Structure of the N Terminus of a Non-Muscle α-Tropomyosin in Complex with the C Terminus: Implications for Actin Binding

Abstract

Tropomyosin is a coiled-coil actin binding protein that stabilizes the filament, protects it from severing, and cooperatively regulates actin’s interaction with myosin. Depending on the first coding exon, tropomyosins are low molecular weight (LMW), found in the cytoskeleton and predominant in transformed cells, or high molecular weight (HMW) found in muscle and non-muscle cells. The N- and C-terminal ends form a complex that allows tropomyosin to associate N terminus-to-C terminus along the actin filament. We determined the structure of a LMW tropomyosin N-terminal model peptide complexed with a smooth/non-muscle tropomyosin C-terminal peptide. Using NMR and circular dichroism we showed that both ends become more helical upon complex formation, but that the C-terminal peptide is partially unfolded at 20 °C. The first five residues of the N terminus that are disordered in the free peptide are more helical and are part of the overlap complex. NMR data indicate residues 2-17 bind to the C terminus in the complex. The data support a model for the LMW overlap complex that is homologous to the striated muscle tropomyosin complex in which the ends are oriented in parallel N terminus-to-C terminus with the plane of the N-terminal coiled coil perpendicular to the plane of the C terminus. The main difference is that the overlap spans 16 residues in the LMW tropomyosin complex compared to 11 residues in the HMW striated muscle overlap complex. We discuss the relevance of a stable but dynamic intermolecular junction for high-affinity binding to actin.

A dynamic actin cytoskeleton is required for cellular movement, determination and maintenance of cellular shape, and numerous actin-dependent functions. The pleiotropic properties of the actin filament depend on its ability to bind a variety of proteins, many of which are the final targets in signaling pathways. In recent years tropomyosin has been recognized as universal regulator of the actin filament (1).

Tropomyosins are encoded by a multi-gene family and are expressed in most eukaryotes in cell- and tissue-specific manners. In mammals there are four genes, with additional isoforms generated by alternative exon expression, by the use of alternate promoters and by formation of heterdimers (1). The alternate promoters result in two tropomyosin classes: high molecular weight (HMW) tropomyosins that are ~284 residues long and are expressed in both muscle and non-muscle cells, and low molecular weight (LMW) tropomyosins (~247 residues long) found in cellular cytoskeletons. In LMW tropomyosins the first exon, exon 1b, replaces exon 1a and 2 in the HMW tropomyosins. Although the sequences of the two classes of N-terminal ends are overall conserved within each class, relatively minor sequence differences can have a major effect on function (2). While much remains to be learned about the specific functions of non-muscle tropomyosin isoforms, it has long been known that oncogenic transformation of cells is accompanied by down-regulation of certain HMW tropomyosins (1).

The importance of the ends of tropomyosin for actin binding and regulatory function is well-established. Both ends are encoded by alternatively-expressed exons and are the binding sites for other proteins including tropomodulin at the N terminus (3) and troponin T in striated muscle isoforms (4). The extreme ends of striated muscle α- tropomyosin are required for cooperative binding to filamentous actin, Ca²⁺-dependent regulation by troponin in muscle cells, and modulate the cooperative interaction of myosin with the actin filament (5, 6).

The structures of isolated C- and N- terminal domains of tropomyosins have been solved in model peptides using NMR and X-ray crystallography (7-12). In the solution structure of the complex formed between the N- and C-terminal domains of rat striated muscle α-tropomyosin (1aNTD and 9aCTD, respectively) the chains of the C-terminal end spread apart to allow insertion of 11 residues of the N-terminal coiled coil into the resulting cleft (13). The structure of the C terminus in the complex is close to that found in crystal structures in which the chains form an antiparallel complex with another C-terminal molecule (10, 12). The 1aNTD in the complex is similar in structure to the uncomplexed peptide, but its chains are slightly spread apart (7).

In the 1aNTD/9aCTD complex, the plane of the N-terminal coiled-coil is rotated 90° relative to the plane of the C terminus. A consequence of the geometry is that when the overlap complex is modeled into the X-ray structure of tropomyosin the orientation of postulated periodic actin binding sites (14) on the coiled-coil surface is retained from one molecule to the next along the actin filament (13).

The LMW α-tropomyosins with exon 1b-encoded N-termini bind with higher affinity to actin than HMW forms, despite being shorter, such that one molecule has only six instead of seven periodic repeats, the length of six actin subunits in the filament (5). The structure of the N terminus of unbound rat α-TM encoded by exon 1b was solved in a chimeric peptide, αTM1b_1-19Zip (1bNTD), using NMR (9). The main difference between the structures of αTM1a_1-14Zip (1aNTD) and 1bNTD is that exon 1b encodes an N-terminal extension of five residues that is disordered in the solution structure (9) and whose specific function is unknown.

Here we have used circular dichroism to analyze the affinity and folding properties of two peptides that model the intermolecular junction of short tropomyosins, 1bNTD, that contains residues encoded by exon 1b and the C-terminal peptide, 9dCTD, that contains residues encoded by exon 9d. We determined the structure of 1bNTD in complex with the C terminus and the residues that interact with the C terminus using heteronuclear NMR, and compared it to the structure of unbound 1bNTD. Based on the NMR results we propose a model for the non-muscle complex in which the two peptides form a dynamic complex with a structure similar to but longer than that of striated muscle tropomyosin. We discuss the possible relevance to actin affinity of a stable complex that contains unfolded regions.

MATERIALS AND METHODS

Peptides

Unlabeled 1bNTD was synthesized at the Keck Facility at Yale University (New Haven, CT). Unlabeled αTM9a_246–284 (9a39), αTM9d_246–284 (9d39) and AcTM1a_1-14Zip and GlyTM1a_1-14Zip were purchased from SynPep (Dublin, CA). 1bNTD labeled with ¹⁵N or ¹⁵N and ¹³C was expressed and purified as previously described (9). Unlabeled 9dCTD was synthesized by the Tufts University Core Facility (Boston, MA). The peptide was cross-linked by disulfide formation catalyzed by the presence of CuSO4 and K₃FeCN₆ (15). Oxidized and reduced peptides were separated using HPLC on a C18 reverse phase column (Vydac 218TP10, Hesperia, CA). The concentrations of the peptides were determined using the microbiuret procedure of Goa (16). The peptides were dissolved in 0.01 mM sodium phosphate, 100 mM NaCl, pH 6.5 for all CD and NMR studies.

Circular dichroism

Circular dichroism measurements were carried out as previously described (2). Data obtained as a function of temperature were fit to curves for the unfolding of monomers (CTD), dimers (NTD) and heterotrimers (NTD/CTD complex) (17). The helical content of the peptides were determined using the program CDNN (18). To determine the binding constant of the 1bNTD/9dCTD complex, the 1bNTD peptide was titrated with a solution containing the same concentration of 1bNTD peptide and excess 9dCTD peptide to maintain constant 1bNTD concentration, as described by Greenfield (19). The binding constant was determined by the method of Engel (20) with a linear term to correct for the contribution of unbound 9dCTD. The combined equation is:

$$\mathrm{y}=\mathrm{M}\{(1+\mathrm{nkx}+\mathrm{kP})/2\mathrm{kP}­{[{\left((1+\mathrm{nkx}+\mathrm{kP})/2\mathrm{kP}\right)}^2­\mathrm{nx}/\mathrm{P}]}^{½}\}+\mathrm{mx}+\mathrm{b}$$

where y is the ellipticity as a function of total added 9dCTD, M is the maximal increase in ellipticity when 9dCTD binds to 1bNTD compared to the sum of the unmixed components, n is 1/number of binding sites, x is the total concentration of 9dCTD, k is the binding constant for the 1bNTD/9dCTD complex, P is the total concentration of 1bNTD, m is the change in ellipticity of unbound 9dCTD as a function of concentration and b is the intercept of the unbound 9dCTD on the Y axis. Curve fitting using the Levenberg-Marquardt algorithm (21) was performed using SigmaPlot 8.0 (Systat Software, Inc. San Jose, CA), which was also used to create all the graphs in this paper.

Nuclear Magnetic Resonance

The NMR data were collected on a Varian Inova 600 spectrometer (Varian, Inc., Palo Alto, CA) equipped with a pulse-field gradient HCN cryoprobe. All measurements were performed in susceptibility-matched NMR tubes (Shigemi, Inc., Allison Park, PA). The sequential assignments and major structural studies were carried out at 20 °C, although HSQC and NOESY data of the complexes were also collected at 10 °C. Data processing and peak picking were performed as previously described (13). The chemical shifts of the C^α and C^β atoms of 1bNTD were only slightly changed by complex formation compared to those of unbound 1bNTD (9). Therefore it was possible to determine the sequential assignments of the H^N, C^α, C^β, H^α and H^β resonances using a minimal set of NMR experiments: ¹⁵N-¹H HSQC (22, 23), HNCO (24), H(CA)(CO)NH, (CA)(CO)NH (25, 26), CBCA(CO)NH (27) and HBHA(CO)NH (28). The other side-chain resonances were determined using HCCH-COSY (29, 30) and ¹³C-¹H HSQC (22) spectra.

Structure Determination

Distance constraints to determine the changes in structure of ¹³C¹⁵N-labeled 1bNTD upon complex formation with unlabeled 9dCTD were obtained by analysis of ¹⁵N edited PFG NOESY-HSQC(31), ¹³C edited PFG NOESY-HSQC (32) and X-filtered ¹³C-edited NOESY (33) experiments as previously described for uncomplexed 1bNTD and 9dCTD (9, 11) with the following modifications. To identify all interchain NOEs, a heterodimer of 1bNTD was prepared with one chain labeled with ¹⁵N and ¹³C and one chain unlabeled. 1bNTD heterodimer was mixed with unlabeled 9dCTD. X-filtered ¹³C-edited NOESY crosspeaks from this complex are due to both intermolecular NOEs between the unlabeled 9d peptide and the labeled chain of the 1b peptide and intramolecular NOEs from the unlabeled to the labeled chain of the 1b peptide. To identify the intermolecular NOEs between the 9dCTD and 1bNTD, an X-filtered ¹³C-edited NOESY experiment was performed in which both chains of the 9d peptide were unlabeled and both chains of the 1b peptide were labeled with ¹³C and ¹⁵N.

These intermolecular NOEs were removed from the data set obtained when only one chain of the 1bNTD peptide was labeled to identify the intramolecular-interchain crosspeaks. Only these were used for the structure determination of bound 1bNTD. The structure was calculated using AutoStructure v1.1.2 (34, 35) which calls DYANA (36) as a subroutine as previously described for the uncomplexed 1bNTD peptide (9). The structure was refined using CNS (37), as previously described (13).

Molecular Modeling

The 1bNTD/9dCTD complex was modeled on the structure of the complex of the striated muscle 1aNTD/9aCTD complex. First, the helical residues from 1aNTD (1-29) and 1bNTD (6-34) were aligned using the coordinates from Model 1 of the 1aNTD/9aCTD complex (RCSB data bank 2g9j) and the coordinates from the first model of the 1bNTD structure in the presence of unlabeled 9dCTD determined herein. The residues were aligned using MolMol by removing the chain assignments and “TER” codes from the PDB files. The combined file was exported as a single pdb file. The coordinates of the 1aNTD were removed, the “TER” codes were added to separate the chains and the hybrid 1bNTD/9aCTD molecule was imported into the program Sybyl v8.0 (Tripos International, St. Louis MO). The coordinates from the 9aCTD (αTM_251-284) were changed to those of the 9dCTD (αTM_252-284) and the energy of the complex was minimized.

A molecular model of full-length LMW tropomyosin (LMW TM) was created based on model 1 of the X-ray structure of striated muscle tropomyosin (PDB accession code 1c1g, (38)) using Sybyl v8.0. In LMW tropomyosins exon 1b, which encodes 44 residues, is expressed in place of exons 1a and 2 that encode residues 1-80 of HMW tropomyosins. The initial Met in the LMW rat α-tropomyosins (TM-5a and TM-5b (39)) is cleaved following translation, thus the first residue is A1. To model the N terminus encoded by exon 1b, residues 1-37 from chains A and B were deleted. Residues 38 to 80 were changed to residues 1 to 43 from rat TM-5a. To model the non-muscle C terminus, residues 258-284 encoded by exon 9a were replaced by exon 9d-encoded residues and the energy of the structure was minimized. To link two of the LMW TM models with the LMW overlap complex, the coordinates of residues 1-21 of the N-terminus of one LMW TM model were replaced with those of the 1bNTD peptide complexed with the 9dCTD peptide and the coordinates of the last 24 residues of a second LMW TM molecule were replaced with those from the model of the 9dCTD in the complex, as described previously for the 1aNTD/9aCTD complex (13).

RESULTS

Design of the peptide model system

The model system for determining the solution structure of the intermolecular junction between the ends of non-muscle LMW tropomyosins is based on the model for the overlap region of striated muscle tropomyosin (Table 1) (13). The N-terminal domain peptide (1bNTD) is αTM1b_1-19Zip, the first 19 residues encoded by exon 1b followed by the last 18 residues of the leucine zipper region of the yeast transcription factor, GCN4 (9). The C-terminal domain peptide (9dCTD) is α-TM9d_252-284, the last 35 residues of the C-terminus of smooth and non-muscle α-tropomyosins in which residues 252-257 are those encoded by part of the constitutively-expressed exon 8, and residues 258-284 by exon 9d (based on the residue numbering of a HMW, 284-residue TM). The addition of GCGG at the N terminus allows oxidative crosslinking of the peptide.

TABLE 1.

Peptide Sequences

^aThe residues which derive from tropomyosin are in upper case and the residues from GCN4 or added to allow cross-linking are in lower case.

^bThe first methionine residue encoded by exon 1b is cleaved following translation so the first residue in LMW tropomyosin is A1.

^cThese peptides contain the mutation N279K, which increases peptide stability (11).

^dResidues 252-284 encoded by exon 9d in HMW rat α-TM correspond to residues 215-247 of rat LMW tropomyosins.

The C- and N-terminal domains of short tropomyosins form a stable complex but the C-terminal domain in the complex is partially disordered

Figure 1A shows the circular dichroism spectra of the 1bNTD and 9dCTD peptides, a mixture of the two peptides and addition of the individual curves. At 20 °C the 1bNTD is ~65% helical but the 9d peptide is only ~15 % helical. Although it is mainly unfolded at 20 °C, the 9dCTD peptide binds tightly to the 1bNTD and the binding is accompanied by a large increase in helical content. A one-to-one mixture of the peptides has a helical content of 43%, while addition of the curves for the 1bNTD and 9dCTD would give a helical content of only 33%. The increase in helical content upon binding was used to determine the dissociation constant of the 1bNTD/9dCTD complex (see Materials and Methods). When 1bNTD, 7.9 μM, was titrated with 9dCTD at 20 °C (Figure 1B), the ellipticity saturated at a 0.94±0.03 molar ratio, indicating formation of a 1:1 complex with a K_d of <0.5 μM, too tight to measure accurately at the concentration used for the CD measurements.

Figure 1.

Dissociation constant and stability of the 1bNTD/9dCTD complex. (A) The ellipticity of (●) 9dCTD, 7.9 μM, (□) 1bNTD, 7.9 μM, (▲) mixture of the 1bNTD and 9dCTD peptides, (Δ) addition of the curves of the unmixed 1bNTD and 9dCTD peptides in 10 mM sodium phosphate, 100 mM NaCl, pH 6.5 at 20 °C. (B) Titration of 1bNTD, 7.9 μM with increasing concentrations of 9dCTD plus 1bNTD, 7.9 μM at 20 °C. (○) 250-208 nm, (●) 250-222 nm. The data were not corrected for the contribution of unbound 9dCTD. The data were fit to a curve for the formation of a binary complex ((20). (C) The ellipticity of (●) 9dCTD, 35 μM, (□) 1bNTD, 35 μM, (▲) mixture of the 1bNTD and 9dCTD peptides, (Δ) addition of the curves of the unmixed 1bNTD and 9dCTD peptides in 10 mM sodium phosphate, 100 mM NaCl, pH 6.5 at -5 °C. (D) The change in ellipticity as a function of temperature. Symbols and conditions same as 1C. The unfolding of 9dCTD is independent of protein concentration since it is crosslinked by a disulfide bond. (E) The ellipticity of the mixed 1bNTD and 9dCTD peptides minus the ellipticity of the individual peptides as a function of temperature. The unfolding data of 1bNTD and 9dCTD in 1D (all points) were fit by equations for the unfolding of a dimer and a monomer, respectively. The data in D and E for the unfolding of the 1bNTD/9dCTD complex were fit by curves for the unfolding of a monomer followed by the unfolding of a heterotrimer (17). Symbols show every 8th point in A and C, all points in B, every 15th point in D, every 10th point in E. The data in 1A and 1C were smoothed (63).

Even though the binding is tight at 20 °C, the complex is not fully folded at this temperature. Figure 1C shows the spectra of the mixed and unmixed 1bNTD and 9dCTD peptides, 35 μM, at -5 °C. The helical content of 1bNTD is 75% helical, consistent with the solution structure (9). The helical content of the unbound 9dCTD is ~25%, and it unfolds with a TM of < 0 °C (Figure 1D). The helical content of the 1:1 mixture is 73%, higher than that calculated for the addition of the spectra of the unbound peptides (~50%) indicating binding. The unfolding and dissociation of the 1bNTD/9dCTD complex is not cooperative (Figures 1D, 1E). The complex begins to unfold before the chains dissociate. The major T_M of unfolding of the 1bNTD/1dCTD complex is 14.7 °C (Figure 1E) but this unfolding transition is independent of peptide concentration, suggesting that it does not represent dissociation of the NTD/CTD complex. In contrast, the 9aCTD is greater than 75 % helical at 10 °C and unfolds with a T_M of ~25 °C (11), and the unfolding and chain dissociation of complexes containing the exon 9a-encoded C-terminus are cooperative.

The low stability of the exon 9d-encoded sequence is well established from studies of full-length tropomyosins and other model TM peptides (6, 40, 41). Moreover, it has been shown by comparing differential scanning calorimetry and CD that unfolding is non-cooperative in full length tropomyosins containing the N terminus encoded by exon 1b and the C terminus encoded by exon 9d (42), with substantial unfolding occurring without chain dissociation.

Non-cooperative unfolding and dissociation of the complex of the C- and N-terminal domains of the non-muscle short tropomyosin were also observed in another peptide model for the exon 9d-encoded C terminus (6), described in Supplemental Material, Figure S1. Based on the results with both peptides we conclude that the C terminus encoded by exon 9d binds with high affinity to the N terminus encoded by exon 1b, but that the complexes begin to unfold and lose helical content before the chains dissociate. At 20 °C, the temperature used for determination of the solution structure, the two peptides bind tightly to each other but extensive regions of the bound 9dCTD are disordered.

The ¹⁵N-¹H HSQC spectra of 1bNTD in complex with 9dCTD

The ¹⁵N-¹H HSQC spectra of ¹⁵N¹³C-1bNTD or ¹⁵N-1bNTD in complex with 9dCTD allowed initial characterization of the 1bNTD structure in the complex. Spectra were collected at 10, 15 and 20 °C. Many resonances were broadened at the lower two temperatures, but at 20 °C, all the resonances were visible although some overlapped (Figure 2). There is only one set of resonances for all of the NH crosspeaks showing that the intermolecular junction is symmetric, as is the striated muscle 1aNTD/9aCTD complex (13).

Figure 2.

¹⁵N-¹H HSQC spectra of ¹⁵N-1bNTD in the absence (black) and presence (cyan) of unlabeled 9dCTD at 20 °C. The crosspeaks from the free ¹⁵N-1bNTD peptide are labeled in black. The crosspeaks from the peptide complex are labeled in cyan when they are displaced from those of the free peptide. The peptides were 0.35 mM in 100 mM NaCl, 10 mM sodium phosphate, pH 6.5. There is only one set of crosspeaks for both the free and complexed 1bNTD showing that the complex is symmetric.

The changes in chemical shifts and relative intensities of the crosspeaks of homologous residues of the 1aNTD and 1bNTD peptides differ in the two complexes (Figure 3). When 1bCTD was in a complex with 9dCTD, most of the crosspeaks near the N terminus gained in relative intensity, whereas those near the center of the molecule shifted (Figure 3A) and were broadened (Figure 3B). The H^N resonances from residues 14-17 in the ¹⁵N-¹H HSQC spectra of 1bNTD were the most broadened upon complex formation with 9dCTD. A similar pattern of broadening was noted at 10 °C, but the relative broadening of residues 12-17 was more severe at the lower temperatures. Residues 6-36 of unbound 1bNTD are fully helical and have similar heteronuclear ¹H-¹⁵N HNOE ratios and stabilities (9). The secondary structure of these residues does not change upon complex formation (see below). Therefore, the source of the exchange broadening of the resonances from residues 14-17 of 1bNTD most likely arises from their interactions with residues from the C terminus of the 9dCTD peptide undergoing chemical exchange between multiple conformations. The HSQC data are consistent with the CD results that show the bound 9dCTD peptide is partially unfolded at 20 °C.

Figure 3.

The changes in the chemical shifts and relative intensities of the crosspeaks in the ¹⁵N-¹H-HSQC spectra of ¹⁵N-1bNTD and ¹⁵N-1aNTD complexed with unlabeled 9dCTD and 9aCTD, respectively. (A & C) The changes in chemical shifts upon complex formation of the 1b crosspeaks at 20 °C and the 1a crosspeaks 10 °C, respectively. (B & D) The change in relative intensity upon complex formation of the NH crosspeaks of the 1b and 1a peptides, respectively. The relative intensity changes of the crosspeaks were normalized to the intensities of the crosspeaks of G30 in the 1aNTD spectra and G35 in the 1bNTD spectra, which were set as 1.0. The changes in the intensities of the crosspeaks of 1bNTD upon complex formation suggest that the resonances from residues 12 and 14-17 are broadened due to interaction with residues from the 9dCTD, which are undergoing chemical exchange between multiple conformations. In contrast, there is no evidence of exchange broadening of the crosspeaks in the 1aNTD/9aCTD complex.

The changes in the 1bNTD crosspeaks upon binding 9dCTD contrast with those of 1aNTD in complex with 9aCTD. The changes in the chemical shifts of homologous residues of ¹⁵N-1bNTD are much larger than those of 1aNTD upon complex formation (Figure 3C, 3D; residue 1 of 1aNTD is homologous to residue 6 of 1bNTD). In addition, the relative intensities of all crosspeaks in the 1aNTD HSQC spectrum from the helical residues are similar although lower than those of the uncomplexed peptide. This suggests the broadening is due to slower tumbling of the entire molecule upon complex formation, rather than exchange broadening. The lack of exchange broadening of the 1aNTD spectrum upon complex formation with the 9aCTD is consistent with the small changes in the T1, T2 and HNOE data of the 1aNTD in complex with the 9aCTD previously reported (10), which show that most of the NTD residues in the 1a9a complex are ordered and have similar HNOEs and relaxation rates.

The secondary structure of ¹⁵N¹³C-1bNTD complexed with unlabeled 9dCTD

The H^α, H^N, C^α and C’ chemical shift displacements of 1bNTD in complex with 9dCTD are almost identical to those of the unbound 1bNTD (9). There were slight changes in the chemical shifts of two of the ¹³C^α resonances suggesting that S4 and S5 become slightly more helical upon binding, but the rest of the secondary structure is almost unchanged. Residues 6-36 are helical and have coiled-coil interactions as in the free 1bNTD structure (9). This indicates that most of the increase in helical content upon complex formation observed using CD (Figure 1) is due to induction of helix in 9dCTD upon binding to 1bNTD.

Tertiary structure of 1bNTD in the 1bNTD-9dCTD complex

The structure of 1bNTD in a complex with unlabeled 9dCTD, was solved using AutoStructure 1.1.2 (34, 35) as described in Materials and Methods. When 1bNTD was bound to 9dCTD there were no observed intramolecular NOEs between the chains of the bound 1bNTD molecule for residues 1 to 15, but there were many intermolecular NOEs from the unlabeled 9dCTD to the labeled 1bNTD. The intermolecular CTD to NTD resonances could be uniquely assigned as arising from interactions between the CTD and NTD chains from the X-filtered ¹³C-edited NOESY spectrum collected when both chains of 1bNTD were labeled and both chains of 9dCTD were unlabeled. The intermolecular NOEs were removed from the data set before the structural calculations were performed. Despite having no intramolecular-interchain NOEs between the chains of the 1bNTD molecule for residues 1-15, there were sufficient intrachain and interchain NOEs from residues 16-37 to define its structure and to determine the conformational changes in the 1bNTD upon binding. Table 2 gives the statistics for the fit of the structure to the NMR data.

Table 2.

Summary of NMR Constraint Data and Structure Quality Statistics for 1bNTD Bound to 9dCTD

NMR distance and dihedral constraints
Distance constraints
Total number of NOE-derived constraints	3276
Intra-residue	1544
Inter-residue
Sequential (|i-j| = 1)	740
Medium-range (1 < |i - j| <= 5)	864
Long-range intra-chain (|i-j| > 5)	0
Long-rang inter-chain interactions	128
Hydrogen bond constraints	112
Dihedral angle restraints
ϕ	128
χ	128
Constraint Violations (mean and s.d.)	0
Average number of distance constraint violations 0.2-0.5 (Å)	0
Total number of close contact violations > 0.2 Å	2
Deviations from idealized geometry
Bond lengths (Å)	0.007
Bond angles (ο̱)	0.7
Impropers (ο̱)	0
Average pairwise r.m.s.d. (Å)
All residues 10 models out of 200
Heavy atoms	2.25
Backbone atoms	2.01
Helical Residues, (residues 6-33), 10 Models out of 200
Heavy atoms	1.48
Backbone atoms	0.87
Ramachandran Plot Summary from Procheck (62)
Most favored regions	94.6%
Additionally allowed regions	4.6%
Generously allowed regions	0.9%
Disallowed regions	0.1%
Other Plot Statistics from Procheck (62)

	Number of data points	Mean Parameter Value	Comparison Typical Values	Value Band Width	Equivalent resolution (Å)
H-bond energy st. dev.	10	0.7	1.3	0.2	1.4 Å
Chi-1 pooled st. dev.	10	12.7	33.5	4.8	1.3 Å
Chi-2 trans st. dev.	10	8.4	31.2	5	1.0 Å

In unbound 1bNTD at 8 °C, residues 1-5 are disordered and there are no sequential NOEs from these residues, indicative of a disordered conformation (9). In contrast, the NOESY spectra for the 1bNTD in complex with unlabeled 9dCTD at 10 °C showed many NOEs characteristic of α-helices for residues 2 to 5. There were strong H^Ni to H^Ni+1 NOEs between G2 and S3, S3 and S4, S5 and L6 and several H^α_i to H^β_i+3 and H^α_i to H^N_i+3 from S4 to E7 and S5 to A8, showing that these residues are ordered when 1bNTD is bound to 9dCTD. We were not able to determine the structure of the bound 1bNTD at 10 °C because the H^N resonances from residues 10 to 16 were too broad due to intermediate chemical exchange to permit assignment of the resonances from the side chains. In addition, there were no unique intramolecular-interchain NOEs arising from residues 1-26 that could be assigned by AutoStructure. At 20 °C the spectra sharpened and almost all the resonances of the molecule could be assigned, allowing determination of the structure, but the H^N_i to H^N_i+1 NOEs arising from residues 3 to 5 were broadened. Sequential i to i+3 and i to i+4 NOEs characteristic of α-helices between residues S4 with E8 and S5 with A8 were observed, suggesting that S4 and S5, which are disordered in the unbound 1bNTD structure, are exchanging between helical and non-helical conformations at 20 °C.

The backbone structure of 1bNTD bound to 9dCTD is shown in Figure 4A (in cyan) and compared to the free peptide (black). When 1bNTD binds to 9dCTD residues S4 and S5 become helical and residues 6-18 splay apart to bind to the 9dCTD. The structure of bound 1bNTD is less well defined than free 1bNTD because the intermolecular-interchain NOEs to residues 1-16, needed to constrain the fits of the data, were not identified since the 9dCTD was unlabeled. In addition, some resonances were broadened due to chemical exchange. The N-terminal five residues are probably more ordered than they appear. For example there were clear intermolecular NOEs to the α-protons of G2 and β-protons of S3 from a leucine methyl group (possibly L274), which would constrain its conformation, but they could not be used in the structure determination since they could not be positively identified. There are numerous intermolecular NOEs to residues 1-16 showing that these residues are making contact at least some of the time when 1bNTD binds to 9dCTD. The refined structure has no upper limit distance constraint violations and no interatomic clashes. The ¹⁵N-edited NOESY spectrum suggested that at 10 °C more of the N-terminal extension is helical in the complex, at least transiently, than at 20 °C. Availability of labeled 9dCTD would improve the structure determination of 1bNTD in the complex. We constructed a 9dCTD clone but were unable to obtain significant expression in E. coli under a variety of conditions.

Figure 4.

Changes in (A) the structure of 1bNTD upon binding to 9dCTD and (B&C) comparison of the bound structure with that of the 1aNTD/9aCTD complex. (A) (black) 1bNTD alone (9), RCSB file 1ihq; (cyan) 1bNTD in the 1bNTD/9dNTD complex, RCSB file 2g9j. In unbound 1bNTD residues 1-5 are disordered and residues 6-19 form a canonical coiled coil, while residues 20-24 from the GCN4 portion of the molecule are also coiled coil. Upon complex formation, residues 4 and 5 become helical, and residues 6-22 splay apart. (B) The backbone structure of (cyan) 1bNTD when complexed to 9dCTD compared to the structure of (red) 1aNTD complexed to 9aCTD. Residues 1-29 of 1aNTD are aligned with the homologous residues 6-34 of 1bNTD. In A and B, the bonds from the peptide backbone are illustrated and the residues in a and d positions of the 1bNTD coiled-coil heptad repeat are labeled. C. Ribbon depictions of the mean backbone structures of the (red) 1aNTD complex compared with the (cyan) 1bNTD complex. The helical residues from the GCN4 portions of the molecules are aligned (residues 15-29 for 1aNTD/9aCTD and 20-34 for the 1bNTD/9dCTD complex). The coordinates of the 1aNTD/9aCTD complex are from the RCSB file 2g9j. The structures of the two NTDs are similar except the helical region of 1bNTD is longer. The figures were drawn using MolMol (64).

The structure of 1bNTD bound to 9dCTD is similar to that of 1aNTD in complex with 9aCTD, the striated muscle tropomyosin complex (Figures 4B, 4C). Figure 4B depicts the backbone bonds of 10 structures of 1bNTD (cyan) and 1aNTD (red) where the homologous helical residues are aligned. Figure 4C shows ribbon models of the mean structures aligned by the GCN4 regions of the molecules. The 1bNTD has a longer helical region since S4 and S5, which have no counterparts in 1aNTD, are helical in the complex. The tropomyosin region of the 1bNTD is slightly more splayed than the 1aNTD; the GCN4 portions of the two molecules are almost identical.

Molecular modeling of the 1bNTD/9dCTD complex

The intermolecular NOEs between the 9dCTD and 1bNTD gave sufficient information to define the length of the region of the 1bNTD that interacts with the 9dCTD and the residues that are in the intermolecular interface. The experimental data showed that residues G2 to Q17 from the 1bNTD are the binding site for the 9dCTD. (1) The residues of the labeled 1bNTD molecule that have interactions with the unlabeled 9dCTD were identified from their assigned ¹³C and ¹H resonances; there were strong intermolecular NOEs from the 9dCTD to residues G2, S3, L6, A8, V9 and I13 from the 1bNTD. (2) The cross-peaks in the ¹H-¹⁵N HSQC spectrum from S4 to L16 of the 1bNTD were greatly shifted upon binding to the 1bCTD. (3) The chemical shifts of the methyl groups from L6, V9 and I13 and L16 were all displaced upon complex formation. (4) The crosspeaks of the amide side-chain of Q17 were displaced upon formation of the complex. Since the secondary structure of the 1bNTD was unchanged upon binding, these results provide strong evidence that residues G2-Q17 are all in the intermolecular interface. There was also only one set of NOEs observed for each residue, showing that the complex is symmetric. This finding eliminated a non-interleaved overlap structure.

The structures of bound 1bNTD and bound 1aNTD are similar, the 1bNTD/9dCTD complex is symmetric, and the hydrophobic residues of the 1bNTD with intermolecular NOEs are homologous to the hydrophobic residues the 1aNTD that interact with 9aCTD to form an overlap complex. Based on these findings, the structure of the 1bNTD/9dCTD complex was homology modeled using the coordinates of the NMR structure of the 1aNTD/9aCTD complex (13) (see Materials and Methods) (Figure 5). The overall structure of the model of the 1bNTD/9dCTD is similar to that of the solution structure of the 1aNTD/9aCTD complex (13). The main difference is that the overlap region in the 1bNTD/9dCTD model is longer than in the 1aNTD/9dCTD complex. Residues 2-17 of 1bNTD interact with ~residues 267-284 from 9dCTD while residues 1-11 of 1aNTD interact with residues 274-284 of 9aCTD in the striated muscle overlap complex.

Figure 5.

Comparison of a model of the 1bNTD/9dCTD complex with the structure of the 1aNTD/9aCTD determined by NMR spectroscopy (13). The backbones are illustrated by ribbons. (A) Homology model of the 1bNTD/9dCTD complex based on model 1 of the 1aNTD/9aCTD complex, RCSB pdb file 2g9j. (B) 1aNTD/9aCTD. The NTDs are in cyan and the CTDs are in brown. The overall structures of the complexes are similar but the overlap region is ~17 residues compared to ~11 for the 1aNTD/9aCTD complex. The figures were drawn using MolMol (64).

The interatomic distances < 5 Å in the model structure were calculated using MolMol. The observed intermolecular NOEs to the 1bNTD are mainly to hydrophobic a and d residues predicted to be at the NTD/CTD interface: L6, V9, I13 and L16. The observed intermolecular NOEs arising from residues of 1bNTD are consistent with most of the predicted interactions, suggesting that the model is a fair approximation of the actual interface structure and that the non-muscle overlap region is homologous to the striated muscle overlap complex determined by NMR (13). Since the 9dCTD was unlabeled, we evaluated whether the observed intermolecular NOEs from the unlabeled 9dCTD to the labeled 1aNTD were consistent with what would be predicted from the model. The methyl group proton chemical shifts of H^ε of Met (~ 2.1 ppm), H^γ of Thr (~1.3 ppm) and H^δ of Leu (~0.9 ppm) are distinctive, so some of the 9dCTD a and d interface residues in the overlap region, could be tentatively identified.

Table 3 compares the residues which were predicted to have intermolecular NOEs in the model and the observed NOEs, with the observed NOEs in the 1aNTD/9aNTD complex. The model predicts strong interactions of L6 with M273, L274, L278 and T277, A8 with L278 and L279, V9 with L278, L281 and M284 and I13 with L278, L281 and M284. NOEs between resonances of 1bNTD with resonances having chemical shifts consistent with these interactions were observed.

Table 3.

Intermolecular NOEs Consistent with the Model of the 1bNTD/9dCTD Complex.

Residue	Consistent Predicted NOEs	Predicted NOEs Not Observed	Homologous Residue in 1aNTD/9aCTD Complex	Observed NOEs in 1aNTD/9aCTD Complex
L6	M273, L274 and/or L278, T277		M1	I270, E273, L274, A277, L278
A8	L278		A3	L278
V9	L278 and/or L281, M284	T277	I4	L278, M281, I284
R10	L281, E280	T277	K5	A277, L278, M281, I284
K12*	M284	L278, N282	K7	M281, T282, I284
I13	L278 and/or L281, M284		M8	M281, I284
L16*		M284	L11	M281, I284
Q17#	L281, M284		K12	I284

^*Crosspeaks from K12 and L16 were broadened upon complex formation.

^#The crosspeaks from the side-chain amides of Q17 were shifted upon complex formation.

The major differences between the predicted and observed NOEs in our model were in the N-terminal extension, A1-S5, that vary in conformation among the ten NMR models (Figure 5A); we did not include these interactions in Table 3. Some of the predicted interactions were not observed, and interactions between S3 and a leucine residue (L274 or L278) were observed that were not predicted. The major differences were that predicted interactions of L16 with M284 were not observed, as well as extensive predicted interactions with K12. Intramolecular crosspeaks from the side-chains of these two residues in the NOESY spectra were weak, suggesting these residues undergo exchange broadening at 20 °C. The crosspeaks from both of these residues, however, were displaced from those of the unbound 1bNTD, suggesting they do indeed form part of the 1bNTD/9dCTD interface. Predicted NOEs between resonances of V9 with T277 were also not observed, possibly because T277 in 9dCTD is less hydrophobic than A277 in 9aCTD.

While the pattern of NOEs in the 1bNTD/9dCTD complex is consistent with major interactions between L6-L16 and residues 274-284 of the 9dCTD (homologous to the structure of the 1aNTD/9aNTD overlap complex), L6 to L16 would have a similar pattern of NOEs if L6 were aligned with T277 rather than L274. Both models have a similar spectrum of predicted NOEs, although the relative intensities of the various peaks would be different. The pattern of NOEs, however, is inconsistent with L6 being aligned with N267, M270 or M281. The correct alignment must await determination of the full structure. We have elected to show the model with the longer overlap region since it incorporates the sequence homology between 1aNTD and 1bNTD.

The solution structures of 1bNTD in the complex and the striated muscle overlap complex are not similar to the crystal structure of the striated muscle overlap complex (RCSB 2z5i, (43)) in which one molecule of an N-terminal TM model peptide dimer interacts with two molecules of a C-terminal TM model peptide dimer. The extensive hydrophobic intermolecular interactions of L6 and I13 with the 9dCTD and observed intermolecular NOEs from the 9dCTD to resonances from the side-chains of R10 are incompatible with the 1bNTD/9dCTD complex having such a structure.

DISCUSSION

A distinguishing characteristic of LMW tropomyosins expressed in non-muscle cells, in addition to being shorter, is the presence of a N-terminal extension of five residues in mammalian tropomyosins that is disordered in solution (9). When an N-terminal model peptide, 1bNTD, binds to a C-terminal peptide, 9dCTD, the extension becomes more helical. Residues 2-5, along with the following 11 residues, are engaged in binding that also induces helical formation of 9dCTD. The homologous regions of both HMW and LMW N-terminal domains have similar coiled coil structures.

A homology model of the LMW intermolecular overlap complex was constructed using the coordinates of the HMW striated muscle complex (13), initially based on sequence homology and supported by NMR data. The structures of the two complexes are similar overall in being symmetric and in the perpendicular orientation of the planes of the coiled coils in the complex. The major difference is that the overlap region of the LMW complex is 16 residues, compared to 11 for the HMW complex because of the involvement of the N-terminal extension. Second, although the affinity of the LMW ends for each other is higher than that of the HMW ends, at physiological temperature much of the C-terminal region is disordered and exchanges between ordered and disordered states. That is, the unfolding of the complex is uncoupled from dissociation, in contrast to other complexes we have studied.

A major outcome of the solution structure of the striated muscle tropomyosin junctional complex is the insight gained into how the proposed periodic actin binding sites on tropomyosin relate to the actin subunits in the actin filament (14). The seven-fold periodicity in the striated muscle tropomyosin sequence was recognized when the sequence became available (14, 44, 45). The pitch of the supercoil is variable along the length of the molecule (38), and one might intuitively expect there to be seven half-turns of the supercoil per molecule that correspond to the seven actin subunits along its length of one HMW tropomyosin molecule. Our analysis of the 7 Å structure using TWISTER (46) showed there are only ~5.75 half turns of the supercoil (13). The perpendicular orientation of the planes of the coiled coils of the molecules that form the complex “corrects” for the quarter turn. Modeling of tropomyosin molecules joined by the overlap complex on the actin filament showed that the proposed actin binding sites from one tropomyosin molecule to the next present a similar face to the actin subunits, even though the repeats do not correspond to supercoil repeats. Were the planes parallel, as in earlier models (47, 48), this would not be the case.

Does the longer N terminus of LMW tropomyosin change the alignment of the proposed actin binding sites with respect to the actin filament? To address this question we modeled the LMW overlap complex into the striated muscle tropomyosin structure (See Materials and Methods; Compare Figures 6A and 6B). The arrangement of the surface residues in tropomyosin with the LMW junction is comparable to that with the HMW forms (Compare Figure 6C and 6E with 6D and 6F, respectively).

Figure 6.

Models of full-length striated and non-muscle tropomyosin molecules associated head-to-tail. (A) Two molecules from the 7 Å structure of striated muscle TM (38) linked via one model structure of the NMR ensemble of the 1aNTD/9aCTD complex (13). The model is shown in ribbon format (cyan) with the side chains (carbon atoms) of “consensus residues” in each of the seven proposed actin binding sites shown in spheres (black). The “consensus” residues are a repeated motif of seven polar and non-polar surface residues that have the periodicity and spacing to be actin binding sites (14). Mutagenesis experiments have shown that they are required for tropomyosin to bind to F-actin (51). (B) Model of two molecules of LMW α-tropomyosin linked by the 1bNTD/9dCTD complex. (C and D) Details of three sequential proposed actin binding sites including the one in the intermolecular junction. (C) 1aNTD/9aCTD. (D) 1bNTD/9dCTD. (E and F) Views of the junctional regions of the 1aNTD/9aCTD model (E) and the 1bNTD/9dCTD model (F) highlighting (in black) single side chains of the first residue from consensus sites 7, 1 and 2 in the 1aNTD/9aCTD complex and sites 6, 1 and 2 in the 1bNTD/9dCTD complex. The side chains are shown on one side of each coiled coil. In the 1aNTD/9aCTD complex they are R244, K6 and K48, respectively, showing their similar orientations. In the 1bNTD/9dCTD short tropomyosin complex, the homologous residues are R207, R11 and R53. The models show that the consensus residues side chains of the three sequential periods have similar orientations relative to the axis of the coiled coil. In both models, the planes of the N termini relative to the planes of the C termini are rotated by 90°. As a result, the orientation of the proposed consensus residues (shown in black) of the actin binding sites on the surface is retained from one molecule to the next along the filament. The figures were drawn using UCSF-Chimera (65).

We have considered how the recombinant LMW α-tropomyosins have higher actin affinity than HMW α-tropomyosins even though there are six versus seven postulated actin binding sites. Both circular dichroism and NMR data show that although the ends encoded by exon 1b and exon 9d form a tight inter-molecular junction, the overlap region is relatively disordered and undergoes chemical exchange between multiple conformations. Conformational flexibility (instability) is essential for formation of the 1aNTD/9aCTD complex and for troponin T binding (49) as well as for the binding of tropomyosin to actin (50, 51). The disordered 9d C-terminus may facilitate strong actin binding (5, 52) since the instability gives the conformational freedom to find the correct binding partners. We also suggest that the longer N-terminal region encoded by exon 1b contributes to the greater actin affinity of recombinant LMW rat α-tropomyosins (5) and the influence of the N terminus on the regulatory state of the actin filament (53). Support for the idea that the N-terminal region (AGSSSLEAVRRKIR) may encode an actin binding site is its homology to an internal cofilin sequence (rat cofilin residues 114-127, IYASSKDAIKKKLT) that is helical in solution (54) and in X-ray structures of yeast and bovine cofilins (55). The region is part of and/or close to an actin binding site on cofilin (56-58). Cofilin binds with high affinity to the actin filament. Tropomyosin is a competitive inhibitor of cofilin binding and severing (59-61). Tropomyosin’s ability to inhibit cofilin binding and severing may reflect direct competition for binding sites on the actin filament, as well as its generic ability to stabilize the actin filament against severing proteins.

Abbreviations

1aNTD (αTM1a_1-14Zip) and 1bNTD (αTM1b_1-19Zip)

model peptides containing residues 1-14 of rat striated muscle α-TM encoded by exon 1a or residues 1-19 of rat non-muscle α-TM encoded by exon 1b, respectively and the last 18 residues of GCN4. Both peptides have a glycine at the N-terminus replacing the native acetyl group

9aNTD (αTM9a_251-284)

a peptide containing residues 251-284 of rat striated α-striated muscle tropomyosin encoded by exon 8 and exon 9a with the sequence GCG at the N-terminus to allow cross-linking

9dCTD (αTM9d_252-284)

a peptide containing residues 252-284 of rat smooth and some non-muscle α-tropomyosins encoded by exon 8 and exon 9d with the sequence GCGG at the N-terminus to allow cross-linking

CD

circular dichroism

GCN4

a yeast transcription factor

HMW

high molecular weight

HSQC

heteronuclear single quantum coherence

LWM

low molecular weight

NOE

nuclear Overhauser effect

NOESY

NOE spectroscopy

NOE crosspeaks

proton/proton interactions detected in nuclear Overhauser effect spectroscopy

r.m.s.d.

root-mean-square deviation

TM

tropomyosin