Herpes Simplex Virus Type 1 2-Kilobase Latency-Associated Transcript Intron Associates with Ribosomal Proteins and Splicing Factors

Transcription

1 JOURNAL OF VIROLOGY, Dec. 2001, p Vol. 75, No X/01/$ DOI: /JVI Copyright 2001, American Society for Microbiology. All Rights Reserved. Herpes Simplex Virus Type 1 2-Kilobase Latency-Associated Transcript Intron Associates with Ribosomal Proteins and Splicing Factors MARYAM AHMED AND NIGEL W. FRASER* Department of Microbiology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania Received 4 April 2001/Accepted 19 September 2001 During latency of herpes simplex virus type 1 in sensory neurons, the transcription of viral genes is restricted to the latency-associated transcripts (LATs). The stable 2-kb LAT intron has been characterized previously and has been shown to accumulate to high levels in the nuclei of infected neurons. However, in productively infected tissue culture cells, this unique intron is also found in the cytoplasm. Although deletion mutant analysis has suggested that the region of the gene from which the intron is spliced plays a role in maintenance of latency or in reactivation from latency, no well-defined function has been ascribed specifically to the 2-kb LAT intron. Nevertheless, previous work has shown that it associates with 50S particles in the cytoplasm of acutely infected cells. Our studies tested the ability of the 2-kb LAT to dissociate from cytoplasmic protein complexes under various salt conditions. Results indicated that this association, which had been speculated to be mrna-like, is actually more similar to the affinity of rrnas for translational complexes. Furthermore, by immunoprecipitation analysis, we demonstrate that the 2-kb LAT associates with ribosomal as well as with splicing complexes in infected cells. Our results suggest that the 2-kb LAT is processed similarly to mrnas in the nuclei of infected cells. However, in the cytoplasm, the 2-kb LAT may play a structural role in the ribosomal complex, similar to that of the cellular rrnas, and therefore affect the functioning of the translational machinery. The pathogenic human alphaherpesvirus herpes simplex virus type 1 (HSV-1) causes lifelong latent infections interrupted by recurrent episodes of viral production. The virus initially replicates at the periphery, where it infects nerve endings and travels to sensory ganglia. Once the virus reaches the nuclei of ganglionic neurons, it can establish a latent infection. Upon stress, the viral genome becomes transcriptionally active and reactivation of HSV-1 from latency occurs. In contrast to what occurs in the acute infection, viral transcription during latency is limited. In fact, the diploid gene encoding the latency-associated transcripts (LAT) is the only gene transcribed during the latent state (for reviews, see references 11, 40, and 46). The LAT gene maps to the long terminal repeat regions of the HSV-1 genome, and the most abundant LAT species detected is the 2-kb LAT intron (Fig. 1A and B) (10, 38, 43, 47), which is also expressed during productive infections (43). Interestingly, the subcellular localizations of the 2-kb LAT intron during productive and during latent infections are different. During latency in neurons, the 2-kb LAT intron is found predominantly in the nucleus, whereas during productive infections of tissue culture cells and murine brain stems, the 2-kb LAT is also found in the cytoplasm (13, 32, 43, 47). Since the LATs are abundantly transcribed during latent infections, their role in the establishment, maintenance, and reactivation from latency has been examined extensively. Early studies proposed that the 2-kb LAT is involved in an antisense * Corresponding author. Mailing address: Department of Microbiology, University of Pennsylvania School of Medicine, Philadelphia, PA Phone: (215) Fax: (215) suppression mechanism because it overlaps the 3 end of ICP0 mrna, which expresses a potent and promiscuous transactivator of viral and cellular gene expression (47). Other studies have shown that several LAT deletion viruses exhibit a delayed reactivation phenotype in various animal models, suggesting that LATs play a role in efficient reactivation from latency (3, 18, 23, 45, 50). Work by Sawtell and Thompson suggested that LATs play a role in promoting efficient establishment of latency in trigeminal ganglia (42). It has also been proposed that LATs may facilitate the establishment of latency by reducing productive viral gene expression (12, 25). Most recently, experiments have suggested that LAT promotes neuronal survival after HSV infection by reducing apoptosis in infected cells (37). This antiapoptotic phenotype of LAT may ensure that latent infection is maintained and allow for the efficient reactivation of the virus under conditions of stress (37). Although the 2-kb LAT is an intron (10, 54), analysis of its sequences indicates that there are two potential open reading frames which are conserved among the different HSV-1 strains (44). However, as of yet, no protein product has convincingly been ascribed to any of the LATs in vivo. Using a combination of biochemical and molecular biology techniques, the potential of the 2-kb LAT as a substrate for translation was examined by determining its association with ribosome-sized particles in sedimentation experiments (32). Results indicated that the majority of the 2-kb LAT intron comigrates with ribosome-sized subunits rather than polysome-sized complexes in productively infected tissue culture cells (32). These data support previous data (44) indicating that the 2-kb LAT is not efficiently translated during productive infections and show that the 2-kb LAT is in association with translational machinery-sized particles 12070

2 VOL. 75, 2001 INTERACTION OF LAT WITH RIBOSOMES AND SPLICING FACTORS FIG. 1. HSV-1 latency-associated transcripts. (A) Linear map of the HSV-1 genome with its unique long (UL) and unique short (US) regions flanked by inverted repeat (IR) elements. (B) LAT region of the HSV-1 genome. The LAT region is enlarged to show the different LAT transcripts that map to this area, as well as the other RNAs (L/ST s, ICP0, ICP4, ICP34.5, UL54, UL55, UL56). The minor LAT (mlat), the putative 8.5-kb primary transcript, and the potential spliced exons are shown (including 2-kb LAT intron). (C) The location of the BstEII region of plasmid pcdna3 Pst-Mlu, used as a probe for the 2-kb LAT, is shown. during productive infection of tissue culture cells. Thus, the possibility exists that the 2-kb LAT is translated during alternative cell conditions or that it is involved in a novel translational role with ribosomes. Other studies have shown that a portion of the 2-kb LAT is found in polysomal fractions of cell extracts from latently infected ganglia and infected neuronal cells (13). Therefore, if the 2-kb LAT is translated, its translation is probably tightly regulated during the virus infection. Until a polypeptide is identified, other possibilities for the function of the 2-kb LAT must be considered. For example, it is conceivable that the LATs may affect the translational machinery of cells or may associate with cellular factors to modify their functions. Since the 2-kb LAT localizes preferentially to different compartments during latency and productive infection, it is possible that in the nucleus versus the cytoplasm, LAT may associate with different factors to mediate this localization. In eukaryotic cells, essential posttranscriptional processes are mediated by RNA-binding proteins and by small RNAs as stable ribonucleoprotein (RNP) complexes found both in the nucleus and cytoplasm (4, 7, 8, 16, 22, 48). In addition, splicing of mrna to generate mature RNAs is mediated by protein factors on the introns of protein coding gene transcripts to form spliceosomes (54). Because the 2-kb LAT is an intron, it is likely to interact with components of the spliceosomes during its formation. In addition, it is possible that LAT functions through its interaction with RNPs that are involved in splicing, transport, and other processing pathways in the nucleus. Such cellular interactions with RNA-binding proteins may be important to the functions of the 2-kb LAT during the viral life cycle. However, in the cytoplasm of infected cells, LAT sediments at approximately 50S, a size corresponding to translation initiation complexes (32). Therefore, it is also possible that LAT interacts with ribosomal components in the cytoplasm for translation, or alternatively, to function in the control of the translation process. In this study, we examined the ability of the 2-kb LAT intron to associate with cellular proteins in order to gain a greater understanding of the role of this stable intron during the virus infection. Taking advantage of methods to separate subcellular compartments, the localization of the 2-kb LAT intron during productive infections was examined. The data indicate that during productive infections of HSV-1 in HeLa cells, LAT is distributed throughout the cell, including membrane and nucleolar fractions. However, the major fraction of LAT was found in the nucleoplasm. The fraction of the 2-kb LAT that was found in the cytoplasm appears to interact with ribosomesized complexes with an affinity resembling that of rrnas, rather than that of actively translating cellular or viral mrnas. Furthermore, immunoprecipitation analysis with antibodies to ribosomal proteins revealed that LAT directly associates with ribosomal proteins. These results support the hypothesis that LAT may affect the functioning of the translational machinery or play a structural role in the ribosomal complex. MATERIALS AND METHODS Cell lines and viruses. HeLa cells were grown in Dulbecco s modified Eagle medium supplemented with 10% calf serum. HSV-1 strain 17 was used in all experiments involving virus infections.

3 12072 AHMED AND FRASER J. VIROL. Antibodies. The ribosomal P antibody, HP (Immuno Vision, Springdale, Ark.), recognizes a 38-kDa protein in the human 60S ribosomal subunit. The ribosomal L7/SPA polyclonal antibody (GeneTex, San Antonio, Tex.) recognizes a 27-kDa protein that is also associated with the 60S subunit. To detect particles associated with the spliceosome, the monoclonal Y12 anti-sm antibody (Neomarkers, Fremont, Calif.) was used (55). The anti-rnp antibody (Inno- Genex, San Ramon, Calif.) was used to detect RNP particles in both the nucleus and cytoplasm. The control antibody, anti-myelin basic protein (MBP) was obtained from Zymed (South San Francisco, Calif.). Preparation of 32 P-labeled probes. The 2-kb LAT probe, a 1.0-kb BstEII- BstEII DNA fragment, was generated as previously reported (56) and diagrammed as shown in Fig. 1C. Briefly, it was derived from the pcdna3.pst-mlu plasmid which expresses the 2-kb LAT intron as well as portions of exon 1 and 2. However, this probe is specific for the 2-kb LAT. The pa plasmid contains a 7.3-kb fragment encoding most of the human 28S rdna gene (9). The pa plasmid was digested with BamHI, and the resulting 1.4-kb fragment containing 28S rdna sequences was subcloned into the pgem-3z vector (Promega, Madison, Wis.) to generate the plasmid pgem-28s. The 0.9-kb 28S rdna probe was produced by digesting pgem-28s with BamHI and BglII. The 0.24-kb human -actin probe was generated by PCR amplification of HeLa genomic DNA with primers 5 TACATGGCTGGGGTGTTGAA3 and 5 AAGAGAGGCATCCTC ACCCT3 (34). The HSV-1 gc fragment was PCR amplified using primers previously described (49). Amplified bands and restriction-digested bands were gel isolated and purified with Geneclean II (Bio 101, Inc., Carlsbad, Calif.). DNA fragments were radiolabeled with 32 P using the Rad Prime DNA labeling kit (Gibco-BRL, Grand Island, N.Y.) for detection of the 2-kb LAT, 28S, gc, and -actin mrnas in dot blots. Preparation of cellular extracts. The procedure for the preparation of the cytoplasmic extract, outer nuclear membrane fraction, nucleoplasm, and extract of the nuclear pellet was performed as previously described (17). Basically, cells were mock infected or infected with HSV-1 strain 17 at a multiplicity of infection of 3 PFU/cell. At 16 h postinfection, cells were harvested and centrifuged at 600 g for 5 min. Cells were resuspended in ice-cold EBKL 0.1% NP-40 buffer (25 mm HEPES [ph 7.6], 5mM MgCl 2, 1.5 mm KCl, 2 mm dithiothreitol (DTT), 1 mm phenylmethylsulfonyl fluoride, 4 g aprotinin per ml, and 0.1% NP-40). The cells were then lysed on ice in a Dounce homogenizer (30 tight strokes), and the nuclei were removed by spinning at 600 g for 5 min. The supernatant is the crude cytoplasmic extract. The nuclei were washed in EMBK buffer (25mM HEPES [ph 7.6], 5 mm MgCl 2, 1.5 mm KCl, 75 mm NaCl, 175 mm sucrose, 2 mm DTT, and protease inhibitors) and then washed in EMBK buffer containing 0.5% NP-40. The supernatant from this step was the outer nuclear membrane wash fraction. The nuclei were resuspended in EBKL (0.1% NP-40) and incubated for 10 min and then lysed by the dropwise addition of KCl to 0.2 M final concentration. The lysed nuclei were incubated with DNase for 15 min at 37 C and pelleted at 10,000 g for 10 min. The supernatant (nucleoplasm) was removed, and the pellet containing chromatin, nuclear membranes, and nucleolar material was sonicated in EBMK 0.5% NP-40 buffer, followed by centrifugation at 10,000 g for 10 min. The resultant supernatant was called the extract of the nuclear pellet. Isolation of ribosomal complexes. Cytoplasmic extract was prepared as published previously (20). Briefly, mock-infected and HSV 17 -infected cells were harvested and resuspended in buffered saline (5 mm D-glucose, M NaCl, 5 mm KCl, 7.5 mm MgCl 2, 10 mm HEPES [ph 7.2]). Cells were pelleted at 2,000 rpm for 10 min, and the pellet was washed twice in buffered saline. After the final centrifugation, 1.5 volumes of ice-cold water was added to cells and mixed thoroughly. After a 10-min incubation on ice, the lysate was centrifuged at 10,000 rpm for 20 min to pellet membranes and other cytoplasmic organelles, and the supernatant was collected. Ribosomes were obtained by centrifugation of the cytoplasmic extract at 200,000 g for 4 h. The supernatant from this spin, containing smaller particles in the cytoplasm, is referred to as the supernatant. The pellet was resuspended in the minimum possible volume of low-salt buffer (20 mm Tris-HCl [ph 7.5], 50 mm KCl, 50 mm NaCl, 1 mm EDTA, 1 mm DTT, 2% glycerol, and 1% Triton X-100), or high-salt buffer (20 mm Tris-HCl [ph 7.5], 50 mm KCl, 450 mm NaCl, 1 mm EDTA, 1 mm DTT, 2% glycerol, and 1% Triton X-100). The ribosomes were incubated in the high-salt (500 mm salt) or low-salt (100 mm salt) buffers for 2honiceandwere centrifuged for 50 min at 350,000 g. The supernatant from this spin consists of RNA and protein that have dissociated from the translational complexes, and the pellet contains larger, intact translational complexes. A 200- l aliquot of each (supernatant, dissociated particles, and ribosomal complexes) was mixed with 120 l of 37% formaldehyde, 80 l of 20 SSC (3 M NaCl and 0.3 M sodium citrate [ph 7.5]), and 400 l of deionized formamide. Following a 15-min incubation at 60 C, the mixture was spotted onto a nylon membrane using a dot blot apparatus from Schleicher & Schuell (Keene, N.H.). Dot blots were cross-linked in a Stratalinker (Stratagene, La Jolla, Calif.) and prehybridized as in the Northern blot analysis previously described (43). Hybridization was performed overnight with heat-denatured 32 P-labeled DNA probes for the 2-kb LAT, 28S rrna, and -actin mrna, and blots were washed twice in 1, 0.5, and 0.1 SSPE (1 SSPE is 180 mm NaCl, 10 mm monobasic sodium phosphate [ph 7.7], 1 mm EDTA) with 0.1% sodium dodecyl sulfate (SDS). Filters were exposed to autoradiographic film and were quantitated using phosphorimaging (Molecular Dynamics, Piscataway, N.J.). Immunoprecipitation experiments. Cell lysates for immunoprecipitation studies were prepared by washing mock- or HSV-1-infected HeLa cells once with cold phosphate-buffered saline A. Cells were then washed once with cold Trisbuffered saline (40 mm Tris-Cl [ph 7.4], 150 mm NaCl), resuspended to approximately cells/ml in NET-2 buffer (50 mm Tris-Cl [ph 7.4], 150 mm NaCl, 0.05% Nonidet P-40) at 4 C, and sonicated three times for a total of 1 min. Cellular debris was removed by centrifugation at 14,000 g for 15 min, and the supernatant was stored at 70. For each immunoprecipitation, 100 l of lysate was used. Specific antibody was added to the lysate and incubated at 4 C with gentle agitation on a rotator for 2 h. Protein A-Sepharose (Sigma, St. Louis, Mo.) was used to precipitate the RNP, P, L7, and the MBP antibodies, while protein G-agarose (Sigma) was used precipitate to the Sm antibody. The lysate/antibody complex was incubated with protein A-Sepharose or protein G-agarose, as specified by the manufacturer of the antibody, for 1 h at 4 C, followed by washing six times in radioimmunoprecipitation assay buffer (0.15 M NaCl, 1% deoxycholate, 1% Triton X-100, 0.1% SDS, 10 mm Tris [ph 7.4]) plus 2% SDS. Proteins were eluted from the agarose or Sepharose by adding 2 volumes of Na 2 HPO 4 and citric acid solution (ph 3.5) and incubating for 15 min at room temperature on a rotator. The beads were pelleted, and the supernatant was neutralized with the dropwise addition of 5 N NaOH as determined by litmus paper testing. The supernatant was then blotted onto a nylon membrane and hybridized with probes specific for the 2-kb LAT or the 28S ribosomal RNA as previously indicated. To determine the nonspecific hybridization background, cellular lysates were incubated with protein A-Sepharose or protein G-agarose in the absence of antibody. RESULTS The 2-kb LAT RNA is found predominantly in the nucleoplasm of productively infected HeLa cells, although it is also distributed in the cytoplasm, membrane, and nucleolar fractions. During latent infections in the trigeminal ganglia, the 2-kb LAT is found predominantly in the nucleus (43, 45, 47). However, previous studies have suggested that during productive infections, the major subset of the 2-kb LAT RNA is found in the cytoplasm of infected cells (32), where it may be interacting with components of the translational machinery. In order to get a clearer idea of the types of proteins that may be binding to LAT, the subcellular localization of the 2-kb LAT in productively infected tissue culture cells was examined in detail. Mock-infected and HSV-1-infected HeLa cells were harvested at 16 h postinfection, and cells were separated into the cytoplasm, outer nuclear membrane, nucleoplasm, and an extract of the nuclear pellet (which contains chromatin and nuclear membranes as well as the nucleolar material), as described in Materials and Methods (17). An equivalent cellular amount of each fraction was blotted onto a nylon membrane, and the 2.0-kb LAT RNA, 28S ribosomal RNA, and -actin mrna were detected with 32 P-labeled probes specific for each of the RNAs. Data was quantitated and expressed as percentages of total cellular RNA levels. Figure 2A shows the distribution of cellular RNAs in uninfected HeLa extracts. As expected, the majority of -actin mrna (50%) was found in the cytoplasm, where it was undergoing translation. Less than 20% of total -actin mrna was found in the outer nuclear membrane and the pellet frac-

4 VOL. 75, 2001 INTERACTION OF LAT WITH RIBOSOMES AND SPLICING FACTORS FIG. 2. The subcellular localization of the 2-kb LAT, 28S rrna, and -actin mrna. HeLa cells were mock infected (A) or infected with HSV-1 strain 17 (B) at a multiplicity of infection of 5 PFU/cell. At 16 h postinfection, cells were harvested and separated into the cytoplasm, outer nuclear membrane (ONM), nucleoplasm, and extract of the nuclear pellet. An equal aliquot of each fraction was dot blotted onto a nylon membrane and hybridized with probes specific for the 2-kb LAT RNA, 28S rrna, and -actin mrna. Data was quantitated by both densitometry and phosphorimaging and expressed as percentages of total 2-kb LAT, 28S, or -actin RNA in the cell. tions. The presence of the RNAs in the membrane fraction may be indicative of their migration from the nucleus to the cytoplasm. The 28S rrna was equally distributed in the nucleus and cytoplasm in uninfected cells (30% each). Although the majority of 28S rrna was found in the cytoplasm and nucleoplasm fractions, about 20% of 28S rrna was also found in the extract of the nuclear pellet, which contains nucleolar material. This is anticipated since the 28S ribosomal RNA is synthesized in the nucleoli, where it is assembled with ribosomal proteins as an integral part of the mature translational complex (for a review see reference (39). The levels of -actin mrna in HSV-infected cells were significantly depressed (data not shown) due to the shutoff of host transcription during the viral infection (14), and a greater percentage of ( -actin) mrna was now detected in the nucleus (35% in uninfected cells to 55% in infected cells, as shown in Fig. 2). This decrease in cellular mrna levels and the nuclear restriction of cellular mrnas have been observed previously and may be due to the impairment of host cell splicing by the viral ICP27 protein (15, 41). Similar to the cellular mrna retention in the nucleus, studies have shown that rrnas are also retained in the nucleus at later times postinfection (2). However, although there is a slight increase in the nuclear 28S rrna in infected cells compared to uninfected cells in our experiments, in contrast to -actin mrna levels and distribution, the 28S ribosomal RNA is not altered significantly (our data and reference (14). These data indicate that at 16 h postinfection, the 28S rrna is relatively stable in

5 12074 AHMED AND FRASER J. VIROL. infected cells since it is necessary to maintain the integrity of the ribosomal complex for viral mrna translation. We found that the majority of the 2-kb LAT in acutely infected tissue culture cells (HeLa) localized to the nucleus (Fig. 2B). In fact, our results show that approximately half of total LAT was found in the nucleoplasm alone. Since previous work dissecting the localization of LAT in CV-1 cells does not clearly quantitate the localization of LAT in the cytoplasm versus the nucleus (32), this is the first work establishing quantitation of LAT found both in the nucleus and cytoplasm in productively infected tissue culture cells. A significant portion of the 2-kb LAT was also found in the cytoplasm and extract of the nuclear pellet (25 and 20% of total LAT, respectively). Previous studies have shown that LAT migrates at 50S in sucrose density gradients, which corresponds to translation initiation complexes in the cytoplasm (32). Therefore, LAT may potentially interact with members of the translational machinery in the cytoplasm. In addition, since 20% of LAT was found in the pellet fraction in infected cells, it is possible that LAT migrates to the nucleoli and interacts with the ribosomal complexes that are being assembled in that compartment. In fact, Fig. 2 shows that the nuclear and cytoplasmic distribution of LAT was more similar to that of the 28S rrna than it was to that of the -actin mrna. Therefore, if LAT plays a role in infected cells, these data would suggest a structural role for LAT that is similar to that of rrnas during assembly of the ribosomal complexes. The 2-kb LAT RNA associates with ribosomal subunits with the affinity of rrnas an association that has greater affinity than that of cellular mrnas and actively translating viral mrnas. The 28S rrna plays a critical role in the biogenesis of the translational complex and is an integral component of the ribosomal complex (52). If the 2-kb LAT is similar to the 28S rrna in terms of its interaction with ribosomal complexes, perhaps it affects the function of the translational machinery. Therefore, we determined the affinity of LAT with translational complexes and compared it to that of the 28S rrna and -actin mrna (Fig. 3). For these experiments, cells were mock infected or infected with HSV-1 strain 17 virus. At 16 h postinfection, cells were harvested and ribosomal complexes were obtained by high-speed centrifugation (5). The ribosomes were then resuspended in buffers containing 100 mm or 500 mm salt, spun at high speeds, and both supernatant and pellet were collected. Buffer containing 100 mm salt was referred to as the low-salt buffer since it is closer to physiological salt levels, and the buffer containing 500 mm salt was referred to as the high-salt buffer in these experiments. The supernatant collected after the high-speed spin consists of particles, either RNA or protein, that have dissociated from the translation complexes after the salt washes, whereas the pellet contains RNA and protein that are maintained as part of the translational complex. The presence of the 2-kb LAT RNA and cellular RNAs (28S and -actin RNAs) in the supernatant before the salt washes and in the supernatant and pellet after the salt washes was quantitated by dot blot analysis as described in Materials and Methods. In uninfected cells (Fig. 3A), at 100 mm salt conditions, approximately 70% of -actin mrna dissociated from the translational complexes in the pellet, while 20% was associated with these complexes. However, under these conditions, close to 65% of the 28S rrna is found in the pellet. Since the 28S rrna is a key component of the ribosomal complex, it is expected that the affinity of the 28S rrna for ribosomes is greater than that of mrnas at physiological conditions. At higher salt concentrations (500 mm), the interaction of -actin mrna with the translational machinery was similar to that at the lower-salt conditions. Under these same conditions, a higher percentage of the 28S rrna was dissociated from translational complexes, leaving 40% of 28S bound to the complexes. However, the affinity of the 28S ribosomal RNA for the translational machinery was greater than that of -actin mrna at both salt conditions. As mentioned earlier, during HSV-1 infections cellular mrna synthesis, including that of -actin mrna, is severely depressed (14). Of the -actin mrna that did remain in the cytoplasm, 50% did not associate with ribosomes or the translational machinery and was found in the supernatant fraction (Fig. 3B). Although the remaining 50% of -actin could be spun down with heavier complexes, it was no longer associated with the intact translational machinery. However, it is possible that -actin still interacts with individual ribosomal proteins or components of the translational machinery. These data indicate that during HSV-1 infection, -actin mrna does not have access to the cellular translation apparatus and protein synthesis is impaired. In contrast, close to 50% of the 28S rrna remains associated with the intact translational machinery, reinforcing the idea that maintenance of efficient translation is critical for viral RNA expression. At least 55% of LAT RNA was dissociated from the translational machinery under these conditions, while 30% remains bound to the translational apparatus. However, the majority of LAT can be spun down with the translational complexes, indicating that at salt conditions close to physiological levels, LAT RNA is associated with the translational machinery. Under stringent conditions (500 mm salt), the profile of -actin mrna levels was similar to those at the lower-salt conditions. However, a greater percentage (70%) of the 28S rrna was dissociated from the translational complexes, while 20% remained bound. Surprisingly, the affinity of LAT for these complexes did not change significantly from the low-salt conditions. These results also demonstrate that the interaction of LAT with ribosomal complexes is as stable as the interaction of the 28S rrna with these complexes, since high-salt conditions are not able to disrupt this association. In contrast, cellular mrnas are not as tightly bound to the translational complexes in infected cells. Thus, the affinity of LAT for translational complexes is comparable to that of -actin mrna in uninfected cells. However, in infected cells, the interaction of LAT for the ribosomal complexes is similar to that of rrnas and not to cellular mrnas. Since our data indicate that LAT has a greater affinity for translational complexes than do cellular messages in infected cells, we wanted to test whether this was also the case for viral mrnas (Fig. 4). As mentioned earlier, studies from our lab have shown that the 2-kb LAT comigrates with ribosomal subunits on sucrose density gradients (32). In contrast, the viral glycoprotein C (gc) mrna sediments with polysomes during productive infection of CV-1 cells. Furthermore, LAT distribution is unaffected by EDTA or puromycin treatments, whereas gc mrna is shifted to lower-molecular-weight com-

6 VOL. 75, 2001 INTERACTION OF LAT WITH RIBOSOMES AND SPLICING FACTORS FIG. 3. Affinity of 2-kb LAT, 28S rrna, and -actin mrna for translational complexes. HeLa cells were mock infected (A) or infected with HSV-1 strain 17 (B) at a multiplicity of infection of 5 PFU/cell. At 16 h postinfection, the cytoplasmic fraction of the cells was subjected to low-salt (100 mm) or high-salt (500 mm) washes as described in Materials and Methods. Supernatant, fraction collected before the salt washes; salt-sup, supernatant fraction that was collected after each of the salt washes (this fraction contains particles [RNA and protein] that have dissociated from the translational complex after the salt washes); salt-pellet, fraction containing the intact translational machinery after the salt washes. An equal aliquot of each fraction was dot blotted onto a nylon membrane and hybridized with probes for the 2-kb LAT, 28S rrna, and -actin mrna. Results were quantitated by densitometry and phosphorimaging and expressed as percentages of total 2-kb LAT, 28S, or -actin RNA in the cell. plexes or ribosomal subunits. These results suggest that unlike gc, LAT may not be efficiently translated during the virus infection and that the interaction of LAT with cellular factors may be distinct from that of viral mrnas. Therefore, the experiment described in Fig. 3 was repeated, but fractions were probed for the gc mrna to determine whether the affinity of LAT for these complexes would resemble that of actively translating viral mrnas. gc is found on the surfaces of virions and infected cells during the viral infection. Results depicted in Fig. 4 show that at low-salt conditions, gc binds the ribosomal complexes with an affinity closely resembling that of LAT. Approximately 35% of the RNAs were found associated with the intact translational machinery, whereas greater than 50% were dissociated. However, under stringent conditions, we can see that 90 to 100% of gc can be dissociated from the translational machinery, whereas LAT levels stay stable. Therefore, LAT has a greater affinity for ribosomal complexes than does the gc viral mrna. In fact, all results indicate that the interaction of LAT with ribosomal complexes is more similar to that of the 28S rrna compared to both cellular mrnas and translating viral mrnas. These data again support the hypothesis that LAT may play a role in the cell that resembles that of the 28S rrna. The 2-kb LAT RNA interacts with ribosomal proteins and splicing factors in vivo. Our studies indicate that LAT associates strongly with members of the translational complex in the cytoplasm (Fig. 3 and 4). However, it has not been shown which cellular complexes, or specific proteins, interact with LAT. In vitro experiments done in our lab indicate that the 2-kb LAT RNA binds cellular proteins in both the nucleus and

7 12076 AHMED AND FRASER J. VIROL. FIG. 4. Affinity of the 2-kb LAT and gc mrna for translational complexes. HeLa cells were infected with HSV strain 17 virus for 16 h and subjected to low-salt (100 mm) and high-salt (500 mm) washes as indicated for Fig. 3. Supernatant, fraction collected before the salt washes; salt-sup, supernatant collected after the salt washes; salt-pellet, fraction containing the translational complexes that sediment at high speeds after the salt washes. Results were again quantitated by densitometry and phosphorimaging and expressed as percentages of total 2-kb LAT, 28S, or -actin RNA in the cell. the cytoplasm (M. Ahmed et al., unpublished data). However, it is not know which proteins are involved in these interactions, the major limitation being the inability to in vitro transcribe LAT RNA while maintaining its stable intron structure. However, these studies suggested that the 2-kb LAT associates with general RNA-binding proteins in infected cells. To directly determine the interaction of LAT with proteins in vivo, we performed an immunoprecipitation analysis (Fig. 5) using total cell extract from HeLa and SY5Y cells. Due to the technical limitations of our assay, we were unable to quantitate the percentage of LAT, 28S, and gc RNA in each compartment that was recovered in the immunoprecipitates with the various antibodies. Nevertheless, we were able to get an indication of the overall ability of the 2-kb LAT and the other RNAs to bind to several RNA-binding proteins. In these experiments, antibodies against specific proteins were utilized to determine whether LAT associates with ribosomes. In addition, since the majority of LAT was found in the nucleoplasm of infected cells, where it sediments between 40 and 60S (data not shown), we wanted to determine whether LAT interacts with the splicing machinery as well as with general RNA-binding proteins. Antibodies used in this section were directed against RNPs, ribosomal P and L7 proteins, and general splicing factors. The ribosomal P protein is part of the 60S ribosomal subunit and is thought to bind the GTPase domain of the 28S rrna to aid in the accessibility of the rrna for elongation factors (52). Therefore, if LAT is similar to the 28S rrna and plays a structural role in the ribosomal complex, this ribosomal protein would be a likely candidate for binding to LAT. The ribosomal L7 protein is associated with the large subunit of eukaryotic ribosomes and is involved in regulating protein translation (31, 39). HnRNPs are proteins that bind nascent RNAs and are involved in RNA processing (8, 16, 22). RNPs are involved in the interaction of hnrnas with nuclear structures, in nucleo-cytoplasmic transport of mrna, and in other important cellular processes. Therefore, these proteins are likely candidates for binding to the 2-kb LAT during the viral life cycle. In addition, sera from patients with autoimmune diseases were also used in this assay since these sera contain antibodies against general RNA-binding proteins as well as splicing factors (16, 53). Immunoprecipitates were blotted onto a nylon membrane and hybridized with probes specific for the 2-kb LAT, 28S ribosomal RNA gc viral mrna, and -actin mrna (Materials and Methods). The MBP antibody was used as a negative control for binding to the 2-kb LAT, whereas the binding of the 28S rrna to the ribosomal P protein was used as a positive control for the assay. Results depicted in Fig. 5 and 6 are expressed as ratios of the binding of LAT, 28S rrna, or gc mrna to the indicated antibody over the nonspecific background binding levels. In HeLa cells (Fig. 5A), the 28S rrna binds to the ribosomal P protein as expected. However, the 28S rrna is not immunoprecipitated by antibodies to RNPs, autoimmune antigen, the L7 ribosomal protein, or the Sm proteins. This positive control verifies that the immunoprecipitation conditions are optimal in this assay for the specific binding of antibodies to their antigen. In contrast to the 28S rrna, the 2-kb LAT was pulled down with several proteins in our assay. LAT was immunoprecipitated by the antibody to the ribosomal P protein at levels similar to that of the 28S rrna, further indicating that in the cytoplasm of HSV-infected cells, LAT interacts with ribosomal factors. LAT was also immunoprecipitated at high levels with the Sm antibody, which recognizes splicing factors. However, we found that the association of LAT with Sm fac-

8 VOL. 75, 2001 INTERACTION OF LAT WITH RIBOSOMES AND SPLICING FACTORS FIG. 6. Immunoprecipitation of LAT, 28S rrna, and gc mrna with specific antibodies in neurone-like SY5Y cells. The same experiment as that depicted in Fig. 5A was carried out, except that SY5Y cellular extract was used. Results are expressed as the ratios of the levels of binding of LAT, 28S, or gc over the background binding levels. Values above 1 indicate positive association, whereas values below 1 are not significant for association. FIG. 5. Immunoprecipitation of LAT, 28S rrna, and gc mrna with specific antibodies in HeLa cells. HeLa cellular extract was prepared from HSV-1 strain 17-infected cells, and 100- l aliquots were incubated with the RNP, P, L7, Sm, and MBP antibodies and autoimmune sera for 2 h at 4 C. Protein A-Sepharose and protein G-agarose were incubated with extract alone to control for background binding to the beads. The lysate-antibody mixture was then incubated with protein A or protein G for 1 h and washed six times in radioimmunoprecipitation assay buffer. Proteins were eluted from the agarose or Sepharose beads with the addition of low-ph buffer, and the supernatant was collected and neutralized. The supernatant was blotted onto a nylon membrane and hybridized with probes for the 2-kb LAT, 28S rrna, and gc mrna (A). (B) Immunoprecipitation with the P and Sm antibodies was carried out three consecutive times to determine the amount of 2-kb LAT or 28S rrna remaining in the extract after each addition of antibody. Results are expressed as the ratios of the levels of binding of LAT, 28S, or gc over the background binding values. Values above 1 indicate positive association, whereas values below 1 are not significant for association. tors is not as strong as its interaction with the ribosomal complex since high-salt conditions easily disrupted LAT-Sm interactions (data not shown). On the other hand, gc, an actively translated mrna, associates weakly with both ribosomal proteins. Since these proteins are not known to directly bind mrna during translation, we would not expect high levels of interaction between gc and P or L7 proteins. In addition, since gc is not spliced, its interaction with splicing factors was minimal as compared to the 2-kb LAT intron. A question that arises from these immunoprecipitation experiments is whether these results are accurate, or if the antibodies are saturated such that the level of binding of each antibody to its antigen, in Fig. 5A, is not authentic. To determine if the level of binding we measured was accurate, an immunoprecipitation analysis was carried out in which equal amounts of P or Sm antibodies were added to the same cellular extract three consecutive times. This would allow us to determine whether additional LAT, gc mrna, or 28S rrna can be pulled down with the addition of new antibody to the extract. Figure 5B indicates that there is a decreasing linear relationship between binding of each of the RNAs and the addition of antibody at consecutive times. Therefore, by the third addition of antibody, LAT, gc, or 28S RNAs could no longer be pulled down by the antibodies. These data indicate that after the first addition of antibody, the binding sites of the antibody are mainly saturated and most of the RNAs have been pulled down, thus generating results that are similar to those shown in Fig. 5A. The association of LAT to cellular proteins in undifferentiated neuron-like SY5Y cells (Fig. 6) was also analyzed. Results similar to that in HeLa cells, with minor differences, were observed. In SY5Y cells, the binding of LAT to splicing factors was similar in level to its interaction with the ribosomal P protein, whereas in HeLa cells, at least twice as much LAT associated with splicing factors as did with the ribosomal P protein. Furthermore, gc bound more strongly to antigens found in autoimmune sera in SY5Y cells than in HeLa cells. Since autoimmune sera contain numerous RNA-binding proteins that are important in cellular processing, these results suggest that gc mrna synthesis and processing may be more active in neuronal cells.

9 12078 AHMED AND FRASER J. VIROL. We conclude from these results that the interaction of LAT with ribosomal proteins, splicing factors, and RNA-binding proteins is distinct from that of both the viral gc mrna and the 28S rrna. It associates at high levels with splicing factors in vivo, unlike the 28S rrna and gc mrna, but interacts with the ribosomal P proteins at levels similar to that of the 28S rrna, again suggesting a structural role for LAT in the translational machinery. DISCUSSION Transport of the 2-kb LAT to the cytoplasm of infected cells. In this paper, we demonstrated that in productively infected HeLa cells, the major fraction of the 2-kb LAT was found in the nucleoplasm, while a smaller fraction (30%) was found in the cytoplasm. Although previous studies demonstrated that LAT is present in the cytoplasm of acutely infected tissue culture cells and SCID mouse brain stems (32), these earlier studies did not clearly establish what proportion of LAT was found in the cytoplasm versus the nucleus. Therefore, this is the first instance where the amount of 2-kb LAT in different cellular compartments was measured in detail. The separation of HSV-1-infected cells into subcellular compartments in Fig. 1 also indicates that LAT is found in membrane fractions. Although only 5% of total LAT is found in the outer nuclear membrane fraction, this may be indicative of the proportion of LAT that is transported to the cytoplasm from the nucleus at a given time. In contrast to LAT, we saw that -actin mrna was not found in any of the membrane fractions in infected cells, indicating that transport of cellular mrnas during HSV-1 infection is impaired. As mentioned earlier, this is due to the virus-induced nuclear retention of cellular mr- NAs due to the impairment of cellular splicing by the viral ICP27 protein (15, 41). As of now, it is not known how LAT is transported to the cytoplasm, but it is possible that during its migration to the cytoplasm, LAT may interact with components of the nuclear membrane involved in transport, such as the nucleoporins. Alternatively, it is conceivable that LAT is transported to the cytoplasm by heterologous nuclear ribonucleoproteins, or hnrnps. HnRNPs are an abundant family of proteins that play essential roles in pre-mrnas processing (8). In fact, several hnrnps remain associated with nuclear mrnas after the completion of splicing and are involved in the nucleo/cytoplasmic transport of RNAs (6, 19). Results, not shown in this paper, indicate that in infected cells, the 2-kb LAT comigrates with an RNP that migrates at 40 kda on an SDS-polyacrylamide gel (Ahmed et al., unpublished). This size corresponds to the A1 hnrnp, which is known to contain nuclear export signals to aid in the active transport of mrnas to the nucleus (4). The immunoprecipitation results presented in Fig. 5 demonstrate that LAT associates minimally with antigens to the monoclonal RNP antibody. Therefore, these data suggest that if 2-kb LAT does bind the A1 hnrnp, it may do so below the level of detection in our experiments. The association of LAT with translation factors. Although previous work (32) has shown that the 2-kb LAT sediments at approximately 50S in the cytoplasm, corresponding to translation initiation complexes, in this paper we have demonstrated that LAT associates with ribosomal proteins found in the 60S subunit (Fig. 5 and 6). Since our sucrose gradient analyses show an overlap between the LAT and 28S rrna peaks (data not shown), it is possible that a portion of LAT is found in the 60S ribosomal complex, whereas another fraction of LAT is found with translational initiation complexes. Alternatively, LAT may interact with ribosomal components to form a novel 50S subunit, which is distinct from the standard cellular 60S ribosomal complexes. Studies have shown that portions of the 2-kb and 1.5-kb LATs are also found in polysomal fractions of cell extracts from latently infected ganglia and infected neuronal cells (13). Under these conditions, it has been speculated that LAT may have access to the translation machinery of the cell. However, despite numerous efforts by several groups, an in vivo-expressed LAT protein has not been demonstrated (reviewed in reference (46). An alternative possibility is that LAT is a functioning RNA similar to the adenovirus-associated (VA) RNAs. The VA RNAs function to block the activation of PKR, the double-stranded RNA-activated inhibitor of protein synthesis, to aid in the efficient translation of cellular and viral proteins late after infection (26, 33). LATs do not appear to have a function similar to that of the VA RNAs. Nevertheless, an RNA-dependent translational function for LATs cannot be ruled out. An obvious interpretation of the interaction of the 2-kb LAT with ribosomal proteins, as indicated in Fig. 3, 5, and 6, is that it plays a role in the virus-induced shutoff of host protein synthesis by repressing the action of the translational machinery. However, we have determined that at a gross level, the shutoff of host protein synthesis after wild-type HSV-1 infection was no different from infection with LAT deletion mutants (data not shown). Still, it is possible that LAT selectively inhibits or enhances translation of certain proteins with specific functions. On the other hand, the possible interaction of LAT with specific ribosomal factors may serve an alternative role during ribosomal biogenesis. Is the 2-kb LAT in the nucleoli of infected cells? Surprisingly, a significant subset of the 2-kb LAT was found in the extract of the nuclear pellet, containing nucleolar material (Fig. 2). Previous studies have not clearly identified whether the 2-kb LAT is found in the nucleoli of infected cells. However, it is interesting to speculate what role, if any, LAT may play in the nucleoli. Since the nucleolus is the center of rrna transcription and ribosomal complex formation (2, 4, 51), perhaps the presence of LAT in the nucleoli is indicative of its interaction with the translational complex. Figure 3 shows that the affinity of LAT for ribosomes is comparable to that of the 28S rrna. Furthermore, its affinity for these complexes is greater than that of cellular messages and viral messages (Fig. 3 and 4). Therefore, it is possible that LAT plays a structural role in ribosome folding similar to that played by rrnas. To reinforce this idea, immunoprecipitation studies, illustrated in Fig. 5 and 6, conducted with both HeLa and SY5Y cells, indicated that LAT associates with the ribosomal P protein at levels similar to that of the 28S rrna, and less strongly to the ribosomal L7 protein. The ribosomal P proteins are known to bind the GTPase domain of the 28S ribosomal RNA protein during complex formation in order to make the 28S rrna accessible to elongation factors (52). The ribosomal L7 protein is involved in regulating protein translation and can bind both

10 VOL. 75, 2001 INTERACTION OF LAT WITH RIBOSOMES AND SPLICING FACTORS RNA and double-stranded DNA. Therefore, both proteins are important players in the translational complex. The interaction of LAT with either one of these proteins could affect ribosomal complex formation during biogenesis in the nucleoli or the function of these proteins in the cytoplasm. The function of LAT and its interaction with ribosomal proteins. Although several functions for the LATs have been investigated in detail, the most recent hypothesis states that LAT blocks virus-induced neuronal apoptosis to facilitate the survival of infected cells (37). The increased viability of infected nerve cells would thus be advantageous for the virus in promoting efficient production and reactivation from latency. This hypothesis is supported by previous studies showing that LAT deletion mutants are less efficient at establishing latency (36, 42), more neurovirulent (35), and less effective at reactivating from latency (3, 18, 23, 50). In addition to LAT, there are at least four other HSV genes, ICP27, US3, US5 (gj), and US6 (gd), that have been reported to have antiapoptotic functions during lytic infection (1, 21, 24, 57). However, it is possible that at later times postinfection, or during latency, LAT is the predominant player in protecting cells from apoptosis since it is the only transcript synthesized at that time. Although there have been many speculations over the roles of the LATs during the viral life cycle, mechanisms for their functions have not been dissected. It is possible that the LATs protect cells from apoptosis by interacting directly with members of the apoptotic pathway. Since LAT can protect cells from a host of apoptotic inducers, it is thought that LAT may affect a downstream regulator of apoptosis (37). Alternatively, LAT may interact with certain ribosomal proteins or the intact translational machinery to inhibit apoptosis. Recently, studies have suggested a correlation between levels of ribosomal proteins and activation of apoptosis. For example, inhibiting the expression of enhanced levels of the ribosomal protein S3a has been found to be directly related to apoptotic induction in certain cell lines (28, 29, 30). Furthermore, constitutive expression of L7 ribosomal protein has also been linked to activation of the apoptotic process in Jurkat cells (28). It is possible that disrupting the balance of factors involved in the translation complex may lead to a reduced efficiency in the expression of certain cellular antiapoptotic factors. Data in our paper indicate that the 2-kb LAT interacts with the ribosomal P protein, and to a lesser degree with the ribosomal L7 protein (Fig. 5 and 6). The association of LAT with ribosomal proteins may serve to stabilize the translational machinery and aid in the production of specific antiapoptotic factors, or inhibition of apoptotic proteins, under conditions of virus-induced apoptosis. This hypothesis is reinforced by our finding that LAT binds to the translational machinery with a greater affinity than cellular or actively translating viral mrnas (Fig. 3 and 4). In addition, the affinity of LAT for these complexes is similar to that of the 28S rrna (Fig. 3), which plays a role both structural and functional in the translation complex. The association of LAT with splicing factors. The main proportion of the 2-kb LAT is found in the nucleoplasm of infected cells as seen in Fig. 2. Therefore, is interesting to speculate on the types of proteins that may be binding to 2-kb LAT in the nuclei. Although 2-kb LAT is a nonpolyadenylated transcript, it is processed in the nuclei as are cellular or viral mrnas. Therefore, it may have access to hnrnps that are involved in essential cellular processing functions. Since 2-kb LAT is a spliced intron, the most obvious candidates for binding to LAT are the splicing factors. Figure 5 indicates that 2-kb LAT binds to Sm antigens, which are mainly splicing factors, about threefold over background. Furthermore, more than twice as much LAT per cell can be pulled down with the anti-sm antibody as with the ribosomal antibodies. This suggests that a higher percentage of 2-kb LAT associates with splicing factors than to ribosomal proteins in vivo. However, the affinity of 2-kb LAT for these splicing factors is not as great as that of LAT for ribosomal proteins (data not shown). Overall, these results suggest that LAT is processed similarly to mrnas in the nuclei of infected cells. In order to get a clearer picture of LAT s interaction with nuclear factors, we must further dissect these interactions by utilizing antibodies to specific proteins involved in transcription, splicing, and transport. ACKNOWLEDGMENTS We acknowledge Darby Thomas and Cathie Miller for critically reading the manuscript. This work was supported by the Public Health Service Program Project grant NS33768 from the National Institutes of Health. REFERENCES 1. Aubert, M., and J. Blaho The herpes simplex virus type 1 regulatory protein ICP27 is required for the prevention of apoptosis in infected human cells. J. Virol. 73: Besse, S., and F. Puvion-Dutilleul Intranuclear retention of ribosomal RNAs in response to herpes simplex virus type 1 infection. J. Cell Sci. 109: Block, T. M., S. L. Deshmane, J. Masonis, J. Maggioncalda, T. Valyi-Nagy, and N. W. Fraser An HSV LAT null mutant reactivates slowly from latent infection and makes small plaques on CV-1 monolayers. Virology 192: Burd, C. G., and G. Dreyfuss Conserved structure and diversity of functions of RNA-binding proteins. Science 265: Clarke, P. A., M. Schwemmle, J. Schinckinger, K. Hilse, and M. J. Clemens Binding of Epstein-Barr virus small RNA EBER-1 to the doublestranded RNA-activated protein kinase DAI. Nucleic Acids Res. 19: Cullen, B. R Connections between the processing and nuclear export of mrna: evidence for an export license? Proc. Natl. Acad. Sci. USA 97: Dreyfuss, G., Y. D. Choi, and S. A. Adam Characterization of heterogeneous nuclear RNA-protein complexes in vivo with monoclonal antibodies. Mol. Cell. Biol. 4: Dreyfuss, G., M. Matunis, S. Pinol-Roma, and C. Burd hnrnp proteins and the biogenesis of mrna. Annu. Rev. Biochem. 62: Erickson, J. M., C. L. Rushford, D. J. Dorney, G. N. Wilson, and R. D. Schmickel Structure and variation of human ribosomal DNA: molecular analysis of cloned fragments. Gene 16: Farrell, M. J., A. T. Dobson, and L. T. Feldman Herpes simplex virus latency-associated transcript is a stable intron. Proc. Natl. Acad. Sci. USA 88: Fraser, N. W., T. M. Block, and J. G. Spivack The latency-associated transcripts of herpes simplex virus: RNA in search of function. Virology 191: Garber, D., P. Schaffer, and D. Knipe A LAT-associated function reduces productive-cycle gene expression during acute infection of murine sensory neurons with herpes simplex virus type 1. J. Virol. 71: Goldenberg, D., N. Mador, M. J. Ball, A. Panet, and I. Steiner The abundant latency-associated transcripts of herpes simplex virus type 1 are bound to polyribosomes in cultured neuronal cells and during latent infection in mouse trigeminal ganglia. J. Virol. 71: Greco, A., A.-M. Laurent, and J.-J. Madjar Repression of -actin synthesis and persistence of ribosomal proteins synthesis after infection of HeLa cells by herpes simplex virus type 1 infection are under translational control. Mol. Gen. Genet. 256: Hardwicke, M. A., and R. M. Sandri-Goldin The herpes simplex virus regulatory protein ICP27 contributes to the decrease in cellular mrna levels during infection. J. Virol. 68: Hassfeld, W., E. K. L. Chan, D. A. Mathison, D. Portman, G. Dreyfuss, G. Steiner, and E. M. Tan Molecular definition of heterogeneous nuclear ribonucleoprotein R (hnrnp R) using autoimmune antibody: immunolog-