The Sequence of the Alternatively Spliced Sixth Exon of α-Tropomyosin Is Critical for Cooperative Actin Binding but Not for Interaction with Troponin*

Tropomyosins, a family of highly conserved coiled-coil actin binding proteins, can differ as a consequence of alternative expression of several exons (Lees-Miller, J., and Helfman, D. (1991) BioEssays 13, 429–437). Exon 6, which encodes residues 189–213 in long, 284-residue tropomyosins, has two alternative forms, exon 6a or 6b, both highly conserved throughout evolution. In α-tropomyosin, exon 6a or 6b is not specific to any one of the nine isoforms. Exon 6b encodes part of a putative Ca2+-sensitive troponin binding site in striated muscle tropomyosins, suggesting that the exon 6-encoded region may be specialized for certain tropomyosin functions. A series of recombinant, unacetylated tropomyosin exon 6 deletion and substitution mutants and chimeras was expressed in Escherichia coli to determine the requirements of exon 6 for tropomyosin function. Functional properties of the tropomyosins were defined by actin affinity measured by cosedimentation, troponin T affinity using a newly developed biosensor assay, and regulation of the actomyosin MgATPase. The region of tropomyosin encoded by exon 6 affects actin affinity but not thin filament assembly, troponin T binding, or regulation with troponin. The tropomyosins with exon 6a or 6b function normally whether a striated muscle exon 9a or smooth/non-muscle exon 9d is present. However, the effect of deleting 21 amino acids encoded by exon 6 or replacing it with a GCN4 leucine zipper sequence depends on the COOH-terminal sequence.

Tropomyosins, a family of highly conserved coiled-coil actin binding proteins, can differ as a consequence of alternative expression of several exons (Lees-Miller, J., and Helfman, D. (1991) BioEssays 13, 429 -437). Exon 6, which encodes residues 189 -213 in long, 284-residue tropomyosins, has two alternative forms, exon 6a or 6b, both highly conserved throughout evolution. In ␣-tropomyosin, exon 6a or 6b is not specific to any one of the nine isoforms. Exon 6b encodes part of a putative Ca 2؉sensitive troponin binding site in striated muscle tropomyosins, suggesting that the exon 6-encoded region may be specialized for certain tropomyosin functions.
A series of recombinant, unacetylated tropomyosin exon 6 deletion and substitution mutants and chimeras was expressed in Escherichia coli to determine the requirements of exon 6 for tropomyosin function. Functional properties of the tropomyosins were defined by actin affinity measured by cosedimentation, troponin T affinity using a newly developed biosensor assay, and regulation of the actomyosin MgATPase. The region of tropomyosin encoded by exon 6 affects actin affinity but not thin filament assembly, troponin T binding, or regulation with troponin. The tropomyosins with exon 6a or 6b function normally whether a striated muscle exon 9a or smooth/non-muscle exon 9d is present. However, the effect of deleting 21 amino acids encoded by exon 6 or replacing it with a GCN4 leucine zipper sequence depends on the COOH-terminal sequence.
Tropomyosins (TM) 1 are a family of highly conserved coiledcoil actin binding proteins present in most eukaryotic cells. At least 15 different isoforms arise through the use of alternative promoters and alternative RNA splicing of the transcripts of a small number of genes (three to four in vertebrates; reviewed in Ref. 1). These isoforms are expressed in developmentally and tissue-specific patterns and differ in actin affinity.
A function common to all TMs is cooperative binding to F-actin (2). Tropomyosin molecules are aligned head-to-tail in the grooves of the helical actin filament (3,4). The role of TM is best understood in striated muscle where it regulates Ca 2ϩ -dependent muscle contraction with Tn (reviewed in Refs. [5][6][7]. Structural studies have shown that Tn, found only in striated muscles, extends along at least the COOH-terminal third of the TM molecule (8 -13). Troponin I, TnC, and the COOH terminus of TnT are positioned near Cys-190 of TM. The elongated NH 2 terminus of TnT extends beyond the COOH terminus of one TM to the NH 2 -terminal 10 -30 residues of the next TM along the actin filament (13,14).
Troponin greatly increases the affinity of TM for actin in the presence of Ca 2ϩ , with a further increase upon removal of Ca 2ϩ (15). Based on binding studies with TM and Tn peptides, Mak and Smillie (16) proposed that the ␣-TM-Tn interaction in the region of Cys-190 is weakened in the presence of Ca 2ϩ (Ca 2ϩsensitive), whereas that at the COOH terminus and the overlap region is strong in the presence and absence of Ca 2ϩ (Ca 2ϩindependent). Substantial evidence has since supported this model.
The presence of tissue-specific alternatively expressed exons implies distinct functions for the regions of TM they encode. For example, of four exons in the ␣-TM gene encoding the COOH terminus (9a-9d), exon 9a, important for Tn interaction on the thin filament (17,18), is uniquely expressed in striated muscles. Of the two sixth exons, only exon 6b is expressed in striated ␣and ␤-TMs, although it is not restricted to striated isoforms. The exon 6b-encoded region of striated muscle TM (residues 189 -214) is part of the putative Ca 2ϩ -sensitive Tn binding site.
To determine the requirements of the alternatively spliced exon 6 for TM function, we expressed a series of recombinant, unacetylated TM exon 6 deletion and substitution mutants and chimeras in Escherichia coli. Functional properties of the modified TMs were defined by actin affinity, TnT affinity, assembly with Tn on the thin filament, and regulation of the actomyosin MgATPase. We have shown that the region of TM encoded by exon 6 affects actin affinity but not TnT binding, assembly with Tn on the thin filament, or thin filament regulation. Portions of this work have been reported in a preliminary form (19).

DNA Constructions and Protein
Purification-General recombinant DNA techniques were performed as described in Sambrook et al. (20) or as recommended by the supplier.
The oligonucleotides were synthesized on an Applied Biosystems DNA synthesizer, purified using an NENSORB cartridge, and phosphorylated enzymatically using phage T4 polynucleotide kinase (24). The synthesis reaction was carried out for 30 min at 37°C. Competent E. coli DH5a cells were transformed with a portion of the synthesis reaction, then plated for colony isolation. Single colonies were reisolated and then used for DNA preparation. The DNA was screened for mutants using restriction analysis. The complete cDNA sequence was determined using the dideoxynucleotide chain termination method with ␣-35 S-dATP (25).
Conformational Analysis-Circular dichroism spectroscopy was carried out by Dr. N. J. Greenfield to measure the folding of TM as a function of temperature (31,32). The TM exon 6 variants showed multiple transitions with virtually the same degree of folding at 20°C, with the exception of TM⌬6-9d which is only about 80% folded (Table I) (32). The COOH-terminal 27 residues encoded by the striated musclespecific exon 9a is associated with an increase of the overall thermal stability of TM, the magnitude depending on the exon 6 variant (Table  I). In general, the differences can be understood in terms of changes in the hydrophobicity of the a and d residues of the heptad repeat that form the interface between the two ␣-helices of the coiled coil (32). It has not been possible to relate actin affinity to the thermal stability of these or other recombinant TMs.
Actin Binding Assays-Binding of TM to actin was measured by cosedimentation at 25°C in a Beckman model TL-100 centrifuge as described previously (33) with modifications (34). The bound and free TM were determined by quantitative densitometry of SDS-polyacrylamide gels stained with Coomassie Blue (34).
Apparent binding constants (K app ) and Hill coefficients (␣ H ) were determined by using SigmaPlot (Jandel Scientific) to fit the data to the equation The TM/actin ratio determined using densitometry (arbitrary units) was normalized using the n reported by SigmaPlot. We have shown that saturation corresponds to a TM:actin molar ratio of 0.14, a stoichiometry of 1 TM:7 A.
Real Time Kinetic (BIAcore) Analysis of TnT Binding to TM Exon 6 Variants-A BIAcore (Pharmacia Biotech Inc.) biosensor was used to measure binding of TM in solution to immobilized TnT. Chicken muscle TnT was dialyzed against 10 mM Hepes, pH 7.0, 500 mM NaCl, 0.5 mM DTT at 4°C overnight prior to immobilization on a sensor chip CM5 in the BIAcore system using an amine coupling kit (Pharmacia, Biosensor). Thirty l of TnT diluted to 0.33 mg/ml in 10 mM sodium borate buffer, pH 8.5, 300 mM NaCl, were immobilized on the carboxylated matrix of a sensor chip activated by a 20-l injection of a solution containing a 1:1 mixture of 0.2 M N-ethyl-NЈ-(3-diethylaminopropyl)carbodiimide and 0.05 M N-hydroxysuccinimide. Excess reactive groups were blocked with a 30-l injection of ethanolamine hydrochloride.
Assays were carried out at 25°C in a 5 l/min flow of buffer (150 mM NaCl, 10 mM Tris-HCl, pH 7.5, 2 mM MgCl 2 , 0.5 mM DTT). The density of TnT immobilized on the surface measured 3358 response units (RU) for the exon 6 variants and 2652 RU for the exon 9 variants, corresponding to 12-15 ng/mm 2 . Based on the RU when TM binds, the 1:1 stoichiometry of TM:Tn, and knowing that the surface plasmon resonance response is proportional to the surface concentration of bound protein (35), approximately 30% of the surface was active. When striated muscle TM was repeatedly injected, with regeneration of the surface between injections, the amount of bound TM was reproducible to within a few percent (730 Ϯ 13 RU).
An analytical cycle consisted of initiation of the association phase at 120 -260 s by injection of 20 l of TM (0.5-12 M) at 25°C in 150 mM NaCl, 10 mM Tris-HCl, pH 7.5, 2 mM MgCl 2 , 0.5 mM DTT at 5 l/min across a surface of immobilized TnT. The dissociation phase was initiated at 360 -380 s by switching to continuous flow of buffer without TM, and finally, regeneration of the surface with 10 l of 2.5 M NaCl at 5 l/min.
BIAevaluation software (Pharmacia, Biosensor, version 2.1) was used for data processing. Dissociation rate constants (k d ) were determined by fitting the dissociation data to the equation: R t ϭ R 0 e Ϫkd1(tϪt0) ϩ(R 0 ϪR 1 )e Ϫkd2(tϪt0) which calculates the dissociation rate of two parallel reactions, where R t is the relative response (RU) (representing TM bound to TnT) at time t; R 0 is the response at the start of dissociation; k d1 is the dissociation rate constant (s Ϫ1 ) for the first phase of the biphasic curve; k d2 is the dissociation rate constant (s Ϫ1 ) for the second phase of the biphasic curve; t 0 is the start time for dissociation. The part of the sensorgram usable for analysis is the most linear region of a plot of ln(abs(dY/dX)) versus time after the initial bulk refractive index "hump." Generally, t 0 was about 10 -20 s and t was about 160 -240 s after the start of elution (indicated by the region between the solid vertical lines in Figs. 5 and 6). Association rate constants (k a ) were calculated from the calculated k t values at different TM concentrations according to the following association type one model: R t ϭ Req1(1 Ϫ e Ϫka1⅐C⅐nlϩkd1(tϪt0) ) ϩReq2(1 Ϫ e Ϫka2⅐C⅐n2ϩkd2(tϪt0) ), which calculates the association rate constants k a1 and k a2 for a heterogenous interaction, Req is steady state, C is the molar concentration of TM; n is the steric interference factor. Generally, t 0 was about 10 -15 s and t was about 130 -185 s after injection of TM (indicated by the region between the broken vertical lines in Figs. 5 and 6). Apparent binding constants (K app ) were calculated by dividing the average k a (determined at different concentrations of each TM variant) by the average k d . Applying the tests recommended by Schuck and Minton (36), our data are valid when evaluated using a two-site model.
Actomyosin MgATPase Assay-The actomyosin MgATPase was measured as a function of TM concentration in a thermoequilibrated Molecular Devices ThermoMax microtiter plate reader (18). The amount of inorganic phosphate released was determined colorimetrically according to White (37) and the plates were read in a Molecular Devices ThermoMax microtiter plate reader with a 650-nm filter. Time courses had been carried out previously to show that phosphate liberation was linear. The assays were single point determinations.

Rationale-
The alternatively spliced exon 6 of TM encodes residues 189 -214 in striated TMs and is part of the putative Ca 2ϩ -sensitive Tn binding site. To define the requirements of the exon 6-encoded region for TM function, we made three exon 6 variants by deleting or replacing exon 6b in smooth TM (TM6b-9d, Fig. 1A). 1) Residues 191-212 of exon 6b were deleted (TM⌬6-9d). Only 21 of the 25 residues were deleted to maintain the heptapeptad repeat of hydrophobic residues important for the stability of the coiled coil. The deletion corresponds to one-half an actin binding period, based on there being seven quasiequivalent periodic sites (40). 2) Residues 191-211 were replaced with the corresponding sequence from the alternative ␣-TM exon 6a (TM6a-9d).
3) The entire sixth exon, residues 189 -213, was replaced with a non-TM coiled-coil sequence from the yeast transcriptional activator, GCN4 (TM6zip-9d) (23). Residue 212 in the original GCN4 sequence was changed from Gly to Ala to assure an uninterrupted coiled coil. These three exon 6 variants were expressed in E. coli to produce unacetylated TMs with two different COOH-terminal exons: exon 9d, found in smooth and most non-muscle TMs (Fig. 1A) and exon 9a, unique to striated TM (Fig. 1B). All of the exon 6 variants contain the unique smooth ␣-TM exon 2a.
Actin Affinity of Tropomyosin Exon 6 Variants-The actin affinity of the TM exon 6 variants was measured by cosedimentation (Fig. 2, Table II). TM6b-9d (smooth muscle ␣-TM) bound to actin with high affinity, as previously reported (17,18). Replacing 21 residues of exon 6b with 6a (TM6a-9d) resulted in a 2-fold increase in actin affinity, consistent with previous results (41). Replacement of exon 6b with a GCN4 leucine zipper sequence (TM6zip-9d) or deletion of half an actin binding period (TM⌬6-9d) reduced the actin affinity so it could not be accurately measured. The effect of the leucine zipper replacement was surprising because the 284 residue length of the molecule and the coiled-coil ␣-helix were maintained.
All of the exon 6 variants with an exon 9a-encoded COOH terminus bound weakly compared with TM6a-9d or TM6b-9d (Fig. 2, Table II), as does unacetylated striated ␣-TM (17,18). Surprisingly, deletion of exon 6 had little effect while replacement with the GCN4 leucine zipper sequence significantly reduced actin affinity. It is unclear how the effect of deleting 21 residues depends on exon 9.
Actin Affinity of Tropomyosin Exon 6 Variants in the Presence of Troponin-If the region encoded by exon 6 contributes to the Tn binding site on TM, one might expect the TM exon 6 variants to differ significantly in the effect of Tn on thin filament assembly. Troponin in the presence of Ca 2ϩ caused a Ϸ100-fold increase for all TM variants containing exon 9a (Fig.  3A), whereas it had only a small effect on the affinity of exon 9d-containing TMs, except TM6zip-9d and TM⌬6-9d which did not change (Fig. 3B). The 4-fold lower actin affinity of TM6zip-9a compared with TM6a-9a was a similar effect to that reported for striated ␣-TM with GCN4 sequence inserted in the second or third periods of the molecule (42). Removal of Ca 2ϩ increased the affinity of all exon 9a-and 9d-encoded variants, independent of exon 6 identity (Fig. 3, A and B, and Table II).
To measure accurately the actin affinities in the absence of Ca 2ϩ , the NaCl concentration was increased from 150 to 300 mM to weaken actin affinity (Fig. 4). Together the results in Figs. 2-4 and Table II show that the relationship of the different exon 6 variants in terms of actin affinity remained constant with Tn in the presence and absence of Ca 2ϩ , as well as in the absence of Tn. Clearly, alterations in exon 6, or deletion of exon (f), TM6zip-9a. The data shown are from three experiments for TMs that bound well (TM6b-9d, TM6a-9d) and from one experiment for TMs that did not bind well (TM⌬6-9d, TM6zip-9d, TM6a-9a, TM⌬6-9a, TM6zip-9a). 6, primarily affect actin affinity, not Tn interaction on the thin filament. The results are consistent with previous reports showing the importance of exon 9 on Tn-enhanced TM affinity for actin (17,18).
Interaction of Tropomyosin Exon 6 and Exon 9 Variants with Troponin T-From the actin binding experiments, we can only infer the requirements of exon 6 for Tn interaction because Tn binds to both actin and TM. Existing methods used to evaluate TM-Tn interaction are qualitative or require covalent modification of TM (16, 43, 44). Thus, we have developed a solid phase affinity assay employing biosensor technology to measure directly the binding of TM with TnT. In this assay, TnT was immobilized on a dextran matrix attached to a gold-coated sensor chip surface to detect binding of different TM variants in a flow chamber. The amount of TM which bound to the immobilized TnT was calculated from the surface plasmon resonance signal measured as a function of time (45). Using this technique, we have measured TnT binding of the exon 6 variants described here, as well as of a series of previously reported exon 9 variants (18).
Qualitative comparison of sensorgrams obtained from different TMs binding to the same surface can be quite informative. Figs. 5, A and B, shows the results when a 1.5 M solution of each TM variant was allowed to bind to immobilized TnT. The data between the vertical broken lines were used to compare association rates and that between the vertical solid lines to compare dissociation rates. Fig. 5A shows sensorgrams of the exon 6 variants. Comparing the association phases (data between the vertical broken lines) clearly shows that TM⌬6-9a and TM6a-9a had the highest binding rates (steepest increase). Comparing the data between the vertical solid lines shows that TM6a-9a, TM⌬6-9a, and TM6zip-9a had similar (parallel) dissociation rates with significant residual signal at 480 s, reflecting measurable binding. In contrast, all the TM variants with exon 9d showed a very weak signal from 380 -480 s, reflecting poor binding. In general, dissociation and rebinding (an incline between the two vertical solid lines rather than a decline) was observed for the recombinant TMs which bound poorly to TnT. Fig. 5B shows sensorgrams of a series of exon 9 variants, chimeras and deletion mutants in which the last nine residues of exons 9a and 9d were exchanged or deleted (18) (described here in the legend to Fig. 5B. All expressed exons 2b and 6b). We previously reported that the last 9 residues encoded by exon 9d are important for actin binding whereas the striated specific exon 9a-encoded COOH terminus, especially the first 18 residues, are important for Tn-dependent thin filament assembly of unacetylated TM (18). Qualitative analysis of the sensorgrams in Fig. 5B shows that only TM9a and TM9a/9d, as well as acetylated striated ␣-TM isolated from muscle, bound to TnT in this assay. Replacing the first 18 residues of exon 9a with 9d (TM9d/9a), or deleting the last 9 residues (TM9a/) resulted in a major loss of TnT affinity, in contrast to deletion or replacement of exon 6 which had only a small effect. TM9d is particularly interesting because it appeared to have an extremely fast on-rate as well as an extremely fast off-rate, resulting in virtually no TM bound.
The five unacetylated TMs that bound measurably to TnT in Fig. 5, A and B, were quantitatively analyzed to obtain binding constants. On (k a ) and off (k d ) rate constants were determined at four different concentrations of each TM (Fig. 6). The apparent binding constant (K app ) was determined by dividing the average k a by the average k d (Table III). Evaluation of the sensorgrams suggested some rebinding of TM during the dissociation phase. This problem is normally addressed by adding the immobilized protein to the dissociation buffer. However, this was not possible because TnT is insoluble in 150 mM NaCl.
A two-site model fit the TnT binding data well for all of the exon 6 variants with an exon 9a-encoded COOH terminus with an apparent affinity on the order of 10 5 M Ϫ1 for the initial binding phase and 10 6 M Ϫ1 for the second phase, whereas the TMs with an exon 9d encoded COOH terminus bound too weakly to measure. The higher affinity of the second phase is generally attributed to a slower off rate than observed for the initial phase. The initial phase constitutes less than 30% of the total binding response.
The biphasic nature of the association and dissociation data has different possible explanations. Edwards et al. (46) have found that at low concentrations of analyte a single exponential model is sufficient to describe the binding data, whereas at high concentrations a double exponential model is required. This has been attributed to the dextran surface of the biosensor chip. However, we have found that a double exponential equation (2-site model) provides the best fit to our data at all concentrations of TM. Control experiments showed no significant binding of TM to the dextran surface alone, ruling out potential artifacts due to the dextran surface of the sensor chip. Steric hindrance on a crowded surface or inactivation by amine coupling of some TnT lysines required for TM binding can also result in a biphasic curve as high surface occupancy is approached (i.e. a 1:1 stoichiometry). It would be necessary to compare the binding properties of TnT immobilized at different densities using different coupling chemistries to rule out these possibilities. Furthermore, a conformational change may be necessary before or after an initial binding event between TM and TnT, resulting in a biphasic curve. Although not detected with steady state fluorescent assays (43,44), our data may reflect binding which requires more than one step. Because the on rate increases with increasing TM concentration, our data are consistent with a conformational change occurring before the initial binding (46). Finally, two separate binding sites with different affinities may exist. However, we only detected binding to variants with exon 9a suggesting both sites must involve this domain.
Substitution of exon 6 with a non-TM coiled-coil sequence (TM6zip-9a) did not affect either the first or second binding phase. Deletion of the sixth exon (TM⌬6-9a) slightly reduced K app1 but increased K app2 ϳ3-fold compared with the other exon 6 variants containing exon 9a. Clearly, the exon 6 sequence has little effect on TM affinity for TnT. The major determinant for TnT interaction is the COOH-terminal exon 9a, in particular the first 18 residues. The TM must be fulllength, as deletion of the last 9 residues from exon 9a resulted in loss of binding. These results, summarized in Table III, are consistent with the actin binding studies (this work) (18).
In control experiments, binding of TM directly to a blank dextran matrix was 0 -2 RU, demonstrating no measurable ionic interaction between the chip surface and TM. Therefore the fact that TMs which differ in sequence by as few as nine amino acids differ in the ability to bind TnT provides the best evidence that binding of TM variants to TnT is specific (compare TM9a with TM9a/ in Fig. 5 and Table III).
The acetylated muscle TM was not analyzed in detail because it polymerizes at 150 mM NaCl, complicating the interpretation and data analysis. By analyzing the sensorgram in Fig. 5B, it is estimated that acetylated TM binds TnT about 3-fold stronger than the unacetylated form (K app2 ϭ 8.5 ϫ 10 5 M Ϫ1 versus 2.3 ϫ 10 6 M Ϫ1 . K app1 values are equal). The affinity is similar to that reported by Lehrer and colleagues using a fluorescence assay (43,44). This results suggests that acetylation of striated ␣-TM has only a small effect on TnT affinity, in contrast to the large effect (Ϸ100-fold) on actin affinity (33,34,47).
Calcium Regulation of the Actomyosin MgATPase by Tropomyosin Exon 6 Variants with Troponin-The regulatory function of the TM exon 6 variants was evaluated by measuring their ability to confer calcium-dependent regulation of the actomyosin MgATPase with Tn. All TM variants that bound to actin with Tn inhibited the actomyosin ATPase in a Ca 2ϩsensitive manner with similar effectiveness (Fig. 7). In contrast, TM⌬6-9d and TM6zip-9d, which did not bind to actin in the conditions of the ATPase assay, did not regulate. In the presence of Ca 2ϩ , the ATPase activity was the same with all variants. For simplicity, only TM6b-9d is shown in the presence of Ca 2ϩ . These results are consistent with previous reports from this laboratory that TMs that can bind to actin with Tn are similar in their ability to regulate the actomyosin ATPase (17,18,27,48).

DISCUSSION
Analysis of a series of TMs that differ in the sequence encoded by exon 6 (residues 189 -214) has shown that this region of TM influences actin affinity but not TnT binding or assembly and regulation with Tn on the thin filament. The two native TM sequences, encoded by exon 6a and 6b, both function normally with a striated or smooth/non-muscle COOH terminus, although there is about a 2-fold difference in actin affinity, as previously reported (41). However, the effect of deleting exon 6 or replacing it with a leucine zipper-encoding sequence depends on the COOH-terminal sequence; neither can bind to actin with an exon 9d-encoded COOH terminus in any condition tested. The results show that the sequence in the middle of TM can have a long range effect on the functional activity of the ends.
Of the 21 residues (residues 191-211) of exon 6b substituted with exon 6a sequence, 8 were identical, including four in a or d interface positions. Interestingly, whereas replacement of exon 6a for 6b in ␣-TM alters actin affinity (these results) (41), in ␤-TM it does not affect actin affinity (49), even though exon 6a and 6b are about 70 and 85% identical, respectively, in rat ␣versus ␤-TM. Substitution of all 25 residues of exon 6 with the GCN4 leucine zipper sequence changed all but 5 residues, 2 of which are in interface positions. The observed loss of actin affinity of the 6zip variants in the absence of Tn was as great as previously reported for modifications at the NH 2 -or COOH terminus of TM (17,18,33,47,50). The results show that a native TM sequence in the exon 6-encoded region is critical for cooperative binding to actin. Without high resolution structural information about the TM structure, and its binding site on actin, further speculation is premature.
The most surprising result is that deletion of 21 residues of exon 6 abolished actin binding of TM with exon 9d but had little effect with a 9␣-encoded COOH terminus. In previous work from this laboratory, a 21-residue deletion within exon 2b of a fusion striated ␣-TM (with exons 6b, 9a and an 80 residue fusion peptide on the NH 2 terminus) resulted in loss of actin affinity (48). Because 21 residues correspond to one quarter of a turn of the coiled coil, we postulated that the failure of the 21-residue deletion to bind actin may be a consequence of altered geometry of the ends of adjacent molecules with respect to each other on the actin filament. However, the more deleterious effect of the 21-residue deletion in exon 6 in combination with exon 9d than exon 9a shown in this study implies that the structures of the ends are significant, but not simply in terms of the coiled-coil geometry. With our present understanding of TM structure we cannot explain how a 21-residue deletion has variants which all have exon 2b and exon 6b. In TM9a and 9d the COOH-terminal 27 amino acids are encoded by exons 9a and 9d, respectively. In TM9a/9d, the first 18 residues are encoded by exon 9a, the last 9 by exon 9d. In TM9d/9a, the first 18 residues are encoded by exon 9d, the last 9 by exon 9a. In TM9a/ and TM9d/ the last 9 amino acids have been deleted. such a severe effect with an end that allows high affinity actin binding, such as NH 2 -terminal fusion peptide in our initial studies or exon 9d here, but not exon 9a.
Our results clearly show that the sequence encoded by exon 6 (residues 189 -214) is not critical for Tn interaction. Troponin in the presence of Ca 2ϩ greatly increased the affinity of all TMs containing exon 9a, independent of the exon 6 sequence, but had only a small effect on those with exon 9d-encoded termini, consistent with previous results (17,18). Removal of Ca 2ϩ caused an increase in actin affinity of all TM variants with Tn (independent of exon 6 or 9 sequence). The differences in actin affinity among the exon 6 variants in the presence of Tn seem to relate more to actin affinity than troponin interaction because the relative differences were maintained in the three binding conditions (no Tn, Tn ϩ Ca 2ϩ , Tn Ϫ Ca 2ϩ ). The effect of the leucine zipper replacement was similar to that reported for leucine zipper replacements at other sites in TM (42). The exception was TM6zip-9d where Tn had no effect in the presence of Ca 2ϩ .
The effect of Tn on TM binding to actin correlates well with the binding of TnT to TM measured using biosensor technology where binding could be measured in all variants containing exon 9a, or the first 18 residues encoded by exon 9a in a full-length TM, independent of the exon 6-encoded sequence.
The region of TM encoded by exon 6 (residues 189 -213) has been implicated in Tn binding based on structural studies (11), binding studies with TM and Tn peptides (16), TnI, TnC, and TnT cross-linking to Cys-190 of TM (51,52), and probes attached to Cys-190 responding to Tn and TnT binding to TM alone and on the actin filament (43,44,53,54). However, the present results indicate that the Tn interaction with TM is indifferent to the exon 6-encoded sequence ruling out this region of TM as a specific Ca 2ϩ -sensitive Tn binding site. The interaction with Tn in this region must be much weaker than at the COOH terminus, and cross-linking and fluorescence studies may simply be detecting the close proximity between the exon 6-encoded region of TM and the Tn components.
Although exon 6 clearly is not required for interaction with Tn (unique to striated muscles), it may be important for interaction with caldesmon which is found in smooth and nonmuscle cells. Some reports have suggested that a region of TM including exon 6 is important for caldesmon binding (55) and that the sequence of exon 6 influences the effect of caldesmon on the affinity of TM for actin (49). Also, further work will be necessary to understand the significance of isoform specific TM function in terms of its cooperative interaction with specific myosin isoforms on the actin filament. This is an essential TM function for which the structural requirements have not been investigated.  Fig. 6 were fit to the equation R t ϭ R 0 e Ϫkd1(tϪt0) ϩ (R 0 Ϫ R 1 )e Ϫkd2(tϪt0) which calculates the dissociation rate of two parallel reactions. The reported parameters were calculated as described under "Materials and Methods."  . Specific activity is expressed as mol of P i /mg of myosin/min. In the absence of TM, the mean specific activity of the actin-Tn-myosin ATPase for all experiments pooled was 0.13 Ϯ 0.02 mol of P i /mg of myosin/min (range ϭ 0.09 -0.16 mol of P i /mg of myosin/min; n ϭ 35). Each curve contains normalized data pooled from three to four experiments. Data was normalized to the mean specific activity of the actin-Tn-myosin ATPase in the absence of TM for each experiment. Only TM6b-9d is shown with Tn and 0.2 mM Ca 2ϩ , since it is representative of all the TM exon 6 variants.