The 8-kDa Dynein Light Chain Binds to Its Targets via a Conserved (K/R)XTQT Motif*

Cytoplasmic dynein is a large, multisubunit molecular motor that translocates cargoes toward the minus ends of microtubules. Proper functioning of the dynein motor requires precise assembly of its various subunits. Using purified recombinant proteins, we show that the highly conserved 8-kDa light chain (DLC8) binds to the intermediate chain of the dynein complex. The DLC8-binding region was mapped to a highly conserved 10-residue fragment (amino acid sequence SYSKETQTPL) C-terminal to the second alternative splicing site of dynein intermediate chain. Yeast two-hybrid screening using DLC8 as bait identified numerous additional DLC8-binding proteins. Biochemical and mutational analysis of selected DLC8-binding proteins revealed that DLC8 binds to a consensus sequence containing a (K/R)XTQT motif. The (K/R)XTQT motif interacts with the common target-accepting grooves of DLC8 dimer. The role of each conserved amino acid residue in this pentapeptide motif in supporting complex formation with DLC8 was systematically studied using site-directed mutagenesis.

The 8-kDa light chain (DLC8) of cytoplasmic dynein was originally identified as a light chain of Chlamydomonas outer arm axonemal dynein (5-7). The protein was subsequently shown to be a stoichiometric component of cytoplasmic dynein and of actin-based motor myosin V (8). Mutational studies showed that DLC8 is required for proper functioning of cytoplasmic dynein including retrograde intraflagellar transport (9), nuclear migration, and motor complex localization (10). However, it is not known how DLC8 is assembled into the cytoplasmic dynein motor complex, although it was suggested that the protein is associated with DIC in the axonemal dynein complex (6).
DLC8 contains 89 amino acid residues and is highly conserved throughout evolution (Ͼ90% amino acid sequence identity from Caenorhabditis elegans to humans). Other than binding to certain subunits of motor proteins, DLC8 has also been shown to interact with several cellular targets with diverse functions. DLC8 binds to neuronal nitric-oxide synthase (11,12), proapoptotic Bcl-2 family protein Bim (13), Drosophilia mRNA localization protein Swallow (14), transcriptional regulator IB (15), and postsynaptic scaffold protein GKAP (16). The DLC8-binding domain of nNOS was mapped to a short stretch of amino acid residues (12,17). The other known DLC8binding proteins neither contain homologous DLC8-interacting sequences found in nNOS nor bear obvious amino acid sequence similarities among each other. Elucidation of the molecular basis of the interactions between DLC8 and its diverse target proteins represents an important step in understanding the function of this versatile protein. Structural studies showed that DLC8 contains two identical target-binding grooves located at opposite faces of the protein dimer interface (18). DLC8 is capable of binding to short peptide fragments of ϳ10 amino acid residues from its targets. The target peptides bind to DLC8 in an antiparallel ␤-strand structure by pairing with the ␤-strand located at the base of each target-accepting groove (17,18).
We report here that DLC8 binds to a conserved (K/R)XTQT amino acid sequence motif in a wide variety of protein partners. Such a motif is present in DIC of cytoplasmic dynein and mediates association of DLC8 and DIC in the dynein complex.

MATERIALS AND METHODS
Yeast Two-hybrid Screen of DLC8-binding Proteins-Yeast-two hybrid screening was performed as described previously (16). Briefly, DLC8 was subcloned into pBHA (LexA fusion vector) and used to separately screen ϳ1 ϫ 10 6 clones each from human and rat cDNA libraries constructed in pGAD10 (GAL4 activation vector; CLONTECH).
Expression Constructs-The full-length mouse DIC gene was constructed by assembling three partially overlapping genomic clones using standard PCR and DNA ligation methods. The DIC fragments were generous gifts from Dr. L. C. Tsui (20). The full-length DIC coding DNA fragment was inserted into the EcoRI/BamHI sites of an in-house-modified pET32a plasmid vector (Novagen). Various truncation and point mutants of DICs were cloned into pGEX-4T-1 vector (Amersham Pharmacia Biotech) using standard recombinant DNA methods. The plasmids harboring the full-length DIC gene and various mutant DIC genes were transformed into Escherichia coli BL21(DE3) host cells for fusion protein production.
The C-terminal fragment of c82 (gene product of clone 82 identified from a yeast two-hybrid screening for DLC8-binding proteins; see Table  I) containing amino acid residues 119 -199 of the native protein was amplified by PCR using the coding strand primer 5Ј-CCGGAATTC-GAGAGATTGCAGGGTCTG-3Ј and noncoding strand primer 5Ј-CCGCTCGAGCTAGAGGCTGGGTCTAC-3Ј. The PCR-amplified c82 fragment was inserted into the EcoRI/XhoI sites of pGEX-4T-1. Various GST-fused c82 truncation mutants used in this study were generated using standard PCR techniques. The GST-fused BS69 fragment was prepared using a method similar to that described for c82, except that two primers specific for BS69 were used in the PCR amplification.
Bacterial expression vectors of GST-fused Bim L (long form) and Bim S (short form), both lacking the hydrophobic transmembrane region and the BH3 domain, were constructed by PCR using full-length Bim L and Bim S (gifts from Dr. A. Strasser) as templates, respectively, using a pair of primers: 5Ј-CGGGGATCCATGGCCAAGCAAC-3Ј (sense), and 5Ј-CCGCTCGAGTCACTCCTGTGCGATCC-3Ј (antisense).
Expression and Purification of the Fusion Proteins-To express the full-length DIC and its truncated mutants, host cells transformed with the DIC expression plasmids were grown in LB medium at 37°C until A 600 reached ϳ1.2. The expression of the proteins was induced by the addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of ϳ0.5 mM. Protein expression continued for ϳ3 h at 37°C, and the cells were then harvested by centrifugation and stored at Ϫ80°C prior to protein purification. The full-length DIC containing a hexahistidine tag at the N terminus was expressed in inclusion bodies. The pelleted bacterial cells were resuspended in 50 mM Tris-HCl buffer (pH 7.5) containing 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, and 1 g/ml antipain prior to cell lysis using sonication. The inclusion bodies were then washed extensively using 50 mM Tris-HCl buffer (pH 7.5) containing 1 M urea and 0.5% Triton X-100. The washed inclusion bodies were solubilized in 50 mM Tris-HCl buffer (pH 7.9) containing 5 mM imidazole, 0.5 M NaCl, and 6 M guanidine HCl at room temperature. The denatured protein was then passed through a Ni 2ϩnitrilotriacetic acid affinity column following the procedure described by the manufacturer (Novagen) for proteins under denaturing conditions. Refolding of full-length DIC was achieved by a single step dialysis of the denatured protein against 1ϫ phosphate-buffered saline buffer (pH 7.4). The refolded protein was recovered from the supernatant after centrifugation.
Various GST-fused DIC (as well as c82 and BS69) truncated and point mutants were expressed in soluble forms and purified using GSH-Sepharose affinity columns (Amersham Pharmacia Biotech) following the instructions of the manufacturer. The purified GST fusion proteins were dialyzed against 1ϫ phosphate-buffered saline buffer (pH 7.4) to remove GSH, and the protein samples were directly used for DLC8 binding assay experiments. To express GST⅐Bim L and GST⅐Bim S , host cells containing the Bim plasmid constructs were grown in LB medium at 37°C, reaching an A 600 of ϳ1.0. The temperature of the bacterial cultures was then lowered to 25°C before induction of GST⅐Bim expression. Purification of GST⅐Bim proteins followed the procedure described for the purification of the GST⅐DIC mutant proteins. Preparation of pure and untagged DLC8 was described in our earlier work (12).
Pull-down and Peptide Competition Experiments-Direct interactions between DLC8 and various GST-fused proteins were assayed in phosphate-buffered saline buffer (pH 7.4). Equal molar amounts of DLC8 and one of the GST fusion proteins (0.6 nmol each) were mixed in 100 l of the assay buffer. The GST fusion protein-DLC8 complexes were pelleted by 30 l of fresh GSH-Sepharose beads. The pellets were beads were used as a negative control in the binding experiment (B). The Ni 2ϩ -nitrilotriacetic acid beads contain some nonspecific binding to DLC8. To avoid possible confusion, the rest of the DIC fragments were prepared as GST fusion proteins. Purified GST was used as the negative control for the binding interactions between DIC fragments and DLC8. The lane numbers in C correspond to the construct numbers in A. MW, molecular weight markers.
washed three times with 0.5 ml of the assay buffer and subsequently boiled with 30 l of 2ϫ SDS-PAGE sample buffer. The intensity of the DLC8 band on SDS-PAGE gels was used to judge the strength of the interaction between DLC8 and various GST fusion proteins.
An 11-residue synthetic peptide (VSYSKETQTPL), corresponding to amino acid residues 147-157 of DIC, was commercially synthesized. To assay competition between the synthetic peptide and DIC for DLC8, increasing amounts of the peptide (from 0 to 200 molar ratio amounts of DIC) were included in the GST⅐DIC/DLC8 mixture. Residual GST⅐DICbound DLC8 was assayed using GSH-Sepharose affinity "pull-down" followed by SDS-PAGE as described above.
NMR Experiments-1 H, 15 N HSQC spectra of 15 N-labeled DLC8 complexed with various synthetic peptides were acquired on a Varian Inova 500 MHz spectrometer equipped with a z-gradient shielded triple res-FIG. 2. The DIC peptide specifically binds to DLC8. 1 H, 15 N HSQC spectra of 15 N-labeled free DLC8 (A) and the protein complexed with the DIC peptide (B). The assignments of selected amino acid residues of both free DLC8 and its complex with the DIC peptides are labeled with amino acid residue numbers and names. The large chemical shift changes of DLC8 resulting from DIC peptide binding indicate that DLC8 undergoes a significant conformational change upon binding to the DIC peptide. C, chemical shift changes of DLC8 resulting from the binding of the DIC peptide expressed using the minimal shift perturbation approach (19). The combined 1 H and 15 N chemical shift changes are defined according to Equation 1.
The scaling factor (␣ N ) used to normalize the 1 H and 15 N chemical shifts is 0.17. The secondary structure of DLC8 is also included in the figure. D, the DIC peptide competes with DIC for DLC8 in a dose-dependent manner. In this experiment, equimolar amounts of GST⅐DIC and DLC8 were mixed with increasing amounts of the DIC peptide. The remaining DLC8 complexed with GST⅐DIC was pulled down by GSH-Sepharose affinity beads and analyzed by SDS-PAGE followed by Coomassie Blue staining. MW, molecular weight markers. onance probe. All NMR spectra were recorded at 30°C, with a protein concentration of ϳ0.5 mM dissolved in 100 mM potassium phosphate buffer, pH 7.0.

DLC8 Binds to an 11-Residue Fragment Located within a Highly Conserved Region of the N-terminal Domain of DIC-To
understand how DLC8 is assembled in cytoplasmic dynein, we set out to identify the binding partner of DLC8 within the dynein complex. Given the overall similarity between axonemal and cytoplasmic dynein complexes, we suspected that DLC8 may also bind to DIC in cytoplasmic dynein. The gene encoding full-length DIC was assembled from three overlapping genomic clones (corresponding to the long form of DIC, GenBank TM accession number AF063229). His-tagged fulllength DIC was expressed in E. coli cells and purified in its denatured form. DIC was successfully refolded, and the refolded DIC was assayed for its binding to DLC8. Affinity pulldown assay indicates that the full-length DIC binds directly to DLC8 (Fig. 1B). The circular dichroism spectrum of the refolded, full-length DIC indicated that the protein has a well folded structure in solution (data not shown).
To further define the DLC8-binding region, we created a series of truncation mutants of DIC as GST fusion proteins (Fig. 1A). The purified proteins were subsequently assayed for binding to DLC8 by simple mixing and GSH-Sepharose bead pull-down. The DLC8-binding region was mapped within the N-terminal domain of DIC (Fig. 1C). An 11-residue fragment corresponding to amino acid residues 147-157 (VSYS-KETQTPL) of DIC was sufficient to bind to DLC8 (Fig. 1C, We verified the interaction between DLC8 and the 11-residue peptide fragment of DIC derived from the truncation mapping experiment using two different approaches. In the first experiment, we titrated 15 N-labeled DLC8 with a synthetic peptide corresponding to residues 147-157 of DIC (the DIC peptide). Fig. 2, A and B show the 1 H, 15 N HSQC spectra of DLC8 before and after binding to the DIC peptide. The chemical shift changes of DLC8 resulting from the DIC peptide binding are summarized using the minimal shift perturbation approach (Fig. 2C). These data show that the DIC peptide specifically binds to DLC8 and that the peptide-binding region is located at the target-binding groove formed by the ␤2, ␤3 strands, the loop linking ␤2 and ␤3, and part of the ␣2 helix (see "Discussion" for details). The observation of slow chemical exchange between the apo-DLC8 and peptide-bound DLC8 during titration indicates that the DIC peptide binds to DLC8 with a high affinity (data not shown). NMR titration experiments also showed that the DIC peptide binds to DLC8 with a 1:1 stoichiometry. In the second approach, we performed a binding competition experiment using the DIC peptide. Increasing molar amounts of the DIC peptide specifically competed with DIC protein for binding to DLC8 (Fig. 2D). As a negative control, an 11-residue peptide containing a partial Tctex-1-binding domain of DIC (amino acid sequence LGRRLHKLGVS) was found not to compete with DIC for binding to DLC8 (data not shown and Ref. 28).
The 11-Residue Peptide Fragment Is the Only DLC8-binding Domain of DIC-The results presented in Figs. 1 and 2 demonstrate that the 11-residue peptide fragment (residues 147-157) of DIC is sufficient to bind to DLC8. Is this 11-residue fragment in DIC the only binding site for DLC8? To address this question, we disrupted the 11-residue DLC8-binding site by deleting residues 151-155 (KETQT) of DIC. This deletion mutant did not bind to DLC8, indicating that the 11-residue peptide fragment mapped in this study is the only DLC8binding site in DIC (Fig. 3).
(K/R)XTQT Is a Common DLC8 Recognition Motif-To identify additional proteins that interact with DLC8, we performed a yeast two-hybrid screen using DLC8 as bait with both human and mouse brain cDNA libraries. A large number of specific interactors were isolated in the screen (see a partial list in Table I), many of which represent previously unidentified, potential DLC8-binding partners. We focus on a subset of the genes in Table I to study their interaction with DLC8.
The first gene we chose was clone 82 (c82), because the full-length sequence of c82 was recently deposited in Gen-Bank TM . The fragment isolated from the two-hybrid screen encodes the C-terminal half of the c82 protein (residues 119 -199). The interaction between c82 and DLC8 was confirmed in vitro using purified recombinant proteins (Fig. 4). Using a series of deletion mutants, the DLC8-binding domain of c82 was mapped to a 14-residue peptide fragment with amino acid sequence 159 VGMHSKGTQTAKEE 172 (Fig. 4).
The cDNA fragment of BS69 isolated from the DLC8 twohybrid screen encodes amino acid residues 396 -424 of the protein. GST-fused BS69-(396 -424) robustly bound to DLC8 in vitro, indicating that the DLC8-binding region in BS69 is contained within this 29-amino acid segment (Fig. 5A). A stretch of amino acids similar to the DLC8-binding regions of DIC and c82 was noted within this 29-residue fragment, as was 410 MLHRSTQTTN 419 .
The proapoptotic protein Bim was previously shown to interact specifically with DLC8, and the DLC8-binding domain was  mapped to a region of ϳ30 amino acid residues that were absent in the short form (Bim S ) of the alternatively spliced gene product (13). We confirmed this interaction between Bim L and DLC8 in vitro using purified recombinant proteins (Fig.  6A). Additionally, we synthesized three overlapping peptides to map the minimal DLC8-binding domain of Bim L (Fig. 6B). NMR titration of 15 N-labeled protein showed that all three peptides bound to the protein in an essentially identical manner (data not shown). The shortest peptide used in this experiment contains only nine amino acid residues (MSCDKSTQT). The 1 H, 15 N HSQC spectrum of this 9-residue peptide-bound form of DLC8 is shown in Fig. 6B; the data suggest that this 9-residue peptide forms a stable complex with DLC8 (18). We aligned the amino acid sequences of the DLC8-binding domains of various targets mapped in this study (Fig. 7). In this alignment, we also included putative DLC8-binding domains of Swallow and rabies virus P protein, two recently identified DLC8-binding proteins (14,21,22). The (K/R)XTQT motif was present in each of these DLC8 interactors and probably represents the consensus DLC8-binding sequence of these proteins (Fig. 7).
Mutational Analysis of the Consensus DLC8-binding Motif-To gain more insights into the interaction between DLC8 and the (K/R)XTQT motif, we mutated each consensus amino acid in the motif and assayed for the effect. To facilitate the binding assay, we fused the DLC8-binding domain of DIC with GST. In GST pull-down assays, mutation of the Gln residue in the (K/R)XTQT motif to either Ala or Gly completely abolished binding of the peptide to DLC8 (Fig. 8B). Mutation of either Thr residue in the motif to Gly also abolished interaction between the peptide and DLC8. Mutation of the first Thr (Nterminal to the Gln) to Ala greatly inhibited the binding, sug-gesting that the hydroxyl group of the Thr plays an active role in the complex formation. Mutation of the second Thr to Ser also significantly weakened the interaction, pointing to a possible role of the methyl group in the Thr side chain in peptide-DLC8 complex formation. Mutations of the Lys residue to a neutral Ala or a negatively charged Glu weakened the binding of the peptide to DLC8. However, the magnitude of the inhibition by the Lys mutations was lower than those observed with other point mutations. Taken together, the data in Fig. 8 indicate that all four consensus amino acid residues in the (K/R)XTQT motif play active roles in supporting productive complex formation between DLC8 and its target peptides. The data also indicate that the Thr-Gln-Thr tripeptide is likely to play a more dominant role in the peptide-DLC8 interaction. DISCUSSION As a multicomponent macromolecular complex, precise organization of various subunits of the dynein complex is essential for the motor to function properly. Recent molecular analysis of the cytoplasmic dynein complex has provided a detailed picture of the assembly of the DHC, DIC, and DLIC subunits in the motor complex (23,24). However, little is known regarding the assembly of the light chains. Both cytoplasmic and axonemal dyneins share the highly conserved DLC8. In Chlamydomonas outer arm dynein, DLC8 was suggested to associate with the intermediate chains of the motor complex (7). By analogy, it was suggested that DLC8 is also assembled into the cytoplasmic dynein by binding to DIC subunits of the motor complex (8), although the flagellar and cytoplasmic DICs share limited sequence identity (particularly in the N-terminal half of the proteins). Using purified recombinant proteins, we show here that DLC8 can indeed bind to the full-length intermediate chain of cytoplasmic dynein (Fig. 1). An 11-residue fragment in the N-terminal half of DIC is necessary and sufficient for binding to DLC8. Our data is consistent with an earlier observation that DLC8 and DIC exist at a 1:1 stoichiometric ratio in cytoplasmic dynein (8).
Mammalian DIC is encoded by two different genes, and each gene generates multiple protein products by alternative splicing (4, 26). The 11-residue DLC8-binding domain of mouse DIC mapped in this study contains the amino acid sequence 148 VSYSKETQTPL 157 , and this fragment is located C-terminal to the second alternative splicing site of DIC. Within this 11residue fragment, the 5-residue KETQT motif is likely to be responsible for the interaction of DIC to DLC8 (Figs. 6 and 8; see also Ref. 18). This 5-residue KETQT motif is present in all splice isoforms of both DIC gene products. Furthermore, the KETQT motif is highly conserved in cytoplasmic DICs throughout evolution (27). Given the fundamental roles that DLC8 plays in dynein motor function (9, 10), it might be expected that the DLC8-binding site should always remain available in various forms of DIC (and hence the dynein complex). We further note that the DLC8-binding domain does not overlap with the dynactin-and Tctex-1-binding sites of DIC, because dynactin binds to the first 120-amino acid residue fragment of DIC (4,28). If DLC8 indeed acts as a dynein cargo adaptor, a single dynein complex may thus interact simultaneously with multiple cargoes.
The amino acid sequence of DLC8 is highly conserved throughout evolution, and the protein is ubiquitously expressed in various tissues. As such, DLC8 was proposed to be a multifunctional regulatory protein, in addition to functioning as a light chain of the dynein complex (11,12,25). To gain more insights into the range of proteins that interact with DLC8, we performed a yeast two-hybrid screen using DLC8 as bait. No obvious amino acid sequence homology or functional relationship can be identified among the proteins listed in Table I or among other proteins previously identified as binding to DLC8. Given that DLC8 can bind to short peptide fragments of ϳ10 amino acid residues (12,17,18), we performed detailed mapping analysis of DLC8-binding domains in several additional DLC8-binding proteins either identified from our yeast twohybrid screen (c82 and BS69) or previously discovered (Bim).
The DLC8-binding domains of these proteins are contained within short stretches of amino acids (9 -29 amino acid residues). Sequence alignment analysis of the DLC8-binding domains immediately suggests that the 5-residue (K/R)XTQT sequence is a common DLC8 recognition motif in target proteins (Fig. 7). We also predict that DLC8 is likely to bind to Swallow via the 291 KATQT 295 sequence immediately C-terminal to the coiled-coil domain of Swallow. Our prediction is consistent with experimental data showing that the DLC8binding domain lies within amino acid residues 197-295 of Swallow (14). Recently, DLC8 was shown to interact with rabies virus P protein (the ␣ subunit of RNA polymerase) via residues 138 -172 of P protein (21,22). Sequence alignments of P proteins from various strains of rabies virus showed that all contain a conserved KSTQT motif at the center of the protein.
We suggest that DLC8 binds to P protein via this KSTQT motif. In the protein data base there are a large number of additional viral proteins containing a potential DLC8-binding (K/R)XTQT motif. For example, poly(A) polymerase catalytic subunit from vaccinia virus contains a 84 KQTQT 88 motif. Human papillomavirus probable E4 protein contains a KQTQT motif in its N terminus. Several viral proteases (e.g. human adenovirus endoprotease and Haemophilus influenzae immunoglobulin A1 protease precursor) contain potential DLC8-binding KSTQT motifs. It is possible that DLC8 is responsible for dyneinmediated retrograde transport of these viral proteins along microtubules in infected cells (e.g. retrograde axonal transport of viral proteins in virus-infected neurons).
It is important to point out that the (K/R)XTQT motif is not the only DLC8 recognition sequence in target proteins. For example, the amino acid sequence of the DLC8-binding domain of nNOS is clearly different from the (K/R)XTQT motif (Fig. 7). We were not able to find a (K/R)XTQT motif in a number of other known DLC8-binding proteins including myosin V, GKAP, AIBC1, and KIAA0710. We showed earlier that DLC8 binds to a (K/R)XTQT motif-containing peptide and a peptide encompassing the DLC8-binding domain of nNOS with remarkably similar mechanisms, although the two peptides share little amino acid sequence identity (18). We have also shown that the structural and dynamic properties of the targetbinding site of DLC8 is uniquely suited to bind to peptide fragments with diverse amino acid sequences without sacrificing target binding specificity and affinity. 2 Therefore, it is perhaps not surprising that DLC8 can bind a number of other target proteins, even though these proteins do not contain a (K/R)XTQT motif. We would also like to emphasize that the DLC8-binding proteins listed in Table I are Fig. 6) to DLC8. In this figure, the DLC8 dimer is shown in a surface model, and the peptide is in an explicit atom model. The positively charged amino acids in DLC8 are shown in blue; negatively charged residues are in red; hydrophobic residues are in yellow; and polar residues are in gray. The amino acid residues of DLC8 that form intimate electrostatic and hydrogen-bonding interactions are labeled with amino acid residue numbers and names (the numbers with and without a prime are used to differentiate residues from two different monomers of DLC8). The amino acid sequence of the peptide is labeled with residue name and number using double primes. The coordinates of the structure were taken from our earlier work (Ref. 18, Protein Data Bank code 1F95).
DLC8 targets, because we only used partial fragments of the proteins for binding studies. Further in vitro and in vivo studies are required to substantiate whether these proteins are genuine DLC8-binding targets under physiological conditions.
Our earlier structural studies showed that the (K/R)XTQT motif forms a ␤-strand structure in the DLC8-target peptide complex ( Fig. 8C; see also Ref. 18). To establish a better structure-function relationship of the (K/R)XTQT motif-mediated DLC8 binding, the four conserved residues in the motif were systematically mutated. The structure of the DLC8 complexed with a (K/R)XTQT motif-containing peptide suggested that the binding affinity of the peptide to DLC8 is due to a combination of backbone hydrogen bonding and a number of hydrogenbonding and electrostatic interactions between the side chains of the peptide and DLC8 (18). For example, strong hydrogenbonding interactions between the side chain of the Gln residue in the peptide and side chains of Glu 35 and Lys 36 of DLC8 were observed. Mutation of the Gln to Ala, which eliminates the hydrogen-bonding capacity of the side chain of this amino acid residue, completely abolished peptide binding to DLC8 (Fig. 8). We believe that the Gln residue plays a central role in promoting the formation of the peptide-DLC8 complex. Structurebased sequence alignment of the nNOS peptide and the (K/R)XTQT motif peptide showed that a Gln residue in the nNOS peptide occupies the same position as the Gln residue in the (K/R)XTQT peptide upon formation of complexes with DLC8 (18). The total loss of the interaction observed in the individual Thr to Gly mutations is probably a combination of the loss of side chain interactions between the peptide and DLC8 and destabilization of the backbone ␤-strand structure of the peptide. Mutations of the Lys residue to either Ala or Glu lead to somewhat decreased binding affinity between the peptide and DLC8, suggesting that the Lys residue in this position has a relatively small contribution to the binding affinity of the peptide. The mutagenesis study reinforces our earlier conclusion that the interactions between DLC8 and the (K/R)XTQT motif-containing peptide are dominated by a combination of side chain and backbone hydrogen-bonding and electrostatic interactions (18).
In summary, we have shown that DLC8 binds to the Nterminal domain of cytoplasmic DIC via a highly conserved (K/R)XTQT motif. The data firmly establish that DLC8 is assembled into the cytoplasmic dynein by binding to its intermediate chain located at the base of the motor complex. In addition, we further show that the (K/R)XTQT motif represents a general DLC8 recognition sequence in its diverse target pro-teins. The experimental data presented in this work, together with our earlier structural studies, reinforce the hypothesis that DLC8 probably acts as a multifunctional regulatory protein by binding to a large number of functionally unrelated proteins.