Inhibiting STAT5 by the BET Bromodomain Inhibitor JQ1 Disrupts Human Dendritic Cell Maturation

Several mechanisms are involved in the regulation of the immune response to prevent excessive activation of the immune system, tissue damage, and autoimmunity. Dendritic cells (DCs), a specialized subset of APCs, play a major role in the balance between tolerance and immunity (1, 2). DCs are responsible for triggering and modulating the immune response against invading pathogens and certain malignant cells while keeping the immune system in a standby condition against self-Ags (3). One critical factor determining the effectiveness of the immune response is the maturation status of the DCs. In the absence of danger signals, immature DCs expressing few MHC and costimulatory molecules induce anergy of T cells or induce and activate regulatory T cells (Tregs) that provide a check on the immune response (4). In contrast, danger signals induce maturation of DCs, which express higher levels of T cell–activating molecules, including MHC, CD80, CD86, CD40, and CD83, and secrete proinflammatory cytokines, thus effectively priming T cell responses (1, 3).

Because of the critical role of DCs in the immune response, they have been studied intensely as a tool to treat cancer and immune-related diseases (5–7). However, cell-based immunotherapies as a treatment modality for immune disorders still require further development. Most cytokines and growth factors implicated in the differentiation and maturation of DCs culminate in the activation of the Jak/STAT signaling pathway (8, 9). Thus, understanding the function of STAT transcription factors in the physiology of DCs is important to further reveal basic physiologic mechanisms of these cells, as well as to target STATs therapeutically to modulate the immune response.

The differentiation of DCs from human monocytes in vitro depends on IL-4 and GM-CSF (10). Although IL-4 signals via STAT6, GM-CSF can activate STAT1, STAT3, and STAT5 (9, 11, 12). The importance of STAT5 in the development of DCs was demonstrated by studies showing that GM-CSF–activated STAT5 promotes differentiation of myeloid DCs by inhibiting the development of plasmacytoid DCs (12, 13). Further evidence showed that DCs differentiated at low doses of GM-CSF become resistant to maturation stimuli afforded by LPS, TNF, and CD40L, leading to the generation of immature (tolerogenic) DCs (11). However, the particular role of STAT5 during the maturation of DCs remains unclear.

It was shown that the selective bromodomain inhibitor JQ1 blocks STAT5 function (14). JQ1 was designed as an inhibitor of bromodomain and extraterminal domain (BET) family members of bromodomain-containing reader proteins, which include BRD2, BRD3, BRD4, and BRDT. These proteins specifically recognize acetylated chromatin sites and facilitate gene expression by recruiting transcriptional activators (15, 16). It was found that JQ1 reduced STAT5 function in leukemia and lymphoma cells through inhibition of BRD2, which is a critical mediator of STAT5 activity (14). JQ1 also was found to decrease STAT5 phosphorylation (and exert an antitumor effect) in acute lymphoblastic leukemia cells through suppression of transcription of IL-7R (17). In addition to its promising role in treating cancer, JQ1 showed anti-inflammatory properties in murine macrophages (18, 19). Although tyrosine kinase inhibitors are used to treat immune-mediated diseases, this strategy is hampered by a lack of specificity and extensive suppression of immune responsiveness, leading to serious adverse effects, such as infections or malignances (20). Therefore, the development of more selective agents with reduced adverse effects would be a major step forward.

In this study, we aimed to determine the effect of JQ1 in human monocyte-derived DCs (Mo-DCs) as a potential inhibitor of STAT5 function. Additionally, we explored the role of STAT5 during the maturation of DCs induced by LPS. Our findings demonstrate that JQ1 can modulate adaptive-immune responses, at least in part through STAT5. Our results provide new insight into the mechanism of STAT5 signaling during Mo-DC maturation and indicate that JQ1 may be used for the rational design of new strategies for the treatment of immune-related disorders.

PBMCs isolated from leukapheresis products from healthy donors were obtained through a Dana-Farber Cancer Institute Institutional Review Board–approved protocol. Volunteers provided informed consent in accordance with the Declaration of Helsinki. PBMCs were isolated by Ficoll-Paque density gradient centrifugation. Human Mo-DCs were generated from PBMCs by adherence to plastic for 2 h at 37°C in 5% CO₂. Adherent monocytes were cultured in RPMI 1640 complete medium (10% heat-inactivated FBS, 1% GlutaMAX, 1 mM sodium pyruvate, 0.5% MEM–amino acids, 1% MEM-Vitamin, 0.07 mM 2-ME, 1% penicillin/streptomycin [Life Technologies, Grand Island, NY]) supplemented with GM-CSF (50 ng/ml) and IL-4 (50 ng/ml; both from PeproTech, Rocky Hill, NJ). After 5 d, immature Mo-DCs (Mo-iDCs) were induced to mature with LPS (Escherichia coli; 100 ng/ml; Sigma-Aldrich, St. Louis, MO). At day 6, mature Mo-DCs (Mo-mDCs) were harvested for further experiments.

JQ1 was provided by James Bradner (Dana-Farber Cancer Institute) (16), and Jak inhibitor 1 (Jaki) was obtained from EMD Millipore (Billerica, MA). The drugs were dissolved in DMSO and added to the culture media for Mo-DC differentiation at day 5 for 1 h before LPS stimulation. JQ1 was diluted to a final concentration of 0.25 μM (unless otherwise noted), and Jaki was used at a final concentration of 1 μM. In each case, equal amounts of DMSO were added as a control. In the experiments involving JQ1 treatment and LPS activation in the presence of rGM-CSF, Mo-iDCs were treated with JQ1 (0.25 μM) for 1 h. Then, GM-CSF (50 ng/ml) was added, and cells were stimulated with LPS (100 ng/ml) for 24 h.

RNA was harvested using the RNeasyPlus Mini Kit from QIAGEN (Valencia, CA). cDNA was synthesized using the TaqMan Reverse Transcription kit (Applied Biosystems, Foster City, CA). Quantitative PCR (qPCR) was performed in triplicate using SYBR Select Master Mix on an ABI Prism 7500 Sequence Detection System (both from Applied Biosystems). RNA expression was normalized to 18S RNA, and the fold increase was determined by dividing the expression in each sample by that of Mo-mDC controls. Data are expressed as mean fold increase and SEM. Primer sequences are provided in Supplemental Table I.

Total protein lysates were prepared using lysis buffer (50 mM Tris [pH 8], 250 mM NaCl, 0.5% Nonidet P-40) containing protease and phosphatase inhibitors. Nuclear and cytoplasmic fractionation was performed according to the manufacturer’s protocol (Active Motif, Carlsbad, CA). Protein lysates were resolved by 8% SDS-PAGE and immunoblotted with primary Abs specific for p-STAT5, p-STAT3, p-STAT1, and poly(ADP-ribose) polymerase (Cell Signaling, Boston, MA); STAT5, STAT3, and STAT1 (Santa Cruz Biotechnology, Santa Cruz, CA); and actin or tubulin (Sigma-Aldrich). Band intensity was quantitated using ImageJ software (National Institutes of Health).

Chromatin immunoprecipitation (ChIP) was performed as described (21). Briefly, cells were fixed in 1% formaldehyde for 10 min and sonicated using a Qsonica sonicator, and lysates were immunoprecipitated overnight with normal rabbit IgG (Caltag, Burlingame, CA) and anti-STAT5 (sc-835; Santa Cruz Biotechnology). qPCR was performed in triplicate using primers for CSF2RA (5′-TTTGCATGTGGTCTTTGAGG-3′ and 5′-TTCTTGACAACACCCAGCAC-3′), CisH (5′-CCCGCGGTTCTAGGAAGAC-3′ and 5′-CGAGCTGCTGCCTAATCCT-3′), Socs2 (5′-AGGCCGATTCCTGGAAAG-3′ and 5′-CGACGAGACTTGGCAAGAG-3′), CD83 (5′-CTGGCCCTCAAATTCTTTCA-3′ and 5′-TGAGACGTTAGCCAGTGGAA-3′), CD80 (5′-CCAAATCTTCACCCCACCTA-3′ and 5′-CTGAGGAAAAGCGAATGGAA-3′), or rhodopsin (5′-TGGGTGGTGTCATCTGGTAA-3′ and 5′-GGATGGAATGGATCAGATGG-3′). The results were normalized to the input and expressed relative to binding to a negative control region in the rhodopsin gene. Match Browser (http://www.gene-regulation.com/cgi-bin/pub/programs/match/bin/match.cgi) and the UCSC Genome Browser (http://genome.ucsc.edu) were used to identify potential STAT5 binding sites in the regulatory region of genes.

The eGFP-expressing lentiviral vector, SiEW, expressing a constitutively activated mutant form of STAT5a (caSTAT5) was cloned and produced as described (22). Immature Mo-DCs were infected with lentivirus expressing caSTAT5 or empty vector at day 5 of culture in the presence of 4 μg/ml Polybrene (Sigma-Aldrich) and centrifuged at 2500 rpm at 32°C for 30 min. Medium was changed 150 min after transduction, and Mo-iDCs were treated with JQ1 or DMSO for 1 h, followed by LPS stimulation (100 ng/ml). After 24 h, Mo-mDCs were analyzed by flow cytometry or harvested for RNA analysis. The transduction efficiency was between 60 and 70%.

Mo-iDCs were generated in the presence of IL-4 and GM-CSF, as described above. On day 5 of culture, culture media were removed, and fresh media were added to eliminate residual GM-CSF. Then, 10 μg anti–GM-CSF Ab (BioLegend, San Diego, CA) was added to Mo-iDCs, and cells were stimulated with LPS (100 ng/ml) for 24 or 48 h.

Viable cells were measured by adenosine triphosphate–dependent bioluminescence using the CellTiter-Glo assay (Promega, Madison, WI). The combination index to measure drug interactions was calculated using CalcuSyn software (Conservion, Ferguson, MO).

Mo-mDCs pretreated with JQ1, Jaki, or DMSO were harvested and washed to remove residual drug at day 6 of culture. These cells were cocultured with bead-purified allogeneic CD3⁺ T cells labeled with CellTrace Violet (Molecular Probes). Cells were cultured at 37°C in 5% CO₂ for 5 d. Cell proliferation was quantified by flow cytometry. The division index and quantitation of cell division were calculated with FlowJo 8.7 software (TreeStar).

Cells were stained on ice in PBS containing 2% FBS for 20 min, using the following mAbs: FITC anti-CD11c; PERCP anti-CD14; PE-Cy7 anti-CD80; allophycocyanin anti-CD83; allophycocyanin-Cy7 anti-CD40; Pacific blue anti-CD86; and V500 anti–HLA-DR (BD Biosciences); anti-CD4 (allophycocyanin, allophycocyanin-Cy7, and Alexa Fluor 488); anti-CD8 (PE and PERCP); allophycocyanin-Cy7 anti-CD25; and PE-Cy7 anti-CD127 (BioLegend). For intracellular cytokine staining, cells were stimulated with Leukocyte Activation Cocktail with GolgiPlug (BD Biosciences) for 5 h at 37°C in 5% CO₂. Cells were permeabilized using a BD Cytofix/Cytoperm Fixation/Permeabilization Solution Kit (BD Biosciences). Intracellular staining was performed with PE anti–IL-4, allophycocyanin anti–IFN-γ, PE-Cy7 anti–IL-10, and FITC anti-TNF (BioLegend). To quantitate Tregs (CD4⁺CD127⁻CD25⁺Foxp3⁺), cells were stained for intracellular Foxp3 using the allophycocyanin Anti-Human Foxp3 Ab Kit (eBioscience). Flow cytometry was performed using a FACSCanto II (Becton Dickinson, Franklin Lakes, NJ), and the results were analyzed with FlowJo 8.7 software.

Secretion of IL-12p70, TNF, IL-10, and IFN-γ in Mo-mDC supernatant or coculture supernatant was determined using ELISA MAX Deluxe (BioLegend), following the manufacturer’s instructions. Analysis was performed using SpectraMax M3 (Molecular Devices).

Data are shown as mean ± SEM. Comparison of results was carried out using a two-tailed paired Student t test when there were only two groups or one-way ANOVA followed by the Tukey posttest for multiple comparisons. Analyses were performed using GraphPad Prism 6 software, and differences were considered significant at p < 0.05.

To understand the role of STAT5 in the maturation of human Mo-DCs and to determine the potential of targeting this transcription factor therapeutically, we focused on pharmacological modulators of STAT5 in hematopoietic cells. Previous data demonstrated that the bromodomain inhibitor JQ1 reduces STAT5 function through the inhibition of BRD2 protein in leukemia and lymphoma cells (14). Therefore, we probed whether JQ1 could also inhibit STAT5 activity in Mo-DCs following LPS stimulation. LPS stimulation prominently increased the activating tyrosine phosphorylation of STAT5 in Mo-DCs. Treatment with JQ1 abrogated this effect, although it did not significantly alter STAT5 protein expression (Fig. 1A). The ratio of phosphorylated STAT5 to total STAT5 (p-STAT5/STAT5) was reduced by JQ1 in Mo-DCs induced to mature with LPS to the same levels observed in immature Mo-DCs (Fig. 1B). LPS can lead to activation of STAT5, as well as STAT1 and STAT3 (23, 24). However, JQ1 had no significant effect on either tyrosine phosphorylation or total levels of STAT1 or STAT3 in Mo-mDCs, indicating that BET inhibition by JQ1 selectively reduces STAT5 activation without influencing the activation of other STATs (Fig. 1B).

Following tyrosine phosphorylation, STATs translocate from the cytoplasm to the nucleus. Therefore, we next considered whether inhibition of STAT5 phosphorylation was reflected in a decrease in nuclear localization of STAT5. We isolated the nuclear fraction from Mo-DCs induced to mature by LPS that had been pretreated with JQ1 or vehicle alone. As a positive control, cells also were treated with Jaki, which completely blocks the Jak kinases upstream of STATs. In Mo-DCs, the nuclear accumulation of both total STAT5 protein and its phosphorylated form were increased following LPS stimulation. However, JQ1 treatment reduced nuclear localization of both phosphorylated and total STAT5 (Fig. 1C). These findings further confirm that JQ1 inhibits STAT5 activation and, hence, decreases STAT5 nuclear localization, which could inhibit its transcriptional activity.

Therefore, we next determined whether JQ1 affected the transcriptional activity of STAT5 in Mo-mDCs and how this compared with its transcriptional activity in Mo-iDCs. We assessed the mRNA expression of endogenous STAT5 target genes, including CisH, Socs2, Socs3, Bcl-x, Bcl-6, Bcl-2, and Bcl-3 (Fig. 1D, 1E). Consistent with the induction of STAT5 phosphorylation, LPS stimulation led to increased expression of Socs2, Socs3, Bcl-x, Bcl-2, and Bcl-3, all of which were abrogated by JQ1 treatment. Bcl-6, which is repressed by STAT5, showed decreased expression following LPS stimulation, and JQ1 treatment tended to reverse this effect. Although LPS-induced maturation did not further increase CisH mRNA levels, JQ1 was found to repress the expression of CisH compared with control Mo-mDCs. Furthermore, ChIP analysis showed that maturation of Mo-DCs was associated with increased binding of STAT5 to the regulatory regions of CisH, SOCS2, and Bcl-3, and this recruitment of STAT5 was reduced following JQ1 treatment (Fig. 1F). To determine whether these alterations in STAT5 binding were associated with changes in transcription, we evaluated whether the regions of STAT5 binding corresponded to regions of transcriptionally active chromatin by performing ChIP assays for acetylated H4 histones (acetyl H4). There was a trend toward increased acetyl H4 in all of these regions following LPS stimulation and a decrease in acetyl H4 in the presence of JQ1 (Supplemental Fig. 1A). Taken together, these results show that JQ1 directly inhibits STAT5 binding to the regulatory region of target genes and that this is associated with a decrease in transcriptionally active chromatin marks.

It is known that GM-CSF, signaling through STAT5, plays an important role in the differentiation of Mo-DCs, although its role during Mo-DC maturation is not certain. This raised the possibility that JQ1 could be inhibiting STAT5 by downregulating GM-CSF signaling in Mo-DCs stimulated with LPS. To address this question, we first evaluated whether the expression of GM-CSF and the specific α subunit of its receptor (CSF2RA) is affected by STAT5 inhibition upon JQ1 treatment of Mo-DCs. Although LPS induced upregulation of both GM-CSF and CSF2RA mRNA expression, JQ1 completely blocked this effect (Fig. 1G). Consistent with this finding, the increased surface expression of the GM-CSFR (or CD116) induced by LPS was significantly abrogated by JQ1 treatment (Fig. 1G). To determine the mechanism for this effect, we analyzed the CSF2RA regulatory region and identified a strong potential STAT binding site. We then performed ChIP to assess whether STAT5 could bind to this site during LPS-induced maturation. We found that STAT5 was recruited to this site in the CSF2RA promoter region following LPS stimulation, and this was inhibited by JQ1 treatment (Fig. 1H). To determine whether STAT5 binding in the CSF2RA promoter region is associated with chromatin marks of active transcription, we performed ChIP for acetyl H4. Similar to the STAT5 binding pattern, the levels of acetyl H4 trended toward an increase following LPS stimulation and were at basal values in the presence of JQ1, indicating that STAT5 binding in the CSF2RA regulatory region is likely functional (Supplemental Fig. 1B). These data reveal CSF2RA as a novel STAT5 target gene in Mo-DCs.

Collectively, these results show that JQ1 selectively inhibits STAT5 tyrosine phosphorylation and decreases STAT5 nuclear translocation, resulting in decreased expression of STAT5 target genes. This decrease in STAT5 activity by JQ1 is associated with inhibition of expression of GM-CSF and the GM-CSFR.

Although accumulating evidence shows that STAT5 is required for in vitro differentiation of Mo-DCs, this transcription factor is also thought to be important for DC maturation (9, 25). Having found that JQ1 inhibits STAT5 activity in Mo-DCs, we next evaluated the effect of this drug on the expression of surface markers of Mo-DCs following LPS-induced maturation. Mo-iDCs were treated with JQ1 for 1 h and stimulated by LPS for another 24 h, at which point they were analyzed by flow cytometry. The increased expressions of the activation marker CD83 and the costimulatory molecules CD80 and CD86 that occur in Mo-DCs stimulated by LPS were strongly attenuated by JQ1 treatment and reverted toward the expression pattern seen in Mo-iDCs (Fig. 2A). Moreover, JQ1 inhibited the increased expression of these cell surface proteins in a dose-dependent manner (Fig. 2B). This was not a general effect of JQ1, because the expression of other molecules, such as CD14, CD11c, HLA-DR, and CD40, remained unaffected (Fig. 2A, Supplemental Fig. 2B). The unaltered expression of these molecules indicates that the cells retain the characteristics of myeloid Mo-DCs and maintain the capability for Ag presentation (signal 1, based on HLA-DR expression) and the ability to respond to the maturation signal dependent on CD40. Importantly, JQ1 did not affect the viability of Mo-DCs at any dose tested (Supplemental Fig. 2A). These findings show that JQ1-treated Mo-DCs display an impaired maturation capacity, although not a complete block of maturation.

The decreased induction of these key cell surface molecules raised the possibility that JQ1 decreased the maturation of Mo-DCs by directly inhibiting the interaction of STAT5 with the genomic regulatory sequences of these genes. To address this issue, we identified a potential STAT5 binding site in the regulatory region of CD83, and we focused on a previously described binding site in the CD80 regulatory region (26). Interestingly, we did not identify STAT5 binding sites in the proximal regulatory region of CD86, raising the question of whether STAT5 regulates CD86 through a distant enhancer locus or regulates this gene indirectly through other transcription factors. We then performed ChIP to determine the relative binding of STAT5 at the CD83 and CD80 regulatory regions. Although LPS treatment increased STAT5 binding to both the CD83 and CD80 regulatory regions, pretreatment with JQ1 reduced STAT5 recruitment, consistent with the decreased STAT5 phosphorylation and nuclear translocation observed with JQ1 treatment (Fig. 2C). Moreover, we performed ChIP for acetyl H4 to evaluate a marker of transcriptionally active chromatin in the regulatory regions of CD83 and CD80. We observed that acetyl H4, which was increased after LPS stimulation, was present at only basal levels in Mo-mDCs pretreated with JQ1 (Supplemental Fig. 1C).

Taken together, these data demonstrate that JQ1 specifically decreases CD83 and CD80 expression and reduces STAT5 recruitment to their regulatory regions, which display evidence of less transcriptionally active chromatin. This represents one significant mechanism by which JQ1 treatment can lead to the impaired maturation of Mo-DCs.

To further evaluate the effect of JQ1 on the phenotype of Mo-DCs, we analyzed whether JQ1 could also interfere with the production of proinflammatory (IL-12 and TNF) and anti-inflammatory (IL-10 and TGF-β) cytokines. As expected, Mo-iDCs had lower mRNA expression of IL-12 and TNF, but higher expression of IL-10 and TGF-β, compared with Mo-DCs following LPS-induced maturation (Fig. 2D). JQ1 blocked the increase in IL-12 and the decrease in TGF-β that occurs with LPS-induced maturation (Fig. 2D). Notably, Mo-iDCs treated with JQ1 showed no change in IL-12 or TGF-β expression (Fig. 2D). In contrast to IL-12, the increase in TNF expression with maturation was unchanged by JQ1. JQ1 treatment reduced IL-10 expression in both immature and mature Mo-DCs. This decreased expression was of similar magnitude (∼50%) for Mo-iDCs and Mo-mDCs treated with JQ1 compared with vehicle treatment (Fig. 2D). Consistent with these findings, we observed that the secretion of IL-12p70 and IL-10 by Mo-mDCs treated with JQ1 was significantly reduced compared with that of control Mo-mDCs. In contrast, the amount of TNF measured in Mo-DC supernatant was unaltered by JQ1 (Fig. 2E). These data confirm that JQ1 impairs the maturation of Mo-DCs.

Because JQ1 can inhibit the effects of other transcription factors dependent on bromodomain-containing proteins, we wished to test the hypothesis that the effect of JQ1 on DC maturation was dependent on its effect on STAT5. JQ1 repressed GM-CSF expression in Mo-mDCs (Fig. 1G). Therefore, we considered the possibility that the addition of exogenous GM-CSF could increase STAT5 activity in JQ1-treated Mo-DCs and, consequently, enhance the maturation of Mo-DCs. To address this, Mo-DCs pretreated with JQ1 were stimulated by LPS in the presence of human rGM-CSF, and the expression of CD80, CD86, and CD83 was analyzed by flow cytometry. Notably, the addition of GM-CSF did not increase the expression of CD80, CD86, and CD83 on Mo-mDCs, which suggests that decreased expression of GM-CSFR functionally restricts this pathway (Fig. 3A).

To directly assess the role of STAT5 in Mo-DC maturation, we transduced Mo-DCs with a lentiviral vector encoding eGFP alone or eGFP and caSTAT5. The caSTAT5 mutant has constitutive STAT5 transcriptional activity and can overcome the inhibitory effects of JQ1 (14). After transduction, Mo-DCs were treated with JQ1, and maturation was induced with LPS. To confirm the functional activity of caSTAT5, we measured expression of the STAT5 target genes GM-CSF and CSF2RA, both of which were prominently induced (Fig. 3B). Then, we investigated the expression of CD80, CD86, and CD83 in the caSTAT5-transduced Mo-DCs, considering only the eGFP⁺CD14⁻HLA-DR⁺ subpopulation in the analysis. Even in the presence of JQ1, caSTAT5 significantly enhanced the expression of CD86 and CD83 (although not CD80) compared with empty vector–transduced Mo-DCs (Fig. 3C). Furthermore, the levels of IL-12 were notably increased by caSTAT5-transduced Mo-DCs in the presence of JQ1, whereas TGF-β remained unchanged (Fig. 3D). Additionally, the expression of caSTAT5 resulted in the upregulation of the STAT5 target genes CisH and Bcl-x, which were previously shown to be repressed by JQ1 (Fig. 3E). These data show that the effect of JQ1 in decreasing the maturation of Mo-DCs is dependent, at least in part, on its effect on Stat5.

It was shown that stimulation of human monocytes with LPS leads to activation of STAT5, which is dependent on GM-CSF secreted by these cells (23). To determine whether a similar mechanism occurs in the maturation of Mo-DCs stimulated with LPS, we first investigated whether the activation of STAT5 induced by LPS occurs as a consequence of GM-CSF produced by Mo-mDCs or directly by LPS itself. We performed a detailed time course of STAT5 phosphorylation following LPS stimulation. We did not detect enhanced STAT5 phosphorylation at any time point up to 180 min, as would be expected to occur with a direct inducer of STAT5 activity like a cytokine (Fig. 4A). In contrast, we detected increased STAT5 phosphorylation 24 h following LPS addition, suggesting an indirect effect. Consistent with such a mechanism, we found upregulation of GM-CSF and CSF2RA occurred no earlier than 1 and 3 h, respectively, following LPS stimulation (Fig. 4B). These data suggested that LPS stimulation induces the production of both GM-CSF and GM-CSFR, which, in turn, triggers STAT5 phosphorylation.

To determine the importance of GM-CSF during the maturation of Mo-DCs induced by LPS, we used an Ab to neutralize the activity of GM-CSF and analyzed whether this affected STAT5 phosphorylation and the expression of CD80, CD86, and CD83. Mo-DCs were differentiated with IL-4 and GM-CSF for 5 d, the culture media were replaced with fresh media without GM-CSF, and Mo-DCs were stimulated with LPS in the presence or absence of anti–GM-CSF–neutralizing Ab. Blocking GM-CSF during Mo-DC maturation decreased STAT5 phosphorylation compared with control Mo-mDCs and significantly reduced the upregulation of GM-CSFR (CD116) (Fig. 4C, 4D). Furthermore, the increased expression of CD80, CD86, and CD83 induced by LPS was attenuated by anti–GM-CSF (Fig. 4E). These results show that a GM-CSF autocrine loop contributes to the maturation of Mo-DCs.

Having demonstrated that JQ1 inhibited the expression of cytokines and cell surface markers associated with Mo-DC maturation, we next wished to test the physiologic role of JQ1 inhibition on Mo-DC function. To do this, we assessed the ability of JQ1-treated Mo-mDCs to induce allogeneic T cell responses in vitro, by analyzing T cell proliferation. We cultured allogeneic CD3⁺ T cells with Mo-mDCs that had been treated with JQ1 or vehicle and analyzed the proliferation of T cells through CellTrace dilution with flow cytometry. JQ1-treated Mo-mDCs supported significantly less proliferation of both CD4⁺ and CD8⁺ T cells compared with that induced by vehicle-treated Mo-mDCs (Fig. 5). These findings show that JQ1-treated Mo-DCs only weakly stimulate T cell proliferation, consistent with their impaired maturation capacity.

To explore the pattern of the T cell–mediated response that is being induced by Mo-mDCs treated with JQ1, we next measured the cytokine profile in the coculture supernatant. Allogeneic T cells stimulated by JQ1-treated Mo-mDCs produced less inflammatory cytokines, such as IFN-γ and TNF, than did T cells cultured with untreated Mo-mDCs. Furthermore, the production of the anti-inflammatory cytokine IL-10 was significantly decreased (Fig. 6A).

Depending on the signal provided during the interaction with DCs, T cells differentiate toward proinflammatory cells (Th1) or anti-inflammatory cells (Th2 and Tregs) (27). Therefore, we determined whether JQ1-treated Mo-mDCs would alter CD4⁺ T cell polarization. We assessed the percentage of IFN-γ and TNF-producing (Th1-type) or IL-4 (Th2-type) and IL-10–producing cells by intracellular cytokine staining. Compared with the enhanced IFN-γ– and TNF-producing T cells induced by untreated Mo-mDCs, CD4⁺ T cells displayed a low percentage of cells producing IFN-γ and TNF when stimulated by JQ1-treated Mo-mDCs (Fig. 6B). In contrast, JQ1 did not alter the ability of Mo-mDCs to induce IL-4– and IL-10–producing CD4⁺ T cells (Fig. 6B).

We also found that JQ1 treatment impaired the ability of Mo-mDCs to induce IFN-γ– and TNF-producing CD8⁺ T cells (Fig. 6B). Moreover, Mo-mDCs treated with JQ1 did not modify the induction/expansion of CD4⁺CD127⁻CD25⁺ Tregs (Fig. 6C). These data indicate that JQ1-treated Mo-DCs display a decreased Th1 polarization activity while maintaining an unaffected ability to induce Th2 cells and Tregs.

To determine whether the effect of JQ1 on Mo-mDCs was causally related to STAT5 inhibition, we treated Mo-DCs with Jaki and then induced their maturation with LPS. We analyzed the phenotype and function of these Mo-mDCs through the expression of surface markers and their ability to stimulate T cell proliferation, respectively. Similar to Mo-mDCs treated with JQ1, the increased expression of the molecules CD80, CD86, and CD83 was inhibited by Jaki treatment, whereas expression of CD11c and HLA-DR was not significantly changed (Fig. 7A). Interestingly, CD40 expression also was downregulated by Jaki treatment, whereas JQ1 did not significantly affect CD40 expression (Fig. 2A). This may reflect the effect of Jaki on signaling pathways other than STAT5. Like JQ1, Jaki-treated Mo-mDCs induced less T CD4⁺ and T CD8⁺ proliferation compared with untreated Mo-mDCs (Fig. 7B). These data show that the specific inhibition of STAT5 by JQ1 was comparable to the inhibition of STATs by Jaki and indicate that STAT5 plays a key role in maintaining phenotype and function of Mo-DCs as efficient APCs (Fig. 8).

The manipulation of DCs to generate cells able to modulate the immune response is an appealing approach to treat patients with excessive inflammation and autoimmunity. In this article, we show that the BET bromodomain inhibitor JQ1 has immunomodulatory effects on DCs that could be exploited for this purpose. The presence of JQ1 during LPS-induced maturation of Mo-DCs leads to a dose-dependent reduction in CD80, CD86, and CD83 expression. In addition, JQ1-treated Mo-mDCs exhibit markedly reduced production of IL-12p70 and IL-10 but not TNF. These effects of JQ1 on Mo-mDCs depend on its inhibition of STAT5, because introduction of a constitutively active form of STAT5 can reverse the effects of JQ1. The activation of STAT5 in maturing Mo-DCs occurs through the production of GM-CSF induced by LPS stimulation, which activates a positive-feedback GM-CSF/GM-CSFR/STAT5 autocrine loop. Coculture of JQ1-treated Mo-mDCs with allogeneic T cells results in significantly reduced IFN-γ and TNF production, whereas IL-4 is unchanged. Also, the ability of JQ1-treated Mo-mDCs to trigger CD4⁺ and CD8⁺ T cell responses is lower compared with control Mo-mDCs.

The precise relevance of STAT5 for the maturation of DCs has been uncertain. Because STAT5 is a mediator of the biological effects of GM-CSF, previous reports focused more on studying the role of STAT5 during the differentiation of DCs (9, 25, 28, 29) rather than during their maturation. STAT5 is continuously activated during the maturation of murine DCs into immunogenic APCs, suggesting its involvement in this process (9, 25). We used JQ1 as a potentially clinically relevant inhibitor of STAT5 function to investigate its role in human Mo-DCs, differentiated in the presence of GM-CSF and IL-4, which are induced to mature under LPS stimulation. JQ1 was reported to inhibit STAT5 phosphorylation and transcriptional activity in a range of cancer cell lines (14, 17), but its effect in human DCs has not been described. We show that, in human LPS-matured Mo-DCs, JQ1 inhibits STAT5 phosphorylation and subsequent nuclear translocation, thereby inhibiting the transcription of target genes, including CisH, Socs2, Bcl-x, and Bcl-3. Confirming the role of JQ1 in disrupting STAT5 regulation of these genes, we found decreased DNA binding of STAT5 to the regulatory region of CisH and Bcl-3, which was correlated with regions of chromatin less transcriptionally active. CisH, a well-described specific STAT5 target gene (25, 28), can drive murine DC differentiation toward an immunogenic phenotype through GM-CSF–activated STAT5 (25). Similarly, Socs2, a regulatory molecule that belongs to the same family as CisH, appears to be essential for DC maturation induced by LPS (30). These data support our findings that the downregulation of STAT5-mediated expression of CisH and Socs2 by JQ1 could account for the decrease in Mo-DC maturation. The inhibition of STAT5 activity in Mo-mDCs by JQ1 is associated with the ability of this drug to prevent GM-CSF signaling, which is critical for LPS-induced STAT5 activation (23). We found that JQ1 blocked the ability of LPS to induce expression of both GM-CSF and its receptor (CSF2RA). Moreover, we found that STAT5 binds to the CSF2RA promoter, suggesting that CSF2RA is a novel STAT5 target gene that is directly repressed by JQ1 treatment.

The effects of JQ1 on the function of Mo-mDCs may be due to its inhibition of the expression of cell surface molecules on these cells. CD80 and CD86 upregulation was decreased by JQ1 in LPS-stimulated Mo-DCs. This resulted in defective costimulatory capacity, the second classical signal needed for naive T cell activation (31). Similar findings were demonstrated in mice lacking STAT5 in CD11c⁺ cells, which showed impaired CD80 and CD86 expression and inhibition of Th2 response upon thymic stromal lymphopoietin treatment (32, 33). Moreover, JQ1-treated Mo-mDCs also displayed reduced CD83 expression, an important maturation marker described as essential for the induction of T cell proliferation and IFN-γ production (34, 35). Interestingly, JQ1 directly decreased the recruitment of STAT5 to the CD83 regulatory region, indicating that CD83 is a potential STAT5 target gene. Contributing to the lower immunostimulatory activity of JQ1-treated Mo-mDCs, the production of IL-12 by these cells was markedly reduced compared with control Mo-mDCs, whereas TGF-β was produced at elevated levels. Interestingly, TNF was unaffected by JQ1 treatment of Mo-mDCs, and IL-10 was decreased, indicating that these molecules are not responsible for the reduced maturation capacity of these cells. These characteristics of JQ1-treated Mo-mDCs strongly suggest that this drug is impairing the full maturation of Mo-DCs, even upon LPS stimulation.

Because of the reduced expression of CSF2RA in Mo-mDCs treated with JQ1, the addition of human rGM-CSF simultaneously with LPS was unable to induce STAT5 activation or enhance the expression of CD80, CD86, or CD83 in these cells. However, introduction of caSTAT5 restored the expression of GM-CSF and CSF2RA, suggesting a positive-feedback loop involving the induction of expression of these molecules by STAT5. Notably, when activated STAT5 was restored, the defective phenotype of JQ1-treated Mo-mDCs was reversed, at least in part, as demonstrated by increased expression of CD86 and CD83 and increased IL-12 levels, thus confirming the important role for STAT5 in Mo-DC activation induced by LPS. Although we found that STAT5 binds to the CD80 regulatory region, the introduction of caSTAT5 did not rescue the expression of this molecule in the presence of JQ1, suggesting that other cooperating factors are necessary. It was reported recently that the presence of both STAT5a and STAT5b is necessary for the induction of CD80 expression in cutaneous T cell lymphoma cells (26). Thus, one possibility is that the constitutively activated STAT5a expressed in this experiment was not sufficient to mediate expression of CD80 without the presence of an activated form of STAT5b. It is also possible that STAT5 must cooperate with other transcription factors dependent on bromodomain-containing proteins, which are inhibited by JQ1.

Because STAT5 activation contributes to the expression of CD83, costimulatory molecules, and IL-12 in LPS-stimulated Mo-DCs, our data support the hypothesis that STAT5 is important for Mo-DC differentiation, as described previously (11, 13), as well as for their complete maturation. We also found that LPS induces expression of GM-CSF and GM-CSFR prior to the time when phosphorylation of STAT5 can be detected and that neutralization of GM-CSF during Mo-DC maturation blocks STAT5 activation and decreases the upregulation of GM-CSFR, CD80, CD86, and CD83, similar to the effects of JQ1. Furthermore, activated STAT5 directly induces increased GM-CSF and GM-CSFR levels, which consequently increases STAT5 signaling. These findings clearly demonstrate that this GM-CSF/STAT5 positive-feedback loop is required for full maturation of Mo-DCs.

As expected from their cell surface phenotype and cytokine-secretion pattern, JQ1-treated Mo-mDCs had a compromised ability to induce T cell proliferation. Indeed, both CD8 and CD4 T cells stimulated by JQ1-treated Mo-mDCs produced lower amounts of the proinflammatory cytokines IFN-γ and TNF. This is in agreement with the decreased IL-12 production by JQ1-treated Mo-mDCs. Furthermore, JQ1-treated Mo-mDCs did not affect Th2 generation or the induction and expansion of CD4⁺CD127⁻CD25⁺ Tregs. These results are in agreement with the direct effect of JQ1 on T cells, in which JQ1 does not interfere with the differentiation of human or murine Th2 or Tregs from naive CD4⁺ T cells (36). Thus, JQ1-treated Mo-mDCs can inhibit Th1 differentiation (inflammatory response) but not the differentiation of other T cell subtypes.

In summary, we showed that the BET inhibitor JQ1 has immunosuppressive properties. The selectivity of JQ1 in inhibiting the activating tyrosine phosphorylation of STAT5, but not other STATs, as previously reported (14), is sufficient to disrupt the phenotype and function of Mo-mDCs. Compounds with the potential to downregulate pathogenic immune responses while preserving protective immunity are highly desirable to the prevent serious adverse effects that are caused by tyrosine kinase inhibitors. Thus, JQ1 might be such a selective agent that could be used to treat immune-mediated disease. JQ1 was shown to have antioncogenic properties (14, 17) and the potential to inhibit innate immune responses by murine macrophages (18, 19). In this article, we report that JQ1 may also profoundly affect the initiation of an adaptive-immune response by reducing costimulation (signal 2) and production of proinflammatory cytokines (signal 3) in human Mo-DCs stimulated by LPS (Fig. 8). Mechanistically, the induction of STAT5 phosphorylation occurs in two phases. First, LPS induces GM-CSF and GM-CSFR (CSF2RA) expression, which initially activates STAT5. Consequently, STAT5 leads to the expression of maturation markers of Mo-DCs, as well as induces further expression of GM-CSF and GM-CSFR. This autocrine production of GM-CSF and GM-CSFR further enhances STAT5 signaling, generating a positive-feedback mechanism (Fig. 8). This is important to enable Mo-DCs to acquire a more mature phenotype, and it can be inhibited by JQ1 through targeting STAT5. Thus, our data support a significant role for GM-CSF/GM-CSFR–mediated STAT5 phosphorylation in human Mo-DC maturation and indicate that this transcription factor is the likely target of JQ1 (Fig. 8), although the involvement of other transcription factors cannot be completely ruled out. Taken together, these findings suggest that JQ1 has the potential to be a novel immunomodulatory drug that could have beneficial effects in patients with T cell–mediated inflammatory diseases.

The authors have no financial conflicts of interest.