The transcription factor Zbtb32 controls the proliferative burst of virus-specific natural killer cells responding to infection

MCMV infection induces Zbtb32 expression in NK cells

To identify BTB-ZF genes that might regulate antigen-specific NK cell responses, we first compared expression of 47 BTB-ZF genes in sorted Ly49H⁺ NK cells from MCMV-infected and uninfected animals by microarray. Remarkably, only three (Zbtb32, Hic1 and Bcl6) were significantly modulated in NK cells on day 1.5 post-infection (p.i.), and of these three, Zbtb32 was the most highly upregulated (Fig. 1a). Indeed, of the >35,000 genes evaluated on the microarray (GEO code: GSE15907)²⁶, Zbtb32 was among the top 30 most highly induced on day 1.5 p.i. (Fig. 1b). The microarray data were confirmed by quantitative reverse-transcription polymerase chain reaction (qRT-PCR), which revealed a >100-fold upregulation of Zbtb32 transcript in Ly49H⁺ NK cells at day 2 p.i. with MCMV (Fig. 1c), and by flow cytometry, which showed elevated Zbtb32 protein expression on day 2 and day 3 p.i. (Supplementary Fig. 1). Expression of Zbtb32 transcript and protein were transient, and both returned to baseline abundance by day 4 p.i., suggesting an equally rapid down-regulation of this transcription factor following its induction during viral infection.

Zbtb32 is required for NK cell anti-viral immunity

Given its rapid upregulation following viral infection, we hypothesized that Zbtb32 might regulate the function and/or phenotype of activated antigen-specific NK cells. It has previously been shown that adoptively transferred Ly49H⁺ NK cells can rescue NK cell-deficient or -impaired animals from otherwise fatal doses of MCMV⁸. To test the protective capacity of Zbtb32^−/− NK cells, we transferred Ly49H⁺ wild-type or Zbtb32-deficient NK cells into neonate mice, which lack mature B, T and NK cells and are highly susceptible to viral infection, and assessed the ability of the transferred cells to protect against lethal MCMV challenge^8,27. As expected, wild-type Ly49H⁺ NK cells had a significantly protective effect, with 33% of animals surviving past 3 weeks (median survival of 15 days; Fig. 2a). In contrast, neonates receiving Zbtb32^−/− NK cells died early after infection (median survival of 8 days), similar to control animals that did not receive NK cells. To further confirm the requirement for Zbtb32 in protective NK cell responses, we compared the ability of wild-type and Zbtb32^−/− Ly49H⁺ NK cells to protect adult Ly49H-deficient mice against lethal doses of recombinant vesicular stomatitis virus expressing the MCMV-derived m157 protein (VSV-m157). In this model, VSV-m157-infected animals rapidly succumb to paralysis and death in the absence of transferred Ly49H⁺ NK cells. Whereas adoptively transferred wild-type Ly49H⁺ NK cells protected ∼50% of infected animals from death, all recipients of Zbtb32^−/− Ly49H⁺ NK cells died rapidly after infection, at a time comparable to animals that did not receive NK cells (Fig. 2b). Thus, Zbtb32 is required for protective anti-viral responses by antigen-specific NK cells.

Zbtb32 is dispensable for NK cell activation and effector function

NK cells respond to MCMV infection by rapidly producing cytolytic proteins and pro-inflammatory cytokines, and for those expressing the Ly49H receptor, by proliferating to enlarge the overall pool of effector cells^1,7. Given the protective defect of Zbtb32^−/− NK cells, we sought to delineate which of these effector functions was controlled by Zbtb32. Zbtb32 has previously been shown to suppress cytokine production in activated T cells^28–32. Zbtb32^−/− CD4⁺ T cells expressed more T_H2 cytokines, including IL-4, IL-5 and IL-13, following activation in vitro²⁹ or in the context of allergen-induced models of airway and contact hypersensitivity in vivo^31,32. Similarly, activated Zbtb32^−/− CD8⁺ T cells produced more IFN-γ than their wild-type counterparts²⁹. In contrast to the findings in T cells, Zbtb32^−/− and wild-type NK cells were indistinguishable in their ability to produce IFN-γ in response to MCMV infection in vivo (Fig. 3a), and following stimulation in vitro with pro-inflammatory cytokines or via cross-linking of activating receptors (Supplementary Fig. 2a). Furthermore, wild-type and Zbtb32^−/− NK cells from MCMV-infected mice similarly upregulated expression of the potent cytotoxic protein Granzyme B (Fig. 3b) and CD107a, a marker of degranulation (data not shown). In addition, NK cells in Zbtb32^−/− animals were similar to those in wild-type animals with respect to their ability to kill adoptively transferred target cells (Supplementary Fig. 2b). Following MCMV infection, wild-type and Zbtb32^−/− NK cells were comparable with respect to expression of the activation markers CD69 and KLRG1 (Fig. 3c), and both populations exhibited similar conversion to fully mature CD27^loCD11b^hi cells by day 7 p.i. (Fig. 3d). Taken together, our data indicate that the protective deficiency of Zbtb32^−/− NK cells does not stem from a defect in killing capacity, cytokine production or maturation following viral infection.

Cell-intrinsic requirement for Zbtb32 in NK cell expansion

Given that effector responses were largely intact in Zbtb32-deficient NK cells, we postulated that their protective defect may arise from an impairment in antigen-driven proliferation. To test this hypothesis, we co-transferred equal numbers of Zbtb32^−/− and wild-type Ly49H⁺ NK cells into Ly49H-deficient hosts and evaluated the ability of the transferred Ly49H⁺ cells to expand after MCMV infection (Supplementary Fig. 3a). Whereas transferred wild-type Ly49H⁺ NK cell populations rapidly expanded and reached peak numbers and percentages in the spleen by day 7 p.i., transferred Zbtb32^−/− Ly49H⁺ NK cells were barely detectable at this time point (Fig. 4a,b). Similar results were obtained when Zbtb32^−/− and wild-type Ly49H⁺ NK cells were transferred separately (Fig. 4c), indicating that wild-type NK cells were not simply out-competing their Zbtb32-deficient counterparts. The low numbers and percentages of Zbtb32^−/− NK cells in the spleen were not due to aberrant trafficking, since a similar deficiency was observed in the liver, lymph node and lung (Fig. 4d). Moreover, the relative percentages of wild-type and Zbtb32^−/− NK cells (∼95% vs. ∼5%) remained constant from day 5 through day 45 p.i., ruling out delayed population expansion kinetics and a role for Zbtb32 in memory NK cell formation and maintenance (Fig. 4e). The requirement for Zbtb32 in antigen-specific NK cell expansion was not limited to MCMV infection, since the Zbtb32^−/− Ly49H⁺ NK cell population also exhibited an expansion defect following infection with m157-expressing VSV or m157-expressing VacV, although the magnitude of this defect was directly proportional to number of cellular divisions driven by each virus (Fig. 4f and data not shown).

To test whether there exists a gene dosage effect for Zbtb32 in antiviral NK cell responses, NK cells from Zbtb32^−/−, Zbtb32^+/−, or Zbtb32^+/+ littermates were individually co-transferred along with congenic wild-type NK cells into Ly49H-deficient hosts. As expected, Ly49H⁺ NK cells from Zbtb32^+/+ animals increased as well as those from wild-type donors, whereas Ly49H⁺ NK cells from Zbtb32^−/− littermates barely increased at all, confirming that Zbtb32-deficiency, and not background-related effects, was responsible for the expansion defect (Fig. 4g). Surprisingly, Ly49H⁺ NK cells containing only one Zbtb32 allele (Zbtb32^+/−) were almost as dysfunctional as fully-deficient Zbtb32^−/− NK cells, underscoring that maximal Zbtb32 expression is required for NK cell population expansion after infection.

Because Zbtb32 is also upregulated in antigen-specific CD8⁺ T cells following viral or bacterial infection (data not shown), and has previously been reported to regulate T cell proliferation in vitro (although these studies suggested a restrictive role for Zbtb32 in TCR-driven proliferation)^29,33, we measured the expansion of antigen-specific CD8⁺ T cells in wild-type:Zbtb32^−/− mixed bone marrow chimeras during MCMV infection (Fig. 4h). In contrast to our findings in NK cells, Zbtb32-deficient antigen-specific CD8⁺ T cell populations expanded similarly or better than their wild-type counterparts. Thus, taken together, our data indicate an essential cell-intrinsic and lineage-specific function for Zbtb32 in promoting NK cell expansion after infection.

Zbtb32 controls NK cell proliferation, but not survival

We next investigated the extent to which Zbtb32 controlled the various cellular processes of proliferation, survival and apoptosis in activated NK cells, any one of which could explain the observed population expansion defect. Transfer of CFSE-labeled Ly49H⁺ NK cells confirmed that Zbtb32-deficiency significantly impaired MCMV-driven proliferation in vivo (Fig. 5a). The impaired proliferative capacity of Zbtb32^−/− Ly49H⁺ NK cells was further corroborated by their decreased expression of the proliferation marker Ki67 (Fig. 5b) and by their inability to incorporate BrdU after short-term pulses on day 4 and day 6 p.i. (Fig. 5c). This defect was mirrored at a molecular level by a failure of Zbtb32-deficient NK cells to upregulate a suite of genes involved in cell cycle progression (Fig. 5d).

Because Zbtb32 was found to be upregulated to a similar degree in both Ly49H⁺ and Ly49H⁻ NK cells after infection (Supplementary Fig. 3b), we tested whether Zbtb32 might also regulate the low-level NK cell proliferation that occurs independently of Ly49H-m157 engagement. Indeed, even Ly49H⁻ NK cells showed impaired proliferation in the absence of Zbtb32, as did Ly49H⁺ NK cells responding to infection with MCMV Δm157 (Supplementary Fig. 3c,d), although this defect was subtle given that all NK cells proliferated very little in these settings.

In contrast to their proliferation defect, Zbtb32^−/− NK cells exhibited no evidence of enhanced apoptosis by staining for Annexin V, the pro-apoptotic molecule Bim, or activated caspases (Fig. 5e and data not shown). In T cells, the pro-survival protein Bcl-2 is downregulated in effector cells with the highest proliferative capacity, and subsequent apoptosis of these highly proliferative Bcl-2^lo cells marks the contraction phase of the effector T cell response^34,35. Consistent with a defect in proliferation but not survival, Zbtb32^−/− NK cells failed to undergo activation-induced downregulation of Bcl-2 (Fig. 5f). Thus, these findings highlight a specific requirement for Zbtb32 in the proliferation but not survival of NK cells following viral infection.

Zbtb32 is dispensible for development and homeostasis

The NK cell compartment of Zbtb32^−/− animals was phenotypically indistinguishable from that of wild-type animals, suggesting that Zbtb32-deficiency did not impair NK cell development or abrogate proliferation in that context. The percentage and number of mature NK cells in peripheral organs and of NK cell developmental subsets in the bone marrow of Zbtb32-deficient animals was comparable to that observed in wild-type animals (Supplementary Fig. 4a,b and data not shown). Similarly, NK cells in Zbtb32^−/− animals expressed normal amounts of markers typically associated with NK cell identity and function, including Ly49H, Ly49D, NK1.1, DX5, Ly49C or Ly49I and Ly49A (Supplementary Fig. 4c,d).

In addition, wild-type and Zbtb32-deficient NK cells repopulated the hematopoietic compartment of mixed bone marrow chimeric animals with similar efficiencies (Supplementary Fig. 5, left panel), further confirming that Zbtb32 was dispensible for NK development. However, when wild-type and Zbtb32^−/− Ly49H⁺ NK cells from mixed bone marrow chimeric animals were adoptively transferred into naive Ly49H-deficient recipients, only the wild-type population was able to expand following MCMV challenge, reaffirming that even Zbtb32^−/− NK cells that developed in a Zbtb32-sufficient environment exhibited an expansion defect after infection (Supplementary Fig. 5, middle and right panels). Consistent with a dispensable role in NK cell proliferation during development, Zbtb32^−/− NK cells exhibited normal homeostatic proliferation when transferred into lymphocyte-deficient hosts (Fig. 5g) and when expanded ex vivo with IL-2 and IL-15, cytokines important for NK cell proliferation and survival during development and at steady-state (Fig. 5h). Thus, Zbtb32 does not appear to be necessary for NK cell proliferation in a non-inflammatory environment. Collectively, our data indicate that Zbtb32 is dispensible for development and homeostasis-associated proliferation, but rather specifically regulates NK cell proliferation in the context of an infectious or inflammatory setting.

Inflammation drives Zbtb32 expression in NK cells

Because maximal levels of Zbtb32 mRNA in NK cells were observed at early time points in vivo (by ∼48 h p.i.), when inflammation is high but viral dissemination (and thus antigen availability) is still lagging, and because both Ly49H⁺ and Ly49H⁻ NK cells were able to upregulate Zbtb32 after infection, we hypothesized that signals from proinflammatory cytokines, rather than interactions between the Ly49H receptor and the m157 viral antigen, might control Zbtb32 expression in activated NK cells. Indeed, resting NK cells incubated ex vivo with the proinflammatory cytokines IL-12, IL-18, or IFN-α/β induced high expression of Zbtb32 mRNA, with IL-12 and IL-18 co-treatment exerting a strongly synergistic effect (Fig. 6a). In contrast, Zbtb32 mRNA was barely detectable following crosslinking of the activating receptors Ly49H, Ly49D, NKG2D or NKp46. To test whether signals from pro-inflammatory cytokines were required for Zbtb32 induction in vivo, we utilized mixed bone marrow chimeric animals harboring both wild-type NK cells and NK cells lacking receptors for either type I IFNs, IL-12, or IL-18, or for both type I IFNs and IL-12. Compared to their wild-type counterparts, cytokine receptor-deficient NK cells were markedly impaired in their ability to upregulate Zbtb32 mRNA following MCMV infection in vivo, with NK cells lacking both the IL-12 and the type I interferon receptors showing the greatest defect (Fig. 6b).

Given that pro-inflammatory cytokines induce Zbtb32, we postulated that specific transcriptional activators downstream of cytokine receptors might directly regulate Zbtb32 expression in NK cells. Analysis of the Zbtb32 promoter revealed several conserved non-coding sites (CNS) within a DNase I hypersensitivity region previously identified in human NK cells³⁶ (Fig. 6c). Chromatin immunoprecipitation (ChIP) experiments showed enrichment of acetylated histone marks at these sites (specifically, H3K27Ac, which marks transcriptionally active promoters), consistent with a likely role in transcriptional activation of Zbtb32 (Fig. 6d). In support of direct regulation by cytokine receptor signaling, the CNS regions of the Zbtb32 promoter were found to harbor putative binding sites for the transcription factor STAT4, which acts downstream of the IL-12 receptor to induce target gene expression in lymphocytes (Fig. 6c). Similar to IL-12R-deficient NK cells, NK cells lacking STAT4 were also significantly impaired in MCMV-induced Zbtb32 expression (Fig. 6e). Using STAT4 ChIP, we observed substantial binding of STAT4 at a CNS in the Zbtb32 promoter approximately 750 base pairs upstream of the transcriptional start site, indicating that Zbtb32 may be a direct target of STAT4, acting as a transmitter of IL-12 signaling (Fig. 6f). Decreased Zbtb32 expression correlates with the impaired proliferation recently observed in IL-12R- or STAT4-deficient NK cells³⁷, consistent with a central role for Zbtb32 in transmitting pro-proliferative signals downstream of inflammatory cues. Thus, signals from proinflammatory cytokines, but not activating receptors, are sufficient to induce and are required for maximal Zbtb32 expression in primary NK cells.

Zbtb32 promotes proliferation by antagonizing Blimp-1

We next sought to identify the specific pathway(s) regulated by Zbtb32 in proliferating NK cells. Zbtb32-deficiency did not affect expression or phosphorylation of STAT1, STAT3, STAT4 or T-bet in activated NK cells, transcription factors with proven or suspected roles in NK cell effector function^37–39, nor did it influence the expression of Gata3, a known target of Zbtb32 in T cells^28,30 (Supplementary Fig. 6a). These data are consistent with our findings that the non-proliferative features of NK cell activation, including production of IFN-γ and cytotoxic molecules, are intact in the absence of Zbtb32.

Because cell cycle genes are dysregulated in Zbtb32^−/− NK cells during MCMV infection (Fig. 5d), we investigated the mechanism(s) by which Zbtb32 may be promoting cell cycle progression in NK cells. In T and B cells, proliferation of effector lymphocytes (for example, germinal center B cells) depends upon suppression of the anti-proliferative/pro-terminal differentiation factor, Blimp-1 (encoded by Prdm1)⁴⁰. Blimp-1 is upregulated in NK cells during development, where it functions to restrict proliferation, and remains highly expressed in mature resting NK cells. Although the BTB-ZF protein, Bcl-6, is the main antagonist of Blimp-1 in B and T cells⁴⁰, it does not play this role in developing NK cells⁴¹ and is unlikely to do so in activated NK cells, since Bcl-6 is actually down-regulated after infection (Fig. 1b and Supplementary Fig. 6b). Given the high degree of homology between Bcl-6 and Zbtb32, we hypothesized that Zbtb32 rather than Bcl-6 may function in activated NK cells to suppress Blimp-1 expression or function and thereby facilitate antigen-driven proliferation. This idea was supported by a recent report showing a physical interaction between Zbtb32 and Blimp-1 in B cells⁴² and our own observation that Prdm1 expression was elevated in Zbtb32^−/− Ly49H⁺ NK cells (Fig. 7a). However, to formally test this hypothesis, we generated mice harboring NK cells in which both Zbtb32 and Prdm1 have been genetically deleted (Zbtb32^−/−Prdm1^fl/flNkp46-Cre⁺). We predicted that if the primary function of Zbtb32 was to suppress or antagonize Blimp-1 function, then deletion of Blimp-1 in Zbtb32-deficient NK cells should result in a partial or complete restoration of Ly49H⁺ NK cell expansion during MCMV infection. Indeed, we found that whereas NK cells from Zbtb32^−/−Prdm1^fl/flNkp46-Cre^- (Zbtb32-deficient only) donors were severely impaired in their ability to expand after infection, NK cells from Zbtb32^−/−Prdm1^fl/flNkp46-Cre⁺ (Zbtb32 and Prdm1 double-deficient) littermates were fully rescued from this defect and proliferated as well as wild-type cells (Fig. 7b,c). Thus, consistent with our hypothesis, Zbtb32 was no longer required for MCMV-driven NK cell proliferation when Blimp-1 was not present. These findings provide a molecular explanation for how Zbtb32 acts downstream of pro-inflammatory signals to promote a pro-proliferative state in activated NK cells, and reveal a novel antagonistic interaction between Zbtb32 and Blimp-1 that regulates NK cell expansion during infection (Supplementary Fig. 7).

Mice

Mice were bred at MSKCC in accordance with the guidelines of the Institutional Animal Care and Use Committee (IACUC). The following strains were used in this study: C57BL/6 (CD45.2⁺; The Jackson Laboratory), B6.SJL (CD45.1⁺; Taconic), Rag2^−/− × Il2rg^−/− (Taconic), Il18r1^−/−54, Il12rb2^−/−55, Stat4^−/−56, Ifnar1^−/−57, Klra8^−/− (Ly49H-deficient)⁵⁸, B2m^−/− (Taconic), m157-transgenic⁵⁹, Nkp46^iCre (herein called Nkp46-Cre)⁶⁰, Prdm1^fl/fl61, and Zbtb32^−/−31 mice. Femurs from Il12rb2^−/− × Ifnar1^−/− mice were provided by S. Way (University of Minnesota) for making mixed bone marrow chimeras. Zbtb32^−/−Prdm1^fl/flNkp46^iCre/wt animals and littermate controls were generated by breeding at MSKCC. Unblinded experiments were conducted using age- and gender-matched animals in accordance with approved institutional protocols.

Mixed bone marrow chimeric mice were generated as described⁸. Briefly, host C57BL/6 × B6.SJL animals (CD45.1⁺CD45.2⁺) were lethally irradiated with 900 grays of radiation, and reconstituted with a 1:1 mixture of bone marrow cells from B6.SJL WT (CD45.1⁺) and knock-out donor animals (CD45.2⁺), co-injected with anti-NK1.1 (clone PK136) to deplete any residual donor or host mature NK cells. Host NK cells were excluded from all analyses.

Virus infections

MCMV (Smith Strain) and MCMVΔm157 were obtained from L. Lanier (UCSF) and U. Koszinowski (Max von Pettenkofer-Institute), respectively. MCMV was passaged serially through BALB/c hosts two times, then viral stocks were prepared by douncing the salivary glands of infected animals 3 weeks p.i. For neonate protection studies, 4-day old Ly49H-deficient neonates received 1 × 10⁶ WT or Zbtb32^−/− Ly49H⁺ NK cells by intraperitoneal (i.p.) injection and were infected by i.p. injection of 2 × 10³ PFU of MCMV on the following day. WT, Zbtb32^−/− and mixed bone marrow chimera mice were infected by i.p. injection of 7.5 × 10³ PFU of MCMV. For the adoptive transfer studies, animals were infected by i.p. injection of 7.5 × 10² PFU of MCMV one day after receiving approximately 1 × 10⁶ Ly49H⁺ NK cells.

VSV-Indiana and recombinant VSV-m157, made by cloning the coding sequence for the MCMV glycoprotein m157 into the parental VSV-Indiana strain⁶² were provided by K. Schluns (MD Anderson). For the VSV protection studies, adult Ly49H-deficient mice were infected by intravenous (i.v.) injection of 1.5 × 10⁹ PFU of VSV-m157 one day after receiving 1 × 10⁶ WT or Zbtb32^−/− Ly49H⁺ NK cells. In other adoptive transfer experiments, mice were infected by i.v. injection of 1 × 10⁷ PFU of VSV-m157 or VSV-Indiana one day after receiving approximately 1 × 10⁶ Ly49H⁺ NK cells.

VacV-m157 was provided by G. Gasteiger (Memorial Sloan-Kettering Cancer Center) and animals were infected by i.p. injection with 1 × 10⁷ PFU.

Flow cytometry and cell sorting

Cell surface staining was performed using the following fluorophore-conjugated antibodies (purchased from BD Biosciences, eBioscience, BioLegend, and R&D Systems): NK1.1 (PK136), CD11b (M1/70), CD27 (LG.3A10), CD49b/DX5 (DX5), KLRG1 (2F1), NKp46 (29A1.4), CD69 (H1.2F3), Ly49H (3D10) Granzyme B (16G6), CD107a (1D4B), CD45.1 (A20), CD45.2 (104), CD8α (53-6.7), CD122 (TM-β1), TCRβ (H57-597), CD3ε (145-2C11), IFN-γ (XMG1.2), Ly49D (4E5), Ly49A (A1/Ly49A), Ly49C/I (5E6), T-bet (4B10), Bcl-2 (3F11), Ki67 (B56), Gata3 (L50-823), Stat1-pY701 (4a), Stat3-pY705 (4/P-STAT3), Stat4-pY693 (38/p-Stat4) and peptide-loaded MHC class I tetramers (MCMV peptides m45, m38, or m139; NIH Tetramer Facility). Unless otherwise indicated, NK cells were defined as TCRβ^-NK1.1⁺ or CD3^-NK1.1⁺ cells. Intracellular staining was performed by fixing and permeabilizing with the Foxp3/Transcription Factor Staining Kit (eBioscience) for staining intranuclear proteins, including Zbtb32; with Cytofix/Cytoperm™ Plus (BD) for staining cytokines; or with formaldehyde and methanol for staining phosphorylated STAT proteins. For the detection of intracellular Zbtb32, permeabilized cells were stained with 4 μg/ml of a rat anti-Zbtb32 primary antibody (#853111, generously provided by R&D Systems) for 45 min at 4 °C, then with a FITC-conjugated anti-rat secondary antibody for 10 min at 4 °C. Pan-caspase staining was carried out using the FAM FLICA™ in vitro Poly Caspase Kit (Immunochemistry Technologies) per the manufacturer's instructions. Positive control cells were first incubated with 20 μg/ml of anti-mouse CD95 (clone Jo-2; BD Pharmingen) for 2.5 – 4 h at 37 °C (data not shown). For the BrdU labeling experiments, animals were injected i.p. with 1.5 mg of BrdU in PBS on days 3.5 and 5.5 p.i. with MCMV, and BrdU labeling was determined approximately 16 hours later using the APC BrdU Flow Kit (BD Biosciences).

Flow cytometry and cell sorting were performed on the LSR II and Aria II cytometers (BD Biosciences), respectively. For experiments involving qRT-PCR, cell populations were sorted to >95% purity. Data were analyzed with FlowJo software (Tree Star).

NK cell enrichment and adoptive transfer

T, B, and red blood cells were labeled with 10 μg per spleen of rat monoclonal antibodies against CD4 (GK1.5), CD8 (53.6.72), CD19 (1D3) and Ter119 (obtained from the UCSF Core Facility) and magnetically depleted from total splenocyte suspensions using anti-rat IgG-coupled magnetic beads (Qiagen)³⁷. Approximately 1 × 10⁶ enriched NK cells (typically 25-70% of the total cells) were injected i.v. into adult or i.p into neonate recipients one day before infection. In some experiments, splenocytes were labeled prior to transfer with 5 μM Vybrant™ CFDA SE (Molecular Probes).

Ex vivo stimulation of NK cells

Approximately 1 × 10⁵ sorted NK cells were stimulated for 18 hours in RPMI containing 5% fetal bovine serum with recombinant mouse IL-12 (20 ng/ml; R&D Systems), IL-18 (10 ng/ml; R&D Systems), or IFN-α/β (400 U/mL) or with plate-bound antibodies (10 μg/ml; eBioscience) against the activating NK cell receptors Ly49H (3D10), Ly49D (4E5), NKp46 (29A1.4), or NK1.1 (PK136).

For ex vivo NK cell expansion experiments, 2.5 × 10⁵ sorted NK cells (>98% purity) were stimulated in RPMI supplemented with 10% fetal bovine serum, penicillin/streptomycin, L-glutamine and β-mercaptoethanol, and containing recombinant mouse IL-2 (20 ng/ml; R&D Systems) and IL-15 (40 ng/ml; R&D Systems). Fresh media and cytokines were added on alternating days of culture.

Quantitative RT-PCR

Cells were lysed in Tri-Reagent (Ambion). RNA purification and cDNA synthesis were carried out with the Qiagen RNeasy kit (with on-column DNase I treatment), and MuLV reverse transcriptase and Oligo dT₍₁₆₎ primers (Applied Biosystems). iQ Sybr Green SuperMix (BioRad) was used for qRT-PCR. Data were normalized to Actb and expressed as relative target abundance via the ΔΔCT method. Supplementary Table 1 lists relevant primer sequences. For Fig. 5d, the Mouse Cell Cycle PCR Array, RT² SYBR Green qPCR Master Mix, and Mouse Cell Cycle RT² PreAMP cDNA synthesis kits (Qiagen) were used.

Chromatin Immunoprecipitation (ChIP)

Approximately 5 × 10⁶ enriched splenic NK cells were purified by sorting on TCRβ⁻NK1.1⁺ cells. Proteins were cross-linked to chromatinized DNA for 8 min at 25 °C by addition of 1% formaldehyde to the medium. ChIP was carried out as previously described⁶³ using 10 μg of rabbit polyclonal anti-STAT4 (Santa Cruz, sc-468, clone C-20) or 10 μg of rabbit serum IgG antibody as a control (data not shown; Sigma, I5006), or 0.5 μg of anti-H3K27Ac (Abcam, 4729). Relative abundance of regulatory sequences in the Zbtb32 promoter or, as controls, in the Actb and Gapdh promoters or a gene desert ∼50 kb upstream of the Foxp3 gene was measured by qPCR in the antibody-precipitated DNA. Percent input was calculated by 100 × 2̂(CT^{adjusted input} – CT^target), where the CT^input was adjusted from 5% to 100% by subtracting Log₂ 20 CTs. Primer sequences are listed in Supplementary Table 2.

Statistical Analyses

For graphs, data are shown as mean ± SEM and, unless otherwise indicated, statistical differences were evaluated using a two-tailed unpaired Student's t-test, assuming equal sample variance. P values < 0.05 were considered significant and are denoted by an asterisk (*). Statistical differences in survival were determined by Gehan-Breslow-Wilcoxon Test analysis. Graphs were produced and statistical analyses were performed using GraphPad Prism.