X-Ray Fiber Diffraction Recordings from Oriented Demembranated Chlamydomonas Flagellar Axonemes

Abstract

The high homology of its axonemal components with humans and a large repertoire of axonemal mutants make Chlamydomonas a useful model system for experiments on the structure and function of eukaryotic cilia and flagella. Using this organism, we explored the spatial arrangement of axonemal components under physiological conditions by small-angle x-ray fiber diffraction. Axonemes were oriented in physiological solution by continuous shear flow and exposed to intense and stable x rays generated in the synchrotron radiation facility SPring-8, BL45XU. We compared diffraction patterns from axonemes isolated from wild-type and mutant strains lacking the whole outer arm (oda1), radial spoke (pf14), central apparatus (pf18), or the α-chain of the outer arm dynein (oda11). Diffraction of the axonemes showed a series of well-defined meridional/layer-line and equatorial reflections. Diffraction patterns from mutant axonemes exhibited a systematic loss/attenuation of meridional/layer-line reflections, making it possible to determine the origin of various reflections. The 1/24 and 1/12 nm⁻¹ meridional reflections of oda1 and oda11 were much weaker than those of the wild-type, suggesting that the outer dynein arms are the main contributor to these reflections. The weaker 1/32 and 1/13.7 nm⁻¹ meridional reflections from pf14 compared with the wild-type suggest that these reflections come mainly from the radial spokes. The limited contribution of the central pair apparatus to the diffraction patterns was confirmed by the similarity between the patterns of the wild-type and pf18. The equatorial reflections were complex, but a comparison with electron micrograph-based models allowed the density of each axonemal component to be estimated. Addition of ATP to rigor-state axonemes also resulted in subtle changes in equatorial intensity profiles, which could report nucleotide-dependent structural changes of the dynein arms. The first detailed description of axonemal reflections presented here serves as a landmark for further x-ray diffraction studies to monitor the action of constituent proteins in functional axonemes.

Introduction

By virtue of the x ray’s small wavelength (∼0.1 nm), the x-ray diffraction technique can potentially resolve the structure of biomolecules or their assemblies with atomic resolution. X-ray diffraction is especially useful when target molecules are periodically arranged because x rays scattered by such a molecular array interfere with each other to generate strong signals. Moreover, because all the parameters needed for analysis—such as the x-ray wavelength, specimen-to-detector distance, and pixel size of the detector—can be accurately defined, the x-ray diffraction technique is especially effective for precisely analyzing periodic structures.

In fact, the high penetration of x rays allows the technique to be applied to naturally occurring periodic structures while they remain functional in an aqueous environment, including DNA (1), actin filaments (2,3), microtubules (4), fibrous viruses (5), bacterial flagella (6), and skeletal muscles (7–9). Such data can be collected free of fixation-induced artifacts. Especially in studies of cross-bridge movement in live muscles in situ, intense and coherent synchrotron x rays have been used, and the diffraction technique has provided information about cross-bridge movements with nanometer- and millisecond precision (9). These achievements show the advantages of x-ray diffraction over another high-resolution technique, electron microscopy.

Cilia and flagella are motile organelles found in various types of eukaryotes. Their scaffold, the axoneme, is composed of very regular arrays of proteins. In the axoneme, one-dimensional movement of microtubules driven by dynein arms is converted to complex three-dimensional waveforms to cause a flow of the surrounding medium (10–12). The mechanism of this conversion is a long-standing unresolved issue, and structural information at near-atomic-level resolution in an aqueous environment is required. If the dynein arms in the axonemes change their positions three-dimensionally, their movement should be reflected in the diffraction pattern. Therefore, the axoneme is a potential target for x-ray diffraction. However, electron microscopy has been practically the only means used to obtain structural information about the axonemal components since their 9 + 2 architecture was first described in the 1950s (13), with the exception of a few pioneering x-ray diffraction studies (14).

Although the application of x-ray diffraction to axonemes has been expected to elucidate how the dynein arms move and how axonemal structures change their positions, use of this technique has been hampered by the small size of the material and by the difficulty of orientating the samples uniformly. Typically, a single axoneme is ∼150 nm in diameter and <100 μm in length, and scatters x rays too weakly for conventional techniques to record. Recently, however, intense x-ray beams generated by third-generation synchrotron radiation facilities have made it feasible to analyze axonemal structure by means of small-angle x-ray diffraction.

We employed two approaches to overcome the difficulties in sample preparation. One is to quick-freeze single axonemes and irradiate them with x-ray microbeams (diameter ∼2 μm) long enough to obtain sufficient signals. Biological materials are much more resistant to radiation damage when they are frozen at liquid nitrogen temperature (15). We successfully applied this technique to the long flagellar axoneme of Drosophila melanogaster sperm tail (16). The other approach is to use a suspension of a large number of axonemes. In this case, strong signals can be obtained without freezing the material, and thus measurements can be obtained under more physiological conditions. However, since the axonemes are randomly oriented, the resulting x-ray scattering contains only one-dimensional information.

Sugiyama et al. (17) developed a technique to align fibrous biological materials by applying shear flow to the suspension. This method very effectively aligns axonemes of sea urchin sperm tails (18) and Chlamydomonas reinhardtii flagella (19). Using this shear-flow-alignment method with a repertoire of flagellar mutants of Chlamydomonas, we sought to provide a detailed description of the x-ray reflections from Chlamydomonas flagellar axonemes and to characterize them. The use of mutants lacking specific components of the axoneme allowed us to investigate the origin of observed reflections. We also examined the effects of nucleotides simply by adding particular nucleotides to the same samples. The information provided herein is expected to serve as the basis for further x-ray diffraction studies using axonemes of eukaryotic flagella and cilia. The basic theory of diffraction from axonemes (20) and an introductory guide to applying x-ray diffraction to axonemes (21) have already been produced.

Materials and Methods

Specimen preparation

Chlamydomonas reinhardtii strains were cultured and their flagella were isolated and demembranated as described elsewhere (21). The axonemes prepared in this way were collected by centrifugation and resuspended in an HMDEKP solution containing 30 mM HEPES-K, 5 mM MgSO₄, 1 mM EGTA, 1 mM dithiothreitol, 50 mM potassium acetate, 0.5% (v/v) poly(ethyleneglycol), and 2% (w/v) methylcellulose (pH 7.4; M0512; Sigma, St. Louis, MO) at final axoneme concentrations of 5–10 mg/ml. Methylcellulose is used to greatly improve the alignment of axonemes under the shear flow (17).

Shear-flow alignment

The apparatus for shear-flow alignment was described previously (17,21). Briefly, it consisted of two round coverslips (diameter 17 mm) facing each other with a small gap (0.1–0.35 mm) between them that was filled with a suspension of axonemes (Fig. 1). The suspension remained in place because of surface tension. One of the coverslips remained stationary while the other was rotated by a motor. X-ray beams were irradiated close to the edge of the coverslips (at a distance of ∼6 mm from the center of the rotating disc; r in Fig. 1), where the shear rate was nearly at its greatest. Under this continuous shear flow, the axonemes were aligned tangential to the direction of the rotation of the disc. The number of axonemes in the volume through which the beam passed (1 × 10⁻⁸ L) was estimated to be 4−16 × 10⁴. The angular deviation of the alignment of the axonemes can be estimated from the distribution of the equatorial peak intensities to be <5°. Taking into account the beam shape (ellipsoidal, 0.3 mm (h) × 0.2 mm (v)), we chose the irradiation point at which the longitudinal axes of the axonemes were oriented vertically, as shown in Fig. 1. Therefore, the reflection on the meridian (the axis parallel to the longitudinal axis of the axoneme) appeared vertically on the x-ray detector. The equator is the axis at a right angle to the meridian.

Figure 1.

The experimental system used for the shear-flow alignment of axonemes, modified from Sugiyama et al. (17) and Oiwa et al. (21). (a and b) The x-ray beam passes through an area 6 mm off the center (r) of a pair of tubes. The suspension of axonemes (2–5 mg/ml) is placed in the gap (0.1–0.35 mm) between the two parallel discs (coverslips shown in yellow, glued onto the openings of the tubes). (c) The highlighted area schematically shows the alignment of axonemes under shear flow and in the beam. (a and d) One of the discs (a) is rotated by a DC motor and a rubber drive-belt (d). The x-ray diffraction was measured downstream through a vacuum chamber. (e) X-ray fiber diffraction patterns from the axonemes are dominated by a sharp reflection on the meridian (the axis parallel to the axonemal axis), an indication of microtubules running along the axonemal axis. The equator is the axis at right angles to the meridian.

To test the effects of nucleotides on the axonemal components in the same samples, we added a small volume of concentrated nucleotides (ATP + vanadate, ADP, or AMPPNP) to the axoneme suspension in the space between the spinning discs. By applying high shear flow (1000–5000 s^–1) to these axonemes, we found that alignment of the axonemes was accomplished within a few seconds after the start of spinning. On the other hand, the alignment gradually decayed over 1 min after cessation of the spin. Therefore, after running a single x-ray recording to check the axonemes in the absence of nucleotides, we observed the same sample under continuous shear flow after addition of a nucleotide. These features allowed us to compare directly the effects of nucleotides on the axonemal structures in the same specimen.

X-ray diffraction recording and data processing

X-ray diffraction recording was performed at the BL45XU beamline of SPring-8 (22). The setting was that used for small-angle scattering, with a specimen-to-detector distance of either 2.0 m (with an x-ray wavelength of 0.09 nm) or 3.5 m (wavelength 0.15 nm). Unless otherwise specified, the detector was a cooled CCD camera (C4880; Hamamatsu Photonics, Hamamatsu, Japan; 1000 × 1018 pixels, pixel size = 136 μm) used in combination with a 6-inch x-ray image intensifier (VP5445; Hamamatsu Photonics). Imaging plates (IPs; Fujifilm) were also used for more precise measurements. The specimen-to-detector distance used for IPs was 3.3 m (with an x-ray wavelength of 0.09 nm). Since IPs have a large area, a high dynamic range, and no distortion of the image, they are suitable for highly accurate static measurements.

The x-ray beams were not attenuated (beam flux, 2 ×10¹¹ photons × s⁻¹) and they would cause radiation damage to stationary biological samples within ∼1 s. Because of continuous rotation, however, the axonemal specimens kept moving and withstood >50 exposures of 1 s. In a single run of exposures, 50 frames were usually collected. The frames collected in this way were summed afterward, after correcting for inclination and folding the four quadrants to improve the signal/noise ratio. Background scattering was subtracted and reflection intensities were determined in the same manner as described previously for processing patterns from muscle (23,24).

Model building, calculation, and fitting

The procedures used for model building, calculation, and fitting were similar to those used for Drosophila axonemes (16). Model axonemal structures were built from electron micrographs of the cross sections of wild-type Chlamydomonas axonemes. The unit structures, each containing a single doublet microtubule with dynein arms and a radial spoke, were extracted from eight electron micrographs. The averaged unit structures were rearranged in a 9-fold rotational symmetry. After the densities of the components were changed, the model was subjected to fast Fourier transform and rotary averaged to calculate the equatorial reflection intensities. The calculated intensity profiles were compared with the observed intensity profiles to obtain the best fits. To evaluate the models, we used modified χ² test functions. Details are described in the Supporting Material.

Results

General description of the diffraction pattern from wild-type Chlamydomonas axonemes

As was previously shown for axonemes from sea urchin sperm (17,18), Chlamydomonas axonemes were aligned well by shear-flow in solutions containing 1–2% methylcellulose. Fig. 2 shows diffraction patterns from wild-type Chlamydomonas axonemes, nonmotile but relaxed in the presence of ATP and vanadate. The patterns were obtained with two different x-ray settings: one with a long specimen-to-detector distance (3.5 m) and a long x-ray wavelength of 0.15 nm (Fig. 2 A; long-camera settings), and one with a shorter specimen-to-detector distance (2 m) and a wavelength of 0.09 nm (Fig. 2 B; short-camera settings).

Figure 2.

(A and B) Diffraction patterns from flow-oriented axonemes of wild-type Chlamydomonas flagella, recorded with (A) a long specimen-to-detector distance (3.5 m) and a long x-ray wavelength (0.15 nm) (long-camera settings), sum of 50 frames (0.7 s exposure each); and (B) a short specimen-to-detector distance (2 m) and a short x-ray wavelength (0.09 nm) (short-camera settings), sum of 40 frames (0.8 s exposure each). The diffusive layer-line reflections (double arrows) were observed at 1/4 nm⁻¹, representing the first microtubule-based reflection. As shown in Fig. 1, we chose the area in which the longitudinal axes of axonemes were oriented vertically in the system. Background scattering was subtracted from both patterns as previously described (23,24).

With the long-camera settings (Fig. 2 A), details were resolved in the small-angle region, but reflections in the wider-angle region were not recorded. The diffraction pattern consists of two series of reflections: equatorial and meridional. The very intense equatorial reflections occur in the direction perpendicular to the longitudinal axis of the axonemes. Their peaks are not well separated from each other, because they directly represent the peaks of continuous Bessel functions of different orders originating from cylindrically arranged objects (20). The nine doublet peripheral microtubules make the greatest contributions to the intensities, but dynein arms, radial spokes, and other periodically arranged structures should also make some contribution. The intensity profiles of the equatorial reflections are described, modeled, and discussed in more detail below.

The weaker meridional reflections are the ladder-like reflections aligned along the meridian (corresponding to the longitudinal axis of the axonemes). When viewed along the meridian, the peaks of the meridional reflections are sharp and well-separated from each other, and are indexable to a basic axial repeat of 96 nm, i.e., the periodicity for the radial spokes and inner dynein arms (25,26). Although the first-order reflection at 1/96 nm⁻¹ is too close to the beamstop and is not clearly recognizable, reflections of up to the 12th (at 1/8 nm⁻¹; Fig. 2 A) and 24th (at 1/4 nm⁻¹; Fig. 2 B) orders are observed.

In contrast to the case with equatorial reflections, microtubules are not expected to contribute to the meridional reflections because they have little contrast of density along their longitudinal axis (20). The first microtubule-based reflection to occur is the layer-line reflection at 1/4 nm⁻¹ (labeled as 4 nm-LL and indicated by four double arrows in Fig. 2 B). In contrast, the sharp meridional reflections do not have the distinct features of layer lines, which are expected to appear on a cross passing through the origin and 4 nm-LLs. Therefore, the meridional reflections are considered to originate exclusively from periodic structures aligned along the microtubules.

On the other hand, if periodic structures are helically arranged in a whole axoneme, they are expected to generate layer-line reflections, i.e., their intensities are distributed off-meridional (intensities separate to both sides of the meridian). Although axonemal components such as the radial spokes have been reported to have a helical arrangement in Chlamydomonas (27,28) and other organisms (25), such layer-line-like features are not evident in the patterns from wild-type axonemes. As will be described later, however, such layer-line-like features are evident in some of the reflections from mutant axonemes, and it is clear that at least some of the axonemal components have helical symmetries when seen in a whole axoneme.

With the short-camera settings, a wider scattering angle is covered but individual peaks of reflections, especially those of equatorial reflections, may not be separated well (Fig. 2 B). In this pattern, a weak but sharp meridional reflection occurs at 1/4 nm⁻¹, and this is considered to be the 24th-order reflection from the 96 nm basic repeat. Fig. 2 B also shows a diffuse layer-line reflection at 1/4 nm⁻¹. This is considered to be the first-order reflection from the helical arrangement of tubulin monomers within microtubules. The same layer-line reflection has been observed for purified oriented microtubules (4) (Kamimura et al., manuscript in preparation).

Effect of ligand binding

Fig. 3 compares diffraction patterns from axonemes of wild-type Chlamydomonas flagella in rigor, i.e., in the absence of nucleotide (Fig. 3 A), in the presence of ATP and inorganic vanadate (Fig. 3 B), and in the presence of AMPPNP (a nonhydrolyzable analog of ATP) (Fig. 3 C). The most obvious effect of adding ATP and vanadate to axonemes in rigor is the dramatic improvement in the axonemal orientation, as evidenced by the reduced arcing of meridional reflections, reduced spread of equatorial reflections, and intensification of reflections in the higher-angle region. AMPPNP had a similar effect (Fig. 3 C).

Figure 3.

(A–C) Effects of nucleotides on the patterns from axonemes of wild-type Chlamydomonas flagella in (A) the absence of nucleotide, (B) the presence of 1 mM ATP and 100 μM vanadate, and (C) the presence of 3 mM AMPPNP. Recorded with short-camera settings.

Since we obtained diffraction patterns from the same axoneme suspensions before and after the addition of nucleotides, we can directly compare the effects of the nucleotides on the axonemal components and extract information about structural changes in the dynein arms and their resultant effects on the gross axonemal structure. However, interpretation of these diffraction data is hampered by difficulties in measuring precise diffraction intensities, especially of meridional reflections, owing to overlapping of the off-meridional peaks derived from a helix with a large radius (see below). In addition, because of the change in the extent of orientation (as evidenced by the arcing of meridional reflections), a precise assessment of the effect of nucleotides on the integrated intensities of individual meridional reflections would require further detailed experiments. On the other hand, nucleotide binding also caused a small but consistent effect on the intensity profile of the equatorial reflections, as will be described later, and this may report structural changes in the dynein arms.

Variation of diffraction patterns from axonemes lacking specific components

An advantage of using axonemes of Chlamydomonas flagella is the availability of a variety of mutants lacking specific axonemal components. Here we recorded diffraction patterns from axonemes of oda1 (lacking the whole outer dynein arm), oda11 (lacking the α-heavy chain of the outer arm dynein), pf14 (lacking the radial spokes), and pf18 (lacking the central apparatus (CA)). These patterns are shown in Fig. 4 along with that of the wild-type.

Figure 4.

Diffraction patterns from axonemes of wild-type and mutant strains of Chlamydomonas. (A) Wild-type. (B) oda11 (lacking the α-heavy chain of the outer dynein arm). (C) oda1 (lacking the whole outer dynein arm). (D) pf14 (lacking the radial spokes). (E) pf18 (lacking the CA). All of the patterns were recorded in the presence of 1 mM ATP and 100 μM vanadate. Blue arrows, the 4th and 8th (of 96 nm repeat) meridional reflections, to which the outer dynein arms are considered to make major contributions. Green arrows, the 3rd and 7th meridional reflections, to which the radial spokes are considered to make major contributions. Magenta arrows, the 6th and 12th meridional reflections, to which the inner dynein arms and/or the dynein regulatory complex are considered to make major contributions. Recorded with short-camera settings.

When compared with the wild-type axonemes (Fig. 4 A), the 1/24 nm⁻¹ (fourth of the 96 nm repeat) and the 1/12 nm⁻¹ (eighth) meridional reflections (blue arrows) are progressively weaker in oda11 (lacking the α-heavy chain of the outer dynein arm; Fig. 4 B) and oda1 (lacking the whole outer dynein arm; Fig. 4 C) mutants. These data indicate that the outer dynein arms, which are known to have a 24 nm repeat, make major contributions to these reflections. However, the reflections do not completely disappear in oda1, meaning that the radial spokes and/or the inner dynein arms also make some contribution. Interestingly, the intensity of the 1/24 nm⁻¹ reflection is apparently weaker on the meridian than in its off-meridional parts, suggesting that these structures are helically arranged in the axonemes.

In the patterns from pf14 (Fig. 4 D), which lacks the radial spokes, the third (1/32 nm⁻¹) and seventh (1/13.7 nm⁻¹) reflections (green arrows) are visibly weaker, meaning that the intensities of these reflections mainly come from the radial spokes. The pattern from pf18 (Fig. 4 E), which lacks the CA, is very similar to that from the wild-type, indicating that the contribution of the CA to diffraction patterns is limited. In the presence of nucleotide, under shear force, the orientation of pf18 axonemes tends to be better than that of wild-type axonemes. This suggests that the CA has some influence on the stiffness of axonemes and possibly a role in bend formation.

The 6th (1/16 nm⁻¹) and 12th (1/8 nm⁻¹) reflections (Fig. 4, magenta arrows) are very stable and are consistently observed for the mutants as well as for the wild-type. Therefore, these reflections may come from the inner dynein arms and/or the dynein regulatory complex, which is also known to have a 96 nm repeat (27). The meridional part of the second reflection (1/48 nm⁻¹, cyan arrow) becomes weaker in pf14, but its off-meridional part is clearly visible. Its appearance is like a widespread layer line, suggesting that it comes from a helix with a small diameter. However, it is unlikely that it comes from a helical structure within the CA, because the layer line is also observed in pf18. Therefore, this layer line may also originate from inner dynein arms and/or the dynein regulatory complex.

Axial arrangements of the axonemal components on the doublet microtubules

We completed a further series of experiments using IPs. Because of their large area, wide dynamic range, and no distortion of the image, IPs allow us to measure peak positions with high accuracy. The precise positions of the major meridional reflections are summarized in Table 1. Since the first-order and second-order reflections at 1/96 nm⁻¹ and 1/48 nm⁻¹ are too close to the beamstop under the condition used here (a specimen-to-detector distance of 3.3 m and an x-ray wavelength of 0.09 nm), we focused on the 3rd, 4th, 6th, and 12th reflections. We calculated the peak positions and the full width at half-maximum (FWHM) of the peaks by fitting Lorentzian curves to each peak. The peaks appear at integer multiples of 8.24 nm, indicating that the axonemal components are arranged on microtubule lattices with an axial repeat of 8.24 nm, which is similar to the value reported in the structural unit of a microtubule, the α-β tubulin heterodimer, by cryo-electron microscopy (29). Furthermore, small variations in the peak positions and the sharpness of the peaks (small FWHMs) between the wild-type and mutants’ diffractions show that most axonemal components are arranged precisely along the microtubule lattice repeat, but the arrangement of the axonemal components is not significantly disturbed by mutations that cause the lack of some axonemal components. To interpret the results of mutant studies regarding flagellar motility, we should assume structural integrity: the mutations do not lead to severe and global structural perturbation of an axoneme (30). The x-ray diffraction patterns of the mutants we used here show that the structure of the axoneme, as expected, is invulnerable to mutations. This feature confirms the validity of mutant studies of the mechanism of flagellar beating.

Table 1.

Diffraction maxima positions in real space

Reciprocal Space (nm−1)	Wild-Type		oda1 (Outer Armless)		pf14 (Radial Spokeless)
	Peak (nm)	FWHM (nm)	Peak (nm)	FWHM (nm)	Peak (nm)	FWHM (nm)
1/32 (3rd)	32.97	1.00	32.93	1.72	N.A.	N.A.
1/24 (4th)	24.74	0.52	24.67	0.96	24.74	0.46
1/16 (8th)	16.49	0.34	16.46	0.44	16.40	0.42
1/8 (12th)	8.28	0.12	8.25	0.14	8.26	0.08

The meridional reflections were recorded with IPs. To achieve better alignment of the axonemes in a shear flow, the measurements were obtained in the presence of 1 mM ATP and 100 μM vanadate, with the long-camera settings (L = 3285 mm, λ = 0.09 nm). The Lorentzian peak curve was fitted to the intensity profiles of individual peaks, and the peak position and FWHM of the peak were obtained.

Analysis of equatorial reflections

The equatorial reflections from an ensemble of axially oriented axonemes represent a rotary-averaged profile of the Fourier transform of the cross section of the axoneme (16) and are described as the sum of Bessel functions of various orders (20). In contrast to meridional reflections, which can be approximated as discrete δ functions, Bessel functions are continuous and their peaks are not separated. The microtubules are expected to make the largest contributions to their intensities, but other axonemal components may modulate their intensity profiles. The intensity profile is so complex that, at present, comparisons between observed and model-calculated profiles are the only practical means of analysis.

Comparisons between mutants

Fig. 5 shows the intensity profiles of the equatorial reflections from axonemes of wild-type flagella and three mutant Chlamydomonas strains, in the presence of ATP and vanadate, recorded with the long-camera settings. The profiles are similar to the rotary-averaged profiles from the end-on diffraction patterns of axonemes of Drosophila sperm (16). In the smallest-angle region (>50 nm), three intense peaks are observed. The spacing for the innermost observable peak (second peak of J₀, the 0th-order Bessel function) is beyond 100 nm, and its profile could not be determined accurately because of the beamstop. The intensity ratio of the second and third peaks, the mixture of J₀ and J₉, varied between mutants. These are the lowest-resolution peaks and are unlikely to be affected by structural details, and therefore may be used to estimate the diameter of the axonemes.

Figure 5.

Comparison of the intensity profiles of equatorial reflections between the wild-type and various mutants. Gray, wild-type; blue, oda1; red, pf14; green, pf18. All of the profiles were recorded in the presence of 1 mM ATP and 100 μM vanadate. Intensities are in an arbitrary unit. Inset: profiles in the higher-angle region (d < 40 nm) are also shown at 10× magnified intensity on the same axis of d-spacing. Data were recorded with the long-camera settings. Due to the presence of the beamstop, the profile of the innermost peak is not correctly represented.

This group of intense peaks is followed by a group of weaker peaks at 12–30 nm, where even higher-order Bessel functions (J₁₈+) are involved. This group of peaks should reflect finer details of the axonemal structure, but the peak compositions are very similar between different mutants. This suggests that the microtubules are still the major determinants of intensity profiles in this region of spacing. Depending on the particular mutant, a small peak is observed between these two groups at ∼40 nm.

Comparison with models

As was previously done for the profiles of Drosophila sperm axonemes (16), we compared the observed intensity profiles with those calculated from a model based on electron micrographs of Chlamydomonas flagella. Fig. 6 shows how the model was built. The CA was omitted from the model, which has a 9 + 0 structure. The nonmicrotubule parts were divided into three areas (outer dynein arms, inner arms, and radial spokes), and their densities independently varied in six steps (0%, 40%, 80%...200%). In this way, a total of 6³ = 216 models were generated, and the best model was sought to reproduce each of the observed intensity profiles.

Figure 6.

Model building from electron micrographs of axonemes from Chlamydomonas flagella. (A) One of the micrographs used for modeling. (B) An average structural unit of the axoneme, taken from eight micrographs, consisting of a doublet microtubule, dynein inner and outer arms, and a radial spoke. The circle drawn in yellow connects the centers of doublet microtubules, the diameter of which represents that of the axoneme. (C) A 9 + 0 model structure of the axoneme, in which nine units shown in B are arranged in a ninefold rotational symmetry. The CA is not included in the model. (D) The Fourier transform (structure factor) of the model axoneme in C. An equatorial intensity profile was obtained by squaring and rotary averaging this pattern. Scale bar in A–C, 100 nm.

The results of the fitting are shown in Fig. 7. For the wild-type axoneme (Fig. 7 A), good fittings are obtained if both the outer and inner arms of dyneins have densities. An unexpectedly low density is assigned to the radial spokes, and good fittings are obtained even if there are no radial spokes. In the fittings shown in Fig. 7, the peaks in the higher-angle region (<30 nm) are generally reproduced well along with the strong peaks in the low-angle region. We evaluated the likelihood of the combination of densities by using the sum of squared residuals. The error surface was not bumpy but had a global minimum with limited combinations of densities (see Fig. S2).

Figure 7.

Comparison between observed equatorial intensity profiles and those calculated from the best-fit models. Black and gray curves represent the observed and calculated intensity profiles, respectively. Profiles in the higher-angle region (d < 40 nm) are also shown magnified ×10 (inset boxes). (A) Wild-type. In the best fit model, the densities of the inner dynein arm, outer arm, and radial spokes are 120%, 120%, and 0% of the density of the microtubule, respectively. The axonemal diameter is estimated to be 191.3 nm (diameter of the circle connecting the centers of doublet microtubules shown in Fig. 6B). (B) oda1. The densities of components in the best-fit model are 80%, 0%, and 0% (same order as in A). The estimated axonemal diameter is 192.2 nm. (C) pf14. The densities of components in the best-fit model are 80%, 40%, and 80%. The estimated axonemal diameter is 192.9 nm. (D) pf18. The densities of components in the best-fit model are 80%, 40%, and 40%. The estimated axonemal diameter is 191.7 nm. All of the observed curves were recorded in the absence of nucleotide with the long-camera settings.

For the whole outer-dynein armless mutant, oda1, a model that lacks the outer arm, gave a good fit (Fig. 7 B). The observed profile for pf14 (lacking the radial spokes) is noisier than the others, and its fit to the models is poorer than that observed for the other mutants (Fig. 7 C). The best-fit model for pf14 had some density in the radial spokes, but this would be expected to disappear if better-quality diffraction patterns were available. For the CA-lacking mutant, pf18, a model that had densities in all of the three nonmicrotubule components produced a good fit (Fig. 7 D).

Effect of nucleotides

Besides their effects on axonemal orientations, nucleotides also affect the intensity profiles of the equatorial reflections. In oda1, small reciprocal intensity changes in the second and third low-angle peaks (with the long-camera settings) are observed upon addition of ATP and vanadate (Fig. 8 A). In the higher-angle region, the peaks at ∼25 and ∼15 nm tend to increase relative to the peaks in the middle (∼18 and ∼22 nm). Addition of AMPPNP causes smaller changes in the equatorial intensity profiles (Fig. 8 B). These changes may reflect a detachment of microtubule-binding sites of dynein arms from the adjacent microtubules or nucleotide-induced structural changes. Although these changes are not evident in other mutants with the long-camera settings, similar changes in the higher-angle region are observed in pf14 and pf18 with short-camera settings (Fig. 8, C and D).

Figure 8.

Effects of nucleotides on the equatorial intensity profiles. Black and gray curves represent the intensity profiles recorded in the absence and presence of nucleotides, respectively. Profiles in the higher-angle region (d < 40 nm) are also shown magnified 10× (inset boxes). (A) Effect of 1 mM ATP and 100 μM vanadate on the intensity profiles from oda1. (B) Effect of 3 mM AMPPNP on oda1. (C) Effect of 1 mM ATP and 100 μM vanadate on pf14. (D) Effect of 1 mM ATP and 100 μM vanadate on pf18. The curves in A and B were taken with the long-camera settings and the rest were taken with the short-camera settings.

Discussion

Here we have provided, to the best of our knowledge, the first detailed description of x-ray diffraction patterns from axonemes of eukaryotic flagella. We demonstrated that by using the technique of shear-flow alignment, one can record strong x-ray diffraction signals from suspensions of these submicrometer-sized organelles. Although many exposure frames were summed to improve the signal/noise ratios, a single exposure time was typically <1 s, and in this single frame, major features from axonemal structures are already evident. For a single experiment, only an ∼200 μl suspension of ∼1 mg axonemes was typically required, from which ∼50 exposure frames were recorded without causing radiation damage. These observations demonstrate that x-ray diffraction from a suspension of flow-aligned axonemes is a practical means to analyze their structures in the native state.

Comparison with other techniques

One of the technologies for investigating structural biology that has advanced rapidly in recent years is cryo-electron tomography. Three-dimensional information is obtained in real space by tilting the sample, embedded in a thin layer of vitreous ice, within an electron microscope. Using this technique, investigators have been able to obtain details about the three-dimensional architecture of axonemes in sea urchin sperm tail (27) and Chlamydomonas flagella (31–33) (the spatial resolution currently available is ∼3.5 nm). In recent studies, individual inner dynein arm molecules and the dynein regulatory complex were resolved (32). However, images obtained with conventional electron microscopy are essentially snapshots of specimens at the moment of fixation.

The practicality of directly determining three-dimensional structures of noncrystalline biological materials using x-ray diffraction patterns is still being assessed (34). Therefore, in this study we relied on the conventional method of analyzing x-ray diffraction patterns in reciprocal space. The interpretation of signals in reciprocal space is not always straightforward and requires knowledge about the theory of diffraction. Nevertheless, x-ray diffraction has advantages over cryo-electron tomography in that it allows one to extract structural information from unfixed biological materials that remain functional in an aqueous environment. Because of the noninvasive nature of the technique, dynamic or time-resolved measurements are possible, and the structures of a single biological sample can be compared before and after experimental intervention. In fact, many of the data presented here were taken from the same axonemal suspensions before and after the addition of nucleotides (Fig. 8). In preliminary experiments, the time course of a structural change of axonemes was followed after flash photolysis of caged ATP with a 3.4 ms time resolution (unpublished results).

Another advantage of x-ray diffraction is that the physical dimensions of samples in their native state, such as the long-range average repeat of periodically arranged monomers (e.g., outer dynein arms and radial spokes) and axonemal diameters, and their variations under different experimental conditions, may be estimated accurately. In this study, we estimated the diameters of axonemes from various mutants by fitting the observed intensity profiles of equatorial reflections to those calculated from models (Fig. 6). Furthermore, by using electron microscopes, previous studies showed that high Mg²⁺ tends to stabilize rigor bridges (35) and the Mg²⁺ concentration may have a strong effect on the diameter of the axoneme (36). By examining the diffraction patterns at low (<1 mM) Mg²⁺, we can determine whether the reflections attributed to the dynein arms undergo a change and evaluate the effect of Mg²⁺ on the structure.

Interpretations of meridional reflections

In reciprocal space, signals that originate from structures with the same spatial frequencies overlap with each other. To correctly interpret x-ray diffraction data, therefore, it is important to identify the structure that gives rise to a particular reflection, and if the reflection originates from more than one structure, to determine the relative contributions of those structures. For example, the outer dynein arms are known to have a 24 nm periodicity along the axonemal axis, and thus are expected to give rise to a series of meridional reflections indexable to that repeat. At the same time, however, these reflections can be higher-order reflections indexable to, for instance, a 96 nm repeat of the radial spokes, or the inner dynein arms. In this study, we solved this problem by using various Chlamydomonas mutants that lack components of the axoneme.

The comparison of diffraction patterns from wild-type, oda11, and oda1 (lacking a part of or the entire outer dynein arm) demonstrates that most of the intensity of the meridional reflection at d = 24 nm comes from the outer dynein arm. This d-spacing also corresponds to the fourth order of the 96 nm repeat of the radial spokes or inner dynein arms, and the diffraction pattern from oda1 axonemes shows that this reflection is relatively weak. An important observation is that this reflection is split in the middle, suggesting a helical arrangement of the radial spokes or inner dynein arms. At present, there is no obvious sign that the signal from the outer dynein arm is split in the middle, meaning that the outer arms may not be helically arranged in the axoneme. The split of reflections is currently observed only in lower-order reflections (the second and third orders of the 96 nm repeat), suggesting that the helical order of the radial spokes or inner dynein arms is not very regular.

Although they have the same basic repeat of 96 nm, the radial spokes and inner dynein arms (and/or the dynein regulatory complex) seem to make different contributions to individual meridional reflections. The radial spokes make greater contributions to the 3rd and 7th meridional reflections, whereas the inner dynein arms (and/or dynein regulatory complex) seem to contribute more to the 6th and 12th reflections. These differential contributions of different components are considered to reflect the difference in their structure factors (Fourier transforms of these components calculated along the axonemal axis) because the observed intensities of the meridional reflections are expressed as products of the structure factors and the periodicity function (Laue function).

Having established the relative contributions of various axonemal components to the respective meridional reflections, one may, in principle, obtain information about structural changes of these components upon addition of ligands and other biologically active substances by analyzing changes in the intensities of these reflections. However, these ligands also affect the alignment of the axonemes, which in turn substantially affects the apparent intensities of the reflections. The primary cause of inferior alignment in the absence of nucleotides, as observed in this work, is considered to be the curvature of axonemes. Establishing procedures to reduce this curvature will improve the outcomes of future studies.

Interpretations of equatorial reflections

As stated above, the doublet microtubules make the greatest contributions to the intensities of equatorial reflections. Unlike meridional and layer-line reflections, whose intensities are discrete (δ) functions when viewed along the meridian, the intensities of equatorial reflections are continuous Bessel functions and it is not meaningful to discuss the origin of each peak. However, this study demonstrates that one can extract structural information about dynein and other nonmicrotubule components by using the method applied to the Drosophila axoneme (16), i.e., by comparing the intensity profiles calculated from models with experimentally observed profiles. By using models derived from electron micrographs of Chlamydomonas axonemes, we reproduced peaks of the higher-angle regions (d = 15–30 nm) as well as the innermost stronger peaks when we assigned proper mass to each axonemal component (Fig. 6). The fitting procedure correctly assigned the relative masses of the dynein outer and inner arms. The best-fit curve for oda1 (which lacks the whole outer dynein arm) assigned no mass for the outer arm, whereas for the wild-type and other mutants, the best-fit curves assigned masses to both the outer and inner arms.

An unexpected result is that the mass for the radial spokes was not well assigned. For the best-fit models for the wild-type and oda1, no mass was assigned for the radial spokes, although some of the meridional reflections were identified as being of radial-spoke origin. This contrasts with the case of intact Drosophila axonemes (not demembranated), in which mass was clearly assigned for the radial spokes (16). This difference may arise from different masses of individual radial spokes in the two organisms: in electron micrographs, the radial spokes are clearly visible for Drosophila (37), whereas the staining is much weaker for Chlamydomonas (Fig. 5). At present, it is unclear whether there is any difference in the constituent proteins of radial spokes in the two organisms, but in Chlamydomonas at least 23 radial spoke proteins have been identified (38), and whole-genome information is available for Drosophila. Therefore, direct comparison of masses will be possible in the near future when radial spoke proteins have been identified in Drosophila.

The fitting procedure described above also yields estimates of the axonemal diameter. Such an approach has already been used for Drosophila axonemes (16). However, in this study, peaks in the higher-angle regions (d = 15–30 nm) were also used for fitting, and therefore the estimates should be more accurate. The estimated axonemal diameters (of the circles connecting the center of doublet microtubules) were ∼190 nm, which is close to the estimates from electron micrographs.

Although the effects of ligand binding (e.g., ATP) are expected to be more subtle than the effects of missing specific axonemal components in mutants, the equatorial reflections seem to be affected by nucleotide binding (Fig. 7). If these changes were caused by improved axonemal alignment, the intensities would simply increase with increasing scattering angle (because intensities were integrated within a narrow strip with a constant width); however, the changes actually observed are more complex. Therefore, the observed changes may reflect nucleotide-induced detachment of the dynein stalk from the neighboring microtubules and/or a conformational change of dynein molecules as a whole. Future projects will address modeling of these structural changes.

Conclusions

We have described x-ray diffraction patterns from flow-aligned flagellar axonemes prepared from wild-type and mutant strains of Chlamydomonas. By comparing the patterns from different strains, we were able to identify the origins of many reflections. The information provided here will be useful for pursuing the dynamic aspects of axonemal structures in future studies.

Author Contributions

S.T., H.I., S.K., and K.O. designed and performed experiments and analyzed data. I.H. and K.O. wrote the article.

Editor: Jennifer Ross.