A 2.9-Kilobase Noncoding Nuclear RNA Functions in the Establishment of Persistent Hz-1 Viral Infection

Abstract

Differential viral gene expression during both productive and persistent infections of Hz-1 virus in insect cells was elucidated. Despite more than 100 viral transcripts being expressed during productive viral infection, massive viral gene shutoff was observed during viral persistency, leaving the 2.9-kb persistence-associated transcript 1 (PAT1) as the only detectable viral RNA. Persistence-associated gene 1 (pag1), which encodes PAT1, was cloned and found to contain no significant open reading frames. PAT1 is not associated with the cellular translation machinery and is located exclusively in the nucleus. Further experiments showed that PAT1 is functional in the establishment of persistent Hz-1 viral infection in the cells. All the evidence collectively indicates that PAT1 is a novel nuclear transcript of viral origin. Our results showed that although PAT1 and XIST RNA, a mammalian X-inactive specific transcript, are transcribed by different genes, they have interesting similarities.

Persistent viral infection has been reported to occur naturally in insects and cultured cells. Changes in rearing temperature, the presence of high humidity, a decrease in food quality, superinfection with different viruses, and/or other stimuli may activate persistent infections (6, 11, 21, 22, 34, 46). However, due to the difficulty in establishing persistent viral infections in laboratory stocks of insects, persistent viral infections are usually studied only after unexpected viral activation from previously healthy-looking insects or insect cells (21, 34, 46).

Hz-1 virus (also called Hz-1 baculovirus or Hz-1V) was originally identified in a persistently infected IMC-Hz-1 cell line from the ovarian tissue of Heliothis zea (15). It is capable of establishing persistent infection in various insect cells (6, 10, 36). Such persistent infection may be reactivated to resume productive infection upon infection by heterologous viruses (6). Persistently infected cell cultures are resistant to superinfection by homologous viruses due to the induction of apoptosis (26). Like baculoviruses (2, 44), Hz-1 virus is rod shaped with a circular, double-stranded 228-kb DNA genome (9, 20). It was previously referred to as the type species of the subfamily Nudibaculovirinae in the family of Baculoviridae (45). Recently, Hz-1 virus and other nonoccluded baculoviruses were removed from the baculovirus family, and they are temporarily unclassified (44).

Hz-1 virus can establish both productive and persistent infections in several lepidopteran cell lines (6). Upon productive infection, more than 100 different viral transcripts can be detected (10), and the infected cells eventually die by necrosis (26). However, a very small proportion of the infected cells, usually less than 0.2%, grow and become persistently infected clones. In these cells, only one 2.9-kb viral transcript was detectable in Sf9, Sf21, and TN368 cells that were persistently infected with Hz-1 virus (reference 10 and data not shown), and this RNA species was named persistence-associated transcript 1 (PAT1).

While little is known concerning persistent viral infections in insects, there are several relatively well-studied examples in the herpesvirus system. For instance, approximately 12 genes are expressed during latent infection by Epstein-Barr virus, but more than 50 viral genes are expressed in its lytic phase of viral growth (24, 31). Herpes simplex virus probably expresses more than 70 genes during productive viral infection, but only 3 related latency-associated transcripts (LATs) are detectable during latent viral infection (39, 42). The study of similar differential viral gene expression in viruses other than members of the herpesvirus family would provide useful information for understanding the molecular basis of persistent or latent viral infections in eukaryotic cells.

Previously, we localized the region transcribing PAT1 within the EcoRI-M fragment of the Hz-1 viral genome (10). In the present study, persistence-associated gene 1 (pag1), which encodes PAT1, was cloned and characterized. We found that PAT1 is not associated with the cellular translation machinery and, interestingly, that it is located exclusively in the nucleus, where PAT1 functions in the establishment of persistent Hz-1 viral infection.

MATERIALS AND METHODS

Cell lines.

Cell line TN368 was derived from Trichoplusia ni (18), and the other two cell lines, Sf21 and Sf9, were from Spodoptera frugiperda. Persistently infected cells, TNP3 cells (10), were derived from TN368 cells; and SfP2 cells were derived from Sf21 cells. All cells were maintained at 26°C in TNM-FH medium supplemented with 8% fetal bovine serum (Gibco BRL Life Technologies).

DNA sequencing, primer extension, and RNase protection analysis.

Progressive deletion clones were constructed in both directions from the viral genomic EcoRI-M fragment by using an exonuclease III-mung bean nuclease technique (17). The nucleotide sequence was determined directly from double-stranded plasmid DNA by the dideoxynucleotide chain termination method (12). Both strands were sequenced at least twice. Computer open reading frame (ORF) analysis of pag1 was done by using the Sequence Analysis Software Package of the Genetics Computer Group (GCG) (University of Wisconsin Biotechnology Center). Clusters of direct repeats were evaluated by self-comparison analysis of the pag1 sequence with a Dotplot program from the same software package.

The transcription start site was determined by primer extension. A 35-bp primer from nucleotide 1109 (3′) to 1143 (5′), antisense to PAT1, was used. Total RNAs (25 μg) extracted from both productively infected TN368 cells and persistently infected TNP3 cells were used as templates for the experiments.

The 3′ end of PAT1 was mapped by RNase protection assays. Plasmid pHzEM-C, which contains only subfragment C of the EcoRI-M viral genome fragment, was used (Fig. 1A). Antisense probes were transcribed, using T3 polymerase, from pHzEM-C, which was restricted with HpaI and hybridized to the total RNAs (30 μg) from persistently infected cells. After RNase A and T₁ digestion, the protected fragments were fractionated on 6% polyacrylamide–urea gels.

FIG. 1.

Location and sequence of pag1. (A) Map and location of PAT1 coding region. The first line represents percentages of the viral genome, the EcoRI map of the linearized 228-kb Hz-1 viral genome (9) is shown in the second line, the KpnI map of the EcoRI-M fragment is shown in the third line, the region sequenced is shown in the fourth line, and the orientation and transcriptional region of PAT1 are shown in the fifth line. The letters above the lines denote the restricted viral DNA fragments, and those below the lines denote restriction sites (Kpn, KpnI; Eco, EcoRI). (B) Nucleotide sequence of pag1. The putative AP1 consensus sequence and putative GATA, CAAT, and TATA motifs are underlined. CAGT, the conserved transcription start sequence for baculovirus early genes, is boxed. The transcription start site is indicated by an arrow and the transcription termination site is marked by an asterisk. (C) Primer extension was used to determine the transcription start site of PAT1. Lanes 1 to 4, sequence ladders. The extended (78-bp) and the primer (35-bp) bands derived from the total RNAs extracted from productively infected TN368 cells (lane 5), persistently infected TNP3 cells (lane 6), and healthy parental TN368 cells (lane 7) are shown. (D) RNase protection was used to map the 3′ end of PAT1. A ³²P-labeled RNA probe was transcribed by using T3 polymerase from subfragment C of viral EcoRI-M fragment. Before in vitro transcription, subfragment C was digested with restriction enzyme HpaI to generate an EcoRI-HpaI single-stranded antisense RNA probe (see panel A). Lanes 4 to 7, are sequence ladders. Three closely associated bands were protected for productively infected TN368 cells (lane 1) and persistently infected cells TNP3 (lane 2). The major protected 82-bp band is marked. No protection was observed for RNA extracted from uninfected TN368 cells (lane 3). nt, nucleotides.

Cloning of overlapping cDNA fragments by RNA PCR (reverse transcriptase PCR).

cDNA fragments were amplified by RNA PCR. By using paired primer sets, the H, M1, M2, M, and T fragments (see Fig. 3) were amplified, cloned, and sequenced. The region covered by the P fragment (see Fig. 3) is the promoter region, which cannot be generated by reverse transcriptase PCR and thus served as a negative control.

FIG. 3.

Amplification of PAT1 cDNA fragments by RNA PCR. Overlapping fragments covering the entire 2.8-kb cDNA were amplified by using paired primers. Regions of the amplified fragments H, M, M1, M2, and T were sequenced. The experiments, from RNA PCR to sequencing, were repeated twice. The P fragment, which resides in the promoter region, was subjected to RNA PCR amplification to serve as a negative control.

Analysis of the promoter region of pag1.

Progressive deletions from both ends of the 5′ regulatory region of pag1 were generated by PCR and further verified by DNA sequencing. For DNA deletion analysis of the region specifying the 5′-end sequence of PAT1, five progressive deletion fragments were synthesized and ligated separately upstream of an intact luciferase-coding region (see Fig. 4), and then 5 × 10⁵ Sf9 cells were cotransfected with 1 μg of plasmid DNA containing each of the above-mentioned deletion constructs and a construct containing a chloramphenicol acetyltransferase (CAT)-coding sequence driven by the Drosophila actin promoter (0.25 μg). The latter construct was used as an internal control to normalize the efficiency of transfection.

FIG. 4.

Promoter activity analysis of pag1. (A) Transfection of pag1 into host cells, indicating that viral factors are not essential for PAT1 transcription by the pag1 promoter. Plasmid pHzE-M (10), which contains the entire pag1 gene, was transfected into Sf9 cells. Total RNAs were harvested and analyzed by Northern hybridization. PAT1 signals can be found at 4 and 8 h after transfection. Total RNAs harvested from parental TN368 cells and persistently infected TNP3 cells were also used as negative and positive controls, respectively. The probe used for this analysis was a gel-purified subfragment E derived from the viral genomic EcoRI-M fragment (see Fig. 1A). Numbers on the left are in kilobases. (B) Progressive deletion analysis at the 3′ end of the upstream regulatory region of pag1. These fragments were ligated with the protein-coding region of the luciferase gene, and the activity of this enzyme was determined after the transfection of the constructs into Sf9 cells. All of the constructs were cotransfected with another construct containing a Drosophila actin promoter-driven CAT gene as an internal control. Data (means ± standard deviations) were collected from triplicate assays in three independent experiments. (C) Progressive deletion analysis at the 5′ end of the upstream regulatory region of pag1. All the fragments ended at +29 bp and were ligated with the protein-coding region of the lacZ gene. CAAT, GATA (TTATC), and TATA motifs are shown. Following transfection of the constructs into Sf9 cells, LacZ activities were determined. All of the constructs were cotransfected with a construct containing a Drosophila actin promoter-driven CAT gene to serve as negative controls. One unit of LacZ activity is equal to the intensity emitted by 0.1 nM 4-methylumbelliferone. Data (means ± standard deviations) were collected from triplicate assays in three independent experiments.

The 5′ regulatory region further upstream from the transcription start site, within positions −727 to +29, was also analyzed by progressive deletion analyses and ligated separately upstream of a full-length lacZ coding sequence in the plasmid pTSV-2 (28). Sf9 cells (5 × 10⁵) were then cotransfected with two different plasmids: plasmid pTSV-2, containing progressive deletions of the pag1 promoter region driving an intact lacZ coding sequence (1 μg), and a construct containing the CAT-coding sequence driven by the Drosophila actin promoter (0.25 μg). The latter construct was again used as an internal control to normalize the efficiency of transfection. LacZ expression was determined by assaying β-galactosidase activity as previously described (28). Briefly, 40 μl of cell lysate was mixed with 160 μl of reaction cocktail (containing 25 mM Tris-HCl [pH 7.5], 125 mM NaCl, 2 mM MgCl₂, 12 mM β-mercaptoethanol, and 0.3 mM 4-methylumbelliferyl-β-d-galactoside) and incubated at 37°C for 30 min, and then 100 μl of the sample was aspirated into 2 ml of glycine-carbonate reagent and the fluorescence was read with the TKO-100 Mini Fluorometer (Hoefer Scientific Instruments).

Polysome fractionation.

The procedures for isolation and analysis of polysomes described by Schmidt and Merrill (37) were used. TNP3 cells (10⁷) were lysed and subjected to sucrose gradient centrifugation. One-tenth aliquots of the fractionated RNA were assayed by Northern blotting.

Nuclear localization.

Cells were first fractionated into nuclear and cytoplasmic fractions by the procedure of Summers and Smith (43). Nuclear RNA was extracted from the nuclear pellets with guanidinium thiocyanate. Five micrograms of each of the resultant RNAs was serially diluted and slot blotted. Two in vitro-transcribed probes containing either pag1 (Fig. 1, EcoRI-M fragment) or actin sequences from the genome of Drosophila (16) were used for hybridization by both Northern and slot blotting.

Preparation of RNA probes for fluorescence in situ hybridization.

An RNA probe was made for fluorescence in situ hybridization of the persistently infected or stable transfected cell lines as follows: the 0.7-kb subfragment E (restricted by KpnI) of the EcoRI-M fragment (Fig. 1A) was cloned into plasmid pBluescript KSM+ (Stratagene), and a 0.7-kb RNA probe labeled with digoxigenin-11-UTP was produced by in vitro transcription with T3 RNA polymerase (Boehringer Mannheim Biochemicals).

PAT1 detection by fluorescent in situ hybridization.

Sf9, SfP2, and SfPAG1-1 cells were resuspended in fresh medium at a concentration of 10⁶ cells/ml, and then 50 μl of this suspension was pipetted onto the center of the premarked area on an in situ PCR glass slide (Perkin-Elmer Applied Biosystems). Slides were then incubated at 26°C for 1 h, followed by 4% paraformaldehyde fixation in phosphate-buffered saline (PBS) for 30 min. After fixation, paraformaldehyde was inactivated by washing the slides twice (2 min each) in PBS. The labeled RNA probe in hybridization buffer (2× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 5% dextran sulfate, 0.2% bovine serum albumin, and 50% formamide) was applied on each monolayer cell and assembled into AmpliCover discs and AmpliCover clips (Perkin-Elmer Applied Biosystems). Each assembled slide was transferred to a GeneAmp in situ PCR System 100 set for a 40°C soak program and incubated overnight. After hybridization, the slides were washed three times with 2× SSC at 65°C. For the detection of the labeled probe, slides were incubated with 0.25% Triton X-100 in PBS for 5 min and blocked for 30 min with 10% fetal bovine serum (Gibco BRL Life Technologies) in PBS containing 0.25% Triton X-100. Subsequently, the slides were incubated with 0.4 μg of antidigoxigenin fluorescein-conjugated antibodies per ml diluted in 0.25% Triton X-100 in PBS at room temperature for 1 h and then washed three times in PBS containing 0.25% Triton X-100. Finally, slides were incubated in PBS containing 1 μg of DAPI (4′,6′ diamidino-2-phenylindole) (Molecular Probes) per ml for 5 min and washed briefly with PBS.

Establishment of stably transfected cell lines SfPAG and SfPKN.

Two plasmids, pPAGN and pKN, were used to separately transfect Sf9 cells. Plasmid pPAGN contains two genes. The first is a full-length pag1 gene driven by its own promoter and is derived from the EcoRI-M fragment of Hz-1 virus (10) (Fig. 1). The second is a neomycin resistance gene driven by the heat shock promoter of Drosophila (41). Plasmid pKN contains only a neomycin resistance gene driven by the hsp70 promoter (41). Cellfectin (Gibco BRL Life Technologies) was used to transfect Sf9 cells (2 × 10⁵ cells per well of a 24-well plate) with 1 μg of either pPAGN or pKN. Twelve hours after transfection, the cells were replaced with fresh medium and incubated at 26°C for 24 h. Following this, the cells were maintained for 10 days in TNM-FH medium containing 2 mg of G418 per ml. The stably pag1-transfected clones SfPAG1-1, SfPAG1-2, SfPAG2-1, and SfPAG2-2 were established and isolated from different transfection experiments.

Challenge of stably or transiently pag1-transfected cells with Hz-1 virus.

Stably transfected pag1 cells were seeded at 4 × 10⁴ cells per well onto 96-well plates. Cells were challenged with Hz-1 virus (multiplicity of infection [MOI] = 1). The number of persistently infected cell colonies was determined at 10 days after viral infection (dpi). In transient-transfection assays, Sf9 cells were transfected with pPAGN and pKN plasmids separately. Twelve hours after transfection, the cells were infected with Hz-1 virus (MOI = 1), and the number of persistently infected colonies was again calculated at 10 dpi. The criteria we used to score persistently infected colonies were as follows: (i) the cells survived productive viral infection; (ii) the cells grew and formed a colony; (iii) the colony contained more than five cells at 10 dpi (colonies containing fewer than five cells/clone have little chance of survival and are thus difficult to propagate and assay regardless of whether they are persistently infected by Hz-1 virus); and (iv) the cells were resistant to superinfection with Hz-1 virus.

Simultaneous detection of viral DNA and antigen in cells infected by Hz-1 virus.

Parental Sf9 cells, productively Hz-1 virus-infected Sf9 cells (MOI = 2; 18 h postinfection), persistently infected SfP2 cells, and newly established persistently infected SfPAG1-1P cells were used. These cells were derived from the infection of SfPAG1-1 cells with Hz-1 virus. After establishment, cells were mixed and passed four times before in situ analyses. They were first fixed in ethanol-acetic acid (3:1) at −20°C for 30 min. One drop of cell suspension was added to 70% ethanol-cleaned dry slides. After spreading of the cells, the slides were dried at 60°C for 1 h followed by dehydration in ethanol at increasing concentrations (70, 80, 90, and 100%). DNA was denatured by incubation in 2× SSC–70% formamide (pH 7.0) for 2 min at 70°C. After being rinsed for 2 min in 2× SSC, the slides were dehydrated again as described above.

A 0.3-kb HincII subfragment derived from HindIII-K of the viral genome (9) was cloned into plasmid pBluescript KSM+ (Stratagene Cloning System), and a 0.3-kb RNA probe labeled with digoxigenin-11-UTP was produced from this plasmid by in vitro transcription with T3 RNA polymerase (Boehringer Mannheim Biochemicals). Probes diluted in hybridization buffer (2× SSC, 5% dextran sulfate, 0.2% bovine serum albumin, and 50% formamide) were then denatured at 75°C for 5 min, chilled on ice, and placed on the slide, and then the AmpliCover discs and AmpliCover clips (Perkin-Elmer) were reassembled. For hybridization, slides were placed at 40°C overnight in the GeneAmp in situ PCR System 100.

After hybridization, slides were washed four times for 5 min each in 2× SSC–50% formamide at 60°C and three times for 10 min each in 1× SSC at 60°C. Slides were incubated with 0.3% Triton X-100 in PBS for 5 min and blocked for 30 min with 10% fetal bovine serum (Gibco BRL Life Technologies) in PBS containing 0.3% Triton X-100. Similar techniques with modifications were used for the detection of the viral antigen (26). Briefly, slides were incubated for 1 h at room temperature with rabbit polyclonal anti-Hz-1 virus antibody diluted at 1:1,000 in PBS containing 0.3% Triton X-100. After three washes with PBS–0.3% Triton X-100, Cy5-conjugated goat anti-rabbit immunoglobulin G (Jackson ImmunoResearch Laboratories) was added at a dilution of 1:500 in PBS containing 0.3% Triton X-100 and incubated for 1 h. Subsequently, the slides were incubated with 0.4 mg of antidigoxigenin fluorescein-conjugated antibodies per ml diluted in 0.3% Triton X-100 in PBS at room temperature for 1 h and then were washed in three changes of 0.3% Triton X-100 in PBS. Finally, slides were incubated in PBS containing 0.5 mg of propidium iodide (Molecular Probes) per ml for 5 min, washed briefly with PBS, and visualized with a Laser Scanning confocal microscope (Zeiss LSM 310).

Nucleotide sequence accession number.

The sequence reported in this paper has been deposited in the GenBank database (accession no. U03488).

RESULTS

No significant ORFs except clustered repeated sequences were found in the PAT1 coding region.

Previously we showed that the PAT1 sequence resides in the EcoRI-M fragment of the viral genome (10). This fragment was further subdivided into A, B, C, D, and E fragments by using the restriction enzyme KpnI (Fig. 1A, EcoRI-M fragment). The coding region of PAT1 and the putative promoter region were sequenced (Fig. 1A and B).

The initiation and termination sites of PAT1, which were roughly analyzed previously by using RNase protection assays in agarose gels, were found to localize in fragments B and C, respectively (Fig. 1A) (10). To position both the 5′ and 3′ ends of PAT1 precisely, primer extensions and RNase protection analyses in a high-resolution polyacrylamide gel were undertaken.

A 78-base extension product (Fig. 1A and C) was observed by using a 35-base primer beginning at position 1143 (Fig. 1A), suggesting that the transcript was initiated from nucleotide A at position 1066 (Fig. 1B). This position is very close to a conserved transcription start sequence, CAGT, of baculovirus early transcripts (Fig. 1B, boxed). For the transcription termination site of PAT1, the RNase protection experiment indicated the protection of a major 82-base band (Fig. 1A and D), which can be mapped to nucleotide C at position 4002 (Fig. 1B), and several closely associated weaker bands (Fig. 1D).

Analysis with the GCG Sequence Analysis Software Package showed, interestingly, that the coding region of PAT1 contains no significant ORF in any forward or reverse translation frames (Fig. 2A). GCG Dotplot analysis showed that pag1 contains several clustered direct repeats. These repeats were organized primarily into three clusters within nucleotides in the regions of positions 1400 to 1550, 1800 to 2000, and 2100 to 2200 (Fig. 2B and C). The lack of a significant ORF together with unusually clustered repeats suggested that PAT1 may not encode a protein.

FIG. 2.

DNA sequence analysis of pag1. (A) Computer ORF analysis of pag1. ORFs are indicated as open boxes. ATG initiation codons are indicated by a vertical line, and termination codons are indicated by a vertical line bisecting the horizontal. The orientations of the top three frames are the same as for PAT1. The orientations of the bottom three reading frames are opposite to those for PAT1. The transcriptional region of PAT1 is indicated by an arrow. (B) Relative locations of clustered repeats (a, b, and c) on PAT1. (C) Sequences and positions of three major clustered direct repeats (a, b, and c). Only those repeats larger than 10 bases are included.

The cDNA sequence of PAT1 is identical to that of genomic DNA.

Overlapped cDNAs of PAT1 were amplified and cloned from persistently infected TNP3 cells (10) by using an RNA PCR technique. Multiple primers were used to amplify H (251 bp), M (1,638 bp), M1 (870 bp), M2 (809 bp), and T (1,320 bp) cDNA fragments which run across the entire PAT1 coding region (Fig. 3). Since it was previously found that no transcript other than PAT1 was detectable during persistent viral infection, a pair of primers was also used to amplify the promoter region to serve as a negative control. The sequence of PAT1 cDNAs was determined and was found to be identical to the sequence of genomic DNA, indicating that RNA editing or splicing does not occur after transcription.

Viral factors are not essential for pag1 transcription.

It has been shown previously that PAT1 can be detected very early during productive viral infection and that it is the only virus-specific transcript detectable during persistent viral infection (10). To test whether the expression of PAT1 is independent from the expression of other viral genes, a plasmid, pHzE-M, which contains only the putative promoter and the PAT1 coding region (10) was transfected into Sf9 cells. At 4 and 8 h after transfection, total RNA was extracted and analyzed by Northern hybridization. Figure 4A shows that PAT1 was detected 4 h after transfection and that the intensity of the signal had increased greatly by 8 h after transfection. These results indicated that host factors alone were sufficient for the transcription of PAT1, although it was still possible that some viral factors further modulate the expression of pag1 upon infection with Hz-1 virus.

The pag1 promoter is close to the transcriptional start site of PAT1.

To further identify and characterize the promoter of pag1, progressive deletions from both upstream and downstream sequences were constructed and analyzed. For downstream progressive deletions, regions between fixed upstream position −727 and various downstream positions, i.e., +1, +6, +9, +29, and +198, were each cloned and fused to a luciferase-coding sequence (Fig. 4B). Luciferase activity was analyzed after transfection of these constructs into Sf9 cells. The results showed that upstream sequences between nucleotide −727 and the start site at nucleotide +1 gave rise to weak luciferase activity. The luciferase activity gradually increased as the 3′ end of the promoter region was extended to nucleotide +29. However, the promoter activity dropped again upon further extension to nucleotide +198, where two ATG codons are found between nucleotides +29 and +198 (Fig. 4B). The results showed that nucleotides +1 to +29 downstream of the transcription start site were required for better expression of the ligated luciferase sequence (Fig. 4B).

To analyze the upstream sequence required for promoter activity, the region between nucleotides −727 and +29 was first ligated to a LacZ-coding sequence (Fig. 4C). Plasmids containing progressive deletions from the upstream region were transfected separately into Sf9 cells, and the intensity of lacZ expression from each promoter deletion construct was analyzed. Rather similar levels of promoter activity were observed in the constructs containing nucleotides −727/+29 to −315/+29. Activity gradually increased when the construct was deleted up to nucleotide −158. The best promoter activity was observed when the construct was further deleted to nucleotide −90, which still retained the putative CAAT and TATA boxes and a GATA motif. Further deletion into the nucleotide −90/+29 region abolished the activity of the promoter, indicating that the closely associated TATA box and CAAT and GATA motifs were crucial for PAT1 expression (Fig. 4C).

PAT1 is not associated with ribosomes and is localized predominantly in the nucleus.

Although the GCG computer program predicted that PAT1 lacks protein-coding potential (Fig. 2A), it was necessary to test whether PAT1 is associated with the cellular protein synthesis machinery. We therefore examined polysome profiles in postmitochondrial fractions obtained from persistently infected TNP3 cells. As a control, polysomes were dissociated from mRNA by adding EDTA. Gradient fractions were then assayed for PAT1 RNA by Northern blot analysis. A low level of PAT1 signal was detected throughout gradient fractions either without (Fig. 5A) or with (Fig. 5B) EDTA treatment. These results indicated that PAT1 was not associated with ribosomes. In contrast, actin mRNA was detected mainly in the heavy polysome regions when EDTA was omitted (panel c in Fig. 5A). However, in the presence of EDTA, the majority of actin mRNA shifted dramatically to the free ribosome fractions (panel c in Fig. 5B).

FIG. 5.

Polysome fractionation of persistently infected TNP3 cells. (A) Postmitochondrial lysates collected from 10⁷ persistently infected TNP3 cells were subjected to sucrose gradient centrifugation. (a) Profile of optical density at 254 nm of postmitochondrial lysates. Each fraction was then collected and analyzed by Northern hybridization with either a pag1 (b) or actin (c) probe. One-tenth of the total RNA extracted from 10⁷ TNP3 cells was loaded into the control lanes prior to fractionation to serve as a control. (B) (a) Ribosome knockout by EDTA treatment in the polysome fractionation experiment. A profile of the optical density at 254 nm of the gradient isolated from 10⁷ cells (a) and Northern analysis of each fraction with either a pag1 (b) or an actin (c) probe are shown. One-tenth of the total RNA extracted from 10⁷ TNP3 cells was loaded into the control lanes prior to fractionation to serve as a control. The probe used for pag1 hybridization was the same as in Fig. 4A, and the actin probe used was the 0.6-kb BglII/SalI subfragment of the original 1.8-kb Bombyx mori actin clone in pGem (32).

Total signal intensities of PAT1 from all postmitochondrial fractions were found to be much lower than PAT1 signals in the control lanes which contained unfractionated total cellular RNAs (panel b in Fig. 5). These results are contradictory to those for the control actin signals, in that the intensities of all postmitochondrial fractions should be 10 times that of the control lanes (panel c in Fig. 5), thus suggesting that the majority of PAT1 may not be in the postmitochondrial fractions. To determine the distribution of PAT1 in the cell, RNAs from both isolated nuclei and pooled cytoplasmic fractions of TNP3 cells were analyzed separately in slot blots. The results indicated that PAT1 was present almost exclusively in the nuclear fraction, with less than 1% of the PAT1 signal detected in the cytoplasm (Fig. 6A, panel a). In contrast, over 90% of actin RNA was found in the cytoplasmic fraction (Fig. 6A, panel b). Results of fluorescent in situ hybridization provided additional proof that PAT1 is indeed localized in the nucleus regardless of whether it is expressed by the genome of the virus (Fig. 6B, panel a) or by the stably transfected pag1 gene in the cells (Fig. 6B, panel b).

FIG. 6.

PAT1 is localized in the nucleus. (A) Slot blots of nuclear (Nu) and cytoplasmic (Cyt) RNAs from persistently infected TNP3 cells. Slot blots with a series of 10× dilutions of nuclear and cytoplasmic RNAs start from 5 μg per slot. The blots were hybridized with a pag1 (a) or actin (b) probe. The pag1 and actin probes used for this analysis were the same as for Fig. 5. (B) Fluorescent in situ hybridization showing that PAT1 is localized in the nucleus. (a1, b1, and c1) PAT1 fluorescent in situ hybridization of persistently infected SfP2 cells, stably pag1-transfected SfPAG1-1 cells, and parental Sf9 cells, respectively. (a2, b2, and c2) DAPI-stained nuclei of the cells in panels a1, b1, and c1, respectively.

PAT1 functions in establishing persistent viral infection.

The function of PAT1 was originally tested by replacing pag1 of Hz-1 virus with the green fluorescent protein (GFP) gene of the jellyfish Aequorea victoria (7, 8). After cotransfection of the GFP gene-containing transfer plasmid and the viral DNA, the supernatants containing viruses were harvested. Emission of green fluorescence in individual cells could be detected when cells were infected at a high multiplicity of viruses from the supernatant. This observation suggested that recombinant viruses containing the GFP gene were formed. However, green fluorescence was no longer detectable in any cells when the supernatants were highly diluted, suggesting that pag1 is likely to be an essential gene in the life cycle of the virus. In addition, we reported previously that there are at least two other large viral transcripts which traverse the coding region of PAT1 during productive viral infection (10). If one or both of these transcripts are essential for the completion of productive viral infection, then the removal of pag1 from the viral genome will also be fatal to viral replication. This observation may explain why the pag1-deleted Hz-1 virus could not be constructed even in cells stably transfected with pag1.

Because pag1-deleted Hz-1 virus could not be constructed in cells stably transfected with pag1, another approach was taken. pag1 was ligated with a neomycin resistance gene (Fig. 7A) and transfected into Sf9 cells which were free of infection with Hz-1 virus. Sf9 cell clones stably transfected with pag1 were isolated and propagated in monolayers before infection with the Hz-1 virus. Fluorescent in situ hybridization showed that PAT1 was properly expressed and retained in the nuclei of these stably pag1-transfected cells (Fig. 6B, panel b). Most of the Sf9 cells died when infected with Hz-1 virus, leaving only a small percentage of the cells that became persistently infected (Fig. 7B, Sf9). The percentage of persistently infected cells did not increase in cells stably transfected with the neomycin resistance gene alone (Fig. 7B, SfPKN3H and SfPKN4H). However, the number of persistently infected cell clones increased drastically when Sf9 cell clones containing a stably integrated pag1 gene were infected with Hz-1 virus (Fig. 7B, SfPAG1-1, SfPAG1-2, SfPAG2-1, and SfPAG2-2).

FIG. 7.

The pag1 gene functions in the establishment of persistent viral infection. (A) Map of the plasmids as it appeared in this experiment. Arrows indicate the regions of the promoters and their directions of gene expression. Phsp70, hsp70 promoter; Ppag1, pag1 promoter; neo, neomycin resistance gene; pag1, PAT1 coding region; pKSMII, bacterial plasmid vector (Stratagene). (B) Generation of persistently infected clones by Hz-1 virus infection in the parental and stably pag1-transfected cells. Sf9, untransfected host cells. SfPKN3H and SfPKN4H are two cell lines stably transfected with only the neomycin resistance gene. The stably pag1-transfected clones, SfPAG1-1, SfPAG1-2, SfPAG2-1, and SfPAG2-2, were established from different transfection experiments. Sf9 cells transiently transfected with plasmids containing pag1 (pPAGN) were also tested. In these experiments, cells (4 × 10⁴) were challenged with Hz-1 virus, and the numbers of surviving persistently infected cell clones were calculated. Data (means ± standard deviations) were collected from triplicate assays of three independent viral infection experiments. (C) Colonies which were established by the infection of Hz-1 virus in the parental Sf9 cells (a) and the stably pag1-transfected SfPAG1-1 cells (b).

Cloning and propagation of G418-resistant cells that were stably transfected with the pag1 gene required a significant amount of time, and many of the cell clones died during clonal expansion. We were concerned that the death of part of the cells might bias the response of the remaining clones to infection by Hz-1 virus, so the pag1 function was also tested by transient pag1 gene transfection. The results showed that persistently infected cell clones again increased significantly in Sf9 cells transiently transfected with pag1 but not in Sf9 cells transiently transfected with the neomycin resistance gene or in untransfected cells (Fig. 7B and data not shown).

We found that the increase in persistently infected colonies in pag1-containing cells was not merely due to a pag1 function of increasing either colony formation or viral resistance, because our results showed that the colony formation rates were the same between a plasmid containing only a neomycin resistance gene or another plasmid, with further ligation of the pag1 gene (data not shown). Furthermore, cells with or without pag1 in their genomes did not show differences in resistance to Hz-1 virus infection according to assays of yields of viral progeny (data not shown).

To determine if all or only some of the persistently infected cells contain viral DNA and if the clones established by Hz-1 virus infection of pag1-containing cells were persistently virus infected, these cells were analyzed by fluorescent in situ hybridization (Fig. 8). The results showed that in cells productively infected with Hz-1 virus, viral DNA was detectable in essentially all of the cells, and viral antigens were also detectable in many of the cells having undergone productive viral infection (Fig. 8B). In the persistently infected SfP2 cells, viral DNA was detectable in all of the cells (Fig. 8C), whereas viral antigens were not visible (Fig. 8C, inset). Persistently infected clones established by the infection of SfPAG1-1 cells were analyzed after they were mixed and cultured for four passages. Similar to the case for the persistently infected SfP2 cells established by using non-pag1-containing cells, viral DNA (Fig. 8D) but no viral antigen (Fig. 8D, inset) was again detectable in all persistently infected cells established from SfPAG1-1 cells. These results suggest that although upon viral infection the number of clones was increased in cells stably transfected with pag1, all of these cells was persistently virus infected, as were the persistently infected clones derived from the parental Sf cells.

FIG. 8.

Simultaneous detection of viral DNA and antigens in the Sf cells. Cells were hybridized with a 0.3-kb probe derived from the viral genomic HindIII-K fragment and then labeled with an antibody against Hz-1 virus. (A and A′) Uninfected Sf9 cells; (B and B′) Sf9 cells infected by Hz-1 virus; (C and C′) persistently infected SfP2 cells; (D and D′) mixture of cloned cells established by the infection of SfPAG-1-1 cells with Hz-1 virus. (A to D) Images of double DNA and antigen detections; (A′ to D′) bright-field photographs of cells prepared in parallel, corresponding to panels A to D, respectively. Some of the Hz-1 viral DNAs (green) are indicated by arrowheads to assist identification. The host chromosomes (red) were stained with providium iodide. When these two stains overlapped, the viral DNA signal becomes yellow. Viral antigen was stained with Cy5 (pink to purple, depending on the overlapping dyes resulting from the other signals). The insets in panels A to D are the overlapping colors of only two images, propidium iodide and Cy5, to show the positions and intensities of viral antigens in different cells. Numbers 1 to 3 in panel B indicate some of the cells with labeled signals of viral antigen.

DISCUSSION

The pag1 gene of Hz-1 virus was found to transcribe a unique PAT1 RNA. Its lack of significant ORFs, lack of polysome binding, and localization in the nucleus argue that PAT1 may not encode a peptide. So far, PAT1 or related sequences have not been reported to exist in any other viruses, including the type baculovirus Autographa californica nuclear polyhedrosis virus. A computer gene bank search also did not find any other DNA or RNA species which have significant sequence homology to pag1.

Although the PAT1 transcript initiates from position +1, progressive deletions of the downstream promoter region showed that the promoter activity increased drastically when the CAGT motif, a conserved motif found in baculovirus early genes (1, 35), was included (Fig. 4B, positions −727 to +1 versus −727 to +6). This observation suggested that this CAGT motif may play a crucial role in the proper expression of pag1 promoter. Progressive deletion of the upstream promoter region showed that a 90-base region containing putative CAAT, GATA (TTATC), and TATA motifs (Fig. 4C, positions −90 to +29) is also important for promoter activity. Although high levels of LacZ expression were observed for the progressively deleted pag1 promoter constructs containing upstream regions from positions −493 to −158 (Fig. 4C, constructs −493 to +29, −403 to +29, −315 to +29, −212 to +29, and −158 to +29), further increases were observed with a short −90 promoter construct in which all upstream regions were removed (Fig. 4C, −90 to +29). These expression patterns suggest that a repressor and/or activator from host cells (or virus) may play a role in the expression of the pag1 promoter.

Collectively, the involvement of pag1 transcription with the TATA box- and CAGT motif-containing regions and its termination at dinucleotide CA, which is 25 bases downstream from the AATAAA motif, revealed that although pag1 does not have a peptide-coding potential, it is still likely to be transcribed by RNA polymerase II. Furthermore, peptides can be properly synthesized if protein-coding genes, e.g., the luciferase gene and lacZ, are ligated to the pag1 promoter (Fig. 4B and C).

PAT1 is the only detectable viral transcript during persistent Hz-1 virus infection (10). Similar global viral gene shutoff is also observed on latent infection of herpesviruses in mammals, where viral gene expression is limited to the transcription of only one latency-associated gene which gives rise to three nuclear-localized RNAs, the LATs (39, 42). Mutational analysis has demonstrated that LATs are not responsible for the initiation of latent infection (19, 40). Rather, they could be involved in herpes simplex virus type 1 reactivation (13, 29, 40), although contradictory results have also been reported (3, 19, 30). The promoter predicted for LATs is over 660 bases upstream from their 5′ ends, suggesting that LATs may be introns of a larger unstable 8.3-kb RNA which is transcribed only 28 bases downstream from the promoter (14, 48). This suggestion is also supported by the observation that some LAT RNA species contain lariat structures (47).

The pag1 gene is located in a heavily transcribed region, and many other transcripts traverse the PAT1 coding region in the same orientation (10). However, PAT1 is unlikely to be an intron of another longer transcript for the following reasons. (i) Only PAT-1, and no other overlapping long transcripts, is detectable during persistent viral infection (10). (ii) Unlike the case for LATs, the TATA box of the pag1 promoter is only 25 bases upstream from the 5′ end of PAT1. (iii) PAT-1 is readily detectable when a viral EcoRI-M fragment which contains only the promoter and coding region of pag1 is transiently transfected (Fig. 4A) or stably transfected (Fig. 6B) into virus-free cells. Fragment EcoRI-M contains the pag1 promoter but is not long enough to contain the promoters or the transcription start sites of the other transcripts initiating upstream from the PAT1 coding region (10). (iv) The 90-base region directly upstream from the transcription start site, which contains the TATA, CAAT, and GATA motifs, is essential for promoter activity. (v) The cDNA sequence combined from PCR-generated subfragments of PAT1 is the same as that of the genomic DNA (Fig. 3 and data not shown). These results and observations show that the sequence of PAT1 is most likely to be identical to that of pag1, although we can not exclude the possibility that spliced minor PAT1-related RNA species exist.

PAT1 has some similarities with the human X-inactive specific transcript (XIST) (5) and its mouse counterpart (Xist) (4), two recently described genes which map to the X-chromosome inactivation center of mammals. This gene is expressed only from the inactive X chromosome in which the majority of X-linked genes are inactivated (4, 5, 23) and is required for X-chromosome inactivation (27, 33; for a review, see reference 38). The 17-kb human XIST and 15-kb mouse Xist RNAs lack significant ORFs. XIST/Xist RNA is not associated with the translational machinery and is located almost exclusively in the nucleus (4, 5). A unique feature of the XIST/Xist sequence is the presence of several regions comprised of direct tandem repeats. These repeats are conserved in both mice and humans, suggesting that they may have functional significance (5).

It was suggested that the clustered repeats of XIST may serve either as binding sites for nuclear attachment or as a factor for X-chromosome inactivation to occur (5). Xlsirts, a family of interspersed repeat RNAs that contain from 3 to 13 repeat units, is the Xenopus laevis homolog to the mammalian XIST transcript. The Xlsirt RNA repeat sequences were found to be required for translocation of RNAs to the vegetal cortex (25). Currently, we lack information regarding the function of clustered repeats of PAT1. They may function as signals for nuclear localization or retention. Another possibility is that they may serve as domains for PAT1 to bind to the activator protein or to the viral genome. Alternatively, they may be functional domains in the genome of the virus for interaction with PAT1 or may serve as origins for genomic inactivation. Experiments for further elucidating all of these possibilities are in progress.

Although XIST RNA and PAT1 differ in sequence and it is likely that their modes of function are different, similarities between PAT1 and XIST RNA are still evident (Table 1). The sequence homology between XIST RNA and PAT1 was estimated to be 49% by computer analysis. However, the homology is mainly due to the AT-rich nature of these two transcripts, and highly homologous stretches were not found. At present, it is not known whether PAT1 is directly responsible for the establishment of persistent viral infection or only enhances this process. Even if the latter is true, the multifold enhancement of persistent viral infection is significant and warrants further investigation of its mechanisms. In addition, further experiments are necessary to determine whether pag1 functions only in the establishment of persistent viral infection and viral gene shutoff is a later consequence or, alternatively, whether pag1 functions directly in the shutoff of viral gene expression, which then results in persistent viral infection.

TABLE 1.

Comparisons between PAT1 and XIST RNA

Property	PAT1	XIST RNA
Localization	Nucleus	Nucleus
Significant ORFs	No	No
Clustered repeats	Yes	Yes
Expression upon genome inactivation	Yes	Yes
Functions linked to genome inactivation	Yes	Yes
Sequence homologya (%)	49	49
Size (kb)	2.9	15–17

^aSequence homology in a window of 20 bases through a 2.9-kb region between PAT1 and XIST.