The N-terminal peptide of the Kaposi's sarcoma-associated herpesvirus (KSHV)-cyclin determines substrate specificity.

Cyclin-dependent kinases (Cdks) are activated by cyclin binding and phosphorylation by the Cdk-activating kinase (CAK). Activation of Cdk6 by the D-type cyclins requires phosphorylation of Cdk6 by CAK on threonine 177. In contrast, Cdk6 is activated by the Kaposi's sarcoma-associated herpesvirus (KSHV)-cyclin in the absence and presence of CAK phosphorylation. The activity of Cdk6.KSHV-cyclin complexes was investigated here by analyzing mutants of the KSHV-cyclin and Cdk6 in vitro as well as in U2OS cells. Deletion of the N terminus of the KSHV-cyclin affects the substrate specificity indicating that the N terminus is required for phosphorylation of histone H1 but not for other substrates. Mutation of residues in the region 180-200 of the KSHV-cyclin decreases the binding affinity to Cdk6 in U2OS cells but increases the activity of Cdk6.KSHV-cyclin complexes in vitro indicating that low affinity binding of cyclins to the Cdk subunit might favor increased on- or off-rates of Cdk substrates. Expression of high levels of p16(INK4a) in cells leads to the formation of a heterotrimeric complex composed of Cdk6, KSHV-cyclin, and p16(INK4a). Some of the Cdk6 .KSHV-cyclin.p16 complexes were found to be active indicating that there might be different modes of p16 binding to Cdk6.cyclin complexes.

The activities of cyclin-dependent kinases (Cdks) 1 are thought to promote the transitions between the different phases of the cell cycle. For this reason, Cdk activities are tightly controlled by several mechanisms: cyclin binding, activating phosphorylation by the Cdk-activating kinase (CAK), inhibitory phosphorylation by Wee1/Myt1 kinases, dephosphorylation by the dual specificity phosphatase Cdc25, transcriptional regulation, ubiquitin-mediated degradation, subcellular localization, binding of inhibitors.
Eleven different Cdks have been identified in mammals. Of these, Cdk4 and Cdk6 bind to the three D-type cyclins and are active in the G 1 phase. Cdk2 binds to E-and A-type cyclins and has been thought to promote entry into S-phase and DNA replication. Cdc2 binds to the A-and B-type cyclins and is responsible for entry into mitosis. In general, the activities of Cdk4 and Cdk6 are hard to differentiate although there are indications that they do not have identical functions. For example, Cdk4 knock-out mice display specific defects (1, 2) that are not compensated by the presence of Cdk6.
Cyclin binding to the Cdk catalytic subunit is an important step in the activation of Cdks since it leads to a realignment of the activation segment/T-loop (substrate binding region) and the P-loop (ATP binding site) (3). For example, cyclin A binds to the PSTAIR helix, the T-loop, and the C-terminal lobe of Cdk2. For full activation of Cdks, threonine 177 in Cdk6 (Thr-160 in Cdk2) needs to be phosphorylated by CAK (for review see Ref. 4). Phosphorylation of the activating threonine causes further movement of the T-loop (5). Mutation of the activating threonine in Cdks leads to inactivation in vitro (6 -10) and in vivo (11,12).
The substrate specificity of Cdk⅐cyclin complexes is an important determinant of their function and is an area we know very little about. It has been shown that the cyclin subunits can influence substrate specificity since binding of different cyclins to the same Cdk can change its substrate specificity (32)(33)(34). Furthermore, there is a substrate binding site in the cyclin subunits, termed Cy or RXL motif (35,36), where p27 binds. Mutation of the Cy/RXL motif leads to decreased phosphorylation of Rb without affecting other substrates (e.g. histone H1) (37).
Here I investigate how the KSHV-cyclin activates Cdk6 using a series of KSHV-cyclin mutants. Several point mutations in the KSHV-cyclin lead to increased kinase activity in vitro and in cells and this is associated with decreased binding capacity between Cdk6 and the KSHV-cyclin. Interestingly, deletion of the N terminus of KSHV-cyclin leads to decreased phosphorylation of histone H1 whereas other substrates are not affected. To study the complex formation of p16, Cdk6, and KSHV-cyclin, p16 was expressed at high levels in U2OS cells, and a heterotrimeric complex of p16⅐Cdk6⅐KSHV-cyclin was detected. Nevertheless, p16 was not bound to KSHV-cyclin mutants because of the weak interaction with Cdk6 in cells.
Treatment of Cdk⅐Cyclin Complexes-Buffer treatment: 0.1 g insect cell produced GST-Cdk6 was incubated with 0.5 g of wild-type or mutant GST-KSHV-cyclin for 4.5 h at room temperature in a total volume of 10 l of buffer A (80 mM ␤-glycerophosphate, pH 7.3, 20 mM EGTA, 15 mM MgCl 2 , 150 mM NaCl, 1 mg/ml ovalbumin, and 1ϫ protease inhibitors (10 g/ml each of leupeptin, chymostatin, and pepstatin (Chemicon, Temecula, CA)). 1.5 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences, BLU502A), 0.375 mM ATP, and 1.5 g of histone H1 (Roche Applied Science) in a total volume of 6 l was added, and the reaction was continued for 15 min. ATP treatment: identical to the buffer treatment with the exception that 0.375 mM ATP was present with GST-Cdk6 and GST-KSHV-cyclin for 4.5 h instead of only 15 min at the end of the reaction. CAKϩATP treatment: 0.1 g of GST-Cdk6 was incubated with 0.375 mM ATP and 16 ng of GST-Cak1p for 4.5 h at room temperature in a total volume of 10 l of buffer A. 0.5 g of wild type or mutant GST-KSHV-cyclin, 1.5 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences, BLU502A), 0.375 mM ATP, and 1.5 g histone H1 (Roche Applied Science) was added, and the reaction was continued for 15 min. Reactions were terminated by adding 8 l of 5ϫ SDS sample buffer and were run on 12.5% SDS-polyacrylamide gels (Bio-Rad). Gels were dried and analyzed by phosphorimaging (Molecular Probes, Storm820).
Immunoprecipitation-48 h after transfection, each 10-cm dish of U2OS cells was scraped with a razor blade, washed with phosphate-buffered saline, and cells were lysed in 100 l of buffer B (phosphatebuffered saline, pH 7.4, 0.2% Triton X-100, 10% glycerol, 2 mM NaF, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 1ϫ protease inhibitors (20 g/ml each of leupeptin, chymostatin, and pepstatin (Chemicon)). For measurement of Myc-KSHV-cyclin or Cdk6-associated activity in vitro, the complexes were immunoprecipitated for 4 h at 4°C using 10 l of anti-Myc agarose (Clontech/BD, 3843-1) or anti-HA agarose (Roche Applied Science, 1815016). Immunocomplexes were washed four times with the buffer A containing 0.5% Nonidet P-40 followed by two washes with buffer A. For stringent conditions, immunoprecipitates were washed four times with RIPA buffer (50 mM Tris, pH 8.0, 1% Nonidet P-40, 0.5% sodium desoxycholate, 0.1% SDS, 150 mM NaCl, 1ϫ protease inhibitors (20 g/ml each of leupeptin, chymostatin, and pepstatin (Chemicon)), and twice with buffer A. Kinase reactions were performed in the presence of 1.5 Ci of [␥-32 P]ATP and 375 M ATP for 15 min at room temperature using 5 g of GST-Rb-(605-928) (prepared according to Ref. 65) or 1.5 g of histone H1 (Roche Applied Science) as substrates in a total volume of 16 l of buffer A. Phosphorylated proteins were analyzed on 12.5% SDS-polyacrylamide gels followed by autoradiography.

RESULTS
Activation of Cdk6 by KSHV-cyclin-Cyclin-dependent kinases are activated both by cyclin binding and by phosphorylation of an activating threonine residue (Thr-177 in Cdk6 or Thr-160 in Cdk2) by CAK (for review see Ref. 4). The KSHVcyclin has been shown previously to activate Cdk6 in the absence of CAK (49). I aimed to investigate the mechanism of Cdk6 activation by the KSHV-cyclin under identical conditions in the presence or absence of CAK (Fig. 1A). The assay consists of a 4.5-h preincubation period followed by 15 min labeling of substrate in the presence of radiolabeled ATP (for details see "Experimental Procedures"). CAK phosphorylated baculovirus produced GST-Cdk6 was readily activated by bacteria produced GST-KSHV-cyclin (Fig. 1B, lane 3). Interestingly, incubation of Cdk6 with the KSHV-cyclin and ATP resulted almost in the same degree of activation as CAK phosphorylation ( Cdk6⅐KSHV-cyclin Activity (Fig. 1B, lane 5). Next, I investigated if these results were related to the intrinsic properties of the purified Cdk6 or KSHV-cyclin preparations. The KSHV-cyclin could only activate CAK-phosphorylated Cdk2 (Fig. 1B, lane 9), as has been reported before (49). Furthermore, Cdk6 was not activated by cyclin A-(174 -432) (Fig. 1B, lanes 10 -12). Thus, activation of Cdks in the absence of CAK is restricted to the combination of Cdk6 and the KSHV-cyclin. The fact that ATP-activated Cdk6⅐KSHV-cyclin complexes indicated at least two possibilities: (1) Cdk6⅐KSHV-cyclin complexes autophosphorylate at one or several residues and this contributes to the activation; and/or (2) ATP causes conformational changes in Cdk6⅐KSHVcyclin complexes, which activate these complexes. When Cdk6⅐KSHV-cyclin complexes were incubated in vitro with increasing amounts of ATP, a substantial mobility shift of the KSHV-cyclin was observed (Fig. 1C, compare lane 1 to lanes 2 and 3). In addition, when Cdk6⅐KSHV-cyclin complexes were immunoprecipitated from transfected cell extracts and incubated in vitro with radiolabeled ATP (Fig. 1D, lane 2), both Cdk6 and the KSHV-cyclin incorporated radiolabeled phosphate. The KSHV-cyclin alone (in the absence of transfected Cdk6) was not labeled (Fig. 1D, lane 1). These results were consistent with the possibility that Cdk6⅐KSHV-cyclin phosphorylation might be related to their activation. To investigate this possibility further, I generated mutations in a number of potential phosphorylation sites.
Mutation of the KSHV-cyclin Affects Its Ability to Activate Cdk6 -Bacteria produced GST-KSHV-cyclin proteins bearing one or more mutations were assayed for activation of baculovirus produced GST-Cdk6 using the conditions described in Fig.  1A (buffer, ATP, and CAKϩATP treatment). All mutant KSHV-cyclin proteins were able to activate Cdk6 toward the substrates Rb ( Fig. 2A) and histone H1 (Fig. 2B). GST-KSHVcyclin ⌬N1-19 , ⌬N1-45, and T96A/S98A displayed similar characteristics as wild-type KSHV-cyclin; i.e. buffer treatment led to minimal activation, and activation by CAKϩATP was more effective compared with the ATP treatment. For a number of KSHV-cyclin mutants (T3A, T56A, S60A, S89A, S109 -117A, S125A/S127A, T146A, T152A, S166A, S215A/S217A, S226A/ S233A, and T244 -247A) similar results were obtained (data not shown). One of the mutants, KSHV-cyclin S41-48A , activated Cdk6 in the absence and presence of CAK equally strong. In contrast, GST-KSHV-cyclin S105A , S178A, T181A, T189 -198A, and S178 -198A activated Cdk6 more readily after ATP treatment compared with the CAKϩATP treatment. Other KSHVcyclin mutants (S9A/T15A, T34 -37A, T155A/S156A) displayed similar characteristics (data not shown). Interestingly, the activity after buffer treatment was increased in the KSHV-mutants T181A, T189 -198A, and S178 -198A. This indicates that the region of KSHV-cyclin around amino acids 180 -200, contributes to the amount of time it takes to activate Cdk6 (essentially during the 15 min of labeling) (see Table III and also discussion). When the KSHV-cyclin mutants were assayed for activation of Cdk6 toward the substrate histone H1, very similar results were obtained (Fig. 2B). There was one notable Cdk6⅐KSHV-cyclin Activity difference; deletion of the N terminus of KSHV-cyclin (⌬N1-19 and ⌬N1-45) resulted in minimal histone H1 activity. Given that these two mutants displayed normal activity toward Rb (see Fig. 2A), this was not a general defect. This change in substrate specificity of the N-terminal deletion mutants of the KSHV-cyclin was further investigated (see below, Fig. 3).
The differences observed for the KSHV-cyclin mutants (see Fig. 2) could be due to the time it takes to activate Cdk6 or due to changed binding affinity. Therefore, I performed experiments where I varied the length of incubation or the amount of KSHV-cyclin added in the reaction (Table III). To achieve 50% of the maximal Cdk6 activation, it required incubation of ATP with Cdk6 and wild-type KSHV-cyclin for 160 min using the substrate Rb or 311 min using the substrate histone H1 (Table  III). Most of the KSHV-cyclin mutants required only 10 -30 min for activation of Cdk6 for both Rb and histone H1. The exception was KSHV-cyclin S41-48A , which required approximately half the time than wild-type KSHV-cyclin needed. KSHVcyclin ⌬N1-19 needed ϳ9 times longer to activate Cdk6 toward histone H1 than Rb, consistent with a specific defect when the substrate histone H1 was used (see also Fig. 3). An inverse effect was observed with KSHV-cyclin S178A , which took four times longer to activate Cdk6 toward Rb compared with histone H1. The difference between these two mutants was KSHVcyclin S178A ⅐Cdk6 complexes reached similar maximal activity for Rb and histone H1 as all other mutants. In contrast, the theoretical maximal activity for ⌬N1-19 using the substrate histone H1 was ϳ6-fold lower than all other mutants (and wild type) KSHV-cyclin. 0.5 g of wild-type KSHV-cyclin was required to achieve 50% Cdk6 activity toward histone H1. Two KSHV-cyclin mutants (T34 -37A and T181A) required 10 -15 times less protein (0.03-0.04 g) compared with wild-type KSHV-cyclin. KSHV-cyclin T56A required two times more (1 g), whereas most of the other mutants required 2-4 times less protein compared with wild-type KSHV-cyclin (Table III). Fig. 2, it was shown that deletion of the N terminus of KSHV-cyclin resulted in low levels of histone H1 activity whereas the activity toward Rb was comparable to wild-type KSHV-cyclin. To investigate this phenomenon in more detail, the activity toward five different substrates, Rb, histone H1, p27, p53, and Cdc25 was compared. Wild-type KSHV-cyclin⅐Cdk6 complexes phosphorylated all substrates (Fig. 3, lanes 1-3). Similar results were obtained with two other mutants, T181A and S178 -198A (Fig. 3, lanes 7-12). The mutant KSHV-cyclin ⌬N1-45 phosphorylated Rb to similar levels as wild-type KSHV-cyclin in complex with Cdk6 (Fig. 3A, lanes  4 -6). In contrast, histone H1 was poorly phosphorylated by ⌬N1-45, leading to a 10 -15-fold decreased phosphorylation efficiency (Fig. 3B, lanes 4 -6). Nevertheless, phosphorylation of p27, p53, or Cdc25 by Cdk6/KSHV-cyclin ⌬N1-45 complexes was similar as wild type. These results suggested that the N terminus of the KSHV-cyclin contributes to the recognition of the substrate histone H1 (see also discussion).

The N Terminus of the KSHV-cyclin and Its Involvement in Substrate Specificity-In
To further investigate the function of the N terminus of the KSHV-cyclin, a synthetic peptide corresponding to the first 20 amino acids was synthesized (NH 2 -MATANNPPSGLLDPTL-CEDR-COOH). This peptide was used in competition experiments at concentrations up to 1 mM and had no effect on the ability of Cdk6⅐KSHV-cyclin complexes to phosphorylate either Rb or histone H1 (data not shown). Furthermore, the N terminus of cyclin D1 was replaced by the N terminus of the KSHVcyclin. Such a cyclin D1 mutant was able to phosphorylate Rb but not histone H1 (data not shown). These results indicated that the N terminus of the KSHV-cyclin is not sufficient to confer the ability to phosphorylate histone H1 in the context of cyclin D1. Most likely other domains of the KSHV-cyclin in addition to the N terminus contribute to the substrate specificity.
Expression of KSHV-cyclin Mutants in U2OS Cells-Although the KSHV-cyclin mutants tested in vitro (see Fig. 2) did not prevent activation of Cdk6⅐KSHV-cyclin complexes in the absence of CAK, I aimed to test them in a cellular system. The  Cdk6⅐KSHV-cyclin Activity drawback of the cellular system is that autophosphorylation cannot be investigated since there is plenty of CAK activity and high levels of ATP present. In the following experiments, it can be assumed that Cdk6 is phosphorylated at Thr-177 at all times. To test KSHV-cyclin mutants in cells, Myc-tagged KSHV-cyclin was co-transfected with GFP-Cdk6 or Cdk6-HA into U2OS cells. Cell extracts were assayed for their total kinase activity toward Rb (Fig. 4A, third panel from the bottom). Transfection of the vector without coding region resulted in little background activity (Fig. 4A, lane 1). Cell extracts expressing wild-type Myc-KSHV-cyclin together with GFP-Cdk6 yielded robust activity toward Rb (Fig. 4A, lane 2). Most of the KSHV-cyclin mutants displayed activity similar to wildtype KSHV-cyclin. However, four mutants displayed decreased activity; T34A/S35A/S37A, K106E, T189 -198A, and T244 -247A (Fig. 4A, lanes 5, 12, 21, and 24). Myc-KSHV-cyclin⅐GFP-Cdk6 complexes were immunoprecipitated and assayed for their activity toward Rb (Fig. 4A, top panel) and histone H1 (Fig. 4A, second panel from top). The kinase activities from the immunoprecipitates were similar as the ones from the lysates. To analyze the cyclin⅐Cdk complexes, I analyzed the amount of Cdk6 protein bound to Myc-KSHV-cyclin (Fig. 4A, third panel  from top). Similar amounts of Cdk6 were bound to most mutant KSHV-cyclin proteins compared with wild-type KSHV-cyclin. There were only four mutant KSHV-cyclins (T189 -198A, S215A/S217A, S226A/S233A, and T244 -247A) that displayed decreased Cdk6 binding, although the expression of Myc-KSHV-cyclin was also decreased in the extracts (Fig. 4A,  bottom panel, lanes 21-24).
To investigate if mutations in the KSHV-cyclin had any effect on the binding to Cdk6, I assayed immunoprecipitates using stringent conditions. Myc-KSHV-cyclin was co-transfected with Cdk6 and cell lysates were prepared. Myc-KSHV-cyclin⅐Cdk6 complexes were then immunoprecipitated in RIPA buffer (which contains Triton X-100, sodium desoxycholate, and SDS; see "Experimental Procedures") and immunoprecipitates were analyzed for their kinase activity and Cdk6 binding (Fig. 4B). Cdk6 did not bind to Myc-KSHV-cyclin T181A (Fig. 4B,  third panel from top, lane 11) and bound to a lesser extent to Myc-KSHV-cyclin T189 -198A, T244 -247A, T155A/S156A, K106E (lanes 12, 15, 9, and 7). All other mutant Myc-KSHVcyclin proteins bound similar amounts of Cdk6 as wild-type Myc-KSHV-cyclin. The amount of Cdk6 bound to Myc-KSHVcyclin directly correlated with the observed kinase activity toward Rb (Fig. 4B, top panel) and histone H1 (Fig. 4B, second  panel from top).
Expression of Cdk6 mutants in U2OS Cells-To learn more about the interaction of the KSHV-cyclin and Cdk6, I investigated the binding and resulting kinase activity of Cdk6/KSHVcyclin complexes. Cdk6-HA was co-transfected with wild-type GFP-KSHV-cyclin into U2OS cells and after 48 h of incubation, cell extracts were prepared. Such lysates were analyzed for their ability to phosphorylate GST-Rb-(605-928) or histone H1 in vitro. When Cdk6-HA or KSHV-cyclin alone was transfected, only background activity was detected (data not shown and see Fig. 4A, lane 1). Extracts prepared from cells co-transfected with wild-type Cdk6-HA and Myc-KSHV-cyclin resulted in robust Rb kinase activity (Fig. 5A, third panel from top, lane 1) and expressed Cdk6 (Fig. 5A, fourth panel from top) and Myc-KSHV-cyclin (Fig. 5A, third panel from bottom) at similar level. The Cdk6 mutants, D163N, T177A, and T177A/S178A displayed only low activity. All other mutants, displayed similar activity when compared with wild-type Cdk6. To exclude contribution of other kinases that are present in cell lysates and of the endogenous wild-type Cdk6, Myc-KSHV-cyclin⅐Cdk6 complexes were immunoprecipitated and analyzed for kinase activity (Fig. 5A, top panel). After immunoprecipitation, all Cdk6 mutants displayed similar activity as wild-type Cdk6 with the exception of D163N, T177A, and T177A/S178A. Cdk6 D163N was completely inactive, as expected and previously described (49,58,63). Low levels of Rb kinase activity were detected for Cdk6 T177A and Cdk6 T177A/S178A (Fig. 5A, top panel, lanes [11][12], consistent with previous findings (49,51). I also verified that similar amounts of Cdk6-HA were co-immunoprecipitated with KSHV-cyclin (Fig. 5A, second panel from top). As comparison, activation of Cdk6 by cyclin D1 was investigated. When
Since mutations in Cdk6 had little effect on the activity of KSHV-cyclin⅐Cdk6 complexes, the immunoprecipitation experiments were repeated under stringent conditions (Fig. 5B). Using these stringent conditions, KSHV-cyclin bound poorly to mutant Cdk6 S57A , Cdk6 T58A , Cdk6 T95A , Cdk6 T177A , and Cdk6 S290A (Fig. 5, third panel from top, lanes 3, 4, 9, 11, and  14). These results were different from Fig. 5A where gentle conditions (0.5% Nonidet P-40) were used. Nevertheless, this indicated that these mutations affected the ability of Cdk6 to bind to KSHV-cyclin. The resulting kinase activity toward Rb (Fig. 5B, top panel) and histone H1 (Fig. 5B, second panel from  top) correlated with the amount of bound KSHV-cyclin to Cdk6 (Fig. 5B, third panel from top).
Heterotrimeric Complex of Cdk6⅐KSHV-cyclin⅐p16 -From crystallographic studies, it has been shown that Cdk6, KSHVcyclin, and p16 form a heterotrimeric complex in vitro (52). Here, I tried to confirm these results when these proteins were expressed in mammalian cells. First of all, I checked if high levels of p16 were able to inhibit Myc-KSHV-cyclin⅐Cdk6 complexes in U2OS cells. High levels of p16 inhibited wild type Myc-KSHV-cyclin⅐Cdk6 but even more so Myc-KSHVcyclin ⌬N1-19 and Myc-KSHV-cyclin S178 -198A toward the substrates Rb (Fig. 6B, top panel) and histone H1 (Fig. 6B, second  panel from top). Wild-type Myc-KSHV-cyclin bound equal amounts of Cdk6 independent on the levels of p16 (Fig. 6B,  third panel from top). Furthermore, the heterotrimeric complex of wild-type Myc-KSHV-cyclin⅐Cdk6⅐p16 was observed ( Fig. 6B;  third, fourth, and fifth panel from top, lanes 3 and 4). Interestingly, the stoichiometry of the heterotrimeric complex did not change when lanes 3 and 4 were compared but there was no detectable kinase activity in lane 4. This indicates that the lack of kinase activity is not a result of p16 disrupting the wild-type Myc-KSHV-cyclin⅐Cdk6 complex. The two mutant Myc-KSHVcyclins were much more sensitive to p16 and in that case the binding of Cdk6 to the Myc-KSHV-cyclin was decreased. There-fore, Cdk6 binding to mutant Myc-KSHV-cyclin ⌬N1-19 or S178 -198A is weakened, especially in the presence of p16. DISCUSSION In this study, I have investigated the activation of Cdk6 by the KSHV-cyclin. Several mutants of the KSHV-cyclin were able to activate Cdk6 more effectively in the absence than in the presence of CAK. None of the mutations (or combination of mutations) rendered the KSHV-cyclin inactive or nonfunctional. Deletion of the N terminus of the KSHV-cyclin led to a change in substrate specificity. Expression of high levels of p16 in U2OS cells resulted in inactivation of KSHV-cyclin⅐Cdk6 complexes while forming a heterotrimeric complex.
Mutants were generated of the KSHV-cyclin that could only activate CAK-phosphorylated Cdk6 but not Cdk6 activated by ATP treatment. Some of the mutant KSHV-cyclins were activated preferentially by CAK phosphorylated Cdk6, others displayed higher activity after ATP treatment but none of them were activated only by one treatment. There are many potential reasons for this outcome. First of all, I have speculated that the effect of the ATP treatment among other possibilities could be caused by autophosphorylation or an ATP induced conformational change. To peruse phosphorylation events, I made a number of mutants that could not be phosphorylated at specific residues but these mutations had little effect on the activation of Cdk6⅐KSHV-cyclin complexes. It is possible that the KSHVcyclin is phosphorylated at several residues and that I have not mutated the right combination of residues. Another possibility is that both the KSHV-cyclin and Cdk6 are phosphorylated. Here I have not tested a large number of Cdk6 mutants since Cdks consist of only a minimal kinase domain and are notorious of being inactivated by even single residue mutations, which is not related to phosphorylation but rather to folding. Therefore, this question remains to be answered in the future. Nevertheless, some of the KSHV-cyclin mutants turned out to be quite interesting since the mutated residues affect the binding affinity to Cdk6. Furthermore, it is interesting to note that all KSHV-cyclin mutants remained active even when homologous cyclin D mutants were not active (e.g. T156A, Ref. 66 ATP treatment activated Cdk6 to similar levels than CAK phosphorylation in vitro (see Figs. 1 and 2). In contrast, when Cdk6 T177A was expressed in U2OS cells only a low level of activity was detected (Fig. 5, A and B, lane 11). Similar results have been described previously (49,51). There are two distinct possibilities that could lead to this difference: (1) the mutation at residue 177 could lead to structural distortions in the T-loop, which are unrelated to the phosphorylation state of threonine 177; and (2) in vitro, ATP treatment of Cdk6⅐KSHV-cyclin complexes results in (auto)-phosphorylation of Thr-177 in Cdk6. By phosphopeptide mapping, we have shown that Cdk6 was phosphorylated on Thr-177 in these conditions (data not shown). It is possible that both possibilities apply under certain conditions.
The determination of the substrate specificity of Cdk⅐cyclin complexes is of great importance since only few substrates have been identified. Cyclins have been implied to be determinants of the substrate specificity (32)(33)(34). For example, Cdk6⅐cyclin D complexes phosphorylate Rb only, whereas Cdk6⅐KSHV-cyclin complexes phosphorylate a wide range of substrates (see Introduction). It was not known which regions of the KSHV-cyclin contribute to the change in substrate specificity. I have shown here that two different deletions of the N terminus of the KSHV-cyclin display reduced phosphorylation of histone H1 whereas phosphorylation of other substrates was unchanged (see Fig. 3). This indicates that the N terminus of the KSHVcyclin is required to phosphorylate histone H1. It is interesting to note that the N terminus of the KSHV-cyclin directly interacts with the T-loop of Cdk6 (53). Therefore, it is possible that deletion of the N terminus of the KSHV-cyclin leads to a conformational change of the T-loop, which then affects histone H1 phosphorylation. Nevertheless, this possibility is unlikely since phosphorylation of other substrates (Rb, p27, p53, Cdc25) was not affected (see Fig. 3). The N terminus of cellular cyclins has not been shown to affect substrate specificity. Deletion of the N terminus of cyclin A has no obvious effect on phosphorylation of histone H1 (3,69) or Rb 2 but deletion of the N terminus of cyclin D (70) or cyclin H (71) abolishes activity. Nevertheless, it is difficult to compare cyclin A to the KSHV-cyclin (or to the D-type cyclins) since cyclin A contains an N-terminal extension of 150 amino acids. Histone H1 does not (detectably) physically interact with the Cdk⅐cyclin complexes whereas Rb and p27 bind to the Cdk subunit and the Cdk and cyclin subunits, respectively. Mutations in cyclins have been described that weaken the interaction with the Rb or p27 and therefore decrease phosphorylation of these substrates (35)(36)(37). Deletion of the N terminus of the KSHV-cyclin provides the first hint for a determinant of a non-binding substrate like histone H1. It can be assumed that the majority of Cdk substrates are rather non-binding (like histone H1) than binding (like Rb) substrates since these substrates display minimal processivity due to a low dissociation constant.
Co-expression of the KSHV-cyclin and Cdk6 in U2OS cells leads to high levels of cell death (54,55). I have tested the mutant KSHV-cyclins (see Fig. 2) for their ability to induce apoptosis when co-expressed with Cdk6. All mutants were able to induce cell death in U2OS cell (judged by nuclear morphology; data not shown), which is consistent with the fact that all mutant KSHV-cyclin displayed kinase activity (see Fig. 2). I also measured the activity of caspase-3 and caspase-9 in these cells (assay described in Refs. 72 and 73) and found that the caspase activity was unchanged compared with untransfected cells but was lower than in UV-treated cells (data not shown). This indicates that the cell death induced by the KSHV-cyclin⅐Cdk6 is independent on caspase activation, which is consistent with Bcl-2 not being able to rescue this phenotype (55).
The Ink4 inhibitor p16 binds to the N-terminal lobe of Cdk6 (or Cdk4) (29,30). Immunoprecipitation of p16 from cell extracts indicated that only Cdk4 bound but cyclin D was not co-immunoprecipitated (24). This led to the conclusion that p16 would displace cyclin D when it binds to Cdk6 (or Cdk4). Nevertheless, it is also possible that cyclin D is such an unstable protein (68) that its levels are too low to detect in such experiments. In vitro, p16 was shown to form a heterotrimeric complex with purified Cdk6 and the KSHV-cyclin (49, 52) but p16 does not interact directly with the cyclin subunit (52). In this study, p16 binds to the Cdk6⅐KSHV-cyclin complex in cells (see Fig. 6B). In addition, this heterotrimeric complex can be active or not (Fig. 6B, compare lane 3 to lane 4) indicating that there might be two different modes of p16 binding to the Cdk6⅐KSHV-cyclin complex. Interestingly, deletion of the N terminus or mutation of the region around residue 180 -200 of the KSHV-cyclin led to destabilization of the heterotrimeric p16 complex formation.