High-Resolution KIR and HLA Genotyping in Three Chinese Ethnic Minorities Reveals Distinct Origins

Abstract

Polymorphism of Killer-cell Immunoglobulin-like Receptors (KIR) and their HLA class I ligands impacts the effector activity of cytotoxic NK cell and T cell subsets. Therefore, understanding the extent and implications of KIR and HLA class I genetic polymorphism across various populations is important for immunological and medical research. In this study, we conducted a high-resolution investigation of KIR and HLA class I diversity in three distinct Chinese ethnic minority populations. We studied the She, Yugur, and Tajik, and compared them with the Zhejiang Han population (Zhe), which represents the majority Southern Han ethnicity. Our findings revealed that the Tajik population exhibited the most diverse KIR copy number, allele, and haplotype diversity among the four populations. This diversity aligns with their proposed ancestral origin, closely resembling that of Iranian populations, with a relatively higher presence of KIR-B genes, alleles, and haplotypes compared to the other Chinese populations. The Yugur population displayed KIR distributions similar to those of the Tibetans and Southeast Asians, whereas the She population resembled the Zhe and other East Asians, as confirmed by genetic distance analysis of KIR. Additionally, we identified 12.9% of individuals across the three minority populations as having KIR haplotypes characterized by specific gene block insertions or deletions. Genetic analysis based on HLA alleles yielded consistent results, even though there were extensive variations in HLA alleles. The observed variations in KIR interactions, such as higher numbers of 2DL1-C2 interactions in Tajik and Yugur populations and of 2DL3-C1 interactions in the She population, are likely shaped by demographic and evolutionary mechanisms specific to their local environments. Overall, our findings offer valuable insights into the distribution of KIR and HLA diversity among three distinct Chinese ethnic minority populations, which can inform future clinical and population studies.

Introduction

Killer-cell immunoglobulin-like receptors (KIR) are expressed on the surface of natural killer (NK) cells and subsets of cytotoxic T cells.^1,2 Through interaction with polymorphic human leukocyte antigen (HLA) class I molecules expressed on tissue cells,¹ polymorphic inhibitory and activating KIR have critical roles in regulating the function of these vital immune effector cells. Genomic diversity of KIR and HLA is thus implicated in modulating the susceptibility and progress of infection, autoimmunity, and tumors as well as outcome of fetal implantation, organ transplantation and immunotherapies.^{3, 4, 5}

The human KIR locus spans 150–350 kbp of the leukocyte receptor complex on the long arm of chromosome 19 (19q13.4).⁶ The KIR gene family contains 13 functional genes KIR2DL1, KIR2DL2/3, KIR2DL4, KIR2DL5A, KIR2DL5B, KIR3DL1/S1, KIR3DL2, KIR3DL3, and KIR2DS1–2DS5, and two pseudogenes, KIR2DP1 and KIR3DP1.^{7, 8, 9} KIR genes are characterized by high sequence homology across the locus, coupled with genomic structural variation and allelic polymorphism. Unlike most genes, which usually have only one copy per haplotype, some KIR genes have from 0–3 copies per haplotype.¹⁰ This haplotype diversity is enhanced through meiotic recombination, with resultant duplications, deletions, gene fusions and novel alleles, each having capacity to directly affect effector cell functions.^{10, 11, 12, 13, 14} All these characteristics make KIR highly complex to analyze and interpret.

There are four framework genes (KIR3DL3, KIR3DP1, KIR2DL4 and KIR3DL2), which flank the ‘centromeric’ and ‘telomeric’ oriented regions of most KIR haplotypes. Aside from this similarity, the genes are organized into two haplotype classes, termed KIR-A and KIR-B.^{10, 11, 12} KIR-A carry a fixed content that includes inhibitory receptors specific for HLA-A, -B and -C. KIR-A haplotypes carry only one activating-receptor gene (KIR2DS4), whereas KIR-B haplotypes can carry one or more of six genes encoding activating KIR (KIR2DS1-S5, KIR3DS1) and fewer encoding inhibitory receptors. Thus as example, KIR2DL2 is specific to KIR-B haplotypes, whereas KIR2DL1 can be carried by either haplotype class.

KIR genes, alleles, and haplotypes correlate directly with NK cell function through affecting receptor specificity, binding strength, signal transduction and NK cell development.^{15, 16, 17, 18, 19} As examples, KIR3DS1 associates with slower progression to AIDS in HIV infected patients,^{20, 21} and KIR2DL4 participates in the NK-mediated maternal/fetal interface control during pregnancy.^{22, 23} KIR-A haplotypes often contribute to the regulation of immune responses against infectious diseases. In contrast, KIR-B haplotypes can provide protection against NK cell-mediated pregnancy syndromes and graft rejection events.^{4, 16, 24, 25} It is important to note that the influence of KIR genes on NK cell function is not solely determined by KIR. These receptors typically interact differentially with subsets of HLA class I allotypes, relying on specific outward-facing amino acid motifs. These motifs include the Bw4 epitope, spanning residues 77–83, which is present on multiple HLA-A and HLA-B allotypes, and the mutually exclusive C1 or C2 epitope, defined by residues 77–80 of HLA-C.^{26, 27, 28, 29}

Accumulating evidence shows that the distribution of KIR and HLA genes, alleles, and haplotypes varies tremendously worldwide, likely contributing to diversity in immunity and susceptibility to immune-mediated diseases within and across populations. In general, the extent of this diversity correlates with genome-wide diversity, with exceptions indicating population-specific natural selection.^{30, 31}

Due to their importance in the regulation of immunity, as well as disparities in the incidence rates of immune-mediated diseases across populations, a thorough understanding of worldwide KIR and HLA diversity is critical. In our previous study,³² we investigated the distribution of KIR and HLA in Zhejiang Han (Zhejiang Han population; abbreviated here as Zhe), representing the Southern Han, which is the largest ethnic group in China. However, there are 56 officially recognized nationalities in China. According to the main data of the seventh national population census,³³ the Han population accounts for 91.11% of the people living in China, while the total component of the other 55 ethnic groups is relatively small, accounting for 8.89% of the total population. These so-called ethnic minorities tend to have their own language and culture. To address the lack of knowledge of KIR and HLA diversity within the ethnic minorities of China, we applied a next generation sequencing (NGS) method to analyze all aspects of KIR and HLA genomic variation for three of the populations, the Tajik, Yugur, and She (Figure 1). This study will help towards a deeper understanding of the diversity and function of KIR and HLA in regulating NK cells across populations.

Figure 1. The map of China shows the distribution of She, Yugur, and Tajik population.

The colour keys at the bottom display the location of the She, Yugur, and Tajik regions.

Material and Methods

Study Population

A total of 332 whole blood specimens were collected from unrelated healthy adult individuals representing three ethnic minorities in China: Yugur (n=100), She (n=135), and Tajik (n=97). Specimen collection was approved by the regional ethics committee in the Blood Center of Zhejiang Province. All participants gave informed consent. All specimens were stored below −20°C before use. Genomic DNAs were extracted using commercial MagNA Pure LC DNA Isolation Kits (Roche Diagnostics, Indianapolis, IN, USA) according to the manufacturer’s instruction. The final DNA concentration was adjusted to 60 ng/μl and optical density at 260/280 was approximately 1.8.

High-resolution next-generation sequencing for KIR

According to the previously reported NGS method, a group of oligonucleotide probes that capture all KIR genes are used to enrich genomic libraries prior to sequencing using the Illumina platform.³⁴ After sequencing, the data were analyzed using Pushing Immunogenetics into the Next Generation program to obtain KIR gene content and allelic genotypes.³⁵ The copy number of each gene was calculated by the ratio of reads mapping to each KIR gene to those mapping to KIR3DL3, which is a reference gene with only one copy on each haplotype. When ambiguous allele results were obtained using PING analysis, they were reanalyzed by manual inspection for the final assignment.

Estimate of KIR Haplotype

The KIR haplotypes of each individual were estimated according to the copy number of each KIR, the linkage disequilibrium among KIR genes, allelic results, and expectation-maximization algorithm using Arlequin software 3.5.2.2.³⁶ Each complete haplotype was divided into two parts: centromeric-oriented (Cen) and telomeric oriented (Tel). Each haplotype was named according to previous reports^{14, 37} based on gene-content, for those that had not been identified before, we named them here as known haplotypes with an inserted or deleted gene or block.

Frequency calculation of KIR alleles, genotypes, and haplotypes

KIR genotypes were named based on the presence or absence of KIR genes according to the allelefrequencies.net database.³⁸ Carrier frequencies of KIR genes and genotypes were calculated as their percentage of the total numbers of individuals. The frequencies of alleles or haplotypes were calculated by direct counting, and the number observed divided by 2N (alleles duplicated on a single haplotype were not included for frequency calculations, and absence was counted as a distinct allele). Cluster analysis was conducted using the Hiplot website (https://hiplot.com.cn/dendrogam plot) based on the KIR distribution data obtained from various populations.

HLA genotyping

All specimens were genotyped for HLA-A, -B, and -C loci using the AllType^™ NGS kit (One Lambda Inc, Canoga Park, CA, USA), as detailed in our previous report.³⁹ In brief, the HLA-A, -B, and -C loci were amplified through a single multiplex PCR. Subsequently, the amplified fragments were fragmented after the addition of adapters, and fragments ranging from 300 to 1000 bp were selected and pooled into a single tube. Library preparation was conducted using the Ion Shear Plus Reagents Kit and Ion Plus Fragment Library Kit (One Lambda Inc, Canoga Park, CA, USA). Amplicon purification, dilution, and library pooling were carried out using the automated Microlab STAR system (Hamilton, Bonaduz, Switzerland). Sequencing was performed on the Ion Torrent S5 platform (ThermoFisher Scientific, Waltham, MA, USA).

HLA genotyping was determined using the HLA TypeStream Visual Software version 2.0 (One Lambda Inc, Canoga Park, CA, USA). The frequencies of HLA alleles, as well as the composition and frequencies of HLA haplotypes, were assessed using Arlequin 3.5.2.2.³⁶ Additionally, the presence of KIR ligands, namely A3/11, Bw4, and C1/C2, was determined based on the corresponding HLA alleles. Cluster analysis was conducted using the Hiplot website based on the HLA allele distribution data obtained from various populations.

KIR/HLA interaction analysis

The interactions between inhibitory KIR and HLA ligands were quantified for each individual, as previously reported.^{30, 32} Broadly, KIR2DL1 recognizes HLA-C2, KIR2DL2/3 recognizes HLA-C1, as well as some HLA-C2 allotype. KIR3DL1 identifies the Bw4 motif based on residues 77–80 on some HLA-A and HLA-B allotypes. Meanwhile, KIR3DL2 recognizes the A3/11 motif, which is present in HLA-A*03 and HLA-A*11 allotypes.^{27, 40} To provide a summary measure of these interactions, we computed the mean number of distinct ligand-receptor allotype pairs for each individual.

Statistical analysis

Hardy-Weinberg equilibrium (HWE) was determined for each gene using Fisher’s exact test implemented in the Arlequin software 3.5.2.2.³⁶ The distributions of KIR and HLA in these three populations were compared to that of Zhe population described previously.³² Besides, the mean number of KIR/HLA interaction were compared with sub-Saharan Africa Ghanaian,⁴¹ European,⁴² Māori,⁴³ Amerindian Yucpa,¹⁰ and other Asian populations from Malaysia,⁴¹ Japan,⁴⁵ Shenzhen Chinese.³⁰ Differences between populations were assessed and performed using GraphPad software by means of the χ2 test for categorical variables. The p values were calculated using Fisher’s exact test (pf). Bonferroni correction for multiple comparisons was applied. p<0.05 was regarded as significant.

Results

KIR-B haplotype genes show significant frequency variation across the ethnic groups of China.

All 13 KIR genes (KIR2DL1, KIR2DL2/3, KIR2DL4, KIR2DL5A, KIR2DL5B, KIR2DS1, KIR2DS2, KIR2DS3, KIR2DS4, KIR2DS5, KIR3DL1/S1, KIR3DL2, KIR3DL3) and two pseudogenes (KIR2DP1 and KIR3DP1) were genotyped to allelic resolution for 332 unrelated individuals representing three Chinese ethnic minorities (the Yugur, She, and Tajik). The results were compared with the Zhe, representing the majority population of Chinese Han, genotyped using the same methods.³² Initial analysis of KIR gene content showed that the distribution of KIR-B haplotype genes varies greatly across the ethnic groups (Supplementary Table 1). For example, KIR2DL2, KIR2DS2, having been observed at 18.18% in Zhe,³² are found at approximately 30% in She and Yugar and almost 60% in Tajik. Indeed, the frequencies of all characteristic KIR-B haplotype genes were highest in Tajik (Figure 2A). The differences are most striking for Cen-B genes (KIR2DL2, KIR2DS2, KIR2DS3, and KIR2DL5B), which are at low frequency in Zhe and observed here at intermediate frequencies in Yugur and She (Figure 2A). For the Cen-B genes, the differences of frequencies observed in Tajik compared to each of the other studied populations were all statistically significant, where only KIR2DS3 lost statistical significance after correction in She and Zhe (Supplementary Table 2). As for the Tel-B genes, they displayed a tendency toward statistical significance (Supplementary Table 2). By contrast, the frequencies of characteristic KIR-A haplotype and KIR-A/B shared genes were less variable across the populations, being present with high frequencies reaching 88.7–100% (Supplementary Table 1).

Figure 2. Comparison of KIR Gene Distribution among Four populations.

A. Shown is the distribution of KIR-B specific genes in the four populations. Colour-coding at the bottom indicates the corresponding populations.

B. Genetic distance among the four populations. The genetic distance was calculated according to the distribution frequency of all KIR genes. Shades of colour represent the distance, from furthest (blue) to closest (red).

C. Shown is the cluster analysis depicting KIR gene frequencies across 17 different populations.

D. Shown is the copy number distribution of KIR-B genes (top) and KIR-A and A/B shared genes (bottom) across the four populations. Colour key at the right shows the copy number.

We calculated genetic distance measurements based on the observed distribution of KIR genes (Figure 2B). This analysis indicated the She to be closer to Zhe, and Tajik to be distant from She, Zhe, and Yugur. We subsequently compared the KIR distribution of these three populations to 13 other populations of African,^{41, 46} European,^{42, 47, 48} Asian,^{44, 45, 49, 50, 51} Oceanian⁵² or Amerindian origin⁵³ (Figure 2C). This cluster analysis indicated that She, Yugur, and Tajik each group with distinct sets of populations. The Tajik showed a short genetic distance to Iranian.⁴⁹ In contrast, the Yugur population exhibited close genetic affinities with the Chinese Tibetan population, as well as with Southeast Asian populations such as Malays⁴⁴ and Thai⁵¹. She shared genetic similarities with Malaysian Chinese,⁴⁴ Zhe, and Japanese,⁴⁵ all of which belong to the Eastern Asian genetic cluster. In summary, the distribution of KIR-B haplotype genes showed significant variation among Chinese ethnic minorities. Notably, the Tajik population stood out, with KIR-B frequencies resembling those found in Iranians and demonstrating proximity to European populations.

Diversity in KIR Gene Copy Number and Genotypes Among Ethnic Groups in China

Gene copy number variation was observed across the KIR genes and four Chinese populations analyzed. In general, KIR-B genes were the most likely to be absent from a given individual (0 or 1 copy per individual, Figure 2D). By contrast the characteristic KIR-A and KIR-A/B shared genes tended to occur as normally diploid (2 copies per individual, Figure 2D). Again, the copy number variability was most obvious for the Tajik, with those individuals carrying three copies of one or more of the genes at relatively high frequency, 6.19% for KIR3DP1 and KIR2DL4, 4.12% for KIR2DL1 and KIR2DP1, and 1.03% for KIR3DL1 and KIR2DS4 (Supplementary Table 3). Interestingly, KIR2DS4 has not been reported at more than two copies in a Chinese population previously. Individuals having either one or three copies of both KIR2DL4 and KIR3DP1 were observed in all four populations, indicating that insertion or duplication haplotypes are present.

Based on the presence, absence and KIR gene copy number, the genotypes were classified according to Allelefrequencies.net database³⁷ denoted categories (Figure 3A). The AA genotype was identified in all four populations (Figure 3A), albeit at significantly lower frequency (16.5%) in Tajik than in Zhe, She and Yugur (49%−56.6%) (Figure 3B). This difference remained after splitting the haplotypes into centromeric and telomeric segments (Figure 3C). The distribution of Bx genotypes also exhibited a wide distribution, with the Tajik accordingly showing higher numbers in total Bx genotypes, and specifically of Bx2, Bx4, Bx5, Bx6, Bx72 and Bx73 (Figure 3A, 3D). Furthermore, two genotypes previously unreported in East Asians (new1, new2) were identified in two Tajik individuals. One of these genotypes (new1) lacked the entire block from KIR2DL1 to KIR3DL1/S1 and KIR2DS3 to KIR2DS4, spanning from centromeric to telomeric KIR regions (Figure 3A). For further analysis, it was observed that its 2DS5 might be located in the centromeric segment, causing the absence of the entire block from 2DL1 through to 2DS1. The other genotype (new2) lacked KIR2DL3, KIR2DP1 to KIR2DL4, and KIR2DS3. They both lack the KIR2DL4 and KIR3DP1 framework genes, which is rare.³⁷ For Tajik, 24 genotypes including 85.57% individuals carried all four genes encoding HLA class I-specific inhibitory KIR (KIR2DL1, KIR2DL2/3, KIR3DL1 and KIR3DL2). In Yugur there were 23 and She, 22 distinct genotypes that carried all four of these genes, accounting for 93% and 97.1% individuals, respectively (Figure 3A).

Figure 3. Distribution of genotypes derived from the four populations.

A. At the top is shown the gene organization of the KIR locus. Underneath is shown is the comparison of genotypes derived from the four populations. Genotype ID is assigned according to allele frequencies net database, based on the presence or absence of KIR genes. The observed numbers and frequencies are shown on the right. The colours of the boxes indicate the gene copy numbers, as given in the key at the bottom. The frequencies (%) of the corresponding genotypes are listed at the right.

B-C. Shown are the comparisons of KIR-AA genotypes (panel B), Cen-AA and Tel-AA genotypes (panel C) derived from the four populations. The x axis represents the carrier frequencies.

D. Shown are the carrier frequencies of KIR-Bx2, Bx3, Bx4, Bx5, Bx6, Bx7, Bx8, Bx9, Bx72, and Bx73 genotypes in the four populations. The coloured circles represent the populations, as shown at the bottom. The Y axis represents the carrier frequencies.

The diversity of KIR alleles Among Ethnic Groups in China

The KIR alleles were determined with high resolution at the five-digit level. The data for all alleles are shown in Supplementary Table 4. The distribution of the alleles for each KIR gene was consistent with Hardy-Weinberg equilibrium. Among the three populations, the Tajik showed the highest number of alleles (n=157), there were 131 alleles detected in Yugur, and the least number of alleles were identified in She (n=108). In the Tajik, we detected 115 alleles of inhibitory KIR and 21 alleles of activating KIR, with 91 and 21 alleles in Yugur, and 73 and 20 alleles in She, respectively. Thus, like other populations studied worldwide,^{30, 41, 44, 45, 54} inhibitory KIR are more polymorphic than activating KIR. KIR3DL3 was the most polymorphic studied with 38, 25, and 21 alleles observed in Tajik, Yugur, and She, respectively. Noteworthy were the higher frequencies in Tajik than the other three populations, of characteristic Cen-B haplotype alleles, such as KIR2DL2*00101, KIR2DS2*00101, KIR2DL5B*00201, KIR2DL5B*00801, KIR2DS3*00103 (Figure 4). Similarly, the frequency spectra of telomeric KIR genes in Tajik were distinct from the other populations, where KIR2DL4*00801, KIR3DL1*00101, KIR2DS4*00301, and KIR3DL2*00101 were the most frequent alleles observed. Again, this distribution of alleles is similar to that of Iranian populations.⁵⁴ The most frequent telomeric KIR alleles in the other three Chinese populations were KIR2DL4*00102, KIR3DL1*01502, KIR2DS4*00101, and KIR3DL2*00201 and these were observed at significantly lower frequency in Tajik (Supplementary Table 5). These findings suggest a greater distinction in the telomeric than centromeric KIR between the Tajik and the other populations analyzed. The distribution of alleles among the Zhe, She, and Yugur populations did not show significant difference except KIR2DL4*01101, KIR3DL1*00501, KIR3DL2*00201, KIR3DL3*00301, and KIR2DS4*010, which exhibited much higher frequency and significantly lower KIR3DL2*00701 in She than Zhe or Yugur (Supplementary Table 5).

Figure 4. Comparison of KIR alleles discovered in the four populations.

Shown is the comparison of KIR alleles discovered in this study, only alleles with frequency greater than 4% in at least one population are shown here. Colour key at the bottom shows the populations studied.

Diverse Distribution of KIR Haplotypes

Haplotypes were assigned per individual according to the copy number, alleles of each KIR gene (except KIR3DL3 and KIR3DP1) and the expectation-maximization algorithm. The centromeric and telomeric haplotypes were analyzed separately, which showed 29 and 37, 20 and 36, 35 and 44 motif haplotypes in Yugur, She, and Tajik, respectively (Supplementary Table 6). Only four of these centromeric and seven telomeric haplotypes were common to Yugur, She, and Tajik populations. Of those, three shared centromeric haplotypes (C1, C2, and C9, Figure 5A) correspond to a cumulative frequency of 83.5%, 84.6%, and 69.1% in Yugur, She, and Tajik, respectively. For the 7 shared ones in the telomeric KIR region, they comprise a cumulative frequency of 78%, 68.7%, and 57.2% in Yugur, She, and Tajik, respectively. All these common telomeric haplotypes displayed greater variation in Tajik than those of the other three populations. Five of the 7 shared telomeric haplotypes were all tA01 by gene-content and differed only in their alleles. The most frequent telomeric haplotype was KIR2DL4*00102–3DL1*01502–2DS4*00101–3DL2*00201, with frequency of only 8.8% in Tajik but around 40–45.45% in other populations (Figure 5A). However, the frequencies of KIR2DL4*00801–3DL1*00101–2DS4*00301–3DL2*00101 and KIR2DL4*00501–3DS1*01301–2DS5*00201–2DL5A*00101–2DS1*00201–3DL2*00101 reached 19.6% and 18.% in Tajik but was less than 5% and 10% in the other three Chinese populations (Figure 5A). The other shared haplotype KIR2DS1*00201–3DL2*00701, which is a tB01 deletion type that lacks the entire block of KIR3DP1–2DL4–3DS1–2DS5–2DL5B, showed a frequency as high as 6% and 6.2% in Yugur and Tajik. This haplotype was also identified as frequent in Europeans,^{36, 55} but less than 1.5% in the She and Zhe populations. In summary, we found that the KIR-B haplotype had a higher allelic diversity than the KIR-A haplotype motifs in both the centromeric and telomeric segments of the KIR locus. However, the KIR-B haplotype was less frequent than the KIR-A haplotype in both segments, though it was slightly higher in Tajik than the other three populations (Figure 5B).

Figure 5. KIR haplotype diversity in the four populations.

A. Shown are the centromeric and telomeric KIR haplotypes and their frequencies identified in the four populations, only haplotypes with frequency greater than 2% in at least one population are shown here. KIR-A haplotypes are shaded red and KIR-B haplotypes are blue.

B. Shown is the distribution of centromeric (Cen) and telomeric (Tel) KIR-A and B haplotypes in the four populations. KIR-A haplotypes are shaded red and KIR-B haplotypes are blue.

C. Shown are the structural variant haplotypes identified. Color-coding is used to indicate the ‘deletion’ or ‘insertion’. Grey shading indicates the additional segment in relation to more frequent haplotypes, as described above each with ‘deletion’. Purple shading indicates ‘insertion’ structural.

More than 90% of haplotypes can be classified into one of three centromeric (cA01, cB02, and cB01) and two telomeric (tA01 and tB01) gene-content motifs (Supplementary Table 7). Additionally, the remaining observed haplotypes are characterized by insertions or deletions. Two Tajik individuals were identified with a centromeric haplotype, which has inserted KIR2DS3/5c-2DL5B-2DP1–2DL1 compared with cA01. The other five were all deletion types, with one (KIR2DP1–2DL1) occurring with high frequency in Yugur and Tajik populations, suggesting that it is endemic. Regarding the telomeric haplotypes, apart from the tB01-based haplotype featuring 3DP1–2DL4–3DS1–2DS5–2DL5B, which was identified in all populations, the KIR3DP1–2DL4–3DL1/S1 insertion was also common, being found in all three populations here and widely reported previously.^{23, 32, 44, 55}

Interestingly, two Yugur individuals were found to have deleted the entire telomeric region from KIR3DP1 to KIR3DL2, which is a rare occurrence that has not been reported before in East Asian populations. We characterized to high resolution 43 haplotypes showing large-scale insertion or deletion events, which corresponded to 12.9% of individuals (Figure 5C). The majority of the deletion haplotypes had deleted a large block from KIR2DP1 to KIR2DS4, corresponding to 12 haplotypes (S1-S11, S16, Figure 5C) and 25 individuals in total. One haplotype (S16, Figure 5C) from a She individual had 5 genes remaining after deletion, and one haplotype (S15, Figure 5C) from two Yugur individuals only had 3 genes remaining due to the deletion of the whole telomeric region and a large part of the centromeric region. For those with haplotypes showing duplications, most were insertions of KIR3DP1–2DL4–3DL1/S1 or longer blocks of KIR2DS3/5c-2DL5B-2DP1–2DL1–3DP1–2DL4–3DL1/S1, but one individual (S17, Figure 5C) was identified with a KIR2DS4 duplication which has not been previously reported. From both the insertion and deletion haplotypes, we can deduce that haplotype exchanges are common during evolution, caused by recombination, especially between the centromeric and telomeric regions.

Distinct HLA distribution and inhibitory KIR/HLA interactions.

HLA genotyping revealed an extensive array of alleles, totaling 76 (19 for HLA-A, 37 for HLA-B, 20 for HLA-C), within the Tajik population. Similarly, the Yugur population exhibited 81 alleles (21 for HLA-A, 37 for HLA-B, 23 for HLA-C), while the She population exhibited 55 alleles (15 for HLA-A, 23 for HLA-B, 17 for HLA-C) (Supplementary Table 8). The frequency distributions of all HLA genes conformed to the principles of Hardy-Weinberg equilibrium (Supplementary Table 9). Despite the She population showing the least polymorphism, it stood out as the most distinctive among the studied populations. Notably, the most prevalent alleles in She for HLA-A, -B, and -C respectively were A*02:03 (22.6%), B*46:01 (19.6%), and C*07:02 (27.4%), all of which demonstrated significantly higher frequencies compared to both Yugur and Tajik (Figure 6A, Supplementary Table 10). Furthermore, B*38:02, B*40:01, and C*01:02 also exhibited significantly higher frequencies in the She population in contrast to the Yugur and Tajik populations. Conversely, the frequency of allele C*06:02, relatively more frequent in the Yugur and Tajik populations, was significantly lower in the She population.

Figure 6. Diversity of HLA alleles in the four populations.

A. Shown are the HLA-A, -B, and -C alleles identified in this study and a comparison across the four populations. The colour key located at the top of the figure indicates the respective populations under investigation. Additionally, different coloured stars are used to denote significant statistical differences observed among these populations. Red star indicates a significant difference between the Zhe and Tajik populations; black star - between the Zhe and Yugur populations; orange star - between the Zhe and She populations; green star - between the Yugur and Tajik populations; blue star - between the She and Tajik populations; purple star - between the She and Yugur populations.

B. Shown is the cluster analysis depicting HLA allele frequencies across 17 different populations.

In terms of Yugur and Tajik populations, their allele distributions largely aligned, diverging only in B*52:01, which displayed a significantly higher frequency in the Tajik population compared to the Yugur population (Figure 6A, Supplementary Table 10). Interestingly, despite their geographical proximity, the She and Zhe populations exhibited noteworthy distinctions in their allele profiles with a total of 9 alleles showing significant differences. Among these, three alleles (A*02:03, A*11:02, B*38:02) exhibited significantly higher frequencies in the She population, while six alleles (A*02:01, A*30:01, A*31:01, A*68:01, C*06:02, and C*08:01) demonstrated significantly lower frequencies or absent in the She population (Figure 6A, Supplementary Table 10).

Initially, we found that the She population differed significantly from other populations due to its notable allele differences. However, cluster analysis of 13 other populations of African,^{41, 46} European,^{42, 48, 56} Asian,^{44, 45, 49, 50, 51} Oceanian⁵² or South American origin⁵³ revealed that the She and Zhe populations still shared a close relationship with the Japanese⁴⁵ and Malasia Chinese population⁴⁴ (Figure 6B). On the other hand, similarly to KIR cluster analysis, the Tajik population exhibited a closer genetic affinity to Iranian,⁵⁴ indicating a common ancestral link with Europeans, including Italians⁵⁶ and Belgians⁵⁷. Moreover, Yugur showed a close distance to Tibetans⁵⁰, and was close to the Southeast Asian population, Thai⁵¹ and Malays⁴⁴.

Highlighting the significant variation in HLA alleles across populations, the HLA haplotypes also exhibited considerable diversity. We identified a total of 77, 103, and 94 distinct HLA-A-B-C haplotypes among the She, Yugur, and Tajik populations, respectively (Supplementary Table 11). Interestingly, only three haplotypes were shared among these populations, and these were among the most frequent haplotypes overall, as illustrated in Figure 7A. The linkage disequilibrium(LD) for each pair of HLA loci was also analyzed and listed in Supplementary Table 12, and their haplotype frequencies were compared only when two loci were shown as LD. The difference was found to be remarkable (Supplementary Table 13), for example, She population had 9 haplotypes with frequency higher than 10%, however, Yugur had only one, and Tajiks didn’t have haplotypes greater than 10%.

Figure 7. Distribution of HLA and KIR-HLA Interactions Across Four Populations.

(A) Shown are the ten the ten most prevalent HLA class I haplotypes observed within the four populations. Colour shading highlights HLA class I alleles that encode KIR ligands, as detailed in panel B. (B) Shown are the frequency spectra of HLA-A, -B, and -C allotypes across the four populations. Coloured segments represent the allotypes that serve as KIR ligands: Yellow-A3/11 (HLA-A3 and -A11); Green-Bw4; Red-C1; Blue-C2; Gray-allotypes that are not KIR ligands. The corresponding number and frequency of allotypes acting as KIR ligands are listed in the frame below. (C) Shown are the combined frequencies of HLA class I haplotypes encoding one (blue), two (orange), or three (green) KIR ligands. (D) Shown is the mean number of distinct interactions per individual observed for each of the inhibitory KIR that interact with polymorphic HLA class I. The values are presented for eight representative populations: Tajik, Yugur, She, Zhe, European, Malay, Shenzhen Han, Japanese.

The She population exhibited the least amount of polymorphism, with the cumulative frequency of the top ten haplotypes reaching 57.6%. Notably, three haplotypes displayed frequencies exceeding 9%, particularly A*02:03-B*38:02-C*07:02, which reached a frequency of 13.3% (Figure 7A). Interestingly, this specific haplotype was absent in the Yugur and Tajik populations and appeared infrequently in the Zhe population. In contrast, the combined frequency of the top ten haplotypes ranged from 27% to 35% in the remaining three populations. Moreover, the most prevalent haplotypes in these populations also varied, with frequencies falling below 7%.

While the She population displayed the lowest HLA-class I allele and haplotype polymorphism, it is noteworthy that both She and Tajik populations exhibited a higher count of allotypes that function as ligands for KIR, in comparison to Yugur and Zhe populations (Figure 7B). Notably, the Tajik population demonstrated an intriguing pattern, where over half of the HLA-A and HLA-B allotypes are KIR ligands. Within the KIR ligands encoded by HLA-B, Bw4 was the prevalent subtype, whereas the presence of C1 encoded by HLA-B was relatively rare (Figure 7B). In contrast, the She population stood out for encoding a substantial number of C1 ligands derived from HLA-B, primarily due to the elevated frequency of B*46:01 within this population. Consistent with other previously studied Asian populations,^{27, 50} the occurrence of C1 ligands encoded by HLA-C remained relatively low, accounting for 10.7% and 17.6% in the She and Zhe populations respectively. Notably, this frequency was higher in the Yugur and Tajik populations, exceeding 35% in both populations.

The mean number of inhibitory KIR-HLA pairs per individual was analyzed and compared with other Asian populations, including the Japanese, Malaysians, Shenzhen Han Chinese, as well as other populations of African,⁴¹ European,⁴² Asian,^{30, 44, 45} Oceanian⁵² or Amerindian⁵³ (Figure 7C–7D). The analysis revealed notable findings. Specifically, the mean number of interacting KIR2DL1-C2 and KIR2DL2-C1/C2 pairs was higher in the Tajik population. This finding was similar to Europeans⁴² or Ghana⁴¹ from Africa, which could be attributed to the elevated presence of KIR2DL2 in Tajik individuals. Conversely, the She population exhibited a remarkably higher incidence of KIR2DL3-C1 interactions per individual, reaching 2.2. This value surpassed that of all other populations. This heightened interaction is attributed to the increased prevalence of HLA-C1, particularly that provided by HLA-B*46:01 (Figure 7A).

Discussion

Our study is the first to apply the NGS method to describe high-resolution (5-digit) KIR genotypes, alleles, and haplotypes for all functional KIR genes in three ethnic minority populations from China. These minorities are Yurgu, Tajik, and She populations, who are geographically distant from each other. Tajiks mainly live in the Taxkorgan Autonomous County of Xinjiang Uygur Autonomous Region, China and in 2021, the population of Tajiks was 50,896.³³ Historically, Tajiks who have lived in vast areas of Xinjiang since ancient times, and Tajiks who migrated from the west of Pamirs to the east and settled in Tashkurgan at different times, are the ancestors of Chinese Tajiks. The ethnic origin of the Tajik ethnic minority can be traced back to the Iranian tribes distributed in the eastern part of the Pamir Plateau several centuries BC. ⁷⁰ Yugur people are mainly engaged in animal husbandry and the population of Yugur ethnic minority in China was 14,706 in 2021.³³ They mainly live in the Sunan Yugur Autonomous County. The Yugur originated from the Uighur nomads in the Orhun River valley in the Tang Dynasty. Referring to She ethnic minority, more than 90% of them live in the vast mountain areas of Fujian and Zhejiang. They reside mainly in the Jingning She autonomous county, and the total population was 746,385 in 2021.³³ There are conflicting opinions about the origin of She ethnic minority. However, at the beginning of the 7^th century in the Sui and Tang dynasties, the She ethnic minority lived in southern Fujian, Chaoshan and other places at the junction of Fujian, Guangdong and Jiangxi provinces (https://www.gov.cn/test/2006-04/14/content_254231.htm). We have previously reported the distribution of KIR and HLA in Zhejiang Han population, however, the distribution of KIR and HLA for Chinese ethnic minorities was unknown. Analyzing these ethnic minority populations provides a deep insight into the alleles and haplotypes as well as new variation present in these groups, which will extend genetic resources for the studying the role of KIR and HLA in multiple immune-mediated diseases.

The Tajik population we studied showed a relatively high frequency of KIR-B haplotypes. Genetic distance analysis revealed that the Tajik group is particularly close to Iranians, and distinct from the other three Chinese populations studied. This finding is consistent with the origin of the Tajik people being traced back to Iranian-speaking tribes in the eastern Pamirs several centuries BC. These tribes were distributed in multiple places south of the Tianshan Mountains in China since ancient times, suggesting that Tajik and Iranians may have descended from the same ancestral population. Previous studies have suggested that the KIR-B haplotype may be undergoing strong diversification selection, which could be related to a variety of factors, such as reproduction, the carrying of unfavorable genes associated with autoimmune risk, or the generation or loss of new functional genes to combat infection.^{12, 58} Therefore, the high KIR-B haplotype of Tajiks may be the result of natural selection due to their unique plateau living environment, distinct from other populations in China, as well as their distinct ancestry. In addition to the high frequency of KIR-B haplotypes, the most frequent telomeric KIR-A haplotype in Tajik was KIR2DL4*00801–3DL1*00101–2DS4*00301–3DL2*00101 (19.58%), which only expresses inhibitory KIR, indicating that the inhibition signal plays a dominant role in Tajik individuals carrying Tel-A haplotypes. KIR2DL4*00801 does not express a receptor on NK cells due to a single nucleotide deletion at position 811 in Exon 7, and KIR2DS4*00301 expresses a non-functional KIR2DS4 allotype.⁵⁹ In addition to the similarities to Iranians, Tajiks differ from other Chinese in the frequency of KIR3DL1*002. KIR3DL1*002 is an allele highly expressed on the surface of NK cells.⁴⁰ It is widely distributed in European²⁹ but rare in Chinese and other Asians.^{27, 45, 50} Here, we did not find KIR3DL1*002 in Tajik populations, suggesting that KIR3DL1*002 has been specifically targeted by negative selection in Tajik. By contrast, the She and Zhe population showed similar allele frequency spectra, which is also consistent with the She living in the mountain areas of Zhejiang province since 1,200 years ago. However, the Yugur also showed similarities to Tibetan or Southeast Asian populations.

The KIR haplotypes exhibited a high degree of variation in each population. Structural patterns caused by frequent duplication and large deletions were identified, accounting for 12.9% of individuals across the three populations. These haplotypes tended to be characterized by KIR3DP1 and KIR2DL4, being duplicated on some haplotypes and deleted from others. Copy number analysis also indicated that 14% Yugur and 11.34% Tajik individuals in the cohort showed only one copy of KIR3DP1/KIR2DL4. Two novel genotypes were identified in two Tajik individuals, who were absent of KIR3DP1 and KIR2DL4 genes. These findings are consistent with strong linkage between KIR3DP1 and KIR2DL4, and a possible recombination hotspot.^{12, 32} KIR3DP1 is a pseudogene, but KIR2DL4 has been considered to play a special role in the NK-mediated control of the maternal/fetal interface during pregnancy,^{22, 23, 60} and its lack in maternal NK cells has been reported to be compatible with successful pregnancy.^{22, 23, 60} The relatively high occurrence of KIR2DL4 deletion in these two populations may be related to the natural selection of successful pregnancy in high altitude environments, which needs to be further investigated. In addition to deletion of KIR3DP1 and KIR2DL4, all deletion type haplotypes identified in the study showed large deletions from centromeric to telomeric, and some even lacked the whole telomere motif. This variation has the potential to provide insight into the different migration processes of these populations and the functional properties of the KIR in NK-mediated immunity.

The cluster analysis of KIR genes revealed that Tajik populations are genetically close to the Iranian population and European populations. When considering the distribution of HLA alleles, the She population appeared notably different. However, the cluster analysis of HLA alleles yielded similar results. It indicated that the She and Zhe populations share a common genetic ancestry, despite variations in allele distribution. Additionally, the Tajik showed genetic affinities with Western Asian populations Iranians in both KIR and HLA cluster analysis,⁶¹ aligning with their historical connections. However, the Yugur population clustered within the same branch as the Tibetan population, a pattern also consistent with their historical interactions.

Notably, the HLA-B*73:01 allele, previously associated with Western Asian populations,³⁷ was identified in the Yugur population, confirming their genetic links with Western Asia. Some alleles in the She population exhibited significant differences from the Zhe population, despite their geographical proximity. Examples include HLA-A*02:03, HLA-B*38:02, and HLA-B*46:01. This divergence might be attributed to the She population’s isolated, mountainous habitat and their tendency to intermarry and reproduce within their community, resulting in the enrichment of certain alleles. The higher prevalence of HLA-B*46:01 in the She population corresponds with an increase in HLA-C1. Consequently, we observed a higher frequency of 2DL3-C1 interactions in the She population. Although the 2DL3-C1 interaction is generally reported to be weak,^{4, 12, 26} it has been associated with better prognosis during Hepatitis C virus (HCV) infections,⁶² as it triggers activating signals in NK cells during viral infections. The pronounced presence of the 2DL3-C1 interaction in the She population may suggest localized selection favoring protection against HCV or similar infections, a topic deserving further study.

In contrast, the interaction between 2DL1 and C2 was found to be more common in the Yugur and Tajik populations, while the 2DL2-C1/2 interaction exhibited higher frequency in the Tajik population. It is important to note that both of these interactions are known to be strong. These observed variations in KIR interactions highlight the unique genetic profiles of the Yugur and Tajik populations. The prevalence of 2DL1-C2 and 2DL2-C1/2 interactions suggests distinct immune response mechanisms in these groups, which may have implications for their susceptibility to certain diseases or their ability to mount effective immune responses against specific pathogens. Further research is needed to explore the functional consequences of these strong KIR-HLA interactions in the context of immune responses and disease outcomes among the Yugur and Tajik populations. Understanding these interactions can provide valuable insights into the genetic basis of immune-related diseases and contribute to personalized medicine approaches for these populations.

In summary, our study comprehensively explored NK cell diversity attributable to the polymorphism of the KIR and HLA genes, including copy number, alleles, haplotypes, and KIR-HLA interactions, within three distinct Chinese ethnic minority populations. The distribution of KIR and HLA in the Tajik population closely resembles that of the Iranian population, indicating shared genetic characteristics, which supports the previously reported the origin of the Tajik. In contrast, the Yugur population shares similarities with Tibetans in both KIR and HLA gene distributions, indicating affinities with populations from Southeast Asia and East Asia. Meanwhile, the She population clusters with East Asian populations, highlighting its genetic connections within this regional group. These findings provide a comprehensive understanding of the genetic diversity and historical relationships among these ethnic minority populations in China, illuminating their unique genetic profiles and potential implications for health and disease.

Supplementary Material

Supinfo

Supplementary Table 1. The distribution of KIR genes in the four populations

Supplementary Table 2. The statistical comparison of the frequencies of KIR genes among the four populations.

Supplementary Table 3. The distribution of KIR copy numbers in the four populations.

Supplementary Table 4. Distribution of KIR alleles in She, Tajik, and Yugur populations.

Supplementary Table 5. The statistical comparison of the frequencies of KIR alleles among the four populations.

Supplementary Table 6. Distribution of centromeric (A) and telomeric (B) KIR haplotypes in the She, Tajik, and Yugur populations.

Supplementary Table 7. The distribution of KIR haplotypes by gene-content in the four populations.

Supplementary Table 8. The distribution of HLA alleles in the four populations.

Supplementary Table 9. HWE test (p-values) for each HLA locus.

Supplementary Table 10. The comparison of HLA alleles among the four populations.

Supplementary Table 11. Distribution of HLA haplotypes in She, Tajik, and Yugur populations.

Supplementary Table 12. Table of Chi-square P values of Linkage disequilibrium test between each pair of HLA loci.

Supplementary Table 13. Distribution and comparison of HLA haplotype at each pair of HLA loci only when two loci were shown as LD in the three ethnic minorities.

Data Availability Statement

The data that support the findings of this study are freely available in the Supplementary tables.