Differential Actin-regulatory Activities of Tropomodulin1 and Tropomodulin3 with Diverse Tropomyosin and Actin Isoforms*

Background: Tropomodulins (Tmods) cap pointed ends of actin filaments in a tropomyosin (TM)-dependent manner. Results: Tmod1 and Tmod3 similarly cap actin filaments with diverse TM and actin isoforms, but only Tmod3 sequesters β- and γcyto-actin monomers. Conclusion: Isoform-specific actin monomer sequestration by Tmod3 may provide a mechanism for actin remodeling in TM-deficient regions of cells. Significance: Defining the actin-regulatory activities of Tmods illuminates cytoskeletal dynamics. Tropomodulins (Tmods) are F-actin pointed end capping proteins that interact with tropomyosins (TMs) and cap TM-coated filaments with higher affinity than TM-free filaments. Here, we tested whether differences in recognition of TM or actin isoforms by Tmod1 and Tmod3 contribute to the distinct cellular functions of these Tmods. We found that Tmod3 bound ∼5-fold more weakly than Tmod1 to α/βTM, TM5b, and TM5NM1. However, surprisingly, Tmod3 was as effective as Tmod1 at capping pointed ends of skeletal muscle α-actin (αsk-actin) filaments coated with α/βTM, TM5b, or TM5NM1. Tmod3 only capped TM-coated αsk-actin filaments more weakly than Tmod1 in the presence of recombinant αTM2, which is unacetylated at its NH2 terminus, binds F-actin weakly, and has a disabled Tmod-binding site. Moreover, both Tmod1 and Tmod3 were similarly effective at capping pointed ends of platelet β/cytoplasmic γ (γcyto)-actin filaments coated with TM5NM1. In the absence of TMs, both Tmod1 and Tmod3 had similarly weak abilities to nucleate β/γcyto-actin filament assembly, but only Tmod3 could sequester cytoplasmic β- and γcyto-actin (but not αsk-actin) monomers and prevent polymerization under physiological conditions. Thus, differences in TM binding by Tmod1 and Tmod3 do not appear to regulate the abilities of these Tmods to cap TM-αsk-actin or TM-β/γcyto-actin pointed ends and, thus, are unlikely to determine selective co-assembly of Tmod, TM, and actin isoforms in different cell types and cytoskeletal structures. The ability of Tmod3 to sequester β- and γcyto-actin (but not αsk-actin) monomers in the absence of TMs suggests a novel function for Tmod3 in regulating actin remodeling or turnover in cells.


Tropomodulins (Tmods) are F-actin pointed end capping proteins that interact with tropomyosins (TMs) and cap TMcoated filaments with higher affinity than TM-free filaments.
Here, we tested whether differences in recognition of TM or actin isoforms by Tmod1 and Tmod3 contribute to the distinct cellular functions of these Tmods. We found that Tmod3 bound ϳ5-fold more weakly than Tmod1 to ␣/␤TM, TM5b, and TM5NM1. However, surprisingly, Tmod3 was as effective as Tmod1 at capping pointed ends of skeletal muscle ␣-actin (␣ sk -actin) filaments coated with ␣/␤TM, TM5b, or TM5NM1. Tmod3 only capped TM-coated ␣ sk -actin filaments more weakly than Tmod1 in the presence of recombinant ␣TM2, which is unacetylated at its NH 2 terminus, binds F-actin weakly, and has a disabled Tmod-binding site. Moreover, both Tmod1 and Tmod3 were similarly effective at capping pointed ends of platelet ␤/cytoplasmic ␥ (␥ cyto )-actin filaments coated with TM5NM1. In the absence of TMs, both Tmod1 and Tmod3 had similarly weak abilities to nucleate ␤/␥ cyto -actin filament assembly, but only Tmod3 could sequester cytoplasmic ␤and ␥ cyto -actin (but not ␣ sk -actin) monomers and prevent polymerization under physiological conditions. Thus, differences in TM binding by Tmod1 and Tmod3 do not appear to regulate the abilities of these Tmods to cap TM-␣ sk -actin or TM-␤/␥ cytoactin pointed ends and, thus, are unlikely to determine selective co-assembly of Tmod, TM, and actin isoforms in different cell types and cytoskeletal structures. The ability of Tmod3 to sequester ␤and ␥ cyto -actin (but not ␣ sk -actin) monomers in the absence of TMs suggests a novel function for Tmod3 in regulating actin remodeling or turnover in cells.
Actin filament assembly and turnover are essential for the mechanical strength and physiological function of virtually all eukaryotic cells and are regulated by a host of actin-binding proteins, including side-binding and capping proteins. Recently, members of the tropomodulin (Tmod) 2 family of actin filament (F-actin) pointed end capping proteins have attracted attention as key regulators of actin filament architecture and organization (1,2). Of the four known Tmod isoforms (Tmods1-4), the best characterized isoforms are Tmod1 and Tmod3, both of which are present in diverse cells and tissues (2). Tmod1 is predominantly expressed in postmitotic, terminally differentiated cells where it caps the pointed ends of actin filaments in stable, long lived cytoskeletal assemblies, such as the sarcomeres of striated muscle cells (3) and the short actin filaments of the spectrin-actin membrane skeleton in red blood cells, where it regulates precise actin filament lengths and stability (2). Unlike Tmod1, Tmod3 is predominantly found in more dynamic cellular contexts, such as the dendritic actin network in the leading lamellipodia of migrating endothelial cells where it acts as a negative regulator of F-actin assembly and cell migration rates (4). Tmod3 is also present in proliferating erythroblasts prior to enucleation with Tmod3 deficiency leading to impaired differentiation, aberrant F-actin organization, and reduced enucleation of mouse erythroblasts (5). However, in a subset of its contexts, Tmod3 can be incorporated into a stable F-actin membrane skeleton in a manner similar to Tmod1, such as on the lateral membranes of polarized epithelial cells (6) or in the sarcoplasmic reticulum (SR) of skeletal muscle fibers (7). In these contexts, Tmod3 depletion leads to F-actin cytoskeletal disruption and impaired membrane mechanics (6,7) similar to Tmod1 depletion phenotypes in red blood cells (8) and ocular lens fibers (9). Moreover, Tmod3 levels decrease whereas Tmod1 levels increase as erythroid precursors prepare to enucleate and begin to synthesize membrane skeleton components during terminal erythroid differentiation (5), highlighting the divergent roles of Tmod1 versus Tmod3 in regulating dynamic versus stable actin filament architectures in proliferating versus terminally differentiated cells, respectively.
Tmods consist of an N-terminal tropomyosin (TM)/pointed end actin capping (TM-Cap) domain containing two TM-binding motifs and a TM-dependent actin pointed end capping motif followed by a C-terminal leucine-rich repeat/Cap domain containing five leucine-rich repeat motifs and a TM-independent actin pointed end capping motif (see Fig. 1A) (2). TMs are ␣-helical, rodlike F-actin side-binding proteins that bind to the TM-Cap domain of Tmods via Tmod recognition motifs located in the variable N-terminal exons of TM isoforms (see Fig. 1B) (10 -17). Tmod1 caps the pointed ends of TM-decorated actin filaments with an affinity more than 1,000-fold greater than for bare actin pointed ends (K d ϳ0.2 M for pure actin and in the nM-to-pM range for TM-actin filaments) (18 -20). In addition, mutations in Tmod1 that disable striated muscle ␣/␤TM binding lead to reduced Tmod1 capping of ␣/␤TMcoated actin filaments in vitro (21)(22)(23) as well as ineffective targeting of Tmod1 to thin filament pointed ends in the sarcomeres of both cultured cardiac and skeletal myocytes (23,24). Moreover, various combinations of Tmod and TM isoforms show distinct binding affinities (20,(25)(26)(27), and Tmod1 caps actin filaments coated with nonmuscle TM5a more strongly than filaments coated with striated muscle ␣/␤TM in vitro (20), suggesting that TM isoforms may selectively modulate the pointed end capping activities of Tmods in cells. However, the abilities of different TM isoforms to regulate the pointed end capping of Tmod1 versus Tmod3 have not been studied.
Although it is known that TM isoforms can differentially govern the pointed end capping function of Tmod1 (20), it remains unknown whether different actin isoforms can govern the pointed end capping function of any Tmod. Although the six actin isoforms share at least 93% sequence identity, they show differential expression patterns and are associated with distinct cytoskeletal assemblies (28). In skeletal muscle fibers, Tmod1 caps the pointed ends of the skeletal muscle ␣-actin (␣ sk -actin) that polymerizes into the sarcomeric thin filaments, whereas Tmod3 is associated with the cytoplasmic ␥-actin (␥ cyto -actin) that polymerizes into the SR-associated membrane skeleton (7). Skeletal muscle fibers also contain a costamere-associated cytoplasmic ␤-actin network (29), although which Tmod isoforms (if any) associate with this network remain unknown. The segregation of Tmod1 versus Tmod3 with sarcomeric ␣ skactin filaments versus SR-associated ␥ cyto -actin filaments, respectively, suggests that Tmod1 and Tmod3 may exhibit different capping activities for the pointed ends of TM-coated ␣ sk -actin filaments as compared with ␤-actin or ␥ cyto -actin filaments. On the other hand, there does not seem to be a simple one-to-one correspondence between Tmods and actin iso-forms because Tmod1 is associated with the TM5b/TM5NM1coated ␤-actin (not ␣ sk -actin) filaments in the spectrin-actin membrane skeleton of red blood cells (30 -34). In addition to capping pointed ends, in vitro biochemical studies have shown that, in the absence of TMs, Tmods can also bind ␣ sk -actin monomers and weakly nucleate ␣ sk -actin assembly (35,36). Tmod1 and Tmod3 have similar abilities to nucleate ␣ sk -actin polymerization (36), and both Tmod1 and Tmod3 can bind to ␣ sk -actin monomers at low ionic strength (36). However, only Tmod3 has been shown to reduce ␣ sk -actin polymerization at physiological ionic strength likely by sequestering ␣ sk -actin monomers (35). Neither Tmod1 nor Tmod3 interactions with cytoplasmic ␤and ␥ cyto -actin have been studied either in the presence or in the absence of TMs.
Here, we compare Tmod1 and Tmod3 with respect to Tmod-TM binding, Tmod-TM-actin pointed end capping, and Tmod/actin interactions using various combinations of Tmod, TM, and actin isoforms. Tmod1 bound to muscle (␣/␤TMs) and nonmuscle TMs (TM5NM1 and TM5b) ϳ5-fold better than Tmod3, but Tmod1 and Tmod3 showed no significant differences in pointed end capping of ␣ sk -actin filaments coated with any of these TMs. Moreover, both Tmod1 and Tmod3 were equally effective at capping platelet ␤/␥ cytoactin filaments coated with TM5NM1. Thus, Tmod1 and Tmod3 do not discriminate between TM isoforms or actin isoforms with respect to high affinity capping of pointed ends, and the TM-actin pointed end capping activities of Tmod1 and Tmod3 fail to explain the differential associations of these Tmods with diverse cytoskeletal structures in cells. However, in the absence of TMs, Tmod3 (but not Tmod1) sequestered cytoplasmic ␤and ␥ cyto -actin monomers and inhibited polymerization but had relatively little effect on steady state ␣ sk -actin polymerization in a sedimentation assay. We propose that Tmod3 functions to sequester ␤and ␥ cyto -actin monomers and inhibit filament assembly in dynamic regions of cells with reduced levels of TMs, such as protruding lamellipodia (37,38). This explains previous observations that Tmod3 levels are inversely correlated with lamellipodial F-actin levels, providing a mechanism for negative regulation of endothelial cell migration rates by Tmod3 (4). In other regions of cells containing TMs, TM binding to polymerizing ␤and ␥ cyto -actin filaments may switch Tmod3 from relatively low affinity (M) actin monomer binding to high affinity (nM) pointed end capping, thereby stabilizing Tmod3⅐TM⅐actin filament complexes.

EXPERIMENTAL PROCEDURES
Proteins-␣ sk -Actin was prepared from rabbit muscle acetone powder as described previously (39). N-(1-Pyrene)iodoacetamide-labeled ␣ sk -actin was prepared and stored as described previously (40). For nonmuscle actin capping and nucleation experiments, human platelet actin (ϳ85% ␤-actin, 15% ␥ cytoactin) and pyrene-labeled human platelet actin were obtained from Cytoskeleton, Inc. (Denver, CO), solubilized to prepare G-actin, and pyrene-labeled as described previously (40). For actin monomer sequestration experiments, ␤-, ␥ cyto -, and insect actins were expressed using a baculovirus-driven expression system in insect cells and purified as described previously (41). Protein concentrations of actins were determined from the absorbance at 290 nm and an extinction coefficient of 0.63 M Ϫ1 cm Ϫ1 (41). Capping protein of chicken skeletal muscle (CapZ) was a gift from Dorothy A. Schafer (University of Virginia, Charlottesville, VA) and Sumiko Kimura (Chiba University, Japan). Rabbit skeletal muscle ␣/␤TM was prepared as described previously (42). Recombinant rat TM5b (TM5b cDNA in the pJC20 vector), a gift from Michael A. Geeves (University of Kent, Canterbury, UK), was expressed in BL21 Escherichia coli and purified as described previously (43). A clone of human TM5NM1 was a gift from Andrea Bacconi (The Scripps Research Institute, La Jolla, CA) and amplified from a pEGFP-TM5NM1 expression vector using PCR primers designed to introduce NcoI and XhoI restriction sites for cloning into the bacterial pET-14b expression vector (Novagen). Recombinant human TM5NM1 was expressed in BL21 E. coli and purified in the same manner as for TM5b. Rat ␣TM2 was a gift from Sarah E. Hitchcock-DeGregori (Rutgers University, Piscataway, NJ) (44). TM protein concentrations were determined using a BCA Protein Assay kit (Pierce). Human Tmod1, mouse Tmod3, and their TM binding-incompetent mutants (Tmod1-L27G/I131D and Tmod3-L29G/L134D) were expressed as glutathione S-transferase fusion proteins in BL21 E. coli and purified by thrombin cleavage followed by affinity chromatography using glutathione-Sepharose and ion exchange chromatography as described previously (36). Concentrations of purified Tmods and mutants were determined spectroscopically as described previously (36).
TM-Tmod Binding Assays-We determined the relative strengths of Tmod/TM interactions by performing blot overlay assays as described previously (45,46). Briefly, increasing amounts of purified wild-type or mutant Tmod1 and Tmod3 were electrophoretically separated by SDS-PAGE on 12% Laemmli minigels (47) and transferred to nitrocellulose in methanol-containing transfer buffer. Parallel gels were stained with Coomassie Blue to control for protein loading or transferred to nitrocellulose membranes to assess TM binding. Blots were incubated with 5 g/ml purified TMs, and TMs bound to immobilized Tmods were then detected using anti-TM antibodies as follows: TM311 (1:1,000; Abcam) for rabbit skeletal muscle ␣/␤TM, AB5441 (1:1,000; Millipore) for TM5b, and AB5447 (1:1,000; Millipore) for TM5NM1. After washing, blot overlays were incubated with HRP-conjugated secondary antibodies (1:10,000; Invitrogen). Bands were detected using an enhanced chemiluminescence kit (Pierce) according to the manufacturer's instructions and exposed to film. Background-corrected band intensities were quantified in NIH ImageJ.
To confirm that the Tmod-TM binding avidities determined in the blot overlay assays were not confounded by SDS-PAGE treatment and denaturation of Tmods, we also performed dot blot overlay assays in the absence of detergents. Briefly, 200 ng of wild-type Tmods were diluted in 400 l of 5 g/ml BSA in a Hepes-buffered salt solution and then directly spotted onto a nitrocellulose membrane using a Schleicher & Schuell dot blot apparatus. BSA alone was also spotted as a negative control for TM binding, and membranes were stained with Ponceau S to control for total protein loading. Incubation with TMs and detection with anti-TM antibodies were then performed as described in the previous paragraph. After washing, blots were incubated with 680LT-conjugated secondary antibodies (1:20,000; LI-COR Biosciences) and imaged on a LI-COR Odyssey infrared imaging system (resolution, 84 m). Backgroundcorrected dot intensities on blots were quantified in NIH ImageJ. For background correction, the intensity of the "BSA alone" dot was subtracted from the intensities of the Tmod1 and Tmod3 dots for each TM. The intensity of the BSA alone dot was never more than ϳ2% of the total (uncorrected) signal intensity of TM bound to each Tmod.
Co-sedimentation Assay for TMs Binding to F-actin-The binding affinity of TMs to F-actin was measured by co-sedimentation as described previously (48)  were ultracentrifuged in a Beckman TLA100 rotor at 150,000 ϫ g for 20 min, and supernatants were removed. Supernatants and pellets were solubilized in SDS gel sample buffer, adjusted to the same volume, and analyzed by SDS-PAGE on 4 -20% gradient minigels. Gels were stained with Coomassie Blue, and background-corrected band intensities were quantified by densitometry using NIH ImageJ.
Fluorescence Assays for Actin Polymerization-Elongation rates at actin filament pointed ends were measured as described previously (18,49) but with slight modifications. Briefly, 1.8 M G-actin (8% pyrene-labeled for ␣ sk -actin or 10% pyrene-labeled for human platelet ␤/␥ cyto -actin) was converted to Mg 2ϩ -actin as described previously (50). CapZ-capped actin filament nuclei (CapZ:actin, 1:100) were prepared by mixing 10 M G-actin and 100 nM CapZ in G-buffer followed by addition of 1 ⁄ 10 volume 10ϫ polymerization buffer and incubation overnight on ice. Seeds (final concentration, 1.6 nM CapZ, 160 nM actin) were incubated with the indicated concentrations of Tmods and TMs added to the indicated concentrations of G-actin for 5 min at 25°C and polymerization was initiated immediately by addition of 1 ⁄ 10 volume of 10ϫ polymerization buffer (200 mM Hepes, pH 7.5, 1 M KCl, 20 mM MgCl 2 , 2 mM DTT, 5 mM ATP, 10 mM EGTA). Fluorescence measurements (excitation, 366.5 nm; emission, 407 nm) were performed using a Fluoromax 3 fluorometer (HORIBA Jobin Yvon, Edison, NJ) or an Envision 2103 Multilabel Reader (PerkinElmer Life Sciences). Capping activities for Tmod1 or Tmod3 were obtained from initial elongation rates over the first 1 min as determined from the slopes of the polymerization traces by linear regression analysis using Microsoft Excel. Rates in the presence of increasing concentrations of Tmods or TMs were divided by the rate in the absence of Tmod, yielding a series of rate/control rate ratios, which were converted into pointed end capping efficiencies using the following formula: Pointed end capping efficiency ϭ (1 Ϫ Rate/ Control rate) ϫ 100 (%). Thus, when rate/control rate ϭ 0, 100% of pointed ends are capped.
Nucleation of platelet ␤/␥ cyto -actin was measured as described previously for ␣ sk -actin (36) but with the following modifications. 3.4 M G-actin (5% pyrene-labeled) was converted to Mg 2ϩ -actin as described previously (36), and then the indicated concentrations of Tmod proteins were added and incubated for 1 min. Polymerization was then initiated as above. Fluorescence measurements were performed using an Envision 2103 Multilabel Reader as described above.
G-actin Binding Assays-Non-denaturing PAGE was performed as described by Safer (51). G-actin and increasing concentrations of either Tmod1 or Tmod3 were incubated in buffer A (2 mM Tris, pH 8.0, 0.2 mM CaCl 2 , 0.2 mM DTT, 0.2 mM ATP) for 20 min at 25°C. Protein mixtures were then supplemented with 0.25 volume of loading buffer (50% glycerol, 0.05% bromphenol blue) and electrophoresed using a Bicine/triethanolamine buffer system. Proteins were visualized by staining with Coomassie Blue. In sedimentation assays for G-actin sequestration, G-actin was mixed with Tmod1 or Tmod3, and polymerization was initiated by adding 10ϫ polymerization buffer. After polymerizing to steady state for 16 h at 25°C, protein mixtures were separated by ultracentrifugation in a Beckman TLA100 rotor at 300,000 ϫ g for 20 min. Supernatants and pellets were solubilized in SDS gel sample buffer, adjusted to the same volume, and analyzed by SDS-PAGE on 12% gels followed by Coomassie Blue staining. In both types of G-actin binding assays, background-corrected protein band intensities were quantified using NIH ImageJ.

RESULTS
Sarcomeric and Cytoskeletal TMs Bind to ␣ sk -Actin Filaments with Varying Affinities-TM isoforms bind to ␣ sk -actin filaments with distinct affinities (13), affecting TM polymerization along F-actin and the probability that the terminal TM will associate with the pointed end (52,53). In its pointed end position, TM influences the actin pointed end capping activities of Tmods (18 -20, 25, 26). Therefore, we assembled a panel of muscle and cytoskeletal TM isoforms with varying affinities for F-actin, including native striated muscle ␣/␤TM, recombinant rat ␣TM2, and two cytoskeletal TMs, recombinant TM5NM1 from the ␥TM gene and TM5b from the ␣TM gene (Fig. 1). We examined the F-actin binding affinity of these TM isoforms for ␣ sk -actin by standard co-sedimentation assays followed by SDS-PAGE, Coomassie Blue staining, and densitometry. The data were fit to the Hill equation to determine the affinity (K app ) of TMs for F-actin and their Hill coefficients, indicative of cooperativity. Binding of all TMs showed sigmoidal saturation curves, indicating their cooperative binding to F-actin, as expected ( Fig. 2A) (52,53). ␣/␤TM and TM5b were the strongest F-actin-binding TMs, achieving saturation at 1.0 -1.5 M TMs, whereas TM5NM1 and recombinant ␣TM2 were weaker binding TMs, achieving saturation between 2.0 and 2.5 M TMs. Their K app values ranged from the strongest at 2.20 Ϯ 0.11 M Ϫ1 for native ␣/␤TM to the weakest at 0.88 Ϯ 0.10 M Ϫ1 for recombinant ␣TM2, whereas their Hill coefficients ranged from 2.12 Ϯ 0.13 for TM5NM1 to 2.97 Ϯ 0.31 for TM5b (Fig.  2B). The strong F-actin binding of native ␣/␤⌻⌴ and TM5b is consistent with previous reports (54 -56) as is the weak F-actin binding of bacterially expressed recombinant ␣TM2 (57). Previous studies have demonstrated that long, exon 1a-containing muscle TMs, such as ␣TM2, require N-terminal acetylation for their high affinity binding to F-actin (which is not present in the bacterially expressed protein), whereas the short, exon 1b-containing TMs, such as TM5b, do not (57,58). Overall, the relative binding strengths of these TMs for ␣ sk -actin were from strongest to weakest: Tmod1 Interacts More Strongly with Both Muscle and Nonmuscle TMs than Does Tmod3-Similar to TM/␣ sk -actin interactions, TM/Tmod interactions are also isoform-specific (20,(25)(26)(27). Therefore, we compared the relative strengths of Tmod1 versus Tmod3 binding to ␣/␤TM, TM5b, and TM5NM1. ␣TM2 was not included based on previous studies showing undetectable binding of Tmods to recombinant unacetylated ␣TM2 (59). Binding was evaluated semiquantitatively using blot overlay assays in which each TM was incubated with immobilized Tmods or Tmod mutants on nitrocellulose blots followed by antibodies to detect the bound TMs (45,46). The amount of each TM bound was normalized to the level of Tmod or mutant protein by performing Coomassie Blue staining of a duplicate parallel gel (Fig. 3A) (45). These experiments showed that skeletal muscle ␣/␤TM bound more strongly to Tmod1 than Tmod3 with a ϳ5-fold difference in binding (Fig. 3, A and C), consistent with previous studies (45). TM5b and TM5NM1 also demonstrated stronger binding to Tmod1 as compared with Tmod3 with a similar ϳ5-fold difference in avidity (Fig. 3, B and C). SDS-PAGE treatment and resultant denaturation of the immobilized Tmods are not the reason for the differential interactions of Tmod1 and Tmod3 with the various TMs because dot blot overlays in which wild-type Tmods were spotted onto nitrocellulose membranes in the absence of detergents and then incubated with TMs yielded TM-Tmod binding avidities similar to those obtained with the conventional blot overlays with all TMs binding ϳ4 -5-fold more strongly to Tmod1 than to Tmod3 (Fig. 3, D-G).
To verify the specificity of TM binding in these assays, we also tested TM binding to Tmod1 and Tmod3 mutants in which both their first and second TM-binding sites were disabled by mutations in their TM-binding ␣1and ␣3-helices (21,22). As expected, all TMs exhibited significantly less binding to the double mutant Tmod1-L27G/I131D, demonstrating specificity of binding (Fig. 3, A--C). TM binding was also barely detectable with Tmod3-L29G/L134D, a mutant corresponding to Tmod1-L27G/I131D (Fig. 3, A-C), indicating that TMs recognize Tmod3 via TM-binding sites homologous to those in Tmod1 (Fig. 1A), as expected from previous studies with the TM-binding peptides of Tmods (60). All TMs tested displayed similar relative binding to the wild-type and mutant Tmod proteins with wild-type Tmod1 Ͼ Ͼ wild-type Tmod3 Ͼ Tmod1-L27G/ I131D Ͼ Tmod3-L29G/L134D. Although these data do not address differences in affinities of Tmod1 and Tmod3 for ␣/␤TM, TM5NM1, or TM5b (60), they do show that Tmod1 recognizes each TM better than does Tmod3. This is in agreement with measurements of K d values for binding of Tmod1 and Tmod3 fragments containing TM-binding Site 1 or Site 2 ( Fig. 1) to the N-terminal peptides ␣TM1a, ␣TM1b, and ␥TM1b corresponding to the ␣/␤TM, TM5b, and TM5NM1 proteins, respectively (60).
Tmod1 and Tmod3 Exhibit Similar Pointed End Capping Efficiencies for ␣ sk -Actin and ␤/␥ cyto -Actin Filaments Regardless of the TM Isoform Present-To compare the pointed end capping efficiencies of Tmod1 versus Tmod3 with various TM isoforms, we first tested the abilities of Tmods to inhibit initial elongation rates from ␣ sk -actin filament pointed ends. In these experiments, we measured pointed end elongation rates from CapZ-capped, TM-coated, pyrene-labeled ␣ sk -actin filament seeds (Fig. 4, A-D) (18,19,49). Pointed end capping efficiencies were expressed as the percent inhibition of the initial elongation rate in the presence of Tmod with respect to the control rate without Tmod (Fig. 4, E-H). Surprisingly, despite substantial differences in TM binding (Fig. 3), both Tmod1 and Tmod3 displayed similar efficiencies of pointed end capping for ␣ skactin filaments in the presence of ␣/␤TM, TM5b, or TM5NM1 (Fig. 4, A-C and E-G). ␣/␤TMand TM5b-coated ␣ sk -actin filaments were capped most strongly by Tmods with ϳ40 -50% capping efficiency at 2 nM Tmods, whereas TM5NM1-coated ␣ sk -actin filaments were capped at ϳ30% efficiency at 2 nM Tmods (Fig. 4, A-C and E-G). A difference between pointed end capping by Tmod1 and Tmod3 was only observed with recombinant ␣TM2-coated ␣ sk -actin filaments with Tmod1 capping pointed ends more than twice as efficiently as Tmod3 (Fig. 4, D and H). Although Tmod1 was reasonably efficient at pointed end capping of ␣TM2-coated ␣ sk -actin filaments (80% capping at 10 nM Tmod1), Tmod3 was only ϳ30% efficient at 10 nM Tmod3 (Fig. 4H). Because all the TM concentrations in these assays were at or close to saturation for TM binding to actin (Fig. 2), the similarities or differences in Tmod1 and Tmod3 capping in the presence of the various TMs are most likely not directly related to their relative TM-actin binding affinities. Instead, differences in high affinity capping of TMactin filament pointed ends by Tmod1 and Tmod3 may only be revealed in the case of an exceptionally weak Tmod/TM interaction as is the case for recombinant unacetylated ␣TM2 and Tmod1 (59). However, because muscle TMs are acetylated in vivo, differences in Tmod1 and Tmod3 capping of ␣/␤TM-␣ skactin filaments are unlikely to explain selective assembly of Tmod1 with sarcomeric ␣/␤TM-␣ sk -actin thin filaments (3).
In another approach to compare capping of ␣ sk -actin filaments by Tmod1 and Tmod3 in the presence of ␣/␤TM, TM5b, or TM5NM1, we tested high affinity Tmod capping as a function of TM concentration. We reasoned that, at subsaturating TM concentrations, there would be a reduced probability that TM molecules would be located at the pointed end position to enhance the capping affinities of Tmods (18 -20, 52, 53). Therefore, we compared the pointed end capping efficiency of Tmods over a range of TM concentrations with ␣ sk -actin filaments. With the exception of the 0.5 M concentration, the pointed end capping activities of Tmod1 and Tmod3 were indistinguishable for ␣/␤TM (Fig. 5A). Furthermore, both Tmod1 and Tmod3 displayed similar extents of pointed end capping at all concentrations of TM5b and TM5NM1 (Fig. 5, B and C). It is likely that the high affinity interaction of Tmods with TM-actin pointed ends favors TM binding along the pointed end regions of actin filaments over all TM concentrations. In this experiment, both Tmod1 and Tmod3 achieved 70 -80% capping with 2.5 nM Tmod at 0.5 M TM5b and 1.0 M TM5NM1, which are TM concentrations that half-maximally saturate binding to F-actin as measured in our co-sedimentation assays (Fig. 2). Thus, although the blot overlay assays indicated that Tmod1 binds ϳ4 -6-fold more strongly than Tmod3 to all TMs tested (Fig. 3), this is not reflected by similar differences in high affinity pointed end capping of Tmod1 and Tmod3 in the presence of these TMs. Collectively, these data fail to identify noteworthy differences in high affinity capping of ␣ sk -actin filaments by Tmod1 and Tmod3 in the presence of different TMs. Therefore, the TM-dependent pointed end capping activities of

Differential Interactions among Tmods, TMs, and Actins
Tmod1 and Tmod3 are unlikely to account for the preferential association of Tmod1 with ␣ sk -actin structures, such as the thin filaments in skeletal muscle sarcomeres (3).
Next, we tested whether Tmod1 or Tmod3 differentially capped ␤/␥ cyto -actin filaments coated with a nonmuscle TM. In these experiments, we used purified human platelet ␤/␥ cytoactin. CapZ-capped ␤/␥ cyto -actin filament seeds were coated with TM5NM1, and pointed end elongation rates were measured by adding ␤/␥-actin monomers (10% pyrene-labeled) in the presence of saturating concentrations of TM5NM1. Results show that both Tmod1 and Tmod3 capped pointed ends of TM5NM1-coated ␤/␥ cyto -actin filaments with similar efficiencies at both 5 and 10 nM Tmod (Fig. 6). Based on these experiments with ␤/␥ cyto -actin and the experiments described above with ␣ sk -actin, no significant differences were observed between Tmod1 and Tmod3 with respect to high affinity capping of TMactin filaments. Thus, differences in Tmod1 and Tmod3 capping of actin filaments composed of ␤/␥ cyto -actin and nonmuscle TMs are unlikely to account for preferential Tmod3 assembly with ␥ cyto -actin filaments (containing TM5NM1 and TM4) in the SR of striated muscle (7) or Tmod1 assembly with ␤-actin filaments (containing TM5NM1 and TM5b) in the erythrocyte membrane skeleton (8).
Tmod1 and Tmod3 Similarly Nucleate ␤/␥ cyto -Actin Filaments in the Absence of TMs-Tmods can promote ␣ sk -actin polymerization in the absence of TMs by stabilizing spontaneously formed actin nuclei with activity at submicromolar concentrations of Tmods (35,36,49). Tmod1 and Tmod3 similarly promote ␣ sk -actin filament assembly with a 200 nM concentration of either Tmod nucleating assembly of ϳ3-4fold increased numbers of barbed ends compared with ␣ sk -actin alone (36). We tested whether Tmod1 and Tmod3 might be different at promoting F-actin assembly by comparing the kinetics of spontaneous ␤/␥ cyto -actin polymerization in the presence of increasing concentrations of Tmod1 or Tmod3. Using pyrene-actin polymerization assays, we observed that 50 -300 nM Tmod1 or Tmod3 similarly enhanced the rate of 3.4 M ␤/␥ cyto -actin polymerization in a dose-dependent fashion (Fig. 7, A and B). Although higher concentrations of Tmod3 (400 nM) promoted even higher rates of ␤/␥ cyto -actin polymer-  ization, the final level of polymerized actin was lower (Fig. 7B) most likely due to ␤/␥ cyto -actin monomer sequestration by Tmod3 (see Figs. 8 and 9). This only becomes evident at later time points because of the slow on-rate of Tmod3 for G-actin (2.9 ϫ 10 Ϫ3 M Ϫ1 s Ϫ1 ) compared with the faster on-rate of Tmod3 for the pointed ends of spontaneously formed actin filament nuclei (35,36).
Calculation of the concentrations of free barbed ends in these assays showed that Tmod1 and Tmod3 had similar dose-dependent actin-nucleating activities with 300 nM Tmods leading to ϳ5-6-fold increases in the number of barbed ends over ␤/␥ cyto -actin alone (Fig. 7C). When we compared these data with previous data obtained for nucleation experiments using ␣ sk -actin (36), we found that neither Tmod1 nor Tmod3 nucleated ␣ sk -actin assembly as efficiently as ␤/␥ cyto -actin with only ϳ3-fold increases in the number of barbed ends at 400 nM Tmods with ␣ sk -actin (compare Fig. 7C and Ref. 36). For both ␣ sk -and ␤/␥ cyto -actin, Tmod3 promoted assembly of slightly more barbed ends than Tmod1, but the differences were statistically insignificant. Thus, the similar nucleating activities of Tmod1 and Tmod3 with both ␣ sk -and ␤/␥ cyto -actins indicate that differences in actin nucleation are unlikely to explain the selectivity of Tmod1 or Tmod3 for cytoskeletal actin filaments comprising different actin isoforms.
Tmod3 Sequesters ␤and ␥ cyto -Actin Monomers More Strongly than Does Tmod1-Similar to other actin-nucleating proteins, Tmods can also bind ␣ sk -actin monomers with Tmod3 having greater monomer binding activity than Tmod1 (35,36). However, the monomer binding interactions of Tmods with ␤and ␥ cyto -actin monomers have not been investigated. In these assays, we tested the binding of Tmods to native rabbit skeletal muscle ␣ sk -actin as well as to human cytoplasmic ␤and ␥ cytoactins obtained using a baculovirus-driven expression system in insect cells (41). The purified cytoplasmic actins were contaminated with 20 -25% insect actin (41), so purified insect actin was also tested as an additional control. First, we examined the interaction of Tmods with actin monomers using nondenaturing PAGE at low ionic strength as described previously (36,51). In these assays, Tmod co-migrates with the Tmod⅐Gactin complex, and formation of the Tmod⅐G-actin complex is reflected by a decreased intensity of the G-actin band as the Tmod concentration is increased (Fig. 8, A and B). Tmod1 and Tmod3 bound all G-actin isoforms (36) with Tmod3 displaying stronger overall binding than Tmod1 based on enhanced complex formation at lower concentrations of Tmod3 than Tmod1 (Fig. 8, A and B). Densitometric quantification of the percentage of actin complexed with Tmod at various Tmod concentrations did not reveal any dramatic differences between Tmod1 and Tmod3 in terms of their ability to bind the various G-actin isoforms (Fig. 8, C and D). However, non-denaturing PAGE is performed at low ionic strength, and assaying Tmod-G-actin binding with this methodology may not reflect physiological interactions.
To examine Tmod-G-actin binding in the presence of physiological salt, we performed G-actin sequestration assays. ␣ sk -Actin, cytoplasmic ␤and ␥ cyto -actins, and insect actin were polymerized to steady state in the presence of increasing concentrations of Tmod1 or Tmod3, and the F-actin was then ultracentrifuged to separate actin monomers from filaments followed by SDS-PAGE and Coomassie Blue staining of supernatants and pellets. On these gels, Tmod1 migrates slightly above actin, whereas Tmod3 migrates slightly below actin. Results show that Tmod1 had little effect on the amount of G-actin in the supernatant regardless of which actin isoform was tested, indicating that Tmod1 does not significantly bind or sequester G-actin at physiological salt conditions (Fig. 9, A and  C). This is in agreement with previous studies using pyreneactin polymerization assays showing that Tmod1 does not sequester ␣ sk -actin (18,19,35). By contrast, Tmod3 dramatically increased the amount of unpolymerized actin in the supernatant for cytoplasmic ␤and ␥ cyto -actins and insect actin in a Tmod3 concentration-dependent manner (Fig. 9, B and D). At the highest concentration of Tmod3 tested (6 M) and 4 M actins, nearly all of the ␤and ␥ cyto -actins and insect actin FIGURE 7. Tmod1 and Tmod3 similarly promote spontaneous platelet ␤/␥ cyto -actin polymerization. A and B, Tmod1 or Tmod3 was mixed with 3.4 M platelet ␤/␥ cyto -G-actin (5% pyrene-labeled), and salts were added to initiate polymerization. Actin polymerization kinetics were monitored by increased pyrene-actin fluorescence as described under "Experimental Procedures." AU, arbitrary units. In A, 300 nM Tmod3 was tested to allow direct comparison with 300 nM Tmod1 in a single experiment. C, concentration of ␤/␥ cyto -actin barbed (B) ends was calculated from the polymerization rate when 50% of monomers were polymerized and then normalized to the concentration of barbed ends in the "no Tmod" control. Data shown are mean Ϯ S.D. (error bars) of three experiments. remained in the supernatant. By contrast, Tmod3 had only a relatively small effect on the amount of ␣ sk -actin in the supernatant fraction. These results indicate that, first, Tmod3 binds to ␤and ␥ cyto -actins more strongly than Tmod1 and, second, Tmod3 binds to ␤and ␥ cyto -actins more strongly than to ␣ skactin. Note that the monomer sequestration activity of Tmod3 observed here operates in the micromolar concentration range (Fig. 9D), well above the nanomolar concentrations of Tmod3 required for high affinity pointed end capping of TM-coated ␣ sk -or ␤/␥ cyto -actin filaments (Figs. 4 -6). Thus, Tmod3 sequestration of ␤and/or ␥ cyto -actin might provide a mechanism for dynamic regulation of ␤and ␥ cyto -actin monomer-polymer levels in regions of cells that lack TMs, such as protruding lamellipodia (37,38).

Tmod-TM Isoform Binding and Capping of TM-coated Actin
Filaments-This study revealed differences between Tmod1 and Tmod3 in terms of their TM binding avidities for a diverse assortment of TM isoforms with Tmod1 binding ␣/␤TM, TM5NM1, and TM5b ϳ5-fold more strongly than Tmod3. The conclusion that Tmod1 is a stronger TM binder than Tmod3 is consistent with published measurements of K d values for the TM-binding Site 1 and Site 2 fragments of Tmod1 and Tmod3 binding to N-terminal ␣TM1a, ␣TM1b, and ␥TM1b synthetic peptides corresponding to the ␣/␤TM, TM5b, and TM5NM1 proteins, respectively, tested in our study (see Ref. 60). However, the differences that we observed for Tmod1 and Tmod3 binding to TMs are not directly comparable with those measured by Uversky et al. (60) because of the different assays as well as the fact that Uversky et al. (60) measured TM binding for the Tmod Site 1 and Site 2 fragments independently, whereas we studied TM binding to full-length Tmod1 and Tmod3.
In the case of Tmod1, a large body of evidence indicates that interaction of Tmod1 with TMs via the two TM-binding sites in the N-terminal domain of Tmod1 promotes its high affinity pointed end capping of TM-actin filaments (nM-pM K d ; Refs. 16, 18, 19, and 22-26). It has been proposed that the binding of the ␣1and ␣3-helices of Tmod1 to the terminal TMs at the actin filament pointed end may enhance the interaction of the actincapping site of Tmod1 with actin by providing two additional Tmod-binding sites at the pointed end (16,21). This is supported by experiments using various N-terminal Tmod1 fragments containing helix-disrupting or charge-altering mutations leading to concurrent loss of TM binding and TMdependent capping activity (15,21,22). Moreover, a Tmod1-R11K/D12N/Q144K triple mutant has greatly weakened binding to skeletal muscle ␣/␤TM along with severalfold reduced Tmod1 capping of ␣/␤TM-coated actin filaments (23,61). Although the TM-dependent actin pointed end capping of Tmod3 has not been studied previously, the sequence conservation observed between Tmod1 and Tmod3 in their TM-binding Sites 1 and 2 (60) predicts that the TM-dependent high affinity capping of Tmod3 utilizes similar sites and mechanisms as does that of Tmod1. Yet, surprisingly, despite the differences between Tmod1 and Tmod3 in TM binding ( Fig. 3 and Ref. 60), we did not observe differences in capping of TM-actin pointed ends for Tmod1 and Tmod3 either for either ␣ sk -actin filaments coated with skeletal muscle ␣/␤TM or cytoskeletal TM5NM1 and TM5b or for ␤/␥ cyto -actin filaments coated with TM5NM1. Only in the case of the bacterially expressed rat ␣TM2 was a difference in pointed end capping between Tmod1 and Tmod3 observed with Tmod3 roughly one-third as effective at capping ␣TM2-actin pointed ends as Tmod1 (Fig. 4). Recombinant muscle ␣TM2 is unacetylated with undetectable Tmod binding in binary assays (59), suggesting that severe attenuation of Tmod-TM binding may be required to detect differences in TM-actin filament pointed end capping by Tmods. Nevertheless, because this study shows that recombinant ␣TM2 still promotes relatively high affinity pointed end capping by Tmod1 (80% capping at 10 nM Tmod1), ␣TM2 may bind Tmod1 better than Tmod3 in the context of the actin filament pointed end. Alternatively, the presence of ␣TM2 may lead to a conformational change in the actin filament that enhances the interaction of Tmod1 more than that of Tmod3 with actin filament pointed ends.
The fact that Tmod1 and Tmod3 do not display significant differences in terms of their high affinity TM-actin capping activities (that are correlated with differences in TM binding) raises the possibility that Tmod/TM interactions may have additional functions at the actin filament pointed end. Using skinned muscle fiber mechanics and x-ray diffraction approaches, we recently showed that substitution of Tmod3 for Tmod1 at the thin filament pointed end in Tmod1 Ϫ/Ϫ mouse skeletal muscle impairs ␣/␤TM strand movement across the thin filament and reduces recruitment of myosin cross-bridges (62). It is possible that Tmod1 (but not Tmod3) interactions with ␣/␤TM and ␣ sk -actin at the pointed end induce long range conformational changes in ␣/␤TM and/or the actin filament that facilitate the cooperative azimuthal movement of ␣/␤TM during thin filament Ca 2ϩ activation (62). Future structural and biochemical studies of thin filament regulation with purified proteins in reconstituted systems will be required to evaluate this possibility. It is also known that myosin cross-bridge cycling can induce actin polymerization (63,64), pointing toward a novel (but untested) mechanism of isoform-specific regulation of actin dynamics by Tmod1 and Tmod3 via modulation of TM strand movement and myosin head activity.
The biochemical data generated in this study provide clues to how in vitro Tmod/TM/actin isoform interactions may underlie the in vivo characteristics of diverse Tmod-TM-actin structures in cells. For example, the similar pointed end capping activities of Tmod1 and Tmod3 with various TM and actin isoforms imply that, in skeletal muscle, TM and actin isoform differences per se are unlikely to account for Tmod1 selectively capping ␣/␤TM-coated ␣ sk -actin thin filaments and Tmod3 selectively capping SR-associated, TM5NM1/TM4-coated ␥ cyto -actin filaments (7,45). Instead, it is more likely that the association of Tmod3 with the SR is driven by its interaction with the membrane-associated protein small ankyrin 1.5 (sAnk1.5) at the M-line (7), whereas as-yet-unknown factors are involved in targeting and assembly of Tmod1 to the thin filaments. Such a mechanism is supported by investigations of developing skeletal muscle where Tmod1 is observed on pointed ends of the thin filaments during the assembly of myofibrils relatively early in skeletal muscle development (45,65), whereas Tmod3 remains cytoplasmic prior to its later adoption of a striated staining pattern signifying assembly into the SR (7,45). By contrast, in red blood cells where Tmod1 and Tmod3 are co-expressed late in differentiation (5), the stronger binding of Tmod1 to TM5NM1 and TM5b as compared with Tmod3 could account for the preferential assembly of Tmod1 with TMs on the ␤-actin filaments of the membrane skeleton (although a caveat is that their TM-␤-actin capping affinities are similar; Fig. 6) (8). Notably, Tmod3 does assemble with the ␤-actin filaments in Tmod1 Ϫ/Ϫ red blood cells but at reduced (one-fifth) levels with respect to Tmod1 in wild-type red blood cells (8), which may also reflect the weaker binding of Tmod3 to TM5NM1 and TM5b. Such differences in Tmod assembly could be exacerbated if Tmod concentrations were limiting during TM-actin filament assembly in vivo. Additional studies are needed to explain how or why the unique intermolecular affinities of various combinations of Tmod, TM, and actin isoforms contribute to the co-assembly of these isoforms in some cell types but not in others.
Tmod/Actin Isoform Interactions-We found that Tmod3 interacts with and sequesters cytoplasmic ␤and ␥ cyto -actin monomers more strongly than ␣ sk -actin monomers, reducing steady state ␤and ␥ cyto -actin polymerization. We also show that Tmod1 lacks a similar actin monomer-sequestering ability for any actin isoform in agreement with previous studies on Tmod1 with ␣ sk -actin (35). However, our previous experiments showed that Tmod3 also reduced pyrene-labeled ␣ sk -actin polymerization at steady state (35), which we did not observe here in co-sedimentation assays measuring ␣ sk -actin polymerization (Fig. 9). The reason for these differences is unclear, although one possibility is that pyrene-iodoacetamide labeling of ␣ sk -actin on Cys-373 (40) may have enhanced the interaction of Tmod3 with ␣ sk -actin, promoting monomer sequestration by Tmod3. Alternatively, the ␣ sk -actin used in the previous study (35) could have been contaminated with ␤and ␥ cytoactins, which are also present in skeletal muscle tissues.
Thus, Tmod3 (but not Tmod1) is a new member of a growing subset of actin-binding proteins that are capable of discriminating among actin isoforms, a group that includes annexin-5a (66), profilin (67), cofilin (68), thymosin-␤4 (69), L-plastin (70), ezrin (71), and ␤Cap73 (72) (for a review, see Ref. 28). Tmod3 does not discriminate between actin isoforms in nucleation assays or in pointed end capping assays (this study and Refs. 35 and 36), suggesting that the monomer-sequestering activity of Tmod3 may rely on divergent actin isoform residues that are exposed in the monomer but not at the polymer pointed end. Our previous mutagenesis study showed that mouse Tmod3 binds to ␣ sk -actin monomers via the TM-Cap domain (36). Chemical cross-linking and LC-MS/MS experiments have indicated that Tmod3 residues Glu-74 and Glu-79 cross-link to ␣ sk -actin subdomain 2 (actin Lys-61), whereas Lys-53, Lys-94, Lys-96, and Lys-169 cross-link to actin subdomain 1 (actin Asp-24, Asp-25, Glu-99, and Glu-100) (36). All of these residues are conserved among all Tmod and actin isoforms, suggesting that our studies did not detect the critical interacting residues that could confer selective recognition between Tmod3 and ␤and ␥ cyto -actins. Notably, skeletal muscle ␣ sk -actin and ␥ cyto -actin share 93% identity with their principal sequence differences residing in their ϳ10 N-terminal residues (73). The N-terminal ends of actins are located in actin subdomain 1 where Tmod3interacting residues are located (36), suggesting that differences in the N-terminal sequences of actin isoforms may directly affect isoform-specific Tmod/actin interactions. Additionally, mutagenesis of N-terminal residues in yeast actin to muscle actin residues leads to allosteric changes in subdomain 2 (74,75), another Tmod3-binding site (36). Future biochemical and structural studies will be required to identify the interaction sites for Tmod3 on cytoskeletal ␤and ␥ cyto -actins.
Could Tmod3 act as a monomer-sequestering protein in vivo? Immunoprecipitation of FLAG-tagged Tmod3 from HEK293 cells in the presence of latrunculin A has demonstrated that Tmod3 can interact with ␤/␥ cyto -actin monomers in a cellular context (35). Moreover, Tmod3 is enriched in the leading lamellipodia of migrating human microvascular endothelial cells where F-actin levels are inversely proportional to Tmod3 levels in GFP-Tmod3 overexpression and Tmod3 siRNA depletion experiments (4). This is as expected from a monomer-sequestering protein and provides a simple explanation for reduced rates of cell migration resulting from Tmod3 overexpression; namely, increased actin monomer sequestration by Tmod3 would reduce the supply of actin available for polymerization and protrusion of lamellipodia, thereby reducing cell migration rates (4,76). Such a mechanism is expected to depend on reduced TM levels as TM binding to actin filaments would otherwise provide high affinity Tmod3-binding sites, leading to Tmod3 capping and stabilization of the filaments and more F-actin, which was not observed. Indeed, previous studies have reported reduced levels of TM in lamellipodia of migrating cells (37,38), although this phenomenon may depend on cell type or lamellipodial properties. Note that bulk actin sequestration in cells by Tmod3 is unlikely based on a global Tmod3 concentration of ϳ0.5 M in migrating human microvascular endothelial cells (4) as compared with a significantly higher concentration of actin, which constitutes 5-15% of the total protein content of endothelial cells (77). However, Tmod3 may be present at a higher local concentration in the lamellipodia of migrating cells based on immunolocalization indicating enrichment of Tmod3 in this region (4). By comparison, in Tmod1 Ϫ/Ϫ red blood cells where Tmod3 is increased and present at onefifth normal Tmod1 levels, there is no decrease in F-actin in the membrane skeleton or increase in ␤-actin monomers in the cytosol (8). In this markedly different cellular context (i.e. a highly stable actin cytoskeleton), the ␤-actin filaments in the membrane skeleton of the Tmod1 Ϫ/Ϫ red cells are associated with TM5NM1 and TM5b, which would favor high affinity Tmod3 capping of pointed ends and thereby preclude monomer sequestration by Tmod3. Thus, we propose that Tmod3 operates as a low affinity (M) monomer-sequestering protein in the case of TM-free actin filaments in dynamic cellular contexts but switches to a high affinity (nM) pointed end capping