Arginase 1–Based Immune Modulatory Vaccines Induce Anticancer Immunity and Synergize with Anti–PD-1 Checkpoint Blockade

Cancer cells can directly inhibit anticancer immune mechanisms and corrupt immune cells to generate and uphold an immunosuppressive microenvironment. To evade immune surveillance, tumor cells can promote recruitment of myeloid-derived suppressor cells (MDSC) or differentiation of tumor-associated macrophages (TAM), which together contribute to impairing anticancer immunity through various mechanisms (1–3). Indeed, MDSCs and TAMs inhibit the activation, proliferation, and cytotoxicity of effector T cells and natural killer cells, as well as induce the differentiation and expansion of regulatory T cells (Treg). One of the most effective mechanisms exploited by MDSCs/TAMs for inhibiting T-cell fitness and activation is the aberrant consumption of essential amino acids such as tryptophan, cysteine, and L-arginine in the tumor microenvironment (4). Arginase 1 (ARG1) is expressed by MDSCs and TAMs and catalyzes the conversion of the amino acid L-arginine into L-ornithine and urea. Many studies have shown increased ARG1 activity in cancer, including head and neck cancer (5), breast cancer (6), renal cell carcinoma (7), and non–small cell lung cancer (8), and cancer cells can express ARG1 (9). Unsurprisingly, numerous approaches to inhibit ARG1 have been developed, and ARG1 inhibitors have shown promising results in different mouse models (10, 11). For instance, an ARG1 inhibitor was shown to block tumor growth in a mouse lung carcinoma model with a subpopulation of mature tumor-associated myeloid cells that express high ARG1 (10); analogously, ARG1 transcriptional inhibition by either AT38 ([3-(aminocarbonyl) furoxan-4-yl] methyl salicylate; ref. 12) or phosphodiesterase 5 inhibitors (13) increases the therapeutic impact of immunotherapy. Finally, genetic knockout of ARG1 improves survival in tumor-bearing mice receiving adoptive transfer of tumor-specific cytotoxic T cells (14).

We speculated on the possibility to evoke an immune response against ARG1 to limit the expansion of immunosuppressive elements in the tumor microenvironment. Our previous data report the existence of intrinsic T-cell immunity against ARG1-derived antigens in the peripheral blood of both cancer patients and healthy subjects (15–17), suggesting the presence of an endogenous T-cell receptor (TCR) repertoire toward ARG1 epitopes that can be utilized by immunotherapy. Indeed, ARG1-specific T cells specifically recognized ARG1-expressing immune cells in an ARG1-dependent manner (15), and the T-cell responses against ARG1 were part of the T-cell memory repertoire (17). To further extend our understanding of the role of ARG1-specific T cells, we examined the possibility to activate and expand in vivo ARG1-specific T cells by peptide-based vaccination. We also evaluated the effectiveness of this treatment in controlling tumor progression, either alone or in combination with checkpoint blockade targeting programmed cell death protein 1 (PD-1).

Female C57BL/6 mice were either of own breeding (CCIT-DK, 10–18 weeks old), purchased from Taconic (8–12 weeks old) or Charles River Laboratories Inc. Female BALB/c mice were purchased from Charles River Laboratories Inc. OT-1 TCR transgenic mice (C57BL/6-Tg (TcraTcrb)1100Mjb/J) and CD45.1⁺ congenic mice (B6.SJL-PtrcaPepcb/BoyJ) were purchased from Jackson Laboratories. All animal work was conducted under the approval of either the Danish Ethics Committee on experimental animal welfare and performed according to the Danish guidelines or approved by Verona University Ethical Committee according to the Italian guidelines (protocol no. 12722 approved by the Ministerial Decree No. 14/2012-B of January 18, 2012, and protocol no. BR15/08 approved by the Ministerial Decree No. 925/2015-PR of August 28, 2015). For both facilities, all animal experiments were conducted according to the guidelines of Federation of European Laboratory Animal Science Association (FELASA) and ARRIVE, European laws and regulations, and in accordance with the Amsterdam Protocol on animal protection and welfare. Mice were monitored daily and euthanized when displaying excessive discomfort; for example, decreased activity, piloerection, and an ungroomed appearance.

Six different 8 mer–20 mer ARG1 peptides (mARG29, mARG56, mARG212, mARG290, ARG1_261–280, and ARG1_191–210) and were predicted from the murine (Q61176) and human (P05089) sequence using the prediction server from the University of Tübingen (available at http://www.syfpeithi.de/bin/MHCServer.dll/EpitopePrediction.htm) or by manual screening of the sequence (15). Of these six peptides, only ARG1_261–280 was shared between the murine and human sequences of ARG1, and only ARG1_191–210 was predicted from the human sequence. ARG1_261–280 (TEEIYKTGLLSGLDIMEVNP), ARG1_191–210 (KTLGIKYFSMTEVDRLGIGK), mARG212 (MEETFSYL), and mARG290 (KSTVNTAVAL) were all reconstituted in 10 mmol/L DMSO. mARG56 (VDVPNDSSF) and mARG29 (AALRKAGLL) were reconstituted in 2 mmol/L H₂O. Peptides were purchased at Schäfer or Pepscan. Ovalbumin (OVA_257–264, SIINFEKL), the influenza virus hemagglutinin (HA_512–520, IYSTVASSL), or vehicle for peptide in Montanide ISA 51 VG (Seppic, cat. no. 36362Z) emulsion served as a control vaccine. Control peptides were purchased from JPT Peptide Technologies.

The following tumor cell lines were used for the studies reported in this paper: MC38, B16F10, CT26, and MCA205. MC38 and B16F10 were acquired from the NCI–DCTD Repository (NIH, Frederick, MD) in 2016. The mouse CT26 colon cell line was purchased from ATCC (CRL-2638) in 2019, whereas the mouse fibrosarcoma MCA205 cell line was received as a gift from Dr. Laurence Zitvogel (Institut Gustave Roussy, Université de Paris Saclay, INSERM) in 2017. Cell lines were thawed from primary stocks maintained under liquid nitrogen and cultured for a maximum of 3 weeks, during which time all experiments were performed. Cell cultures were regularly tested for Mycoplasma using the MycoAlert LookOut Mycoplasma PCR Detection Kit (Sigma-Aldrich). The cell lines have not been authenticated after acquisition.

An inoculum of 5 × 10⁵ or 8 × 10⁵ tumor cells were injected subcutaneously (s.c.) on the right flank in 100 μL DMEM media (Life Technologies; cat. no. 31966-047). After tumor inoculation, treatment with subcutaneous vaccines and anti–PD-1 therapy were initiated as indicated in figures. MC38 tumor cells were cultured in DMEM supplemented with 10% heat-inactivated fetal bovine serum (FBS; Invitrogen), 1% streptomycin/penicillin (Life Technologies), 10 mmol/L HEPES (Life Technologies), 0.1 mmol/L nonessential amino acids (Life Technologies), and 1 mmol/L sodium pyruvate. B16-F10 tumor cells were cultured in DMEM supplemented with 10% FBS and 1% streptomycin/penicillin. MCA205 tumor cells and CT26 tumor cells were cultured in DMEM supplemented with 2 mmol/L L-glutamine (Euroclone), 10 mmol/L HEPES (Euroclone), 20 μmol/L β-mercaptoethanol (Sigma-Aldrich), streptomycin (150 U/mL; Euroclone), penicillin (200 U/mL; Euroclone), and 10% FBS (Invitrogen). After thawing, tumor cells were always passaged a maximum of five times before inoculation. Tumor growth was measured three to four times a week using a digital caliper, and tumor sizes are presented as the mean ± the standard error of the mean (SEM). Tumor volume was calculated according to the following equation: V (mm³) = (d2 × D)/2, where d (mm) and D (mm) are the smallest and largest perpendicular tumor diameters, respectively, assessed by a caliper measurement. The maximum tumor volume allowed by the Danish Ethics Committee for these studies were 1,500 mm³, and by the Italian regulation 1,700 mm³ was allowed.

For tumor rechallenge, an inoculum of 5×10⁵ MC38 tumor cells was injected s.c. on the left flank in 100 μL DMEM, and tumor growth was assessed as described above. Seven tumor-naïve C57BL/6 mice were used as controls for tumor growth.

Animals were vaccinated subcutaneously, as indicated in figures, at the base of the tail with 100 μg peptide in DMSO/H₂O in a 1:1 emulsion with Montanide ISA 51 VG or vaccinated using Covax. Covax is based on the combination of an intraperitoneal injection of 100 mg anti-CD40 (FGK45.5, Bio X Cell) with 100 μg peptide in saline at the tail base and 50 mg of imiquimod 5% cream (Aldara 55 cream, Meda) applied on the vaccination site (14).

Animals were treated with anti-mouse PD-1 (clone: RMP1-14, BioSite; cat. no. BE0146), as indicated in figures, intraperitoneally (i.p.) with 250 μg per dose in multiple-dose studies. Anti–PD-1 was diluted to 250 μg/200 μL in phosphate-buffered saline (PBS) and 200 μL were administered per i.p. injection.

Spleens and draining lymph nodes (inguinal lymph nodes) were collected immediately after euthanasia and transferred to RPMI-1640 media supplemented with 10% heat-inactivated FBS, streptomycin/penicillin (100 μg/mL), and streptomycin (100 μg/mL). Each organ was processed through a 70-μm cell strainer, and spleens were treated with red blood cell lysis buffer (Qiagen; cat. no. 158904). All organs were washed twice in RPMI-1640 (Sigma-Aldrich; cat. no. R1145-500 mL), 10% FBS, 1% streptomycin/penicillin. Tumors were collected immediately after euthanasia and transferred to digestion buffer consisting of collagenase type I (2.1 mg/mL; Worthington; cat. no. LS004196) and DNase I (75 μg/mL; Worthington; cat. no. S002139) in RPMI-1640. The tumors were cut into small pieces using scissors and left overnight at 4°C on an end-over-end rotor (18). The following day, the tumors were shaken for 15 minutes at 37°C before they were passed through a 70-μm cell filter and washed in PBS. The tumors were treated with red cell lysis buffer and washed in PBS.

ELISPOT plates (Millipore; cat. no. MSIPN4W50) were coated with mouse IFNγ-specific capture antibody (AN18; Mabtech; cat. no. 3321-3-1000) at a concentration of 12 μg/mL in PBS overnight at 4°C. Splenocytes and cells from the draining lymph nodes from C57BL/6 or BALB/c mice were seeded in the ELISPOT wells in triplicates (8–10 × 10⁵ splenocytes per well and 6 × 10⁵ cell from the draining lymph nodes per well) with and without the respective ARG1 peptide (5 μmol/L) and incubated overnight. Concanavalin A (5 μmol/L; Sigma-Aldrich; cat. no. C5275) was used as positive controls in all setups. The following day, the cells were discarded, and the wells of the plates were washed with 300 μL PBS using a CAPP Plate Washer before the biotinylated secondary IFNγ antibody (1:1,000; R4-6A2-biotin; Mabtech, cat. no. 3321-6-1000) was added at the concentration of 1 μg/mL in ELISPOT buffer (PBS, 0.5% BSA and NaN₃) for 2 hours at room temperature. The plates were washed again before addition of streptavidin-AP (1:1,000; Mabtech, cat. no. 3310-10) for 1 hour at room temperature and washed for the last time before the enzyme substrate BCIP/NBT (Mabtech; cat. no. 3650-10) was added to the wells at room temperature for 1 to 5 minutes to visualize IFNγ-secreting cells. The spots were counted using the CTL ImmunoSpot S6 Ultimate-V analyzer with ImmunoSpot software, version 5.1. For phenotyping of CD4⁺ and CD8⁺ T cells using IFNγ ELISPOT, CD4⁺ cells were initially sorted using CD4 (L3T4) microbeads (Miltenyi MACS; cat. no. 130-117-043) for positive selection of CD4⁺ T cells. CD8⁺ cells were sorted from the CD4-negative fraction using CD8a (Ly-2) microbeads (Miltenyi MACS; cat. no. 130-117-044). For antigen-presenting cells, splenocytes from C57BL/6 mice were added to a concentration of 2:1 to the CD4⁺/CD8⁺ cells in the ELISPOT wells. All samples were performed in technical triplicates, and all ELISPOT data were normalized to 10⁶ cells.

Antibodies to CD45-PE/Cy7 (cat. no. 103113), CD11b-BV421 (cat. no. 101235), F4/80-APC (cat. no. 123115), CD206-PE (cat. no. 141705), PE-IgG2a, k isotype control (cat. no. 400211), CD3-FITC (cat. no. 100203), CD4-BV421 (cat. no. 100437), and CD8a-APC (cat. no. 100711) were purchased from BioLegend. Viability was assessed by Zombie Aqua Fixable Viability Kit (cat. no. 423101) purchased from BioLgend. Unspecific antibody binding to Fc-receptors was avoided by using mouse FcR blocking reagent (1:10; Miltinyi Biotec, 130-092-575). Samples were washed and stained for 30 minutes at 4°C, washed and resuspended in 100 μL FACS buffer for acquisition. All flow cytometry was performed on either the BD Biosciences FACSCanto or the ACEA NovoCyte Quanteon. Data were analyzed using FlowJo version 10.6 software. Overview of the gating strategy can be found in Supplementary Fig. S1.

For IHC and immunofluorescence, MCA205 tumors were collected from tumor-bearing C57BL/6 mice and fixed in 1% paraformaldehyde for 1 hour and frozen in a cryo-embedding medium (OCT; Bio-Optica, cat. no. 05-9801). Frozen sections were cut, and 5-μm slides were fixed in ice-cold acetone for 10 minutes and incubated with rabbit polyclonal antiarginase I (1:50; Santa Cruz, cat. no. sc20150) overnight, or rabbit monoclonal anti-CD3 (1:150, Abcam; cat. no. ab16669) for 1 hour. For immunofluorescence, primary antibody incubation was followed by secondary goat antirabbit Alexa Fluor 546 (1:500; Invitrogen, cat. no. A11010), and nuclei were stained with Dapi (Sigma; cat. no. D9542). Image acquisition was performed using Zeiss LSM 800 confocal microscope. ARG1 intensity was expressed as arithmetic mean intensity measured with Zen 2.3 Lite software and was evaluated on digital images (5 × 400 microscopic fields per sample). For IHC, primary antibody incubation was followed by secondary goat antirabbit (1:500; Jackson Immuno Research, cat. no. 111-065-144), and immunoreactive antigens were detected using streptavidin peroxidase (Thermo Scientific; cat. no. TS-125-HR) and the DAB Chromogen System (Dako; cat. no. K3468). The number of CD3⁺ cells was evaluated on digital images (3–5 × 200 microscopic fields per sample), acquired with Leica DMRD optical microscope (Leica).

Immunosuppressive activity of myeloid cells isolated from tumor-bearing mice was evaluated as previously described (19). Briefly, spleen-derived CD11b⁺ cells were isolated from MCA205 tumor–bearing mice by immunomagnetic sorting using CD11b Microbeads (Miltenyi Biotec) according to the manufacturer's instructions, and their purity was evaluated by flow cytometry. For all separations, the positive fraction was obtained with a purity of ≥ 95%. Purified CD11b⁺ cells were cocultured at different concentration in presence of CD45.2⁺ splenocytes from OT-I transgenic mice (Jackson Laboratories), labeled with 1 μmol/L CellTrace (Thermo Fisher Scientific) and diluted 1:10 with CD45.1⁺ splenocytes (Jackson Laboratories), in the presence of OVA peptide (1 μg/mL final concentration). After 3 days of coculture, cells were stained with APC-Cy7-conjugated anti-CD45.2 (clone 104, eBioscience, Thermo Fisher Scientific) and PerCP-Cy5.5-conjugated anti-CD8 (clone SK1, eBioscience, Thermo Fisher Scientific). CellTrace signal of gated lymphocytes was used to analyze cell proliferation. Samples were acquired with the BD Biosciences FACSCanto II, and data were analyzed using the FlowJo software.

C57BL/6 mice received four subcutaneous ARG1 peptide vaccination (as preciously described). Thirty-six days after the last vaccination, the mice were decapitated, and approximately 500 μL blood was collected in a 1-mL Eppendorf tube. Samples were left 30 minutes at room temperature to allow for blood coagulation. Samples were centrifuged for 10 minutes at 2000 × g. Sera (around 100–500 μL) were collected and stored at −20°C. Sera were tested with two kits: AST Activity Assay kit (Sigma-Aldrich, cat. no. MAK055-1KT) and ALT Activity Assay kit (Sigma-Aldrich, cat. no. MAK052-1KT) according to the manufacturer's protocols.

ELISPOT data were analyzed using the distribution-free resampling (DFR) method (20). For analysis of triplicates DFR, P < 0.05 (*) and DFRx2, P < 0.01 (**) were considered statistically significant. Analysis of flow cytometry data and ALT/AST levels was performed using Welch unpaired t tests. Antitumor effects over time were analyzed by using a mixed-effects model, whereas antitumor effects on specific times points were analyzed using Welch unpaired t test. Survival curves were analyzed using the log-rank (Mantel–Cox) test. An event in the survival data was determined as the maximum tumor size allowed by the Danish Ethics Committee for these studies (864 mm³ or 1,500 mm³). Error bars denote the SEM. The DFR analysis was performed in RStudio (RStudio Team (2019). RStudio: Integrated Development for R. RStudio, Inc.; http://www.rstudio.com/), and all other statistical analyses were performed using the GraphPad Prism 9.0.0 software. Sample sizes were chosen based on power calculations from pilot experiments.

We first examined the immunogenicity of two previously identified 20 amino acid long peptides (ARG1_261–280 and ARG1_191–210; ref. 15) after one vaccination with the individual peptides. ARG1_191–210 is derived from the human ARG1 sequence but only differs in one amino acid (R₂₀₅ to L₂₀₅ substitution in murine sequence), whereas ARG1_261–280 is conserved in both humans and mice. Upon vaccination with ARG1_261–280 or ARG1_191–210, we examined ex vivo IFNγ responses by ELISPOT assay (Fig. 1A). C57BL/6 mice vaccinated with the ARG1_261–280 peptide showed a significant ARG1_261–280-specific release of IFNγ (Fig. 1B), as well as BALB/c mice vaccinated with ARG1_191–210 peptide, which showed a significant release of IFNγ to ARG1_191–210 (Fig. 1C). To initially assess the impact of vaccination in controlling tumor growth, MCA205 tumor–bearing mice were vaccinated with ARG1_261–280 and ARG1_191–210 peptide in two different adjuvant settings, e.g., Montanide or anti-CD40 in combination with imiquimod (hereafter indicated as Covax; ref. 21). We found that ARG1-specific IFNγ responses were independent of the vaccine adjuvant and ARG1 epitopes (Supplementary Fig. S2) almost in all C57BL/6 (H-2^b) mice, even though only one epitope predicted to bind to H-2^b showed measurable immune responses (Supplementary Fig. S3). Thus, we decided to focus on the two long ARG1_261–280 and ARG1_191–210 peptides. To further characterize ARG1_261–280-specific immune responses, CD4⁺ and CD8⁺ T cells were isolated from splenocytes of vaccinated mice by magnetic bead sorting and evaluated for their ability to release IFNγ in presence with ARG1-derived peptide by ELISPOT assay (Fig. 1D). The phenotype of the sorted fractions was confirmed using flow cytometry (Supplementary Fig. S1). In all analyzed mice, CD4⁺ T cells secreted more IFNγ compared with CD8⁺ T cells (Fig. 1E). Despite ARG1-specific IFNγ responses, the vaccines did not associate with any detectable systemic toxicity, as vaccinated mice did not show weight loss over time, no increased immune activity in the liver, or aberrant levels of liver enzymes in the blood were detected (Fig. 2A–G).

To assess the therapeutic effectiveness of ARG1_261–280 vaccination in tumor-bearing mice, we evaluated two different cancer settings (Fig. 3). In both models, tumor cells were subcutaneously injected on day 0 prior to the first weekly vaccination. Melanoma (B16-F10) and colon adenocarcinoma (MC38) mice vaccinated with ARG1_261–280 had significantly reduced tumor growth compared with the control groups (B16-F10: P = 0.0001, MC38: P = 0.0098; Fig. 3; Supplementary Fig. S4A and S4B), and increased infiltration of CD4⁺ T cells was found in MC38 tumors of vaccinated mice compared with the control group (Fig. 3E).

Monoclonal anti–PD-1 has previously been shown to be functional in the MC38 mouse tumor model as monotherapy (22–24). Thus, we examined the combination of anti–PD-1 therapy and ARG1_261–280 vaccination in the MC38 cancer model. MC38 cells were inoculated prior to the first weekly vaccination with ARG1_261–280, and anti–PD-1 therapy was initiated on day 7. Both therapies were administered three times (Fig. 4A). Combination therapy with ARG1_261–280 vaccinations and anti–PD-1 therapy showed increased antitumor effects (P = 0.0001; Fig. 4B; Supplementary Fig. S4C) compared with either monotherapy with anti–PD-1 or ARG1_261–280 vaccination. The combination therapy also significantly increased the survival of tumor-bearing mice (P = 0.0042; Fig. 4C).

To understand the cellular mechanisms underlying this observed therapeutic combination effect, we analyzed the immune cell composition of the microenvironment. ARG1_261–280 vaccinations induced tumor infiltration of CD45⁺ cells both as a monotherapy and in combination with anti–PD-1 (Fig. 4D). The combination therapy also induced a significant ARG1-dependent increase in the ratio of M1-like macrophages to M2-like macrophages (Fig. 4E), thus indicating a shift from an immunosuppressive to a more proinflammatory tumor microenvironment.

The therapeutic effectiveness of anti–PD-1 therapy in combination with ARG1_261–280-based vaccination was confirmed in a different tumor setting using an anti–PD-1–sensitive sarcoma model (MCA205). MCA205 tumor–bearing mice were vaccinated 3 days after tumor challenge. On day 7, mice were boosted with ARG1 vaccine, whereas anti–PD-1 treatment was infused by four iterative intraperitoneal administrations every second day (Fig. 4F), as previously described (25). ARG1_261–280-based immunization partially controlled tumor progression without inducing a significant increase in the survival in tumor-bearing mice (Fig. 4G and H; Supplementary Fig. S4D). On the contrary, the combination of ARG1 vaccination with anti–PD-1 showed a therapeutic synergy, promoting both increased antitumor efficacy and significantly prolonged survival (Fig. 4G and H). To elucidate the impact of ARG1 vaccine on the myeloid compartment, an immunosuppression assay (19) was conducted. CD11b⁺ cells, containing immunosuppressive MDSCs, were purified from splenocytes of either vaccinated or unvaccinated tumor-bearing mice and cocultured in vitro at different cell ratios with cell-trace–labeled OT1 splenocytes in the presence of immunodominant OVA peptide (19). The immunoregulative properties of CD11b⁺ cells isolated from vaccinated mice were significantly impaired compared with cells from controls (P < 0.036), suggesting a potential reduction in ARG1-expressing myeloid cells by our developed immunization (Fig. 4I).

To further confirm the effectiveness of ARG1-based vaccination on controlling tumor progression, we also tested the ARG1_191–210-based vaccine in combination with anti–PD-1 treatment in the MCA205 tumor setting (Fig. 5A). In agreement with the previous results, the ARG1_191–210 vaccinations showed a direct antitumor effect compared with control groups and a similar therapeutic synergism in combination with anti–PD-1 treatment (Fig. 5B; Supplementary Fig. S4E). To investigate whether ARG1 targeting could be sufficient to reprogram the tumor framework, we evaluated tumor-infiltrating cells using IHC. We observed not only an increase in CD3⁺ T cells (Fig. 5C and D) in the ARG1_191–210 vaccination group compared with control groups, but an overall significant reduction in ARG1⁺ expression (Fig. 5C and E). In order to ensure that the antitumor effect was not limited to the C57BL/6 background, the same study was performed in the CT26 colon carcinoma model in BALB/c mice. Hemagglutinin (HA) peptide-based vaccination was utilized as a control. Similar antitumor effects by ARG1_191–210 vaccinations as monotherapy, as well as the synergy of the combination with anti–PD-1 treatment, were evident (Fig. 5A and F; Supplementary Fig. S4F).

The therapeutic combination of the ARG1_261–280-based vaccine and anti–PD-1 (Fig. 4B and C) resulted in complete tumor regression in several of the mice. To study whether the treatment had induced protective memory responses in these mice, they were rechallenged with 5×10⁵ MC38 tumor cells on the left flank on day 51 after primary tumor inoculation and treatment initiation (Fig. 6A). As a control to the rechallenge group, seven tumor- and treatment-naïve mice were inoculated with MC38 cells on the left flank. None of the mice were treated during the tumor rechallenge. Seven days following rechallenge initiation, tumors were measurable on all controls, whereas none of the rechallenged mice developed tumors (Fig. 6B; Supplementary Fig. S4G). Mice that cleared tumors were followed until day 185 after primary tumor inoculation, and spleens and draining lymph nodes were collected for IFNγ ELISPOT. Results showed a significant ARG1-specific IFNγ release in three of five mice treated with the combination therapy in the draining lymph nodes (Fig. 6C) and in the spleens of all mice treated with the combination therapy (Fig. 6D). Supernatants were collected from the ELISPOT cultures, and IFNγ ELISAs showed a significant release of IFNγ upon stimulation with the ARG1_261–280 peptide in all mice treated with the combination therapy (Fig. 6E).

In the current study, we examined the antitumor effect of ARG1-targeting vaccines in four different in vivo models of cancer. We showed that it was possible to induce CD4⁺ T-cell response against a peptide derived from ARG1 after just a single immunization using a simple vaccine based on a long peptide mixed with an adjuvant. We described that vaccinations with the ARG1_261–280 epitope had a significant antitumor effect on the growth of B16-F10 tumors, as well as MC38 tumors, and we also found an increase in CD4⁺ T cells in the tumor microenvironment in ARG1 peptide–vaccinated mice. In line with these results, we found increased CD3⁺ T-cell infiltration and decreased ARG1 expression in MCA205 tumors upon vaccination with the ARG1_191–210 epitope mixed with adjuvant. The antitumor effects of the ARG1_191–210-based vaccination were also evident in CT26 tumor–bearing mice with BALB/c background. Thus, the therapeutic and immunomodulatory effects of ARG1-targeting immunotherapy are independent of genetic background and cancer type, which allows us to speculate that this approach could be a broadly applicable treatment in different cancer settings. ARG1 expression is one of the main characteristics of both MDSCs and TAMs. ARG1-expressing myeloid cells play a major role in the development of a suppressive microenvironment, as they prevent effector lymphocyte proliferation at the tumor site (4, 26, 27). We previously described that ARG1-specific proinflammatory T cells are naturally occurring in the memory T-cell repertoire of the human immune system (17). We described that such T cells were naturally activated in T_H2 cytokine environments mimicking a tumor microenvironment (28). Hence, ARG1-specific T cells expanded in response to IL4 without any other specific stimulation. ARG1-specific T cells that release IFNγ in response to an ARG1-expressing M2 macrophages may therefore drive the immune response away from T_H2 and back into the T_H1 pathway. Here, we expanded these finding by describing that the activation of ARG1-specific T cells by peptide vaccines can directly be used to modulate the tumor microenvironment. We also previously reported the existence of other autoreactive, antigen-specific proinflammatory T cells [defined as antiregulatory T cells (anti-Tregs); ref. 29] that can react to immunosuppressive cells and therefore are able to counteract the many different immunosuppressive feedback mechanisms mediated by such regulatory cells. Hence, we have characterized anti-Tregs, which specifically recognize HLA-restricted epitopes derived from proteins expressed in regulatory immune cells, e.g., indoleamine 2,3-dioxygenase (IDO; refs. 30–33), IDO2 (34), tryptophan 2,3-dioxygenase (35), programmed death-ligand 1 (PD-L1; refs. 36–39), CCL22 (40), arginase 2 (ARG2; ref. 41), in addition to ARG1 (15–17) epitopes.

We previously described the existence of proinflammatory effector T cells that recognize ARG2 (41), and ARG2 is, therefore, also a potential target for novel immune modulatory vaccines. We do not yet know, however, if ARG2, like ARG1, can be utilized as a target for anticancer vaccinations. Because ARG1 and ARG2 are expressed heterogeneously, the combination of ARG1 and ARG2 for vaccination might capture different immunosuppressive arginase-expressing cells in the tumor microenvironment and benefit more patients. Activation of ARG1-specific T cells with a vaccination differentiates from other ways of targeting ARG1 in a cancer setting by its immunomodulatory effects, as it should cause T-cell infiltration at the tumor site converting an immunosuppressive environment into a proinflammatory environment. Thus, in contrast to ARG1 pharmacologic inhibition, ARG1 vaccination combines both TAM depletion (especially if it successfully activates ARG1-specific CD8⁺ T cells) and TAM reprogramming (42, 43) by introducing proinflammatory cytokines, such as IFNγ, into the immunosuppressive microenvironment. Indeed, we found that not only the lymphoid compartment was affected by the vaccines, but also the myeloid compartment, as the MDSCs in the tumor microenvironment were less immunosuppressive upon vaccination with ARG1_261–280. Accordingly, this rebalancing should increase the effect of T-cell–enhancing drugs, like immune-checkpoint blockers. MDSCs are under the influence of hypoxia-inducible factor 1 alpha (HIF1α) and A2 adenosine receptor (A2AR)–mediated immunosuppressive transcription and signaling. These are among the most important mechanisms used by MDSCs to decrease antitumor immunity and contribute to tumor immune escape (44). Thus, synergistic combination of ARG1 vaccines with the blockade of the hypoxia A2AR immunosuppressive pathway (45–47) could be an interesting future treatment modality.

To avoid immune-mediated clearance, advanced tumors exploit an array of immune suppressive pathways. These pathways are unlikely to be overcome when only interfering with signaling checkpoints (48, 49). We showed that ARG1 vaccination synergizes with anti–PD-1 therapy in multiple in vivo tumor models. TAM infiltration in the tumor microenvironment has been described as a major reason that checkpoint blockers show limited effects in most patients with cancer (50–54). ARG1-targeting vaccination induces a T_h1-associated inflammation that favors the expression of certain immune-related molecules such as PD-L1 in both cancer, immune, and stroma cells. Therefore, this immunotherapeutic approach could generate a microenvironment more sensitive to anti–PD-1/PD-L1 blockade therapy, as well as a long-term protective memory immune response as described in this study.

In conclusion, we demonstrated that an ARG1-targeting vaccine can activate the endogenous antitumor immunity in synergy with checkpoint blockade through the induction of a robust shift toward a more proinflammatory microenvironment, but without the development of systemic toxicity in the host. We previously showed similar effects of another immune modulatory vaccine based on epitopes from IDO with Montanide in a mouse setting (55). Promising clinical efficacy has been reported in a nonrandomized phase I/II study with IDO/PD-L1 peptide vaccines with Montanide, in combination with nivolumab, in patients with progressive metastatic melanoma (ClinicalTrials.gov; NCT03047928, abstract published; ref. 56). Hence, combining these peptide vaccines with checkpoint blockade therapy may be the new general way forward in treating many different cancer patients. We are currently examining the safety and immunologic effect of ARG1-derived vaccines with Montanide in two early vaccination trials at our institution (ClinicalTrials.gov: NCT03689192, NCT04051307).

M. Aaboe Jørgensen reports grants from Copenhagen University Hospital, Herlev, and Danish Council for Independent Research during the conduct of the study. E. Martinenaite reports grants from Innovation Fund Denmark during the conduct of the study. M. Zocca reports personal fees from Valo Therapeutics and nonfinancial support from Unikum Therapeutics outside the submitted work. V. Bronte reports grants and personal fees from IO Biotech during the conduct of the study. M.H. Andersen reports grants and personal fees from IO Biotech during the conduct of the study; grants and personal fees from IO Biotech outside the submitted work; and a patent for immunogenic arginase peptides pending to IO Biotech, a patent for arginase 1 polpypeptides pending to IO Biotech, and a patent for immunogenic arginase 2 polypeptides pending to IO Biotech. No disclosures were reported by the other authors.

The funders had no role in the study design, collection of data, data analysis, decision to publish, or manuscript preparation.

M. Aaboe Jørgensen: Data curation, formal analysis, investigation, writing–original draft, project administration. S. Ugel: Data curation, formal analysis, investigation, writing–review and editing. M. Linder Hübbe: Investigation, writing–review and editing. M. Carretta: Supervision, investigation. M. Perez-Penco: Investigation. S.E. Weis-Banke: Investigation. E. Martinenaite: Investigation, writing–review and editing. K. Kopp: Investigation. M. Chapellier: Investigation. A. Adamo: Investigation. F. De Sanctis: Investigation. C. Frusteri: Investigation. M. Iezzi: Investigation. M.-B. Zocca: Funding acquisition. D. Hargbøll Madsen: Data curation, supervision, writing–review and editing. A. Wakatsuki Pedersen: Supervision, writing–review and editing. V. Bronte: Supervision, funding acquisition, writing–review and editing. M.H. Andersen: Conceptualization, data curation, supervision, funding acquisition, writing–review and editing.

The authors thank Merete Jonassen for excellent technical support. This work was supported by Copenhagen University Hospital, Herlev, and the Danish Council for Independent Research.

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.