2017 Volume 42 Issue 2 Pages 131-140
Actin, a major component of microfilaments, is involved in various eukaryotic cellular functions. Over the past two decades, actin fused with fluorescent protein has been used as a probe to detect the organization and dynamics of the actin cytoskeleton in living eukaryotic cells. It is generally assumed that the expression of fusion protein of fluorescent protein does not disturb the distribution of endogenous actin throughout the cell, and that the distribution of the fusion protein reflects that of endogenous actin. However, we noticed that EGFP-β-actin caused the excessive formation of microfilaments in several mammalian cell lines. To investigate whether the position of the EGFP tag on actin affected the formation of filaments, we constructed an expression vector harboring a β-actin-EGFP gene. In contrast to EGFP-β-actin, cells expressing β-actin-EGFP showed actin filaments in a high background from the monomer actin in cytosol. Additionally, the detergent insoluble assay revealed that the majority of the detergent-insoluble cytoskeleton from cells expressing EGFP-β-actin was recovered in the pellet. Furthermore, we found that the expression of EGFP-β-actin affects the migration of NBT-L2b cells and the mechanical stiffness of U2OS cells. These results indicate that EGFP fused to the N-terminus of actin tend to form excessive actin filaments. In addition, EGFP-actin affects both the cellular morphological and physiological phenotypes as compared to actin-EGFP.
Key words: actin, GFP, cytoskeleton and probe
In most eukaryotic cells, cytoplasmic actin plays essential roles in numerous, fundamental cellular processes, such as cell adhesion, cell migration, cytokinesis, and membrane trafficking (Le Clainche and Carlier, 2008; Pantaloni et al., 2001). In mammalian cells, there are two types of cytoplasmic actin genes, ACTB and ACTG1, which encode β-actin and γ-actin, respectively (Perrin and Ervasti, 2010), and differ only with respect to the conservative substitutions of four amino acid residues near the N-terminus. Monomeric actin, often called G-actin, is a globular protein and polymerizes to form filaments (F-actin). F-actin and its binding proteins are known to compose actin-based cellular structures, such as stress fibers, focal adhesions, cellular cortices, lamellipodia, and filopodia, which regulate various cellular events (Le Clainche and Carlier, 2008). The regulation of such cellular processes is thought to be under the control of both the actin cytoskeleton rearrangement, via polymerization and depolymerization and interaction with specific actin-binding proteins in response to physiological signals.
A variety of methods have been developed for the microscopic visualization of the actin cytoskeleton in order to understand the role of actin. For example, immunofluorescence staining using specific antibodies for actin (Lazarides and Weber, 1974) and phalloidin labeling (Wulf et al., 1979) have been widely used to visualize actin in fixed cells. To observe the dynamics of the actin cytoskeleton in living cells, methods to inject actin labeled with a fluorescent dye were developed (Kreis et al., 1979; Taylor and Wang, 1978; Yumura, 1996). However, microinjected fluorescent actin was unstable in living cells. About 20 years ago, pivotal papers reported the visualization of the live dynamics of actin filaments by observations of fusion proteins, composed of actin and green fluorescent protein (GFP) from Aequorea jellyfish, in yeast (Doyle and Botstein, 1996) and Dictyostelium (Westphal et al., 1997). GFP-actin was also found to accumulate in the lamellipodia and stress fibers of mammalian cells (Choidas et al., 1998). Subsequently, vectors harboring the EGFP-actin gene became commercially available and have been widely used to visualize actin filaments in living cells. During the early development stage of GFP-actin probes, the GFP moiety was fused to the N-terminus of actin (GFP-actin) and expressed in mammalian and Dictyostelium cells (Choidas et al., 1998; Fischer et al., 1998; Westphal et al., 1997). Thereafter, to the best of our knowledge, EGFP-actin, in which EGFP was fused to the N-terminus of actin, was used to visualize cytoskeletal actin in living mammalian cells in all reports in which information on the organization of the fusion protein was provided, with the exception of two cases (Dugina et al., 2009; Hodgson et al., 2000).
However, the fusion of EGFP to the N-terminus of actin may be problematic, because the N-terminus of β-actin is post-translationally modified in vivo. First, the initiating methionine of β-actin is removed by an aminopeptidase, and the newly generated N-terminal amino group of aspartic acid is then acetylated by an acetyltransferase (Rubenstein and Martin, 1983). These modifications of the actin N-terminus appear to be essential for the maturation of β-actin in higher eukaryotic cells (Abe et al., 2000; Mayer et al., 1989). In Drosophila, inhibition of the N-terminal modification of actin by the removal of the responsible aminotransferase causes an aggregation of actin filaments (Schmitz et al., 2000). Moreover, an arginyltransferase commonly adds arginine to the N-terminus of β-actin via a peptide bond, which is known to regulate the actin cytoskeleton and cell motility (Karakozova et al., 2006).
These studies led us to investigate whether the fusion of fluorescent protein to the N-terminus of actin affects the physiological properties of actin. More specifically, we compared the physiological differences between EGFP-actin and actin-EGFP, and found that EGFP-actin exhibited excessive actin polymerization in cells compared with actin-EGFP. These results suggest that the behavior of the EGFP-actin fusion protein may not faithfully represent that of endogenous actin. Moreover, the expression of EGFP-actin affects the mechanical properties of cells, warranting caution when interpreting data derived from cells expressing this widely used actin probe.
All mammalian cell lines were obtained from the RIKEN Bioresource Center. HeLa, HEK293, and U2OS cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM; Sigma-Aldrich, Tokyo, Japan) containing 10% fetal bovine serum (FBS; Daiichi Pure Chemicals, Tokyo, Japan). NBT-L2b cells were grown in Minimum Essential Media (MEM; Sigma-Aldrich) containing 10% FBS, 0.1 mM nonessential amino acids (Wako, Osaka, Japan), 0.5 mM sodium pyruvate (Wako), and a mixture of antibiotics including penicillin and streptomycin. Transfection of cells was performed using the Polyethylenimine Max reagent (Cosmo Bio, Tokyo, Japan), according to the manufacturer’s instructions.
The established Arabidopsis cell line, T87, was obtained from the RIKEN Plant Cell Bank. T87 cells were cultured in Jouanneau and Péaud-Lenoël medium (Jouanneau and Péaud-Lenoël, 1967) with gentle suspension at 22°C under continuous white light (Yamada et al., 2004). Transient gene expression was performed using a protocol for transient expression in Arabidopsis mesophyll protoplasts based on the polyethylene glycol (PEG)-calcium transfection method (Yoo et al., 2007).
The axenic Dictyostelium cell line Ax2 was obtained from the National BioResource Project Nenkin (University of Tsukuba, Japan). Dictyostelium cells were cultured in HL-5 medium supplemented with a 6 μg/mL penicillin/streptomycin mixture (Wako) and 20 μg/mL folate (Wako). Transformants were generated by electroporation, as described previously (Nagasaki et al., 2002), and maintained in HL-5 medium containing 12 μg/mL G418 (Nacalai Tesque, Kyoto, Japan).
Molecular cloningThe pEGFP vectors were obtained from Clontech Laboratories (Takara, Shiga, Japan). Full-length human β-actin cDNA was cloned from a human cDNA library by RT-PCR. The EGFP-β-actin expression vector was generated by the insertion of human β-actin cDNA between the EcoRI and HindIII sites of pEGFP-C3 (Clontech). The expression vector for β-actin-EGFP was created by cloning a human β-actin PCR product followed by insertion of a flexible linker (GGSGGS) between the HindIII and PstI sites of pEGFP-N1 (Clontech). The mCherry-β-actin fusion protein expression vector was described in our previous report (Nagasaki et al., 2008). For the Lifeact-EGFP expression vector, Lifeact/pEGFP-N1, a Lifeact cDNA fragment was synthesized and subcloned between the BamHI and BglII sites of the pEGFP-N1 plasmid. The Lifeact-mKate2 expression vector was generated by replacement of EGFP cDNA in Lifeact/pEGFP-N1 vector with a PCR fragment of mKate2 cDNA using the BamHI and NotI sites (Evrogen, Moscow, Russia). To generate vector harboring EGFP-β-actin with the same linker of β-actin-EGFP, DNA fragment of the linker coding 21 amino acids between β-actin and EGFP in pEGFP-N1 plasmid was amplified by PCR with a pair of primers, 5'-CTCGAGGGTGGCTCTGGAGGCT-3' and 5'-AAGCTTGGTGGCGACCGGTGGA-3'. The PCR fragment was inserted in XhoI and HindIII sites of the multiple cloning site between β-actin and EGFP cDNA in pEGFP-C3 vector. For the expression of exogenous actin without an EGFP tag, we subcloned Hs β-actin cDNA between the EcoRI and HindIII sites in a pcDNA3 vector (Invitrogen).
The GFP-actin fusion protein expression vector for Dictyostelium cells has been described in our previous report (Asano et al., 2004). To create a vector to express the actin-GFP fusion protein, we used the pTIKL-ART vector (Noguchi et al., 2007). A GFP cDNA fragment following a 3× GSS linker sequence was inserted between the NheI and SacI sites, downstream of the actin coding sequence and upstream of the terminator sequence, in the pTIKL-ART vector.
The expression plasmid for the plant cell line T87 was a kind gift from Dr. Sam-Geun Kong. This expression vector is based on the pUC plasmid, which contains an expression unit for plant cells driven by the 35S promoter and containing the nos terminator (Kong et al., 2006). To investigate the localization of actin in plant cells, we cloned Arabidopsis vegetative actin, ACT7. ACT7 cDNA was obtained from pTIKLART-ACT15P-ACT7 (Kijima et al., 2016) by PCR using specific primers. Coding sequences for ACT7 and GFP were inserted between the 35S promoter and the nos terminator of this expression vector, separated by a 6× GSS linker.
Triton-insoluble cytoskeleton assayTo estimate the relative amounts of cellular F-actin and G-actin, we performed a Triton X-100-insoluble cytoskeleton assay. HEK293 cells were transfected with a vector that expressed either EGFP-β-actin or β-actin-EGFP. Forty-eight hours after transfection, cells were gently washed twice with phosphate-buffered saline (PBS). The cells were treated with lysis buffer [10 mM HEPES (pH 7.0), 50 mM NaCl, 1% Triton X-100, 1 mM MgCl2, and 2.5 mM EGTA], and the samples were harvested using a rubber scraper. After incubation for 10 min at room temperature, the samples were centrifuged at 8,000×g for four min. Soluble and insoluble fractions were subjected to SDS-PAGE (Sodium dodecyl sulfate-Polyacrylamide gel electrophoresis) and analyzed by western blotting using monoclonal anti-actin antibodies (Clone C4, Millipore, Billerica, MA, USA). We used the Student’s t-test for Western blotting data to compare between control cells (EGFP) and each cell line expressing GFP fusion proteins.
MicroscopyTo investigate the localization of actin fused to EGFP in each mammalian cell line, cells were transfected with either the EGFP-β-actin or β-actin-EGFP expression vector. Transfected cells were incubated for at least 24 h and were then transferred to a collagen-coated glass-bottom dish (Matsunami, Osaka, Japan). After at least 6 h of incubation, the medium was exchanged with DMEM/F12 without phenol red (Dulbecco’s Modified Eagle Medium: Nutrient Mixture F-12; Sigma-Aldrich), containing 10% FBS to reduce background fluorescence. For confocal fluorescence imaging, we used an inverted microscope (IX71; Olympus, Tokyo, Japan) equipped with a confocal scanning unit (CSU10; Yokogawa, Tokyo, Japan) and a cooled CCD camera (ORCA-ER; Hamamatsu Photonics, Hamamatsu, Shizuoka, Japan). All confocal images, except for those shown in Fig. 5 and Supplementary Figures, were Z-projection images from Z-series stacks processed by the ImageJ software. Confocal images of Dictyostelium and Arabidopsis cells expressing fusion proteins were obtained using our previous method (Nagasaki et al., 2008). Briefly, cells expressing fusion proteins were transferred to a new glass-bottom dish. After at least 20 minutes, Dictyostelium and Arabidopsis cells were observed by confocal microscopy using 60× and 100× objective lenses, respectively, as described above.
Cell motility assays were performed, as described previously (Onuki-Nagasaki et al., 2015). Briefly, for the time-lapse recording of migrating cells expressing actin fused to EGFP, cells were observed with a cooled CCD camera (ORCA-AG; Hamamatsu Photonics) attached to an inverted microscope (IX71), with an incubator (Tohkai Hit, Shizuoka, Japan) set at 37°C on the stage. Time-lapse images were acquired for 8 h at 300-s intervals. Time-lapse images were analyzed by the ImageJ software with a manual tracking plug-in, and the migration speeds of different cell lines were compared using the Steel-Dwass test (statistical significance recognized at P<0.01).
For rhodamine-phalloidin staining, cells were initially fixed with PBS containing 3.7% formaldehyde for 30 min at room temperature; cells were then permeabilized and stained with PBS containing 0.1% Triton X-100 and 100 nM rhodamine-labeled phalloidin (Molecular Probes, Eugene, OR, USA) for 30 min at room temperature.
Atomic force microscopyFor the estimation of membrane stiffness of cells expressing fusion proteins, we used atomic force microscopy (AFM; Nanowizard II BioAFM; JPK Instruments, Berlin, Germany), combined with an inverted microscope (Olympus). A normal pyramidal AFM tip (ATEC-Cont, Nanosensors, Neuchatel, Switzerland) was etched to form a cylinder-shape AFM tip, 2.5 μm in diameter and 7.0 μm in height, using a focused ion beam (SMI500, Hitachi High-Tech Science, Tokyo, Japan), which was then used for the determination of mechanical stiffness. U2OS cells were cultured on a plastic dish and transfected with expression vectors for each actin probe. Transformants exhibiting green fluorescence were indented with the cylindrical AFM tip at an approach velocity of 10 μm/s, avoiding the nuclear region, until a repulsive force of 10 nN was reached. The resulting force-indentation curves were fitted by the following equation of the Hertz model for the cylindrical tip:
where F, a, E, I, and ν represent force, the radius of the cylinder, Young’s modulus, the indentation depth, and Poisson’s ratio; we set at 0.5 in this study referring to a previous study (Chiou et al., 2013), to evaluate Young’s moduli of the tested cells as parameters of cell stiffness (Obataya et al., 2005). We used the Student’s t-test comparing between control cells (EGFP) and each cell line expressing GFP tags for actin.
For the visualization of actin dynamics in living cells, the EGFP-β-actin fusion protein has been widely used as a standard probe. However, some reports have suggested that the EGFP-β-actin fusion protein alters the mechanical responses of actin in transfected cells (Feng et al., 2005; Deibler et al., 2011). Furthermore, post-translational modifications of the N-terminus of actin are required for the maturation of the actin protein (Terman and Kashina, 2013). Since the N-terminus of actin is tagged with GFP in GFP-β-actin, the GFP-β-actin fusion protein may not be able to properly mature.
To investigate this issue, a vector that expresses β-actin-EGFP was constructed and transfected into four mammalian cell lines. It was immediately obvious that the distribution of the fusion proteins was very different in cells expressing EGFP-β-actin and those expressing β-actin-EGFP (Fig. 1A). As already known, EGFP-β-actin displayed prominent localization along stress fibers with relatively low diffuse cytoplasmic distribution in HeLa and U2OS cells. Such a distribution is very similar to the staining pattern of rhodamine-phalloidin (compare the top and bottom rows in Fig. 1A). In contrast, HeLa and U2OS cells expressing β-actin-EGFP displayed highly diffuse cytoplasmic distribution, which presumably represents the distribution of monomeric (G-actin) or oligomeric actin-β-EGFP. NBT-L2b cells were initially isolated from the NBT-II cell line as a highly invasive clone (Nishi et al., 1997), and these cells move rapidly and persistently in one direction on collagen-coated surfaces, maintaining a semi-circular shape with a large lamellipodium. The distribution of β-actin-EGFP in NBT-L2b cells was similar to that in HeLa and U2OS cells, in that β-actin-EGFP localized mainly in the cytoplasm. In HEK293 cells, prominent stress fibers were not observed; both EGFP-β-actin and rhodamine-phalloidin were concentrated in peripheral protrusions (white arrows in Fig. 1A, right column), whereas β-actin-EGFP was uniformly distributed in the cytosol. Actin-containing structures in cells expressing β-actin-EGFP were not as clearly visible as in cells expressing EGFP-β-actin (Fig. 1A, middle). Nonetheless, as shown in enlarged images of the boxed regions of Fig. 1A, weak fluorescence of β-actin-EGFP was observed along stress fibers in HeLa and U2OS cells, and in peripheral projections of NBT-L2b and HEK293 cells. Cortical actin arcs were visualized in HeLa cells expressing β-actin-EGFP (white arrows in Fig. 1A, left column). Furthermore, when cells expressing the actin probes were fixed and permeabilized with formaldehyde and Triton X-100 to wash out monomeric and oligomeric proteins, actin-containing structures became clearly visible, not only in cells expressing EGFP-β-actin and Lifeact-EGFP, but also in cells expressing β-actin-EGFP (Fig. 1B). Next, we compared distribution of actin fusion proteins tagged with EGFP with that of pan-F-actin labeled with rhodamine phalloidin in the same fixed cells (Supplementary Fig. 1). The distribution of F-actin containing EGFP fusion proteins is consistent with that of rhodamine phalloidin, and line scan analyses confirmed that the fluorescence intensities of EGFP fused with actin and rhodamine phalloidin fluctuated in similar patterns (Supplementary Fig. 1A and B). These results indicate that β-actin-EGFP is able to copolymerize with endogenous actin to form normal actin-containing structures, albeit much less efficiently than EGFP-β-actin. Furthermore, we generated U2OS cells expressing both EGFP actin fusion protein and Lifeact-mKate2 for the visualization of filaments of pan-actin in the same living cells (Supplementary Fig. 2). F-actin in the cells expressing β-actin-EGFP was hardly recognizable in the EGFP channel, due to the high background intensity in comparison to cell expressing EGFP-β-actin, and these results are consistent with the fluorescent images shown in Fig. 1.
Localization of fusion proteins in mammalian cells. (A) Cells were transfected with an expression vector for either EGFP-β-actin (upper panel) or β-actin-EGFP (middle panel). The lower panel shows each cell line labeled with rhodamine-phalloidin. White boxes indicate enlarged inset images. White arrows indicate actin structures described in the text. (B) The upper and lower panels show living and fixed U2OS cells expressing actin probes, respectively. Cells were fixed and permeabilized with 3.7% formalin and 0.1% Triton X-100 for 30 min. Scale bars represent 10 μm.
Distribution of mCherry-β-actin and β-actin-EGFP in living cells. (A) Cells were transfected with vectors that express both β-actin-EGFP (left) and mCherry-β-actin (middle). The merged image of β-actin-EGFP (green) and mCherry-β-actin (red) is shown on the right. Scale bar represents 10 μm. Panel B shows the fluorescent intensity profiles of β-actin-EGFP (left) and mCherry-β-actin (right) along the white line shown in A.
Among the four mammalian cell lines tested, U2OS cells appeared to have the largest difference in the G-actin:F-actin ratio of actin tagged with EGFP, as assessed by the fluorescence intensity in the cytoplasm and the actin-containing structures between EGFP-β-actin and β-actin-EGFP (Fig. 1). Thus, we generated U2OS cells expressing both mCherry-β-actin and β-actin-EGFP for a direct comparison of the distribution of the two fusion proteins in the same cells. In live cells, mCherry-β-actin was concentrated along stress fibers but was barely detectable in the cytoplasm, whereas β-actin-EGFP was distributed both in the cytoplasm and along stress fibers. The intensity profile also indicated that the cytoplasmic fluorescence of β-actin-EGFP was more intense than that of mCherry-β-actin (Fig. 2B). These results suggest that the amounts of G-actin and F-actin differed between mCherry-β-actin and β-actin-EGFP in living cells.
Determination of F-actin and G-actin contents in mammalian cellsNext, we attempted to determine the amounts of cellular G-actin and F-actin using a Triton X-100-insoluble cytoskeleton assay. We also investigated whether β-actin-EGFP or EGFP-β-actin functions more like endogenous actin in living cells. By adding buffer containing Triton X-100 to cells and centrifuging the cell lysates at low speed, G-actin and F-actin were separated into soluble and insoluble fractions, respectively. Western blotting results show that the amounts of actin in the soluble and insoluble fractions are similar in all four cell lines (Fig. 3A). Next, we investigated the amounts of fusion proteins in the soluble and insoluble fractions of transiently transfected HEK293 cells (Fig. 3B). As expected from the microscopic studies (Fig. 1, Fig. 2 and Supplementary Fig. 2), the insoluble fraction of EGFP-β-actin was much greater than the soluble fraction (Fig. 3B and C). Notably, the amount of endogenous actin in the insoluble fraction was somewhat greater than in the soluble fraction, unlike in control HEK293 cells. In contrast, the amounts of soluble and insoluble fractions of β-actin-EGFP were similar, which was also the case for endogenous actin (Fig. 3B). The distribution of endogenous actin was unaffected by the expression of β-actin-EGFP. On the basis of these results, we conclude that the dynamic equilibrium between G-actin and F-actin of β-actin-EGFP, but not that of EGFP-β-actin, is relatively normal in the cells.
Determination of the amounts of F-actin and G-actin in mammalian cells. Western blot analysis of Triton X-100-soluble and -insoluble fractions of cells expressing the individual actin probes using anti-actin antibodies. SUP (supernatant, soluble) and PPT (precipitate, insoluble) fractions are assumed to contain G-actin and F-actin, respectively. Panel A shows endogenous actin in four mammalian cell lines, whereas panel B shows endogenous actin and the individual actin probes in HEK293 cells expressing GFP, EGFP-β-actin, or β-actin-EGFP. Representative data from more than three independent experiments are shown. (C) Effects of fusion protein expression on F-actin levels. Each point is the mean±standard deviation of six independent experiments. * and ** indicate statistically significant differences with P<0.01 and P<0.05, respectively (Student’s t-test).
Next, we examined the possible impact of the expression of EGFP and actin fusion proteins on the morphological and physiological phenotypes of the cells. As described above, rat bladder carcinoma NBT-L2b cells are convenient for studying cell migration, because this cell line has a high migration activity and stable unidirectional motility on surfaces coated with type-I collagen (Nishi et al., 1997). In this study, we compared the effects of several actin probes on NBT-L2b cell migration. Cells transfected with vectors for EGFP, Lifeact-EGFP, EGFP-β-actin, or β-actin-EGFP expression were transferred to collagen-coated dishes, and GFP fluorescence was recorded for 8 h using fluorescent microscopy. The migration speeds of cells expressing EGFP, Lifeact-EGFP, or β-actin-EGFP were approximately 100 μm/h, whereas the migration speed of cells expressing EGFP-β-actin was 80% of the control cell speed (Fig. 4A); this difference was statistically significant (*P<0.01, N>30). These results indicate that EGFP-β-actin expression partially inhibits the migration of NBT-L2b cells, whereas the expression of Lifeact-EGFP or β-actin-EGFP does not significantly affect NBT-L2b cell migration.
Effects of fusion protein expression on NBT cell migration and U2OS cell cortical stiffness. (A) Migration speeds of NBT-L2b cells expressing EGFP-β-actin, β-actin-EGFP, Lifeact-EGFP, or EGFP on collagen-coated plastic dishes. Bars depict means±standard deviations (N>30). * indicates a statistically significant difference (P<0.01 by Steel-Dwass test). (B) AFM analysis of the cortical stiffness of U2OS cells expressing each actin probe. ** P<0.05 comparing EGFP and EGFP-β-actin (N=30, Student’s t-test).
A previous study reported that treatment of cells with cytochalasin D inhibits actin polymerization, reduces the amount of F-actin and decreases stiffness in living Ishikawa cells (Salker et al., 2016). We thus measured the mechanical stiffness of U2OS cells expressing actin probes by AFM. Cells expressing one of several actin probes or EGFP alone were cultured on collagen-coated plastic dishes, and the cells were indented with a cylinder-shape AFM tip to measure cell stiffness (Fig. 4B). Cells expressing EGFP-β-actin were stiffer than the control cells expressing EGFP (P<0.05, N=30), which can be explained by the excessive formation of F-actin due to the expression of EGFP-β-actin. On the other hand, the stiffness of cells expressing β-actin-EGFP was not significantly different from that of control cells expressing EGFP alone.
Distribution of GFP-actin and actin-GFP in Arabidopsis and Dictyostelium cellsFinally, we investigated the distribution of actin fused with GFP in cells of phylogenetically distant species, that is, in the cellular slime mold Dictyostelium discoideum and the higher plant Arabidopsis thaliana, and whether the phenomenon observed in mammalian cells was conserved across such widely divergent eukaryotic cells. As in mammalian cells, GFP-ACT7 expressed in Arabidopsis T87 cells was incorporated into distinct actin bundles (Fig. 5A, left). In contrast, filaments of ACT7-GFP were unrecognizable because of high cytoplasmic fluorescence (Fig. 5A, middle). Nonetheless, treatment of those cells with the actin-polymerizing drug jasplakinolide induced the formation of numerous short actin bundles, indicating that ACT7-GFP is polymerization competent (Fig. 5A, right).
In Dictyostelium, GFP-actin has been widely used for the visualization of the actin cytoskeleton since the early days of the development of GFP technology (Aizawa et al., 1997; Westphal et al., 1997). In the pioneering report as an unpublished observation, actin-GFP displayed only weak signal and no specific structures were visualized in the cells (Westphal et al., 1997). We then tried to establish Dictyostelium cells expressing actin-GFP protein. As shown in Fig. 5B, cells expressing actin-GFP displayed low fluorescence when compared with cells expressing GFP-actin. Contrary to expectations, however, differences in the localization of actin-GFP were not observed as in mammalian and plant cells. These results suggest that GFP-actin affects the assembly of long bundled actin filaments, such as stress fibers in mammalian cells and F-actin bundles in plant cells, as shown in Fig. 1A and Fig. 5A, but does not affect the polymerization of shorter actin filaments found in the crowns, cup-shaped actin-containing structures implicated in macropinocytosis, of Dictyostelium cells.
Distribution of GFP-actin and actin-GFP in plants and Dictyostelium cells. (A) Arabidopsis T87 cells expressing fusion proteins of GFP and actin. Scale bars represent 100 μm. (B) Dictyostelium cells expressing GFP-actin (left) and actin-GFP (right). Scale bar represents 10 μm.
Several microscopic tools have been developed to observe the dynamics of the actin cytoskeleton. Until the advent of GFP technology, phalloidin-labeled fluorescent dyes, such as rhodamine-phalloidin, were used widely to observe the cytoskeleton but could only be used in fixed cells. The utilization of GFP fusion proteins enabled the visualization of the actin cytoskeleton in living cells, organs, and whole organisms in real time (Chalfie et al., 1994); however, the effects of the position of the GFP tag on actin functions were not sufficiently considered.
In this study, we expressed actin fused with EGFP at either the N- or C-terminus; we also explored the differential effects of fusion protein geometry on the behavior of both the labeled actin and the cells expressing the labeled actin. Our investigation revealed that EGFP fused to the N-terminus of actin augments the formation of F-actin, whereas EGFP fused to the C-terminus of actin does not significantly alter the cellular G-actin:F-actin ratio. EGFP-β-actin and β-actin-EGFP used in the above observations had different linker sequences between actin and EGFP. However, EGFP-β-actin with exactly the same linker sequence as β-actin-EGFP did not show the high background in cytosol (Supplementary Fig. 3A and B), demonstrating that the difference in the linker sequence was not the cause of different polymerization properties of the two actin probes. Furthermore, we confirmed that overexpression of exogenous actin does not affect cellular G-actin:F-actin ratio (Supplementary Fig. 4). Our finding is consistent with reports that the expression of EGFP-actin can influence cellular processes, cytoskeleton rearrangement and adhesion dynamics (Deibler et al., 2011; Feng et al., 2005). To avoid such adverse effects of expressing EGFP-actin, EGFP fusion probes were developed using the actin-binding domains of various F-actin-binding proteins, such as Lifeact and others (Belin et al., 2014; Burkel et al., 2007; Pang et al., 1998; Riedl et al., 2008). These EGFP probes for actin detect only F-actin, like fluorescent phalloidin, in the absence of cytosolic G-actin staining. Furthermore, certain F-actin probes, such as Lifeact, do not significantly alter the cellular G-actin:F-actin ratio, as reported by Riedl et al. and confirmed in this study. Interpretation of cellular fluorescence requires caution, however, since certain F-actin probes, including Lifeact, do not stain all cellular F-actin with equal efficiency (Belin et al., 2014; Riedl et al., 2008; Uyeda et al., 2011).
As shown in this study, the expression of EGFP-actin probes in some cell lines resulted in the excessive formation of F-actin. The amount of cellular F-actin can be increased by different molecular mechanisms, including increased nucleation of filament polymerization, increased rate of polymerization, and decreased rate of filament severing and/or depolymerization. Among these possibilities, the fact that endogenous actin also tended to excessively polymerize in cells expressing EGFP-actin is inconsistent with the second possibility, because the polymerization competence of EGFP-actin is unlikely to promote the polymerization of endogenous actin in the same cell. In fission yeast, EGFP-actin is hardly incorporated into certain actin-containing structures that are dependent on the activity of formin (Wu and Pollard, 2005), arguing against the first possibility. We thus favor the possibility that chimeric F-actin, containing both EGFP-actin and endogenous actin, either depolymerizes more slowly than does normal F-actin or is more resistant to the severing activity of cofilin. Further studies are needed to examine these possibilities experimentally.
We thank Dr. Sam-Geun Kong of Kongjyu University for his support in plant cell culture. We also thank Dr. Reiko Nagasaki for proofreading and for the critical discussions. This work was supported in part by Grants-in-aid from the Ministry of Education, Culture, Sports, Science and Technology (No. 24117008). The authors also thank RIKEN CELL BANK and NBRP for providing cell lines used in this study.
