Introduction

The clinical course of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection displays wide heterogeneity, ranging from asymptomatic to critical illness and death [1]. While demographic and clinical factors such as age, comorbidities, and blood group influence coronavirus disease 2019 (COVID-19) severity, host genetics may also play a substantial role in disease outcomes [2–4].

Among immunogenetic factors, Killer Immunoglobulin-like Receptors (KIRs), encoded on chromosome 19q13.4, have been implicated in modulating antiviral responses. Anthropologically KIR genes, which mostly interact with specific human leukocyte antigen (HLA) class I ligands, evolved relatively recently compared to HLA loci and exhibit extensive polymorphism and copy number variations. Previous studies, including our own, have associated specific KIR genotypes and haplotype motifs with COVID-19 severity, highlighting the importance of innate immune signaling diversity, particularly its role in the clearance of SARS-CoV-2 infected cells through NK cell activation mediated by KIR-HLA ligand interactions [5–7].

Another evolutionary aspect of immune responsiveness involves Neanderthal-introgressed genomic regions. Zeberg and Pääbo (2020) identified a Neanderthal-derived haplotype on chromosome 3p21.31, encompassing genes such as SLC6A20, LZTFL1, and CCR9, that confers increased risk of severe COVID-19 [8]. However, a subsequent study by the same group [9] revealed a distinct Neanderthal-inherited haplotype at the OAS1/2/3 locus on chromosome 12q24.13, which instead provides protection against severe disease. These findings underscore the dualistic impact of Neanderthal ancestry on COVID-19 outcomes, with both risk-enhancing and protective genomic segments persisting in modern human populations.

The OAS1/2/3 gene cluster encodes oligoadenylate synthetases that activate RNase L-mediated degradation of viral RNA, constituting a key arm of the interferon-stimulated antiviral response. Functional studies have shown that the Neanderthal-derived allele at OAS1 (rs10774671-G), which preserves an ancestral splice variant, results in higher enzymatic activity and enhanced viral restriction [10,11]. In addition to rs10774671, the Neanderthal-introgressed OAS haplotypes include three missense variants (rs2660, rs1293767 and rs1859330) and two synonymous variants (rs1859329 and rs2285932), all of which have been associated with an altered course of various viral infections [9,12]. Considering that the protective Neanderthal alleles at OAS1/2/3 are annotated as the reference alleles in GRCh38 (OAS1: rs10774671-G and rs2660-G; OAS2: rs1293767-C; OAS3: rs1859330-G, rs1859329-C and rs2285932-T), individuals lacking the alternate variants in OAS1/2/3 are expected to mount more effective innate immune responses to RNA viruses such as SARS-CoV-2.

Given these parallel findings on KIR variability and Neanderthal-introgressed OAS1/2/3 alleles, we hypothesized that their interaction might synergistically modulate COVID-19 outcomes. In the present study, we systematically evaluated the combined influence of KIR haplotype motifs, OAS1/2/3 protective variants, and additional polymorphisms in immune-related genes (IFITM3: restriction of virus replication [13]; DPP4: viral entry, inflammatory response and insulin resistance [14]; TLR7: interferon induction [15]; APOE: immunometabolic regulation [16]) on COVID-19 symptomatology and severity. Notably, individuals were classified as OAS1/2/3(-) if they carried at least one Neanderthal-derived protective allele (rs10774671-G, rs2660-G, rs1293767-C, rs1859330-G, rs1859329-C or rs2285932-T), consistent with prior functional and population genetic evidence. By integrating these immunogenetic factors within multivariate predictive models, we sought to identify key genetic determinants of disease progression in a cohort of unvaccinated individuals infected with SARS-CoV-2.

Methods

Results

Clinical and demographic features

The study included 175 unvaccinated individuals with confirmed SARS-CoV-2 infection: 65 were asymptomatic, 47 had mild-to-moderate disease, and 63 experienced severe or critical illness requiring intensive care unit (ICU) admission. Clinical severity was associated with increasing age, male gender, and higher comorbidity burden (Table 1). Blood group A, previously linked to infection susceptibility [19,20], was paradoxically more frequent among asymptomatic individuals in this cohort. Frequency distributions were validated using external controls including 75 donors from Ankara University Donor Registry and data from the Turkish Genome Project Data Sharing Portal (TUSEB) (n = 557) (https://tgd.tuseb.gov.tr/en/. Accessed 29 Dec 2024).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Patient demographics and frequency distributions across disease severity groups.

https://doi.org/10.1371/journal.pone.0345137.t001

KIR motifs and immune variant frequencies

KIR/KIR-ligand motifs tAA/wL and tAB1/wL displayed divergent distributions across disease categories. The tAA/wL motif was enriched in symptomatic cases, while the tAB1/wL motif was more frequent among asymptomatics. Additional genotyping included SNPs in OAS1/2/3, IFITM3, DPP4, TLR7, and APOE (Table 2). Variants were grouped based on known functional or ancestral profiles. In particular, individuals were classified as OAS1/2/3(-) if they carried at least one Neanderthal-derived allele previously shown to increase antiviral OAS activity.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Genotype frequencies of key immunogenetic variants across COVID-19 severity groups.

https://doi.org/10.1371/journal.pone.0345137.t002

The minor allele of rs12252 (IFITM3) was associated with symptomatic disease, and rs3788979 (DPP4) with severe disease. The APOE ε3ε3 genotype was more common among asymptomatic individuals and appeared protective. Rare APOE genotypes (ε2ε2, ε2ε4, ε4ε4) were not observed. A complete list of analyzed variants, with rsIDs, allele designations, and genomic context, is provided in S1 Table.

KIR-OAS synergy and multivariate analysis

To investigate potential interactions between KIRs and SNPs, frequency distribution of each pair of KIR genotypes and SNPs were compared within the asymptomatic/symptomatic and the non-severe/severe patient groups. Binary Logistic Regression was performed to analyze the individual effects of KIR genotypes, SNPs, and their combined effects. A significant interaction has been found between tAB1/wL and OAS1/2/3, favoring their synergistic ability to discriminate between asymptomatic vs. symptomatic groups (P = 0.032). Further analysis revealed that individuals lacking tAB1/wL while carrying OAS1/2/3(+) alleles were significantly more likely to be symptomatic (OR 3.471, 95% CI: 1.429–8.432, P = 0.006) and to develop severe disease (OR 2.409, 95% CI: 1.051–5.519, P = 0.038), compared to other genotype combinations (Table 3). This synergistic genotype was significantly rare in asymptomatics but present in over 40% of symptomatic patients (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Binary logistic regression models identifying independent predictors of the presence of the symptoms and the severity of COVID-19.

https://doi.org/10.1371/journal.pone.0345137.t003

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Distribution of OAS1/2/3 and tAB1/wL genotype combinations among A) asymptomatic vs. symptomatic and B) non-severe vs. severe patients.

https://doi.org/10.1371/journal.pone.0345137.g001

To crosscheck the independent effects of OAS and KIR genotypes and to assess any potential misleading effects due to collinearity, frequency distributions were analyzed across the entire cohort, including both patients and registry population. Cross-tabulation of OAS1/2/3 with tAB1/wL or tAA/wL genotypes, regardless of the study groups, revealed no significant differences in their frequencies, suggesting that the associations with disease progression are not due to the uneven distribution of the genotypes (S2 Table).

Multivariate logistic regression identified age, blood group A, IFITM3 rs12252, APOE ε3ε3, tAA/wL status, and the OAS1/2/3(+)/tAB1/wL(-) interaction as independent predictors of symptomatic infection. For severe disease, the final model retained age, male gender, blood group A, DPP4 rs3788979, and the same synergistic OAS1/2/3(+)/tAB1/wL(-) genotype (Table 3, Fig 2).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Forest plots showing odds ratios and 95% confidence intervals for predictors of symptomatic (left) and severe (right) COVID-19.

https://doi.org/10.1371/journal.pone.0345137.g002

Model performance was robust: The symptomatic vs. asymptomatic classifier achieved 75.4% accuracy with an AUC of 0.849, sensitivity of 81.4%, specificity of 69.2%, and a positive predictive value (PPV) of 76.6%. The severe vs. non-severe model achieved 78.9% accuracy with an AUC of 0.861, sensitivity of 79.4%, specificity of 78.4%, and a PPV of 73.9%. Full performance metrics are provided in S1 Fig.

Discussion

Our study demonstrates, for the first time, a synergistic effect between the absence of the tAB1/wL KIR motif and the presence of Neanderthal-derived OAS variants [OAS1/2/3(-)] in predicting both symptomatic and severe COVID-19 progression. Age, male gender, blood group A, and SNPs in IFITM3 and DPP4 further contributed to disease outcomes. These findings advance the understanding of how both recent immune gene evolution (KIR polymorphisms) and ancient introgressed variants (OAS haplotypes) jointly modulate human viral susceptibility. To provide a biological framework for the observed genetic associations, we generated a gene–gene interaction network using the GeneMANIA platform [21]. This network illustrates known connections between KIR-associated HLA signaling, interferon-mediated antiviral pathways, and immunometabolic regulators, supporting the plausibility of gene–gene interactions underlying COVID-19 symptomatology and severity and is presented as a conceptual framework to provide biological context for the observed genetic associations rather than as direct evidence derived from the present cohort (Fig 3).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Gene–gene interaction network generated using the GeneMANIA platform illustrating known functional, physical, and pathway-based relationships among genes investigated in this study [21].

The network highlights interactions linking KIR-associated HLA signaling with interferon-stimulated antiviral pathways (OAS1, OAS2, OAS3, IFITM3, TLR7) and immunometabolic regulators (APOE, DPP4). Edges represent previously reported co-expression, physical interaction, pathway co-membership, or shared protein domains.

https://doi.org/10.1371/journal.pone.0345137.g003

In this study, we evaluated a panel of immune-related SNPs previously associated with COVID-19 severity and mortality, rather than conducting an untargeted genome-wide analysis. By assessing these variants across clinically defined categories – asymptomatic vs symptomatic and non-severe vs severe – we aimed to contextualize their individual contributions to disease progression.

Our findings suggest a layered genetic influence along the symptomatic-to-severe disease spectrum. IFITM3 rs12252-GG and APOE ε3ε3 genotype were associated primarily with the presence and absence of symptoms, respectively; while DPP4 rs3788979-TT was more strongly linked to progression toward severe disease. These findings are consistent with the proposed roles of these gene polymorphisms in the course of infection. While the IFITM3-GG genotype has been associated with impaired viral restriction [13], APOE ε4 variants have been linked increased cholesterol levels and enhanced accumulation of ACE2 and TMPRSS2 within cholesterol-enriched domains [22]. Although these factors alone are insufficient to explain the entire mechanism, it is reasonable to suggest that patients lacking APOE ε3ε3 genotype and carrying IFITM3-GG variants may be more prone to a symptomatic course of infection, as host immunity may fail to effectively prevent viral entry and replication. DPP4, on the other hand, is suggested to play multiple roles in COVID-19 pathogenesis [14]. In addition to its involvement in viral entry, its physiological effects on T cell activation, cell adhesion and apoptosis may exert a greater influence on disease severity rather than on symptomatic determination. Although rare deleterious TLR7 variants have been associated with severe COVID-19 in young males, their absence in our cohort is consistent with their very low population frequency and the limited sample size. Notably, the interaction between OAS1/2/3(+) and absence of the protective tAB1/wL KIR motif was a shared predictor of both symptom onset and severity, suggesting it may represent a broader axis of innate immune dysregulation relevant throughout the clinical course. Moreover, cross-sectional frequency distributions of OAS1/2/3 and tAB1/wL combinations among the significant predictors of symptomatic and severe disease confirm that the predisposing role of OAS1/2/3(+)/tAB1/wL(-) genotype was preserved despite the lack of additional risk factors, i.e., older age, male gender, IFITM3 rs12252, or DPP4 rs3788979 status (S2 Fig).

While some variants may plausibly influence both stages of disease (e.g., IFITM3 or DPP4), the lack of statistical significance in one of the comparisons may reflect power limitations rather than true absence of effect. These observations support the hypothesis that distinct genetic mechanisms may underlie the transition from viral containment (asymptomatic vs symptomatic) and from immune control failure (non-severe vs severe), with partially overlapping yet non-identical contributors.

Consistent with prior studies [23], advanced age was the strongest predictor of severity. Comorbidity burden correlated with severity in univariate models but was not retained in multivariate models, likely reflecting cohort size limitations. While blood group A has been associated with infection susceptibility [19,20], there have been contradictory conclusions in worldwide studies regarding its association with disease severity [4]. The distribution of blood groups in our cohort is consistent with the earlier reports and the 2024 update from Turkish Red Crescent [24,20]; nevertheless, the protective role of blood group A against symptomatic and severe COVID-19 observed in our study is novel within Turkish populations and warrants further study in larger populations.

Our findings align with studies showing inhibitory KIR motifs (e.g., tAA) predispose to worse outcomes, whereas activating motifs (e.g., tAB1) are protective [6,25]. In contrast to previous studies examining individual KIR genes or general haplotype groups, such as the study by Littera et al. (2021) [6], our analysis focused on composite KIR genotypes defined by specific activating or inhibitory motifs (tAA and tAB1) in combination with their respective HLA ligands. This approach was chosen to reduce dimensionality and preserve statistical power in the context of a limited sample size. Moreover, evaluating ligand-associated KIR motifs allowed us to target broader NK cell regulatory effects, rather than isolated gene-level signals. While this genotypic definition increases specificity, it may reduce generalizability and comparability with studies using broader haplotype categories. Therefore, rather than testing all possible motif combinations, we focused on genotypes previously associated with clinical outcomes in our earlier cohort [5]. Differences in KIR association results across studies likely reflect ethnic-specific KIR-HLA diversity and variable analysis strategies (allelic vs haplotypic).

Supporting Zeberg and Pääbo (2021), carriers of Neanderthal-protective OAS1/2/3 alleles exhibited reduced risk of symptomatic or severe infection [9]. However, this protection was overcome when combined with absence of the protective tAB1/wL genotype – highlighting gene-gene interactions in innate antiviral defense.

The rs12252 variant in IFITM3 and the rs3788979 variant in DPP4 were associated with symptomatic infection and disease severity, respectively, consistent with prior meta-analyses [14,22]. APOE ε3ε3 conferred relative protection, though associations involving rare APOE variants (e.g., ε2ε3) require cautious interpretation due to small sample sizes. No rare TLR7 loss of function variants were detected, consistent with their extremely low frequency; larger male-enriched cohorts would be needed for detection.

From the evolutionary perspective, the interaction between Neanderthal OAS alleles and KIR genotypes supports the concept that introgressed innate immune alleles and recent KIR diversification both shaped modern human responses to RNA viruses. Such dual-layered genetic architecture likely conferred historical survival advantages but now influences modern disease susceptibilities.

This study has several notable strengths. By integrating KIR haplotype motifs, Neanderthal-inherited OAS1/2/3 variants, and modern immune-related SNPs within the same predictive models, we provide a comprehensive immunogenetic evaluation of COVID-19 outcomes. The use of a homogenous population, validated against national genomic databases, minimizes the risk of population stratification bias. Furthermore, the application of multivariate logistic regression, ROC curve analysis, and cross-validation with healthy donor controls enhances the robustness of our findings.

Earlier, the population-based distributions of Neanderthal-inherited SNPs or KIR genes have been published [9,18,26]. Likewise, susceptibility against COVID-19 has also been shown to be influenced by ethnic background. In our earlier study, KIR genotype frequencies in our geographical region were similar to those of Europeans [5]. According to Zeberg and Pääbo (2021), frequencies of Neanderthal-introgressed OAS haplotypes have increased over time, i.e., OAS3 rs1156361 from below 10% to ~30%, in Eurasia in the last 20,000 years [9]. In this study, Neanderthal-inherited allele frequency of OAS SNPs was ~ 57% in the registry donors, similar to that in the TUSEB (https://tgd.tuseb.gov.tr/en/. Accessed 29 Dec 2024), consistent with the idea of positive selection of innate immunity gene variants in modern humans.

Nevertheless, some limitations should be acknowledged. First, our genotyping approach did not resolve KIR copy number variation, which could contribute additional phenotypic variability. Second, the relatively small sample size limited our power to detect associations involving rare genetic variants, particularly in APOE and TLR7. Third, while population homogeneity strengthens internal validity, it also restricts the generalizability of our results to broader, multiethnic populations. Finally, by combining mild-intermediate and severe cases into a single “symptomatic” group for some analyses, we may have partially obscured biological differences between truly mild and critically severe disease trajectories. These limitations notwithstanding, the observed synergy between KIR and OAS1/2/3 genotypes offers novel insights into the layered evolutionary architecture of innate antiviral immunity.

Conclusions

In this study, we investigated potential evolutionary interactions among the multifactorial components of immune defense by extending our previous findings on KIR - COVID-19 associations. We uncovered a novel immunogenetic interaction between KIR motifs and Neanderthal-derived OAS1/2/3 variants impacting COVID-19 symptomatology and severity. The validity of our findings may differ in diverse populations which require further investigation, incorporate deeper KIR characterization (including CNV), and explore functional consequences of these gene–gene interactions in antiviral immunity. After validation in independent cohorts, these immunogenetic signatures may contribute to genetic risk stratification and, in selected settings, inform personalized preventive or monitoring strategies for susceptible individuals.

Supporting information