Mini-thin filaments regulated by troponin–tropomyosin

Abstract

Striated muscle thin filaments contain hundreds of actin monomers and scores of troponins and tropomyosins. To study the cooperative mechanism of thin filaments, “mini-thin filaments” were generated by isolating particles nearly matching the minimal structural repeat of thin filaments: a double helix of actin subunits with each strand approximately seven actins long and spanned by a troponin–tropomyosin complex. One end of the particles was capped by a gelsolin (segment 1–3)–TnT fusion protein (substituting for normal TnT), and the other end was capped by tropomodulin. EM showed that the particles were 46 ± 9 nm long, with a knob-like mass attributable to gelsolin at one end. Average actin, tropomyosin, and gelsolin–troponin composition indicated one troponin–tropomyosin attached to each strand of the two-stranded actin filament. The minifilaments thus nearly represent single regulatory units of thin filaments. The myosin S1 MgATPase rate stimulated by the minifilaments was Ca²⁺-sensitive, indicating that single regulatory length particles are sufficient for regulation. Ca²⁺ bound cooperatively to cardiac TnC in conventional thin filaments but noncooperatively to cardiac TnC in minifilaments in the absence of myosin. This suggests that thin filament Ca²⁺-binding cooperativity reflects indirect troponin–troponin interactions along the long axis of conventional filaments, which do not occur in minifilaments. Despite noncooperative Ca²⁺ binding to minifilaments in the absence of myosin, Ca²⁺ cooperatively activated the myosin S1-particle ATPase rate. Two-stranded single regulatory units therefore may be sufficient for myosin-mediated Ca²⁺-binding cooperativity. Functional mini-thin filaments are well suited for biochemical and structural analysis of thin-filament regulation.

Molecular motors produce force and movement by interacting with large filamentous protein assemblies such as the actin-based thin filaments along which myosin motors translocate thick filaments or organelles. In vertebrate and many invertebrate striated muscles, thin filaments also contain tropomyosin and troponin, which regulate actin–myosin interactions and thus muscle contraction. The present work describes the isolation and characterization of “mini-thin filaments,” which approximate single regulatory units of the thin filament and represent an approach for investigating these large assemblies and their regulation.

Striated muscle thin filaments typically are ≈1 μm long, comprise several hundred actin monomers, and contain one tropomyosin and one troponin for every seven actins. The actin/tropomyosin–troponin ratio is determined by the near correspondence between the 38.5-nm span of seven actin monomers along each long-pitch helical strand of the actin filament (1) and the ≈40-nm length of the tropomyosin coiled coil (2). We reasoned that short thin filaments ≈40 nm in length might be assembled by placing proteins that cap actin filament ends at opposite ends of tropomyosin. Because actin filaments are two-stranded helices, such particles would contain 14 actins, two tropomyosins, and two troponins if precisely constructed. These minifilaments would allow investigation of Ca²⁺-, myosin-, and troponin–tropomyosin-induced cooperativity independently of tropomyosin–tropomyosin end-to-end linkage that occurs on normal-length thin filaments. They also would make possible approaches to thin-filament structural analysis. We report here the design and purification of mini-thin filaments with the intended composition and compare their function to the function of conventional-length thin filaments.

Ca²⁺ regulates muscle contraction in the heart and in skeletal muscle by binding to specific site(s) in the NH₂ domain of the troponin subunit, TnC. Significantly, Ca²⁺ activates tension very cooperatively (3, 4) even in cardiac muscle, in which each TnC has only one regulatory Ca²⁺ site (5). This observation implies that multiple troponins on each thin filament bind Ca²⁺ interdependently in contracting muscle. Because mini-thin filaments have only two troponins, one on each actin strand, they should provide an opportunity to investigate this cooperative mechanism. As shown in the results below, thin-filament cooperativity is complex. Some aspects of cooperative regulation were disrupted in the particles, but other aspects were retained. The findings suggest which processes depend on end-to-end contacts between multiple contiguous units in the thin filament. More generally, the results provide perspectives on the regulation of muscle contraction and on the function of the thin filament as a large protein assembly.

Materials and Methods

Design of a Gelsolin–TnT Fusion Protein. By using standard techniques, the cDNA encoding residues Met-1–Tyr-418 of human gelsolin [a gift from D. Kwiatokowski, Harvard Medical School, Boston (6)] was inserted into the NcoI site of pET3d and succeeded in frame by bovine cardiac TnT cDNA (7) (minus the initiating ATG of TnT). Gelsolin contains an ≈26-residue linker connecting domain 3 to domain 4 (8). In the construct, this linker plus the structurally variable N-terminal part of domain 4 form the connection between gelsolin domains 1–3 and the TnT N terminus. Coding regions were deemed to be without errors by automated DNA sequencing.

Protein Purification. The gelsolin–TnT fusion protein was expressed in BL21 (DE3). After washing twice with 50 mM Tris·HCl (pH 8.5)/2 M urea/1 mM EDTA, inclusion bodies were dissolved in 50 mM diethanolamine (pH 8.9)/8 M urea/1 mM EDTA/10 mM DTT and applied to a Q-Sepharose FastFlow column. Protein elution used a 0–0.8 M NaCl gradient in column buffer [25 mM diethanolamine (pH 8.9)/5 M urea/1 mM EDTA/1 mM DTT]. Chicken E tropomodulin was purified by using a pGEX-KG expression plasmid obtained as a gift from V. Fowler (The Scripps Research Institute, La Jolla, CA) (9). Tropomodulin was purified further by FPLC (Resource Q column). Bovine cardiac troponin, troponin subunits, and tropomyosin and rabbit fast skeletal muscle myosin S1 and actin were purified as described (10).

Preparation of Cardiac TnC Labeled on Cys-84 with 2-(4′-(Iodoacetamido)anilino)naphthalene-6-sulfonic Acid (IAANS). Recombinant cardiac TnC (C35S) (a gift from M. Regnier, University of Washington, Seattle) was expressed in BL21 (DE3) cells. The crude lysate was directly applied to a Q-Sepharose FastFlow column, and eluting TnC was resolved further on an FPLC Resource Q column. The purified protein was dialyzed into 0.1 mM DTT/0.2 M KCl/30 mM Mops (pH 7.0)/2 mM EGTA, followed by dialysis without DTT. The TnC was adjusted to a concentration of 1–2 mg/ml, and 6 M urea plus a 4-fold excess of IAANS were added to effect labeling of Cys-84 at 4°C overnight. After quenching with DTT and either dialysis or G-25 desalting, final protein concentration was determined by protein assay with unlabeled TnC as a standard. Labeling stoichiometry was 0.88 mol of fluorophore/mol of protein, as determined by IAANS E₃₂₅ = 24,900 M^-1·cm^-1.

Reconstitution and Purification of a Gelsolin–Troponin Complex. Gelsolin–TnT and bovine cardiac muscle TnI and TnC (or IAANS-labeled TnC for experiments so indicated) were mixed under denaturing conditions in 1:1:1 ratios in the presence of 2 M urea/1 M KCl/20 mM Tris·HCl (pH 7.8)/1 mM DTT. Protein molarities were calculated by absorbance (11). Mixtures were dialyzed in consecutive solutions containing 1, 0.7, 0.5, 0.3, 0.2, and 0.06 M KCl plus 10 mM Tris·HCl (pH 7.8)/1 mM DTT. Samples were clarified by centrifugation, applied to an FPLC Resource Q column equilibrated with the same buffer, and eluted with a 0–0.8 M NaCl gradient.

Assembly and Isolation of Mini-Thin Filaments. First, 15 μM G-actin, 3 μM tropomyosin, 3 μM tropomodulin, 3 μM gelsolin–troponin, and 15 μM phalloidin were mixed in the presence of 10 mM Mops (pH 7.0)/0.2 mM DTT/0.2 mM ATP, and then 2 mM MgCl₂ and 100 mM KCl were added. After incubation at room temperature for 30 min, the 3- to 3.5-ml solution was chromatographed over Sephacryl HR S-500 equilibrated in 10 mM Mops (pH 7.0)/2 mM MgCl₂/0.2 mM DTT/0.2 mM ATP and 50 mM or 100 mM KCl. Fractions containing mini-thin filaments with concentrations at least 50% of the peak concentration were pooled and characterized. Of ≈2 mg of initial actin, ≈1 mg was recovered as minifilaments. To determine protein ratios, SDS/PAGE of the particles was compared quantitatively to SDS/PAGE standard curves of actin (42 kDa), tropomyosin (dimer, 66 kDa), and TnI (23.5 kDa) as described (12). (The actin standard curve also included tropomodulin at a 1:14 ratio relative to actin.) Results indicated a 7:(0.97 ± 0.11):(0.97 ± 0.10) ratio of actin/tropomyosin/gelsolin–troponin T.

Electron Micrographs and Image Analysis. Solutions containing mini-thin filaments were applied to holey carbon grids or holey carbon coated with an additional carbon layer and negatively stained with 1% uranyl acetate (13). Unchromatographed samples were viewed and recorded with a Hitachi (Tokyo) H-7000 transmission electron microscope. Sephacryl HR S-500 chromatographed particles were examined under low-dose conditions (≈12 electron/Ų) in a Philips (Eindhoven, The Netherlands) CM120 electron microscope. Magnifications were calibrated to 2% accuracy by using a diffraction grating with 2,160 lines per mm, and unselected filaments were measured particle by particle by using the program image j.

Myosin S1 ATPase Assay. Myosin S1 ATPase activity was determined as described in ref. 14. Conditions were: 25°C, 1 μM myosin S1/5 mM imidazole (pH 7.1)/3.5 mM MgCl₂/40 mM KCl/1 mM DTT/1 mM ATP/0.5 mM dibromo-1,2-bis(2-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid, and either mini-thin filaments (0.9 μM actin) or conventional thin filaments (0.9 μM F-actin, 0.4 μM tropomyosin, 0.4 μM troponin, and 0.2 μM phalloidin). Free Ca²⁺ concentrations were controlled by addition of CaCl₂ in varying amounts (15). The MgATPase rate of myosin S1 alone (0.032 s^-1) was subtracted. Neither thin filaments nor minifilaments had detectable MgATPase rates under these conditions in the absence of myosin S1. The program scientist (MicroMath, Salt Lake City) was used for nonlinear least-squares fitting MgATPase rate vs. Ca²⁺ data to the Hill equation. Curve fitting used averaged data for each Ca²⁺, weighted according to variance of three to five measurements. However, similar fits resulted when all measurements were fit as one data set.

ATPase measurements were made in the presence of low actin concentrations, resulting in rates very far below the V_max of myosin S1. The low rates resulted from two interrelated causes: relatively low concentrations of the particles were available, and S1-particle binding was inhibited by the presence of KCl required during particle preparation (particle isolation required 100 mM KCL, of which 40 mM KCl remained under ATPase conditions).

In control experiments, three minifilament preparations were examined by electron microscopy in the presence of ATP and myosin S1 under conditions identical to those of the ATPase experiments. No long filaments were observed, and no effect of myosin S1 on the particles was detected. Thus, myosin S1 did not induce filament elongation in the presence of ATP, and ATPase results reported below were due to the interactions between myosin S1 and short thin-filament particles.

Fluorescence Measurement. Mini-thin filaments labeled with IAANS TnC were prepared as described for unlabeled TnC and examined at an actin concentration of 1 μM. Fluorescence measurements were performed on 1.8-ml samples by using a FluoroMax-3 spectrofluorometer. Calcium titrations were performed by sequential addition of CaCl₂. Free calcium concentrations were calculated as described (16). Conditions were 25°C and 20 mM Mops (pH 7.0)/1 mM DTT/3 mM MgCl₂/0.1 M KCl/0.2 mM ATP/1 mM EGTA. Conventional thin filaments also were studied by reconstitution of purified proteins to final concentrations of 1.4 μM F-actin, 0.5 μM tropomyosin, 0.2 μM IAANS-troponin, and 1.4 μM phalloidin, incubated together for 1 h at room temperature. These concentrations were designed to maximize saturation of actin with troponin–tropomyosin so that cooperativity could be examined. Ca²⁺ increased the fluorescence intensity of these thin filaments by 36%, similar to the 39% increase for other samples, which were designed to maximize the fraction of troponin that was bound to the thin filament (5 μM F-actin, 1.8 μM tropomyosin, and 0.2 μM IAANS-troponin). Fluorescence intensities were corrected for progressive dilution (≈3% maximum). Ca²⁺ decreased the fluorescence intensity of IAANS-troponin in the absence of actin by 2% (data not shown), similar to previous findings (17). scientist was used for nonlinear least-squares fitting to the Hill equation or to a noncooperative binding isotherm by using pooled data from multiple fluorescence titrations.

Results

Strategy for the Creation of Mini-Thin Filaments. Gelsolin caps the “barbed” (“plus”) end of the two-stranded actin filament, preventing actin monomer addition and dissociation. The N-terminal three domains of gelsolin bind to the end of one actin strand, and domains 4–6 bind to the end of the other strand (8, 25). We therefore reasoned that short F-actin particles could be produced by using troponin to position gelsolin domains 1–3 on each of the two strands, at seven-actin intervals along thin filaments. More specifically, gelsolin domains 1–3, fused to the TnT N terminus, might cap thin filaments near the site of normal end-to-end overlap of adjacent tropomyosins in long thin filaments. The other, “pointed” (“minus”) end of the filaments then could be capped with tropomodulin, which interacts with the tropomyosin N terminus (18).

We considered that this approach might succeed for the following reasons: The troponin tail domain, comprised of N-terminal 170–200 amino acids of TnT, attaches to tropomyosin near the C terminus of the latter (4, 19, 20); hence, an appropriately engineered gelsolin–TnT construct would be positioned to make it feasible, in principle, to cap F-actin just beyond the end of troponin–tropomyosin. In addition, gelsolin fragments comprising domains 1–3 cap filaments regardless of Ca²⁺ concentration, presumably with one fragment linked to the barbed end of each actin strand (21, 22). Finally, TnC and TnI attach to TnT by means of the C-terminal region of the latter (24), indicating that attachment of a fusion protein to the TnT N terminus is unlikely to interfere with formation of a complex between TnT and the two other troponin subunits.

Before attempting to generate mini-thin filaments, cardiac TnC, cardiac TnI, and the cardiac TnT–gelsolin fusion protein were combined to form a ternary, gelsolin–troponin complex, which was purified by FPLC. SDS/PAGE indicated successful complex formation (Fig. 1). Mini-thin filaments then were grown from monomeric G-actin polymerized by addition of KCl and MgCl₂ in the presence of tropomyosin, gelsolin–troponin, and tropomodulin. Phalloidin was included to help stabilize the minifilaments. Electron microscopy showed that under these conditions only short filaments (mostly <<100 nm) were formed (Fig. 2C). Similar results were obtained in the presence or absence of Ca²⁺, i.e., with the addition of either CaCl₂ or EGTA (data not shown). Tropomyosin, tropomodulin, and gelsolin–troponin were each required for minifilament formation. Omission of tropomyosin (Fig. 2 A) or tropomodulin (Fig. 2B) yielded relatively long filaments (several hundred nanometers), still shorter than pure F-actin filaments but not useful for the present experiments. Omission of gelsolin–troponin yielded filaments with normal lengths, little different in appearance from F-actin–tropomyosin filaments (data not shown). Multiple attempts to form particles by severing F-actin with gelsolin–troponin rather than limiting growth from G-actin were unsuccessful and yielded mixtures of completely depolymerized actin and relatively long filaments.

Fig. 1.

SDS/PAGE of troponin and gelsolin–troponin fusion protein. Lane 1, bovine cardiac troponin prepared by reconstitution from isolated subunits TnT, TnI, and TnC followed by ion-exchange chromatography; lanes 2–3 (duplicates), similar to lane 1 except the reconstituted complex contained the gelsolin–TnT fusion protein instead of TnT; lane 4, native bovine cardiac troponin; Stnd, molecular weight standards. Results are from a 14% acrylamide gel.

Fig. 2.

Electron microscopy of mini-thin filaments. Conditions were as follows: 5 mM Mops, pH 7.0/2 mM MgCl₂/100 mM KCl/0.2 mM ATP/0.2 mM DTT. Proteins were incubated in the following concentrations: 5 μM G-actin, 1.2 μM tropomyosin, 1 μM gelsolin–troponin, 1 μM tropomodulin, and 5 μM phalloidin. The samples were diluted 5-fold immediately before application to grids. (A) Tropomyosin was omitted. (B) Tropomodulin was omitted. (C) All components were present. (D) Higher magnification of minifilaments prepared with all components as in C and then chromatographed by using Sephacryl S-500 (see Materials and Methods).

Electron Microscopy of Mini-Thin Filaments. After size-exclusion chromatography (Fig. 3 Lower) mini-thin-filament structure was examined by negative-stain electron microscopy. Minifilament width was relatively constant and was indistinguishable from normal F-actin–tropomyosin width. However, particle lengths (Fig. 4) were much shorter than those of normal thin filaments, with a mean of 45.7 ± 8.8 nm (n = 583), which corresponds to F-actin filaments approximately seven to eight monomers long, with capping proteins on both ends. The average particle size therefore approximated the seven-actin span of a single troponin–tropomyosin complex, as intended. Virtually all the particles (97%) were <80 nm and therefore could not accommodate two end-to-end tropomyosin lengths. Nevertheless, the length distribution (Fig. 4) was broad, indicating that the particles were not uniform.

Fig. 3.

Isolation of mini-thin filaments using size-exclusion chromatography. HR S500 chromatography (Lower) and gradient gel SDS/PAGE (Upper) of mini-thin filaments are shown. The minifilament peak is indicated on the chromatogram, performed with a column volume of ≈90 ml. Other peaks were variable in size from preparation to preparation and contained low protein concentrations. OD is in arbitrary units. Lanes 1–3 show column fractions on the leading edge, peak, and trail of the minifilament peak, respectively. pre, sample before column loading; Tm, tropomyosin.

Fig. 4.

Mini-thin-filament particle length distribution. After size-exclusion chromatography, filaments were negatively stained and examined by electron microscopy. Particle lengths are shown as binned into 2-nm groups and expressed as percent total. The major peak corresponds to a Gaussian distribution of 45.7 ± 8.8 nm (SD).

Except for their short length, minifilaments appeared similar to reconstituted F-actin or native thin filaments (Fig. 2D). A knob-like mass, however, was observed on one end of many particles (arrows). This feature cannot be troponin, because the globular, core domain of troponin binds closer to the middle than to the end of tropomyosin (19, 24) and thus would not be expected to contribute to density at the end of the particles. Instead, the extra density primarily must represent the gelsolin portions of gelsolin–troponins, one on each strand, each with a mass (≈45 kDa) comparable to that of an actin monomer. In structural models of gelsolin binding to F-actin, domains 1–3 protrude from the filament (8, 25). In contrast, models of tropomodulin binding (26) do not suggest a large protrusion of this type.

SDS/PAGE of Mini-Thin Filaments. After size-exclusion chromatography to remove any free tropomodulin, gelsolin–troponin, tropomyosin, and monomeric actin, the composition of the minifilaments was examined by SDS/PAGE (Fig. 3). The particles contained actin, tropomyosin, TnC, TnI, and TnT–gelsolin in a 7:1:1:1:1 ratio (Materials and Methods). Thus, on average, the actin, tropomyosin, and troponin composition of the particles resembled the composition of the repeating structural unit of the thin filament.

The molecular weight of tropomodulin is nearly identical to that of actin, and the two proteins closely comigrate on SDS/PAGE (data not shown). To test for the presence of tropomodulin on minifilaments, the actin band (including any proteins that comigrate with actin) was excised from the gel and digested with trypsin, and the proteolytic fragments were analyzed by MALDI-TOF mass spectrometry. Six tropomodulin peptides were detected, along with 11 actin peptides, confirming that the particles contained tropomodulin. The heights of the most prominent tropomodulin peaks (corresponding to residues 315–325 and 170–189) were similar to tropomodulin peak heights for a control sample containing actin and tropomodulin in a 14:1 ratio (peak heights measured relative to actin peptide peaks). However, the observed peak ratios were variable, and therefore the amount of tropomodulin per particle could not be measured easily.

Mini-Thin Filaments Activate the MgATPase Activity of Myosin. Conventional thin filaments activate the myosin S1 MgATPase rate, and Ca²⁺ regulates this activation. Similarly, minifilaments activated myosin S1 ATPase, and Ca²⁺ controlled this activation (Fig. 5, filled symbols and solid line). Composite data from three preparations showed overall dependence on Ca²⁺ in which the ATPase rate was almost 3-fold higher in the presence of calcium than in its absence: 0.049 ± 0.006 vs. 0.127 ± 0.004 s^-1 (ATP split per S1/s). Long filaments, therefore, are not necessary for Ca²⁺-sensitive control.

Fig. 5.

Mini-thin filaments activate myosin S1 ATPase activity. Addition of mini-thin filaments to myosin S1 increased ATPase activity in a Ca²⁺-dependent manner, with a 2.6-fold overall activation by Ca²⁺ (filled circles), which indicates that short filaments, only one regulatory unit long, were sufficient for regulation. Each point corresponds to an average of three to five measurements. The results indicate a cooperative transition (solid line) with n_H = 2.4 ± 0.5 and Ca²⁺ K_app = 1.34 ± 0.14 × 10⁶ m^-1. Adjusted data for conventional thin filaments are shown also (open circles), after normalization and scaling to the same maximum and minimum rates (see Results).

Fig. 5 also shows that Ca²⁺ sensitivity (i.e., apparent Ca²⁺ affinity) was similar for thin-filament particles (filled circles, K_app = 1.34 ± 0.14 × 10⁶ M^-1) and for conventional thin filaments (open circles, K_app = 7.4 ± 0.5 × 10⁵ M^-1). To facilitate long filament vs. minifilament Ca²⁺-sensitivity comparison, ATPase activation data for conventional filaments was normalized and then rescaled for Fig. 5 to match the maximum and minimum rates for minifilaments. When not so scaled, conventional thin filaments revealed slightly lower rates in the absence of Ca²⁺ (0.03 ± 0.01 s^-1) and a 2-fold higher maximal rate in the presence of saturating Ca²⁺ (0.28 ± 0.09 s^-1). These differences may reflect the presence of a small fraction of particles either shorter or longer than 40 nm (see Fig. 4). Tropomyosin would not be expected to bind properly to the very short particles, and similarly, those with lengths between 40 and 80 nm might bind only one tropomyosin, which in both cases could lead to actins that were unregulated by troponin–tropomyosin.

Surprisingly, minifilament ATPase activation by Ca²⁺ was cooperative, as evidenced by a Hill coefficient (n_H) of 2.4 ± 0.5 (Fig. 5, solid line), where n_H > 1 suggests cooperative site–site interactions. Long filaments exhibit very similar cooperativity (and Ca²⁺ K_app), as shown in Fig. 5 and as reported (4, 10, 15, 27). Myosin is known to increase TnC Ca²⁺ affinity, and myosin can bind cooperatively to the seven actins contacted by a single troponin–tropomyosin complex (3, 28–31). These properties may have influenced the Fig. 5 data. Could they explain the cooperativity? To evaluate this possible explanation, one can consider a minifilament model in which the above-mentioned properties are postulated in the extreme: (i) myosin binding is maximally cooperative for each actin strand (i.e., either seven myosins are bound or none); (ii) bound myosin increases the Ca²⁺ affinity of that strand's TnC by 10-fold; (iii) the ATPase rate is proportional to myosin binding and also requires Ca²⁺ binding to TnC site II; and (iv) the two actin strands do not affect each other. Simulations of this steady-state model (not shown) show that the magnitude and Ca²⁺ sensitivity of the ATPase rate depend on myosin concentration, and Ca²⁺ and myosin affect the binding of each other. Nevertheless, the simulations confirm a noncooperative n_H = 1 for ATPase rate vs. Ca²⁺ concentration, which is what cooperativity theory (33) would predict, because there is only one Ca²⁺-binding site in this illustrative system. Therefore, returning to the present data, cooperative myosin binding to the multiple actins within each strand cannot be excluded but nevertheless is disfavored as an explanation for n_H > 1 in Fig. 5. An alternative interpretation is that the troponins on opposite strands of the particles interact in some manner. This possibility is discussed further below.

Ca²⁺ Binding to Thin Filaments and to Mini-Filaments. Ca²⁺ binds cooperatively to the regulatory sites (TnC site II) of isolated cardiac thin filaments (27), as monitored by the fluorescence intensity of TnC labeled with IAANS. However, two Cys residues are labeled with this approach (34), introducing ambiguity in interpretation, and results are condition-dependent (27). A potential improvement is to use single-Cys TnC, specifically labeled on Cys-84 (17, 34, 35). Single-Cys TnC has been used to monitor Ca²⁺ binding to isolated TnC and to TnC incorporated into muscle fibers but has not been used previously to study thin filaments in the absence of myosin. We prepared both conventional and mini-thin filaments containing Cys-84–IAANS TnC. Ca²⁺ cooperatively increased the fluorescence intensity of Cys-84 IAANS-labeled control thin filaments (Fig. 6, open symbols), with n_H = 1.77 ± 0.05. Because there is only one regulatory site on each cardiac TnC, this cooperativity implies that troponins interact with each other in the absence of myosin. An identical Ca²⁺ titration was performed by using IAANS-labeled thin-filament particles (Fig. 6, filled symbols). In this case, the fluorescence transition was noncooperative (n_H = 0.999 ± 0.005), which suggests that Ca²⁺-binding cooperativity in conventional thin filaments reflects interactions along the long filament axis, absent in the particles.

Fig. 6.

Ca²⁺ binding to the regulatory sites of thin filaments and to regulatory sites of thin-filament particles. Ca²⁺ binding was monitored by titration of thin filaments (open circles) or minifilament particles (filled circles), in each case fluorescently labeled on cardiac TnC at Cys-84. Solid lines are best-fit curves and indicate a qualitatively different transition shape for the long thin filaments. Unlike the findings for thin-filament particles, the effect of Ca²⁺ on long thin filaments was cooperative, producing a steeper transition. Steadystate fluorescence results from three experiments are superimposed. Intensities were normalized to the fractional change produced by saturating Ca²⁺. The maximal fluorescence increase for short thin filaments was two-thirds of the maximal increase for conventional thin filaments (i.e., a 24% increase rather than 36% after Ca²⁺ addition). Ca²⁺ binding to the particles was slightly stronger (K_Ca²⁺ = 4.5 ± 0.2 × 10⁶ m^-1 for the particles vs. 2.78 ± 0.05 × 10⁶ m^-1 for thin filaments).

Discussion

The design and isolation of mini-thin filaments allowed us to distinguish aspects of cooperative thin-filament activation that are inherent in a single regulatory unit (i.e., a double actin helix with seven actins and a troponin–tropomyosin complex in each strand) from those that depend on end-to-end contacts between multiple contiguous units. There are three principal findings. First, in the absence of myosin, Ca²⁺ bound cooperatively to the regulatory sites of long thin filaments but noncooperatively to those of mini-thin filaments, which suggests that (indirect) troponin–troponin interactions along the longitudinal filament axis affect Ca²⁺ binding in the absence of myosin. Second, minifilaments activated the myosin S1 MgATPase rate in a Ca²⁺-regulated manner, indicating that short-length particles are sufficient for Ca²⁺-mediated control of myosin. Finally, ATPase regulation by Ca²⁺ was cooperative, which suggests the possibility that troponin–troponin interactions also may occur across the thin filament (i.e., between its two strands) when myosin is present.

Although attempts to understand the cooperative regulation of muscle contraction lack consensus, they have a common theme emphasizing the extended linear organization of the thin filament. Both structural and biochemical data imply that there are no less than three states of the thin filament (e.g., see refs. 30, 36, and 37). In some models (38, 39), the key regulatory phenomenon is a propagated change in the conformation of many adjacent actins. Functionally important changes in actin may occur, but the regulatory mechanism in our view instead should emphasize structural and kinetic evidence that strong myosin–actin attachment requires tropomyosin to shift position on actin (30, 37, 40). Therefore, we concur with others who consider myosin-binding cooperativity to involve the tendency of the tropomyosin strand to shift position not at a single actin or even at seven actins but rather coordinately on actins within several adjacent regulatory units (31, 37, 41, 42). Such shifts are consistent with three-state models, because Ca²⁺ and myosin have different effects on the position of tropomyosin on actin. Transitions among thin-filament states tend to be cooperative.

Results shown in Fig. 6 strongly suggest that the effect of Ca²⁺ on the thin filament involves cooperative effects between troponins located successively along the long axis of the actin filament. If, instead, cooperative Ca²⁺ binding to conventional thin filaments were caused by interactions between troponins opposite each other on the two actin strands, then similar cooperative Ca²⁺ binding to the mini-thin filaments would be expected, contrary to observation. Therefore, we conclude that regulatory sites on troponins adjacent to each other along the long axis of (conventional) thin filaments tend to bind Ca²⁺ cooperatively, despite their ≈40-nm separation. From the observed n_H, the cooperative free energy affecting Ca²⁺ binding can be estimated as ΔGc = -RT ln (n_H²) = 0.7 kcal/mol (1 cal = 4.18 J) (27).

Notably, the cooperative activation of the myosin S1-minifilament ATPase rate by Ca²⁺ (Fig. 5) contrasts with the noncooperative Ca²⁺ binding observed in the absence of myosin (Fig. 6). Low ATPase rates and limited material thus far have precluded testing whether myosin binding itself is cooperative in the presence of ATP. Nevertheless, n_H ≈ 2 suggests that myosin S1 ATPase activity requires that site II of both TnCs on minifilaments bind Ca²⁺ and that the sites strongly affect each other. Previously, we suggested that myosin's ≈10,000-fold strengthening of tropomyosin–actin binding reflects an effect of myosin on the actin inner domain, the actin site at which tropomyosin binds to myosin-decorated actin. The current results can be explained if the effects of myosin on actin also alter the interface between monomers on the two strands of the filament. If so, then myosin binding to actin on one strand would influence the troponin–tropomyosin on the opposite strand by means of linked actin–actin and actin–tropomyosin effects. However, additional evidence will be needed to determine whether myosin binding to one actin strand affects the actin monomers, tropomyosin, and/or troponin on the opposite actin strand. If strand–strand interactions are as important as suggested by the current data, then additional supporting results should emerge.

It is important to consider whether experimental capping of the minifilament ends spuriously affected the fluorescence and/or ATPase results. Ca²⁺ binding by the gelsolin fragments might have altered troponin-mediated effects of Ca²⁺ or even mediated strand–strand communication at the particle ends. We consider such phenomena unlikely to have had major effects. Conventional and mini-thin filaments exhibited similar Ca²⁺ affinity (Figs. 4 and 5), fluorescence change, and ATPase regulation. The similar Ca²⁺ affinities suggest that the gelsolin and troponin Ca²⁺ sites do not interact strongly, and the data generally suggest that the particles comprise a valid experimental model for studying the properties of the thin filament. Also, cooperativity typically reflects quaternary structure shifts that involve specific protein–protein contacts. Therefore, we view end effects or other spurious features of the particles as unlikely (albeit plausible) explanations of the ATPase cooperativity.

Finally, there are two interrelated rationales for isolating thin-filament particles: to investigate function as in the present report and to facilitate determination of structure. Current structural models of the thin filament (and of bare F-actin) are informative, but these models are not at atomic resolution and are notably poor in information about troponin. Also, the structure of gelsolin bound to F-actin is less well elucidated than is the structure of gelsolin bound to G-actin. Single-particle analysis of electron microscope images of the minifilaments and potentially x-ray crystallography of minifilaments offer solutions to unraveling the structural organization of actin and actin-binding proteins.