Cdx2 Protein Expression in Normal and Malignant Human Tissues: An Immunohistochemical Survey Using Tissue Microarrays Christopher A. Moskaluk, M.D., Ph.D., Hong Zhang, M.D., Steven M. Powell, M.D., Lisa A. Cerilli, M.D., Garret M. Hampton, Ph.D., Henry F. Frierson, Jr., M.D. Departments of Pathology (CAM, LAC, HFF), of Biochemistry and Molecular Genetics (CAM), and of Medicine (SMP), University of Virginia Health System, Charlottesville, Virginia; Department of Pathology (HZ), Anhui Medical University, Hefei, China; and Genomics Institute of the Novartis Research Foundation (GMH), San Diego, California KEY WORDS: Cdx2, immunohistochemistry, RNA Cdx2 has been identified as a marker of colon can- profiling, Tissue microarray. cer in RNA-profiling experiments. We show here Mod Pathol 2003;16(9):913–919 that the detection of Cdx2 protein by immunohis- tochemistry correlates well with RNA transcript lev- Cdx2 is a homeobox gene that has been shown to els as detected by oligonucleotide microarrays. Us- play a role in the development of the small and ing tissue microarrays containing most normal large intestine in mammals and in the differentia- tissue types and an antibody to the Cdx2 protein, tion of intestinal epithelial cells. The Cdx2 gene is strong diffuse Cdx2 staining was only seen in the expressed in all but the most distal portions of the nuclei of small and large intestinal epithelium and intestinal tract during development (1, 2). In a cell portions of the pancreatic duct system. In tissue culture model, Cdx2 has been shown to decrease microarrays containing 745 cancers from many an- cellular replication and induce differentiation to atomic sites, colonic adenocarcinomas showed mature intestinal epithelium (3). Cdx2 acts as a strong extensive staining in 90% of cases, with ade- transcription factor, increasing the expression of nocarcinomas of the stomach, esophagus, and several gene products associated with mature intes- ovary (endometrioid and mucinous types) showing tinal epithelial cells (3, 4). Perhaps not surprisingly, extensive staining in only 20–30% of cases. Other Cdx2 has recently been identified as a gene whose types of carcinomas showed extensive staining in expression had a high predictive value for the diag- only <1% of cases. Of 30 neuroendocrine tumors nosis of colonic adenocarcinoma in microarray examined, carcinoids of the midgut and hindgut gene expression analyses (5, 6). The use of Cdx2 as had the most cases with extensive staining (73% and an immunohistochemical marker has been de- 44%, respectively), thus paralleling the distribution scribed recently in studies of human gastric and of Cdx2 expression in adenocarcinomas. Cdx2 colon cancer, but no comprehensive survey of Cdx2 shows a limited range of expression in the spectrum protein expression has been reported (7–9). To fol- of human tissues and neoplasia and thus may have low up on previous RNA transcript profiling data, utility in determining the site of origin of tumors in we report our findings using immunohistochemis- certain clinical situations. try on tissue microarrays to comprehensively pro- file the protein expression of Cdx2 in a broad range of human carcinomas and normal tissues. Copyright © 2003 by The United States and Canadian Academy of Pathology, Inc. VOL. 16, NO. 9, P. 913, 2003 Printed in the U.S.A. MATERIALS AND METHODS Date of acceptance: May 29, 2003. Hong Zhang is an ITREID fellow in the Center for Global Health at The University of Virginia and is supported by NIH Fogerty Grant Tissue Specimens D43TW00909-05. Address reprint requests to: Christopher A. Moskaluk, M.D., Ph.D., De- Formalin-fixed, paraffin-embedded tissue blocks partment of Pathology, University of Virginia Health System, Rm. 3890 were obtained from the archives of the Department Old Medical School Bldg., Jefferson Park Ave., Charlottesville, VA 22908; fax: 434-243-6757; e-mail: [email protected]. of Pathology of the University of Virginia Health DOI: 10.1097/01.MP.0000086073.92773.55 System with institutional review board approval. 913 Tissue Microarrays TABLE 1. Cdx2 Immunohistochemical Staining in Normal Tissues Targeted tissue areas were marked on H&E- No staining Peripheral nerve stained sections corresponding to the paraffin Adipocytes Pituitary blocks. One to three 0.6-mm tissue cores from each Adrenal cortex Placental membranes case were removed using a manual microarray de- Adrenal medulla Placental villi Anal squamous epithelium Prostate vice (Beecher Instruments, Silver Spring, MD) and Astroglia Salivary gland, minor inserted into recipient paraffin blocks. The tissue Bartholin gland Salivary gland, parotid microarrays were sectioned at a 4-␮m thickness Bile duct epithelium Salivary gland, submandibular Bladder transitional epithelium Seminal vesicle and placed on charged slides. Blood vessels Skeletal muscle Bone Skin, squamous epithelium Bone marrow Skin, sweat glands Breast (duct epithelium) Smooth muscle, uterus Immunohistochemistry Breast (lobular epithelium) Spinal cord Bronchial epithelium Spleen Tissue microarray and standard tissue histologic Brunner glands Stomach, antrum sections were deparaffinized, rehydrated, and Cartilage (bronchus) Stomach, fundus Endocervical epithelium Synovium blocked with methanolic 3% hydrogen peroxide. Endometrial epithelium, Tendon The slides were immersed in 10 mM citrate buffer, proliferative pH 6.0 and heated in a 1200W microwave oven at Endometrial epithelium, secretory Testis, Leydig cells Endometrial stroma Testis, rete the highest power setting. Evaporated liquid was Endometrial stroma, decidualized Testis, seminiferous tubules replenished at 5 minutes, then the slides were Epidiymis Thymus heated at high power for an additional 5 minutes. Esophagus, squamous epithelium Thyroid Esophagus, submucosal glands Tonsil The slides were left in the buffer for an additional 10 Fallopian tube Ureter minutes before being removed. The antibody used Gallbladder Urethra, male was a mouse monoclonal antibody raised against Ganglia, peripheral Vagina, squamous epithelium Heart, myocardium Vas deferens full-length recombinant Cdx2 protein (AM392—5 M; Kidney, cortex Vulva, squamous epithelium BioGenex, San Ramon, CA). Immunostaining was Kidney, medulla Umbilical cord performed by hand using the prediluted antibody Lacrimal gland Liver, hepatocytes 2ϩ staining solution and a 30-minute incubation. After wash- Liver, portal tracts Pancreatic ducts, intermediate ing, antibody binding was visualized using the to large cts avidin-biotin-peroxidase technique (Vectastain Lung alveoli Lymph node 4ϩ staining Elite ABC kit, Vector Laboratories, Burlington, VT), Lymphatics Appendiceal epithelium followed by incubation with 3,3'-diaminobenzidine Melanocytes Colonic epithelium tetrahydrochloride. The slides were counterstained Mesothelium Pancreatic centroacinar cells Neurons (cerebellum) Pancreatic ducts, interacinar with hematoxylin. Neurons (cerebrum) Rectal epithelium The immunohistochemical stains were evalu- Ovarian stroma Small intestinal epithelium ated by two pathologists (HFF and CAM). Only Ovary, corpus luteum Ovary, primary follicle and germ nuclear staining was considered positive, and it cells was scored on the following basis: 0 (no detect- Pancreatic acinar cells able staining); 1ϩ (Ͻ25% positive cells); 2ϩ (25– Pancreatic islets Parathyroid 49%); 3ϩ (50–74%); 4ϩ (Ͼ75%). In general, cases showing 3ϩ and 4ϩ staining also had strong in- tense staining, so intensity was not factored into the scoring. RNA Gene Expression Profiling The list of normal tissues examined is given in Gene expression profiles of a subset of tumor Table 1. Focal heterogeneous Cdx2 expression cases were obtained on U95a GeneChips (Af- was present in the pancreatic tissue microarray fymetrix, Santa Clara, CA), as reported previously samples, and standard histologic sections of this (6). organ were also examined for confirmation of this finding. RESULTS The list of tumors examined is given in Table 2. The following tumors were absent from or not The degree of Cdx2 protein expression, as deter- well represented in the arrays, and additional mined by immunohistochemistry, was compared cases were obtained and examined as standard with the level of Cdx2 RNA transcript levels for a histologic sections: germ cell tumors (12 addi- subset of tumors. Transcript levels determined by tional cases), ampullary carcinoma (6 cases), and hybridization to GeneChip oligonucleotide arrays neuroendocrine carcinomas of the gastrointesti- (Affymetrix Inc.) were available for 20 breast ade- nal tract, biliary tract, lung, and pancreas (30 nocarcinomas, 19 colon adenocarcinomas, 2 gastric cases). adenocarcinomas, 12 lung adenocarcinomas, 12 914 Modern Pathology TABLE 2. Immunohistochemical Staining of Cancers Staining Pattern 0 1ϩ 2ϩ 3ϩ 4ϩ Ampullary carcinoma 21111 Bladder/ureter transitional cell carcinoma 70000 Breast adenocarcinoma 70 0000 Carcinoid tumors Lung 52000 Common bile duct 11000 Foregut (stomach) 30000 Midgut (ileum & appendix) 00117 Hindgut (rectum) 22131 Cholangiocarcinoma 87200 Colorectal adenocarcinoma 042549 Gallbladder adenocarcinoma 42000 Gastroesophageal adenocarcinoma 88 23 20 15 18 Germ cell tumors Dysgerminoma 31000 Embryonal Carcinoma 30000 Seminoma 60000 Yolk sac tumor 11210 Hepatocellular carcinoma 12 0000 Lung carcinoma Adenocarcinoma 33 1100 Large cell undifferentiated carcinoma 10000 Small cell/Neuroendocrine carcinoma 20000 Squamous cell carcinoma 13 0000 Melanoma 10000 Mesothelioma 28 0000 Ovarian tumors Clear cell ca 11000 Endometrioid