Analysis of KIR and HLA polymorphism in Chinese individuals with COVID-19

Abstract

Killer-cell immunoglobulin-like receptor (KIR) interactions with HLA class I have crucial roles in modulating NK cell function in response to viral infections. To explore the correlation between KIR/HLA and susceptibility to SARS-CoV-2 infection, we analyzed polymorphism of KIR genes, haplotypes, HLA allotypes, and the interplay between KIR and HLA in individuals diagnosed with COVID-19. Compared to a population control group, we observed a significantly increased frequency of KIR3DL3*00802 in the COVID-19 group. When encoded by the HLA-B gene, the frequency of HLA-Bw4, a ligand for KIR3DL1, was at lower frequency in the COVID-19 group. Additionally, significantly elevated frequencies of KIR-Bx3, KIR3DL3*00301, 3DL3*048, and C1⁺HLA-C were identified in the COVID-19 group before multiple test correction, suggesting associations with susceptibility to SARS-CoV-2 infection. Our findings indicate that the KIR3DL3*00802 allele may be a high-risk factor for SARS-CoV-2 infection, while Bw4 encoded by HLA-B gene may confer protective effects against the infection.

INTRODUCTION

Novel coronavirus disease 2019 (COVID-19) is an acute respiratory tract infection caused by novel coronavirus (SARS CoV-2) (1). Since its first discovery in December 2019, it has spread rapidly, becoming a major threat to global public health security (2–3). Since the World Health Organization declared COVID-19 a global pandemic on March 11, 2020, it has become one of the most serious public health threats of this century. In the process of continuous self-replication, it is prone to frequently mutate, producing many variant strains, including Delta variant with strong virulence, and Omicron variant with strong transmission but relatively weaker virulence (4–5).

Natural Killer (NK) cells are essential components of the innate immune system and are crucial for effective antiviral immune responses against viruses like HIV, HBV, and HCV (6–9). Their function is regulated by killer cell immunoglobulin-like receptors (KIR), which interact with polymorphic HLA class I molecules (10).

KIRs are encoded by 15 highly polymorphic genes situated within the leukocyte receptor complex on chromosome 19q13.4 (11–12). Of these, KIR2DP1, and 3DP1 are pseudogenes. The KIR2DS1-2DS5 and 3DS1 genes encode activating receptors, KIR2DL1-3, 2DL5A/B and 3DL1-3 encode inhibitory receptors, and 2DL4 has dual function. The KIR genes are positioned in a head-to-tail manner, forming two distinct haplotype classes, termed KIR-A and KIR-B (13–14). The KIR-A haplotypes are characterized by genes encoding four inhibitory KIR (KIR2DL1, 2DL3, 3DL1, 3DL2), one activating KIR (KIR2DS4), and one bifunctional KIR (KIR2DL4) that interact specifically with HLA class I ligands: HLA-C1 group for 2DL3, HLA-C2 group for 2DL1, multiple HLA-C and A*11 for 2DS4, HLA-Bw4 group for 3DL1, HLA-A3/11 group for 3DL2, and HLA-G for 2DL4 (15–17). KIR-B haplotypes exhibit a greater variability in the number of genes than KIR-A, containing 2DL2 and additional genes encoding activating KIR specific for HLA-class I: HLA-C1 group for 2DL2, HLA-C2 group for 2DS1, and HLA-C1 for 2DS2. KIR-A and -B haplotypes both also encode inhibitory 3DL3, which binds B7 family member B7H7(18).

Being the most polymorphic regions in the human genome, KIR and HLA have been reported to relate to many diseases. Our previous study reported that the HLA-C*07:29 and B*15:27 alleles may be associated with the occurrence of COVID-19, with an odds ratio of 130.2 and 3.59 for SARS CoV-2 infection (19). However, the data on the influence of KIR and KIR-HLA interaction to COVID-19 are limited. A recent study showed KIR2DS4 to be associated with development of severe COVID-19 (20), consistent with functional data implicating this activating KIR (21–22).

In this study, we explored the relationship between KIR genes and their corresponding HLA ligands in Chinese Han individuals with the original strain SARS CoV-2 infection and compared them with randomly collected individuals from before the COVID-19 outbreak. This research has the potential to enhance our comprehension of the association of KIR and HLA with SARS CoV-2 infection and gain insight into the mechanisms of NK cell activity against COVID-19.

MATERIALS AND METHODS

Patients and controls

The patients included 119 individuals with original strain SARS CoV-2 infection (Defined as COVID-19 group). All of them had mild or severe conditions, with none in a critical state. They were confirmed positive for SARS CoV-2 viral RNA by fluorescence quantification polymerase chain reaction (PCR) test. The patients were recruited to donate plasma after recovery and the plasma was prepared to use for treatment of other severe or critical COVID-19 patients from February to April 2020. The patients’ age ranges from 20 to 54 years old. All samples were collected during the process of plasma donation, with informed consent being obtained from each patient. The control individuals were randomly collected from Zhejiang Han population, China, as our previous report (23) (Defined as control group, n=176). Approval for this study was granted by the regional ethics committee at the Blood Center of Zhejiang Province. Genomic DNA was extracted from peripheral blood utilizing QuickGene DNA whole blood kits (FujiFilm Corporation, Tokyo, Japan) following the manufacturer’s guidelines. The extracted DNA was then employed for KIR and HLA genotyping.

KIR and HLA genotyping

The KIR and HLA genes were sequenced by the NGS platform based on a biotinylated DNA probe-based capture method as described (24). Briefly, Genomic DNA (500ng from each sample) was fragmented and labeled uniquely using custom dual-index adapters (Illumina Inc, San Diego, CA). The fragments from each specimen were pooled equally after dual size selection, then enriched using capture probes according to the modified version of Nextera Rapid Capture enrichment protocol (Illumina Inc, San Diego, CA) as outlined in the provided description (24), modified according to previous report (25). The library concentration was assessed using a Qubit instrument, adjusted to final concentration of 12pmol/L, subsequently subjected to sequencing on a MiSeq instrument utilizing the v3 Reagent Kit (600-cycle; Illumina, Inc. San Diego, CA, USA).

The Pushing Immunogenetics to the Next Generation (PING) pipeline was employed to produce a detailed assessment of KIR gene content and genotype at the allele level with high resolution (24, 26). HLA genotypes were determined using the TypeStream Visual Software version 2.0 (One Lambda Inc.) and the ambiguous results of HLA genotype were resolved using the HLA common and well-document (CWD) alleles principle according to our previous report (27).

Distribution patterns of KIR genotypes, alleles, and haplotypes

Genotype designations, reflecting the presence or absence of KIR genes, were assigned in accordance with the nomenclature guidelines provided by Allelefrequencies.net database (28). Carrier frequencies of KIR genes and gene-content genotypes were determined by expressing the percentage of individuals with a positive result relative to the total number of individuals in each group (number of individuals positive divided by n). Allele and haplotype frequencies were calculated through direct counting (the number observed divided by 2n) as reported previously (29). The centromeric haplotype comprises all genes from KIR3DL3 through KIR3DP1, while the telomeric haplotype includes all genes from KIR2DL4 to KIR3DL2. Determination of the centromeric and telomeric haplotypes was based on the linkage disequilibrium among KIR genes and the copy number of each KIR gene in an individual. HLA allele frequencies were assessed using Arlequin 3.5 software (30).

Carrier frequencies of HLA allotypes

HLA allotypes were defined based on the HLA allele, and the population frequencies of each HLA allotypes were computed. This included A3/11, encoded by HLA-A*03 and HLA-A*11; Bw4, encoded by HLA-A and HLA-B alleles carrying the Bw4 motif at residues 77–83; C1, encoded by HLA-B or -C with asparagine at position 80; and C2, encoded by HLA-C with lysine at this position. The frequencies of these HLA allotypes were calculated as the number of observed HLA allotypes divided by the total number of alleles (2n).

HLA/KIR interactions

The potential interactions between KIR and HLA within individuals were evaluated, including 2DL1/C2⁺, 2DL2/C1⁺, 2DL3/C1⁺, 2DL2/C1⁺HLA-B, 2DL3/C1⁺HLA-B, 3DL1/Bw4⁺HLA-A, 3DL1/Bw4⁺HLA-B, 3DL1/Bw4⁺, 3DL2/A3/11⁺, 2DS1/C2⁺, 2DS2/C1⁺. The carrier frequency of HLA/KIR cognate pairs were determined by direct counting those positive for both KIR and cognate HLA ligand divided by total number of individuals (n). The mean number of KIR-HLA pairs per individual was calculated by summing the known viable interactions between KIR and HLA allotypes based on each individual’s KIR and HLA allele, then divided by the total number of individuals for each group as previously reported (24,–25).

Statistical analyses

Hardy-Weinberg equilibrium (HWE) was evaluated through maximum likelihood using the Arlequin software version 3.5 (30). Differences between the COVID-19 group and the control group were analyzed using the χ2 test for categorical variables. Odds ratios (95% confidence interval [CI]) and p-values were computed using Fisher’s exact test in Prism 5.0 software (GraphPad, San Diego, California). Corrected P-values (pc) were determined by applying the Bonferroni correction method, multiplying the number of alleles at each locus. The significance threshold for the pc-value was set at 0.05.

RESULTS

To elucidate the potential role of KIR genes and HLA class I ligands in genetic susceptibility to SARS CoV-2 infection, we compared KIR gene frequencies, genotypes, alleles, haplotypes, KIR ligands, and KIR-HLA interactions between the COVID-19 group and the control group. Analysis of KIR gene carrier frequencies revealed that framework genes (KIR3DL3, 3DP1, 2DL4, and 3DL2) were present in all individuals from both groups, while centromeric haplotype B genes (KIR2DS2, 2DL2, and 2DL5B) exhibited a modest increase in the COVID-19 group compared to the control group (9.2% to 22.7% vs. 7.9% to 18.2%) (supplemental table 1). Conversely, the frequencies of the telomeric haplotype B genes (KIR3DS1, 2DL5A, and 2DS1) were slightly decreased in the COVID-19 group, ranging from 30.3% to 31.9%, compared to 34.1% to 35.2% in the control group. The frequencies of the telomeric haplotype A genes KIR3DL1 and KIR2DS4 were marginally higher in the COVID-19 group. However, the differences did not reach statistical significance.

Genotypes were classified into group AA and Bx based on the presence or absence of KIR genes (www.allelefrequencies.net). Twenty distinct KIR genotypes were identified in the control group, with 17 genotypes in the COVID-19 group (Figure 1A, supplemental table 2). AA1 was the only AA genotype observed in both groups, constituting the most prevalent genotype with frequencies of 53.4% in the control group and 62.2% in the COVID-19 group, exhibiting no significant difference. Bx genotype frequencies varied widely in both groups, ranging from 0% to 14.2% in the control group and 0% to 8.6% in the COVID-19 group. Bx2 was the most frequent Bx genotype in both groups, notably less frequent in the COVID-19 group without statistical significance. Interestingly, Bx3, the second most frequent Bx genotype, exhibited a nominal association with the COVID-19 group (8.4% vs 2.3%, p=0.023, OR=3.95 (Figure 1B), although significance was lost after Bonferroni correction. No other associations were found with Bx genotypes.

Figure 1. Comparison of KIR genotype frequencies between the COVID-19 and the control group.

A. The color bar below indicates the groups, with the pink bar representing the COVID-19 group and the purple bar representing the control group. The carrier frequencies calculated as N/n*100%, where n is the total number of individuals tested, and N represents the number of individuals positive for the genotypes. B. Shown is the forest plot visualizing the odds ratios (OR) with 95% confidence intervals (CIs) for the genotypes from the two groups. The red lines represent the 95% CIs, and the blue dots indicate the ORs. The vertical dashed line at OR=1 indicates the neutral effect. The data on the right side provide details on the ORs, 95% CIs, and p-values for each genotype comparison between the two groups.

Alleles of each KIR gene, excluding KIR3DP1, were analyzed, identifying 111 alleles in the COVID-19 group and 108 alleles in the control group (supplemental table3). The inhibitory KIR exhibited greater polymorphism than activating KIR in both groups. KIR3DL3 was the most polymorphic gene, with 24 alleles in the COVID-19 group and 23 in the control group (Figure 2A). Remarkably, frequencies of the alleles KIR3DL3*00802 (OR=4.276; P=0.0008), KIR3DL3*048 (OR=2.892; P=0.0166), and KIR3DL3*00301 (OR=4.572; P=0.0172) were significantly increased in the COVID-19 group (Figure 2B). Notably, the frequency difference for the KIR3DL3*00802 allele remained statistically significant after Bonferroni correction (pc=0.0248). KIR2DS4 exhibited high polymorphism. Due to its distinction between the full-length and deleted forms (representing functional and non-functional variants, respectively), we conducted a detailed analysis of their distributions. It indicated that the full-length KIR2DS4 was present in 54.6% of the COVID-19 group compared to 52.3% in the control group, while the deleted form was found in 27.8% of the COVID-19 group and 26.5% of the control group. No significant differences were observed.

Figure 2. Distribution and comparison of KIR3DL3 alleles in the COVID-19 and control group.

A. Allele frequencies were calculated as N/2n*100%, where n is the total number of individuals tested and N represents the number of corresponding alleles. B. Shown is the forest plot visualizing the odds ratios (OR) with 95% confidence intervals (CIs) for the alleles from the two groups. The red lines represent the 95% CIs, and the blue dots show the ORs. The vertical dashed line at OR=1 indicates the neutral effect. The data on the right side detail the ORs, 95% CIs, and p-values for each allele comparison between the two groups.

Centromeric and telomeric haplotypes were analyzed based on the alleles and copy number of each KIR gene, with detailed distribution presented in supplemental table 4. Across both groups, the frequency of the A haplotype exceeded that of the B haplotype in each motif. While the Cen-A haplotype exhibited lower frequency and Tel-A slightly higher frequency in the COVID-19 group compared to the control group (supplemental table 5), no statistically significant differences were observed.

The HLA alleles were successfully typed for 111 out of 119 COVID-19-infected individuals, while 8 samples did not yield conclusive results. Among the HLA allotypes, all HLA-C allotypes, as well as certain HLA-A and HLA-B allotypes, serve as ligands for KIR. Analysis of corresponding ligands (HLA-C1, C2, Bw4, and A3/11) was conducted based on the allotype of the encoded HLA alleles (supplemental table 6). The frequency of the Bw4 allotype encoded by HLA-B was 29.3% in the COVID-19 group and 38.6% in the control group (Figure 3A, supplemental table 7), reflecting a statistically significant difference (OR=0.6575; p=0.024) (Figure 3B). Conversely, the frequency of the C1 allotype encoded by HLA-C was 88.3% in the COVID-19 group and 81.9% in the control group (OR=1.74; P=0.027), though lacking statistical significance after P correction. A3/11, a ligand for KIR3DL2 (16), was found in 35.1% of the COVID-19 group and 29.8% of the control group, displaying no statistically significant difference. Further analysis revealed that HLA-A*11, not HLA-A*03, substantially contributed to the high proportion of A3/11 in both groups, with HLA-A*11 accounting for 32.9% and 27.8% in the COVID-19 group and control group, respectively. This is due to the fact that A3 is much less frequent than A11 in South China (23, 29). HLA homozygosity genotypes analysis showed that the homozygosity rates were 17.1% and 14.8% for HLA-A, 8.1% and 6.3% for HLA-B, and 7.2% and 7.9% for HLA-C in the COVID-19 and control groups, respectively. At the allele level, the most frequent homozygous genotype was HLA-A*11:01, with frequencies of 7.2% and 9.1% in the COVID-19 and control groups, respectively. No significant differences in these homozygosity distributions were observed between the two groups (supplemental table 8).

Figure 3. Comparison of HLA allotypes between the COVID-19 group and the control group.

A. shown is a bar chart comparing the distribution of HLA allotypes between the COVID-19 and the control group. The specific distribution frequencies are provided in Supplemental Table 7. B. Shown is the forest plot visualizing the odds ratios (OR) with 95% confidence intervals (CIs) for the allotypes from the two groups. The red lines represent the 95% CIs, and the blue dots show the ORs. The vertical dashed line at OR=1 indicates the neutral effect. The data on the right side detail the ORs, 95% CIs, and p-values for each allotype comparison between the two groups.

The KIR-HLA interaction of the two groups was analyzed. Compared to the control group, lower frequencies of the 2DL1/C2⁺ and 3DL1/Bw4⁺HLA-B were observed in the COVID-19 group, alongside higher proportions of the 2DL2/C1⁺, 3DL1/Bw4⁺HLA-A, and 3DL2/A3/11⁺ (supplemental table 9). When concerning the mean number of viable interactions per individual for each receptor ligand pair (supplemental table 10) indicated a trend towards higher mean interaction of inhibitory functional units per individual in the COVID-19 group, such as 2DL2/C1⁺, 2DL3/C1⁺, 2DL3/C1⁺HLA-B, and 3DL2/A3/11⁺, but showed no significant difference. These findings, all pertaining to inhibitory KIR-HLA interactions, suggest a potential increase in inhibitory effects in the COVID-19 group.

DISSCUSION

Since 2020, an increasing number of studies have delved into various aspects of COVID-19, ranging from factors influencing disease progression to immune control and infection mechanisms (31–32). Comorbidities, advanced age, and reduced lymphocyte count have consistently emerged as risk factors for COVID-19, as confirmed in previous reports (33–34). Moreover, global case-control studies have pinpointed specific genes associated with susceptibility to COVID-19 (35–36). Among these genes, HLA and KIR genes stand out as critical regulators of immune molecules, boasting high genetic polymorphism within the human genome (20, 37). They have been implicated in the onset of numerous diseases and are recognized as pivotal genetic factors influencing disease susceptibility (10, 38–39). Studies have increasingly highlighted the role of these genes in SARS-CoV-2 infection, disease progression, immune response to viral strains, and post-COVID-19 symptoms. However, consensus remains elusive in these investigations (40–43). For instance, studies from Saudi Arabia (40) suggested an association between KIR2DS4 and 3DL1 genes and an increased risk of severe COVID-19. This finding is consistent with phenotyping studies identifying an excess of KIR2DS4+ NK cells in infected tissues, and with a large cohort study from Italy (20–22) but was not replicated in patients from Sicily by Ligotti et al. (41). Similarly, Maruthamuthu et al. (42) reported an association between the presence of KIR2DS1+KIR2DS5+ and elevated susceptibility to severe COVID-19. Discrepancies in these findings may stem from population differences and other factors.

In our study, while no significant difference in the distribution of KIR genes was observed between the COVID-19 and control groups, a similar trend of slightly higher KIR2DS4 and KIR3DL1 frequencies was noted in the COVID-19 group, akin to previous findings (20–22, 40). Additionally, we observed an increased frequency of centromeric B genes in the COVID-19 group, albeit not reaching statistical significance. Notably, a significantly higher frequency of the Bx3 genotype was detected in the COVID-19 group. The Bx3 genotype encompasses all KIR genes except KIR2DS3, potentially explaining its higher frequency in the COVID-19 group due to the low occurrence of KIR2DS3 in these individuals. KIR2DS3 has been reported as significantly associated with protective effects against COVID-19 (44). Although the incidence rate of KIR2DS3 did not significantly decrease in COVID-19 patients in our study, and it is not known if KIR2DS3 binds to any HLA molecules, we speculate that its absence in the Bx3 genotype may compromise the protective ability against COVID-19.

Given the limited use of KIR high-resolution genotyping, only a few reports have explored the correlation between COVID-19 and KIR alleles. Our study addressed this gap by integrating KIR alleles into the COVID-19/KIR association analysis. Notably, three KIR3DL3 alleles, KIR3DL3*00802, 3DL3*048, and 3DL3*00301 were associated with COVID-19 risk, particularly KIR3DL3*00802, which remained significance after p correction. KIR3DL3, devoid of a reported HLA ligand, binds instead B7 family member B7H7 and is an inhibitory receptor expressed predominantly by tissue resident T cells (18, 45). Distinguished from other genes, KIR3DL3 holds a unique status as the sole KIR gene present in every individual across diverse populations and haplotypes, suggesting it plays a vital role in human survival (46). The identification of these three KIR3DL3 alleles positively associated with COVID-19 underscores a significant improvement in understanding KIR3DL3 function, warranting further investigation.

Previous studies (47–48) have reported a higher degree of HLA homozygosity in COVID-19 groups, which may be associated with disease susceptibility. We compared the distribution of HLA-A, -B, and -C homozygosity between the COVID-19 and control populations but found no significant differences. This lack of difference could be attributed to variations in the conditions such as severity of COVID-19 infection among the patients.

Generally, KIR do not regulate NK cell activity alone, but through interaction with HLA on target cells, thereby influencing antiviral and immune mediating functions. In the COVID-19 group, we observed more frequent interactions like 2DL2/C1⁺, 3DL1/Bw4⁺HLA-A, 3DL2/A3/11⁺, whereas 2DL1/C2⁺ and 3DL1/Bw4⁺HLA-B were less frequent. Therefore, we deduced that the inhibitory signals of 2DL2/C1⁺, 3DL1/Bw4⁺HLA-A, and 3DL2/A3/11⁺ were dominant in the COVID-19 group over other inhibitory interactions. We speculate that these KIR-HLA interactions may conduct stronger inhibitory signals and consequently repress NK cell activity, resulting in the virus escape from the immune surveillance of NK cells, and favor the development of COVID-19. This finding is in accordance with the previous reports for the COVID-19 patients, particularly those experiencing severe symptoms, where a consistent observation of lower NK cell counts was documented (49). Additionally, there is evidence suggesting an exhausted NK cell phenotype in COVID-19 (50), wherein the interplay of KIR and HLA may contribute and influence the progression of the disease.

A previous study reported that HLA-C*05 could recognize and bind viral peptide, subsequently being recognized by functional KIR2DS4, leading to immune over activation, which is significantly associated with a higher risk of COVID-19 and mortality (51). Allele frequency of full-length KIR2DS4 is inversely correlated with the frequency of HLA-C*05:01 (52). In our population, as well as for most East Asian populations, HLA-C*05 is rare, while frequency of full-length KIR2DS4 is high. However, besides HLA-C*05, KIR2DS4 also interacts with other ligands, such as HLA-A11 and certain HLA-C1/C2 encoded by HLA-C*01 and -C*04, but no significant differences were observed in these ligands and interaction with 2DS4 between the COVID-19 and control. Additionally, another HLA allotype, HLA-Bw4, has been demonstrated associated with the protective functions against viral infections and tumor metastasis, especially in HIV infection, which slowed the procession of HIV infection and reduced risk of HIV transmission (53). Our study revealed a significant decrease in Bw4 encoded by HLA-B in the COVID-19 group, suggesting a potential role in resistance against COVID-19. It’s worth noting that Bw4 can be encoded by both HLA-A and HLA-B, and the significant decrease in Bw4⁺HLA-B in the COVID-19 group implies that Bw4 encoded by HLA-A and HLA-B may have distinct roles in regulating NK cell activity, thus warranting further investigation.

Due to limitations in sample size, we did not subcategorize individuals infected with SARS CoV-2 by severity. Focusing on the original strain of SARS-CoV-2, our study investigated the relationship between highly polymorphic KIR and HLA genes and susceptibility to or protection against COVID-19. Our results highlighted the significant association of KIR3DL3*00802 with COVID-19 risk and the protective nature of Bw4⁺HLA-B allotype. Additionally, KIR-Bx3, KIR3DL3*00301, 3DL3*048, and C1⁺HLA-C showed nominal associations with COVID-19. Because we used population controls collected before the pandemic, it is unknown if any of the control cohort contracted SARS-CoV-2 infection. Our study is therefore underpowered for detecting disease association effects. Thus, considering the potential conservatism of p correction, we should not rely solely on statistically significant factors after correction. Our findings will contribute to exploring the mechanisms underlying SARS CoV-2 infection and outcomes of the COVID-19 patients. Further research is warranted to fully elucidate these mechanisms.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in the article/Supplementary Material and allelefrequencies.net database.