A20 Controls Macrophage to Elicit Potent Cytotoxic CD4+ T Cell Response

Emerging evidence indicates that CD4+ T cells possess cytotoxic potential for tumor eradication and perforin/granzyme-mediated cytotoxicity functions as one of the important mechanisms for CD4+ T cell-triggered cell killing. However, the critical issue is how the cytotoxic CD4+ T cells are developed. During the course of our work that aims at promoting immunostimulation of APCs by inhibition of negative regulators, we found that A20-silenced Mф drastically induced granzyme B expression in CD4+ T cells. As a consequence, the granzyme-highly expressing CD4+ T cells exhibited a strong cytotoxic activity that restricted tumor development. We found that A20-silenced Mф activated cytotoxic CD4+ T cells by MHC class-II restricted mechanism and the activation was largely dependent on enhanced production of IFN-γ.


Introduction
CD8 + T cells are the most cytotoxic T lymphocytes (CTLs) that directly destroy virus-infected or malignant cells. CD4 + T cells are recognized for their coordinated orchestration by production of various cytokines, such as T helper (Th)1 producing interferon (IFN)-c to promote cellular immunity, Th2 producing interferon (IL)-4 to potentiate humoral immune response, and Th17 producing IL-17 to facilitate inflammation and autoimmune diseases. Recent studies further identified different subsets of CD4 + regulatory T cells which perform immune regulation on effector T cells by expressing transcription factor FoxP3 or by secreting anti-inflammatory cytokine IL-10 or transforming growth factor (TGF)-b. However, emerging evidence indicates that CD4 + T cells also develop cytotoxic activity to directly participate in cytolysis of tumor or infected cells. For instance, tumor-reactive CD4 + T cells were found to develop cytotoxic activity and eradicate large established melanoma after transfer into lymphopenic hosts [1,2]. The critical issue is how these cytotoxic CD4 + T cells are developed.
Macrophages (M s) are initially recognized as phagocytic cells responsible for pathogen elimination and housekeeping function in homeostasis and tissue repair. The classically known M s, which are activated by microbial products or interferon (IFN)-c, produce large amounts of proinflammatory cytokines, express high levels of MHC molecules, and function as a potent killer of pathogens and tumor cells [3]. Dependent on the anatomical location and the physiological or pathological context, M s can be alternatively activated by anti-inflammatory cytokines such as IL-4 or IL-13 [4]. The alternatively activated M s produce high amounts of IL-10, express scavenger receptors, and exhibit anti-inflammatory and tissue repair functions [5]. Recent studies suggest that M s represent a very plastic cell population that play an essential role in the regulation of the pro-inflammation vs anti-inflammation and in the coordination of the pro-tumorgenesis vs. anti-tumorgenesis [6]. Classically activated M s and alternatively activated M s represent two extremes in the spectrum of the phenotype and functionality of M s [5,7].
To promote the antitumor activity of M , we used an A20 silencing strategy to enhance the classical activation of M . This was based upon the published studies that A20, a zinc-finger ubiquitinmodifying enzyme, inhibits several upstream signaling pathways of NF-kB in a feedback manner by degradation or deactivation of signaling molecules via its dual functions of ubiquitination and deubiquitination [8,9,10]; A20-deficient M s display prolonged NF-kB activity [8,10]; A20-silenced dendritic cells (DCs) express higher levels of costimulatory molecules and proinflammatory cytokines, and display a superior immunostimulatory ability [11]. We found that A20-silenced M not only enhances expression of perforin and granzyme B in CD8 + T cells and Natural Killer (NK) cells, also drastically upregulate these cytotoxic molecules in CD4 + T cells. As a consequence, the granzyme-highly expressing CD4 + T cells exhibited cytotoxic activity in vitro/vivo. We further defined that A20-silenced M activated cytotoxic CD4 + T cell response by MHC class-II restricted mechanism, and the activation was largely dependent on enhanced IFN-c production.

A20 Controls M Maturation and Immunostimulatory Activity
To investigate whether A20 controls maturation of M , bone morrow-derived M s (BMM s) were transduced with adenovirus Ad-A20shRNA (Ad-shA20) or Ad-GFPshRNA (Ad-con). Downregulation of A20 expression by Ad-shA20 was confirmed via quantitative RT-PCR (qRT-PCR) at the level of mRNA and via intracellular staining (ICS) at the level of protein (Fig.S1A&B). Flow cytometric assay shows that Ad-shA20-transduced BMM s expressed higher levels of CD80, CD86, CD40 and MHC class-II molecule I-A/I-E than Ad-con-BMM s under the stimulation of LPS (Fig.1A). ELISA results show that Ad-shA20-BMM s, but not Ad-con-BMM s, spontaneously produced large amounts of inflammatory cytokines such as IL-6, TNF-a, IFN-c and IL-12p40, and produced larger amounts of these cytokines in response to LPS stimulation (Fig.1B). Adenoviral vector which induces maturation of antigen-presenting cells per se [12] may contribute to the observed ''spontaneous'' cytokine production by A20-silenced BMM s. A20-silenced BMM s also produced higher level of nitric oxide than the control M s (Fig.1C). Despite the reported anti-apoptotic role of A20 in TNF-treated cells [9], A20-silenced BMM s showed a comparable viability to Ad-con-BMM s in cell culture (Fig.S2). Taken together, these results imply that A20 negatively regulates the maturation and cytokine production of BMM s.
Next, we tested if A20-silenced BMM s possess an enhanced immunostimulatory activity. The transduced BMM s were pulsed with H2-K b -restricted OT-I (SIINFEKL) or OT-II (ISQAV-HAAHAEINEAGR) peptide and then co-cultured with CD8 + OT-I or CD4 + OT-II cells isolated from Ovalbumin (OVA)specific TCR transgenic mice. Results showed that CD8 + OT-I cells cocultured with A20-silenced BMM s expressed enhanced levels of CD25 and CD44 in comparison with those cocultured with the control BMM s (Fig.S3, left). Moreover, the cocultured OT-I cells with A20-silenced BMM s produced higher levels of IFN-c and TNF-a (Fig.S3, right) In parallel, A20-silenced BMM s also more potently activated CD4 + OT-II cells, as evidenced by enhanced expression of CD25 and CD69, and heightened production of IFN-c by the OT-II cells cocultured with Ad-shA20-BMM s (Fig.S4). A20-silenced BMM s also modestly enhanced proliferation of both CD8 + OT-I or CD4 + OT-II cells, as tested by 3 H-Thymidine Incorporation Assay (data not statistically significant and not shown). These results support that A20-silencing endowed BMM s with an enhanced immunostimulatory activity.

A20 Controls M to Elicit a Cytotoxic CD4 + T Cell Response
We examined the potential of A20-silenced BMM to activate cytotoxic cell responses by testing expression of cytotoxic molecules in the cocultured T cells by ICS. As shown in Fig.2A, A20-silenced BMM enhanced expression of granzyme B in co-cultured CD8 + OT-I T cells (upper), but also significantly enhanced granzyme B expression in co-cultured CD4 + OT-II cells (lower). In the meantime, we also detected an enhanced expression of perforin in these co-cultured T cells with A20silenced BMM (Fig. S5). To rule out that the observed result is derived from the adenoviral transduction of M , BMM s were Figure 1. A20 controls maturation and cytokine production of M . A. Expression of costimulatory molecules and MHC class II molecule on the adenoviral-transduced BMM in response to stimulation of LPS. B. Production of inflammatory cytokines by the adenoviral-transduced BMM s, as tested by ELISA. C. NO production by adenoviral-transduced BMM s, as tested by Griess assay. Experiments were repeated three times with similar results. *p,0.05, **p,0.01 Ad-shA20-vs. Ad-con-transduced M . doi:10.1371/journal.pone.0048930.g001 nucleofected with recombinant plasmid pshuttle-shA20 or pshuttle-shGFP according to the manufacturer's instruction (Amaxa), which reached ,40% transfection efficiency, as monitored by Ad-GFP nucleofection in parallel (data not shown). The nucleofected BMM s were then co-cultured with freshly isolated OT-II T cells in the presence of the OT-II peptide. ICS assay showed that pshuttle-shA20-nucleofected BMM s display a more potent ability to elicit expression of granzme B in the cocultured OT-II cells (Fig.S6). Furthermore, we also tested the potential of A20-silenced BMM immunization to induce cytotoxic cell responses in mouse model. C57BL/6 mice were i.p. immunized with OT-I/OT-II peptides-pulsed, Ad-shA20 or Ad-con-transduced BMM s or PBS twice. 7-10 days after the 2 nd immunization, spleens and lymph nodes (LNs) were harvested to analyze granzyme B expression in effector cells by ICS. In agreement with the in vitro study, ICS assay explored that A20-silenced BMM s significantly enhanced expression of granzyme B and perforin in CD4 + and CD8 + T cells as well as NK cells derived from inguinal lymph nodes (LNs) (Fig.2B & Fig. S5) or spleen (data not shown) of the immunized C57BL/6 mice. qPCR assay further confirmed an enhanced level of granzyme B expressed in CD4 + T cells derived from OT-II (not OT-I)-pulsed, A20-silenced BMM -immunized mice (Fig.2C). To exclude the possibility that the OT-I/OT-IIpulsed, A20-silenced BMM s have any different propensity of releasing the loaded antigen to endogenous APCs, we in vitro cultured OVA protein-pulsed, differently transduced BMM s for one or three days. ELISA analysis revealed that an identical amount of cell-free OVA protein is present in the culture media of differently transduced or Mock BMM s (data not shown).
To determine cytolytic activity of these effector cells, the splenocytes were isolated from the immunized mice and cultured overnight for the NK-mediated cytotoxicity assay or 5-6 days in the presence of OT-I or OT-II peptide for CD8 + or CD4 + T cellmediated cytotoxicity assay. Due to the low expression of MHC class-II molecule on the targeted cell, a murine Burkitt lymphoma cell line B6SJ003, the splenocytes cultured with OT-II peptide were selected using anti-CD4 beads prior to the cytotoxicity assay. As shown in Fig.3, A20-silenced BMM immunization enhanced the activity of NK cells, CD8 + T cells, and CD4 + T cells in killing their specific target cell compared with control BMM or PBS immunization. The killing specificity of CD8 + T cells and NK cells was confirmed by failure of the cytotoxic cells to kill the irrelative control, such as EL-4 cells. We also found that freshly isolated CD4 + T cells from A20-silenced BMM -immunized mice displayed a relatively high non-specific cytolytic activity against the target cell EL-4, but the in vitro culture of these CD4 + T cells in the presence of OT-II peptide 5-6 days led these cells to largely lose their non-specific killing activity. Concanamycin A (CMA) acidifies intracellular vacuolar granules to degrade the content in the exocytotic granules [13]. Ethyleneglycotetracetic acid (EGTA) chelates extracellular free calcium to inhibit exocytosis of cytolytic granules and pore formation by perforin [14]. To confirm the CD4 + T cell-associated cytotoxicity is mediated by cytotoxic molecules, CMA and EGTA were included for blocking perforin/ granzyme activity in some of those cocultures. Data showed that both CMA and EGTA drastically reduced the cytotoxic activity of CD4 + T cells (both specific and non-specific), as well as that of CD8 + T cells derived from A20-silenced BMM -immunized mice. Moreover, we also directly demonstrated the role of granzyme B in CD4 + T cell-mediated cytotoxicity in the A20-silenced BMMimmunized mice. OT-II (not OT-I)-pulsed, differently transduced BMM s were used to immunize C57BL/6 mice and splenocytes were harvested for CTL assay after the 2 nd immunization. Result showed that CD4 + T cells derived from the A20-silenced BMM -immunized mice killed OVA-expressing B6SJ003 with a higher efficiency, however, Z-AAD-CMK, a weak and specific granzyme B inhibitor, reduced the CD4 + T cells-mediated CTL activity when included into the coculture of OVA-B6SJ003 and CD4 + T cells derived A20-silenced BMM -immunized mice in the CTL assay (Fig. S7). The results strengthen our contention that the expressed cytotoxic molecules contribute to CD4 + T cell-mediated cytotoxicity, as they do in CD8 + T cell-mediated killing.
A20 Controls M to Trigger CD4 + T Cell-mediated Antitumor Immune Protection C57BL/6 mice were immunized with OT-I/OT-II-pulsed, control BMM or A20-silenced BMM , or PBS. The immunized mice were challenged with EG-7 tumor cells two weeks after the 2 nd immunization as described [15]. Fig.4A shows that A20silenced BMM s fully protect the immunized mice from EG-7 challenge. We further tested the A20-silenced BMM -triggered immune protection by challenging the immunized mice with a more aggressive, OVA-expressed melanoma cell line, M05. Fig.4B shows that A20-silenced BMM s were still superior to control M in protecting the immunized mice from the M05 challenge.
Recent studies indicated that tumor-reactive CD4 + T cells have a potential to up-regulate expression of MHC class-II on melanoma B16 cells, and thereby reject the cells by an MHC-II restricted mechanism in a mouse model [1,2]. To demonstrate contribution of CD4 + T cells to A20-silenced BMM -triggered immune protection, OT-II-pulsed, A20-silenced BMM s were used to immunize CD4 2/2 mice and the wildtype littermates followed by a challenge of melanoma M05 cells two weeks after the 2 nd immunization. Fig.4C shows that, in contrast to wild-type mice, which were protected from tumor occurrence with 80% efficiency, CD4 2/2 mice only achieved 20% of protection after A20-silenced BMM immunization.
To directly confirm cytotoxic CD4 + T cell-mediated immune protection, naïve C57BL/6 mice were inoculated with 6610 5 OVA-expressing B6SJ003 followed by adoptive transfer of 5610 6 in vitro primed CD4 + OT-II cells with OT-II-pulsed, A20silenced BMM or control BMM . T cell adoptive transfer was repeated once at a one-week interval. Fig.4D shows that OT-II cells primed by A20-silenced BMM are superior to those primed by control BMM in inhibiting onset and growth of the engrafted OVA-expressed B6SJ003 tumor. However, treatment of A20silenced BMM /OT-II coculture with 100 nM of CMA for 1 hr prior to OT-II adoptive transfer ablates the superior ability of the OT-II cells in rejection of the engrafted tumor. Taken together, the results support that A20-silenced BMM s not only elicit CD8 + T cells and NK cell to combat tumor, also effectively trigger cytotoxic CD4 + T cell response for anti-tumor immune protection.
A20 Restricts M to Trigger Cytotoxic CD4 + T Cell Response by Limiting IFN-c Production As described above, A20-silenced BMM s not only express enhanced proinflammatory cytokines, also prime the cocultured T cells to produce higher levels of proinflammatory cytokines. To determine whether the enhanced cytokine expression relates to the distinct activity of M in triggering a cytotoxic CD4 + T cell response, the control, but not A20-silenced, BMM s were cocultured with CD8 + OT-I or CD4 + OT-II T cells in the presence of varying doses of IFN-c, IL-12, or IL-6. As shown in Fig.5A, while the addition of IL-6 did not promote BMM to trigger granzyme B expression in the cocultured CD4 + OT-II cells and the addition of IL-12 promoted BMM to trigger granzyme B expression in the cocultured CD4 + T cells at a medium level, addition of IFN-c drastically enhanced BMM to trigger granzyme B expression in the cocultured CD4 + T cells. Addition of IFN-c also enhanced the ability of BMM to trigger perforin + -CD4 + T cell response (data not shown), but the result is not so convincing likely due to the antibody's limitation in recognizing perforin in cocultured T cells. Furthermore, addition of IFN-c was found to endow BMM with a comparable ability to A20-silenced BMM in eliciting expression of granzyme B in CD8 + T cells, but the overall granzme B level in the cocultured CD8 + T cells is much lower than those in the cocultured CD4 + T cells (Fig.5B &  Fig. 2A). These results suggest that enhanced production of IFN-c by A20-silenced BMM s may contribute to priming of the cytotoxic T cells, especially to priming of cytotoxic CD4 + T cells.
To verify the effect of the cytokines, the coculture of A20silenced BMM s with T cells was added with anti-IFN-c or anti-IL-12 to neutralize activity of these cytokines. Fig.6A showed that neutralization of IFN-c, but not IL-12, dramatically reduced A20silenced BMM to stimulate production of granzyme in the cocultured OT-II cells. Fig.6B showed that neutralization of either cytokine IL-12 or IFN-c reduced A20-silenced BMM to produce granzyme-expressing OT-I cells to a certain extent. As individually neutralizing IL-12 or IFN-c does not reduce expression of the cytotoxic molecule to the level in cocultured OT-I with con-BMM s (data not shown) or OT-I culture alone (Fig.6B), a synergistic effect of these cytokines may be required for BMM to optimally stimulate a cytotoxic CD8 + T cell response, at least on the cellular level. The results suggest that A20-silenced BMM s provoke cytotoxic CD8 + /CD4 + T cells likely through different mechanisms. A20-silenced BMM s have a superior ability to trigger a cytotoxic CD4 + T cell response largely by enhancing the production of both autocrine and paracrine IFN-c.
To confirm the observed in vitro effect of IFN-c in immunized mice, groups of C57BL/6 mice were immunized twice as the indicated in Fig.7. All the BMM s were pulsed with OT-I/OT-II prior to immunization. Antibody (250 ug/mouse) was administrated (i.p) one day before BMM immunization, or IFN-c (1 ug/ mouse) administered on the same day as the BMM immunization and two days later. ICS analysis of the inguinal LNs showed that immunization of control BMM s with the IFN-c coadministration dramatically activated granzyme B expression in CD4 + T cells, whereas, immunization of A20-silenced BMM with the anti-IFN-c co-administration drastically reduced granzyme B expression in these CD4 + T cells (Fig. 7A). In parallel, co- administration of IFN-c was found to enhance control BMM to stimulate CD8 + T cells, while co-injection of anti-IFN-c attenuated A20-silenced BMM to stimulate CD8 + T cell response (Fig.7B). Injection of IFN-c alone did not achieve significantly cytotoxic T cell responses (Fig.7A&B). A similar but not identical response pattern was obtained from analysis of splenic CD4 + / CD8 + T cells (Fig.S8). These results highlight that IFN-c is critical for M to activate a cytotoxic CD4 + T cell response and that A20 controls M to activate cytotoxic T cells by limiting IFN-c production.
A20-silenced M Elicits a Cytotoxic CD4 + T Cell Response by Activation of IFN-c Signaling as Well as by an MHCclass-II-restricted Mechanism IFN-c exerts its effects on cells by interacting with a specific receptor composed of two subunits, IFNGR1 and IFNGR2, and thereby phosphorylating Jak/Stat1 signaling molecules [16]. To demonstrate A20-silenced BMM s provoking potent cytotoxic T cell response through activation of IFN-c signaling, A20-silenced BMM s and control pulsed with OT-I/OT-II were used to immunize IFNR1 2/2 mice and their wildtype littermates. ICS analysis of the inguinal LNs showed that A20-silenced BMM s had an equivalent or higher efficacy than the control BMM s to induce CD4 + /CD8 + cytotoxic T cell responses in IFNGR1 2/2 mice, but had a significantly lower efficacy compared with what they did in wildtype mice (Fig. 8A). The result implies that IFN-c receptor is required for A20-silenced BMM to elicit cytotoxic T cell responses, but other signaling pathways also contribute some to the function of A20-silenced BMM s. Furthermore, A20silenced or control BMM s were used to immunize Stat1 2/2 mice in parallel with their wildtype littermates. As Stat1 2/2 mice are under the 129S background, OVA protein instead of the peptides was used to pulse the BMM for immunization. Again, ICS showed that A20-silenced BMM had an equivalent or higher efficacy than the control BMM to induce CD4 + /CD8 + cytotoxic T cell responses in Stat1 2/2 mice, but the efficacy is significantly lower than what they did in wildtype mice (Fig. 8B), which supports that IFN-c-triggered Stat1 signaling is required but not the only for A20-silenced BMM to elicit cytotoxic T cell responses. Indeed, Zimmermann et al reported that IFN-c directly activates Stat2 signaling for the antiviral potency [17]. We also Figure 3. A20-silenced M immunization enhances NK cell-, CD8 + T cell-and CD4 + T cell-mediated cytotoxicity. Splenocytes pooled from 2-3 immunized mice were cultured overnight for NK-mediated cytotoxicity assay or 5-6 days in the presence of OT-I or OT-II peptide for T cellsmediated cytotoxicity assay. The splenocytes cultured with OT-II peptide were selected using anti-CD4 beads prior to cytotoxicity assay. Cytotoxic activities were analyzed by LDH release assay as described in Material and Methods. Experiments were repeated three times with similar results. *p,0.05, Ad-shA20-M immunization vs. Ad-con-M immunization for specific killing. doi:10.1371/journal.pone.0048930.g003 analyzed splenocytes from the immunized IFNR 2/2 mice and Stat1 2/2 mice and obtained similar but not identical results (Fig.  S9A&B).
Ultimately, we tested whether A20-silenced BMM uses a MHC class-II-restricted mechanism to induce cytotoxic T cell response. BMM s were prepared from MHCII 2/2 mice or wildtype littermates. The OT-I/OT-II-pulsed, adenoviral-transduced BMM s were used to immunize wildtype C57BL/6 mice as described. ICS analysis of inguinal LNs shows that A20-silenced MHCII 2/2 M , equivalent to the control MHCII 2/2 M , displayed a significantly lower efficacy than their wild-type counterpart in the activation of cytotoxic CD4 + T cells. However, A20-silenced MHCII 2/2 M s barely lost their ability in activation of cytotoxic CD8 + T cells when compared with A20-silenced wildtype BMM s (Fig. 8C). A similar but not identical result was obtained from ICS analysis of the immunized splenocytes (Fig.  S9C). These results support that A20-silenced BMM s activate a cytotoxic CD4 + T cell response in an MHC class-II restricted manner. A20 controls M s to activate cytotoxic T cell responses largely by limiting IFN-c signaling.

Discussion
Cytotoxic CD4 + T cells were detected in both mouse and human over 20 years ago. The early evidence claimed that distinct from cytotoxic CD8 + T cells, CD4 + T cells use the FAS/FAS ligand system for the cytolytic activity [18,19]. Recent studies strongly supported that granule exocytosis of perforin/granzymes represents the main pathway of cytotoxicity in both CD4 + and CD8 + T cells [20,21,22,23,24,25]. In line with these studies, our study suggested that granzyme B as well as possible perforin can be induced in CD4 + T cells by A20-silenced M s and the resultant CD4 + T cells rejected engrafted tumors in a perforin/granzymedependent manner. Although freshly isolated CD4 + T cells from A20-silenced M immunized mice display some nonspecific cytotoxicity, the isolated CD4 + T cells after in vitro re-stimulation use MHC class-II restricted mechanism to kill tumor cells. CD4 + Figure 4. A20-silenced M immunization induces enhanced immune protection. A & B. C57BL/6 mice (5-6 mice/group) were immunized twice. The mice were s.c. injected with 5610 5 EG-7 (A) or M05 (B). Tumor growth was monitored on the indicated days. * p,0.05, Ad-shA20-M immunization vs. Ad-con-M immunization. C. CD4 2/2 C57BL/6 or the wildtype littermates (5-6 mice/group) were immunized with OT-II-peptidepulsed, Ad-shA20-transduced BMM s twice followed by s.c. injection of 5610 5 M05 tumor cells. Tumor occurrence and growth were monitored on the indicated days. **p,0.01, wild-type mice vs. CD4 2/2 mice. D. Transferred OT-II-specific immune pretection. In vitro primed OT-II T cells (5610 6 ) were transplanted into naïve RAG 2/2 C57BL/6 mice (5 mice/group) by retro-orbital injection following s.c injection of OVA-expressed B6SJ1003 tumor cells (6610 5 ). The transplantation of OT-II T cells was repeated one week later. One group of mice were transplanted with CMA-treated, Ad-shA20transduced M -primed OT-II T cells. Tumor growth was monitored on the indicated days. *p,0.05, Ad-shA20-M -primed OT-II T cell transfer vs. Adcon-M -primed OT-II T cell transfer, or Ad-shA20-M -primed OT-II T cell transfer vs. Ad-shA20-M -primed OT-II T cell+ CMT transfer. All the experiments were repeated with similar results. doi:10.1371/journal.pone.0048930.g004 T cell killing of infected or malignant cells in MHC-class IIrestricted manner has been reported in several studies [23]. Quezada et al. and Xie et al. recently further claimed that tumorreactive CD4 + T cells secrete a copious amount of IFN-c to upregulate expression of MHC-class-II molecules on tumor cells and make them the target of cytotoxic CD4 + T cells after transfer into lymphopenic hosts [1,2]. Thus, our reported, A20-silenced M induced, CD4 + T cells exhibit common functional features to those in vivo or ex vivo differentiated cytotoxic CD4 + T cells. It is worth mentioning here that throughout the whole study, we persistently detected a higher level of perforin in either stimulated or immunized T cells by A20-silenced M s and the expressing pattern of perforin in these T cells resembled the expression of    A. Adenoviral-transduced BMM s were used to immunize IFNGR 2/2 mice or the wildtype littermates (2-3 mice/group) twice. The inguinal LNs were harvested for analyzing expression of granzyme B in CD4 + or CD8 + T cells by ICS. p,0.01 Ad-shA20-IFNGR KO mice vs. Ad-ShA20 WT mice. B. Adenoviral-transduced BMM s were used to immunize Stat1 2/2 mice or the wild-type littermates twice (2-3 mice/group). The LNs were harvested for analyzing expression of granzyme B in CD4 + (p,0.05, Ad-shA20-Stat1 KO mice vs. Ad-shA20-WT mice) or CD8 + T cells by ICS. C. BMM s were prepared from MHCII 2/2 mice or the wild-type littermates. The adenoviral-transduced BMM s were used to immunize wild-type mice (2-3 mice/group) twice. The LNs were harvested for analyzing expression of granzyme B in CD4 + (p,0.01, Ad-shA20-MHC-II KO M immunization vs. Ad-shA20-WT M immunization) or CD8 + T cells by ICS. Experiments were repeated with similar results. doi:10.1371/journal.pone.0048930.g008 granzyme B, but the results may not be convincing due to the antibodies' limitation.
Cytotoxic CD4 + T cell differentiation occurs under different physiological or pathological conditions. Recent studies further investigated cytotoxic CD4 + T cells by adoptive cellular transfer (ACT) of antigen-specific CD4 + T cells or creation of antigenspecific TCR-transgenic mice. Brown et al. explored that virusspecific TCR transgenic CD4 + cells acquired perforin-mediated cytolytic activity after adoptive transfer into influenza-infected mice, and that the perforin-dependent cytolysis represents one of the important mechanisms to protect mice from lethal influenza infection [26]. Xie et al. and Quzezada et al. reported that naïve tumor-specific CD4 + T cells develop cytotoxic activity and eradicated established melanoma after transfer into lymphopenic hosts [1,2]. Corthay et al. unveiled that primary antitumor immune response can be triggered by transgenic ID-specific CD4 + T cells in immune deficient SCID mice [27]. All these studies revealed a dominant type-I immune response environment associated with the cytotoxic CD4 + T cell differentiation. For example, EBV-specific CD4 + T cells represent one of the earliest defined cytolytic CD4 + T lymphocytes. Paludan et al. reported that EBV infection triggers CD4 + T cell to primarily differentiate into IFN-c-producing Th1-type [28]. Xie et al and Quzezada et al adoptively transferred tumor antigen-specific CD4 + T cells into lymphopenic mice. Their studies also claimed that Th1 polarization is a default pathway in lymphopenic host [1,2]. Corthay et al found that transgenic ID-specific CD4 + T cells infiltrate into tumors and produce Th1 cytokines in mice with an immune deficient background [27]. Recently, Muranski et al. discovered that Th17-polarized tumor-reactive CD4 + T cells are capable of rejecting established melanomas [29]. Their subsequent study informed that Th17 cells are metastable and able to gradually acquire a Th1-like phenotype secreting less IL-17A and more IFNc [30]. Our reported A20-silenced M s produce high levels of proinflammatory cytokines and preferentially prime IFN-c/TNFa-producing T cells, which further supports type-I immune environment promotes cytotoxic CD4 + T cell development.
Our study further defined that IFN-c is crucial for A20-silenced M to induce cytotoxic CD4 + T cell differentiation. IFN-c impact on cytotoxic CD4 + T cell responses has been implicated in many published studies. Mumberg et al. reported that anti-IFN-c treatment abolishes the CD4 + T cell-mediated rejection of the tumor cells in SCID mice [31]. Corthay explored that CD4 + T cells mediate tumor rejection by producing IFN-c to activate Massociated antitumor activity [27]. Perez-diez et al. revealed that CD4 + T cells obtain the maximal antitumor effect by partnering with NK cells, an innate source of IFN-c [32]. Furthermore, both Xie et al. and Quezada et al. defined that IFN-c facilitates cytotoxic CD4 + T cells to reject malenoma by up-regulation of MHC class-II expression on tumor cells [1,2]. In our present study, IFN-c is found to directly promote expression of cytotoxic molecules in CD4 + T cells, which is consistent with an early report that activation of IFN signaling was required for expression of perforin and granzyme in CD8 + T cells and NK cells in melanoma patients [33]. Thus, IFN-c exhibits comprehensive functions associated with cytotoxic CD4 + T cell response, while our present result suggested a novel mechanism for IFN-c functioning CD4 + T cell-mediated cytotoxicity. Our study further indicated that A20silenced M -induced cytotoxic CD4 + T cell differentiation is MHC class-II restricted, which coincides with published studies that tumor-reactive CD4 + T cells develop cytotoxic activity in an MHC class-II-dependent manner [34] and priming of tumorreactive CD4 + T cells requires MHC class-II expression on recipient or host cells, not on tumor cells [1,2,27]. Most intriguingly, Corthay et al identified that tumor infiltrated macrophages are an important component to re-activate tumorspecific CD4 + T cells by presenting tumor-derived peptides on their MHC-II molecules [27]. Our study further suggested that the re-activation step also triggers CD4 + T to express and exocytose cytotoxic molecules for directly killing MHC-II-restricted tumor cells and MHC-II-non-restricted tumor cells in the close proximity.
Ex vivo generated, tumor-reactive, autologous CD4 + T cell clones have successfully been used to treat melanoma patients [35]. Our study may provide a platform for in vitro generating antigen-specific cytotoxic CD4 + T cells for adoptive tumor immunotherapy.

Peptides, Proteins and Cell Lines
H2-K b -restricted OT-I and OT-II peptides were synthesized by Genemed Synthesis. OVA protein was purchased from Sigma-Aldrich. The B6SJ003 Burkitt lymphoma cell line (H2-K b , MHC-II-expressed) was kindly provided by Herbert C. Morse III at the NIAID/NIH [36]. OVA-expressing B6S1003 was generated by stable transfection of OVA gene. B16-OVA melanoma cell line M05 (H2-K b ) was kindly provided by R. Dutton at the Trudeau Institute [37]. Lymphoma cell EG-7 (H2-K b ) which engineeringly expresses OVA was purchased from ATCC.

M Immunization and Tumor Models
Mouse BMM s were generated by culturing BM cells in the presence of macrophage colony-stimulating factor (M-CSF). The differentiated BMM s were incubated with Ad-shA20 or Ad-con at a multiplicity of infection (MOI) of 500, which allows ,60% of M s to be transduced as demonstrated by Ad-GFP transduction of M in parallel (data not shown). The transduced M s were pulsed with H2-K b -restricted OT-I or OT-II peptide, followed by ex vivo maturation with LPS (100 ng/ml). The M s (0.5-1610 6 ) were then i.p. injected into C57BL/6 mice twice at a one-week interval. For tumor challenge, two weeks after the 2 nd immunization, the mice received s.c. injection of 5610 5 EG-7 or M05. Tumor onset and growth were monitored weekly.

In vitro T Cell Priming
T cells were purified from OT-I or OT-II transgenic mice using the MACS CD8 + or CD4 + T cell isolation kits (Miltenyi Biotec). 5610 4 purified T cells and 5610 3 adenoviral-transduced, OT-I or OT-II peptide-pulsed BMM were cocultured in RPMI 1640 medium supplemented with 10 U/ml of IL-2. In some experiments, anti-IFN-c or anti-IL-12 was added into the co-cultures at the final concentration of 2.5 ug/ml, 10 ug/ml, or 20 ug/ml, or IFN-c, IL-12 or IL-6 was added at the final concentration of 2.5 ug/ml or 10 ug/ml. After 3-5 days of coculture, T cells were harvested to analyze the indicated cytokines by ICS assay.

Adoptively Transfer Assay
The isolated OT-II cells were cocultured with adenoviraltransduced, OT-II peptide-pulsed BMM s for 3-5 days at M :T ratio of 1:10. The cocultured OT-II cells (5610 6 ) were harvested and transplanted into naïve RAG 2/2 C57BL/6 mice by retroorbital injection followed by tumor challenge. The transplantation of OT-II T cells was repeated one week later.

Flow Cytometric Analysis
For ICS assay, lymphocytes were harvested from draining lymph nodes or spleens of immunized mice and cultured with 20 ug/ml of OT-I or OT-II peptide for 6-10 hours at 37uC in the presence of GolgiPlug (BD Biosciences/Pharmingen). After surface staining with anti-CD8 or anti-CD4, cells were permeabilized and stained for intracellular cytokines, as previously described [38,39]. All the antibodies and matched isotype controls were purchased from BD PharMingen or eBioscience. Stained cells were analyzed on a FACSaria (Becton Dickinson) flow cytometer and FloJo software.

CTL and NK Assays
Different numbers of effector cells (5610 5 , 2.5610 5 or 1.25610 5 ) were cocultured with a certain number (5000 cells) of Yac-1 (for NK assay), EG-7 (for CD8 + T cell assay), or OVAexpressed B6SJ1003 (for CD4 + T cell assay) for 5 hrs. EL-4 tumor cell line was used as a non-specific control. Some of the cocultures were added with 3 nM CMA or 1 mM EGTA to inhibit activity of perforin and granzyme. The supernatants were harvested and analyzed by LDH release assay (Roche Diagnostics).

Statistical Analysis
We used the Student's t-test. A 95% confidence limit was used to assess results for statistical significance, defined as P,0.05. Results are typically presented as means 6 standard error. Figure S1 Ad-shA20 reduces expression of A20 mRNA in transduced BMM . BMM s were transduced with Ad-shA20, Ad-con, or PBS. 24 hr later, the M s were stimulated with 100 ng/ml LPS or none for overnight. A, relative expression of A20 mRNA in the transduced BMM s was evaluated by qRT-PCR. * p,0.05, Ad-shA20-M vs. Ad-con-M . B, A20 protein expression in the transduced BMM s was evaluated by ICS. The anti-A20 was purchased from Santa Cruz. Experiments were repeated twice with similar results. (TIF) Figure S2 Ad-shA20 barely enhances apoptosis of the transduced BMM s. BMM s were transduced with Ad-shA20 or Ad-con. 24 hr later, the M s were stimulated with PBS, anti-CD40 (10 ug/ml), or LPS (100 ng/ml) for overnight. The treated BMM s were analyzed with Annexin V-APC Apoptosis Detection Kit (BD Bioscience). Experiments were repeated with similar results.  Figure S6 pshuttle-shA20-transfected M s prime cytotoxic OT-II T cell response in vitro. BMM s were neuclofected with pshuttle-shGFP or pshuttle-shA20. 24 hrs later, the transfected BMM s were cocultured with freshly isolated OT-II T cells in the presence of OT-II peptide for 3-5 days. OT-II T cells were harvested for analyzing expression of granzyme B and perforin by ICS. Experiment was repeated once with similar results. (TIF) Figure S7 Z-AAD-CMK inhibited CTL activity mediated by A20-silenced M -immunzed CD4 + T cells. OT-II (not OT-I)-pulsed, differently transduced BMM s were used to immunize C57BL/6 mice and splenocytes were harvested and restimulated with OT-II peptide for 5-6 days. Various ratios of the splenocytes and target cells (OVA-expressing B6SJ003) were cocultured with or without 75 uM of Z-AAD-CMK for 6 hrs. Cytotoxic activities were analyzed by LDH release assay as described in Material and Methods. Experiments were repeated once. *p,0.05, Ad-shA20-M immunization vs. Ad-shA20-M immunization plus the Z-AAD-CMK treatment. (TIF) Figure S8 IFN-c impacts MF to trigger cytotoxic T cell responses in immunized mice. C57BL/6 mice were immunized twice with 1, PBS plus IgG; 2, PBS plus IFN-c; 3, Ad-con-M ; 4, Ad-con-M plus IFN-c; 5, Ad-shA20-M plus IgG; or 6, Ad-shA20-M plus anti-IFN-c. Antibody (250 ug/ mouse) was i.p administrated one day before M immunization, and IFN-c (1 ug/mouse) was given on the same day as the M immunization and two days later. Two weeks after the 2 nd immunization, splenocytes were harvested for intracelluar granzyme staining of CD4 T cells (A) or CD8 T cells (B). (TIF) Figure S9 A20-silenced M elicits a cytotoxic CD4 + T cell response via activation of IFN-c signaling and by an MHC-class-II-restricted mechanism. A. Adenoviral-transduced BMM s were used to immunize IFNGR 2/2 mice or the wild-type littermates twice. Splenocytes were harvested for analyzing expression of granzyme B in CD4 + or CD8 + T cells by ICS. B. Adenoviral-transduced BMM s were used to immunize Stat1 2/2 mice or the wild-type littermates twice. Splenocytes were harvested for analyzing expression of granzyme B in CD4 + or CD8 + T cells by ICS. C. BMM s were prepared from MHCII 2/2 mice or wild-type littermates. The adenoviral-transduced BMM s were used to immunize wild-type mice twice. Splenocytes were harvested for analyzing expression of granzyme B in CD4 + or CD8 + T cells by ICS. Experiments were repeated with similar results. (TIF)