CD83 Gene Polymorphisms Increase Susceptibility to Human Invasive Cervical Cancer

Research Article

Zhengyan Zhang,1 Ingrid Borecki,2 Loan Nguyen,1 Duanduan Ma,2 Kimberly Smith,1 Phyllis C. Huettner,3 David G. Mutch,1 Thomas J. Herzog,1 Randall K. Gibb,1 Matthew A. Powell,1 Perry W. Grigsby,4 L. Stewart Massad,5 Enrique Hernandez,6 Patricia L. Judson,7 Elizabeth M. Swisher,8 Sara Crowder,9 Jianduan Li,1 Daniela S. Gerhard,2 and Janet S. Rader1

Departments of 1Obstetrics and Gynecology, 2Genetics, 3Pathology and Immunology, and 4Radiation Oncology, Washington University School of Medicine, St. Louis, Missouri; 5Department of Obstetrics and Gynecology, Southern Illinois University School of Medicine, Springfield, Illinois; 6Department of Obstetrics, Gynecology and Reproductive Sciences, Temple University School of Medicine, Philadelphia, Pennsylvania; 7Department of Obstetrics, Gynecology, and Women’s Health, University of Minnesota Medical School, Minneapolis, Minnesota; 8Department of Obstetrics and Gynecology, University of Washington, Seattle, Washington; and 9Mid-Missouri Gynecologic Oncology, Columbia, Missouri

Abstract itself because an estimated 30% to 60% of sexually active women We previously mapped a nonrandom frequent loss of are infected with genital HPV yet do not develop cancer (1). The 10-year cumulative incidence rates of immediate cancer precursor heterozygosity (LOH) region in cervical cancers to 1 Mb of 6p23. Here, we describe the identification of a novel cervical cervical intraepithelial neoplasia 3 (CIN3) or invasive cervical cancer susceptibility gene, CD83. The gene was identified by cancer (ICC) are 17.2% for women infected with HPV16 and 13.6% several complementary approaches, including a family-based for those infected with HPV18 (2). Furthermore, the transition from association study, comparison of transcript expression in CIN to invasive cancer has a long latency period, and only a normal and cancerous tissue, and genomic sequencing of minority of those infected go on to develop ICC. Moreover, a large candidate. CD83 encodes an inducible glycoprotein in the proportion of women clear HPV infection (3). Therefore, identifying immunoglobulin superfamily and is a marker for mature nonviral factors that permit the virus to cause cancer is important dendritic cells. The association study that includes 377 family for understanding viral-associated malignancies. trios showed that five single nucleotide polymorphisms (SNP) Compelling evidence has implicated genetic factors in susceptibility to CIN3 and ICC. Using the Swedish national registers, within 8 kb of its 3¶-end showed significant allelic association that was strengthened in a subgroup of women with invasive Magnusson et al. found a significant familial clustering of the cancers infected by high-risk human papillomavirus type 16 disease. The consistent pattern of decreasing familial relative risk and 18 (rs9296925, P = 0.0193; rs853360, P = 0.0035; rs9230, with decreasing degree of genetic relationship pointed to genes as P = 0.0011; rs9370729, P = 0.0012; rs750749, P = 0.0133). the major cause of familial aggregation of cervical cancer (4). Of the Investigation of CD83 uncovered three alternative transcripts several susceptibility markers identified, the association of HLA in cervical tissue and cell lines, with variant 3 (lacking exons 3 alleles with increased risk of CIN3 or ICC is most often reported (5, 6). A recent report of genomic mutations in the genes TMC6 and and 4) being more frequent in cervical cancer than in normal TMC8 cervical epithelium (P = 0.0181). Genomic sequencing on 36 in patients with epidermodysplasia verruciformis, a rare paired normal and cervical tumors revealed several somatic dermatosis associated with a high risk of HPV5-related skin cancer, mutations and novel SNPs in the promoter, exons, and introns highlights the importance of genetics in progressive viral- of CD83. LOH was confirmed in >90% of cervical cancer associated cancers (7). specimens. Immunofluorescence colocalized CD83 protein to Candidate genes for carcinogenesis can be identified in areas of nonrandom chromosomal aberrations. We and others have the Golgi apparatus and cell membrane of cervical cancer cell lines. None of seven nearby genes was differentially expressed identified distinct chromosomal alterations on chromosomes 3p, in cervical cancer. The importance of CD83 in epithelial versus 6p, and 11q linked to cervical carcinogenesis using methods such dendritic cells needs to be determined, as does its role in as loss of heterozygosity (LOH), comparative genomic hybridiza- promoting cervical cancer. [Cancer Res 2007;67(23):11202–8] tion, and cDNA expression microarrays (8–12). Loss of 6p23 is an early event, which is detected in CIN (10, 13). To identify causative candidate genes within 6p23, we first Introduction identified new protein-coding segments within this region by Human papillomavirus (HPV) infection is necessary for the comparative analysis of human and mouse genomic sequence. Then, development of cervical cancer. However, it is not sufficient by we evaluated seven known genes and nine unknown genes with at least one splicing event within and around this region. Finally, we conducted an association study using tag single nucleotide Note: Current address for D.S. Gerhard: Office of Cancer Genomics, National polymorphisms (SNP; ref. 14) to investigate whether germ-line Cancer Institute, Bethesda, MD 20892. sequence variants associate with susceptibility to ICC and CIN3. Requests for reprints: Janet S. Rader, Division of Gynecologic Oncology, Department of OB/GYN, Washington University School of Medicine, 660 South Euclid Avenue, Campus Box 8064, St. Louis, MO 63110. Phone: 314-362-3181; Fax: 314- Materials and Methods 362-2893; E-mail: [email protected]. I2007 American Association for Cancer Research. Choice of candidate genes and expressed sequence tags for doi:10.1158/0008-5472.CAN-07-2677 expression studies. In previous work, we detected a small (1 Mb) deletion

Cancer Res 2007; 67: (23). December 1, 2007 11202 www.aacrjournals.org

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2007 American Association for Cancer Research. CD83 Polymorphism Increases Susceptibility to Cervical Cancer on 6p23 of ICC patients between markers D6S429 and D6S1578 (13). This Because the 6p23 markers in LOH and microarray studies associate with region has only one known gene, CD83, which lies at its telomeric end. We both CIN3 and ICC (8, 13), we hypothesized that the locus may promote started looking for protein-coding segments in 6p23 by comparing it with genetic susceptibility to the disease. To identify relevant genetic variants, we the mouse sequence. Because this work began before the mouse genome performed a region-based association study using family samples. Poly- was completely sequenced, we used the RPCI-23 mouse library to develop a morphisms used for genotyping were either selected from the International bacterial artificial chromosome contig centered on CD83. Cold Spring HapMap Project database11 or identified by us during sequencing of Harbor Sequencing Center sequenced three overlapping clones covering candidate genes. For this study, we chose gene-based tag SNPs, defining f650,000 bp of that region. We then determined the homology of the haplotype blocks by a minimum r2 of 0.8. The tag SNPs had to have a minor mouse DNA with human DNA using PipMaker,10 which was developed by allele frequency of >5%. All putative functional mutations we identified were Dr. W. Miller’s laboratory at Pennsylvania State University. Percent identity genotyped in the samples. The 24 tag SNPs span across 0.68 Mb from SIRT5 plots (Pips) provide complementary information to GenScan and highlight to CD83. An additional 31 SNPs were added around RNF182 and CD83, database hits for finding exons and candidate regulatory elements. close to markers with significant P value (P < 0.05). The genomic positions We selected CD83 and nine expressed sequence tags (EST) centromeric for all SNPs came from dbSNP build 127. to that gene within the LOH region to explore expression patterns in a panel We performed genotyping with several methods. One, Taqman SNP of human cervical tissues and other tissue types. To further investigate Genotyping Assays (Applied Biosystems) using the primers and probe of the potential candidate genes around the critical region, we expanded the company. Two, pyrosequencing across the polymorphic sites (Biotage, Inc.) survey region to 5 Mb from D6S470 to D6S1578. The expression of six more using primers designed and validated in house (available on request). The genes [PAK1IP1, NEDD9, SIRT5, RANBP9, nucleolar protein 7 (NOL7), and PCR profile and reaction conditions were tested and optimized using RNF182] telomeric to CD83 and with relevant biological functions was 24 individual control DNAs and pooled Caucasian or African-American evaluated in the same panel of tissues. DNAs. Genotyping success rates were >98% with the Taqman assays and Subjects for the association study. We obtained informed consent >95% with pyrosequencing. Each genotyping plate contained water and from 377 family trios. Each trio consisted of a proband—a woman with ICC DNA controls. The final genotypes were analyzed for transmission (255) or CIN3 (122, CIN3 and/or adenocarcinoma in situ)—and either her consistency between parents and offspring, and genotypes that showed biological parents or one parent and one or more siblings. Three hundred Mendelian error were deleted and not used in analysis. and forty-one trios were Caucasian, and 36 were African-American. Blood or RNA isolation, reverse transcription-PCR, and sequencing alterna- buccal samples were obtained from all the subjects. Invasive cancers or tively spliced transcripts of CD83. RNA was extracted from tissue using a CIN3 lesions from the probands were obtained for HPV typing. Among the Micro RNA Isolation kit (Stratagene) as per the manufacturer’s protocol. 255 ICC, 167 tumors were squamous, 62 were adenocarcinoma, 9 were Cell line RNAs were extracted using Trizol reagent (Invitrogen). Reverse adenosquamous, and 17 were of other histologies. Of the 303 cervical tissues transcription-PCR (RT-PCR) was performed as described previously (9). typed for HPV, 259 tissues are classified as HPV16 or HPV18 related. The Primers specific to the mRNA sequences were designed, and they spanned a study was approved by the Washington University’s Human Studies small intron when possible (primer information supplied on request). Committee. Primers were designed to capture all reported transcripts (SIRT5, three Cervical tissues and cell lines. Frozen specimens included 8 normal variants; NEDD9, two variants). RT-PCR products were identified by cervical biopsies, 11 normal cervical epithelia, 14 ICC, and cancers and ethidium bromide staining. matching nonmalignant tissues from lung, liver, colon, spleen, prostate, cDNAs representing the entire coding region of CD83 (NM_004233) were kidney, breast, brain, thyroid, muscle, and lymph node. The normal cervical generated by RT-PCR using primer sets (forward, 5¶-ATGTCGCGCGGICC- biopsies in our panel are enriched for the stratified epithelium near the TICCAGCTTC-3¶; reverse, 5¶-GCTCATAICCAGTTCTGTCTTGTGAGGAGTC- transformation zone, where carcinogenesis is thought to begin. RNA from 3¶) with Platinum Taq DNA polymerase (Invitrogen). The amplified the epithelium of fresh-frozen normal cervix and ICC was microdissected by fragments were separated on 2% agarose gel, visualized by ethidium laser capture (PixCell IIe Laser Capture Microdissection System). The cell bromide, and excised using a Gel Extraction kit (Qiagen, Inc.). The lines included two HPV16 E6/E7-transformed cervical epithelia cell lines, sequences were confirmed by bidirectional sequencing on an Applied 14 cervical cancer–derived cell lines, 3 ovarian cancer cell lines, 4 breast Biosystems 3730 Genetic Analyzer as described previously (16). cancer cell lines, and 3 prostate cancer cell lines. The following cell lines Identifying DNA polymorphisms and mutations in CD83 and NOL7. were purchased from the American Type Culture Collection: CRL2614, CD83 spans 19 kb of genomic sequence and contains five exons. NOL7 CRL2615, HeLa, SiHa, CaSki, C4I, HT3, ME180, C33A, SW756, DoTc2 4510, spans 5.57 kb of genomic sequence and contains eight exons. To detect the SKOV3, CAOV3, and MCF7. They were grown in the recommended medium. most functional sequence variations, we designed primer pairs that were Cell lines CCI, 931, 954, and 964 were generated in our lab and grown in adequate distances from exon/intron boundaries. PCR products from CD83 serum-free medium (Invitrogen). Cell lines LKP31 (HPV-transformed were analyzed either by direct sequencing or by denaturing high- keratinocytes), CIN612 (9Ep12, derived from a CIN1 lesion biopsy), and performance liquid chromatography (dHPLC) with the WAVE Nucleic Acid FC16 (HPV-transformed fetal cervix) were established by Dr. L.A. Laimins Fragment Analysis System (Transgenomic, Inc.). Those from NOL7 were (Northwestern University, Chicago, IL; ref. 15). Prostate cancer cell lines analyzed by direct sequencing. The resulting sequences were compared with 22Rv1, LNCaP, and PC-3 and breast cancer cell lines MDA-MB231, MDA- the corresponding reference gene sequences12 using DNAStar software MB435, and MDA-MB468 were provided by Dr. C.J. Anderson (Washington (DNASTAR, Inc.). Thirty-six tumors and matched normal genomic DNA University School of Medicine, St. Louis, MO) and grown in recommended were evaluated for CD83 and 12 pairs for NOL7. Pooled genomic DNAs from medium. the Caucasian and African-American populations were used to distinguish SNP selection and genotyping. Genomic DNA was isolated from blood novel SNPs from mutations as described previously (16). by overnight digestion with proteinase K and extraction with phenol/ Immunofluorescence analysis. Immunofluorescence staining of CD83 chloroform and from buccal cells using Puregene DNA Purification kits protein was performed according to the method of Klein et al. (17). Briefly, (Gentra Systems, Inc.) using protocol provided by the manufacturer. The f2 105 cells from HeLa, CaSki, ME180, and CRL2614 lines were seeded DNA was quantified by NanoDrop/ND-1000 Spectrophotometer and on coverslips in six-well culture dishes and incubated overnight. Each prepared at a concentration of 0.5 ng/AL and then loaded on 384-well coverslip was washed with PBS (without calcium and magnesium) thrice PCR plates (3 AL/well) using the automated Apricot Personal Pipettor and fixed with 1 to 2 mL of cold (20jC) 70% methanol for 5 min. The cells pp-150-MS (Perkin-Elmer). were then reacted with anti-CD83 antibody (1:20, MHCD8300, clone HB15e,

11 http://www.hapmap.org 10 http://bio.cse.psu.edu 12 http://www.ncbi.nlm.nih.gov/ www.aacrjournals.org 11203 Cancer Res 2007; 67: (23). December 1, 2007

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2007 American Association for Cancer Research. Cancer Research a mouse monoclonal antibody to the human CD83 antigen; Invitrogen, we reached several important conclusions. First, CD83 showed Caltag Laboratories) or with the isotype control (mouse IgG1 purified; strong homology at most of its five exons. Second, there was high Invitrogen, Caltag Laboratories) for 1 h at room temperature in TBS homology (>75%) for the exons of several predicted genes of containing 1% bovine serum albumin (BSA). After three TBS washes, the unknown function and this high level of homology provided the cells were reacted for 1 h with Alexa Fluor 488 goat anti-mouse antibody in reason why the ESTs were evaluated for their expression. TBS containing 1% BSA (50 AL/slide). For colocalization, cells were reacted with anti-human golgin-97 mouse monoclonal antibody (Invitrogen) for Expression of the candidate genes and EST in 6p23. In total, 1 h at room temperature and then with Alexa Fluor 594 goat anti-mouse seven known genes and nine ESTs within 6p23 were evaluated for PAK1IP1, NEDD9 SIRT5 antibody for 1 h. For nuclear staining, the same cells were treated with 4¶,6- expression by RT-PCR. (NM_006403), diamidino-2-phenylindole (DAPI; 1 Ag/mL; Sigma) for 5 min. The stained variant 1, NOL7, and RANBP9 were not expressed differentially in cells were analyzed with a Nikon Eclipse TE200-Uinverted microscope normal or cancerous cervix. The short variant of NEDD9 (L43821) using QImage software (QImaging Corp.). was expressed at a lower level in normal cervix than in ICC. The HPV typing. HPV typing was performed on tissue sections snap frozen expression of CD83, SIRT5 variant 2, and RNF182 was higher in ICC in OCT or formalin-fixed, paraffin-embedded blocks modified from Li et al. or cervical cancer cell lines than in normal cervix. None of the nine (18). A HPV-negative control sample was cut between every tenth specimen ESTs that were centromeric to CD83 was expressed at significant and no cross-contamination was found in the HPV typing. DNA was levels in normal or cancerous tissue. extracted from the OCT blocks by the standard SDS-proteinase K ¶ procedure. It was extracted from the paraffin blocks using a Puregene SNPs in the 3 region of CD83 confer risk of ICC. We analyzed DNA Purification kit. The DNA was amplified with primers to the conserved 24 SNPs across 0.68 Mb on 6p23 in the LOH region that spans regions of HPV L1 (L1C1/L1C2M) and E6 (E6-L/E6-R). Aliquots of PCR were SIRT5 to CD83 in the discovery set of 121 families with ICC. An run on agarose gel and dHPLC and then sequenced. If tissue was infected additional 31 SNPs were selected around CD83 for denser coverage with multiple HPV types (indicated by multiple peaks on dHPLC), the PCR where significant SNPs were identified at P < 0.05 in discovery set. fragments were first separated by the fragment collector and then Among the 55 SNPs analyzed, 7 were significant (P < 0.05). Along sequenced. Families were grouped according to the HPV type detected in with a few adjacent SNPs, they were genotyped in the remaining the cervical neoplasia of the probands at diagnosis. HPV16-related types are 134 trios with ICC. The validation set comprised all 255 ICC trios HPV16, HPV31, and HPV52. HPV18-related types include HPV18 and and showed that five of the seven SNPs remained significant. The HPV45. SNP association was strengthened when the families were subset Association analysis. We used the family-based test of association that is implemented in the program TRANSMIT (19) because it is robust to by high-risk HPV types defined by probands with HPV16- and population stratification and our association study included trios of HPV18-related types (Table 1; Fig. 1). The five SNPs with significant different ethnicities. A two-stage design was used to evaluate genetic P values were located within an 8-kb region of the 3¶-end of CD83. variation in the LOH region. First, 24 SNPs were screened in a discovery set However, these five SNPs were not significant in a smaller sample of 121 trios. SNPs significant at a nominal P = 0.05 were then typed in of CIN3 trios (Table 1). To eliminate the possibility that the CD83 validation set of 134 trios. We reported the results of stage I discovery SNPs conferred susceptibility to HPV16- and HPV18-related testing and our analysis of the combined stage I and II samples for optimal infections, we examined the distribution of the alleles in the five CD83 power (20). After having surveyed the 6p LOH region, SNPs within SNPs in women with HPV16- and HPV18-related cancers compared showed the strongest association; thus, we identified 31 additional tag SNPs with other HPV types. We found no significant difference in the located around CD83, assuming a loose block size (r2 = 0.5). Five SNPs distribution of the alleles between groups. Haplotype analyses of within CD83 showed marginal significance of the transmission/disequilibrium test of association. We examined haplotypes composed of these five these SNPs showed that a haplotype comprising markers rs9230- P SNPs, as well as all three SNP combinations, to determine whether these rs9370729-rs853362 was associated with ICC (0.001 < < 0.009). haplotypes showed stronger association than the individual SNPs. These The haplotype analysis did not add additional information to tests can suggest whether the haplotypic background is important in analysis of individual SNPs. tagging the functional variant. To identify possible heterogeneity of risk, we Polymorphisms and mutations in CD83. We identified poly- conducted follow-up tests by subdividing the sample into HPV16- and morphisms in genomic DNA and also identified several somatic HPV18-related types, as supported by numerous studies (6, 21, 22). mutations in a panel of ICC and cell lines. First, one tumor contained a 1-bp deletion (174 C/) 5 bp before the start codon Results ATG. Second, the cell line SW756 contained a two-nucleotide Human and mouse DNA are highly homologous at 6p23. We deletion (TT591-592/ ). This frameshift deletion creates a searched for homology between the human and mouse genome by nonsense mutation, truncating the protein to 137 amino acids. studying f650 kb of DNA centered on 6p23, first masking repeated Several novel SNPs were also identified in the promoter region and sequences in the human DNA with the RepeatMasker program.13 exons (Table 2). The polymorphisms were validated by methods The mouse sequence was obtained from British Columbia Genome other than sequencing, such as dHPLC. The dHPLC results and Sequencing Center before the completion of the mouse genome sequencing confirmed LOH in the region because among CD83 sequence. The mouse sequence was aligned and compared with the informative SNPs in 21 of 23 ICC indicated that only one human sequence using PipMaker. The protein-coding exons of allele was preserved. CD83 many human and mouse genes show an average homology of Alternatively spliced mRNAs. We identified three CD83 f85% (23, 24). We plotted percent identity (between 50% and constitutively expressed transcripts of in the cervical cancer CD83-1 100%) of the sequences from the two species using coordinates of cells. The full-length five-exon transcript of 618 bp ( ) and the human sequence, such as genes and repeats, along the two splice variants were confirmed by direct sequencing. Variant 2 CD83-2 horizontal axis. After plotting the homology of 17 predicted genes, ( ) lacked entire exon 3 and had only 389 bp. Variant 3 (CD83-3) lacked both exon 3 and exon 4 and had only 282 bp. The three alternatively spliced transcripts are predicted to encode three different proteins. CD83-1 would encode a protein of 205 amino 13 http://repeatmasker.genome.washington.edu acids and was expressed in all of the cell lines and human tissues

Cancer Res 2007; 67: (23). December 1, 2007 11204 www.aacrjournals.org

Table 1. Significant SNPs in discovery and validation sets for association study within 6p23

SNP All family trios Family trios with HPV16 and HPV18types

ICC CIN3 (n = 122) ICC CIN3 (n = 75)

Discovery (n = 121) Validation (n = 255) Discovery (n = 98) Validation (n = 184)

rs10498684 0.0854 0.0073 0.96243 0.0439 0.0512 0.87316 rs9296925 0.1540 0.3737 0.99948 0.0089 0.0193 0.54217 rs853360 0.1068 0.0116 0.59319 0.0129 0.0035 0.49491 rs9230 0.0098 0.0099 0.92108 0.0041 0.0011 0.63475 rs9370729 0.1193 0.0254 0.33736 0.0405 0.0012 0.06173 rs853362 0.2097 0.5969 0.71412 0.0477 0.1095 0.89239 rs750749 0.1119 0.1534 0.72441 0.0223 0.0133 0.78779

tested, including cervical, ovarian, breast, and prostate cancers. of staining in the cell membrane and perinuclear staining in the CD83-2 (79 amino acids) was detected in 6 of 12 ICC cell lines and 5 cytoplasm. Colocalization staining was performed on cell lines of 8 ICC tumors. CD83-3 (93 amino acids) was seen in 4 of 8 ICC HeLa, CaSki, ME180, CAVO3, SKOV3, and MCF7 using golgin-97 but in none of the 11 samples of normal cervical epithelium monoclonal antibody (red). It revealed that CD83 localized in the (P = 0.0181; Table 3; Fig. 2). This variant showed a similar Golgi apparatus in HeLa and CaSki cells, in the cytoplasm but not differential expression trend between normal (4 of 31 positive, in the Golgi apparatus in SKOV3 cells, in both the cytoplasm and 12.9%) and cancerous (4 of 20 positive, 20%) of other tissue types. adjacent membrane in ME180 cells, and in adjacent membrane in CD83 protein is localized to the cytoplasm of cervical cancer CAVO3 cells (Fig. 3). These results clearly confirm the presence of cells. Because our microdissected tumor samples should have intracellular CD83 in all the cell lines tested. contained few dendritic cells and RT-PCR identified transcripts in Polymorphisms and mutations in NOL7. NOL7 has been cervical tissues and cell lines, we hypothesized that CD83 may have shown to suppress tumor growth in a cervical cancer mouse a role in cancer cells. Therefore, we investigated expression of CD83 xenograft model and is located 0.5 kb from the peak association protein using a specific monoclonal antibody in two immortalized and LOH region (25). In addition, to evaluate expression for normal cervical epithelial cell lines (CRL2614, ectocervix; CRL2615, NOL7 as noted above, we evaluated this gene for mutations/ endocervix), six cervical cancer cell lines (HeLa, CaSki, ME180, polymorphisms in 12 women with ICC. We found several new SiHa, HT3, and C4I), two ovarian cancer cell lines (CAVO3 and polymorphisms in exon 1, 3¶-untranslated region (UTR), and SKVO3), and one breast cancer cell line (MCF7). A weak CD83 introns and three somatic mutations in exon 1 in two of the ICC signal was detected in all the cell lines tested, but signal intensities samples. The first mutation was G26A, which created a new start and locations differed. Double staining with CD83 monoclonal codon, ATG, upstream of the start ATG and produced a frameshift. antibody (green) and nucleus-specific DAPI (blue) revealed traces The putative transcript encodes a protein containing 257 amino

Figure 1. A, schematic view of 55 SNPs from the association study. The horizontal black bar on the top of the figure is the critical LOH region identified previously (13). Gray diamonds, markers from the total set of 55 SNPs; black diamonds, focus set of 17 SNPs. Validated SNPs with evidence of association (P < 0.05) are labeled with rs#. X axis, chromosome position. The unit is Mb. B, CD83 gene structure shown in a genomic context. Black boxes, exons; solid lines, introns. The unit in X axis is kb.

www.aacrjournals.org 11205 Cancer Res 2007; 67: (23). December 1, 2007

Table 2. Previously unidentified mutations and polymorphisms in cervical tumors in genes NOL7 and CD83

Gene Location Frequency in Nucleotide Nucleotide Amino acid Predicted Chromosome 6 c b the samples position* change change effect position

NOL7 mutation Exon 1 1/12 26 G>A n/a New ATG 13723563 Exon 1 1/12 141 G>A Leu37Ile 13723573 Exon 1 1/12 254 G>A Ala74Ala 13723791 CD83 mutation Exon 1 1/30 72 C>T 5¶-UTR 14225915 Exon 1 1/30 170 G>A 5¶-UTR 9 bp before ATG 14226013 Exon 1 1/30 174 C/ 5¶-UTR 5 bp before ATG 14226017 Intron 1 4/30 n/a C>G or G>C n/a 14226117 Exon 4 1/36x 591–592 TT/ Phe138 Stop Truncation (TTT>TAA) 14241889–14241890 Intron 4 1/36 n/a C>T n/a 14242062 Intron 4 1/12 n/a A>G n/a 14242513 Exon 5 1/33 1660 T>C 3¶-UTR 14244310 Exon 5 3/23 1757 A>C 3¶-UTR 14244407 Exon 5 1/33 1953 A>G 3¶-UTR 14244603 Exon 5 1/27 2038 T>C 3¶-UTR 14244688 Exon 5 3/27 2343 C>T or T>Ck 3¶-UTR 14244993 CD83 SNPs Promoter 5/11 n/a C/T n/a 14225258 Promoter 1/30 n/a C/T n/a 14225702 Promoter n/a 12 bp deletion n/a 14225799–14225810 Exon 1 10/30 170 A/G 5¶-UTR 14226013 Exon 1 1/30 201 T/C Leu8Pro 14226044 Intron 4 30/30 n/a T/ Intron 4 14242193 Exon 5 1/30 1268 C/T 3¶-UTR 14243918 Exon 5 1/22 1748 G/T 3¶-UTR 14244398 Exon 5 1/27 2224 A/G 3¶-UTR 14244874

Abbreviation: n/a, not applicable. *Nucleotide position for CD83 is based on NM_004233; for NOL7, it is based on NM_016167. cNucleotide change is compared with matched normal genomic DNA. bSome tumor is T>C, some C>T. xThis mutation was found in cervical cancer cell line SW756. kChromosome position is based on data from http://www.ncbi.nih.gov/snp dbSNP build 127.

acids, whereas the mutated one would encode a protein of 59 gene signature and could be responding to the increased activity amino acids. The second mutation was C141A, which causes a seen in the stroma at CIN2/3 (26). However, the limited number of nonsynonymous change, Leu37Ile. The third was G254A, which CIN3 sample may not be statistically sufficient to rule out the caused a synonymous change, Ala74Ala (Table 2). association of CD83 with development of CIN3. Clearly, this needs to be evaluated in further studies. Discussion Our data show that five SNPs within the 3¶-end of CD83 are Table 3. Expression of CD83 transcripts in cervical tissue overtransmitted in women with ICC in a family-based association and cell lines as determined by RT-PCR study, and thus, CD83 may represent a genetic risk factor for the disease, particularly in women with high-risk HPV16- and HPV18- Cervical cell lines or cervical tissue CD83-1 CD83-2 CD83-3 related type infections. This interaction with HPV types parallels our previous study showing that the HLA DQB1 0303 allele is Normal cervix (epithelium) 7/11 3/11 0/11* associated with an increased risk of cervical cancer in women with Cervical cancers 7/8 5/8 4/8 tumors containing HPV16, HPV18, HPV31, or HPV33 (6). No Cervical cancer cell lines 12/12 6/12 4/12 CRL2614 (HPV-transformed ectocervix) 1/1 1/1 1/1 significant differences were found in the distribution of the CD83 CRL2615 (HPV-transformed endocervix) 1/1 1/1 1/1 alleles among women with HPV16- and HPV18-related cancer FC16 (HPV-transformed fetal 1/1 0/1 0/1 compared with those who had other HPV types, indicating that cervical cell line) these SNPs are not associated with susceptibility to HPV16- or 9Ep12 (transformed CIN cell line) 1/1 0/1 0/1 HPV18-related type infection. In this study, the association with invasive cancer but not with CIN3 further suggests that CD83 might have an effect later in carcinogenesis. Our recent publication *Lack of CD83-3 in normal cervical epithelium is significant (P = of genomic signatures of cervical carcinogenesis showed that 0.0181) when compared with that in tumors. CIN3/squamous carcinoma transition coincided with a proinvasive

Cancer Res 2007; 67: (23). December 1, 2007 11206 www.aacrjournals.org

Berchtold et al. (36) narrowed down the highest promoter activity to the 261 bp region in CD83. This chromosomal fragment contains four SP1 binding sites and one NF-jB element. TNFa needs the latter to induce the CD83 promoter, and EBV needs it to induce CD83 expression on the surface of B cells (36, 37). One of 30 cervical tumors contained a mutation within the NF-jB element, and deletion polymorphisms were identified in the SP1 region (Table 2). The G170A SNP is nine nucleotides upstream of the translational start site (TSS), and loss of the G allele in cancerous versus normal cervix is highly significant (P < 0.025). In addition, 1 of 30 tumors had a single nucleotide deletion at 5 bp from the TSS (Table 2). Prechtel et al. (38) identified a structural element—the posttran- scriptional regulatory element (PRE)—in exons 4 and 5 of the CD83. PRE binds HuR (ELAVL1) and helps export CD83 mRNA from the nucleus. Binding of HuR to CD83 PRE does not affect the decay of CD83 transcripts but seems to alter the level of nucleocytoplasmic Figure 2. Representative alternative splicing of CD83. Full-length CD83 translocation via the CRM1 (XPO1) pathway. We identified three (618bp) is the major transcript in all samples. The sample names are on the top. somatic mutations in exons 4 and 5. The 8-kb region containing the 1, full-length CD83; 2, CD83-2, which lacks all of exon 3; 3, CD83-3, which ¶ CD83 lacks all of exon 3 and all of exon 4. A 100-bp DNA marker is at the left. five significant SNPs lies within PRE and the 3 -end of . RNA splicing is essential for protein diversity. It can also have regulatory functions, and alternative splicing patterns can be Analyses of these data were carried out including all families in cancer specific. The variant CD83-3 that lacks exons 3 and 4 was the study, although 341 of 377 families are Caucasian. The use of identified more often in cervical cancers and cell lines than in family-based controls, as in the transmission/disequilibrium test, normal cervixes (P = 0.0181). Moreover, this alternative transcript is allows us to pool over population groups because the methods are the only splice product secreted into the supernatant and elevated robust to the effects of population stratification (27). Nonetheless, in sera from leukemic patients (39). we also repeated the analyses using only Caucasian families, Our interest in the 6p23 region stems from extensive fine and the results were qualitatively the same. In this study, we also mapping of LOH in microdissected epithelial cervical cancer cells. showed that CD83 was transcribed and translated in cervical Therefore, it may seem surprising that the current study identified cancer cells. Somatic mutations in ICC, novel SNPs, and splice an immune marker as a susceptibility gene for the disease. variants were also identified within CD83. However, epithelial cells, once viewed principally as a mechanical CD83 is an inducible glycoprotein belonging to the immunoglobulin superfamily. Its exact function is unknown, but it seems to have an important role in T-cell immunity mediated by dendritic cells and to be a marker for mature dendritic cells (28). However, recent literature suggests a broader functional role and wider expression pattern (29–31). Transgenic expression of CD83- immunoglobin fusion protein impairs CD4+ T-cell development and T-cell activation, whereas CD83 knockout mice exhibit a profound block in the development of CD4+ T cells (32, 33). In addition, a mutation in exon 5 substantially reduced CD4+ T-cell development in mice. Moreover, the T cells that did develop failed to respond normally to allogenic stimulation, and their pattern of cytokine expression was skewed (30). Cell surface or immobilized CD83 has been shown to enhance T-cell activation (31), whereas soluble CD83 inhibits dendritic cell–mediated T-cell proliferation (31, 34). Human cytomegalovirus (HCMV) seems to exploit this mechanism to evade host immunity. Mature monocyte-derived dendritic cells that are infected with HCMV lose surface CD83 but continue to express the protein in the cytoplasm. The soluble CD83 is released, and it inhibits the stimulatory activity of noninfected mature dendritic cells and/or T-cell proliferative responses (35). Several mechanisms might potentially affect CD83 regulation or lead to inactivation. They include mutations, splice variants, allelic loss (LOH), and cancer-associated variants (Tables 2 and 3). dHPLC and sequencing confirmed LOH in the CD83 region because 21 of 23 Figure 3. Subcellular colocalization of CD83 in cervical cancer cell lines (91%) tumors with informative SNPs had retained only one allele at (HeLa, CaSki, and ME180) and in ectocervical epithelium transformed by HPV16 specific SNP markers compared with matched normal genomic E6/E7 (CRL2614). Red, location of the Golgi apparatus; green, location of DNA. The sequence and structure of the 5¶-UTR of the mRNA CD83; yellow, colocalization of Golgi and CD83, which could be seen in the Golgi. In HeLa, CD83 staining was seen in the cytoplasm and Golgi. In CaSki and transcript is usually very important in regulating protein synthesis. CRL2614, it was seen in the cytoplasm. In ME180, it was found both in the Using electrophoresis mobility shift assays and deletion mutants, cytoplasm and on the cell surface. www.aacrjournals.org 11207 Cancer Res 2007; 67: (23). December 1, 2007

Downloaded from cancerres.aacrjournals.org on September 30, 2021. © 2007 American Association for Cancer Research. Cancer Research barrier, are now regarded as having roles that are more complex in HPV is an epitheliotropic virus that infects epithelial cells and cellular immunity. They contribute to the local inflammatory and subverts the immune response. However, infection with this virus immune responses by secreting or responding to various cytokines, must persist if a precursor lesion is to progress to ICC. Thus, growth factors, and chemokines. Toll-like receptors on cervical escape from immune surveillance emerges as one important step epithelial cells and dendritic cells provide the first-line response to in the progression of HPV-linked tumors (42). We have identified a pathogens by stimulating the production of defense proteins and candidate for genetic susceptibility that may act through immune mediating the influx of inflammatory cells into the infected site or responses. Clearly, functional assays and replication studies in cancer (40, 41). However, it is unclear whether CD83 could play a independent cohorts are needed. Furthermore, identifying other role in epithelial cells or dendritic cells or both. Further work may genetic modifications to host cells that impair immune responses establish the mechanism for the increased association of CD83 to HPV infection, permit infection to persist, and facilitate the with ICC and determine which systems are involved. progression to cancer will be important for the development of NOL7 is a newly cloned gene, which located near our LOH therapeutic interventions. region. A recent study showed that NOL7 showed a tumor-specific loss of a single allele. Transfection of NOL7 into cervical carcinoma cells inhibited their growth in mouse xenografts by the decrease in Acknowledgments the production of the angiogenic vascular endothelial cell growth Received 7/16/2007; revised 9/11/2007; accepted 10/3/2007. Grant support: National Cancer Institute grants 5R01CA094141 and factor, as an increase in the production of the naturally occurring 5R01CA095713. inhibitor of angiogenesis thrombospondin-1 (25). Previous inves- The costs of publication of this article were defrayed in part by the payment of page tigations and our newly identified somatic mutations make NOL7 a charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. potential candidate tumor suppressor gene in late-stage develop- We thank the patients and their families who participated in this study and Bradley ment of cervical cancer. Coleman for his work on initial phase of Pips project.

References 15. Chang YE, Pena L, Sen GC, Park JK, Laimins LA. 30. Garcia-Martinez LF, Appleby MW, Staehling- Long-term effect of interferon on keratinocytes that Hampton K, et al. A novel mutation in CD83 results in 1. Saslow D, Castle PE, Cox JT, et al. American Cancer maintain human papillomavirus type 31. J Virol 2002;76: the development of a unique population of CD4+ T cells. Society Guideline for human papillomavirus (HPV) 8864–74. J Immunol 2004;173:2995–3001. vaccine use to prevent cervical cancer and its precur- 16. Zhang Z, Huettner PC, Nguyen L, et al. Aberrant 31. Scholler N, Hayden-Ledbetter M, Dahlin A, et al. sors. CA Cancer J Clin 2007;57:7–28. promoter methylation and silencing of the POU2F3 gene Cutting edge: CD83 regulates the development of 2. Khan MJ, Castle PE, Lorincz AT, et al. The elevated in cervical cancer. Oncogene 2006;25:5436–45. cellular immunity. J Immunol 2002;168:2599–602. 10-year risk of cervical precancer and cancer in women 17. Klein E, Koch S, Borm B, et al. CD83 localization in a 32. Luthje K, Cramer SO, Ehrlich S, et al. Transgenic with human papillomavirus (HPV) type 16 or 18 and the recycling compartment of immature human monocyte- expression of a CD83-immunoglobulin fusion protein possible utility of type-specific HPV testing in clinical derived dendritic cells. Int Immunol 2005;17:477–87. impairs the development of immune-competent CD4- practice. J Natl Cancer Inst 2005;97:1072–9. 18. Li J, Gerhard DS, Zhang Z, et al. Denaturing high- positive T cells. Eur J Immunol 2006;36:2035–45. 3. Tindle RW. Immune evasion in human papillomavi- performance liquid chromatography for detecting and 33. Fujimoto Y, Tu L, Miller AS, et al. CD83 expression rus-associated cervical cancer. Nat Rev Cancer 2002;2: typing genital human papillomavirus. J Clin Microbiol influences CD4+ T cell development in the thymus. Cell 59–65. 2003;41:5563–71. 2002;108:755–67. 4. Magnusson PK, Sparen P, Gyllensten UB. Genetic link 19. Clayton D, Jones H. Transmission/disequilibrium 34. Lechmann M, Krooshoop DJ, Dudziak D, et al. The to cervical tumours. Nature 1999;400:29–30. tests for extended marker haplotypes. Am J Hum Genet extracellular domain of CD83 inhibits dendritic cell- 5. Wang SS, Hildesheim A. Chapter 5: viral and host 1999;65:1161–9. mediated T cell stimulation and binds to a ligand on factors in human papillomavirus persistence and 20. Skol AD, Scott LJ, Abecasis GR, Boehnke M. Optimal dendritic cells. J Exp Med 2001;194:1813–21. progression. J Natl Cancer Inst Monogr 2003;31:35–40. designs for two-stage genome-wide association studies. 35. Senechal B, Boruchov AM, Reagan JL, Hart DN, 6. Neuman RJ, Huettner PC, Li L, et al. Association Genet Epidemiol 2007;31:776–88. Young JW. Infection of mature monocyte-derived between DQB1 and cervical cancer in patients with 21. WrightJD,LiJ,GerhardDS,etal.Human dendritic cells with human cytomegalovirus inhibits human papillomavirus and family controls. Obstet papillomavirus type and tobacco use as predictors of stimulation of T-cell proliferation via the release of Gynecol 2000;95:134–40. survival in early stage cervical carcinoma. Gynecol soluble CD83. Blood 2004;103:4207–15. 7. Orth G. Genetics of epidermodysplasia verruciformis: Oncol 2005;98:84–91. 36. Berchtold S, Muhl-Zurbes P, Maczek E, et al. insights into host defense against papillomaviruses. 22. Burger RA, Monk BJ, Kurosaki T, et al. Human Cloning and characterization of the promoter region Semin Immunol 2002;18:362–74. papillomavirus type 18: association with poor prognosis of the human CD83 gene. Immunobiology 2002;205: 8. Fitzpatrick MA, Funk MC, Gius D, et al. Identification in early stage cervical cancer. J Natl Cancer Inst 1996; 231–46. of chromosomal alterations important in the develop- 88:1361–8. 37. Dudziak D, Kieser A, Dirmeier U, et al. Latent ment of cervical intraepithelial neoplasia and invasive 23. Okazaki Y, Furuno M, Kasukawa T, et al. Analysis of the membrane protein 1 of Epstein-Barr virus induces carcinoma using alignment of DNA microarray data. mouse transcriptome based on functional annotation of CD83 by the NF-nB signaling pathway. J Virol 2003;77: Gynecol Oncol 2006;103:458–62. 60,770 full-length cDNAs. Nature 2002;420:563–73. 8290–8. 9. Zhang Z, Gerhard DS, Nguyen L, et al. Fine mapping 24. Waterston RH, Lindblad-Toh K, Birney E, et al. Initial 38. Prechtel AT, Chemnitz J, Schirmer S, et al. Expression and evaluation of candidate genes for cervical cancer on sequencing and comparative analysis of the mouse of CD83 is regulated by HuR via a novel cis-active 11q23. Genes Chromosomes Cancer 2005;43:95–103. genome. Nature 2002;420:520–62. coding region RNA element. J Biol Chem 2006;281: 10. Umayahara K, Numa F, Suehiro Y, et al. Comparative 25. Hasina R, Pontier AL, Fekete MJ, et al. NOL7 is a 10912–25. genomic hybridization detects genetic alterations dur- nucleolar candidate tumor suppressor gene in cervical 39. Dudziak D, Nimmerjahn F, Bornkamm GW, Laux G. ing early stages of cervical cancer progression. Genes cancer that modulates the angiogenic phenotype. Alternative splicing generates putative soluble CD83 Chromosomes Cancer 2002;33:98–102. Oncogene 2006;25:588–98. proteins that inhibit T cell proliferation. J Immunol 11. Rader JS, Gerhard DS, O’Sullivan MJ, et al. Cervical 26. Gius D, Funk MC, Chuang EY, et al. Profiling 2005;174:6672–6. intraepithelial neoplasia III shows frequent allelic loss in microdissected epithelium and stroma to model ge- 40. Andersen JM, Al-Khairy D, Ingalls RR. Innate 3p and 6p. Genes Chromosomes Cancer 1998;22:57–65. nomic signatures for cervical carcinogenesis accommo- immunity at the mucosal surface: role of toll-like 12. Rader JS, Kamarasova T, Huettner PC, et al. dating for covariates. Cancer Res 2007;67:7113–23. receptor 3 and toll-like receptor 9 in cervical epithelial Allelotyping of all chromosomal arms in invasive 27. Zhao H. Family-based association studies. Stat cell responses to microbial pathogens. Biol Reprod 2006; cervical cancer. Oncogene 1996;13:2737–41. Methods Med Res 2000;9:563–87. 74:824–31. 13. Rader JS, Li Y, Huettner PC, Xu Z, Gerhard DS. 28. Lechmann M, Zinser E, Golka A, Steinkasserer A. 41. Scott M, Nakagawa M, Moscicki AB. Cell-mediated Cervical cancer suppressor gene is within 1 cM on 6p23. Role of CD83 in the immunomodulation of dendritic immune response to human papillomavirus infection. Genes Chromosomes Cancer 2000;27:373–9. cells. Int Arch Allergy Immunol 2002;129:113–8. Clin Diagn Lab Immunol 2001;8:209–20. 14. de Bakker PI, Yelensky R, Pe’er I, et al. Efficiency and 29. Breloer M, Kretschmer B, Luthje K, et al. CD83 is a 42. zur Hausen H. Papillomaviruses and cancer: from power in genetic association studies. Nat Genet 2005;37: regulator of murine B cell function in vivo. Eur J basic studies to clinical application. Nat Rev Cancer 1217–23. Immunol 2007;37:634–48. 2002;2:342–50.

Cancer Res 2007; 67: (23). December 1, 2007 11208 www.aacrjournals.org

Zhengyan Zhang, Ingrid Borecki, Loan Nguyen, et al.

Cancer Res 2007;67:11202-11208.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/67/23/11202

Cited articles This article cites 42 articles, 10 of which you can access for free at: http://cancerres.aacrjournals.org/content/67/23/11202.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/67/23/11202.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/67/23/11202. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.