Eur. J . Biochem. 128, 193- 197 (1982) cc FEBS 1982

Abundant Messenger RNA Sequences Common to Transplanted Mouse Tumours of Different Tissue Origins

Gilles MERCIER, Martine GUILLIER, and Jacques HAREL

Group de Recherche de Carcinologie Expkrimentale (Groupe 8 du Centre National de la Recherche Scientifique) et Institut Gustave-Roussy, Villejuif

(Received April 26/July 15, 1982)

Radioactive cDNA, cell-free-synthesised on poly(A)-rich mRNAs from mouse FLS-ascites tumour cells, were separated into two fractions corresponding to abundant mRNA species and to rare mRNA species respectively, the former being nearly 100-times more abundant per cell than the latter. Aliquots of each fraction were hybridized to a large excess of RNA of different origins: three ascites tumours (the FLS, Krebs 2 and Ehrlich tumours) all transplanted in animals; one established line of Friend erythroleukemia cells serially passaged in culture; adult organs and whole mouse embryos. It was shown that all or nearly all the mRNA species, in particular the abundant ones, were present and similarly distributed in the three transplanted tumour strains studied, despite their different tissue origins. The great majority of abundant mRNA species were also present in Friend erythroleukemia cells but about two-times less frequent relative to transplanted tumours. Much greater differences were found as compared with non-malignant cells. In particular, although the majority of the abundant mRNA species were also found in all normal tissues studied, they were on the average 7- &times less abundant in adult liver or in 10-days-old embryos, and 10-times less abundant in adult brain, than in tumour cells. Finally about 10 of the abundant RNA species common to all transplanted tumours appeared to be absent in normal cells, whatever their growth rates or stage of differentiation. The possible implications of these findings are discussed.

In a recent article we reported molecular hybridization data showing an extensive homology between poly(A)-rich RNA sequences from different strains of mouse malignant cells established for many years [I]. Using as standard probes highly labelled cDNAs, cell-free-synthesised on cytoplasmic poly(A)- rich RNAs from FLS-ascites tumour cells, very few differences were found in the total number and frequency distribution of RNA species in three ascites-tumour strains of various tissue origins, continuously transplanted in mice. The differences were slightly more pronounced as compared with RNAs from tissue-cultured Friend erythroleukemia cells and were much greater as compared with RNAs from non-growing adult tissues, terminally differentiated, as well as RNAs from em- bryonic tissues in the process of active growth. In particular some of the abundant poly(A)-rich RNAs common to all the malignant strains studied appeared to be either absent or much less abundant in 10-days-old embryos or in brain, liver and kidneys of adult animals. The present studies were undertaken in order to determine the precise frequency distribution of homologous and non-homologous poly(A)-rich RNAs in malignant cells and normal tissues.

MATERIALS AND METHODS

Cells and Tissues

The cell lines and tissues utilized in the present studies were the same as previously described [I]. Therefore only a few essential points will be mentioned again. The FLS [2], Ehrlich [3] and Krebs [4] ascites tumours, originally derived from solid tumours of different tissue origins, were continuously trans- planted in young adult Swiss mice and collected at the end of the exponential growth period. The line 745 of Friend erythro-

leukemia cells [5] was serially passaged in culture flasks and cells were collected after 3 days. This cell line has retained the capacity to undergo erythrocytic differentiation under the effects of various agents. The brain and liver were immediately removed from sacrified Swiss mice 3 - Cmonths old and total embryos were removed by Caesarean operation, 10 - 1 1 days after the onset of pregnancy in Swiss mice 2 - 3-months old.

Nucleic Acids

The procedures utilized to prepare and characterize nuclear DNA, total cellular RNAs or cytoplasmic RNAs, total cellular or cytoplasmic poly(A)-rich RNAs and cell-free-syn- the~ised[~H]cDNA copies from FLS-cell poly(A)-rich RNAs, were described [I]. I should simply be recalled that the percentages of poly(A)-rich RNAs purified by three successive runs of oligo(dT)-cellulose chromatography were the same as previously reported and the mean size of purified cytoplasmic poly(A)-rich RNAs remained equivalent to 1800- 2000 bases in all cells and tissues studied. Also the final amounts of [3H]cDNAs synthesised corresponded to 8 - 10 %, of the quan- tity of RNA templates, the peak fractions of cDNA consisted of chains 800 - 1000-nucleotides long and their specific radioac- tivity was approximately 2 x 10' counts min-' pg-'.

Fractionation of the [3H]cDNA Probes into Abundant and Rare Sequences

One sample of 5 x lo6 counts/min, corresponding to 0.25 pg of the FLS-cell [3H]cDNAs, was annealed to 100-fold excess of FLS-cell cytoplasmic poly(A)-rich RNAs and allowed to hybridize under previously specified conditions [ 11. After attaining rot values of 0.7-0.8 mol RNA * 1-' . s the sample

1 94

was subjected to hydroxyapatite chromatography as described [6]. The fractions eluting as single-stranded molecules and double-stranded molecules were separately pooled. Each frac- tion, which contained about 50 of the input [’HH]cDNA (as shown by measuring the acid-precipitable radioactivity on duplicate aliquots), was adjusted to 0.1 M NaOH and in- cubated overnight at 37 “C in order to obtain complete degradation of RNA. Thereafter the solution was neutralized and passed through a Sephadex G-50 column. The void volume was adjusted to 0.3 M NaCl and [3H]cDNA was precipitated with 20 kg of carrier Escherichia coli DNA by adding 2.5 vol. ethanol. The precipitate was collected by high-speed centrif- ugation. The fractionated cDNAs were subjected to a second cycle of annealing to 25 pg of FLS-cell cytoplasmic poly(A)- rich RNA followed by hydroxyapatite chromatography. The final fractions, which were eluted as double-stranded and single-stranded molecules, were considered as ‘abundant’ and ‘rare’ sequences, respectively.

Aliquots of each fraction were hybridized to unlabelled nucleic acids in excess as indicated in the figure captions. The percentages hybridized were determined by use of S, nuclease as previously indicated [I].

RESULTS

Distribution of the ‘Abundant’ and ‘Rare’ Sequences in Cytoplasmic Poly( A ) - Rich RNAs of FLS-Tumour C e h

When aliquots of the ‘abundant’ and ‘rare’ FLS-cDNA fractions were hybridized to FLS-cell cytoplasmic poly(A)-rich RNA in great excess, reassociation kinetics similar to those shown in Fig. 1 were regularly obtained. The difference in rotl,2 values (which were about 0.048 and 4.0 mol . 1-l . s for the abundant and rare sequences respectively), indicated that the relative abundance of the former class of sequences is nearly 100-fold greater than that of the latter class. Following a rough estimate, based on hybridization kinetics as previously de- scribed [I], the abundant probes may correspond to 100- 110 distinct sequences of 1800 bases each on average. Also as shown in Fig. 1, the hybridization curves obtained with the recycled cDNA fractions were consistent with the intermediate curve obtained with unfractionated cDNAs. Furthermore the latter curve encompassed log rot values differing by more than 4 while each of the former curves encompassed log rot values differing by 3 or less. This clearly indicated that the recycled fractions were much more homogenous than the unfractionated probes, in terms of frequency distribution.

It should be noted that even after reaching saturation levels, no more than 15 ”/, of the recycled abundant probes could be reassociated with homologous poly(A)-rich RNAs instead of 85 - 87 ”/, for unfractionated cDNAs. A likely explanation for this decrease is that single-stranded segments are lost at each step of the recycling procedure.

Distribution of the Abundant und Rare Sequences in Total Poly(A)-Rich RNA from FLS-Tumour Cells

As shown in Fig. 2, when aliquots of recycled cDNA fractions were annealed to poly(A)-rich RNAs isolated from the total FLS-cell RNA, the respective hybridization curves of the abundant and rare sequences were almost identical to those obtained with cytoplasmic poly(A)-rich RNAs, except for a very slight shift to higher rot values. As previously discussed [I] this shift may be explained by the fact that total poly(A)-rich RNA contained intranuclear RNA sequences, which arc more complex than cytoplasmic poly(A)-rich RNAs.

80 -1 70 -

3 60 - I

u .- : 50- 2 n

Q Z

L O -

n 0 30-

20 - 10 -

’- I /

/

0 I I I ’

-3 -2 -1 0 1 2 3 log r, f

Fig. 1. Hybridization of the abundant andrars c D N A fiactioiis w i thpo /y (A I - rich RNA,from FLS-tumour cells. Aliquots of 2000 counts/min (equivalent to 0.1 ng) of the recycled abundant FLS-cell cDNA probes (r----V) and rare FLS-cell cDNA probes (v -v) were annealed to unlabclled cytoplasmic poly(A)-rich RNA. The concentrations of RNAs varied over 0.5- 10 Fg/ml for abundant cDNAs and 100- 330 pg/ml for rare cDNAs. At each indicated rot value (mol RNA . I - ’ . s) the percentage hybridized were determined by the use of S 1 nuclease, the control values (at most 4- 5 %) were substracted. The broken line represents the curve obtained with unfractionated cDNAs [l]

80

log ro t Fig. 2. Hybridization of abundant and r w s cDNA fkuctions kvith t o i d cyloplcumicpoly ( A )-rich RNAsfrom FLS-tumour cells. Same expcrimental conditions as those shown in Fig. 1 , except that the solid lines represent the hybridization curves obtained with total poly(A)-rich RNAs and abundant cDNA (r-r) or rare cDNA (V-V), compared with cytoplasmic poly(A)-rich RNAs (broken line)

Distribution of Ihe Abundant and Rare Sequences in Poly(A)-Rich RNAs of Neoplastic Cells

When aliquots of the recycled abundant and rare cDNA fractions were annealed to poly(A)-rich RNAs from the other

195

70

60 - s - 50 U 111 N .-

P LO n

2 30

20

x S

8

10

0

-3 -2 -1 0 1 -1 0 1 2 3 log r, t

Fig. 3. Hybridization of the abundant and rare cDNA fractions with cytopla.smicpoly(A)-rich RNAs from different neoplastic cells. Abundant (A) and rare (B) FLS-cell cDNAs were hybridized to RNA from Ehrlich-tumour cells (0-n), Krebs-2-tumour cells (-) or Friend erythroleukemia cells (m- 1). The broken lines represent the hybridization curves obtained with RNAs from FLS-tumour cells. Same experimental conditions as those indicated in the legend to Fig. 1

log r, t

Fig. 4. Hybridization of abundant and rare r D N A fracfions with totalpoly(A)-rich RNAs from normal tissues. Abundant (A) and rare (B) FLS-cell cDNA were hybridizcd to RNAs from brain (w), liver (M) or total mouse embryos (M). The broken lines represent the hybridization curves obtained with total RNAs from FLS-tumour cells. Same experimental conditions as those indicated in the legend to Fig. 1, except that the amounts of RNA from normal cells were 2.5 greater than those of RNA from FLS-tumour cells

ascites tumours or poly(A)-rich RNAs from Friend erythro- leukemia cells, hybridization kinetics similar to those shown in Fig. 3 were obtained. The reaction rates, as well as the maximum amounts of abundant sequences hybridized (Fig. 3A), were nearly the same with poly(A)-rich RNAs from FLS-tumour, Ehrlich-tumour and Krebs-Ztumour cells. If the marginal differences observed at apparent saturation levels are real, they could mean that one or two of the abundant poly(A)- rich RNA species accumulated into FLS-tumour cells was (were) absent or much less abundant in the other ascites tumours. The slight but reproducible differences observed with poly(A)-rich RNA from Friend erythroleukemia cells indicated that four or five of the abundant poly(A)-rich RNA species accumulated into FLS-tumour cells appear to be missing in Friend cells and that most of the other abundant sequences may be about two-times less abundant in Friend cells than in FLS cells.

The hybridizations kinetics of the rare cDNA sequences (Fig. 3B) were again almost the same with RNAs from the three ascites tumours studied and slightly different with RNAs from Friend cells. However, the small but reproducible differ- ences in the maximum percentage hybridized could signify that a few hundred of poly(A)-rich RNAs sequences of low abundance present in FLS-tumour cells, were not present in other neoplastic cells.

Distribution of the Abundant and Rare Sequences in Poly(A )- Rich Sequences from Non-malignant Cells

When aliquots of the abundant and rare cDNA fractions were annealed to poly(A)-rich RNAs from adult tissues or total embryos, hybridization curves similar to those shown in Fig. 4 were regularly obtained; this confirmed the results obtained with unfractionated cDNA [ 11. Here again, greater differences

196

/ I LO

- J - n 30 - 0 n x

4 z 0 0

= 20

10

0 0 1 2 3

log ro t Fig. 5. Hybridizalion of’ abundant and rare cDNA fractions with mouse genomic DNA. Aliquots of 2000 counts/min of the abundant (0, 0) and rare (0, m) FLS cDNAs were hybridized to 2 x 105-fold excess amounts of nuclear DNA (sonicated to 400 - 500-nucleotide-long fragments) from FLS-tumour cells (0, m) or from normal liver (0,o). At each indicated cot value (mol DNA 1- . s) the percentages hybridized wcrc dctcrmined by the use of S 1 nuclease as described [ l ]

were found in the reaction rates of abundant sequences with RNAs from brain than with RNAs from liver or RNAs from total embryos (Fig. 4A). Here again, superimposable hy- bridization curves were found with RNAs from liver and RNAs from embryos. However these new results demonstrate more clearly that the majority of poly(A)-rich RNA sequences which accumulate into ascites-tumour cells are unequally distributed and, on the whole, relatively much less abundant in normal cells than in tumour cells. In particular, even when hybridization assays were performed using excess amounts of normal-cell RNAs (2.5-times greater than those of FLS-cell RNAs), about 10 of the abundant sequences hybridizable to tumour-cell RNA, could not be reassociated with RNAs from adult or embryonic tissues.

As shown in Fig. 4B, the hybridization curves of the rare cDNA probes with poly(A)-rich RNAs from normal tissues also confirmed our previous data [l]. Here again the greatest difference was obtained with brain RNA and superimposable curves were found with liver RNA and total embryonic RNA.

Reassociation of the Abundant and Rare cDNA Sequences with Genomic DNA

We had previously found that a minor fraction of the unfractionated cDNAs could corresponded to repetitive gene sequences, while the rest of cDNAs correspond to non-repeated gene sequences [l]. In order to confirm this finding, each of the recycled cDNA fractions was hybridized to nuclear DNA in excess. On the whole, the abundant cDNA probes hybridized significantly faster than the rare cDNA probes (Fig. 5). This difference is not due to some dissimilarity in the probe dimensions because when analyzed as described [l] both cDNA fractions exhibited the same size distribution. This tends to confirm the finding that genes which encode the abundant RNA sequences are more repeated than those which encode the rare RNA sequences [7, 81. The fact that superimposable hybridization curves were obtained with DNA from FLS tumour and DNA from normal liver suggests that abundant poly(A)-rich RNA sequences did not represent transcripts of eenes snecificallv amdified in the tumour cells. However. the

methods utilized in the present studies are not sensitive enough to detect some limited gene amplification.

DISCUSSION

The present work both confirms and completes previous data obtained with the same cell strains and normal tissues 111, which were chosen for several reasons. Ascites tumours and tissue-cultured leukemia cells offer cell populations both more homogeneous and easier to control than those of solid tu- mours. Adult brain and liver contain differentiated cell popu- lations, essentially non-proliferating, endowed with a great variety of functions. Whole embryos, 10- 11-days old, repre- sent a mixture of cell lineages still in the process of active growth and engaged in various differentiation pathways. These previous data showed only very small differences in the hybridizations kinetics of total FLS-cell cDNAs, with RNA from the three ascites tumours studied. Slightly greater differ- ences were found with Friend cell RNA and much more pronounced differences were repeatedly found with RNA from adult organs or embryonic tissues. The effects of methodologi- cal factors, such as systematic changes in the size of poly(A)- rich RNA or differences in the purity of RNA preparations from one cell system to another, could be excluded [I]. Using total cellular RNAs or total cytoplasmic RNAs instead of purified poly(A)-rich RNA, the cDNA/RNA hybridization curves showed the same homologies in the case of neoplasic cells and the same differences as compared with non-malignant cells. This demonstrated that our data could not be attributed simply to systematic differences in the synthesis or processing of poly(A) chains associated to mRNA molecules. It was also shown that the maximum amounts of FLS cDNA hybridizable to RNA from other ascites tumour cells was only about 1 %less than those hybridized to homologous cell RNA, but constantly 13% less with RNA from all normal tissues either adult or embryonic. However, the use of unfractionated cDNA probes made it impossible to decide for certain whether the latter differences concerned a great number of RNA sequences of low abundancy rather than a small number of RNA sequences of high abundancy or both sorts of RNA sequences. The new results presented here turn the scale in favour of a high- abundancy class. They indicate that nearly all the rare poly(A)- rich RNA species detectable in FLS-tumour cells exist with very similar frequency distributions in the other neoplastic strains studied. They are also present in normal tissues regardless of their growth potential but with a lower frequency distribution, probably because they are diluted among more abundant RNA species which control specific cell functions. These data are in agreement with previous reports, also based on cDNA/RNA hybridizations, which favoured the conclusion that the total number of transcripting gene sequences is practically the same in neoplastic cells as in normal cells of the same lineages [9 - 131. However this conclusion cannot be accepted as well founded since it is now clear that cDNA/RNA hybridization methods can fail to detect some of the rare mRNA sequences which may correspond to numerous genes. In fact, former studies based on saturation hybridization of unique-copy genomic DNA [14], as well as recent ones based on more sophisticated recycling procedures, support an opposite conclusion in showing that in well defined cell systems malig- nancy may involve changes in the expression of several thousand different gene sequences [I5 - 171.

Nevertheless cDNA/RNA hybridization methods remain the most adequate for determining the overall freauencv

Y , I distribution of: mRNAs and analyzing abundant mRNA

197

species. In this respect the results obtained with the abundant FLS-cell cDNA fraction may be considered as reliable. These results demonstrate that all or nearly all abundant poly(A)-rich RNA species accumulated in FLS-tumour cells are also present with a very similar frequency distribution in Krebs and Ehrlich- tumour cells. If the very small difference in the maximum amounts hybridized is real, it could mean that out of 100- 110 RNA sequences of high abundancy present in FLS tumour cells, only one or two are absent in Ehrlich-tumour cells. The slightly greater difference reproducibly obtained with RNA from Friend cclls may signify that four or five of the abundant RNA species are absent in the latter cells. It should also be recalled that our previous results were in agreement with those of Affara et al. [IS] who compared mRNA populations in murine leukemia cells and in teratocarcinoma cells.

Finally our new results demonstrate that about 90 % of the most abundant RNA species accumulated in FLS-tumour cells are also present in adult or embryonic normal cells, but these common sequences are relatively 7 - 8-times less abundant in adult liver and whole embryos and 9- 10-times less abundant in adult brain than in FLS-tumour cells. On the other hand, 10 or 11 of the abundant RNA species appear to be totally absent in non-malignant cells. These differences cannot simply reflect the normal expression of mitosis-specific genes because super- imposable hybridization curves were obtained with the abun- dant cDNA fraction, annealed to RNA from adult liver where cell proliferation is negligible and to RNA from total embryos, still in the process of active growth. However, since the mitosis rates were probably not quite equivalent in all the main embryonic tissues, certain differences in the cDNA/RNA reassociation rates, but probably not the most pronounced difference, could be related to cell multiplication.

A possible role of the endogenous viruses integrated into the genome of all mouse cells [19, 201 in the formation of abundant mRNA species ought also to be envisaged. In fact there are many reports on the production of retroviruses, mainly of C and A types, in non-virus-induced mouse tumours [21 - 241 but this may be also true for fetal tissues [25,26]. In our laboratory, repeated assays failed to detect any consequent level of retrovirus production in the strain of FLS-tumour cells utilized to prepare the cDNA probes, but viral RNAs could be accumulated in the absence of virus maturation. In fact, following current studies (performed in Centre Hayem, Hopital Saint Louis Paris) the synthesis of type-A-virus RNA was found to occur in a variety of adult or fetal tissues from normal mice of three different strains (C. Escot-Theillet and R. Emanoel Ravicovitch, personal communication). Whatever the possible role of endogenous viruses, it is now known that abundant RNA species direct specific cell functions [7]. Therefore our data support the view that common specific functions characterize all the tumour cell strains studied,

despite their different tissue origin. It is probable that some of these functions concern the capacity to grow indefinitely in a liquid medium but it is tempting to assume that they are mainly involved in the maintenance of malignancy. New lines of evidence tend to demonstrate that constitutive expression of certain genomic sequences, the so-called onc genes, may play an essential role in carcinogenesis (see a recent review by G. Klein [27]). We postulate that at least some of the abundant RNA species selectively accumulated in tumour cells represent onc gene transcripts. Further investigations aimed at isolating and characterizing these abundant sequences are being undertaken.

REFERENCES

1. 2.

3. 4. 5.

6.

7. 8.

9. 10.

1 1 . 12. 13.

14.

15.

16. 17.

18.

19.

20. 21.

22. 23.

24. 25. 26. 27.

Mercier, G. & Harel, J. (1982) Eur. J . Biochem. 123, 407-414. Lacour, F., Lacour, J. , H a d , J. & Huppert, J. (1960) J. Narl Cancer

Klein, G. (1951) Exp. Cell Res. 2, 518-563. Klein, G. & Klein, E. (1951) Cancer Res. 11, 468-469. Friend, C. , Patuleia, H. C. & de Harven, E. (1966) Not1 Cancer 61s t .

Tapiero, H., Leibovitch, S. A, , Shaool, D., Monier, M. N. & Harel. J .

Hastie, N. D. & Bishop, J. 0. (1976) Cell, 9, 761 - 774. Alonso, A, Winter, H. & Krieg, L. (1981) Biochim. Biophys. Acta, 652,

Williams, J. G., Hoffman, R. & Penman, S. (1977) Cell, I / , 901 -907. Getz, M. J. , Reiman, H. M., Jr. Siegal. G. P.. Quinlan, T. J . , Proper J. ,

Rolton, H. A, , Birnie, G. D. & Paul, J . (1977) Cell Differ. 6, 25- 39. Supowit, S. C. & Rosen, J. M. (1980) Biochembtry, 19, 3452-3460. Sholla, C. A,, Pctropoulos, C. J . , Becker, F. F. & Fausto, H. (1981)

Grady, L. J . &Campbell, W. P. (1975) Science f Wush. DCI 254,356-

Groudine, M. & Weintraub, H. (1980) Proc. Null Acad. Sci. USA, 77,

Supowit. S . C. & Rosen, J . M. (1981) Cancer Res. 41, 3827-3834. Hanania, N., Shaool, D., Poncy, C. & Harel, J . (1981) Proc. Nafl Acad.

Affara,N. A., Jacquet, M.,Jakob, H., Jakob, F. &Gros, F. (1977) Cell,

Pincus, T. (1980) in Molecular Biology of RNA Tumour Viruses (Stephenson, J . R., ed.) pp. 77-130, Academic Press, New York.

Lueders, K. K. & Kuff, E. L. (1977) Cell, 12, 963-972. Gross, L. (1970) in Oncogenic Viruses, 2nd ed, pp. 281 - 601, Pergamon

Wive], N . A. & Smith, G . H . (1971) Int . J . Cancer 7, 167-175. Kakefuda, T., Roberts, E. & Suntzeff, V. (1970) Cuncrr Res. 30,101 1 -

Lueders, K. K., Segal, S. & Kuff, E. (1977) Cell, / I , 83-94. Chase, D. G. & Piko, L. (1973) J. Nut1 Cancer Inst. 51, 1971 - 1975. Strand, M.,August, J . T. &Jaenish, R. (1977) Virology, 76,886-890. Klein, G. (1981) Nature (Lond.) 294, 313-318.

I ~ I s ~ . 24, 301 - 327.

M o ~ u R ~ . 22, 505 - 520.

(1976) Nucleic Acids Res. 3, 953-963.

72-81.

Elder, P. K. & Moses, H. L. (1977) Cdl, I / , 909-921.

Biochemistry, 20, 381 5 - 3821.

358.

5351 - 5354.

Sci. USA, 78, 6504-6508.

12, 509-520.

Press.

1019.

G. Mercier, M. Guillier, and J. Harel, Groupe de Recherche de Carcinologie Experimentale, Institut Gustave-Roussy, 16bis Avenue Paul-Vaillant-Couturier, F-94800 Villejuif, Val-de-Marne, France