[CANCER RESEARCH 42, 601-608. February 1982]0008-5472/82/0042-OOOOS02.00

Antigens of Human Pancreatic Adenocarcinoma Cells Defined by MurineMonoclonal Antibodies1

Richard S. Metzgar,2 Melissa T. Gaillard, Steven, J. Levine, Francis L. Tuck, Edward H. Bossen, and

Michael J. Borowitz

Departments of Microbiology and Immunology (R. S. M., M. T. G., S. J. L., F. L. T.] and of Pathology [E. H. B., M. J. B.¡,Duke University Medical Center, Durham.North Carolina 27710

ABSTRACT

We have elicited and characterized the serological specificityof five murine monoclonal antibodies (DU-PAN-1, 2, 3, 4, and

5) to a human pancreatic tumor cell line, HPAF. The antibodiesare not detecting HLA-associated antigens since all of the

monoclonals failed to react with human lymphoid and myeloidcell lines and uncultured cells. All of the monoclonals exceptDU-PAN-5 reacted with four of five pancreatic tumor cell linesand two of two uncultured pancreatic tumors. An immuno-

peroxidase technique was used to determine the presence ofthe antigens detected by the monoclonal antibodies in frozensections of tumor and adult and fetal normal tissues. DU-PAN-1 antigen was detected on pancreatic tumors, and a transitionalcell carcinoma of the bladder, but was detected on no otheradult or fetal normal tissues including pancreas. One of theantigens (DU-PAN-2) defined by the monoclonals was present

on pancreatic ductal epithelial cells and showed a restricteddistribution on tumor cells from some other carcinoma patientsand on cells from certain fetal tissues. DU-PAN-3 antigen was

present on adult and fetal pancreatic cells and certain tumorcells but could not be detected on cells of other fetal or adultnormal tissues. DU-PAN-4 and DU-PAN-5 antigens have a

more widespread distribution on normal or tumor cell types.

INTRODUCTION

Cancer of the pancreas is frequently a fatal disease whichcurrently ranks fourth among causes of death due to cancer.One major obstacle in the management of pancreatic cancerhas been the lack of an effective method for early detection.Recently, advances in our knowledge of tumor-associated antigens have prompted the application of immunological techniques to the diagnosis and management of malignant disease.For example, measuring serum levels of CEA3 has been dem

onstrated to be a useful adjunct both in the diagnosis and inthe monitoring of therapy of certain cancers, including pancreatic cancer (6). Unfortunately, the presence of cross-reacting antigens in normal individuals and in patients with variousnonneoplastic diseases has limited the usefulness of CEAdeterminations as a screening test of cancer.

A number of other human pancreatic tumor-associated antigens have been characterized using various xenogeneic anti-sera. These include pancreatic oncofetal antigens (1, 4) and

' This research was supported in part by Grants CA 08975 and AM 08054

from the NIH.2 To whom requests for reprints should be addressed.3 The abbreviations used are: CEA, carcinoembryonic antigen; EMEM, Eagle's

minimal essential medium; PBS, phosphate-buffered saline (0.01 M sodiumphosphate:0.15 M sodium chloride, pH 7.2).

Received July 21, 1981; accepted November 5, 1981.

other pancreas tumor-associated antigens (5, 7-9). While

these antisera are of considerable interest, none has detecteda defined antigen with the specificity required to provide ameaningful diagnostic test. Some of the specificity problems inthese studies were probably due to the inherent difficulties inobtaining monospecific reagents from absorbed xenoantisera.Hybridoma technology has allowed production of monospecificantibodies which can be used to evaluate the distribution ofpotential tumor-associated or organ-specific antigens. In this

report, we describe the production of 5 monoclonal antibodiesreactive with human pancreatic carcinoma antigens. We thendiscuss the distribution of these antigens on a number of tissueculture cell lines and on normal, malignant, and fetal humantissues.

MATERIALS AND METHODS

Tissue Culture Cell Lines. The cell line HPAF was used for immunization of BALB/c mice for the production of the monoclonal antibodies described in this report. This line was established in our laboratory and derived from the ascitic fluid of a patient with pancreaticadenocarcinoma. The cells in culture ranged from small mononuclearcells to large multinucleate cells. Signet ring cells were noted afterstaining for mucin. Vacuole formation was a prominent feature, andthere was a striking similarity between the tissue culture cells (Fig. 1,a and b) and the cells present in the ascitic fluid of the living patient(Fig. lc). After 20 passages, most of the cultures contained predominately the large multinuclear cells shown in Fig. 1a. These culturesthen became static and could not be passaged further. In subsequentstudies from frozen stock of early passaged HPAF cells, we haveconsistently seen this effect. The largest number of passages we haveachieved with this line has been 22. The cell line is maintained onEMEM with 10% fetal calf serum and was Mycoplasma free during theentire course of these studies. Many of the other established monolayerand suspension-cultured human cells used for screening and specificity

testing have been maintained in our laboratory for years and have beenused in our previous publications. Additional monolayer cultures derived from human tumors were obtained from the following sources.Lines with a COLO designation were obtained from Dr. George Moore,Denver, Colo. Colon carcinoma cell lines 1116 and 948 were providedby Dr. Zenon Steplewski, Philadelphia, Pa. The Mia-PACA line wasobtained from Dr. Adel Yunis, Miami, Fla., and the CAPAN-1 andCAPAN-2 lines were obtained from Dr. Jörgen Fogh, New York, N. Y.

Two lines derived from melanoma patients (LR and CT), line MB derivedfrom a patient with lung cancer, and line MS derived from a stomachcancer patient were obtained from Dr. Seigler and Dr. Stuhlmiller ofour institution.

Nude Mouse Studies. Monolayers of HPAF cells (passage 5) weretrypsinized, and the cells were washed 3 times in sterile 0.9% NaCIsolution. The cells (6 x 105 to 1.5 x 107) were resuspended in 0.1 ml

0.9% NaCI solution and injected s.c. into a posterior flank position ofcongenially athymic nu/nu mice. Tumors were palpable after 1 weekand reached a 1-cu cm mass after 3 to 4 weeks. As few as 6 x 105

FEBRUARY 1982 601

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

R. S. Metzgar et al.

HPAF-cultured cells were able to produce tumor in some of the animals,whereas all animals and sites receiving 1.25 x 106 cells developed

tumor by 3 weeks.The tumor formed mucin-producing glands or ducts when introduced

into nuce mide (Fig. 2a). The nude mouse tumor resembled the histology of the liver metastasis of the patient (Fig. 20). The HPAF tumor hasthus far been serially passaged 3 times in nude mice with no diminutionof tumor formation.

Production of Monoclonal Antibodies. BALB/c mice were given¡.p.injections with 1 to 5 x 106 HPAF tissue culture cells. The animals

were given 2 or 3 i.p. booster injections 2 to 3 weeks apart and a finalinjection 3 to 4 days prior to the fusion experiments. Spleen cells fromthese animals were fused with either P3 x 63 Ag8 or PS-NS1 -1 -Ag14-

1 BALB/c myeloma cells according to the method of Galfre ef a/. (3).Fused cells were plated in 96-well flat-bottomed plates in EMEM

containing hypoxanthine:aminopterin:thymidine (3) at cell concentrations between 2.5 x 105/ml and 106/ml. Plates with growth in 20 to

40% of the wells were selected for antibody-screening assays.

Hybrid cultures selected for cloning were suspended at densitiesranging from 10' /ml to 103/ml in 1.66% methylcellulose (Methocel Me

4000 cP; Tridom-Fluka, Hauppage, N. Y.) in EMEM containing hypo-

xanthine:aminopterin:thymidine. Colonies were picked for subculturingafter 10 to 14 days.

Cloned hybrid cells were used to produce ascitic fluid in BALB/cmice by i.p. injection of 1 to 5 x 106 hybrid cells.

'"(-labeled Antiglobulin Binding Assay. Target cells were attached

with poly-L-lysine (50 fig/ml) to 96-well flexible polyvinyl chloride plates(Dynatech Laboratories, Inc., Alexandria, Va.) for 1 hr at 4°.The plates

were then washed twice with PBS and then incubated for 30 min withPBS with 1% normal goat serum to reduce the amount of nonspecificbinding during the assay. The wells were then washed twice with PBS,and 50 fil of hybridoma supernatant were added to each well. Theplates were incubated at room temperature for 1 to 2 hr, washed 3times with PBS, and incubated for an additional 1 to 2 hr with 125I-labeled goat anti-mouse F(ab)'2. Plates were again washed 3 times

with PBS and dried, and each well was counted for radioactivity (cpm).Indirect Immunofluorescence. Test cells (106) were washed 3 times

with PBS containing 1% bovine serum albumin and 0.02% NaN3.Pellets were resuspended in 100 fi\ hybridoma supernatant and incubated at room temperature for 30 min. The cells were then washed 3times with PBS buffer and incubated for 30 min at room temperaturewith fluorescein isothiocyanate-conjugated goat anti-mouse (7S) im-

munoglobulin.When monolayer cultures were used, the tissue-cultured cells were

removed from the flask by incubation with 0.25% trypsin and thenwashed and resuspended in EMEM with 10% fetal calf serum for 30 to60 min. It was established that trypsin treatment of HPAF cells did notalter their ability to react with the monoclonal antibodies used in thisreport. Cells that were removed from the monolayer with a rubberpoliceman and then dispersed manually gave the same titers andintensity of fluorescence as the trypsinized cells that were resuspendedin EMEM with 10% fetal calf serum.

Immunoperoxidase Testing. A modified, indirect immunoperoxi-

dase procedure was used (2). After preincubation with 10% nonimmunegoat serum, tissue sections were incubated for 60 min at room temperature with supernatant from either one of the hybridoma cultures orthe P3 myeloma cell line. A second 45-min incubation was performedwith a 1:50 dilution of peroxidase-conjugated goat anti-mouse IgG

(Cappel Laboratories, Inc., Cochranville, Pa.). Slides were developedby incubation for 15 min with a solution of diaminobenzidine (SigmaChemical Co., St. Louis, Mo.) (0.5 mg/ml):hydrogen peroxide(0.001%) in 0.05 M Tris buffer, pH 7.6, and counterstained withhematoxylin.

Collection of Tissue. Normal and malignant adult tissues werecollected from surgery in the fresh state and were processed asdescribed below within 4 hr after surgical removal. In all cases, ahistological diagnosis was made by the attending surgical pathologist.

Fetal tissues were obtained from a prostaglandin-induced aborted fetusof 20 weeks gestation and processed after overnight storage at 4°.

Tissue Processing. Tissue was cut into small blocks and embeddedin an isotonic solution of 7.5% gelatin. The blocks were snap frozen bysubmersion in isopentane precooled in liquid nitrogen. Sections werecut at 5 to 7 firn thickness on a microtome-cryostat (American OpticalCorp., Buffalo, N. Y.), thaw mounted onto gelatin-coated slides, andimmediately fixed in acetone at —¿�20°for 15 to 30 sec. Slides wereeither stored at -70° or used immediately for immunoperoxidase

staining. One section was stained with hematoxylin and eosin andreviewed by one of us (M. J. B.) to ensure that our frozen sample wasrepresentative of the original material.

Isotyping of Murine Monoclonal Antibodies. Fluids from spentmedium from cloned hybrid cultures were tested by immunodiffusionwith rabbit antisera to mouse lgG1, lgG2, IgM, and IgA. In order toconfirm the immunodiffusion data and to distinguish lgG1 from theparent P3 x 63/Ag8 line from lgG1 from hybrid cultures, HPAF cellswere sensitized with the culture fluids and/or ascitic fluids from thecloned lines and tested by indirect immunofluorescence with fluorescein isothiocyanate-conjugated rabbit antisera to mouse lgG1, lgG2,

IgM, and IgA. Antisera were obtained from Meloy Laboratories, Inc.,Springfield, Va.

RESULTS

Production and Initial Testing of Monoclonal Antibodies.More than 200 hybrid cultures from each of 2 fusions of spleencells of immunized mice with NS-1 and P3 cells, respectively,

were tested by the antiglobulin binding assay. One culture fromthe NS-1 fusion and 4 cultures from the P3 fusion demonstrated2-fold or greater binding to HPAF cells and no significantbinding to T- and B-lymphoblastoid cell lines. These 5 positive

culture fluids also reacted with HPAF cells by indirect immunofluorescence. The positive hybrid cultures were then clonedin methylcellulose. At least 2 clones from each initial culture,when grown, had culture fluid which showed the same serolog-ical specificity as that from the parent lines, and these wereselected for further studies. Both subclones from each of theoriginal 5 hybrid lines showed identical serological reactivity insubsequent specificity testing. In this report, results will begiven for a single subclone of each of the 5 lines. Ascitic fluidobtained from mice given i.p. injections with the cloned linesshowed the same specificity as the culture supernatants. Thedata in this report are from culture supernates only. Indirectimmunofluorescence titers of culture supernates from the various clones against HPAF cells ranged from 5 to 50, while titersof ascites were 100 or greater.

The isotypes of the murine monoclonals as determined byimmunodiffusion and immunofluorescence analysis are: DU-PAN-1 and DU-PAN-3, lgG2; DU-PAN-4 and DU-PAN-5, lgG1;and DU-PAN-2, IgM.

Reactivity with Tissue Culture Cell Lines. Tissue culturesupernatant fluids of the 5 clones reactive with HPAF cellswere tested by indirect immunofluorescence with a variety ofmonolayer and suspension-type tissue culture cell lines. Anantiserum dilution in 5- to 10-fold excess of the end point titer

with HPAF cells was used.The following lymphoblastoid cell lines were nonreactive with

all 5 monoclonal antibodies: Molt-4 (T-cell line); SB and RSM(B-cell lines); HL-60 (myeloid type); K562 (erythroid-myeloid

type); and U937 (monocytic type). The reactivity of the antibodies with monolayer cultures derived from various humantumors and with skin fibroblasts of a normal donor is shown in

602 CANCER RESEARCH VOL. 42

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

Human Pancreatic Adenocarcinoma Antigens

Table 1

Indirect immune/fluorescence reactivity of pancreatic tumor monoclonal antibodies with human tissueculture cell lines

Reactivity" of monoclonal antibody

CelllinesHPAFCOLO

357MIA-PACA-2CAPAN-1CAPAN-2Derived

from othertumorsA204Fibrosarcoma

(DFS-CL)Melanoma1(LR)Melanoma2(CT)Melanoma3 (COLO381)Melanoma4 (COLO298)Lung

carcinoma 1(MB)Lungcarcinoma 2 (COLO338)Stomach

carcinoma(MS)Renalcarcinoma (COLO293)Coloncarcinoma 1 (COLO206F)Coloncarcinoma 2 (COLO320)Coloncarcinoma3(1116)Coloncarcinoma 4(948)Normal

skin fibroblastDU-PAN-13

+2+02+2+000000000000000DU-PAN-23+3+03+3+00000002+00001

+i-t

eiDU-PAN-32

+1+1+1+1

+1-2

+01

+1+001

+2+NTb0001

+1+0DU-PAN-43

+3+3+3+3+3+NT2+3+2

+1+3+3+NT02+03+3+0DU-PAN-53

+000001-2

+000000000002

+2-3

+

Intensity of fluorescence graded from weak but distinct fluorescence over controls, 1 + ; moderateintensity. 2 + ; strong bright fluorescence. 3 + ; no increase in fluorescence over control. 0.

NT. not tested.

Table 1. DU-PAN-1 antibodies showed reactivity with 4 of 5

cell lines derived from pancreatic tumors and no reactivity withany other tumor-derived cell line tested. DU-PAN-2 antibodies

also reacted with 4 of 5 pancreatic tumor cell lines but showedsome cross-reactivity with one lung carcinoma cell line andweak reactivity with 2 colon carcinoma cell lines. DU-PAN-3monoclonal reacted to some degree with all 5 pancreatic tumor-

derived cell lines and with 7 of 13 cell lines from patients withother tumor types. DU-PAN-4 monoclonal was strongly reactivewith all 5 pancreatic tumor lines and was broadly cross-reactivewith cell lines from most but not all other tumors tested. DU-PAN-1 , 2, 3, and 4 antibodies failed to react with normal skinfibroblasts. DU-PAN-5 antibodies did react with normal skinfibroblasts but showed reactivity with only 2 of 13 nonpan-creatic tumor-derived and 1 of 5 pancreatic tumor-derived celllines. DU-PAN-1 antibodies reacted with 70% of HPAF cells,and DU-PAN-3, 4, and 5 antibodies reacted with 90 to 100%of these cells. In contrast, DU-PAN-2 antibodies reacted withonly 10% of HPAF cells but reacted with 60 to 90% of CAPAN-

1 cells.All 5 monoclonal antibodies failed to react with enriched T-

and B-blood lymphocytes from 10 normal donors or with bloodGroups A and B erythrocytes.

Immunoperoxidase Reactivity with HPAF Cells and NudeMouse-passaged HPAF Cells. In order to determine the reactivity of the monoclonals with uncultured human adult and fetaltissues, we investigated the conditions which would allow us touse the indirect immunoperoxidase technique on fixed cryostattissue sections. Cryostat sections of pelleted HPAF cells andtumors of HPAF cells in nude mice were subjected to short-term cold-acetone fixation and immunoperoxidase testing asdescribed in "Materials and Methods." The sectioned cell

pellets failed to react with the P3 x 63 Ag8 control but reactedstrongly with all 5 monoclonals. The nude mouse HPAF tumorreacted almost identically (Fig. 3), failing only to react with the

DU-PAN-5 antibody. This indicates that these antigens werestable to this short-term fixation procedure. The DU-PAN-2

monoclonal antibodies, which reacted with 10% of HPAF cellsby ¡mmunofluorescence, reacted with a similar percentage ofcells in sections of the pelleted tissue culture cells but reactedwith most of the adenocarcinoma cells in the HPAF nude mousetumor sections.

Immunoperoxidase Reactivity with Pancreatic Tumor, Normal Adult Pancreas, and Fetal Pancreatic Tissue. The 5monoclonal antibodies were then tested by the immunoperoxidase technique with sections from 2 pancreatic tumors, normaladult pancreas and fetal pancreas (Table 2). A dilution of eachmonoclonal antiserum was selected for these specificity studieswhich was in 5- to 10-fold excess of the end point titer with the

HPAF nude mouse tumor sections. Pancreatic carcinoma 1was metastatic to the liver. Four of the 5 monoclonals werereactive with the adenocarcinoma cells, and all failed to reactwith the normal hepatocytes or connective tissues in the samesection (Fig. 4). Pancreatic carcinoma II was obtained from apancreatic mass. All of the monoclonals except DU-PAN-5

reacted with the lumenal surface of the glands formed byadenocarcinoma cells, and in some cases, there was reactivitywith the intralumenal contents. Again, the connective tissuecomponents of the sections failed to react with any of themonoclonals.

DU-PAN-2, 3, and 4 were reactive with normal adult pancreas. DU-PAN-2 and 3 reacted only with ductal epithelial cells,whereas DU-PAN-4 (Fig. 5, b and c) reacted with ductal epithelium and some acinar cells (Table 2). DU-PAN-1 antibodies

failed to react with normal pancreas by the immunoperoxidasemethods (Fig. 5a). The various cells in the Islets of Langerhansfailed to react with any of the monoclonal antibodies.

The pattern of reactivity of the monoclonals with fetal pancreas was in most cases different from that seen with eitheradult normal pancreas or pancreatic carcinoma. DU-PAN-4

FEBRUARY 1982 603

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

R. S. Metzgar et al.

reacted intensely with fetal pancreas, as it had with othernormal pancreatic tissues and tumors. However, the antigendetected by the DU-PAN-5 monoclonal, which was not detectedon either of the carcinomas or on normal pancreas, was presenton fetal pancreas. The antigens detected by DU-PAN-1 andDU-PAN-2 monoclonals, which were present on pancreaticcarcinomas and normal pancreas (DU-PAN-2), could not bedetected on fetal pancreas. DU-PAN-3 antibodies reacted

weakly with fetal pancreas. Specific ductal versus acinar localization could not be determined with certainty in the sectionsof 20-week fetal pancreas.

Tissue sections from the following additional normal adulttissues were tested with the monoclonal antibodies: colon;small intestine; lung; kidney; breast; prostate; salivary gland;bladder; liver; lymph node; blood vessels; fat; smooth muscle;and endometrium. Antibodies DU-PAN-1, 2, and 3 failed toreact with any of the cells from these tissues. Antibody DU-PAN-5 reacted with prostatic epithelium, salivary gland acini,and renal tubules and weakly with small intestine. DU-PAN-4

antigen had the broadest distribution. It was present on renalglomeruli, breast, prostate, lung, endometrium, and blood vessels of many tissues.

Similarly, the following tissues from a 20-week fetus were

tested by the immunoperoxidase assay: colon; small intestine;stomach; lung; kidney; liver; salivary gland; thyroid; thymus;and blood vessels. DU-PAN-1 and 3 monoclonals failed toreact with any cells of the fetal tissues tested. DU-PAN-2

antibodies reacted only with secretory cells of the fetal smallintestine and salivary gland. The DU-PAN-5 monoclonal re

acted only with tubular epithelial cells in the fetal kidney,whereas DU-PAN-4 antibodies reacted with glomeruli.

Immunoperoxidase Reactivity with Nonpancreatic Tumors. The 5 monoclonal antibodies were tested by the immu

noperoxidase technique on tissue sections of a wide variety ofnonpancreatic tumors (Table 3). DU-PAN-1 antibodies reacted

only with cells from one of 2 patients with transitional cellcarcinoma of the bladder. DU-PAN-2 antibodies showed nocross-reactivity with any of the other tumors tested. The DU-PAN-3 monoclonal gave weak but distinct reactivity with one

of 2 lung and renal carcinomas and one of 3 colon carcinomas.DU-PAN-4 antibodies showed strong cross-reactivity with tu

mor cells from many different types of epithelial tumors butfailed to react with specimens of other carcinomas of the samehistological type. They were also negative when tested against2 liposarcomas and 2 lymphomas. DU-PAN-5 antiserum re

acted only with tumor cells from 2 of the 3 renal carcinomapatients.

DISCUSSION

The 5 monoclonal antibodies to the pancreatic tumor cell lineare directed to non-HLA antigens, since they fail to react with

a variety of cells which express antigens coded for by thevarious loci of this region. The antigens detected by each ofthe 5 monoclonals appear to be different by virtue of theirunique distribution on the various target cells used in thisreport. The failure of DU-PAN-1 and DU-PAN-2 antibodies to

react with all HPAF cells by indirect immunofluorescence suggests that either some of the antigens are differentiation-type

antigens which are cell cycle dependent or the HPAF line is aheterogeneous cell population. We are attempting to clone theHPAF line in order to partially resolve these 2 alternativeexplanations. The differences in serological specificity as exemplified by antigen distribution are summarized in Table 4. Itmust be emphasized that, although a significant number of celllines, tumors, and normal tissues have been tested to establish

Table 2

Immunoperoxidase reactivity of pancreatic tumor monoclonal antibodies with pancreatic cancer andnormal and fetal pancreas

Reactivity3 of monoclonal antibodies

TissuePancreatic

carcinomaIPancreaticcarcinoma II

AdultpancreasFetalpancreasDU-PAN-12100DU-PAN-2321*0DU-PAN-322

2"1DU-PAN-433 33DU-PAN-500 02

0, negative, (no enzyme reactivity noted over P3 control); 1. weak but distinct localization of enzymereactivity in certain cells of the tumor or tissue: 2. medium intensity staining of certain cells in the tissuesections: 3. strong staining of certain cells in the tissue sections.

6 Reactive with ductal epithelium only.

Table 3

Immunoperoxidase reactivity of pancreatic tumor monoclonal antibodies with nonpancreatic tumorsReactivity3 of monoclonal antibodies

TumortypeBladder

carcinomaColoncarcinomaLungadenocarcinomaRenalcarcinomaBreastcarcinomaEsophagealcarcinomaMalignantmelanomaLiposarcomaLymphomaNo.

tested232221222DU-PAN-13/1600000000DU-PAN-2000000000DU-PAN-301/11/11/100000DU-PAN-43/13/13/13/13/13/13/100DU-PAN-50003/200000

0, negative (no tumor of this category showed any reactivity over P3 controls): 1, weak but distinctlocalization of enzyme reactivity in certain cells of the tumor or tissue; 2, medium intensity staining of certaincells in the tissue sections; 3. strong staining of certain cells in the tissue sections.

0 Intensity of immunoperoxidase reaction/number of this type of tumor showing this reactivity.

604 CANCER RESEARCH VOL. 42

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

Human Pancreatic Adone-carcinoma Antigens

Table 4Summary of immunoperoxidase reactivity of monoclonal antibodies to human

pancreatic tumor cells

TissueNormal

adultpancreasFetal

pancreasNormaladulttissuesNormal

fetaltissuesPancreatictumorsNonpancreatictumorsDU-PAN-10000+1/186Reactivity"DU-PAN-2±00++0/18of

monoclonalantibodiesDU-PAN-3

DU-PAN-4+

++

+0+0

++

+3/18

7/18DU-PAN-50+++02/18

0. no reactivity with any of the cells or tissues tested; ±.weak reactivitywith some of the cells or tissues tested; + , distinct reactivity with some of thecells or tissues tested.

Number of tumors showing distinct reactivity with the monoclonal antibodies/number of tumors tested.

the general distribution of the antigens defined by these mon-

oclonals and antibodies, the number of patients or donorsstudied in each category is low. Thus, the true distribution andpresence of some of the antigens may be evident only aftermore detailed immunoperoxidase testing of normal and tumortissues from a large number of donors. This is especially truefor establishing the distribution of the antigens on pancreatictumors. Only 2 pancreatic adenocarcinomas have thus farbeen available for serological studies. The relatively low incidence of the tumor and the inoperable nature of the diseasemake tissue acquisition difficult.

DU-PAN-1 antigen has thus far been detected only on tumor

cells and not on normal adult or fetal tissues. Moreover, thedistribution of this antigen on tumor cells is quite restricted.Both pancreatic tumors reacted, but only one of 18 nonpan-

creatic tumors has the antigen. Although the reactive tumorwas a transitional cell carcinoma of the bladder, a secondbladder tumor of this same histological cell type was negative.

DU-PAN-2 antigen was weakly detected on normal adult

pancreatic ductal epithelial cells, on villi from the small intestinein the fetal intestinal tract, and on fetal salivary gland cells. Thisantigen was not detected on any of the 18 nonpancreatictumors thus far studied but was found on some nonpancreatictumor cell lines. DU-PAN-2 antibodies may detect an organ-

specific antigen of the pancreas, which on tissues is presentonly on pancreatic ductal epithelial cells. Since most pancreaticexocrine tumors are of ductal cell origin, tumors of this typeshould express the antigen. Since the antigen is present onfetal small intestine cells, it is possible that some other digestivetract tumors could also express the antigen. We have not yethad the opportunity to study an example of the rare primaryadenocarcinoma of the small intestine; therefore, we simplymay not have seen the appropriate histological tumor type thatmight express this antigen.

The DU-PAN-3 antigen may also be an example of an organ-

specific antigen associated with the pancreas. It is present onpancreatic ductal epithelial cells, but unlike DU-PAN-2 antigen,

it is expressed on fetal pancreatic cells and not on other fetalgastrointestinal tract cells. The DU-PAN-3 antigen distributionon tumor cells is also different from that of DU-PAN-2. The DU-PAN-3 antigen is present on tumor cells of one example eachof colon, lung, and renal carcinoma in addition to pancreatic

tumor cells. The DU-PAN-3 antigen was absent from tumor

cells from the other 15 cancer patients thus far studied.Of the 5 monoclonals tested, the antibody DU-PAN-4 detects

an antigen that is least specific for pancreatic carcinoma.Judging from the intensity of immunoperoxidase staining, however, it detects either a highly antigenic molecule or a substance present in high concentration on reactive cells. Itsdistribution is broad but not ubiquitous, and it seems particularly well represented on exocrine glands of both fetal andadult tissues including pancreas, prostate, breast, and salivaryglands. It is also present on blood vessels in many organs,including renal glomeruli. The significance of a shared antigenon these tissues is not clear. DU-PAN-4 antigen is also present

on 7 of 18 carcinomas, but there was no simple relationshipbetween DU-PAN-4 positivity of a tumor and its cell of origin.

DU-PAN-5 antigen appears not to be pancreatic tumor re

lated, since it could not be detected on cells from the 2pancreatic carcinoma patients studied and was detected onnormal adult and fetal tissues as well as on 2 of 18 nonpancreatic tumor cells. It is curious that this antigen is present onthe HPAF cell line but is lost when HPAF is grown as a tumorin nude mice. In fact, the only tumors in which it was detectedwere 2 examples of renal cell carcinomas. Among normal andfetal tissues, the antigen has a fairly restricted distributionwhich includes renal tubules. Thus, this reagent may be helpfulin studies of renal carcinomas.

Preliminary studies on the molecular characterization of theantigens detected by the 5 monoclonals indicate that DU-PAN-

3 antigen is a glycopeptide with a molecular weight of 70,000.The DU-PAN-2 antigen also appears to be a molecule with

some indication that sialic acid is essential for its antigenicity.Treatment of HPAF cells with neuraminidase resulted in loss ofactivity with the DU-PAN-2 monoclonal antibodies but did not

inhibit the serological reactivity of the cells with the other 4monoclonals. Radioimmunoprecipitation and polyacrylamidegel electrophoresis studies demonstrated an antigen band inthe region of the gel with a molecular weight of 110,000 forDU-PAN-5. No data are yet available on the nature of the DU-PAN-1 and DU-PAN-4 antigens. Details of the molecular characteristics of the antigens defined by the DU-PAN monoclonals

will be included in separate reports. The antigen data are toopreliminary for a reliable comparison with the antigens definedpreviously by the various polyclonal xenoantisera (5, 7-9).

Although only the DU-PAN-1 antigen thus far meets thecriteria for being considered a tumor-associated antigen, DU-PAN-2 and DU-PAN-3 antigens are not yet found on normal

adult tissues other than pancreas and thus may be valuableorgan- or tissue-specific markers of adenocarcinomas. How

ever, as was the case with CEA, the distribution or specificityof the antigen may well change as the techniques of antigendetection evolve. The serological testing thus far has been byimmunoperoxidase and immunofluorescence techniques, bothof which are difficult to evaluate in terms of their sensitivity ofantigen detection. Absorption experiments with adult and fetalpancreas are hampered by problems of proteases present inthis tissue and the difficulty in getting adequate suspensions ofa given cell type, especially ductal epithelial cells. Futurestudies with inhibition radioimmune assays combined with antigen identification and purification will lead to more sensitiveand quantifiable assays. We have already established an inhibition radioimmune assay for DU-PAN-2 antigen and have

FEBRUARY 1982 605

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

R. S. Metzgar et al.

detected antigen in serum and ascites from patients with pancreatic, breast, ovarian, and colon carcinomas.4

REFERENCES

1. Banwo, 0., Versey, J., and Hobbs, J. E. New oncofetal antigen for humanpancreas. Lancet, 1: 643-645, 1974.

2. Borowitz, M. J.. Croker, B. P., and Metzgar, R. S. Comparison of histologieand immunologie heterogeneity of non Hodgkins lymphomas. Am. J. Pathol.,105: 97-106, 1981.

4 This work, "Radioimmune Assay Studies of Serum and Ascites from Pan

creatic Cancer Patients for an Antigen (DU-PAN-2) Detected by a MonoclonalAntibody" by Vicki N. Daasch and Richard S. Metzgar, was published as an

abstract and presented orally at the joint meeting of the American PancreaticAssociation. Inc., and the National Pancreatic Cancer Project. November 5 and6, 1981. in Chicago, III.

3. Galfre, G., Howe, S. C., Milstein, C., Butcher, G. W.. and Howard, J. C.Antibodies to major histocompatibility antigen produced by hybrid cell lines.Nature (Lond.). 266. 550-552, 1977.

4. Gelder, F. B., Reese, C. J., Moossa, A. R., Hall, T., and Hunter, R. Purification, partial characterization, and clinical evaluation of a pancreatic oncofetal antigen. Cancer Res., 38: 313-324, 1978.

5. Holyoke, E. D., Douglass, H. O., Goderosen, M. H., and Chu, T. M. Tumormarkers in pancreatic cancer. Semin. Oncol., 6: 347-356, 1979.

6. Ona, F. V., Zamcheck. N., Dhar, P.. Moore. T., and Kupchik, H. Z. Carci-noembryonic antigen (CEA) in the diagnosis of pancreatic cancer. Cancer(Phila.), 37. 324-327, 1972.

7. Schultz, D. R., and Yunis, A. A. Tumor associated antigen in human pancreatic cancer. J. Nati. Cancer Inst., 62. 777-785, 1979.

8. Shimano, T., Loor, R. M., Papsidero. L. D., étal.Isolation, characterization,and clinical evaluation of a pancreas cancer associated antigen. Cancer(Phila.), 47: 1602-1613. 1981.

9. Wood, R. A. B., and Moossa, A. R. The prospective evaluation of tumorassociated antigens for the early diagnosis of pancreatic cancer. Br. J.Surg.. 64: 718-720, 1978.

606 CANCER RESEARCH VOL. 42

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

Human Pancreatic Adenocarcinoma Antigens

l .]Q

T. »• oi••;.•:-

¿*.* •¿�

•¿�

5'* ^'-^^i' * W *»'**J tt * «7^:; ^

t *í¿-Af •¿�'/•.'*v?«£:#?JS¿r x -t* -f' .~v>.'._

Fig. 1. a, HPAF tissue culture. 2 multinucleate HPAF cells are seen. One has large multiple vacuoles while the other has small vacuoles giving the cytoplasm afoamy appearance. Smaller mononuclear and multinuclear cells are also present. Papanicolaou stain, x 250. t>. HPAF tissue culture growth during passage 16. H& E, x 25. c, vacuolated tumor cell. Cells stand out against a background of inflammatory cells From ascitic fluid obtained during life. Papanicolaou stain, x 400.

Fig. 2. a. histology of nude mouse HPAF tumor tissue culture cell implant. H & E x 250. b, histology of liver tumor biopsy of patients. Compare with Fig. 2a. H& E, X 250.

FEBRUARY 1982 607

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

R. S. Metzgar et al.

- ., .,y- .

-Â̂»'....--r1

•¿�í"í rt*0f•¿�>•i* "*"»'•*•

-rtT#«*

^- •¿�

>•*sát ^

VV\ A •¿�^

.

5a

^v-v-.--^'^; -t -ív:> ' •¿�¿Õ::--Õ*J

*\Of * x ' •¿�»'f/ -W •¿�•'«"V*

^^•^^-'-•-•A^l^Jt;íSv^fb

Fig. 3. Frozen section of an HPAF tumor in a nude mouse stained by immunoperoxidase with DU-PAN-1 antibody (a) and with P3 control (£>)Note the irregularconfiguration of glands, most of which show intense staining for DU-PAN-1. Hematoxylin counterstain, x 400.

Fig. 4. Frozen section of pancreatic carcinoma metastatic to liver stained by immunoperoxidase with DU-PAN-2 antibody. Clusters of positively staining tumorglands (arrows) are shown invading the nonstaining hepatic parenchyma. At higher magnification (inset), the apical staining pattern of a single tumor gland can beappreciated. Hematoxylin counterstain, x 1700; inset, x 520.

Fig. 5. Frozen section of normal pancreas stained by immunoperoxidase with DU-PAN-1 (a) and DU-PAN-4 (b and c) antibodies. There is no significant stainingfor DU-PAN-1 antigen, but DU-PAN-4 antigen is present on small ductules (short arrows) and on the apical portion of larger ducts (long arrow). At higher magnification(c). clusters of acinar cells (A) also show DU-PAN-4 staining, a and b. hematoxylin counterstain. x 100; c, x 325.

608 CANCER RESEARCH VOL. 42

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from

1982;42:601-608. Cancer Res   Richard S. Metzgar, Melissa T. Gaillard, Steven J. Levine, et al.   by Murine Monoclonal AntibodiesAntigens of Human Pancreatic Adenocarcinoma Cells Defined

  Updated version

  http://cancerres.aacrjournals.org/content/42/2/601

Access the most recent version of this article at:

   

   

   

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

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/42/2/601To request permission to re-use all or part of this article, use this link

Research. on February 28, 2021. © 1982 American Association for Cancercancerres.aacrjournals.org Downloaded from