TREM2 promotes natural killer cell development in CD3−CD122+NK1.1+ pNK cells

Background Triggering receptor expressed on myeloid cells 2 (TREM2) signaling is considered to regulate anti-inflammatory responses in macrophages, dendritic cell maturation, osteoclast development, induction of obesity, and Alzheimer’s disease pathogenesis. However, little is known regarding the effect of TREM2 on natural killer (NK) cells. Results Here, we demonstrated for the first time that CD3−CD122+NK1.1+ precursor NK (pNK) cells expressed TREM2 and their population increased in TREM2-overexpressing transgenic (TREM2-TG) mice compared with that in female C57BL/6 J wild type (WT) mice. Both NK cell-activating receptors and NK cell-associated genes were expressed at higher levels in various tissues of TREM2-TG mice than in WT mice. In addition, bone marrow-derived hematopoietic stem cells (HSCs) of TREM2-TG mice (TG-HSCs) successfully differentiated into NK cells in vitro, with a higher yield from TG-HSCs than from WT-HSCs. In contrast, TREM2 signaling inhibition by TREM2-Ig or a phosphatidylinositol 3-kinase (PI3K) inhibitor affected the expression of the NK cell receptor repertoire and decreased the expression levels of NK cell-associated genes, resulting in significant impairment of NK cell differentiation. Moreover, in melanoma-bearing WT mice, injection of bone marrow cells from TREM2-TG mice exerted greater antitumor effects than that with cells from WT control mice. Conclusions Collectively, our data clearly showed that TREM2 promoted NK cell development and tumor regression, suggesting TREM2 as a new candidate for cancer immunotherapy. Supplementary Information The online version contains supplementary material available at 10.1186/s12865-021-00420-0.

Here, we demonstrated that overexpression of TREM2 promoted NK cell differentiation and enhanced their cytotoxicity toward tumor cells in vivo and in vitro. Conversely, treatment with TREM2-Ig or a PI3K inhibitor inhibited NK cell differentiation, suggesting that activation of the PI3K pathway by TREM2/DAP12 signaling plays a crucial role in both the differentiation and effector function of NK cells.

NK cell populations are increased in TREM2overexpressing transgenic (TREM2-TG) mice
To investigate the effect of TREM2 on NK cell development, we analyzed NK cell populations in previously generated TREM2-overexpressing transgenic (TREM2-TG) and wild type (WT) mice using flow cytometry [33]. The expression of NK receptor repertoires, NK1.1+ population percentage, and their absolute numbers in the spleen, BM, and liver were higher in TREM2-TG mice than in WT ( Fig. 1a and b). Furthermore, the absolute number of NK cells expressing the NKG2A/ NKG2C/NKG2E receptor was higher in the spleens (Additional file 1; Fig. S1, left panel) and livers (right panel) of TREM2-TG mice than that in those of WT mice. A slight increase in the BM of TREM2-TG mice (middle panel) was also observed. Similarly, the percentage and absolute number of Ly49C/F/H/I + and Ly49D + NK cells in the spleens and BMs of TREM2-TG mice were significantly higher than in those of WT ( Fig. 1b and Additional file 1, Fig. S1).

In vivo inhibition of TREM2 signaling reduces NK cell populations
To verify the effect of TREM2 on NK cell development in vivo, we inhibited TREM2 signaling in WT mice via intraperitoneal injection of a TREM2-Ig fusion protein or a humanized (hu)-Ig control. Three days after injection, the spleen, BM, and liver cells were isolated, and NK cell populations and NKspecific receptors expression were analyzed by flow cytometry. The frequency and absolute number of NK1.1 + cells, in the total splenocyte population, were reduced in the spleens of TREM2-Ig-injected mice, compared to hu-Ig-injected control mice (3.525% ± 0.32% vs. 5% ± 0.5%; Fig. 2a and Additional file 1, Fig.  S4). Similarly, the percentage and absolute number of NK1.1 + cells in the BM and liver of TREM2-Iginjected mice were lower than in hu-Ig-injected control mice (Fig. 2a). Furthermore, in the spleen, the percentage of NK1.1 + NKG2A/C/E + cells was lower in TREM2-Ig-injected mice (1.9%) than in hu-Ig-injected control mice (2.4%) (Additional file 1, Fig. S4, left panel). Both the NK1.1 + NKG2A/C/E + (0.6% vs. 1.1%) and NK1.1 + Ly49D + (0.5% vs. 1%) populations were reduced in the BMs of TREM2-Ig-injected mice when compared with the control group; the absolute number of NK1.1 + cells was also decreased by TREM2 signaling inhibition ( Fig. 2b and Additional file 1, Fig.  S4, middle panel). In addition, the frequency of NK1.1 + NKG2A/C/E + (4.9% vs. 6.3%) and NK1.1 + Ly49D + (2% vs. 2.4%) populations in the liver, as well as the absolute number of NK1.1 + cells, was lower in TREM2-Ig-injected mice than in control mice ( Fig. 2b and Additional file 1, Fig. S4, right panel). However, the absolute numbers of NK1.1 + NKA/C/E + , Ly49C/F/ H/I + , and Ly49D + cells were significantly decreased in the spleens (Fig. 2b, left panel) and BMs (Fig. 2b, middle panel), but not in the livers (Fig. 2b, right panel), of TREM-Ig-injected mice, when compared with those values found in the hu-Ig-injected mice ( Fig. 2b, right panel). These data collectively indicate that in vivo inhibition of TREM2 signaling by TREM2-Ig decreases the number of NK cells and the expression of their signature surface receptors.

TREM2 promotes NK cell differentiation and direct cytotoxic activity in vitro
Our results showed that TREM2 signaling increased the number of NK cells in vivo. However, this is not sufficient to conclude that TREM2 enhances commitment to the NK cell fate and differentiation of the NK cell lineage. Therefore, to determine the effects of TREM2 on the differentiation of NK cells, we isolated c-kit + Lin − HSCs from BMs of WT and TREM2-TG mice and differentiated them into pNK and mNK cells in vitro. During NK cell differentiation, pNK cells were treated with TREM2-Ig or hu-Ig to inhibit TREM2 signaling. As a result, the percentage and absolute numbers of NK1.1 + NKG2A/C/E + cells were approximately 2-fold higher in the mNK cells derived from hu-Ig-treated pNK cells of TREM2-TG mice (25%) than in their counterparts derived from hu-Ig-treated pNK cells of WT mice (14%) (Fig. 3a). However, the NK1.1 + NKG2A/C/E + cell population dramatically decreased when WT-pNK (46.8% ± 0.8 to 10.6% ± 2.7%) and TREM2-TG-pNK (53.8% ± 2.8 to 13.3% ± 0.88%) cells were treated with TREM2-Ig during differentiation ( Fig. 3a and b, upper panel). The absolute number of NK1.1 + cells decreased by 3.97-(WT) and 4.77-folds (TREM2-TG) after treatment with TREM2-Ig (Fig. 3c). Additionally, the percentage and absolute number of NK cells expressing either Ly49C/F/H/ I or Ly49D were reduced in both NK cells derived from WT-pNK (6.7 to 1.8% and 3.9 to 2.2%, respectively) and TREM2-pNK (7.3 to 4.7% and 7.1 to 3.1%, respectively) after TREM2-Ig treatment during differentiation ( Fig. 3a and c). In contrast, the difference between the total number of cells differentiated from TREM2-TG-pNK or WT-pNK cells was not significative, regardless of TREM2-Ig treatment (Fig. 3b, lower panel). Subsequently, we performed RT-PCR analyses to identify the expression of NK cell-associated genes that are regulated by TREM2 ( Fig. 3d and Additional file 1, Fig. S5). NK cells that differentiated from TREM2-TG-pNK cells treated with hu-Ig showed increased Ifng (3.97 ± 0.63fold) and Fas ligand (Faslg) (4.5 ± 0.3-fold) expression compared with those NK cells derived from WT-pNK hu-Ig-treated cells. We also observed increased expression levels of Gzmb (1.8 ± 0.36-fold), Prf1 (1.25 ± 0.05fold), and TNF-related apoptosis-inducing ligand (Trail) (1.9 ± 0.22-fold) in NK cells derived from TREM2-TG-pNK cells treated with hu-Ig, when compared with those derived from WT-pNK hu-Ig treated cells. In contrast, the expression levels of these genes were reduced in NK cells derived from both WT-pNK and TG-pNK cells, when TREM2 signaling was inhibited by TREM2-Ig (Fig. 3d). Moreover, the expression levels of E4bp4, Id2, CD122, and CD123 increased in TREM2-TG-pNK-derived NK cells when compared to those in WT-pNKderived NK cells treated with hu-Ig, while in differentiated NK cells, it decreased significantly after TREM2-Ig treatment (Additional file 1, Fig. S5). Next, to elucidate the effect of TREM2 on the cytotoxicity of differentiated NK cells, we performed an in vitro NK cell cytotoxicity assay. The mNK cells differentiated from TREM2 TG-pNK treated with TREM2-Ig or hu-Ig showed significantly higher specific cytolytic activity (32 ± 1.2% and 41 ± 2.4%, respectively) against target cells at a 20:1 (Effector: Target) ratio (Fig. 3e) than those differentiated from WT-pNK cells treated with hu-Ig. At a 5:1 ratio, we observed that NK cells differentiated from TREM2 TG-pNK hu-Ig or TREM2-Ig treated cells showed a significative higher level of target cell death than cells from WT-pNK that received the same treatments (Fig. 3e). Therefore, these data suggest that TREM2 signaling increases in vitro cytotoxic activity of differentiated NK cells.

TREM2 signaling inhibits tumor progression
As described above, the inhibition of TREM2 signaling pathway by TREM2-Ig reduced NK cell receptor and NK cell-associated gene expression, as well as the absolute number of NK cells in vitro (Fig. 3). To confirm whether TREM2 affects tumor progression in vivo, we injected TREM2-TG or WT mice with B16F10 melanoma cells after intraperitoneal injection of hu-Ig or TREM2-Ig. As shown in Fig. 4a, on day 25, the tumor volume in WT mice treated with TREM2-Ig (WT + TREM2-Ig) was significantly higher than that in WT mice treated with hu-Ig (WT + hu-Ig), and these differences became even more prominent after day 25.  . 4b). Furthermore, the number of metastatic melanomas in the lungs of TREM2-Ig-injected WT mice was higher (24 ± 4, B16F10 cell spots) than that of hu-Iginjected WT mice (3 ± 1, B16F12 cell spots) (Fig. 4c). Surprisingly, B16F10 melanoma cells were rarely observed in the lungs of hu-Ig-injected TREM2-TG mice, whereas melanoma cells were apparent in the lungs of TREM2-Ig-injected TREM2-TG mice (19 ± 1, B16F10 cell spots).

Adoptive transfer of TREM2-TG BM cells promotes tumor regression
As mentioned above, TREM2-TG mice showed a significantly lower tumor volume and rare metastatic tumor spots compared with WT mice when they were injected with B16F10 melanoma cells. This may be related to the effects of TREM2-overexpressing monocytes/macrophages or DCs, which secrete cytokines and indirectly activate T cells and NK cells. To investigate whether  (Fig. 5a). These data indicated that TREM2-TG mice have a larger NK cell population than WT mice. Then, we used an in vivo tumor model to determine whether TREM2 signaling affects the antitumor effect of BM-derived immune cells. We subcutaneously injected B16-F10 melanoma cells into WT mice (CD45.1) The tumor volume measured 21 days post-inoculation in mice transplanted with a TREM2-TG-BM was lower (TG to WT, 1079 ± 221.5 mm3) than that of mice that received a WT-BM (WT to WT controls, 3122.7 ± 1269 mm3) (Fig. 5b). Furthermore, 27 days post-inoculation, the survival rate (75%) of tumor-bearing mice transplanted with a TREM2-TG-BM was significantly higher than that of WT-BM-transplanted mice (0%) (Fig. 5c).

TREM2 regulates NK cell differentiation via PI3K signaling
TREM2-DAP12 signaling, triggered by TREM2 ligand binding, may promote or inhibit proinflammatory responses, induce obesity [33], and mediate neurodegeneration [34,38]. DAP12, an adaptor protein of TREM2, mediates downstream signaling via the cytoplasmic ITAM domain, which recruits SYK and activates PI3K, phospholipase C, and Vav signaling cascades [42]. To investigate how TREM2 signaling regulates NK cell differentiation, we treated pNK cells differentiated from WT-HSCs or TG-HSCs with the PI3K inhibitor Ly294002 or with dimethyl sulfoxide (DMSO) as control vehicle during their differentiation into mNK cells. After 14 days, differentiated mNK cells were stained with NK-specific markers and analyzed via flow cytometry (Additional file 1, Fig. S6A). In the absence of the PI3K inhibitor, population of NK1.1 + NKG2ACE + cells differentiated from TREM2-TG-pNK cells Additional file 1, Fig. S6A, lower panel) was 2-fold higher than that of NK cells differentiated from WT-pNK cells (Fig. S6A, upper panel). NK1.1 + NKG2ACE + cell populations derived from both WT-and TREM2-TG-pNK cells decreased (10-fold) after Ly294002 treatment during NK cell maturation. We also analyzed the expression of NK cell-associated genes in mNK cells differentiated in the presence or absence of PI3K inhibitor. The expression levels of Ifng, Prf1, and Gzmb increased by 4-to 5-fold in mNK cells differentiated from TREM2-TG-pNK cells compared to mNK cells differentiated from WT-HSCs. Similarly, the expression levels of Faslg, Trail, and IL-15Rα were higher in mNK cells differentiated from TREM2-TG-pNK cells than those in cells differentiated from WT-pNK cells. With the exception of E4bp4 and IL-15Rα, NK cell-related gene expression levels significantly decreased in mNK cells treated with Ly49294002 during NK cell differentiation (Additional file 1, Fig. S6). In particular, the expression level of Id2 decreased by more than 2-fold in mNK cells after treatment with Ly294002.

Discussion
TREMs have emerged as critical immune regulators that modulate inflammatory responses in macrophages, glial cells, and DCs [41,[43][44][45]. Recently, several groups reported TREM2 as a novel tumor suppressor in colorectal and hepatocellular carcinoma [40,46]. However, the The survival rate of tumor-bearing WT and TREM2-TG mice. Seven mice per group were used. Survival probabilities were analyzed using the Kaplan-Meier method. The significance of differences between groups assessed using the log-rank test. All statistical tests were two-sided, with *P < 0.05 taken to indicate significance. Significance of difference between samples was determined using two-way ANOVA analysis with Bonferroni posthoc test. WT + hu-Ig vs. TG + hu-Ig (**P < 0.01), WT + hu-Ig vs. WT + TREM2-Ig (*P < 0.05), and TG + hu-Ig vs. TG + TREM2-Ig ( †P < 0.05, † † †P < 0.001) were compared. c Representative photographs of lungs with metastatic colonies from mice of each group (top panel). Graph quantitates the total number of metastatic colonies in the lungs of each group treated with hu-Ig or TREM2-Ig (bottom panel). WT + hu-Ig vs. TG + hu-Ig ( # P < 0.01) WT + hu-Ig vs. WT + TREM2-Ig (***P < 0.05), and TG + hu-Ig vs. TG + TREM2-Ig ( † † †P < 0.001) were compared. Significance of difference between samples was determined using two-way ANOVA analysis with Bonferroni posthoc test  [47], such as difficulties in NK cell expansion ex vivo and the severe side effects observed after IL-2 treatment for NK cell activation [40]. Therefore, it is necessary to find new ways to regulate NK cell function. It is known that NK cells are modulated by their activating receptors, which present binding motifs for the adaptor protein DAP12. The structure of TREM2 is largely similar to that of other NK cell receptors that transmit intracellular signals via DAP12, although its function in NK cells remains unclear.
In this study, we found that TREM2 is expressed in CD3 − CD122 + NK1.1 + pNK cells and that NK cell population abundance in the BMs of TREM2-TG mice was higher than that in the BMs of WT mice. It was recently reported that liver lymphocytes express DAP12, as well as low levels of TREM2; this information supports the findings of the current study [48].
The late NK cell maturation stage (CD3 − NK1.1 + ) can be subdivided into four distinct subsets: CD27 lo Mac-1 lo , CD27 hi Mac-1 lo , CD27 hi Mac-1 hi , and CD27 lo Mac-1 hi [49][50][51]. The intermediate CD27 hi Mac-1 hi population shows the strongest cytotoxicity and cytokine secretion [52], and it has a higher proliferation potential and an enhanced ability to interact with DCs [51]. Conversely, CD27 lo Mac-1 hi NK cells, the most abundant mNK cells, are effective killer cells in vivo, and are particularly effective against MHC class I-negative tumor cells [49]. According to our data, TREM2 overexpression affected CD27 hi Mac-1 lo , CD27 hi Mac-1 hi , and CD27 lo Mac-1 hi populations in the BM, suggesting that TREM2 promoted NK cell maturation in the BM with strong expression of TREM2 ligands (Additional file 1, Fig. S7). In contrast, in the peripheral blood, the number of NK1.1 + CD3 − cells was higher in TREM2-TG mice (8.2%) than in WT mice (5.7%), whereas NK1.1 + CD3 − CD27 hi Mac-1 hi cell population presence was similar in the two groups of mice (Additional file 1, Fig. S7B). These data suggested that TREM2 did not affect NK cell maturation in the peripheral blood, although more NK cells were released into the blood in TREM2-TG mice, as NK cell maturation was promoted in the BMs of TREM2-TG mice.
In addition, we established a tumor-bearing mouse model to demonstrate that the NK cell population increased via TREM2 overexpression, which reduced tumor progression. The tumor volume in tumor-bearing WT mice transplanted with BMs from TREM2-TG mice was lower than that in WT mice transplanted with BMs from WT mice. The metastasis of tumor cells in the lung tissue was reduced, and the survival rate of WT mice transplanted with BMs from TREM2-TG mice was higher than that of mice transplanted with BMs from WT mice. NK cells promote the maturation of DCs via IFN-γ, an important proinflammatory cytokine [53], while DCs stimulate NK cytotoxicity and cytokine secretion via IL-12 [54]. This bidirectional crosstalk between NK cells and DCs is an important mechanism in innate and adaptive immune responses [55,56]. Therefore, TREM2 reduces tumor progression in vivo by directly improving NK cell cytotoxicity, and TREM2-overexpressing DCs and macrophages could develop and stimulate NK cell function in vivo.
We co-cultured pNK cells derived from WT-HSCs and TG-HSCs with OP9 stromal cells, which support hematopoiesis by secreting growth factors, to induce further differentiation into mNK cells. Bartosz et al. [57] reported that NK cells may be derived from myeloid progenitors. Thus, we hypothesized that TREM2 in pNK cells or myeloid progenitors enhanced both the differentiation of NK cells in vitro and their cytotoxicity in the presence of OP9 cells. Recently, apolipoprotein E has been reported as a ligand of TREM2, although this is now considered controversial [58]. Consequently, the identification of TREM2 ligands is still necessary in order to develop NK cell therapies.
Previous studies have demonstrated that PI3K, and not PLC-γ, plays a critical role in the development of mNK cells [59]; moreover, while the absence of PLC-γ does not disrupt NK cell development, it causes defects in NK cell cytotoxicity [60,61]. However, Tassi et al. [62] have demonstrated that PLC-γ2 is crucial for the development of the NK cell receptor repertoire. In the current study, expression of the NKG2A/NKG2C/NKG2E receptor in mNK cells, differentiated from WT-pNK or TREM2-TG-pNK cells, decreased significantly from 40.68 to 8.96% after PLC-γ inhibitor treatment (Additional file 1, Fig. S8A). The expression of Id2 was upregulated by TREM2 in mNK cells differentiated from WT-pNK or TREM2-TG-pNK cells, and it was downregulated upon PI3K inhibitor treatment. Moreover, the expression of E4bp4, an essential factor for NK cell development [18], was upregulated by TREM2 in NK cells differentiated from pNK cells and was downregulated following PLC-γ inhibitor treatment (Additional file 1, Fig. S8B), indicating that TREM2 influenced NK cell differentiation not only via the PI3K signaling pathway but also via PLC-γ signaling.

Conclusions
In conclusion, we demonstrated that TREM2 played an important role not only in myeloid cells but also in CD3 − CD122 + NK1.1 + pNK cells. Furthermore, TREM2 promoted NK cell differentiation, as well as the expression of NK cell receptor repertoires and cytokines, suggesting that TREM2 might be an effective candidate for new NK cell therapies.

Mice
Five to seven-week-old female C57BL/6 J WT and TREM2-TG mice were used in this study as described previously [33]. The TREM2-TG mice were generated using pcDNA3.1(+) expression vector containing pCMV promoter. Overexpression of the TREM2 gene was observed in all tissues. All animal experiments were carried out following the guidelines of the Institutional Animal Care Committee of Chonnam National University (CNU IACUC-YB-2017-19).

Differentiation of HSCs into mNK cells
Murine HSCs were sorted from BM cell populations by negative or positive selection using a magnetic-activated cell sorter (MACS), as described previously [10]. Briefly, total BM cell samples were prepared by flushing the femurs from C57BL/6 mice, followed by filtration through a 70-μm cell strainer (Falcon, San Jose, CA, USA). Total BM samples were cleared of erythrocytes with Erythrocyte Lysis Buffer (Sigma-Aldrich, St. Louis, MO, USA) treatment. The suspensions of single BM cells were labeled using a cocktail of biotinylated antibodies against lineage (Lin + ) markers (CD11b, Gr-1, B220, NK1.1, CD2, and TER-119), which were then incubated on streptavidinmagnetic beads. The samples were depleted of magnetically labeled Lin + cells by retention on CS column beads in the magnetic field of a VarioMACS Separator (Miltenyi Biotec, Sunnyvale, CA, USA). c-Kit + cells among the Lin − cells were positively selected with magnetic beadconjugated antibodies against c-kit, and then, the cell suspension was run through an MS magnetic column Separator (Miltenyi Biotec, Sunnyvale, CA, USA).
HSCs were stimulated to differentiate into NK cells, as described previously [11]. In brief, purified Lin − c-kit + HSCs were plated on a 24-well plate (Corning, ME, USA) at 1 × 10 6 cells/well and cultured in RPMI medium supplemented with a mixture of IL-7 (0.5 ng/mL), stem cell factor (30 ng/mL), Flt3-L (50 ng/mL), indomethacin (20 μg/mL), and gentamycin (20 μg/mL) at 37°C and 5% CO 2 . Three days later, half of the culture supernatant was removed and replaced with fresh medium containing the same cytokines. After 7 days, the cells were cocultured with or without OP9 stromal cells (American Type Culture Collection, Manassas, VA, USA) in the presence of mouse IL-15 (20 ng/mL). Three days later, half of the culture medium was changed with fresh medium containing the same cytokines, and the cells were cultured for an additional 7 days. To determine purity, the cells were stained on days 0, 7, and 14 with stage-specific antibodies during the differentiation of NK cells and analyzed via flow cytometry.

Isolation of NK1.1 + cells from the spleen
Splenocytes were isolated from the spleens of the mice, and the cell suspension was filtered through a 20-μm cell strainer. After removal of erythrocytes via treatment with erythrocyte lysis buffer (Sigma-Aldrich, St. Louis, MO, USA), the single cells in suspension were first incubated with a biotinylated antibody against NK1.1 (BD Pharmingen, San Diego, CA, USA), followed by incubation with streptavidin-magnetic beads. NK1.1 + cells were then purified using MACS (Miltenyi Biotec) according to the manufacturer's instructions.

Flow cytometry analysis
To determine the developmental status of NK cells differentiated from HSCs of WT and TREM2-TG mice, we performed flow cytometry analysis of HSCs, pNKs, and mNKs co-cultured with OP9 cells using antibodies against the markers with stage-specific expression during the differentiation of NK cells. In brief, HSCs were stained with 1 μL of fluorescein isothiocyanate (FITC)-conjugated anti-c-kit, phycoerythrin (PE)-conjugated anti-Sca-1, and biotin/streptavidin/cytochrome-conjugated anti-IL-7Rα. pNKs were stained with 1 μL of FITC-conjugated anti-CD122 and 0.5 μL of PE-conjugated anti-NK1.1 antibodies. mNKs were stained with FITC-conjugated anti-NKG2A/C/E, anti-Ly49C/F/H/I, and anti-Ly49D and PE-conjugated anti-NK1.1 antibodies. The cells were incubated with the antibodies for 30 min on ice and then washed twice with staining buffer (phosphatebuffered saline containing 3% fetal bovine serum and 0.1% NaN 3 ). The cells were analyzed using a FACS Calibur flow cytometer (BD Bioscience, San Jose, CA, USA) and Cell Quest software. The data shown in the histograms or dot plots are representative of replicates.

BM adoptive transfer
WT (CD45.1) recipients were irradiated with 6.5 Gy, followed by injection with WT (CD45.2) or TREM2-TG (CD45.2) BM cells (1 × 10 6 ) intravenously. Four to eight weeks after the cells were transplanted, the BM and spleen were harvested, and single-cell suspensions were prepared as described above. Erythrocytes were lysed, lymphoid cell populations were counted, and the number of NK cells was assessed using flow cytometry with antibodies against NK1.1, CD45.1, or CD45.2.

Tumor models
To determine the tumor volume and survival rate, we subcutaneously injected B16F10 melanoma cells (5 × 10 5 cells/mouse) (ATCC, VA, USA) into the left flank of WT mice, TREM2-TG mice, BM-transplanted WT mice, and mice intraperitoneally injected with TREM2-Ig or hu-Ig (as a control). After B16F10 cell injection, the tumor volume was measured every 2 days.

RT-PCR
Total cellular RNA was extracted using Trizol B reagent (Tel-Test, Friendswood, TX, USA) according to the manufacturer's instructions. Aliquots of total RNA were transcribed into cDNA at 37°C for 1 h in a total reaction volume of 20 μL with 2.5 U of Moloney murine leukemia virus reverse transcriptase (Roche, Mannheim, Germany). Reversetranscribed cDNA was added to a PCR mixture consisting of 10× PCR buffer, 0.2 mM dNTP, 0.5 U Taq DNA polymerase (Bioneer, Daejeon, Korea), and 10 pmol of primers for each gene. For β-actin amplification, 27 cycles were performed, and for all other genes, 30 or 35 cycles were performed. The amplification profile included denaturation at 95°C for 1 min, primer annealing at 55°C for 1 min, and extension at 72°C for 10 min. PCR products were electrophoresed and visualized via ethidium bromide staining.

Cytotoxicity assay
The lactate dehydrogenase-release assay kit (Promega, WI, USA) was used to measure the cytotoxicity of NK cells, according to the manufacturer's instructions. In brief, NK cells were stimulated with 20 ng/mL of recombinant murine IL-2 for 48 h, washed twice with phosphate-buffered saline, and seeded into 96-well round-bottom microtiter tissue culture plates at various effector:target cell ratios. Target cell samples (1 × 10 4 cells per well) were tested in triplicate. The cells were incubated for 4 h at 37°C in a 5% CO 2 humidified incubator. Culture supernatants (50 μL) were then collected and combined with 50 μL of the substrate. The plates were covered with aluminum foil for protection against light and incubated at room temperature for 30 min, after which 50 μL of stop solution was added to each well. Absorbance at 490 nm was measured within 1 h of adding the stop solution. The results are expressed as the percentage of specific release based on the following formula: percent specific release = [(experimental release -spontaneous release)/(maximum release -spontaneous release)] × 100.

Statistical analysis
All values are expressed as mean ± standard error of the mean (SEM). All experiments were repeated at least three times, independently. Student's t-test and analysis of variance were performed using GraphPad Prism 5 (San Diego, CA, USA). Differences were considered statistically significant at P < 0.05.
Additional file 1: Fig. S1. The number of NK1.1 + cells, from TREM2-TG mice, expressing NK cell receptors is higher than in WT mice. Graphs show the absolute number of cells that express the NK cell receptors in the spleen, BM, and liver of WT (opened bar) and TREM2-TG mice (solid bar). Data are shown as mean ± SED of three independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student's t-test. Fig. S2. TREM2-TG mice show enhanced NK cell cytotoxicity. (A) Real-time qPCR analysis to determine expression of the TREM2, Ifng, Prf1, and Gzmb (granzyme B) mRNAs in splenic NK1.1 + cells of WT or TREM2-TG mice. (B) LDH assay to measure cytotoxicity of NK1.1 + cells purified from splenocytes harvested from WT (open square) and TREM2-TG (solid circle) mice. Three independent experiments were performed (A-B). *P < 0.05, **P < 0.01, and ***P < 0.001 by Student's t-test. Fig. S3. CD4 + T cell, CD8 + T cell, and B220 + B cell frequency and absolute number in WT and TREM2-TG mice are similar. Percentage (A) and absolute number (B) of CD4, CD8, and B cells in spleen of WT and TREM2-TG mice determined by flow cytometry (N = 5). Three independent experiments were performed, but no significant differences were observed between WT and TG mice group. Fig.  S4. Inhibition of TREM2 signaling reduces the NK cell pool in vivo. Representative flow cytometry plots of expression of NK cell-specific receptors (NKG2A/C/E, Ly49C/F/H/I, and Ly49D) on surface of cells isolated from the spleen, BM, and liver of WT mice injected (i.p.) with 100 μg of TREM2-Ig or hu-Ig (control) twice per week for 4 weeks. Fig. S5. TREM2 signaling enhances NK cell-related gene expression in differentiated NK cells in vitro. Quantitative real-time PCR analysis of NK cell-associated genes using mRNA isolated from mNK cells derived from WT or TREM2-TG HSCs. Data are shown as mean ± SED of three independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001 vs. WT+ hu-Ig, and †P < 0.05, † †P < 0.01, † † †P < 0.01 vs. TG + hu-Ig, based on two-way ANOVA with Bonferroni post-hoc test. Received: 15 December 2020 Accepted: 6 April 2021