Comparison between Human Cytomegalovirus pUL97 and Murine ...

JOURNAL OF VIROLOGY, Nov. 2000, p. 10729–10736 0022-538X/00/$04.00⫹0 Copyright © 2000, American Society for Microbiology. All Rights Reserved.

Vol. 74, No. 22

Comparison between Human Cytomegalovirus pUL97 and Murine Cytomegalovirus (MCMV) pM97 Expressed by MCMV and Vaccinia Virus: pM97 Does Not Confer Ganciclovir Sensitivity MARKUS WAGNER,1 DETLEF MICHEL,2 PETER SCHAARSCHMIDT,2 BIANCA VAIDA,2 STIPAN JONJIC,3 MARTIN MESSERLE,1 THOMAS MERTENS,2* AND ULRICH KOSZINOWSKI1 Max von Pettenkofer Institut fu ¨r Hygiene und Medizinische Mikrobiologie, Ludwig-Maximilians-Universita ¨t Mu ¨nchen, Munich,1 2 and Abteilung Virologie, Institut fu ¨r Mikrobiologie, Universita ¨t Ulm, Ulm, Germany, and Department of Histology and Embryology, Faculty of Medicine, University of Rijeka, Rijeka, Croatia3 Received 26 June 2000/Accepted 18 August 2000

The UL97 protein (pUL97) of human cytomegalovirus (HCMV) is a protein kinase that also phosphorylates ganciclovir (GCV), but its biological function is not yet clear. The M97 protein (pM97) of mouse cytomegalovirus (MCMV) is the homolog of pUL97. First, we studied the consequences of genetic replacement of M97 by UL97. Using the infectious bacterial plasmid clone of the full-length MCMV genome (M. Wagner, S. Jonjic, U. H. Koszinowski, and M. Messerle, J. Virol. 73:7056–7060, 1999), we replaced the M97 gene with the UL97 gene and constructed an MCMV M97 deletion mutant and a revertant virus. In addition, pUL97 and pM97 were expressed by recombinant vaccinia virus to compare both for known functions. Remarkably, pM97 proved not to be the reason for the GCV sensitivity of MCMV. When expressed by the recombinant MCMV, however, pUL97 was phosphorylated and endowed MCMV with the capacity to phosphorylate GCV, thereby rendering MCMV more susceptible to GCV. We found that deletion of pM97, although it is not essential for MCMV replication, severely affected virus growth. This growth deficit was only partially amended by pUL97 expression. When expressed by recombinant vaccinia viruses, both proteins were phosphorylated and supported phosphorylation of GCV, but pUL97 was about 10 times more effective than pM97. One hint of the functional differences between the proteins was provided by the finding that pUL97 accumulates in the nucleus, whereas pM97 is predominantly located in the cytoplasm of infected cells. In vivo testing revealed that the UL97-MCMV recombinant should allow evaluation of novel antiviral drugs targeted to the UL97 protein of HCMV in mice. Human cytomegalovirus (HCMV) is the major cause of congenital viral infections and an important pathogen of immunocompromised patients. One of the compounds used in the chemotherapy of HCMV infections is the nucleoside analog ganciclovir (GCV). The activation of the prodrug requires phosphorylation to the GCV triphosphate, whereby the UL97 protein (pUL97) of HCMV directs phosphorylation to the GCV monophosphate. Prolonged use of the compound can result in the emergence of resistant viruses. Resistance is often associated with mutations in either pUL97 or the HCMV DNA polymerase (UL54) or both (8). The UL97 gene product represents a virion component (29) which acts as a serine/threonine kinase (9), is located in the nuclei of infected cells, and has the capacity for autophosphorylation and for phosphorylation of cellular elongation factor 1␦ (11). Furthermore, pUL97 has the unusual ability to direct the phosphorylation of GCV and other nucleoside analogs even in the absence of other HCMV genes (16, 28, 32). GCV phosphorylation is dependent on pUL97 autophosphorylation (15). The HCMV UL97 gene is homologous to protein kinases of other alpha-, beta-, and gammaherpesviruses (2, 25). It has been shown that pUL97 expressed by a recombinant herpes simplex virus type 1 deficient for the UL13-encoded kinase conferred GCV sensitivity on the recombinant virus in Vero cells but not in thymidine kinase-minus cells (18). HCMV replication is restricted to human cells, and studies * Corresponding author. Mailing address: Abteilung Virologie, Institut fu ¨r Mikrobiologie, Universita¨t Ulm, Albert-Einstein-Allee 11, 89081 Ulm, Germany. Phone: 49 (0) 731 502 3340. Fax: 49 (0) 731 502 3337. E-mail: [email protected].

in the natural host are limited. Mouse CMV (MCMV) provides a useful model with which to study various aspects of CMV infection and disease. MCMV is susceptible to GCV, and consistent with the overall similarity of the genomes, the M97 gene is located in the MCMV genome at a position homologous to that of UL97 in the HCMV genome (24). The gene was found conserved in all field isolates of MCMV (25), but no experimental data concerning the function of pM97 have been published so far. Recently, we pioneered the cloning of herpesvirus genomes as bacterial artificial chromosomes (BACs) by introducing F plasmid functions into the viral genome (13) and constructed an MCMV-BAC which systematically loses all bacterial sequences upon transfection into cells and regains wild-type (wt) MCMV functions (30). These techniques allow mutagenesis of any gene of interest, since genetic engineering is carried out in Escherichia coli prior to testing of characterized mutated genomes for their potential to reconstitute infectious virus progeny (1). Here, we tested whether the M97 gene of MCMV can be functionally complemented by the homologous HCMV gene UL97. We reasoned that if this mutant was viable, it could be used to study pUL97 functions in a small-rodent model. Such a model could be useful for the development of alternative antivirals targeted to HCMV pUL97. Functional differences analyzed by the use of recombinant vaccinia viruses (rVVs) could provide further clues to the biological function of these proteins. MATERIALS AND METHODS Viruses and cell culture. The AD169 strain of HCMV was propagated on human foreskin fibroblasts (HFF) as reported before (16, 17). The Smith strain

10729

10730

WAGNER ET AL.

of MCMV (ATCC VR-194) and the recombinant MCMVs were propagated on BALB/c mouse embryonic fibroblasts (MEF) in Dulbecco’s modified Eagle’s medium supplemented with 5% (vol/vol) newborn calf serum. Generation of rVV and propagation of 143B thymidine kinase-deficient (tk⫺) and CV1 cells were performed as described previously (14, 16). For construction of an rVV expressing the M97 protein that is C-terminally tagged with StrepTag II (amino acids TrpSerHisProGlnPheGluLys) (Institut fu ¨r Bioanalytik GmbH, Go ¨ttingen, Germany), the M97 open reading frame (ORF) was amplified by PCR using MCMV Smith DNA as a template. The amplicon was cloned into a plasmid that already contained the Strep Tag II-encoding sequence. From the resulting plasmid, the M97 ORF was excised and inserted into the rVV plasmid p7.5K131 (14). Recombinant vaccinia virus rVV-UL97, which expresses pUL97 of HCMV, has been described previously (17). Plasmid construction and BAC mutagenesis. Plasmid cloning was performed by standard methods (27). Construction of the shuttle plasmids pSTUL97 and pST⌬M97 made use of plasmid HindIII G that contains the MCMV sequence from nucleotides (nt) 130741 to 150754 (7, 24). For generation of a recombination plasmid, the mutated sequences have to be flanked with DNA fragments that are homologous to upstream and downstream sequences at the intended insertion site in the viral genome. First, a fragment of 3.8 kbp (nt 137785 to 141622) (24) was isolated by PmlI digestion from the HindIII G fragment and inserted into the SmaI site of pUC19. Second, a 2.9-kbp fragment (nt 141861 to 144765) was isolated from HindIII G by digestion with AsuII and PstI and subcloned into SmaI- and PstI-digested pUC19. After destroying the KpnI site within the multiple cloning site (MCS) of the first plasmid construct by mung bean nuclease digestion and religation, the plasmid was cut with SmaI followed by partial digestion with PstI in the MCS. From the second plasmid, the entire insert was excised by EcoRI-HindIII digestion, and both ends were blunt ended. Then, the fragment was cut with PstI within the MCS, and the released fragment was cloned into the first plasmid. The resulting plasmid was cut with KpnI (MCMV nt 141991), partially digested with BstXI (MCMV nt 140146), and religated in the presence of NotI linkers. The resulting plasmid, pUC-⌬M97, contains 2.4 kbp (nt 137785 to 140146) and 2.7 kbp (nt 141991 to 144765) of sequences upstream and downstream of the M97 gene, respectively, but lacks the M97 ORF. For construction of plasmid pUC-UL97, the entire UL97 coding region was isolated from the previously described plasmid p212 (16, 17) using XbaI and EcoRI. After filling in the protruding ends, NotI linkers were ligated to both ends, followed by NotI digestion. The fragment was then inserted into the NotI site of plasmid pUC-⌬M97, resulting in plasmid pUC-UL97. The inserts of pUC-UL97 and pUC-⌬M97 were then isolated and inserted into the shuttle plasmid pST76K (21), resulting in pST⌬M97 and pSTUL97, respectively. Mutagenesis of the MCMV BAC plasmid pSM3fr (30) with shuttle plasmids pST⌬M97 and pSTUL97 was performed by a two-step replacement strategy using homologous recombination in the E. coli strain CBTS (12) as described previously (13, 19, 30), resulting in BAC plasmids p⌬M97-MCMV and pUL97MCMV, respectively. For reconstitution of viral progeny, MEF were transfected with 1.5 ␮g of MCMV BAC plasmids by the calcium phosphate precipitation technique as described (27). Five hours posttransfection, the MEF were treated with glycerol (15% glycerol in HEPES-buffered saline) for 2.5 min as described (27). Isolation and analysis of viral DNA. Plasmids and MCMV BAC plasmids were isolated from E. coli cultures using an alkaline lysis procedure as described (13, 27). Total DNA from infected cells and viral DNA from MCMV virions were isolated as published (7). Southern blot analysis was performed essentially as shown elsewhere (5, 27). The probe used represented nt 138859 to 143639 of the MCMV genome. Plaque formation and plaque reduction assay. MEF were infected with wt MCMV or recombinant MCMV at a multiplicity of infection (MOI) of 0.001. After a 1-h period of adsorption, cells were washed with phosphate-buffered saline (PBS) and overlaid with minimal essential medium (Gibco) containing 10% (vol/vol) fetal calf serum and 0.75% carboxymethyl cellulose (Sigma) to prevent formation of secondary plaques. At 6 days postinfection (p.i.), plaque sizes were determined. Cells were washed with PBS and stained for 1 min with Giemsa solution (Merck, Darmstadt, Germany). Pictures were taken with a light microscope at a magnification of ⫻50. For determination of plaque sizes in square millimeters, the horizontal and vertical diameters of a plaque were measured and the space was calculated. For each virus, the sizes of at least 20 plaques were determined. The sensitivity of MCMV recombinants to inhibition by GCV was tested in a quantitative microtiter plaque reduction assay according to Pavic et al. (20). Briefly, 100 ␮l of GCV solution (serial twofold dilutions to reach final concentrations of from 400 to 0.4 ␮M in duplicates) and 100 ␮l of virus (50 PFU) were distributed into 96-well plates containing 5 ⫻ 104 cells per well. Development of plaques was monitored by light microscopy for 4 to 7 days. For better visualization and quantification, cells were stained for 5 min with Giemsa solution. GCV sensitivity was expressed as the 50% inhibitory concentration (IC50). The IC50 represents the concentration (micromolar) of GCV which results in 50% reduction in plaque number in cultures compared to control cultures without GCV. HPLC analysis of GCV phosphorylation. High-pressure liquid chromatography (HPLC) analysis of cell extracts was performed as described previously (16,

J. VIROL. 32). MEF were mock infected or infected with the different MCMV recombinants at an MOI of 5 and harvested at 36 h p.i. Protein kinase assay. Phosphorylation of pUL97 and pM97 in MCMV-infected and rVV-infected cells was determined according to He et al. (9) and Michel et al. (16, 17). MEF were mock infected or infected with the different MCMV recombinants at an MOI of 5 and harvested at 36 h p.i. The phosphorylated proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (12% acrylamide) and visualized by autoradiography. Western blot analysis. Western blot analysis for detection of pUL97 expressed from recombinant MCMV and rVV was performed as previously described using a pUL97-specific polyclonal antiserum (16, 17). Total-cell lysates were extracted from mock-, MCMV-, and HCMV-infected cell cultures at 24 or 48 h p.i. Expression of the tagged pM97 by construct rVV-M97 in CV1 cells was detected at 18 h p.i. using StrepTactin-horseradish peroxidase conjugate and chemiluminescence according to the manufacturer’s recommendations (Institut fu ¨r Bioanalytik GmbH, Go ¨ttingen, Germany). Indirect immunofluorescence. Indirect immunofluorescence was performed as described previously (16). Briefly, CV1 cells were mock infected or infected with the indicated rVV at an MOI of 0.1 and fixed at 24 h p.i. with methanol-acetone (1:1). The pUL97 antiserum was used at a dilution of 1:100, and the secondary fluorescein isothiocyanate (FITC)-conjugated immunoglobulin G (IgG) F(ab⬘) antibody was diluted 1:30. For better visualization, the cells were counterstained with Evans blue. The tagged M97 protein was visualized by the specific antiserum anti-TagII-polyIV (Institut fu ¨r Bioanalytik GmbH) (dilution of 1:200) and a secondary Texas red-conjugated donkey anti-rabbit IgG F(ab⬘) (dilution of 1:75). Animal experiments. BALB/c (H-2d haplotype) and C57BL/6 mice (H-2b haplotype) were bred at the Central Animal Facilities at the Medical Faculty, University of Rijeka. Neonatal mice, 24 h postpartum, were injected intraperitoneally (i.p.) with 103 PFU of recombinant or wt MCMV in a volume of 50 ␮l of diluent. In vivo depletion of CD4⫹ and CD8⫹ T-lymphocyte subsets was performed by i.p. injection of monoclonal antibodies (rat anti-mouse IgG) to CD4 (YTS 191.1) and/or CD8 (YTS 169.4) molecules (3). Newborn mice received 250 ␮g of antilymphocyte antibodies at the time of injection and every fifth day throughout the experiment. The efficacy of T-lymphocyte depletion was greater than 95% as assessed by cytofluorometric analysis of spleen cells using FITC- or phycoerythrin-conjugated antibodies directed against mouse CD4 and CD8 molecules (Becton Dickinson, Heidelberg, Germany, catalog nos. 1333 and 1447). Detection of infectious MCMV in tissues was performed by plaque assays as described previously (26). Statistics. To compare GCV sensitivity, GCV phosphorylation, and plaque size the Kruskal-Wallis test (chi-square approximation) and the Wilcoxon twosample test were used. A P value of ⬍0.05 was considered significant.

RESULTS Construction of MCMV recombinants. To generate the UL97-MCMV recombinant, the BAC pSM3fr containing the complete MCMV genome (30) was mutated in E. coli at the site of the M97 gene. For insertion of the UL97 gene of HCMV, shuttle plasmid pSTUL97 was constructed, which contains the UL97 ORF flanked by authentic MCMV sequences. In this plasmid, the M97 ORF was replaced by the UL97 ORF. Homologous recombination with the BAC plasmid pSM3fr resulted in the MCMV BAC plasmid pUL97-MCMV, which was used for reconstitution of recombinant UL97-MCMV (Fig. 1C and D). The successful reconstitution of infectious virus revealed that either the M97 gene is not essential for MCMV replication or UL97 can complement its function. Restriction enzyme analysis served to characterize the DNA of the mutant (Fig. 1A and B). Hybridization of the NotI-cleaved UL97-MCMV DNA with a probe specific for the MCMV M97 gene and flanking sequences (Fig. 1C) confirmed the presence of fragments that were predicted after allelic exchange. The M97-deficient MCMV mutant was created by homologous recombination between shuttle plasmid pST⌬M97 and BAC plasmid pSM3fr, resulting in BAC plasmid p⌬M97MCMV. In this BAC plasmid, almost the entire M97 ORF is deleted. Transfection of this BAC plasmid into MEF resulted in plaque formation. Although the kinetics of plaque formation was delayed and required about 10 instead of 5 days for the wt MCMV-BAC plasmid, the successful reconstitution of recombinant ⌬M97-MCMV demonstrated that pM97 is not essential for virus replication and virus growth in vitro. The

10731 COMPARISON OF pM97 AND pUL97 FUNCTIONS VOL. 74, 2000

FIG. 1. Structural analysis of recombinant MCMV genomes. Recombinant MCMV genomes were compared with wt MCMV and revertant rMCMV genomes by restriction enzyme analysis and agarose gel electrophoresis (A) and Southern analysis (B). Specific fragments expected after cleavage with NotI and SfiI are listed in panel D and marked in panels A and B. (C) Genome structure of the virus constructs within the HindIII-G fragment that contains the M97 ORF. The probe used for Southern blotting is shown at the top. The genome structures of the different viruses and the locations of the NotI and SfiI restriction sites in the M97 gene region are depicted below.

10732

WAGNER ET AL.

FIG. 2. Kinase assays with cytoplasmic and nuclear extracts of MCMV-infected cells. MEF were infected with the indicated viruses. Cell fractionation was performed at 36 h p.i., and cytoplasmic (A) and nuclear matrix fractions (B) were prepared and treated for protein kinase activity. Phosphorylated proteins were resolved by SDS-PAGE and visualized by autoradiography. Differences in the phosphorylation patterns are indicated by arrows. (C) Parallel samples were used for Western blotting with a pUL97-specific antiserum. The putative pM97 phosphoprotein is marked by the white arrowhead.

correct deletion of the M97 ORF was analyzed and confirmed as described above (Fig. 1A and B). Finally, to be able to unequivocally attribute phenotypic properties of the UL97-MCMV to the gene exchange, a revertant virus with a wt MCMV genome (rMCMV) was generated. To this end, MEF were cotransfected with the BAC plasmid pUL97-MCMV and a linear MCMV fragment overlapping the M97 gene region (nt 138859 to nt 143639) to allow recombination between the wt MCMV fragment and the recombinant UL97-MCMV genome in eukaryotic cells. Selection for a revertant virus with wt properties was based on the different speed of plaque formation. Hence, the introduction of a positive selection marker into the viral genome which is then difficult to remove was not needed for isolation of a revertant. Cotransfection resulted in the expected early plaque formation phenotype indicative of wt MCMV. The correct reinsertion of the M97 ORF within the rMCMV DNA derived from such plaques was confirmed by comparison of the NotI DNA patterns of rMCMV, recombinant UL97-MCMV, and wt MCMV (Fig. 1A and B). Expression and phosphorylation of pUL97 and pM97 by MCMV recombinants. Expression of pUL97 by MCMV was tested by Western blot analysis in extracts from UL97-MCMVinfected MEF using a pUL97-specific antiserum (16). As demonstrated in Fig. 2C, pUL97 was expressed and showed the characteristic molecular mass of 80 kDa. No signals were detectable in extracts from mock-infected cells or MEF infected with wt MCMV or the M97 deletion mutant. This result proves that pUL97 is expressed by the recombinant UL97-MCMV and that the UL97 antiserum does not cross-react with the pM97 of MCMV or proteins from mouse cells. There is evidence that autophosphorylation of pUL97 is essential for the function of the protein (15). We therefore studied whether the pUL97 expressed by the UL97-MCMV recombinant is phosphorylated in infected MEF. Furthermore,

J. VIROL.

we asked whether we could detect pM97 by comparing the patterns of phosphoproteins in cells infected with wt MCMV and the M97 deletion mutant. A clear phosphorylation signal at 80 kDa could be seen when cytoplasmic extracts were tested (Fig. 2A), and an even stronger one was found when nuclear matrix extracts from cells infected with UL97-MCMV were analyzed (Fig. 2B). The signal was found at the position detected by Western blot analysis performed in parallel with the nuclear matrix extracts (Fig. 2C). However, no prominent differences between phosphorylation patterns of the proteins from mock-infected, wt MCMV-infected, and ⌬M97-MCMVinfected MEF cells were detectable between 70 and 80 kDa, the range of the expected molecular mass of pM97. Only comparison with the size of the rVV-expressed pM97 (see below) hints at its presence in the expected position (see also Fig. 3). On the other hand, an intense signal at 40 kDa appeared in cytoplasmic and nuclear matrix extracts from wt MCMV-infected MEF. This signal was much weaker in samples from ⌬M97-MCMV- and UL97-MCMV-infected cells and in all preparations from nuclear matrix. The cellular elongation factor 1␦ (EF-1␦), a protein of about 40 kDa, is hyperphosphorylated in cells infected with alpha-, beta-, and gammaherpesviruses, and the modulation of EF-1␦ phosphorylation is a property of pUL97 (11). Western blotting using an antiserum raised against the human EF-1␦ (kind gift from G. M. Janssen, Leiden, The Netherlands) detected a protein that migrates at 40 kDa (data not shown). However, without further analysis, we can only speculate that the 40-kDa phosphoprotein represents the kinase-induced modification of the cellular translation factor. Furthermore, a signal between 20 and 30 kDa could be observed in extracts from UL97-MCMV-infected cells but not in the other extracts (Fig. 2B). The nature of this band also remains unclear. Taken together, the results show that pM97 is either poorly expressed by MCMV or poorly phosphorylated under conditions (15, 16) which clearly allow phosphorylation of the MCMV-expressed pUL97. Properties of pM97 and pUL97 expressed by rVV. Properties of pUL97, GCV phosphorylation, pUL97 autophosphorylation, and nuclear transport of the protein were tested by using rVV. Especially, the quantification of GCV phosphorylation can be more easily standardized after expression of the protein by rVV rather than by analysis of HCMV-infected fibroblasts (15–17, 32). Hence, rVV carrying the M97 ORF were constructed for direct comparison with the rVV-expressed pUL97. The M97 protein was tagged with StrepTagII in order to facilitate its detection. Western blot analysis of nuclear and cytoplasmic extracts from rVV-M97-infected CV1 cells were performed. Kinase assays were carried out in parallel to analyze potential phosphorylation of the proteins. As shown in Fig. 3A, a phosphorylated protein with a molecular mass of 73 kDa was detected in cells infected with rVV-M97. In contrast to pUL97, which, as reported before (16, 17), is predominantly localized in the nuclear matrix, the M97 signal was visible almost exclusively in the cytoplasmic extract. This was confirmed by comparing Western blots performed with either StrepTactin-horseradish peroxidase (Fig. 3B) or the pUL97 antiserum (Fig. 3C). No specific signals were visible in control extracts from wt vaccinia virus Copenhagen- or mockinfected cells. To further confirm the cytoplasmic localization of pM97, immunofluorescence with CV1 cells infected with the rVV was carried out (Fig. 4). pM97 was again detected in the cytoplasm (Fig. 4B). Thus, after overexpression in the vaccinia virus system, pM97 is phosphorylated but is distributed to the cytoplasm of the infected cells. GCV phosphorylation mediated by pM97 and pUL97. pM97- and pUL97-directed GCV phosphorylation in the con-

VOL. 74, 2000

FIG. 3. Protein kinase assays with cytoplasmic and nuclear extracts of rVVinfected cells. CV1 cells were infected with the indicated vaccinia viruses and harvested 24 h p.i. For protein kinase assays, cell fractionation was performed. (A) Kinase assays were done with extracts from cytoplasm and nuclear matrix in parallel. For Western blotting, aliquots of these assays were separated by SDS– 10% PAGE. Both the tagged M97 protein (B) and pUL97 using a polyclonal antiserum (C) were visualized by chemiluminescence. pM97 is indicated by white arrows, and pUL97 is marked by black arrows. cop, vaccinia virus strain Copenhagen; M97, rVV-M97 carrying the M97 ORF; UL97, rVV-UL97 carrying the UL97 ORF.

text of MCMV infection was compared. To this end, extracts from MEF infected at an MOI of 5 with wt MCMV, UL97MCMV, and the M97 deletion mutant were analyzed by HPLC. The results obtained from five independent experiments are summarized in Fig. 5 (solid bars). The Smith strain of MCMV has been reported to be sensitive to GCV (25). However, no major differences were found between phosphorylation of GCV in mock-infected control cells, cells infected with MCMV wt, and cells infected with ⌬M97-MCMV. Due to the cytopathic effect of virus infection, GCV phosphorylation in wt MCMV- and M97-minus mutant MCMV-infected cells was even reduced compared to the mock-infected controls. However, as expected, introduction of the UL97 coding region caused a significant increase in the phosphorylation of the

COMPARISON OF pM97 AND pUL97 FUNCTIONS

10733

compound in cells infected with UL97-MCMV (P ⫽ 0.01 for UL97-MCMV versus wt MCMV and versus ⌬M97-MCMV). These results proved that pUL97 can also direct the phosphorylation of GCV when expressed by a recombinant MCMV. To test whether a low level of pM97-directed GCV phosphorylation remained undetected due to the relatively high cellular background phosphorylation in uninfected MEF, we performed GCV phosphorylation assays using the rVV on tk⫺ 143B cells. By using this more sensitive test system, we could confirm the pUL97-mediated GCV phosphorylation, which was significantly higher than after infection of MEF with UL97-MCMV. In this assay we could also observe a small but definitive GCV phosphorylation in extracts from rVV-M97infected 143 tk⫺ cells. This phosphorylation activity was approximately 10% of the activity measured in extracts from rVV-UL97-infected cells (Fig. 5, open bars). Mock-infected cells and cells infected with the wt vaccinia virus Copenhagen had no activity. In conclusion, we could show that, unlike pUL97-directed GCV phosphorylation, pM97 has only very low activity. GCV sensitivity of wt MCMV and recombinants ⌬M97MCMV and UL97-MCMV. We studied whether ⌬M97-MCMV was more resistant than wt MCMV and whether the expression of pUL97 resulted in higher GCV sensitivity in the recombinant. MEF were infected with the different viruses, and the IC50s were determined for GCV using a plaque reduction assay. As expected, wt MCMV exhibited a GCV-sensitive phenotype, with an IC50 of 5.2 ⫾ 0.8 ␮M (standard error of the mean of five independent experiments performed in duplicate). Interestingly, deletion of the M97 coding region did not result in GCV resistance of the deletion mutant. The IC50 of the ⌬M97-MCMV recombinant (3.2 ⫾ 1.4 ␮M), as determined by seven independent experiments, was not significantly different from that of wt MCMV (P ⫽ 0.17). The introduction of the UL97 ORF and expression of pUL97 increased the GCV sensitivity of the MCMV recombinant approximately two- to threefold (1.2 ⫾ 0.7 ␮M) compared to wt MCMV (P ⫽ 0.04). Thus, the GCV sensitivity of MCMV is probably due to relatively high cellular GCV phosphorylation rather than to the weak activity of pM97. Altered biological properties of the MCMV recombinants in cell culture. Plaque formation by the recombinants was studied after infection of MEF at an MOI of 0.001. Plaque sizes were determined at 6 days p.i. after staining with Giemsa. Plaques

FIG. 4. Indirect immunofluorescence of CV1 cells infected with vaccinia viruses. Cells were infected and indirect immunofluorescence was performed at 24 h p.i. (A and B) A primary antibody directed against the StrepTactinII tag and a secondary antibody conjugated with Texas red were used. Cells were infected with (A) vaccinia virus strain Copenhagen or (B) rVV-M97. (C) Cells were infected with rVV-UL97. pUL97 was visualized with a polyclonal UL97-specific antiserum and a secondary antibody conjugated with FITC, and cells were counterstained with Evans blue.

10734

WAGNER ET AL.

J. VIROL.

FIG. 6. Reduced virus release in cells infected with ⌬M97-MCMV and UL97-MCMV. MEF were infected with the indicated viruses at an MOI of 1 (A) or 0.01 (B). Supernatants were harvested daily from days 0 to 4 p.i. Virus titers were determined by standard plaque assay.

FIG. 5. GCV phosphorylation in MCMV- and rVV-infected cells. The phosphorylation of GCV in MEF and 143B (tk⫺) cells after mock infection or infection with the indicated viruses was determined at 36 h p.i. for MCMV or 24 h p.i. for rVV. Mean values from five independent experiments ⫾ standard error are shown. VV cop, Copenhagen strain.

formed by ⌬M97-MCMV were more than fivefold smaller than those formed by wt MCMV (P ⫽ 0.0001). UL97-MCMV produced plaques that were smaller than wt MCMV plaques (P ⫽ 0.001) but threefold larger than plaques of ⌬M97-MCMV (P ⫽ 0.0001). The revertant rMCMV showed wt plaque sizes again. To study the importance of the M97 gene and the UL97 gene in the UL97-MCMV recombinant for virus replication in vitro, replication kinetics of the deletion mutant and the UL97MCMV recombinant were compared to that of wt MCMV and rMCMV. MEF were infected with the four viruses at an MOI of 1, and viral titers in supernatants were determined at different time points after infection (Fig. 6A). Viral titers in supernatants of ⌬M97-MCMV- and UL97-MCMV-infected cells were found to be reduced about 10-fold compared to those of wt MCMV and rMCMV. There was no difference between wt MCMV and rMCMV. Similar results were obtained when testing the titer of cell-associated virus (data not shown). The differences were more prominent after infection at an MOI of 0.01 (Fig. 6B). At 4 days p.i., the difference in titer between wt MCMV and ⌬M97-MCMV was close to 100,000-fold, whereas UL97-MCMV yielded about 10-foldhigher titers than ⌬M97-MCMV. Thus, deletion of the M97 gene strongly compromised viral growth in vitro, and its substitution with pUL97 could only partially complement this deficiency. Virus replication in mice. Given the in vitro replication deficiency of ⌬M97-MCMV and UL97-MCMV, we wondered whether UL97-MCMV could be used for infection of mice to provide an in vivo system for future functional testing of the pUL97 from HCMV and of the mutated UL97 protein in a

small-animal model. Newborn mice depleted of CD4⫹ and CD8⫹ T cells were infected with recombinant viruses. Depletion was done to avoid the contribution of immune control mechanisms and thereby increase the productivity of virus infection. Mice were injected with 103 PFU, and groups of four to six mice were sacrificed at 3, 7, and 14 days p.i. Virus titers were tested in liver, spleen, lungs, and salivary glands. Virus was found in all organs, including the salivary glands. In Fig. 7 the representative titers of the three viruses are shown for two organs. While wt MCMV reached the highest and ⌬M97MCMV the lowest titers, again, UL97-MCMV complements to some extent the loss of the M97 gene in this in vivo system. At day 14, all wt MCMV-infected mice had succumbed to infection, whereas the mice in the other groups were still alive. At that time the highest titers were found in lungs and liver (data not shown). Thus, UL97-MCMV is clearly able to grow in vivo and to seed the relevant organs.

FIG. 7. Virus titers in salivary glands and lungs of infected mice. BALB/c mice (five animals per group) were injected i.p. with 1,000 PFU of the indicated viruses. One day p.i., mice were depleted of CD4⫹ and CD8⫹ T lymphocytes. Virus titers in organs were determined at 7 days p.i. using a standard plaque assay. The significance of the titers was determined as follows: salivary glands, significant (P ⬍ 0.005 for all titers compared); lungs, wt MCMV versus ⌬M97MCMV significant (P ⬍ 0.005); wt MCMV versus UL97-MCMV and UL97MCMV versus ⌬M97-MCMV not significant. DL, detection limit.

VOL. 74, 2000

COMPARISON OF pM97 AND pUL97 FUNCTIONS

DISCUSSION Herpesviruses encode homologs of eukaryotic protein kinases, one example of which is represented by HCMV pUL97 (2, 15, 25). In alphaherpesviruses, pUL97 homologs are dispensable for viral replication in cell culture (4, 6, 10, 23), whereas their function is thought to be relevant for betaherpesviruses (9, 16, 22, 31). Although the role of pUL97 in the biology of HCMV is still unknown, the protein has attracted much interest since it has been shown to activate GCV as well as other antiviral nucleoside analogs by converting them to their monophosphates and because specific mutations in the UL97 ORF confer drug resistance on the virus. Since HCMV is strictly species specific, no animal model is available to test HCMV replication. For many aspects, MCMV serves as a model for HCMV. The fact that MCMV is GCV sensitive and encodes a homolog of HCMV pUL97 led to the idea that the mechanism of antiviral action may be identical. On this premise and to directly compare the functions of pM97 and pUL97, taking advantage of our recently described method of mutating herpesvirus genomes cloned as infectious bacterial plasmids (1, 13, 30), we constructed an MCMV which encodes HCMV pUL97 instead of pM97. To our surprise, comparison of the GCV sensitivity of wt MCMV and ⌬M97-MCMV did not reveal any major differences. Insertion of the UL97 gene clearly increased the GCV sensitivity of MCMV. We therefore concluded that GCV phosphorylation by pM97 contributes, if at all, only marginally to the GCV sensitivity of wt MCMV. These results were confirmed by data obtained with the vaccinia virus-expressed proteins, showing that under identical conditions, pM97 phosphorylates GCV to about a 10-fold lower extent than pUL97. This low activity is comparable to that of pUL97 mutants of GCV-resistant clinical HCMV isolates (17). The sensitivity of herpesviruses to GCV and other nucleoside analogs with a similar mechanism of action is caused not only by the function of a viral kinase, but also by the affinity of the triphosphates for the viral DNA polymerase. This is true, e.g., for herpes simplex virus (HSV) and aciclovir (ACV), which is only minimally phosphorylated by the HSV tk but is nevertheless highly effective due to the high affinity of the ACV triphosphate to the HSV DNA polymerase. Interestingly, after substitution of the HSV-1 UL13 gene with HCMV UL97, the transfer of GCV sensitivity is only detectable in tk⫹ cells (18). Unfortunately, the tk⫺ cell line is refractory to MCMV replication and could not be tested. Therefore, a plausible explanation for the GCV sensitivity of wt MCMV is that the low but significant level of GCV phosphorylation by MEF suffices to inhibit the MCMV DNA polymerase and to restrict MCMV replication. We suggest that the MCMV DNA polymerase can be inhibited by smaller amounts of GCV triphosphate than the DNA polymerase of HCMV. Since in MCMV-infected cells GCV phosphorylation is lower than in mock-infected mouse fibroblasts, it is not reasonable to assume that another MCMV function is responsible for GCV phosphorylation. These differences in the mechanism of prodrug activation indicate that wt MCMV is only of limited value for testing nucleoside analogs which are thought to serve as therapeutics in HCMV disease. In future, UL97-MCMV should provide a useful tool for directly testing pUL97-specific antiviral compounds. We have reported that autophosphorylation of pUL97 is a prerequisite for GCV phosphorylation (15). Notably, phosphorylation or autophosphorylation by pM97, in contrast to pUL97, was hardly detectable after physiological expression by MCMV. To detect pM97 expressed by rVV, the protein was C-terminally tagged with a sequence of eight amino acids.

10735

Using this construct, we were able to detect pM97 at its expected position. Expression of pM97 by MCMV was associated with the expression of a strongly phosphorylated protein migrating at 40 kDa. We considered that this might represent a hyperphosphorylated form of the cellular elongation factor EF-1␦, a known substrate of herpesvirus kinases, which migrates at this position in gels (11). Alternatively, this protein could also represent a processed form of pM97 due to protein cleavage. However, such processing did not occur after expression of the isolated pM97 by rVV. This argues against this interpretation, unless other MCMV proteins are involved in this cleavage. Further studies are required to resolve this issue. Recently, by providing complementing cells expressing the UL97 gene product, Prichard et al. (22) isolated two independent deletion mutants of HCMV lacking more than 70% of the UL97 ORF. The HCMV UL97 deletion mutants grew poorly in tissue culture and with a delayed kinetics, in particular at a low MOI. Here, we found that in MCMV, the absence of the M97 gene also severely hampers plaque formation and virus replication. The marked decrease in virus yield is particularly prominent after infection at a low MOI. Therefore, pM97 plays an important, albeit not essential role during virus replication. Wt virus properties were obtained only when the UL97MCMV genome was reverted again by insertion of the M97 gene. Thus, the growth deficiency of the ⌬M97-MCMV recombinant and of the UL97-MCMV recombinant is causally related to the targeted mutagenesis of M97 and not to any other putative mutation elsewhere in the genome. Deletion of the UL97 homolog in HSV-1, UL13, caused only a mild growth disadvantage, and the HCMV UL97 protein rescued the replication deficiency of UL13-negative HSV-1 and phosphorylated the same viral and cellular target proteins despite the fact that HSV and HCMV belong to different herpesvirus subfamilies (11, 18). The homology of genes between members of the same subfamily of herpesviruses is even higher than the homology between genes from different herpesvirus subfamilies. The domains of strongest homology between pM97 and pUL97 are indeed those relevant for nucleotide binding and for substrate recognition (15, 25). It was therefore surprising that the replacement of pM97 with pUL97 complemented the replication deficiency only partially. The recombinant UL97-MCMV yielded a plaque morphology intermediate between those of wt MCMV and ⌬M97-MCMV, and evidence for rescue of the growth properties by pUL97 complementation was only seen at a low MOI. This suggests that in betaherpesviruses, the specific protein kinase has a much more vital function. We would assume that this deficit was not due to insufficient expression of pUL97 or insufficient kinase activity. The discrepant result after transfer of UL97 into an unrelated alpha- and a homologous betaherpesvirus is not to be expected at first glance. However, we observed that while pUL97 redistributed mainly in the nucleus, pM97 remained in the cytosol. Notably, even the presence of the autologous HCMV UL97 expressed by cells in trans to the UL97-minus mutants of HCMV could not fully complement the growth defect (22). Apparently, already subtle changes in the correct spatial and temporal interaction with the physiological target proteins of the viral kinase affect the efficacy of the function. The growth deficiency of the mutants compared to wt MCMV in vivo and in vitro revealed unexpected aspects. In vitro the titer differences 4 days p.i. amounted to more than 10,000-fold, whereas the difference in vivo was much less and reached only about 100-fold after 7 days of infection. Further, supplementation with pUL97 provided an intermediate phe-

10736

WAGNER ET AL.

notype. This strongly suggests that the requirement for pM97 function involves functions provided by the infected cell type. With both ⌬M97-MCMV- and UL97-MCMV-infected mice, virus replication can be detected in the same organs that are infected by wt MCMV. All that is needed is a certain degree of immunosuppression to improve the replication chance of the pM97-deficient virus mutants in its target tissues in vivo. Thus, despite some loss of virulence, UL97-MCMV can probably serve as a small-rodent model with which to test chemicals targeting pUL97. Altogether, we show that HCMV UL97 can substitute for MCMV M97 in certain respects. With regard to GCV phosphorylation and GCV sensitivity, the phenotype of the MCMV mutants should be defined by the substituted gene. Mutated UL97 genes can be easily inserted into the MCMV background and tested for altered function. The BAC clone of MCMV offers the possibility of testing new compounds directed against the biological function of pUL97 in vitro and in vivo. ACKNOWLEDGMENTS M. Wagner and D. Michel contributed equally to this work. We thank Kirsten Wunderlich for skillful technical assistance. P.S. was supported by the DFG-Graduiertenkolleg Biomolekulare Medizin; D.M. was supported by a grant from the BMBF (01KI9603). The work was supported in part by a grant from the European Community (PL9604721) and by Sonderforschungsbereich 455 of the Deutsche Forschungsgemeinschaft. REFERENCES 1. Brune, W., M. Messerle, and U. H. Koszinowski. 2000. Forward with BACs: new tools for herpesvirus genomics. Trends Genet. 16:254–259. 2. Chee, M. S., G. L. Lawrence, and B. G. Barrell. 1989. Alpha-, beta- and gammaherpesviruses encode a putative phosphotransferase. J. Gen. Virol. 70:1151–1160. 3. Cobbold, S. P., A. Jayasuriya, A. Nash, T. D. Prospero, and H. Waldman. 1984. Therapy with monoclonal antibodies by elimination of T-cell subsets in vivo. Nature 312:548–550. 4. Coulter, L. J., H. W. Moss, J. Lang, and D. J. McGeoch. 1999. A mutant of herpes simplex virus type 1 in which the UL13 protein kinase gene is disrupted. J. Gen. Virol. 74:387–395. 5. Crnkovic-Mertens, I., M. Messerle, I. Milotic, U. Szepan, N. Kucic, A. Krmpotic, S. Jonjic, and U. Koszinowski. 1998. Virus attenuation after deletion of the cytomegalovirus Fc receptor is not due to antibody control. J. Virol. 72:1377–1382. 6. De Wind, N., J. Domen, and A. Berns. 1992. Herpesviruses encode an unusual protein-serine/threonine kinase which is nonessential for growth in cultured cells. J. Virol. 66:5200–5209. 7. Ebeling, A., G. M. Keil, E. Knust, and U. K. Koszinowski. 1983. Molecular cloning and physical mapping of murine cytomegalovirus DNA. J. Virol. 47:421–433. 8. Erice, A. 1999. Resistance of human cytomegalovirus to antiviral drugs. Clin. Microbiol. Rev. 12:286–297. 9. He, Z., Y. He, Y. Kim, L. Chu, C. Ohmstede, K. K. Biron, and D. M. Coen. 1997. The human cytomegalovirus UL97 protein is a protein kinase that autophosphorylates on serines and threonines. J. Virol. 71:405–411. 10. Heineman, T. C., and J. I. Cohen. 1995. The varicella-zoster virus (VZV) open reading frame 47 (ORF47) protein kinase is dispensable for viral replication and is not required for phosphorylation of ORF63 protein, the VZV homolog of herpes simplex virus ICP22. J. Virol. 69:367–370. 11. Kawaguchi, Y., T. Matsumura, B. Roizman, and K. Hirai. 1999. Cellular elongation factor 1delta is modified in cells infected with representative alpha-, beta-, or gammaherpesviruses. J. Virol. 73:4456–4460. 12. Kemkes, B., R. Pich, B. Zeidler, B. Sugden, and W. Hammerschmidt. 1995. Immortalization of human B lymphocytes by a plasmid containing 71 kilobase pairs of Epstein-Barr virus DNA. J. Virol. 69:231–238.

J. VIROL. 13. Messerle, M., I. Crnkovic, W. Hammerschmidt, H. Ziegler, and U. H. Koszinowski. 1997. Cloning and mutagenesis of a herpesvirus genome as an infectious bacterial artificial chromosome. Proc. Natl. Acad. Sci. USA 94: 14759–14763. 14. Metzger, C., D. Michel, K. Schneider, A. Lu ¨ske, H.-J. Schlicht, and T. Mertens. 1994. Human cytomegalovirus UL97 kinase confers ganciclovir susceptibility to recombinant vaccinia virus. J. Virol. 68:8423–8427. 15. Michel, D., S. Kramer, S. Ho ¨hn, P. Schaarschmidt, K. Wunderlich, and T. Mertens. 1999. Amino acids of conserved kinase motifs of the cytomegalovirus protein UL97 are essential for autophosphorylation. J. Virol. 73:8898– 8901. 16. Michel, D., I. Pavic, A. Zimmermann, E. Haupt, K. Wunderlich, M. Heuschmid, and T. Mertens. 1996. The UL97 gene product of the human cytomegalovirus is an early-late protein with a nuclear localization but is not a nucleoside kinase. J. Virol. 70:6340–6347. 17. Michel, D., P. Schaarschmidt, K. Wunderlich, M. Heuschmid, L. Simoncini, D. Mu ¨hlberger, A. Zimmermann, I. Pavic, and T. Mertens. 1998. Functional regions of the human cytomegalovirus protein pUL97 involved in nuclear localization and phosphorylation of ganciclovir and pUL97 itself. J. Gen. Virol. 79:2105–2112. 18. Ng, T. I., C. Talarico, T. C. Burnette, K. Biron, and B. Roizman. 1996. Partial substitution of the functions of the herpes simplex virus 1 UL13 gene by the human cytomegalovirus UL97 gene. Virology 225:347–358. 19. O’Connor, M., M. Pfeifer, and W. Bender. 1989. Construction of large DNA segments in Escherichia coli. Science 244:1307–1312. 20. Pavic, I., A. Hartmann, A. Zimmermann, D. Michel, W. Hampl, I. Schleyer, and T. Mertens. 1997. Flow cytometric analysis of herpes simplex virus type 1 susceptibility to acyclovir, ganciclovir, and foscarnet. Antimicrob. Agents Chemother. 41:2686–2692. 21. Posfai, G., M. D. Koop, H. A. Kirkpatrick, and F. R. Blattner. 1997. Versatile insertion plasmids for targeted genome manipulations in bacteria: isolation, deletion, and rescue of the pathogenicity island LEE of the Escherichia coli O157:H7 genome. J. Bacteriol. 179:4426–4428. 22. Prichard, M. N., N. Gao, S. Jairath, G. Mulambia, P. Krosky, D. M. Coen, B. O. Parker, and G. S. Pari. 1999. A recombinant human cytomegalovirus with a large deletion in UL97 has a severe replication deficiency. J. Virol. 73:5663–5670. 23. Purves, F. C., and B. Roizman. 1992. The UL13 gene of herpes simplex virus 1 encodes the functions for posttranslational processing associated with phosphorylation of the regulatory protein alpha 22. Proc. Natl. Acad. Sci. USA 89:7310–7314. 24. Rawlinson, W. D., H. Farrell, and B. G. Barrell. 1996. Analysis of the complete DNA sequence of murine cytomegalovirus. J. Virol. 70:8833–8849. 25. Rawlinson, W. D., F. Zeng, H. E. Farrell, A. L. Cunningham, A. A. Scalzo, T. W. M. Booth, and G. M. Scott. 1997. The murine cytomegalovirus (MCMV) homolog of the HCMV phosphotransferase (UL97(pk)) gene. Virology 233:358–363. 26. Reddehase, M. J., F. Weiland, K. Mu ¨nch, S. Jonjic, A. Lu ¨ske, and U. H. Koszinowski. 1985. Interstitial murine cytomegalovirus pneumonia after irradiation: characterization of cells that limit viral replication during established infection of the lungs. J. Virol. 55:264–273. 27. Sambrook, J., T. Maniatis, and E. F. Fritsch. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. 28. Sullivan, V., C. L. Talarico, S. C. Stanat, M. Davis, D. M. Coen, and K. K. Biron. 1992. A protein kinase homologue controls phosphorylation of ganciclovir in human cytomegalovirus-infected cells. Nature (London) 358:162– 164. 29. van Zeij, M., J. Fairhurst, E. Z. Baum, L. Sun, and T. R. Jones. 1997. The human cytomegalovirus UL97 protein is phosphorylated and a component of virions. Virology 231:72–80. 30. Wagner, M., S. Jonjic, U. H. Koszinowski, and M. Messerle. 1999. Systematic excision of vector sequences from the BAC-cloned herpesvirus genome during virus reconstitution. J. Virol. 73:7056–7060. 31. Wolf, D. G., A. Honigman, J. Lazarovits, E. Tavor, and A. Panet. 1998. Characterization of the human cytomegalovirus UL97 gene product as a virion-associated protein kinase. Arch. Virol. 143:1223–1232. 32. Zimmermann, A., D. Michel, I. Pavic, W. Hampl, A. Luske, J. Neyts, E. De Clercq, and T. Mertens. 1997. Phosphorylation of aciclovir, ganciclovir, penciclovir and S2242 by the cytomegalovirus UL97 protein: a quantitative analysis using recombinant vaccinia viruses. Antivir. Res. 36:35–42.