Unique progenitors in mouse lymph node develop into CD127⁺ NK cells: thymus-dependent and thymus-independent pathways

Natural killer (NK) cells are lymphocytes of the innate immune system. They rapidly respond to tumors, viral infections, and allogeneic BM grafts by direct cytotoxicity and cytokine production.^1-3  Mouse NK cells are identified by cell surface markers, including NK1.1 (NKRP1C), CD49b (Dx5), and NKp46.⁴  They also express multiple inhibitory receptors, including the Ly49 family and CD94-NKG2A heterodimer, which recognize classic and nonclassic MHC class I, respectively,⁵  as well as the stimulatory receptor NKG2D recognizing stress-induced ligands.⁶  NK cells acquire these receptors as they develop from hematopoietic stem cells in the BM.⁷  Committed NK cell progenitors (NKPs) in the BM have been identified by the lack of the mature NK-cell markers (NK1.1 and DX5) and the expression of the IL-2/IL-15 receptor β chain (CD122).^8,9  They become immature NK cells expressing NKG2D, NK1.1, CD94-NKG2A, NKp46, and CD127 (IL-7Rα). Immature NK cells are neither cytotoxic nor do they produce IFNγ.⁸  As they further mature, they acquire the pan-NK cell marker CD49b (Dx5) and Ly49 as well as cytotoxicity and cytokine production capacities.¹⁰  Mature NK cells are heterogeneous and can be divided into subsets on the basis of their cell-surface receptors, including CD11b (Mac-1) and CD27.¹¹  Although the relationship between different subsets of mature NK cells is still unclear, they are thought to indicate further maturation of NK cells.^11,12 

Most studies on mouse NK cells focus on spleen NK cells, and it is generally thought that they develop in the BM. However, NK cells are widely distributed, and those in various tissues differ from spleen NK cells in receptor expression patterns. Of particular interest are NK cells in the lymph node (LN). Approximately 20% of LN NK cells, whereas only ∼ 5% of splenic NK cells, express CD127 (α-chain of the IL-7 receptor). CD127⁺ LN NK cells have less Ly49 receptor expression, produce larger amounts of IFNγ, and are less cytolytic than CD127⁻ NK cells.¹³  Most thymic NK cells are CD127⁺. Moreover, transplantation of newborn mouse thymus results in donor-derived CD127⁺ NK cells in the recipient mice, and athymic nude mice have significantly less CD127⁺ NK cells.¹³  Therefore, it has been suggested that CD127⁺ NK cells derive from the thymus.

We have previously reported that TCRγ genes are rearranged in some mouse NK cells.¹⁴  Although TCRγ gene rearrangement is rare (< 5%) among NK cells in most tissues, ∼ 20% of LN NK cells have rearranged TCRγ genes. Furthermore, those NK cells are undetectable in athymic mice, indicating that they are thymus dependent.¹⁵  Although CD127⁺ NK cells in the LN are thought to be thymus derived, TCRγ gene rearrangement is detectable at similar levels in both CD127⁺ and CD127⁻ LN NK cells.¹⁵  Thus, the relationship between CD127⁺ LN NK cells and those with rearranged TCRγ genes has been unclear.

In this study, we reexamined CD127⁺ NK cells in the LN and found that athymic mouse LNs have normal percentages of CD127⁺ NK cells, indicating that most CD127⁺ LN NK cells are thymus independent. We then found a population of NKPs in the LN cells that does not express mature lymphoid and myeloid cell markers (lineage markers; Lin) and CD122 but express CD49b and CD127. They differentiate into CD127⁺ NK cells in vitro as well as in vivo. Our results suggest that at least some CD127⁺ LN NK cells develop from novel Lin⁻CD49b⁺CD127⁺ NKPs in the LN.

C57BL/6J (B6), B6-Ly5.1-Pep3b (B6.Pep3b), and NOD/SCID/IL-2Rγ-deficient (NSG) mice were bred and maintained in the Animal Resource Center of the British Columbia Cancer Research Center. Recombination activating gene 1 (RAG1)–deficient and Foxn1-deficient (nude) mice of the B6 background were obtained from The Jackson Laboratory. All mice were kept under specific pathogen-free conditions in the Animal Resource Center of the British Columbia Cancer Research Center. All animal use was approved by the animal care committee of the University of British Columbia, and animals were maintained and killed under humane conditions in accordance with the guidelines of the Canadian Council on Animal Care.

Single-cell suspensions prepared from spleen, LN, BM, and blood; erythrocytes were analyzed by flow cytometer as described.¹⁵  The following FITC, PE, R-PE-cyanine 7 (Cy7), peridinin chlorophyll protein complex (PerCP)–Cy5.5, or allophycocyanin (APC)–conjugated monoclonal Abs (mAbs; BD Biosciences) were used (clone names given in parentheses): anti-CD3ϵ (145-2C11), anti-CD4 (GK1.5), anti-CD8α (53-6.7), anti-CD11b (M1/70), anti-CD11c (HL3), anti-CD25 (PC61), anti-CD45.2 (104), anti-CD49b (HMα2 and Dx5), anti-CD117 (2B8), anti-CD122 (TM-β1), anti-B220 (anti-RA3-6B2), TCRβ (H57-597), anti-TCRγδ (GL3), anti-Ly49D (4E5), Ly49G (4D11), anti–Ter-119 (Ly76), anti–Gr-1 (RB6-8C5), and anti-NK1.1 (PK136); from eBioscience: anti-CD24 (M1/69), anti-CD127 (A7R34), anti–stem cell antigen-1 (Sca-1; D7), and anti-NKp46 (29A1.4). Anti-Ly49H mAb (3D19) was a gift from Dr W. Yokoyama (Washington University, St Louis, MO).

Genomic PCR to determine TCRγ rearrangement was carried out as described previously¹⁴  with modified primers. The following primers were used: forward 5′-CAGTTGGACATGGGAAGTTGGAG-3′ and reverse 5′-CAGAGGGAA TTACTATGAGCTTAGTT-3′.

BM and LN cells were incubated with 2.4G2 hybridoma supernatant to block Fc receptors. LN cells were stained with FITC-conjugated Lin antibodies (CD3ϵ, CD4, CD8α, CD11b, CD11c, CD19, CD122, B220, NK1.1, Gr-1, TCRβ, TCRγδ, and Ter-119), CD127-PE, CD49b-APC, and Sca-1–Pe–Cy7. BM cells were stained with a different set of FITC-conjugated antibodies for the isolation of BM NK progenitors (CD3ϵ, CD4, CD8α, CD11b, CD11c, CD19, B220, Gr-1, TCRβ, TCRγδ, and Ter-119), CD122-PE, and NK1.1-APC. Cells were sorted on a FACSAria II (BD Biosciences) for 2 rounds to a purity of > 99%. For bulk NK-cell differentiation cultures, the sorted cells were cultured in the presence of stem cell factor, Flt-3 ligand, IL-7, and IL-15 in 24-well plates containing a preformed stroma layer of OP9 cells as described previously.¹⁵  One-half of the media was replaced every 4-5 days, and cultures were analyzed after 12 days. For limiting dilution, the secondary sort was done in purity mode into 96-well plates at a low flow rate on preformed OP9 stroma (2 × 10³ cells/well).

NK cells were cultured in 150 μL of RPMI 1640 medium with 10% FCS, 2mM l-glutamine, 100 U/mL penicillin, 100 μg/mL streptomycin, 2 × 10⁻⁵M β-mercaptoethanol in the presence of 1 ng/mL IL-12 and 0.5 ng/mL IL-18 or were left unstimulated for 18 hours at 37°C. IFN-γ production in culture supernatant was determined by the capture ELISA, following the manufacturer's instructions (R&D Systems). The detection limit was 62.5 pg/mL.

The NK-sensitive cell lines RMA-S, YAC-1, and NK-resistant RMA cells were stained with CellTacker Green CMFDA (Invitrogen). The CFSE-labeled target cells (10⁴) were mixed with effector NK cells at a 1:1, 1:2.5, and 1:5 ratio in 150 μL of RPMI 1640, 10% FCS on a 96-well plate. After 4 hours of incubation, cells were washed, stained with 1 μg/mL 7-aminoactinomycin D, and analyzed by FACS on a FACSCalibur. The level of cytotoxicity was determined as the percentage of 7-aminoactinomycin D–positive cells among CFSE⁺ target cells minus background.

LN and BM NKPs from B6 mice were sorted to high purity (> 99%), and 1 × 10³ cells were injected intravenously into NSG mice in 300 μL of PBS. Blood, spleen, and BM of injected mice were analyzed after 28 days.

Splenocytes from B6.RAG1 knockout (KO) mice were either injected into B6.Pep3b mice or labeled with CellTacker Green CMFDA, following the manufacturer's instructions (Invitrogen) and injected into wild-type (WT) B6 mice. Approximately 18 hours later, single-cell suspensions were prepared from BM, spleen, blood, and LNs and analyzed by flow cytometry. For B6.Pep3b mice receiving unlabeled RAG1 KO spleen cells, cells were stained with anti–CD45.2-PerCP-Cy5.5, anti–NKp46-FITC, and anti–CD127-PE. For B6 mice receiving CFSE-labeled spleen cells, cells were stained with anti–NK1.1-PerCP-Cy5.5, anti–CD49b-APC, and anti–CD127-PE.

Vosshenrich et al¹³  have previously reported that the percentages of CD127⁺ cells among splenic and LN NK cells are much lower in athymic mice than in WT B6 mice. In the study, NK cells were identified by CD3⁻CD122⁺NK1.1⁺. Here, we reexamined the frequency of CD127⁺ NK cells with the use of the natural cytotoxicity receptor NKp46 to identify NK cells because it is thought to be a specific marker for mouse NK cells.¹⁶  WT and athymic nude (Foxn1^−/−) B6 mice did not differ significantly in the frequencies of CD127⁺ cells among CD3⁻NKp46⁺ NK cells in the LN, spleen, and BM. Very few NK cells in the liver expressed CD127, and the frequency of CD127⁺ cells was slightly lower in nude liver than that of B6 liver (P < .05) (Figure 1A). As reported previously, CD127⁺ NK cells were relatively rich in the LN, whereas they comprise only a minor population in the spleen, BM, and liver. Analysis of young (4 weeks old), middle-aged (11 weeks old), and old (29 weeks old) WT and nude B6 mice (sex and age matched) showed that the frequency of LN CD127⁺ NK did not change with the age of the mice (data not shown). Individual analysis of axillary, inguinal, cervical, and popliteal LNs showed no significant difference among different LNs (data not shown). Therefore, LN cells were pooled for subsequent studies. Despite the lack of T cells, the LNs of nude mice contained significantly more lymphocytes than WT counterparts (63.0 ± 23.3 × 10⁶ vs 27.5 ± 10.8 × 10⁶; P < .01). NK cells were 2.7% of nude mouse LN cells and only 0.6% of B6 mouse LN cells. Thus, 2.1 ± 0.8 × 10⁵ CD127⁺ NK cells are found in nude mouse LN compared with 2.1 ± 0.5 × 10⁴ CD127⁺ NK cells in B6 mouse LN. Although most CD3⁻NKp46⁺ cells coexpressed high levels of NK1.1 or CD49b, some CD3⁻NKp46⁺ cells did not express NK1.1 or CD49b. Similarly, some CD3⁻NK1.1⁺ or CD3⁻CD49b⁺ did not express NKp46 (data not shown). Therefore, CD127 expression on LN NK cells defined by CD3⁻NK1.1⁺ or CD3⁻CD49b⁺ was also tested. Again, there was no significant difference between WT and athymic B6 mice in the frequency of CD127⁺ NK cells in the LN (Figure 1B). We further analyzed the phenotype of CD127⁺ and CD127⁻ NK cells in the LN of WT and nude B6 mice and found no significant difference in the expression of Ly49s, CD122, KLRG1, CD11b, CD27, NK1.1, and DX5 with the only exception being CD117 (c-Kit), which was expressed at significantly (P < .05) higher levels in nude mouse LN NK cells (Figure 1C).

In the above-described analysis, we detected LN cells expressing CD49b but not NKp46. Some CD3⁻CD49b⁺NKp46⁻ LN cells expressed CD127 (Figure 2A). Our initial characterization of this population showed that it contained CD11b⁺ cells, CD19⁺ cells, and other undefined cells. Therefore, we further characterized the population with the use of a cocktail of mAbs to mature hematopoietic cell markers (Lin), including CD122, NK1.1, CD3ϵ, CD4, CD8α, TCRβ, TCRγδ, CD11b, CD11c, CD19, B220, Gr-1, and Ter-119. Note that the Lin antibody cocktail in this study included anti-CD122, which would exclude conventional Lin⁻CD122⁺ NKPs. The frequency of Lin⁻CD49b⁺ cells was 0.05% ± 0.01% (n = 5) in the LN, 1.0% ± 0.1% (n = 3) in the BM, 0.14% ± 0.01% (n = 3) in the blood, 0.09% ± 0.02% (n = 3) in the spleen, and 0.13% ± 0.02% (n = 3) in the liver of normal B6 mice. Although most Lin⁻CD49b⁺ cells in the LN expressed CD127, only a small fraction of those in other tissues tested was CD127⁺ (Figure 2B). Conversely, most Lin⁻CD49b⁺ cells in the BM expressed CD117 (c-kit), but only a minor subset in the LN was CD117⁺. Lin⁻CD49b⁺ cells in the spleen were similar to those in the BM, although they had higher percentages of CD127⁺ and CD117⁻ cells. Lin⁻CD49b⁺ cells in the LN are B220⁻ but express CD27 and 2B4 (Figure 2C). To examine the relationship between Lin⁻CD49b⁺ LN cells and the conventional Lin⁻CD122⁺ NKPs, another Lin cocktail of mAbs without anti-CD122 was used to identify Lin⁻CD122⁺ cells in the LN. Those cells were mostly CD127⁺, and only ∼ 25% expressed CD49b. Similarly, only < 20% of Lin (minus CD122)⁻CD49b⁺ LN cells were CD122⁺ (supplemental Figure, available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Therefore, Lin⁻CD122⁻CD49b⁺LN cells are different from conventional Lin⁻CD122⁺ NKPs.

Lin⁻CD49b⁺CD127⁺ LN cells were sorted from normal B6 mice to high purity (> 99%) and cultured on a confluent layer of OP9 stroma cells in the presence of standard NK differentiation cytokine cocktail.¹⁷  Conventional NKPs (Lin⁻CD122⁺)⁸  in the BM were also sorted (Lin antibody cocktail excluded anti-CD122) and cultured for NK cell differentiation for comparison. Both Lin⁻CD49b⁺CD127⁺ LN cells and conventional BM NKPs efficiently differentiated into NK cells expressing NKp46 and CD122 at the end of the 12 days of culture (Figure 3A). Further analyses of NK cells generated in vitro from Lin⁻CD49b⁺CD127⁺ LN cells (LN NKPs hereafter), and conventional BM NKPs showed most NK cells generated from the former expressed significantly higher levels of CD127 than those generated from the latter, whereas the expression of NKG2D, NKG2A, and Ly49 was the same. Limiting dilution assays showed that ∼ 1 in 36 Lin⁻CD49b⁺CD127⁺ LN cells and 1 in 20 BM NKPs differentiated into NK cells in vitro (Figure 3B).

The functional maturity of LN NKP-derived NK cells was also assessed. They efficiently killed the prototypic NK cell target YAC-1 (Figure 3D). They also killed MHC class I–deficient RMA-S cells but not control RMA cells expressing MHC class I, indicating that the cytotoxicity was inhibited by MHC class I (data not shown). On stimulation with IL-12 and IL-18, they produced a large amount of IFN-γ (Figure 3E). Taken together, LN NKP-derived NK cells had the functions of mature NK cells.

To assess the NK-cell potential of LN NKPs in vivo, we grafted 1 × 10³ LN NKPs from B6 mice (CD45.2) into lymphopenic NSG mice (CD45.1). Four weeks later, we assessed the level of chimerism by staining for CD45.2⁺ cells in the recipient mice and characterized their maturation level. Because the recipient mice have no LNs, we recovered LN NKP-derived CD45.2⁺ cells from the BM, spleen, and blood. The percentages of NK cells among CD45.2⁺ cells varied, ranging from 12% to 38.8%, whereas no mature T and B cells could be detected (CD3 and CD19, respectively; Figure 3F). Remarkably, ∼ 40% of donor-derived NK (NKp46⁺) cells expressed CD127. These results show that LN NKPs differentiate into CD127⁺ NK cells in vitro as well as in vivo.

To determine whether Lin⁻CD49b⁺CD127⁺ LN NKPs derive from the thymus, we compared LN NKPs from WT and athymic nude (Foxn1^−/−) B6 mice. Lin⁻CD49b⁺ cells were detected in nude mouse LNs at a higher frequency than in B6 mouse LNs (Figure 4A left). The total numbers of Lin⁻CD49b⁺ LN cells were ∼ 1.1 ± 0.3 × 10⁵ in nude mouse and 1.1 ± 0.4 × 10⁴ in WT B6 mouse. Both B6 and nude LN NKPs were CD127⁺CD117⁻. Further analysis showed a significant difference in the expression of Sca-1. Although approximately one-half (45.8% ± 3.2%; n = 7) of WT B6 LN NKPs expressed low levels of Sca-1, most nude mouse LN NKPs are Sca-1^hi with few (7.0% ± 0.7%; n = 4) Sca-1^lo cells. To test their NK-cell potential, nude mouse LN NKPs, as well as Sca-1^lo and Sca-1^hi LN NKPs, from WT mice were purified by cell sorting and cultured for NK-cell differentiation. They all differentiated into NK cells expressing NKp46, CD122, and variable levels of CD127 (Figure 4B). They were also functionally mature and efficiently killed YAC-1 cells (Figure 4C). Limiting dilution analysis of B6 LN NKP subsets showed that 1 in 21 Sca-1^lo and 1 in 72 Sca-1^hi LN NKPs differentiated into NK cells (data not shown).

We have previously shown that ∼ 20% of LN NK cells in WT B6 mice, but not nude mice, have rearranged TCRγ genes, indicating that TCRγ gene rearrangement is a unique marker for thymus-dependent NK cells.¹⁵  Therefore, we tested whether TCRγ genes are rearranged in NK cells generated in vitro from Sca-1^lo and Sca-1^hi subsets of LN NKPs. As expected from our previous report, NK cells generated from both LN NKPs of nude mice showed no TCRγ gene rearrangement. Interestingly, bulk B6 LN NKP-derived NK cells, but not BM NKP-derived NK cells, were positive for TCRγ rearrangement, and only those derived from the Sca-1^lo subset of LN NKPs contained cells positive for TCRγ rearrangement (Figure 4D). B6 thymocytes were used as positive control for the genomic PCR analysis. To determine whether TCRγ genes are rearranged in NKPs or during the in vitro differentiation, Sca-1^hi and Sca-1^lo B6 LN NKPs were purified and tested for rearranged TCRγ genes. The results showed that freshly isolated Sca-1^lo, but not Sca-1^hi, LN NKPs have rearranged TCRγ genes. Those results collectively showed thymus-dependent Sca-1^lo and thymus-independent Sca-1^hi NKPs in the mouse LN differentiate into mature CD127⁺ NK cells.

The above-described results suggested that CD127⁺ NK cells in steady state LN may develop within the LN from the novel Lin⁻CD49b⁺CD127⁺ LN NKPs. Alternatively, CD127⁺ NK cells may develop elsewhere and preferentially migrate into the LN. To test the latter possibility, we intravenously injected CFSE-labeled spleen cells from RAG1-deficient mice into nonirradiated normal B6 mice and examined the migration of donor NK cells (CFSE⁺NK1.1⁺CD49b⁺) after ∼ 18 hours. The recovery of CFSE⁺ cells was the highest in the spleen (mean = 3.2 × 10⁴/10⁶ injected, SD = 1.6 × 10⁴; n = 6) and lowest in the LN (Figure 5A left). When axillary, cervical, and inguinal LNs were individually isolated and analyzed, no preferential migration of CFSE⁺ cells into a specific LN was found. Therefore, all LNs of the recipients were pooled and further analyzed for donor NK cells. On average, 218 ± 50 (n = 6) CFSE⁺ cells were recovered in the LNs of recipients from 10⁶ injected cells, and of those 68% ± 17% were NK cells (Figure 5A right). Approximately 33% of RAG1 KO spleen cells were NK cells. Thus, of ∼ 3.3 × 10⁵ NK cells among 10⁶ injected cells, only ∼ 148 NK cells (0.04%) were recovered in LNs. In this experiment, CD127 expression on CFSE⁺ NK cells in the LN was not reliably analyzed because of difficulty in compensating strong CFSE fluorescence. Therefore, we also injected unlabeled RAG1 KO spleen cells (CD45.2) into normal B6.Pep3b mice (CD45.1) and analyzed CD127 expression on CD45.2⁺NKp46⁺ NK cells in the injected mice. Because some RAG1 KO spleen cells were not strongly stained by anti-CD45.2, the total number of injected cells recovered in the recipients' lymphoid organs could not be determined in this analysis. However, the donor and host NK cells were readily identified by a combination of anti–CD45.2-PerCP-Cy5.5 and anti–NKp46-FITC staining (Figure 5B dot plots). The percentages of CD127⁺ cells among donor NK cells recovered in various lymphoid organs, including the LN of the recipients, were similar to that of the injected spleen cells (Figure 5B contour plots). These results showed that CD127⁺ NK cells do not preferentially migrate to the LN.

Here, we report a novel population of NKPs in the mouse LN. Unlike conventional NKPs in the BM that are Lin⁻CD122⁺, the newly identified LN NKPs do not express CD122 but express CD127 and CD49b (pan-NK cell marker DX5/HMα2, very late antigen-2). NK cells generated in vitro from LN NKPs and those from BM NKPs differ as well. The former express CD127, whereas the latter are mostly CD127⁻. Although CD127 is expressed on immature NK cells, CD127⁺ NK cells generated from LN NKPs are functionally mature because they efficiently kill YAC-1 cells and produce IFN-γ on stimulation. On transplantation into lymphopenic NSG mice, LN NKPs differentiate into both CD127⁺ and CD127⁻ NK cells, whereas most of the transplanted Lin⁻CD49b⁺CD127⁺ cells retain the expression of CD49b and CD127. Furthermore, LN NKPs can be divided into Sca-1^hi and Sca-1^lo subsets, both of which differentiate into CD127⁺ NK cells in vitro. Sca-1^lo LN NKPs are greatly reduced in athymic Foxn1^−/− mice, suggesting that most of them derive from the thymus.

Veinotte et al¹⁵  have previously detected thymus-dependent NK cells marked by TCRγ gene rearrangement, which are most abundant in the LN but undetectable in athymic mice. Interestingly, TCRγ gene rearrangement is not restricted to CD127⁺ NK cells because it is detected in both CD127⁺ and CD127⁻ NK cells at similar levels. Consistent with those previous findings, in our current study TCRγ gene rearrangement is detectable in NK cells generated in vitro from B6 mouse LN NKPs but not athymic mouse LN NKPs. When B6 LN NKPs are divided into Sca-1^hi and Sca-1^lo subsets, TCRγ gene rearrangement is only detected in Sca-1^lo NKPs and NK cells generated from them. Although most NK cells generated in vitro from LN NKPs are CD127⁺, both CD127⁺ and CD127⁻ NK cells are generated from LN NKPs in mice that receive a transplant. Taken together, these results suggest that thymus-derived Sca-1^lo LN NKPs, some of which have rearranged TCRγ genes, differentiate into CD127⁺ and CD127⁻ NK cells. We have previously found that thymus-derived Lin⁻CD44⁺ NKPs in the LN resembling immature thymocytes develop into NK cells with rearranged TCRγ genes. CD127 was detected in only a minor subset of the Lin⁻CD44⁺ NKP population in this earlier study, and whether they develop into CD127⁺ NK cells was not tested. LN NKPs (Lin⁻CD127⁺CD49b⁺) in the current study are all CD44⁺ (data not shown) and seem to partially overlap with the Lin⁻CD44⁺ NKP population. It is possible that Sca-1^lo LN NKPs derive from Lin⁻CD44⁺Kit^loCD127⁺Sca-1^lo thymic progenitors reported by Terra et al¹⁸  However, the origin of Sca-1^hi LN NKPs is currently unknown. They may directly migrate from the BM or alternatively develop from more immature lymphoid progenitors in the LN. Indeed, we have recently identified lymphoid progenitors that have T-, B-, and NK-cell potential in the LN (K.W., C.L., and F.T., manuscript in preparation), and some of the LN NKPs in this study may derive from those lymphoid progenitors.

Vosshenrich et al¹³  reported that most thymic NK cells are CD127⁺, athymic mice have greatly reduced CD127⁺ NK cells, and transplantation of thymus results in donor-derived CD127⁺ NK cells. On the basis of those findings they suggested that CD127⁺ NK cells are thymus derived. However, our current study did not find a reduction in CD127⁺ NK cells in athymic mice. The reason for the difference is still unclear. Whereas at least some CD127⁺ NK cells in WT B6 mice, particularly those with rearranged TCRγ genes, are thymus dependent, those in athymic mice are obviously thymus independent. It is possible that CD127⁺ NK cells in the athymic mouse LN may develop from LN NKPs. Alternatively, they may develop elsewhere and preferentially migrate into the LN. However, our migration study suggests that CD127⁺ NK cells do not preferentially migrate into the LN. Therefore, a relatively high percentage (∼ 20%) of CD127⁺ NK cells in the LN compared with spleen (∼ 5%) and peripheral blood (∼ 4%) cannot be explained by preferential migration of CD127⁺ NK cells from the circulating pool. Thus, it seems probable that CD127⁺ NK cells in the LN develop, at least in part, from thymus-dependent Sca-1^lo and thymus-independent Sca-1^hi LN NKPs.

Although some hematopoietic progenitors in the BM have been shown to express CD49b,¹⁹  the phenotype and lineage potential of LN NKPs is different from that of multipotent hematopoietic progenitors²⁰  and the common lymphoid progenitor^21,22  in the BM. LN NKPs do not express CD135 (fms-like tyrosine kinase receptor-3) and CD117 (c-Kit), and they only have NK potential. The developmental relationship between LN NKPs and conventional BM NKPs is still unclear. Although LN NKPs are CD122⁻, they acquire CD122 as they differentiate into mature NK cells while retaining CD127.

Our in vivo engraftment studies were performed in NSG mice that lack B, T, and NK cells and fail to develop lymph nodes during ontogeny. However, LN NKPs seem to have limited proliferative capacity in lymphopenic NSG mice compared with BM NKP. We assume that the poor engraftment potential was because of the absence of LNs in NSG. LN stroma cells are a source for IL-7,²³  which seems to be required for the development of CD127⁺ NK cells.¹³  Remarkably, a large proportion of donor-derived cells that were recovered after 4 weeks shared a similar phenotype (CD49b⁺CD127⁺) with the injected LN NKPs, suggesting that they proliferate while maintaining their phenotype. Of note, the LN NKPs we described here are phenotypically different from mouse CD4⁺ lymphoid tissue inducer cells.^24,25 

Thus far, the physiologic relevance of LN NKP-derived NK cells is unknown. Under steady state conditions, NK cells in the LN localize to the interfollicular and cortex regions and stay in close proximity to dendritic cells, possibly allowing them to regulate initial immune responses in a timely manner.²⁶  LN-resident NK cells expressing CD127 might also be involved in the maintenance of LN homeostasis by regulating the availability of IL-7 produced by stroma cells, a cytokine critical for B- and T-cell homeostasis.

We thank C. Benz, D. Ko, and W. Xu for advice and excellent technical support.

This work was supported by grants from the Canadian Cancer Society Research Institute and the Deutsche Forschungsgemeinschaft (grant LU 1529/1-1; C.L.).

Contribution: C.L. designed and performed the research, analyzed the data, and wrote the paper; K.W. designed and performed the research, analyzed the data, and wrote the paper; and F.T. designed the research, analyzed the data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Fumio Takei, Terry Fox Laboratory, British Columbia Cancer Agency, 675 West 10th Ave, Vancouver, BC, Canada; e-mail: ftakei@bccrc.ca.