Structural Analysis of the Regulation of the DYNLL/LC8 Binding to Nek9 by Phosphorylation*

Background: Phosphorylation regulates binding of the mitotic kinase Nek9 to LC8, controlling signal transduction through the Nek9/Nek6/7 module. Results: Crystal structures reveal the basis for the reduced interaction of LC8 for Nek9 upon phosphorylation on Nek9(Ser944). Conclusion: The reduced binding affinity of Nek9 for LC8 is due to diminished enthalpic interactions. Significance: Disclosing phosphorylation as a novel mechanism that directly regulates LC8 protein-protein interactions. The NIMA family protein kinases Nek9/Nercc1, Nek6, and Nek7 constitute a signaling module activated in early mitosis involved in the control of spindle organization. DYNLL/LC8 (dynein light chain 8) was originally described as a component of the dynein complex, but the recent discovery of multiple interaction partners for LC8 has suggested that it has a general role as a dimerization hub that organizes different protein partners. Recent experiments suggested that LC8 binding to Nek9 was regulated by Nek9 autophosphorylation on Ser944, a residue immediately located N-terminal to the LC8 conserved (K/R)xTQT binding motif, and that this was crucial for the control of signal transduction through the Nek/Nek6/7 module. In the present work, we present two crystal structures of LC8 with a peptide corresponding to the Nek9 binding region with and without a phosphorylation on Ser944. Structural analysis of LC8 with both Nek9 peptides, together with different biophysical experiments, explains the observed diminished binding affinity of Nek9 to LC8 upon phosphorylation on Ser944 within the Nek9 sequence, thus shedding light into a novel phosphorylation regulatory mechanism that interferes with LC8 protein·protein complex formation.

The NIMA family protein kinases Nek9/Nercc1, Nek6, and Nek7 constitute a signaling module activated in early mitosis involved in the control of spindle organization. DYNLL/LC8 (dynein light chain 8) was originally described as a component of the dynein complex, but the recent discovery of multiple interaction partners for LC8 has suggested that it has a general role as a dimerization hub that organizes different protein partners. Recent experiments suggested that LC8 binding to Nek9 was regulated by Nek9 autophosphorylation on Ser 944 , a residue immediately located N-terminal to the LC8 conserved (K/R)x-TQT binding motif, and that this was crucial for the control of signal transduction through the Nek/Nek6/7 module. In the present work, we present two crystal structures of LC8 with a peptide corresponding to the Nek9 binding region with and without a phosphorylation on Ser 944 . Structural analysis of LC8 with both Nek9 peptides, together with different biophysical experiments, explains the observed diminished binding affinity of Nek9 to LC8 upon phosphorylation on Ser 944 within the Nek9 sequence, thus shedding light into a novel phosphorylation regulatory mechanism that interferes with LC8 protein⅐protein complex formation.
Nek9 (also known as Nercc1) belongs to the mammalian NIMA family of protein kinases, named after the Aspergillus nidulans NIMA kinase (1). Mammals contain 11 members of the NIMA family or Neks, some of which are modular proteins with different protein domains, whereas some others only contain the kinase domain that has a sequence identity of ϳ40% for all members of the family (2). Mammalian Nek members seem to be specialized in functions related to the control of the microtubule and cilliary machineries (3). Nek6 and Nek7, together with Nek9, constitute a signaling module activated early in mitosis and involved in the regulation of the mitotic spindle (4). Nek9 (also called Nercc1) is a 120-kDa modular protein composed of an N-terminal kinase domain, with a strong sequence identity to the other members of the Nek family, followed by a domain homologous to RCC1 (the exchange factor for the small G protein Ran) and a C-terminal domain that contains a coiled-coil motif involved in dimerization (1). Nek9 is inactive during interphase and is activated at centrosomes and spindle poles during mitosis through a two-step mechanism involving CDK1 and Plk1 (1,5,6). In mitosis, Nek9 specifically interacts with the highly similar Nek6/7 (2,7) and is able to directly phosphorylate and activate these two kinases (8). Thus, Nek9, together with Nek6 and Nek7, form a phosphorylation signaling module that is activated in mitosis and has been shown to be essential for correct mitotic progression, controlling spindle formation and different aspects of the centrosomal cycle during early mitosis (1, 4, 7, 9 -11).
A recent report using two-hybrid experiments uncovered DYNLL/LC8 as a protein partner of Nek9 (12). LC8 was shown to interact with the C-terminal tail of Nek9, immediately after the coiled-coil region, in a conserved consensus motif (12). DYNLL/LC8 is a highly conserved ubiquitous eukaryotic protein with many protein partners involved in a great variety of cellular functions (13). Partners of LC8 include, among other proteins, dynein, myosin V, neuronal nitric oxide synthase, the proapoptotic protein BimL, and transcription factors Swallow and Trps1 (14 -20). Although DYNLL/LC8 was initially proposed to be a cargo adapter for the molecular motor dynein and myosin V, the existence of multiple protein partners revealed LC8 as a regulatory hub protein that interacts with a linear consensus binding motif normally located in intrinsically disordered protein regions (21).
Binding of LC8 normally promotes dimerization and a gain of structure of the protein partner, in some examples by facilitating the formation of coiled-coil structures immediately N-terminal to the LC8 binding region.
LC8 is a homodimeric structure that contains two hydrophobic binging grooves for linear peptides located in opposite sites of the central ␤-sheet interface (22). As shown by crystal structures of LC8 complexes, two peptides can interact simultaneously with the two hydrophobic grooves forming each one an extra ␤-strand to the central ␤-sheet of LC8 homodimer (23)(24)(25). Structural and functional data indicate that the homodimer structure of LC8 is essential for the interaction with protein partners, which are normally formed by dimeric structures (26,27). Signaling inputs such as LC8 phosphorylation, pH variations, or the cellular redox state have been shown to regulate LC8 function by controlling its dimerization (28 -30). Interestingly, it has been recently shown that phosphorylation of Ser 88 in LC8, which promotes the disruption of the LC8 homodimer, can regulate the binding of protein partners such as dynein intermediate chain (27,31).
Two different protein sequence motifs in target proteins can be recognized by LC8: (K/R)XTQT and G(I/V)QVD (where X is any amino acid) (32,33). In addition to these short sequences, the binding region of the peptide can be extended at the N terminus, thus completing a total of nine residues forming the extra ␤-strand of the interface. Only a few LC8 partners contain non-canonical binding motifs lacking the most conserved Gln residue (25). The binding affinity of LC8 to monomeric peptides is moderately weak (K d between 0.1 and 40 M) (13); however, bivalent peptides linking two consensus motifs have been shown to increase significantly the binding affinity for LC8 (34), indicating the importance of the dimer-dimer interaction in the complex of LC8 with protein partners. Despite the strong conservation in the LC8 binding region, recent biophysical data indicates an inherent plasticity that allows the interaction with these different consensus motifs (24).
LC8 has been proposed to participate in the regulation of the Nek9/Nek6/Nek7 kinase signaling module as a negative regulator of Nek6/7 binding and activation (12). LC8 binding to the C-terminal region of Nek9 interferes with the interaction of Nek9 with Nek6 and Nek7, thus impeding Nek9 activation of Nek6/7. In vivo experiments suggest that autophosphorylation of Nek9 can regulate this process by regulating the binding of Nek9 to LC8. Ser 944 , which is immediately located N-terminal to the LC8 consensus motif, has been proposed to be phosphorylated upon Nek9 activation and directly promote the LC8 dissociation from Nek9 in vivo, although such direct effect of Ser 944 phosphorylation on LC8 binding has never been measured. The conservation in this position of a serine residue and the presence of other Ser/Thr residues in LC8 consensus motifs of many protein partners might indicate that phosphorylation could represent a general manner of regulation for the LC8 binding. To gain insights in this regulation, we have crystallized LC8 in complex with a Nek9 peptide corresponding to the LC8 binding region and with a similar phosphopeptide modified on the equivalent of Nek9(Ser 944 ). Comparison of both crystal structures sheds light into the diminished LC8 binding of Nek9 upon phosphorylation on Ser 944 . Additionally, in vitro LC8 binding experiments using different biophysical methods demonstrate that phosphorylation on Ser 944 reduces the affinity to form a complex with LC8, thus directly confirming that phosphorylation of the partner could be a general way to regulate interactions with LC8.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-The cDNAs encoding full-length LC8 and the Nek9-CC 5 constructs (Nek9, residues 893 to 974) were cloned into a pET28b vector with the fusion protein Smt3 tagged with six histidines at the N-terminal. Constructions were expressed in Escherichia coli BL21(DE3) for 4 h at 37°C after induction with 1 mM isopropyl 1-thio-␤-D-galactopyranoside, and recombinant proteins were purified by nickel-nitrilotriacetic acid agarose resin (Qiagen) and dialyzed against 20 mM Tris-HCl (pH 8.0), 250 mM NaCl, and 1 mM ␤-mercaptoethanol in the presence of SENP2 SUMO protease overnight at 4°C. After protease cleavage, LC8 and Nek9-CC were purified by ion exchange chromatography using a Resource Q (GE Healthcare) column and by gel filtration using a Superdex75 column (GE Healthcare).
Single point mutants of NEK9-CC, including Q948A and S944E, were constructed using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. All sequences were sequenced after generation. Nek9-CC mutants were produced and purified as the wild-type form.
Peptide Design-The Nek9 peptide contains the amino acids between the Nek9 positions 940 to 950 resulting in the sequence VGMHSKGTQTA. In case of the Nek9 P-peptide, with a phosphate group on Ser 944 , the resulting phosphopeptide contains the sequence VGMH(pS)KGTQTA. Both peptides were provided by the company Pepnome Limited.
Circular Dichroism-CD spectra were measured in a Jasco-715 spectropolarimeter thermostatted at 25°C. Spectra were recorded from 260 to 200 nm at 0.5-nm intervals, 1-nm bandwidth, and scan speed of 10 nm/min. Ten accumulations were done for each spectrum. LC8 concentration for each scan is 11.2 M buffered with 100 mM NaCl, 10 mM sodium phosphate, pH 7.5, and 1 mM ␤-mercaptoethanol. In each spectrum, we vary the Nek9 peptide or the Nek9 P-peptide concentration ranged between 0 and 26.6 M.
In the case of LC8 binding to Nek9-CC (Nek9 residues from 893 to 974) wild-type and point mutants, LC8 and Nek9-CC concentrations were 10 M and 6.66 M, respectively, buffered as mentioned above. Same concentrations were used for the spectra of the LC8⅐Nek9-CC complexes. Data were normalized to [] (molar ellipticity) using the formula shown in Equation 1, where [] means molar ellipticity, indicates measured ellipticity, M represents molarity of the sample, N e indicates the number of peptide bonds, and l indicates the cuvette path length in cm.

Determination of the Interaction Constants (K a ) by Circular
Dichroism-To determine the interaction constants between LC8 and the peptides, we measured the ellipticity at 220 nm in every spectrum. Because the interaction of LC8 with the peptides is characterized by moderate-to-high affinity and the concentrations are comparable with the expected dissociation constant, no approximation for the free ligand concentration can be made. The total concentration of ligand is expressed as follows: where [L] T and [P] T are the total concentration of the peptide and LC8, respectively, K a is the association constant, and [L] is the concentration of the free ligand. Solving this equation for the unknown [L] will allow calculating the concentration of the LC8⅐peptide complex, [PL], for each experimental point, from which the change in circular dichroism signal can be calculated in Equation 4, where A 0 is the circular dichroism signal at zero peptide concentration, and A accounts for the maximal change of the signal at high peptide concentration.
Thermal Denaturation of LC8 and LC8⅐Peptide Complexes-Thermal transition curves of LC8 and LC8⅐Nek9-peptide complexes were obtained in a spectrofluorometer (Cary Eclipse) at a heating rate of 0.2°C/min following changes of intrinsic fluorescence at 338 nm while excited at 280 nm. The LC8 concentration was 11.2 M in a buffer containing 100 mM NaCl, 10 mM sodium phosphate, pH 7.5, and 1 mM ␤-mercaptoethanol. To obtain the thermal denaturation curves for the LC8 in the presence of the two peptides, the experiments were performed in the presence of 11.2 M concentration of the Nek9 peptide and P-peptide.
Thermodynamic Parameters from Thermal Unfolding-The thermal transition curves were fitted to obtain the unfolding thermodynamic parameters using Equation 5, where m j and b j are the slope and the intercept with y axis in the pretransition region, m u and b u are the slope and the intercept with y axis in the post-transition region, ⌬H m is the unfolding enthalpy, T is the absolute temperature, T m is the temperature at the melting point, and R is the universal gas constant.
Thermodynamic Binding Profile by Isothermal Titration Calorimetry-Binding interaction between LC8 and both peptides was assessed by high-sensitivity isothermal titration microcalorimeter AutoITC200 (MicroCal, GE Healthcare). Protein and peptide samples were properly degassed and care-fully loaded into the cells to avoid bubble formation during stirring. Experiments were performed in 100 mM NaCl, 10 mM sodium phosphate, or Tris-HCl, pH 7.5, and 1 mM ␤-mercaptoethanol. Experiments were carried out titrating 20 M LC8 in the calorimetric cell with 200 -300 M peptide in the injection syringe. The heat evolved after each ligand injection was obtained from the integral of the calorimetric signal. The heat due to the binding reaction was obtained as the difference between the reaction heat and the corresponding heat of dilution, and the latter was estimated as a constant heat throughout the experiment and included as an adjustable parameter in the analysis. The association constant (K a ) and the enthalpy change (⌬H) were obtained through non-linear regression of experimental data to a model for a protein with a single binding site. Data were analyzed using software developed in our laboratory implemented in Origin (version 7, OriginLab). The dissociation constant (K d ), the free energy change (⌬G), and the entropic change (⌬S) were obtained from basic thermodynamic relationships. Experiments performed in two buffers with different ionization enthalpies (0.86 kcal/mol phosphate, 11.4 kcal/mol Tris) allowed eliminating the contribution of proton exchange processes to the binding interaction and determining bufferindependent thermodynamic interaction parameters (enthalpy and entropy) (35,36). Experiments performed at two temperatures, at 15 and 25°C, allowed estimating the change in heat capacity upon binding, which is a binding parameter associated with changes in solvent accessible (polar and non-polar) surface area and, therefore, provides fundamental information on the binding interaction and possible conformational changes coupled to the complex formation.
Crystallization and Data Collection-Crystals of DYNLL/ LC8 (dynein light chain 1) in complex with Nek9 peptide were obtained at 18°C by sitting drop vapor diffusion methods. The reservoir solution contained 35% PEG 1500, 0.1 M MIB buffer (malonate, imidazole, and boric acid 2:3:3), pH 4. Crystals of LC8 in complex with Nek9 P-peptide were obtained at 18°C by sitting drop vapor diffusion methods. The reservoir solution contained 30% PEG 1500, 0.1 M MIB buffer, pH 4.5. Single crystals appeared after 1 week from equal volumes of protein solution (12 mg/ml in 5 mM Tris, pH 8.0, 50 mM NaCl) and reservoir solution. All crystals were cryo-protected in reservoir buffer containing 15% glycerol and flash-frozen in liquid nitrogen prior to diffraction analysis. Diffraction data were recorded from cryo-cooled crystals (100 K) at the ALBA synchrotron in Barcelona (BL13-XALOC beamline). Data were integrated and merged using XDS (37) and scaled, reduced, and further analyzed using CCP4 (Table 1) (38).
Structure Determination and Refinement-The structure of LC8 in complex with Nek9 peptide and P-peptide were determined from the x-ray data by molecular replacement using a previous LC8 structure (Protein Data Bank code 2P1K) as a model. The initial electron density maps produced from molecular replacement programs were manually improved to build up complete models for LC8 and peptides using the program COOT (39). Model refinement was performed with Refmac (38) and Phenix (40). LC8 in complex with Nek9 P-peptide contained six LC8 molecules assembling three homodimers in the asymmetric unit, and the Ramachandran analysis shows 91.06% of residues (499 residues) are in preferred regions, 7.12% of residues (39) are in allowed regions, and 1.82% of residues (10) are in outlier regions. LC8 in complex with Nek9 peptide also contained six LC8 molecules assembled in three homodimers in the asymmetric unit, and the Ramachandran analysis shows 96.91% of residues (534 residues) are in preferred regions, 1.81% of residues (10) are in allowed regions, and 1.27% of residues (7) are in outlier regions. Refinement and data statistics are provided in Table 1. Structural representations were prepared with PyMOL (41).
Accession Codes-Coordinates and structure factors from the two structures were deposited in the Protein Data Bank with accession codes 3ZKE and 3ZKF.

Circular Dichroism Experiments on LC8
Binding to the Nek9 C-terminal Region-Dimerization of Nek9 has been proposed to occur by means of the predicted C-terminal coiled-coil region (residues 891-939), which is located immediately upstream of the LC8 consensus binding region (residues 940 -950). In addition to Nek9 dimerization through its C-terminal coiled-coil region, LC8 has been suggested to contribute to Nek9 oligomerization as it does for other proteins (12). To analyze in vitro the interaction between LC8 and Nek9, CD experiments were carried out with LC8 and a fragment of the C-ter-minal region of Nek9, which includes the consensus LC8 binding region and the predicted coiled-coil region (Nek9-CC, residues 893-939) (Fig. 1). The concentrations of LC8 and Nek9-CC used in the CD experiments were 10 M and 6.7 M, respectively. CD spectra of the individual proteins show an expected ␣-␤ profile for LC8, and interestingly, a rich ␣-helix profile for the Nek9-CC fragment, which could be indicative of the presence of a coiled-coil structure. This is supported by the higher ordered multimeric structures that appear during the gel filtration purification of the recombinant Nek9-CC protein (12).
Remarkably, differences can be observed after comparison of the CD spectra of a mixture of the two proteins, LC8 and Nek9-CC, and the sum of their individual spectra (Fig. 1C). These changes in the ellipticity values between 200 and 240 nm suggest variations in the composition of secondary structure elements; in particular, the increase in ellipticity ϳ222 nm may indicate changes in the ␤/␣ structure. These results are in agreement with the previously reported binding between Nek9 and LC8 (12), and with crystal structures of LC8⅐peptide complexes, that indicated the formation of an extra ␤-strand (22).
To analyze the binding between LC8 and Nek9, we produced two Nek9-CC point mutant constructs. Nek9-CC Q948A can destabilize the binding by perturbing the highly conserved Gln948 interaction with the LC8 dimer; and Nek9-CC S944E, a phosphomimetic that can destabilize the complex by mimicking phosphorylation on Ser 944 , as it has been recently shown in vivo (12). Individual CD spectra of both Nek9-CC point mutants show similar profiles as the Nek9-CC wild-type form, probably indicating the formation of a coiled-coil structure (Fig. 1B). However, using the same protein concentrations as the wild-type form, the CD spectra of the mixture of LC8 with both Nek9-CC point mutants show differences in the CD profile compared with Nek9-CC wild-type, in particular in the region ϳ222 nm. This reduction in the ellipticity at 222 nm could indicate a decrease in the affinity of LC8 for both Nek9 mutant constructs under these experimental conditions. Thus, the destabilization of the binding with LC8 can occur either by the mutation on the conserved Gln 947 residue, which has already been established to play an important role in LC8 binding, or by a mutation on Ser 944 , a site proposed to regulate LC8 binding to Nek9 upon phosphorylation.
Crystal Structures of LC8 in Complex with Nek9 Peptide and Nek9 P-peptide-To gain insight at the atomic level into the Nek9 binding to LC8, we synthesized two different peptides corresponding to the Nek9 binding consensus region, with one of them carrying a phosphoserine modification at position 5 corresponding to position Ser 944 in the Nek9 sequence, 940 VGMHSKGTQTA 950 . Both peptides displayed good solubility properties and could be dissolved in aqueous buffers at relatively high concentrations. LC8 was also produced at milligrams level amounts, as reported previously (12). Purification of recombinant LC8 by gel filtration chromatography indicated unequivocally the presence of a homodimer organization in the quaternary structure of the protein.
Crystals of LC8 with both Nek9 peptides were produced in a similar reservoir condition but with different crystallographic space groups. Orthogonal protein crystals of LC8 with the unmodified Nek9 peptide diffracted beyond 2.2 Å resolution, whereas hexagonal protein crystals of LC8 with Nek9 P-peptide (phosphopeptide) diffracted at 2.6 Å resolution (see Table 1 for details). In both crystal forms, the asymmetric unit contained three homodimers, with each LC8 molecule bound to a Nek9 peptide (a total number of 12 distinct molecules form the asymmetric unit). Both Nek9 peptides could be completely traced in the electron density maps in the three homodimers and only in some cases lacked the flexible terminal residue.
Both crystal forms display a similar LC8 homodimeric structure, which constitutes the biological unit and has been well established in the recent years by different structures, either in the apo form or in complex with peptides (13,(22)(23)(24)(25). Basically, the LC8 structure is composed by two monomers with a characteristic fold formed by five five-stranded antiparallel ␤-sheets at the center of the homodimer, swapping four strands from one monomer with a fifth strand from the other monomer ( Fig.  2A). Additionally, each LC8 monomer contains a pair of ␣-helices flanking the central ␤-sheet region at opposite faces of the homodimer. The peptide-binding region is located in identical opposite grooves at the two edges of the dimerization interface. As observed in other LC8 complex structures, the bound peptide interacts in an extended orientation and forms an extra ␤-strand with LC8, increasing the total number of strands of the central ␤-sheet. The parallel orientation of the two Nek9 peptides in the LC8 homodimer would be compatible with the presence of the coiled-coil structure immediately before the LC8 binding region ( Fig. 2A).
As observed in previous LC8⅐peptide complexes, the Nek9 peptide and Nek9 P-peptide interact extensively with the central ␤-sheet of the LC8 homodimer (Figs. 2 and 3), establishing several backbone hydrogen bonds and side chain interactions with the ␤3 strand of one of the LC8 subunit. The only interaction of the peptide with the opposite subunit of the LC8 homodimer is mainly established by Gln 947 from the highly conserved KXTQTX motif, with its side chain buried in a hydrophobic pocket formed by Ile 34 , Glu 35 , and Lys 36 from the ␣2 helix.
Most of the interactions are basically conserved between in both complex structures, although two hydrogen bonds are disrupted in the Nek9 P-peptide structure. In particular, the backbone hydrogen bond of the carbonyl group of Gly 941 is not formed in some of P-peptide molecules found in the asymmetric unit. A second disrupted interaction is the hydrogen bond between Ser 944 and Thr 67 (at 2.69 Å in the case of the Nek9 peptide structure), which is in all cases disrupted by the presence of the phosphate group at Ser 944 . The phosphate group of Ser 944 is located at the homodimer interface, in contact with the side chain of Tyr 65 , from the ␤3 strand, and next to the side chains of Lys 43 and Lys 44 from the opposite LC8 subunit (Fig. 2).
Interestingly, structural superposition of the six different peptides found in the asymmetric unit show that the Nek9 P-peptides display a high degree of flexibility when compared with the unmodified Nek9 peptides, which displays an almost identical superposition of peptides (Fig. 3). This increase of flexibility displayed by the Nek9 P-peptide is more prominent at the N-terminal part of the P-peptide (from Val 940 to His 943 ), compared with the KXTQTX C-terminal part (from Lys 945 to Thr 949 ). These results suggest that phosphorylation on Ser 944 , which is located four positions N-terminal to the central Gln 948 , affects the interaction of the Nek9 peptide to LC8 by producing a destabilization of the complex. Our detailed structural analysis supports the hypothesis that phosphorylation on Ser 944 can modulate and weaken the binding of LC8 to Nek9 in the cellular context, in which protein concentrations are much lower than in crystal structures and the phosphorylation outcome may be more evident.
Another remarkable difference found after comparison of both crystal structures is a shift produced in the LC8 homodimer structure upon binding to the Nek9 P-peptide, in particular in one of the LC8 monomers with respect to the other monomer (Fig. 3C). This shift might be attributed to the presence of a phosphate group at Ser 944 , in which its negative charge might somehow interfere in the dimer interface. It is possible that the lysine residues from the opposite subunit next to the phosphate group are affected by the FIGURE 2. Crystal structures of LC8 with Nek9 peptide and Nek9 P-peptide. a, schematic representation of the LC8⅐Nek9 peptide structure. ␣-Helices and ␤-strands of each LC8 subunit are labeled. The two Nek9 peptides binding the homodimer are depicted in red. b, surface representation of the LC8 homodimer (yellow and orange) in complex with the Nek9 peptide, which is labeled and shown in stick representation. c, surface representation of the LC8 homodimer (yellow and orange) in complex with the Nek9 P-peptide, which is labeled and shown in stick representation. d, detailed stick representation of the interaction between the N-terminal regions of the Nek9 peptide (gray) with LC8 (blue). e, detailed stick representation of the interaction between the N-terminal region of the Nek9 P-peptide (yellow) with LC8 (blue). The figure was prepared with PyMOL (43).
negative charge of the phosphate group. In contrast to the structural comparison between LC8⅐ neuronal nitric oxid synthase and LC8⅐Swa complexes (42) and despite the aforementioned LC8 subunit shift, we have not observed an enlargement of the peptidebinding cleft in the Nek9 P-peptide complex. This movement in the homodimer quaternary structure upon Ser 944 phosphorylation could be significant to perturb the binding and stability of the Nek9 peptide to LC8.
Circular Dichroism Experiments of the Binding of LC8 to Nek9 Peptides-Because the binding partners increase the ␤-sheet signal by adding an extra strand to the central five-stranded ␤-sheet of LC8, CD represents an appropriate method to calculate the binding affinity in titration experiments. The CD spectrum of LC8 shows a characteristic curve for a ␣/␤-rich protein, displaying a maximal negative ellipticity value ϳ220 nm (Figs. 1B and 4). Using fixed LC8 concentrations, titration experiments were conducted by measuring the CD spectra of increasing concentrations of the unmodified Nek9 peptide (Fig.  4A). The observed augment of the ellipticity ϳ220 nm, which signals changes in the ␣/␤-secondary structure, might indicate an increase of ␤-structure produced by the binding of the Nek9 peptide. CD spectra comparisons indicate a major increase of FIGURE 3. Comparison of the interaction of LC8 with Nek9 peptide and Nek9 P-peptide. a, stereo representation of the superposition of the six Nek9 peptides from the asymmetric unit. Nek9 peptide and interacting LC8 residues are labeled and shown in stick representation. b, stereo representation of the superposition of the six Nek9 P-peptides from the asymmetric unit. Nek9 P-peptide and interacting LC8 residues are labeled and shown in stick representation. c, stereo representation of the ribbon superposition of the LC8⅐Nek9 peptide and LC8⅐Nek9 P-peptide structures. Both Nek9 peptides are labeled and shown in stick representation. The two LC8 subunits of the homodimer are labeled monomers 1 and 2.
When identical CD titration experiments were carried out with the Nek9 P-peptide, a smaller increase in ␤-signal was observed, in contrast to the unmodified Nek9 peptide (Fig. 4A). The LC8 titration with the Nek9 P-peptide shows a smoother increase at 220 nm and, unlike the unmodified Nek9 peptide, the signal is not saturated ϳ1.9 M. The ␤-signal of the Nek9 P-peptide spectra seems to increase more gradually, probably indicating a reduced binding affinity of the Nek9 peptide upon phosphorylation.
To calculate the different binding capabilities between the two peptides, the change in ellipticity at 220 nm has been represented as a function of the increase of peptide concentration (phospho-and unmodified Nek9 peptide) (Fig. 4B). The two curves were analyzed considering that the concentrations of protein and peptide are comparable (see "Experimental Procedures"), and the dissociation constant (K d ) for each peptide was determined. Using this approach, we have estimated K d values ϳ 0.2 and 3.7 M for the unmodified and the Nek9 P-peptide, respectively, thus confirming our hypothesis that the unmodified peptide binds tightly to LC8. These K d values are in the range of reported LC8-peptide interactions (13) and suggest that phosphorylation of the Nek9 peptide results in a 18-fold reduction of binding to LC8, thus supporting conclusions drawn from crystal structures.
Thermal Denaturation of the LC8⅐peptide Complex-Protein stability of the LC8⅐peptide and LC8⅐P-peptide complexes was analyzed by changes of the intrinsic fluorescence during thermal denaturation. Based on our CD experiments, fixed concentrations of Nek9 peptides (11 M) were chosen to form a complex with LC8. At this concentration, CD analysis showed a partial binding of the Nek9 P-peptide to LC8 in comparison with the Nek9 peptide at 20°C. The transition curves of the heat-induced fluorescence emission changes of LC8 complexes are shown in Fig. 5. These curves include LC8 alone, FIGURE 4. Circular dichroism analysis of LC8 binding to Nek9 peptide and Nek9 P-peptide. a, right: circular dichroism spectra of the titration of LC8 with an increasing concentration of the Nek9 peptide. Left: circular dichroism spectra of the titration of LC8 with an increasing concentration of the Nek9 P-peptide. b, plot of the variation of the ellipticity values from the CD spectra of a and b at 220 nm, which follows the change of ␤-secondary structure. Circles, Nek9 peptide; squares, Nek9 P-peptide. and LC8⅐peptide and LC8⅐P-peptide complexes. Single cooperative transition curves were observed, and the data were fitted to a two-state temperature-induced protein-unfolding model (all curves with R Ͼ 0.999). As shown in Table 2, the transition temperatures of unfolding obtained for the three curves are quite similar (T m ϳ333 K), indicating minor changes produced in the context of a global denaturation. However, the different extracted thermodynamic enthalpies support the existence of significant differences for both Nek9 peptides (Table 2).
In the case of LC8 alone and LC8⅐Nek9 P-peptide complex, the enthalpies for the thermal denaturation unfolding experiments are comparable, ϳ156.8 and 155.8 kcal/mol for each, respectively. However, these similar values are in contrast with the higher enthalpy value (ϳ174.2 kcal/mol) shown for the LC8⅐Nek9 peptide complex. This higher enthalpy value could be an indication of a large number of disrupted bonds during the denaturation experiment, thus suggesting the formation of a tighter complex between LC8 and Nek9 peptide that would support our previous observations. Thermodynamic Profile of the LC8 Complex Formation-Thermodynamic binding parameters for LC8 in complex with the two Nek9 peptides were determined by isothermal titration calorimetry. Experiments performed in two buffers with different ionization enthalpies indicated that the binding of the peptides is coupled to a minor proton uptake process. (The number of exchanged protons, n H , is small and positive (Table 3).). Titration experiments of both Nek9 peptides on LC8 showed that the affinity of the Nek9 P-peptide for LC8 is reduced by ϳ6-fold in the dissociation constant (K d is 0.16 and 0.96 M, for both Nek9 peptide and P-peptide, respectively) (Fig. 6). This decrease of the K d in the Nek9 P-peptide is due to a reduction of the enthalpy in the complex with LC8 (Table 3), probably by a reduced number of bonds formed between LC8 and Nek9 peptide.
Thus, the decrease of the affinity for LC8 for the Nek9 P-peptide is basically caused by less favorable enthalpic interactions (⌬⌬H ϭ 1.7 kcal/mol) combined by a partial compensation from more favorable entropic interactions (ϪT⌬⌬S ϭ Ϫ0.6 kcal/mol) ( Table 3). These two factors produce a global reduction of the affinity, as observed in the difference of the binding Gibbs energy value (⌬⌬G ϭ 1.1 kcal/mol), producing the afore-    the Nek9 P-peptide, respectively, employing a well known structural parameterization of the binding energetics in proteins (⌬C P ϭ 0.45 ϫ ⌬SASA nonpolar Ϫ 0.26 ϫ ⌬SASA polar ) (43) and the crystallographic structures of the complexes and apoLC8. These values indicate that there are not substantial conformational changes in LC8 upon binding of the peptides.

DISCUSSION
Post-translational modification of proteins by phosphorylation is one the most, if not the most, important manner of regulating protein-protein interactions and many signaling pathways inside of the cell (44). DYNLL/LC8 has been recently postulated to be a dimerization hub that can control the function of other proteins by regulating their oligomerization state, normally by inducing higher oligomerization states (13,21). This might be the case for the protein kinase Nek9, which is involved in the control of early events of mitosis, and it has been shown to interact with LC8 through a canonical consensus region located in a partially disordered region of the C-terminal tail. In addition, LC8 negatively controls the binding of Nek9 to the related kinases Nek6/7 and thus their activation (12), suggesting that LC8 binding may have additional functions in the context of specific complexes besides regulating oligomerization. Thus, the regulation of LC8 binding might represent an important event to control the function of the target protein, in this case Nek9 and the formation of the mitotic spindle. In a recent work, we showed that Nek9 autophosphorylation on Ser 944 may impair the interaction with LC8 in vivo, thus supporting the view that phosphorylation of LC8 binding partners can have a major effect in regulating the LC8 interaction (12). A similar LC8-mediated regulation was described for the proapoptotic Bim, which is sequestered to the LC8⅐dynein motor complexes but can be released upon phosphorylation by JNK kinase on the LC8 binding consensus motif. In this case, phosphoryl-ation occurs on threonines adjacent to the central glutamine residue of the LC8 binding motif (45).
Examples describing the interaction of different partners to LC8 such as Dynein intermediate chain and Swallow (46,47), indicate that LC8 seems to facilitate the folding and increase the ␣-helical content by stabilization of coiled-coil structures in those proteins. This "chaperone-like activity" of LC8 normally occurs in proteins with coiled-coil regions next to the LC8 binding consensus motif; those regions are normally intrinsically or partially disordered. In those examples, binding of LC8 can thus facilitate the formation of coiled-coil structures and thus dimerization of protein targets. In the case of dynein, in which LC8 was first described to act as a cargo adapter, this chaperone-like activity effect of LC8, which indirectly affects the cargo binding, has also been described (46,48). In this case, the dynein intermediate chain is partially disordered and gains structure upon binding to LC8. In Nek9, our CD experiments show a high content of ␣-helical structure at the C-terminal region (Nek9-CC constructs, Fig. 1), probably indicating the presence of the predicted coiled-coil structure (1). In our LC8/ Nek9-CC binding experiments with two point mutants for the LC8 consensus-binding motif, we observed different CD spectra when compared with the wild-type Nek9-CC construct. It might be that LC8 binding to Nek9 produces changes in the stability of the predicted Nek9 coiled-coil, thus affecting the Nek9 oligomerization state, as suggested in previous experiments (12).
We have analyzed the LC8-Nek9 interaction by using different biophysical methods. CD is an optimal approach to determinate directly the interaction of the peptide to LC8 by the measurement of changes in the ␤-signal, which is the type of secondary structure formed upon binding of the peptide (22). The ϳ18-fold reduction in the K d extracted from the CD experiments with the Nek9 P-peptide is a direct indication of the perturbation of the interaction produced by the presence of the phosphate group. In the thermal denaturation unfolding of LC8 complexes, chased by intrinsic fluorescence, the major enthalpy of the complex between LC8 and the unmodified Nek9 peptide indicates a major energy released for the disruption of the complex in comparison with the Nek9 P-peptide. Finally, the thermodynamic values extracted from isothermal titration calorimetry experiments also show a 6-fold decrease of the dissociation constant (K d ) for the Nek9 P-peptide, due basically to less favorable enthalpic interactions, probably indicating a reduction in the number of bonds formed in the complex. All three methods point to a direct perturbation of the complex between LC8 and Nek9 upon phosphorylation on Ser 944 .
The crystal structures of LC8 with both Nek9 peptides shed light to the binding differences observed in the biophysical experiments. The interaction of both Nek9 peptides to LC8 is carried out by a similar ␤-like antiparallel strand, which starts seven N-terminal positions to Gln 948 , namely from Gly 941 to Thr 949 , forming a total number of eight backbone ␤-sheet hydrogen bonds between the Nek9 peptide and ␤3 strand of LC8 (see Fig. 2). However, comparisons between the six LC8 monomers in the asymmetric unit indicate that the phosphate negative charge on Ser 944 perturbs the interaction with LC8, displaying a high degree of flexibility that affects mainly the binding of the first half of the peptide, from positions Ϫ7 to Ϫ4 (Gly 941 to Ser 944 ). The position of Ser 944 in the LC8 binding motif of Nek9 is located four positions N-terminal to the central Gln 948 and varies from Asp, Ser, Asn and Thr in the consensus sequence analysis for LC8 binding partners (13). Interestingly, some of them are either residues susceptible of being phosphorylated or residues with an inherent negative charge.
Despite the lower affinity for the Nek9 P-peptide and the important binding differences described in vivo (12), the high protein concentration in crystals favors the formation of fairly similar LC8⅐Nek9 complex structures. However, such high concentration of proteins are difficult to occur in the cellular context, and the reported K d values strongly suggest that the differences between the unmodified and the Nek9 P-peptide are sufficient to perturb binding in vivo. The slight shift observed in the LC8 homodimer interface, the two disrupted hydrogen bonds in the Nek9 P-peptide complex and the higher flexibility of the Nek9 P-peptides produces a binding reduction on LC8 upon Ser 944 phosphorylation.
We thus provide for the first time in vitro experimental data directly indicating that phosphorylation can regulate the binding of LC8 to a particular protein partner, in this instance, Nek9. Regulation of protein-protein interaction by phosphorylation is one of the most common ways to regulate many cellular signaling pathways. In the case of the mitotic Nek9/Nek6/ Nek7 signaling module, LC8 seems to interfere in this pathway by a direct binding competition with the downstream kinases Nek6 and Nek7, regulation that can be reversed by direct phosphorylation on the interface between the two proteins. The fact that LC8 has multiple binding partners, that it is ubiquitously present, and that many of its protein partners contain Ser/Thr residues in the consensus binding motif, suggest that phosphorylation on this motif might be a general mechanism of controlling LC8 interaction in different cellular contexts.