KLF10-deficient mice with C57Bl/6 (B6) background were kindly provided by Dr. Woon Kyu Lee (Inha University, Incheon, South Korea) (28). B6.Rag1-deficient mice and B6.CD45.1 congenic mice were obtained from The Jackson Laboratory. All animals were bred and maintained under specific pathogen-free conditions at the Institute of Laboratory Animal Resource Seoul National University and treated in accordance with institutional guidelines that were approved by the Institutional Animal Care and Use Committee (SNU-140930-4-1).

Mouse peripheral lymph nodes (cervical, axillary, brachial, and inguinal), mesenteric lymph node, spleen, thymus, and lung were homogenized by mechanical disaggregation, strained through a 70-µm strainer (BD Biosciences), and washed in RPMI 1640 medium containing 10% (vol/vol) fetal bovine serum (FBS). Peritoneal cells were obtained from peritoneal lavage in cold phosphate-buffered saline (PBS) containing 5% FBS.

Single-cell suspensions were first blocked with anti-CD16/32 antibody (93; eBioscience) and then stained with antibodies at 4°C for 20 min in staining buffer (1 × PBS containing 0.1% bovine serum albumin and 0.1% sodium azide). For intracellular cytokine staining, the cells were stimulated for 4 h with 50 ng/ml phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich) and 750 ng/ml ionomycin (Sigma-Aldrich) in the presence of brefeldin A (BD Biosciences). The cells were then fixed, permeabilized with a BD Cytofix/Cytoperm Kit according to the manufacturer’s instructions (BD Biosciences), and stained for IL-17 and IFN-γ. Intracellular phosphorylated protein staining was performed as described (29) with modifications. For assays of intracellular Ca²⁺ mobilization, cells from WT and knockout (KO) mice were first surface-stained with anti-CD45.2-PerCP-Cy5.5 and anti-CD45.2-Alexa 700, respectively, for subsequent identification of the two strains. Cells from both strains were mixed together at a ratio of 1:1 and loaded for 45 min at 37°C with the membrane-permeable fluorescent Ca²⁺ indicator dye, Indo-1 AM (Invitrogen), at a concentration of 4 µM in RPMI medium plus 5% (vol/vol) FBS. Cells were then stained for surface markers and were kept on ice. Before stimulation, cell aliquots were allowed to equilibrate to 37°C for 5 min and analyzed by flow cytometry. After acquisition of background intracellular Ca²⁺ concentrations for 30 s, cells were stimulated with biotinylated anti-CD3ε antibody and crosslinked by the addition of streptavidin. Samples were acquired on Canto II or Aria II (BD Biosciences) and data were analyzed with FlowJo software (TreeStar). Antibodies used for flow cytometric analyses were fluorochrome-labeled mAbs against mouse γδTCR (GL3), CD3ε (145-2C11), CD27 (LG.3A10 or LG.7F9), CD25 (PC61), CD69 (H1.2F3), CD5 (53-7.3), CD24 (M1/69), CD103 (2E7), CD122 (TM-b1), CD132 (TUGm2), CD127 (eBioSB/199), Ly6C (HK1.4), CD28 (E18), CD44 (IM7), CD62L (MEL-14), CCR6 (29-2L17), NK1.1 (PK136), CD45.1 (A20), CD45.2 (104), Vγ1 (2.11), Vγ4 (UC3-10A6), IL-17 (TC11-18H10), IFN-γ (XMG1.2), CD4 (RM4-5), CD8α (53-6.7), TCRβ (H57-597), pErk1/2 (D13.14.4E), pZap70 (n3kobu5), pSTAT3 (D3A7), and pSTAT5 (47/Stat5) (from BD Biosciences, Biolegend, eBioscience, or Cell Signaling Technology).

After blocking with anti-CD16/32, cells were negatively selected from pooled pLN and splenocytes using anti-TCRβ biotin, anti-CD45R biotin (RA3-6B2), and anti-CD11b (M1/70) antibodies together with MagniSort™ Streptavidin Negative Selection Beads according to the manufacturer’s instructions (eBioscience). The negatively selected cells were then stained with fluorochrome-labeled anti-γδTCR, anti-CD27, and/or anti-Vγ4 antibodies. γδ subsets (Vγ4^{− γδ27+}, Vγ4+ γδ27+, Vγ4^{− γδ27−}, and Vγ4^{+ γδ27−}) were sorted using an Aria II with a purity of at least 98%. Alternatively, total γδ cells were manually isolated by negative selection as described previously (30) with modification. Biotin-conjugated mAbs against the following proteins were used: TCRβ, CD4, CD8α, CD45R, CD11b (M1/70), CD11c, Ly-6G (RB6-8C5), TER-119 (TER-119), and CD49b (DX5) (all from eBioscience).

Cells were cultured in RPMI 1640 containing 10% (vol/vol) FBS, 50 µM 2-mercaptoethanol, 1% (vol/vol) non-essential amino acids, 10 mM HEPES, and 1% (vol/vol) antibiotics/antimycotic solution (all from Invitrogen). For TCR stimulation, the cells were incubated for the indicated times with plate-bound anti-CD3ε mAb (0.1, 1, or 10 µg/ml; PeproTech) only or in combination with plate-bound anti-CD28 mAb (1 or 10 µg/ml; PeproTech). For cytokine stimulation, IL-7 (20 ng/ml; R&D Systems), TGF-β (10 ng/ml; R&D Systems), IL-6 (20 ng/ml; R&D Systems), IL-1β (10 ng/ml; PeproTech), and IL-23 (20 ng/ml; PeproTech) were added to the medium.

Bone marrow cells (5 × 10⁶ cells) from B6.CD45.2 wild-type (WT) or B6.CD45.2 KLF10-deficient mice were intravenously transferred to B6.CD45.1 mice, which were irradiated at 900 cGy. For mixed BM chimeras, B6.CD45.1/2 mice were lethally irradiated at 900 cGy and intravenously administered 5 × 10⁶ B6.CD45.1 WT cells mixed 1:1 with 5 × 10⁶ B6.CD45.2 KLF10-deficient BM cells. Recipient mice were sacrificed and analyzed for reconstituted γδ T cells after at least 12 weeks.

Single-cell suspensions of pLN cells obtained from B6.CD45.1 WT and B6.KLF10 knockout (KO) mice were stained with Cell Trace Violet (CVT; Invitrogen) and injected intravenously with 2 × 10⁶ cells into Rag-1-deficient mice. After 5 days, pLNs from the recipient mice were collected and examined for CTV dilution of the transferred T cells.

RNA was purified from cells using a Qiagen RNeasy kit. After reverse transcription into cDNA, PCR was performed with a StepOnePlus real-time PCR system (Applied Biosystems) and iTaq SYBR Green Supermix (Applied Biosystems) and relative expression was displayed in arbitrary units or as a percent of maximum expression, normalized to Eef1a1 (encoding eukaryotic translation elongation factor 1α1; called “Efa1” hereafter) via the ΔΔCt method. The following primers were used: Efa1 forward, 5′-TCCACCGAGCCACCATACA-3′, reverse, 5′-CCAACCAGAAATTGGCACAA-3′; Klf10 forward, 5′-ACCCAGGGTGTGGCAAGAC-3′, reverse, 5′-AGCGAGCAAACCTCCTTTCA-3′; Rorc forward, 5′-TCAGCGCCCTGTGTTTTTCT-3′, reverse, 5′-CAAATTGTATTGCAGATGTTCCA-3′; Sox13 forward, 5′-CTGCCACCTGGGTTACTTTGA-3′, reverse, 5′-GAGTGGCGTGATGAACATGTG-3′.

Prism software (GraphPad) was used for all statistical analyses. All quantitative data are shown as mean ±  standard deviation (SD) unless otherwise indicated. The two-tailed, paired t-test was used for BM chimeras. The two-tailed, unpaired t-test or two-way ANOVA followed by a Bonferroni post hoc test were used for all other data sets. Mean fluorescence intensity (MFI) indicates geometric MFI. Robust coefficient of variation (CV) defines 100 × 1/2 [intensity (at 84.13 percentile) − intensity (at 15.87 percentile)/Median] and was determined by FlowJo software. The robust CV is a normalized SD not as skewed by outlying values as the CV.

KLF10 has been reported to control Treg and Th17 cell induction (1, 3). Nonetheless, alterations of these cells in KLF10-deficient mice are still controversial (2, 3) and even the general T-cell status of the KO mice is not clearly defined (31). Therefore, we first examined the frequencies and absolute numbers of αβ and γδ T cells in the pLN, spleen, lung, and peritoneal cavity of KO mice under specific pathogen-free conditions. Compared with WT mice, levels of γδ T cells, but not conventional CD4⁺ and CD8⁺ αβ T cells, were significantly increased in pLN and lung of KO mice (Figures S1A,B in Supplementary Material). Analysis of discrete LNs (cervical, axillary, brachial, inguinal, and mesenteric; Figure S1C in Supplementary Material) confirmed the higher frequencies of γδ T cells, with the exception of mesenteric LN (mLN). Interestingly, the increase in γδ T cells was largely attributable to γδ²⁷⁻ cell augmentation (Figures 1A,B). As expected, higher numbers of γδ²⁷⁻ cells were observed in all LNs except for mLN (Figure S1D in Supplementary Material), in which innate-like γδ-17 cells are reported to be absent (5).

CD27 expression discriminated γδ subsets with CD44, CD62L, NK1.1, and CCR6; pLN γδ²⁷⁻ cells were delineated as NK1.1⁻CCR6⁺CD44^hiCD62L⁻ effector memory-phenotype cells (11, 32). Consistent with the augmentation of γδ²⁷⁻ cells in KO mice, there were considerably more effector memory-phenotype (CD44^hiCD62L⁻) γδ T cells in the analyzed organs (Figure 1C). By contrast, similar frequencies of effector memory-phenotype CD4⁺ T and CD8⁺ T cells in these organs were observed from both strains (Figure S2A in Supplementary Material). Intriguingly, the frequencies of Tregs (Figure S2B in Supplementary Material) and Th17 cells (Figure S2C in Supplementary Material, left column) among CD4⁺ T cells were unchanged in KO mice relative to those in WT mice. In summary, these data suggested that the generation of γδ²⁷⁻ cells was enhanced in KO mice.

To explore the nature of γδ²⁷⁻ cell augmentation in KO mice, we investigated whether KLF10 deficiency affected homeostatic proliferation of γδ²⁷⁻ cells. Considering that the absolute number of KO pLN γδ²⁷⁻ cells was about twice higher than that in WT pLN cells (Figure 1B), we transferred CD45.1 WT and CD45.2 KO pLN cells together into lymphopenic mice at 2:1 ratio; we confirmed that there were the same number of γδ²⁷⁻ cells from both strains (data not shown). The homeostatic expansion of KO γδ²⁷⁻ cells was superior to that of WT γδ²⁷⁻ cells, with a ratio as great as threefold (Figures 1D,E). By contrast, we observed a similar expansion pattern of αβ T cells from both strains in recipient mice transferred with the pLN mixture at 1:1 ratio (Figures S2D–F in Supplementary Material). Therefore, these data suggested that KLF10 specifically inhibited γδ²⁷⁻ cell expansion under lymphopenic conditions.

To determine phenotypic traits of γδ²⁷⁻ cells expanded in a KLF10-deficient condition (Figures 1A–C), we investigated the expression of surface molecules involved in innate and adaptive immune features of γδ T cells. Interestingly, KO γδ²⁷⁻ cells showed lower expression of CD69, CD28, and CD5, but higher expression of CD103 and CD127, than their WT counterparts (Figure 2A). To understand these changes in the light of Vγ chains, we examined the Vγ usage and found a Vγ4-biased composition of KO γδ²⁷⁻ cells (Figure 2B, left). Indeed, Vγ4⁺ γδ²⁷⁻ cells expressed lower CD5 and CD28 but higher CD127 and CD103 than Vγ4⁻ γδ²⁷⁻ cells (data not shown), which demonstrated that the change in these surface molecules on γδ²⁷⁻ cells (Figure 2A) could be attributed to the preferential distribution of a Vγ4 subset. In addition, only Vγ4⁺ γδ²⁷⁻ cells were considerably enhanced in pLN, whereas the abundance of Vγ1⁺ γδ²⁷⁻ and Vγ1⁻Vγ4⁻ γδ²⁷⁻ cells was unchanged in KO mice compared with that in WT mice (Figure 2B, right). Collectively, our results indicated that Vγ4⁺ γδ²⁷⁻ cells were selectively enriched under KLF10 deficiency.

As γδ²⁷⁻ cells are innate-like IL-17-producing γδ T cells (γδ-17) (6), we examined intracellular levels of IL-17A. Consistent with the greater abundance of Vγ4⁺ γδ²⁷⁻ (Figure 2B) and Vγ4⁺CCR6⁺ cells (data not shown) in pLN of KO mice, considerably more IL-17⁺ cells were observed in γδ²⁷⁻ cells from KO pLN cells after stimulation with PMA plus ionomycin (Figures 2C,D). This was further confirmed by the increased number of IL-17⁺CCR6⁺ or IL-17⁺Vγ4⁺ γδ T cells (Figure 2E). Of note, IL-17 production by IL-17⁺Vγ4⁺CCR6⁺ γδ T cells was higher in the KLF10-deficient condition than in the normal condition, as measured by the MFI of intracellular IL-17 (Figure 2F), indicating that KLF10 constrained IL-17 production by Vγ4⁺ γδ²⁷⁻ cells. Finally, KO Vγ4⁺ γδ²⁷⁻ cells contained considerably higher levels of Rorc than their WT counterparts (Figure 2G). These results collectively suggested that KLF10 impaired the size of the innate-like Vγ4⁺ γδ²⁷⁻-17 population together with their IL-17-producing capacity.

Peripheral homeostasis and IL-17 production of innate-like γδ-17 cells are mainly controlled by innate signaling triggered by cytokines, such as IL-7 and IL-1β plus IL-23 (8, 9). Thus, we determined whether KLF10 was involved in cytokine-signaling on Vγ4⁺ γδ²⁷⁻ cells. Under IL-7 treatment the absolute number of Vγ4⁺ and Vγ1⁻Vγ4⁻ γδ²⁷⁻ cells from KO mice was considerably increased compared with their WT counterparts (Figure 3A). Both γδ²⁷⁻ subsets of KO mice exhibited greater proliferation (Figure 3B) and there were higher frequencies of IL-17⁺ expanding cells among KO γδ²⁷⁻ subsets (Figure 3C). Because IL-7 enriches γδ²⁷⁻-17 cells by activating STAT3 rather than STAT5 (9), we assessed STAT phosphorylation triggered by IL-7. IL-7 substantially activated STAT5 in both Vγ4⁺ and Vγ4⁻ γδ²⁷⁻ cells from WT mice, whereas the phosphorylation of STAT3 was slightly induced by IL-7 (Figure 3D), consistent with the previous report showing the capacity of IL-7 to activate STAT3 in γδ²⁷⁻ cells (9). Of note, compared to the normal condition, the KLF10 deficiency increased the level of phospho-STAT3 (pSTAT3), but not of pSTAT5, under IL-7 treatment (Figure 3D). Meanwhile, we observed the pSTAT3 induction in Vγ4⁻ γδ²⁷⁻ cells when treated with IL-6, but not in Vγ4⁺ γδ²⁷⁻ cells (Figure S3A in Supplementary Material). These data suggested that STAT3 activation might be involved in the hyper-responsiveness of KO γδ²⁷⁻ cells to IL-7. On the other hand, contrary to the aforementioned in vivo Vγ4⁺ γδ²⁷⁻-17 cell-specific regulation of KLF10 (Figure 2), in vitro hyper-responsiveness to IL-7 was observed in KO γδ²⁷⁻ cells regardless of Vγ4 (Figures 3A–C). In addition, KLF10 deficiency promoted γδ²⁷⁻ cells, irrespective of Vγ4, providing an advantage for homeostatic expansion in lymphopenic conditions (Figure 3E). Collectively, these results suggest a critical role for KLF10 in IL-7 signaling-mediated homeostasis of innate-like γδ²⁷⁻-17 cells.

We next examined whether KLF10 is also involved in the reactivity of γδ²⁷⁻ cells to inflammatory conditions such as IL-1β plus IL-23 that are known to induce IL-17 production by γδ T cells (8). Consistent with the results of IL-7 stimulation, γδ²⁷⁻ cells from KO mice were hyper-responsive to IL-1β plus IL-23 (Figures S3B,C in Supplementary Material), which indicated that KLF10 also inhibited the activation of γδ²⁷⁻ cells triggered by inflammatory stimuli. Of note, Klf10 expression was increased in γδ²⁷⁻ cells stimulated with IL-1β plus IL-23, as well as those stimulated with IL-7 (Figure 3F). On the other hand, KLF10 deficiency did not lead to an alteration in the calcium fluxes directly triggered by stimulation with PMA plus ionomycin, ruling out the possibility that the hyper-responsiveness of KO γδ²⁷⁻ cells might be qualitatively unspecific to cytokines (Figure S3D in Supplementary Material). Together, these data suggested that homeostatic and inflammatory cytokine signaling could induce KLF10, which in turn, as a negative feedback signaling factor, might impair γδ²⁷⁻ cell responsiveness to these innate stimuli.

Next, we sought to determine whether KLF10 might be involved in TCR-triggered activation of peripheral γδ T cells (13, 32). γδ²⁷⁺ cells could readily expand under TCR/CD28 stimuli, whereas γδ²⁷⁻ cells showed different patterns of proliferation and, especially, Vγ4⁺ γδ²⁷⁻ cells hardly expanded (Figure 3G), supporting a preceding report that innate-like γδ²⁷⁻ cells display hypo-responsive TCR signaling (13). Although Klf10 could be significantly induced by TCR activation in total γδ T cells, in which γδ²⁷⁺ cells accounted for up to 85% of the cells (Figure 3H), there were similar expansions of each γδ subset between WT and KO mice (Figure 3G). As previously reported (13), a rapid and transient increase in cytosolic calcium concentration triggered by TCR engagement was readily detected in γδ²⁷⁺ cells, but not in γδ²⁷⁻ cells; however, there were no differences in calcium fluxes between the strains (Figure 3I). Most of all, a Vγ4⁺ γδ²⁷⁻ subset almost completely failed to phosphorylate ERK after TCR stimulation (Figure S3E in Supplementary Material). Therefore, KLF10 has a minor role in TCR-triggered activation of peripheral γδ subsets. To further investigate whether the level of engagement by the antigen or costimulatory receptor influenced KLF10 involvement in the TCR response, we treated the cells with different doses of agonist antibodies. Vγ4⁺ γδ²⁷⁻ cells from both strains responded similarly to a high level of antigen engagement even in the absence of costimulatory signaling, but still showed an impaired proliferative response compared to that of γδ²⁷⁺ cells (Figure S3F in Supplementary Material); similar results were found with costimulatory stimuli. Therefore, neither the strong antigen receptor nor costimulatory signaling caused a substantial discrepancy in TCR response of peripheral γδ subsets between the two strains.

The expression level of CD5, a stable indicator of TCR strength, was lower in γδ²⁷⁻ cells than in γδ²⁷⁺ cells (Figure 2A), indicating that γδ²⁷⁻ cells might receive a relatively weak TCR strength compared with γδ²⁷⁺ cells; this is in line with the fact that a weak TCR-signal strength is required for thymic development of innate-like γδ-17 cells (11, 33). Intriguingly, when we monitored CD5 expression (Figure 4A), γδ²⁷⁻ cells contained two different populations (CD5^high and CD5^low), meaning that they are a heterogeneous group receiving discrete TCR-signal strengths. Such a two-peak pattern was also observed for CD127 (IL-7 receptor-α, IL-7Rα), one of the markers identifying innate-like γδ-17 cells (Figure 4A) (9, 13), allowing us to distinguish CD5^loCD127^hi and CD5^hiCD127^lo cells (Figure 4B) and further presume that CD5^loCD127^hi γδ²⁷⁻ cells might represent the innate-like γδ-17 cells. Indeed, the CD5^loCD127^hi γδ²⁷⁻ subset was greater in KO mice, whereas the CD5^hiCD127^lo γδ²⁷⁻ subset was present in normal numbers (Figure 4C).

Interestingly, CD5^hiCD127^lo γδ²⁷⁻ cells closely resembled γδ²⁷⁺ cells rather than CD5^loCD127^hi γδ²⁷⁻ cells in terms of surface protein expression (Figure 4D) and Vγ usage (Figure 4E). We noted that the frequency of CD5^hiCD127^lo cells among γδ²⁷⁻ cells was very low (3.4%) in young mice (2 weeks old) but greatly increased (45.5%) in the adult (8 weeks old) (Figure 4F). These cells mainly appeared within mature (CD24^lo) γδ²⁷⁻ cells of the thymus (Figure S4 in Supplementary Material) and pLN (Figure 4G), and were accompanied by higher levels of Vγ1⁺ cells. In addition, CD5^hiVγ1⁺ cells emerged among immature (CD24^hi) γδ²⁷⁻ cells of the adult thymus (Figure S4 in Supplementary Material), in accordance with the previous report on sequential Vγ waves with age (34) and possibly indicating that CD5^hiCD127^lo γδ²⁷⁻ cells might be generated in the adult rather than the fetal/neonatal thymus.

In contrast to CD5^hiCD127^lo γδ²⁷⁻ and γδ²⁷⁺ cells, CD5^loCD127^hi γδ²⁷⁻ cells had common γ chain receptor signal-dependent and epithelial-homing phenotypes with Vγ4⁺ subset dominance (Figures 4D,E). In particular, IL-17⁺ γδ²⁷⁻ cells were CD5^loCD127^hi (Figure 4H), which ultimately validated CD5^loCD127^hi γδ²⁷⁻ cells as “virtually” innate-like γδ-17 cells.

By scrutinizing the expression patterns of CD5 and CD127 on γδ²⁷⁻ cells, depicted in a pseudocolor plot (Figure 4B), we could recognize another peak of CD5 that was relatively high in CD5^loCD127^hi γδ²⁷⁻ cells under KLF10 deficiency, but otherwise was barely detectable under normal conditions. Thus, we further distinguished CD5^int cells from CD5^loCD127^hi γδ²⁷⁻ cells (Figure 5A) and found that the CD5^intCD127^hi subgroup was expanded in a KLF10-deficient condition (Figure 5B). This enriched sub-group skewed toward a Vγ4 chain (Figure 5C), consistent with the selective enrichment of a Vγ4⁺ γδ²⁷⁻ subset in KO mice (Figure 2B). Indeed, IL-17 production was observed in both CD5^lo and CD5^int CD127^hi γδ²⁷⁻ cells, and the latter population (IL-17⁺CD5^intCD127^lo) was genuinely augmented by KLF10 deletion (Figure 5D). Taken together, these findings indicated that KLF10 deficiency selectively expanded innate-like IL-17-competent CD5^intVγ4⁺CD127^hi γδ²⁷⁻ cells.

In normal conditions, CD5^lo and CD5^int CD127^hi γδ²⁷⁻ subgroups exhibited preferential distribution of Vγ4⁺ and Vγ1⁻Vγ4⁻ subsets, respectively (Figure 5C), raising the possibility that KLF10-deficient Vγ4⁺ γδ²⁷⁻ cells might express higher amounts of surface CD5 than their WT counterparts. Indeed, we observed elevated CD5 expression on Vγ4⁺ γδ²⁷⁻ cells of pLN from KO mice (Figure 5E; Figure S5 in Supplementary Material). It is noting that the considerable increase of CD5 was detected only in Vγ4⁺ γδ²⁷⁻ cells, but not in other immune cells, including naive (CD44^loCD62L^hi), effector (CD44^loCD62L^lo), and memory (CD44^hiCD62L^hi as central, CD44^hiCD62L^lo as effector) phenotypes of CD4⁺ or CD8⁺ T cells (data not shown). Because CD5 is positively associated with the strength of TCR signaling that T cells received during their selection within the thymus (35), we next assessed the expression level of CD5 on thymic γδ T cells in neonates in which Vγ4⁺ γδ²⁷⁻-17 cell maturation actively occurred (22). Interestingly, we observed considerably higher surface CD5 expression on mature (CD24^loCD44^hi) thymic Vγ4⁺ γδ²⁷⁻ cells under conditions of KLF10 deficiency, whereas the difference in CD5 was equivocal on the immature (CD24^hiCD44^lo) cells (Figures 5F,G). However, consistent with the peripheral observation (Figure S3E in Supplementary Material), TCR-triggered phosphorylation of ERK in neonatal thymic Vγ4⁺ γδ²⁷⁻ cells was quite similar between both strains regardless of their maturation; ERK was hardly activated by TCR stimulation in Vγ4⁺ γδ²⁷⁻ cells, contrary to the significant activation of ERK in γδ²⁷⁺ cells (Figure 5H). These data collectively suggested that KLF10 deficiency resulted into CD5^intVγ4⁺CD127^hi γδ²⁷⁻ cell development by controlling thymic maturation of Vγ4⁺ γδ²⁷⁻ cells in neonates.

Next, we determined whether the selective expansion of CD5^intVγ4⁺ γδ²⁷⁻-17 cells under KLF10 deficiency was developmental. As expected, Vγ4⁺ γδ²⁷⁻ cell maturation was enhanced in the neonatal thymus of KO mice, as measured by the frequency of mature (CD24^loCD44^hi) cells among Vγ4⁺ γδ²⁷⁻ cells and their absolute number (Figures 6A,C). Intriguingly, frequencies of IL-17⁺ cells were higher in both Vγ1⁻Vγ4⁻ and Vγ4⁺ γδ²⁷⁻ cells of KO neonatal thymus (Figure 6B) and this abundance was even observed at the immature (CD24^hiCD44^lo) stage (Figure 6D), indicating a general involvement of KLF10 in the IL-17-producing capacity of γδ²⁷⁻ cells before thymic maturation. However, only Vγ4⁺ γδ²⁷⁻-17 cells were significantly more abundant in the KO neonatal thymus at both immature and mature stages (Figure 6E), confirming the preferential restraint of KLF10 on Vγ4⁺ γδ²⁷⁻-17 cell development.

Consistent with the difference in CD5 expression on γδ subsets in the periphery (Figures 5C and 7A; Figure S5 in Supplementary Material), CD5 expression on each γδ thymic subset was unique and distinct based on the expression of Vγ chains and CD27 (Figure 7B). These results might suggest that different strength of TCR signal is required for thymic development of γδ subsets (11, 26). Regardless of CD27, the intensity of CD5 expression on immature Vγ4⁺ subsets was at a low level and distinct from that of immature Vγ1⁺ or Vγ1⁻Vγ4⁻ subsets, suggesting a requirement for weaker TCR signaling for thymic emergence of Vγ4⁺ cells (Figure 7B). We noted that immature Vγ4⁺ γδ²⁷⁻ thymocytes expressed the lowest level of CD5, which was similarly maintained in the periphery (Figure 7A).

We observed that thymic maturation induced a general decrease in CD5 expression on γδ T cells (Figure 7B). However, the CV of CD5 dramatically decreased in only Vγ4⁺ thymocytes after maturation (Figure 7C), suggesting their restricted spectrum of TCR-signal mode. To determine the basal activity of TCR signaling, we directly assessed intracellular levels of pZap70 in each γδ subset (Figure 7D). Interestingly, in contrast to the Vγ chain-specific clustering of CD5 MFI (Figure 7B), the level of pZap70 in immature γδ subsets clustered according to the expression of CD27 (Figure 7D), suggesting that each immature Vγ subset possessed basal TCR signaling that was influenced by CD27 costimulation (14, 32). Nonetheless, the level of pZap70 in immature Vγ4⁺ γδ²⁷⁻ thymocytes was low, similar to that in immature Vγ1⁻Vγ4⁻ γδ²⁷⁻ thymocytes, and then slightly decreased in the Vγ4⁺ γδ²⁷⁻ subset and increased in the Vγ1⁻Vγ4⁻ γδ²⁷⁻ subset after maturation; thus, a Vγ4⁺ γδ²⁷⁻ subset could acquire a relatively lower activity of basal TCR signaling (Figure 7D). Together, these data emphasized a TCR-signaling modality unique to the emergence and maturational transition of a Vγ4⁺ γδ²⁷⁻ subset, endorsing a weak TCR-signaling requirement for innate-like γδ-17 differentiation (10, 36). It is important to note that Klf10 could be induced by both TCR-dependent and -independent signaling pathways (Figures 3F,H), which were differently engaged in γδ development (24). On the other hand, CD127 (IL-7Rα) was abruptly induced in Vγ4⁺ or Vγ1⁻Vγ4⁻ γδ²⁷⁻ subsets after thymic maturation (Figure 7B), consistent with the notion of mature stage-specific acquisition of cytokine receptor-mediated regulation of γδ effector differentiation (25). Therefore, a series of weak TCR-signaling engagements with subsequent initiation of cytokine signaling seemingly cooperate for the function of KLF10 specific to Vγ4⁺ γδ²⁷⁻ thymic development by primarily fine-tuning KLF10 transcription.

Transcriptional profiling of thymic Vγ subsets revealed that immature Vγ4⁺ subsets are distinct from other immature Vγ subsets (Vγ1⁺, Vγ1.1⁺Vδ6.3⁺, and Vγ5⁺) but, interestingly, closely similar to CD4⁺CD8⁺ double-positive (DP) CD69⁺ cells of the αβ lineage, based on low expression of genes involved in metabolism and energy production (25). Of note, a gene constellation browser publicly provided by the Immunological Genome Project (ImmGen; www.immgen.org) reported that Klf10 in γδ T cells was closely correlated with genes encoding metabolic molecules, which showed lower expression in the immature Vγ4⁺ subset than in other Vγ subsets (data not shown) (25, 37). When we assessed the expression of CD5 on DP cells, KO DP cells transiently displayed relatively higher CD5 expression than their WT counterparts at post-positive selection, but not at the fully matured (TCRβ^hiCD69^–) stage (Figure 7E), accompanied by a delayed CD4 lineage choice (Figure 7F). However, maturation of CD4 and CD8 T cells was normal in both strains, as measured by the percentage of CD24^loQa-2^hiCD62L⁺CD69⁻ cells among TCRβ^hi CD4⁺ or CD8⁺ cells (Figure 7G). Thus, these data suggested that the putative mechanism involved in mature stage-specific alteration of surface CD5 on the Vγ4⁺ γδ²⁷⁻ thymic subset by KLF10 deficiency (Figures 5F,G) might be related to a metabolic process that is common to both Vγ4⁺ subsets and DP CD69⁺ cells.

Finally, we examined whether KLF10 extrinsically or intrinsically controlled homeostasis of Vγ4⁺ γδ²⁷⁻-17 cells. We reconstituted irradiated CD45.2⁺ WT or KO mice with congenic WT BM cells and analyzed CD45.1⁺ γδ²⁷⁻ cells after at least 12 wks (Figure S6B in Supplementary Material). Reconstitution of Vγ4⁺ γδ²⁷⁻ cells and CD5^loCD127^hi γδ²⁷⁻ cells in KO recipients was achieved at a level comparable to that in WT recipients, excluding the hematopoietic system-extrinsic effect of KLF10 on the homeostasis of Vγ4⁺ γδ²⁷⁻ 17 cells (Figure 8A). By contrast, mixed BM chimera experiments in which a 1:1 mixture of CD45.1⁺ WT and CD45.2⁺ KO BM cells was injected into lethally irradiated CD45.1/2⁺ WT mice (Figure S8C in Supplementary Material) showed a higher proportion of KO BM-derived cells among total CD3ε⁺ cells (data not shown). This suggested that KLF10-deficient BM cells outcompeted their WT counterparts during reconstitution of hematopoietic-derived cells, with the fact that KLF10 transcripts are at high level in long- and short-term repopulating hematopoietic stem cells (www.immgen.org).

We next investigate whether γδ cell-intrinsic KLF10 role is responsible for the restraint of Vγ4⁺ γδ²⁷⁻ cells. In the mixed BM chimera setting, a proportion of Vγ4⁺ γδ²⁷⁻ cells in KO BM-derived γδ T cells was relatively higher than that of their WT counterpart (Figure 8B). Although the total γδ T cells from both WT and KO BM cells contained γδ²⁷⁻ cells similarly, the KO BM-derived Vγ4⁺ cells generated γδ²⁷⁻ cells more than their WT counterpart did (Figure 8B); as expected, there were comparable proportions of γδ²⁷⁻ cells in Vγ4⁻ cells and of Vγ4⁺ cells in γδ²⁷⁺ cells between the both origins (Figure S8D in Supplementary Material). Indeed, KO BM-derived γδ²⁷⁻ cells contained higher proportions of Vγ4⁺ cells and CD5^loCD127^hi cells than their WT BM-derived counterparts (Figure 8C). Of note, KO BM-derived Vγ4⁺ γδ²⁷⁻ cells had greater frequencies of IL-17⁺ cells (Figure 8C). We also found a higher level of Klf10 expression in Vγ4⁺ γδ²⁷⁻ cells compared to the other γδ subsets (Figure 8D), possibly supporting the preferential engagement of KLF10 for the homeostasis of Vγ4⁺ γδ²⁷⁻ cells. Collectively, these results suggest that KLF10 serves as an intrinsic negative regulator to constrain Vγ4⁺ γδ²⁷⁻-17 cells and their production of IL-17.