Increased bone marrow allograft rejection by depletion of NK cells expressing inhibitory Ly49 NK receptors for donor class I antigens

Natural killer (NK) cells are large granular lymphocytes capable of spontaneously killing various tumor cells and virally infected cells.1-3 Murine NK cells can also mediate the acute rejection of allogeneic bone marrow cells (BMCs) in lethally irradiated mice but not solid tissue allografts.3-7 They also exhibit a phenomenon called “hybrid resistance” in which H2 homozygous parental BMC grafts are rejected by irradiated H2 heterozygous F₁ hybrid mice.7 The recognition of major histocompatibility complex (MHC) class I molecules by murine NK cells is mediated by 2 families of cell surface receptors and has at least partially explained how NK cells can detect BMC allografts.6,8-18 One receptor family is called Ly49, which is composed of C-lectin type II membrane glycoproteins.6,8-18 Individual members of these Ly49 receptors have an affinity for particular MHC class I determinants.16,17 Several Ly49 receptors have been identified and cDNAs have been sequenced in mice with strain-specific allelic forms being reported.18-21 Most Ly49 molecules are “inhibitory receptors,” because binding to their specific targets transmits inhibitory intracellular signals through an immunoreceptor tyrosine-based inhibitory motif (ITIM) in their cytoplasmic domain.9,10,16,17 The ITIM recruits tyrosine phosphatase, which results in a potent inhibitory signal being sent to the NK cell.15-17 Three Ly49 receptors (Ly49D, Ly49H, and Ly49P) are characterized as “activating receptors,” because they lack the inhibitory motif (ITIM) in their cytoplasmic domain and instead are dependent on a noncovalently associated molecule “DAP12,” which contains an immunoreceptor tyrosine-based activating motif (ITAM).15,22,23 Recent studies have shown that NK cytotoxicity is regulated by a balance between expression of specific activating and inhibitory receptors.17 A second group of MHC class I–specific receptors on NK cells is composed of disulfide bonded heterodimers termed CD94/NKG2.16,17 CD94 has a short cytoplasmic tail with no signaling function and it forms a heterodimer with various NKG2 proteins A, B, C, D, and E.10,17 

We and others have shown the role of inhibitory Ly49A, C/I, and G2 NK subsets as well as the activating Ly49D NK subset in bone marrow allograft rejection in lethally irradiated mice.22-30Although a small percentage of T cells also express Ly49 receptors, mice with severe combined immunodeficiency (SCID), deficient in T and B cells, also mediate class I–specific bone marrow allograft rejection, indicating that NK cells alone can mediate the specificity seen in BMC rejection.5 

Our previous studies were designed to delineate the role of various Ly49 NK subsets in mediating BMC rejection. These studies involved infusing relatively small numbers of allogeneic BMCs that were rejected by untreated irradiated mice unless specific depletion of a Ly49 NK subset directly mediating the rejection was performed, which instead resulted in BMC engraftment.26,27,29 During those studies using some Ly49 monoclonal antibodies (mAbs) as controls for examining specificity of functions of other Ly49 subsets in mediating rejection, we made a novel and unexpected preliminary observation in mice. We observed that depletion of large subsets of NK cells expressing inhibitory receptors for a given H2 type of BMC had 2 effects: (1) increased resistance to the H2 type BMCs that these receptors recognized to receive negative signals and (2) weakening of rejection against BMC grafts that were not recognized by the particular NK receptor. For example, depletion of Ly49G2⁺ NK cells, which receive negative signals from H2^d but not H2^bclass I antigens, led to decreased ability of H2^dhost mice to reject H2^b BMCs.27 We describe here in more detail the in vivo effect of depletion of Ly49 NK subsets, which do not directly mediate the rejection of H2^d or H2^b haplotype BMCs, on the rejection of higher numbers of BMCs and discuss the potential relevance of these findings in clinical bone marrow transplantation (BMT).

BALB/c (H2^d), C.B-17 SCID (H2^d), C57BL/6 (H2^b), and BALB/c × C57BL/6 F₁(CB6F₁, H2^bxd), mice were obtained from the Animal Production Area, National Cancer Institute (Frederick, MD). All mice were kept under specific pathogen-free conditions until use at 8 to 12 weeks of age.

Groups of 5 to 10 C57BL/6 (H2^b) mice were injected intraperitoneally with 500 μg mAb 5E6 (anti-Ly49C/I) or 200 μg mAb 4D11 (anti-Ly49G2) or 0.2 mL normal rat serum (NRS). Seventy-two hours later spleens were harvested and single-cell suspensions were prepared. Four-color flow cytometric analysis was performed to determine the percentage of Ly49C/I⁺, Ly49G2⁺, and Ly49D⁺ NK cells. Cells were incubated with phycoerythrin (PE)–conjugated mAbs 5E6 (anti-Ly49C/I) or 4E5 (anti-Ly49D), fluorescein isothiocyanate (FITC)–conjugated mAb 4D11 (anti-Ly49G2) or goat-antimouse/rat IgG, allophycocyanin (APC)–conjugated NK pan marker NK1.1 and peridinin chlorophyll protein (PCP)–conjugated anti-CD3 for 30 minutes at 4°C. After 2 washes, the cells were analyzed on an LSR flow cytometer (Becton Dickinson, San Jose, CA). All secondary antibodies were titered and tested for their ability to stain for primary antibodies.

Groups of 4 to 5 conventionally housed recipient mice were injected intraperitoneally with 500 μg mAb 5E6 (anti-Ly49C/I), 200 μg mAb 4D11 (anti-Ly49G2), 500 μg of mAb PK136 (anti-NK 1.1), 20 μL antiasialo GM1 (anti-ASGM1; Wako Chemicals, Richmond, VA), 0.2 mL NRS, or 120 μg polyinosinic/polycytidylic (poly I:C), 2 days before irradiation. The engraftment of donor BMCs was assessed by an in vitro colony assay for hematopoietic cell growth.26 Two days after antibody injection, mice were lethally irradiated with a¹³⁷Cs source (BALB/c at 850 cGy, C57BL/6 at 950 cGy, C.B-17 SCID at 350 cGy, and CB6F₁ at 1000 cGy) and later injected intravenously with 0.5 × 10⁶ to 25 × 10⁶BMCs in 0.5 mL phosphate-buffered saline (PBS). At day 8, spleens were gently crushed in medium and single-cell suspensions were prepared. Then, 10⁶ spleen cells were plated in 0.3% sea plaque agar in 35-mm dishes containing optimal concentrations of recombinant murine cytokines (interleukin 3 [IL-3; 10 ng/mL] and granulocyte-macrophage colony-stimulating factor [GM-CSF; 10 ng/mL]). Cytokines were obtained from the Biological Resources Branch (Frederick, MD). Plates were incubated at 37°C for 7 days and the colonies with more than 50 cells were counted (colony-forming unit-culture [CFU-c]). The data are presented as mean ± SD total CFU-c per spleen, which was obtained by multiplying number of colonies by the cellularity. The Student t test was performed to determine if the mean values were significantly different (P < .05). Each experiment was performed 3 to 4 times with at least 4 mice per group.

We have reported earlier that in BALB/c (H2^d) mice, the Ly49G2⁺ NK subset can mediate the rejection of allogeneic H2^b BMCs and conversely Ly49C/I⁺ NK cells are responsible for the rejection of H2^d BMCs in C57BL/6 (H2^b) mice as ascertained by the in vivo depletion of these subsets.25,26,29 While studying the in vivo role of Ly49G2⁺ NK subset in BALB/c (H2^d) mice, we observed an interesting phenomenon by using antibodies directed against the Ly49C/I⁺ NK cell subset as control for the specificity of Ly49G2⁺ NK subset removal.26 Depletion of Ly49C/I⁺ NK cells in BALB/c (H2^d) mice, which is the NK subset that receives negative signals from H2K^band therefore cannot reject H2^b BMCs, resulted in the ability of the recipient mice to reject higher numbers of allogeneic H2^b BMCs.26 Therefore, depletion of a population of NK cells that were considered indifferent to H2^b BMCs nevertheless resulted in enhanced rejection by H2^d mice.

We carried out further transplantation studies to determine the effects of in vivo depletion of this and other Ly49 NK subsets that are not responsible for the rejection of a particular BMC allograft. To do this, we wanted to assess the effect of NK subset removal in instances where increasing BMC numbers could overcome resistance and result in engraftment. The number of allogeneic BMCs transplanted varied because mice differ in their BMC rejection capability.

We examined the effect of in vivo depletion of Ly49C/I⁺ or G2⁺ NK subsets on allograft rejection in BALB/c (H2^d) mice. Mice were injected with 200 μg mAb 4D11 (anti-Ly49G2) or 500 μg mAb 5E6 (anti-Ly49C/I) to deplete the respective Ly49 NK subset and 48 hours later mice were lethally irradiated and allogeneic BMCs were infused. After 8 days, the splenic CFU-c assay, which quantitates hematopoietic progenitors as reflected by growth in agar, was performed to assess BMC engraftment in mice. The CFU-c assay was chosen over the (¹²⁵I) UdR radioisotope incorporation assay because though both assays detect progenitor cell function, the CFU-c assay, in addition, detects the production of CFU-c precursor/progenitor cells derived from the inoculum, whereas the other detects any proliferating marrow-derived cells in S-phase of the cycle. The results presented in Figure1A demonstrate that irradiated BALB/c mice were capable of resisting 0.5 × 10⁶ H2^bBMCs, whereas the syngeneic control H2^b C57BL/6 mice engrafted H2^b BMCs. When the total NK cell population was removed by injecting ASGM1 antibodies, engraftment of allogeneic H2^b BMCs was enhanced (P < .0001; Figure 1A). Removal of Ly49C/I⁺ NK subset, which does not mediate the rejection of allogeneic H2^b BMCs as they receive negative signals from H2K^b BMCs, led to rejection of the BMCs but depletion of Ly49G2⁺ NK cells (which receive negative signals from H2D^d BMCs and therefore can mediate the rejection of H2^b BMCs) significantly increased the engraftment of allogeneic H2^b BMCs compared with the control group (NRS treated).26 However, when rejection was overridden by infusing larger numbers of (8 × 10⁶) allogeneic H2^b BMCs (Figure 1B-C), engraftment of the infused BMCs was noted in the control group (NRS treated), similar to syngeneic control (Figure 1B) and anti-ASGM1–treated group (Figure 1B-C), which had all of the NK cells depleted, suggesting that no resistance was occurring. Removal of Ly49G2⁺ NK cells, which mediates the rejection of H2^b BMCs, did not increase the engraftment of allogeneic H2^b BMCs compared with the control group (NRS treated) also indicating that no appreciable resistance was occurring (Figure 1B-C). In vivo activation of NK cells by poly I:C, which involves interferon α/β (IFN-αβ) production and increases BMC rejection ability of mice, prior to BMC infusion resulted in inhibition of engraftment (P < .007) compared with the control group (NRS treated; Figure 1C). However, in vivo removal of the supposedly irrelevant Ly49C/I⁺ NK subset (which does not mediate the rejection of allogeneic H2^bBMCs because it receives negative signals from H2K^b BMCs) resulted in significant rejection (P < .0001) of H2^b BMCs compared with either of the control groups (NRS or ASGM1 treated; Figure 1B-C) or poly I:C-treated group (P < .002; Figure 1C). Thus, removal of Ly49C/I⁺ NK cells in H2^d mice enhances the capability of the mice to reject H2^b BMCs and even greater than if agents commonly used to activate NK cells (ie, poly I:C) are used, which is not noticeable when low numbers of allogeneic H2^b BMCs are transplanted. This suggests that in H2^d (BALB/c) mice, Ly49C/I⁺ NK cells exert an inhibitory effect on the cells responsible for the rejection of H2^b BMCs.

Next, we performed the reciprocal experiment. We examined the effect of depletion of the Ly49C/I⁺ or Ly49G2⁺NK subset on the rejection of higher numbers of allogeneic H2^d BMCs in irradiated C57BL/6(H2^b) mice to assess whether Ly49G2⁺ NK cells were able to inhibit the ability of mice to reject H2^d BMCs.23,24 In lethally irradiated H2^b mice, the Ly49C/I⁺ NK subset has been shown to mediate the rejection of allogeneic H2^d BMCs when transplanted in limited (0.5 × 10⁶) numbers (Figure2A).25,29 However, when allogeneic H2^d BMCs were infused in larger numbers (25 × 10⁶ BMCs), the control group (NRS treated) could no longer reject the bone marrow allograft because the CFU-c content was similar to mice depleted of all NK cells by NK1.1 mAb and syngeneic control (Figure 2B-C). Under such circumstances, depletion of the Ly49C/I⁺ NK subset (containing Ly49D⁺ cells) also did not affect the engraftment of H2^d bone marrow allografts when compared with the control group (NRS-treated group; Figure 2C). However, the C57BL/6 mice treated with the anti-Ly49G2 mAb 4D11 demonstrated significant augmentation of H2^d bone marrow allograft rejection (Figure 2B-C; P < .003) compared with the control recipients and this could be reversed if all the NK1.1⁺ cells were removed by injecting pan NK mAb, PK136 (αNK1.1) along with anti-Ly49G2 mAb 4D11, suggesting that BMC graft rejection was mediated by NK1.1⁺ cells (Figure2C).

Thus, in H2^b mice, the converse of the effect seen with NK subset removal in H2^d mice (Figure 1B) occurs. Removal of the Ly49 NK subset that receives negative signals from H2^dsignificantly enhances the ability of the remaining NK1.1⁺cells to reject H2^d BMCs.

NK cells in F₁ (H2^bxd) hybrid mice can reject both parental bone marrow allografts when transplanted in limited numbers, by a phenomenon called “hybrid resistance.”7 In F₁ (H2^bxd) mice, Ly49C/I and G2 receptors are also exposed to their “self-ligands” H2^b and H2^d, respectively, in vivo, which transmit inhibitory signals to NK cells. Therefore, it was of interest to examine the effect of depletion of Ly49 NK subsets in a hybrid resistance model, that is, depleting the NK subset that does not mediate the rejection of a particular haplotype BMC (Ly49C/I for H2^b and Ly49G2 for H2^d BMCs) on the rejection of higher numbers of those parental BMCs in F₁(H2^bxd) mice.

We have reported earlier, and also show in Figure3A, that in CB6F₁(H2^bxd) hybrid mice NK cells expressing inhibitory Ly49G2 receptors mediate the rejection of parental H2^b BMCs (0.5 × 10⁶) because depletion of the Ly49G2⁺NK subset in vivo results in the engraftment of the BMCs.26 However, when higher numbers of H2^bBMCs (20 × 10⁶) were transplanted to override the resistance, removal of Ly49G2⁺ NK cells (which plays a role in mediating the rejection of H2^b BMCs in heterozygous F₁ [H2^bxd] and homozygous H2^dmice) did not augment allograft engraftment when compared to the control group (NRS-treated group; Figure 3B). Interestingly, similar to the results in homozygous H2^d mice (Figure 1B), significant rejection of parental H2^b bone marrow allografts occurred (P < .006) in Ly49C/I⁺ NK subset–depleted group when compared with anti-NK1.1–treated group, where all NK cells were depleted.

Weak resistance to parental BALB/c(H2^d) bone marrow allografts has been reported to occur by F₁(H2^bxd) hybrid mice and this rejection can be enhanced by using BMCs that are depleted of T cells.31 We have used C.B-17 SCID (H2^d) BMCs as donors because it has been shown that the rejection pattern of congenitally T cell-deficient C.B-17 SCID BMCs is identical to BALB/c BMCs that were depleted of T cells.31 The BMT results presented in Figure4A demonstrate that as reported earlier, in CB6F₁ (H2^bxd) hybrid mice, NK cells expressing inhibitory Ly49C/I receptors mediate the rejection of parental H2^d BMCs (0.5 × 10⁶) as depletion of Ly49C/I⁺ subset in vivo results in the engraftment of the C.B-17 SCID (H2^d) BMCs.26 Removal of the other Ly49 NK subset, that is, the Ly49G2 NK subset that does not mediate the rejection of H2^d BMCs instead resulted in the rejection of the bone marrow allograft, similar to the control group, whereas syngeneic controls engrafted the BMCs (Figure 4A). Increasing the C.B-17 SCID (H2^d) BMCs to 4 × 10⁶ also resulted in rejection by the host (Figure 4B). However, when parental 4 × 10⁶ BALB/c (H2^d) BMCs were transplanted there was engraftment of H2^d BMCs in the control group (NRS treated) and the reason is weak resistance to parental BALB/c (H2^d) bone marrow allografts by F₁(H2^bxd) hybrid mice.31 Removal of Ly49C/I⁺ NK cells (H2^d rejecting) did not augment allograft engraftment when compared with the control group (NRS-treated group; Figure 4C). Interestingly, as observed in homozygous H2^b mice (Figure 2B-C), removal of Ly49G2⁺ NK cells, which does not mediate the rejection of H2^d BMCs, resulted in enhanced rejection of H2^dbone marrow allografts (P < .02) by the irradiated F₁ recipients (Figure 4C).

Hence, the ability of F₁ (H2^bxd) hybrid mice to reject larger numbers of either parental BMCs (H2^d or H2^b) is enhanced on the depletion of a specific Ly49 NK subset, Ly49G2⁺ and Ly49C/I⁺, respectively.

To ascertain whether enhanced rejection of allogeneic BMCs is indeed due to NK cells versus NK/T or T cells, we performed the BMT experiment in C.B-17 SCID (H2^d) mice, which are deficient in T and B cells.5 Results presented in Figure5 demonstrate that depletion of Ly49C/I⁺ NK subset in C.B-17 SCID (H2^d) mice enhanced their ability to reject large numbers of allogeneic H2^b BMCs when compared with the control group treated with NRS (P < .0001). This is consistent with results also observed in BALB/c (H2^d) mice (Figure 1B). Removal of Ly49G2⁺ NK cells that mediate the rejection of H2^b BMCs in C.B-17 SCID (H2^d) mice did not augment the engraftment further compared with the control group because large numbers (7.5 × 10⁶) of allogeneic BMCs were transplanted. The results with C.B.-17 SCID (H2^d) mice suggest that NK cells, and not T cells, are responsible for the increased marrow rejection as a result of Ly49 NK subset depletion.

We wondered whether the enhanced allograft resistance in the NK subset–depleted mice might be due to an expansion of the Ly49 NK subset, capable of mediating rejection after depletion of the nonrejecting subset. C57BL/6 (H2^b) mice were injected with 200 μg anti-Ly49G2 (4D11 mAb) or 500 μg anti-Ly49C/I (5E6 mAb) and 48 hours later half the mice were lethally irradiated. Seventy-two hours after mAb treatment (24 hours after irradiation) spleen cells were stained from nonirradiated and irradiated mice with various Ly49 fluoresceinated mAbs. Staining with a goat–anti-mouse antibody (to detect mAb 5E6 bound to NK cells in mAb 5E6-treated mice) or goat–anti-rat antibody (to detect mAb 4D11 bound to NK cells in mAb 4D11–treated mice) was completely negative, suggesting that antibody treatment resulted in depletion of the subsets (Figure6A). Results presented in Figure 6B and6C, demonstrate that in vivo treatment of C57BL/6 mice with the mAb 4D11 (anti-Ly49G2) or mAb 5E6 (anti-Ly49C/I) depleted 90% or more of splenic Ly49G2⁺ and Ly49C/I⁺ NK cells, respectively. However, no increase in absolute numbers of splenic Ly49C/I⁺ NK cells was detected in Ly49G2 NK subset–depleted C57BL/6 (H2^b) mice and vice versa. A slight decrease in the total Ly49C/I⁺ NK cell numbers was noted, most likely due to removal of double-positive Ly49C/I⁺G2⁺ cells. Therefore, these data suggest that no compensatory expansion of surviving subsets occurred, at least within the time frame of 3 days.

Interestingly, in C57BL/6 (H2^b) mice in which the Ly49G2 NK subset was removed by mAb 4D11, (Figure 6B,C) a significant increase in the percentage of Ly49C/I⁺ (P < .0001) and Ly49D⁺ NK cells (P < .0001) was noted (control, 25.9% ± 2.4% Ly49C/I⁺ and 19.8% ± 1.0% Ly49D⁺ cells; mAb 4D11 treated, 51.1% ± 2.3% Ly49C/I⁺ and 47.8% ± 2.8% Ly49D⁺ cells). Therefore, enhanced rejection of H2^d BMCs observed in Ly49G2 subset–depleted C57BL/6 (H2^b) mice (Figure 2B-C) may be due at least in part to an increased likelihood of allogeneic BMCs coming in contact with NK cells responsible for rejection.

In agreement with these data, mice receiving the mAb 5E6 (anti-Ly49C/I) also had a significant increase in the percentage of Ly49G2⁺ NK cells (Figure 6B-C), which correlated with the increased ability of H2^d mice to reject H2^b BMC allograft after Ly49C/I⁺ NK subset depletion (Figure 1B-C). The activating receptor for H2^b BMC grafts has not been identified but Ly49H is a candidate because it is similar to Ly49C and Ly49I in the extracellular domain.32 

Next, we did the flow cytometric analysis of Ly49 NK subsets 72 hours following depletion in the spleen cells of lethally irradiated mice. The results presented in Figure 7indicate that the percentage of NK cells following irradiation increased in the untreated (nonirradiated, 3.3%; irradiated, 23.4%) as well as Ly49C/I NK subset–depleted (nonirradiated, 1.4%; irradiated, 6.9%) or Ly49G2 NK subset–depleted group (nonirradiated, 2.6%; irradiated, 15.5%) because of killing of radiosensitive cells by irradiation and removal of NK subsets by Ly49 antibody treatments (Figure 7A). However, the total number of NK cells also decreased in the untreated control group following irradiation (nonirradiated, 2.09 × 10⁶; irradiated, 0.86 × 10⁶) and even further in mAb-treated groups (Figure 7B).

Similar to the effects observed in nonirradiated C57BL/6 (H2^b) mice (Figure 6), in lethally irradiated C57BL/6 (H2^b) mice, Ly49G2 NK subset removal by mAb 4D11 led to a significant increase in the percentage of Ly49C/I⁺ cells, whereas the Ly49C/I NK subset removal by 5E6 mAb treatment led to a significant increase in the percentage of Ly49G2⁺ NK cells (Figure 8A-B). These results indicate that antibody depletion of the subsets was not altered when irradiation was performed.

We have previously described the roles of various Ly49 NK subsets (inhibitory and activating) in bone marrow allograft rejection in lethally irradiated mice.26,27,29 However, all previous studies involved transplanting limiting numbers of allogeneic BMCs, which were rejected by the control (untreated) irradiated mice. During those studies, we observed that in H2^d mice, removal of the inhibitory Ly49C/I NK subset, which played no role in the rejection of allogeneic H2^b grafts, augmented rejection of larger numbers of H2^b BMCs.26 This suggested that such NK subsets are potentially “immunoregulatory” and may functionally regulate the NK cells that mediate marrow graft rejection.

Our previous studies examined the ability of Ly49 NK subset removal to abrogate bone marrow allograft rejection and the work presented here suggests that NK subset modulation can also enhance allograft rejection. We have demonstrated that large numbers of allogeneic BMCs can be strongly rejected by mice of a particular haplotype if the appropriate inhibitory Ly49 NK cells are depleted in vivo. Thus, in homozygous H2^d or heterozygous H2^bxd mice, removal of inhibitory Ly49C/I⁺ NK cells, which do not mediate the rejection of allogeneic H2^b bone marrow grafts, significantly augmented the capability of the mice to reject H2^b BMCs. The opposite effect was observed with homozygous H2^b or heterozygous H2^bxd mice, where removal of inhibitory Ly49G2⁺ NK cells, which do not mediate the rejection of allogeneic H2^d bone marrow grafts, significantly augmented the capability of the mice to reject allogeneic H2^d BMCs. Although in F₁ (H2^bxd) mice, both Ly49 receptors (Ly49C/I and G2) are exposed to their “self-ligands” (parental MHC class I, H2^d and H2^b, respectively) during development, which on interaction transmits inhibitory signals to NK cells, the NK cells were strongly capable of mediating allograft rejection despite living in an environment expressing inhibitory MHC.

It is possible that the enhanced rejection observed by removal of either Ly49G2⁺ or C/I⁺ NK cells is due to either the expansion in cell numbers or the enrichment of the NK subset directly mediating the rejection. Depletion of the Ly49G2⁺NK subset, in H2^b mice (Figures 6 and 7) or the Ly49C/I⁺ NK subset, in H2^d mice (data not shown) did not result in expansion of cell numbers of the other NK subset but did result in enrichment of Ly49C/I⁺ and Ly49D⁺ NK cells (Figures 6 and 8). Presumably, Ly49C/I⁺D⁺ NK cells mediate rejection of H2^d BMCs. Conceivably, the alterations induced by depletion of NK cells expressing a given inhibitory receptor(s) increases the frequency of activating receptor-positive cells or removes inhibition by those inhibitory receptor-positive cells in NK cells capable of rejection.

It is well documented that BMC rejection in mice can be enhanced by treatment with poly I:C in vivo.33,34 Poly I:C treatment activates the macrophages, which produce IFN-α/β, and that enhances NK cell cytotoxicity.33,34 Interestingly, removal of the appropriate Ly49 NK subset enhances the BMC rejection capability of the remaining NK1.1⁺ cells significantly greater than poly I:C–activated cells. NK cells also produce a variety of cytokines on activation, for example, GM-CSF, IFN-γ, tumor necrosis factor α, transforming growth factor β, and so forth.35 Therefore, one mechanism may be novel NK-NK cell immunoregulation whereby one Ly49 NK subset influences another Ly49 NK subset(s) mediating rejection possibly through the production of cytokines or other factors. Removal of that subset relieves the inhibition of NK cells mediating the rejection and results in enhanced rejection of allogeneic BMCs. These studies suggest that caution must be exercised when using other Ly49 mAbs because negative controls in in vivo assays examining the role or function of a particular NK subset because activity of the remaining subsets may be heightened.

We have previously observed that in H2^d mice, Ly49C/I⁺ NK cells produce larger amounts of the growth-promoting cytokine (GM-CSF) than the growth inhibiting cytokine (IFN-γ).36 Similarly, in H2^b mice, Ly49G2⁺ NK cells produce larger amounts of the growth-promoting cytokine (GM-CSF) than growth inhibiting cytokine (IFN-γ; unpublished data, January 1994). Therefore, depletion of Ly49C/I⁺ or Ly49G2⁺ NK cells in a particular haplotype strain of mice may create condition(s) unfavorable for the engraftment of the BMCs.

It is conceivable that following removal of either Ly49 NK subset (Ly49C/I or G2) more than one mechanism may be involved in augmenting the ability of the mice to reject larger numbers of BMCs. Our data suggest that NK subsets are complex in that a particular inhibitory Ly49 NK subset known to play no role in rejection of BMCs of a particular haplotype can apparently prevent rejection of allogeneic marrow. Human NK cells recognizing HLA-C may play a role in marrow graft rejection.37 This study has implications for human NK subset depletion for BMT. Removal of inappropriate NK subsets could have disastrous consequences by enhancing the recipient's ability to reject bone marrow allografts. It is perhaps imperative to create reagents that deplete or inhibit activating but not inhibitory NK cell receptors.

The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. Government. Animal care was provided in accordance with the procedures outlined in the Guide for the Care and Use of Laboratory Animals (NIH publication no. 86-23, 1985).

We thank Drs. Joost Oppenheim, Frank W. Ruscetti,and Crystal Y. Koh for critically reviewing the manuscript and Steve Stull for the superb technical assistance. We are especially grateful to Ms Laura Knott for her excellent secretarial assistance.

We thank Dr Llewellyn Mason for providing 4D11 and PK136 monoclonal antibodies.