The impact of tropomyosins on actin filament assembly is isoform specific

Abstract

Tropomyosin (Tpm) is an α helical coiled-coil dimer that forms a co-polymer along the actin filament. Tpm is involved in the regulation of actin's interaction with binding proteins as well as stabilization of the actin filament and its assembly kinetics. Recent studies show that multiple Tpm isoforms also define the functional properties of distinct actin filament populations within a cell. Subtle structural variations within well conserved Tpm isoforms are the key to their functional specificity. Therefore, we purified and characterized a comprehensive set of 8 Tpm isoforms (Tpm1.1, Tpm1.12, Tpm1.6, Tpm1.7, Tpm1.8, Tpm2.1, Tpm3.1, and Tpm4.2), using well-established actin co-sedimentation and pyrene fluorescence polymerization assays. We observed that the apparent affinity (K_d(app)) to filamentous actin varied in all Tpm isoforms between ∼0.1–5 μM with similar values for both, skeletal and cytoskeletal actin filaments. The data did not indicate any correlation between affinity and size of Tpm molecules, however high molecular weight (HMW) isoforms Tpm1.1, Tpm1.6, Tpm1.7 and Tpm2.1, showed ∼3-fold higher cooperativity compared to low molecular weight (LMW) isoforms Tpm1.12, Tpm1.8, Tpm3.1, and Tpm4.2. The rate of actin filament elongation in the presence of Tpm2.1 increased, while all other isoforms decreased the elongation rate by 27–85 %. Our study shows that the biochemical properties of Tpm isoforms are finely tuned and depend on sequence variations in alternatively spliced regions of Tpm molecules.

Keywords:

• actin
• actin elongation
• actin affinity
• co-sedimentation
• isoforms
• tropomyosin

Introduction

Actin is one of the most abundant and highly conserved proteins in eukaryotic cells. Actin filaments are involved in a plethora of critical cellular processes, including cell division, motility, intracellular transport, and cell adhesion.¹ These diverse functions of actin filaments are regulated by over 150 actin-binding proteins. Among these, tropomyosin (Tpm) is an α-helical coiled-coil protein dimer, which forms a continuous head-to-tail polymer along the major grooves of the actin filament. The Tpm polymer stabilizes actin filaments and is well known for its role in regulating muscle contraction. It has now become clear that Tpm is also a key regulator of the actin cytoskeleton whereby the ∼40 mammalian Tpm isoforms associated with functionally distinct actin filament populations in the cell control access of other actin binding proteins to the filament in an isoform-specific manner.² There are 4 Tpm genes (TPM1–4) and various isoforms arise from each gene via the use of 2 alternative promoters and the differential splicing of exons 2, 6 and 9. On the basis of their molecular weight, Tpms are classified into high molecular weight (HMW) isoforms that are encoded by 9 exons and typically contain 284–285 amino acids, and low molecular (LMW) isoforms that do not utilize exon 2 and consist of 245–248 amino acids.^3,4 Despite intensive research efforts to determine the physicochemical properties, cellular localizations and diverse roles of Tpm isoforms, the molecular basis for sorting of isoforms to different actin filament populations in the cell is still not well understood.

In vitro studies with purified proteins have revealed distinct biochemical and biophysical properties associated with different isoforms^5,6 while cellular and animal studies have shown isoform-specific localizations and functions⁷ leading to the proposal that sequence variation between exons could play a role in the sorting mechanism.⁸ A systematic comparison of isoforms under identical conditions is needed to identify sequences associated with the assembly and properties of different actin-Tpm filaments, especially because the biochemical properties of Tpms are highly dependent on experimental conditions which can vary between laboratories. In this study we compare basic biochemical properties of 4 HMW isoforms (Tpm1.1; Tpm 1.6; Tpm1.7 and Tpm2.1) and 4 LMW isoforms (Tpm1.8; Tpm1.12; Tpm3.1 and Tpm4.2) (Fig. 1). This list includes most of the commonly studied isoforms and represents a broad range of muscle and cytoskeletal Tpms involved in a variety of cellular processes as illustrated by the examples below. Tpm1.1 is the most extensively studied isoform. It is found in striated muscle where it is involved in the regulation of muscle contraction.⁹ The cytoskeletal isoforms Tpm1.6, Tpm1.7, Tpm3.1 and Tpm4.2 play different roles in the assembly of contractile stress fibers,¹⁰ a process that involves the recruitment of myosin II by Tpm4.2. These isoforms are also associated with a range of other structures and functions. For example, Tpm3.1 is required for the survival of tumor cells, and this dependency has been exploited to develop anti-cancer drugs targeting this isoform.¹¹ Tpm1.8 is a cytoskeletal isoform that is localized to the apical surface of epithelial cells and regulates the insertion of transport proteins into the cell membrane.¹² Tpm1.12 is a major neuronal isoform encoded by the TPM1 gene, which promotes binding of ADF/cofilin to the filament¹³ and is involved in neuronal morphogenesis and control of membrane protrusions.¹⁴ The only product of the TPM2 gene used in our study, Tpm2.1, is found in the cytoskeleton where it is involved in the regulation of the stability of focal adhesions and dorsal stress fibers,¹⁰ sensing of the mechanical properties of the external environment¹⁵ and the regulation of anoikis (detachment induced apoptosis).¹⁶ Tpm2.1 is also found in the contractile apparatus in smooth muscle cells but predominantly in the form of heterodimers.¹⁷

Figure 1. Exon usage and sequence alignment of selected Tpm isoforms. (A, B) Exon usage of the (A) HMW and (B) LMW Tpm isoforms used in this study (adapted from Geeves et al., 2015). Shades of gray denote the different TPM genes. (C) Amino acid sequence comparison of Tpm isoforms. The amino acid sequences were aligned using ClustalX2. Exons are indicated above the sequence in the gray bars. The background color groups residues according to their physicochemical properties: orange = G; blue = hydrophobic (A, V, F, M, I, L) and C; blue-green = hydrophilic (Y, H); purple = acidic (D, E); red = basic (R, K); green = hydroxyl (S, T) and amide (N, Q).

Here we show that a general purification protocol that does not require affinity purification tags is suitable for the purification of 8 Tpm isoforms and compare the binding of these isoforms to actin filaments using co-sedimentation analysis. We found that both HMW and LMW isoforms bind to skeletal and cytoskeletal actin with similar affinity but binding of HMW isoforms to skeletal F-actin shows higher cooperativity than the other combinations. Pyrene actin polymerization assays showed that Tpm isoforms could inhibit or promote the rate of actin elongation. Both properties showed no clear dependence on exon usage suggesting that actin-Tpm filament assembly is isoform specific but not dictated by a specific exon in a predictable manner.

Results

Tropomyosin purification

We optimized published Tpm purification protocols^18-20 to obtain a general purification process for Tpm without purification tags expressed in E. coli (Fig. 2A) that yielded highly pure protein for all isoforms used in this study with only minor modifications in the last step. The first step took advantage of the well-known heat stability of the protein and Tpm was obtained in the supernatant after heating the cell lysate and removing cell debris together with denatured proteins by centrifugation. Several rounds of precipitation at low pH followed by re-solubilization then further enriched the protein. Next we added a subtractive anion exchange chromatography step to the protocol to remove highly anionic contaminants such as nucleic acids, whereby Tpm was applied to the column in a high salt buffer, passed through the column without binding and was collected in the flow-through (Fig. 2B). Dilution of the sample before re-applying to the anion exchange resin allowed Tpm to bind to the column and the protein was eluted in a salt gradient at 150–200 mM NaCl (Fig. 2C). These first steps were the same for all Tpm isoforms and yielded a highly enriched Tpm preparation, which typically contained a major contaminating band migrating at ∼15 kDa on SDS PAGE (Fig. 2D). This band was already present in the cell lysate after induction of Tpm expression (Fig. 2A) and was attributed to a proteolysis product of Tpm. This band and other minor contaminants were removed during a final purification step using a ceramic hydroxyapatite column whereby the proteolysis product eluted prior to Tpm in the sodium phosphate gradient (Fig. 2E/F). Each isoform had a different elution profile on the ceramic hydroxyapatite column requiring optimization of the sodium phosphate gradient for elution. This step yielded highly purified Tpm for all isoforms included in this study (Fig. 2G).

Figure 2. Expression and purification of Tpm4.2. (A) SDS PAGE with Coomassie Blue staining of total cell protein with and without induction of Tpm expression with IPTG. (B) Elution profile of the subtractive anion exchange chromatography step for removal of highly anionic contaminants such as nucleic acids. Tpm (monitored by UV light absorption at 280 nm and 230 nm) was collected in the flow through (peak between ∼30 and 100 ml elution volume). Bound nucleic acids (monitored by UV light absorption at 260 nm) was eluted with buffer B at ∼280 ml elution volume. (C, D) Elution profile (C) and SDS PAGE with Coomassie Blue staining (D) of the anion exchange chromatography step using a QFF column. Peak fractions containing highly enriched Tpm typically eluted at 35% buffer and were combined as indicated by the red bar. (E, F) Elution profile (E) and SDS PAGE with Coomassie Blue staining (F) of the chromatography using a ceramic hydroxyapatite column. Peak fractions containing pure Tpm were combined as indicated by the red bar. Note that in panels (B), (C) and (E), the green line represents concentration of buffer B (%) and the magenta line represents conductivity (mS/cm). (G) SDS PAGE with Coomassie Blue staining of all purified Tpm isoforms used in this study.

Binding of Tpm isoforms to F-actin

The four TPM genes are highly conserved and variation between Tpm isoforms arises primarily from differences in exon usage.²¹ Thus, the amino acid sequences exhibit variability primarily in the N-terminal region (exons 1 and 2), middle region (exon 6) and the C-terminal region (exon 9) (Fig. 1). We used co-sedimentation assays to determine whether the differences in sequence and/or the size (HMW versus LMW) of recombinant Tpm isoforms have an effect on their binding to skeletal (α) F-actin (Fig. 3 A and B). In this assay, binding was measured from the appearance of Tpm in the pellet fraction collected by centrifugation after incubation of F-actin with various Tpm concentrations. The apparent equilibrium dissociation constants (K_d(app)) and the Hill coefficients (a measure of cooperativity) for Tpm isoforms were obtained by fitting the Hill equation to the curves derived from densitometry analysis of protein in the pellet and supernatant fractions resolved on SDS PAGE (Table 1). Tpm1.1 showed the strongest affinity for actin filaments with a K_d(app) of 0.12 ± 0.01 μM for skeletal F-actin, in agreement with previous measurements conducted under similar conditions.^22,23 The alanine-serine extension added to the N-terminus of the construct for this isoform mimics the N-acetylation of the native protein²⁴ and Tpm1.1 with this modification has been shown to bind F-actin with the same affinity as the native protein isolated from muscle.²² The weakest affinity was determined for Tpm2.1 (K_d(app) of 5.51 ± 0.41 μM for skeletal F-actin), an isoform found in smooth muscle (predominantly as a heterodimer) and in the cytoskeleton. Previous studies have shown even weaker binding for recombinant Tpm2.1 (K_d(app) > 20 μM) whereby addition of the alanine-serine extension strongly increases its affinity for F-actin (K_d(app) of 0.3 μM).²⁵ While acetylation of Tpm in muscle is required for strong binding to actin,²⁶ it is less clear what role acetylation plays in the cytoskeleton. All other HMW and LMW isoforms in this study are associated with the cytoskeleton and showed high affinity binding to F-actin with a K_d(app) in the range of ∼0.4–1.0 μM (Table 1). The values for some Tpm isoforms differed considerably from published studies, in particular for Tpm1.12, which has previously been found to bind weakly to F-actin.^6,27 These variations may reflect subtle differences in the way the experiments were performed in the different laboratories. This serves to emphasize the value of making such comparisons within a single experimental set (Table 1). While there was overall no pronounced difference in affinity between HMW isoforms and LMW isoforms, the Hill coefficients derived from the binding curves with skeletal F-actin was on average 3-fold higher for HMW isoforms than for LMW isoforms (Table 1).

Figure 3. Binding curves of various Tpm isoforms to F-actin. (A) Binding of HMW Tpm isoforms to skeletal F-actin for Tpm1.1 (▪), Tpm1.6 (▴), Tpm 1.7 (•), and Tpm2.1 (▾); (B) Binding of LMW Tpm isoforms to skeletal F-actin for Tpm1.12 (□), Tpm 1.8 (Δ), Tpm3.1 (○), and Tpm4.2 (∇). (C) Comparison of the binding of selected HMW and LMW Tpm isoforms to either skeletal or cytoskeletal F-actin: Tpm1.1 to skeletal (▪), and cytoskeletal (▴) F-actin; Tpm3.1 to skeletal (○), and cytoskeletal (∇) F-actin. The best fit of the Hill equation to the data is shown as a solid line (skeletal F-actin) or dashed line (cytoskeletal F-actin). The apparent equilibrium dissociation constants (K_d(app)) and the Hill coefficients (h) are listed in Table 1. Buffer conditions: 150 mM NaCl, 10 mM Tris-HCl pH 7.5, 2 mM MgCl₂, 0.5 mM DTT.

Table 1. Apparent equilibrium dissociation constants and Hill coefficients for binding of HMW and LMW tropomyosin isoforms to F-actin. Values for K_d(app) determined in previous studies for comparison: (a) Tpm1.1 with Ala-Ser extension: 0.18 μM,²⁵ 0.20 μM,²³; (b) Tpm1.6: 0.82 μM,³⁵ 0.32 μM⁶; (c) Tpm2.1: >20 μM²⁵; (d) Tpm1.8: 0.27 μM,³⁵ 0.09 μM⁶; (e) Tpm1.12: 3.48 μM,²⁷ >10 μM⁶; (f) Tpm3.1: 0.1 μM²⁷

		skeletal actin		cytoskeletal actin
Tpm isoform		Kd(app) (μM)	h	Kd(app) (μM)	h
HMW	Tpm1.1^(a)	0.12 ± 0.01	5.7 ± 0.8	0.18 ± 0.05	1.3 ± 0.4
	Tpm1.6^(b)	0.97 ± 0.08	3.2 ± 0.5	1.06 ± 0.40	2.4 ± 0.9
	Tpm1.7	0.55 ± 0.04	4.0 ± 1.3	0.59 ± 0.08	2.7 ± 0.8
	Tpm2.1^(c)	5.51 ± 0.41	6.4 ± 3.1	5.75 ± 0.70	3.8 ± 1.2
LMW	Tpm1.8^(d)	0.37 ± 0.05	1.8 ± 0.5	–	–
	Tpm1.12^(e)	0.71 ± 0.21	1.3 ± 0.4	–	–
	Tpm3.1^(f)	0.46 ± 0.11	1.8 ± 0.6	0.48 ± 0.8	2.5 ± 1.0
	Tpm4.2	0.96 ± 0.21	1.4 ± 0.3	–	–

We then determined the binding of all HMW isoforms and one LMW isoform (Tpm3.1) to cytoskeletal (85 % β/15 % γ) F-actin. Importantly, the values for K_d(app) and for the Hill coefficient determined for each isoform with cytosketetal F-actin were essentially the same as the corresponding values determined with skeletal F-actin. Thus, Tpm isoforms did not show preferential binding to skeletal or cytoskeletal actin isoforms, which presumably reflects the high surface conservation of mammalian actin isoforms.²⁸ The only exception was Tpm1.1, which showed a higher Hill coefficient with skeletal F-actin than with cytoskeletal F-actin.

Effects of Tpm isoforms on actin elongation

Tpm can modulate the kinetics of in vitro actin assembly from G-actin in an isoform-dependent manner in the absence^27,29 and presence of F-actin seeds.³⁰ To examine the effect of the 8 Tpm isoforms studied here on this process, we measured the increase of fluorescence intensity resulting from incorporation of pyrene actin into filaments. Pyrene-labeled G-actin (2 μM, 10 % pyrene label) was copolymerized with Tpm isoforms at saturating concentrations in the presence of non-labeled F-actin seeds. Under these conditions actin nucleation does not influence the reaction kinetics and the rate of pyrene actin incorporation is dominated by filament elongation occurring at the barbed end.

Recombinant Tpm1.1 strongly decreased the elongation rate of filaments by ∼80 % relative to elongation in the absence of Tpm (Fig. 4), which was consistent with previous studies where Tpm obtained from muscle (containing mainly Tpm1.1) inhibited the seeded actin assembly rate by ∼40 %.³⁰ In similar experiments it has been shown that muscle Tpm²⁷ and recombinant Tpm1.1 with N-terminal alanine-serine extension²⁹ also strongly inhibit spontaneous actin assembly in the absence of F-actin seeds. Other HMW isoforms from the TPM 1 gene (Tpm1.6 and Tpm1.7) showed a milder inhibitory effect on actin elongation (∼30 % decrease) while Tpm2.1 resulted in a small but significant increase in actin growth rate by about 15 % (Fig. 4). The LMW isoforms examined here slowed down actin elongation to differing degrees (Fig. 4) whereby the strongest inhibitor was Tpm1.8 (∼85 %) followed by Tpm3.1 (∼70 %). We have observed a robust inhibitory effect of Tpm3.1 on F-actin growth under similar conditions previously but to a smaller extent³¹ while others have reported either no effect²⁹ or a slight acceleration²⁷ in the presence of Tpm3.1. Stimulatory effects have also been reported previously for Tpm1.6 and Tpm1.8, which was opposite to the inhibition observed for these isoforms in our study. In general we observed that the effect of Tpm isoforms on actin assembly was sensitive to changes in experimental conditions and the batch of actin and Tpm used for the assay. Nevertheless, individual isoforms tested under the same conditions had distinct effects on actin elongation.

Figure 4. Effect of HMW and LMW Tpm isoforms on actin polymerization. Normalized polymerization rates of 2 μM G-actin (10 % pyrene labeled) in the presence of 0.5 μM unlabelled F-actin seeds and saturating concentrations of various Tpm isoforms. The Tpm concentration was 2.5 μM for all isoforms with exception of Tpm2.1 (7 μM). Data were obtained from at least 3 independent experiments and the errors are given as mean ± SD. Buffer conditions: 100 mM KCl, 2 mM MgCl₂, 2 mM Tris-HCl pH 8, 0.1 mM CaCl₂, 0.2 mM ATP, 0.5 mM DTT, 1 mM NaN₃. The mean rate in the absence of Tpm was significantly different to the means in the presence of Tpm (one-way ANOVA with Tukey's test, p < 0.01 for actin vs. Tpm2.1, p < 0.001 for actin vs. Tpm1.12 and p < 0.0001 for actin vs. all other Tpm isoforms).

Discussion

Here we expressed and purified 8 recombinant Tpm isoforms and characterized their interactions with actin in vitro. We used human cytoskeletal isoforms Tpm1.6, Tpm 1.7, Tpm1.8, Tpm1.12, Tpm2.1 (present also in smooth muscle), Tpm3.1, Tpm4.2 and rat skeletal muscle isoform Tpm1.1. We were able to purify all Tpm isoforms in sufficient purity and quantities for in vitro biochemical assays without the need for affinity purification tags by using one general method with multiple chromatography steps. This method allowed us to test one of the most comprehensive sets of Tpm isoforms under the same experimental conditions. Our study provides an insight into the biochemical properties of Tpm in relation to actin-Tpm filament assembly, based purely on differences in size and amino acid sequence in specific Tpm regions. On the basis of our results we draw the following general conclusions: First, the affinity for most Tpm isoforms varies over a narrow range and does not differ with size but cooperativity of binding is higher for HMW isoforms (using exon 1a and 2b) than for LMW isoforms (using exon 1b). Second, cytoskeletal Tpm isoforms bind with similar affinity independent of the actin source but the differences in cooperativity between HMW and LMW isoforms are greater for skeletal actin. Third, Tpm regulates actin assembly in an isoform specific manner in vitro but this regulation is not linked to a specific pattern of exon usage.

Our data allow comparisons between isoforms derived from the same gene that differ in the usage of just one exon (Fig 1A and B). Regardless of whether isoforms differ at the N-terminus (exon 1a/2b in Tpm1.6 vs. exon 1b in Tpm1.8), in the middle of the molecule (exon 6b in Tpm1.6 versus exon 6a in Tpm1.7) or at the C-terminus (exon 9d in Tpm1.8 vs. exon 9c in Tpm1.12), the affinity is typically within a factor of 2−3. The skeletal muscle isoform Tpm1.1 (the only isoform in this set containing exon 9a) showed the highest affinity to F-actin with an approximately 8-fold reduction in K_d(app) compared to the cytoskeletal isoform Tpm1.6 (containing exon 9d). However, Tpm1.1 is also the only isoform with an alanine-serine extension at the N-terminus. Without this modification, binding of Tpm1.1 is weak²⁴ suggesting that N-acetylation can tune the affinity of the Tpm-actin interaction to at least the same extent as differences in isoform usage. While N-acetylation of muscle Tpm isoforms is well documented, there is no comprehensive data on the extent of N-acetylation of Tpm isoforms in the cytoskeleton. The isoforms Tpm1.6 and Tpm1.7 use the same exon 1a as Tpm1.1 and could in principle be acetylated by the same enzyme in the cell. However, unlike Tpm1.1. the unmodified versions of Tpm1.6 and Tpm1.7 showed strong binding to F-actin, and it has been observed previously that N-acetylation of Tpm1.6 leads to only a slight increase in its association with F-actin.³² Thus, acetylation has isoform-specific effects on actin affinity.

In a similar fashion we can compare isoforms that use the same combination of exons but originate from different genes. HMW isoforms Tpm1.7 and Tpm2.1 are 82.7 % identical (93.7 % similarity) with sequence variations between the isoforms occurring along the entire length of the molecule (particularly in exons 6 and 9). The K_d(app) determined for Tpm2.1 is ∼10-fold higher compared to Tpm1.7, whereby specific mutations could affect electrostatic interactions with the actin filament (e.g. K213 in Tpm1.7 to E213 in Tpm2.1³³), properties of the coiled coil structure (e.g., via insertion of polar residues into the hydrophobic core in Tpm2.1 at Q43, S158, S179) and/or the overlap complex between N- and C-terminus. It should be noted that Tpm2.1 with the alanine-serine extension exhibits a K_d(app) of 0.3 μM,²⁵ suggesting that N-acetylation can tune the affinity of this isoform to be in the same range as observed for other isoforms. LMW isoforms Tpm1.8 and Tpm4.2 (82 % identity/92.3 % similarity) show a similar level of conservation as the HMW pair discussed above but vary in affinity by a factor of 2−3.

Overall we find that cytoskeletal Tpm isoforms bind with similar affinity to F-actin (K_d(app) typically within a factor of 4). It has been shown that single point mutations of conserved Tpm residues involved in actin binding can change the affinity to actin by 2−10-fold,³³ i.e. single residue changes can have the same or a larger effect than multiple differences that distinguish the different isoforms. Thus, residue changes between isoforms have either little effect on affinity and/or that positive and negative effects of individual residues present in different regions of the molecule largely compensate for each other. Further, the effect of a particular exon on affinity has to be assessed in the context of which other exons and modifications (N-acetylation) are present. Taken together these observations suggest that the narrow range observed for K_d(app) (with the exceptions discussed above) may reflect a general property of the polymer system that is conserved across most (cytoskeletal) isoforms.

While we did not observe a dependence of affinity on size, we found that cooperativity was stronger for HMW isoforms with a value for h on average ∼3-fold higher than for LMW isoforms in their binding to skeletal actin. The same trend has been observed in previous studies comparing HMW isoform Tpm1.7 and LMW isoform Tpm3.1.^34,35 This difference is unlikely to be driven by the size of the molecule (HMW isoforms interact with 7 actin monomers while the shorter LMW isoforms interact with 6 actin monomers) but may be related to the differences at the N-terminus, i.e., the usage of exon1a/2b versus exon 1b. These differences might affect properties of the overlap complex between N- and C-terminus adjacent Tpm molecules in the polymer, which are thought to be crucial for cooperative binding.³⁶ There was no evidence in our data that N-terminal modification with alanine-serine further increases cooperativity relative to the cytoskeletal isoforms. It should be noted, however, that the different impact of skeletal vs. cytoskeletal actin suggests that there are isoform specific differences in the interaction of Tpms with actin surface residues, at least for skeletal actin.

Finally, a comparison of the effect of the Tpm isoforms included in this study on actin assembly showed that Tpm isoforms could inhibit or (for Tpm2.1) stimulate this process. The ability of Tpm isoforms to exert these opposing effects on actin assembly has also been observed in previous studies but pronounced differences exist for the effects of individual isoforms.^27,29,30 Importantly there was no clear pattern of dependence on exon usage and inhibition of actin assembly was not strongly dependent on affinity. The seeded actin polymerization assay used here is dominated by actin elongation at the barbed end and largely excludes the contribution of nucleation to the assembly kinetics. It has been proposed that Tpm binding affects the actin polymerization rate in this type of assay simply by stabilizing actin filaments and thereby preventing random breakages due to thermal fluctuations or mechanical forces that would otherwise generate new barbed ends as additional sites for elongation.^37,38 A prediction from this model is that actin polymerization would slow down in the presence of all isoforms. This prediction is contrary to our and previous³⁰ observations of acceleration for particular isoforms suggesting that other (additional) mechanisms are at play. For example, decoration with Tpm isoforms may impart a specific structural state to the actin filament in an isoform-dependent manner (or at least to the barbed end)³⁶ that modulates the rate of binding and/or dissociation of monomeric actin. Alternatively, the extent to which different Tpms overhang the actin barbed end during co-polymerization may influence the rate of monomer addition based on steric considerations. While the differential effect of Tpm isoforms on actin polymerization observed here may not be related to the physiological roles of the isoforms, they may report on structural differences between filaments that could contribute to isoform sorting.

What are the broader implications of our study for understanding the mechanisms that lead to functionally distinct actin filament populations decorated with specific Tpm isoforms observed in the cell? Our results point to a complex interplay of multiple interactions between the 2 polymer systems that can be tuned by exon usage and N-acetylation whereby general properties, such as affinity, can be maintained in a tight range while allowing variations in more specialized properties, such as isoform specific regulation of interactions with actin-binding proteins.² Maintaining similar affinities may be important to prevent strongly binding isoforms from outcompeting or displacing weakly binding isoforms. While the Tpm polymer may provide some sorting information, e.g. via structural changes of the actin filament, these mechanisms are unlikely to be sufficient to explain the precise sorting observed in cells. Indeed, our data shows that muscle and cytoskeletal isoforms do not bind preferentially to their respective “correct” native actin isoform. Other components of the filament system are thus likely to regulate the assembly of specific actin-Tpm filaments and several potential sorting machineries have been identified. Biochemical studies have suggested that actin nucleators from the formin family are involved in selecting specific Tpm isoforms for incorporation into growing filaments³⁰ and this sorting mechanism was recently demonstrated in fission yeast.³⁹ In the lamellipodium Tpm sorting may involve regulation by the Arp2/3 complex and cofilin.⁴⁰ While actin-Tpm filaments have traditionally been thought of as stable units such as that observed in the sarcomere, it is now becoming clear that Tpm in the cytoskeleton can be highly dynamic.^10,31 Maintaining the “molecular identity” (i.e. association with a particular Tpm isoform) and regulating turnover are also likely to depend on a combination of Tpm-intrinsic and context-dependent properties such as the presence of regulatory factors and biomechanical forces. The emerging view that specific functions associated with actin filaments depend on the presence of the correct regulatory factors has recently also been demonstrated for filament severing.⁴¹ Ultimately determining to what extent Tpm isoform sorting is intrinsic to the polymer system and defining the precise role of other molecular machinery in this process will require experimental approaches that can resolve competitive binding of different isoforms at the level of individual filaments in real time such as observation of filament-associated dynamics using fluorescence microscopy.

Materials and methods

Protein preparation

Rabbit skeletal muscle actin was prepared according to a published method.⁴² G-actin was further purified by gel filtration using a HiLoad 16/600 Superdex 75 column (GE Healthcare Life Sciences, USA) equilibrated with a buffer solution containing 2 mM Tris-HCl pH 8.0, 0.2 mM ATP, 0.2 mM CaCl₂, 0.2 mM DTT. Purified G-actin was labeled with pyrene by reaction of Cys-374 with N-(1-pyrenyl)iodoacetamide (P-29 Molecular Probes, USA) as described previously.⁴³ Free pyrene was removed by gel filtration as described above. The degree of labeling was typically ∼30 %.

Constructs for the expression of rat Tpm1.1 and human Tpm1.6, Tpm1.7, Tpm2.1, Tpm3.1 using the pET bacterial plasmid system were available from previous work.⁴⁴ The construct for expression of Tpm1.1 contains an Ala-Ser extension at the N-terminus, which has been shown to mimic the N-acetylation of the native Tpm.²⁴ The amino acid sequence of rat Tpm1.1 differs from the human sequence by only one residue (R220 in human, K220 in rat). The open reading frames for human Tpm1.8, Tpm1.12 and Tpm4.2 were synthesized and subcloned into a pDEST14 vector (GeneArt). All recombinant Tpms were expressed in E. coli BL21 (DE3) grown in 3 L (5 × 0.6 L) of 2YT broth containing 100 μg/ml ampicillin at 37°C until the cell density reached OD₆₀₀ ∼1. Protein expression was then induced by addition of isopropylthio-β-D galactosidase (IPTG, Gold Biotechnology) to a final concentration of 1 mM and the cells were grown for an additional 3 h (Fig. 2A). Cells were harvested by centrifugation (8,280 g, 4°C for 15 min, SLA-3000 rotor, Thermo Scientific) and the pellet was stored at −20°C until purification. The pellet was resuspended in lysis buffer solution (20 mM sodium phosphate pH 7.5, 500 mM NaCl, 5 mM MgCl₂, 1 mM NaN₃) containing 2 μg/ml DNAse, “Complete” protease inhibitor cocktail (Roche) and 1 mM PMSF. Cells were lysed on ice by sonication (30 s pulse on/30 s pulse off cycle, repeated 4 times). The lysate was heated in a water bath set at 80°C for 8 min and then cooled to room temperature. The lysate was centrifuged at 47,810 g, 4°C for 45 min (SS-34 rotor, Sorvall RC6+, Thermo Scientific) to remove cell debris. The supernatant was acidified to pH 4.7 and precipitated material was collected by centrifugation at 3900 g, 4°C for 12 min (SX 4250 rotor, Alegra X-22 R, Beckman Coulter). The pellet containing Tpm was resuspended in a buffer solution containing 500 mM NaCl, 100 mM Tris, 5 mM MgCl₂, 1 mM DTT, 1 mM NaN₃, pH 7.5 and the acidification procedure was repeated once or twice until the pellet was white. The resuspended pellet was then filtered (0.22 μm) and dialysed overnight against the loading buffer (20 mM Tris, pH 7.5, 500 mM NaCl, 0.5 mM DTT). After adjusting the conductivity to 45 mS/cm² (labChem-CP TPS conductivity meter), the supernatant was loaded onto a 40 ml QFF anion exchange column (GE Healthcare) equilibrated with 20 mM Tris, pH 7.5, 500 mM NaCl, 0.5 mM DTT and Tpm was collected in the flow through (Fig. 2B). The QFF resin was cleaned using 5 column volumes (CV) of 1 M NaCl and 5 CV of 1 M NaOH to remove DNA and other contaminants. The flow through containing Tpm was diluted 5-fold with 20 mM Tris pH 7.5, 0.5 mM DTT and reloaded on the QFF column equilibrated with the same buffer. Tpm was eluted with a linear gradient from 0–500 mM NaCl over 10 CV (Fig. 2C). Protein peak fractions were analyzed by SDS-PAGE (Fig. 2D) and pooled. The protein solution was then applied to a ceramic hydroxyapatite column (CHT type I, 5ml Cartridge, BioRad) equilibrated with a 10 mM sodium phosphate buffer pH 7.0 containing 1 M NaCl and 0.5 mM DTT. The sample was eluted using a linear gradient (10–160 mM) of sodium phosphate over 30 CV (Fig. 2E). Protein peak fractions were analyzed by SDS-PAGE (Fig. 2F), pooled and dialysed against a buffer solution containing 20 mM Tris pH 7, 100 mM KCl, 5 mM MgCl₂, 1 mM NaN₃, 0.5 mM DTT. Purified proteins were snap-frozen, lyophilized and stored at −20°C.

Actin co-sedimentation assays

Co-sedimentation assays were performed according to a published protocol⁵ with modifications. Increasing concentrations of Tpm (0.5–16 μM for Tpm2.1 and 0.2–6 μM for all other Tpm isoforms) were mixed with 3 μM rabbit skeletal F-actin or human cytoskeletal actin (85 % β-actin and 15 % γ-actin purified from platelets, Cytoskeleton, Inc., Cat. #APHL99) in a buffer solution containing 10 mM Tris-HCl pH 7.5, 150 mM NaCl, 2 mM MgCl₂ and 0.5 mM DTT in a total volume of 50 μL. Samples were incubated for 20 min at room temperature and then centrifuged at 435,680 g for 30 min at 20°C (TLA-120.1 rotor, Beckman-Coulter) to pellet F-actin and associated Tpm. Pellet and supernatant fractions were separated on SDS-PAGE, visualized by Coomassie blue staining and the protein bands were quantified by densitometry (Epson Perfection V750 Pro scanner and ImageJ). The ratio of the Tpm/actin density was plotted as a function of the concentration of free Tpm in the supernatant determined by densitometry. The Hill equation (Equation (1(1) ${\displaystyle v=\text{ n }{\left[\text{Tpm}\right]}^h/\left({\left({\text{K}}_{\text{d}(\text{app})}\right)}^h+{\left[\text{Tpm}\right]}^h\right)}$(1) )) was used to fit the binding curves:(1) ${\displaystyle v=\text{ n }{\left[\text{Tpm}\right]}^h/\left({\left({\text{K}}_{\text{d}(\text{app})}\right)}^h+{\left[\text{Tpm}\right]}^h\right)}$(1) where v is the Tpm/actin density ratio, [Tpm] is the concentration of free Tpm , n is the maximum Tpm/actin density ratio, h is the Hill coefficient and K_d(app) is the apparent equilibrium dissociation constant for Tpm binding to F-actin.

Pyrene-actin polymerization assays

Fluorescence assays were conducted using a Multi-mode microplate reader EnSpire® (Perkin Elmer, USA). The excitation and the emission wavelengths for pyrene-actin were set at 365 nm and 407 nm, respectively. Actin polymerization (2 μM of 10 % pyrene-labeled actin) was measured in a buffer solution containing 2 mM Tris-HCl pH 8, 0.1 mM CaCl₂, 0.2 mM ATP, 0.5 mM DTT, 1 mM NaN₃. Prior to all kinetic experiments, Ca-G-actin was exchanged to Mg-G-actin by addition of 0.2 mM EGTA and 0.05 mM MgCl₂ followed by incubation at room temperature for 10 min.^27,30 Polymerization was initiated by addition of 100 mM KCl and 2 mM MgCl₂ (final concentration) in the presence of 2.5 μM Tpm and 0.5 μM of unlabelled F-actin seeds. The only exception was Tpm2.1 which was mixed with actin at higher concentration of 7 μM because of its low affinity to actin (K_d(app) = 5.5 μM). Normalized data were fit to a single step association exponential model (Equation (2(2) ${\displaystyle \text{y }=\text{ A }\left(1–e^{-\text{kt}}\right)}$(2) )).(2) ${\displaystyle \text{y }=\text{ A }\left(1–e^{-\text{kt}}\right)}$(2) where the rate constant k was taken as an estimate for the polymerization rate k_obs (s⁻¹).

Data fitting: The data were fit to the equations using nonlinear least squares regression, using the EzyFit toolbox written by Frederic Moisy (http://www.mathworks.com/matlabcentral/fileexchange/10176-ezyfit-2-42) for Matlab (The MathWorks, Inc., Natick, Massachusetts, USA). The optimization minimised the chi squared difference between the data and fit using the Matlab function fminsearch which employs the simplex method of Lagarias,⁴⁵ a direct search procedure. Each data set was fitted individually. The parameter estimates were then averaged and the standard deviation determined. Simultaneous optimization of the all data together was also performed, however the variance of the parameters was found to be better estimated from the average of the individual fits.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

PWG is a non-executive Director of Novogen Pty Ltd, a company which is commercialising anti-tropomyosin drugs for the treatment of cancer.

Acknowledgments

We thank Shane P. Whittaker for help with purification of Tpm isoforms.

Funding

This research was supported by Australian Research Council grants FT100100411 (TB) and DP130100936 (TB, AC) and NHMRC grants 1004188 (PWG) and 1098870 (TB, PWG). PWG was also supported by The Kids Cancer Project.