The enhanced expression of T cell Ig and mucin protein-3 (TIM-3) on tumor-associated dendritic cells (DCs) attenuates antitumor effects of DNA vaccines. To identify a potential target (or targets) for reducing TIM-3 expression on tumor-associated DCs, we explored the molecular mechanisms regulating TIM-3 expression. In this study, we have identified a novel signaling pathway (c-Src→Bruton’s tyrosine kinase→transcription factors Ets1, Ets2, USF1, and USF2) necessary for TIM-3 upregulation on DCs. Both IL-10 and TGF-β, which are produced in the tumor microenvironment, upregulated TIM-3 expression on DCs via this pathway. Suppressed expression of c-Src or downstream Bruton’s tyrosine kinase, Ets1, Ets2, USF1, or USF2 blocked IL-10– and TGF-β–induced TIM-3 upregulation on DCs. Notably, in vivo knockdown of c-Src in mice reduced TIM-3 expression on tumor-associated DCs. Furthermore, adoptive transfer of c-Src–silenced DCs in mouse tumors enhanced the in vivo antitumor effects of immunostimulatory CpG DNA; however, TIM-3 overexpression in c-Src–silenced DCs blocked this effect. Collectively, our data reveal the molecular mechanism regulating TIM-3 expression in DCs and identify c-Src as a target for improving the efficacy of nucleic acid–mediated anticancer therapy.
In view of the initial discovery of T cell Ig and mucin protein-3 (TIM-3) as a Th1-specific cell surface protein (1), earlier studies primarily focused on how TIM-3 regulates Th1 responses (2, 3). However, deciphering the role of TIM-3 in innate immunity has recently gained considerable interest due to the fact that TIM-3 is also expressed on innate effectors, including dendritic cells (DCs) (4, 5). In DCs, TIM-3 acts as a critical regulator of antitumor immunity (6, 7), but the molecular mechanisms that regulate TIM-3 expression remain obscure. Although the transcription factors T-bet in T cells and SMAD2 and SMAD4 in a mast cell line have been proposed to regulate HAVCR2 (which encodes TIM-3) expression (8, 9), only T-bet has been shown (in T cells) to bind to the HAVCR2 promoter (8). In the case of DCs, reports on transcription factors regulating HAVCR2 expression are much more limited. As yet, only T-bet has been implicated in the regulation of HAVCR2 mRNA expression; however, T-bet deficiency does not affect TIM-3 protein expression in DCs (8). Furthermore, whether T-bet binds to the HAVCR2 promoter in DCs is not known. Accordingly, the events regulating HAVCR2 expression in general and specifically in DCs are ill defined. Moreover, because TIM-3 impacts DC-mediated antitumor immunity, the identification of transcription factors regulating HAVCR2 expression in DCs may reveal potential new targets for cancer treatment.
In the field of cancer immunotherapy, DNA vaccination has emerged as a promising therapeutic approach (10). A recent study showed that the antitumor efficacy of DNA vaccines is limited by elevated TIM-3 expression on tumor-associated DCs (TADCs) (6). However, the potential molecular target for reducing TIM-3 expression on TADCs remains unidentified. Accordingly, we set out to unravel the molecular mechanism regulating TIM-3 expression in DCs. In this study, we provide evidence for a novel pathway in which the immunosuppressive cytokines IL-10 and TGF-β, found in the tumor milieu (11), upregulate TIM-3 expression on DCs. Specifically, we demonstrated that both IL-10 and TGF-β upregulated TIM-3 surface expression on DCs via a common signaling pathway that involved sequential activation of c-Src and Bruton’s tyrosine kinase (Btk; a member of the Tec nonreceptor tyrosine kinase family), leading to the recruitment of Ets and USF transcription factors to the HAVCR2 promoter. We also showed that silencing of the proximal mediator c-Src reduced TIM-3 expression on TADCs and significantly improved the in vivo antitumor effects of immunostimulatory CpG DNA in animal models. These findings provide a valuable insight for improving the efficacy of nucleic acid–mediated antitumor therapy.
Materials and Methods
The following Abs and reagents were purchased from Santa Cruz Biotechnology: anti-Ets1, anti-Ets2, anti-USF1, anti-USF2, anti–TGF-β type I receptor (TGF-βRI), anti–IL-10 receptor 1 (IL-10R1), anti–β-actin, HRP-conjugated anti-rabbit IgG, and protein A/G PLUS-Agarose beads. Abs to c-Src, Btk, p-c-Src (Tyr416), and p-Btk (Tyr223) were obtained from Cell Signaling Technology. For immunoprecipitation and chromatin immunoprecipitation (ChIP) assays, rabbit IgG was purchased from R&D Systems. Neutralizing anti–IL-10 and anti–TGF-β Abs were from eBioscience and R&D Systems, respectively. PE-conjugated anti-mouse TIM-3 and anti-human TIM-3 and allophycocyanin-conjugated anti-mouse CD11c were from BioLegend. The following isotype-matched control Abs were used: mouse IgG1 (R&D Systems); rat IgG1 (BioLegend); and PE-conjugated rat IgG1, rat IgG2, and mouse IgG1 (all from BioLegend). Recombinant mouse GM-CSF and IL-4 were from PeproTech; mouse IL-10, human IL-10, and human TGF-β were from R&D Systems; and human GM-CSF and IL-4 were from Miltenyi Biotec. Anti-mouse CD11c MicroBeads and anti-human CD14 MicroBeads were obtained from Miltenyi Biotec. The SMARTpool small interfering RNAs (siRNAs) targeting mouse and human c-Src, Btk, Ets1, Ets2, USF1, or USF2 were obtained from Dharmacon. Nontargeting scrambled control siRNA was purchased from Santa Cruz Biotechnology.
All animal experiments were done with the approval of the Institutional Animal Ethics Committee of Institute of Microbial Technology. For human monocyte-derived DC (HuMoDC) preparation, the buffy coats of healthy donors were obtained from Fortis Hospital (Mohali, India), with approval of the Biosafety Committee of Institute of Microbial Technology and the Ethics Committee of Fortis Hospital. Informed consents were obtained from all blood donors.
BALB/c and C57/BL6 (B6) mice were maintained at the Institute of Microbial Technology animal facility and used at 8–12 wk of age.
Tumor cell culture and ELISA
MC-38 (gifted by G. Shurin; University of Pittsburgh) and CT26-wt (American Type Culture Collection) tumor cells (1 × 106 cells/ml) were cultured as described (12). After 48 h, supernatants were assayed for IL-10 and TGF-β using mouse IL-10 and TGF-β ELISA kits (eBioscience) and subsequently used for DC treatment.
DC preparation and treatment
Mouse bone marrow–derived DCs (BMDCs), splenic DCs (sDCs), and HuMoDCs were prepared as described (13). BMDCs and sDCs (5 × 106 cells/well) were treated with mouse IL-10 (25 ng/ml) or human TGF-β (10 ng/ml), and HuMoDCs (5 × 106 cells/well) were treated with human IL-10 (20 ng/ml) or TGF-β (40 ng/ml) in complete RPMI 1640 medium (10% FBS, penicillin/streptomycin, l-glutamine, and 2-ME). In some experiments, DCs were incubated with MC-38 or CT26-wt tumor cell supernatants (50% of total medium) in the presence or absence of 10 μg/ml neutralizing anti–IL-10 or anti–TGF-β Ab or respective isotype controls (rat IgG1 for anti–IL-10 and mouse IgG1 for anti–TGF-β) .
Primer extension analysis
Primer extension analysis was done as described (14) using mouse or human HAVCR2-specific [32P]-end–labeled antisense primers, the sequences and locations (relative to the translation initiation sites) of which were as follows: mouse HAVCR2, primer A (+9 to −21) 5′-TGAAAACATGAGTACTTGGCAGGGGAAATC-3′ and primer B (−11 to −37) 5′-CAGGGGAAATCCAAGGACAGCTCTGTG-3′; human HAVCR2, primer C (−57 to −76) 5′-CAAATGGACTGGGTACTTCT-3′, and primer D (−72 to −92) 5′-CTTCTTCCAACTGTCTACTCC-3′. The extracted primer extension products were run on a 10% sequencing gel along with the appropriate sequencing ladders.
The cDNA synthesis and quantitative RT-PCR analyses were performed using the SuperScriptIII Platinum SYBR Green one-step qRT-PCR kit (Invitrogen) and gene-specific primers (Supplemental Table I). Expression was quantified by the change in threshold method (ΔΔCT) and normalized to the Actb mRNA (encoding β-actin) expression.
In vivo footprinting
Nuclear extracts were prepared as described (16). EMSA was performed using various 32P-labeled DNA probes (Supplemental Table I) specific for mouse and human HAVCR2 promoters. An OCT-1 probe, 5′-TGTCGAATGCAAATCACTAGAA-3′, was used as control. Bands were visualized using a phosphoimager (Fujifilm FLA-9000; Fujifilm).
ChIP assays were performed with rabbit IgG or Abs to Ets1, Ets2, USF1, or USF2 using the ChIP-IT kit (Active Motif). Enrichment of specific DNA fragments was measured by quantitative PCR using primers mentioned in Supplemental Table I. Results were normalized to ChIP with rabbit IgG (control) and input DNA.
Immunoprecipitation, DNA pulldown assay, and immunoblot analysis
DCs were lysed with the cell lysis buffer (Cell Signaling Technology). Immunoprecipitation and immunoblot analysis were performed as described (13). DNA pulldown assay of nuclear extracts of DCs using biotinylated oligonucleotides (Supplemental Table I), followed by immunoblot analysis was done as described (17).
Densitometry analysis was performed using Scion Image software (Scion Corporation).
Custom GLuc-ON dual-reporter constructs containing wild-type mouse HAVCR2 promoter fragment [−1298 to +43 region; HAVCR2(Wt)pro] or similar promoter fragment containing mutated Ets- or USF-binding site [HAVCR2(MutEts)pro and HAVCR2(MutUSF)pro, respectively] were obtained from GeneCopoeia. These reporter plasmids encoded the secreted Gaussia luciferase under control of the HAVCR2 promoter and the secreted alkaline phosphatase (SEAP) under control of the CMV promoter. RAW264.7 mouse macrophage cells (1 × 106) were transfected in triplicates with either of the above-mentioned reporter constructs (800 ng) along with 800 ng mammalian expression vectors encoding Ets1 (pcDNA3.1-Ets1), Ets2 (pcDNA3.1-Ets2), USF1 (pCMV-USF1), or USF2 (pCMV-USF2) or respective empty vector using the TransIT-2020 transfection reagent (Mirus). At 36 h after transfection, supernatants were assayed for the activities of Gaussia luciferase and SEAP using the Secrete-Pair Dual Luminescence Assay kit (GeneCopoeia). The Gaussia luciferase activity was normalized to the activity of SEAP. The Ets1 and Ets2 constructs were provided by T.M. Nowling (Medical University of South Carolina) and USF1 and USF2 constructs by M.D. Galibert (Univèrsité Rennes, Rennes, France).
DCs were transfected with 60 nmol siRNA using Lipofectamine RNAiMAX reagent (Invitrogen).
Delivery of vivo-morpholinos and assessment of TIM-3 expression on TADCs
B6 and BALB/c mice were injected s.c. in right flank with MC-38 and CT26-wt tumor cells (1 × 106 cells/mouse), respectively. Mice were then injected i.v. with nontargeting control vivo-morpholino (Ctrl vivo-MO, 5′-CCTCTTACCTCAGTTACAATTTATA-3′) or c-Src–targeting vivo-morpholino (c-Src vivo-MO, 5′-CTTGCTCTTGTTGCTGCCCAT-3′; Gene Tools) at 12.5 mg/kg body weight or with PBS for 5 consecutive d (day 21–25 posttumor inoculation). As a control, tumor-free mice were injected with PBS at corresponding time points. After 26 d of tumor inoculation, single-cell suspensions were prepared from spleens and established tumors. Viable cells were enriched using a dead cell removal kit (Miltenyi Biotec). TIM-3 expression on sDCs and TADCs was assessed via flow cytometry after gating on CD11c+ population.
Tumor tissues were excised and then fixed overnight in 4% paraformaldehyde (Sigma-Aldrich). Sectioning of paraffin-embedded tissues and immunohistochemical staining were done at IMGENEX India (Bhubaneswar, India). Expression of c-Src in tumor tissue sections was detected by immunohistochemical staining with anti–c-Src Ab (1:50; Sigma-Aldrich) using MACH 1 detection kit (Biocare Medical) following the manufacturer’s protocol. Sections were counterstained with hematoxylin (Vector Laboratories).
DC transfer experiments
B6 or BALB/c BMDCs transfected with control siRNA or c-Src–specific siRNA were injected intratumorally (1 × 106 DCs/mouse) into MC-38 or CT26-wt tumor-bearing syngeneic mice (on days 12, 13, 15, and 16 posttumor cell inoculation). Mice were then injected intratumorally with control oligodeoxynucleotide (Ctrl-ODN) or CpG-ODN (30 μg/mouse; Invivogen) on days 13, 14, and 16 after tumor inoculation. In some experiments, prior to intratumoral administration, c-Src siRNA-treated BMDCs (1 × 106) were transfected with 2 μg empty vector or TIM-3–expressing vector (B6 TIM-3/pMKITneo or BALB/c TIM-3/pMKITneo; gifted by H. Akiba, Juntendo University, Tokyo, Japan) using the TransIT-2020 transfection reagent (Mirus). Tumor volume was measured as described (18). Expression of IL-12 in tumor lysates was assessed using mouse IL-12 ELISA kit (eBioscience) on day 28 after tumor inoculation.
Flow cytometry was performed with a C6 Accuri flow cytometer (BD Biosciences). Data were analyzed with FlowJo software (Tree Star).
One-way ANOVA (SigmaPlot 11.0 program) was used for all statistical analyses. A p value <0.05 was considered significant.
Tumor cell–secreted IL-10 and TGF-β upregulate TIM-3 expression on DCs
Because IL-10 and TGF-β are generally expressed in the tumor microenvironment (11), we first assessed whether tumor cell–derived IL-10 and TGF-β influenced TIM-3 expression by DCs. For these experiments, we used the colon carcinoma cell lines MC-38 and CT26-wt (19, 20), which secrete high levels of IL-10 and TGF-β (Fig. 1A). We treated BMDCs, established from B6 mice, with culture supernatants from MC-38 tumor cells (H2Kb) and assessed TIM-3 expression. Whereas treatment with MC-38 culture supernatants upregulated TIM-3 expression on B6 BMDCs, the addition of IL-10– or TGF-β–neutralizing Ab diminished TIM-3 induction (Fig. 1B). TIM-3 expression was also upregulated on BALB/c BMDCs in a TGF-β–dependent manner by CT26-wt cell (H2Kd) supernatants (Fig. 1C). Furthermore, DC stimulation with rIL-10 or TGF-β was sufficient to increase TIM-3 expression at both mRNA and protein levels (Fig. 2). Cotreatment with IL-10 and TGF-β had no additive or synergistic effect on TIM-3 expression (data not shown). Because MC-38 and CT26-wt tumor cell supernatants contain several other factors as well as IL-10 and TGF-β (6, 21, 22), we used rIL-10 and TGF-β in subsequent experiments. These results demonstrate that TIM-3 expression by DCs is upregulated by tumor cell–derived IL-10 and TGF-β.
TIM-3 upregulation on DCs requires Ets and USF
As an initial step to determine the transcription factors regulating HAVCR2 expression, we sought to define the transcriptional start sites for mouse and human HAVCR2. Primer extension analysis using two different primers mapped the transcription start site of the mouse HAVCR2 to a C residue and of the human HAVCR2 to a G residue, which were located 52 and 115 nt upstream of the translational start codon (ATG), respectively (Fig. 3). Hence, we designated the respective C and G residues of the mouse and human HAVCR2 as +1 nt.
Given that both cytokines upregulated TIM-3 expression by DCs, we investigated whether IL-10 and TGF-β induced binding of any common transcription factor (or factors) to the proximal region of the HAVCR2 promoter. In vivo footprint analysis revealed DMS-protected G residues at positions −103 and −104 on the coding strand and at positions −106 and −117 on the noncoding strand of the mouse HAVCR2 promoter in both IL-10– and TGF-β–treated BMDCs but not in untreated BMDCs (Fig. 4A, top). Examination of the sequences encompassing these DMS-protected G residues with the TFSEARCH program indicated putative binding sites for the Ets and USF transcription factors (−106CTGGAGG−100 and −119AACGTG−114, respectively; Fig. 4A, bottom). EMSA using mouse HAVCR2 promoter-specific Pr1 and Pr2 probes, which contained putative Ets- and USF-binding sites, respectively (Fig. 4B), demonstrated an increased binding of nuclear proteins to each of these probes after IL-10 or TGF-β treatment of mouse BMDCs and sDCs (Fig. 4C–E). In contrast, mutation of the Ets (MutEts-Pr1)- and USF (MutUSF-Pr2)–binding probes blocked this nuclear protein binding (Fig. 4B, 4F). Furthermore, Pr1 and Pr2, but not MutEts-Pr1 and MutUSF-Pr2, successfully precipitated Ets and USF proteins, respectively, from nuclear extracts of BMDCs treated with IL-10 or TGF-β (Fig. 4G). Correspondingly, ChIP assays showed an increased recruitment of Ets1, Ets2, USF1, and USF2 to the HAVCR2 promoter after IL-10 or TGF-β treatment (Fig. 4H). We further observed that the human HAVCR2 promoter also has Ets- and USF-binding sequences (−129CAGGATG−123 and −244CATCTG−239, respectively) that in turn were bound by these transcription factors upon IL-10 or TGF-β treatment of HuMoDCs (Supplemental Fig. 1A–E). These results demonstrate that both IL-10 and TGF-β induce Ets and USF binding to the HAVCR2 promoter in mouse and human DCs.
Next, we tested whether Ets and USF transactivate the mouse HAVCR2 promoter in a reporter assay. Overexpression of Ets1, Ets2, USF1, or USF2 in RAW264.7 mouse macrophages strongly enhanced the activity of wild-type HAVCR2 promoter (Supplemental Fig. 2). However, HAVCR2 promoter activity was drastically reduced when the Ets- or USF-binding site in the HAVCR2 promoter was mutated (Supplemental Fig. 2). Furthermore, downregulating the expression of Ets1, Ets2, USF1, or USF2 by siRNA inhibited IL-10– and TGF-β–induced TIM-3 upregulation by BMDCs (Fig. 4I, 4J) and HuMoDCs (Supplemental Fig. 1F, 1G). Collectively, these data demonstrate that Ets and USF positively regulate HAVCR2 promoter activity, leading to increased TIM-3 expression by DCs upon IL-10 and TGF-β stimulation.
c-Src–Btk signaling drives TIM-3 upregulation on DCs
We then examined the signaling events upstream of Ets and USF recruitment to the HAVCR2 promoter upon IL-10 or TGF-β stimulation. Computational analysis using the Scansite 3 program (http://scansite3.mit.edu) predicted c-Src (a nonreceptor tyrosine kinase) as a potential common interaction partner for the intracellular domains of both mouse and human IL-10 and TGF-β receptors. Therefore, we determined whether c-Src activation, as measured by tyrosine 416 (Tyr416) phosphorylation of c-Src (23), was induced by both IL-10 and TGF-β. Within 2.5 min after IL-10 or TGF-β treatment, phosphorylation of c-Src was enhanced in BALB/c BMDCs and sustained up to 6 h (Fig. 5A). Phosphorylation of c-Src was also induced in BALB/c sDCs and HuMoDCs after IL-10 or TGF-β treatment (Fig. 5B, 5C). Thus, both IL-10 and TGF-β induce c-Src activation in mouse and human DCs.
It is reported that Btk can function either as an upstream regulator or downstream effector of c-Src (13, 24). Accordingly, we verified whether IL-10 and TGF-β induced Btk activation in DCs. We assessed Btk activation by measuring phosphorylation of Btk at Tyr223 (13). Phosphorylation of Btk was enhanced in BALB/c BMDCs at 15 min after IL-10 or TGF-β treatment and persisted up to 6 h (Fig. 5D). IL-10 and TGF-β similarly induced Btk phosphorylation in BALB/c sDCs and HuMoDCs (Fig. 5E, 5F). Thus, Btk, like c-Src, is activated by both IL-10 and TGF-β in DCs. Furthermore, both c-Src and Btk were immunoprecipitated with IL-10R1 and TGF-βRI after IL-10 and TGF-β treatment, respectively (Fig. 5G, 5H). Because the kinetic analyses showed an early activation of c-Src relative to Btk (Fig. 5A, 5D), c-Src could be an upstream regulator of Btk in IL-10 and TGF-β signaling. To test this possibility, we silenced c-Src and Btk expression by siRNA (Supplemental Fig. 3A, 3B) and analyzed the effect of these silencing on IL-10– and TGF-β–stimulated activation of Btk and c-Src. Whereas c-Src silencing blocked Btk phosphorylation after IL-10 and TGF-β stimulation, Btk silencing had no effect on c-Src phosphorylation (Supplemental Fig. 3C, 3D). Moreover, c-Src silencing prevented Btk to interact with IL-10R1 or TGF-βRI despite IL-10 or TGF-β stimulation, respectively (Supplemental Fig. 3E). These results confirmed that c-Src acts upstream of Btk in both IL-10 and TGF-β signaling pathways.
Next, to verify whether c-Src and Btk were required for IL-10– and TGF-β–induced binding of Ets and USF to the HAVCR2 promoter and upregulation of TIM-3 expression, we silenced c-Src and Btk expression in BALB/c BMDCs and HuMoDCs by siRNA (Fig. 5I). Silencing of c-Src or Btk inhibited Ets and USF binding to the HAVCR2 promoter and prevented TIM-3 upregulation on BALB/c BMDCs (Fig. 5J, 5L) and HuMoDCs (Fig. 5K, 5M) upon IL-10 and TGF-β stimulation. Together, these data suggest that both IL-10 and TGF-β induce c-Src–Btk signaling, which triggers Ets and USF binding to the HAVCR2 promoter and upregulation of TIM-3 expression by DCs.
Silencing of c-Src reduces TIM-3 expression on TADCs and enhances antitumor effects of CpG DNA
Having shown that c-Src activation is an early event required for IL-10– and TGF-β–induced TIM-3 upregulation on DCs (Fig. 5, Supplemental Fig. 3), we determined whether c-Src silencing affected TIM-3 expression on TADCs, which often encounter IL-10 and TGF-β in the tumor microenvironment (11). Notably, IL-10 and TGF-β secreted in MC-38 and CT26-wt tumor cell supernatants induced c-Src phosphorylation in BMDCs (Fig. 6A, 6B). Moreover, silencing of c-Src by siRNA blocked the upregulation of TIM-3 on BMDCs mediated by the MC-38 or CT26-wt cell supernatants (Fig. 6C). These data indicate that tumor cell–secreted IL-10 and TGF-β upregulate TIM-3 expression on DCs in a c-Src–dependent manner.
Next, we assessed the in vivo role for c-Src in the regulation of TIM-3 expression by TADCs. For this, we delivered c-Src–specific vivo-MO i.v. into MC-38 or CT26-wt tumor-bearing mice, which in turn reduced c-Src expression in the tumors (Fig. 7A, 7B). Compared with sDCs from tumor-bearing or tumor-free mice, TADCs from MC-38 or CT26-wt tumors expressed much higher levels of TIM-3 (Fig. 7C, 7D, PBS-treated panels). However, TIM-3 expression on TADCs was greatly reduced after c-Src silencing through vivo-MO–mediated interference (Fig. 7C, 7D). These results revealed c-Src as a critical mediator of enhanced TIM-3 expression on TADCs in vivo.
Because high expression of TIM-3 by TADCs suppresses nucleic acid–mediated antitumor responses (6), it seemed likely that the downregulation of TIM-3 on DCs upon c-Src silencing would improve the antitumor effects of nucleic acids. To confirm this scenario, we transferred c-Src–silenced DCs and control (control siRNA-transfected) DCs into syngeneic mice bearing MC-38 or CT26-wt tumor. We then injected these mice with a nucleic acid–based adjuvant CpG-ODN, a potential therapeutic agent for cancer treatment (25). Although treatment with CpG-ODN led to inhibition of tumor growth in the presence of control DCs, the efficacy of CpG-ODN was enhanced by transfer of c-Src–silenced DCs (Fig. 8A). In contrast, forced expression of TIM-3 in c-Src–silenced DCs blocked the tumor growth inhibitory effect of CpG-ODN (Fig. 8A). Consistent with these findings, CpG-ODN–induced intratumoral expression of the potent antitumor cytokine IL-12 (26) was higher in the presence of c-Src–silenced DCs than in the presence of control DCs (Fig. 8B, 8C). However, the ability of c-Src–silenced DCs to enhance CpG-ODN–induced IL-12 production was blocked upon forced expression of TIM-3 in these cells (Fig. 8B, 8C). Together, these results suggest that c-Src plays a key role in suppression of the antitumor effects of CpG-ODN and that c-Src mediates this inhibitory effect by increasing the expression of TIM-3 on TADCs (Fig. 9).
The level of TIM-3 expression on TADCs influences the efficacy of antitumor therapies (6). However, the regulatory mechanism of TIM-3 expression is ill defined. In this study, we have delineated the signaling pathway responsible for TIM-3 upregulation on DCs and identified c-Src as a molecular target to reduce TIM-3 expression on TADCs and enhance the antitumor effects of immunostimulatory DNA.
In the tumor microenvironment, TIM-3 expression is upregulated on DCs via various tumor cell–secreted immunosuppressive factors, including IL-10 (6). Our findings demonstrate that TIM-3 expression is also upregulated by TGF-β. Furthermore, we provide evidence that the transcription factors Ets and USF play a critical role in IL-10– and TGF-β–induced upregulation of TIM-3 expression by DCs. This differs from T cells in which constitutive TIM-3 expression is dependent on T-bet (8). Until now, the role of Ets and USF in the regulation of TIM-3 expression has remained unaddressed. Our findings therefore demonstrate a previously unrecognized function for Ets and USF.
Our second finding that Btk is necessary for upregulation of TIM-3 expression on DCs establishes a new role for Btk in DC immunoregulation. Indeed, both IL-10 and TGF-β required Btk to induce Ets and USF binding to the HAVCR2 promoter. So far, only a few studies have addressed the role for Btk in DC immunobiology. For instance, Btk has been shown to negatively regulate LPS-induced DC maturation (27). Also, our previous studies have implicated Btk in c-MET (hepatocyte growth factor receptor) and TIM-3 signaling that leads to DC suppression (5, 13). However, the role of Btk in IL-10 and TGF-β signaling remains unknown. In this study, we describe Btk as an important mediator of both IL-10 and TGF-β signaling pathways. In addition, to the best of our knowledge, our findings are the first to demonstrate the regulation of Ets and USF activity by Btk.
Another key observation made in this study is that c-Src activation is an early event induced by IL-10 and TGF-β that is needed for downstream Btk activation and subsequent recruitment of Ets and USF to the HAVCR2 promoter. Although activation of c-Src by TGF-β has been reported in other cell types (28), a similar role for IL-10 is not known. Our findings also suggest that c-Src plays a key role in regulating TIM-3 expression by TADCs in vivo. For instance, silencing of c-Src expression blocked upregulation of TIM-3 by TADCs. In addition, we found that blocking c-Src expression and thereby preventing increased expression of TIM-3 by TADCs enhanced the antitumor effects of CpG DNA in vivo. This later finding demonstrates the therapeutic potential of targeting c-Src. Studies have reported that the antitumor effects of CpG DNA are largely dependent on its ability to induce IL-12 secretion from TADCs (6, 29). However, the elevated TIM-3 expression by TADCs greatly impairs CpG DNA-induced IL-12 secretion (6). TIM-3 can mediate this inhibitory effect in two possible ways. First, elevated TIM-3 expression by TADCs may interfere with endosomal localization of CpG DNA, thus preventing the interaction between CpG DNA and TLR9 (6). Consequently, CpG DNA cannot trigger NF-κB signaling (via TLR9), which is necessary for subsequent IL-12 production (6, 30). Second, TIM-3 may transduce inhibitory signals that block CpG DNA-induced NF-κB activation and IL-12 production. This possibility has been spurred by our recent observations that triggering of TIM-3 signaling by anti–TIM-3 Ab blocks LPS-induced NF-κB activation and IL-12 secretion (5). In the same report, we have demonstrated c-Src as a critical component of the TIM-3 signaling pathway that inhibits NF-κB activation and subsequent IL-12 secretion. In addition, we have shown in this study that c-Src was essential for enhanced TIM-3 expression by TADCs. Therefore, downregulation of c-Src expression not only blocks the inhibitory effects of TIM-3 signaling but also reduces the TIM-3 levels on TADCs (Fig. 9). In the absence of TIM-3, CpG DNA can readily trigger TLR9-mediated NF-κB signaling and elicit IL-12 secretion from TADCs. In fact, our data have shown that CpG DNA-induced IL-12 production in the tumor milieu was significantly increased after c-Src silencing that reduced TIM-3 expression on TADCs. In this way, suppression of c-Src in DCs can enhance the antitumor efficacy of CpG DNA.
In summary, our study has established a pivotal role for c-Src in the regulation of TIM-3 expression in DCs, and identifies the pathway of c-Src→Btk→Ets1/2, USF1/2 as controlling this process. In addition, we have demonstrated that c-Src downregulation in TADCs potentiates the antitumor effects of CpG DNA. Earlier we had shown that c-Src is also required for TIM-3–induced inhibition of DC activation and maturation (Fig. 9) (5). Notably, inhibition of TADCs by TIM-3 has been linked to an impaired nucleic acid–mediated antitumor response (6). These reports together with our present study suggest that the suppression of c-Src in DCs augments the antitumor effects of nucleic acids in two mutually nonexclusive ways: 1) by downregulating TIM-3 expression on DCs; and 2) by blocking the inhibitory effects of TIM-3 signaling on DCs (Fig. 9). Collectively, our results provide a molecular blueprint for targeting c-Src to improve the antitumor efficacy of DNA vaccines.
We thank Drs. Ashwani Kumar (Institute of Microbial Technology) and Prafullakumar B. Tailor (National Institute of Immunology) for reading the manuscript, Dr. Galina V. Shurin (University of Pittsburgh) for providing the MC-38 cell line, Drs. Tamara Nowling (Medical University of South Carolina), Marie-Dominique Galibert (Univèrsité Rennes), and Hisaya Akiba (Juntendo University) for sharing plasmid constructs, Fortis Hospital (Mohali, India) for providing human buffy coat, and the Institute of Microbial Technology Animal House Facility for providing mice for experimentation.
This work was supported by grants from the Council of Scientific and Industrial Research and the Department of Biotechnology, Government of India.
The online version of this article contains supplemental material.
Abbreviations used in this article:
bone marrow–derived dendritic cell
Bruton’s tyrosine kinase
human monocyte-derived DC
IL-10 receptor 1
secreted alkaline phosphatase
small interfering RNA
TGF-β type I receptor
T cell Ig and mucin protein-3
The authors have no financial conflicts of interest.