Binding of HTm4 to Cyclin-dependent Kinase (Cdk)-associated Phosphatase (KAP)·Cdk2·Cyclin A Complex Enhances the Phosphatase Activity of KAP, Dissociates Cyclin A, and Facilitates KAP Dephosphorylation of Cdk2*

Cyclin-dependent kinase 2 (cdk2) activation requires phosphorylation of Thr160 and dissociation from cyclin A. The T-loop of cdk2 contains a regulatory phosphorylation site at Thr160. An interaction between cdc-associated phosphatase (KAP) and cdk2 compromises the interaction between cdk2 and cyclin A, which permits access of KAP, a Thr160-directed phosphatase, to its substrate, cdk2. We have reported that KAP is bound and activated by a nuclear membrane protein, HTm4. Here, we present in vitro data showing the direct interaction between the HTm4 C terminus and KAP Tyr141. We show that this interaction not only facilitates access of KAP to Thr160 and accelerates KAP kinetics, but also forces exclusion of cyclin A from the KAP·cdk2 complex.

HTm4 (MS4A3) is the third member of subfamily A in an extensive membrane-spanning four-domain gene family. These genes are only loosely related at the sequence level, but their encoded proteins share a common four-transmembrane topology, including CD20 (MS4A1) and Fc⑀RI (MS4A2). To date, few functions for the MS4 family of proteins have been ascribed. However, a diverse functionality is beginning to emerge. These functions include roles such as cell surface signaling receptors and intracellular adapter proteins (1)(2)(3)(4)(5)22).
Cell cycle progression is regulated by the sequential activation, and inactivation, of the cyclin-dependent kinases (cdks). 1 Cdks, themselves, are substrates for regulatory phosphoryla-tion and dephosphorylation. The phosphorylation status of cdks is controlled by a discrete class of regulatory kinases and phosphatases (6). Cdk activity is controlled by activatory and inhibitory proteins, the latter exemplified by p21 and p27. Binding of inhibitors such as p21 and p27 to different cyclin⅐cdk complexes is sufficient to arrest the cell cycle (7)(8)(9)(10)(11)(12). Conversely, inactive, monomeric cdk can be activated via association with a specific cyclin and the concurrent phosphorylation of a conserved and essential threonine residue, such as threonine 160 (Thr 160 ) in cdk2, which is located within the activation segment (T-loop) of the kinase.
In addition to cyclin association, full activation of cdk2 requires phosphorylation of Thr 160 and dephosphorylation of Thr 14 and Tyr 15 . Thr 160 of cdk2 is phosphorylated by CAK (cdk-associated kinase), whereas its dephosphorylation is critical for inactivation. This dephosphorylation is executed by a serine/threonine-directed phosphatase, KAP. It has been shown that exogenous expression of KAP slows the G 1 phase cell cycle progression in HeLa cells and that aberrant KAP transcripts are detected in some hepatocellular carcinomas (13,14). These observations suggest that KAP has the same biological effects as cdk inhibitors, although their modes of actions are different. KAP can bind to cdk2 either in the presence or absence of cyclins (15)(16)(17). However, KAP can only dephosphorylate cdk2 when cyclin A is degraded or disassociated, in a mechanism that may control access of KAP to its substrate Thr 160 (16). Cyclin binding may therefore control access of KAP to its substrate phosphothreonine 160. We reported previously that HTm4 interacts directly with the KAP via its C terminus (18,19) and that exogenous expression of HTm4 leads to dephosphorylation of cdk2 and cell cycle arrest at the G 0 /G 1 phase. Here, we show that the presence of HTm4 in the KAP⅐cdk2⅐cyclin A complex controls cdk2 activity in a dual fashion. First, HTm4 binding causes exclusion of cyclin A from its interaction with cdk2. Second, HTm4 binding potentiates KAP enzymatic activity and causes conformational changes that regulate access to Thr 160 .
Recombinant HTm4 was purified from transformed JM109 Escherichia coli as follows. Protein production was induced by the addition of isopropyl 1-thio-␤-D-galactopyranoside to bacterial liquid culture in logarithmic phase. Bacteria were harvested by centrifugation at 6,000 rpm for 30 min. The bacterial pellet was lysed by sonication on ice (20 KW, 15 min) in 50 ml of N buffer (20 mM Tris-HCl, pH 8.0, containing 0.1 M NaCl, 1 mM phenylmethylsulfonyl fluoride, 1 mM ␤-mercaptoethanol). The bacterial lysate was centrifuged at 20 ϫ 10 4 g at 4°C for 30 min. The remaining pellet was lysed in 40 ml of N buffer containing 6 M guanidine HCl. Proteins were bound to a 20-ml nickel-chelated Sepharose column (Amersham Biosciences) and were refolded by exposure to a 6 -0 M urea gradient at 0.2 ml/min. The refolded protein sample was eluted with a gradient of N buffer containing 0.5 M imidazole and 0.3% deoxycholate. The peak fraction containing HTm4 was collected and purified by XArrest agarose after His tag cleavage with 4 units of Factor Xa (Novagen). Wild-type KAP, inactive KAP (C140S), and mutant KAP were expressed and purified as described previously (20).
Western Blotting of HTm4 -For the Western blot, 10 l of purified protein was boiled in SDS sample buffer and resolved by 15% PAGE. After electrotransfer to polyvinylidene difluoride, the membrane (Bio-Rad) was blocked followed by either incubation with either anti-His tag antibody diluted at 1:5,000 (MBL) or anti-HTm4 antibody (Zymed Laboratories Inc.) and horseradish peroxidase-conjugated anti-rabbit IgG. Signals were detected using ECL (Amersham Biosciences).
Protein Interaction Analyses (Surface Plasmon Resonance Assay)-The immobilization of proteins to sensor chips (CM-5 chip) was performed using a carbodiimide covalent linkage protocol (Amersham Biosciences). KAP was coupled to the sensor chip at 2,800 resonance units (RU). Nonspecific human immunoglobin was coupled to the surface of the sensor chip (anti-␤2m antibody BBM.1) at 2,900 RU. Protein interaction assays were carried out using the BioSensor Biacore 2000 (Biacore). An HTm4 C-terminal peptide composed of 22 amino acids (CNANCCNSREEISSPPNSV) was synthesized. Either full-length HTm4 protein (20 M), or HTm4 C-terminal peptide (1 mM) in 50 mM HEPES containing 1 mM EDTA, 1 mM DTT, 5% glycerol, and 0.1% Tween 20 was examined at a flow rate of 10 l/min.
Protein Activity Assays-KAP phosphatase activity was examined by the detection of hydrolysis of p-nitrophenyl phosphate (pNPP) (WAKO) as described previously (20). The time course of phosphatase activity for KAP was examined. 20 l of reaction mixture containing 50 mM Tris-HCl, pH 7. Kinetic Measurements-To investigate the effect of HTm4 on KAP, the initial reaction velocity of KAP was assessed using different concentrations of active cdk2⅐cyclin A as the substrate in the presence or absence of HTm4. The results were analyzed further using a Lineweaver-Burk plot.
KAP Dephosphorylation of Peptide or Protein Substrates-KAP dephosphorylation of phosphorylated proteins or peptide substrates was examined using the Ser/Thr Phosphatase Assay Kit 1 (Upstate Biotechnology) according to the manufacturer's instructions. Phosphorylated Thr 160 of cdk2 and K-R-pT-I-R-R peptide were used as substrates for the assay. Concentrations of KAP and HTm4 were both 20 M; concentrations of p-cdk2 and peptide substrates (K-R-pT-I-R-R) were 10 and 25 M, respectively. Reactions were performed in pNPP Ser/Thr Assay Buffer, containing 50 mM Tris-HCl, pH 7.0, and 100 M CaCl 2 . To examine the importance of the HTm4 C-terminal region and to confirm the KAP phosphatase activity enhancement effect of HTm4, an HTm4 C-terminal peptide composed of 22 amino acids (CNANCCNSREEIS-SPPNSV) was synthesized. The phosphatase enhancement activity of the HTm4 C-terminal region was examined using the assay protocol described above. All assays were performed in triplicate, with the average values and standard deviations displayed on the plot (see Figs. 2

and 3).
Circular Dichroism Spectroscopy-Circular Dichroism (CD) spectroscopic measurements were performed on a Jasco J-700 (Jasco, Japan). Spectra measurements were recorded from 250 to 320 nm. The spectra change of KAP was measured by comparing two different KAP samples in the same solution (20 mM Tris-HCl, pH 7.2, 0.15 M NaCl). Sample A contained 5 M KAP and an excess amount of HTm4 C-peptide (50 M). The contents of sample B were identical to those in sample A, except for HTm4 C-peptide. The difference in spectra was further converted to parameter (mean residue ellipticity) and plotted on the graph in Fig. 4.
Cyclin A Dissociation Assay-The status of the cdk2⅐cyclin A complex was examined in the presence and absence of HTm4. Recombinant cdk2⅐cyclin A (Upstate) and His-tagged KAP were combined in vitro and incubated in the absence or presence of either 1 or 10 M HTm4. The samples were added to a nickel magnet bead matrix (Toyobo) to capture His-tagged KAP protein and any associated binding partners. The bound fractions were resolved by SDS-PAGE and Western blotted for the presence of cyclin A (Santa Cruz Biotechnology). An alkaline phosphatase conjugated to anti-rabbit IgG (EY Laboratory) was used to visualize the amount of bound cyclin A present in the KAP⅐cdk2 complex. 10 l of 10 M cdk2⅐cyclin A and 10 l of 10 M His-tagged KAP were mixed in 20 mM Tris-HCl, pH 7.5, buffer containing 0.15 M NaCl, 1 mM ␤-mercaptoethanol, and 0.1% Tween 20. After incubating at room temperature for 5 min, either 10 l of 10 M HTm4 (without His tag) or its buffer was added to the reaction mixture. After incubating at room temperature for 5 min, 5 l of nickel magnet beads was added to each mixture. The fractions bound to His-tagged KAP were washed with the reaction buffer and boiled for SDS-PAGE. The gel was stained with a silver staining kit (ATTO).

Direct Binding of KAP and HTm4 in Vitro-
We have shown previously that HTm4 and KAP phosphatase may be coimmunoprecipitated from hematopoietic cells (18). Because these data form the basis for a novel model where HTm4 is a component of the cell cycle machinery, it is important that we determine whether the interaction is direct or if it involves an intermediary protein. We have since established a purified protein-protein interaction system that allows us to examine whether HTm4 and KAP interact directly.
His-tagged HTm4 was produced in E. coli and purified by fast protein liquid chromatography (Fig. 1a). The eluted peak corresponding to HTm4 was visualized by Coomassie Blue staining (Fig. 1b) and then analyzed by Western blot. Both anti-HTm4 and anti-His antibodies detected a 27-kDa band corresponding to HTm4 (Fig. 1, c and d). Surface plasmon resonance was used to detect direct binding of purified HTm4 to KAP. Full-length KAP protein was coupled to a sensor chip CM-5 shell (Amersham Biosciences). Compared with either uncoupled or nonspecifically coupled chips (coupled to anti-␤2m antibody BBM.1), an immediate increase in mass was observed when either full-length HTm4 (Fig. 2a) or a C-terminal peptide from HTm4 ( Fig. 2b) was passed over KAP-coupled sensor chips. Moreover, premixing HTm4 with KAP abolished the KAP/HTm4 selective binding (data not shown). These in vitro data confirm that HTm4 and KAP interact directly and that the HTm4 C terminus is necessary and sufficient to mediate binding to KAP.
KAP Phosphatase Activity Is Enhanced in the Presence of HTm4 -HTm4 is a direct binding partner for KAP, as shown in our previous yeast two-hybrid data. We hypothesize that HTm4 may regulate the enzymic activity of the KAP phosphatase. In the current experiment, KAP phosphatase activity was measured using pNPP as a generic substrate for KAP. In the presence of HTm4 and the cdk2⅐cyclin A-binding complex, a bipha-sic kinetic for KAP activity was observed. As shown in Fig. 3a, after the addition of pNPP substrate, an exponential phase (in the first 20 min) was followed by a plateau phase (after 30 min). The presence of HTm4 linearly increased the phosphatase activity of KAP (y ϭ 1.45x ϩ 0.01; r 2 ϭ 0.98) (Fig. 3b). Without HTm4, the V max and K m for KAP phosphatase activity were 1.30 nM/min and 2.68 M, respectively (Fig. 3, c and d). However, when HTm4 was present, the V max of KAP phosphatase activity increased dramatically to 17.5 nM/min. A parallel increase in K m (to 8.33 M) was also observed. Taken together, these data suggest that the presence of HTm4 directly affects the activity of the KAP phosphatase.
KAP Dephosphorylation of Peptide or Protein Substrates-KAP dephosphorylation of phosphorylated proteins or peptide substrates was examined using the Ser/Thr Phosphatase Assay Kit 1 according to the manufacturer's instructions. Phosphorylated Thr 160 of cdk2 (p-cdk2) and K-R-pT-I-R-R peptides were used as substrates for the assay. Concentrations of KAP and HTm4 were both 20 M; concentrations of p-cdk2 and peptide substrates (K-R-pT-I-R-R) were 10 and 25 M, respectively. Reactions were performed in pNPP Ser/Thr Assay Buffer, containing 50 mM Tris-HCl, pH 7.0, and 100 M CaCl 2 . To examine the necessity of the HTm4 C-terminal region and to confirm the KAP phosphatase activity enhancement effect of HTm4, a peptide containing the last 22 amino acids of the HTm4 C terminus (CNANCCNSREEISSPPNSV) was synthesized. The phosphatase enhancement activity of the HTm4 C-terminal region was examined using the assay protocol described above. All assays were performed in triplicate, with the average values and standard deviations displayed on the plot.
We further evaluated the major factors affecting KAP phosphatase activity. As we have shown, KAP phosphatase activity is significantly enhanced in the presence of HTm4. When an inactive KAP mutant (C140S) is substituted for the wild-type KAP, the phosphatase activity is almost completely abolished. The inactive KAP mutant thus serves as a negative control (15,20). We found that when HTm4 was absent or was replaced with a nonrelated peptide, KAP phosphatase activity was almost completely inhibited compared with control reactions containing wild-type KAP, HTm4, and cdk2⅐cyclin A complex (Fig.  3e). It is interesting to note that a cdk2⅐cyclin A complex, rather than either a monomeric cdk2 or cyclin A, was required for a stronger KAP phosphatase activity induced by HTm4. Additionally, our data suggest that KAP activity is optimal in the context of a multicomponent protein complex that contains its substrate, cdk2, and the HTm4 protein. Our data demonstrate that wild-type KAP and HTm4 are essential for KAP phosphatase activity.
Based on our previous yeast two-hybrid and immunoprecipitation data, we aimed to confirm whether the C-terminal region of HTm4, alone, was sufficient to mediate the HTm4-KAP interaction and to induce the KAP phosphatase activity. To this end, we utilized the same synthesized peptide containing the hypothesized KAP binding domain (CNANCCNSREEISSP-PNSV), as described above. When introduced into our in vitro KAP activity assay, the HTm4 C-peptide showed a comparable enhancement effect on KAP phosphatase activity compared with HTm4 wild-type protein (Fig. 3e). These data both confirm our previous observation that the C terminus of HTm4 mediates direct binding of HTm4 to KAP and show that the C terminus is the functional domain of HTm4 which regulates KAP activity.
HTm4 Controls the Phosphatase Activity of KAP toward Its Physiological Substrates-The data presented above demonstrate that HTm4 enhances the activity of the KAP phosphatase toward a generic substrate compound, pNPP. We asked whether HTm4 could also modify the activity of KAP toward its physiological substrates. We evaluated the ability of KAP to dephosphorylate two of its substrates (Fig. 3f). First, we examined cdk2 with a phosphorylated Thr 160 (p-cdk2), which acts as a natural substrate for KAP. We then used a synthesized threonine phosphopeptide, K-R-pT-I-R-R (Upstate). We found that HTm4 increases the dephosphorylating activity of KAP toward phosphorylated cdk2 by more than 6-fold, when cdk2 is presented in the context of the cdk2⅐cyclin A complex. We also demonstrated that HTm4 potentiates dephosphorylation of the synthesized phosphopeptide K-R-pT-I-R-R. However, no KAP dephosphorylation activity could be detected if HTm4 was not presented. Interestingly, no dephosphorylation effect was observed when phosphorylated cdk2 was presented in the cdk2⅐cyclin A-binding complex. These data imply that the presence of HTm4 confers a substrate-specific enhancement of KAP phosphatase activity.
Structural Change of KAP by Adding C-peptide of HTm4 -Based on our observation that C-peptide of HTm4 protein enhanced KAP activity, we hypothesized that the binding of HTm4 C-peptide to the KAP molecule induced KAP conformational HTm4. e, effects of HTm4 on KAP phosphatase activity toward pNPP substrate. Activities of either wild-type (WT) or C140S inactive mutant KAP were examined in the different reaction mixtures, with or without 20 M HTm4 (its C-terminal peptides/nonrelated peptides at 0.1 mM) or with 4 M active cdk2⅐cyclin A (cdk2 monomer). f, effects of HTm4 on KAP phosphatase activity with phosphorylated cdk2 or cdk2-derived peptide as its substrate. Phosphorylated threonine 160 from cdk2⅐cyclin A, active, and phosphopeptide (K-R-pT-I-R-R) were used as substrates to assess KAP activity. Assays were performed using the Ser/Thr Phosphatase Assay Kit 1. Concentrations of KAP and HTm4 were both 20 M, and concentrations of p-cdk2 and peptide (K-R-pT-I-R-R) were 10 and 25 M, respectively. Reactions were made in the pNPP Ser/Thr Assay Buffer, which contains 50 mM Tris-HCl, pH 7.0, and 100 M CaCl 2 . The reaction was allowed to proceed for 30 min at 30°C, and absorbance was measured at 650 nm.
change. To confirm this hypothesis, we examined conformational changes of KAP in the presence or absence of the C-peptide by using CD spectroscopy. The CD spectrum of a protein in the "near-ultraviolet" spectral region (250 -350 nm) can be sensitive to certain aspects of tertiary structure. At these wavelengths the chromophores are the aromatic amino acids and disulfide bonds, and the CD signals they produce are sensitive to the overall tertiary structure of the protein. Signals in the region from 250 to 270 nm are attributable to phenylalanine residues, signals from 270 to 290 nm are attributable to tyrosine, and those from 280 to 300 nm are attributable to tryptophan. Disulfide bones give rise to broad weak signals throughout the near-ultraviolet spectrum. As shown in Fig. 4, we found a spectral change at 270 -280 nm in the difference spectrum of KAP between sample A (with HTm4 C-peptide) and sample B (without C-peptide). The 270 -290 nm difference spectrum demonstrated a conformational change in tyrosine when HTm4 C-peptide was added into KAP solution. Because HTm4 C-peptide has no tyrosines residues, this major change in the spectrum must be caused by conformational changes in the KAP tyrosine residues. Hence, this result demonstrated that HTm4 could cause direct conformational/tertiary structural change in KAP after its binding to KAP by the C terminus.
HTm4 Dissociates Inhibitory Cyclin A from the cdk2⅐cyclin A-binding Complex-The activity of cdk2 is controlled by its protein-protein interactions with KAP and cyclin A. The binding of KAP phosphatase to the cdk2⅐cyclin A complex dephosphorylates, and hence inactivates, cdk2. However, the binding of cyclin A inhibits the ability of KAP to dephosphorylate cdk2. Dissociation of cyclin A by HTm4 facilitates cdk2 dephosphorylation by KAP. Interestingly, HTm4 enhances the KAP dephosphorylation effect even on phosphorylated cdk2 that is presented in a complex with cyclin A. This result suggests that HTm4 may facilitate cdk2 inactivation by alleviating the inhibitory effect of cyclin A. Our model suggests that HTm4 may dually regulate cdk2 activity, via KAP, and by physically excluding cyclin A from its inhibitory interaction with cdk2.
To verify our hypothesis, we used an affinity purification method to evaluate the cdk2-cyclin A interaction both with and without HTm4. We isolated His-tagged KAP protein using a Ni 2ϩ binding column. Without HTm4, cyclin A remained in the His-purified KAP⅐cdk2⅐cyclin A-binding complex (Fig. 5a). FIG. 4. Plot of KAP CD spectra change. Spectra measurements were recorded from 250 to 320 nm. The spectra change of KAP was measured by comparing two different KAP samples in the same solution (20 mM Tris-HCl, pH 7.2, 0.15 M NaCl). Sample A contains 5 M KAP and an excess amount of HTm4 Cpeptide (50 M). Sample B was formulated as Sample A, with the exception of the HTm4 C-peptide. The difference in spectra was further converted to parameter (mean residue ellipticity) and plotted.
FIG. 5. Effects of formation of the cdk2⅐cyclin A complex in the presence or absence of HTm4. a, in the absence of HTm4. 10 l of 10 M cdk2⅐cyclin A and 10 M His-tagged KAP were mixed, and the same amount of 10 M HTm4 (left), 1 M HTm4 (middle), or its buffer (right) was added to the mixture. After that, 5 l of nickel magnet beads was added. Bound fractions were electrophoresed by SDS-PAGE and then detected by Western blotting with anti-cyclin A antibody as the primary antibody, and alkaline phosphatase-conjugated anti-rabbit IgG was used as the secondary antibody. The band of cyclin A is indicated with an arrow. b, HTm4 released cyclin A. 10 l of 10 M cdk2⅐cyclin A and 10 l of 10 M His-tagged KAP were mixed in 20 mM Tris-HCl, pH 7.5, buffer containing 0.15 M NaCl, 1 mM 2-mercaptoethanol, and 0.1% Tween 20. After incubation at room temperature for 5 min, 10 l of 10 M HTm4 (without His tag) or its buffer was added to the mixture and allowed to incubate at room temperature for 5 min. 5 l of nickel magnet beads was added to each mixture. The bound fractions through His-tagged KAP were washed with the Tris buffer and boiled for SDS-PAGE. The gel was then silver stained. Bands of cyclin A, cdk2, and KAP are indicated with arrows. FIG. 6. KAP, cdk2, and cyclin A complex model. a, KAP, cdk2, and cyclin A interaction. Coordinates from cdk2⅐cyclin A and cdk2⅐KAP from the PDB have been structurally aligned. A CPK trace has been labeled. Both phosphorylated Thr 160 residues of the T-loops are shown on the CPK model. The red T-loop is from cdk2⅐KAP, and the blue T-loop is from cdk2⅐cyclin A. Extension of the blue T-loop is inhibited by an ionic interaction between Arg 157 of cdk and cyclin A Glu 268 , as shown in Fig. 8a. b, stereo view of model complex cdk2 (yellow)⅐KAP (blue) and cyclin A (pink). This view shows the reverse side of Fig. 8c, in a CPK model. The green molecule is Thr 160 , and phosphorus is labeled with red. The cyclin A molecule is viewed from the back of the cdk2⅐KAP complex.
However, when purified HTm4 (after His tag cleavage) was added to the KAP⅐cdk2⅐cyclin A reaction mixture, cyclin A dissociated from the complex in a dose-dependent manner, and the remaining protein complex was found to contain only KAP⅐cdk2⅐HTm4 (Fig. 5b). These observations demonstrate that HTm4 inactivates cdk2 by activating KAP and by causing concomitant dissociation of cyclin A from cdk2. Together with the observation that HTm4 enhances KAP phosphatase activity, we suggest that HTm4 is positioned as a key regulator in the activation of KAP and for cdk2 activity.

Mechanism for HTm4-induced Dissociation of Cyclin A from
Cdk2⅐Cyclin A-In silico modeling was used to probe the potential nature of the HTm4 regulatory interaction that promotes KAP phosphatase activity and potentially leads to the exclusion of cyclin A. Based on our findings, we show that KAP undergoes conformational changes following allosteric binding to the HTm4 C-terminal domain and that these conformational changes may subsequently facilitate the interaction between the KAP substrate and its active site. To examine this hypothetical model, we produced a steric structure model using the coordinates of cdk2⅐cyclin A (1JST) (21) and cdk⅐KAP (1FQ1) (20) from the Protein Data Bank (PDB). These structures were overlapped utilizing Homology in Insight II software (MSI). The resulting model complex is shown in Fig. 6a. Both cdk structures significantly overlap, and the overall RMS deviation is 0.92 Å, with the exception of the T-loop region. In this region, marked differences in cdk structure can be noted, depending on whether coordinates are derived from the cdk2⅐cyclin A (blue loop) or the cdk2⅐KAP complex (red loop). There is a deep concave face, as shown in Fig. 6b. We hypothesize that the concave face forms a potential docking site for the HTm4 C terminus. On the reverse side of the interface, phosphothreonine 160 of cdk2 can be inserted into the entrance of the active Pro and the phosphorylation site of Thr are colored with pink and yellow. Conserved Arg 157 is indicated with an arrow. c, stereo view of the T-loop side in cdk2⅐cyclin A (1JST). Pro 155 of cdk2 (carbon, green; oxygen, red; nitrogen, dark blue) is seen in the deep concave. The Pro residue seems not to be accessible by an isomerase. Right, cyan is cdk2; left, white molecule is cyclin A. d, this view is a reverse side of Fig. 6b (KAP, cdk2 active site) and is also shown by CPK model. Thr 160 of cdk2 (yellow) and cyclin A (pink) are seen on the back side of the cdk2⅐KAP complex (KAP, blue; cdk2, white). We can stereographically see a broad concave that HTm4 C-terminal domain would bind. Tyrosines residues 87 (dark blue) and 141 (green) of KAP represent the candidate conformational change sites after the HTm4-C-terminal binding shown by CD difference spectrum (Fig. 4). site of KAP in the absence of cyclin A, whereas a small molecule of substrate such as pNPP may still be able to enter the entrance even in the presence of cyclin A. This finding may explain why KAP activity toward pNPP is enhanced at moderate levels in the presence of the cdk2⅐cyclin A complex.

Mode of Structural Change of the Cdk2 T-loop in the Absence and Presence of HTm4 -
The T-loop is a conserved segment in cdks, which plays an important role in cdk activation. In the cdk2⅐cyclin A complex, phosphorylation is associated with the mobility of the T-loop (12). However, viral cyclins (K-, H-cyclins) can still activate cdk6 without Thr 177 phosphorylation, which corresponds to Thr 160 phosphorylation in cdk2. The degree of interaction between the T-loop of cdk6 and V-cyclin is shown to be critical because the interaction area between cdk6 and V-cyclin was 20% larger than that of cdk2 and cyclin A (15). This increased interaction is contributed mainly by the T-loop of cdk6. Hence, the movement of the T-loop is critical for cdk activation.
In the cdk2⅐cyclin A activation model, van der Waals interactions exists between Ala 151 , Phe 152 , and Tyr 159 of the cdk2 T-loop and Phe 267 , Ile 182 , and Ile 270 of cyclin A (Fig. 7a) (6). However, in our KAP⅐cdk2⅐cyclin A complex model, the second cdk2-cyclin A interaction (Phe 152 -Ile 182 ) is completely abolished if the T-loop is moved because of the interaction of KAP (Fig. 7b). This result demonstrates that binding of KAP to cdk2 weakens the interactions between cdk2 and cyclin A. Moreover, it is reported that interactions between the T-loop of cdk2 and KAP comprise van der Waals interactions between residues Tyr 141 and Phe 53 of KAP with residue His 161 of cdk2 (20).
Besides the van der Waals interaction between the T-loop of cdk2 and cyclin A, a specific ionic interaction also exists between the -NH at Arg 157 of cdk2 and the -OH of Glu 268 of cyclin A in cdk2⅐cyclin A complex. This interaction disappears when the cyclin A molecule is dissociated, as seen in the cdk2⅐KAP complex (Fig. 8a). Thus, the extension of the T-loop in cdk2⅐cyclin A (blue) is inhibited by this ionic interaction.
In PDB models, more than 20 sets of coordinates are deposited for cdk2, both with and without cyclin A or its inhibitors. We investigated each structure of the T-loop and measured dihedral angles between Val 154 and Pro 155 , which are conserved in several cdks. Fig. 8b shows amino acid multialignments for several cdks. When cdk2 is associated with cyclin A, the T-loop is shortened with cis-Pro at position 155. Conversely, without cyclin A, the loop is extended with trans-Pro (Table I). When bound to KAP (without cyclin A), the T-loop of cdk2 shows extension in the trans form. According to the predicted crystal structure of cdk2⅐cyclin A, this cis-trans conversion at Pro 155 is thought to occur automatically because the residue Pro 155 would no longer be accessible to an isomerase (Fig. 8c).
When cyclin A dissociates from cdk2, the T-loop will extend through cis-trans conversion at the Pro 155 residue, and phosphothreonine 160 can then be inserted into the active site of KAP. Thus, the susceptibility of phosphothreonine 160 to KAP phosphatase occurs in response to HTm4 binding to KAP. Pro 155 commonly exists in cdk3, cdk5, but not cdk4, cdk5 as shown in Fig. 8b, whereas its counterpart residue Glu 258 is conserved in cyclins A, B, E, and C but not in cyclin D. This disparity suggests that mechanisms for cdk inactivation by KAP may differ between cdk species and hence between different phases of the cell cycle.
Binding Site of HTm4 -In Fig. 8d, we show the close allosteric position between Thr 160 of cdk2 and Tyr 84 , Tyr 141 of KAP, which suggests that the concave face of KAP accepts the HTm4 C terminus via the Tyr 141 residue of KAP. Immediately following HTm4 binding, the conformational change of the Tyr 141 residue of (Fig. 4) will weaken the van der Waals interaction between cdk2 and cyclin A and then affect continuously alongside T-loop (from residues 142 to 165) involving Ala 151 , Phe 152 , and Tyr 159 . This HTm4 binding eventually dissociates cyclin A from cdk2, thereby inducing a structural change of cdk2. At the reverse side of the interface, phosphothreonine 160 of cdk2 is inserted into the entrance of the active site of KAP, after cyclin A is released.
Model of Action Mechanism for HTm4 -HTm4 has four transmembrane domains (22). Topologically, this structure provides two tails that are available for protein-protein interactions (2). HTm4 localizes to the nuclear membrane. We hypothesize that the tails of HTm4 extend into the nuclear lumen and tether the KAP⅐cdk2⅐cyclin A complex (Fig. 9). The interaction between HTm4 and the KAP⅐cdk2⅐cyclin A complex is likely reversible and may occur only at certain stages in the cell cycle. HTm4 significantly affects the activity of KAP, cdk2, and cyclin A. HTm4 also regulates cdk2 activity in a dual fashion, by concurrently activating KAP activity and facilitating the accessibility of Thr 160 to KAP by causing dissociation of inhibitory cyclin A (Fig. 9). In conclusion, we show that HTm4 regulates the cell cycle in hematopoietic cells through its ability to control cdk2 status. T. Tsukihara and Dr. K. Ogasawara of Osaka University for CD measurement.