Although immune checkpoint blockade (ICB) has shown remarkable clinical benefit in a subset of patients with melanoma and lung cancer, most patients experience no durable benefit. The receptor tyrosine kinase AXL is commonly implicated in therapy resistance and may serve as a marker for therapy-refractory tumors, for example in melanoma, as we previously demonstrated. Here, we show that enapotamab vedotin (EnaV), an antibody–drug conjugate targeting AXL, effectively targets tumors that display insensitivity to immunotherapy or tumor-specific T cells in several melanoma and lung cancer models. In addition to its direct tumor cell killing activity, EnaV treatment induced an inflammatory response and immunogenic cell death in tumor cells and promoted the induction of a memory-like phenotype in cytotoxic T cells. Combining EnaV with tumor-specific T cells proved superior to either treatment alone in models of melanoma and lung cancer and induced ICB benefit in models otherwise insensitive to anti–PD-1 treatment. Our findings indicate that targeting AXL-expressing, immunotherapy-resistant tumors with EnaV causes an immune-stimulating tumor microenvironment and enhances sensitivity to ICB, warranting further investigation of this treatment combination.

Significance:

These findings show that targeting AXL-positive tumor fractions with an antibody–drug conjugate enhances antitumor immunity in several humanized tumor models of melanoma and lung cancer.

Immune checkpoint blockade (ICB) has revolutionized the treatment of several cancer indications, particularly melanoma and lung cancer, most notably by inhibition of the PD-1/PD-L1 pathway (1–4). Combining ICB regimens, such as anti–PD-1 together with anti–CTLA-4 therapy, further increases response rates in melanoma and non–small cell lung cancer (NSCLC; refs. 5, 6). However, both innate and acquired resistance to ICB, conferred by several mechanisms, remain a challenge for the majority of patients (7). For instance, mutations in B2M, loss of HLA expression or other defects in the antigen presentation machinery can lead to immune evasion (8–10). In addition, an immunosuppressive tumor microenvironment can influence tumor inflammation status and impede response to checkpoint inhibition (11–14). Therefore, there is a dire need for biomarkers that can identify therapy-resistant tumor fractions, as well as for novel treatment options to combat immunotherapy-refractory tumors (15).

In recent years, the receptor tyrosine kinase AXL has emerged as a promising oncology target, especially in the context of therapy resistance. We and others previously reported that increased AXL expression marks resistance to targeted tumor therapy in melanoma and lung cancer (16, 17), possibly via its association with epithelial-to-mesenchymal transition (18). For instance, melanoma cells that express high levels of AXL are commonly insensitive to BRAF and MEK inhibition (16, 19). Recently, reports in breast and head and neck cancer indicated that AXL may also have an active role in immune suppression, for instance by lowering HLA class I expression and inducing several immunosuppressive cytokines, as well as PD-L1 expression (20, 21). Therefore, AXL may serve not only as a biologically relevant marker for (immuno)therapy-resistant cell fractions, but also as an attractive therapeutic target (20, 22).

Consistent with this notion, we recently showed that AXL-expressing melanoma cells can be effectively eliminated by the AXL-targeting antibody–drug conjugate (ADC) enapotamab vedotin (EnaV), and that EnaV can synergize with BRAF+MEK inhibition in melanoma by eliminating therapy-resistant cell fractions (18). On the basis of these findings, and the notion that AXL expression may serve as a marker of resistance to ICB (22), we hypothesized that immunotherapy-resistant, AXL-expressing cell fractions could similarly be targeted by EnaV, enabling immune cell antitumor responses. Targeting AXL by means of EnaV, which consists of an antibody conjugated to the microtubule inhibitor MMAE, may have an additional benefit, because cytotoxic payloads such as MMAE have previously been shown to mediate primary and adaptive immunity associated with immunogenic cell death (23), leading to combined efficacy with ICB (24). Therefore, we set out to mechanistically explore the immunogenic effects of EnaV, as well as to determine the combinatorial efficacy of EnaV, adoptive human T-cell transfer and ICB in several human xenograft models of melanoma and lung cancer.

Cell lines and cell culture conditions

All melanoma and lung cancer cell lines were obtained from the Peeper laboratory cell line stock, or from the ATCC (Genmab experiments). Melanoma cell lines were cultured in DMEM (Gibco), and lung cancer cells were cultured in RPMI (Gibco), supplemented with FBS (Sigma), 100 U/mL penicillin and 0.1 mg/mL streptomycin (both Gibco) or 10% Donor Bovine Serum with Iron (Life Technologies), respectively, under standard conditions. Cell lines were regularly confirmed to be Mycoplasma free.

Isolation and generation of T-cell receptor–specific CD8 T cells

MART-1 (1D3) T-cell receptor (TCR) retrovirus was produced in a packaging cell line as described previously (25). Peripheral blood mononuclear cells were isolated from healthy donor buffy coats (Sanquin) by density gradient centrifugation using Lymphoprep (Stem Cell Technologies). CD8+ T cells were purified using CD8 Dynabeads (Thermo Fisher Scientific), activated for 48 hours on a non-tissue culture treated 24-well plate that was precoated overnight with αCD3 and αCD28 antibodies (eBioscience, 16–0037–85 and 16–0289–85) at 2 × 106 per well. Activated CD8 T cells were harvested and mixed with TCR retrovirus and spinfected on a Retronectin coated (Takara, 25 μg per well) non-tissue culture treated 24-well plate for 2 hours at 2,000 × g. After 24 hours, T cells were harvested and maintained in RPMI (Gibco) containing 10% human serum (One Lamda), 100 U per mL of penicillin, 100 μg per mL of streptomycin, 100 U per mL IL2 (Proleukin, Novartis), 10 ng per mL IL7 (ImmunoTools), and 10 ng per mL IL15 (ImmunoTools).

HLA-A2 and MART-1 transduction in tumor cells

MART-126–35 and HLA-A2 were introduced using lentiviral (for BLM and LCLC103H) and retroviral (SkMel-147) constructs. Constructs for lentivirus were packaged using two helper plasmids (psPax and MS2G, Addgene) in HEK293T cells. Constructs for retrovirus were produced in a packaging cell line (FLY cells). Viral supernatant was either snap frozen or immediately used for infection. MART-126–35–Katushka–positive cells were sorted by flow cytometry. Except for the AXL expression experiment of Fig. 2A and immunogenic cell death (ICD) experiments, all experiments with BLM, SkMel-147, and LCLC-103H were performed using the MART-1 + HLA-A2–transduced cell lines.

In vitro cytotoxicity assays

For EnaV assays, 1.2 × 103 cells were plated in 96-well plates and the drugs were added 1 to 3 hours after seeding with the HP D300 Digital Dispenser (Tecan, Giessen, Germany). Phenylarsine oxide (PAO) was used as a positive control. After 5 days of incubation, the medium was replaced by a dilution of CellTiter Blue (Promega) in medium. Fluorescence was measured by the Infinite M200 microplate reader (Tecan) after 3 hours. The percentage of living cells was calculated using the following equation: % living cells = (signal treated–signal PAO)/(signal untreated–signal PAO) × 100%.

For T-cell killing colony formation assays, a total of 2 × 105 tumor cells were plated per well on a 6-well plate or 1 × 105 per well on a 12-well plate. CD8 T cells were admixed simultaneously in the indicated ratios and washed away after 24 hours. After 2 days the plates were washed, fixed, and stained for 1 hour using a crystal violet solution containing 0.1% crystal violet (Sigma) and 50% methanol (Honeywell). For quantification, remaining crystal violet was solubilized in 10% acetic acid (Sigma). Absorbance of this solution was measured on an Infinite 200 Pro spectrophotometer (Tecan) at 595 nm.

Animal studies and in vivo drug treatments of SkMel-147, BLM, and LCLC103H

All in vivo drug combination experiments were performed in either NSG or NSG-β2Mnull mice (The Jackson Laboratory). Mice were inoculated subcutaneously at the right flank with 1 × 106 cells in 1:1 Matrigel (Corning) and normal medium. Tumor size was measured three times weekly with a caliper, and tumor volume was calculated using the following formula: ½ × length (mm) × width (mm)2. Randomization occurred in a blinded fashion, when tumors reached approximately 100 mm3. Mice were treated intravenously with 5 × 106 untransduced or MART-1–specific T cells, diluted in 0.2 mL PBS per mouse. Enapotamab vedotin (Genmab) or control ADC were combined in this mixture and were dosed on body weight of the mice in a range of 1 to 4 mg/kg (indicated per experiment in the figure legends). On the moment of T-cell injection, these T cells were supported intraperitoneally by 100,000 IU IL2 (Novartis) for 3 days. The experiment ended for individual mice either when the tumor size exceeded 1 cm3, the tumor showed ulceration, in case of serious clinical illness, when the tumor growth blocked the movement of the mouse, or when tumor growth assessment had been completed. Anti–PD-1 treatment (pembrolizumab, SelleckChem, A2005, 5 mg/kg) was given weekly from the start of randomization via intraperitoneal injections. Differences in mean tumor volumes were compared between treatment groups using the Kruskal–Wallis test. Mantel–Cox analysis of Kaplan–Meier curves was performed to analyze statistical differences in overall survival time with a general tumor size cutoff value of 500 or 1,000 mm3 (indicated per experiment in figure legends).

Animal studies with patient-derived xenograft models

Tumor fragments from donor mice bearing patient‐derived NSCLC xenografts (LXFA526 or LXFA677, both EGFR wild type) were used for subcutaneous inoculation of 4- to 6-week-old male NMRI nu/nu mice (experiments performed by Charles River Discovery Research Services). Randomization of animals was performed as follows: Animals bearing a tumor with a volume of about 200 mm3 were distributed across 7 experimental groups (3 animals per group), considering a comparable median and mean of group tumor volume. IgG1-b12 (isotype control) and EnaV were dosed at 4 mg/kg and were intravenously injected on the day of randomization (day 0). At the indicated time points, the mice were sacrificed and tumor samples were collected for molecular profiling.

The lung cancer patient-derived xenograft (PDX) model LU5401, which harbors EGFR variant R521K, not considered an activating mutation (26), was established in huCD34NSG-SGM3 mice with a humanized immune system (HIS). These HIS mice had been generated by inoculation with human CD34+ hematopoietic stem cells at 3 weeks of age, giving rise to a humanized immune system with >45% huCd45 cells and >11% huCD3+ cells at 21 weeks of age, at which point the mice were implanted with LU5401 NSCLC PDX tumors. After tumor establishment (average tumor volume per treatment group 355–428 mm3), mice were treated with either 4 mg/kg EnaV or IgG1-b12, once a week for 2 consecutive weeks. Next, tumors were harvested for molecular profiling 3 days after treatment start or within an average of 30 days (range, 28–35 days) after initiation of treatment.

Animal husbandry and study approval

Animal experiments of SkMel-147, BLM, and LCLC103H were approved by the animal experimental committee of the institute and performed according to Dutch law. In vivo experiments with PDX LU5401 were performed at CrownBio, and experiments with all other PDX were conducted by Charles River Discovery Research Services. These experiments were approved by the local authorities, and were conducted according to all applicable international, national and local laws and guidelines. Only animals with unobjectionable health were selected to enter testing procedures. Animals were routinely monitored, twice daily on working days and daily on Saturdays and Sundays. Specific tumor xenograft characteristics with experiment-limiting impact (e.g., cachexia-inducing tumors) were considered according to available metadata information. Monitoring routinely included mortality checks, assessment of animal welfare, and tumor growth by observation, control of feed and water supply.

Calreticulin surface expression

LCLC-103H cells were grown to 70% to 80% confluence in 6-well plates, washed with serum-free medium and incubated with EnaV (2 μg/mL), free monomethyl auristatin E (MMAE, SAFC, cat. no. SG10; 10 nmol/L), or isotype control ADC (IgG1-b12-vcMMAE; 2 μg/mL) in serum-free medium (RPMI-1640, Lonza BE17–603E) for 48 hours. After incubation, supernatant and cells were collected and washed in PBS/0.1% BSA/0.02% azide (FACS buffer). Cells were then incubated with a phycoerythrin (PE)-conjugated mouse anti-human calreticulin antibody (Enzo, cat. no. ADI-SPA-601PE-D, lot. 10021835) for 30 minutes at 4°C in the dark. A PE-conjugated mouse isotype control antibody (Enzo, cat. no. ADI-SAB-600PE-D, lot. 08071817) was included as a control. After washing with FACS buffer, cells were analyzed with flow cytometry in the presence of DAPI (BD Biosciences; cat. no. 564907, lot. 8012653) or TO-PRO-3 (Thermo Fisher, cat. no. T3605, lot. 1976612). Calreticulin expression was determined as PE-positive cells and the percentage of calreticulin-positive cells over isotype control was obtained on DAPI/TO-PRO-3-negative cells. Relative calreticulin expression was calculated and expressed as fold change over untreated control. The ratio of mean PE fluorescence intensity (MFI) of viable (DAPI negative) cells over the PE MFI of viable, untreated control cells was calculated.

Extracellular ATP secretion

LCLC-103H cells were seeded in quadruplicate at 25,000 cells per well in 24-well plates and grown for 3 hours at 37°C. Subsequently, EnaV (2 μg/mL), free monomethyl auristatin E (MMAE; 10 nmol/L), or isotype control (IgG1-b12-vcMMAE; 2 μg/mL) were added per well. Plates were incubated for 48 hours at 37°C and after incubation each well was washed with 200 μL PBS. After the final wash, PBS was replaced with 200 μL fresh PBS for 10 minutes, harvested and transferred to 96-well plate, and then centrifuged at 1,000 rpm for 2 minutes at room temperature. Supernatants were transferred to a new 96-well (optiwhite) plate and ATP was measured using CellTiterGlo Luminescent Cell Viability Assay (Promega, cat. no. G7570) according to the manufacturer's instructions. All steps were performed with ice-cold PBS and in the dark. Relative ATP secretion was calculated and expressed as fold change over untreated control.

Extracellular HMGB1 secretion

LCLC-103H cells were seeded in quadruplicate at 25,000 cells/well in 24-well plates and grown 3 to 4 hours at 37°C. Subsequently, free MMAE at 10 nmol/L, EnaV (2 μg/mL) or isotype control ADC (2 μg/mL) were added to the wells and plates were incubated for 48 hours at 37°C. After incubation cells were pelleted at 1,000 rpm for 3 minutes at room temperature and supernatants were transferred to a new 96-well plate. HMGB1 was measured in the supernatant using an HMGB1-specific ELISA kit (IBL, cat. no. ST51011) according to the manufacturer's instructions. Relative HMGB1 secretion was calculated and expressed as fold change over untreated control.

Western blotting and antibodies

Cell pellets were lysed in RIPA buffer (50 mmol/L TRIS pH 8.0, 150 mmol/L NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with complete protease and phosphatase inhibitor cocktail (Thermo Fisher). Protein concentration was determined with the BCA Protein Assay Kit (Pierce). Western blotting was performed with standard techniques using 4% to 12% Bis-Tris polyacrylamide-SDS gels (NuPAGE, Life Technologies) and nitrocellulose membranes using the iBlot 2 system (Licor). Blots were blocked in 4% milk in PBS plus 0.2% Tween 100 and incubated with primary antibody: Cleaved caspase-3 (1:1,000, 9446T, Cell Signaling Technology), caspase-3 (1:1,000, 14220S, Cell Signaling Technology), PARP (1:1,000, 9542, Cell Signaling Technology) or tubulin (1:5,000, 9026, Sigma). The following secondary antibodies were used: Goat anti-rabbit peroxidase conjugate (1:5,000, G21234) and goat anti-mouse (1:5,000, G21040), both purchased from Invitrogen. Western blots were incubated in a 1:1 dilution of solution 1 (0.1 mol/L Tris pH 8, 2.5 mmol/L luminol, 0.4 mmol/L p-Coumaric acid, all Sigma) and solution 2 (0.1 mol/L Tris pH 8, 30% H2O2, all Sigma) and chemiluminescent signal was visualized using the Bio-Rad ChemiDoc imaging system.

Flow cytometry

For assessment of TCR transduction efficiency, the mouse TCR β chain (expressed within the construct) was stained (BD Pharmingen, 553172). For in vivo melanoma-infiltrating T-cell analyses, cells were first single-cell dissociated ex vivo using collagenase and mechanical dissection. Afterwards, the cell pellet was stained with CD8-Pacific Blue (558207, BD Biosciences) and PD-1-APC (17–5983–42, eBioscience) for 30 minutes at 4°C and analyzed at LSRII or LSR Fortessa (BD Biosciences).

Proteomic and RNA profiling of lung cancer tumor and melanoma xenografts

Fresh-frozen tumor samples were subjected to tissue lysis, protein extraction, and trypsin digestion. Isolated peptides were then quantified and used for construction of sample-specific spectral library using shotgun LC-MS/MS technology; samples were then analyzed by high-reaction monitoring (HRMTM) mass-spectroscopy for the unbiased detection of all detectable proteins and peptides (Biognosys).

For NGS RNA profiling of the lung cancer PDX (GenomeScan), rRNA-depleted RNA was purified from formalin-fixed, paraffin-embedded (FFPE) tumor tissue, and subjected to paired-end RNA sequencing (150 bp fragment length) on the NovaSeq 6000 platform (Illumina, Inc.). For melanoma tumors, fresh-frozen tumor samples were subjected to tissue lysis and RNA extraction and 65bp single-end sequencing was performed on an Illumina HiSeq2500.

Figure 1.

EnaV induces an inflammatory response in PDX models of lung cancer in vivo and in vitro. A, Heat map of the EnaV-associated gene signature based on RNA sequencing of lung cancer PDX models treated with either control (Ctrl; IgG1-b12) or EnaV (4 mg/kg) for 6 days. B, GSEA of EnaV signature, showing significantly induced hallmark gene sets. C, Proteomics analysis of tumors (at day 6) used for RNA profiling. A Q value of <0.1 and fold change of >1.5 were used as a cutoff value. Green dots indicate significant differentially expressed proteins involved in the inflammatory response. D–F, Expression of three established ICD markers in LCLC-103H cells in response to EnaV treatment. All samples were normalized to untreated control. Free MMAE at 10 nmol/L was used as a positive control. Error bars, SD of three independent replicates. Statistical analysis by one-tailed Student t test comparing samples with untreated control or IgG1-b12-MMAE. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure 1.

EnaV induces an inflammatory response in PDX models of lung cancer in vivo and in vitro. A, Heat map of the EnaV-associated gene signature based on RNA sequencing of lung cancer PDX models treated with either control (Ctrl; IgG1-b12) or EnaV (4 mg/kg) for 6 days. B, GSEA of EnaV signature, showing significantly induced hallmark gene sets. C, Proteomics analysis of tumors (at day 6) used for RNA profiling. A Q value of <0.1 and fold change of >1.5 were used as a cutoff value. Green dots indicate significant differentially expressed proteins involved in the inflammatory response. D–F, Expression of three established ICD markers in LCLC-103H cells in response to EnaV treatment. All samples were normalized to untreated control. Free MMAE at 10 nmol/L was used as a positive control. Error bars, SD of three independent replicates. Statistical analysis by one-tailed Student t test comparing samples with untreated control or IgG1-b12-MMAE. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Close modal

Quantitative RT-PCR

RNA was extracted using the Isolate II RNA Mini Kit (Bioline) and 1 μg of total RNA was reverse transcribed into cDNA using the Maxima First Strand cDNA Kit (Thermo Scientific). Primers are listed in Table 1 (below). RPL13 was used as control housekeeping gene. Real-time quantitative PCR amplification was performed using a 96-well plate system (OneStepPlus Real-Time PCR System, Thermo Fisher Scientific).

Table 1.

Primer sequences used in this study.

GenePrimer sequence
IFIT1 F: ATTTACAGCAACCATGAGTACAAA 
 R: TCCCACACTGTATTTGGTGTC 
IFIT2 F:CTGCAACCATGAGTGAGAACA 
 R: GGTTGCACATTGTGGCTTTGA 
IFIT3 F: CAGAGGGCAGTCATGAGTGAGG 
 R: GCCAACAAGTTGTACATTGTAGCTT 
MX1 F: CAGCTCAGGGGCTTTGGAAT 
 R: CCTTGGAATGGTGGCTGGAT 
RPL13 F: GAGACAGTTCTGCTGAAGAACTGAA 
 R: TCCGGACGGGCATGAC 
GenePrimer sequence
IFIT1 F: ATTTACAGCAACCATGAGTACAAA 
 R: TCCCACACTGTATTTGGTGTC 
IFIT2 F:CTGCAACCATGAGTGAGAACA 
 R: GGTTGCACATTGTGGCTTTGA 
IFIT3 F: CAGAGGGCAGTCATGAGTGAGG 
 R: GCCAACAAGTTGTACATTGTAGCTT 
MX1 F: CAGCTCAGGGGCTTTGGAAT 
 R: CCTTGGAATGGTGGCTGGAT 
RPL13 F: GAGACAGTTCTGCTGAAGAACTGAA 
 R: TCCGGACGGGCATGAC 

IHC

IHC for AXL, CD8, and PD-L1 was performed manually using the following protocol: Slides were deparaffinized and antigen retrieval was performed using Tris/EDTA for 15 minutes in the pressure cooker. Slides were incubated with primary antibody AXL (1:100, C89E7, CST), CD8 (1:100, Dako Cytomation; M7103), or PD-L1 (1:200, 13684P, Cell Signaling Technology) overnight at 4°C. Secondary antibody for AXL was polymer-HRP anti-rabbit envision (K4011, Dako) and visualization was performed using DAB (Sigma). For CD8 and PD-L1 the secondary antibody was polymer-HRP anti-mouse envision (K4007, Dako) and visualization was performed using DAB (Sigma). A counterstain with hematoxylin was performed, and tissue slides were manually analyzed and scored by a certified animal pathologist.

For F4/80 macrophage marker expression, FFPE tumor sections were analyzed by IHC for F4/80 and quantified using digital pathology. In brief, after deparaffinization and antigen retrieval, sections were stained with rat anti-mouse monoclonal F4/80 antibody (Bio-Rad/AbD Serotec, MCA497). F4/80 staining specificity was controlled by incorporating an isotype control antibody on an adjacent tumor section. All slides were digitalized using the Nanozoomer scanner (HAMAMATSU) in brightfield condition (objective ×20), and the xenograft tumor area used in the analysis was manually delineated using a hematoxylin and eosin counterstained guiding slide. The resulting labeling surface of F4/80 was quantified using HALO (Indica Labs) software. The results are expressed as the percentage (%) of F4/80 staining positivity: % = (F4/80 positively stained area*100)/total tumor area.

CD3 protein expression was evaluated by automated immunohistochemical staining of paraffin-embedded tumor specimens on the Roche Ventana Discovery autostainer platform with the use of a human CD3 ready-to-use IHC assay (rabbit-anti-human CD3 clone 2GV6, 790–4341, Roche). Tissue sections of standard thickness (4 μm) were subjected to standard protocols for deparaffinization, antigen retrieval (Tris-EDTA buffer pH 7.8), endogenous peroxidase activity block (S2003, Dako Agilent), and primary antibody detection [OmniMap anti-rabbit HRP (5269679001) and ChromoMap DAB (5266645001), both from Roche]. CD3 staining specificity was controlled by incorporating isotype control stainings on consecutive tissue sections. Tissues were counterstained with hematoxylin. Sections were mounted in ClearVue (4212, Thermo Fisher). Whole-tissue slides were scanned with a standardized scanning profile on Axio Scan Z1 (Zeiss). Digital images were analyzed for CD3 quantitation within viable intratumoral regions with tailored image analysis algorithm in HALO software (Indica Labs). CD3 quantitation readouts were generated by calculating the number of CD3+ cells per surface area (mm2).

Bioinformatic analyses

Sequence samples were mapped to the human genome (Homo.sapiens.GRCh38.v82) using STAR(2.6.0c) in two-pass mode with default settings. The mapped PDX samples were filtered for contaminating sequence reads of mouse using XenofilteR (27). The (filtered) samples were used to generate read count data using HTSeq-count (28). Normalization and statistical analysis of the expression of genes was performed using DESeq2 (29). Gene set enrichment analysis (GSEA) Preranked was performed using the BROAD javaGSEA standalone version (http://www.broadinstitute.org/gsea/downloads.jsp). Gene ranking was performed using the signal-to-noise ratio and run with 10,000 permutations. The T-cell gene set (Supplementary Fig. S4) was taken from Supplementary Table S3 from Tirosh and colleagues (30), and the memory + effector gene sets (Supplementary Fig. S4B) from Sarkar and colleagues (31). The gene sets from Supplementary Fig. S4C were generated using MCP Counter. The clinical datasets were taken from Gide and colleagues (ENA/SRA database: PRJEB23709; ref. 32) and Riaz and colleagues (ENA/SRA database: PRJNA356761; ref. 33). Fastq files were downloaded and processed using the identical pipeline as for the PDX and melanoma samples as described above. Clinical data were taken from the supplementary tables from the original articles. Response to ICB was based on RECIST criteria as described previously in the articles (responders: complete response/partial response/stable disease, nonresponders: progressive disease).

Statistical analysis

The data of in vivo experiments were analyzed at the indicated time points in legends by the nonparametric Mann–Whitney test for 2 conditions, or Kruskal–Wallis test when >2 conditions were compared. Survival analyses on Kaplan–Meier curves were analyzed using the log-rank Mantel–Cox test. All analyses were performed using the Prism GraphPad software. The data of in vitro experiments on melanoma cell lines were analyzed using the Mann–Whitney test for 2 conditions or Kruskal–Wallis test for >2 conditions. Induction of the molecular ICD markers was analyzed using one-tailed student t tests against untreated control samples.

Data availability

RNA sequencing data from this study have been deposited at the European Genome-phenome Archive (https://ega-archive.org/) under study number EGAS00001004562.

EnaV induces an inflammatory response in PDX models of lung cancer in vivo and in vitro

To examine whether targeting AXL with EnaV may promote a pro-inflammatory tumor microenvironment, we first investigated the transcriptional and proteomic changes upon EnaV treatment in human cancer models. To this end, we implanted two AXL-positive lung cancer PDX models into mice and, after tumor establishment, treated them with either EnaV or control antibody. In line with our previous findings (18), two rounds of EnaV treatment were sufficient to significantly inhibit tumor growth in these AXL-positive models (Supplementary Fig. S1A and S1B). To look at the early effects of EnaV treatment, we harvested tumors 3 and 6 days after start of therapy and performed RNA and proteomic profiling (Supplementary Fig. S1C). First, we computationally dissected human from mouse reads in the RNA sequencing dataset to specifically investigate the effects of EnaV on the human (tumor cell) compartment, using the XenofilteR algorithm we previously developed (27). Principal component analysis (PCA) revealed that whereas the first two principal components differentiated the two PDX models from each other, the third principal component separated EnaV-treated tumors from control ones in both PDX models (Fig. 1A; Supplementary Fig. S1D–S1F; Supplementary Table S1). GSEA of PC3 showed that the two highest enriched gene sets were both involved in cell-cycle checkpoints (Fig. 1B). This is consistent with the mechanism of action of MMAE, the cytotoxic moiety linked to the AXL antibody of EnaV, which causes G2–M cell-cycle arrest (34). Other top enriched gene sets were dominated by inflammatory processes, such as interferon response pathways (Fig. 1B). These results were confirmed by proteomic profiling of the tumors, where we also observed that interferon response proteins such as MX1, IFIT1, and IFIT2 were significantly induced upon EnaV treatment (Fig. 1C; Supplementary Fig. S1G; Supplementary Table S2). This was validated by quantitative RT-PCR for IFIT 1 and IFIT2 on tumors obtained from treated mice, indicating that these proteins were transcriptionally induced by EnaV (Supplementary Fig. S1H).

We also assessed whether a similar phenomenon of tumor-associated inflammation would be observed in mice bearing multiple components of the immune system. As EnaV is not cross-reactive with mouse AXL, we could not employ syngeneic mouse models to answer this question (18). Furthermore, for not fully understood reasons, mouse-derived tumor cell lines are intrinsically resistant to MMAE compared with human cells (Supplementary Fig. S2A). Therefore, we used human immune system (HIS) mice, which harbor human hematopoietic stem cells, giving rise to a broad spectrum of human immune cells (35). We implanted a lung cancer PDX in these HIS mice and assessed the tumor response to EnaV (Supplementary Fig. S2B). In this tumor model, too, we observed an enrichment of the PC3 gene set, corresponding to induction of similar proinflammatory gene sets as seen in the other PDX models (Supplementary Fig. S2C and S2D). These results indicate that tumor-associated inflammation upon EnaV treatment is a robust phenotype detected across a variety of tumor and mouse models.

To further investigate this observed inflammation-associated response to EnaV in cancer cells, we assessed the induction of ICD markers in vitro. ICD has been proposed to be induced upon cytotoxic cell stress, including by certain ADC payloads such as MMAE (23), and may reflect tumor cell death–induced inflammation (36). We treated the LCLC-103H lung cancer cell line with either EnaV, control ADC or free, unconjugated MMAE, and assessed several known markers of an in vitro ICD phenotype. Treatment with either EnaV or free MMAE led to a pronounced increase in extracellular ATP and HMGB1 release in the supernatant of the cell culture, which are both damage-associated molecular patterns and indicative of ICD (Fig. 1D and E). An increase was observed also for calreticulin translocation to the cell surface, another marker of ICD (Fig. 1F). Together, these results indicate that EnaV elicits an immunogenic response in tumor cells, as indicated by the induction of inflammatory genes and proteins by tumor cells in vivo, as well as by the induction of ICD hallmarks in vitro.

EnaV and tumor-specific T cells cooperate in vitro to induce tumor cell killing in melanoma and lung cancer cell lines

As EnaV induced several ICD hallmarks and a proinflammatory phenotype in tumor cells, we hypothesized that this could sensitize tumor cells toward T-cell–mediated killing. First, we analyzed a panel of melanoma and NSCLC cell lines expressing AXL (Fig. 2A). This cohort indeed displayed sensitivity to EnaV, and as expected not to a control ADC (IgG1-b12-MMAE; Fig. 2B), in line with our previous results (18). Next, we assessed the in vitro and in vivo sensitivity to T-cell attack within this panel of tumor cell lines. We made use of an overexpression system that we recently optimized in which we constitutively express an HLA type (HLA-A*02:01) and an antigen (MART-1) in tumor cells, which allows for recognition by cognate cytotoxic T cells carrying a MART-1–specific TCR (Fig. 2C for graphic overview; refs. 25, 37). All cell lines displayed dose-dependent sensitivity to MART-1, tumor-specific T cells in vitro (Fig. 2D; Supplementary Fig. S3A). We then asked if the extent of T-cell killing could be increased by combining with EnaV. Indeed, cooperative sensitivity to combined treatment with tumor-specific T cells and EnaV was observed in vitro (Fig. 2E). Mechanistically, this was apoptosis-mediated, because we observed a cooperative increase in cleaved PARP and cleaved caspase-3 upon the combination of EnaV and MART-1 T-cell treatment (Fig. 2F; Supplementary Fig. S3B).

Figure 2.

EnaV and tumor-specific T cells cooperate in vitro to induce tumor cell killing in melanoma and lung cancer cell lines. A, Flow cytometry of cell surface AXL expression in indicated cell lines. B, Cytotoxicity assay in three cell lines for EnaV versus control (Ctrl) ADC (IgG1-b12-MMAE). Figure shows a representative experiment. Error bars represent SD of three technical replicates, repeated three biological times. C, Graphic overview of matched tumor-specific T-cell and tumor model. Tumor cell lines were transduced with HLA-A2 and MART-1–encoding sequences, and T cells were retrovirally transduced with a TCR recognizing MART-1 loaded on HLA-A2. D, Quantification of T-cell sensitivity in a panel of tumor cell lines that were transduced with MART-1 + HLA-A2. Error bars, SD of three independent replicates. Viability was normalized to Ctrl T-cell treatment. E, Colony formation assay quantification of indicated treatments on cell lines transduced with MART-1 + HLA-A2. Tumor cells were treated for 4 days and stained with crystal violet. 2:1 tumor cell:T-cell ratio and 0.1 μg/mL EnaV was used. Error bars, SD of three independent replicates. Statistical analysis by the Kruskal–Wallis test. **, P < 0.01; ****, P < 0.0001; ns, not significant. F, Western Blot analysis of cleaved PARP in MART-1 + HLA-A2–transduced BLM cells treated with EnaV at 2 mg/mL and/or MART-1–specific T cells (or control IgG1-b12-MMAE and Ctrl T cells) in a 1:4 T-cell:tumor cell ratio for 20 hours. Tubulin was used as a loading control.

Figure 2.

EnaV and tumor-specific T cells cooperate in vitro to induce tumor cell killing in melanoma and lung cancer cell lines. A, Flow cytometry of cell surface AXL expression in indicated cell lines. B, Cytotoxicity assay in three cell lines for EnaV versus control (Ctrl) ADC (IgG1-b12-MMAE). Figure shows a representative experiment. Error bars represent SD of three technical replicates, repeated three biological times. C, Graphic overview of matched tumor-specific T-cell and tumor model. Tumor cell lines were transduced with HLA-A2 and MART-1–encoding sequences, and T cells were retrovirally transduced with a TCR recognizing MART-1 loaded on HLA-A2. D, Quantification of T-cell sensitivity in a panel of tumor cell lines that were transduced with MART-1 + HLA-A2. Error bars, SD of three independent replicates. Viability was normalized to Ctrl T-cell treatment. E, Colony formation assay quantification of indicated treatments on cell lines transduced with MART-1 + HLA-A2. Tumor cells were treated for 4 days and stained with crystal violet. 2:1 tumor cell:T-cell ratio and 0.1 μg/mL EnaV was used. Error bars, SD of three independent replicates. Statistical analysis by the Kruskal–Wallis test. **, P < 0.01; ****, P < 0.0001; ns, not significant. F, Western Blot analysis of cleaved PARP in MART-1 + HLA-A2–transduced BLM cells treated with EnaV at 2 mg/mL and/or MART-1–specific T cells (or control IgG1-b12-MMAE and Ctrl T cells) in a 1:4 T-cell:tumor cell ratio for 20 hours. Tubulin was used as a loading control.

Close modal

EnaV and tumor-specific T cells cooperatively inhibit tumor growth in vivo

Next, we determined the in vivo sensitivity of the same panel of cell lines to T cells and EnaV. First, we established an adoptive T-cell transfer model, where we implanted human tumor cells subcutaneously, followed by an intravenous injection of either control (Ctrl) or MART-1–specific T cells after tumor establishment. Interestingly, whereas all models displayed sensitivity to T cells in vitro, the in vivo efficacy of MART-1 T cells was different between models: BLM and LCLC-103H melanoma and lung cancer tumors showed moderate responses to MART-1–specific T cells, but SkMel-147 melanoma tumors were completely insensitive (Fig. 3A–C). Of note, untransduced T cells did not alter the growth kinetics compared with no treatment (Supplementary Fig. S4). None of these models showed a curative or even durable response, nor did they respond as strongly to T cells compared with other sensitive models that we recently established (37, 38), indicating that all three tumor models were partially or fully resistant to T-cell attack in vivo. When we analyzed the tumors from each experiment at endpoint, we observed that high AXL expression was maintained in all three models (Fig. 3D). Interestingly, however, SkMel-147 harbored no T cells, whereas the BLM tumors were still heavily infiltrated with T cells at this time (Fig. 3D). LCLC-103H showed few T cells infiltrating the tumor (Fig. 3D, arrows indicate T cells). In contrast, PD-L1 was highly expressed in all three tumor models (Fig. 3D).

Figure 3.

EnaV and tumor-specific T cells cooperatively inhibit tumor growth in vivo. A–C, T-cell sensitivity of the cell line panel transduced with MART-1 + HLA-A2 in vivo. Tumors were treated at 100 mm3 with Ctrl or MART-1 T cells. n = 6 per group for SkMel-147 and BLM, and n = 4 for LCLC103H Ctrl T cells versus n = 8 for the MART-1 T-cell group. All mice received 5 million of either Ctrl or MART-1 T cells. Error bars represent SD. Statistical analysis by the Mann–Whitney test; **, P < 0.01; ns, not significant. D, Histology (H&E, hematoxylin and eosin) and IHC analysis for AXL, PDL-1, and CD8 in the three tumor cell line xenografts. Representative images of each marker in endpoint tumors. Scale bars, 50 μm. Arrows, T cells. E and F, Hematoxylin SkMel-147 in vivo sensitivity to EnaV (2 mg/kg) and/or MART-1–specific T cells (5 million per mouse). Treatment started on the day indicated by arrow (day 9). n = 10 mice per group except for the Ctrl group, which had n = 9 mice. The cutoff tumor volume for survival curve was at 500 mm3. Error bars, SD. Statistical analysis by the Mann–Whitney test; *, P < 0.05; **, P < 0.01; ns, not significant. Statistical analysis of survival curve by the log-rank Mantel Cox test; ***, P < 0.001.

Figure 3.

EnaV and tumor-specific T cells cooperatively inhibit tumor growth in vivo. A–C, T-cell sensitivity of the cell line panel transduced with MART-1 + HLA-A2 in vivo. Tumors were treated at 100 mm3 with Ctrl or MART-1 T cells. n = 6 per group for SkMel-147 and BLM, and n = 4 for LCLC103H Ctrl T cells versus n = 8 for the MART-1 T-cell group. All mice received 5 million of either Ctrl or MART-1 T cells. Error bars represent SD. Statistical analysis by the Mann–Whitney test; **, P < 0.01; ns, not significant. D, Histology (H&E, hematoxylin and eosin) and IHC analysis for AXL, PDL-1, and CD8 in the three tumor cell line xenografts. Representative images of each marker in endpoint tumors. Scale bars, 50 μm. Arrows, T cells. E and F, Hematoxylin SkMel-147 in vivo sensitivity to EnaV (2 mg/kg) and/or MART-1–specific T cells (5 million per mouse). Treatment started on the day indicated by arrow (day 9). n = 10 mice per group except for the Ctrl group, which had n = 9 mice. The cutoff tumor volume for survival curve was at 500 mm3. Error bars, SD. Statistical analysis by the Mann–Whitney test; *, P < 0.05; **, P < 0.01; ns, not significant. Statistical analysis of survival curve by the log-rank Mantel Cox test; ***, P < 0.001.

Close modal

Because SkMel-147 was completely nonresponsive to tumor-specific T cells in vivo, and given the antitumor activity of EnaV in vitro, we asked whether EnaV has a therapeutic impact, alone or in a combination setting, on this immunotherapy-resistant tumor. Therefore, we implanted SkMel-147 cells in a new cohort of mice and after tumor establishment treated them with either EnaV, MART-1 T cells or the combination. As expected, based on our previous findings and the fact that this tumor expresses high levels of AXL, SkMel-147 was sensitive to even a single dose of EnaV (Fig. 3E). Strikingly, however, combining EnaV with MART-1–specific T cells proved superior over EnaV alone in both tumor growth inhibition and extending animal survival, even though these tumor-specific T cells had no efficacy as a single agent in this model (Fig. 3E and F). Therefore, we concluded that, first, EnaV proved effective in causing regression of tumors that were otherwise unresponsive to immune attack (albeit temporarily, likely because only a single EnaV dose was given), but more importantly, that EnaV and T cells cooperate in inhibition of tumor growth and prolonging animal survival.

EnaV promotes immune infiltration and induces phenotypic changes in T cells

Consistent with the proinflammatory tumor phenotype observed in response to EnaV treatment in lung cancer PDX models (Fig. 1A and B), we observed that melanoma tumors treated with EnaV displayed enrichment of remarkably similar gene sets associated with tumor inflammation (Fig. 4A and B). This observation prompted the question whether the induction of a tumor-intrinsic inflammatory response affects the influx and phenotype of T cells in the tumor and could thus provide mechanistic support for increased tumor kill in the presence of EnaV. We compared the BLM melanoma samples that were treated with MART-1 T cells ± EnaV. We did not observe an enhanced T-cell influx in the tumor based on T-cell–specific gene expression (Supplementary Fig. S5A), possibly due to the fact that this model was already highly infiltrated (Fig. 3D). However, when we assessed the phenotype of these T cells, we found that they were now expressing high levels of the activation marker CD137 (TNFRSF9), without significantly enhancing their exhaustion phenotype as assessed by PD-1 expression (PDCD1; Fig. 4C). In addition, in response to EnaV T cells were skewed toward a memory-like phenotype (Tmem cells), rather than presenting with a classical effector phenotype as seen in the T cells alone condition (Supplementary Fig. S5B). Tmem cells have been shown to promote long-term T-cell fitness and antitumor efficacy (39–41), which is consistent with the increased antitumor activity that we observe here.

Figure 4.

EnaV promotes immune infiltration and induces phenotypic changes in T cells. A, Graphical overview of set-up of experiment. Mice were injected with BLM melanoma cells transduced with MART-1 + HLA-A2 and randomized at 100 mm3 into the following treatment arms: Ctrl ADC + Ctrl T cells, EnaV (4 mg/kg) + Ctrl T cells, Ctrl ADC + MART-1 T cells, or EnaV + MART-1 T cells. For each treatment arm, 5 million T cells (Ctrl or MART-1) were injected per mouse. Tumors were harvested and RNA sequenced on day 7 after start of treatment. B, GSEA of comparison between Ctrl ADC (IgG1-b12-MMAE) versus EnaV, showing significantly induced inflammation-associated hallmark gene sets. C,TNFRSF9 (CD137) expression versus PD1 gene expression in BLM tumors derived from xenografted mice treated with MART-1 T cells ± EnaV; tumors were harvested after 7 days of treatment. D, Heat map of the relative abundance of imputed T-cell subsets in a lung cancer PDX model treated with either control IgG1-b12 or EnaV. T-cell abundance was imputed on the basis of gene expression using MCP Counter. E,TNFRSF9 (CD137) expression versus PD1 expression in HIS mice treated with control IgG1-b12 or EnaV. Mice carried LU5401 lung cancer PDX tumors, and tumors were harvested at an early time point (3 days) or late time point (28–35 days) after treatment. F, GSEA of comparison between responding (R) versus nonresponding (NR) patients at baseline for anti–PD-1 immunotherapy in clinical datasets for melanoma. An FDR value of <0.05 was used as cutoff for significance (all depicted gene sets are significant).

Figure 4.

EnaV promotes immune infiltration and induces phenotypic changes in T cells. A, Graphical overview of set-up of experiment. Mice were injected with BLM melanoma cells transduced with MART-1 + HLA-A2 and randomized at 100 mm3 into the following treatment arms: Ctrl ADC + Ctrl T cells, EnaV (4 mg/kg) + Ctrl T cells, Ctrl ADC + MART-1 T cells, or EnaV + MART-1 T cells. For each treatment arm, 5 million T cells (Ctrl or MART-1) were injected per mouse. Tumors were harvested and RNA sequenced on day 7 after start of treatment. B, GSEA of comparison between Ctrl ADC (IgG1-b12-MMAE) versus EnaV, showing significantly induced inflammation-associated hallmark gene sets. C,TNFRSF9 (CD137) expression versus PD1 gene expression in BLM tumors derived from xenografted mice treated with MART-1 T cells ± EnaV; tumors were harvested after 7 days of treatment. D, Heat map of the relative abundance of imputed T-cell subsets in a lung cancer PDX model treated with either control IgG1-b12 or EnaV. T-cell abundance was imputed on the basis of gene expression using MCP Counter. E,TNFRSF9 (CD137) expression versus PD1 expression in HIS mice treated with control IgG1-b12 or EnaV. Mice carried LU5401 lung cancer PDX tumors, and tumors were harvested at an early time point (3 days) or late time point (28–35 days) after treatment. F, GSEA of comparison between responding (R) versus nonresponding (NR) patients at baseline for anti–PD-1 immunotherapy in clinical datasets for melanoma. An FDR value of <0.05 was used as cutoff for significance (all depicted gene sets are significant).

Close modal

We further investigated this phenomenon in the lung cancer PDX model LU5401, which was implanted in HIS mice. Interestingly, we observed that EnaV promoted gene expression changes suggestive of immune infiltration of mostly T cells in this model. This was further confirmed by IHC analysis showing increased infiltration of CD3+ T cells in these tumors following EnaV treatment (Fig. 4D; Supplementary Fig. S5C and S5D). Similar to the melanoma model, also these T cells were skewed toward a more activated phenotype, assessed by CD137 expression (Fig. 4E).

To extend these findings, we also investigated whether EnaV treatment affects endogenous mouse immune cells in our NSCLC PDX models implanted in nude mice, which lack T cells but have other immune cells such as myeloid cells (Supplementary Fig. S1C). First, we analyzed the tumors for expression of the mouse-specific macrophage marker F4/80 by IHC. We observed that in both models, EnaV treatment enhanced macrophage tumor influx (Supplementary Fig. S5E). Protein expression analysis of tumors treated with EnaV also revealed that there was an induction of mouse B2m, consistent with an increased antigen presentation phenotype (Supplementary Fig. S5F).

Together, these data indicate that EnaV elicits an inflammatory tumor phenotype that can have a promoting effect on immune cells in several different ways. Corroborating these data clinically, we also observed that the inflammatory gene sets that were induced by EnaV in both lung cancer and melanoma models, correlated with anti–PD-1 immunotherapy outcome in two independent clinical datasets of melanoma (Fig. 4F; refs. 32, 33). These clinical data support the emerging hypothesis from our mouse studies that the inflammatory microenvironment induced by EnaV treatment in tumors may be beneficial for ICB response.

EnaV enhances ICB benefit in human melanoma and lung cancer models in vivo

Given the observation that EnaV induced an inflammation-associated phenotype, and potentiation of T cells associated with a memory-like phenotype, we asked whether ICB further enhances the antitumor activity of these T cells in the presence of EnaV. We tested this hypothesis in xenograft models of both melanoma and lung cancer.

First, we confirmed that MART-1 T cells were able to infiltrate the tumors, and importantly, that they expressed PD-1. Indeed, tumors from both models harbored T cells seven days after T-cell injection, assessed by flow cytometry (Fig. 5A and B). This was in concordance with the IHC results we obtained (Fig. 3D). Moreover, high PD-1 expression was observed on these T cells, whereas the control T cells were hardly expressing any PD-1 (Fig. 5A and B).

Figure 5.

EnaV enhances ICB benefit in human melanoma and lung cancer models in vivo. A and B, T-cell infiltration quantification and PD-1 MFI of MART-1–specific T cells versus control T cells in BLM and LCLC-103H xenografts. T cells were injected at 100 mm3, and flow cytometry–based readout was performed on endpoint tumors (average 1,000 mm3). C and D,In vivo sensitivity of BLM melanoma cells (transduced with MART-1 + HLA-A2) to EnaV (4 mg/kg) and/or MART-1–specific T cells (5 million per mouse) and/or anti–PD-1 (5 mg/kg). Treatment started on day 7. All treatments were given once except for anti–PD-1, which was given weekly. Ctrl group, n = 10 mice; MART-1 + anti–PD-1 ± EnaV groups, n = 13; all other groups, n = 11. The cutoff tumor volume for survival curve was at 1,000 mm3. Error bars, SD. Statistical analysis by the Mann–Whitney test; *, P < 0.05. Statistical analysis of survival curve by the log-rank Mantel–Cox test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. E and F,In vivo sensitivity of LCLC-103H (transduced with MART-1 + HLA-A2) to EnaV (1 mg/kg) and/or MART-1–specific T cells (5 million per mouse) and/or anti–PD-1 (5 mg/kg). Treatment started on day 16. Ctrl group, MART-1 T-cell group, and MART-1 + anti–PD-1 group, n = 7; other groups, n = 11. EnaV/Ctrl ADC and anti–PD-1 were given weekly. The cutoff tumor volume for survival curve was at 500 mm3. Error bars, SD. Statistical analysis by the Mann–Whitney test; *, P <0.05. Statistical analysis of survival curve by the log-rank Mantel–Cox test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Figure 5.

EnaV enhances ICB benefit in human melanoma and lung cancer models in vivo. A and B, T-cell infiltration quantification and PD-1 MFI of MART-1–specific T cells versus control T cells in BLM and LCLC-103H xenografts. T cells were injected at 100 mm3, and flow cytometry–based readout was performed on endpoint tumors (average 1,000 mm3). C and D,In vivo sensitivity of BLM melanoma cells (transduced with MART-1 + HLA-A2) to EnaV (4 mg/kg) and/or MART-1–specific T cells (5 million per mouse) and/or anti–PD-1 (5 mg/kg). Treatment started on day 7. All treatments were given once except for anti–PD-1, which was given weekly. Ctrl group, n = 10 mice; MART-1 + anti–PD-1 ± EnaV groups, n = 13; all other groups, n = 11. The cutoff tumor volume for survival curve was at 1,000 mm3. Error bars, SD. Statistical analysis by the Mann–Whitney test; *, P < 0.05. Statistical analysis of survival curve by the log-rank Mantel–Cox test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. E and F,In vivo sensitivity of LCLC-103H (transduced with MART-1 + HLA-A2) to EnaV (1 mg/kg) and/or MART-1–specific T cells (5 million per mouse) and/or anti–PD-1 (5 mg/kg). Treatment started on day 16. Ctrl group, MART-1 T-cell group, and MART-1 + anti–PD-1 group, n = 7; other groups, n = 11. EnaV/Ctrl ADC and anti–PD-1 were given weekly. The cutoff tumor volume for survival curve was at 500 mm3. Error bars, SD. Statistical analysis by the Mann–Whitney test; *, P <0.05. Statistical analysis of survival curve by the log-rank Mantel–Cox test; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001.

Close modal

Next, we asked whether melanoma tumors would respond better to the combination of T cells and anti–PD-1 in the presence of EnaV. Because SkMel-147 displayed no T-cell infiltration at all upon treatment, we instead focused on the BLM model. We treated established BLM melanomas with either EnaV, tumor-specific T cells, or both, and combined these treatments with PD-1 antibody treatment. Although EnaV treatment showed a potent antitumor effect, a single dose was insufficient to cause prolonged tumor inhibition (Fig. 5C). Tumor-specific T cells were less potent than EnaV and caused only partial tumor control, even when combined with anti–PD-1 therapy. Combining a single-dose EnaV treatment with MART-1 T cells proved superior to each agent alone (Fig. 5C), similarly to what was observed in the SkMel-147 model (Fig. 3E). Most importantly, anti–PD-1, which failed to show benefit in the absence of EnaV treatment, significantly potentiated this effect when administered with EnaV. As a result, mice receiving the triple treatment survived significantly longer (Fig. 5D).

Finally, we determined whether these results could be recapitulated in a lung cancer model. Whereas EnaV had only a partial effect at a low dose tested in this model, combining it with both tumor-specific T cells and anti–PD-1 again proved superior to either treatment alone, leading to a significantly prolonged survival of the mice (Fig. 5E and F). In conclusion, these results indicate that EnaV potentiates the antitumor effect of T-cell attack in vivo and creates de novo sensitivity to ICB in immunotherapy-resistant lung and melanoma tumor models.

We show here that targeting AXL using an ADC approach promotes antitumor T-cell activity and enhances the efficacy of ICB in models of both melanoma and lung cancer. First, EnaV induced an inflammatory environment in tumors in vivo, increased T-cell influx and tumor penetration, and skewed cytotoxic T cells toward a CD137+ phenotype, all of which may augment antitumor efficacy. Second, EnaV exerted an antitumor effect in cancer models that were otherwise insensitive to tumor-specific T cells and anti–PD-1 ICB. Finally, EnaV potentiated anti–PD-1 therapy in melanoma and lung cancer models, leading to de novo ICB benefit in these models.

The relevance of this work is reinforced by several findings from recently published work. We previously showed that EnaV has effective antitumor activity in preclinical lung cancer models (42). Of note, we also previously reported an acceptable toxicology profile of EnaV as a single agent in cynomolgus monkeys (18). Moreover, EnaV in cooperation with BRAF pathway inhibition has demonstrated potent inhibitory effects in therapy-refractory melanoma tumors (18). AXL expression in preclinical models of breast cancer has been associated with immune resistance due to several mechanisms (20). Mechanistically, it has recently been suggested that AXL-positive cancer cells secrete immunosuppressive cytokines and they commonly express low levels of MHC class I and high levels of PD-L1 (20, 21). Moreover, AXL may serve as a biomarker for ICB resistance in clinical melanoma samples (22). AXL is also thought to drive resistance to cytotoxic lymphocyte-mediated killing in lung cancer cells in vitro (43). As such, targeting AXL-positive cancer cells by EnaV may be beneficial due to the killing of immunotherapy-refractory cell fractions.

Recent attempts to target AXL-expressing tumor cells have also focused on small-molecule–based inhibition of AXL, amongst others (43). However, this requires AXL to be an oncogenic driver of tumor biology, and this seems not, or only partially, to be the case for several tumor indications (16, 38, 44). An advantage of EnaV may be that it elicits an antitumor effect in two distinct ways. First, it eliminates AXL-positive cancer cells not by inhibition of AXL, but merely by using AXL as an address to deliver its cytotoxic payload. Hence, it is predicted to be efficacious also in tumors that do not rely on AXL signaling for their proliferative or survival capacity. Second, EnaV also induces MMAE-mediated bystander killing and has immune-promoting effects. Besides triggering a tumor-intrinsic inflammation phenotype through immunogenic cell death, it has been shown that microtubule-disrupting agents such as MMAE can also have direct stimulatory effects on immune cells, through enhancement of immune subsets, including antigen presentation of dendritic cells (DC) as well as direct T-cell proliferation (23, 24, 45). Also induction of cytokine-responsive genes such as the IFIT genes, which were potently induced in our models upon EnaV treatment, has been shown to contribute to antitumor immunity (46). Of note, those studies were largely performed in a syngeneic mouse setting; here, we show that EnaV potentiates anti–PD-1 therapy in a setting with human tumor xenografts and humanized immune system. It is likely that multiple of the above-described mechanisms contribute to the cooperative antitumor effect we observed for EnaV and immunotherapy.

Although we did not particularly investigate this in the current study, AXL can also be expressed on other cell subsets besides tumor cells, such as myeloid-derived suppressor cells (MDSC; ref. 47). In fact, AXL expression on MDSCs can drive a protumorigenic phenotype. It would thus be interesting to investigate whether EnaV can target or reverse the phenotype of such subsets. Similarly, AXL may also be expressed on DCs (48) and protumorigenic natural killer (NK) cells (49). More work is needed to investigate the effects of EnaV on these diverse AXL-positive immune subsets, as well as their contribution to ICB resistance.

Another novel aspect of this study is the finding that EnaV induces a CD137-high phenotype in cytotoxic T cells and that it may also be associated with memory-skewing as well as expansion of intratumoral immune cells. It has been shown preclinically that fully effector-like T cells are actually less potent in vivo in adoptive T-cell models compared with less differentiated, more memory-like T cells (39). Clinically, such memory-like T cells have been shown to promote long-term T-cell fitness and antitumor efficacy. They correlate with reduced metastatic invasion and increased overall survival, and have been shown to expand in melanoma patients in response to PD-1 blockade (40, 41). In line with these findings, our data also suggest that T cells in in vivo models can develop a dysfunctional T-cell state (indicated by high PD-1 expression and low CD137 expression; refs. 50, 51), which can be rescued by EnaV. In addition, upregulation of CD137 reflects expansion of activated, tumor antigen–specific T cells as previously reported (52). It is conceivable that these traits together contribute to the EnaV-mediated enhanced sensitivity of tumors to tumor-specific T cells as well as to reverting resistance to the benefit of PD-1 blockade.

In conclusion, we have shown here not only that targeting of AXL-positive melanoma and lung cancer fractions is beneficial for immunotherapy-refractory tumors, but also that targeting AXL by EnaV enhances the efficacy of T cells and can create de novo sensitivity to ICB. These results warrant further investigation of the combination of anti–PD-1 therapy and AXL-targeting ADCs.

J. Boshuizen reports a patent for WO2017009258A and WO2017121877 licensed. N. Pencheva, P. Garrido Castro, E. Gresnigt-Van den Heuvel, M.L. Janmaat, and M. Jure-Kunkel report salaried employment with Genmab that includes relevant stocks and warrants during the conduct of this study and outside the submitted work. M.L. Janmaat also reports a patent for WO/2019/197506 pending. M.A. Ligtenberg reports other from Immagene BV (shareholder) outside the submitted work, as well as employment with Immagene BV. D.S. Peeper reports grants from Genmab during the conduct of the study; is listed on and receives fees from Genmab patents relating to AXL-ADC; and is co-founder of, shareholder of, and adviser for Immagene BV, which is unrelated to the submitted work. No disclosures were reported by the other authors.

J. Boshuizen: Conceptualization, data curation, formal analysis, writing–original draft, writing–review and editing. N. Pencheva: Conceptualization, data curation, investigation, methodology, writing–review and editing. O. Krijgsman: Conceptualization, data curation, software, formal analysis. D. D'Empaire Altimari: Data curation, validation, investigation. P. Garrido Castro: Resources, data curation, formal analysis, writing–review and editing. B. de Bruijn: Investigation, mouse experiments, writing–review and editing. M.A. Ligtenberg: Conceptualization, data curation. E. Gresnigt-Van den Heuvel: Data curation. D.W. Vredevoogd: Conceptualization, data curation, investigation. J.-Y. Song: Visualization. N. Visser: Data curation. G. Apriamashvili: Data curation. M.L. Janmaat: Methodology, writing–review and editing. T.S. Plantinga: Data curation. P. Franken: Data curation. M. Houtkamp: Resources. A. Lingnau: Data curation. M. Jure-Kunkel: Supervision, writing–review and editing. D.S. Peeper: Conceptualization, funding acquisition, writing–review and editing.

The authors thank all the members of the Peeper and Blank laboratory for their valuable input and the FACS and animal facilities at the NKI for their support. They acknowledge the NKI-AVL Core Facility Molecular Pathology and Biobanking (CFMPB) for supplying NKI-AVL Biobank material and laboratory support. This study has been funded by a research grant from Genmab; the European Research Council under the European Union's Seventh Framework Programme (FP7/2007–2013)/ERC Synergy Grant agreement number 319661 COMBAT CANCER; and grants NKI 2014–7241, NKI 2013–5799, and NKI 2017–10425 from the Dutch Cancer Society (KWF). The Peeper lab is a member of the Oncode Institute, which is partly financed by the Dutch Cancer Society.

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.

1.
Hodi
FS
,
Chiarion-Sileni
V
,
Gonzalez
R
,
Grob
JJ
,
Rutkowski
P
,
Cowey
CL
, et al
Nivolumab plus ipilimumab or nivolumab alone versus ipilimumab alone in advanced melanoma (CheckMate 067): 4-year outcomes of a multicentre, randomised, phase 3 trial
.
Lancet Oncol
2018
;
19
:
1480
92
.
2.
Valsecchi
ME
. 
Combined nivolumab and ipilimumab or monotherapy in untreated melanoma
.
N Engl J Med
2015
;
373
:
23
34
.
3.
Garon
EB
,
Rizvi
NA
,
Hui
R
,
Leighl
N
,
Balmanoukian
AS
,
Eder
JP
, et al
Pembrolizumab for the treatment of non–small cell lung cancer
.
N Engl J Med
2015
;
372
:
2018
28
.
4.
Reck
M
,
Rodriguez-Abreu
D
,
Robinson
AG
,
Hui
R
,
Csöszi
T
,
Fülöp
A
, et al
Pembrolizumab versus chemotherapy for PD-L1–positive non–small cell lung cancer
.
N Engl J Med
2016
;
375
:
1823
33
.
5.
Larkin
J
,
Chiarion-Sileni
V
,
Gonzalez
R
,
Grob
JJ
,
Cowey
CL
,
Lao
CD
, et al
Combined nivolumab and ipilimumab or monotherapy in untreated melanoma
.
N Engl J Med
2015
;
373
:
23
34
.
6.
Hellmann
MD
,
Ciuleanu
TE
,
Pluzanski
A
,
Lee
JS
,
Otterson
GA
,
Audigier-Valette
C
, et al
Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden
.
N Engl J Med
2018
;
378
:
2093
104
.
7.
Sharma
P
,
Hu-Lieskovan
S
,
Wargo
JA
,
Ribas
A
. 
Primary, adaptive, and acquired resistance to cancer immunotherapy
.
Cell
2017
;
168
:
707
23
.
8.
McGranahan
N
,
Rosenthal
R
,
Hiley
CT
,
Rowan
AJ
,
Watkins
TBK
,
Wilson
GA
, et al
Allele-specific HLA loss and immune escape in lung cancer evolution
.
Cell
2017
;
171
:
1259
71
.
9.
Sucker
A
,
Zhao
F
,
Real
B
,
Heeke
C
,
Bielefeld
N
,
Maßen
S
, et al
Genetic evolution of T-cell resistance in the course of melanoma progression
.
Clin Cancer Res
2014
;
20
:
6593
604
.
10.
Burr
ML
,
Sparbier
CE
,
Chan
KL
,
Chan
YC
,
Kersbergen
A
,
Lam
EYN
, et al
An evolutionarily conserved function of polycomb silences the mhc class i antigen presentation pathway and enables immune evasion in cancer
.
Cancer Cell
2019
;
36
:
385
401
.
11.
Spranger
S
,
Bao
R
,
Gajewski
TF
. 
Melanoma-intrinsic β-catenin signalling prevents antitumour immunity
.
Nature
2015
;
523
:
231
5
.
12.
Wellenstein
MD
,
Coffelt
SB
,
Duits
DEM
,
van Miltenburg
MH
,
Slagter
M
,
de Rink
I
, et al
Loss of p53 triggers WNT-dependent systemic inflammation to drive breast cancer metastasis
.
Nature
2019
;
572
:
538
42
.
13.
Mariathasan
S
,
Turley
SJ
,
Nickles
D
,
Castiglioni
A
,
Yuen
K
,
Wang
Y
, et al
TGFβ attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells
.
Nature
2018
;
554
:
544
8
.
14.
Peng
W
,
Chen
JQ
,
Liu
C
,
Malu
S
,
Creasy
C
,
Tetzlaff
MT
, et al
Loss of PTEN promotes resistance to T-cell–mediated immunotherapy
.
Cancer Discov
2016
;
6
:
202
16
.
15.
Boshuizen
J
,
Peeper
DS
. 
Rational cancer treatment combinations: an urgent clinical need
.
Mol Cell
2020
;
78
:
1002
18
.
16.
Müller
J
,
Krijgsman
O
,
Tsoi
J
,
Robert
L
,
Hugo
W
,
Song
C
, et al
Low MITF/AXL ratio predicts early resistance to multiple targeted drugs in melanoma
.
Nat Commun
2014
;
5
:
5712
.
17.
Zhang
Z
,
Lee
JC
,
Lin
L
,
Olivas
V
,
Au
V
,
Laframboise
T
, et al
Activation of the AXL kinase causes resistance to EGFR-targeted therapy in lung cancer
.
Nat Genet
2012
;
44
:
852
60
.
18.
Boshuizen
J
,
Koopman
LA
,
Krijgsman
O
,
Shahrabi
A
,
Van Den Heuvel
EG
,
Ligtenberg
MA
, et al
Cooperative targeting of melanoma heterogeneity with an AXL antibody–drug conjugate and BRAF/MEK inhibitors
.
Nat Med
2018
;
24
:
203
12
.
19.
Konieczkowski
DJ
,
Johannessen
CM
,
Abudayyeh
O
,
Kim
JW
,
Cooper
ZA
,
Piris
A
, et al
A melanoma cell state distinction influences sensitivity to MAPK pathway inhibitors
.
Cancer Discov
2014
;
4
:
816
27
.
20.
Aguilera
TA
,
Rafat
M
,
Castellini
L
,
Shehade
H
,
Kariolis
MS
,
Hui
ABY
, et al
Reprogramming the immunological microenvironment through radiation and targeting Axl
.
Nat Commun
2016
;
7
:
13898
.
21.
Skinner
HD
,
Giri
U
,
Yang
LP
,
Kumar
M
,
Liu
Y
,
Story
MD
, et al
Integrative analysis identifies a novel AXL-PI3 kinase-PD-L1 signaling axis associated with radiation resistance in head and neck cancer
.
Clin Cancer Res
2017
;
23
:
2713
22
.
22.
Hugo
W
,
Zaretsky
JM
,
Sun
L
,
Song
C
,
Moreno
BH
,
Hu-Lieskovan
S
, et al
Genomic and transcriptomic features of response to anti–PD-1 therapy in metastatic melanoma
.
Cell
165
:
35
44
23.
Müller
P
,
Martin
K
,
Theurich
S
,
Schreiner
J
,
Savic
S
,
Terszowski
G
, et al
Microtubule-depolymerizing agents used in antibody–drug conjugates induce antitumor immunity by stimulation of dendritic cells
.
Cancer Immunol Res
2014
;
2
:
741
55
.
24.
Müller
P
,
Kreuzaler
M
,
Khan
T
,
Thommen
DS
,
Martin
K
,
Glatz
K
, et al
Trastuzumab emtansine (T-DM1) renders HER2
+
breast cancer highly susceptible to CTLA-4/PD-1 blockade
.
Sci Transl Med
2015
;
7
:
315ra188
.
25.
Jorritsma
A
,
Gomez-Eerland
R
,
Dokter
M
,
Van De Kasteele
W
,
Zoet
YM
,
Doxiadis
IIN
, et al
Selecting highly affine and well-expressed TCRs for gene therapy of melanoma
.
Blood
2007
;
110
:
3564
72
.
26.
Hsieh
YY
,
Tzeng
CH
,
Chen
MH
,
Chen
PM
,
Wang
WS
. 
Epidermal growth factor receptor R521K polymorphism shows favorable outcomes in KRAS wild-type colorectal cancer patients treated with cetuximab-based chemotherapy
.
Cancer Sci
2012
;
103
:
791
6
.
27.
Kluin
RJC
,
Kemper
K
,
Kuilman
T
,
de Ruiter
JR
,
Iyer
V
,
Forment
JV
, et al
XenofilteR: computational deconvolution of mouse and human reads in tumor xenograft sequence data
.
BMC Bioinformatics
2018
;
19
:
366
.
28.
Anders
S
,
Pyl
PT
,
Huber
W
. 
HTSeq-a Python framework to work with high-throughput sequencing data
.
Bioinformatics
2015
;
31
:
166
9
.
29.
Love
MI
,
Huber
W
,
Anders
S
. 
Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2
.
Genome Biol
2014
;
15
:
550
.
30.
Tirosh
I
,
Izar
B
,
Prakadan
SM
,
Wadsworth
MH
,
Treacy
D
,
Trombetta
JJ
, et al
Dissecting the multicellular ecosystem of metastatic melanoma by single-cell RNA-seq
.
Science
2016
;
352
:
189
96
.
31.
Sarkar
S
,
Kalia
V
,
Haining
WN
,
Konieczny
BT
,
Subramaniam
S
,
Ahmed
R
. 
Functional and genomic profiling of effector CD8 T-cell subsets with distinct memory fates
.
J Exp Med
2008
;
205
:
625
40
.
32.
Gide
TN
,
Quek
C
,
Menzies
AM
,
Tasker
AT
,
Shang
P
,
Holst
J
, et al
Distinct immune cell populations define response to anti–PD-1 monotherapy and anti–PD-1/anti–CTLA-4 combined therapy
.
Cancer Cell
2019
;
35
:
238
55
.
33.
Riaz
N
,
Havel
JJ
,
Makarov
V
,
Desrichard
A
,
Urba
WJ
,
Sims
JS
, et al
Tumor and microenvironment evolution during immunotherapy with nivolumab
.
Cell
2017
;
171
:
934
49
.
34.
Buckel
L
,
Savariar
EN
,
Crisp
JL
,
Jones
KA
,
Hicks
AM
,
Scanderbeg
DJ
, et al
Tumor radiosensitization by monomethyl auristatin E: mechanism of action and targeted delivery
.
Cancer Res
2015
;
75
:
1376
87
.
35.
De La Rochere
P
,
Guil-Luna
S
,
Decaudin
D
,
Azar
G
,
Sidhu
SS
,
Piaggio
E
. 
Humanized mice for the study of immuno-oncology
.
Trends Immunol
2018
;
39
:
748
63
.
36.
Kroemer
G
,
Galluzzi
L
,
Kepp
O
,
Zitvogel
L
. 
Immunogenic cell death in cancer therapy
.
Annu Rev Immunol
2013
;
31
:
51
72
.
37.
Vredevoogd
DW
,
Kuilman
T
,
Ligtenberg
MA
,
Boshuizen
J
,
Stecker
KE
,
de Bruijn
B
, et al
Augmenting immunotherapy impact by lowering tumor TNF cytotoxicity threshold
.
Cell
2019
;
178
:
585
99
.
38.
Boshuizen
J
,
Vredevoogd
DW
,
Krijgsman
O
,
Ligtenberg
MA
,
Blankenstein
S
,
de Bruijn
B
, et al
Reversal of pre-existing NGFR-driven tumor and immune therapy resistance
.
Nat Commun
2020
;
11
:
3946
.
39.
Gattinoni
L
,
Klebanoff
CA
,
Palmer
DC
,
Wrzesinski
C
,
Kerstann
K
,
Yu
Z
, et al
Acquisition of full effector function in vitro paradoxically impairs the in vivo antitumor efficacy of adoptively transferred CD8
+
T cells
.
J Clin Invest
2005
;
115
:
1616
26
.
40.
Pagès
F
,
Berger
A
,
Camus
M
,
Sanchez-Cabo
F
,
Costes
A
,
Molidor
R
, et al
Effector memory T cells, early metastasis, and survival in colorectal cancer
.
N Engl J Med
2005
;
353
:
2654
66
.
41.
Ribas
A
,
Shin
DS
,
Zaretsky
J
,
Frederiksen
J
,
Cornish
A
,
Avramis
E
, et al
PD-1 blockade expands intratumoral memory T cells
.
Cancer Immunol Res
2016
;
4
:
194
203
.
42.
Koopman
LA
,
Terp
MG
,
Zom
GG
,
Janmaat
ML
,
Jacobsen
K
,
Van Den Heuvel
EG
, et al
Enapotamab vedotin, an AXL-specific antibody–drug conjugate, shows preclinical antitumor activity in non–small cell lung cancer
.
JCI Insight
2019
;
4
:
e128199
.
43.
Terry
S
,
Abdou
A
,
Engelsen
AST
,
Buart
S
,
Dessen
P
,
Corgnac
S
, et al
AXL targeting overcomes human lung cancer cell resistance to NK- and CTL-mediated cytotoxicity
.
Cancer Immunol Res
2019
;
7
:
1789
802
.
44.
Axelrod
HD
,
Valkenburg
KC
,
Amend
SR
,
Hicks
JL
,
Parsana
P
,
Torga
G
, et al
AXL is a putative tumor suppressor and dormancy regulator in prostate cancer
.
Mol Cancer Res
2019
;
17
:
356
69
.
45.
Martin
K
,
Müller
P
,
Schreiner
J
,
Prince
SS
,
Lardinois
D
,
Heinzelmann-Schwarz
VA
, et al
The microtubule-depolymerizing agent ansamitocin P3 programs dendritic cells toward enhanced antitumor immunity
.
Cancer Immunol Immunother
2014
;
63
:
925
38
.
46.
Mbofung
RM
,
McKenzie
JA
,
Malu
S
,
Zhang
M
,
Peng
W
,
Liu
C
, et al
HSP90 inhibition enhances cancer immunotherapy by upregulating interferon response genes
.
Nat Commun
2017
;
8
:
451
.
47.
Holtzhausen
A
,
Harris
W
,
Ubil
E
,
Hunter
DM
,
Zhao
J
,
Zhang
Y
, et al
TAM family receptor kinase inhibition reverses MDSC-mediated suppression and augments anti–PD-1 therapy in melanoma
.
Cancer Immunol Res
2019
;
7
:
1672
86
.
48.
Seitz
HM
,
Camenisch
TD
,
Lemke
G
,
Earp
HS
,
Matsushima
GK
. 
Macrophages and dendritic cells use different Axl/Mertk/Tyro3 receptors in clearance of apoptotic cells
.
J Immunol
2007
;
178
:
5635
42
.
49.
Paolino
M
,
Choidas
A
,
Wallner
S
,
Pranjic
B
,
Uribesalgo
I
,
Loeser
S
, et al
The E3 ligase Cbl-b and TAM receptors regulate cancer metastasis via natural killer cells
.
Nature
2014
;
507
:
508
12
.
50.
Chen
Z
,
Ji
Z
,
Ngiow
SF
,
Manne
S
,
Cai
Z
,
Huang
AC
, et al
TCF-1–centered transcriptional network drives an effector versus exhausted CD8 T-cell fate decision
.
Immunity
2019
;
51
:
840
55
.
51.
Jacquelot
N
,
Roberti
MP
,
Enot
DP
,
Rusakiewicz
S
,
Semeraro
M
,
Jégou
S
, et al
Immunophenotyping of stage III melanoma reveals parameters associated with patient prognosis
.
J Invest Dermatol
2016
;
136
:
994
1001
.
52.
Ye
Q
,
Song
DG
,
Poussin
M
,
Yamamoto
T
,
Best
A
,
Li
C
, et al
CD137 accurately identifies and enriches for naturally occurring tumor-reactive T cells in tumor
.
Clin Cancer Res
2014
;
20
:
44
55
.