American ournal oj/Patlolog,tb ol. 143, NAo 1,/u/v 1993

copyright ©) American Societ/obr- Investigative Pathology,

Derivation of Ductlike Cell Lines from aTransplantable Acinar Cell Carcinoma of theRat Pancreas

0. S. Pettengill,* R. A. Faris,t R. H. Bell, Jr.,tE. T. Kuhlmann,* and D. S. Longnecker*Department of Pathology, * Dartmouth Miedical School.Lebanon, Neuw Hampshire; Department of MedicalOncology,t Rhode Islan-d Hospital-Brown U1niversity,Providence, Rhode Island; anid Department of Surgery,+Unziversity of Cincinnati, Cincinnati, Ohio

Two cel lines were derivedfrom a transplantableacinar cell carcinoma that had been establishedfrom a primary carcinoma ofthe pancreas in anazaserine-treated Lewis rat. The cultured tumorceUs initially produced amylase, but productionof exocrine enzymes ceased after 1-2 weeks inculture. The cultured cells were tumorigenic inLewis rats, and one line produced solid tumorscomposed of ductlike structures surrounded bydense fibrous tissue. The second cell line pro-duced partially solid and partially cystic tumorswith a mixedphenotype ofsquamous, mucinous,and glandular areas when it grew in vivofollow-ing regrafting. Both ceU lines lost structural andimmunohistochemical acinar cell markers whileacquiring duct cell markers during culture andregrafting. These studies provide strong supportfor the hypothesis that ductlike carcinomas canarisefrom neoplastic pancreatic acinar cells inrats. (AmJ Pathol 1993, 143:292-303)

Approximately 75% of carcinomas of the human exo-crine pancreas are classified as ductal adenocarci-nomas, while several other variant histological typesare also considered to be of ductal origin.1 As animalmodels of pancreatic carcinoma have been devel-oped, distinct species differences in the response tochemical carcinogens have become apparent. Bothspontaneous and chemically induced carcinomas inrats are predominantly of acinar cell phenotype,whereas in hamsters, most carcinomas are of ductal

phenotype. Accordingly, most carcinomas induced inrats by azaserine are of the acinar cell type.2 As thisand other rat models were described, the question ofrelevance to the human was raised because of theacinar cell phenotype.

Since 80% of the mature pancreas is composed ofacinar cells, it is reasonable to assume that these cellsare likely to be a target for carcinogens. This is clearlythe case in rat models of pancreatic carcinogenesisbecause early focal proliferative lesions and nodulesmaintain acinar cell differentiation. Acinar cell ade-nomas are relatively common, while proliferative le-sions within ducts are virtually unknown in the rat.Some acinar cell carcinomas have a minor compo-nent with ductal phenotype.3 Bockman has de-scribed sequential changes in acinar tissue leadingto the formation of tubular complexes.4 These obser-vations suggest that tumors with a ductal phenotypemay be derived from neoplastic acinar cells that haveundergone metaplasia to a ductal phenotype.The experiments described here are based on an

azaserine-induced acinar cell carcinoma of the ratpancreas that in early passages was the best differ-entiated of six such carcinomas that have been suc-cessfully transplanted. This carcinoma was placed inculture with the goal of establishing a well-differentiated cell line that might be found useful instudies of acinar cell biology. Although the tumor cellsgrew in culture and initially released amylase and li-pase, they failed to maintain production of exocrineenzymes. The cultured cells were tumorigenic in ratsand have yielded tumors with ductal characteristics.These studies provide strong support for the hypoth-esis that ductlike carcinomas can sometimes arisefrom transformed acinar cells.

Supported by National Institute of Environmental Health Servicesgrant ES03687 and NIH/National Cancer Institute grant CA47327.

Accepted for publication January 29, 1993.

Address reprint requests to Dr. Daniel S. Longnecker, DartmouthHitchcock Medical Center, Department of Pathology, Lebanon,NH 03756.

292

Ductlike Cell Lines from Acinar Cells 293AJPJuly 1993, Vol. 143, No. 1

Materials and MethodsThe same basic experiment has been performedtwice with minor variations. These experiments willbe referred to as culture cycles 1 and 2. The alter-nating sequence of culture, regrafting, culture, etcis summarized in Figure 1.

Transplantable Tumor

Lewis rats were obtained from Charles River Labo-ratories (Wilmington, MA). A transplantable tumordesignated as DSL-6 was derived in 1986 from a

CYCLE I

TEumor ID Growth

DSL6A

DSL6A/C 1

DSL6A/T1

DSL6A/C2

DSL6A/T2

Phenotype

Acinar cells

Transient amylasesecretion

Ductal, scirrhous

I EZ2 No amylase

Ductal, scirrhous

CYCLE I1

DSL6B

I-DSL6B/C 1

DSL6B/TI1

DSL6B/C2

I-

DSL6B/T2

Acinar, moderatelydifferentiated

Transient amylasesecretion

Adenosquamous

No amylase

AdenosquamousFigure 1. Flou chart history of Cycle I, DSL-6A atnd Cvcle II, DSL-6Bcell linies in vitro anid in vivo.

primary acinar cell carcinoma of the pancreas. Thecarcinoma developed in a male Lewis rat (DSL-101-79) that was given three intraperitoneal 30 mg/kg in-jections of azaserine when the rat was 2, 3, and 4weeks of age. This rat was autopsied at 2 years ofage, and a 7-mm tumor was sampled for histologi-cal evaluation and subcutaneously transplanted intoLewis rats. Histologically, the tumor was a moder-ately well-differentiated acinar cell carcinoma. Thistumor was transplanted into three rats at first andsubsequent passages. The transplanted tumorshave maintained a consistent phenotype similar tothat of the primary tumor through 12 passages.These transplants served as the source of cells forexperiments in culture cycles 1 and 2. Aliquots oftumor have been cryopreserved at several pas-sages and are maintained in liquid nitrogen.

Cell Lines

In culture cycle 1 a cell line (DSL-6A/C1) was ob-tained from the DSL-6 tumor tissue taken at theeighth transplantation. Tissue was minced andplated in Waymouth's MB 752/1 medium with 10%heat-inactivated fetal bovine serum (FBS) and 0.1mg soybean trypsin inhibitor (SBTI)/ml, penicillin,and streptomycin and incubated in 5% C02 and airat 37 C in a Corning tissue culture flask. SBTI wasomitted from the growth medium after the initial plat-ing. Cultures were passaged with 0.2 mmol/L EDTAcontaining 0.25% trypsin. In culture cycle 2 DSL-6B/C1 was developed using the same protocol.AR42J cells from the American Type Culture Col-

lection (Rockville, MD) were grown in Ham's F12Kmedium with 15% FBS.

Variation of Culture Conditions

A variety of culture conditions have been tested onestablished cell cultures to evaluate their effect onphenotype, both morphologically and as reflectedby amylase or lipase secretion.Medium components were tested in Waymouth's

MB 752/1 medium with 10% heat-inactivated FBS(HyClone Laboratories, Logan, UT). Logsdon's me-dium,5 designed specifically for mouse pancreaticacinar cells, contained 100 U/ml penicillin, 100mg/ml streptomycin, 0.5 mmol/L IBMX (isobutyl-methylxanthine), 0.2 mg/ml SBTI, 1 mmol/L insulin,1 nmol/L epidermal growth factor, 1 nmol/L carba-chol, and 10 nmol/L corticosterone. All reagentsand growth media were obtained from SigmaChemical Co. (St. Louis, MO). In addition, a simpler

294 Pettengill et alA/PJuly^ 1993, Vol. 143, No. I

supplemented Waymouth's medium with 10% FBS,10 nmol/L caerulein (a gift from Dr. R deCastiglione,Farmitalia Carlo Erba, Milan, Italy), 10 mg/ml dexa-methasone, and 0.1 mg/ml SBTI was tested. SBTIwas tested alone in Waymouth's/10% FBS medium.

Several substrates were also tested. Corning cellculture flasks (Corning Glass Works, Corning, NY)were used as a control, to compare with Falcon Pri-maria flasks (Becton-Dickinson, Lincoln Park, NJ)and plastic flasks coated with Matrigel (Collabora-tive Research, Waitham, MA) or Collagen (rat tailcollagen prepared in the laboratory as described).6In addition, some primary cultures were plated onPermanox slides (Nunc, Naperville, IL).

Tumorigenicity of Cultured Cells

Approximately 2 x 107 cultured tumor cells sus-pended in 0.4 ml growth medium were inoculatedsubcutaneously into three syngeneic weanling rats.

Light Microscopy

The primary tumor and transplanted tumors werefixed in Bouin's fixative or 10% neutral buffered for-malin, and embedded in paraffin, and sections werestained with hematoxylin and eosin (H&E). DSL-6A/C1 and C2 cells were grown on glass coverslipsor Permanox slides, rinsed, and fixed in 10% buff-ered formalin, followed by staining with H&E. Toprepare sections of cells for comparison with tumorsections, monolayers of cells were scraped, centri-fuged, fixed in buffered formalin, and embedded inparaffin for sectioning.

Antibodies

In the present investigation monoclonal antibodies(MAbs) and commercially available anticytokeratins(anti-CKs) which distinguished pancreatic acinar(OC.1) and ductal cells (OV-6, CK-7, CK-19) wereused to characterize tumors and cell lines derivedin culture. In addition to these cell-specific markers,OC.2, an acinar/ductal marker, and CK-14, a markerof squamous epithelium, were employed for someexperiments. The production and characterizationof MAbs have been described previously.7'8 MAbOV-69'10 was kindly provided by Dr. Harold Duns-ford (University of Texas Medical Branch,Galveston, TX). All anti-CKs were obtained fromBoehringer Mannheim (Indianapolis, IN). Some ofthe MAbs are not suitable for the study of fixed,

paraffin-embedded tissues and could not be ap-plied retrospectively at all stages of the study.

Immunohistochemistry

Indirect immunofluorescence (IIF) or immunoperoxi-dase analysis was carried out on cell pellets andcultured cells as previously described.11 Culturedcells were scraped into a pellet and either frozen inhexane chilled by a dry ice acetone bath or fixed inOmnifix (Xenetics Biomedical, Tustin, CA) or coldacetone and embedded in paraffin. For some stud-ies, minced primary tumor tissue was plated on Per-manox culture slides (Nunc), and at various inter-vals after plating, monolayer cultures were rinsedand fixed in cold acetone. IIF analysis was also car-ried out on frozen sections of azaserine-inducedtransplantable pancreatic tumors. Briefly, subcuta-neous tumors were excised, and small blocks of tu-mor were frozen as described above and stored at-70° until analyzed. Acetone-fixed serial sectionswere stained with MAbs followed by affinity-purified,fluorescein-conjugated, goat anti-mouse immuno-globulin (Sigma) as the secondary antibody. Non-specific staining was assessed by examining tissuesections stained with culture supernatants fromP3x63Ag8 myeloma cells. Sections were examinedusing a Nikon Microphot FX equipped with epifluo-rescence and a 35-mm camera.

For some experiments, double-labeling IIF analy-sis of cultured cells and frozen tissue sections wascarried out using MAbs in combination with isotypespecific secondary antibodies conjugated to eitherfluorisothiocyanate or Texas red.12 Photomicro-graphs were recorded with Ektachrome 200 filmand images were printed on black and white KodakPanalure paper. With this technique, reactive areasappear as dark staining cells.

Electron Microscopy

Small cubes (1-2 mm3) of tumor tissue or mono-layer cultures were fixed in 4% buffered glutaralde-hyde. Cultures were scraped after fixation and pel-leted. Both types of specimen were postfixed inosmium tetroxide and embedded for sectioning.Thin sections were examined using a Jeol 2000FXelectron microscope.

Amylase and Lipase Activity

Amylase or lipase activity of culture medium wasmeasured with an Olympus Demand analyzer and

Ductlike Cell Lines from Acinar Cells 295A//1pli' 1993. Vol. 143, No. 1

Technicon Amylase reagent (Technicon InstrumentsCo., Tarrytown, NY) or on a Kodak Ektachem Ana-lyzer (Kodak, Rochester, NY).

Doubling Time in Vitro

DSL-6A/C1 cells were plated in triplicate cultures at5 x 105 cells/ml. At feeding intervals (3 or 4 days)during the next 10 days, triplicate cultures wererinsed, dissolved in 3 N KOH, and precipitated with10% trichloroacetic acid and an aliquot used formeasurement of total protein by the Bio-Rad re-agent (Bio-Rad Laboratories, Richmond, CA). Dou-bling time was calculated graphically.

CCK Receptor Assay

Assays were performed using the method of vonSchrenk13'14 and frozen samples of DSL-6A/C1 atthe fifth passage level, transplanted DSL-6, andAR42J, a cell line which was isolated from DSL-1,

another azaserine-induced transplantable tumor.15Cultured cells were harvested by scraping and pel-leted by centrifugation.

Regrafting of AR42J Cells

AR42J cells were harvested by scraping, and ap-proximately 107 cells were inoculated subcutane-ously in male Lewis rats. Tumors developed after2-3 weeks in vivo and were harvested at autopsyfor histological study.

ResultsThe histological phenotype of the first DSL-6 trans-plant was similar to that of the primary tumor, andtumors from serial transplantation, sometimes ofcryopreserved tumor, continue to maintain an acinarcell phenotype to the present-growing as moder-ately to poorly differentiated acinar cell carcinomas.The primary acinar cell carcinoma did not contain

Figure 2. DSL-6, tbe fiOst transplant establishedftom a carcinoma inidulced bi azaserine in amale Leo'i.s rat. Acinar structores are preseiitthroiioghontt A feo acinar himens in the upperleft corner are dilated and conitainz homiioge-nieonis secretion (arrows). H&E. X 1 58.Figure 3. Serial tumor tranisplanit of the DSL-6tuoior as used for initiation oqfcycles 1 anid 2.The tionior maintains acinar phbenotipe. Al-thouigh lobuilcs of tuortco) are separated bh /i-broies stroma. the tumor is not tin/ ldesmnoplas-tic. H&IL X 158.Figure 4. DSL-6A cell line in cuiltlue at the.C2" stag,e, ie, after- regrafting anid reculture.The polygonal cells have proniinenit nucleoli.Too mitotic nuclei are present. H&E. X 1 58.

296 Pettengill et alA/P juld 1993, V,ol. 143, Ao. 1

ductal elements that could be identified in H&E-stained sections. The transplantable tumor line de-rived from this tumor maintained acinar differentia-tion but exhibited a microcystic growth pattern insome areas (Figure 2). The microcystic phenotypediminished with time, but acinar formation continues(Figure 3). Focal necrosis is noted centrally in largetumors. The transplanted DSL-6 tumors are not des-moplastic.

Cell Lines

We have assumed that they did not represent aseparate phenotype but resulted in some way fromgrowth on a plastic substrate. In addition, appar-ently dead, flat plaques of cells lifted off the mono-layer, suggesting the loss of differentiated cells.This morphology was exhibited as long as therewere any normal fibroblasts remaining in the culturebut is not present in the continuous cell line.

Enzyme Secretion of Primary Cultures

In the isolation (cycle 1) of cell line DSL-6A, theearly days of the culture were characterized by theproduction of large amounts of acid in the medium,necessitating medium changes at 1-2-day intervals.Epithelioid plaques appeared in the culture, whichwere not overgrown by endogenous fibroblasts. Atthe first subculture, 2 months later, no amylase waspresent in the medium. Within a month, after thenext passage level, a marked change in morphol-ogy occurred that was coincident with a loss of nor-mal stromal elements. Tumor cells that had beengrowing as colonies on the fibroblastoid monolayernow attached to the plastic. A continuous cell lineemerged which has retained the same morphology.By 3 months in vitro, all normal fibroblastoid cellshad disappeared. The doubling time was measuredas 4 days on passage 7 cells. This cell line has nowbeen carried through 18 subcultures, over a periodof several years. Various passage levels have beencryopreserved.A second cell line, DSL-6B/C1, was established

using the same protocol (culture cycle 2). A similarcell line emerged as described for DSL-6A/C1.Cells were grown in Waymouth's MB 752/1 mediumwith 10% heat-inactivated FBS (Way/10) withand without SBTI and, in addition, in Logsdon'smedium,5 which was developed specifically for mu-rine pancreatic acinar cells. No differences wereobserved either in amylase secretion or in cell mor-phology when compared to Way/10, the mediumselected for routine use.

Morphology of Cultured Cells

The early cultures in both cycles were characterizedby islands of cells appearing to contain two epithe-lioid cell types: a denser population of smaller cellsbordered the edge of the islands, which were com-posed of large flat cells (Figure 4). The smaller cellsdid not increase in number but remained at the pe-riphery of the island until they became confluent.

All cultures initially contained high levels of amylaseactivity in spent medium (>1000 units/L) collectedafter 4-day culture. Amylase levels quickly declinedand were absent when tested several months later.None of the alternative media conditions affectedthe level of amylase activity, nor was any morpho-logical change in phenotype observed by phasemicroscopy in those cultures compared to unsup-plemented Way/10 (data not shown). Collagensubstrate provided no advantage in either growth orenzyme secretion over untreated plastic cultureflasks. Cells appeared to attach 1 day sooner onMatrigel than on the other substrates or uncoatedplastic, and amylase secretion was retained for afew days longer in these cultures (Figure 5). None

-J

0~0)

E

50

40

0

c

0~:3

30

20

10

0 1 0 20Days

Figure 5. Ejftct of vanring coultre flas.k matrices on amyla.se anld li-pase secretioni in primnanr cuilturi-e of DSL-6. 7otal levels of activity on

day of collectioni are illutstrated.

10 200

Ductlike Cell Lines from Acinar Cells 297A/P,jull 1993. TVW. 143, Ao. I

of these measurements were related to cell numberor protein because the original cultures were estab-lished from minced tissue. Therefore, quantities ofenzyme activity are not as significant as the lengthof time of their persistence in the cultures. All condi-tions supported cell attachment by the fourth day invitro. Neither medium variations nor use of specialsubstrates altered the morphological phenotype ofthe cells in vitro as detected by phase microscopy,nor did they support the maintenance of the acinarphenotype.

In one of the experiments described in Figure 5,a portion of the original tumor was homogenized inphosphate-buffered serum at 1:10 (w/v) dilution andassayed for amylase and lipase with the followingresult: amylase:230,218 U/L; 387 U/mg tumor tis-sue; lipase: >4.05 U/mg tumor tissue. For compari-son, cells from DSL-6A/C3 were assayed similarly. A0.1-ml volume of packed cells, diluted to 1.0 ml inTris buffer, produced no detectable amylase and<0.01 U lipase/ml.

Growth of Cultured Cells in Vivo

DSL-6A cells at the sixth passage, 8 months afterthe tumor was placed in culture, were regraftedsubcutaneously in Lewis rats and were tumorigenic.The resulting tumors (DSL-6A/T1) were solid andvery firm, suggesting a high content of fibrous tis-sue. Histologically this is a desmoplastic ductlikecarcinoma (Figure 6), in contrast to the acinar mor-phology observed in the DSL-6 tumor from whichthe DSL-6A cell line was obtained. Following 3months' growth in vivo, DSL-6A/T1 tumor tissue wasquickly readapted to in vitro conditions in a mannersimilar to that of the parent DSL-6A/C1 culture andwas designated as DSL-6A/C2. DSL-6A/C2 cellsdid not secrete amylase at any time and expressedthe ductal phenotype when they were regrafted asecond time. After this passage through rats andreturn to culture, the cells were similar to DSL-6A/C2.

Figure 6. DS?L-6A/TI regrafted tmnor g,roLringl4sOihciitanieoiisli' in dense fibrous tis.sie oJ' a

Lea is rat. Du/ctlik.e striuctiures are pres-weitthrouighobit. H&EZ X 79.Figure 7. DSL-6B cell linCe growing subhcitanei-ouily' O/InnXoing regrafting. A. dilated. iiii.cI%sfilled .%paces partial/, iinecd bhi' glandldar epithe-limnn (X 104). B. sqviiaunuois area in the sametMinor (H158. X 1 58).

298 Pettengill et alA/P JulI 199-3, Ikl. 143, No. 1

Cells from cycle 2 (DSL-6B) were regrafted at thefourth passage. The carcinomas were partially solidand partially cystic. Histologically, the tumor pre-sented a mixed phenotype but retained no areas ofacinar cell differentiation. Some areas consisted ofpoorly differentiated adenocarcinoma, other areaswere mucin secreting (Figure 7A), and several ar-eas showed a stratified squamous pattern (Figure7B). There was moderate desmoplasia in the muci-nous areas. The tumor cells exhibited the same se-quence of events in vitro when returned to cultureas DSL-6A/C2, ie, attachment to the flask andgrowth of tumor cells as small colonies among fibro-blasts, which gradually disappeared. A mixture ofphenotypes suggestive of secretory (signet ringcells) and squamous (pavement epithelium) cellswere observed in vitro.

Regrafted AR42J cells exhibited a glandular phe-notype characteristic of poorly differentiated acinarcell carcinoma. There was no desmoplasia, ductformation, mucin secretion, or squamous metapla-sia (Figure 8).

Figure 8. In vivo grouth fAR421 ccell.s as a s.ubcutane.is tunloorfol-louwing regraJiinig )f the ciultiu-red cells A feua rudimentary acmi,i are

preseit, huit the tiunor geier-all) groos a.s a s'olicl ti(1,o0 uthbowt for-

mation o?f glands or ducts Tbhe tiomzor- cootaiuis no0 fibroi.s tissueH&EL X.230

Ultrastructure

The DSL-6 transplantable carcinoma shows ultra-structural features of acinar cells, including the for-mation of glandular spaces with microvilli on the lu-menal surface with intercellular tight junctionalcomplexes near the lumen (Figure 9). The cyto-plasm contained moderate numbers of densemembrane-bound granules consistent with zy-mogen, abundant free polysomes, scattered pro-files of rough endoplasmic reticulum (RER), andabundant mitochondria. Microfilament bundles wereprominent in some specimens.

Cells fixed for electron microscopy at passage 2,3 months after initiation of DSL-6A/C1, lacked zy-mogen granules, contained free ribosomes, formedmicrovilli on their surfaces, and were linked by tightjunctions (Figure 10). They mimic ductal cell mor-phology. These have been compared and found tobe similar to cells obtained after the second pas-sage through the rat, DSL-6A/C2, at the second cul-ture passage. Cells from the second cell line, DSL-6B, were similar in culture.

Regrafted cells from the DSL-6A cell line retainedseveral features exhibited in culture, forming mi-crovilli on lumenal surfaces with intercellular tightjunctional complexes. The cytoplasm containedpolysomes but not RER. One specimen from re-grafted DSL-6B/C1 cells generally showed less evi-dence of glandular differentiation, forming relativelyfew lumens. The cells were generally small andformed few lumens. Fewer than 10% of the cellscontained a few dense membrane-bound granulesof variable size.

Immunocytochemical Studies

The results obtained by using several antibodies tostain selected specimens from cycle 1 suggestedthat when acinar cells from the transplanted tumorwere passaged in vitro, a subpopulation of thesecells "dedifferentiated," losing acinar cell pheno-type, and began to acquire a ductal phenotype. Al-ternatively, the possibility existed that DSL-6 con-tained rare ductal cells that were selectivelyexpanded in culture and subsequently producedductal tumors when reimplanted into rats. To ad-dress this question, the "cycle" experiment was re-peated, and MAbs that distinguished acinar andductal cells were employed to characterize subcu-taneous tumors and representative stages in thederivation of cell lines. In general, results from IIFand immunoperoxidase studies (Table 1) show a

Ductlike Cell Lines from Acinar Cells 299A/P julyl 1993, VWo. 143, An. 1

Figure 9. DSL-6 transplanilable cicinlCIr ce/lcarcinom'iiia(/fiom ibiclb the cell lines Wlei-ederived. Alicrol'illi linle c lillnienl (top). Acjac-Celi cvtOplasmui Conitains cleuise. biuiuiinge-iieoiis. uuieuiJbiciuie- boiiicl grcnidiiies. Jlicrn/ilc-nicile7 biuidlesc/ir conispiciuous. 7I7e cytoplaslmcontainis aibuncldcant muitc)condria. tiee pol'-5somes. (111(1cialeft pr/files 0/'RER. X 16. 775.Figure 10. DSL-61A (,'1 cells (in culture). Th1eCC//.S /bave ueuii scrapeicld/i)uoi tle diiS?) 71T lIp-per (niedici exposed) sl{iirfce is cotereui b,1 mi-crovilli. t7e cvtoplaismi contaiis poli'snomies )iitlio secretori' gr,inrmdies onr RER X 54-i5.

Table 1. Indirect imnmnwnofiourescenzce anialysis of niormal aild nzeoplastic panicreatic cells

Markers

Cells OC.1 OC.2 OV-6 CK-7 CK-19 CK-14

Acinar cells +++ +++Ductal cells - +++ +++ +++ +++DSL-6 - +++ (R)* (R) (R) NDDSL-6/C1 - +++ (R) (R) (R) NDDSL-6/T1 - +++ +++ +++ +++ NDDSL-6/C2 - +++ +++ +++ +++ NDDSL-6/T2 - +++ ++ ++ ++ +++

(R) indicates the presence of rare positive cells. ND, not done.

trend for loss of acinar cell markers and acquisitionof ductal markers in the cell lines and regrafted tu-mors. The parent tumor line, DSL-6, was composedof cells that expressed OC.2 but lacked detectablelevels of OC.1. Although a subpopulation of theseOC.2+ cells (<10%) weakly expressed OV-6, CK-7,and CK-19 in vitro (Table 1, Figure 11), a pheno-typic characterization of DSL-6B/C1 demonstratedthat cell lines derived from DSL-6 in culture werecomposed primarily of OC.2+ cells that lacked the

ductal specific markers OV-6, CK-7, and CK-19.However, when DSL-6B/C1 was regrafted into syn-genic hosts, the resulting tumors (eg, DSL-6B/T1)were classified as adenosquamous tumors thatcontained areas of ductal metaplasia that ex-pressed ductal specific markers (Figure 12). Thesquamous areas of the regrafted tumor were posi-tive for CK-14. The stability of the ductal phenotypewas confirmed by the observation that these tumorcells continued to express OC.2, OV-6, CK-7, and

300 Pettengill et alA/JP jildy 199.3, 1V). 143, No. 1

C DFigure 11. Dolble labeling IIF anablsis o'L)SL-6 andel D)SL-6/62 Acetonie-fixed fiozen sectionls ?/' the DSL-6 paencreatic transplantable tumor- were

ieacted nlUb 01-6 (A) (10n1 C2 (B), and positie cell/s veiisiializedl Isisiig isot)pespecific seconidaii' anitibodis.s .llicrographs nere photographedlusinig L'ktlacbromne 2000fil/. cndiiegatines were printed oni Kodaelk Panal/ure paper l sing tbis tecnpiqe. reactive airtas aippeair (is dark stainingcells 711h tiliior stroiig' expresses OC62,Arrows inidicate rare O6.2 positive cells bich ieakli express the dIiictal iiarker desigiiated 01 6- (X 100)C aId D, acetonie/-ccel ciiltires oIn 6(2 cells reacted with 016 (C) riiid 0(2 (D). hi coiitrast to c/luitres of' 0(,'2 I)5S1 -661 cells (notshoin), this seconidari cc'll /i)iie ai/so cocxvprsses thedillictil mlirker design)iatedl 017-6 (X 180)

CK-19 (Table 1, Figure 11) when reintroduced intoculture (eg, DSL-6B/C2). It should be noted that alltumors and cultured cell lines lacked detectablelevels of OC.1, an observation consistent with theloss of this acinar cell marker by early eosinophilicpreneoplastic lesions and primary tumors (R. A.Faris, unpublished observations).

CCK Receptor Levels

Although DSL-6 exhibits both high- and low-affinityreceptors for CCK, DSL-6A/C1 does not bind CCKwith high affinity (Table 2). AR42J cells, utilized as a

positive control in this experiment, did exhibit bothhigh- and low-affinity receptors for CCK.

Discussion

Similar but slightly divergent results were obtainedin experimental cycles 1 and 2 (Figure 1). In bothexperiments, the cell lines derived from the trans-plantable acinar cell carcinoma lost most markers

of acinar cell differentiation and acquired new phe-notypic characteristics of ductal differentiation overthe course of about 1 year. Regrafted tumor in cycle1 mimics the appearance of human ductal adeno-carcinoma and in cycle 2 mimics an adenosqua-mous carcinoma, which in humans has beenclassed as a variant of ductal carcinoma.1 Regraft-ing of the cell lines elicits a desmoplastic host re-sponse. Desmoplasia characteristically was presentin the ductlike areas but not in the acinar compo-nents of primary carcinomas that developed inazaserine-treated rats.16

Regrafting of the established cell lines into ratsdid not cause a reappearance of the acinar cellphenotype. When such tumors were returned to cellculture they did not produce amylase in vitro but in-stead expressed the ductal phenotype morphologi-cally. To this time, we have not observed a reversalof the ductal phenotype with reexpression of theacinar phenotype either in vitro or in vivo.

The best characterized cell line derived previ-ously from an azaserine-induced carcinoma is

Ductlike Cell Lines from Acinar Cells 301A/P Jull/ 1993lVol. 14_3, No. 1

11 A

. .: :

.! 'I .m ..: ;, ., .: 1: .1 --.pmi. ,: i. .: !:,..'.!..:!:.:.'a.:uz. .: 11 .. .. .: :, y

. ' ' .

I~~~~~~~~~~~~~~~.j::F ,A

Figure 12. Double labelin,g IIF analysis o?f DSL-6BITi. Acetonie-fixed frozen ections qf the DSL-6BITi ltulolr were reacted sinultanieolusl, withOC.2 (A, C, anid E) aind CK-7 (B), 01 6- (D) orJ CK-14 (F). Positive cells uwer-e i'isualized u ih iso/ipe specific secondary anitibodies. Micrograpbswiere' phbtographecl OI Eklachroie 200 film cidcl printed o'n Kocdak Panalure paper With this tecbhiquie. reactive earc-as appcear- as Clark staiuiingcells. Glaniciilar epithelia u ithin thhe aclenosquamious cLaciioncia /stroilgh express OC.2. CK- 7. anid 0'-6. In conitrast, squcamcols-like epithelial cellsappear to lose hboth OC2 (C anid E) aniid 016 (D) an d acquire CK-14 (E). Arrows indlicate areas q/C iiSCli0im% epitbelium. A, B, E, F, x 100. C

D, X200

Table 2. CCK receptor levels in DSL-6A anid AR441 cell linies

High-affinity Low-affinity receptor

Specimen Kd nmol/L Bmax Kd(pmol/L) Bmax

DSL-6 0.33 ± 0.04 122 ± 23 0.08 ± 0.08 423 ± 320DSL-6A/C1 Not detected 0.45 ± 0.32 444 ± 377AR42J 0.30 ± 0.05 85 ± 11 0.04 ± 0.03 399 ± 227

'k.i.

t,.i.

i-

i.

:U.

Y

a..

t,

u

i'i

I

-ibs

302 Pettengill et alA/P J1uly 1993 lu/. 14g3 No. 1

AR42J. It has maintained acinar cell differentiation,including the capacity to make and secrete amy-lase in culture for years,17 and maintains an acinarphenotype when grown in vivo. AR42J exhibitshigh- and low-affinity receptors for CCK. WhenAR42J was regrafted into Lewis rats, the phenotypeof the resulting tumor was a poorly differentiatedacinar cell carcinoma. The in vivo growth of AR42Jdid not yield either ductlike or adenosquamousphenotypes. Thus, there is a distinct difference be-tween AR42J and the cell lines derived from theDSL-6 tumor in regard to the tendency to changephenotype. Cells from a transplanted azaserine-induced carcinoma (DSL-1) formed ductlike struc-tures when they were cultured on seminiferous tu-bules,18 but in contrast, a transplantable pancreaticacinar cell carcinoma derived by Reddy from aF-344 rat maintained an acinar cell phenotype whencultured under similar conditions.19 These observa-tions suggest that individual tumors vary in regardto the stability of phenotype.Two mechanisms could explain the derivation of

neoplastic ductal cell lines from the DSL-6 tumor inthese experiments. It is generally accepted thatpancreatic tumors arise from azaserine-altered aci-nar cells in this rodent model.20 The most likely pos-sibility is that neoplastic acinar cells may lose aci-nar cell differentiation and begin to express ductalmarkers, ie, undergo metaplasia. The second possi-ble mechanism is that overgrowth of a minor andundetected ductal component coincides with com-plete loss of neoplastic acinar cells in culture. Al-though we found that DSL-6 tumors contained arare population of cells expressing ductal markers,these tumors lacked a histologically identifiableductal component. Moreover, immunocytochemicalcharacterization of DSL-6/C1 demonstrated thatductlike cells were not selectively propagated inculture. In the conceptual context of the clonal ori-gin of the tumors, the DSL-6 cell lines employed inthe present study would be considered to be of aci-nar cell origin under either mechanism. This view issupported by the finding that transplantable tumorscontinue to express amylase when passaged invivo. Although the possibility exists that the meta-plastic process may begin in vivo, our findingsdemonstrate that acquisition of the ductal pheno-type is accelerated by adapting the tumor cells toculture, a manipulation that resulted in the acquisi-tion of ductal phenotype when these cells were re-grafted.

Our findings suggest that loss of the differenti-ated phenotype of neoplastic acinar cells in cultureallows these cells to differentiate along a ductal

pathway when regrafted into a syngeneic host.Since acquisition of the ductal phenotype does notinitially occur in culture, our findings further suggestthat progression of these less differentiated cellsalong a ductal pathway is determined, in part, bythe in vivo microenvironment of the subcutaneoustransplantation site. This observation is in contrastto the findings of De Lisle and Logsdon,21 whofound that normal adult mouse acinar cells dediffer-entiate in culture to a more ductlike cell that ex-presses duct-specific antigens.The studies described here provide additional

evidence to support the view that some pancreaticcarcinomas with a ductal phenotype might be de-rived by transformation and dedifferentiation ormetaplasia of acinar cells. Other studies that offersupport for this hypothesis have been done primar-ily in rats,4 but a similar result is reported for guinea

22 2pig and mouse.23 Carcinomas that develop in theDMBA-induced model of pancreatic cancer in ratsare considered to be of acinar cell origin, althoughtheir histological phenotype is ductal.4 A similar sit-uation pertains in methylnitrosourea-treated guineapigs in which a sequence of formation ofpseudoductules from acinar cells preceded the de-velopment of ductlike carcinomas.22 Ela-1-myctransgenic mice that were created using a pseudo-gene with an elastase 1 promoter/enhancer linkedto myc have developed tumors with both acinar andductal elements.23 The model is based on the ex-pectation that myc will be overexpressed specifi-cally in acinar cells. This is supported by prior stud-ies in the ELSV transgenic mouse model based onthe elastase 1-simian virus 40 T antigen pseudo-gene in which the vast majority of exocrine tumorsretain some degree of acinar cell phenotype.24The implications of these studies for humans are

uncertain. It is commonly assumed that ductal car-cinomas arise as a result of transformation and neo-plastic progression of duct cells. The finding of se-vere atypia of carcinoma in situ grade in thepancreatic ductal epithelium supports the validity ofthis pathway.25 Foci of atypical (hyperplastic or dys-plastic) acinar cells resembling those induced byazaserine in rats have been described in the humanpancreas,16 but at this time there is no firm evi-dence to link such foci with the genesis of ductalcarcinomas. Ductal complexes in lobular tissue ofhuman pancreas have been considered to arise bymetaplasia of acinar cells26 and to provide a path-way for the origin of tumors of ductal phenotypefrom acinar cells. The observations in animal mod-els reported or reviewed here suggest that the mat-ter of histogenesis of human carcinomas with a

Ductlike Cell Lines from Acinar Cells 303AJPJulV 1993. V'ol. 143, No. 1

ductal phenotype should remain an open questionthat requires further evaluation.

References1. Cubilla AL, Fitzgerald PJ: Tumors of the exocrine pan-

creas. Atlas of Tumor Pathology, Second Series, Fasci-cle 19. Washington DC, Armed Forces Institute of Pa-thology, 1984

2. Longnecker DS: Experimental models of exocrinepancreatic tumors. The Exocrine Pancreas: Biology,Pathobiology and Diseases. Edited by VLW Go. NewYork, Raven Press, 1986, pp. 443-458

3. Longnecker DS, Roebuck BD, Yager JD Jr, Lilja HS,Siegmund B: Pancreatic carcinoma in azaserine-treated rats: induction, classification and dietary mod-ulation of incidence. Cancer 1981, 47:1562-1572

4. Bockman DE: Cells of origin of pancreatic cancer: ex-perimental animal tumors related to human pancreas.Cancer 1981, 47:1528-1534

5. Logsdon CD, Williams JA: Pancreatic acinar cells inmonolayer culture: direct trophic effects of caerulein invitro. Am J Physiol 1986, 250:440-447

6. Freshney RI: Culture of Animal Cells: A Manual of Ba-sic Technique. New York, A. R. Liss, 1987

7. Hixson DC, Faris RA, Thompson NL: An antigenic por-trait of the liver during carcinogenesis. Pathobiology1990, 58:65-77

8. Hixson DC, Faris RA, Yang L, Novikoff P: Antigenicclues to liver development, renewal, and carcinogene-sis: an integrated model. The Role of Cell Types inHepatocarcinogenesis. Edited by AE Sirica. Boca Ra-ton, FL, CRC Press, Inc, 1992, pp 151-182

9. Dunsford HA, Sell S: Production of monoclonal anti-bodies to preneoplastic liver cell populations inducedby chemical carcinogens in rats and to transplantableMorris hepatomas. Cancer Res 1989, 49:4887-4893

10. Dunsford HA, Karnasuta C, Hunt JM, Sell S: Differentlineages of chemically induced hepatocellular carci-noma in rats defined by monoclonal antibodies. Can-cer Res 1989, 49:4894-4900

11. Hixson DC, Allison JP: Monoclonal antibodies recog-nizing oval cells induced in the liver of rats by N-2-fluorenylacetamide or ethionine in a choline-deficientdiet. Cancer Res 1985, 45:3750-3760

12. Yang L, Faris RA, Hixson DC: Long term culture andcharacteristics of normal rat liver bile duct epithelialcells. Gastroenterology 1993, 104:840-852

13. von Schrenk T, Moran TH, Heinz-Erian P, Gardner JD,Jensen RT: Cholecystokinin receptors on gallbladdermuscle and pancreatic acinar cells: a comparativestudy. Am J Physiol 1988, 255:G512-G521

14. Bell RH, Kuhlmann ET, Jensen RT, Longnecker DS:Overexpression of cholecystokinin receptors in

azaserine-induced neoplasms of the rat pancreas.Cancer Res 1992, 52:3295-3299

15. Jessop NW, Hay RJ: Characteristics of two rat pancre-atic exocrine cell lines derived from transplantable tu-mors. Abst. In Vitro 1980, 16:212

16. Longnecker DS, Shinozuka H, Dekker A: Focal acinarcell dysplasia in human pancreas. Cancer 1980, 45:534-540

17. Logsdon CD: Glucocorticoids increase cholecystoki-nin receptors and amylase secretion in pancreaticacinar AR42J cells. J Biol Chem 1986, 261:2096-2101

18. Reddy JK, Kanwar YS, Rao MS, Watanabe TK, ReddyMK, Parsa I, Longnecker DS, Tafuri S: Duct-likemorphogenesis of Longnecker pancreatic acinarcarcinoma cells maintained in vitro on seminiferoustubular basement membranes. Cancer Res 1986, 46:347-354

19. Watanabe TK, Hansen LJ, Reddy MK, Kanwar YS,Reddy JK: Differentiation of pancreatic acinar carci-noma cells cultured on rat testicular seminiferous tu-bular basement membranes. Cancer Res 1984, 44:5361-5368

20. Longnecker DS: The azaserine-induced model of pan-creatic carcinogenesis in rats. Experimental Pancre-atic Carcinogenesis. Edited by DG Scarpelli, JKReddy, DS Longnecker. Boca Raton, FL, CRC Press,Inc, 1987, pp 117-130

21. De Lisle RC, Logsdon CD: Pancreatic acinar cells inculture: expression of acinar and ductal antigens in agrowth-related manner. Eur J Cell Biol 1990, 51:64-75

22. Rao MS, Reddy JK: Histogenesis of pseudo-ductularchanges induced in the pancreas of guinea pigstreated with N-methyl-N-nitosourea. Carcinogenesis1980, 1:1027-1037

23. Sandgren EP, Quaife CJ, Paulovich AG, Palmiter RD,Brinster RL: Pancreatic tumor pathogenesis reflectsthe causative genetic lesion. Proc Natl Acad Sci USA1991, 88:93-97

24. Glasner S, Memoli V, Longnecker DS: Characterizationof the ELSV transgenic mouse model pancreatic car-cinoma: histologic type of large and small tumors. AmJ Pathol 1992, 140:1237-1245

25. Mizumoto K, Tsutsumi M, Kitazawa S, Tsujita S, Na-kayama M, Tsujii T, Kanehiro H, Nakajima Y, NakanoH, Konishi Y: Intraductal carcinoma in a surgically re-sected pancreas with chronic pancreatitis. Int J Pan-creatol 1990, 7:279-285

26. Parsa I, Longnecker DS, Scarpelli DG, Pour P, ReddyJK, Lefkowitz M: Ductal metaplasia of human exocrinepancreas and its association with carcinoma. CancerRes 1985, 45:1285-1290