The Impact of KIR2DS4 Alleles and the Expression of KIR in the Development of Acute GVHD After Unrelated Allogeneic Hematopoietic SCT

ORIGINAL ARTICLE The impact of KIR2DS4 alleles and the expression of KIR in the development of acute GVHD after unrelated allogeneic hematopoietic SCT

XJ Bao1, LH Hou2, AN Sun1, QC Qiu1, XN Yuan1, MH Chen2, ZX Chen1 and J He1

1Jiangsu Institute of Hematology, Key Laboratory of Thrombosis and Hemostasis of Ministry of Health, The First Afﬁliated Hospital of Soochow University, Jiangsu, China and 2Department of Pediatric, CW Bill Young Marrow Donor Recruitment and Research Program, Georgetown University Medical Center, Washington, DC, USA

The role of killer Ig-like receptors (KIR) in SCT was transformed cells and may have a role in allogeneic analyzed. A total of 75 Chinese patients were transplanted hematopoietic SCT (allo-HSCT).1,2 Although the most with T-depleted hematopoietic stem cells from unrelated important criteria for donor selection is the match for HLA donors. Among the 75 donor–recipient pairs, 60 were class I and class II, other relevant factors such as the HLA 10/10 matched and 15 had some mismatches at receptors of NK cells are currently a major focus of HLA-C. Transplants from KIR haplotype B/x group ongoing research. The ability of NK cells to recognize and donors showed significantly higher overall survival rates kill target cells is commonly determined by the action of compared with those from KIR haplotype A/A donors different surface receptors that either trigger or inhibit NK (relative risk (RR) 3.1 (95% confidence interval (CI) 1.1– activities. The inhibitory killer Ig-like receptors (iKIRs; 8.6), P ¼ 0.007). In the haplotype A/A group, a higher that is, KIR2DL and KIR3DL) recognize MHC class I risk of acute GVHD (aGVHD) (RR 9.0 (95% CI 1.2– molecules and restrain NK cell activation. HLA group C2 66.9), P ¼ 0.01), especially grade III–IV aGVHD alleles are the ligands for KIR2DL1 while HLA group C1 (P ¼ 0.006), was observed when the donor was homo- alleles are ligands for KIR2DL2/L3 and HLA-Bw4 for zygous for the full-length expressed KIR2DS4*00101 KIR3DL1. There are six members (KIR2DS1, KIR2DS2, allele. Real-time PCR showed that a high expression of KIR2DS3, KIR2DS4, KIR2DS5 and KIR3DS1) in the inhibitory KIR (2DL1 and 3DL1) in the early stages (o90 family of activating KIRs (aKIRs). KIR genes vary greatly days) after transplantation correlated with the development between different individuals. Not all KIR haplotypes of aGVHD (z ¼ 2.558, P ¼ 0.011). Our findings indicated a encode the six aKIR genes. Diverse KIR gene combinations significant association of full-length KIR2DS4 or KIR2DL1/ can be simplified into two groups of genes often called 3DL1 expression with the occurrence of aGVHD. In haplotypes: A and B. The group A haplotype encodes aggregate these results suggested that combining KIR and mainly for inhibitory receptors and appears to carry HLA genotyping could help in the selection of transplant KIR2DS4 as its only stimulatory gene while group B donors and improve the outcome of transplantation. Dynamic haplotypes have variable gene content including additional detection of KIR2DL1/3DL1 expression would be beneficial activating receptor genes. KIR2DS4 has eight alleles for prediction of aGVHD after transplantation. currently described: only 2DS4*001 encodes for a cell Bone Marrow Transplantation (2010) 45, 1435–1441; surface receptor while the remaining alleles, 2DS4*003, doi:10.1038/bmt.2009.357; published online 11 January 2010 *004, *006, *007, *008, carry a deletion in the exon Keywords: NK cell; killer Ig-like receptor; KIR2DS4 encoding the membrane-proximal extracellular domain allele; acute GVHD; allogeneic hematopoietic SCT altering their reading frame.3 In humans, Shilling et al.4,5 showed that the NK cell receptor repertoire, which emerges after allo-HSCT is determined by the KIR genes in the donor cells. In these newly developing NK cells, which may be activated, the inhibitory signal from iKIRs could be lacking because of Introduction the absence of ligands on the recipient’s cells. Even in fully matched HLA transplants, because of the ‘missing ligands’ Natural killer (NK) cells are lymphocytes that act early in NK cells may become uninhibited in the early post the immune response against virus-infected or tumor- transplant period and lyse target cells when MHC class I are absent from recipients’ cells.6,7 Recent advances in understanding of NK signaling have provided a hypothesis Correspondence: Dr J He, Jiangsu Institute of Hematology, Ministry of suggesting that allogeneic transplants selected for lack of Health; The First Affiliated Hospital of Soochow University, 188 Shizi recipient HLA ligands for donor iKIRs allow the develop- Street, Suzhou, Jinagsu 215006, China. E-mail: [email protected] ment of allo-reactive donor-derived NK cells that could kill Received 1 May 2009; revised 26 October 2009; accepted 11 November host tumor (GVL), because of the triggering effect of 2009; published online 11 January 2010 aKIR. When not impeded by the iKIRs, binding of the 2DS4 alleles and KIR expression in acute GVHD XJ Bao et al 1436 aKIRs to their ligands results in NK cell stimulation. Chen recurrence of disease after transplantation, as based on et al.8 reported that presence of more aKIR genes in donors cytomorphological or histological criteria. TRM was were associated with a lower transplant-related mortality defined as death without relapse. Overall survival (OS) (TRM) and a better survival. However, this NK reactivity rate was defined as the percentage of patients who were may also have a negative effect on outcome: Gagne et al.9 alive for a period of time between transplantation and reported an increase of GVHD in unrelated transplanta- death because of any cause. tions when the donor has more aKIR. These data highlight the effect of donor-aKIR genes in allo-HSCT. KIR PCR amplification Several recent trials using KIR-ligand mismatch to Genomic DNA was extracted from PBMC using a Gentra exploit NK cell allo-reactivity in the allo-HSCT have Puregene Blood Kit (QIAGEN, Valencia, CA, USA), yielded discrepant results.10,11 Some discrepancies may be following manufacturer’s instructions. KIR (17 loci) explained by the inconsistent use of T-cell depletion. As genotyping was performed by PCR with sequence-specific donor, T cells may override NK cell effects. In this study, primers using a KIR Genotyping SSP Kit (Invitrogen, we assessed the relative effects of donor KIR and host Carlsbad, CA, USA) to test for gene presence and absence. environment on clinical outcomes for T-depleted unrelated PCR amplification reactions were performed using a Perkin allo-HSCT. Elmer GeneAmp 9700 thermal cycler (Perkin Elmer, Waltham, MA, USA). To identify KIR2DS4 allelic sequences, the KIR2DS4 Materials and methods gene was amplified from genomic DNA by PCR with two pairs of primers designed to yield overlapping amplicons Patients, donors and transplant characteristics covering most of the exons as described earlier.12 Nested Samples from 75 donor–recipient pairs from unrelated PCR was used to isolate KIR2DS4 alleles with and without donors HSCT performed to treat leukemia were provided a known exon 5 deletion in heterozygotes as described by the First Affiliated Hospital of Soochow University. The earlier. For individuals who carried both 2DS4-full-length individuals (150) were Chinese Han. The study protocol allele and 2DS4-deleted variant, PCR products were cloned was approved by the local ethics committee, and all using the TOPO TA cloning kit (Invitrogen). patients gave written informed consent for their participa- tion in the study, which followed the ethical guidelines of Killer Ig-like receptor sequencing the Declaration of Helsinki. The patients in this group had Sequencing primers were positioned to obtain the sequence a median age of 25 and included 19 women and 56 men. of both strands of each amplicon.12 Sequencing was All the patients were undergoing transplantation for performed using Applied Biosystems’ BigDye Terminator n n n CML ( ¼ 24), AML ( ¼ 19), ALL ( ¼ 29) and other Ready Reaction mix according to the manufacturer’s malignancies (n ¼ 3). Selection of an unrelated donor for protocol (Applied Biosystems, Foster City, CA, USA). HSCT was based primarily on HLA matching. All of the 75 The reactions were purified using CleanSEQ (Agencourt donor–recipient pairs were HLA-A, B, C, DRB1 and Bioscience, Beverly, MA, USA) according to the manu- DQB1 low-resolution typing identical. Of the 75 cases, 15 facturer’s protocol. Sequencing products were detected had an allelic mismatch at the HLA-Cw locus by high- using an Applied Biosystems 3730XL DNA analyzer. resolution typing. Transplant characteristics for the 75 Sample files were analyzed using Sequencher (Genecodes patients is shown in Table 1a. Acute and chronic GVHD Corp., Ann Arbor, MI, USA) and Assign SBT 3.2.7 were diagnosed according to standard criteria. Patients (Conexio Genomics, Applecross, Western Australia) soft- with grade 0 acute GVHD (aGVHD) were defined as no ware. The sequences were compared with locus-specific aGVHD while those with grade I–IV aGVHD were all databases (ImmunoPolymorphismDatabase-KIR Release classified into the aGVHD group. Relapse was defined as 2.0.0) of known KIR sequences13 created using Library Builder software (Conexio Genomics) to identify alleles. In this report, the numbering of nucleotides and codons is based Table 1a Patient characteristics for the 75 transplants, from on ImmunoPolymorphismDatabase-KIR unless noted. haplotype A/A donors (n ¼ 35) and haplotype B/x donors (n ¼ 40)

Haplotype Haplotype Haplotype assignment A/A donor B/x donor Detection of at least one of the KIR B haplotype-deﬁning Recipient and transplant characteristics loci (KIR2DL5, 2DS1, 2DS2, 2DS3, 2DS5 or 3DS1) in a Number of patients 35 40 sample dictated that the genotype contains at least one B Female 8 (22.9%) 11 (27.5%) haplotype. Such samples were assigned the genotype B/x. Age, median (range) 27 (4–47) 25 (5–43) ALL 15 (42.9%) 14 (35.0%) We assumed A/A haplotypes when the only activating gene AML 4 (11.4%) 15 (37.5%) present in an individual was KIR2DS4 and the sample CML 13 (37.1%) 11 (27.5%) lacked all KIR B loci. Others 3 (8.6%) 0 (0.0%) Follow-up, month, median (range) 22 (5–31) 23.5 (7–34) 10/10 HLA match with donor 27 (77.1%) 33 (82.5%) Transplant grouping 1 HLA-C loci mismatch with donor 8 (22.9%) 7 (17.5%) Transplants were divided into two groups depending on the Possession of allo-reactive iKIRs 31 (88.6%) 35 (87.5%) donor’s KIR haplotypes:

Bone Marrow Transplantation 2DS4 alleles and KIR expression in acute GVHD XJ Bao et al 1437 1. Group A/A: Transplants from genotype A/A donors the copy number of KIR gene in every 10 000 ABL gene 2DS4-full subgroup: Transplants from donors who copies from all of the nucleated cells in blood. carried two full-length 2DS4 alleles (KIR2DS4*001) 2DS4-deletion subgroup: Transplants from donors who Statistical analysis carried one or two 2DS4 alleles carrying the deletion Comparison between two groups was performed using (that is, KIR2DS4*003, 004, 006, 007, 008). Fisher’s exact test. We considered six outcomes in the 2. Group B/x: Transplants from genotype B/x donors. analysis: OS, event free survival, relapse, TRM, aGVHD and chronic GVHD. Allo-reactive inhibitory killer Ig-like receptors definition Kaplan–Meier and Log-rank tests were used to estimate Under normal circumstances, every NK cell with killing the survival curve and compare the difference of death risk capacity must have an inhibitory receptor that recognizes at between transplants from KIR haplotype A/A and haplo- least one of the MHC class I gene products on the cell type B/x donors. Compared with haplotype B/x group, surface, and this does indeed appear to be the case.14 NK relative risk (RR) of the other five events after HSCT from cells from stem cell donors may have the capacity during haplotype A/A donors and 95% confidence interval (CI) the early recovery phase to react to missing ligands even in were estimated by cohort study. HLA completely matched unrelated recipients6,7 and have a Compared with 2DS4-deletion subgroup, RR of all the capacity of killing target cells in HLA identical allo-HSCT. six outcomes after HSCT from 2DS4-full subgroup donors The combination of incoming aKIR signals with the lack of and 95% CI were estimated by cohort study. inhibition of NK cells tips the balance of NK regulation to Continuous variables were compared between groups by favor the reactive state.15 Therefore, we examined the allo- Mann-Whitney U-test. reactive iKIRs in these 75 transplants. HLA–KIR mis- The P-values o0.05 were considered statistically sig- matching status was determined using high-resolution nificant. Two-sided tests were used throughout all the HLA-B and HLA-Cw typing. Allo-reactive KIRs were analyses. The SAS software package (version 9.1.3; SAS defined as donor 2DL1 lacking HLA group C2 alleles on Institute, Cary, NC, USA) was used for data management. the recipient’s cells as well as 2DL2/L3 lacking HLA group C1 alleles and 3DL1 lacking HLA-Bw4 alleles. Results Real-time (RT) PCR analysis Thirty of the 75 pairs were analyzed by RT-PCR. All the Predicted killer Ig-like receptor alloreactivity selected transplants carried allo-reactive iKIR in the donor- in the recipient vs-recipient direction. The donor samples were drawn HLA typing of 75 donor–recipient pairs showed that 66 before mobilization while the recipient samples at days pairs (88. 0%) lacked at least one recipient HLA ligand for donor iKIRs, defined as having potentially allo-reactive þ 30, þ 60, þ 90, þ 120, þ 180, þ 360 post transplantation. The transplantations were divided into two groups: KIRs. The percentage of transplants with ‘missing ligands’ patients with grade 0 aGVHD were defined as no aGVHD in A/A group and B/x group was 88.6 and 87.5%, group while those with grade I–IV aGVHD were all respectively. In A/A group, this percentage for KIR2DS4- classified into the aGVHD group. full subgroup and KIR2DS4-deletion subgroup was 85.7 RNA samples were isolated using TRIzol reagent and 90.5%, respectively. Multivariate analysis was performed to examine the percentage of transplants with (Invitrogen) and were stored at À80 1C. RT-PCR was used for quantification of KIR2DL1, 2DS1, 3DL1, 3DS1 and ‘missing ligands’ on clinical outcomes and there was no the control ABL gene, using the primers and probes as statistical difference. published.16 We used TaqMan MGB Probe (Applied Of the 66 pairs with potential allo-reactive KIR, 53 pairs Biosystems) to label RT-PCR products. The human house had no C2 group allele in the recipient for donor keeping gene, ABL, was used as a control to assess the KIR2DL1; 2 pairs, no C1 group allele for donor KIR2DL2 quality and quantity of the RNA. To construct a standard and/or 2DL3; and 24 pairs, no HLA-Bw4 in the recipient for donor KIR3DL1. As mRNA expression of 3DL2 is curve for the quantification of each KIR transcript, for 16 each assay, 10-fold dilutions were performed. Results undetectable in most individuals, KIR2DL1 and 3DL1 are the main potential allo-reactive iKIRs in Chinese Han showed that standard curves of KIR2DL1, 2DS1, 3DL1 17 and 3DS1 paralleled that of the ABL gene, suggesting the populations. feasibility of using ABL gene-based standard curves. RNA was first reverse transcribed into cDNA using 2 mg/ml total Killer Ig-like receptor genotyping RNA from nucleated cells. The resulting cDNA was used A majority (139) of the 150 samples from the Chinese Han as templates for subsequent PCR amplification. Reactions population (92.7%) were positive for KIR2DS4. Sequen- were performed using the TaqMan Universal PCR Master cing of most of the coding region of this gene identified four Mix and DNA Engine Opticon 2 (MJ Research, Waltham, of the eight known KIR2DS4 alleles, KIR2DS4*00101, MA, USA). *003, *004 and *007. The KIR2DS4-deleted variants were The quantity of target gene (KIR) in an unknown sample found in 44% of the 150 individuals. Of the 139 2DS4 was expressed as a KIR/ABL ratio. The final copy number positive individuals, 66 were homozygous for the A of the target gene (KIR) was calculated by the following haplotype and 73 were classified into the B/x group. In formula: (KIR/total ABL ratio Â 10 000). It was defined as the A/A group, 12.1% (8 of 66) individuals had the

Bone Marrow Transplantation 2DS4 alleles and KIR expression in acute GVHD XJ Bao et al 1438 KIR2DS4-deleted versions only, compared with 30.1% (22 subgroup (0/21) (P ¼ 0.006). On statistical analysis, 2DS4- of 73) in the B/x group. It has been reported that the 2DS4- full-length homozygous donors had a significantly in- deletion alleles have a 22-bp deletion in exon 5, which creased rate of aGVHD in unrelated donors transplants causes a frame shift, yielding a KIR2DS4 protein with loss compared with the donors who carried one or two 2DS4- of the transmembrane and cytoplasmic domains. The deleted varients. No statistical difference was observed in deletion alleles do not encode for a functional receptor OS, event free survival, relapse, TRM and chronic GVHD. but a soluble form of the protein that is potentially The transplant characteristics and clinical outcomes are secreted.3 We noticed the fact that the individual who is shown in Tables 2a and 2b. homozygous for the A haplotype and with two deleted version of KIR2DS4 may not have any KIR functional activating gene. In this study, a significantly decreased frequency of carrying two 2DS4-deleted variants was Table 1b The clinical outcomes of transplants from haplotype A/A observed in g A/A individuals (8 of 66), who carried donors (n ¼ 35) in comparison with haplotype B/x donors (n ¼ 40) KIR2DS4 as the only aKIR gene, in comparison with the Outcomes (A/A compared with B/x) RR (95% CI) P-value B/x group (22 of 73) (P ¼ 0.013), which was similar to the results reported by of Middleton et al.3 aGVHD 1.4 (0.5–4.3) 0.579 CGVHD 0.9 (0.2–3.7) 1.000 Relapse 0.3 (0.1–1.5) 0.162 The analysis of killer Ig-like receptor haplotype TRM 2.0 (0.6–6.2) 0.265 in donor and clinical outcomes EFS 1.3 (0.5–3.2) 0.646 In this study, transplants from KIR haplotype B/x donors (n ¼ 40) showed a higher OS rate compared with those from Abbreviations: aGVHD ¼ acute GVHD; cGVHD ¼ chronic GVHD. haplotype A/A donors (n ¼ 35) (RR 3.1 (95% CI 1.1–8.6), P ¼ 0.007), a result also observed in another study18 (see Figure 1). No statistical difference was observed in event Table 2a Patient characteristics for transplants from haplotype A/A homozygous donors (n ¼ 35), characterized by KIR 2DS4-full or- free survival, relapse, TRM, aGVHD and chronic GVHD deletion allele between the two groups (see Table 1b). KIR2DS4-full KIR2DS4-deletion donora donora The analysis of KIR2DS4 subtypes in haplotype A/A donors and clinical outcomes Recipient and transplant characteristics Of the 75 recipient/donor pairs, 35 cases who received Number of patients 14 21 Female 2 (14.3%) 4 (19.0%) transplants from KIR haplotype A/A donors were then Age, median (range) 29.5 (9–45) 25 (12–47) further analyzed. In transplants, when donors carried two ALL 5 (35.7%) 14 (66.7%) KIR2DS4-full-length alleles (n ¼ 14), the aGVHD rate was AML 1 (7.1%) 1 (4.8%) much higher, compared with those in which the donors CML 7 (50.0%) 6 (28.6%) Others 1 (7.1%) 0 (0.0%) carried one or two 2DS4-deleted variants (n ¼ 21) (RR 9.0 Follow-up, month, median 21.5 (8–34) 21 (5–32) (95% CI 1.2–66.9), P ¼ 0.01). The risk of grade III–IV (range) aGVHD was also increased significantly in the KIR2DS4- 10/10 HLA match with donor 10 (71.4%) 15 (71.4%) full subgroup (5/14), compared with the 2DS4-deletion 1 HLA-C loci mismatch with 4 (28.6%) 6 (28.6%) donor Possession of allo-reactive 12 (85.7%) 19 (90.5%) iKIRs Survival proportions aThe donor 2DS4-full group was defined as transplants from KIR2DS4- 150 Donor KIR haplotype full-homozygous donors; the donor 2DS4-deletion group was defined as A/A transplants from KIR2DS4-deletion-homozygous or KIR2DS4-full/ P=0.007 B/x -deletion heterozygous donors. 100 Table 2b The clinical outcomes of the 2DS4-full subgroup (n ¼ 14) in comparison with 2DS4-deletion subgroup (n ¼ 21), after URD 50 HSCT from haplotype A/A homozygous donors (n ¼ 35)

Percent survival Percent Outcomes (2DS4-full 2DS4-full 2DS4-deletion RR P-value compared with subgroup subgroup (95% CI) 0 -deletion) (n) (n) 0 500 1000 1500 aGVHD 6/14 1/21 9.0 (1.2–66.9) 0.01 Days after transplant Grade III–IV aGVHD 5/14 0/21 — 0.006 Figure 1 Overall survival (OS) rates of 75 transplantations, from killer CGVHD 1/14 3/21 0.5 (0.1–4.3) 0.635 Ig-like receptor (KIR) haplotype A/A or B/x donors. The number of each Relapse 3/14 1/21 4.5 (0.5–39.0) 0.279 group: 35 cases transplanted from KIR haplotype A/A donors and 40 cases TRM 7/14 6/21 1.8 (0.7–4.1) 0.288 from KIR haplotype B/x donors. The OS of the patients correlated with the EFS 4/14 12/21 0.5 (0.2–1.2) 0.166 KIR haplotypes of the donors. Transplantations from KIR haplotype A/A OS 7/14 15/21 0.7 (0.4–1.3) 0.288 donors had a lower OS rate (P ¼ 0.007) compared with those from KIR haplotype B/x donors. Abbreviations: aGVHD ¼ acute GVHD; cGVHD ¼ chronic GVHD.

Bone Marrow Transplantation 2DS4 alleles and KIR expression in acute GVHD XJ Bao et al 1439 Table 3 Patient characteristics of the aGVHD group (n ¼ 12) and P=0.021 no aGVHD group (n 18) in real-time (RT)-PCR analysis ¼ 3000 aGVHD No aGVHD group group

Recipient and transplant characteristics 2000 Number of patients 12 18 Female 4 (33.3%) 4 (22.2%) Age, median (range) 26 (11–49) 26.5 (16–38) Chronic myeloid leukemia 6 (50.0%) 5 (27.8%) Acute leukemia 6 (50.0%) 11 (61.1%) 1000 Others 0 (0.0%) 2 (11.1%) Follow-up, month, median (range) 17.5 (17–46) 16.5 (11–40) 10/10 HLA match with donor 11(91.7%) 17 (94.4%) 1 HLA-C loci mismatch with donor 1 (8.3%) 1 (5.6%)

Possession of allo-reactive iKIRs 12 (100%) 18 (100%) mRNA expression of 2DL1 (copies) 0

Conditioning TBI/Cy 1 (8.3%) 2 (11.1%) aGVHD Mortiﬁed Bu/Cy 11 (91.7%) 16 (88.9%) No aGVHD

GVHD P=0.005 CSA+MTX (d1, d3, d6, 10 (83.3%) 16 (88.9%) 5000 d11)+MMF+ATG 4000 Abbreviations: ATG ¼ antithymocyte globulin; MMF ¼ mycophenolate mofetil. 3000

RT-PCR detection of KIR post transplant 2000 Of the 75 pairs, KIRs from 30 patients were analyzed by RT-PCR for the genes 2DL1, 2DS1, 3DL1 and 3DS1. The 1000 mRNA expression was reliably detected by (RT) PCR whenever the gene was present. All of the 30 unrelated recipient/donor pairs carried allo-reactive iKIR in the mRNA expression of 3DL1 (copies) 0 GVH direction. This means the NK cells of donor-origin possessed at least one iKIR for which a ligand was missing aGVHD in the recipient. The transplant characteristics and clinical No aGVHD outcomes are shown in Table 3. For the 30 transplants, the Figure 2 Difference of the mRNA expression of inhibitory killer Ig-like mRNA expression of 2DL1 increased signiﬁcantly in receptor (iKIR) in the no acute GVHD (aGVHD) group (n ¼ 18) and patients with aGVHD 30–90 days after transplantation the aGVHD group (n ¼ 12). (a) The mRNA expression of KIR2DL1 in the compared with patients without aGVHD (P ¼ 0.021). The early stages post-HSCT (1–3 month) of the two groups (P ¼ 0.021). (b)The same difference was observed for 3DL1 (P ¼ 0.005). The mRNA expression of KIR3DL1 in the same period of the two groups (P ¼ 0.005). results suggested the association of a high expression of KIR2DL1 and 3DL1 in early stages (oday 90) post transplantation with the occurrence of aGVHD (Figure 2). This conclusion was more obvious when the recipient/ knowledge that NK cells would kill any target cells that donor pairs were all KIR haplotype A/A. A signiﬁcant lacked MHC class I molecules.24 It has been known for statistical difference of 2DL1 and 3DL1 copy numbers many years that NK cells have a predilection for killing could be found by 90 days post transplantation between the hematopoietic cells in mismatched BMT.25 In allo-HSCT, groups with and without aGVHD (z ¼ 2.558, P ¼ 0.011) the inhibitory signal for donor’s iKIRs could be lacking (Table 4). because of the absence of their HLA class I ligands on recipient’s cells, resulting in NK cell stimulation as illustrated by the model of activating pathways.26 Usually, the activation pathways are kept in check by signals coming Discussion from inhibitory receptors.21,22 In the absence or down- regulation of self MHC class I on the target cells after HLA Recognition of MHC class I ligands by inhibitory NK mismatched allo-HSCT, these stimulatory signals are no receptors is a critical part of NK cell education or longer suppressed, resulting in NK cell responses including maturation and is required for NK cells to recognize cytokine production and granule release causing cytotoxi- missing ligand.19,20 Once engaged with classical MHC class city.27,28 The advantage for patients receiving T-cell- I, the inhibitory KIR sends a signal to the NK cell that depleted transplants in which the NK allo-reactivity exists dominates over any activating signal and prevents NK cells in the graft-versus-leukemia direction has been described by from killing the MHC class I expressing cell.21–23 Experi- Ruggeri et al.29 In the absence of NK cell allo-reactivity in ments in mouse models of BM graft rejection led to the HLA-identical HSCT it appears to be due to ‘missing

Bone Marrow Transplantation 2DS4 alleles and KIR expression in acute GVHD XJ Bao et al 1440 Table 4 The relationship of patient 2DL1, 3DL1 mRNA expres- related donor allo-HSCT.7 A recent study reported that a sion and aGVHD when both the donor and recipient are haplotype A missing HLA-C2 ligand for donor inhibitory KIR2DL1 homozygous (n ¼ 10) was significantly associated with an increased risk of acute 38 2DL1 3DL1 GVHD. Our findings provide evidence of an association of high expression of KIR2DL1/3DL1 in the early stage No aGVHD group (n ¼ 4) after transplantation with aGVHD. Although the effect of Donor 251±378 517±558 KIR expression on the occurrence of GVHD seemed Patient significant in our study, this hypothesis needs to be further Day 30 191±186 407±289 confirmed in an expanded cohort of patients. Day 60 175±157 484±107 In conclusion, the analysis of KIR genes in stem cell ± a ± b Day 90 724 165 766 167 donors showed a reduced survival if transplanted with aGVHD group (n ¼ 6) a donor who was KIR homozygous for haplotype A. Our Donor 217±193 815±521 findings showed a significant association of the expression of full-length KIR2DS4 or a high expression level of Patient KIR2DL1/3DL1 in the early stages after allo-HSCT with Day 30 309±479 707±992 Day 60c 1066 2812 aGVHD. Taken together, these results suggested that the Day 90 2203±756a 3741±481b combination of KIR and HLA genotyping could help select transplant donors and improve the outcome of transplan- Abbreviations: aGVHD ¼ acute GVHD; cGVHD ¼ chronic GVHD. tation. Dynamic detection of KIR2DL1/3DL1 expression a Statistical difference of the 2DL1 expression in patient at day 90 between would be beneficial for prediction of aGVHD after the no aGVHD group and the aGvHD group (z ¼ 2.558, P ¼ 0.011). bStatistical difference of the 3DL1 expression in patient at day 90 between transplantation. In addition, one may ask whether trans- the no aGVHD group and the aGVHD group (z ¼ 2.558, P ¼ 0.011). plants from KIR haplotype B donors would have a higher cOnly one sample was drawn at day 60 in the aGVHD group. risk of GVHD because the donors have more aKIR genes. However, our study did not show much difference. A majority of these haplotype B consists of multiple genes ligands’ in the recipients,6,7 as well as the presence coding for aKIR and although there was no difference in of activating receptors’ affinity to their ligands.30 The aGVHD, and OS rate was better. combination of incoming aKIR signals with the lack of inhibition of NK cells in the post transplant period tips the balance of NK regulation to favor the reactive state.15 Conflict of interest The aKIR can deliver an activating signal, leading to IFN-g production, cytotoxicity and proliferation.23,31 In The authors declare no conflict of interest. this study, we found that transplants from haplotype B/x donors who carried more aKIRs had a better OS, which was consistent with Chen et al. and Cooley et al.8,18 Acknowledgements However, a recent study also confirmed that the production of IFN-g was associated with the incidence of The KIR2DS4 sequencing was supported by a grant to Drs aGVHD.32 The role of NK cells in the pathophysiology of Carolyn Hurley and Jennifer Ng (CW Bill Young Marrow aGVHD remains to be elucidated. Acute GVHD is a T-cell- Donor Recruitment and Research Program, Georgetown Uni- mediated disease, but there is evidence showing that it can versity Medical Center). We thank them for their support of this be modified by IFN-g production of NK cells and by the project. We thank Prof Yueping Shen (Medical College of kinetics of KIR receptor reconstitution.32 NK cells posses- Soochow University) for his help in statistical analysis. This sing more activating receptors were characterized by the work was supported by the grants from The Science Foundation of Department of Health (No: wkj2006-2-023); The Talent Man ability to produce IFN-g and thus induce a strong cytolytic 8 Foundation of Jiangsu Province (No: 07-B-21); Natural Science response. The aKIR in donors was considered a potential Foundation of Jiangsu Province (No: 07kj320106); and The 9 high-risk marker for aGVHD. Our results showed that Cooperation Foundation of International (No: SWH0716). 2DS4 full-length allele, as the only aKIR gene in donor, was closely related with an increasing risk of aGVHD, 33 especially III–IV aGVHD, in group A/A. When donors References possessed two 2DS4-full alleles, there might be more functional activating receptors on their NK cells compared 1 Ruggeri L, Capanni M, Casucci M, Volpi I, Tosti A, Perruccio with those who possessed less 2DS4-full alleles. The NK K et al. Role of natural killer cell alloreactivity in HLA- cells, activated through missing ligand mechanisms, could mismatched hematopoietic stem cell transplantation. Blood thus create a setting in which T cells cause aGVHD.34 For 1999; 94: 333–339. example, activated NK cells secrete IFN-g that contributes 2 Ruggeri L, Capanni M, Urbani E, Perruccio K, Shlomchik et al. to DC activation and thus indirectly lead to antigen-specific WD, Tosti A Effectiveness of donor natural killer 35 bright cell alloreactivity in mismatched hematopoietic transplants. T-cell priming. CD 56 NK-cells also contribute Science 2002; 295: 2097–2100. 36 directly to T-cell priming. Besides, alloreactive NK cells 3 Middleton D, Gonzalez A, Gilmore PM. Studies on the might directly attack non-hematopoietic cells and so expression of the deleted KIR2DS4*003 gene product and contribute to aGVHD.37 It is illustrated that missing KIR distribution of KIR2DS4 deleted and nondeleted versions in ligands are associated with increased GVHD after un- different populations. Hum Immunol 2007; 68: 128–134.

Bone Marrow Transplantation 2DS4 alleles and KIR expression in acute GVHD XJ Bao et al 1441 4 Shilling HG, McQueen KL, Cheng NW, Shizuru JA, Negrin histocompatibility complex class I molecules. Nature 2005; RS, Parham P. Reconstitution of NK cell receptor repertoire 436: 709–713. following HLA-matched hematopoietic cell transplantation. 21 Bryceson YT, March ME, Ljunggren HG, Long EO. Activa- Blood 2003; 101: 3730–3740. tion, coactivation, and costimulation of resting human natural 5 Shilling HG, Young N, Guethlein LA, Cheng NW, Gardiner killer cells. Immunol Rev 2006; 214: 73–91. CM, Tyan D et al. Genetic control of human NK cell 22 Vivier E, Nunes JA, Vely F. Natural killer cell signaling repertoire. J Immunol 2002; 169: 239–247. pathways. Science 2004; 306: 1517–1519. 6 Hsu KC, Keever-Taylor CA, Wilton A, Pinto C, Heller G, 23 Warren HS, Campbell AJ, Waldron JC, Lanier LL. Biphasic Arkun K et al. Improved outcome in HLA-identical sibling response of NK cells expressing both activating and inhibitory hematopoietic stem-cell transplantation for acute myelogenous killer Ig-like receptors. Int Immunol 2001; 13: 1043–1052. leukemia predicted by KIR and HLA genotypes. Blood 2005; 24 Murphy WJ, Kumar V, Bennett M. Rejection of bone marrow 105: 4878–4884. allografts by mice with severe combined immune deficiency 7 Miller JS, Cooley S, Parham P, Farag SS, Verneris MR, (SCID). Evidence that natural killer cells can mediate the McQueen KL et al. Missing KIR ligands are associated with specificity of marrow graft rejection. J Exp Med 1987; 165: less relapse and increased graft-versus-host disease (GVHD) 1212–1217. following unrelated donor allogeneic HCT. Blood 2007; 109: 25 Ciccone E, Viale O, Pende D, Malnati M, Biassoni R, 5058–5061. Melioli G et al. Specific lysis of allogeneic cells after activation 8 Chen C, Busson M, Rocha V, Appert ML, Lepage V, Dulphy of CD3- lymphocytes in mixed lymphocyte culture. J Exp Med N et al. Activating KIR genes are associated with CMV 1988; 168: 2403–2408. reactivation and survival after non-T-cell depleted HLA- 26 Campbell KS, Cella M, Carretero M, Lopez-Botet M, identical sibling bone marrow transplantation for malignant Colonna M. Signaling through human killer cell activating disorders. Bone Marrow Transplant 2006; 38: 437–444. receptors triggers tyrosine phosphorylation of an associated 9 Gagne K, Brizard G, Gueglio B, Milpied N, Herry P, protein complex. Eur J Immunol 1998; 28: 599–609. Bonneville F et al. Relevance of KIR gene polymorphisms in 27 Ljunggren HG, Karre K. In search of the ‘missing self’: MHC bone marrow transplantation outcome. Hum Immunol 2002; molecules and NK cell recognition. Immunol Today 1990; 11: 63: 271–280. 237–244. 10 Malmberg KJ, Schaffer M, Ringden O, Remberger M, Ljunggren 28 Smyth MJ, Cretney E, Kelly JM, Westwood JA, Street SE, HG. KIR-ligand mismatch in allogeneic hematopoietic stem cell Yagita H et al. Activation of NK cell cytotoxicity. Mol transplantation. Mol Immunol 2005; 42: 531–534. Immunol 2005; 42: 501–510. 11 Vampa ML, Norman PJ, Burnapp L, Vaughan RW, Sacks 29 Ruggeri L, Mancusi A, Capanni M, Urbani E, Carotti A, SH, Wong W. Natural killer-cell activity after human renal Aloisi T et al. Donor natural killer cell allorecognition of transplantation in relation to killer immunoglobulin- missing self in haploidentical hematopoietic transplantation like receptors and human leukocyte antigen mismatch. for acute myeloid leukemia: challenging its predictive value. Transplantation 2003; 76: 1220–1228. Blood 2007; 110: 433–440. 12 Hou L, Steiner NK, Chen M, Belle I, Kubit AL, Ng J et al. 30 Ogasawara K, Benjamin J, Takaki R, Phillips JH, Lanier LL. Limited allelic diversity of stimulatory two-domain killer cell Function of NKG2D in natural killer cell-mediated rejection immunoglobulin-like receptors. Hum Immunol 2008; 69: 174–178. of mouse bone marrow grafts. Nat Immunol 2005; 6: 938–945. 13 Robinson J, Waller MJ, Stoehr P, Marsh SG. IPD–the 31 Spaggiari GM, Contini P, Dondero A, Carosio R, Puppo F, Immuno Polymorphism Database. Nucleic Acids Res 2005; Indiveri F et al. Soluble HLA class I induces NK cell apoptosis 33 (Database issue): D523–D526. upon the engagement of killer-activating HLA class I receptors 14 Valiante NM, Uhrberg M, Shilling HG, Lienert-Weidenbach through FasL-Fas interaction. Blood 2002; 100: 4098–4107. K, Arnett KL, D’Andrea A et al. Functionally and structurally 32 Cooley S, McCullar V, Wangen R, Bergemann TL, Spellman distinct NK cell receptor repertoires in the peripheral blood of S, Weisdorf DJ et al. KIR reconstitution is altered by T cells in two human donors. Immunity 1997; 7: 739–751. the graft and correlates with clinical outcomes after unrelated 15 Khakoo SI, Thio CL, Martin MP, Brooks CR, Gao X, donor transplantation. Blood 2005; 106: 4370–4376. Astemborski J et al. HLA and NK cell inhibitory receptor 33 Uhrberg M, Valiante NM, Shum BP, Shilling HG, Lienert- genes in resolving hepatitis C virus infection. Science 2004; 305: Weidenbach K, Corliss B et al. Human diversity in killer cell 872–874. inhibitory receptor genes. Immunity 1997; 7: 753–763. 16 Cooley S, Xiao F, Pitt M, Gleason M, McCullar V, 34 Sun JY, Dagis A, Gaidulis L, Miller MM, Rodriguez R, Bergemann TL et al. A subpopulation of human peripheral Parker P et al. Detrimental effect of natural killer cell blood NK cells that lacks inhibitory receptors for self-MHC is alloreactivity in T-replete hematopoietic cell transplantation developmentally immature. Blood 2007; 110: 578–586. (HCT) for leukemia patients. Biol Blood Marrow Transplant 17 Bao XJ, He J, Chen ZX, Wu DP, Yao L, Yuan XN et al. Study 2007; 13: 197–205. on the behavior of NK cell KIRs of donor/recipient pairs in 35 Lanier LL, Phillips JH, Hackett Jr J, Tutt M, Kumar V. HLA matched unrelated allo-HSCT. Zhonghua Xue Ye Xue Za Natural killer cells: definition of a cell type rather than a Zhi 2007; 28: 510–513. function. J Immunol 1986; 137: 2735–2739. 18 Cooley S, Trachtenberg E, Bergemann TL, Saeteurn K, Klein 36 Cudkowicz G, Bennett M. Peculiar immunobiology of bone J, Le CT et al. Donors with group B KIR haplotypes improve marrow allografts. II. Rejection of parental grafts by resistant relapse-free survival after unrelated hematopoietic cell trans- F 1 hybrid mice. J Exp Med 1971; 134: 1513–1528. plantation for acute myelogenous leukemia. Blood 2009; 113: 37 Vivier E, Tomasello E, Baratin M, Walzer T, Ugolini S. 726–732. Functions of natural killer cells. Nat Immunol 2008; 9: 503–510. 19 Anfossi N, Andre P, Guia S, Falk CS, Roetynck S, Stewart CA 38 Ludajic K, Balavarca Y, Bickeboller H, Rosenmayr A, Fae I, et al. Human NK cell education by inhibitory receptors for Fischer GF et al. KIR genes and KIR ligands affect occurrence MHC class I. Immunity 2006; 25: 331–342. of acute GVHD after unrelated, 12/12 HLA matched, 20 Kim S, Poursine-Laurent J, Truscott SM, Lybarger L, Song hematopoietic stem cell transplantation. Bone Marrow Trans- YJ, Yang L et al. Licensing of natural killer cells by host major plant 2009; 44: 97–103.

Bone Marrow Transplantation