NK Cells Rapidly Reject Allogeneic Bone Marrow in the Spleen Through a Perforin- and Ly49D-Dependent, but NKG2D-Independent Mechanism

Authors

  • K. Hamby,

    1. The Emory Transplant Center, Department of Surgery, Emory University School of Medicine, Atlanta, GA
    Search for more papers by this author
  • A. Trexler,

    1. The Emory Transplant Center, Department of Surgery, Emory University School of Medicine, Atlanta, GA
    Search for more papers by this author
  • T. C. Pearson,

    1. The Emory Transplant Center, Department of Surgery, Emory University School of Medicine, Atlanta, GA
    Search for more papers by this author
  • C. P. Larsen,

    1. The Emory Transplant Center, Department of Surgery, Emory University School of Medicine, Atlanta, GA
    Search for more papers by this author
  • M. R. Rigby,

    1. The Emory Transplant Center, Department of Surgery, Emory University School of Medicine, Atlanta, GA
    2. Division of Critical Care Medicine, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA
    Search for more papers by this author
  • L. S. Kean

    Corresponding author
    1. The Emory Transplant Center, Department of Surgery, Emory University School of Medicine, Atlanta, GA
    2. Division of Hematology/Oncology/BMT, The Aflac Cancer Center and Blood Disorders Clinic, Department of Pediatrics, Emory University School of Medicine, Atlanta, GA
      * Corresponding author: Leslie S. Kean, leslie.kean@oz.ped.emory.edu
    Search for more papers by this author

* Corresponding author: Leslie S. Kean, leslie.kean@oz.ped.emory.edu

Abstract

We have used a sensitive and specific in vivo killing assay to monitor the kinetics, anatomic location and mechanisms controlling NK-mediated rejection of Balb/c bone marrow by C57BL/6 natural killer (NK) cells. We find that NK killing of fully allogeneic bone marrow is a rapid, highly efficient process, leading to substantial rejection of transplanted marrow within 6 h of transplant and elimination of 85% of the transplanted cells within 2 days. NK-mediated rejection occurred predominantly in the spleen, with sparing of rejection in the bone marrow and lymph nodes. Rejection was dependent on Perforin gene function, but was independent of interferon-gamma. Finally, rejection of Balb/c bone marrow by B6 NK cells required signaling through the Ly49D receptor, but occurred despite blockade of NKG2D, which distinguishes these results from previous studies using semiallogeneic transplant pairs. These results identify NK cells as highly active mediators of bone marrow rejection, and suggest that inhibiting NK function early during transplantation may increase the efficiency of engraftment and allow successful engraftment of limiting doses of donor bone marrow.

Introduction

In order for stable donor hematopoiesis to occur after allogeneic bone marrow transplantation (BMT), two distinct processes must be accomplished. First, donor cells must home to and engraft in available bone marrow niches. Second, the donor cells must avoid rejection by the recipient immune system. We and others have shown that natural killer (NK) cells may function early after transplant, to reduce the effective stem cell innoculum that is available for engraftment (1–11). This NK effect may become clinically significant during BMT with limited donor cells (such as with umbilical cord blood transplants for hematologic disease (12–14)), or with deceased donor transplants (as could be envisioned with potentially tolerogenic combined bone marrow and solid organ transplantation). In these clinical scenarios, early rejection responses may impact the effective dose of transplanted stem cells, and thus contribute to delayed or even failure of engraftment. Our previous findings indicating that depletion of NK cells can result in successful engraftment after transplant of up to 20-fold fewer bone marrow cells (7) support the hypothesis that this cell population may be of significant clinical importance during transplants involving limiting numbers of stem cells.

That NK cells can recognize and eliminate donor bone marrow has been shown rigorously in rodents (1–11), and by inference in nonhuman primate experiments (15) and in clinical transplantation (16,17). This recognition relies on signaling through both inhibitory and activating receptors. In murine systems, both the Ly49 and CD94 families of inhibitory receptors have been identified as being crucial for recognition and acceptance of self-bone marrow through binding to self-MHC (18–26). The inhibitory receptors also serve an important function during bone marrow rejection: parental bone marrow is rejected during parent-to-offspring transplantation by virtue of the fact that these donor cells are missing the self-MHC recognized by inhibitory NK receptors (27). The identification of the inhibitory receptors in mice ultimately led to the identification and characterization of the functionally analogous class of inhibitory KIR receptors in human NK cells (28). KIR receptor-mediated signaling has been shown to play a role in NK-mediated rejection of hematopoietic cells and in selected studies of BMT for myeloid leukemia, KIR incompatibility allowed greater NK alloreactivity, leading to increased NK-versus-tumor effects (29–31).

In addition to inhibitory receptor-mediated allorecognition, the importance of activating receptors to the mechanisms controlling NK-mediated recognition and elimination of donor cells is being increasingly appreciated (32,33). Both murine and human activating receptors have been identified, and their contribution to NK-mediated recognition and rejection of a variety of foreign cell types has been documented (reviewed in (34)).

The identification of NK activating receptors may have clinical significance, as positive signaling molecules may be one class of receptors that could be potentially targeted to increase bone marrow engraftment by inhibiting NK-mediated stem cell rejection. In murine systems, the Ly49D activating receptor has been shown to be a crucial component of positive signaling resulting in NK alloreactivity when Balb/c bone marrow is transplanted into B6 animals (35–37). NKG2D represents a distinct functional class of activating receptor from the Ly49 family and has recently gained prominence as an important component to both T-cell and NK cell signaling (38–40). This receptor has recently been shown to be important during NK-mediated rejection of B6 parental bone marrow when this marrow is transplanted into F1 offspring that had previously been treated with the simulated viral NK activator, poly-IC (41). A possible mechanism for this NKG2D-mediated rejection of parental bone marrow was also identified, since B6 hematopoietic stem cells were shown to express the NKG2D ligand, RAE-1. These studies had important implications for potential therapeutic targeting of NKG2D during transplantation, as it appeared to be necessary for NK-mediated rejection of donor stem cells. Whether the contribution of NKG2D-mediated rejection extends to unrelated BMT, and in the absence of prior activation of NK cells with poly-IC has not previously been determined.

In the event that the matrix of positive and negative signals integrate into a final cue for NK cytotoxicity, NK cells utilize a variety of mechanisms to promote target cell killing, including perforin-granzyme, Fas-FasL, and cytokines, including IFN-gamma, and TNF-alpha (42–45). In studies using multiple strain combinations and BMT preparative regimens (but all who have used engraftment of donor cells as the read-out for recipient NK-mediated rejection) the extent to which each of these possible modes of cytotoxicity has contributed to rejection has proved to be highly variable, and which have both implicated and refuted the contribution of each mechanism as contributing to NK-mediated bone marrow rejection. However, none of these studies have examined been able to examine in vivo, the mechanisms involved in rapid NK-mediated rejection of transplanted cells.

In this report, we use a fully allogeneic murine BMT model and describe the timing, place, and molecular mechanism of NK-mediated bone marrow rejection. This study benefits from a highly sensitive and rigorous assay of NK alloreactivity able to measure NK killing in vivo, without requiring prior NK activation with agents such as poly I:C. In addition to the killing assay, we've also assessed the mechanisms controlling NK alloreactivity during rejection of transplanted bone marrow. A large variety of transplant scenarios have been used to measure NK alloreactivity. Some have studied transplant between fully allogeneic donors and recipients in the context of an irradiation-based preparative regimen (in which NK cells are less radio-sensitive than T cells, thus, NK alloreactivity persists despite the preparative regimen) (46). Others have used ‘hybrid resistance’ transplant systems, in which parental bone marrow is transplanted into offspring recipients. In this scenario, T-cell rejection is negligible, but NK-based rejection persists (2,41). In both our previous work (7) and in this study, we have utilized a nonmyeloablative, T-cell costimulation blockade-based BMT system in which to study the effect of NK cells on transplant rejection. Costimulation blockade effectively eliminates T-mediated rejection, while NK-based rejection responses remain (7) thus providing a model of NK alloreactivity with particular relevance for the growing field of non-myeloablative BMT and also for chimerism-based models of tolerance-induction for solid organ transplantation The studies described herein identify a highly active NK-driven machinery for bone marrow rejection between unrelated allogeneic transplant pairs that is independent of NKG2D, in contrast to mechanisms previously described for parent-to-offspring transplants, that required NKG2D. While independent of NKG2D-based signaling, NK-dependent rejection in this system did require both Ly49D- and perforin-based signaling in order to effectively eliminate transplanted allogeneic target cells.

Methods

Animals

C57BL/6 (B6), Balb/c, C57BL/6-Prf1<tm1Sdz>J (Perforin), B6Smn.C3-Fasl<gld>/J (Fas-Ligand), and B6.129S7-Ifngtm1Ts/J (GKO) were purchased from Jackson Laboratories (Bar Harbor, ME). The lymph node deficient alymphoplasia (aly/aly) mice (47) were a kind gift from Dr. Kenneth Newell, Emory University. Splenectomy was performed on B6 mice through a small incision on the left lateral aspect of the abdomen, followed by suture closure of the peritoneal cavity and the overlying skin.

Antibodies and other reagents

PK136 (anti-NK1.1), GK1.5 (anti-CD4), 53.6.72 (anti-CD8), MR1 (anti-CD154), and UC10-4F10-11 (CTLA4-Ig) were purchased from BioExpress Inc (West Lebanon, NH). R4-6A2 (anti-IFN-gamma) was produced at by Dr. Robert Mittler of Emory University. CX5 (anti-NKG2D) was purchased from eBioscience (San Diego, CA). 4E5 (anti-Ly49D) was purchased from BD Pharmingen (San Diego, CA). Antibody dosing prior to in vivo killing experiments or transplant experiments was as followed: Anti-CD4, anti-CD8, anti-NK1.1: 100 μg intraperitoneally (ip) on days −2, −1 and 0 relative to transplant for the killing assays, and 100 μg intraperitoneally (ip) on days −2, −1, 0, 2, 4, 6, 14, 28 relative to transplant for donor chimerism assays. Anti-IFN-gamma, anti-CD154 and CTLA4-Ig: 500 μg ip on days −2, −1 and 0 relative to transplant for the killing assays and 500 μg ip on days −2, −1, 0, 2, 4, 6, 14, 28 relative to transplant for donor chimerism assays. Anti-NKG2D and anti-Ly49D: 200 μg ip on days −2 and 0 relative to transplant for the NK killing assays or on days −2, 0 and 6 relative to transplant for donor chimerism assays. All antibodies used for flow cytometric analysis were purchased from Pharmingen Inc (San Diego, CA). CFSE and Far-Red were purchased from Invitrogen (Eugene, OR).

NK in vivo killing assay

The in vivo NK alloreactivity assay, which was adapted from previously described memory T-cell and NK cell killing assays (8,48), used B6 mice as transplant recipients and both B6 and Balb/c mice as bone marrow donors. Bone marrow was prepared from both B6 and Balb/c mice, and labeled with 5 μM CFSE and 1 μg/ml Far-Red (B6 bone marrow) or 5 μM CFSE alone (Balb/c bone marrow). 10 × 106 cells of both the labeled B6 and the labeled Balb/c bone marrow were combined and transplanted together into recipient animals. Flow cytometric analysis was performed on this mixture to document the delivery of equivalent cell doses to recipients (Figure 1). At designated times after transplant (1, 6 h, 24 h and 48 h) splenocytes or bone marrow cells were prepared from recipient animals, and analyzed flow cytometrically on a FACScalibur flow cytometer (Becton Dickinson, San Diego, CA). Two million events were analyzed for each sample, to ensure adequate quantification of the adoptively transferred populations. The B6 and Balb/c cells were distinguished as follows: B6 cells were CFSE+/Far-Red+ and Balb/c cells were CFSE+/Far-Red-negative. As can be seen in Figure 1, CFSE and Far-red both labeled the bone marrow populations differentially, such that there were multiple CSFE+/Far-Red+ B6 sub-populations and multiple CFSE+/Far-Red-negative Balb/c sub-populations noted on flow cytometric analysis. This likely occurred due to the heterogeneous nature of the bone marrow product, with cells of varying sizes and densities and in multiple phases of the cell cycle, all which will contribute to the efficiency of dye uptake. Control experiments in which only B6 bone marrow was labeled and then transferred into B6 recipients confirmed the identification of all CFSE/Far-Red-double positive cells as B6 in origin (with Far-Red fluorescence >101, Figure 1). Thus, although the bone marrow labeled heterogeneously with CFSE and Far-Red, the identification of the origin (B6 vs. Balb/c) of the labeled cells was unambiguous, and thus differential killing of the two populations could be unambiguously determined. The percent of B6 and Balb/c cells present in the fluorescent population was determined using WinList analysis software (Verity, Topsham, ME) or FlowJo software (Ashland, OR). The percent killing was determined through the following formula:

image
Figure 1.

Figure 1.

In vivo killing assay: (A) As described in Methods, we utilized an in vivo killing assay to measure NK cytotoxicity in an allogeneic bone marrow transplant model. Specific labeling of B6 bone marrow with CFSE + Far-Red and Balb/c bone marrow with CFSE alone allowed flow cytometric distinction of the two populations. Top row: Representative flow cytometric dot plots of labeled populations prior to transplantation are shown: Left panel: B6 alone; Middle panel: Balb/c alone; Right panel: Adoptive transfer of an equal number of B6 and Balb/c bone marrow cells. Bottom row: Representative flow cytometric dot plot of B6 bone marrow alone transferred into a B6 recipient and analyzed 1 day after transfer. This panel shows that despite a loss of Far-Red fluorescence after adoptive transfer, B6 cells are still double positive (CFSE+/Far-Red+) and thus can be distinguished from transferred Balb/c cells. (B) Representative flow cytometric dot plots demonstrating in vivo killing results. Syngeneic B6 bone marrow is identified as CFSE+/Far-Red+. Allogeneic Balb/c bone marrow is identified as CFSE+/Far-Red-negative. Dot plots for representative animals receiving no antibody pretreatment, NK depletion, T-cell depletion and both NK and T-cell depletion are shown. (C) The in vivo killing assay is NK-specific. The percentage-specific killing of transplanted allogeneic Balb/c bone marrow by B6 recipients was calculated as described in Methods. Shown is the percentage killing (±SEM) for animals receiving no antibody pretreatment (n = 46), NK depletion (n = 18), T-cell depletion (n = 28) and both NK and T-cell depletion (n = 29). (D) Flow cytometry showing the effect of T-cell depletion with anti-CD4 and anti-CD8 antibodies or of NK depletion with anti-NK1.1. Top panels: representative flow cytometry showing CD4+ and CD8+ cells (previously gated on CD3+ cells) either before anti-CD4 and anti-CD8 antibody treatment (left panel) or after 3 days of antibody treatment (right panel). Bottom panels: representative flow cytometry showing CD3 and NK1.1 labeling either before treatment with anti-NK1.1 antibody (left panel) or after 3 days of antibody treatment (right panel).

Figure 1.

Figure 1.

In vivo killing assay: (A) As described in Methods, we utilized an in vivo killing assay to measure NK cytotoxicity in an allogeneic bone marrow transplant model. Specific labeling of B6 bone marrow with CFSE + Far-Red and Balb/c bone marrow with CFSE alone allowed flow cytometric distinction of the two populations. Top row: Representative flow cytometric dot plots of labeled populations prior to transplantation are shown: Left panel: B6 alone; Middle panel: Balb/c alone; Right panel: Adoptive transfer of an equal number of B6 and Balb/c bone marrow cells. Bottom row: Representative flow cytometric dot plot of B6 bone marrow alone transferred into a B6 recipient and analyzed 1 day after transfer. This panel shows that despite a loss of Far-Red fluorescence after adoptive transfer, B6 cells are still double positive (CFSE+/Far-Red+) and thus can be distinguished from transferred Balb/c cells. (B) Representative flow cytometric dot plots demonstrating in vivo killing results. Syngeneic B6 bone marrow is identified as CFSE+/Far-Red+. Allogeneic Balb/c bone marrow is identified as CFSE+/Far-Red-negative. Dot plots for representative animals receiving no antibody pretreatment, NK depletion, T-cell depletion and both NK and T-cell depletion are shown. (C) The in vivo killing assay is NK-specific. The percentage-specific killing of transplanted allogeneic Balb/c bone marrow by B6 recipients was calculated as described in Methods. Shown is the percentage killing (±SEM) for animals receiving no antibody pretreatment (n = 46), NK depletion (n = 18), T-cell depletion (n = 28) and both NK and T-cell depletion (n = 29). (D) Flow cytometry showing the effect of T-cell depletion with anti-CD4 and anti-CD8 antibodies or of NK depletion with anti-NK1.1. Top panels: representative flow cytometry showing CD4+ and CD8+ cells (previously gated on CD3+ cells) either before anti-CD4 and anti-CD8 antibody treatment (left panel) or after 3 days of antibody treatment (right panel). Bottom panels: representative flow cytometry showing CD3 and NK1.1 labeling either before treatment with anti-NK1.1 antibody (left panel) or after 3 days of antibody treatment (right panel).

As shown in Figure 1, the fluorescence intensity of both the CFSE and Far-Red tracking dyes did decrease after adoptive transfer and incubation in the recipient animals for up to 48 h after transfer. However, the two populations were still easily distinguishable, given the fact that the Balb/c cells were Far-Red-negative, and the B6 cells were Far-Red-positive (albeit with lower fluorescence intensity than was present pre-transplant). To verify the NK-specificity of the killing assay, various antibody treatments were used: Each experiment included animals left untreated, as well as those treated with anti-CD4 and anti-CD8 antibodies to deplete T cells (thus isolating cytotoxicity to the non-T-cell populations). The NK-specificity of killing was ascertained by including recipients in each killing experiment that had NK cells depleted with the PK136 anti-NK1.1 antibody.

BMT technique

BMT was performed as previously described (7,49). Briefly, antibody pretreatment of recipient B6 animals was performed as described above. Then, on the day of transplant a suboptimal dose of 2 × 106 Balb/c bone marrow cells (whole bone marrow, obtained from donor by flushing tibias and femurs with Hanks balanced salt solution (Sigma, St. Louis, MO) by using a 25-guage needle and syringe) were transplanted into B6 mice that had not been pretreated with antibodies, or into B6 mice pretreated with the depleting and/or inhibitory antibodies described above. In addition to B6 recipients, some experiments used mutant strains (all on the B6 background) including Perforin-deficient mice, Fas-ligand-deficient mice and IFN-gamma-deficient (GKO) mice. To induce T-cell tolerance to the graft, costimulation blockade with 500 μg/dose of the anti-CD154 antibody and CTLA4-Ig was given on days −2, −1, 0, 2, 4, 6, 14 and 28 relative to the transplant, and pretransplant nonmyeloablative preparation with 20 mg/kg Busulfan was given on day-1 relative to the bone marrow transplant. This treatment alone (in the absence of NK-depletion) did not affect NK cytotoxicity (data not shown). WBC chimerism was monitored by flow cytometric analysis of CD45+ cells (total WBC chimerism), B220+ cells (B cells chimerism), CD3+ cells (T-cell chimerism) and Mac-1/GR1+ cells (macrophage/granulocyte chimerism) bearing the Balb/c-specific MHC allele, H-2Kd using lineage-specific and H-2Kd antibodies (Beckton Dickinson).

Statistical analysis

The Prism software system (Graphpad Software Inc., San Diego CA) was used to apply the students T test and the Mann-Whitney test to determine the statistical significance of the differences in percent killing measured with the various treatment conditions.

Results

NK cells rapidly reject fully allogeneic bone marrow

Using an in vivo killing assay with the capacity to rapidly and specifically measure NK alloreactivity we were able to determine the kinetics of NK-mediated killing of fully allogeneic bone marrow in a murine model. In this model, the donor cells were Balb/c bone marrow cells which were transplanted into recipient B6 animals along with equal numbers of syngeneic B6 bone marrow cells. The simultaneous transplant of these two populations allowed allo-specific NK alloreactivity to be measured in each experiment. To differentiate the two cell populations, the syngeneic bone marrow was fluorescently labeled with both the CFSE and Far-Red fluorescent tracking dyes, while the allogeneic cells were labeled with CFSE only (see Methods for details of labeling and population identification). Thus, differential killing of allogeneic (CFSE+/Far-Red-) versus syngeneic (CFSE+/Far-Red+) bone marrow could be easily performed (Figure 1). By comparison with the ratio of allogeneic versus syngeneic bone marrow in the transplanted innoculum (determined flow-cytometrically on the day of transplant, Figure 1A) the% killing of allogeneic cells was determined as described in Methods.

The extent to which killing was NK-specific was determined by incorporating 4 treatment conditions into each experiment (Figure 1B–D): (1) No antibody pretreatment (2), Pretreatment with anti-CD4 and anti-CD8 antibodies to deplete T cells, thus revealing the extent of killing that was T cell-independent (3). Pretreatment with the anti-NK1.1 antibody PK136 either alone or (4) in addition to T-cell depletion with anti-CD4 and anti-CD8 antibodies to document that the killing measured was NK-specific. Figure 1D confirms the efficiency of NK depletion with anti-NK1.1 (82%) and T-cell depletion with anti-CD4 and anti-CD8 antibodies (92%). As shown in Figure 1B–C, when measured at 48 h posttransplant, the majority of the killing of allogeneic cells was NK-specific. Thus, very similar amounts of killing occurred when either no antibody pretreatment occurred, or in the setting of specific T-cell depletion using anti-CD4 and anti-CD8 antibodies (85% vs. 84% killing respectively, Figure 1B–C). That this killing was NK-mediated was confirmed by the inhibition of killing that occurred if NK cells were depleted prior to bone marrow transfer (1% killing, Figure 1B–C).

The kinetics of NK-mediated killing was very rapid. Forty-eight percent allo-specific killing occurred within 6 h after bone marrow transplant, with the majority of killing occurring within 24 h after transplant, when killing was measured in the spleen (Figure 2A). Recipient bone marrow represented a relative sanctuary from NK killing, with significantly less killing apparent in the bone marrow at 6 h after transplant (25% in the bone marrow vs. 48% in the spleen at 6 h, 48% killing in the bone marrow vs. 82% in the spleen at 24 h, Figure 2A). By 48 h, the amount of killing in the bone marrow was reflective of what was occurring elsewhere in the animal.

Figure 2.

Rapid NK-mediated killing of allogeneic targets occurs, predominantly in the spleen. (A) Time-course of in vivo killing in the bone marrow and spleen. Shown is the percentage killing measured at 1, 6, 24 and 48 h after transplant. Shown are results for a single representative experiment, n = 3 ± standard deviation. (B) Representative flow cytometric analysis of in vivo killing in splenectomized B6 mice and the aly/aly lymph node-deficient mutant strain. Shown are representative dot plots demonstrating the relative survival of transplanted syngeneic B6 bone marrow and allogeneic Balb/c bone marrow (measured in the bone marrow 48 h after transplantation). (C) NK killing occurs predominantly in the spleen. Shown in columns, left to right, in the figure, is the% killing of Balb/c target cells in the spleens of normal B6 mice (n = 46, ± SEM); aly/aly mice with killing measured in the spleen (n = 6, ± SEM); B6 mice with killing measured in the bone marrow (n = 12, ± SEM); B6 mice that had been previously splenectomized (killing measured in the bone marrow, n = 8, ± SEM); and aly/aly mice with killing measured in the bone marrow (n = 6, ± SEM).

The spleen represented a major anatomic site for NK killing

In addition to identifying the bone marrow as a relative sanctuary site for the transplanted allogeneic bone marrow, the contribution of the lymph nodes and the spleen to NK-mediated killing were both tested directly. To determine the requirement for lymph nodes in NK-killing of allogeneic targets, lymph-node deficient aly/aly mutant mice (47) were used as transplant recipients. To determine the contribution of the spleen to NK killing, mice were surgically splenectomized 5 days prior to transplant and performance of the NK killing assay. As shown in Figure 2B–C, splenectomized mice displayed a significant inhibition of their NK cells' ability to reject allogeneic bone marrow while aly/aly mice demonstrated equivalent killing efficiency as wild-type mice when killing was measured either in the bone marrow or spleen. Thus, splenectomized mice demonstrated only 12% killing was measured in the bone marrow 48 h after bone marrow transplant, compared to 76% in the nonsplenectomized mice (Figure 2C). These results have important implications for the mechanism of NK-mediated bone marrow rejection, especially in light of recent reports of important interactions occurring between NK cells and T cells that occur in the secondary lymph organs, including lymph nodes (50).

Our results reveal an independence from T-cell mediated effects, which is supported both by the observation of efficient killing in the setting of T-cell depletion and in the independence of NK killing on the existence of lymph nodes.

NK-killing of fully allogeneic target cells is significantly inhibited in the absence of perforin gene function, but is independent of IFN-gamma

In viral immunity models, NK cells have been shown to have two mechanisms of producing cytotoxicity: through direct cytotoxic effects (often mediated through the Perforin/granzyme pathway) and through cytokine-mediated effects, during which NK-derived cytokines can have direct cytopathic effects and can also serve an immunomodulatory role on other immune cells (reviewed in (51)). Interferon-gamma has been identified as a key immunomodulatory cytokine produced by NK cells, and serves as part of the intercellular signaling cascade linking NK cells both to other members of the innate immune system, as well as to T cells and other members of the adaptive immune system (52,53). In NK-mediated rejection of allogeneic bone marrow cells, the mechanism by which stem cell elimination occurs has remained undetermined. To explore the mechanism by which NK cells kill allogeneic targets, we pretreated recipient animals with anti-IFN-gamma monoclonal antibodies and also performed killing assays in the IFN-gamma deficient GKO mice, as well as in Perforin-deficient, and Fas-ligand-deficient mutant strains. As shown in Figure 3 A–G, while IFN-gamma did not play a significant role in NK-killing of allogeneic target cells (92% and 87% killing despite antibody-mediated or genetic absence of IFN-gamma, respectively), Perforin was involved in NK-mediated killing of allogeneic bone marrow (killing was inhibited by 53% in the setting of perforin-deficient NK cells). NK-mediated killing was inhibited to a small, but statistically significant extent in the Fasl<gld>/J Fas-ligand mutant strain (79% killing compared to 89% in the B6 control strain, p = .0015). As shown in Figure 3H–I (and in agreement with ours and others previous results, (1–11)), the decrease in target cell killing that occurred with NK depletion lead to increased bone marrow engraftment. This engraftment was multilineage (data not shown) and was stable for >180 days after transplant (Figure 3H). Figure 3I compares WBC chimerism at 30 days in transplant recipients lacking NK cells (through depletion with the anti-NK1.1 PK136 antibody), lacking perforin gene function or lacking Fas-ligand gene function. As noted above, while Fas-ligand mutants exhibited a small inhibition of NK-mediated target cell killing, the clinical significance of this Fas-ligand-dependent small decrease in killing was likely negligible, as no increases in chimerism-induction were observed in the FasL-mutant strain compared to B6 controls (Figure 3I). In contrast, the decrease in NK-mediated killing noted in the perforin-deficient strain did translate into an effect on bone marrow engraftment: As shown in Figure 3I, perforin deficient mice exhibited increased levels of donor total WBC (CD45+) chimerism (44%) compared to B6 mice (8%) when measured after 30 days in an NK-dependent model of transplant rejection.

Figure 3.

Figure 3.

NK-mediated killing is Perforin-dependent. (A–F) In vivo killing analysis was performed in B6 mice receiving the following: (A) no antibody pretreatment (n = 46, ± SEM); (B) NK depletion with PK136 (n = 18, ± SEM); (C) anti-IFN-gamma monoclonal antibody treatment (n = 6, ± SEM); (D) IFN-gamma knock-out mice (GKO, n = 6, ± SEM); (E) Fas-ligand mutant mice (n = 9, ± SEM); and (F) in Perforin mutant mice (n = 9, ± SEM). Figure 3A–F depicts dot plots showing survival of syngeneic B6 bone marrow and allogeneic Balb/c bone marrow in the different strains of mice and with the different antibody treatments. Figure 3G shows the average percentage killing for each of the different mouse strains with statistical analysis also shown. (H) Time-course of Balb/c chimerism in B6 recipients in the presence or absence of NK depletion. Balb/c donor (H-2Kd+) chimerism in CD45+ cells was measured flow cytometrically at the time-points indicated after transplantation of limiting numbers (2 × 106) of whole bone marrow as described in Methods. This was performed in animals that were either unmanipulated (open circles) or after NK-cell depletion (closed circles). Shown are the averages for each time-point, ± SEM (n = 4–5 mice per time-point). (I) Significant donor chimerism-induction in Perforin mutant mice when transplanted with limiting doses of Balb/c bone marrow. Mice from B6, Perforin deficient, and Fas-ligand deficient strains were transplanted using a nonmyeloablative transplant strategy and a suboptimal dose of donor bone marrow. Peripheral blood WBC chimerism on day 28 posttransplant is shown (n = 4–5 per group, ± SEM with statistical significance also indicated).

Figure 3.

Figure 3.

NK-mediated killing is Perforin-dependent. (A–F) In vivo killing analysis was performed in B6 mice receiving the following: (A) no antibody pretreatment (n = 46, ± SEM); (B) NK depletion with PK136 (n = 18, ± SEM); (C) anti-IFN-gamma monoclonal antibody treatment (n = 6, ± SEM); (D) IFN-gamma knock-out mice (GKO, n = 6, ± SEM); (E) Fas-ligand mutant mice (n = 9, ± SEM); and (F) in Perforin mutant mice (n = 9, ± SEM). Figure 3A–F depicts dot plots showing survival of syngeneic B6 bone marrow and allogeneic Balb/c bone marrow in the different strains of mice and with the different antibody treatments. Figure 3G shows the average percentage killing for each of the different mouse strains with statistical analysis also shown. (H) Time-course of Balb/c chimerism in B6 recipients in the presence or absence of NK depletion. Balb/c donor (H-2Kd+) chimerism in CD45+ cells was measured flow cytometrically at the time-points indicated after transplantation of limiting numbers (2 × 106) of whole bone marrow as described in Methods. This was performed in animals that were either unmanipulated (open circles) or after NK-cell depletion (closed circles). Shown are the averages for each time-point, ± SEM (n = 4–5 mice per time-point). (I) Significant donor chimerism-induction in Perforin mutant mice when transplanted with limiting doses of Balb/c bone marrow. Mice from B6, Perforin deficient, and Fas-ligand deficient strains were transplanted using a nonmyeloablative transplant strategy and a suboptimal dose of donor bone marrow. Peripheral blood WBC chimerism on day 28 posttransplant is shown (n = 4–5 per group, ± SEM with statistical significance also indicated).

NK killing of fully allogeneic targets was dependent on the LY49D activating receptor, but occurred despite blockade of NKG2D

The Ly49D activating receptor (present on approximately 50% of B6 NK cells, Figure 4F and (54)) has previously been shown to be an important positive regulator of NK-mediated alloreactivity, especially towards Balb/c target cells (reviewed in (36)). Recently, the NKG2D receptor (present on the vast majority of B6 NK cells, Figure 4A and (55)) has been implicated as being crucial for NK-mediated recognition of Balb/c stem cells in a modified hybrid-resistance model of NK-mediated bone marrow rejection (41). In that study, NK cells were first activated by poly-IC, an agent known to mimic mechanisms of viral-mediated NK activation. We sought to measure the requirement of NKG2D during fully allogeneic bone marrow transplant in a system devoid of external NK activation. As shown in Figure 4, while we were able to efficiently block NKG2D using the CX5 anti-NKG2D monoclonal antibody (Figure 4A), NK cells were still able to eliminate Balb/c target cells despite this blockade (Figure 4B–C). While the extent of in vivo killing that occurred in the presence of anti-NKG2D was slightly lower than without blockade (82 ± 4% with anti-NKG2D versus 89%± 1% without, which, given the large number of animals tested (10 animals with anti-NKG2D, 46 animals without) did reach statistical significance (p = 0.01)), blockade of this molecule did not recapitulate the inhibition of killing observed with NK depletion (Figure 4C). Further supporting an NKG2D-independent mechanism of NK-mediated rejection of fully allogeneic bone marrow cells was the observation that blockade of NKG2D with the CX5 antibody did not appreciably increase the engraftment efficiency of limiting doses of Balb/c bone marrow, while whole-scale NK depletion did allow engraftment after transplant with limiting donor bone marrow doses (Figure 4D–E). The confirmation that in our assay system, blockade of the Ly49D NK activating receptor with anti-Ly49D antibodies did result in significant inhibition of NK-killing as has been previously reported (without NK depletion, Figure 4F–H) underscores the specificity of these results. These two experiments support a model in which in the fully allogeneic setting, while Ly49D is an important component of NK-mediated signaling and cytotoxicty, these cascades can function without signaling through NKG2D.

Figure 4.

Figure 4.

Mechanism of NK killing of allogeneic bone marrow.Figure 4A–D: NK killing of Balb/c bone marrow by a B6 recipient occurs despite blockade of NKG2D. (A) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and NKG2D expression on CD3-negative cells after either no antibody pretreatment, NK depletion, or blockade of NKG2D.(B) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-NKG2D. (C) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-NKG2D (n = 10 ± SEM). (D) Blockade of NKG2D does not increase donor chimerism. Top panel: Average donor WBC chimerism at 30 days after transplant in animals receiving no pretreatment (n = 9 ± SEM), NK depletion (n = 8 ± SEM), or treatment with anti-NKG2D (n = 6 ± SEM). Bottom panel: Representative flow cytometric analysis of Balb/c chimerism (H-2kd) in the three experimental conditions. Plots of CD45 versus the Balb/c-specific MHC marker H-2Kd are shown. (E) Time-course of Balb/c chimerism in B6 recipients in the presence or absence of NK depletion or blockade of NKG2D. Balb/c donor (H-2Kd+) chimerism in CD45+ cells was measured flow cytometrically at the time-points indicated after transplantation of limiting numbers (2 × 106) of whole bone marrow as described in Methods. This was performed in animals that were either unmanipulated (open circles), after NK-cell depletion (closed circles), or after blockade of NKG2D (open squares). Shown are the averages for each time-point, ± SEM (n = 4–5 mice per time-point). Figure 4F–H: NK killing of Balb/c bone marrow by a B6 recipient requires Ly49D. (F) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and Ly49D expression on CD3-negative cells after, either no antibody pretreatment, NK depletion or blockade of Ly49D. (G) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-Ly49D. (H) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-Ly49D (n = 6 ± SEM).

Figure 4.

Figure 4.

Mechanism of NK killing of allogeneic bone marrow.Figure 4A–D: NK killing of Balb/c bone marrow by a B6 recipient occurs despite blockade of NKG2D. (A) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and NKG2D expression on CD3-negative cells after either no antibody pretreatment, NK depletion, or blockade of NKG2D.(B) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-NKG2D. (C) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-NKG2D (n = 10 ± SEM). (D) Blockade of NKG2D does not increase donor chimerism. Top panel: Average donor WBC chimerism at 30 days after transplant in animals receiving no pretreatment (n = 9 ± SEM), NK depletion (n = 8 ± SEM), or treatment with anti-NKG2D (n = 6 ± SEM). Bottom panel: Representative flow cytometric analysis of Balb/c chimerism (H-2kd) in the three experimental conditions. Plots of CD45 versus the Balb/c-specific MHC marker H-2Kd are shown. (E) Time-course of Balb/c chimerism in B6 recipients in the presence or absence of NK depletion or blockade of NKG2D. Balb/c donor (H-2Kd+) chimerism in CD45+ cells was measured flow cytometrically at the time-points indicated after transplantation of limiting numbers (2 × 106) of whole bone marrow as described in Methods. This was performed in animals that were either unmanipulated (open circles), after NK-cell depletion (closed circles), or after blockade of NKG2D (open squares). Shown are the averages for each time-point, ± SEM (n = 4–5 mice per time-point). Figure 4F–H: NK killing of Balb/c bone marrow by a B6 recipient requires Ly49D. (F) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and Ly49D expression on CD3-negative cells after, either no antibody pretreatment, NK depletion or blockade of Ly49D. (G) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-Ly49D. (H) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-Ly49D (n = 6 ± SEM).

Figure 4.

Figure 4.

Mechanism of NK killing of allogeneic bone marrow.Figure 4A–D: NK killing of Balb/c bone marrow by a B6 recipient occurs despite blockade of NKG2D. (A) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and NKG2D expression on CD3-negative cells after either no antibody pretreatment, NK depletion, or blockade of NKG2D.(B) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-NKG2D. (C) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-NKG2D (n = 10 ± SEM). (D) Blockade of NKG2D does not increase donor chimerism. Top panel: Average donor WBC chimerism at 30 days after transplant in animals receiving no pretreatment (n = 9 ± SEM), NK depletion (n = 8 ± SEM), or treatment with anti-NKG2D (n = 6 ± SEM). Bottom panel: Representative flow cytometric analysis of Balb/c chimerism (H-2kd) in the three experimental conditions. Plots of CD45 versus the Balb/c-specific MHC marker H-2Kd are shown. (E) Time-course of Balb/c chimerism in B6 recipients in the presence or absence of NK depletion or blockade of NKG2D. Balb/c donor (H-2Kd+) chimerism in CD45+ cells was measured flow cytometrically at the time-points indicated after transplantation of limiting numbers (2 × 106) of whole bone marrow as described in Methods. This was performed in animals that were either unmanipulated (open circles), after NK-cell depletion (closed circles), or after blockade of NKG2D (open squares). Shown are the averages for each time-point, ± SEM (n = 4–5 mice per time-point). Figure 4F–H: NK killing of Balb/c bone marrow by a B6 recipient requires Ly49D. (F) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and Ly49D expression on CD3-negative cells after, either no antibody pretreatment, NK depletion or blockade of Ly49D. (G) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-Ly49D. (H) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-Ly49D (n = 6 ± SEM).

Figure 4.

Figure 4.

Mechanism of NK killing of allogeneic bone marrow.Figure 4A–D: NK killing of Balb/c bone marrow by a B6 recipient occurs despite blockade of NKG2D. (A) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and NKG2D expression on CD3-negative cells after either no antibody pretreatment, NK depletion, or blockade of NKG2D.(B) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-NKG2D. (C) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-NKG2D (n = 10 ± SEM). (D) Blockade of NKG2D does not increase donor chimerism. Top panel: Average donor WBC chimerism at 30 days after transplant in animals receiving no pretreatment (n = 9 ± SEM), NK depletion (n = 8 ± SEM), or treatment with anti-NKG2D (n = 6 ± SEM). Bottom panel: Representative flow cytometric analysis of Balb/c chimerism (H-2kd) in the three experimental conditions. Plots of CD45 versus the Balb/c-specific MHC marker H-2Kd are shown. (E) Time-course of Balb/c chimerism in B6 recipients in the presence or absence of NK depletion or blockade of NKG2D. Balb/c donor (H-2Kd+) chimerism in CD45+ cells was measured flow cytometrically at the time-points indicated after transplantation of limiting numbers (2 × 106) of whole bone marrow as described in Methods. This was performed in animals that were either unmanipulated (open circles), after NK-cell depletion (closed circles), or after blockade of NKG2D (open squares). Shown are the averages for each time-point, ± SEM (n = 4–5 mice per time-point). Figure 4F–H: NK killing of Balb/c bone marrow by a B6 recipient requires Ly49D. (F) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and Ly49D expression on CD3-negative cells after, either no antibody pretreatment, NK depletion or blockade of Ly49D. (G) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-Ly49D. (H) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-Ly49D (n = 6 ± SEM).

Figure 4.

Figure 4.

Mechanism of NK killing of allogeneic bone marrow.Figure 4A–D: NK killing of Balb/c bone marrow by a B6 recipient occurs despite blockade of NKG2D. (A) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and NKG2D expression on CD3-negative cells after either no antibody pretreatment, NK depletion, or blockade of NKG2D.(B) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-NKG2D. (C) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-NKG2D (n = 10 ± SEM). (D) Blockade of NKG2D does not increase donor chimerism. Top panel: Average donor WBC chimerism at 30 days after transplant in animals receiving no pretreatment (n = 9 ± SEM), NK depletion (n = 8 ± SEM), or treatment with anti-NKG2D (n = 6 ± SEM). Bottom panel: Representative flow cytometric analysis of Balb/c chimerism (H-2kd) in the three experimental conditions. Plots of CD45 versus the Balb/c-specific MHC marker H-2Kd are shown. (E) Time-course of Balb/c chimerism in B6 recipients in the presence or absence of NK depletion or blockade of NKG2D. Balb/c donor (H-2Kd+) chimerism in CD45+ cells was measured flow cytometrically at the time-points indicated after transplantation of limiting numbers (2 × 106) of whole bone marrow as described in Methods. This was performed in animals that were either unmanipulated (open circles), after NK-cell depletion (closed circles), or after blockade of NKG2D (open squares). Shown are the averages for each time-point, ± SEM (n = 4–5 mice per time-point). Figure 4F–H: NK killing of Balb/c bone marrow by a B6 recipient requires Ly49D. (F) Phenotypic analysis of the NK compartment after antibody treatment. Shown are representative dot-plots of NK1.1 expression and Ly49D expression on CD3-negative cells after, either no antibody pretreatment, NK depletion or blockade of Ly49D. (G) Representative flow cytometric analysis of in vivo killing after either no antibody pretreatment, depletion of NK cells, or treatment with anti-Ly49D. (H) Average in vivo killing in the presence of no antibody pretreatment (n = 46 ± SEM), NK depletion (n = 46 ± SEM), or treatment with anti-Ly49D (n = 6 ± SEM).

Discussion

We have used both an in vivo assay capable of measuring NK-mediated killing of allogeneic targets and a BMT model that has previously shown to be a sensitive indicator of NK-mediated rejection (7) to determine the mechanisms by which NK cells target and kill allogeneic bone marrow. Our results show that NK cells are highly active mediators of early rejection of donor bone marrow after transplantation and are consistent with a model of bone marrow engraftment that places NK cells as the ‘forward guard’ in the gauntlet of first-pass metabolism that stem cells must face before successfully engrafting in a bone marrow niche. That they are highly effective in their role as early mediators of rejection of allogeneic cells is supported by the observation that a significant portion of the transplanted innoculum was eliminated within the first 6 h after transplant and almost half of the transplanted cells were eliminated within 24 h after transplant. This efficient NK-mediated killing occurred in the absence of T cells, and independently of lymph nodes, implying a highly autonomous rejection function in which the spleen functions as a major site of the ‘first-pass’ effect. Consistent with this is the fact that rejection did not require the cytokine IFN-gamma, and functioned largely through a Perforin-mediated mechanism, with a small (but potentially clinically insignificant) contribution of signaling through the Fas-ligand pathway. The Perforin-dependence and IFN-gamma independence of the killing is consistent with its rapid tempo, and support a model of direct NK-mediated allo-recognition followed by rapid cytotoxicity.

Unlike recent studies that implicated the NKG2D receptor in NK-mediated rejection of Balb/c bone marrow from a B6xBalb/c F1 offspring animal (41), our results did not support an over-riding role for NKG2D in rejection of fully allogeneic Balb/c bone marrow from a B6 recipient. Our system did identify Ly49D, which has been previously established as a Balb/c-specific NK activating receptor, as important to NK-mediated killing of Balb/c bone marrow. This result confirms the authenticity of our assay system and supports the conclusion that in this system, NKG2D activity is not essential for NK-mediated rejection. Given others previous results showing a central role for NKG2D in rejection of Balb/c bone marrow in a hybrid resistance model (41), we believe that two explanations can explain the different conclusions drawn from the two studies: The first is that in the fully allogeneic system, unlike in the parent-to-offspring transplants explored previously, rejection may occur due to the action of a large array of activating receptors, which may include, but does not strongly rely on signaling through NKG2D. The increased genetic similarity of the parent-to-offspring transplants described previously may limit the number of activating ligands expressed on the transplanted bone marrow cells, revealing an important role for NKG2D that is not apparent in the fully allogeneic setting. This ‘rheostat’ effect of the activating and inhibitory receptor array on NK cells has been previously demonstrated in an in vitro system, in which differential expression of inhibitory Ly49 molecules significantly affected signaling through NKG2D (56). It is also possible that the pre-activation of NK cells with poly-IC that was performed in the previous study induced an NKG2D-mediated pathway that would not be functional without poly-IC. Thus, while NKG2D may function in the rejection response in poly-IC activated NK cells, it may not be as crucial in bone marrow rejection in the absence of poly-IC treatment. The identification of the array of NK receptors that are active during bone marrow rejection is of crucial importance to the potential targeting of these receptors in order to either induce NK activation (to facilitate graft-vs.-tumor effects) or NK inhibition (to inhibit NK-mediated rejection of limiting bone marrow doses). Our results provide further insight into the NK receptor and signaling arrays that are active during rejection of fully allogeneic bone marrow transplants.

The fact that only 10%–20% of the bone marrow transplant remained in the circulation (or in the bone marrow) within 2 days of transplant implies that NK cells effectively rejected 80%–90% of transplanted cells prior to their successful engraftment into a marrow niche. Thus, with effective inhibition of NK function, the number of cells required for engraftment may be significantly decreased. The ability to target NK receptors during BMT may have important implications during transplantation using limited stem cell numbers, such as are available in umbilical cord units, or would be available from a deceased bone marrow donor (in the setting of combination bone marrow and organ transplantation). In order to efficiently target NK-mediated bone marrow rejection, a comprehensive analysis of NK activation cascades thus remains a priority, as these molecules may become important drug development targets. The ultimate goal would be efficient, transient inhibition of NK alloreactivity, in order to increase the success of transplant and the availability of transplantable stem cell units.

Acknowledgments

LSK is supported by an NIH K08 Award # 1K08AI065822 – 01A1, by a Burroughs Wellcome Fund Career Award in the Biomedical Sciences, by the McKelvey Lung Transplant Center, The AFLAC Cancer Center and Blood Disorders Service and an NIH Pediatric Loan Repayment Award. MRR is supported by the American Diabetes Association and the McKelvey Lung Transplant Center. CPL is supported by the Carlos and Marguerite Mason Trust and by the McKelvey Lung Transplant Center. KH, AT, TCP, CPL, MRR and LSK are supported by the following NIH grants: #5P01-AI044644-07, 5U19-AI051731-02 and 5R01-AI40519-07.

Ancillary