Possible Role of Arginase-1 in Concomitant Tumor Immunity

The expression of Adenovirus serotype 2 or serotype 5 (Ad2/5) E1A in tumor cells reduces their tumorigenicity in vivo by enhancing the NK cell mediated and T cell mediated anti-tumor immune response, an activity that correlates with the ability of E1A to bind p300. We determined if E1A could be used as a molecular adjuvant to enhance antigen-specific T cell responses to a model tumor antigen, ovalbumin (OVA). To achieve this goal, we stably expressed a fusion protein of E1A and OVA (MCA-205-E1A-OVA), OVA (MCA-205-OVA) or a mutant version of E1A unable to bind p300 and OVA (E1A-Δp300-OVA) in the B6-derived, highly tumorigenic MCA-205 tumor cell line. MCA-205-E1A-OVA tumor cells were over 10,000 fold less tumorigenic than MCA-205-OVA, MCA-205-E1A-Δp300-OVA, or MCA-205 in B6 mice. However, immunization of B6 mice with live MCA-205-OVA, MCA-205-E1A-Δp300-OVA and MCA-E1A-OVA tumor cells induced nearly equivalent OVA-specific CD4 T cells and CD8 CTL responses. Further studies revealed that mice with primary, enlarging MCA-205-OVA or MCA-205-E1A-Δp300-OVA tumors on one flank exhibited OVA-specific anti-tumor T cell responses that rejected a tumorigenic dose of MCA-205-OVA cells on the contralateral flank (concomitant tumor immunity). Next we found that tumor associated macrophages (TAMs) in progressive MCA-205-OVA tumors, but not MCA-205-E1A-OVA tumors that expressed high levels of arginase-1, which is known to have local immunosuppressive activities. In summary, immunization of mice with MCA-205 cells expressing OVA, E1A-Δp300-OVA or E1A-OVA induced equivalent OVA-specific CD4 and CD8 anti-tumor responses. TAMs found in MCA-205-OVA, but not MCA-205-E1A-OVA, tumors expressed high levels of arginase-1. We hypothesize that the production of arginase-1 by TAMs in MCA-205-OVA or MCA-205-E1A-Δp300-OVA tumor cells leads to an ineffective anti-tumor immune response in the tumor microenvironment, but does not result in inhibition of a systemic anti-tumor immunity.


Introduction
Expression of the Adenovirus E1A oncoprotein in primary cells results in cellular immortalization [1]. Cells stably expressing E1A and the helper protein E1B have been shown to be oncogenic in immunosuppressed rodents [2,3]. Paradoxically, in rodent models the expression of Adenovirus serotype 2 or serotype 5 (Ad2/5) E1A in tumor cell lines significantly reduces tumorigenicity [4] (we now refer to Ad2/5 E1A as simply E1A). The ability of E1A to reduce tumorigenicity is dependent on the induction of a robust NK cell and T cell antitumor immune response [5] and correlates with the ability of E1A to bind the transcriptional co-adaptor molecule p300 or CBP [6]. p300 and CBP are highly homologous co-activators of transcription with intrinsic histone-acetyl transferase activity and will hereafter be referred to as simply p300 [7]. The expression of E1A, but not mutant forms of E1A that do not bind p300 (E1A-Dp300), also upregulates NKG2D ligands [8] and sensitizes cells to lysis by macrophages, NK cells and immune effector molecules utilized by these cells [9][10][11][12][13].
Based on these anti-tumorigenic activities of E1A, we sought to determine if E1A could be used to enhance antigen specific, anti-tumor T cell responses to MCA-205 tumor cells that express a model tumor antigen, ovalbumin (OVA). MCA-205 tumor cells that expressed a fusion protein of E1A and OVA elicited an effective anti-tumor T cell response and were rendered non-tumorigenic. Surprisingly, immunization of mice with live MCA-205-OVA or MCA-205-E1A-Dp300-OVA tumor cells elicited a robust anti-tumor immune response, despite forming progressive tumors at the primary site of immunization (concomitant tumor immunity). Further studies examined a possible mechanism whereby immunization of B6 mice with MCA-205-OVA or MCA-205-E1A-Dp300-OVA could induce systemic anti-tumor immunity but fail to clear a local tumor burden.

Flow cytometry
Flow cytometry was performed with a LSR II (BD biosciences, San Jose, CA) using BD FACSDiva software. Flow cytometry analysis was performed using Flow Jo software (Tree Star, Ashland, OR). Antibodies specific to mouse CD3e

Cloning strategy
The wild-type Adenovirus 5 E1A gene was cloned from Adenovirus 5 (GenBank ID: AY147066.1) bp 44-596,713-1029 Forward primer: 59-CGT ACT GAA TTC TAA GGT ACC ATG GGC TCC ATC GGT GCA GC-39, Reverse primer: 59-GCT GCA CCG ATG GAG CCT GGC CTG GGG CGT TTA CAG CT-39 by PCR. A mutant version of E1A unable to bind p300 (E1A-Dp300) was cloned from the Adenovirus E1A mutant dl1104 using the same primers as E1A. E1A-Dp300 has a deletion in amino acids 48-60, which eliminates p300 binding [16]. The OVA gene was cloned from the pAC-OVA plasmid (GenBank ID: J00895.1), coding sequence (CDS) Forward The resulting OVA gene expresses a slightly truncated OVA protein from AA 3-383 that included the relevant H-2K b and I-A b epitopes. The truncated version of OVA was made, as we found it could readily be expressed in MCA-205 cells, whereas, we could not express full-length OVA 1-386 in MCA-205 cells (data not shown). The respective genes were cloned into pLXSN using XHOI and EcoRI restriction sites. The E1A-OVA and E1A-Dp300-OVA chimeric genes were generated through overlapping PCR using the following primers: E1A forward and reverse primer (listed above) plus the E1A-OVA primer 59-AGC TGT AAA CGC CCC AGG CCA GGC TCC ATC GGT GCA GC-39 and OVA reverse: 59-CGT ACT CTC GAG TTA AGG GGA AAC ACA TCT-39. The chimeric gene was inserted into pLXSN using EcoRI and Xho-I (New England Biolabs, Ipswich, MA) restriction sites. DNA sequencing by the Human and Molecular Genetics Sequencing core at the Medical College of Wisconsin verified the fidelity of OVA, E1A-OVA, and E1A-Dp300-OVA genes.

Tumor induction studies
Quantitative tumor induction studies were performed as previously described [8]. Briefly, mice were administered serial log dilutions of tumor cells subcutaneously (s.c.) in the flank and monitored for tumor growth twice weekly using digital calipers. Animals were euthanized by CO 2 followed by cervical dislocation when tumors reached 20 mm in diameter, if tumors ulcerated, or at the end of a 12-week monitoring period. Tumor producing dose 50 (TPD 50 ) values, which are the log 10 of the number of cells required to form tumors, were calculated using the Spearman-Karber Formula.
Quantification of OVA specific CD8 and CD4 T cells OT-I transgenic CD8 T cells were harvested from the spleens of OT-I + RAG 2/2 CD45.1 + mice and OT-II transgenic CD4 T cells were harvested from the spleens of OT-II + RAG 2/ 2 CD45.1 + mice. 1610 5 CD45.1 + OT-I CD8 T cells or 1610 6 CD45.1 + OT-II CD4 T cells were administered i.v. via retro-orbital injection into B6 mice. The following day, mice were administered 1610 5 live MCA-205-OVA, MCA-205-E1A-OVA or MCA-205-E1A-Dp300-OVA cells s.c. in the hock (the lateral tarsal region just above the ankle). Five days (OT-I) or nine days (OT-II) following tumor injection the popliteal lymph nodes were removed and the CD45.1 + OT-I or OT-II T cells were quantitated by flow cytometry by staining for CD45.1 + CD3 + CD8 + T cells or CD45.1 + CD3 + CD4 + T cells, respectively. The absolute number of cells was determined by multiplying the percentage of the target cell population by the total number of cells in the lymph node.
In vivo CTL assay B6 mice were primed with 1610 6 live tumor cells s.c. in the flank. Seven days later an in vivo CTL assay was performed on the primed mice [18]. B6 splenocytes pulsed with OVA 257-264 peptide were used as targets. OVA pulsed splenocytes were labeled with a low dose (1 mM) of CFSE for one minute in 5% FBS PBS. Untreated splenocytes were labeled with a high dose (10 mM) CFSE for one minute in 5% FBS PBS. The two CFSE labeled target splenocytes groups were mixed equally and injected i.v. into primed mice. 10610 6 total target cells were administered to the mice. Four hours later, spleens were removed and splenocytes were analyzed by flow cytometry for CFSE expression. The ratio of OVA pulsed splenocytes (low CFSE) to unpulsed splenocytes (high CFSE) was used to determine specific killing. Specific killing was calculated as follows: Specific lysis = 1-R Naive /R exp X 100; R = % OVA pulsed/% non-pulsed. Immune infiltrate of MCA-205-OVA tumors B6 mice were administered 1610 6 live MCA-205-OVA cells s.c. in the flank. When the tumor reached a diameter of 10 mm (, two weeks) tumors were excised. Tumors were minced with scissors and digested with collagenase Type I 5 U/mL (Sigma-Aldrich, St Louis, Missouri), Deoxyribonuclease I 50 U/mL (Sigma-Aldrich, St Louis, Missouri), and Hyaluronidase, Type II 5 U/mL (Sigma-Aldrich, St Louis, Missouri) in 10 mL of RPMI-5 at 37uC for two hours. The digestion was stopped by the addition of 5 mL of 10 mM EDTA and incubated at 37uC for 15 minutes. A single cell suspension was generated by crushing the tissue with glass slides and passing through 40 mm filters. Cells were then characterized by flow cytometry. Tumor associated macrophage arginase expression Tumors were excised from mice and a single cell suspension was generated as described above. Macrophages were purified using FACS and lysed in.4% Triton-X lysis buffer at a concentration of 1610 5 cells per 100 mL of lysis buffer. Arginase expression was quantified using the QuantiChrom Arginase Assay kit (Bioassay systems, Hayward, CA), following manufacturer's directions.

Statistics
Statistical differences between groups were calculated using the ANOVA and Dunnett multiple comparison tests. Mann-Whitney tests were used to compare two sets of data. Survival curves were compared using the log-rank (Mantel Cox) test; p values less than.05 were considered significant. Unless otherwise stated, all data are presented as mean 6 standard error of mean (SEM). Statistical analysis was done using Prism version 5.0a software (GraphPad Software, La Jolla, CA). (*, p,0.05; **, p,0.01; ***, p,0.001).

Results
Expression of E1A-OVA, OVA or E1A-Dp300-OVA in MCA-205 cells In an effort to determine if E1A could augment anti-tumorspecific T cell responses to OVA, we first generated MCA-205 tumor lines (see Materials and Methods) that expressed a slightly truncated OVA 3 (Figure 1 B). OVA was detected at the expected molecular weight of approximately 43 kDa, whereas E1A-OVA and E1A-Dp300-OVA bands were detected at significantly higher molecular weights due to the addition of the E1A or E1A-Dp300 proteins.
Next, we confirmed that the OVA protein was processed and presented by the class I molecule H-2K b by staining the various cell lines with the 25-D1. 16 (Figure 1 C). These data demonstrate that MCA-205 lines were generated that stably express E1A and/or OVA and that the OVA protein is processed and presented on the cell surface.     TAMs from MCA-205-E1A-OVA tumors express significantly less arginase than TAMs from MCA-205-OVA tumors Our data suggests that the systemic anti-tumor T cell response was similar following challenge with tumorigenic MCA-205-OVA cells and non-tumorigenic MCA-205-E1A-OVA cells. Therefore, we investigated the MCA-205-OVA tumor microenvironment at the primary site of inoculation in an effort to determine why these mice failed to reject primary MCA-205-OVA tumors. We first quantitated the types of immune cells (macrophages, CD4 and CD8 T cells, NK cells and myeloid-derived suppressor cells) present in the MCA-205-OVA tumors by flow cytometry (Figure 7). We found that macrophages were the predominant inflammatory cell in MCA-205-OVA tumors contributing to nearly 55% of the CD45 + cells; this was nearly four-fold higher than CD8 T cells, the next most frequent immune cell in tumors (Figure 7). CD4 T cells (2%) and regulatory T cells (,1%) were notable for their relative absence in progressive MCA-205-OVA tumors.
Tumor associated macrophages (TAMs) are known to be protumorigenic and known to actively suppress the immune system through the production of suppressive immune mediators [20][21][22]. One major mechanism whereby TAMs suppress anti-tumor immune responses is through the production of arginase-1. Arginase-1 expression by TAMs has been associated with the suppressive tumor microenvironment and shown to inhibit antitumor T cell responses [23][24][25]. Importantly, arginase-1 is usually only suppressive in the tumor microenvironment locally and does not lead to systemic immunosuppression. Because our data showed that the systemic anti-OVA T cell response in mice with MCA-205-OVA tumors was intact, arginase-1 production by TAMs was investigated. To compare arginase-1 expression between TAMs from MCA-205 or MCA-205-E1A tumors, we grew MCA-205-OVA tumors in B6 or B6 RAG 2/2 mice and purified TAMs by FACS. MCA-205-E1A-OVA cells are non-tumorigenic in WT B6 mice; therefore, RAG 2/2 mice were used to generate MCA-205-E1A-OVA tumors. Lysates from the different TAMs were analyzed for arginase-1 activity (Materials and Methods). These results showed that TAMs from MCA-205-OVA tumors grown in both WT and RAG 2/2 mice produced large amounts of arginase-1 ( Figure 8). Similar results were obtained with MCA-205 tumor cells (data not shown). In contrast, TAMs from MCA-205-E1A-OVA tumors produced negligible amounts of arginase-1 in RAG 2/2 mice.
High arginase expression by TAMs is associated with low Larginine levels within the tumor. L-arginine is a critical amino acid for T cells, and one of the effects of low arginine levels on T cells is the reduction of CD3e on the cell surface [25]. Therefore, we compared surface expression of CD3e on T cells found in the tumor, the draining lymph node and the spleen of MCA-205-OVA tumor bearing mice. Our results show that CD8 T cells from MCA-OVA tumors expressed significantly less surface CD3e than CD8 T cells from the spleen, but were not significantly different than CD8 T cells from the draining lymph node (Figure 9 A, B). CD4 T cells exhibited no change in the surface expression of CD3e between the tumor, draining lymph node and spleen (Figure 9 C). Collectively, these data are consistent with the hypothesis that arginase-1 producing TAMs present in the tumor microenvironment induce local, but not systemic, suppression of anti-tumor immunity following injection of MCA-205-OVA tumor cells. In contrast, E1A expression in MCA-205 cells inhibits arginase-1 expression by TAMs allowing for an effective local antitumor immune response.

Discussion
In this study we used OVA as a model tumor antigen to evaluate the ability of E1A expression to augment antigenspecific anti-tumor T cell responses in mice. We found that the expression of an E1A-OVA fusion protein rendered MCA-205 cells essentially non-tumorigenic in normal B6 mice. In contrast, the tumorigenicity of MCA-205 cells was not substantially changed by the expression of OVA or E1A-Dp300-OVA in normal B6 mice. Immunization with either MCA-205-OVA, MCA-205-E1A-Dp300-OVA, or MCA-205-E1A-OVA tumor cells induced a robust OVA-specific anti-tumor T cell response. For example, following injection of either MCA-205-OVA, Based on the observation of concomitant tumor immunity in the presence of primary tumor formation, we investigated the tumor microenvironment. We found that TAMs were the predominant inflammatory cell in progressive MCA-205-OVA tumors in WT B6 mice. TAMs isolated from MCA-205-OVA tumors, in either WT B6 mice or B6 RAG2/2 mice, expressed large amounts of arginase-1. In contrast, TAMs from MCA-205-E1A-OVA tumors from B6 RAG2/2 mice expressed negligible amounts of arginase-1. TAMs are often phenotypically similar to alternatively activated macrophages, producing high levels of arginase-1, which depletes the local environment of L-arginine [26]. T cells in an environment depleted of L-arginine express lower levels of CD3e on the cell surface, reduced total levels of CD3f, become arrested in the G 0 -G 1 growth phase and display a decrease in global protein translation [25,27]. We found that CD8 T cells which had infiltrated MCA-205-OVA tumors had significantly lower surface CD3e than CD8 T cells from the spleen of MCA-205-OVA tumor bearing mice, a finding that is consistent with high arginase-1 activity. Collectively, our results are consistent with the hypothesis that E1A expression in tumors may preserve antigen-specific antitumor T cell responses in the local tumor environment by inhibiting the production of arginase-1 by TAMs, a biological activity of E1A not previously described. Alterations in arginase-1 Future studies need to directly examine the role of arginase-1 production by TAMs, concomitant tumor immunity and the anti-tumorigenic effect of E1A. Definitive studies examining the role of arginase-1 production by TAMs and anti-tumor immunity are currently hampered by a relative lack of arginase-1 specific reagents. The most commonly used in inhibitor of arginase is N(v)-hydroxy-nor-L-arginine (nor-NOHA). Nor-NOHA inhibits both arginase-1 and arginase-2 and in conventionally used doses the inhibition of arginase-1 in vivo is less than 50% [28]. Mice genetically deficient in arginase-1 [29] or conditionally-induced to be deficient in arginase-1 [30], succumb to an illness that mimics human arginase deficiency far too rapidly to perform tumor induction experiments examining concomitant tumor immunity. The development of mice conditionally deficient of arginase-1 in macrophages or better arginase-1 inhibitors will facilitate progress in this important area.
The mechanism whereby tumor cells that express E1A lead to the decreased expression of arginase-1 by TAMs is unknown and also needs to be explored. Arginase-1 activity has been postulated as one of the mechanisms for the failure of adoptive cellular immune therapy to be effective against solid tumors. Determining how E1A is able to inhibit tumor cells from inducing arginase-1 in TAMs could have important implications in augmenting local anti-tumor immune responses in the setting of progressive tumor enlargement. Additionally, novel uses of E1A could be considered to augment local anti-tumor immune responses. For example, studies could be performed to determine if administration of liposomes with E1A protein into a tumor could augment local tumor anti-immune responses and lead to tumor rejection.
Finally, the use of the MCA-205-OVA, MCA-205-E1A-Dp300-OVA, or MCA-205-E1A-OVA tumor lines in the B6 mouse is a new model of concomitant tumor immunity that also allows for the quantitation of antigen-specific T cell responses. In the B16 melanoma concomitant tumor immunity model developed by Turk et. al. [31], concomitant tumor immunity was observed only after manipulating B16 melanoma cells to express GM-CSF, or by depleting/inhibiting regulatory T cells. In our MCA-205-OVA model, concomitant tumor immunity occurred by day 20, and required no further experimental manipulation of the tumor cells or B6 mice. To our knowledge, this model of concomitant tumor immunity is unique and may more accurately replicate ongoing systemic anti-tumor immune responses in a human in the face of enlarging primary or secondary tumors.