DR5 is expressed by clinically aggressive breast tumors and correlates with lymph node metastases and poor survival.¹⁵ While TRAIL has exhibited promise as a monotherapy, combinations with TRAIL-sensitizing agents, including chemotherapy drugs, have also improved activity against TRAIL-resistant cancer cell lines.^16–18 Additionally, TRAIL proteins have been incorporated onto nanoparticles to take advantage of the improved circulation times.^19–21

Because of cost and stability issues associated with proteins and antibodies,²² TRAIL-mimetic peptides have also been investigated as therapeutic agents that bind to either DR4 or DR5.^23–26 The binding and in vitro activity of bivalent and trivalent forms of these peptides, conjugated covalently or non-covalently, have been found to increase 1000-fold relative to monomeric structures.²³ This enhancement through multivalency is likely related to the previous finding that TRAIL activates the apoptotic pathway through the oligomerization of multiple DR5 receptors.^27–28 We have investigated here the binding and induction of apoptosis by supramolecular nanostructures containing TRAIL-mimetic peptides given their highly polyvalent nature.

In this work we have incorporated a TRAIL-mimetic peptide sequence into a peptide amphiphile (PA) to generate by supramolecular self-assembly a targetable therapeutic nanostructure. PAs developed earlier in our laboratory form one-dimensional nanostructures through a combination of hydrophobic collapse and β-sheet hydrogen bonding among especially designed peptide domains.^29–31 They have been used previously as antitumor therapies by functionalization with a cationic, membrane-lytic peptide sequence.^32–33 While effective as both in vitro and in vivo therapies, these cytotoxic PAs are not specific to cancer cells, unlike a potential TRAIL-based therapy could be. Additionally, the ability of PA nanostructures to deliver small molecule drugs, either through encapsulation or covalent attachment, was demonstrated in previous work.^34–37 Efforts to increase the drug release in tumors by peptide nanostructures have focused on environmentally induced changes to the peptide’s supramolecular structure.^38–41 However, the peptides used in these studies lacked the chemistry to target binding to specific receptors on cancer cells.

In this report we demonstrate encapsulation of the chemotherapy drug paclitaxel within nanostructures that externally display TRAIL-mimetic peptides as proof-of-principle for the rational design of multifunctional nanotherapeutics with enhanced cytotoxic activity directed at cancer cells.

Experimental (Materials and Methods)

Peptide Synthesis

TRAIL PAs and peptides were synthesized on a Tentagel S Ram low loading resin (Peptides International, 0.18 mmol/g) using standard Fmoc solid-phase synthesis. For the hydrocarbon tail of TRAIL PAs, lauric acid was coupled to the ε-amine of lysine(Mtt) (4-methyltrityl), which was deprotected five times in 2% trifluoroacetic acid (TFA) and 5% triisopropylsilane (TIPS) in CH₂Cl₂ three times for two minutes. After synthesizing the Fmoc-K(Fmoc)K(Boc)K(Boc)A₆K(C₁₂) on an automated synthesizer (CEM Microwave Synthesizer), the peptide was deprotected in 30% piperidine in DMF, washed, and subsequently an Fmoc-EL-NH₂ peptide was coupled using 2 equivalents of peptide, and 1.95 equivalents of PyBop. The Fmoc-EL-NH₂ was used to preserve the binding sequence from the previously described work,²³ which contained a linker between peptides on the second amino acid from the C-terminus. Fmoc-EL-NH₂ was synthesized by reacting Fmoc-Glu(OtBu)-COOH with Leu-NH₂ with 2 equivalents EDC and 1 equivalent Oxyma Pure in CH₂Cl₂ (Figure S1). The product was then washed in 0.1 M HCl and brine, and dried in vacuo. The crude product was recrystallized in ethyl acetate to yield Fmoc-E(OtBu)L-NH₂, and then deprotected in 20% TFA, 5% TIPS in CH₂Cl₂. After coupling on Fmoc-EL-NH₂ to the PA, standard Fmoc SPPS was resumed on the automated synthesizer, using double the equivalents of reactants because of the two reaction sites on the branched compound. Synthesis of PEG PA (PEG₂₀₀₀-E₂A₆K(C₁₂), referred to as PA 4) was carried out as described previously using Fmoc-chemistry on a Tentagel S Ram LL resin.³³ COOH-PEG₂₀₀₀-MeO was synthesized as previously described⁴² and was coupled onto the N-terminus of the PA on resin, using 4 fold excess of PEG with DIC and Oxyma pure (Novabiochem) in 25% DMF, 75% DCM overnight.

Cleavage was then performed using a TFA/Triisopropylsilane/H2O/2,2’-(Ethylenedioxy)diethanethiol mixture (90:2.5:2.5:5). PA purification was performed on a Varian Prostar 210 HPLC system, eluting with 2% acetonitrile (MeCN) to 100% MeCN in water with 0.1% TFA on a Phenomenex C18 Gemini NX column (150 × 30 mm). HPLC fractions with the PA of interest were confirmed by ESI mass spectrometry (Agilent 6510 Q-TOF LC/MS), combined, and lyophilized after removing MeCN by rotary evaporation. Cyclization was performed at a concentration of 1 mg/mL in 20% DMSO, 0.1% TFA in H₂O and monitored using Ellman’s reagent for the dimeric version of TRAIL PA (referred to as PA 1), dimeric TRAIL peptide (peptide 2), and the monomeric version of the PA (PA 3). After 48 hours, cyclized PA was directly injected onto the HPLC. Using an isocratic hold after a slow ramp to 38% MeCN over 30 minutes, the cyclized fractions containing the expected mass were collected and lyophilized after removal of MeCN.

Because each dimeric TRAIL PA and peptide had four cysteines, multiple disulfide formations were possible. The HPLC peaks with the correct mass were distinguished by the location of disulfide bonds using trypsin degradation, a technique that has been described in detail previously for determining the location of disulfide bonds in proteins.⁴³ The purified products were dissolved in PBS pH 6.5 with 0.1wt% PA 1 and 10 uL of trypsin immobilized on beads (Thermo Scientific). After leaving the reaction shaking overnight, the solution was centrifuged at 500 g to remove trypsin, and kept at 4°C until injection by LC-MS in acidic conditions. Additionally, 50 mM TCEP was added to the solutions to reduce the disulfide bonds and then LC-MS was performed to determine the masses of the individual components.

PAs with encapsulated paclitaxel, either PA 4 alone or PAs 1 & 4, were prepared using the same protocol as described previously with camptothecin.³⁴ Briefly, PAs were dissolved in a solution of (1,1,1,3,3,3)-hexafluoroisopropanol (HFIP) and mixed with a concentrated stock of paclitaxel in DMSO, evaporated to a film in vacuo. The films were then resuspended in water, centrifuged 1000g to remove any precipitated paclitaxel, and lyophilized before use. Total concentrations of PA (PAs 1 & 4 or PA 4 alone) were kept the same for paclitaxel encapsulation.

Materials Characterization

Cryogenic TEM (cryo-TEM) specimens were prepared using an FEI Vitrobot by blotting in 95% humidity and subsequently plunging lacey carbon grids into liquid ethane. Images were taken for cryo-TEM using a JEOL 1230 transmission electron microscope operating at 100 keV equipped with a Gatan camera. For the targeted and coassembled nanofibers, the samples were dissolved at 500 uM PA in PBS prior to plunging. Critical aggregation concentrations (CACs) were determined by measuring maximum emission wavelength of the hydrophobic dye Nile red. A stock solution of Nile red in ethanol was added to solutions of PAs or peptides for a final Nile Red concentration of 100 nM. In PBS, Nile Red had a maximum emission wavelength of 658 nm, which was found to be independent of dye concentration over the range used in this study. Fluorescence was measured using a NanologHJ Fluorometer.

Surface plasmon resonance experiments were performed on a Reichert SR7500DC SPR spectrometer. Using a dextran chip, 2000 response units (RU) of DR5 (Enzo) was immobilized onto the surface with EDC/NHS chemistry in a pH 5.6 15 mM Na Acetate buffer with 10 ug/mL of protein at a rate of 10 uL/min. 1M Ethanolamine was then flowed over the chip to react with any free carboxylate groups still present on the surface. The chip was then regenerated several times to remove any non-covalently attached protein with 25 mM HCl for 30 s each time. A buffer of 10 mM phosphate, 150 mM NaCl, 3 mM EDTA, and 0.01 v/v % Tween 20, pH 7.0 was used as the running buffer. The standard running buffer of HEPES 10 mM, pH 7.4 resulted in aggregation of the PA on the surface, so a slightly more acidic PBS solution was used. Associations were run for 8 min, and dissociations for 10 min at a flow rate of 30 uL/min. Regenerations were performed after each analysis using 25 mM HCl for 30 s. Before analysis, concentrations of PAs and peptide were determined using the absorbance at 280 nm of a Nanodrop analyzer.

In vitro Cell Viability and Apoptosis Assays

Cell viability was measured using the CellTiter 96® Aqueous One Solution Cell Proliferation Assay (Promega, Madison WI). The Cell Titer Assay is a 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS)-based assay and was used according to the supplier’s instructions. For each well to be assayed in a 96-well plate containing 100 μl of media per well, 20 μl of the CellTiter96® Aqueous One Solution was added. The plate was incubated for 1–3 h at 37 °C, and absorbance was read using a Molecular Devices microplate reader (490 nm). For apoptosis studies, Annexin V–positive cells were identified using the Annexin-PE Apoptosis Detection Kit I (BD Bioscience) according to the manufacturer’s instructions except that 4’,6-diamidino-2-phenylindole (DAPI) was used.

In vivo Orthotopic Model of Breast Cancer

1 × 10⁶ MDA-MB-231 human breast ductal carcinoma cells were injected intraductally into the 4^th mammary fat pads of 4–5-week old female athymic nude mice (Harlan Sprague-Dawley, Madison, WI) to establish orthotopic xenograft tumors. Four weeks after tumor implantation, mice were randomized into treatment groups (8 mice per group) and administered either PBS vehicle (Sigma-Aldrich), PA 1 3.5 mg/kg/dose mixed with two molar equivalents of PA 4, PA 4 (three molar equivalents) containing 1 mg/kg encapsulated paclitaxel, or PAs 1 and 4 containing 1mg/kg paclitaxel by tail vein injection twice per week for six total injections. Free paclitaxel was administered in a stock solution of DMSO via intraperitoneal injection also twice per week for six total injections. Tumors were measured after each injection with Vernier calipers, and tumor volume was calculated using the equation: tumor volume (mm³) = (length x width²) x 1/2 as previously described.⁴⁴ All animal experiments were conducted under protocols approved by the Animal Care and Use Committee of Northwestern University. Statistical significance was assessed using Graph-Pad Prism 5.0b software using the appropriate statistical test: two-way ANOVA with Bonferroni post-tests for in vivo tumor cytotoxicity.

Results

To mimic the dimer structure described by Pavet et al.,²³ a branched, cyclic peptide was synthesized that included both the targeting and self-assembly domains (using standard Fmoc chemistry with solid-phase peptide synthesis (SPPS) (Figure 1A)). The branched PA was synthesized using a lysine residue as the branching point, allowing for the TRAIL-mimetic peptide sequence to be covalently conjugated to the PA. The TRAIL-mimetic peptide sequence reported by Pavet et al., WDCLDNRIGRRQCVKL, was attached to the internal K residue, which made direct coupling of the sequence via SPPS impossible. In order to preserve the same binding sequence and point of attachment, an Fmoc-EL-NH₂ linker was synthesized and coupled onto both amines of the lysine on resin via the side chain on the linker’s glutamic acid (Figure S1). The rest of the dimeric, TRAIL-mimetic PA (referred to as PA 1) was synthesized with standard SPPS protocols and cleaved in the presence of thiols, creating a PA with two linear DR5 binding sequences (Figure S2). This synthetic route was chosen because solution-phase functionalization of the cyclic binding sequence to a branched PA was inaccessible due to steric hindrance created by aggregated molecules (data not shown).

Figure 1.

General synthesis and structure of PA 4. (A) The branched structure is synthesized by solid phase peptide synthesis, purified with four deprotected thiols, and then oxidized in the presence of DMSO to form a mix of cyclization products. (B) The isomeric, cyclized products are separated by HPLC and subsequently identified by treatment with trypsin, followed by LC/MS to determine the fragment masses, which differ based on the location of the disulfide bonds. (C) Schematics of the dimeric, self-assembling PA 1, the dimeric peptide 2, monomeric PA 3, and a PEGylated PA 4.

After purification by HPLC, cyclization was achieved using 20% DMSO in acidic water. This reaction produced two major products, with the desired cyclization product eluting before the other cyclized PA (Figure 1B, S3A). Because the two cyclization products have the same mass (Figure S3B–C), each was verified using a previously established technique that incorporated trypsin degradation and LC-MS (Figure 1B, S4A).⁴³ Depending on where the disulfides formed, the masses of the cleaved compounds would be different as measured by LC-MS. Indeed, the degradation products of the two major HPLC peaks had the expected masses of the desired cyclized product and the cyclization side reaction (Figures S4B and S4D, respectively). LC-MS of the reduced molecule resolved the individual components, proving that the cyclization was the result of disulfide formation (Figure S4B–S4C). Overall, our alternative synthetic method incorporated multiple cyclizations without the need for different thiol-protecting groups to create a dimeric PA. Using similar methods, a dimeric TRAIL-mimic without the self-assembly domain (Peptide 2), a monomeric TRAIL-mimic PA (PA 3), and a PEGylated PA, PEG_2k-E₂A₆-K(C₁₂) (PA 4) were also synthesized (Figure 1C).

Characterizing the morphology of the PA 1 nanostructure was essential to understand the effect of the targeting moiety on the one-dimensional assembly. Cryogenic transmission electron microscopy (cryo-TEM) was used to determine the structure of each component used in subsequent in vitro and in vivo studies. PA 1 by itself seemed to form very short, aggregated structures at a concentration of 100 μM (Figure 2A). PA 4, which was similar in structure to previously described molecules,³³ formed nanofibers in PBS that ranged from tens to hundreds of nanometers in length (Figure 2B). Peptide 2, which lacked the self-assembly domain present in PA 1, did not appear to form any defined structures as revealed by cryo-TEM (Figure S6). PA 3 formed long, one-dimensional structures that were similar to the canonical PA nanofibers (Figure 2C). Because PA 3, which contains a monomeric binding peptide, forms long nanofibers, while the dimeric PA 1 does not, the dimeric version of the cyclic epitope likely disrupts fiber formation, resulting in loosely aggregated structures. However, when PA 1 and PA 4 were mixed, they formed nanoscale cylindrical structures that were several microns in length, which were significantly longer than either PA 1 or 4 alone (Figure 2D). We chose to coassemble PAs 1 and 4 because incorporation of PA 4 into PA assemblies was previously shown to result in enhanced tumor accumulation in vivo.³³ Coassembly with PA 4 had also been shown to increase the plasma concentration of the coassembled PA as measured by area under the curve over the first six hours after injection.³³ Additionally, a 2:1 molar ratio of PA 4:PA 1 was used in all experiments involving coassembled PAs, which was above the ratio necessary to limit proteolysis in our prior studies.³³ Previous results had shown the same long fiber formation after co-assembly of a cationic PA with a negatively-charged PEG PA,³³ suggesting that both PA molecules were coassembled into the same nanostructure.

Figure 2.

Characterization of TRAIL-mimetic and PEG PAs. (A-C) Cryo-TEM of PA 1 (A), PA 4 (B), and PA 3 (C), respectively, at 100 μM. (D) Cryo-TEM of PA 1 coassembled with PA 4 at a total PA concentration of 50 μM. Insets show representative cartoons of the molecules being imaged.

Critical aggregation concentrations (CACs) were determined for the PAs alone and the mixed PA 1 and PA 4 combination to determine at what critical concentration the PA molecules began to assemble. Using the encapsulation of Nile red, which has a blue-shifted emission maximum in hydrophobic environments, the CAC was determined for PA 4 alone, PA 1 alone, and the PA 1 & 4 mixture (Figure 3A). The CACs for both PAs 4 and 1 were approximately 3 and 10 μM, respectively, indicating the onset of self-assembly at a similar concentration. When coassembled in the same 2:1 PA 4:PA 1 ratio, the CAC of the total PA concentration was also found to be roughly 3 μM. Because samples containing coassembled molecules had approximately the same CAC as PA 4 molecules alone (in spite of lower concentrations of the individual components in co-assemblies) these results suggest that the individual PAs are combining together onto the same supramolecular structure. On this basis we believe these results strongly suggest the co-assembly of PA 4 with PA 1 molecules in solution. In spite of the relatively high CAC of PAs, previous studies have shown that PA assemblies retain a nanofiber morphology in the presence of FBS.³⁷

Figure 3.

(A) Critical micelle concentration measurements using the maximum emission wavelength of the fluorophore Nile red as a function of total PA concentration, plotted as a blue shift relative to PBS alone. (B) Surface plasmon resonance curves of PA 1 (blue), peptide 2 (grey), and coassembled PAs 1 and 4 at a concentration of 62 nM PA 1.

To probe the effect of supramolecular structure on the targeting capabilities of PA 1, we performed surface plasmon resonance (SPR) against surface-immobilized human recombinant DR5. The dimeric peptide bound with similar affinities to DR5 compared to previously reported findings,²³ suggesting that the new linker covalently attached to the glutamic acid side chain did not negatively impact binding (Figure 3B). We did not expect to see any enhancement in binding kinetics of the PA due to the multivalency effect arising from self-assembly because analysis was performed at concentrations below the CAC. Indeed, the overall binding curves were similar between the peptide and PA versions of the dimeric TRAIL peptide, and coassembly with PA 4 only induced a minor decrease in the rate of association. Additionally, by itself, PA 4 did not show any binding to DR5 (Figure S5C).

When the peptide 2 binding data were fit using a 1:1 Langmuir binding model, the K_D was found to be 2.1 nM (Figure S8A). A 1:1 Langmuir binding model did not fit the data well for PA 1 (Figure S8C, E), but when a mass transport model was used, the fit was improved and yielded a K_D of 2.4 and 2.2 nM for PA 1 and coassembled PAs 1 and 4, respectively (Figure S8D, F). While the PA was at concentrations below the CAC, oligomerization of a few PAs by hydrogen bonding might have still decreased the diffusion rate relative to non-assembling peptides. The addition of a mass transport model did not improve the fitting for the peptide sensograms, suggesting that diffusion did not limit the binding of peptide 2 to DR5 (Figure S8B). We speculate that even at low concentrations, PA-PA interactions decreased the diffusion coefficient of the PA, causing an additional limiting step in interaction with the DR5 ligand. The monovalent PA 3 showed decreased levels of binding relative to the dimeric binding peptides, as seen specifically by the faster dissociation rate (Figure S7C). These results demonstrated that the monovalent display of the cyclized peptide on a supramolecular structure did not enhance the binding affinity to levels that were comparable to the covalently linked dimers. Overall, these results show that both PAs and peptides displayed a similar high affinity for DR5, while also exhibiting significantly different binding kinetics.

To test for non-specific adsorption, we measured the binding of the peptide and PAs to DR4, using the same conditions that were used for DR5. The peptide sequence chosen here has been shown not to interact with death receptor 4, nor decoy receptors 1 and 2.²³ As expected peptide 2 did not display any binding to DR4 (Sup. Fig. 9A); however, PA 1 did show some non-specific adsorption to DR4 at higher concentrations, but the response was significantly lower relative to DR5 (Sup. Fig 9B). Interestingly, the combination of PAs 1 and 4 showed much lower non-specific adsorption at the same concentrations. Even though the PAs were below their CAC according to Nile Red measurements, these results suggest that there was still an interaction between PAs 1 and 4 that limited non-specific binding to DR4.

In addition to the bioactivity provided by binding to DR5, PAs were also combined with the hydrophobic chemotherapy paclitaxel using previously described methods³⁴ because this molecule has been shown to enhance the pro-apoptotic effects of TRAIL receptor-targeted therapies.⁴⁵ Previous reports using camptothecin with PA nanofibers had shown encapsulation efficiencies of approximately 70% and release of greater than 75% of the encapsulated drug after seven days.³⁴ The incorporation of paclitaxel resulted in fusion of the nanostructures into larger, aggregated structures in the case of both coassembled PAs 1 and 4, as well as PA 4 alone (Figure S10A–B). Treatment of the coassembled PAs 1 and 4 with DMSO to mimic the encapsulation protocol, but in the absence of paclitaxel, did not display any fusion, suggesting that inclusion of paclitaxel caused the change in nanostructure dimensions (Figure S10C). The hydrophobic paclitaxel molecule may be linking hydrophobic regions between fibers, resulting in higher order bundling. However, the addition of paclitaxel did not change the binding kinetics of PA 1 to DR5 by SPR (Figure S10D). This change in supramolecular dimensions with paclitaxel demonstrated the effect of a hydrophobic additive to PA components.

To determine if the differences in binding to DR5 had an effect on cell death, in vitro cytotoxicity experiments were performed using MDA-MB-231 human breast cancer cells. Cytotoxicity was observed in treatments that incorporated PA 1, and PA 4 with PA 1 and encapsulated paclitaxel, referred to as PAs 1+4(tax) (Figure 4A). Because the dose of paclitaxel is well above its IC50 value, the cell viabilities after treatments of PA 1 with paclitaxel and free paclitaxel remained constant across different dosages of PA 1. PA 1 with and without coassembled PA 4 was cytotoxic at concentrations above 1 μM. These results, combined with the SPR sensograms that show specific binding to DR5, suggest that binding to DR5 initiates a cell death pathway. Encapsulation of paclitaxel by PAs 1 and 4 enhanced the cytotoxicity of these nanostructures at higher concentrations, demonstrating that the two therapies are effective when combined within a single nanostructure. T47D breast cancer cells, which have previously been shown to be resistant to TRAIL-based therapies,⁴⁶ were also treated with these nanostructures. Across this concentration range, no decrease in cell viability was observed by treating T47D cells with PA 1 (Figure 4B). Moreover, encapsulation of paclitaxel into the TRAIL-mimetic nanostructures did not enhance their cytotoxicity. These results suggest that the cytotoxicity of PA 1 may require cell surface expression of DR5 on target cells.

Figure 4.

In vitro cytotoxicity of dimeric TRAIL nanostructures. (A) Cell viability of MDA-MB-231 breast cancer cells as a function of PA 1 concentration. Groups with encapsulated paclitaxel, marked as “+ tax,” are compared to free paclitaxel administered in DMSO. For experimental groups not containing PA 1, the equivalent dose of paclitaxel was used compared to the other experimental groups. (B) Cell viability of TRAIL-resistant T47D breast cancer cells in the presence of PA 1. Flow cytometry of Annexin V positive, apoptotic MDA-MB-231 cells (C) and T47D cells (D) at a concentration of 1 μM PA 1. (* p < 0.05, ** p < 0.01, *** p < 0.001).

The effectiveness of PA 1 was further characterized in vitro using Annexin V staining to measure apoptosis. Based on the cytotoxicity data, we chose to test PAs at a sub-lethal concentration of 1 μM in order to be able to detect PA nanostructure and paclitaxel synergy. Only the multivalent PAs 1+4(tax) PA demonstrated statistically significant increased annexin V positivity (Annexin V staining was not statistically different for any of the other treatments) (Figure 4C). Addition of ZVAD, a caspase inhibitor, significantly reduced cell death in the presence of PAs 1+4(tax) PA, demonstrating that cell death was the result of caspase-dependent apoptosis. Consistent with the MTS results, the treatment of T47D cells with TRAIL-mimetic nanostructures did not increase cell death. Paclitaxel treated cells showed a statistically significant increase in annexin V binding; however, we did not observe a statistically significant ZVAD reversibility, suggesting the mechanism of paclitaxel cytotoxicity was not dependent on caspase activation (Figure 4D). These results show that PA 1 induces caspase-dependent apoptosis in TRAIL-responsive breast cancer cells and can synergize with paclitaxel. Combining our results from SPR and from the in vitro cell studies, we have demonstrated that PA 1 binds with high affinity to DR5 and can initiate apoptosis and cell death in MDA-MB-231 breast cancer cells.

We used an orthotopic xenograft model of breast cancer with MDA-MB-231 cells⁴⁷ to investigate in vivo the antitumor effects of supramolecular nanostructures targeting the DR5 receptor. PBS vehicle, PAs 1+4 at 3.5 mg/kg/dose, PA 4(tax) containing 1 mg/kg encapsulated paclitaxel, PAs 1+4(tax), or free paclitaxel was administered by tail vein injection, or in the case of the free drug via intraperitoneal injection, twice per week for six total injections. Mouse weights remained constant throughout the experiment, suggesting that all treatments were well tolerated (Figure S11). The PAs 1+4 nanostructures with encapsulated paclitaxel inhibited mammary tumor growth, and were significantly more effective than paclitaxel alone and saline controls (Figure 5A–B). There was a trend towards smaller tumors in the mice treated with PAs 1+4(tax) nanostructures compared to PAs without paclitaxel encapsulation or PA 4(tax) ones, but this difference was not statistically significant. Moreover, the combination of PAs 1 and 4 without paclitaxel showed a similar trend against PBS control, but this trend was also not statistically significant. Based on the enhancements in efficacy over free paclitaxel, our findings demonstrate that the co-assembled PAs 1 and 4 nanostructures can also serve as effective delivery vehicles for the hydrophobic chemotherapy drug paclitaxel. Future PA-based therapies could combine other TRAIL-sensitizing drugs with PA 1 to create a nanoscale therapy that would be effective against TRAIL-resistant tumors. The systemic delivery of a targeted multifunctional nanostructure demonstrates the modularity of the PA platform and its broad translational potential for personalized cancer therapy.

Figure 5.

In vivo efficacy of PA 1 nanostructures. (A) Tumor volume plotted over five weeks as the result of treatments that included different combinations of PA 1, PA 4, and paclitaxel. (B) Histogram of tumor volumes after five weeks shows significant differences in tumor volume for PAs 1 and 4 with paclitaxel treatment (** p < 0.01, *** p < 0.001).

Conclusions

Apoptosis-inducing supramolecular nanostructures that are targeted to cancer cells offer new opportunities as antitumor therapeutics. In this work, a dimeric bioactive PA was synthesized with high affinity for the pro-apoptotic DR5 death receptor and capacity to coassemble with other PAs into supramolecular nanostructures. The hydrophobic compartment of these nanostructures enabled encapsulation of a chemotherapy drug, and the synergy with pro-apoptotic bioactivity led to in vivo efficacy in a breast cancer model. Our results establish the translational potential of PAs as multifunctional personalized nanomedicines that can be rationally designed to incorporate targeting epitopes and multiple therapeutic agents tailored to the molecular characteristics of the tumor.

ABBREVIATIONS

DR5

Death Receptor 5

DR4

Death Receptor 4

TRAIL

Tumor Necrosis Factor Related Apoptosis Inducing Ligand

CAC

critical aggregation concentration

PA

peptide amphiphile

SPPS

solid phase peptide synthesis