176 blood samples were collected from healthy blood donors in the Blood Center of Zhejiang province, China. The ethnic background of all individuals is Zhejiang Han. Informed consent was obtained from all participants. The Zhejiang Han are a population of approximately 60 million individuals living on the Southeast coast of China and are a subgroup of the Chinese Southern Han ethnic group. This study was approved by the regional ethics committee, Blood Center of Zhejiang Province. Genomic DNA was extracted using MagNA Pure LC DNA Isolation Kits (Roche Diagnostics, Indianapolis, IN, USA) according to the manufacturer’s instructions. The final DNA concentration was adjusted to 60 ng/μl and optical density at 260/280 was approximately 1.8.

A biotinylated DNA probe-based capture method was applied to sequence KIR and HLA genes to high-resolution, as described previously (33). Briefly, 500 ng genomic DNA for each sample was randomly fragmented using enzyme digestion (New England Biolabs, Boston, MA) and then labeled with TruSeq DNA CD indexes (Illumina Inc, San Diego, CA). Fragments of 800–1,200 bp length were obtained by size selection and the individual samples pooled at equal quantity into one tube. Fragments containing target HLA and KIR genes were captured using specific probes, prepared to a final concentration of 12 pmol/L, as determined by Qubit instrument, then sequenced using a MiSeq instrument and a v3 Reagent Kit (600-cycle; Illumina Inc, San Diego, CA, USA).

KIR gene content, copy numbers, and allele genotypes were determined using the Pushing Immunogenetics to the Next Generation pipeline (PING) as previously reported (33). The copy number was determined from the ratio of reads mapping to each KIR gene to those mapping to KIR3DL3, a reference gene present only one copy on each haplotype. Alleles for 13 KIR genes were analyzed, of which KIR2DL2 and KIR2DL3 are alleles at the same locus (KIR2DL2/3), as are KIR2DS3 and KIR2DS5 (KIR2DS3/5), and KIR3DL1 and KIR3DS1 (KIR3DL1/S1). KIR3DL3 alleles were determined using PHASE 2.1 (38). The HLA alleles were determined using TypeStream Visual NGS Analysis Software (TSV2.0, One Lambda Inc, Canoga Park, CA). The presence of KIR ligands, A3/11, Bw4, and C1/C2 was determined according to the HLA alleles.

KIR genotypes were named based on the presence or absence of KIR genes according to the allelefrequencies.net database (39). Carrier frequencies of KIR genes and genotypes were calculated as their percentage of the total number 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, and absence was counted as a distinct allele). The composition and frequencies of HLA haplotypes were determined using Arlequin 3.5.2.2 (40).

Interactions between inhibitory KIR and HLA ligands were analyzed. The number of distinct allotype interactions was counted for each KIR for every individual and the mean number was calculated per population. Broadly, KIR2DL1 recognizes C2⁺HLA-C, and KIR2DL2/3 recognizes C1⁺HLA-C, and some C2⁺HLA-C. KIR3DL1 recognizes the Bw4 motif at residues 77–80 of some HLA-A and HLA-B allotypes. KIR3DL2 recognizes the A3/11 motif carried by HLA-A*03 and HLA-A*11 allotypes (41, 42). Major exceptions are KIR2DL1*022, which interacts with C1⁺HLA, and HLA-B*46 and B*73, which carry C1 motifs (20, 26).

Hardy–Weinberg equilibrium (HWE) was determined for each gene using Fisher’s exact test implemented in the Arlequin software 3.5.2.2. The distributions of genes, alleles, haplotypes, and KIR/HLA interactions in the Zhejiang Han population were compared to other populations as described. Differences between populations were assessed using GraphPad software, by means of the χ² test for categorical variables. The p values were calculated using Fisher’s exact test (pf), applied when appropriate. Bonferroni correction for multiple comparisons was applied. p <0.05 was regarded as significant.

To analyze the KIR locus of the Zhejiang Han at high resolution, we first established which KIR genes are present, and in how many copies, for every individual. The four framework genes [KIR3DL3, KIR2DL4, KIR3DP1 and KIR3DL2 (15)] were detected in all 176 individuals of the cohort ( Figure 1A ). KIR2DL4 and KIR3DP1 were each observed with one to three copies and one individual in the cohort has only one copy of KIR3DL2 ( Figure 1A ). Only four individuals have less than two copies of any of these framework genes, six individuals harbor three copies, and none were observed with four copies. These observations indicate that, although structural diversity of the KIR locus is present, it is rare in the Zhejiang Han population. KIR2DL1, KIR2DL3, KIR3DL1, KIR2DS4, and KIR2DP1 were all observed at a frequency greater than 90% and at two copies in at least 64.8% of the population ( Figures 1A, B ). These five genes are not present on every KIR haplotype, and when they occur together, they characterize the KIR A haplotype (15, 51). KIR2DL3, which occurs in the centromeric KIR region, showed higher frequency than KIR3DL1 and KIR2DS4, which are located in the telomeric KIR region (81.8% vs 64.8%), indicating a higher frequency of centromeric KIR-A haplotype motifs (Cen-A) than telomeric KIR-A motifs (Tel-A). Accordingly, the genes characteristic of KIR-B haplotypes have frequencies less than 25% in centromeric and approximately 35% in telomeric KIR region ( Figure 1B ). We compared the distribution of KIR genes of Zhejiang Han with populations representing Asia (37, 43, 46, 52–54), Europe (47, 48, 55, 56), Africa (49, 57), America (50) and Oceania (58, 59) ( Supplementary Table 1 ). As a control, this analysis showed the KIR gene frequencies to be similar to our previous study of an independent Zhejiang Han cohort that was performed at lower resolution (60). The KIR gene frequencies are also similar to those of other populations from East Asia including Japan (37), South Korea (43) and Singapore (53), but distinct from Southeast Asians (46) and populations from other continents ( Supplementary Table 1 ). Accordingly, the frequencies of Cen-B specific KIR genes (KIR2DS2, KIR2DL2, KIR2DL5B) in the Zhejiang Han are significantly lower than all populations outside East Asia ( Supplementary Table 1 ).

As determined by the copy number of each of the 13 KIR genes, twenty-three distinct genotypes were observed ( Figure 1C ). Of these genotypes, KIR-AA is the most frequent (53.4%) in Zhejiang Han ( Figure 1C ). The KIR-AA genotype is common across East Asia, also being observed frequently in other Chinese Southern Han (36), Taiwan Chinese (44), South Korean (43), and Japanese (37). The KIR-AA genotype is less frequent outside East Asia, as illustrated by populations representing Southeast Asia, Europe, South America, and sub-Saharan Africa ( Figure 1D ). The other 22 KIR genotypes observed in Zhejiang Han are designated KIR-Bx, the distributions of which are also similar among East Asians and distinct from populations of other continents ( Figure 1D ). For example, Bx2, which is the most frequent KIR-B genotype in Zhejiang Han, was not observed in sub-Saharan Africans, whereas Bx5 is frequent in sub-Saharan Africans, but very rare in Zhejiang Han ( Figure 1D ). Another example is Bx3, which was observed with high frequency in Amerindians, but almost absent from all other populations ( Figure 1D ). These examples of distinct genotype profiles characterizing major human population groups illustrate the rapid evolution of the KIR locus (27, 61).

A total of 107 KIR alleles were observed in the Zhejiang Han cohort ( Figure 2A and Supplementary Table 2 ) and no novel alleles were identified. The allele frequency distributions for all the KIR genes were in accordance with Hardy–Weinberg equilibrium. The inhibitory KIR are more diverse and polymorphic (88 alleles in total) than the activating KIR (14 alleles). The most diverse gene encoding an activating KIR is KIR2DS4, which is the only activating KIR carried by the KIR-A haplotype, with five alleles observed ( Figure 2B ). KIR2DS4*00101 is the most frequent allele and is the only one that can encode a functional allotype on the cell surface, whereas the remainder are non-expressed alleles because of a 22bp deletion in exon 5 (25). Other activating KIR are encoded by KIR-B haplotypes and show little diversity. KIR2DS3/5 can be present on either a Cen-B or Tel-B haplotype; two KIR2DS3 alleles were observed, one is KIR2DS3*00103 observed in Cen-B, and the other is KIR2DS3*00201 observed in Tel-B. Although the activating KIR are less polymorphic, their allele frequency distributions do vary across populations. For example in Zhejiang Han, functional KIR2DS4 allotypes are the most frequent, whereas in some populations, such as in Oceania (62), the non-expressed KIR2DS4 alleles predominate.

Among genes encoding inhibitory KIR, those with three domains show greatest variability (25, 32). We observed 23 alleles each of KIR3DL3 and KIR3DL2 and 11 alleles of KIR3DL1/S1, which encode 16, 17, and 11 allotypes, respectively ( Supplementary Table 2 ). Thus, each allele of KIR3DL1 we observed here ( Figure 2C ) encodes a distinct allotype and the most abundant have high-expressing, high-binding, phenotypes (37, 63, 64). KIR3DL1*015 is the most frequent KIR3DL1 allotype in Zhejiang Han. Comparison with other representative populations shows that KIR3DL1*015 is present worldwide, reaching highest frequency in East Asians ( Figure 3 ). By contrast, KIR3DL1*004, which is a poorly expressed phenotype found in Europeans and Africans, was not observed in Zhejiang Han ( Figure 3 ) and neither was it observed in another Chinese Southern Han group (36). We previously reported KIR3DL1 allele frequencies from an independent cohort of Zhejiang Han (66). Because KIR3DS1 was not analyzed in that cohort, the reported KIR3DL1 allele frequencies are higher than in the present cohort, however, the frequencies are similar when KIR3DS1 is considered. The present analysis also showed that KIR3DL3 has the most extensive cross-population diversity. Here, KIR3DL3*010 clearly characterizes East Asian populations, whereas 3DL3*003 for example characterizes Amerindians, but is rare in East Asians ( Figure 3 ).

Although KIR2DL1 is the most polymorphic two-domain KIR, having 173 known alleles [IPD-IMGT/KIR Database version 2.10.0, December 2020 (32)], it has low diversity in Zhejiang Han, with five alleles observed ( Figure 2A ). The most prominent of these alleles is KIR2DL1*00302 at an allele frequency of 77.8% ( Supplementary Table 2 ), corresponding with a similar frequency of KIR2DL3*00101. This strong linkage disequilibrium between the two alleles is shared across multiple populations worldwide (34, 37, 45, 50, 62, 65). KIR2DL4, which can be an activating receptor specific for HLA-G (67, 68), has six alleles in Zhejiang Han ( Supplementary Table 2 ). Of 107 KIR2DL4 alleles, encoding 54 distinct polypeptide sequences, 18 do not express a receptor on the cell surface due to a single nucleotide deletion at position 811 in Exon 7 (69). The non-expressed alleles include KIR2DL4*008 and KIR2DL4*011, which were observed at a total frequency of 20.2% in Zhejiang Han ( Figure 2B ). The four expressible KIR2DL4 alleles observed thus constitute a majority. This majority varies little across human populations, the most extreme being Amerindians, where >95% KIR2DL4 alleles are expressible, and the exception being KhoeSan, where there is an approximately equal ratio between expressible and non-expressible forms of KIR2DL4 ( Figure 3 ).

Based on the copy number and allele information obtained from high resolution analysis, 54 centromeric haplotypes and 37 telomeric haplotypes were observed in Zhejiang Han ( Supplementary Table 3 ). As defined by gene content, three common centromeric haplotype groups, cA01, cB01, and cB02, and two telomeric groups, tA01 and tB01, were observed, as well as rarer forms having duplicated or deleted segments ( Figure 4 ). When accounting for allele diversity, there are 41 haplotypes in the cA01 group ( Supplementary Table 3A ), and 23 in the tA01 group ( Supplementary Table 3B ). The most frequent cA01 haplotype at the allele level in Zhejiang Han is KIR3DL3*01002-2DL3*00101-2DP1*00201-2DL1*00302, with a frequency of 19.03% ( Figure 5A ). The five most common cA01 haplotypes (64.5% in total) differ only in their KIR3DL3 alleles ( Figure 5A ). The high polymorphism of KIR3DL3 alleles diversifies the centromeric region, and the number of the centromeric haplotypes would decrease from 54 to 19 if KIR3DL3 alleles were excluded ( Supplementary Table 3A ). The most frequent tA01 haplotype at the allele level in Zhejiang Han is KIR2DL4*00102-3DL1*01502-2DS4*00101-3DL2*00201 ( Figure 5B ), with a frequency of 45.5%. Within the telomeric haplotypes, KIR2DS3/5t and KIR2DL5A alleles show strong linkage disequilibrium, in which KIR2DS5*00201 and KIR2DL5A*00101 are always found together, as are KIR2DS3*00201 and KIR2DL5A*00501. These patterns of linkage disequilibrium are the same as those found in other populations outside of Africa.

Six individuals having an insertion/duplication of KIR3DP1-2DL4-3DS1 were observed in Zhejiang Han ( Figure 5C ). The corresponding haplotypes comprised two distinguished at the gene content level, and three as determined by allele content. We also identified four individuals with haplotypes lacking KIR3DP1-2DL4-3DL1/S1 ( Figure 5D ), with three of these haplotypes (individuals 1, 2 and 4: Figure 5D ) distinguished only by their alleles of KIR3DL3. These haplotypes likely formed by uneven exchange during meiotic recombination, and have also been observed in other populations (17, 45, 70, 71). Although it remains to be confirmed through full haplotype sequence analysis, the insertions/duplications and the deletions occur in approximately the same location 5’ from KIR3DP1, suggesting the crossovers can occur frequently through shared mechanisms (72). In addition, we identified a unique haplotype that lacks any KIR genes in the telomeric region 3’ from KIR2DL4, leaving only six genes on the entire KIR haplotype (individual 5; Figure 5D ). That the allele is KIR2DL4*00102 suggests the parental haplotype that lost these genes was similar to the most frequent KIR-A in Zhejiang Han ( Figure 5B ).

Alleles for highly polymorphic HLA-class I (HLA-A, -B, -C) were obtained from high resolution sequencing. A total of 25 HLA-A alleles, 43 HLA-B alleles, and 27 HLA-C alleles were observed ( Supplementary Table 4 ), and the distributions for all the HLA genes complied with Hardy–Weinberg equilibrium. Only some HLA-class I allotypes act as ligands for KIR (9, 27). The two most frequent alleles of HLA-A, -B, -C respectively in Zhejiang Han are A*11:01 (25.3%) and A*24:02 (13.1%), B*40:01 (11.6%) and B*46:01 (10.5%), C*07:02 (18.2%) and C*01:02 (16.8%). All the allotypes they encode can interact with KIR, except for B*40:01. Every HLA-C allotype is a ligand for KIR, and in total, 44.1% of HLA-A and 48.6% of HLA-B allotypes by frequency in the Zhejiang Han can act as KIR ligands ( Figures 6A, B ). Most of the HLA-C allotypes (82.4%) contain a C1 motif (C1⁺HLA-C) in Zhejiang Han and HLA-B*46:01, which also carries a C1 motif was also observed at high frequency (10.5%). HLA-B*46:01 is one of only two HLA-B allotypes that carry a C1 motif (C1⁺HLA-B), the other is HLA-B*73:01, and both have geographical distributions focused in Asia, B*46:01 in Southeast Asia and B*73:01 in West Asia (39). C1⁺HLA-C and C1⁺HLA-B were both observed at higher frequencies than non-Asian populations ( Figure 6C ). Accordingly, HLA-C2 shows extremely low frequency in Zhejiang Han, and significantly lower than other populations ( Figure 6C , Supplementary Table 5 ). Moreover, A3/11, specifically encoded by A*11:01, was observed at higher frequency in Zhejiang Han than any other populations. We identified 166 HLA haplotypes ( Supplementary Table 4 ). Most of the observed haplotypes (77.3%) encode more than two KIR ligands, including A*11:01, which was detected in five of the ten most frequent haplotypes ( Figure 6D ). These findings are similar to those recently reported for Chinese Southern Han from Shenzhen (36).

The high frequency of the KIR-A haplotype, which encodes four inhibitory receptors able to interact with HLA class I ligands (KIR2DL3, KIR2DL1, KIR3DL1 and KIR3DL2), indicates inhibitory KIR-HLA interactions predominate in Zhejiang Han. Here, KIR2DL1*00302, KIR2DL3*00101, KIR3DL1*01502 and KIR3DL2*00201 are the major alleles carried by the KIR-A haplotype, and all the allotypes they encode have strong interactions with their HLA ligands compared to other allotypes (20, 37, 41, 73, 74). Because of the high frequency of KIR-A, Zhejiang Han are rich in KIR-AA genotypes, and each of these individuals therefore has eight strong inhibitory KIR allotypes that can interact with HLA. The mean number of inhibitory KIR-HLA pairs per individual was analyzed, revealing that the mean interaction of KIR2DL3-C1⁺HLA per individual (2.2) is higher than other populations ( Figure 6E ). This high frequency is caused by the high frequencies of KIR2DL3 and C1⁺HLA-C, and the extra C1⁺HLA-B allotype present ( Figure 6A ). By contrast, due to the low frequency of the C2⁺HLA-C allotypes, the frequency of interactions with C2⁺HLA is low in Zhejiang Han ( Figure 6E ). In these regards the Zhejiang Han also show very similar repertoires to the Shenzhen Chinese Southern Han (36). Further diversifying NK cell interactions with polymorphic HLA, conserved HLA-E can present peptides derived from HLA-A, C and some B allotypes. The peptides are derived from allotypes possessing methionine at the second residue of the leader sequence (-21M). HLA-E is recognized by inhibitory CD94:NKG2A expressed by NK cell and T cell subsets (75, 76). In Zhejiang Han, the total frequency of HLA-B alleles encoding -21M is 11.6% ( Figure 6F ). Among them, only HLA-B*38 also encodes a Bw4 ligand for KIR and is the most frequent at 3.7%.