Insulin–InsR signaling drives multipotent progenitor differentiation toward lymphoid lineages

To explore the role of insulin signaling in hematopoiesis, we first checked expression levels of InsR in the hematopoietic system. InsR was constitutively expressed in all the hematopoietic progenitors and had a higher expression level in MPPs than other populations (Fig. 1 A and Fig. S1 A). The expression levels of InsR in LSKs were further validated by confocal microscopy (Fig. 1 B). We then generated MxCre⁺;Insr^flox/flox mice by crossing MxCre mice with Insr^flox/flox mice (Brüning et al., 1998). After injecting poly(I:C) three times in 5 d to induce expression of the Cre recombinase, Insr was deleted thoroughly in the BM of Insr conditional knockout mice (Fig. 1 C; hereafter called Insr^+/+ for control mice and Insr^−/− for KO mice). Insr^−/− mice showed no abnormal phenotypes, and Insr^−/− body weights were similar to their littermate Insr^+/+ control mice. Importantly, Insr^−/− mice displayed enlarged spleens (Fig. 1 D), with expansion of spaces between lymphoid follicles (Fig. 1 E). Moreover, Insr^−/− spleens had an expansion of myeloid cells but a reduction of lymphoid cells (Fig. 1 F). The skewing myeloid/lymphoid ratio was also observed in BM examined by flow cytometry using markers for myeloid cells (Fig. 1 G) and lymphoid cells (Fig. 1 H).

Dynamic analysis of BM cells showed that myeloid cells in BM increased gradually after Insr deletion (Fig. 1 I), accompanied by declined lymphoid cells (Fig. 1 J). Moreover, mature cells positive for specific myeloid markers were significantly increased (Fig. 1 K), whereas B cells were markedly decreased in Insr KO mice (Fig. 1 K). When we checked peripheral blood compositions, we found that myeloid-derived cells were dramatically increased, but lymphoid-derived cells were decreased (Table 1 and Fig. 1 L). Additionally, absolute numbers of lymphoid cells were markedly declined in Insr KO BM 16 wk after Insr deletion (Fig. 1 M). Collectively, these results indicate that Insr deficiency increases myeloid cells but reduces lymphoid cells.

We stained Insr^+/+ and Insr^−/− BM cells with several cellular surface markers that could distinguish between self-renewable stem cells and differentiated progenitor cells (Fig. 2 A and Fig. S1 B). As expected, progenitors for granulocyte/monocyte progenitors (GMPs) were remarkably increased in Insr^−/− BM cells (Fig. 2 B), whereas CLPs were decreased. However, cell numbers of self-renewable hematopoietic stem cells (HSCs) and MPPs in the LSK population were unchanged after Insr deletion (Fig. 2 B). We then intraperitoneally injected BrdU (7.5 mg/kg) into Insr^+/+ and Insr^−/− mice for 18 h, and then examined BrdU incorporation in isolated GMPs and CLPs. We found the percentages of BrdU-positive cells were comparable between Insr^+/+ and Insr^−/− cells (Fig. 2 C). Moreover, the S/G2/M state was also comparable between Insr^+/+ and Insr^−/− cells through cell cycle analysis with Hoechst 33342 and pyronin Y staining (Fig. 2 D). Finally, we checked the activation state of caspase-3, an early marker of apoptosis. We found that the apoptotic cells were comparable between Insr^+/+ and Insr^−/− cells (Fig. 2 E). These results suggest that the shifted ratio of the GMPs and CLPs might be caused by differentiation commitment of the LSK population.

Self-renewal is essential for HSCs to maintain their population long term, and altered self-renewal activity can affect the numbers of downstream populations (Riddell et al., 2014; Tamplin et al., 2015). To test whether the self-renewal property was changed in Insr^−/− LSK cells, we used a long-term culture initiating cell (LT-CIC) assay to detect the long-term self-renewal ability of Insr deficient cells. Consistent with the unchanged composition of LSKs determined by flow cytometry, percentages of colony-forming cells (CFCs) were comparable between Insr^−/− and Insr^+/+ LSKs (Fig. 2 F), indicating that the skewed myeloid/lymphoid ratio in Insr^−/− mice was not caused by abnormality in self-renewal of LSKs. Moreover, when we sequentially transplanted Insr^+/+ and Insr^−/− LSKs into recipient mice, similar numbers of reconstituted LSKs were observed (Fig. 2 G), suggesting that the self-renewal capacity of LSKs was not affected by Insr deletion. We then used an in vitro assay to test the differentiation potential of Insr^−/− LSKs. Insr^−/− LSKs exhibited enhanced differentiation potential to myeloid cells in CFC assays (Fig. 2 H), but decreased lymphoid differentiation potential when co-cultured with OP9 cells (Fig. 2 I). However, the differentiation potential toward erythrocytes was not changed after Insr depletion (Fig. 2 J). These data suggest that Insr deletion biases LSKs to differentiate into myeloid cells.

To further verify that the biased differentiation potential of Insr^−/− LSKs was intrinsic, we transplanted BM cells from untreated MxCre⁺;Insr^flox/flox mice to lethally irradiated CD45.1 recipient mice. After successfully reconstituted the recipient BM, poly(I:C) was administrated to induce deletion of the Insr gene. The numbers of myeloid and lymphoid cells were examined by flow cytometry. Expectedly, mice reconstituted with MxCre⁺;Insr^flox/flox LSKs had elevated numbers of GMPs, but decreased numbers of CLPs (Fig. 2 K). Subsequently, neutrophils were significantly increased, whereas lymphocytes were decreased in the peripheral blood of MxCre⁺;Insr^flox/flox LSK-reconstituted mice (unpublished data). Moreover, in competitive transplantation experiments, in which 100 Insr^−/− LT-HSC cells and 3 × 10⁵ CD45.1 BM cells were together transplanted into lethally irradiated CD45.1 mice, the skewed myeloid/lymphoid ratio was still observed (Fig. 2 L). Collectively, our results suggest that Insr-deficient LSKs tend to differentiate into myeloid cells but not lymphoid cells.

We next wanted to determine whether insulin could regulate LSK differentiation through InsR engagement. We in vitro cultured isolated LSKs with or without insulin. We found that insulin stimulation could cause InsR activation in LSKs (Fig. 3 A). After being cultured in CFC-assay medium for 10 d, MPPs pretreated with insulin had a declined myeloid colony number, whereas cells pretreated without insulin had an increased myeloid colony number (Fig. 3 B). However, insulin challenge had no effect on the differentiation potential of Flk2⁻LSKs (Fig. 3 B). In contrast, when co-cultured with OP9 for lymphoid differentiation, MPPs pretreated with insulin had more lymphoid cells, whereas MPPs without insulin had fewer lymphoid cells (Fig. 3 C). In addition, insulin incubation did not cause the changes of cell cycle state and apoptotic cell rate in Flk2⁻LSKs and MPP cells (Fig. 3, D and E). To further determine the physiological role of insulin signal in MPP differentiation, we injected insulin in mice to elevate insulin levels. Importantly, we found that insulin injection caused significantly decreased myeloid progenitor cells (Fig. 3 F), while dramatically increasing lymphoid cells in the BM.

To more clearly delineate the effect of insulin on LSKs in vivo, we monitored the insulin levels of mice during one feeding cycle (Casanova-Acebes et al., 2013). The insulin level was higher at the end of the night (8:00 am), while it was lower at the end of the day (8:00 pm; unpublished data). Consequently, there were more lymphoid cells but fewer myeloid cells at the end of the night compared with those of the day (unpublished data). When mice were starved to decline their circulating insulin levels, we found that starvation-treated mice had more myeloid cells but fewer lymphoid cells than those of control-treated mice (unpublished data), suggesting that the insulin–InsR signaling might be required for the fate decisions of LSKs.

To further determine whether insulin–InsR signaling indeed directs differentiation of LSKs, Insr^−/− LSKs rescued with InsR were transplanted into lethally irradiated CD45.1-recipient mice (Fig. 3 G). After successfully reconstitution of the recipient BM, mice were injected with insulin to elevate the circulating insulin level. Only mice reconstituted with InsR could rescue Insr^−/− LSKs for differentiation with insulin stimulation (Fig. 3 H), which was similar to WT LSKs. These data suggest that insulin–InsR signaling affects the fate decision of LSKs. We generated diabetic mice that impaired insulin production (Yan et al., 2015; Fig. 3, I and J). More importantly, these mice had more GMPs but fewer CLPs than those of control mice (Fig. 3 K). Finally, proliferation and apoptosis states of GMPs and CLPs displayed no difference in diabetic mice and WT control mice (Fig. 3, L and M). Notably, IGF1R, the other insulin-binding receptor, was almost undetectable in MPPs (unpublished data). Altogether, the insulin–InsR signaling governs the differentiation of LSKs into lymphoid cell lineages.

It has been reported that HSCs are heterogeneous with myeloid-biased progenitors in CD150^highCD34⁻LSKs (Morita et al., 2010). To test whether the insulin signaling is involved in the specification of these cells, we treated CD150^highCD34⁻LSKs and CD150^negCD34⁻LSKs with insulin stimulation, and then transplanted these cells together with 3 × 10⁵ CD45.1 BM cells into lethally irradiated mice. Blood constitutions were examined 16 wk later. We observed that insulin stimulation did not alter the myeloid or lymphoid differentiation potential of CD150^highCD34⁻LSKs or CD150^negCD34⁻LSKs (Fig. 4, A and B), suggesting that insulin signaling did not affect the differentiation of myeloid-biased progenitors of CD150^highCD34⁻LSKs. However, when we transplanted Insr^+/+ or Insr^−/− CD150^highCD34⁻LSKs and CD150^negCD34⁻LSKs into recipient mice, we noticed that both Insr^−/−CD150^highCD34⁻LSKs and Insr^−/−CD150^negCD34⁻LSKs promoted myeloid commitment (Fig. 4, C and D), but declined lymphoid differentiation potential. In addition, the overall donor chimerism between these cell populations was similar (Fig. 4, E–H). Moreover, numbers of CD150^highCD34⁻LSKs in Insr^+/+ and Insr^−/− mice were comparable (Fig. 4 I). Additionally, numbers of CD150^high, CD150^low, and CD150^neg CD34⁻Flk2⁻LSKs in Insr^+/+ and Insr^−/− mice were also comparable (Fig. 4 J). In sum, these data suggest that the insulin signaling is involved in the skewing myeloid/lymphoid differentiation of MPPs, but not in the myeloid-biased HSCs.

Lymphopoiesis requires lineage-specific signature genes, such as Ikaros, IL-7R, Lck, and Ly-6D (Akashi et al., 2003; Nodland et al., 2011; Brownlie and Zamoyska, 2013), to direct restricted lineage differentiation. We checked whether insulin initiates expression of these lymphoid signature genes for lymphoid differentiation. Through analyzing enrichment/depletion of lineage signatures (Akashi et al., 2003), we found that lymphoid-related genes tended to be down-regulated, whereas myeloid-related genes were up-regulated in Insr^−/− MPPs (Fig. 5, A and B), but not in Insr^−/− LT-HSCs or Insr^−/− ST-HSCs compared with Insr^+/+ counterpart populations. Ikaros has been defined as critical for lymphoid differentiation (Iwasaki and Akashi, 2007). Therefore, we focused on the modulation of Ikaros in lymphopoiesis for insulin–InsR signaling. We first examined expression of Ikaros in the BM. Ikaros exhibited the highest expression level in CLPs and a modest level in MPPs (Fig. 5 C), which was consistent with previous reports (Yoshida et al., 2006; Iwasaki and Akashi, 2007). Interestingly, Insr knockout only affected the expression level of Ikaros in MPPs (Fig. 5 C), but remained unchanged in other hematopoietic progenitors.

We then determined Ikaros expression in MPPs under insulin stimulation. We found that insulin promoted the expression of Ikaros in MPPs (Fig. 5 C), but not in LT-HSCs or ST-HSCs. Additionally, Insr deficiency could not activate Ikaros expression in MPPs (Fig. 5 C). In parallel, starved mice with a low level of insulin showed reduced expression of Ikaros, and refeeding mice exhibited elevated expression of Ikaros (Fig. 5 D). Moreover, Ikaros was remarkably decreased in MPPs in the starved mice (Fig. 5 E), which was in agreement with the lower insulin level (not depicted). Finally, we found that insulin incubation could promote Ikaros expression in all MPP cells we examined, but not in Flk2⁻LSKs (Fig. 5 F), further validating a restrictive effect of insulin signaling on the Ikaros expression in MPPs.

To further explore the role of Ikaros in the fate decision of MPPs, we reconstituted lethally irradiated mice with either Ikaros-silenced or -overexpressed MPPs. Expectedly, Ikaros knockdown drove MPPs to differentiate into myeloid cells (Fig. 5, G and H), whereas Ikaros overexpression in MPPs led to more lymphoid cells (Fig. 5, I and J), suggesting that Ikaros activation was indispensable for the lymphoid differentiation of MPPs. To further determine the relationship between InsR and Ikaros, we introduced Ikaros to Insr^−/− MPPs for rescue experiments (Fig. 5 K). Importantly, Insr^−/− MPPs with Ikaros restoration generated fewer myeloid colonies than empty vector-treated Insr^−/− MPPs (Fig. 5 L). However, Insr^−/− MPPs with Ikaros restoration produced many more lymphoid cells (Fig. 5 M). Additionally, Ikaros restoration was able to rescue colony numbers of Insr^−/− MPPs to similar levels with those of Insr^+/+ MPPs either in a CFC assay or in an OP9 co-culture system (Fig. 5, L and M). These data suggest that Ikaros acts downstream of insulin–InsR signaling to control the fate decision of MPPs.

The mechanistic target of rapamycin (mTOR) is a serine/threonine protein kinase that acts as a master regulator of cellular growth and metabolism in response to nutrient and hormonal regulation (Laplante and Sabatini, 2012; Johnson et al., 2013). mTOR was a key regulator downstream of the insulin–InsR signaling pathway (Chi, 2012; Inoki et al., 2012; Powell et al., 2012). To examine whether mTOR is involved in regulation of the expression of Ikaros, we overexpressed mTOR in MPPs and checked the expression level of Ikaros. Interestingly, we found that mTOR overexpression in MPPs dramatically increased the expression level of Ikaros upon insulin stimulation (Fig. 6, A and B). However, without insulin stimulation, mTOR overexpression did not enhance Ikaros expression in MPPs. Moreover, we knocked down mTOR expression in MPPs and checked Ikaros expression (Fig. 6 C). We observed that mTOR knockdown remarkably reduced Ikaros expression in MPPs, which was comparable to that of shCtrl MPPs, even with insulin stimulation (Fig. 6 D). These data indicate that insulin–InsR signaling-mediated mTOR activation participates in the transcriptional activation of Ikaros in MPPs.

mTOR functions in two different complexes, mTOR complex 1 (mTORC1) and mTOR complex 2 (mTORC2; Lamming et al., 2012; Bar-Peled et al., 2013). Rapamycin inhibits mTORC1 from suppressing mTOR activity (Yip et al., 2010). Importantly, rapamycin addition could inhibit Ikaros expression in MPPs with insulin stimulation (Fig. 6 E), suggesting that mTORC1 is involved in the expression of Ikaros in MPPs. Moreover, through RT-PCR analysis, mTOR dramatically promoted Ikaros expression (Fig. 6 F), whereas the kinase-dead mTOR mutant (mTOR-KD) had no such activity. These observations verify that mTOR is indeed involved in the transcriptional activation of Ikaros in MPPs. We next determined the effect of rapamycin on MPP differentiation using an in vitro differentiation assay. As expected, MPPs pretreated with rapamycin produced more myeloid cells than vehicle-treated cells (Fig. 6 G); however, rapamycin pretreatment significantly decreased the numbers of lymphoid cells (Fig. 6 H). A reversible inhibitor of PI3Ks, LY294002, acts as an upstream inhibitor of mTOR (Wang et al., 2013). LY294002 could also drive MPPs to myeloid cell differentiation (Fig. 6, I and J). We transplanted shCtrl or mTOR-silenced MPPs into normal or diabetic mice to determine whether mTOR acted downstream of insulin signaling. Intriguingly, we observed that mTOR-silenced MPPs were prone to differentiate into myeloid cells in normal mice (Fig. 6 K). In contrast, this trend did not appear in insulin-suppressed diabetic mice, whose mice had more myeloid cells and fewer lymphoid cells compared with normal mice. In sum, these results suggest that mTOR enhances Ikaros expression that acts downstream of the insulin–InsR signaling in lymphopoiesis.

To identify the transcription factors that are responsible for Ikaros transcription downstream of insulin–InsR–mTOR signaling in MPPs, we silenced several transcription factors that were known to be regulated by mTOR (Chi, 2012; Powell et al., 2012). Surprisingly, among several factors we checked, only Stat3 knockdown abolished the enhanced expression of Ikaros after insulin stimulation (Fig. 7 A), which was comparable to that of shCtrl-treated cells. Importantly, Stat3 associated with the promoter of Ikaros in MPPs upon insulin stimulation through chromatin immunoprecipitation (ChIP) assays (Fig. 7 B). Moreover, Stat3 could vigorously activate Ikaros expression with insulin administration (Fig. 7, C and D). To further determine the physiological role of Stat3 in Ikaros expression, we generated Stat3 conditional knockout mice in the hematopoietic system. Stat3 was completely deleted in the BM (Fig. 7 E). More importantly, Stat3 deletion impaired the expression of Ikaros upon insulin stimulation (Fig. 7 F). These data suggest that Stat3 binds to the promoter of Ikaros to initiate its expression with insulin stimulation.

Stat3 is activated by a variety of cytokines and growth factors (Casanova et al., 2012). Stat3 is phosphorylated at its two key residues, Tyr705 and Ser727. Ser727 phosphorylation is involved in the modulation of Stat3 activity, whereas Tyr705 phosphorylation participates in its dimerization and activation (Wen et al., 1995; Calò et al., 2003). To determine whether mTOR activates Stat3 upon insulin stimulation in MPPs, we performed immunoblotting with MPP cell lysates obtained from different treatments. Interestingly, we observed that Stat3 was phosphorylated at serine 727 (S727; Fig. 7 G), instead of the canonical site tyrosine 705 (Y705), with insulin stimulation in MPPs. More importantly, rapamycin completely blocked S727 phosphorylation (Fig. 7 G). These data suggest that mTOR acts as an upstream factor to activate Stat3 by its phosphorylation. Furthermore, the serine 727 to alanine mutation (S727A) of Stat3 failed to augment the activation of Ikaros (Fig. 7, H and I), whereas the Y705F Stat3 mutant still augmented Ikaros activation. Additionally, rapamycin addition completely abolished the activation of Ikaros, even when cotransfected with WT Stat3 (Fig. 7 I). Importantly, we observed that Stat3 translocated from the cytoplasm to the nucleus upon insulin stimulation (Fig. 7 J), whereas rapamycin impaired this activity. These results suggest that Stat3 is activated by mTOR with insulin signaling to initiate the transcription of Ikaros.

We found that Stat3 deficiency reduced Ikaros expression upon insulin stimulation (Fig. 7 F). We next detected the differentiation potential of Stat3^−/− MPPs. Stat3^−/− MPPs produced many more myeloid cells compared with those of WT Stat3^+/+ cells (Fig. 7 K). In contrast, Stat3^−/− MPPs generated fewer lymphoid cells (Fig. 7 L). We then transplanted Stat3^−/− or Stat3^+/+ MPPs into normal or diabetic mice to determine whether Stat3 is the transcription factor downstream of the insulin–InsR signaling. Importantly, we found that Stat3^−/− MPPs generated more myeloid cells in normal mice (Fig. 7 M), whereas fewer lymphoid cells. However, in diabetic mice, Stat3^−/− MPPs exhibited no such difference. Together, these results indicate that Stat3 acts downstream of insulin–InsR–mTOR signaling to activate the expression of Ikaros in MPPs, leading to lymphoid lineage specification.

The lineage commitment of HSCs results in distinct myeloid and lymphoid differentiation pathways, leading to the generation of common myeloid progenitors and CLPs (Iwasaki and Akashi, 2007; Rossi et al., 2012). However, the underlying mechanisms that modulate the lineage specification properties of early progenitors have not been defined yet. In this study, we show that InsR is constitutively expressed in multipotent hematopoietic progenitors. Deletion of InsR in murine BM causes biased differentiation of MPPs toward myeloid cells. Insulin–InsR signaling initiates the expression of Ikaros, which is required for lymphoid differentiation. Without insulin signaling, Ikaros expression is suppressed and MPPs are prone to differentiate into myeloid cells. mTOR acts as a downstream effector that modulates MPP differentiation specification. mTOR activates Stat3 by phosphorylation at serine 727 under insulin stimulation, which translocates to the nucleus of MPPs to bind to the promoter of Ikaros, leading to its transcription priming. Our findings demonstrate that insulin–InsR signaling plays a critical role in governing MPPs toward lymphoid lineages in early lymphopoiesis.

HSCs have an open chromatin structure that allows the transcription of many genes, most of which are myeloid-related (Akashi et al., 2003; Laurenti et al., 2013). Lymphoid-related genes come later during the lineage specification process (Akashi et al., 2003). It is therefore suggested that HSCs are prone to differentiate into myeloid cells and extrinsic signals are required for priming lymphoid lineage commitment. HSCs differentiate to lymphoid-primed MPPs, then to lymphoid-restricted progenitors, accompanied by the loss of erythroid-megakaryocyte and myeloid potential (Adolfsson et al., 2005). More recently, many studies reported the lineage-instructive capacities of several cytokines on hematopoietic stem/progenitor cells. For example, M-CSF directly induces expression of the myeloid master regulator PU.1 to direct myeloid cell-fate specification in mouse HSCs (Mossadegh-Keller et al., 2013). Erythropoietin (Epo) triggers erythroid lineage skewing at all lineage bifurcations between HSCs and erythroid progenitors (Grover et al., 2014). Here, we show that InsR is constitutively expressed in the multipotent hematopoietic progenitors of BM and that extrinsic insulin signaling is required for lymphoid lineage specification in early lymphopoiesis. Notably, we obtained quite high counts of peripheral white blood cells in Insr^+/+ and Insr^−/− mice, which might be caused by a sensitive blood cell counter we used. In reality, Insr-deficient mice displayed much higher peripheral white blood cell counts than WT mice. Our study also demonstrates that insulin-suppressed diabetic mice have more myeloid cells and fewer lymphoid cells compared with normal WT mice. A recent study showed that high fat diet–induced obesity produces skewed myeloid progenitors that potentiate generation of macrophages (Singer et al., 2014), augmenting inflammatory responses in metabolic tissues. Interestingly, prolonged fasting can reduce circulating IGF-1 levels in LT-HSCs and niche cells, which promote self-renewal, and lineage-balanced regeneration of HSCs (Cheng et al., 2014). Of note, Ding and Morrison (2013) reported that lymphoid/myeloid-skewed HSCs occupy distinct BM niches (Oguro et al., 2013), suggesting that a specialized microenvironment in which progenitors reside also plays a critical role in hematopoiesis.

The membrane InsR binds to insulin to turn on the PI3K–Akt pathway (Fruman and Rommel, 2014). Akt can phosphorylate to activate mTOR, which is present in two distinct protein complexes: mTORC1 and mTORC2 (Shimobayashi and Hall, 2014). mTORC1 phosphorylates its main effector substrates, eukaryotic translation initiation factor 4E (eIF-4E)–binding protein 1 (4E-BP1) and 70-kD ribosomal S6 protein kinase (S6K), to potentiate protein translation and synthesis. Besides its role in metabolism, mTORC1 is also involved in mitochondrial biosynthesis and autophagy (Cunningham et al., 2007; Yu et al., 2010). Rapamycin binds to FKBP12 by binding to the FRB site on mTOR, which blocks the ability of Raptor to bind to mTOR to inhibit the activity of mTORC1 (Powell et al., 2012). Prolonged treatment with rapamycin for some tissues and cells are able to suppress the mTORC2 activity as well (Sarbassov et al., 2006). In this study, we found that mTOR phosphorylates Stat3 at serine 727, a novel substrate for mTOR, which is indispensable for the transcriptional activation of Ikaros in MPPs. Both mTOR knockdown and rapamycin treatment are able to suppress the expression of Ikaros in MPPs. How mTORC1 and mTORC2 function in early lymphopoiesis remains to be further investigated.

Ikaros plays an essential role in the modulation of lymphoid development and differentiation (Iwasaki and Akashi, 2007). Neonatal Ikaros-null mice appear to be a complete defect in fetal thymocyte development. Adult Ikaros-deficient mice block lymphoid differentiation and exhibit thymocyte development skewed to CD4 T cells (Urban and Winandy, 2004). Interestingly, the numbers of myeloid cells are increased in Ikaros deficient mice. In mature CD4⁺T cells, Ikaros modulates a variety of processes including Th2 differentiation and cytokine production (Bandyopadhyay et al., 2007; Quirion et al., 2009). In B cells, the transcription factor FoxO1 plays a critical role in Ikaros expression, whereas FoxO1 does not directly initiate Ikaros transcription. However, the transcriptional activation of Ikaros is poorly understood in hematopoietic lineage specification. Here, we show that InsR is highly expressed in hematopoietic progenitors, with the highest level in MPPs. Extrinsic insulin–InsR signaling directly activates Stat3 in MPPs to initiate Ikaros transcription.

Stat3 is well known to maintain the pluripotency of embryonic stem cells under the control of LIF signaling (Yu et al., 2014). Stat3 also participates in sustaining HSC self-renewal (Chung et al., 2006). Mice with Stat3 conditional deletion in the hematopoietic system display increased numbers of myeloid cells and decreased numbers of lymphoid cells, a phenomenon similar to ageing mice (Mantel et al., 2012). Reactive oxygen species (ROS) levels are elevated in HSCs and hematopoietic progenitor cells. Lymphopoietic activity becomes compromised during ageing (Montecino-Rodriguez and Dorshkind, 2006). The earliest lymphoid progenitor pools are declined in aged BM (Miller and Allman, 2005). Senescence may decrease the sensing ability to insulin signaling due to insulin resistance. Disrupted insulin secretion in mouse diabetes models generates more myeloid cells but fewer lymphoid cells, resulting in a skewed myeloid/lymphoid ratio in diabetic mice. It remains to be delineated whether insulin resistance decreases lymphoid lineage differentiation in diabetic patients, which may result in damage of the immune system and increased susceptibility to infections. We expect that our findings will provide useful clues for translational study for diabetic patients. In summary, our observations reveal that insulin–InsR signaling drives MPP differentiation into lymphoid lineages in early lymphopoiesis, which is essential for maintaining a balanced immune system for an individual organism.

The following commercial antibodies were used: mouse hematopoietic lineage eFlour 450 cocktail, PerCP-Cy5.5-anti-CD45.1, FITC-anti-CD45.2, Alexa Fluor 700-anti-IL-7Rα, FITC-anti-Ly6A/E (Sca-1), PE-anti-CD117 (c-Kit), APC-eFluor780-anti-CD48, anti-CD3, anti-CD19, FITC-anti-CD11b, PE-anti-Gr1, FITC-anti-B220, Pecy5-anti-CD3ε, APC-anti-Flt3/CD135, anti-F4/80, anti-Gr-1, and anti-CD34 were purchased from eBioscience. PE-cy7-anti-CD150 was obtained from BioLegend. PerCP-Cy5.5–conjugated goat anti–rat IgG and APC-Cy7–conjugated goat anti–rabbit IgG were purchased from Santa Cruz Biotechnology, Inc. Anti–β-actin was from Sigma-Aldrich. Antibodies against Insr-β, phosphorylated Insr-β (Tyr1150/1151), insulin, mTOR, Stat3, S727 phosphorylated Stat3, Y705 phosphorylated Stat3, S6K, and phosphorylated S6K were purchased from Cell Signaling Technology. Donkey anti–rabbit or anti–mouse secondary antibodies conjugated with Alexa Fluor 488, 594, or 405 were purchased from Molecular Probes. HRP-conjugated secondary antibody was obtained from Santa Cruz Biotechnology, Inc. Propidium iodide (PI), Annexin-V, insulin, rapamycin, and streptozotocin (STZ) were purchased from Sigma-Aldrich.

For HSC or MPP culture, cells were cultured in StemPro-34 medium (Invitrogen), containing 4 mM l-glutamine, 100 µg/ml streptomycin, 100 U/ml penicillin, and the following cytokines (all from PeproTech): 10 ng/ml IL-3, 25 ng/ml SCF, 25 ng/ml Ftl-3L, 10 ng/ml GM-CSF, 25 ng/ml IL-11, 4 U/ml Epo, and 25 ng/ml Tpo.

Insr^flox/flox mice were purchased from The Jackson Laboratory (B6.129S4(FVB)-Insrtm1Khn/J). Insr^flox/flox mice were crossed with MxCre⁺ mice to get MxCre⁺;Insr^flox/+ mice. The MxCre⁺;Insr^flox/+ mice were then crossed with Insr^flox/flox mice to generate MxCre⁺;Insr^flox/flox mice. MxCre⁺;Insr^flox/flox mice were intraperitoneally injected with 300 µg polyinosine-polycyticylic acid (poly(I:C)) every other day for a total of three times. The recombinant efficacy was examined by Western blotting 5 d after the last poly(I:C) administration using BM cells. Stat3^flox/flox mice (Takeda et al., 1999) were crossed with Vav-Cre mice to generate mice with conditional deletion of Stat3 in the hematopoietic system. Mouse experiments complied with ethical regulations and were approved by the Institutional Animal Care and Use Committees at the Institute of Biophysics, Chinese Academy of Sciences.

Spleens were fixed in 4% paraformaldehyde (PFA; Sigma-Aldrich) for 12 h. Femurs were fixed in buffer containing 10% formaldehyde for 12 h, and then decalcified in decalcifying buffer (10% EDTA in PBS [wt/wt], pH 7.4) for another 12 h. Fixed tissues were washed twice using 75% ethanol and embedded in paraffin, followed by sectioning and staining with hematoxylin and eosin according to standard laboratory procedures.

Flow cytometry was performed as previously described (Xia et al., 2015b). Generally, mice were euthanized and BM cells were flushed out from femur in PBS (phosphate buffered saline) buffer. Spleen cells were generated by mashing the spleen in PBS buffer. Cells were sifted through 50-µm cell strainers after removing red blood cells by suspending cells in ammonium-based red cell lysis buffer. For flow cytometry analysis, cells were stained with fluorophore-conjugated antibodies, followed by detecting or sorting on an Influx cell sorter (BD). Data were analyzed using FlowJo 7.6.1 software (Tree Star).

RNA interference was performed as previously described (Xia et al., 2015a). In brief, RNA interference sequences used in this research were designed according to pSUPER system instructions (Oligoengine). H1 promoter and targeting sequences were further cloned to lentivirus vector pSIN-EF2. Lentiviruses were generated by transfecting pSIN-EF2-shRNA and packaging vectors to HEK293T cells, followed by concentration using ultracentrifugation at 50,000 g. EML or sorted cells were infected with lentiviruses for 36 h before examination or transplantation. shRNA sequences were: Ikaros, #1, 5′-CAGTGACACTCCAGATGAA-3′; #2, 5′-GGAAGAATGTGCAGAGGAT-3′; #3: 5′-GAGGCATTCGACTTCCTAA-3′; mTOR, #1, 5′-TGCCAACTACCTTCGAAAC-3′; #2, 5′-AGGAAATGCAGAAGCCTCA-3′; #3, 5′-CCGGCACACATTTGAAGAA-3′; Stat3, #1, 5′-GAGTCAAGACTGGGCATAT-3′; #2, 5′-CCAGCAATATAGCCGATTC-3′; #3, 5′-CCTCTATCAGCACAACCTT-3′. Total RNA was extracted from sorted cells using TRIzol reagent, and cDNA was reverse-transcribed using Superscript II (Invitrogen). RT-PCR was performed using the following primers: Ikaros, forward, 5′-TCGGGAGAGAAAATGAATGG-3′, reverse, 5′-AGGCCGTTCACCAGTATGAC-3′; Insr, forward, 5′-AAAGTTTGCCCAACCATCTG-3′, reverse, 5′-GTGAAGGTCTTGGCAGAAGC-3′. Expression was normalized to that of housekeeping gene β-actin. Single-cell RT-PCR was performed as previously described (Luc et al., 2008).

Transient knockdown was performed as described previously (Wang et al., 2013). In brief, H1 promoter together with shRNA sequences were first merged with 2 × TetO2 by overlapping PCR and then cloned to pcDNA4/TO/Myc-His B replacing the CMV promoter. For overexpression experiments, cDNA sequences were cloned to the multiple cloning site of pcDNA4/TO/Myc-His B. pcDNA4/TO/Myc-His B was cotransfected with pcDNA6/TR into MPPs by electroporation, followed by addition of doxycycline (1 µg/ml) for 36 h (for knockdown) or 18 h (for overexpression).

Cells were fixed in 1% formaldehyde for 20 min and then quenched with 0.125 M lysine, followed by swelling in lysis buffer (50 mM Hepes, pH 7.5, 140 mM NaCl, 1% Triton X-100, 0.1% NaDeoxycholate, and protease inhibitors) for 30 min on ice. Chromatin was sheared to an average length of 400 bp by sonication. After being de-cross-linked by RNase, proteinase K, and heat, input genomic DNA was precipitated with ethanol and quantified in a GeneQuant 100 spectrophotometer (GE Healthcare). Chromatin was precleared with protein A/G-agarose (Santa Cruz Biotechnology, Inc.), followed by incubation with the indicated antibodies at 4°C overnight and further incubation with protein A/G-agarose for 2 h. Beads were washed with washing buffer (10 mM Tris, pH 8.0, 250 mM LiCl, 0.5% NP-40, 0.5% NaDeoxycholate, 1 mM EDTA) three times and eluted with elution buffer (50 mM Tris, pH 8.0, 1% SDS, 10 mM EDTA). Eluates were de-cross-linked by RNase, proteinase K, and heat, and DNA was extracted with phenolchloroform, followed by ethanol precipitation. For each ChIP experiment, 2 × 10⁴ cells were used. 5% of nuclear extracts served as inputs. Immunoprecipitated DNAs were further analyzed by real-time PCR. Signals were normalized to input DNA.

BM transplantation was performed as described previously (Xia et al., 2014). In brief, donor BM cells were separated as described above and injected alone or with recipient BM cells into lethally irradiated (10 Gy) CD45.1 mice. Reconstituted mice were fed with water containing 1 g/liter ampicillin for 2 wk before switching to regular water. For analysis of peripheral blood cells, blood was obtained from mouse tail vein and stained as described above.

2 × 10⁴ BM cells were seeded in 35-mm dishes containing IMDM, supplemented with 1.2% methylcellulose (STEMCELL Technologies), 40 µM 2-mercaptoethanol, 0.5 mM haemin, 30% FBS, 2 mM l-glutamine, 20 ng/ml rmIL-3, 6 U/ml recombinant human EPO, and 20 ng/ml recombinant mouse SCF. Colonies of CFU-GEMM, CFU-GM, and BFU-E were calculated 10 d after incubation at 37°C. LT-CIC assay was performed as previously described (Méndez-Ferrer et al., 2008).

OP9 cells were cultured in α-MEM supplemented with 20% FBS, 40 µM 2-mercaptoethanol, and 2 mM l-glutamine. For lymphoid differentiation, sorted LSKs or MPPs were plated onto 80% confluent OP9 cells in α-MEM supplemented with 5 ng/ml rmFlt3L and 5 ng/ml rmIL-7 (PeproTech). Cells were stained with anti-CD19 antibody, followed by flow cytometry analysis. Samples were run out to determine the numbers of lymphoid cells (positive for CD19) by flow cytometry.

For circadian rhythm, mice were fed regularly during the night (8:00 p.m. to 8:00 a.m. on the following day) and with water only in the day (8:00 a.m. to 8:00 p.m.). For the starvation/refeeding cycle, mice were supplied with water only from 8:00 a.m. for 48 h, followed by refeeding normally for another 48 h. For generation of diabetic mice, mice were injected with streptozotocin (80 mg/kg i.p. for four consecutive days; vehicle, 0.1 M sodium citrate, pH 4.5) to induce diabetes.

Student’s t test was used as statistical analysis by using Excel (Microsoft; Xia et al., 2013).

Fig. S1 shows the gating strategy used in this study.

We thank Xuan Yang for protein expression and Yan Teng for technical support.

This work was supported by the National Natural Science Foundation of China (81330047, 31471386, 91419308, and 31300645), 973 Program of the MOST of China (2010CB911902), the Strategic Priority Research Programs of the Chinese Academy of Sciences (XDA01010407), and the China Postdoctoral Science Foundation (2015M571141).

The authors declare no competing financial interests.

Author contributions: P. Xia designed and performed experiments, analyzed data, and wrote the paper; S. Wang performed experiments and analyzed data. G. Huang constructed plasmids; Z. Fan initiated the study, and organized, designed, and wrote the paper.