Certain Killer Immunoglobulin-Like Receptor (KIR)/KIR HLA Class I Ligand Genotypes Influence Natural Killer Antitumor Activity in Myelogenous Leukemia but Not in Acute Lymphoblastic Leukemia: A Case Control Leukemia Association Study

Objective: Natural killers (NK) cell function is mainly controlled by the expression of killer immunoglobulin-like receptors (KIRs) and their ligation with the corresponding ligands. The objective of this study was to investigate the putative association of KIRs, HLA class I ligands, and KIR/ligand combinations with rates of development of acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), and chronic myeloid leukemia (CML). Materials and Methods: The KIR/HLA I genotypes of 82 patients with leukemia (ALL, n=52; AML, n=17; and CML, n=13) were determined by PCR-SSP method and compared with genotypes of healthy controls (n=126). Results: KIR genotype frequency differed significantly between myelogenous leukemia patients and healthy controls for KIR2DL5A (17.6% vs. 47.7%, p=0.02), KIR3DS1 (17.6% vs. 47.6%, p=0.02), and KIR2DS4*001 (36.6% vs. 20.2%, p=0.017). The incidence of homozygous HLA-BBw4 (31.0% vs. 12.5%, p=0.042) and HLA-Bw4Thr80 Thr80 (13.0% vs. 1.2%, p=0.01) was significantly elevated in myeloid leukemia patients compared to healthy controls. KIR/HLA class I ligand profile KIR3DS1(+)/L (-) was decreased and KIR3DL2(+)/HLA-A3/11(-) was increased among myeloid leukemia cases compared to controls. Conclusion: These data suggest that the activity of NK cells as determined by inherited KIR/HLA class I ligand polymorphisms influences the susceptibility to myelogenous leukemia, but not to lymphoblastic leukemia. Additionally, the KIR genotype characterized by the absence of the inhibitory KIR2DL2 and the activating KIR2DS2 and KIR2DS3 (ID2) was found at a lower frequency in patients compared to controls, which confirmed the need for complex analysis based on all possible KIR/HLA class I ligand polymorphism combinations.


Introduction
In 2015, 4062 patients with leukemia were treated and/or monitored in the Republic of Bulgaria [1]. Similarly to other neoplastic diseases, the precise etiopathogenetic mechanism that leads to the transformation of "normal" hematopoietic cells into blast cells has not yet been elucidated. The idea of a combined influence of external environmental factors and "internal" factors in the onset and development of malignancies is increasingly being discussed.
NK cells play an important role in antitumor immune defense [2,3]. Their function is controlled by the interaction of cell surface receptors with appropriate ligands, among which the most important and best studied are killer immunoglobulin-like receptors (KIRs) and their HLA class I ligands [4,5,6,7,8]. Seventeen KIR genes and pseudogenes have been described so far [4]. KIRs are named based on the number of their extracellular Ig-like domains (2D or 3D) and by the length of their cytoplasmic tail (long [L], short [S], or pseudogene [P]) [9]. Ligands for most KIRs are specific patterns of the HLA class I molecules [6,7,8,9,10]. HLA C molecules with amino acid residues Ser and Asn at positions 70 and 80, respectively (Ser77 and Asn80), form the HLA-C1 KIR ligand group, which specifically binds KIR2DL1 and KIR2DS2. The HLA-C2 KIR ligand group (Asn77 and Lys80) interacts with KIR2DL2/2DL3 and KIR2DS1 [5,6,10,11,12,13]. HLA-B class I molecules with the Bw4 epitope are ligands for KIR3DL1 and KIR3DS1 [7,8,14,15]. The strength of the KIR/ligand binding is determined by the amino acid residue at position 80 in the Bw4 molecule (Bw4 Ile80 are stronger ligands for their specific KIRs than Bw4 Thr80 ) [7,8]. Data suggest that HLA-A alleles with Bw4 epitopes may be ligands for KIR3DL1 [11]. KIR3DL2 specifically recognizes HLA-A3 and HLA-A11 specific to certain peptides such as Epstein-Barr virus peptides [16,17].
The first immunogenic studies investigating the effect of KIRs and their HLA class I ligands and their relation to the development of oncohematological diseases were conducted by Demanet et al. [18] and Verheyden et al. [19] in 2004. There are increasing data on the association of genetic polymorphisms of KIRs and their HLA ligands with a predisposition to various hematological malignancies, although a specific polymorphism clearly associated with leukemia development has not been identified [20]. Indirect evidence for the role of NK cells in leukemia defense includes the proved decreased incidence of relapse and increased leukemia-free surveillance in the setting of allogenic KIR HLA class I ligand donor-recipient incompatibility hematopoietic stem cell transplantation [21,22]. These data additionally suggest that NK cells may play a major role in the control and clearance of leukemia. It seems that the interplay between the inhibitory KIR signals and/or the predominance of activating ones is critical for the NK-mediated antileukemic effect. Therefore, in the present study, we studied the polymorphism of NK-cell receptors, namely KIRs and their HLA class I ligands, in patients with leukemia and healthy controls and investigated the possible association of these immunogenic factors with different leukemias in the Bulgarian population.

Study Groups
Eighty-two patients with primary leukemia and 126 healthy controls were included after they provided signed informed consent. The healthy individuals were randomly selected from unrelated volunteers from the Bulgarian population, without chronic diseases and with negative family history of hereditary diseases, autoimmune diseases, or malignancies (46.1% male and 53.9% female, mean age 46.6±11.9 years). The patient group consisted of patients with acute lymphoblastic leukemia (ALL) (n=52, 65.4% male and 34.6% female, mean age 20.8±12.5 years), acute myeloid leukemia (AML) (n=17, 64.7% male and 35.3% female, mean age 36.9±11.3 years), and chronic myeloid leukemia (CML) (n=13, 84.6% male and 15.4% female, mean age 36.8±13.4 years).
KIR genotyping was performed by PCR-SSP methods (Olerup SSP™ KIR and KIR/HLA Ligand Kit, Sweden) according to the manufacturer's instructions. In brief, 24 locus-specific primer sets in the KIR genotyping kit allow detection of 16 KIR genes and pseudogenes and discrimination of KIR2DL5A, KIR2DL5B, and KIR3DL1*004 alleles and both groups of KIR2DS4 alleles (KIR2DS4*001 from KIR2DS4*003/004/006/007). The typing was interpreted with the worksheet provided with the kit. KIR/HLA ligands were determined as previously described [9].

Statistical Analysis
Individual KIR genes, KIR HLA class I ligands, and KIR/HLA class I ligand combination frequencies were determined by direct counting. KIR haplotypes and genotypes were defined in accordance with the Allele Frequency Net Database [20]. Subsequently, individual KIR genotype frequencies were also determined by direct counting. The established frequencies of each of the factors studied were compared between patients and healthy controls using the Pearson chi-square test and Fisher exact test. Odds ratios (ORs) with 95% confidence interval (CIs) were assigned to variables with significant differences determined at a threshold of p<0.05. All statistical analyses were performed with SPSS 16.0 for Windows (SPSS Inc., Chicago, IL, USA).
The frequency of individual KIR genes and the comparison between the patients and healthy controls are presented in Table 1.
To determine and analyze whether there were findings specific to a particular leukemia type, the KIR frequencies were compared between patients with ALL, AML, and CML. The applied intragroup analysis did not show appreciable differences in the distribution of any KIRs (data not shown).
CML: Chronic myeloid leukemia, ALL: acute lymphoblastic leukemia, ML: myeloid leukemia, AML: acute myeloid leukemia. and healthy controls, whose characteristics and frequency are presented in

KIR/HLA Class I Ligand Combinations
The frequencies of individual KIR/HLA class I ligand combinations of inhibitory receptor with the appropriate ligand and the activating counterparts are presented in In the next step, the KIR/HLA class I ligand combinations were investigated taking into account the ligand and the combination of its appropriate inhibitory KIR and activating KIR counterpart (inhibitory KIR/activating KIR/HLA class I ligand). Significant differences were not found between patients and healthy controls in this assay (data not shown). The subgroup analysis, depending on the cytological variant of leukemia, also showed no differences (data not shown).

Discussion
Particular KIRs and KIR HLA class I ligand polymorphisms associated with a variety of tumors have been reported but the precise disease-predisposing mechanisms have not been elucidated [20].
The KIR2DS4*001 allele was found to be significantly more frequent in the leukemia group compared to healthy controls with the difference being more prominent for AML. Two additional KIRs were identified as protective for AML: KIR2DL5A and KIR3DS1. The protective effect of KIR3DS1 that we established supports the hypothesis that genetic imbalance between activating and inhibitory KIRs in the direction of decreased activation/increased inhibition may contribute to tumorigenesis. These results are in line with the data from a similar disease-associated study in AML patients from Iran [23], as well as its reported protective effect associated with solid tumors [24] and Hodgkin's disease [25]. However, this hypothesis cannot explain the lower incidence of inhibitory KR2DL5A and the higher incidence of KIR2DS4*001 in the AML group, logically associated with decreased inhibitory and increased NK cell-activating function. KIR2DL5 has been found less frequently in patients with oncohematological diseases such as B-cell chronic lymphocyte leukemia [26] and Hodgkin's lymphoma patients [24]. Similarly, higher incidence of KIR2DS4 associated with leukemia was reported independently by Giebel    for acute leukemia [29,33]; in other words, their absence in the KIR genotype ID2 can be interpreted as absence of a genetically predetermined disease susceptibility factor and higher tumor resistance.
Analysis of disease susceptibility by testing the inherited KIRs ligands showed a higher incidence of the HLA-BBw4 (HLA-BBw4/Bw4) homozygous genotype in patients compared to healthy controls. Particularly at risk were patients in the myeloid leukemia group. HLA-BBw4 is a ligand for both KIR3DL1 and KIR3DS1, the former binding it with greater affinity than its activating counterpart. Moreover, KIR3DL1 is significantly more frequent than KIR3DS1. It can be assumed that the expression of HLA-Bw4 ligands maintains NK cells in a state of hyporesponse rather than contributing to NK activation by KIR3DS1/HLA-BBw4. Thus, HLA-BBw4/Bw4 appears to be a risk factor for leukemia development. Two independent studies by Middleton et al. [34] and de Smith et al. [35] also reported homozygous HLA-BBw4 as a risk factor for AML and CML development. de Smith et al. [35] also showed that the KIR3DL1+/HLA-BBw4/ Bw4 combination was associated with ALL. Additionally, carriers of two HLA-Bw4 alleles with threonine at the 80 position (HLA-Bw4Thr80 Thr80) were found with a higher frequency in myeloid leukemia cases compared to healthy controls. Bw4Thr80 binds KIR3DL1 with a lower affinity than Bw4Iso80, which results in a lower inhibitory signal to the NK cells by HLA-Bw4Thr80 homozygous individuals. Association of the KIR HLA-Bw4 ligand according to amino acid at position 80 was also reported by Shen et al. [36], who demonstrated significantly higher frequencies of HLA-Bw4Iso80 in the prognostically "poor" AML risk group compared to those with "favorable" risk. In contrast to other studies of KIR HLA class I ligands in leukemia, no other differences were observed [36,37].
Analysis of KIR/KIR HLA class I ligand combinations first confirmed that the investigated polymorphic gene systems have the highest importance for myeloid leukemia susceptibility, whereas their role in disease predisposition to ALL could not be confirmed. The KIR3DL2(+)/HLA-A3/11(-) combination was found significantly more frequently in myeloid leukemia patients than in the healthy population. KIR3DL2 belongs to the framework of KIR genes and is present almost ubiquitously [20]. Thus, the myeloid leukemia-associated genotype KIR3DL2(+)/ HLA-A3/11(-) very likely indicates lack of NK cell activity mediated by the inhibitory receptor. Another KIR/HLA class I ligand combination where the activating KIR3DS1 is expressed, but not its putative HLA-A/BBw4 ligand, KIR3DS1(+)/L(-), was found at a lower frequency in the myeloid leukemia group compared to the healthy individuals. On the contrary, La Nasa et al. reported an increased risk of Hodgkin's disease associated with genotype KIR3DS1(+)/HLA-Bw4(-) [38]. It should be noted, however, that in our study, KIR3DS1 was found to be significantly less common among patients with AML, which may be an explanation for the observed differences and logically raises the question of whether KIR3DS1 is an independent protective factor for leukemia development or the receptor-ligand combinations in which it participates are also important. There is support for both hypotheses in the available literature [23,25]. For the second possible mechanism [26,30], the discussions are only addressing the presence of the binding ligand.
In summary, it is obvious that the patient group is distinguished from healthy controls by the presence of both individual KIR genes and some KIR HLA class I ligands and KIR/HLA class I ligand combinations. These data support the hypothesis of the complex influence of various polymorphic gene systems, in particular KIRs and KIR HLA class I ligands, in the genetically regulated NK immune response. On the other hand, the differences found are valid for patients with myeloid leukemia, but not for the ALL group. These results are not unusual considering the higher susceptibility of myeloblastic cells compared to lymphoblastic cells to NK-mediated cytolysis [39,40,41] and the presumed direct involvement of NK cells in the antitumor response in hemoblastosis of myeloid origin.

Study Limitations
As a limitation of this study, most importantly, the group of patients analyzed was heterogeneous and included three different types of leukemia characterized by different clinical evolution and prognosis. When dividing patients into separate groups, depending on the type of leukemia (ALL, AML, and CML), the number of subjects analyzed in each group was limited, and highly variable factors such as KIR genotypes did not allow the comparison of each subgroup of leukemia with healthy controls.

Conclusion
The leukemia susceptibility factors we have found confirm the importance of KIR/HLA class I ligand gene systems in NKmediated antitumor response in patients with myeloid leukemia. The understanding of the mechanisms of their influence on NK cell function remains limited. Interpretation of the results obtained in the context of the hypothesis of different NK cell activity predetermined at the genetic level depending on the inherited inhibitory/activating potential is difficult due to the poorly studied ligand specificity of the KIR genes as well as their functional activity.

Authorship Contributions
Surgical