Donor natural killer cells trigger production of β-2-microglobulin to enhance post–bone marrow transplant immunity

Allogeneic hematopoietic cell transplantation is the most powerful therapy for high-risk leukemia.^1,2 Unresolved issues are leukemia relapse, graft-versus-host disease, and prolonged immune incompetence. In pioneering major histocompatibility complex (MHC) haplotype-mismatched (“haploidentical”) transplantation,³ we discovered that donor-versus-recipient natural killer (NK) cell alloreactions play a beneficial role.^4,5 Human NK cell function is finely tuned by clonally distributed cell surface receptors, including inhibitory receptors termed “killer-cell immunoglobulin-like receptors” (KIRs), that recognize human leukocyte antigen (HLA) class I allele groups (“KIR ligands”), such as Bw4, C1, and C2.^4-13 In transplants that are KIR ligand-mismatched in the donor-versus-recipient direction, certain NK cells in the donor repertoire express KIR(s) for HLA class-I allele group(s) present in the donor but absent in the recipient. Such cells sense the missing expression of the inhibitory self HLA class-I KIR ligand and are activated to kill recipient targets. NK cell alloreactions reduce leukemia relapse and improve survival.^4,5,13 In murine F1 H-2^d/b→parent H-2^b transplants or in H-2^d→H-2^b transplants, donor NK cells that do not express the H-2^b–specific Ly49C/I inhibitory receptor (but instead bear H-2^d–specific Ly49A/G2 receptors) cannot be blocked by the mismatched recipient MHC haplotype and are activated to kill the recipient’s targets.^4,5 In this model, the pretransplant infusion of donor-versus-recipient alloreactive NK cells ablates leukemic cells, recipient T cells responsible for graft rejection, and recipient dendritic cells (DCs), that trigger graft-versus-host disease.^4,5 

Here, in murine bone marrow transplant (BMT) models and in a human cell culture system, we show that donor-versus-recipient alloreactive NK cells triggered recipient DCs to synthesize a protein that played a key role in stimulating production of 2 master regulators of lymphocyte development, interleukin-7 (IL-7) and cKIT ligand (cKIT-L), that greatly accelerated post-BMT recovery of donor-derived B and T lineage cells and DCs. Proteomics analyses demonstrated the molecule produced by DCs to be β-2-microglobulin (B2M). Genetics analyses strengthened this finding and showed the in vivo role of B2M in immune reconstitution.

Experiments were performed in accordance with the Italian Ethics Approval Document for Animal Experimentation. Initial experiments investigated the effects of alloreactive NK cell infusions on post-BMT immune reconstitution. Six- to 8-week-old female C57BL/6 (H-2^b) mice (Charles River Laboratories, Calco, Italy) were conditioned with lethal total-body irradiation (TBI) (8 Gy). One day later, mice received an IV infusion (through the tail vein) of alloreactive NK cells from 6- to 8-week-old female Balb/c (H-2^d) mice (Charles River Laboratories). Control mice were given nonalloreactive (ie, syngeneic) NK cells. NK cells were obtained from splenocytes by Ficoll-Hypaque gradient centrifugation and immunomagnetic selection using a mouse NK cell isolation kit (Miltenyi, Bergisch Gladbach, Germany). Before infusion, NK cells were cultured for 4 days in the presence of 2000 IU/mL human IL-2 (Miltenyi) at the concentration of 2 × 10⁶ cells/mL in RPMI culture medium supplemented with 10% fetal calf serum (Invitrogen, CA). DX5 antibody staining of murine NK cell–specific α2 integrin showed ≥98% purity of ex vivo–expanded, IL2-activated NK cells used for infusions in bone marrow–transplanted mice (DX5⁻ cells were CD3⁺ T cells) (supplemental Figure 1). Before infusion, NK cells were washed. NK cell infusions contained 10⁶ alloreactive NK cells in a final volume of 500 μL phosphate-buffered saline. NK cells were analyzed for expression of inhibitory receptors for MHC class I by fluorescein isothiocyanate–conjugated anti-Ly49A, fluorescein isothiocyanate–conjugated anti-Ly49G2, and phycoerythrin-conjugated anti-Ly49C/I monoclonal antibodies (BD Biosciences, CA). Multicolor immunofluorescence was analyzed by flow cytometry using a 2-laser FACScanto (BD Biosciences). Such an analysis demonstrated that the NK cells from H-2^d donor mice contained a population (∼40%) that did not carry the H-2^b–specific Ly49C/I inhibitory receptor and, consequently, was potentially alloreactive against recipient H-2^b targets. Alloreactivity was tested using H-2^b mouse concanavalin A (Sigma-Aldrich, Missouri) T-cell blasts as targets in a ⁵¹Cr-release cytotoxicity assay.⁴ One day after the NK cell infusion, mice received 10 × 10⁶ T-cell–depleted bone marrow (BM) cells collected by flushing the femur and tibia shafts of H-2^d mice. BM cells were T-cell–depleted by negative immunomagnetic selection using anti-CD5 microbeads (Miltenyi) as previously described.⁴ 

In order to investigate whether a specific recipient cell type triggered alloreactive NK cells to accelerate post-BMT immune reconstitution, 4 types of recipient chimeras were constructed in which hematopoietic and nonhematopoietic tissues differed in their MHC class I types so as to make tissues potentially susceptible (H-2^b) or resistant (H-2^d or H-2^d/b) to alloreactivity mediated by NK cells from the H-2^d donor mouse. Chimera 1 displayed NK-susceptible nonhematopoietic tissues and NK-resistant hematopoietic cells. Chimera 2 displayed NK-resistant nonhematopoietic tissues and NK-susceptible hematopoietic cells. Chimera 3 displayed DCs that were potentially susceptible to donor alloreactive NK cells, whereas all other recipient hematopoietic and nonhematopoietic cells were resistant. Chimera 4 displayed NK-resistant DCs and NK-susceptible hematopoietic and nonhematopoietic cells. BM graft and NK cell numbers were the same as in the transplants described above. For a detailed description, see “Construction of transplantation chimeras” in supplemental Methodology in supplemental Material (available on the Blood Web site).

In a further series of experiments, mice received an infusion of supernatants obtained from NK/DC cocultures. Such experiments were designed to investigate whether soluble factor(s) contained in alloreactive NK/DC coculture supernatants mediated biological effects. Therefore, these experiments were intentionally performed using either allogeneic or syngeneic donor-recipient transplant pairs, with identical results. The details of such experiments are extensively outlined under “In-vivo infusions of NK/DC coculture supernatants” in supplemental Methodology in supplemental Material.

For statistical analyses, see supplemental Methodology.

Immune reconstitution was evaluated by multicolor immune-fluorescence, enzyme-linked immunosorbent assay, and quantitative polymerase chain reaction (qPCR). For a detailed description of methodologies, see “Immune reconstitution” in supplemental Methodology in supplemental Material.

For methodologies dealing with cloning of human alloreactive NK and T cells, cytotoxicity assays against allogeneic DCs,¹³ generation of NK/DC coculture supernatants, and the human thymic epithelial cells (TEC)/thymocyte culture system,¹⁴ see “Human cell cultures” in supplemental Methodology in supplemental Material.

The biochemical analyses were designed to identify a newly synthesized, biologically active protein (“the immune rebuilding factor”) and to define its biochemical features. In order to biochemically define the “immune rebuilding factor,” NK/DC coculture supernatants were subjected to hydrophobicity chromatography (HIC), reverse phase (RP) chromatography, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), and autoradiography. SDS-PAGE bands displaying immune rebuilding activity were excised and subjected to protein digestion and peptide extraction before nano–liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. For the detailed description of such experiments, see “Proteomics” in supplemental Methodology in supplemental Material.

In order to document the role of B2M in accelerating posttransplant immune reconstitution, supernatants obtained from cocultures of alloreactive NK-DCs or nonalloreactive NK/DC culture combinations were B2M immunodepleted using the polyclonal rabbit anti-B2M FL-119 Ab (Santa Cruz Biothecnology), followed by a secondary goat anti-rabbit Ab subsequently adsorbed on a Sepharose column. A non-B2M immune rabbit serum was used as a negative control. In order to obtain genetic evidence for B2M’s role in accelerating immune reconstitution, the following experiments were performed: (1) B2M-knockout (KO) H-2^b mice (Jackson Laboratories) were lethally irradiated and were given an infusion of 10⁶ alloreactive wild-type (WT) H-2^d NK cells before T-cell–depleted BMT (10 × 10⁶ cells) from WT H-2^d mice. (2) Lethally irradiated WT H-2^b mice were given an infusion of 500 μL supernatant from cocultures of 5 × 10⁶ alloreactive WT H-2^d NK cells and 5 × 10⁶ DCs from B2M-KO H-2^b mice. (3) B2M-KO mice received an infusion of supernatants obtained from cocultures of alloreactive WT H-2^d NK cells and WT H-2^b DCs. In order to obtain genetic evidence of the B2M’s role in the human system, the B2M gene was silenced in human DCs. Human DCs were transfected using Amaxa P3 Primary Cell 4D-Nucleofector X Unit (Lonza, Basel, Switzerland) in accordance with the manufacturer’s instructions. Aliquots of 2 × 10⁶ DCs were electroporated with 20 pmol of B2M-specific or scramble short interfering-RNA, resuspended in cell culture medium, and incubated overnight at 37°C/5% CO₂. B2M-silenced DCs or DCs electroporated with scramble RNA or untreated DCs were cocultured with alloreactive NK cell clones. NK/DC coculture supernatants were added to the human TEC/thymocyte culture.

In order to obtain information about the downstream pathways triggered by B2M on stromal cells, B2M KO mice were used in order to avoid any interference by endogenous B2M. They were lethally irradiated, and 7 days later, they were infused with supernatants obtained from cocultures of alloreactive WT H-2^d NK cells and WT H-2^b DCs or nonalloreactive (syngeneic) NK/DC coculture combinations. After 12 hours, RNA from murine BM was subjected to gene expression profiling. For the detailed description of such experiments see, “Gene expression profiling” in supplemental Methodology in supplemental Material.

H-2^b recipient mice were conditioned with TBI and received an infusion of IL-2–activated NK cells from H-2^d mice. Such NK cells contain a population that lacks the H-2^b–specific Ly49C/I inhibitory receptor and, consequently, is alloreactive against H-2^b lympho-hematopoietic cells.⁴ Subsequently, mice were transplanted with extensively T-cell–depleted BM from H-2^d mice (contaminating T cells in the BM graft were <0.2%) (Figure 1). The engrafted hematopoietic stem cells quickly gave rise to a greatly accelerated recovery of major players of the immune system, such as developing and mature T and B cells, as well as of DCs in the thymus, BM, and spleen, which quickly reached values comparable with those of donor mice (see Figure 1 and its legend for a detailed description of the various developing T, B, and DC subsets). We next investigated whether interactions between donor NK cells and specific recipient cell types were responsible for accelerated post-BMT immune reconstitution. Because alloreactive NK cells are preferentially activated by hematopoietic cells,⁴ we constructed recipient chimeras in which hematopoietic lineage cells and nonhematopoietic tissues differed in their MHC class I types so as to make tissues potentially susceptible (H-2^b) or resistant (H-2^d or H-2^d/b) to alloreactivity mediated by donor H-2^d NK cells (Figure 2A). Recipient chimeras were lethally irradiated, received an infusion of H-2^d NK cells, and subsequently received a T-cell–depleted BMT from the H-2^d donor mice. In chimeras with NK-susceptible nonhematopoietic tissues and NK-resistant hematopoietic cells, post-BMT immune reconstitution was not accelerated (Figure 2A: chimera 1). In contrast, chimeras with NK-resistant nonhematopoietic tissues and NK-susceptible hematopoietic cells displayed accelerated post-BMT immune reconstitution (Figure 2A: chimera 2). Because of the well-known key role of DCs in the activation of NK cell effector functions,^15,16 we constructed recipient chimeras in which DCs were the only recipient cell type that were either resistant or susceptible to donor NK cell alloreactivity (Figure 2A: chimera 3 and chimera 4). Accelerated post-BMT immune reconstitution did not occur in chimeras that had NK-resistant DCs (Figure 2A: chimera 3). In contrast, accelerated post-BMT immune reconstitution did occur in chimeras with NK-susceptible DCs (Figure 2A: chimera 4). Thus, recipient NK-susceptible DCs were necessary and sufficient for accelerated post-BMT immune recovery to occur. To investigate whether accelerated immune reconstitution was mediated by soluble factors released as a consequence of NK/DC interactions, we cocultured alloreactive NK cells from H-2^d mice with NK-susceptible DCs from H-2^b mice and infused the coculture supernatant into lethally irradiated recipient mice prior to the infusion of T-cell–depleted BMT (Figure 2B). As such, experiments were designed to investigate the potential effects of alloreactive NK/DC coculture supernatants; they were intentionally performed using either allogeneic or syngeneic donor-recipient transplant pairs, with identical results. Remarkably, unlike the nonalloreactive NK/DC coculture supernatant (denoted “NK^syn + DC” in Figure 2B), the alloreactive NK/DC coculture supernatant (denoted “NK^allo + DC” in Figure 2B) promoted accelerated reconstitution of donor thymocytes, B lineage cells, and DCs (Figure 2B). Thus, the interaction between alloreactive NK cells and NK-susceptible DCs resulted in the production of a soluble factor that accelerated post-BMT immune reconstitution. The factor appeared to be a newly synthesized protein as the supernatant’s ability to accelerate donor immune reconstitution was abolished by trypsin treatment and because preformed intracellular proteins obtained by lysing NK cells or DCs did not promote accelerated immune reconstitution (supplemental Figure 2A). To further demonstrate that the protein was newly synthesized and to identify the cellular source of the protein (NK cells vs DCs), DNA transcription was blocked with actinomycin D either in NK cells or in DCs before their coculture. Coculture supernatants were infused into conditioned mice before BMT. Supernatants in which DNA transcription was blocked in DCs, but not in NK cells, failed to accelerate post-BMT immune reconstitution (Figure 2B). These results demonstrated that the “immune rebuilding factor” was synthesized by DCs upon attack by alloreactive NK cells. Only the infusion of supernatants obtained by coculturing DCs with alloreactive NK cells, and not, for example, with alloreactive cytotoxic T cells, accelerated immune reconstitution (supplemental Figure 2B). Thus, it would appear that an NK cell–specific signaling was required in order to trigger DCs to activate DNA transcription and synthesis of the “immune rebuilding factor.” Moreover, because the capacity to accelerate reconstitution of the immune system was lacking in supernatant of cultures in which perforin-KO NK cells were used as effectors, it appears that the NK cell–mediated killing of allogeneic DCs was indispensable for the release of the “immune rebuilding factor” in the supernatant (supplemental Figure 2B).

To probe the existence of a human homolog of “the immune rebuilding factor,” NK cell clones were generated from individuals who possessed both HLA class I C1 and C2 group alleles (C group heterozygous individuals) and, therefore, possessed NK cells that were potentially alloreactive against targets from HLA class I C1 or C2 allele group homozygous individuals.^4-6,13 NK clones that were alloreactive against DCs from individuals homozygous for either the HLA C1 or C2 allele groups were cocultured with target DCs. When added to cultures of human TECs and human thymocytes,¹⁴ supernatants from such NK/DC cocultures increased the cellularity of thymocytes (Figure 2C). As in mice, DNA transcription blockade in DCs, but not in NK cells, prevented the increase in human thymocyte counts (Figure 2C). Also, trypsin treatment of human alloreactive NK/DC coculture supernatants abolished their ability to increase thymocyte cellularity (supplemental Figure 2C). Likewise, NK cell or DC lysates were not effective at increasing thymocyte cellularity (supplemental Figure 2C). Finally, only supernatants obtained by coculturing DCs with alloreactive NK cells, and not, for example, cytotoxic T cells, were able to increase thymocyte counts (supplemental Figure 2D).

To identify newly synthesized proteins with immune rebuilding activity, we performed 2 parallel NK/DC coculture experiments: 1 in the presence and the other in the absence of ³⁵S-methionine. Culture supernatants were fractionated by HIC and RP chromatography. One fraction (out of the 30 murine and 30 human fractions that were obtained in the nonradioactive experiment) displayed the most robust biological effect. Such fraction enhanced murine thymocyte (Figure 3A) and T-cell, B-cell, and DC counts in vivo (not shown) and increased human thymocyte cellularity in vitro in the human TEC/thymocyte cultures (Figure 3B). Then, HIC+RP chromatography fractions from the radioactive experiment were subjected to SDS-PAGE and autoradiography. An 11 to 12 kDa molecular weight (MW) protein was detected in the murine and human fractions that corresponded to the fractions that exerted maximal immune rebuilding activity in the nonradioactive experiment (Figure 3C). Two-dimensional electrophoresis showed the 11 to 12 kDa MW protein had pI of 7.5 and 8 in murine samples and 6 to 7 in human samples (Figure 3D). Subsequently, proteins were extracted from the area of interest of the nonradioactive SDS-PAGE gel, digested, and analyzed by high-resolution mass spectrometry. To identify the molecular nature of the “immune rebuilding factor” among the hundreds of proteins detected by mass spectrometry, the following ranking criteria were applied: (1) ≥1.5-fold increase in protein content in supernatants from alloreactive NK/DC cocultures in comparison with supernatants from nonalloreactive NK/DC cocultures, (2) 11 to 12 kDa MW, (3) pI of 6 to 7 for the human and 7.5 and 8 for the murine protein, and (4) detection of the protein displaying the above biochemical features in all murine (n = 3) and human (n = 2) experiments. Among the 30 top-ranking proteins (supplemental Table 1) out of the 853 that were detected by mass-spectrometry, we preliminarily tested FKBP1A because it is the target of tacrolimus and has a plethora of immune-modulatory functions,¹⁷ in spite of the fact that it did not fulfill all the ranking criteria because it had pI that differed from those detected in our experiments. We immune-depleted alloreactive NK/DC coculture supernatant with anti-FKBP1A antibodies. The immune rebuilding effect of the supernatant was not affected (not shown).

Only 1 protein, B2M, fulfilled all the ranking criteria^18,19 (Figure 3E; supplemental Table 1). We next sought functional evidence for B2M’s role in accelerating immune reconstitution. In the mouse, anti-B2M antibody depletion reduced the immune rebuilding ability of alloreactive NK/DC coculture supernatants, thus indicating B2M played a role in posttransplant immune reconstitution (Figure 4A). Most importantly, genetic evidence was obtained in B2M-KO H-2^b mice (Figure 4B). B2M-KO H-2^b mice were lethally irradiated and were given alloreactive NK cells and a T-cell–depleted BMT from WT H-2^d mice. In this model, accelerated immune reconstitution was not observed. Accelerated post-BMT immune rebuilding was also not observed when lethally irradiated WT H-2^b mice were given an infusion of supernatants from cocultures of alloreactive WT H-2^d NK cells and B2M-KO H-2^b DCs. In contrast, infusion of supernatants obtained from cocultures of alloreactive WT H-2^d NK cells and WT H-2^b DCs endowed B2M-KO mice with the capacity to undergo accelerated post-BMT immune reconstitution (Figure 4B).

Given that multiple cells of the hematopoietic lineage, such as B lineage cells, developing T cells, and DCs, were involved in the immune rebuilding effect, we hypothesized that the BM environment played a primary role in mediating the accelerated immune recovery. Therefore, to identify the molecular mechanisms underlying the regenerative effects induced by B2M, we performed RNA-seq analyses (shown as a volcano plot in Figure 4C) followed by gene set enrichment analysis on BM from mice that had been lethally irradiated but not rescued by BMT and had received an infusion of alloreactive NK/DC coculture supernatants vs infusion of nonalloreactive (syngeneic) NK/DC coculture supernatants. The list of differentially regulated genes is shown in the BM gene expression profiling table (BM gene expression profiling, supplemental Data 1). Among the 20 most upregulated genes, IL-7 and cKIT-L stood out because they are well-known master regulators of lymphocyte development.^20-22 Indeed, IL-7 and cKIT-L messenger RNA quantification by qPCR confirmed that these cytokines were stably upregulated (for 1 week) in the BM and thymus of lethally irradiated, alloreactive NK cell–treated mice (Figure 4D). IL-7 and cKIT-L were indispensable for faster post-BMT immune reconstitution as combined infusion of blocking anti–IL-7 and anti–cKIT-L antibodies hindered accelerated immune reconstitution of BMT-transplanted mice (Figure 4E). Interestingly, gene set enrichment analysis revealed that several biological processes, including cell cycle phase transition, response to cytokine, biosynthetic metabolic processes, and leukocyte immunity, were induced after infusion of alloreactive NK/DC coculture supernatants (supplemental Figure 3 and GEAS table results, supplemental Data 2).

Comparable results were obtained using human alloreactive NK cells cocultured with human DCs in which B2M gene expression was silenced by short interfering-RNA (Figure 5A). Supernatants from cocultures of alloreactive NK cells and B2M-silenced DCs failed to increase thymocyte cellularity in the human TEC/thymocyte culture system (Figure 5B). Further evidence for the involvement of B2M in the accelerated immune rebuilding was obtained by the use of recombinant B2M. Unfortunately, the use of murine recombinant B2M did not provide informative results (data not shown), probably because commercially available murine recombinant B2M molecules are histidine-tagged, a feature that is known to result in diminished or altered biological activity.^23,24 In contrast, a commercially available human recombinant B2M, which is not histidine-tagged and is expressed in Escherichia coli as the 14.0 kDa B2M precursor,²⁵ did exert a dose-dependent effect on the increase in thymocyte cellularity in the TEC/thymocyte culture system. However, the effect mediated by 14.0 kDa B2M precursor was smaller (one-third) than that obtained by using alloreactive NK/DC coculture supernatants (Figure 5C).

In the human TEC/thymocyte culture system, alloreactive NK/DC coculture supernatants promoted IL-7 production by TECs (Figure 5D), and the addition of an anti–IL-7 antibody prevented the increase in thymocyte cellularity, thus indicating that thymocyte expansion promoted by alloreactive NK/DC coculture supernatants was IL-7 dependent in this in vitro system (Figure 5E).

The present study uncovers a novel, NK cell–based immunotherapeutic intervention to obviate post–bone marrow transplant immune deficiency. In murine allogeneic bone marrow transplantation models, the infusion of ex vivo–expanded donor alloreactive NK cells triggered B2M production by recipient DCs that in turn elicited the release of IL-7 and cKIT-L from thymic and BM stroma. IL-7 and cKIT-L greatly accelerated reconstitution of donor-type, developing, and mature T and B cells, as well as of DCs in the thymus, BM, and spleen (see Figure 6 for a graphic summary of data). Comparable results were obtained in a human in vitro thymocyte/TEC culture system; that is, a human alloreactive NK/DC coculture supernatant promoted IL-7 production by TECs, which in turn supported thymocyte cellularity.

Evidence for the unprecedented involvement of B2M in immune reconstitution was obtained by the observation that (1) chromatography fractionation of alloreactive NK/DC coculture supernatants identified a protein with MW and pI of B2M. (2) A high-sensitivity mass spectrometry analysis identified a protein with the amino acid sequences specific of B2M. (3) Anti-B2M antibody depletion of NK/DC coculture supernatants abrogated the immune rebuilding effect. (4) The B2M gene is indispensable for accelerated immune reconstitution. B2M KO mice were unable to undergo accelerated immune reconstitution, and the infusion of WT NK/DC coculture supernatants reversed their inability to undergo accelerated immune reconstitution. Similarly, silencing the human B2M gene in human DCs before coculture with alloreactive NK cells prevented the increase in thymocyte cellularity in the human in vitro system. (5) The human recombinant 14.0 kDa B2M precursor²⁵ increased thymocyte cellularity in an in vitro TEC/thymocyte culture system. The partial immune rebuilding effect exerted by the human recombinant B2M, compared with the B2M synthesized by DCs, might be due to the fact that the recombinant B2M had a molecular weight of 14.0 kDa because it included the whole precursor sequence with the signal peptide for secretion. Moreover, in the E. coli expression system, no posttranslational modifications are usually inserted in the expressed proteins due to the diversity of the molecular machinery for posttranslational modifications, including proteolytic cleavage. According to our analysis, the protein candidate responsible of the immune reconstitution and identified as B2M had a molecular weight of ∼12.0 kDa, corresponding to the mature sequence of B2M as confirmed by SDS-PAGE and mass spectrometry.²⁵ It is, therefore, possible that the actual effector is the mature protein and not the precursor. Furthermore, 2-dimensional electrophoresis experiments showed that B2M is present as 2 spots with different pI, a possible indication of DC-specific posttranslational modifications that may happen before secretion and may be involved in the immune reconstitution effect.

B2M, recently shown to play a role in inflammation,^26,27 is primarily known as a key component of MHC class I family molecules; it is necessary for their correct folding and cell surface expression enabling the formation of a stable peptide binding groove and, consequently, for recognition by CD8⁺ T cells and CD8⁺ T-cell clonal expansion. Interestingly, B2M deficiency in humans supports the idea that B2M is involved in the maturation of various components of the immune system because B2M-deficient patients exhibit broad immune incompetence involving not only CD8⁺ T cells but also CD4⁺ and B cells.²⁸ Thus, it is conceivable that the NK/DC “crosstalk”¹⁶ and the consequent NK cell activation and killing of autologous DCs¹⁵ may lead to B2M production even under physiological conditions (an effect that our control experiments using autologous NK/DC combinations might not have been sensitive enough to detect).

To improve immune reconstitution after allogeneic hematopoietic cell transplantation, a variety of approaches have been proposed, such as adoptive T-cell therapy with nonalloreactive and/or pathogen-specific T cells, transfer of lymphoid progenitor cells, thymic grafts, enhancement of thymopoiesis by sex-steroid blockade, administration of IL-7, keratinocyte growth factor, fms-like tyrosine kinase 3 ligand, IL-22, or growth hormone.^29-38 Approaches thus far tested clinically have significant limitations. Here, for the first time, we demonstrate that infusion of large doses of alloreactive NK cells promotes striking acceleration of post–bone marrow transplant immune reconstitution. As it might have been expected, we did not find any difference in immune reconstitution of patients transplanted from NK alloreactive donors and patients transplanted from non-NK alloreactive donors. In fact, it should be noted that there are very remarkable differences between our clinical transplantation protocol^3,4 and the murine experiments. Human transplants were performed using purified CD34⁺ hematopoietic progenitor cells, which give rise to the reconstitution of NK cells that initially contain alloreactive cells in low frequencies. In contrast, the murine experiments were performed with the infusion of very large numbers of ex vivo–expanded, IL2-activated donor NK cells containing high frequencies (40%-50%) of alloreactive cells.

In conclusion, our studies uncover a novel therapeutic principle for the treatment of post–bone marrow transplant immune deficiency and suggest that future protocols that will include the adoptive transfer of large numbers of cytokine-activated, ex vivo–expanded donor alloreactive NK cells may improve the survival rate of patients undergoing haploidentical hematopoietic transplantation.

The authors thank Temistocle Ragni (Ospedale Santa Maria della Misericordia, Perugia) and Adriano Carotti (Ospedale Bambino Gesù, Roma) for thymic tissue samples obtained during corrective cardiovascular surgery. The authors also thank Massimo F. Martelli, Professor Emeritus, University of Perugia, for critical review of the manuscript and helpful advice.

This work was supported by grants from the Italian Association for Cancer Research (AIRC), the Swiss National Science Foundation, the Leukemia and Lymphoma Society, and the Italian Ministry for Further Education. Vrije Universiteit Medical Center-Cancer Center Amsterdam is acknowledged for support of the mass spectrometry infrastructure. E.V. was supported by grants from the Amy Strelzer Manasevit Research Program and AIRC.

Contribution: L.R. designed and performed experiments and wrote paper; E.U., S.C., and R. S. performed murine transplant experiments and human cell cultures; F.S. and P.L.O. performed biochemical analyses; E.B. performed murine immune reconstitution analyses; D.C. designed and performed proteomics analyses and wrote paper; S.R.P. performed proteomics analyses; S.B. performed human RNA silencing experiments; D.R. contributed to human thymocyte culture experiments; S.P. as heart surgeon at the Ospedale Santa Maria della Misericordia, Perugia, provided human thymic tissue samples obtained during corrective cardiovascular surgery; L.B. conceived, designed, and supervised biochemical analyses and critically revised the manuscript; C.R.J. supervised proteomics analyses and critically revised the manuscript; G.A.H. designed experiments, provided constructive suggestions, and wrote paper; E.V. and A.C performed GSEA and revised the manuscript; A.P. and F.L. revised the manuscript; and A.V. conceived and supervised the project, designed experiments, and wrote the paper.

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

Correspondence: Loredana Ruggeri, Ospedale Santa Maria della Misericordia di Perugia, Piazzale Menghini 1, 06132 Perugia, Italy; e-mail: loredana.ruggeri@ospedale.perugia.it; and Andrea Velardi, Division of Hematology and Clinical Immunology and Bone Marrow Transplantation Program, Department of Medicine and Surgery, University of Perugia, Piazza Gambuli 1, 06132 Perugia, Italy; e-mail: andrea.velardi@unipg.it.