RNA sequencing data was obtained from a previously generated dataset [38]. RNA-seq reads were aligned to the mm9 Ensembl transcript annotation (release 65) using the PRADA pipeline (10.1093/bioinformatics/btu169), and FPKM expression values were determined using Cufflinks [39] with mm9 RefSeq gene annotations. FPKM values were log2-transformed and then used to calculate p values.

Triple immunofluorescence (3plex IF) stains were carried in the Leica Bond-Rx fully automated staining platform (Leica Biosystems Inc., Norwell, MA). Slides were dewaxed in Bond™ Dewax solution (AR9222) and hydrated in Bond Wash solution (AR9590). Epitope retrieval for all targets were done for 30 or 20 min in Bond-epitope retrieval solution 1 pH6.0 (AR9661) or solution 2 pH9.0 (AR9640) as shown in Additional file 1 : Table S1. The epitope retrieval was followed with 10 min endogenous peroxidase blocking using Bond peroxide blocking solution (DS9800). The application order of the primary and secondary antibodies, dilutions are shown in Additional file 1: Table S1; between the stains the appropriate antigen retrieval (20 min) and peroxide blocking steps were inserted. Stained slides were counterstained with Hoechst 33258 (# H3569) and mounted with ProLong® Diamond Antifade Mountant (#P36961) Life Technologies (Carlsbad, CA). Positive and negative controls (no primary antibody) and single stain controls were done for 3plex IF when one primary antibody was omitted to make sure that cross reactivity between the antibodies did not occur.

ID8 cells were provided by Gordon Freeman (DFCI). Cells were cultured in DMEM with 10% FBS (Heat inactivated). The rest of the cell lines were received from ATCC and cultured in their recommended media. Splenocytes were cultivated in SILAC RPMI (Thermo # 89984) media supplemented with: 10% Heat Inactivated FBS, Sodium pyruvate 1 mM, Glutamine 1 mM, Penicillin (50 U/ML) Strptomycin (50μg/ml), 2-Mercaptoethanol; 0.00005 M, (Sigma M7522), IL7: 5 ng/ml (PeproTech, #217–17 murine IL7), IL15: 100 ng/ml (PeproTech, #210–15 murine IL15), N-acetyl L Cysteine: 0.5 M (Sigma A9165), Lysine 218 uM and Arginine: 75 uM (or as indicated). Recombinant arginase was purchased from (R&D Biosystems, 5868AR010). Purified anti-mouse CD3 (cat# 100202) and purified anti-mouse CD28 (cat# 102102) antibodies were purchased from BioLegend. IFNg ELISA was performed using Biolegend (Mouse IFN-γ ELISA MAX).

Splenocytes were prepared by standard techniques. Briefly, spleen was minced with syringe plunger in cell culture dish with 5 mL of splenocyte medium. The splenocytes were rinsed filtered through 40um strainer and resuspended thoroughly with 10 mL PBS + 5% FBS. Cells were spun at 400 g for 5 min. The cell pellet was resuspended in 5 mL 1X RBC Lysis Buffer and incubated at 37 C for 5 min. Lysis was stopped by adding ~ 10 mL PBS, then spinning for 5 min at 400 g. Cells were washed in 5 mL PBS, and spun for 5 min at 400 g, and resuspended in splenocyte media. IFNg secretion was stimulated by treatment with CD3 (5μg/ml) and CD28 (1μg/ml) antibodies. Sorted CD11b + myeloid cells or IL4/PGE2 treated peritoneal macrophages were added to the incubation mix when indicated. Supernatants were collected 24 h after cell culture/compound addition, and 10 to 20 ul of the sample (supernatants) was evaluated by ELISA. CD11b + myeloid cells were sorted from ID8 tumor bearing mouse ascites. Cells were isolated from ascites fluid and stained with CD45 and Cd11b antibodies. CD45+ and CD11b + cells were sorted by flow cytometry. Peritoneal macrophages were isolated by standard techniques from wild type mice after treatment with 3% thioglycolate (Sigma, T0632) for 4 days. PM were treated with IL4 (50 ng/ml) and pGE2 (Sigma P0409) (2.6 uM) for indicated times in RPMI medium (2% FBS).

Compound 9 (PubChem CID: 66833213) was previously described by another group [40]. We purchased the compound from Wuxi LabNetwork (Shanghai, China) for in vivo studies and from MedChemExpress for in vitro studies. Compound 9 was dissolved in DMSO and used in the IC50 assays in the indicated doses. Cell titer glo (Promega) was used per the manufacturer’s instructions to determine cell viability. Cisplatin was purchased from UT Southwestern Medical Center pharmacy and diluted in saline, and stausporine was purchased from Cell Signaling Technologies.

Total protein lysates were prepared with NP40 lysis buffer (Boston BioProducts) with Protease inhibitors (Complete, Roche). Proteins were quantitated and 50 μg protein was resolved on SDS gel. Proteins were then transferred to PVDF membrane overnight in cold room. Blot was incubated with blocking buffer (5% milk in 1X TBST) for one hour at room temperature and then with Arginase 1 (Sigma, SAB2108087) and tubulin antibodies (Cell Signaling, #2148) overnight in cold room on a shaking platform. After washing 3 times with 1X TBST (each 10 min at room temperature), HRP conjugated secondary antibody was added to the blot for another hour at room temperature. The blot was again washed 3 times with 1X TBST, then ECL reagent was used to reveal the signals on Fluro Chem (Protein Simple). Arg1 mRNA expression information was obtained from the Cancer Cell line Encyclopedia (CCLE) [41].

Mice were treated with compound 9 at 30 mg/kg (dissolved in water) daily, by oral gavage. MRI was performed using the 7 Tesla Bruker MRI machine at the Lurie Family Imaging Center of DFCI. Tumor volumes were quantified using 3D Slicer software. All mouse experiments were performed with the approvals of IACUC at the Dana-Farber Cancer Institute and UT Southwestern Medical Center. Mice were of mixed 129/Sv, C57Bl/6, and BALB/c background. Immune analysis of mouse tumor tissues were performed by generating single cell suspension of tumors by collagenase treatment, lysing red blood cells, and then staining with the pertinent antibodies. Antibodies used are listed in Additional file 1: Table S2 and detailed tissue processing and staining protocol is described in Additional file 1: supplementary methods.

To evaluate the expression of arginase in different cellular compartments in lung tumors, we sorted tumor associated macrophages (CD11c+ CD11b- CD103-) and tumor associated neutrophils (CD11b+, LY6G+) from the tumors of a classical mouse model for lung adenocarcinoma-(KRAS^G12D GEMM) and isolated RNA. Combination of these two populations add up to approximately 80% of all CD45+ cells in the tumor bearing lungs in our model. Lymphocytes are about 15% of all CD45+ cells, leaving only 5–7% of CD45+ to the rarer myeloid cell types. Since the majority of the myeloid cell population fall into these two major groups, we focused our analysis on these subsets (Additional file 2: Figure S1).

We performed RNA sequencing and validated the cell type specific markers, Cd11b, Cd11c, Cd45 and Epcam [38]. Based on RNA-sequencing data, as compared to normal healthy control mice, neutrophils and macrophages from the tumors showed significantly increased expression of Arg 1 (Fig. 1a, p values 0.02 and 0.03 for macrophages and neutrophils respectively). Arg2 expression was in general higher than Arg1 in all the sorted samples and only significantly elevated in macrophages (p values are 0.01 and 0.07 for macrophages and neutrophils).Fig1Fig. 1

Arginase expression in the lung tumors. a RNA Sequencing analysis of sorted tumor (Epcam+), macrophage (CD11c+ CD11b- CD103-) and neutrophils (CD11b+, LY6G+) from Kras^G12D lung tumors. T denotes cells isolated from tumor carrying mice and c from healthy control littermates. Heatmaps denote Log2 of fragments per kilobase of transcript (FPKM) per each mouse sample. p values are 0.02 and 0.03 for macrophages and neutrophils for Arg1 and 0.01 and 0.07 for macrophages and neutrophils for Arg2. Log2 (FPKM) values were also used for statistical analysis by student’s T test. Heatmaps represent unscaled values. b High power image of ARG1, CD3, CD66b multiplex immunohistochemistry. c Low power representative images from ARG1, CD3, CD66b multiplex immunohistochemistry staining of the TMA for varying levels of Arg1 and CD3 staining. Left: Arg1 high CD3 low, middle: Arg1 and CD3 high, and right: Arg1 low CD3 high tumor sample. d Correlation of CD66b and ARG1 in lung tumor samples. P < 0.0001, Spearman correlation. e Automated quantification of the ARG1 + CD66b + density in adjacent matched-normals, LUAD, and LUSQ p < 0.0001, determined by Mann-Whitney test f Correlation of CD66b+ ARG1+ double staining and CD3 staining in NSCLC, p < 0.0264, Spearman correlation. G. Kaplan Meier survival analysis for patients whose tumors were CD66b+/ARG1+ high or low density based on multiplex IHC. p = 0.09 (Log Rank test). The thresholds set for fluorescence positive signals for CD66b and CD3 are higher than that for ARG1 therefore the quantity of the expression is relative not exact

Arginase inhibitors have been developed as tools to evaluate the contribution of arginine deprivation mediated immune-suppression to the tumor escape from the immune response. A recently developed arginase inhibitor previously displayed therapeutic efficacy in an in vivo model of myocardial ischemia/reperfusion injury [40]. This inhibitor (Cpd9) (R)-2-amino-6-borono-2-(2-(piperidin-1-yl)ethyl) hexanoic acid, inhibited enzymatic function human arginases I and II with IC50s of 223 and 509 nM, respectively, and was active in urea release cellular assay using chicken hamster ovary (CHO) cells overexpressing human arginase I (IC50 8uM). Our characterization of Cpd9 provided similar results in enzymatic assays (human ARG1 IC50 1.1uM; human ARG2 IC50 2.7 uM) and in a HepG2 urea release assay (IC50 8 uM) (Data not shown). Given the importance of arginase activity in cancer, we further characterized this compound in ex vivo assays designed to assess the effects of arginine deprivation in T-cell responses.

T-cell functional responses can be evaluated ex vivo by monitoring IFNγ secretion by mouse splenocytes. Arginine deprivation diminished T-cell functionality by impairing T-cell proliferation and reducing IFNγ secretion (Fig. 2a). In our studies, IFNγ secretion by mouse splenocytes was optimal when arginine levels were above 10uM, showing a significant reduction in media lacking arginine (Fig. 2a). Since the physiological concentrations of arginine are 75–150 uM [44], all IFNγ secretion assays were performed at concentrations in this range. In humans, arginine deprivation could also result from the action of secreted ARG1 to the microenvironment [45]. Addition of human ARG1 to mouse splenocytes reduced IFNγ secretion by ~ 90%, which was reversed by addition of 5–50 uM of Cpd9 (Fig. 2b and Additional file 2: Figure S5A). We have also confirmed the reversal of IL-2 and Granzyme B production with Cpd9 in these splenocytes (Fig. 2b and Additional file 2: Figure S5A).Fig2Fig. 2

Compound 9 reverses arginine depletion mediated interferon gamma production. a Arginine depletion prevented IFNg secretion by splenocytes. Splenocytes were activated with CD3/28 in the presence of media with arginine levels of 0, 2.5, 10, 40, 160 and 1000 uM. After 24 h IFNg was determined by ELISA in the culture supernatants b-d. Compound 9 mediated restoration of T cell function. Splenocytes were isolated from mixed background mice, dissociated and activated with CD3/28 antibodies in the presence of in the presence of arginine 40 uM. Recombinant Arginase (ARG1, 1 μg/ml) and Cpd9 (50 uM) were added to the indicated wells at the same time. b IFNg was determined by ELISA in the culture supernatants after 24 h c Granzyme B expression CD8 T cells were determined by flow cytometry d IL-2 expressing CD8 T cells were determined by flow cytometry. *p < =0.05, **p < =0.01, and ***p < =0.001. Graphs show representative results from experiments performed at least three times

Previous studies suggested arginase 2 expression and activity is important for breast and renal cancer cell growth in vitro by replenishing the intracellular polyamine pools and other mechanisms yet to be discovered [50, 51]. We surveyed ARG1 expression in panel of twelve human cancer cell lines from originating from different tissues to first determine whether cancer cells are dependent on arginase activity in vitro and then to determine whether tumor cell arginase expression correlates with sensitivity to Cpd9. In line with previous studies, we did not find any expression of arginase 1 across several human tumor cell lines by western blot (not shown). We detected arginase 2 in several lines by western blot (Additional file 2: Figure S7A). Compound 9 sensitivity of different human cell lines correlated with their arginase 2 expression (CCLE, Additional file 2: Figure S7B-D and data not shown).

To evaluate the cytotoxic activity of compound 9 specifically in lung cancer cell lines, we treated a panel of murine and human lung cancer cell lines with compound 9. In vitro treatment of syngeneic murine lung cancer line Lewis Lung Carcinoma (LLC), KP9–1, a cell line derived from concurrent Kras and p53 (Kras^G12D, p53^−/−) mutant mouse lung tumor, and human lung adenocarcinoma cancer cell lines A549 and H460 revealed no cytotoxic activity (Additional file 2: Figure S7D and data not shown).

To evaluate the contributions of immunosuppression mechanisms driven by arginine deprivation to evasion of the anti-tumor immune response in vivo, we studied the effect of Cpd9 in an immune-competent genetically engineered mouse model of Kras driven lung adenocarcinoma. Kras is the most common oncogene detected in lung cancer patients [5]. LSL-Kras G12D transgenic mice develop sporadic pulmonary adenocarcinomas following intranasal administration of Cre-expressing adenovirus vector to activate the latent oncogene Kras G12D allele. We induced tumor formation in the LSL-Kras^G12D and monitored tumor growth by magnetic resonance imaging (MRI). After confirming tumor growth, we initiated treatment studies. Upon administration of 30-mg/kg of Cpd9 daily to GEMMs G12D, we detected a reduction in tumor volume as early as one week measured by MRI (Figs. 4a, b). Upon extension of this treatment up to four weeks, we detected a significant difference in the tumor growth rate as compared to baseline between the Cpd9 vs vehicle treated animals (Figs. 4a, b). Cpd9's effects on tumor growth correlated with its pharmacodynamic effects on tumor and blood arginine levels. Cpd9 treatment significantly increased the levels of arginine in both tumor (p < 0.05) and blood (p < 0.001) (Fig. 4c). These results are consistent with Arginase inhibition preventing arginine deprivation, which in turn leads to reduction in tumor size in transgenic Kras G12D mice.Fig4Fig. 4

Inhibiting arginase activity has therapeutic efficacy in Kras mutant Gemms by creating an immune favorable microenvironment. a Representative MRI image of Kras mutant mouse lung tumors from mice treated with Compound 9 for one week. b Quantification of MRI images of lung tumors of mice either treated with vehicle or compound 9 for up to four weeks, each dot represents an individual mouse. p values for each of the 1,2 and 4 week time points are indicated on the graph c Measurement of arginine levels in the lung tumor tissue and blood of mice either treated with vehicle or compound 9 by metabolomics. d % CD3, CD4, and CD8 T cells in all CD45 cells, and e CD8 to regulatory T cell (FoxP3+) ratios in the lung tumor microenvironment of mice treated with vehicle or compound 9. determined by multicolor flow cytometry analysis. f Quantification of interferon gamma production by flow cytometry of ex vivo stimulated and enriched lymphocytes from the lung tumors in mice treated with vehicle or compound 9 for one week. All p values in this figure were determined by Student’s t test. Each data point in the graphs is from a different mouse