Antibody–drug conjugates (ADC) have revolutionized the field of cancer therapy. ADCs combine the high specificity of tumor-targeting monoclonal antibodies with potent cytotoxic drugs, which cannot be used alone because of their high toxicity. Till date, all ADCs have either targeted cell membrane proteins on tumors or the tumor vasculature and microenvironment. Here, we investigate ADCs of APOMAB (DAB4, or its chimeric derivative, chDAB4), which is a mAb targeting the La/SSB protein, which is only accessible for binding in dying or dead cancer cells. We show that DAB4-labeled dead cells are phagocytosed by macrophages, and that the apoptotic/necrotic areas within lung tumor xenografts are bound by DAB4 and are infiltrated with macrophages. We show that only DAB4–ADCs with a cleavable linker and diffusible drug are effective in two lung cancer models, particularly when given after chemotherapy. These results are consistent with other recent studies showing that direct internalization of ADCs by target cells is not essential for ADC activity because the linker can be cleaved extracellularly or through other mechanisms. Rather than targeting a tumor cell type specific antigen, DAB4–ADCs have the advantage of targeting a common trait in most solid tumors: an excess of post-apoptotic, necrotic cells either adjacent to hypoxic tumor regions or distributed more generally after cytotoxic therapy. Consequently, any antitumor effects are solely the result of bystander killing, either through internalization of the dead, ADC-bound tumor cells by macrophages, or extracellular cleavage of the ADC in the tumor microenvironment.
The recent clinical development of antibody–drug conjugates (ADC) has revolutionized the field of cancer treatment by coupling technologies that link the exquisite tumor-targeting specificity of mAbs with extremely potent cytotoxic drugs. Four ADCs have current FDA approval, with gemtuzumab ozogamicin (GO, MYLOTARG; Pfizer) being the first ADC to be approved after receiving accelerated approval in 2000 as a stand-alone treatment for patients ≥60 years of age with relapsed, CD33-positive acute myeloid leukemia. GO was voluntarily withdrawn from the market in 2010 before re-gaining FDA approval in late 2017 with an altered treatment schedule and age demographic. Ado-trastuzumab emtansine (KADCYLA; Roche/Genentech) is indicated for the treatment of patients with HER2-positive, metastatic breast cancer (1), and brentuximab vedotin (ADCETRIS; Seattle Genetics) for the treatment of relapsed or refractory CD30-expressing Hodgkin lymphoma (2) and anaplastic large cell lymphoma (3, 4). Finally, inotuzumab ozogamicin (BESPONSA; Pfizer) was recently approved for the treatment of adult patients with relapsed or refractory CD22-positive B-cell acute lymphoblastic leukemia. The clinical and commercial success of these ADCs has accelerated investment in the clinical evaluation of other ADCs comprising various combinations of mAbs and drugs (5).
An ADC consists of a tumor-targeting mAb coupled to a potent cytotoxin through a chemical linker. The linker dictates when the drug is released from the antibody and can be divided into two broad classes of cleavable and noncleavable linkers. Noncleavable linkers rely on internalization of the ADC by the target cell followed by catabolism of the antibody to release the active drug. Cleavable linkers include disulfide linkers, which are cleaved by reducing agents such as glutathione; dipeptide linkers, which are cleaved by specific proteases including cathepsin B (CTSB); and hydrazone linkers, which are pH dependent. When cleaved from the linker, and depending on the chemistry of the drug used, the cleaved drug can diffuse out away from the target cell and be taken up by and kill cells surrounding the target cell, a phenomenon called bystander killing (6).
APOMAB® (DAB4) is a mouse mAb, which targets the Lupus-associated (La)/Sjögren Syndrome-B (SSB) antigen, which is overexpressed in a variety of different cancers (7–11). La/SSB only becomes available for antibody binding in cells that have lost membrane integrity, particularly in apoptotic and necrotic cancer cells, making DAB4 a dead tumor cell–targeting mAb (7, 12–17). As La/SSB is highly conserved between mice and humans, DAB4 binds both mouse and human forms of La/SSB. By radiolabeling DAB4 with α- or β-emitting radionuclides, we have been able to specifically target syngraft and xenograft tumors in mice, particularly after prior treatment with chemotherapy, effectively irradiating the tumor from the inside out (7, 13, 17). These studies demonstrate that radiation cross-fire dose, the dose delivered to cells surrounding the antibody-bound target cells, can have potent antitumor effects.
Targeting dead tumor cells for treatment of surrounding, viable tumor cells is not a new concept. Epstein and colleagues (18) first used antibodies directed against single-stranded DNA/histone epitopes to identify necrotic areas in tumors, which technology has been termed tumor necrosis therapy (TNT). The Chinese State Food and Drug Administration approved the TNT3 antibody labeled with Iodine-131 (Vivatuxin) as a treatment for patients with advanced lung cancer (19, 20). Nevertheless, characteristics of the in vitro binding of the DAB4 mAb to dead tumor cells compared with a TNT mAb help to explain tumor accumulation of DAB4. In vitro, DAB4 preferentially binds to dead tumor cells, which die because of DNA-damaging antitumor treatment; this binding is saturable, rapid, high-avidity, irreversible, and involves protein–protein crosslinking of DAB4 with its target antigen, La/SSB, during tumor cell death (7, 12–16). TNT3 has been used preclinically as an effective vehicle for the delivery of immunostimulatory CpG to the tumor site in mouse xenograft tumor models (21) and as a fusion protein for tumor delivery of IFNγ (22). Similarly, targeting melanin, which only becomes available for antibody binding in dead tumor cells, with a radiolabeled antibody has been shown to be effective in preclinical cancer models (23, 24). However, it is not known if targeting apoptotic and necrotic cells with ADCs would be effective.
Although targeting necrotic tumor cells alone is not beneficial because these tumor cells are already dead, apoptotic and necrotic areas of cancers are associated with high levels of CTSB (25) as well as reducing agents (26), and thus the extracellular cleavage of ADCs would enable bystander killing of surrounding, viable tumor cells (6). Furthermore, the direct internalization of ADCs or small molecule–drug conjugates (SMDCs) by the target cell may not be essential for release of drugs with cleavable dipeptide (27–31) or disulfide linkers (32, 33). Here, we investigate whether DAB4–ADCs with different drug/linker combinations were tolerable and effective in controlling tumor growth in two preclinical cancer models when given alone and in combination with chemotherapy, and examine which linker and drug types were required for anticancer efficacy. In these proof-of-concept experiments, we demonstrate that single treatments of DAB4–ADCs are effective, more so when given after chemotherapy, but only if a cleavable linker and diffusible drug combination is used. Furthermore, these treatments were tolerable irrespective of prior chemotherapy.
Materials and Methods
Cell culture, antibodies, and free cytotoxins
Mouse macrophage lines RAW264.7 and J774A.1, mouse Lewis lung (LL2) tumor cells, purchased from CellBank, and human A549 lung cancer cells, purchased from ATCC, were cultured in RPMI1640 (Sigma-Aldrich) with 5% FCS (Bovogen Biologicals). DAB4 is a mouse IgG2a mAb, which is a subclone of the anti-La 3B9 hybridoma originated by Dr Michael Bachmann (34), which was selected on the basis of higher binding to a defined epitope of the La antigen. 3B9 was the gift of Prof. Tom Gordon, SA Pathology, Flinders Medical Centre, Adelaide, Australia. DAB4 and the isotype-specific control mAb (Sal5, targeting the irrelevant Salmonella antigen) were produced and purified as previously described (7). All cell lines were used for fewer than 3 months after receipt or resuscitation from cryopreservation, and no further cell authentication was performed. Cells were routinely checked for mycoplasma contamination using MycoAlert Mycoplasma Detection Kit (Lonza) and were negative. Chimeric DAB4 (chDAB4), resulting from the genetic fusion of the murine DAB4 F(ab)2 and human IgG1 Fc, was created at CSIRO Manufacturing, and subsequently produced using the CHO-XL99 system at the National Biologics Facility, Australian Institute for Bioengineering and Nanotechnology, University of Queensland (Queensland, Australia). Free drugs of monomethyl aurostatin E (MMAE) and Duocarmycin DM were purchased from Levena Biopharma and Spirogen provided SG3199 (35).
Flow cytometry and immunofluorescence phagocytosis analysis
LL2 or A549 cells were killed by treatment with 20 μg/mL cisplatin for 48 hours. The dead cells were collected and labeled with 0.5 μmol/L CellTrace carboxyfluorescein succinimidyl ester (CFSE; ThermoFisher Scientific) following the manufacturer's protocol. After washing, CFSE-labeled dead cells were added to CellTrace Far Red-labeled macrophage cells (RAW264.7 and J774A.1 cells) at varying ratios and incubated for 24 hours. Adherent macrophages were washed to remove any unbound dead cells, collected, and analyzed by flow cytometry using a BD Accuri flow cytometer. Macrophages were identified as being CellTrace Far Red+, and the phagocytosis index determined as the percentage of Far Red+ CFSE+ cells.
For fluorescence microscopy analysis, macrophages were grown on coverslips, labeled with CellTrace Far Red and varying concentrations of dead, CFSE+ tumor cells were added and incubated for 2 hours. Adherent macrophages were washed to remove any unbound dead cells and were fixed, stained with 0.5 μg/mL DAPI, imaged using a Leica TCS SP8 STED confocal microscope, and images were analyzed using Fiji software. In a separate experiment, unlabeled dead LL2 cells and A549 cells were incubated with 5 μg/mL chDAB4 labeled with FITC (ThermoFisher Scientific) following the manufacturer's instructions and cocultured with macrophages for 2 hours. Macrophage monolayers were washed, fixed with neutral buffered formalin, permeabilized using ice-cold methanol, blocked with 5% BSA, and incubated overnight with 2 μg/mL LAMP-1 rabbit polyclonal antibody (Abcam), followed by washing and incubation with 4 μg/mL Alexa Fluor 594 anti-rabbit IgG, counterstained with DAPI, and imaged using a Leica TCS SP8 STED confocal microscope.
Treatment of cancer cells with free drug
LL2 or A549 cells were seeded at 104 cells/well in 96-well plates. Drugs (MMAE, Duocarmycin DM, SG3199) were resuspended in DMSO and diluted to appropriate concentrations in cell culture medium, added to cells, and incubated for 3 days at 37°C with 5% CO2. The medium was removed and 0.4 μg/mL Thiazolyl Blue Tetrazolium Bromide (Sigma) added and incubated for 2 hours at 37°C with 5% CO2. The resulting formazan crystals were dissolved using isopropanol and absorbance measured at 570 nm. Absorbance readings were normalized and expressed as percentage change in cell survival.
96-well high-binding plates were coated overnight with 1 μg/mL of a peptide related to the La/SSB epitope published for the 3B9 mAb (36). The following day, plates were washed with PBS, blocked with 3% BSA, and incubated with different concentrations of antibody or ADC for 2 hours at 37°C. Following washing, plates were incubated with anti-mouse IgG alkaline phosphatase or antihuman Fc-specific alkaline phosphatase antibodies, washed and developed using SIGMAFAST p-nitrophenyl phosphate tablets following the manufacturer's instructions and absorbance read at 405 nm.
Drug conjugations of DAB4 and Sal5 were performed at Levena Biopharma. Conjugation of ADCs with the SPDB or SMCC linker was performed by targeting reactive lysines as described in Example 1 of US patent 6441163B1 (37), and conjugation of ADCs with the MC-VC-PAB linker was performed via reduction of interchain cysteine thiols for conjugation as described in Example 5 of US patent 7837980B2 (38). The conjugation of SG3249 (35) to chimeric DAB4 and the human IgG1 isotype control (B12, targeting the glycoprotein gp120 of HIV) was performed at Spirogen as previously described (39). Information for the ADCs used is shown in Supplementary Table S1 and the structure of MC-PEG4-VC-PAB-Duocarmycin DM is shown in Supplementary Fig. S1A. ADCs were aliquoted and frozen at −80°C until required.
All animal experiments were approved by the SA Pathology Animal Ethics Committee, Adelaide, Australia, and conducted following institutional ethical guidelines. Six- to 8-week-old female C57Bl/6 mice were inoculated subcutaneously in the right flank with 106 LL2 cells in PBS and 6- to 8-week-old female Balb/c nude mice were inoculated subcutaneously in the right flank with 5 × 106 A549 cells in 1:1 matrigel/PBS. Tumor size was measured using electronic calipers and tumor volume determined using the calculation: tumor volume = (a2 × b)/2, where a is the shortest diameter and b is the longest diameter of the tumor. Mice were randomly allocated to treatment groups when tumors reached approximately 40 to 60 mm3. Mice were monitored daily using a clinical record sheet and body weight and tumor volume measured at least three times per week. Mice were humanely euthanized when a clinical score of 5 was reached, weight loss was >15% (cf. day 1), or tumor volume was >600 mm3.
LL2 tumor-bearing mice were treated intravenously with 50 mg/kg gemcitabine (Hospira) on days 1 and 2 and 2.5 mg/kg cisplatin (Hospira) on day 1. A549 tumor-bearing mice were treated intraperitoneally with 150 mg/kg gemcitabine and 3 mg/kg cisplatin on Day 1. The day after chemotherapy, mice were treated with an intravenous injection of ADC. Mice that did not receive chemotherapy were treated with ADC on the same day.
Tumors from untreated mice or mice euthanized 24 hours after administration of 100 μg chDAB4 were removed and snap frozen in Tissue-Tek OCT cryoprotectant (Sakura FineTek). Five micrometer sections were fixed with 10% neutral-buffered formalin, washed with PBS, and blocked and permeabilized with 5% BSA and 0.3% Triton-X 100 (Sigma-Aldrich). Using In Situ Cell Death Detection Kit (Sigma-Aldrich) and following the manufacturer's protocol, dead cells were detected using the TUNEL method. Macrophages were detected by incubating for 1 hour at 37°C with 2 μg/mL CD11b-Alexa Fluor 488 (BD Biosciences) and 4 μg/mL antihuman Alexa Fluor 647 or antihuman Alexa Fluor 488 to detect chDAB4. Sections were washed, counterstained with 0.5 μg/mL 4′,6-diamidino-2-phenylindole (DAPI), and examined using an Olympus IX71 microscope with CellSens Standard (v1.6) software and analyzed using ImageJ (v1.45) software (NIH).
Statistical analyses were performed using GraphPad Prism (v7.0) software. Kd values from ELISA data were determined using nonlinear regression analysis. IC50 values were determined by log transformation and nonlinear regression analysis. Comparison of two groups was performed by two-way t test or intergroup comparisons made by two-way ANOVA. Data are shown as mean ± SEM. Kaplan–Meier median survival curves were compared using log-rank (Mantel–Cox) test. Statistical significance was reached when P < 0.05, with * representing P < 0.05 and ** representing P < 0.01.
Dead cells are phagocytosed by macrophages in vitro
To determine whether macrophages phagocytosed chemotherapy-induced dead tumor cells, LL2 or A549 lung tumor cells were treated with 20 μg/mL cisplatin for 48 hours, washed, labeled with CFSE, and co-cultured with macrophages (J774A.1 or RAW264.7). For fluorocytometric analyses, CellTrace Far Red+ macrophages were selected, and phagocytotic macrophages identified as being Far Red+ and CFSE+ (Fig. 1A). As the number of dead CFSE+ LL2 tumor cells (Fig. 1B) or dead CFSE+ A549 tumor cells (Fig. 1C) increased in coculture with macrophages, so did the percentage of CFSE+ macrophages, indicating that the macrophages were phagocytosing the dead CFSE+ cells. This was confirmed by fluorescence microscopy where macrophages had engulfed the dead cells, with punctate staining of CFSE-labeled dead cells within the cytoplasm of the macrophages cells observed (Fig. 1D).
After confirmation that chDAB4 bound to cisplatin-treated dead LL2 cells, with at least 85% of the dead cells being labeled with chDAB4 (Fig. 2A), we conjugated chDAB4 with FITC, incubated it with dead LL2 cells, and examined whether macrophages phagocytosed the chDAB4-labeled dead cells. chDAB4–FITC–labeled dead cells were detected within the cytoplasm of macrophages and colocalized with the LAMP-1 lysosomal marker (Fig. 2B), indicating internalization and lysosomal processing of chDAB4-labeled dead cells by macrophages in vitro. This process was specifically due to phagocytosis because when macrophages were incubated at 4°C, rather than at 37°C, internalization of chDAB4-labeled dead cells was not detectable. We also showed that tumor cells, which were killed with cisplatin, Duocarmycin DM, SG3199, or MMAE, are bound by chDAB4 and subsequently are phagocytosed by macrophages in vitro (Supplementary Fig. S1B–S1D).
chDAB4 binds within the necrotic areas of tumors containing macrophages
We have previously shown that DAB4 binds within the necrotic and apoptotic areas of solid tumors including LL2 tumors (7, 13, 17). We also saw a similar tumor uptake pattern of chDAB4 in A549 tumors, with chDAB4 taken up primarily within the TUNEL-positive necrotic/apoptotic regions of the tumors (Supplementary Fig. S2). These regions of the tumors were also densely populated with cells expressing the common macrophage marker, CD11b, which colocalized with chDAB4 uptake (Fig. 3). Furthermore, use of chemotherapy in the LL2 xenograft model also produced an increased number of tumor-associated macrophages (TAM) and increased intratumoral CTSB expression (Supplementary Fig. S3). These attributes of the tumor microenvironment may favor the use of the DAB4 mAb in ADC therapy, particularly in combination with chemotherapy.
Effect of free drugs on cell survival
Next, we examined how effective the drugs used in the ADCs were at decreasing the survival of the LL2 and A549 cells. MMAE, duocarmycin DM, and SG3199 were added as free drugs to cells and, after 3 days of treatment, the effect on cell survival was examined and IC50 values determined (Fig. 4A). For the LL2 cells, SG3199 was the most potent with an IC50 of 1 pmol/L, followed by Duocarmycin DM (142 pmol/L) and MMAE (687 pmol/L) compared with 1.5 μmol/L for cisplatin. Although A549 cells were more resistant, SG3199 was also the most potent inhibitor of A549 cell survival (IC50 16 pmol/L), followed by duocarmycin DM, MMAE, and cisplatin (IC50 399 pmol/L, 844 pmol/L, and 6.8 μmol/L respectively; Fig. 4A).
Efficacy of DAB4–ADC using tubulin-targeting drugs in lung cancer models
ADCs of DAB4 or chDAB4 were made using noncleavable (SMCC), disulfide (SPDB), or dipeptide (VC or VA) linkers with drugs varying in potency. Drug-to-antibody ratio (DAR), unconjugated antibody, and the percentage of aggregate for each preparation are shown in Supplementary Table S1. ELISA using La peptide was used to confirm the binding of each ADC to the target antigen (Fig. 4B).
We examined whether DAB4–ADCs would be effective in LL2 syngraft and A549 xenograft models when given as a single dose, either alone or after chemotherapy. In the LL2 tumor model, we first performed dose escalation studies with DAB4-MMAE and DAB4-Duocarmycin DM either given alone or with chemotherapy. ADCs were given as a single dose at 2.5, 5, and 10 mg/kg and both ADCs were well-tolerated either alone or combined with chemotherapy. The combination with chemotherapy had better antitumor responses and resulted in reduced tumor growth (Supplementary Fig. S4). There was no significant difference in survival for mice treated with 5 or 10 mg/kg ADC either alone or combined with chemotherapy, so a dose of 5 mg/kg was chosen for further studies.
Next, a comparison study was performed to compare the efficacy of DAB4–ADCs with the tubulin-targeting drug, MMAE, versus the DNA minor-groove binding drug, duocarmycin DM, alone or combined with chemotherapy in the LL2 (Fig. 5A) or A549 (Fig. 5B) tumor models. Control ADCs consisting of Sal5 conjugated with the identical linker/drug combinations showed no further benefit compared with untreated mice or mice treated with chemotherapy alone, nor did an ADC of DAB4 using the val-cit linker and the MMAF drug, which is less cell-permeant then MMAE. Both DAB4-MMAE and DAB4-Duocarmycin DM showed similar potency in vivo with combination of chemotherapy and ADC resulting in better treatment outcomes (Fig. 5), and significant increases in survival (Supplementary Fig. S5). The treatments were well tolerated, with no adverse clinical signs evident, and with only minor and reversible weight loss seen when chemotherapy alone or chemotherapy plus ADC was given (Fig. 5). We also compared DAB4–ADCs conjugated to DM1 or DM4 using the SMCC (noncleavable) or SPDB (cleavable) linkers, respectively, and found that a cleavable linker was required for antitumor activity (Supplementary Fig. S6).
Efficacy of chDAB4–ADC using a DNA-targeting drug in lung cancer models
Because the DNA minor-groove binding drug of the pyrrolobenzodiazepine (PBD) dimer class, SG3199, showed the highest potency in vitro, we next examined whether it would be more potent when combined with chDAB4 as an ADC (chDAB4-SG3249). SG3199 is the PBD dimer drug in the linker-payload tesirine (SG3249). We first performed a dose escalation study with chDAB4-SG3249 in the LL2 tumor model given alone or combined with chemotherapy. ADCs were given at 1, 3, and 5 mg/kg, with all doses being well-tolerated either alone or combined with chemotherapy. There was a dose-dependent increase in tumor control using chDAB4-SG3249, particularly when combined with chemotherapy (Supplementary Fig. S7). When compared with DAB4-DuoDM, chDAB4-SG3249 was more effective at slowing tumor growth when given alone at the same dose (5 mg/kg), which was more evident when combined with chemotherapy (Fig. 6A). Of the ten mice treated with chemotherapy and 5 mg/kg chDAB4-SG3249 in the two experiments, two mice were tumor-free up to 90 days after treatment commenced, with no evidence of metastasis. Three mice were euthanized 36, 48, and 52 days after treatment because of adverse clinical scores, which were associated with metastatic disease and which occurred before subcutaneous tumors reached the tumor size endpoint. Treatment of A549 tumor-bearing mice with chDAB4-SG3249 also resulted in a notable reduction in tumor growth, especially when combined with chemotherapy (Fig. 6B) and significantly increased survival (Supplementary Fig. S5), although no cures were observed. All treatments were well tolerated with only minimal and reversible treatment-induced weight loss (Fig. 6).
To assess the therapeutic efficacy of ADCs of APOMAB®, we compared DAB4 conjugated with different linkers and drugs to assess which combinations resulted in the greatest antitumor response in two lung cancer models: LL2 tumors hosted by syngeneic, immunocompetent mice and A549 tumors hosted by xenogeneic, immunodeficient mice. We established that a cleavable linker was a requirement for there to be any therapeutic effect with the ADCs examined in this study, and that the greater the potency of the drug, the greater the tumor response observed.
We, and others, have shown that La is overexpressed in malignancy (7–10, 12) and that DAB4 specifically targets tumors by preferentially binding to dead tumor cells among a mass of viable tumor cells. The reasons include overexpression of La by tumor cells, enhanced DAB4 binding after tumor cell death caused by DNA-damaging treatment, the relatively inefficient clearance of dead tumor cells, and protein–protein crosslinking in dead tumor cells of La antigen with its specific antibody (12). This makes DAB4 a useful tool for not only detecting the induction of tumor cell death after cancer treatment, but also as a vehicle for the targeted tumor delivery of radioisotopes or cytotoxins.
Because DAB4 cannot be internalized by the dead target cells, we examined whether the dead tumor cells are phagocytosed by macrophages in vitro. Indeed, dead tumor cells, and chDAB4-labeled dead cells, were phagocytosed by macrophages. Moreover, our in vitro data showing macrophage-dependent phagocytosis particularly of LL2 cells killed by the DNA-binding payloads suggest that free payload drug, which is released by chDAB4–ADC with these payloads, may generate dead tumor cells to which additional chDAB4-ADC molecules bind before a further round of phagocytosis, thus producing a self-amplifying effect. We also confirmed that the necrotic areas of the tumor, which DAB4 targets in vivo, were densely populated with macrophages, a common pathology of solid tumors (40). As macrophages are critical effectors in response to antibody-based cancer therapies (reviewed by ref. 41), we postulated that ADCs based on DAB4 would also be effective preclinically.
We identified antitumor activity of DAB4–ADCs, but only when a cleavable linker and diffusible drug combination was used, that is only ADCs capable of bystander killing were effective. This effect was potentiated further when chemotherapy was given before ADC treatment. The observed tumor responses reflect the activity of free drug, which we hypothesize is generated in greater amounts after chemotherapy both because of increased tumor cell death resulting in increased DAB4–ADC tumor uptake (7) and the post-chemotherapy tumor microenvironmental changes including increased CTSB expression and macrophage infiltration that together promote intratumoral processing of the ADC. Hence, given that the SG3199 cytotoxin is most potent as a free drug, we expect that the chDAB4-SG3249 ADC would have the greatest antitumor activity, and this was observed in both tumor models, with two of 10 LL2-tumor-bearing mice cured after chemotherapy. This response was greater than what we have seen for DAB4 radiolabeled with either β- or α-emitting radionuclides in the same model (7, 13, 17). Interestingly, only the isotype antibody ADC containing the SG3199 linker/drug showed some antitumor activity compared with untreated control mice. This control antibody contains the human IgG1 Fc domain, and Li and colleagues (42) have shown that TAMs can process nontargeting human IgG1 ADC through FcγR to reduce tumor growth.
It was initially thought that the internalization of an ADC was a prerequisite for the resulting cleavage of the linker or degradation of the antibody to release the free, active drug. Although this still remains true for non-cleavable linkers, it is now becoming clear that this may not be essential for ADCs with cleavable linkers. For example, poorly internalizing or non-internalizing ADCs with cleavable linkers are effective in vivo (31, 32, 43, 44). ADCs targeting the non-internalizing splice-isoforms of fibronectin or tenascin-C comprising either disulfide or dipeptide-cleavable linkers attached to drug have been shown not to internalize but yet are still effective in controlling tumor growth (30, 31, 33, 45, 46). Similarly, targeting carbonic anhydrase IX, which is not effectively internalized, using either antibody or SMDCs with a dipeptide linker was effective in preclinical cancer models (27). Therefore, in the tumor microenvironment, glutathione and other reductants can cleave disulfide linkers while proteases can cleave dipeptide linkers. It is also worth noting that dipeptide linkers are not solely cleaved by cathepsin B; other cathepsins, such as K, L, and S (47) and carboxylesterase 1C (48) can also cleave dipeptide linkers.
We found that cures with the PBD dimer-containing DAB4–ADC only occurred in immunocompetent rather than in immunodeficient tumor-bearing mice. PBD dimers are known to result in immunogenic cell death of tumor cells (49), so it is possible that an antitumor T-cell–mediated immune response contributed to tumor clearance in some immunocompetent mice, and this is a hypothesis we are currently testing. Furthermore, although these ADCs are not directly internalized by the target cell, they may be internalized indirectly and still undergo lysosomal processing (6). For example, TAMs can process cleavable ADCs through FcRγ-mediated interactions to produce free drug, resulting in antitumor activity (42). Indeed, we saw an abundance of TAMs within the necrotic core of these tumors, which is a common feature of solid tumors (40, 50), and one which may enable release of free drug because we show that antibody-bound dead cells are internalized by macrophages in vitro. Others have indicated that the release of reducing agents by dead tumor cells can cleave disulfide linkers (45), which may be one mechanism by which DAB4–ADC with disulfide linkers are cleaved independently of internalization. In future experiments, we will explore whether depletion of TAMs alters the treatment response to ADC to elucidate whether TAM processing of dead, ADC-bound tumor cells is a key mechanism of action.
In conclusion, here we show that DAB4 ADCs are effective are controlling tumor growth, particularly when given after chemotherapy, when a diffusible drug and cleavable linker are used. As this ADC cannot be cleaved by direct internalization, future studies aimed at enhancing conditional activation of this ADC in the tumor microenvironment may help further to increase its activity.
Disclosure of Potential Conflicts of Interest
M.P. Brown is an inventor of APOMAB-related patents owned by AusHealth Research, as agent on behalf of Royal Adelaide Hospital. P.H. van Berkel has Ownership Interest (including stock, patents, etc.) in ADC Therapeutics SA. No potential conflicts of interest were disclosed by the other authors.
Conception and design: A.H. Staudacher, D. Chin, M.P. Brown
Development of methodology: A.H. Staudacher, Y. Li, O. Dolezal, T.E. Adams, P.H. van Berkel, M.P. Brown
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.H. Staudacher, V. Liapis
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A.H. Staudacher, Y. Li, V. Liapis, P.H. van Berkel, M.P. Brown
Writing, review, and/or revision of the manuscript: A.H. Staudacher, Y. Li, V. Liapis, J.J. Cheng Hou, D. Chin, T.E. Adams, P.H. van Berkel, M.P. Brown
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A.H. Staudacher, J.J. Cheng Hou, M.P. Brown
Study supervision: A.H. Staudacher, M.P. Brown
This work was supported by National Health and Medical Research Council, Australia (Project Grant ID 1126304, awarded to A.H. Staudacher) and by in-kind contributions from ADC Therapeutics UK Ltd. We acknowledge support for the National Biologics Facility (AIBN and CSIRO) from the Australian Government's National Collaborative Research Infrastructure Scheme.
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.