genomic EBV DNA - Wiley Online Library

Int. J . Cancer: 45,163-712 (1990) 0 1990 Wiley-Liss, Inc.

Publication of the International Union Against Cancer Publication de I‘Union lnternationale Contre le Cancer

ESTABLISHMENT OF IMMORTALIZED PRIMATE EPITHELIAL CELLS WITH SUB-GENOMIC EBV DNA Loraine KARRAN’,Chong Gee TEO’, David KING’,Mary M. HITT’, Yanning GAO’, Nina WEDDERBURN~ and Beverly E. GIUFFIN’,~ ‘Department of Virology, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, London W12 ONN; and 2Departmentof Pathology, Royal College of Surgeons, Lincoln’s Inn Fields, London WC2A 3PN, UK. The genetic information in a sub-fragment of EBV DNA, designated p31 (containing less than a quarter of the viral genome and derived from a recombinant DNA cosmid library) allows epithelial cells from primary monkey and human kidney cultures to escape senescence under standard tissue culture conditions. A number of epithelial cell lines, designated M1131, 483/31, 199/31 and HW31, have been established and characterized following transfection of primary cells with p31 DNA. They share many properties, although morphologically they are not all identical. The cultures are immortalized but not fully transformed or tumorigenic. They appear to be phenotypically stable, although DNA hybridization studies indicate that genotypic alterations, including amplification, occur subsequent to transfection with p31 DNA and the establishment of a continuously proliferating epithelium. All cell lines consistently express high levels of cytokeratin 18 and varying amounts of cytokeratin 7, demonstrating their epithelial origin. From a single marmoset kidney (designated 199) a series of related immortalized cells, with subtle phenotypic differences, have been generated by p31 or sub-fragments of it. Although hallmarks of a “hit-and-run” mechanism are apparent in all of our studies, 2 different techniques (in situ hybridization or selection for cell survival in semi-solid media, followed by nucleic acid hybridization) show that, in late-passaged cultures, a small proportionof the cells still contain some viral DNA. The studies focus on genetic information within the BarnHl A and I regions as being relevant to immortalization. The role of the EBV DNA fragment in the genesis of epithelial cell lines is considered.

The well-characterized association of Epstein-Barr virus (EBV) with human malignancies has motivated much research aimed at localizing immortalization and/or transforming functions associated with this virus. To date, such efforts have been largely disappointing. Although it has been shown in 2 laboratories (Wang et al., 1985; Baichwal and Sugden, 1988) that the virally-encoded latent membrane protein (LMP, the gene product of BNLF- 1) can transform established rodent fibroblast cells in vitro and produce tumours in immuno-compromised mice, as can the gene product of BARF1 (Wei and Ooka, 1989), it is not clear what relevance this has to the action of the virus in vivo where the recipient cells acted upon are of human B- or epithelial-cell origins. Several years ago we embarked on a programme to examine the ability of large cloned fragments of EBV DNA to alter in vitro the phenotypic properties of primary cells from primate sources. Initially, we used among a variety of cells those from African green monkey kidneys (AGMK) because it was reported that they could be transfected with high efficiency (Gorman et al., 1983). We showed (Griffin and Karran, 1984) that epithelial cells from kidneys of these animals could be induced to proliferate continuous in vitro under standard tissue culture conditions following transfection with 2 overlapping subgenomic regions of EBV DNA. Control untreated AGMK cells (or cells treated with other EBV or carrier DNAs), after several passages, consisted exclusively of fibroblasts which eventually senesced. It is well known that Old-world monkeys carry their own endogenous viruses, including a simian herpesvirus, and our findings might have been the consequence of reactivation

of one or other of these viruses. This stimulated us to investigate the generality of our observations. If our results were reproducible, not only would this rule out a reactivation hypothesis and provide data concerning immortalizing functions encoded within the EB viral genome and the nature of the interaction of EBV with epithelium, but it would generate useful epithelial cell lines for studying functions associated with immortalization and the genesis of carcinomas. We chose for our present studies primary human cultures, or similar cells from the New World common marmoset (Callithrix jacchus) monkeys, because (unlike rodent cells) neither appear to undergo spontaneous transformation, and both have stable chromosome populations (diploid chromosome numbers of 46). Thus, any observed phenotypic alterations should reflect input information from the EBV genome. Further, the common marmoset provides a useful model for in vivo experiments, having been shown previously (Wedderburn et al., 1984) to accommodate infection by EBV without tumour formation. Here we show that transfection with DNA from one of the 2 overlapping EBV fragments that immortalized AGMK cells (designated p3 l), but not apparently from the other (p13/33), immortalizes epithelial cells in primary cultures from both New-World monkey or human sources. Cell lines have been derived from several morphologically distinct epithelium. None of these overgrow to produce foci, nor are they tumorigenic in nude mice. As earlier, untreated cells or cells transfected with other fragments from EBV recombinant DNA libraries generally resulted in the short-term propagation of fibroblasts. Nucleic acid hybridization data suggest that the mechanism by which this occurs is more reminiscent of a “hitand-run’’ phenomenon than due to the continuous expression of a viral gene function. That is, input viral DNA can be readily found in cells during the establishment of an allepithelial culture but as cells continue to be passaged it becomes increasingly difficult to demonstrate the presence of the EBV genome. Thus, the presence of the EBV gene information seems to be essential for the estabzishment of the immortalized cell but perhaps not for its maintenance. MATERIAL AND METHODS

Cells Primary cultures were prepared from fresh young “common marmoset” monkey (Callithrix jucchus) or 3-month human foetal kidneys, essentially as described by Freshney, (1983) and were used as primary cultures. In general, explant cultures were preferable to enzymatic tissue dissociation for generating viable robust epithelial cell mixtures, with fewer contaminating fibroblasts. From a single animal, depending upon viability, 10-16 50-mm dishes of primary cultures were obtained for use in transfection experiments. All non-human primates used

3T0whom reprint requests should be addressed. Received: November 27, 1989.

764

KARRAN ET AL.

were shown by serology to be negative for EBV early and viral capsid antigens. Human A431 cells (Giard et al., 1973) were obtained from the PHLS Centre for Applied Microbiology and Research, Salisbury, UK. During and after establishment, all cells were maintained in Dulbecco’s MEM supplemented only with 5% foetal calf serum, 2 nm glutamine, penicillin (100 unitdml), streptomycin (100 pg/ml) and fungizone (0.25 pg/ml) (Gibco, Paisley, UK), unless otherwise stated. Sources of DNA EBV DNA used in transfections was obtained from a library of large (about 40-kb) size-fractionated fragments of strain B95-8 virus DNA cloned as BamHI partial digestion products into the cosmid pHC79 (Griffin and Kman, 1984), or from ClaI or KpnI subfragments of the p31 cosmid fragment recloned in pAT153. Linear cosmid p31 DNA was prepared by in vivo packaging in E . coli strain BHB 3175 (Little and Cross, 1985). DNA was stored at 4°C in TE buffer (10 nm Tris-HC1, PH 8.0, 1 m EDTA) prior to use.

glass slides. The chromosome preparations were hybridized with biotinylated p31 EBV DNA (Teo and Griffin, 1987). (c) Cytological hybridization. Slides with cytospun cells were heated in a convection oven for 10 min at 80°C-lOO”C, then placed on an ice-cooled metal tray. The hybridization mixture consisted of 50% formamide, 10% (w/v) dextran sulphate, 2X SSC (1X = 0.15 M NaClI15 m sodium citrate), 25 mM sodium phosphate (PH 6.5), 2X Denhardt’s solution (1 X = 0.02% polyvinylpyrrolidone/O.02% Ficoll/0.02% bovine serum albumin), 250 pg/ml herring sperm DNA and 0.1 to 5.0 pg/ml biotinylated probe p31 or pHC79 cosmid DNAs. Ten microlitres of the mixture were applied to each cytospun area of cells and a 22 X 22-mm glass coverslip was placed over the mixture; the edges were sealed with rubber solution (Chemico, County Chemical Co. Ltd., Shirley, Solihull, UK). Denaturation of probe and target DNAs were performed as described by Teo and Griffin (1987). Hybridization was carried out at 37°C for 1 6 2 0 hr. Slides were then washed in 2 x SSC, 3 times for 10 min each at room temperature, then 2X SSC at 60°C for 30 min and 0.2X SSC at 42°C for 30 min. Cytochemical visualization was carried out as described by Teo and Griffin (1987).

Establishment of epithelial cell lines Primary cells were transfected either by the calcium phosphate precipitation technique or by electroporation. Calciumphosphate-mediated transfections (van der Eb and Graham, Detection of cellular cytokeratins 1980) were carried out with supercoiled circular or linear Indirect immunofluorescence was performed essentially as DNAs using 5-10 p g of DNA/90-mm dish of slightly sub- described (Lane et al., 1985; Ramaekers et al., 1987) using the confluent cells, generally without carrier DNA. Once conflu- mouse monoclonal antibodies (MAbs) LE61 and C 0 4 ent, cells were subcultured at a 1:5 dilution and, while growing (a-keratin 18), LP2K and BA16 (a-keratin 19), LHPl and actively (24-40 hr post initial transfection), were retransfected RKSE60 (a-keratin lo), and C35 (a-keratin 7). LE61 and C 0 4 under the same conditions as before and with the same DNA. recognize most simple epithelia, C35 recognizes some simple The rationale for this was the assumption that the viral DNA epithelia, LP2K and BA16 recognize occasional simple epiload could be increased by homologous recombination with thelia and some squamous cells, and LHPl and RKSE60 are input DNA. In practice, this procedure proved advantageous. specific for cornifying epidermal epithelium (Lane et al., Subsequently, cells were maintained at near confluence and 1985). subcultured ( 1 5 dilution) initially at 2- to 3-week intervals. Electro-transfection was carried out on cells in suspension us- Cell growth in culture: plating eflciencies ing an apparatus modelled on that of Neumann et al. (1982). Plating efficiencies were determined as described by FreshPrimary cells were trypsinized, washed and suspended at a ney (1983) using initial concentrations varying from 10 to 500 density of 3 X lo7 cells/ml in HBS buffer (140 nm, NaC1, 25 cells/ml. When an immortalized cell line was being examined, m HEPES-NaOH, 0.75 KIM Na,HPO,, PH 7.1). DNA was colonies were counted after 3 weeks and, in the case of a added to a final concentration of 18 pg/ml, then the suspension transformed cell line, after 10 days, following staining with was cooled to 0°C and exposed to an electrical field of 7.5 Leishman’s dye solution. kv/cm. Two short pulses, separated by 10 sec, were employed. With marmoset or AGMK cells, little cell death resulted from Cell growth this treatment, as determined by Trypan blue exclusion; human Cells were plated onto 20 dishes (50 mm) (in duplicate) at embryo cells were more sensitive. Cells were replated directly 1-2 X lo5 per dish, on day zero. An average count from 2 at 1.5-3 x lo6 cells/90-mm dish. Subsequently, dishes were dishes was obtained at 2- or 3-day intervals for a period of up maintained in a near-confluent state by subculturing (1:4) every to 17-20 days or, for serum response, as stated. Medium was 10-14 days. To assess the maximum electroporation efficiency not changed during this period. in these experiments, marmoset cells were independently treated with either supercoiled circular or linearized (EcoRI Soj? agar assays digested) SV40 DNA and assayed after several days for TThe method of Bouck et al. (1978) was modified to allow for antigen expression by indirect immunofluorescence. One to growth or survival of epithelial cells. To assess for growth, two per cent of cells were T-antigen-positive. These eventually suspensions of lo4 cells/ml were made in 0.8% normal agarose became fully-transformed lines. dissolved in Dulbecco’s MEM, supplemented as described for cell culture. The mixtures were poured into tubes and capped Analysis of genomic DNA porn immortalized cell lines tightly. To assess for cell survival, ultra-low melting agarose (a) Hybridization of Southern blots. Total cellular DNA was (SeaPrep 15/45, FMC Bioproducts Europe, Denmark) was prepared by standard procedures and digested with restriction used and the tubes were refrigerated at 4°C for 45 min, to allow enzymes according to the manufacturers’ instructions. Samples gelling to occur. Capped tubes were incubated at 37°C. Feed(10 kg) were separated by electrophoresis on 0.8% SeaKem ing was not carried out except in those assays incubated for agarose gels, transferred to nitrocellulose (Schleicher and more than 4-6 weeks. In experiments selecting for cell surSchuell, Dassel, FRG) and analysed by hybridization with vival, although there were no visible colonies, cells were piCX-[~~P]-~C nick-translated TP probes under conditions de- petted out of soft agar onto plastic dishes after 2 months and scribed earlier for EBV DNA (Rymo et al., 1979; Griffin et viable adherent cells were cultured as described above. al., 1981). (b) In situ hybridization of chromosomes. Confluent cells RESULTS were incubated with colcemid (0.5 pg/ml) for 12 hr. Mitotic Characterization o f the structure of p31 DNA cells were selectively shaken off, subjected to hypotonic treatA cosmid DNA library was generated with partial BamHI ment, fixed with acetic acid-methanol (1:3) and dripped onto

765

EPITHELIAL CELL IMMORTALIZATION WITH EBV DNA

restriction fragments of EBV viral DNA from the transforming strain, B95-8, cloned in the vector pHC79 (Griffin and Karran, 1984). Most clones selected to make up a comprehensive genome library contained at least one copy of the large internal repeat, BamHI W (probably present at high copy number in the digest and selected during ligation and cloning to make up the 40-kbp size required for efficient packaging into the X phage head). Because of its properties as an epithelial cell growthstimulating fragment, the precise structure of p31 was determined by restriction enzyme analysis. As shown schematically (Fig. l), p31 contains a large contiguous segment of EBV DNA (BamHI D-A) flanked by the smaller BamHI W and L fragments, the latter in reverse orientation to that normally found in the EBV genome. There is a deletion in the genome which removes BamHI V, as indicated. Whether this represents a cloning artefact or a subpopulation of EBV DNA in B95-8 cells is not known. Cell growth in culture following transfection of EBV DNA The ability of an EBV recombinant DNA (p31) to immortalize primate epithelial cells from primary cultures in vitro, without stimulating growth of fibroblasts, has been assessed in long-term studies. The p31 cosmid DNA was used either in its supercoiled form, or as linear DNA (cleaved within the lambda cos sequences of pHC79; Little and Cross, 1985), or as subfragments generated by restriction enzyme digestion (as indicated in Fig. 1). In the case of ClaI and KpnI digests, individual fragments were recloned into the plasmid pAT153 for use in transfection studies. In experiments with EcoRI, Hind11and BgZII, fragment mixtures were used. In one set of experiments, Hind11 digests of p31 [which retain a large (approx. 20 kbp) fragment from p311 were transfected onto primary marmoset cells together with the BamHI-W repetitive DNA (on the assumption that an enhancer function within the repeat, either acting in trans or combined intercellularly following transfec-

tion, might increase the efficiency of the immortalization process and counteract losses of efficiency arising from the use of linear DNA). As controls, cells were transfected with other members of the EBV DNA cosmid library, with the cosmid vector pHC79, or with calf thymus DNA. The cultures used were derived from African green monkey (Griffin and Karran, 1984) or common marmoset kidneys, or from human foetal kidney or breast milk (Chang et al., 1982). A very large number of experiments have been carried out and a consistent cell growth pattern has been observed: (i) In general among the cosmid recombinant EBV DNA library (Griffin and Karran, 1984), only transfection with p31 DNA has resulted in the production of continuously proliferating (immortal) marmoset or human epithelial cells. (ii) However, not all the epithelial cells that appeared initially to be “growthstimulated” to produce colonies on plastic progressed to become immortalized; that is, of the colonies observed on a single dish of cells in subconfluent early cultures, some would continue to proliferate whereas others would senesce. (iii) After about 5 or 6 passages (i.e., several months) in culture, surviving cells showed an enhanced growth rate and could then be subcultured weekly at a 1:4-15 dilution; once cultures were fully established, they could sustain further dilution. (iv) The generation of a solely epithelial, continuously growing culture generally required between 4 and 6 months. (v) The most efficient stimulant for epithelial cell growth was supercoiled p31 DNA. In most of the experiments where cells were transfected with other fragments of the EBV genome or full-length or sub-fragments of p31 (see above and Fig. l), fibroblasts eventually became the only surviving cells and these generally senesced after 5 or 6 passages in culture. (vi) The frequency with which epithelial cell lines were generated from a primary culture from individual species was 30% in the case of human cells (1:3), 50% in the case of marmoset kidneys (3:6) and 100%(1:l) for AGMK cells. In cases in which no cell line was U

A.

P

JI

../ ’0

0 , 0 , 0 0

0

H 0 4

,

0

0 ‘0 ,

,=- /. ,. 0

0

I I

0

eR

cbXd

YII

II

.,/.’.. JJ

0 , 0,

,,/’ , , ’ ,

a S

I

,”,

0

0

I

J

\ \

0

0

\

0 0 / / ’ 0

\ \

\

6. BainliI E H Bg

C

K

BarnHl

a

BDD

I

b a@

D

766

KARRAN ET AL.

established, including that involving human milk cells, transfection of p31 DNA conferred a “growth advantage” to epithelial cells in that their survival time in culture was considerably prolonged in relation to control cells (Griffin et al., 1984). Since the raison d’2tr-e of our experiments was mainly to localize and identify epithelial cell growth-enhancing functions within the EB viral genome, hormone stimulants were not added to the culture media. p31 DNA itself proved adequate for growth stimulation of epithelium, without any obvious response by fibroblasts. Once an epithelial cell line has been established, the addition of exogenous stimulants (insulin, hydrocortisone or prostaglandin) to the culture medium had little or no effect on growth. Selection by co-transfection of p31 DNA with genes encoding drug resistance markers proved in our hands to be counter-productive since epithelial cells were more drug-sensitive than fibroblasts. Table I shows the cell lines that have been generated following transfection with p3 1 DNA; cells were only deemed to be “immortal” when they had been passaged in culture for 1 year or more. Properties of the marmoset and human epithelial cell lines Rheinwald and O’Connell (1985) classified human renal cortex cells into 2 morphologically distinct classes. We have adapted their classification to correspond with our findings, sub-dividing type-I cells into those with a swirling, unstratified epithelioid morphology (IA) and those with a discrete cuboidal morphology showing little evidence of swirling (IB). Type-I1 cells have a more fibroblastoid morphology.

( I ) The MI131 marmoset kidney cell line The MU31 line was established from kidneys taken from a young adult animal. Morphologically, it falls into the class-I1 category. Cells show strong immunofluorescence with MAbs specific for cytokeratin 18 and, to a lesser extent, cytokeratin 7, indicative of poorly differentiated epithelium (Fig. 2, panels c and d; Table 11), probably derived from renal tubules (Quinlan et al., 1985). Electron microscopy showed the absence of gap junctions and confirmed the poorly differentiated state of the cells (data not shown). MU31 cells grow as flat monolayers; they do not grow to high density (not exceeding 2.5 x lo6

FIGURE2 - Light microscopy and immunofluorescent staining of marmoset kidney cells. (a) Confluent monolayer of the primary mixture of fibroblast and epithelial cells from kidneys prior to transfection, showing about 30% cytokeratin positively staining cells with the LE61 MAb that recognizes cytokeratin 18. (The percentage of epithelial cells in primary cultures generally varied between 30% and 60% depending upon the age of the animal and the method of culture.) (b) Control (untransfected) cells after several months in culture, containing mainly fibroblasts, which are not stained by LE61. (c, d) Cells of the established marmoset epithelial cell line (M1/31) with type-II morphology as observed by light microscopy and immunofluorescence with LE61, respectively; similarly, ( e ) and If), the type-IB 483 marmoset cells as observed by light microscopy and immunofluorescence. Immunofluorescence, bar = 100 km. Light micrographs, bar = 100 pm.

TABLE I - CELL CULTURES AND CELL LINES Source of Drimarv cells

African green monkey kidney Marmoset kidney 1 Marmoset kidney 483 Marmoset kidney 199 Human foetal kidney

lines

AGMW3 1 AGMKl 13 MU31 M1/5/31

Transfected DNA5

P3 1 p 13/p33 P3 1 p31 + p5

Morphology1

~~~~f~~~

Maintained

Calcium phosphate

IB IB

2 2

Calcium phosphate

Variable

-

Method Of transfection

483/3 1 P3 1 483lSV SV40 Electroporation 199/31 P3.1 199/31HD p3lHindII 199/31HD/W p3 1HindII6 199/31L/W Linear p316 Calcium 199/23/W ~ 2 3 ~ phosphate HW3 1

P3 1

Calcium phosphate

I1

2

25 Passages3

-

+

IB IA IA IA IA IA

-

4

I1

-

23 Passages3

2

4

4 4 4

‘IA, similar to morphology I of Rheinwald and O’Connell(1985); IB, similar, but without a swirling pattern, more cuboidal; 11, similar to morphology I1 of these authors.-*For more than 2 years in continuous ~ulture.-~Cells frozen away at this stage, but still ~iable.-~Morethan 1 year in culture, still being pr0pagated.-~p5, p31, p13/p33 and p23 are clones from the cosmid library (Griffin and Karran, 1984).-6Means cotransfection with the large EBV internal repeat, BamHI W.

767

EPITHELIAL CELL IMMORTALIZATION WITH EBV DNA TABLE U - CYTOKERATIN STATUS OF P31-IMMORTALIZEDEPITHELIAL CELLS Cell line

LE61 (Keratin 18)

C04 (Keratin

BA16 (Keratin

LPZK (Keratin

18)

19)

1%

MU31 48313 1 199131 HW3 1

++ ++ ++ +

+++ +++ ++ ++ +

Staining intensity indicated ( - to

-

-

-

-

-

-

c35 (Keratin 7)

+

+ +f l +-

RKSE60 (Keratin 10)

LHPl (Keratin 10)

-

ND ND

-

-

ND

+ + +). ND, not determined.

cells/50-mm dish) nor, even after several consecutive years in culture, do they produce foci. When left at confluence over several weeks MU31 cells do not overgrow, nor do they produce visible colonies in soft agar or tumours in nude mice. The EBV genetic information present in total chromosomal DNA from the MU31 epithelial cell lines, as determined by Southern blot analysis, is illustrated in Figure 3a. At the 10th passage (about 4 months in culture), all the BamHI fragments of p3 1 viral DNA were found to be present in the chromosomal DNA digest. At this stage it could be shown that the bulk of the viral DNA was probably integrated since no material corresponding in size to input plasmid DNA could be detected in Hirt extracts (Hirt, 1967) or in uncleaved total chromosomal DNA (data not shown). Evidence of viral genome alterations were found, however, notably in the marked difference in copy number of the BamHI W fragment, where one copy is present in the input DNA (Fig. 3a; see also Fig. 1) compared with several copies present in chromosomal DNA isolated during early stages in the establishment of the cell line (Fig. 3b, lane c and 3c, lane a). In the latter, a band corresponding to the input pHC79 cosmid sequence is notably absent. By passage 22 no obvious EBV DNA could be detected in the total chromosomal DNA digests (Fig. 3c, lane b). The latter finding is consistent with those obtained earlier for AGMK epithelial cell lines (Griffin and Karran, 1984).

FIGURE3 - (a) Ethidium bromide staining of BamHI restriction digests of EBV and related DNAs. Tracks contain fragments from: (a) p31 DNA (Fig. 1); (b) a simulated EBV virion BamHI DNA pattern obtained by mixing fragments from the EBV recombinant cosmid library (Griffin and Karran, 1984); (c) the pattern obtained after digestion of B95-8 virion DNA (a transforming strain of EBV). (b) Southern blot analysis of BamHI DNA fragments hybridized with '*P-labelled p31 DNA. Tracks contain fragments from: (a) p31; (c) MU31 chromosomal DNA at 10th passage; (b and d) marmoset and calf thymus chromosomal DNAs, respectively. In the immortalized cell line (lane c) the increased copy number of the BamHI W fragment, compared with the cosmid p31 DNA (lane a), is apparent, and the full-length cosmid vector (pHC79) is absent, as indicated. (c) As in panel (b) Tracks contain: (a) chromosomal DNA from the MU31 cell line at early (10th) and (b) late (22nd) passage in culture; (c and d) calf thymus and marmoset chromosomal DNAs, respectively. The faintly positive band seen in lanes (a) and (b) and present in (d) appears to be cross-hybridization either between EBVs or between plasmid DNAs and marmoset DNA.

The apparent absence of viral DNA in late-passaged cells could reflect either a total loss of viral DNA, or the retention of only very short sequences, or cellular heterogeneity with viral DNA retained in a minority of the cells. To investigate this question a sensitive in situ nucleic acid hybridization technique, that allows low copy number genes to be recognized and localized, was employed. In this experiment, late-passage MU 31 cells were spun onto glass slides and hybridized with biotinylated p31 or pHC79 (control) DNAs. Signals were amplified as described by Teo and Griffin (1987). Namalwa Burkitt lymphoma cells, which have been found by solution hybridization to contain 2-3 copies of EBV DNA per cell (Adams, 1979), were used as controls in this experiment and the results, plotted as a histogram, are given in Figure 4. This technique, which showed a great diversity of viral genome copies in Namalwa, with a median number of 3, could only identify EBV DNA in about 5% of the MU31 cells. 23

Nemolwa (9

14

o

1

18

14

z

3

4

5

e

7

8

80

n

o

l

e

RGURE 4 - Histogram showing hybridization of Namalwa Burkitt

lymphoma cells (control) and the MU31 marmoset kidney epithelial cells with biotinylated p31 and pHC 79 probes, as indicated. Stained grains were counted as described by Teo and Griffin (1987). For Namalwa, although individual nuclei gave variable numbers of signals, the average number was consistent with data from solution hybridization (Adams, 1979); the average p31 copyhucleus = 2.88; average pHC 79 copylnucleus = 0.01. For M1/31, at passage 15 the average p31 copyhucleus = 0.07; average pHC 79 copy1 nucleus = 0.02. 1000 nuclei were used for each cytospin preparation.

768

KARRAN ET AL.

In an attempt to select for the cells carrying EBV DNA in late-passage M113 1 cultures, a modification of the soft-agar assay (Bouck et al., 1978) was adopted. Although, under normal culture conditions in dishes, only senescence was observed, some cells underwent limited growth in soft agar and could survive for a number of weeks in a sealed culture-tube. Such cells could be recovered eventually into adherent culture with no apparent alteration in their growth properties. Analysis of DNA from 3 such selected cultures (Fig. 5) by Southern blot hybridization showed the presence of EBV BamHI A DNA in one (panels 3) and BamHI I DNA (panels 4) in another. In the third case, no differences were detected between the parental (panels 1) and a third selected (panels 2) line. The data (Figs. 4 and 5) suggest that there is considerable heterogeneity in the original MU31 immortalized cell line, at least with regard to viral DNA content, and show that there is retention of a part of p31 DNA in cells that, when unselected, appear to contain no viral DNA (by Southern blot analyses). Only sequences from BamHI A and I fragments were observed in this experiment; other regions of p31 DNA were not detected.

(2) The HI131 human foetal kidney epithelial cell line This is also a type-I1 cell line. In most of its properties, it resembles the marmoset MU31 line (Tables I and 11). A growth curve for the H1131 line is shown in Figure 6 which illustrates not only the fact that the cells do not grow to high density, but also that they are serum-dependent. Fluorescence staining with C04 (cytokeratin 18), C35 (cytokeratin 7) and BA16 (cytokeratin 19, negative) MAbs is shown in Figure 7. (3) The 483131 marmoset kidney epithelial cell line This differs from the above cell lines mainly in having a type-IB morphology (Fig. 2 ) . The cytokeratin staining pattern of 483/31 is given in Table 11. Like the MU31 cell line, once the cells were growing in culture as a fully established epithelial line, no EBV DNA could be detected by Southern blot hybridization of total chromosomal DNA cleaved with BamHI and probed with 32P-labelledp31 DNA (data not given). In this case, in order to assess whether residual viral DNA was present, but at levels too low to be detected; the in situ hybridization technique previously devised to visualize EBV genes present in B-cells in low copy numbers (Teo and Griffin, 1987) was applied to the epithelial cell line. Using metaphase chromosomes and a biotinylated p3 1 cosmid DNA probe, scaring as positive those chromosomes in which distinct grains were observed on both chromatids. a small fraction of the cells

0

,

4 8 12 16 Time (days post plating)

20

FIGURE6 - Growth attems of an immortalized prototype marmoset cell line (483131) &-A) compared with SV40-transformed 483 marmoset kidney cells (O--O), and the immortalized HW31 human line (A-A) compared with the A-431 human carcinoma line (U Cells ) were .not serum-fed during this experiment. There is a shorter lag period between plating and growth for fully-transformed cells than for corresponding immortalized lines. Transformed cells consistently grew to higher density than immortalized cells and were less serum-dependent. Plating efficiency experiments also showed that it took longer (3 weeks) for colonies that could be counted to develop from the immortalized lines than from fully-transformed lines (I0 days). Moreover, plating efficiencies of the former were very concentration-dependent, whereas this was not the case with the SV40transformed marmoset cells. Plating efficiencies observed for the different cell types (using 200 cellslml) were 3.2% for 483/31, 1.1%for HW31, 6.5% for A431 and 10% for SV40-transformed kidney cells, respectively.

in the 483/31 line could clearly be identified as carriers of EBV DNA. Seven out of over 150 metaphase spreads examined (2 of which are shown, Fig. 8) were found to contain EBV DNA of the size expected for an average gene. Thus, although Southern blot analyses revealed no viral DNA, the more sensitive procedure showed retention of viral information in some cells. Assuming that the grains identified reflect viral integration, as indicated by analysis of Hirt extracts (data not shown), it appears to be a random occurrence.

(4) The 483/SV cell line Kidney cultures from animal 483, transfected with SV40 DNA, produced a transformed cell line that grew as foci on plastic or as colonies in soft agar. Growth curves of the p31immortalized and SV40 fully-transformed cells are shown in Figure 6.

1 2 3 4

FIGURE5 - Hybridization of Southern blots of BamHI fragments from late-passaged M1/31 marmoset cells. Tracks contain: (1) uncloned cells and ( 2 4 ) lines selected for growth-survival in lowmelting agarose (“Material and Methods”). (As this is a very lengthy procedure, only the MU31 line has as yet been sub-cloned by this method.) The probes used were 32P-labelledEBV fragments derived from: (left panel) EcoRI C (that should recognize EBV BamHI fragments X through A, see Fig. 1) as well as plasmid DNA; (center panel) purifigd (vector-free) BamHI A, and (right panel) purified BamHI I. The lgqations of BamHI A (+) and I are indicated. No signifi nt differences were observed between one of the cloned lines (tracks ) and the uncloned population of cells (tracks l), whereas another cloned line (left and center panels, tracks 3) can be observed to have a band that hybridizes strongly to BamHI A, and the third (left and right panels, tracks 4) to Bum HI I.

Yi

3-1061

(c)

(5) 199 marmoset kidney cell lines A unique exception to the general observation that only intact circular p3 1 DNA immortalized epithelial cells was shown by kidneys from a marmoset designated 199. In this case, epithelial cell lines were generated following co-transfection of primary kidney cultures with intact p3 1 (line 199/3l ) , with p3 1 that had been cleaved with H i n d u (199/31HD), or with p31 linearized at the cohesive (cos) sites of the vector or cleaved with HindIII, co-transfected with BamHI W DNA (line 199/ 31UW and 199/31HD/W, respectively) (Table I). All of the 199 family of cell lines were morphologically of the type-IA class and were strongly fluorescent with MAbs to cytokeratin 18 and 7 (Table 11). The growth characteristics of all but one of the 199 cell lines were similar. In DMEM containing 5% foetal calf serum, the lines reached confluence around 5 X lo6 cells/

769


FIGURE 7 - Immunofluorescent staining of established epithelial cell lines (Table I) with various a-keratin antibodies. (a-c) HW31 human cells stained with: (a) C04; (b) C35; and ( c ) BA16 MAbs and (d-f) 199/31 marmoset cells stained with: (d) C04; (e) C35; and (f) RKSE 60 (Fig. 2). Antibody specificities are given in Table 11.

I

3-1

FIGURE8 -In situ hybridization of 2 of the positively staining metaphase chromosomes from the 483/3 1 cell line, at 22nd passage in culture, using biotinylated p31 DNA as a probe and a procedure capable of detecting single-copy genes on metaphase chromosomes (Teo and Griffin, 1987). Arrowheads point to chromosomes symmetrically labelled at both chromatids of a homologue in a single spread. Of 7 positively staining spreads in this experiment (out of a total of 150 examined) no 2 were identical, indicative of random integration of the EBV DNA. Nucleic acid hybridization of total cellular DNA from 483/31 cells at this late passage (as in Fig. 5) failed to show the presence of viral DNA (data not given).

50-mm dish in culture; 199/31HD/W only reached 2.5 x lo6 cells/50-mm dish when confluent (Fig. 9). On the other hand, neither the rate of growth nor the density of the latter were appreciably affected when propagated in 1% serum, whereas all other cells were markedly serum-dependent (as illustrated, for 199/31HD). Similar data regarding serum dependence, at least for concentrations between 1% and lo%, were observed in serum titration experiments (Table 111). Although no spontaneously-occurring lines were derived from the 199 kidney, an epithelial cell line was also produced subsequent to co-transfection with the cosmid recombinant DNA, p23 (Fig. 1 and Griffin and Karran, 1984) and BamHI W DNA. This is the first time that immortalization has been observed with p23. This line (199/23/W) will be characterized further since p23 DNA encodes 3 factors known to be involved in transcriptional activation of EBV early promoters, at least in B-cells (Chevallier-Greco et al., 1986; Hardwick et al., 1988). DISCUSSION

Primate epithelial cells have proved remarkably difficult to establish in long-term culture compared with other cell types, such as fibroblasts or B-lymphocytes. Indeed, Tucker et al.

I

1

3

5 7 9 11 13 15 17 Time (days post plating)

FIGURE 9 - Serum response of 2 different marmoset cell lines derived from the same kidney. Cells were plated at 1 X 1O5/5O-mmdish. Curves represent: 199/31 HD, propagated in 5% serum (D-0)and 1% serum (A--A); similarly, 199/31HD/W cells with 5 % (U) and 1% serum (A-A). See Table I for description of cell lines and Table I11 for further serum studies. TABLE III - SERUM RESPONSE

Cell line

199/31HD

199/31HD/W

Serum concentration (percentage)

Cell count1 Day 4

Day 8

Cell growth response

0.1 1.o 3.0 5.0 10.0 0.1 1.o 3.0 5.0 10.0

1.1 5.0 3.6 5.5 5.9 1.2 2.6 1.4 3.7 5.8

7.62 4.0 3.8 6.8 11.8 6.02 4.0 2.1 7.1 9.0

Decrease Decrease Xl.1 x1.2 X2.0 Decrease X1.5 X1.5 X1.9 X1.7

l x lo5,unless otherwise noted. Cells were plated at 2 X 1O5/5O-mm dish in the serum concentration indicated and propagated without change of media. Data given are average counts from 3 dishes.JX 104.

(1984) have shown that proliferation of kidney epithelial cells can actually be inhibited by a protein (related or identical to TGF-P) secreted by the cells. Here we show, however, that the EBV DNA sequence present in the recombinant cosmid clone

770

KARRAN ET AL.

p31 carries information that can confer a growth advantage on EBV primate (common marmoset or human) epithelial cells. In many cases, cells have been established as continuously proliferating (immortalized) lines (Table I), expressing sub-sets of cytokeratins (Figs. 2 and 7; Table 11). Cell lines so generated I 0 have proved to be phenotypically stable; over a 2- to 3-year period of observation, none produced in the course of our work have progressed spontaneously to a “fully transformed” phe- p31 region L D cb T X I A W pHC79 notype. In this way they differ markedly from immortalized rodent cells which tend to become transformed on passage in culture, probably as a consequence of chromosomal alter4c 4 4 ations. In the single marmoset case examined in detail (line TK Pol LF3 LFI 483131, data not shown), cells have remained diploid in culture e over a long period of time. gB* RF“18.8 Open reading frames in p3 1 DNA which might specify genes relevant to the immortalization of epithelial cells are given in Figure I(c). The precise function(s) in EBV responsible for HSV-1 creating or inducing the cellular alterations have not been idenBglll tified but our data point toward information encoded within the BarnHI A and I sub-fragments of p31 as being relevant to MRTI , immortalization. That is, DNA from these regions has been region 0.31 0.34 0.36 0.38 0.40 0.42 found in late-passaged cultures (Fig. 5) and Hind11 digestion MuX BamHlg of p31, which preserves intact this same region has, in one -+ case, resulted in the generation of several immortalized epiTKPOI ---+ + thelial cell lines. In comparison studies, we (Hitt et al., 1989), 40K gB DBP as well as others (Rabb-Traub et al., 1983; Tugwood et al., 1987), have shown that in nasopharyngeal carcinoma cells the FIGURE10 - Comparisons of immortalization and morphological p31 region of EBV is transcriptionally active. For example, in transformation regions within the EBV and HSV-1 genomes, respeca cDNA library derived from an NPC tumour propagated in tively. The major repeat elements are indicated by open bars ( 0 ). nude mice, we found the p31 region to be strongly represented, For EBV (at top), the locations of genes for the 6 known EBNAs whereas transcription from regions encoding all the EBV ( 4 ), the latent membrane protein ( ) the major viral origin of EBNA genes (except that for EBNA-1) was not observed (Hitt replication (0) as well as the location of p31 DNA on the physical et al., 1989). Moreover, a hitherto unrecognized open reading map are indicated (see also Fig. 1). Known functions that reside within frame, designated “18.8”, was identified among the major the contiguous viral sequence in p31 include the thymidine kinase transcripts. Within the p31 region a large number of open (TK), DNA polymerase (Pol), major DNA binding protein (DBP), reading f r a m e s 4 n e or more of which may be relevant to glycoprotein related to HSV-1 gB, and BALFl function (RF1). Also shown is the recently identified RF “18.8” open reading frame which immortalization-are found. Functions which might induce the is highly transcribed in an NPC tumour (Hitt ei al., 1989), as well as phenotypic alteration, within p31 DNA, include the DNA 2 other open reading frames (LF1 and LF3) for which no function has polymerase (see below), BARF1 (Wei and Ooka, 1989), RF been identified (Baer et al., 1984; Farrell, 1987). For HSV-1 (bot“ 18.8’ ’ and functions within undesignated open frames such as tom), the location of the morphological transformation region (MRTI) LF1 and LF3 (Fig. 10). and the functions it encodes are shown in relation to the physical map. When the role of EBV in B-cell immortalization in vitro is Functions include a BamHIg region which acts as a mutagen in the considered, the EBNA functions, especially those of EBNA-2, Ames test, the MuX fragment which can activate endogenous retroappear to be prime candidates in inducing the cellular pheno- viruses, as well as the thvmidine kinase TTK). viral Dolvmerase (Pol). binding protein (DBP) and glycoprotein; gB. Dati adapted‘from typic alterations (Skare et al., 1985; Hammerschmidt and Sug- DNA Macnab (1987). den, 1989). EBNA-1 is of undoubted relevance also, due to its function in maintaining the EB viral genomes as episomes (Yates et al., 1985). Whether EBV DNA also integrates into the host chromosome during the genesis of a tumour is not in the light of results with herpes simplex virus which show known in the majority of cases: studies on this topic are frus- these genes to be an essential component of the observed abiltrated by the large number of extrachromosomal forms of the ity of HSV, when superinfected into cells, to amplify nonviral genome that obscure analyses. In our epithelial cell stud- herpes virus target genes within host cell chromosomes (Matz ies, the EBNA-1 gene is not present in the p31 plasmid and et al., 1984). Overall, a remarkably similar relationship exists viral DNA appears to integrate into the host chromosome. between the EBV p3 1 region and the morphologically “transCellular immortalization then takes place in the absence of all forming region” of HSV-1 (Fig. 10) where numerous genes the known EBNAs. Our findings are corroborated by the stud- with apparently similar functions are clustered (Macnab, ies on the NPC tumour (C15) which show that (except for 1987). Gene amplification of endogenous cellular functions EBNA-1) the EBNA genes are transcriptionally silent (Hitt et could also be an important aspect of the immortalization proal., 1989). Taken together, the results allow us to conclude that cess that accompanies transfection of p31 EBV DNA into epEB virally-induced immortalization of B- and epithelial cells ithelial cells. During the establishment of the MU31 epithelial need not arise by a common mechanism. Therefore, when cell line, taking the viral DNA as a “reporter” sequence, considering immortalizing functions, we must examine genes genotypic alterations observed include an amplification of the other than those which encode EBNAs. The latent membrane large repetitive (Barn HI-W) EBV sequence and loss of the protein (LMP) is not considered here since its gene does not plasmid sequence (Fig. 3); amplification of the former suggests reside within the p3 1 recombinant DNA; LMP is expressed in that the presence of the repetitive sequence in p3 1 DNA may in many, but not all, nasopharyngeal carcinomas (Fihraeus et al., itself be significant in establishing immortalized cells. 1988). The viral DNA polymerase and other functions associOnce epithelial cell lines have been fully established followated with replication, such as the DNA binding protein (found ing EBV DNA transfection, Southern blot analyses indicate within the large Hind11 and BarnHI A and I fragments; Figs. that maintenance does not appear to require the continued pres1 and 10 and Zhang et al., 1988) warrant special consideration ence of viral information. This is the case with all the cell lines

/ \

------


generated in this study, although data (Fig. 3) are only given for one. Moreover, even in early passaged cells where the presence of EBV genetic information could be identified, no viral transcripts (Northern blots) could ever be found (data not given); this is possibly of little significance, however, since low copy number transcripts are often not observed with this relatively insensitive technique. We are thus obliged to consider the possibility that the observed cellular immortalization occurs as a consequence of a “hit-and-run” mechanism, as proposed previously for HSV- 1 and 2 (Galloway and McDougall, 1983). Before pursuing this hypothesis, however, it seemed highly desirable to ascertain that EB viral information really did not persist in established cell lines. Two approaches were adopted to search for the presence of viral DNA in cells where none could be detected by conventional (DNA or RNA) hybridization procedures: sensitive in situ hybridization methods (Teo and Griffin, 1987) were employed to detect viral sequences in nuclei or metaphase chromosomes of cells from long-term cultures. In 2 such studies, our data show that large sequences of viral DNA did persist but only in a small proportion of cells (Figs. 4 and 8). They confirm that the immortalized cells are heterogeneous (Fig. 8), at least with regard to EB viral DNA and, assuming input DNA is integrated as suggested by our results, integration is random. In the other approach, late-passaged MU31 cells were selected for their ability to survive for several months in semi-solid media, and in 2 such experiments, but not in a third, sequences corresponding to the BamHI A and I fragments (Fig. 5) of EBV could then be shown by Southern blots to be present in cells surviving this selection process; these data are consistent with the in situ findings and

77 1

show that viral DNA is still present in immortalized cells even when it is undetectable by hybridization analyses. We do not yet know whether this is extraneous, non-functional DNA representing mere “ghosts” of genetic information required for earlier events, or whether the viral information retained is indeed essential, for example, allowing some epithelial cells to express factors necessary for prolonging the life of other cells in culture. Thus, at this stage, although most of our data point to a “hit-and-run” mechanism, there is no compelling evidence to support this notion, particularly as in our study, unlike that of Galloway and McDougall (1983) involving HSV, we have never succeeded in immortalizing cells with sub-gene amounts of EBV DNA. Thus, at present, there is no reason to prefer this to other mechanisms which, initially at least, might involve viral functions in the immortalization of epithelial cells. ACKNOWLEDGEMENTS

We thank Drs. B . Lane, Y. Bartek, F. Ramaekers and P. Little and Ms. S.H. Cross for advice and generous gifts of materials, Dr. J. Keeling for human embryonic kidneys, Mr. C. Kelley, FRCS, for surgical assistance in obtaining marmoset materials, and Ms. K. Coates for help with growth curves. M.H. was supported by a National Research Service Award, C.G.T. by a fellowship from the National University, Singapore, and D.K. by a bursarship from the ICRF. We thank our colleagues for helpful discussions and acknowledge generous support from the Cancer Research Campaign throughout this project.

REFERENCES ADAMS,A., The state of the virus genome in transformed cells and its relationship to host cell DNA. In: M.A. Epstein and B.G. Achong (eds.), The Epstein-Burr virus, Springer, Berlin, pp. 155-182 (1979). BAER,R., BANKIER, A., BIGGIN,M., DEININGER, P., FARRELL,P., GIBS., SEGUIN,C., SON, T., HATFULL,G., HUDSON,S., SATCHWELL, TUFFNELL,R. and BARRELL,B., DNA sequence and expression of the B95-8 Epstein-Barr virus genome. Nature (Lond.), 310, 207-21 1 (1984). BAICHWAL, V.R. and SUGDEN,B., Transformation of BALB 3T3 by the BNLF-I gene of Epstein-Ban virus. Oncogene, 2, 461-467 (1988). BOUCK,N., BEALES,N., SHENK,T., BERG,P. and DIMAYORCA, G., New region of the simian virus 40 genome required for efficient viral transformation. Proc. nat. Acad. Sci. (Wash.), 75, 2473-2477 (1978). CHANG,S.E., KEEN,J., LANE,E.B. and TAYLOR-PAPADIMITRIOU, J., Establishment and characterization of SV40-transformed human breast epithelial cell lines. Cancer Res., 42, 2040-2053 (1982). CHEVALLIER-GRECO, A., MANET,E., CHAVRIER,P., MOSNIER,C., A., Both Epstein-Barr virus (EBV)-encoded DAILLIE,J. and SARGEANT, trans-acting factors, EB 1 and EB2, are required to activate transcription from an EBV early promoter. EMBO J . , 5, 3243-3249 (1986). F ~ R A E UR., S ,Hu, L., ERNBERG, I., FENKE, J., ROWE,M., FALK,K., KLEIN, G., NILSON,K., TURSZ,T., BUSSON, P. and KALLIN,B., Expression of Epstein-Barr virus encoded protein in nasopharyngeal carcinoma. Int. J. Cancer, 42, 329-338 (1988). FARRELL,P.J., Epstein-Barr virus (B95-8 strain). In: S.J. O’Brien (ed.), Genetic maps, pp. 99-107, Cold Spring Harbor Lab. New York pp. 99107 (1987). FWHNEY,R.I., Culture of animal cells, pp. 104-118 and 204205 A.R. Liss, New York (1983). GALLOWAY, D.A. and MCDOUGALL, J.K., The oncogenic potential of herpes simplex viruses: evidence for a “hit and run” mechanism. Nature (Lond.), 302, 21-24 (1983). GIARD,D.J., AARONSON, S.A., TODARO,G.J., ARSTEIN,P., KERSEY, J.H., DOSIK,H. and PARKS,W.P., In vitro cultivation of human tumors: establishment of cell lines derived from a series of solid tumors. J. nut. Cancer Inst., 51, 1417-1423 (1973). GORMAN, C., PADMANABHAN, R. and HOWARD,B.H., High efficiency DNA-mediated transformation of primate cells. Science, 221, 55 1-553 (1983). GRIFFIN, B.E., B J ~ R C KE., , BJURSELL, G. and LINDAHL,T., Sequence

complexity of circular Epstein-Barr virus DNA in transformed cells. J . Virol., 40, 11-19 (1981). GRIFFIN, B.E. and KARRAN,L., Immortalization of monkey epithelial cells by specific fragments of Epstein-Ban virus DNA. Nature (Lond.), 309, 78-82 (1984). GRIFFIN, B.E., KARRAN,L., KING,D. and CHANG,S.E., Immortalizing genes encoded by Epstein-Barr virus In: P.W.J. Rigby and N.M. Wilkie (eds.), Viruses and cancer, pp. 93-1 10, Cambridge University Press, Cambridge (1984). HAMMERSCHMIDT, W. and SUGDEN,B., Genetic analysis of immortalizing functions of Epstein-Barr virus in human B lymphocytes. Nature (Lond.), 340,393-397 (1989). HARDWICK,J.M., LIEBERMAN, P.M. and HAYWARD, S.D., A new Epstein-Barr virus transactivator, R, induces expression of a cytoplasmic early antigen. J . Virol., 62, 22762284 (1988). HIRT,B., Selective extraction of polyoma DNA from infected mouse cell cultures. 1.mol. Biol., 26, 365-369 (1967). HITT, M., ALLDAY,M., HARA,T., KARRAN,L., JONES,M.D., BUSSON, I., EBV gene expression in P., TURSZ,T., GRIFFFIN, B .E. and ERNBERG, an NPC-related tumour. EMBO J . , 8, 2639-2651 (1989). LANE,E.B., BARTEK,J . , PURKIS,P.E. and LEIGH,I.M., Keratin antigens in differentiating skin. Ann. N.Y. Acad. Sci., 455, 241-258 (1985). LITTLE,P.H.R. and CROSS,S.H., A cosmid vector that facilitates restriction enzyme mapping. Proc. nut. Acad. Sci. (Wash.), 82, 3159-3163 (1985). MACNAB,J.C.M., Herpes simplex virus and human cytomegalovirus: Their role in morphological transformation and genital cancers. J. gen. Virol., 68, 2525-2550 (1987). MATZ,B., SCHLEHOFER, J.R.and ZUR HAUSEN,H., Identification of a gene function of herpes simplex virus type 1 essential for amplification of simian virus 40 DNA sequences in transformed hamster cells. Virology, 134, 328-337 (1984). NEUMANN, E., SCHAEFER-RIDDER, M., WANG,Y. and HOFSCHNEIDER, P.H.,Gene transfer into mouse lyoma cells by electroporation in high electric fields. EMBO J., 1, 841-845 (1982). QUINLAN,R.A., SCHILLER,D.L., HATZFELD,M., ACHSTATTER, T., MOLL,R., JORCANO, J.L., MAGIN,T.M. and FRANKE,W.W., Patterns of expression and organization of cytokeratin intermediate filaments. Ann. N.Y. Acad. Sci., 455, 282-306 (1985). RAAB-TRAUB, N., HOOD,R., YANG,C-S., HENRY,B. and PAGANO, J.S.,

772

KARRAN ET AL.

Epstein-Ban virus transcription in nasopharyngeal carcinoma. J. Virol., 48, 580-590 (1983). RAMAEKERS, F., HUYSMANS, A,, SCHAART,G., MOESKER,0. and VOOIJS,P., Tissue distribution of keratin 7 as monitored by a monoclonal antibody. Exp. Cell Res., 170, 235-249 (1987). RHEINWALD, J.G. and O'CONNELL,T.M., Intermediate filament proteins as distinguishing markers of cell type and differentiated state in cultured human urinary tract epithelia. Ann. N.Y. Acud. Sci., 455,259-267 (1985). RYMO,L., LINDAHL,T. and ADAMS,A,, Sites of sequence variability in Epstein-Ban virus DNA from different sources. Proc. nut. Acad. Sci. (Wash.), 83, 5096-5100 (1979). SKARE,J., FARLEY,I., STROMINGER, J.L., FRESEN,K.O., CHO,M.S. and ZUR HAUSEN,H., Transformation by Epstein-Ban virus requires DNA sequences in the region of BurnHI fragments Y and H. J . Virol., 55, 286297 (1985). TEO, C.G. and GRIFFIN, B.E., Epstein-Barr virus genomes in lymphoid cells: Activation in mitosis and chromosomal location. Proc. nat. Acud. Sci. (Wash.),84, 8473-8477 (1987). TUCKER, R.F., SHIPLEY,G.D., MOSES,H.L. and HOLLEY,R.W., Growth inhibitor from BSC-1 cells closely related to platelet type p transforming growth factor. Science, 226, 705-707 (1984). TUGWOOD,J.D., LAW,W-H., SAI-KI,O., TSAO,S-Y., CRAIGMARTIN,

W.M., SHIU, W., DESGRANGES, C., JONES, P.H. and ARRAND,J.R., Epstein-Barr virus transcription in normal and malignant nasopharyngeal biopsies and in lymphocytes from healthy donors and infectious mononucleosis patients. J. gen. Virol., 68, 1081-1091 (1987). VAN DER EB, A.J. and GRAHAM, F.L., Assay of transforming activity of tumor virus DNA. Meth. Enzymol., 65, 826-839 (1980). WANG,D., LIEBOWTZ,D. and KIEFF, E., An EBV membrane protein expressed in immortalized lymphocytes transforms established rodent cells. Cell, 43, 831-840 (1985).

WEDDERBURN, N., EDWARDS,J.M.B., DESGRANGES, C., FONTAINE, C., COHEN.B. and DE THB. G.. Infectious mononucleosis-like reswnse in common marmosets infected'with Epstein-Barr virus. J. If. &., 150, 878-882 (1984). WEI, M.X. and OOKA,T.,A transforming function of BARF1 gene encoded by Epstein-Ban virus. EMBO J . , 8, 2897-2903 (1989). YATES,J.L., WARREN, N. and SUGDEN, B., Stable replication of plasmids derived from Epstein-Barr virus in a variety of mammalian cells. Nature, 313, 812-815 (1985). ZHANG,C.X., DECAUSSIN, G., DAILLIE,J., and OOKA,T., Altered expression of two Epstein-Barr virus early genes localized in BumHI-A in nonproducer Raji cells. J. Virol., 62, 1862-1869 (1988).