Leukemia-associated activating mutation of Flt3 expands dendritic cells and alters T cell responses

Heterozygous Flt3^ITD/+ mice are healthy and have normal hematopoiesis, whereas homozygous Flt3^ITD/ITD mice display prominent myeloproliferation but only rare overt leukemia (Lee et al., 2007). As expected, young adult (5–14-wk-old) Flt3^ITD/+ mice contained normal fractions and numbers of major hematopoietic cell types, with only a slight (greater than twofold) increase in monocytes in the BM only (unpublished data). On the other hand, cDCs were significantly expanded several-fold in all lymphoid organs, including the BM, spleen, LNs, and thymus (Fig. 1, A and B). The cDC populations in nonlymphoid tissues, such as the small intestine, showed a similar expansion (unpublished data). pDCs were also significantly expanded in all organs except the thymus (Fig. 1, C and D). Homozygous Flt3^ITD/ITD mice showed an even greater increase in all DC populations (Fig. 1, A–D), although other defects such as myeloproliferation and impaired B cell development were also observed as expected (unpublished data). Overall, DC expansion represented the most prominent hematopoietic phenotype in the heterozygous Flt3^ITD/+ mice (Fig. 1 E).

To analyze early DC development, we generated Flt3^ITD/+ mice carrying the Cx₃cr1^GFP reporter allele that has been widely used to visualize early DC progenitors (Geissmann et al., 2010). DC progenitors in the resulting Cx₃cr1^GFP animals were defined as shown in Fig. S1. Whereas the earliest progenitors of monocytes and DCs (macrophage/DC progenitors) were not increased significantly, CDPs and preDCs were expanded in absolute numbers (Fig. 1 F). In contrast, Cx₃cr1^GFP-negative progenitors, including the HSC population and erythromyeloid progenitors were unaffected (Fig. S1). We conclude that the specific DC expansion in Flt3^ITD/+ mice commences at the stage of Flt3-expressing common DC progenitors.

To test whether the effect of Flt3-ITD on DCs is cell intrinsic, we generated competitive BM chimeras and tracked the donor contribution with the CD45 congenic marker. Irradiated recipients (CD45.1⁺/CD45.2⁺) were injected with a 1:1 mixture of BM cells from wild-type mice (CD45.1⁺) and either Flt3^+/+, Flt3^ITD/+, or Flt3^ITD/ITD donor (CD45.2⁺) mice. 14 wk after transfer, the fractions of cDCs from Flt3^ITD/+ and Flt3^ITD/ITD donors were expanded, whereas wild-type CD45.1⁺ competitor DCs were unaffected or reduced (Fig. 2, A and B). Similarly, the fraction of Flt3^ITD/+ donor-derived pDCs was significantly expanded (Fig. 2 C). The fraction of pDCs derived from Flt3^ITD/ITD donors was not increased in the BM or spleen, likely caused by the expansion of other lineages in these organs. Similar trends were observed with absolute numbers of cDCs and pDCs (unpublished data).

The analysis of DC progenitor populations in recipient mice revealed the prominent expansion of CDPs and preDCs derived from Flt3^ITD/+ donor BM (Fig. 2 D). Although the analysis of DC progenitors derived from Flt3^ITD/ITD donor BM was confounded by the absence of surface Flt3 (see below), an alternative gating strategy revealed a donor-specific expansion of Flt3^ITD/ITD CDPs (Fig. S2). Thus, the Flt3-ITD mutation confers a cell-intrinsic competitive advantage to mature DCs and their progenitors.

Myeloproliferation in Flt3^ITD/ITD mice is accompanied by the loss of surface Flt3 expression and was proposed to occur independently of Flt3L (Kharazi et al., 2011). However, a role of exogenous Flt3L in leukemogenesis caused by Flt3-ITD has been demonstrated (Zheng et al., 2011; Bailey et al., 2013). We therefore examined the effect of Flt3-ITD on Flt3 expression and Flt3L signaling in the DC lineage. In the aforementioned competitive chimeras, Flt3 expression was significantly down-regulated in Flt3^ITD/ITD donor-derived cDCs and pDCs but not in the corresponding wild-type cells (Fig. 3, A and B). A similar trend toward the down-regulation was observed in Flt3^ITD/+ donor-derived cells, although it reached significance only in pDCs. Moreover, CDPs from Flt3^ITD/+ and Flt3^ITD/ITD donor BM showed a progressive reduction of surface Flt3 (Fig. 3, C and D).

To test how the observed down-regulation of Flt3 expression affects Flt3L-induced DC development, we cultured whole BM cells from Flt3^+/+, Flt3^ITD/+, and Flt3^ITD/ITD mice with increasing concentrations of Flt3L. Flt3^ITD/+ BM showed a dose-dependent reduction of pDC development, whereas Flt3^ITD/ITD BM was unable to generate cDCs or pDCs (Fig. 3 E). These data suggest that Flt3-ITD down-regulates Flt3 surface expression on DCs and renders them insensitive to exogenous Flt3L signaling. This appears to be detrimental in vitro, where the emergence and differentiation of DC progenitors are driven solely by Flt3L. In contrast, it appears to be compatible with DC development in vivo, where DC progenitors may be supported by other cytokines and Flt3 is partially redundant (Waskow et al., 2008; Ginhoux et al., 2009).

Mature cDCs in lymphoid organs comprise several genetically and functionally distinct subsets that can be identified using the Cx₃cr1^GFP reporter strain (Bar-On et al., 2010; Lewis et al., 2011). These include GFP^lo Esam^hi CD11b⁺ cDCs that are efficient at CD4⁺ T cell priming; GFP^hi Esam^lo CD11b⁺ cDCs that preferentially produce cytokines; canonical GFP⁻ CD8⁺ cross-presenting cDCs; and noncanonical GFP^hi CD8⁺ cDCs (nc-CD8⁺) of unknown function. We analyzed the effect of Flt3-ITD on the subsets of cDCs using Flt3^+/+ or Flt3^ITD/+ mice carrying the Cx₃cr1^GFP reporter (Fig. 4 A). Although all four cDC subsets were significantly expanded, the GFP^hi nc-CD8⁺ cDCs were particularly increased (Fig. 4, B and C). In contrast, Flt3L administration highly and preferentially expanded the canonical GFP⁻ CD8⁺ cDC subset (Fig. 4 C), as reported previously (Bar-On et al., 2010). Thus, Flt3-ITD causes a unique pattern of DC subset expansion that is distinct from exogenous Flt3L and favors the nc-CD8⁺ cDCs.

Global gene expression analysis (Bar-On et al., 2010; and this study) suggested that nc-CD8⁺ cDCs preferentially express PD-L1, a ligand of the inhibitory receptor PD-1 expressed on T cells. In combination with CD86 or Dec205, which are expressed on the canonical CD8⁺ cDCs, PD-L1 identified the nc-CD8⁺ cDC subset independent of Cx₃cr1^GFP (Fig. 4 D and not depicted). Indeed, the PD-L1⁺ nc-CD8⁺ cDCs were preferentially expanded in Flt3^ITD/+ mice (Fig. 4 E). To confirm this notion, we crossed Flt3^ITD/+ mice with Batf3-deficient mice that lack the canonical CD8⁺ cDCs (Hildner et al., 2008; Bar-On et al., 2010). The resulting Flt3^ITD/+ Batf3^−/− mice lacked the CD86⁺ CD8⁺ cDCs as expected, but showed the same prominent expansion of PD-L1⁺ nc-CD8⁺ cDCs (Fig. 4, F and G). The canonical Batf3-dependent cDCs are required for the establishment of infection by intracellular bacterium Listeria monocytogenes (Lm; Edelson et al., 2011), and their expansion facilitates the infection (Sathaliyawala et al., 2010). Flt3^ITD/+ mice showed a significant greater than twofold increase in Lm titers in the spleen (Fig. 4 H), consistent with the corresponding increase in canonical CD8⁺ cDCs (Fig. 4, B and C). In contrast, Flt3^ITD/+ Batf3^−/− mice were completely resistant to Lm infection, suggesting that the expanded nc-CD8⁺ cDCs cannot substitute for the canonical CD8⁺ cDCs (Fig. 4 H).

We analyzed the properties and expression profile of Flt3^ITD/+ DCs, including pDCs and the four cDC subsets (canonical CD8⁺, nc-CD8⁺, CD11b⁺ Esam^hi, and CD11b⁺ Esam^lo). All Flt3^ITD/+ DCs showed normal incorporation of BrdU during a 2-d pulse, revealing that their proliferation was unchanged (Fig. 5 A). Together with increased CDP numbers (Fig. 1 F) and enhanced DC production by Flt3^ITD/+ myeloid progenitors in vitro (unpublished data), this suggests that DC expansion in Flt3^ITD/+ mice occurs at the level of progenitors rather than mature DCs. Similarly, surface expression of MHC proteins and co-stimulatory molecules such as CD80 and CD40 was normal on all Flt3^ITD/+ cDCs (Fig. 5 B). Next, we characterized global gene expression profiles of the five DC subsets from wild-type and Flt3^ITD/+ mice by RNA sequencing (RNA-Seq). Wild-type DC subsets showed the expected differential expression of signature markers and transcription factors (Fig. 6 A). Principal component analysis (Fig. 5 C) confirmed that pDCs, canonical CD8⁺ cDCs, and CD11b⁺ cDCs (both subsets) have distinct expression profiles, whereas nc-CD8⁺ resemble CD11b⁺ cDCs and pDCs (Bar-On et al., 2010). All Flt3^ITD/+ DC subsets clustered closely with their wild-type counterparts, suggesting that their expression profiles were largely preserved.

Given the dominant nature of the Flt3-ITD mutation, we focused specifically on the genes induced in Flt3^ITD/+ DCs. Several genes were induced in multiple DC subsets, including Cdh2 and Socs2, a negative feedback regulator of cytokine signaling and a common target of Flt3-ITD in AML (Sen et al., 2012; Fig. 6 B and supplemental dataset). Pairwise comparison of expression profiles showed that canonical CD8⁺ cDCs and nc-CD8⁺ cDCs harbored the largest number of induced genes (Fig. 6 B). Although most of the genes were induced only modestly, we noticed the induction of several genes associated with immunological tolerance and/or T reg cell function in both CD8⁺ cDC subsets (Fig. 6, C and D). These included amphiregulin (Areg; Zaiss et al., 2015), adrenomedullin (Adm; Rullé et al., 2012), Cox-2 (Ptgs2; Zelenay et al., 2015), and adenosine 2A receptor (Adora2a; Li et al., 2012). We conclude that Flt3^ITD/+ DCs have largely normal phenotypes and gene expression, with the few induced changes in canonical CD8⁺ cDCs and nc-CD8⁺ cDCs consistent with a tolerogenic function.

Given the important role of DCs in T cell homeostasis and especially in T reg cell maintenance, we examined T cell populations in Flt3-ITD–carrying mice. No significant differences in the numbers of conventional T (T conv) cells were observed in the BM, spleens, or LNs of these mice (Fig. 7 A). However, the fraction and absolute numbers of CD4⁺ FoxP3⁺ T reg cells were progressively increased in the BM of Flt3^ITD/+ and Flt3^ITD/ITD mice (Fig. 7 B). To test whether the observed effects on T cells were cell intrinsic, we examined T cell populations in the competitive chimeras described in the legend to Fig. 2. Donor-derived T conv and T reg cell populations were unchanged for Flt3^ITD/+ donors and markedly reduced for Flt3^ITD/ITD donors, consistent with impaired lymphopoiesis in the latter (Fig. 7, C and D). In contrast, the fraction of wild-type T conv cells in the recipients of Flt3^ITD/+ cells was significantly increased in the BM and, to a lesser extent, in the spleen (Fig. 7 C). Moreover, wild-type T reg cells were also expanded in the BM, spleen, and LN (Fig. 7 D), causing a significant increase in the T reg cell fraction out of total T cells (Fig. 7 E). No increases of competitor T cells were observed in the recipients of Flt3^ITD/ITD cells, likely reflecting the profound decrease of donor-derived T cells. We conclude that heterozygous Flt3-ITD mutation causes a cell-extrinsic expansion of T cells that favors the T reg cell population.

Given the cell-intrinsic effect of Flt3-ITD on DCs, we hypothesized that these cells may mediate the cell-extrinsic effect of the mutation on T cells. To test this notion, we set up allogeneic mixed leukocyte reactions (MLRs) with Flt3^ITD/+ splenic DCs as stimulators and wild-type T cells as responders. To differentiate conventional and regulatory T cell populations, we used T cells from the FoxP3^GFP reporter mice in which T reg cells express GFP. Compared with control Flt3^+/+ DCs, similar numbers of Flt3^ITD/+ DCs induced greater expansion of allogeneic CD4⁺ T conv, CD8⁺ T conv, and T reg cells (Fig. 8 A). Notably, the fraction of T reg cells among total responder T cells was significantly increased as a result (Fig. 8 B). Accordingly, all T cell populations stimulated with Flt3^ITD/+ DCs showed more extensive proliferation as measured by the dilution of the tracer dye Cell Trace Violet (CTV; Fig. 8 C). Isolated cDC subsets also stimulated T cell proliferation but did not yield increased T reg cell proportions (unpublished data), suggesting that the latter may require cross talk between several subsets.

To further dissect the cell-extrinsic effect of Flt3-ITD on T cells in vivo, we generated Flt3^ITD/+ Rag1-deficient mice that lack endogenous T or B cells. The transfer of wild-type T cells into lymphopenic mice causes a rapid T cell expansion that depends on microbiota-derived antigens, and a slower cytokine-driven homeostatic proliferation (Takada and Jameson, 2009). We transferred CTV-labeled wild-type T cells from FoxP3^GFP mice into control Flt3^+/+ Rag1^−/− or Flt3^ITD/+ Rag1^−/− mice. The resulting numbers of all T cell populations were significantly increased in Flt3^ITD/+ lymphopenic recipients (Fig. 8 D). The proportion of T cells that underwent homeostatic proliferation was not substantially changed (Fig. 8 E), suggesting that Flt3-ITD facilitates both the proliferation and survival of T cells in vivo in a cell-extrinsic manner.

Consistent with the preferential expansion of T reg cells by Flt3L administration, they were also preferentially expanded in the presence of Flt3-ITD–expressing cells (Fig. 7 E and Fig. 8 B). To directly analyze T reg cell homeostasis in Flt3^ITD/+ mice, we used the FoxP3^DTR-GFP strain that allows both the detection of T reg cells by GFP expression and their specific depletion by administration of diphtheria toxin (DT; Kim et al., 2007). The resulting sensitive detection of T reg cells in naive Flt3^ITD/+ FoxP3^DTR-GFP mice revealed a greater proportion of T reg cells out of T cells in the spleen (Fig. 9 A), but not in the thymus or blood (Fig. 9, A and B). Upon DT-mediated depletion, Flt3^ITD/+ mice supported a more robust reconstitution of blood T reg cells at days 10–28 (Fig. 9 B). Although T conv cells also underwent compensatory expansion, the fraction of T reg cells among total T cells was significantly increased in Flt3^ITD/+ mice on day 21 (Fig. 9 B). At the 28 d endpoint, the fraction of T reg cells among T cells was significantly increased in the spleens and thymi of Flt3^ITD/+ mice (Fig. 9 A).

Flt3L-mediated T reg cell expansion has been associated with reduced alloreactive T cell response during acute experimental GVHD (Swee et al., 2009). To test whether Flt3-ITD has a similar tolerogenic effect, we transferred wild-type B6 lymphocytes (CD45.2⁺) into nonirradiated (B6xFVB) F₁ hosts (CD45.1⁺/CD45.2⁺) that were Flt3^+/+ or Flt3^ITD/+. 1 wk later, the frequency and number of the transferred T cells were significantly reduced in Flt3^ITD/+ compared with Flt3^+/+ recipients (Fig. 9, C and D). This was accompanied by an increased fraction of T reg cells among host T cells (unpublished data). Given the expansion of PD-L1⁺ CD8⁺ cDCs (Fig. 4), we tested whether the reduced alloreactive response was dependent on the PD-L1–PD-1 pathway by treating recipients with anti–PD-L1 or anti-PD-1 antibodies. As expected (Blazar et al., 2003), PD-1 blockade led to a greater expansion of alloreactive T cells (Fig. 9 E). However, T cell expansion was still severely reduced in Flt3^ITD/+ compared with Flt3^+/+ recipients treated with either anti–PD-L1 or anti–PD-1, suggesting that the effect of Flt3-ITD is not solely dependent on this pathway.

Finally, we asked whether the dampening of T cell responses against self-antigens in Flt3^ITD/+ recipients is indeed dependent on T reg cells. To this end, we crossed Flt3^ITD/+ mice with FoxP3^sf (scurfy) mice that lack functional T reg cells and develop lethal T cell–mediated autoimmunity. If Flt3-ITD directly impaired self-reactive effector T cell responses, one would expect that the autoimmunity in Flt3^ITD/+ FoxP3^sf mice would be ameliorated. In contrast, we found that these mice died significantly sooner than Flt3^+/+ FoxP3^sf littermates (Fig. 9 F). Collectively, these results suggest that Flt3-ITD mutation causes a preferential expansion of T reg cells, which is accompanied by dampened T cell responses against self-antigens.

Constitutive activation of the growth factor receptor Flt3 represents the most common genetic abnormality in AML. Therefore, Flt3 signaling has been extensively studied in the context of leukemogenesis, where mutant variants such as Flt3-ITD are thought to promote the growth of Flt3-expressing transformed progenitors. Separately, Flt3 signaling is known as a critical regulator of DC development and homeostasis, and DC expansion represents the most striking consequence of Flt3 activation by the exogenous ligand. Here, we combined the two facets of Flt3 activation by testing the effect of Flt3-ITD on DC development and its immunological consequences.

We found that the endogenous Flt3-ITD caused an early cell-intrinsic expansion of all DC subsets and their progenitors (CDP). Because CDP did not support enhanced DC development in vitro (unpublished data), the expansion is likely to originate in the earlier Flt3⁺ progenitors, such as macrophage/DC or even HSC/multipotent progenitors. As in the case of myeloid progenitors (Kharazi et al., 2011), Flt3-ITD in DCs caused the loss of surface Flt3 expression and of the responsiveness to Flt3L, suggesting that its primary function is independent of Flt3L. The effect of Flt3-ITD on DCs was allele dosage-dependent and was highly significant even in heterozygous Flt3^ITD/+ mice that appear normal otherwise. Thus, DC lineage expansion represents a prominent consequence of the germ-line Flt3-ITD mutation in the steady state, thereby recapitulating the effects of Flt3L administration.

Exogenous Flt3L is known to expand DC subsets to various extents, particularly favoring the canonical cross-presenting CD8⁺ cDCs (O’Keeffe et al., 2002; Varol et al., 2009; Bar-On et al., 2010). The canonical CD8⁺ cDCs were expanded by only approximately twofold in Flt3^ITD/+ mice, along with the corresponding increase in bacterial titers upon Listeria infection as expected (Sathaliyawala et al., 2010). In contrast, the greatest expansion occurred in the noncanonical subset of CD8⁺ cDCs, which appear to be related to pDCs yet lack the cytokine-producing abilities of pDCs or canonical CD8⁺ cDCs (Bar-On et al., 2010). Flt3L–Flt3 signaling in DCs is mediated primarily by Stat3 (Laouar et al., 2003) and by the PI(3)K–mTOR pathway (Sathaliyawala et al., 2010). However, Flt3-ITD activates multiple pathways, including PI(3)K, Stat5, and Ras–MAPK, and their relative balance likely differs from that of the normal receptor (Parcells et al., 2006). Therefore, our results suggest that genetic and pharmacological activation of Flt3 are similar but not identical, and may involve different downstream effectors. Importantly, both CD8⁺ cDC subsets showed the largest effect of Flt3-ITD on gene expression and, notably, up-regulated a set of genes associated with tolerance.

The expansion of DCs by Flt3L administration has a complex indirect effect on T cells, both enhancing and inhibiting various T cell responses (Sela et al., 2011; Anandasabapathy et al., 2014). Importantly, it has been demonstrated to increase the number of T reg cells and facilitate their function both in mice (Darrasse-Jèze et al., 2009; Swee et al., 2009) and humans (Klein et al., 2013). The effects on T reg cells may be caused by the numerical expansion of DCs, as well as by modified functionality of Flt3L-expanded DCs on a per cell basis (Sela et al., 2011). As a result, Flt3L administration dampens T cell responses to self-antigens in the contexts of tissue inflammation and GVHD (Chilton et al., 2004; Darrasse-Jèze et al., 2009; Swee et al., 2009; Collins et al., 2012; Svensson et al., 2013). Similarly, we found that FLT3-ITD–expressing DCs more efficiently support T cell proliferation in vitro and in vivo, and particularly favor the expansion of T reg cells as confirmed in a model of specific T reg cell ablation. The observed T reg cell expansion was accompanied by reduced alloreactive T cell response during GVHD; in contrast, Flt3-ITD enhanced rather than dampened autoreactivity in the absence of T reg cells. These data suggest that FLT3-ITD resembles pharmacological Flt3 activation in creating a tolerogenic environment via a DC-mediated enhancement of T reg cell function. As such, they provide robust genetic evidence for the role of DCs in T reg cell homeostasis (Darrasse-Jèze et al., 2009; Swee et al., 2009; Bar-On et al., 2011).

Recent genome-wide analysis of human AML showed that FLT3 mutation is a relatively late event in AML progression (Jan et al., 2012; Shlush et al., 2014). Nevertheless, it may occur sufficiently early so that the resulting FLT3-ITD mutation is transmitted to normal hematopoietic hierarchy. Critically, peripheral blood DCs in AML patients were shown to express the mutant FLT3-ITD receptor and manifest higher frequencies and functional abnormalities (Rickmann et al., 2011, 2013). Indeed, the short-lived DCs are constantly replenished (Liu et al., 2007), therefore they should quickly inherit any mutation from the stem/progenitor compartment. Furthermore, increased frequency of T reg cells has been demonstrated in AML, and its reduction by various therapeutic approaches represents a promising treatment strategy for the disease (Schick et al., 2013; Yang and Xu, 2013; Bachanova et al., 2014; Govindaraj et al., 2014). These emerging data suggest that, similar to the animal model described here, FLT3-ITD in human AML patients is expressed in the DC lineage and may cause its expansion, leading to enhanced T reg cell function.

In conclusion, our study documents a novel major effect of the leukemogenic Flt3 mutation on normal hematopoiesis, including the modulation of DC and T cell numbers and function. This effect appears to precede overt leukemogenesis and involve both cell-intrinsic and -extrinsic components. Together with the described impairment of HSC maintenance (Chu et al., 2012) and of lymphopoiesis (Mead et al., 2013), our results reveal a profound modulation of normal hematopoiesis by FLT3-ITD. It is tempting to speculate that the observed modulation of DC and T reg cell function might impair immunosurveillance, thereby facilitating the escape of the mutant leukemic clone. This scenario would become directly testable with the advent of experimental models of immunosurveillance in leukemia. In addition, the observed effects on GVHD may be relevant for graft-versus-leukemia response that eliminates residual leukemic cells after allogeneic BM transplantation. This hypothetical mechanism of immunoevasion would be unique to hematopoietic tumors, which originate from the same source as the immune system and therefore might subvert the latter through leukemogenic mutations.

All animal studies were performed according to the investigator’s protocol approved by the Institutional Animal Care and Use Committee of Columbia University and New York University Langone Medical Center. Flt3^ITD/+ (Lee et al., 2007), CX3CR1^GFP/GFP (Jung et al., 2000), Batf3^−/− (Hildner et al., 2008), FoxP3^DTR-GFP (Kim et al., 2007), FoxP3^GFP (Fontenot et al., 2005), FoxP3^Sf, and Rag1^−/− mice were obtained from The Jackson Laboratory. All mice were on pure C57BL/6 (B6) background and were maintained by crossing to wild-type B6 mice. Unless indicated otherwise, Flt3^+/+ littermates of Flt3^ITD/+ or of Flt3^ITD/ITD animals were used as wild-type controls. To study allogeneic T cell responses, Flt3^ITD/+ mice were crossed to FVB/N mice (Taconic) to generate wild-type of Flt3^ITD/+ (B6xFVB) F₁ littermates. For T reg cell depletion model, FoxP3^{DTR-GFP/DTR-GFP} females were crossed to Flt3^ITD/+ males, and male progeny were studied. For T cell–mediated autoimmunity model, FoxP3^sf/+ females were bred to Flt3^ITD/+ males, and only male FoxP3^sf/y progeny were studied. Mice were analyzed at 5–14 wk of age unless otherwise noted.

Lymphoid organs were digested with collagenase D (1 mg/ml) and DNase I (20 µg/ml) in DMEM/10% FCS for 30–60 min at 37°C before generating single-cell suspensions, and red blood cells were lysed. For Flt3L-driven DC development in vitro, total BM cells were plated in triplicates in a 24-well plate (2 × 10⁶ per well) in DMEM/10% FCS with the indicated concentrations of recombinant murine Flt3L (PeproTech) and analyzed 7 d later.

Cells were stained with the following fluorochrome- or biotin-conjugated antibodies obtained from eBioscience (or another manufacturer, as indicated): anti-CD11c (N418), MHC II (M5/114.15.2), CD11b (M1/70), CD8α (53–6.7), Gr-1 (RB6-8C5), TCRβ (H57-597), CD3 (17A2), B220 (RA3-6B2), NK1.1 (PK136), Ter119 (TER-119), CD49b (DX5), CD4 (RM4-5 and GK1.5), FoxP3 (FJK-16s), CD24 (M1/69), Flt3 (A2F10), c-Kit (2B8), Sca-1 (D7), CD115 (AFS98), IL7-Rα (A7R34), CD45.1 (A20), CD45.2 (104), PD-L1 (MIH5), CD86 (GL1), DEC205 (NLDC-145; BioLegend), mPDCA-1/Bst2 (Miltenyi Biotec), Sirpα (P84, BD), CD80 (16-10A1), CD40 (1C10), and MHC I (AF6-88.5.5.3). Intracellular staining of FoxP3 was performed per the manufacturer’s instructions (eBioscience). Cell acquisition was done on LSR II or LSR Fortessa (BD) and analysis was done using FlowJo (Tree Star).

For sorting, spleens were harvested from Flt3^+/+ or Flt3^ITD/+ FVB-B6 F₁ mice and sorted on either FACSAria (BD) or MoFlo (Beckman Coulter) instruments. DC populations were sorted as follows: pDC (Bst2⁺ B220⁺), cDCs (CD11c^hi MHCII⁺ B220⁻) and subsets thereof; canonical CD8⁺ (CD8⁺ SSC^hi DEC205⁺ or CD86⁺); noncanonical CD8⁺ (CD8⁺ SSC^lo DEC205⁻ or CD86⁻); Esam^hi (CD11b⁺ Esam^hi); and Esam^lo (CD11b⁺ Esam^lo).

Recipient animals represented F₁ progeny of 129SvEvTac and B6.SJL mice (Taconic). Lethally irradiated (129xB6.SJL) F₁ recipients (CD45.1⁺/CD45.2⁺) were injected i.v. with a mixture of 10⁶ B6.SJL (CD45.1⁺) BM cells and 10⁶ Flt3^+/+, Flt3^ITD/+, or Flt3^ITD/ITD (CD45.2⁺) BM cells.

Cx₃cr1^GFP/GFP mice were injected i.p. with PBS or 5 µg recombinant Flt3L-Fc (LakePharma) every 3 d (days 0, 3, and 6) and analyzed on day 9 after initial treatment.

Flt3^+/+ or Flt3^ITD/+ mice were pulsed with an initial intraperitoneal injection with 1 mg BrdU (Sigma-Aldrich), followed by a continuous administration of 0.8 mg/ml BrdU in the drinking water for 2 d. Spleens were harvested after 2 d, treated with collagenase D (1 mg/ml) and DNase I (20 µg/ml), and analyzed for BrdU incorporation using the APC BrdU Flow kit (BD) per the manufacturer’s instructions.

Cells were sorted from splenocytes pooled from two to three animals of each genotype, and samples from independent experiments were used as biological replicates in RNA-Seq (three for CD8⁺ cDCs, two for CD11b⁺ cDCs, and two for pDCs). Sorted cells (3–5 × 10³) were pelleted or sorted directly into TRIzol reagent (Thermo Fisher Scientific), and RNA was extracted using the Arcturus PicoPure kit (Thermo Fisher Scientific). For TRIzol samples, equal volume of 70% ethanol was added to the aqueous phase of TRIzol samples and applied to columns from the PicoPure kit. Up to 250 µl of ethanol/aqueous phase mix was loaded onto the column and spun at 100 g for 2 min for each load. Bound RNA was washed, treated with DNase I (QIAGEN), and eluted as per the manufacturer’s instructions. To remove phenol contamination, eluate was resuspended in 100 µl of Wash Buffer 1 and reloaded onto a fresh column followed by elution.

cDNA libraries were prepared using the SMART-seq v4 Ultra Low Input RNA kit and Low Input Library Prep kit (Life Sciences) and were sequenced on an Illumina HiSeq 2500. Reads were mapped to Mus musculus mm10 genome using FastQ Groomer (v1.0.4) and Bowtie2 (v0.4) within the Galaxy platform (Giardine et al., 2005; Blankenberg et al., 2010; Langmead and Salzberg, 2012). Principal component analysis and identification of differentially expressed genes were performed using DESeq2 (Love et al., 2014). Differential expression was selected based on an adjusted p-value of <0.1 (false discovery rate of 10%), exhibiting at least a twofold difference in expression levels (log₂ fold change ≥1) and P < 0.05 for statistical significance. Principal component analysis was performed using calculated transformed counts (regularized logarithm) calculated by the DESeq2 package. Plots were generated using the R Studio software package.

Mice were infected with 5 × 10³ CFU of Lm expressing OVA in PBS and injected i.v. For bacterial titration, spleens were weighed and then pressed into nylon strainers to make single-cell suspensions in sterile PBS. Cells were lysed with an equal volume of sterile water with 0.1% Triton X-100 and plated on brain–heart infusion agar plates (Difco; BD) for overnight incubation at 37°C. Colonies were counted the next day.

Spleen and LNs were isolated from FoxP3^GFP reporter mice and T cells were enriched on MACS columns (Miltenyi Biotec) by negative selection with antibodies against CD11b, CD11c, B220, Gr-1, CD49b, and Ter119. T cells were stained with 5 µM CTV (Life Technologies) in PBS for 20 min at 37°C, washed, and injected i.v. into Rag1^−/− Flt3^+/+or Rag1^−/− Flt3^ITD/+ recipients (2 × 10⁶ per mouse). Splenocytes from the recipients were analyzed 5 d later.

Splenic DCs were prepared from (B6xFVB) F₁ Flt3^+/+ or Flt3^ITD/+ littermates by positive selection using biotinylated anti-CD11c, streptavidin microbeads, and MACS columns (Miltenyi Biotec). T cells from FoxP3^GFP/GFP spleen and LNs were enriched by magnetic negative selection and labeled with CTV as above. Labeled T cells were plated in a 96-well plate (5 × 10⁴/well) in RPMI/10% FCS with the indicated numbers of CD11c-enriched stimulator cells and cultured for 4 d.

FoxP3^DTR-GFP/y Flt3^+/+ and FoxP3^DTR-GFP/y Flt3^ITD/+ mice were injected i.p. with 50 µg/kg DT (Sigma-Aldrich) on two consecutive days. Blood was analyzed for the presence of GFP⁺ T reg cells on days 0 (before treatment), 2, 10, 14, 21, and 28, and the mice were sacrificed and analyzed on day 28 after injection.

RNA-seq data are available in the NCBI Gene Expression Omnibus database under the accession no. GSE76132.

Spleen and LNs were isolated from C57BL/6 (H-2^b; CD45.2) mice, pooled, and injected i.v. (3 × 10⁷ cells/mouse) into nonirradiated (B6xFVB) F₁ recipients (H-2^b/H-2^q; CD45.2/CD45.1). Spleens were harvested and analyzed 1 wk later. For antibody blockade studies, 200 µg of control IgG, anti-mPD-L1 (10F.9G2), or anti-mPD-1 (RMP1-14) were injected i.p. on the same day before graft transfer on days 0, 2, and 5, and analyzed on day 7 or 8.

Online supplemental material

Fig. S1 shows the definition of DC progenitor populations. Fig. S2 shows the definition of DC progenitors in Flt3^ITD/ITD mice. A supplemental dataset, available as an Excel file, shows differentially expressed genes between Flt3^ITD/+ and Flt3^+/+ cells in each DC subset.

We thank K. Liu, I. Ivanov, K. Sawai, K. Lewis, X.Y. Qu, M. Warren, M.R. Smith-Raska and S. Weisberg for advice and support.

This work was supported by the National Institutes of Health (NIH) grant AI072571 (B. Reizis) and training grants CA009503 and AI100853 (C.M. Lau). Cell sorting was performed at the CCTI Flow Cytometry Core, which is supported in part by NIH grant OD020056. Additional cell sorting was performed at the New York University Cytometry and Cell Sorting Core, and preparation of cDNA libraries and sequencing was performed at the New York University Genome Tech Center, both of which are supported by NIH grant CA016087.

The authors declare no competing financial interests.