Sostdc1 Regulates NK Cell Maturation and Cytotoxicity

Natural killer cells are innate lymphocytes that are important for early immune defense against tumors and virally infected cells. Since the initial discovery of NK cells in 1975, studies from many groups have identified NK cell receptors that are involved in self/nonself recognition, NK cell precursors and stages of maturation, cytokines and transcription factors that are critical for NK cell development and function, and evidence for NK cell immune memory (1–9). Despite over 40 years of NK cell history, the molecular and cellular mechanisms that drive and integrate these processes is still unclear. In particular, how the microenvironment regulates NK cell maturation and function is still an area of ongoing investigation.

Conventional NK cells develop in the bone marrow (BM) from hematopoietic stem cells following a well-established sequence of maturational stages, and egress to the peripheral organs to fully mature and function (10–12). NK cell maturation (Fig. 1A) originates with the immature NK (iNK) (CD27⁺CD11b⁻) cells, which progresses to the transitional NK (tNK) cell stage (CD27⁺CD11b⁺, also known as DP [13], and not to be confused with tissue resident NK cells [14]), then to the final mature NK (mNK) cell stage (CD27⁻CD11b⁺) (13, 15–17). As NK cells progress through these stages, they lose proliferative and cytokine-producing capability but gain cytotoxic ability against target cells (12, 18, 19). Although the BM microenvironment is critical for NK cell development, how the peripheral microenvironment regulates NK cell maturation and cytotoxicity is incompletely understood and requires further investigation.

Sclerostin domain–containing-1 (Sostdc1), also known as Wise, Ectodin, Usag-1, and Sost-like, has been studied in the context of tooth development, kidney disease, hair follicle formation, and bone fracture (20–26). Sostdc1 can function as an antagonist of both bone morphogenetic protein and canonical Wnt signaling pathways (21, 22, 24). Sostdc1 expression is highly expressed in skin, brain, and intestine as well as in skeletal muscles, kidney, lungs, and vasculature (21, 23, 24). Most recently, we found it also to be expressed in the bone periosteum and mesenchymal stem cells to support bone formation and fracture remodeling (27). In this study, we reveal Sostdc1’s cell-extrinsic roles in the regulation of NK cell maturation, Ly49 receptor expression, and cytotoxic function in the BM and spleen.

Sostdc1^−/− mice have been described (23, 27) and were both bred and transferred from Lawrence Livermore National Laboratories (LLNL) to University of California (UC) Merced to begin an independent breeding colony. C57B6/J (CD45.2⁺/5.2⁺) and B6.SJL-Ptprca Pepcb/BoyJ (CD45.1⁺/5.1⁺) and B6.129P2-B2mtm1Unc/J β-2 microglobulin knockout (KO) (β₂m^−/−) mice were obtained from The Jackson Laboratory. Mice of 28–38 wk of age and of both sexes were used. No differences between sexes nor mice from LLNL and UC Merced colonies have been observed. All mice were housed in conventional housing with autoclaved feed. Mice were euthanized by carbon dioxide asphyxiation followed by cervical dislocation. All animal procedures were approved by the UC Merced and LLNL Institutional Animal Care and Use Committees.

Isolation of spleen and BM cells were performed and stained for flow cytometry (FCM) as described (27). Abs against CD161 (also known as NK1.1, PK136), CD11b (M1/70), CD27 (LG.3A10), CD19 (6D5), CD3 (2C11), Gr1-Ly6C/G (Gr1), Ly49G2 (4D11), Ly49I (YL1-90), Ly49H (3D10), Ly49D (eBio4E5), CD45.2 (104), CD45.1 (A20), CD45 (30-F11), CD4 (GK1.5), CD8 (2.43), Ter119 (TER119), CD107a (1D4B), rat IgG2a κ isotype control (RTK2758), IFN-γ (XMG1.2), rat IgG1 κ isotype control (RTK2071), GM-CSF (MP1-22E9), granzyme B (GB11), perforin (eBioOMAK-D) IL-12/IL-23 p40 (C15.6), TNF-α (TN3-19.12), and BUV395 streptavidin, eFluor 780 fixable viability dye, eFluor 506 fixable viability dye were purchased from eBioscience, BioLegend, Miltenyi Biotec, and BD Biosciences. Staining of all cells included a preincubation step with unconjugated anti-CD16/32 (clone 2.4G2 or clone 93) mAb to prevent nonspecific binding of mAbs to FcγR. For extracellular staining, the cells were washed and incubated with a panel of mAbs for 15–20 min at 4°C or on ice and then washed again. Intracellular staining was performed using the BD Cytofix/Cytoperm Fixation/Permeabilization Kit (BD Biosciences) per the manufacturer’s instructions. To purify NK cells and NK cell subsets, enrichment of NK cells was first achieved by staining with biotinylated anti-“lineage” mixture (anti-CD3, -CD4, -CD8, -CD19, -Gr1, and -Ter119) followed by magnetic separation using EasySep Positive Selection Kit (STEMCELL Technologies), After enrichment, cells were stained with streptavidin-FITC, anti-CD27, and additional anti-CD3, -CD19, -Gr1, -CD45, -NK1.1, and -CD11b. Lineage-negative CD45⁺ NK1.1⁺ iNKs, tNKs, and mNKs were sorted on the FACSAria II (Becton Dickinson). Single-color stains were used for setting compensations, and gates were determined by historical data in addition to fluorescent-minus-one control stains. Flow cytometric data were acquired on the BD LSR II or FACSAria II cell sorter (Becton Dickinson). The data were analyzed using FlowJo version 7.6 or 10 (Tree Star).

Sostdc1^−/− and B6 control mice received 200 μg of polyinosinic-polycytidylic acid (poly[I:C]) (Sigma-Aldrich) via injection into the i.p. cavity. Thirty-six hours later, splenic cells from age- and sex-matched β₂m^−/− and β-2 microglobulin sufficient (β₂m^+/+) (wild-type [WT] or Sostdc1^−/−) mice were harvested and processed to a single-cell suspension in media (medium 199, 2% FCS, 2mM l-glutamine, 100 U/ml penicillin, 100 μg/ml streptomycin, 25 mM HEPES) and counted using a hemocytometer. β₂m^+/+ (WT or Sostdc1^−/−) and β₂m^−/− control target cells were stained with anti-CD45 conjugated to either allophycocyanin or PE in media for 20 min on ice. A total of 5 × 10⁶ β₂m^−/− stained splenic cells were mixed with 5 × 10⁶ β₂m^+/+ stained control cells at a 50:50 ratio, thus providing a method to track each target cell type by FCM. Stained cell target cell mixtures were injected i.v. by retro-orbital injection. Fourteen hours later, spleens from Sostdc1^−/− and WT recipients were then harvested and processed for FCM. NK cell lysis of targets was determined by FCM and calculated by the ratio of live β₂m^+/+ (WT), β₂m^+/+ (Sostdc1^−/−), and β₂m^−/− targets over WT control cells in the same mouse.

NK cells were enriched by magnetic bead sorting as described above. NK cell stimulation with IL-2 and feeder cells was performed as described (9). On day 4, spleen cells from β₂m^+/+ (WT) and β₂m^−/− mice were harvested and labeled with anti-CD45 allophycocyanin and used as targets in separate wells. NK cells and targets were cocultured at specific E:T ratios with a minimum of 1 × 10⁵ NK cells and 1 × 10⁵ targets per culture well. NK cell lysis of targets was determined by FCM and quantified using the ratio of live β₂m^−/− CD45-allophycocyanin–positive targets to WT control CD45-allophycocyanin–positive cells in the cocultures.

Splenic cells from Sostdc1^−/− and B6 mice were isolated, and 2 × 10⁶ cells were transferred to a flat-bottom plate in the presence of anti-CD107a or isotype control mAb. Cells were stimulated with PMA and ionomycin at a final concentration of 100 ng/ml and 1000 ng/ml and incubated at 37°C and 5% CO₂ conditions for 4 h (28). Protein transport inhibitors brefeldin A (BioLegend) and monensin (BioLegend) were added 1 h poststimulation at a final concentration of 1×. Cells were washed and cell surface stained for CD27, CD3, CD19, Gr1, NK1.1, CD11b, fixable viability dye, and Fc block. Cells were then fixed and stained intracellularly with anti–IFN-γ or isotype mAb control using the BD Cytofix/Cytoperm Kit (BD Biosciences) according to the manufacturer’s suggested protocol. Cells were analyzed using FCM on the LSR II (Becton Dickinson).

Age- and sex-matched Sostdc1^−/− and B6 control mice received 200 μg of poly(I:C) (Sigma-Aldrich) via injection into the i.p. cavity. Sixteen hours later, splenic cells were harvested, processed to a single-cell suspension in RPMI1640 media supplemented with 10% FBS 3, 0.09 mM nonessential amino acids, 2 mM l-glutamine, 1 mM sodium pyruvate, 100 U/ml penicillin, 100 μg of streptomycin, 0.025 mM BME, and 0.01 M HEPES buffer, and RBCs were lysed with ammonium-chloride-potassium buffer. A total of 5 × 10⁶ splenic cells were additionally stimulated in 96-well flat-bottom plates with 100 ng/ml of PMA and 1000 ng/ml of ionomycin and incubated at 37°C and 5% CO₂ conditions for 5 h. Brefeldin A (BioLegend) was added 1 h poststimulation at a final concentration of 1×. Cells were washed and stained with Abs against CD3, CD19, Gr1, NK1.1, fixable viability dye, and Fc block. Cells were then fixed and stained intracellularly with anti–GM-CSF, –TNF-α, –IFN-γ, –granzyme B, -perforin, or –isotype mAb control according to the BD Cytofix/Cytoperm Kit (BD Biosciences) manufacturer’s suggested protocol.

Five million splenic cells from Sostdc1^−/− and B6 control mice were stimulated in 96-well flat-bottom plates with 100 ng/ml LPS (Sigma-Aldrich) and incubated at 37°C with 5% CO₂ for 3 h with brefeldin A (BioLegend) at a final concentration of 1×. Cells were washed and stained with Abs specific for CD8α, CD11c, CD45, fixable viability dye, and Fc block. Cells were then fixed and stained intracellularly with anti–IL-12 or isotype mAb control according to the BD Cytofix/Cytoperm Kit (BD Biosciences) manufacturer’s suggested protocol.

Cells were pelleted and resuspended in RNeasy Lysis Buffer with 2-ME (Qiagen). Total RNA was purified using Qiagen RNeasy Mini Kit (Qiagen) according to manufacturer’s protocol. RNA concentration and purity was analyzed using a NanoDrop ND-1000 Spectrophotometer (Thermo Fisher Scientific). iScript cDNA Synthesis Kit was used (Bio-Rad Laboratories) according to the manufacturer’s protocol. Real-time quantitative PCR (qPCR) performed using the iTaq Universal SYBR Green Supermix kit (Bio-Rad Laboratories) and ran on a Stratagene Mx3000P thermocycler (Thermo Fisher Scientific) using the following conditions: 1 cycle at 9°C for 30 s, followed by 40 cycles of 95°C for 5 s and 60°C for 30 s, and a final cycle at 95°C for 1 min, 55°C for 30 s, and 95°C for 30 s to end the run. The PCR products were visualized on a 2% agarose gel and imaged under UV light using a ChemiDoc (Bio-Rad Laboratories) with SYBR Safe (Invitrogen) stain. The genes and primer sequences used are as follows: β-galactosidase: forward 5′-ACGGCCAGGACAGTCGTTTG-3′, reverse 5′-CCGCTCATCCGCCACATATC-3′; Sostdc1: forward 5′-CACCCTGAATCAAGCCAGGA-3′, reverse 5′-TAGCCTCCTCCGATCCAGTT-3′; Axin2: forward 5′-ACGCACTGACCGACGATTC-3′, reverse 5′-CCATGCGGTAAGGAGGGAC-3′; Myc: forward 5′-GCTGTTTGAAGGCTGGATTTC-3′, reverse 5′-GA TGAAATAGGGCTGTACGGAG-3′; Tcf7: forward 5′-AAGGTCATTGCTGAGTGCACAC-3′, reverse 5′-TGCATGCCACCTGCGAC-3′; Lef1: forward 5′-AAGGCGATCCCCAGAAG GAG-3′, reverse 5′-AGGGTGTTCTCTGGCCTTGT-3′; Eomesodermin (Eomes): forward 5′-TCCTAACAC TGGCTCCCACT-3′, reverse 5′-GTCACTTCCACGATGTGCAG-3′; Id2: forward 5′-GTCC TTGCAGGCATCTGAAT-3′, reverse 5′-TTCAACGTGTTCTCCTGGTG-3′; Tbx12: forward 5′-CAACCAGCACCAGACAGAGA-3′, reverse 5′-ACAAACATCCTGTAATGGCTTG-3′; and GAPDH: forward 5′-TCACCACCATGGAGAAGGC-3′, reverse 5′-GCTAAGCAGTTGGTGGTGCA-3′.

Whole BM cells were aseptically isolated from B6 (CD45.1 or CD45.2), WT, or Sostdc1^−/− (CD45.2) mice, and 5 × 10⁶ cells were transferred via retro-orbital injection into lethally (10 Gy) irradiated recipients 4 h after irradiation using a cesium irradiator. Mice were given neomycin-containing drinking water for 2 wk posttransfer. Chimeras were analyzed 14 wk posttransplant.

Student t test with a two-tailed distribution and with two-sample equal variance (homoscedastic test) was used to determine differences in means between groups using GraphPad Prism software. A p value < 0.05 was considered to be statistically significant.

Our previous studies demonstrated that femurs of Sostdc1^−/− mice display a 21% increase in BM cavity volume compared with WT controls (27). Consistent with this, the total BM cellularity of Sostdc1^−/− bones was increased (Fig. 1B). Sostdc1^−/− mice also displayed higher total splenic cell numbers (Fig. 1C). The increased cellularity suggested that the Sostdc1^−/− BM and spleen microenvironments may be altered and that immune cell development may also be affected by the loss of Sostdc1. To test this, we performed FCM and observed no differences in frequencies or absolute numbers of CD19⁺ B lymphocytes, CD3⁺ T lymphocytes, and CD11b⁺ Gr1⁺ granulocytes (data not shown). However, the frequency of CD11b⁺ Gr1⁻ cells in the Sostdc1^−/− spleen was reduced. To determine if the CD11b⁺ Gr1⁻ cells were monocytes or NK cells, we performed more detailed analysis with anti-NK1.1. Total NK (live, CD3⁻, CD19⁻, Gr1⁻, NK1.1⁺) frequencies were not affected, but total NK cell numbers were increased only in the BM of Sostdc1^−/− mice (Fig. 1D–G). We then investigated if lack of Sostdc1 affected NK cell maturation (Fig. 1A) and discovered that Sostdc1^−/− mice exhibit a partial block between the tNK (NK1.1⁺ CD11b⁺ CD27⁺) and mNK (NK1.1⁺ CD11b⁺ CD27⁻) cell stages in both the BM and spleen, as demonstrated by the increase in frequency and number of tNK cells in the BM (Fig. 1H, 1I) and spleen (Fig. 1J, 1K) and the decreased frequency of mNKs in the spleen. These data indicated that Sostdc1 is required for full developmental progression from the tNK to the mNK cell stages.

We also examined if NK cells in Sostdc1^−/− mice expressed different levels and distributions of inhibitory (Ly49I and Ly49G2) and activating (Ly49D and Ly49H) Ly49 receptors. FCM analysis of the Ly49 repertoire on iNKs, tNKs, and mNKs in Sostdc1^−/− mice revealed decreased frequencies of Ly49G2⁺ cells at all NK cell stages in the BM (Supplemental Fig. 1A–C) and spleen (Fig. 1L–N). In contrast, frequencies of Ly49H⁺ iNK and tNK cells in the Sostdc1^−/− BM and spleen were higher than controls, but the frequencies of Ly49H⁺ mNK cells were reduced in both tissues (although only statistically significant in the spleen). Similarly, frequencies of Ly49D⁺ in iNK and tNK cells were increased and reduced among the mNKs in the Sostdc1^−/− BM and spleen (Fig. 1L–N, Supplemental Fig. 1A–C). The frequencies of Ly49I⁺ in iNK and tNK cells were similar to controls, but frequencies of Ly49I⁺ mNK cells were decreased in Sostdc1^−/− mice (Fig. 1N, Supplemental Fig. 1C). The median fluorescent intensity of staining for Ly49G2 and Ly49H was reduced on Sostdc1^−/− BM mNK cells only, indicating a relatively minor effect of Sostdc1 on cell surface Ly49 receptor expression levels (Supplemental Fig. 1D–K).

Because it is theorized that NK cell activity is governed by the combined set of Ly49 receptors expressed on a given NK cell, we further compared the frequencies of WT and Sostdc1^−/− NK cells that express different combinations of Ly49 receptors (29) (Supplemental Fig. 2). Higher frequencies of iNK and tNK cells expressing more activating than inhibitory receptors (i.e., “activating repertoires”) were observed in Sostdc1^−/− mice (Supplemental Fig. 2D, 2E). However, lower frequencies of mNK cells with activating repertoires were observed (Supplemental Fig. 2F). Taken together, these data show the lack of Sostdc1 influences the Ly49 receptor repertoire.

The reduced frequency of splenic mNK with activating repertoires suggested that NK cell cytotoxicity in Sostdc1^−/− mice would be impaired. To determine if the alterations in NK Ly49 repertoire correlated with NK cell killing ability, we analyzed Sostdc1^−/− NK cell cytotoxicity with FCM-based in vivo and in vitro killing assays (Fig. 2, Supplemental Fig. 3A–E). β₂m^−/− cells express little to no cell surface MHC class I (MHC-I) molecules and therefore are sensitive targets for NK cell killing (30). To test NK cell killing in vivo, we preactivated NK cells in Sostdc1^−/− and WT control mice with poly(I:C) (31) (Fig. 2A) and challenged them with equal numbers of β₂m^−/− and β-2 microglobulin-sufficient (β₂m^+/+) target cells, each labeled with two different fluorochromes (Fig. 2B). β₂m^+/+ target cells from WT and Sostdc1^−/− mice were both included as negative “self” controls (Fig. 2C–E and data not shown). After 14 h of target cell challenge, we quantified the remaining β₂m^−/−, WT (β₂m^+/+), and Sostdc1^−/− (β₂m^+/+) targets by FCM to determine the frequency of live cells in each target population (Fig. 2A, 2C, 2D) and calculated the ratio of WT (β₂m^+/+), Sostdc1^−/− (β₂m^+/+), and β₂m^−/− targets in each setting (Fig. 2E). An increased proportion of β₂m^−/− targets remained in the Sostdc1^−/− mice compared with WT controls (Fig. 2E). We confirmed that there was no effect on fluorophore labeling on β₂m^−/− cell target killing with reciprocal labeling of targets of opposite fluorophore (Fig. 2C, 2D).

This result was confirmed using an in vitro NK cell killing assay (Supplemental Fig. 3) using enriched NK cells from Sostdc1^−/−and WT mice (32) challenged with fluorescently labeled β₂m^−/− or β₂m^+/+ targets for 4 h in E:T ratios of 1:1, 2:1, and 4:1 (Supplemental Fig. 3A–C). As expected, WT NK and Sostdc1^−/− NK cells did not lyse β₂m^+/+ targets at any E:T ratio (Supplemental Fig. 3D). However, as shown in Supplemental Fig. 3E, Sostdc1^−/− NK cells have reduced capacity to lyse β₂m^−/− targets, even at the highest 4:1 E:T ratio, indicating their hyporesponsiveness to β₂m^−/− targets. FCM analysis of Sostdc1^−/− NK cells to measure the cytokine IFN-γ after stimulation revealed comparable levels to WT controls (Supplemental Fig. 3F, 3G). Surprisingly, activated Sostdc1^−/− NK cells at all developmental stages expressed significantly increased levels of the degranulation marker CD107a (Supplemental Fig. 3H, 3I). Furthermore, the impaired killing ability of NK cells in Sostdc1^−/− mice is independent of the expression of NK cell proinflammatory and cytotoxic cytokines GM-CSF, TNF-α, and IFN-γ as well as granzyme B and perforin expression, as we confirmed their levels are similar between Sostdc1^−/− and WT NK cells (Fig. 2F–H). We further investigated if Sostdc1^−/− mice display decreased IL-12 expression in the spleen (33–35) and found no significant difference in expression between WT and Sostdc1^−/− IL-12–producing CD8-CD11c^(hi) and CD8⁺CD11c⁺ dendritic cells (DCs) (Supplemental Fig. 3J, 3K). Taken together, these results suggest that Sostdc1^−/− NK cell cytotoxicity is impaired despite their ability to produce comparable levels of IFN-γ, similar levels of inflammatory cytokine production, and evidence of elevated accumulation of cytotoxic granules at the cell surface and similar levels of IL-12 production in the microenvironment.

Given that Sostdc1 is a known antagonist to canonical Wnt signaling, we hypothesized that expression of canonical Wnt pathway transcription factors would be increased in Sostdc1^−/− NK cell subsets. We purified iNK, tNK, and mNK cells by FCM and analyzed expression of Wnt pathway genes Tcf7 (3, 36), Lef1 (37), Axin2 (38), and Myc (39) by real-time qPCR. Our results showed that relative to WT subsets, Sostdc1^−/− splenic tNK and mNK cells express significantly higher levels of Tcf7, and tNK cells also show significantly increased expression of Lef1 (Fig. 3A), consistent with our hypothesis. We did not observe any remarkable difference in Axin2 and Myc expression in any Sostdc1^−/− NK cell subset (Fig. 3A). Together, these results support a role for Wnt signaling by Tcf7 and Lef1 in NK cells of Sostdc1^−/− mice.

We also analyzed expression of transcription factors that govern NK cell maturation. T-box family members Eomes and T-box protein 21 (Tbx21) have been shown to play a crucial role in early iNK and mNK cell maturation (40, 41). Additionally, Eomes-deficient NK cells have reduced Ly49A, Ly49D, Ly49G2, and Ly49H frequencies (40). Inhibitor of DNA-binding 2 (Id2) is an early transcription factorinvolved in NK and innate lymphoid cell lineage commitment (41–43). Id2-deficient mice have fewer mNK cells and impaired cell killing in vitro (44). Because Sostdc1^−/− mice display a partial maturation block at the tNK cell stage and altered Ly49 receptor frequencies (Fig. 1), we hypothesized that we would observe decreased expression of Eomes, Tbx21, and Id2 at distinct NK cell stages (40–47). We found Id2 expression was decreased in iNK and tNK cells in Sostdc1^−/− mice (Fig. 3B). These results suggest a strong regulation of Id2 by Sostdc1 at early NK cell stages, whereas Eomes and Tbx21 are not regulated by Sostdc1.

To determine if and how Sostdc1 within specific microenvironmental cell types contributes to the partial block in NK cell maturation and changes in Ly49 repertoires, we performed whole BM transplantation experiments. To investigate if Sostdc1 in nonhematopoietic cells influenced NK cell development, we first transplanted whole BM cells from WT (CD45.1⁺/5.1⁺) donors into lethally irradiated Sostdc1^−/−(CD45.2⁺/5.2⁺) recipients to create WT→KO chimeras. WT(CD45.1⁺/5.1⁺)→WT(CD45.2⁺/5.2⁺) control chimeras were also prepared (Fig. 4A). Fourteen weeks post–BM transplantation, we analyzed donor-derived NK cell subsets and Ly49 receptor frequencies by FCM. Splenic NK cell numbers were increased (Fig. 4B), and WT→KO chimeras displayed a partial block between tNK and mNK cell stages in the spleen (Fig. 4C–E) and BM (Supplemental Fig. 4C–E), similar to the phenotype that was observed in the nontransplanted Sostdc1^−/− mice (Fig. 1H–K). Ly49H⁺ mNK cells were decreased in WT→KO, similar to nontransplanted Sostdc1^−/− mice in the spleen (Figs. 1N, 4F). In contrast, WT→KO chimeras contained decreased frequencies of splenic Ly49H⁺ iNK and tNK cells compared with WT→WT controls, a result that was the opposite of the increased frequencies of Ly49H⁺ cells within these NK subsets of nontransplanted Sostdc1^−/− mice (Figs. 1L, 1M, 4F). In addition, no differences in Ly49G2 and Ly49D subsets were observed between the WT→KO and control chimeras, a result that also differed from the nontransplanted Sostdc1^−/− mice in the spleen (Figs. 1L–N, 4F) and BM (Supplemental Fig. 4F). Taken together, these results suggested that Sostdc1 in nonhematopoietic cells controls progression from tNK to mNK stages and NK cellularity, but plays a smaller role in shaping the Ly49 receptor repertoire.

We next prepared reciprocal KO→WT chimeras (whole BM cells from Sostdc1^−/− (CD45.2⁺/5.2⁺) donors transplanted into lethally irradiated WT (CD45.1⁺/5.1⁺) recipient mice to determine how Sostdc1^−/− NK cells mature and if their Ly49 receptor frequency was changed in a Sostdc1-sufficient microenvironment (Fig. 5A). Remarkably, splenic NK cell numbers (Fig. 5B) and maturation (Fig. 5C–E) were not affected in KO→WT chimeras, in contrast to the nontransplanted Sostdc1^−/− mice and the WT→KO chimeras (Figs. 1H, 1I, 4B–E, respectively). However, the BM analysis showed an increase in NK cell numbers (Supplemental Fig. 4H) and a similar tNK cell accumulation as observed in the nontransplanted Sostdc1^−/− mice (Fig. 1H–K, Supplemental Fig. 4K). Furthermore, analysis of donor-derived Ly49-expressing NK cell subsets in the spleens in KO→WT chimeras demonstrated some similar patterns as nontransplanted Sostdc1^−/− mice, such as the increase in frequencies of Ly49H⁺ and Ly49D⁺ tNK cells and a decrease in the frequencies of Ly49G2⁺ iNK cells and decreased frequencies of Ly49G2⁺ and Ly49I⁺ mNK cells (Figs. 1L–N, 5F). However, higher frequencies of mNK cells expressing Ly49H and Ly49D were observed in the KO→WT spleens, whereas these populations were decreased in nontransplanted Sostdc1^−/− mice (Figs. 1L–N, 5F).

The Ly49 frequency patterns observed in the KO→WT chimeras strongly suggested that Sostdc1 in NK cells regulated the Ly49 repertoire in a cell-intrinsic fashion. To confirm this, we examined Sostdc1 expression in sorted WT iNK, tNK, and mNK cells by qPCR. Surprisingly, iNK, tNK, and mNK cells do not express Sostdc1 (Fig. 5G). To determine alternative possible hematopoietic sources of Sostdc1, we examined other populations, such as lineage^-Sca1⁺Kit⁺ (LSK), lineage^-Kit⁺Sca1⁻ (LK), and common lymphoid progenitors, NKT cells, macrophages, B cells, and granulocytes, which were all negative for Sostdc1 (Fig. 5G). Only CD4⁺ and CD8⁺ T cells displayed high levels of Sostdc1 expression (Fig. 5G). This expression pattern was confirmed in CD4 and CD8 T cells from Sostdc1^-/- mice using PCR for LacZ (27) (data not shown). Collectively, these results indicate that Sostdc1 does not regulate splenic NK cell development in a NK cell–intrinsic manner and identifies Sostdc1-positive T cells as putative NK “niche cells” that may contribute to shaping of the Ly49 repertoire (Fig. 6).

To our knowledge, we have uncovered novel roles of the Sostdc1 gene in NK cell maturation and function through two distinct mechanisms. Our working model is illustrated in Fig. 6. Our data support that Sostdc1 from two distinct sources, nonhematopoietic stromal cells and hematopoietic cells (in particular, CD4⁺ and CD8⁺ T lymphocytes), regulate NK cell maturation versus Ly49 receptor expression and frequencies, respectively, in somewhat independent manners, and that this occurs through the control of Wnt signaling activation. Our developmental and functional NK cytotoxicity assay results lead us to conclude that several NK niche cell populations exist that require Sostdc1 expression to produce a healthy NK cell repertoire that can distinguish between self and nonself.

The microenvironmental cytokines may influence the development and functions of NK cells in relation to the niche, stromal, and T cells. In the spleen, the abundance of microenvironment cytokines, such as IL-2, IL-12, IL-15, IL-18, and IL-21, maintain NK cells in a steady-state and promote cytotoxic function (48). IL-15 supports NK cell development and function and is secreted by hematopoietic cells such as DCs, macrophages, and monocytes as well as nonhematopoietic stromal cells (49, 50). Alternatively, in an inflammatory microenvironment, activated T cells and DCs secrete abundant sources of cytokines that enhance NK cell cytotoxic function, such as IL-2, IL-12, IL-18, and IL-21 (35, 48, 51). We show that the hyporesponsiveness of Sostdc1^−/− NK cells toward β₂m^−/− cell targets is not mediated by an inability of DCs to produce IL-12 in the spleen (Supplemental Fig. 3). We have also concluded that production of inflammatory cytokines IFN-γ, GM-CSF, and TNF-α by NK cells is not affected by loss of Sostdc1. Detailed investigation of other cytokines such as IL-2, IL-7, IL-15, and IL-21 in the microenvironment are necessary to fully determine how Sostdc1 by hematopoietic and nonhematopoietic cells influence NK cytotoxicity and development.

Perhaps the most surprising finding from our studies is that Sostdc1 is not expressed in NK cells themselves and that Ly49 receptor expression in the spleen is possibly controlled in a cell-extrinsic manner by CD4⁺ and CD8⁺ T cells. Evidence that T cells shape the NK Ly49 receptor profile exist, but the mechanisms underlying this are still unclear. Jeannet et al. (2006) show that TCRβδ^−/− mice have increased frequencies of Ly49G⁺ and Ly49I⁺ NK cells, but frequencies of Ly49D⁺ NK cells are similar to controls. RAG-1^−/− mice also have increased NK cell Ly49I⁺ frequencies but no changes in Ly49G⁺ or Ly49I⁺ populations (52). In Sostdc1^−/−→WT chimeras, in which Sostdc1 is absent from T cells, we have observed decreased frequencies of Ly49G and Ly49I and increased frequencies of Ly49D⁺ mNK cells (Fig. 5F), opposite and distinct from the pattern found in TCRβδ^−/− and RAG-1^−/− mice. Taken together, our data and the literature show that complete lack of T cells is not essential for NK cell development and suggest that a specific subset of T cells may regulate the NK cell repertoire. Recent studies have defined Sostdc1 as a marker of relatively rare memory PD-1⁺ CD4⁺ T cells (53, 54) and T follicular helper cells found in the Peyer’s Patches and peripheral lymph nodes (10) but only cite Sostdc1’s role on humoral immunity. We observed high expression of Sostdc1 in bulk sorted splenic CD4⁺ and CD8⁺ cells from the spleen, and we assume that most of the Sostdc1 expression is coming from PD1⁺ CD4⁺ T cells and T follicular helper cell subsets based on these published studies. We have not observed any obvious block in T cell development in the Sostdc1^−/− mice (data not shown). Undeniably, further experimentation is required to definitively connect the role of these T cell subsets in NK cell regulation and to determine whether the T cells are regulating Ly49 receptor expression via directly binding to NK cells or mediating their effects indirectly by secreting Sostdc1 in a paracrine fashion.

Wnt signaling has been well studied in the framework of hematopoietic stem cells to help promote proliferation, differentiation, and homeostasis (55, 56). It is now evident that canonical Wnt signaling plays a crucial role in the regulation of many immune cells (55). Based on the antagonist role of Sostdc1 on Wnt signaling and our discovery of increased Tcf7 and Lef1 expression in NK cells from Sostdc1^−/− mice, we conclude that canonical Wnt signaling plays a crucial role in NK cell development and function, particularly at the tNK cell stage. In the absence of Sostdc1, tNK cells are partially blocked in their maturation, express dysregulated Ly49 frequencies in the BM and spleen, and appear to be more reliant on Wnt activation. Our observations that one of the coactivators of canonical Wnt signaling, Tcf7, was significantly upregulated in tNK and mNK cells, and the secondary coactivator, Lef1, was significantly upregulated in the mNK cell stage, suggest that a critical period exists in which tNK cells require downregulation of Tcf7 and Lef1 to progress to the mNK cell stage. Our results and interpretation are consistent with a recent study that downregulation of Tcf1 (encoded by Tcf7) is required for full NK cell maturation and cytotoxicity (3).

Based on our current findings, we cannot rule out whether the impaired killing of β₂m^−/− targets by NK cells from Sostdc1^−/− mice is due to insufficient numbers of mNK cells, inefficient execution of the perforin and granzyme pathways, the dysregulation of Ly49 receptor frequencies among NK cells, or a combination of all of these possible mechanisms. NK cells at tNK and mNK stages express genes involved in cytotoxic function (11–13). It has been shown that Ly49 receptor expression is required for NK cell cytotoxicity (57), which is consistent with our observations that the splenic mNK cells in the Sostdc1^−/− mice contain decreased frequencies of all Ly49-expressing subsets and the killing ability of Sostdc1^−/− NK cells toward β₂m^−/− targets is poor. Because Sostdc1^−/− mice express increased proportions of tNK cells with an “activating” repertoire (58) and high CD107a levels, we would have expected enhanced target cell killing by the Sostdc1^−/− NK cells, but we observed them to be hyporesponsive. Taken together, our results suggest that NK cell cytotoxicity is universally disabled in the absence of Sostdc1 but is caused by distinct mechanisms in tNK and mNK cells. Further experiments, in which the killing ability of purified tNK and mNK cells from Sostdc1^−/− mice is specifically examined, are necessary to definitely demonstrate this. Additional work is also needed to dissect the specific roles of nonhematopoietic and hematopoietic cells on NK cell cytotoxicity. Understanding the details of the basic biology underlying the development and regulation of NK cell cytotoxicity and how these processes are quantitatively integrated could be applied to manipulate these processes in a controlled fashion to produce specific numbers of NK cells with enhanced killing ability and perhaps impact the production of NK cell–based cancer immunotherapies (1–6).

We thank the staff of the Department of Animal Research Services and the Flow Cytometry Core of the Stem Cell Instrumentation Foundry at University of California Merced for excellent animal care and technical support, and Dr. Anna Beaudin and Dr. Marcos E. García-Ojeda for their comments on the manuscript.

The authors have no financial conflicts of interest.