Dendritic cells (DCs) are composed of multiple subsets that collectively provide early innate immunity, leading to subsequent adaptive immunity.¹  Plasmacytoid DCs (pDCs) and CD8α⁺ DCs produce signature cytokines, type I IFNs, and IL12 and help to establish antiviral antimicrobial states in the body. Other functions, including antigen presentation, are carried out more efficiently by CD4⁺ DCs and CD4⁻CD8α⁻ DCs. The origin of DC subsets and the mechanism underlying their development are not fully understood.^2,3  Recent studies show that IRF4 and IRF8, transcription factors of the IRF family, play critical roles in DC subset development.^4,,,,–9  Whereas the development of CD8α⁺ DCs and pDCs requires IRF8, that of the remaining subsets is dependent on IRF4. Other IRF members are also involved in modulating DC development and function.^10,–12 

BXH2 is a recombinant inbred mouse strain displaying a phenotype similar to IRF8 KO mice.¹³  They have splenomegaly with increased granulocytes and develop spontaneous myeloid leukemia. BXH2 mice are susceptible toM bovis (BCG), and are defective in IL12p40 production.¹³  Recently, Turcotte et al¹⁴  showed that BXH2 mice carry a point mutation in theIrf8/Icsbp gene that changes arginine (R) to cysteine (C) in position 294 (IRF8^R294C). The mutation is within the IRF association domain (IAD) important for the interaction of IRF8 with partner proteins.^12,15  Here we investigated the effects of the IRF8^R294C mutation on DC development.

BXH2 mice obtained from National Cancer Institute were crossed with C57BL/6J (B6) female mice for 3 generations to eliminate maternal transmission of ecotropic murine leukemia viruses (MuLVs). Progeny mice were crossed to generate mice homozygous for IRF8^R294C and wild-type IRF8 (IRF8^WT), genotyped as inFigure S2, and were used for experiments at 8 to 12 weeks of age.

Magnetic bead separation of splenic DCs, flow cytometry, and the generation of bone marrow–derived DCs (BMDCs) in fms-like tyrosine kinase ligand (Flt3L) were described.^6,9  DCs were stimulated with indicated Toll-like receptor (TLR) ligands, and induction of IL12p40 and IFNα was measured as described.^6,7 

Electrophoretic mobility shift assays (EMSAs) were performed using in vitro–transcribed and – translated IRF8^WT, IRF8^R294C, and IRF8^R289E cloned in pcDNA as described.⁶  Chromatin immunoprecipitation (ChIP) analysis was performed for BMDCs (seeFigure S4 for details).^9,16  For gene transfer experiments, IRF8^R294C in the pMSCV-puro retroviral vector was transduced into BM progenitor cells from IRF8 KO mice, and tested by flow cytometry and for cytokine expression on day 8 or day 9.^6,9 

Maternal transmission of MuLV facilitates leukemia progression in BXH2 mice initiated by the IRF8 mutation, suggesting that the phenotypes resulting from the IRF8 mutation may be modulated by MuLV.^13,14  Since our aim was to study the effect of the mutation, independent of MuLV production, we generated mice carrying IRF8^R294C without virus by crossing BXH2 mice with B6 females preventing vertical transmission of the virus.Figure 1A-D depicts flow cytometry analysis of DC subsets in IRF8^WT and IRF8^R294C spleens. The most striking feature of IRF8^R294C DC subsets was the almost total absence of CD8α⁺ DCs detected in total spleen and CD11c^high populations (Figure S1, available on theBlood website; see the Supplemental Materials link at the top of the online article). In contrast, B220⁺ pDCs expressing PDCA1 and Gr-1 were present in somewhat greater numbers in IRF8^R294C spleens. Similarly, CD4⁺ DCs were increased in the mutants, both in number and frequency. On the other hand, levels of CD4⁻CD8α⁻ DCs and the recently identified IFN-producing natural killer DCs (CD11c^int, B220⁺ CD122⁺, PDCA1⁻, Gr-1⁻)¹⁷  were similar in IRF8^WT and IRF8^R294C spleens. The total number of splenic CD11c⁺ DCs was higher in IRF8^R294C mice than in IRF8^WT mice (Figure 1C). BXH2 mice producing MuLV showed a very similar phenotype (Figure S2). Together, the representation of DC subsets in IRF8^R294C mice is markedly different from that of IRF8 KO mice, which lack both pDCs and CD8α⁺ DCs.^4,7,9  These data indicate that CD8α⁺ DC lineage development is selectively obliterated by the IRF8^R294C mutation. The increase in pDCs and CD4⁺ DCs in the mutant mice may suggest that their developmental pathways are influenced by that of CD8α⁺ DCs.

Flt3L supports the generation of CD24^high DCs in vitro, which correspond to splenic CD8α⁺ DCs, although CD8α itself is not expressed in DCs cultured in vitro.¹⁸  In line with data for splenic DCs, CD24^high cells were present in IRF8^WT cultures, but were absent in IRF8^R294C cultures (Figure 1E). B220⁺ pDCs, present in IRF8^R294C spleens, were also generated from mutant BM cells at levels similar to those of IRF8^WT mice. Following TLR stimulation by CpG, LPS, and Poly IC, IRF8^WT BMDCs expressed CD8α, as reported,⁶  while IRF8^R294C BMDCs remained negative for this marker (Figure 1F), further supporting the inability of the mutant to generate CD8α⁺ DCs.

TLR ligands stimulate IL12p40 expression in CD8α⁺ DCs, often at higher levels than in other DCs.^9,18,19  In contrast, type I IFNs are expressed at higher levels in pDCs than in other DCs.²⁰  As seen inFigure 2A, IL12p40 transcripts and protein were essentially undetectable in IRF8^R294C BMDCs following TLR stimulation, while high levels of IL12p40 were induced in IRF8^WT DCs by the same stimulation. In contrast to their failure to induce IL12p40, IRF8^R294C DCs expressed IFNα transcripts and protein in response to TLR stimulation. Thus, the IRF8^R294C mutation selectively abolished cytokine induction predominant in CD8α⁺ DCs, without affecting that in pDCs. These results reinforce the concept that the IRF8^R294C mutation selectively affects the development of CD8α⁺ DCs. The differential effect of the mutation on cytokine induction may explain why BXH2 mice are susceptible to certain pathogens but not others.¹³ 

A differential effect of the IRF8^R294C mutation on DC subset development was corroborated by gene transfer experiments in IRF8 KO BM progenitors. InFigure 2B, introduction of expression vectors for IRF8^WT but not for IRF8^R294C led to the generation of CD8α⁺ DCs, although both IRF8^WT and IRF8^R294C generated B220⁺ DCs at comparable levels. Similarly, IL12p40 induction was rescued only by IRF8^WT, while IFNα induction was rescued by both IRF8^WT and IRF8^R294C (Figure S3).

IRF8 binds to certain DNA motifs such as ISRE and Ets/IRF composite elements by interacting with partner proteins that include IRF members and Ets family proteins.^12,15  Although IRF8 can regulate gene expression through other mechanisms, the interaction with partners is thought to be a critical requirement for macrophage development.¹⁶  In EMSA analyses shown inFigure 2C, IRF8^WT interacted with IRF2, PU.1, and SpiB²¹  to bind the probes. In contrast, IRF8^R294C and the engineered IAD mutant, IRF8^R289E, did not interact with the partners and failed to bind to these elements.⁶  Thus, the R294C mutation disabled partner-dependent DNA-binding activity of IRF8 in vitro. To ascertain this defect in vivo, we performed ChIP analysis first for theCsc3 (cystatin C) promoter. Cystatin C is expressed highly in CD8α⁺ DCs and its expression depends on the IRF8-PU.1 interaction.^16,18  In agreement, cystatin C expression was high in IRF8^WT DCs, but was greatly diminished in IRF8^R294C DCs (Figure 2D). ChIP analysis showed that whereas IRF8^WT bound to the Csc3 promoter well, IRF8^R294C showed meager binding despite that PU.1 binding was comparable in both of the samples. We also noted that IRF8^R294C bound to the IFNa4 promoter, albeit less efficiently than IRF8^WT, in agreement with the expression of IFNa4 in IRF8^R294C DCs (Figure S4). Further, IRF8^R294C failed to bind to the IL12p40 promoter, again in agreement with the failure of the mutant to support cytokine expression (Figure S4). Together, our data indicate that IRF8 by interacting with partners may regulate many genes specifically required for the development and functional elicitation of CD8α⁺ DCs.^22,23  Based on the fact that the IRF8^R294C mutation did not affect pDC development, an alternative mechanism(s) of IRF8 action is likely to operate in directing pDC development and their function.¹² 

In summary, analyses of BXH2 mice revealed that IRF8 uses distinct mechanisms to direct the development of CD8α⁺ DCs and pDCs (Figure 2E). These mice will serve as a useful model to further study the role of IRF8 in DC subset gene expression, and to investigate the mechanisms of its action in directing DC and lymphocyte development.

We thank H. Dasenbrock, M. Smith, and B. Oldfield for genotyping and technical assistance; T. Kuwata for breeding suggestions; and B. Levi for discussions.

This work was supported by the Intramural Research Programs of the NIH, NICHD, and NIAID.

National Institutes of Health

Contribution: P.T., T.T., H.C.M., and K.O. designed experiments; P.T. carried out the experiments; all authors contributed to data analysis and discussions.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Keiko Ozato, Laboratory of Molecular Growth Regulation, NICHD, National Institutes of Health, Bldg 6, Rm 2A01, 6 Center Dr, Bethesda, MD, 20892-2753; e-mail:[email protected].