INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Human cytomegalovirus (HCMV), a -herpesvirus, is a major cause of congenital malformation and a very frequent opportunistic agent in transplant recipients and AIDS patients. In the course of acute infection with viremia, circulating leucocytes and endothelial cells, though carrying the viral genome, harbor a restricted set of viral proteins (35, 54). A major viral antigen found mostly in mono- and polymorphonuclear leukocyte nuclei (26, 57), pp65 lower matrix phosphoprotein (pUL83) (Fig. 1A), is supposed to result not from de novo synthesis but from direct injection from viral particles, since the majority of positive cells does not exhibit pp65 gene expression (25).

View larger version (29K): [in this window] [in a new window]   FIG. 1.   Summary of yeast THS experiments implicating Plk1 as a pp65 partner. (A) Top: UL83/pp65 and UL82/pp71 ORFs position within HCMV genome map (61). TR, terminal repeats; IR, inverted repeat; US, unique short region; UL, unique long region. Center: schematic representation of pp65 protein sequence. Dashed box, major hydrophilic region; double grey boxes, bipartite nuclear localization signals; vertical bars, CKII-like phosphorylation sites (20, 53). Bottom: Gal4-pp65 hybrid proteins; DB, DNA binding domain; AD, acidic activation domain. (B) Plk1 domain organization (top) and THS Gal4-Plk1 hybrid proteins (bottom). (C) THS tests. HF7c yeast cells were transformed with plasmid combinations to express the indicated proteins. Cells were plated onto SD synthetic medium-agar plates lacking the selection amino acids (Trp and Leu for double transformants; Trp, Leu and His to select for Gal4 activity). After 4 days at 30°C, triple selective plates were inspected for colony growth. Colonies from double selection plates were filter assayed for -gal activity. Alternatively, single colonies were reinoculated into liquid SD selection medium, and shaken at 30°C overnight. Cells were lysed for soluble protein extraction according to the vortexing-glass beads method. All the above procedures followed Clontech Matchmaker kit protocols. Total protein was assayed with the BCA reagent (Pierce) and was -gal assayed by the method of Ausubel et al. (3). Activity values (average of at least three colonies from at least two independent transformations) are expressed as percentages of positive control (the activity in cells harboring reference partners, the strongly interacting hybrids Gal4-DB/p53 [aa 72 to 390; unrelated protein 1 {UP1} in panel C] and Gal4-A/SV40 TAg [aa 84 to 708; unrelated protein 2 {UP2} in panel C]).

In spite of being an abundant viral particle protein (5, 33, 34, 68) and a dominant T-cell antigen (58, 75), pp65 is dispensable for virus growth in cultured fibroblasts (15, 65). pp65 gene knock-out, however, abolishes the production of dense bodies (large noninfectious enveloped particles filled with pp65 protein) which are probably the major source of input pp65 in circulating cells (65). In addition, in the pp65-negative mutant (RVAd65) one of the protein kinase activities normally associated with viral particles (47, 59) is strongly depressed (65). pp65 is phosphorylated (59) on casein kinase II (CKII)-like sites (20, 53) (Fig. 1A), and pp65 itself has been regarded as a serine-threonine protein kinase, since anti-pp65 antibodies immunoprecipitate a CKII-like activity (7, 48, 49, 67). Also, pp65 has been shown to be essential for a block in HCMV IE1-p72 immediate-early protein presentation with infected cell major histocompatibility complex 1 (MHC-1); the escape mechanisms involves an IE1 threonine phosphorylation event favored by pp65 (21).

pp65 is efficiently targeted to the cell nucleus, both as a cytoplasmically injected viral particle content early in infection and as a newly synthesized protein at later infection stages (and in transfected cells) (8, 14, 20, 63, 64, 74), thanks to redundant nuclear localization signals (20, 64) (Fig. 1A). In addition, inside the nucleus most of neosynthesized pp65 sticks to the nuclear lamina and becomes experimentally unextractable, probably due to direct interaction with insoluble nuclear lamins (63). In mitotic astrocytoma cells, a recombinant pp65 was found associated to condensed chromosomes (14).

It appears that the pp65 sequence sums up the molecular information needed to cope with the multiple tasks of entering the nucleus, interacting with the nuclear scaffold, oligomerizing and interacting with other HCMV structural proteins to be eventually packaged into the amorphous tegument of virions and dense bodies, and possibly deploying a kinase activity. The last function, however, is controversial, since the protein lacks a recognizable kinase consensus. The only clearly identifiable pp65 homologs are pp71 (11), the product of the HCMV UL82 gene, and mouse cytomegalovirus pp86, related to both HCMV proteins (13). The UL82/pp71 gene is near to UL83/pp65 in HCMV genome (61) (Fig. 1A), and the two genes, probably deriving from an ancient duplication, are transcribed in a bicistronic mRNA (15, 61). However, in addition of being an abundant virion tegument element (5, 33, 34, 68), pp71 acts as a transcriptional transactivator (44) and is important for replication in vitro (4, 15), in contrast to pp65.

To extend knowledge of pp65 function, we screened for genes encoding pp65 cell protein partners by use of the yeast two-hybrid system (THS). Polo-like kinase 1 (Plk1) cDNA emerged as the most frequent pp65 partner in our screen. Plk1 is essential for a normal mitosis and cytokinesis progression and is endowed with a CKII-like substrate specificity (22, 41). Plk1 copurified with pp65 from COS7 cells transfected with the pp65 gene and from HCMV-infected fibroblasts. Furthermore, the cell kinase is present in wild-type (WT) HCMV (AD169) particles; by contrast, it is absent in pp65-negative mutant virions. A possible conclusion from the above findings is that one function of pp65 is to bind a cell kinase, and to vehiculate it into viral particles. The possibility that Plk1 plays a role in p72/IE1 threonine phosphorylation at early infection times, or that Plk1-pp65 interaction can interfere with normal Plk1 function in the cell, will be discussed.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Cells, viruses, and antibodies. Diploid human embryonic lung fibroblasts (HELF), and COS7 and HeLa cell lines were maintained in Dulbecco's modified Eagle's medium Glutamax (Gibco), 10% fetal bovine serum (FBS) and antibiotics at 37°C. HCMV strain AD169 and its pp65-negative derivative RVAd65(65) were propagated on HELF. Viral titers were measured by an immunohistochemical assays on HELF as follows. Cells in 96-well microtiters were infected with 10-fold serial dilutions of virus stocks and incubated at 37°C for 24 h. Cells were fixed, incubated with anti-IE1/2 monoclonal antibody (MAb) followed by a horseradish peroxidase (HRP)-conjugated anti-mouse immunoglobulin G (IgG) antibody, stained in HRP substrate, and inspected for positive cells.

Plasmid constructions. UL83/pp65 open reading frame (ORF), the natural ATG included (61), was amplified from purified HCMV AD169 DNA by using Pfu polymerase (Stratagene) and the primer pair 5'-ACTGACGG ATC CGC AGC ATG GAG TCG CGC GGT CGC CG (upstream) and 5'-ACTGACCTCGAGCTCAACCTCGGTGCTTTTTGGGCG-3' (downstream). The amplification product was digested with BamHI and XhoI (sites underlined in the primer sequences) and cloned downstream from, and in frame with, Gal4-DB coding region into pGBT9 vector (Clontech) BamHI and SalI sites, to create pGBT9-65. The same insert was introduced into pGAD424 vector (Clontech), creating pGAD-65, to express a Gal4-AD-pp65 fusion; and it was also introduced into pBluescriptSK(+), creating pBSc65. The same fragment was coligated with the phosphorylated PstI-BamHI oligonucleotide adapter      5'-GC ATG GAC TAC AAG GAC GAC GAT GAC AAG GG-3'        |||||||||||||||||||||||||||||||
3'-ACG TCG TAC CTG ATG TTC CTG CTG CTA CTG TTC CCC TAG-5',

Yeast THS. Saccharomyces cerevisiae HF7c, carrying HIS3 and lacZ markers under the control of Gal4 binding sequences, was transformed with pGBT9-65 according to the lithium acetate protocol and selected for leucine prototrophy on SD-Leu plates. Alternatively, yeast cells were double transformed with pGBT9-65 and either plasmid pGAD424 or pTD1, driving the synthesis of Gal4-A alone or Gal4-A fused to aa 84-708 of simian virus 40 (SV40) TAg, respectively; or pGAD-65. Cells were selected for leucine and triptophan prototrophy on SD plates lacking both amino acids. Yeast colonies were then subjected to a filter test for -galactosidase (-gal) expression (19, 60). Both pGBT9-65 single transformants and the double transformants were found to be negative for -gal production, which suggested that pp65 fusion could neither activate Gal4 responsive reporter alone (mimicking an activation domain) nor bind spuriously Gal4-A or Gal4-A fusions. A commercial cDNA library from HeLa cells cloned into pGAD-GH vector (Clontech) was overtransformed into yeast harboring pGBT9-65, and double transformants were plated for selection on SD medium without Trp, Leu, or His and were supplemented with 10 mM 3-aminotriazole. An aliquot was plated on SD-LeuTrp to estimate the number of screened double transformants. Colonies grown on triple selective medium were filter assayed for -gal expression, and the library plasmid from -gal-positive clones was isolated by bacterial rescue. The cDNA insert was automatically sequenced at both ends for a 250- to 400-nucleotide extention and compared to DNA and protein data banks by use of the FASTA and BLAST algorithms.

Recombinant proteins. Escherichia coli DH5- cells harboring pGEX-Plk1-C' were induced with 1 mM isopropyl--D-thiogalactopyranoside (IPTG) for 3 h, and the corresponding GST-Plk1-C' product was purified from cleared bacterial lysates by glutathione-Sepharose (Pharmacia) chromatography, according to the manufacturer's instructions, in the presence of protease inhibitors (10 µg/ml each of leupeptin, pepstatin, aprotinin and 1 mM Pefabloc and benzamidine). GST fusions were either left adsorbed to the resin or eluted with 50 mM reduced glutathione in 25 mM Tris-HCl (pH 7.5) and were dialyzed against phosphate-buffered saline (PBS).

Electrophoresis and Western blotting. In experiments involving L-[³⁵S]-methionine-labelled pp65 (see below), proteins were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (10% polyacrylamide in the resolving gel) and visualized and quantified by use of a phosphorimager (445 SI; Molecular Dynamics). In nonradioactive experiments, proteins were electroblotted onto nitrocellulose membrane (ProTranBA83; Schleicher & Schuell). This was saturated firstly in 1% bovine casein in 50 mM Tris-HCl (pH 7.5)-50 mM NaCl-0.15% Tween 20 (WB buffer) at room temperature for 1 h, and subsequently in the same buffer but replacing casein with 5% skim milk (Difco) for another hour. It was then incubated at room temperature for 1 h with appropriate dilutions of primary antibody. After three washes in WB-skim milk buffer, membranes were incubated for 1 h at room temperature with HRP-conjugated secondary antibodies and revealed with an HRP chemiluminescent substrate (Super Signal; Pierce). Blots were exposed to X-ray film, and signals within the film's linear range were quantified with an imaging densitometer (GS670; Bio-Rad).

In vitro binding assays. For pull-down assays, GST-Plk1-C' or the GST controls was incubated with natural pp65 (0.5 µg) and immobilized on protein G-Sepharose beads through a 4C1 MAb bridge in binding buffer (25 mM HEPES [pH 7.4], 50 mM NaCl, 2.5 mM CaCl₂, 1 mM MgCl₂, 0.1% Triton, 1% bovine serum albumin (BSA) mixed with the protease inhibitor cocktail) for 1 h at 4°C, and the resin was washed four times in the same buffer lacking BSA. The resin-bound pp65 was obtained by lysing in binding buffer HCMV AD169-infected HELF at 10 days postinfection and subjecting the cleared lysate to indirect immunoprecipitation (see below). Resin-bound complexes were boiled in SDS-PAGE sample buffer and analyzed by Western blotting. Alternatively, GST-Plk1-C' bound to glutathione-Sepharose resin was incubated with in vitro translated and labeled pp65 or control for 1 h at 4°C in binding buffer, and the washed complexes were electrophoresed.

Cell transfection and analysis. In a typical experiment, 2.5 µg of pMT-2 or pMT-65 plasmid, or no plasmid DNA (mock experiment), was used to transform a 70% confluent COS7 monolayer in a 25-cm² flask, using the standard calcium phosphate procedure. Cells were harvested 48 h after transfection and directly lysed in prewarmed (95°C) SDS-PAGE sample buffer. Alternatively, after transfection, cells were washed twice with ice-cold PBS plus the protease inhibitor cocktail, scraped with a rubber policeman in the same buffer, pelleted at 650 × g and 4°C, and resuspended in cold CSK buffer (10 mM PIPES [pH 6.8], 100 mM NaCl, 300 mM sucrose, 3 mM MgCl₂, 1 mM Pefabloc, 2 mM vanadyl adenosine, 0.5% Triton X-100) for cell/subnuclear fractionation experiments. These were performed essentially as described previously (18, 63), with minor modifications. After incubation on ice for 5 min, lysate was pelleted at 650 × g, the supernatant (S1) was removed, and the nuclear pellet was serially extracted in the following buffers: CSK with 250 mM ammonium sulfate in place of NaCl (5 min at 4°C, then centrifugation as described above [S2]); CSK with 50 mM NaCl, 100 µg of DNaseI/ml (20 min at 20°C, then ammonium sulfate added to 250 mM final and centrifugation at 1,000 × g [S3]); CSK with 25 µg of RNase A/ml and vanadyl adenosine omitted (10 min at 20°C, then centrifugation as above [S4]); CSK with 1.7 M NaCl (5 min on ice, then centrifugation [S5], and the final insoluble nuclear matrix pellet [NP]). The treatments are expected to remove cytosol and readily extracted nucleosol proteins (S1), salt-extractable nuclear and cytoskeletal proteins (S2), chromatin proteins (S3), RNA associated proteins (S4), and high-salt extractable nuclear matrix proteins (S5).

Phosphorylation assays. In vitro phosphorylation reactions using purified Plk1 and either pp65 fragment 357-475 (20) or dephosphorylated bovine casein (Sigma) as the substrates were performed as described previously (43). For phosphoamino acid analysis, phosphorylated ³²P-labeled pp65-357-475 was separated by SDS-PAGE, transferred onto a polyvinylidenfluoride membrane, and visualized by autoradiography. The radioactive band corresponding to the phosphoprotein was cut out, and the membrane piece was incubated at 110°C for 1 h in 200 µl of 6 N HCl under nitrogen. The amino acids were lyophilized and washed repeatedly with water. The hydrolysate was mixed with 1 µg each of nonradioactive phosphoserine, phosphothreonine, and phosphotyrosine (Sigma) and separated by high-voltage electrophoresis on thin-layer chromatography (TLC) plates in 5% acetic acid-0.5% pyridine [pH 3.5], at 750 V for 20 min. The radioactive signal was detected by phosphorimaging, and the nonradioactive phosphoamino acids were detected by staining with ninhydrin.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

A yeast two-hybrid screening (19) for cellular proteins that interact with HCMV pp65 protein was performed. Yeast expressing a fusion protein comprising the DNA binding (DB) domain of Gal4 transcriptional activator and pp65 (AD169 strain) (Fig. 1A) was overtransformed with a library of human cDNAs from HeLa cells, fused to the sequence encoding the Gal4 activation (A) domain. The bait fusion was preliminarily tested to rule out that it could spuriously activate the THS reporters (Fig. 1C).

pp65 was also fused to Gal4-A, and the Gal4-DB/65 and Gal4-A/65 constructs cotransformed into the yeast reporter strain (Fig. 1A). This cross also did not activate Gal4-responsive reporters (Fig. 1C), which suggested that pp65 cannot dimerize in the yeast THS. This result was quite unexpected, since pp65 seems to extensively oligomerize inside cell to be packaged into virion and dense body tegument. The result might reflect a false-negative result due to the interference of the fused Gal4 domains.

Out of 5 × 10⁶ independent transformants, 47 were selected on the basis of their ability to activate transcription of both HIS3 (selectable) and lacZ (screenable) reporter genes under the control of Gal4 binding sites. Partial sequences at both ends of the cDNA inserts of each clone were obtained, and aligned to DNA and protein sequences in data banks. This analysis showed that 18 clones (38.3% of positives) contained partial cDNAs for Plk1 fused in frame to Gal4-A. Two fully sequenced Plk1 clones, no. 142 and 15, contained codons 318 to 603 and 367 to 603, respectively (Fig. 1B), with no discrepancy with respect to published sequences (24). The tract covered by these clones roughly corresponds to the C-terminal four-fifths of Plk1 nonenzymatic domain (Plk1-C'), including the so-called Polo homology 2 (PH2) domain, highly conserved throughout the family of Polo kinases (22, 41) (Fig. 1B).

Another recurrent isolate (14 clones) was found to coincide with a number of expressed sequence tags (ESTs) in DNA data banks. Two fully sequenced cDNAs were found to bear a continuous ORF covering the cloned fragment through the 3' polyadenylation signal. Intriguingly, the predicted aa sequence for this putative partner displays an extensive homology to pp65 itself (30.5% identity, 59.1% similarity).

Both Gal4-A/Plk1-C' no. 142 and 15 (Fig. 1C), as well as Gal4-A/EST (data not shown) were strong activators of the lacZ reporter (50% activation relative to a standard ths interaction control, Gal4-DB/p53 × Gal4-A/TAg) in the original cross with Gal4-DB/65. Both Plk1-C' (Fig. 1C) and the EST product (not shown) passed standard tests for false positive exclusion, showing that neither protein could spuriously activate the reporter. Thus, both proteins appear as bona fide pp65 partners in the yeast THS.

These results suggested that pp65 can recognize at least two distinct cell proteins as interaction partners. Subsequent work, reported hereafter, was aimed at a more extensive characterization of pp65 interaction with the most frequent isolate, Plk1.

The Plk1 tract originally isolated as a pp65 partner in the THS screen lacks Plk1 kinase domain (Fig. 1B), which is then dispensable for the interaction in the yeast system. To test whether the full-length Plk1 also was capable of interacting with pp65, a complete Plk1 coding sequence (24) was inserted into the Gal4-A yeast plasmid (Fig. 1B) and crossed with the pp65 bait construct. This cross activated the yeast reporter, though appreciably less then that using the Plk-C' fusion (Fig. 1C). By contrast, a fusion including only the Plk1 N-terminal kinase domain (Fig. 1B) failed to activate in a cross with pp65 as the bait (Fig. 1C). In addition, a pp65 internal deletion mutant, fused to Gal4-DB, was checked against the Plk-C' construct. The deletion removed an internal tract overlapping the larger hydrophilic tract of pp65 (20, 53) (Fig. 1A), and a putative kinase nucleotide binding motif (21). The pp65 mutant failed to activate (Fig. 1C).

Taken together, the THS data indicate that Plk1 can bind pp65 via its C-terminal regulatory domain, and that an integer pp65 sequence is required for the interaction. Since Plk1 was the only member of the mammalian Polo family to be repeatedly fished out in our screening, a possible inference is also that, within the C-terminal domain, the PH2 tract is not the important one for pp65 binding in THS.

In order to verify that pp65 and Plk1 can physically interact independently of the yeast genetic assay, their association was first analyzed by in vitro binding experiments.

In one pull-down assay, a GST-Plk-C' fusion and a GST attached to a Plk1-unrelated protein (UP) were compared for their ability to be captured by a natural pp65 immunopurified from HCMV-infected fibroblast lysates. GST-Plk1-C' was obtained by subcloning the Plk1 cDNA fragment as found in ths clone 15 (Fig. 1B) into E. coli pGEX-2TK expression vector, and by purifying from bacterial cells the corresponding fusion protein. The integrity of the full-length hybrid protein was verified by finding that it was recognized by Western blotting by affinity-purified rabbit anti-Plk1 antibodies, raised against Plk1 C-terminal tridecapeptide (Fig. 2A). The GST proteins were incubated with a pp65 immobilized onto protein A-Sepharose resin through an anti-pp65 MAb. The amount of GST proteins captured by pp65 was revealed by Western blotting with anti-GST antibodies. Results from these experiments (Fig. 2B) showed that the GST-Plk-C' fusion has an affinity for pp65 much higher than that of the control. The extensive degradation forms of the GST-Plk1-C' chimera, which prevail in the input protein preparation, were selectively lost in the captured fraction, and those still visible were a subset of those present in the input lane. At least in part, they may represent chimeras which have lost most of GST, and the C-terminal extremity of the Plk1 moiety; they would be still reactive to the polyspecific anti-GST antiserum, and invisible to the anti-Plk1 C-terminal antibodies (Fig. 2A). Thus, part of the degraded forms seen in the pull-down lane might in turn be specifically captured.

View larger version (44K): [in this window] [in a new window]   FIG. 2.   Recombinant Plk1 C-terminal domain binds both natural and recombinant pp65 in vitro. The indicated GST fusions were analyzed by Western blotting with anti-Plk1 C-terminal tridecapeptide rabbit antibodies (A). Alternatively, they were incubated with pp65-anti-pp65 MAb immunocomplexes adsorbed to protein G-Sepharose resin and were analyzed as described in Materials and Methods (B). GST-UP: GST fused to a Plk1-unrelated protein (human immunodeficiency virus Nef protein) (60). (C) GST-Plk1-C', adsorbed to glutathione-Sepharose resin, was incubated with either in vitro translated, [35S]methionine-labeled pp65, or a control (HBV L) (in vitro translated hepatitis B virus large surface protein (20a)]. The retained prey proteins were revealed by SDS-PAGE and fluorography. pd, pulled down. In panels A to C, positions and molecular masses (in kilodaltons) of markers are shown.

A similar result was obtained using a recombinant pp65, produced by in vitro translation of the cloned pp65 gene, in place of the natural pp65. In this case the assay was reversed, the GST-Plk1-C' fusion being adsorbed to a glutathione-Sepharose resin and the captured [³⁵S]methionine-labeled protein quantified by phosphorimaging. GST-Plk1-C' precipitated significantly more in vitro translated pp65 than an unrelated control (hepatitis B virus L protein, HBV L) (Fig. 2C).

To check whether Plk1 and pp65 can also interact in intact cells, the pp65 gene was cloned into pMT-2 eukaryotic vector to be expressed in COS7 cells. An N-terminal epitope tag was fused to pp65 (Flag65). The fused tag was assumed not to interfere on pp65 properties and subcellular localization. This point was preliminarily checked with respect to pp65 nuclear localization. Flag65 protein was verified by IF analysis to be strictly nuclear, whether revealed with the anti-pp65 MAb or with Flag-specific rabbit IgGs (Fig. 3A). Thus, the appended epitope did not disturb pp65 subcellular localization, as expected on the basis of available information on pp65 nuclear localization signals (20, 64) and in agreement with the recent demonstration that pp65 subcellular localization is not altered by the N-terminal fusion of an entire protein, the green fluorescent protein (63). IF analysis also allowed to measure the efficiency of transfection, which in repeated experiments was found to be 8 ± 4%.

View larger version (37K): [in this window] [in a new window]   FIG. 3.   Flag65-Plk1 copurification from transfected COS7 cells. (A) Subcellular localization of epitope-tagged pp65 (Flag65). COS7 cells transiently transfected with pMT-Flag65 plasmid were reseeded on glass coverslips and processed by indirect IF 48 h posttransfection. Cells were fixed in 3% paraformaldehyde, permeabilized, and incubated with the indicated primary antibody and then with either anti-rabbit or anti-mouse FITC-conjugated secondary antibodies. Cells were observed under a Zeiss epifluorescence microscope connected to a CCD camera. N, nucleus, C, cytoplasm. Original magnification, ×1,000. (B) Flag65 and Plk1 subcellular fractionation. Transfected cells were subjected to a serial fractionation-centrifugation protocol, and the obtained fractions were analyzed by Western blotting with the indicated antibody. S1, detergent-soluble cell extract (cytosol, membranes and readily extracted nuclear proteins); S2 to S5, nuclear fractions released after nuclei treatment with 250 mM ammonium sulfate, DNase, RNase and 1.7 M NaCl, respectively; NP, insoluble nuclear pellet. (C) Anti-Plk1 antibodies do not cross-react against pp65. Total cell lysates of either COS7 cells or insect cells (Sf9/Bac65) overproducing a baculovirus vector-driven pp65 (r-pp65) were blotted and stained with Ponceau red and then analyzed by Western blotting with anti-Plk1 antibodies. (D) Anti-Plk1 antibodies detect Plk1 in Flag65 immunoprecipitates. Lysates from COS7 cells transfected with either pMT-Flag65 or the empty vector (pMT-2) were immunoprecipitated with the indicated antibodies, and captured proteins analyzed by Western blotting. In the lanes labeled COS7 lysate, lysate samples corresponding to 1/200 and 1/100 of the amount subjected to immunoprecipitations were run for comparison. The MAbs used for immunoprecipitation, resolved by SDS-PAGE as monomeric heavy (IgGhc) and light (IgGlc) chains, were cross-recognized by anti-rabbit secondary antibodies; their positions are indicated. ccp, cross-reacting cell protein, unknown cellular protein recognized by anti-Plk1 antibodies. Positions and molecular masses (in kilodaltons) of markers are shown.

Secondly, the extractability of Flag65 was studied. As evidenced by the in vitro binding experiments described above, an amount of pp65 sufficient for a small scale immunopurification could be extracted in the presence of nonionic detergent from HCMV-infected fibroblasts (see also Materials and Methods). A more quantitative Western blot analysis on transfected COS7 cells showed that, in analogy with natural pp65, a minor fraction of Flag65 (approximately one-sixth) was readily extracted in a mildly extractive buffer containing 0.5% nonionic detergent, while the rest could withstand treatments of increasing harshness and followed the insoluble nuclear matrix fraction (Fig. 3B, left).

Applying the same fractionation procedure, Plk1 was extracted in nonionic detergent buffer for approximately nine-tenths of the total, in both untransfected and transfected COS7 cells. The residual proportion cofractionated with the insoluble nuclear material and might in turn be associated to the nuclear matrix (Fig. 3B, right). As expected, the rabbit anti-Plk1 antibodies used for this analysis revealed an unidentified cross-reactive 50-kDa cellular protein (Fig. 3) along with the kinase, as reported previously (43). No change in Plk1 distribution between soluble and unsoluble fraction could be appreciated in transfected cells relative to the untransfected ones (data not shown).

Based on this preliminary characterization, the nonionic extraction buffer was adopted to coextract Plk1 and soluble Flag65 and verify their interaction by coimmunoprecipitation. As a most preliminary control, the anti-Plk1 IgGs were tested in WB for a possible cross-reactivity against pp65, to rule out that a spurious recognition of the viral protein could be misinterpreted as detection of the comigrating Plk1. To this purpose, a recombinant pp65 (r-pp65) produced by a baculovirus vector in insect cells, which do not contain Plk1 nor a strict homolog, was utilized. In the insect/baculovirus system, r-pp65 undergoes a posttranslational processing similar to that taking place in mammalian cells, including nuclear targeting, nuclear insolubilization and phosphorylation, thus reflecting some salient properties of the natural counterpart (20). No recognition of the overexpressed pp65 band was observed (Fig. 3C). For Flag65/Plk1 coprecipitation analysis, after lysate immunoprecipitation with anti-pp65 MAb and Western blotting, the membrane was incubated with either anti-Plk1 or anti-Flag IgGs and revealed with the appropriate secondary antibodies. Controls in this experiment were cells transfected with the empty vector and a pp65-unrelated MAb used for immunoprecipitation. The result (Fig. 3D) showed that pp65 was immunoprecipitated only by the specific monoclonal. More importantly, Plk1 was coimmunoprecipitated with pp65, contrary to the cellular protein against which anti-Plk1 antibodies cross-react. The amount of Plk1 copurifying with pp65 under this conditions was calculated to be 1/200th of the extract content (Fig. 3D), that is, roughly 1/20th of the content of the transfected cells.

These results from transfection experiments demonstrate that a specific partnership connects pp65 and Plk1, corroborating THS and biochemical data. The interaction was necessarily assessed on the soluble fraction of both proteins, leaving unanswered the question whether they can associate in the insoluble fraction. A further point made by the sum of the experiments reported thusfar was that pp65-Plk1 interaction does not require other viral products besides pp65 itself. However, it was important to confirm the pp65-Plk1 association in the context of a natural HCMV infection, when a complex set of regulatory and structural HCMV proteins are coexpressed with, and potentially influence, either partner. To test this, cultured human lung fibroblasts were infected (MOI, 2) with HCMV strain AD169, and the monolayers were lysed and analyzed for pp65-Plk1 interaction at increasingly later times postinfection. As an appropriate negative control, the HCMV-AD169 derivative lacking a functional pp65 gene (RVAd65), previously obtained by one of us (65), was studied in parallel. It turned out that Plk1 could be coimmunoprecipitated with pp65 at late (72 h) times postinfection but not at early times (3 h), from AD169-infected fibroblasts (Fig. 4A), although pp65 was efficiently immunoprecipitated at both times (Fig. 4B). In fibroblasts, the amount of Plk1 polypeptide detected in cell lysates is lower than that found in COS7 cells, as expected for nontransformed primary cells. However, the estimate of the amount found in association with pp65 is comparable (1/25th; data not shown).

View larger version (36K): [in this window] [in a new window]   FIG. 4.   pp65-Plk1 copurification from HCMV AD169-infected fibroblasts. Human embryonic lung fibroblasts infected with either the reference HCMV strain AD169, or the pp65-negative RVAd65 mutant, were lysed 3 h or 72 h (3 days) postinfection. Lysates were subjected to immunoprecipitation with anti-pp65 MAb and analyzed by Western blotting with anti-Plk1 (A) or anti-pp65 (B) antibodies. Alternatively, total cell lysates were directly analyzed by Western blotting with anti-IE1/2 antibodies (C). Marker positions and molecular masses are shown.

As expected, from RVAd65-infected cells neither pp65 nor the partner kinase could be precipitated at any time by anti-65 MAb (Fig. 4A and B). For comparison, infection progression was followed by measuring the intracellular levels of the immediate-early products p72/IE1 and p86/IE2 by direct Western blot analysis of lysates with a specific MAb. In this case, in confirmation of the earlier report (65) p72 and p86 levels were found to grow from early to late times postinfection, with no difference between cells infected with WT or mutant virus (Fig. 4C).

A possible explanation of these observations is that Plk1 is scavenged by de novo synthesized pp65 at late infection stages when the viral protein is being accumulated to be packaged into viral particles, mainly into dense bodies. An implication of this interpretation is that Plk1 might be vehiculated into HCMV particles in a pp65-dependent manner. To test this possibility, sorbitol gradient-purified HCMV particles from either WT AD169- or RVAd65-infected fibroblasts were directly analyzed by Western blotting for Plk1 content. As shown earlier (65), the mutant virions, albeit lacking pp65, exhibited a conserved profile of other structural proteins as detected by total protein staining (Fig. 5A, left) or by Western blotting with a hyperimmune human serum (Fig. 5A, center). In accordance with the theory, Plk1 was detected within AD169, but not RVAd65, particle preparation (Fig. 5A, right). By contrast, two control cell proteins, the cyclin-dependent kinase p34^cdc2 and the hnRNP protein A1, both much more abundant than Plk1, could not be detected with specific antibodies in the AD169 particle lysate, suggesting that Plk1 detection was not due to cell protein contamination of the viral preparation (Fig. 5B). Based on Western blotting, including that with GST-Plk1-C' dilutions as internal standard, Plk1 can be grossly estimated to be incorporated into the viral particles at the ratio of 1 molecule per 7 × 10² pp65 molecules.

View larger version (34K): [in this window] [in a new window]   FIG. 5.   Plk1 associates with HCMV viral particles. (A) Lysates (2 µg of total protein) of sorbitol gradient-purified AD169 and RVAd65 viral particles were electrophoresed and blotted, and total protein stained with Ponceau red (left). Bands corresponding to pp71 higher matrix protein, and pp150 tegument protein (pUL32) plus the major capsid protein (MCP; pUL86) which are not resolved in standard SDS-PAGE gels (34, 68) are clearly identifiable in both WT and mutant virus lysate, whereas the prominent pp65 band is absent in RVAd65 sample; the insensitive Ponceau staining strongly underrepresents the remaining viral particle protein pattern. Western blot analysis was subsequently completed by incubating the membrane with either an hyperimmune human anti-HCMV antiserum (center), or the anti-Plk1 antibody (right). (B) A 2-µg quantity of AD169 particles lysate compared with 5 to 10 µg of lysate of HELF for the content in p34cdc2 and A1 proteins by Western blot analysis with antisera of the corresponding specificity. (C) Separation of AD169 virions (V) and noninfectious enveloped particles (N) from dense bodies (D) on positive density (tartrate)-negative viscosity (glycerol) gradients (33, 71) is schematized (white arrow, direction of the centrifugal force [Cfg]), and Western blot analysis of Plk1 (top) and pp65 (bottom) content in V+N and D fractions is displayed below.

Since, in light of previous results, correlation between pp65 and Plk1 presence inside viral particles was thought to descend from physical interaction, Plk1 was also expected to follow pp65 into the particle type, dense bodies, where most of the tegument protein appears to be packaged. To test this additional point, the viral particles were further fractionated on a positive density-negative viscosity gradient (33, 71), separating capsid-containing viral particles (infectious virions and noninfectious enveloped particles) from dense bodies. In agreement with the prediction, Plk1 was found to be largely associated with the dense body fraction (Fig. 5C, top), although significant pp65 amounts were revealed also in the virion fraction (Fig. 5C, bottom). This raises the intriguing possibility that the kinase is selectively partitioned with dense bodies, with a sorting mechanism which would involve more than the relative pp65 abundance in the particle types.

With reference to the functional significance of pp65-Plk1 interaction, we attempted at first to define whether pp65 can be a substrate of Plk1. pp65 is extensively phosphorylated on CKII sites clustered within an hydrophilic region (Fig. 1A) (20, 53), and Plk1 has been shown to modify acidic protein substrates, like casein, with a CKII-like substrate preference (23, 43). HeLa cells were exploited as the source of active Plk1. Since Plk1 activity has a peak at early mitosis (23, 43), cells were arrested in mitosis by nocodazole treatment. Plk1 was immunoprecipitated from the cell lysate with anti-Plk1 C-terminal rabbit antibodies and was eluted from immunocomplexes with a peptide mimicking the C terminus. The cell kinase was mixed with a bacterially-produced pp65 fragment spanning the phosphorylated hydrophilic region of pp65 (aa 357 to 475) in a kinase assay buffer including [³²-P]ATP, and the reaction mixture was analyzed by SDS-PAGE and phosphorimaging. Results showed that pp65 hydrophilic fragment is accepted as a substrate for Plk1 (Fig. 6A, top), like alpha-casein control (Fig. 6A, bottom). A weak signal in correspondence of Plk1 position was also observed (data not shown), confirming previous reports that the cell kinase self-phosphorylates in these conditions (23, 43). Phosphoaminoacid analysis on the modified pp65 fragment demonstrated that it was labeled on serines (Fig. 6B). These results suggest that pp65 is a suitable Plk1 substrate in vitro, although the relevance of this finding to the in vivo setting is unknown, and kinases other than Plk1 have the chance to phosphorylate pp65, as suggested by the in vivo pp65 phosphorylation in insect cells and by the efficient in vitro labeling of pp65 hydrophilic domain by CKII.

View larger version (39K): [in this window] [in a new window]   FIG. 6.   In vitro phosphorylation experiments. (A) Approximately 1, 5, or 10 ng of Plk1, immunopurified from mitotically arrested HeLa cells (I.P. Plk1) was incubated with either 5 µg of a bacterially produced fragment (aa 357 to 475) spanning the pp65 major hydrophilic region (top) or with the same amount of dephosphorylated bovine casein in the presence of [32-P]ATP (43). Reaction mixtures were resolved by SDS-PAGE (15% resolving gel for pp65-357-475; 12% resolving gel for casein) and visualized by phosphorimaging. In panel B, 32P-labeled pp65 fragment from panel A was hydrolyzed and subjected to phosphoamino acid analysis by TLC electrophoresis and phosphorimaging. The position of standard phosphoamino acids is schematized at right. Pi, inorganic phosphate; par., partially hydrolyzed peptides; or., migration origin.

Finally, we tried to directly check whether pp65 harbors an ATP-binding cassette, the minimal property for a protein kinase. To this purpose we exploited the 5'-p-fluorosulfonylbenzoyl adenosine (FSBA) assay. FSBA is an ATP analogue known to irreversibly bind to and identify ATP binding sites of protein kinases (12). The usual assay scheme implies the incubation of the test protein with FSBA, either in the presence or in the absence of an excess of nucleotide competitor. FSBA modification is regarded as specific if it is quantitatively quenched by competitors. Lysates of WT HCMV particles were exposed or mock-exposed to 0.125 mM FSBA, either without competition or in the presence of a 5 mM nucleotide-10 mM Mg²⁺ competitor (ATP, GTP, or the nonhydrolizable ATP analogue adenyl-imidodiphosphate). Reactions were stopped, immunoprecipitated with anti-pp65 MAb, and analyzed by Western blotting with anti-FSBA IgGs. Unfortunately, FSBA labeling of pp65 was found to be insensitive to nucleotide competition (data not shown). This result will be discussed below.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

Several recent studies have shown that the kidnapping of cellular proteins is a widespread viral strategy. Besides the common case of cell membrane proteins trapped in viral envelopes, a number of cell proteins of other classes (enzymes, cytoskeletal components, molecular chaperones) appear to be included in the interior of virions via specific protein-protein interactions with viral matrix or capsid. Kinases are overrepresented among enzymes, which is not surprising in view of the central role played by phosphorylation in protein function regulation. In some cases, the presence of a previously unsuspected cell protein cargo in virions has been disclosed by the interaction-trap cloning approach, as was the case with cyclophilin A incorporation into human immunodeficiency virus particles (46).

HCMV particles have been previously reported to harbor a number of cell proteins: beta-2-microglobulin (27), annexin II (76) and complement regulating surface proteins CD55-CD59 (69), thought to be associated with virion envelope, and implicated in virion adsorption/penetration; DNA polymerase (47); a cellular actin-like protein (5); and capsid-associated protein phosphatase PP2A (50), whose reinjection into the infected cell might contribute to the strong downregulation of intracellular phosphorylation at early infection stages. The present work provides another example of entrapment of a cellular protein within virus particles, discovered thanks to the yeast THS. The bait protein in our yeast screening, HCMV pp65 lower matrix protein, has for many years had a role in clinical practice as an HCMV antigenic marker of choice, by virtue of the ease of its revelation in circulating leukocytes and endothelial cells (35, 54). The energetic expense in pp65 production and the strict conservation of a functional pp65 gene in primary viral isolates strongly support a relevant role of pp65 in the viral infection. However, a HCMV pp65 knock-out mutant (RVAd65) has been demonstrated to replicate efficiently on cultured fibroblasts (65). RVAd65 viability in vitro demonstrated that pp65 is an obliged component of dense bodies, not of infectious virion tegument, in agreement with earlier immunoelectron microscopy studies (32, 39), which failed to detect pp65 in virions. pp65 might then represent an accessory protein with an important function in HCMV spreading in the human host. A proof in this sense has come from the demonstration that IE1-p72 protein, an immediate early viral factor with a pivotal role in the establishment of infection, escapes MHC-1-restricted surface presentation thanks to a threonine-phosphorylation event which correlates with pp65 expression (21). This specific MHC-1 elusion mechanism might be crucial in the establishment of HCMV infection-latency in vivo, since it precedes the deployment by HCMV of a more general MHC-1 downregulation-decoy substitution strategy (31, 55). By hypothesis, this pp65 in trans effect has been linked to the action of pp65 as a serine-threonine protein kinase. A number of laboratories have in fact demonstrated that a serine-threonine protein kinase activity is coimmunoprecipitated by anti-pp65 antibodies (7, 48, 49). This has brought to the designation of pp65 itself as a PK68 kinase (67). An obstacle to this interpretation is the lack of the conserved protein kinase motif signature in pp65 sequence (29). The HCMV genome, however, seems to encode at least two such anomalous kinases, the product of UL97 gene (which controls ganciclovir phosphorylation but also autophosphorylates) (30) and IE1-p72 itself (which self-phosphorylates on serines and phosphorylates E2F transcription factors and pocket proteins) (52).

Moving from this background, we have searched for pp65 cellular protein partners, as a direct way to pinpoint the basis of its molecular action. The outcome of our screen can be summarized as follows. Two proteins recurred as pp65 prey. One, a novel protein partially homologous to pp65, is presently under scrutiny. The other (the most frequent one), coincided with the C-terminal regulatory domain of Plk1. The refinement of ths analysis indicated that Plk1-pp65 association was specific in the context of yeast THS, that full-length Plk1 also could interact with pp65, while the N-terminal kinase domain alone could not, and that the deletion of an hydrophilic tract within pp65 protein abolished the interaction (Fig. 1). The association between pp65 and Plk1 C-terminal domain was further established in pull-down experiments in vitro (Fig. 2). Next, the in vivo association of Plk1 and pp65 was explored, and the cell kinase was found to be coimmunoprecipitated with a recombinant pp65 expressed in COS7 cells (Fig. 3), as well as with the natural pp65 present in fibroblasts infected with the reference HCMV strain AD169 at late infection times (Fig. 4). Plk1 was detected also in lysates of HCMV AD169 particles, especially in dense bodies, whereas it could not be found in the particles of RVAd65 mutant, which indicates that Plk1 is specifically incorporated into HCMV particles through contacts with pp65 protein (Fig. 5).

Altogether, these data support the contention that a fraction of intracellular pp65 is joined in a physical union with at least one cell kinase, Plk1, and that it shuttles it into HCMV particles. It is possible to envision a model (a producer cell model) describing the interaction as the result of Plk1 capture by neosynthesized pp65, import of the cell kinase into HCMV particles, and, possibly, reintroduction of the captured enzyme into host cells at the next infectious cycle, by passive injection from viral particles.

A function of Plk1 in healthy cells seems to be that of signaller of a centrosome maturation checkpoint, at the G₂/M boundary of cell cycle (23, 28, 43, 73). In fact, interference with Plk1 function by antibody microinjection leads to abnormal (mono- or multipolar) mitotic spindle assembly, an effect particularly severe in nontransformed cells like diploid fibroblasts (40). A putative substrate of Plk1, by analogy with the cognate Xenopus protein Plx1 (38), might be Cdc25, the phosphatase activating the mitotic cyclin-dependent kinase p34^cdc2, a key regulator of mitosis progression. Recent work on Plk1, Plx1, and on the budding yeast Cdc5 homolog, furthermore, suggests a complex regulatory interplay between Cdc25, the polo kinase, and APC/cyclosome components, implicating the kinase in mitosis exit also (1, 10, 16, 37, 42, 56, 66). The role in mitosis progression is matched to a sudden increase of Plk1 kinase activityin the face of a moderate increase in Plk1 synthesisat the onset of mitosis (23, 28, 43, 73).

In addition, Plk1 changes location repeatedly during mitosis, concentrating on the dividing centrosome at prophase-metaphase, then moving to the spindle equatorial plane at anaphase, and eventually reaching the midbody at diakynesis. During interphase Plk1 is seen at numerous cytoplasmic spots, on the centrosome, and at intranuclear spots (23, 43). The erratic Plk1 subcellular localization might be explained by interactions with proteins associated to the microtubular apparatus (mitotic kinesin-like proteins [2, 43] or microtubule-associated proteins [72]). Based on these notions, one can surmise that nuclear pp65 binds the Plk1 subpopulation residing in the nucleus of interphase cells, and/or that pp65 polypeptides can scavenge cytoplasmic Plk1 before nuclear transfer. In our experiments, the interaction was necessarily assessed on the extractable fraction of both proteins, leaving unanswered the question of whether the majority of pp65 molecules that are sequestered on the nuclear cytoskeleton can also bind Plk1. One might suspect that this is the case, though, since pp65 adhesion to the nuclear lamina is possibly a prerequisite for its incorporation into the budding viral particles, to which Plk1 is eventually found associated.

The functional significance of pp65-Plk1 interaction can be at present only matter for speculation. Distinct, but not mutually exclusive, hypotheses can be envisaged. For instance, (i) Plk1 serves to phosphorylate pp65 or other HCMV particle proteins during the assembly process or inside the particles; (ii) Plk1 serves to phosphorylate viral or cellular proteins at or just after cell penetration, at the beginning of the infectious cycle; (iii) Plk1 is targeted and potentially phosphorylated by the viral protein, which interferes with the normal cell kinase function.

Our results do not allow us to decide among these possibilities, although clues in favor of each can be found in the literature and in our experiments. The previously mentioned IE1-specific immune evasion mechanism requiring pp65 coexpression is triggered by an IE1 threonine phosphorylation which might well be produced by a pp65-associated Ser/Thr kinase, like Plk1, in alternative to pp65 itself; or, which might be the result of a more complex network of regulatory phosphorylations (as mentioned above, IE1 itself is a kinase [52]; thus, all the hypothesized partners might phosphorylate one another).

Plk1 substrate specificity makes it a potential pp65 modifying kinase. Accordingly, the pp65 region which is target of phosphorylations on CKII-like sites in natural pp65 and in vitro was found to be modified in vitro by Plk1 (Fig. 6A) on serines (Fig. 6B).

On the other hand, Plk1 activation in the natural setting is thought to entail phosphorylation(s), which would lead to the detachment of a self-inhibitory, C-terminal pseudosubstrate peptide from the kinase active center (51). HCMV infection has been reported to subvert the normal cell cycle progression, halting cycling cells at the G₁/S boundary (or at the G₂/M boundary if cells are infected after the S phase) (6, 17, 36, 45, 62) so that in most infected cells, Plk1 should be expected to be hypophosphorylated and inactive. But, since pp65-Plk1 interaction involves the C-terminal domain of the cell kinase, the interaction itself might be activating for Plk1; alternatively, pp65 might activate Plk1 by phosphorylation.

All in all, the assignment of the roles of modifying enzyme and substrate in the Plk1-pp65 partnership is an issue left unresolved by our experiments. The attempt to define, through an affinity labeling procedure, whether pp65 carries an ATP-binding site as must be for a protein kinase has been frustrated by the finding that pp65 modification by FSBA cannot be competed by an excess of ATP, GTP, or of a nonhydrolyzable ATP analog (data not shown). This could be due to an aspecific labeling problem, or, alternatively, might be explained by pp65 containing a binding cassette with preference for adenine compounds (e.g., ADP and SAM) not assayed as competitors. Further nucleotide binding studies, as well as the use of a procaryotically produced pp65 in kinase assays, are needed to explore this aspect.

Finally, the induction of chromosomal aberrations and aneuploidy (consequences of an altered mitotic chromosome segregation apparatus) is a well-known effect of HCMV infection, and specific defects of centrioli-spindle poles in dividing cells have been reported (reference 9 and references therein). Given the functional and physical connection between Plk1 and the centrosome, a role of altered Plk1 activity or subcellular localization in these events warrants specific investigation.