Walleye Dermal Sarcoma Virus: OrfA N-Terminal End Inhibits the Activity of a Reporter Gene Directed by Eukaryotic Promoters and Has a Negative Effect on the Growth of Fish and Mammalian Cells

Abstract

Walleye dermal sarcoma virus (WDSV) is a fish retrovirus causing a skin tumor termed walleye dermal sarcoma, which develops and regresses on a seasonal basis. The WDSV genome contains three short open reading frames designated orfA, orfB, and orfC in addition to the viral structural genes, gag, pol, and env. orfA and orfB transcripts are detected in tumors by reverse transcription-PCR. Recently, OrfA, whose amino acid sequence is similar to that of cyclins A and D, has been shown to complement a cyclin-deficient yeast strain. We report that expression of the accessory gene orfA inhibited nonspecifically the activity of a reporter gene directed by various eukaryotic promoters. In addition, stable transfection with the wild-type orfA generated substantially fewer G418-resistant colonies in both fish and mammalian cells than the parent vector. An orfA mutant expressing only the first N-terminal 49 residues of the full-length protein had the same negative effect on the activity of the reporter gene and on the number of stably transfected colonies as the full-length OrfA. Thus, OrfA inhibits cell growth and/or causes cell death, and the first 49 N-terminal residues of this protein are sufficient to cause these negative effects.

The walleye dermal sarcoma virus (WDSV) is a retrovirus etiologically associated with a skin tumor termed walleye dermal sarcoma (WDS) that is endemic in walleye fish throughout North America (2, 14). In contrast to other tumors induced by animal retroviruses, WDS cyclically develops and regresses (12, 13). Tumors develop in the fall, when viral expression and virion production are minimal (3), and then persist and increase in size through winter until spring, when viral expression and virion production are maximal and when tumors synchronously undergo coagulation necrosis and show a second type of cell death morphologically consistent with apoptosis (13). Viral transmission presumably occurs at that time, when fish congregate for spawning (2, 18, 19).

Retroviruses are currently classified into simple and complex retroviruses. The latter group includes lentiviruses and spumaviruses, whose larger genome contains accessory genes in addition to the structural genes gag, pol, and env. Accessory genes are involved in pathogenicity, in oncogenesis, and in the control of viral gene expression (5, 6, 10). The large size of WDSV genome (12.708 kb [DNA]) and the presence of three accessory genes place WDSV in the complex retrovirus group (8, 12).

orfA and orfB are two nonoverlapping open reading frames located immediately downstream of the env gene. Their high transcription levels in regressing tumors and their low transcription levels in developing tumors support that they play a major role in tumor development and regression (11, 16).

The similarity of orfA to cyclin D at the amino acid sequence level is reflected functionally by its ability to complement a cyclin-deficient yeast strain mutant. Since cyclins are involved in both neoplasia and cell death, orfA might play a role in the benign biological behavior of WDS in nature (11). The objective of the present study was to determine the role of orfA in tumor development and/or regression.

W12 cells, derived from WDS tissue, were cultured at 25°C without CO₂ supplement. The firefly (Photinus pyralis) luciferase gene was used as a reporter gene and was placed under the control of the simian virus 40 (SV40) early promoter in pSV40 (Promega, Madison, Wis.). The pCMV reporter plasmid was constructed by placing the HindIII-BamHI 2.7-kb fragment containing the luciferase gene from pGL2-basic (Promega) under the control of the cytomegalovirus (CMV) immediate early promoter in pcDNA3 (Invitrogen, Carlsbad, Calif.). pActin was derived from the plasmid pFV6a-CAT (provided by P. B. Hackett [University of Minnesota]), which contains the promoter and the first intron of the carp β-actin gene. pActin was constructed by replacing the CAT gene of pFV6a-CAT with the BamHI-NheI 2.7-kb fragment containing the luciferase gene. pMMTV was derived from the pMSG vector (Pharmacia Biotech Inc., Baie D’Urfé, Quebec, Canada) by placing the NheI-SalI fragment containing the luciferase gene downstream of the mouse mammary tumor virus (MMTV) full-length promoter. To construct pLTR, two primers, LTR-F (forward, 5′-CTCGGTACCAAATGAGAAACTAA) and LTR-R (reverse, 5′-CGGAAGCTTTGTTAATTCAAATT), were used to PCR amplify the WDSV long terminal repeat (LTR) (590 bp) from a WDSV clone (12). The 5′ ends of the LTR-F and LTR-R primers contained a KpnI site and a HindIII site, respectively. The purified PCR amplicon was inserted in the polylinker sequence upstream of the firefly luciferase gene in the promoterless vector pGL2-basic (Promega). To construct porfA (Fig. 1), the entire orfA gene (894 nucleotides [nt]) was PCR amplified from a WDSV full-length clone (12) by using the primers ORFA-F (forward, 5′-ATAAGACTACTACAGGGTACGTCC) and ORFA-R (reverse, 5′-AGTTATCCTATTGGATCGACGACG) and subcloned into pSVK3 (Pharmacia Biotech Inc.). To generate pEcoRV, EcoRV was used along with SmaI to remove the orfA ATG initiation codon. pNdeI was generated by NdeI digestion of porfA, end repair with Klenow fragment, and intramolecular ligation; the resulting 2-bp insertion created a +2 frameshift at 144 bp (NdeI site) and fused five out-of-frame codons.

FIG. 1.

(A) WDSV genomic structure and restriction sites of orfA (894 bp). The WDSV genome is 12.7 kb long (DNA) and contains three accessory genes, orfA, orfB, and orfC. (B) Genetic organization and mutants of the WDSV accessory gene orfA. The enzymes used to generate deletions are shown on the top of the full-length orfA (porfA). Dark boxes indicate the original open reading frame, whereas empty boxes illustrate disruption of the original open reading frame. The mutant pNdeI was generated by insertion of 2 bp at the NdeI site, resulting in the disruption of the open reading frame and fusing 5 out-of-frame codons [wt orfA: 139-GCA (A) ACC (T) CAT (H) ATG (M) GTC (V) CTG (L) TTA (L) AAA (K)-163; pNdeI: 139-GCA (A) ACC (T) CAT (H) ATA (I) TGG (W) TCC (S) TGT (C) TAA (stop codon)-163]. The fidelity of the mutants was confirmed by sequencing.

W12 and NIH 3T3 cells were transfected with Lipofectamine (GIBCO BRL, Gaithersburg, Md.), and luciferase activity was determined with a Lumat LB 9501/16 (Berthold). pCMV, which contains a neo selection marker gene, was used to cotransfect W12 and NIH 3T3 cells (from the American Type Culture Collection) for stable transfection with wild-type (wt) orfA and orfA mutants. Transfected cells were selected with G418 for 12 days, fixed, and stained with Giemsa, and G418-resistant colonies were counted. Total cellular RNA was isolated from cells stably transfected with porfA and pNdeI and treated with DNase. cDNA was generated and PCR amplified with the primers ORFA-F and ORFA-R. The reverse transcription-PCR products were analyzed by Southern blotting with an orfA-specific probe generated by PCR from a WDSV full-length clone (12). In order to detect apoptosis, we examined transiently transfected cells to detect DNA fragmentation by agarose gel electrophoresis. Cells floating in medium and adherent cells were collected, pelleted, and resuspended in 20 μl of phosphate-buffered saline containing 50 μg of RNase per μl. The suspension was directly loaded into a 1.8% agarose gel well containing 10 μl of 4% sodium dodecyl sulfate. The well was sealed with low-melting-point agarose and incubated at room temperature for 45 min. Following electrophoresis, genomic DNA was examined under UV light.

The BLAST program was used to search databases. Amino acid alignments were carried out visually. The amino acid substitutions used to determine sequence similarities were as follows: D = E, F = Y = W, M = L = V = I, R = K = H, and S = A, S = T.

orfA expression nonspecifically inhibits luciferase activity directed by eukaryotic promoters.

In fish cells, orfA expression inhibited the luciferase activity directed by WDSV-LTR and the carp β-actin promoter 18-fold and 3.5-fold respectively (Fig. 2). To further define this negative effect, several reporter vectors containing eukaryotic promoters (CMV immediate early promoter, SV40 early promoter and MMTV full-length LTR) were constructed and cotransfected with porfA. Wt orfA expression decreased the luciferase activity directed by the CMV, SV40 and MMTV promoters 278-, 34-, and 317-fold, respectively (Fig. 2).

FIG. 2.

Effect of orfA expression on luciferase reporter gene activity driven by different promoters in a fish cell line, W12. The effector plasmid, porfA, was cotransfected with reporter genes consisting of the luciferase gene directed by the CMV immediate-early promoter, the MMTV full-length LTR, SV40 early promoters, the WDSV full-length LTR, and a carp β-actin promoter. The parent plasmid of porfA (pSVK3) was used as a negative control.

To further characterize OrfA functionally, a deletion mutant (pEcoRV) and an insertion mutant (pNdeI) were constructed. pEcoRV lost orfA inhibitory activity in both NIH 3T3 and W12 cells, while in contrast, pNdeI showed the same inhibitory activity as the full-length orfA (Fig. 3).

FIG. 3.

Mutational analysis of orfA gene. The luciferase reporter plasmid pCMV and pLTR were respectively cotransfected with full-length orfA and the orfA mutants. The parent plasmid of porfA (pSVK3) was used as a negative control. (A) W12 cells; (B) NIH 3T3 cells.

orfA expression decreases the number of G418-resistant cell colonies.

To determine if orfA expression affects cell growth, we conducted stable-transfection experiments. In W12 cells, the number of G418-resistant colonies resulting from cotransfection with orfA or pNdeI was reduced by 34% compared to the number of colonies stably transfected with the parent vector pSVK3. In NIH 3T3 cells, the number of G418-resistant colonies was reduced 4 times by the expression of wt orfA and 5 times by the pNdeI mutant (Fig. 4; Table 1).

FIG. 4.

Effect of orfA expression on the number of G418-resistant cells transfected with the neo resistance gene. Full-length orfA (porfA) (panels 2 and 5) and the mutant pNdeI (panels 3 and 6) were cotransfected with plasmid pcDNA3, which expresses the selection marker gene, neo. Transfected cells were selected for two weeks with G418. G418-resistant colonies were fixed with methanol and stained with Giemsa. Transfections in panels 1, 2, and 3 were conducted in NIH 3T3 cells and transfections in panels 4, 5, and 6 were carried out in W12 fish cells. In panels 1 and 4, cells were transfected with the parental plasmid pSVK3.

TABLE 1.

Numbers of G418-resistant colonies after transfection with full-length orfA and orfA mutants

Plasmid	No. of G418-resistant colonies ina:
	NIH 3T3 cells	W12 cells
pSVK3	425 ± 52	1,547 ± 29
porfA	107 ± 27	1,025 ± 156
pNdeI	87 ± 33	1,059 ± 87

^aValues are means ± standard deviations. 

orfA is expressed in G418-resistant cells.

Total cellular RNA was isolated from G418-resistant cells after 2-week drug selection, and cDNAs were synthesized by using a poly(dT) primer and PCR amplified. Hybridization of PCR products with a WDSV orfA probe demonstrated that orfA was expressed in G418-resistant cells generated by porfA and pNdeI transfection (Fig. 5). We observed DNA fragmentation into oligonucleosome-sized DNA fragments in some regressing tumors (20), indicating that tumor regression may occur through apoptosis. Next, we asked if the reduction in the number of G418-resistant colonies resulting from orfA expression was the result of apoptosis. NIH 3T3 cells were transfected with porfA and the mutant pNdeI. Transfected cells were collected at different times (0, 6, 12, 18, 24, 30, and 36 h) posttransfection. Control cells treated with actinomycin showed typical DNA fragmentation. No DNA fragmentation was detected from cells transfected with either wt orfA or the mutant pNdeI (data not shown).

FIG. 5.

Southern blot analysis of PCR-amplified orfA cDNA expressed in G418-resistant cells. orfA cDNA was synthesized from total RNA isolated from G418-resistant W12 cells and PCR amplified with the primer pair ORFA-F–ORFA-R. The PCR product was electrophoresed, blotted onto a nylon membrane, and hybridized with a WDSV orfA-specific probe. Lane 1, cells transfected with the parent plasmid; lanes 2 and 3, cells transfected with porfA and pNdeI, respectively. The molecular size of orfA is indicated on the right.

We demonstrated that transfecting cells from widely different classes of vertebrates with the full-length orfA inhibited the activity of a reporter gene directed by various eukaryotic promoters and that this inhibition did not require any other viral gene products (Fig. 2 to 4). In addition, orfA expression resulted in fewer piscine and mammalian cells surviving G418 selection. Removal of the orfA initiation codon (pEcoRV) abolished orfA inhibitory activity, and a frameshift at codon 50 preserved the inhibitory activity shown by the full-length orfA. Considered together, these observations indicate that orfA expression inhibits cell growth and/or causes cell death and that the first 49 N-terminal residues of this protein are sufficient to cause these negative effects. These observations also suggest that these effects are mediated by OrfA protein, not by orfA mRNA. The expression of orfA in cells that survived G418 selection is best explained by an arrest in the cell cycle, by low levels of orfA expression, or by the selection of orfA mutants.

A database search using the BLAST program showed that the pNde mutant sequence was similar to that of the N-terminal helix, the α1 helix, and a large part of the α2 helix of the first cyclin box of cyclin A. The first N-terminal amino acids of OrfA corresponded exactly to the second N-terminal amino acid sequence of the human cyclin A N-terminal helix (amino acids [aa] 179 to 190), and the first 12 aa of OrfA were very similar to those of the cyclin A N-terminal helix (Fig. 6; Table 2). Other similarities were the presence of alanines 235, 259, and 264, characteristic of the cyclin fold (numbering from the human sequence), the presence of the corresponding alanine 363 and alanine 359 present within the second repeat, the presence of lysine 266 and glutamic acid 295, which contact cdk2 in the cyclin A-cdk2 complex, and the presence of a hydrophobic patch in the putative α1 helix, corresponding to the MRAIL sequence region of the human cyclin A (9) (Fig. 6). The degree of similarity between OrfA and the first cyclin box of human cyclin A (aa 208 to 303) as a whole was the same as that obtained with cyclin D (11). However, the similarity between OrfA amino acid sequence and that of cyclin D is low in the N-terminal region, upstream of the cyclin D α helix A (11). In contrast, this region was very similar to the N-terminal region (including the N-terminal helix) of cyclin A (Fig. 6; Table 2), and the similarity of cyclin A α3 and α5 helices with the corresponding regions of OrfA is greater than the similarity of cyclin D helices C and E with the corresponding regions of OrfA (Table 2). Thus, the OrfA amino acid sequence is most comparable to that of cyclin A truncated at its N-terminal end.

FIG. 6.

Amino acid sequence alignment of the orfA accessory gene in WDSV, WEHV1, and WEHV2 with cyclin A of humans (Hum), goldfish (Gfish), clam, Chlorohydra viridissima (Hydra), and Drosophila melanogaster (Dros.). Identical and similar amino acids are in boldface. The 10 α helices that compose the cyclin box are indicated by boxes. Conserved residues are indicated by asterisks, notably alanines 234 and 263, lysine 265, and glutamic acid 284 (numbering is from the human sequence).

TABLE 2.

Comparison of OrfA amino acid sequences with those of D- and A-type cyclins within the third and fifth α-helix of the first repeat of the cyclin box

OrfA region (size)	% Identity/% similaritya with sequence from:
	Human cyclin D helices			Cyclin A helices from:					WEHV1	WEHV2
	D1	D2	D3	Human	Goldfish	Clam	Hydra	Drosophila
α helix C (25 aa)	28/44	28/48	28/48						36/48	28/32
α helix E (20 aa)	25/35	20/35	15/30						25/35	25/35
N terminus (12 aa)				33/42	33/42	33/42	33/42	42/50	25/42	25/42
α3 helix (16 aa)				37/56	33/56	50/62	50/62	50/56	43/62	44/50
α5 helix (15 aa)				20/47	27/40	20/47	20/40	27/47	40/60	20/40

^aPercent similarity is the sum of allowed (silent) substitutions and identical amino acids. 

Cyclin A has been one of the first cyclins identified as being involved in cancer (4). Paradoxically, cyclin A expression and cdk2 activation have also been associated with apoptosis (7, 15). Similarly, OrfA might arrest the cell cycle or cause apoptosis, or on the contrary trigger cell proliferation (11). Strikingly, a region of ODV-EC27 (EC27), a protein of Autographa californica nucleopolyhedrovirus, which is very similar to the cyclin box α1 and α2 helices, corresponds exactly to orfA mutant pNdeI. EC27 arrests the cell cycle and has functional features of both cyclin B (it binds cdc2) and cyclin D (it binds cdk6) (1). By binding cdc2, EC27 might be responsible for cell cycle arrest at G₂/M by forming an active cyclin B-like–cdc2 complex that resists degradation (1). Alternatively, a putative OrfA-cdk complex may be inactive. In the cdk2-cyclin A complex, the N-terminal cyclin box fold (residues 208 to 306) and the N-terminus helix are both involved in contacting cdk2. Thus, the OrfA N terminus (pNde), which contains the N-terminal helix and a large part of the first cyclin box of cyclin A, might bind cdk2 and inactivate it, thereby preventing mitosis. Since cyclins determine the substrate specificity of cyclin-cdk complexes, OrfA could also modify cdk2 specificity by recruiting different substrates (17) (Fig. 6; Table 2).

Our inability to detect apoptosis in the current transfection studies might be due to a number of factors. orfA constitutive expression might cause apoptosis only at specific steps of the cell cycle. Since cell cycles were not synchronized in the current study, transfected cells would become apoptotic only as they enter a specific step of the cell cycle, and thus only a small number of transfected cells would be apoptotic at any given time. It is also possible that the low transfection efficiency (5%), as determined by transfection with the pCMV-β-Gal plasmid, did not allow the detection of apoptosis. OrfA might trigger apoptosis after a period longer than that used in our assay. Alternatively, OrfA may cause cell death by nonapoptotic mechanisms. Finally, it is possible that other proteins, viral or cellular, are involved in the apoptosis observed in regressing tumors in addition to, or instead of, orfA.