Tropomodulin Contains Two Actin Filament Pointed End-capping Domains*

Tropomodulin 1 (Tmod1) is a ∼40-kDa tropomyosin binding and actin filament pointed end-capping protein that regulates pointed end dynamics and controls thin filament length in striated muscle. In vitro, the capping affinity of Tmod1 for tropomyosin-actin filaments (Kd ∼ 50 pm) is several thousand-fold greater than for capping of pure actin filaments (Kd ∼ 0.1 μm). The tropomyosin-binding region of Tmod1 has been localized to the amino-terminal portion between residues 1 and 130, but the location of the actin-capping domain is not known. We have now identified two distinct actin-capping regions on Tmod1 by testing a series of recombinant Tmod1 fragments for their ability to inhibit actin elongation from gelsolin-actin seeds using pyrene-actin polymerization assays. The carboxyl-terminal portion of Tmod1 (residues 160–359) contains the principal actin-capping activity (Kd ∼ 0.4 μm), requiring residues between 323 and 359 for full activity, whereas the amino-terminal portion of Tmod1 (residues 1–130) contains a second, weaker actin-capping activity (Kd ∼ 1.8 μm). Interestingly, 160–359 but not 1–130 enhances spontaneous actin nucleation, suggesting that the carboxyl-terminal domain may bind to two actin subunits across the actin helix at the pointed end, whereas the amino-terminal domain may bind to only one actin subunit. On the other hand, the actin-capping activity of the amino-terminal but not the carboxyl-terminal portion of Tmod1 is enhanced several thousand-fold in the presence of skeletal muscle tropomyosin. We conclude that the carboxyl-terminal capping domain of Tmod1 contains a TM-independent actin pointed end-capping activity, whereas the amino-terminal domain contains a TM-regulated pointed end actin-capping activity.

respectively), which are about 60 -70% identical to one another and are expressed in a tissue-specific and developmentally regulated fashion (1)(2)(3). Tmod function in vivo is best understood in striated muscle cells, where Tmod1 or Tmod4 is associated with the free, pointed ends of thin filaments in sarcomeres and functions to regulate pointed end dynamics and control filament length (1, 4 -7). Whereas the functions of Tmods in non-muscle cells are less well understood, recent work from our laboratory indicates that the Tmod3 isoform is present in the leading lamellipodia in human microvascular endothelial cells, where it reduces pointed end disassembly and negatively regulates cell migration (8). In the brain, the Tmod2 isoform may play a role in synaptic plasticity, based on characterization of behavioral deficits and enhanced long term potentiation in a Tmod2 knockout mouse (9).
Tmods are unique in several respects as compared with all other capping proteins. First, Tmods bind specifically to actin filament pointed ends but not to actin monomers, sides of filaments, or barbed filament ends (10,11). Second, Tmods cap pointed ends transiently, thus slowing down monomer addition and leading to an increased proportion of filaments with terminal ADP-actin subunits that have a lower affinity for ends (12). The consequence is that at high Tmod concentrations, there is net actin depolymerization and filament shortening until a new steady state is achieved. Third, Tmods are unique among all capping proteins in that they also bind to tropomyosins (TM) (10,13,14) and their capping activity is enhanced several thousand-fold in the presence of TM (1,11,12). 2 However, Tmod capping of TM-actin filament pointed ends does not affect the pointed end critical concentration, because the tightly capped TM-actin ends do not exchange actin subunits and thus are silent. We have proposed that TM enhances the actin-capping activity of Tmod by providing a second binding site for Tmod at the filament pointed end. However, it is also possible that TM binding to Tmod leads to a conformational change in the Tmod that increases its affinity for actin at the pointed ends. The role of TM binding for Tmod function in vivo is not known.
Recent biophysical and biochemical studies indicate that the amino-terminal region of Tmod1 is highly protease-sensitive and unstructured, whereas the carboxyl-terminal portion is compact, folded, and protease-resistant (15)(16)(17)(18). The carboxylterminal portion (residues 160 -344) of chicken Tmod1 (residues 1-359) has been crystallized, and its structure has been solved to a resolution of 1.45 Å (19). This reveals that the carboxyl-terminal domain is composed of a series of five leucine-rich repeats, each consisting of an ␣-helix/␤-sheet pair. The five leucine-rich repeats are followed by a nonhomologous, associated ␣-helix (␣ 6 ) composed of residues 322-344. The structure of the final 15 amino acids (345-359) at the carboxyl terminus of Tmod1 is not known, since they were not present in the recombinant protein used for crystallization by Krieger et al. (19). The TM-binding region of Tmod1 and Tmod4 is located in the amino-terminal portion between residues 1 and 130 (15,18,20,21). It was proposed some time ago that the carboxylterminal half of Tmod1 contained the actin pointed end-capping domain, based on the ability of a monoclonal antibody (mAb9) that bound to this region of Tmod1 to inhibit pointed end-capping activity (5).
In this study, we have identified directly the actin pointed end-capping regions on Tmod1 by designing a series of recombinant fragments based on the domain structure of Tmod1 and testing them for their ability to inhibit actin elongation from the pointed ends of gelsolin-capped actin filaments in pyreneactin polymerization assays. These experiments demonstrate that a TM-independent pointed end-capping activity for pure actin filaments is associated with the extreme carboxyl-terminal end of Tmod1, whereas a TM-regulated actin-capping activity is located in the unstructured TM-binding region in the amino-terminal portion of Tmod1. Unexpectedly, we also report that the full-length Tmod1 protein as well as the carboxylterminal domain enhances spontaneous actin nucleation, whereas the amino-terminal domain does not. These results suggest that 1) Tmod1 has two actin binding domains that interact with two different sites on actin filament pointed ends and 2) TM-binding by Tmod1 most likely enhances Tmod1 capping by enhancing the actin capping affinity of the Tmod1 amino-terminal actin-capping domain rather than by providing a separate TM-binding site for Tmod1 at the filament end. These biochemical results will provide a foundation for future exploration of the role of TM in regulation of actin-capping activities by Tmod1 in vivo.

Construction of cDNAs and Expression of Recombinant Tmod1
Fragments-Full-length chicken Tmod1 (E-Tmod) (accession number L36678) was inserted in frame in the pGEX-KG vector so as to code for a fusion protein with glutathione S-transferase (GST) on the aminoterminal end of Tmod1 (20). cDNAs encoding GST fusion proteins with various Tmod1 fragments were inserted in the EcoRI and/or NcoI sites in the linker region of the pGEX-KG vector using the polymerase chain reaction with appropriate primers. The use of different insertion sites in the linker region for the truncated Tmod1 cDNAs resulted in some amino acid differences in the sequence of the linker remaining on the amino-terminal end of the purified Tmod1 after removal of the GST moiety as well as a few additional vector-derived amino acids at the carboxyl-terminal end of some of the Tmod1 fragments (Table I).
Transformed Escherichia coli (BL21(DE3)) cells were grown to an A 605 of 1.2-1.5 before inducing with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 2 h at 37°C, followed by lysis and purification using a modification of a previously described procedure (20). After affinity purification of the GST fusion protein on a glutathione column, the Tmod1 or fragment was released from the GST moiety by cleavage with thrombin for 30 min at 22°C, using 12.5 units thrombin for each 2 liters of bacterial extract loaded onto 10 -15 ml of beads. Thrombin concentration and cleavage times were adjusted for each fragment to maximize release from the GST moiety while minimizing internal proteolysis by thrombin. After elution from the beads, 50 units of recombinant hirudin (Calbiochem) were added to terminate thrombin cleavage and to prevent Tmod1 proteolysis by residual thrombin during subsequent purification steps. Tmod1 or fragments were dialyzed into 10 mM NaCl, 10 mM Tris-HCl, pH 8.0, 1 mM dithiothreitol (DTT), 2.5 mM EDTA and purified by anion exchange chromatography on a Resource Q column (Amersham Biosciences), eluting with a 10 -300 mM NaCl gradient in the above buffer, followed by chromatography on a Mono Q column (Amersham Biosciences), eluting with the same gradient. Tmod1 fragments 130 -359, 160 -359, and 160 -344 with neutral or basic pI values were dialyzed into 20 mM MES, pH 6.5, buffer followed by purification on a Mono S (Amersham Biosciences) cation exchange column, eluting with a 10 -300 mM NaCl gradient. Fractions containing Tmod1 or frag-ments were identified by SDS-polyacrylamide gel electrophoresis and Coomassie Blue staining, pooled, and dialyzed into 20 mM HEPES, pH 7.3, 80 mM KCl, 1 mM DTT, and 0.02% sodium azide and stored frozen at Ϫ80°C. All of the fragments used in this study were soluble and well behaved, as expected from the previous biophysical studies, and yields ranged from 2 to 10 mg from 2 liters of bacterial extract, depending on the fragment. Before use in experiments, proteins were dialyzed and centrifuged for 30 min at 100,000 ϫ g to remove minor amounts of aggregated material resulting from freezing and thawing, and their concentrations were redetermined. Protein concentrations were determined by light absorption at A 280 based on the extinction coefficients calculated from the amino acid composition of the purified fragments as described by Gill and Von Hippel (22) using the ProtParam tool available on the World Wide Web at expasy.org/tools/protparam.html. The molecular weights, pI values, extinction coefficients, and amino-and carboxyl-terminal sequences for Tmod1 and Tmod1 fragments used in this study are summarized in Table I.
Preparation of Actin, TM, and Gelsolin-Rabbit skeletal muscle actin was prepared from rabbit muscle acetone powder by the method of Spudich and Watt (23) and further purified by gel filtration over a Superose 6 column (Amersham Biosciences) as previously described (24). G-actin was stored by flash freezing in liquid nitrogen and defrosted as described (25). Before use, actin was dialyzed for 3-4 days at 4°C into 4 mM Tris-HCl, pH 8.0, 0.1 mM CaCl 2, 0.2 mM ATP, 1.0 mM DTT, 0.02% sodium azide) followed by centrifugation for 30 min at 450,000 ϫ g in a Beckman TLA 100.3 rotor to remove filaments and small oligomers that had formed during freezing and thawing. G-actin was stored on ice and used for up to 6 days. Pyrene-labeled actin was prepared as previously described (26,27) and stored and treated as for unlabeled actin. Chicken skeletal muscle TM was prepared according to Smillie (28) and stored as the lyophilized powder. Gelsolin was a gift from J. Bryan (Baylor College of Medicine, Dallas, TX) and was prepared as described (29). Protein concentrations were determined for actin and gelsolin by light absorption, using E 290 ϭ 24.9 mM Ϫ1 cm Ϫ1 and E 280 ϭ 150 mM Ϫ1 cm Ϫ1 , respectively, and for TM by Lowry's method, using bovine serum albumin as a standard.
Actin Polymerization Measurements-Measurements of elongation rates at the pointed end were carried out using 8 -12% pyrenyl-actin and gelsolin-capped actin filaments as nuclei for polymerization (11,12). In initial experiments, fluorescence was followed on a PerkinElmer 650S fluorometer and detected using a chart recorder (excitation ϭ 366.5 nm, emission ϭ 407 nm). Photobleaching was minimized by setting the excitation slit width to 2 nm and using a neutral density filter in the excitation light path (30). In subsequent (most) experiments, fluorescence was followed using a Spex Fluoromax-3 fluorometer (excitation slit 0.5 nm, emission slit 5 nm) (Jobin Yvon Horiba, Edison, NJ), and data were collected using DataMax software.
For assays in the absence of TM, gelsolin-actin seeds were prepared by copolymerizing 10 M G-actin with 1 M gelsolin in the presence of calcium (actin/gelsolin, 10:1) and used for up to 4 days. For each time course, Ca 2ϩ -actin was first converted to Mg 2ϩ -actin by incubation for 10 min at 20°C in a buffer containing 10 mM imidazole, pH 7.0, 0.05 mM MgCl 2 and EGTA present at 2-3-fold in excess over the CaCl 2 contributed by the actin buffer (25). Tmod1 or fragments and gelsolin/actin seeds were added in succession, and then polymerization was initiated by the addition of one-sixteenth volume of 16ϫ polymerizing salts (1.65 M KCl, 33 mM MgCl 2 , 8.0 mM ATP, 3.3 mM CaCl 2 ) and followed at 20°C in the fluorometer. The final concentrations of components in the assays were as follows: 2.5 M G-actin, 10 -20 nM gelsolin/actin seeds, and Tmod1 or fragments as indicated in the figure legends. The buffer was 10 mM imidazole, pH 7.0, 0.1 M KCl, 2.0 mM MgCl 2 , 0.5 mM ATP, 0.2 mM CaCl 2 , 1 mM DTT. Capping activities for Tmod1 and fragments were obtained from the initial elongation rates, measured directly from the slopes of the polymerization traces over the first 30 s to 1 min of polymerization. Rates in the presence of Tmod1 or fragments were divided by the rate for actin in the absence of Tmod1, giving a rate/ control rate. The K d values for full-length Tmod1 and each fragment were then calculated from the x intercept of a double reciprocal plot of 1/(1 Ϫ (rate/control rate)) versus 1/Tmod1 concentration. Although the absolute K d values varied somewhat between different experiments, the relative differences in capping activities of the fragments as compared with full-length Tmod1 were very similar in each experiment. The variability for determination of the K d values in these assays is most likely due to a variable efficiency of barbed end capping by gelsolin from one experiment to the next. Any free barbed ends will lead to a background of fluorescence increase due to barbed end elongation, which will lead to a falsely low readout for the pointed end-capping activity by Tmod1 and thus an apparently higher K d value.
Experiments measuring the effects of Tmod1 fragments on elongation rates in the presence of TM were generously carried out by Dr. Annemarie Weber at the University of Pennsylvania (Philadelphia, PA). Actin/gelsolin seeds (150:1) were copolymerized with TM, and additional TM was also added along with the Tmod1 or fragments in the polymerization mixtures to ensure that newly elongating filaments were also coated with TM (12). Fluorescence was followed using a Photon Technology International fluorometer (Princeton, NJ). Elongation rates in the presence of TM and Tmod1 were measured from the flat portion of the curve after the inhibition of the polymerization rate by Tmod1 was maximal (e.g. see Fig. 4 in Ref. 12), and K d values were calculated as above.
The effects of Tmod1 or fragments on spontaneous actin polymerization were examined by polymerizing 5 M G-actin (8 -10% pyrenylactin) in the presence or absence of Tmod1 or fragments. G-actin was first converted from Ca 2ϩ -actin to Mg 2ϩ -actin as described above, and then the indicated concentrations of Tmod1 or fragments were added (see Fig. 8 legend) followed immediately by one-tenth volume of 10ϫ polymerizing buffer (1.0 M KCl, 20 mM MgCl 2 , 5.0 mM ATP, 50 mM EGTA) and incubation at 20°C in the spectrofluorometer cuvette. The time course of fluorescence increase was monitored in a Fluoromax-3 fluorometer with the excitation slit at 0.25 nm and the emission slit at 10 nm to minimize photobleaching. The concentration of barbed ends was calculated from the elongation rate (measured by the rate of polymerization, where 50% of monomers were polymerized) using the equation, Electrophoresis Procedures-Electrophoresis of proteins was on 12% SDS-polyacrylamide gels using a running gel pH of 8.6 (10). Proteins were transferred to nitrocellulose, and blots were preheated at 65°C in PBS followed by staining with Ponceau S to detect proteins (32) before labeling with a monoclonal antibody to tropomodulin (mAb9) (5), followed by rabbit anti-mouse IgG (Sigma) and detection by standard chemiluminescence procedures.
CD Experiments-CD measurements of fragments 160 -344 and 160 -359 were performed on an Aviv model 62 D spectrophotometer in 100 mM NaCl, 10 mM sodium phosphate buffer, pH 6.5, at 10°C. The secondary structure of these fragments were estimated using the programs CDNN (33) and Selcon 1 (34). Two different sets of standards were used with the Selcon program. The first contained 17 proteins where secondary structures were calculated as described by Kabsch and Sander (35), and the second had 33 reference proteins where the secondary structures were calculated as described by Toumadje et al. (36) as reviewed by Greenfield (37).
Determination of the Free Energy of Folding of Peptide Fragments-The free energy of folding of fragments 160 -344 and 160 -359 were determined as described by Santoro and Bolen (38), assuming that the free energy of folding is linearly dependent on the concentration of a chemical denaturant. In these experiments, a stock solution of 6 M guanidine HCl containing 10 M peptide, 100 M NaCl, and 10 mM sodium phosphate, pH 6.5, was used to titrate a 10 M solution of each peptide, and the ellipticity ( obs ) was recorded at each concentration of guanidine [Gn i ]. The data were fit to the equation, where ⌬G o is the free energy of folding in the absence of denaturant, R is the gas constant, T is the temperature (Kelvin), obs is the observed ellipticity at any concentration of guanidine HCl, [Gn i ], M is slope of the linear change in free energy as a function of [Gn i ], and f and u are the ellipticity of the fully folded and fully unfolded peptides at 10°C. To solve the equation's initial values of ⌬G o , M, f , and u were estimated, and the best fits to the data were evaluated using the Levenberg-Marquardt (39) algorithm implemented in SigmaPlot 8.0 (SpSS Inc., Chicago, IL).

Recombinant Tmod1
Fragments-We chose to prepare a series of fragments from the chicken Tmod1 isoform (E-Tmod), since most of the previous structural and functional studies have been performed with this protein. Fig. 1 is a schematic depicting the locations of the fragments that we tested, with respect to the known structural features of Tmod1. In general, most fragments corresponded to the domain organization of Tmod1. Thus, the fragments 35-359, 95-359, and 130 -359 all contained the compact, folded carboxyl-terminal domain but were missing increasingly greater amounts of the amino-terminal unstructured, TM-binding portion of Tmod1 (15,18). The 160 -344 fragment was the compact carboxyl-terminal domain that was crystallized by Krieger and colleagues (19), whereas 160 -359 contained an additional 15 amino acids of unknown structure derived from the Tmod1 carboxyl terminus. The fragments 1-322, 1-292, 1-238, 6 -187, 1-156, and 1-130 all contained the unstructured amino-terminal portion of Tmod1 but were missing varying portions of the compact carboxyl-terminal domain; note that the carboxyl termini of these fragments were located at the boundaries of adjacent LRRs ( Fig. 1; see Ref. 19).
These fragments were expressed in E. coli as GST fusion proteins, and 1-mg quantities were purified to homogeneity after removing the GST moiety by thrombin cleavage (20) (Table I and Fig. 2). All of the recombinant proteins contained a 13-16-amino acid extension at their amino terminus that was contributed by the linker region in the vector and that remained after thrombin cleavage from the GST moiety (Table I). We showed previously that this amino-terminal extension had no effect on the actin-capping activity of Tmod1 (11). Several fragments (1-130, 1-156, 6 -187, and 35-128) also contained some additional amino acids at their carboxyl terminus, which also appeared to have no effect on actin capping based on comparison of their activities with the other fragments (see below). Production and purification of the correct fragments was confirmed by amino-terminal sequencing and mass spectrometry using matrix-assisted laser desorption/ionization time-of-flight, which showed that the actual molecular mass corresponded to the molecular mass predicted from the amino acid sequence of fragment plus the extra amino-and carboxylterminal sequences (data not shown).
The Carboxyl-terminal Domain of Tmod1 Contains a Pointed End-capping Activity for Pure Actin Filaments-We showed previously that a monoclonal antibody (mAb9) that reduced the ability of Tmod1 to cap actin filament pointed ends bound to the carboxyl-terminal half of chicken Tmod1 (residues 190 -359) (5). We mapped the mAb9 epitope further by immunoblotting a series of Tmod1 fragments that had been successively truncated from the carboxyl-terminal end (Fig. 3). These experiments showed that mAb9 recognized full-length Tmod1 (residues 1-359) as well as Tmod1 fragments 160 -359 and 160 -344 but not 1-322, 1-292, 1-238, or 1-156, demonstrating that residues between 323 and 344 contribute to the mAb9 epitope on Tmod1. The amino-terminal residues 1-130 of Tmod1 were determined to be unstructured by CD spectroscopy (15,16,18), and the carboxyl-terminal domain of Tmod1 was determined to consist of five LRRs by x-ray crystallography (19). The location of the TMbinding domain is from Refs. 15, 18, and 20. To investigate the role of this carboxyl-terminal region in actin capping by Tmod1, we first compared the actin-capping activities of Tmod1s that had been successively truncated from the carboxyl-terminal end ( Fig. 4 and Table II). Fig. 4 depicts a representative experiment with the 1-322, 1-238, and 1-130 fragments and shows that about 16 times as much 1-322 protein was required to achieve half-maximal inhibition of the initial rate of actin polymerization as compared with fulllength Tmod1 (Fig. 4A). However, further truncation from the carboxyl-terminal end to remove several or all of the LRRs had no further effect, since 1-238 and 1-130 fragments both appeared to have about the same activity as did the 1-322 fragment. Determination of the K d values for these fragments from double reciprocal plots revealed that the K d values for 1-322, 1-238, and 1-130 were all about 1.80 M, as compared with a K d of 0.11 M for full-length Tmod1 in this experiment ( Fig. 4B and Table II). In other experiments, we also determined the K d values of 1-292 and 1-156 to be about 2 M (Table II and data   not shown). We conclude from these experiments that residues between 323 and 359 at the carboxyl-terminal end of Tmod1 are necessary for full actin pointed end-capping activity. In contrast, the five LRRs, which are located between residues 160 and 322, do not appear to be directly involved in capping actin pointed ends, since the activity of 1-130 (or 1-156) is similar to that of 1-322 (Fig. 4, Table II).
To determine directly whether the carboxyl-terminal domain of Tmod1 can cap actin pointed ends in the absence of the amino-terminal domain, we measured the actin-capping activities of Tmod1 fragments 160 -359 and 160 -344. The results of these experiments showed that 160 -359 was active at capping actin pointed ends in the absence of the amino-terminal portion of Tmod1, albeit about 4-fold more weakly than full-length Tmod1 (K d ϳ0.4 and ϳ0.1 M, respectively) ( Fig. 5 and Table  II). However, Tmod1 fragment 160 -344 was considerably less active than 160 -359, with a K d value of about 4 M (Fig. 5 and Table II). This weak activity was also observed for a 160 -344 fragment prepared by limited proteolysis of recombinant Tmod1 1-344 that was generously provided by A. Kostyukova and Y. Maeda (15) (data not shown).
We used CD spectroscopy to investigate whether the 10-fold decrease in capping activity upon removal of the carboxylterminal 15 amino acids could be due to altered folding or decreased stability of 160 -344 as compared with 160 -359 (Fig.  6). Comparison of the CD spectra for these two fragments revealed that they were very similar to one another (Fig. 6A) and to the spectra reported previously for the 160 -344 proteolytic fragment studied by Kostyukova et al. (15). Estimation of the secondary structure using CDNN and Selcon (see "Experimental Procedures") showed that 160 -359 had about 41 Ϯ 1% ␣-helix and 15 Ϯ 2% ␤-sheet, compared with about 37 Ϯ 1% ␣-helix and 14 Ϯ 2% ␤-sheet for 160 -344, with the remainder in ␤-turns and regions without defined secondary structure. Note that the x-ray structure of fragment 160 -344 (19) is reported to be 42% ␣-helical and 8% ␤-structure. From these percentages, 160 -359 can be calculated to have about 84 residues with an ␣-helical conformation, whereas 160 -344 can be calculated to have about 70 residues with an ␣-helical conformation. This suggests that the 15 residues that are missing from the carboxyl terminus of 160 -344 may be in an ␣-helical conformation, with the caveat that the relatively small differences in helical content are within the limits of accuracy of CD in estimating secondary structure (37).
Comparison of the stability of 160 -359 and 160 -344 by unfolding in increasing concentrations of guanidine-HCl revealed very similar denaturation curves (Fig. 6B). The free energies of unfolding were calculated to be Ϫ4.15 and Ϫ3.83 kcal/mol for 160 -359 and 160 -344, respectively, indicating that 160 -344 is only about 8% less stable than 160 -359. These data indicate that the decrease in capping activity of the carboxyl-terminal domain upon deletion of residues 345-359 is not due to significant effects on secondary structure of residues 160 -344. Thus, we conclude that the 15 amino acid residues at the carboxyl terminus of Tmod1 most likely participate in binding directly to actin.
These experiments demonstrate first that the carboxyl-terminal domain of Tmod1 directly binds to and caps actin filament pointed ends and, second, that residues between 345 and 359 at the extreme carboxyl-terminal end are required for full activity of the carboxyl-terminal domain. Third, residues in the ␣ 6 -helix that spans residues 323-344 (19) also appear to participate in actin capping by the carboxyl-terminal domain of Tmod1, based on the residual capping activity of 160 -344 (Fig.  5), together with the truncation experiments in the context of full-length protein (Fig. 4), and the mapping of the mAb9 epitope (Fig. 3).
The Amino-terminal Domain of Tmod1 Also Contains an Actin Pointed End-capping Activity-The results of the carboxyl-terminal deletion analyses described above showed unexpectedly that the unstructured, amino-terminal TM-binding domain of Tmod1 (residues 1-130) had pointed capping activity for pure actin filaments (K d ϳ 1.8 M) (Fig. 4), albeit less than the carboxyl-terminal domain (residues 160 -359; K d ϭ 0.4 M) (Fig. 5) (Table II). To further investigate the regions in Tmod1 required for this amino-terminal actin-capping activity, we  compared the ability of a series of amino-terminal deletion fragments to inhibit elongation from gelsolin-capped actin filaments. Fig. 7 depicts a representative experiment and showed that, overall, the fragments fell into two categories. Fragment 35-359 was as effective at capping actin pointed ends as was full-length Tmod1, whereas about 4-fold higher concentrations of the other amino-terminal deletion fragments (95-359, 130 -359, and 160 -359) were required to achieve 50% inhibition of actin pointed end elongation as compared with full-length Tmod1 (Fig. 7A). Determination of the K d values of these fragments by double reciprocal plots revealed that the K d values of 35-359 and full-length Tmod1 were both ϳ0.09 M in this experiment, whereas the K d values for all of the other aminoterminal deletion fragments were ϳ0.4 M (Fig. 7B, Table II). Additionally, we also found that the actin-capping activity of Tmod1 fragment 35-128 was similar to that of 1-130 (Table II). Therefore, residues between 35 and 95 appear to be most important for the pointed end actin-capping activity of the aminoterminal 1-130 region of Tmod1.
The Carboxyl-but Not the Amino-terminal Actin-capping Domain of Tmod1 Stimulates Actin Filament Nucleation-The presence of two distinct actin pointed end-capping domains in Tmod1 raised the possibility that Tmod1 might be able to bind to both of the actin subunits that are exposed at the pointed end of the filament. In this case, one might expect that Tmod1 would also bind to the "pointed ends" of spontaneously formed actin nuclei and stabilize them, analogous to a nucleating mechanism proposed for the barbed end actin-capping protein, CapZ (40). The ability of Tmod1 to nucleate actin filament assembly has not been investigated previously (11,12). We tested the effect of increasing concentrations of Tmod1 and the Tmod1 fragments 1-130 and 160 -359 on spontaneous actin polymerization in a standard actin nucleation assay (Fig. 8). These experiments showed that Tmod1 was a very weak nucleator of actin polymerization, in that micromolar amounts were required to observe an enhancement of the polymerization rate, and a lag period was still observed even at the highest Tmod1 concentrations tested (Fig. 8A). A plot of the numbers of barbed ends as a function of the Tmod1 concentration showed that the maximum concentration of barbed ends generated was about 0.4 nM at the 5 M concentration of G-actin used in these experiments (Fig. 8D). The presence of a lag time even at high Tmod1 concentrations (Fig. 8A) and the apparent saturation of the numbers of barbed ends generated (Fig. 8D) is consistent with Tmod1 stimulating actin nucleation by binding to and stabilizing spontaneously formed actin nuclei during polymerization.
Surprisingly, contrary to our initial expectations, the nucleating activity of Tmod1 did not appear to be due to the com- bined binding activities of its two actin-capping domains. The amino-terminal domain 1-130 was completely inactive at stimulating actin nucleation, even at concentrations as high as 8 M (Fig. 8, C and D) or 20 M (data not shown). In contrast, the carboxyl-terminal domain 160 -359 was active at stimulating actin nucleation by generating new barbed ends, similar to full-length Tmod1 (Fig. 8, B and D). Interestingly, the plot of barbed end concentration versus the concentration of 160 -359 did not appear to reach saturation, unlike the barbed end plot for Tmod1 (Fig. 8D). This might mean that 160 -359 (but not full-length Tmod1) can recruit additional monomers to create new nuclei, rather than simply stabilizing spontaneously generated actin nuclei in these assays. Additional experiments to test nucleating activity and the numbers of barbed ends generated as a function of the actin monomer concentration will be required to investigate this possibility.
We also tested the 160 -344 fragment and found that 8 M 160 -344 was required to achieve a similar extent of actin nucleation as 1 M 160 -359 (data not shown). The weak nucleating activity of 160 -344 parallels its weak pointed endcapping activity as compared with 160 -359 ( Fig. 5 and Table  II). Therefore, the carboxyl-terminal 15 amino acids of Tmod1 appear to be required for the actin nucleating activity of the carboxyl-terminal domain 160 -359. In conclusion, the differences in nucleating ability of the amino and carboxyl-terminal domains are consistent with the possibility that the carboxylterminal capping domain may bind "across the actin helix" to two actin subunits at the pointed filament end, whereas the amino-terminal actin-capping domain may interact with only one actin subunit at the filament pointed end.
TM Regulates the Actin-capping Activity of the Amino-but Not the Carboxyl-terminal Domain of Tmod1-The amino-terminal actin-capping activity of Tmod1 is coextensive with the region on Tmod1 that has been shown to bind TM (15,18,20). Therefore, we examined whether TM could enhance the ability of the amino-or carboxyl-terminal actin-capping domains to inhibit elongation from the pointed ends of gelsolin-capped filaments. As we observed previously, the actin pointed endcapping activity of Tmod1 for TM-actin filaments (K d ϳ 0.06 nM) was more than 1000-fold greater than the capping activity of Tmod1 for pure actin filaments (K d ϳ 120 nM) (Table III). TM also dramatically enhanced the pointed end-capping activity of the 1-238 or 6 -187 fragments that contain the entire aminoterminal TM-binding domain but are missing some or all of the  (Table III). Tmod1 fragment 6 -187, which was missing all of the LRRs in the carboxyl-terminal domain, also had a similar high affinity for capping of TM-actin filaments (K d ϳ 0.71 nM). However, both 1-238 and 6 -187 were less effective at capping of TM-actin pointed ends than was full-length Tmod1 (Table III). This suggests that the carboxyl-terminal actincapping region does contribute to capping of TM-actin filaments in the context of full-length Tmod1.
Not surprisingly, deletion of residues 1-130 from the amino terminus to remove the entire TM-binding domain from Tmod1 completely eliminated the ability of TM to enhance the pointed end-capping activity of Tmod1. Thus, 130 -359 had a K d of ϳ 380 nM for TM-actin, similar to a K d of ϳ280 nM for pure actin filaments (Tables II and III). However, TM was still able to enhance the actin-capping activity of fragment 95-359 by about 160-fold. 95-359 had a K d of ϳ 1.6 nM for capping of TM-actin filaments as compared with a K d of 260 nM for capping of pure actin filaments (Table III). TM also enhanced the actin-capping activity of Tmod1 fragment 35-359 (data not shown). These observations are surprising, since neither 35-359 nor 95-359 fragments were able to bind to skeletal muscle TM in the absence of actin (20) (see "Discussion").
In conclusion, these experiments show that the carboxylterminal domain on its own is completely ineffective at capping TM-actin filaments with a high affinity, due to the absence of the TM-binding activity contributed by the amino-terminal domain of Tmod1. In contrast, the capping activity of the amino-terminal domain of Tmod1 is dramatically up-regulated in the presence of TM to become a high affinity pointed end cap. DISCUSSION We describe here two structurally and functionally distinct actin-pointed end-capping activities in Tmod1, one that is present in the unstructured amino-terminal domain utilizing residues between 35 and 130 and another that is associated with the compact carboxyl-terminal domain utilizing residues between 323 and 359 at the carboxyl terminus of the protein (Fig.  9). The actin-capping activity of the carboxyl-terminal domain is somewhat greater than that of the amino-terminal domain, based on K d values of ϳ0.4 M for the carboxyl-terminal domain (residues 160 -359) and ϳ2 M for the amino-terminal domain (residues 1-130). Strikingly, the actin pointed endcapping activity of the amino-terminal domain is enhanced over 1000-fold for TM-actin filaments, similar to full-length Tmod1, whereas the activity of the carboxyl-terminal domain is unaffected by the presence of TM. Thus, the carboxyl-terminal capping domain represents a TM-independent actin pointed end-capping activity, whereas the amino-terminal domain is a TM-regulated pointed end actin-capping activity (Fig. 9).
Interaction of the Carboxyl-terminal Domain with Actin Pointed Ends-Results from deletion analysis and epitope mapping with a function blocking monoclonal antibody (mAb9) indicate that residues between 323 and 359 at the carboxyl terminus are important for TM-independent pointed end capping by the carboxyl-terminal domain. This carboxyl-terminal actin-capping region of Tmod1 includes the extended ␣-helix, 322-344, that is associated with the LRR domain of Tmod1 (19), as well as an additional 15 amino acids (residues 345-359) of unknown structure at the carboxyl terminus (Fig. 9). Interestingly, mobile carboxyl-terminal extensions ("tentacles") on the ␣ and ␤ subunits of the barbed end capping protein CapZ (capping protein) are responsible for binding to and capping the actin barbed filament end (41,42).
Comparison of the carboxyl-terminal sequence of chicken Tmod1 with the carboxyl-terminal sequences of Tmod1 to -4 reveals that residues between 323 and 344 are highly conserved among all the ϳ40-kDa Tmods but that Tmod2, -3, and -4 are missing the final 15-amino acid "tail" at the carboxyl terminus that is important for the full actin-capping and nucleating activity of chicken Tmod1 (Fig. 10). However, all the Tmods that have been tested in actin polymerization assays for pointed end capping (chicken Tmod4, human Tmod1, and human Tmod3) are as effective as chicken Tmod1 at inhibiting elongation from the pointed ends of gelsolin-capped actin filaments (1,8) (Fig. 10). A common feature of the carboxyl-terminal ends of these four Tmod isoforms is the presence of at least 8 basic residues that are grouped into two or three clusters. These basic residues may be involved in actin binding, similar to basic residues in the actin-binding regions of other actin binding proteins, such as in the carboxyl-terminal tentacle of the ␤ subunit of capping protein (41)(42)(43). If so, one might expect that the clusters of basic residues would be oriented similarly in the three-dimensional structures of the carboxyl termini of the Tmods. The importance of these basic (or other) residues in actin pointed end capping will require site-directed mutagenesis and actin-capping assays along with structural studies of the carboxyl-terminal 15-amino acid tail of Tmod1.
What can our data tell us about the site(s) on actin with which the carboxyl-terminal actin-capping domain interacts? The ability of the carboxyl-terminal domain to enhance spontaneous nucleation of actin suggests that this domain may interact with two actin subunits at the "pointed end" of spontaneously forming actin dimers so as to stabilize them. In this case, the carboxyl-terminal domain may bind to the pointed end in the position of the next incoming actin monomer so that the carboxyl-terminal domain can bind to the interface between the terminal and penultimate actin subunit at the pointed end. This would be similar to the proposed mechanism for capping by the carboxyl-terminal tentacles of CapZ (41,42). Binding of Tmod1 in the position of the incoming actin monomer was previously proposed based on the kinetics of Tmod1 inhibition of actin pointed end elongation (11,12). However, the actin nucleating activity of Tmod1 is unlikely to be biologically significant in that relatively high micromolar concentrations are required to observe significant increases in the numbers of additional barbed ends.  9. Schematic of locations of Tmod1 amino acid residues involved in TM-regulated and TM-independent actin pointed end-capping activities with respect to Tmod1 structural domains. TM-regulated actin capping involves residues between 35 and 130 in the unstructured region at the amino-terminal end, whereas TM-independent actin pointed end capping involves the ␣ 6 -helix 322-344 (19), together with the carboxyl-terminal tail, residues 345-359. Krieger et al. (19) have proposed an attractive model for docking Tmod1 onto the pointed end of an actin filament based on their crystal structure of the Tmod1 fragment 160 -344 and a structural model of the actin filament (44). In this model, the positively charged ␤-surface of the LRR domain 160 -322 is proposed to interact with the negatively charged actin helix Ala 181 -Glu 195 , placing the tip of the ␣ 6 -helix at the entrance to the major groove of the actin filament. The cluster of basic residues (RKRR; residues 340 -343) at the tip of the ␣ 6 -helix is proposed to interact with the highly negatively charged groove on the actin filament, and a hydrophobic region further up this helix is proposed to interact with the hydrophobic "plug" on actin. Our results are consistent in part with this model, in that our data show that the ␣ 6 -helix is likely to be involved in actin capping and that the carboxyl-terminal portion of Tmod1 binds to the pointed end in the position of the incoming actin monomer. The docking model also agrees with our suggestion that the basic residues in the ␣ 6 -helix probably play a role in binding to actin. However, our results comparing 160 -359 with 160 -344 show additionally that residues 345-359 distal to this ␣ 6 -helix are critical for actin capping. These residues were not in the fragment used for crystallization and thus were not included in the docking model.
Our results also disagree in part with the proposed importance of the LRRs in the interaction of Tmod1 with actin, in that the affinity we observed for 1-322 in capping actin was the same as that for 1-156 or 1-130, suggesting that residues 160 -322 containing the LRRs are unlikely to play a significant role in binding directly to actin. However, it is possible that the LRRs may play a role in the folding of the ␣ 6 -helix and carboxyl-terminal tail regions of Tmod1 so as to influence the structural presentation of the actin binding residues. On the other hand, we cannot rule out the possibility that removal of the ␣ 6 -helix (residues 323-344) may destabilize the folding of the LRRs (residues 160 -322) to such an extent that this region no longer has any actin-capping activity. We have not investigated the structure or capping activity of a 160 -322 fragment. Alternatively, rather than binding to actin, the LRRs may mediate interactions of Tmod1 with other ligands, such as nebulin (45) or filensin (46). Indeed, 160 -344 was shown recently to bind directly to the M123 peptide Tmod1 binding peptide from the NH 2 -terminal end of nebulin (19). Future progress in defining the molecular interaction of the carboxyl-terminal actin-capping domain of Tmod1 with actin pointed ends will require mutagenesis in combination with actin-capping assays as well as direct structural information on a Tmod1-actin complex.
Interaction of the Amino-terminal Domain with Actin and Basis for Regulation by TM-We proposed previously that the higher affinity of Tmod1 for TM-actin pointed ends as compared with pure actin pointed ends was due to TM providing a second binding site for Tmod1 at the filament pointed end (12).
However, the amino-terminal domain 1-130 can bind TM and cap actin pointed ends (15,18,20), and fragments 6 -187 and 1-238 that contain this amino-terminal domain but not the carboxyl-terminal domain exhibit high affinity capping of TMactin filaments (Table III). Interestingly, CD spectroscopy indicates that 1-130 is unstructured on its own in solution and undergoes a transition to a more ␣-helical conformation upon binding to an amino-terminal TM peptide that contains the Tmod1 binding site (18). This raises the intriguing possibility that TM may enhance the actin-capping activity of Tmod1 by inducing a conformational change in the amino-terminal actincapping domain that leads it to bind to actin with a higher affinity. In the context of full-length protein, binding of the carboxyl-terminal domain to actin also contributes to high affinity capping in the presence of TM, since the affinity of 1-238 or 6 -187 for TM-actin is significantly less than that of fulllength Tmod1 for TM-actin filaments.
Interestingly, the actin and TM binding sites in the aminoterminal domain do not appear to be coextensive. For example, deletion of residues 1-34 had no effect on actin pointed end capping ( Fig. 7 and Table II) but abolished binding of Tmod1 to skeletal muscle TM as assayed by dot blots and competition binding assays (20) or by CD spectroscopy experiments. 3 95-359 also does not bind to skeletal muscle TM (20) but caps TM-actin filaments about 160-fold better than it does pure actin filaments (Table III). One possible explanation is that there is a secondary binding site for skeletal muscle TM between Tmod1 residues 95 and 130 that is only revealed in the presence of actin. The idea that Tmod1 has more than one TM-binding site was proposed earlier, based on observations that erythrocyte TM could bind to Tmod1 fragments 35-359 and 95-359 (20). Alternatively, it is conceivable that skeletal muscle TM may enhance the pointed end-capping activity of the Tmod1 amino-terminal domain via a conformational change in actin induced by TM binding to actin.
What can our data tell us about the site on actin at the pointed end to which the TM-regulated, amino-terminal actincapping domain of Tmod1 might bind? First, the inability of 1-130 to promote spontaneous actin nucleation suggests that 1-130 may interact with only one actin subunit at the pointed end. Second, the dramatic enhancement of actin capping in the presence of TM suggests that 1-130 might interact with an actin pointed end subunit near the TM binding site on actin. If the 1-130 were to bind near the TM-binding site on the penultimate actin subunit, this could position the 1-130 domain in a position where it might partially block the addition of the next actin monomer to the pointed end, thus explaining its weak actin-capping activity. Residues are numbered with respect to chicken Tmod1, and basic residues are highlighted in boldface type. Residues 322-344 form an extended ␣-helix in the crystal structure of the Tmod1 carboxyl-terminal fragment 160 -344 (19), whereas the structures of residues 345-359 of Tmod1 are not known. Tmod1, -2, -3, and -4 proteins have different length carboxyl-terminal "tails," but all Tmod isoforms contain 8 basic residues in the region spanning residue 323 to their carboxyl terminus, and all have similar K d values for capping actin filament pointed ends (this study and Refs. 1, 12, and 46).

Relationship of the Amino and Carboxyl-terminal Actin
Pointed End-capping Activities in Tmod1-In an attempt to estimate the relative contributions of the amino-and carboxylterminal capping activities to the actin-capping activity of fulllength Tmod1, we calculated the relative binding energies for the Tmod1 fragments, 1-130 and 160 -359, as compared with full-length Tmod1 (⌬G Fragment/⌬G Tmod1). 4 Based on this analysis, 1-130 contains ϳ85% of the total binding energy, whereas 160 -359 contains ϳ90% of the total binding energy. Thus, by this simple analysis, the binding energy of the aminoterminal 1-130 actin-capping domain plus that of the carboxylterminal 160 -359 actin-capping domain add up to ϳ170% of the binding energy measured for full-length Tmod1. One possibility to explain these observations is that the sites on actin to which the amino-and carboxyl-terminal domains bind at the pointed filament end are partially overlapping so that in the full-length protein, neither domain can bind with high affinity. For example, binding of the amino-terminal domain to the penultimate actin subunit so that it partially blocks the incoming actin monomer could sterically interfere with the binding of the carboxyl-terminal domain in the position of the incoming actin monomer at the pointed end.
A second related possibility is that the two actin-capping domains in the full-length protein are oriented with a less than optimal configuration with respect to their complementary sites on the actin pointed end. This could lead to a strain in the Tmod1 molecule upon its binding to actin pointed ends, thus reducing its apparent affinity. In this scenario, the strain would be relieved upon physical separation of the two portions of Tmod1 so that each binding domain could then interact optimally with its actin binding site, leading to the higher observed affinities of each separate domain for actin pointed ends. A reduced affinity of a protein for the substrate is often observed in proteins with two binding domains, as, for example, in the interaction of the double zinc finger protein, Xenopus transcription factor IIIA, with two adjacent sequence binding motifs in its DNA substrate (47). A third possibility is that the amino-terminal domain of Tmod1 could interact physically with the carboxyl-terminal domain and directly inhibit its actin binding activity in cis. This type of interdomain inhibition of actin binding activity has been described for proteins of the ezrin-radixin-moesin family (48) and would suggest intriguing regulatory possibilities.