Specific cross-linking of the SH1 thiol of skeletal myosin subfragment 1 to F-actin and G-actin.

Recently, we reported that (maleimidobenzoyl)-G-actin (MBS-G-actin), which was resistant to the salt and myosin subfragment 1 (S-1) induced polymerizations, reacts reversibly and covalently in solution with the S-1 heavy chain at or near the strong F-actin binding region [Bettache, N., Bertrand, R., & Kassab, R. (1989) Proc. Natl. Acad. Sci. U.S.A. 86, 6028-6032]. Here, we have readily converted the MBS-G-actin into MBS-F-actin in the presence of phalloidin and salts. The binding of S-1 to the two actin derivatives carrying on their surface free reactive maleimidobenzoyl groups was investigated comparatively in cross-linking experiments performed under various conditions to probe further the molecular structure of the actin-heavy chain complex before and after the polymerization process. Like MBS-G-actin, the isolated MBS-F-actin, which did not undergo any intersubunit cross-linking, bound stoichiometrically to S-1, generating two kinds of actin-heavy chain covalent complexes migrating on electrophoretic gels at 180 and 140 kDa. The relative extent of their production was essentially dependent on pH for both G-and F-actins. At pH 8.0, the 180-kDa species was predominant, and at pH 7.0, the amount of the 140-kDa adduct increased at the expense of the 180-kDa entity. The cross-linking of MBS-F-actin to S-1 led to the superactivation of the MgATPase substantiating the ability of this derivative to stimulate the S-1 ATPase as the native protein.(ABSTRACT TRUNCATED AT 250 WORDS)     

Temperature-induced changes in the flexibility of the loop between SH1 (Cys-707) and SH3 (Cys-522) in myosin subfragment 1 detected by cross-linking

The ability of dibromobimane to cross-link SH1 (Cys-707) in the 21-kDa C-terminal segment to SH3 (Cys-522) in the 50-kDa middle segment of the myosin S1 heavy chain has been examined as a function of nucleotide binding and temperature. The results obtained indicate that, while the reagent rapidly reacts with SH1 at both 25 and 4 degrees C, its ability to cross-link to SH3 is highly dependent on temperature. At 25 degrees C, substantial cross-linking from monofunctionally labeled SH1 to SH3 occurs, in agreement with recent work of Mornet, Ue, and Morales (1985, Proc. Natl. Acad. Sci, USA 82, 1658-1662) and of Ue (1987, Biochemistry 26, 1889-1894) and with their conclusion that a loop, allowing SH1 and SH3 to reside at the cross-linking span of dibromobimane, preexists in the protein. At 4 degrees C, however, negligible amounts of cross-linking are observed whether or not a nucleotide is present, despite indications that SH1 is labeled rapidly by the reagent at this temperature. The inability to form this cross-link is not due to an alternate cross-link between monofunctionally labeled SH1 and another thiol in the 21-kDa segment. These results indicate that this loop exists at 25 degrees C and does not exist (or exists only transiently) at the lower temperature.     

Spatial proximity of the glycine-rich loop and the SH2 thiol in myosin subfragment 1

Subfragment 1 (S1) prepared from rabbit skeletal muscle myosin was digested with trypsin to cleave the 95K heavy chain into three pieces, i.e., the 23K, 50K, and 20K fragments. The trypsin-treated S1 was then cross-linked with p-nitrophenyl iodoacetate. The cross-linker bridged one of the reactive thiols (SH2) in the 20K fragment and a lysine residue in the 23K fragment [Hiratsuka, T. (1987) Biochemistry 26, 3168-3173]. Location of the lysine residue was mapped along the 23K fragment by "end-label fingerprinting", which employed site-directed antibodies against the N-terminus of the 23K fragment and against the C-terminus of the 24K fragment (the 23K fragment plus nine extra residues at its C-terminus). The mapping revealed that Lys-184 or Lys-189 was the residue cross-linked with SH2. Since the cross-linker used here spans only several angstroms, the result indicates that Lys-184 or Lys-189 is very close to SH2 in the three-dimensional structure of myosin head. Examination of the primary structure of the 23K fragment has revealed that these lysine residues are in and very close to the so-called "glycine-rich loop", whose sequence is highly homologous to those of nucleotide-binding sites of various nucleotide-binding proteins.     

Electron microscopic visualization of the SH1 thiol of myosin by the use of an avidin-biotin system

One of the reactive thiols in the myosin head, SH1, was covalently labeled with a biotin derivative, N-iodoacetyl-N'-biotinylhexylenediamine. When 50% of the SH1 thiol was modified with the biotin reagent as judged from measurements of ATPase activities, the biotinylated myosin bound one mole of avidin per mole of myosin at the saturating level. The avidin-myosin complex was readily formed in the presence of MgADP or MgATP. Peptide maps of the biotinylated myosin revealed that SH1 is actually the site of biotinylation with N-iodoacetyl-N'-biotinylhexylenediamine. Electron microscopic examination of the avidin-myosin complex showed that the attachment site of avidin on the myosin head is 130 A from the head-rod junction, indicating that the SH1 thiol is located there.     

Thiols of myosin. IV. "Abnormal" reactivity of S1 thiol and the conformational changes around S2 thiol.

The flexibility of the tertiary structure around the active site of myosin ATPase [EC 3.6.1.3] was studied using the reactivity of two specific thiol groups, S1 and S2, as a structural probe. The following four maleimide derivatives were used as thiol-directed reagents: N-ethylmaleimide (NEM), N-(4-methoxy-2-benzimidazolyl methyl) maleimide (MBM), N-(p-(2-benzimidazolyl)phenyl)maleimide (BIPM) and N-(4-dimethyl-amino-3,5-dinitrophenyl)maleimide (DDPM). 1. All the maleimide derivatives used activated the Ca2+-ATPase activity and inhibited the EDTA-ATPase activity, like NEM, indicating that they modified S1. The rate of modification of S1 by NEM and BIPM increased with increasing pH, while that by DDPM decreased. BIPM simultaneously modified S1 and S2. 2. S1 showed much higher reactivity toward the maleimides, except for BIPM, than did N-acetylcysteine (N-Ac-Cys) a low molecular-weight model compound. The extremely small pKa value of S1, 6.28, accounted for this high reactivity. In addition, the ATP-induced increase in its reactivity inducated that S1 was in a buried state. Kinetic analysis showed that the teritiary structure around S1 at alkaline pH differed from that at acidic pH. 3. The apparent rate constant of S2-modification with NEM was approximately one seven-hundredth and one four-hundredth of those of S1 and N-Ac-Cys, respectively. Fluorimetric studies using BIPM revealed that S2 in the buried state was exposed upon adding ATP; this was compensated by the burying of some other thiol group(s) (Sp). Non-linearity of the Arrhenius plots of the reaction rate of S2 suggested that the S2 region of myosin had different conformations at high and low temperatures, the transition temperature being 10--15degrees. This non-linearity completely disappeared in the presence of Mg2+-ATP. On the other hand, Arrhenius plots for the thiols reactive to BIPM did not show non-linearity in the presence or absence of ATP.     

The MgADP-induced decrease of the SH1-SH2 fluorescence resonance energy transfer distance of myosin subfragment 1 occurs in two kinetic steps

The fluorescence resonance energy transfer distance between 5-[2-[iodoacetyl)amino)ethyl]aminoaphthalene-1-sulfonic acid covalently attached to the SH1 thiol of myosin subfragment 1 as the energy donor and N-(4-dimethylamino-3,5-dinitrophenyl)maleimide attached to SH2 as the energy acceptor has been found to decrease by about 7 A in the presence of MgADP (Dalby, R. E., Weiel, J., and Yount, R. G. (1983) Biochemistry 22, 4696-4706; Cheung, H. C., Gonsoulin, F., and Garland, F. (1985) Biochim. Biophys. Acta 832, 52-62). Fluorescence stopped-flow experiments on the same system have yielded biphasic traces which are resolvable into a fast and slow component, k1 and k2, respectively. Results of experiments in which k1 and k2 were measured as a function of excess ADP concentration showed: 1) a nonlinear dependence of the apparent rate constants on [ADP]; 2) k1 is a factor of 10 faster than k2. These results are consistent with the 3-step mechanism previously proposed for nucleotide binding to myosin S1 (Garland, F., and Cheung, H. C. (1979) Biochemistry 18, 5281-5289). Kinetic experiments in which the anisotropy of the donor was monitored show this quantity to be unchanged over the course of the reaction. The biphasic decrease of donor intensity is assigned to an increase in energy transfer efficiency which, from the above results, is due to a decrease in donor-acceptor distance, occurring in two steps. The fast step is associated with a 4-5-A decrease of the donor-acceptor separation, while the slow step is associated with a further decrease of approximately 2 A.     

Structural changes in myosin subfragment 1 by mild denaturation and proteolysis probed by antibodies

The perturbations in the structure of myosin subfragment 1 (S1) by mild denaturation or proteolysis were investigated by measuring the inhibition of the binding of antibodies to immobilized S1 by treated S1 in a solution-phase competitive immunochemical assay. The structural changes in S1 were probed by using anti-50-kDa segment, anti-N-terminus, anti-27-kDa segment, and anti-A1 light chain monoclonal antibodies (MAbs). Methanol and heat denaturation increased MAb binding to the 50-kDa segment. MAb binding to regions in the 27-kDa segment was also promoted, slightly by methanol and more drastically by heat. Proteolysis also induced structural alterations in 50- and 27-kDa segments as shown by increased MAb binding to these regions in cleaved S1. These results indicate that mild denaturation and proteolysis induce structural perturbations which alter the epitope accessibility in 50- and 27-kDa segments of S1 and that antibody binding studies afford a sensitive probe to such perturbations.     

Actin and temperature effects on the cross-linking of the SH1-SH2 helix in myosin subfragment 1

Past biochemical work on myosin subfragment 1 (S1) has shown that the bent alpha-helix containing the reactive thiols SH1 (Cys(707)) and SH2 (Cys(697)) changes upon nucleotide and actin binding. In this study, we investigated the conformational dynamics of the SH1-SH2 helix in two actin-bound states of myosin and examined the effect of temperature on this helix, using five cross-linking reagents that are 5-15 A in length. Actin inhibited the cross-linking of SH1 to SH2 on both S1 and S1.MgADP for all of the reagents. Because the rate of SH2 modification was not altered by actin, the inhibition of cross-linking must result from a strong stabilization of the SH1-SH2 helix in the actin-bound states of S1. The dynamics of the helix is also influenced by temperature. At 25 degrees C, the rate constants for cross-linking in S1 alone are low, with values of approximately 0.010 min(-1) for all of the reagents. At 4 degrees C, the rate constants, except for the shortest reagent, range between 0.030 and 0.070 min(-1). The rate constants for SH2 modification in SH1-modified S1 show the opposite trend; they increase with the increases in temperature. The greater cross-linking at the lower temperature indicates destabilization of the SH1-SH2 helix at 4 degrees C. These results are discussed in terms of conformational dynamics of the SH1-SH2 helix.     

Nucleosomal structure as probed by H3 histone thiol reactivity

Two H3 histone variants are found in equal amount in HeLa cells, and they have been characterized by two-dimensional gel electrophoresis followed by reaction with specific antibodies. These molecules are the only cysteine-containing histones, and they have been used as the target for thiol-specific reagents, in intact nuclei, isolated nucleosomes, histone complexes, and purified histones. Cysteine residues are available toN-ethylmaleimide only when histones are disassembled from the core particles. Upon reaction with these reagents, one of the H3 variants undergoes profound conformational changes, as revealed by an altered electrophoretic mobility.     

Nucleotide-induced change of the interaction between the 20- and 26-kilodalton heavy-chain segments of myosin adenosinetriphosphatase revealed by chemical cross-linking via the reactive thiol SH2

When myosin subfragment 1 (S-1) reacts with the bifunctional reagents with cross-linking spans of 3-4.5 A, p-nitrophenyl iodoacetate and p-nitrophenyl bromoacetate, the 20-kilodalton (20-kDa) segment of the heavy chain is cross-linked to the 26-kDa segment via the reactive thiol SH2. The well-defined reactive lysyl residue Lys-83 of the 26-kDa segment was not involved in the cross-linking. The cross-linking was completely abolished by nucleotides. Taking into account the recent report that SH2 is cross-linked to a thiol of the 50-kDa segment of S-1 using a reagent with a cross-linking span of 2 A [Chaussepied, P., Mornet, D., & Kassab, R. (1986) Proc. Natl. Acad. Sci. U.S.A. 83, 2037-2041], present results suggest that SH2 of S-1 lies close to both the 26- and 50-kDa segments of the heavy chain. The data also encourage us to confirm our previous suggestion that the ATPase site of S-1 residues at or near the region where all three segments of 26, 50, and 20 kDa are contiguous [Hiratsuka, T. (1984) J. Biochem. (Tokyo) 96, 269-272; Hiratsuka, T. (1985) J. Biochem. (Tokyo) 97, 71-78].     

Proximity and ligand-induced movement of interdomain residues in myosin subfragment 1 containing trapped MgADP and MgPPi probed by multifunctional cross-linking

The conformations of myosin subfragment 1 containing trapped MgADP or MgPPi have been studied by investigating the spatial disposition of the remainder of the subfragment 1 structure to the covalently bridged ATPase-related thiols SH1 and SH2. This has been done by synthesizing a trifunctional photoactivatable reagent 4,4'-bis(N-maleimido)benzophenone and reacting it with subfragment 1 in the presence of these ligands. Modification of subfragment 1 by this reagent mimics closely the changes in the ATPase properties as noted previously for modification with p-phenylenedimaleimide. In addition, noncovalent trapping of nucleotide also results, presumably by the bridging of the SH1 and SH2 thiols. On photolysis, cross-linking from the reagent bridging the thiols to other regions in subfragment 1 can be observed, but the extent and course of the photoinduced cross-linking depend on the nature of the trapped ligand. For subfragment 1 with trapped MgADP, a high efficiency cross-linking occurs between the 21-kDa segment and the 50-kDa segment. With MgPPi as the trapped ligand, low efficiency cross-linking occurs between the bridged thiols and either the 27-kDa N-terminal or the 50-kDa segments of the heavy chain. These results indicate that without the adenosine moiety, the binding of MgPPi to subfragment 1 leaves the protein in a flexible state so that residues in both the 27-kDa and the 50-kDa segment can move within the cross-linking span of the activated benzophenone triplet. The trapping of MgADP apparently results in a more rigid state for the subfragment 1 in which residues in the 50-kDa segment are spatially close to the bridged thiols, thus enabling photocross-linking to proceed with higher efficiency.     

Chromatin structure of Drosophila nasutoides satellites as probed by cross-linking DNA in situ

Chromatin structure can be probed by cross-linking DNA in situ using trioxsalen and irradiation with UV light. Presumably DNA within a nucleosome is protected from cross-linking so that this region appears as a single-strand loop in the electron microscope under a condition in which single-strands and double-strands are distinguished. Unprotected regions appear as duplex due to cross-linking.We have used this approach to investigate the structure of chromatins containing satellite DNAs of Drosophila nasutoides. We have previously shown that D. nasutoides has an unusually large autosome pair which is almost entirely heterochromatic. Its nuclear DNA reveals four major satellite components amounting up to 60% of the total genome. All of them are localized in this large heterochromatic chromosome. We wish to ask whether chromatins containing different satellite sequences have different arrangements of nucleosomes. Our results from cross-linking experiments show that all DNA components including main band DNA have different patterns of protected and unprotected regions: (a) The length distributions of protected regions show multiple peaks with the smallest unit lengths being 200 nucleotides for main band DNA, 180 for satellites I, II and III, and 160 for satellite IV. (b) The amounts of unprotected regions, presumably internucleosome DNA, vary from 16% for main band DNA to 60% for satellite IV, suggesting that satellite chromatins have fewer nucleosomes per given length of chromatin than main band DNA chromatin. The spacings between nucleosomes appear to be random in satellite chromatins.     

Catalytic cooperativity induced by SH1 labeling of myosin filaments

Modifications of SH1 groups on isolated myosin subfragment 1 (S-1) and myosin in muscle fibers affect differently the acto-S-1 ATPase and the fiber properties. Consistent with the findings of earlier work on fibers, the modification of SH1 groups in relaxed myofibrils with phenylmaleimide caused a loss of their shortening. This loss paralleled the decrease in the Vmax of extracted myosin but was not linear with the extent of SH1 labeling. Strikingly, the decrease in Vmax of S-1 prepared from the modified myofibrils was directly proportional to the extent of SH1 labeling. The specificity of SH1 labeling in myofibrils was verified by ATPase activities, thiol titrations, radiolabeling experiments, and comparisons to myosin labeled on SH1 in solution. To test for intermolecular interactions in the myosin filaments and their contribution to the differences between S-1 and myosin, the catalytic properties of copolymers of myosin were examined. Copolymers of myosin and rod minifilaments were formed in 5 mM citrate-Tris (pH 8.0) buffer, and their homogeneity was verified by sedimentation velocity analysis. The inhibition of actomyosin ATPase by rod particles was related to the decrease in the Km value. When rod particles were replaced in these minifilaments by SH1-modified myosin, the ATPase of the copolymers was increased over that of the combined ATPases of the individual filaments. The actomyosin ATP turnover rates on the unmodified heads were increased severalfold by the modified heads.(ABSTRACT TRUNCATED AT 250 WORDS)     

Ionic interaction of myosin loop 2 with residues located beyond the N-terminal part of actin probed by chemical cross-linking

To probe ionic contacts of skeletal muscle myosin with negatively charged residues located beyond the N-terminal part of actin, myosin subfragment 1 (S1) and actin split by ECP32 protease (ECP-actin) were cross-linked with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC). We have found that unmodified S1 can be cross-linked not only to the N-terminal part, but also to the C-terminal 36 kDa fragment of ECP-actin. Subsequent experiments performed on S1 cleaved by elastase or trypsin indicate that the cross-linking site in S1 is located within loop 2. This site is composed of Lys-636 and Lys-637 and can interact with negatively charged residues of the 36 kDa actin fragment, most probably with Glu-99 and Glu-100. Cross-links are formed both in the absence and presence of MgATP.P(i) analog, although the addition of nucleotide decreases the efficiency of the cross-linking reaction.     

Structural dynamics of the actomyosin complex probed by a bifunctional spin label that cross-links SH1 and SH2

We have used a bifunctional spin label (BSL) to cross-link Cys⁷⁰⁷ (SH1) and Cys⁶⁹⁷ (SH2) in the catalytic domain of myosin subfragment 1 (S1). BSL induces the same weakened ATPase activity and actin-binding affinity that is observed when SH1 and SH2 are cross-linked with pPDM, which traps an analog of the post-hydrolysis state A·M·ADP·P. Electron paramagnetic resonance showed that BSL reports the global orientation and dynamics of S1. When bound to actin in oriented muscle fibers in the absence of ATP, BSL-S1 showed almost complete orientational disorder, as reported previously for the weakly bound A·M·ADP. In contrast, helical order is observed for the strongly bound state A·M. Saturation transfer electron paramagnetic resonance showed that the disorder of cross-linked S1 on actin is nearly static on the microsecond timescale, at least 30 times slower than that of A·M·ADP. We conclude that cross-linked S1 exhibits rotational disorder comparable to that of A·M·ADP, slow rotational mobility comparable to that of A·M, and intermediate actin affinity. These results support the hypothesis that the catalytic domain of myosin is orientationally disordered on actin in a post-hydrolysis state in the early stages of force generation.     

Interaction of myosin subfragment 1 with fluorescent ribose-modified nucleotides. A comparison of vanadate trapping and SH1-SH2 cross-linking

The environment near the ribose binding site of skeletal myosin subfragment 1 (S1) was investigated by use of two adenosine 5'-diphosphate analogues with fluorescent groups attached at the 2'- and 3'-hydroxyls of the ribose ring. We have compared steady-state and time-resolved fluorescent properties of the reversibly bound S1-nucleotide complexes and the complexes generated by N,N'-p-phenylenedimaleimide (pPDM) thiol cross-linking or vanadate (Vi) trapping. A new fluorescent probe, 2'(3')-O-[N-[2-[[[5-(dimethylamino)naphthyl]sulfonyl] amino]ethyl]carbamoyl]adenosine 5'-diphosphate (DEDA-ADP), which contains a base-stable carbamoyl linkage between the ribose ring and the fluorescent dansyl group, was synthesized and characterized. For comparison, we performed parallel experiments with 2'(3')-O-(N-methylanthraniloyl)adenosine 5'-diphosphate (MANT-ADP) [Hiratsuka, T. (1983) Biochim. Biophys. Acta 742, 496-508]. Solute quenching studies indicated that both analogues bound reversibly to a single cleft or pocket near the ribose binding site. However, steady-state polarization measurements indicated that the probes were not rigidly bound to the protein. The quantum yields of both fluorophores were higher for the complexes formed after trapping with pPDM or Vi than for the reversibly bound complexes. Both DEDA-ADP and MANT-ADP, respectively, had nearly homogeneous lifetimes free in solution (3.65 and 4.65 ns), reversibly bound to S1 (12.8 and 8.6 ns), and trapped on S1 by pPDM (12.7 and 8.7 ns) or Vi (12.8 and 8.6 ns). In contrast to the quantum yields, the lifetimes were not increased upon trapping, compared to those of the reversibly bound states. These results suggested that static quenching in the reversibly bound complex was relieved upon trapping. Taken together, the results suggest that there was a conformational change near the ribose binding site upon trapping by either pPDM or Vi. On the basis of the quantum yield, lifetime, polarization, and solute accessibility studies, we could not detect differences between the S1-pPDM-nucleotide analog complex and the S1-Vi-nucleotide analogue complex for either analogue. Thus, previously observed differences with the adenine modified nucleotide analogue 1,N6-ethenoadenosine diphosphate (epsilon ADP) could not be detected with these ribose-modified probes, indicating that structural differences may be localized to the adenine binding site and not transmitted to the region near the ribose ring.     

Intramolecular cross-linking of myosin subfragment 1 with bimane

We previously showed that the fluorescent inter-thiol cross-linker dibromobimane (DBB) [Kosower, N. S., Kosower, E. M., Newton, G. L., & Ranney, H. M. (1979) Proc. Natl. Acad. Sci. U.S.A. 76, 3382-3386] cross-links two [50 and 20 kilodaltons (kDa)] of the three major fragments of myosin subfragment 1 (S-1); on intact S-1, DBB quenches tryptophans and inhibits all ATPases [Mornet, D., Ue, K., & Morales, M. F. (1985) Proc. Natl. Acad. Sci. U.S.A. 82, 1658-1662]. Here we characterize the modification chemically: DBB cross-links Cys-522 (50 kDa) with Cys-707 (20 kDa), thereby sealing a large preexisting heavy-chain loop containing important functionalities. Cross-linking rate is insensitive to nucleotides, but apparently sterically, either monobromobimane or DBB reduces Ca2+-ATPase to low, nonzero levels.     

Nucleotide- and temperature-induced changes in myosin subfragment-1 structure.

The effects of nucleotide binding and temperature on the internal structural dynamics of myosin subfragment 1 (S1) were monitored by intrinsic tryptophan phosphorescence lifetime and fluorescence anisotropy measurements. Changes in the global conformation of S1 were monitored by measuring its rate of rotational diffusion using transient electric birefringence techniques. At 5 degrees C, the binding of MgADP, MgADP,P and MgADP,V (vanadate) progressively reduce the rotational freedom of S1 tryptophans, producing what appear to be increasingly more rigidified S1-nucleotide structures. The changes in the luminescence properties of the tryptophans suggest that at least one is located at the interface of two S1 subdomains. Increasing the temperature from 0 to 25 degrees C increases the apparent internal mobility of S1 tryptophans in all cases and, in addition, a reversible temperature-dependent transition centered near 15 degrees C was observed for S1, S1-MgADP and S1-MgADP,P, but not for S1-MgADP,V. The rotational diffusion constants of S1 and S1-MgADP were measured at temperatures between 0 and 25 degrees C. After adjusting for the temperature and viscosity of the solvent, the data indicate that the thermally induced transition at 15 degrees C comprises local conformational changes, but no global conformational change. Structural features of S1-MgADP,P, which may relate to its role in force generation while bound to actin, are presented.     

FliG subunit arrangement in the flagellar rotor probed by targeted cross-linking

FliG is a component of the switch complex on the rotor of the bacterial flagellum. Each flagellar motor contains about 25 FliG molecules. The protein of Escherichia coli has 331 amino acid residues and comprises at least two discrete domains. A C-terminal domain of about 100 residues functions in rotation and includes charged residues that interact with the stator protein MotA. Other parts of the FliG protein are essential for flagellar assembly and interact with the MS ring protein FliF and the switch complex protein FliM. The crystal structure of the middle and C-terminal parts of FliG shows two globular domains joined by an alpha-helix and a short extended segment that contains two well-conserved glycine residues. Here, we describe targeted cross-linking studies of FliG that reveal features of its organization in the flagellum. Cys residues were introduced at various positions, singly or in pairs, and cross-linking by a maleimide or disulfide-inducing oxidant was examined. FliG molecules with pairs of Cys residues at certain positions in the middle domain formed disulfide-linked dimers and larger multimers with a high yield, showing that the middle domains of adjacent subunits are in fairly close proximity and putting constraints on the relative orientation of the domains. Certain proteins with single Cys replacements in the C-terminal domain formed dimers with moderate yields but not larger multimers. On the basis of the cross-linking results and the data available from mutational and electron microscopic studies, we propose a model for the organization of FliG subunits in the flagellum.