| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* Division of Structural Biology, Biozentrum, University of Basel, Basel, Switzerland;
European Molecular Biology Laboratory, Hamburg Outstation, c/o DESY, Hamburg, Germany and Institute of Crystallography, Russian Academy of Sciences, Moscow, Russia; and
Biochemisches Institut, Universität Zürich, Zürich, Switzerland
1Correspondence: Division of Structural Biology, Biozentrum, University of Basel, Klingelbergstr. 70, Basel CH-4056, Switzerland. E-mail: olga.mayans{at}unibas.ch
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: kinase • coiled-coil • protein assembly • protein structure • X-ray crystallography • multiangle static light scattering • small angle X-ray scattering
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
, Rho-associated kinases I and II (ROCKI/ II, also known as ß and 
, respectively) (3
, 4)
, C. elegans LET-502 (5)
, Drosophila Genghis Khan kinase (Gek) (6)
, murine citron kinase (7)
, and rat Cdc42 binding kinase (MRCK) (8)
, among others. Most of these kinases act as effectors of small Rho GTPases (Rho, Rac, and Cdc42 subclasses) known to regulate cytoskeletal dynamics. Accordingly, several members of the family are functionally linked to cytoskeletal reorganization events, as illustrated by the involvement of Gek in actin polymerization (6)
and the role of citron kinase, LET-502, and ROCKII in regulating the cleavage furrow contraction during cytokinesis (9
10
11)
.
DMPK is regarded as the defining member of this kinase family. It is most closely related to MRCK and ROCKs, with which it shares sequence similarity across the N-terminal catalytic fraction but not across CC moieties (Fig. 1
). By analogy to other family members, DMPK is suspected to regulate cytoskeletal events (12
, 13)
and may act as a downstream effector in the regulatory pathways of RhoA (12)
and Rac1 (14)
. Other reports, however, defend that DMPK intervenes in the modulation of the electrical excitability of the plasma membrane (15
, 16)
. Taken together, functional data suggest that DMPK might be related to muscle stiffness and hyperexcitability, heart conduction defects, and cognitive deficit in a progressive neuromuscular disorder known as myotonic dystrophy (DM1), which is the most common form of muscular dystrophy in adults (17)
. DM1 patients suffer from DMPK insufficiency due to the retention of mRNA transcripts in the nucleus (18
, 19)
. However, it is not currently understood how low levels of this kinase affect cellular function, and thus its involvement in disease remains polemic.
|
Oligomerization via CC segments is thought to be central to the regulation of DMPK-related kinases. CC motifs consist of two to five amphipathic -helices that wind around each other to form, typically, a left-handed supercoil. They mediate self-assembly events in proteins but can also be implicated in refolding processes and molecular recognition (for a recent review, see ref 20
). CC sequences are characterized by a repeat pattern of seven amino acids denoted a to g. Positions a and d classically harbor hydrophobic residues, which constitute the structural core of the motif. Positions e and g often host charged groups that form networks of intra- and interhelical salt bridges crucial to fold stability and interchain recognition. CC domains from DMPK-related kinases are believed to form dimeric assemblies and thereby to mediate dimerization in these enzymes. To date, only two closely related CC fragments from this family have been characterized. These correspond to the C-terminal, Rho binding motifs of ROCKII-CC and ROCKI-CC that share 52% sequence identity (Fig. 1)
. For these, crystal structures have been reported in free form (21)
and complexed to RhoA (22)
, respectively, confirming that both coils form parallel dimers. In a similar way, only a few analytical studies have reported on the oligomerization state of full-length kinases. Currently, data are available only for DMPK (23)
and ROCKs (24
, 25)
, which have been shown to adopt dimeric states. Despite the similarity of these kinases, different mechanisms have been proposed for their dimerization. While data on DMPK defended a self-assembly driven by its CC motif (23)
, the reports on ROCKs prove that CC moieties are not required for their oligomerization. The latter agrees with a biochemical study of MRCK, which also reported self-association in the absence of CC fractions (26)
. The associative interactions in ROCKs are the best understood so far due to the recent availability of the crystal structure from the dimeric N-terminal fraction (i.e., lacking CC segments) from ROCKI (25)
. This shows how head-to-head dimers interact via a small, N-terminal domain comprising the -helical LR region (Fig. 1)
as well as a tail, C-terminal to the kinase domain, which wraps around the catalytic core to reach the subunit interface. Given that dimerization segments show a certain homology across DMPK, ROCKs, and MRCK (Fig. 1)
, it would be reasonable to expect that related mechanism of assembly could be supported by these kinases, even if conflicting data have been reported (23)
.
Moreover, different biochemical roles have been attributed to the CC moieties of these kinases. The CC domains of ROCKs and citron kinase act as intrasteric inhibitors of their respective kinase domains and house at their C termini binding motifs for small Rho GTPases, where binding of the latter releases the inhibition (7
, 27)
. CC segments of MRCK (CC2/CC3) also act as intrasteric inhibitors but do not host a GTPase binding region (26)
, and physiological release factors are yet to be identified. On the contrary, the CC fraction of DMPK (DMPK-CC) has been proposed to act as an enhancer of its kinase activity (23)
, although it is unknown whether this involves transphosphorylation events as those observed in tyrosine kinases. In summary, both the functionality of CC domains from this family and the extent to which they govern molecular assembly remain unclear. Given that the CC from DMPK appears to differ from those of ROCKs and MRCK both in its regulatory properties and in its role during assembly (23)
, we have investigated the architectural details of DMPK to assess whether a common association principle exists in this family of kinases.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. DMPK
VR was cloned into the EcoRI and SalI sites of the insect cell expression vector pFastBac1 (Invitrogen, San Diego, CA, USA). A His6-tag and a TEV (tobacco etch virus) protease cleavage site were inserted N-terminal to the target construct. DMPKCC/VR was cloned into pET-15b (Novagen, Madison, WI, USA) using primers 5'-GGAATTCCATATGTCAGCCGAGGTGCG-3' and 5'-CGGGATCCTTACCCCGTGACAGCTGTGG-3' with NdeI and BamHI restriction sites, respectively. This vector adds a His6-tag and a thrombin cleavage site N-terminal to the target protein. Cloning of DMPK-CC has been previously reported (28)
. DMPK-CCN+ was cloned into the pETM-11 vector, a modified version of pET-24d (Novagen) that includes a His6-tag plus a TEV protease cleavage site N-terminal to the expression construct. This used primers 5'-CGTTCGAAGACTACATGTTGGTGGAGGACGGGCTC-3' and 5'-GGGGTACCTTACCCCGTGACAGCTGTGG-3' carrying BbsI (NcoI-compatible) and KpnI restriction sites.
Protein production
Sf9 cells were transfected with the bacmic cDNA construct of DMPKVR according to the manufacturer’s instructions (Life Technologies Bac-to-Bac system). The transfected cells were incubated for 3 days at 27°C and the supernatant, containing recombinant virus for protein expression, was collected. Amplified virus (10 ml) was added to 1 x 106 Sf9 cells and incubated at 27°C for 3 days. Lysis was performed by incubation on ice for 30 min in the presence of DNase and protease inhibitors (Complete EDTA-free; Roche, Nutley, NJ, USA) in 50 mM Tris pH 8.0, 300 mM NaCl, 5 mM ß-mercaptoethanol, 2% glycerol. A clarified supernatant was obtained by centrifugation at 20,000 rpm and 4°C for 1 h. The protein product was purified by affinity chromatography on a His-Trap HP column (GE Healthcare) equilibrated in lysis media. The bound protein was eluted by supplementing the medium with 200 mM imidazole. Subsequent purification used size exclusion chromatography on a Superdex 200 Tricorn 10/30 GL column (GE Healthcare) in 50 mM Tris pH 8.0, 150 mM NaCl, 5 mM ß-mercaptoethanol, 2% glycerol.
DMPKCC/VR was expressed in E. coli Rosetta (DE3) (Novagen) grown in LB medium supplemented with 34 µg/ml of chloramphenicol and 100 µg/ml ampicillin. Induction used 1 mM IPTG and incubation was overnight at 25°C. Cells were harvested by centrifugation at 8000 g at 4°C for 15 min. Lysis was by French pressing in the presence of DNase and protease inhibitors in 50 mM Tris pH 8.0, 50 mM NaCl, 5 mM ß-mercaptoethanol. The clarified supernatant, obtained by centrifugation at 20,000 rpm (4°C), was purified by affinity chromatography and subsequent gel filtration (as described for DMPKVR) in lysis media. Expression and purification of DMPK-CC were as reported (28)
. Production of DMPK-CCN+ used protocols as for DMPK-CC, except that all media were supplemented with 5 mM ß-mercaptoethanol. All measurements used fresh preparations and sample identification was by peptide-digest mass spectrometry.
Structure determination
Crystallization, data collection, data statistics, and experimental phasing of DMPK-CC have been described (28)
. Briefly, crystals grew from 100 mM Tris-HCl pH 7.5, 2.0 M Na+/K+ tartrate and 5–10 mg/ml protein solutions. X-ray diffraction data collected at beamline ID14–2 (ESRF, Grenoble) revealed the crystals to belong to space group P212121 with unit cell dimensions of a = 39.1 Å, b = 46.2 Å, c = 143.5 Å, and three polypeptide chains per asymmetric unit. Experimental phasing used 3
MAD on an Se-Met derivative. Phase refinement was in SHARP (29)
and automatic model improvement in ARP/warp v6.1 (30)
. Manual model building used the software O (31)
and refinement was in REFMAC (32)
. For cross-validation, crystallographic data were divided into a working and a test set using FREERFLAG (33)
. Restraints for non-crystallographic symmetry were applied during the initial phase of refinement and released at later stages. Geometrical parameters of intermediate and final models were assessed using PROCHECK (34)
. The final model lacks several N- and C-terminal residues that could not be resolved in electron density maps, namely 461–465/530–537 in chain A, 461–463/528–537 in chain B and 461–465/527–537 in chain C. This amounts to 
20% of disorder in the structure and explains the slightly elevated R-factor values, which are somewhat above those characteristic of this resolution. Coordinates have been deposited at the Protein Data Bank with accession number 1WT6.
Analytical ultracentrifugation
Sedimentation velocity and equilibrium data were recorded using an Optima XL-A analytical ultracentrifuge (Beckman Instruments, Fullerton, CA, USA) equipped with 4 and 12-mm Epon double-sector cells in an An-60 Ti rotor and absorption optics. Runs were performed at 20°C in 1) 50 mM Tris pH 7.5, 50 mM NaCl and 2) 100 mM sodium phosphate pH 7.0 using detection wavelengths of 277 nm and 234 nm. Sedimentation velocity experiments were carried out at 54,000 rpm and 0.65 mg/ml sample concentration. Sedimentation equilibrium measurements were performed at 19,000, 21,000, 24,000, and 26,000 rpm, and protein concentrations of 0.16–3.3 mg/ml for DMPK-CC and 0.09–1.4 mg/ml for DMPK-CCN+, where optical density values remained in a linear range. Average molecular masses were evaluated using SEGAL (G. Machaidze and A. Lustig, unpublished software), which adjusts the baseline absorbance to obtain the best linear fit of ln(absorbance) vs. radial distance square. A protein partial specific volume of 0.73 ml/g was used, while solution density was taken as 1.003 g/ml and viscosity as 1.02 centipoise.
Circular dichroism spectroscopy
Thermal denaturation of DMPK-CC was monitored at 222 nM on a 62A DS circular dichroism spectropolarimeter (AVIV) equipped with a temperature-controlled quartz cell of 2 mm optical path over a temperature range of 5–90°C. The scan speed was 0.5°C/min with a spectral bandwidth of 2.0 nm.
Binding assays
The binding of DMPK-CC to DMPKCC/VR was assayed in 50 mM Tris-HCl pH 7.5, 100 mM NaCl by 1) affinity pull-down assays using His6-tagged DMPKCC/VR immobilized in a His-Trap HP column (GE Healthcare) vs. a flow of untagged DMPK-CC; 2) simultaneous coexpression of both constructs followed by attempts of copurification via metal affinity plus size exclusion chromatography (protocols as above).
Size exclusion chromatography combined with multiangle light scattering (SEC-MALS)
The oligomeric state of DMPKVR and DMPKCC/VR in solution was determined via SECMALS measurements performed on an ÅKTA explorer 10 system (GE Healthcare) connected to a triangle light scattering detector and a differential refractometer (miniDAWN Tristar and Optilab, respectively; Wyatt Technology, Santa Barbara, CA, USA). A Superdex 200 HR 10/30 column (GE Healthcare) was used in 50 mM Tris-HCl pH 8.0, 50 mM NaCl at a flow rate of 0.5 ml/min. Sample volumes of 50 µl (DMPKVR) and 100 µl (DMPKCC/VR) were injected at a concentration of 2 mg/ml. A specific refractive index increment (dn/dc) value of 0.185 ml/g was used. The data were recorded and processed using ASTRA software (Wyatt Technology). To determine the detector delay volumes and normalization coefficients for the MALS detector, a BSA sample (Sigma A-8531) was used as reference. Neither despiking nor band broadening correction was applied.
Small angle X-ray scattering (SAXS)
Synchrotron radiation SAXS data were collected on the X33 camera (35)
of EMBL (DESY, Hamburg). Solutions of DMPKCC/VR were measured at protein concentrations of 1.4, 3.8, 5.7, 8.5, and 9.3 mg/ml in 50 mM TrisHCl pH 8.0, 50 mM NaCl, 2.5 mM ß-mercaptoethanol at 4°C. Measurements were recorded using a MAR345 IP detector at a sample detector distance of 2.7 m and wavelength = 1.5 Å in the range of momentum transfer 0.011 < s < 0.45 Å–1. Exposure time for a single measurement was 5 min. To check for radiation damage, two consecutive 3 min exposures at the highest protein concentration were compared and no changes were observed. The data were normalized, the scattering of the buffer subtracted, and the difference curves extrapolated to zero solute concentration using the program package PRIMUS (36)
. The maximum particle dimension Dmax, forward scattering I (0) and the radius of gyration Rg were evaluated using GNOM (37)
. The molecular mass (MM) of the solute was calculated by comparing its forward scattering with that of a reference solution of BSA (MM=66 kDa). The volume of the hydrated particle was computed using the Porod invariant equation (38). The low resolution shape of the molecule was computed ab initio using the program DAMMIN (39)
, which uses an assembly of densely packed beads to represent the protein molecule. Simulated annealing was used to build a compact, interconnected configuration of beads inside a sphere with the diameter Dmax that fits the experimental data Iexp(s) to minimize the discrepancy:
 |
(sj) are the calculated intensity and the experimental error at the momentum transfer sj, respectively.
The scattering from the crystallographic model of ROCKI (protein fraction of PDB entry 2ETR; 25
) was calculated using CRYSOL (40)
. The model of dimeric DMPKCC/VR was constructed based on the structure of ROCKI using the program BUNCH (41)
. The dimerization domains consisting of the 70 N-terminal and 20 C-terminal residues of ROCKI remained fixed during the calculations, so that the subunit interface stayed unaltered. The kinase domain was allowed to move as a rigid body while the missing C-terminal tail in DMPKCC/VR was represented as an interconnected chain of dummy residues (for domain definition, see
Fig 6a
) (42)
. A simulated annealing protocol was used to position the kinase domain and find the configuration of the C-terminal fragment that could fit the experimental scattering while obeying the P2 symmetry of the dimer.
|
|
|
|
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
) has been elucidated by X-ray crystallography up to 1.6 Å resolution. X-ray data statistics and model refinement parameters are given in Table 1
. Crystals contain three polypeptide chains in the asymmetric unit, which form a consecutive coil of 86.5 Å length, spanning residues 466–527 and comprising eight complete heptad repeats. They build a parallel, left-handed trimeric motif (Fig 2
a).
|
The architecture of DMPK-CC is restrained at its terminal sections while a central region exhibits moderate structural fluctuations as indicated by crystallographic temperature factors and interchain conformational agreement (Fig 2a, b
). The relative structural slackness at the middle point of the domain correlates with a widening of the coil at that position (Fig 2c
), which is likely to be caused by the presence of a sequence motif, 493-IRTD-496, with low propensity for CC formation (estimated using MultiCoil; 43). Also noticeable is a residue located upstream, R483, which contributes a charged group in core position a with a reduced Crick-phase (–2.8° instead of the commonly encountered value of –40°, as calculated using Twister; 44). This leads to overtwisting of the subsequent coil portion (Fig 2c
) and an associated suboptimal hydrophobic core packing in that region. Thus, central coil widening and pitch parameters appear not to simply reflect crystal lattice restraints but to be an intrinsic structural feature of this CC domain.
Residue arrangements around the 3-fold axis of DMPK-CC are mostly symmetrical, an exception are the side chains of F500, an aromatic residue in position d. The presence of aromatic groups in core positions is extremely rare in dimeric CCs, but it is observed in trimeric motifs (45)
. For example, the trimeric coil of lung surfactant protein D (46)
contains F225 and Y228 in a and d positions, respectively, which exhibit similar asymmetrical arrangements of their side chains. This is likely to reflect the difficulty of packing bulky groups in the limited interior of a trimeric coil. Remarkably, DMPK-CC also houses several solvent molecules in its core. In particular, its C-terminal section comprises an internal cavity filled by solvent that stabilizes a polar residue, N511, in a core position. A similar arrangement has been found in other trimeric CCs, such as bacteriophage T4 fibritin (47)
and a domain of a coat protein from MoMuLVirus (48)
. Such buried polar groups in CCs are thought to contribute to the specificity of the motif and to maintain chain registry (49)
. Other relevant interactions in DMPK-CC (Fig. 3
) include 6 interchain and 2 intrachain salt bridges. Interchain bridges help defining the association state of CCs, while intrachain bridges are important for helix stability (50)
. The interchain bridges of DMPK-CC belong both to the classical types eg’RE and eg’ER and to unusual core contacts of type ad’RE (notation as in ref 40
; where ’ designate residues in a subsequent heptad repeat). Intrachain salt bridges in DMPK-CC are of type bfRE. It is striking that the sequence of DMPK-CC is rich in charged residues (16E, 9R, vs. 2D, 0K per chain), although less than half of these are involved in interactions (8E, 4R, 0D). Thus, availability of reactive groups in the surface of this domain is high (predominantly of a negatively charged character and toward the NH2 terminus of the domain), so that it could support electrostatic binding to other molecular components.
Characterization of DMPK-CC in solution
In agreement with crystallographic data, the association state of DMPK-CC in solution is trimeric. One single oligomeric species could be detected using native-PAGE, size exclusion chromatography (Fig 4
a), and a sedimentation velocity study, which produced a single migrating boundary corresponding to a S20W of 2,1. Thus, the sample population in solution appeared to be highly homogeneous, with no dynamic states of assembly being observed. Sedimentation equilibrium analysis of DMPK-CC yielded an average molecular mass (MM) of 33 kDa over the entire protein concentration range assayed (0.16–3.3 mg/ml), consistent with the single presence of trimers in the solutions (theoretical MM for a trimer of this construct 34.9 kDa; Fig 4b
). Since the sample remained fully assembled at the lowest concentration assayed, the Kd of the interaction must be lower than 4 µM.
These studies carried out at moderate ionic strength (50–100 mM NaCl) and limited protein concentration are in good agreement with crystallographic data obtained from high salt solutions (2.0 M Na+/K+ tartrate) or polyethylene glycol media and concentrated protein samples (5–10 mg/ml), both at pH 4.6 and 7.5 (28)
. This suggests that the trimeric formations of DMPK-CC are stable in a broad range of media conditions. To gain further insight into the stability of this domain, it was investigated by thermal denaturation monitored by circular dichroism. A monophasic melting curve was recorded characteristic of a cooperative helix-coil transition of the three chains as a single domain, with a T1/2m of 46°C (Fig 4c
). This measurement was carried out in 100 mM sodium phosphate buffer pH 7.0, 100 mM NaCl at a protein concentration of 0.6 mg/ml. To account for specific buffer effects, measurements were repeated in 50 mM bis-Tris propane pH 7.5, 100 mM NaCl with equivalent results. This indicates a strong association of the component chains and confirms the robustness of the assembly. Therefore, it can be concluded that DMPK-CC has a strong tendency to self-associate into well-defined trimeric formations.
Oligomeric state of DMPK variants containing and lacking the CC domain
The assembly level of a DMPK construct comprising the LR sequence, the kinase domain, and the CC fraction, DMPKVR (Fig. 1)
, was investigated using multiangle light scattering coupled to size exclusion chromatography (SEC-MALS). This construct closely resembles an expressed DMPK spliceoform and is a suitable representative of this kinase in the cell. The study yielded MMs of 122'7 kDa and 124'5 kDa for protein concentrations estimated using UV-A280 and refractive index, respectively. These values are in excellent agreement with the calculated MM of 120'2 kDa for a dimer of this sample (Fig. 5
). This result is consistent with previous sedimentation equilibrium data on a comparable construct of this kinase (23)
.
To investigate the contribution of DMPK-CC to the dimerization of this kinase, a truncated variant lacking the CC fraction, DMPKCC/VR (Fig. 1)
, was examined. Data from size exclusion chromatography and native PAGE indicated the presence of a single species of DMPKCC/VR in solution, and SEC-MALS measurements confirmed that this corresponded to dimeric formations (Fig. 5)
. The experimental MM estimated for this sample was 101'4 kDa and 107'4 kDa using UV-A280 and refractive index to estimate the protein concentration, respectively. This is in close agreement with the theoretical MM of 106'6 kDa for a dimer of this construct. Furthermore, samples remained monodisperse and data were consistent through injected protein concentrations of 0.3–2 mg/ml, indicating that results are concentration independent (data not shown). The assembly state of DMPKVR coincides with that of DMPKCC/VR but not DMPK-CC, showing that the association level of this kinase is not dictated by the oligomerization properties of its CC domain and that segments associated with the kinase domain bear this function instead.
Analysis of DMPKCC/VR dimers by small angle X-ray scattering
The kinase domain of DMPK is flanked by an N-terminal region of 65 residue length (LR) and a C-terminal tail containing 120-residues that links it to the CC domain (Fig. 1)
. Recent crystallographic data on ROCKI (25)
show that, in that kinase, sequences both N- and C-terminal to the kinase domain pack together to form a dimerization domain. It has been shown that the single presence of the -helical LR region is not sufficient to secure assembly in this case, so that the removal of the kinase C-terminal tail results in 80% monomeric species (24)
. Given that ROCKI and DMPK share homology in the mentioned regions (22% and 36% seq. id. in the LR domain and kinase tail, respectively; Fig 6
a), we investigated whether DMPK might follow the same architectural principles of ROCKI by analyzing the overall conformation of DMPKCC/VR dimers using SAXS.
The experimental SAXS pattern of DMPKCC/VR is displayed in Fig 6b
. The estimated MM, 95 ± 10 kDa, confirms that the protein is dimeric under the SAXS experimental conditions. This is further corroborated by the volume of the hydrated particle, 210 ± 10 x 103Å (3)
, noting that, for globular proteins, the hydrated volume in Å3 should be approximately twice the MM in daltons. The radius of gyration Rg and the maximum particle size Dmax were 39.9 ± 0.6 Å and 130 ± 10 Å, respectively, suggesting an elongated structure. This agrees with the low-resolution shape of DMPKCC/VR reconstructed ab initio, which fits neatly the experimental data with a discrepancy 
= 0.95 (Fig 6b
). A comparison of this with the crystal structure of ROCKI indicates that DMPKCC/VR has a similar overall conformation, also appearing to form head-to-head dimers (Fig 6c
). The low-resolution model of DMPKCC/VR, however, displayed extra electron density that was not satisfied by the ROCKI structure, as can be expected from the fact that each protomer of DMPKCC/VR has 65 additional residues at its C terminus (Fig 6a
). The missing residues account for 14% of the total scattering mass of this sample. They play a significant role in the scattering of this molecule, possibly due to their contribution to an anisometric molecular shape (see below), so that the patterns computed from ROCKI yielded a poor fit to the experimental data (=6.6; Fig 6b
). To construct a model of DMPKCC/VR, the missing C-terminal portion was generated for each monomer using beads and maintaining a 2-fold symmetry as described in Materials and Methods. Although during the calculation the kinase domain was allowed to drift and rotate as a rigid body, its position was practically unchanged with respect to ROCKI. Models of DMPKCC/VR so generated showed that the C-terminal portions of each protomer formed rather compact formations, suggesting the existence of a protein domain at this position. A typical model of DMPKCC/VR in solution provides a good fit to the experimental data (=1.75; Fig 6b
) and agrees well with the molecular shape calculated ab initio as revealed by a comparative superimposition (Fig 6c
). A deviation between models, however, shows that the volumes occupied by the C-terminal fraction are represented somewhat differently by the two modeling methods. It should be noted that the ab initio model was obtained without imposing any prior knowledge or symmetry restrictions, whereas the reconstructed model applies a 2-fold symmetry that may not be entirely adhered to by a potentially flexible C-terminal tail. Such flexibility could underlie localized deviations between otherwise consistent models and scattering patterns.
CC domains can be found in proteins in combination with other motifs that influence their association. The most prominent example is that of the trimeric CC of fibritin, which is followed by a C-terminal extension of 30 residue length that folds into a ß-hairpin (51)
. This intertwines with equivalent ß-hairpins from other subunits, being essential to achieve CC trimerization in that case. Given that SAXS models of DMPKCC/VR suggested the presence of a structural formation immediately prior to the CC fraction, we assayed whether this might act as an association motif that could modulate CC stoichiometry. For this, we constructed an N-terminally extended version of DMPK-CC, DMPK-CCN+ (Fig. 1)
, and analyzed its oligomeric state by sedimentation equilibrium (Fig 4b
). Results indicated that DMPK-CCN+ remained trimeric (estimated MM 56.1 kDa compared to a theoretical MM of 63.5 kDa for a trimer of this construct), suggesting that possible structural elements of the kinase-CC linker sequence do not influence the oligomerization of this coil.
DMPK-CC is not involved in domain interactions
The CC fractions of ROCKs and MRCK have been shown to interact directly with their respective kinase domains, thereby acting as intrasteric inhibitors of kinase activity (26
, 27)
. In ROCKs, the interaction involves the Rho binding site at the C terminus of its long CC domain, whereas in MRCK the binding has been mapped to a fragment spanning CC2/CC3, whose deletion resulted in about a 3-fold increase in activity. Given that DMPK-CC and MRCK-CC3 share certain sequence similarity (up to 31% identity according to isoforms) (Fig 2d
), we investigated the possible interaction of DMPK-CC with other domains in this kinase using pull-down assays as well as coexpression/copurification experiments on DMPK-CC and DMPKCC/VR. Binding between these species could not be detected by either procedure, which yielded independent original species. Hence, DMPK-CC appears to neither interact with the kinase domain nor form a hetero-coil against the -helical LR sequence. This indicates that a CC association/dissociation regulatory model such as that proposed for MRCK and ROCKs is unlikely to be applicable to DMPK and that the kinase and CC fractions component of DMPK are likely to be arranged as architecturally distinct fractions, resulting in a globular portion and a filamentous shaft.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 22)
. Contrary to the dimeric character of the latter fragments, DMPK-CC forms a trimeric motif that is characterized by rigid termini and a central widening of its coil. It is currently unknown whether this feature might reflect functional properties of this domain. As proven by sedimentation equilibrium data, the assembly level of DMPK-CC in solution coincides with that in the crystal. No other oligomeric state has been detected in the current study under a broad range of protein concentration and ionic strength conditions. Furthermore, thermal denaturation studies have shown this assembly to be stable. Thus, our data prove that the oligomeric states of CC domains from this kinase family are more diversified than previously expected and that more coils must be investigated before general conclusions on their self-association properties can be drawn.
Experimental results on DMPK-CC contrast markedly with sequence-based predictions, which suggests this coil forms dimeric assemblies. For example, the software MultiCoil (43)
estimates 82% probability of dimer vs. 10% trimer formation. The structural determinants of CC association remain poorly understood, and thus such predictions should be considered with caution. Only a few systematic analyses have addressed the basis of CC association experimentally, focusing on either the role of hydrophobic core residues (52
, 53)
or salt bridges (e.g., 54
, 55
). Consequently, cases are described where experimental data disagree with predictions (for example, see ref 46
). To understand the molecular basis of DMPK-CC trimerization, we analyzed its interchain contacts. The only factor that might be proposed at this time to favor formation of trimers over dimers is the presence of an aromatic residue, F500, at core position a, since this is rarely observed in dimeric coils (45)
. The contribution of the several salt bridges in DMPK-CC to its oligomerization cannot be evaluated based on current knowledge.
Given that CCs are established oligomerization motifs, these were initially thought to be indispensable for the molecular assembly of DMPK-related kinases. However, recent studies of ROCKs have shown that other domains mediate self-association in these kinases, namely, the N-terminal LR region and a tail C-terminal to the kinase domain (24
, 25)
. Similarly, in the current study, a truncated DMPK variant lacking the CC moiety formed dimeric assemblies according to SEC-MALS data. A low-resolution structural analysis of these dimers using SAXS indicated that DMPK and ROCKI share similar dimerization principles involving head-to-head protomer interactions, even though a certain number of residues involved in the dimerization of ROCKI are not conserved in DMPK (Fig 6a
). Our data on the truncated DMPK variant differ from those of a previous study (23)
, where a similar DMPK construct was argued to be monomeric according to sucrose density gradient sedimentation and pull-down assays. The dimeric form in our study (residues 1–460) contains 29 more residues at its C terminus than the construct previously reported. The truncation in that case might have disrupted a possible structural element C-terminal to the kinase domain, which has been insinuated by current SAXS data. More studies are required to identify the exact structural details of this segment of DMPK and to establish whether its integrity is essential for the assembly of this kinase. Contrary to that former study, we conclude that the CC domain of DMPK is not required for its self-association and that a common architectural design is likely to be shared by the closest members of this family: DMPK, MRCK, and ROCKs.
The fact that an interaction between DMPK-CC and other domains in DMPK could not be identified in the current study suggests it is unlikely that the assembly properties of this CC are altered by a possible embedding within the kinase core. Neither is the influence of flanking structural elements expected, given that an N-terminally extended CC construct retained the same oligomerization concentration. Thus, we speculate that the stoichiometry of this coil within the context of full DMPK must be dominated by the prior association of the N-terminal catalytic fraction into dimers. Prediction data confirm that the DMPK-CC sequence is geometrically compatible with a dimeric association, and it is known that CC oligomerization can be dynamically altered by local factors (20)
.
Oligomerization is crucial for the activation of numerous kinases. Often self-assembly is directed to ensure efficient autophosphorylation in trans by increasing the local concentration of catalytic domains (56)
. Given that DMPK-CC does not determine the self-association of this kinase and that, contrary to MRCK and ROCKs, it does not act as an intrasteric inhibitor but as an activator, further investigations are needed to clarify whether its function might relate instead to the achievement of an active conformation or even substrate docking. Moreover, given the heterogeneity in CC sequences and their diversified interaction patterns across DMPK-related kinases, studies are required to reveal the exact role of these moieties and to establish the principles of their regulatory mechanisms.
|
ACKNOWLEDGMENTS |
|---|
Received for publication December 3, 2005. Accepted for publication February 3, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Windows-PC based system for small-angle scattering data analysis. J. Appl. Crystallogr. 36,1277-1282[CrossRef] program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl. Crystallogr. 28,768-773[CrossRef]This article has been cited by other articles:
|
|
 |
|
  S. H. Choi, G. Czifra, N. Kedei, N. E. Lewin, J. Lazar, Y. Pu, V. E. Marquez, and P. M. Blumberg Characterization of the Interaction of Phorbol Esters with the C1 Domain of MRCK (Myotonic Dystrophy Kinase-related Cdc42 Binding Kinase) {alpha}/{beta} J. Biol. Chem., April 18, 2008; 283(16): 10543 - 10549. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||
