KIR genotypic diversity can track ancestries in heterogeneous populations: a potential confounder for disease association studies

Abstract

Killer cell immunoglobulin-like receptors (KIR) are encoded by highly polymorphic genes that regulate the activation of natural killer (NK) cells and other lymphocyte subsets and likely play key roles in innate and adaptive immunity. Association studies increasingly implicate KIR in disease predisposition and outcome but could be confounded by unknown KIR genetic structure in heterogeneous populations. To examine this, we characterized the diversity of 16 KIR genes in 712 Northern Californians (NC) stratified by self-assigned ethnicities and compared the profiles of KIR polymorphism with other US and global populations using a reference database. Sixty-eight distinct KIR genotypes were characterized: 58 in 457 Caucasians (NCC), 17 in 47 African Americans (NCAA), 21 in 80 Asians (NCA), 20 in 74 Hispanics (NCH), and 18 in 54 "other" ethnicities (NCO). KIR genotype patterns and frequencies in the 4 defined ethnicities were compared with each other and with 34 global populations by phylogenetic analysis. Although there were no population-specific genotypes, the KIR genotype frequency patterns faithfully traced the ancestry of NCC, NCAA, and NCA but not of NCH whose ancestries are known to be more heterogeneous. KIR genotype frequencies can therefore track ethnic ancestries in modern urban populations. Our data emphasize the importance of selecting ethnically matched controls in KIR-based studies to avert spurious associations.

Fig. 1

KIR genotypes observed among North Californian (NC) ethnicities; Caucasian (NCC), African American (NCAA), Asian (NCA), Hispanic (NCH) and “others” (NCO). The first column labels and stratifies by frequency the distinct KIR genotypes generated from 712 subjects by KGP. The KIR genotypes are depicted in rows. The respective KIR inhibitory, activating and pseudo genes are ordered in columns, from left to right. By convention, black cells designate the presence and white cells the absence, of respective genes. The ethnic breakdown of the dataset is shown in the next 5 columns, NCC-NCO. Adjacent to the respective ethnicity headings, within square brackets [], are the total number of individuals separated by (/) from the total number of KIR genotypes in each ethnic group. In each column, the number and percent (in parentheses) of individuals possessing respective KIR genotypes is shown. The frequency rank of respective KIR genotypes in each ethnic group is shown in the 5 right columns. NC1, alone, can be seen as the most frequent genotype in all ethnicities. The genotype shaded in grey (NC28) occurred only in NCC. The 10 non-NCC genotypes (NC59 to NC68) were found only among non-Caucasians in our NC cohort.

Fig. 2

Phylogenetic tree constructed using KIR genotype frequencies of 4 ethnicities sampled from NC (this study) and SC. Ethnic abbreviations are shown in Table 2. Respective Caucasian, Asian and African American clades are indicated.

Fig. 3

Phylogenetic trees based on frequencies of 162 KIR genotypes in respective populations (Cohort 4 in Table 3). Ethnic abbreviations are shown in Table 2. In clockwise order, African, European, Native American and Asian clusters are respectively enclosed within distinct outlines, as shown. The NC and SC populations are clustered in proximity to their native ancestries.