Functional Architecture of the Outer Arm Dynein Conformational Switch*

Background: Ciliary dyneins monitor and respond to the mechanical state or curvature of the microtubular axoneme. Results: NMR chemical shift mapping and binding assays define functional subdomains within a key component (LC1) of this system. Conclusion: LC1 provides a direct tether between a dynein motor unit and the microtubule that modulates motor function. Significance: This study provides insight into the mechanism by which dyneins are coordinated during ciliary beating. Dynein light chain 1 (LC1/DNAL1) is one of the most highly conserved components of ciliary axonemal outer arm dyneins, and it associates with both a heavy chain motor unit and tubulin located within the A-tubule of the axonemal outer doublet microtubules. In a variety of model systems, lack of LC1 or expression of mutant forms leads to profound defects in ciliary motility, including the failure of the hydrodynamic coupling needed for ciliary metachronal synchrony, random stalling during the power/recovery stroke transition, an aberrant response to imposed viscous load, and in some cases partial failure of motor assembly. These phenotypes have led to the proposal that LC1 acts as part of a mechanical switch to control motor function in response to alterations in axonemal curvature. Here we have used NMR chemical shift mapping to define the regions perturbed by a series of mutations in the C-terminal domain that yield a range of phenotypic effects on motility. In addition, we have identified the subdomain of LC1 involved in binding microtubules and characterized the consequences of an Asn → Ser alteration within the terminal leucine-rich repeat that in humans causes primary ciliary dyskinesia. Together, these data define a series of functional subdomains within LC1 and allow us to propose a structural model for the organization of the dynein heavy chain-LC1-microtubule ternary complex that is required for the coordinated activity of dynein motors in cilia.

The beating of cilia and flagella is driven by the inner and outer rows of dynein arms associated with the axonemal doublet microtubules. This motile behavior requires that a wave of dynein motor activity be propagated along the structure, usually from base to tip, and theoretical models suggest that one or more mechanosensitive switch systems might act to monitor axonemal curvature (1)(2)(3). In Chlamydomonas, evidence for the existence of such a switch mechanism came from experiments where paralyzed flagella of mutant strains lacking vari-ous axonemal substructures were induced to beat in a rhythmic oscillatory manner for one or more cycles following mechanical activation (4). Based on this analysis, it was further proposed that the flagellum contains two mechanosensory systems that monitor and respond to the mechanical state (i.e. curvature) of the axoneme. The first of these involves the central pair microtubule complex and radial spokes and controls the activity of the inner dynein arm system, whereas the second appears to be distinct at the molecular level and impinges on the outer arms.
Chlamydomonas outer arm dynein with its associated docking structures is a highly complex motor consisting of at least 20 distinct molecular components with a total mass in excess of 2 MDa (for a review, see Ref. 5). This dynein is built around three heavy chains (HCs 2 ; ␣, ␤, and ␥), which exhibit different ATPase and motor properties (6 -8). The ␥ HC (encoded at DHC15 (9)) is also distinct in that it is the only axonemal dynein HC known to have additional components bound to the motor unit: specifically, cross-linking and vanadate-mediated photocleavage experiments suggested that one copy of light chain 1 (LC1) is bound to the ATP hydrolytic subdomain AAA1, and a second copy is bound to either AAA3 or AAA4 (10). 3 Further evidence that LC1 is associated with the motor unit came from analysis of the oda2-t mutant that expresses a truncated 150-kDa form of the ␥ HC retaining only the N-terminal region; LC1 is not present in oda2-t outer arm dynein, whereas all other components are (11). LC1 also binds to tubulin located within the A-tubule of the outer doublets, and thus, the ␥ HC is tethered to the A-tubule via both its N-terminal domain that associates with the intermediate chain-LC complex and its motor unit (12). Furthermore, a third distinct tether comes into play under high Ca 2ϩ conditions when the LC4 calmodulin homologue (encoded at DLE1 (9)) releases its Ca 2ϩ -dependent binding site on the ␥ HC N-terminal domain (it remains bound to this ␥ HC region through a second Ca 2ϩ -insensitive site) and instead binds to intermediate chain 1 within the intermediate chain-LC complex (13,14). LC1 (encoded at DLU1 (9)) is a 22-kDa leucine-rich repeat (LRR) protein that is one of the most conserved components within outer arm dynein. In humans, a homozygous mutation within the last LRR of the LC1 orthologue DNAL1 has recently been demonstrated to cause primary ciliary dyskinesia (15). This LC consists of an N-terminal helix followed by six ␤␤␣ motifs derived from the LRRs and a C-terminal helical domain that protrudes from the main protein axis defined by the LRR barrel (16). The orientation of the C-terminal helical region is controlled by two residues (Met 182 and Asp 185 ) that exhibit high backbone dynamics; the remainder of the protein is relatively rigid with the exception of both termini and a flexible leucine spine within one interaction surface (17,18). The C-terminal region is also highly charged: in mammals, the ␣9 helix is very acidic, whereas in Chlamydomonas and Trypanosoma, there are two conserved basic residues (Arg 189 and Arg 196 in Chlamydomonas) in addition to several Asp/Glu residues.
Previously we found that expression of mutant forms of LC1 in which these Arg residues or the Met/Asp exhibiting high dynamics were altered yielded dominant negative phenotypes in a wild-type background (12). These Chlamydomonas cells swam slowly as they were unable to coordinate their flagella movement due to random stalling during the beat cycle and especially at the power/recovery stroke transition. In addition, these strains had an aberrant response to viscous load. Under low viscosity conditions, they generated the low propulsive force characteristic of outer armless mutants (19). However, as load increased (and beat frequency decreased), the force output approached that of wild-type cells, a response that essentially phenocopied the oda2-t mutant. Introduction of different mutant forms of LC1 produced graded effects: for example, mutation of Met 182 to Ala was much less disruptive than alterations to either Gly or Pro, which were predicted to change the dynamic behavior. Mutation of the equivalent residues in Trypanosoma brucei LC1 also caused major motility defects (20).
Some of the key issues for understanding this dynein control system are to identify the functional subdomains within LC1 and to determine their orientation with respect to the motor unit and microtubule. Here we have used chemical shift mapping to define the perturbed regions within five mutant forms of LC1 that exhibited a range of phenotypic effects in vivo. To gain further insight into the mechanisms involved, we have also used targeted mutagenesis to delineate the segment of this molecule involved in associating the dynein motor unit with the outer doublet A-tubule and have characterized the structural consequences of the PCD-causing mutation within the terminal LRR. Together, these data allow functional subdomains within LC1 and the overall geometry to be defined. The results implicate the LC1 C-terminal domain in controlling motor function through direct interactions within the HC AAA domains.

EXPERIMENTAL PROCEDURES
Molecular Biology and Mutagenesis-A cDNA containing the entire LC1 coding region in pBluescript II SKϪ was obtained previously (10). The M182A, D185G, D185P, R189E, and R196A mutations used for in vivo experiments (12) , and the  additional K3E, K8E, R12E, R17E, K18E, K39E, K62E, R79E, K83E, K84E, K126E, K135E, and N150S alterations were generated by oligonucleotide-directed mutagenesis using standard methods. The mutated coding regions were then obtained by PCR, subcloned into the pET16B vector (Novagen), and transformed into Escherichia coli BL21(DE3). When expressed, these proteins contain an N-terminal His 10 tag and linker that may be cleaved from the target molecule by Factor Xa digestion leaving a single additional His residue at the N terminus (designated residue Ϫ1).
Preparation of LC1 Proteins-For 15 N labeling, cells were grown at 37°C in 1 liter of M9 minimal medium containing [ 15 N]NH 4 Cl to an optical density at 600 nm of ϳ0.6. The culture was then supplemented with isopropyl ␤-D-1-thiogalactopyranoside to a final concentration of 1 mM and grown for an additional 3 h to allow for protein expression. Double labeled samples of D185G and R189E LC1 were prepared in a similar manner from cells grown in the presence of [U-13 C]glucose and [ 15 N]NH 4 Cl. For unlabeled samples, cells were grown in LB complete medium, and protein expression was induced for 2 h.
Cells were harvested by centrifugation; resuspended in 50 ml of 20 mM Tris⅐Cl, pH 8.0, 50 mM NaCl; and frozen at Ϫ20°C. Samples were subsequently defrosted, and cells were lysed by sonication. Following centrifugation to remove insoluble material, total soluble protein was applied to a Ni 2ϩ resin column and washed with 20 mM Tris⅐Cl, pH 8.0, 500 mM NaCl, 20 mM imidazole. His-tagged protein was subsequently eluted with 20 mM Tris⅐Cl, pH 8.0, 500 mM NaCl, 250 mM imidazole. Peak fractions were pooled and dialyzed extensively against 20 mM Tris⅐Cl, pH 8.0, 50 mM NaCl, 10 mM EDTA to remove any paramagnetic Ni 2ϩ . Following dialysis, the His 10 tag was cleaved from LC1 proteins by addition of 100 units of Factor Xa and incubation at room temperature overnight; samples were subsequently concentrated using Amicon Ultra-4 ultrafiltration units (10,000 molecular weight cutoff). Proteins were subjected to a final purification step and buffer exchange by gel filtration in a Superdex 200 10/300 GL column attached to an ÄKTApurifier chromatography work station and equilibrated with 2.5 mM Tris⅐Cl, pH 6.8, 50 mM NaCl; this buffer was used previously for chemical shift assignments and collection of other NMR data necessary for structure determination (21). All NMR experiments and binding assays utilized only monomeric LC1.
NMR Spectroscopy-Purified 15 N-labeled LC1 samples were concentrated to a final volume of 600 l and a concentration of ϳ1 mM (ϳ22 mg/ml) or greater using Amicon Ultra-4 ultrafiltration units. Thirty-five microliters of D 2 O was then added, and two-dimensional 1 H-15 N HSQC spectra were collected using a Varian 800-MHz NMR spectrometer equipped with a cryogenic probe. The NMR spectra and chemical shift alterations were analyzed with NMRViewJ. To obtain the complete backbone assignments for the D185G form of LC1, the 1 H-15 N HSQC, HNCA, HNCO, and HBHA(CO)NH spectra were obtained from a 13 C, 15 N-labeled sample using a Varian 500-MHz NMR spectrometer equipped with a cryogenic probe; the HNCACB experiment did not yield a usable spectrum due to the large size of LC1. Peak assignments were made with XEASY, and the backbone torsion angles were predicted from the chemical shifts using TALOSϩ (22). The 1 H-15 N HSQC spectrum from 13 C, 15 N-labeled R189E LC1 was sufficiently poorly resolved that it was not feasible to reassign the entire backbone for this mutant form.
Circular Dichroism Spectroscopy-The concentrations of purified fully denatured wild-type, D185G, and N150S LC1 proteins in 2.5 mM Tris⅐Cl, pH 6.8, 50 mM NaCl were determined by absorbance at 280 nm using an extinction coefficient (⑀) of 17,810 liters⅐mol Ϫ1 ⅐cm Ϫ1 . To determine the melting temperature of these LC1 proteins, the ellipticity signals at 222 nm ( 222 ) of 3 M samples were measured in the temperature range 4 -86°C (65°C for N150S) using a Jasco J-715 spectropolarimeter as described (23).
Axoneme Binding Assay-The Chlamydomonas reinhardtii wild-type cc125 and mutant oda6 that lacks outer dynein arms were grown in R medium as described previously (24). Cells were deflagellated with dibucaine, and flagellar axonemes were purified by standard methods. Wild-type axonemes were depleted of dynein arms by extraction with 0.6 M NaCl. Axonemes were resuspended in 20 mM Tris⅐Cl, pH 8.0, 5 mM MgSO 4 , 0.5 mM EDTA, 50 mM NaCl prior to addition of LC1 proteins. Following incubation at room temperature for 60 min, axonemes were pelleted in a microcentrifuge, washed with buffer, and prepared for electrophoresis. Samples were The secondary structure derived from the Chlamydomonas protein structure (Protein Data Bank code 1M9L) is indicated above the alignment. The asterisk indicates the conserved Asn residue that is mutated to Ser in the PCD patients. b, color-coded alignment of the C-terminal helix: polar, yellow; hydrophobic, green; acidic, red; basic, blue; glycine, white. c, the cross-species conservation is displayed on the LC1 surface: the graded colors represent the degree of conservation (identity) from red (Ն90%) to white (Յ10%). The arrows indicate major highly conserved regions of the molecular surface. separated in 12.5% polyacrylamide gels containing SDS and stained with Coomassie Blue. Stained gels were imaged using an ImageQuant LAS4000 digital imaging system (GE Healthcare), and band intensities were quantified using ImageQuant TL software.
Other Computational Methods-The alignment of LC1 orthologues was performed using ClustalW, and the output was processed with BOXSHADE. The level of sequence conservation was titrated onto the LC1 structure with VENN (25). Spectral overlays were prepared using Adobe Photoshop CS4. The structural model of D185P mutant LC1 was generated with SWISS-MODEL (26).

RESULTS
Structural Orthology of LC1 Proteins-LC1 is one of the most highly conserved components of axonemal outer arm dyneins ( Fig. 1, a and b); for example, the Chlamydomonas and human orthologues share 55% identity (70% similarity). Mapping the degree of sequence identity onto the Chlamydomonas LC1 structure revealed two major areas of surface conservation (Fig.  1c, arrows). The first is located within the midregion of the molecule centered on Trp 99 (this exposed hydrophobic region may represent part of the ␥ HC binding surface), whereas the second is on the opposite face near the N terminus and includes a series of charged residues. Table 1) suggested that the C-terminal domain might represent a key functional subdomain of LC1, although intriguingly there are significant sequence differences observed within this region when assessed across a broad phylogenetic spectrum ( Fig. 1, a-c). As long range perturbations could disrupt other segments of the molecule, we set out to determine the extent to which the introduced mutations affected the structure of LC1 and thus define the region responsible for the observed phenotypic effects. The chemical shifts (␦) for individual atoms are exquisitely sensitive to their local environment, and so NMR-based chemical shift mapping was used to identify those regions of the mutant proteins that undergo alterations in their chemical environment compared with wild type (Fig. 2). To achieve this, we prepared 15 N-labeled wild-type and M182A, D185G, D185P, R189E, and R196A mutant forms of LC1 and collected two-dimensional 1 H-15 N HSQC spectra using an 800-MHz NMR spectrometer equipped with a cryogenic probe (Fig. 2). These particular mutations were chosen as they represent all residues altered previously and generated a graded series of phenotypic effects in vivo (12). All six molecules yielded well resolved peaks (Fig. 2, lower panel, insets), and the spectra exhibit large chemical shift dispersion in both 1 H and 15 N dimensions indicative of well folded proteins (Fig. 2, lower panels).

Perturbations Caused by Dominant Negative Mutations in LC1 C-terminal Domain-Previous single residue mutagenesis experiments (Ref. 12 and see
Superficially the spectra for mutant forms of LC1 appear similar to wild type. However, on careful examination, a significant number of chemical shift alterations (⌬␦) were identified. In addition, the observed signals for many peaks from the R189E mutant (and to a lesser extent M182A) showed evidence of considerable line broadening and a concomitant decrease in peak height that is usually indicative of transitions between structural states occurring on the NMR time scale (i.e. in the s to ms range); this line broadening was not due to aggregation as the sample remained completely monomeric even after many months. The broadening is illustrated in Fig. 3, which compares the R189E spectrum at different contour levels with wild type. The line broadening is especially evident in the low contour map where many R189E peaks were merged together. At a high contour, many R189E LC1 peaks were missing with the exception of those near the termini, which are highly dynamic (17).
To assess the degree and extent of structural perturbations caused by these various mutations, the 1 H and 15 N ⌬␦ values were clustered into four color-coded groups (Յ0.1, Յ0.2, Յ0.3, and Ͼ0.3 ppm or peak missing), and the corresponding side chains are displayed on the LC1 structure (Fig. 4); the shift differences for every residue in both dimensions are shown in Fig. 5. For several mutations, e.g. D185G, significant changes occurred only in residues relatively close in three-dimensional space to the mutation site, and indeed all changes for this mutant were confined to the C-terminal domain. Even so, the number of residues exhibiting significant ⌬␦ values suggests that the observed phenotypes cannot necessarily be ascribed to Only monomeric LC1 was used for binding assays. c The values quoted were determined by quantitative densitometric analysis of the Coomassie Blue-stained gels shown in Fig. 9 using ImageQuant TL. Background subtraction utilized the rolling ball method. d ND, not determined. e Monomeric R189E LC1 was very stable, and only 0.47% was found in oligomeric form following Ͼ6 months at 4°C. the mutation of a particular residue but rather may reflect a more global perturbation or reorientation of this domain. To address this directly, we reassigned the backbone for 13 C, 15 Nlabeled D185G LC1 from 1 H-15 N HSQC, HNCA, HNCO, and HBHA(CO)NH spectra and performed a TALOSϩ backbone chemical shift analysis. This revealed that the overall helical structure of the C-terminal domain was not greatly perturbed but that the and angles for residues surrounding the mutation site were significantly changed (Fig. 6).
Several mutations (R189E and to a lesser extent M182A and R196A) caused perturbations that were propagated along the two ␤ sheet faces and intervening loop region of the ␤␤␣ barrel toward the N terminus; in contrast, the ␣ helical face of the barrel was essentially unaffected. The R189E transition was especially disruptive, and resonances for many residues within the C-terminal domain were greatly altered or missing. In addition, significant perturba-tions extended along the large ␤ sheet face that is thought to provide the interaction surface with the ␥ HC. Thus, the dramatic phenotypic consequences observed for this mutation in vivo (Table 1) may reflect disruption of both the C-terminal domain and other LC1 regions.
We previously altered both Met 182 and Asp 185 to Pro to reduce the high backbone dynamics at these positions that were predicted to control the orientation of the terminal ␣9 helix. Analysis of the D185P HSQC spectrum revealed that two amide peaks of unequal intensity were present for the preceding residue Val 184 ; note that Pro residues do not contain a free amide and thus do not yield a peak in this NMR experiment. In contrast, alteration of Asp 185 to Gly to increase backbone mobility resulted in a single Val 184 peak that was shifted to a different location. Pro residues can exist in either cis or trans conformations (Fig. 7), and molecular modeling revealed that a cis-Pro residue is readily accommodated within the wild-type structure  15 N HSQC spectra for wild-type and the M182A, D185G, D185P, R189E, and R196A mutant forms of LC1 are shown in the lower panels. Each peak represents an amide resonance derived from the peptide backbone, Asn/Gln side chains, or the Trp indole ring. All spectra reveal a similarly high degree of peak dispersion in both 1 H and 15 N dimensions, suggesting that the mutant proteins are well folded as is wild-type LC1. The insets show an enlargement of a small region (red box in wild-type spectrum) to illustrate the high resolution achieved for all LC1 proteins using the 800-MHz spectrometer. with little change in the location of Val 184 . Consequently, the less intense Val 184 peak in the D185P spectrum that is very close to the wild-type position likely represents this isoform. This in turn suggests that the trans-Pro isomer, which would cause a significant alteration in the orientation of the C-terminal ␣9 helix, is the predominant form in solution.
Microtubule Interaction Surface-In addition to an interaction with the ␥ HC, LC1 also associates directly in situ with tubulin located in the A-tubule of the outer doublets (12). Defining the region of LC1 that mediates this association is important as it determines the geometry of the conformational switch and orientation of the LC1 C-terminal domain that has an impact on outer arm function. As the LC1-tubulin interaction is disrupted by high salt and readily detected using the zero-length cross-linking reagent 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide to generate an M r 66,000 LC1-tubulin product (10), the interface must include both acidic and basic residues that interact directly. The exposed tubulin C-terminal domain is highly acidic, and therefore, we predicted that LC1 provides one or more basic residues to this interaction. To test this, we prepared 12 mutant forms of LC1 in which a single basic residue on the face opposite the presumptive ␥ HC interaction surface was changed to Glu (Fig. 8, top left  panel). Interestingly, many of these basic residues are conserved between Chlamydomonas, trypanosomes, and humans and form a helical stripe that spans the short axis of the LC1 LRR barrel (Fig. 8, top right panel). With one exception, all these mutant proteins were soluble and could be purified in monomeric form; the R79E version of LC1 was obtained only as larger oligomers and could not be used further (Fig. 8, lower panel). Once purified, the mutant proteins were stable and remained almost completely monomeric even after many months at 4°C (Table 2).
Axonemal tubulins are highly modified by several processes, including polyglutamylation, polyglycylation, and acetylation; these modifications can have a major impact on dynein-axonemal microtubule interactions e.g. (27). Consequently, to assess the ability of mutant LC1 proteins to interact specifically with axonemal microtubules, we prepared axonemes from the oda6 mutant, which lacks outer dynein arms, or depleted wild-type axonemes of dynein arms by high salt extraction. Thus, the axoneme samples have the native LC1 interactions sites available. Monomeric LC1 does not sediment during centrifugation in the absence of axonemes (Fig. 9a), whereas wild-type LC1 bound oda6 axonemes in a dose-dependent manner (Fig. 9b). Furthermore, binding of LC1 to a fixed amount of dynein-depleted wild-type axonemes became saturated even in the presence of a large excess of LC1 (Fig. 9c). However, three mutants (K18E, K39E, and K62E) showed severe decreases (to Ͻ25% of wild type) in their ability to bind oda6 axonemes, and several others (K3E, K8E, R12E, R17E, and K126E) were significantly reduced from wild-type levels ( Fig. 9d and Table 2). In contrast, the K83E, K84E, and K135E alterations and mutations in the C-terminal domain had little or no negative effect on microtubule binding activity (Tables 1 and 2).
Identification of the K126E mutation as affecting axoneme binding was unexpected in that it is located far from other alterations that yielded a similar effect. K126E LC1 was stable; the purified protein did not aggregate and indeed remained completely monomeric even after several months at 4°C (data not shown). However, this residue is located close to Arg 79 , which caused severe disruption and essentially complete aggregation when altered to Glu, suggesting that changes propagated through the structure might be responsible for the decrease in binding activity.
Consequences of the N150S PCD Mutation-Asn 150 is part of the terminal LRR consensus sequence, and the side chain of this residue packs within the protein core (16). In humans, the homozygous N150S mutation causes PCD and has been predicted to convert the ␣7 helix to a loop (15). To directly assess the consequences of this disease-causing alteration, we determined the extent of the perturbation by chemical shift mapping and thermal titration circular dichroism spectroscopy. N150S LC1 could be purified in monomeric form and yielded a dis-

The Outer Arm Dynein Conformational Switch
persed, well resolved 1 H-15 N HSQC spectrum (Fig. 10, a and  insets) even though significant precipitation occurred during data collection at 25°C. Resonances for Ile 105 , Ala 106 , Met 122 -Lys 126 , Thr 128 , Leu 146 -Gly 149 , and Leu 152 -Asn 154 either could not be identified or were shifted by Ͼ0.2 ppm (Fig. 10b). These residues map to the ␤7, ␤8/␤9, and ␤10 strands of the fourth, FIGURE 6. TALOS؉ analysis of D185G LC1 structure. The backbone assignments for D185G LC1 were obtained from a 13 C, 15 N-labeled sample, and the backbone torsion angles were predicted from the chemical shifts using TALOSϩ. The upper panels indicate the predicted secondary structure (blue, ␤ sheet; red, ␣ helix) and order parameter (S 2 ). Ramachandran plots of / angles for best matches in the database for the mutation site and immediately adjacent residues are shown in the lower panels. The results illustrate that this mutation does not dramatically alter the helical structure of the C-terminal domain. However, the / angles for Gly 185 (Ϫ89 Ϯ 23°and 152 Ϯ 18°) are very different from those for Asp 185 in wild-type LC1 (48.2°and 4.7°), suggesting that the orientation of the terminal helix is altered. fifth and sixth LRRs, respectively, and the ␣7 helix that is immediately adjacent to the mutation site (Fig. 10c). Although some of these residues are exposed, the side chains of Ile 105 , Met 122 , Asn 125 , Leu 147 , and Leu 152 are located within the core, and thus, this mutation disrupts the internal packing. Consistent with this, we observed that the melting temperature as assessed by circular dichroism spectroscopy decreased from 51.5°C for wild-type LC1 to 44°C for the N150S mutant (Fig. 10d). In con-trast, the T m for D185G LC1 was slightly increased over wild type. The unfolding profile for LC1 appears to be two-state.

DISCUSSION
LC1 is one of the most highly conserved components of axonemal outer arm dyneins, and multiple lines of evidence have now revealed that it plays a key role in ciliary motility. In Chlamydomonas, dominant negative mutations in the C-termi-FIGURE 8. Mutagenesis of basic residues within central LRR barrel. The 12 Arg/Lys residues mutated to Glu to assess their involvement in LC1-microtubule interactions are indicated in the top left panel. These residues form two clusters: a group near the N-terminal end of the central barrel and a helical stripe that extends along the barrel and crosses the protein short axis. The conservation of these residues between Chlamydomonas and humans is mapped onto the structure in the top right panel. The color code is as follows: identical, deep salmon; conservative substitution, olive; not conserved, green. Part of the N-terminal basic patch and much of the helical stripe is highly conserved. The lower panel shows the gel filtration profiles for K8E, K18E, and R79E forms of LC1 that exhibit distinct propensities for oligomerization: the peak eluting at ϳ16 ml represents monomeric LC1, whereas oligomers elute earlier. The insets show the final purified mutant proteins separated in 12.5% SDS-polyacrylamide gels (Coomassie Blue stain).
nal domain caused flagellar stalling and an abnormal response to viscous load (12). Similar mutations in Trypanosoma also lead to defective motility (20), whereas RNAi-driven knockdowns caused failure of outer arm dynein assembly (28). In the context of a ciliated epithelium, reduction of LC1 levels in the planarian Schmidtea mediterranea resulted in lower beat frequency and disruption of the hydrodynamic coupling necessary for generating metachronal synchrony but did not appear to affect dynein assembly (29).
Furthermore, a single residue alteration within the terminal LRR of the human orthologue DNAL1 has been demonstrated to cause PCD (15), and lack of a LC1/DNAL1 orthologue has also been reported to yield non-flagellate zoospores in the pathogenic fungus Phytophthora nicotianae (30).
Subdomain Organization of LC1 and Orientation of Conformational Switch-LC1 interacts with both the ␥ HC motor domain and tubulin located within the outer doublet A-tubule. . Axoneme binding activity of mutant LC1 proteins. a, the supernatants (S) and pellets (P) for wild-type and K84E mutant LC1 spun in a microcentrifuge in the absence of axonemes. This panel demonstrates that monomeric LC1 does not sediment during centrifugation. b illustrates the dose-dependent binding of LC1 to axonemes from the outer armless mutant oda6. The numbers on the image indicate the relative amounts of LC1 bound to the axoneme sample (0.83 mg/ml) as the LC1 concentration was increased from 0 to 0.18, 0.94, and 1.88 mg/ml. c, increasing amounts of D185P LC1 (0, 0.07, 0.14, 0.28, 0.42, and 0.7 mg/ml final concentration) were added to dynein-depleted wild-type axonemes (0.7 mg/ml). The pellets from the binding assay are shown in the top panel, and equivalent loadings of the corresponding supernatants are shown in the lower panel. The numbers on the upper panel indicate the relative amounts of LC1 bound to a fixed amount of axonemes and demonstrate that the binding becomes saturated. d illustrates the effects of individual point mutations on LC1-axoneme binding; only pellets from the binding assay are shown. All assays contained the same concentration of both axonemal (0.9 mg/ml) and LC1 (0.56 mg/ml) proteins except for M182A (0.81 mg/ml). Several of the Lys 3 Glu mutations severely affected the ability of LC1 to bind axonemes, whereas the more C-terminal mutations did not: these data are quantified in Tables 1  and 2. All samples were separated in 12.5% SDS-polyacrylamide gels and stained with Coomassie Blue. Value quoted is the average of two separations. Only monomeric LC1 was used for binding assays. b Purified monomeric LC1 proteins were stored at 4°C for Ͼ4 months, and the percentage of aggregation occurring was determined by gel filtration in a Superdex 200 10/ 300 GL column. c The values quoted were determined by quantitative densitometric analysis of the Coomassie Blue-stained gels shown in Fig. 9 using ImageQuant TL. Background subtraction utilized the rolling ball method. d Essentially no monomeric R79E LC1 was evident in the chromatograms (see Fig. 6). e ND, not determined.
Chemical shift mapping indicates that perturbation of the C-terminal helical domain accounts for the dominant negative effects observed in vivo. Indeed, for one of the most disruptive alterations (D185G), all identified chemical shift changes were within the C-terminal region. Furthermore, the interaction with the ␥ HC is hydrophobic and so likely involves one or both of the exposed Trp residues. The mutagenesis experiments described here also allow the region responsible for binding tubulin to be localized to one face of the N-terminal region of the LRR barrel. These data are mapped onto the LC1 structure in Fig. 11a and together define discrete functional subdomains.
As at least one of the two copies of LC1 appears to span the ϳ6-nm gap between the ␥ HC AAA ring domain and the A-tubule (12,31), it must be aligned approximately along its long axis of 69.8 Å; the two short axes are only 28.4 and 30.0 Å (17). Thus, taken together, the geometric considerations and functional data suggest that one copy of LC1 is arranged with the  15 N dimensions, respectively. b, the N150S mutation (green) and residues with chemical shifts that are Ͼ0.2 ppm different from wild type (orange) or missing/unidentified (red) are illustrated on the LC1 ribbon structure with surface mesh. Significant changes surround the mutation site and are propagated into the fourth and fifth LRRs. c, the thermal stability of wild-type, N150S, and D185G LC1 was assessed by circular dichroism spectroscopy in the range 4 -65°C. N150S LC1 was considerably destabilized and exhibited a T m of 44°C as compared with 51.5 and 52.5°C for wild-type and D185G mutant LC1, respectively. N-terminal domain binding the A-tubule and the hydrophobic region(s) interacting with the AAA1 domain (Fig. 11b). This in turn requires that the LC1 C-terminal region is also in direct contact with the HC motor unit; it might either be directly inserted into the cleft between AAA domains or aligned along an exposed face. One key question is now to identify the residues on the ␥ HC with which this LC1 subdomain interacts as this will provide additional clues as to how LC1 is involved in the control of outer arm function.
Cross-linking and vanadate-mediated photocleavage data support association of an LC1 monomer with AAA1 with a second copy bound to either AAA3 or AAA4 (10) 3 ; the ambi-guity is derived from assignment of the V2a and V2b cleavage sites (32) to specific AAA domains. Based on the structural data that are now available for both the HC and LC1, it is uncertain whether both copies of LC1 could be bound to the A-tubule at the same time, although they might interact sequentially during the mechanochemical cycle, or indeed how LC1 if bound to AAA4 could even interact with the A-tubule at all if the ␥ HC motor unit is arrayed in the same orientation as others within the outer arm (33,34). Thus, the possibility remains that this second copy of LC1 is associated with the HC in a different manner and/or performs a function distinct from LC1 bound to AAA1. FIGURE 11. Functional subdomains within LC1 and architecture of dynein conformational switch. a, three views of the LC1 long axis and views from either terminus of the molecular surface are shown at the top. The tubulin-binding region is defined by the basic residues (blue) that disrupt this activity, whereas the C-terminal domain (orange) represents the alterations present in the D185G mutation that exhibits a strong phenotype in vivo with the least structural perturbation. These two sections are at opposite ends of the molecule, strongly suggesting that the C-terminal region indeed interacts directly with the ␥ HC AAA domains. This latter association is known to be hydrophobic, and two potential regions, a hydrophobic patch centered on Trp 99 and a smaller region involving Trp 130 and Tyr 156 , which may individually or together mediate this interaction, are indicated in green. The PCD-associated mutation site is shown in red. b, a model for the overall geometry of LC1-mediated association of the ␥ HC motor domain with the outer doublet A-tubule is illustrated as van der Waals surface representations of the dynein motor unit (rainbow color from the N-terminal linker in blue to AAA6 in red; Protein Data Bank code 3QMZ) (38), LC1 (from a), and the tubulin dimer (yellow-orange; Protein Data Bank code 3IZO) (39); the structures are all illustrated at the same scale. One copy of LC1 is predicted to be bound to AAA1 through the hydrophobic patch(es) with the C-terminal domain interacting with the AAA ring. The second copy of LC1 appears to associate with AAA3 or AAA4. The precise relative orientations of the components remain uncertain.
Microtubule-binding Subdomain-Mutation of single Lys residues within the LRR barrel had a pronounced effect on the ability of LC1 to interact with axonemal microtubules. With one exception (discussed further below), all these residues form a patch near the N terminus (Fig. 11a). Identification of multiple Lys residues involved in binding the A-tubule also provides an explanation for the very high yield M r 66,000 product containing LC1 and tubulin that is generated by treatment of wildtype axonemes with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (10). In effect, multiple cross-links between Lys residues on LC1 and their interacting acidic residues on tubulin are possible, any one of which generates the same dimeric product as assessed by gel electrophoresis.
We also observed that alteration of Lys 126 led to decreased microtubule binding even though this residue is located far from others that yielded a similar phenotype. This residue is adjacent to Arg 79 , which caused protein aggregation when mutated to Glu. Although K126E LC1 was stable, a possible explanation is that altering the Arg 79 /Lys 126 region causes widespread disruption to a large portion of the LC1 surface that in the case of R79E yields aggregation but for K126E results only in decreased microtubule binding affinity.
Function of LC1 C-terminal Domain-In metazoans, the C-terminal helical domain is highly charged, containing five or more acidic residues, whereas in protists (i.e. Chlamydomonas and Trypanosoma), there are also two conserved basic residues (Arg 189 and Arg 196 in Chlamydomonas) (Fig. 1b). Structural analyses indicate that the orientation of this C-terminal helix is controlled by two residues (Met 182 and Asp 185 in Chlamydomonas) that show high backbone dynamics (17). Binding of LC1 to the ␥ HC requires hydrophobic interactions, and most single residue alterations within the C-terminal domain did not affect LC1-␥ HC associations. The one exception was D185G, which resulted in a slight increase in detergent solubility of LC1 (12). Thus, it is unlikely that this LC1 region is merely acting as part of a tether attaching the motor unit to the A-tubule. Rather, given the strong phenotypic consequences observed from minor alterations, it more likely makes key contacts within the AAA domain that lead to the modulation of motor activity. This suggestion does not negate the possibility that LC1 also plays an important role in stabilizing the motor domain with respect to the A-tubule. This might ensure that the AAA ring remains a constant distance from the A-tubule within a bent region of the flagellum when the interdoublet gap increases (35) and/or prevent or modulate the ability of the AAA ring to move with respect to the axonemal long axis (36,37); this latter feature might convert the ␥ HC motor into an ATP-dependent brake that limits sliding.
Why Does N150S LC1 Cause Primary Ciliary Dyskinesia?-Asn 150 is part of the LRR consensus, and in wild type, this side chain is packed within the protein core. In humans, altering this conserved residue to Ser resulted in a partial outer arm assembly defect. Molecular modeling suggested that the N150S alteration might convert the ␣7 helix to a loop and disturb the terminal LRR fold (15). Our chemical shift mapping revealed that the perturbation is much more pervasive and extends through the last three LRRs, disrupting the core. Consequently, N150S LC1 is much less stable than wild type or any of the other mutant forms we have made. Indeed, significant precipitation occurred within the few hours at 25°C necessary to collect the 1 H-15 N HSQC spectrum, and this protein had a melting temperature that was 8°C lower than wild type. Thus, at physiological temperature in humans, it is likely that a significant fraction of this PCD-causing mutant form does not fold correctly and/or even assemble within the dynein complex. As the lack of LC1 would lead to exposure of a hydrophobic region on one HC, this mutation might therefore result in the partial failure of outer arm assembly due to indirect effects mediated through aggregation of other component(s) that are no longer stable in the absence of LC1. This scenario would be consistent with the observed reduction of outer arms in trypanosomes following LC1 RNAi (28), although similar knockdowns in planaria did not appear to affect arm assembly (29).
In conclusion, we have defined a series of functional subdomains within the LC1 component of outer arm dynein. These data provide insight into the mechanism by which this highly conserved protein modulates dynein motor function within the ciliary axoneme.