Selection of variant hepatoma cells in liver-specific growth media: regulation at the mRNA level

Differentiation (1992) 50:25-33 Ootogeny, Nwplasia and Differentiation Therapy

0 Springer-Verlag 1992

Selection of variant hepatoma cells in liver-specific growth media : regulation at the mRNA level Lysiane Armbruster’, Catherine Cavard *,Pascale Briand ’, and Roger Bertolotti

*

’ Laboratory of Cell Biology, National Institute of Mental Health, National Institutes of Health, Bethesda, MD 20892, USA Institut Cochin de GCnitique MolCculaire, GCnCtique et Pathologie Experimentale, INSERM 22, F-75014 Pans, France

Accepted in revised form January 27, 1992

Abstract. Three liver-specific growth media, respectively free of arginine (Arg-), tyrosine (Tyr -) and glucose (G-), have been used to characterize cells of the rat H4IIEC3, human HepG2 and mouse BW hepatoma lines. Cells of clone FaO, a derivative of line H4IIEC3, freely grew in Tyr- and G - media, and gave rise to stable variants in Arg- conditions. Cells of line HepG2 and clone BWTG3, a derivative of line BW, degenerated in all three media. Arg and tyr variants were however derived from HepG2 cells ; their genesis appeared to be pathway specific, illustrating the complexity of the regulatory loops that are implicated in the control of the differentiated state. No variant was ever obtained with BWTG3 cells, demonstrating the stability of their deficiency in the post-natal hepatic functions that are involved in Arg-, Tyr- and G- selections. Variant clones of HepG2 and F a 0 cells that have been isolated in Argmedium were characterized in details for liver-specific urea-cycle enzyme activities and mRNA. These variants were shown to be controlled at the mRNA level, most likely at transcription. Isolation of stable F a 0 and HepG2 variant clones as well as the converse demonstration of the stable deficiency of BWTG3 cells in postnatal hepatic functions were aimed at expression cloning. Our results are thus discussed in terms of transfection with full-length cDNA expression libraries and cloning of regulatory genes that could activate or extinguish liver specific genes.

Introduction The pattern of specific genes a differentiated cell is able to express is thought to be stably set by determination during development. In vitro, under appropriate conditions, this ability appears to be maintained through clonal propagation [24]. However, as emphasized by Ephrus-

* To whom offprint requests should be sent at his present address: Mutagenkse el CancCrogenkse, Laboratoire d’Enzymologie, CNRS, F-91198 Gif-sur-Yvette, France

si [18], what is inherited is not merely a differentiated state as such, but the commitment to a certain type of differentiation, to the exclusion of all others. Indeed, the evaluation of the full set of differentiated functions a cell can express is usually achieved through modifications of the culture media, mainly with hormones and growth factors [44]. A complementary approach involves selective nutritional media based on tissue-specific enzymes or metabolic pathways. In this latter case, a differentiated cell is selected by its specific ability to synthesize an essential nutrient that has been omitted from growth medium [2,5,21,26,40,50]. Under these conditions, cells commited to the appropriate differentiated state either grow easily [2, 5, 261 or give rise to rare variants that would not be detectable without selective media [5, 211. We have used liver-specific growth media to characterize well-differentiated hepatoma cell lines and to generate constitutive variant clones. Liver-specific growth media were first designed by Leffert and Paul to maintain short-term primary cultures of hepatocytes and to prevent overgrowth by non-parenchymal liver cells [33]. They used an arginine-free medium for this purpose, and, in addition, did some preliminary experiments with a glucose-free one. This type of selective nutritional media found one of its main applications in somatic cell genetics of differentiation. A tyrosine-free medium, first devised for the selection of neuroblastoma cells that synthesize tyrosine hydroxylase [51, has been used to grow rat hepatoma cells that express phenylalanine hydroxylase, an hepatic enzyme that catalyzes the conversion of phenylalanine to tyrosine [ 10, 261. Then, growth in glucose-free medium was shown to be restricted to well-differentiated hepatoma cells and correlated to their gluconeogenic capability [2]. This property permitted an efficient screening for hepatoma cell hybrids that re-expressed the liver-specific gluconeogenic enzymes [3]. In addition, sequential selection of hybrid cells in medium containing either dihydroxyacetone or oxaloacetate demonstrated that survival in glucose-free conditions requires two liver-specific enzymes, fructose bisphosphatase and either triokinase or phosphoenolpyruvate carboxykinase. Arginine-free medium

26

has been used more recently with hepatoma cells [21, 40, 501. It contains ornithine and selects for the two liver-specific enzymes from the urea cycle that are involved in the conversion of ornithine to arginine, i.e. carbamoylphosphate synthetase I and ornithine transcarbamoylase. Cells of the well-differentiated rat hepatoma line H4IIEC3 and derivatives freely grow in tyrosine-free or glucose-free medium [2, 10,261. They express the appropriate liver-specific enzymes, i. e. phenylalanine hydroxylase (EC 1.14.16.1), fructose bisphosphatase (Fru-I ,6Pase, EC 3.1.3.1I), phosphoenolpyruvate carboxykinase (PEPCK, EC 4.1.1.32) and triokinase (EC 2.7.1.28). On the other hand, when inoculated in arginine-free medium, these rat hepatoma clones quickly stop growing and show extensive cell death [21, 40, 501; they express carbamoylphosphate synthetase I (CPS-I, EC 2.7.2.5) but lack ornithine transcarbamoylase activity (OTC, EC 2.1.3.3). However, with a low frequency of about they give rise to variant clones that to 3 x do grow without arginine and show significant CPS-I and OTC activities [21]. In this paper, we describe how we have used the aforedescribed liver-specific growth media to characterize clone FaO, a derivative of the H4IIEC3 line, together with cells from the human HepG2 and mouse BW hepatoma lines. After a general description of the growth properties of these three hepatoma lines, we focus on the variants that we have isolated in arginine-free medium and show that CPS-I and OTC are controlled at the mRNA levels in these hepatoma cells.

Methods Cell lines and culture conditions. The human HepG2 line was supplied by Drs Knowles and Aden who adapted these cells to in vitro culture from an hepatoblastoma biopsy [31]. Clone F a 0 is an H PRT- (hypoxanthine-guanine phosphoribosyl transferase negative) derivative of the rat hepatoma line H4ITEC3 [42]; it is resistant to 8-azaguanine and to ouabain [17]. BWTG3 is an HPRTclone derived from the mouse hepatoma line BW [35]; for sake of cell homogeneity, subclone BTG9A [49] was used in the present work. Rat FR3T3 [46] and human MRCS [28] and W138 [27] are fibroblastic lines. Cells were maintained in modified Ham's F12 medium [13], supplemented with 5% dialyzed fetal calf serum in a 5% C 0 2 incubator. Fetal calf serum was dialyzed as previously described [2]. Tyrosine-free, arginine-free and glucose-free media were supplemented with 0.3 mM L-phenylalanine, 1 mM [.-ornithine and 2 mM oxaloacetate respectively. Selection experiments were performed in 10 cm culture dishes that were inoculated with 1-2 x lo6 cells. Individual variant colonies were transferred to 24 well plates for clonal expansion.

OTC and CPS-I assuyys. Frozen samples (- 80" C) were homogenized in cold 0.1 YOcetyltrimethyl ammonium bromide and assayed for enzyme activities [8]. OTC was determined according to the method of Snodgrass [47] at pH 8.0 or 9.0 with 0.2 M triethanolamine-HCI buffer, 5 m M ornithine and 5 mM carbamoylphosphate. In control kinetic studies, one of the two substrates was at a 5 mM final concentration while the other ranged from 0.04 to 5 m M . Citrulline concentration was measured with an automated colorimetric method.

CPS-I was assayed at pH 7.5 as described by Brown and Cohen [6] where the conversion of the end-product of the reaction, carbamoylphosphate, to citrulline is monitored by L-ornithine and purified OTC (Preortan, Laboratoires Leuquin, Paris). Controls include reactions without purified OTC, N-acetylglutamate and C03HNH4 respectively. Protein concentrations were determined according to the method of Lowry et al. [34]. Northern blots. Cytoplasmic RNA was prepared from cell cultures that were lysed with 0.5% (vol/vol) Nonidet P-40, centrifuged to remove nuclei and organelles, and extracted with phenol and chloroform. Total RNA was extracted from liver samples according to Okayama et al. [41]. Poly(A)+ RNA was selected by two cycles of binding to oligo(dT)-cellulose. RNA was denatured and electrophoresed through agarose formaldehyde gels, blotted onto Genescreen membranes and hybridized to 32P-labelledprobes according to standard procedures [37]. The different probes that we used comprise: pAct-I [48], Sub A [45], A7h [38]. pCPSr4 [I], and pMN 152 [391 and its pBOTC derivative [9].

Southern blots. Genomic DNA was prepared according to a modification of the method of Wigler et al. [51]. Gel electrophoresis was followed by alkaline transfer on Zetaprobe membranes [43] and blot hybridization was performed as described by Church and Gilbert [ll].

Results Growth properties of FaO, HepG2 and B WTG3 hepatoma cells in liver-specific culture media

The three clones of well-differentiated hepatoma cells we used were first adapted to growth in 5% dialyzed serum. Then, they were plated in Arg-, Tyr- and G-OA, the aforedescribed liver-specific growth media free of arginine, tyrosine and glucose respectively. In the present work, only one glucose-free medium has been used; it contains 2 mM oxaloacetate (G-OA). As shown in Table 1, only rat F a 0 cells are able to survive in all three media. As reported for other derivatives of the H4IIEC3 line [2, 10, 21, 26, 40, 501, F a 0 freely grew in Tyr- and G-OA media, and gave rise to rare variants in Arg- conditions. Arg variants arose at a frequency By contrast, cells of clone BWTG3, a of about Table 1. Selective growth of well-differentiated hepatoma lines in liver-specific media. Cells (1-2 x 10') were suspended in the appropriate selective medium and plated in 10 cm culture dishes. Sizeable colonies appeared 10-21 days after inoculation. Each selection involved 107-108 cells. V indicates extensive cell death and selective growth of rare variants. Values in parentheses correspond to variant frequencies Cell lines Selective media

-

Arg-

Tyr-

F a 0 (rat) HcpG2 (human) BWTG3(mouse) Control fibroblasts *

v (10-5) v

v

-

+ -

G-OA

+

-

* Control fibroblasts comprised rat FR3T3, and human MRCS and W138

27 HepG2

no survivor

Hep-tyr5

1 1

Hep-erg1

ARG-

Hep-tyr5-arg A

1-

Hep-argl -tyrA

thioguanine

Hep-5A- 8 1

1c

EBNA-APH.11

G418

Hep- E BN A-2

Fig. 1. Genealogy of HepG2 derivatives. Selection of variant cells was performed either in Tyr-, Arg- or G-OA medium, or in standard medium containing either thioguanine (0, 15 pg/ml) or G418 (300 pg/ml). The left part of the figure has been previously described [36]. EBNA-APH.11 stands for transfection with plasmid pSVoB-H2.9 [53] that comprises the EBNA-1 gene and the bacterial APH.11 gene encoding aminoglycoside 3'-phosphotransferase 11. APH.11 is under the control of the promotor and polyadenylation signal of the thymidine kinase gene from the Herpes simplex virus and confers resistance to G418 [ 121

Table 2. Cell growth in arginine-free medium and CPS-I/OTC activities

Cell lines

derivative of the mouse BW line, did not survive in any of the three selective media; no variant has ever been obtained out of the 3 x 107-108 cells that have been plated in each medium. The human HepG2 line stands in between F a 0 and BWTG3. HepG2 cells degenerate in all three selective media but gave rise to tyr and arg variants at a frequency of about More than 10' HepG2 cells have been challenged without success to G-OA medium. As expected, rat FR3T3 and human MRC5 and W138 control fibroblasts died in all three liver-specific growth media. To obtain evidence regarding putative cross-regulations between the two pathways that are involved for survival in Tyr- and Arg- respectively, HepG2 were subjected to sequential selection in both media. As outlined in Fig. 1, HepG2 cells were first inoculated in Tyror Arg-. In both cases, variants appeared at a frequency when 2 x lo6 cells were plated in 10 cm of about culture dishes. Ten such tyr and arg clones were isolated, expanded in Tyr- and Arg - conditions respectively, and shown to be stable. One of each, Hep - tyr5 and Hep argl, were transferred to the converse media. In both cases, extensive cell death occurred and variants arose at a frequency of about suggesting that the genesis of tyr and arg variants are independent processes. Five of each were then isolated, and representative Heptyr5 -argA and Hep - argl - tyrA variant clones were further characterized. Clone Hep - tyr5 - argA was used to generate cellular tools aimed at the expression cloning of regulatory genes [36]. First, in order to promote eftiEnzyme activity * Growth and

CPS-I

Survivor frequency . .

OTC in arginine-free medium

Rat Hepatomas Fa0 FAG1 Human Hepatomas HepG2 Hep - argl Hep -argl -tyrA Hep - tyr5 Hep-tyr5 -argA Hep- 5A - 01 Hep - EBNA-2 Mouse Hepatoma BWTG3 Fibroblasts MRC-5 (Human) FR3T3 (Rat) Liver Human Rat Mouse

3.0+2 3.8f2 <0.1

-

< 0.1 25+ 4

27+ 325 35+ 30+ 47+ 45+

2 2 2 2 2 4

+ -

10-5 1 10-4

110+10

+ + + + +

t0.1

-


<0.1

<0.1 <0.1

-

< 10-

<0.1

2.0f 1 2.5 & 2 <0.1 6.0+2 3.7 k 1

16.0+5 <0.1

30.0 f8 32.0+6 24.0+7

592 f160 917+ 83 1250+170

N.T. N.T. N.T.

1 1

10-5 1 1 1


N.T. N.T. N.T.

* Specific activities are expressed as nmol citrulline/min per mg protein. At least two independent cell extracts have been used for each clone; in the case of liver, values were generated from ten independent samples. N.T.. not tested

28

cient co-transfection experiments, the HPRT- marker (hypoxanthine-guanine phosphori bosyl transferase deficiency) was introduced into these cells and clone Hep 5A - 81 was isolated. Then, Hep - EBNA-2 was derived that constitutively expresses the Epstein-Barr nuclear antigen and can, therefore, accomodate EBV-oriP shuttle vectors. Double arg - tyr variants such as Hep - 5A - 81 and Hep- EBNA-2 cells were checked for their surviving capability in G-OA medium. Like the original HepG2 cells, they did not give rise to any variant colony out of 4 x lo7 plated cells [36].

1

2

3

4

5 4-0

- 28s

- 18s CPS-I and OTC activities in FaO, HepG2, BWTG3 and their vuriant cells Growth in Arg- medium requires the simultaneous presence of both CPS-I and OTC activities. A precursor of the rat hepatoma clone F a 0 and its arg variants were previously analzyed for their CPS-I and OTC activities [21]. We have shown that, like its precursor, F a 0 lacks OTC activity only; it exhibits a fair CPS-I activity that corresponds to about 10% that of the rat liver (see Table 2). Five arg variant clones were isolated from FaO. One of them, FAGI, was characterized in detail. Like the arg variants described by Goss [21], it expresses a significant level of OTC specific activity, about 3% that of the rat liver. By contrast, CPS-I activity is just slightly increased as compared to FaO. The converse situation holds true for HepG2 and Hep-tyr5 cells that are, like F a 0 cells, unable to grow in Arg- medium. They have a fair amount of OTC but are devoid of any detectable CPS-I activity. Interestingly, comparing CPS-I and OTC activities from Hep- tyr5 to those of the original HepG2 cells shows that selection in Tyr- did not affect significantly the level of these urea cycle enzymes. As expected, all arg variants exhibit a significant amount of CPS-I activity, from 10% to 50% that of the human liver. The original level of OTC specific activity in HepG2 and Hep- tyr5 is compatible with growth in Arg- medium; it corresponds at least to FAGl level. However, OTC activity is increased in most arg variants. Interestingly, Hep - EBNA-2 cells that express the Epstein-Barr nuclear antigen possess the highest activity described to date in vitro for both CPS-I and OTC. Neither CPS-I nor OTC activity was ever detected in mouse BWTG3 hepatoma cells. Although they are differentiated hepatoma cells, they compare to control MRC-5 and FR3T3 fibroblasts rather than to F a 0 or HepG2 cells. Northern blot analysis of CPS-I and OTC from rodent hepatoma cells and control fibroblasts

Northern blots were first performed with cytoplasmic RNA extracted from FaO, FAG1, BWTG3 and from FR3T3 control fibroblasts. Total RNA prepared from rat liver was used as a reference. Blots were probed either

Fig. 2. Northern blot analysis of cytoplasmic CPS-I RNA from FaO, FAG1, BWTG3 and FR3T3 rodent cells. Cytoplasmic RNA from FR3T3 fibroblasts (lane 1 , 10 pg), BWTG3 ( / m e2, 10 pg), Fa0 (lane 3, 10 pg) and FAGl (lane 4, 5 pg) and total RNA from rat liver (lane 5,0.5 pg) were electrophoresed through a 0.8% agarose formaldehyde gel, blotted onto a Genescreen membrane, hybridized with 32P-labelled pCPSr4 probe and autoradiographed for 4 days at -70" C with an intensifying screen. Adjacent lanes were stained with ethidium bromide to determine the positions of 18 Sand 28 S rRNA bands, i. e. 2.1 and 5.1 kb respectively

for CPS-I or OTC. Under these conditions, only CPS-I mRNA were readily detected (Fig. 2). As expected, the CPS-I probe hybridized to a single major mRNA species corresponding to about 5.7 kb, found in RNA extracted from liver (lane 5), FAGl (lane 4) and F a 0 (lane 3), and not in RNA from BWTG3 (lane 2) and FR3T3 fibroblasts (lane 3 ) . The relative intensities of the bands roughly correspond to CPS-I activities presented in Table 2. The tenfold difference in CPS-I activity between rat liver and FAGl or F a 0 is maintained at the mRNA level since the intensity of signal is equivalent with 0.5 pg of liver RNA (lane 5) and 5 pg of FAGl RNA (lane 4) and to half the 10 pg of F a 0 RNA (lane 3). The sensitivity of the method suggests that the CPS-I mRNA content of BWTG3 and FR3T3 is less than 5% that of FaO. In order to clearly characterize OTC mRNA from these rodent cell lines, total RNA were fractionated through oligo(dT)-cellulose columns and poly(A)+ RNA were used for Northern blot analysis. As illustrated on Fig. 3, the OTC probe hybridized with an expected 1.8 kb mRNA species that is specific to liver (lane 4) and FAG 1 (lane 3). Again, the relative abundances of mRNA grossly correspond to OTC relative activities from Table 2. A 40-fold difference in OTC activity distinguishes rat liver from FAG1 and the signal from 0.2 pg of rat liver poly(A)+ RNA probed with 32P-OTC (lane 4) roughly compares to twice the 4 pg of FAGl poly(A)+ RNA (lane 3). On the other hand, in agreement with enzyme activities shown in Table 2, OTC

--ALBUMIN

OTC

1

2

3

4

1

2

3

29

and 2). This blot was also probed for p-actin, a ubiquitous protein common to most cell types, and for albumin, a liver-specific protein that is expressed by the original clone of FaO. Albumin was chosen as a reference for hepatic differentiation because it is not involved in Argselection. In addition, its mRNA is about the same size as that of OTC (2.3 kb) and is therefore transferred to the Genescreen membrane with an efficiency that can compare OTC. The second panel (Fig. 3) demonstrates that albumin mRNA is indeed present in FaO, FAGl and rat liver, and is absent from FR3T3 fibroblasts. Albumin mRNA appears less abundant in FAGl than in FaO, even if we account for the fact lane 2 (FaO) was loaded with 12pg of poly(A)+ RNA and lane3 (FAG1) with 4 pg only. fi-actin was used as a quantitative reference for all mRNA and a qualitative control for FR3T3 mRNA. The third panel from Fig. 3 shows that this ubiquitous mRNA migrated at its expected position (2.1 kb), between the OTC and albumin mRNA species, and that it is more abundant in FR3T3 fibroblasts (lane 1) than in rat hepatoma cells (lane 2 and 3). The weak signal from hybridization with rat liver poly(A)+ RNA reflects the amount of mRNA loaded on the gel.

ACT1N

4

1

2

3

4

5 4-0

b

- 28s 0

Fig. 3. Northern blot analysis of poly(A)+ RNA from FaO, FAGl and FR3T3 rat cells. Poly(A)+ RNA from FR3T3 fibroblasts (lane I , 12 pg), F a 0 (lane 2, 12 pg), FAGl (lane 3, 4 pg) and from rat liver (lane 4 0 . 2 pg) were electrophoresed through a 1% agarose formaldehyde gel, blotted onto a Genescreen membrane, successively hybridized with 32P-labelledprobes (pBOTC, albumin sub A and pAct-1, respectively) and autoradiographed either at -70" C with an intensifying screen (OTC for 5days, albumin for 14h) or at room temperature (actin for 30 h). Adjacent lanes containing liver cytoplasmic RNA were stained with ethidium bromide to visualize the 18 S and 28 S rRNA markers

Northern blot analysis of CPS-I from human hepatoma cells and control fibroblasts Preliminary Northern blot experiments were performed with cytoplasmic RNA extracted from the original HepG2 cells, some of their arg variants and control human MRC5 fibroblasts. They demonstrated that CPS-I mRNA was absent from HepG2 and control MRCS cells, and present in Hep - argl, Hep - argl - tyrA, Hep- 5A - 81 and Hep- EBNA-2. In order to improve the sensitivity of mRNA detection and the quality of the blots, poly(A) RNA were prepared from represen-

mRNA was undetectable in F a 0 (lane 2), FR3T3 (lane 1) and BWTG3 (not shown); overexposed autoradiograms would have detected 10% of the OTC contained in the 4 pg of FAGl RNA from lane 3, i.e. 3% of the FAGl content with a 12 pg RNA load (lanes 1

--CPS

1

2

3

4

.

+

APO B

5

r

..-

6

4

7

c

1

2

3

4

ACTIN

5

6

7

1

2

3

4

5 6 7

0

- 28s

- 18s

Fig. 4. Northern blot analysis of poly(A)+ RNA from HepG2, Hep - 5A - 81, Hep - EBNA-2 and MRCS human cells. Poly(A)+ RNA from MRCS fibroblasts (lane 1, 10 pg), HepG2 (lane 2, 10 pg), Hep - SA - 81 (lane 3, 5 pg), Hep-EBNA-2 (lane 4, 5 pg) and from human liver (lanes 5-7, 0.25 pg, 0.5 pg and 2.5 pg respectively) were electrophoresed through a 0.8% agarose formaldehyde gel, blotted onto a Genescreen membrane, successively hybridized with 32P-labelled probes (pCPSr4, apoB A7h and pAct-I, respectively) and autoradiographed either at -70" C with an intensifying screen (CPS-I for 5 days, apoB for 10 h) or at room temperature (actin for 26 h)

-

30

tative HepG2 clone and subclones, from MRCS fibroblasts and from human liver. The first panel from Fig. 4 shows that a 5.7 kb RNA species corresponding to CPSI is indeed restricted to human liver (lanes 5-7), HepEBNA-2 (lane 4) and Hep - 5A - 81 (lane 3). This RNA is undetectable in HepG2 (lane 2) and MRCS (lane 1) preparations although the corresponding lanes were loaded with twice the amount of poly(A)+ RNA as Hep - 5A - 81 and Hep - EBNA-2. Again, the relative intensities of the bands roughly correspond to CPS-I activities from Table 2. The 2.5 pg of human liver RNA from lane 7 compare to the 5 pg of Hep - EBNA-2 RNA from lane 4, i.e. a twofold difference (see Table 2). Likewise, there is an apparent two to fourfold difference between Hep- EBNA-2 and Hep - 5A - 83 (lanes 4 and 3 respectively). The sensitivity of the techniques suggests that the CPS-I mRNA content of HepG2 and MRCS is at most 5% that of Hep- 5A - 81. Control hybridizations were performed for the ubiquitous B-actin and, in this case, for apolipoprotein B (apoB) encoded, like CPS-I, by a high molecular weight mRNA. HepG2 cells (lane 2, middle panel) contain the 14.1 kb apoB liver mRNA (lane 7) and, in addition, a 7.5 kb mRNA that is usually restricted to the intestine [32, 381. ApoB mRNA is maintained in HepG2 derivatives (lanes 3, 4), and the relative abundance of both forms fluctuates from one clone to the other. The third panel in Fig. 4 demonstrates that p-actin mRNA is equally well-expressed in human liver (lane 7), in HepG2 (lane 2) and its subclones (lanes 3 and 4).

BamH I

EcoR I

1 2 3

4

A

7

2

3

Taq I

2

A

c

2

3

2 * O

-- 239..14kb - 6.6 - 4.4

u

-

2.3 -2.0

Fig. 5. Southern blot analysis of genomic OTC from FaO, FAGl and FR3T3 rat cells. Genomic DNA was extracted from FR3T3 fibroblasts (lane I ) , F a 0 (lane 2), FAGl (lune3) and rat liver (lane 4, aliquoted and digested to completion with EcoR1, BurnHI and TaqI respectively. Ten pg of each DNA digest was loaded on a 0.7% agarose gel, electrophoresed in TRIS-borate buffer, transferred to a Zetaprobe membrane, hybridized to the 32P-labelled Xha I - Nsi I OTC insert of pMN 152 and autoradiographed for 7 days at -70" C with an intensifying screen. Molecular weight markers correspond to 1 DNA cut with HindIII

and human controls on the other hand. Blots were hybridized with an OTC and a CPS-I probes, respectively. A typical OTC autoradiogram of a Southern blot of rat DNA digests is shown on Fig. 5. Lanes 1-4 contain genomic DNA from FR3T3 fibroblasts, FaO, FAG1 and liver, respectively, digested to completion either with EcoRI, BamHI or TaqI. Interestingly, in each panel, all four samples present the same patterns of bands and, on repeated experiments, do not exhibit any

Southern blot analysis of rat genomic OTC and human genomic CPS-Ifrom hepatoma cells and their arg variants

To evaluate if the genesis of the arg variants resulted from real genetic regulation or from gene amplification or chromosome rearrangements, Southern blot analysis were performed on FaO, FAGl and rodent controls on the one hand, and on HepG2, representative variants

--EcoR I

1 2 3 4

Pst I

5

1 2 3 4

Taq I

5

1 2 3 4

5

4-0

MMbw Y

**I.! '

k t

( 1

* 4 r 4 0

b4

- 236 kb

prthr q -- : ; 4.4

. ?

r r y w

e

-

2.3 2D

Fig. 6. Southern blot analysis of genomic CPS-I from HepG2, Hep - 5A - 81, Hep - EBNA-2 and MRCS human cells. Genomic DNA was extracted from MRCS fibroblast (lane I), HepG2 (lane 2), Hep - 5A - 81 (lune 3), Hep - EBNA-2 (lane 4 ) and human liver ( h e 5). aliquoted and digested to completion with EcoRI, PstI and Taql respectively. Seven pg of EcoRI digests, 10 pg of PstI digests and 8 kg of HindIII digests were loaded on a 0.7% agarose gel, electrophoresed in TRlS-borate buffer, transferred to a Zctaprobe membrane, hybridized to the 32P-labelled 850 bp PstI fragment or pCPSr4 and autoradiographed for 24 h at -70" C with an intensifying screen. Molecular weight markers correspond to 1. DNA cut with Hind111

31

significant difference in intensity. The weak hybridization of lane 4 (rat liver) from the EcoRI panel was the result of underloading and was not found on a sister blot. With these genomic blots, no significant alteration was ever found in any sample that we have analyzed. Southern blots of human DNA digests are illustrated on Fig. 6. A typical CPS-I autoradiogram is shown for EcoRI, PstI and TuyI digests. Lanes 1-5 correspond to MRC5 fibroblasts, HepG2, Hep - 5A- 01, HepEBNA-2 and liver respectively. As emphasized for rat OTC, in each panel all samples show the same patterns of bands and do not exhibit any significant difference in intensity. Lane 5 (human liver) from the EcoRI panel and lane 2 (HepG2) from the PstI panel were overloaded in the gel presented on Fig. 6 .

Discussion Terminal differentiation is achieved during development through a stepwise process consisting mainly of the sequential and discontinuous establishment of clusters of differentiated functions [25]. Discrete ontogenic stages have been thus defined for many cell lineages, in particular for hepatocytes. The well-differentiated hepatoma lines that we are using exhibit specific patterns of liver differentiation : the H4IIEC3 rat hepatoma line expresses neonatal-adult functions while the mouse BW line appears to be restricted to fetal ones [7, 201. Although phenotypically very close to the H4IIEC3 hepatoma, the human HepG2 line synthesise a-fetoprotein [31, 361. We have used liver-specific growth media to delineate some of the potentialities of these three hepatoma lines and to evaluate the stability of their phenotype. Consistent with the data gained with somatic cell hybrids [7] and recently published results [19, 201, mouse BWTG3 cells were shown, in the present work, to be unable to survive in Tyr-, Arg- or G-OA medium that select for neonatal-adult liver functions. No variant was ever obtained, demonstrating the stability of their deficiency in these post-natal hepatic functions. By contrast, rat F a 0 cells freely grew in Tyr- and G-OA, and gave rise to variants in Arg- conditions. HepG2 cells stand in between; the majority of them degenerate in all three media but they provided for arg and tyr variants. Thus, BWTG3 and HepG2 cells are amenable to transfection experiments where hepatic gene activations will be attempted using full-length cDNA expression libraries of adult liver (Bertolotti, Armbruster and Okayama, submitted). In addition to the cloning of putative monospecific trans-acting genes, this approach offers an attractive opportunity to isolate some hepatic counterparts of the MyoD/myogenin genes [14, 521 that could promote a pseudo-ontogenic transition from an immature hepatic phenotype to a more mature one. A gene of this type should monitor the activated expression of a group of late hepatic functions and, possibly, the silencing of some early functions such as cr-fetoprotein synthesis. This approach appears all the more feasible with BWTG3 cells as the silent neonatal hepatic genes that

have been analyzed can be activated under appropriate conditions [19, 201. Northern blot analysis showed that CPS-I and OTC mRNA are undetectable in BWTG3 and fibroblastic cells. Thus, as expected [16], the CPS-I and OTC genes, which are not included in the differentiation programs of the aforedescribed cells, are controlled at the RNA level, most likely at the transcriptional level. In the case of the cells than can generate variants, our data clearly show that the regulation of OTC or CPS-I is also at the mRNA level. Southern blot analysis of genomic DNA did not reveal any obvious gene amplification or chromosome rearrangements in the variant cells. We deal, therefore, with real genetic regulations. The fact that 5-azacytidine can induce OTC expression in H4IIEC3 clones [ 15,231 suggests, indeed, that transcriptional regulations are directly involved in the genesis of the arg variants. This drug generates stable hypomethylation of DNA and, in doing so, is believed to lead to transcriptional activation [ 301. In addition, the apparent independence of the genesis of arg and tyr variants of HepG2 illustrates the complexity of the regulatory loops that control the differentiated state. Hormonal supplements are indispensable for CPS-I production in normal liver cells in vitro [22]. Thus, growth in Arg- medium appears to select for rare hormone-independent variants that constitutively express the CPS-I and OTCase genes. The arg variants of HepG2 and of F a 0 that we have isolated are stable. They continue to grow in Arg- medium even after extensive passages in complete medium. In addition, no significant difference in CPS-I and OTC specific activities were ever found when variant cells such as FAGl or Hep- 5A - 61 were maintained as mass culture either in arg- or complete medium. Preliminary experiments with Hep - 5A -01, Hep - EBNA-2 and FAGl suggest that the frequency of revertants is about l o w 6under normal growth conditions. Such a stability will be reinforced with hormonal supplements [20] and should accomodate experiments aimed at the cloning of extinguishing genes for the urea cycle [4, 291. Acknowledgments. We wish to thank H. Okayama and M. Brownstein for their interest and support, W. O’Brien for providing plasmid pCPSr4, and I. Levilliers for critical reading of the manuscript. This work was initiated at the Centre de Gtnetique Molkculaire du CNRS (Gif-sur-Yvette, France) where it was supported by grant to R.B. from the CNRS (ATP Biologie Moleculaire du Gene), INSERM (contrat libre 821002) and the Association pour la Recherche SUT le Cancer (contrat 6137). C.C. and P.B. received support from the Association pour la Recherche sur le Cancer (contrat 6744) and from the Ligue Nationale contre le Cancer. A.L. was a recipient of a Lavoisier felloswhip from the French Government and of a postdoctoral fellowship from the Association pour la Recherche sur le Cancer. R.B. was a recipient of an Eleanor Roosevelt International Cancer Research fellowship funded by the American Cancer Society.

References 1. Adcock M, OBrien W (1984) Molecular cloning of cDNA for

rat and human carbamyl phosphate synthetase I. J Biol Chem 259: 13471-13476

32 2. Bertolotti R (1977) A selective system for hepatoma cells producing gluconeogenic enzymes, Somatic Cell Genet 3 : 365-380 3. Bertolotti R (1977) Expression of differentiated functions in somatic cell hybrids : selection in glucose-free media of segregated hybrid cells which reexpress gluconeogenic enzymes. Somatic Cell Genet 3 :579-602 4. Boshart M, Weih F, Nichols M, Schutz G (1991) The tissuespecific extinguisher locus TSEl encodes a regulatory subunit of CAMP-dependent protein kinase. Cell 66: 849-859 5. Breakefield X, Niremberg M (1974) Selection for neuroblastoma cells that synthesize certain transmitters. Proc Nat Acad Sci USA 71 :2530-2533 6. Brown G, Cohen P (1959) Comparative biochemistry of urea synthesis. I. Methods for the quantitative assay of urea cycle enzymes in liver. J Biol Chem 234: 1769-1774 7. Cassio D, Weiss M (1979) Expression of fetal and neonatal hepatic functions by mouse hepatoma-rat hepatoma hybrids. Somatic Cell Genet 5:719-738 8. Cavard C, Cathelineau L, Rabier D, Briand P (1987) Immunochemical analysis of nineteen ornithine transcarbamoylase deficiencies. Enzyme 22 :375L383 9. Cavard C, Grimber G, Dubois N, Chasse JF, Bennoun M, Minet-Thuraux M, Kamoun P, Briand P (1988) Correction of mouse ornithine transcarbamylase deficiency by gene transfer into the germ line. Nucleic Acids Res 16:2099-2110 10. Choo K, Cotton R, Danks D (1976) Phenylalanine hydroxylation and tyrosine requirement of cultured cells. Exp Cell Res 101 :370-382 11. Church G, Gilbert W (1984) Genomic sequencing. Proc Natl Acad Sci USA 81 :1991-1995 12. Colb&re-Garapin F, Horodniceanu F, Kourilsky P, Garapin A (1981) A new dominant hybrid selective marker for higher eukaryotic cells. J Mol Biol 15O:l-14 13. Coon HG, Weiss MC (1969) A quantitative comparison of formation of spontaneous and virus-produced viable hybrids. Proc Natl Acad Sci USA 62:852-859 14. Davis RL, Weintraub H, Lassar AN (1987) Expression of a single transfected cDNA converts fibroblasts to myoblasts. Cell 5 1 :987-1000 15. Delers A, Szpirer J, Szpirer C, Saggiorno D (1984) Spontaneous and 5-azacytidine-induced reexpression of ornithine carbamoyl transferase in hepatoma cells. Mol Cell Biol4:809-812 16. Derman E, Krauter K, Walling L, Weinberger C, Ray M, Darnell J (1981) Transcriptional control in the production of liverspecific mRNA. Cell 23:731-739 17. Deschatrette J, Weiss M (1974) Characterization of differentiated and dedifferentiated clones from a rat hepatoma. Biochimie 56:1603-1611 18. Ephrussi B (1972) Hybridization of Somatic Cells. Princeton University Press, Princeton N.J. 19. Fanner AA, Goss SJ (1991) BWTG3 hepatoma cells can acquire phenylalanine hydroxylase, cystathionine synthetase and CPS-I without genetic manipulation, but activation of the silent OTC gene requires cell fusion with hepatocytes. J Cell Sci 98: 533-538 20. Faust D, Imaizumi-Scherrer T, Fulchignoni-Lataud MC, Catherin AM. lost I, Weiss M (1991) Activation of phenylalanine hydroxylase expression following genomic DNA transfection of hepatoma cells. Differentiation 44:74-79 21. Goss SJ (1984) Arginine synthesis by hepatomas in vitro. I. The requirements for cell growth in medium contalning ornithine in place of arginine and the isolation and characterization of variant hepatomas auxotrophic for arginine. J Cell Sci 68:285-303 22. Goss SJ (1984) Arginine synthesis by hepatomas in vitro. 11. Isolation and characterization of Morris hepatoma variants unable to convert ornithine to arginine, and modulation of ureacycle enzymes by dexamethasone and cyclic-AMP. J Cell Sci 68 :30S319

23. Goss SJ (1984) The associated reactivation of two X-linked genes. The spontaneous and azacytidine-induced reexpression of ornithine transcarbamoylase and glucose-6-phosphate dehydrogenase in a rat hepatoma. J Cell sci 72:241-257 24. Green H, Todaro G (1967) The mammalian cell as a differentiated microorganism. Annu Rev Microbiol 21 :573400 25. Greengard 0 (1971) Enzymic differentiation in mammalian tissues. Essays Biochem 7: 159-205 26. Haggerty D, Young P, Buese J (1975) A tyrosine-free medium for the selective growth of cells expressing phenylalanine hydroxylase activity. Dev Biol44: 158-168 27. Hayflick L (1965) The limited in vitro lifetime of human diploid cell strains. Exp Cell Res 37:614-636 28. Jacobs J, Jones C, Baille J (1970) Characteristics of a human diploid cell designated MRC-5. Nature 227 :168-1 70 29. Jones KW, Shapero MH, Chevrette M, Fournier REK (1991) Substrative hybridization cloning of a tissue-specific extinguisher: TSEl encodes a regulatory subunit of protein kinase A. Cell 66:861-872 30. Jones P, Taylor S (1980) Cellular differentiation, cytidine analogs and DNA methylation. Cell 20: 85-93 31. Knowles BB, Howe CC. Aden DP (1980) Human hepatocellular carcinoma cell lines secrete the major plasma proteins and hepatitis B surface antigen. Science 209 :497-499 32. Law S, Monge J, Lackner K, Hospattankar A, Grant S, Higuchi K, Gregg R, Lee N, Brewer H (1986) Human apolipoprotein B: cloning and analysis of liver and intenstinal mRNA and their expression in patients with abetalipoproteinemia. In : Gallo LL (ed) Cardiovascular disease 86. Molecular and cellular mechanisms, preventions, treatment. Plenum Press, pp 21-32 33. Leffert H, Paul D (1972) Studies on primary cultures of differentiated foetal liver cells. J Cell Biol 52:559-568 34. Lowry 0, Rosebrough N , Farr A, Randall R (1951) Protein measurement with the Folin phenol reagent. J Biol Chem 193~265-275 35. Lutfalla G, Blanc H, Bertolotti R (1985) Shuttling of integrated vectors from mammalian cells to E. coli is mediated by head-totail multimeric inserts. Somat Cell Mol Genet 11 :223-238 36. Lutfalla G, Armbruster L, Dequin S, Bertolotti R (1989) Construction of an EBNA-producing line of well-differentiated human hepatoma cells and of appropriate Epstein-Barr virusbased shuttle vectors. Gene 76:27-39 37. Maniatis T, Frisch EF, Sambrook J (1982) Molecular cloning, a Laboratory manual. Cold Spring Harbor laboratory, Cold Spring Harbor, N.Y. 38. Mehrabian M, Schumaker V, Fareed G, West R, Johnson D, Kirchgessner T, Lin HC, Wang X, Ma Y, Mendiaz E, Lusis A (1985) Human apolipoprotein B: identification of cDNA clones and characterization of mRNA. Nucleic Acids Res 13:6937-6953 39. Nguyen M, Argan C, Lusty C, Shore G (1986) Import and processing of hybrid proteins by mammalian mitochondria invitro. J Biol Chem 261 : 800-805 40. Niwa A, Yamamoto K, Yasumura Y (1979) Establishment of a rat hepatoma cell line which has ornithine carbamyltransverase activity and grows continuously in arginine-deprived medium. J Cell Physiol98: 177-1 84 41. Okayama H, Kawaichi M, Brownstein M, Lee F, Yokota T, Arai K (1987) High-efficiency cloning of full-length cDNA, construction and screening of cDNA expression libraries for mammalian cells. Methods Enzymol 154: 3-28 42. Pitot M, Peraino C, Morse P, Potter VR (1964) Hepatomas in tissue culture compared with adapting liver in vitro. Natl Cancer Mongr 12:229-242 43. Reed KC, Mann D (1985) Rapid transfer from agarose gels to nylon membrane. Nucleic Acids Res 13: 7207-7221 44.Rudnicki M, McBurney M (1987) Cell culture methods and induction of differentiation of embryonal carcinoma cell lines. In: Robertson EJ (ed) Teratocarcinomas and embryonic stem cells, a practical approach. TRL Press, pp, 19-49

33 45. Sargent T, Jagodzinski L, Yang M, Bonner J (1981) Fine structure and evolution of the rat serum albumin gene. Mol Cell Biol 1 :871-883 46. Seif R, Cuzin F (3977) Temperature-sensitive growth reguiation in one type of transformed rat cells induced by the tsa mutant of polyoma virus. J Virol24: 721-728 47. Snodgrass P (1968) The effects of pH on the kinetics of human liver ornithine-carbamylphosphate transferase. Biochemistry 7: 3047-3051 48. Spiegelman B, Frank M, Green H (1983) Molecular cloning of mRNA from 3T3 adipocytes J Biol Chem 258: 10083-10089 49. Szpirer C, Szpirer J (1975) A mouse hepatoma cell line which secretes several serum proteins including albumin and a-foetoprotein. Differentiation 4 : 85-91

50. Widman L, Golden J, Chasin L (1979) Immortalization of normal liver functions in cell culture: rat hepatocyte-hepatoma cell hybrids expressing ornithine carbamoyl transferase activity. J Cell Physiol 100:391-400 51. Wigler M, Sweet R, Sim GK, Wold B, Pellicer A, Lacy E, Maniatis T, Silverstein S, Axel R (1979) Transformation of mammalian cells with genes from procaryotes and eucaryotes. Cell 16: 777-785 52. Wright WE, Sassoon DA, Lin VK (1989) Myogenin, a factor regulating myogenesis, has a domain homologous to MyoD. Cell 56 :607-61 7 53. Yates J , Warren N, Sudgen B (1985) Stable replication of plasmids derived from Epstein-Barr virus in various mammalian cells. Nature 313:812-815