The Protein Made from a Common Allele of KIR3DL1 (3DL1*004) Is Poorly Expressed at Cell Surfaces due to Substitution at Positions 86 in Ig Domain 0 and 182 in Ig Domain 1

KIR3DL1 constructs

The 3DL1*002 and 3DL1*004 were isolated from PBMCs of different individuals. RNA was obtained using RNA-Bee (Tel-Test, Friendswood, TX), as per manufacturer’s instructions, and cDNA was synthesized using an oligo(dT) primer. PCR was conducted with primers that contained NotI and XbaI restriction endonuclease sites for subcloning: sense primer, K3D5ut2-NotI, 5′-ccgaatgcggccgcaccggcagcaccatgt-3′, and the antisense primer, K3D3ut-XbaI, 5′-ccgctctagagargagcgatsccctaaga-3′, in which the underlined sequences correspond to the restriction endonuclease sites. The K3D3ut-XbaI is a degenerate primer that was designed to pick up all 3DL1 alleles; the r (adenosine or guanosine) and s (cytidine or guanosine) correspond to the International Union of Pure and Applied Chemistry/International Union of Biochemistry and Molecular Biology codes. PCR products were cloned into TOPO TA cloning kit (Invitrogen, San Diego, CA) and sequenced using big dye terminators and automated sequencing on an ABI 377 instrument (Perkin-Elmer/Applied Biosystems, Foster City, CA). Error-free clones were digested with the restriction endonucleases NotI and XbaI and subcloned into the expression vector pcDEF3 (43).

A recombinant amplification was used to generate 3DL1*002 and 3DL1*004 FLAG-tagged constructs. The FLAG epitope, having the sequence DYKDDDDK (single amino acid code), was inserted between the leader peptide and the D0 domain of each variant. The PCR for the leader-FLAG sequence was performed with the sense primer K3D5ut2-NotI and the antisense primer FLAG-R, 5′-cttatcgtcgtcatccttgtaatctggaccggccctctggaccaa-3′, in which the underlined sequence corresponds to the FLAG epitope. The PCR for the FLAG-D0 sequences used sense primers FLAG-F, 5′-gattacaaggatgacgacgataagcacgtgggtggtcaggacaa-3′, for the 3DL1*002 allele, and FLAG-F.2, 5′-gattacaaggatgacgacgataagcacatgggtggtcaggacaa-3, for 3DL1*004, and the antisense primer 3DL1-Eco47IIIb, 5′-caccacagcgctgggccagg-3′, in which the sequence underlined is the restriction site for the endonuclease Eco47III. The leader-FLAG and FLAG-D0 amplicons were then used together as a template in a recombinant PCR using the primers K3D5ut2-NotI and the 3DL1-Eco47IIIb. The resulting recombinant leader-FLAG-D0 amplicon and the 3DL1*002 and 3DL1*004 constructs were digested with the NotI and Eco47III. The leader-FLAG-D0-digested amplicon was ligated into the 3DL1*002- and 3DL1*004-digested constructs.

Reciprocal recombinant chimeric constructs of 3DL1*002 and 3DL1*004 were made by a recombinant amplification of the region encoding the leader-D0-D1 domains of one allele, with the region encoding the D2-transmembrane-cytoplasmic domains of the second allele. The amplification for the leader-D0-D1 region was performed with the K3D5ut2-NotI sense primer and the antisense primer Re5–3g, 5′-caggacaaggtcacgctctc-3′, that primes in exon 5. The D2-transmembrane-cytoplasmic region was amplified with the sense primer Fe4–3e, 5′-acatcgtggtcacaggtcc-3′, that primes in exon 4, and the antisense primer K3D3ut-XbaI. Recombinant PCR was conducted using the two amplicons as template using the K3D5ut2-NotI sense primer and the antisense primer K3D3ut-XbaI. The two reciprocal recombinant products were cloned into the TOPO 4.0 vector (Invitrogen), and multiple clones were sequenced. Error-free clones were identified, digested with NotI and XbaI, and subcloned into the pcDEF3 expression vector.

A similar recombinant PCR approach was used to make chimeric constructs of KIR3DL1 and enhanced green fluorescent protein (EGFP), in which EGFP was attached to the C terminus of KIR3DL1. The amplification of KIR3DL1 was performed with the sense K3D5ut2-NotI primer and the antisense 3DL1-EGFP-3 primer, 5′- tcctcgcccttgctcaccattgggcaggagacaactttgg-3′, that overlap the 5′ end of EGFP; and for EGFP the sense 3DL1-EGFP-5 primer, 5′-ccaaagttgtctcctgcccaatggtgagcaagggcgagga-3′, that overlaps the 3′ end of the KIR3DL1, and the antisense GFP-XbaI-R primer, 5′-ccgctctagaccgctttacttgtacagctc-3′, in which the underlined sequence corresponds to the XbaI restriction endonuclease site. The template for the EGFP amplification was the pEGFP-N3 vector (Clontech Laboratories, Palo Alto, CA). A recombinant amplification was performed using the KIR3DL1 and EGFP amplicons as a template with the K3D5ut2-NotI and GFP-Xba-R primers and cloned into the TOPO 4.0 vector. KIR3DL1-EGFP error-free clones were digested with NotI and XbaI restriction endonucleases and subcloned into the pcDEF3 expression vector.

Point mutations in KIR3DL1-FLAG and KIR3DL1-EGFPconstructs were generated using the Quick Change mutagenesis kit (Stratagene, La Jolla, CA) and oligonucleotide primers containing the relevant mutations, per the manufacturer’s protocol.

Transfection

Transfection of Jurkat cells was performed by electroporation using an ECM-830 square pulse electroporator (BTX Instruments Division, Holliston, MA). Six micrograms of each KIR3DL1 and KIR3DL1-FLAG construct were cotransfected with 1 μg of the EGFP-N3 vector into 10⁷ Jurkat cells with two pulses of 130 V, 20 ms, and 1-s gap, using a 2-mm cuvette (BTX). Six micrograms of each KIR3DL1-EGFP construct were transfected into 10⁷ Jurkat cells using the same parameters described above. The HEK293T cell line was transfected with 1 μg of each DNA construct by the Lipofectamine method (Life Technologies, Grand Island, NY), per the manufacturer’s instructions. All DNA constructs used for transfection were prepared using the EndoFree Plasmid Maxi Kit (Qiagen, Valencia, CA) and sequenced, and only error-free clones were used for transfections.

Cell lines, Abs, and flow cytometric analysis

The human T cell line Jurkat was cultured in RPMI 1640 supplemented with 10% (v/v) bovine calf serum. HEK293T cells were cultured in DMEM supplemented with 10% (v/v) bovine calf serum. mAbs used in this study were: DX9-PE (a gift from L. Lanier, Cancer Research Center, University of California, San Francisco, CA) and Z27-PE (Beckman-Coulter-Immunotech, Brea, CA), both specific for KIR3DL1; 5.133, specific for KIR3DL1 and KIR3DL2 (a gift from M. Colonna, Washington University School of Medicine, St. Louis, MO); and the anti-FLAG M2 (Sigma-Aldrich, St. Louis, MO). F(ab′)₂ from goat anti-mouse PE (Beckman-Coulter-Immunotech) were also used. Intracellular staining was conducted with the Cytofix/Cytoperm kit (BD PharMingen, San Diego, CA). Flow cytometric analysis was performed on a FACScan flow cytometer using CellQuest software (BD Biosciences, San Jose, CA).

Data analysis

For the KIR3DL1 and KIR3DL1-FLAG experiments, a cotransfection with the pEGFP-N3 vector (Clontech Laboratories) was performed to maximize the readout of the transient transfection experiments. Thus, after 48 h of transfection, flow cytometric analysis of cell surface expression was conducted by gating on the EGFP-positive cells (usually ∼20–40% of total cells).

For the KIR3DL1-EGFP experiments, the EGFP fluorescence was used to normalize the levels of cell surface expression data obtained with monoclonal anti-KIR3DL1 Abs. The geometric mean fluorescence intensity (MFI) was used for comparing the flow cytometric analysis data, as it is a more accurate measure for log-amplified data. As the amount of green fluorescence correlates with the total translation of the chimeric protein in each transfection, we could then calculate the cell surface expression of KIR3DL1 relative to EGFP as the ratio between the MFI of the cell surface-staining anti-KIR3DL1 Ab and the MFI of the EGFP.

Confocal microscopy

Cells were grown in a Lab-Tek II Chamber Coverglass System (Nalge Nunc International, Naperville, IL) for 24 h and assessed in a confocal laser-scanning microscope. Imaging was performed at the Cell Sciences Imaging Facility (Stanford University) on the Zeiss LSM 510 confocal laser-scanning microscope (Zeiss, Thornwood, NY) using a 488-nm argon laser and a Zeiss Plan-Apo ×63 oil immersion lens (NA 1.4 mm). Staining of the cell membrane was performed with the DiD lipophilic carbocyanide tracer (Molecular Probes, Eugene, OR), per the manufacturer’s instructions.

KIR3DL1 structure model

Prediction of the KIR3D structure was performed using the Swiss model automated protein structure homology-modeling server (44) accessible via the ExPASy web server (http://www.expasy.org/swissmod/SWISS-MODEL.html). The amino acid sequence of 3DL1*002 (accession U31416) was used for generation of the three-dimensional model of D1 and D2 domains. The D0 domain was modeled separately and added to the D1-D2 three-dimensional model using the Swiss-Pdb (Protein Data Bank) viewer computer program.

The 3DL1*004 protein is made, but is poorly expressed at the cell surface

Constructs containing the coding sequence of 3DL1*004 (no DX9-binding allele) or 3DL1*002 (high DX9-binding allele) were transiently transfected into the human T cell line Jurkat. The cells were cotransfected with a limiting amount of a vector encoding the EGFP, and flow cytometric analysis for KIR3DL1 expression was confined to the population of cells expressing EGFP. As expected, this analysis showed that Jurkat cells transfected with 3DL1*002 bound the DX9 Ab to a high level, whereas cells transfected with 3DL1*004 did not bind the DX9 Ab (Fig. 1⇓A). Similar results were obtained with two other mAbs: Z27, which is specific for KIR3DL1, and 5.133, which recognizes both KIR3DL1 and KIR3DL2 (data not shown). The phenotypes of the Jurkat cells transfected with 3DL1*002 and 3DL1*004 reproduce those observed for NK and T cells isolated from the peripheral blood of donors having these KIR3DL1 alleles (16). Furthermore, to be able to measure small differences on KIR3DL1 cell surface expression, we prepared constructs encoding chimeric proteins in which EGFP was attached to the C terminus of KIR3DL1 and only cells expressing the EGFP were studied. Jurkat cells stably transfected with the 3DL1*002-EGFP construct gave high binding with the DX9 Ab (mean of three experiments; mean fluorescence intensity (MFI) = 603 ± 17; data not shown) compared with untransfected Jurkat cells (mean of three experiments; MFI = 12.4 ± 0.4). Strikingly, cells transfected with the 3DL1*004-EGFP construct showed a poor, but statistically significant expression (mean of three experiments; MFI = 14.5 ± 0.3) (Fig. 1⇓B), which was 0.34% of that for 3DL1*002-GFP. This experiment demonstrates directly that 3DL1*004 gives rise to the no-binding phenotype seen for PBLs (16).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Flow cytometry analysis of Jurkat cells after 48 h of transfection with 3DL1*002 or 3DL1*004 constructs. To maximize the readout of these experiments, we cotransfected each construct with the pEGFP-N3 vector. Only cells that were positive for the EGFP fluorescence were examined. The negative control samples were Jurkat cells transfected with the pEGFP-N3 vector alone. A, Transient transfection of 3DL1*002 and 3DL1*004 into the Jurkat cells, followed by gating on the EGFP-positive cells, demonstrated that most of the cells transfected with 3DL1*002 expressed 3DL1*002 on the cell surface, whereas the cells transfected with 3DL1*004 did not express 3DL1*004 at the cell surface, as detected by DX9 binding. A stable transfectant for a chimeric construct containing EGFP (3DL1*004-EGFP) was compared for DX9 binding to untransfected Jurkat cells (mean of three experiments; MFI = 12.4 ± 0.4) (B); the results from three different experiments showed that the low cell surface expression of 3DL1*004-EGFP detected (mean of three experiments; MFI = 14.5 ± 0.3) was significant (p < 0.002 by Student’s t test); error bars depict the SE of these three experiments. FLAG-tagged constructs (3DL1*002F and 3DL1*004F) transfected into Jurkat cells were analyzed for binding with the anti-FLAG M2 Ab inside the cell (C) or at the cell surface (D).

Two possible mechanisms could explain the cell surface phenotype associated with the 3DL1*004 allele: first, the 3DL1*004 variant lacks the epitopes recognized by the three mAbs; second, expression of the 3DL1*004 allele does not lead to expression of a 3DL1*004 protein on the cell surface. To help differentiate between these possibilities, constructs of 3DL1*002 and 3DL1*004 were made in which a sequence encoding the 8-aa FLAG epitope was added between the sequences encoding the leader sequence and the D0 domain. This results in the FLAG epitope being attached to the N terminus of the mature KIR3DL1 protein. Transfectants containing these constructs were then analyzed for expression of both the FLAG epitope and the epitope recognized by DX9. Expression was independently assessed at the cell surface with anti-KIR3DL1 and anti-FLAG Abs and inside the cell with only the anti-FLAG Ab (Fig. 1⇑). (Several attempts to perform intracellular staining with anti-KIR3DL1 Abs failed to give staining above the background.) In cells expressing the 3DL1*002F construct, the anti-FLAG Ab stained intracellular and cell surface molecules. This control showed that the FLAG epitope did not grossly interfere with the assembly and structure of 3DL1*002F. In cells expressing the 3DL1*004F construct, the anti-FLAG Ab gave intracellular staining (Fig. 1⇑C), but no cell surface reaction (Fig. 1⇑D). These results indicated that 3DL1*004F was transcribed and translated in the transfected cells, but that the protein made did not reach the surface in detectable amount.

A potential complication to this interpretation would be if 3DL1*004F had premature termination in the translation of 3DL1*004F molecules. To address this possibility, we analyzed the chimeric proteins in which EGFP was attached to the C terminus of KIR3DL1, and FLAG was not present. In cells transfected with these constructs, the detection of EGFP is contingent on complete translation of the associated KIR3DL1. Examination by confocal microscopy of the fluorescence due to EGFP in the transfectants showed that cells expressing 3DL1*004-EGFP or 3DL1*002-EGFP contained comparable amounts of EGFP. Striking, however, were the differences in the cellular distribution of EGFP molecules in the two transfectants. In cells expressing 3DL1*004-EGFP, the fluorescence localized in the perinuclear region, whereas in cells expressing 3DL1*002-EGFP it was found also at the plasma membrane (Fig. 2⇓). These data further support the interpretation that the 3DL1*004 variant is poorly expressed on the cell surface, retained within the cell, and thus barely detectable with anti-KIR3DL1 Abs.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Confocal microscopy shows that 3DL1*004 is retained inside the cell with a perinuclear distribution. Confocal microscopy of the midsection of Jurkat cells stably transfected with the chimeric constructs of 3DL1*002-EGFP (A and B) or 3DL1*004-EGFP (C and D) (left panels). Both transfectants were stained for the plasma membrane with the lipophilic carbocyanide DiD (middle panels). Merged images show colocalization of KIR3DL1-EGFP and the plasma membrane for 3DL1*002, but not for 3DL1*004 (right panels).

The human embryonic kidney cell line 293T was transfected with the 3DL1*002-EGFP and 3DL1*004-EGFP constructs. Flow cytometric analysis of the transient transfectants gave similar data to those obtained from the corresponding Jurkat transfectants (data not shown). Thus, the intracellular confinement of 3DL1*004 is not a special property of lymphocytes.

The unusual properties of 3DL1*004 are due to residues in the extracellular D0-D1 domains

The 3DL1*002 and 3DL1*004 differ by 12-aa substitutions, which are spread throughout the sequence (Fig. 3⇓). As a first step to define which of these substitutions are responsible for the phenotype of 3DL1*004, a pair of reciprocal recombination mutants was constructed in which the D0 and D1 domains of one allotype are associated with the D2, stem, transmembrane region, and cytoplasmic domains of the other allotype. These mutants were made from the FLAG-tagged 3DL1*002F and 3DL1*004F constructs (Fig. 4⇓A). The mutant constructs were transfected into Jurkat cells, and cell surface and intracellular expression of KIR3DL1 was assessed by flow cytometric analysis using the anti-FLAG Ab. As expected, both recombinant molecules were detected with intracellular staining. With cell surface staining, the recombinant having the D2, stem, transmembrane region, and cytoplasmic domains of 3DL1*004 was detectable, whereas the recombinant having the D0 and D1 domains of 3DL1*004 was not (Fig. 4⇓B). These results demonstrate that the intracellular confinement of 3DL1*004 is due to one or more of the 7-aa substitutions that distinguish its D0 + D1 domains from those of 3DL1*002.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

The amino acid substitutions and levels of DX9 binding that distinguish KIR3DL1 allotypes. All KIR3DL1 allotypes are compared with the consensus sequence (3DL1*001). Only the positions of difference are shown. Dashes indicate amino acid identity. The allotypes 3DL1*004 and 3DL1*002 compared in this investigation are shown shaded in gray.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

The amino acid substitutions that distinguish the cell surface phenotypes of 3DL1*002 and 3DL1*004 localize to the extracellular D0 and D1 domains. Reciprocal recombinant mutants were made between 3DL1*002 and 3DL1*004 in which the recombination was made between the D1 and D2 domains (A). The recombinants were transfected into Jurkat cells and tested by flow cytometry for surface expression (B, left panel) and intracellular expression (B, right panel) with the anti-FLAG Ab. The negative control samples were Jurkat cells transfected with the pEGFP-N3 vector alone.

Leucine 86 in the D0 domain is the principal factor preventing cell surface expression of 3DL1*004

At positions 44 and 86 in the D0 domain, 3DL1*004 has residues (glycine and leucine, respectively) that are unique to this allotype, making them the best candidates for preventing cell surface expression. Also implicated is serine 182 in the D1 domain, which is found only in 3DL1*004 and 3DL1*005, an allotype with low DX9 binding (16). The other four positions of difference (residues 2, 31, 47, and 54) were considered poor candidates because in each case the residue present in 3DL1*004 is also present in one or more of the high DX9-binding allotypes. We therefore chose to construct and examine the properties of mutants having substitution at residues 44, 86, and 182.

For each of these positions, the residue in 3DL1*004F was replaced by the corresponding residue in 3DL1*002F and vice versa. The mutants were transfected into Jurkat cells and analyzed by flow cytometry with anti-KIR3DL1 and anti-FLAG Abs. The most dramatic effects were seen for the point substitutions at position 86 in the D0 domain. When leucine was replaced by serine at position 86 in 3DL1*004F, high levels of the mutant 3DL1*004-86S protein were detected at the cell surface with anti-KIR3DL1 (Fig. 5⇓) and anti-FLAG Abs (data not shown). Conversely, changing leucine for serine at position 86 was the only substitution in 3DL1*002F that severely prevented cell surface expression of 3DL1*002F (Fig. 5⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Amino acid substitution at position 86 is the principal factor that distinguishes the cell surface phenotype of 3DL1*004 from 3DL1*002. Three point mutants each of 3DL1*002 and 3DL1*004 were made in which the amino acids at positions 44, 86, and 182 were swapped. The six mutants and the two parental alleles were transfected into Jurkat cells and analyzed for binding with the DX9 Ab. The two top panels compare the cell surface expression of the parental KIR3DL1 alleles (WT for wild type), while the three panels below compare the mutants at positions 44, 86, and 182.

The serine present at position 86 in all KIR3DL1 allotypes, except 3DL1*004, is part of a 5-aa motif, WSXPS, of which variants are found in homologous position in all KIR Ig-like domains (Fig. 6⇓). A similar motif, WSXWS, is a characteristic of the hemopoietic receptor family. The three-dimensional structures of KIR and hemopoietic receptors are related, and these motifs form a similar β-bulge in which the two serines form hydrogen bonds to the adjacent β-strand. This arrangement distinguishes these Ig-like domains from the more familiar situation in which hydrogen bonds are made between β-strand backbones (9, 45). In 3DL1*004, the first serine in the WSXPS motif is replaced by leucine, which cannot hydrogen bond. Systematic mutation of the first serine of the WSXWS motif in the erythropoietin receptor caused the erythropoietin receptor to be retained within the cell, except when the substituted residue was threonine or glycine (46). By analogy, it seems likely that leucine 86 in 3DL1*004 prevents normal folding of the molecule leading to its retention within the cell. In this context, 3DL1*004 would be considered an allele encoding a nonfunctional receptor. Further emphasizing the functional significance of the substitution of leucine for serine at position 86 was the finding that point mutations made at 17 other positions within the D0 domain of KIR3DL1 had no qualitative effect upon either cell surface expression or interaction with HLA-B ligand (11). Serine 86 is not essential for binding anti-KIR3DL1 Abs because Rojo et al. (10) showed that a soluble KIR3DL1 mutant that lacked D0 bound DX9 with a strength about half that of a soluble KIR3DL1 retaining D0.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Leucine 86 of 3DL1*004 corrupts the WSAPS motif of the D0 domain. A model of the three-dimensional structure of KIR3D is shown. The WSXPS-like motifs in the D0, D1, and D2 domains are indicated. The table compares the WSXPS-like motifs for 3DL1*002 and 3DL1*004.

Serine 182 in the D1 domain also contributes to the unusual 3DL1*004 phenotype

Comparison of the binding of anti-KIR3DL1 Abs to Jurkat transfectants expressing 3DL1*002, 3DL1*002F, and 3DL1*002-EGFP showed that presence of the FLAG sequence at the N terminus of 3DL1*002 reduced expression and/or Ab binding, whereas presence of EGFP at the C terminus had no effect. To better quantify the differences in expression of mutant KIR3DL1, we therefore chose to use KIR3DL1-EGFP chimeras. Such chimeras were made for the six swap point mutations at positions 44, 86, and 182. (To shorten the name of the EGFP mutants, we will refer to them henceforth as 3DL1*002 or 3DL1*004, plus the amino acid change.) For transfectants expressing these constructs, the EGFP fluorescence was used to normalize the level of binding obtained with Abs DX9 and 5.133 (Fig. 7⇓). This method of analysis confirmed the importance of the substitution at position 86. It also revealed an additional smaller effect due to the polymorphism at position 182 in the D1 domain. Thus, mutant 3DL1*004–182P has a small increase of Ab binding compared with 3DL1*004, and mutant 3DL1*002–182S has Ab binding reduced by about one-half compared with 3DL1*002. Glycine at position 44 does not contribute to the lack of cell surface expression of 3DL1*004. In fact, introduction of glycine 44 slightly increased the cell surface expression of 3DL1*002, and arginine at this position in 3DL1*004 had no effect.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Normalized comparison of the levels of cell surface expression of 3DL1*002, 3DL1*004, and 10 point mutants. In addition to the six swap point mutants described in Fig. 5⇑, point mutants of 3DL1*002 and 3DL1*004 having alanine at positions 86 and 182 were analyzed. The parental alleles and the point mutants were transfected into Jurkat cells and analyzed by flow cytometry for binding the DX9 (left panel) and 5.133 (right panel) Abs. The relative cell surface expression of each mutant is shown. Values are the mean of the normalized geometric mean fluorescence intensities for the 5.133 and DX9 Abs of four independent experiments, and the error bars depict the SE of these four experiments.

The effects of substitution at positions 86 and 182 were further studied using mutants having alanine substitutions at these positions (Fig. 7⇑). Alanine at position 86 in either 3DL1*002 or 3DL1*004 prevented cell surface expression, indicating that serine at this position is necessary for cell surface expression and that the intracellular retention of 3DL1*004 caused by leucine at this position was not specific to this amino acid residue. Alanine at position 182 decreased the expression of 3DL1*002 to an extent similar to that caused by serine substitution at this position. Moreover, alanine at position 182 in 3DL1*004 produced no change in cell surface expression. These results show that presence of proline at position 182 is therefore a second factor that contributes to the normal level of cell surface expression of 3DL1*002. Implied by these observations is that the low DX9-binding phenotype of the 3DL1*005 allotype (16) is due, at least in part, to the presence of serine 182.

Polymorphism at positions 44, 86, and 182 makes independent, additive contributions to cell surface expression of KIR3DL1

To assess how polymorphism at positions 44, 86, and 182 interacts in determining the level of cell surface expression of KIR3DL1, six mutants were made that represent all pairwise combinations of the residues found in 3DL1*002 and 3DL1*004. Thus, three double mutants were made from the EGFP chimera of each allele. The double mutants were transfected into Jurkat cells, and their levels of expression, as measured by Ab binding normalized to EGFP fluorescence, were compared with those of the parent wild-type molecules and the single mutants (Fig. 8⇓).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 8.

Substitution at position 86 in the D0 domain and 182 in the D1 domain makes separate and additive contribution to the difference in the 3DL1*002 and 3DL1*004 cell surface phenotype. For positions 44, 86, and 182, eight double swap mutants were made. In each mutant, two of the three positions were changed to that of the other allotype. Jurkat cells were transfected with each mutant and analyzed for binding to the DX9 (left panels) and 5.133 (right panels) Abs. The levels of cell surface expression are compared with the parental allotypes and the point mutants in three combinations. Comparison of position 86 and 182 mutations (A), comparison of position 44 and 86 mutations (B), and comparison of position 44 and 182 mutations (C). Values are the mean of the normalized geometric mean fluorescence intensities for the 5.133 and DX9 Abs of four independent experiments, and the error bars depict the SE of these four experiments.

Importantly, surface expression of double mutant 3DL1*004-86S-182P was higher than that of either single mutant, 3DL1*004-86S and 3DL1*004-182P, being 2-fold higher than the latter and greater than that of 3DL1*002. Reciprocally, the surface expression of double mutant 3DL1*002-86L-182S was reduced below the level obtained by either single mutant (3DL1*002-86L and 3DL1*002-182S), and was at background level like 3DL1*004 (Fig. 8⇑A). This experiment shows that polymorphism at residues 86 and 182 makes independent and additive contributions to the differences in cell surface expression of the 3DL1*002 and 3DL1*004 allotypes.

This analysis also revealed small, but reproducible effects due to the polymorphism at position 44. For this position, glycine, which is naturally present in 3DL1*004, increases cell surface expression when introduced into 3DL1*002, as can be seen in both single and double mutants (Fig. 8⇑, B and C). Conversely, arginine, which is naturally present in 3DL1*002, is seen to decrease cell surface expression of the 3DL1*004 double mutants 3DL1*004-44R-86S and 3DL1*004-44R-182P compared with their respective single mutants 3DL1*004-86S and 3DL1*004-182P (Fig. 8⇑, B and C). The results for all the constructs are accountable in terms of consistent and additive contributions from the alternative substitutions at positions 44, 86, and 182. That the data obtained with each Ab leads to the same interpretation strengthens that interpretation, as it does not depend upon the properties of a single Ab.

Consistent differences in the reactivity of the anti-KIR3DL1 Ab (DX9) and the anti-KIR3DL1 and KIR3DL2 Ab (5.133) were observed. Overall, the reactions, relative to EGFP fluorescence, obtained with 5.133 were stronger than those of DX9 (Figs. 7⇑ and 8⇑). In addition, the two Abs showed differential sensitivities to mutation. For example, 5.133 reacted equivalently with 3DL1*004-86S and 3DL1*002, whereas DX9 bound more strongly to the latter (Fig. 8⇑A). These observations are consistent with the DX9 and 5.133 Abs having different affinities and distinctive target epitopes on the KIR3DL1 molecules.

Concluding remarks

This study has shown that the 3DL1*004 allele is transcribed and translated to give normal levels of protein. What distinguishes the product of the 3DL1*004 allele is that the protein is sequestered within the cell and is barely present at the cell surface. This unusual property can be accounted for by a combination of two substitutions that distinguish 3DL1*004 from 3DL1*002 and other allotypes that are well expressed at the cell surface. Of these substitutions, leucine 86 in the D0 domain makes the major contribution to the 3DL1*004 phenotype, and serine 182 in the D1 domain makes the minor contribution. Leucine 86 corrupts the WSAPS motif found in the D0 domain of all other KIR, a substitution that when present in a related protein, the erythropoeitin receptor, prevented proper folding (46). Serine 182 is also close to the homologous LSAPS motif of the D1 domain. These observations are consistent with a model in which 3DL1*004 fails to fold properly and is retained in the endoplasmic reticulum with little or no expression at the cell surface.

A general feature of the polymorphism at the KIR3DL1 locus is to produce different levels of cell surface expression of the protein (16, 42). Although functional differences have not been associated with these differences, they have the capacity to affect the strength of the inhibitory signal generated by engagement of KIR3DL1 with a Bw4⁺ HLA-B ligand. We show in this study that 3DL1*004 is the most extreme example of a KIR3DL1 allotype with low cell surface expression, and it clearly is different in function from 3DL1*002.

That 3DL1*004 is at a frequency of ∼20% in the Caucasian population points to this allele having been the target for natural selection. One possibility is that it performs its function within the cell, in contrast to other KIR3DL1 allotypes. Alternatively, 3DL1*004 could represent an inactivated form of a KIR3DL1 allotype that was expressed at the cell surface and became the target for subversion by a viral ligand (47). By this means, the virus could have inhibited the antiviral NK cell response. In this circumstance, there would be competitive advantage to a variant allele whose protein was sequestered inside the cell and unable to interact with a viral protein expressed on the surface of infected cells.