The Disruption of ND10 during Herpes Simplex Virus Infection Correlates with the Vmw110- and Proteasome-Dependent Loss of Several PML Isoforms

Plasmids and bacteria.

The pCI expression plasmids used in this study were based on the pCIneo vector (Promega), in which the human cytomegalovirus enhancer-promoter region lies upstream of an intron sequence, a multicloning region, and the simian virus 40 late poly(A) signal. A PIC1 coding region with an N-terminal c-Myc epitope tag sequence (3) was inserted into the cloning region to give plasmid pCIPIC1. Plasmid pPIC1GST was constructed by using oligonucleotides to link the C-terminal double-glycine codons of the N-terminally Myc-tagged PIC1 coding region to the initiating methionine codon of glutathione S-transferase (GST) in a T7 expression vector based on pET8c, thus allowing the expression of a fusion protein with PIC1 linked to GST in a way analogous to the ubiquitin-GST (Ub-GST) fusion protein expressed by plasmid pRB307 (a kind gift from Rohan Baker). Plasmids pACYC-HAUSP and pACYC-UBP2 express HAUSP and UBP2, respectively, from pACYC replicons with tac promoters (17). Plasmids were grown in Escherichia coli DH5, and large-scale preparations were made by using the boiling method and CsCl purification.

Viruses and cells.

HSV-1 strain 17 syn+ was the wild-type strain used in these studies. Vmw110 mutant viruses dl1403, FXE, D12, D13, D14, and E52X have been described previously (13, 43, 54). Additional mutant viruses E58X, A8X, and A78 have lesions as described in the text and Table 1; their detailed characterization will be described elsewhere. All viruses were grown and titrated in baby hamster kidney (BHK) cells propagated in Glasgow modified Eagle’s medium (MEM) containing penicillin (100 U/ml), and streptomycin (100 μg/ml) and supplemented with 10% newborn calf serum and 10% tryptose phosphate broth. HEp-2 cells were grown in Dulbecco’s MEM supplemented with 10% fetal calf serum and antibiotics as described above. HFL cells were grown in Dulbecco’s MEM supplemented with 5% fetal calf serum, 5% newborn calf serum, 1% nonessential amino acids, and antibiotics as described above.

Electroporation.

HEp-2 cells were trypsinized, resuspended in complete medium, pelleted, and washed twice with serum-free medium before being resuspended in serum-free medium at a concentration of 6 × 10⁶ cells per ml. Plasmid DNA (20 μg) and 0.8 ml of cells were added to a 4-mm electroporation cuvette, incubated on ice for 10 min then mixed again, and pulsed in a Hybaid electroporator at a setting of 400 V. The cells were incubated on ice for a further 10 min before being diluted into fresh complete medium and seeded into Linbro wells at a nominal density of 2 × 10⁵ cells per well for Western blotting samples and half of that for immunofluorescence. The cells were used for experimentation the following day.

Infections and cell extracts.

HEp-2 cells, seeded in Linbro wells the previous day, were infected with HSV-1 by adsorption in 0.1 ml of medium at multiplicities as indicated in the figure legends. After 1 h, an additional 0.5 ml of medium was added and infection was allowed to proceed at 37°C. When lactacystin lactone (Boston Biochem Inc.) or MG132 (Calbiochem) was used, the cells were preincubated with medium containing the drug for 30 min before infection; then all subsequent incubations were performed in the presence of drug. Samples for use in immunofluorescence were treated as described below. Samples for Western blotting were washed in phosphate-buffered saline (PBS) and then resuspended directly into sodium dodecyl sulfate (SDS)-gel loading buffer. When proteasome inhibitors had been used, the loading buffer was supplemented with 5 mM N-ethylmaleimide.

Western blotting.

SDS-polyacrylamide gels (7.5% acrylamide) were prepared and run in the Bio-Rad MiniProtean II apparatus, and then proteins were electrophoretically transferred to nitrocellulose membranes according to the manufacturer’s recommendations. After blocking in PBS containing 0.1% Tween 20 (PBST) and 5% dried milk overnight at 4°C, the membranes were incubated with primary antibody in PBST–5% dried milk at room temperature for 4 h and then washed in PBST at least six times before incubation with horseradish peroxidase-conjugated secondary antibody in PBST–2% dried milk at room temperature for 1 h. After extensive washing, the filters were soaked in Amersham ECL (enhanced chemiluminescence) reagent and exposed to film. Antibodies were stripped from the membranes following the Amersham ECL protocol, and the membranes were reprobed as necessary.

Antibodies.

Anti c-Myc monoclonal antibody (MAb) 9E10 was purchased from Santa Cruz Biotechnology Inc. Anti-PML MAb 5E10 (55) was a generous gift from R. van Driel. The anti-UL42 and anti-Vmw110 MAbs Z1F11 and 11060 and the anti-PML rabbit serum r8 have been described previously (3, 15, 51). Horseradish peroxidase-conjugated sheep anti-mouse and goat anti-rabbit antibodies were purchased from Sigma Immunochemicals.

Immunofluorescence.

HEp-2 cells were fixed with formaldehyde (5% [vol/vol] in PBS containing 2% sucrose) and permeabilized with 0.5% Nonidet P-40 in PBS with 10% sucrose. The primary antibodies were diluted in PBS containing 1% newborn calf serum. Anti-Vmw110 MAb 11060 and anti-PML rabbit serum r8 were used at dilutions of 1/2,000 and 1/1,000 respectively. After incubation for 1 h, the coverslips were washed several times in PBS–1% calf serum and then incubated with goat anti-mouse fluorescein isothiocyanate-labeled and goat anti-rabbit tetramethylrodamine isothiocyanate-labeled secondary antibodies (Sigma) at dilutions of 1/100. After staining, the coverslips were mounted and examined by using appropriate narrow band filters in a Nikon Microphot-SA microscope adapted with a Digital Pixel CCD digital camera. Captured images were prepared for printing by using Photoshop.

Cleavage of ubiquitin and PIC1 fusion proteins in bacteria.

E. coli Novablue (DE3) strains carrying the substrate plasmids pPIC1GST and pRB307 were established, and derivatives which carried in addition either pACYC-HAUSP or pACYC-UBP2 were made. The former two plasmids express ampicillin resistance on pBR322 replicons and are therefore compatible with the pACYC184 replicons expressing chloramphenicol resistance of the latter two plasmids. The bacteria were grown to mid-log phase, and expression of substrate and enzyme activities were induced by the addition of isopropyl-β-d-thiogalactopyranoside (IPTG). The bacteria were harvested 2 h later. With the PIC1-GST substrate, whole-cell extracts were analyzed on a 12.5% polyacrylamide gel and the Myc-tagged proteins were detected by Western blotting using MAb 9E10. With the tag at the N-terminal end, this detects the PIC1-GST fusion protein and any liberated N-terminal fragments but not the C-terminal GST polypeptide. To analyze cleavage of the Ub-GST fusion protein, soluble protein extracts were prepared and the Ub-GST substrate and GST product were purified on glutathione-agarose beads as described previously (42). The purified proteins were analyzed on 10% polyacrylamide gels and detected by staining with Coomassie blue.

Detection of a PIC1-RanGAP1 hydrolase activity in Xenopus egg extracts and immunodepletion of a HAUSP-related Xenopus protein.

A GST-PIC1 fusion protein expression construct was produced by using a pGEX-2TK vector (Pharmacia) and a PIC1 cDNA (obtained from the IMAGE consortium [35]) that had been mutated to truncate the last four amino acids from its C terminus (49). Full-length PIC1 must be processed to remove the C-terminal four amino acids in order to produce a form of the protein that can be conjugated to target substrates (26). The truncated form of GST-PIC1 is readily conjugated to substrate protein in reticulocyte lysates and Xenopus egg extracts (49). Bacterial expression and purification of GST-PIC1 were carried out as described previously (49). Membrane-free soluble Xenopus protein extracts were fractionated from interphase egg extracts as described previously (8, 52). Samples of the extracts (50 μl) were depleted of HAUSP by using 5 μl of anti-HAUSP rabbit serum r201 and 20 μl of protein A-Sepharose beads; after incubation at 4°C for 1 h, the beads were removed by centrifugation. Mock depletion was carried out by using the same protocol, replacing r201 with the corresponding preimmune serum. The extracts were stored at −80°C. To detect RanGAP1-PIC1 isopeptidase activity using GST-PIC1, a Xenopus RanGAP1 cDNA (48) was cloned into a pT7 vector to give an open reading frame initiated at the original methionine start codon. Radiolabeled RanGAP1 protein was produced by using a TnT-coupled transcription-translation kit (Promega) in the presence of GST-PIC1 (50 ng/μl) and l-[³⁵S]methionine for 1.5 h at 30°C. The products of the reaction were purified on glutathione-Sepharose beads and washed extensively with PBS with 0.1% Triton X-100. Isopeptidase activity was assayed by incubation of 5 μl of the glutathione-Sepharose beads carrying GST-PIC1-RanGAP1 conjugates with 2 μl of untreated extract, mock-depleted extract, or HAUSP-depleted extract for 10 min at 30°C. The products were immediately analyzed on a SDS–4 to 20% polyacrylamide gel followed by autoradiography.

HSV-1 infection causes the rapid loss of high-molecular-weight isoforms of PML.

Infection by HSV-1 results in the rapid disruption of ND10 and the apparent dispersal of the constituent proteins (16, 38). While it has been clear that PML-derived immunofluorescence in infected cells is substantially reduced, the fate of the PML protein was uncertain. Because two reports have shown that both the PML-RARα fusion protein (59) and PML itself (60) are subject to degradation following treatment of cells with all-trans retinoic acid and arsenic oxide, respectively, we set out to determine if HSV-1 infection also caused the actual loss of PML. The primary transcript of the PML gene is subject to extensive alternative splicing, and the encoded proteins are subsequently modified posttranslationally to produce a plethora of isoforms with gel mobilities varying from about 65 to over 180 kDa (45, 55). These can be detected by Western blotting using MAb 5E10, which recognizes an epitope in a region of PML present in most isoforms (31, 55).

HEp-2 cells were infected with HSV-1 at increasing multiplicities of infection and harvested 4 h later. Western blotting showed that many multiple high-molecular-weight PML isoforms were lost or substantially reduced following virus infection and that the effect could be seen at a moderate virus dose of 5 PFU per cell (Fig. 1A). The reduction in the band intensities was substantial even as early as 2 h postinfection (data not shown), and it is pertinent that this timing correlates well with the observed virus-induced disruption of ND10 (38). It is noteworthy that these results are not an artifact arising from the use of MAb 5E10 since a similar pattern of high-molecular-weight bands is recognized by anti-PML rabbit serum r8 and these bands are also eliminated during virus infection (data not shown). There appears to be some isoform- or modification-dependent specificity to this process since major PML isoforms at about 70 kDa, and particularly the major bands at about 130 kDa, were not as greatly affected. However, appropriate exposure of the blots indicated that these bands also decreased in intensity in response to virus infection (see also Fig. 5).

An important question concerns the nature of these endogenous PML bands. The complexity of the PML isoforms detected on Western blots, the low amount of total PML in the cell, and the difficulty of isolating it in soluble form have hampered attempts to study the nature and biochemical properties of endogenous PML. However, three recent studies used various methods to induce increased expression of PML (including induction of endogenous PML with interferon, induction from stably integrated or transiently transfected genes, or use of a cell line engineered to express PML at high levels), with or without simultaneous increased expression of PIC1. All three describe the formation of a number of high-molecular-weight PML-PIC1 conjugate bands with gel mobilities that appear extremely similar to the pattern of endogenous high-molecular-weight PML bands shown in Fig. 1 (27, 44, 53). Importantly, a similar pattern of endogenous PML bands was found in untreated cells after prolonged exposure of the Western blots (53), conditions which closely parallel those used in our experiments. Accordingly, comparison with these other studies leads to the reasonable conclusion that the high-molecular-weight isoforms of PML that are lost during virus infection represent PML-PIC1 conjugates. A more direct proof of this conclusion would require immunoprecipitation of the endogenous PML-PIC1 conjugates, but the poor solubility of the protein, its low amounts in the cell, and the fact that many PIC1 conjugates appear to be extremely labile in vitro have so far made this experiment impossible.

Expression of Vmw110 and its ability to interact normally with ND10 are essential for efficient loss of the PML isoforms.

Since the HSV-1 IE protein Vmw110 initially colocalizes with PML at ND10 and then causes their disruption, it seemed possible that Vmw110 is involved in the loss of the PML isoforms. HEp-2 cells were infected with wild-type and several Vmw110 mutant viruses, whose genotypes are explained in Table 1 and depicted in Fig. 2C. The cells were harvested 4 h after infection and analyzed by Western blotting. The results (Fig. 2A) clearly show that (i) Vmw110 is absolutely essential for the loss of the high-molecular-weight isoforms of PML (because dl1403 had no effect), (ii) the RING finger domain of Vmw110 is also essential (virus FXE had no effect), (iii) the complete C-terminal region of Vmw110 containing the HAUSP binding and ND10 localization sequences is also very important (virus E52X had little effect), (iv) the ability of Vmw110 to migrate to ND10 is required for normally efficient loss of the PML isoforms (viruses E58X, A8X, D13, and D14 were all less active than wild type), but (v) Vmw110 mutants which bind HAUSP poorly yet migrate to ND10 are still capable of inducing the loss of the PML isoforms, although somewhat less efficiently than wild-type Vmw110 (compare viruses D12 and A78 with wild type). Probing of the same filter with MAb 11060 indicated that the mutant viruses expressed similar levels of Vmw110 (Fig. 2B). An important factor of this experiment is that the differences between the mutant viruses is partly kinetic, in that virus E52X (for example) induces increased loss of the PML isoforms as infection progresses. However, viruses dl1403 and FXE were completely defective in this assay, even late in infection (data not shown). These results generally correlate well with the ability of the mutant forms of Vmw110 to disrupt ND10 early in virus infection (39, 43; see also below). The possible role of HAUSP in this process is discussed further below.

Involvement of the proteasome in the loss of the PML isoforms.

Since the loss of the high-molecular-weight PML isoforms occurred rapidly, it seemed likely that this was a proteasome-dependent process. The ubiquitin-mediated protein degradation pathway is being increasingly recognized as a common mechanism of regulating the amount of a protein within the cell, and its activation can lead to very rapid elimination of the target protein (reviewed in references 20 and 57). The pathway involves initial conjugation of a ubiquitin monomer to the target protein, followed by the polymerization of the ubiquitin chain to produce a tag which targets the substrate to the proteasome for rapid degradation. Inhibition of the proteolytic activity of the proteasome results in the accumulation of multiubiquitinated polypeptides.

Addition of the proteasome inhibitors lactacystin lactone or MG132 to the culture medium eliminated the virus-induced loss of the PML isoforms (Fig. 3). Although there was no evidence of accumulation of a ladder of ubiquitinated forms of PML in the drug-treated samples in response to viral infection, it is possible that their appearance is masked by the size and complexity of the various isoforms. Probing of the same blot with MAb Z1F11 indicated that at the multiplicity of virus infection used in this experiment, the inhibitors had no effect on the efficiency of expression of UL42, a viral protein of the early kinetic class (data not shown). This finding demonstrates that the effect of MG132 was not due to interference with viral gene expression.

Proteasome inhibitors stabilize ND10 in virus-infected cells.

The correlation between the virally induced destruction of ND10 and the Vmw110-induced, proteasome-mediated loss of the high-molecular-weight isoforms of PML suggests that proteasome activity may be essential for the effect of the virus on ND10. HEp-2 cells were infected with HSV-1 in the presence and absence of MG132 and lactacystin lactone and stained 4 h later for PML and Vmw110. Comparison of the infected and uninfected cells in each situation clearly showed that both proteasome inhibitors greatly reduced the ability of the virus to disrupt ND10, thus providing a direct mechanistic link between the loss of the PML isoforms and the ND10 disruption. In the absence of drug, the uninfected cells (upper right and middle-leftmost cells in Fig. 4A) showed the normal punctate accumulations of PML in ND10, while most of the infected cells had no trace of ND10. In the presence of lactacystin lactone, the infected cells clearly retained ND10 (Fig. 4C), and this effect was even more striking in the presence of MG132 (Fig. 4E). Interestingly, MG132 by itself clearly affected ND10, causing an increase in their number and prominence (compare the uninfected cells in Fig. 4A and E). Virus infection in the presence of MG132 led to extensive colocalization of PML with Vmw110 (Fig. 4E and F), a situation reminiscent of RING finger mutant forms of Vmw110 which migrate to ND10 but do not disrupt them. Taken with the results of Fig. 3, these observations suggest that the loss of the PML isoforms correlates with and probably precedes ND10 disruption.

Virus-induced disruption of ND10 correlates with the loss of high-molecular-weight PML isoforms in different cell types.

Many of the results presented above suggest an attractive model which would connect the loss of the PML isoforms and the proteasome-dependent disruption of ND10 by Vmw110 with its ability to bind to the ubiquitin-specific protease HAUSP. However, Fig. 2 shows that the role of HAUSP in this effect is not clear, since the HAUSP binding-deficient mutants D12 and A78 retain the ability to stimulate the loss of the PML bands, at least in HEp-2 cells. In a previous publication we showed that the HAUSP binding-negative mutant D12 was defective in the disruption of ND10 in BHK cells (43), and at first sight this result is not consistent with either the loss of the PML isoforms or the time course of ND10 disruption in HEp-2 cells infected with virus D12 (Fig. 2 and data not shown). Therefore we have investigated the correlation between loss of the PML isoforms and ND10 disruption using the HAUSP binding-negative mutant D12 in different cell types. In these experiments, we compared the results for primary human fibroblasts (HFL cells) with those for the BHK cells that had been used previously. The wild-type virus efficiently disrupted ND10 in both cell types, and this correlated with the loss of the PML isoforms (Fig. 5A to F). Figures 5C and D show a group of six cells, three of which have been infected. The PML staining indicates clear loss of ND10 from the infected cells but not from their uninfected neighbors. Similarly, compare the three infected HFL cells (the middle one of which expresses only very low amounts of Vmw110) with the uninfected cell at the top of Fig. 5D. In contrast, virus D12 had a markedly reduced ability both to disrupt ND10 and induce the loss of the isoforms in BHK cells but not HFL cells (Fig. 5A, B, and G to J). Figure 5A shows that virus D12 induced only a marginal loss of the PML isoforms in BHK cells, and this correlated with the retention of ND10 in most cells 4 h after infection (for example, the lowest cells in Fig. 5G and H). In contrast, all D12-infected HFL cells, even those with expressing low amounts of the D12 protein (compare the lowest two cells in Fig. 5I and J), had lost ND10. The Western blot analysis performed in parallel showed that D12 was as efficient as the wild type in stimulating the loss of the PML bands from HFL cells (Fig. 5B). A feature of the D12 phenotype is an increase in the rate at which the protein accumulates in the cytoplasm (Fig. 5G). However, the failure to disrupt ND10 and stimulate PML loss is not due this property since at earlier times of infection the localization of D12 mutant protein in the nucleus and at ND10 is similar to that of the wild-type protein in HEp-2, BHK, and HFL cells (reference 43 and data not shown). These results further underline the correlation between ND10 disruption and PML isoform loss, and they suggest that the ability of Vmw110 to bind to HAUSP contributes to these effects in a cell-type (or species)-dependent manner. It should be noted that BHK cells express a protein of the same size and antigenicity as HAUSP which also binds to Vmw110 (reference 42 and results not shown).

The reduced ability of virus D12 to disrupt ND10 and induce the degradation of the PML isoforms in BHK but not HFL cells correlates with its plaque-forming abilities in the two cell types; unlike most Vmw110 mutant viruses, it forms plaques more efficiently in HFL than BHK cells (data not shown).

HSV-1 infection results in the loss of several high-molecular-weight PIC1 conjugates.

Given that the high-molecular-weight PML isoforms which disappear during HSV-1 infection have gel mobilities extremely similar to those of defined PML-PIC1 conjugate proteins, and that PIC1 can also be conjugated to a number of other proteins (26), we decided to investigate the fate of other PIC1 conjugates during virus infection. Plasmid pCIPIC1 (which expresses an epitope-tagged version of PIC1; see Materials and Methods) was electroporated into HEp-2 cells which were subsequently infected with wild-type virus, the Vmw110 deletion mutant dl1403, and the RING finger mutant FXE. Samples were taken at 4, 8, and 22 h after infection and analyzed by Western blotting.

Uninfected electroporated cells expressed a PIC1 protein of about 18 kDa (the size expected considering the gel mobility of the endogenous monomeric PIC1 plus the epitope tag) and a number of higher-molecular-weight bands culminating in a complex pattern of a multitude of high-molecular-weight polypeptides (Fig. 6). The prominent band at about 90 kDa probably corresponds to a form of RanGAP1 which is covalently conjugated to PIC1 (36, 37), while the higher-molecular-weight bands have been identified as other PIC1 conjugates (26).

In response to HSV-1 infection, many of the high-molecular-weight PIC1 conjugate bands disappeared (Fig. 7). This effect was slower than that seen with the high-molecular-weight isoforms of PML, becoming almost complete by 22 h after infection; however, it should be noted that these cells express artificially high levels of PIC1, and so endogenous PIC1 conjugates in untreated cells may be eliminated much faster. It is evident that there is a degree of specificity to this process since there was no reduction in a number of the PIC1 bands, including that likely to correspond to RanGAP1. This result is consistent with the observation that the PIC1 conjugation status of endogenous RanGAP1 (as determined by Western blotting with a RanGAP1-specific antibody) is not affected during virus infection (data not shown). The results with the Vmw110 mutant viruses showed that this effect was again totally dependent on Vmw110 and the integrity of its RING finger (Fig. 7). The same set of virus mutants as used for Fig. 2 was analyzed in a similar experiment, and all were less active than the wild type in reducing the PIC1 conjugate bands (data not shown).

The loss of the PIC1 conjugates was inhibited by MG132 (Fig. 8), an observation which suggests that a similar mechanism may operate for both this effect and the Vmw110-induced loss of the PML isoform bands. A striking feature of this experiment was that MG132 resulted in greatly increased levels of the PIC1 conjugate bands. This result implies that these polypeptides are subject to turnover, directly or indirectly, by a proteasome-dependent pathway. This point is discussed further below. Lactacystin lactone did not inhibit the degradation of the PIC1 conjugates, but the drug would be expected to be hydrolyzed to an inactive form early in the relatively long time course of this experiment (10).

Does HAUSP cleave PIC1 conjugates?

Because of the observed colocalization of PIC1 and HAUSP in ND10, and the direct implication of the ubiquitin pathway in the stability of PIC1 conjugates and ND10 themselves, it is a reasonable and intriguing question whether PIC1 conjugates are substrates for HAUSP. Indeed, precisely this question has been posed by other workers in two recent publications (23, 44). We have taken four approaches to this question. First, we asked whether HAUSP could bind to immobilized PIC1 in vitro, but the results were negative (data not shown). Second, there was no evidence that HAUSP could cleave the C-terminal four residues from the primary PIC1 translation product in an assay where clear cleavage of the analogous ubiquitin precursor by HAUSP was evident (data not shown). Third, we constructed a PIC1-GST model substrate expression plasmid and compared the ability of HAUSP to cleave it in parallel with an equivalent Ub-GST model substrate. The results clearly showed that while HAUSP cleaved Ub-GST, it could not cleave PIC1-GST (Fig. 9A and B). It is possible that cleavage of PIC1 conjugates by HAUSP is context dependent, requiring a natural PIC1 conjugate or an isopeptide linkage. We were able to test these possibilities in one instance since there is an activity in Xenopus egg extracts which efficiently cleaves PIC1 from RanGAP1, and also a protein which is strongly related to HAUSP in terms of size, immunoprecipitation by anti-HAUSP r201 serum (Fig. 9D), and interaction with the C-terminal region of Vmw110 (data not shown). Depletion of this protein from egg extracts by immunoprecipitation had no effect on the PIC1 cleavage activity (Fig. 9C and D), and there was no increased activity in the HAUSP immunoprecipitate (data not shown). However, these results do not exclude the possibility that any hypothetical PIC1 cleavage activity of HAUSP is both context dependent and highly substrate specific.