A natural killer cell receptor takes sharp aim at the world of bacteria

Killer cell Ig-like receptors (KIR) are expressed on natural killer (NK) cells, lymphocytes of innate immunity and reproduction. The principal KIR ligands are the polymorphic HLA-A, HLA-B, and HLA-C glycoproteins of the human major histocompatibility complex. The KIR form 4 phylogenetic lineages. Lineage III KIR are most numerous and include all KIR that recognize epitopes of HLA-C, the most important KIR ligands. Well established is the role of inhibitory lineage III KIR in educating NK cells to be tolerant of healthy cells and ready to kill infected or malignant cells. By comparison, functions for the activating lineage III KIR have proved elusive, particularly for KIR2DS4, the most common activating KIR. As reported in PNAS (1), Sim et al. find KIR2DS4 has a binding site with much narrower specificity for HLA-C and peptide than the inhibitory KIR. In prokaryote genomes, they identify a highly conserved nonamer sequence in bacterial recombinase A. The corresponding peptide binds HLA-C*05:01, forming a ligand that activates human NK cells bearing KIR2DS4. The potential of this ligand−receptor combination is to make NK cell responses against many different bacterial infections. That the KIR2DS4 ligand is now clearly seen to be a combination of HLA-C*05:01 and a common bacterial peptide is a discovery that should encourage and invigorate future research on other orphan KIR.

NK Cells Are Controlled by Receptors for MHC Class I

NK cells were first studied in the context of killing tumor cells that escaped CD8 T cell attack by reducing their surface expression of MHC class I (2). During development, human NK cells acquire activating and inhibitory receptors that recognize variable HLA-A, HLA-B, and HLA-C, and conserved HLA-E. Inhibitory HLA class I receptors educate developing NK cells (3), creating a dynamic intracellular balance between activating and inhibitory signals. This balance ensures that NK cells do not disturb healthy cells but promptly respond to the lower amounts of HLA class I on malignant and infected cells. Human NK cells have 2 unrelated types of inhibitory receptor that recognize HLA class I and educate NK cells in complementary ways. One is the conserved CD94:NKG2A receptor that recognizes conserved HLA-E; the other is the diverse family of KIR that recognize highly variable HLA-A, HLA-B, and HLA-C ligands.

KIR recognition of HLA-C focuses on 2 epitopes defined by dimorphism at positions 77 and 80 of the α₁ domain. Serine 77 and asparagine 80 define the C1 epitope recognized by inhibitory KIR2DL2 and KIR2DL3, whereas asparagine 77 and lysine 80 define the C2 epitope recognized by KIR2DL1, KIR2DS1, and some KIR2DS5 allotypes (4, 5). In the KIR nomenclature, 2D and 3D refer to the number of Ig-like domains. Letters L and S refer to the length of the cytoplasmic tail, which is Long in the inhibitory receptors and Short in the activating receptors.

The C1 and C2 specificities were first distinguished during the 1990s, using cellular assays of NK cell killing (4, 5). Subsequently, an in vitro binding assay was used, in which KIR-Fc fusion proteins were tested for binding to panels of beads, each coated with one HLA-A, HLA-B, or HLA-C allotype (6). Although each bead is coated with one species of HLA, its bound peptides are very heterogeneous, because the HLA class I was isolated from human cells and is therefore loaded with thousands of different peptides. This approach defined the precise HLA class I specificities for KIR2DL1, KIR2DL2, KIR2DL3, and KIR2DS1 and KIR2DS5, but gave no significant reactions with KIR2DS2 and KIR2DS3, and only weak heterogeneous reactions with KIR2DS4. Following these studies in the mid-1990s, KIR2DS2, KIR2DS3, and KIR2DS4 still languished as orphan receptors some 20 y later.

KIR2DS4 Recognizes Complexes of HLA-C*05:01 and a Nonamer Peptide

Sim et al. (1) purified and denatured HLA-C*05:01 to release the bound peptides, which were then separated and sequenced. As the peptides, mostly nonamers, arise from intracellular degradation of human proteins, they are called “self” peptides to distinguish them from peptides derived from microbial pathogens. Aliquots of a cell line deficient in peptide loading were individually incubated with one of 46 synthetic peptides corresponding to peptides isolated from HLA-C*05:01. In each aliquot, HLA-C*05:01 assembled with a different peptide, whereupon the complexes were transported to the cell surface and could be tested for binding to KIR2DS4-Fc. None of the 46 complexes bound to KIR2DS4. However, by mutating residues 7 and 8 of IIDKSGSTV, a self peptide isolated from HLA-C*05:01, Sim et al. (1) obtained strong binding of KIR2DS4-Fc to the P2-AY mutant peptide, in which alanine 7 and tyrosine 8 replace serine 7 and threonine 8 of the natural peptide. Extending the analysis to a total of 61 peptides, the only other peptides to bind HLA-C*05:01 and engage KIR2DS4 were variants of P2-AY in which tyrosine 8 was replaced by another aromatic residue: tryptophan in mutant P2-AW and phenylalanine in mutant P2-AF. The strength of the interaction increased with the size of the side chain at position 8: tryptophan> tyrosine> phenylalanine.

To assess the functional effect of KIR2DS4 interaction with HLA-C*05:01, NK cells were isolated from human blood and incubated with target cells expressing C*05:01 loaded with the P2-AW, P2-AY, or P2-AF peptide. Using flow cytometry, the NK cells were assayed for degranulation, a measure of NK cell activation and its capacity to kill a target cell. Degranulation was observed for the subset of NK cells that express KIR2DS4, but not for the many NK cell subsets that lack KIR2DS4. Other measures of NK cell activation, such as cytokine production, correlate with the extent of degranulation (1). A most unexpected observation is that interaction between peptide-loaded HLA-C*05:01 and KIR2DS4 is, by itself, sufficient to activate NK cells and trigger degranulation. Previously, the general observation was that 2 or more different pairs of target cell ligand and NK cell receptor are required to activate an NK cell (7).

HLA-C*05:01 has the C2 epitope, the ligand for KIR2DL1. In maturing NK cells, the KIR genes are expressed stochastically. For individuals who have HLA-C*05:01, KIR2DL1, and KIR2DS4, one subset of NK cells expresses KIR2DL1 and KIR2DS4, a second subset expresses KIR2DL1 but lacks KIR2DS4, a third subset expresses KIR2DS4 but lacks KIR2DL1, and a fourth subset expresses neither KIR2DL1 nor KIR2DS4. Sim et al. (1) examined the interplay between the KIR2DS4 and KIR2DL1 receptors. When NK

As reported in PNAS, Sim et al. find KIR2DS4 has a binding site with much narrower specificity for HLA-C and peptide than the inhibitory KIR.

cells expressing both KIR2DS4 and KIR2DL1 were cultured with target cells expressing complexes of HLA-C*05:01 and P2-AW, P2-AY, or P2-AF, the inhibitory signals coming from KIR2DL1 overwhelmed the activating signals transduced by KIR2DS4. In contrast, the subset of NK cells that expresses KIR2DS4, but not KIR2DL1, is activated by complexes of these peptides with HLA-C*05:01. Unexpectedly, Sim et al. find that NK cells expressing KIR2DS4, but not KIR2DL1, give a similarly strong response to HLA-C*05:01 and peptide, irrespective of whether the NK cell was educated or not. Such indifference to education fits with the capacity of KIR2DS4 to activate NK cells without the assistance of signals coming from other ligand receptor interactions.

KIR2DS4 Recognizes a Conserved Bacterial Peptide Bound to HLA-C*05:01

Having shown P2-AW was their best peptide antigen for binding to HLAC*05:01 and engaging KIR2DS4, Sim et al. (1) searched among the peptides eluted from HLA-C:05:01 for ones having tryptophan at position 8. Twelve were found, synthesized, and tested for interaction with KIR2DS4. Two formed complexes with HLA-C*05:01 that were recognized by KIR2DS4. The interaction with peptide MSDVQIHWF from the TM9SF4 cargo protein was weak, whereas SNDDKNAWF from E3 ubiquitin ligase conferred a strong interaction with KIR2DS4, comparable to that of P2-AW.

Sim et al. (1) also searched prokaryote genomes for sequences like P2-AW. Striking homology was found with the recombinase A protein of Helicobacter fenneliae. Positions 283 to 291 are identical to the first 8 residues of P2-AW, and the carboxyterminal residue has a conservative substitution. Further analysis concentrated on bacterial recombinases having an acidic residue at position 286, the anchor residue needed for a peptide to bind HLA-C*05:01. This selection identified 14 candidate peptides, which were made and tested for interaction with HLA-C*05:01 and KIR2DS4. The 6 peptides that formed complexes with HLA-C*05:01 have C-terminal phenylalanine, whereas the 8 peptides that did not form complexes have C-terminal tyrosine. The 6 different complexes of HLA-C*05:01 and bacterial peptide have variable affinity for KIR2DS4. By defining residues that correlate with high or low binding to KIR2DS4, and comparing bacterial recombinase sequences, Sim et al. identify >100 bacterial species for which antigen processing in human cells should give a peptide that binds HLA-C*05:01 and is recognized by KIR2DS4.

The work of Sim et al. (1) on KIR2DS4 and NK cell recognition of bacterial antigens is complemented by that of Naiyer et al. (8) on viral antigens. They showed that HLA-C*01:02 presents a conserved peptide of flaviviruses, such as the Hepatitis C, Dengue, Yellow Fever, Zika, and Japanese encephalitis viruses, and the complex activates NK cells that express KIR2DS2. Comparison of 63 flavivirus sequences predicts that 61 of the viruses can give rise to a complex of viral peptide and HLA-C*01:02 that is a ligand for the KIR2DS2 of NK cells.

Population Genetics of KIR2DS4

Although humans and chimpanzees are closely related, KIR2DS4 is the only lineage III KIR they share (Fig. 1). In separating from their common ancestor, the human ancestors experienced a severe population bottleneck that retained only 2 lineage III KIR. One was KIR2DS4, and the other was an inhibitory C1 receptor that gave rise to all modern, human-specific lineage III KIR, before becoming the nonfunctional pseudogene KIR2DP1 (9). Thus, KIR2DS4 is 5 million years older than any other functional human lineage III KIR. This history points to KIR2DS4 having valuable functions that contributed to early human survival and still applies to modern humans. KIR2DS4 protects pregnant women from preeclampsia, a disorder of poor placentation (10). Implied by Sim et al. (1) is that KIR2DS4 protects against infection by a world of bacterial pathogens.

Fig. 1.

The diagram summarizes phylogenetic relationships between human and chimpanzee lineage III KIR. They form 4 clades. Oldest are the chimpanzee C1-specific KIR that have lysine 44. Second oldest are the chimpanzee C2-specific KIR that have methionine 44. Third oldest is the clade containing human and chimpanzee KIR2DS4, as well as 2 C2-specific receptors: one with methionine 44 and the other with glutamate 44. Most recently evolved are all of the 7 other human lineage III KIR, which all derive from a C1-specific progenitor.

Disadvantages also come with KIR2DS4, including promotion of HIV pathogenesis (11). Consequently, all human populations have 2 high-frequency KIR2DS4 alleles: one functional and one not (12). In human populations, Sim et al. (1) find that the frequencies of functional KIR2DS4 and its HLA-C*05:01 ligand are inversely correlated. This balance limits the frequency of individuals who have both KIR2DS4 and HLA-C*05:01 and can use them to respond to bacteria. Such balance can arise because the KIR and HLA loci segregate on different chromosomes, and its likely cause is competing selection pressures imposed by immunity and reproduction.