Tektins Are Heterodimeric Polymers in Flagellar Microtubules with Axial Periodicities Matching the Tubulin Lattice*

Tektins are proteins that copartition with tubulin in a stable ribbon of three protofilaments from ciliary and flagellar microtubules. After purification, tektins A, B, and C from sea urchin sperm flagellar microtubules appear as extended relatively insoluble filaments, <5 nm in diameter. We used cross-linking reagents to investigate the associa-tions and structural organization of subunits within tektin polymers isolated from stable protofilament ribbons of Strongylocentrotus purpuratus. We show by SDS-poly- acrylamide gel electrophoresis, hunoblots, and trans-mission electron microscopy that tektins are continuous heteropolymers in the stable protofilament ribbons, and thus flagellar microtubules. Our results also provide evi- dence for the arrangement of Werent tektin polypeptides within “core” filaments containing equimolar tektins Aand B. Treatment of these core filaments with bis(sulfosuccin-imidy1)suberate and with l-ethyl-3-(3-dimethylaminopro- py1)carbodiimide yielded a predominant cross-linked -106-kDa heterodimer of tektins A and B similar results were obtained by glutaraldehyde cross-linking of tektins solubilized under mild conditions. Finally, cross-linking with 3,3’-dithiobis(sulfosuccinimidylpropionate) revealed a 16 nm periodicity in isolated tektin AB filaments buffer and negatively stained. Specimens were examined in a JEOL lOOCX electron microscope operated at an acceleration voltage of 80 kV. Dimensions and spacings in negative stain images were determined by computed Fourier transforms, using tobacco mosaic virus as a standard.

Tektins A (-55 ma), B (-51 m a ) , and C (-47 kDa) are biochemically defined as the 0.5% Sarkosyl + 2 M urea-insoluble, filamentous fraction from sea urchin sperm flagellar axonemes (6). Antibodies specific for all three tektins stain intermittently along the length of all nine outer doublet microtubules fixed for immunofluorescence microscopy (10) and by immuno-EM anti-tektin antibodies label filaments protruding from the ends of pf-ribbons (8, 11). Since pf-ribbons contain a significant amount of tektins, these immunolabeling results were interpreted to suggest that tektins were longitudinal polymers associated with the stable tubulin protofilaments; however, direct evidence for this model has been lacking. Of more general interest, immunological studies have suggested the presence of tektin-like proteins in other stable microtubule organelles, including basal bodies, centrioles, centrosomes, mitotic spindles, and midbodies from species, including echinoderms, molluscs (Spisula), and cultured mammalian cells (10,(12)(13)(14).
Tektins have biochemical, immunological, and structural similarities to intermediate filament proteins but little sequence identity (15)(16)(17), and thus they are a distinct class of coiled-coil proteins. Tektin filaments are -70% a-helix (7) and yield strong a-type x-ray diffraction patterns (18). cDNA sequence analysis of tektins A and B indicates that each polypeptide chain has a central rod of approximately 310 residues composed of four major regions strongly predicted to form coiled coils (17,19). In our present study we used cross-linking reagents to analyze Sarkosyl-insoluble protofilament ribbons and isolated tektin filaments. Our results begin to elucidate the molecular arrangement of tektin AB filaments within the microtubule and are consistent with an important role for tektins in the organization of cilia and flagella. EXPERIMENTAL PROCEDURES Protein Purification-All steps were conducted at 0 4 "C, unless otherwise stated. Sperm flagellar axonemes were prepared from the sea urchin Strongylocentrotus purpuratus, as modified from Gibbons and Fronk (20). Protein determinations were made according to Lowly et al. (21). Pf-ribbons were prepared by extracting axonemes with a solution of 0.5% Sarkosyl (W. R. Grace), 10 mM Tris, 1 m M EDTA, pH 8.0, for 2 h, and centrifuging at 100,000 x g for 90 min; the pellet was reextracted with this same solution and recentrifuged to collect the pellet of purified pf-ribbons. Pf-ribbons that were not used immediately were stored in 50% glycerol, 50 m~ Tris, 1 m M EDTA, 1 m M DTT, pH 8.0, at -20 "C or -80 "C. Tektin filaments were isolated from fresh or glycerinated pfribbons by double extractions with 0.5% Sarkosyl, 2 or 4 M urea (as specified), 50 mM Tris, 1 m M EDTA, 1 m~ DTT, pH 8.0, for 1-2 h, followed by centrifugation as before. Experiments using fresh or glycerinated ribbons gave identical results. Soluble tektin protein was obtained by incubating tektin filaments in 1 mM HEPES, 0.1 mM EDTA, 10 m~ DTT, pH 8.0, for 30 min, followed by centrifugation at 300,000 x g for 10 min.
Chemical Cross-linking-Microtubule fractions were resuspended after centrifugation to a final concentration of 1 mg/ml in 10 mM HEPES, 1 mM EDTA, pH 8.0 (for BS3 and DTSSP) or in 100 mM MES, pH 6.0 (for EDC). The cross-linkers BS3, DTSSP, and EDC were obtained from Lune 3, the Sarkosyl + 2 M urea-insoluble fraction from cross-linked pf-ribbons. Lune 4 , the SarkosyVurea-insoluble fraction from crosslinked pf-ribbons after DTSSP cleavage. Samples in lunes 1 and 4 were reduced with 2-mercaptoethanol prior to electrophoresis; samples in lunes 2 and 3 were not reduced to prevent DTSSP cleavage. Lane 1 shows the completely resolved pf-ribbon polypeptides. In lune 2 the staining at the top is due to protein complexes that did not enter the gel; however, most of the a-and P-tubulin did not appear to be cross-linked into this complex. In lune 3 the SarkosyUurea-insoluble cross-linked proteins did not enter the gel. In lune 4 this same material is resolved into tektins A, B, and C after cleavage of the cross-linker.
the addition of 1 M Tris, pH 8.0, to a final concentration of 50 mw Cross-linked specimens were centrifuged as above for SDS-PAGE or applied to carbon grids for EM.
EM-In most cases samples of pf-ribbons or tektin filaments were prepared in tubes and then applied to carbon-coated grids. Grids were rinsed with 10 mM Tris, 1 mM EDTA, pH 8.0, and stained for 30 s with 1% uranyl acetate. In some cases samples were prepared in situ on grids: pf-ribbons or filaments were applied to grids and the excess material washed off. The ribbons and filaments that remained sparsely attached to the carbon film were rigorously extracted by washing rapidly with five drops of SarkosyVurea solution, incubated on a sixth drop for 1 h a t 4 "C, followed by two more drops of Sarkosyl/urea and seven drops of Tris-EDTA buffer. The grid sample was then negatively stained directly or processed for cross-linking.
Grids with samples to be cross-linked in situ were washed several times with 10 mM HEPES, 1 mM EDTA, pH 8.0. After removing excess buffer, each grid was placed on a drop of HEPES-EDTA buffer containing 1 mg/ml DTSSP (diluted from 30 mg/ml in dimethyl sulfoxide) and incubated a t room temperature for 1-2 h. Grids were then washed with Tris-EDTA buffer and negatively stained. Specimens were examined in a JEOL lOOCX electron microscope operated at an acceleration voltage of 80 kV. Dimensions and spacings in negative stain images were determined by computed Fourier transforms, using tobacco mosaic virus as a standard.

Flagellar Axonemes Are Differentially Extracted by
Sarkosyll U r e a S . purpuratus sperm flagellar axonemes can be fractionated by 0.5% Sarkosyl to yield stable ribbons of three to four protofilaments (6). Since this is the starting material for our investigations, the fractionation and SDS-PAGE profiles are show in Fig. 1; the principal polypeptides include a-and p-tubulin, tektins A, B and C, and 77 and 83-kDa polypeptides. Sarkosyl; note that one of the ribbons appears to break down, leaving a stable filament. 11, tektin ARC filaments isolated from pf-ribbons with Sarkosyl + 2 11 urea. c, cross-linked pf-ribbons, which frequently appear paired after cross-linking. d, filaments isolated from cross-linked pf-ribbons with SarkosyYurea; note that the filaments have associated globular material. e, tektin ABC filaments, first isolated and then cross-linked; note the presence of globular structures. f , the filaments in e after incubation with 50 mM DTT to cleave DTSSP; globules are no longer present on the filaments. Samples for a and c were prepared in tubes and then applied to EM grids; samples for 6 and d-fwere prepared directly on grids. Bar, 100 nm (a+.

Tektins Are Heterodimeric Polymers
The non-tektin proteins can be selectively solubilized from the pf-ribbons by inclusion of urea in the Sarkosyl solution. Extraction of pf-ribbons with Sarkosyl + 2 M urea yields filamentous preparations composed exclusively of tektins A, B, and C in equimolar amounts (9). For the purposes of our current studies, i t was necessary to further fractionate tektin filaments; we found that Sarkosyl + 4 M urea was optimally effective a t solubilizing tektin C (Fig. 1, lane 4 ) , yielding filaments composed exclusively of tektins A and B in equimolar amounts. The negative stain EM appearance of pf-ribbons and tektin filaments is discussed later.

Tektins A, B, and C Are Selectively Cross-linked within Pf-
ribbons by DTSSP-DTSSP is a homobifunctional N-hydroxysuccinimidyl ester conjugation reagent that has been used to cross-link keratins (26). It is cleaved by reducing agents. Using SDS-PAGE and EM, we examined the effects of DTSSP on pf-ribbons and on cross-linked pf-ribbons that were subsequently extracted with Sarkosyl + 2 M urea (Fig. 2). Following treatment with DTSSP, most of the non-tubulin proteins were cross-linked into complexes that were too large to be resolved by SDS-PAGE (lane 2 ); however, most of the tubulin appeared not to be cross-linked into this complex. After extraction of the cross-linked ribbons with Sarkosyl + 2 M urea (lane S), the insoluble fraction was collected by centrifugation and found to contain only high molecular weight makrial, which collld not be further separated by SDS-PAGE under nonreducing conditions. Incubation of the cross-linked SarkosyVurea-extracted material with 2-mercaptoethanol and subsequent analysis by SDS-PAGE demonstrated that the insoluble high molecular weight material was composed of tektins A, R, and C ( l a m 4 ) .
Thus, tektins A, B, and C were selectively cross-linked in the pf-ribbon by DTSSP.
Immunoblots confirmed that cross-linked tektins A, B, and C remaincd insoluble essentially in the absence of other pf-ribbon proteins (Fig. 3). Before cross-linking, the major protein components of control pf-ribbons could be recognized by well characterized monospecific antibodies (10,14,25). Pf-ribbons were cross-linked and extracted with Sarkosyl + 2 M urea; the insoluble cross-linked material was then examined by immunoblotting (following hydrolysis of the DTSSP cross-linker prior to electrophoresis) . Lanes 1-7 show that the insoluble material is composed principally of tektins A, B, and C and a small amount of cosedimenting tubulin. This amount of residual tubulin was not detected by SDS-PAGE with protein staining (see Fig. 1, lanes 3-4, and Fig. 2, lane 4). EM Analysis-The structure of pf-ribbons, tektin filaments, and the cross-linked products were examined by negative stain EM. The polypeptide composition of these specimens can be correlated to the SDS-PAGE analysis in Figs. 1 and 2. As a control, non-cross-linked S. purpurutus flagellar axonemal microtubules are fractionated by Sarkosyl into characteristic ribbons of three protofilaments (Fig. 4a). Sarkosyl + 2 M urea extraction further fractionates pf-ribbons into <5-nm diameter filaments and bundles of filaments (Fig. 4b) composed exclusively of tektins A, B, and C in equimolar amounts (cf. Fig. 1,  lane 3 ) ; we refer to these as tektin ABC filaments. The same approach was used to examine the effects of DTSSP. Fig. 4c shows that cross-linked pf-ribbons were similar to control ribbons; however, perhaps due to the cross-linking the ribbons were frequently paired. When cross-linked pf-ribbons were subsequently extracted with Sarkosyl + 2 M urea, <5-nm diameter filaments appeared that were associated with globular material (Fig. a). We initially attributed the globular material to the presence of cross-linked non-tektin proteins. This hypothesis was disproved when tektin ABC filaments were first isolated and then cross-linked with DTSSF? While untreated tektin ABC filaments appeared as bundles of smooth <5-nm filaments (Fig. 4b), DTSSP-cross-linked tektin ABC filaments again displayed, only more clearly, a series of axially repeating globules (Fig. 4). The globules disappeared after incubation of the cross-linked tektin filaments with 50 m~ DTT to cleave the DTSSP (Fig. 4f). Taken together with SDS-PAGE and immunoblot data (cf. Figs. 2 and 31, these results demonstrate directly that tektins preexist as axially continuous filaments in pf-ribbons (and thus flagellar microtubules) and were stabilized by cross-linking.
Tektin AB Filaments-To determine whether or not tektin C gave rise to the associated globules along cross-linked filaments, we took advantage of the fact that tektin C can be selectively solubilized, leaving stable tektin AB filaments (Fig.  1, lane 4, and Fig. 5). Prior to cross-linking, tektin AB filaments (Fig. 5a) had a smooth appearance similar to tektin ABC filaments (cf. Fig. 4b). After cross-linking tektin AB filaments with DTSSP, structural repeats were once again seen only more clearly (Fig. 5b), indicating that tektins A and/or B form the basis of the globular repeats. The presence of tektin C seemed to occlude or disrupt this high degree of order. The globules have a uniform diameter of -10 nm and an axial repeat of 16 tektin AB filaments. n. tcktin AB filamcrlts nppcorcd similar tn con-1%:. A K filaments ( r f . Fig. 46). h. nftcr DTSSP-cross-linking. tektin AI3 lilamcnts again displnycd axially repeating glnhulas, with diameters of -10 nm and a periodicity of 16 nm. FiItlnwn1,s li-cqurnt.ly appeared double nr intertwining. c, the 16 nm periodicity was determined from computed Fourier transforms of single straight filaments with globules, similar to those in h. Samples were prepared in tubes and then applied to grids. Bar, 100 nm (a, b). nm, as determined from computed Fourier transforms of single, straight, cross-linked filaments (Fig. 5c). Frequently there is an appearance of two filaments wound around each other or of a filament with a double spiral of globules.

. Negntivc stain EM of non-cross-linked and cro=-linkt=d trnl tektin
Tektin AB Heterodimers Are Resolved from Tektin Filaments-We cross-linked tektin AB filaments with low concentrations of DTSSP to elucidate their intermolecular structure (Fig. 6). Prior to cross-linking, tektin AB filaments were solubilized by SDS in nonreducing conditions into oligomeric complexes of -120 and -200 kDa, apparently due to pre-existing disulfide bonds. Each of these bands is composed of tektins A and B, as determined by immunoblot analysis (Fig. 6, c  and d). In the absence of lower molecular mass oligomers, the -120-kDa band would appear to represent a tektin AB heterodimer (predicted mass of 106 kDa). Incubation with DTSSP initially increased the appearance of the putative dimer while decreasing the amount of tektin A and B monomers. Again, tektins A and B were present in the presumed dimeric complex by immunoblot analysis (c and d ) , indicating that tektin AB heterodimers could be resolved from cross-linked and noncross-linked tektin AB filaments.
In the absence of a reducing agent we were unable to make accurate molecular mass determinations of tektin oligomers by SDS-PAGE. BS3 is a nonhydrolyzable N-hydroxysuccinimidyl ester cross-linking reagent that has a similar spacer arm length to DTSSP (11.4 and 12 A, respectively). Low concentrations of BS3 cross-linked tektin AB filaments into prominent -107-kDa complexes, which by immunoblot analysis contained both tektins A and B (Fig. 7). The predicted mass of tektin AB heterodimers is 106 kDa. Oligomers corresponding to predicted tektin AA or BB homodimers (110 or 102 kDa, respectively) were not observed with BS'. The higher molecular mass complexes presumably represented tektin trimers, tetramers, and higher oligomers. Repeating globular structures were not present on tektin filaments cross-linked by BS3 (data not shown). Similar results were also obtained with the zero length crosslinker EDC, which yielded a predominant dimer of 106 kDa composed of tektins A and B (Fig. 8); however, with EDC we did observe a faint band corresponding to a 102-kDa BB dimer but did not detect 110-kDa AA dimers.
To examine whether the heterodimer exists in solution under mild conditions, tektin filaments were solubilized in low salt buffer plus DTT and examined by cross-linking with glutaraldehyde (since DTT interferes with DTSSP and BS3). As shown in Fig. 9, SDS-PAGEhmmunoblotting indicated the presence of a heterodimer.

DISCUSSION
Tektins Exist as Longitudinal Filaments Associated with the Walls of Flagellar Microtubules-From the data presented it is clear that tektins are cross-linked into extended polymers composed a t least of tektins A and B and probably also tektin C. Previous results have suggested that tektins exist as filaments associated with ciliary and flagellar microtubules (8, 10, 11); however, the possibility could not be excluded that the observed "- tektin filaments were aggregation artifacts, following the extraction of flagellar microtubules with SarkosyVurea. In our current investigation we used cross-linking reagents to specifically cross-link tektins in pf-ribbons before SarkosyVurea extraction, eliminating potential extraction artifacts. The presence of individual filaments (rather than bundles) composed of tektins A, B, and C, following the cross-linking and extraction of pf-ribbons, demonstrates that tektins naturally exist as axially continuous filaments in the pf-ribbons and thus flagellar microtubules. These filaments might either be associated with certain tubulin protofilaments or they might exist as a nontubulin protofilament in the A-tubule wall (7, 8, 11).

Tektin Filaments Are Composed of Tektin AB Heterodimers-
Cross-linking and solubilization experiments would be expected to identify homo-andor heterodimers as adjacent neighbors, if they are present in the tektin filament structure.
The apparent molecular masses of tektins A and B by SDS-PAGE are 55 and 51 kDa, respectively. After cross-linking with BS3, DTSSP, and EDC, a new predominant band of 106-107 kDa appears that is stained by monospecific antibodies against tektin A and tektin B (Figs. 6-8). With DTSSP and BS', 102and 110-kDa bands, corresponding to BB and AA homodimers, respectively, are not detected; however, with EDC we detected faint BB dimers but not AA dimers. Thus, tektins A and B are cross-linked predominantly as heterodimers in the filament structure. Although these results indicated that tektins A and B are adjacent neighbors in the polymer, the question remained as to whether the cross-linked subunits represent physiological dimers. The existence of a native heterodimer was confirmed by the solubilization of tektins under mild conditions, followed by cross-linking with glutaraldehyde ( Fig. 9). Even in non-crosslinked (and nonreduced) samples there is faint evidence of the heterodimer (Fig. 6), which is presumably due to strong disulfide interactions between the A and B polypeptide chains; such disulfides may be important in the stabilization of the dimer.
Sequence comparison of tektin B to tektin A (17,19) indicated that both polypeptide chains contained four major regions predicted to form coiled coils along a central rod. Our results now confirm the earlier prediction that coiled-coil heterodimers may be more favorable than either type of homodimer (19). These results may be analogous to keratins, which are obligate heteropolymers and which exist in solution as heterodimers (26-28). In addition, higher molecular weight oligomers are also seen in cross-linked samples (Figs. 6-81, as would be expected from a polymer (28-30).

45
Tektin Filaments Have a 16 nm Axial Periodicity-The evidence indicates that the globular structures repeating along DTSSP-cross-linked tektin AB filaments arise from tektin A, B, or both (Fig. 51, but it is not entirely clear whether the globules are native structures that are preserved, or whether they are induced by the DTSSP. Such globules were not observed on non-cross-linked AB filaments stained with phosphotungstic acid instead of uranyl acetate or after rotary shadowing; the globules were also not observed on negatively stained AB filaments cross-linked with glutaraldehyde, BS3, or EDC. Thus, our data would suggest that the globules are not present on tektin AB filaments isolated from SarkosyVurea but are induced by cross-linking with DTSSP. Whether these structures are real or induced, we think that their appearance reflects an inherent structural periodicity of the tektin filament. The size of the globules (-10 nm in diameter) and their axial repeat (16 nm) are reproducible and regular (Fig. 5). This periodicity is of particular interest since it is precisely twice the 8 nm repeat of the tubulin dimer in the microtubule lattice (31-35). The 16 nm periodicity also matches the observed spacings of various microtubule-associate components (36-40) and correlates to a 48 nm spacing along tektin filaments (11). Thus, while we have not yet been able to obtain direct evidence for the interaction of tektin and tubulin, the 16 nm spacing along tektin filaments may provide a structural basis for this interaction andor for the attachment of microtubule-associated components.