Synthesis and Evaluation of a Monomethyl Auristatin E—Integrin αvβ6 Binding Peptide–Drug Conjugate for Tumor Targeted Drug Delivery

Many anticancer drugs exhibit high systemic off-target toxicities causing severe side effects. Peptide–drug conjugates (PDCs) that target tumor-specific receptors such as integrin αvβ6 are emerging as powerful tools to overcome these challenges. The development of an integrin αvβ6-selective PDC was achieved by combining the therapeutic efficacy of the cytotoxic drug monomethyl auristatin E with the selectivity of the αvβ6-binding peptide (αvβ6-BP) and with the ability of positron emission tomography (PET) imaging by copper-64. The [64Cu]PDC-1 was produced efficiently and in high purity. The PDC exhibited high human serum stability, integrin αvβ6-selective internalization, cell binding, and cytotoxicity. Integrin αvβ6-selective tumor accumulation of the [64Cu]PDC-1 was visualized with PET-imaging and corroborated by biodistribution, and [64Cu]PDC-1 showed promising in vivo pharmacokinetics. The [natCu]PDC-1 treatment resulted in prolonged survival of mice bearing αvβ6 (+) tumors (median survival: 77 days, vs αvβ6 (−) tumor group 49 days, and all other control groups 37 days).


■ INTRODUCTION
Many of the current cancer treatment options are non-targeted and lack selectivity, affecting both the cancer and normal tissue. 1 This uncontrolled killing of healthy cells results in high systemic off-target toxicity, severe side effects, and poor quality of life for patients. 1 To overcome these challenges several tumor-targeting strategies have been explored including antibody−drug conjugates (ADCs) and peptide−drug conjugates (PDCs). Since 2019 only 9 ADCs have been FDA approved, including brentuximab vedotin, enfortumab vedotin, and polatuzumab vedotin, which are conjugated to monomethyl auristatin E (MMAE), 1−6 while no PDC has yet gained regulatory approval. 7,8 Although ADCs have demonstrated great promise, several challenges remain, notably the controlled site-specific chemical conjugation of the drug to the antibody, which often leads to ADC instability, poor antibody target affinity, and purification challenges. 2−4 In addition, their large size can result in poor tumor penetration and long blood residence times, thereby further increasing systemic toxicity. 2−4 To overcome some of these limitations, peptides have been investigated as delivery vehicles for the delivery of cytotoxic agents. Peptides are relatively easily synthesized by solid-phase peptide synthesis (SPPS), can be prepared in large quantities, and are readily modified to fine-tune affinity, selectivity, stability, and pharmacokinetics. 9 The ease of modification makes them an ideal platform as a PDC, and their smaller size permits better tumor penetration and a shorter blood residence time which can reduce systemic toxicity.
Many tumor-specific cell surface receptors have been identified as therapeutic targets, among them the integrins which are a family of cell surface receptors that are involved in cell migration and invasion. 10,11 Recently, the integrin α v β 6 has garnered much attention as a target for both the detection as well as the treatment of cancers. The integrin α v β 6 is an epithelial-specific cell surface receptor with low-to-no expression on healthy adult epithelium, but is highly overexpressed in many cancers, including some of the most lethal malignancies such as pancreatic cancer. 12−14 Studies have shown that the integrin α v β 6 plays a key role in carcinogenesis, where it is involved in cellular invasion, migration, angiogenesis, and adhesion to the extracellular matrix. 15 Importantly, it has been identified as a prognostic indicator, with high expression level correlating to poor prognosis and overall survival for patients. 15 Consequently, our group has developed and extensively studied the integrin α v β 6 -binding peptide (α v β 6 -BP), a peptide with nanomolar affinity and highly selective binding to integrin α v β 6 . The fluorine-18-labeled α v β 6 -BP was translated into the clinic to detect tumors in patients with breast, colon, lung, and pancreas cancer. 16 The α v β 6 -BP rapidly binds to and is internalized into α v β 6 -expressing cells, 17−19 and therefore, we now propose to use it as a chaperone for the selective delivery of the highly potent cytotoxic agent MMAE.
Caspase-3/7 Activity Assay. The caspase-3/7 activity ( Figure 4) is a measure of programmed cell death, and it was shown to correlate with the WST-1 cell viability assay ( Figure  3). The treatment of cells with PDCs (NH 2 -PDC-1, [ nat Cu]PDC-1) showed an α v β 6 -dependent increase in caspase-3/7 activity: for the DX3puroβ6 (+) and DX3puro (−) pair it resulted in a >5 times higher activity at 24 h for the DX3puroβ6 cells, with no observed change for the DX3puro cells ( Figure 4A Figure 4D, red and blue, respectively), with levels remaining close to the untreated cells ( Figure 4D, yellow). The staurosporine (purple) or free MMAE (green) provided increased caspase-3/7 activity in all cell lines regardless of integrin α v β 6 expression, again showing the lack of integrin α v β 6 selectivity for these non-targeted agents. The peptides containing no MMAE, i.e., NH 2 -2 (gray) and [ nat Cu]2 (black), showed no effect on caspase-3/7 activity in all cell lines and were indistinguishable from untreated cells (yellow). Notably, a slight increase in caspase-3/7 activity was observed for the MIA PaCa-2 (−) cells when treated with the [ nat Cu]PDC-1 ( Figure 4D, blue), which was not entirely unexpected since it had shown some initial toxic effect at the highest concentration by WST-1, however, the treatment with the NH 2 -PDC-1 (red) resulted in no significant increase of caspase-3/7 activity at any time point ( Figure 4D).
PET Imaging and Biodistribution. [ 64 Cu]PDC-1 showed integrin α v β 6 -dependent targeting and accumulation with clear visualization of both the DX3puroβ6 (+) and BxPC-3 (+) tumors by positron emission tomography (PET) imaging, along with no observable uptake in the DX3puro (−) tumor ( Figure 5). The PET images further showed high uptake in the kidneys, and some uptake in the gastrointestinal tract (stomach, small and large intestines, Figure 5). The biodistribution of [ 64 Cu]PDC-1 confirmed the α v β 6 -selective tumor accumulation, with 4.46 ± 0.91% ID/g in the DX3puroβ6 (+) tumor at 4 h vs 0.56 ± 0.12% ID/g in the DX3puro (−) tumor (ratio = 8:1; Figure 6A, Table S3). The BxPC-3 (+) tumor also exhibited a similarly high accumulation (4.61 ± 1.44% ID/g at 4 h; Figure 6B and Table S4). Moderate tumor washout was observed at later time points for both tumor models; it did reach significance at 48 h for the DX3puroβ6 tumor (4.46 ± 0.91% ID/g at 4 h to 3.39 ± 0.56% and 2.53 ± 0.37% ID/g at 24 h and 48 h, respectively, 4 to 48 h: P = 0.0002). For the BxPC-3 tumor, the uptake went from 4.61 ± 1.44% ID/g at 4 h to 3.73 ± 0.44 and 2.93 ± 0.80% ID/g, at 24 and 48 h, respectively (4 to 48 h: P = 0.054; Figure  6). Uptake of [ 64 Cu]PDC-1 was successfully blocked by preadministration of DOTA-2 (205 nmol) 10 min prior to administration of [ 64 Cu]PDC-1, resulting in 87−91% reduced uptake in the α v β 6 (+) tumors down to the level of the DX3puro (−) tumor (0.42 ± 0.04% ID/g; vs DX3puroβ6: 0.39 ± 0.04% ID/g and BxPC-3: 0.61 ± 0.05% ID/g at 4 h post injection; p.i.), thus demonstrating integrin α v β 6 -selective targeting in vivo (Table S6, Figure S24 ). Clearance from the blood was rapid, resulting in α v β 6 (+) tumor/blood ratios of   (Table S5). [ 64 Cu]PDC-1 primarily cleared through the kidneys, from 50 to 64% ID/g at 4 h to ≤25% ID/ g at 48 h ( Figure 6). Some uptake was observed in the gastrointestinal tract ( Figure 6), with the stomach dropping from 9% ID/g at 4 h to ≤3% ID/g at 48 h, the large intestines from 6% ID/g at 4 h to 3% ID/g at 48 h, and the small intestines from 4 to 5% ID/g at 4 h to 1.5 %ID/g at 48 h with elimination in the fecal matter (12% ID/g at 4 h to 1.5% ID/g at 48 h). Accumulation in the liver remained steady between 1.5 and 2.2% ID/g at 4 to 48 h, and uptake in other organs such as muscle (≤0.9% ID/g) and pancreas (≤0.5% ID/g) was low at all time points ( Figure 6).  Figure 7A). The mean tumor volume at day 37 for the DX3puroβ6 (+) bearing mice treated with [ nat Cu]PDC-1 was significantly >2.75 times smaller than the equally treated DX3puro (−) tumors (P = 0.0099; Figure 7A); at the same time point the [ nat Cu]PDC-1-treated DX3puroβ6 (+) mean tumor volume was >4 times smaller than all other treatment groups (saline, [ nat Cu]2, MMAE). All mice in the groups treated with saline, non-drug bearing peptide [ nat Cu]2, or free, non-targeted MMAE had met an end point criterion (≥2 cm in any direction and/or tumor ulceration) by 56 days, 70 days, and 64 days from start of treatment, respectively, with all these groups having the same median survival of 37 days ( Figure 7B). The DX3puro (−) tumor bearing mice treated with [ nat Cu]PDC-1 had a median survival of 49 days, with all mice reaching an end point at 95 days, while those bearing DX3puroβ6 (+) tumors treated with [ nat Cu]PDC-1 had a median survival of 77 days, and a 20% survival at the end of the study (day 122, Figure 7B). No significant differences of the average body weight were observed between any of the groups, indicating no significant adverse events or high systemic toxicity from the [ nat Cu]PDC-1 ( Figure S25).

■ DISCUSSION
Most standard chemotherapies do not distinguish cancerous cells from healthy cells, leading to less than ideal therapeutic efficacy and high systemic off-target toxicity. Tumor-targeted drug delivery approaches, such as PDCs, can improve accumulation of the therapeutic in the diseased tissue, reduce damage to healthy tissues and minimize unwanted side-effects. PDCs have been developed for targeting a wide range of receptors, including integrins, 27−42 with a variety of cytotoxic agents including doxorubicin (Dox), paclitaxel (PXT), camptothecin (CPT), and MMAE. 1,7,27,43−45 One emerging therapeutic target in oncology is the integrin α v β 6 , a cell surface receptor highly overexpressed in a wide range of malignancies with little to no expression on normal tissue. 13−16 The integrin α v β 6 is present in approximately 90% of pancreatic cancers and nearly all cases of metastatic disease. 13−16 Pancreatic cancer  remains one of the most lethal malignancies worldwide with a 5 year survival of less than 10%, 46 in part due to limited treatment options. Surgery is the only cure and unfortunately less than 20% of patients are eligible for resection at the time of diagnosis due to the presence of metastasis. 13−16 A clear unmet need for more effective and targeted treatments exists. We previously demonstrated that the α v β 6 -BP identified both primary and metastatic disease in a range of carcinomas. 16 These data suggest that the development of an integrin α v β 6targeted PDC based on the α v β 6 -BP for selective delivery of highly cytotoxic agents like MMAE holds great promise.
MMAE inhibits tubulin assembly with cytotoxic activity in the picomolar range and is extremely lipophilic, preventing its use as a therapy due to high systemic toxicity. 3,4,47 Efforts to overcome these high systemic toxicities include linking peptides to MMAE via protease-cleavable linkers. The linker choice is important because it governs the successful release of the cytotoxic agent. If the linker is too stable, release of the cytotoxic agent will be hindered, providing poor efficacy, 3,48 and if the linker has low stability, non-specific release of the cytotoxic agent will occur, leading to increased systemic offtarget toxicities and ineffective treatment. 3,48 We chose the Mc-Val-Cit-PABC cleavable linker because it combines high stability in human plasma 49 with rapid hydrolysis by lysosomal enzymes such as cathepsin B, an enzyme that is upregulated in cancer cells, 20−23 resulting in the release of MMAE in its unaltered form. 21 Standard SPPS combined with a site-specific Michael addition enabled the robust synthesis of the α v β 6 -BPlinker-MMAE-conjugate (PDC-1), and radiolabeling with copper-64 yielded [ 64 Cu]PDC-1 which enabled the quantitative assessment of cell binding, internalization, and in vivo pharmacokinetics.
[ 64 Cu]PDC-1 demonstrated integrin α v β 6 receptor selective binding and internalization in vitro. [ nat Cu]PDC-1 also demonstrated integrin α v β 6 selective cytotoxicity; for example, the DX3puroβ6 cells, having the highest integrin α v β 6 expression, had an EC 50 = 0.058 ± 0.003 nM, the intermediate integrin α v β 6 -expressing BxPC-3 had an EC 50 = 65.1 ± 10.6 nM, the low expressing MIA PaCa-2 cells showed low cytotoxicity (EC 50 > 250 nM) and the non-expressing DX3puro cells exhibited no observable cytotoxic effects. In contrast, the free, non-targeted MMAE was highly cytotoxic to all cells, having an EC 50 of 0.14−0.5 nM. The in vitro efficacy of [ nat Cu]PDC-1 was comparable to the integrin α v β 6 -targeting PDC containing the cytotoxic drug tesirine (PDC, SG3299) that was previously reported to have an EC 50 = 4.19−5.37 nM in α v β 6 -expressing cells, including in the engineered melanoma cell line A375Pβ6 and the pancreatic Capan-1 (EC 50 = 4.19 ± 3.76 and 5.37 ± 5.23 nM, respectively). 42 The tesirine-PDC (SG3299), when compared to the non-targeting scrambled PDC, tesirine-PDC (SG3511), provided a 15:1 ratio for selective cytotoxicity toward A375Pβ6 (+) melanoma cells, but the targeting tesirine-PDC (SG3299) also had relatively high cytotoxicity to α v β 6 -null engineered melanoma cells A375Ppuro and Panc-1 cells (EC 50 = 30.6 ± 18.8 nM and 175.6 ± 115.7 nM, respectively). 42 By comparison, in the present study, NH 2 -PDC-1 was >31-fold and [ nat Cu]PDC-1 was >86-fold more cytotoxic toward the melanoma DX3pur-oβ6 (+) than the DX3puro (−) cells. Other integrin α v β 3 and α v β 5 targeting camptothecin (CPT) PDCs have shown less favorable in vitro efficacy of EC 50 = 0.16−27 μM, 34 with some integrin α v β 3 targeting α-amanitin-PDCs exhibiting nonselective cytotoxicity. 41 Piarulli et al. showed that MMAE-PDCs targeting integrin α v β 3 produced cytotoxicities with EC 50 = 11−400 nM, concluding they had a promising candidate for in vivo experiments to obtain evidence of accumulation at the tumor site. 40 Indeed, few studies show biodistribution data for the PDCs, with limited examples including tritium or iodine-125 radiolabeled PDCs; however, these have limitations for noninvasive imaging and tracking. 29 Building on the encouraging in vitro and in vivo data suggesting selective integrin α v β 6 targeting, the PDC-1 was further evaluated for therapeutic efficacy. To permit a direct side-by-side comparison, this was done with DX3puroβ6 (+) or DX3puro (−) tumor bearing mice. Treatment with [ nat Cu]PDC-1 suppressed DX3puroβ6 (+) tumor growth and prolonged median survival to 77 days, compared to 49 days for the DX3puro (−) tumor-bearing mice, and >2-fold longer than other treatment groups (saline, non-drug bearing peptide [ nat Cu]2, or free, non-targeted MMAE: median survival 37 days). The [ nat Cu]PDC-1 treated DX3puroβ6 tumor cohort had 20% remaining alive at the end of the study (122 days). Notably, the [ nat Cu]PDC-1 treatment did not cause adverse systemic side-effects when administered four times at 6 mg/kg (0.9 μmol/kg), as the mice maintained healthy body weight during the course of the study. This concentration corresponds to 0.64 mg/kg of free MMAE, i.e., close to the LD 50 for free MMAE of 1 mg/kg (1.4 μmol/kg), 51 thus highlighting the successful administration of a highly cytotoxic agent safely as part of a targeted PDC at concentrations that would be systemically toxic when administered alone.

■ CONCLUSION
We developed the [ 64 Cu]PDC-1 by combining the highly cytotoxic drug MMAE with the highly selective integrin α v β 6 -BP, with the goal to reduce off-target toxicity of the drug whilst retaining therapeutic efficacy. In vitro testing demonstrated integrin α v β 6 -dependent binding, internalization, and cytotoxicity with high stability in human serum at 37°C. PET/CT imaging of [ 64 Cu]PDC-1 showed integrin α v β 6 -selective tumor accumulation and visualization, and the biodistribution confirmed a favorable pharmacokinetic profile with rapid blood clearance and renal excretion. In vivo therapeutic efficacy studies displayed >2-fold improved overall survival of mice bearing DX3puroβ6 (α v β 6 +) tumors compared to the control groups. Different dosing regimens are currently under evaluation with the goal to develop a highly effective, integrin α v β 6 -targeted PDC therapeutic for a wide range of carcinomas.

■ EXPERIMENTAL SECTION
Reagent lists and commercial sources along with additional method details are described in the Supporting Information (S4−S36).
Analytical Methodology. Characterization of purity and stability were confirmed using an analytical C 12 -reverse-phase (RP) highpressure liquid chromatography (HPLC) column (Jupiter Proteo, 250 mm × 4.6 mm × 4 μm; Phenomenex, Torrance, CA) at a 1.5 mL/min flow rate. All reverse phase high performance liquid chromatography (RP-HPLC) was carried out on a Beckman Coulter Gold HPLC equipped with a 2 mL injection loop. RP-HPLC was monitored by UV detector at a wavelength of 220 nm; a serially connected γdetector was used to monitor radioactivity. The mobile phase was a gradient starting at 9% acetonitrile in water containing 0.05% trifluoroacetic acid (TFA; EMD, Merck Millipore, Burlington, MA) held for 2 min, followed by linear ramp up to 81% acetonitrile over 30 min (for a total run time of 32 min till reaching 81%, Table S1). Purification of peptides was done by semi-preparative RP HPLC (C 12 : Jupiter Proteo column, 250 mm × 10 mm × 10 μm, Phenomenex) at a flow rate of 3 mL/min using the same gradient solvent system. After HPLC purification all peptides were confirmed by analytical HPLC to be >95% pure, and identity was confirmed by mass spectrometry at the UC Davis Mass Spectrometry Facility using a MALDI-TOF spectrometer (UltraFlextreme; Bruker, Billerica, MA) in positive ionization mode with a sinapic acid matrix (Sigma-Aldrich).
Radiochemical Synthesis. DOTA-2 (5 μg, 0.0009 μmol) was dissolved in metal free water (10 μL) and added to a solution of , after which the media was removed; cells were washed twice with media (200 μL) and re-incubated in media (37°C, 5% CO 2 ) for 24 h. The media was then removed and the WST-1 reagent was added to each well, and the cells were incubated for 2 h at 37°C. The 96 well plates were read at 450 nm by a Multiscan Ascent microplate reader. The percent cell viability was normalized to untreated cells (set as 100% viability) for each cell line.
Caspase-3/7 Activity Assay. Caspase-3/7 activity was analyzed using an ApoTox-Glo Triplex Assay kit. Cells were seeded in a 96 well plate at the same density and using the same respective media as described for the WST-1 assay and incubated overnight (37°C, 5%  24 Untreated cells (media) were used as a measure of endogenous caspase-3/7 activity (normalized to 1). Cells were treated (n = 4/cell line/compound/time) for 24, 48, or 72 h (37°C , 5% CO 2 ) prior to washing. After treatment, the media was removed, cells were washed, and Caspase-Glo 3/7 reagent was added, and incubated for 1 h at room temperature. Caspase-3/7 activity was analyzed by measuring luminescence with a Fluoroskan FL microplate reader according to the manufacturer's protocol.
In Vivo Studies. All animal procedures conformed to the Animal Welfare Act and were approved by the University of California Davis Institutional Animal Care and Use Committee. All mice used for in vivo work were female athymic nude mice (6−8 weeks old) purchased from Charles River Laboratories (Wilmington, MA). For PET imaging and biodistribution studies, female athymic nude mice (6− 8 weeks old) were injected subcutaneously with 3 × 10 6 DX3puro and 3 × 10 6 DX3puroβ6 cells in serum free DMEM on the right and left flank, respectively, or with 5 × 10 6 25 The [ nat Cu]PDC-1 treatment groups consisted of n = 10/tumor model, while all other groups (saline, [ nat Cu]2, and MMAE) consisted of n = 4/tumor model. All groups received four doses (on days 0, 3, 6, and 9) via i.v. tail vein injection of the above dose dissolved in saline (100 μL). The mean weights and standard deviation of each group was 25.9 ± 1.2 g (saline), 25.6 ± 1.5 g ([ nat Cu]2), 25.8 ± 2.5 g (MMAE), 24.9 ± 1.9 g ([ nat Cu]PDC-1, DX3puroβ6 tumors), and 25.7 ± 2.0 g ([ nat Cu]PDC-1, DX3puro tumors) at day 0. Tumor volumes and body weights (to assess possible systemic toxicity) were measured starting on day 0, and once a week thereafter until the end of the study. Tumor volume (V) was determined according to the equation V = (π/6) × L × W × H, where L is the longest axis, W is the axis perpendicular to L, and H is perpendicular to the plane of L and W. End point determination criteria were: any axis >2 cm, active ulceration, or compromised health of the mouse (>20% loss of body weight from the start of the study). All data are represented as the mean ± SD and are plotted beginning at day 0. Survival curves were determined by Kaplan− Meier method.
Statistical Analysis. Quantitative data are reported as mean ± SD. Statistical significance was determined with paired two-tailed Student's t tests to give a significance value (P-value) at 95% confidence interval. A P-value of ≤0.05 was considered statistically significant.