Abstract
We have developed a highly active and well-tolerated camptothecin (CPT) drug-linker designed for antibody-mediated drug delivery in which the lead molecule consists of a 7-aminomethyl-10,11-methylenedioxy CPT (CPT1) derivative payload attached to a novel hydrophilic protease-cleavable valine–lysine–glycine tripeptide linker. A defined polyethylene glycol stretcher was included to improve the properties of the drug-linker, facilitating high antibody–drug conjugate (ADC) drug loading, while reducing the propensity for aggregation. A CPT1 ADC with 8 drug-linkers/mAb displayed a pharmacokinetic profile coincident with parental unconjugated antibody and had high serum stability. The ADCs were broadly active against cancer cells in vitro and in mouse xenograft models, giving tumor regressions and complete responses at low (≤3 mg/kg, single administration) doses. Pronounced activities were obtained in both solid and hematologic tumor models and in models of bystander killing activity and multidrug resistance. Payload release studies demonstrated that two CPTs, CPT1 and the corresponding glycine analog (CPT2), were released from a cAC10 ADC by tumor cells. An ADC containing this drug-linker was well tolerated in rats at 60 mg/kg, given weekly four times. Thus, ADCs comprised of this valine–lysine–glycine linker with CPT drug payloads have promise in targeted drug delivery.
Introduction
Over several decades, considerable research has focused on the use of monoclonal antibodies (mAb) directed against tumor-associated antigens for the delivery of cytotoxic agents to cancer cells (1–3). Significant progress has been made toward advancing antibody–drug conjugates (ADC) into treatment regimens for both hematologic and solid tumor indications (4–9), with nine that are approved (four within the last year), and many in early- and late-stage clinical development (10). The payloads used in approved ADCs are comprised of a limited set of drug classes. These include microtubule-disrupting drugs such as the auristatins and maytansines present within brentuximab vedotin (11), enfortumab vedotin (12), polatuzumab vedotin (13), belantamab mafodotin (14), and ado-trastuzumab emtansine (6); a DNA-damaging agent present within gemtuzumab ozogamicin (9) and inotuzumab ozogamicin (7); and topoisomerase 1 inhibitors within trastuzumab deruxtecan (15, 16) and sacituzumab govitecan (17). Enfortumab vedotin, polatuzumab vedotin, and trastuzumab deruxtecan all use protease-cleavable linkers, a linker class first validated by brentuximab vedotin (BV). Although existing ADCs have had considerable clinical impact across a broad array of malignancies, additional work is needed to further improve the profile of drug-linker combinations. ADC potency, immunologic specificity, stability, pharmacokinetic profile, and manufacturability are all potential areas for improvement. Our goal is to develop new technologies that can be applied for the generation of highly active ADCs with broad therapeutic indices. We have previously reported on ADCs comprised of camptothecin (CPT) analogs (18, 19). Here, we report on the development of a novel hydrophilic and protease-cleavable val–lys–gly tripeptide linker that may be used for potent CPT payloads attached to mAbs. ADCs conjugated to this drug-linker combination release active CPT derivatives through proteolytic cleavage with high antigen specificity. The ADCs are broadly active in in vitro and in vivo models, including models of multidrug resistance (MDR) and bystander killing activity. They display a broad preclinical therapeutic index, and the technology is well positioned for clinical development.
Materials and Methods
Synthesis
Solid-phase peptide synthesis of MP-PEG4-VK(Boc)G-OH
Unprotected glycine pre-loaded (1.1 mmol/g) on 2-chlorotryityl resin was purchased from BAChem. The resin (1 gram) was added to reaction vessel and washed with DMF 4 times and drained completely. The resin was swelled by shaking in DMF for 30 minutes, and drained. Using the general coupling procedure, Fmoc-Lys(Boc)-OH was coupled to the resin. The Fmoc group was deprotected using the general deprotection procedure described below. Using the general coupling procedure Fmoc-Val-OH was coupled to the resin, followed by the general Fmoc deprotection procedure. MP-PEG4-OH was coupled using the general coupling procedure. The resin was then washed with DCM 3 times, followed by MeOH 3 times, and placed under high vacuum overnight. The peptide was cleaved from the resin by stirring the resin in a solution of 1 mL acetic acid, 2 mL hexafluoroisopropanol, and 7 mL DCM for 1 hour. The resin was then filtered and rinsed with DCM 3 times, and then the solution was concentrated in vacuo. The resulting white powder was dissolved in 2:1 DMA:H2O (3 mL) and purified by preparative HPLC using a 30 × 250 mm Phenomenex Max-RP 4-μm Synergi 80Å reversed-phase column using a 5%–95% gradient elution of MeCN (0.05% TFA) in aqueous 0.05% TFA. Fractions containing the desired product were lyophilized to afford a white powder (354 mg, 0.442 mmol, 40%). Rt = 1.39 minutes, general method UPLC. MS (m/z) [M + H]+ calc. for C36H59N6O14 801.42, found 801.02 (Supplementary Fig. S2).
UPLC-MS analysis was performed on a Waters single quad detector mass spectrometer interfaced to a Waters Acquity UPLC system using a Waters CORTECS C18 1.6 μmol/L, 2.1 × 50 mm column. The general method was a linear gradient from 3% to 60% B over 1.70 minutes, followed by a linear gradient to 95% B at 2.00 minutes, with an isocratic hold at 95% B to 2.50 minutes, followed by a linear gradient to 3% B at 2.80 minutes and isocratic hold at 3% B to 3.00 minutes (flow rate: 0.6 mL/min, Mobile Phase A: 0.1% aqueous formic acid, Mobile Phase B 0.1% formic acid in acetonitrile).
General Fmoc deprotection procedure
A solution of 20% piperidine in DMF (10 mL) was added to the resin, shaken for 1 minute, and drained. Another 10 mL of 20% piperidine in DMF was added to the resin, shaken for 30 minutes, and drained. The resin was washed with DMF 4 times and drained completely.
General peptide coupling procedure
A solution was prepared in DMF (10 mL) of Fmoc Amino Acid (3 mmol), HATU (3 mmol), DIPEA (6 mmol). The solution was added to the resin and shaken for 60 minutes. The reaction vessel was drained and washed with DMF 4 times.
Synthesis of CPT-Lb
MP-PEG4-VK(Boc)G-OH (90.0 mg, 0.112 mmol) was dissolved in anhydrous DMF (0.3 mL) and DIPEA (0.05 mL, 0.30 mmol) was added. TSTU (67.6 mg, 0.224 mmol) was added to the reaction vessel, and conversion to the N-hydroxysuccinimide (OSu) activated ester was monitored by UPLC-MS. Complete conversion was observed after 5 minutes. The reaction was acidified with acetic acid (0.05 mL, 0.874 mmol), and purified by Biotage flash chromatography using a 10G Ultra silica gel column with a gradient elution of 0%–10% MeOH in DCM. Fractions containing the desired product were concentrated in vacuo to afford a white solid of MP-PEG4-VK(Boc)G-OSu (91.2 mg, 0.102 mmol, 90%). Rt = 1.48 minutes general method UPLC. MS (m/z) [M + H]+ calc. for C40H62N7O16 898.44, found 898.33.
A solution of 7-aminomethyl-10,11-methylenedioxy CPT (CPT1, 24 mg, 0.057 mmol) dissolved in anhydrous DMF (0.48 mL) was added directly to the reaction vessel with the MP-PEG4-VK(Boc)G-OSu (50 mg, 0.056 mmol). DIPEA (0.05 mL, 0.303 mmol) was added to the reaction vessel. Reaction progress was monitored for completion by UPLC-MS. Complete conversion to the desired coupled product was observed after 5 minutes. The mixture was acidified with acetic acid (0.05 mL, 0.87 mmol) and purified by filtration through a silica gel column with a gradient elution of 0%–10% MeOH in DCM. The eluent was concentrated in vacuo to afford a yellow solid which was the desired product MP-PEG4-VK(Boc)G-CPT1 (32 mg, 0.027 mmol, 48%). Rt = 1.59 minutes General Method UPLC. MS (m/z) [M + H]+ calc. for C58H77N9O19 1,204.54, found 1,204.25.
MP-PEG4-VK(Boc)G-CPT1 was dissolved in 20% TFA in DCM. The reaction was monitored for completion by UPLC-MS. Complete conversion was observed after 10 minutes. The reaction was concentrated in vacuo, reconstituted in 10% acetic acid in 2:1 DMA:H2O, and purified by preparative HPLC using a 21.2 × 250 mm Phenomenex Max-RP 4-μm Synergi 80Å reversed-phase column using a 5% to 95% gradient elution of MeCN (0.05% TFA) in aqueous 0.05% TFA. The fractions containing the product were lyophilized to afford CPT–Lb as a yellow powder (33 mg, 0.030 mmol, 80%). Rt = 1.12 minutes General Method UPLC. MS (m/z) [M + H]+ calc. for C53H69N9O17 1,104.49, found 1,104.70.
Modifications of these methods were also used for the synthesis of drug-linkers CPT–La, CPT–Lc, and CPT–Ld.
Conjugation
The method for ADC synthesis previously described (20) was used with several modifications. Briefly, a PBS, pH 7.4 solution of fully reduced mAb (10 mg/mL) was diluted 2-fold with propylene glycol (PG). Likewise, a 10 mmol/L DMSO stock solution of a CPT drug-linker (1.3 equiv./thiol) was diluted 2-fold with PG and the resulting drug-linker solution was added to the mAb solution in quarter-portions over a 5-minute period, resulting in drug-linker conjugation onto reduced interchain cysteine residues. After 15 minutes, excess drug-linker was removed by treating the mixture with activated charcoal (1 mg of charcoal to 1 mg of mAb) followed by vortexing for 30 seconds. The charcoal was then removed via filtration and the resulting ADC was buffer exchanged using a NAP5 or a PD10 size-exclusion column into 5% trehalose in pH 7.4 PBS. ADC drug-loading was determined by PLRP-MS (representative data for h00–CPT–Lc shown in Supplementary Fig. S1), and yield was determined using a LUNATIC instrument measuring the 280-nm absorbance and subtracting the 280-nm absorbance contribution from the CPT payloads (representative yield, 96% for h00–CPT–Lc ADC). Sequences for the cAC10, h1F6, and h00 mAbs used in this study are included in Supplementary Table S1.
Hydrophobic interaction chromatography
Samples of ADC (approximately 75 μg) were injected onto a Butyl hydrophobic interaction chromatography (HIC) NPR column (2.5 μm, 4.6 mm × 35 mm, Tosoh Bioscience, PN 14947) at 25°C and eluted with a 12-minute linear gradient from 0% to 100% B at a flow rate of 0.8 mL/minute (Mobile Phase A: 1.5 mol/L ammonium sulfate in 25 mmol/L potassium phosphate, pH 7; Mobile Phase B: 25 mmol/L potassium phosphate, pH 7, 25% isopropanol). A Waters Alliance HPLC system equipped with a multi-wavelength detector and Empower3 software were used to quantify ADC chromatography results.
Plasma stability
All ADC solutions were normalized to 2.5 mg/mL in PBS, and stock solutions in mouse plasma were prepared by adding the ADCs (50 μL) to mouse plasma (200 μL). Plasma samples in duplicate were incubated at 37°C for t = 0, 6, 24 hours and 3 and 7 days. At each timepoint, samples were retained and stored at −80°C for processing and analysis. To process the samples, a 50% slurry of IgSelect in PBS was prepared and 50 μL was added to a 3-μm filter plate under vacuum. The resin was washed (2 × 1 mL, PBS) and the stability samples (180 μL) were applied. The filter plate was shaken (1,200 rpm for 1 hour at 4°C) and plasma was removed under vacuum. The resin was washed with PBS (1 mL) and water (1 mL), with vacuum applied after each wash, and then the sample plate was centrifuged at 500 × g for 2 minutes over a Waters 350 μL collection plate. The ADCs were eluted from the resin by treatment with pH 3.0 glycine buffer (100 mmol/L, 50 μL), agitated at 500 rpm for 2 minutes at 4°C, and centrifuged at 500 × g for 3 minutes into a 350 μL 96-well plate, each well containing 10 μL of 1 mol/L Tris pH 7.4 buffer. The elution sequence was then repeated. The ADC concentration of each sample was determined by UV. The samples were treated with PNGase for 1 hour at 37°C to affect deglycosylation. Each sample was reduced by treatment with dithiolthreitol (12 μL, 100 mmol/L) followed by incubation for 15 minutes at 37°C. Samples (50 μL/injection) were analyzed by PLRP-MS to assess drug-linker composition of the light and heavy chains.
In vitro cytotoxicity
In vitro potency of CPT payloads and ADCs was assessed on multiple cancer cell lines, namely L540cy derivative of the Hodgkin lymphoma (HD) line L540 provided by Dr. Philip Thorpe, University of Texas Southwestern Medical Center, Dallas, TX, 2012), L428 (HD, DSMZ, 2013), Karpas299 [anaplastic large cell lymphomas (ALCL), DSMZ, 2015], DEL (ALCL, DSMZ, 2013), and DEL-BVR (BV-resistant; ref. 21; ALCL, 2013), 786-O [renal cell carcinoma (RCC), ATCC, 2015], UM-RC-3 (RCC, ATCC, 2018), ACHN (RCC, ATCC, 2013), Caki-1 (clear cell carcinoma, ATCC, 2013), Raji (Burkitt's lymphoma, ATCC, 2013). All cell lines were authenticated by STR profiling at IDEXX Bioresearch and cultured for no more than 2 months after resuscitation. Cells cultured in log-phase growth were seeded for 24 hours in 96-well plates containing 150 μL RPMI-1640 supplemented with 20% FBS. Serial dilutions of ADCs in cell culture media were prepared at 4x working concentrations, and 50 μL of each dilution was added to the 96-well plates. Following addition of test articles, cells were incubated for 4 days at 37°C, after which growth inhibition was assessed by the addition of CellTiter-Glo (Promega) and luminescence was measured on a plate reader. The IC50 value, determined in triplicate, is defined here as the concentration that results in 50% reduction in cell growth relative to untreated controls.
Mouse xenograft studies
All rodent animal studies, including mouse xenograft models, rat toxicology and rat PK experiments were conducted according to a written study protocol and approved by the Institutional Animal Care and Use Committee in a facility fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Efficacy experiments were conducted in the renal cell carcinoma 786-O, HD L540cy, HD L428 xenograft models, and in an ALCL Karpas 299/Karpas BVR bystander killing xenograft model. Tumor cells, as suspensions, were implanted subcutaneously in SCID or nude mice. Upon tumor engraftment, mice were randomized to study groups (5 mice per group) when the average tumor volume reached about 100 mm3. The ADC or vehicle controls were dosed once via intraperitoneal injection. Tumor volume was determined using the formula (L x W2)/2. Animals were euthanized when tumor volumes reached 750 mm3. Mice showing durable regressions were terminated after 10–12 weeks post implant.
Rat toxicology
A total of 12 naïve female Sprague-Dawley (Hsd:Spraque Dawley SD) rats (Envigo), 3 rats per group, were administered a weekly dose of either vehicle (1X PBS pH 7.4) or ADC at 10, 30, or 60 mg/kg IV. Body weights were measured on the day of dosing and weekly thereafter. Blood for routine clinical pathology evaluation was collected from the jugular or caudal vena cava under isoflurane anesthesia and analyzed using the Advia 2120i (Siemens). Hematology was evaluated on study days 8, 15, 22, and 29.
Pharmacokinetics
Total human IgG was detected in plasma using the Gyrolab platform (Gyros AB). Assay standards and quality control samples (QCs) were prepared using ADC diluted in pooled female Sprague-Dawley rat plasma. Standards, QCs, and study samples were diluted 10-fold into Rexxip buffer (Gyros AB). Briefly, a biotinylated murine anti-human IgG was captured onto streptavidin-coated beads within the Gyrolab Bioaffy CD. After being captured, human IgG was detected with an Alexa Fluor 647 (Thermo Fisher Scientific) labeled version of the same anti-human IgG. The fluorescence signal (in response units) was read at the 1% photomultiplier tube (PMT) setting. Unknown sample concentrations were determined by interpolating against a standard curve fit with a 5-parameter logistic function weighted by 1/y2 using the Gyrolab Evaluator Software (Version 3.4.0.24).
Cell monolayer permeability assay
The method was previously described (22). Briefly, MDCK II and MDR1-MDCK II cells (obtained from Piet Borst at the Netherlands Cancer Institute) were seeded onto polyethylene membranes in a 96-well plate at 2.3 × 105 cells/cm2 for 4–7 days until confluent cell monolayer formation. Test and reference compounds were diluted with transport buffer (HBSS with 10 mmol/L HEPES, pH 7.4) from stock solution to a concentration of 2 μmol/L and applied to the apical or basolateral side of the cell monolayer. Permeation of the test compounds from A to B direction or B to A direction was determined in duplicate over a 150-minute incubation at 37°C and 5% CO2 at saturated humidity without shaking. Test and reference compounds were quantified by LC/MS/MS analysis based on the peak area ratio of analyte/IS.
The apparent permeability coefficient Papp (cm/s) was calculated using the equation:
Where dCr/dt is the cumulative concentration of compound in the receiver chamber as a function of time (μmol/L/s); Vr is the solution volume in the receiver chamber (0.075 mL on the apical side, 0.25 mL on the basolateral side); A is the surface area for the transport, that is, 0.0804 cm2 for the area of the monolayer; C0 is the initial concentration in the donor chamber (μmol/L).
Cathepsin B release assay
To 1 μL of a 10 mmol/L DMSO drug-linker stock solution was added 1 μL 100 mmol/L PBS N-acetylcysteine. After 5 minutes, the mixture was diluted with 25 mmol/L sodium acetate/1 mmol/L EDTA buffer (pH 5.0) to 100 μL total to make a 0.1 mmol/L stock solution. A stock solution of bovine spleen cathepsin B was prepared by dissolving the lyophilized solid (1.2 mg, ca. 36% protein) in 1 mL of 25 mmol/L sodium acetate/1 mmol/L EDTA buffer (pH 5.0). For each assay, a portion of the enzyme solution (6 μL) was activated at room temperature with a solution of 30 mmol/L DTT/15 mmol/L EDTA (0.012 mL) for 15 minutes. The activated cathepsin B was then diluted with 0.40 mL of a 25 mmol/L sodium acetate/1 mmol/L EDTA buffer (pH 5.0, preincubated at 37°C). The 0.10 mL solution of a 0.1 mmol/L sodium acetate quenched drug-linker solution was added to the enzyme solution. The mixture was incubated at 37°C, and 10 μL aliquots were diluted in 50 μL cold MeOH at various time points. The MeOH diluents were centrifuged and the supernatant was injected on the UPLC-MS.
In vitro drug release and quantitation
The CD30-positive cell lines, Karpas299 and L540cy, were grown in suspension culture in RPMI-1640 supplemented with 20% FBS in a humidified environment of 5% CO2 at 37°C. Cell volumes were calculated on the basis of an average viable cell diameter determined with a Vi-Cell XR2.03 cell viability analyzer (Beckman Coulter). Cell volume calculations assumed a spherical cell with no correction made for nuclear volume. Samples of cells (5 × 105 cells/mL) in triplicate were washed into fresh media; typical culture volumes were 10 mL. ADC was added to each culture at a final concentration of 100 ng/mL. The cultures were incubated at 37°C in a humidified 5% CO2 atmosphere for 24 hours. A Vi-Cell counter was used to determine cell count, diameter, and circularity. A known volume of cells was harvested by centrifugation (750 × g, 3 minutes, 4°C). After centrifugation, a volume of supernatant was collected, taking care to not disrupt the pellet, and frozen at −80°C. Cells were washed with an equal volume of ice-cold PBS, re-pelleted (750 × g, 3 minutes, 4°C) and supernatant aspirated. This wash step was repeated, with the final pellet being stored in 100 μL of fresh PBS and frozen at −80°C.
Medium and cell pellets not treated with ADC were used to establish standard curves and were prepared following the same precipitation procedures. Eight-point standard curves were made using varying amounts of CPT1 and CPT2, and a constant internal standard.
Cell pellets for samples and standards were acidified with 5% formic acid in water, vortexed vigorously, and incubated on ice for 5 minutes. These were then precipitated with 4 volumes of ice-cold methanol containing 0.1% formic acid and an internal standard. Cell pellet samples and standards and were centrifuged at high speed to remove protein, and the supernatants were removed and transferred to a protein precipitation plate (Impact Protein Precipitation Plate; Phenomenex). The resulting eluate was evaporated under nitrogen.
Media samples and standards were acidified on ice with 1% formic acid in water with an internal standard and then vortexed. These samples were then precipitated with 10 volumes of ice-cold methanol, centrifuged to remove protein, and transferred to a plate to evaporate under nitrogen.
Cell pellet and media samples were reconstituted in 10% acetonitrile with 0.1% formic acid and examined by LC-MS/MS.
To derive a method for the quantitation of released drug in the experimental samples, the LC peak area for the CPT1 and CPT2 standards was divided by the peak area obtained for the internal standard. The resultant ratios were plotted as a function of the internal standard concentration for each data points. These data were fitted using a non-linear regression. The peak area ratios were converted to drug concentrations using the derived equation.
Results
Drug-linker design
For our drug-linker constructs, we used a highly cytotoxic CPT payload 7-aminomethyl-10,11-methylenedioxy analog (CPT1, Fig. 1; ref. 23) with cellular IC50 values in the low single digit nmol/L range (Table 1). The presence of a 7-aminomethyl substituent on CPT1 serves as a site for linker attachment and is amenable for investigating a variety of linker constructs. Our choice of linkers was driven by the goal to produce potent ADCs with high stability, minimal aggregation, facile payload release inside cancer cells, and an unperturbed pharmacokinetic (PK) profile relative to the unconjugated antibody. Peptide linkers were explored in the work described here, because they are known to facilitate all these desired characteristics and have been successfully used in clinically validated ADCs (24), with the mc–val–cit–PABC–MMAE (vedotin) drug-linker being a notable example (25). To accommodate drug hydrophobicity and the consequent potential for ADC aggregation, we used peptide sequences that were intrinsically hydrophilic, together with polyethylene glycol (PEG) stretcher units. Incorporation of short, defined PEG units into linker constructs has been shown to significantly reduce aggregation and improve the PK properties of ADCs (26–28). Our choice to investigate val–lys and PEG-containing drug-linkers of CPT1 was influenced by earlier investigations with another hydrophobic drug class comprised of a duocarmycin-related minor groove binder (26). In that study, we demonstrated that applying a PEGylated val–lys linker to such hydrophobic drugs led to minimal ADC aggregation. The impact of the PEG unit was striking, in that without the PEG unit the ADCs were fully aggregated, whereas incorporation of a short PEG stretcher between the maleimide unit and val–lys dipeptide gave ADCs with minimal aggregation. We anticipated that applying this strategy would enable us to achieve ADCs with high drug-to-antibody ratios (DAR), while mitigating aggregation, thereby optimizing potency and ADC homogeneity (27). To minimize the complexity of the drug-linker construct, we also chose to directly attach the PEG–peptide linkers to CPT1 and not rely on the use of a self-immolative spacer unit to facilitate payload release. It was envisioned that this strategy might increase the stringency for payload release. The drug-linkers based on potent CPT1 using hydrophilic and cleavable peptide-based linkers were designed for high in vitro and in vivo activities, low levels of aggregation and optimal biophysical properties.
Camptothecin payloads and linkers.
CPT payload (IC50 nmol/L) and ADC (IC50 ng/mL) in vitro potency on CD30-positive HD and ALCL cancer cell lines.
A series of four val–lys-containing drug-linkers were prepared, each using CPT1 as the core payload unit (Fig. 1, Supplementary Table S2). Drug-linkers CPT–La and CPT–Lb were comprised of val–lys, and val–lys–gly peptide sequences, respectively, and each incorporated a PEG unit (n = 4) attached to the N-terminal valine residue. CPT–Lc was analogous to CPT–Lb but used a slightly longer PEG (n = 8). Each drug-linker CPT–La through CPT–Lc was terminated in a maleimidopropionyl (MP) residue for antibody conjugation at reduced mAb hinge cysteine residues. A final drug-linker, CPT–Ld, comprised of a val–lys–gly sequence and terminated directly in a maleimideocaproyl (MC) spacer (no PEG stretcher), was prepared to assess the impact of the PEG stretcher on aggregation. The peptide linker components were prepared on resin using solid-phase synthesis, cleaved from resin, coupled to CPT1, and the lysine ε-amine BOC protecting groups were removed to yield drug-linkers CPT–La through CPT–Ld (Supplementary Fig. S2). Each drug-linker was conjugated to the fully reduced chimeric anti-CD30 antibody cAC10 for the preparation of nearly homogeneous ADCs (representative data, Supplementary Fig. S1). Drug-linkers CPT–La, CPT–Lb, and CPT–Lc, incorporating a PEG stretcher in their structures, afforded DAR8 ADCs with low levels of high molecular weight species (0.9%, 2.9%, 2.2%, respectively) as determined by size-exclusion chromatography (Fig. 2A). In contrast, antibody conjugation with the non-PEGylated drug-linker CPT–Ld resulted in significant levels of high molecular weight species (10%) even at a DAR of 2.7 (Supplementary Fig. S3). Albeit a multivariable comparison, these data demonstrate the favorable impact that PEG stretchers can have on ADC biophysical properties and these data are consistent with previous findings (26).
ADC pharmacokinetic and physiochemical measurements. A, Size-exclusion chromatography traces for CPT–Lb and CPT–Lc cAC10 ADCs. B, Hydrophobic interaction chromatography. cAC10, cAC10—CPT–Lb, and cAC10-DT were injected onto a Butyl HIC NPR column (2.5 μm, 4.6 mm × 3.5 cm, Tosoh Bioscience, PN 14947) at 25°C and eluted with a 12-minute linear gradient from 0% to 100% B at a flow rate of 0.8 mL/min (Mobile Phase A, 1.5 mol/L ammonium sulfate in 25 mmol/L potassium phosphate, pH 7; Mobile Phase B, 25 mmol/L potassium phosphate, pH 7, 25% isopropanol). C, Kupffer cell uptake assay. Kupffer cells were incubated with AlexaFluor 647-labeled ADCs at a concentration of 0.1 mg/mL in cell culture media for 24 hours. ADC uptake into cells determined by measuring mean fluorescent intensity (MFI), D, Pharmacokinetic profile of h00 mAb (non–cross-reactive), and h00–CPT–Lb in rats. Rats were injected with 1 mg/kg of parental h00 mAb or h00–CPT–Lb ADC. Samples from scheduled blood draws were obtained, and the h00 antibody and h00–CPT–Lb ADC were captured from plasma via a biotin-conjugated murine anti-human light chain kappa mAb and streptavidin-coated beads. The h00 antibody and ADC were quantified via ELISA using an AF647-anti–human kappa detection reagent.
In vitro evaluation
The anti-CD30 cAC10 and non-binding h00 DAR8 ADCs based on drug-linkers CPT–La, Lb and Lc were tested for cytotoxic activity on CD30-positive HD and ALCL lines (Table 1). For the high CD30-expressing lines, L540cy, Karpas299, DEL and DEL BVR (21), the potency for CPT–Lb and CPT–Lc cAC10 ADCs was in the low ng/mL range. Interestingly, the ADC based on the val–lys drug-linker CPT–La showed activity only on the DEL and DEL BVR lines and was inactive on the L540cy and Karpas299 cell lines. All three cAC10 ADCs were inactive on the L428 cell line, which compared with Karpas299, L540cy and DEL BVR lines, has relatively low CD30 antigen expression (70k CD30 copies/cell). Activity on the BV-resistant DEL BVR line was maintained with all three CPT1-based ADCs, despite the overexpression (8-fold increase relative to the DEL line) of the MDR p-glycoprotein (MDR1) induced by prolonged BV treatment; the BV payload MMAE is transported out of the cell by the MDR1 protein (29). All three CPT1 ADCs had high immunologic specificities, as demonstrated by low cytotoxic activity with the non-binding h00 ADCs against these cell lines.
The activities and immunologic specificities of DAR8 ADCs based on deruxtecan (DT) and govitecan (GT; Fig. 1) were also assessed. These drug-linker components are contained within the recently FDA-approved ADCs ENHERTU (15, 16) and TRODELVY (17). DT and GT were conjugated to both cAC10 and h00 using the same methods used for the new CPT drug-linkers described here. The L428 and the Karpas299 lines were insensitive to the cAC10-DT ADC. In the cell lines where the cAC10-DT ADC was active (L540cy, DEL and DEL/BVR), potency was lower than corresponding ADCs comprised of CPT–Lb and CPT–Lc. The DT ADCs were immunologically specific, with the non-binding h00-DT ADC showing no activity on L540cy, DEL, and DEL/BVR lines. In contrast, cAC10-GT and h00-GT ADCs were equipotent, devoid of immunologic specificity on all five cell lines tested. This profile is a consequence of the known instability of the GT ADCs (30) and suggests that targeted delivery of the SN-38 payload may not contribute to the in vitro activities observed.
To further test the in vitro activity, CPT–Lb was linked to the anti-CD70 mAb h1F6 (31), and the resulting ADC was highly active on cell lines with high CD70 antigen expression levels (Supplementary Table S3). Taken together, these data show that drug-linkers CPT–Lb and CPT–Lc provide ADCs that are immunologically specific and potent against a variety of cancer cell types, including lymphoma and solid tumor indications and cell lines overexpressing the MDR transporter.
ADC physiochemical properties
Hydrophilicity of CPT–Lb was evaluated by hydrophobic interaction chromatography (Fig. 2B; ref. 32). The HIC analysis compared cAC10–CPT–Lb, unconjugated cAC10, and cAC10-DT. cAC10–CPT–Lb eluted with a retention time between the cAC10 mAb and the cAC10-DT ADC, demonstrating reduced hydrophobicity of an ADC based on the CPT–Lb drug-linker relative to the DT drug-linker. Similarly, a Kupffer cell uptake assay (Fig. 2C), an in vitro proxy for non-specific liver uptake that can be a significant ADC clearance mechanism (33), demonstrated that uptake of the AlexaFluor 647-labeled cAC10–CPT–Lb was only marginally higher than that of the corresponding N-ethylmaleimide (NEM)–conjugated cAC10. Uptake of labeled cAC10–CPT–Lb was significantly lower than that of AlexaFluor 647-labeled cAC10-DT ADC. These data demonstrate that drug-linker CPT–Lb leads to the generation of ADCs with favorable biophysical characteristics, namely that they have a low propensity to aggregate (Fig. 2A) and have low hydrophobicity (Fig. 2B and C).
We also investigated ADC stability to proteolysis and retro-Michael deconjugation, as stability can impact activity and toxicity (34). The experiment was designed to test ADC stability under conditions where proteases may play a role. Similar studies were used to determine the stability of peptide-linked auristatin conjugates (35). Using an MP residue as the thiol-reactive functional group for both CPT–Lb and CPT–Lc, resulted in ADCs with high ADC stabilities. Incubation of h00–CPT–Lb and h00—CPT–Lc ADCs in BALB/c mouse plasma was followed by capture of the ADCs from plasma with IgSelect, deglycosylation with PNGase, and analysis by PLRP-MS. This led to the finding that 22% and 12% of drug-linkers were lost at the 24-hour timepoint from the two ADCs, respectively (Supplementary Fig. S4), with drug loss due solely to retro-Michael deconjugation. No proteolytic drug loss was observed by PLRP-MS. Beyond this timepoint, the ADCs were stable to further drug loss, consistent with previous findings with maleimidocaproyl-based conjugates (36). The ADC stability profiles for the CPT–Lb and CPT–Lc drug-linker constructs compared favorably with those containing DT and GT (Supplementary Fig. S4).
Finally, a rat PK study (Fig. 2D) with the non–cross-reactive h00–CPT–Lb ADC showed an exposure profile coincident with unconjugated h00 mAb. Considering the mouse plasma stability data, we conclude that conjugation to a high stability CPT1 drug-linker has minimal impact on ADC PK clearance.
In vivo evaluation
To begin our in vivo assessment, we compared cAC10 ADCs based on CPT–Lb, CPT–Lc and GT in the CD30-positive HD model L540cy. The cAC10 CPT–Lb and CPT–Lc ADCs cured the animals with a single 1 mg/kg dose (Fig. 3A). In contrast, the cAC10-GT ADC demonstrated only modest activities in the L540cy xenograft model when given as a single dose at 10 and 30 mg/kg (Fig. 3B). Previously, we have shown that even high stability drug-linkers based on SN-38 were not able to achieve cures in this model (19). This comparison with GT demonstrates the impact of using a highly potent CPT payload together with a stable and enzyme-cleavable linker.
In vivo xenograft studies. A and B (data from separate studies), Activity of cAC10 ADCs based on CPT–Lb, CPT–Lc, and GT drug-linkers in L540cy HD xenograft model. C, Comparison of cAC10–CPT–Lc and DT ADCs in CD30-low L428 HD model. D, Comparison of cAC10–CPT–Lc and cAC10-DT in Karpas 299 (CD30-positive)/Karpas BVR (CD30-negative) bystander killing activity model. All experiments were single-dose studies.
Because the activities of linkers CPT–Lb and CPT–Lc were equivalent in the L540cy model, we focused attention on the drug-linker CPT–Lc, which uses a longer PEG unit, for subsequent efficacy studies; direct comparisons were made with the DT drug-linker in two tumor models. Active dose ranges were determined empirically and based on experience with other ADCs. In the CD30 low L428 xenograft model, both the 1 and 3 mg/kg dose groups of cAC10–CPT–Lc were active and gave tumor regressions, whereas the corresponding cAC10–DT ADC (Fig. 3C) gave only minimal tumor growth delays. It is noteworthy that these ADCs were inactive against this cell line in a 96-hour in vitro assay, indicating a divergence in factors such as time and concentration exposure differences, the presence of effector cells in vivo, and differences in tumor morphology that can impact cell kill in vitro and in vivo. We also compared the cAC10–CPT–Lc and cAC10–DT ADCs in a bystander killing activity model designed to assess the ability of the ADCs to kill antigen negative tumor cells in proximity with antigen-positive cells. This model has been a described previously (37), and is comprised of a 1:1 mixture of Karpas 299 and Karpas BVR cells implanted subcutaneously into SCID mice (Fig. 3D). When the mixed tumor implants reached 100 mm3 in size, a single dose of each ADC was given at 3 and 10 mg/kg. The non-binding h00–CPT–Lc ADC was also given at 10 mg/kg. We found that the ADC based on CPT–Lc was >3-fold more active than the DT-based ADC in this model and that tumors in the h00–CPT–Lc non-binding control ADC group grew rapidly. These data demonstrate that CPT–Lc provides ADCs with immunologically specific bystander killing activity, which may be important in treating tumors with heterogeneous antigen expression (38).
Finally, potent antitumor activities were also observed in nude mice bearing MDR-positive 786-O renal cell carcinoma tumors (39, 40) when treated with the anti-CD70 ADCs h1F6–CPT–Lb and h1F6–CPT–Lc at 1 and 3 mg/kg (Supplementary Fig. S5) further demonstrating the potential for treating solid and MDR-positive tumors.
Payload release and characterization
To characterize drug release, the maleimide unit of val–lys–gly drug-linker CPT–Lb was quenched with N-acetyl cysteine (NAC), and treated with cathepsin B at pH 5.0. The enzyme was able to cleanly cleave the linker to afford the glycine adduct CPT2 as the only released CPT analog (Supplementary Fig. S6). In contrast, exposing the NAC-quenched val–lys drug-linker CPT–La to the same conditions showed no cleavage to CPT1 over a 19-hour period (Supplementary Fig. S7). The difference in enzyme cleavage profiles between these two linkers is noteworthy, considering the impaired activity of cAC10–CPT–La compared with cAC10–CPT–Lb ADCs in vitro (Table 1). Taken together, these data suggest that CPT–Lb may be a more facile substrate for enzymatic cleavage within lysosomes.
To further explore drug release, CD30-positive Karpas 299 and L540cy cells were treated with the cAC10–CPT–Lc ADC (100 ng/mL) for 24 hours, followed by extraction of released drug from pelleted cells and supernatant. LC-MS/MS analysis of the catabolites demonstrated that both CPT1 and CPT2 were released, with CPT2 being the predominant catabolite (Fig. 4A). This observation led us to explore the functional differences between these two CPT analogs. In a cytotoxicity assay against a panel of cancer cell lines (Table 1), CPT1 is significantly more potent than CPT2. The permeability coefficients for CPT1 and CPT2, determined by measuring the ability of each compound to cross a monolayer of Madin–Darby canine kidney cells, are 2.6 × 10−6 (moderate permeability) and 0.5 × 10−6 (low permeability), respectively (22). Thus, the difference in cytotoxic potency between the CPT1 and CPT2 analogs can be explained by differential membrane permeability. A topoisomerase 1 biochemical assay (23) that assesses the ability of each catabolite to trap the topoisomerase 1-DNA cleavage complex revealed that CPT1 and CPT2 have similar EC50 values (Fig. 4B). Consequently, both agents are expected to be highly cytotoxic when released intracellularly.
Characterization of CPT drug release and biochemical potency. A, CD30-positive Karpas 299 and L540cy cells were treated with the cAC10–CPT–Lc ADC (100 ng/mL) for 24 hours, followed by extraction of released drug from pelleted cells and supernatant, and drug quantitation by LC-MS/MS. B, Supercoiled DNA-topoisomerase 1 complex trapping with CPT payloads CPT1 and CPT2.
Tolerability
Tolerability of a non-targeted ADC based on CPT–Lb was assessed in Sprague-Dawley rats (3 animals/group). Using a non–cross-reactive ADC for tolerability assessment eliminates complications that arise from target-mediated effects. The h00–CPT–Lb ADC was evaluated with 4 weekly doses of 10, 30, and 60 mg/kg. All doses were well tolerated, and hematology changes were minimal in the 60 mg/kg group (Fig. 5A–C). Average platelet and reticulocyte values were consistently higher than vehicle-treated animals at each timepoint. Although the changes may indicate a response to the ADC, these values are within normal platelet and reticulocyte values. Weight loss in the animals was also minimal (Fig. 5D).
Rat toxicology results: hematology and weight. A, Absolute neutrophil (x103/μL). B, Absolute reticulocyte (cells/μL). C, Platelet counts (x103/μL) for q7dx4 treatment of h00–CPT–Lb at 60 mg/kg/dose, compared with vehicle (PBS) control animals. D, The percentage of animal weight loss for 10, 30, and 60 mg/kg/dose of h00–CPT–Lb, q7dx4, compared with vehicle control.
Discussion
CPTs intercept the covalent adduct formed between the enzyme topoisomerase 1 and DNA (41). The resulting ternary complex prevents DNA phosphate backbone re-ligation, resulting in double-strand breaks and ultimately in cell death. Irinotecan and topotecan are clinically approved for the treatment of solid tumors, most notably colorectal and ovarian carcinomas (42). Despite their clinical utilities, these therapeutics induce significant gastrointestinal and hematologic toxicities, including neutropenia and anemia. We previously reported work on ADCs with CPT payloads (18, 19). Applying ADC technology to a CPT payload can improve efficacy through the antigen-mediated delivery to the tumor, while reducing toxicity by minimizing systemic drug exposure (10). Recent clinical data with ADCs using CPT-based technologies, namely the HER2-targeted ADC ENHERTU (15, 16), and the TROP2-targeted ADC TRODELVY (43) have demonstrated strong antitumor activity in difficult to treat indications; both have recently gained FDA approval.
One context in which ADCs having CPT payloads may provide an activity advantage over other ADC technologies is in tumors that express p-glycoprotein, or the MDR1 protein. This transporter has been implicated in resistance to chemotherapeutics in the clinic (44), and the CPT drug class can overcome MDR1 (45). ADCs having CPT payloads, as demonstrated here, may be effective against tumors in which this mechanism has been induced by prior chemotherapy treatment, or when the transporter is upregulated.
ADCs based on CPT–Lb and CPT–Lc, each comprised of CPT1 and novel pegylated val–lys–gly linker sequences, are highly active in multiple tumor models, including models of MDR and bystander killing activity. The broad and potent antitumor activities in mouse xenograft models is complemented by high multidose tolerability in rat, framing a sizable therapeutic window. Design features within these constructs, including the use of a highly potent CPT payload and a high stability and hydrophilic protease-cleavable linker, likely contribute to the pronounced activities and tolerability profiles observed. One of the interesting findings surrounding cAC10–CPT–Lc is that both CPT1 and CPT2 are released within cells. Although these drugs are equally potent topoisomerase 1 inhibitors, they differ in membrane permeability. It is expected that CPT1 would be more active in the bystander setting, whereas CPT2, due to its impaired permeability, would be retained within tumors where it is generated. It is possible that these individual properties can be further optimized through novel drug-linker designs, and such studies are currently underway. A further finding was that incorporation of a short PEG sequence within the linker allowed for the generation of highly substituted and uniform ADCs with low aggregation levels. The control of overall ADC hydrophobicity minimizes non-specific normal cell uptake (33) and mitigates accelerated clearance (28).
The preclinical activity profile for ADCs based on CPT–Lc compares favorably with matched ADCs based on the DT and GT. Specifically, CPT–Lc ADCs have pronounced serum stability, immunological specificity, and demonstrate potent antitumor activities in mouse xenograft models, which contrasts with corresponding GT-based ADCs. Relative to DT, CPT–Lb and CPT–Lc ADCs have improved potency (both in vitro and in vivo) and improved bystander killing activity. By using a val–lys–gly linker, we have eliminated the need for a self-immolative unit to achieve facile drug release and the desired pharmacologic properties. Including a glycine residue into the cleavage sequence improved the activity profile when compared with the val–lys–CPT1 (CPT–La) construct. Differences between the val–lys and the val–lys–gly constructs were also observed in enzyme experiments. Together, these data suggest that facile enzymatic release plays a strong role in achieving ADCs with optimal activities.
BV (ADCETRIS), the anti-CD30 ADC based on the cAC10 mAb and the potent microtubule-disrupting agent MMAE is approved for use in frontline HD and peripheral and cutaneous T-cell lymphomas. As shown here, cAC10–CPT–Lc has a promising preclinical profile that may complement or even combine favorably with BV. Additional ADCs based on the CPT–Lc drug-linker are currently under evaluation in a variety of other cancer indications.
Authors' Disclosures
R.D. Lyski reports a patent for WO 2019195665 pending. D.W. Meyer reports a patent for 62/653961 pending to Seagen Inc. J.H. Cochran reports a patent for 62/653961 pending. N.M. Okeley reports personal fees from Seagen Inc. (employment) outside the submitted work. S.C. Jeffrey reports a patent for provisional patent application no. 62/653961 pending. No disclosures were reported by the other authors.
Authors' Contributions
R.D. Lyski: Conceptualization, investigation, project administration. L.B. Bou: Investigation, methodology. U.Y. Lau: Investigation. D.W. Meyer: Investigation, methodology, project administration. J.H. Cochran: Investigation, methodology. N.M. Okeley: Supervision, investigation, methodology. K.K. Emmerton: Investigation. F. Zapata: Investigation. J.K. Simmons: Investigation, writing–review and editing. E.S. Trueblood: Investigation. D.J. Ortiz: Investigation, methodology. M.C. Zaval: Investigation. K.M. Snead: Investigation. S. Jin: Investigation. L.M. Farr: Investigation, methodology. M.C. Ryan: Supervision, writing–original draft. P.D. Senter: Conceptualization, writing–review and editing. S.C. Jeffrey: Conceptualization, supervision, investigation, methodology, writing–original draft, writing–review and editing.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
Note: Supplementary data for this article are available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
Mol Cancer Ther 2021;20:329–39
- Received June 26, 2020.
- Revision received August 14, 2020.
- Accepted November 9, 2020.
- Published first December 3, 2020.
- ©2020 American Association for Cancer Research.
References
Volume 20, Issue 2
