KLF12 Regulates Mouse NK Cell Proliferation

Patients with genetic mutations resulting in diminished NK cell numbers or function succumb to recurrent herpesvirus and papillomavirus infections (1–4), highlighting the importance of NK cells in controlling certain viral infections. NK cells are cytotoxic lymphocytes that have the unique ability to recognize and lyse target cells without prior exposure. NK cells also secrete cytokines, such as IFN-γ, to activate other immune cells to coordinate appropriate immune responses against pathogens (5).

Mouse CMV (MCMV) infection is an ideal model to study NK cell activation, expansion, and effector function. At the onset of infection, IL-12 production by dendritic cells is critical for early NK cell production of IFN-γ and control of viral load (6–8). A subset of NK cells expressing the activating Ly49H⁺ receptor in C57BL/6 mice specifically recognizes the MCMV-encoded glycoprotein, m157 (9, 10). Ly49H⁺ NK cells expand, contract, and persist after MCMV infection (11). These cells conferred specific protection against MCMV rechallenge and not other heterologous infections, indicating that these are MCMV-specific memory NK cells (12, 13).

NK cells share expression of many genes with their lymphocyte counterparts; therefore, we sought to find genes preferentially expressed by NK cells in the hematopoietic cell lineage to understand their unique activation and cytotoxic capabilities. From the Immunological Genome (ImmGen) Consortium, we identified kruppel-like factor 12 (KLF12), a novel transcription factor, to be preferentially expressed in mouse NK cells. KLF12 is a zinc finger transcription factor in the Kruppel-like factor family. Similar to KLF3 and KLF8, KLF12 has a conserved PVDLS domain at the N terminus that binds to the corepressor, CtBP1 (14–16). Klf12 transcripts are found in the kidney, endometrial stromal cells, primary gastric tumors, and various cancer cell lines (15, 17–19). Prior studies have demonstrated that KLF12 binds to a conserved CACCC sequence and functions as a transcriptional repressor or activator, suggesting that the function of KLF12 is context and cell type specific (17, 20, 21). KLF12 target genes are largely unknown but include NR4A1 (Nur77), TRAP2A, FOXO1, SLC14A2, and EGR1 (17, 20, 22–25).

In this study, we assessed the role of KLF12 in mouse NK cells as a potential transcriptional regulator of NK cell development and/or effector functions. To address this, we generated a mouse with floxed Klf12 loci and crossed these the mice expressing β-actin Cre recombinase to delete KLF12 expression. We assessed the development, proliferation, and effector functions of KLF12-deficient NK cells in response to in vitro stimulation and MCMV infection.

Mice were obtained from the following sources: wild-type (WT) C57BL/6 and C57BL/6 CD45.1 mice were purchased from the National Cancer Institute (Frederick, MD), Rosa26-Flippase C57BL/6 mice from Dr. R. Locksley and β-actin Cre transgenic C57BL/6 mice from Dr. M. McManus, University of California San Francisco (UCSF), and Klra8^−/− (Ly49H-deficient) C57BL/6 mice from Dr. S. Vidal, McGill University. The Klf12 targeting vector was purchased from the International Mouse Knockout Consortium and electroporated into E14-129/Ola embryonic stem cells. Selected clones were then microinjected into C57BL/6 females, and heterozygotes were backcrossed at least nine generations onto the C57BL/6 background. Klf12-floxed mice were genotype by PCR using the following primer pairs: WT, forward, 5′-CACAGCGAGTTCCCCAAGAT-3′, reverse 1, 5′-GGACGCACATACAGCTTCCT-3′, reverse 2, 5′-AGGGAAAGGGTCGAGAGACA-3′; Flox, forward, 5′-CACAGCGAGTTCCCCAAGAT-3′, reverse 3, 5′-TAGGAGGTGTGGCTTTGCTG-3′, reverse 4, 5′-CCCTTTGTTGTCCGCCTACT-3′. Mice were bred and housed in a specific pathogen-free facility and experiments were performed according to UCSF Institutional Animal Care and Use Committee and National Institutes of Health guidelines.

Total RNA from splenocytes or indicated cellar populations was prepared using the Ambion RNAqueous kit and converted into cDNA. Quantitative PCR analysis with SYBR Green master mix (Roche) was performed on the cDNA following standard conditions with the following primer pairs: Klf12, forward, 5′-CTGGCGAACCACATAGGCCCAG-3′, reverse, 5′-CGGCGGCCTACATTTACGTGAT-3′, Btg3, forward, 5′-TGCTGCCGGTATGGAGAGAA-3′, reverse, 5′-GGTCACCTTATCCAGAGCCC-3′; Hprt, forward, 5′-CACAGGACTAGAACACCTGC-3′, reverse, 5′-GCTGGTGAAAAGGACCTCT-3′. Primer pairs to confirm disrupted Klf12: Klf12 exons 2-3, forward, 5′-GCTAATGCTTGATGGAATGCC-3′, reverse, 5′-AGTTGTGGACGTTTGGAGAC-3′; exons 5-6, forward, 5′-ACATCCATCCCCGGTATCCA-3′, reverse, 5′-TGGCGTCTTGTGCTCTCAAT-3′. Expressions were normalized to HPRT.

Genomic DNA (gDNA) from selected stem cell clones was processed using the Promega Wizard gDNA purification kit. gDNA was digested overnight with EcoRV, transferred onto a membrane, probed with α-³²P dATP against the 5′ arm of the Klf12 targeting vector, and exposed to film. Probes were amplified using the following primer pair: forward, 5′-TCTCCCTCTTGGTGGTCACT-3′, reverse, 5′-GATGCCTGAAAACCGCACAG-3′. The 3′ arm of the targeting vector was amplified by PCR using Takara PrimeSTAR GXL DNA polymerase with the following primers: forward, 5′-GGATCTCATGCTGGAGTTCTTCGCC-3′, reverse 1, 5′-CCAAAGCCCCTATACCCTTCCCCGC-3′, and reverse 2, 5′-ATCTGGCGTGGGCGGCCAGCAGTTC-3′.

Splenocytes were resuspended in RPMI 1640 supplemented with 10% FCS, 50 μM 2-ME, 100 μM nonessential amino acids, 1 mM sodium pyruvate, 50 mM HEPES, and 2 μM l-glutamine. One million cells were stimulated for 6 h in the presence of BD GolgiStop with 20 ng/ml recombinant mouse (rm) IL-12 (no. 419-ML/CF; R&D Systems) and 10 ng/ml rmIL-18 (no. B0025; R&D Systems), with 20 ng/ml PMA and 200 ng/ml ionomycin or with 10 μg/ml of plate-bound anti-NK1.1 Ab (clone PK136), were fixed and permeabilized with BD Cytofix/Cytoperm and then stained for IFN-γ and CD107a. For p-STAT5 (no. 612599; BD Biosciences) and p-STAT3 (no. 557814; BD Biosciences) staining, 1 million cells were stimulated for the indicated times and concentrations with recombinant human (rh) IL-15 (no. 247-ILB; R&D Systems) or rmIL-21 (ZymoGenetics), fixed with 1% paraformaldehyde in PBS, permeabilized with methanol, and stained. For proliferation assays, 1 million CellTrace Violet (no. C34557; Invitrogen)–labeled splenocytes were cultured in 10 ng/ml rhIL-15, 1 × 10⁵ paraformaldehyde-fixed RMA lymphoma cells, or MCMV m157–transduced RMA cells supplemented with 50 U/ml rhIL-2 (National Cancer Institute Preclinical Repository) or 100 ng/ml rmIL-21.

Donor bone marrow (BM) was isolated from femurs and tibias of 8- to 10-wk-old WT and Klf12^F/F C57BL/6 mice and mixed 1:1 in sterile PBS, and 5 × 10⁶ cells were injected i.v. into WT recipients lethally irradiated with 1200 Gy. Recipient mice were injected i.p. with 200 μg anti-CD4 (GK1.5), 200 μg anti-CD8 (2.43), and 100 μg anti-NK1.1 (PK136) depleting Abs at the time of transfer and maintained on antibiotic pellets (S0443; Bio-Serv) for 2 wk. Recipients reconstituted for at least 8 wk were used for experiments.

NK cells were enriched from spleens using mAbs against Ter119, Gr-1, CD4, CD8, CD19, and CD5. Qiagen anti-rat IgG magnetic beads were used to deplete the indicated populations. Enriched NK cells from WT or KLF12-deficient mice for adoptive transfer were mixed 1:1 and injected i.v. into Ly49H-deficient C57BL/6 recipients 1 d prior to MCMV infection.

Mice were injected i.p. with 1 × 10⁴ PFU or 2.5 × 10³ PFU of Smith strain MCMV for adoptive NK cell transfer experiments. For MCMV quantitation, oral lavage was collected by washing the sublingual cavity with sterile saline solution and used directly in quantitative PCR analysis (26). DNA was isolated from blood using a ReliaPrep Blood gDNA Miniprep System kit (Promega) and 1 μl was used for quantitative PCR analysis. MCMV primer pairs: forward, 5′-AGCCACCAACATTGACCACGCAC-3′ and reverse, 5′-GCCCCAACCAGGACACACAACTC-3′.

Total RNA from 6-wk-old Klf12^+/+ and Klf12^F/F β-actin Cre⁻ littermate males was isolated from CD3⁻NK1.1⁺ NK cells, CD4⁺TCRβ⁺CD25⁻ T cells, CD8⁺TCRβ⁺CD25⁻ T cells and B220⁺IgM⁺IgD⁺ B cells following the ImmGen standard protocol (https://www.immgen.org/). One thousand cells (≥99% purity) were double sorted directly into 96-well plates containing TCL buffer (Qiagen) and 1% 2-ME, frozen, and analyzed by the Broad Technology Labs for SmartSeq2 library preparation and NextSeq500 sequencing. Transcripts were quantified using Cuffquant and then normalized using DEseq (27–31). Raw data were interpreted by Database for Annotations, Visualization and Integrated Discovery (DAVID), analyzing all differentially expressed genes using default parameters (32).

Single-cell suspensions were incubated with anti-CD16 plus CD32 (2.4G2) mAb to block Fc receptors 15 min on ice and then stained with mAbs to cell surface or intracytoplasmic Ags of interest. mAbs were purchased from BD Biosciences, BioLegend, eBioscience, and Tonbo Biosciences. Samples were acquired on an LSR II or LSR Fortessa (BD Biosciences) and analyzed on FlowJo software (Tree Star).

The unpaired, two-tailed Student t test was used to analyze results in Prism GraphPad. A p value ≤0.05 was considered significant. Error bars represent SD.

ImmGen results revealed that mouse NK cells preferentially expressed a novel transcription factor, KLF12. Consistent with ImmGen data, quantitative PCR from highly purified lymphocyte populations from splenocytes of C57BL/6 mice confirmed that NK cells expressed 25-fold more Klf12 transcripts than CD4⁺, CD8⁺ T cells, and CD19⁺ B cells (Fig. 1A). Klf12 expression increased as splenic NK cells transitioned from semimature CD27⁺CD11b⁺ double positive (DP) to fully mature CD27⁻CD11b⁺ NK cells (Fig. 1B). Analysis of microarray and RNA sequencing (RNA-Seq) data from ImmGen indicated absence or low transcription of Klf12 in all other hematopoietic cell types, including macrophages, monocytes, dendritic cells, granulocytes, mast cells, innate lymphoid cell (ILC)2, ILC3, γδ T cells, thymocytes, invariant NKT cells, hematopoietic stem cells, and virus-activated CD8⁺ T cells (https://www.immgen.org). Furthermore, ILC1 do not express Klf12 (33). Taken together, mouse NK cells preferentially express Klf12.

To investigate the role of KLF12 in NK cells, we generated Klf12 knockout mice using a targeted vector to excise exon 3 of Klf12 (Fig. 2A). Integration of the targeting vector was confirmed by the presence of a 6.44-kb band in the Southern blot of selected stem cell clones and a 5.7-kb band detected in mice by long range PCR (Fig. 2B, 2C). The lacZ and neomycin cassettes were removed by crossing to Rosa26-Flippase mice, resulting in progeny mice bearing loxP sites flanking exon 3 of Klf12 (Fig. 2A). β-Actin Cre recombinase excised exon 3 of Klf12 and the progeny were genotyped by PCR (Fig. 2D). Klf12 transcripts without exon 3 were detected in whole splenocytes from Klf12^F/F mice (Fig. 2E) that resulted in a premature translational stop codon (Fig. 2F). The putative truncated protein encodes a 65 aa peptide lacking all functional KLF12 protein domains. However, as none of the Abs currently available are KLF12 specific, we were unable to measure KLF12 protein in WT or knockout mice. The Klf12-null mice bred in predicted Mendelian ratios, and there were no abnormalities observed in growth or weight compared with WT or heterozygous littermates. T cell development in the thymus and spleen and T cell development in BM, spleen, lymph nodes, and the peritoneal cavity are normal in KLF12-deficient mice when compared with WT and heterozygous littermates (Supplemental Fig. 1). Additionally, stimulation of CD4⁺ and CD8⁺ T and B cells isolated from lymph nodes through their Ag receptors by immobilized anti-CD3 or anti-IgM, respectively, resulted in downstream signaling of ERK (Supplemental Fig. 2) and proliferation (Supplemental Fig. 3) equivalent to WT or heterozygous littermates.

Phenotypic analysis of Klf12^+/+, Klf12^F/+, and Klf12^F/F NK cells was performed to determine whether KLF12 is required for NK cell development. We examined NK developmental subsets as defined by expression of CD27 and CD11b and expression of Ly49 and NKG2 activating and inhibitory receptors in the BM and spleen. Klf12^F/+ and Klf12^F/F mice had normal frequencies and numbers of all NK cell developmental subsets in the BM and spleen, equivalent to Klf12^+/+ littermate controls. Expression of activating and inhibitory receptors was also similar to Klf12^+/+ controls (Fig. 3A). Furthermore, the number and frequency of liver ILC1 and NK cells was equivalent in Klf12^+/+ and Klf12^F/F mice (data not shown). Therefore, KLF12 deficiency does not affect NK cell development or expression of NK receptors.

We assessed whether KLF12 deficiency affected NK cell effector functions. Klf12^F/+ and Klf12^F/F NK cells produced IFN-γ and degranulated similar to Klf12^+/+ NK cells upon stimulation with IL-12 and -18, anti-NK1.1, or PMA and ionomycin (Fig. 3B). Furthermore, the in vitro proliferative capacity of Klf12^F/+ and Klf12^F/F NK cells upon stimulation with IL-15, the RMA lymphoma cell line, and MCMV m157 (ligand of the activating Ly49H receptor)–transduced RMA cells was comparable to Klf12^+/+ NK cells (Fig. 3C). Therefore, KLF12 deficiency does not affect the in vitro NK cell effector functions and cytokine- and Ag-driven proliferation in the knockout mice.

We also tested in vivo NK cell responses upon MCMV infection. Klf12^+/+, Klf12^F/+, and Klf12^F/F mice were infected with Smith strain MCMV, and expansion of Ly49H⁺ NK cells and viral titers were monitored over 28 d. Naive, uninfected Klf12^+/+, and Klf12^F/F mice had a similar percentage of Ly49H⁺ NK cells in the blood. After MCMV infection, there was comparable expansion of Ly49H⁺ NK cells in Klf12^+/+, Klf12^F/+, and Klf12^F/F mice. Viral titers in the blood and oral lavage were indistinguishable (Fig. 3D). Therefore, KLF12 deficiency does not alter NK cell effector functions in vitro and upon in vivo MCMV challenge.

We hypothesized that redundancy or compensatory mechanisms with other KLF members may mask the effects of KLF12 deficiency in NK cell development and function. Therefore, we performed RNA-Seq to determine the effects of KLF12 deficiency on the transcriptome of Klf12^F/F mice. Total RNA was collected from purified T cells, B cells, and NK cells from 6-wk-old Klf12^+/+ and Klf12^F/F β-actin Cre⁻ littermate male mice. Klf12^F/F NK cells had increased transcripts involved in regulating cellular transcription, NF-κB activity, and cellular division compared with Klf12^+/+ NK cells (Fig. 4A). Of the differentially expressed genes, we confirmed increased expression of B cell translocation gene 3, Btg3, encoding an antiproliferative protein, in Klf12^F/F NK cells but not in Klf12^F/F T cells, B cells, or Klf12^+/+ NK cells (Fig. 4B). Furthermore, Klf12^F/F NK cells isolated from mixed BM chimeric mice had more Btg3 transcripts than Klf12^+/+ NK cells purified from the same mice (Fig. 4C). Altogether, KLF12-deficient NK cells intrinsically express more Btg3 transcripts. By contrast, no significant transcriptional differences were observed comparing Klf12^+/+ and Klf12^F/F T cells and B cells, indicating that the alterations in Klf12^F/F NK cells were cell intrinsic.

Although in the intact KLF12-deficient mice we observed no marked alteration in NK cell phenotype, in many cases gene deficiencies in knockout mice are only revealed when the gene-deficient cells are in competition with WT cells. Therefore, we generated mixed BM chimeras of Klf12^+/+ and Klf12^F/F cells at a 1:1 ratio to assess the role of KLF12 in NK cells in a competitive setting. After reconstitution of the hematopoietic cells for about 2 mo, we observed reduced absolute numbers of Klf12^F/F NK cell developmental subsets in the BM and spleen (Fig. 5A). The levels of expression (as reflected by mean fluorescence intensity [MFI]) of activating and inhibitory NK receptors were comparable between Klf12^+/+ and Klf12^F/F cells (Fig. 5A). In a competitive setting, Klf12^F/F NK cells produced less IFN-γ upon IL-12 plus IL-18 stimulation but more IFN-γ upon anti-NK1.1 stimulation. Klf12^F/F NK cells also degranulated more than Klf12^+/+ NK cells upon anti-NK1.1 stimulation (Fig. 5B). Thus, KLF12 deficiency alters NK cell development and effector function, but not receptor expression, when the NK cells are developed in a competitive setting. By contrast, Klf12^F/F and WT CD4⁺ and CD8⁺ T cells responded equivalently to MCMV infection (Supplemental Fig. 4).

Because we observed increased expression of the antiproliferative gene, Btg3, in Klf12^F/F NK cells, we directly assessed whether in vitro proliferation of Klf12^F/F NK cells was preferentially affected in BM chimeras compared with Klf12^+/+ NK cells. We observed a statistically significant proliferative impairment of Klf12^F/F NK cells in BM chimeric mice in response to IL-15 or coculture with m157-expressing RMA lymphoma cells compared with Klf12^+/+ NK cells (Fig. 5C). Klf12^F/F NK cells also had impaired proliferation in vivo when BM chimeric mice were infected with MCMV. At day 7 after MCMV infection, we observed a statistically significant reduction in the numbers of Ly49H⁺ Klf12^F/F NK cells compared with Ly49H⁺ Klf12^+/+ NK cells (Fig. 5D). Furthermore, this observation was recapitulated when a mixture of mature Ly49H⁺ Klf12^+/+ and Klf12^F/F NK cells were adoptively transferred into Ly49H-deficient mice and infected with MCMV (Fig. 5E). Together, these findings reveal that KLF12 deficiency impairs NK cells to proliferate in response to cytokines and Ags when in competition with WT NK cells.

The common γ-chain (CD132) cytokine receptor family is important for lymphocyte development during BM reconstitution and for lymphocyte activation. We observed slightly decreased CD132 expression on Klf12^F/F NK cells compared with Klf12^+/+ NK cells in BM chimeric mice, whereas expression of the β-chain (CD122) was identical. Furthermore, CD132 expression was significantly decreased on all Klf12^F/F splenic NK cell developmental subsets compared with Klf12^+/+ NK cells (Fig. 6A). Expression of CD132, but not CD122, was decreased on KLF12-deficient NK cells from mixed BM chimeric mice.

We examined which members of the common γ-chain cytokine receptor family might be affected by the decreased expression of CD132 on KLF12-deficient NK cells. We hypothesized that the responsiveness of IL-15 might be affected by KLF12 deficiency because it is important for NK cell development (34, 35). We examined the phosphorylation of STAT5 in NK cells from BM chimeric mice upon IL-15 stimulation as a functional measure of responsiveness to IL-15. The percentages of NK cells responding and MFI of p-STAT5 were similar between Klf12^+/+ and Klf12^F/F NK cells upon culture with low and high concentrations of IL-15. Thus, responsiveness to IL-15 remains intact and unaffected by KLF12 deficiency.

The IL-21R is another member of the common γ-chain cytokine receptor family that has been implicated in NK cell maturation and proliferation (36–38). Although IL-21R is dispensable for mouse NK cell development, IL-21 induces their maturation but inhibits their proliferation (37). Furthermore, IL-21 enhances NK cell responses against tumors expressing ligands for the activating NK receptor, NKG2D (39). Given the inhibitory role of the IL-21R in NK cell proliferation, we assessed IL-21R expression on NK cells in BM chimeric mice. Interestingly, we observed higher levels of IL-21R expression on splenic Klf12^F/F NK cells. This was also evident on all splenic Klf12^F/F NK cell developmental subsets. In fact, the most immature subset of NK cells, lacking expression of CD27 and CD11b (termed double negative [DN]), expressed the highest levels of the IL-21R (Fig. 7A). Thus, expression of the IL-21R is increased on mouse NK cells by KLF12 deficiency.

We measured p-STAT3 expression upon IL-21 stimulation to determine whether increased IL-21R expression on KLF12-deficient NK cells resulted in increased IL-21–induced signaling. At a low concentration of IL-21, we detected increased percentages and levels of expression (MFI) of p-STAT3 in Klf12^F/F NK cells compared with Klf12^+/+ NK cells in the mixed BM chimeras (Fig. 7B). Furthermore, it was the most immature subset of NK cells (DN) that upregulated p-STAT3 the most compared with other developmental subsets. This is not surprising because the DN subset expressed the highest levels of IL-21R and therefore should have increased p-STAT3 upregulation.

Subsequently, we evaluated whether increased IL-21R expression and signaling in KLF12-deficient NK cells correlated with greater IL-21–mediated inhibition of proliferation. We cultured splenocytes from mixed BM chimeric mice with IL-15 in the presence or absence of IL-21 and monitored the maturation and percentage of NK cells for 7 d. We observed reduced percentages of Klf12^F/F NK cells upon IL-15 culture compared with Klf12^+/+ NK cells (Fig. 7C). By day 7 of IL-15 culture, all NK cells had proliferated (measured by the percentage of divided cells), and we were unable to observe differences in proliferation as we observed on day 5 of IL-15 culture (Fig. 5C). Addition of IL-21 to the culture reduced the overall percentages of NK cells and inhibited the proliferation of Klf12^F/F NK cells more than Klf12^+/+ NK cells. In fact, the most immature Klf12^F/F NK cell DN subset was significantly reduced in percentage by IL-21, but there was no difference in proliferation of total NK cells, possibly because the Klf12^F/F NK cell DN subset was driven to mature in response to the IL-21. Although there was a higher percentage of Klf12^F/F semimature CD27⁺CD11b⁺ DP subset of NK cells, they did not proliferate as well as their Klf12^+/+ counterparts. KLF12 deficiency in NK cells resulted in increased expression and signaling of the IL-21R. This enhanced IL-21R signaling correlated with greater inhibition of NK cell proliferation, particularly in the most immature DN NK cell subset.

Given that KLF12-deficient NK cells have impaired proliferation, intrinsically express more Btg3 transcripts, and express higher levels of the IL-21R, we sought to determine whether IL-21 might inhibit NK cell proliferation via BTG3. To address this, we cultured enriched WT NK cells in IL-21 and assessed Btg3 expression. We observed a significant increase in Btg3 expression after 2 h of IL-21 stimulation and then a progressive decrease in expression with time. Even after 24 h of continual IL-21 stimulation, Btg3 expression remains significantly increased (Fig. 8). Thus, IL-21 directly induces Btg3 expression in NK cells and correlates with the decreased proliferative activity.

Unlike their other lymphocyte counterparts, NK cells have the unique ability to recognize and lyse target cells without prior exposure via expression of their germline-encoded receptors. However, there are no NK cell-specific genes that are exclusively expressed by all NK cells. Transcription factors, signaling components downstream of NK activating or inhibitory receptors, and even expression of certain NK activating receptors are shared among NK cells and other lymphocytes. From the ImmGen studies, we identified Klf12, encoding a novel transcription factor, to be preferentially expressed in mouse NK cells and not in ILC1, ILC2, ILC3, T cells, or B cells. In this study, we generated Klf12 knockout mice to assess its role in NK cell development or effector function. We used β-actin Cre recombinase to delete KLF12 in all cells. We observed normal lymphocyte development, proliferation, and activation in KLF12-deficient mice (Supplemental Figs. 1–4). We also generated mice in which KLF12 was conditionally deleted only in NK cells using Ncr1-Cre mice with similar results (data not shown). However, we found that in competitive mixed BM chimeras, KLF12-deficient NK cells demonstrated less robust proliferation that correlated with higher levels of expression of Btg3, an antiproliferative gene, which is upregulated by IL-21. Notably, KLF12-deficient NK cells express higher levels of IL-21R and have elevated p-STAT3 signaling in response to IL-21 compared with WT NK cells.

In the absence of competition with WT NK cells, KLF12 is dispensable for NK cell lineage commitment, development, and effector functions. KLF12-deficient NK cells produced IFN-γ and degranulated equivalently as WT NK cells. Furthermore, NK cell proliferation in vitro and in vivo in the context of MCMV infection was also unperturbed by KLF12 deficiency in the Klf12^F/F mice. The lack of a NK cell phenotype may be due to compensatory mechanisms or redundancy among other members of the KLF family, particularly KLF3 and KLF8. Both KLF3 and KLF8 share the conserved PVDLS domain with KLF12 (14, 16), and KLF3 is highly expressed in mouse NK cells (https://www.immgen.org/). Interestingly, KLF12-deficient NK cells have similar expression of KLF3 and KLF3 transcripts as WT NK cells (data not shown, GSE128962). However, we were unable to address whether compensatory mechanisms masked KLF12 deficiency in vivo because of embryonic lethality in C57BL/6 KLF3-deficient mice and KLF3 and KLF8–double knockout mice (Ref. 40 and personal communication with M. Crossley). Further studies generating NK cell-specific double or triple knockout mice would be required to uncover potential compensatory pathways in KLF12-deficient NK cells.

Recent studies have shown that KLF12 regulates proliferation of many cancer cell lines. Overexpression of KLF12 in endometrial and lung cancer cell lines correlated with decreased apoptosis, increased cellular proliferation, and increased in vivo tumor growth (41–43). Conversely, downregulation of KLF12 resulted in a proliferative defect in multiple cancer cell lines (19, 23, 44–47). These observations were recapitulated in primary human cancer cells and mouse kidney cells (48, 49). In this study, we observed a proliferative defect in KLF12-deficient NK cells in a competitive setting in BM chimeric mice upon Ag- and cytokine-mediated proliferation. Although we observed a proliferative impairment upon IL-15 stimulation, the expression of p-STAT5 was comparable between KLF12-deficient and WT NK cells. Interestingly, we detected increased p-STAT3 expression in KLF12-deficient NK cells upon IL-21 stimulation, which correlated with greater IL-21–mediated inhibition of proliferation. Similarly, overexpression of KLF12 in an endometrial adenocarcinoma cell line resulted in decreased p-STAT3 expression upon LIF stimulation, which inhibits cellular differentiation (42). The proliferative defect in KLF12-deficient NK cells might be an intrinsic effect of p-STAT3 expression.

Alternatively, the proliferative impairment observed in KLF12-deficient NK cells may be due to the increased expression of an antiproliferative transcription factor, Btg3. BTG3 is a member of the antiproliferative BTG/Tob protein family that inhibits entry into the S-phase of cell cycle progression. It has been shown that upon DNA damage, p53 binds to the BTG3 promoter and induces its expression to regulate cell cycle checkpoints (50). Downregulation of BTG3 expression is associated with enhanced cell proliferation, growth, and migration (51, 52). Conversely, overexpression of BTG3 is associated with suppressed proliferation, reduced cancer invasiveness, and cellular apoptosis in primary cancers and cancer cell lines (53–55).

During MCMV infection, IL-21 induces mouse NK cells to produce IL-10, which affects dendritic cell activation and CD8⁺ T cell responses (38, 56–58). We also observed proliferative impairment of KLF12-deficient NK cells during MCMV infection in the mixed BM chimeric mice and in the adoptive transfer of mature NK cells. IL-21 present during MCMV infection may be inhibiting NK cell proliferation and inducing IL-10 production to modulate adaptive immune responses. However, we observed comparable numbers of KLF12-deficient and WT CD4⁺ AND CD8⁺ naive, effector memory, and MCMV-specific NKG2D⁺ CD8⁺ T cells in MCMV-infected BM chimeric mice (Supplemental Fig. 4), suggesting that the T cell responses against MCMV are normal.

Although the IL-21R is not required for mouse NK cell development, IL-21 induces NK cell maturation, inhibits their proliferation, and enhances their NKG2D-mediated antitumor response (36–39). In this study, we noted increased expression of the IL-21R on KLF12-deficient NK cells compared with WT NK cells in mixed BM chimeric mice. Increased IL-21R expression resulted in enhanced IL-21 signaling and correlated with inhibition of KLF12-deficient NK cell proliferation. It appears as though IL-21 is driving NK cell maturation, but in the absence of KLF12 NK cell maturation may be arrested at the semimature DP stage. This is consistent with the fact that KLF12 is upregulated at the semimature DP stage. Moreover, IL-21 alone induces Btg3 expression in NK cells, which may reinforce the proliferative defect in KLF12-deficient NK cells in competitive situations. Conditional deletion of Btg3 and Il21r in NK cells will be needed to definitively address this hypothesis.

We thank the UCSF Cell and Genome Engineering Core for assistance and Dr. Hong-Erh Liang in the Dr. Richard Locksley laboratory for guidance in generating Klf12 conditional knockout mice, the ImmGen Consortium for performing the RNA-Seq experiment, Dr. Alice Chan in the Dr. Mark Anderson laboratory for technical guidance, and members of the Lanier Lab for helpful discussions.

The authors have no financial conflicts of interest.