JOURNAL OF PATHOLOGY, VOL.

165 3 3 4 1 (1991)

CHARACTERIZATION OF INTEGRIN CHAINS IN NORMAL AND NEOPLASTIC HUMAN PANCREAS PETER A. HALL*, PHILIP C O A T E S ~ NICHOLAS , R. LEMOINE~AND MICHAEL A. H O R T O N ~

*Department of Histopathology, Royal Postgraduate Medical School, DuCane Road, London W12 OHS, U.K.; ?Department of Histopathology, St Bartholomew’s Hospital, West Smithfield, London EC1 A 7BE, U.K.; SICRF Molecular Oncology Group, Hammersmith Hospital, DuCane Road, London W12 OHN, U.K.; SICRF Haemopoiesis Research Group, Dominion House, St Bartholomew’s Hospital, West Smith3eld, London ECIA 7BE, U.K. Received8 January 1991 Accepted 18 April 1991

SUMMARY Integrins are a complex family of non-covalently linked heterodimeric glycoproteins which function as cell adhesion molecules, interacting with extracellular matrix molecules such as laminin, fibronectin, vitronectin, and collagen, and also having a role in intercellular adhesion. Each integrin subfamily is characterized by a common chain associated with variable a chains. We have examined, using immunohistological methods, the expression of the VLA (very late activation) family comprisingpl in association with al-6, and also a6 in association w i t h p , the L F A p c h a i n P , and the vitronectin receptor, in association with 81 or p5 and as the complex avp3. Cryostat sections of normal pancreas, pancreatic adenocarcinomas, and ampullary tumours were studied together with six pancreatic carcinoma cell lines. Normal pancreas showed expression o f p l in all parenchyma. a2 and a6 had a similar distribution whereas a3 expression was confined to ducts, including the very smallest radicles. Staining along the basement membranes of ducts was seen with 84 and the anti-vitronectin av chain receptor antibody 13C2. Islet cells failed to stain with any antibody. No staining of epithelial components was seen with antibodies to a l , a4, a5, or to the avp3 form of the vitronectin receptor (j73 and avp3 using the antibody 23C6). Pancreatic adenocarcinomas and ampullary tumours showed expression of a2, a3, (16, PI, B4, and the vitronectin receptor (av associated with PI or P5). Well differentiated cell lines (BXPC-3 and Capan-2) showed an identical phenotype, but less well differentiated lines showed variable loss of one or more integrin chains. The functional role of integrins was quantitated by adhesion assays in which we could assess the ability of pancreatic carcinoma cells to stick to the extracellular matrix substrates fibronectin, collagen, and laminin but not to fibrinogen or BSA. Binding of well differentiated cell lines to fibronectin and collagen was, in part, inhibited by exogenous RGD peptide. The distribution of integrins in normal and neoplastic cells and their functional properties were also shown to correlate well with the spatial distribution of laminin, fibronectin, and type IV collagen as shown by immunohistology in normal and neoplastic pancreas. KEY

woms-Integrins,

extracellular matrix, basement membrane, differentiation, pancreas.

INTRODUCTION Interactions between epithelia and associated mesenchymal tissues are believed to be fundamental processes during embryogenesis and development, in normal growth, and also in healing and neoAddressee for correspondence: Prof. P. A. Hall, Department of Histopathology, St Thomas’s Hospital (UMDS), Lambeth Palace Road, London, SE17EH, U.K.

0022-3417/91/090033-09 $05.00 0 1991 by John Wiley & Sons, Ltd

plasia.’,* The role of extracellular matrix components in regulating epithelial cell differentiation has been revealed by numerous in vivo and in vitro experiment^.^,^ Grobstein and others have demonstrated the important role that mesenchymal factors have in regulating the differentiation of the developing pan~reas.’,~ The response of pancreatic acinar carcinoma cells to basement membrane’ and recent observations of the induction of differentiation in pancreatic carcinoma cells’ cultured in extra-

34

P. A. HALL ET AL.

cellular matrix or on mesenchymal cells further support the role of mesenchyme in regulating cellular differentiation in the pancreas. Epithelial-mesenchymal interactions are believed to be largely mediated through extracellular matrix proteins such as collagens, laminin, fibronectin, and a wide range ofglycoproteins interacting with a range of specific receptors that have been identified on normal and neoplastic cells. Several classes of receptor have been identified including those with immuno lobulin-like structures such as N-CAM and DCC'.'' and those participating in calcium-dependent interactions termed cadherins.' The third major class of receptors is the integrin family 12.13 of non-covalently linked heterodimeric proteins which function as cell adhesion molecules, interacting with extracellular matrix molecules such as laminin, fibronectin, vitronectin, and collagen. Each integrin subfamily is characterized by a common p chain associated with different a chains. That integrin-mediated interactions are involved in cell differentiation has been suggested by their spatial distribution in tissues and demonstrated in a range of experimental model system^.'^'^ It has been proposed that loss or abnormal function of integrins (and possibly other adhesion molecules) could account for some aspects of neoplastic cell behaviour.'' We have examined the expression of integrins using immunohistochemical methods on normal pancreas, pancreatic and ampullary tumours, and in pancreatic carcinoma cell lines. We have correlated these observations with the distribution of extracellular matrix components, and have also demonstrated the ability of pancreatic carcinoma cell lines to bind to matrix components in an RGD-dependent manner.

'

MATERIALS AND METHODS Immunohistochemistry Fresh material was obtained from operations for pancreatic carcinoma, ampullary tumours (Whipple's procedure), and from cadaver organ donors. Cryostat sections (6pm) of normal pancreas (n = 4), chronic pancreatitis (n= 4), adenocarcinoma of the pancreas (n= 6), or from ampullary tumours ( n = 3 ) were cut and placed on silanecoated multiwell slides (C. A. Hendley, Essex, U.K.), air-dried, and stored at -20°C in air-tight boxes with desiccant. Cryostat sections of nude mouse xenografts of pancreatic carcinoma cell lines

were also examined. For immunostaining, sections were warmed to room temperature, fixed in acetonemethanol (1:l) for 5 min, and air-dried. Immunostaining was performed with the primary antibodies listed in Table 1'9-30using the indirect immunoperoxidase method. After incubation with primary antibody for 1 h, sections were washed three times with phosphate-buffered saline (PBS) and stained with peroxidase-conjugated rabbit anti-mouse (Dakopatts Ltd, High Wycombe, U.K.) for 1 h. After further washes in PBS, diaminobenzidineH,O, was used as chromogen and sections were lightly counterstained with haematoxylin. All chemicals and reagents were obtained from Sigma Ltd., Poole, U.K. unless otherwise stated. The distribution of laminin, fibronectin, and type IVcollagen in normal pancreas, chronicpancreatitis, and pancreatic carcinoma was determined by immunohistochemistry. The avidin-biotin complex (ABC; Dakopatts Ltd.) method was employed on 4 pm sections of formalin-fixed, wax-embedded material digested for 25 min in 0.05 per cent protease type XXIV in PBS. Primary antibodies were obtained from Gibco-BRL, Paisley, U.K. (laminin), Dakopatts Ltd. (fibronectin), and Eurodiagnostics Ltd., Reading, U.K. (type IV collagen). Cell culture

Pancreatic carcinoma cell lines (Capan-2, BXPC3, PANC-I, MIA PaCa-2, ASPC-1) were obtained from ATCC, U.S.A. The cells were grown in RPMI 1640 medium supplemented with 10 per cent fetal calf serum (Gibco Ltd., Paisley, U.K.) and incubated at 37°C in 5 per cent CO,. The phenotype of the cell lines has been extensively characterized in terms of antigenic expression and in terms of functional responses to culture on and in extracellular matrix components (Hall e f al., in preparation). For immunostaining, cells were grown on multiwell glass slides, and when just sub-confluent they were rinsed briefly in PBS and fixed in acetonemethanol (1:l) for 5 min before air-drying. Immunostaining was then performed using the indirect immunoperoxidase method as described for tissue sections. Adhesion assays

Tissue culture dishes or 96 well plates (Falcon, Becton Dickinson, NJ 07035, U.S.A.) were coated with fibronectin, laminin, fibrinogen, type IV collagen, or type I collagen (all from Sigma Ltd., Poole,

35

INTEGRINS IN THE PANCREAS

Table I-The

antibodies used and their specificities

Integrin chain

Antibody

VLA a l VLA a2 VLA a3 VLA a4 VLA a5 VLA a6

TS2/7.1.1. 12FI 5143 B5610 P1F8 135 L36

Chain structure

Ligand(s) Collagen, laminin Collagen Collagen, laminin, fibronectin Fibronectin CS- 1, VCAM- 1 Fibronectin Laminin

Reference 19 20 21 22 23 24

??

81 ln 83 84 85

25 26 27 28 29

K20 60.3 Y2/5 114 439-98 anti-C-terminal peptide

Vitronectin receptor (aV)

13C2

Vi tronec tin receptor

23C6

aV associates with /31 o r p 5 in neural and epithelial tissues to form alternate vitronectin receptors with differing ligand binding properties

wli3

Promiscuous

30

30

RESULTS U.K.). This was achieved by placing the matrix protein at a range of concentrations (0.1-100 pg/ml) on the plastic and incubating at room temperature for Integrin expression in normal and neoplastic 1 h, after which the dishes were washed extensively pancreas in PBS. The dishes were then incubated with a 1 mg/ The results of integrin expression are tabulated in ml solution of bovine serum albumin in PBS for 2 h, Table 11. Normal pancreas showed expression of81 to block potential binding sites on the plastic itself, in all exccrine parenchyma and fibroblasts. a2 and and then washed extensively in PBS. Control dishes a6 had a similar distribution whereas a3 expression were blocked with BSA without prior treatment was confined to ducts, including the very smallest with the matrix protein. radicles (Fig. 1). Staining along the basement memCells for the adhesion assay were resuspended in brane domain of epithelial cells in ducts was seen serum-free RPMI and placed in pretreated dishes or with 84, the anti-vitronectin receptor av chain plates (Falcon, Becton Dickinson). Cells were incu- specific antibody 13C2, and also 85. Islet cells failed bated at 37°C for varying periods of time or in the to stain with any antibody at the concentrations presence of GRGDSP peptide or control GRGESP employed. Pancreatic adenocarcinomas and peptide (both from Peninsula Labs Inc., Belmont, ampullary tumours showed expression of 81,84, a2, CA, U.S.A.) at l(r400 ,ug/ml (approximately 20- a3, a6, the vitronectin receptor av chain (1 3C2), and 800 PM). At the end of the experiments, unattached 85 (Figs 2 and 3). No staining of epithelial cells were washed off by repeated rinses in PBS, components was seen with 82,a l , a4, or a5. fixed in 1 per cent glutaraldehyde in PBS, and counted using a phase-contrast microscope. Alternatively, cells were stained with Giemsa and Integrin expression in pancreatic carcinoma cell counted using a conventional light microscope. The lines Well differentiated cell lines (BXPC-3 and number ofcells attached in ten medium power fields (final magnification x 400) was counted. The two Caplan-2) showed an identical phenotype to that seen in primary pancreatic neoplasms (Table 11; see methods were found to give identical results.

36

P. A. HALL E T A L .

of immunostaining o n normal and neoplastic tissues and cell lines

Table 11-Results

a2 Normal pancreas ( n = 4) Ducts Acini Islets Chronic pancreatitis (n = 3 ) Tumours ( n = 9) Cell lines BXPC-3

ASPC-I

CL X CL X CL X CL X

PANC-I PSN- 1 PaCa-2

p4

a l , a4, a5 j32,83, aV/j?3

85

V n R (13C2) aV/Bl o r pS

++ +

++ + -

-

+ + + + + + + + + ++ + ++ -

-

++

++

++

++

++

++

++

-

++

++

++

++

++

++

++

-

+ + + + F + + + + + + +

+ + + + F + + F + +/+

+ + + + + + + + + + + +

+

+ + ND + ND +

+ + + + + + + F + + +

-

X CL

81

+

+

CL

Capan-2

a6

a3

x

+ +

N D = N o t done; + + + =strong expression; F = focal expression.

-

-

+ + + F ++ F + +/+

-

ND

ND ND ND

+

ND

-

-

+ + =moderate expression; + =weak expression; - =no detectable expression;

Fig. I-Strong a3 integrin staining in all duct radicles including the smallest radicles

Fig. 2-a2

immunoreactivity in a ductal pancreatic carcinoma

INTEGRINS IN THE PANCREAS

37

Fig. 3-a3 staining of an ampullary carcinoma. Notice the intercellular staining

Fig. 4 for example of staining) but the less well differentiated lines showed variable loss of one or more integrin chains. Heterogeneous staining with only focal positivity was seen in ASPC-1, PSN-1, and MIA PaCa-2. The cell line PANC-1 showed increased expression of some integrin chains when grown as nude mouse xenografts, suggesting some form of regulation by surrounding matrix components. Functional properties of integrins in pancreatic carcinoma cell lines The functional role of integrins and other adhesion molecules was quantitated by adhesion assays in which the ability of pancreatic carcinoma cell lines BXPC-3, Capan-2, and PANC-1 to adhere and spread on a range of extracellular matrix substrates was determined. Both BXPC-3 and Capan-2 had similar properties, binding to fibronectin, laminin, and collagen type I in a time- and concentration-dependent manner (Fig. 5). Neither cell line bound to bovine serum albumin nor to fibrinogen. The ability to bind to fibrinonectin and to collagen type I could be inhibited, in part, by addition of RGD-containing peptide in a concentrationdependent manner (Fig. 6). Such data areconsistent with a role for integrins in matrix binding, although experiments with blocking antibodies will be required to demonstrate this conclusively. The cell line PANC-I also bound to collagen, laminin, and

Fig. +Easily identifiable intercellular staining with 8 1 integrin on the cell line BXPC-3 (a). Compare with negative control (b)

fibronectin in a similar manner to that reported by Cheresh et ~ 1 . Binding ~' of PANC-I cells to fibronectin, collagcn, or laminin was not altered in the presence of RGD peptide, confirming the report by Cheresh et aL3' Expression of basement membrane components in the pancreas The distribution of integrins in normal and neoplastic cells was found in a basement membrane

38

P. A. HALL ET AL.

100-

-5

A

--

80-

U

$m

-

-

6040-

0

2o0

Fibrinogen

I BSA

m

a

Laminin

Collagen I Fibronectin

EG&=&=L / 0

60

180

120

240

Time (minutes)

100 ~~~

-$

Fibronectin

80

BSA

U

-s

Fibrinogen

60

Laminin Collagen

m

40 0

20 0 0.1

0.3

1

3

10

30

Concentration (pg/ml)

Fig. 5-Capan-2 cells attach to a range of substrates in a time- (a) and substrate concentration-dependent (b) manner. Note that there is no significant binding to BSA or to fibrinogen. Similar results were obtained with the cell lines BXPC-3 and PANC-1

distribution. This correlates well with the spatial distribution of laminin, fibronectin, and type IV collagen as shown by immunohistology in normal and neoplastic pancreas. DISCUSSION In this study we have defined the pattern of integrin expression in the pancreas. The integrin phenotypes of normal ducts, ductular elements in chronic pancreatitis, and in ductal adenocarcinoma of the pancreas (in vivo and in vitro) are similar with expression of a2, 0.3, and a6 together with p l , 84, and p.5 chains. This phenotype is essentially similar to that reported in other simple epithelia including

gut,'* b r e a ~ t ,and ~ ~ kidney.35 .~~ Some epithelia and their tumours express an av chain together with a p l chain, forming a fibronectin r e ~ e p t o r ~or ~ .with ~' a p5 chain, forming a vitronectin r e ~ e p t o r , ~ ' , ~ ' , ' ~ rather than as the more usual p3 complex.40 This arrangement is seen in the pancreatic ducts; they are positive for av (monoclonal 13C2-positive), pl, and p5 but negative for 83 and av p3 complex (that is, monoclonal 23C6-negative). The distribution of a6 and p4 chains is identical to that reported recently by Sonnenberg et ~ 1 . The ~ ' distribution of integrins follows, for the most part, that of basement membrane. It is of note that no integrin expression was detectable immunohistologically in islets, except in association with capillaries. The distribution of laminin, collagen type IV, and fibronectin is similar

INTEGRINS IN THE PANCREAS

01 1

I

,

I

70

100

1000

Peptide (Pgiml)

Fig. &Binding of Capan-2 cells to fibronectin was partly inhibited by exogenous RGD-containing peptide in a concentrationdependent manner. Similar results were obtained with binding to collagen type I. The cell line BXPC-3 had similar properties but the binding of PANC-1 cells could not be inhibited

to that reported previously43 and is essentially coincident with the distribution of integrins as shown in the present study. No significant difference could be detected between expression of integrin chains in pancreatic ducts and that seen in adenocarcinomas of the pancreas and ampullary neoplasms. Pignatelli et al. " have proposed that loss of functional integrins, or other cell adhesion molecules, may be associated with neoplastic progression. Loss of integrin chains has been described in poorly differentiated breast and colorectal carcinomas.'8*33-34 In pancreatic carcinoma cell lines, there is a clear relationship between expression of integrins and other manifestations of differentiation. The current results with clinical material cannot refute this notion since the small number of tumours examined are all well differentiated. Moreover, there may not necessarily be a relation between immunologically detectable integrin expression and function.16 Finally, the role of other receptors has not yet been defined. It is worthy of note that pancreatic cancer has a remarkable propensity for perineural spread and that this could reflect a role for integrins or other cell adhesion molecules. Intercellular staining was identified with antibodies to a2 and a3 chains similar to that reported previously in keratino~ytes.'~ Here we report that intercellular staining also occurs with an antibody that recognizes the a6 chain. The functional significance of the intercellular distribution of integrin chains remains uncertain. It could conceivably represent homotypic interactions between integrins

39

and RGD sites present on integrin molecules. Alternatively, it may reflect interactions between integrins and 'bridging' multivalent ligands, although there is no evidence for the presence in intercellular spaces of the limited range of extracellular matrix proteins examined (fibronectin, laminin, or collagen). A final possibility is that the integrins present in the intercellular region are not functional, although this seems unlikely. We have shown that pancreatic carcinoma cell lines can bind to a range of extracelluar matrix molecules, and the kinetics and matrix concentration effects observed in this study are comparable to other reports.44 In two cell lines (BXPC-3 and Capan-2) which are well differentiated, both in terms of morphology and immunophenotype (Hall et al., in preparation), binding to fibronectin and collagen type I is in part mediated by R G D mechanisms, and this probably involves integrins. This inhibition is not seen with binding to laminin. The lack of binding to fibrinogen reflects the absence of the classical avp3 form of the vitronectin receptor which is known to bind f i b r i n ~ g e nR. ~G~D peptide could not inhibit binding of the poorly differentiated cell line PANC-1 to any substrate, confirming the observations of Cheresh et aL3' Using affinity chromatography, a series of collagen and laminin binding proteins have been identified from the human pancreatic carcinoma cell line PaTu II.46The existence of diverse receptors for laminin and collagen in the pancreas, and elsewhere, indicates considerable redundancy of receptors in the ability of cells to bind extracellular substrates (Hall et al., in preparation). It is of interest that those cell lines (BXCP-3 and Capan-2) capable of responding to culture in extracellular matrix components by forming duct-like structures' are those that express the full range of integrins at high levels, as assessed by immunohistochemistry. This differentiative response is inhibited, in part, by R G D peptide (Hall et al., in preparation), further suggesting the involvement of integrins in this process. While extracellular matrix molecules may interact with integrins (and other receptors) to regulate patterns of cellular gene expression and consequently differentiation, it should be noted that other molecules may play important roles. For example, it has recently been shown that TGFa can act as a morphogen inducing tubule formation in renal epit h e l i ~ r nand ~ ~ TGFa is expressed at high levels by pancreatic ductal epithelium4' and other epithelia that form tubules. In addition, there is good evidence that mesenchymal cells play an important role

40

P. A. H A L L ET AL.

in regulating epithelial cell differentiati~n.~ Consequently, it is clear that we are only beginning to characterize the molecular basis of interactions between epithelia and cellular and acellular components of mesenchyme.

19.

20.

21.

ACKNOWLEDGEMENTS

We thank Professor R. Williamson, Mr R. C . G. Russell, and Sister B. Theis for their help in obtaining clinical material and Dr Jo Adams for encouragement. We thank GenenTech for thegift of anti@ antibody. This work was supported by the Imperial Cancer Research Fund.

22. 23.

24.

REFERENCES 25.

I . Bissell MJ, Hall G H , Parry G. How does the extracellular matrix direct gene expression? J. Theor Biol 1982;99:3 1-68. 2. Liotta LA. Tumour invasion and metastasis. Cancer Res 1986; 4 6 1-7. 3. Haffen K, Kedinger M, Simon-Assman P. Mesenchyme-dependent differentiation of epithelial progenitor cells in the gut. J Pediarr Gastro Nulr 1987; 6 1 4 2 3 . 4. Del Buono R, Pignatelli M , Hall PA. Control of differentiation in a rectal adenocarcinoma cell line. The role of diffusible and cell-associated factors. J fathol (in press). 5. Golosow N , Grobstein C. Epitheliomesenchymal interactions in pancreatic morphogenesis. Des B i d 1962;4: 242-255. 6. Rutter WJ, Pictet RL, Hardin JD, Chirgwin JM, MacDonald RJ, Przybala AE. An analysis of pancreatic development: role of mesenchymal factor and other extracellular factors. In: Papaconstantinou J, Rutter WJ, eds. Molecular Control of Proliferation and Differentiation. New York: Academic Press, 1978;205-227. 7. Ingber DE, Madri JA, Jamieson JD. Basement membrane as a spatial organiser of polarised epithelia. Am JPathol1986; 122: 129-139. 8. Hall PA, Coates PJ, Del Buono R, el a/. Patterns of cellular differentiation in human fetal, adult and pathological exocrine pancreas: correlation with in vitro and xenograft studies. J Parhol 1990; 161: 319a. 9. Fearon ER. Cho KR, Nigro JM, et a/. Identification o f a chromosome I8q gene that is altered in colorectal cancers. Science 1990;247: 49-56. 10. Rutishauser U. Acheson A, Hall AK, Mann D M , Sunshine J. The neural cell adhesion molecule (N-CAM) as a regulator of cell-cell interactions. Science 1988; 240 53-57. 11. Takeichi M. The cadherins: cell-cell adhesion molecules controlling animal morphogenesis. Dev Biol 1988; 1 0 2 639455. 12. Hynes RO. Integrins: a family ofcell surface receptors. Cell 1987; 4 8 549-554. 13. Ruoslahti E, Pierschbacher MD. New perspectives in cell adhesion: R G D and integrins. Science 1987; 2 3 8 491497. 14. DeStrooper B, van den Scheurer B, Jaspers M, et a/. Distribution of81 subgroup of integrins in human cells and tissues. J Hisrochem Cyrochem 1989; 37: 299-307. 15. Menko AS, Boettinger D. Occupation of the extracellular matrix receptor, integrin, is a central figure in myogenic differentiation. Cell 1987; 51: 51-57. 16. Adams J, Watt FM. Changes in keratinocyte adhesion during terminal differentiation: reduction in fibronectin binding precedes a5Pl loss from cell surface. Cell 1990; 63: 425435. 17. Carter WG, Wayner EA, Bouchard TS, Kaur P. The role of integrins a281 and a3pl in cell-cell and cell-substrate adhesion of human epithelial cells. J CellBiol 1990; 1 1 0 1387-1404. 18. Pignatelli M, Smith MEF, Bodmer WF. Low expression of

26. 27.

28.

29. 30.

31.

32.

33. 34.

35.

36.

37. 38.

collagen receptors in moderate and poorly differentiated colorectal adenocarcinomas. Br J Cancer 1990; 61: 636-638. Hemler ME, Sanchez-Madrid F , Plotte TJ, el a/. Glycoproteins of 210,000 and 130,000 molecular weight on activated Tcells: cell distribution and antigenic relation to components on resting cells and T cell lines. JImmunol1984; 132: 301 1-3018. Santoro SA, Rajpara SM, Staaz WD, Woods VL. Isolation and characterisation of a platelet surface collagen binding complex to VLA-2. Biochem Biophys Res Commun 1988; 153: 217-223. Hemler ME, Huang C , Schwartz L. The VLA protein family. Characterisation of five distinct cell surface heterodimers each with a common 130,000 molecular weight b subunit. J Biol Chem 1987; 262: 330k3303. Hemler ME, Huang C, Takada Y, Schwartz L, Stroinger JL, Clabby ML. Characterisation of the cell surface heterodimer VLA-4 and related peptides. J Biol Chem 1987; 2 6 2 11478-1 1485. Wayner EA, Carter WG, Piotrowicr RS, Kunicki TJ. The function of multiple extracellular matrix molecules in mediating cell adhesion to extracellular matrix: preparation of monoclonal antibodies to the fibronectin receptor that specifically inhibit cell adhesion to fibronectin and react with platelet glycoprotein Ic-Ha. J CeN Biol 1988; 107: I88 1-1 89 I . Kennel SJ, Foote LJ, Falcioni R, ei a/. analysis of the tumour associated antigen TSP-I 80. Identity with aG84 in the integrin superfamily. JBiolChem 1989; 264: 15515-15521. Bernard A, Gay-Bellile V, Amiot M, Caillou B, Charbord P, Boumsell L. A novel human leukocyte differentiation antigen: monoclonal antiD44 defines a 28 k D molecule present in immature hematologic cells and a subpopulation of mature T cells. J. Immunol 1984; 132: 2338-2344. Beatty PG, LedbetterJA, Master PJ, Price TH, Hansen JA. Definition of a common leukocyte cell-surface antigen (Lp95-150) associated with diversecell-mediated functions. JImmunol1983; 131: 291 3-2918. Albert AG, Von dem Borne K R , Modderman PW, Admiral LG, Nieuwenhuis HK. Platelet antibodies. In: Knapp W, et a/. Leukocyte Typing IV. Oxford: Oxford University Press, 1990. Hemler ME, Crouse C, Sonnenberg A. Association of the VLA a6 subunit with a novel protein. A possible alternative to the common VLA 81 subunit on certain cell lines. J Biol Chem 1989; 264: 6529-6535. McLean JW, Vestal DJ, Cheresh DA, Bodary SC. cDNA sequence of the human85integrinsubunit.JCeN Bio11990;265: 1712&17131. Davies J, Warwick JMB, Totty N, Philp R, Horton MA. The osteoclast functional antigen is biochemically related to the vitronectin receptor and is implicated in the regulation of bone resorption. J Cell B i d 1989; 1 0 9 1817-1826. ChereshDA, SmithJW,CooperHM, QuarantaV.Anovelvitronectin receptor integrin (a,&) is responsible for distinct adhesive properties ofcarcinoma cells. Ceif 1989; 57: 59-69. Choy M-Y, Richman PI, Horton MA, MacDonald TT. Expression of the VLA family of integrins in human intestine. J farhol 1990; 160: 3540. Pignatelli MN, Hanby AN, Stamp GWH. Low expression of 81, a2 and a3 subunits of VLA integrins in malignant mammary tumours. Submitted for publication. Santoro SA, Statz WD, Zuter M M , McGoff MA, Sicard G A . The a2bl integrin (VLA-2) and the platelet membrane glycoprotein IIbIlla complex: implications for pathobiology (abstract A 030). J Cell Biochem 1990; Suppl 14a: 148. Korhonen M, Ylanne J, Laitinen L, Virtanen I. The ul-a6 subunits of integrins are characteristically expressed in distinct segments of developing and adult human nephron. J Cell B i d 1990; 111: 1245-1254. Bodary SC, McLean JW. The integrin beta I subunit associates with the vitronectin receptor alpha v subunit t o form a novel vitronectin receptor for a human embryonic kidney cell line. J Bio/ Chem 1990; 2 6 5 593-5941, Vogel BE, Tarone G, Giancotti FG, Gailit J, Ruoslahti E. A novel fibronectin receptor with an unexpected subunit composition (nvpl). J Biol Chem 1990;265: 59345937. Smith JW, Vestal DJ, Irwin SV, Burke TA, Cheresh DA. Purification and functional characterisation ofintegrin alpha v beta 5. An adhesion receptor for vitronectin. JBiolChem 1990; 2 6 5 11008-1 1013.

INTEGRINS IN THE PANCREAS 39. Ramaswamy H, Hemler ME. Cloning, primary structure and properties of a novel human integrin beta subunit. E M B O J 1990; 9: 1561-1568.

40. Pytela R, Pierschbacher M D , Ruoslahti E. A 125/115 I D cell surface receptor specific for vitronectin interacts with the arginine-glycineaspartatic acid adhesion sequence derived from fibronectin. Proc Nail AcadSci USA 1985; 82: 576G5770. 41. Sonnenberg A, Linders CJT, Daams JH, Kennel SJ. The a681 (VLA6) and a684 protein complexes: tissue distribution and biochemical properties. J Cell Sci 1990; 9 6 207-21 7. 42. Step MA, Spurr-Michaud S, Tisdale A, Elwell J, Gipson IK. u684 integrin heterodimer is a component of hemidesmosomes. Pror Nail Acad Sci USA 1990; 87: 897CL8974. 43. Kern HF. Zorr A, Theiss UY, e l a/. Cell biology of human pancreatic adenocarcinoma: review oftumor markers. In: Preece PE, Cuschieri A, Rosin RD, eds. Cancer of the Bile Ducts and Pancreas. Philadelphia: WB Saunders, 1989; 139-164.

41

44. Haberern CL, Kupchik HZ. Diversity of adhesion to basement membrane components of human pancreatic adenocarcinomas. Cancer Rrs 1986; 45: 5245-5251. 45. Cheresh DA, Berliner SA, Vincente V, Rugger ZM. Recognition of distinct adhesive sites on fibrinogen by related integrins on platelets and endothelial cells. Cell 1989; 5 8 945-953. 46. Mai M, Brune K, Jacoby B, Kern HF, Molenhauer J. Laminin interactions with ductal pancreatic adenocarcinoma cells: identification of laminin and collagen binding proteins. J Cell Sci 1990; 95: 65-74. 47. Taub M, Wang Y, Szczesny TM, Kleinman HK. Epidermal growth factor or transforming growth factor a is required for kidney tubulogenesis in matrigel in serum free conditions. Proc Nut/ AcadSci USA 1990; 87: 40024006. 48. Barton C, Hall PA, Hughes CM, Gullick WJ, Lemoine NR. Transforming growth factor u and epidermal growth factor in human pancreatic cancer. J Paihol1991; 163 I 1 1-1 16.