A small molecule inhibitor of tropomyosin dissociates actin binding from tropomyosin-directed regulation of actin dynamics

The tropomyosin family of proteins form end-to-end polymers along the actin filament. Tumour cells rely on specific tropomyosin-containing actin filament populations for growth and survival. To dissect out the role of tropomyosin in actin filament regulation we use the small molecule TR100 directed against the C terminus of the tropomyosin isoform Tpm3.1. TR100 nullifies the effect of Tpm3.1 on actin depolymerisation but surprisingly Tpm3.1 retains the capacity to bind F-actin in a cooperative manner. In vivo analysis also confirms that, in the presence of TR100, fluorescently tagged Tpm3.1 recovers normally into stress fibers. Assembling end-to-end along the actin filament is thereby not sufficient for tropomyosin to fulfil its function. Rather, regulation of F-actin stability by tropomyosin requires fidelity of information communicated at the barbed end of the actin filament. This distinction has significant implications for perturbing tropomyosin-dependent actin filament function in the context of anti-cancer drug development.

Tropomyosins (Tpms) form end-to-end polymers along actin filaments and determine the functional properties of the filament in fungi 1 , flies 2 and mammals 3 . They belong to a highly conserved family of proteins with the greatest sequence divergence occurring at the N-and C-terminal ends due to alternative promotor use and exon splicing 4 . The N-and C-termini of adjacent Tpm molecules form an overlap complex that is required for Tpm to form cables along both sides of the helical actin filament 5 . It is not clear how the isoform-specific sequence information contained within the overlap complex contributes to differences in the way Tpms bind to and regulate actin.
Functionally distinct actin filament populations, characterised by their Tpm isoform composition, directly regulate a wide range of physiological processes in mammals 6 . In malignancy the Tpm profile is significantly altered, concomitant with dramatic rearrangements in actin cytoskeleton architecture 7 . Despite a down-regulation in high-molecular weight Tpm isoforms, actin filaments incorporating the low molecular weight isoform Tpm3.1 persist in all malignant cell types and are required for tumour cell survival in, at least, melanoma and neuroblastoma 8,9 . Studies implicating Tpm3.1-containing actin filaments in focal adhesion stability 10 , ERK mediated proliferation 11 and myosin-dependent mechanical tension 12 may speak to the specific reliance on Tpm3.1 in malignancy. How Tpm3.1 achieves these isoform-specific functions at the molecular level remains unknown.
We reported the preferential targeting of Tpm3.1-containing actin filaments by the small molecule TR100 in vivo 9 . We found that the actin filament disrupting effect of TR100 was shown to specifically depend on the C terminus of Tpm3.1, encoded by exon 9d, an alternatively spliced exon. Cells transfected with a Tpm3.1 chimera containing an alternative striated muscle-specific C terminus (exon 9a), demonstrated significant resistance to the effects of TR100. In this report we use TR100 as a molecular tool to dissect-out the contribution of the Tpm overlap region to actin filament regulation. We find that TR100 perturbs Tpm3.1-regulated actin filament depolymerisation from the barbed end without affecting the capacity of Tpm3.1 to regulate barbed end actin elongation or bind F-actin. The distinction made between actin disassembly and actin binding, using Tpm3.1 as proof-of-principle, may be intrinsic to all Tpm isoforms and provide an explanation for the functional significance of sequence variation in the overlap region.

Results
Both muscle and non-muscle Tpm isoforms have been characterised for their ability to regulate actin filament kinetics [13][14][15] . It has been suggested that inhibition of actin polymerisation by Tpm may be a direct result of changing the barbed end elongation rate 14,15 or the ability of Tpm to promote end-to-end annealing of actin filaments, effectively reducing the number of barbed ends primed for actin monomer addition 15,16 . Here we find that saturating concentrations of Tpm3.1 slow the incorporation of pyrene-labelled actin monomers into the growing filament when compared to the spontaneous polymerisation of actin alone (Fig. 1a,b). The capacity of Tpm3.1 to regulate actin polymerisation was investigated in the presence of TR100. Pre-incubation of Tpm3.1 with TR100 prior to mixing with G-actin and initiating polymerisation did not affect its ability to reduce actin polymerisation (Fig. 1a). The initial rate of actin polymerisation was approximately half the rate observed for actin alone both in the presence and in the absence of TR100 (Fig. 1b).
The critical concentration of actin assembly is the concentration of monomeric actin at steady state when the rates of polymerisation and depolymerisation are equal such that there is no net filament growth. It is determined by rates of binding and dissociation of ATP-actin at the filament ends. As observed for other Tpm isoforms 17,18 , the critical concentration (~0.1 μ M for the barbed end) was not affected in the presence of saturating concentrations of Tpm3.1 (Fig. 1c) and TR100 did not affect critical concentration when pre-incubated with Tpm3.1 prior to binding to actin filaments (Fig. 1c).
In addition to actin elongation kinetics Tpms have also been shown to inhibit actin filament depolymerisation [19][20][21] and this property was recently extended to Tpm3.1 9,22 . We previously reported that the stabilising effect of Tpm3.1 on actin filament depolymerisation is negated by pre-incubating Tpm3.1 with TR100 prior to filament saturation 9 . We questioned whether this effect would manifest if the actin filament was initially saturated with Tpm3.1 prior to drug exposure. To address this we compared the effect of TR100 on actin filament stability preand post-saturation with Tpm3.1 (Fig. 2). As demonstrated previously, dilution-induced depolymerisation of F-actin, which is dominated by dissociation of ADP-actin from the barbed end of the filament, was significantly slower for filaments coated with Tpm3.1 (DMSO control, Fig. 2a,d). Pre-incubating Tpm3.1 with 50 μ M TR100 prior to polymer formation largely negated the effect of Tpm3.1 on actin depolymerisation (Fig. 2b) and the initial depolymerisation rate (Fig. 2c). As a consequence, Tpm3.1/actin filaments depolymerised to a similar extent as bare actin filaments. This effect on depolymerisation however, did not manifest when F-actin filaments, fully coated with Tpm3.1 and exposed to 50 μ M TR100 for 10 minutes, were diluted into F-actin buffer to initiate depolymerisation ( Fig. 2d-f). This suggests that TR100 binds to Tpm free in solution and that the drug binding site is inaccessible once Tpm has polymerised on the actin filament. Furthermore, these data suggests that the functionality of the head-to-tail overlap complex makes a significant contribution to the actin-filament stabilising property of Tpm.
It is possible that the impact of TR100 on Tpm-regulated actin filament kinetics reflects a change in the binding capacity of Tpm3.1 for actin. To investigate this the affinity of Tpm3.1 for F-actin was measured by co-sedimentation. Increasing concentrations of Tpm3.1 were mixed with a constant amount of F-actin and pelleted by high speed centrifugation to separate out the actin-bound (pellet) and unbound (supernatant) fractions of Tpm3.1 (Fig. 3a). As expected, Tpm3.1 bound to skeletal F-actin with positive cooperativity as indicated by the sigmoidal shape of the binding-curve (Fig. 3b). Of note, we predetermined that Tpm3.1 binds to non-muscle beta F-actin with a similar binding affinity and cooperativity (alpha-actin k app 0.46 ± 0.11, α H 1.8 ± 0.6 vs. beta-actin k app 0.48 ± 0.80, α H 2.5 ± 1.0, data not shown). Surprisingly, at an identical concentration of TR100 used to elicit effects on Tpm3.1-regulated actin depolymerisation (50 μ M), we did not observe any effect on the affinity of Tpm3.1 for F-actin or the cooperativity of binding (Fig. 3a,b). This indicates, that despite changes to Tpm3.1-regulated actin filament kinetics in the presence of TR100, Tpm3.1 does not lose the capacity to bind at the actin interface in a cell-free assay.
To corroborate these findings in vivo, FRAP analysis was employed to measure the recovery kinetics of tagged Tpm3.1 following bleaching. Mouse embryonic fibroblasts (MEFs) transfected with Tpm3.1-mNeonGreen were treated with 25 μ M TR100 for 1 hour prior to FRAP analysis. At this concentration and treatment time an obvious change in cell morphology was observed (Fig. 3c). Specifically cells became less spread and there was a reduction in the appearance of large actin bundles with TR100 treatment. Time-lapse recordings of Tpm3.1-containing stress fibers following photobleaching show that TR100 does not affect the exchange of Tpm3.1 into filaments (Fig. 3d) or the recovery kinetics (Fig. 3e). Together, the co-sedimentation and FRAP data indicate that Tpm3.1 binds to actin equally well in the presence and absence of TR100.

Discussion
We propose that TR100 acts to compromise the integrity of Tpm cables rather than prevent overlap complex formation. Our data suggests that TR100 is incorporated into the growing actin-Tpm co-polymer given that its effects cannot be observed on pre-formed Tpm3.1/actin filaments. Certainly, Tpm3.1 can form a continuous polymer with actin in the presence of TR100 which must involve both Tpm3.1-actin binding and Tpm3.1 head-to-tail cooperative binding 23 . These results therefore dissociate the ability of Tpm3.1 to bind along an actin filament from its ability to regulate actin filament stability.
The C terminus of Tpm is helical and a coiled coil but contains a hinge near the end to enable the helical ends to splay apart and form the overlap complex with the coiled coil N terminus. The C terminus must be flexible in order to interact with the N terminus 24 . Upon formation of the overlap complex both ends are stabilised, though the overlap complex remains dynamic 25,26 . Therefore, we propose a mechanism of action in which TR100 binds to the uncomplexed C terminus of Tpm3.1 in a conformation permissible for N-terminal binding. It is a formal possibility that the presence of C-terminal bound TR100 introduces steric hindrance in the overlap complex leading to reduced flexibility in this region. In both striated and smooth muscle isoforms the overlap domain is characterised by a degree of flexibility 25,[27][28][29] which is likely a governing factor in how information is communicated along the Tpm polymer as well as between the Tpm polymer and the actin filament. Finally, given that the binding capacity of Tpm3.1 for actin is unaffected by TR100, exactly how actin dynamics is altered remains a subject of intense interest. One possibility is the existence of different conformational states of actin induced by Tpm binding 23 . Due to the highly cooperative nature of the actin polymerisation/depolymerisation process, small conformational changes to the actin filament would likely result in dramatic changes to kinetic assembly and disassembly.
Unlike striated muscle Tpm which requires N-terminal acetylation to associate end-to-end and for cooperative binding to filamentous actin 30 , bacterially expressed non-muscle Tpms are capable of self-association and actin binding in the absence of acetylation [31][32][33] . Given that N-terminal acetylation has been shown to stabilise the alpha-helical conformation of an N-terminal striated muscle peptide, albeit with little effect on the overall stability of full length Tpm 34 , it may still be important to investigate whether this conformational change extends to non-muscle Tpm isoforms and the potential implications for cooperative binding to actin. It should be noted however, that in the fission yeast model, both acetylated and non-acetylated subpopulations of a single Tpm isoform (Cdc8) exist and are associated with distinct actomyosin pools in the cell 35 . It is therefore possible that acetylation status serves a similar purpose amongst mammalian Tpm isoforms, making it important to compare the effects of anti-Tpm compounds on both modified and non-modified forms of Tpm.
Concomitant with disrupted actin filament integrity in vivo, TR100 demonstrates strong cytotoxicity against neuroblastoma and melanoma cell lines grown in 2D and 3D culture 9 . Interestingly, the dose-response curves for TR100 are characterised by a steep slope, indicating a narrow concentration range over which TR100 acts to reduce cell viability. This threshold response may correspond to loading a sufficient amount of TR100 into the filament, which correlates with our current finding that TR100 is gradually incorporated into the polymerising actin-Tpm3.1 filament. This would suggest that the threshold is based on two states of the filament; functional versus non-functional, rather than TR100 having a graded effect on filament function.
Tpm3.1 localises strongly to stress fibers and both nascent and mature focal adhesions 36,37 , however a clear consensus on the preferential association of Tpm3.1 for beta versus gamma-actin does not exist in an absolute sense. The association of Tpm3.1 with either actin isoform appears to be cell type specific 38,39 . Nevertheless, the demonstration that beta and gamma actin have intrinsic functional differences in their ability to activate different myosin II motors opens the possibility that Tpm3.1-containing filaments composed of different actins may respond differently to TR100 in different cellular contexts 40 .

Methods
Reagents. TR100 was designed by T.A. Hill and A. McCluskey as described previously 9 . TR100 was synthesised by SynMedChem and prepared as a 50 mmol/L stock in DMSO.
Protein purification. Rabbit skeletal muscle actin was sourced commercially (Cytoskeleton Inc.), with the exception of actin used in the spontaneous actin polymerisation and critical concentration assay which was prepared and labelled with pyrene as described previously 41 .
The human homologue of Tpm3.1 was expressed in E. coli BL21 (DE3) cells and purified according to Actin/Tpm co-sedimentation assay. The affinity of Tpm for F-actin was measured by co-sedimentation according to Heald and Hitchcock 42 , with modifications. Prior to co-sedimentation, Tpm3.1 was pre-incubated with 50 μ M TR100 compound or 1% (v/v) DMSO vehicle control for 10 min. Increasing concentrations of Tpm (0.2-5.5 μ M) were mixed with 3 μ M rabbit skeletal F-actin in buffer containing 150 mM NaCl, 10 mM Tris-HCl pH 7.0, 2 mM MgCl 2 and 0.5 mM DTT. Samples were incubated for 20 min at room temperature before pelleting at 50,000 rpm for 30 min at 20 °C. Supernatant and pellet were resolved on a 12.5% polyacrylamide gel and proteins visualised with Coomassie Blue. Gels were scanned using a LAS4000 imager and densitometry was performed in Image J (Image J software, NIH). The ratio of Tpm to actin in the pellet was normalised to 1.0 by dividing the Tpm to actin ratio obtained from densitometry by the observed maximal ratio at saturation. The concentration of free Tpm in the supernatant was calculated from a standard curve of Tpm run in parallel with the samples. The Tpm to actin ratio was plotted against the concentration of free Tpm. The binding constant (K app ) of Tpm for actin and the Hill coefficient (α H ) were determined by fitting the experimental data to a non-linear regression model (4 parameter, variable slope) in GraphPad Prism 5.0.
Fluorescence recovery after photobleaching. To create the Tpm3.1-mNeonGreen construct, human Tpm3.1 cDNA (NCBI mRNA accession# NM_153649) followed by the sequence encoding mNeonGreen at the C-terminus (a gift from Jiwu Wang), separated by a linker motif (GGGGSGGGGS), were subcloned into pcDNA3.1 containing a CMV promotor (GeneArt, Invitrogen). Primary MEFs were maintained in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% foetal bovine serum (FBS) at 37 °C, 5% CO 2 . Transfection with Tpm3.1-mNeonGreen construct was performed using Lipofectamine 3000 reagent (Life Technologies) according to manufacturer's instructions. Tpm3.1-mNeonGreen expressing MEFs were used for imaging on days 2 and 3 post-transfection. Live cell imaging of transfected MEFs was performed on a Nikon A1 inverted scanning confocal microscope equipped with NIS Elements software, incubation chamber (37 °C) and CO 2 control. Photobleaching was achieved with a single 120.38 ms pulse using a 488-nm laser, focused using a Nikon Plan Apochromat λ 60×/1.4 (oil) objective. For FRAP analyses, a circular region of interest (5 μ M diameter), containing mostly stress fibers, was selected for photobleaching per cell. Pre-bleached scans (3-5 per cell) were taken to determine initial fluorescence intensity. Following a single bleach pulse images were acquired every 1 s for 2 min. To obtain control FRAP curves, select zones containing stress fibers were first photobleached prior to the addition of TR100. Cells were then incubated with 25 μ m TR100 for 1 h and the same cell was photobleached at an alternate target zone containing stress fibers. FRAP curves generated were normalised to minimum and maximum fluorescence values. Data from normalised FRAP curves were processed to obtain standard error measurement values for each respective time point.