Antisense Downregulation of N-myc1 in Woodchuck Hepatoma Cells Reverses the Malignant Phenotype

Abstract

Cell line WH44KA is a highly malignant woodchuck hepatoma cell line. WH44KA cells contain a single woodchuck hepatitis virus (WHV) DNA integration in the 3′ untranslated region of exon 3 of the woodchuck N-myc1 gene. The highly rearranged WHV DNA contains WHV enhancers which activate the N-myc promoter, and a hybrid N-myc1–WHV mRNA is produced, which leads to a high steady-state level of N-myc1 protein. To investigate whether continuous N-myc1 expression is required to maintain the tumor phenotype, we knocked out N-myc expression using a WHV–N-myc1 antisense vector. We identified two WH44KA antisense cell lines, designated 4-5 and 4-11, in which steady-state N-myc1 protein levels were reduced by 95 and 80%, respectively. The growth rates of both antisense cell lines were reduced in comparison to those of wild-type and vector controls. The phenotype of 4-5 and 4-11 cells changed to a flattened appearance, and the cells exhibited contact inhibition. Colony-forming ability in soft agar was reduced by 92% for 4-5 cells and by 88% for 4-11 cells. Cell line 4-11 formed only small, slow-growing tumors in nude mice, consistent with a low level of N-myc1 remaining in the cells. In contrast, 4-5 cells, in which N-myc protein was reduced by greater than 95%, failed to form tumors in nude mice. The integrated WHV DNA contained the complete WHV X gene (WHx) and its promoter; however, we did not detect any WHx protein in the cells by using a sensitive assay. These data demonstrate that N-myc overexpression is required to maintain the malignant phenotype of WH44KA woodchuck hepatoma cells and provide a direct function for integrated WHV DNA in hepatocarcinogenesis.

Hepatocellular carcinoma (HCC) is one of the most common human cancers, and epidemiological studies have established a causal relationship between persistent infection with hepatitis B virus (HBV) and primary HCC (3). A closely related animal virus model for persistent HBV infection is infection with woodchuck hepatitis virus (WHV) (33). In fact, persistent WHV infection is associated with a nearly 100% incidence of HCC in WHV carrier woodchucks (26, 31, 32).

Hepatocarcinogenesis in WHV-carrier woodchucks is a multistep process, as with other cancers, in which precancerous lesions with altered gene expression profiles can be identified (2, 8, 35). The carcinogenesis process in WHV carriers is believed to be driven initially by a limited immune response which begins a cycle of cell death and regeneration in the liver (28, 33). In addition, the release of toxic oxygen radicals in the liver in response to inflammatory reactions during persistent infection can also increase DNA damage (10, 18, 19) and hepadnavirus DNA integration (25), which increases the risk of cancer (10, 33).

In many virus-associated cancers, viral DNA directly participates in oncogenesis by the process of insertional activation of proto-oncogenes (23, 29). The first proto-oncogene shown to be activated by viral DNA integration was the c-myc gene in bursal lymphomas induced by avian leukosis virus (12, 21). The discovery of clonally propagated integrations of WHV DNA in woodchuck hepatomas led to their cloning and a search for a common integration site for WHV DNA (24). Initial studies revealed highly rearranged WHV integrations which contained liver-specific enhancers of viral origin and the WHV X (WHx) gene (24). Additional cloning studies led to the discovery that WHV DNA was integrated within, or upstream of a unique N-myc retroposon (N-myc2) present in the woodchuck genome, which has in addition the normal N-myc gene (designated N-myc1) (6, 7, 34). The presence of this second functional N-myc gene in woodchucks may greatly increase their risk for hepatocarcinogenesis.

One study demonstrated that while N-myc activation was the most common event in WHV-associated HCC, c-myc activation occurred in those HCCs in which N-myc was not activated (11). Thus, one member of the myc gene family appears to be overexpressed in nearly 100% of woodchuck HCCs. The mechanism of N-myc activation is through a cryptic promoter in the N-myc2 retroposon by insertion of the strong WHV DNA liver-specific enhancer either in the 3′ untranslated region of N-myc2 exon 3 or upstream of the N-myc2 promoter (34). Interestingly, a second common WHV integration site, called WIN, has also been identified approximately 200 kb upstream from N-myc2 (5). Integration of WHV DNA at this distant upstream site is also associated with activation of N-myc2 transcription. The WIN chromosomal integration site does not contain a transcribed open reading frame but has matrix attachment sites for chromatin (5). Thus, WIN site integrations may activate N-myc2 transcription via alteration of chromatin structure.

While the case for myc genes as gatekeepers (15) for woodchuck HCC is very strong, many of the WHV integrations in N-myc also contain the WHx gene (24, 28). Transgenic mouse data suggest that both the WHx and the HBV X (HBx) proteins can act as tumor promoters (4, 16, 30). However, while WHx protein is present in all chronically infected livers, WHx protein was not detected in woodchuck HCCs that were nonpermissive for viral replication (4). Thus, continuous WHx expression may not be necessary for the maintenance of the malignant phenotype of woodchuck HCCs. Although HCC was induced in two HBx transgenic lines (13, 16), numerous other X gene transgenic lines did not develop HCC (references 4 and 30 and unpublished data).

In order to further study the role of N-myc and WHx in hepatocarcinogenesis in woodchucks, we identified a woodchuck hepatoma cell line, WH44KA (1), which contains a WHV DNA integration in the N-myc1 gene. While N-myc2 is a functional retroposon, N-myc1 is the normal N-myc gene in woodchucks, as determined by the presence of introns in the genomic sequence (6). Integrations of WHV DNA in N-myc1 have also been reported in primary woodchuck HCCs, albeit at a much lower frequency than that of integrations in N-myc2 (7).

In this report we describe the structure of the WHV DNA integration in WH44KA cells. The integration occurred in the 3′ untranslated region of N-myc1 exon 3 and leads to the production of a hybrid N-myc1–WHV mRNA. While the entire WHx open reading frame is present in the mRNA, no WHx protein is detectable in cultured cells. Using antisense vectors, we knocked out the accumulation of N-myc proteins and observed a loss of the malignant phenotype. Subclones of WH44KA cells which received ineffective antisense vectors maintained their malignant phenotype. These data support the conclusion that the continuous presence of N-myc proteins is required to maintain the malignant phenotype of WH44KA hepatoma cells and support the gatekeeper concept (15) of myc genes in WHV-associated hepatocarcinogenesis.

MATERIALS AND METHODS

Cells and culture conditions.

The WH44KA cell line was kindly provided by Kenji Abe and Toshio Shikata, Tokyo, Japan (1). WH44KA cells are maintained with Dulbecco modified Eagle medium (GIBCO BRL) containing 10% fetal bovine serum (GIBCO BRL) at 37°C in 5% CO₂. WLC-3 cells were a gift from T. Kitagawa. WLC-3 cells are a nonmalignant woodchuck hepatocyte progenitor cell line (17), maintained as described above. All tissue culture media contained 100 U of penicillin, 10 μg of streptomycin, and 250 ng of amphotericin B per ml.

Detection of N-myc1 protein.

Cells were lysed with sample buffer (62.5 mM Tris HCl [pH 6.8], 2% sodium dodecyl sulfate, 5% glycerol, 2% β-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 2 μg of leupeptin per ml, 2 μg of pepstatin A per ml). Cell lysates were passed through a 26-gauge needle three times to shear the genomic DNA. The total protein concentrations of samples were determined with a Coomassie dye-based kit (Bio-Rad). Lysates were stored at −70°C until use. Standard sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed through 10% resolving gels with a Hoefer minigel apparatus and 15 μg of total protein per lane. Proteins were transferred to polyvinylidene difluoride membranes (Millipore) with an LKB electrophoresis unit. A human N-myc monoclonal antibody (Nmyc ab-1 OP13; Oncogene Science) was used to detect woodchuck N-myc1 proteins (63 kDa) by Western blotting followed by enhanced chemiluminescence (ECL) detection (Amersham) according to the manufacturer’s instructions. The blots were exposed to XAR-5 films (Kodak) for 5 or 10 s at room temperature.

Nucleic acid analysis.

DNA and RNA extractions and Southern and Northern blotting were performed as previously described (8, 27, 35). The WHV DNA and N-myc probes were radiolabeled with a random-primer-labeling system (Amersham). The 3.3-kb cloned genome of WHV was used to make the virus-specific probe. The N-myc1-specific probe, including a section of the 3′ end of N-myc1 exon 1 plus the flanking intron, and the N-myc2 exon 3 probes were kindly provided by Genvieve Fourel and Marie Annick Buendia (6, 7). Riboprobes were synthesized and labeled by in vitro transcription (Promega) with SP6 RNA polymerase to synthesize antisense N-myc RNA from the pGEM N-myc vector. Sense-strand N-myc probe was made by primer extension with the Klenow fragment enzyme (Boehringer Mannheim) and the pGEMN-myc vector.

Cloning of the WHV integration in WH44KA cells.

An N-myc1 exon 3 coding region oligonucleotide primer, 5′AAAGCCTGTGAGTATGTCCAC3′ (6) (oligonucleotide a), and a WHV primer spanning WHV minus-strand nucleotides 1373 to 1393, 5′TCGGGAGGGGGAAAGCGAAAG3′ (oligonucleotide b), were used to amplify the left virus DNA-cell DNA junction of the WHV integration. A 3′ N-myc1 exon 3 untranslated region primer homologous to the antisense strand, nucleotides 1776 to 1804 (5′GTGGGTACCTAATGTCCCAGCTGAATCT3′) (oligonucleotide e), and a plus-strand primer spanning WHV nucleotides 241 to 261 (5′ATCCACCATATTGTCTCCTCC3′) (oligonucleotide c) were used to generate the right viral DNA-cellular DNA junction. Amplified junction fragments were subcloned into the pGEM-T vector (Promega) and sequenced with an automatic sequencer (Applied Biosystems, Inc.). The DNA sequence in the integrated WHV DNA was determined by sequence walking in both directions from the left and right junctions with WHV plus- and minus-strand primers and an automated sequencing system (Chanin Cancer Center) to sequence the entire integration in both directions. Since the internal WHV sequences were rearranged, care was taken to use primers which would give unambiguous sequences for individual reactions and some portions of the sequences were subcloned for sequencing.

Generation of N-myc antisense constructs.

Recombinant plasmid clones were constructed by subcloning PCR-amplified fragments from WH44KA genomic DNA with primer pair 1 and 2 for antisense vector 1 (see Fig. 3C), primer pair 1 and 3 for antisense vector 2, primer pair 1 and 4 for antisense vector 3, and primer pair 1 and 5 for antisense vector 4.

FIG. 3.

(A) Map of the integrated WHV DNA in WH44KA cells determined by sequencing the entire integrated WHV DNA. The selected WHV open reading frames are illustrated in order to identify the sequences within the WHV map. The WHV nucleotide numbers at each WHV rearrangement point in the integration are noted below the map in small numbers (9). The N-myc1 nucleotide numbers at the integration junctions are noted in large numbers below the map (6). Ex3, exon 3; open boxes, WHV; shaded boxes, N-myc1; En1, WHV enhancer 1; X, WHx gene; C, woodchuck hepatoma C gene; P/S, entire Pre S-S open reading frame; Pre-S1, woodchuck hepatoma Pre S gene; a to e, oligonucleotides used for PCR amplification of the integration (see Materials and Methods); star, left junction sequence shown in panel B. (B) Nucleotide sequence map of the left N-myc1–WHV junction. The N-myc1 sequence is in lowercase letters, and WHV sequences are in uppercase letters and indicated by lines. The map illustrates the inverted repeat of WHV sequences at the N-myc junction. (C) Structures of the WHV and N-myc DNA sequences from the integration which were included in each antisense vector. Each of the sequences was inserted in the inverse orientation in the antisense vectors in order to synthesize antisense RNAs. The antisense vector was pCR3 (see Materials and Methods). The arrows in the figure denote the transcriptional direction to produce a sense-strand RNA. The constructs were inserted into the vector in the opposite direction to produce an antisense RNA.

Primer 1 was N-myc1 exon 3 nucleotides 1 to 21, 5′TGTCTCATGAATGTTCCTCCA3′. Primer 2 was N-myc1 exon 3 antisense-strand nucleotides 926 to 947, 5′ACTCAGTTGTTTGAAAACTTGG3′. Primer 3 was WHV minus-strand nucleotides 1373 to 1393, 5′TCGGGAGGGGGAAAGCGAAAG3′. Primer 4 was WHV minus-strand nucleotides 1906 to 1926, 5′ACGGAAGTCGCATGCATTTAT3′. Primer 5 was WHV minus-strand nucleotides 326 to 345, 5′TACACCACCTGTAATCCTGC3′.

The PCR-amplified fragments were introduced into the pCR3 expression vector (Invitrogen) in an antisense orientation relative to that of a cytomegalovirus promoter (confirmed by sequencing). In Fig. 3C, the arrows depict the directions of transcription for the sense strand. However, the fragments were inserted in the vectors in the opposite direction, which led to production of an antisense RNA. The gene for neomycin resistance was used to select antisense cell lines.

DNA transfection.

WH44KA cells were transfected with 10 μg of antisense plasmid DNA comprising vectors 1 to 4 prepared with a Qiagen kit by lipofection (GIBCO BRL) according to the manufacturers’ instructions. After 48 h, cells were diluted fivefold and selection for the ability to grow in G418-containing medium was performed. The colonies resistant to G418 (400 μg/ml) were isolated 2 weeks later and grown as individual colonies, which were maintained under G418 selection for all further experiments.

Determination of cell growth.

Parental and antisense cell lines were seeded in triplicate wells in six-well plates (Falcon) for 5 days. Cells were harvested by trypsinization, resuspended in Isotonic diluent (Hematall), and counted with a Coulter Electronic counter.

Colony formation in soft agar.

Ten milliliters of molten 0.5% Noble agar (Difco) in complete medium was first added to plates (Falcon) and allowed to harden to form the bottom agar. Cells from control lines and antisense lines were harvested by trypsinization, 2 × 10⁴ cells were resuspended in 8 ml of complete medium and then 2 ml of 1.7% molten agar (45°C) was added, and this mix was immediately laid over the bottom medium. Triplicate plates were incubated at 37°C under 5% CO₂, and the number of macroscopic colonies per plate was counted after 3 weeks.

Measure of tumorigenicity in nude mice.

The tumorigenic capacities of control and antisense cell lines were assayed by subcutaneous injection of 2 × 10⁶ or 1 × 10⁶ cells into two sites in opposite flanks of 4-week-old Swiss nude mice (NIHS-nufDF). Injection sites were monitored for the appearance of tumors 2 weeks after injection.

RESULTS

Characterization of the WHV integration in WH44KA cells.

WH44KA cells were isolated from a primary HCC from a WHV-carrier woodchuck (1). This cell line has a highly malignant phenotype, exhibited by its ability to rapidly form tumors in nude mice, to form colonies in soft agar, and to grow without contact inhibition in cell culture with a doubling time of 21 h (Table 1). The WH44KA cell line does not harbor replicating WHV, and attempts to establish WHV replication in this cell line failed. The nonpermissive nature of WH44KA for WHV replication is a common phenotype of woodchuck tumors observed in vivo.

TABLE 1.

Summary of the growth characteristics of N-myc antisense, vector-only, and parental WH44KA woodchuck hepatoma cells

Constructa	Cell lineb	N-myc protein level (%)	Growth rate (doubling time, h)	No. of colonies on soft agar/100 mm2
4	4-5	5	76.0	15.3
4	4-11	20	35.0	22.7
Vector	V-7	100	24.0	170.3
Parental cell	WH44KA	100	21.0	427.7

^a4, antisense vector used to generate the cell line. 

^b4-5 and 4-11, antisense cell lines generated with antisense vector 4; V-7, pCR3 vector-only control cell line; WHK44A, parental woodchuck HCC cell line from which the WHV integration in N-myc1 was cloned. N-myc1 protein in WHK44A cells was used as the 100% standard. 

KpnI does not cleave the WHV genome, and Southern blot analysis of WH44KA DNA revealed a single WHV integration (Fig. 1A). To determine whether the integration occurred in N-myc1 or N-myc2, we used a N-myc1-specific hybridization probe and identified a unique N-myc1 fragment using HindIII digestion. This fragment cohybridized with WHV probe and was an N-myc1–WHV junction fragment. We cloned the left and right junction fragments and sequenced the virus-cell junctions (Fig. 2), confirming that the WHV integration had occurred in N-myc1. Northern blot analysis identified a 3.9-kb RNA that cohybridized with both WHV and N-myc probes (Fig. 1B). Western blot analysis, with a commercial antibody which identifies only N-myc1 and not N-myc2, identified an abundant 63-kDa N-myc protein in WH44KA cells and not in a nontumor woodchuck liver epithelial cell line (Fig. 1C).

FIG. 1.

Characterizations of WHV DNA and N-myc and WHV X proteins in WH44KA cells. (A) Southern blot illustrating WHV DNA hybridization to a single 7-kb genomic DNA fragment of WH44KA DNA (KpnI digested); (B) Northern blot illustrating WHV DNA and N-myc1 cohybridization to the same 3.9-kb RNA plus WHV hybridization to lower-molecular-weight RNA species; (C) Western blot illustrating a high steady-state level of 63-kDa N-myc1 protein in WH44KA cells and its absence in WLC3 woodchuck liver epithelial cells.

FIG. 2.

Nucleotide sequences of the left (A) and right (B) N-myc1-WHV junctions of the single WHV integration in the 3′ untranslated region of N-myc1 exon 3. A comparison of N-myc2, N-myc1, and WHV sequences with the left and right junction sequences is illustrated. Note the 2-bp homology between WHV and N-myc1 sequences at the integration site, which is a common feature of hepadnavirus integrations. Underlined sequences denote the inverted repeat of WHV sequences.

We mapped the WHV DNA integration and sequenced the integrated WHV DNA plus the additional N-myc1 flanking sequences. Sequence analysis revealed a highly rearranged WHV genome (Fig. 3A). A 19-bp inverted repeat was present at the left-hand virus-cell junction (Fig. 3B). The entire open reading frames for the WHx gene and the Pre S, Middle S, and S genes were also present in the integrated WHV DNA.

Since hepadnavirus X genes have been implicated in hepatocarcinogenesis, we tested for the presence of WHx protein (pX) in the cells. We used a sensitive immunoprecipitation-Western blot-ECL method for detecting pX (4). We did not detect any pX, while parallel assays of samples from chronically infected woodchuck livers routinely detected 10⁴ molecules of pX per hepatocyte. Thus, we concluded, as we have observed in a previous report (4), that WHx is not required for the maintenance of the malignant phenotype. Thus, the presence of WHx was not a complicating factor in our experiments, which were aimed at knocking out the N-myc1 protein to test for its role in the maintenance of the malignant phenotype.

Generation of N-myc antisense vectors and antisense-N-myc-expressing cell lines.

In order to determine which antisense sequences were most effective in knocking out N-myc1, we constructed a series of antisense vectors in which we included N-myc1 exon 3 sequences plus various lengths of the WHV sequences homologous to the integrated WHV DNA (Fig. 3C). The sequences were inserted into the pCR3 vector in the antisense direction under the control of the cytomegalovirus promoter. We transfected the antisense vectors or a control vector, together with a neomycin selection plasmid, and selected 29 cell lines exhibiting G418 resistance. We were able to establish seven cell lines with antisense vectors 1 to 3 and only two cell lines with antisense vector 4. Analysis of the N-myc1 protein expression in the cell lines selected with antisense vectors 1 to 3 revealed normal levels of N-myc1 (Fig. 4A, lanes 1 to 4) and that these cells maintained a malignant phenotype. In contrast, the N-myc1 levels were reduced 95 and 80% in cell lines 4-5 and 4-1, respectively, which were produced with antisense vector 4 (Fig. 4A, lanes 5 and 6), compared to levels in positive control cells with the vector only (Fig. 4A, lane 9) and parental WH44KA cells (Fig. 4A, lane 7). A normal woodchuck liver cell line (WC-3) did not contain N-myc (Fig. 4A, lane 8). Northern blot analysis with an antisense N-myc1 probe detected the major 3.9-kb sense-strand hybrid N-myc–WHV RNA in cell lines 4-5 and 4-11, demonstrating that the N-myc gene was still transcribed as in wild-type WH44KA cells (Fig. 4B). However, by hybridizing the blots with a sense-strand probe, we detected N-myc1 antisense RNA only in the 4-5 and 4-11 cell lines, as expected (Fig. 4C). It was difficult to assess the molar ratios of N-myc1 sense and antisense transcripts in the cells due to variables such as probe-specific activity and efficiency of transfer and hybridization. However, the data demonstrate that antisense RNAs were uniquely present in the antisense vector cell lines 4-5 and 4-11.

FIG. 4.

(A) Western blot illustrating the suppression of N-myc protein accumulation in WH44KA cells transfected with antisense vector 4 and not in cells transfected with antisense vector 1, 2, or 3 or in vector-only control cells. Lanes 1, 2, and 4, lysates from cell lines transfected with antisense vectors 1, 2, and 3, respectively; lanes 5 and 6, lysates of cell lines 4-5 and 4-11, respectively, transfected with antisense vector 4; lanes 3 and 9, lysates from cells transfected with pCR3 vector only; lane 7, lysate from WH44KA woodchuck hepatoma cells; lane 8, lysate from WLC-3 cells (a woodchuck liver epithelial cell line serving as a negative control). (B) Northern blot hybridized with an antisense N-myc probe illustrating the continued presence of sense N-myc1 RNA in antisense cell lines, along with positive and negative controls. Lane 1, positive control WH44KA cells; lane 2, positive control cells (WHK44A cells transfected with pCR3 vector); lane 3, antisense cell line 4-5 transfected with vector 4 (Fig. 3C); lane 4, antisense cell line 4-11 transfected with vector 4 (Fig. 3C); lane 5, negative control cells (WLC-3 cells). (C) Hybridization with a sense-strand N-myc probe to detect the antisense N-myc RNA produced by antisense vector 4 in cell lines 4-5 (lane 3) and 4-11 (lane 4).

Reversal of the malignant phenotype in cell lines expressing N-myc1 antisense vector 4.

We assessed the malignant phenotype of the antisense cell lines 4-5 and 4-11 by comparing it with the vector-only and parental phenotypes. Both antisense lines exhibited a reduction in growth rate compared to that of the vector-only control cells (Fig. 5). The doubling time of the antisense lines increased from approximately 24 to 35 or 76 h (Table 1). The phenotype of the 4-5 antisense line reflected a clear alteration to a flattened and enlarged appearance, as well as a clear reduction in the propensity of the cells to pile up in culture (Fig. 6). The ability of the antisense lines 4-5 and 4-11 to form colonies in soft agar was also reduced by 92 and 88%, respectively (Table 1).

FIG. 5.

Growth curves of control and N-myc1 antisense cell lines. V--7, pCR3 vector-only control; 4--5 and 4--11, N-myc1 antisense cell lines in which the steady-state N-myc protein level was reduced 95 and 80%, respectively.

FIG. 6.

(A) Morphology of vector control cell line V-7. (B) Morphological change in cell line 4-5, which shows flattened cells compared to those of the V-7 cell line. Phase contrast; magnification, X340.

The tumorigenic capacities of the antisense lines 4-5 and 4-11 were also compared to those of the vector-only and parental cells by analyzing the results of their subcutaneous injection into nude mice. When either 1 or 2 million cells were injected subcutaneously, tumors developed rapidly at all injection sites for parental cells (WH44KA) and vector-only cells (V-7) (Table 2). The 4-11antisense line, which did not exhibit a complete suppression of N-myc expression, formed small tumors in the nude mice which were 14 to 39% of the sizes of control tumors in different experiments (Table 2). In contrast, the 4-5 antisense cell line, with nearly complete N-myc1 suppression, did not form tumors in nude mice. A small growth was observed at one inoculation site, and we were not able to confirm its origin. Thus, antisense vector 4, when its genes were expressed in WH44KA cells, was able to block the malignant phenotype by all criteria tested.

TABLE 2.

Suppression of tumorigenicity in nude mice of cell lines expressing N-myc1 antisense vectors^a

Cell line	No. of cells injected (106)	No. of mice with tumors/total no. of mice (%)	Mean vol (mm3)
WHKA44	2	5/5 (100)	126.0
V-7	2	5/5 (100)	84.0
4-11	2	5/5 (100)	18.5
4-5	2	1/5 (20)	1.5
WHKA44	1	5/5 (100)	49.2
V-7	1	5/5 (100)	40.5
4-11	1	5/5 (100)	19.5
4-5	1	0/5	0.0

^aCells were transplanted and tumors were analyzed as described in Materials and Methods. Cell lines are as described in Table 1. 

DISCUSSION

The WHV DNA integration in WH44KA cells is somewhat unusual in that it occurred in N-myc1 instead of the N-myc2 retroposon, which is the preferred integration site for WHV (7). The high frequency of N-myc integrations suggests that selection for N-myc overexpression in woodchuck liver is very strong. In addition, N-myc genes may be located at a chromosomal site which is more susceptible to illegitimate recombination. The rearranged structure of the integrated WHV DNA sequences in WH44KA cells is consistent with results of previous studies (24, 27).

Northern blot analysis revealed a single N-myc1–WHV fusion RNA in WH44KA cells. We observed several additional small WHV RNAs in the cells which did not cohybridize with N-myc (data not shown). Therefore, these RNAs must originate and terminate within the integrated WHV sequences. The integrated WHV DNA contains promoters for the pregenomic RNA, WHx gene mRNA, Pre S mRNA, and S mRNA and poly(A) addition signals which may give rise to at least seven WHV RNAs. The specific promoters that were active in the integrated DNA were not determined.

The WHx and HBx genes have been implicated as tumor promoters in transgenic-mouse studies (4, 13, 16, 30). Since the integrated WHV DNA sequences contained an intact WHx gene promoter and a WHx open reading frame, we determined whether WH44KA cells contained any detectable pX. Using a sensitive immunoprecipitation-Western blot-ECL detection method, we were not able to detect any pX in WH44KA cells. This observation is consistent with our previous observation that woodchuck HCCs, which are nonpermissive for WHV replication, do not contain WHx protein (4). Thus, even though the WHx open reading frame is present in the integrated WHV DNA, it does not appear to be functional in this cell line.

The mechanism of action of antisense RNA is thought to involve hybridization of antisense RNA to a target mRNA or pre-RNA (20, 22). In mouse L cells, thymidine kinase (TK) antisense expression induced a retention of TK sense RNA in the nucleus without any alteration in total cellular TK sense RNA (14). In WH44KA cells, downregulation of N-myc1 protein occurred in the continued presence of N-myc1 mRNA, suggesting that the block was at the translational level. However, in the 4-5 cell line the steady-state level of N-myc1 mRNA was reduced to approximately half. This suggests that the antisense RNA, which contains WHV enhancer sequences, may also have a direct effect on the action of the WHV enhancer that is needed to activate the N-myc1 promoter.

The degree of downregulation of N-myc1 was 95% or greater in antisense line 4-5 and approximately 80% in line 4-11. A reversal of the malignant phenotype was more complete in line 4-5, which did not form tumors in nude mice. However, inhibition of colony-forming ability in soft agar was reduced by approximately 90% in both the 4-5 and 4-11 cell lines compared to that of vector-only controls. Our data are consistent with the hypothesis that N-myc proteins affect the activity of a set of genes affecting growth and differentiation of woodchuck hepatoma cells. Therefore, only when the N-myc level is virtually eliminated do the cells exhibit flattening and a complete loss of tumor-forming ability.

During hepatocarcinogenesis in WHV carrier woodchucks, N-myc overexpression in precancerous lesions is commonly observed (35). In woodchuck liver cell cultures N-myc overexpression in the absence of growth factors can cause apoptosis. However, addition of insulin-like growth factor II to the media blocks apoptosis and promotes colony formation (36). Insulin-like growth factor II is coordinately expressed with N-myc during woodchuck hepatocarcinogenesis, suggesting that the two gene products complement each other’s function during hepatocarcinogenesis.

A previous distinction has been made between precancerous lesions that express a moderate level of N-myc and are permissive for WHV replication and foci within lesions that express very high levels of N-myc and are nonpermissive for WHV replication (35). This distinction has been interpreted as evidence for selection of the high-level-N-myc, low-level-WHV phenotype in malignant woodchuck HCCs. The data in this paper illustrate that a nearly complete elimination of N-myc is necessary to reverse the malignant phenotype. The data also establish a direct role for N-myc in the maintenance of the malignant phenotype.