Tuberous Sclerosis Complex Protein 2-Independent Activation of mTORC1 by Human Cytomegalovirus pUL38

Plasmids and reagents.

LentiCRISPR (plasmid 49535; Addgene) is a lentiviral vector expressing cas9 and single guide RNA (sgRNA) (32). The following lentiviral overexpression vectors were derived from pLKO.DCMV.tetO (33). pLKO-HA-UL38, expressing the UL38 open reading frame, and pLKO-HA-UL38_25–331, expressing a pUL38 segment from amino acid 25 to amino acid 331, both carried a hemagglutinin (HA) epitope at their amino termini. pLKO-Flag-UL38 and pLKO-Flag-UL38_25–331 both carried a 3× Flag tag at their N termini. Each pair of amino acids between amino acids 13 and 24 of pUL38 were mutated to alanine by overlapping PCR and cloned into pLKO.DCMV.tetO vector, namely, pLKO-HA-UL38 SL/AA, pLKO-HA-UL38 LD/AA, pLKO-HA-UL38 E/A, pLKO-HA-UL38 EW/AA, pLKO-HA-UL38 RQ/AA, and pLKO-HA-UL38 TQ/AA. Primers for introducing these mutations are listed in Table 1. pLKO-Flag-GFP expressed the green fluorescent protein (GFP) gene with a 3× Flag tag at its N terminus. pLKO-TSC2 expressed the TSC2 consensus coding sequence (CDS), and pLKO-Flag-TSC2 expressed TSC2 with a 3× Flag tag at its N terminus.

Primary antibodies used in the present study included anti-actin (AC15; Abcam); anti-immediate-early protein IE1/2 (a generous gift from Jay Nelson, Oregon Health & Science University); anti-IE1 (23), anti-pUL38 (23), anti-pp28 (34), and anti-pp71 (35) (gifts from Thomas Shenk, Princeton University); anti-ribosomal S6 kinase (S6K) (Proteintech); anti-cleaved poly(ADP-ribose) polymerase (PARP) (Cell Signaling); anti-phosphorylated S6K (Thr-389) (Cell Signaling); anti-HA and anti-Flag (Abmart); anti-TSC2 (1824-1; Abcam); and anti-TSC1 (4906; Cell Signaling).

Viruses and cell lines.

Primary human foreskin fibroblasts (HF), human glioblastoma-astrocytoma cell line U373-MG (kindly provided by James Alwine, University of Pennsylvania), and HEK293T cells were maintained at 37°C in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS). HEK293T cells were transfected with the indicated plasmid together with the packaging plasmids 9.2ΔR and VSV-G to produce lentivirus stocks. HF cells or U373-MG cells were transduced with lentivirus supplemented with 5 μg/ml Polybrene and selected with 2 μg/ml puromycin (Sigma-Aldrich) to generate a pool of cells expressing the protein of interest.

Bacterial artificial chromosome (BAC)-HCMV clones were created and used to make recombinant HCMV. HA-UL38, HA-UL38_12–331, HA-UL38_25–331, and HA-UL38_35–331 were amplified by PCR, and we inserted them into the BAC pAD-GFP to replace the full-length UL38 by linear recombination in the bacterial strain SW105 as previously reported (36). The primers used for linear recombination are available upon request. pADinHA-UL38, pADtrunHA-UL38_12–331, pADtrunHA-UL38_25–331, and pADtrunHA-UL38_35–331 were used to reconstitute ADinHA-UL38, ADtrun HA-UL38_12–331, ADtrun HA-UL38_25–331, and ADtrunHA-UL38_35–331 mutant viruses in which the N-terminal amino acid residues of pUL38 were deleted as indicated. pAD-GFP was used to produce wild-type virus (ADwt), and pADpmUL38, a recombinant BAC derived from pAD-GFP in which the expression of pUL38 was disrupted, was used to produce ADpmUL38 as previously reported (23, 30).

To reconstitute the virus, 2 μg of the BAC DNA and 1 μg of the pp71-overexpressing plasmid were transfected into HF cells as described previously (37). Twenty-four hours later, culture medium was replaced. The supernatant was collected when ∼100% of the cell monolayer was lysed.

Viral growth analysis.

HF cells or HF cells expressing a protein of interest were seeded in 12-well plates overnight, and cells were incubated with recombinant HCMV for 1 h at an MOI of 0.01 or 3. The inoculum was removed, the monolayer was washed with phosphate-buffered saline (PBS), and fresh medium was added. Medium from infected cells was harvested at the indicated times. The virus titer was determined by 50% tissue culture infectious dose (TCID₅₀) assay in HF or HF-HA-UL38 cells.

Protein analysis.

Proteins were analyzed by immunoblotting as previously described (30). Briefly, cells were washed with phosphate-buffered saline (PBS) and lysed in the protein sample buffer supplemented with protease inhibitor cocktails (PIC; Roche) and phosphatase inhibitor cocktail IV (PHIC; BioVision, Inc.). Proteins were resolved by electrophoresis and transferred to a polyvinylidene difluoride membrane, and then they were hybridized with primary antibodies and horseradish peroxidase-conjugated secondary antibodies and visualized by using enhanced chemiluminescence (ECL) reagents (Bio-Rad).

Immunoprecipitation.

Immunoprecipitations were done either in HF cells infected with HCMV at an MOI of 3 at 72 hpi or in HEK293T cells transfected with plasmids of interest. Cell samples were lysed in lysis buffer, containing 120 mM NaCl, 40 mM HEPES, 1 mM EDTA, 0.3% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), PIC, and PhIC, for 1 h and centrifuged at 16,000 × g for 15 min at 4°C. The supernatants were incubated with Flag antibody-conjugated beads (Sigma-Aldrich) for 1 h at room temperature in rotation, washed with wash buffer (120 mM NaCl, 40 mM HEPES, 1 mM EDTA, 0.3% CHAPS), and eluted with 200 μg/ml flag peptide. The eluents were analyzed by immunoblotting.

Assays for mTORC1 activation.

Immunoblotting of phosphorylated S6K was used in this study as the readout for mTORC1 activation. To determine mTORC1 activation in infected cells, HF cells were infected with various HCMV at an MOI of 3 and collected at various hours postinfection for immunoblot analysis. To determine mTORC1 activation in expressing cells, HF cells or U373-MG cells overexpressing proteins of interest were serum starved for 72 h or 48 h, respectively. Cell lysates were collected for immunoblot analysis.

shRNA-mediated TSC2 knockdown in HF cells.

The control (shCtrl) short hairpin RNA (shRNA) and TSC2-targeting shRNA (shTSC2-1 or shTSC2-2) expressing lentiviral vectors were constructed based on the PLKO.1 vector. The shRNA targeting sequences are 5′-CAA CAA GAT GAA GAG CAC CAA-3′ (shCtrl), 5′-CAA TGA GTC ACA GTC CTT TGA-3′ (shTSC2-1), and 5′-GGA TTA CCC TTC CAA CGA AGA-3′ (shTSC2-2). Lentivirus preparation and transduction were the same as those described above for viruses and cell lines. Pools of HF cells stably expressing the respective shRNAs were used as control or TSC2 knockdown cells.

Generation of TSC2 knockout U373-MG cell lines.

Genome-engineering experiments using the CRISPR-Cas9 system were performed as described previously (38). Briefly, two sgRNAs targeting TSC2 were cloned into lentiCRISPR vector. Lentivirus was produced in HEK293T cells. U373-MG cells were transduced with lentivirus and selected with 2 μg/ml puromycin. TSC2 knockout cell clones were isolated by limited serial dilution. Finally, TSC2 knockout cell clones were confirmed by sequencing and immunoblot analysis. The sgRNA sequences targeting TSC2 were 5′-CAC CGT GGC CTC AAC AAT CGC ATC-3′ (site 1) and 5′-CAC CGA GCA CGC AGT GGA AGC ACT C-3′ (site 2) (32).

Construction and analysis of serial N-terminal-truncated mutants of HCMV pUL38.

We previously identified the domains or motifs responsible for the distinct activities of HCMV protein pUL38 by mutagenesis (31). Deletion of the first 56 residues was associated with rapid degradation of the protein, making it hard to dissect the structural and functional relationships of this region. Therefore, we generated mutant viruses expressing small, serial N-terminal deletions of pUL38, with the aim of obtaining stable mutant proteins with altered functions in the context of HCMV infection. Mutations were introduced into the virus genome using a bacterial artificial chromosome (BAC) recombineering technique (36). Recombinant virus ADinHA-UL38 carried the entire UL38 open reading frame (ORF) with an HA tag on its N terminus, ADtrunHA-UL38_12–331 carried the UL38 ORF lacking the 11 N-terminal amino acids, ADtrunHA-UL38_25–331 carried the UL38 ORF lacking the 24 N-terminal amino acids, and ADtrunHA-UL38_35–331 carried the UL38 ORF lacking the 34 N-terminal amino acids. All of the truncated variants of pUL38 were tagged with HA at their N termini to facilitate protein detection. The parental HCMV-BAC clone pAD-GFP, derived from the HCMV AD169 strain (23), was used to produce wild-type virus (ADwt), which was used as a positive control. The integrity of the recombinant BAC clones was examined by digestion with EcoRI and BamHI (Fig. 1A and B). The digestion patterns were consistent with the predictions, indicating that the recombinant BAC clones carried the intact viral genome and the precise modifications at the correct loci.

The recombinant viruses were reconstituted, and multistep viral growth assays were performed to detect the effects of the serial pUL38 N-terminal truncations on viral growth. ADinHA-UL38 and ADtrunHA-UL38_12–331 replication was similar to that of the wild-type virus, ADwt (Fig. 1C). However, ADtrunHA-UL38_25–331 replication was about 10-fold lower than that of the wild-type virus, and replication was almost absent from ADtrunHA-UL38_35–331. The growth defect of ADtrunHA-UL38_25–331 and the wild-type replication kinetics of ADtrunHA-UL38_12–331 indicated the importance of amino acids 12 to 24. In addition, the dramatic growth defect of ADtrunHA-UL38_35–331 suggested that this variant lost pUL38 function and that amino acids 25 to 34 were necessary for functional pUL38 and viral growth.

Given that pUL38 can inhibit premature cell death and antagonize TSC2 to facilitate HCMV replication, we determined if the truncated mutant proteins pUL38_12–331, pUL38_25–331, and pUL38_35–331 retained the ability to block cell death during virus infection. Immunoblot analysis showed that pUL38_12–331 was accumulated to a slightly higher level than HA-tagged pUL38 and pUL38_25–331 protein was expressed at a level similar to that of HA-tagged pUL38, but pUL38_35–331 was almost undetectable during infection (Fig. 2A). As expected, ADinHA-UL38 inhibited premature cell death, unlike ADpmUL38 (23, 30), a mutant virus that does not express pUL38 (Fig. 2B). The cell death morphology is indicated by more round floating cells and fewer cells adhering to the dishes. ADtrunHA-UL38_12–331 and ADtrunHA-UL38_25–331 also inhibited cell death, while ADtrunHA-UL38_35–331 induced cell death late postinfection (Fig. 2B). Consistent with these results, PARP cleavage was inhibited in HF cells infected with ADinHA-UL38, ADtrunHA-UL38_12–331, and ADtrunHA-UL38_25–331 but induced in cells infected with ADtrunHA-UL38_35–331 and ADpmUL38 (Fig. 2A). Thus, the phenotype of ADtrunHA-UL38_35–331 virus mimics that of ADpmUL38, probably as a result of the loss of pUL38 protein expression. Therefore, in the rest of the manuscript, we focused on pUL38_25–331, which was sufficiently expressed and blocked cell death (Fig. 2) but resulted in a 10-fold reduction in virus growth (Fig. 1C).

pUL38_25–331 induces S6K phosphorylation without binding to TSC2 during HCMV infection.

We previously showed that pUL38 suppressed cell death independently of its ability to induce mTORC1 activation (31). ADtrunHA-UL38_25–331 virus retained the ability to block cell death, but virus growth was reduced by about 10-fold. Therefore, we determined if the N-terminal-truncated mutants of pUL38 retained the ability to activate mTORC1. We used S6K phosphorylation (phos-S6K) as an indicator of mTORC1 activation, as described previously (31). The ADpmUL38 mutant virus induced apparent S6K phosphorylation, consistent with our previous report that HCMV could activate mTORC1 in a pUL38-independent manner (31). However, ADinHA-UL38 and ADtrunHA-UL38_12–331 viruses induced S6K phosphorylation to much greater levels, indicating that pUL38 was required for full mTORC1 activation and that pUL38_12–331 was fully capable of activating mTORC1 (Fig. 3A). ADtrunHA-UL38_25–331 virus induced S6K phosphorylation at levels greater than those by ADpmUL38 but lower than those by ADinHA-UL38, suggesting only partial pUL38-dependent activation of mTORC1 (Fig. 3A). pUL38 was reported to induce mTORC1 activation through binding to TSC2 (14). Therefore, we examined the interactions of pUL38 and its mutants with TSC2 during virus infection (Fig. 3B). As expected, HA-tagged full-length pUL38 (HA-UL38) and pUL38_12–331 pulled down TSC2 effectively. However, pUL38_25–331 failed to pull down TSC2 in the assay, despite being able to induce S6K phosphorylation.

We characterized the ADtrunHA-UL38_25–331 mutant virus further by comparing mTORC1 activation and viral protein expression of HA-UL38 and HA-UL38_25–331 at multiple time points. Both ADinHA-UL38 virus and ADtrunHA-UL38_25–331 mutant virus induced S6K phosphorylation, but ADtrunHA-UL38_25–331 mutant virus caused less S6K phosphorylation at all time points than ADinHA-UL38 virus infection (Fig. 3C). Therefore, reduced mTORC1 activation by the ADtrunHA-UL38_25–331 mutant virus was not time dependent, suggesting that during infection pUL38 was able to activate mTORC1 independently of TSC2 binding but that maximal mTORC1 activation required its interaction with TSC2. Viral immediate-early protein (IE1/2) and early protein (pUL38) expression levels were similar during infection with mutant ADtrunHA-UL38_25–331 and ADinHA-UL38 viruses (Fig. 3D). The expression levels of late proteins (pp28 and pp71) were slightly reduced during ADtrunHA-UL38_25–331 mutant virus infection compared with those for ADinHA-UL38 infection (Fig. 3D). We further compared the virus growth of ADtrunHA-UL38_25–331 and ADinHA-UL38 at high or low multiplicity of infection (MOI). As shown in Fig. 3E, ADtrunHA-UL38_25–331 grew almost as well as the wild-type virus expressing the full-length pHA-UL38 at an MOI of 3 but produced ∼10-fold less progeny virus at an MOI of 0.01. The data indicate that the N-terminal 24 amino acids of pUL38, its interaction with TSC2, and the full activation of mTORC1 are required for optimal virus growth in an MOI-dependent manner.

pUL38_25–331 activates mTORC1 independently of TSC2 binding when expressed in isolation.

The previous study demonstrates that pUL38 is sufficient to mediate TSC2-dependent activation of mTORC1 in isolation (14). We were interested in determining whether pUL38_25–331 also could mediate TSC2-independent activation of mTORC1 in the absence of any other viral proteins. To test this, we created HF cells stably expressing empty vector, pHA-UL38, or pHA-UL38_25–331. These cells were serum starved for 3 days and then collected to test for mTORC1 activity by detecting the levels of phos-S6K. Immunoblot analysis showed that pHA-UL38 was capable of inducing high levels of phos-S6K, indicating high mTORC1 activation, consistent with previous results (Fig. 4A) (31). Importantly, pHA-UL38_25–331 also was able to induce relatively high levels of phos-S6K compared to those of the empty vector control, although they were lower than those for pHA-UL38, suggesting that the mutant protein retained the ability to activate mTORC1. Furthermore, when coexpressed, full-length pUL38 and TSC2 could pull down each other while pUL38_25–331 (either HA or Flag-tagged) failed to do so (Fig. 4B and C). pUL38_25–331 did not pull down TSC1, a protein complexed with TSC2 to negatively regulate mTORC1, either (Fig. 4B). These results suggest that amino acids 1 to 24 are essential for the interaction between pUL38 and TSC2, and that pUL38, lacking these amino acids, still was able to activate mTORC1 outside the context of HCMV infection.

A TQ motif of pUL38 is required for pUL38-TSC2 interaction.

As pHA-UL38_12–331 interacts with TSC2 but pHA-UL38_25–331 failed to do so (Fig. 3B), we hypothesized that the TSC2-binding motif was located among residues 13 to 24 of pUL38. Therefore, we mapped the key amino acids required for TSC2 binding by constructing six recombinant clones of pUL38, sequentially replacing two neighboring amino acids at a time to alanines among residues 13 to 24. The mutants were SL/AA, LD/AA, E/A, EW/AA, RQ/AA, and TQ/AA, all of which were tagged with HA at their N termini to facilitate protein detection (Fig. 5A). Coimmunoprecipitation analysis showed that the mutants SL/AA, LD/AA, EW/AA, and RQ/AA still were able to combine with TSC2 (Fig. 5B). However, the E/A mutant demonstrated reduced binding to TSC2, while the TQ/AA mutant completely lost the ability to interact with TSC2 (Fig. 5B). These data indicate that the E residue is important and the TQ motif is necessary for pUL38 binding to TSC2. We tested if the TQ mutant that failed to bind to TSC2 was able to activate mTORC1. As shown in Fig. 5C, the levels of phos-S6K were increased after expressing pHA-UL38 TQ/AA and pHA-UL38_25–331 in HFs, although phosphorylation was less than that in cells expressing full-length HA-UL38 (Fig. 5C). As expected, when cells were treated with rapamycin, an mTORC1-specific allosteric inhibitor, the phos-S6K induced by pUL38 isoforms was completely blocked (Fig. 5D). These data demonstrated that the TQ motif was required for pUL38 to interact with TSC2, and that mTORC1 could be activated in a TSC2-binding-independent manner by pUL38 in isolation.

The site-specific mutants mimic the function of pHA-UL38_25–331.

In order to study the role of site-specific mutants (TQ/AA and E/A) in the context of virus infection, we made HF cells stably expressing these mutant proteins, infected them with pUL38-deficient virus ADpmUL38, and then tested virus growth, cell death, and mTORC1 activity. The full-length pUL38-expressing cells (HF-HA-UL38) known to complement ADpmUL38 were used as the positive control, and the cells expressing an empty vector (HF vector) were used as the negative control. As expected, pUL38-deficient virus ADpmUL38 was greatly growth attenuated in control cells (HF vector) but grew efficiently in cells expressing wild-type pUL38 (HF-HA-UL38) (Fig. 6A and B). At a low MOI of 0.01, the growth of pUL38-deficient virus was partially rescued in cells expressing pUL38 mutants (HF-UL38_25–331, HF-TQ/AA, and HF-E/A) with an ∼10-fold lower efficiency than that in HF-HA-UL38 cells (Fig. 6A). High-MOI growth curve analysis revealed the virus replicated equally well in all of the pUL38 mutant-expressing cells (Fig. 6B), suggesting that the TSC2 binding of pUL38 and the full activation of mTORC1 are not required for maximum HCMV replication on HFs on a single-step growth curve. The site-specific mutants inhibited cell death (Fig. 6C) and activated mTORC1 (Fig. 6D) similarly to pHA-UL38_25–331. Taking these findings together, we showed that the site-specific mutants unable to interact with TSC2 mimic the function of pHA-UL38_25–331.

pUL38 further increases mTORC1 activity in TSC2 knockout cells.

To provide additional evidence that pUL38 was able to activate mTORC1 independently of TSC2 interaction, we tested the ability of pUL38 to increase mTORC1 activity in the absence of TSC2. The CRISPR/Cas9 system (38) was used to knock out TSC2 in U373-MG human glioblastoma cells. To minimize off-target effects, we chose two sgRNAs (Fig. 7A) targeting TSC2 at two different gene locations, which were predicted to have minimal off-target activities by an optimized algorithm (32). Cas9- and TSC2-specific sgRNAs were introduced into U373-MG cells by lentiviral vector transduction, and TSC2-null cell lines were cloned by limited serial dilution, as described in Materials and Methods. Two clones, 1-17 and 2-7, derived from each sgRNA, were validated by sequencing the genome-targeting site (data not shown) and by immunoblotting (Fig. 7B). As shown in Fig. 7B, TSC2 protein expression was completely disrupted in both clones. The level of phos-S6K was dramatically upregulated in both clones compared to levels in cells expressing Cas9 only, consistent with a negative regulatory role of TSC2 on mTORC1. When pUL38 was expressed in TSC2-null cells, phos-S6K was further increased compared to levels in cells expressing empty vector (Fig. 7B and C). pUL38_25–331 and pUL38-TQ/AA behaved similarly to pUL38 in TSC2-null cells to further increase mTORC1 (Fig. 7C). To test if this also is true in HFs, we used RNA interference (RNAi) to knock down TSC2. CRISPR-mediated gene knockout requires single-cell cloning, which is not applicable to primary human fibroblasts. As shown in Fig. 7D, shTSC2-2 potently suppressed TSC2 expression, and pUL38 and its variants further increased S6K phosphorylation in the TSC2 knockdown HFs. The results indicate that pUL38 can activate mTORC1 through pathways other than antagonizing TSC2. Overall, we have identified residues in pUL38 critical for its binding to TSC2 and demonstrated that pUL38 was able to activate mTORC1 independently of TSC2 binding.