Expression of the cell adhesion molecule CD44 in gastric adenocarcinomas

Expression of the Cell Adhesion Molecule CD44 in Gastric Adenocarcinomas KAY

WASHINGTON,

MARILYN

CD44,

an integral

MD,

J. TELEN,

membrane

PHD,

MARCIA

R.

glycoprotein

expressed by many cell hyahuonate receptor

and may be a determinant of metastatic and invasive behavior in carcinomas. The expression of CD44 in 23 gastric adenocarcinoma and 12 peptic ulcer disease (PUD)

resection specimens and gastric

carcinoma cell lines HS746t and KATO III was examined by immnnobistochemistry using the mnrine monoclonal antibody A3D8 on fortissue or cells. Western blot analysis

of whole cell lysates of KATO

III and HS746t cells showed protein

bands at 85 to 90 kd with KATO

III cells expressing an additional

band at 145 kd. In normal stomach gastric epithelimn was negative. In PUD foveolar epithelinm was focally positive, but staining did not correlate with the extent of gas&is.

In carcinoma cases intensity of

staining was progressively stronger comparing intestinal metaplasia with dysplasia with intramucosal carcinoma. Invasive carcinoma was invariably more strongly positive than dysplasia or intramucosal carcinoma. Twelve adenocarcinomas

AND

lymphoma.;‘~” Metastatic spread requires that the tumor cells interact with the extracellular matrix and with other cells, presumably partly through tumor cell surface adhesion molecules. Because acquisition of the ability to metastasize is an important step in tumor progression that in part may be mediated by expression of cell surface adhesion molecules, CD44 expression in gastric adenocarcinomas was studied using two previously described monoclonal antibodies to CD44. A3D8 and AlGX These antibodies have been shown to recognize closely related but nonidentical epitopes of CD44.’ A3D8 has been shown by Western blot techniques to recognize an 80-kd membrane form of CD44 strongly expressed on a variety of leukocytes.

were weakly positive and 11 were

strongly positive. The staining intensity of metastases

(12 cases)

MATERIALS AND METHODS

was

the same or weaker than the primary tumor. For the 12 patients whose

Patient Selection

carcinomas were weakly positive, mean length of snrvival for the six who died was 23.3 months. Five of the 11 patients whose carcinomas strongly expressed CD44 died within the study period with a mean length of surGval of 11.O months. A key consequence of CD44 overexpression in gastric carcinomas may be development

of the invasive

phenotype and strong expression may indicate a poorer prognosis. HUM PATHOL 25:1043-1049.

MD,

MD

types. serves as the principal transmembrane

maim-fixed, paraffinembedded

GOTTFRIED,

Copyright 0 1994 by W.B. Saunders

Company

Members of the CD44 family of molecules are important in cell-cell and cell-matrix adhesion, and are expressed by many types of cells.’ The CD44 molecule expressed on lymphocytes (also denoted as Hermes antigen) is important in the adhesion and homing of lymphocytes to high endothelial venules in lymph nodes.’ (:D44 also may modulate adhesive functions exerted by other membrane glycoproteins and serves as the principal transmembrane hyaluronate receptor.” Recent work indicates that a variant of CD44 containing an additional extracellular domain of 162 amino acids confers metastatic potential to rat pancreatic and breast carcinoma cell lines.” High expression of CD44 has been shown to correlate with widespread dissemination and poor prognosis in non-Hodgkin’s

From the Departments of Pathology and Medicinr. Duke Univer\Ity Medical (:rncrr. Drwham. NC. .4rcepted for publication March w. 1093. Supported in pwt hv a College of American Pathologists Foundat~on Scholars Awar-d (KM’.). Northfield, 11.. and bv HI, X%72 and III. 0229.1, National Institutes of Hralth (M.J.T.), B&the&. MD. Kq wowf\: gastric adenocarcinoma. CD44. immunohistorhelnie tw. cell adhesion molecrtltx Addrrs correspondence and reprint requests to ha) Washingeon. MD, PhD, 130s 5712, Duke Ilniversity Medical Gntrr. Durham, NC: 27710. (:opyri@t CC)t 994 by W.B. Saunders Companv 004(i-x I77 ./I~-t,/?1,10-000s$:i.oo//0

Twenty-three consecutive cases of gastric adenocarcinomas surgically resected at Duke University Medical (Ienter between November 1986 and November 1991 were reviewed. None of the patients had received chemotherapy before surgery. The mean patient age was 65.3 years (range, 26 to 82 years). Fourteen of these 23 patients were male (Table 1). Nine patients had no metastases at surgery, 19 had regional lymph node metastases, one had liver merastases, and one had invasion of the pancreas. For the cases wit.h tumors length of survival after the time of surgical resection was recorded as well as observation of regional metasrases noted by the surgeon in the operative note. A control group of 12 partial gastric resections for peptic ulcer disease (PUD) from the Same time period also was stained for CD44. The mean age of this group was .X3.7 years

(range. ST to 72 years). Ten of these l? patients were male.

lmmunohistochemical

Staining

Tissue on all cases was fixed in buffered formalin and embedded in paraffin. Four-micrometer thick sections were stained with hematoxylin-eosin and with capillary action immunoprroxidase technology’ using the murine monoclonal antibody A3D8 or AlG3 without proteolytic enzyme treatment. (It had been determined from initial antibody characterization that pepsin had detrimental effects on the CD44 antigen present in the tissue section, thus causing negative staining results.) All cases were stained with ASD8: 10 carcinoma cases and five peptic ulcer cases also were stained with AlG3. The monoclonal antibodies were obtained as mouse ascitic fluid and used at a l:l,OOO dilution with 1 hour incubation at room temperature. The unlabeled bound primary antibodies were detected with biotin/streptavidin methodology and visualized with 3,3’-diaminobenzidine chromagen. Normal murine serum was used as the negative control. Gastric carcinoma cell lines KATO III and HS746t oh tained from American Type Culture Collection were grown in tissue culture flasks and collected bv gently scraping cells

1043

HUMAN

PATHOLOGY

Volume 25, No. 10 (October

1994)

TABLE 1. CD44 Expression in Gastric Adenocarcinomas Case No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Age/Sex 70/M 74/F 82/F 30/F 76/M 37/M 71/F 69/M 65/M 72/M 26/M 76/M 78/M 77/M 77/F 61/F 73/F 48/M 79/M 56/M 70/M 68/M 67/F

Metastases None Lymph Lymph Lymph Lymph Liver None Lymph Lymph None Lymph Lymph None Lymph Pancreas None None None None Lymph Lymph Lymph None

node node node node

node node node node node

node node node

Patholom

CD44 Intensity

Intestinal Diffuse Intestinal Diffuse Intestinal Intestinal Diffuse Diffuse Diffuse Intestinal Intestinal Intestinal Intestinal Intestinal Intestinal Intestinal Intestinal Intestinal Diffuse Intestinal Intestinal Diffuse Diffuse

Strong Weak Weak Strong Weak Weak Weak Weak Weak Strong Weak Weak Strong Strong Strong Strong Strong Strong Weak Strong Weak Strong Weak

? Variation Yes No Yes Yes No Yes Yes No No Yes Yes No Yes No Yes Yes Yes Yes Yes Yes Yes Yes No

Membrane Pattern Yes No Yes Yes No Yes Yes No Yes No No No No No Yes No No No Yes Yes No Yes Yes

Length of Survival (mos)

Alive Alive,

Alive, Alive. Alive Alive, Alive Alive, Alive,

Alive

14 24 with mets. 32 12 NED 42 7 5.5 4.9 1.5 NED 55 53 NED 32 0.5* with mets 26 NED 22 m with mets 10 NED 18 2 NED 19 16 1.5* 11 with mets 2

Abbreviations: M. male: F, female; ? Variation, ? increased staining with invasion; mets, metastases; NED, no evidence of disease. * Died of postoperative complications.

from the flask and pipetting to form a suspension. Trypsin was not used to harvest cells. The cells were then briefly fixed in buffered formalin and centrifuged, and the resulting pellet was collected and embedded in paraffin. Four-micrometer thick sections were stained with A3D8 at a l:l,OOO dilution as above, except that 3-amino, S-ethyl carbazole was used as the chromagen. For one of the tumor cases, one of the peptic ulcer cases, and one case of normal fundus taken from an esophagogastrectomy specimen with squamous cell carcinoma of the esophagus, tissue was snap frozen in liquid nitrogen and sections were fixed in cold acetone before staining in addition to examination of routinely submitted formalin-fixed, paraffinembedded sections.

Detection of A3D8 Antigen on Gastric Carcinoma Cell Lines by Western Blot Technique Cells from gastric carcinoma cell lines KATO III and HS746t were solubilized in detergent. Whole cell lysates were separated by 5% to 12% linear gradient polyacrylamide gel electrophoresis and transferred to nitrocellulose by standard methods.” The nitrocellulose paper was incubated in 10% bovine milk powder for 1 hour to block nonspecific binding sites and incubated with A3D8 at 1:500 dilution for 1 hour at room temperature, followed by incubation with immunoglobulin (Ig) G anti-mouse antibody. The protoblot alkaline phosphatase detection system (Promega Biotec, Madison, WI) was used to visualize the bands. Normal human erythrocyte ghosts served as a positive control. Buffer only and P3x63/Ag8 ascitic fluid (P3) served as negative controls.

Histological

Review

All slides were reviewed separately in a blinded fashion by two authors (K_W. and M.R.G.) and the tumors were scored as weakly positive or strongly positive. The staining intensity

1044

of nerve fibers was used as an internal control to assess the adequacy of the stain and also to provide a reference for the semiquantitative scoring of the tumors. Variation in staining intensity within individual tumors was noted separately. Membrane or cytoplasmic staining pattern was recorded. The carcinomas were classified as intestinal (15 cases) or diffuse (eight cases). Staining intensity of areas of intramucosal carcinoma. high grade dysplasia, and low grade dysplasia was graded separately.

RESULTS

CD44 EXPRESSION IN GASTRIC ADENOCARCINOMAS (Washington et al)

TABLE 2.

Distribution of A3D8 Antigen in Normal Human Stomach Mucosa

Tissue (iat dia Fovcolar epithrlium (&mds Fulldus Fovrolar epithelium (:hirf cells I’arietal ~~11% Antrum Fovrolar epithclium Neck cells .Intral glands Pylorus and Ihlodenunl Surface eplthelium Goblet cells Drunner’s glands Other cell ty~pes Smooth murcle \dipocytes Fihrohlast.5 Lymphocytes Peripheral nerve

Staining

invasive carcinoma. The metaplastic “enterocytes” had a difftlse cytoplasmic staining pattern, and staining of the brush border was observed in well developed intestinal metaplasia. Staining was usually stronger in intestinal metaplasia than in normal pylorus or duodenum. Goblet cells were negative in all cases. In sections of duodenum Brunner’s glands were positive with a cytoplasmic pattern.

Intensity

-

_

Gastric Carcinomas

+ (C)*

All gastric carcinomas were at least weakly positive with A3D8 (Table 1). Ten of these cases were stained with AlG3; all showed the same staining patterns and slightly weaker intensity than A3D8. Sixteen of the carcinomas had a similar pattern of variation in staining intensiq; the invasive areas of the carcinoma stained more strongly positive than the intramucosal portion (Fig 1). Areas of high grade dysplasia tended to be weaker than areas of intramucosal carcinoma. Areas of low grade dysplasia were weakIy positive or negative. Areas of intestinal metaplasia were variably weakly positive to negative and were always weaker than areas of dysplasia and carcinoma (Fig 2). Nine of the 23 carcinoma cases had, in addition to the diffuse cytoplasmic staining pattern, a membrane pattern. This pattern was observed in both intestinal and diffuse carcinomas. In three of the diffuse carcinomas a microvesicular staining pattern was present in signet ring-type tumor cells. Metastases in lymph node (10 cases), Iiver (one case), and pancreas (one case) were examined. In nine of the 10 cases with lymph node metastases the staining intensity was the same or weaker than the intensity of the original carcinoma. In one case the staining intensity appeared stronger. In the pancreas and liver metastases the staining was the same intensity as the primary tumor. In the liver a band of hepatocytes adjacent to the tumor exhibited membrane positivity whereas hepatocytes distant from the tumor were negative.

_ i/+/‘+ (C) and +hrush _

border

+ (C) + CC) + (m) + CC) +/Cm) ++ (m)

Abbreviations: C. c.ytvplasmic staining; m, membrane staining pattern; ++, strong stammg: +. weak staining; +/-. variable weak to negative staining: -, lack of staining above background. *-Parietal cells were negative on frozen section. Positive staining 011 paraffin sections may represent artifact.

Gastric Carcinoma

Areas of intestinal metaplasia observed in 12 tumor cases varied in staining intensity but always stained less strongly than dysplasia, intramucosal carcinoma, and

1045

Cell Lines

We used similar methods to examine two gastric carcinoma cell lines, KATO III and HS746t. Sections cut from paraffin-embedded pellets of these cell lines showed strong membrane staining with A3D8. Many cells also showed dense cytoplasmic staining (Fig 3). Western blot analysis of the two gastric carcinoma cell lines was performed to confirm that A3D8 antibody was binding specifically to CD44. These studies (Fig 4) showed that A3D8 bound to a protein band at approximately 85 to 90 kd. For HS746t cells the band was located at 85 kd whereas the band from the lysate of KATO III cells had a slightly larger molecular weight of 90 kd. For KATO III cells a diffuse band at approximately 145 kd also was present. For normal red blood cell ghosts A3D8 antibody identified a doublet with the major band at 88 kd and a minor band at 80 kd. These bands were not present on lanes run as negative controls (buffer alone or P3 instead of A3D8) and are consistent with the CD44 isoforms known to be expressed by epithelial cells and hematopoietic cells, respectively.

HUMAN

PATHOLOGY

Volume 25, No. 10 (October

1994)

FIGURE 1. The invasive carcinoma cells were more strongly positive for CD44 than the intramucosal portion of the tumor (top). Both areas showed a membrane staining pattern in this case. (A3D8; original magnification x40.)

FIGURE 2. (A) Areas of dysplasia stained less strongly than intramucosal or invasive carcinoma (A3D8: original magnification x80.1 (6) Intestinal metaplasia (top) always stained less strongly than dysplasia with a focal cytoplasmic staining pattern. (A3D8; original magnification x 132 .)

CD&l EXPRE~ION IN GASTRtC ADENOCARCINO~AS ~ashjngton

C

A

et al)

ses found at the time of surgery were more common in the patients with weak CD44 expression (nine of 12) than in patients with strong CD44 expression (five of 11) in the primary tumor. If only cases with membrane staining pattern are counted as positive, the average length of survival for those with membrane staining is 13 months, in contrast to those with no membrane staining with an average length of survival of 26.7 months, excluding the two postoperative deaths. Thus, the survival advantage of weak CD44 expression appeared to be independent of the presence of metastatic disease at diagnosis in this small study.

DISCUSSION

69

-

460

30, I

21.5 -

,

FIGURE 4. Detection of A3D8 antigen activity with Western blotting. Whole cell lysates of two gastric adenocarcinoma cell lines (Lane A, HS746t; Lane B. KATO III) were electrophoresed through la discontinuous sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose paper. The separated lanes of nitrocellulose paper were then incubated with A308 and stained using an alkaline phosphatase detection system. Normal red blood cell ghosts (Lane C) served as a positive control.

CD44 is a multiform transmembrane protein with many functions in the normal cell. The role of epithelial variants of CD44 is not well understood, although it has been hypothesized that CD44 serves to anchor the cell to hyaluronate in the basement membrane. promoting normal epithelial polarity.” For lymphocytes CD44 serves as a cell adhesion moiecule involved in homing to high endotheIia1 venules in lymph nodes.‘” Recent work has shown that CD44 may play a key role in tumor metastasis, perhaps by interaction with the extracellular matrix. A study by Gunthert et al4 suggests that high levels of a CD44 isoform (p Meta-1) with an additional 162 amino acid extracellular domain are necessary for the development of the metastatic phenotype in rat pancreatic and breast carcinoma cell lines. In addition, Heider et al’” have shown by immunohistochemistry that a human homologue of this CD44 variant is expressed in adenomatous polyps and adenocarcinoma of the colori but is expressed only weakly in crypts of normal colonic mucosa. An immull~~histoch~mical study of CD44 expression in breast cancer showed that CD44 positivity was associated with aggressive histological features, such as high mitotic count and poor differentiation.‘,5 CD44 is expressed in a variety of cell surface forms; the most common form is an 85 to 90 kd molecular weight form expressed by many cell types, including hematopoietic cells.“j Many forms are derived from a

TABLE 3.

Survival Data

WeakExpression N = 12

Survival Data The survival data for patients with gastric carcinomas are shown in Tables 1 and 3. In general, patients whose tumors showed strong expression of CD44 had a shorter mean survival duration than those whose tumors showed weak expression. The apparent survival advantage was despite the fact that abdominal metasta<

Died of carcinoma Mean length of survival Alive, NED Metastases at surgery Lymph node Other None

6 23.3 mos ?I

Strong

Expression N = 11 5

I 1.0 mos 3

9

5

8 1 3

4 1 6

FIGURE 3. HS746t gastric adenocarcinoma cells were fixed in formalin and the cell pellet was embedded in paraffin, lmmunohistochemical staining with A3D8 antibody showed strong membrane staining in all cells and diffuse cytoplasmic staining in many ceils. (A3D8: original magn~~cationx 160.) (B) Negative control (normal mouse serum; original magnification x 132.) 1047

HUMAN PATHOLOGY

Volume 25. No. 10 (October

single protein core of 37 kd by posttranslational modifications, such as the addition of chondroitin sulfate moieties that may result in cell-specific forms of the molecule.” The most common form of CD44 has a 21-23 amino acid transmembrane domain, a relatively hydrophilic external domain of 248 amino acids, and a highly charged cytoplasmic domain of 72 amino acids. Other CD44 isoforms caused by alternate splicing of the CD44 gene have been described.4~‘sg1 These variants are similar in form and contain a 132-162 amino acid insert. The form of CD44 expressed by epithelial cells and the variant p Meta-l belong to this group. In this study expression of CD44 in gastric adenocarcinomas and two gastric carcinoma cell lines was examined by immunohistochemical methods on formalin-fixed, paraffin-embedded tissue. The tissue distribution of CD44 in normal stomach was similar to that reported by other investigators using frozen sections; fibroblasts, smooth muscle, endothelial cells, and many lymphocytes were positive whereas normal gastric mucosa was largely negative. These results imply that the A3D8 antigen is preserved in routinely processed tissue, and this method is potentially useful in examining archival material. Both membrane and cytoplasmic staining was present in sections of the two adenocarcinoma cell lines. Cytoplasmic staining without a membrane pattern was observed in areas of intestinal metaplasia and dysplasia. In the normal cell CD44 is a membrane antigen. It may be argued that the purely cytoplasmic staining pattern is nonspecific and should not be interpreted as representing increased expression of CD44. Possible explanations for cytoplasmic staining in carcinoma cells are that carcinoma cells may overproduce CD44 to such an extent that not all CD44 is incorporated into the cell membrane or that aberrant forms of the molecule are produced. However, this would not explain purely cytoplasmic staining in areas of dysplasia, and further investigation with isoform specific antibodies to CD44 is indicated. Western blot analysis of whole cell lysates from these carcinoma cell lines were similar to immunoblotting results for normal cells with broad diffuse protein bands of 85 to 90 kd. In addition, KATO III cell lysates contained an additional protein band of 145 kd, which may represent an isoform of CD44 protein with chondroitin sulfate moieties or a higher molecular weight splicing variant of CD44. In this study all gastric adenocarcinomas examined were found to express CD44 more strongly than normal or inflamed gastric mucosa, and 11 were found to have a strong staining intensity by immunoperoxidase methods. The intensity of the staining increased with invasion, but not with metastasis; intramucosal portions of the carcinomas tended to be less strongly positive than the deeply invasive portion of the tumor. Lymph node and hematogenous metastases were not found to have a higher level of expression of CD44 than the invasive portion of the carcinoma, similar to data reported by Heider et al” in their study of expression of CD44 splice variants in gastric adenocarcinoma. Areas of epithelial dysplasia stained less strongly than areas of intramucosal carcinoma in the same case, and areas of intestinal

1994)

metaplasia were weaker than dysplastic mucosa. However, staining in metaplastic and dysplastic mucosa had a cytoplasmic pattern and may represent nonspecific staining. Strength of expression of CD44 did not correlate with intestinal or diffuse histological type or with degree of tumor differentiation in contrast to the results reported in breast cancer.‘” These findings suggest that expression of CD44 may be important in tumor invasion, perhaps by interaction of the molecule with the extracellular matrix. High levels of CD44 expression could facilitate invasion by tumor cell migration or increased adherence to the extracellular matrix. However, the role of CD44 in epithelial metastases is not clear. Although some studies suggest that a high level of expression of a variant form of CD44 may be important in the development of metastases, overall CD44 expression, as recognized by the A3D8 and AlG3 antibodies, did not appear to be increased relative to the invasive portion of the adenocarcinoma. However, A3D8 and AlG3 do not distinguish between the hematopoietic and epithelial forms of CD44. The possibility of expression of a variant form of CD44, or of one that reacted poorly with A3D8, in the metastatic lesions cannot be ruled out. Minor differences in CD44 expression were found in normal and inflamed gastric mucosa. In normal gastric mucosa there was little or no staining of the foveolar surface of the gastric antrum and fundus. Focal cytoplasmic positivity was present in cases of gastritis. This observation may be related to the loss of mucin commonly observed in reactive epithelium adjacent to ulcers or in gastritis because neither diffuse gastric-type mucin nor goblet cell mucin stained with these antibodies, or it may represent nonspecific staining. In two cases, one in the PUD group and one in the tumor group, antral mucous glands had a membrane staining pattern. Severe chronic active astritis was present in both of these cases. Volpes et alf 2 demonstrated a similar finding in the liver in cases of acute hepatitis. Hepatocytes are normally negative for CD44, but hepatocytes with membrane staining were found in areas of inflammation. These results suggest that CD44 expression may be upregulated in some epithelial cell types in the stomach in the setting of chronic active gastritis and PUD, similar to changes found in the liver in hepatitis. In summary, CD44 expression, evaluated by intensity of immunoperoxidase staining with the monoclonal antibodies A3D8 and AlG3, is increased in gastric adenocarcinomas relative to normal gastric mucosa. Staining intensity was greater in invasive carcinomas than in adjacent intramucosal carcinoma and dysplasia. Lymph node metastases did not stain more strongly that the primary tumor, suggesting that a key consequence of CD44 overexpression in gastric carcinomas may be the development of the invasive phenotype rather than metastases. Although the number of cases in each group was too small to allow definite conclusions about a possible correlation between CD44 expression and prognosis, weaker expression of the antigen appeared to be associated with slightly longer survival. Further study of

1048

CD44 EXPRESSION

expression explanation

‘of specific isoforms for these findings.

IN GASTRIC

may help elucidate

ADENOCARCINOMAS

the

We thank Jim Burchette for assisting in Ackno7ukdgmunf. adapting the immunohistochenkal procedure for formalinfixc*d tissue and for performing the immunohistochemical stains, and I>r Neer?ia Rd0 for assisting in performing the Western bloc assavs.

REFERENCES I. Picket 1J. Nakachr M. But&cl- EC: Monoclonal antibodies to human Ilmphocyte homing receptors define a novel class ot adhesion molcculcs on diverse cell vpes. .J Cell Biol 109:927-937. 19x9 2. Haynw BF. Trlen MJ, Hale l.P, et al: CD44-A molecule ~nvolvcd in lrukoc\,tr adherence and T-cell activation. Immunol Today 10:423-428. 19X9 3. Miyakr I(. llndrrhill CA. Lesley J. et al: Hyaluronatc can furlction as a cell adhesion molecule and CD44 participates in hvahlronatc recognition. J Exp Med 172:69-75. 1990 4. (&nthcrt I!, Hofmann M, Rudy MT. ct al: A new variant of glycoprotein CD44 confers metastatic potential to rat carcinoma cell lines. Cell 65:13-24. 1991 .i. Horst E. Meijer CJLM, Radaszkiewicz T. et al: Adhesion molecules in the prognosis of diffuse large-cell lymphoma: Expression of a lymphocyte homing receptor (CD44). LFA-1 (CDlla/l8). and ICXM-I (CD!%). Leukemia 4:595-599, 1990 6. Jalkanen S. Joensuu H. Soderstrom KO, et al: I.ymphocyte homing and clinical behavior of non-Hodgkin’s lymphoma. J Clin Invest 87:183’5-1840, 1991 7. Telen MJ. Shehata H, Haynes BF: Human medullaq thymowe ~80 antigen and In(Lu)- related ~80 antigen reside on the same protein. Hurn Immunol 17:31 l-324, 1986 8. Brigati DJ, Budgeon LR. Unger ER, et al: Immunocytochemistn is automated: Development of a robotic workstation based upon the capillary action principle. J Histotechnol 11:165-183, 1988 9. Towbin H. Staehelin T. Gordon J: Electrophoretic transfer of pr-oteins from polvaclylamide gels to nitrocellulose sheets: Procedure and some applications. Proc Nat1 Acad Sci USA 76:4350-4354. 1979 10. Telen MJ, Eisenbarth GS, Haynes BF: Human erythrocyte

1049

(Washington

et al)

antigen\: Regulation ot expansion of a novel CITthrw. vtv ,urtace antigen hv the inhibitor I.utheran In (Ix) gcnr. J (Iin Inwst 71:187X1886, 1983 11. Hcider K-H, Dammrich J? Ski-och-Angel P. t’t al: Dilfercntial expression of (X)44 splice variants in intestinal alid diffuse-type human gastric carcinc,mas and normal gastric mucosa. t.:,m Rcs i3:41!)/4203. 1993 12. Arufto A, Stamenkovic 1. Mclnick M, ct al: (:1)44 is the pl-illcipal cell surface rrceptor tor hyaluronate. (:ell 61: IX)?-1313. 1990 13. Berg EL, Goldstein W.Jutil.1 MA. ct aI: Homing I-rccptc~~~~ and vascular addwssins: Cell adhcsioll moleculrs th:11 dirrct Ivmphocyte traffic,. Immunol Rev 10X:5-18. 1989 14. Hrider K-H, Hofman M, Hors E, CI al: .I human bomologur (hi’the mtc Incraarasis-associated variant ot(:D41 is expressed in colarectal carcinomas and adenomatous polvps. J (:rll Biol 120:227-233.

1003 I!i. Jocnsuu

H, Klrmi P. ToikkaIlen S, et dl: (;I>( oprottin (:I)-14 expression and its association with wwival in bwa\t cat,< er. Am ,J Pathol 143867-854, 1993 16. Stamenkwic I, Amiot M. Prsando JM. et al: .I lymphocytr molecule implicated in lymph node homing ia .I n1cmhc.r of thr cartilage link protein famdy. Cell 56: 1057-I 062, 1SW 17. Jalkdnen S, Jalkanen M, Bargatzc R, et al: Biochemical propel-ties of glycoproteins involved in lvmphocyte recognitiolr of high endothelial venules in man. J Immunol 141:1615-1623, 19X8 18. Goldstein LA, Zhou DFH. Picker LJ. et al: 11 human lymphocvte homing receptor, the Hermes antigen, is related to cartilage proteoglycan core and link proteins. Cell 5ti:1063-1t)72, 1989 19. Sramenkovic I, Aruffo A. Amiot M, et al: The hematopoietic and epithelial foorms of CD44 are distinct polypeptides with different adhesion potentials for hyaluronate-hearing cells. EMBO J 10:333348. 1991 ?O. Dougherty GH. Lansdorp PM, Cooper DL. et al: Molecular cloning of CD44Rl and CD44R2, two novel isoforms of the human CD44 lymphocyte “homing” receptor expressed by hematopoietic cells. J Exp Med 1741-5. 1991 21. Brown TA, Bouchard T, St. John 7. vt al: Human keratinocvtes express a new CD44 core protein (CD44E) as a heparin-sulfate intrinsic membrane proteoglycan with additional exons. J Cell Biol 113:207-221, 1991 22. Volpes R, van den Oord 11. Desmet 1’1: Vascular adhesion molecules in acute and chronic-liver inflan;matmn. Hepdtolog, 15269-275, 1992