A KIR2DL5A*001 cDNA clone with a FLAG epitope (DYKDDDDK) inserted between the leader peptide and the D0 Ig-like domain (6) was used to generate, by site-directed mutagenesis, three additional constructs bearing each of the missense mutations that distinguish the KIR2DL5A*001 and *005 primary structures—constructs N152D (152 AAT → GAT, asparagine to aspartate), G174S (GGC → AGC, glycine to serine), and KIR2DL5A*005 (both changes). Plasmids were purified using the EndoFree Plasmid Maxi and Midi Kits (Qiagen, Valencia, CA, USA) and sequenced with the universal primer SP6 and the internal primer R5e61 (5′-gttttggagcttggttcag-3′). Only error-free clones were used for transfection. Transfection experiments were replicated using up to five different plasmid batches per construct to control for random variations in expression attributable to DNA quality.

Human embryonic kidney (HEK)-293T cells were cultured in DMEM (Corning Inc., Corning, NY, USA) supplemented with 2 mM l-glutamine, 1% penicillin–streptomycin, and 10% FBS (Gibco Life Technologies, Carlsbad, CA, USA), and transiently co-transfected by the calcium phosphate method with 5 µg of each KIR2DL5 construct and 0.1 µg of pEGFP-N1 vector (Clontech Laboratories, Palo Alto, CA, USA). Jurkat cells were cultured in RPMI-1640 Glutamax (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10% FBS and 1% penicillin–streptomycin, and transiently transfected with 2 µg of each KIR2DL5 construct along with 0.1 µg of pEGFP-N1, using Solution V and the X-01 program of a Nucleofector I apparatus (Amaxa Biosystems, Cologne, Germany).

Forty-eight hours after transfection, 2 × 10⁵ HEK-293T cells were lysed in 1% Non-idet P-40 Substitute (Sigma-Aldrich, St. Louis, MO, USA), 20 mM Tris–HCl, pH 8, 137 mM NaCl, 2 mM Na₂EDTA lysis buffer containing 1% protease inhibitor cocktail (Roche Diagnostics GmbH, Mannheim, Germany). Proteins were reduced and denatured in 5× Laemmli buffer, run in 10% SDS-PAGE, and blotted to nitrocellulose membranes (iBlot Gel Transfer Stacks Nitrocellulose, Mini, Novex Life Technologies, Grand Island, NY, USA), which were treated for 1 h with blocking buffer (Li-Cor Bioscience, Lincoln, NE, USA). KIR2DL5–FLAG protein bands were detected with 1:1,000 dilution of anti-FLAG mAb M2 (Sigma-Aldrich, St. Louis, MO, USA), and 1:5,000 secondary antibody directed against mouse IgG (IRDye800, Li-Cor Bioscience). Beta-actin was detected, as a positive control, with 1:2,000 rabbit anti-human β-actin antibody (Abcam, Cambridge, UK) and the secondary reagent IRDye700 (Li-Cor Bioscience). Signals were detected with the Odyssey Infrared Imaging System (Li-Cor Bioscience). For N-deglycosylation, 400 µg of total cell lysates were denatured for 10 min at 100°C and treated with 2.5 U of Peptide-N-Glycosidase F (PNGase F, New England Biolabs, Ipswich, MA, USA) for 5 h at 37°C before Western blotting.

PBMCs or IL-2-expanded NK cells from healthy donors of known KIR genotype were incubated at 4°C with the following antibodies: APC anti-CD56 (eBioscience, Inc., San Diego, CA, USA), PE/Cy7 anti-CD3 (Biolegend, San Diego, CA, USA), and either PE anti-KIR2DL5 (UP-R1, Biolegend) or PE anti-KIR2DL2/L3/S2 (DX27, Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). Isotype-matched negative controls were PE IgG1 (clone MOPC-21, Sigma-Aldrich) and PE IgG2a (clone S43.10, Miltenyi). Flow cytometry analysis was performed in MACSQuant Analyzer (Miltenyi) using MACSQuantify software. Percentages of KIR2DL5-positive cells were determined after gating on the CD3⁻CD56⁺ NK-cell population. The genotype of each donor was assessed using previously published polymerase chain reaction primers (14) targeting specific polymorphisms in the KIR2DL5 coding and promoter regions; the latter primers also enabled to rule out potential confounding presence of transcribed KIR2DL5B alleles. This study was carried out in accordance with the recommendations of the Ethical Committee of Clinical Research of Instituto de Investigación Sanitaria Puerta de Hierro, which approved the protocol (270910-258). All participating volunteers provided written informed consent in accordance with the Declaration of Helsinki.

Primary mAbs used for flow cytometry of KIR2DL5-transfectants were anti-FLAG mAb M2 (Sigma-Aldrich), KIR2DL5-specific mAb UP-R1 (6), and mouse IgG1 isotype-matched negative control MOPC-21 (Sigma-Aldrich). PE-conjugated goat anti-mouse IgG + IgM, F(ab′)₂ fragments (Jackson ImmunoResearch, West Grove, PA, USA) were used as secondary Ab. Jurkat and HEK-293T cells were stained 24 and 48 h post-transfection, respectively, and transient surface expression of KIR2DL5–FLAG was determined in Epics XL (Coulter Electronics, Hialeah, FL, USA) or MACSQuant (Miltenyi) flow cytometers, using Expo32 and MACSQuantify software, respectively. EGFP-positive cells ranged ~8–12% (Jurkat) and 30–40% (HEK-293T). Mean fluorescence intensities (MFIs) of different KIR2DL5 constructs, calculated after gating on EGFP⁺ cells, were compared using the paired Student’s t-test. To assess the effect of KIR2DL5 polymorphisms on recognition by mAb UP-R1, we calculated in each experiment the ratio of the MFIs obtained with KIR2DL5A*001 and each transfectant using M2 (anti-FLAG) and UP-R1, and ratios obtained for each mAb were compared using the non-parametric Wilcoxon signed-rank test.

Forty-eight hours after transfection, HEK-293T cells were stained with anti-FLAG M2 mAb, fixed in PBS with 4% paraformaldehyde, and incubated with Alexa546-conjugated anti-mouse IgG (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA) to detect KIR2DL5–FLAG molecules on the surface. Then, cells were treated with 0.3% Triton X-100, and re-incubated with anti-FLAG M2 mAb, followed by Alexa448-labeled anti-mouse IgG (Invitrogen), to detect intracellular KIR2DL5 molecules. After this staining strategy, based on Ref. (21), cells were visualized on poly-l-lysine-coated glass-bottom dishes on a confocal laser-scanning microscope (TCS SP5, Leica Microsystems CMS GmbH, Mannheim, Germany) using Argon (488 nm) and Helium-Neon (543 nm) lasers and a 20×/0.5 lens. Images were acquired using the LAS AF SP5 software (Leica).

We based on the KIR2DL1 X-ray structure (22) (PDB accession code: 1NKR) to predict the structures of the D2 Ig-like domains of KIR2DL5A*001 and *005, using the SWISS-MODEL automated protein structure homology-modeling system (23), accessible via the ExPASy web server.¹ Quality of each model was assessed with the QMEAN Z-score (24), which describes the likelihood that a model quality is comparable to high-resolution experimental structures.² Structures were visualized and edited using the PyMOL Molecular Graphics System (version 1.7.4, Schrödinger, LLC).

Flow cytometry assays of PBMCs from donors with a KIR2DL5A*001 allele identify discrete subpopulations of NK cells reacting with the only available anti-KIR2DL5 mAb, UP-R1 (average of positive NK cells in 11 donors ± SD: 5.18 ± 2.15%; range: 1.71–7.89%), consistently with the clonal expression pattern reported for the receptor (4, 6). In contrast, much fewer UP-R1-positive NK cells are seen in donors carrying allele KIR2DL5A*005 (0.50 ± 0.24%; range: 0.23–0.93%; n = 9; p = 0.0001), their proportions being not significantly different from the background seen in donors lacking a transcribed KIR2DL5 gene (0.33 ± 0.20%; range: 0.15–0.78%; n = 11). In fact, the flow cytometry plots of donors pertaining to the latter two categories are hardly distinguishable on an individual basis (Figure 1), indicating that NK cells of KIR2DL5A*005 subjects are, for some reason, unreactive with UP-R1 in these assays.

KIR2DL5A*001-positive NK cells expanded in vitro express higher levels of the receptor on the surface than resting cells, as it happens with other KIR (Figure 2). In contrast, NK cells expanded from individuals carrying KIR2DL5A*005 appear to upregulate no product detectable with UP-R1, remaining indistinguishable from those of a KIR2DL5A-negative donor (Figure 2). This argues against the possibility of a KIR2DL5A*005 product being expressed at low levels on the surface of freshly isolated NK cells. Given that KIR2DL5A*005 mRNA is as readily detected as that of the common allele KIR2DL5A*001 (14), these results suggest that the putative KIR2DL5A*005 product is not transported to the cell surface or fails to react with mAb UP-R1.

Because of the crucial role of protein posttranslational modifications on several cellular processes, including maturation and protein location, we analyzed whether potential KIR2DL5A posttranslational processing could be predictably affected by its allelic polymorphism. The two substitutions that distinguish the primary structures of KIR2DL5A*001 and *005 are located in the D2 Ig-like domain and involve amino acids with side chains susceptible to glycosylation—asparagine/aspartate-152 and glycine/serine-174. Therefore, we performed a predictive search of possible O- and N-glycosylation motifs in the sequences of their D2 domains. None of the predicted O-glycosylation sites involves KIR2DL5 polymorphic positions (not shown). In contrast, substitution of asparagine-152 by aspartate in KIR2DL5A*005 destroys an N-glycosylation site (Figure 3) highly conserved in all KIR (25), whose loss might affect KIR2DL5A*005 surface expression.

We also explored possible presence in the KIR2DL5A*005 primary structure of sequence motifs that could affect its subcellular location or folding. However, we found no signals for endoplasmic reticulum (ER) retention, such as di-basic (i.e., arginine/lysine) motifs, or variants of an H/KDEL sequence seen in ER-resident proteins (26). Furthermore, KIR2DL5A*005, like allele *001, has intact WSXPS motifs in each of its Ig-like domains, mutations of which have been reported to hamper correct folding and surface expression of the KIR3DL1*004 allele (27), as well as cytokine receptors (28).

To reliably track the expression of KIR2DL5A*001 and *005, we tagged cDNA constructs of these alleles with a FLAG epitope which should enable us to detect their products independently of UP-R1 recognition. Such constructs were transfected into HEK-293T cells, and expression was verified by western blot with an anti-FLAG mAb. As shown in Figure 4A, specific bands of 50–60 kDa were seen in both KIR2DL5A*001 and *005-transfected cells, in comparison with mock-transfected HEK-293T. Differences in the relative mobility of KIR2DL5A*001 and *005 were observed, and they are attributable to their differential N-glycosylation, since they disappear after treatment of the cell lysates with PNGase F (Figure 4B). Also, due to variable transfection efficacy, the relative amounts of each synthesized protein varied between experiments (e.g., experiments 1 and 2 of Figure 4A). To compensate for such variations, transfections were replicated six times using different plasmid batches, EGFP was co-transfected in all experiments, and KIR2DL5 expression was evaluated on EGFP⁺ cells.

Flow cytometry analysis of KIR2DL5A-transfected HEK-293T cells with anti-FLAG mAb revealed discrete subpopulations of cells showing specific surface expression of both KIR2DL5A*001 and *005 (Figures 5A,B). Density of the former molecule was, however, consistently and significantly higher than that of the minor allele KIR2DL5A*005 (average MFI of six experiments ± SD: 38 ± 11 vs. 9 ± 8; p = 0.0004). Similarly, flow cytometry with the KIR2DL5-specific mAb UP-R1 showed that KIR2DL5A*001-transfected cells are stained more intensely than those expressing KIR2DL5A*005 (MFI 47 ± 16 vs. 4 ± 6; p = 0.0001, Figures 5C,D). These results roughly replicate the observed natural behavior of these alleles, except for the fact that KIR2DL5A*005 is only detectable on transfected cells. Furthermore, comparison of the staining profiles obtained with anti-FLAG M2 and anti-KIR2DL5 UP-R1 mAbs showed that, despite following similar trends, differences in apparent expression levels of KIR2DL5A*001 and *005 were somewhat more pronounced when cells were stained with UP-R1 mAb (Figure 5). For instance, the MFI values of KIR2DL5*001 cells were 5.1- to 75.0-fold higher than those of *005 ones when measured with UP-R1, but only 2.0- to 21.6-fold higher when assessed with the anti-FLAG mAb, the difference being statistically significant (p < 0.05) and consistently seen in every experiment. This behavior suggests that KIR2DL5A*005 detection by UP-R1 is limited, not only by lower expression of the molecule but also by poorer recognition by the antibody.

To assess the relative contribution of the KIR2DL5A*005 polymorphisms to reduced surface expression and UP-R1 mAb recognition of this allele, we introduced separately the G174S and N152D mutations into the KIR2DL5A*001 coding region, and analyzed expression on transfected HEK-293T cells as above, using anti-FLAG and anti-KIR2DL5 mAbs.

Examination of the mutants flow cytometry profiles with the anti-FLAG antibody (Figures 5A,B) showed that substitution of aspartate for asparagine-152 induced only a modest effect on KIR2DL5A expression levels (MFI: 26 ± 11 vs. 38 ± 11 in KIR2DL5A*001, p = 0.0695). In contrast, the serine for glycine-174 mutant displayed a significant reduction in surface expression in comparison with the wild-type allele (MFI: 18 ± 10 vs. 38 ± 11, p = 0.0013). Western blot analyses of the two mutants revealed comparable levels of KIR2DL5 protein synthesis (not shown).

Analysis of expression with the KIR2DL5-specific mAb UP-R1 replicated the results obtained with the anti-FLAG antibody, except for the facts that both mutants differed significantly from KIR2DL5A*001 (Figure 5D); and that, as it happened in the comparison of KIR2DL5A*001 and *005, the decrease in recognition of the G174S mutant was more pronounced with UP-R1 than with the anti-FLAG antibody (2DL5A*001/G174S MFI ratios: UP-R1, 2.7–29.5; anti-FLAG, 1.4–4.6, p < 0.05). In fact, G174S mutants showed an average MFI similar to that of the low-expressed allele KIR2DL5A*005 (8 ± 11 vs. 4 ± 6), supporting that serine-174 hampers recognition by the KIR2DL5-specific antibody besides reducing expression levels.

To further validate our findings, we replicated the flow cytometry analysis of each KIR2DL5 construct after nucleofection into Jurkat, a T-cell line previously used in studies of KIR expression (27, 29, 30). In line with the less effective transfection of Jurkat, expression levels of all KIR2DL5 constructs emulated more closely those seen in NK cells of KIR2DL5A*001- and *005-positive subjects, being 2- to 10-fold lower than those obtained with HEK-293T cells. Despite this difference, the relative expression of each construct in comparison with KIR2DL5A*001, determined with either UP-R1 or anti-FLAG mAbs, roughly reproduced that previously obtained in HEK-293T cells; i.e., percentages of cells with appreciable receptor density on the surface were conspicuously lower for KIR2DL5A*005 and the G174S mutant than for KIR2DL5A*001, whereas the N152D mutation had only a minor effect on KIR2DL5 expression and recognition by UP-R1 (Figure 6).

Our previous results show that KIR2DL5A*005 translation is not followed by efficacious expression on the cell surface. To assess whether the KIR2DL5A*005 product is retained intracellularly, we performed confocal microscopy experiments of HEK-293T cells transfected with FLAG-tagged constructs of alleles KIR2DL5A*001 and *005, and with the above described intermediate mutants N152D and G174S.

As shown in Figure 7, remarkable differences were seen in the cellular distribution of the four transfected constructs, as assessed with an anti-FLAG mAb. In cells expressing KIR2DL5A*001 or the N152D mutant, fluorescence was localized on the plasma membrane, without appreciable intracellular staining. In contrast, KIR2DL5A*005, as well as mutant G174S, were predominantly detected inside transfected cells, despite a small amount of molecules colocalizing with the plasma membrane, in agreement with the flow cytometry results. Unfortunately, KIR2DL5-specific mAb UP-R1 behaved poorly after cell fixation–permeabilization, which precluded subsequent studies of intracellular staining with this mAb to track KIR2DL5A*005 localization in NK cells expressing naturally this molecule.

To gain insight on the mechanism by which a serine for glycine-174 change in the D2 Ig-like domain could hinder KIR2DL5A surface expression, we analyzed this substitution in a predicted three-dimensional model of the receptor. To that end, and since the crystallographic structure of KIR2DL5 has not yet been solved, we modeled its D2 domain on the structure of KIR2DL1 (31), which, among all KIR, shares with KIR2DL5 the highest amino acid sequence identity in that region (86.73% with no gaps). QMEAN6 Z-scores, estimating the absolute quality of the KIR2DL5*001 and *005 models, were 0.25 and −0.43 SDs, respectively, from the KIR2DL1 experimental structure. KIR2DL5 residue 152 (Asn/Asp) is exposed on the protein surface (segment connecting β-strands D and E), while amino acid 174 (Gly/Ser) is partially hidden inside the receptor architecture (strand F), confronted with phenylalanine-129 (loop 5, near strand C). Replacement of glycine-174 in the KIR2DL5A*001 model by serine of KIR2DL5A*005 induces displacement of phenylalanine-129 (Figure 8), as a consequence of the bigger size of the serine hydroxymethyl group in comparison with the hydrogen side chain of glycine-174. Of note, a similar displacement of KIR2DL1 tyrosine-134 (paralogous of KIR2DL5 Phe-129, Figure 3), resulting in intracellular retention, has been attributed to an identical Gly-to-Ser substitution in allele KIR2DL1*014 residue 179 (32), paralogous of KIR2DL5 amino acid 174. KIR2DL5 substitutions G174S and N152D appear to affect no other bonds or interactions between surrounding amino acids, and we found no additional clues to the KIR2DL5A*005 behavior by modeling on KIR2DL4 (PDB accession code 3WYR, results not shown), receptor that, in contrast with KIR2DL1, shares with KIR2DL5 the D0–D2 Ig-like domain organization, but slightly lower percentage of identity in the primary structure.