Zeb2 regulates commitment to plasmacytoid dendritic cell and monocyte fate

Xiaodi Wua, Carlos G. Briseñoa, Gary E. Grajales-Reyesa, Malay Haldara, Arifumi Iwataa, Nicole M. Kretzera, Wumesh KCa, Roxane Tussiwanda, Yujiro Higashib, Theresa L. Murphya, and Kenneth M. Murphya,c,1

aDepartment of Pathology and Immunology, School of Medicine, Washington University, St. Louis, MO 63110; bDepartment of Perinatology, Institute for Developmental Research, Aichi Human Service Center, Kasugai, Aichi 480-0392, Japan; and cHoward Hughes Medical Institute, School of Medicine, Washington University, St. Louis, MO 63110

Contributed by Kenneth M. Murphy, November 14, 2016 (sent for review July 20, 2016; reviewed by Christophe Benoist and Richard A. Flavell) Dendritic cells (DCs) and monocytes develop from a series of bone- have impaired migration from tissues (12, 13). Notch2 is required for marrow–resident progenitors in which lineage potential is regu- cDC2s in the spleen and mesenteric lymph node (LN) to acquire lated by distinct transcription factors. Zeb2 is an E-box–binding expression of CD4 and ESAM and produce IL-23 in response to associated with epithelial–mesenchymal transition and is pathogens (14–16). expression in cDC2s is required to induce widely expressed among hematopoietic lineages. Previously, we protective TH2 responses to Schistosoma mansoni infection (17). observed that Zeb2 expression is differentially regulated in pro- A recent study has argued that the transcription factor Zeb2 genitors committed to classical DC (cDC) subsets in vivo. Using (Sip1, Zfhx1b) regulates commitment to the cDC2 lineage by re- systems for inducible deletion, we uncover a requirement pression of Id2 (18). Zeb2 interacts with Smad and contains for Zeb2 in the development of Ly-6Chi monocytes but not neutro- N- and C-terminal domains flanking a Smad-binding phils, and we show a corresponding requirement for Zeb2 in ex- domain, homeodomain, and a C-terminal–binding protein in- pression of the M-CSF in the bone marrow. In addition, teraction domain (19). Zeb2 represses E-cadherin and other com- we confirm a requirement for Zeb2 in development of plasmacy- ponents of cell junctions during epithelial–mesenchymal transition toid DCs but find that Zeb2 is not required for cDC2 development. (20, 21), and germline deletion of Zeb2 leads to embryonic lethality Instead, Zeb2 may act to repress cDC1 progenitor specification in in mice (22, 23). Heterozygous Zeb2 defects in humans are associ- the context of inflammatory signals. ated with Hirschprung’s disease and Mowat–Wilson syndrome, and Zeb2 expression is dysregulated in several human cancers (19). In monocyte | plasmacytoid dendritic cell | transcription factor the nervous system, Zeb2 controls myelination by modulating the activity of Smads activated by bone morphogenetic proteins, mem- endritic cells (DCs) comprise several related lineages that bers of the TGF-β superfamily (24). In oligodendrocyte precursors, Dinitiate and regulate immune responses (1). Classical DCs where Zeb2 expression is low in abundance, activated Smads bind (cDCs) present antigens to prime naive T cells and produce cyto- the coactivator histone acetyltransferase p300 and activate the ex- kines to activate T cells and innate lymphoid cells. They can be pression of negative regulatory such as Id2 and Hes1;by categorized into two distinct lineages, termed cDC1 and cDC2 (2), contrast, in differentiating oligodendrocytes, expression of Olig1 and that rely on different transcription factors for their development and Olig2 induces Zeb2, which binds Smad–p300 complexes and re- + function. The cDC1 subset includes lymphoid-resident CD8α presses expression of Id2 and Hes1 (24). Within the hematopoietic + system, Zeb2 cooperates with Tbx21 (T-bet) to promote terminal cDCs and tissue-resident CD103 cDCs that function in cross- + presentation of viral antigen and defense against intracellular patho- maturation of natural killer (NK) cells and CD8 T cells (25–27), gens. The cDC2 subset includes heterogeneous populations of and its inactivation results in broadly dysregulated hematopoiesis + CD172a (Sirp-α) cDCs that promote TH17-type responses to bac- with prominent neutrophilia and loss of B cells and monocytes (28).

teria and fungi and TH2-type responses to parasites. Plasmacytoid DCs INFLAMMATION (pDCs) are a lineage distinct from cDCs identified by surface ex- Significance IMMUNOLOGY AND pression of CD45R (B220), Siglec-H, and CD317 (Bst2). They do not function directly in T-cell priming (3) but are specialized for produc- Distinct transcription factors regulate the development of im- tion of large quantities of type I IFN in response to infection (4–6). mune cell lineages, and changes in their expression can alter the DCs arise from a series of progenitors with progressively restricted balance of cell types responding to infection. Recent studies − + − − potential (1). Within lineage (Lin) Kit Sca-1 IL-7Rα bone marrow have identified Zeb2 as a transcription factor important for the + + (BM) cells, CD16/32 (FcγRII/III)loCD34 common myeloid pro- final maturation of natural killer cells and effector CD8 T cells. + genitors give rise to all myeloid lineages through FcγRII/IIIhiCD34 In this study, we show that Zeb2 is required for the development − granulocyte–macrophage progenitors (GMPs) and FcγRII/IIIloCD34 of two myeloid cell types, the monocyte and the plasmacytoid megakaryocyte–erythrocyte progenitors. Macrophage–DC pro- dendritic cell, and clarify that this factor is not required for the genitors (MDPs) differ from GMPs by decreased expression of Kit development of classical dendritic cells. and increased expression of the chemokine receptor CX3CR1. MDPs express the receptors M-CSFR and Flt3 and give rise Author contributions: X.W., W.K., T.L.M., and K.M.M. designed research; X.W., C.G.B., + + − + to Kit M-CSFR Flt3 Ly-6C –committed monocyte progenitors G.E.G.-R., M.H., A.I., N.M.K., R.T., and T.L.M. performed research; Y.H. contributed new int + + reagents/analytic tools; X.W., C.G.B., A.I., R.T., T.L.M., and K.M.M. analyzed data; and (7) and to Kit M-CSFR Flt3 common DC progenitors (CDPs). X.W. and K.M.M. wrote the paper. int − α− + From CDPs, pDCs develop via Kit M-CSFR IL-7R Flt3 Reviewers: C.B., Harvard Medical School; and R.A.F., Yale School of Medicine, Howard progenitors (8). Committed progenitors of cDCs also develop Hughes Medical Institute. from CDPs, and progenitors committed to either the cDC1 or the The authors declare no conflict of interest. cDC2 lineage have been identified in the BM and blood (9, 10). Freely available online through the PNAS open access option. Several transcription factors are required for development of DCs Data deposition: Gene expression microarray data have been deposited in the Gene (11). cDC1 development requires Irf8, Nfil3, Id2, and Batf3, whereas Expression Omnibus (accession nos. GSE87882, GSE87883, and GSE87884). pDC development requires Irf8 and Tcf4 (E2-2). cDC2 development 1To whom correspondence should be addressed. Email: [email protected]. was thought to require Irf4; however, recent analysis has shown that This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. cDC2s develop in the absence of Irf4 but lack CD4 expression and 1073/pnas.1611408114/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1611408114 PNAS | December 20, 2016 | vol. 113 | no. 51 | 14775–14780 Downloaded by guest on September 26, 2021 Previously, we and others have observed that Zeb2 is down- regulated upon specification of the CDP to the cDC1 lineage (9, 29). Id2 is induced by TGF-β and is required for development of cDC1s but is not required for development of cDC2s (30, 31). Furthermore, the balance between Id2 and E2-2 influences cDC1 and pDC development (32–34), and exogenous TGF-β applied to BM progenitors accelerates differentiation to cDCs rather than pDCs (35). Modest decreases in pDC and cDC2 frequency have been observed in mice with conditional deletion of Zeb2 in + CD11c cells, leading to the interpretation that Zeb2 regulates commitment of pDC and cDC2 lineages by controlling Id2 ex- pression (18). However, expression of CD11c occurs coordinately with lineage specification or, in the case of the committed cDC1 progenitor, actually occurs after specification (9). Thus, condi- + tional deletion of Zeb2 in CD11c cells may not fully eliminate the actions of that transcription factor during lineage specification. To address these issues, we used several systems to control the timing of Zeb2 deletion during DC development, and we find that, in contrast to Scott et al. (18), deletion in early progenitors regulates specification to the pDC lineage but not to the cDC2 lineage. This finding is consistent with reports that Id2 is required for the de- velopment of cDC1s but not cDC2s (30, 36). Finally, we found that loss of Zeb2 impaired both the expression of M-CSFR and the development of Ly-6Chi monocytes, implicating Zeb2 activity in the diversification of multiple myeloid lineages. Results We generated mice in which Zeb2 is conditionally deleted in cells expressing Cre recombinase driven by the Itgax promoter (CD11c- Cre) (14). Compared with Zeb2-sufficient (Zeb2fl/fl)mice,CD11c- Cre–driven Zeb2-deficient [Zeb2fl/fl;CD11c-Cre(tg)] mice showed substantially decreased pDC frequency in the spleen and skin- draining LN but not in the BM (Fig. 1A). Within the cDC com- partment, Zeb2-deficient mice showed an increased ratio of splenic cDC1s to cDC2s (identified by expression of CD24 and Sirp-α, respectively) and an increased ratio of equivalent cell types in other organs (Fig. 1 B and C). Neither spleen nor BM cellularity was significantly different between the two groups (Fig. S1). Using an inducible model of Zeb2 deficiency driven by the type I IFN- Fig. 1. Zeb2 is required for pDC development in vivo. (A)Samplesprepared inducible Mx1-Cre, pDCs were ablated in vivo 7–9 d after two from the spleen, skin-draining LN (SLN), and BM, harvested from mice of the treatments with poly(I:C) (Fig. 1D), and this ablation persisted for indicated genotypes, are compared for frequency of pDCs. Shown are repre- + + at least 1 y (Fig. 1E). By contrast, the ratio of CD24 :Sirp-α cDCs sentative two-color histograms comparing splenic populations (Top)andaplot was perturbed only slightly in that system (Fig. 1F). Neither spleen displaying frequency of pDCs as a proportion of all singlet lymphocytes (Bot- nor BM cellularity was significantly different between the two tom). Numbers at the Top indicate percentage of cells within the indicated gate; dots at the Bottom each represent a distinct biological replicate and are groups in the second week after poly(I:C) treatment (Fig. S1), representative of multiple independent experiments. (B and C) Samples pre- although splenomegaly was grossly evident in Zeb2-deficient mice pared in A are compared for frequency of cDCs as a proportion of all singlet + + 1 y after treatment. In summary, robust deletion of Zeb2 using lymphocytes (B) and for frequency of CD24 cDCs and Sirp-α cDCs as a pro- Mx1-Cre showed a complete dependence for this factor in pDC portion of cDCs (C). Shown is a table (B) and representative two-color histo- development and not in cDC development. grams (C). Numbers in C indicate the percentage of cells within the indicated We overexpressed Zeb2 by retroviral transduction in Kithi BM gate. (D) Samples prepared from the indicated organs, harvested from mice of progenitors cultured in the presence of Flt3L (Fig. 2A). Trans- the indicated genotypes 7–9 d after administration of poly(I:C), are compared duction efficiency was consistently lower for Zeb2-expressing for frequency of pDCs. Shown is a plot displaying frequency of pDCs as a retrovirus compared with control retrovirus, yet the transduced proportion of all singlet lymphocytes. Dots each represent a distinct biological replicate and are representative of multiple independent experiments. fraction showed markedly increased frequency of pDCs with no α (E) Samples prepared from the indicated organs, harvested from mice of the detectable change in the ratio of CD24:Sirp- cDCs (Fig. 2A). indicated genotypes 1 y after administration of poly(I:C), are compared for To test whether the requirement for Zeb2 was intrinsic to de- frequency of pDCs. Shown are representative two-color histograms (n = 3mice + + +/− veloping pDCs, we mixed CD45.1 CD45.2 Zeb2 BM cells with per group pooled over two independent, consecutive experiments). Numbers + − CD45.1 CD45.2 wild-type (WT) BM cells and cultured them indicate the percentage of cells within the indicated gate. (F) Samples pre- in vitro in the presence of the cytokine Flt3L. In this setting, we pared in D are compared for frequency of cDCs as in C. res., resident. observed that Zeb2 haploinsufficiency produced a partial defect in + the development of pDCs, whereas we observed no defect in Sirp-α − − + Next, we analyzed the development of CD19 Ly-6G KithiFlt3 cDC development from Zeb2-haploinsufficient BM (Fig. 2B). We + progenitors in the presence of Flt3L and 4-OHT, mixing CD45.2 used another model of conditional Zeb2 deletion driven by the fl/fl iCre/iCre + 4-hydroxytamoxifen (4-OHT)–inducible Gt(ROSA)26Sor-Cre- Zeb2 ;R26 cells with congenically marked CD45.1 Zeb2- ERT2 (R26-iCre) to study the effect of complete Zeb2 deficiency sufficient cells (B6.SJL) (Fig. 2D). We observed that Zeb2-defi- in vitro. As expected, 4-OHT treatment eliminated pDC devel- cient progenitors were unable to support substantial pDC devel- + + opment in Flt3L-treated cultures of Zeb2fl/fl;R26iCre/iCre BM but did opment, but CD24 cDCs and Sirp-α cDCs developed from both not perturb pDC development in cultures of WT BM (Fig. 2C). Zeb2-deficient and Zeb2-sufficient progenitors (Fig. 2D). Thus,

14776 | www.pnas.org/cgi/doi/10.1073/pnas.1611408114 Wu et al. Downloaded by guest on September 26, 2021 + source of CD24 cDCs (Fig. 3B). Thus, loss of Zeb2 in DC pro- genitors treated with IFN-α not only abrogated development of + pDCs but diverted progenitors to the CD24 cDC lineage at the + expense of Sirp-α cDC development. We examined targets of Zeb2 regulated during DC develop- ment by gene expression microarray analysis of WT and CD11c- + Cre–driven Zeb2-deficient splenic Sirp-α cDCs (Fig. 4). Few genes showed more than a threefold change between groups. Among those were Itgad, which encodes CD11d and is part of the CD11c-Cre transgenic integration (14), and Ms4a1, which encodes CD20 (Fig. S2A). Expression of Lyz2 (encoding lyso- zyme) was decreased more than twofold and that of Csf1r + (encoding M-CSFR) was halved in Zeb2-deficient Sirp-α cDCs compared with WT (Fig. 4A and Fig. S2A). Of transcription- factor–encoding genes, Ifi203 and Id2 were each increased in + expression less than twofold in Zeb2-deficient Sirp-α cDCs compared with WT (Fig. 4B). By reverse transcription quanti- tative PCR (RT-qPCR), we confirmed that Id2 expression was + increased in CD11c-Cre–driven Zeb2-deficient Sirp-α cDCs + compared with WT Sirp-α cDCs (Fig. 4C). Because Ifi204 encodes a protein known to antagonize the function of Id2 by cytoplasmic translocation (40, 41), we overex- pressed Ifi204 in BM cultured in the presence of Flt3L and analyzed its effect on DC development. However, expression of Ifi204 in- duced no substantial change in the frequency of pDCs or cDCs (Fig. S3A). Because Zeb2 promotes transcription of Smad7 in the central nervous system (24), we also overexpressed Smad7 in BM cultured in the presence of Flt3L. Again, however, we observed no sub- Fig. 2. Plasmacytoid DCs require Zeb2 in a dose-dependent and cell-intrinsic stantial change in the relative frequency of DC subsets (Fig. S3B). manner. (A) Single-cell suspensions of Kithi BM progenitors isolated from WT Using the type I IFN-inducible Mx1-Cre system, we also observed mice were cultured in the presence of Flt3L and then retrovirally transduced long-term changes in the development of other myeloid lineages 1 d later to overexpress Thy1.1 alone (Empty RV) or Zeb2 and Thy1.1 (Zeb2 RV). after deletion of Zeb2. Poly(I:C) treatment induced transient loss of Shown are two-color histograms comparing pDC and cDC frequencies among Ly-6Chi blood monocytes in both Zeb2fl/fl;Mx1-Cre(tg) mice and + transduced (Thy1.1 ) cells or 7 d after transduction. Data are representative of Zeb2fl/fl control mice; however, monocytes were restored to pre- at least two independent experiments. (B) Single-cell suspensions of whole BM + − treatment frequencies within 3 d in control mice but remained isolated from WT or Zeb2 / mice were mixed with congenically marked fl/fl + − depleted in Zeb2 ;Mx1-Cre(tg) mice (Fig. 5A). Similarly, mono- (CD45.1 CD45.2 ) single-cell suspensions of whole BM isolated from WT B6.SJL cytes were depleted in the BM and spleen of Zeb2fl/fl;Mx1-Cre(tg) mice and cultured in the presence of Flt3L. Shown is the proportion of cells fl/fl expressing CD45.2 (blue) or not expressing CD45.2 (orange) among progeny mice compared with Zeb2 control mice 1 wk after poly(I:C) within the indicated subsets. (C) Single-cell suspensions of whole BM isolated treatment (Fig. S4). By contrast, neutrophil frequency was not de- from mice of the indicated genotypes were cultured in the presence of Flt3L creased in Zeb2-deficient mice (Fig. S4). We found that depletion (FL) and either 100 nM 4-OHT dissolved in ethanol or ethanol alone (vehicle). of monocytes persisted for at least 1 y after poly(I:C)-induced Zeb2 Shown are representative two-color histograms comparing a proportion of deletion (Fig. 5B). Consistent with these observations, the fre- + + Siglec-H pDCs among progeny after 9 d of culture (n > 2 biological replicates quency of M-CSFR cells in the BM was substantially decreased hi + per group over at least two independent experiments). (D)BMKit Flt3 – INFLAMMATION + after Mx1-Cre driven Zeb2-deletion (Fig. 5C). By contrast, the IMMUNOLOGY AND progenitors isolated from R26iCre/ mice or Zeb2fl/fl;R26iCre/iCre mice were mixed − − + γ + + hi + frequency of GMPs (Lin Sca1 Kit Fc RII/III ) was not decreased with congenically marked (CD45.1 )BMKit Flt3 progenitors isolated from 1 wk after Mx1-Cre–driven Zeb2 deletion compared with Zeb2- WT B6.SJL mice and cultured in the presence of Flt3L and 4-OHT. Shown are two-color histograms comparing the proportion of cells within the indicated sufficient control mice (Fig. 5C). Using a Zeb2-GFP fusion protein reporter mouse (42), we found that Zeb2 was expressed in subsets expressing CD45.1 or CD45.2. − + all Lin M-CSFR cells in the BM, supporting the notion that Zeb2 may be required for M-CSFR expression (Fig. S5). the selective requirement for Zeb2 in pDC but not cDC devel- To determine if commitment to the monocyte lineage was im- opment was cell-intrinsic to progenitors. paired by loss of Zeb2, we used gene expression microarrays to We also examined the effect of complete Zeb2 deficiency on DC compare neutrophils and Ly-6Chi monocytes from Zeb2fl/fl control fl/fl development in vitro using the type I IFN-inducible Mx1-Cre system mice or poly(I:C)-treated Zeb2 ;Mx1-Cre(tg) mice (Fig. 6A). because type I IFN itself can induce pDC death (37) or promote Although Zeb2-deficient and Zeb2-sufficient neutrophils were − − hi + nearly indistinguishable from each other, residual Ly-6Chi mono- pDC development (38, 39). We cultured CD19 Ly-6G Kit Flt3 fl/fl progenitors in vitro with Flt3L and in the absence or presence of cytes developing in poly(I:C)-treated Zeb2 ;Mx1-Cre(tg) mice α showed increased expression of numerous genes compared with IFN- . In this setting, Zeb2-sufficient progenitors skewed away hi + Zeb2-sufficient Ly-6C monocytes (Fig. 6A). In particular, Zeb2- from CD24 cDC development and toward pDC development deficient Ly-6Chi monocytes showed increased expression of ∼50 (Fig. 3A). However, IFN-α treatment of Zeb2fl/fl;Mx1-Cre(tg) + genes that as a group are characteristic of WT neutrophils (Fig. progenitors diverted progenitors toward CD24 cDC development + 6B). For example, Ltf (encoding lactotransferrin), Mmp9 (encod- with nearly complete loss of pDCs and substantial loss of Sirp-α fl/fl ing matrix metallopeptidase 9), and Camp (encoding cathelicidin cDCs (Fig. 3A). Mixed cultures of Zeb2-sufficient and Zeb2 ; antimicrobial peptide) were more abundantly expressed in Zeb2- Mx1-Cre(tg) progenitors showed that this action of Zeb2 was cell- deficient Ly-6Chi monocytes and in WT neutrophils than in WT intrinsic because Zeb2-sufficient cells contributed exclusively to Ly-6Chi monocytes. Zeb2-deficient Ly-6Chi monocytes also + pDC development and to the majority of Sirp-α cDC develop- showed increased expression of several genes characteristic of the ment, whereas Zeb2-deficient progenitors were the predominant GMP (43) and not highly expressed in neutrophils (Fig. 6 B–D).

Wu et al. PNAS | December 20, 2016 | vol. 113 | no. 51 | 14777 Downloaded by guest on September 26, 2021 ""

Fig. 3. Type I IFN diverts Zeb2-deficient progenitors + + to the CD24 cDC lineage. (A)BMKithi Flt3 progen- itorsisolatedfrommiceoftheindicatedgenotypes were cultured in the presence of Flt3L (FL) and IFN-α. Shown are representative two-color histograms com- paring progeny of those isolated cells as analyzed after 7–7.5 d of culture (n ≥ 3 biological replicates per group over two independent experiments). (B)BM + Kithi Flt3 progenitors isolated as in A were mixed + + with congenically marked (CD45.1 )BMKithi Flt3 progenitors isolated from WT B6.SJL mice and cul- tured as in A. Shown are representative two-color histograms comparing their progeny as analyzed after 7 d of culture (Top) and one-color histograms in- + dicating the proportion of pDCs (magenta), CD24 + cDCs (cyan), and Sirp-α cDCs (orange) lacking ex- pression of CD45.1 (n ≥ 3 biological replicates per group over two independent experiments). In all panels, numbers indicate the percentage of cells within the indicated gate.

For example, Mpo (encoding myeloperoxidase), Elane (encoding cDC1s to pDCs (32–34), our evidence supports a model in which neutrophil-expressed elastase; the transcript is not highly expressed Zeb2 regulates the choice between pDC and cDC1 fate through in neutrophils), and Prtn3 (encoding proteinase 3) were more repression of Id2 (Fig. S6). abundantly expressed in Zeb2-deficient Ly-6Chi monocytes com- Previously, the similarly incomplete abrogation of pDC and pared with WT Ly-6Chi monocytes and neutrophils (Fig. 6C), and cDC2 development resulting from CD11c-Cre–driven Zeb2 de- they are also more abundantly expressed in GMPs than in Ly-6Chi letion suggested a role for this transcription factor in commitment monocytes or neutrophils (Fig. 6D). to both lineages (18). Using inducible systems to delete Zeb2 in The transcription-factor–encoding genes Id2 and are each early progenitors, we observed that Zeb2 was not essential for increased in expression in Zeb2-deficient Ly-6Chi monocytes cDC2 lineage specification and survival. Instead, we found only a compared with Zeb2-sufficient Ly-6Chi monocytes (Fig. 6E). twofold decrease in cDC2 frequency in Zeb2-deficient mice Other transcription factor–encoding genes such as Cebpe and that showed abrogation of pDC development. Furthermore, Zeb1 also showed increased expression in Zeb2-deficient Ly-6Chi Zeb2-deficient cDC2s closely resembled their WT counterparts monocytes compared with Zeb2-sufficient counterparts and are in global gene expression. Moreover, because cDC2s appear to also more highly expressed in neutrophils than in Zeb2-sufficient persist normally in Id2-deficient mice (30, 36), defects in their Ly-6Chi monocytes. Several transcription-factor–encoding genes were expressed less abundantly in Zeb2-deficient Ly-6Chi mono- cytes than in Zeb2-sufficient Ly-6Chi monocytes, including Tcf4, Prdm1 (also known as Blimp1), Klf4 [a target of Irf8 essential for monocyte development (44–46)], Irf4, Fosb,andAtf3 (Fig. 6E). In summary, deletion of Zeb2 resulted in loss of M-CSFR expression in the BM and in the long-term ablation of Ly-6Chi monocytes. Residual Ly-6Chi monocytes that develop in the ab- sence of Zeb2 showed increased expression of Id2 along with expression of genes more characteristic of neutrophils or GMPs. Discussion Here, we describe an action of Zeb2 in promoting M-CSFR expression and repressing expression of neutrophil genes in favor of monocyte development. After inducing deletion of Zeb2, we observed severe loss of M-CSFR expression on progenitor cells in the BM and a long-term impairment of monocyte develop- hi ment in the peripheral blood. Residual Zeb2-deficient Ly-6C + Fig. 4. Zeb2-deficient Sirp-α cDCs express more abundant Id2 mRNA. monocytes showed increased expression of Id2 compared with hi (A) Scatterplot comparing changes in gene expression between Zeb2-deficient WT Ly-6C monocytes, and they expressed genes characteristic Sirp-α+ cDCs and WT Sirp-α+ cDCs (y axis) to changes in gene expression be- + + of WT neutrophils and GMPs. These findings agree with pre- tween WT CD24 cDCs and WT Sirp-α cDCs (x axis). Both axes are logarithmic, vious reports that Id2 represses PU.1-mediated induction of and probe sets with less than 1.9-fold changes in expression along either di- Csf1r (47) and that M-CSFR–deficient (op/op) mice have se- mension are omitted for clarity. (B) Heat map showing relative expression in the verely decreased blood monocyte frequency (48). indicated populations as determined by gene expression microarray analysis, In agreement with a recent study (18), we confirm a re- filtered for probe sets assigned to genes encoding nuclear protein products with the greatest increase in expression in Zeb2-deficient Sirp-α+ cDCs compared with quirement for Zeb2 in the development of pDCs. We observed + WT Sirp-α cDCs. In A and B, probe sets in gray have scant expression (linear + that deletion of Zeb2 abrogated pDC development in vitro and normalized value < 64) in Zeb2-deficient Sirp-α cDCs (n ≥ 3 biological replicates in vivo. We also found that expression of Id2 was increased in per group pooled over two independent experiments). (C)Plotshowingrela- + cDC2s lacking Zeb2 in agreement with previous studies (18, 24). tive expression of Id2 mRNA normalized to Hprt mRNA in Sirp-α cDCs isolated Because Id2 overexpression inhibits pDC development (49), and from mice of the indicated genotypes as determined by RT-qPCR; each dot because the balance between Id2 and E2-2 regulates the ratio of represents a biological replicate averaged from technical replicates.

14778 | www.pnas.org/cgi/doi/10.1073/pnas.1611408114 Wu et al. Downloaded by guest on September 26, 2021 which this transcription factor regulates the development of immune lineages. Materials and Methods Mice carrying the conditional Zeb2fl [B6;129(Cg)-Zfhx1btm1.1Yhi] allele (22) were derived from biological material provided by the RIKEN BioResource Center through the National BioResource Project of the Ministry of Edu- cation, Culture, Sports, Science and Technology, Japan. Other mouse strains were obtained from The Jackson Laboratory as described in SI Materials and Methods. Mice were bred and maintained in a specific-pathogen–free

Fig. 5. Zeb2 is required for monocyte development in vivo. (A) Blood from Zeb2fl/fl and Zeb2fl/fl;Mx1-Cre(tg) mice is compared for neutrophil and monocyte frequency before and after administration of poly(I:C). Shown are two-color histograms analyzed on the indicated days. Numbers indicate the percentage of cells within the indicated gate. (B) Blood from poly(I:C)-treated mice of the indicated genotypes are compared using two gating schemes for monocyte frequency 1 y after administration of poly(I:C). Shown are representative two- color histograms (n = 3 mice per group pooled over two independent, con- secutive experiments). Numbers indicate percentage of cells within the in- dicated gate. (C) BM from poly(I:C)-treated mice are compared for M-CSFR − − + + expression and frequency of GMPs (identified as Lin Sca-1 Kit FcγRII/III )7–9d after treatment. Shown are representative two-color histograms (Top)anda plot displaying frequency of cells identified by the indicated surface markers as a proportion of all singlet lymphocytes (Bottom). Numbers at Top indicate percentage of cells within the indicated gate; dots at Bottom each represent a distinct biological replicate and are pooled from independent experiments.

development would not be expected to arise from changes in INFLAMMATION expression of Id2. As such, our results are more consistent with IMMUNOLOGY AND the interpretation that Zeb2 control of Id2 expression regulates specification between pDC and cDC1 fates. Addition of type I IFN substantially increased the pDC output hi of WT Kit progenitors in vitro. The cytokine itself can either hi Fig. 6. Zeb2-deficient Ly-6C monocytes express neutrophil-associated induce pDC death (37) or promote pDC development (38, 39), genes. (A) Volcano plots showing changes in gene expression between Zeb2- and our observation raises the possibility that a developmental deficient and Zeb2-sufficient Ly-6Chi monocytes (mo.) or between Zeb2- feed-forward loop could promote pDC development in the deficient and Zeb2-sufficient neutrophils, all isolated from poly(I:C)-treated context of viral infection. Such a mechanism would potentiate mice. (B) Scatterplot comparing changes in gene expression between Zeb2- the well-studied molecular feed-forward loop that promotes type deficient Ly-6Chi monocytes and Zeb2-sufficient Ly-6Chi monocytes (y axis) to fl/fl changes in gene expression between Zeb2-sufficient neutrophils and Zeb2- I IFN production (50, 51). In contrast, treatment of Zeb2 ;Mx1- hi Cre(tg) Kithi progenitors with type I IFN diverted nearly all sufficient Ly-6C monocytes (x axis). Compared with their expression in Zeb2-sufficient monocytes, some probe sets are more abundantly expressed progenitors to the cDC1 fate. This observation is not explained in Zeb2-deficient monocytes but not in Zeb2-sufficient neutrophils (solid by selective death of pDCs and cDC2s because we observed that green), more abundantly expressed in Zeb2-sufficient neutrophils but not in Zeb2-deficient progenitors outcompete WT progenitors in gen- Zeb2-deficient monocytes (solid red), or more abundantly expressed both in erating cDC1s in mixed cultures. Thus, pDC development re- Zeb2-deficient monocytes and in Zeb2-sufficient neutrophils (dashed yel- quires cell-intrinsic Zeb2 expression and is not compensated by low). Numbers in each outlined region indicate absolute probe set count. type I IFN signaling. Instead, Zeb2 acts to repress specification (C) Heat map showing relative expression of 22 probe sets outlined in B of progenitors to the cDC1 fate under these conditions. (solid green) among the indicated populations. (D) Relative expression of Taken together, our findings suggest that Zeb2 activity may nine probe sets clustered in C among the indicated populations in the BM (data from ImmGen). (E) Sparklines showing relative expression of tran- engage similar mechanisms to suppress alternative fates in scription-factor–encoding genes more (Top)orless(Bottom) abundantly multiple developing lineages. As lymphoid progenitors not pre- expressed in Zeb2-deficient Ly-6Chi monocytes than in Zeb2-sufficient Ly-6Chi sented here express and require Zeb2 (28), those cells and their monocytes. In all panels, results shown are from at least four biological progeny represent further avenues to explore the mechanisms by replicates in each group pooled over two independent experiments.

Wu et al. PNAS | December 20, 2016 | vol. 113 | no. 51 | 14779 Downloaded by guest on September 26, 2021 animal facility according to institutional guidelines and under protocols ACKNOWLEDGMENTS. We thank J. Michael White and the technical staff approved by the Animal Studies Committee of Washington University in of the Transgenic Knockout Micro-Injection Core in the Department of St. Louis. Pathology and Immunology at Washington University School of Medicine for mouse rederivation and the Genome Technology Access Center in the Flow cytometry samples were stained in magnetic-activated cell-sorting Department of Genetics at Washington University School of Medicine. The γ (MACS) buffer at 4 °C and, unless staining for Fc RII/III, in the presence of Fc Genome Technology Access Center is partially supported by National Can- Block (2.4G2, BD Biosciences). Antibodies and other reagents were pur- cer Institute Cancer Center Support Grant P30 CA91842 to the Siteman chased as described in SI Materials and Methods. Cells were analyzed using a Cancer Center and by Institute of Clinical and Translational Sciences/ FACSCanto II, LSR II, LSR Fortessa, FACSAria II, or FACSAria Fusion flow Clinical & Translational Science Award Grant UL1 TR000448 from the Na- cytometer (BD), and data were analyzed using FlowJo software (FlowJo). All tional Center for Research Resources, a component of the NIH, and by NIH Roadmap for Medical Research. This work was supported by the Howard gating strategies incorporated size and doublet discrimination based on Hughes Medical Institute (K.M.M.), NIH Grants 1K08AI106953 (to M.H.) forward and side scatter parameters. and 1F31CA189491-01 (to G.E.G.-R.), American Heart Association Grant Induction of gene deletion, cell preparation, cell culture, microarray analysis, 12PRE12050419 (to W.K.), and the Burroughs Wellcome Fund Career and RT-qPCR were performed as described in SI Materials and Methods. Award for Medical Scientists (to M.H.).

1. Satpathy AT, Wu X, Albring JC, Murphy KM (2012) Re(de)fining the dendritic cell 27. Omilusik KD, et al. (2015) Transcriptional repressor ZEB2 promotes terminal differ- + lineage. Nat Immunol 13(12):1145–1154. entiation of CD8 effector and memory T cell populations during infection. J Exp Med 2. Guilliams M, et al. (2014) Dendritic cells, monocytes and macrophages: A unified 212(12):2027–2039. nomenclature based on ontogeny. Nat Rev Immunol 14(8):571–578. 28. Li J, et al. (September 28, 2016) The EMT transcription factor Zeb2 controls adult 3. Colonna M, Trinchieri G, Liu YJ (2004) Plasmacytoid dendritic cells in immunity. Nat murine hematopoietic differentiation by regulating cytokine signaling. Blood, – Immunol 5(12):1219 1226. 10.1182/blood-2016-05-714659. 4. Cervantes-Barragan L, et al. (2007) Control of coronavirus infection through plas- 29. Miller JC, et al.; Immunological Genome Consortium (2012) Deciphering the tran- – macytoid dendritic-cell-derived type I interferon. Blood 109(3):1131 1137. scriptional network of the dendritic cell lineage. Nat Immunol 13(9):888–899. 5. Blasius AL, Krebs P, Sullivan BM, Oldstone MB, Popkin DL (2012) Slc15a4, a gene re- 30. Hacker C, et al. (2003) Transcriptional profiling identifies Id2 function in dendritic cell quired for pDC sensing of TLR ligands, is required to control persistent viral infection. development. Nat Immunol 4(4):380–386. PLoS Pathog 8(9):e1002915. 31. Ginhoux F, et al. (2009) The origin and development of nonlymphoid tissue CD103+ 6. Cervantes-Barragan L, et al. (2012) Plasmacytoid dendritic cells control T-cell response DCs. J Exp Med 206(13):3115–3130. to chronic viral infection. Proc Natl Acad Sci USA 109(8):3012–3017. 32. Cisse B, et al. (2008) Transcription factor E2-2 is an essential and specific regulator of 7. Hettinger J, et al. (2013) Origin of monocytes and macrophages in a committed plasmacytoid dendritic cell development. Cell 135(1):37–48. progenitor. Nat Immunol 14(8):821–830. 33. Ghosh HS, Cisse B, Bunin A, Lewis KL, Reizis B (2010) Continuous expression of the 8. Onai N, et al. (2013) A clonogenic progenitor with prominent plasmacytoid dendritic transcription factor e2-2 maintains the cell fate of mature plasmacytoid dendritic cell developmental potential. Immunity 38(5):943–957. cells. Immunity 33(6):905–916. 9. Grajales-Reyes GE, et al. (2015) Batf3 maintains autoactivation of Irf8 for commitment + 34. Ghosh HS, et al. (2014) ETO family protein Mtg16 regulates the balance of dendritic of a CD8α( ) conventional DC clonogenic progenitor. Nat Immunol 16(7):708–717. – 10. Schlitzer A, et al. (2015) Identification of cDC1- and cDC2-committed DC progenitors cell subsets by repressing Id2. J Exp Med 211(8):1623 1635. β reveals early lineage priming at the common DC progenitor stage in the bone mar- 35. Felker P, et al. (2010) TGF- 1 accelerates dendritic cell differentiation from common row. Nat Immunol 16(7):718–728. dendritic cell progenitors and directs subset specification toward conventional den- – 11. Murphy TL, et al. (2016) Transcriptional control of dendritic cell development. Annu dritic cells. J Immunol 185(9):5326 5335. α+ Rev Immunol 34:93–119. 36. Seillet C, et al. (2013) CD8 DCs can be induced in the absence of transcription factors 12. Bajaña S, Roach K, Turner S, Paul J, Kovats S (2012) IRF4 promotes cutaneous dendritic Id2, Nfil3, and Batf3. Blood 121(9):1574–1583. cell migration to lymph nodes during homeostasis and inflammation. J Immunol 37. Swiecki M, et al. (2011) Type I interferon negatively controls plasmacytoid dendritic 189(7):3368–3377. cell numbers in vivo. J Exp Med 208(12):2367–2374. + 13. Kumamoto Y, et al. (2013) CD301b dermal dendritic cells drive T helper 2 cell- 38. Li HS, et al. (2011) Cell-intrinsic role for IFN-α-STAT1 signals in regulating murine Peyer mediated immunity. Immunity 39(4):733–743. patch plasmacytoid dendritic cells and conditioning an inflammatory response. Blood 14. Caton ML, Smith-Raska MR, Reizis B (2007) Notch-RBP-J signaling controls the ho- 118(14):3879–3889. meostasis of CD8- dendritic cells in the spleen. J Exp Med 204(7):1653–1664. 39. Chen YL, et al. (2013) A type I IFN-Flt3 ligand axis augments plasmacytoid dendritic 15. Lewis KL, et al. (2011) Notch2 receptor signaling controls functional differentiation of cell development from common lymphoid progenitors. J Exp Med 210(12):2515–2522. dendritic cells in the spleen and intestine. Immunity 35(5):780–791. 40. Liu CJ, et al. (2005) The interferon-inducible p204 protein acts as a transcriptional 16. Satpathy AT, et al. (2013) Notch2-dependent classical dendritic cells orchestrate intestinal coactivator of Cbfa1 and enhances osteoblast differentiation. J Biol Chem 280(4): immunity to attaching-and-effacing bacterial pathogens. Nat Immunol 14(9):937–948. 2788–2796. 17. Tussiwand R, et al. (2015) Klf4 expression in conventional dendritic cells is required for 41. Luan Y, et al. (2008) p204 protein overcomes the inhibition of – T helper 2 cell responses. Immunity 42(5):916 928. alpha-1-mediated osteogenic differentiation by Id helix-loop-helix proteins. Mol Biol 18. Scott CL, et al. (2016) The transcription factor Zeb2 regulates development of con- Cell 19(5):2113–2126. – ventional and plasmacytoid DCs by repressing Id2. J Exp Med 213(6):897 911. 42. Nishizaki Y, Takagi T, Matsui F, Higashi Y (2014) SIP1 expression patterns in brain 19. Vandewalle C, Van Roy F, Berx G (2009) The role of the ZEB family of transcription investigated by generating a SIP1-EGFP reporter knock-in mouse. Genesis 52(1):56–67. – factors in development and disease. Cell Mol Life Sci 66(5):773 787. 43. Paul F, et al. (2015) Transcriptional heterogeneity and lineage commitment in mye- 20. Comijn J, et al. (2001) The two-handed E box binding zinc finger protein SIP1 loid progenitors. Cell 163(7):1663–1677. downregulates E-cadherin and induces invasion. Mol Cell 7(6):1267–1278. 44. Feinberg MW, et al. (2007) The Kruppel-like factor KLF4 is a critical regulator of 21. Vandewalle C, et al. (2005) SIP1/ZEB2 induces EMT by repressing genes of different monocyte differentiation. EMBO J 26(18):4138–4148. epithelial cell-cell junctions. Nucleic Acids Res 33(20):6566–6578. 45. Alder JK, et al. (2008) Kruppel-like factor 4 is essential for inflammatory monocyte 22. Higashi Y, et al. (2002) Generation of the floxed allele of the SIP1 (Smad-interacting differentiation in vivo. J Immunol 180(8):5645–5652. protein 1) gene for Cre-mediated conditional knockout in the mouse. Genesis 32(2): 46. Kurotaki D, et al. (2013) Essential role of the IRF8-KLF4 transcription factor cascade in 82–84. murine monocyte differentiation. Blood 121(10):1839–1849. 23. Van de Putte T, et al. (2003) Mice lacking ZFHX1B, the gene that codes for Smad- 47. Ji M, et al. (2008) Id2 intrinsically regulates lymphoid and erythroid development via interacting protein-1, reveal a role for multiple neural crest cell defects in the etiology – of Hirschsprung disease-mental retardation syndrome. Am J Hum Genet 72(2): interaction with different target proteins. Blood 112(4):1068 1077. 465–470. 48. Wiktor-Jedrzejczak WW, Ahmed A, Szczylik C, Skelly RR (1982) Hematological char- 24. Weng Q, et al. (2012) Dual-mode modulation of Smad signaling by Smad-interacting acterization of congenital osteopetrosis in op/op mouse. Possible mechanism for – protein Sip1 is required for myelination in the central nervous system. Neuron 73(4): abnormal macrophage differentiation. J Exp Med 156(5):1516 1527. 713–728. 49. Spits H, Couwenberg F, Bakker AQ, Weijer K, Uittenbogaart CH (2000) Id2 and Id3 + 25. van Helden MJ, et al. (2015) Terminal NK cell maturation is controlled by concerted inhibit development of CD34( ) stem cells into predendritic cell (pre-DC)2 but not into actions of T-bet and Zeb2 and is essential for melanoma rejection. J Exp Med 212(12): pre-DC1. Evidence for a lymphoid origin of pre-DC2. J Exp Med 192(12):1775–1784. 2015–2025. 50. Sato M, et al. (1998) Positive feedback regulation of type I IFN genes by the IFN- 26. Dominguez CX, et al. (2015) The transcription factors ZEB2 and T-bet cooperate to inducible transcription factor IRF-7. FEBS Lett 441(1):106–110. program cytotoxic T cell terminal differentiation in response to LCMV viral infection. 51. Hall JC, Rosen A (2010) Type I interferons: Crucial participants in disease amplification J Exp Med 212(12):2041–2056. in autoimmunity. Nat Rev Rheumatol 6(1):40–49.

14780 | www.pnas.org/cgi/doi/10.1073/pnas.1611408114 Wu et al. Downloaded by guest on September 26, 2021