Shared and Distinct Functions of the Transcription Factors IRF4 and IRF8 in Myeloid Cell Development

Interferon regulatory factor (IRF) 8 and IRF4 are structurally-related, hematopoietic cell-specific transcription factors that cooperatively regulate the differentiation of dendritic cells and B cells. Whilst in myeloid cells IRF8 is known to modulate growth and differentiation, the role of IRF4 is poorly understood. In this study, we show that IRF4 has activities similar to IRF8 in regulating myeloid cell development. The ectopic expression of IRF4 in myeloid progenitor cells in vitro inhibits cell growth, promotes macrophages, but hinders granulocytic cell differentiation. We also show that IRF4 binds to and activates transcription through the IRF-Ets composite sequence (IECS). Furthermore, we demonstrate that Irf8 -/- Irf4 -/- mice exhibit a more severe chronic myeloid leukemia (CML)-like disease than Irf8 -/- mice, involving a disproportionate expansion of granulocytes at the expense of monocytes/macrophages. Irf4 -/- mice, however, display no obvious abnormality in myeloid cell development, presumably because IRF4 is expressed at a much lower level than IRF8 in granulocyte-macrophage progenitors. Our results also suggest that IRF8 and IRF4 have not only common but also specific activities in myeloid cells. Since the expression of both the IRF8 and IRF4 genes is downregulated in CML patients, these results may add to our understanding of CML pathogenesis.


Introduction
Cell differentiation requires appropriate changes in gene expression patterns, which are tightly regulated by cell typespecific transcription factors. In case of hematopoiesis, dysregulation of these processes can result in hematopoietic disorders such as leukemias [1]. Myeloid progenitor cells, defined as granulocytemacrophage progenitors (GMPs) [2], give rise to granulocytes (such as neutrophils) or monocytes/macrophages. A number of transcription factors including PU.1, C/EBPs and Interferon Regulatory Factor 8 (IRF8) have been shown to regulate this process. While PU.1 is essential for macrophage differentiation in particular, C/EBPa and C/EBPe are the critical drivers of granulocyte differentiation [3]. We have shown previously that IRF8, a hematopoietic cell-specific factor belonging to the IRF family, directs macrophage differentiation whilst it inhibits myeloid cell growth and neutrophil differentiation [4,5]. We have also previously identified an IRF8's target DNA element termed the IRF-Ets Composite Element (IECS; represented by GAAANN[N]GGAA) and multiple direct target genes including those encoding Blimp-1, Cathepsin C and Cystatin C [6,7]. Importantly, mice lacking the Irf8 gene (Irf8 -/mice) develop a chronic myelogenous leukemia (CML)-like syndrome, in which there is a disproportionate expansion of neutrophils at the expense of monocytes/macrophages [8,9,10]. Furthermore, cells from human CML patients lack the expression of IRF8 [11], suggesting that its loss is a key aspect of human CML pathogenesis. Conservation of IRF8's function between mice and humans has been proven by a recent study demonstrating that a loss-offunction mutation in the human IRF8 gene also results in a very high neutrophil count and an absence of circulating monocytes and dendritic cells [12].
IRF4 is another hematopoietic cell-specific IRF and has the highest amino acid sequence similarity with IRF8. Consistent with this structural similarity, both IRFs have an ability to interact with the Ets transcription factor PU.1, required also for B cell differentiation, and to activate transcription via the Ets-IRF Composite Element (EICE; GGAANNGAAA) [13]. The EICE is another DNA sequence targeted by IRF and PU.1, and is active at the promoters of B cell-specific genes such as immunoglobulin light chain genes. In fact, IRF4 and IRF8 are expressed in B lineage cells and cooperatively stimulate the development of B cells [14]. In dendritic cells (DCs) consisting of multiple subsets, these two IRFs are expressed in a subset-selective manner and govern the generation of corresponding subsets [15,16] via their common and specific activities [16]. It has been reported also that IRF4, like IRF8, is expressed in macrophages [17]. However, the role of IRF4 in myeloid cell development remains poorly understood. In our current study, we have examined whether IRF4 has any roles in regulating myeloid cell growth and differentiation through gene introduction experiments and through the analysis of mice lacking Irf4 and/or Irf8.

Ethics statement
All animal experimentations were conducted in accordance with NIH and Public Health Service (PHS) policy or the Guidelines for Proper Conduct of Animal Experiments (Science Council of Japan), and all protocols were approved by Eunice Kennedy Shriver NICHD Animal Care and Use Committee (Protocol #08-010) or institutional review boards in Yokohama City University (Protocol #09-127, 10-122).

Quantitative RT-PCR
Total RNA was prepared using RNAiso Plus (Takara Bio), treated with DNase I (Invitrogen), and reverse transcribed using Primescript (Takara Bio) in accordance with the manufacturer's instructions. Quantitative PCR (qPCR) was performed in triplicate using the THUNDERBIRD SYBR qPCR Mix (Toyobo) and the ABI 7500 or StepOnePlus real-time PCR systems (Applied Biosystems) according to the manufacturers' protocols. The following primers were used: Msr1 (sense, 59-ATC ACC AAC  GAC CTC AGA CT-39; antisense, 59-CCG ATC ACC TTT  AAC ACC T- . The primers used to analyze the other genes tested in this study were as previously described [6,7,16]. Each of the primer sets produced a unique product. Data were analyzed using either the DDCT method or the standard curve method, and normalized against the Gapdh expression levels.

Reporter Assay
Reporter assays were performed as described previously [7] with slight modifications. Briefly, cells were transduced with SIRV-IECS-Ld40-GFP or SIRV-mIECS-Ld40-GFP reporter constructs, selected with puromycin, and then transduced with MSCV-CD8t vectors harboring IRF4FLAG or IRF8FLAG. Transduced cells were stained with anti-human CD8 conjugated with Cy-Chrome (BD Pharmingen). The promoter activities were analyzed using a FACSCanto II on day 2 after the transduction of MSCVs to acquire GFP signals in CD8 + cells. The data were analyzed using FlowJo software (TreeStar).

Chromatin immunoprecipitation assay
Chromatin immunoprecipitation (ChIP) assays were performed as described previously [7] with slight modifications. Briefly, cell lysates were sonicated six times for 30 sec each at 1 min intervals using Bioruptor (Cosmo Bio) to shear the genomic DNA into 200to 1000-base pair fragments. Immunoprecipitations were then performed using 2 mg of normal goat immunoglobulin G (IgG), goat anti-IRF8 antibody (C-19; Santa Cruz Biotechnology) or goat anti-IRF4 antibody (M-17; Santa Cruz Biotechnology). The primers used for qPCR were as described previously [7]. Data were analyzed using the DDCT method.

IRF4 induces macrophage differentiation and cell cycle arrest
We have previously shown that the introduction of IRF8 into Irf8 -/myeloid progenitor cell lines such as Tot2 causes their differentiation into mature macrophages concomitant with a cell cycle arrest at the G 0 /G 1 phase. To examine whether IRF4 has a similar function, we introduced this factor into Tot2 cells using a MSCV retrovirus (MSCV-IRF4FLAG-puro). As shown in Figure 1A, IRF4-transduced cells manifested morphologic changes typical of macrophages, i.e., enlargement of the cytoplasm filled with vacuoles, and shrinkage and condensation of nucleus by day 6, which were similar also to those observed in cells transduced with IRF8-FLAG. Flow cytometry analysis demonstrated that these morphological changes accompanied the expression of CD11b, F4/ 80 and M-CSF receptor, and also the loss of Gr-1 expression ( Figure 1B). Furthermore, IRF4-transduced cells exhibited strong phagocytic activity that was comparable to IRF8-transduced cells, whilst control MSCV-transduced cells did not show this activity ( Figure 1C). These results indicate that the transduction of IRF4 or IRF8 cells results in their differentiation into bona fide macrophages. Immunoblotting analysis using an anti-FLAG antibody showed that the expression level of the introduced IRF4 was considerably lower than that of IRF8, suggesting a comparable or greater ability of IRF4 than IRF8 to cause morphological macrophage differentiation ( Figure 1D). Of note, Tot2 cells do not express detectable levels of endogenous Irf4 by semi-quantitative RT-PCR (data not shown), which would explain the strong dependence of this cell line on exogenous IRF8 or IRF4 for macrophage differentiation (also see below for the detection of Irf4 expression in GMPs using a more sensitive qRT-PCR method).
qRT-PCR analysis revealed that various macrophage-related genes such as those encoding Cathepsin C (Ctsc), Cystatin C (Cst3), CSF-1 receptor/M-CSF receptor (Csf1r), Scavenger receptor (Msr1) and Blimp-1 (Prdm1), known to be induced by IRF8 [6,7], are also strongly induced upon the introduction of IRF4, whereas IRF2 failed to induce Ctsc or Cst3 (Figure 2A). On the other hand, both IRF4 and IRF8 inhibited the expression of the Cebpe gene that encodes C/EBPe, a transcription factor essential for neutrophil differentiation. In the case of Cst3, its induction by IRF4 was somewhat weaker than that by IRF8 but was still reproducibly detectable. Blimp-1 is a transcription factor that represses the Myc gene. Myc expression was in fact suppressed, and the cell cycle largely arrested at G 0 /G 1 , in cells transduced with IRF4 or IRF8 ( Figure 2B). Interestingly, we further found that IRF4 and IRF8 induce the gene encoding IRF5, another IRF that is preferentially expressed in immune cells including macrophages (see below).
These results indicate that IRF4 has abilities that are similar to IRF8 in directing myeloid progenitor cells to differentiate towards macrophages and inducing cell growth arrest.

Activation of IECS-mediated transcription by IRF4
To next examine the mechanism by which IRF4 induces macrophage differentiation, we analyzed whether IRF4 targets the IECS, which was originally identified as the DNA element targeted by IRF8 in differentiating macrophages [7]. We first performed reporter assays using the self-inactivating retrovirusbased reporter, SIRV-GFP [7]. Tot2 cells were first transduced with SIRV-IECS-Ld40-GFP (in which GFP transcription is driven by three copies of the IECS from Cst3, a direct target gene of IRF8, followed by the minimal promoter Ld40) or SIRV-mIECS-Ld40-GFP (in which the core IRF binding sequence GAAA in the IECS is mutated), and further transduced with IRF4 or IRF8. The promoter activity levels were then quantified by measuring the GFP expression levels. The results showed that both IRFs strongly induced transcription via the IECS but not the mIECS containing promoters ( Figure 3A).
We next performed chromatin ChIP assays using the same system, i.e. Tot2 cells transduced with SIRVs and IRFs. As shown in Figure 3B, both IRF4 and IRF8 bound to the IECS but not mIECS within the SIRV reporter cassette. As expected, this binding was not observed in empty MSCV-transduced cells. Likewise, both IRF4 and IRF8 bound to the endogenous Cst3 gene promoter that contains an IECS ( Figure 3C). Thus, IRF4 and IRF8 show a common ability to bind to, and thereby activate transcription, via the IECS.

Specific activities of IRF4 and IRF8
We wished to next determine whether the activities of IRF4 and IRF8 in myeloid cells are equivalent. To address this question, we screened the expression of various genes in Tot2 cells 3 days after the transduction of IRF4, IRF8, or an empty vector. The Tot2 system was used in order to compare the activities of these IRFs in an early phase of macrophage differentiation on the same platform. Although many macrophage-related genes are induced by both IRF4 and IRF8 as shown in Figure 2A, we found that there are genes specifically induced by only one of these two IRFs. For example, the Mmp12 and Mrc1 genes, encoding matrix metalloproteinase-12 and the CD206/mannose receptor and CD169, respectively, were found to be specifically induced by IRF4, whereas Itgae (encoding integrin aE) was only activated by IRF8 ( Figure 4A). These results suggest that IRF4 and IRF8 also have separate functions in macrophages.

Regulation of the innate immune responses by IRF4
To compare the impact of IRF4 and IRF8 upon TLR signaling, Tot2-derived macrophages generated by the transduction of either IRF4 or IRF8 were stimulated with the TLR4 ligand lipopolysaccharide (LPS) or the TLR9 ligand CpG DNA. The induction of Il12b and Il6 mRNAs was then measured. Consistent with the known role of IRF8 in Il12b transcription and TLR9 signaling, macrophages generated via IRF8 displayed much higher induction of Il12b upon both LPS and CpG treatment, and of Il6 upon CpG treatment, compared with those generated by IRF4, although the latter cells showed low but measurable responses despite the lack of IRF8 expression ( Figure 4B). Surprisingly, the induction of Il6 upon LPS treatment was higher in IRF4-transduced macrophages than in IRF8-transduced macrophages. The empty virus-transduced Tot2 cells did not show any induction of either gene. These results suggest that IRF4 may have a positive role, in addition to its previously reported negative role [18,19], in TLR signaling and/ or the transcriptional regulation of cytokine genes.
As a possible mechanism, we found that both IRF4 and IRF8 strongly induce Irf5 transcripts in Tot2 cells (Figure 2A). It has been shown that IRF5 binds to MyD88 upon recognition of various TLR stimuli, and then activated IRF5 functions as a direct transcriptional activator of multiple cytokine genes including Il6 and Il12b [20]. Thus, IRF5 may, in part, mediate the positive effects of IRF4 and IRF8 on the innate immune responses in macrophages.

IRF4 inhibits neutrophil differentiation
We next sought to determine whether IRF4 has an ability to regulate neutrophil differentiation in a similar manner to IRF8. To this end, we employed 32Dcl.3 myeloid progenitor cells, which differentiate towards neutrophils when the supplemented cytokine is switched from IL-3 to G-CSF. In the presence of IL-3, the transduction of IRF4 or IRF8 did not affect cell viability or the proliferation of 32Dcl.3 cells (Figure 5A and C). In empty vectortransduced control cells, a 7-day culture with G-CSF showed morphological changes typical of neutrophils such as band/ segmented nuclei in approximately 40% of the cells, whereas in cells transduced with IRF4 or IRF8, the same G-CSF treatment resulted in the appearance of mostly immature intermediate cells ( Figure 5A and D). The expression levels of the introduced IRF4 and IRF8 were comparable ( Figure 5B) and qRT-PCR analysis of the IRF4 or IRF8-transduced cells revealed that the induction of the gene encoding G-CSF receptor (Csf3r) upon G-CSF treatment was three-fold lower than in the control cells ( Figure 5E). Indeed, the growth response to G-CSF was approximately seven-fold less than that of the control cells ( Figure 5F). These results indicate that both IRF4 and IRF8 have an ability to inhibit neutrophil growth and differentiation.
A more severe CML-like disease in mice doubly deficient for Irf4 and Irf8 To gain insight into the role of endogenous IRF4 in myeloid cell development, we analyzed mice that are deficient in Irf8, Irf4 or both (Irf8 -/-, Irf4 -/or DKO, respectively). Splenomegaly, known to occur in Irf8 -/mice [8], was not observed in Irf4 -/mice at the age examined (7-9 weeks). However, DKO mice exhibited a more severe splenomegaly than Irf8 -/mice ( Figure 6A), indicating a possible role of IRF4 that might be ''hidden'' if Irf8 is intact. Flow cytometric analysis of splenocytes and bone marrow cells revealed that the increase in both the percentages and absolute numbers of CD11b + Gr1 + granulocytes, known to occur in Irf8 -/mice, was markedly augmented in DKO mice, whilst Irf4 -/mice did not show an obvious increase ( Figure 6B and C). On the other hand, the F4/80 + macrophage counts, known to decrease in Irf8 -/mice, were more severely diminished in DKO mice. Accordingly, the ratios of the granulocyte to macrophage numbers in the spleens of WT, Irf8 -/-, Irf4 -/-, and DKO mice were 3.3, 88.1, 6.0, and 1023.5, respectively ( Figure 6C). Taken together, our data reveal that DKO mice exhibit a more severe myeloid development abnormality resembling CML compared with Irf8 -/mice, in which there is a disproportionate expansion of granulocytes at the expense of monocytes/macrophages. After we completed this work, Ren and colleagues also reported the expansion of granulocytes in Irf4 -/-Irf8 -/mice [21]. In addition, the authors investigated the long-term fate of the mutant mice to show cooperative tumor suppressive functions for IRF4 and IRF8 in both myeloid and lymphoid cells.

Expression of Irf4 and Irf8 mRNAs at various differentiation stages in hematopoietic cells
To better understand why the loss of Irf4 alone does not cause obvious abnormalities in myeloid cell development, whilst DKO mice show more severe defects than Irf8 -/mice, we examined the expression of endogenous Irf4 and Irf8 in CMPs, GMPs, macrophages and granulocytes, along with several other types of hematopoietic/immune cells. Whereas Irf8 was clearly expressed in CMPs and GMPs, the expression of Irf4 was very low in both of these progenitors ( Figure 7A). This result provides a basis of the role of IRF8 being predominant and that of IRF4 becoming visible only when IRF8 is absent in myeloid progenitor cells. The Irf4 transcript levels in Irf8 -/-GMPs were found to be comparable with those in WT GMPs, indicating that it is unlikely that IRF8 functions as an IRF4 inducer ( Figure 7B). Consistent with the common functions of these IRFs in stimulating and inhibiting the differentiation of macrophages and granulocytes, respectively, the expression of both Irf4 and Irf8 was found to be increased in macrophages but decreased to very low levels in granulocytes ( Figure 7A).

Discussion
We demonstrate in our current study that IRF4 has similar functions to IRF8 in the regulation of differentiation and growth in myeloid cells. Moreover, our present data suggest that each of these IRFs also has separate functions in macrophages.
In vitro differentiation experiments showed that IRF4, like IRF8, inhibits myeloid cell growth, and tunes the balance of lineage selection in myeloid progenitor cells by stimulating macrophage differentiation whilst inhibiting neutrophil differentiation. The results of reporter and ChIP assays revealed that IRF4 directly targets the IECS to activate the transcription of macrophagerelated genes, clearly indicating that IRF4, as well as IRF8,  functions intrinsically in myeloid cells. In support of these in vitro findings, Irf8 -/-Irf4 -/mice display more severe CML-like symptoms than Irf8 -/mice. However, mice singly deficient in Irf4 did not show any obvious abnormalities in their granulocyte and macrophage counts. One possible reason for this is the differential expression levels of IRF4 and IRF8 in myeloid progenitors (CMPs and GMPs). We observed by qRT-PCR that the expression level of Irf8 in CMPs and GMPs is far higher than that of Irf4, whilst both IRF genes were clearly expressed in macrophages and at very low levels in granulocytes. We propose that during lineage selection in GMPs i.e. differentiation into macrophages or granulocytes, the role of IRF8 predominates but IRF4, expressed at a low level, can partially compensate for the absence of IRF8.
The activities of IRF4 and IRF8 in myeloid cells are not fully equivalent however. Whereas many macrophage-related genes such as Ctsc, Cst3, Csf1r, Msr1, Prdm1, and Irf5 are induced both by IRF4 and IRF8, several genes were found to be induced by only one of the two IRFs. Specifically, IRF4 induces Mmp12 and Mrc1, whilst only IRF8 induces Itgae. The specific induction of Mmp12 and Itgae is reminiscent of a similar observation made previously in dendritic cells [16]. Mannose receptor is an M2 macrophagerelated molecule [22] (see also below). Global gene expression profiling by microarray analysis supports the presence of both common and specific downstream genes for IRF4 and IRF8 (A.N., M.Y. and T.T., unpublished results). These results suggest that they have both common and specific activities, which are likely to confer basic features and functional diversity, respectively, upon macrophages.
It has been reported that IRF4 and IRF8 play distinct roles in innate immune responses. For example, IRF4 and IRF8 have been shown to affect the TLR-MyD88 pathway negatively and positively, respectively [18,19,23,24]. Only IRF8 is induced by IFN-c, the cytokine once referred to as the macrophage-activating factor, and participates in the transcriptional activation of IL-12p40 [25,26]. Our present data indeed demonstrate that IRF8 is far more potent than IRF4 in supporting TLR9 signaling and inducing the Il12b gene. Whilst the accumulating evidence suggests that IRF8 is essential for M1 macrophage polarization via the induction of IL-12p40, Akira and colleagues have reported very recently that IRF4 is essential for the M2 polarization [27]. We also observed that IRF4 but not IRF8 induces Mrc1, that encodes an M2 macrophage-related molecule [22]. Yet, our current study indicates a more complex picture of the role played by IRF4; it stimulates LPS (but not CpG) induction of the gene encoding the M1 cytokine IL-6 more strongly than IRF8 [22]. The ''positive'' effects of IRF4 on the innate immune responses may be mediated, at least in part, by the induction of Irf5 because IRF5 is essential for the signaling that occurs through various TLRs and is implicated in the transcriptional activation of multiple cytokine genes including Il6 [20]. However, the induction of Irf5 occurs also via IRF8, suggesting that this represents a previously unrecognized ''common and positive'' effect of IRF4 and IRF8 on the innate immune responses. The mechanism underlying the ''specific and positive'' potential of IRF4 requires further investigation. Because the heterogeneity of macrophages has been progressively uncovered, it will be important to further investigate the usage of IRF4 and IRF8 in different macrophage subsets.
It is noteworthy that IRF4 and IRF8 are essential for the development of all professional antigen presenting cells (APCs) including macrophages (this study), DCs and B cells [14,16,28]. In these APCs, IRF4 typically stimulates Th2 responses, whilst IRF8 induces Th1 responses [16,27,29,30]. Furthermore, IRF4 and IRF8 also play a role in T cells where IRF4 is required for Th2, Th17 and Th9 responses, whereas IRF8 inhibits Th17 responses [31,32,33,34]. Hence, the similar but distinct transcription factors IRF4 and IRF8, which are present only in vertebrates, make a critical contribution to diverse immune responses by acting in professional APCs as well as in T cells, both of which are the hallmark of the vertebrate immune system.
It has been shown previously that not only IRF8 but also IRF4 transcript levels are significantly diminished in human CML patients [11,35,36]. Moreover, IRF8 and IRF4 expression correlates with the cytogenetic response to IFN-a. Importantly, these observations are not secondary phenomena due to the expansion of neutrophils that do not express these IRFs, because the expression of IRF8 and IRF4 is also diminished in sorted B and T cells, respectively. The Bcr/Abl kinase inhibitor imatinib has replaced IFN-a as the first-line therapy for CML. However, this inhibitor cannot effectively eliminate leukemic stem cells [37,38], and if the drug is discontinued, most patients eventually relapse. The next generation of therapies for CML is thus eagerly awaited. Notably, it is well established that CML cells are highly sensitive to T cell-mediated immunity [39]. Naïve T cells are activated by professional APCs, particularly DCs and macrophages, whose differentiation and function are cooperatively regulated by IRF8 and IRF4, as revealed by the current and previous studies. In this regard, it will be interesting to more closely examine whether CML patients have any defects in the development and function of their professional APCs. Thus, the lack of IRF8 and IRF4 expression is likely to be critically involved in the pathogenesis of Values are mean 6 standard deviation from measurements of 5 to 6 spleens of each genotype. *P,0.01 (Student's t-test). (B) Flow cytometric analysis of granulocytes (CD11b + Gr1 + ) and macrophages (F4/80 + ) in bone marrow cells (upper panels) and splenocytes (lower panels). Numbers indicate the percentages of granulocytes and macrophages. Data are representative of three independent experiments with similar results. (C) The absolute numbers of granulocytes and macrophages per femur (upper part) or spleen (lower part). The ratios of granulocytes to macrophages are shown in the right panels. Values are mean 6 standard deviation from 3 to 5 mice of each genotype. *P,0.01 and **P,0.05 (Student's t-test). doi:10.1371/journal.pone.0025812.g006 human CML, and seeking a way to restore the expression and function of these IRFs could be a powerful new approach to the improvement of CML therapy.