INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The proteins encoded by herpes simplex virus 1 (HSV-1) possess two interesting characteristics. First, their electrophoretic mobility in denaturing gels is usually slower than that predicted by their molecular weight. This property may be related to the relatively high proline content of most HSV-1 proteins. The second property of many protein products of viral genes is that they form multiple bands on electrophoresis through denaturing polyacrylamide gels (28). This property is particularly prevalent among regulatory proteins (e.g., infected-cell protein 0 [ICP0], ICP4, ICP22, ICP27, etc.) (1, 2). The mechanisms underlying the formation of these isoforms is generally unknown. To investigate one case in some detail, we chose the products of the U_L3 gene.

The HSV-1 U_L3 gene is predicted to encode a protein of 235 amino acids reported to contain a signal sequence, a potential hydrophobic domain, and a probable N-glycosylation site (21). The U_L3 open reading frame (ORF) contains six in-frame methionine codons; the first two are 10 codons apart. Ghiasi et al. (12) reported that in baculovirus, U_L3 was unmodified by tunicamycin, that it was present in both cytoplasm and nuclei, and that it formed two bands with M_rs of 27,000 and 33,000 corresponding to nonphosphorylated and phosphorylated forms of U_L3, respectively. It has been also reported that a single mRNA of 2.4 kb spans the U_L3 ORF (29). The attraction of U_L3 for the studies that we had in mind was twofold. First, U_L3 is dispensable for growth in cells in culture (3), and therefore genetic manipulations of the gene would not cause significant selective pressure for emergence of second-site, compensatory mutants. Second, preliminary studies showed that the protein products of U_L3 form at least six bands differing in electrophoretic mobility in denaturing gels.

In this report, we show that (i) U_L3 protein forms at least six bands with apparent M_rs ranging from 34,000 to 42,000, (ii) neither monensin nor tunicamycin affects the electrophoretic mobility of the U_L3 protein isoforms contained in lysates of infected mammalian cells, (iii) the isoforms arise predominantly by translation of the mRNA from the second methionine codon, and (iv) the translation product initiated from the second methionine codon underwent at least two posttranslational modifications. Phosphorylation of the U_L3 protein mediated by the viral U_L13 protein kinase accounted for the slower-migrating protein bands inasmuch as they were absent from lysates of cells infected with U_L13-deficient (U_L13) virus.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and viruses. Vero and HEp-2 cell lines were obtained from the American Type Culture Collection. Virus stocks were grown in HEp-2 cells [HSV-1 strain F {HSV-1(F)}] or Vero cells (mutants), and their titers were determined on Vero cells. All transfections were done in rabbit skin cells originally obtained from J. McLaren. The human 143 thymidine kinase-deficient (143TK) cells, originally obtained from Carlo Croce, were used for selection of tk recombinant viruses. Cell lines were grown in Dulbecco's modified Eagle medium supplemented with either 5% newborn calf serum (Vero and rabbit skin cells) or 5% fetal calf serum (HEp-2 and 143TK cells) supplemented with 40 µg of bromodeoxyuridine per ml of medium for 143TK cells. Infected cells were maintained in medium 199V, consisting of mixture 199 supplemented with 1% calf serum.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   Schematic diagram of the sequence arrangement of the genomes of HSV-1(F) and relevant domains of the recombinant viruses. (A) Construction of recombinant viruses containing a CMV epitope. Line 1, HSV-1 genome. The lines represent unique long (UL) and short (US) sequences, whereas the rectangles represent the inverted repeats ab b'a', a'c', and ca, as marked. Line 2, representation of the Kpn J fragment, which shares portions of the b sequence with the KpnI G fragment. Line 3, positions of UL3 and UL4 ORFs in the KpnI G fragment. Line 4, relevant restriction endonuclease cleavage sites in the KpnI G fragment. Line 5, representation of the 27-tk gene (open rectangle, tk gene; filled rectangle, 27 promoter) and the poly(A) signal (line) inserted into the BamHI site located between UL3 and UL4 to create R7205. Line 6, restriction sites within the R7205 virus and newly created KpnI fragments 1 and 2 detected by a KpnI-HindIII probe shown in line 9. Lines 7 and 8, location of the oligonucleotide encoding the CMV epitope containing a diagnostic KpnI restriction site inserted within the predicted UL3 ORF at the NarI (R8105) and SfiI (R8103) restriction sites. The insertion of the oligonucleotide containing the KpnI site divides the KpnI G fragment into two roughly equivalent fragments predicted in this figure and identified in Fig. 2 as bands 3-4 and 5-6. (B) Construction of the UL3 virus R7211 and its repair virus R8106. Line 10, restriction endonuclease cleavage sites in the KpnI G fragment of HSV-1(F). Cleavage at three EcoRV restriction endonuclease sites within the KpnI G fragment is predicted to result in two EcoRV, fragments O and N, which contain portions of the UL3 ORF. Line 11, chimeric 27-tk gene and the polyadenylation signal inserted into the BamHI site between UL3 and UL4 ORFs in R7205, creating new EcoRV fragments, 7 and 8, detected by an EcoRV-HindIII probe (line 14). Line 12, sequence arrangement of R7211 recombinant virus after the deletion of the UL3 ORF downstream of the EcoRV site and the accompanying tk sequence (fragment 7 in line 11). Line 13, sequence arrangement of R8106 recombinant in which the natural sequences in the UL3 and UL4 regions have been restored to that of HSV-1(F). Line 14, location of the EcoRV-HindIII fragment used as a radiolabeled probe. (C) Line 15, sequence arrangement of HSV-1(F) DNA; line 16, relevant restriction endonuclease sites and EcoRV fragments detected by the BamHI-Q probe shown in line 18; line 17, the EcoRV fragment 9 created in HSV-1(F)305 DNA. Abbreviations: B, BamHI; Bg, BglII; H, HindIII; K, KpnI; M, MseI; N, NarI; R, EcoRV; S, SfiI, Sa, SacII.

Reagents and antisera. Restriction endonucleases were from New England Biolabs; T4 DNA ligase was from United States Biochemicals; the Klenow enzyme and calf intestinal alkaline phosphatase were obtained from Boehringer Mannheim; tunicamycin and monensin were obtained from Sigma Chemical Co. (St. Louis, Mo.)

Construction of plasmids. Plasmid pRB442 contains a 6.3-kbp fragment encoding the U_L1 to U_L4 genes and portions of the U_L5 gene of HSV-1(F) DNA. pRB447, a subclone of pRB442, contains only U_L3 and U_L4 in their entirety, specifically from the XbaI site at nucleotide (nt) 10637 pairs to the HindIII site at nt 12717 in the vector pGEM3Z. Both plasmids are described elsewhere (3).

Production of the GST-U_L3 fusion protein and rabbit polyclonal antiserum. Plasmid pRB5250 was constructed by the in-frame insertion of the EcoRV-ApoI fragment from pRB447 encoding the last 192 codons of the U_L3 ORF into the SmaI-EcoRI sites of the glutathione S-transferase (GST) protein expression vector pGEX2T (Pharmacia Biotech). The junction between the pGEX2T and the U_L3 sequences was sequenced to verify that the coding sequences of the two proteins are in frame (data not shown). Escherichia coli BL21 cells transformed with this vector were grown at 30°C to an optical density of between 0.7 and 1.2 and induced with 0.1 mM isopropyl--D-thiogalactopyranoside for 2 h, and the expressed fusion protein was affinity purified as instructed by the manufacturer (Pharmacia Biotech). Antibody to the eluted chimeric protein was made in rabbits at Josman Laboratories (Napa, Calif.) according to their standard protocol.

Multiple sequence alignment. The amino acid sequence alignment was compiled initially by using the Wisconsin Package (Genetics Computer Group, Madison, Wis.) and then optimized by hand. The U_L3 homologs were from HSV-1 (21), HSV-2 (20), equine herpesvirus (EHV) (30), canine herpesvirus (CHV) (27), bovine herpesvirus (BHV) (16), varicella-zoster virus (VZV) (7), pseudorabies virus (PRV) (8), Marek's disease virus (MDV) (32), and infectious laryngotracheitis virus (10). The consensus line shows residues shared among at least four of the nine U_L3 homologs.

Construction of recombinant viruses. The plasmids and viral DNAs used in the construction of the recombinant viruses described in this report are listed in Table 1. Viral DNAs were prepared from potassium acetate gradients as described elsewhere (13). Recombinant viruses R8103 and R8105 were constructed by cotransfection of R7205 DNA with plasmids pRB4722 and pRB4657, respectively, into rabbit skin cells by the DEAE-dextran method as described elsewhere (23). R8106 was constructed by cotransfection of R7211 DNA with pRB442 plasmid DNA into rabbit skin cells. R8108 was constructed by cotransfection of R8105 DNA with pRB165 plasmid DNA containing the entire BamHI Q fragment. tk progeny were selected on 143TK cells in the presence of bromouracil deoxyriboside. tk⁺ viruses were selected on 143TK cells overlayed with HAT medium, consisting of Dulbecco's modified Eagle's medium supplemented with hypoxanthine, aminopterin, thymidine, and 5% fetal bovine serum, as described elsewhere (22, 23). Viral isolates were plaque purified at least three times on Vero cells, and repair of the tk gene was verified by hybridization.

Southern analyses of viral DNA. Viral DNA was prepared from cytoplasmic extracts of infected Vero cells and purified by phenol-chloroform extraction (EcoRV digests) or potassium acetate gradients (KpnI digests) was digested with EcoRV or KpnI, electrophoretically separated in agarose gels, and transferred to Zeta-Probe blotting membranes (Bio-Rad Laboratories, Hercules, Calif.) overnight by the alkaline blotting method. The DNAs bound to the membrane were hybridized with the nick-translated probe prepared as instructed by the manufacturer (Promega).

Polyacrylamide gel electrophoresis and immunoblotting. Rabbit skin cells in 25- or 150-cm² flasks were mock infected or exposed to 10 PFU of the virus per cell. The cells were harvested, rinsed once with phosphate-buffered saline, solubilized in disruption buffer containing sodium dodecyl sulfate (SDS) and -mercaptoethanol, subjected to electrophoresis on SDS-polyacrylamide gels, and transferred to a nitrocellulose sheet. The electrophoretically separated proteins were blocked in 5% milk in phosphate-buffered saline for 1 h, exposed to antibody for 2 h, rinsed three times for 10 min each in buffered 5% milk, exposed to the secondary antibody for 1 h with alkaline phosphatase-conjugated goat anti-rabbit or anti-mouse serum (Bio-Rad), washed again with buffered saline, rinsed in alkaline phosphatase buffer, and reacted with alkaline phosphatase substrates nitroblue tetrazolium and 5-bromo-4-chloro-3-indolylphosphate (Bio-Rad).

Immunoprecipitation. Replicate 25-cm² cultures of rabbit skin cells were infected with either HSV-1(F) or R7353 (U_L13/U_S3) in medium 199V and then incubated in phosphate-free Eagle minimal essential medium. After 1 h, the cells were incubated in the same medium but supplemented with [³²P]orthophosphate (100 µCi per 25-cm² culture). The cells were harvested at 16 h after infection. U_L3 was precipitated from cell lysates with the HSV-2 U_L3 antibody as described elsewhere (31).

Transfection-superinfection assay. Plasmid DNA from bacterial clones pRB5253 through pRB5259 prepared by the Wizard maxiprep DNA purification system (Promega) were transfected into 25-cm² cultures of rabbit skin cells with the aid of the Lipofectamine Plus system (Life Technologies). Subconfluent cultures were transfected with 3 µg of plasmid DNA and 21 h later infected with the U_L3 virus, R7211. Replicate untransfected cultures infected with HSV-1(F) or R7211 served as positive and negative controls. Cells were harvested 20 h after infection.

In vitro transcription-translation assay. Two different PCR products were generated to link a T7 promoter upstream of the U_L3 ORF. The first PCR product encoding the entire U_L3 ORF was generated by standard techniques and primers 5'-TAATACGACTCACTATAGGGAGACCACATCGAATTCATGGTTAAACC-3' and 5'-GTTTTTCAGTATCTACGATTCATAGATCTCTGCAGGTCGACGGATCC-3' for the 5' and 3' ends, respectively. The second PCR product was generated in the same way except that the sequence for the 5' primer contained a GAG codon instead of the ATG codon underlined above. The resulting PCR products were transcribed and translated by using the TNT coupled reticulocyte lysate system (Promega) and the T7 polymerase according to manufacturer's directions.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Construction of recombinant viruses R8105 and R8103, containing an epitope-tagged U_L3 ORF. To identify the coding sequence of the U_L3 ORF and characterize its product, we inserted into the U_L3 ORF oligonucleotide sequences encoding the CMV epitope recognized by monoclonal antibody CH-28-2. To construct the viruses containing the epitope tags inserted into the middle or at the 3' terminus of the U_L3 ORF, rabbit skin cells were cotransfected with R7205 DNA and plasmid pRB4657 or pRB4722 (Table 1). The recombinant virus R7205, which served as a parent for the construction of the epitope-tagged viruses, contained the chimeric 27-tk gene inserted into the BamHI site between U_L3 and U_L4 (Fig. 1, line 6). In recombinant progeny, the 27-tk gene was replaced with the U_L3 gene carrying the tag. Positions of the tags in the sequence of the U_L3 protein are indicated by the central KpnI site in Fig. 1 (lines 7 and 8).

View larger version (53K): [in this window] [in a new window]   FIG. 2.   Autoradiographic images of electrophoretically separated restriction endonuclease digests of DNAs from HSV-1(F) and recombinant viruses transferred to Zeta-Probe membranes and probed with radiolabeled fragments of the relevant HSV-1(F) domain. The letters and numbers beside the lanes in panels A to C denote the DNA fragments generated by restriction endonuclease cleavages indicated in Fig. 1A to C, respectively. (A) Viral DNA was cleaved with KpnI and probed with a radiolabeled KpnI-HindIII fragment (Fig. 1A, line 9). (B and C) Viral DNA was cleaved with EcoRV and probed with a radiolabeled EcoRV-HindIII fragment (Fig. 1B, line 14) or BamHI-Q fragment (Fig. 1C, line 18). Markers for fragment sizes are indicated for and between panels B and C.

U_L3 protein forms numerous isoforms in denaturing polyacrylamide gels. Electrophoretically separated lysates of rabbit skin cells mock infected or infected with HSV-1(F), R8105 (epitope tagged), or R7211 (U_L3) were reacted with either the monoclonal antibody directed against the CMV tag or the polyclonal antibody made against the GST-U_L3 fusion protein. The products of the U_L3 ORF detected by the anti-U_L3 polyclonal rabbit antibody (Fig. 3) formed at least six bands designated 1 to 6 and characterized by apparent M_rs of 34,000, 35,000, 38,000, 40,000, 41,000, and 42,000, respectively. These bands were absent from lysates of mock-infected cells (Fig. 3A, lane 6; Fig. 3B, lane 3) or of cells infected with the R7211 mutant (Fig. 3A, lane 5; Fig. 3B, lane 4). The U_L3 proteins from lysates of cells infected with the R8105 recombinant (Fig. 3A, lane 3) and detected by the anti-U_L3 antibody migrated more slowly than the wild-type proteins, reflecting the increase in the size of the proteins due to the insertion of the epitope. All of these bands were also detected by the antibody to the CMV epitope (data not shown).

View larger version (42K): [in this window] [in a new window]   FIG. 3.   Photographs of immunoblots of electrophoretically separated lysates of mock-infected or infected rabbit skin cells reacted with UL3 polyclonal antibody. (A) Cells were infected in the presence (lanes 2 and 4) or absence (lanes 1, 3, 5, and 6) of PAA. (B) Cells were mock infected (lane 3) or infected with HSV-1(F) (lane 1), R8106 (UL3 repair; lane 2), or R7211 (UL3; lane 4).

U_L3 isoforms are ₂ proteins. Replicate cultures of rabbit skin cells infected with HSV-1(F) or R8105 were exposed at 2 h after infection to phosphonoacetate (PAA; 300 µg/ml of medium; gift from Abbott Laboratories). This concentration of the drug is sufficient to totally block viral DNA synthesis (14). U_L3 was absent from HSV-1(F)-infected cells treated with PAA (Fig. 3A, lane 2), indicating that U_L3 was regulated as a ₂ gene. The epitope-tagged proteins were also expressed as ₂ proteins (Fig. 3A, lane 4).

U_L3 isoforms are not glycosylated. The presence of the putative N-linked glycosylation site NKS at codon 162 of the predicted sequence of the U_L3 protein prompted us to determine whether this protein is glycosylated in mammalian cells. Earlier studies (12) on U_L3 protein in baculovirus were not compelling inasmuch as they were not done in the context of infected cells and it could be argued that U_L3 protein must associate with another viral protein to undergo glycosylation. In these experiments, replicate cultures of rabbit skin cells exposed to HSV-1(F) for 2 h were either mock treated or overlaid with medium 199V containing 10 µM monensin or tunicamycin (5 µg/ml). The cells harvested 19 h after infection were solubilized, electrophoretically separated on a denaturing gel, transferred to a nitrocellulose sheet, and reacted with polyclonal antibodies to both U_L3 and gM (U_L10). The drugs had no effect on the mobility of U_L3 proteins, whereas as expected the electrophoretic mobility of gM encoded by the U_L10 gene was affected by both tunicamycin and monensin (Fig. 4, lanes 2 and 3). These results do not support the hypothesis that the U_L3 is N glycosylated.

View larger version (68K): [in this window] [in a new window]   FIG. 4.   Photograph of an immunoblot of HSV-1(F)-infected cell lysates probed with antibodies to UL3 and to gM. Lane 1, untreated infected cells; lane 2, cells treated with tunicamycin; lane 3, cells treated with monensin.

U_L13 mediates posttranslational modification of U_L3. The ladder-like appearance of the U_L3 protein bands in denaturing polyacrylamide gels prompted us to test the hypothesis that they represent differential posttranslational modifications mediated by one or both HSV-1 protein kinases (U_L13 and U_S3). In these experiments, electrophoretically separated lysates of rabbit skin cells mock infected or infected with HSV-1(F), R7356 (U_L13), R7041 (U_S3), R7353 (U_L13/U_S3), or R7358 (U_L13R) were transferred to a nitrocellulose sheet and reacted with the U_L3 antibody. The electrophoretic profiles of U_L3 proteins in Fig. 5A indicate that the slowest-migrating three isoforms represent posttranslational modifications mediated by the U_L13 protein kinase. Specifically, isoforms 4 to 6 were absent from the lysates of U_L13 virus-infected cells but were present in cells infected with wild-type virus, U_S3, or the virus in which the deleted U_L13 sequences had been restored.

View larger version (47K): [in this window] [in a new window]   FIG. 5.   (A) Photograph of an immunoblot of electrophoretically separated lysates of cells mock infected or infected with HSV-1(F) or recombinant viruses reacted with the antibody to UL3. (B) Photograph of an autoradiogram of electrophoretically separated UL3 protein isoforms immunoprecipitated from [32P]orthophosphate-labeled HSV-1(F)- or R7353-infected cells with the UL3 HSV-2 antibody.

Analyses of isoforms of U_L3 by tagging the gene at middle or 3' terminus of the ORF. To determine whether the rapidly migrating U_L3 isoforms represent proteolytic cleavage of the translation product, we inserted in frame at two sites in the U_L3 ORF a sequence encoding the epitope of a known monoclonal antibody (Table 1; Fig. 1A, lines 7 and 8). In R8105 and R8108 (repair of tk at its natural locus), the epitope was inserted in frame into the NarI restriction site approximately in the middle of the U_L3 ORF, whereas in R8103 the epitope was inserted in frame into the SfiI restriction site two codons from the 3' terminus of the U_L3 ORF. The electrophoretically separated lysates of cells infected with HSV-1(F), HSV-1(F)305, or the tagged viruses were reacted with the monoclonal antibody to the CMV epitope (Fig. 6, lanes 1 to 6) or the rabbit polyclonal antibody to the HSV-1 U_L3 protein (lanes 7 to 12). The protein products of the U_L3 ORF present in lysates of cells infected with R8103, R8105, or R8108 visualized with the antibody to the epitope tag were identical to each other and to those visualized with the antibody to U_L3. The electrophoretic profiles in lanes 7 to 12 show, however, the expected decrease in electrophoretic mobility caused by the insertion of the epitope tag compared to the untagged proteins (compare lanes 2 to 4 and 8 to 10 with lanes 11 and 12). Inasmuch as the pattern of protein products of the lysates from cells infected with the virus containing the carboxyl-terminal tag (R8103) were identical to those of cells infected with the tag in the center of the U_L3 ORF (R8105 and R8108), these results do not support the hypothesis that the two fastest-migrating forms of U_L3 were products of proteolytic cleavage at or near the carboxyl terminus.

View larger version (46K): [in this window] [in a new window]   FIG. 6.   Photograph of immunoblots of electrophoretically separated lysates of cells mock infected or infected with HSV-1(F) or recombinant viruses reacted with the antibody (Ab) to UL3 or to the epitope tag, as indicated. In addition to the six isoforms of UL3 typically observed in these studies, in this immunoblot a protein migrating faster than isoform 1 was faintly detected by both antibodies.

U_L3 protein is translated largely from the second in-frame methionine codon of the ORF. To identify the initiator methionine, we constructed a series of plasmids containing the U_L3 ORF driven by the 27 promoter. Included in this series were plasmids carrying the substitutions M1E (pRB5253) or M12N (pRB5254) or both M1E and M12N (pRB5256). In addition, plasmids containing single methionine codon substitutions were repaired, to recreate the E1M and N12M codons (pRB5258 and pRB5259), respectively, to ensure that the phenotype observed with these plasmids was a result of the substitution of the ATG codon rather than to an adventitious mutation elsewhere in the plasmid.

View larger version (48K): [in this window] [in a new window]   FIG. 7.   Photograph of immunoblots probed with antibody to UL3 protein. (A) Cells were infected with HSV-1(F) (lane 3) or R7211 (UL3; lane 1) or first transfected with the indicated plasmid and then infected with R7211 (lanes 2 and 4 to 8). The letters and numbers assigned to individual protein bands are described in the text. (B) Immunoblots of UL3 proteins obtained by in vitro translation of UL3 transcripts of plasmids containing an intact M1 ATG (lane 2) or a M1E substitution (lane 3). Lysates of cells transfected and superinfected with pRB5258 and R7211 or infected with HSV-1(F) alone are shown in lanes 1 and 4, respectively.

U_L3 isoforms 1 and 2 result from initiation at the second and first ATG codons of U_L3, respectively. Although the transfection-superinfection assay indicated that the second ATG codon was the preferred initiator codon for the U_L3 protein products, it was not clear whether isoform 3 was a modification of isoform 1 or whether isoform 1 was a cleavage product of form 3, as they were always present in pairs. To determine the answer to this question, the U_L3 ORF was translated in a rabbit reticulocyte lysate system. Two sets of PCR primer pairs were designed so as to allow amplification of the entire U_L3 ORF attached to a T7 promoter. In one set of primers, the first ATG was left intact (PCR product WT [wild type]) whereas in the second set it was substituted (M1E). The protein products of this assay were electrophoretically separated on an SDS-polyacrylamide gel alongside lysates from HSV-1(F)-infected cells, and cells transfected with pRB5258 and infected with R7211. The electrophoretically separated proteins from cell lysates were transferred to nitrocellulose and probed with the antibody to U_L3. The results were as follows.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The ORF designated U_L3 by McGeoch et al. (21) could be expected to yield a protein of 235 amino acid residues with an expected molecular weight of 26,500. The striking feature of the protein products of the U_L3 ORF were the number of isoforms. We have identified at least six isoforms designated 1 to 6 and characterized as having apparent M_rs of 34,000, 35,000, 38,000, 40,000, 41,000, and 42,000, respectively. Worrad and Carradonna (31) demonstrated multiple forms of U_L3 in HSV-2 with apparent M_rs of 28,000, 30,500, and 33,000. HSV proteins frequently migrate in denaturing gels as proteins of higher apparent molecular weights, and moreover, they commonly form multiple isoforms (28). This is especially true of regulatory proteins exemplified by ICP0, ICP4, ICP22, ICP27, etc. (1, 2). The presence of isoforms raise two questions: are the various isoforms the products of posttranslational modification of the product of a single coding domain, and does each isoform perform a unique function not shared with other isoforms? U_L3 appeared to be an ideal target of an initial study largely because the gene is dispensable for viral replication in cells in culture. Attempts to unravel the origin of the isoforms led to the following observations.

(i) The U_L3 ORF was expressed as a ₂ gene in that the gene products were not detected in infected cells overlaid with medium containing sufficient PAA to block DNA synthesis. This finding is in accord with the report of Singh and Wagner (29) that the domain of the U_L3 gene is transcribed late in infection. This observation makes it highly unlikely that the synthesis of the isoforms is differentially regulated.

(ii) Exposure of infected cells to tunicamycin or monensin in concentrations sufficient to block N-linked glycosylation of glycoprotein M had no effect on the electrophoretic mobility of U_L3 protein products.

(iii) U_L3 proteins are posttranslationally modified by U_L13 protein kinase but not by the U_S3 protein kinase. Thus, the slow-migrating isoforms of U_L3 proteins are missing from lysates of cells infected with U_L13 recombinant virus.

(iv) Isoforms 1 and 3 appear to be the products of translation initiated from the second methionine (M12) in the U_L3 ORF. Thus cells transfected with a plasmid carrying the substitution of M12N did not contain either product. We have excluded the possibility that band 1 contains the cleavage products of proteins present in band 3 inasmuch as band 1 was the only product of in vitro transcription-translation of the U_L3 ORF lacking the first methionine (Fig. 7B, lane 3). On the basis of this finding and other data present in this report, we conclude that band 3 contains a posttranslational modification of band 1 protein that is not mediated by the U_L13 protein kinase.

(v) The U_L3 band 2 appears to contain the protein product initiated at the first methionine. This conclusion is based on the observation that it was absent from lysates of cells transfected with the plasmid containing the substitution M1E. It is of interest that band 2 was overexpressed in lysates of cells transfected with wild-type U_L3 ORF driven by the 27 promoter but present in much smaller amounts in cells infected with wild-type parent virus. We cannot exclude the possibility that the products of the first methionine were processed more rapidly than those initiated at the second methionine. We should note, however, that band A was present at significant levels only in lysates containing large amounts of band 2 (e.g., lysates of cells transfected with plasmids containing an intact M12 [Fig. 7A, lanes 2, 5, 7, and 8]). Band A therefore may contain a posttranslationally modified band 2 protein. The necessary conclusion is that in cells infected with wild-type virus, the proteins initiated at the second methionine (M12) accumulate in larger amounts than those initiated at the first methionine.

(vi) None of the U_L3 protein products accumulating in cells infected with wild-type virus are products of proteolytic cleavage at or near the carboxyl-terminal domain of the translated product of the U_L3 ORF. This conclusion is based on the observation that all isoforms of U_L3 protein contained the epitope tag inserted two codons from the 3' terminus of the ORF.

(vii) We have observed several fast-migrating forms of the U_L3 protein in lysates of cells transfected with plasmids lacking the second in-frame methionine codon of the U_L3 ORF. One hypothesis to explain their presence is that they represent products initiated at methionine codons located 3' to M12.

In infected cells, the U_L3 mRNA appears to be preferentially translated beginning with the second methionine codon. The apparent preference for the second codon may be due to the presence of a small upstream ORF (uORF) of five codons (uORF3), located just upstream of the first methionine codon of the U_L3 ORF. uORF3 is in a different reading frame than the U_L3 ORF; moreover, the uORF3 stop codon, TAA, shares a nucleotide with the first methionine start codon, ATG. Numerous examples have been identified in which the reading frame of a small uORF that overlaps another ORF can prevent efficient usage of the first methionine codon of the downstream ORF (reviewed in references 11 and 18). In support of the hypothesis that uORF3 interferes with the utilization of the first methionine codon in the U_L3 ORF is the observation that the first ORF was used very efficiently in cells transfected with plasmids lacking uORF3. Thus, the band corresponding to band 2 of infected cells was the predominant product of the U_L3 ORF contained in pRB5255. It is noteworthy that the DNA sequence of U_L3 in HSV-2 also contains a uORF (predicted to be 17 amino acids in size) whose stop codon TGA overlaps the ATG of the first methionine of HSV-2 U_L3.

In addition to uORF3, other factors may promote the strong usage of the second methionine codon for initiation of protein synthesis. In the context of the viral genome sequence, the first and second methionine codons each have one optimal base in one of the two critical bases 3 and +4 with A of ATG as +1 (reviewed in reference 18). In the environment of the first U_L3 ATG, TTAATGGTT, the 3 nucleotide is a T (poor) and the +4 is a G (good), whereas in the environment of the second U_L3 ATG, GTGATGTCG, the 3 nucleotide is G (good) and the +1 is T (adequate). These environments were retained in the plasmids used for the in vitro transcription-translation or transfection-superinfection assays. Thus, although the exact sequence 5' to the first ATG was altered to GAATTCATGGTT, the 3 nucleotide is still a T (poor) and the +1 remains a G (good). The sequence 3' of the M12 codon, however, does contain an interesting pattern AGAGGGAGTTCCCTCT. Using the underlined nucleotides as unpaired components of a loop, this sequence could form a hairpin 14 nt downstream of the M12 codon. Hairpins in this position, 12 to 15 nt downstream of the AUG codon, have been reported to allow the 40S subunit of the ribosome to stall long enough to recognize and initiate at the second methionine (17). In contrast, the sequence 12 to 15 nt downstream of the first AUG do not contain a sequence consistent with the formation of a hairpin; rather, the sequence containing MseI sites flanking the M1, TTAATGGTTAA, forms a complementary sequence which could form a weak hairpin hiding the first ATG.

Examination of the DNA sequence and predicted amino acid sequences of the U_L3 protein homologs in other members of the Herpesviridae family is consistent with the hypothesis that the first 11 codons of U_L3 may not play a significant coding role. A multiple sequence alignment of U_L3 homologs of members of the Alphaherpesvirinae is shown in Fig. 8. The alignment shows that in both HSV-1 and HSV-2 U_L3 ORFs, the first two methionines are separated by only 10 codons, of which 7 are identical. There is no identity with those of PRV and little (1 of 10) with those of MDV. Rather, a high degree of identity among the homologs of U_L3 begins to be apparent at the sixth amino acid from the second methionine of HSV-1 and HSV-2 (Fig. 8). Furthermore, both MDV and PRV contain ATG codons downstream of the first predicted initiation codon, and these are located prior to the first region in which all viruses show a strong homology. In addition, the sequences encoding the U_L3 homologs for EHV, CHV, BHV, and VZV all begin downstream of the second HSV-1 ATG codon.

View larger version (52K): [in this window] [in a new window]   FIG. 8.   Multiple sequence alignment of the HSV-1- and HSV-2-encoded UL3 ORF and of UL3 homologs of other members of the Alphaherpesvirinae subfamily (EHV [ORF60], CHV [ORF7], BHV, VZV [ORF58], PRV, MDV, and infectious laryngotracheitis virus [ILTV]). The bottom line shows the amino acids shared by at least four of the nine UL3 homologs. Capital letters in other lines indicate that the amino acid is identical to the consensus amino acid for that position; the amino acid position indicated is for the consensus alignment and does not intentionally correspond to that of any individual viral sequence. A highly conserved nuclear localization motif is located at amino acids 211 to 215 (underlined). The highly conserved N-linked glycosylation motif NKS is located at amino acids 183 to 185 (underlined); the CMV tag in R8103 is located near the end of the C terminus (arrow); the CMV tag in R8105 is located just upstream of the N-linked glycosylation site (arrow). Note that the DNA sequence encoding the amino-terminal portion of the UL3 protein homolog in CHV was not available.

In this report, we have not specifically addressed the functions of the various isoforms of U_L3 protein products. We should note, however, that the virus has evolved what appears to be an elaborate mechanism to use predominantly, but not exclusively, M12 as the initiator methionine codon. Moreover, there appear to be at least two independent posttranslational modifications, those mediated by U_L13 protein kinase (bands 4 to 6) and those independent of protein kinase U_L13 (band 3). For heuristic reasons, it is convenient to consider two hypotheses to explain the accumulation of isoforms of the M12 protein products. The first is that posttranslational modifications are sequential and cumulative, leading from band 1 (nascent protein) to band 6 (presumably the most extensively posttranslationally modified). Given that U_L13 protein kinase is responsible for protein accumulating in bands 4 to 6, the amounts of U_L3 proteins in band 3 to 5 would be expected to fluctuate in time and from one experiment to the next. The alternative hypothesis is that the modifications are mutually exclusive and totally dependent on the interactive protein partners. Implicit in the interactive-partner hypothesis is the notion that each isoform has a unique function defined by both the partner and the posttranslational modification it bears. We prefer the second hypothesis based largely on other HSV examples. ICP22, for example, also forms multiple isoforms dependent on both U_L13 and U_S3 protein kinases (26). ICP22 has been shown to interact with several proteins (5, 6). Of particular interest is the observation that only the posttranslationally underprocessed form of ICP22 interacts with a host protein designated p60 (5).