The product of ORF O located within the domain of herpes simplex virus 1 genome transcribed during latent infection binds to and inhibits in vitro binding of infected cell protein 4 to its cognate DNA site

Abstract

The partially overlapping ORF P and ORF O are located within the domains of the herpes simplex virus 1 genome transcribed during latency. Earlier studies have shown that ORF P is repressed by infected cell protein 4 (ICP4), the major viral regulatory protein, binding to its cognate site at the transcription initiation site of ORF P. The ORF P protein binds to p32, a component of the ASF/SF2 alternate splicing factors; in cells infected with a recombinant virus in which ORF P was derepressed there was a significant decrease in the expression of products of key regulatory genes containing introns. We report that (i) the expression of ORF O is repressed during productive infection by the same mechanism as that determining the expression of ORF P; (ii) in cells infected at the nonpermissive temperature for ICP4, ORF O protein is made in significantly lower amounts than the ORF P protein; (iii) the results of insertion of a sequence encoding 20 amino acids between the putative initiator methionine codons of ORF O and ORF P suggest that ORF O initiates at the methionine codon of ORF P and that the synthesis of ORF O results from frameshift or editing of its RNA; and (iv) glutathione S-transferase–ORF O fusion protein bound specifically ICP4 and precluded its binding to its cognate site on DNA in vitro. These and earlier results indicate that ORF P and ORF O together have the capacity to reduce the synthesis or block the expression of regulatory proteins essential for viral replication in productive infection.

The herpes simplex viruses 1 and 2 (HSV-1 and HSV-2) cause two kinds of infections. At the portal of entry into the body, these viruses express >80 genes, assemble infectious progeny, and ultimately destroy the infected cells (1). Latent infections usually take place in neurons of dorsal root ganglia. Thus, the virus present at the portal of entry infects nerve endings and is transported retrograde to the neuronal nucleus where it multiplies and destroys the neuron or remains latent (reviewed in ref. 2). In latent state, viral DNA forms an episome and only a small portion of the genome is transcribed. The transcribed domain contained within the inverted repeats flanking the unique long (U_L) sequence is ≈8.5 kbp. In productively infected cells, the antisense strand of this domain expresses two ORFs encoding infected cell protein 0 (ICP0) and γ₁34.5. In latently infected neurons transcripts of the entire 8.5-kbp DNA stretch (major latency-associated transcript) are present in low abundance (3–7). Transcripts of a 1.5- and 2.0-kbp stretch are present in large abundance in nuclei and probably represent stable introns (4, 5, 8–10). The role of the 1.5- and 2.0-kbp transcripts in the establishment of latency is uncertain. Deletion of the promoter or coding domains that give rise to the stable intron RNAs decrease the number of neurons harboring latent virus and decrease the capacity of latently infected neurons to reactivate virus but do not affect the establishment of latency (11–13). In earlier studies from this laboratory we reported that the 8.5 kb of DNA transcribed during latency contains 16 ORFs encoding at least 50 codons, which we have designated ORF A through ORF P. We reported that one of the five ORFs tested, ORF P, situated almost totally antisense to the γ₁34.5 gene, was repressed in productive infection and expressed only under conditions in which the major regulatory protein ICP4 was nonfunctional or if the binding site was mutagenized such that ICP4 could not bind to its cognate site at the transcription initiation of ORF P (14–16). We should note that the promoter of ORF P was discovered by Bohenzky et al. (17), and the existence of the 2.3-kbp mRNA species coterminal with the 8.5-kb latency-associated transcript, encoding what was subsequently shown to be ORF P was described by Yeh and Schafer (18).

At the time ORF P was discovered, we could not reproduciby show the expression of ORF O because its expression was many times lower than that of ORF P. In subsequent studies we discovered that ORF O was expressed, but that the coding domain was smaller and totally overlapped the domain of ORF P (Fig. 1, line 2). Specifically, the nucleotide sequence of ORF O predicts that the initiator methionine of ORF O is located in the TATA box of ORF P. We show that in fact this methionine is not used and that the only methionine in a reasonable location to initiate ORF O translation is that which initiates translation of ORF P, suggesting that ORF O is expressed by a frameshift or editing process within the first 35 codons of ORF P mRNA. We also report that fusion proteins comprising ORF O sequences interact specifically with ICP4 and interfere in vitro with the binding of ICP4 to its cognate site.

Figure 1.

Schematic representations of sequence arrangements of recombinant virus genomes. Lines 1 and 3, representation of the HSV-1 (F) genome. The lines represent unique long (U_L) and short (U_S) sequences that are flanked by inverted repeats ab and b′a′ and a′c′ and ca, respectively, shown as rectangles. Line 2, representation of recombinant virus R7520 (15), which contains the cytomegalovirus (CMV) epitope inserted in the DraIII site, in-frame with ORF O. The rectangular boxes and arrows represent gene domains and direction of transcription of the ORF O, ORF P, and γ₁34.5 genes in the inverted repeat sequence b′a′. The identical sequences in the inverted orientation map in the ab repeat. The closed circle denotes a wild-type ICP4 binding site. Line 5, the corresponding domain of R3659 (16). The StuI–BstEII sequences encoding the ORF O, ORF P, and γ₁34.5 genes were replaced in both repeats by the chimeric α27-tk gene (21). Line 7, the corresponding domain of the recombinant virus R7540. The α27-tk gene of the recombinant R3659 was replaced with sequences containing a mutated ICP4 binding site with a diagnostic EcoRI endonuclease site, and a CMV glycoprotein B (gB) epitope containing a diagnostic EcoRI site in the DraIII (Dr) site in-frame with ORF O. Line 9, the corresponding domain of the recombinant virus R7548. Here the α27-tk of R3659 was replaced with the CMV epitope in the DraIII (Dr) site in frame with ORF O, a mutated ICP4 binding site, and an additional CMV gB epitope inserted in the SacI site, also in-frame with ORF O. The second insertion was within the predicted ORF O gene, upstream of the sequences subsequently shown to be the ORF P coding sequences. Lines 4, 6, 8, and 10, the expected sizes of fragments detected by hybridization of the 1,800-bp NcoI fragment with electrophoretically separated digests of viral DNAs with NcoI–EcoRI (the first band set per virus), diagnostic of the ICP4 binding site mutation; or NcoI–XbaI (the second band set per virus), diagnostic of the CMV epitope tag insertion. The arrows in these lines denote restriction cleavage sites present in the respective viruses and therefore the DNA fragment boundaries. HSV-1(F) would be expected to yield bands A and B, respectively; R3659 would be expected to yield bands C and D; R7540 would be expected to yield bands E, F, and G; R7548 would be expected to yield bands H, J, K, L, and M, respectively. Dr, DraIII; St, StuI; Nc, NcoI; Bs, BstEII; Sc, SacI; Ec¹, the introduced EcoRI site.

MATERIALS AND METHODS

Cells and Viruses.

Rabbit skin and 143tk⁻ cells were originally obtained from J. McClaren (University of Arizona) and Carlo Croce (Thomas Jefferson Medical College), respectively. Vero and HeLa cells were from American Type Culture Collection. HSV-1(F) is the prototype HSV-1 strain used in this laboratory; as is the case with fresh HSV-1 isolates with limited history of replication outside the human host, the α4 gene of HSV-1(F) is temperature sensitive and does not repress itself or ORF P at 39.5°C (19). The recombinants derived from HSV-1(F) used in this study fall into two categories, those in which the ICP4 binding site at the transcription initiation site of ORF P was destroyed by mutagenesis, as described elsewhere (14, 16) (e.g., R7530, R7540, R7548), and those which retain the binding site (15) (R7519 and R7520). In addition, several recombinants were tagged with a sequence encoding 20 codons from the glycoprotein B of human CMV and for which a monoclonal antibody (CH28-2) (20) was available from the Goodwin Cancer Research Institute, Plantation, FL. A list of recombinants used in this study is given in Table 1.

Table 1.

Properties of recombinant viruses

Virus no.*	Designation	ICP4 site	Epitope tag and frame
R3659†		Deleted	None
R7519	CMV–ORF P	Native	BstEII/ORF P
R7520	CMV–ORF O	Native	DraIII/ORF O
R7530	P++/O++	Mutated	None
R7540	P++/O++ CMV I	Mutated	DraIII/ORFO
R7548	P++/O++ CMV 1+2	Mutated	DraIII/ORF O
			SacI/ORF O

^*R7519 and R7520 were described in ref. 15; R7530 was described in ref. 14. For R7540 and R7548, see Materials and Methods. 

^†In R3659 the ORF P/ORF O and γ₁34.5 genes were replaced by the chimeric α27-tk gene (14). 

Plasmids.

Plasmid pRB4794 (15) containing the 1,800-bp NcoI fragment of BamHI S, spanning the region between the start codons of the α0 and γ₁34.5 genes, was used as a probe for analysis of ORF P and ORF O recombinant viral DNA; pRB4791 (15) containing a CMV tag in the DraIII site of BamHI S, in the ORF O reading frame, was used to construct virus R7520; pRB4855 (14) containing a 4-base change in the ICP4 binding site which has been shown to derepress ORF P expression. pRB4923 was made by replacing the BspEI–BstEII fragment from pRB4855 with the corresponding fragment from pRB4791. It was used to construct the virus R7540. pBR5181 contains a CMV tag inserted into the SacI sites of pRB4923, in the ORF O reading frame. The CMV tag sequence was GAAAGGACAAAAGCCCAACCTTCTAGACCGACTCCGACATAGAAAGAACGGGTACCGACACGCAGCT and its complement. It was used to construct virus R7548. pRB5179 contains the 287-bp XmaI fragment from BamHI S inserted in the XmaI site of pGEX-3X (Pharmacia) and was used to purify the glutathione S-transferase (GST)–ORF P_C fusion protein (ORF P amino acids 130–225). pRB5180 contains the 420-bp XmaI fragment from BamHI S inserted in the XmaI site of pGEX-4T3 (Pharmacia) and was used to purify GST–ORF P_N fusion protein (ORF P amino acids −10 to 130). pRB4936 contains the same 420-bp XmaI fragment inserted into the XmaI site of pGEX-2T and was used to purify GST–ORF O fusion protein [amino acids 45 to the C terminus of the HSV-1(F) sequence predicted for ORF O].

Construction of R7540 and R7548.

Recombinant R7540 contains a mutated ICP4 binding site and a CMV tag in the DraIII site of BamHI S, in-frame with ORF O. R7548 contains both of these mutations plus an additional CMV tag insertion in the SacI sites located between the predicted ORF O initiator methionine and the downstream ORF P initiator methionine. The procedure for the construction of the recombinant viruses R7540 and R7548 was similar to that previously described (14, 21, 22). Viral DNA was isolated from infected cells and purified on a 5–20% KAc gradient as described (23). The mutations were verified by sequencing and the recombinant viral DNA was probed for the presence of diagnostic restriction endonuclease sites. In the case of R7540, an EcoRI restriction endonuclease site, diagnostic of the ICP4 binding site mutation, and an XbaI restriction endonuclease site, diagnostic of the CMV epitope insertion (Fig. 1, lines 7 and 8), were present. Digestion of R7540 viral DNA with NcoI–EcoRI resulted in a 930/940-bp doublet (Fig. 2, lane 5, bands E/E′), whereas digestion with NcoI–XbaI yielded DNA fragments of 1,225 and 625 bp (Fig. 2, lane 6, bands F and G). In contrast, digestion of HSV-1(F) viral DNA with NcoI–EcoRI or NcoI–XbaI results in 1,800-bp DNA fragments (Fig. 2, lanes 1 and 2, bands A and B). Digestion of the parental R3659 viral DNA (Fig. 1, lines 5 and 6) produced 700-bp fragments (Fig. 2, lanes 3 and 4, bands C and D) corresponding to the sequence upstream of the thymidine kinase (tk) replacement. As for R7548, an EcoRI restriction endonuclease site, diagnostic of the mutation in the ICP4 binding site, and two XbaI restriction endonuclease sites, diagnostic of the two CMV tag insertions (Fig. 1, lines 9 and 10) were present. Digestion of R7548 with NcoI–EcoRI produced 930- and 1,000-bp DNA fragments (Fig. 2, lane 7, bands H and J), whereas cleavage with NcoI–XbaI produced DNA fragments of 1,050, 255, and 625 bp (Fig. 2, lane 8; bands K, L, and M, respectively). All fragment sizes in Fig. 2 correspond to the predicted patterns shown in Fig. 1.

Figure 2.

Autoradiographic image of electrophoretically separated viral DNA fragments containing sequences in the domain of the ORF O/ORF P/γ₁34.5 genes. Viral DNAs were digested with either NcoI and EcoRI (lanes 1, 3, 5, and 7) or NcoI and XbaI (lanes 2, 4, 6, and 8). They were then subjected to electrophoresis on a 28-cm, 0.85% agarose gel and transferred to a Zeta probe (Bio-Rad) by capillary action in 0.5 M NaOH. The membrane was rinsed in 2× SSC (0.3 M NaCl/0.015 M Na citrate), prehybridized in 30% formamide, 6× SSC, 1% milk, 1% SDS, and 100 μg single-stranded calf thymus DNA per ml for 30 min at 68°C. A total of 10⁶ cpm of denatured, ³²P-labeled pRB4794 was then added overnight and the blot was rinsed as recommended by the manufacturer. Autoradiographic images on Kodak XAR-5 film were overexposed to detect smaller fragments. The expected sizes of the fragments generated by cleavages (bands A through M) are shown in Fig. 1.

Production of Polyclonal Anti-ORF O Antibody.

pRB4936, described above, was transformed into Escherichia coli BL21, and protein was expressed and purified as recommended by the manufacturer (Pharmacia). Two rabbits were inoculated subcutaneously with 1 mg each of purified fusion protein at 14-day intervals, as per the normal protocol at Josman Laboratories (Napa, CA). The sera used in this study were collected two weeks after the final immunization.

RESULTS

ORF O Is Expressed Under the Same Conditions as ORF P.

Earlier studies have shown that ORF P is expressed in cells infected and maintained at 39.5°C, the nonpermissive temperature for ICP4 in HSV-1(F), or maintained at permissive temperatures after infection with mutants in which the ICP4 binding site at the transcription initiation site of ORF P was destroyed by mutagenesis (14–16). The results in Fig. 3 show that the expression of ORF O protein with an apparent M_r of 20,000 is regulated in the same fashion as that of ORF P. Specifically, the tagged ORF O protein was detected only in Vero cells infected with the R7520 (CMV–ORF O) recombinant (Fig. 1, line 2), which contains a CMV epitope in-frame with ORF O, and maintained at 39.5°C (Fig. 3 A, lane 7; B, lanes 6 and 9). Inasmuch as both ORF P and ORF O were tagged with the same amino acid sequence and detected with the mAb directed against the epitope, we conclude that the results shown in Fig. 3A, lanes 7 and 8, also indicate that the ORF O protein is made in quantities far smaller than those of ORF P. To test whether the same ICP4 cognate site was responsible for repression of ORF P and ORF O, we characterized the recombinant virus R7540 (P++/O++, CMV 1), which combines the CMV epitope insertion in-frame with ORF O with the ICP4 binding site mutation previously shown to derepress the expression of ORF P (16). The construction and verification of R7540 (P++/O++, CMV 1) are described in Materials and Methods. Vero cells infected with the R7540 (P++/O++, CMV 1) recombinant and maintained at both permissive (Fig. 3B, lanes 4 and 7) and nonpermissive temperatures (Fig. 3B, lane 10) expressed ORF O. Like ORF P protein (15), the ORF O protein continued to accumulate until at least 22 h after infection. The lower abundance, faster migrating ORF O band is not reproducible and probably represents a degradation product.

Figure 3.

Photograph of infected cell proteins electrophoretically separated on a denaturing polyacrylamide gel and reacted with the CH28-2 mouse mAb to the CMV epitope. (A) Replicate Vero cell cultures grown in 25-cm² flasks were infected with 10 pfu of HSV-1(F), R7519 (CMV–ORF P), or R7520 (CMV–ORF O) per cell, maintained at 37°C for 4 or 18 h (lanes 1–5) or maintained at 39.5°C for 24 h (lanes 6–8), harvested, solubilized, electrophoretically separated on 15% polyacrylamide denaturing gels, transferred to a nitrocellulose sheet, and reacted first with mouse mAb CH28-2 (20) and second with goat anti-mouse IgG conjugated to horseradish peroxidase, and horseradish peroxidase electrochemiluminescent substrate. Blots were exposed to Kodak XAR-5 film for 1 min. (B) Replicate Vero cell cultures grown in 25-cm² flasks were infected with 10 pfu of HSV-1(F), R7520 (CMV–ORF O), or R7540 (P++/O++, CMV 1) per cell, maintained at 37°C for 2, 12, or 22 h (lanes 1–7) or maintained at 39.5°C for 22 h (lanes 8–10), harvested, solubilized, electrophoretically separated on 12.5% polyacrylamide denaturing gels, transferred to a nitrocellulose sheet, and reacted with mouse monoclonal antiserum to CMV gB, then goat anti-mouse IgG conjugated to alkaline phosphatase, and alkaline phosphatase substrate.

The results shown here indicate that the expression of ORF O is repressed during productive infection by ICP4 and that mutagenesis of the ICP4 binding site at the transcription initiation site of ORF P derepresses both ORF P shown earlier (14–16) and ORF O.

The First Methionine of the ORF Is Not the Initiator Methionine of the ORF O Protein.

In this series of experiments, we constructed the recombinant virus R7548 in which the ICP4 binding site at the transcription initiation site of ORF P was destroyed by mutagenesis, and two sequences encoding the CMV epitope were inserted in-frame with ORF O, one at the DraIII site as in R7540 (CMV 1) and one at the SacI sites located between the predicted methionine codon at the 5′ end of the ORF and the initiator methionine codon of the ORF P gene (Fig. 1 line 9). The epitope tag contains no methionine codons in any frame that could alter the translation of downstream sequences. The prediction of this experiment was that the second epitope would increase the apparent molecular weight of ORF O as a consequence of the addition of 19 amino acids to the protein. The lysates of cells infected with wild-type and recombinant viruses were subjected to electrophoresis in a denaturing polyacrylamide gel, transferred to a nitrocellulose sheet, and probed with either the monoclonal antibody to the CMV epitope or the rabbit polyclonal antibody made against the GST–ORF O chimeric protein described in Materials and Methods. As shown in Fig. 4, the rabbit polyclonal antibody specifically reacted with the wild-type, singly and doubly tagged ORF O proteins (Fig. 4A) whereas the mAb reacted only with the tagged proteins (Fig. 4B). ORF O with a single CMV tag (R7540 CMV 1) migrated more slowly than the wild-type ORF O (R7530 P++/O++), consistent with the insertion of a 20 amino acid epitope into the coding region of ORF O (Fig. 4A). The surprising finding was that the electrophoretic mobility of the singly (R7540 CMV 1) and doubly (R7548 CMV 1+2) tagged ORF was identical. The necessary conclusion is that the second tag was inserted into the sequence that is not translated and by extension, that the initiator codon is downstream from the site of insertion of the second CMV tag. The only methionine between the untranslated and translated epitope tags is the initiator methionine of ORF P, suggesting that this is also the initiator methionine of ORF O. The protein then shifts into the ORF O frame by amino acid 35, the site of the expressed CMV tag insertion, by an unknown mechanism.

Figure 4.

Photograph of infected cell proteins electrophoretically separated on a denaturing polyacrylamide gel and reacted with (A) rabbit polyclonal antiserum specific for GST–ORF O or with (B) mAb CH28-2. Replicate Vero cell cultures grown in 25-cm² flasks were infected with 10 pfu of HSV-1(F), R7530 (P++/O++), R7540 (CMV 1), or R7548 (CMV 1+2) per cell, maintained at 37°C for 22 h, harvested, solubilized, electrophoretically separated on 12.5% polyacrylamide denaturing gels, transferred to a nitrocellulose sheet, and reacted with (A) rabbit polyclonal antiserum to ORF O or (B) mouse monoclonal antiserum to CMV gB followed by secondary antibodies conjugated to alkaline phosphatase as described in the legend to Fig. 3. R7540 (CMV 1) and R7548 (CMV 1+2) both have mutated ICP4 binding sites, designated (P++/O++).

ORF O Protein Interacts with ICP4 and Blocks the Binding of ICP4 to its Cognate DNA Site.

The purpose of these studies was to determine whether ORF P or ORF O proteins interact with any viral proteins expressed during productive infection. In the first series of experiments, replicate HeLa cells were labeled with [³⁵S]methionine from 4 to 8 h after infection or mock-infection. The cell lysates were precleared with GST and reacted with either GST–ORF P_C, GST–ORF P_N, or GST–ORF O as described in the legend to Fig. 5. The results (Fig. 5A lane 8) show that only the GST–ORF O chimeric protein brought down a set of labeled proteins with the apparent M_r predicted for ICP4. The bands reacted with the mAb to ICP4 (Fig. 5B, lane 8). GST–ORF O bound the two forms of ICP4 detectable on this immunoblot with relatively equal affinity. In the second series of experiments, anti-ICP4 mAb brought down ICP4 and GST–ORF O but not GST–ORF P fusion proteins (data not shown).

Figure 5.

(A) Autoradiographic image of [³⁵S]methionine-labeled infected cell proteins from total cell lysates or of proteins bound to GST fusion proteins electrophoretically separated on denaturing gels. (B) Photograph of the same gel reacted with monoclonal antiserum to ICP4. Replicate HeLa cells grown in 150-cm² flasks were mock-infected (lanes 1–4) or infected with HSV-1(F) (lanes 5–8) and maintained at 37°C. At 4 h after infection the medium was replaced with mixture 199 lacking methionine but supplemented with 1% calf serum and 50 μCi (1 Ci = 37 GBq) of [³⁵S]methionine. After additional incubation for 4 h, the cells were harvested, solubilized by vigorous pipetting in lysis buffer (10 mM Hepes, pH 7.6/250 mM NaCl/10 mM MgCl₂/1% Triton X-100/0.5 mM phenylmethylsulfonyl flouride/2 mM benzamidine). The lysates were precleared with 20 μg of GST and divided into three identical samples to which were added 10 μg of GST–ORF P_C (lanes 2 and 6), GST–ORF P_N (lanes 3 and 7), or GST–ORF O (lanes 4 and 8) fusion proteins, respectively, and reacted at 4°C for 3 h. Glutathione-conjugated agarose beads were then added to the mixtures and allowed to react for 30 min at 4°C. The beads were collected by low speed centrifugation and washed four times with 50 volumes of lysis buffer. The proteins adhering to the beads were then solubilized, electrophoretically separated on a denaturing 15% polyacrylamide gel, transferred to a nitrocellulose sheet, and exposed to Kodak XAR-5 film for 2 days. (B) Photograph of the blot shown in A after reaction first with mouse mAb specific for ICP4, H1114 (30) (Goodwin Cancer Research Institute), followed by goat-anti-mouse antibody conjugated to alkaline phosphatase.

The purpose of the third series of experiments was to determine whether the ORF O fusion protein which interacts with ICP4 affects the interaction of ICP4 with its cognate DNA sequence (Fig. 6). ICP4 binds to both high affinity sites consisting of a conserved consensus sequence and weak affinity sites for which no clear consensus sequence has been derived (24, 25). We selected for these studies the strong binding site present at the transcription initiation site of ORF P (designated probe) and the corresponding DNA fragment containing a mutagenized ICP4 binding site (designated probeΔICP4bs). Reaction of the probe DNA with nuclear extracts of HSV-1(F)-infected cells yielded a specific ICP4–DNA complex that was supershifted by monoclonal antibody H943 to ICP4 (26) (Fig. 6, lanes 2 and 3, respectively). Concentrations of GST–ORF O greater than 250 ng blocked the binding of ICP4 to the probe (lane 8) whereas 3 μg of GST–ORF P_C had no effect (lane 10). ICP4 did not bind to the mutagenized sequence (lane 12), consistent with earlier results (25, 27). Because GST–ORF O did not bind the DNA probe (lane 9), we conclude that ORF O precludes the binding of ICP4 to DNA rather than competes with ICP4 for the DNA binding site.

Figure 6.

Autoradiographic image of a gel retardation assay showing the interaction of ICP4 with a high affinity DNA site in the presence or absence of GST-ORF O. Nuclear extracts (1.5 μg) were reacted with 2 × 10⁴ cpm of a ³²P-labeled probe DNA in 25 μl of a solution containing 20 mM Tris (pH 7.6), 50 mM KCl, 0.05% Nonidet P-40, 5% glycerol, 1 mM EDTA, 1 mg BSA per ml, 10 mM 2-mercaptoethanol, and 3 μg poly(dI⋅dC). The DNA probes were as follows: the 112-bp MscI–SacI fragment from pRB4794 (13), which is nucleotides −87 to +25 relative to the ORF P transcription initiation site (lanes 1–11), and probeΔICP4bs (lanes 12 and 13), which is the same fragment as above from pRB4855 (14), which contains a mutated ICP4 binding site. The probes were dephosphorylated by shrimp alkaline phosphatase and 5′ labeled with [γ-³²P]dATP using T4 polynucleotide kinase. The ICP4–DNA complex was supershifted by the addition of the mAb H943 to the reaction mixture as originally described by Kristie and Roizman (26). The contents of each reaction mixture is described above the corresponding lane. NE, nuclear extract; GST-ORF P, GST-ORF P_C (Fig. 5). The electrophoretically separated samples in a 4% nondenaturing polyacrylamide gel were dried and exposed to Kodak XAR-5 film for 4 h.

DISCUSSION

The key findings reported here are that (i) a second ORF, designated ORF O and mapping in the domain of the HSV-1 genome transcribed during latency, is expressed only in the absence of a functional ICP4 in cells infected and maintained at nonpermissive temperature, or after mutagenesis of the ICP4 binding site at the transcription initiation site of ORF P; (ii) ORF O protein accumulates in significantly lower amounts than ORF P; (iii) ORF O and ORF P may share the same initiator methionine codon but differ in size and predicted amino acid sequence beginning at an amino acid residue N terminal to amino acid 35; and (iv) in vitro ORF O protein interacts with ICP4 and blocks it from binding its cognate DNA sequence. Relevant to this report are the following.

(i) In our initial studies, ORF O protein was not detected reproducibly in large part because its abundance was much lower than that of ORF P protein. In numerous assays done subsequently, it became apparent that ORF O protein could be readily and reproducibly detected by adjusting the assay conditions to take into account the low abundance of the ORF O protein made in the cell lines tested. ORF O therefore represents the second gene mapping in the inverted repeats flanking the U_L sequences and which is expressed only under conditions in which ICP4 is defective or cannot bind to its cognate site at the transcription initiation site of ORF P.

(ii) The coding domain of the ORF O protein became somewhat of a mystery because the methionine codon at the 5′ terminus of the ORF is upstream of the transcript shown to be derepressed by mutation of the ICP4 cognate site (14, 18). We present evidence suggesting that the methionine codon at the 5′ terminus of the ORF is not the initiator methionine of ORF O protein. The key evidence supporting the hypothesis that ORF P and ORF O may share the same N-terminal methionine is the observation that a tag inserted into the sequence between the amino terminal codons of the two ORFs did not increase the apparent molecular weight of ORF O protein. The first downstream methionine codon is that of ORF P; no other methionine codon exists within the ORF O gene. Insertion of the epitope tag in-frame with ORF O at codon 35 of ORF P yielded a protein of a size consistent with the apparent M_r of ORF O protein tagged with a 20-amino acid sequence. Finally, antibody made against the fusion protein consisting of GST fused to the C terminus of ORF O protein reacted with a protein of the predicted size and found in lysates of cells infected with a virus carrying a mutation at the ICP4 binding site at the transcription initiation of ORF P, but not in lysates of cells infected with wild-type virus and maintained at the permissive temperature. The observation that ICP4 represses ORF O as strongly as it does ORF P is consistent with the hypothesis that the ICP4 binding site is at the transcription initiation site of both genes. The significantly lower levels of accumulation of ORF O protein are consistent with a post transcriptional event such as frameshift or editing of the mRNA encoding ORF O protein. The limited distance between the mapped site of the shift of frames (a maximum of 34 codons, 102 nt) and the absence of consensus splice acceptor/donor sites does not support the hypothesis that the synthesis of ORF O is directed by a spliced mRNA. Parenthetically, we have detected only one derepressed mRNA corresponding to the ORF P transcript (data not shown), this datum is less persuasive than those listed above because ORF O mRNA would be difficult to detect if its abundance reflected that of the protein.

(iii) It could be predicted that if productive infection were to be inhibited at a very early stage, it would be necessary to preclude at least the synthesis or function of the α proteins, particularly ICPs 0, 4, 22, and 27, because these proteins control all subsequent events in the viral replicative cycle. ICP4 is the major regulatory protein of the virus and its expression is required for the expression of both β and γ genes expressed later in infection. ICP4 regulates expression of viral genes both positively and negatively. ICP0 is a promiscuous transactivator required for efficient expression of viral genes, ICP27 regulates posttranscriptional processing of mRNAs, and ICP22 regulates the expression of both early and late gene expression (reviewed in ref. 2). It could also be expected that if HSV-1 encodes proteins responsible for establishment of latency, the expression of these proteins would be repressed during productive infection. ORF P and ORF O meet these predictions in that (a) both genes are repressed during productive infection, and (b) whereas ORF P protein appears to have a role in blocking the synthesis of ICP0 and ICP22 (28), the ORF O protein, at least in vitro under conditions tested, appears to affect the ability of ICP4 to bind its cognate site on HSV-1 DNA.

(iv) The central question whether ORF O and ORF P are either necessary or sufficient to establish latent infections remains unresolved. The experiments designed to determine the function of ORF P and ORF O have been done mostly in cultured cells and using in vitro assays. One problem encountered in in vivo studies is that expression of ORF P and ORF O and of the antisense gene, γ₁34.5, are mutually exclusive (14, 29). Elucidation of the role of these genes in latent infection may require the construction of novel viruses to dissociate the sense from the antisense genes. In addition, several more ORFs within the domain transcribed during latency remain to be explored for their contributions to the establishment or maintenance of the latent state.

ABBREVIATIONS

HSV-1

herpes simplex virus 1

GST

glutathione S-transferase

ICP

infected cell protein

tk

thymidine kinase

CMV

cytomegalovirus