Enhanced Therapeutic Activity of Non-Internalizing Small-Molecule-Drug Conjugates Targeting Carbonic Anhydrase IX in Combination with Targeted Interleukin-2

Cytotoxic drugs are frequently used for cancer chemotherapy (1), but the unfavorable tissue distribution properties of these molecules often prevent dose escalation to therapeutically active regimens (2). A suboptimal uptake of anticancer drugs in the neoplastic masses has been confirmed in cancer patients, using isotope-labeled preparations of drugs and positron emission tomography methodologies (3). This pharmacokinetic limitation of conventional chemotherapy makes it more difficult to induce objective responses in patients and contributes to side effects (4).

In order to overcome the limitations of conventional cytotoxic agents, it is attractive to couple potent drugs to monoclonal antibodies (5, 6) or to small organic ligands (2, 7, 8), which recognize a cognate antigen in the tumor environment and may therefore facilitate a preferential uptake at the site of disease. While it has long been thought that internalization of drug conjugates into the tumor cell may represent an indispensable requirement for potent anticancer activity (9), experimental evidence suggests that it may be possible to release cytotoxic payloads in the tumor extracellular space (10–15). Four antibody–drug conjugates (ADC) have gained marketing authorization (Kadcyla, Adcetris, Besponsa, and Mylotarg) for cancer therapy (16), while small-molecule-drug conjugates (SMDC) are still being investigated both at the preclinical and clinical levels (2, 7). SMDCs could offer certain advantages compared with ADCs, such as a more rapid and uniform diffusion into the tumor mass (17), lower cost of goods (18), and lack of immunogenicity (19).

Certain folate analogues (20), prostate-specific membrane antigen binders (21), somatostatin antagonists (22), and carbonic anhydrase IX ligands (11, 23, 24) have been considered as delivery vehicles for SMDC development activities, based on promising tumor-targeting performance in Nuclear Medicine studies (25, 26) and on encouraging preclinical results.

Carbonic anhydrase IX (CAIX) is a homodimeric membrane protein that is virtually undetectable in most normal adult tissues, exception made for certain gastrointestinal structures (27). However, CAIX is upregulated in hypoxic cells and is overexpressed in approximately 90% of clear cell renal cell carcinomas, as a result of von Hippel–Lindau inactivation (28). Interestingly, CAIX is also strongly expressed in a portion of patients with other malignancies, including colorectal, urothelial, lung, stomach, pancreas, breast, head and neck, ovaries, brain, and cervix cancer (29). Antibodies specific to CAIX efficiently localize to metastatic renal cell carcinoma lesions but not to normal gastrointestinal structures, as revealed by immuno-PET studies in cancer patients (30). Acetazolamide (an approved drug for the treatment of idiopathic intracranial hypertension, metabolic alkalosis, and glaucoma) exhibits a high binding affinity to CAIX (31). The development of charged derivatives of acetazolamide prevents the crossing of the cellular membrane and the binding to abundant intracellular carbonic anhydrase isoforms (11). Quantitative biodistribution studies performed in nude mice, bearing SKRC-52 human kidney cancer xenografts, revealed that ^99mTc-labeled charged derivatives of acetazolamide would achieve tumor to blood ratios of ∼100:1, six hours after intravenous administration (24, 32).

Our group has previously described the synthesis and potent antitumor activity of non-internalizing anti-CAIX SMDCs, based on acetazolamide conjugates to DM1, monomethyl auristatin E (MMAE), or the ultrapotent nemorubicin derivative PNU-159682, featuring the use of cleavable disulfide bridges or dipeptide linkers (11, 33). However, we have recently identified the 4,4-bis(4-hydroxyphenyl)valeric acid moiety as a structural feature, capable of increasing the binding affinity of acetazolamide derivatives to CAIX. This allowed the generation of a novel acetazolamide derivative with a higher affinity for CAIX (hereafter named AAZ⁺; ref. 23), which displays an improved tumor-targeting performance, compared with acetazolamide (24, 34). Encouraged by favorable biodistribution results, we here describe the synthesis and anticancer activity of a novel SMDC, based on the high-affinity AAZ⁺ CAIX ligand, a cleavable peptide linker and monomethyl auristatin E as cytotoxic payload. We observed that the new SMDC was well tolerated and potently active in preclinical models of cancer that are typically resistant to therapeutic agents which are commonly given to cancer patients. Durable complete cancer eradications could be observed in all treated mice when the new SMDC product was combined with L19-IL2, a clinical-stage fusion protein that delivers IL2 to the alternatively spliced EDB domain of fibronectin, typically expressed in the subendothelial extracellular matrix of tumor blood vessels (35–39). Collectively, our results provide a rationale for the clinical development of acetazolamide–drug conjugates for the treatment of patients with CAIX-positive malignancies, as well as for their combination with immunostimulatory drugs, such as L19-IL2.

Detailed synthetic procedures and characterization of the presented compounds are described in the Supplementary Information.

Peptidic precursors of compounds 1, 2, 3 (Fig. 1), 5, and 6, as well as compounds 4, 7, and 8 were synthesized by solid phase peptide synthesis (SPPS), using Fmoc-protected amino acids. Suitable resins (300–500 mg) were swollen with DMF (10 mL) for 15 minutes inside a syringe equipped with a filter pad. Fmoc deprotection was achieved by shaking the resin with 20% v/v piperidine in DMF (3 × 5 mL × 10 minutes). The appropriate Fmoc-protected amino acid (3 eq) was activated with HATU (3 eq) and DIPEA (6 eq) in DMF (5 mL) for 15 minutes at 0°C. After this time, the solution was allowed to react with the resin for 1 hour. Couplings and deprotections were alternated with DMF washing steps (4 × 10 mL), until the desired peptide sequence was obtained on solid phase.

Alkynes bearing acetazolamide or a free amide were then reacted by CuAAC “click” reaction with the corresponding peptide derivatives (carrying an azide moiety) on the resin. Side chain deprotection, cleavage from the resin, treatment with TCEP (30 eq) and RP-HPLC purification allowed to yield pure peptides, as confirmed by LC/MS analysis.

Commercially available MC-ValCit-PAB-MMAE (1 eq) was reacted with the corresponding precursor (4 eq) in degassed phosphate buffered saline (PBS; 50 mmol/L phosphate, 100 mmol/L NaCl, pH 7.4) added with DMF to help solubility. The reactions were stirred at room temperature until completion (monitored by LC/MS). The solvents were removed under vacuum and the final products were purified through RP-HPLC and lyophilized. The purity (>95%) of the final compounds was assessed by LC/MS.

The human renal cell carcinoma cell line SKRC-52 was kindly provided by Professor E. Oosterwijk (Radbound University Nijmegen Medical Centre, Nijmegen, the Netherlands). Upon thawing, cells were kept in culture for no longer than 10 passages. SKRC-52 were maintained in RPMI medium (Invitrogen) supplemented with fetal calf serum (10%, FCS; Invitrogen) and antibiotic–antimycotic (1%, AA; Invitrogen) at 37°C and 5% CO₂. For passaging, cells were detached using Trypsin-EDTA 0.05% (Invitrogen) when reaching 90% confluence and reseeded at a dilution of 1:6.

The murine colorectal carcinoma cell line CT26.wt (ATCC) was maintained in RPMI medium (Invitrogen) supplemented with FCS (10%; Invitrogen) and AA (1%; Invitrogen) and cultured at 37°C and 5% CO₂. For passaging, confluent cells were detached using Trypsin-EDTA 0.05% (Invitrogen) and reseeded at a dilution of 1:5. Transfected CT26 cells were kept in the same culture conditions as CT26.wt cells.

Surface plasmon resonance (SPR) experiments were performed on a BIAcore S200 instrument (GE Healthcare) at room temperature. For all measurements, CM5 chips (Series S) and filtered PBS pH 7.4 with DMSO (5% v/v) as flow buffer were used. Human CAIX was immobilized on the chip to 500 response units (RU) using EDC·HCl and NHS following manufacturer instructions. Serial dilutions of compounds 1 and 2 (Supplementary Fig. S3) in running buffer at a flow rate of 20 μL/min were used as analytes. After each cycle, the sensor surface was regenerated by a short treatment with DMSO (50% v/v) in PBS. Sensorgrams were solvent corrected, and the binding kinetics were analyzed with the BIAcore S200 evaluation software using a 1:1 Langmuir.

SKRC-52 cells were seeded in 96-well plates in RPMI added with 10% FCS (100 μL) at a density of 5 × 10³ cells per well and allowed to grow for 24 hours. The medium was replaced with medium containing different concentrations of test substance (1:3 dilution steps) and plates were incubated under standard culture conditions. After 72 hours, the medium was removed, MTS cell viability dye (20 μL, Promega) was added in 150 μL of fresh medium, the plates were incubated for 2 hours under standard culture conditions and the absorbance at 490 nm was measured on a Spectra Max Paradigm multimode plate reader (Molecular Devices; background correction was performed by measuring the absorbance at 630 nm). Experiments were performed in triplicates, and the average cell viability was calculated as measured background corrected absorbance divided by the absorbance of untreated control wells. IC₅₀ values were determined by fitting data to the four-parameter logistic equation, using a Prism 6 software (GraphPad Software) for data analysis.

For cellular expression analysis of human CAIX, cells were detached with 50 mmol/L EDTA and 5 × 10⁵ cells were stained with a human anti-CAIX–specific antibody (500 nmol/L, 1 hour, 4°C) in a volume of 100 μL FACS buffer (0.5% BSA, 2 mmol/L EDTA in PBS). For signal amplification, the anti-CAIX antibody was detected with an anti-human AlexaFluor647–labeled antibody (Invitrogen; 1:200, 45 minutes, 4°C). In-between and after staining, cells were washed with 100 μL FACS buffer and centrifuged at 1,100 rpm for 3 minutes. Stained cells were analyzed with a 2L-Cytoflex (Beckman-Coulter). SKRC-52 cells were used as positive control and CT26.wt cells as negative control. Results were analyzed with FlowJo 9 (FlowJo LLC).

QIFIKIT (Agilent) was used to quantify the copy number of human CAIX on SKRC-52 and CT26.3E10 cells. The cells were detached with 50 mmol/L EDTA, and 10⁶ cells were stained with a human anti-CAIX–specific antibody (500 nmol/L, 1 hour, 4°C) in a volume of 100 μL FACS buffer (0.5% BSA, 2 mmol/L EDTA in PBS). For signal amplification, the anti-CAIX antibody was detected with an anti-human AlexaFluor488–labeled antibody (Invitrogen; 1:200, 45 minutes, 4°C). In-between and after staining, cells were washed twice with 3 mL of FACS buffer and centrifuged at 1,100 rpm for 5 minutes. Setup and calibration beads present in the commercial kit were stained with an Alexa488 antibody (Invitrogen; 1:200, 45 minutes, 4°C). Cells and beads were analyzed with an LSR Fortessa (Becton Dickinson). Results were analyzed with FlowJo 9 (FlowJo LLC). Calibration curve was obtained plotting the fluorescent intensity with the number of antigen of the different bead populations, and CAIX expression determined based on the fluorescent intensity of stained SKRC-52 and CT26.3E10 cells (Supplementary Fig. S11).

The gene for human CAIX was cloned into the mammalian expression vector pcDNA3.1(+) (Fig. 5A) containing an antibiotic resistance for G418 Geneticin. A total of 6 × 10⁷ CT26.wt cells were transfected with 60 μg of pcDNA3.1(+) containing the human CAIX gene using the Amaxa 4D-Nucleofector (Lonza) with the SG Cell Line 4D-Nucleofector X Kit L (Lonza) and reseeded in complete growing medium. Three days after the transfection, the medium was replaced with RPMI (10% FCS, 1% AA) containing 0.5 mg/mL G418 (Merck) to select a stably transfected polyclonal cell line. To yield a monoclonal cell line, the stable cell line was stained as described for FACS analysis and single cell sorting was performed using a BD FACSAria III. Different clones were expanded and checked for antigen expression. Clone CT26.3E10 was selected for CAIX expression by FACS, immunofluorescence and confocal microscopic, and used for further in vivo experiments.

All animal experiments were conducted in accordance with Swiss animal welfare laws and regulations under the license number 27/2015 granted by the Veterinäramt des Kantons Zürich.

SKRC-52 cells were grown to 80% confluence and detached with Trypsin-EDTA 0.05% (Life Technologies). Cells were washed with Hank's Balanced Salt Solution (HBSS, pH 7.4) once, counted, and resuspended in HBSS to a final concentration of 3.4 × 10⁷ cells/ml. Aliquots of 5 × 10⁶ cells (150 μL of the suspension) were injected subcutaneously in the right flank of female athymic BALB/c nu/nu mice (8–10 weeks of age, Janvier).

CT26.3E10 cells were grown to 80% confluence and detached with Trypsin-EDTA 0.05% (Life Technologies). Cells were washed with HBSS (pH 7.4) once, counted, and resuspended in HBSS to a final concentration of 6.0 × 10⁷ cells/ml. Aliquots of 6 × 10⁶ cells (100 μL of the suspension) were injected subcutaneously in the right flank of female BALB/c mice (8–10 weeks of age, Janvier).

Female BALB/c mice bearing subcutaneous CT26.3E10 or CT26.wt tumors were injected intravenously with acetazolamide labeled with the near-infrared dye moiety IRDye680RD (AAZ-IRDye680RD; compound 6; 250 nmol/kg), dissolved in sterile PBS (100 μL). Mice were anesthetized with isoflurane and fluorescence images acquired on an IVIS Spectrum imaging system (Xenogen, exposure 1 second, binning factor 8, excitation at 675 nm, emission filter at 720 nm, f number 2, field of view 13.1). Images were taken before the injection and after 5 minutes, 1 hour, 3 hours, and 6 hours. Food and water were given ad libitum during that period. Mice were sacrificed after the last picture by cervical dislocation, and organs were extracted and imaged using the indicated parameters.

Radiolabeling procedures with technetium-99m were performed following the described procedures (24). Briefly, compounds 7, 8, and 9 (60 nmol/L) in TBS pH 7.4 (50 μL) were separately mixed with SnCl₂ (Sigma Aldrich, 200 μg) and sodium glucoheptonate (TCI, 20 mg) in H₂O (150 μL). Tris-buffered saline at pH 7.4 (600 μL) was added and the resulting solution degassed for 5 minutes by bubbling with nitrogen gas. The eluate from a ^99mTc generator (200 μL, ca. 200 MBq, Mallinckrodt) was added, and the reaction mixture heated to 90°C for 20 minutes. After cooling to room temperature, analytic aliquots were analyzed by RP-HPLC (XTerra C18, 5% MeCN in 0.1% aq. TFA to 80% over 20 minutes on a Merck-Hitachi D-7000 HPLC system equipped with a Raytest Gabi Star radiodetector) to determine radiolabelling yields. Technetium-99m incorporations >95% were routinely achieved.

SKRC-52 tumor cells were implanted into female BALB/c nu/nu mice (Janvier) as described above and allowed to grow for 3 weeks to an average volume of 250 mm³. Mice were randomized (n = 3 per group) and injected intravenously with radiolabelled preparations of compounds 7, 8, and 9 (1–3 MBq, 30 nmol/kg). Mice were sacrificed 6 hours after the injection by CO₂ asphyxiation and organs extracted, weighed and radioactivity measured with a Packard Cobra γ-counter. Values are expressed as %ID/g ± SD (Supplementary Fig. S1).

The tumor-targeting performance of L19-IL2 in SKRC-52 bearing nude mice was evaluated by quantitative biostribution analysis of the radioiodinated protein as described previously (40). Briefly, radioiodinated protein was injected into the lateral tail vein of mice (n = 3). The animals were sacrificed after 24 hours, organs were excised, and radioactivity was counted in Cobra gamma counter (Packard). Values are expressed as %ID/g ± SD (Supplementary Fig. S5).

SKRC-52 or CT26.3E10 tumor cells were implanted into female BALB/c mice (Janvier) as described above and allowed to grow to an average volume of 100 mm³. Mice were randomly assigned into therapy groups of 4 or 5 animals and treatment started by injecting a solution of AAZ-ValCit-MMAE (compound 1), AAZ⁺-ValCit-MMAE (compound 2), untargeted control (compound 3), L19-IL2, combination or vehicle (PBS containing 1% of DMSO) intravenously (lateral tail vein) at the doses and with the schedules indicated in the text and in Fig. 2, 3, and 6. Compounds 1, 2, and 3 were injected as solutions in sterile PBS containing 1% DMSO. L19-IL2 was injected as solution in appropriate sterile formulation buffer (Philogen). Animals were weighed, and tumor sizes were measured daily with an electronic caliper. The tumor volume was calculated according to the formula (long side) × (short side) × (short side) × 0.5. Animals were sacrificed when the termination criteria were reached. Prism 6 software (GraphPad Software) was used for data analysis (regular two-way ANOVA followed by Bonferroni test).

SKRC-52 tumors were excised, snap-frozen in OCT medium (Thermo Scientific) and stored at −80°C. Cryostat sections (7 μm) were cut, fixed with acetone, and stained with hematoxylin and eosin (H&E) using routine methods or subjected to immunofluorescence and immunohistochemistry. Primary staining for NK cells was performed with a rat anti-murine NKp46 IgG (NCR1; Biolegend). Primary staining for apoptosis was performed with a rabbit anti-caspase 3 IgG (C8487; Sigma Aldrich). Donkey anti-rat IgG-AlexaFluor594 and goat anti-rabbit IgG-AlexaFluor488 (Molecular Probes by Life Technologies) were then used for microscopic detection as secondary antibodies. Inflammatory cell infiltration of the tumors was assessed by immunostaining with a rat anti-CD45 monoclonal antibody (ab25386, Abcam; 1:2,000). The reaction was visualized using 3,3′-diaminobenzidine (DAB) as chromogen, followed by light counterstain with hematoxylin). Double immunohistochemistry was applied to frozen tumor sections to detect neutrophils (anti-myeloperoxidase rabbit polyclonal antibody, A0398, Dako-Agilent Technologies; 1:1,000) and macrophages (anti-CD68 rabbit polyclonal antibody, ab125212, Abcam; 1:200). The reaction was visualized using 3-amino-9-ethylcarbazole (AEC) and DAB as chromogens for neutrophils and macrophages, respectively, followed by hematoxylin counterstain.

Images were obtained using an Axioskop2 mot plus microscope (Zeiss) or the NDP.view2 software (Hamamatsu Photonics K.K., Japan), after the slides were scanned using a digital slide scanner (NanoZoomer-XR C12000, Hamamatsu Photonics K.K.).

For further evaluation of human CAIX expression, 10⁵ cells were cultured overnight on a sterile cover slip. Cells were fixed with ice-cold acetone, blocked with PBS (2% BSA), and stained with 5 μg/mL of a human antibody anti-CAIX in PBS (1% BSA). Detection was performed with an anti-human AlexaFluor488–labeled antibody (1:200 in PBS containing 1% BSA). The cytoskeleton was stained with Phalloidin-AlexaFluor647 (Thermo Fisher), and cell nuclei were stained with DAPI (MolecularProbes). Microscopic analysis was performed with an Axioskop 2 mot plus microscope (Zeiss; Supplementary Fig. S10).

To determine CAIX internalization after transfection, CT26.3E10 cells were seeded into 4-well cover slip chamber plates (Sarstedt) at a density of 10⁴ cells per well in RPMI medium (1 mL, Invitrogen) supplemented with 10% FCS, 1% AA and 10 mmol/L HEPES and allowed to grow for 24 hours under standard culture conditions. The medium was replaced with medium containing AAZ-Alexa594 (compound 5; 120 nmol/L; Supplementary Information), after 1 hour, Hoechst 33342 nuclear dye (Invitrogen) was added, and randomly selected colonies were imaged on a SP8 confocal microscope equipped with an AOBS device (Leica Microsystems).

The tumor-targeting performance of AAZ⁺, an affinity-matured version of acetazolamide discovered using encoded self-assembling chemical libraries and previously characterized by near-infrared fluorescence imaging (23), was confirmed in mice with subcutaneously grafted SKRC-52 tumors, using a radiolabeled and fluorescently labeled AAZ⁺ derivative (Supplementary Figs. S1 and S2). The product accumulated with 40% ID/g in the tumor mass 6 hours after intravenous administration (with a tumor to blood ratio of 80:1), while a similar compound featuring acetazolamide (AAZ) as CAIX binder exhibited a 18% ID/g at the same time point. Encouraged by these favorable biodistribution results, we synthesized a novel SMDC product, featuring the AAZ⁺ structure as tumor-homing moiety.

SMDC products were designed following the general ligand–spacer–linker–drug scheme depicted in Fig. 1. The CAIX-ligand acetazolamide was installed by copper-catalyzed azide-alkyne cycloaddition (“Click” reaction) on solid phase onto different peptides, which featured the presence or the absence of a 4,4-bis(4-hydroxyphenyl)valeric acid moiety, responsible for high-affinity CAIX binding (23). The resulting SMDCs 1 and 2 were subsequently cleaved and purified. In addition, compound 3 (which served as negative control) was synthesized in a similar way, replacing the acetazolamide moiety with an amide and thus generating a structure devoid of CAIX binding (Fig. 1). Reduction of a C-terminal cysteine residue allowed the coupling to a linker–payload, featuring a Val-Cit cleavable linker, a self-immolating para-aminobenzyl carbamate and the potent anti-tubulin monomethyl auristatin E (MMAE), which is also used in Adcetris. All products were obtained as highly pure (>95% purity by UPLC) lyophilized material (Supplementary Information), characterized for the binding to the cognate antigen (Supplementary Fig. S3) and used for the subsequent biological assays.

The cytotoxic activity of compound AAZ⁺-ValCit-MMAE 2 was tested on SKRC-52 cells and compared with the antiproliferative properties of free MMAE, used as reference compound (Supplementary Fig. S4). As expected for acetazolamide derivatives of MMAE, compound 2 acted as a prodrug and exhibited a substantially higher IC₅₀ value (440 nmol/L) compared with MMAE (IC₅₀ = 0.9 nmol/L). This experimental observation is in line with what had previously been published for several non-internalizing SMDC derivatives bearing the same linker–payload structure (33).

We compared the anticancer activity of compounds AAZ-ValCit-MMAE (1) and AAZ⁺-ValCit-MMAE (2) in BALB/c nude mice, bearing subcutaneously grafted SKRC-52 renal cell carcinomas. Compound 3, devoid of the CAIX-targeting moiety but otherwise identical to compound 2, served as negative control. The SMDCs were administered at the maximum tolerated dose, which had been found to be 250 nmol/kg in preliminary experiments (Supplementary Fig. S5), following the schedule depicted in Fig. 2. A fourth group of mice (“presaturation” group) was injected with a 50-fold higher dose (12.5 μmol/kg) of ligand AAZ⁺, which was directly followed by the administration of AAZ⁺-ValCit-MMAE (2). Both compounds 1 and 2 showed a potent anticancer activity (P < 0.0001 compared with the group of mice treated with vehicle), while negative control compound 3 did not display a difference from the vehicle-treated group. Presaturation with AAZ⁺ partially reduced therapeutic activity. Compound 2 was very well tolerated (Fig. 2B) and exhibited the most potent tumor growth retardation effect, but none of the mice enjoyed a durable complete response.

In a second therapy experiment performed in the same cancer model, compound 2 (AAZ⁺-ValCit-MMAE) was tested in combination with the clinical-stage antibody–cytokine fusion protein L19-IL2 (refs. 36–39; Fig. 3). A selective uptake of L19-IL2 at the tumor site had been confirmed in the SKRC-52 model, using biodistribution studies and a radioiodinated protein preparation (Supplementary Fig. S6). While neither L19-IL2 nor compound 2 was able to eradicate SKRC-52 xenografts when used as single agent, the combination of the two agents was well tolerated and induced complete remissions in all mice (Fig. 3; Supplementary Fig. S7). Cured mice were rechallenged 17 days after the therapy by a second subcutaneous injection of cancer cells and tumors regrew in all cases, as expected considering the lack of functional T cells in nude mice. Macroscopically, the combination of SMDC with L19-IL2 induced flattening of the tumor and hematoma formation at the site previously occupied by the solid tumor. Histologically, xenografts from mice that had received the vehicle or the monotherapy were composed of coalescing sheets and nests of densely packed neoplastic epithelial cells, separated by inconspicuous fibrous stroma and mild infiltration of inflammatory cells (Fig. 4; Supplementary Fig. S8). Tumors from mice exposed to the combination therapy exhibited instead a marked increase in the number of CD45-positive inflammatory cells infiltrating the xenograft. Most inflammatory cells were p46-positive NK cells (Fig. 4; Supplementary Fig. S9). Moreover, combination treatment led to a potent and homogeneous induction of apoptosis within the tumor mass, as revealed by cleaved caspase-3 staining (Fig. 4). Infiltration by myeloperoxidase-positive neutrophils and CD68-positive macrophages was low in all groups and restricted to the periphery of the xenografts (Fig. 4).

Approximately 30% of patients with colorectal cancer express high levels of CAIX in their tumor, but the immunocompetent mouse models of the disease tested by our group were negative for the antigen (see for example FACS analysis on CT26.wt murine cancer cells included in Fig. 5B; similar results were obtained for C51.wt and for RENCA tumor cells). In order to establish an immunocompetent murine model of colorectal cancer expressing CAIX, we stably transfected colorectal CT26.wt cancer cells with a gene coding for human CAIX and a second gene conferring antibiotic resistance against Neomycin (Fig. 5A). After antibiotic selection and single-cell sorting, monoclonal cell lines were tested for CAIX expression. Clone CT26.3E10 (but not CT26.wt) showed a shift in fluorescence intensity in FACS analysis using an anti-CAIX monoclonal antibody. The FACS shift was similar to the one observed for SKRC-52 cells (Fig. 5B). Fluorescence microscopy experiments confirmed CAIX expression on the membrane of CT26.3E10 and SKRC-52 cells, but not in wild-type CT26 cells (Fig. 5C; Supplementary Fig. S10). CAIX quantification by FACS analysis indicates that CT26.3E10 has a slightly lower surface density of CAIX, compared with SKRC-52 cells (Supplementary Fig. S11). Furthermore, mice bearing CT26.3E10 tumors showed a preferential uptake of the near-infrared fluorophore conjugate AAZ-IRdye680RD in the neoplastic mass 6 hours after intravenous injection, compared with similar experiments performed in mice bearing CT26.wt tumors (Fig. 5D).

Therapy experiments were then performed using compound 2, alone or in combination with the immunocytokine L19-IL2, in immunocompetent BALB/c mice bearing subcutaneous CT26.3E10 tumors. Also in this case, the antitumor effect was strongest when both therapeutic agents were used in combination (P < 0.0001), leading to durable complete responses in 100% of treated mice (Fig. 6; Supplementary Fig. S12). Thirty-seven days after complete tumor eradication, cured mice were rechallenged by a second subcutaneous injection of CT26.3E10 cancer cells and tumors regrew in all cases. The observation that the combination treatment was not able to confer protective immunity to the treated animals is in keeping with the predominant role played by NK cells, following L19-IL2 accumulation in the neoplastic mass (35).

An acetazolamide derivative (named AAZ⁺), which recognizes CAIX with dissociation constant in the subnanomolar range, was coupled to the linker–payload combination of Adcetris (an approved ADC product), featuring the valine-citrulline dipeptide, a self-immolative linker and the anti-tubulin agent MMAE. The new SMDC product mediated a potent tumor growth inhibition activity, but was not able to induce durable complete remissions, when used as single agent in tumor-bearing mice. However, the same product, used in combination with L19-IL2, mediated complete cancer eradication in all nude mice with xenografted SKRC-52 renal cell carcinomas, as well as in all immunocompetent mice bearing a syngeneic CT26 colorectal cancer, transfected with human CAIX.

L19-IL2 is a fully human antibody–cytokine fusion protein (35), which has been extensively studied in preclinical models of cancer and in clinical trials in patients with various types of malignancies (36, 39). Encouraged by the results observed in phase II studies (37, 38), the product is currently being investigated in a phase III clinical trial as combination agent for the treatment of resectable stage IIIB and C melanoma (NCT02938299), which is authorized in the United States, Germany, Italy, and Poland.

From a mechanistic viewpoint, the antibody-based delivery of IL2 to the tumor environment promotes the influx of leukocytes into the neoplastic mass (35, 41, 42) and an increase in vascular permeability at the tumor site (41, 43), favoring a preferential uptake of other therapeutic agents used in combination. The observation that L19-IL2 displays a similar activity in nude mice and in immunocompetent mice bearing the same murine tumor suggests that, at least in mice, the product mediates an anticancer effect by inducing the migration, proliferation, and activation of NK cells (35, 42). The use of tumor-targeting antibody fusions as an avenue to increase the therapeutic index of IL2 has also been described in reports published by other groups (44, 45).

The targeted delivery of IL2 to the subendothelial extracellular matrix in solid tumors and in certain hematologic malignancies has been shown to synergize with other therapeutic modalities, such as the use of chemotherapy (41), radiotherapy (46), intact antibodies capable of triggering an antibody–dependent cell cytotoxicity (ADCC; ref. 42), as well as various forms of armed antibody derivatives. A synergistic therapeutic activity has been reported for the combination of IL2-based immunocytokines with other antibody–cytokine fusion proteins (e.g., those based on TNF or on IL12; refs. 47, 48) and for ADC products (49).

CAIX has initially been assumed to be an internalizing antigen, suitable for the intracellular delivery and release of potent cytotoxic payloads coupled to CAIX-specific antibodies (50). However, subsequent studies have clearly shown that CAIX does not efficiently internalize, at least when bound to small molecule ligands (11). The targeted delivery of linker–payload combinations to the extracellular environment is attractive, as it may promote an efficient drug release process, triggered by high local concentration of glutathione (released by dying tumor cells; refs. 11, 12) or by proteases (10, 33). Potent cytotoxic drugs may diffuse into the tumor mass, thus potentially acting also on neighboring tumor cells, endothelial cells, and/or tumor-resident leukocytes.

The combined use of L19-IL2 with the SMDC product 2 (AAZ⁺-ValCit-MMAE) leads to cancer eradication in two mouse models, which do not respond to conventional chemotherapy or to other experimental therapeutic strategies (11). Because the corresponding targets (i.e., the alternatively spliced EDB domain of fibronectin and CAIX) are strongly expressed in the majority of patients with clear cell renal cell carcinoma and in various types of other malignancies (including colorectal cancer), the preclinical results presented in this study provide a rationale for the clinical investigation of this combination modality. A companion diagnostic, featuring the use of acetazolamide derivatives labeled with technetium-99m, has exhibited promising results in biodistribution studies and has recently received authorization to start clinical studies, with SPECT imaging of cancer patients (EUDRACT no. 2016-004909-13). The study will be important, in order to elucidate whether similar tumor:organ and tumor:blood ratios can be observed both in mice and in patients. Ideally, the SMDC + L19-IL2 combination should be administered to those subjects, who show an adequate uptake of acetazolamide derivatives in the neoplastic masses, using Nuclear Medicine procedures (25).

S. Cazzamalli is a co-inventor in a patent application assigned to Philogen claiming CAIX-targeted SMDC/Immunocytokine combination. D. Neri is an employee of and has ownership interests (including patents) at Philogen. No potential conflicts of interest were disclosed by the other authors.

Conception and design: S. Cazzamalli, F. Pretto, S. Wulhfard, D. Neri

Development of methodology: S. Cazzamalli, B. Ziffels, D. Neri

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Cazzamalli, B. Ziffels, F. Widmayer, P. Murer, G. Pellegrini, F. Pretto, S. Wulhfard

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S. Cazzamalli, B. Ziffels, F. Widmayer, G. Pellegrini, D. Neri

Writing, review, and/or revision of the manuscript: S. Cazzamalli, B. Ziffels, G. Pellegrini, D. Neri

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S. Cazzamalli, P. Murer, D. Neri

Study supervision: D. Neri

We thank Dr. Nikolaus Krall for his contribution to the biodistribution experiments of Supplementary Fig. S1. We acknowledge the positive and constructive criticisms made by Dr. Alberto Dal Corso. The authors gratefully acknowledge financial support from ETH Zürich, the Swiss National Science Foundation (Projects No. 310030B_163479/1, SINERGIA CRSII2_160699/1), ERC Advanced Grant “Zauberkugel” (Grant Agreement 670603), and Kommission für Technologie und Innovation (Grant No. 17072.1).

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.