The Expression of the Imprinted Genes H19 and IGF-2 in Choriocarcinoma Cell Lines

The expression of the imprinted genes H19 and IGF-2 in choriocarcinoma cell lines. Is H19 a tumor suppressor gene?

O Lustig-Yariv1, E Schulze2, D Komitowski3, V Erdmann4, T Schneider4, N de Groot1 and A Hochberg1

1The Department of Biological Chemistry, The Silberman Institute of Life Sciences, The Hebrew University of Jerusalem, Jerusalem 91904, Israel; 2The Zoologisches Institut-Entwicklungs biologie, Gottingen University, Gottingen, Germany; 3The Department of Histodiagnostics & Pathomorphology, DKFZ, Heidelberg, Germany; 4The Department of Chemistry, Institute for Biochemistry, Freie Universitat, Berlin, Germany

H19 is a paternally imprinted gene with unknown nearly all precursors of mRNAs). It is bound in 28S function. It is located in close proximity to the cytoplasmic RNPs and not to ribosomes (Brannan et maternally imprinted IGF-2 gene on chromosome al., 1990). Its function is enigmatic. Genetic analysis of 11p15.5. In this study no consistent relationship between the 11p15.5 region revealed a frequent correlation the expression of these two genes in clones derived from between the loss of the maternally inherited 11p15.5 JEG-3 and JAr cell lines could be detected. Nor could a region and many cases of Wilm's tumor (Junien, 1992) consistent relationship be detected between the expres- and some embryonal rhabdomyosarcomas (Scrable et sion levels of these two genes and between certain al., 1990). These correlations and other experimental characteristic tumorigenic properties of these clones. We ®ndings can be explained by assuming the existence of included in this study clones, expressing low H19 levels, a paternally imprinted (maternally expressed) tumor which after transfection with an H19 expression suppressor gene in this chromosomal area. In order to construct highly expressed the H19 gene. In tumors, provide evidence for the proposal that H19 functions formed by the injection of cells of JAr or JEG-3 clones as a tumor suppressor gene, Hao et al. (1993) into nude mice, the H19 expression was high and introduced a construct expressing the H19 gene under irrelevant to the expression level in the cells before the the control of the metallothionein promoter into cells injection. The same phenomenon was found for IGF-2 derived from a kidney tumor. Under conditions of high expression during tumorigenesis caused by cells of H19 expression, the cells, which showed prominent dierent JEG-3 clones and in some but not all JAr morphological changes, grew at a slower rate, had a derived clones. Both H19 and IGF-2 are biallelicly much lower anchorage independent growth rate in soft expressed in all the JAr and JEG-3 clones. In summary, agar, and did not develop tumors when injected into our observations point to the conclusion that H19 is not nude mice as did cells prior to their transfection with a tumor suppressor gene. However, its high expression in the H19 expression construct. However, recently, all the tumors formed after injection of cells of the JAr ®ndings have been reported, which are contradictory and JEG-3 clones, leaves its role, if any, in choriocarci- to the proposed tumor suppressor function for the H19 nogenesis an open question. gene. H19, which is not expressed in normal adult tissues, is expressed in several human cancers such as Keywords: H19; IGF2; choriocarcinoma cell lines choriocarcinoma and placental site trophoblastic tumor (Ariel et al., 1994), bladder carcinoma (Cooper et al., 1996; Elkin et al., 1995), lung cancer (Kondo et al., 1995), breast adeno-carcinoma (Dugimont et al., 1995). Introduction Cervical carcinoma (Douc-Rasy et al., 1996) and esophageal cancer (Hibi et al., 1996). Moreover, in H19 and IGF-2 belong to the family of imprinted tumors formed by the injection of cells of a genes whose expression depends upon their parental choriocarcinoma derived cell line, JEG-3, into nude origin. The two genes are located in close physical mice, the level of H19 RNA was 3 ± 5 times higher than proximity both in human (11p15.5) and mice (distal in the cells before injection (Rachmilewitz et al., 1995). chromosome 7) (Zemel et al., 1992) and are Also, cells of human bladder carcinoma derived cell reciprocally imprinted. The H19 gene is expressed lines which do not express H19 (or express it at a very from the maternally inherited allele and the IGF-2 low level) form tumors in nude mice in which H19 from the paternally derived allele (Rainier et al., 1993). RNA can easily be detected (Elkin et al., 1995). The H19 gene is highly expressed in many embryonic However, all these observations leaves the question tissues but not expressed in neural tissues (Goshen et of the role of H19 in tumorigenesis open. al., 1993; Lustig et al., 1994). H19 expression is down The IGF-2 gene is highly expressed in the embryo regulated postnatally in nearly all tissues. and in some fetal tissues. The tissue-speci®c expression No protein product is known to be produced by the pattern of IGF-2 is very similar to that of the H19 gene H19 gene, nevertheless its primary transcript undergoes (Rappolee et al., 1992; Goshen et al., 1993) and, its processing (capping, polyadenylation and splicing like expression is also postnatally depressed in many adult tissues (Rechler and Nissley, 1990). Prominent excep- Correspondence: N de Groot tions are the adult liver and the choroid plexus which Received 8 November 1996; revised 24 March 1997; accepted 24 continue to express IGF-2 at relatively high levels March 1997 postnatally. It has to be mentioned here that the IFG-2 H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 170 gene can be transcribed from four dierent promoters, and IGF-2 expression, we followed IGF-2 expression three of which are active during fetal life and in cancer in the same JEG-3 and JAr clones. As reported before, tissues and are imprinted (P2, P3, P4). The fourth (P1)) no expression of IGF-2 was detected in cells of the is not imprinted, and is responsible for IGF-2 original JAr cell line (de Groot et al., 1994). Similarly expression in the adult liver and in the choroid plexus in cells of the JAr derived clones IGF-2 expression was (Holthuizen et al., 1993; Robertson, 1995; Ekstrom et not detected. The sole exception was clone JAr-C4 al., 1995). which showed a very low level of expression (data not The relationship between the regulation of IGF-2 shown). It is very interesting to note that no consistent and H19 is a complex one. It has been proposed that in mice H19 and IGF-2 expression are under the control of a common enhancer, which is *3 kb downstream a from the H19 coding region (Sasaki et al., 1992; Ferguson-Smith et al., 1993; Batholomei et al., 1993). JEG-3 C1 C2 C3 C4 C5 C6 This enhancer, or protein(s), bound to the enhancer region, can activate the H19 promoter on the maternally inherited gene or the IGF-2 promoters on the paternally inherited allele because paternally 28S — inherited H19 gene has a more fully methylated promoter which is unable to interact with the H19

downstream enhancer. However, this proposed me- 18S — ¨ H19 chanism can only partly explain the regulation of the H19 and IGF-2 expression. The presence of gene and tissue-speci®c transcription factors is decisive for the actual expression levels of the two genes, and therefore b loss of imprinting of either the H19 or the IGF-2 gene is not necessarily accompanied by an increase in the gene's expression. The relationship between H19 and IGF-2 expression in human diers from that in mouse 28S — (Adam et al., 1996). It has been claimed that in the human both genes can be expressed from the same chromosome (Eversole-Cire et al., 1995; Jinno et al., 18S — 1995). In order to establish more clearly the relationship between the level of H19 and IGF-2 expression and tumorigenicity, we purposely compared the tumori- c genic properties of choriocarcinoma cell line derived clones which express H19 and IGF-2 at dierent levels. However, since we were aware that those clones were still heterogeneous in many other properties, we 28S — introduced an H19 expressing construct into some of the very low H19-expressing clones and looked for phenotypic dierences related exclusively to the change 18S — in H19 expression.

Figure 1 Expression of H19 and IGF-2 in JEG-3 derived clones. Results Autoradiogram of Northern blots containing RNA isolated from the original JEG-3 cells and their derived clones (C1 ± C6). The Individual clones were isolated from JAr and JEG-3 positions of 28S and 18S ribosomal RNAs are indicated. (a) original cell cultures. Special care was taken during the Expression of H19 (exposure time 3 h). (b) Expression of IGF-2. (exposure time 72 h). (c) Relative amounts of RNA loaded in cloning procedure in order to ensure that each clone is each lane. Visualized by methylene blue staining of the blot derived from one cell only. (Sambrook et al., 1989) In Figure 1a we show H19 expression in several of the JEG-3 clones. H19 RNA appears as 2.3 kb band corresponding to the sum of the ®ve exons of the H19 gene (Brannan et al., 1990). In the majority of the blots JAr done by us and others a second band of approximately C1 C2 C3 C4 C5 C6 C7 4 kb is visible. The clones C1, C3 and C6 express H19 approximately at the same level as the original culture 28S —

from which they were derived, but other clones showed very dierent levels of H19 expression, such as clones ¨ H19 C2 and C5, in which the H19 level was de®nitely lower 18S — and clone C4 which expresses H19 at a signi®cantly higher level. A similar phenomenon was also revealed in clones derived from JAr cell lines (Figure 2). (i.e. in Figure 2 Expression of H19 in JAr derived clones. Autoradio- JAr-C5 and JAr-C7 high expression, in JAr-C4 low gram of Northern blots containing RNA isolated from JAr cells expression). In order to study the relation between H19 and their derived clones (C1 ± C7). Exposure time was 2 h H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 171 correlation was found between the levels of H19 and (Figure 1a). Also no correlation was obvious between IGF-2 expression in the dierent JEG-3 clones (Figure the level of IGF-2 expression and the growth rate of 1b). For example, high H19 and IGF-2 expression were JEG-3 clones. All the JAr clones that express high H19 noted in clone JEG-C1, while high H19 expression and grew at the same rate as the original culture (Figure absence of IGF-2 expression was found in clone JEG- 3B) but one clone which expresses H19 poorly, JAr-C4, C4. In contrast, low H19 and IGF-2 expression were showed a much slower growth rate. found in clone JEG-C2 and low H19 expression and To assess a possible correlation between H19 very high IGF-2 expression was found in clone JEG- expression and cell transformation phenotype, the C5. These results may indicate that the H19 RNA has clones were analysed for growth in soft agar. Cells no direct in¯uence on the IGF-2 expression level in were seeded in parallel in liquid medium culture and in choriocarcinoma derived cells and IGF-2 expression medium containing soft agar. The number of colonies very likely does not determine the level of the H19 in soft agar was determined and normalized to the transcript. frequency of colony formation in liquid medium. The original JAr culture produces larger anchorage- independent colonies and at a much greater frequency Growth rates and anchorage independent growth than the original JEG-3 cells (Rachmilewitz et al., High growth rates and anchorage independent growth 1995). However, all the clones derived from the potential in soft agar of cells are frequently considered original JAr and JEG-3 cultures produced anchorage- as characteristic properties of tumor cells. independent colonies at frequencies similar to that of As can be seen in Figure 3A there is no obvious the original cultures despite greatly dierent levels of correlation between H19 expression and growth rate of H19 expression. The only exception is JAr-C4 which the JEG-3 clones. For example JEG-C5 expresses less showed a signi®cantly lower anchorage-independent H19 than JEG-C1 and it is growing faster, but the growth than the other JAr clones. opposite situation exists between JEG-C2 and JEG-C4 Allelic mode of expression of H19 and IGF-2 In many cases of cancers, including choriocarcinoma (Ariel et al., 1993) and bladder carcinoma (Elkin et al., 1995) biallelic expression of H19 was detected. JAr and JEG-3 cells were analysed for their allelic mode of H19 expression. In order to inquire about the possible relationship between the level and the allelic mode of H19 expression, we studied 14 clones of JAr cells and 13 clones of JEG-3 cells. All the clones showed biallelic expession of H19 (Figure 4a), notwithstanding that some of the clones expressed H19 at a high level and other clones hardly express H19 at all (Figures 1 and 2). The JEG-3 clones were also tested for IGF-2 allelic mode of expression and all of them were found to express it biallelicly (Figure 4b).

a b 1 2 1 2

— 653bp

— 235bp

Figure 4 Allelic mode of IGF-2 and H19 expression in JAr and JEG-3 cell lines and their derived clones. (a) H19/RsaI site in JAr and JEG-3 cells. (b) IGF-2/ApaI site in JEG-3 cells. Lanes 1 ± RT ± PCR. Lanes 2 ± restriction analysis. Identical results have been obtained with all the derived clones examined. The relevant molecular weight markers are indicated (Boehringer Mannheim No. VI). The appearance of two bands in lanes 2 proves the biallelic expression of the H19 (Zhang and Tycko, 1992) and IGF-2 genes (Tadokoro et al., 1991). The appearance of the two Figure 3 Growth rates of JEG-3, and JAr cell lines and their bands in lanes 2 was absolutely dependent on the RT steps. No derived clones. (A) JEG-3 and its derived clones. (B) JAr and its products at all appeared on the gels when the RT reaction was derived clones. The assay was performed as described in Materials omitted, indicating that the RNA preparations used were not and methods contaminated with genomic DNA H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 172 Transfection of a non-H19 expressing clone with an H19 the tumors derived from JEG-3 clones, especially in the tumors derived from JEG-C5 clone which showed a expression vector dramatic increase (Figure 6, lanes 8 and 9). In order to investigate the possible relationship between We found approximately the same level of H19 the tumorigenic phenotype and the levels of the H19 RNA in the tumors formed by the dierent JEG-3 gene product, we transfected clones that did not express clones notwithstanding that the clones in vitro showed H19 with an expression vector containing the H19 gene a wide range of H19 expression level. H19 RNA level under the control of the metallothionein promoter. is high in JEG-C4 clone and did not show a signi®cant Transfected and non-transfected cells of the clone JAr- C4 were cultivated in the presence of increasing

concentrations of ZnSO4 (Figure 5, lanes 2 ± 9). As can be seen, the zinc ions activated the metallothionein a promoter which leads to zinc dependent H19 expression 1 2 3 4 5 6 7 8 9 in the transfected cells. The Zn2+ ion concentration in the culture medium is sucient to activate the metallothionein promoter to some extent (lanes 6 and 7). However in the transfected cells a very signi®cant 28S — increase in the H19 RNA level was observed at 50 and 100 mM Zn2+ ion concentration but no increase in the H19 RNA level could be detected in the original JAr 18S — culture (lanes 8 and 9). The same results were obtained with JEG-3 clones C2 and C5 (data not shown). We asked whether any phenotypic change can be related to the H19 expression by comparing the growth rate and anchorage-independent growth in soft agar of JAr-C4, b the clone which does not express H19, versus JAr-C4- 1 2 3 4 5 6 7 8 9 H19 in the presence of 100 mM Zn2+ ions, the same clone expressing the H19 gene from the metallothionein promoter. However, neither a dierence in the growth rate nor in the anchorage-independent growth were detected between the two clones. 28S —

Tumorogenicity potential of non-H19 and H19 expressing clones in athymic nude mice 18S — The following experiments were performed, in order to further evaluate the possible correlation between tumorigenesis and H19 gene expression. Four H19 expressing clones (JAr-C5, JAr-C7, JEG-C1 and JEG- c C4), three H19 low expressing clones (JAr-C4, JEG-C5 and JEG-C2) and their pMEP4-H19 transfectants were injected subcutaneously into athymic nude mice. In order to further stimulate the metallothionein promoter

we added 25 mM ZnSO4 to the mice drinking water. In order to ascertain that the zinc by itself has no eect on the mice and on the tumorogenicity of the clones we injected all the clones and transfectants both to mice with Zn2+ and without Zn2+ in their drinking water. No signi®cant dierences were observed in the tumorigenic potentials of the dierent clones and their corresponding transfectants as measured by the time interval between the injection and the time the tumor became visible (16 ± 18 days) and by the weight of the tumor after 25 days. The weight of all the tumors was Figure 5 Expression of H19 in a JAr-C4 clone before and after in the 0.25 ± 0.30 gram range. transfection with an H19 expression construct, under the control of the metallothionein promoter. (a) Autoradiogram of Northern blots. The clones were incubated in a usual medium for 24 h, 2+ Expression of H19 and IGF-2 in subcutaneous tumors in afterwards Zn was added as indicated and the cells incubated for another 48 h. Exposure time was 3 h. (b) Relative amounts of athymic nude mice RNA loaded in each lane. Visualized by methylene blue staining of the blot. (c) Histogram showing the relative abundance of H19 We measured the H19 and IGF-2 expression in the RNA. The intensity of the H19 hybridization signal shown in a cells before and after the tumor formation. was measured using a bio-imager analyzer and was normalized to As was shown previously (Rachmilewitz et al., 1995), the 28S ribosomal fraction in each lane. Lane 1 ± original JAr 3 ± 5-fold increase in the level of H19 RNA is seen in the culture. Lanes 2 ± 5 JAr-C4 and lanes 6 ± 9 JAr-C4 transfected tumors derived from JEG-3 cells as compared to the level with PMEP4-H19. Lanes 2 and 6 ± normal medium +10% FCS. Lanes 3 and 7 ± normal medium +2% FCS. Lanes 4 and 8 ± in the cells of the original culture (Figure 6, lanes 1 ± 3). normal medium +2% FCS + 50 mgZn2+. Lanes 5 and 9 ± The same phenomenon of increase was also shown in all normal medium +2% FCS + 100 mgZn2+ H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 173 increase in the tumors. Also the same RNA level was JEG-3 and JEG-3 clones. Its expression rose in tumors measured in tumors derived from JEG-C5 transfected derived from JEG-3 cells of the original cell line and with an H19 expression construct which had a high from clones which expressed IGF-2 in low amounts level of H19 RNA. such as the JEG-3 clones C1 and C4 (Figure 8, lanes The same phenomenon was observed in tumors 1 ± 6), and no change in its expression level was shown derived from JAr cells and their derived clones. The in tumors that arose from JEG-C5 which highly abundance of H19 transcripts in tumors induced by expressed the gene (Figure 8, lanes 7 ± 8). JAr cells was as high as in the original JAr cells (Figure Dierent results were observed concerning the 7, lanes 1 ± 3), and their clones (example Figure 7, lanes expression of IGF-2 in tumors derived from JAr cells 4 and 5). JAr-C4 is the only JAr clone which expresses as can be seen in Figure 9, JAr cells and their derived H19 at a low level, formed tumors with an H19 RNA clones do not express IGF-2 (or express it at a very low abundance equal to the H19 RNA abundance in the level; lanes 1, 3, 5 and 7). Also the tumors derived from tumors induced by the other JAr clones (Figure 7, the cells of the original JAr cell culture did not express lanes 6 and 7). It was shown that 25 mM of ZnSO4 in IGF-2 (lane 2). Similarly, no expession could be the drinking water can activate the metallothionein detected in tumors induced by JAr-C7 cells (lane 8). promoter in transgenic mice (Palmiter et al., 1983). However, in tumors derived from two dierent JAr However, Zn2+ did not seem either to in¯uence the clones, IGF-2 RNA could be detected; a prominent growth or the H19 expression level in the tumors that expression was observed in tumors derived from JAr- arose from original and transfected cells (Figure 6, C4 (lane 4) and a lower expression was detected in JAr- lanes 2 and 3; Figure 7, lanes 2 and 3). C5 induced tumors (lane 6). A very interesting observation was that the IGF-2 gene showed a similar expression pattern as H19 in Discussion

1 2 3 4 5 6 7 8 9 10 11 12 13 Actually nothing is known about the biological function of the H19 transcript. H19, very likely, 28S — acts as an RNA, not only no-H19 protein could be

18S — 1 2 3 4 5 6 7 8

Figure 6 Expression of H19 in JEG-3 cells and derived clones, 28S — and in the tumor formed by these cells in nude mice. Autoradiogram of Northern blot, exposure time 3 h. Lane 1 ± JEG-3 original culture. Lane 2 ± tumors induced by JEG-3. Lane 2+ 3 ± tumor arising from JEG-3 in mice which had 25 mM Zn in their drinking water. Lane 4 ± JEG-C1. Lane 5 ± tumors induced 18S — by JEG-C1. Lane 6 ± JEG C4. Lane 7 ± tumors induced by JEG- C4. Lane 8 ± JEG-C5. Lane 9 - tumors induced by JEG-C5. Lane 10 - JEG-C5 after transfection with pMEP4-H19. Lane 11 ± JEG- C5-pMEP4-H19 supplemented with 100 mm of ZnSO4. Lane 12 ± tumors induced by JEG-C5-pMEP-H19 in mice. Lane 13 ± tumors Figure 8 Expression of IGF-2 in JEG-3 cells and derived clones, and in the tumor formed by these cells in nude mice. induced by JEG-C5-pMEP-H19 in mice which had 25 mM ZnSO4 in their drinking water Autoradiogram of Northern blot, exposure time 24 h. Lane 1 ± JEG-3 original culture. Lane 2 ± tumors induced by injection JEG- 3. Lane 3 ± JEG-C1. Lane 4 ± tumors induced by JEG-C1. Lane 5 ± JEG-C4. Lane 6 ± tumors induced by JEG-C4. Lane 7 ± JEG- 1 2 3 4 5 6 7 8 9 C5. Lane 8 ± tumors induced by JEG-C5

28S — 1 2 3 4 5 6 7 8

18S — 28S —

18S — Figure 7 Expression of H19 in JAr cells and derived clones and in the tumors formed by these cells in nude mice. Autoradiogram of Northern blot, exposure time 3 h. Lane 1 ± JAr original culture. Lane 2 ± tumors after injection of JAr. Lane 3 ± tumors induced by Figure 9 Expression of IGF-2 in JAr cells and derived clones, JAr in mice which had 25 mM ZnSO4 in their drinking water. and in the tumor formed by these cells in nude mice. Lane 4 ± JAr-C5. Lane 5 ± tumors induced by JAr-C5. Lane 6 ± Autoradiogram of Northern blot, exposure time 72 h. Lane 1 ± JAr-C4. Lane 7 ± tumors induced by JAr-C4. Lane 8 ± tumors JAr original culture. Lane 2 ± tumors induced by JAr. Lane 3 ± induced by JAr-C4 transfected with pMEP4-H19. Lane 9 ± tumors JAr-C4. Lane 4 ± tumors induced by JAr-C4. Lane 5 ± JAr-C5. induced by JAr-C4 transfected with pMEP4-H19 in mice which Lane 6 ± tumors induced by JAr-C5. Lane 7 ± JAr-C7. Lane 8 ± had 25 mM ZnSO4 in their drinking water tumors induced by JAr-C7 H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 174 detected but the mouse and human genes do not phenomenon is completely contradictory to the role of share an ORF with an amino acid sequence a tumor suppressor gene which has been assigned to homology. Our knowledge about the H19 RNA the H19 gene. does not provide us with any indication as to its Does the H19 gene take part in the process of possible biological role. Recently it was shown that tumorigenesis in cancers such as choriocarcinoma and certain mRNAs may have functions additional to bladder carcinomas? This question remains an open being a template for protein synthesis (Rastinejad et question. Alternative pathways may transform a cell of al., 1993). Whatever its actual function, the H19 gene a certain cell type, and H19 expression may be on one is highly expressed in embryonal and fetal tissues and of those pathways promoting tumorigenesis. The in several tumors, but is not or only marginally highest levels of the H19 gene product are found in expressed in normal adult tissues. Therefore, the tissues during early development and in several cancers. function of H19, if any, seems to be important in Till now we cannot distinguish between the possibility fetal and tumor development. In this work we have that H19 expression promotes carcinogenesis or that its tried to de®ne a relationship between the level of increased expression is a result of the carcinogenic H19 and IGF-2 expression and some tumorigenic process. properties of choriocarcinoma derived cells. However, The relationship between the expression levels of the we could not detect any consistent relationship H19 and the IGF-2 genes has been the subject of many between the H19 and IGF-2 RNA levels in these investigations. In the mouse either H19 or IGF-2 cells (which dier greatly among the clones (Figures expression seems to exclude the expression of the 1 and 2)) and some of their characteristic tumorigenic second gene from the same allele, and as a result, the properties, such as growth rate, anchorage-indepen- expression of the imprinted H19 gene on the dent growth in soft agar and their capacity to induce maternally inherited chromosome leaves the IGF-2 tumors in nude mice (Figure 3). This failure may be gene in an inactive state. However, the actual due to dierent reasons. The ®rst one is simply the expression level will be determined by the availability possibility that no such relationship exists. The of H19 and IGF-2 speci®c transcription factors. The possibility also exists that the experimental condi- IGF-2/H19 relationship in the human gene seems to be tions did not allow us to detect such a relationship, dierent from that in the mouse (Adam et al., 1996). assuming that it does exist. JAr and JEG-3 cell We did not detect a consistent relationship between the cultures are very heterogeneous populations, and the levels of H19 and IGF-2 expression in the JEG-3 clones we isolated from these cell lines may dier clones. Also we failed to detect a correlation between each from the other not only in their H19 and IGF-2 the level of IGF-2 expression and the tumorigenic expression, but also, simultaneously and indepen- properties of the cells as mentioned above for the H19 dently, in the expression level of other genes. All gene. these variables between the clones may make it We measured the IGF-2 mRNA levels in the dicult to relate any special dierence between the induced tumors. The level of IGF-2 mRNA in properties of the two clones to dierences in the tumors induced by cells of the JEG-3 cell line or by levels of the H19 and IGF-2 RNAs. JEG-3 derived clones was higher than in the cells Therefore, we transfected clones isolated from the from which they originated. This phenomenon is JAr and JEG-3 cell lines with low H19 expression very similar to that obtained with the H19 gene. levels with a H19 expression construct under the However, a somewhat dierent pattern of IGF-2 control of the metallothionein promoter. As a result, expression was observed in cells from the JAr cell we obtained a pair of JAr and a pair of JEG-3 clones, lines (see Figure 8). In one of the JAr clones and in one of each pair expressing H19 at a low level, and the original JAr culture we did not detect an the second one expressing H19 at a high level. All the increase in the IGF-2 RNA level as a result of other possible dierences between the cells of such a tumor formation. In two other clones we found a pair are due, directly or indirectly to the dierent H19 substantial increase in the IGF-2 RNA level in the RNA levels, except for unspeci®c eects caused by the tumors as compared to that in the cells before their presence of the plasmid only. However, no such injection into the nude mice. These results show that plasmid eects were detected. But again no signi®cant H19 and IGF-2 expression are regulated by dierent correlation was found between the level of H19 pathways, as indicated before. Moreover, one may expression and the tumorigenic properties mentioned conclude that IGF-2 is not involved in choriocarci- above. These ®ndings are however in line with our noma formation. observations made with the non-transfected clones We found biallelic expression of H19 and IGF-2 in some with relatively high H19 RNA levels. all the clones tested. Biallelic expression of both H19 On comparing the levels of H19 RNA in the cells of and IGF-2 was previously observed in testicular germ the tumors to those in the cells grown in culture before cell tumors (van Gurp et al., 1994; Verkerk et al., 1997) their injection, one particular phenomenon drew our H19 undergoes loss of imprinting in choriocarcinoma attention. Upon injection of cells of those clones with a in vivo, therefore the biallelic expression status is low H19 RNA level, tumors were formed with a higher preserved in cells derived from choriocarcinoma. H19 RNA level (including cells from the original JEG- The situation with the IGF-2 gene is somewhat more 3 cell line). However, during tumorigenesis induced by complicated because the IGF-2 gene has four cells of clones which had a relatively high level of H19 promoters and one of them, the P1 promoter, which RNA there was no increase in the H19 RNA level. is responsible for IGF-2 expression in the human adult Therefore, the H19 RNA level in all the tumors formed liver and in the choriod plexus is not imprinted. was nearly the same, regardless of the H19 RNA level Therefore we cannot be sure if the results described in in the cells before their injection into the mice. This Figure 4b are due to a loss of imprinting or to H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 175 promoter switching from an imprinted to the non- dish. Clones were picked from wells in which the imprinted P1 promoter. probability to ®nd at most one cell was one to ®ve. In summary we studied two sets of clones, one Clones were expanded and kept frozen. derived from the choriocarcinoma cell line JAr and the second from the JEG-3 cell line. Notwithstanding Isolation of RNA and DNA the great variance in the level of H19 expression Total cellular RNA was isolated from the various tissues by among the clones we did not detect a direct the guanidinium thiocyanate method (McDonald et al., relationship between the level of the H19 in the 1987) and from cells according to Chomczynski and Sacchi cells on the one hand and their growth rate, (1987). DNA was isolated as described (Ausubel et al., 1993). clonogeneity in soft agar and their tumorigeneity on the other hand. An important conclusion can be Cell growth assay drawn from our ®ndings in this investigation, the H19 is not a tumor suppressor gene. Other data 36104 cells were plated in triplicates in wells of 24 well suggest that a gene other than H19 on the short arm tissue culture dishes. Cells were assayed at 24 h intervals of chromosome 11 might function as a tumor over a period of 5 days. To assay cell number, the medium was removed, the cells were washed twice with Hank's suppressor gene (Koi et al., 1993) which may be buered salt solution (HBSS), 0.5 ml of distilled water was the recently identi®ed cyclin-dependent kinase inhi- added, and the cells were frozen. After thawing, aliquots of bitor p57kip2, that causes G1 arrest (Matsuoka et al., the lysate were taken for protein determination by the Bio- 1995). This gene is paternally imprinted and Rad protein assay dye reagent. maternally expressed (Kondo et al., 1996; Hatada et al., 1996; Matsuoka et al., 1996), but some leakage of the paternal allele has been found. Some lung cancers Anchorage-independence assay show selective loss of the maternal allele of this gene, To assess anchorage-independent growth, cells were without inactivating mutation of the retained allele suspended in soft agar (Difco) according to the procedure (Kondo et al., 1996). In a number of Wilms' tumors described (Freshney, 1990). Each cell type was tested in the level of p57kip2 expression was found to be duplicate for growth in liquid medium and growth in soft reduced (Hatada et al., 1996). agar. The number of colonies with dense centers was Of special interest is our ®nding that JAr-C4, the scored for each plate after 2 weeks. The number of colonies only JAr clone with low H19 expression, had a in liquid medium allowed correction for viability of the cells of each cell type. signi®cantly lower growth rate and lower ability to grow in soft agar. These properties did not change after the H19 level in those cells was raised as a result Tumorigenicity assay of their transfection with an H19 expressing construct. Cells were suspended in culture media without serum at Moreover, the injection of the JAr-C4 cells into nude approximately 56106 cells/ml. Each nude mouse received mice caused tumor formation as fast as the transfected two subcutaneous injections in the dorsal ¯ank region JAr-C4 cells. Also, the H19 level in those tumors was (0.2 ml; 16106 cells/site). Cells from each dierent clone as high as in all the other tumors. The results with this were injected into three mice at two sites (six injections of clone in particular highlights our conclusion that H19 each cell population). The mice were females, CD-1 is not a tumor suppressor gene. athymic nude (Nu/Nu) mice, 50 ± 55 days of age. The H19 is an oncofetal gene, a gene expressed in fetal time the tumors became visible was determined, but tumors tissues and in tumors derived from those tissues. continued to grow until the animals were sacri®ced. The However its role, if any, in the process of tumorigen- wet weights of the tumors were determined immediately esis remains an enigma. after dissection. Portions of the tumors were preserved at 7808C for RNA extraction.

Materials and methods Preparation of H19 transcription vector 800 bp from the human H19 gene 3' region were subcloned at EcoRI site into Bluescript II KS plasmid (Stratagene) Cell culture behind the T7 and T3 RNA polymerase binding sites. In The JAr and JEG-3 cell lines, purchased from the vitro RNA transcription with T7 RNA polymerase was American Type Culture Collection, Rockville, USA, were used to produce antisense H19 cRNA from linearized maintained in DMEM-F12 (1 : 1) medium containing 10% plasmid DNA. Sense H19 RNA, prepared with T3 fetal calf serum, 25 mM HEPES (pH 7.4), penicillin polymerase, was used for control. All the described (180 units/ml), streptomycin (100 mg/ml) and amphotericin hybridization experiments, using sense H19 RNA failed B(0.2mg/ml). 46104 cells/cm2 were plated in polystyrene to produce any signal. culture dishes (NUNC). Every 4 days the cells were trypsinized with 0.05% trypsin-EDTA solution (Beit Haemek) for 15 min and replated again at the same initial Preparation of probes for Northern blots density. 32P-labeled RNA transcripts were produced in vitro using the Amersham Kit and RNA polymerases from Boehringer (Mannheim). Linearized plasmid was prepared by digestion Cloning of cells with HindIII for the antisense probe according to the Cells were trypsinized with 0.05% trypsin-EDTA solution manufacturer's instructions. The fragments were separated for 15 min, vigorously mixed in a vortex, centrifuged and from unincorporated nucleotides by ethanolic precipitation. resuspended in the same initial medium. The cells were IGF-2 40 mer oligonucleotide probes were labeled by the counted and diluted in serial logarithmic dilutions. Each 5'-end labeling procedure, using [g-32P]ATP according to the dilution was plated in 36 wells of 96 well tissue culture instructions of the supplier (Oncogene Science). H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 176 Northern blotting Plasmids 5or10mg of each RNA sample was separated by 1% pMEP-4H19B which contains the H19 cDNA under the agarose-formaldehyde gel electrophoresis and transferred control of the metallothionein promoter was a kind gift to Hybond-N Nylon ®lters (Amersham, England). The from Prof. Benjamin Tycko, Columbia University, New blots were stained with methylene blue (Sambrook et al., York and described in Hao et al. (1993). 1989), in order to ascertain that equal amounts of RNA had been loaded in each lane. The blots were Transfections prehybridized at 608C in 50% formamide, 56SSPE, 56Denhardt solution, 0.1% SDS with 0.1 mg/ml herring Atotalof46105 cells in 30 mm dishes were transfected testes DNA followed by hybridization with speci®c with 14 mg of plasmid using the calcium phosphate cDNA probes. After hybridization with the IFG-2 precipitation method (Shen et al., 1982). The precipitates oligonucleotide probe the blots were washed in 26SSC; were added to the cells with 1 ml of the usual growth 0.1% SDS and twice in 0.16SSC; 0.1% SDS at 658C medium. After 16 h of incubation medium was changed to and exposed to AGFA Curix ®lm at 7808C(Oncogene fresh normal medium. In the case of stable transfections, Science procedure). Phosphoimaging (Bio-imaging Analy- cells were trypsinized and seeded in six well plates, and zer, NIH-image 1.58) was used for quantitative analysis 24 h later 125 ± 250 ml of Hygromycin (Calbiochem.) was of the hybridization signal. added to the media to initiate selection.

H19 and IGF-2 allelic mode of expression PCR conditions for each of the genes were as follows: 30 Acknowledgements or 35 cycles of ampli®cation of H19 was carried out as This work was supported by a grant from the US-Israel described by Zhang and Tycko (1992). 30 ± 40 cycles of Binational Science Foundation; by the Cooperation ampli®cation of IGF-2 were carried out as described Program in Cancer Research of the Deutsches Krebs- (Tadokoro et al., 1991). Aliquots were analysed for allelic forschungzentrum (DKFZ) Heidelberg, Germany and of mode of expression using polymorphic sites detected by Israel's Ministry of Science and the German-Israeli- RsaI in the H19 gene (Zhang and Tycko, 1992) and by Palestinian Authority Trilateral grant sponsored by the ApaI in the IGF-2 gene (Tadokoro et al., 1991). Deutsche Forschungs Gemeinschaft (DFG).

References

Adam GIR, Cui HM, Miller SJ, Flam F and Ohlsson R. HaoY,CrenshawT,MoultonT,NewcombEandTyckoB. (1996). Develop., 122, 839 ± 847. (1993). Nature, 365, 764 ± 767. Ariel I, Lustig O, Oyer CE, Elkin M, Gonik B, Biran H, Hatada I, Inazawa J, Abe T, Nakayama M, Kaneko Y, Jinno Rachmilewitz J, Goshen R, de Groot N and Hochberg A. Y, Niikawa N, Ohashi H, Fukushima Y, Lida K, Yurani (1994). Gyn. Oncol., 53, 212 ± 219. C, Takahashi S-I, Chiba Y, Ohishi S and Mukai T. (1996). AusubelFM,BrentR,KingstonRE,MooreDD,Seidman Hum. Molec. Genet., 5, 783 ± 788. JG, Smith JA and Struhl K. (1993). (eds). In: Current Hibi K, Nakamua H, Hirai A, Fiyikakale Y, Kasai Y, protocols in molecular biology, Vol 1 pp. 2.21-2.23. New Akiyama S, Ito K and Takagi H. (1996). Cancer Research, York: John Wiley & Sons. 56, 480 ± 482. Bartolomei MS, Webber AL, Brunkow ME and Tilghman Holthuizen PE, Cleutjens CBJM, Veenstra GJC, Van der Lee SM. (1993). Genes Dev., 7, 1663 ± 1673. FM, Koonen-Reenst AMCB and Sussenbach JS. (1993). Brannan CI, Dees EC, Ingram RS and Tilghman SM. (1990). Regul. Pept., 48, 77 ± 89. Mol. Cell. Biol., 10, 28 ± 36. Jinno Y, Ikeda Y, Yun K, Maw M, Masuzaki H, Fukuda H, Chomczynski P and Sacchi N. (1987). Anal. Biochem., 162, Inuzuka K, Fujishita A, Ohtani Y, Okimoto T, Ishimaru T 156 ± 159. and Niikawa N. (1995). Nature Genet., 10, 318 ± 324. Cooper MJ, Fisher M, Komitowski D, Shevelev A, Schulze Junien C. (1992). Curr. Opinion Genet. Dev., 2, 431 ± 438. E, Ariel I, Tykocinski M, Miron S, Ilan J, de Groot N and Koi M, Johnson LA, Kalikin LM, Little PFR, Nakamura Y Hochberg A. (1996). J. Urol., 155, 2120 ± 2127. and Feinberg AP. (1993). Science, 260, 361 ± 364. de Groot N, Rachmilewitz J, Ariel I, Goshen R, Lustig O and Kondo M, Suzuki H, Ueda R, Osada H, Takagi K and Hochberg A. (1994). Trophoblast Research, 8, 285 ± 302. Takahashi T. (1995). Oncogene, 10, 1193 ± 1198. Douc-Rasy S, Barrois M, Fogel S, Ahomadegle JC, Stehelin Kondo M, Matsuoka S, Uchida K, Osada H, Nagatake M, D, Coll J and Rio G. (1996). Oncogene, 12, 423 ± 430. Takagi K, Harper JW, Takahashi T and Elledge SJ. DugimontT,CurgyJ-J,WernetN,DelobelleA,RaesM-B, (1996). Oncogene, 12, 1365 ± 1368. Joubel A, Stehelin D and Coll J. (1995). Mol. Cell., 85, Lustig O, Ariel I, Ilan J, Lev-Lehman E, de Groot N and 117 ± 124. Hochberg A. (1994). Mol. Rep. Dev., 38, 239 ± 246. Elkin M, Shevelev A, Schulze E, Tykocinski M, Cooper M, Matsuoka S, Edward MN, Bai C, Parker S, Zhang P, Baldini Ariel I, Pode D, Kopf E, de Groot N and Hochberg A. A, Harper JW and Elledge SJ. (1995). Genes Develop., 9, (1995). FEBS Lett., 374, 57 ± 61. 650 ± 662. Ekstrom TJ, Cui H, Li X and Ohlsson R. (1995). Matsuoka S, Thompson JS, Edward MN, Barletta JM, Development, 121, 309 ± 316. Grundy P, Kalikin LM, Harper JW, Elledge SJ and Eversole-Cire P, Ferguson-Smith AC, Surani MA and Jones Feinberg AP. (1996). Proc.Natl.Acad.Sci.USA,93, PA. (1995). Cell Growth Di., 6, 337 ± 345. 3026 ± 3030. Ferguson-Smith AC, Sasaki H, Cattanach BM and Surani McDonald RJ, Swift GH, Przybyla AE and Chirgwin JM. MA. (1993). Nature, 362, 751 ± 755. (1987). In: Berger SL and Kimmel AR. (eds). Methods in Freshney RI. (1990). Culture of animal cells ± A manual of Enzymology, New York: Acad Press, 152, 219 ± 227. basic technique (second ed.), pp. 142 ± 144. New York: Palmiter RD, Norstedt G, Gelinas RE, Hammer RE and Wiley Liss. Brinstet RL. (1983). Science, 222, 809 ± 814. Goshen R, Rachmilewitz J, Schneider T, de-Groot N, Ariel I, Palti Z and Hochberg A. (1993). Mol. Reprod. Dev., 34, 374 ± 379. H19 and IGF-2, in choriocarcinoma cell lines O Lustig-Yariv et al 177 Rachmilewitz J, Elkin M, Rosensaft J, Gelman-Kohan Z, Scrable HJ, Sapienza C and Cavenee WK. (1990). Advances Ariel I, Lustig O, Schneider T, Goshen R, Biran H, de in Cancer Research, 54, 25 ± 62. Groot N and Hochberg A. (1995). Oncogene, 11, 863 ± 870. Shen YM, Hirschhorn RR, Mercer WE, Surmacz E, Tsutsui Rainier S, Johnson LA, Dobry CJ, Ping AJ, Grundy PE and Y, Soprano KJ and Baserga R. (1982). Mol. Cell. Biol., 2, Feinberg AP. (1993). Nature, 362, 747 ± 749. 1145 ± 1154. Rappolee DA, Sturm KS, Behrendtsen O, Schultz GA, Tadokoro K, Fujii H, Inoue T and Yamada M. (1991). Nucl. Pedersen RA and Werb Z. (1992). Genes Dev., 6, 939 ± 952. Acids Res., 19, 6967. Rastinejad F, Conboy M, Rando T and Blau HM. (1993). Van Gurp RJLM, Oosterhuis JW, Kalscheuer V, Mariman Cell, 75, 1107 ± 1117. ECM and Looijenga LHJ. (1994). J. Natl. Cancer Inst., 86, Rechler MM and Nissley SP. (1990). In: Sporn MB and 1070 ± 1075. Roberts AB (eds). Peptide growth factors and their Verkerk AJMH, Ariel I, Dekker MC, Schneider T, van Gurp receptors Vol. I, pp. 263 ± 367. New York: Springer Verlag. RJHLM, de Groot N, Gillis AJM, Oosterhuis JW, Robertson EJ. (1995). Developmental Biol., 6, 293 ± 299. Hochberg A and Looijenga LHJ. (1997). Oncogene, 14, Sambrook J, Fritsch EF and Maniatus T. (1989). In: 95 ± 107. Molecular Cloning. A laboratory manual (2nd ed.). USA: Zemel S, Bartholomei MS and Tilghman SM. (1992). Nature Cold Spring Harbor Laboratory Press. Genet., 2, 61 ± 65. Sasaki H, Jones PA, Chaillet JR, Ferguson-Smith AC, Zhang Y and Tycko B. (1992). Nature Genet., 1, 40 ± 44. Barton SC, Reik W and Surani MA. (1992). Genes Dev., 6, 1843 ± 1856.