EFA6A encodes two isoforms with distinct biological activities in neuronal cells

The regulation of GTPases forms the basis of a wide variety of cellular processes, such as differentiation, migration and proliferation. The ADP-ribosylation factor (ARF) family of GTPases comprises six polypeptides (ARF1-ARF6) that are involved in membrane trafficking and maintenance of organelle structure (D'Souza-Schorey and Chavrier, 2006). The ARF6-signaling pathway has a role in endosomal-plasma membrane recycling, and in cortical actin cytoskeleton remodeling (D'Souza-Schorey and Chavrier, 2006), suggesting that its activity could coordinate membrane and cytoskeleton dynamics. ARF6 has also been demonstrated to be involved in regulated exocytosis (Caumont et al., 1998; Yang and Mueckler, 1999), in clathrin-dependent and - independent endocytosis (Naslavsky et al., 2003), in phagocytosis (Niedergang et al., 2003), in cytokinesis (Schweitzer and D'Souza-Schorey, 2002), in coordinating the transition of epithelial cells from a stationary to a motile state (Santy and Casanova, 2001; Turner and Brown, 2001) and in tumor cell invasion (Hashimoto et al., 2004; Tague et al., 2004).

ARF6 is a ubiquitously expressed GTPase (Yang et al., 1998). Several molecules that appear to act in vitro as GEFs or GAPs for ARF6 have been identified, indicating that the regulation of ARF6 activity is finely tuned (Randazzo and Hirsch, 2004; Shin and Nakayama, 2004). Distinct GEFs and GAPs could regulate the diverse activities of ARF6 by localizing at specific subcellular compartments and mediating ARF6 regulation at distinct sites (D'Souza-Schorey and Chavrier, 2006; Shin and Nakayama, 2004). Localization of the ARF6 GEFs and GAPs could be under the control of diverse signals (upon binding of different phosphoinositides via the PH domains, or via protein-protein interactions) and in turn might regulate different biological processes via the recruitment of specific effectors.

ARF6 GEFs include the EFA6 family and the ARNO/cytohesin family (Cox et al., 2004; D'Souza-Schorey and Chavrier, 2006); these molecules are characterized by the presence of a Sec7 domain, mediating the GEF activity, a PH domain and a coiled-coil region. The EFA6 gene family includes four members: EFA6A, EFA6B, EFA6C and EFA6D (Derrien et al., 2002; Franco et al., 1999; Matsuya et al., 2005; Sakagami et al., 2006; Sakagami, 2008). Among the putative ARF6 GEFs, some, such as ARNO (Kim et al., 1998), EFA6B (Derrien et al., 2002) and EFA6D (Sakagami et al., 2006), are widely expressed, whereas EFA6A and EFA6C appear to be tissue specific, and are both highly expressed in the brain (Matsuya et al., 2005; Perletti et al., 1997; Suzuki et al., 2002), underlying a fundamental role for ARF6 regulation in neural tissue (Jaworski, 2007).

The process of neurite extension and remodeling requires a close coordination between the cytoskeleton and the cell membranes, to allow polarized growth of neuronal processes (Bretscher and Aguado-Velasco, 1998; da Silva and Dotti, 2002). Cytoskeletal remodeling during neuronal development underlies axonal and dendritic extension and branching, apical dendrite development, spine formation and synaptic plasticity.

ARF6 is expressed in developing and adult rat brain (Choi et al., 2006; Suzuki et al., 2002). The ARF6 GEF ARNO has been shown to regulate dendritic and axonal development in hippocampal neurons (Hernandez-Deviez et al., 2002; Hernandez-Deviez et al., 2004). Also, ARF6 has been shown to be involved in neurite extension in chicken retinal neurons (Albertinazzi et al., 2003), and in regulating spine formation (Choi et al., 2006; Miyazaki et al., 2005).

A role for EFA6A in neuronal morphogenesis is suggested by the finding that a GEF-defective EFA6A mutant enhances dendrite formation (Sakagami et al., 2004), and the observation that EFA6A overexpression promotes spine formation in hippocampal neurons (Choi et al., 2006). EFA6A is expressed in neural tissue during development, and in adult brain EFA6A mRNAs are detected in the cerebral cortex, the hippocampus and the dentate gyrus (Sakagami et al., 2004; Suzuki et al., 2002). Northern analysis has shown two distinct transcripts, of about 4 kb and 2 kb, respectively, both in human (Perletti et al., 1997) and rat (Suzuki et al., 2002) brain RNA. Both transcripts are expressed in neural tissue during embryogenesis and in postnatal life (Suzuki et al., 2002).

In the current study, we isolated and characterized the murine cDNAs corresponding to the two EFA6A mRNA transcripts, and analyzed their functional activity in HeLa cells and in neuronal cells. Our findings indicate that the two isoforms encoded by EFA6A have distinct effects on cell morphogenesis, and suggest that their regulated expression in neuronal cells might have a role in the process of neurite elaboration.

To isolate murine EFA6A, we screened mouse cDNA libraries using the 3′ region of human EFA6A (Perletti et al., 1997) as a probe, followed by the RACE technique to extend 5′ the cloned sequences. Through this approach, a 3785 bp cDNA was isolated. The murine EFA6A mRNA has the potential to encode a 1025 amino acid polypeptide, encompassing two proline-rich sequences, a Sec7 and a PH domain, and two putative coiled-coil motifs in the C-terminal region (identified by the COILS algorithm) (Lupas, 1996) (Fig. 1A; supplementary material Fig. S1A). The initiation methionine codon is preceded by an in-frame stop codon. A murine cDNA clone of 3950 bp (Accession no. NM_028627) is almost identical to the murine EFA6A cDNA that we cloned (no mismatches in 3782 bp overlap). Our cDNA only diverged from this sequence for a 3 bp insertion in position 1645, which results in the insertion of a serine in position 519 of the encoded protein. As this occurs at a site of splicing, and the three-nucleotide insertion is also found in several ESTs, it is likely to be a splicing variant. A murine EFA6A cDNA identical to the NM_028627 GenBank sequence, encoding a 1024 amino acid polypeptide has been recently reported by Sakagami et al. (Sakagami et al., 2007).

By screening a day 11.5 mouse embryo cDNA library we also isolated a different 1850 bp cDNA, which appeared to be identical to the EFA6A cDNA in the 3′ region, up to nucleotide 2185, and then diverged. This cDNA has the potential to encode a 393 amino acid polypeptide. The putative methionine initiation codon is not preceded by a stop codon; we have been unable to isolate clones extending 5′ by RACE experiments. The nucleotide sequence surrounding the predicted initiating methionine conforms to the Kozak consensus for efficient translation (Kozak, 1987). The encoded polypeptide has an N-terminal unique stretch of 66 amino acids (supplementary material Fig. S1B), followed by 327 amino acids identical to EFA6A, encompassing the PH domain and the coiled-coil regions (Fig. 1A; supplementary material Fig. S1A). We named this polypeptide EFA6As (for short EFA6A). EFA6As lacks the Sec7 domain, endowed with the GEF activity on ARF6; it could act as a scaffolding protein, tethering interacting proteins via its coiled-coil motifs.

The sequence encoding EFA6As appears to be the result of transcription from an alternative internal promoter. The EFA6A cDNA is encoded by 16 exons; the first exon of EFA6As, encoding the 66 N-terminal amino acids, is contained in a large 3.4 kb intron, between exons 9 and 10 of EFA6A (supplementary material Fig. S1C). The region 5′ to the start of the EFA6As sequence contains a CpG island; in this region we have identified a 200 bp fragment that displays promoter activity in neuronal cells (G.F., unpublished).

The identification of two distinct EFA6A cDNA sequences, identical in the 3′ region, is in line with the two distinct transcripts, of about 4 kb and 2 kb, detected by Northern analysis, upon hybridization of human (Perletti et al., 1997), rat (Suzuki et al., 2002) or murine (Fig. 1B,C) brain RNA with probes encompassing the 3′ region of EFA6A.

Using BLAST analysis, we identified a human fetal brain cDNA that has the potential to encode the EFA6A short form (Accession no. CR616163.1), as well as several human ESTs encompassing the 5′ region of EFA6As cDNA. In addition, a cDNA clone of 1832 bp from murine visual cortex (Accession no. AK158851.1), identical to the sequence that we isolated, is present in GenBank. Interestingly, searching GenBank with the N-terminal amino acid sequence of EFA6As, several chicken, fish and Xenopus EST sequences encoding highly homologous polypeptides were identified, indicating that the transcript encoding EFA6As appears to be highly conserved in vertebrates.

We performed a Northern analysis of total RNA from adult murine tissues, using as a probe a 520 bp fragment, common to cDNAs encoding EFA6A and EFA6As (encompassing the sequence encoding 118 amino acids of the C-terminal region and 166 nucleotides of the 3′ untranslated region). Two distinct transcripts, of about 4 kb and 2 kb, respectively, were detected in brain RNA, whereas stomach, intestine, uterus and ovary mainly expressed the 2 kb transcript (Fig. 1B). No specific signal was detectable in tongue, thymus, spleen, lung, heart, liver and kidney. These results are in line with those observed in the analysis of RNA from adult human tissues using a probe from human EFA6A (Perletti et al., 1997). Both transcripts were detected in embryonic, early postnatal and adult mouse brain (Fig. 1C). The mRNA transcript encoding EFA6A peaked at postnatal day 4-8, and it was downregulated in the adult, whereas the highest level of the mRNA encoding EFA6As was detected in adult brain. These data closely parallel what was observed in rat brain (Sakagami et al., 2004; Suzuki et al., 2002).

The distinct temporal regulation of the levels of mRNAs encoding EFA6A and EFA6As suggests that the two polypeptides might perform different functions in neuronal cells.

To analyze the expression of endogenous EFA6A and EFA6As in murine tissues, we raised two distinct polyclonal antibodies, by immunizing rabbits against a histidine-tagged fusion protein encompassing the C-terminal 155 amino acids, or against a 13 amino acid peptide corresponding to amino acids 988-1000 of murine EFA6A. By western blot analysis, both antibodies appeared to detect specific bands in extracts from murine brain tissues (Fig. 1D), with apparent molecular masses of about 120 kDa and 45 kDa, comigrating with EFA6A and EFA6As overexpressed in transfected HeLa cells. These same bands were not present in extracts from tissues that do not express EFA6A, such as kidney or liver (Fig. 1D).

We also attempted to generate anti-EFA6As polyclonal antibodies, directed against the N-terminal 66 amino acids unique to EFA6As; however, the antibodies obtained detected the overexpressed EFA6As in extracts of transfected cells, but not the endogenously expressed protein. The lack of success in generating high-titer specific antibodies is probably the result of the remarkable sequence conservation of the 66 N-terminal amino acids across species (i.e. 100% identity between the murine and the human sequence; 86% identity between the murine and the chicken sequence).

We tested the biological activity of EFA6A and EFA6As in HeLa cells, by transfection of the cDNAs cloned in the pcDNA3 expression vector, followed by immunofluorescence analysis. Cells transfected with EFA6A displayed membrane ruffles at the cell periphery, and a substantial loss of stress fibers (Fig. 2C-F,M), compared with untransfected or EGFP-transfected cells (Fig. 2A,B,M); EFA6A was detected mostly at the cell membrane, and it colocalized with F-actin at membrane ruffles. The biological activity of EFA6A appears to require the C-terminal region of EFA6A, because a ΔC mutant (Fig. 1A), encompassing the first 883 amino acids of EFA6A, but lacking the 142 C-terminal residues, does not induce morphological changes or loss of stress fibers (Fig. 2I,J,M). This finding indicates that the C-terminal region, common to EFA6A and EFA6As, is essential for cytoskeletal remodeling in HeLa cells.

Transfection of HeLa cells with a plasmid encoding a polypeptide lacking the first 374 amino acids of murine EFA6A (construct 375) (Fig. 1A) induced a similar phenotype, indicating that the N-terminal region appears to be dispensable for biological activity in transfected HeLa cells (Fig. 2G,H,M).

Cells overexpressing EFA6As were characterized by an elongated morphology, and by an increase of the short microvilli-like actin-rich protrusions on the dorsal surface (Fig. 2K,L,N). The majority of the protein appeared to be localized on the plasma membrane and on these protrusions; no substantial loss of stress fibers was detected in EFA6As-transfected cells (Fig. 2M). Deletion mutants of EFA6As lacking the C-terminal 142 amino acids did not induce elongation of transfected HeLa cells, indicating that biological activity of EFA6As also requires the C-terminal region (data not shown).

Our results with HeLa cells indicate that both isoforms encoded by EFA6A induce cytoskeletal remodeling, but lead to distinct morphological changes in transfected cells: EFA6A promotes the formation of membrane ruffles and the loss of stress fibers, whereas EFA6As induces cell elongation.

To assess the biological function of murine EFA6A and EFA6As in neuronal cells, we transfected the EFA6A cDNA in rat pheochromocytoma PC12 cells. After 3-4 hours of incubation with the DNA-pEi complexes, cells were incubated for 40 hours in the absence or in the presence of NGF, fixed and analyzed by immunofluorescence. Scoring of the morphology of transfected cells, alongside with control EGFP-transfected cultures (Fig. 3A; Fig. 4A), showed that 13.6% of EFA6A-transfected cells displayed neurites, which were rich in filopodia (Fig. 3B; Fig. 4A). Moreover, the majority of EFA6A-transfected cells displayed numerous short protrusions, and increased spreading when compared with non-transfected cells (Fig. 3B).

In EFA6A-transfected cells exposed to NGF, the fraction of neurite-bearing cells was about 25% (approximately equal to the sum of EFA6A-induced and NGF-elicited neurite outgrowth), suggesting that the NGF effect was additive, but not synergistic, with that of EFA6A (Fig. 4A). Neurites displayed a high density of filopodia (Fig. 3B). NGF-treated EFA6A-overexpressing cells often showed multiple neurites (Fig. 3B); this phenotype was observed in 65% of the cells, whereas only 30% of the EGFP-transfected cells had more than one neurite. EFA6A activates ARF6, leading to activation of phosphatidylinositol(4)P 5-kinase and phospholipase D, and of the Rac GTPase (D'Souza-Schorey and Chavrier, 2006). In addition, in melanoma cells, ARF6 has been shown to increase Erk phosphorylation (Tague et al., 2004). Activation of the Ras-MAPK pathway and Rac activation have a major role in NGF-dependent neurite extension in PC12 cells (Reichardt, 2006). Thus, a crosstalk between the NGF- and EFA6A-signaling pathways in neurite outgrowth in PC12 cells might take place.

The 375 mutant (deleted in the N-terminal region), and the ΔC mutant (lacking the 142 C-terminal amino acids) displayed reduced neurite-promoting activity (Fig. 4A).

To evaluate the role of the ARF6 pathway in neurite induction by EFA6A, we generated a putative GEF-defective mutant, by replacing the highly conserved Glu622 in motif 1 with Lys (EFA6A-E622K, represented as GEF^–) (Fig. 1A). A mutation in this position of the Sec7 domain of human EFA6A has been shown to impair GEF activity on ARF6 (Franco et al., 1999). Transfection of GEF^– EFA6A led to a loss of neurite induction by 60.2% (Fig. 4B). To further assess the role of ARF6 in the biological activity of EFA6A in neuronal cells, we cotransfected PC12 cells with a dominant-negative ARF6 mutant (ARF6T27N), and observed a decrease in neurite outgrowth similar to that observed with the GEF^– mutant (Fig. 4B).

Therefore, the ARF6 pathway appeared to have a role in neurite outgrowth induced by EFA6A; however, interfering with the ARF6 pathway did not completely block the biological activity of EFA6A in neuronal cells. Similarly, a constitutively active ARF6 mutant (ARF6Q67L) was weak in inducing neurite outgrowth in PC12 cells, suggesting that further molecular events elicited by EFA6A have a role in the extension of neurites (Fig. 4B). The results observed with the ΔC deletion mutant suggest that, besides the Sec7 domain (endowed with the GEF activity on ARF6), the C-terminal region of the molecule might be engaged in molecular interactions contributing to EFA6A-induced neurite outgrowth.

To explore the function of the Rho family GTPases, which are known to have a fundamental role in the growth and remodeling of neuronal processes (Govek et al., 2005), we cotransfected EGFP-tagged dominant-negative mutants of Cdc42 and Rac1 or constitutively active RhoA, along with EFA6A. Dominant-negative Cdc42N17 or constitutively active RhoAV14 were able to substantially inhibit neurite outgrowth by EFA6A in PC12 cells, whereas coexpression of dominant-negative Rac1N17 resulted in a partial inhibition (64.4%) of EFA6A-induced biological activity (Fig. 4C). Transfection of EFA6As did not induce neurite outgrowth (Fig. 3C, Fig. 4A); in cells treated with NGF, multiple protrusions, resembling short neurites, were observed in about 30% of EFA6As-transfected PC12 cells (Fig. 3C).

Therefore, in PC12 cells, EFA6A overexpression promotes neurite extension, which is inhibited by interfering both with the ARF6 as well as with the Rho-GTPase pathway, whereas EFA6As has a modest biological activity, inducing the formation of protrusions upon exposure to NGF.

To assess the role of EFA6A and EFA6As in primary neurons, E18.5 rat cortices were dissociated, and cells were transfected after 2 days in culture. Cultures were fixed 20-24 hours later, and the morphology of transfected neurons was evaluated after immunofluorescence staining. The majority of EGFP-transfected control cells displayed a pyramidal morphology, with a long apical dendrite, and 4-5 short basal dendrites (Fig. 5A,B), whereas only about 7% of the neurons had a nonpyramidal phenotype, characterized by 5-6 extended processes (Fig. 6A). By contrast, in EFA6A-transfected cells, the fraction of nonpyramidal neurons was highly increased, to 53% (Fig. 5C,D; Fig. 6A). The N-terminal deletion 375 mutant also promoted a substantial increase of nonpyramidal neurons, whereas only 17.8% of the neurons expressing the ΔC mutant, lacking the C-terminal 142 amino acids, displayed multiple extended processes.

To assess the involvement of ARF6 in EFA6A activity in primary neurons, cells were transfected with the GEF^– mutant or cotransfected with EFA6A and dominant-negative ARF6. The fraction of nonpyramidal neurons was significantly reduced both in cells transfected with the GEF^– mutant, and upon cotransfection with EFA6A and dominant-negative ARF6, indicating a role for the ARF6 pathway in the extension of processes (Fig. 6B). Moreover, in cells transfected with the GEF^– mutant, or cotransfected with EFA6A and dominant-negative ARF6, increased dendrite branching was observed (Fig. 6D). This is in line with the studies of Hernandez-Davies and colleagues, who found increased dendritic branching upon transfection of hippocampal neurons with a dominant-negative mutant of the ARF6-GEF ARNO (Hernandez-Deviez et al., 2004).

As Rho family GTPases have been shown to have a fundamental role in growth and remodeling of neuronal processes in primary cortical cultures (Threadgill et al., 1997), we assayed the effect of cotransfecting dominant-negative Cdc42, Rac1 or constitutively active RhoA on the phenotype induced by EFA6A. In cells cotransfected with EFA6A and RhoAV14, no increase in the fraction of nonpyramidal neurons was observed (Fig. 6C), indicating that low RhoA activity is required for processes extension induced by EFA6A. A significant decrease in the fraction of nonpyramidal neurons was also observed upon cotransfection of Cdc42N17 or Rac1N17 (Fig. 6C).

Transfection of EFA6As did not result in changes in the fraction of nonpyramidal neurons (Fig. 6A), whereas a significant increase in the branching of the basal dendrites was observed in transfected cells (Fig. 5E,F; Fig. 6E). Cotransfection of Cdc42N17 or RhoAV14 resulted in a drastic reduction of dendrite branching (Fig. 6E). In cells cotransfected with Rac1N17, dendrite branching induced by EFA6As was also significantly reduced (Fig. 6E).

Therefore, in primary neurons the two isoforms have distinct biological activities: EFA6A promotes neurite extension, which requires both ARF6 and Rho-GTPase activity, whereas EFA6As induces dendrite branching that is significantly reduced by interfering with the Rho-GTPase pathway. Representative images of cortical neurons transfected with the different constructs are shown in supplementary material Figs S2 and S3.

In line with the results of the overexpression studies, siRNAs directed against EFA6A led to a significant increment in the fraction of transfected neurons displaying increased neurite branching (Fig. 7A; supplementary material Fig. S4). Upon interfering with the EFA6As transcript, an increase in the fraction of neurons with a nonpyramidal multipolar morphology was observed (Fig. 7B; supplementary material Fig. S4).

We addressed the biological effect of co-overexpression of EFA6A and EFA6As in primary cortical neurons (supplementary material Fig. S5). The fraction of neurons cotransfected with both EFA6A and EFA6As displaying a nonpyramidal morphology was lower than that observed in cells transfected with EFA6A (Fig. 8A); this reduction appeared more substantial in cells transfected with a higher EFA6As:EFA6A ratio. These data indicate that, when expressed at high levels, EFA6As could inhibit EFA6A biological activity.

However, in primary cortical neurons cotransfected with EFA6A and EFA6As, a significant reduction of EFA6As-induced dendrite branching was observed, suggesting that coexpression of EFA6A could interfere with the biological activity of the short isoform (Fig. 8B). Together, these data suggest a possible role of the balance between EFA6A and EFA6As expression levels in the regulation of neuronal morphogenesis.

To investigate whether the short isoform could have a role in modulating the EFA6A GEF activity toward ARF6, we performed ARF6 activation assays in transfected HeLa cells. Cells were transfected with wild-type ARF6 and the N-terminal deletion 375 construct (Fig. 1A) or EFA6A, and the effect of cotransfection of EFA6As on the levels of ARF6-GTP was tested. As shown in Fig. 8C, the levels of activated ARF6 were not significantly changed by coexpression of EFA6As, indicating that the short isoform does not modulate ARF6 activation elicited by the EFA6A Sec7 domain.

In this study we report that EFA6A encodes two polypeptides, EFA6A and EFA6As, which each have distinct biological activities. In HeLa cells, EFA6A overexpression gives rise to membrane ruffles and to loss of stress fibers, whereas overexpression of EFA6As, the short isoform that we identified, induces cell elongation and actin-rich protrusions.

Induction of membrane ruffles and loss of stress fibers upon human EFA6A overexpression have been reported in previous studies in epithelial CHO cells (Franco et al., 1999), as well as in fibroblastic BHK cells (Derrien et al., 2002). The human construct used in those studies lacks the 374 amino acids at the N-terminus, similarly to our 375 mutant, which also induces ruffles and loss of stress fibers: these data together indicate that the N-terminal region does not appear to significantly contribute to cytoskeletal remodeling induced by EFA6A.

In neuronal cells, the cell type that physiologically expresses the EFA6A-encoded polypeptides, overexpression of EFA6A induces neurite extension, whereas EFA6As leads to increased dendrite branching in primary cortical neurons. This is the first report describing the biological activity of wild-type EFA6A in neuronal cell lines and in primary cortical neurons at early stages in culture. We have observed that neurite extension by EFA6A is significantly decreased by cotransfection of dominant-negative ARF6, or by impairing the EFA6A GEF activity upon specific mutation of the Sec7 domain.

Extension and elaboration of neuronal processes require coordination of directed membrane growth, modulation of adhesion, and dynamic changes in the cytoskeleton, biological activities that the ARF6-GTPase appears to modulate in a variety of cell systems (Donaldson, 2003). Several studies have indicated a role for ARF6 in neurite development: ARF6 is expressed both in developing and in adult brain (Suzuki et al., 2001), and overexpression of dominant-negative mutants in neuronal cells affects neurite development (Albertinazzi et al., 2003; Hernandez-Deviez et al., 2002; Hernandez-Deviez et al., 2004). Moreover, recent studies indicate a role for ARF6 also in the regulation of dendritic spine formation (Miyazaki et al., 2005; Choi et al., 2006).

The GEF activity of EFA6A on ARF6 is not enough to efficiently induce extension of neuronal processes, in line with the observation that constitutively active ARF6 does not promote neurite extension in hippocampal or in retinal neurons (Hernandez-Deviez et al., 2002; Albertinazzi et al., 2003). Neurite extension by EFA6A requires additional biochemical interactions mediated by the C-terminal region. The C-terminal region of EFA6A has been shown in vitro to interact with the K⁺ channel TWIK1 (Decressac et al., 2004), and in the recent study by Sakagami and co-workers, it has been reported to interact with the actin-binding protein α-actinin (Sakagami et al., 2007).

The short isoform EFA6As that we cloned and characterized promotes increased branching of the basal dendrites, similarly to that observed upon transfection of the EFA6A GEF^– mutant (Sakagami et al., 2004) (and this study), suggesting that indeed the main functional difference between the two isoforms of EFA6A lies in the GEF activity displayed by the Sec7 domain. EFA6As is likely to act as a scaffold protein, and elicit its effects on cytoskeletal dynamics via the recruitment of interacting molecules.

The Rho-GTPase pathway, which has a fundamental role in regulating the morphogenesis of neuronal processes (Govek et al., 2005), appears to be involved in the biological activities of both EFA6A and EFA6As, which are blocked by overexpression of dominant-negative Cdc42 or constitutively active RhoA, and significantly reduced upon cotransfection of dominant-negative Rac1. The C-terminal region, common to EFA6A and EFA6As, is essential for their biological activity in transfected cells, and it is probably involved in protein-protein interactions, which could impinge on the Rho-GTPase pathway.

Neurotrophins and other secreted factors, neuronal activity, adhesion molecules, changes in the actin and microtubule cytoskeleton all have an important role in the formation and/or stabilization of dendritic branches, a process that is essential for the establishment of the neuronal circuitry (Jan and Jan, 2003). Increased dendritic (Hernandez-Deviez et al., 2002) and axonal (Hernandez-Deviez et al., 2004) branching has been observed upon expression of catalytically inactive ARNO, or dominant-negative ARF6, suggesting a role for ARNO and ARF6 signaling in negatively regulating neurite branching. As EFA6As overexpression does not appear to modulate the levels of active ARF6, EFA6As-promoted dendrite branching in primary cortical neurons could involve negative modulation the ARF6-signaling pathway by interfering with its effectors.

What is the functional significance of the two different isoforms encoded by EFA6A? The two isoforms of EFA6A appear to be conserved in evolution, because, in the NCBI database, ESTs encompassing the 5′ region of EFA6As are also detected in chicken, fish and Xenopus. The transcripts encoding EFA6A and EFA6As are detected in developing and adult brain, with a distinct temporal regulation: mRNA encoding the EFA6A isoform peaks at postnatal day 4-8, whereas expression of the 2 kB transcript, encoding EFA6As, is highest in mature neural tissue (Sakagami et al., 2004) (and our data). The expression data, along with the biological activity of the two isoforms in neuronal cells, could suggest distinct functions for these proteins in the formation and remodeling of the neuronal processes. The role of EFA6A could be that of promoting processes elongation at earlier stages of development, whereas EFA6As activity could contribute to elaboration and remodeling of the arbor at later embryonic and postnatal stages of neural development. In adult life, EFA6As might be involved in formation and remodeling of dendritic spines. The expression of EFA6A and EFA6As appears to be driven by different promoters (G.F., unpublished), and thereby the transcripts encoding the two isoforms could be subjected to a distinct regulation by external stimuli.

We have observed that when EFA6A and EFA6As are co-overexpressed in transfected cortical neurons, at high levels of EFA6As expression, neurite extension, and thereby increase in the fraction of nonpyramidal neurons promoted by EFA6A, is inhibited. Coexpression of EFA6As does not modulate the GEF activity of the EFA6A Sec7 domain on ARF6, suggesting that EFA6As does not act as a dominant-negative inhibitor of EFA6A by interfering with its ability to activate the ARF6 GTPases. However, EFA6As might compete with EFA6A for binding to the same interactor/s via the C-terminal region, which is identical in the two isoforms. Indeed, dendrite branching promoted by EFA6As is significantly reduced by coexpression of EFA6A, supporting the hypothesis that the two isoforms could compete for common effectors. The temporal regulation of the expression levels of EFA6A and EFA6As in neuronal cells in vivo could contribute to the regulation of neuromorphogenesis during development and differentiation. Owing to the significant sequence conservation of the C-terminal region of the EFA6 family members (Derrien et al., 2002), it is also possible that EFA6As could modulate the biological activity of other coexpressed EFA6 proteins.

The finding that the polypeptides encoded by the EFA6A mRNA transcripts, which are regulated during neural development, are endowed with distinct biological activities in neuronal cells, support a role for EFA6A in the regulation of neuronal morphogenesis. Further studies will clarify the molecular mechanisms underlying the function of EFA6A and EFA6As in neuronal cell arborization.

To isolate murine EFA6A, we screened mouse embryo cDNA libraries using the 3′ region of human EFA6A (Perletti et al., 1997) as a probe. Several positive clones were isolated and characterized. The longest clone isolated was of 2.3 kb, and lacked an initiation methionine codon and an in frame stop codon. For the isolation of the complete sequence of EFA6A cDNA (3785 bp), 5′ RACE was performed using specific oligonucleotide primers and murine brain polyA⁺ RNA. Murine EFA6As cDNA was isolated from an E11.5 mouse embryo cDNA library upon hybridization with the 3′ region of human EFA6A.

EFA6A and EFA6As cDNAs were tagged at the C-terminus with a FLAG epitope by PCR, sequenced and cloned in the pcDNA3 expression vector; 375 and ΔC FLAG-tagged mutants were generated by PCR using specific primers, sequenced and cloned in pcDNA3; EFA6A-E622K (GEF^–) was obtained mutagenizing the FLAG-tagged EFA6A cDNA by PCR using specific primers and the QuikChange site-directed mutagenesis kit from Stratagene.

Murine ARF6 was isolated by PCR from murine brain cDNA, Myc tagged at the C-terminus, and cloned in the pcDNA3 expression vector; to generate the ARF6T27N dominant negative and ARF6Q67L constitutively active mutants, mutagenesis was performed with specific oligos using the QuikChange site-directed mutagenesis kit from Stratagene.

EGFP-tagged dominant-negative Rac1 and Cdc42 and constitutively active RhoA in the pCB6 expression vector were a kind gift from Michael Way (Cancer Research UK London Research Institute, London, UK).

Total RNA from mouse tissues was purified using a standard guanidinium thiocyanate-phenol-chloroform extraction procedure. RNA was separated on 1% agarose gels and transferred to Hybond N nylon membranes by capillary transfer; hybridization was performed using as a probe a 520 bp fragment of murine EFA6A (bp 2722-3242) labeled with ³²P by random priming.

Rabbit polyclonal anti-EFA6A antibodies were generated upon immunization with a 13 amino acid peptide corresponding to amino acids 988-1000 of murine EFA6A or with a histidine-tagged fusion protein encompassing the C-terminal 155 amino acids, followed by purification of IgGs by affinity chromatography on protein-A-Sepharose beads.

Anti-FLAG polyclonal antibody and anti-MAP2 monoclonal antibody HM-2 were from Sigma, anti-Myc monoclonal antibody 9E10 was from Santa Cruz Biotechnology, anti-GFP monoclonal antibody was from Roche. Alexa-Fluor-594 and -488 goat anti-rabbit and goat anti-mouse IgGs, Alexa-Fluor-488 and -594 phalloidin were from Molecular Probes.

Tissues were dissected, snap frozen in liquid nitrogen, pulverized, lysed in RIPA buffer [10 mM Tris-HCl (pH 7.4), 0.15 M NaCl, 1% sodium deoxycholate, 1% NP40, 0.1% SDS, 1 mM EDTA, 10 mM KCl, protease inhibitors] or 50 mM Tris-HCl (pH 7.4)-2% SDS, centrifuged at 10,000 × g, and the protein concentration of the supernatant was estimated. HeLa cells (4×10⁵ cells/60 mm plate) were transfected by pEi (pEi; average M_r ∼25,000, Aldrich), with 10 μg EGFP, EFA6A or EFA6As expression plasmids, and lysed the next day in RIPA buffer. 25-50 μg protein lysate/sample were adjusted into gel loading buffer (50 mM Tris-HCl, pH 6.8, 100 mM DTT, 2% sodium dodecyl sulfate, 10% glycerol, 0.1% bromophenol blue), heated at 90°C for 5 minutes, separated on 8% SDS-polyacrylamide gels, transferred to polyvinylidene difluoride (PVDF). Membranes were stained with Ponceau red to control for even loading, destained and processed according to standard procedures. The anti-EFA6A antibodies were diluted in 5% non-fat dry milk at 1 μg/ml.

HeLa cells were grown in DMEM supplemented with 10% fetal calf serum. Cells (2×10⁴/well in 24-well plates) were plated on collagen-coated glass coverslips and transfected the next day using polyethylenimine (pEi; average M_r ∼25,000, Aldrich), with 1 μg plasmid DNA per well. After 3-4 hours of incubation with the DNA-pEi complexes, fresh medium was added, and 20-24 hours later cells were fixed with 3% paraformaldehyde and processed for immunofluorescence. Loss of stress fibers and cell elongation were evaluated in several independent experiments; 50 transfected cells per construct were scored for each experiment. Statistical analysis was performed by one-way ANOVA.

PC12 cells were grown in DMEM supplemented with 10% horse serum and 5% fetal calf serum. For differentiation, cells were grown in DMEM supplemented with 2% horse serum, 1% fetal calf serum and 50 ng/ml NGF (Sigma). Cells (7×10⁴/well in 24-well plates) were plated on collagen-coated glass coverslips and transfected the next day by pEi, with 1-2 μg plasmid DNA per well. After 3-4 hours of incubation with the DNA-pEi complexes, fresh medium (growth or differentiation medium) was added, and 40 hours later cells were fixed and processed for immunofluorescence. Neurites were scored when their length was at least double the diameter of the cell body. Data were collected from several independent transfection experiments; at least 50 transfected cells per construct were evaluated for each experiment; the results are presented in percent ratios ± s.d. Statistical analysis was performed by one-way ANOVA followed by Tukey's post-hoc test for multiple comparisons. Significance level was taken as P<0.05.

To establish cultures of primary cortical neurons, cortices from six to eight E18.5 CD rat embryos were dissected, fragmented and processed as described (Treadgill et al., 1999). Cells were counted, and 1×10⁶ cells/well in 24-well plates were plated on glass coverslips coated with laminin and poly-L-lysine in Eagle's basal medium (BME) supplemented with 5% fetal calf serum and 1% N2 supplement (Gibco). Two days later, cells were transfected by pEi (1-2 μg plasmid DNA per well, 2-3 hours of incubation with the DNA-pEi complexes). Cells were fixed and processed for immunofluorescence 20-24 hours later. Neuronal cells were identified by staining with anti-MAP2 monoclonal antibody. Processes were scored when their length was five times or more the diameter of the cell body. Data were collected from several independent transfection experiments; at least 50 transfected neurons per construct were evaluated for each experiment; the results are presented in percent ratios ± s.d. Collaterals were scored as branches when their length was equal or more than half the diameter of the cell body. Statistical analysis was performed by one-way ANOVA followed by Tukey's post-hoc test. Significance level was taken as P<0.05.

Cells were fixed in 3% paraformaldehyde and 2% sucrose in PBS for 10 minutes at room temperature, permeabilized in 0.2% Triton X-100, incubated with primary antibodies diluted in PBS with 0.2% BSA for 1 hour at 37°C, washed three times with PBS-0.2% BSA and incubated in 2% BSA at 37°C for 15 minutes. Cells were then incubated with fluorescently labeled secondary antibodies and/or fluorescent phalloidin diluted in PBS-0.2% BSA for 1 hour at 37°C, washed three times with PBS-0.2% BSA, stained with Hoechst 33258, and mounted on glass slides with Mowiol 4-88 (Calbiochemical). Images were acquired using a Zeiss Axiophot microscope and a Hamamatsu C4742-95 digital camera, using the Hipic32 software.

Chemically synthesized, duplex siRNAs were purchased from Eurofins MWG. The following RNA interference sequences were used: siEFA6A1, 5′-ATTGGTGGCTGGCGAGTAT-3′ and siEFA6A2, 5′-GAACAATGACTTCAGCAAA-3′ of rat EFA6A mRNA (GenBank accession no. XM_001066749); siEFA6AsA, 5′-GCATGATCGGCGTCAACAG-3′ and siEFA6AsB, 5′-GCCGCCTGCAGAGCCGCAA-3′ of rat EFA6As (AC 096363.8). As a control, the Luciferase GL2 siRNA, 5′-CGTACGCGGAATACTTCGA-3′ (siLuc2) was used. Complementary small interfering RNA oligonucleotides were annealed, and 100 nM of the siRNA oligonucleotides were cotransfected with EFA6A or EFA6As expression plasmids in HeLa cells or in primary neurons by Lipofectamine 2000, to test the efficiency of the siRNAs in silencing EFA6A or EFA6s expression.

Primary cortical neurons (1.2×10⁶ cells/well) were cotransfected with siRNAs and pEGFP (100 nM siRNA oligonucleotides, 50 ng pEGFP plasmid per well in 24-well plates) by Lipofectamine 2000. Cells were processed for immunofluorescence with anti-GFP and anti-MAP2 antibodies 24 hours later. GFP-positive neurons (50 cells per experiment) were scored for morphology.

The assay was performed following the protocol as described (Klein et al., 2006). HeLa cells (1.4×10⁶ per 60 mm plate) were transfected with Lipofectamine 2000 (Invitrogen), and 20 hours later, cells were lysed on ice with 500 μl of lysis buffer [1% Triton X-100, 50 mM Tris-HCl (pH 8), 100 mM NaCl, 10 mM MgCl₂, 0.05% sodium deoxycholate, 0.005% SDS, 10% glycerol, 2 mM dithiothreitol, and protease inhibitors]. Lysates were clarified by centrifugation at 13,000 × g for 10 minutes and incubated with 30 μg GST-ARHGAP10 Arf-binding domain (Dubois et al., 2005) bound to glutathione-Sepharose beads (Amersham Biosciences) supplemented with 0.5% BSA for 40 minutes at 4°C. The beads were washed three times with lysis buffer without sodium deoxycholate and SDS, and bound proteins were eluted in 30 μl of SDS sample buffer. The presence of Arf6-GTP was detected by immunoblotting using an anti-Myc monoclonal antibody (9E10, Santa Cruz Biotechnology).