Natural Killer Cells from Patients with Recombinase-Activating Gene and Non-Homologous End Joining Gene Defects Comprise a Higher Frequency of CD56bright NKG2A+++ Cells, and Yet Display Increased Degranulation and Higher Perforin Content

Mutations of the recombinase-activating genes 1 and 2 (RAG1 and RAG2) in humans are associated with a broad range of phenotypes. For patients with severe clinical presentation, hematopoietic stem cell transplantation (HSCT) represents the only curative treatment; however, high rates of graft failure and incomplete immune reconstitution have been observed, especially after unconditioned haploidentical transplantation. Studies in mice have shown that Rag−/− natural killer (NK) cells have a mature phenotype, reduced fitness, and increased cytotoxicity. We aimed to analyze NK cell phenotype and function in patients with mutations in RAG and in non-homologous end joining (NHEJ) genes. Here, we provide evidence that NK cells from these patients have an immature phenotype, with significant expansion of CD56bright CD16−/int CD57− cells, yet increased degranulation and high perforin content. Correlation was observed between in vitro recombinase activity of the mutant proteins, NK cell abnormalities, and in vivo clinical phenotype. Addition of serotherapy in the conditioning regimen, with the aim of depleting the autologous NK cell compartment, may be important to facilitate engraftment and immune reconstitution in patients with RAG and NHEJ defects treated by HSCT.

Mutations of the recombinase-activating genes 1 and 2 (RAG1 and RAG2) in humans are associated with a broad range of phenotypes. For patients with severe clinical presentation, hematopoietic stem cell transplantation (HSCT) represents the only curative treatment; however, high rates of graft failure and incomplete immune reconstitution have been observed, especially after unconditioned haploidentical transplantation. Studies in mice have shown that Rag −/− natural killer (NK) cells have a mature phenotype, reduced fitness, and increased cytotoxicity. We aimed to analyze NK cell phenotype and function in patients with mutations in RAG and in non-homologous end joining (NHEJ) genes. Here, we provide evidence that NK cells from these patients have an immature phenotype, with significant expansion of CD56 bright CD16 −/int CD57 − cells, yet increased degranulation and high perforin content. Correlation was observed between in vitro recombinase activity of the mutant proteins, NK cell abnormalities, and in vivo clinical phenotype. Addition of serotherapy in the conditioning regimen, with the aim of depleting the autologous NK cell compartment, may be important to facilitate engraftment and immune reconstitution in patients with RAG and NHEJ defects treated by HSCT.
cD56, interferon-γ, degranulation inTrODUcTiOn Natural killer (NK) cells are effector and immunomodulatory cells of the innate immune system. Through recognition of HLA class I molecules via activating and inhibitory killer cell immunoglobulin-like receptors (KIRs), and upon release of cytolytic granules, NK cells mediate killing of tumor and virus-infected cells and editing of dendritic cells, and play an important role in graft-versus-leukemia and graft rejection after hematopoietic stem cell transplantation (HSCT) (1)(2)(3)(4). Furthermore, they can modulate immune responses by secreting chemokines and cytokines (5,6).
Human peripheral blood contains two major NK cell subsets that can be distinguished based on the density of CD56 and CD16 expression on the cell surface: CD56 bright CD16 −/low and CD56 dim CD16 bright cells. These two NK cell subsets differ for the expression pattern of various other cell surface and intracellular molecules (7). In particular, CD56 bright cells express NKp46, CD94/NKG2A, and CCR7 at higher levels than CD56 dim NK cells, whereas CXCR1 and KIRs are more abundantly expressed by CD56 dim cells. Furthermore, CD56 bright and CD56 dim NK cells have distinct functional properties, with CD56 bright cells being potent producers of cytokines, and CD56 dim cells being active mediators of natural and antibody-dependent cellular cytotoxicity, as also reflected by higher intracellular levels of perforin and granzymes (8,9). In healthy adults, CD56 bright cells comprise a minority (5-10%) of all circulating NK cells, but because they express CCR7, they are attracted to secondary lymphoid organs where they represent the predominant NK subset (10,11). A subset of CD56 low KIR + NK cells, expressing CD57 represent terminally differentiated NK cells, whereas a further subset expressing the CD56 − CD16 + CD57 + KIR + phenotype are thought to represent exhausted NK cells (12).
RAG1 and RAG2 mutations in humans are associated with a broad spectrum of clinical and immunological phenotypes, including T − B − severe combined immune deficiency (SCID) (13), Omenn syndrome (OS) (14), atypical SCID (AS) (15)(16)(17), and combined immune deficiency with granuloma and/or autoimmunity (CID-G/A) (18)(19)(20)(21). We have previously shown that the severity of the clinical and immunological phenotype in patients with RAG mutations correlates with the residual recombination activity of the mutant recombinase-activating gene (RAG) protein (22), which may differently affect diversity and composition of T and B cell receptor repertoires (23), whereas NK cell differentiation proceeds unaffected. For patients with severe RAG mutations presenting with SCID, OS, or AS, HSCT represents the only option of definitive cure; however, an increased rate of allograft rejection has been observed as compared to patients with other forms of SCID (24,25). An important role of NK lymphocytes in mediating rejection of bone marrow allografts has been known for decades (26), but why patients with RAG deficiency would have a higher risk of graft rejection than other forms of NK + SCID (such as IL7R or CD3 deficiency) remains unknown.
Although RAG genes are not required for NK cell development, data in mice indicate that Rag deficiency affects NK cell phenotype and function. It has been shown that expression of the RAG genes begins in common lymphoid progenitor cells that give rise to T, B, and NK cells (27)(28)(29). Studies in mice harboring transgenic reporters for Rag expression or recombinase activity have demonstrated the existence of two populations of mature NK cells: those that have been exposed to transient Rag expression during lymphoid differentiation (here termed as Rag past ) and NK cells that were not previously exposed to Rag (Rag naive NK cells) (30). These two populations differ for their proliferative capacity and interleukin-2 (IL-2)-mediated Stat5 phosphorylation, and a progressive decrease in the proportion of Rag past cells has been observed during NK cell differentiation (29). Furthermore, Rag naive NK cells display an activated phenotype, increased cytotoxicity, and enhanced apoptosis, thereby resulting in poor survival and impaired DNA damage response as compared to their Rag past counterpart (30). It has been postulated that Rag expression in lymphoid progenitors would favor selection of cells with optimal levels of expression of proteins involved in DNA break repair, including ARTEMIS and DNA ligase 4 (LIG4), thereby marking functionally distinct subsets of NK cells, and providing Rag past NK cells with improved survival and "fitness. " If confirmed also in humans, the hypothesis that RAG deficiency results in the presence of NK cells with a distinctive phenotype and enhanced cytotoxic potential, would provide novel mechanistic insights to account for the high rate of primary graft failure, incomplete T cell reconstitution, and lack of B cell engraftment that are frequently observed following haploidentical HSCT for RAG deficiency, unless chemotherapy is used (25). To test this hypothesis, we have analyzed NK cell phenotype and function in a large cohort of patients with mutations in RAG and in other genes involved in non-homologous end joining (NHEJ), presenting with a variable severity of clinical and immunological manifestations, as compared to healthy donors of comparable age, and to patients with other forms of T or B cell lymphopenia.

Patients and controls
The patient population consisted of 66 subjects with molecularly confirmed biallelic mutations in gene involved in V(D)J recombination, in particular 35 patients with RAG1, 11 with RAG2, 15 with DCLRE1C (ARTEMIS), 3 with LIG4, and 2 with NHEJ1 (Cernunnos/XLF) mutations. Based on the clinical and immunological phenotype ( Table 1), these patients were assigned to the following subgroups: T − B − NK + SCID (n = 19), OS (n = 11), AS (n = 13), and delayed onset combined immune deficiency (CID, n = 23). For the purposes of analysis, patients with OS and with AS, were combined into a single group (OS/ AS, n = 24).
In addition, 22 patients with various other conditions characterized by numerical and/or functional defects of T lymphocytes served as a control for T cell deficiency. This group of other T cell defects (TCDs) included four patients with CD3G mutations, two patients with Di George syndrome, two patients with IL7R deficiency, and one patient each with mutations in MSN, JAK3, IL2RG, CD3D, CD3E, RMRP genes, or with MHC class II deficiency. In seven cases with T cell deficiency, the underlying diagnosis was unknown, including two infants with severe T cell lymphopenia diagnosed at birth with low T cell receptor excision circle (TRECs). Based on established criteria (31), 7 of these patients met definition of SCID (with <300 T cells/μL), and the remaining 15 had either less severe numerical TCDs or had functional T cell abnormalities. None of these patients with TCD carried mutations in RAG1/2, DCLRE1C, PRKDC, LIG4, or NHEJ1 genes.
As a control for B cell lymphopenia, nine patients with molecularly confirmed mutation in the BTK gene, causing X-linked agammaglobulinemia (XLA), were also studied.
Finally, 19 healthy infants (age ≤2 years) and 29 healthy subjects of >2 years of age served as normal controls for patients with SCID and OS/AS, or for patients with CID, respectively.
The study was approved by the Institutional Review Boards of Boston Children's Hospital, the National Institute of Allergy and Infectious Diseases (protocols 93-I-0119 and 16-I-N139), and of all other referring centers. Blood samples were obtained upon written informed consent of the subject or, in the case of minors, of their parents or legal guardians. Because of limitations in blood volume and/or in the number of cells, not all analyses could be performed in each individual included in the study.

analysis of nK cell Phenotype
Peripheral blood was collected in EDTA-or heparin-containing vacutainers. Red blood cells were lysed from 200 µl of EDTAblood using RBC lysis buffer (eBioscience). Cells were washed with FACS buffer (PBS containing 2% FBS), then equally distributed in five tubes and incubated for 1 h at 4°C with primary antibodies (Ab), as described below. After washing with FACS buffer, cells were incubated for 30 min at 4°C with secondary antibodies, washed again with FACS buffer, and incubated for 30 min at 4°C with conjugated antibodies. After washing, cells were re-suspended in FACS buffer and immediately analyzed on LSR Fortessa Flow Cytometer (BD) using FACS Diva v.6.1.3 software (BD). NK cells were defined as CD56 + CD3 − CD20 − CD14 − cells. Final analysis on the expression of NK cell markers was performed using FloJo v.10.2 (TreeStar). The antibodies used to define NK cell phenotype were either commercially available or were generated in the laboratories of Alessandro Moretta and Silvia Parolini.
To analyze the expression of NK cell surface markers, five different tubes were prepared ( Table 2), each of which contained A. Moretta

statistical analysis
Statistical analysis of the in vitro recombination activity of mutant RAG and ARTEMIS proteins, and of the expression of NK cell markers in patients and controls was performed using Mann-Whitney test for non-parametric variables.

Demographic and clinical Features
At the time of evaluation, mean age (±SEM) was significantly lower in patients with SCID (7. A clinical history of significant infections was present in 60 patients with RAG/NHEJ gene defects ( Table 1). Candida (n = 11), Pseudomonas (n = 7), human papillomavirus (HPV, n = 7), adenovirus (n = 6), varicella zoster virus (VZV, n = 6), Pneumocystis jiroveci (n = 5), Bacillus Calmette-Guerin (BCG, n = 5), and rhinovirus (n = 5) were the most common pathogens. In particular, P. jiroveci pneumonia was observed only in patients with SCID or OS/AS, whereas severe VZV infection was mostly restricted to patients with CID. Cytomegalovirus (CMV) and Epstein-Barr virus (EBV) infections were documented in 13 patients each, but were clinically significant only in 4 and 3 patients, respectively. Two SCID patients (P4 and P11) were diagnosed at birth following newborn screening, and infections were effectively prevented with isolation and prophylactic antimicrobials.
Autoimmunity and/or autoantibody production were documented in 23 patients, and were more common in patients presenting with CID (14 out of 21 patients) than in those with SCID (0/19) or with OS/AS (9/24). Cytopenias were the most frequent manifestation of autoimmunity and were documented in 11 patients (4 with OS/AS and 7 with CID). Alopecia and/or vitiligo were observed in six CID patients. Granulomatous disease was present in eight patients with CID. Finally, 12 patients (P20-P31) had typical features of OS (generalized erythroderma, lymphadenopathy, hepatosplenomegaly).

immunological Phenotype
The count of circulating CD3 + , CD19 + , and CD3 − CD56 + cells in patients belonging to the various groups (SCID, OS/AS, and CID) is shown in Figure 1. With the exception of one SCID patient (P19) who had a high count of maternal T cells (13997 cells/μl), all others had severe T cell lymphopenia (Figure 1A). By contrast, a higher count of autologous T cells was recorded in patients with CID and especially in patients with OS/AS. B cell lymphopenia was observed in the vast majority of patients, irrespective of the clinical phenotype ( Figure 1B). By contrast, NK cell count was either normal or increased in most patients ( Figure 1C).
Data on in vitro proliferation to phytohemagglutinin (PHA) were available in 51 patients. Response was absent (<10% of lower limit of normal, LLN) in 17, markedly reduced (10-30% of LLN) in 11, reduced (>30% of LLN, but lower than LLN) in 14, and normal in 9 patients. T cell proliferation to PHA was better preserved in patients with CID; among 18 such patients for whom information was available, 15 had either normal (n = 7) or reduced (n = 8) response.
Low serum levels of IgA and IgM were detected in the majority of SCID patients ( Table 1); serum IgG in this group was less informative because of the presence of maternally derived immunoglobulins. Immunoglobulin levels were more variable in patients with OS/AS and in those with CID. In particular, among 23 CID patients, 17 had either low serum IgG (n = 9) or were under immunoglobulin replacement therapy (n = 8) and 15 had undetectable serum IgA.

correlation between clinical Phenotype and recombination activity sustained by the Mutant alleles
Molecular analysis of the patients included in this study identified a total of 41 RAG1, 16 RAG2, 11 DCLRE1C, 4 LIG4, and 2 NHEJ distinct gene mutations ( Table 1). Using a flow cytometrybased assay, we have previously reported correlation between recombination activity supported by RAG and DCLRE1C mutant alleles and the clinical and immunological disease phenotype (22,32). For the patient cohort included in this study, data on in vitro recombination activity were available for 36 RAG1, 9 RAG2, and 6 DCLRE1C mutant alleles. Upon plotting for each patient the recombination activity associated with each of the two mutant alleles, a significant genotype-phenotype correlation was observed, with higher residual recombination activity for mutant alleles associated with a CID phenotype than for those associated with OS/AS or SCID (Figure 2).

circulating nK cells from healthy infants comprise a high Proportion of cD56 bright cells
According to a widely accepted scheme, human NK cell differentiation is marked by changes in the expression of cell surface and intracytoplamsic markers, so that five distinct stages of NK cell development are recognized (33). In peripheral blood from healthy controls, both stage 4 (CD56 bright CD16 − ) and stage 5 (CD56 + CD16 hi ) NK cells are detectable. In particular, in normal adults, CD56 bright CD16 − cells account for only 5-10% of all circulating NK cells. By contrast, a higher percentage of CD56 bright cells was detected in healthy infants (Figures 3A,B). Moreover, the proportion of CD16 + cells was lower in infants than in healthy controls greater than 2 years of age (Figures 3A,C). Finally, NK cells from healthy infants included a lower proportion of cells expressing CD57 (Figures 3A,D), KIRs (Figures 3A,E), and CXCR1 (Figures 3A,F), whereas the percentage of NK cells expressing NKG2A was higher in infants than in older healthy controls (Figures 3A,G).

Markedly increased Proportion of cD56 bright nK cells in the Peripheral Blood of Patients with rag and nheJ Defects correlates with the clinical Phenotype
A recent study in Rag −/− mice has unexpectedly disclosed abnormalities of NK phenotype and function, with increased proportion of mature, activated cells with increased cytotoxic activity and reduced survival (30). In order to assess whether similar abnormalities exist in patients with RAG or NHEJ defects, and to establish whether such abnormalities, if present, correlate with the severity of the clinical phenotype and with residual function of the mutant protein, we have analyzed the expression of CD56 and CD16 on the surface of CD3 − CD14 − CD20 − peripheral blood NK cells from patients and controls. As shown in Figures 4A,B, patients with SCID due to RAG/NHEJ defects had a high proportion of CD56 bright CD16 − NK cells (mean ± SEM: 34.7 ± 4.4), that exceeded what was observed in healthy infants (13.9 ± 3.3; p = 0.0007). Similarly, the proportion of CD56 bright CD16 − NK cells was higher in patients with CID than in healthy controls greater than 2 years of age (10.4 ± 1.8 vs. 5.1 ± 0.7; p = 0.0431; Figures 4A,B). Among infants with RAG/NHEJ defects, the proportion of CD56 bright CD16 − NK cells was higher in those presenting with SCID than in those with OS/AS (p = 0.0008; Figure 4B). In order to investigate whether the increase in the proportion of CD56 bright NK cells was secondary to numerical and/or functional abnormalities of T and/or B cells, we also analyzed a group of patients with other forms of TCD or with XLA. The proportion of CD56 bright CD16 − NK cells in these groups was similar to that of healthy controls, and significantly lower than in patients with SCID due to RAGH/NHEJ defects ( Figure 4B). Among patients with TCD, no correlation was observed among the proportion of CD56 bright NK cells and the severity of T cell lymphopenia, indicating that the increase of CD56 bright NK cells observed in SCID patients with RAG/NHEJ defects is not simply secondary to T cell lymphopenia. Furthermore, the proportion of CD56 bright CD16 int NK cells, that are considered to represent an intermediate step of differentiation between CD56 bright CD16 − and CD56 dim CD16 hi cells, was significantly higher in infants with SCID due to RAG/ NHEJ defects than in healthy infants, patients with TCD or with XLA (p < 0.0001; Figure 4C). A higher percentage of CD56 bright CD16 int NK cells was also observed in patients with CID than in healthy controls greater than 2 years of age (p < 0.05; Figure 4C). Finally, patients with SCID due to RAG/NHEJ defects had a lower proportion of CD56 dim CD16 hi NK cells when compared to healthy infants (p = 0.0007; Figure 4D). Similar results were obtained when comparing patients with CID versus healthy controls older than 2 years of age (p = 0.0011).

cD56 bright nK cells from infants with rag/ nheJ Defects have an immature Phenotype
The data reported above on CD56 and CD16 surface expression indicate that patients with RAG/NHEJ defects have an abnormal distribution of NK cell subsets, with an increased proportion of more immature cells as compared to what is observed in healthy controls. In order to further confirm this, we analyzed the expression of additional surface markers that are differentially expressed in CD56 bright versus more mature CD56 dim NK cells. As compared to healthy controls >2 years of age, patients with CID had a lower proportion of CD57 + NK cells (p = 0.0022; Figures 5A,B). Furthermore, patients with SCID had fewer CD57 + NK cells than patients with other forms of TCD (p = 0.0041), and a similar trend was also observed when compared to infant controls, although the difference did not reach statistical significance ( Figure 5B).
In normal subjects, circulating stage 4 CD56 bright NK cells are characterized by high levels of expression of NKG2A and of the chemokine receptor CCR7, which facilitates homing to lymphoid tissues. By contrast, expression of KIRs and of the chemokine receptor CXCR1 is virtually absent in CD56 bright NK cells. As compared to healthy controls, the proportion of NK cells expressing NKG2A at high density was increased in patients with SCID and CID due to RAG/NHEJ defects (Figures 5C,D), and by contrast, the percentage of CXCR1 + NK cells was reduced (Figures 5E,F). Furthermore, patients with SCID due to RAG/NHEJ defects had a lower proportion of NK cells expressing KIR than healthy infants, patients with OS and those with other forms of TCD (Figures 5G,H). While the proportion of CCR7 + cells was higher in healthy infants than in healthy controls >2 years of age (Figure 6), no significant differences were observed in the percentage of CCR7 + NK cells between patients with RAG/NHEJ defects and healthy controls.
Notably, CCR7 expression was downregulated during the transition from CD56 bright CD16 − to CD56 bright CD16 int cells (data not shown). Furthermore, the proportion of NK cells expressing Siglec-7, leukocyte Ig-like receptor (LIR)-1, or the activation marker CD69 was similar in patients and healthy controls, although interestingly a high proportion of Siglec-7 + NK cells were detected in the control group of patients with XLA ( Figure 6). Overall, these data support the notion that the CD56 bright NK cells that are more abundant in patients with RAG/NHEJ defects represent immature NK cells and provide evidence for a genotype-phenotype correlation in this group of diseases.

impact of Viral infections on nK cell Phenotype
Recombinase-activating gene and NHEJ defects lead to increased susceptibility to viral infections, including CMV, EBV, and VZV. It has been reported that viral infections (and CMV in particular) may lead to significant changes of NK cell phenotype and function, with accumulation of cytotoxic CD56 dim NKG2C + NKG2A − KIR + CD57 + cells (34). Expansion of these cells does  panels (a,c,e,g), representative examples are shown for one patient each belonging to the severe combined immune deficiency (SCID), Omenn syndrome (OS), and delayed-onset combined immune deficiency (CID) subgroups. For each marker, positivity was defined as fluorescence intensity above that of isotype control. Bars represent mean ± SD values. *p < 0.05; **p < 0.01; ***p < 0.005; ****p < 0.001.
not require T cells, as shown in an IL7R-deficient SCID infant after CMV infection (35). A significant proportion (26/66, 39.4%) of the patients with RAG/NHEJ defects included in this study had a documented history of CMV, EBV, VZV or severe JC virus (JCV) infection ( Table 1). The proportion of NK cells expressing NKG2C was not statistically different in patients with  SCID, OS/AS, or CID versus healthy controls ( Figure 7A). Furthermore, when patients were divided into two groups according to their history of CMV, EBV, VZV, or severe JCV infection, no differences were observed in the proportion of NKG2C + NK cells (Figure 7B). Even when patients with SCID/ OS/AS or with CID were divided into two subgroups based on their CMV infection history, no difference was observed in the proportion of NKG2C + (Figure 7C) and of CD56 dim (Figure 7D) NK cells. Finally, although a higher percentage of "exhausted" NK cells with a CD56 − CD16 + CD57 + phenotype was detected in patients with SCID, OS/AS, and TCD than in healthy infants (Figure 7E), no correlation was observed with a history of viral infections. cD56 bright cells from Patients with rag/ nheJ Defects are Potent Producers of iFn-γ Circulating CD56 bright NK cells serve an immunoregulatory function through secretion of various cytokines (IFN-γ, TNF-α, IL-10) (36). In order to assess this aspect of NK cell function, we performed in vitro stimulation with IL-12 and IL-18, and analyzed intracellular expression of IFN-γ upon gating separately on CD56 bright vs. CD56 dim NK cells. As expected, in both patients and controls CD56 bright cells were more potent producers of IFN-γ than CD56 dim cells, but both subsets were capable of producing this cytokine (Figure 8). However, CD56 dim cells from patients with SCID and with OS/AS due to RAG/NHEJ defects showed impaired IFN-γ production when compared to equivalent cells from healthy infants or patients with TCD. A similar trend was observed in patients with CID versus healthy controls greater than 2 years of age (Figure 8).
nK cells from Patients with sciD due to rag/nheJ Defects have increased Degranulation capacity in the absence of il-2-Mediated stimulation, and express a higher amount of intracellular Perforin than nK cells from healthy infant controls Previous data had shown that NK cells from rag −/− mice have increased cytotoxic activity (30). To investigate the cytolytic machinery of NK cells, we have analyzed NK cell degranulation in response to K562 target cells, in the absence or presence of IL-2. It is well known that human CD56 bright NK cells have reduced cytotoxic activity when compared to CD56 dim cells (8). In spite of the markedly increased proportion of CD56 bright NK cells, we observed that in the absence of stimulation with IL-2, NK cells from patients with SCID due to RAG/NHEJ defects have increased degranulation capacity when compared to NK cells from healthy infants (Figure 9A, left panel). In particular, CD56 dim NK cells from patients with SCID showed increased degranulation when compared to the equivalent subset of NK cells from healthy infants (Figure 9A, middle panel). A similar trend was observed also for CD56 bright NK cells, but the difference between SCID patients and controls did not reach statistical significance (Figure 9A, middle panel). Stimulation with IL-2 significantly increased NK cell degranulation capacity to similar levels in patients and controls (Figure 9A, right panel). Although the amount of intracellular perforin could be measured only in a small number of patients with SCID, a significantly higher mean fluorescent intensity (MFI) was observed in unstimulated NK cells from patients than from healthy infants ( Figure 9B).

DiscUssiOn
In this manuscript, we have demonstrated that NK cells from healthy infants comprise a higher proportion of CD56 bright CD16 −/int CD57 − cells than what is observed in older healthy subjects, confirming previous observations (37)(38)(39). Several lines of evidence indicate that CD56 bright cells represent precursors to CD56 dim cells. In particular, it has been demonstrated that appearance of CD56 bright cells precedes that of CD56 dim cells after HSCT (40) and that it is possible to induce differentiation of CD56 bright into CD56 dim NK cells in vitro in response to signaling via fibroblast growth factor receptor 1 (41). On the other hand, the model according to which CD56 bright cells precede CD56 dim cells during NK cell development has been challenged by NK cell lineage tracking experiments with genetic bar coding in macaques, which have suggested that these subsets may have distinct ontogeny (42).
In any case, the demonstration that composition of the peripheral NK cell compartment varies with age indicates that appropriate age-matched controls should be used when analyzing NK cell phenotype in pathological conditions. In particular, by performing an extensive phenotypic analysis, we have shown that NK cells from patients with RAG/NHEJ defects comprise a higher proportion of CD56 bright CD16 −/int NKG2A +++ cells, and a reduced percentage of CD56 dim CD16 hi cells expressing CD57, KIRs, and CXCR1, than observed in agematched healthy controls. Altogether, these observations suggest that NK cells from patients with RAG/NHEJ defects have a more immature phenotype when compared to age-matched healthy controls, in spite of the fact that NK cells from CID patients showed signs of progressive differentiation as compared to NK cells from patients with SCID/OS/AS. These data contrast with recent findings in Rag −/− mice, whose peripheral NK cell compartment is characterized by a predominance of cells expressing KLRG1, a marker of mature NK cells (30). Furthermore, peripheral NK cells from Rag −/− mice are largely CD62L − , indicating an in vivo activated phenotype (30). By contrast, we did not observe increased expression of CD69, an activation marker (43), in NK cells from patients with RAG/NHEJ defects, including those with OS, a condition characterized by accumulation of in vivo-activated autologous T cells. While the reasons for these discrepancies are not clear, murine and human NK cells differ substantially, also in regard to phenotypic markers that are differentially expressed during development.
By affecting the process of V(D)J recombination, mutations in RAG and NHEJ genes profoundly affect generation of T and B lymphocytes. Consistent with this, patients with RAG/NHEJ gene defects had a markedly reduced number of circulating B and T cells, with the exception of patients with OS, some of which had a normal or even increased number of oligoclonal and activated T cells, as typically observed in this disease. Previous studies in TCRβδ −/− and in μMT −/− mice had demonstrated that lack of T and B cells have opposite effects on the count of splenic NK cells. In particular, T cell deficiency is associated with a twofold to threefold increase, and B cell deficiency with a twofold to threefold decrease of NK cell count as compared to C57BL/6 wild-type mice (44). The patients with RAG and NHEJ defects included in this study had either normal or increased count of circulating NK cells. In order to rule out the possibility that their abnormalities of NK cell phenotype could be secondary to T and B cell lymphopenia, we have included in this study patients with XLA (who lack circulating B cells) and with other forms of TCD, including seven patients with SCID not due to RAG/ NHEJ defects. Even when compared to these two control groups, patients with SCID due to RAG/NHEJ gene defects had a markedly increased proportion of CD56 bright CD16 − CD57 − NKG2A +++ cells. Similar differences were also observed also when comparing NK cells from patients with OS/AS and those with TCD or XLA, with an increased percentage of CD56 bright CD16 − CD57 − cells in the former. Altogether, these data indicate that the increased representation of immature cells within the peripheral NK cell compartment of patients with RAG/NHEJ defects is not secondary to T and B cell lymphopenia.
Because of their severe T cell immunodeficiency, patients with RAG/NHEJ defects are highly prone to viral infections, including CMV. LIR-1 is a member of the immunoglobulin superfamily which has been shown to bind the human CMV MHC class I homolog UL-18 protein (45). NKG2A and NKG2C represent inhibitory and activating forms of CD94, recognizing non-classical HLA-E molecules (46). CMV infection can drive expansion of KIR + and/or LIR-1 + , NKG2A − NKG2C + NK cells (34). Furthermore, rapid differentiation of CD56 dim NKG2C + NKG2A − CD57 + cells has been observed after CMV reactivation in recipients of HSCT (47). It has been shown that upon transfer from CMV-seropositive donors into seropositive HSCT recipients, NKG2C + cells undergo expansion and produce high amounts of IFN-γ as compared to NKG2C + cells transfused into CMV-seronegative recipients, implying that NKG2C + cells may represent "memory" NK cells capable of a prompt response upon re-exposure to CMV (48). Finally, rapid expansion of NKG2C + NK cells has been reported in infants acquiring perinatal CMV infection (39). Although several of the patients included in this study had a history of CMV infection, no particular expansion of NKG2C + and/or LIR-1 + cells was observed. Monocytes are apparently required to allow expansion of adaptive/memory-like NK cells (49). Lack of expansion of NKG2C + cells in CMVinfected, RAG-/NHEJ mutated SCID patients may reflect a requirement also for T cell help in this process and may further contribute to poor control of CMV infection in these patients.
CD56 bright and CD56 dim NK cells differ in their immunomodulatory and effector function. In particular, CD56 bright cells are potent producers of IFN-γ, but have lower content of perforin and display reduced cytotoxic activity as compared to CD56 dim cells. Analysis of IFN-γ production by CD56 bright and CD56 dim NK cells from patients and controls confirmed that CD56 bright cells were more potent producers of IFN-γ, irrespective of the underlying diagnosis. However, the proportion of CD56 dim NK cells producing IFN-γ was lower in patients with SCID/OS/AS due to RAG/ NHEJ defects than in healthy infants or in patients with other forms of TCD, and a similar trend was observed for CD56 dim cells from CID patients. These results suggest that RAG/NHEJ defects are associated with abnormalities of NK cells that are not restricted to their phenotype, but may also involve NK cell function. In this regard, it is particularly significant that NK cells from the patients displayed enhanced degranulation in the absence of previous stimulation with IL-2, and that CD56 bright cells from patients with SCID/OS/AS had a higher content of perforin than CD56 bright cells from healthy infants. Although we did not perform a chromium release assay to investigate NK cytolytic function, these data suggest that RAG/NHEJ defects are associated with enhanced effector function, similarly to what was previously reported for Rag −/− mice (30).
Expression of RAG may initiate prior to T and B cell differentiation, as indicated by the presence of incomplete rearrangements at the immunoglobulin heavy chain or T cell receptor loci in a minority of murine and human NK cells (29,50). It has been suggested that introduction of DNA double strand breaks in lymphoid progenitor cells would facilitate selection of cells with more efficient DNA repair machinery (30). In the absence of this mechanism, NK cells from Rag-deficient animals and humans would have reduced cellular fitness. Our data indicate that indeed patients with mutations in RAG or in NHEJ genes share similar abnormalities of NK cell phenotype and function. While it is not known whether the immunoglobulin and T cell receptor loci represent the only sites in the genome that are targeted by RAG during early stages of lymphoid development, it is interesting to note that selective loss of CD56 dim cells, with relative preservation of the CD56 bright NK cell compartment, has been reported in patients with mutations of MCM4 and POLE2 genes, all of which control DNA replication (51)(52)(53).
In conclusion, we have identified for the first time abnormalities of NK cell phenotype and function in patients with mutations in RAG and in genes involved in NHEJ. These abnormalities are more pronounced in patients with severe mutations associated with a more severe clinical and immunological phenotype, and are not secondary to T and B cell lymphopenia. Demonstration of enhanced degranulation activity and higher amount of perforin in NK cells from patients with SCID due to RAG/NHEJ defects suggests that inclusion of serotherapy targeting NK cells in the HSCT conditioning regimen may help reduce the risk of graft rejection that has been reported in these diseases.

eThics sTaTeMenT
The study was approved by the Institutional Review Boards of Boston Children's Hospital, the National Institute of Allergy and Infectious Diseases (protocols 93-I-0119 and 16-I-N139), and of all other referring centers. Blood samples were obtained upon written informed consent of the subject or, in the case of minors, of their parents or legal guardians.