Expression of Variant CD44 Messenger RNA in Colorectal Adenocarcinomas and Adenomatous Polyps in Humans

Transcription

1 GASTROENTEROLOGY 1996;110: Expression of Variant CD44 Messenger RNA in Colorectal Adenocarcinomas and Adenomatous Polyps in Humans FUMIO IMAZEKI,* OSAMU YOKOSUKA,* TAKETO YAMAGUCHI,* MASAO OHTO,* KAICHI ISONO, and MASAO OMATA *First Department of Medicine and Second Department of Surgery, Chiba University School of Medicine, Chiba; and Second Department of Internal Medicine, Faculty of Medicine, Tokyo University, Tokyo, Japan C D44 molecule is a transmembrane glycoprotein expressed on a wide range of different cells and seems to have various functions such as a lymphocyte homing receptor on circulating lymphocytes 1 ; binding to collagen, fibronectin, and hyaluronate to confer cell-matrix contacts 2 4 ; and mediation of cell-cell adhesion. 5 CD44 molecules have heterogeneous isoforms, which are attributed both to variable exon usage with alternative splicing 4,6,7 and to differential glycosylation within the extracellular domain. Molecular cloning of CD44 disclosed the genomic region coding for exons 6 15, which are components of the extracellular domain of CD44 and can be incorporated alternatively into CD44 messenger RNA (mrna) (Figure 1). 8,9 Recent studies suggest that there are various isoforms of CD44 generated by this alternative splicing of exons 6 15 (variant exons Background & Aims: Recent studies have shown that some variant forms of CD44, a transmembrane glycoprotein expressed on various cell surfaces, might be involved in tumor progression or tumor metastasis. The aim of this study was to analyze the expression of CD44 messenger RNA (mrna) in colorectal cancer and colorectal adenoma to further elucidate the role of CD44 in colorectal tumorigenesis. Methods: The ex- pression of CD44-mRNA was examined in 90 specimens from 44 patients with colorectal cancer or colo- rectal adenomatous polyps and in the peripheral blood leukocytes from 7 healthy volunteers by reverse tran- scription polymerase chain reaction and Southern blot hybridization. Results: Strong expression of the epithelial form and variant forms containing exons 11 or 12 of the CD44 gene, which conferred metastatic behavior to rat cells, was detected in primary and metastatic tumor tissues, whereas it was very weak or not detectable in normal colonic mucosae, normal liver tissue, or peripheral blood leukocytes. However, adenomatous colorectal polyps also showed as strong an expression of epithelial and variant forms of CD44 as primary and metastatic tumor tissues. Conclusions: These results suggest that variant forms of CD44-mRNA might be expressed in an early stage of colorectal carcinogene- sis. 1 10). 10,11 One is expressed in hematopoietic cells with a molecular weight of kilodaltons and represents the basic unit of the CD44 proteins, the so-called hematopoietic or standard form. 4,6 Exons 6 15 are all spliced in this form of CD44. Another isoform, called the epithelial form, 4,6 is a protein with a molecular weight of kilodaltons, expressed weakly in normal epithelium but strongly in some carcinomas, which contains 135 amino acids more than the hematopoietic form in the extracellular portion of the molecule near the transmembrane region. Exons 6 12 are spliced in this form of CD44. Other variant forms are created by alternative splicing of the mrna and the addition of new exons to the extracellular domain near the transmembrane region of the hematopoietic form. Analysis of the comple- mentary DNA sequence of variant CD44 molecules, which is expressed on the surface of the metastatic rat pancreatic carcinoma cell line, has shown that the protein product of exons 11 and 12 (variant exons 6 and 7) of the CD44 gene is involved in the epitope of the metastasis of cancer cells and is sufficient to confer metastatic potential to nonmeta- static rodent cancer cells. 12,13 The expression of CD44 variant forms has been described in various human cancers, such as non-hodgkin s lymphoma, 14 breast cancers, 15 colorectal cancers, gastric cancers, 19,20 and uterine cervical cancers 21 by immunohistochemistry or reverse transcription polymerase chain reaction (RT-PCR). We report that CD44 splice variants are expressed in adenomatous polyps as much as in colorectal cancers by RT-PCR, suggesting that the expression of CD44 variant forms could be an early event in colorectal carcinogenesis. Materials and Methods Tumors and Tissues A total of 97 specimens were examined. Sixty-eight specimens obtained during surgical resection from 33 patients Abbreviations used in this paper: E/H ratio, epithelial/hematopoi- etic form ratio; RT-PCR, reverse transcription polymerase chain reac- tion by the American Gastroenterological Association /96/$3.00

2 February 1996 CD44-mRNA IN COLORECTAL CANCERS AND ADENOMAS 363 Figure 1. Schematic representation of the CD44 gene., Hematopoietic (standard) CD44;, exons that are spliced alternatively; Ø, untranslated regions;, transmembrane region. Exons 6 15 corre- spond to variant exons The location and direction of P1 and P2 oligonucleotide primers used for RT-PCR and P3, P4, and P5 for hybridization probe are shown. 37 C. The complementary DNA products were then used as templates for PCR in 100 ml volume of the reaction mixture containing 10 mmol/l Tris-HCl (ph 8.3), 50 mmol/l KCl, 1.5 mmol/l MgCl 2, 0.1 mmol/l deoxyribonucleoside triphosphate, 1 mmol/l primers P1 (5 -GACACATATTGCTTC- AATGCTTCAGC-3 ) (Figure 1), and P2. Amplification was performed with 2.5 U of Ampli-Taq polymerase (Takara Shuzo Corp., Shiga, Japan) in a DNA thermal cycler 480 (Perkin Elmer Cetus Corp., Norwalk, CT) after an initial 5-minute 94 C denaturation step, by 30 cycles of denaturation at 94 C for 1 minute, annealing at 55 C for 1 minute, and elongation at 72 C for 2 minutes. 15,24 Negative controls without total RNA in the reaction mixture were reverse-transcribed, followed by PCR amplification with every batch. Southern Blot Hybridization The PCR products were electrophoresed on 1.2% agarose gel and transferred to Hybond N / (Amersham In- (average age, 60 years) with colorectal cancer or liver metastasis ternational, Little Chalfont, Buckinghamshire, England) included 27 primary tumors, 1 polyp, 23 corresponding nor- nylon membranes for consecutive hybridization with mal colon mucosae, 9 metastatic liver tumors, and 8 corre- 32 P-labeled oligonucleotide probes P5 (5 -TGAGATTGGsponding normal liver tissue specimens (Table 1). Patients 1 GTTGAAGAAATC-3 ), P4 (5 -CCAGGCAACTCCTAG- 24 were without liver metastasis at the time of surgery. Pa- TAGTACAA-3 ), and P3 (5 -CCTGAAGAAGATTGTtients had liver metastasis at the time of surgery, and ACATCAGTCACAGAC-3 ). The locations of these both primary and metastatic tumors were analyzed. Metastatic primers are shown in Figure 1. Hybridization was perliver tumors only were obtained from patients whose formed using Rapid-Hyb buffer (Amersham), and the filprimary colorectal tumors had been resected 1 3 years earlier. ters were exposed to Kodak XAR-5 film with an intensi- Twenty-two specimens obtained during colonofiberscopy from fying screen, followed by immersion in boiling water for 11 patients (average age, 61 years) with colon polyps included 5 minutes before rehybridization with another probe. Two 5 adenocarcinomas, 15 adenomas, and 2 normal colon mucosae bands corresponding to hematopoietic and epithelial forms (Table 2). All of these samples were frozen quickly after exci- were quantitated with an Image Analyzer BAS 2000 (Fujix, sion and stored at 080 C until use. In addition, normal periph- Tokyo, Japan), and the ratio of the epithelial form to the eral blood leukocytes were obtained from 7 healthy volunteers. hematopoietic form (E/H ratio) was calculated. Statistical Tumorous and nontumorous parts of tissues and polyps were analysis was performed with the Wilcoxon s rank sum test. also fixed in 10% formalin and stained with H&E for histological examination. Colorectal carcinomas were staged according Results to the TNM classification 22 and histologically according to the Expression of CD44-mRNA in Colorectal degree of dysplasia in the polyps from mild to severe; the Tumors and Normal Colonic Mucosae histological grade of the carcinomas ranged from well differentiated to poorly differentiated. The clinical and pathological RNA was amplified with RT-PCR and hybridized features of these patients are described in Tables 1 and 2. with exon-specific probes to analyze the difference of RT-PCR Amplification expression of CD44-mRNA between tumors and normal tissues. Using primers P1 and P2, located 5 upstream Total cellular RNA was extracted from tumorous and and 3 downstream of the insertion site of the variant nontumorous tissues by the method of Chomezynski and Sac- exons (Figure 1), the 0.48-kilobase PCR product, indicachi. 23 Two micrograms of total RNA was used as a template tive of the hematopoietic form, was detected with a P3 for the synthesis of complementary DNA in reactions that oligonucleotide probe both in tumors and normal tissues. comprised 20 mmol/l Tris-HCl (ph 8.4), 50 mmol/l KCl, In contrast, several additional splice variants could be 2.5 mmol/l MgCl 2, 0.1 mg/ml bovine serum albumin, 0.5 amplified from primary and metastatic tumors (Figure mmol/l deoxyribonucleoside triphosphate, and 10 mmol/l dithiothreitol. First-strand synthesis was performed for 50 minthe epithelial form was amplified more frequently from 2A). The 0.88-kilobase PCR product corresponding to utes at 42 C in the presence of 200 U of SuperScript reverse transcriptase (GIBCO BRL Life Technologies, Inc., Gaithers- tumor specimens than from normal tissues (Figure 2A). burg, MD) with the specific primer P2 (5 -GATGCCAAG- The ratios of the epithelial to the hematopoietic form ATGATCAGCCATTCTGGAAT-3 ) (Figure 1) followed by (E/H) of CD44 were examined for a quantitative determinaincubation with 2 U of ribonuclease H for 20 minutes at tion of the epithelial form, one of the variant forms, in

3 364 IMAZEKI ET AL. GASTROENTEROLOGY Vol. 110, No. 2 Table 1. Clinicopathologic Characteristics of Patients With Colorectal Cancer E/H ratio Patient Age (yr)/sex Site Size (mm) Clinical stage a Histology T b M c N d 1 49/M Sigma Villotubular /M Rectum Well e /M Sigma Well /F Rectum Well /F Ascending Moderately f /M Descending Moderately 0.2 ND g 7 78/M Ascending Moderately /M Sigma Moderately /F Sigma Moderately /F Sigma Moderately /F Rectum Moderately /M Sigma Well /M Rectum Moderately 0.2 ND 14 79/M Sigma Well /F Cecum Moderately /M Transverse Moderately /F Rectum Moderately /M Sigma Moderately /M Sigma Well /F Ascending Well /F Cecum Moderately 0.5 ND 22 52/F Rectum Well 0 ND 23 61/M Sigma Moderately /M Sigma Moderately /M Sigma Moderately /F Ascending Moderately /F Sigma Mucinous /M Liver h ND 1.6 ND 29 52/M Liver ND 0.2 ND 30 55/F Liver ND 0.8 ND 31 38/M Liver ND 1.4 ND 32 57/M Liver ND 1.1 ND 33 69/M Liver ND 0.7 ND a Staged according to the TNM classification 22 : 0, carcinoma in situ; 1, tumor invades submucosa or muscularis propria without regional lymphnode and distant metastasis; 2, tumor invades through muscularis propria or perforates visceral peritoneum or directly invades other organs without regional lymphnode and distant metastasis; 3, tumor with regional lymphnode metastasis but no distant metastasis; 4, tumor with distant metastasis. b Primary colon cancer. c Metastatic liver tumor. d Nontumorous colon epithelium. e Well-differentiated adenocarcinoma. f Moderately differentiated adenocarcinoma. g Not done. h Metastasis. from whom specimens representing primary tumors and corresponding normal tissue specimens were simultaneously analyzed, displayed higher E/H ratios in their primary tumors than in their normal mucosae, and none of the 23 patients revealed higher E/H ratios in normal mucosae than in colorectal tumor tissue. We studied the correlation be- tween this E/H ratio and tumor size and clinical stage and histological grade. The E/H ratio was 0.6 { 0.5 in the tumor that was õ6 cmand0.8{ 0.6 in the tumor that was ú6 cm(p Å NS). The E/H ratio was 0.8 { 0.6 in stages 1 and 2 compared with 0.6 { 0.4 in stages 3 and 4(P Å NS). The E/H ratio was 1.0 { 0.6 in the well- primary and metastatic tumors and normal tissues. The mean values of E/H ratios in primary tumors, metastatic tumors, and normal colon mucosae were 0.7 { 0.5, 0.7 { 0.6, and 0.1 { 0.2, respectively, with both primary and metastatic tumors revealing higher ratios than normal mucosae (P õ 0.001). The liver metastatic tumor specimens displayed almost the same E/H ratios as the primary tumor specimens (Figure 3) and showed rather lower ratios than the primary tumor specimens in 3 patients from whom specimens representing normal mucosae, primary tumors, and liver metastatic tissue specimens were all simultaneously available (Figure 2A). Nineteen of 23 patients (83%),

4 February 1996 CD44-mRNA IN COLORECTAL CANCERS AND ADENOMAS 365 Table 2. Clinicopathologic Characteristics of Patients With Colorectal Polyps E/H ratio Patient Age (yr)/sex Site Size (mm) Histology P b N c 34 33/M Transverse Adenoma (mild) a 0.9 ND d 35 61/M Cecum Adenoma (moderate) 1.6 ND Transverse Adenoma (severe) 0.9 ND 36 61/M Rectum Adenoma (moderate) 1.4 ND 37 61/F Ascending Adenoma (mild) 1.9 ND Transverse Adenoma (mild) 2.9 ND Rectum Adenoma (severe) 2.1 ND 38 63/M Cecum Adenoma (mild) 1.9 ND Ascending Adenoma (mild) 2.0 ND 39 56/M Transverse Adenoma (moderate) /M Sigma Adenocarcinoma 1.9 ND 41 63/M Sigma Adenocarcinoma 1.4 ND 42 68/M Sigma Adenocarcinoma 1.8 ND Rectum Adenoma (mild) 0.8 ND 43 63/M Ascending Adenocarcinoma /M Ascending Adenocarcinoma 2.6 ND Transverse Adenoma (moderate) 3.1 ND Sigma Adenoma (moderate) 2.8 ND Sigma Adenoma (severe) 1.3 ND Sigma Adenoma (moderate) 1.9 ND a Parentheses show the degree of dysplasia in adenoma tissue. b Colon polyp. c Corresponding normal colon mucosa. d Not done. differentiated type and 0.5 { 0.4 in the moderately differen- fiberscopy, with 5 patients showing focal adenocarcinomas tiated type (P õ 0.05). and the other 15 showing adenomas (Table 2). The Hybridization with primers P4 and P5, located at expression of CD44-mRNA was also examined in these exons 11 and 12, respectively (Figure 1), showed that samples and, interestingly, the epithelial form and splic- the PCR products at 1.15 kilobases and larger sizes were ing variant forms containing exons 11 and 12 were detected easily detectable in primary and metastatic tumor tissues in addition to the hematopoietic form not only in but were only very weakly detectable or not detectable the early cancers but also in the adenomas (Figure 3A in normal tissues (Figure 2B and C). The expression of C and Table 2). Normal colon mucosae obtained from these variant forms correlated well with that of the epi- patients 39 and 43 displayed only the hematopoietic thelial form. No differential expression was shown in form (lane 9 in Figure 3A C). primary tumors without liver/lymph node metastasis and The E/H ratios were determined in these samples. with metastasis (lanes 1 and 3 vs. lanes 5, 7, and 9 in Their average in the adenomas with mild, moderate, and Figure 2). severe dysplasia and in the adenocarcinomas was 1.7 { Patient 24 had colorectal cancer with familial adenomatous 0.8, 2.0 { 0.7, 1.4 { 0.6, and 1.9 { 0.4, respectively; polyposis, and 1 of the multiple polyps, 15 mm that in the adenomas of õ10 mm and ú10 mm in in diameter, was analyzed for CD44-mRNA. The expression diameter was 1.9 { 0.9 and 1.7 { 0.6, respectively of CD44-mRNA in this polyp, with no evidence of (P Å NS). There was no correlation between these ratios cancer, was as strong as in the corresponding primary and histological grade (degree of dysplasia) or polyp size. tumor (data not shown). Peripheral blood leukocytes and The expression of alternative splicing variants also normal liver specimens expressed the hematopoietic form showed no correlation with the degree of dysplastic of CD44 but not the epithelial form or splicing variants change of the polyps (Figure 3A C). Some of the samples (Figure 2A C). were analyzed in duplicate from RNA extraction and the Expression of CD44-mRNA in Colorectal results were reproducible (data not shown). Polyps Discussion Twenty colorectal polyps ranging from 3 to 20 Our study showed that the expression of CD44 mm in size were obtained from 11 patients during colono- in colorectal carcinomas differed dramatically from that

5 366 IMAZEKI ET AL. GASTROENTEROLOGY Vol. 110, No. 2 in normal mucosae, but there was no differential expression in benign adenomatous polyps, malignant colorectal tumors, or metastatic liver tumors. The combination of RT-PCR and Southern blot hybridization provides a very sensitive method for analyzing the alternative splicing isoforms of CD44, but it does not allow for the precise quantification of various isoforms. Then, the E/H ratios of CD44 were calculated to quantitate the level of the epithelial form, one of the variant forms of CD44, as shown by Tanabe et al. 17 The E/H ratios in colorectal cancer were higher than those in corresponding normal mucosae in 19 of 23 cases (83%), but the epithelial form or splicing variant forms were detected to a lesser extent Figure 2. Analysis of CD44-mRNA in colorectal cancer by RT-PCR and Southern blot hybridization. Amplified products were electrophoresed on 1.2% agarose gel, and, after transfer to Hybond N / membranes, Figure 3. Analysis of CD44-mRNA in colorectal polyp by RT-PCR and the same filters were hybridized consecutively with (C) primer P5, (B) Southern blot hybridization. Amplified products were examined in the primer P4, and (A) primer P3 probes. T, primary colorectal tumor; N, same way as in Figure 2. Consecutive hybridization probes were (C) normal colorectal mucosa; M, metastatic liver tumor; L, normal liver; primer P5, (B) primer P4, and (A) primer P3. n, normal colon mucosa; PBL, peripheral blood leukocytes from normal volunteers. E, epithelial m, mild dysplasia; mo, moderate dysplasia; s, severe dysplasia; ca, form of CD44; H, hematopoietic form of CD44. Lanes 1 and 2, patient adenocarcinoma. E, epithelial form of CD44; H, hematopoietic form 4; lanes 3 and 4, patient 9; lanes 5 and 6, patient 14; lanes 7 and of CD44. Lane 1, patient 36; lanes 2 4, patient 37; lane 5, patient 8, patient 17; lanes 9 12, patient 25; lane 13, patient 30; lane 14, 41; lanes 6 and 7, patient 42; lanes 8 and 9, patient 43; lanes 10 patient , patient 44.

6 February 1996 CD44-mRNA IN COLORECTAL CANCERS AND ADENOMAS 367 ant exon 7) was present in early adenoma but had no correlation with colorectal tumor progression. Dis- cordance between the expressions of mrna and the protein product may be involved in these differences of CD44 expression, as also reported in the keratin gene expression in the salivary gland. 27 The hematopoietic form is the most widely expressed type of CD44 and has been shown to be a major cell surface receptor for hyaluronate. 4 It has been discussed that the hyaluronate binding activity of CD44 on some T cells was induced by a CD44-specific antibody, 28 suggesting that physiological mechanisms may exist for the activation of CD44 function. Although the precise role and the regulation of the CD44 variants remain to be defined, a similar process might be involved in the induc- tion of variant CD44 activation. Guo et al. 29 reported that the concentration of soluble CD44 in serum was elevated in patients with advanced colon cancer and correlated with tumor metastasis and tumor burden. If elevated CD44 levels in the serum would be related to tumor volume rather than metastasis and there was no differential expression of CD44 in benign adenomatous polyps and malignant colorectal tu- mors, patients with familial adenomatous polyposis, with no evidence of cancer, may have elevated serum CD44 levels to the same degree as patients with colon cancer. Mulder et al. 30 reported that the expression of the rat metastasis-related variant of CD44 (exon v6) in colon tumors by immunohistochemistry was associated with an unfavorable prognosis. Finn et al. 31 also showed by RT-PCR assay that increasing levels of variant CD44 forms were associated with poor survival. However, CD44v6 expression and variant molecules of CD44 were detected in adenomatous polyps by immunohistochemis- try and Western blotting. 16,26 By the RT-PCR method, we also showed the overexpression of variant CD44 at the stage of benign adenomatous polyps that precedes colorectal cancer. The mechanism of CD44 production remains to be clarified. The extensive genomic alterations seen in colo- rectal tumors might cause the variant mrna of CD44, although various isoforms of CD44 are suggested to be generated by an alternative splicing mechanism. 10,11 Analysis of genomic DNA from matched pairs of colon cancer and normal colon mucosa by Southern blot hybridization may very well confirm that the variant mrna is caused by splicing rather than genomic dele- tion or rearrangement. In conclusion, our result suggests that the expression of the CD44 variants could be an early event in colorectal carcinogenesis similar to the genetic changes of APC or K-ras genes that have been exemplified in the colorectal in 10 of 23 (43%) normal colon mucosae, indicating that the expression of these variant forms is not specific for cancer. 15,17,25 Interestingly, the E/H ratios in adenomatous polyps showed no difference from those in colorectal cancers (Figure 3A and Table 2), and the level of these ratios was not different among primary tumors with liver/ lymph node metastasis (stages 3 and 4) or without metastasis (stages 1 and 2) and liver metastatic specimens (Figure 2A and Table 1). Hence, the overexpression of the epithelial form was not specific for primary cancer or for metastasis. The E/H ratios were different for the studies on cancers and adenomas, approximately 0.7 for the former and 1.8 for the latter. This was because CD44 expression in the study of polyps showed a decrease in hematopoietic form rather than an increase in epithelial form and some variant forms (Figure 3A), although the reason for this is not clear. The result was reproducible when we examined several samples of cancers and polyps simultaneously from RT-PCR in the same batch by Southern blot hy- bridization on the same filter (data not shown). Our study showed that the E/H ratio in well-differentiated colorectal cancer was higher than that in the moderately differentiated type. Our results are compatible with the report by Kim et al. 26 that variant molecules of CD44 were detected more frequently in the adenomatous portions than in the accompanying carcinomas in the study of colorectal adenomas. Matsumura et al. 15 reported that there was a marked difference in the number and size of the bands and the intensity of the signal containing an exon-12 (variant exon 7) sequence with RT-PCR among samples from liver metastatic tumors and primary tumors with metastasis or without metastasis. Our results using RT-PCR showed that CD44 containing an exon-11 or exon-12 sequence had no differential expression in metastatic tu- mors or in primary tumors with or without liver/lymph node metastasis. Our results also indicated that the size and number of the signals containing exon 11 were identical to those containing exon 12 in the cases of colorectal cancers and adenomas (Figures 2 and 3). Because we could not differentiate the adenomatous polyp from can- cer by the intensity and size of the bands of variant CD44, analysis of clinical samples for CD44 variants by RT-PCR is unlikely to be of practical value for cancer diagnosis. Wielenga et al. 18 showed immunohistochemically that the overexpression of CD44 isoforms carrying epitopes of exon 11 (variant exon 6) was well correlated with the progression of colorectal tumors, i.e., with the severity of dysplasia in adenoma and with Dukes stage in invasive carcinoma. They also reported that the expression of CD44 carrying epitopes encoded by exon 12 (vari-

7 368 IMAZEKI ET AL. GASTROENTEROLOGY Vol. 110, No. 2 adenoma-carcinoma sequence, 32 and a novel expression cancer diagnosis and disease evaluation. Lancet 1992; 340: of the CD44 gene at an early stage of carcinogenesis 16. Heider K-H, Hofmann M, Hors E, van den Berg F, Ponta H, Herrlich might provide favorable conditions for cell proliferation. P, Pals ST. A human homologue of the rat metastasis-associated variant of CD44 is expressed in colorectal carcinomas and adenomatous polyps. J Cell Biol 1993;120: References 17. Tanabe KK, Ellis LM, Saya H. Expression of CD44R1 adhesion 1. Jalkanen S, Bargatze RF, de los Toyos J, Butcher EC. Lymphocyte molecule in colon carcinomas and metastases. Lancet 1993; recognition of high endothelium: antibodies to distinct epitopes 341: of an kd glycoprotein antigen differentially inhibit lympho- 18. Wielenga VJM, Heider K-H, Offerhaus GJA, Adolf GR, van den cyte binding to lymphnode, mucosal, or synovial endothelial cells. Berg FM, Ponta H, Herrlich P, Pals ST. Expression of CD44 variant J Cell Biol 1987;105: proteins in human colorectal cancers is related to tumor progres- 2. Carter WG, Wayner EA. Characterization of the class III collagen sion. Cancer Res 1993;53: receptor, a phosphorylated, transmembrane glycoprotein ex- 19. Heider K-H, Dammrich J, Skroch Angel P, Muller Hermelink H, pressed in nucleated human cells. J Biol Chem 1989;263: Vollmers HP, Herrlich P, Ponta H. Differential expression of CD splice variants in intestinal- and diffuse-type human gastric carci- 3. Jalkanen S, Jalkanen M. Lymphocyte CD44 binds the COOHnomas and normal gastric mucosa. Cancer Res 1993;53:4197 terminal heparin-binding domain of fibronectin. J Cell Biol ;116: Mayer B, Jauch KW, Gunthert U, Figdor CG, Schildberg FW, Funke I, Johnson JP. De-novo expression of CD44 and survival in gastric 4. Stamenkovic I, Aruffo A, Amiot M, Seed B. The hematopoietic and cancer. Lancet 1993;342: epithelial forms of CD44 are distinct polypeptides with different 21. Peter D, Heider K-H, Hekele A, von Minckwitz G, Kaufmann M, adhesion potentials for hyaluronate-bearing cells. EMBO J Ponta H, Herrlich P. Surface protein expression and messenger 1991;10: RNA-splicing analysis of CD44 in uterine cervical cancer and nor- 5. Koopman GY, van Kooyk Y, de Graaff M, Meyer CJLM, Figdor mal cervical epithelium. Cancer Res 1994;54: CG, Pals ST. Triggering of the CD44 antigen on T lymphocytes 22. Beahrs OH, Henson DE, Hutter RVP, Myers MH. Manual for stagpromotes T cell adhesion through the LFA-1 pathway. J Immunol ing of cancer. 3rd ed. Philadelphia: Lippincott, 1988: ;145: Chomczynski P, Sacchi N. Single-step method of RNA isolation 6. Stamenkovic I, Amiot M, Pasedo JM, Seed B. A lymphocyte mole- by acid guanidinium thiocyanate phenol-chloroform extraction. cule implicated in lymph node homing is a member of the carti- Anal Biochem 1987;162: lage link protein family. Cell 1989;56: Saiki RK, Gelfand DH, Stoffel S, Scharf SJ, Higuchi R, Horn GT, 7. Jalkanen S, Jalkanen M, Bargatze R, Tammi M, Butcher EC. Bio- Mullis KB, Erlich HA. Primer-directed enzymatic amplification of chemical properties of glycoproteins involved in lymphocyte rec- DNA with a thermostable DNA polymerase. Science 1988; ognition of high endothelial venules in man. J Immunol 1988; 239: : Fox SB, Gatter K, Jackson DG, Screaton GR, Bell MV, Bell JI, 8. Screaton GR, Bell MV, Jackson DG, Cornelis FB, Gerth U, Bell Harris AL, Simmonds D, Fawcett J. CD44 and cancer screening. JI. Genomic structure of DNA encoding the lymphocyte homing Lancet 1993;342: receptor CD44 reveals at least 12 alternatively spliced exons. 26. Kim H, Yang X-L, Rosada C, Hamilton SR, August JT. CD44 ex- Proc Natl Acad Sci USA 1992;89: pression in colorectal adenomas is an early event occurring prior 9. Tolg C, Hofmann M, Herrlich P, Ponta H. Splicing choice from ten to K-ras and p53 gene mutation. Archives Biochem Biophys variant exons establishes CD44 variability. Nucleic Acids Res 1994;310: ;21: Su L, Morgan PR, Harrison DL, Waseem A, Lane EB. Expression 10. Jackson DG, Buckley J, Bell JI. Multiple variants of the human of keratin m-rnas and proteins in normal salivary epithelia and lymphocyte homing receptor CD44 generated by insertions at a pleomorphic adenomas. J Pathol 1993;171: single site in the extracellular domain. J Biol Chem 1992; 28. Lesley J, He Q, Miyake K, Hamann A, Hyman R, Kincade PW. 267: Requirements for hyaluronic acid binding by CD44: a role for 11. Hofmann M, Rudy W, Zoller M, Tolg C, Ponta H, Herrlich P, Gunththe cytoplasmic domain and activation by antibody. J Exp Med 1992;175: ert U. CD44 splice variants confer metastatic behavior in rats: 29. Guo Y-J, Liu G, Wang X, Jin D, Wu M, Ma J, Sy M-S. Potential homologous sequences are expressed in human tumor cell lines. use of soluble CD44 in serum as indicator of tumor burden and Cancer Res 1991;51: metastasis in patients with gastric or colon cancer. Cancer Res 12. Gunthert U, Hofmann M, Rudy W, Reber S, Zoller M, Haussman 1994;54: I, Matzku S, Wenzel A, Ponta H, Herrlich P. A new variant of 30. Mulder J-W R, Kruyt PM, Sewnath M, Oosting J, Seldenrijk CA, glycoprotein CD44 confers metastatic potential to rat carcinoma Weidema WF, Offerhaus GJA, Pals ST. Colorectal cancer prognocells. Cell 1991;65: sis and expression of exon-v6 containing CD44 proteins. Lancet 13. Rudy W, Hofmann M, Schwartz Albiez R, Zoller M, Heider K-H, 1994;344: Ponta H, Herrlich P. The two major CD44 proteins expressed on 31. Finn L, Dougherty G, Finley G, Meisler A, Becich M, Cooper DL. a metastatic rat tumor cell line are derived from different splice Alternative splicing of CD44 pre-mrna in human colorectal tuvariants: each one individually suffices to confer metastatic be- mors. Biochem Biophys Res Commun 1994;200: havior. Cancer Res 1993;53: Fearon ER, Vogelstein B. A genetic model for colorectal tumori- 14. Koopman G, Heider K-H, Horst E, Adolf GR, van den Berg F, genesis. Cell 1990;61: Ponta H, Herrlich P, Pals ST. Activated human lymphocytes and aggressive non-hodgkin s lymphomas express a homologue of Received February 6, Accepted September 25, the rat metastasis-associated variant of CD44. J Exp Med Address requests for reprints to: Fumio Imazeki, M.D., First Depart- 1993;177: ment of Medicine, Chiba University School of Medicine, Inohana Matsumura Y, Tarin D. Significance of CD44 gene products for 8-1, Chuou-ku, Chiba 260, Japan. Fax: (81)