Natural killer (NK) and invariant NK T (iNKT) cells are critical in host defense against pathogens and for the initiation of adaptive immune responses. miRNAs play important roles in NK and iNKT cell development, maturation, and function, but the roles of specific miRNAs are unclear. We show that modulation of miR-150 expression levels has a differential effect on NK and iNKT cell development. Mice with a targeted deletion of miR-150 have an impaired, cell lineage–intrinsic defect in their ability to generate mature NK cells. Conversely, a gain-of-function miR-150 transgene promotes the development of NK cells, which display a more mature phenotype and are more responsive to activation. In contrast, overexpression of miR-150 results in a substantial reduction of iNKT cells in the thymus and in the peripheral lymphoid organs. The transcription factor c-Myb has been shown to be a direct target of miR-150. Our finding of increased NK cell and decreased iNKT cell frequencies in Myb heterozygous bone marrow chimeras suggests that miR-150 differentially controls the development of NK and iNKT cell lineages by targeting c-Myb.
The innate immune system provides early host defense against viruses, bacteria, and tumor cells. Two of its cell lineages, NK and invariant NK T (iNKT) cells, do not require prior sensitization for effector function and play critical roles in both initially fighting infection and subsequently activating the adaptive immune response.
NK cells develop in the BM and share a common lymphoid progenitor with T and B cells (Kondo et al., 1997). The committed NK cell precursors (NKPs) express the shared IL-2 and IL-15 receptor common β subunit (CD122; Rosmaraki et al., 2001). NK cell development is critically dependent on IL-15, as mice lacking IL-15 or any component of its receptor or signaling have a block in NK cell differentiation in the BM (DiSanto et al., 1995; Suzuki et al., 1997; Kennedy et al., 2000; Gilmour et al., 2001; Vosshenrich et al., 2005). As NK cells mature, they sequentially acquire the cell surface receptors NK1.1 (in C57BL/6 mice), CD94-NKG2A, Ly49 receptors, and the α2-integrin CD49b (DX5; Hesslein and Lanier, 2011). Terminal maturation is accompanied by loss of CD27 expression and up-regulation of CD11b expression, together with optimal cytolytic function and IFN-γ production. NK cell specificity is fine-tuned by stochastic expression of activating and inhibitory receptors of the Ly49 family. Some activating Ly49 members can detect specific virus-encoded products (e.g., Ly49H recognizes the mouse cytomegalovirus m157 glycoprotein; Arase et al., 2002; Smith et al., 2002). Individual inhibitory Ly49 members recognize distinct MHC class I allotypes (e.g., Ly49C recognizes H2b in C57BL/6 mice) and are important in self-recognition, whereas other inhibitory Ly49 genes produce receptors that do not recognize the autologous MHC class I (e.g., Ly49A in C57BL/6 mice; Orr and Lanier, 2011).
NKT cells share properties with both NK cells and T cells in that they express receptors of the NK cell lineage, including NK1.1 and members of the Ly49 family, along with a TCR. The majority of NKT cells express an invariant TCRα (Vα14Jα18 in mice) that pairs with a limited repertoire of TCRβ chains (Vβ8, Vβ7, or Vβ2 in mice; Godfrey et al., 2010). These NKT cells, also referred to as iNKT cells, can be identified using CD1d tetramers loaded with the glycosphingolipid antigen α-galactosylceramide (Matsuda et al., 2000). iNKT cells develop from CD4+CD8+ double-positive Vα14Jα18 TCR+ thymic precursors (Godfrey et al., 2010). After positive selection, iNKT-cell precursors down-regulate their expression of CD24 (HSA), and transition through several maturation stages accompanied by the expression of NK1.1, CD44, and CD122. After CD122 up-regulation, iNKT cells proliferate in response to IL-15 and migrate from the thymus to the periphery, where they are most abundant in the spleen, liver, and BM.
Although much has been learned about the transcriptional regulation of NK and iNKT lineage development (Godfrey et al., 2010; Hesslein and Lanier, 2011), less is known about the posttranscriptional mechanisms that regulate NK and iNKT cells. microRNAs (miRNAs) are noncoding RNAs, expressed from endogenous genes, which act on protein-encoding mRNAs, targeting them for translational repression or degradation (Bartel, 2004). Many miRNAs are expressed in a stage- and cell-specific fashion in the hematopoietic system (Kuchen et al., 2010), and emerging evidence suggests that they regulate lymphocyte differentiation and function (Muljo et al., 2005; Cobb et al., 2006; Koralov et al., 2008; X. Zhou et al., 2008; Liston et al., 2008; Zhou et al., 2009; Fedeli et al., 2009). Deletion of the miRNA-processing enzymes Dicer or Dgcr8 leads to defects in NK cell activation, survival, and function during mouse cytomegalovirus infection (Bezman et al., 2010). In addition, loss of miRNAs in the BM or thymus leads to impaired iNKT cell development and function (Fedeli et al., 2009; Zhou et al., 2009; Seo et al., 2010). These results, and the description of a subset of miRNAs differentially expressed in NK and iNKT versus other hematopoietic cells (Fedeli et al., 2009; Bezman et al., 2010; Kuchen et al., 2010), raise a question as to how individual miRNAs present in high amounts in these cell types contribute to specific aspects of their biology.
miR-150 is expressed in both mature NK and iNKT cells (Fedeli et al., 2009; Bezman et al., 2010; Fehniger et al., 2010) and, as in B and T lymphocytes, miR-150 is up-regulated during cellular maturation. Constitutive expression of miR-150 blocks B cell development at the transition from the pro–B to pre–B cell developmental stage (Zhou et al., 2007). miR-150–deficient mice have an accumulation of B-1 B cells in the spleen and peritoneal cavity (Xiao et al., 2007). In nonlymphoid lineages, where it drives megakaryocyte differentiation at the expense of erythropoiesis, miR-150 regulates the cell fate of megakaryocyte–erythroid progenitors (Lu et al., 2008). The transcription factor c-Myb has been shown to be a direct and functional target of miR-150 (Xiao et al., 2007; Lu et al., 2008). In this study, we show that miR-150 differentially affects the development of NK and iNKT cell lineages. These results show a critical developmental role for miR-150 in the innate immune system and suggest that there are distinct molecular pathways during the maturation stages of NK and iNKT cells.
RESULTS
Mice lacking miR-150 have cell-intrinsic defects in the development and maturation of NK cells
To define the role for miR-150 in NK cells, we generated mixed BM chimeras by reconstituting lethally irradiated mice with BM from WT (CD45.1+) and miR-150−/− (CD45.2+) mice (Xiao et al., 2007). Analysis of the resulting chimeric animals (WT:miR-150−/−) allowed us to study the cell-intrinsic effects of miR-150 deficiency on NK cells. We confirmed the absence of miR-150 in CD45.2+ NK1.1+ TCRβ− NK cells (Fig. 1 A). Flow cytometric analyses of WT and miR-150−/− compartments showed a significant reduction in the percentage of miR-150−/− NK cells in the spleen (WT, 1.1 ± 0.1% vs. miR-150−/−, 0.8 ± 0.1%), liver (WT, 3.6 ± 0.5% vs. miR-150−/−, 2.5 ± 0.4%), and blood (WT, 1.1 ± 0.1% vs. miR-150−/−, 0.6 ± 0.1%; Fig. 1 B). Concurrently, the absolute numbers of miR-150−/− NK cells were also reduced compared with that of WT NK cells in the spleen of WT:miR-150−/− chimeras (WT, 0.30 × 106 ± 0.05 × 106 vs. miR-150−/−, 0.10 × 106 ± 0.01 × 106; Fig. 1 C). The frequency of miR-150−/− monocytes was similar to that of WT monocytes in peripheral blood (WT, 7.3 ± 0.2% vs. miR-150−/−, 7.4 ± 0.4%; P = 0.8; n = 10), and was even elevated in spleen of WT:miR-150−/− chimeras (WT, 2.9 ± 0.3% vs. miR-150−/−, 5.0 ± 0.6%; P = 0.008; n = 11). Furthermore, we observed no significant differences between WT and miR-150−/− donor cells in the monocyte compartment, whereas miR-150−/− NK cells were underrepresented (Fig. 1 D). Together, these results suggest that the reduced frequency of miR-150−/− NK cells is not caused by a competitive disadvantage of miR-150−/− hematopoietic stem cells to reconstitute the host.
The maturation status of peripheral NK cells can be classified based on their expression of CD27 and CD11b, following the pathway of CD27+ CD11b− → CD27+ CD11b+ → CD27− CD11b+ (Chiossone et al., 2009). A significant accumulation of more immature CD27+ CD11b− and a corresponding reduction of mature CD27− CD11b+ subsets was observed among miR-150−/− NK cells in chimeric mice (Fig. 1 E). Similarly, miR-150−/− NK cells showed increased expression of c-Kit, a receptor that is expressed on developing BM and immature splenic NK cells but is down-regulated upon maturation (Fig. 1 F; Di Santo, 2006; Chiossone et al., 2009).
To determine whether the reduction of miR-150−/− NK cells was the result of increased cell death, increased expansion of immature NK cells, or both, we stained freshly isolated cells with Annexin V and a viability dye to assess apoptosis. There was no significant increase in apoptotic miR-150−/− compared with WT NK cells (unpublished data). The rate of NK cell turnover at steady state (measured by BrdU incorporation during a 3-d labeling period) was also not different between WT and miR-150−/− splenic NK cells (unpublished data). However, when WT and miR-150−/− NK cells were co-transferred into a lymphopenic environment (Rag2−/− x Il2rg−/− mice) for 3 d, immature CD27+ miR-150−/− NK cells proliferated more than CD27+ WT NK cells, whereas mature CD27− miR-150−/− NK cells proliferated less than their WT counterparts (Fig. 1 G). Thus, the immature phenotype of peripheral miR-150−/− NK cells might be because miR-150−/− NK cells spend more time at the immature stage, and CD27− miR-150−/− NK cells have reduced proliferative capacity compared with WT NK cells.
To determine whether the function of NK cells is affected by miR-150 deficiency, freshly isolated splenic NK cells were stimulated either by cross-linking cell surface receptors, including NK1.1 and Ly49H, or by a combination of IL-12 and IL-18 cytokines. Whereas control NK cells produced high amounts of IFN-γ in response to stimulation, a significantly smaller proportion of miR-150−/− NK cells made IFN-γ (∼30% decrease in the frequency of miR-150−/− IFN-γ+ NK cells; Fig. 2 A). The amounts of IFN-γ produced by individual control and miR-150−/− NK cells were similar (unpublished data). The reduction in IFN-γ–producing cells might be caused by the reduction in the CD27− CD11b+ subset within the NK cell population, because this subset plays a dominant role in cytokine production (Chiossone et al., 2009). We co-stained NK cells with anti-CD11b antibody and measured intracellular IFN-γ levels. Both CD11b+ and CD11b− subsets of miR-150−/− NK cells still produced less IFN-γ compared with that in WT NK cells (Fig. 2 B). Collectively, these experiments provide strong evidence for a cell-lineage intrinsic role for miR-150 in the generation and maturation of NK cells.
miR-150 promotes NK cell development in the BM
Because miR-150 is expressed early in NK cell development, and is up-regulated during maturation (Fig. 3 A), we tested whether the absence of this miRNA influenced the development of NK cells in the BM. NK cells can be identified by the expression of CD122 and the absence of cell surface markers associated with mature hematopoietic lineages, such as CD4, CD8, CD3, CD11c, Ter119, Gr-1, and CD19 (Di Santo, 2006). A significant reduction of miR-150−/− NK cells was observed in the BM of WT:miR-150−/− chimeras (WT, 1.1 ± 0.1% vs. miR-150−/−, 0.5 ± 0.1%; P = 0.003; n = 9; Fig. 3 B). A scheme for the differentiation stages of developing BM NK cells has been refined based on cell surface markers (Di Santo, 2006). NKPs are defined as CD122+ NK1.1−, as well as being lineage marker negative, and are limited to cells differentiating into NK cells (Rosmaraki et al., 2001). Acquisition of NK1.1 marks the beginning of the immature NK cell (iNK) stage (Rosmaraki et al., 2001; Kim et al., 2002). Expression of CD51 is up-regulated at this stage, and then down-regulated upon further maturation into mature NK cells when CD49b is expressed. Whereas the frequency of NKP cell subsets among CD122+ miR-150−/− NK cells was not affected (Fig. 3 C), a significant accumulation of CD51+CD49b− iNK cells was observed among CD122+ NK1.1+ miR-150−/− NK cells (Fig. 3 D). Together with the corresponding increase in miR-150 expression at the iNK cell stage (Fig. 3 A), these results suggest that the reduction of miR-150−/− NK cells occurs during the transition from immature to mature NK cells rather than earlier.
Mature hyperfunctional NK cells accumulate in miR-150Tg mice
Having shown that WT:miR-150−/− chimeras display an impaired development of miR-150−/− NK cells, we hypothesized that increased miR-150 expression in BM could result in the accumulation of NK cells. To test this hypothesis, we used transgenic mice with moderate ectopic expression of miR-150 under the control of the CAG promoter in the context of Rosa26 locus (Xiao et al., 2007). Analogous to our studies with miR-150−/− mice, we performed BM reconstitution where lethally irradiated hosts were reconstituted with an equal mix of miR-150Tg and congenic WT BM cells. miR-150 expression level in sorted splenic miR-150Tg NK cells was 3.0 ± 0.7× higher than the endogenous level in splenic WT NK cells (Fig. 4 A).
We observed a significantly greater frequency of miR-150Tg NK cells in the spleen (WT, 0.5 ± 0.1% vs. miR-150Tg, 5.2 ± 1.1%), liver (WT, 2.6 ± 0.3% vs. miR-150Tg, 21.3 ± 3.6%), and blood (WT, 0.7 ± 0.1% vs. miR-150Tg, 2.7 ± 0.4%) of WT:miR-150Tg chimeric mice (Fig. 4 B). The absolute numbers of splenic miR-150Tg NK cells were also increased compared with that of WT NK cells (WT, 0.10× 106 ± 0.04 × 106 vs. miR-150Tg, 0.30 × 106 ± 0.03 × 106; P = 0.003; n = 6; Fig. 4 C). Among miR-150Tg NK cells, there was a significant reduction in the CD27+ CD11b− subset in the spleen of WT:miR-150Tg chimeras (Fig. 4 C). This decrease in CD27+CD11b− miR-150Tg NK cells was correlated with a concomitant mild increase in CD27+ CD11b+ and CD27− CD11b+ NK populations. Next, we tested the ability of miR-150Tg NK cells to produce IFN-γ upon activation with various stimuli. NK cells isolated from WT:miR-150Tg chimeras were stimulated by immobilized antibodies to several activating receptors or by a combination of IL-12 and IL-18. After either type of stimulation, a larger fraction of miR-150Tg NK cells produced IFN-γ (∼40% increase in miR-150Tg as compared with WT NK cells; Fig. 4 D). Controlling for an increased number of NK cells expressing CD11b, a greater proportion of miR-150Tg NK cells generated IFN-γ compared with WT NK cells (Fig. 4 E). The amounts of IFN-γ produced by individual miR-150Tg NK cells, however, were similar to those made by control NK cells (unpublished data).
Because we observed the accumulation of mature miR-150Tg NK cells in WT:miR-150Tg chimeras, we tested whether this could be caused by faster expansion and/or lower susceptibility to apoptosis of peripheral miR-150Tg NK cells or alterations in the development of BM miR-150Tg NK cells. We did not find a significant difference in either basal turnover rate or Annexin V staining of splenic miR-150Tg NK cells in comparison to WT NK cells (unpublished data). However, when adoptively transferred into Rag2−/− x Il2rg−/− mice, mature CD27− miR-150Tg NK cells underwent more rounds of proliferation compared with CD27− subset of WT NK cells (Fig. 5 A). Furthermore, there was a significant increase in the percentage of lineage− CD122+ miR-150Tg NK cells in the BM (WT, 1.0 ± 0.2% vs. miR-150Tg, 7.8 ± 2.7%; P = 0.04; n = 9) of WT:miR-150Tg chimeras (Fig. 5 B). Among miR-150Tg NK cells in the BM, there was also a significant reduction in iNK cell fraction (Fig. 5 C). Thus, a larger proportion of mature miR-150Tg NK cells in the periphery of WT:miR-150Tg chimeras is likely caused by both increased maturation of miR-150Tg NK cells in the BM and expansion of mature miR-150Tg NK cells in the spleen.
In contrast to the NK cell compartment, the frequency of mature T cells was decreased in the peripheral lymphoid organs in WT:miR-150Tg chimeras (unpublished data). Analysis of thymocytes by flow cytometry revealed a partial block in T cell development, at the double negative-3 (DN3) to the double negative-4 (DN4) transition. These data are in agreement with the phenotype observed in miR-150Tg mice (Xiao et al., 2007).
Partial loss of c-Myb recapitulates the phenotype of miR-150Tg mice and leads to enhanced NK cell development
We explored the mechanism underlying miR-150–dependent NK cell development. Several studies in cell lines, as well as primary cells, have shown that a transcription factor c-Myb is a target of miR-150 (Xiao et al., 2007; Lin et al., 2008; Lu et al., 2008). We tested whether there is an inverse relationship between miR-150 and c-Myb expression in NK cells. CD27+CD11b−, CD27+CD11b+, and CD27−CD11b+ subsets of NK cells were purified from spleens of WT mice and analyzed for their expression of miR-150 and c-Myb mRNA. miR-150 expression level correlates with the progression of NK cell maturation, with the highest being in the most mature CD27− CD11b+ subset of NK cells (Fig. 6 A). The expression of c-Myb mRNA shows a reverse trend with the lowest being in CD27− CD11b+ subset of NK cells, consistent with a previous study showing c-Myb as a transcription factor differentially expressed by CD27+ CD11b− and CD27− CD11b+ subsets of NK cells (Chiossone et al., 2009).
Do changes in miR-150 expression inversely correlate with c-Myb expression? Because c-Myb mRNA levels were already low in freshly isolated splenic NK cells, we asked whether miR-150 deficiency leads to c-Myb up-regulation. We found an increased level of c-Myb mRNA in miR-150−/− NK cells compared with WT cells (Fig. 6 B). Transcription levels of the c-Myb targets c-kit and Bcl2 (Greig et al., 2008) were also increased in sorted miR-150−/− NK cells (Fig. 6 C), suggesting that the c-Myb protein level is likely to be altered in miR-150−/− NK cells as well.
To ascertain whether the enhanced NK cell development induced by miR-150 overexpression might be the consequence of c-Myb repression by miR-150, we studied mice with a partial loss of Myb. We generated two sets of chimeric mice by reconstituting irradiated hosts with an equal mix of BM from WT (CD45.1+) and experimental (Myb+/+ [2 doses of c-Myb gene] or Myb+/− [c-Myb heterozygous knockout]; CD45.2+) mice. The proportion of NK cells was significantly increased in the CD45.2+ compartment of WT:Myb+/− mice, compared with that in the WT:Myb+/+ chimeras (Fig. 6 D, III vs. IV in the flow cytometry plots). The frequency of NK cells among Myb+/− lymphocytes was also significantly higher than the frequency of NK cells among WT lymphocytes in WT:Myb+/− mice (Fig. 6 D, II vs. IV in the flow cytometry plots), thus suggesting a cell intrinsic mechanism for c-Myb regulation of NK cell generation. Furthermore, there was a significant increase in the percentage of lineage− CD122+ Myb+/− NK cells compared with that of Myb+/+ cells (Myb+/+, 0.5 ± 0.1% vs. Myb+/−, 1.3 ± 0.2%; P = 0.005; n = 5; Fig. 6 E) in the BM of WT:Myb+/+ and WT:Myb+/− mice. Splenic Myb+/− NK cells expressed higher amounts of CD11b compared with their Myb+/+ counterparts (Fig. 6 F), suggesting that Myb+/− NK cells resembled more mature NK cells. Together, these results indicate the NK cell developmental phenotype observed in WT:Myb+/− chimeras resembles that in WT:miR-150Tg chimeras, suggesting that the latter could be explained, at least partially, by the control of c-Myb expression by miR-150. This is reminiscent of the observation that Myb+/− mice have fewer mature B-1 B cells and block at the transition from pro–B to pre–B cells, which is similar to the phenotype of miR-150Tg mice (Xiao et al., 2007).
Acquisition of the Ly49A NK cell receptor is dependent on the dosage of miR-150 and c-Myb
NK cells express a repertoire of activating and inhibitory receptors, including members of the Ly49 gene family and CD94 that complexes with NKG2A, NKG2C, or NKG2E. CD27+ CD11b− and CD27− CD11b+ subsets of NK cells express a distinct NK cell repertoire (Hayakawa and Smyth, 2006). In particular, CD27+ CD11b− NK cells display higher expression of NKG2A, whereas CD27− CD11b+ NK cells contain a greater proportion of cells that express inhibitory Ly49 receptors (C and I isoforms), which recognize self-MHC class I molecule, H-2b (C57BL/6 mice), and the activating Ly49 receptor Ly49D. Consistent with the higher frequency of immature CD27+ CD11b− NK cell subset among miR-150−/− NK cells, expression of NKG2A was higher and expression of Ly49C/I was lower in miR-150−/− NK cells (Table I). Although we did not observe a corresponding change in these receptor expressions in miR-150Tg NK cells, they did show an impaired expression of Ly49D. Expression of neither the activating Ly49D or Ly49H, nor the inhibitory NKG2A, Ly49C/I, and Ly49G2 receptors, was affected in Myb+/− NK cells.
The striking exception was Ly49A whose expression was exquisitely sensitive to the dosage of miR-150 and c-Myb. Ly49A is an inhibitory receptor that recognizes MHC class I molecules H2Dd and H2Dk (Orr and Lanier, 2011). The size of Ly49A subset was significantly higher in splenic miR-150−/− NK cells (WT, 14.5 ± 0.8% vs. miR-150−/−, 17.5 ± 0.8%; P = 0.03; n = 14), and lower in splenic miR-150Tg NK cells (WT, 13.2 ± 0.8% vs. miR-150Tg, 4.0 ± 0.3%; P < 0.0001; n = 14; Fig. 7 A). Furthermore, Ly49A expression was lower in splenic Myb+/− NK cells compared with either internal WT control NK cells (WT, 17.1 ± 0.3% vs. Myb+/−, 7.7 ± 0.3%; P < 0.0001; n = 3) or Myb+/+ NK cells (Myb+/+, 15.6 ± 1.2% vs. Myb+/−, 7.7 ± 0.3%; P = 0.004; n = 3; Fig. 7 B). To determine at what point in NK cell development miR-150 mutant NK cells first show altered Ly49A expression, we measured Ly49A frequencies among NK cell populations in the BM of WT:miR-150Tg mice. NK1.1+ CD51+ iNK cells begin to acquire Ly49 receptors as they up-regulate CD49b (Fig. 7 C). miR-150Tg NK cells showed a marked reduction in Ly49A frequency at both intermediate (CD51+ CD49b+) and mature (CD51− CD49b+) stages. This deficit in Ly49A receptor acquisition is specific to Ly49A, as we observed similar expression of Ly49G2 receptor in miR-150Tg and WT NK cells. These data suggest that the percentage of NK cells expressing the Klra1 gene in C57BL/6 mice encoding Ly49A is determined at an early point in NK cell development.
The presence of occasional Ly49A-bearing NK cells among miR-150Tg and Myb+/− NK cells suggests that normal Ly49A expression levels are not dependent on miR-150 or c-Myb dosage. This is evident from the mean fluorescence intensity (MFI) of Ly49A+ NK cells, which is identical among the different miR-150 mutant (Fig. 7 D), and between WT and Myb+/− (Fig. 7 E) NK cells. Therefore, miR-150 and c-Myb are necessary for the acquisition of Ly49A, but they are dispensable for Ly49A cell surface expression once its expression is initiated.
Ly49A is also expressed by a small fraction of memory CD8+ T cells (Coles et al., 2000). We found that Ly49A expression was present in higher fraction of miR-150−/− T cells compared with WT T cells (WT, 0.40 ± 0.04% vs. miR-150−/−, 0.80 ± 0.07%; P = 0.0001; n = 9); accordingly, Ly49A expression was displayed in lower proportion of miR-150Tg T cells compared with control T cells (WT, 0.30 ± 0.02% vs. miR-150Tg, 0.08 ± 0.02%; P < 0.0001 [n = 9]). Together, these data indicate that the dosage of miR-150 and c-Myb influences, in a cell-autonomous manner, the percentage of cells expressing Klra1. Because potential c-Myb binding sites exist in the upstream region of Klra1 (Tanamachi et al., 2004), c-Myb might represent the transcription factor that determines Klra1 gene variegation.
The development of iNKT cells is impaired in miR-150Tg and Myb+/− mice
Previous studies have shown that global miRNA loss as a result of Dicer deficiency results in severe defects in iNKT cell development (Fedeli et al., 2009; Zhou et al., 2009; Seo et al., 2010). iNKT cells developing in the thymus initially acquire Vα14-Jα18 TCR rearrangement, and then down-regulate CD24, and finally up-regulate CD44 and NK1.1 as the last maturation step (Godfrey et al., 2010). Populations of thymic iNKT cells, identified using α-galactosylceramide–loaded CD1d tetramer (CD1d-tet; Matsuda et al., 2000), were purified and analyzed for their expression of miR-150 (Fig. 8 A). Because miR-150 is up-regulated during iNKT cell maturation in the thymus, we hypothesized that miR-150 could also play a role in the development of this lineage. We isolated thymocytes from WT:miR-150−/− and WT:miR-150Tg chimeras and stained for iNKT cells. Although the frequency of thymic miR-150−/− iNKT cells was mildly decreased (1.5 ± 0.2-fold compared with that of WT cells), the proportion of thymic miR-150Tg iNKT cells was significantly reduced (4.6 ± 0.8-fold compared with WT cells; Fig. 8 B). Analysis of CD1d-tet+ thymocytes showed that the majority of WT cells down-regulate CD24 and up-regulate NK1.1 (Fig. 8 C). In contrast, the proportion of CD24hi NK1.1− cells was significantly higher among miR-150Tg iNKT cells as compared with WT cells, whereas the proportion of mature stage CD24lo NK1.1+ miR-150Tg iNKT cells was reduced. Analysis of miR-150−/− iNKT developmental intermediates revealed a mild reduction in the fully mature CD24lo NK1.1+ iNKT cells; however, the data were not statistically significant. As CD24 is normally down-regulated during positive selection, these results suggest that miR-150 is important during the positive selection of iNKT cells. The terminal maturation stages during iNKT cell development involve up-regulation of Ly49 receptors (Gapin et al., 2001) and CD122 (Matsuda et al., 2002) in addition to NK1.1 (Benlagha et al., 2002). miR-150Tg iNKT cells had lower proportion of Ly49G2-expressing cells compared with that among WT iNKT cells (WT 49.2% ± 2.5% vs. miR-150Tg 36.6% ± 3.7%; P = 0.05 [n = 3]), although they had normal levels of CD122 (unpublished data). The observed decrease of miR-150Tg iNKT cells was also found in spleen (3.7 ± 0.6-fold compared with WT cells; Fig. 8 D). Together these data suggest that optimal dose of miR-150 is necessary for the development and maturation of iNKT cells.
Myb deficiency in the thymus results in a complete absence of iNKT cells (Hu et al., 2010). If miR-150 regulates iNKT cell development by regulating c-Myb, partial loss of c-Myb would be predicted to have the same effect on the development of this lineage as the overexpression of miR-150. To test this hypothesis, we stained for iNKT cells in the spleen of WT:Myb+/+ and WT:Myb+/− chimeras. The frequency of Myb+/− iNKT cells was significantly reduced (2.8 ± 1-fold compared with WT cells; Fig. 8 E). We also examined the iNKT cell population in the thymus of WT:Myb+/+ and WT:Myb+/− chimeras and found fewer Myb+/− iNKT cells that exhibited more immature (CD24hi NK1.1− and CD24hi NK1.1−) phenotype (unpublished data). These data suggest that the iNKT cell developmental phenotype observed in miR-150Tg chimeras is at least partially caused by the alteration in c-Myb levels.
DISCUSSION
Although miR-150 was initially appreciated as a key miRNA in B cell development (Zhou et al., 2007), it is becoming increasingly apparent that it also functions in other cell types of the hematopoietic system. miR-150 modulates lineage fate in megakaryocyte progenitors, driving differentiation of megakaryocytes at the expense of erythroid cells (Lu et al., 2008). Overexpression of miR-150 in hematopoietic stem cells leads to a mild increase in granulocytes (Zhou et al., 2007). In this study, we have demonstrated an important role for miR-150 in directing that development and maturation of the NK and the CD1d-restricted NKT cell lineages as well. In the absence of miR-150, peripheral numbers of NK cells are decreased; the cells appear to be halted at a stage before final maturation, and their ability to produce IFN-γ after activation is diminished. That miR-150 promotes NK cell development and maturation was corroborated in experiments in which miR-150 was ectopically expressed in BM. This led to an accumulation of more mature hyperfunctional NK cells. Furthermore, proper dosage of miR-150 is also necessary for the development and maturation of iNKT cells. Importantly, we find that miR-150 acts in a cell-intrinsic manner in both lineages. Because NK cells and iNKT cells come from different committed precursors and develop in different locations (Di Santo, 2006; Godfrey et al., 2010), miR-150 most likely acts independently in both cell types after lineage commitment. Given that miR-150 expression is low in the progenitor populations of each of these lineages compared with mature NK and iNKT cells, it is unlikely that it would regulate the formation of these committed precursors.
miR-150 effects in NK and iNKT cells are at least partially mediated by its target gene Myb. Transcript levels of c-Myb and its target genes c-kit and Bcl2 are increased in miR-150−/− NK cells. Partial ablation of c-Myb phenocopies increased NK cell frequency and impaired iNKT cell development observed in miR-150Tg chimeric mice. It is possible that other miR-150 target genes may also contribute to the NK and iNKT cell developmental phenotypes observed in miR-150−/− and miR-150Tg chimeric mice. Bioinformatics analyses predict hundreds of target genes for each miRNA (Rajewsky, 2006). Only a few of these, however, may represent targets with a significant biological function or be relevant in a particular cellular context. Pim-1, a predicted target of miR-150 (Xiao et al., 2007), is known to stimulate c-Myb activity (Leverson et al., 1998), indicating that miR-150 may suppress multiple genes in the c-Myb pathway. The expression pattern of Pim-1 mimics c-Myb, and is inversely correlated with that of miR-150 during NK cell activation (Heng and Painter, 2008), suggesting that it might be functionally relevant for the regulation of NK cells. Recently, miR-150 was shown to target the transcription factor Egr2 (Wu et al., 2010). Egr2 is an essential regulator of iNKT cell development and is thought to control iNKT cell proliferation and apoptosis (Lazarevic et al., 2009). Interestingly, although expressed in developing NK cells (unpublished data), Erg2 is not necessary for NK cell development (Lazarevic et al., 2009), suggesting that miR-150–mediated regulation of Eg2 levels might have functional consequences for iNKT but not NK cells. Furthermore, another miR-150 target, the purinergic P2X7 receptor (L. Zhou et al., 2008), has been implicated in regulating iNKT cell activation (Kawamura et al., 2006). Targeted mutagenesis of the miR-150–binding sites in the Myb 3′UTR or a genetic rescue of miR-150−/− phenotype by the partial loss of c-Myb will ultimately be required to determine to what extent contributions of target genes other than c-Myb might explain the phenotypes of NK and iNKT cell compartments.
Several transcription factors that regulate the development of NK cells have been described (Hesslein and Lanier, 2011). Members of the Ets family of transcription factors control several aspects of hematopoiesis in general, and NK cell development in particular. PU.1 deficiency results in the generation of reduced numbers of NKP and mature NK cells (Colucci et al., 2001), whereas deficiencies in Ets1 and MEF result in severe defects in the development of NK and iNKT cells (Barton et al., 1998; Walunas et al., 2000). T-bet, a member of the T-box family of transcription factors, controls various aspects of NK cells (Hesslein and Lanier, 2011). Mice lacking T-bet have a profound, cell-intrinsic defect in their ability to generate mature NK and iNKT cells (Townsend et al., 2004). Both cell lineages fail to complete maturation, leading to decreased numbers in the periphery. Bcl11b is a zinc finger protein that plays an essential role in specifying T cell fate by both up-regulating T cell–specific genes and suppressing NK cell genes (Li et al., 2010). When Bcl11b was deleted, T cells acquired expression of NK cell lineage molecules including CD122, T-bet, NKp46, and NKG2A receptors, perforin, and IFN-γ after in vitro culture. Perhaps the accumulation of miR-150Tg NK cells is a result of reprogrammed NK-like cells with altered Bcl11b expression? This scenario is unlikely as the mouse Bcl11b gene does not contain the miR-150 seed sequence in its 3′ or 5′ UTR or its coding sequence (Dweep et al.. 2011), and there is no evidence suggesting that Bcl11b mRNA levels are altered in miR-150 mutant NK cells (unpublished data). c-Myb has recently been shown to be required for the development and function of iNKT cells (Hu et al., 2010). Further investigation of a role for c-Myb in NK cell development requires loss of c-Myb specifically in NK cell lineage. With the development of a NK cell–specific Cre mice (Eckelhart et al., 2011), it will now be feasible to test the hypothesis that c-Myb is a negative regulator of NK cell development.
The factors that control Ly49 gene expression are beginning to be defined. We show here that miR-150 and c-Myb are required for the acquisition of the Ly49A receptor. Several scenarios are possible to explain the reduced frequency of Ly49A+ NK cells in miR-150 transgenic and c-Myb heterozygous mice. First, miR-150 and c-Myb may be involved in the formation of Ly49 receptor repertoire at the level of its adaptation to the MHC class I environment. This is unlikely for Ly49A because Ly49A is not known to recognize the MHC class I H2b allele in C57BL/6 mice (Orr and Lanier, 2011), and our studies were conducted in mixed BM chimeras where both WT and mutant cells develop in the same environment. Alternatively, the reduced frequency of Ly49A+ NK cells might result from differences within the CD27+ CD11b− and CD27+ CD11b− cell subsets between miR-150Tg and Myb+/− NK cells. To this end, we co-stained NK cells from different chimeric mice with anti-CD11b antibody and measured Ly49A expression. Both CD27+ CD11b− and CD27− CD11b+ subsets among miR-150Tg NK cells showed a lower proportion of Ly49A+ NK cells compared with WT cells; similar findings were observed in Myb+/− chimeras (unpublished data). In a third possible scenario, a lack of transcription factors necessary for Ly49A expression could explain the reduction of Ly49A subset in miR-150Tg and Myb+/− NK cells. Gata-3 has been shown to be a direct target of c-Myb (Maurice et al., 2007), and Gata-3 deficiency leads to a reduction of the Ly49A subset of NK cells (Samson et al., 2003). Trans-acting factor transcription factor-1 has been shown to bind the Klra1 promoter and regulate the Ly49A-bearing NK cell subset in a dosage-dependent manner (Held et al., 1999). Putative binding sites for the transcription factors Runx1 and Atf2 have also been suggested to control Ly49A expression (Kubo et al., 1999; Kunz and Held, 2001), but at least in the case of Atf2, the Ly49 receptor repertoire analysis does not seem to indicate a requirement for this transcription factor for Ly49A expression (Kim et al., 2006). Lastly, it is possible that c-Myb itself binds to and regulates the Klra1. A c-Myb–binding sequence is found in the upstream region of Klra1 (Tanamachi et al., 2004), and we find that the size of the Ly49A-expressing NK cell subset is dependent on the dosage of c-Myb. Future studies in c-Myb–deficient NK cells will definitively prove whether c-Myb is necessary for the acquisition of Ly49A.
We find that miR-150 differentially affects the development of NK and iNKT cell lineages. The literature supports the notion that miR-150 plays different roles depending on cell context. Forced expression of miR-150 promoted proliferation of gastric cancer cells (Wu et al., 2010), whereas it increased the rate of apoptosis and reduced proliferation of NK/T lymphoma cells (Watanabe et al., 2011). miR-150Tg mice have a reduced number of splenic B cells, and miR-150Tg B cells from the BM showed increased sensitivity to apoptosis when cultured in vitro (Xiao et al., 2007). In contrast, there was an accumulation of NK cells in miR-150Tg chimeric mice and no difference was seen in the rate of apoptosis between WT and miR-150Tg NK cells. The differences in susceptibility to apoptosis can be partially explained by the higher frequency of mature NK cells and greater percentage of immature B cells (Xiao et al., 2007), but the maturation status alone is probably insufficient to explain the entire phenotype of miR-150Tg NK cells. B cells depend on positive signaling through the B cell receptor to promote survival and terminal differentiation (Meffre et al., 2001). B-1 B cells, which require strong BCR signaling to develop (Hayakawa et al., 1999), are significantly reduced in miR-150Tg mice (Xiao et al., 2007). The opposite is true for NK cells, which rely on negative signaling through the inhibitory receptors to reach a mature, functionally competent state (Fernandez et al., 2005; Kim et al., 2005; Anfossi et al., 2006). Future studies will examine how miR-150 effects the survival of iNKT cells. Whereas c-Myb is necessary for the survival of double-positive thymocytes (Bender et al., 2004), ectopic expression of miR-150 did not significantly effect T cell development in the thymus (Xiao et al., 2007; Zhou et al., 2007). Another unexpected finding was the relatively unchanged number of miR-150−/− iNKT cells. Perhaps developing iNKT cells are exquisitely sensitive to the dose of miR-150 and its regulated genes; too much miR-150 inhibits iNKT cell development, whereas lack of miR-150 also leads to a mild reduction. It is also possible that miR-150 might have both a negative and a positive role at different stages of iNKT cell development, depending on the expression of the target genes it regulates. This context-specific property may perhaps explain why miR-150 expression varies between different lineages and different developmental stages within the hematopoietic system (Kuchen et al., 2010). In conclusion, this is the first study showing the requirement of a specific miRNA for the development and maturation of NK and iNKT cells.
MATERIALS AND METHODS
Mice.
M. Ansel (University of California, San Francisco, San Francisco, CA) provided miR-150−/− mice. K. Rajewsky (Harvard Medical School, Boston, MA) provided miR-150Tg BM cells (Xiao et al., 2007). T.P. Bender (University of Virginia Health System, Charlottesville, VA) provided Myb+/+ and Myb+/− BM cells (Bender et al., 2004). To generate mixed BM chimeras, 6–8-wk-old CD45.1-congenic C57BL/6 mice (National Cancer Institute, Frederick, MD) were lethally irradiated with 1,000 rads, and then reconstituted with a 1:1 mix of CD45.1 and either miR-150−/−, miR-150Tg, Myb+/+, or Myb+/− BM cells. Mixed BM cells were then retroorbitally injected into the host mice. Mice were analyzed 8–14 wk after reconstitution. CD45.1+ C57BL/6 mice were purchased from the National Cancer Institute. Rag2−/− x Il2rg−/− C57BL/6 mice were purchased from Taconic. Experiments were done according to the University of California, San Francisco Institutional Animal Care and Use Committee guidelines.
Flow cytometry.
Fc receptors on cells isolated from peripheral blood, spleen, liver, BM, and thymus were blocked with anti-CD16 + CD32 mAb (clone 2.4G2) at 10 µg/ml before surface staining. Cells were stained with antibodies against NK1.1, TCRβ, CD45.1, CD45.2, CD27, CD11b, NKp46, CD122, CD49b, CD51, CD4, CD8, CD11c, CD19, GR1, Ter119, Ly49H, Ly49D, CD94, NKG2A, Ly49C/I, Ly49G2, Ly49A, CD24, and CD117 (from BD, eBioscience, or BioLegend). Intracellular IFN-γ staining was performed according to the manufacturer’s instructions (BD). iNKT cells were identified with a Pacific blue–conjugated mouse CD1d tetramer either unloaded or loaded with PBS57. For apoptosis analysis, freshly isolated splenocytes were first stained with antibodies against NK1.1, TCRβ, CD45.1, CD45.2, and Live/Dead fixable near-IR stain (Invitrogen), washed, and then stained with PE-conjugated Annexin V (BD), according to the manufacturer’s protocol. For BrdU labeling, mice were injected i.p. with 200 µg of BrdU (Sigma-Aldrich) in PBS every 24 h for 3 consecutive days, and then sacrificed. For the detection of incorporated BrdU, cells were first stained for surface antigens, and then fixed, permeabilized, treated with DNase I, and stained with APC-conjugated anti-BrdU mAb (BD). Flow cytometry was performed on an LSRII, and data were analyzed with FlowJo software (Tree Star).
CFSE labeling and adoptive transfer.
Enriched (60–80% NK1.1+ TCRβ−) cells from WT and miR-150−/− mice (mixed at a 1:1 ratio) or splenocytes from WT:miR-150Tg chimeras were labeled for 8 min with 10 µM CFSE in accordance with the manufacturer’s instructions (Invitrogen), and transferred into Rag2−/− x Il2rg−/− mice. 3 (Fig. 1) or 5 d (Fig. 5) later, NK cells were analyzed by flow cytometry for CFSE dilution and expression of CD27.
Ex vivo stimulation assay.
Splenocytes were stimulated in Immulon microtiter plates (Thermo Fisher Scientific) coated with 10 µg anti-NK1.1, anti-Ly49H, or control IgG in PBS for 4 h at 37°C in the presence of GolgiStop (BD), followed by staining for intracellular IFN-γ (BD). For cytokine stimulation, splenocytes were incubated with 20 ng/ml IL-12 (R&D Systems) and 10 ng/ml IL-18 (R&D Systems).
Cell sorting.
NK cells from spleen were enriched with antibodies against CD5, CD4, CD8, Ter119, Gr-1, CD19 (UCSF Antibody Core), and anti–rat IgG-coated magnetic beads (Miltenyi Biotec). NK cells were stained with antibodies against NK1.1, TCRβ, CD45.1, CD45.2, NK1.1, TCRβ, CD11b, and CD27 and were sorted using a FACSAria (BD). NK populations from the BM were sorted using antibodies against CD4, CD8, CD19, GR-1, CD11c, Ter119, CD122, NK1.1, CD51, and CD49b. Thymic iNKT cells were sorted with PBS57-loaded CD1d tetramer and antibodies against NK1.1, TCRβ, and CD24. The purity of the recovered populations was typically >98%.
Quantitative RT-PCR (qRT-PCR).
Total RNA was extracted from sorted cells with RNAqueous-Micro kit (Invitrogen). miR-150 expression was examined by using the TaqMan MicroRNA Assay kit (Applied Biosystems) with gene-specific stem-loop PCR primers and TaqMan probe (IDT). SnoRNA202 was used as endogenous control. cDNA was generated by using a SuperScript III First-Strand kit (Invitrogen) and real-time PCR was performed on an Applied Biosystems 7500. PCR of cDNA was conducted using ABI TaqMan kits (c-Myb and HPRT) or SYBR Green PCR Master Mix (Invitrogen). The primer sequences for SYBR green PCR reactions were as follows: c-kit, forward, 5′-TCATCGAGTGTGATGGGAAA-3′; reverse, 5′-GGTGACTTGTTTCAGGCACA-3′; Bcl2, forward, 5′-TACCGTCGTGACTTCGCAGAG-3′; reverse, 5′-GGCAGGCTGAGCAGGGTCTT-3′; HPRT, forward, 5′-CACAGGACTAGAACACCTGC-3′; reverse, 5′-GCTGGTGAAAAGGACCTCT-3′. Data were analyzed by the CT method (Livak and Schmittgen, 2001).
Statistical analysis.
All data shown are the mean ± SEM. Comparisons between samples were performed by using a two-tailed unpaired Student’s t test, unless otherwise stated. Statistics were calculated using Prism software (GraphPad Software, Inc.). P values were denoted in the following manner: *, P < 0.05; **, P < 0.01; ***, P < 0.001.
Acknowledgments
We thank K. Rajewsky for BM cells; J.L. Baron for the CD1d-PBS57 tetramer; J. Arakawa-Hoyt for assistance with cell sorting; H. Consengco for assistance with genotyping; M. Ansel (University of California, San Francisco), C. Kim (University of California, San Francisco), and J. Sun (Memorial Sloan-Kettering Cancer Center) for critical review of the manuscript; and members of the Lanier laboratory for discussions.
T. Chakraborty was supported by grants from the National Institutes of Health (NIH) to K. Rajewsky. N.A. Bezman is supported by post-doctoral fellowship (117346-PF-09186-01-LIB) from the American Cancer Society. This study was supported by NIH grant AI068129 and L.L. Lanier is an American Cancer Society Professor.
The authors have no financial conflicts of interest.