Peptomer Linkers Enable Kinetic Control over Co‐Delivery of Multiple Chemotherapeutics

ABSTRACT

Combination cancer therapies offer to minimize toxicity and alleviate patient burden, but a key challenge is independent control over the release of multiple therapeutics, especially in systems with chemotherapeutic drugs of similar size and structure. Due to MMP upregulation in cancer microenvironments, matrix metalloproteinase (MMP)‐degradable linkers are often exploited for targeted release; however, their short substrates exhibit overlap with multiple proteases, which confounds kinetic control. Here, a library of MMP‐responsive “peptomer” drug linkers was developed to control release on fast and slow timescales. Peptomers are hybrid molecules of peptides and non‐natural peptoids (N‐substituted glycines), which hinder proteolytic susceptibility. Systematic variation of peptoid substitutions within a pan‐MMP‐cleavable peptide sequence yielded distinct degradation kinetics to multiple MMPs. Two chemotherapeutics, doxorubicin and geldanamycin, were conjugated to peptomer linkers and incorporated into polyethylene glycol (PEG)‐based hydrogels for sequential delivery. Fluorogenic and mass spectrometry‐based assays demonstrated decoupled release of each drug in response to MMP‐2 and MMP‐9. In vitro studies using MDA‐MB‐231 and A549 cells showed that cell death rates correlated with the order of drug release. These findings highlight peptomers as modular, biocompatible linkers capable of kinetic control over multiple therapeutic agents, providing a versatile platform for improving the precision of combination drug delivery systems.

A key challenge in combinatorial chemotherapeutic drug delivery is independent control over release kinetics, especially with drugs of similar size and structure. Here, peptoid substitutions to proteolytically degradable peptides enabled the design of fast and slow‐releasing drug linkers. Upon incorporation into a hydrogel, these linkers sequentially controlled the delivery of doxorubicin and geldanamycin in response to both exogenous and cell‐secreted proteases.

1. Introduction

Combination therapy has emerged as an important strategy in cancer treatment, offering enhanced efficacy, reduced side effects, and reduced drug resistance when compared to monotherapy [1, 2, 3, 4]. By using two or more therapeutics that target different pathways, each drug can often be administered at a lower dose, thereby reducing the toxicities associated with higher dosages of each of the individual drugs [5, 6, 7]. While combination therapy is effective and commonly implemented, it is associated with increased patient burden due to complex treatment schedules and the financial and psychological stress of managing multiple appointments and administrations [8, 9, 10]. Dual‐release systems capable of delivering multiple therapeutics at distinct times thus hold promise in improving the efficiency, safety, and convenience of combination therapies [11, 12].

Decoupling release kinetics in dual‐release systems is critical for optimizing therapeutic synergy and minimizing toxicity in combination cancer therapies [13]. By independently controlling the rate of each drug's release, these systems ensure that each agent reaches its target at the most effective concentration and in an appropriate sequence. Current strategies in staged co‐delivery utilize hydrogels with photocleavable drug linkers [14], layer‐by‐layer (LbL) films [15, 16], hydrogel/micelle composites [17], and liposomal hydrogels [18, 19, 20], each of which has its own limitations. For example, the shallow tissue penetrability of cytocompatible light significantly limits the translatability of photoresponsive hydrogel drug carriers in the clinic. LbL films are prone to burst effects and rely on electrostatic interactions that are susceptible to interlayer diffusion, which may result in heterogeneous dosing [21, 22]. Hydrogel/micelle composites and liposomal hydrogels have demonstrated independently controlled co‐delivery, but rely on specific cargo‐material affinity, which can be difficult to tune in a systematic way. The design of effective and tunable staged co‐delivery vectors that can be applied to drugs with similar chemical and physical properties thus poses an urgent research barrier.

When coupled with spatial targeting—such as tumor‐specific delivery or microenvironment‐responsive release—dual‐release systems can further enhance therapeutic efficacy by localizing drug activity to diseased tissues while sparing healthy cells [23, 24]. For example, matrix metalloproteinases (MMPs) are proteolytic enzymes largely responsible for extracellular matrix remodeling and are the most prominent family of proteases associated with tumorigenesis [25]. MMP‐2 and ‐9 (the gelatinases) are highly expressed in the tumor microenvironment (TME) [26, 27, 28], with several studies reporting an average 10 fold increase in the expression of genetic or protein levels across a variety of tumor types when compared to healthy tissue [27, 28, 29, 30, 31, 32, 33, 34]. The incorporation of MMP‐degradable substrates into drug delivery carriers is a well‐studied approach to engineering triggerable release to the TME [35], and short, MMP‐degradable peptides have been applied as crosslinkers in hydrogel [36] and nanoparticle [37] systems, peptide‐drug conjugates for self‐assembled nanoparticles [38, 39], or linkers for drug release from hydrogel systems [40]. However, a key limitation is that the small peptide substrates employed for these applications often suffer from substrate overlap and degrade to a wide class of proteases [40, 41, 42, 43, 44], thereby weakening control over kinetic profiles in co‐delivery systems and resulting in nonspecific delivery to healthy tissue. Specific cases related to MMP‐degradable peptides have reflected incidences where scrambled peptide sequences designed as a nondegradable control have still been subject to nonspecific degradation [45], and where drug delivery systems based on MMP‐degradable peptide linkers still reported 50% nonspecific release to other proteases [40].

To address this limitation, synthetic peptide analogs have been explored as a strategy to systematically tune proteolytic degradation to multiple proteases. These analogs include D‐peptides, N‐alkylated peptides, and N‐substituted glycines (peptoids) [43, 46, 47]. Incorporation of these analogs into peptides increases their stability, but D‐amino acids have previously shown cytotoxic effects [48]. In addition, N‐alkylated peptides have limited in vitro data, perhaps due to the need to synthesize each N‐alkylated monomer individually or perform additional on‐resin N‐alkylation chemistries, alongside concerns over changes that may diminish bioactivity entirely [49, 50, 51]. Peptoids, however, have gained attention for biomaterial and pharmaceutical applications due to their enhanced stability and versatility [52], as well as their promising cytocompatibility [53, 54, 55, 56]. The submonomer method used in peptoid synthesis decouples backbone assembly from side chain incorporation, such that virtually any amine can be incorporated, providing access to a wide variety of functionalities [57, 58]. Additionally, this method is compatible with traditional peptide synthesis methods, allowing for the synthesis of peptide‐peptoid hybrid molecules, or “peptomers” [59]. Therefore, we hypothesized that through making intentional peptoid substitutions in MMP‐degradable peptide sequences, we could predictably control degradation timescales, allowing for the design of fast and slow‐release hydrogel linkers for small molecule chemotherapeutics.

In this study, we designed a library consisting of one peptide and three peptomers that degrade at different rates to MMP‐2 and ‐9. After characterizing degradation timescales and cytocompatibility, the peptomers were applied as linkers in a polyethylene glycol (PEG) hydrogel model that demonstrated decoupled release kinetics of orthogonally fluorescent cargo to isolated MMP solutions. We developed conjugation strategies to link the peptomers to two model drugs, doxorubicin and geldanamycin, creating peptomer‐drug conjugates (PDCs). Doxorubicin is a chemotherapeutic drug that is FDA‐approved and widely used for the treatment of various cancer types, due to its intercalation into DNA and disruption of topoisomerase‐II‐mediated DNA repair [60]_. However, there are serious side effects associated with off‐target toxicity, predominantly in the form of cardiotoxicity [61]. In addition, doxorubicin is typically used in combination with other chemotherapeutic drugs, presenting a strong demand for delivery modes to promote localized delivery as well as codelivery. Thus, we also explored geldanamycin, an Hsp90 inhibitor, which has been extensively investigated in clinical trials, though these have been terminated due to hepatotoxicity and poor pharmacokinetics associated with solubility issues [62, 63]. We incorporated both doxorubicin and geldanamycin‐based PDCs with different release rates into PEG hydrogels and investigated their efficacy in vitro. Altogether, our work builds off the synergy achieved when these drugs are used in combination [64, 65, 66] and offers a path for increased drug solubility and kinetic control over chemotherapeutic release profiles.

2. Results and Discussion

2.1. Peptomer Library Design

To systematically control the degradability of therapeutic linkers, we designed a peptomer library around an MMP‐degradable consensus sequence, PAA↑LVA, adapted from the more widely known PAN↑LVA sequence [43, 67]. In peptide‐substrate nomenclature, amino acid positions are designated using the “P” system that matches residues with pockets of the enzyme binding site. The cleavage site is located between the P1 and P1’ residues, and amino acids toward the N‐terminus are classified as non‐prime, while those toward the C‐terminus are classified as prime (Figure 1A). Peptoid substitutions, specifically N‐alanine residues, were made in the P1, P3, and P1 and P3 positions in tandem to generate a library of four peptomers with varying location and number of substitutions: PAA↑LVA (PAALVA), PANAla↑LVA (P1‐NAla), NAlaAA↑LVA (P3‐NAla), and NAlaANAla↑LVA (P3, P1‐NAla) (Figure 1A). Mass spectrometry, analytical HPLC, and NMR confirmed that each peptomer was successfully synthesized in high purity (Figures S1 and S2).

FIGURE 1.

Overview of peptomer library, cleavage kinetics, and cytocompatibility. (A) The MMP‐consensus sequence, PAALVA, is shown along with the peptomer analogs and their associated N‐alanine substitutions. Peptomer sequences were functionalized with FRET pairs, and their cleavage rates were tracked over time via fluorescence increase to 0.13 µg mL⁻¹ of MMP‐2 (B) and 0.50 µg mL⁻¹ of MMP‐9 (C). Cell viability studies reflect peptomer cytocompatibility when incubated in complete media for three days with HDFs (D) and MDA‐MB‐231s (E) over a range of concentrations. Viability was determined via luminescent metabolic assay. All error bars represent the standard deviation from three technical replicates. Data is color‐coded according to the scheme in (A).

We selected the P1 and P3 positions for peptoid substitutions because it was hypothesized that the N‐substitution of the peptoids would not significantly interfere with enzyme‐substrate interactions at those sites. For residues in the P2‐P2’ positions, the nitrogen‐affixed hydrogen atoms play a role in substrate recognition and stabilization (Figure S3) [52, 68]. However, previous work identified that the carbonyl in the P1 position interacts with the zinc‐ion complex of the MMP active site, while the amide is free [43, 69]. Thus, we rationalized that N‐substitutions in the P1 position would still enable proteolytic cleavage [41, 43]. In addition, there is a high incidence of proline residues, which are naturally “N‐substituted” as tertiary amides, in the P3 position for MMP‐degradable sequences, implying other N‐substituted residues would be tolerated in this position [43, 44, 67]. Previous work found that substitutions in the P1 and P3 were among the most well‐tolerated by various MMPs, and that substitutions in the P2 and P3’ positions, as well as substitution combinations, may be used to further tailor degradation rates in future studies [43]. Despite the tolerance of N‐substitutions in P1 and P3, their presence still alters enzyme‐substrate interactions and results in a slowed cleavage rate.

Here, to evaluate the degradation kinetics of each peptomer sequence to MMP‐2 and MMP‐9 in solution, fluorescent resonant energy transfer (FRET) was implemented using a methoxycoumarin (MCA) fluorophore and a dinitrophenyl (DNP) quencher pair. MCA and DNP were successfully conjugated to the N‐terminus and C‐terminus, respectively, of each peptomer (Figure S4). Activity‐normalized concentrations [43] of MMP‐2 (0.13 µg mL⁻¹) and MMP‐9 (0.50 µg mL⁻¹) were mixed with each FRET‐labeled peptomer, and fluorescence increase due to fluorophore liberation was tracked over 4 h, demonstrating cleavage of all peptomer sequences (Figure 1B, C, Table S1). For both MMPs, PAALVA degraded the fastest, as expected, due to its optimization for the gelatinases. The singly‐substituted sequences degraded more slowly, with P3‐NAla showing slower degradation kinetics than P1‐NAla. The tandem substituted sequence degraded the slowest. Thus, both the location and number of peptoid substitutions affected the degradation kinetics. This data is supported by previous work that explored the effect of different peptoid substitutions on substrate degradation rate to an extended number of MMPs, where the trends held for MMP‐1, 2, 8, 9, and 13 [43]. Therefore, this set of linkers demonstrated promise for use in staged release in response to cell‐secreted proteases, as the degradation kinetics slowed systematically across both MMP types.

Previous work explored the use of other non‐natural amino acids, such as D‐amino acids, to systematically control peptide linker degradability; however, sequences with D‐amino acids demonstrated significant cytotoxicity at concentrations greater than 0.1 mm [48, 70]. Thus, before applying the peptomers to dual release, we evaluated their cytocompatibility with MDA‐MB‐231 breast cancer cells and human dermal fibroblasts. Each peptomer was dissolved in complete cell culture media at concentrations ranging from 1 µm to 100 mMm and incubated with cells for 3 days. Cell viability remained greater than 94% for all concentrations tested, across both cell lines, when normalized to untreated controls (Figure 1D, E). These data indicated that any toxicity measured in future experiments would stem from the chemotherapeutic, rather than the peptomer linker.

2.2. Fluorogenic Hydrogel Release Models

We next investigated the ability of the peptomer linkers to control the release of multiple cargoes from a model hydrogel system. We selected the fastest (PAALVA) and slowest (P3, P1‐NAla) degrading sequences and functionalized the N‐terminus with a fluorophore for tracking and the C‐terminus with a cysteine residue for linking to the hydrogel (Figures S5 and S6). MCA and fluorescein (6fam) were selected as orthogonal fluorophores for independent tracking due to their non‐overlapping fluorescent profiles. Polyethylene glycol (PEG) was selected as the polymer backbone for the hydrogel due to its biocompatibility, ease of modifiability, and general promise as a drug delivery carrier [71, 72]. 4‐arm PEG‐norbornene (PEG‐NB, 10 kDa) was synthesized (Figure S7), and 95% of the norbornene groups were crosslinked with a linear PEG‐dithiol (3.4 kDa), while the remaining 5% were reacted with the cysteine of the fluorescently‐labeled peptomers via UV‐initiated thiol‐ene linking (Figure 2A).

FIGURE 2.

Decoupled release kinetics of fluorogenic cargo from a hydrogel. (A) Schematic representing conditions for thiol‐ene linking of the peptomers into PEG‐hydrogel systems. (B) Substrate release from gels formed in 24‐well plates was monitored using a fluorescent area scan. Prior to treatment with MMP, the fluorescent signal was localized in the center of the well. Over time, the peptomers were cleaved, and fluorogenic cargo was released into the surrounding media. The fraction remaining was calculated by monitoring a decrease in fluorescence from the center of the well. Both the MCA (325/393 Ex./Em.) and fluorescein (494/518 Ex./Em.) channels were monitored to track the release of both cargoes. The fraction of fluorogenic substrate remaining over 6 days was monitored for hydrogels loaded with MCA‐labeled peptide and fluorescein‐labeled tandem peptomer (C, E) and fluorescein‐labeled peptide and MCA‐labeled tandem peptomer (D, F) after treatment with MMP‐2 and MMP‐9, respectively. Circles represent MCA‐fluorophores, while squares represent fluorescein. Black corresponds to the peptide sequence, and blue to the tandem‐substituted sequence. Trends were fit to GraphPad Prism 9's built‐in nonlinear regression one‐phase decay model, and half‐lives were calculated for each cargo in response to MMP‐2 (G) and −9 (H) activity. All error bars represent the standard deviation from three technical replicates. Statistical analyses represent the results of GraphPad Prism's ordinary one‐way ANOVA, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001.

For each hydrogel, two different peptomers were incorporated at equal concentrations to compare both fast‐ and slow‐releasing cargo. Both combinations of peptomer and fluorophore were tested to account for differences in cargo size and polarity, which would likely be present in therapeutic applications. For example, hydrogels containing MCA‐PAALVA and 6fam‐P3, P1‐NAla were compared to hydrogels containing MCA‐P3, P1‐NAla and 6fam‐PAALVA (Figure 2A). Hydrogels were treated with 0.50 µg mL⁻¹ of MMP‐2 or ‐9 and peptomer release from the hydrogel was tracked using a fluorescent area scan over the course of 6 days (Figure 2B).

Regardless of the peptomer‐cargo pairing, the fluorophore linked to PAALVA was always released significantly faster than the fluorophore linked to P3, P1‐NAla, for both MMP‐2 (Figure 2C, D) and MMP‐9 (Figure 2E, F). This trend held even when the loading concentration of fluorophore was doubled, as well as conditions when fluorophores were present in non‐equimolar ratios (Figure S8). The shallower curves observed for MMP‐9 reflected the lower activity of the enzyme. For the PAALVA peptomer, the half‐life for release was not impacted by the conjugated fluorophore (Figure 2G, H, Table S2); both fluorophores showed a half‐life of 12 h in response to MMP‐2 and 163 h in response to MMP‐9. The P3, P1‐NAla linker showed different behavior in response to MMP‐2. In that case, the half‐life for the 6fam‐conjugated peptomer was 114 h, whereas the half‐life for the MCA‐conjugated peptomer was 48 h. In response to MMP‐9, however, both P3, P1‐NAla peptomers showed minimal release with a half‐life of 3280 h.

It is possible that the fluorophores lead to different diffusivities for the P3, P1‐NAla degradation products, and that the MMP‐9 activity is too low to tease these differences out on this experimental timescale. To test this possibility, we measured the release of fluorophore‐conjugated P3, P1‐NAla fragments that were simply encapsulated into the hydrogel (Figure S9 and S10). No differences were observed in the half‐life of the fragments with either fluorophore, and furthermore, the half‐life was less than an hour, confirming that the release kinetics of the tethered peptomers were dominated by the enzymatic cleavage rate rather than diffusion. In addition, rheological gelation experiments indicated that peptomers with both fluorophores achieved similar extents of reaction in the hydrogel (Figure S11). Overall, the chemical and physical properties of the cargo did not impact proteolytic trends, prompting us to develop a therapeutically active system.

2.3. Synthesis of Peptomer Drug Conjugates

While the fluorophores were easily conjugated to peptomers via lysine residues or conjugation at the C‐terminus, peptomer linker chemistries were more carefully crafted for the conjugation of doxorubicin and geldanamycin in an effort to balance effective drug loading with drug activity retention. In both cases, the drug was slightly modified using natural chemical handles to generate a linker capable of amide coupling with peptomer termini (Scheme 1A). For doxorubicin, the therapeutic was Fmoc‐protected before the C14 hydroxyl was reacted with glutaric anhydride to create an acid‐terminated ester bridge for facile amide coupling to the N‐terminus of the desired peptomer (Scheme 1B) [73]. The ester bridge was selected for its previously reported intracellular hydrolysis, which retains doxorubicin activity [73, 74]. For geldanamycin, several derivatives have been developed to improve solubility, and the vast majority modify the molecule at the C17 position where the methoxy group acts as a vinylogous ester [75, 76, 77, 78]. Exploiting this reactive handle, we attached a diamine linker at the C17 position and used it for amide coupling to peptomer C‐termini (Scheme 1C). For both drug conjugates, we selected two linkers that comprised the library's kinetic range: 1) a “fast releasing” derivative linked to the peptide sequence, and 2) a “slow releasing” derivative linked to the tandem‐substituted peptomer sequence. A cysteine residue was retained at the opposite terminus to facilitate downstream hydrogel tethering for all PDCs. MALDI and analytical HPLC confirmed the successful synthesis of all PDCs with a purity of at least 96% (Figures S12 and S13). The drugs were attached at opposite termini for synthetic ease, and a control study confirmed that cysteine modification at either terminus did not impact degradation kinetics to MMPs (Figure S14).

SCHEME 1.

Synthesis of doxorubicin and geldanamycin peptomer‐drug conjugates. (A) General anatomy of PDCs, consisting of a chemical linker bridging the drug to the cysteine‐tagged peptomer. (B) Synthesis of the DOXPDC via ester linkage: a) Fmoc‐OSu, DIPEA, DMF, Ar, 4 h, protected from light, b) glutaric anhydride, DIPEA, DMF, Ar, 24 h, c) HATU, DIPEA, peptomer, DMF, Ar, 4 h, d) 20% piperidine in DMF, 0.5 h, e) β‐mercaptoethanol, NMM, DMF, 3hr, Ar. (C) Synthesis of GDPDC via diamine linkage: f) 1,5‐diaminopentane, DCM, 0.5 h g) HATU, DIPEA, DMF, Ar, 4 h. h) DCM/TFA/TIPS (95/2.5/2.5%v/v), 2 h.

Whereas the ester linkage in the doxorubicin peptomer drug conjugate (DOXPDC) enables recovery of the unmodified doxorubicin (Figure 3A), the amide bond in the geldanamycin PDC (GDPDC) implies that a fragment of the peptomer will remain conjugated to the drug post‐cleavage (Figure 3B). We therefore tested the activity of geldanamycin, doxorubicin, and their corresponding PDCs against plated MDA‐MB‐231 cells using a luminescent viability assay to determine IC₅₀, as well as flow cytometry to better distinguish between live and apoptotic cells. All experiments used the PAALVA linker to assess the impact of conjugation on chemotherapeutic activity. IC₅₀ experiments revealed that DOXPDC retained the same activity as unmodified doxorubicin with a 3‐day IC₅₀ value of approximately 0.51 µMm, while GDPDC had an IC₅₀ value approximately 4 times higher than unmodified geldanamycin at 15.3 µm (compared to 3.70 µm for the free drug) (Figure 3C, D). This implies that a significantly higher dosage of the geldanamycin PDC would be required to achieve the same therapeutic effect as geldanamycin. We note that other geldanamycin derivatives (e.g., tanespimycin) have increased IC₅₀ values compared to geldanamycin (Figure S15) and still showed some efficacy in clinical trials [79, 80, 81]. In addition, the diamine linker in these experiments did significantly improve the solubility of geldanamycin such that solubility limitations would not prevent the use of higher doses in aqueous environments. In scenarios where a combination of drugs is administered to achieve synergistic effects (as is being explored for doxorubicin and geldanamycin), the synergy may compensate for the reduced potency from bioconjugation to peptomers, while gaining control over the drug release profile. To explore the relevance of this in our system, the combination index of different concentrations of doxorubicin and geldanamycin used together was calculated. It was verified that synergy is achieved against MDA‐MB‐231 cells, in vitro, over a 3‐day period, with combination indices of ∼0.90 for three tested combinations (Figure S16).

FIGURE 3.

PDC degradation mechanisms and cytotoxicity. (A) Proposed mechanism for DOXPDC degradation, where the ester linkage allows for hydrolysis of the peptomer fragment from the drug post‐cleavage by MMPs. (B) Proposed mechanism for GDPDC release, where the peptomer fragment remains attached to the drug post‐cleavage due to amide bond stability. Dose‐response curves of MDA‐MB‐231s to treatment with doxorubicin and DOXPDC (C) and geldanamycin and GDPDC (D) after a 3‐day incubation period. The data were fit to a sigmoidal curve using Graphpad Prism 9's built‐in log(inhibitor) vs. normalized response function. Error bars represent the standard deviation of three technical replicates. Representative flow cytometry dot‐plots and gating showing cellular response to treatment with 1 µm DOXPDC (E) and 4 µm GDPDC (F) after 3 days. (G) Flow cytometry results showing proportions of healthy cells (‐/‐), early apoptotic cells (‐/+), late apoptotic cells (+/+), and dead cells (+/‐) after a 3‐day treatment with doxorubicin, geldanamycin, and their associated PDCs.

Flow cytometry was used to further evaluate cell response to treatment with the PDCs. Unstained cells (‐/‐) were deemed viable, apotracker‐positive cells (‐/+) were classified as early apoptotic, zombie yellow‐positive cells (+/‐) were considered dead, and double‐positive cells (+/+) were considered late apoptotic (Figure 3E, F). Flow cytometry analysis of each of the drugs and PDCs at the unmodified drug IC₅₀ concentration confirmed the observations from IC₅₀ analysis, where cellular response for doxorubicin and the DOXPDC appear identical, but the GDPDC reflects a decrease in potency (Figure 3G). The retained activity was deemed sufficient to move forward with the model delivery system, and future work will be conducted to optimize linker chemistry for geldanamycin activity retention. Interestingly, the flow cytometry data also provided mechanistic insight into each drug, revealing that doxorubicin appears to trigger the necrotic cell death pathway at the tested concentrations, while geldanamycin facilitates cell death through apoptosis, consistent with what has been observed in the literature [82, 83, 84].

2.4. Independently Controlled Codelivery of Chemotherapeutics

With the functional PDCs in hand, we next investigated the dual release of geldanamycin and doxorubicin from the hydrogels in response to exogenous MMPs via Shimadzu 8060 triple‐quadrupole LC/MS (QQQ LC/MS). Hydrogel formulations were slightly modified from the previous UV‐initiated system due to geldanamycin instability under UV conditions. Rather than a photoinitiated system, 8 wt.% PEG hydrogels were formed using a base‐catalyzed Michael addition (Figure S17). For this experiment, equal concentrations of each drug (0.235 mm) were loaded into the hydrogels, corresponding to the amount of PDC needed to react with 2.5% of the acrylate groups (Figure 4A). Four conditions were tested: dual fast release (both drugs contained PAALVA linkers), dual slow release (both drugs contained P3, P1‐NAla linkers), DOX‐PAALVA (fast)/GD‐ P3, P1‐NAla (slow), and DOX‐ P3, P1‐NAla (slow)/GD‐PAALVA (fast). Hydrogels were treated with 0.50 µg mL⁻¹ of MMP‐2 or ‐9, and release was quantified using QQQ LC/MS over the course of 3 days. Standard curves were developed on the QQQ LC/MS prior to experimentation to relate the area under the curve (AUC) for each analyte to a concentration, covering the entire possible range (Figure S18). Analytes included the doxorubicin‐peptomer fragment, pure doxorubicin, and the geldanamycin peptomer fragment, whose structures are shown in Figure 3A, B. DOX and DOX‐fragment were both considered release products, and their concentration in the supernatant was added together to quantify release at each time point. To control for any release not due to enzymatic activity, an identical set of control hydrogels was synthesized and treated with buffer. All results reflected in Figure 4 are normalized to release from these control hydrogels.

FIGURE 4.

Independently controlled, dual release of doxorubicin and geldanamycin from PEG hydrogels via peptomer linkers. (A) Schematic representing conditions for thiol‐ene linking of PDC‐loaded PEG hydrogels. Each hydrogel was loaded with a fast or slow‐releasing GDPDC paired with a fast or slow‐releasing DOXPDC. (B‐E) Fraction remaining of each drug from each hydrogel formulation in response to 0.50 µg mL⁻¹ MMP‐2 activity over 3 days, as quantified by QQQ LC/MS. Solid bars represent PDC samples from hydrogels with homogeneous peptomer linkers, while checkered bars represent PDC samples from hydrogels with mixed peptomer linkers. (F) Half‐lives for each PDC to MMP‐2 were calculated from an exponential decay fit. Error bars represent the standard deviation from the two samples tested per PDC. (G‐J) Percent remaining of each drug from each hydrogel formulation in response to 0.50 µg mL⁻¹ MMP‐9 activity over 3 days, as quantified by QQQ LC/MS. (F) Half‐lives for each PDC to MMP‐9 were calculated from an exponential decay fit. Error bars represent the standard deviation from the two samples tested per PDC. Statistical analyses represent the results of GraphPad Prism's ordinary one‐way ANOVA, ^* p < 0.05, ^** p < 0.01, ^*** p < 0.001, ^**** p < 0.0001.

Results revealed identical release profiles for the dual‐fast and dual‐slow releasing hydrogels in response to MMP‐2 activity (Figure 4B, C). In hydrogels with both fast and slow degrading peptomer linkers, the drug conjugated to the PAALVA peptomer released faster than those conjugated to P3, P1‐NAla (Figure 4D, E). Half‐lives were calculated for each PDC (Figure 4F, Table S3) to further investigate the release kinetics of each individual PDC. The half‐lives of each PDC remained constant, regardless of whether it was released from a homogenous hydrogel (dual fast/dual slow) or a heterogeneous hydrogel. Additionally, the half‐lives of the fast‐releasing PDCs remained constant across both drugs, at approximately 31 h, reflecting independence from cargo properties or whether the drug was linked to the N‐terminus or the C‐terminus. For the slow‐releasing PDCs, slightly different half‐lives were observed for geldanamycin and doxorubicin PDCs. While the DOX‐P3, P1‐NAla PDCs showed a half‐life of 224 h (9.3 days), the GD‐P3, P1‐NAla showed a half‐life of 116 h (4.9 days). The source of this difference is likely attributed to the short timescale of this experiment, which does not encompass the range calculated for the longer half‐lives. However, the extrapolation is enough to determine that the P3, P1‐NAla sequence significantly slows the rate at which PDCs are released from this system in response to MMP activity.

When MMP‐9 was used instead of MMP‐2, similar results were observed (Figure 4G, K, Table S3). All release rates were slower due to the inherently lower activity of MMP‐9. Once again, all PDCs retained a constant half‐life regardless of whether the hydrogel contained homogeneous or mixed linker types. The half‐life of all fast‐releasing PDCs was 69 h (2.9 days), regardless of the drug attached, and remained within the range of the experimental timeline. The slow‐releasing PDCs still showed some difference in their half‐lives of release: 1,041 h (43.4 days) for DOX‐P3, P1‐NAla, and 1,476 h (61.5 days) for GD‐P3, P1‐NAla. This difference was less than that observed for the slow‐releasing PDCs to MMP‐2. In addition, GD‐P3, P1‐NAla released slightly faster to MMP‐9 compared to DOX‐P3, P1‐NAla, whereas the trend was flipped for MMP‐2. Overall, these results indicate the PDCs preserve the kinetic control imparted by peptomer sequences during release from a hydrogel.

2.5. Evaluation of Dual Release From PDC‐Loaded Hydrogels in Vitro

Finally, MDA‐MB‐231s were incubated with PDC‐loaded hydrogels over a 7 day period to evaluate cellular response to treatment with different PDC formulations. To demonstrate proof‐of‐concept, we loaded the amount of drug required to reach 2xIC₅₀ values in the surrounding supernatant if all the drug were to be released from the hydrogel. IC₅₀ values of the specific drug or PDC were used in all instances. Cells were plated in 24‐well plates, and hydrogels were synthesized in trans‐well inserts that would allow for their degradation to cell‐secreted and serum proteases and diffusion of released drug through the membrane (Figure 5A). This setup also facilitated analysis via flow cytometry. To understand the degradation of the peptomers to serum proteases, the FRET‐functionalized peptomers were incubated in cell culture medium containing 10% fetal bovine serum (Figure S19). The cleavage profiles indicated that serum proteases can cleave the peptomers, but that cleavage trends for PAALVA and P3, P1‐NAla are maintained despite exposure to a mixture of proteases.

FIGURE 5.

Flow cytometry analysis of cellular response to treatment with PDC‐loaded hydrogels. (A) Schematic showing PDC‐loaded hydrogels suspended above MDA‐MB‐231s in trans‐well inserts versus controls, where drugs are dissolved directly into the media. (B) Flow cytometry analysis showing the distribution of healthy, early apoptotic, late apoptotic, and dead cells per condition at 1, 3, 5, and 7‐day time points. (C) Graph representing the fraction of viable cells over time, per condition. Data were fit to a one‐phase exponential decay model.

Several conditions were used to compare the activity of PDC‐loaded hydrogels to free drug in solution, mimicking administration without controlled release mechanisms. Three solution‐phase conditions were included: 1) 2xIC₅₀ of both drugs, 2) a condition that contained 2xIC₅₀ of DOX for the first three days and 2xIC₅₀ of each drug for the final 11 days, and 3) a condition that contained 2xIC₅₀ of GD for the first three days, and 2xIC₅₀ of each drug for the final 11 days. The latter two were designed to mimic the behavior of dually‐loaded hydrogels that release either drug at a faster rate, where exposure to one is higher during earlier timepoints. Unloaded hydrogel controls and untreated cells were also included to control for any non‐drug or hydrogel‐related cell death.

Cell health was quantified via flow cytometry on days 1, 3, 5, 7, and 14 post‐treatment. For the 5‐, 7‐, and 14‐day timepoints, dead cells were stained and fixed every 3 days to prevent skewed data collection resulting from the breakdown of dead cells in solution over time. The proportion of healthy, early apoptotic, late apoptotic, and dead cells was recorded for each condition at each time point (Figure 5B, Figures S20 and S21). On day 1, the hydrogel‐treated cells retained high viability, aside from those treated with the dual‐fast‐releasing hydrogel, which decreased the percentage of healthy cells by 17% after 1 day. On the contrary, each of the drug solutions triggered significant cell death, where only ∼20% of cells remained healthy, except for the geldanamycin‐only treatment condition, which showed a more delayed effect. On day 3, the dual fast‐releasing condition reduced viability to 34%, while the hydrogels with decoupled release kinetics decreased viability by 10–15%. The dual slow‐releasing hydrogels did not have much impact on cell health. Each of the drug solutions resulted in entirely necrotic and apoptotic cell populations by day 3. The trends continued through day 7 and 14, where, of the hydrogel conditions, the dual fast‐releasing hydrogel resulted in the greatest cell deterioration, followed by the fast‐DOX‐releasing system, the fast‐GD‐releasing system, and finally the dual slow‐releasing system (Figure 5C). The experiment was then repeated using A549 human lung carcinoma cells to evaluate whether the trends were cell‐specific. New IC₅₀ values were generated for this cell line in response to doxorubicin, geldanamycin, and GD‐amine, and concentrations used during the experiment were adjusted accordingly (Figure S22). Results mirrored those observed for the MDA‐MB‐231s, aside from the fact that the fast‐GD‐releasing system resulted in a faster decline in cell health than the fast‐DOX‐releasing system for this cell line (Figures S23 and S24).

The cell data indicated that the release kinetics of the drugs were decoupled according to the peptomer sequence, as expected from the hydrogel studies with exogenous MMP; these data also indicated how controlling drug release rates can predictably impact cell viability in vitro. When both drugs were released quickly, necrosis and apoptosis were triggered earlier. When one was released quickly, and the other was slower, this response was delayed. When both drugs were released slowly, dramatic cellular responses were not observed over the 14‐day time course, which aligns with previous half‐life evaluations. An additional experiment was performed to evaluate the amount of drug remaining in the dual‐fast‐releasing hydrogel after the 7‐day time point. Results revealed that not all of the drug had yet been released, indicating that the cell death at day 14 was due in part to additional release of the drug (Figure S25). Delayed, gradual release of the drugs from the controlled release material effectively slowed the rate of cell death when compared to solution conditions at equal concentrations. Thus, the peptomers showed potential to mitigate burst effects and gradually administer the drugs in a way that could translate clinically to reduced side‐effects and less frequent treatments, potentially improving patient burden and adherence. Future studies are needed to optimize drug loading concentrations and scheduling, and to advance the system beyond in vitro models.

3. Conclusions

We have reported the use of proteolytically‐degradable peptomer sequences as an approach to control the sequential release of small molecule chemotherapeutics from a hydrogel system. The peptomers present a facile solution to the substrate overlap and associated nonspecific release often observed for peptides designed for a similar function. Strategic peptoid substitutions resulted in slower cleavage kinetics of the substrate to both MMP‐2 and MMP‐9, enabling the design of fast‐releasing and slow‐releasing drug linkers in a hydrogel. We tested these linkers with two different chemotherapeutic drugs in a hydrogel during culture with cancer cells. Despite the presence of multiple proteases beyond MMP‐2 and MMP‐9, our peptomer linkers still demonstrated distinct release rates of the chemotherapeutics. In vitro studies revealed that the rate of drug release was proportional to cell death rates and that the hydrogel system achieved staged release by mixing fast‐release and slow‐release linkers compared to free drug in solution. Thus, the peptoid substitutions offer a strategy for systematic control of release rate in complex biological environments with mixed protease types.

Our strategy for kinetic control of proteolytic cleavage rate improves on current state‐of‐the‐art release systems that rely entirely on cargo‐material interactions. Furthermore, while we selected a hydrogel model for this work, peptomer drug‐conjugates are versatile and may easily be applied to other drug delivery systems such as polymeric nanoparticles or micelles. Future work will focus on optimizing both the linker chemistry and the carrier architecture to advance the system toward in vivo applicability. Current efforts are directed at improving the GDPDC linker to enhance drug potency while preserving sufficient water solubility. In addition, current efforts are extending this release strategy to alternative delivery modalities, including micelle formulations that have shown efficacy in vivo. Altogether, the results of this study present peptomer linkers as a promising candidate for independently controlled codelivery, an approach that has the power to simplify dosing regimens and alleviate patient burden in the context of cancer therapy.

4. Experimental/Methods Section

4.1. Materials

4.1.1. Cell Culture

All materials were used as received unless otherwise noted. MDA‐MB‐231 human breast adenocarcinoma cell line was purchased from ATCC (batch no. 70029549; Manassas, VA). A549 human lung carcinoma cell line was purchased from Millipore Sigma (batch no. 22E008; Burlington, MA). Human dermal fibroblasts from a healthy, 27 year old female were purchased from Lonza (Basel, Switzerland). Dulbecco's modification of Eagle's Medium (DMEM), Corning Regular Fetal Bovine Serum, and Corning 0.25% trypsin (0.1% EDTA in HBSS w/o calcium, magnesium, and sodium bicarbonate) were purchased from Fisher Scientific (Waltham, MA). CellTiter‐Glo 3D Viability Assay was purchased from Promega (Madison, WI). Apotracker green and zombie yellow flow cytometry stains were purchased from BioLegend (San Diego, CA).

4.1.2. Reagents

All reagents were used as received unless otherwise noted. l‐Amino acids, sarcosine, and modified lysines were all Fmoc‐protected and purchased from Chem‐Impex International, Inc. (Wood Dale, IL), along with Rink amide resin, 2‐chlorotrityl chloride resin, O‐(1H‐6‐chlorobenzotriazol‐1‐yl)‐N, N, N′, N′‐tetramethyluronium hexafluorophosphate (HCTU, ≥99%), N, N′‐diisopropylcarbodiimide (DIC, ≥99%), 1‐hydroxybenzotriazole (HOBt, wetted with not less than 20 wt. % water), and N, N′‐diisopropylethylamine (DIPEA, 99.8%). Acetic anhydride (>99%), pyridine (>99%), beta‐Mercaptoethanol (99%), and triisopropylsilane (TIPS, 98%) were purchased from Acros Organics (Fir Lawn, NJ). N‐Methylmorpholine (NMM, 99%), piperidine (99.5%), and 4‐aminophenylmercuric acetate (APMA, ≥97%), 2,2’‐(Ethylenedioxy) diethanethiol (EDT), triethylamine (TEA, >99.5%), and 2,2,2‐trifluoroethanol (TFE, NMR grade) were purchased from Millipore Sigma (Burlington, MA). NHS‐fluorescein (5/6‐carboxyfluorescein succinimidyl ester, mixed isomer) was purchased from Thermo Fisher Scientific (Waltham, MA). Fmoc N‐hydroxysuccinimide ester (fmoc‐OSu) was purchased from aapptec (Louisville, Kentucky). Sodium hydroxide (NaOH, ≥97%), 1,5‐diaminopentane (99%), and all solvents were purchased from Fisher Scientific (Hampton, NH) at the following purity levels: dimethylformamide (DMF, ≥99.8%), trifluoroacetic acid (TFA, ≥99.5%), acetic acid (AcOH, ≥99%), diethyl ether (ether, ≥99%), acetonitrile (ACN, ≥99.9%), 2‐propanol (IPA, ≥99.5%), dimethylsulfoxide (DMSO, ≥99.7%).

Full‐length recombinant, carrier‐free human matrix metalloproteinases were purchased from R&D Systems, Inc. (Minneapolis, MN). Gibco Collagenase, Type 1, from C. histolyticum was purchased as a lyophilized powder from Thermo Fisher Scientific (Waltham, MA) with a specific activity of 250 units mg⁻¹. Tris‐HCl (≥99%) and sodium chloride (NaCl, ≥99%) were purchased from Fisher Scientific. Brij‐35 and calcium chloride (CaCl₂, ≥96%) were purchased from Acros Organics.

4‐Arm 10 kDa poly(ethylene glycol)‐amine (PEG‐amine, ≥95%) and 4‐arm 10 kDa poly(ethylene glycol)‐acrylate (PEG‐acrylate≥95%) were purchased from JenKem Technology (Plano, TX). 5‐Norbornene‐2‐carboxylic acid (98%) and 3.4 kDa linear PEG‐dithiol were purchased from Millipore Sigma. Lithium phenyl(2,4,6‐trimethylbenzoyl)phosphinate (LAP, 98%), O‐(7‐azabenzotriazol‐1‐yl)‐N, N, N′, N′‐tetramethyluronium hexafluorophosphate (HATU, ≥98%), glutaric anhydride (98%), and doxorubicin hydrochloride (95.0+%) were manufactured by TCI America and purchased from Fisher Scientific. Geldanamycin (99.78%) and Tanespimycin (99.07%) were purchased from MedChemExpress (Monmouth Junction, NJ).

4.2. Peptide Synthesis

Peptides and peptomers were all synthesized on a Prelude X automated peptide synthesizer (Gyros Protein Technologies) using Rink Amide polystyrene resin (0.48 mmol g⁻¹) or chlorotrityl chloride resin (1.07 mmol g⁻¹) at a scale of 250 µmol. Fmoc groups were removed from the Rink Amide resin and subsequent amino acids by washing twice with 20% piperidine in DMF. For the chlorotrityl chloride resin, the first amino acid (500 µmol, 2.0 equiv.) was reacted onto the swollen resin in DCM in the presence of DIPEA (1.0 mmol, 4.0 equiv.); the resin was then capped using a solution of DCM, methanol, and DIPEA (17:2:1 v/v). Fmoc‐protected amino acids (250 mm, 5.0 equiv.) were coupled using HCTU activator (250 mm, 5.0 equiv.) and NMM (500 mm, 10.0 equiv.). Coupling steps were performed twice. Upon completion of synthesis, peptides and peptomers were cleaved from the Rink amide and chlorotrityl chloride resin. For peptides or peptomers with a Cys(trt) residue, a cleavage cocktail composed of TFA, water, EDT, and TIPS (94:2.5:2.5:1%(v/v)) was used for cleavage and deprotection. TFA and water (95:5%(v/v)) were used for all other peptide and peptomer sequences. All cleavages were performed for 2–4 h. Protected peptomers were cleaved from the chlorotrityl chloride resin using a cleavage cocktail composed of DCM, AcOH, and TFE (8:1:1%(v/v)) for 2 h. The resin was then filtered off, and the peptomers were redissolved in acetonitrile/water mixtures for purification.

MCA‐functionalized peptomers were prepared using N ^α‐Fmoc‐N ^ε‐7‐methoxycoumarin‐4‐acetyl‐L‐lysine (fmoc‐Lys(Mca)) at the N‐terminus, and N ^α‐Fmoc‐N ^ε‐2,4‐dinitrophenyl‐L‐lysine (fmoc‐Lys(Dnp), FRET peptomers) at the C‐terminus, which were added during automated peptomer synthesis. For the fluorescein‐functionalized peptomers, NHS‐fluorescein (1.5 equiv.) was reacted onto the free amine of the N‐terminal glycine residue using HOBt (1.5 equiv.) and DIC (1.5 equiv.) in DMF for 2 h prior to cleavage from the resin.

4.3. Peptide Purification

All peptides and peptomers were precipitated into a 10 fold volume of chilled diethyl ether, washed 2x with fresh diethyl ether, then centrifuged to collect the product. Those that did not precipitate out into diethyl ether (the small, non‐functional library) were subjected to rotary evaporation to remove the cleavage cocktail. Peptomers were then dissolved in a mixture of acetonitrile and water (ranging from 10–40% acetonitrile) with 0.1% TFA and purified using a semi‐prep C18 column on a Dionex UltiMate 3000 UHPLC with a 15 min gradient from starting composition to 100% acetonitrile at 10 mL min⁻¹, with gradient adjustments as needed. Peptomers were collected by their 214 nm UV signal, frozen, and lyophilized. Final products were analyzed via high‐performance liquid chromatography (HPLC) with an analytical C18 column and matrix‐assisted laser desorption/ionization time‐of‐flight (MALDI‐TOF) mass spectrometry using a Bruker autoflex maX instrument to assess purity and confirm molecular weight (Figures S1, S4–S6, S9, S12, S13).

4.4. Synthesis of Doxorubicin Peptomer Drug Conjugates

4.4.1. Synthesis of N‐fmoc‐DOX (fmoc‐DOX)

Fmoc protection of amines has been previously described in a number of studies [85, 86, 87]. In this study, doxorubicin hydrochloride (100 mg, 1.0 equiv.) was added to an argon purged flask and dissolved in anhydrous DMF (3 mL). DIPEA (0.124 mL, 5 equiv.) was added, followed by fmoc‐OSu (116 mg, 2 equiv.) in anhydrous DMF (3 mL). The reaction was protected from light and stirred at room temperature for 4 h. After this time, the DMF was removed via rotary evaporation, and product purity was evaluated using analytical HPLC. Yield = 96%.

4.4.2. Synthesis of N‐fmoc‐DOX‐14‐O‐Hemiglutarate (fmoc‐DOX‐hg)

Esterification of the glycolyl group on fmoc‐DOX was carried out as previously described [73]. Briefly, the crude product from the previous step was redissolved in anhydrous DMF (4 mL) in an argon purged flask in the presence of DIPEA (0.374 mL, 14 equiv.). Glutaric anhydride (140 mg, 8 equiv.) was added, and the reaction vessel was protected from light and reacted overnight under argon. DMF was removed by rotary evaporation, and the product was purified using a semi‐prep C18 column on a Dionex UltiMate 3000 UHPLC using acetonitrile and water as the mobile phase. A linear gradient from 33% to 100% ACN was applied using a flow rate of 10 mL min⁻¹. The detection wavelength was set to 470 nm. Yield = 71%.

4.4.3. Synthesis of DOXPDCs

Generally, fmoc‐DOX‐hg (1 equiv.) was dissolved in DMF (5 mg mL⁻¹) and activated using HATU (1.5 equiv.), in the presence of DIPEA (8 equiv.), under argon, while stirring for 15 min at room temperature. StBu‐protected peptide (1.2 equiv.) was dissolved in DMF and added to the reaction vessel, which was stirred at room temperature for 3 h under argon. After 3 h, piperidine was added to a concentration of 20% (v/v) and reacted for 30 min. The product was then precipitated into 10x volume chilled diethyl ether, centrifuged, and washed twice more with fresh diethyl ether. The next day, the crude product was redissolved in the same volume of DMF containing 20% (v/v) 𝛽‐mercaptoethanol and 0.1 M NMM and stirred under argon at room temperature for 4 h to remove the StBu group. The product was precipitated with diethyl ether as described above, then purified via UHPLC using the same method as for fmoc‐DOX‐hg. Yield = 47%. (Figure S13).

4.5. Synthesis of Geldanamycin Peptomer Drug Conjugates

4.5.1. Synthesis of Amine‐Modified Geldanamycin (GD‐Amine)

C17‐modified geldanamycin has been previously synthesized, and this protocol was adapted from a collection of literature sources [76, 77, 88]. Briefly, geldanamycin (70 mg, 1 equiv.) was dissolved in DCM (11 mL) in the presence of 1,5‐diaminopentane (0.015 mL, 1 equiv.) for 30 min exactly, at room temperature. Reaction times longer than 30 min and/or an excess of 1,5‐diaminopentane resulted in the formation of a difunctional side product with amines conjugated at both the C17 and C19 positions on geldanamycin. DCM was removed via rotary evaporation (without heating). The product was purified using a semi‐prep C18 column on a Dionex UltiMate 3000 UHPLC using acetonitrile and water as the mobile phase. A linear gradient of 50% ACN to 100% ACN was applied using a flow rate of 10 mL min⁻¹. The detection wavelength was set to 350 nm. Unreacted geldanamycin was recovered and used for future reactions. Yield = 61%.

4.5.2. Synthesis of GDPDCs

Generally, Trt‐protected peptomers (1.2 equiv.) were dissolved in DMF (5 mg mL⁻¹) and activated by HATU (1.2 equiv.) in the presence of DIPEA (3.5 equiv.) at room temperature under argon for 15 min. GD‐amine (1 equiv.) was added to the flask in DMF and reacted for 3 h. The product was precipitated into water, centrifuged, and dried in a vacuum oven overnight. The next day, the Trt‐protected PDC was deprotected in a solution of DCM, TFA, and TIPS (0.955:0.02: 0.025%(v/v)) for 2 h. The cleavage solution was removed by rotary evaporator, and the PDC was purified via UHPLC using the same procedure reported for the DOX PDCs. Yield = 73%. (Figure S13).

4.6. PEG‐Norbornene Functionalization

4‐arm PEG‐norbornene was synthesized by first activating 5‐norbornene‐2‐carboxylic acid (0.215 mL, 1.76 mmol, 17.6 equiv.) with HATU (608.38 mg, 1.6 mmol, 16 equiv.) in the presence of NMM (0.176 mL, 1.60 mmol, 16 equiv.) for 5 min in anhydrous DMF (2.5 mL) under an argon purge. Next, this solution was added dropwise to an argon purged flask containing 4‐arm, 10 kDa PEG‐amine (1.0 g, 1.0 equiv.) dissolved in anhydrous DMF (4.5 mL), and the reaction proceeded upon stirring for 24 h at room temperature. Upon completion of the reaction, the product was precipitated into chilled diethyl ether (10x volume) and subsequently washed twice with fresh ether, dialyzed for 3 days in distilled water, then lyophilized before use. The product was recovered as a white solid. Yield = 93%, functionalization = 92% (Figure S7). ¹H NMR (dDMSO, 400 MHz): δ 6.20 to 5.86 (m, 2H), δ 3.65 to 3.40 (m, 227H).

4.7. Fluorescence Assays

Peptomer cleavage to MMPs over time was detected through tracking the fluorescent signal of 7‐methoxycoumarin‐4‐acetic acid, which increases upon enzymatic hydrolysis that triggers liberation from the dinitrophenyl quencher. Peptomer substrates were dissolved at 1 mm in DMSO, then diluted to 20 µm in respective MMP buffers (150 mm NaCl, 20 mm tris, 5 m CaCl₂, 0.05 w/v brij, pH 7.4 for MMP‐2, and 150 mMm NaCl, 50 m tris, 10 mm CaCl₂, 0.05 w/v brij, pH 7.5 for MMP‐9) with 10% DMSO. MMP‐2 and MMP‐9 were activated using APMA at a final concentration of 1 mm in buffer for 2 and 24 h, respectively, per the manufacturer's instructions. Immediately upon activation, MMP solutions were diluted in buffer containing 10% DMSO. 50 µL of each peptomer solution, in triplicate, was combined with 50 µL of buffer (controls) or activated MMP, to a final concentration of 10 µm peptomer and 0.13 µg mL⁻¹ MMP‐2 or 0.5 µg mL⁻¹ MMP‐9 in a 96‐well plate. MMP concentrations were selected to be activity normalized [43]. The plate was oscillated for 10 s to mix, then read on a BioTek Synergy H1 Microplate reader at Ex./Em. 325/393 nm over 5 h. Control values were averaged and subtracted from sample values before plotting. This procedure was repeated for the experiment evaluating peptomer cleavage to serum proteases (Figure S19). 10% FBS in phenol‐red‐free media was used in place of MMP solutions.

4.8. Cell Culture

MDA‐MB‐231s, A549s, and human dermal fibroblasts (HDFs) were cultured in DMEM supplemented with 10% FBS in a humidified incubator at 37°C and 5% CO₂. Cells were passaged at 90% confluency and used up to 6 passages. Cells were lifted using 0.25% trypsin with 2.21 mm EDTA and pelleted via centrifugation for 6 min at 130 x g (MDA‐MB‐231s) or 5 min at 220 x g (HDFs, A549s).

4.9. Luminescent Toxicity Assays

MDA‐MB‐231s or A549s were lifted, counted on a hemocytometer, plated at 10,000 cells/well in a 96‐well plate, and incubated for 24 h. After 24 h, cell media was replaced with 100 µL of drug or peptomer solution in complete media in triplicate. The outermost wells were not used and were instead filled with 200 µL of PBS to minimize evaporation around the edges. After 72 h, 100 µL of CellTiter‐Glo Viability Assay was added to each well. The plate was oscillated for 5 min, incubated at room temperature for 30 min, and the luminescence in each well was read using the microplate reader listed above. Viability was determined by subtracting background signals from each well, then normalizing to readout from cells that received no drug treatment.

For dose‐response curves, multiple concentrations of each drug or PDC were used, and the data were plotted as normalized response vs. log(concentration). The data was then fit to a sigmoidal curve using GraphPad Prism's built‐in log(inhibitor) vs. normalized response curve function. The normalized response was inputted as the luminescent output, with 100% signal set to the luminescent value for untreated cells, and 0% set to the background luminescent observed in wells with no cells. The function:

$$Response=\frac{100}{1+{10}^{\log \left(\frac{Concentration}{I{C}_{50}}\right)}}$$

assumes that the dose response curve has a standard slope, equal to a Hill slope (or slope factor) of −1.0. This is the slope expected when a ligand binds to a receptor following the law of mass action, and is the slope expected of a dose‐response curve when the second messenger created by receptor stimulation binds to its receptor by the law of mass action (Figure 3, Figures S15 and S22).

4.10. Combination Index (CI) calculations

MDA‐MB‐231 cells were treated with a series of concentration combinations of doxorubicin and geldanamycin (Figure S16), and viability after a 3 day period was evaluated via luminescent viability assay as described above. The following equation was used for the calculation of the CI index [89]:

$$CI=\frac{D_A}{{D_x}_A}+\frac{D_B}{{D_x}_B}$$

Where D_A and D_B are the doses of drug 1 and drug 2 used in combination to achieve a specific level of cell death, and D_xA and D_xB are the doses of these drugs individually used to achieve the same effect. CI <1 indicates synergism, CI = 1 indicates an additive effect, and CI > 1 indicates antagonism [89]. D_A and D_B were determined experimentally, and D_xA and D_xB were calculated using the sigmoidal dose‐response curve fitted to the cytotoxicity data for each individual drug as described above.

4.11. Fluorescent Hydrogel Release Experiments

PEG gels were formed by combining PEG‐dithiol and PEG‐norbornene at a thiol‐ene ratio of 0.95:1 to achieve 4 wt.% in water with 10% DMSO, which was incorporated to ensure peptomer solubility. For each gel condition, two cysteine‐tagged fluorescent peptomers were added at thiol‐ene ratios of 0.025:1. The combinations of peptomers used were PAALVAC(Mca)/P3, P1‐NAla (fluorescein), and PAALVAC (fluorescein)/ P3, P1‐NAla(Mca). LAP photoinitiator was included at a final concentration of 0.05 wt%. Gelation was induced by exposure to 365 nm light (10 mW cm⁻², 160 s). 30 µL gels were formed in 24‐well plates. Blank gels were synthesized following the same procedure, without the addition of the fluorescent peptomers.

Gels were synthesized in triplicate per condition and swollen in MMP‐2 or ‐9 buffer with 10% DMSO for 24 h to flush out any unreacted material prior to the start of the experiment. Gels were then treated with 1 mL of either MMP‐2 or ‐9 buffer with 10% DMSO for the controls and blanks, or 0.25 µg mL⁻¹ of activated MMP‐2 or ‐9 in their respective buffer solutions (10% DMSO). Fluorescent area scans were taken using the H1 Synergy Biotek plate reader at predetermined time intervals over the course of 72 h to track peptomer cleavage and subsequent diffusion of the fluorophore out of the gel and into the surrounding supernatant. The area scan divided each well into a 9×9 matrix and took a fluorescent reading at Ex./Em. 325/393 nm (7‐methoxycoumarin) and Ex./Em. 494/518 nm (carboxyfluorescein) for each square in the matrix.

Data analysis proceeded following a previously described procedure [43]. The three squares with the highest intensity in the 9×9 matrix at t = 0, corresponding to the center of the gel where the fluorogenic peptomers were originally tethered, were determined for both fluorescent channels in each sample. These were deemed the region of interests (ROI) and used in all subsequent analyses to track decreases in fluorescence that correspond to peptomer release over time. The fluorescence intensity in the ROI of the blank gels was averaged at each time point and subtracted from the control and test gels to account for background fluorescence. Next, each sample was normalized by the fluorescence intensity of the t  = 0 time point, representing the fraction of substrate remaining in the gel. The replicates within each gel were averaged before averaging across the three gels per condition. The averaged MMP‐2 and MMP‐9 samples were divided by the averaged controls at each time point to account for fluorescence fluctuations and any substrate diffusion not due to proteolysis. The trace for each enzyme–substrate combination was fit to GraphPad Prism 9's built‐in nonlinear regression one‐phase decay model

$$y=y_M+\left(y_0-y_0\right)e^{-kt}$$

where y represents inputted values for % substrate remaining, y_M is the plateau value, which was constrained to 0, y₀ is the initial value, which was set to 1, and k represents the rate of reaction (hr⁻¹). Half‐lives (hours) were calculated as ln(2)/k and are reported with their 95% confidence interval.

The same protocol was adapted to analyze the release from gels with different loading ratios. For this experiment, fluorescent peptomers were added into the gel mixture at thiol (fluorophore 1): thiol (fluorophore 2): ene ratios of 0.06:0.03:1. Four conditions were tested: PAALVA(MCA)/P3, P1‐NAla(6fam) at a 2:1 ratio, PAALVA(MCA)/P3, P1‐NAla(6fam) at a 1:2 ratio, PAALVA(6fam)/P3, P1‐NAla(MCA) at a 2:1 ratio, and PAALVA(6fam)/P3, P1‐NAla(MCA) at a 1:2 ratio.

4.12. Flow Cytometry

Generally, MDA‐MB‐231 or A549 cells were plated in 6, 12, or 24‐well plates and allowed to attach overnight. The media was exchanged for drug or PDC solutions in complete media (1% DMSO) or complete media alone (1% DMSO) for controls. At the time of analysis, conditioned media (containing all dead cells) were collected, the cells were lifted, and trypsin was deactivated using the conditioned media to collect both live and dead cells. Cells were centrifuged at 130 x g for 6 min (MDAs) or 220 x g for 5 min (A549s) and washed once with PBS. Cells were stained in a solution comprised of apotracker green (30 nm for MDAs, 35 nm for A549s) and zombie yellow (15 nm for MDAs, 50 nm for A549s) dyes in 10% FBS in PBS at 37°C for 20 min, then washed 3 more times. Zombie yellow controls were generated through heat‐shocking cells at 65°C for 3 min. Apotracker controls were generated through treatment with camptothecin. Data acquisition was performed on a Cytek Northern Lights flow cytometer, and data processing was performed in FlowJo.

4.13. PDC‐Loaded Gel Release Studies

Dually‐loaded PDC gels were formed by combining PEG‐Ac and PEG‐SH in TEA buffer (0.3 m, pH 8) with 10% ACN at a thiol: acrylate ratio of 0.96:1, to a final PEG concentration of 8 wt.%. Each PDC was added to a final gel concentration of 0.235 mm, corresponding to a thiol: acrylate ratio of 0.02:1. Gels formed at room temperature over the course of 1 h (Figure S17) and were swollen overnight in 1 mL of their respective MMP buffers (with 10% ACN) to flush out any unreacted material. The morning of the experiment, gels were washed with buffer once more for 1 h. A supernatant sample was taken after each of the washing steps to correct for the starting concentration of PDC in the gel during downstream analysis. Four combinations were tested: PAALVA(DOX)/PAALVA(GD), PAALVA(DOX)/P3, P1‐NAla(GD), P3, P1‐NAla(DOX)/ PAALVA(GD), and P3, P1‐NAla (DOX)/ P3, P1‐NAla (GD), and 30 µL gels were formed in triplicate for each condition. At t = 0, the buffer was replaced with a fresh buffer (10% ACN, controls, and blanks) or 0.2 µg mL⁻¹ of activated MMP ‐2 or ‐9 solution. At each predetermined time interval, 10 µL were removed from the gel supernatant and diluted 10x in a 10% ACN in buffer solution. All triplicates were analyzed using an 8060 triple quadrupole LC/MS‐MS (Shimadzu, Kyoto). Prior to running the experiment, internal standards were run on the instrument to generate standard curves for targeted analysis of the analytes (Figure S18). Standard curves were generated over a concentration range of 10–2000 ppb using the doxorubicin and geldanamycin peptide cleavage products, as well as pure doxorubicin.

The concentration in the supernatant at each time point was calculated using the standard curve, then converted to a percentage using the initial concentration. The triplicates for each group were averaged and normalized to the control values to account for drug release not due to enzyme activity. The data was fit to GraphPad Prism 9's built‐in nonlinear regression one‐phase decay model as described above.

4.14. In Vitro Release Experiments

MDA‐MB‐231 cells were seeded in 24‐well plates and allowed to attach overnight. The next day cells were treated with one of nine conditions, in triplicate, per time point: control cells (media), control gels (PEG‐acrylate gels, not loaded with any PDC), dual fast releasing gels, dual slow releasing gels, fast (DOX) /slow (GD) gels, fast (GD) /slow(DOX) gels, 2 x IC₅₀ of each drug in solution, DOX → DOX + GD in solution, and GD → DOX + GD in solution. PDC‐loaded PEG‐acrylate gels were made as previously described, but were gelled for 1 h in 24‐well transwell inserts. All gels were allowed to swell in PBS overnight at 37°C, then washed 3x with PBS prior to use in treatment. Gel‐containing inserts were placed in the wells such that the gels were submerged in cell media without any direct contact with the cells. A 3 µm pore size was selected to ensure facile diffusion of proteases and released drugs into the surrounding media. Gels were loaded with appropriate concentrations of PDC such that complete release would result in a concentration of PDC equal to 2 x the IC₅₀ of each drug. The DOX → DOX + GD, and GD → DOX + GD solution conditions were designed to mimic the behavior of gels with heterogeneous release kinetics, and involved the treatment of cells with 2 x IC₅₀ of the first drug for the first 3 days, then 2 x IC₅₀ of each drug from days 3 to 14. At days 1, 3, 5, 7, and 14, cells were lifted and analyzed via flow cytometry as described above. To prevent skewed data collection, for timepoints lasting longer than 3 days, cell supernatant containing any dead cells was collected, stained, and fixed using paraformaldehyde every 3 days,, then incorporated into the cell solution for analysis on the end date to ensure that all dead cells floating in solution were accounted for and not lost to natural degradation. Analysis of the dot plots in FlowJo was accompanied by plots representing the fraction of viable cells over time for each condition, which were fit to GraphPad Prism 9's built‐in nonlinear regression one‐phase decay model as described above (Figure 5, Figures S20–S21).

The procedure was repeated using A549 cells following the appropriate subculture procedures listed above. Data were collected at days 1, 4, and 7 (Figures S23–S24).

4.15. Statistical Analysis

Statistical difference analyses were assessed through t‐test and one‐way ANOVA by GraphPad Prism Version 9.0a software (GraphPad Software). Unless stated otherwise, the data are expressed as mean ± SD, and p < .05 was considered as statistical significance (^* p < .05, ^** p < .01, ^*** p < .001 and ^**** p < .0001).

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supporting File: adhm70691‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.