Research article

Two biochemically distinct and tissue-specific twinfilin isoforms are generated from the mouse Twf2 gene by alternative promoter usage

Elisa M. Nevalainen, Aneta Skwarek-Maruszewska, Attila Braun, Markus Moser, Pekka Lappalainen


Twf (twinfilin) is an evolutionarily conserved regulator of actin dynamics composed of two ADF-H (actin-depolymerizing factor homology) domains. Twf binds actin monomers and heterodimeric capping protein with high affinity. Previous studies have demonstrated that mammals express two Twf isoforms, Twf1 and Twf2, of which at least Twf1 also regulates cytoskeletal dynamics by capping actin filament barbed-ends. In the present study, we show that alternative promoter usage of the mouse Twf2 gene generates two isoforms, which differ from each other only at their very N-terminal region. Of these isoforms, Twf2a is predominantly expressed in non-muscle tissues, whereas expression of Twf2b is restricted to heart and skeletal muscle. Both proteins bind actin monomers and capping protein, as well as efficiently capping actin filament barbed-ends. However, the N-terminal ADF-H domain of Twf2b interacts with ADP-G-actin with a 5-fold higher affinity than with ATP-G-actin, whereas the corresponding domain of Twf2a binds ADP-G-actin and ATP-G-actin with equal affinities. Taken together, these results show that, like Twf1, mouse Twf2 is a filament barbed-end capping protein, and that two tissue-specific and biochemically distinct isoforms are generated from the Twf2 gene through alternative promoter usage.

  • actin dynamics
  • cytoskeleton
  • isoform
  • twinfilin (Twf)


The actin cytoskeleton plays an essential role in numerous cell biological processes, including motility, cell division, polarized growth, cytokinesis, endocytosis and signal transduction. The structure and dynamics of the actin cytoskeleton are both spatially and temporally regulated by a large number of actin-binding proteins, which interact with filamentous and/or monomeric actin (reviewed in [1,2]).

Twf (twinfilin) is an evolutionarily conserved multifunctional actin-binding protein, which is involved in the regulation of motile and morphological processes in organisms from yeasts to mammals [38]. Twf forms a stable, high-affinity complex with ADP-actin monomers, decreases the rate of nucleotide exchange on actin monomers and prevents their assembly to filament ends [9,10]. Interaction between Twf and G-actin can be down-regulated by PIP2 (phosphatidylinositol 4,5-bisphosphate) at least in vitro [10]. Twf also interacts with the heterodimeric capping protein, and at least in budding yeast this interaction is necessary for the correct subcellular localization of Twf to the cortical actin cytoskeleton [1012]. In addition, recent studies have revealed that mouse Twf1 caps filament barbed-ends with a preferential affinity to ADP-bound filament ends. This activity appears to be essential for the role of Twf1 in motile processes [6]. Furthermore, yeast Twf was shown to induce filament severing at low pH and this activity is inhibited by binding to heterodimeric capping protein [13].

Twfs are composed of two ADF-H (actin-depolymerizing factor homology) domains separated by a short linker and followed by a 35-residue C-terminal tail region [14]. The two ADF-H domains are ∼20% homologous with ADF/cofilins and with each other [15]. Studies on mouse Twf1 revealed that the high-affinity actin-monomer-binding site is located in the C-terminal ADF-H domain, whereas the presence of both ADF-H domains is required for filament barbed-end capping [9,16]. The capping-protein-binding site is located in the C-terminal tail region in yeast Twf and in mouse Twf1 [12].

Yeasts, worms and flies have only one Twf isoform. In mammals, however, two Twf genes (Twf1 and Twf2) exist. Twf1 and Twf2 proteins share ∼75% identity. Both mouse Twf isoforms bind actin monomers and capping protein with high affinity, but show different tissue distributions. Twf1 is the major isoform in the developing embryo and in most adult mouse non-muscle tissues, whereas Twf2 expression is most abundant in heart, skeletal muscle and spleen [11].

In the present study we show that the mouse Twf2 gene encodes for two different Twf2 isoforms (designated Twf2a and Twf2b) that are generated by alternative promoter usage. The two Twf2 variants differ only in the very N-terminal region and are identical with each other from Twf2a residue 9 and Twf2b residue 7 onwards. Our results show that Twf2a and Twf2b display distinct tissue-specific expression patterns and differ from each other in respect to interaction with actin monomers. Taken together, these results suggest that alternative promoter usage of the Twf2 gene generates two biochemically distinct Twf isoforms that fulfil specific needs for regulation of actin dynamics in specialized mammalian cell-types.


Plasmid construction

Plasmids for Twf1 and Twf2a expression (pPL144 and pPL182 respectively) as a His-tagged fusion protein in Escherichia coli have been described previously [11,17]. The DNA fragments corresponding to full-length mouse Twf2b, as well as to Twf2a N-terminal domain (residues 1–142), Twf2b N-terminal domain (residues 1–140) and Twf2 C-terminal domain (residues 169–316 in Twf2a and 167–314 in Twf2b) were amplified by PCR from pPL182 plasmid. The oligonucleotides introduced NcoI and HindIII sites at the 5′ and 3′ ends of the PCR products respectively. The PCR fragments were digested with NcoI and HindIII and ligated into the pHAT2 [18] vector to create plasmids pPL481, pPL482, pPL483 and pPL484. The plasmid for expressing mouse α1β2 capping protein in E. coli has been described previously [10]. Generation of the plasmid for expressing Myc-tagged Twf1 in mammalian cells (pPL79) has been described previously [17]. For generation of the constructs for expressing Myc-tagged Twf2a and Twf2b, the corresponding cDNA fragments were amplified by PCR. The oligonucleotides introduced EcoRV and HindIII sites at the 5′ and 3′ ends of the PCR products respectively. The PCR fragments were digested with EcoRV and HindIII and ligated into the Myc-tagged pEGFP-N1 based vector (Clonetech) to create plasmids pPL80 and pPL554 respectively.

Protein purification

The expression and purification of full-length mouse Twf1, Twf2a and Twf2b and the individual domains of Twf2a and Twf2b as His-tagged proteins were carried out as previously described [11]. Mouse α1β2-capping protein was purified as previously described [10]. Actin was purified from rabbit skeletal muscles [19] and labelled by NBD-Cl (7-chloro-4-nitrobenz-2-oxa-1,3-diazole) [20,21] and pyrenyl as described previuosly [22,23]. Recombinant human gelsolin was obtained from Cytoskeleton. Spectrin-actin seeds were isolated from human erythrocytes as described previously [24].

RNA extraction

Total RNA was isolated from various mouse tissues by TRIzol® (Invitrogen) according to the manufacturer's protocol. Briefly, tissue samples were homogenized in a 15 ml tube with Polytron (Glen Mills) in 1.0 ml of TRIzol® per 100 mg of tissue. Homogenates were incubated for 10 min at room temperature (25 °C) and centrifuged at 12000 g for 15 min at 4 °C to pellet insoluble material and high-molecular-mass DNA. Chloroform was added, tubes were shaken for 15 s, incubated for 3 min at room temperature and centrifuged at 12000 g for 15 min at 4 °C. After phase separation, RNA was precipitated with propan-2-ol, washed with 75% (v/v) ethanol and resuspended in an appropriate volume of DEPC (diethyl pyrocarbonate)-treated water. RNA purity and quantity was ascertained from optical density at 260 and 280 nm. The samples were stored at −80 °C until use.

Real-time Q-PCR (quantitative-PCR)

Total RNA (3 μg) was reverse-transcribed at 42 °C for 1 h in 20 μl of solution containing 200 units of SuperScript II RT (reverse transcriptase; Invitrogen) and 500 ng of oligo(dT). The resulting cDNA was either immediately used for non-quantitative RT–PCR or Q-PCR, or stored at −20 °C until use. Q-PCR was applied to determine the expression levels of each isoform in different tissues. Q-PCR reactions were carried out on a LightCycler 480 System (Roche) using the SYBR Green I Master kit (Roche). Reactions had a final volume of 20 μl and contained 4 pmol of each primer and 0.4 μl of cDNA obtained by reverse transcription. Q-PCR was performed with the following temperature conditions: 95 °C for 5 min and 40 cycles at 95 °C for 10 s, 58 °C for 15 s and 72 °C for 15 s. Primers designed for specific detection of different Twf isoforms and GAPDH (glyceraldehyde-3-phosphate dehydrogenase) are listed in Table 1. All primer pairs resulted in a ∼100 bp product. The amplification of the Twf1 cDNA was performed using primers ES110 and ES121, Twf2a with primers ES122 and ES124, and Twf2b with primers ES124 and ES125. GAPDH served as housekeeping gene for the adjustment of relative expression data. The amplification of the Gapdh cDNA was performed using primers ES126 and ES127. To verify the identity of the various PCR products, RT–PCR was performed and the DNA fragments were separated on 2% (w/v) agarose gels, extracted by a NucleoSpinExtract II band extraction kit (Macherey Nagel) and subjected to sequencing.

View this table:
Table 1 Sequences of the oligonucleotides used in Q-PCR

Actin monomer binding assay

Actin monomer binding experiments were performed by monitoring the change in the fluorescence of NBD (7-nitrobenz-2-oxa-1,3-diazole)-labelled G-actin at different Twf concentrations as described previously [9]. ADP-actin was prepared by incubating NBD–actin with hexokinase-agarose beads (Sigma) and glucose for 2 h at 4 °C [25]. The final concentration of actin in these assays was 0.2 μM, and the Twf concentrations ranged from 0.05 to 14 μM. Experiments were carried out at room temperature in F-buffer [20 mM Tris/HCl (pH 8.0), 0.1 mM CaCl2, 0.1 mM DTT (dithiothreitol), 0.2 mM ADP or ATP, 1 mg/ml BSA, 2 mM MgCl2 and 0.1 M KCl]. The reactions were measured using a BioLogic MOS-250 fluorescence spectrophotometer and the data were analysed as described previously [9].

Actin polymerization assays

The effect of Twf on actin filament assembly kinetics was monitored by the increase in pyrenyl-actin fluorescence in the presence of spectrin- or gelsolin-capped filaments as described previously [6]. Experiments were carried out with 2.5 μM actin (10% pyrene labelled) and Twf concentrations ranging from 0.1 to 12 μM. The excitation and emission wavelengths were 365 and 407 nm respectively. Polymerization was monitored at room temperature with a BioLogic MOS-250 fluorescence spectrophotometer at 10 s intervals. The initial rate of filament growth was normalized to the value of 1 measured in the absence of Twf. The data were analysed as described previously using the program SigmaPlot 9.0 [6].

Native gel-electrophoresis assay

Interaction of the Twf isoforms with capping protein was assayed with 6% native polyacrylamide gels. Reactions containing 2 μM wild-type Twf1, Twf2a or Twf2b or a mutant Twf1 lacking the capping-protein-binding capacity and 4 μM α1β2 capping protein in 10 mM Tris/HCl (pH 7.5), 50 mM NaCl and 0.5 mM DTT were incubated for 60 min at room temperature. Reactions (15 μl) were mixed with 5 μl of loading buffer [125 mM Tris/HCl (pH 8.8), 250 mM NaCl, 2.5 mM DTT, 50% glycerol] and loaded on to a 6% native polyacrylamide gel [12]. The gel was run at 100 V for 180 min using native running buffer [25 mM Tris/HCl, 194 mM glycine (pH 8.5) and 0.5 mM DTT], and the proteins were detected by Coomassie Blue staining.

Immunofluorescence microscopy

Isolation, culturing, transfection and immunofluorescence microscopy of neonatal rat cardiomyocytes was carried out as described previously [25a]. The cells were fixed 24 h after transfection. Endogenous Twfs were visualized with an anti-Twf antibody recognizing Twf1 and Twf2 (1:80 dilution), Myc-tagged fusion proteins with mouse anti-Myc antibody (1:500), and secondary antibodies conjugated to fluorescein or rhodamine (Molecular Probes, Invitrogen). Actin filaments were visualized with Alexa Fluor® 488- or rhodamine-conjugated phalloidin at a dilution of 1:300 (Molecular Probes, Invitrogen).


Identification of two Twf2 variants

A previous study identified two Twf isoforms (Twf1 and Twf2) from mammals [11]. To reveal the biological roles of these proteins, we generated Twf2-knockout mice. In the knockout construct, the cassette was inserted into the first coding exon of the Twf2 gene. During our analysis of the knockout mice, we noticed that although Twf2 protein and Twf2 mRNA were absent from all non-muscle tissues examined, Twf2 mRNA and Twf2 protein were still present in striated muscle and heart (E.M. Nevalainen, A Braun, M.K. Vartiainen, M. Moser, P. Lappalainen and R. Fässler, unpublished work). Subsequent EST (expressed sequence tag) database searches revealed two potential mouse Twf2 variants.

In contrast with the previously identified Twf2 variant [11] (which will be referred to as Twf2a), Twf2b uses an alternative translation initiation codon, which is located 18 nucleotides upstream from the 3′ splice acceptor site of the Twf2a intron 1 (Figure 1A). A TATA box located ∼20 residues upstream of the translation initiation codon of Twf2b, as well as a conserved putative promoter region 5′ to the TATA box, was also present (results not shown). This gives rise to a Twf2b mRNA that contains an alternative first exon (referred to as exon 2b in Figure 1A) encoding for a Twf2b protein of 347 residues as compared with Twf2a consisting of 349 residues (Figure 1B). Alignment of the amino acid sequences of mouse Twf2a and Twf2b showed that they are identical with each other, with the exception of the 6/8 N-terminal residues.

Figure 1 Generation of two isoforms from the Twf2 gene

(A) The organization of the Twf2 gene. In contrast with Twf2a, which contains nine exons, the first exon is omitted from the Twf2b mRNA and the translation begins from exon 2 (marked 2b), which is elongated as compared with the corresponding exon in Twf2a. (B) Alignment of mouse Twf2a and Twf2b amino acid sequences. The two sequences differ at the very first 6/8 N-terminal residues. The GenBank® accession numbers of mouse Twfs are AY267188 (Twf1), AY267189 (Twf2a) and AK002699.1 (Twf2b).

Tissue-specific expression patterns of Twf2a and Twf2b

To investigate whether Twf2b mRNA is expressed in vivo and to analyse its expression pattern in comparison with Twf1 and Twf2a, RNA from mouse heart, lung, kidney, spleen, liver, skeletal muscle and brain was extracted. RNAs were reverse-transcribed and the expression of each isoform from these tissues was first examined by a non-quantitative RT–PCR using isoform-specific primer pairs (Table 1). Twf2a was expressed in all tissues examined and Twf1 in all tissues except skeletal muscle; however, based on this experiment, Twf2b expression was restricted to heart and skeletal muscles (Figure 2A). The PCR products were also isolated from an agarose gel and sequenced to verify their identity (results not shown).

Figure 2 Expression profiles of the three mouse Twf isoforms

(A) A non-quantitative RT–PCR analysis of the expression of different Twf and Gapdh mRNAs in selected mouse tissues. (B) Q-PCR analysis from mRNAs isolated from the mouse tissues indicated. The PCR primers used are listed in Table 1. Gapdh amplification was used as a control. Twf2b is the most abundant isoform in heart and skeletal muscles, whereas Twf2a and Twf1 are widely expressed in non-muscle tissues. Twf2a is the predominant isoform in spleen and Twf1 is the predominant isoform in lung, kidney, liver and brain.

The relative amounts of each isoform were subsequently analysed by Q-PCR (Figure 2B). Q-PCR analysis revealed that Twf2a was expressed in all tissues examined and represents the most abundant isoform in spleen. In contrast, Twf2b expression was restricted to skeletal muscle and heart. Interestingly, Twf2b was the dominant isoform in these tissues at the mRNA level when compared with Twf1 and Twf2a expression. As reported previously [11], Twf1 mRNA was widely expressed in non-muscle tissues, but was absent from skeletal muscle (Figure 2B).

Actin monomer binding properties of Twf2a and Twf2b

To compare the biochemical properties of the Twf isoforms, we produced recombinant mouse Twf2a and Twf2b, as well as their individual ADF-H domains, as His-tagged proteins. All proteins were soluble and monomeric according to their elution position from a gel-filtration column (results not shown). We first examined actin monomer binding of all three Twf isoforms under physiological ionic conditions by using a fluorometric assay with NBD-labelled actin monomers. As reported previously for Twf1 and Twf2a [9,11], Twf2b also bound ADP-G-actin with a significantly (>8-fold) higher affinity than ATP-G-actin (Figures 3A–3C). As previously shown for Twf1 [9], the C-terminal ADF-H domain, common to both Twf2 isoforms, bound ADP-G-actin with very high affinity (Figure 3F).

Figure 3 Interaction of mouse Twfs and the isolated ADF-H domains of Twf2 with actin monomers

The increase in the fluorescence of NBD-labelled ATP-G-actin (○) or ADP-G-actin (●) was measured at different concentrations of mouse Twf isoforms and the isolated domains of Twf2a and Twf2b. The experiment was carried out with 0.2 μM actin under physiological ionic conditions. Symbols indicate the data and the solid lines are fitted binding curves for a complex with a stoichiometry of 1:1. (A) Full-length Twf1; (B) full-length Twf2a; (C) full-length Twf2b; (D) N-terminal ADF-H domain of Twf2a; (E) N-terminal ADF-H domain of Twf2b; and (F) C-terminal ADF-H domain of Twf2a/b. Act, actin.

However, the actin monomer binding properties of the N-terminal ADF-H domains of Twf2a and Twf2b differed significantly from each other. Whereas the N-terminal ADF-H domains of Twf2a (Figure 3D) and Twf1 [9] bound ADP-G-actin with a relatively low affinity (KD∼500–700 nM), the corresponding domain of Twf2b bound ADP-G-actin with significantly higher affinity (Figure 3E). It is also important to note that in contrast with all ADF-H domains examined so far [9,2628], the N-terminal ADF-H domain of Twf2a does not show preferential binding to ADP-actin, but instead binds ADP- and ATP-G-actin with equal affinities (Figure 3D).

All mouse Twfs cap actin filament barbed-ends

Previous results have shown that, when using mammalian actin, mouse Twf1 caps actin filament barbed-ends, whereas budding yeast and Drosophila melanogaster Twfs do not display barbed-end-capping activity [6]. Therefore we examined whether mouse Twf2a and Twf2b also display filament barbed-end capping or if this activity is specific for mammalian Twf1. Barbed- and pointed-end-capping activity was assayed by examining the effects of purified Twf2a and Twf2b on the growth of spectrin- or gelsolin-capped actin filaments respectively. We used mouse Twf1, which inhibits pointed-end growth by sequestration and prevents barbed-end assembly by both sequestering actin monomers and capping filament ends, as a reference in the experiments.

Like Twf1, Twf2a and Twf2b inhibited filament barbed-end growth in a range of concentrations substoichiometric to G-actin (Figures 4A–4C). This is not possible by mere sequestering, but is consistent with capping of the barbed-ends in addition to sequestration of actin monomers. The pointed-end growth was inhibited by sequestering actin monomers only. KF values for binding of Twf1, Twf2a and Twf2b to actin filament barbed-ends were 0.14 μM, 0.19 μM and 0.26 μM respectively and KT values for ATP-G-actin sequestration were 0.76 μM, 0.91 μM and 0.81 μM respectively. These results show that Twf2a and Twf2b are also actin filament barbed-end-capping proteins.

Figure 4 All mouse Twf isoforms cap actin filament barbed-ends

In a pyrene-actin polymerization assay, barbed-end growth (●) and pointed-end growth (○) were initiated by using spectrin-actin seeds and gelsolin-actin seeds respectively. A 2.5 μM concentration of 10% pyrenyl-labelled G-actin was used and the Twf concentrations were as indicated on the x-axis. The initial polymerization rates were normalized to the value of 1 measured in the absence of Twf. Similar to Twf1, the inhibition of barbed-end polymerization was stronger at low concentrations of Twf2a and Twf2b than expected from monomer sequestering, indicating barbed-end capping. The effects of Twf isoforms at pointed-ends are accounted for by sequestration of actin in a 1:1 complex. The obtained KT (ATP-G-actin sequestration) and KF values (barbed-end capping) are shown in the Figure.

Twf2a and Twf2b bind heterodimeric capping protein

In addition to actin, Twf1 and Twf2a were shown to bind mouse α1β2 capping protein in vitro [11]. The interaction with heterodimeric capping protein is essential for the correct subcellular localization of Twf in budding yeast [10], but the possible biological role of this interaction in mammalian cells is presently not known. Yeast Twf does not affect the filament barbed-end capping activity of yeast capping protein [12]. However, because mammalian Twf1 and Twf2a also display filament barbed-end capping activity ([6] and Figure 4 of the present study), their possible effects on the barbed-end capping activity of heterodimeric capping protein cannot be reliably examined.

To examine whether Twf2b also binds capping protein, we performed a native gel-electrophoresis assay with purified full-length mouse Twfs and mouse α1β2-capping protein. The capping-protein-binding site resides in the C-terminal tail of yeast Twf and mouse Twf1, and thus a mutant Twf1 construct lacking the C-terminal tail was used as a negative control [12]. Purified capping protein runs as a single band below the migration positions of all full-length Twf isoforms on a native polyacrylamide gel (Figure 5, lane 1). When mixed with each other in a 1:2 molar ratio, a complex of Twf and capping protein forms. On a native gel, this complex migrates above the positions of uncomplexed Twf and capping protein (Figure 5, upper band on lane 3), whereas the band corresponding to the migration position of uncomplexed Twf disappears. As expected, the C-terminally deleted Twf1 protein did not display an interaction with capping protein (Figure 5, lanes 4 and 5); however, both Twf2 isoforms interacted with capping protein similar to Twf1 (Figure 5, lanes 6–9).

Figure 5 All mouse Twf isoforms interact with heterodimeric capping protein

The binding of Twf isoforms to capping protein was studied by a native gel-electrophoresis assay. Mouse full-length Twfs and a mutant Twf1 lacking the capping-protein-binding site were loaded on to a gel either separately or as a mixture with mouse α1β2-capping protein. The concentration of Twfs was 2 μM and the concentration of capping protein was 4 μM. Lane 1 shows the mobility of capping protein alone and lanes 2, 4, 6 and 8 show the mobilities of wild-type Twf1, mutant Twf1, Twf2a and Twf2b respectively. When mixed with capping protein, the Twf1, Twf2a and Twf2b disappear from their original migration position and a new band corresponding to the Twf–capping protein complex appears above the original migration positions of the proteins (lanes 3, 7 and 9). In contrast, the electrophoretic mobility of the mutant Twf1 lacking the capping-protein-binding site [12] is not affected by the presence of capping protein (lane 5).

Subcellular localizations of Twfs in rat cardiomyocytes

In mammalian non-muscle cells, Twf1 and Twf2a show a predominantly punctate cytoplasmic localization that, at least in the case of Twf1, partially overlaps with endosomes. In addition, Twf1 localizes to cell–cell contacts and to lamellipodial actin networks in cells expressing dominant-active forms of Cdc42 (cell division cycle 42) and Rac GTPases [6,11]. To examine the subcellular localizations of Twfs in muscle cells, we first performed immunofluorescence microscopy on cultured neonatal rat cardiomyocytes by using an antibody that recognizes all three Twf isoforms. These cells display regular, contractile myofibrils and thus provide a good model for studying the subcellular localizations of Twfs in heart cells, where all three isoforms are expressed (see Figure 2). Immunofluorescence microscopy revealed that Twfs display predominantly punctate cytoplasmic staining also in cardiomyocytes, although in some cells endogenous Twfs also concentrated to myofibrils (Figure 6A).

Figure 6 Localization of the Twf isoforms in heart muscle cells

(A) Subcellular localization of endogenous Twfs in cultured neonatal rat cardiomyocytes. An antibody recognizing all three Twf isoforms showed punctate cytoplasmic localization and also revealed enrichment of Twfs to myofibrils in a subset of cells. (B) In cells expressing Myc-tagged Twfs, Twf1 and Twf2b displayed mainly punctate cytoplasmic localization, but were also enriched in myofibrils (shown by arrowheads in the insets). Twf2a displayed punctate cytoplasmic localization with some enrichment in the regions between the myofibrils (shown by arrows in the insets). Bars=10 μm.

To compare the subcellular localizations of each Twf isoform in cardiomyocytes, we expressed N-terminally Myc-tagged Twf1, Twf2a and Twf2b in these cells. Similar to endogenous Twf, the Myc-tagged Twfs also displayed mainly punctate cytoplasmic localizations. Twf1 and Twf2b were also enriched along myofibrils in a subset of cells, whereas in the case of Twf2a, significant enrichment to myofibrils was not observed (Figure 6B). Instead, Twf2a was often concentrated to regions between myofibrils (Figure 6B and results not shown).


Twf is an evolutionarily conserved protein that regulates actin dynamics in organisms from yeasts to mammals. Mammals have two Twf genes that were previously shown to encode biochemically very similar proteins with slightly different expression patterns [11]. In the present study we demonstrate that a third mammalian Twf isoform is produced through alternative promoter usage of the Twf2 gene. Interestingly, the two Twf2 variants display distinct tissue-specific expression patterns, suggesting that promoter usage of Twf2 is tightly and tissue-specifically regulated. Whereas the expression of Twf2b is restricted to striated muscle, where it is also the predominant Twf isoform, Twf2a is mainly co-expressed with Twf1 in most non-muscle tissues. In cultured cardiomyocytes Twfs show mainly punctate cytoplasmic localization, although Twf1 and Twf2b also concentrate to myofibrils.

As described in the present study for Twf2, previous studies have shown that the two ADF/cofilin family actin-binding proteins of Caenorhabditis elegans, UNC-60A and UNC-60B, are generated from the unc-60 gene; however, the UNC60A and UNC60B isoforms are generated through alternative splicing instead of alternative promoter usage as shown in the present study for Twf2 [29]. UNC-60A is expressed in various tissues and is required for early embryogenesis, whereas UNC-60B is specifically expressed in body wall muscle and is essential for myofibril assembly [30,31]. Muscle-specific isoforms of central actin-regulating proteins are common in mammals; however, they are usually encoded by a separate gene from their corresponding non-muscle-specific isoforms, mainly as a result of gene duplications during evolution. These include for example muscle-specific ADF/cofilin, cyclase-associated-protein and α-actinins [3236]. In our database searches, we found Twf2b homologues from rat and chicken, suggesting that the alternative promoter usage that produces the two variants of Twf2 is evolutionarily conserved in vertebrates. Furthermore, alternative splicing of the chicken Twf2 gene from another position was also recently reported [37]. Interestingly, we did not find the Twf2b variant from human ESTs, suggesting that this specific alternative promoter usage event may have been lost during the evolution of primates.

The present study demonstrates that all three mammalian Twf isoforms bind capping protein and cap filament barbed-ends in a similar manner. Thus these results provide evidence that actin filament barbed-end capping activity is not restricted to Twf1 only, but may be a general feature of all mammalian Twfs. In addition, all three Twfs bind actin monomers, but display differences in these interactions. The isolated N-terminal ADF-H domain of Twf2b binds ADP-G-actin with approx. 2.5-fold higher affinity compared with the corresponding domain from Twf2a. However, the N-terminal domain of Twf2a displays approx. 2-fold higher affinity to ATP-G-actin than the corresponding Twf2b domain. Consequently, the full-length Twf2a binds ATP-G-actin with higher and ADP-G-actin with lower affinity than Twf2b. These differences in G-actin interactions can be explained by the fact that the N-terminal region, which differs between Twf2a and Twf2b, was shown to play a central role in G-actin binding of Twf1 [38].

In conclusion, we have identified two biochemically distinct variants of the Twf2 protein, and shown that they are expressed differentially. Future cell biological and knockout mouse studies are required to elucidate the biological roles of these two Twf isoforms and to reveal the possible redundant roles of Twf2a and Twf1 in non-muscle tissues. Furthermore, it will be important to examine the role of Twf2b in skeletal and heart muscle cells.


This work was supported by the Sigrid Juselius Foundation and the Finnish Cancer Organization (to P. L.).


We thank Mikko Frilander, Reinhard Fässler, Ville Paavilainen and Martina Serlachius for critical reading of the manuscript prior to submission. We thank Maria Vartiainen for reagents (Cell and Molecular Biology Programme, Institute of Biotechnology, University of Helsinki, Helsinki, Finland).

Abbreviations: ADF-H, actin-depolymerizing factor homology; DTT, dithiothreitol; EST, expressed sequence tag; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; NBD, 7-nitrobenz-2-oxa-1,3-diazole; Q-PCR, quantitative-PCR; RT, reverse transcriptase; Twf, twinfilin


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 25a.
  27. 26.
  28. 27.
  29. 28.
  30. 29.
  31. 30.
  32. 31.
  33. 32.
  34. 33.
  35. 34.
  36. 35.
  37. 36.
  38. 37.
  39. 38.
View Abstract