A Rare Transporter Associated with Antigen Processing Polymorphism Overpresented in HLA^low Colon Cancer Reveals the Functional Significance of the Signature Domain in Antigen Processing

MHC class I molecules are ligands for CTLs on the surface of most tissue cells and are important for the immune recognition of intracellular pathogens and cancer cells. The transporter associated with antigen processing (TAP), a heterodimer of TAP-1 and TAP-2, plays an essential role in the MHC class I–restricted antigen processing and presentation pathway. TAP translocates antigenic peptides ranging from 8 to 15 residues across the membrane of the endoplasmic reticulum. These peptides are typically generated in the cytosol by the 20S proteasome. Upon being translocated into the endoplasmic reticulum lumen, some of the peptides bind to MHC class I heavy chain/β2-microglobulin (β₂M) heterodimers with the help of a peptide-loading complex consisting of tapasin, ERp57, and calreticulin (reviewed in refs. 1, 2). Recent studies by several groups also provided convincing evidence that through the fusion of phagosome and endoplasmic reticulum, exogenous antigens can be presented on MHC class I molecules by professional antigen presenting cells in a TAP-dependent manner (3–5).

TAP-1 and TAP-2 belong to the ATP-binding cassette superfamily, which is comprised of a diverse class of proteins that carry various substances, ranging from ions to proteins, across cellular membranes (6, 7). The nucleotide-binding domains of the ATP-binding cassette transporters are structurally related to nontransporter enzymes that are involved in DNA repair or structural maintenance of chromosomes (6–8). Defects in several ATP-binding cassette transporter members result in human diseases. Mutations in the cystic fibrosis transmembrane conductance regulator, a chloride channel located in the apical membrane of epithelial cells, cause cystic fibrosis in humans (6). Overexpression of P-glycoprotein or multiple-drug resistance protein, a transporter that actively extrudes cytotoxic drugs to maintain low intracellular levels, leads to multiple-drug resistance of certain cancer cells (6). A third medically important ATP-binding cassette transporter is the TAP transporter, a disruption of which results in a lack of cell surface MHC class I expression (human type 1 bare lymphocyte syndrome; ref. 9). Disruption of TAP activity in cancer cells also leads to class I down-regulation on the surface, which facilitates immune evasion (10).

A prototype ATP-binding cassette transporter is composed of two paired membrane integral domains that are thought to form the translocation pore and two paired nucleotide-binding domains that serve as an ATPase to fuel the translocation process. For some ATP-binding cassette transporters, such as cystic fibrosis transmembrane conductance regulator, these domains are fused as a single polypeptide chain, although for others, like TAP, two polypeptides are separately encoded, each containing one transmembrane domain and one nucleotide-binding domain. Members of the ATP-binding cassette transporter superfamily contain several conserved sequence motifs in their nucleotide-binding domains, including the Walker A and Walker B sequences, the signature motif or C loop, Q loop, and D loop. With the exception of the Walker A and B motifs, which are required for ATP binding and ATP hydrolysis, the roles of the other conserved motifs in the ATP-binding cassette transporter function are largely unknown. The well-conserved signature motif with the consensus LSGGQ sequence, usually followed by a basic amino acid R/K, is the most intriguing, because it is conserved in all ATP-binding cassette transporter family members, but is not found outside this superfamily. The importance of the signature motif in human disease is shown by the fact that mutations in this motif lead to loss or impairment of the transporter function of human cystic fibrosis transmembrane conductance regulator (11–14).

Functional abnormalities in TAP have been documented in a variety of cancers and correlate to their clinical prognosis (10). The molecular mechanisms responsible for the defects include depressed expression (15, 16), mutations (17, 18), and increased degradation (18). In addition to these somatic alterations, genetic predisposition due to TAP-1 polymorphism in the nucleotide-binding domain distinct from the signature domain was likely responsible for depressed antigen processing of a lung cancer cell line (19). To address whether germ line or somatic variation in the signature domain contributes to the HLA class I antigen down-regulation frequently observed in cancer cells, we examined the DNA fragments encompassing this region from colon cancer samples. Here we reported the increased frequency of a rare TAP-1 polymorphism among cancer samples that had lower HLA class I antigen expression. Our results showed that this polymorphism and the surrounding cystic fibrosis–like mutations severely impaired the ability of TAP-1 to restore surface HLA expression. These results suggested that a TAP-1 polymorphism adjacent to the signature motif may be a contributing factor for MHC class I down-regulation in colon cancer and that point mutations at or near the signature domain modulate antigen processing.

Isolation of genomic DNA and genotyping. The genomic DNA from the blood of normal population was extracted as described elsewhere (20) and then subjected to the RFLP analysis as below. Tumor tissues were collected either by laser capture microdissection or manually using sterile tweezers. DNA was then isolated using a lysis buffer containing 1× high-fidelity PCR buffer (Roche Diagnostics Corporation, Indianapolis, IN), 1% Tween 20 and 4 mg/mL proteinase K and incubated at 55°C for 72 hours. For the blood DNA samples, RFLP analysis was done. The forward primer, E10 Ahd.f 5′-ACCGTTCTCATCTTGGCCCTTTGCTCTG-3′, generated an AhdI site in the G allele but not in the A allele. Together with the reverse primer, E10 Ahd.r 5′-ATCAATGCTCGGGCCAACGCGACTGCCT-3′, TAP-1 exon 10 was amplified from the genomic DNA. PCR products were then separated by gel electrophoresis and purified using QIAGEN gel extraction kit (Qiagen, Inc., Valencia, CA). The purified PCR products were incubated with AhdI (New England Biolabs, Inc., Beverly, MA) at 37°C overnight and then separated in 5% agarose gel. For cancer DNA samples, nested PCR was designed using the following primers to amplify the exon 10 region of the TAP-1 gene: ouE10.F 5′-GTTCTCATCTTGGCCCTTTGCTCTG-3′ and inE10.R1: 5′-AGAAGATGACTGCCTCACCTGTAAC-3′ for the first-round amplification; E10.F 5′-CCTTTGCTCTGCAGAGGTAGACGAG-3′ and inE10.R2 5′-TGCCTCACCTGTAACTGGCTGTTTG-3′ for the second round. The final PCR products were purified with QIAGEN PCR purification kit (Qiagen) before sequencing.

Cell culture and flow cytometry. Human melanoma cell line SK-MEL-19 was cultured as described (18). For induction of MHC class I expression, the cells were cultured in DMEM supplemented with 10% heat-inactivated FCS, 2 mmol/L l-glutamine, 100 IU/mL penicillin, 100 μg/mL streptomycin (Invitrogen, Inc., Carlsbad, CA) and 1,000 IU/mL recombinant human IFNγ (R&D Systems, Inc., Minneapolis, MN). MHC class I expression on the cell surface of all cell lines was examined by flow cytometry using a phycoerythrin-conjugated anti-HLA-A, B, and C antibody (BD PharMingen, San Diego, CA), as previously described (18). Phycoerythrin-conjugated mouse IgG₁ (BD PharMingen) was used as isotype control.

Generation of TAP-1 variants and stable SK-MEL-19 transfectant cell lines. The cloning of the wild-type TAP-1 cDNA was previously described (18). Constructs for all the TAP-1 variants were made by overlapping PCR and expressed in the pcDNA3.1/Hyg(+) vector (Invitrogen). Nucleotide changes are summarized in Table 1. To ensure that any functional defects of the TAP-1 mutants were due to the mutations we generated, all the constructs were sequenced before further analyses. Stable cell clones were established as previously described (18). Expression of TAP-1 protein was examined by Western blotting using a rabbit anti-human TAP-1 peptide antibody (735-748; Calbiochem, EMD Biosciences, Inc., San Diego, CA). Cell clones expressing TAP-1 protein were selected for flow cytometric analysis of cell surface MHC class I antigen expression.

Immunofluorescence microscopy. Cells were plated on coverslips in a 24-well plate and were allowed to grow overnight. Cells were fixed for 15 minutes in a freshly prepared solution of 4% paraformaldehyde in PBS (pH 7.4), and then permeabilized in washing buffer (PBS with 0.2% Triton X-100) for 5 minutes at room temperature. After three washes, the primary antibodies, rabbit anti-human TAP-1 (Calbiochem), and mouse anti-human calnexin (Affinity Bioreagents, Golden, CO), which were diluted in 200 μL of dilution buffer (PBS containing 3% bovine serum and 0.2% Triton X-100) were added. After incubation for 1 hour at room temperature, coverslips were washed thrice in washing buffer and then incubated with the secondary antibodies, fluorescein-conjugated donkey anti-rabbit immunoglobulin (Amersham Biosciences, Inc., Piscataway, NJ) for TAP-1 and Texas red dye-linked goat anti-mouse immunoglobulin (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) for calnexin, for 1 hour at room temperature. After three washes, the coverslips were dipped in water and then mounted on glass slides in the mounting media with 4′,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA).

Immunoprecipitation. Immunoprecipitation procedures which have been previously described (21) were followed. Cells were extracted for 30 minutes on ice at 2 × 10⁶ cells per milliliter in 1% digitonin (Sigma-Aldrich, Corp., St. Louis, MO) in TBS [10 mmol/L Tris, 150 mmol/L NaCl (pH 7.4)] containing 0.5 mmol/L phenylmethylsulfonyl fluoride (Sigma), 0.1 mmol/L N-tosyl-l-lysine chloromethylketone (Sigma), and 5 mmol/L iodoacetamide (Sigma). Postnuclear supernatant was precleared overnight at 4°C with 5 μL normal rabbit serum (Invitrogen) and 50 μL Zysorbin (Zymed Laboratories, Inc., San Francisco, CA) per milliliter of extract. Aliquots were then incubated with the appropriate antibodies, R.RING4C, a rabbit anti-peptide antibody to the C-terminus of TAP-1 (a gift from Dr. Peter Cresswell; ref. 22), W6/32, a monoclonal antibody (mAb) detecting the β₂M-associated class I heavy chain and an anti-human TAP-2 mAb (BD PharMingen), and protein A-Sepharose for 1 hour each at 4°C. Protein G-Sepharose (Amersham) was used with anti-TAP-2 mAb and W6/32. The immunoprecipitates were washed twice with TBS, 0.1% digitonin and once with distilled water before being separated by SDS-PAGE. TAP-1, TAP-2, and tapasin were detected by Western blotting using the rabbit anti-human TAP-1 peptide antibody (735-748, Calbiochem), the anti-human TAP-2 mAb (BD PharMingen), and R.gp48N, a polyclonal rabbit antibody against the N-terminal regions of tapasin (a gift from Dr. Peter Cresswell; ref. 22). The same antibodies were used for the subsequent Western blot analyses.

Microsome-based peptide translocation assay. Microsomes were prepared as described previously (23). Briefly, 1 × 10⁸ SK-MEL-19 cells that were stably transfected with wild-type TAP-1, TAP-1 G646D, TAP-1 R648Q, and pcDNA3.1/Hyg vector, respectively, were harvested. Cells were washed once in ice-cold PBS and then resuspended in 800 μL of ice-cold cavitation buffer [250 mmol/L sucrose, 25 mmol/L potassium acetate, 5 mmol/L magnesium acetate, 0.5 mmol/L calcium acetate, 50 mmol/L Tris (pH 7.4)] supplemented with a protease inhibitor mixture (Sigma). Cells were lysed by repeatedly drawing the suspension through a 26-gauge needle at least 10 times. After centrifugation at ∼500 × g for 5 minutes, the supernatant was thoroughly mixed with 5 mL of 2.5 mol/L sucrose in gradient buffer [2.5 mol/L sucrose, 150 mmol/L potassium acetate, 5 mmol/L magnesium acetate, 50 mmol/L Tris (pH 7.4)], which was then overlaid with 3 mL each of 2.0 mol/L sucrose in gradient buffer, then 1.3 mol/L sucrose in gradient buffer, and finally with 800 μL of cavitation buffer. This was centrifuged overnight at 80,000 × g, and the microsomal fraction at the interface of the 2.0 and 1.3 mol/L sucrose layers was collected. This fraction was diluted into 5 mL of PBS, 1 mmol/L DTT, and centrifuged for 1 hour at 100,000 × g. The pellet was resuspended in 100 μL of PBS, 1 mmol/L DTT and frozen in aliquots at −70°C. The total protein content was determined by a bicinchoninic acid assay (Pierce Biotechnology, Inc., Rockford, IL). The expressions of TAP-1 and TAP-2 proteins were examined by Western blot. Translocation assays were carried out as previously described (23). An iodinated model peptide, RRYNASTEL, with a specific activity of 100 to 150 μCi/μg was used for the study. Briefly, 30 μg of microsomes containing wild-type TAP-1, TAP-1 G646D, TAP-1 R648Q, or pcDNA3.1/Hyg vector were added to 150 μL of assay buffer [PBS, 0.1% bovine serum albumin, 1 mmol/L DTT (pH 7.3)] containing 10 mmol/L MgCl₂ and 5 mmol/L ATP (+ATP samples) and then incubated with the radioiodinated RRYNASTEL peptide at 37°C for 15 minutes. The samples were then centrifuged at 4°C at 8,800 × g and washed once with 250 μL of the assay buffer. The pellets were subsequently resuspended in 250 μL of lysis buffer [50 mmol/L Tris, 150 mmol/L NaCl, 1% Nonidet P-40 (pH 7.4)] and incubated on ice for 1 hour. After centrifugation at 8,800 × g for 5 minutes at 4°C, the supernatants were transferred to concanavalin A-Sepharose (Amersham) beads and incubated for 2 hours at 4°C with agitation. Beads were washed twice with the assay buffer before the radioactivity was determined using a Beckman 5500 gamma counter.

Colon cancer tissue sections, immunohistochemistry, and statistical analysis. Colon cancer tissue microarray slides were acquired from the National Cancer Institute and Zymed Laboratories. All specimens contained in the array were formalin-fixed and paraffin-embedded. Mouse mAb HC-10 that detects HLA-B and C heavy chain was used in immunohistochemical staining. Tissue samples were graded as MHC class I negative when at least 75% of the tumor cells were not stained or weakly stained. The comparisons of the allelic frequencies were tested using a Fisher exact test. The P values are reported.

Increase of the G1943A allele frequency in HLA^low colon cancer tissues. Down-regulation of cell surface MHC class I expression is one of the strategies tumor cells use to evade immune surveillance by CD8+ T lymphocytes (10, 15), which also poses a major obstacle for T cell–based cancer immunotherapy (24). Disruption of TAP is one of the reasons for the decreased surface MHC class I level that has been frequently encountered in a variety of tumors (10, 18, 19, 25). The signature motif is one of the most conserved motifs in the nucleotide-binding domains of all the ATP-binding cassette family members (examples shown in Fig. 1). To test if functional polymorphisms exist in the signature motif and its surrounding conserved amino acids, tissue microarrays were acquired from the National Cancer Institute and Zymed Laboratories. Individual cancer tissues were collected by laser capture microdissection or manually for DNA extraction. The exon 10 region of the TAP-1 gene encompassing the signature motif was amplified by nested PCR, for which forward primers were designed to reside in intron 9 to avoid contamination from the TAP-1 cDNA constructs. All PCR products were fully sequenced to identify the mutations or polymorphisms in TAP-1 signature motif. One single nucleotide polymorphism G1943A (protein R648Q) was observed in this region of TAP-1.

All the cancer samples were also analyzed for their MHC class I expression by immunohistochemistry using mAb HC-10, which recognizes the heavy chain of HLA-B and C. The standard by which we graded a sample as MHC class I–negative was that at least 75% of tumor cells showed a severely decreased level of MHC class I compared with the normal tissues. Typical patterns of HC-10 staining for MHC class I–positive and negative samples of colon cancer are shown in Fig. 2A-C, respectively. We compared HLA+ and HLA− colon cancer for the allelic frequency of G1943A among the 103 cases of colon cancer in our collection. We found that 88 were MHC class I–positive and 15 were negative, which was similar to the frequency reported previously (26, 27). The frequency of the G1943A allele occurring in the MHC class I–negative samples (10.00%, 3 out of 30) was much higher than that in the positive group (0.57%, 1 out of 176), and the difference was statistically very significant (P = 0.01; Table 2).

Function of TAP-1 R648Q. We tested the function of the protein product of the G1943A TAP-1 allele, R648Q. By comparing the nucleotide-binding domain sequences of ATP-binding cassette transporters (Fig. 1A) and that of TAP-1 from different species (Fig. 1B), we found that either lysine or arginine, both being basic amino acids, are present at position 648 in most cases. Because the residue is just next to the essential LSGGQ/E sequence, we hypothesized that the replacement of the arginine residue by glutamine at position 648 may affect the TAP function. We have previously reported that the SK-MEL-19 melanoma cell line is HLA class I–deficient due to a deficiency in the expression of TAP-1 mRNA even after IFN-γ stimulation. We have identified a single-nucleotide deletion at position +1,489 of the TAP-1 gene that leads to rapid degradation of TAP-1 mRNA (18). We cloned the TAP-1 R648Q by site-directed mutagenesis and transfected it into the SK-MEL-19 cells. Three stable cell clones that expressed comparable levels of TAP-1 protein to the wild-type clone were examined for their surface MHC class I by flow cytometry. As shown in Fig. 3A, transfection of wild-type TAP-1 full-length cDNA reconstituted the HLA class I expression in SK-MEL-19 cells. Western blot showed that TAP-1 and TAP-2 proteins were not expressed in vector-transfected SK-MEL-19 cells, whereas transfection of either wild-type TAP-1 or R648Q restored the protein expression of TAP-1 and also stabilized TAP-2 protein as a result (Fig. 3B). However, even with similar amounts of protein expression of TAP-1 and TAP-2, the SK-MEL-19 cells reconstituted with R648Q TAP-1 only up-regulated the HLA class I expression on the surface to about 28% of that with the wild-type (Fig. 3A and C). We previously have shown that IFN-γ treatment drastically increased the mRNA expression of β₂M, HLA heavy chain, TAP-2, LMP-2, and LMP-7, but not TAP-1 in SK-MEL-19 cells (18). Here we treated the cells with IFN-γ to increase the expression of other components of the HLA class I antigen presentation pathway. We still observed significant defects in HLA class I up-regulation by TAP-1 R648Q (Fig. 3A (bottom) and C).

To test the transporter function of R648Q TAP-1, we prepared microsomes from two of the three stable cell clones expressing R648Q TAP-1. Microsomes containing the wild-type TAP-1 and the cloning vector only were isolated to serve as positive and negative controls, respectively. Wild-type TAP-1 had a ratio of average cpm_+ATP/cpm_−ATP of about 10 (Fig. 3D). The microsomes expressing the vector alone did not show a significant increase of signal when ATP was included in the assays. As expected, the R648Q TAP-1-expressing microsomes did transport the peptide, but with much reduced efficiency. The two microsome preparations we used exhibited an average of 19.81% of transporter activity relative to microsomes containing the wild-type TAP-1. The reduction of TAP activity was similar between two different clones, and these results were reproducible in two independent translocation analyses. We concluded that peptide translocation by R648Q TAP-1 was impaired and resulted in the reduced cell surface expression of MHC class I molecules.

Cystic fibrosis–like mutations in TAP-1 signature motif abrogate TAP-1 activity. To substantiate the functional significance of the TAP-1 signature motif, we introduced the following mutations into the signature motif of TAP-1: S644R, G645R, G646S, and G646D (Table 1) by site-directed mutagenesis. Each of these mutations on cystic fibrosis transmembrane conductance regulators has been shown to cause cystic fibrosis (28). In order to characterize the function of the TAP-1 mutants, we reconstituted the SK-MEL-19 cells, which lack surface MHC class I expression unless a functional TAP-1 is introduced (18). After the stable cell clones were established, the expression of TAP-1 proteins was examined by Western blot using a rabbit anti-human TAP-1 antibody. Cell clones that expressed mutant protein at levels greater than or comparable to wild-type TAP-1 proteins were selected (Fig. 4C) and their cell-surface MHC class I levels were evaluated by flow cytometry. As reported previously, vector-transfected cells expressed little cell surface HLA and wild-type TAP-1 restored HLA expression (Fig. 4A). Cells transfected with mutant TAP-1 constructs had reduced restoration of cell surface HLA, between 1% and 20% compared with cells that expressed comparable levels of wild-type TAP-1 (Fig. 4B and C). These results show that mutations analogous to those that inactivate cystic fibrosis transmembrane conductance regulator function also have a detrimental effect on TAP-1 function.

TAP-1 G646D is expressed in the endoplasmic reticulum and remains associated with the peptide-loading complex. To define the intracellular localization of TAP-1 mutants, we stained stably transfected cells with an antiserum against TAP-1 and an endoplasmic reticulum–resident protein, calnexin, and analyzed the stained cells by immunofluorescence microscopy. As shown in Fig. 5A, a perinuclear reticular staining pattern characteristic of the endoplasmic reticulum was observed in cells expressing both the wild-type and the G646D mutant TAP-1 protein. In addition, both the wild-type and G646D TAP-1 were colocalized with calnexin (red), as shown by the yellow fluorescence in the merged images.

It is known that antigenic peptides transported by the TAP transporters bind to the empty heterodimer of MHC class I heavy chain and β₂M in the lumen of the endoplasmic reticulum with the help of several chaperone proteins, such as ERp57 and calreticulin. Bridged by tapasin, MHC class I molecules associate with the TAP transporter to form the peptide-loading complex (1, 2). We tested whether the G646D mutation in the TAP-1 protein disrupted the formation of interactions within the peptide-loading complex. Immunoprecipitation assays were done using the rabbit anti-TAP-1 antiserum, a monoclonal anti-TAP-2 antibody, and W6/32, a mAb recognizing β₂M-associated MHC class I heavy chains. The precipitated lysates were then analyzed by Western blot to detect TAP-1, TAP-2, and tapasin. As shown in Fig. 5B, the G646D mutation did not interfere with the formation of the tapasin/TAP complex (left and middle). Both wild-type TAP-1 and G646D TAP-1 were detectable in immunoprecipitates with W6/32, indicating that the G646D mutation also did not interfere with the class I/TAP interaction (Fig. 5B, right). The thiol-reductase ERp57 was also detected by sequencing after we separated the cell lysates precipitated by the antiserum against TAP-1 in a two-dimensional SDS-PAGE (data not shown).

It was noticed that without TAP-1 proteins in the cell, TAP-2 proteins could not be detected in the lysates of vector-transfected clones that were precipitated by the antibodies against TAP-1 or TAP-2 (Fig. 5B, left and middle). Only a weak band was observed when the lysates pulled down by W6/32 mAb were blotted with anti-TAP-2 mAb after long-time exposure (data not shown). Previously, we have shown that the TAP-2 mRNA level in the SK-MEL-19 cells was greatly enhanced by IFN-γ (18). However, after IFN-γ treatment at 1,000 units/mL for 48 hours, the TAP-2 protein level was only slightly increased (data not shown). Because the TAP-2 protein levels are greatly increased and support the surface MHC class I expression when TAP-1 is expressed, it seems that the accumulation of TAP-2 protein is TAP-1-dependent.

TAP-1 G646D failed to transport peptide across the endoplasmic reticulum membrane. We then evaluated the transporter function of G646D TAP-1 by a microsome-based peptide translocation experiment described previously (29). Microsomes were prepared from stable cell clones expressing the wild-type TAP-1, G646D TAP-1, and vector control, respectively. Each microsome preparation was quantified for total protein and examined for TAP-1/2 protein expression by immunoblotting (Fig. 6A). As shown in Fig. 6B, microsomes containing the wild-type TAP-1 had a ratio of average cpm_+ATP/cpm_−ATP at about 10, which indicated a successful specific peptide-translocation. The microsomes expressing the G646D TAP-1 or vector only did not show a significant increase in signal when ATP was present in the assays. We concluded that the mutant G646D TAP-1 had lost transporter function completely.

It has been shown that the ATP-binding cassette signature motif and the immediate COOH-terminal amino acids are well conserved among all the ATP-binding cassette transporter family members and its disruption results in the loss of protein function (11, 13, 14, 30). As a first step to determine whether the polymorphism of signature motif region of TAP-1 contributes to variations in TAP-1function, we compared the HLA^low and HLA^high colon cancer samples for the DNA sequence in exon 10 which encodes this region. We observed a 17.5-fold increase in the frequency of a rare allele G1943A (R648Q) among the colon cancer samples that showed reduced cell surface HLA. Functional analysis suggests that TAP-1 encoded by this allele has only 19.81% of transporter activity compared with the wild-type. The surface MHC class I expression restored by this TAP-1 variant was about 30% of the wild-type. Several other single nucleotide polymorphisms in TAP-1, such as I333V and D637G, have been described with varying degrees of association with autoimmune diseases (31–33). However, functional significance of these single nucleotide polymorphisms remains unclear. Our study provides the first link on TAP-1 polymorphism and its functional significance in deficient antigen presentation in human cancer.

Although the significant increase in frequency of the defective G1943A (R648Q) allele among the HLA class I antigen^low colon cancer samples strongly suggests a role of genetic polymorphism of TAP-1 in HLA expression, several caveats deserve consideration. First, when overexpressed, this allele can restore cell surface HLA to about 30% of what was induced by wild-type TAP-1. One may thus wonder whether cells expressing this allele alone may show severe reduction in cell surface HLA. However, it should be pointed out that in our assay, the expression of the variant TAP-1 protein was driven by a cytomegalovirus promoter and consequently overexpressed. In fact, in a previous study (19), no cell surface HLA was detectable in a small cell lung cancer cell line due to selective expression of another TAP-1 variant, R659Q, which was later shown to have as much as 50% of wild-type transporter activity (34).

Based on the G1943A (R648Q) allele frequency derived from the normal population, which is 3.01% in our screening of 166 normal blood donors (data not shown) and 2.6% reported by another study (35), the chance of a homozygous G1943A (R648Q) being identified is <0.1% according to the Hardy-Weinberg equilibrium, which explains the absence of homozygous A allele in our samples. It is not clear how the defective G1943A (R648Q) allele in a heterozygous individual could contribute to MHC class I down-regulation in cancer cells. One potential mechanism would be the loss of heterozygosity (36). Loss of heterozygosity has been found to contribute to β₂M deficiency in melanoma (37) and HLA haplotype loss in colorectal and laryngeal carcinomas (38). Because both alleles were detected in DNA of cancer cells in all cases we studied, loss of heterozygosity is unlikely to have been responsible for defective HLA expression in the colon cancer studied here. Alternatively, it is known that in some tumor cells, only one allele of TAP-1 gene is transcribed. It has been reported previously that a defective TAP-1 allele, but not the wild-type one, was expressed in a human small cell lung cancer cell line, which led to loss of cell surface MHC class I (19). Because the cancer samples used for the current studies were not suitable for RNA analysis, this possibility remains to be tested.

The functional analysis of the polymorphism also suggests that the signature motif likely serves an important function in antigen processing. Recently, one study showed that simultaneous replacement of the conserved serine and second glycine in the signature motif of TAP-1 with alanines resulted in inactivation of TAP-1 transporter function (39), which is also consistent with the essential function of TAP-1 signature motif. Our mutation analyses of the signature domain extend these observations in two ways. First, because our mutants are single amino acid changes, this can be achieved by point mutations. As such, it is much more likely that the function of the signature domain can be modulated through naturally occurring mutations. Second, the mutations we introduced are analogous to those found in inactivating mutations of the cystic fibrosis patients. As such, it is likely that the function of the signature domain is conserved among the ATP-binding cassette family of transporters (40).

Additional studies are required to elucidate the precise roles of the signature motif residues during peptide translocation. The crystal structures of the ATP-binding cassette transporter MJ0796 and the DNA repair enzyme Rad50 (which contains a nucleotide-binding domain that is structurally related to ATP-binding cassette transporter nucleotide-binding domain) show that the signature motif of one nucleotide-binding domain interacts with the Walker A region of a second nucleotide-binding domain in a dimeric arrangement (8, 41). Specific hydrogen bonds are formed between the serine and the second glycine residues of the signature motif and the phosphate oxygens of ATP (8, 41). In this way, the signature motif may not only mediate the interaction between the two nucleotide-binding domains but also act as a sensor for the hydrolysis of ATP γ-phosphate and mediate conformational changes driven by ATP binding and hydrolysis during the substrate translocation process (8, 42, 43).

Loss of cell surface HLA has been reported among 15% of patients with colon cancer (26, 27, 44). The mechanisms described include loss or mutations in β₂M genes (45, 46), loss of HLA heavy chain genes, selective lack of expression of HLA alleles, and regulatory defects in HLA expression including loss of expression of TAP (44). Our data represent the first report of a genetic predisposition among colon cancers with HLA loss. Moreover, variations in or near the signature domain can modulate the function of TAP-1. Given the widespread defects in DNA mismatch repairs in colon cancer (47–49), it would be of great interest to test whether point mutations in the signature domains may account for antigen-presentation defects of colon cancer patients, particularly those with genetic defects in DNA repair.

The authors thank Dr. Soledad Fernandez for the statistical analysis and Lynde Shaw for secretarial assistance; Dr. Peter Cresswell (Yale University, New Haven, CT) for providing R.RING4C rabbit polyclonal antibody against TAP-1 polypeptide and R.gp48N rabbit polyclonal antibody against tapasin; Dr. Kurtis Yearsly for his assistance with laser capture microdissection and Paul Rangel for his help with DNA sequencing; the University of Michigan Department of Reproductive Sciences for the peptide iodination, and the University of Michigan Biomedical Research Core Facilities for peptide synthesis.