Origin and differentiation of natural killer cells. II. Functional and morphologic studies of purified NK-1.1+ cells

Cells bearing the NK-specific marker NK-1.1 were purified from mouse spleens by utilizing a monoclonal anti-NK-1.1 antibody and cell sorting. In normal adult mice, all of the splenic NK activity against YAC-1 cells was found in the NK-1.1+ fraction, whereas NK-1.1- cells were depleted of NK activity. The NK activity of sorted NK-1.1+ cells was enriched 15- to 30-fold over unfractionated spleen cells. Light and electron microscopic studies of purified NK-1.1+ cells showed a homogeneous population of cells, each containing one to four cytoplasmic granules. Mice whose bone marrow has been destroyed by chronic exposure to 17-beta-estradiol have very low NK activity. However, spleen cells of estradiol-treated mice contained a normal frequency of NK-1.1+ cells which bound to YAC-1 cells, but failed to lyse them even after purification and subsequent exposure to interferon-alpha/beta in vitro. It appears, therefore, that in the absence of intact bone marrow, NK-1.1+ cells may be arrested in a nonlytic and interferon-unresponsive state. Spleens of neonatal mice which have low NK activity were analyzed to ascertain whether immature NK-1.1+ cells, similar to those found in estradiol-treated mice, could be demonstrated. Spleens of 8- to 9-day-old mice also contained NK-1.1+ cells which had very low NK activity even after purification. Sorted NK-1.1+ cells were examined for cytotoxicity in mice whose NK activity was suppressed by pretreatment with Corynebacterium parvum (-15 days). In contrast to cells from estradiol-treated and neonatal mice, NK-1.1+ from mice treated with C. parvum had normal functional activity. Similarly, although NK activity of unfractionated bone marrow cells is low, sorted NK-1.1+ cells were greatly enriched for lytic activity. Thus, we conclude that cell sorting with monoclonal anti-NK-1.1 antibody provides a powerful tool for examining the mechanisms underlying various states of low NK activity, and there exist NK-1.1+, nonlytic, interferon-unresponsive cells which apparently require an intact marrow microenvironment for differentiation into mature, lytic NK cells.     

The NK-1.1(-) mouse: a model to study differentiation of murine NK cells

The NK-1.1(-) mouse was constructed by weekly injections of monoclonal anti-NK-1.1 antibody from birth through adulthood. Spleen cells from these mice have decreased NK-1.1+ cells and null (Thy-1- and B220-) cells. Their splenic NK activity to YAC targets was low and was not enhanced by IFN-alpha or IFN-beta. Bone marrow (BM) of these NK-1.1(-) mice have normal precursors to NK cells: 1) NK activity could be generated from NK-1.1(-) BM cells cultured in rIL 2 for 5 to 6 days. These cultured BM cells expressed Qa-5, Thy-1, AsGm-1, and NK-1.1 antigens. The precursor cells of these BM cytotoxic cells are NK-1.1-; 2) transfer of BM cells from the NK-1.1(-) mice reconstituted the NK activity of irradiated, NK-depleted recipients. Lymphokine-activated killer cells could also be generated from spleens of these NK-1.1(-) mice. Therefore, the NK-1.1(-) mice were specifically depleted of mature cytotoxic NK cells, but not the NK-1.1- precursors of NK cells. This mouse model is valuable to study ontogeny and physiologic relevance of NK cells.     

Heterogeneity of natural killer cell subsets in NK-1.1+ and NK-1.1- inbred mouse strains and their progeny.

The 4LO3311 monoclonal antibody, a new NK-specific reagent recently produced in our laboratory, reacts with spleen cells of 11 mouse strains, most of which do not express the NK-1.1 alloantigen recognized by the PK136 mAb. Among positive strains, the susceptibility of spleen cells to the complement-dependent NK-inhibiting activity of the 4LO3311 mAb was variable but independent of the initial NK cell activity level of cells tested. This property was furthermore not modified after poly(I:C) stimulation. The susceptibility of spleen cells to the in vitro 4LO3311 mAb plus complement treatment is however influenced by the absolute number of 4LO3311+ cells as well as by the density of the corresponding alloantigen at the cell surface. Moreover, it was established that the strain-related variations observed also depended upon the relative size of the 4LO3311 cell subset within the lytic NK cell population. Indeed, when C3H (NK-1.1-4LO3311+) mice were inoculated with the 4LO3311 mAb, the lytic activity of their spleen cells was almost unaltered but 4LO3311-reactive cells were no longer detected in the spleen of treated animals and remaining NK cells were totally resistant to the in vitro 4LO3311 mAb plus complement treatment. These findings indicate that the 4LO3311 mAb identifies a subset rather than all NK cells, even in a NK-1.1- strain. Since a NK-1.1-unreactive cell subset was identified in NZB (NK-1.1+4LO3311-) mice inoculated with the PK136 mAb, the NK-1.1+ cell population is not necessarily responsible for all the splenic NK cell activity in all NK-1.1+ strains. In B6C3F1 hybrid mice, a relatively large subset of NK-1.1-4LO3311- cells was found in addition to those expressing the NK-1.1, the 4LO3311 alloantigen, or both. According to these results, NK cell heterogeneity should thus be taken as an evolving concept whose resolution appears more and more complex with the identification of new NK-specific reagents.     

Natural killer (NK) cell subsets in the mouse. NK-1.1+/LGL-1+ cells restricted to lysing NK targets, whereas NK-1.1+/LGL-1- cells generate lymphokine-activated killer cells

Our laboratory has recently identified a novel Ag, LGL-1, that is expressed on a major population of mouse NK cells. Two color immunofluorescence analysis has demonstrated that spleen cells consist of two major subsets of NK cells. We have identified an NK-1.1+/LGL-1+ subset that consists of 50% of the total NK cells and an NK-1.1+/LGL-1- subset comprising the remaining 50%. Because numerous reports have identified NK cells as the major cell type mediating lymphokine-activated killing (LAK), the NK-1.1+/LGL-1+ and NK-1.1+/LGL-1- subsets were examined for their contribution toward LAK generation, as defined by their ability to lyse P815 tumor targets. Antibody plus C depletion experiments with the use of anti-LGL-1 indicated that LGL-1+ cells were not found on LAK precursor or effector cells. Two-color cell sorting experiments were also performed to separate freshly isolated NK-1.1+/LGL-1+ spleen cells from the NK-1.1+/LGL-1- subset. It was found that the vast majority of LAK activity (greater than 95%) is derived from the NK-1.1+/LGL-1- cells. Cell sorting of LAK effectors also demonstrated that the NK-1.1+/LGL-1- cells mediated the vast majority of lysis against P815 targets. Similar results were obtained when NK cell subsets were analyzed for their contribution toward ADCC. These findings may prove important in understanding and further elucidating the contribution of NK cells to the LAK phenomenon. Our data also indicates that subsets of NK cells exist that may function differently in response to stimulation by various lymphokines and cytokines.     

Regulation of the secondary in vitro antibody response by endogenous natural killer cells: kinetics, isotype preference, and non-identity with T suppressor cells

Our previous studies had demonstrated that depletion of endogenous natural killer (NK) cells resulted in an augmented primary antibody response in vivo and in vitro. We have now examined the effect of NK cell depletion on the in vitro secondary response to antigen. Treatment of primed murine spleen cells with anti-NK-1.1 allo-antibody and complement before culture resulted in a significant increase in the magnitude of the antigen-specific plaque-forming cell (PFC) response. This treatment did not affect the proportions of Lyt-2+, L3T4+, or sIg+ cells in the population, however, indicating that the augmentation in PFC was not due to changes in the ratio of T to B cells. Removal of endogenous NK cells had a greater effect on the IgG (indirect) PFC response (100 to 200% increase) than on the IgM (direct) PFC response (25 to 50% increase). In contrast, removal of Lyt-2+ cells before culture affected the IgM and IgG responses similarly. Moreover, the kinetics of augmentation differed between cultures depleted of Lyt-2+ cells and those depleted of NK-1.1+ cells. NK cells appeared to act earlier in the response than did T suppressor cells. The NK-1.1+ cells involved in antibody regulation were not involved in the generation of the in vitro derived T suppressor cells. The conclusion that the regulation of the antibody response by NK-1.1+ cells is distinct from that involving T suppressor cells was confirmed in experiments in which removal of both regulatory cell populations resulted in an increase in PFC that was greater than in cultures depleted of either NK or T suppressor cells.     

NK1.1+ thymocytes. Adult murine CD4-, CD8- thymocytes contain an NK1.1+, CD3+, CD5hi, CD44hi, TCR-V beta 8+ subset.

CD4-, CD8- thymocytes were purified from thymi obtained from normal C57BL/6 mice. By flow cytometry analysis, 5 to 10% of these double negative (DN) thymocytes were found to express NK1.1 on their surface. The NK1.1+ DN thymocytes were demonstrated, by two-color fluorescence, to be CD3lo, CD5hi, CD44hi, J11d-, B220-, MEL 14-, IL2R- with 60% expressing TCR-V beta 8 as determined by the mAb F23.1. In contrast, splenic and peripheral blood NK cells were NK1.1+, CD3-, CD5-, TCR-V beta 8- with 40 to 60% being MEL 14+. Unlike peripheral NK cells, fresh DN thymocytes enriched for NK1.1+ cells were unable to kill YAC-1, the classical murine NK cell target. However, these cells were able to mediate anti-CD3 redirected lysis even when they were assayed immediately after purification, i.e., with no culture or stimulation. These data demonstrate that adult murine thymocytes contain NK1.1+ cells which are distinct, both by function and phenotype, from peripheral NK cells. These data also raise the issue of a possible NK/T bipotential progenitor cell.     

Murine NK cell heterogeneity: subpopulations of C57BL/6 splenic NK cells detected by NK-1.1 and NK-2.1 antisera

NK-1.1 antiserum - (BALB/c X C3H)F1 anti-CE - and NK-2.1 antiserum - NZB anti-BALB/c - detect genetically distinct alloantigens on C57BL/6 natural killer (NK) cells. We have analyzed whether these two alloantigens are associated with functional subsets of NK cells. For this study, nylon wool nonadherent C57BL/6 spleen cells (SC) were treated with complement (C) and NK-1.1 or NK-2.1 antisera and then tested for NK activity against a panel of tumor targets in 6- and 19-hour 51Cr release assays. The NK activity against the prototype NK target YAC-1 was reduced equally by both antisera. Similar reductions by both antisera were also observed when SC were tested against another murine lymphoma target, L5178c127v, against the C57BL/6 melanoma B16, and against the human liver cell line Chang. In contrast, NK activity to the lymphoma FBL-3 and the human erythroleukemia K562 was significantly reduced in SC treated with NK-2.1 antiserum and C, whereas SC treated with NK-1.1 antiserum and C showed either less reduction or no reduction in activity against these two cell lines. With two other targets, E male G2 and RBL-5, neither serum produced significant depletion of activity, Analysis of SC indirectly labeled with either NK-1.1 or NK-2.1 antiserum and fluorescein-labeled goat anti-mouse Ig, however, did not detect significant differences between NK-1+ and NK-2+ cell populations.     

Lymphokine-activated killer (LAK) cells. VI. NK1.1+, CD3+ LAK effectors are derived from CD4-, CD8-, NK1.1- precursors.

Normal murine splenocytes cultured with IL2 for 6, but not 3, days contained an NK1.1+, CD3+ lytically active subset. These lymphocytes were not derived from NK1.1+ precursors since NK1.1+ cells, purified by flow cytometry, failed to express CD3, as determined by the 145-2C11 mAb, on their surface even after culture with IL2 for 6 days. Instead, the precursors of the NK1.1+, CD3+ effectors were contained in a B cell-depleted CD4-, CD8-, NK1.1- splenic subset. Freshly obtained CD4-, CD8-, NK1.1- splenocytes were mostly CD3+, CD5+, B220-, had no spontaneous lytic activity against YAC-1, and were unable to mediate anti-CD3 directed lysis against FcR-bearing target cells. Culture of the CD4-, CD8-, NK1.1- splenocytes with IL2, for 6 days, resulted in the development of NK1.1+, CD3+, B220+ effectors 40% of which were CD5dim and 20-25% of which expressed TCR-V beta 8 as determined by the F23.1 mAb. The acquisition of NK1.1, B220, and lytic activity by this triple-negative subset was readily inhibited by cyclosporine A (CSA). On the other hand, CSA had no effect on the acquisition of B220 or lytic activity by NK1.1+ precursors obtained by flow cytometry sorting. Moreover, all of the NK1.1+ cells generated by IL2 culture of splenocytes obtained from mice depleted of NK1.1+ lymphocytes (by in vivo injection of anti-NK1.1 mAb) coexpressed CD3 on their surface and were thus distinct from classical NK cells. These findings demonstrate that splenic NK cells do not express or acquire CD3; that the NK1.1+, CD3+ LAK effectors are derived from an NK1.1- precursor; and that CSA is exquisitely selective in its inhibitory effect on LAK generation.     

Effect of cyclosporin A on lymphopoiesis. III. Augmentation of the generation of natural killer cells in bone marrow transplanted mice treated with cyclosporin A

The effects of cyclosporin A (CsA) on the generation of NK cells were studied using syngeneic bone marrow transplanted mice subsequently treated with CsA (BMT/CsA mice). In contrast to a severe reduction in T cells that was reported previously, these mice exhibited a marked enhancement of splenic NK activity. The enhanced NK activity was mediated by NK1.1+, Thy-1- cells as assessed by antibody plus complement treatment, and was concomitant with an absolute increase in the numbers of NK1.1+ cells as assessed by flow cytometry. Because the depletion of host-derived, mature NK cells by injection of anti-asialo GM1 antibody before bone marrow reconstitution did not affect the enhancement of NK activity, CsA appeared to augment the generation of NK cells from bone marrow precursors. To investigate a possible relationship between the enhancement of NK activity and the maturational arrest of T cells in the thymus induced by CsA, mice were thymectomized, followed by irradiation, bone marrow reconstitution, and CsA treatment. These mice exhibited as strong enhancement of splenic NK activity as BMT/CsA mice, suggesting that the CsA-induced effect on NK cells is distinct from its effect on T cell development in the thymus. Taken together, these results are the first demonstration of the positive effect of CsA on NK cell generation and may be of importance in clinical bone marrow transplantation.     

Suppressive effects of monocytic cells and transforming growth factor-beta on natural killer cell differentiation in autoimmune viable motheaten mutant mice.

Viable motheaten (mev) mice are homozygous for a recessive single gene mutation at chromosome 6. These mice develop numerous inflammatory and arthritic syndromes and exhibit abnormal B cell functions as well as lower T and NK cell activity. In this study, the differentiation of NK cells in mev mice was examined to elucidate the underlying basis for decreased NK activity. Although NK cells appear to be present in mev mice, their activity was demonstrable only when the spleen cells were enriched by nylon wool passage. Similarly bone marrow cells from these mice could be shown to contain precursors of NK cells when they were passed over nylon wool and transplanted into irradiated recipients. The adherent cells from both the spleen and bone marrow of mev mice suppressed the differentiation of NK cells from normal splenic populations. These suppressive adherent cells were F4/80(+), AsGm-1(+), Qa-5(+), and NK-1.1(+). They were not cytolytic when cultured in IL-2. Antibodies to a number of cytokines, such as IFN-alpha, -beta, and gamma, or TNF-alpha, could not reverse the suppressive effect of the adherent cells. Addition of anti-TGF-beta antibody could, however, overcome the suppression, suggesting that TGF-beta was partly responsible for the defective NK differentiation in the mev mice.     

The presence of NK alloantigens on cloned cytotoxic T lymphocytes

A panel of sera raised against NK-1.1 and NK-2.1 alloantigens was tested for reactivity against a panel of cloned antigen-dependent CTL lines. By using indirect immunofluorescence and flow cytofluorimetry, weak, but clear and consistent, reactivity was found on all CTL. Concordant with the genetics of NK alloantigens, C57BL/6-derived clones were reactive with anti-NK-1.1 and anti-NK-2.1 sera, whereas CBA-derived clones were reactive with anti-NK-2.1 sera but not with anti-NK-1.1 sera. Cloned CTL lines were also able to partially and specifically absorb the antibodies from NK alloantiserum that reacted with splenic NK cells. These results indicate that cloned CTL lines express at least some of the NK alloantigen determinants present on splenic NK cells and have important implications regarding the relationship of CTL and NK cells.     

Production and differentiation of NK lineage cells in long-term bone marrow cultures in the absence of exogenous growth factors.

Neither lytic NK cells nor IL-2-responsive NK precursors were produced in myeloid (Dexter) long-term bone marrow cultures (LTBMC). However, when myeloid LTBMC were switched to lymphoid (Whitlock-Witte) conditions and reseeded ("recharged") with fresh bone marrow cells (BMC), nonadherent cells with NK lytic activity and NK 1.1+ phenotype were produced within 1-2 weeks without the addition of exogenous IL-2 to the cultures. NK- and T cell-depleted BMC proliferated extensively in switched cultures and in 2 weeks generated cells that lysed the NK target YAC-1 but not the LAK target P815. The presence of NK precursors in the cultures was confirmed by reculturing nonadherent cells harvested from recharged LTBMC in fresh medium containing 50 U rIL-2/ml. High levels of NK lytic activity were generated. Sequential expression of NK 1.1 and IL-2 responsiveness followed by lytic activity was demonstrated by harvesting cells early after recharge, prior to the appearance of lytic cells. Elimination of NK 1.1+ cells depleted the ability to respond to IL-2 in secondary culture. Our studies demonstrate that myeloid-to-lymphoid switched LTBMC support the proliferation and differentiation of NK lineage cells from their NK 1.1-, nonlytic progenitors in the absence of an exogenous source of growth factors.     

Splenic NK1.1-negative, TCR alpha beta intermediate CD4+ T cells exist in naive NK1.1 allelic positive and negative mice, with the capacity to rapidly secrete large amounts of IL-4 and IFN-gamma upon primary TCR stimulation

Splenic NK1.1+CD4+ T cells that express intermediate levels of TCR alpha beta molecules (TCRint) and the DX5 Ag (believed to identify an equivalent population in NK1.1 allelic negative mice) possess the ability to rapidly produce high quantities of immunomodulatory cytokines, notably IL-4 and IFN-gamma, upon primary TCR activation in vivo. Indeed, only T cells expressing the NK1.1 Ag appear to be capable of this function. In this study, we demonstrate that splenic NK1.1-negative TCRintCD4+ T cells, identified on the basis of Fc gamma R expression, exist in naive NK1.1 allelic positive (C57BL/6) and negative (C3H/HeN) mice with the capacity to produce large amounts of IL-4 and IFN-gamma after only 8 h of primary CD3 stimulation in vitro. Furthermore, a comparison of the amounts of early cytokines produced by Fc gamma R+CD4+TCRint T cells with NK1. 1+CD4+ or DX5+CD4+TCRint T cells, simultaneously isolated from C57BL/6 or C3H/HeN mice, revealed strain and population differences. Thus, Fc gamma R defines another subpopulation of splenic CD4+TCRint cells that can rapidly produce large concentrations of immunomodulatory cytokines, suggesting that CD4+TCRint T cells themselves may represent a unique family of immunoregulatory CD4+ T cells whose members include Fc gamma R+CD4+ and NK1.1/DX5+CD4+ T cells.     

NK1.1+ T cells in the liver arise in the thymus and are selected by interactions with class I molecules on CD4+CD8+ cells

NK1.1+ T cells represent a specialized T cell subset specific for CD1d, a nonclassical MHC class I-restricting element. They are believed to function as regulatory T cells. NK1.1+ T cell development depends on interactions with CD1d molecules presented by hematopoietic cells rather than thymic epithelial cells. NK1.1+ T cells are found in the thymus as well as in peripheral organs such as the liver, spleen, and bone marrow. The site of development of peripheral NK1.1+ T cells is controversial, as is the nature of the CD1d-expressing cell that selects them. With the use of nude mice, thymectomized mice reconstituted with fetal liver cells, and thymus-grafted mice, we provide direct evidence that NK1.1+ T cells in the liver are thymus dependent and can arise in the thymus from fetal liver precursor cells. We show that the class I+ (CD1d+) cell type necessary to select NK1.1+ T cells can originate from TCRalpha-/- precursors but not from TCRbeta-/- precursors, indicating that the selecting cell is a CD4+CD8+ thymocyte. 5-Bromo-2'-deoxyuridine-labeling experiments suggest that the thymic NK1.1+ T cell population arises from proliferating precursor cells, but is a mostly sessile population that turns over very slowly. Since liver NK1.1+ T cells incorporate 5-bromo-2'-deoxyuridine more rapidly than thymic NK1.1+ T cells, it appears that liver NK1.1+ T cells either represent a subset of thymic NK1.1+ T cells or are induced to proliferate after having left the thymus. The results indicate that NK1.1+ T cells, like conventional T cells, arise in the thymus where they are selected by interactions with restricting molecules.     

Unusual T cell populations in adult murine bone marrow. Prevalence of CD3+CD4-CD8- and alpha beta TCR+NK1.1+ cells

The T cell populations present in normal murine bone marrow have not been previously analyzed in detail, mainly because of their relative rarity. In order to permit such analyses, bone marrow T cells were enriched by depleting Mac1-positive cells, which constitute 65 to 90% of bone marrow cells (BMC), and then studied by two-color flow cytometry. Analysis of the remaining cells revealed that the T cell profile of adult murine bone marrow is markedly different from that of other lymphoid organs. A very high proportion of bone marrow CD3+ cells (approximately one-third) are CD4-CD8-. CD3+CD4-CD8- cells are much more concentrated among BMC T cells than among thymocytes or splenic T cells, suggesting that bone marrow may be either a site of extrathymic TCR gene rearrangement, or a major site to which such cells home from the thymus. The expression of NK1.1 was also evaluated on Mac1-depleted BMC populations. Surprisingly, up to 39% of alpha beta TCR+ BMC were found to express NK1.1. Most alpha beta TCR+NK1.1+ BMC also expressed CD4 or CD8. NK1.1+ alpha beta TCR+ cells represented a much greater proportion of BMC T cells than of other lymphoid (splenocyte or thymocyte) T cell populations. Mac1-depleted BMC of nude mice contained very few cells with this phenotype. These results are consistent with the hypothesis that NK1.1+ alpha beta TCR+ cells are generated primarily in the thymus of normal animals and migrate preferentially to bone marrow, where they may function as regulatory elements in hematopoiesis.     

Antigenic, functional, and molecular genetic studies of human natural killer cells and cytotoxic T lymphocytes not restricted by the major histocompatibility complex

Cytotoxicity not restricted by the major histocompatibility complex (MHC) is mediated by two distinct types of lymphocyte: natural killer (NK) cells and non-MHC-restricted cytotoxic T lymphocytes (CTL). These two types of cytotoxic lymphocytes can be distinguished by antigenic phenotype, function, and molecular genetic studies. In human peripheral blood, NK cells are identified by expression of the Leu-19 and/or CD16 cell surface antigens, and lack of CD3/T cell antigen receptor (Ti) complex expression (i.e., CD3-,Leu-19+). Peripheral blood non-MHC-restricted CTL express both CD3 and Leu-19 (i.e., CD3+, Leu-19+, referred to as Leu-19+ T cells). Both Leu-19+ T cells and NK cells lyse "NK-sensitive" hematopoietic tumor cell targets, such as K562, without deliberate immunization of the host. However, most "NK activity" in peripheral blood is mediated by NK cells, because they are usually more abundant and more efficient cytotoxic effectors than Leu-19+ T cells. The cytolytic activity of both NK cells and Leu-19+ T cells against hematopoietic targets was enhanced by recombinant interleukin 2 (rIL 2). NK cells, but not peripheral blood Leu-19+ T cells, were also capable of lysing solid tumor cell targets after short-term culture in rIL 2. Southern blot analysis of NK cells revealed that both the T cell antigen receptor beta-chain genes and the T cell-associated gamma genes were not rearranged, but were in germ-line configuration. These findings indicate that NK cells are distinct in lineage from T lymphocytes and do not use the T cell antigen receptor genes for target recognition.(ABSTRACT TRUNCATED AT 250 WORDS)     

Heterogeneity of NK1.1+ T cells in the bone marrow: divergence from the thymus.

NK1.1+ T cells in the mouse thymus and bone marrow were compared because some marrow NK1.1+ T cells have been reported to be extrathymically derived. Almost all NK1.1+ T cells in the thymus were depleted in the CD1-/-, beta2m-/-, and Jalpha281-/- mice as compared with wild-type mice. CD8+NK1.1+ T cells were not clearly detected, even in the wild-type mice. In bone marrow from the wild-type mice, CD8+NK1.1+ T cells were easily detected, about twice as numerous as CD4+NK1.1+ T cells, and were similar in number to CD4-CD8-NK1.1+ T cells. All three marrow NK1.1+ T cell subsets were reduced about 4-fold in CD1-/- mice. No reduction was observed in CD8+NK1.1+ T cells in the bone marrow of Jalpha281-/- mice, but marrow CD8+NK1.1+ T cells were markedly depleted in beta2m-/- mice. All NK1.1+ T cell subsets in the marrow of wild-type mice produced high levels of IFN-gamma, IL-4, and IL-10. Although the numbers of marrow CD4-CD8-NK1.1+ T cells in beta2m-/- and Jalpha281-/- mice were similar to those in wild-type mice, these cells had a Th1-like pattern (high IFN-gamma, and low IL-4 and IL-10). In conclusion, the large majority of NK1.1+ T cells in the bone marrow are CD1 dependent. Marrow NK1.1+ T cells include CD8+, Valpha14-Jalpha281-, and beta2m-independent subsets that are not clearly detected in the thymus.