Analysis of biophysical and functional consequences of tropomyosin–fluorescent protein fusions

The dynamic nature of actin polymers is modulated to facilitate a diverse range of cellular processes. These dynamic properties are determined by different isoforms of tropomyosin which are recruited to distinct subpopulations of actin polymers to differentially regulate their functional properties. This makes tropomyosin an attractive target for labelling discrete actin populations. We have assessed the effect of different fluorescent labelling strategies for this protein. Although tropomyosin–fluorescent fusions decorate actin in vivo, they are either nonfunctional or perturb regulation of actin nucleation and cell cycle timings. Thus, conclusions and physiological relevance should be carefully evaluated when using tropomyosin fusions.

The dynamic nature of actin polymers is modulated to facilitate a diverse range of cellular processes. These dynamic properties are determined by different isoforms of tropomyosin which are recruited to distinct subpopulations of actin polymers to differentially regulate their functional properties. This makes tropomyosin an attractive target for labelling discrete actin populations. We have assessed the effect of different fluorescent labelling strategies for this protein. Although tropomyosin-fluorescent fusions decorate actin in vivo, they are either nonfunctional or perturb regulation of actin nucleation and cell cycle timings. Thus, conclusions and physiological relevance should be carefully evaluated when using tropomyosin fusions.
Keywords: acetylation; actin cytoskeleton; Cdc8; fission yeast; Schizosaccharomyces pombe Many dynamic processes within the eukaryotes are driven by the actin cytoskeleton. Actin polymers can generate force to remodel membrane architecture as well as act as tracks for molecular motors. It is the dynamic nature of the actin polymers that is critical for their cellular function. These key properties are regulated by diverse actin-binding proteins, which drive nucleation, stabilisation, cross-linking or even depolymerisation of actin complexes [1].
One such class of actin-binding proteins are the tropomyosins (Tpm), a highly conserved class of dimeric a-helical coiled-coil proteins which associate with and stabilise actin [2]. Tpm dimers interact in an end-toend fashion to generate filaments that associate with and coil around the F-actin polymer. These Tpm cofilaments play a central role in stabilising actin filament dynamics in muscle cells and therefore regulate muscle contraction [3]. In nonmuscle cells, Tpm plays a critical role in stabilising and also defining the biophysical properties of a specific actin polymer as well as modulating its ability to interact with other proteins and thereby regulate its function [4].
Yeast and animal cells express multiple isoforms of Tpm, each of which associate with specific actin structures at a specific place and time, in an exquisitely orchestrated manner, to fine tune the biophysical nature of each actin polymer in a specific way [2]. This makes the different Tpms attractive targets as markers during the analysis of specific actin structures and functions. There is sequence diversity along the length of the different Tpm isoforms, but it is the ends of the protein that seem to be key in defining their function. The termini of the Tpm dimer not only define the structural and charge landscape at the end of the actin-Tpm copolymer but also determine the nature of the end to end contacts between adjacent dimers. This overlap region can vary significantly between different Tpm isoforms and is essential for defining the stability of the Tpm polymer and its ability to bind cooperatively to F-actin. Indeed, seemingly innocuous modifications to the charge on the polypeptide in these overlapping regions can have a profound effect upon the stability of the actin-Tpm complex. For example, in many cell types the addition of acetyl group to the amino-terminal residue of the Tpm stabilises the alpha-helical structure of each coiled-coil structure to facilitate its interaction with carboxyl region of adjacent Tpm molecules [5,6]. Thus, small changes in the charge and shape of the Tpm protein have a dramatic impact upon protein interactions and function.
The molecular-genetic plasticity of yeast model systems provides an excellent setting for studying the impact protein modifications have upon its cellular structure. Yeasts express a smaller number of Tpm isoforms in comparison to metazoan cells, which makes them an attractive organism in which to characterise specific Tpm functions. The fission yeast Schizosaccharomyces pombe has a single tropomyosin, Tpm Cdc8 , which exists in amino-terminally acetylated and nonacetylated forms within the cell [7,8]. Depending upon its acetylation status, Tpm Cdc8 associates with the actomyosin ring during mitosis to facilitate myosin II motor activity and cytokinesis, or with cytoplasmic actin filaments upon which myosin V motors move [9][10][11]. It thus provides a simple system for studying the regulation of function of this essential cytoskeletal protein.
Fluorescent Proteins (FP) provide an attractive method for following protein dynamics within a live cell [12]. FPs have been developed with a variety of spectral properties to facilitate diverse functional analysis. However, FP fusions can impact protein function by inhibiting normal protein folding or affecting interactions with other proteins. Not surprisingly both microtubule and actin cytoskeletons are acutely sensitive to fluorescent labelling of protein components [13]. Diverse FPlabelled markers exist to follow the dynamics of the actin cytoskeleton in a live cell context, however, many have subsequently been shown to alter the behaviour and organisation of the polymers within cells [14][15][16]. Tpms are attractive candidates for markers of specific actin structures and have been used to follow filament dynamics in diverse cell types, in a live cell context [17][18][19][20][21]. However, it is unclear how fusing a FP to the termini of the Tpm protein impacts its normal function.
Here, we describe in vitro and in vivo analyses of amino-and carboxyl-terminal fusions between the fission yeast tropomyosin, Tpm Cdc8 , and a monomeric fluorescent protein. We establish that while Tpm Cdc8carboxyl terminal fusions disrupted the ability of the protein to polymerise or associate with actin in vitro, the amino-terminal fusion formed filaments and associated with actin in a manner similar to amino-terminally acetylated endogenous Tpm Cdc8 . However, while this protein facilitated the formation of a functional contractile actomyosin ring (CAR) and cell growth in cells lacking functional endogenous Tpm Cdc8 , it disrupted the normal timing of cell division and normal myosin V movements. Thus, each Tpm-FP fusion has a significant impact upon the function of this critical cytoskeletal protein.

Cell culture
The yeast strains used in the study were h À leu1.32; h À cdc8.110 myo2.mCherry leu1.32 and h À cdc8.110 myo52.mNeongreen leu1.32. Cell culture and maintenance were carried out according to [24] using Edinburgh minimal medium with Glutamic acid nitrogen source (EMMG). Growth rates were determined from growth curves generated using a BMG Spectrostar Nano plate reader while shaking cells at 36°C. All cells were maintained as early to midlog phase cultures for 48 h before being used for all analyses.

Microscopy
Imaging was undertaken as described previously [9]. Timelapse images of > 10 cells undergoing cytokinesis were used to calculate each CAR constriction rate.

Protein purification
Unlabelled Tpm Cdc8 protein was purified as described previously [8]. Tpm Cdc8 Cerulean3 fusions were expressed from pJC20 Tpm cdc8 -Cerulean3-His 6 and pJC20His 6 -Cerulean3-Tpm cdc8 in either BL21 DE3 or BL21 DE3 pNatB [25] cells. Midlog cultures were grown for 3 h with 100 mgÁL À1 IPTG. Cells were harvested, resuspended in 30 mL lysis buffer (20 mM Tris pH 7.5, 100 mM NaCl, 2 mM EGTA and 5 mM MgCl 2 ), lysed by sonication. Debris and insoluble components were removed by centrifugation and the resulting supernatant was incubated with 10 mgÁL À1 DNase and 10 mgÁL À1 RNase at 4°C for 1 h before isolating His 6 -labelled fusions on nickel-agarose columns and eluting with imidazole. After buffer exchange into FPLC loading buffer (5 mM Tris pH 7.0, 100 mM NaCl) the FP-labelled Tpm Cdc8 was further purified with FPLC using 2 9 5 mL Pharmacia HiTrap-Q columns in tandem, by elution with a 100-900 mM NaCl gradient. Fusions were isolated from appropriate fractions, and concentrated in 5 mM Tris pH 7.0. The purity and mass of the proteins were determined by mass spectroscopy, while parallel Bradford, gel densitometry and spectroscopic analyses were used in parallel to confirm protein concentrations. Rabbit actin was purified as described previously [26].

Circular dichroism
Measurements were made in 1-mm quartz cuvettes using a Jasco 715 spectropolarimeter. Cdc8 proteins were diluted in CD buffer (10 mM Potassium phosphate, 500 mM NaCl, 5 mM MgCl 2 pH 7.0) to a concentration of 0.4 mgÁmL À1 . Thermal unfolding data were obtained by monitoring the CD signal at 222 nm with a heating rate of 1°CÁmin À1 . At completion of the melting-curve the sample was cooled at a rate of 20°CÁmin À1 . CD data are presented as differential absorption (ΔA).

Viscomentry
A Cannon-Manning semimicroviscometer was used to determine the viscosity of 20 lM Cdc8 samples at 20 AE 1°C in 1 mL of viscometry buffer (20 mM MOPS, 5 mM MgCl 2 pH 7.0). NaCl concentration was increased from 0 to 250 mM. Kinematic viscosity was calculated using the manufacturer's predetermined microviscometer kinematic viscosity constant (0.03235) and the average efflux time (typically 30-40 s), calculated from five observations per sample at each NaCl concentration.

Actin-binding assay
Cosedimentation assays were performed at 25°C by mixing 10 lM actin with increasing concentrations of Tpm as described previously [27].

Results
In order to determine the impact fusion of a fluorescent protein to the termini had upon the ability of tropomyosin to self-polymerise, associate with actin and function within a cell, bacteria and fission yeast expression constructs were generated which allowed encoding of amino and carboxyl terminal fusions between the fission yeast tropomyosin, Tpm Cdc8 , and the monomeric fluorescent protein, Cerulean3 [28], each juxtaposed a poly-glycine linker (Fig. 1A). The Cerulean3 fluorophore is a relatively bright, photoand thermostable monomeric fluorescent protein, and due to its low emission wavelength, allows excellent spatial resolution in live cell imaging applications.
To facilitate biochemical analysis of the Tpm proteins, each Cerulean3 fusion (Cerulean3-Tpm Cdc8 and Tpm Cdc8 -Cerulean3) together with the wild-type protein was expressed and purified from E. coli cells (Fig. 1B). In addition, wild-type and the Tpm Cdc8 -Cer-ulean3 carboxyl fusion which had been acetylated upon the amino-methionine resides were isolated using The mass and purity of each protein was confirmed using mass spectroscopy analysis ( Table 1).
As a first step, Circular dichroism (CD) analyses were performed upon each protein to assess the impact the addition of Cerulean3 had on the thermal stability of the dimer. As reported previously [27], broad negative CD spectra peaks at 208 and 222 nm, consistent for a-helical structures were observed for the wild-type and modified forms of the Tpm Cdc8 protein (Fig. 1C). The presence of the Cerulean3 fusion had no significant impact upon the overall CD spectra, as the relative CD signal of an equivalent amount of FP alone is 5% that of that observed for the dimeric coiled-coil Tpm Cdc8 proteins (Fig. 1C). Normalised melting curves for absorbance at 222 nm of each protein are shown in Fig. 2A. Once each melting curve was acquired, the protein was cooled and subjected to two further runs. CD spectra were acquired at the start of each run, and were identical for each protein (not shown), indicating each Tpm Cdc8 was able to rapidly refold to its original state. Figure 2B illustrates the first derivative plots for these data, from which mid-point melting temperatures (T m ) were calculated ( Table 1). The T m of unmodified and amino-terminally acetylated Tpm Cdc8 were equivalent to values established previously [27]. The addition of Cerulean3 to the carboxyl end of Tpm Cdc8 had a minimal effect on the thermostabiltiy of Tpm Cdc8 , increasing the T m from 33.4°C to 34.3°C. However, fusing Cerulean3 to the amino-terminal of the Tpm Cdc8 increased the thermal stability of the protein to equal that of acetylated wild-type Tpm Cdc8 , 35.5°C. Similarly, acetylating the amino-terminal methionine of Tpm Cdc8 -Cerulean3 increased the T m to an equivalent value (T m = 35.7°C). These data confirm that the addition of the Cerulean3 polypeptide to either end of Tpm Cdc8 had no significant detrimental effect upon stability of the Tpm Cdc8 coiled-coil protein.
The additional mass from Cerulean3 at the termini of Tpm Cdc8 has the potential to impact the ability to form end-to-end interactions between the amino and carboxyl termini of Tpm Cdc8 dimers and therefore allow the protein to polymerise. Viscometry was used to investigate the effect the addition of the fluorescent protein had upon the ability of Tpm Cdc8 to form filaments. In this assay, a higher viscosity correlates with a greater capacity to form filaments. Upon the addition of NaCl the strength of ionic interactions between the ends of Tpm Cdc8 proteins was reduced, which is reflected in a reduction in viscosity. Figure 2C shows viscosity data for each Tpm Cdc8 and Tpm Cdc8 fluorescent protein fusion. The viscosity of each protein in the absence of NaCl is shown in Table 1. Acetylation increases the propensity of Tpm Cdc8 to form filaments, which was reflected in an increase in the viscosity from 1.08 centiStokes (cSt) for the unmodified protein, to 1.22 cSt for the amino-terminally modified form. The viscosity data for tagged Tpm Cdc8 indicates the end-toend interactions are affected by the addition of the Cerulean3 fluorophore to either terminus of the protein. In the absence of salt, amino-terminally tagged Tpm Cdc8 formed end-to-end interactions and gave a viscosity reading of 1.17 cSt, which is lower than the wild-type acetylated protein. The carboxyl terminal fusion had a more significant impact upon Tpm Cdc8 filament formation, and abolished the ability of the protein to form end-to-end interactions, as the viscosity values for the unacetylated and acetylated Tpm Cdc8 -Cerulean3 proteins of 0.99 and 1.02 cSt, respectively, are equivalent to buffer only control samples. These data indicate that fusion of the Cerulean3 fluorescent protein to either end of Tpm Cdc8 impacts the end-toend interactions between adjacent tropomyosin dimers. In addition to effects on Cdc8 stability and end-toend contacts, we performed cosedimentation experiments to determine whether the addition of the Cerulean3 fusion had an impact upon the ability of Tpm Cdc8 interactions with actin. Examples of Coomassie-stained SDS/PAGE gels used to determine the binding affinity of Cerulean3-Tpm Cdc8 and both unmodified and amino-terminally acetylated Tpm Cdc8 -Cerulean3 are shown in Fig. 3. In each experiment shown the lower faster migrating bands correspond to actin, with the density remaining approximately constant between each sample. The upper bands are Tpm Cdc8 Cerulean3 fusions and the density increases as dimer concentrations increase from left to right. Figure 3A shows a typical example of a binding experiment for the amino-terminal Cerulean3-Tpm Cdc8 fusion. This protein can be seen to bind strongly to actin as a faint band can be observed in the pellet with as little as 0.5 lM Cerulean3-Tpm Cdc8 , with a much denser band visible at 12 lM Tpm. A typical binding curve from a Cerulean3-Tpm Cdc8 -binding experiment is shown in Fig. 3D. The K 50% for this fusion was calculated to be 0.69 lM, which is similar to the previously reported K 50% value for acetylated Tpm Cdc8 (0.46 lM) [8,27]. In contrast, binding data for both unacetylated (Fig. 3B) and acetylated (Fig. 3C) Tpm Cdc8 -Cerulean3 illustrate both forms of the carboxyl terminal fusion bind weakly to actin. Even at 20 lM Tpm, only a small proportion of the protein is in the pellet fraction, while a large amount of Tpm remains in the supernatant. Both acetylated and unacetylated forms of Tpm Cdc8 -Cerulean3 bound extremely weakly to actin and binding curves could not be generated, as the binding coefficients were estimated to be greater than 20 lM.
In conclusion, in vitro analyses show that the addition of a fluorescent protein fusion to either end of the Tpm Cdc8 did not impact the a-helical content or thermal stability of the protein. However, carboxyl terminal fusions inhibited both the ability of Tpm to form end-to-end contacts, and ability to associate with actin. In contrast, the amino-terminal Cerulean3-Tpm Cdc8 fusion protein has properties that closely mimic the acetylated endogenous protein.
We next explored the ability of each fusion to localise and complement the function of the endogenous Tpm Cdc8 protein within the fission yeast cell. Having previously established that expression levels of GFP-Tpm Cdc8 from the nmt41 promoter reflect Tpm Cdc8 levels from the endogenous locus [8], we chose to express the Tpm Cdc8 fusions using the same repressible promoter [23]. We first examined the localisation of the Cerulean3-Tpm Cdc8 and Tpm Cdc8 -Cerulean3 fusion proteins in wild-type cells. Each protein was seen to associate with the actin polymers within the contractile CAR, when endogenous Tpm Cdc8 was present in the cell (Fig. 4A,B). In contrast, only the Cerulean3-Tpm Cdc8 was seen to localise to and support the formation of the actin ring in cells possessing the temperature-sensitive cdc8-110 allele when held at the restrictive temperature  (Fig. 4C, D) Consistent with these findings comparison of growth kinetics between cdc8-110 cells containing plasmid born copies of wild-type Tpm cdc8 , Tpm cdc8 -Cer-ulean3 or Cerulean3-Tpm cdc8 showed that only the amino-terminal fusion permitted growth, at a rate approaching that of wild-type at the restrictive temperature of 36°C (Fig. 4E, Table 1). However, while Ceru-lean3-Tpm Cdc8 facilitated the formation of an actin ring, cells were significantly longer than wild-type (Fig. 4C, Table 1) demonstrating the normal timing of cell division was disrupted. These data suggest that the fusions may be able to form heterodimers with endogenous unlabelled tropomyosin, as only the amino-terminal Cerulean3-Tpm Cdc8 fusion could form dimers with the capacity to form filaments and associate with actin in the absence of the wild-type protein. This is consistent with the in vitro analysis (Figs 2 and 3).
As Tpm Cdc8 plays a key role in regulating the motor activity of both class II and V myosins [10,11,27], we next determined how the fusions affected the ability of the coiled-coil protein to regulate myosin motor activity in vivo. Myo2 is the only essential myosin within the S. pombe cell, and is required for the assembly of the CAR from actomyosin containing medial nodes [29,30], as well as providing the contractile force required for this organelle's constriction [31-33]. As a readout on the ability of each fusion protein to regulate the contractile activity of myosin II [8,15,34], we measured the rate on CAR constriction in cdc8.110 myo2.mCherry cells at 36°C. Consistent with our previous observation that the myo2.mCherry allele is functional at this temperature [9], these cells formed normal contractile rings which constricted at an average rate of 0.088 AE 0.014 lmÁmin À1 (mean ΔCAR-diameter rate AE SD), when wild-type Tpm Cdc8 was expressed from an episomal plasmid (Fig. 5A). In comparable cells expressing Cerulean3-Tpm Cdc8 , the CAR constricted at a rate of 0.09 AE 0.007 lmÁmin À1 . However, in more than 30% of these mitotic cdc8.110 myo2.mCherry Cerulean3-Tpm cdc8 cells, we observed MyoII incorporated into multiple contractile actomyosin rings (Fig. 5B). Thus, expression of Cerulean3-Tpm Cdc8 disrupted the regulation of node-dependent formation of the CAR, but once formed the actin-Cer-ulean3-Tpm Cdc8 copolymers were capable of regulating normal myosin II-dependent contraction. In contrast, cdc8.110 myo2.mCherry cells expressing Tpm Cdc8 -Ceru-lean3 failed to form a stable CAR when cultured at 36°C. Instead Myo2 foci concentrated to the cell equator and incorporated into randomly organised contractile filaments or aggregates (Fig. 5C). Thus, unable to form a functional CAR these cells failed to complete cytokinesis, leading to a loss of viability (Fig. 5C).
Finally, we explored the impact tagging Tpm Cdc8 had upon the ability of the protein to regulate myosin V function. The fission yeast myosin V, Myo52, associates with interphase actin-Tpm Cdc8 copolymers, and moves towards the barbed end of these filaments to deliver cellular cargoes to the growing tips of the cell [35][36][37]. In yeasts Tpm plays a critical role in modulating the duty-ratio of this dimeric motor protein to promote processive movement upon actin [11,38]. Using a cdc8.110 myo52.mNeogreen strain, we tested the ability of each Tpm Cdc8 fusion to facilitate the actin-dependent movement of the myosin V protein.
Myo52 fails to accumulate at the growing ends of fission yeast cells lacking functional Tpm Cdc8 [39]. Consistent with an inability to associate with forminnucleated tropomyosin associated actin polymers, Myo52 movements were not observed in cdc8.110 myo52.mNeogreen cells expressing Tpm Cdc8 -Cerulean3 and the myosin V was not seen to accumulate at ends of cells (Fig. 5D), but instead localise to randomly dispersed foci throughout the cell. In contrast, Myo52 was seen at the ends of cells expressing wild-type Tpm Cdc8 or the amino-terminal Cerulean3-Tpm Cdc8

Discussion
Over the last decade tropomyosin-fluorescent protein fusions have been used to study the organisation of actin polymers within diverse cell types [17][18][19][20][21]. In each of these studies either the wild-type protein is present within the cell or the Tpm being labelled is nonessential, thus the ability of the fusion-protein to complement its function has not been tested. Given the terminal regions of a Tpm are critical for modulating its ability to (a) form end-to-end contacts; (b) regulate cooperative movements on the actin filament; (c) interact with regulatory proteins; and (d) interact with specific actin polymers at precise cellular locations, we decided to investigate the impact FP-Tpm fusions had upon the biophysical properties and function of this essential cytoskeletal protein. Localisation studies of globular proteins have used tagging strategies where an FP is inserted at a surface loop within the protein sequence, resulting in fusion proteins with varying degrees of functionality [40][41][42]. However, this strategy is not suitable for use with tropomyosins as they are single alphahelices, and insertion of an FP with associated flexible linkers would not only disrupt its packing on the actin surface but would also change the flexibility and shape of the tropomyosin filament, and thereby disrupt its ability to regulate actin function [43]. Having a single Tpm protein, Tpm Cdc8 , makes the fission yeast an attractive model for studying the regulation and function of this conserved dimeric protein. We show that fusing Cerulean3 to the carboxyl terminus of this Tpm disrupts the ability of this protein to form endto-end contacts and abolishes its ability to interact with actin and function within a cell, even when amino-terminally acetylated. In contrast, the addition of the monomeric FP to the amino-terminal of Tpm Cdc8 confers the resultant protein with physical properties equivalent to the acetylated wild-type protein. This is consistent with previous studies on vertebrate tropomyosins which indicate amino-terminal fusions do not interfere with headto-tail self-association and actin-binding properties of tropomyosin [19,44,45]. However, modifications to the amino-terminus of the Tpm protein not only affect interactions with other actin-binding proteins [9,45] but also change the structure at the end of actin-Tpm copolymers.
We have previously shown the cellular distribution of Tpm Cdc8 and its interactions with actin and myosin are modulated by amino-terminal acetylation [9,27], a modification common to all Tpms studied to date [4]. This is consistent with the findings presented here. While Cerulean3-Tpm Cdc8 binds to actin and regulates myosin II activity during CAR constriction to the same extent as wild-type protein, its expression often disrupts initial polymerisation events and results in the disruption of normal actin ring formation and a subsequent delay in cell division. This supports models where the Tpm amino-terminus is crucial during actin-Tpm cofilament nucleation and growth. Thus, the addition of FP tags to the end of the Tpm changes the landscape at the end of the actin polymer, disrupting normal amino-terminal signalling. Consistent with this idea, a single amino-acid substitution within the amino-terminus of Tpm Cdc8 (E6K) has been shown to biophysically mimic the effect of amino-terminal acetylation, however, unlike an amino-terminal fusion, can fully complement endogenous cdc8 function [8,27]. The terminus of the Tpm Cdc8-E6K is almost identical to the wild-type protein, and is acetylated in vivo, again supporting the hypothesis that the biophysical landscape at the amino-terminus of the Tpm Cdc8 protein is crucial for its proper regulation and function.
Both Tpm Cdc8 fusion are capable of localising to the CAR in cells possessing the wild-type cdc8 + allele. However, only Cerulean3-Tpm Cdc8 was able to localise and support growth in the absence of functional endogenous protein. This indicates the carboxyl-terminal fusion, which is unable to form end-to-end polymers or associate with actin alone, is capable of forming heterodimers with endogenous Tpm protein.
Similarly, GFP fusions with budding yeast Tpm1 localises to the S. pombe CAR, but do not complement Tpm Cdc8 function. Thus, the ability of a FP-Tpm fusion to decorate actin polymers in a cell does not provide any indication of its functionality at any level.
In conclusion, while the impact of fusing a fluorescent protein to the terminus of a Tpm is likely to vary between different Tpm isoforms and cell types, it is clear from data presented here that their functionality and biophysical properties should be determined and shown to reflect that of the wild-type protein. Unless this is the case, care should be taken in drawing conclusions from data generated from using them.