Visualizing the in vitro assembly of tropomyosin/actin filaments using TIRF microscopy

Abstract

Tropomyosins are elongated alpha-helical proteins that form co-polymers with most actin filaments within a cell and play important roles in the structural and functional diversification of the actin cytoskeleton. How the assembly of tropomyosins along an actin filament is regulated and the kinetics of tropomyosin association with an actin filament is yet to be fully determined. A recent series of publications have used total internal reflection fluorescence (TIRF) microscopy in combination with advanced surface and protein chemistry to visualise the molecular assembly of actin/tropomyosin filaments in vitro. Here, we review the use of the in vitro TIRF assay in the determination of kinetic data on tropomyosin filament assembly. This sophisticated approach has enabled generation of real-time single-molecule data to fill the gap between in vitro bulk assays and in vivo assays of tropomyosin function. The in vitro TIRF assays provide a new foundation for future studies involving multiple actin-binding proteins that will more accurately reflect the physiological protein-protein interactions in cells.

Introduction

At present, it is not possible to reliably and consistently image single filaments of actin within a living system, due to issues with resolution, labelling and space packing within the cytoplasm. In order to obtain precise kinetic data on single actin filaments, in vitro analysis allows for a consistent tunable system, in which multiple parameters can be defined and measured. Over the past several years, in vitro TIRF microscopy has proven to be an optimal tool for the observation and characterisation of actin dynamics and its interactions with actin-binding proteins (ABPs), from a single filament/molecule to complex multi-component reconstitutions (for review see Mullins and Hansen 2013). This review reflects on several recently published studies using in vitro TIRF microscopy (Bareja et al. 2020; Christensen et al. 2017; Christensen et al. 2019; Gateva et al. 2017; Hsiao et al. 2015; Janco et al. 2018; Nicovich et al. 2016; Palani et al. 2019; Schmidt et al. 2015) that reveal for the first-time complex interactions of various tropomyosin (Tpm) isoforms on a single actin filament (or in addition with other actin-binding proteins) in detail. Newly acquired biochemical data of Tpm in vitro including real-time elongation rates of Tpm on a single actin filament can now be used to build parameterised models (Bareja et al. 2020; Christensen et al. 2017), emphasising the advantages of the TIRF microscopy approach compared with previous biochemical bulk assays. Finally, tropomyosin fluorescent labelling challenges, methods and recommendations for future studies are discussed.

Biochemical in vitro studies with tropomyosin and other actin-binding proteins

Tropomyosins (Tpms) are a family of ABPs, which play an important role in regulating the plethora of cellular functions in both muscle and non-muscle cells. Tpms have also been shown to be involved in cardiac and skeletal muscle diseases (Thierfelder et al. 1994) as well as in the growth and spread of cancer (for reviews see Gunning et al. 2008; Hardeman et al. 2020). After the discovery of Tpm, in vitro studies demonstrated it to be the major regulator of actin dynamics via allosteric conformational changes within the actin filament (F-actin). Tpm forms a head-to-tail polymer alongside F-actin providing structural stabilisation of the actin cytoskeleton, which has been historically regarded as its main function (Khaitlina 2015). However, isoform-specific interactions of Tpms with other ABPs, which include the myosin family, ADF/cofilin, profilin, gelsolin, thymosin beta4, DNase I, CapZ, tropomodulin and Arp2/3 (for review see dos Remedios et al. 2003), underpin the diversity of Tpm functions (for review see Hardeman et al. 2020).

The stabilising role of Tpm on F-actin was suggested from experiments showing that it slows down depolymerisation by lowering the off-rate constant. Using fluorescence assays, the depolymerisation rate in the presence of Tpm averaged 56% of the control rate, while not affecting elongation at the pointed end (Broschat et al. 1989). The degree of this stabilising effect was dependent on the Tpm concentration and filament length, where the longer F-actin (~ 150 monomers) was more effectively protected from depolymerisation than shorter filaments (~ 50 monomers). Nevertheless, Tpm was not capping F-actin as shown by the elongation assays. This indicates that Tpm can bind to the side of the filament and prevent dissociation of monomers from the pointed end without capping it, thus regulating the length of filaments (Broschat 1990). Further evidence of Tpms stabilising role came from experiments with polymerisation-defective yeast actin (GG). It was shown that polymerisation of GG-actin was enabled in the presence of Tpm with the efficiency inversely proportional to Tpm length, while Tpm did not prevent cold-induced depolymerisation (Wen et al. 2000). Tpm also limited the extent and the rate of GG-actin ATPase activity, indicating that Tpm restored polymerisation and stabilised F-actin oligomers, rather than enhancing filament nucleation (Yao and Rubenstein 2001). Additionally, it was suggested that Tpm regulated allosteric strengthening of intermonomer contacts in F-actin (Khaitlina et al. 2017), and more importantly, a recent study showed that Tpm isoforms were involved in the regulation of cofilin activity and induced conformational changes at the longitudinal and lateral actin filament interfaces in an isoform-specific manner (Ostrowska-Podhorodecka et al. 2020).

Evidence of interaction between Tpm and other ABPs impacts the regulation of actin organisation, function and dynamics. Historically, the first studies of Tpm and its function involved the mechanism and the regulation of skeletal muscle (Huxley 1971; Parry and Squire 1973; Spudich et al. 1972; Spudich and Watt 1971), which evolved into a large field of research with focus on the myosin family using various in vitro techniques including TIRF microscopy. For example, it has been shown that the yeast (Saccharomyces cerevisiae) Tpm isoform, Tpm1p, is essential for processivity of the myosin Myo2p (Hodges et al. 2012), whereas the Tpm isoform Tpm1.6 inhibits myosin-Ic and plays an important role in initiation and termination of kinesin-1-driven transport (McIntosh et al. 2015). Additionally, Tpm isoforms (Tpm1.8, Tpm1.12 and Tpm3.1) also have different regulatory effects on non-muscle myosin-2B and F-actin as demonstrated by changes in the frequency, duration, and efficiency of actomyosin interactions (Pathan-Chhatbar et al. 2018). It is therefore not surprising that Tpm binding to actin precedes that of myosin IIA binding. Intravital subcellular microscopy of secretory granule exocytosis revealed that during de novo actin filament formation in live mice, the tropomyosin isoform Tpm3.1 assembles on growing actin filaments, while the recruitment of the myosin IIA follows later (Masedunskas et al. 2018).

Tpms can regulate actin dynamics at the barbed end by activating the binding of actin nucleators (formins) while not affecting the binding of the filament capping protein CapZ (Wawro et al. 2007). F-actin severing proteins such as ADF/cofilin are responsible for the high turnover rates of actin filaments in vivo and bind to F-actin cooperatively inducing a twist in the actin filament, while Tpm competes with ADF/cofilin for binding to actin filaments (Bamburg 1999). Tpm can also interact with the F-actin severing protein gelsolin. It was shown that gelsolin can bind to Tpm via domain 2 (Maciver et al. 2000). While other ABPs were shown to cap the fast growing (barbed) end of F-actin, tropomodulin caps the pointed end. It binds to Tpm and blocks both elongation and depolymerisation at the pointed ends of Tpm-containing F-actin in concentrations equivalent to the concentration of filament ends (Kd ~ 1 nM). In the absence of Tpm, the inhibiting power of tropomodulin is about three order of magnitude lower (Kd ~ 0.1–0.4 μM) (Weber et al. 1994).

Binding of Tpm to F-actin increases the flexural rigidity (stiffness) of actin filaments from 43.7 to 65.3 pN/nm (Kojima et al. 1994). Addition of Tpm renders F-actin filaments more rigid (Lp = 20 μm) in the absence of Ca²⁺ and more flexible (Lp = 12 μm) in the presence of Ca²⁺, suggesting a Ca²⁺-dependent interaction between actin and Tpm (Isambert et al. 1995). Two specific cation binding sites were later identified on actin (situated between adjacent subunits along the long-pitch helix) and classified as “the stiffness site” (dictates bending rigidity of the filament) and “the polymerisation site” (drives actin polymerisation) (Kang et al. 2012).

More recent in vitro studies have extended the details of Tpm behaviour, including the activity and functional specificity of predominantly cytoskeletal Tpm isoforms. It was found that the affinity for F-actin varied between Tpm isoforms (Kd 0.1–5 μM) with similar values for actin filaments formed from skeletal compared with cytoskeletal actin (Janco et al. 2016). Comparison of various dimeric Tpm species revealed that heterodimers of skeletal Tpm have a different F-actin affinity and thermal stability compared with homodimers, while their ability to regulate myosin binding is unaffected (Kalyva et al. 2012). Furthermore, utilizing multicolour in vitro TIRF microscopy, it has been demonstrated that cytoskeletal Tpm isoforms (Tpm1.6, Tpm1.7, Tpm2.1, Tpm3.1 and Tpm4.2) in combination with the set of other ABPs (cofilin, coronin and AIP1) have quantitively distinct abilities to regulate actin filament length and turnover (Jansen and Goode 2019).

In vitro TIRF microscopy of actin-tropomyosin filaments—a comparison of the impact of Tpm labelling and actin-binding surface chemistry

Real-time in vitro TIRF microscopy has been used to study actin filament kinetics and its interactions with actin-binding proteins for many years (Breitsprecher et al. 2009; Kuhn and Pollard 2005). The historical employment of TIRF microscopy using techniques such as the in vitro motility assay and optical tweezer assays (Fanning et al. 1994; Homsher et al. 1996; Kad et al. 2005) contributed towards understanding Tpm activity and cooperativity. However, the first study visualizing Tpm molecules directly labelled with fluorescent tags to investigate interactions with single actin filaments using the in vitro TIRF approach was published only recently (Schmidt et al. 2015).

This delay was caused by the challenges of finding an appropriate methodology to fluorescently label tropomyosin molecules without disrupting binding to actin, and in optimising surface chemistry for actin filament attachment to the surface. The initial study showing tropomyosin interactions with actin filaments (Schmidt et al. 2015) used rabbit skeletal and chicken gizzard smooth muscle Tpms. The Tpms were labelled via maleimide conjugate dyes AlexaFluor-532 and Cy3 at the single native cysteine in these isoforms, Cys-190. The nitrocellulose-coated surface of the flow cell was modified by full-length chicken skeletal muscle myosin treated with N-ethylmalaeimide (the treatment chemically locks myosin into the strong actin-binding state), and bovine serum albumin (BSA) was used as a crowding agent to prevent non-specific binding to the surface (Fig. 1a). This experimental approach measured the binding rates and the effect of salt concentration on Tpm binding to actin and enabled confirmation of previous predictions about the stochastic nature of tropomyosin binding to actin filaments (Holmes and Lehman 2008; Vilfan 2001). The study also identified the events behind formation of the gap-free filament during tropomyosin association and initiation of dissociation from the filament. However, the low resolution of the data prevented observation of the binding events on the opposite sides of the filament, and the use of myosin as a surface linker could have affected the accessibility of actin-binding sites.

Fig. 1.

Schematic representation of the surface chemistry modalities used to anchor actin filaments in tropomyosin studies using TIRF microscopy. a An actin filament attached to the surface by a full-length MyosinII protein. The coverslip surface is modified by bovine serum albumin (BSA) as a crowding agent to prevent non-specific surface binding. b Attachment of the filament to a surface modified with PLL-PEG-biotin and streptavidin via multiple points using biotinylated phalloidin. c Filament grown from the pointed end of a biotinylated spectrin–actin seed and aligned to the surface by the flow inside the microfluidic device. Surface is modified with PLL-PEG-biotin and the spectrin–actin seed is attached to the surface via streptavidin. d An actin filament grown from the barbed end via binding to gelsolin and aligned to the surface by the flow. The surface is modified by the same components as shown in the previous example using a spectrin–actin seed

Two different approaches to actin filament surface attachment and their effects on Tpm binding were evaluated by the Boecking lab (Nicovich et al. 2016). Filaments were grown in the channel of a microfluidics device and attached to the surface via either poly-L-lysine and biotinylated poly(ethylene glycol) (PLL-PEG as a crowding agent) followed by streptavidin and biotinylated phalloidin enabling actin filament attachment at multiple points (Fig. 1b), or by PLL-PEG, streptavidin and biotinylated spectrin–actin seeds for single filament tethering (Fig. 1c). Filaments were grown in situ at the barbed ends of the spectrin–actin seeds and aligned by the flow parallel to the surface (and to each other) as reported previously (Jegou et al. 2011). Skeletal muscle Tpm (Tpm1.1) was labelled by Cy5-maleimide conjugated AlexaFluor-647 dye to Cys-190 with a final degree of labelling of 20%. This approach yielded a majority (81%) of singly labelled tropomyosin dimers binding to actin filaments with a similar affinity to the native Tpm with resolution sufficient to detect binding on both sides of the actin filament. The two attachment modalities showed similar interactions of Tpm with actin (k_obs 0.152 and 0.196 s⁻¹ for fixed and aligned filaments, respectively; and k_off 0.019 and 0.012 s⁻¹ for fixed and aligned filaments, respectively). The full data set of the aligned filaments was later applied to more sophisticated semi-automated MatLab-based script for detailed analysis, allowing measurement of the individual directional rates (towards the barbed or pointed end of the filament) of Tpm association and dissociation to/from actin (Janco et al. 2018).

Another significant challenge in unravelling finely tuned interactions between filamentous actin and Tpm using the in vitro TIRF technique is the difference between individual Tpm isoforms and how they functionally specify actin filament function. This complex riddle was solved in the study by the Lappalainen lab (Gateva et al. 2017), where six high- and low-molecular weight non-muscle Tpm isoforms (Tpm1.6, Tpm1.7, Tpm2.1, Tpm3.1, Tpm3.2 and Tpm4.2) were N-terminally labelled with sfGFP and mCherry and bound to actin filaments in vitro. The use of large fluorescent molecules like sfGFP or mCherry instead of small chemical dyes covalently attached to cysteine was effective because it overcame the problem of the variable number (four cysteines in Tpm3.1 monomer) or a complete lack of cysteines (Tpm1.7) in different Tpm isoforms. It is important to mention that the N-terminal labelling by sfGFP or mCherry most likely affects Tpm end-to-end interactions as reported previously by viscosity measurements of Schizosaccharomyces pombe Tpm^Cdc8 (Brooker et al. 2016), and due to the size of these fluorescent tags the availability of binding sites on the filament for other ABPs may be reduced. The TIRF-based assays in Gateva et al. (2017) revealed that most Tpm isoforms cannot co-localize on a single actin filament (with the exception of Tpm2.1, which co-polymerises with Tpm3.1 and Tpm3.2) providing experimental evidence of a segregation to different actin filaments decorated by a specific isoforms. This study further examined if distinct actin-tropomyosin filaments display functional differences, employing the ABPs cofilin and myosin II. The results showed that cofilin competed with all tested Tpm isoforms for binding to F-actin, associating cooperatively along actin filaments towards the pointed end and simultaneously replacing Tpm. Additionally, pyrene actin–based depolymerisation assays showed that the high molecular weight isoforms Tpm1.6 and Tpm1.7 were able to protect actin filaments from cofilin-mediated disassembly. The effect of different actin/Tpm filaments on myosin II was measured by ATPase activity assay of the non-muscle myosin IIa-HMM rates at steady-state. The results showed an increased actin-activated ATPase activity for low-molecular weight Tpm isoforms Tpm3.1, Tpm3.2 and Tpm4.2, while the high molecular Tpm isoforms Tpm1.6, Tpm1.7 and Tpm2.1 did not show a significant effect on myosin IIA activity (Gateva et al. 2017).

The first study that implemented in vitro TIRF microscopy to elucidate binding of Tpm to branched actin networks regulated by the ABPs cofilin and the Arp2/3 complex was published by the Mullins lab and used in vitro TIRF microscopy to describe the complex cellular processes (Hsiao et al. 2015). In this study, the authors used non-muscle Drosophilla Tpm Tm1A, labelled via maleimide conjugate dyes Cy3 and Cy5 to Cys-82 (endogenous Cys-32 was replaced with alanine and Ser-82 was replaced with cysteine in order to keep the label least disruptive to Tpm function). The coverslip surface for actin filament attachment was modified by PEG-biotin, followed by streptavidin and biotin-phalloidin. The study showed that Tm1A binds preferentially near the pointed end of severed actin filaments composed of ADP-actin and that Tm1A binding to the pointed end was promoted by the actin severing protein cofilin. In contrast, Tm1A does not bind to the dendritic actin network generated in vitro by the Arp2/3 complex, as the Arp2/3 complex blocks the pointed end of the filaments. This interplay between the Arp2/3 complex, cofilin and Tm1A on actin reveals the binding of Tpm to severed actin filaments but not to the branched network.

The Kovar lab used the fission yeast tropomyosin Cdc8, bundling protein fimbrin Fim1, and severing protein ADF/cofilin Adf1 to characterise binding of Cdc8 to actin filaments and to describe the interplay between these three proteins in vitro (Christensen et al. 2017). Firstly, three distinct Cdc8 mutants were created, to introduce engineered cysteine single mutations for maleimide labelling and the mutant Cdc8(I76C) labelled with Cy5 probe that exhibited functional properties closest to the wild-type was selected for the study. Yeast tropomyosin Cdc8 showed stochastic association to the actin filament (elongation rates towards the barbed and pointed ends were 3.2 and 2.4 molecules sec⁻¹ μM⁻¹, respectively, p value = 0.037 s⁻¹) with a high degree of cooperativity. High-data quality allowed the authors to create a model of Tpm association and elongation on actin using the real-time experimental data. Further findings of the study showed that co-incubation of Cdc8 with Fim1 triggered Fim1-mediated bundling displacement of Cdc8 from actin filaments. Cdc8 was also shown to inhibit the initial binding of ADF/cofilin Adf1 to filaments; however, with prolonged incubation time, Adf1 associated and eventually displaced short segments of Cdc8. Fim1 and Adf1 co-localize at both single actin filaments and multi-filament bundles, demonstrating synergistic effects and creating extremely large and dense actin bundles. The synergistic effect of Fim1 and Adf1 allow them to displace Cdc8 from actin filaments. A subsequent study from the Kovar lab (Christensen et al. 2019) used the same experimental settings and showed that fission yeast α-actinin (Ain1) in combination with tropomyosin Cdc8 inhibits binding of Fim1 to actin filaments.

The most recent study utilizing in vitro TIRF microscopy to understand actin-tropomyosin interactions describes the dynamics of non-muscle tropomyosin isoform Tpm1.8 on actin filaments with single molecule resolution (Bareja et al. 2020). Tpm1.8 was N-terminally labelled with the fluorescent proteins mNeonGreen and mRubyII and the actin filament attachment was selected by a single tethering point from either barbed end (via biotinylated spectrin–actin seed; Fig. 1c) or pointed end (biotinylated gelsolin; Fig. 1d) and stretched by the flow in a microfluidic device as described previously (Jegou et al. 2011; Nicovich et al. 2016). The study shows for the first time asymmetric elongation rates of Tpm on an actin filament (1.19 m·M⁻¹·s⁻¹ and 0.64 m·M⁻¹·s⁻¹ towards the barbed end and the pointed end, respectively) as well as asymmetric shrinkage rates from the barbed end (1.98 × 10⁻⁹ m s⁻¹) compared with the pointed end (1.07 × 10⁻⁹ m s⁻¹). This study also reports the discovery of the independent decoration of both actin filament strands by Tpm1.8 and single-molecule kinetics data suggesting that Tpm1.8 binds, elongates and dissociates from actin filaments as a single dimer resulting in the generation of a fully experimentally parameterized stochastic model of Tpm1.8 dynamics on an actin filament.

It has been shown that tropomyosin isoforms associate with the majority of actin filaments in non-muscle cells (Meiring et al. 2018); therefore, in future studies, the characterisation of interactions between actin and ABPs should preferentially be performed in the presence of Tpm isoforms.

Different fluorescent labelling approaches on the functional properties of tropomyosin in vitro

The major challenge of single molecule or single filament in vitro assays in general (in this case mainly from the actin-tropomyosin point of view) is to reconstitute experimental conditions as close as possible to those in vivo, in order to acquire physiologically meaningful data. Tpms are highly conserved proteins (Vrhovski et al. 2008) where the slightest changes in the structure of the protein, such as a single point mutation, might cause devastating functional consequences in vivo (Gupte et al. 2015; Harvey and Leinwand 2011). Therefore, modification of the Tpm molecule for in vitro studies, such as selection and design of fluorescent tags for TIRF microscopy, must be carefully considered and tested.

The most common fluorescent labelling approach of tropomyosin shown in in vitro–based TIRF microscopy studies to date is the use of the maleimide conjugated small chemical dyes (Christensen et al. 2017; Hsiao et al. 2015; Janco et al. 2018; Nicovich et al. 2016; Palani et al. 2019; Schmidt et al. 2015). In order to reach the full potential of the experiment and acquire quality data with suitable resolution, this labelling approach requires the following considerations: (i) it is suitable for Tpm isoforms containing only one cysteine; (ii) cysteine 190 is present in most of the Tpm isoforms containing only one cysteine; however, it has been predicted to be at the position a in the heptad (Brown et al. 2001). Therefore, prior to the labelling, Tpm dimers must be reduced to the monomeric form. The most effective method of generating Tpm monomers is to use a combination of reducing agents (DTT, TCEP), heat (up to ~ 60 °C for Tpm1.1 (Janco et al. 2012) and ~ 45 °C for fission yeast Tpm^Cdc8 (Brooker et al. 2016) according to circular dichroism melting curves for the full unfolding) and high salt concentration (500 mM KCl or NaCl); (iii) the labelling reaction of maleimide with thiols is relatively fast; therefore, prolonged incubation with 5–10x excess of dye is not necessary; (iv) a lower degree of labelling (ideally 20%) is preferential for higher resolution and increased functional properties of Tpm. Additionally, thiol–maleimide reactions are highly specific at pH 6.5–7.0, whereas a higher pH of 8.0 and above might cause an undesired reaction with ε-amines; therefore, buffer conditions should be preferentially at a neutral pH.

Fluorescent fusion proteins are a useful method of visualizing individual Tpm isoforms (Bareja et al. 2020; Gateva et al. 2017). Fluorescent proteins should be placed at the N-terminus of the Tpm protein, as C-terminus labelling with large fluorescent proteins causes a total loss of Tpm binding to actin filaments in vitro (Brooker et al. 2016). The linker between the fluorescent protein and Tpm molecule is also crucial for Tpm binding, and therefore it is recommended to use examples from previously published work or test to ensure that the linker does not interfere with Tpm binding (Bareja et al. 2020; Gateva et al. 2017).

Additionally, there is potential for using other labelling techniques such as SNAP/CLIP tags (selection of numerous fluorescent dyes, covalent interaction to the same substrate sequence) (Bosch et al. 2014) or introducing 6x or 10x His tags to the N-terminal of the Tpm molecule and labelling these relatively small tags via tris-NTA conjugate dyes (Lata et al. 2005). Non-specific binding of Tpm via the NHS ester (or succinimidyl ester) fluorescent dyes is not recommended, as this type of chemistry targets the primary amines (R-NH2) of proteins such as ɛ-amino group of lysine and prevents the Tpm interaction with actin filaments (unpublished data). Additionally there are over 20 lysines available (positions b, c and f of the Tpm heptad) on Tpm1.1 (Brown et al. 2001); therefore, molecules labelled to such a degree would not be suitable for single-molecule analysis. For example, characterisation of the individual binding steps of Tpm requires a similar intensity of every molecule. However, with a heterogenous population of lysine-labelled Tpms (1 to 20 fluorescent tags per molecule in the case of Tpm1.1), the intensity of each single molecule binding to the filament would differ significantly.

In summary, these advances in the biochemical characterisation of Tpm functional assays using in vitro TIRF microscopy show that TIRF microscopy in conjunction with optimized surface and protein chemistry is an optimal tool to fill the gap between conventional in vitro bulk assays and cell based in vivo studies.

Authors’ contributions

All authors contributed to writing the review and in accordance with their place in the authorship list.

Funding information

This work was supported by an Australian Department of Industry, Innovation and Science Cooperative Research Centre Project (CRC-P) grant to P.W.G. and E.C.H. and grants from the Australian Research Council (ARC grant DP160101623), the Australian National Health and Medical Research Council (NHMRC grant APP1100202, APP1079866) and The Kid’s Cancer Project to P.W.G. and E.C.H.

Compliance with ethical standards

Conflict of interest

The authors declare they have no conflict of interest.

Ethics approval

This article does not contain any studies with human participants or animals performed by any of the authors.

Consent to participate

Not applicable.

Consent for publication

Not applicable.