SRC-3 Functions as a Coactivator of T-bet by Regulating the Maturation and Antitumor Activity of Natural Killer Cells

Natural killer (NK) cells, innate lymphocytes with cytotoxic functions, are indispensable for early immunosurveillance of tumors and elimination of infected cells (1, 2). NK cells play an important role in the regulation of the adaptive immune response by secreting cytokines (3). Unraveling the antitumor properties of NK cells is crucial (4, 5), thus a deep understanding of the mechanisms underlying NK-cell differentiation, maturation, and function may advance the development of NK-cell–based immunotherapy.

NK cells are derived from hematopoietic stem cells (6). After development in the bone marrow (BM), NK cells traffic to peripheral organs (7). The commitment of NK-cell–specific differentiation is characterized by the gradual acquisition of CD122, NK1.1, and DX5 expression (8). As NK cells fully mature, CD27 expression gradually decreases and CD11b expression gradually increases (9). NK-cell development, maturation, and function are controlled by a series of intrinsic and extrinsic factors, but these mechanisms are not fully understood (10). T-box transcription factor, T-bet, plays a critical role in this transcriptional regulatory network (11, 12) with many factors, such as Ets1, Foxo1, Tox2, Gata3, and mTOR, modulating NK-cell biology by at least partially affecting T-bet (13–17). However, little is known about how T-bet regulates downstream transcription factors or effector molecules and whether it needs interacting partners in NK cells.

Steroid receptor coactivator 3 (SRC-3), also called NCOA3/ACTR/AIB1/pCIP, is a coactivator of nuclear receptors (18). Similar to other p160 family members, SRC-3 can be recruited by nuclear receptors to modulate the transcription of target genes (19). SRC-3 can regulate the activity of several nonnuclear receptor proteins, including p53, AP-1, and E2F1, indicating a nuclear receptor–independent function (18). SRC-3 has attracted considerable attention due to its oncogenic property, whereas its physiologic function is largely overlooked (20). SRC-3 is involved in many important physiologic processes, such as somatic growth, hematopoiesis, immunoregulation, and energy metabolism (21–23).

In this study, we evaluated the role of SRC-3 in NK cells. On the basis of the observations that SRC-3 displayed increased nuclear translocation during NK-cell terminal differentiation and upon inflammatory cytokine stimulation, we generated mice with hematopoietic and NK-specific SRC-3 deletion. SRC-3 was required to maintain the maturation and effector function of NK cells via modulation of several critical T-bet–dependent genes. Collectively, our study provides new insight into the regulatory mechanism of NK-cell maturation and antitumor activity via SRC-3.

Normal wild-type (WT) C57BL/6J mice were purchased from the Institute of Zoology (Chinese Academy of Sciences, Beijing, China). SRC-3^flox/+ (SRC-3^fl/+) mice were generated at the Shanghai Model Organisms Center. Ncr1-Cre mice were obtained from Beijing Biocytogen Co, Ltd (24). Vav1-Cre and T-bet^−/− mice were purchased from The Jackson Laboratory. SRC-3^fl/fl/Ncr1-Cre (or Vav1-Cre) mice were generated by crossing SRC-3^fl/fl mice with Ncr1-Cre or Vav1-Cre mice. All mice used in the experiment were 6–8 weeks old. All animal experiments were approved by the Animal Care Committee of The Third Military Medical University (Chongqing, China).

NK-92, Yac-1, and B16F10 cells were purchased from Bena Culture Collection (December 2017). NK-92 cells were grown in complete medium containing 75% DMEM Alpha (Gibco; supplemented with 0.2 mmol/L inositol, 0.1 mmol/L 2-mercaptoethanol, 0.02 mmol/L folic acid, and 100–200 U/mL recombinant IL2), 12.5% horse serum (Gibco), and 12.5% FBS (Gibco). Yac-1 cells and B16F10 cells were cultured in 90% RPMI (HyClone) containing 10% FBS. All cell lines were passaged up to 10 times, authenticated by flow cytometry (last authentication, November 2019), and tested for Mycoplasma contamination using a MycoFluor Mycoplasma Detection Kit (Invitrogen; last test, November 2019).

Single-cell suspensions from the BM, spleen, peripheral lymph nodes (pLN), lungs, and liver of mice were obtained as described previously (14, 22). The following flow cytometric antibodies were used: anti-CD3e (145-2C11, #100353, 100321, dilution 1:100), anti-Gr-1 (RB6-8C5, #108417, dilution 1:100), anti-Ter119 (TER-119, #116215, dilution 1:100), anti-B220 (RA3-6B2, #103225, dilution 1:100), anti-CD19 (6D5, #115546, 115521, dilution 1:100), anti-NK1.1 (PK136, #108718, 108732, dilution 1:100), anti-Nkp46 (29A1.4, #137608, dilution 1:100), anti-CD146 (ME-9F1, #134704, dilution 1:100), and anti-CD107a (1D4B, #1216124, dilution 1:50) from BioLegend; anti-c-Kit (2B8, #14-1171-82, dilution 1:100), anti-CD244 (eBio244F4, #25-2441-82, dilution 1:100), anti-CD127 (A7R34, #45-1271-82, dilution 1:100), anti-CD135 (A2F10, #17-1351-82, dilution 1:50), anti-CD122 (TM-beta1, #48-1222-82, dilution 1:100), anti-CD27 (LG.7F9, #12-0271-82, dilution 1:100), anti-CD11b (M1/70, #11-0112-82, 45-0112-82, dilution 1:200), anti-KLRG1 (2F1, #12-5893-8, dilution 1:100), anti-Granzyme B (NGZB, #12-8898-82, dilution 1:50), anti-Perforin (eBioOMAK-D, #12-9392-82, dilution 1:50), anti-T-bet (4B10, #12-5825-82, dilution 1:50), anti-Ki67 (SolA15, #25-5698-82, dilution 1:200), anti-p-STAT3^Tyr705 (LUVNKLA, #12-9033-42, dilution 1:50), anti-p-STAT5^Tyr694 (SRBCZX, #12-9010-42, 1:50), and anti-p-STAT4^Tyr693 (4LURPIE, #17-9044-42, 1:50) from eBioscience; anti-CD43 (S7, #561857, dilution 1:200) and anti-IFN-γ (XMG1.2, #562019, dilution 1:300) from BD Biosciences; anti-S1P5 (#ab214464, dilution 1:50) from Abcam; anti-SRC-3 (F-2, #sc-5305 PE, dilution 1:50) and anti-Zeb2 (E-11, #sc-271984 PE, dilution 1:50) from Santa Cruz Biotechnology; and anti-Blimp1 (C14A4, #9115, dilution 1:50) from Cell Signaling Technology.

Ki67, Granzyme B, Perforin, T-bet, Zeb2, Blimp1, and SRC-3 protein staining was performed using a Foxp3/Transcription Factor Staining Buffer Set (#00-5523-00, eBioscience) according to the manufacturer's instructions. p-STAT3^Tyr705, p-STAT5^Tyr694, and p-STAT4^Tyr693 staining was performed using IC Fixation Buffer (#00-8222-49, eBioscience) according to the manufacturer's instructions. In vivo bromodeoxyuridine (BrdU) incorporation was analyzed using a FITC-BrdU Flow Kit (#559619, BD Pharmingen) according to the manufacturer's instructions. Cellular apoptosis was analyzed with an APC-Annexin V Apoptosis Detection Kit with 7-AAD (#640930, BioLegend) according to the manufacturer's instructions. IFNγ production after stimulation was analyzed using eBioscience Cell Stimulation Cocktail (plus protein transport inhibitors, #00-4975-93) according to the manufacturer's instructions. CD107a expression in NK cells after stimulation with Yac-1 cells was analyzed as we described previously (17).

Flow cytometric detection was conducted using a FACSverse Cytometer (BD Biosciences), and all data were analyzed using FlowJo10.0 Software (TreeStar). FACS was performed using a FACSAriaIII Sorter (BD Biosciences). The detailed gating strategies are shown in Supplementary Fig. S1.

This assay was conducted as described previously (25). In brief, spleens of SRC-3^fl/fl or SRC-3^NKΔ/Δ mice were harvested and grinded gently in RPMI Medium (#SH30809.01, HyClone) using the even end of the syringe, and a single-cell suspension was obtained following red blood cells lysis using a Red Blood Cell Lysis Buffer (#20120, Stem Cell Technologies Inc.) and filtration through a 70-mm filter. Next, spleen cells were stained with NK-cell surface markers (CD3, CD19, NK1.1, and Nkp46) by placing on ice for 20 minutes, and the dilution of these antibodies has been described above. After washing in ice-cold PBS, NK cells (CD3⁻ CD19⁻ NK1.1⁺ NKp46⁺) were sorted by flow cytometry. The detailed gating strategies are shown in Supplementary Fig. S1. Then, freshly sorted cells were cocultured with Yac-1 cells labeled with CFSE (eBioscience) at 37°C at an effector:target (E:T) ratio of 1:1, 5:1, or 10:1. Six hours later, the percentage of TOPRO3⁺ Yac-1 cells was assessed by flow cytometry.

Mice were injected with 1 μg of PE-conjugated anti-CD45 antibody (30F11, #103106, BioLegend, dilution 1:40) via the tail vein (injection volume: 200 μl). Two minutes later, the mice were sacrificed. The BM was collected and stained with NK-cell markers, followed by flow cytometric analysis (26, 27).

SRC-3^fl/fl or SRC-3^NKΔ/Δ mice were injected with 2 × 10⁵ B16F10 melanoma cells via the caudal vein (injection volume: 200 μl). Twelve days later, the mice were sacrificed, and the tumor nodules in the lungs were counted under a dissecting microscope.

Total RNA was extracted from FACS-sorted cells using an RNAqueous Kit (#AM1931, Ambion) according to the manufacturer's instructions. RNA content was measured by NanoDrop 2000 (Thermo Fisher Scientific). Then, 1 μg of total RNA was used to synthesize cDNA, and quantitative RT-PCR (qRT-PCR) was performed as reported previously (22). The relative expression of each gene was normalized to Gapdh and determined by 2^−ΔΔC_t methods. Each reaction was performed in triplicate. Primer sequences are provided in Supplementary Table S1.

NK cells were FACS sorted from the spleen of SRC-3^fl/fl and SRC-3^NKΔ/Δ mice according to the above method, and total RNA was then extracted as described above. Library construction and RNA sequencing (RNA-seq) were conducted at OE Biotech. Co., Ltd. Gene expression was normalized as fragments per kilobase of exon per million fragments mapped values. The combined criteria of a log₂ (fold change) > 0.5 and a P_adj < 0.05 were used to identify differentially expressed genes. The raw RNA-seq data were deposited in the NCBI Sequence Read Archive database (no. PRJNA588159).

This assay was performed using an EZ-ChIP Kit (#17-371RF, Millipore), as described previously (28, 29). In brief, NK cells were freshly sorted from mouse spleen as described above. These cells were fixed with 1% formaldehyde (Sigma) and were then lysed in SDS lysis buffer containing 5 μL of protease inhibitors [included in the chromatin immunoprecipitation (ChIP) kit as mentioned above]. Subsequently, chromatin was sheared to 200–1,000 bp by sonication, and cross-linked protein/DNA was incubated with an SRC-3 (#ab2831, Abcam, dilution 3 μg/10⁶ cells) antibody or IgG control (#ab171870, Abcam, dilution 3 μg/10⁶ cells) overnight at 4°C with rotation. The immunoprecipitated protein/DNA complexes were eluted by elution buffer (included in the ChIP kit as mentioned above) according to the manufacturer's instructions and were reversed to free DNA by incubating at 65°C for 5 hours. Finally, DNA was purified using spin columns and detected by qRT-PCR as described above. Primer sequences are provided in Supplementary Table S2.

Nuclear proteins were isolated using NE-PER Nuclear and Cytoplasmic Extraction Reagents (#78835, Invitrogen) according to the manufacturer's instructions. Then, the nuclear or total protein expression was measured by Western blot analysis as described previously (28). The following antibodies were used: anti-SRC-3 (#ab2831, Abcam, dilution 1:1,000), anti-T-bet (#ab53174, Abcam, dilution 1:1,000), anti-Lamin B1 (#ab16048, Abcam, dilution 1:5,000), and anti-β-actin (#4970, Cell Signaling Technology, dilution 1:1,000).

This experiment was performed using a Dynabeads Protein G Immunoprecipitation Kit (#10007D, Invitrogen) following the manufacturer's instructions. Briefly, mouse primary NK cells (freshly sorted from spleen as described above) or human NK-92 cells were first lysed. The binding of magnetic beads to an SRC-3 antibody (#2126, Cell Signaling Technology, dilution 1:50) or T-bet antibody (#ab53174, Abcam, dilution 1:50) was carried out, and then the magnetic bead–antibody complexes were added to the lysed samples. Subsequently, samples were resuspended and incubated on a shaker (HY-450, Huicheng Biological Technology Co. LTD) at the speed of 75 revolutions per minute for 10 minutes at room temperature to allow the antigen to bind to the magnetic bead–antibody complexes. After washing with washing buffer (included in the immunoprecipitation kit as mentioned above), the target antigen was eluted by elution buffer (included in the immunoprecipitation kit as mentioned above) according to the manufacturer's instructions and subjected to Western blot analysis as mentioned above. To avoid interference from the heavy chain (55 kDa), a secondary antibody against IgG light chain (IPKine HRP, Mouse Anti-Rabbit IgG LCS, #A25022, Abbkine Inc.) was used.

NK-cell subsets freshly sorted from murine spleen were placed on poly-l-lysine–coated slides. After fixation via 4% paraformaldehyde for 15 minutes at room temperature, permeabilization, and blocking, cells were stained with an SRC-3 (#ab2831, Abcam, dilution 1:200) antibody. Then, samples were incubated with a FITC-conjugated secondary antibody (#F-2765, Invitrogen, dilution 1:200) and DAPI (Sigma) and were imaged using a Zeiss LSM800 Confocal Microscope (Carl Zeiss).

Experimental data were analyzed using GraphPad Prism 6.0 software. Two-tailed Student t test and one-way ANOVA were used to compare the differences between two groups and multiple groups, respectively. Data are shown as mean ± SD. P < 0.05 was considered statistically significant.

We first observed that SRC-3 was expressed in NK cells in mouse spleens (Supplementary Fig. S2A), consistent with the data from the BioGPS dataset (Supplementary Fig. S2B). However, neither the mRNA nor the protein expression of SRC-3 changed during NK-cell terminal differentiation (Fig. 1A and B). We observed a gradual increase in nuclear SRC-3 expression in NK cells during the CD27⁺ CD11b⁺ to CD27⁻ CD11b⁺ transition (Fig. 1B). The evident translocation of SRC-3 from the cytoplasm to the nucleus was further confirmed by immunofluorescence staining (Fig. 1C), indicating a potential role of SRC-3 in this process. SRC-3 is increased in response to multiple hormones, growth factors, or inflammatory factors (30, 31). Considering that several inflammatory cytokines, such as IL2, IL12, and IL15 (6, 8), play important roles in regulating NK-cell development and/or function, we tested whether these inflammatory cytokines could also affect the subcellular localization of SRC-3 in NK cells. IL2, IL12, or IL15 stimulation promoted SRC-3 nuclear translocation in a time-dependent manner, but did not affect total SRC-3 expression in mouse NK cells (Fig. 1D–F). Similar results were obtained from human NK-92 cells (Supplementary Fig. S2C–S2E). These findings suggested that SRC-3 may be involved in NK-cell maturation and/or activation.

To assess whether SRC-3 regulated NK-cell lineage development, we established a mouse model with hematopoietic deletion of SRC-3 (SRC-3^fl/fl/Vav1-Cre⁺ mice and the littermate SRC-3^fl/fl/Vav1-Cre⁻ mice were referred to as SRC-3^Δ/Δ and control mice, respectively) by crossing SRC-3^fl/fl mice with Vav1-Cre mice (Supplementary Fig. S3A–S3C). Initial flow cytometric analysis revealed that hematopoietic SRC-3 deficiency did not affect common lymphoid progenitors and NK progenitors (NKP), including pre-NKPs and rNKPs, in mouse BM (Supplementary Fig. S3D and S3E). Hematopoietic deletion of SRC-3 did not significantly alter the percentage and number of Lin⁻ CD122⁺ cells or the proportions of NK1.1⁻ Nkp46⁻, NK1.1⁺ Nkp46⁻, or NK1.1⁺ Nkp46⁺ cells in Lin⁻ CD122⁺ compartments (Fig. 2A). These results suggest that SRC-3 was dispensable for the early stages of NK-cell development. The percentage and number of NK cells were modestly increased in the BM, but significantly decreased in spleen, pLNs, lungs, and liver of SRC-3^Δ/Δ mice (Fig. 2B; Supplementary Fig. S3F). To determine the exact role of SRC-3 in NK cells, we used NK-specific SRC-3–knockout mice (SRC-3^fl/fl/Ncr1-Cre⁺ mice and the littermate SRC-3^fl/fl/Ncr1-Cre⁻ mice are referred to as SRC-3^NKΔ/Δ and SRC-3^fl/fl mice, respectively; Supplementary Fig. S4A and S4B). Indeed, similar results were observed in SRC-3^NKΔ/Δ mice (Fig. 2C; Supplementary Fig. S4C).

We then evaluated whether the decrease in NK cells in the peripheral organs of these mice was because of their blocked egress from the BM. As anticipated, the expression of S1pr5, a key gene that controls NK-cell egress from the BM (27), was markedly reduced in NK cells from SRC-3^NKΔ/Δ mice (Fig. 2D and E). Consistent with this finding, a lower percentage of sinusoidal NK cells was detected in the BM of SRC-3^NKΔ/Δ mice (Fig. 2F). These data illustrated that SRC-3 was needed for the maintenance of a normal NK-cell distribution in mice.

We next assessed whether SRC-3 deficiency affected NK-cell maturation. The results of flow cytometric analyses showed that in the BM, spleen, and pLNs of SRC-3^NKΔ/Δ mice, the proportion of most mature CD27⁻ CD11b⁺ NK cells was decreased, whereas the proportion of CD27⁺ CD11b⁻ and CD27⁺ CD11b⁺ NK cells were increased (Fig. 3A). In SRC-3^NKΔ/Δ mice, these organs exhibited a strongly reduced ratio of CD27⁻/CD27⁺ cells in the CD11b⁺ NK-cell population (Fig. 3B). Terminal NK-cell maturation is associated with the upregulation of CD43 and KLRG1 (17). Specifically, the percentage of CD43⁺ KLRG1⁺ cells was evidently decreased in NK cells in the absence of SRC-3 (Fig. 3C). CD146, another marker of NK-cell maturation (32), was significantly downregulated, and c-Kit (33), a marker of immature NK cells, was markedly upregulated in NK cells when SRC-3 was deleted (Fig. 3D and E). Similar maturational defects were observed in hematopoietic-specific SRC-3–null mice (Supplementary Fig. S5A and S5B).

Immature NK cells have a higher basal proliferation rate than mature NK cells. As expected, NK cells from both the BM and spleen of SRC-3^NKΔ/Δ mice exhibited a significant increase in proliferation as assessed by Ki67 staining and BrdU assays (Supplementary Fig. S6A and S6B). The Annexin V/7-AAD staining results showed that the viability of freshly isolated NK cells from SRC-3^NKΔ/Δ mice was decreased (Supplementary Fig. S6C). The accelerated apoptosis may, to some extent, have limited the number of NK cells in the BM and peripheral organs. However, these observations were not due to an impairment in STATs signaling induced by SRC-3 deficiency (Supplementary Fig. S7). Collectively, these results indicated a role of SCR-3 in facilitating NK-cell maturation and survival.

Our finding that loss of SRC-3 compromised NK-cell maturation prompted us to determine whether SRC-3 was also involved in modulating NK-cell function. We then found that IFNγ production was reduced on a “per-cell” basis as determined by the decrease in the mean fluorescence intensity (MFI) of IFNγ⁺ cells in SRC-3–deficient NK cells after IL12 + IL18 or PMA + ionomycin stimulation, although the percentage of IFNγ⁺ cells was unchanged (Fig. 4A). However, the expression of CD107a, a marker of NK-cell degranulation (34), was substantially reduced in all NK-cell subsets isolated from SRC-3^NKΔ/Δ mice after stimulation with Yac-1 cells (Fig. 4B). In addition, depletion of SRC-3 in NK cells resulted in reduced mRNA and protein production of two critical cytotoxic molecules, granzyme B and perforin (ref. 3; Fig. 4C–E).

Given that SRC-3 functions mainly in the nucleus, we speculated that the demand for nuclear SRC-3 in NK cells may increase when performing their antitumor function. As anticipated, mouse splenic NK cells challenged with Yac-1 cells or B16F10 melanoma cells displayed increased SRC-3 nuclear translocation, although total SRC-3 expression was unchanged (Fig. 4F and G). Consistent with these results, NK cells sorted from SRC-3^NKΔ/Δ mice displayed a compromised ability to specifically lyse target Yac-1 cells in vitro (Fig. 4H). To further verify whether SRC-3 deficiency impaired the tumor surveillance function of NK cells in vivo, we injected an equal number of B16F10 melanoma cells into SRC-3^fl/fl and SRC-3^NKΔ/Δ mice via the caudal vein. As shown in Fig. 4I, compared with their littermate controls, SRC-3^NKΔ/Δ mice exhibited significantly increased metastasis of melanoma cells to the lung. Hence, these findings provide strong evidence that SRC-3 was required to maintain the normal effector function of NK cells.

To further elucidate the underlying mechanism by which SRC-3 regulated NK-cell maturation and function, we performed RNA-seq analysis of NK cells sorted from the spleen of SRC-3^fl/fl and SRC-3^NKΔ/Δ mice (Fig. 5A). By using the criteria [log₂ (fold change) > 0.5, P_adj < 0.05], we identified a total of 413 differentially expressed genes, 260 of which were upregulated and 153 of which were downregulated in NK cells after SRC-3 ablation (Fig. 5B; Supplementary Table S3). Consistent with our above results, the differential expression of several genes (such as S1pr5, Klrg1, Gzmb, Prf1, Kit, etc.) was also observed in these data (Fig. 5B and C). As noted above, the defects observed in SRC-3^NKΔ/Δ mice were reminiscent of those displayed by T-bet^−/− mice (12, 14, 26). However, neither the mRNA nor protein expression of T-bet was changed in NK cells with SRC-3 deficiency (Fig. 5D and E). Two critical T-bet target genes, Zeb2 and Prdm1, which control NK-cell maturation (26, 35), were significantly downregulated in SRC-3–deficient NK cells (Fig. 5C, D, F, and G). The mRNA expression of other NK-associated transcription factors, including Ets1, Id2, Tox, Irf2, Elf4, Nfil3, Gata3, Eomes, Stat5a, Stat5b, Smad4, and Foxo1 (13, 16, 25, 36–41), were similar in SRC-3^fl/fl and SRC-3^NKΔ/Δ NK cells (Fig. 5D). SRC-3 is a coactivator that cannot directly bind to DNA, but regulates the transcriptional activity of nuclear receptors and several nonnuclear receptor proteins (18). Combined with these findings, we reasoned that SRC-3 may be required for T-bet–dependent regulation of the expression of several genes in NK cells, which was in-line with the above findings.

To confirm whether SRC-3 was a coactivator of T-bet in NK cells, we performed a coimmunoprecipitation (Co-IP) assay. Specifically, we observed a physical interaction between the SRC-3 and T-bet proteins in freshly sorted murine NK cells (Fig. 6A). Similar results were observed in human NK-92 cells (Fig. 6B). On the other hand, data from a published ChIP-sequencing (ChIP-seq) study revealed that Zeb2 and Prdm1 loci contain several T-bet–binding sites (ref. 42; Fig. 6C). The recruitment of SRC-3 to prominent T-bet–binding sites in the promoter regions of these genes was observed in normal but not SRC-3–null NK cells (Fig. 6D and E). In addition, studies have reported that S1pr5 is another target gene of T-bet. Indeed, SRC-3 was bound to a T-bet–binding site in the S1pr5 locus (−1.65 kb; Fig. 6C and F). However, these bindings were lost in T-bet^−/− NK cells (Fig. 6G), indicating that the functions of SRC-3 were dependent on T-bet. Finally, we wondered whether there was a regulatory feedback loop between T-bet and SRC-3 in NK cells. Actually, knockout of T-bet did not significantly affect SRC-3 expression in NK cells (Fig. 6H and I). Collectively, our data demonstrated that SRC-3 could coactivate T-bet to regulate the expression of several target genes, which was needed for NK-cell maturation (Fig. 6J).

NK cells are innate immune cells with immunosurveillance and immunoregulatory functions. The development and maturation of NK cells are harmoniously controlled by a complex transcriptional regulatory network (6, 10). Although many critical regulatory factors have been identified, the underlying molecular basis is not completely understood. Here, we showed for the first time that SRC-3 act as a coactivator of T-bet to regulate the maturation and function of NK cells, therefore adding a novel mechanism to T-bet–mediated NK-cell biology.

As a member of the p160 protein family, SRC-3 was widely expressed and involved in many biological processes (18). Here, we observed that SRC-3 was abundant in NK cells in mice. However, using SRC-3^fl/fl/Vav1-Cre (hematopoietic specific) mice, we found that SRC-3 deficiency did not significantly affect the early development of NK cells, as there was normal expression of transcription factors associated with early development of NK cells, such as Nfil3. Targeted ablation of SRC-3 altered the normal distribution of NK cells due to the downregulation of S1pr5. Knockout of T-bet or its target gene, S1pr5, blocks the egress of NK cells from the BM and pLNs (12, 27). However, mice with conditional deletion of SRC-3 displayed reduced NK-cell numbers in pLNs, possibly because of alterations in other NK-cell trafficking–associated molecules, similar to observations in Zeb2^−/− mice (26).

Many transcription factors related to NK-cell maturation, such as T-bet, Zeb2, Aiolos, and Smad4, are gradually upregulated during NK-cell terminal differentiation (12, 26, 41, 43). Although SRC-3 expression was unchanged, its translocation from the cytoplasm to the nucleus is increased during this process, revealing an increased demand for SRC-3 upon NK-cell maturation. As a result, loss of SRC-3 significantly impaired the maturation of NK cells. SRC-3 is present mainly in the cytoplasm under steady-state conditions, and its subcellular distribution may be altered after stimulation (30, 44). SRC-3 displayed increased nuclear translocation upon stimulation with several proinflammatory cytokines. These data indicated that the nuclear translocation of SRC-3 may be a key checkpoint for NK-cell activation. In addition, although IL2, IL12, or IL15 can appreciably induce SRC-3 nuclear translocation, SRC-3 deficiency did not significantly affect the phosphorylation of STAT3, STAT4, or STAT5 in NK cells after stimulation with these cytokines. One reason may be that SRC-3 is downstream of the STATs signaling (16, 36).

SRC-3 promotes cancer cell progression by activating multiple cell proliferation–associated pathways or directly regulating cell-cycle proteins (45). In contrast, our data showed that SRC-3 inhibits the proliferation of NK cells, suggesting that SRC-3 plays a unique role in NK cells. We found a significant decrease in the expression of Prdm1, which has a role in inhibiting NK-cell proliferation, which may also account for the finding observed in SRC-3^NKΔ/Δ mice. However, the enhanced proliferation did not lead to a significant increase in NK-cell number, possibly due to the simultaneous increase in apoptosis induced by SRC-3 deficiency.

NK cells play a crucial role in early immunosurveillance of tumors. Here, we found that CD107a expression was decreased in all subsets of SRC-3–deficient NK cells after Yac-1 cells' stimulation, indicating a reduced “per-cell” cytotoxic potential of SRC-3–null NK cells against Yac-1 target cells. Consistent with this finding, deletion of SRC-3 led to significant decrease in the expression of granzyme B and perforin, critical cytotoxic molecules that kill target cells, in all NK-cell subsets. These findings indicated that SRC-3 may have participated in regulating NK-cell effector function. However, we did not perform direct in vitro cytotoxicity of each NK-cell subset between SRC-3^fl/fl or SRC-3^NKΔ/Δ mice. These assays may have provided evidence supporting our conclusion. It will also be worthwhile to knockout SRC-3 using in vitro–expressed TAT–Cre recombinant protein in NK cells derived from SRC-3^fl/fl mice and analyze the effector functions.

T-box transcription factor, T-bet is a master transcriptional regulatory factor in NK cells (11). Since the role of T-bet in NK cells was first reported by Townsend and colleagues, it has received great attention (12). However, a clear understanding of the molecular mechanisms by which T-bet works is still lacking. SRC/p160 coactivator SRC-2 can regulate the expression of pro-opiomelanocortin (POMC) mediated by another T-box transcription factor, Tpit after hormonal stimulation (46). T-bet can recruit the transcriptional cofactor Bhlhe40 to modulate IFNγ generation in invariant NK T cells (47). Hence, the T-box family of transcription factors may need transcriptional regulators to synergistically control target gene expression in a cell context–specific manner. In this study, we showed that the phenotype of SRC-3–null NK cells resembled that observed in T-bet–knockout mice. We identified an interaction between the SRC-3 and T-bet proteins under physiologic conditions. SRC-3 can be recruited to the promoter regions of Zeb2, Prdm1, and S1pr5, all of which are T-bet target genes (26, 27, 35). However, these effects were abrogated in NK cells when T-bet was knocked out, indicating that SRC-3 acted in a manner dependent on T-bet. Taken together, these results indicated that SRC-3 may be a critical component of the T-bet–dependent transcriptional regulatory complex in NK cells, although we could not exclude other T-bet–independent mechanisms that will need further investigation.

Prdm1 and Zeb2 partially mediate the role of T-bet in terminal NK-cell maturation (26, 35). NK cells from Prdm1^−/− and Zeb2^−/− mice have some maturational defects similar to those from T-bet^−/− mice, which are also observed, albeit to a lesser extent, in SRC-3^NKΔ/Δ mice. The expression of both transcription factors were markedly decreased in SRC-3–deficient NK cells. Hence, we believe that SRC-3 promoted NK-cell maturation at least partially by regulating the T-bet–dependent expression of Zeb2 and Prdm1. T-bet regulates the expression of several target genes in NK cells and T cells in cooperation with Zeb2 (48). Therefore, decreased expression of Zeb2 may further compromise the transcriptional activity of T-bet in SRC-3–deficient NK cells. Indeed, the expression of many target genes coinduced (such as Klrg1, Gzma, Gzmb, Mcam, and Cx3cr1) or corepressed (such as Socx3 and Dusp6) by Zeb2 and T-bet were significantly changed in SRC-3–null NK cells.

In conclusion, our findings demonstrated that SRC-3 was involved in the regulation of the maturation and antitumor function of NK cells at least, in part, by coactivating T-bet. These data not only reveal a new molecular mechanism related to T-bet in NK cells, but also provide a potential target to increase the tumor surveillance and viral elimination functions of NK cells upon SRC-3 nuclear translocation.

No potential conflicts of interest were disclosed.

M. Hu: Conceptualization, formal analysis, validation, investigation, visualization, methodology, writing–original draft. Y. Lu: Formal analysis, validation, investigation, methodology. Y. Qi: Formal analysis, investigation, methodology. Z. Zhang: Formal analysis, investigation, methodology. S. Wang: Software, formal analysis, investigation, visualization. Y. Xu: Funding acquisition, investigation, methodology. F. Chen: Investigation, methodology. Y. Tang: Investigation, methodology. S. Chen: Investigation, methodology. M. Chen: Investigation, methodology. C. Du: Investigation, methodology. M. Shen: Investigation, methodology. F. Wang: Conceptualization. Y. Su: Conceptualization. Y. Deng: Conceptualization, data curation, supervision, writing–review and editing. J. Wang: Conceptualization, data curation, supervision, funding acquisition, validation, project administration, writing–review and editing.

The authors thank Yang Liu for technical support in flow cytometry and Liting Wang for technical assistance in immunofluorescence microscopy. This work was supported by grants from the National Natural Science Foundation of China (grant nos. 81725019, 81930090, 81573084, and 81500087) and the Scientific Research Project of PLA (AWS16J014).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.