Human cytomegalovirus-encoded viral cyclin-dependent kinase (v-CDK) UL97 phosphorylates and inactivates the retinoblastoma protein-related p107 and p130 proteins

The human cytomegalovirus (HCMV)-encoded viral cyclin-dependent kinase (v-CDK) UL97 phosphorylates the retinoblastoma (Rb) tumor suppressor. Here, we identify the other Rb family members p107 and p130 as novel targets of UL97. UL97 phosphorylates p107 and p130 thereby inhibiting their ability to repress the E2F-responsive E2F1 promoter. As with Rb, this phosphorylation, and the rescue of E2F-responsive transcription, is dependent on the L1 LXCXE motif in UL97 and its interacting clefts on p107 and p130. Interestingly, UL97 does not induce the disruption of all p107-E2F or p130-E2F complexes, as it does to Rb-E2F complexes. UL97 strongly interacts with p107 but not Rb or p130. Thus the inhibitory mechanisms of UL97 for Rb family protein-mediated repression of E2F-responsive transcription appear to differ for each of the Rb family proteins. The immediate early 1 (IE1) protein of HCMV also rescues p107- and p130-mediated repression of E2F-responsive gene expression, but it does not induce their phosphorylation and does not disrupt p107-E2F or p130-E2F complexes. The unique regulation of Rb family proteins by HCMV UL97 and IE1 attests to the importance of modulating Rb family protein function in HCMV-infected cells.

effects of p130 Ser-672 phosphorylation on E2F binding or transcriptional repression have not been reported.
Rb is phosphorylated not only by cellular CDKs but also by viral (v) CDKs. The ␤ and ␥ families of the human herpesviruses encode v-CDKs that phosphorylate Rb. The v-CDK proteins are Epstein-Barr virus (EBV) BGLF4, human cytomegalovirus (HCMV) UL97, human herpesvirus type 6 (HHV-6) and type 7 (HHV-7) U69, and Kaposi's sarcoma-associated herpesvirus (KSHV) ORF36 (25). In addition to inactivation by phosphorylation, viruses also inactivate Rb family proteins through other mechanisms (26 -29). For example, the adenovirus E1a protein and the simian virus 40 (SV40) T antigen bind Rb family proteins and disrupt their complexes with E2Fs. In addition, papillomavirus E7 and HCMV pp71 bind to Rb family proteins and induce their proteasomal degradation. In each of these instances, the hypophosphorylated form of Rb is targeted, and the reaction requires a specific motif on the viral protein with an amino acid sequence LXCXE. This LXCXE motif (30) associates in a well defined binding cleft (a shallow groove on the B box) found in all three Rb family proteins (31).
Unlike the cellular CDKs, the HCMV v-CDK UL97 utilizes an LXCXE motif, not a hydrophobic patch, to phosphorylate Rb (32). UL97 has three LXCXE motifs, the first one of which (called L1) is required for full UL97-mediated Rb phosphorylation and for the ability of UL97 to reverse Rb-imposed transcriptional repression of E2F-responsive promoter constructs in reporter assays (32). A recombinant HCMV in which wildtype UL97 is replaced with an L1 mutant UL97 (L1m) where the central cysteine of L1 is changed to a glycine (C151G) does not have a growth defect and does not show diminished induction of E2F-responsive genes (32,33). This is likely because HCMV has multiple and possibly redundant mechanisms to inactivate Rb, p107, and p130 (26, 34 -38). Indeed, the HCMV immediate early 1 (IE1) protein cooperates with UL97-L1m to relieve Rbimposed E2F-responsive promoter repression, but it does not inactivate Rb by itself (32).
Our interest in UL97-mediated Rb protein inactivation stems from both the long-standing paradigm of using viral proteins to probe the mechanisms of cell cycle progression and oncogenesis, as well as our evolving understanding of an association between HCMV infection and human cancers. Although certainly not a classically defined tumor virus, HCMV is found in human tumors of the brain and breast (39 -42), and the virus encodes multiple proteins that induce molecular events that represent the hallmarks of cancer (43)(44)(45). However, HCMV does not overtly transform cells in vitro. Thus it appears that, compared with tumor viruses such as human papillomavirus, HCMV modulates the cell cycle in a more nuanced way that does not lead to transformation but still supports viral infection. We suggest that understanding how HCMV modulates the cell cycle will illuminate potential ways the virus might contribute to cancer. But perhaps more importantly, an understanding of how HCMV manipulates the cell cycle without oncogenically transforming cells can direct us toward growth control pathways specific to cancer cells. Chemotherapeutic inhibition of such pathways might prove effective and better tolerated than more broadly acting drugs.
Here we extend our previous examination of how UL97 inactivates Rb to the Rb-related p107 and p130 proteins. We show that UL97 phosphorylates p107 and p130 in the context of an HCMV infection, when ectopically expressed, and during in vitro kinase reactions. The phosphorylation events require the L1 motif of UL97 and the clefts of p107 and p130, similar to Rb. However, UL97 does not disrupt all transcriptionally repressive complexes formed by p107 or p130, as it does for Rb (32). Furthermore, UL97 forms a stable complex with p107 but not p130 or Rb. Our data indicate that UL97 inactivates the different Rb proteins in different ways that are unique not only when compared with each other but also when compared with the mechanisms used by other viral and cellular proteins to inactivate Rb, p107, and p130.

v-CDK UL97 phosphorylates Rb family members p107 and p130 during HCMV infection
A previous report detected p130 but not p107 phosphorylation during HCMV infection (46). We observed the phosphorylation of both p107 and p130 during HCMV productive infection of primary human fibroblasts by monitoring upward shifts in electrophoretic mobility as well as detection by an antibody specific for Ser-672-phosphorylated p130 (Fig. 1A). The phosphorylated forms of p107 (Fig. 1B) are more definitively separated using phosphate affinity (Phos-tag) gel electrophoresis (47) as confirmed by phosphatase treatment (Fig. 1C), perhaps explaining why they were not observed in the previous study. The percentage of p107 found in the phosphorylated state increased after HCMV infection (Fig. 1B). Consistent with our previous work with Rb (32,48), we show here that the HCMV UL97 protein is responsible for p107 and p130 phosphorylation during infection, as these Rb family members are not phosphorylated during infection with a UL97-null virus (Fig. 1, A and B). Although the viral IE1 protein was reported to possess kinase activity and to phosphorylate p107 and p130 in vitro (49), IE1 is present in UL97-null virus-infected cells, yet p107 and p130 remain hypophosphorylated (Fig. 1, A and B). We conclude that during HCMV infection, IE1 does not phosphorylate p107 or p130 but UL97 does.

Other v-CDKs also phosphorylate p107 and p130
All human herpesviruses encode a kinase orthologous to UL97. The orthologous kinases from the betaherpesviruses D  Whole-cell lysates were prepared at the indicated times postinfection, and equal amounts of protein from each sample were loaded onto a normal gel (A) or phosphate affinity (phos-tag) gel (B) and subjected to Western blotting analysis with the indicated antibodies. The percentage of phosphorylated p107 (compared with total) quantified by ImageJ is presented below the panel. M, mock infection. C, lysates from HCMV-infected HFFs were treated (ϩ) or not (Ϫ) with protein phosphatase (l-PPase) and analyzed as in B. D, Saos-2 cells were transfected with expression plasmids for HA-tagged p107 (HA-p107) together with either an empty vector (EV), one expressing V5-tagged UL97 (97), cyclin A2 (A2)/CDK2 (K2), cyclin E1 (E1)/CDK2, or cyclin D2 (D2)/CDK4 (K4). Lysates harvested 48 h after transfection were analyzed by Western blotting with the indicated antibodies. E, transfections were performed as in D with HA-tagged p130 (HA-p130), p130⌬CDK4, or p130-PM19A. F and G, lysates of Saos-2 cells transfected with plasmids expressing HA-tagged p107 (HA-p107) or p130 (HA-p130) were immunoprecipitated with anti-p107 or p130 antibodies. Either wild-type (WT) or kinase-dead (KD) UL97 was incubated with immuneenriched p107 or p130 substrates in the presence of [␥-32 P]ATP. Kinase reactions were resolved by SDS-PAGE and transferred to a nitrocellulose membrane. They were visualized by both autoradiography and Western blotting with an anti-HA antibody for detection of HA-UL97 and HA-p107 and an anti-p130 antibody for detection of HA-p130. M.W., molecular weight. H, kinase reaction was conducted as in F except normal ATP and bacterially purified GST-tagged p130 were utilized. Samples were analyzed by Western blotting with the indicated antibodies. Experiments were performed in biological triplicate except those in D and E that were performed in biological duplicates.
(HHV-6 and HHV-7) and gammaherpesviruses (EBV and KSHV) also display v-CDK activity and phosphorylate Rb, whereas the orthologs in the alphaherpesviruses (herpes simplex virus type 1 (HSV-1), HSV-2, and varicella zoster virus) do not (25). Similarly, we found that the alphaherpesvirus proteins were unable to phosphorylate p107 ( Fig. 2A) or p130 (Fig. 2B), but the beta-and gammaherpesvirus proteins phosphorylated p107 (Fig. 2C) and p130 ( Fig. 2D) with different efficiencies. Kinase-dead (KD) mutants of these proteins failed to phosphorylate p107 or p130, as expected. We conclude that the v-CDKs encoded by beta-and gammaherpesviruses phosphorylate p107 and p130. This ability highlights the functional relatedness of the v-CDKs as well as the importance of modifying Rb family protein function by phosphorylation for successful infection by these pathogenic human viruses. Understanding how UL97 (the representative v-CDK) targets and inactivates the Rb family proteins not only opens up opportunities for inhibiting this reaction as an antiviral intervention but also illustrates how the Rb family proteins might be inactivated during oncogenic transformation in virus-infected or -uninfected cancer cells.

L1 motif of UL97 and the clefts of p107 and p130 are required for p107 and p130 phosphorylation
UL97 could potentially interact with Rb family proteins (physically or functionally) through two distinct interfaces: one or more of three LXCXE motifs in UL97 (L1, L2, or L3) inter-acting with the binding cleft of the Rb family proteins, or the HP of UL97 interacting with RXL motifs in the Rb family proteins. We previously showed that of all these potential interfaces, only the first LXCXE motif (L1) of UL97 and the cleft of Rb were required for UL97-mediated Rb hyperphosphorylation (32). The rest of the motifs were dispensable. We achieved essentially identical results with p107 and p130. UL97 L2 (C428G) and L3 (C693G) mutants that carry the same central cysteine to a glycine substitution as the L1 mutant phosphorylated p107 (Fig. 3A) and p130 (Fig. 3B) as well as wild type in transfection assays, but the L1 mutant did not. Mutation of all three LXCXE motifs (Tri) also impaired p107 (Fig. 3A) and p130 (Fig. 3B) phosphorylation. Similar results were obtained upon infection of fibroblasts with recombinant HCMVs harboring the UL97 mutant alleles (Fig. 3C). Each LXCXE motif mutant virus and the triple LXCXE mutant virus grow like wild type (32,33). Both the L1 and Tri mutant viruses showed reduced p107 and p130 phosphorylation compared with wild type (Fig. 3C). The L1 mutation-specific decreases in p130 phosphorylation were less dramatic during infection than transfection but were reproducible and statistically significant (Fig. 3D). It is possible the limited p107 and p130 phosphorylation observed in the presence of UL97-L1 mutant results from some residual activity of the mutant kinase toward the substrates. For example, the L1 mutant still efficiently phosphorylates Rb on some cyclin/CDK Values are presented relative to the value in wild-type virus-infected cells (set at 1). Error bars denote the standard deviation. *, p Ͻ 0.05. E, Saos-2 cells were transfected with expression plasmids encoding V5-tagged wild-type UL97 together with either wild-type HA-tagged p107, a p107 cleft mutant (C846F), or a p107 RXL mutant. Forty eight hours later, lysates were harvested and subjected to Western blotting with the indicated antibodies. F, Saos-2 cells were transfected with expression plasmids encoding V5-tagged wild-type UL97 together with either wild-type HA-tagged p130 or a p130 cleft mutant (C894F). Forty eight hours later, lysates were harvested and subjected to Western blotting with the indicated antibodies. G, Saos-2 cells were transfected with expression plasmids encoding HA-tagged wild-type UL97 together with either wild-type HA-tagged p107, a p107 cleft mutant (N935F), or a p107 RXL mutant. Forty eight hours later, lysates were harvested and subjected to Western blotting with the indicated antibodies. H, Saos-2 cells were transfected with expression plasmids encoding HA-tagged wild-type UL97 together with either wild-type HA-tagged p130, a p130 cleft mutant (N1010F), or a p130 RXL mutant. Forty eight hours later, lysates were harvested and subjected to Western blotting with the indicated antibodies. I, Saos-2 cells were transfected with an expression plasmid for cyclin E1 together with HA-tagged wild-type or RXL mutant p107 or p130 expression plasmids. Lysates harvested 48 h after transfection were subjected to immunoprecipitation (IP) with the HA antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. J, Saos-2 cells were transfected with expression plasmids encoding cyclin E1 and CDK2 together with HA-tagged wild-type p107, a p107 RXL mutant, wild-type p130, or a p130 RXL mutant. Forty eight hours later, lysates were harvested and subjected to Western blotting with the indicated antibodies. The phosphorylation-dependent band shifts of both exogenously expressed and endogenous p107 were detected in phosphate affinity (phos-tag) gels. Experiments were performed in at least biological triplicates.

Inactivation of p107 and p130 by HCMV UL97
target sites (32). Alternatively, this restricted phosphorylation may be the result of a cellular kinase.
We tested the necessity of the L1-cleft interaction for UL97mediated p107 or p130 phosphorylation with reciprocal mutations of the cleft domains of p107 and p130 as we did previously for Rb (32). Two mutants were analyzed for each protein. One set of previously studied alleles (p107-C846F (52) and p130-C894F (53)) mimic the Rb-C706F allele found in the small cell lung cancer H209 cell line (54,55). Neither the p107-C846F ( Fig. 3E) nor the p130-C894F (Fig. 3F) cleft mutant protein was efficiently phosphorylated by wild-type UL97. We generated a second set of cleft mutants (p107-N935F and p130-N1010F) corresponding to another previously designed Rb-N757F cleft mutant (56). Neither the p107-N935F (Fig. 3G) nor the p130-N1010F (Fig. 3H) cleft mutant protein was efficiently phosphorylated by wild-type UL97. We conclude that the UL97 L1 motif and the p107 and p130 clefts are required for UL97-mediated p107 and p130 phosphorylation.
The HP mutant showed reduced phosphorylation of p107 ( Fig. 3A) but not p130 (Fig. 3B) in transfection assays. However, results with the HP mutant of UL97 (and with the quadruple mutant combination of L1, L2, L3, and HP motifs) must be interpreted with caution because this specific lesion (W368A) appears to attenuate overall kinase activity and viral replication (32). Therefore, to uncover any role for the HP-RXL interaction in UL97-mediated p107 or p130 phosphorylation, we determined the ability of wild-type UL97 to phosphorylate p107 or p130 alleles lacking RXL motifs, as we did previously for Rb (32). We generated epitope-tagged RXL deletion mutants of p107 (p107-⌬RXL) (57) and p130 (p130-⌬RXL) (58) and expressed them in Saos-2 cells by transient transfection. UL97 phosphorylated the RXL mutant p107 (Fig. 3, E and G) and p130 (Fig. 3H) proteins as well as their wild-type counterparts. Confirming previous results (57,58), cyclin E showed reduced interaction with the RXL mutants of p107 and p130 compared with the wild-type proteins (Fig. 3I). Furthermore, the RXL mutants of p107 and p130 were not phosphorylated as completely as the wild-type proteins by cyclin E/CDK2 (Fig. 3J). Because the p107-and p130-RXL mutants used here show a reduced ability to interact physically and functionally with a hydrophobic patch-containing protein, we conclude the HP-RXL interaction is dispensable for UL97-mediated p107 or p130 phosphorylation.

pE2F1-Luc
Relative luciferase activity (%) Figure 4. UL97 relieves p107-or p130-mediated repression of an E2F-responsive promoter. A, Saos-2 cells were transfected with a luciferase reporter driven by the E2F1 promoter together with either an empty vector (Ϫ) or an expression plasmid for HA-tagged p107, and either an empty vector or an expression plasmid for the indicated allele of UL97. Lysates harvested 48 h after transfection were analyzed for luciferase activity (top) and protein expression with the indicated antibodies (bottom). Luciferase activity was normalized to total protein concentration and is presented relative to the activity of the reporter without p107 or UL97 (set at 100%). B, luciferase and Western blotting analyses were performed as in A except with HA-tagged p130. Luciferase activity is presented relative to the activity of the reporter without p130 or UL97 (set at 100%). C, luciferase and Western blotting analyses were performed as in A except cyclin E1/CDK2 (E/2) was included. Luciferase activity is presented relative to the activity of the reporter with UL97 (97) (set at 100%). D, luciferase and Western blotting analyses were performed as in B except cyclin E1/CDK2 was included. Luciferase activity is presented relative to the activity of the reporter with UL97 (set at 100%). Error bars denote the standard deviation of more than three biological replicates. *, p Ͻ 0.05; **, p Ͻ 0.01; n.s., not significant.

Rb-related p130 protein is required for efficient HCMV productive replication
UL97 also phosphorylates and inactivates Rb, but paradoxically, the Rb protein is required for efficient HCMV infection (59). Conversely, p107, which is also phosphorylated (Fig. 1) and inactivated (Fig. 4) by UL97, is not required for HCMV infection (60). Because p130 had yet to be tested, we monitored HCMV infectious progeny formation in cells in which p130 expression was stably reduced by shRNA-mediated interference. Similar to Rb, stable p130 knockdown cells produced fewer infectious progeny after 4 days of HCMV infection than did control cells transduced with a scrambled shRNA (Fig. 5). We have not tested longer time points but predict the observed decrease in replication would persist because it does in Rbknockdown cells (60). Likewise, because a clinical strain virus also shows decreased replication in Rb-knockdown cells (59), we suspect one would also show defects in p130-knockdown cells, but we have not performed this experiment. Based on this result and previous data, we conclude that Rb and p130, but not p107, are required for efficient HCMV productive replication.

UL97 forms a stable complex with p107 but not with p130
Because Rb and p130 are required for efficient HCMV infection but p107 is not, we explored whether the mechanism of UL97-mediated inactivation of p130 in the context of E2F-dependent gene expression was similar to that of Rb inactivation, and conversely whether UL97-mediated inactivation of p107 was somehow different from Rb inactivation. Our previous work (32) demonstrated that UL97 did not form a stable complex with Rb but did disrupt Rb-E2F complexes. Therefore, we examined UL97 binding to p107 and p130 and the ability of UL97 to disrupt p107-E2F and p130-E2F complexes.
With immunoprecipitation-Western blotting experiments from transfected cells, we detected a strong interaction between UL97 and p107 that required both the L1 motif of UL97 (Fig. 6A) and an intact cleft domain in p107 (Fig. 6B). Unlike cyclin E (Fig. 3I), the RXL motif of p107 was not required for interaction with UL97 (Fig. 6B). However, kinase activity was required as the UL97 point mutant lacking kinase activity failed to interact with p107 (Fig. 6, A and C), even when expressed to higher levels than the wild-type protein (Fig. 6C). We detected the UL97-p107 interaction by Western blotting after immunoprecipitation of p107 (Fig. 6, A and C) or UL97 (Fig. 6D) as well as after covalent capture of UL97 (Fig. 6B). p130 interacted weakly with UL97 in similar experiments (Fig.  6E). Mutating the UL97 L1 motif reduced this interaction (Fig.  6E). The UL97 point mutant lacking kinase activity failed to interact with p130 (Fig. 6E). By using the same antibody (to the HA epitope tag) to precipitate UL97 and the same antibody (FLAG) to detect either p107 or p130, we were able to compare the relative efficiency with which these two pocket proteins interacted with UL97, and we found that UL97 bound 3-fold more p107 than p130 (Fig. 6D). We conclude that UL97 associates strongly with p107 and weakly with p130 through L1cleft interactions in a kinase activity-dependent manner.

Neither UL97 nor IE1 disrupts E2F interactions with p107 or p130
The textbook model for the inactivation of Rb family protein function (61) is that phosphorylation mediated by cellular CDKs or binding by viral oncoproteins disrupts their complexes with E2F transcription factors. Indeed, we previously found that UL97-mediated Rb phosphorylation disrupted its complexes with E2F1, E2F2, E2F3a, and E2F3b (32). However, UL97 was unable to disrupt p107 complexes with E2F1 or E2F4 (Fig. 6, A and F) whereas adenovirus E1a effectively disrupted them (Fig. 6, A and G). UL97 failed to disrupt p107-E2F1 complexes in the same experiment where lower levels of UL97 effectively disrupted Rb-E2F1 complexes (Fig. 6H). Likewise, UL97 was unable to disrupt p130 complexes with E2F4 (Fig. 6, E and F) but adenovirus E1a effectively disrupted them (Fig. 6, E and G). p130 complexes with E2F5 were not detected in our experiments in Saos-2 cells (Fig. 6E). In similar experiments, cyclin E/CDK2 showed a statistically significant ability to disrupt p107-E2F1, p107-E2F4, and p130-E2F4 complexes (Fig. 6, I-K). Although the magnitude of disruption by UL97 was in some cases comparable with cyclin E/CDK2, our results with UL97 never reached statistical significance (Fig. 6F). Therefore, we conclude that UL97-mediated inactivation of p107 and p130 does not occur through disruption of their complexes with the E2F proteins.
The HCMV IE1 protein reversed p107-mediated repression of the E2F-responsive adenovirus E2 promoter during cotransfection/reporter assays in C33A cells (34) indicating that, like UL97, IE1 can inactivate p107. We confirmed the ability of IE1 to inactivate p107 by demonstrating IE1-mediated reversal of p107-instituted repression of a reporter driven by the E2F1 promoter in Saos-2 cells (Fig. 7A). We also demonstrated IE1-mediated inactivation of p130 with a similar assay (Fig. 7B). Despite reports of in vitro kinase activity for the IE1 protein (49), we found no evidence that IE1 induced the phosphorylation of p107 or p130 in vivo (Fig. 7C). However, we did detect a stable interaction between IE1 and both p107 (Fig. 7D) and p130 (Fig. 7E). Like UL97-mediated inactivation of p107 and p130, IE1-mediated inactivation of p107 or p130 did not occur through the disruption of p107-E2F1 (Fig. 7, D and F), p107-E2F4 (Fig. 7, D and F), or p130-E2F4 (Fig. 7, E and F) complexes. Our work mirrors a previous study (62)

Inactivation of p107 and p130 by HCMV UL97
action between IE1 and p107 that did not dissociate E2F4. We conclude that IE1 binds and inactivates p107 and p130 without disrupting their complexes with E2F proteins.

HCMV infection fails to disrupt the transcriptional repressive complexes assembled by p107 or p130
Our inability to detect UL97-or IE1-mediated disruption of p130-E2F complexes during transfection assays (Figs. 6 and 7) is consistent with the previous finding that the p130-E2F4 complex can be captured from HCMV-infected cells (46). We confirmed that p130-E2F4 complexes are stable in serum-starved cells infected with HCMV (Fig. 8A, lane 3) at a multiplicity of infection (m.o.i. ϭ 3) capable of expressing UL97 in most of the cells (Fig. 8B) and that p130-E2F4 complexes are disrupted by serum stimulation of uninfected cells (Fig. 8A, lane 6). Thus in a setting more relevant than transfection assays, UL97 still fails to disrupt p130-E2F4 complexes.
We reasoned that disruption of p130-E2F4-MuvB complexes by UL97-mediated phosphorylation of Ser-672 (with probable simultaneous phosphorylation of other residues) was likely to inactivate the transcriptional repressive function of p130. Therefore, we asked whether the other complex that contains E2F4, MuvB, and p130 phosphorylated at residues other than Ser-672, and that is co-present with UL97, is capable of repressing an E2F-driven reporter. The p130⌬CDK4 protein (T401A/ S627A/S1035A) repressed the E2F1 promoter in Saos-2 cells (Fig. 8F) but was inactivated by UL97 (Fig. 8F). We conclude that, despite the inability of UL97 to disrupt p130-E2F4-MuvB complexes in which Ser-672 is not phosphorylated, UL97 can still functionally inactivate this stable complex. Furthermore, weconcludethatdisruptionofp130complexesviaSer-672phosphorylation is mediated by UL97 but is not necessary for UL97mediated p130 inactivation. Although we cannot test p107 because an antibody specific for the phosphorylated form of the homologous residue (Ser-650) is unavailable, we hypothesize that a similar mechanism would explain the ability of UL97 to inactivate p107 (see Fig. 9). Interestingly, cyclin E/CDK2 was able to both functionally inactivate p130⌬CDK4 (Fig. 8G) and Figure 6. UL97 does not disrupt all p107-E2F or p130-E2F complexes. A, Saos-2 cells were transfected with an expression plasmid for wild-type FLAGtagged p107 together with an empty vector (Ϫ) or an expression plasmid for the indicated allele of UL97 or adenovirus E1a. 97, wild-type UL97; L1, L1 mutant; KD, kinase-dead mutant. Lysates harvested 48 h after transfection were subjected to immunoprecipitation (IP) with a p107 antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. h.c., heavy chain; C, rabbit IgG control. B, Saos-2 cells were transfected with an expression plasmid for the indicated p107 protein together with expression plasmids for HALO-and HA-tandem-tagged UL97 (HALO/HA-97). Lysates harvested 48 h after transfection were subjected to HALO tag capture and cleaved with TEV protease between the HALO and HA sequences to generate HA-tagged UL97 (HA-97). Input lysates and captured proteins were analyzed by Western blotting with the HA antibody. CM1, p107-C846F; CM2, p107-N935F; RXL, p107-⌬RXL. C, Saos-2 cells were transfected with a constant amount (1.6 g) of an expression plasmid for wild-type FLAG-tagged p107 together with differing amounts of expression plasmids for either wild-type or kinase-dead UL97. Lysates were subjected to immunoprecipitation with a p107 antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. D, Saos-2 cells were transfected with an expression plasmid for FLAG-tagged p107 or p130 together with an empty vector (Ϫ) or an expression plasmid for HA-tagged UL97. Lysates harvested at 48 h after transfection were subjected to immunoprecipitation with the HA antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. The levels of FLAG-tagged p107 or p130 bound to HA-tagged UL97 were quantitated using ImageJ, and the values are presented below the panel. E, Saos-2 cells were transfected with an expression plasmid for wild-type FLAG-tagged p130 together with an empty vector (Ϫ) or an expression plasmid for the indicated allele of UL97 or adenovirus E1a. Lysates harvested 48 h after transfection were subjected to immunoprecipitation with a p130 antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. h.c., heavy chain; l.c., light chain; C, rabbit IgG control. F and G, quantification of the levels of bound E2F1 or E2F4 normalized to immunoprecipitated p107 or p130 in the absence (Ϫ) or presence of wild-type UL97 (F) or E1a (G) from A and E performed using ImageJ. Values are presented relative to the value in the absence of UL97 or E1a (set at 1). Error bars denote the standard deviation of more than three biological replicates. *, p Ͻ 0.05; **, p Ͻ 0.01; n.s., not significant. H, Saos-2 cells were transfected with a constant amount (1.8 g) of an expression plasmid for FLAG-tagged p107 or Rb together with differing amounts of expression plasmids for UL97. Lysates harvested 48 h after transfection were subjected to immunoprecipitation with a p107 antibody or an Rb antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. C, rabbit IgG control. Arrow, E2F1-specific band; asterisk, nonspecific band. I and J, Saos-2 cells were transfected with an expression plasmid for FLAG-tagged p107 (I) or p130 (J) together with expression plasmids for cyclin E1/CDK2 (E/2) or HA-tagged UL97. Lysates harvested at 48 h after transfection were subjected to immunoprecipitation with a p107 (I) or p130 antibody (J). Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. K, quantification of the levels of bound E2F normalized to immunoprecipitated p107 or p130 from I and J performed using ImageJ. Values are presented relative to the value in the absence of cyclin E1/CDK2 (set at 1). Error bars denote the standard deviation of more than three biological replicates. *, p Ͻ 0.05. Experiments were performed in at least biological triplicates except those in C and D that were performed in biological duplicates. disrupt its association with E2F4 (Fig. 8, H and I). This suggests that, like with UL97, p130 Ser-672 phosphorylation is not required for inactivation, but unlike UL97, complex disruption may be required for cyclin E/CDK2 to inactivate p130.
We further investigated how UL97-mediated phosphorylation might inactivate p107 and p130 without disrupting their complexes with E2Fs or MuvB. In certain cell types, p107-and p130-assembled complexes contain the chromatin-modifying HDAC1 protein or the chromatin-remodeling BRG1 protein (11). However, we failed to detect p107 or p130 association with either HDAC1 or BRG1 in human fibroblasts (Fig. 8A) or Saos-2 cells (Fig. 6, A and E). Therefore, UL97-mediated inactivation of p107 and p130 is unlikely to occur through or to require disruption of their binding to HDAC1 or BRG1. We conclude that UL97 inactivates p107 and p130 through a unique mechanism that does not require the complete disruption of the transcriptionally repressive complexes assembled by either p107 or p130.
Despite the inability to disrupt the transcriptionally repressive complexes assembled by p107 and p130, HCMV infection induces the accumulation of proteins encoded by E2F-regulated genes such as p107, E2F1, Rb, MCM7, cyclin E1, and cyclin B1 (Fig. 8J). Serum stimulation (Fig. 8J) of mock-infected serum-starved cells also induces the accumulation of these proteins, as well as that of cyclin A2, cyclin D1, and cyclin D2, which are not induced by HCMV infection (63)(64)(65). Maribavir, a small molecule inhibitor of UL97, attenuated the accumulation of the protein products of the tested E2F-regulated genes, except for cyclin E1 (Fig. 8J), indicating UL97 kinase activity is required for most E2F-responsive gene expression in HCMVinfected cells. The L1 LXCXE motif required for UL97-mediated p107 and p130 phosphorylation (Fig. 3) and important for UL97-mediated Rb phosphorylation (32) was not required for the accumulation of the protein products of the tested E2Fregulated genes (Fig. 8J) (32), likely because of complementation by IE1, which independently inactivates p107 (Fig. 7A) (34) and p130 (Fig. 7B), and it cooperates with the L1 mutant UL97 (L1m) to inactivate Rb (32).
In summary, we describe novel ways in which the HCMV v-CDK UL97 inactivates Rb (Fig. 9A), p107 (Fig. 9B), and p130 . Luciferase activity was normalized to total protein concentration and is presented relative to the activity of the reporter without p107 or IE1 (set at 100%). B, luciferase assay and Western blotting analyses were performed as in A except with HA-tagged p130. Luciferase activity is presented relative to the activity of the reporter without p130 or IE1 (set at 100%). Error bars denote the standard deviation of more than three biological replicates. *, p Յ0.05; n.s., not significant. C, Saos-2 cells were transfected with an expression plasmid for HA-p107 or HA-p130 together with either an empty vector (EV) or an expression plasmid for HA-tagged IE1 or V5-tagged UL97. Lysates harvested 48 h after transfection were analyzed by Western blotting with the indicated antibodies. D, Saos-2 cells were transfected with an expression plasmid for HA-tagged p107 and either an empty vector (Ϫ) or an expression plasmid for IE1. Lysates harvested 48 h after transfection were subjected to immunoprecipitation with a p107 antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. E, Saos-2 cells were transfected with an expression plasmid for HA-tagged p130 and either an empty vector (Ϫ) or an expression plasmid for IE1. Lysates harvested 48 h after transfection were subjected to immunoprecipitation with a p130 antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. F, quantification of the levels of bound E2F normalized to immunoprecipitated p107 or p130 from D and E performed using ImageJ. Values are presented relative to the value in the absence of IE1 (set at 1). Error bars denote the standard deviation of biological triplicates. n.s., not significant. (Fig. 9C). Similarities include the requirement for the L1 motif and catalytic activity for UL97 and the cleft domain of the Rb family members. Differences include whether or not associations with E2F proteins are disrupted and whether or not a stable complex between UL97 and the Rb family member is detected.

Discussion
To understand the physiological function of a kinase, it is imperative to know its substrates. Global methods have been utilized to identify potential substrates of the HCMV-encoded v-CDK UL97, including the use of protein arrays (66) and phosphoproteomics (67). Still, novel, important, and confirmed substrates emerge through the classic approach of testing likely candidates such as Rb (48,68), lamin A/C (69), and HDAC1 (70). Through this method, we have identified the Rb-related p107 and p130 proteins as UL97 substrates. Neither p107 nor p130 were detected by more advanced global methods (66,67), reaffirming that such approaches are not comprehensive and that individual candidate interrogation is still viable and valuable.
Our previous work (32) demonstrated that UL97 uses its first LXCXE motif and the LXCXE-binding cleft in Rb to mediate Rb phosphorylation, the disruption of Rb complexes with E2F1, E2F2, E2F3a, and E2F3b, and the reversal of Rb-mediated repression of E2F-responsive promoter reporters (Fig. 9A). Interestingly, we could not correlate UL97-mediated Rb inactivation with E2F dissociation as the UL97-L1 mutant was competent at complex disruption but failed to reverse Rb-mediated reporter repression (32). This represents a mechanism for Rb inactivation and for the stimulation of E2F-dependent transcription that is distinct from those mediated by cellular cyclin-CDK complexes (17) or other viral proteins (26,29,35,36). Here we show that UL97 also inactivates p107 and p130 by seemingly novel mechanisms.
Our immunoprecipitation assays utilize a ratio of UL97 to p130 that is 2-fold higher than that used in our luciferase assays. Despite increasing the UL97/p130 ratio over that which is required to inactivate p130 (Fig. 4B), UL97 still fails to disrupt p130-E2F4 complexes (Fig. 6E). Furthermore, the levels of UL97 achieved by high multiplicity infection of most of the cells in a culture (Fig. 8, A and B) also fail to disrupt p130-E2F complexes. Therefore, we consider it unlikely that a higher level expression of UL97 would lead to complex disruption, and we conclude that UL97 inactivates p130 (and by extension p107) without disrupting their complexes with E2F proteins.
Ser-672 on p130 is analogous to Ser-608 on Rb and Ser-650 on p107. Phosphorylation of those sites weakens the interactions between Rb or p107 and E2F proteins (21,24). We found that when UL97 phosphorylated p130 on Ser-672, p130 failed to interact with E2F4 (Fig. 8). We have not tested the ability of UL97 to phosphorylate p107 on Ser-650, but we hypothesize that such a phosphorylation event occurs and disrupts p107-E2F complexes (Fig. 9B). Interestingly, UL97 phosphorylated a mutant p130 protein with an unphosphorylatable residue at position 672 (Fig. 1) and reversed the ability of this mutant protein to suppress an E2F reporter (Fig. 8). Thus, UL97 must generate at least two differentially phosphorylated forms of p130 that are inactivated in different ways (one by phosphorylation at Ser-672 and complex disruption and one not). This appears to agree with the recent revelation that Rb exists as multiple different mono-phosphorylated forms during the G 1 phase (71). More work is needed to determine whether UL97 mono-phosphorylates the Rb family proteins and to define the physiological roles of mono-phosphorylated Rb family proteins in uninfected cells and perhaps during HCMV infection.
The reasons for the diversity of mechanisms used by UL97 to inactivate Rb, p107, and p130 ( Fig. 9) and for the mechanistic divergence from those used by other cellular and viral proteins are unknown, but they may be related to either the different were incubated with 15% FBS for 18 h before harvesting. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. pS672, p130 phospho-Ser-672 specific antibody. Lanes with immunoprecipitates are numbered below. B, serum-starved HFFs were infected with HCMV as in A. At 45 h post-infection cells were subjected to immunostaining with a UL97 antibody and Hoechst treatment to counterstain DNA. C, U-2 OS cells were transfected with an expression plasmid for FLAG-tagged LIN52 together with an expression plasmid for HA-tagged UL97. Lysates harvested 48 h after transfection were subjected to immunoprecipitation with the FLAG antibody (F) or normal mouse IgG (C) as a control. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. D, Saos-2 cells were transfected with expression plasmids for wild-type FLAG-tagged p107 and wild-type E2F4 together with an empty vector (Ϫ) or an expression plasmid for UL97. Lysates harvested 48 h after transfection were subjected to immunoprecipitation with an E2F4 antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. C, rabbit IgG control. E, Saos-2 cells were transfected as in D except wild-type FLAG-tagged p130 was included. Lysates harvested 48 h after transfection were subjected to immunoprecipitation with an E2F4 or p130 phospho-Ser-672-specific antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. C, rabbit IgG control. Lanes with immunoprecipitates are numbered below. F, Saos-2 cells were transfected with a luciferase reporter driven by the E2F1 promoter together with either an empty vector (Ϫ) or an expression plasmid for p130⌬CDK4 (⌬) and either an empty vector or a wild-type UL97 expression plasmid. Lysates harvested 48 h after transfection were analyzed for luciferase activity (top) and protein expression with the indicated antibodies (bottom). Luciferase activity was normalized to total protein concentration and is presented relative to the activity of the reporter without p130⌬CDK4 or UL97 (set at 100%). Error bars denote the standard deviation of more than three biological replicates. **, p Ͻ 0.01. G, luciferase and Western blotting analyses were performed as in F except expression plasmids for cyclin E1/CDK2 (E/2) were included. H, Saos-2 cells were transfected with an expression plasmid for FLAG-tagged wild-type p130 (WT) or p130⌬CDK4 (⌬) together with expression plasmids for cyclin E1/CDK2 (E/2). Lysates harvested 48 h after transfection were subjected to immunoprecipitation with a FLAG antibody. Input lysates and immunoprecipitates were analyzed by Western blotting with the indicated antibodies. C, mouse IgG control. I, quantification of the levels of bound E2F4 normalized to immunoprecipitated FLAG-tagged p130⌬CDK4 (⌬) from H performed using ImageJ. Values are presented relative to the value in the absence of cyclin E1/CDK2 (set at 1). The error bar denotes the standard deviation of biological triplicate samples. *, p Ͻ 0.05. J, input lysates from A were analyzed by Western blotting with the indicated antibodies. Experiments were performed in at least biological triplicates except those in D and E that were performed in biological duplicates.  Figure 9. Models for Rb, p107, or p130 inactivation by UL97. A, UL97 res-cuesRb-mediatedrepressionofE2F-responsivegeneexpressionbybothphosphorylation-mediated (P) disruption of Rb-E2F complexes as well as through an undefined L1 LXCXE-dependent mechanism. B, UL97 associates with and phosphorylates p107 to rescue p107-mediated repression of E2F-responsive gene expression in a UL97-L1-and p107-cleft-dependent manner but does not disrupt p107-E2F or p107-MuvB complexes. Whether UL97 disrupts some p107 complexes by phosphorylating p107 at Ser-650 (homologous to Rb Ser-608 and p130 Ser-672) is not known (?). C, UL97 phosphorylates p130 to rescue p130-mediated repression of E2F-responsive gene expression in a UL97-L1-and p130-cleft-dependent manner but does not disrupt p130-E2F or p130-MuvB complexes. UL97 phosphorylates p130 Ser-672 disrupting p130-E2F4 complexes, but Ser-672 phosphorylation is not necessary for p130 inactivation by UL97.
roles that the Rb family proteins play during HCMV infection or the different cellular (tumor suppressor) functions of the Rb family proteins. The hypophosphorylated forms of Rb and p130 are degraded by pp71 (36), an HCMV protein delivered directly to cells upon infection by virions (72). The subsequent phosphorylation of Rb and p130 by UL97 appears to protect them from pp71-mediated degradation and helps to maintain their presence in HCMV-infected cells while at the same time inactivating their ability to repress E2F-responsive transcription. The artificial depletion of Rb (59) or p130 (Fig. 5) impairs productive HCMV infection. Therefore, perhaps the Rb and p130 residues phosphorylated by UL97 are selected to simultaneously neutralize pp71-mediated degradation and release E2F repression, yet maintain the functions of Rb and p130 that HCMV requires.
The artificial depletion of p107 did not inhibit HCMV productive replication (60). Unlike Rb and p130 (6,8), p107 is not a strong tumor suppressor (73), and therefore it may affect processes less important for controlling both oncogenesis and viral infection. However, the experiments demonstrating the nonessential nature of p107 for HCMV productive replication were performed in G 0 cells arrested by serum withdrawal where p107 is not initially expressed but is induced by HCMV infection. It remains to be determined whether p107 is required for HCMV infections of S or G 2 phase cells, as well as how the Rb family proteins, presumably in their phosphorylated forms, contribute positively to HCMV infection. This positive contribution seemingly extends beyond the classic paradigm of driving infected cells into the S phase to support the replication of viral DNA genomes by promoting the expression of the genes encoding DNA replication proteins and nucleotide biosynthetic enzymes. This is because HCMV encodes most of its own DNA replication proteins (26), and because deoxyribonucleotides are limiting in HCMV-infected cells even when the Rb family proteins are phosphorylated (74).
A unique feature of all the HCMV proteins that interact with the Rb family members is that they do not lead to overt cellular transformation, whether expressed alone or together during viral infection. Rb-inactivating proteins from other viruses such as adenovirus E1a, SV40 T antigen, and papillomavirus E7 are all highly oncogenic. Thus the different mechanisms that HCMV UL97 uses to inactivate Rb in comparison with viral oncoproteins may be one of the reasons why HCMV is not a classic tumor virus, although it does appear to have a complex relationship with human cancers that is continuing to emerge (39,(43)(44)(45). Understanding the differences between Rb protein inactivation by viral oncoproteins and UL97 may reveal the key functions of Rb, p107, and p130 that protect cells from transformation, perhaps leading to more targeted and better tolerated cancer chemotherapies.

In vivo phosphorylation assays
Saos-2 cells were transfected with expression plasmids for wild-type or mutated p107 or p130 together with wild-type or mutated HCMV UL97 or various cyclin/CDK pairs using Tran-sIT-2020 (Mirus) as described previously (25). At 48 h posttransfection, the transfected cells were harvested and subjected to Western blotting analysis.

In vitro kinase assays
Wild-type and kinase-dead UL97 proteins were purified by transfecting 15-cm plates of HEK293T cells that were at 60% confluence with 32 g of pFC14a-UL97 WT or pFC14a-UL97 KD using polyethyleneimine (PEI) at 1 mg/ml. Cells were harvested at 48 h in purification buffer (1 mM DTT, 0.005% IGE-PAL CA-630, and PBS) containing a protease inhibitor mixture (Promega) and lysed by sonication. Lysates were cleared by centrifugation and then incubated overnight with HaloLink resin (Promega) to bind the HaloTag on the kinase. After three washes, TEV protease was used to release HA-UL97 WT or HA-UL97 KD from the resin. The eluted kinases were stored in purification buffer with 15% glycerol at Ϫ80°C. To generate the substrates for the autoradiography-based in vitro kinase assays, Saos-2 cells (1.6 ϫ 10 6 /100-mm plate) were transfected with 1.8 g of pHAp107 or pHAp130 using TransIT-2020. After 48 h, cells were suspended with modified CSK buffer (100 mM Pipes (pH 6.8); 500 mM NaCl; 300 mM sucrose; 1 mM EGTA; 1 mM MgCl 2 ; 0.1% Triton X-100; 10 g/ml pepstatin A; 25 g/ml leupeptin; 1 mM PMSF; 25 mM NaF; 10 mM ␤-glycerophosphate), collected by centrifugation, and then diluted with an equal volume of ET gel buffer (50 mM Tris (pH 7.4), 0.1% Triton X-100; 1 mM EDTA). The lysates were subjected to immunoprecipitation with anti-p107 or p130 antibodies (2 g) and protein A/G-magnetic beads (Thermo Fisher Scientific). The beads were washed four times with ET gel buffer with 250 mM NaCl and once with ET gel buffer with 150 mM NaCl. To generate the substrate for the Western blotting-based in vitro kinase assays, GST-p130 was purified from Escherichia coli strain Rosetta2(DE3) (Novagen) transformed with pGEX3X-p130 (322-1139 amino acids) as described previously (90). For the kinase reactions, all substrates were washed one final time with kinase buffer (50 mM Tris (pH 8); 5 mM ␤-glycerophosphate; 10 mM MgCl 2 ; 2 mM DTT) and resuspended in the kinase reaction containing ϳ250 ng of kinase; 75 M ATP; 1 Ci of [␥-32 P]ATP; 50 mM Tris (pH 8); 5 mM ␤-glycerophosphate; 10 mM MgCl 2 ; and 2 mM DTT. The reactions were incubated at 37°C for 30 min, stopped with SDS-PAGE loading buffer, and boiled for 10 min. Samples were run on a 6% SDS-PAGE, transferred to a nitrocellulose membrane, and exposed to film for autoradiography. Western blotting was then performed on the membrane to verify the presence of the kinases and p107 using an HA antibody and the presence of p130 using a p130-specific antibody. For detection of phosphorylated Ser-672 on p130, a kinase reaction was performed as described above except in the presence of unlabeled ATP (1 mM) and bacterially purified GST-p130 as a substrate. Western blotting was then performed using a p130 antibody and a phospho-Ser-672specific antibody.

Immunofluorescence analysis
HCMV-infected HFF cells were fixed with 2% paraformaldehyde and stained as described previously (25,92) with a UL97 primary antibody and a secondary antibody conjugated to Alexa Fluor 594 (Molecular Probes). The images were captured using an Olympus Fluoview FV 1000 confocal microscope and software. The specificity of anti-UL97 antibody was confirmed using mock-infected HFF cells.
Author contributions-S. I. and R. F. K. designed the experiments. S. I. performed the experiments. A. C. U. produced the purified UL97 kinases and performed in vitro kinase assays in Fig. 1, F and G. H. R. V. performed the p130 knockdown experiments in Fig. 5. S. I. and R. F. K. wrote the paper with comments from all authors.