Polymorphic HLA-C Receptors Balance the Functional Characteristics of KIR Haplotypes

Abstract

The human killer cell Ig-like receptor (KIR) locus comprises two groups of KIR haplotypes, termed A and B. These are present in all human populations but with different relative frequencies, suggesting they have different functional properties that underlie their balancing selection. We studied the genomic organization and functional properties of the alleles of the inhibitory and activating HLA-C receptors encoded by KIR haplotypes. Because every HLA-C allotype functions as a ligand for KIR, the interactions between KIR and HLA-C dominate the HLA class I-mediated regulation of human NK cells. The C2 epitope is recognized by inhibitory KIR2DL1 and activating KIR2DS1, whereas the C1 epitope is recognized by inhibitory KIR2DL2 and KIR2DL3. This study shows that the KIR2DL1, KIR2DS1, and KIR2DL2/3 alleles form distinctive phylogenetic clades that associate with specific KIR haplotypes. KIR A haplotypes are characterized by KIR2DL1 alleles that encode strong inhibitory C2 receptors and KIR2DL3 alleles encoding weak inhibitory C1 receptors. In striking contrast, KIR B haplotypes are characterized by KIR2DL1 alleles that encode weak inhibitory C2 receptors and KIR2DL2 alleles encoding strong inhibitory C1 receptors. The wide-ranging properties of KIR allotypes arise from substitutions throughout the KIR molecule. Such substitutions can influence cell surface expression, as well as the avidity and specificity for HLA-C ligands. Consistent with the crucial role of inhibitory HLA-C receptors in self-recognition, as well as NK cell education and response, most KIR haplotypes have both a functional C1 and C2 receptor, despite the considerable variation that occurs in ligand recognition and surface expression.

Figure 1. KIR2DL1 and KIR2DS1 form four phylogenetic clades

(A) Shown is a phylogenetic analysis of 33 KIR2DL1 and 2DS1 nucleotide sequences representing the domains encoding amino acids 1-328. The phylogenetic relationships were inferred using three tree-building algorithms which showed broad consensus. Shown is a representative tree created using the Neighbor-Joining Method (34). The analysis identified four clades that have been color shaded for clarity. The optimal tree with sum of branch length = 0.071 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) is shown next to the branches when greater than, or equal to, 50. The evolutionary distances were computed using the Tamura-Nei method and are in the units of the number of base substitutions per site. All positions containing gaps and missing data were eliminated. In the final dataset there was a total of 872 positions. Evolutionary analysis was conducted in MEGA6 (35). (B) Shown is a sequence alignment of the most common allotypes in each of the four KIR2DL1 and 2DS1 clades identified by the phylogenetic analysis. Dots indicate identity with consensus (2DL1*001) and an asterisk indicates a termination codon. The lines beneath the alignment show the structural domains: Ig-like domains (D1+D2), stem (St), transmembrane domain (Tm) and cytoplasmic tail (Cyt). (C) Schematic diagram indicating the likely sites of recombination that define the four phylogenetic clades of KIR2DL1 and 2DS1 identified in (A).

Figure 2. KIR2DL1, KIR2DS1 and KIR2DL2/3 segregate on distinct KIR haplotypes

Shown is a listing of the published associations of KIR2DL1 and 2DS1 and KIR2DL2/3 alleles with the centromeric KIR A (Cen A, red), centromeric KIR B (Cen B, light blue) and telomeric KIR B (Tel B) haplotypes (29, 38-43). Alleles with a dual association are listed under both haplotypes with bold type indicating the less frequent association. Alleles that associate differently from the other alleles in their phylogenetic clade are highlighted with either light blue (phylogenetically Cen B but segregate with Cen A) or light red (phylogenetically Cen A but segregate with Cen B) shading. Alleles are grouped according to their clade that is shown in parentheses to the right of each allele.

Figure 3. KIR2DL2 and KIR2DL3 form four phylogenetic clades

(A) Shown is a phylogenetic analysis of 36 KIR2DL2/3 nucleotide sequences representing the domains encoding amino acids 1-328. The phylogenetic relationships were inferred using three tree-building algorithms. Shown is a representative tree created using the Neighbor-Joining Method (34). The analysis identified four clades that have been color shaded for clarity. The optimal tree with sum of branch length = 0.079 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (500 replicates) is shown next to the branches when greater than or equal to 50. The evolutionary distances were computed using the Tamura-Nei method and are in the units of the number of base substitutions per site. All positions containing gaps and missing data were eliminated. In the final dataset there was a total of 952 positions. Evolutionary analysis was conducted in MEGA6 (35). (B) Shown is an alignment of four KIR2DL2/3 allotypes representing the four clades identified using phylogenetic analysis. Dots indicate identity with the consensus (2DL3*001) and an asterisk indicates a termination codon. The lines beneath the alignment show the structural domains: Ig-like domains (D1+D2), stem (St), transmembrane domain (Tm) and cytoplasmic tail (Cyt). (C) Schematic diagram indicating the likely sites of recombination that define the four phylogenetic clades identified in part (A).

Figure 4. KIR Cen A encodes stronger C2 and weaker C1 receptors than Cen B

KIR2DL1 alleles that segregate on the Cen A haplotype encode receptors that bind to HLA-C2 targets with significantly greater avidity than those that segregate on the Cen B haplotype (two-tailed t-test, p<0.01) (left panel). By contrast, KIR2DL2/3 alleles that segregate on the Cen A haplotype encode receptors that bind to HLA-C2 targets with significantly lower avidity than those on the Cen B haplotype (two-tailed t-test, p<0.001) (center panel). KIR2DL2/3 alleles that segregate on the Cen A haplotype encode receptors that bind to HLA-C1 targets with significantly lower avidity than those on the Cen B haplotype (two-tailed t test, p<0.05) (right panel). The binding avidity of each allotype was assessed using the binding of KIR-Fc fusion proteins to microbeads, each coated with one of nine HLA-C1 and seven HLA-C2 allotypes (Figure S3). Binding values are normalized to that of the W6/32 antibody that binds to all HLA class I allotypes with equal avidity.

Figure 5. Variation at positions 154, 163 and 182 in the D2 domain reduces the avidity of KIR2DL1*004 for HLA-C2

(A) Structural representation of KIR2DL1 (PDB: 1NKR) (48) mapping the location of the four residues at which 2DL1*003 and 2DL1*004 differ. The protein backbone is shown in grey with the four D2 domain positions highlighted in yellow. (B) Mean binding of 16 KIR2DL1-Fc fusion proteins to microbeads, each coated with one of seven C2 HLA-C allotypes. Shown is a sequential mutation analysis in which every possible residue or combination of residues at which 2DL1*003 and 2DL1*004 differ is tested for binding to HLA-C2. The alignment to the left shows the identity of the residues in each KIR-Fc mutant. Binding values are normalized to that of the W6/32 antibody that binds to all HLA class I allotypes with equal avidity. Broken vertical lines indicate the binding of 2DL1*003 (dark blue) and 2DL1*004 (light blue) for comparison. (C) Shown is the mean binding to HLA-C2 of every mutant KIR-Fc containing the listed amino acid residue at positions 114, 154, 163 and 182. KIR-Fc with leucine 114 bound to HLA-C2 with a significantly lower avidity than those with proline 114 (two-tailed t-test, p=0.0047).

Figure 6. The amino-terminal half of the KIR2DL1 transmembrane domain is essential for cell-surface expression

(A) Cell-surface expression of natural and mutant KIR2DL1. Constructs encoding FLAG-tagged KIR2DL1 were transiently transfected into HeLa cells. The binding of anti-FLAG antibody to the transfected cells was measured. Shown are median fluorescence intensity (MFI) values for the KIR2DL1*012, KIR2DL1*026, KIR2DL1*003 and KIR2DL1*014 natural allotypes and for nine KIR2DL1 mutants, each containing a termination codon at the listed residue. Termination codons were placed at position 224 in the stem region (brown), positions 228, 231, 235, 238 and 242 in the transmembrane region (grey) and positions 246, 250 and 256 in the cytoplasmic tail (pale yellow). Error bars give the SD for three separate experiments. Statistically significant differences are denoted by brackets. Shown below is the amino-acid sequence at positions 220-258 in KIR2DL1*003. This sequence encompasses the stem region (brown), the transmembrane region (grey) and the cytoplasmic tail (yellow). Residues at which termination codons were introduced are shown in red. (B) Shown are con-focal microscopy images of HeLa cells 48h after transfection with FLAG-tagged KIR2DL1*012 (upper panels) and KIR2DL1*014 (lower panels). Columns a-e show: the bright field image, staining with Phalloidin AlexaFluor555 to identify the area near the cell surface, staining with FITC conjugated anti-FLAG antibody, a merged image of columns b and c and the co-localization of anti-FLAG and phalloidin. (C) Shown is the quantitative co-localization analysis in 3 dimensions performed between the phalloidin and anti-FLAG channels. The total volume of co-localized voxels per cell was calculated in Volocity (Perkin-Elmer) using eight cells for each of the six transfections performed. (D) Structural representation of KIR2DL1 (PDB: 1NKR) (48) showing the location of glycine 179 (yellow) buried in the interior of the receptor architecture (left panel). As seen in the enlargement in the right panels, structural analysis showed that substitution of glycine for serine at position 179 is predicted to disrupt protein folding as a result of a side-chain interaction between serine 179 and tyrosine 134.

Figure 7. Residues at positions 16 and 148 diversify the binding of two-domain KIR to HLA-C

(A) Structural representation of a two domain KIR (green) bound to HLA-C (blue) (PDB:1EFX) (49). Shown in red and enlarged in the right panel is the position of residues 16 and 148 that occupy the hinge region of the receptor. (B) Alignment showing the amino acid variation at positions 16 and 148 in 61 KIR2DL1 and KIR2DL2/3 allotypes. One representative allele with a unique combination of residues is shown for KIR2DL1 and KIR2DL2/3. The allotypes listed differ at residues other than 16 and 148 (Figure S2A and S2B). (C) For each unique combination of residues at positions 16 and 148, the number of KIR2DL1, KIR2DS1, KIR2DL2 and KIR2DL3 allotypes that encode that combination are listed. (−) indicates that the combination is not present in the listed gene. (D) Shown is the binding of 6 2DL1-Fc, 6 2DL2-Fc and 6 2DL3-Fc fusion proteins to nine HLA-C1 allotypes (white circles) and 7 HLA-C2 allotypes (black triangles). The residues at positions 16 and 148 were mutated to those listed below each KIR-Fc. The prototypical allotypes of each KIR (KIR2DL1*003, KIR2DL2*001 and KIR2DL3*001 respectively) are indicated with red lettering and grey shading. Binding values are normalized to that of the W6/32 antibody that binds to all HLA class I allotypes with equal avidity. (E) Shown is the mean binding to HLA-C2 and HLA-C1 of every KIR containing arginine at either position 16 or 148 (R), cysteine at either position (C) or proline at either position (P). KIR2DL3-Fc containing either R16 or R148 bound to C1 bearing allotypes with significantly greater avidity than those containing either P16 or P148 (two –tailed t test, p=0.01)

Figure 8. Substitutions in the extracellular binding domains regulate the avidity of KIR2DS1 allotypes for HLA-C2

(A) Shown is an alignment of the Ig-like domains (D1+D2) of four allotypes of KIR2DS1 and four allotypes of KIR2DL1. Dots indicate identity with consensus. The position of the structural domain (D1 or D2) is indicated by a line below the alignment. (B) Binding of four naturally occurring KIR2DS1-Fc and KIR2DL1-Fc fusion proteins to microbeads coated with seven C2 HLA-C allotypes. Binding values are normalized to that of the W6/32 antibody that binds to all HLA class I allotypes with equal avidity.