The non-muscle α-actinin isoforms (actinin-1 and -4) are closely related dimeric actin filament cross-linking proteins. Despite high sequence similarity, unique properties have been ascribed to actinin-4 in particular. For example, actinin-4, but not actinin-1, is essential for normal glomerular function in the kidney, is overexpressed in several cancers and can translocate to the nucleus to regulate transcription. To understand the molecular basis for such isoform-specific functions we have, for the first time, comprehensively compared these proteins in terms of alternative splicing, actin-binding properties, heterodimer formation and molecular interactions. We find that the Ca2+-insensitive variant of actinin-4 is expressed only in the nervous system and thus cannot be regarded as a smooth muscle isoform, as is the case for the Ca2+-insensitive variant of actinin-1. The actin-binding properties of actinin-1 and -4 are similar and are unlikely to explain isoform-specific functions. Surprisingly, we reveal that actinin-1/-4 heterodimers, rather than homodimers, are the most abundant form of actinin in many cell lines. Finally, we use a proteomics approach to identify potential isoform-specific interactions. The results of the present study indicate that actinin-1 and -4 can readily form heterodimers composed of monomers that may have different properties and interacting proteins. This significantly alters our view of non-muscle actinin function.

INTRODUCTION

The α-actinins are a major family of actin filament cross-linking proteins. Actinins are dimeric proteins with each monomer consisting of an N-terminal ABD (actin-binding domain), a rod domain composed of four spectrin-like repeats (R1–R4) and a C-terminal CaM (calmodulin-like) domain. Actinin-1 and -4 are regarded as non-muscle isoforms [1]. They are distinguished from muscle actinins by alternative splicing of exon 19, encoding part of the first EF-hand motif of the CaM domain. For actinin-1, splicing of exons 19a and 19b generates isoforms that bind actin in either a Ca2+-sensitive or -insensitive manner respectively (Figure 1A) [2]. This splicing is usually mutually exclusive and the resulting isoforms are commonly referred to as ‘non-muscle’ and ‘smooth muscle’ actinin respectively. Additionally, a brain-specific actinin-1 variant containing both exons 19a and 19b has been described in the rat brain [3]. For clarity we refer to actinin splice variants according to the exons they contain (19a, 19b or 19a+b) rather than their expression patterns. Actinin-4 also has alternate versions of exon 19 that are predicted to confer Ca2+ sensitivity and insensitivity[4], but the expression patterns of the actinin-4 exon 19 splice variants has not been examined. In addition, actinin-4 exhibits alternative splicing of exon 8 (Figure 1B). On the basis of structural studies, alternative splicing of exon 8 may alter actin-binding properties, but this has not been thoroughly investigated [5]. Different splicing patterns or actin-binding characteristics may explain the functional differences between actinin-1 and -4, but these properties of the non-muscle actinins have not been comprehensively compared.

Analysis of the alternative splicing patterns of actinin-1 and -4 exons 8 and 19

Figure 1
Analysis of the alternative splicing patterns of actinin-1 and -4 exons 8 and 19

(A) Schematic depiction of the alternative splicing of exon 19 that encodes part of the first EF-hand of the non-muscle actinins. Mutually exclusive splicing results in the inclusion of either exon 19a or exon 19b to generate Ca2+-sensitive and -insensitive isoforms respectively (black lines). Inclusion of both exons in the mature mRNA transcript can also occur for actinin-1 (grey lines) [3]. The primers flanking exon 19 were used for rtPCR and are indicated by arrows. (B) Schematic depiction of mutually exclusive alternative splicing of exon 8 in actinin-4. Primer pairs in which one primer was specific for either exon 8a or 8b were used for rtPCR and are indicated by arrows. (C) rtPCR analysis of the alternative splicing patterns of actinin-1 and -4 exon 19 and actinin-4 exon 8 that occur in various murine tissues as indicated. The 15 bp size difference between exon 19a and 19b allows these two splice variants to be distinguished. -ve control, negative control; EF1a, elongation factor-1a. (D) rtPCR analysis of the alternative splicing patterns of actinin-1 and -4 exon 19 and actinin-4 exon 8 that occur in normal human brain compared with a panel of human glioblastoma cell lines. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Figure 1
Analysis of the alternative splicing patterns of actinin-1 and -4 exons 8 and 19

(A) Schematic depiction of the alternative splicing of exon 19 that encodes part of the first EF-hand of the non-muscle actinins. Mutually exclusive splicing results in the inclusion of either exon 19a or exon 19b to generate Ca2+-sensitive and -insensitive isoforms respectively (black lines). Inclusion of both exons in the mature mRNA transcript can also occur for actinin-1 (grey lines) [3]. The primers flanking exon 19 were used for rtPCR and are indicated by arrows. (B) Schematic depiction of mutually exclusive alternative splicing of exon 8 in actinin-4. Primer pairs in which one primer was specific for either exon 8a or 8b were used for rtPCR and are indicated by arrows. (C) rtPCR analysis of the alternative splicing patterns of actinin-1 and -4 exon 19 and actinin-4 exon 8 that occur in various murine tissues as indicated. The 15 bp size difference between exon 19a and 19b allows these two splice variants to be distinguished. -ve control, negative control; EF1a, elongation factor-1a. (D) rtPCR analysis of the alternative splicing patterns of actinin-1 and -4 exon 19 and actinin-4 exon 8 that occur in normal human brain compared with a panel of human glioblastoma cell lines. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

The actinin rod domain is responsible for dimerization and forms a spacer of fixed length between the ABDs in the resulting antiparallel dimer [68]. Heterodimer formation has been reported to occur between the muscle actinins (actinin-2 and -3) [9]; however, this has not been examined for the non-muscle isoforms. Thus actinin-1 and -4 are generally regarded as being distinct homodimeric entities, despite the fact that they are co-expressed in many tissues and cell types. A secondary function of the rod domain is to act as a binding site for other proteins. A multitude of interacting proteins have been described for actinins and these bind to the rod as well as the ABD and CaM domains [1,10,11]. Although many actinin-binding partners are likely to be common to multiple actinin isoforms, others may be isoform-specific. However, in many studies the issue of the isoform specificity of actinin ligands has not been tested. Such isoform-specific interactions are likely to confer unique functional properties to different actinins and thus merit a more systematic examination.

Actinin-1 and -4 are very closely related, sharing a 87% amino acid identity[12]. Despite this high degree of sequence similarity, numerous differences in cellular localization and function of the non-muscle actinins have been reported and a number of unique properties have been ascribed to actinin-4 in particular. For example, actinin-4 is the predominant actinin isoform reported to be associated with cancer. Elevated levels of actinin-4 protein are found in a number of cancers, including ovarian [13,14], colorectal [15], pancreatic [16] and astrocytoma [17], and are associated with high-grade tumours and poor patient outcomes [14,16,17]. Furthermore actinin-4 expression promotes cell motility [12,13,17,18] and enhances the metastatic potential of colorectal cancer cells [15]. The functions of actinin-4 in cancer cells are likely to be caused by characteristics of actinin-4 that are not shared by actinin-1. Another actinin-4-specific function is its role in the formation and function of glomeruli in the kidney. Genetic studies identified point mutations in the actinin-4 gene that cause focal segmental glomerulosclerosis [19]. In agreement with this, mice deficient in actinin-4 exhibit podocyte foot-process enfacement and glomerular disease, even though these mice express normal levels of actinin-1 [20]. This suggests that, in the kidney at least, actinin-1 and -4 are not functionally redundant. An additional unique function of actinin-4 is its ability to translocate to the nucleus where it can act as a transcriptional regulator [2124]. The molecular basis for most of these isoform-specific functions of actinin-4 is unclear.

In an effort to understand the molecular mechanisms underlying these unique functions of actinin-4 we compared the alternative splicing, actin-binding affinities, dimerization properties and interacting partners of the non-muscle actinins. We find that the affinity and Ca2+-sensitivity of actin filament binding for the exon 19a variants of actinin-1 and -4 are very similar, but that the expression pattern of the Ca2+-insensitive exon 19b variants differs significantly for actinin-1 and -4. We also describe the actin-binding properties of the brain-specific actinin-1 and actinin-4 splice variants for the first time. Our examination of actinin dimerization reveals, very surprisingly, that actinin-1/-4 heterodimers, rather than homodimers, are the most abundant actinin species in many cell lines. To examine the molecular interactions of actinin in a more systematic manner we have performed yeast two-hybrid screens and analysed affinity-purified actinin-1 and -4 protein complexes to identify putative isoform-specific interacting proteins. This thorough analysis of actinin-1 and -4 provides the basis for further investigation and comparison of the diverse functions of the non-muscle actinins.

MATERIALS AND METHODS

Antibodies, cell lines and reagents

The following antibodies were used: anti-α-Actinin-1 [catalogue number (H-2):sc17829, Santa Cruz Biotechnology], anti-α-actinin-4 (catalogue number IG701, Immunoglobe), anti-α-actinin-2/3 (catalogue number A7811, clone EA-53, Sigma), anti-GFP (green fluorescent protein; catalogue number MAB3580, Millipore), anti-GFP (catalogue number ab290, mAbcam) and anti-FLAG (catalogue number F3165, Sigma). Chemicals were obtained from Sigma unless otherwise stated and restriction enzymes were purchased from New England Biolabs. The cell lines used included A172, U-87MG, U-373MG (Cell Line Services), HEK (human embryonic kidney)-293, MDA-MB-231, DU145, HeLa and MCF-7 (gift from Professor Rosemary O’Connor, University College Cork, Cork, Ireland).

rtPCR (reverse transcription PCR)

Tissues were dissected from C57 BL6J mice and embryos. Animal experiments at University College Cork were approved by the University Ethics Committee and conducted under a license from the Irish Department of Health and Children. RNA was extracted using TriPure Isolation Reagent (Roche Applied Science) according to the manufacturer's instructions. A total of 1 μg of RNA was used per cDNA reaction using the Protoscript AMV First Stand cDNA synthesis kit (New England Biolabs) and 1 μl of cDNA was used per rtPCR. For rtPCR analysis of actinin-4 exon 8, the forward primers that flank exon 8 were used in conjunction with reverse primers specific for exon 8b and reverse primers that flank exon 8 were used in conjunction with forward primers specific for exon 8a. For analysis of actinin-1 and -4 exon 19, primers that flank this exon were used and bands corresponding to exon 19a (128 bp) and 19b (113 bp) were separated on 4% agarose gels. The sequences of primers used for rtPCR are provided in Supplementary Table S1 (at http://www.biochemj.org/bj/452/bj4520477add.htm).

cDNA constructs, protein expression and purification

Details of all cDNA constructs and the methods used for recombinant protein expression and purification are provided in the Supplementary Online Data (at http://www.biochemj.org/bj/452/bj4520477add.htm).

Actin co-sedimentation assays

Human platelet actin (Cytoskeleton) was mixed in G-actin (globulin actin) buffer [5 mM Tris (pH 8.0), 0.2 mM MgCl2, 0.2 mM ATP and 0.5 mM DTT (dithiothreitol)]. Actin and actinin proteins were cleared by ultracentrifugation at 50000 rev./min for 30 min at 4°C using a TLA55 rotor and a Beckman Coulter Optima MAX ultracentrifuge. Actin was polymerized by the addition of 1/100 volume of 100× polymerization buffer (2 M NaCl and 0.1 M MgCl2) and incubated for 30 min at 4°C. Actin and actinin proteins were mixed in 10 mM Tris/HCl (pH 7.5), 100 mM NaCl, 1 mM NaN3, 1 mM MgCl2, 0.1 mM ATP and 0.1 mM DTT and incubated for 30 min at 30°C. To determine the Kd value for actin binding, 2 μM actin was used per assay along with a range of actinin concentrations (0.25–30 μM) in the presence of 0.2 mM EGTA. For the Ca2+-sensitivity analysis 2 μM actin and 2.5 μM actinin were mixed in a variety of free Ca2+ concentrations (100 nM–1 mM) that were obtained using a Ca2+:EGTA buffer system. Polymerized actin was separated by ultracentrifugation at 50000 rev./min for 30 min at 30°C using a TLA55 rotor. The pellets and supernatants were brought to the same total volume of SDS sample buffer, boiled and equal volumes loaded on to SDS/PAGE (12% gels). Assays that omitted actin were used as controls for the non-specific trapping of actinin. For the actin-bundling assays ultracentrifugation was performed at 10 400 rev./min using an F45-24-11 rotor and an Eppendorf 5415 centrifuge.

Analysis of evolutionary conservation of the actinin rod dimerization interface

This analysis was based on the human actinin-2 rod domain for which a crystallographic structure is known [7]. A multiple sequence alignment of this sequence with actinin-1 and -4 sequences from divergent species (human, mouse, frog, zebrafish and chicken) was generated. With this alignment set as the input, the ConSurf server was used to calculate and plot conservation scores on to the actinin-2 rod domain three-dimensional structure [25,26]. Conservation scores that are considered unreliable by ConSurf are coloured yellow and shown in stick rather than space filling representation.

Yeast two-hybrid analysis

Rod domains from actinin-1, -2 and -4 were cloned into both bait and prey vectors as described in the Supplementary Online Data. To assess actinin rod domain heterodimerization, bait and prey constructs were co-transformed into the Saccharomyces cerevisiae L40 strain and activation of the lacZ and HIS3 (histidine 3) reporter genes assayed as described previously [8]. Yeast two-hybrid screens using either actinin-1 or -4 rod domain pLEX-K bait constructs were performed using a P3 mouse brain cDNA library in the pAD-GAL4 vector as described previously [27]. The transformation efficiency for the actinin-1 and actinin-4 screens was 2.4×106 and 6×105 colony forming units respectively. Prey clones identified from the actinin-1 screen were tested for interaction with the actinin-4 bait by co-transforming both plasmids into L40 cells.

In vitro heterodimer-binding assays

GST (glutathione transferase)-tagged ABD-R2 and His6-tagged R3-CaM regions from both actinin-1 and -4 were expressed, purified and dialysed into binding assay buffer [20 mM Tris (pH 7.5), 50 mM NaCl and 5 mM 2-mercaptoethanol]. Experiments were designed to assay for homodimer and heterodimer formation. Briefly, 0.1 mg of GST-tagged ABD-R2 domains of actinin-1 or -4 were incubated with 50 μl of GST beads which were subsequently mixed with 0.1 mg of R3-CaM domains of actinin-1 or -4. Beads were incubated for 30 min at 4°C and washed three times in binding assay buffer for 5 min at 4°C. Proteins were eluted in 50 μl of 2× SDS/PAGE loading dye. Eluted proteins were run on SDS/PAGE (12% gels) and stained with Coomassie Blue. Densitometric analysis was used to compare the levels of R3-CaM proteins bound to ABD-R2 proteins for heterodimer and homodimer formation.

Cell culture and transfections

A172, HeLa, HEK-293, MCF-7, MDA-MB-231 and U-87MG cells were cultured in DMEM (Dulbecco's modified Eagle's medium; catalogue number D6429, Sigma), 10% FBS (fetal bovine serum), 1% penicillin/streptomycin and 1% L-glutamine. DU145 cells were cultured in RPMI 1640 medium (catalogue number R8758, Sigma), 10% FBS, 1% L-glutamine and 1% penicillin/streptomycin. U-118MG and U-373MG cells were cultured in Minimum Essential Medium Eagle (catalogue number M5650, Sigma), 10% FBS, 1% L-glutamine, 1% penicillin/streptomycin, 0.1 mM non-essential amino acids and 1 mM sodium pyruvate. Serum starvation was carried out by growing a confluent cell monolayer in medium containing 10% FBS. Cells were washed twice in PBS and grown for a further 48 h in the absence of FBS. Cell migration was induced by applying multiple scratches to a confluent cell monolayer using a sterile tip ~16 h preharvest. Confluent cell layers were obtained by seeding 800000 cells in a 10-cm-diameter plate 48 h preharvest. Proliferating cells were obtained by seeding 200000 cells ~16 h preharvest.

siRNA (small interfering RNA) knockdown of actinin-4

MCF-7 cells were seeded in six-well dishes so as to be 50% confluent for transfection. Two Silencer Select® Pre-designed siRNAs targeting actinin-4 (s959 and s960, catalogue number 4427037) together with a non-targeting negative control (catalogue number 4390843) were purchased from Ambion (Life Technologies). Cells were transfected in serum-free medium with 10 nM siRNA using 3 μl of Lipofectamine™ purchased from Invitrogen (Life Technologies). Transfections were conducted in 1 ml of DMEM. At 5 h post-transfection the medium was adjusted to 10% FBS. Cells were harvested 72 h post-transfection.

Native PAGE

A standard PAGE protocol was followed with the exception that SDS, 2-mercaptoethanol and the boiling steps were omitted. The PROTEAN II xi Cell electrophoresis instrument was used (Bio-Rad Laboratories). Cells were washed twice in PBS and lysed in 150 mM NaCl, 50 mM Tris/HCl (pH 7.5), 1% NP40 (Nonidet P40), 10% glycerol, 2 mM NaF, 1 mM Na3VO4 and 1 protease inhibitor cocktail tablet/10 ml (Roche). Cells were incubated in lysis buffer for 30 min and centrifuged at 16100 g for 15 min at 4°C. PAGE loading buffer (3×, without SDS and 2-mercaptoethanol) was added to the lysates on ice. Samples were then loaded on to 6% native PAGE (without SDS). Lysates were run for 5 h in PAGE running buffer (without SDS) at a current of 24 mA. The running buffer was refreshed after 2.5 h. Western blot detection was performed on an Odyssey Classic infrared scanner (LI-COR Biosciences).

Determination of the relative amount of actinin-1 and -4 within and between cell lines

Full-length actinin-1/-4 constructs in the pEGFP-C2 vector (Clontech) were transfected into HEK-293 cells using Lipofectamine™ transfection reagent (Life Technologies). In order to determine their expression levels relative to each other Western blots of lysates were probed with an anti-GFP antibody. Subsequently lysates from cells transfected with these GFP-tagged actinin constructs were used to produce a standard curve to which specific actinin-1 and -4 antibody staining could be normalized. Using this standard curve in combination with lysates from various cell lines allowed comparison of the relative amount of actinin-1 and -4 within and between cell lines. Probing native gels with antibodies specific for actinin-1 and -4 followed by densitometric analysis allowed us to quantify the proportion of actinin-1 and -4 involved in heterodimer formation.

TAPs (tandem affinity purifications)

Constructs encoding TAP-tagged actinin-1, -4 and YFP (yellow fluorescent protein) were transfected into HEK-293 cells and grown for four cell passages. Using a fluorescent microscope, YFP/GFP-expressing cells were selected over the course of ~10 passages. In this manner stable cell pools in which >70% of cells were YFP/GFP-positive were obtained. A total of five 15-cm-diameter cell culture dishes were grown to confluence for each round of purification. Cells were washed twice in PBS and lysed in 5 ml of lysis buffer [150 mM NaCl, 50 mM Tris/HCl (pH 7.5), 1% NP40, 10% glycerol, 2 mM NaF, 1 mM Na3VO4 and 1 protease inhibitor cocktail tablet/10 ml] for 30 min and centrifuged at 16100 g for 15 min at 4°C. Lysates were run five times through columns containing 1 ml of IgG beads. Beads were washed three times in 10 ml of lysis buffer. TEV (tobacco etch virus) cleavage buffer (2 ml, lysis buffer plus 0.5 mM EDTA and 1 mM DTT) and TEV protease (0.1 mg/ml) was added to the columns. Columns were incubated for 1.5 h at room temperature (22°C), mixing occasionally. The eluate was adjusted to 2 mM CaCl2 and added to 300 μl of CaM–Sepharose beads. The beads were incubated for 1 h at 4°C and then washed three times in 1 ml of wash buffer (lysis buffer plus 2 mM CaCl2). The bound proteins were eluted in 300 μl of 2× SDS loading dye that was subsequently concentrated down to 50 μl. Eluted proteins were loaded on to SDS/PAGE (12% gels) and run for 2 h at 100V. Each sample lane was divided into six gel segments for MS analysis.

Protein identification by MS

MS analysis was performed at the FingerPrints Proteomics facility at University of Dundee, Dundee, Scotland. Peptides were obtained using an in-gel digestion protocol and extracted prior to analysis by one-dimensional nLC (nano liquid chromatography)-MS/MS (tandem MS) using an LTQ Orbitrap Velos Pro mass spectrometer (Thermo Scientific). MS/MS data were searched against the IPI (International Protein Index)-human database (91464 sequences. 36355611 residues; http://www.ebi.ac.uk) using in-house Mascot software (Matrix Science). Identified proteins were ranked according to Mascot protein scores and listed using protein symbols as identifiers in Excel software. The COUNTIF function of Excel was used to identify proteins present in one or both purified actinin complexes, but not in the control sample. A Mascot protein score of 100 was then applied as a cut-off value to limit results to proteins that had been very reliably identified. This roughly corresponds to proteins with two distinct peptide matches. In the case of proteins present in both actinin complexes, at least one of the protein scores had to be greater than 100. Probable environmental contaminants, such as keratins, were then removed. Information about the subcellular localization and the functions of the remaining proteins was obtained from the UniProt database (http://www.uniprot.org). Proteins that were likely to be false positives on the basis of clear subcellular localizations that differ from the cytoplasmic and nuclear localizations that have been reported for actinins were then eliminated [secreted, mitochondrial matrix and ER (endoplasmic reticulum)/Golgi lumenal proteins] to obtain the final filtered list of identified proteins.

RESULTS

Actinin-1 and -4 show tissue-specific differences in alternative splicing patterns

Although the splicing of exon 19 in actinin-1 has been well characterized, the expression patterns of alternatively spliced actinin-4 isoforms have not been described in detail. Since splicing could confer unique properties to actinin-4, we employed rtPCR to address this deficit in our knowledge. We first compared the tissue specificity of exon 19 splicing between actinin-1 and -4 by examining a panel of murine tissues. As reported previously in the rat [3], the Ca2+-insensitive exon 19b variant of actinin-1 was expressed in the maturing brain and cardiac and skeletal muscle, and in smooth muscle containing tissues such as the stomach, intestine and bladder, whereas the brain-specific exon 19a+b variant was expressed postnatally in the brain and adult spinal cord (Figure 1C). Actinin-1 exon 19a predominated in other tissues and in the immature brain (Figure 1C). For actinin-4, expression of the exon 19a variant was even more widespread and this was the only isoform detected in non-neuronal tissues (Figure 1C). In the brain the actinin-4 exon 19a and 19b variants were equally abundant at all of the developmental stages examined and no 19a+b isoform was detected (Figure 1C). Thus in contrast with actinin-1, it is the Ca2+-sensitive exon 19a version of actinin-4 that is expressed in skeletal and smooth muscle, whereas the Ca2+-insensitive exon 19b variant is restricted to neural tissues.

Actinin-4, unlike actinin-1, also exhibits alternative splicing of exon 8. The expression of actinin-4 exon 8 variants was examined by rtPCR using primers specific against exon 8a and 8b. The 8a variant was detected in all tissues examined (Figure 1c). By contrast the 8b variant was restricted to the brain, spinal cord, skeletal and cardiac muscle, and smooth muscle-rich tissues. We next wanted to assess the relative expression of the 8a and 8b exons in some of these tissues. Since the PCR products obtained with different exon-specific primers cannot be directly compared, we instead used common flanking primers to amplify the exon 8 region of actinin-4 from P28 (postnatal day 28) brain, kidney, lung, intestine, heart and skeletal muscle and directly sequenced several independent PCR products for each tissue. Examination of the sequencing chromatographs revealed that, whereas exon 8b was by far the predominant variant expressed in the brain, it was a very minor or undetectable component compared with exon 8a in the other tissues (results not shown).

Since our analysis in mice revealed that both of the non-muscle actinins are expressed as multiple splice variants in neuronal tissues we wanted to establish if these splicing patterns were conserved in humans and whether alterations in splicing might occur in glioblastoma cells, given that actinin-4 is up-regulated in glioblastomas and other cancers. rtPCR from normal adult human brain tissue showed that alternative splicing patterns for actinin-1 and -4 exon 19 and actinin-4 exon 8 are conserved between the murine and human brain (Figure 1D). Thus the actinin-1 exon 19a+b and 19b variants and actinin-4 exon 19a, 19b, 8a and 8b variants were detected, with exon 8b being the predominant actinin-4 exon 8 variant present as assessed by direct sequencing of the PCR products generated with flanking primers. In the four glioblastoma cell lines examined, we found that there is a switch from exon 19b (Ca2+ insensitive) containing splice variants seen in normal brain to exclusively exon 19a (Ca2+ sensitive) variants of both actinin-1 and -4 (Figure 1D). These cell lines appeared to express exclusively the actinin-4 exon 8a variant as opposed to exon 8b that predominates in normal tissue (Figure 1D).

Actinin-1 and -4 have similar actin-binding properties

The actin-binding properties of actinin-1 and -4 have not been directly compared. Since differences between the proteins in this regard might underlie distinct functions, we employed actin co-sedimentation assays to examine these properties. We first compared the affinity of actin binding for the most widely expressed Ca2+-sensitive actinin-1 (exon 19a) and actinin-4 (exon 8a and 19a) splice variants. In addition, actinin-4 (exon 8b and 19a) was also examined to determine whether the inclusion of the variant exon 8b modulated the actin-binding properties. Full-length actinins were recombinantly expressed in Escherichia coli cells to ensure that pure homodimeric proteins were obtained and the affinity tag used for purification was removed to prevent interference with actin binding. Supernatants and pellets from co-sedimentation assays were analysed on Coomassie Blue-stained polyacrylamide gels and the proportion of actinin in each fraction was quantified (Figure 2A). A single ligand-binding site was assumed and rectangular hyperbolic curves were fitted to plots of bound against free actinin in order to calculate the dissociation constants (Kd). The Kd values calculated for actinin-1 (exon 19a), actinin-4 (exon 8a and 19a) and actinin-4 (exon 8b and 19a) were 1.93±0.56 μM, 2.96±0.38 μM and 3.96±1.19 μM respectively (Figure 2B). The differences in the calculated Kd values between these isoforms did not reach statistical significance (Student's t test). It appears therefore that the non-muscle actinins differ only very slightly in their binding affinity for actin filaments.

Comparison of the actin-binding affinities and Ca2+-sensitivities of actinin-1 and -4

Figure 2
Comparison of the actin-binding affinities and Ca2+-sensitivities of actinin-1 and -4

(A) Representative actin-binding assay for actinin-1 (exon 19a). Increasing concentrations of actinin as indicated were incubated with F-actin and subjected to ultracentrifugation at 50000 rev./min. Bound and free actinin were quantified from Coomassie Blue-stained SDS/PAGE of the pellet and supernatant samples. (B) Calculation of the actin-binding affinity of actinin-1 (exon 19a), actinin-4 (exon 8a and 19a) and actinin-4 (exon 8b and 19a). Representative plots of bound compared with free actinin are shown. The indicated Kd values for the interaction of each isoform with actin were calculated from data from ≥three independent assays. (C) Comparison of the Ca2+-sensitivity of actin binding between actinin-1 (exon 19a) and actinin-4 (exon 8a and 19a). Fixed concentrations of F-actin and actinin were incubated in the presence of the indicated free Ca2+ concentrations. Pellet samples were analysed by Coomassie Blue-stained SDS/PAGE. (D) Graph of bound actinin against Ca2+ concentration was plotted to analyse the Ca2+-sensitivity of actin binding for actinin-1 (exon 19a) and actinin-4 (exon 19b). Error bars represent the S.E.M. (E) Graph of bound actin compared with Ca2+ concentration was plotted to analyse the Ca2+-sensitivity of actin bundling for several non-muscle actinin isoforms. Bundling assays were performed in a similar manner to the actin-binding assays, but with centrifugation at 10000 g.

Figure 2
Comparison of the actin-binding affinities and Ca2+-sensitivities of actinin-1 and -4

(A) Representative actin-binding assay for actinin-1 (exon 19a). Increasing concentrations of actinin as indicated were incubated with F-actin and subjected to ultracentrifugation at 50000 rev./min. Bound and free actinin were quantified from Coomassie Blue-stained SDS/PAGE of the pellet and supernatant samples. (B) Calculation of the actin-binding affinity of actinin-1 (exon 19a), actinin-4 (exon 8a and 19a) and actinin-4 (exon 8b and 19a). Representative plots of bound compared with free actinin are shown. The indicated Kd values for the interaction of each isoform with actin were calculated from data from ≥three independent assays. (C) Comparison of the Ca2+-sensitivity of actin binding between actinin-1 (exon 19a) and actinin-4 (exon 8a and 19a). Fixed concentrations of F-actin and actinin were incubated in the presence of the indicated free Ca2+ concentrations. Pellet samples were analysed by Coomassie Blue-stained SDS/PAGE. (D) Graph of bound actinin against Ca2+ concentration was plotted to analyse the Ca2+-sensitivity of actin binding for actinin-1 (exon 19a) and actinin-4 (exon 19b). Error bars represent the S.E.M. (E) Graph of bound actin compared with Ca2+ concentration was plotted to analyse the Ca2+-sensitivity of actin bundling for several non-muscle actinin isoforms. Bundling assays were performed in a similar manner to the actin-binding assays, but with centrifugation at 10000 g.

The Ca2+-sensitivity of actin binding was also evaluated. Comparison of the actinin-1 (exon 19a) and actinin-4 (exon 8a and 19a) splice variants revealed very similar Ca2+ sensitivities with actin- binding decreasing by 50–60% at free Ca2+ concentrations above 10 μM (Figures 2C and 2D). There was also a dramatic decrease in actin bundling at these free Ca2+ concentrations, that again was similar for both actinins (Figure 2E). By comparison actin bundling by exon 19b-containing isoforms of both actinins was relatively insensitive to Ca2+ at concentrations likely to occur in the cytosol (Figure 2E). The Ca2+ sensitivity of actin binding was also examined for actinin-4 containing exon 8b and the brain-specific actinin-1 splice variant containing both exons 19a and 19b. Both of these isoforms were Ca2+ sensitive and displayed decreases in actin binding of approximately 60% and 40% respectively at high free Ca2+ concentrations (results not shown). Overall the intrinsic actin-binding properties of actinin-1 and -4 are very similar and seem unlikely to account for major functional differences between these proteins.

Non-muscle actinins have the potential to form heterodimers

It is unknown whether the non-muscle actinins have the ability to form heterodimers either with each other or with the muscle actinins. Dimerization of actinins is largely mediated by the four spectrin-like repeats of the rod domain. To explore the potential for heterodimer formation we first examined the evolutionary conservation of the amino acids present at the dimer interface, on the basis of the known crystal structure of the human actinin-2 rod domain [7]. Conservation scores based on an alignment of actinin-1 and -4 sequences from divergent species were calculated and plotted on to the actinin-2 rod domain three-dimensional structure using the ConSurf server [25,26]. This analysis demonstrates that the dimer interface is highly conserved between actinin-1, -2 and -4 (Figure 3A), with very few non-conservative amino acid substitutions. By contrast the exposed surface of the rod shows significantly less sequence conservation. This supports the idea that the dimer interface is almost completely conserved between actinins and that heterodimer formation could occur.

Analysis of the propensity for non-muscle actinins to form heterodimers

Figure 3
Analysis of the propensity for non-muscle actinins to form heterodimers

(A) Evolutionary conservation of the amino acids in the actinin-1, -2 and -4 rod domains. Conservation scores are plotted on to one subunit of the dimeric actinin-2 rod domain structure using the ConSurf server [25,26] (backbone and side chains depicted as spheres). A backbone-only trace of the other subunit is shown. Two views of the structure are shown to highlight the conserved dimer interface (left-hand panel) and less conserved exposed surface of the rod (right-hand panel). Conservation scores are visualized on a scale of nine grades with most variable positions coloured turquoise and most conserved positions coloured maroon. The N-terminus (R1) of the coloured subunit is to the left. (B) In vitro analysis of actinin-1 and -4 homodimeric compared with heterodimeric interactions. GST-tagged actinin-1/-4 ABD-R2 domains were incubated with His6-tagged actinin-1/-4 R3-CaM domains and bound to glutathione–Sepharose. After washing, the eluted proteins were analysed on Coomassie Blue-stained SDS/PAGE. (C) Bound R3-CaM domains were quantified by densitometry and the amount of bound R3-CaM for heterodimeric interactions was expressed as a percentage of that observed for the equivalent homodimeric combination of proteins. Error bars represent the S.E.M.

Figure 3
Analysis of the propensity for non-muscle actinins to form heterodimers

(A) Evolutionary conservation of the amino acids in the actinin-1, -2 and -4 rod domains. Conservation scores are plotted on to one subunit of the dimeric actinin-2 rod domain structure using the ConSurf server [25,26] (backbone and side chains depicted as spheres). A backbone-only trace of the other subunit is shown. Two views of the structure are shown to highlight the conserved dimer interface (left-hand panel) and less conserved exposed surface of the rod (right-hand panel). Conservation scores are visualized on a scale of nine grades with most variable positions coloured turquoise and most conserved positions coloured maroon. The N-terminus (R1) of the coloured subunit is to the left. (B) In vitro analysis of actinin-1 and -4 homodimeric compared with heterodimeric interactions. GST-tagged actinin-1/-4 ABD-R2 domains were incubated with His6-tagged actinin-1/-4 R3-CaM domains and bound to glutathione–Sepharose. After washing, the eluted proteins were analysed on Coomassie Blue-stained SDS/PAGE. (C) Bound R3-CaM domains were quantified by densitometry and the amount of bound R3-CaM for heterodimeric interactions was expressed as a percentage of that observed for the equivalent homodimeric combination of proteins. Error bars represent the S.E.M.

To investigate this possibility directly we tested the ability of actinin rod domains to interact with each other in yeast two-hybrid assays. Rod domains from actinin-1, -2 and -4 were cloned into both bait and prey vectors. Interactions between rod domains were monitored by activation of the HIS3 and LacZ reporter genes (Table 1). Activation of the HIS3 reporter was seen for every combination of rod domains indicating that they are all capable of forming both homo- and hetero-dimers. In all cases the reporter gene expression for heterodimers was comparable with, or greater than, that for homodimers. Activation of the LacZ reporter was more variable, perhaps reflecting differences in expression levels between constructs, but again some homodimeric and heterodimer interactions were detected. Overall, these assays suggest that actinin-1, -2 and -4 rod domains can indeed form heterodimers as well as homodimers.

Table 1
Analysis of actinin homodimer and heterodimer formation by yeast two-hybrid assay

The ability of actinin-1, -2 or -4 rod domains (baits) to interact with actinin-1, -2 or -4 rod domains (preys) was tested using the yeast two-hybrid system. Empty bait and prey vectors were included as negative controls. Interactions between bait and prey were indicated by expression of two reporter genes, His3 and LacZ coding for β-galactosidase (β-gal).

 Prey 
 Actinin-1 Actinin-2 Actinin-4 Empty vector 
Bait His3 β-gal His3 β-gal His3 β-gal His3 β-gal 
Actinin-1 ++ +++ – +++ – – – 
Actinin-2 +++ ++ +++ ++ +++ ++ – – 
Actinin-4 +++ – +++ – – – – 
Empty vector – – – – – – – – 
 Prey 
 Actinin-1 Actinin-2 Actinin-4 Empty vector 
Bait His3 β-gal His3 β-gal His3 β-gal His3 β-gal 
Actinin-1 ++ +++ – +++ – – – 
Actinin-2 +++ ++ +++ ++ +++ ++ – – 
Actinin-4 +++ – +++ – – – – 
Empty vector – – – – – – – – 

To obtain a more quantitative measure of the propensity for homo- compared with hetero-dimer formation we performed in vitro binding assays using purified proteins. To do this we used constructs encoding the N- and C-terminal halves of actinin-1 and -4 (ABD-R2 and R3-CaM). Since spectrin-like repeats R1 and R2 form an interaction surface with spectrin-like repeats R3 and R4 in the rod domain, splitting the rod in half in this manner allows us to compare the affinity of homo- compared with hetero-dimer formation for different actinins. We expressed and purified GST-tagged ABD-R2 and His6-tagged R3-CaM domains for both actinin-1 and -4. These fusion proteins were mixed in various combinations and the ability of R3-CaM to interact with ABD-R2 was assessed by purification of the complex using GST beads (Figure 3B). Densitometric analysis was used to compare the levels of R3-CaM proteins bound to ABD-R2 proteins in each assay. The results show less than 5% differences in bound R3-CaM in homo- compared with hetero-dimeric contexts indicating that non-muscle actinin homo- and hetero-dimers should form with approximately equal affinity (Figure 3C).

Non-muscle actinin heterodimers are prevalent in many cancer cell lines.

The results of the present study described so far demonstrate that the actinin-1 and -4 rod domains can form heterodimers in the yeast two-hybrid system and in vitro. To assess whether the non-muscle actinins form heterodimers in cells in which they are co-expressed, we used native PAGE to maintain non-denaturing conditions in order to preserve and detect heterodimers. We first examined HeLa cells that have endogenous expression of actinin-1 and -4. Western blotting with isoform-specific antibodies revealed that each protein migrated as two bands on native gels. The upper band for actinin-4 co-migrates with the lower band for actinin-1 and we take this band to represent actinin-1/-4 heterodimers and the other bands to represent homodimers. A prominent heterodimer band was observed for confluent, proliferating and migrating HeLa cells as well as in cells that had been deprived of serum (Figure 4A). Thus non-muscle actinin heterodimer formation occurs in HeLa cells grown under a variety of culture conditions. We also detected heterodimers in a number of other cancer cell lines, indicating that this phenomenon was not specific to HeLa cells (Figure 4B). Notably, MDA-MB-231 cells that express only actinin-4 do not show any heterodimer band verifying that the presence of this band is dependent on the presence of both actinin-1 and -4. To further demonstrate this point and to examine the dynamics of heterodimer formation we performed siRNA-mediated knockdown of actinin-4 in MCF-7 cells (Figure 4C). An ~80% knockdown of actinin-4 levels was achieved in these cells, resulting in a large decrease in both actinin-4 homo- and hetero-dimers. This was mirrored by decreased actinin-1 in the heterodimeric state and a consequent increase in actinin-1 levels in the homodimeric state. Taken together these results verify that these non-muscle actinin heterodimers are dependent on the co-expression of both actinins and that a reduction in the expression of either leads to a redistribution of the other towards the homodimeric state.

Actinin-1/-4 heterodimer formation in cultured cells

Figure 4
Actinin-1/-4 heterodimer formation in cultured cells

(A) Actinin heterodimers in HeLa cells grown under a variety of conditions. Heterodimers were detected by native PAGE and Western blotting using antibodies specific against actinin-1 (red) and actinin-4 (green). The upper red band represents actinin-1 present in homodimers, the lower red band represents actinin-1 present in heterodimers, the upper green band represents actinin-4 present in heterodimers and the lower green band represents actinin-4 present homodimers. The merged image shows heterodimer represented by an intermediate orange/yellow band. (B) Native PAGE detection of actinin homodimers and heterodimers present in a panel of cell lines. (C) Native PAGE analysis of lysates from MCF-7 cells following siRNA-mediated knockdown of actinin-4. (D) The relative amounts of actinin-1 and -4 present in the indicated cell lines as determined by SDS/PAGE and quantitative Western blotting. The ratio of actinin-1/actinin-4 is indicated each column. (E) Actinin-1/actinin-4 heterodimers expressed as a the proportion of total dimeric actinin for the indicated cell lines. The proportion of each isoform present as heterodimers was quantified by native PAGE and Western blotting and plotted on to the histogram of the relative actinin levels shown in (D). The percentage of heterodimer for each cell line is indicated above each column.

Figure 4
Actinin-1/-4 heterodimer formation in cultured cells

(A) Actinin heterodimers in HeLa cells grown under a variety of conditions. Heterodimers were detected by native PAGE and Western blotting using antibodies specific against actinin-1 (red) and actinin-4 (green). The upper red band represents actinin-1 present in homodimers, the lower red band represents actinin-1 present in heterodimers, the upper green band represents actinin-4 present in heterodimers and the lower green band represents actinin-4 present homodimers. The merged image shows heterodimer represented by an intermediate orange/yellow band. (B) Native PAGE detection of actinin homodimers and heterodimers present in a panel of cell lines. (C) Native PAGE analysis of lysates from MCF-7 cells following siRNA-mediated knockdown of actinin-4. (D) The relative amounts of actinin-1 and -4 present in the indicated cell lines as determined by SDS/PAGE and quantitative Western blotting. The ratio of actinin-1/actinin-4 is indicated each column. (E) Actinin-1/actinin-4 heterodimers expressed as a the proportion of total dimeric actinin for the indicated cell lines. The proportion of each isoform present as heterodimers was quantified by native PAGE and Western blotting and plotted on to the histogram of the relative actinin levels shown in (D). The percentage of heterodimer for each cell line is indicated above each column.

We next sought to quantify the relative abundance of homo- and hetero-dimers in the cell lines examined. We first quantified the relative amounts of each actinin in our panel of cell lines and calculated the ratio of actinin-1/actinin-4 for each of the cell types (Figure 4D; see the Materials and methods section for details). We then determined the proportion of each actinin that was present as a heterodimer on native gels and used this to calculate the overall percentage of actinin-1/-4 heterodimers for each cell line (Figure 4E). Surprisingly, in those cell lines in which heterodimers were identified, they were more abundant than either homodimeric species and consumed close to 50% of the total non-muscle actinin.

Screening approaches to identify interacting partners of actinin-1 and actinin-4

Given their similar actin-binding properties and propensity to form heterodimers, the isoform-specific functions of the non-muscle actinins are likely to be mediated by isoform-specific interactions with proteins other than actin filaments. Many actinin interacting proteins are known and in most cases for which it has been examined, these interactions are common to multiple actinins. In other cases isoform specificity has not been examined and thus very few validated isoform-specific interactions have been reported. We wanted to identify actinin-interacting proteins and establish the isoform specificity of novel or known actinin-interaction partners in a more systematic manner. We first employed the yeast two-hybrid system. Although full-length proteins could not be used as baits for screening owing to autoactivation of the reporter genes, the rod domains of both actinin-1 and -4 were suitable. Screening a mouse brain cDNA library with the actinin-1 rod domain yielded ten positive clones; whereas, a screen with the actinin-4 rod domain did not generate any positive clones. Among the actinin-1 positive clones were multiple clones for myozenin, a known actinin-2-interacting protein [28]. When directly co-transformed, the myozenin preys were also able to interact with the actinin-4 bait indicating that this is not an isoform-specific interaction. Single clones for two other proteins HIPK1 (homeodomain-interacting protein kinase 1) and the Polycomb protein SCMH1 (sex comb on midleg homologue 1) were identified in our actinin-1 rod domain screen and were found to also interact with actinin-4, but were not investigated further. Overall, given the number of known actinin rod domain interactions, the number of proteins identified in our screen was very low, which suggests that the yeast two-hybrid system was not an efficient screening method for our purpose.

As an alternative approach we used TAP of actinin protein complexes coupled with MS. Heterodimer formation between transfected and endogenous actinin could make it difficult to identify interacting partners unique to either actinin-1 or -4. For this reason we used HEK-293 cells, which do not exhibit significant heterodimer formation (Figure 4B). We established stably transfected cell pools expressing TAP-tagged actinin-1 or -4 as well as control cells expressing TAP-tagged YFP. At least 70% of the cells in these pools expressed the TAP-tagged constructs and Western blotting indicated that expression levels were equivalent to the endogenous actinin levels in HEK-293 cells (results not shown). TAP-tagged actinin-1, -4 and YFP protein complexes were purified and subjected to denaturing gel electrophoresis. The proteins present in each sample were analysed by nLC and MS/MS and identified by searching against the IPI protein database. Non-specific interactions present in TAP-tagged YFP complexes and probable false positives or environmental contaminants were eliminated (see the Materials and methods section) to generate lists of proteins specifically identified in affinity purified actinin-1 and -4 complexes (Table 2). Several well-characterized actinin-interacting proteins were identified, validating the overall approach (Table 2, underlined). Although we used HEK-293 cells owing to their low levels of endogenous heterodimers, there appeared to be some degree of actinin-1/-4 heterodimer formation in our stably transfected cells since actinin-1 was found in actinin-4 complexes and vice versa (Table 2). Interestingly actinin-2 and -3 were also detected in both complexes providing further evidence that all combinations of heterodimers between muscle and non-muscle actinins are possible. The fact that relatively few common proteins were detected in both actinin-1 and -4 complexes suggested that despite the presence of some heterodimers, isoform-specific interactions could still be detected. More novel putative actinin-4-interacting partners were identified compared with actinin-1. Notably, over half of these are proteins that are reported to localize partially or exclusively in the nucleus and have functions related to transcription, RNA binding and mRNA splicing. These observations fit with the reported localization of actinin-4 to the nucleus [12,2123,29,30]. Overall this screen provides a list of putative actinin-1- and actinin-4-specific interacting proteins that might mediate or contribute to some of the reported isoform-specific functions of actinin-4 in particular.

Table 2
Proteomic analysis of actinin-1- and actinin-4-interacting proteins

The proteins reported previously to interact with actinin are underlined and the appropriate reference cited.

(a) Proteins identified only in actinin-1 complexes 
Protein symbol Mascot score Name Subcellular location Function 
DDX17 210 Probable ATP-dependent RNA helicase DDX17 isoform 3 Nucleus Transcriptional regulation 
PALLD [47175 Isoform 2 of Palladin Cytosol Actin cytoskeleton: organization 
CAMK2B [35103 Ca2+/CaM-dependent protein kinase type II subunit β Cytosol Protein kinase 
PDLIM3 [48103 PDZ and LIM domain protein 3 Cytosol Actin cytoskeleton: organization 
PRDX2 100 Peroxiredoxin-2 Cytosol Redox regulation in cells 
(b) Proteins identified only in actinin-4 complexes 
Gene symbol Mascot score Name Subcellular location Function 
LRPPRC 2257 Leucine-rich PPR motif-containing protein, mitochondrial Mitochondria, nucleus, cytosol RNA binding 
IQGAP2 1894 Isoform 1 of Ras GTPase-activating-like protein IQGAP2 Cytosol Cytoskeletal regulation 
DOCK7 588 Dedicator of cytokinesis protein 7 Cytosol Cytoskeletal regulation 
FLNC 544 Isoform 1 of filamin-C Cytosol Actin cytoskeleton: cross-linker 
RAVER1 [49457 Ribonucleoprotein PTB-binding 1 Nucleus, cytosol mRNA splicing, cytoskeletal 
SF3B2 432 Splicing factor 3B subunit 2 Nucleus mRNA splicing 
UBR5 399 E3 ubiquitin-protein ligase UBR5 Nucleus Ubiquitination, transcription 
PRRC2A 353 Isoform 1 of protein PRRC2A Nucleus, cytosol mRNA splicing 
CDKN2AIP 203 CDKN2A-interacting protein Nucleus Activation of p53 
PPFIA1 160 Isoform 1 of liprin-α-1 Cytosol Signalling/scaffolding 
TCEAL4 138 Isoform 2 of transcription elongation factor A protein-like 4 Nucleus Transcription 
ELAVL1 126 Similar to ELAV-like protein 1 Nucleus, cytosol RNA binding 
SPRR2E 117 Small proline-rich protein 2E Cytosol Cytoskeletal component 
SART1 113 U4/U6.U5 tri-snRNP-associated protein 1 Nucleus mRNA splicing 
TCERG1 108 Transcription elongation regulator 1 Nucleus Transcription 
MYH10 104 Isoform 1 of myosin-10 Cytosol Actin cytoskeleton: motor 
SMARCC2 103 Isoform 2 of SWI/SNF complex subunit SMARCC2 Nucleus Transcription 
SF3B4 100 Splicing factor 3B subunit 4 Nucleus mRNA splicing 
MYL6 100 Non-muscle myosin light chain 6 Cytosol Actin cytoskeleton: motor 
(c) Proteins identified both in actinin-1 and -4 complexes 
Gene symbol Mascot scores Name Subcellular location Function 
ACTN1 7522, 3925 α-Actinin-1 Cytosol Actin cytoskeleton: cross-linker 
ACTN4 4814, 6909 α-Actinin-4 Cytosol, nucleus Actin cytoskeleton: cross-linker 
ACTN2 1961, 1114 α-Actinin-2 Cytosol Actin cytoskeleton: cross-linker 
ACTN3 1366, 1071 α-Actinin-3 Cytosol Actin cytoskeleton: cross-linker 
MYH9 103, 391 Isoform 1 of myosin-9 Cytosol Actin cytoskeleton: motor 
PDLIM1 [50208, 38 PDZ and LIM domain protein 1 Cytosol Cytoskeletal, stress fibres 
(a) Proteins identified only in actinin-1 complexes 
Protein symbol Mascot score Name Subcellular location Function 
DDX17 210 Probable ATP-dependent RNA helicase DDX17 isoform 3 Nucleus Transcriptional regulation 
PALLD [47175 Isoform 2 of Palladin Cytosol Actin cytoskeleton: organization 
CAMK2B [35103 Ca2+/CaM-dependent protein kinase type II subunit β Cytosol Protein kinase 
PDLIM3 [48103 PDZ and LIM domain protein 3 Cytosol Actin cytoskeleton: organization 
PRDX2 100 Peroxiredoxin-2 Cytosol Redox regulation in cells 
(b) Proteins identified only in actinin-4 complexes 
Gene symbol Mascot score Name Subcellular location Function 
LRPPRC 2257 Leucine-rich PPR motif-containing protein, mitochondrial Mitochondria, nucleus, cytosol RNA binding 
IQGAP2 1894 Isoform 1 of Ras GTPase-activating-like protein IQGAP2 Cytosol Cytoskeletal regulation 
DOCK7 588 Dedicator of cytokinesis protein 7 Cytosol Cytoskeletal regulation 
FLNC 544 Isoform 1 of filamin-C Cytosol Actin cytoskeleton: cross-linker 
RAVER1 [49457 Ribonucleoprotein PTB-binding 1 Nucleus, cytosol mRNA splicing, cytoskeletal 
SF3B2 432 Splicing factor 3B subunit 2 Nucleus mRNA splicing 
UBR5 399 E3 ubiquitin-protein ligase UBR5 Nucleus Ubiquitination, transcription 
PRRC2A 353 Isoform 1 of protein PRRC2A Nucleus, cytosol mRNA splicing 
CDKN2AIP 203 CDKN2A-interacting protein Nucleus Activation of p53 
PPFIA1 160 Isoform 1 of liprin-α-1 Cytosol Signalling/scaffolding 
TCEAL4 138 Isoform 2 of transcription elongation factor A protein-like 4 Nucleus Transcription 
ELAVL1 126 Similar to ELAV-like protein 1 Nucleus, cytosol RNA binding 
SPRR2E 117 Small proline-rich protein 2E Cytosol Cytoskeletal component 
SART1 113 U4/U6.U5 tri-snRNP-associated protein 1 Nucleus mRNA splicing 
TCERG1 108 Transcription elongation regulator 1 Nucleus Transcription 
MYH10 104 Isoform 1 of myosin-10 Cytosol Actin cytoskeleton: motor 
SMARCC2 103 Isoform 2 of SWI/SNF complex subunit SMARCC2 Nucleus Transcription 
SF3B4 100 Splicing factor 3B subunit 4 Nucleus mRNA splicing 
MYL6 100 Non-muscle myosin light chain 6 Cytosol Actin cytoskeleton: motor 
(c) Proteins identified both in actinin-1 and -4 complexes 
Gene symbol Mascot scores Name Subcellular location Function 
ACTN1 7522, 3925 α-Actinin-1 Cytosol Actin cytoskeleton: cross-linker 
ACTN4 4814, 6909 α-Actinin-4 Cytosol, nucleus Actin cytoskeleton: cross-linker 
ACTN2 1961, 1114 α-Actinin-2 Cytosol Actin cytoskeleton: cross-linker 
ACTN3 1366, 1071 α-Actinin-3 Cytosol Actin cytoskeleton: cross-linker 
MYH9 103, 391 Isoform 1 of myosin-9 Cytosol Actin cytoskeleton: motor 
PDLIM1 [50208, 38 PDZ and LIM domain protein 1 Cytosol Cytoskeletal, stress fibres 

DISCUSSION

Alternative splicing and actin-binding properties of the non-muscle actinins

In the present study we have systematically compared the alternative splicing patterns and actin-binding properties of actinin-1 and -4 for the first time. Having compared the actin-binding properties of the most widely expressed Ca2+-sensitive forms of the two non-muscle actinins, we find that the affinity of actinin-1 (exon 19a), and actinin-4 (exon 8a and 19a) for F-actin (filamentous actin) are quite similar with Kd values of 1.93 μM and 2.96 μM respectively. In addition, their actin-bundling capacity and Ca2+-sensitivity of actin binding and bundling are nearly identical. Actinin-4 was first purified and cloned from the chicken lung and was described as having a low Ca2+-sensitivity [31,32]. The results of the present study using recombinant human actinin do not agree with this observation. This may reflect sequence differences in the EF-hand motifs between human and chick actinin or post-translational modifications in the protein purified from tissue that are not present in our bacterially expressed actinin. The non-muscle actinins are widely co-expressed. Yet actinin-4-knockout mice display specific kidney defects, whereas abnormalities in other tissues were not reported [20]. The similar actin-binding properties that we observe for the two non-muscle actinins are in agreement with this apparent functional redundancy between them in most tissues.

Although the actin-binding properties of the main Ca2+-sensitive non-muscle actinin isoforms are similar, our analysis of alternative splicing patterns reveals significant differences between actinin-1 and -4. In particular we find that the Ca2+-insensitive exon 19b variant of actinin-4 is only expressed in the nervous system. By contrast the exon 19b variant of actinin-1 is regarded as a smooth muscle isoform and is expressed in muscle and smooth muscle as well as in the adult brain. The exon 19a and 19b variants of actinin-4 are co-expressed in the brain at all stages examined; thus, actinin-4 does not exhibit the developmental switching from the exon 19a- to exon 19b-containing isoforms as observed for actinin-1 as the brain matures. Neither was an exon 19a+b variant of actinin-4 detected in the brain.

In muscle and smooth muscle cells, Ca2+-insensitive forms of actinin are thought to facilitate the relatively stable cross-linking of actin filaments despite the continuous fluxes in Ca2+ associated with muscle contraction [33]. The results of the present study show that whereas muscle cells express both the Ca2+-sensitive and -insensitive variants of actinin-1, they express only the Ca2+-sensitive (exon 19a) variant of actinin-4. This Ca2+-sensitive actinin-4 may have distinct functions in the muscle compared with other muscle actinin isoforms. Notably, potential roles for actinin-4 as a transcriptional regulator during muscle differentiation have been described [21,30]. In the brain, actinins are components of the postsynaptic density at synapses [34]. The results of the present study show a complex picture of actinin splicing in the brain with Ca2+-sensitive and -insensitive variants of both actinin-1 (19a, 19a+b and 19b) and actinin-4 (19a and 19b) being co-expressed. Alternative splicing of actinins in neurons may serve to fine tune the cross-linking of actin filaments at neuronal synapses where actinins interact with several key proteins involved in synaptic plasticity [35,36]. In addition to synaptic functions in neurons, actinin-1 and -4 are also expressed in glial cells and up-regulation of actinin-4 has been reported in glioblastoma and is proposed to promote cell migration and metastasis [17,18]. In four glioblastoma cell lines we find a reversion of actinin-1 splicing to the exon 19a isoform observed in the immature brain. These cell lines also predominantly express the exon 19a-containing actinin-4 isoform. Thus for both actinin-1 and -4 only the Ca2+-sensitive actinin isoforms are expressed in glioblastoma cells which may be of significance given the role intracellular Ca2+ plays in the process of cancer cell migration and proliferation [37].

Alternative splicing of exon 8 occurs in several invertebrate lineages [4] and has been conserved in mammals, birds, amphibians and fish for actinin-4, but not the other actinins [4]. This suggests that exon 8b of actinin-4 plays some essential, evolutionarily conserved function. We find that alternative splicing of exon 8b occurs in the nervous system where it predominates over exon 8a, as well as in muscle and smooth muscle-containing tissues where it is a minor species relative to the exon 8a variant. Variations in the amino acid sequence encoded by exon 8 certainly have the potential to alter the actin-bindingproperties and abnormal splicing of exon 8b is a feature of small-cell lung cancer [38,39]. However, the functional significance of alternative splicing of exon 8 has remained unclear. We find that the affinity of the exon 8b variant of actinin-4 for F-actin is not significantly different from that of the exon 8a variant and that both display similar Ca2+-sensitivities. This contrasts with the enhanced actin-binding of the exon 8b variant reported previously by Honda et al. [38]. It should be noted however that Honda et al. [38] measured actin binding of the isolated ABD and that the binding properties of this isoform may be different in the context of the intact actinin dimer. Overall our findings indicate that the actin-binding properties of actinin-1 and the exon 8a and 8b variants of actinin-4 are very similar. Expression of these isoforms may thus serve to fine tune the affinity of actinin for F-actin in particular contexts, but would not alter actin cross-linking properties very dramatically.

Implications of non-muscle actinin heterodimerization

The non-muscle actinins are generally regarded as distinct homodimeric entities. For example actinin-1 and -4 have been reported to have different localizations in cells [12,29,30,40] and to play different, and sometimes opposing, roles in cellular processes such as cell migration, proliferation and adhesion [17,41]. Heterodimer formation is known to occur between the muscle actinins-2 and -3 [9]. To our knowledge the ability of the non-muscle actinins to form heterodimers had not been directly investigated prior to the present study. We present evidence from yeast two-hybrid and in vitro binding assays that actinin-1 and -4 can form heterodimers with each other, as well as with actinin-2, in agreement with the conservation of the dimer interface between these isoforms. Furthermore, we examined the extent of heterodimer formation between endogenously expressed actinins in cells. Significant non-muscle heterodimer formation was observed in six of the eight cell lines examined. Most surprisingly we found that heterodimers are more abundant than either homodimer species in these cell lines. In fact, these proportions of homo- and hetero-dimers are close to what would be expected if the probability of a newly translated actinin monomer forming either a homo- or hetero-dimer is simply proportional to the relative abundance of actinin-1 and -4 in the cell. Thus in a cell line in which actinin-1 and -4 are expressed equally, such as U87-MG (Figures 4D and 4E), there is 25% of each homodimer and 50% actinin-1/actinin-4 heterodimer. This would indicate that preferential homodimer formation between proteins being translated from the same mRNA does not occur. Alternatively homodimer formation may be favoured initially, but monomeric subunits may subsequently be exchanged between dimers until an equilibrium is reached between homo- and hetero-dimers.

In the present study, the degree of heterodimer formation varied between the cell lines, and heterodimers were not observed in HEK-293 cells despite that fact that they express both actinin-1 and -4. Actinin expression in HEK-293 cells is somewhat lower than the other cell lines, but if the homodimer/heterodimer ratio was similar to other cells then heterodimers would still be detectable. This suggests that heterodimer formation may either require high cellular concentrations of actinins or may be regulated in a cell line-specific manner. Notably HEK-293 cells were the only non-cancer cell line examined. Overexpression of actinin-4 has been reported in many types of tumours and is associated with high-grade malignancies and poor outcomes [12,1517]. The results of the present study indicate that when overexpressed, actinin-4 may be capable of forcing actinin-1 into the heterodimer state. This suggests that the cancer phenotype associated with actinin-4 overexpression may not only be a consequence of acquisition or enhancement of an actinin-4-specific function, but may reflect the disruption of actinin-1-specific functions owing to the sequestration of actinin-1 into heterodimers. Alternatively these heterodimers may also have unique tumorigenic or metastatic-promoting properties not possessed by homodimers. Thus the functional consequence of actinin-4 overexpression in a particular tumour may depend on the level of actinin-1 expression and the degree to which actinin-1 is sequestered into heterodimers.

The question of whether actinin-1/actinin-4 heterodimers exist in normal tissues in vivo also arises. Although we have not examined this directly in the present study, the presence of actinin heterodimers in platelets was well characterized by Olomucki and co-workers almost 30 years ago [42,43]. These heterodimers probably contain actinin-1, but at that time actinin-4 had not been cloned, so the exact identity of these heterodimers remained unclear. Using isoform-specific antibodies we find that, in addition to actinin-1, human platelets have high levels of actinin-4 as well as some sarcomeric actinin (actinin-2/3) (Supplementary Figure S1). Taken together this strongly indicates that platelets contain actinin-1/actinin-4 heterodimers and potentially heterodimers formed between muscle and non-muscle isoforms. The co-expression of multiple actinins in many tissues suggests that a variety of heterodimeric actinin species are likely to occur in vivo.

Actinin protein–protein interactions

Our observation that the non-muscle actinins have similar affinities for actin filaments and similar Ca2+ sensitivities indicates that differences in actin-binding properties are unlikely to explain isoform-specific functions, such as those that have been attributed to actinin-4. One possible explanation for such characteristics could be that actinin-4 has unique interacting partners thereby allowing it to carry out functions unmatched by actinin-1. Our proteomics screen revealed a number of putative actinin-4-specific binding proteins that might be linked to the described functions of actinin-4 in cancer and as a nucleo–cytoplasmic shuttling protein (Table 2). We identified cytoskeletal components like myosin-10 and filamin-C as well as two proteins, IQGAP2 (IQ motif-containing GTPase-activating protein 2) and DOCK7 (dedicator of cytokinesis 7), that have the potential to modulate actin dynamics through their regulation of Rho family GTPases. DOCK7 has been shown to act as a guanine-nucleotide-exchange factor for Rac1 and Rac3 [44], whereas IQGAP2 can regulate Rac1 and Cdc42 (cell-division cycle 42), although not by acting as a classical GTPase-activating protein [45]. These cytoskeletal proteins/regulators could potentially mediate some of the effects on cell migration and metastasis that have been ascribed to actinin-4 in the context of cancer. Apart from these cytoskeletal proteins, most of the remaining proteins identified in the TAP-tagged actinin-4 complexes are nuclear proteins. This supports the idea that actinin-4 can shuttle in and out of the nucleus [23]. One of these nuclear proteins, CDKN2AIP (cyclin-dependent kinase inhibitor 2A-interacting protein) is known to activate p53 and thus might connect actinin-4 to cancer phenotypes [46]. The remaining nuclear proteins identified are involved in either transcription or RNA binding/splicing. Although there is no known role for actinin-4 in RNA splicing, several functional interactions of actinin-4 with transcriptional regulators have been described [21,22,24,30]. The proteins identified in the present study may expand this list.

Conclusions

The results of the present study describe the tissue-specific alternative splicing of actinin-4 in detail for the first time. We also report that the actin-binding properties of the non-muscle actinins are virtually identical. Additionally, we find that actinin-1 and -4 do not exist solely as distinct homodimeric entities, but that they can readily form heterodimers composed of monomers that may have different properties and interacting proteins. Overall these findings significantly alter the context in which one views the isoform-specific functions of the non-muscle actinins.

Abbreviations

     
  • ABD

    actin-binding domain

  •  
  • CaM

    calmodulin-like

  •  
  • DMEM

    Dulbecco's modified Eagle's medium

  •  
  • DOCK7

    dedicator of cytokinesis 7

  •  
  • DTT

    dithiothreitol

  •  
  • F-actin

    filamentous actin

  •  
  • FBS

    fetal bovine serum

  •  
  • GFP

    green fluorescent protein

  •  
  • GST

    glutathione transferase

  •  
  • HEK

    human embryonic kidney

  •  
  • HIS3

    histidine 3

  •  
  • IPI

    International Protein Index

  •  
  • IQGAP2

    IQ motif-containing GTPase-activating protein 2

  •  
  • MS/MS

    tandem MS

  •  
  • nLC

    nano liquid chromatography

  •  
  • NP40

    Nonidet P40

  •  
  • rtPCR

    reverse transcription PCR

  •  
  • siRNA

    small interfering RNA

  •  
  • TAP

    tandem affinity purification

  •  
  • TEV

    tobacco etch virus

  •  
  • YFP

    yellow fluorescent protein

AUTHOR CONTRIBUTION

Kate Foley carried out all of the experimental work. Paul Young supervised the study. Kate Foley and Paul Young analysed the experimental data and wrote the paper.

We thank Martha Phelan for help in preparing washed platelets, Anita Murphy, Anjali Pai and Gary Williamson for their contributions to the project while working in the laboratory, and to Kellie Dean for reading the paper before submission.

FUNDING

This work was funded, primarily, by the Health Research Board, PhD Scholars Programme in Cancer Biology [grant number PhD/2007/4].

References

References
1
Otey
C. A.
Carpen
O.
α-Actinin revisited: a fresh look at an old player
Cell Motil. Cytoskeleton
2004
, vol. 
58
 (pg. 
104
-
111
)
2
Waites
G. T.
Graham
I. R.
Jackson
P.
Millake
D. B.
Patel
B.
Blanchard
A. D.
Weller
P. A.
Eperon
I. C.
Critchley
D. R.
Mutually exclusive splicing of calcium-binding domain exons in chick α-actinin
J. Biol. Chem.
1992
, vol. 
267
 (pg. 
6263
-
6271
)
3
Kremerskothen
J.
Teber
I.
Wendholt
D.
Liedtke
T.
Bockers
T. M.
Barnekow
A.
Brain-specific splicing of α-actinin 1 (ACTN1) mRNA. Biochem
Biophys. Res. Commun.
2002
, vol. 
295
 (pg. 
678
-
681
)
4
Lek
M.
MacArthur
D. G.
Yang
N.
North
K. N.
Phylogenetic analysis of gene structure and alternative splicing in α-actinins
Mol. Biol. Evol.
2010
, vol. 
27
 (pg. 
773
-
780
)
5
Franzot
G.
Sjoblom
B.
Gautel
M.
Djinovic Carugo
K.
The crystal structure of the actin binding domain from α-actinin in its closed conformation: structural insight into phospholipid regulation of α-actinin
J. Mol. Biol.
2005
, vol. 
348
 (pg. 
151
-
165
)
6
Djinovic Carugo
K.
Young
P.
Gautel
M.
Saraste
M.
Structure of the α-actinin rod: molecular basis for cross-linking of actin filaments
Cell
1999
, vol. 
98
 (pg. 
537
-
546
)
7
Ylanne
J.
Scheffzek
K.
Young
P.
Saraste
M.
Crystal structure of the α-actinin rod reveals an extensive torsional twist
Structure
2001
, vol. 
9
 (pg. 
597
-
604
)
8
Young
P.
Gautel
M.
The interaction of titin and α-actinin is controlled by a phospholipid-regulated intramolecular pseudoligand mechanism
EMBO J.
2000
, vol. 
19
 (pg. 
6331
-
6340
)
9
Chan
Y.
Tong
H. Q.
Beggs
A. H.
Kunkel
L. M.
Human skeletal muscle-specific α-actinin-2 and -3 isoforms form homodimers and heterodimers in vitro and in vivo
Biochem. Biophys. Res. Commun.
1998
, vol. 
248
 (pg. 
134
-
139
)
10
Djinovic-Carugo
K.
Gautel
M.
Ylanne
J.
Young
P.
The spectrin repeat: a structural platform for cytoskeletal protein assemblies
FEBS Lett.
2002
, vol. 
513
 (pg. 
119
-
123
)
11
Sjoblom
B.
Salmazo
A.
Djinovic-Carugo
K.
α-Actinin structure and regulation
Cell. Mol. Life Sci.
2008
, vol. 
65
 (pg. 
2688
-
2701
)
12
Honda
K.
Yamada
T.
Endo
R.
Ino
Y.
Gotoh
M.
Tsuda
H.
Yamada
Y.
Chiba
H.
Hirohashi
S.
Actinin-4, a novel actin-bundling protein associated with cell motility and cancer invasion
J. Cell Biol.
1998
, vol. 
140
 (pg. 
1383
-
1393
)
13
Barbolina
M. V.
Adley
B. P.
Kelly
D. L.
Fought
A. J.
Scholtens
D. M.
Shea
L. D.
Stack
M. S.
Motility-related actinin α-4 is associated with advanced and metastatic ovarian carcinoma
Lab. Invest.
2008
, vol. 
88
 (pg. 
602
-
614
)
14
Yamamoto
S.
Tsuda
H.
Honda
K.
Takano
M.
Tamai
S.
Imoto
I.
Inazawa
J.
Yamada
T.
Matsubara
O.
ACTN4 gene amplification and actinin-4 protein overexpression drive tumour development and histological progression in a high-grade subset of ovarian clear-cell adenocarcinomas
Histopathology
2012
, vol. 
60
 (pg. 
1073
-
1083
)
15
Honda
K.
Yamada
T.
Hayashida
Y.
Idogawa
M.
Sato
S.
Hasegawa
F.
Ino
Y.
Ono
M.
Hirohashi
S.
Actinin-4 increases cell motility and promotes lymph node metastasis of colorectal cancer
Gastroenterology
2005
, vol. 
128
 (pg. 
51
-
62
)
16
Kikuchi
S.
Honda
K.
Tsuda
H.
Hiraoka
N.
Imoto
I.
Kosuge
T.
Umaki
T.
Onozato
K.
Shitashige
M.
Yamaguchi
U.
, et al. 
Expression and gene amplification of actinin-4 in invasive ductal carcinoma of the pancreas
Clin. Cancer Res.
2008
, vol. 
14
 (pg. 
5348
-
5356
)
17
Quick
Q.
Skalli
O.
α-Actinin 1 and α-actinin 4: contrasting roles in the survival, motility, and RhoA signaling of astrocytoma cells
Exp. Cell Res.
2010
, vol. 
316
 (pg. 
1137
-
1147
)
18
Sen
S.
Dong
M.
Kumar
S.
Isoform-specific contributions of α-actinin to glioma cell mechanobiology
PLoS ONE
2009
, vol. 
4
 pg. 
e8427
 
19
Kaplan
J. M.
Kim
S. H.
North
K. N.
Rennke
H.
Correia
L. A.
Tong
H. Q.
Mathis
B. J.
Rodriguez-Perez
J. C.
Allen
P. G.
Beggs
A. H.
Pollak
M. R.
Mutations in ACTN4, encoding α-actinin-4, cause familial focal segmental glomerulosclerosis
Nat. Genet.
2000
, vol. 
24
 (pg. 
251
-
256
)
20
Kos
C. H.
Le
T. C.
Sinha
S.
Henderson
J. M.
Kim
S. H.
Sugimoto
H.
Kalluri
R.
Gerszten
R. E.
Pollak
M. R.
Mice deficient in α-actinin-4 have severe glomerular disease
J. Clin. Invest.
2003
, vol. 
111
 (pg. 
1683
-
1690
)
21
Chakraborty
S.
Reineke
E. L.
Lam
M.
Li
X.
Liu
Y.
Gao
C.
Khurana
S.
Kao
H. Y.
α-Actinin 4 potentiates myocyte enhancer factor-2 transcription activity by antagonizing histone deacetylase 7
J. Biol. Chem.
2006
, vol. 
281
 (pg. 
35070
-
35080
)
22
Khurana
S.
Chakraborty
S.
Cheng
X.
Su
Y. T.
Kao
H. Y.
The actin-binding protein, actinin α4 (ACTN4), is a nuclear receptor coactivator that promotes proliferation of MCF-7 breast cancer cells
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
1850
-
1859
)
23
Kumeta
M.
Yoshimura
S. H.
Harata
M.
Takeyasu
K.
Molecular mechanisms underlying nucleocytoplasmic shuttling of actinin-4
J. Cell Sci.
2010
, vol. 
123
 (pg. 
1020
-
1030
)
24
Babakov
V. N.
Petukhova
O. A.
Turoverova
L. V.
Kropacheva
I. V.
Tentler
D. G.
Bolshakova
A. V.
Podolskaya
E. P.
Magnusson
K. E.
Pinaev
G. P.
RelA/NF-κB transcription factor associates with α-actinin-4
Exp. Cell Res.
2008
, vol. 
314
 (pg. 
1030
-
1038
)
25
Ashkenazy
H.
Erez
E.
Martz
E.
Pupko
T.
Ben-Tal
N.
ConSurf 2010: calculating evolutionary conservation in sequence and structure of proteins and nucleic acids
Nucleic Acids Res.
2010
, vol. 
38
 (pg. 
W529
-
W533
)
26
Landau
M.
Mayrose
I.
Rosenberg
Y.
Glaser
F.
Martz
E.
Pupko
T.
Ben-Tal
N.
ConSurf 2005: the projection of evolutionary conservation scores of residues on protein structures
Nucleic Acids Res.
2005
, vol. 
33
 (pg. 
W299
-
W302
)
27
Young
P.
Ferguson
C.
Banuelos
S.
Gautel
M.
Molecular structure of the sarcomeric Z-disk: two types of titin interactions lead to an asymmetrical sorting of α-actinin
EMBO J.
1998
, vol. 
17
 (pg. 
1614
-
1624
)
28
Takada
F.
Vander Woude
D. L.
Tong
H. Q.
Thompson
T. G.
Watkins
S. C.
Kunkel
L. M.
Beggs
A. H.
Myozenin: an α-actinin- and γ-filamin-binding protein of skeletal muscle Z lines
Proc. Natl. Acad. Sci. U.S.A.
2001
, vol. 
98
 (pg. 
1595
-
1600
)
29
Bolshakova
A.
Petukhova
O.
Turoverova
L.
Tentler
D.
Babakov
V.
Magnusson
K. E.
Pinaev
G.
Extra-cellular matrix proteins induce re-distribution of α-actinin-1 and α-actinin-4 in A431 cells
Cell Biol. Int.
2007
, vol. 
31
 (pg. 
360
-
365
)
30
Goffart
S.
Franko
A.
Clemen
C. S.
Wiesner
R. J.
α-Actinin 4 and BAT1 interaction with the cytochrome c promoter upon skeletal muscle differentiation
Curr. Genet.
2006
, vol. 
49
 (pg. 
125
-
135
)
31
Imamura
M.
Masaki
T.
A novel nonmuscle α-actinin. Purification and characterization of chicken lung α-actinin
J. Biol. Chem.
1992
, vol. 
267
 (pg. 
25927
-
25933
)
32
Imamura
M.
Sakurai
T.
Ogawa
Y.
Ishikawa
T.
Goto
K.
Masaki
T.
Molecular cloning of low-Ca2+-sensitive-type non-muscle α-actinin
Eur. J. Biochem.
1994
, vol. 
223
 (pg. 
395
-
401
)
33
Blanchard
A.
Ohanian
V.
Critchley
D.
The structure and function of α-actinin
J. Muscle Res. Cell Motil.
1989
, vol. 
10
 (pg. 
280
-
289
)
34
Walikonis
R. S.
Jensen
O. N.
Mann
M.
Provance
D. W.
Jr
Mercer
J. A.
Kennedy
M. B.
Identification of proteins in the postsynaptic density fraction by mass spectrometry
J. Neurosci.
2000
, vol. 
20
 (pg. 
4069
-
4080
)
35
Walikonis
R. S.
Oguni
A.
Khorosheva
E. M.
Jeng
C. J.
Asuncion
F. J.
Kennedy
M. B.
Densin-180 forms a ternary complex with the α-subunit of Ca2+/calmodulin-dependent protein kinase II and α-actinin
J. Neurosci.
2001
, vol. 
21
 (pg. 
423
-
433
)
36
Wyszynski
M.
Lin
J.
Rao
A.
Nigh
E.
Beggs
A. H.
Craig
A. M.
Sheng
M.
Competitive binding of α-actinin and calmodulin to the NMDA receptor
Nature
1997
, vol. 
385
 (pg. 
439
-
442
)
37
Prevarskaya
N.
Skryma
R.
Shuba
Y.
Calcium in tumour metastasis: new roles for known actors
Nat. Rev. Cancer
2011
, vol. 
11
 (pg. 
609
-
618
)
38
Honda
K.
Yamada
T.
Seike
M.
Hayashida
Y.
Idogawa
M.
Kondo
T.
Ino
Y.
Hirohashi
S.
Alternative splice variant of actinin-4 in small cell lung cancer
Oncogene
2004
, vol. 
23
 (pg. 
5257
-
5262
)
39
Weins
A.
Schlondorff
J. S.
Nakamura
F.
Denker
B. M.
Hartwig
J. H.
Stossel
T. P.
Pollak
M. R.
Disease-associated mutant α-actinin-4 reveals a mechanism for regulating its F-actin-binding affinity
Proc. Natl. Acad. Sci. U.S.A.
2007
, vol. 
104
 (pg. 
16080
-
16085
)
40
Araki
N.
Hatae
T.
Yamada
T.
Hirohashi
S.
Actinin-4 is preferentially involved in circular ruffling and macropinocytosis in mouse macrophages: analysis by fluorescence ratio imaging
J. Cell Sci.
2000
, vol. 
113
 (pg. 
3329
-
3340
)
41
Celli
L.
Ryckewaert
J. J.
Delachanal
E.
Duperray
A.
Evidence of a functional role for interaction between ICAM-1 and nonmuscle α-actinins in leukocyte diapedesis
J. Immunol.
2006
, vol. 
177
 (pg. 
4113
-
4121
)
42
Gache
Y.
Landon
F.
Olomucki
A.
Polymorphism of α-actinin from human blood platelets. Homodimeric and heterodimeric forms
Eur. J. Biochem.
1984
, vol. 
141
 (pg. 
57
-
61
)
43
Landon
F.
Gache
Y.
Touitou
H.
Olomucki
A.
Properties of two isoforms of human blood platelet alpha-actinin
Eur. J. Biochem.
1985
, vol. 
153
 (pg. 
231
-
237
)
44
Watabe-Uchida
M.
John
K. A.
Janas
J. A.
Newey
S. E.
Van Aelst
L.
The Rac activator DOCK7 regulates neuronal polarity through local phosphorylation of stathmin/Op18
Neuron
2006
, vol. 
51
 (pg. 
727
-
739
)
45
Brill
S.
Li
S.
Lyman
C. W.
Church
D. M.
Wasmuth
J. J.
Weissbach
L.
Bernards
A.
Snijders
A. J.
The Ras GTPase-activating-protein-related human protein IQGAP2 harbors a potential actin binding domain and interacts with calmodulin and Rho family GTPases
Mol. Cell. Biol.
1996
, vol. 
16
 (pg. 
4869
-
4878
)
46
Hasan
M. K.
Yaguchi
T.
Minoda
Y.
Hirano
T.
Taira
K.
Wadhwa
R.
Kaul
S. C.
Alternative reading frame protein (ARF)-independent function of CARF (collaborator of ARF) involves its interactions with p53: evidence for a novel p53-activation pathway and its negative feedback control
Biochem. J.
2004
, vol. 
380
 (pg. 
605
-
610
)
47
Ronty
M.
Taivainen
A.
Moza
M.
Otey
C. A.
Carpen
O.
Molecular analysis of the interaction between palladin and α-actinin
FEBS Lett.
2004
, vol. 
566
 (pg. 
30
-
34
)
48
Xia
H.
Winokur
S. T.
Kuo
W. L.
Altherr
M. R.
Bredt
D. S.
Actinin-associated LIM protein: identification of a domain interaction between PDZ and spectrin-like repeat motifs
J. Cell Biol.
1997
, vol. 
139
 (pg. 
507
-
515
)
49
Huttelmaier
S.
Illenberger
S.
Grosheva
I.
Rudiger
M.
Singer
R. H.
Jockusch
B. M.
Raver1, a dual compartment protein, is a ligand for PTB/hnRNPI and microfilament attachment proteins
J. Cell Biol.
2001
, vol. 
155
 (pg. 
775
-
786
)
50
Vallenius
T.
Luukko
K.
Makela
T. P.
CLP-36 PDZ-LIM protein associates with nonmuscle α-actinin-1 and α-actinin-4
J. Biol. Chem.
2000
, vol. 
275
 (pg. 
11100
-
11105
)

Supplementary data