Previous studies on small GTP-binding proteins of the Rho family have revealed their involvement in the organization of cell actin cytoskeleton. The function of these GTPases during vertebrate development is not known. With the aim of understanding the possible role of these proteins during neuronal development, we have cloned and sequenced five members expressed in developing chick neural retinal cells. We have identified four chicken genes, cRhoA, cRhoB, cRhoC, and cRac1A, homologous to known human genes, and a novel Rac gene,cRac1B. Analysis of the distribution of four of the identified transcripts in chicken embryos shows for the first time high levels of expression of Rho family genes in the vertebrate developing nervous system, with distinct patterns of distribution for the different transcripts. In particular, cRhoA andcRac1A gene expression appeared ubiquitous in the whole embryo, and the cRhoB transcript was more prominent in populations of neurons actively extending neurites, whereas the newly identified cRac1B gene was homogeneously expressed only in the developing nervous system. Temporal analysis of the expression of the five genes suggests a correlation with the morphogenetic events occurring within the developing retina and the retinotectal pathway. Expression of an epitope-tagged cRac1B in retinal neurons showed a diffuse distribution of the protein in the cell body and along neurites.
Taken as a whole, our results suggest important roles for ubiquitous and neural-specific members of the Rho family in the acquisition of the mature neuronal phenotype.
- Rho GTPases
- neuronal development
- chick embryo
- neural retinal cells
- retinotectal pathway
- dorsal root ganglia
The actin cytoskeleton plays a fundamental role in several aspects of cell life, including adhesion, migration, and cytokinesis. Several actin binding proteins take part in the organization of the actin cytoskeleton and contribute to its dynamic properties. Recent studies have shown that components of the Rho family, which belong to the Ras superfamily of small GTPases, are involved in the reorganization of the actin cytoskeleton and of the associated sites of cell adhesion to the extracellular matrix (Hall, 1994). In particular, microinjection experiments have demonstrated that RhoA is essential for the assembly of focal adhesions and the associated actin stress fibers (Ridley and Hall, 1992) and that Rac1 is required for growth factor-induced membrane ruffling (Ridley et al., 1992), whereas Cdc42 triggers the formation of filopodia (Nobes and Hall, 1995). Neurite outgrowth can be considered as a particular form of cell motility in which actin dynamics during growth cone navigation evolves into stabilization of the cytoskeleton and neurite elongation (Tanaka and Sabry, 1995). The behavior of growth cones can therefore be compared with that of the leading edge of spreading or migrating fibroblasts, in which the dynamic adhesive interactions with the substrate are accompanied by a continuous reorganization of the actin cytoskeleton.
In line with this interpretation, recent evidence has accumulated that suggests a role for the Rho family GTPases in neuritogenesis (Mackay et al., 1995; Luo et al., 1996). In fact, activation of Rho proteins by lysophosphatidic acid, thrombin, or sphingosine-1-phosphate leads to growth cone collapse and retraction of neurites in N1E-115 neuroblastoma cells (Jalink et al., 1994; Postma et al., 1996), and these effects can be prevented by pretreatment of the cells with theClostridium botulinum C3 exoenzyme, which specifically ADP-ribosylates Rho proteins. Furthermore, two Drosophilahomologs of the Rho family GTPases, Drac1 and Dcdc42, are highly expressed in the Drosophila developing nervous system, and mutants of these proteins cause distinct defects in neuronal development (Luo et al., 1994). These data raise the possibility that components of the Rho family of small GTPases may play a role in neuronal development in vertebrates as well. So far, however, the nature and distribution of the different GTPases of the Rho family expressed during vertebrate development remain undefined. With the aim of studying the role of Rho GTPases in the development of the neuronal phenotype, we have now identified by molecular cloning various components expressed in chicken developing neurons. The identified cDNAs have been used for in situ hybridization analysis to look at the expression of the corresponding transcripts in the entire chicken embryo. The data presented in this paper show that developing retinal neurons express the mRNAs of at least five different components of the Rho family of GTPases, including a new Rac protein, which are differentially expressed during the development of the neural retina. Furthermore, the analysis of the distribution of four of the identified transcripts shows that they are strongly and specifically expressed in various areas of the developing CNS and peripheral nervous system.
MATERIALS AND METHODS
Reagents. Fertilized chicken eggs were purchased from Allevamento Giovenzano (Vellezzo Bellini, Italy). Taqpolymerase was from Promega (Madison, WI), Klenow fragment of DNA polymerase was from Pharmacia (Uppsala, Sweden), and restriction enzymes were from Boehringer Mannheim (Mannheim, Germany). [α-35S]dATP and [α-32P]dCTP were from Amersham (Buckinghamshire, UK). Other chemicals were purchased from Sigma–Aldrich (Milan, Italy). Laminin was purified from Engelbreth–Holm Swarm sarcoma as published (Timpl et al., 1979).
Isolation of PCR clones. Total RNA was extracted from E6 retinas by the RNAzolB method (Chomczynski and Sacchi, 1987). Five micrograms of total RNA were converted into complementary DNAs by oligo-dT-priming [oligo-dT12–18; Life Technologies-BRL, Milano, Italy] in the presence of Moloney murine leukemia virus reverse transcriptase (Life Technologies-BRL). Aliquots of the resulting single-stranded cDNA (7.5 ng) were mixed with pairs of degenerate synthetic oligonucleotides and subjected to thermal cycling. Two different sets of degenerate oligonucleotides were used for the amplifications: a set containing the sequences 5′-TTYWSMAARGAYCAGTTCCC (RhoA-1) and 5′-TCACBGGYTCCTGYTTCAT (RhoA-2) coding for amino acid positions 25–31 and 134–140, respectively, of the human RhoA protein, and a more degenerate set of oligonucleotides containing the sequences 5′-AARACNTGYYTNCTSAT (RhoF-1) and 5′-GCHGARCAYTCVADRTA (RhoF-2) coding for amino acid positions 18–23 and 156–161, respectively, of different Rho proteins. PCRs contained 50 μl of 50 mm KCl, 10 mm Tris-HCl, pH 9.0, 1.5 mm MgCl2, 0.1% Triton X-100, 200 μm desoxyribonucleotide triphosphate (dNTP), 100 pmol of each oligonucleotide, 2.5 U of Taq polymerase (Promega), and 1.5 μl of cDNA. The first five cycles were performed under low stringency annealing conditions (94°C, 1 min; 43°C, 1 min; 72°C, 1 min); the following 40 cycles were performed at higher stringency (94°C, 1 min; 50°C, 1 min; 72°C, 1.5 min). The cRhoC clone was isolated by using the “touchdown” PCR technique with RhoF1 and RhoF2 primers (Don et al., 1991). Amplified DNA fragments were subcloned into a pBluescript KS− T-vector (Marchuk et al., 1991). Plasmid clones containing inserts of appropriate size were subjected to direct sequencing by the dideoxy method (Sanger et al., 1977) and analyzed by the GCG Wisconsin Sequence Analysis Package.
Isolation of cDNA clones from λgt10 libraries. Four full-length clones and a partial clone coding for the different chicken Rho and Rac proteins were isolated either from an embryonic day (E) 10 chicken or an E13 chicken brain cDNA library in λgt10 [obtained from Dr. C. Nottenburg (Fred Hutchinson Cancer Research Center, Seattle, WA) and Dr. B. Ranscht (Cancer Research Institute, La Jolla, CA), respectively]. The libraries were plated on Escherichia coli strain LE392, and replica filters were screened in duplicate at high stringency according to a modified procedure of Church and Gilbert (1984). Prehybridization was performed for 3 hr at 65°C in hybridization buffer (125 mmNa2HPO4, 1 mm EDTA, 250 mm NaCl, 7% SDS, 10% PEG-8000, 1% BSA, 100 μg/ml denatured salmon sperm DNA). Hybridizations were performed at 65°C in the same buffer in the presence of 0.5–1 × 106 cpm/ml of 32P-labeled probe. Washings were in 0.2× SSC at 65°C. The cDNAs corresponding to the different chicken PCR products were labeled by random priming (Feinberg and Vogelstein, 1983) at a specific activity between 7 × 108 and 1.2 × 109 cpm/μg. cDNAs inserts from positive purified phages were extracted byEcoRI digestion and subcloned into pBluescript KS−. Both strands were sequenced with the T7 sequencing kit (Pharmacia, Uppsala, Sweden) using different specific oligonucleotide primers.
Northern blot analysis. Total RNA was prepared from E6, E8, E10, and E12 chick neural retinas by the RNAzolB method (Chomczynski and Sacchi, 1987). Northern blot analysis of total RNA (20 μg/lane) was performed as described previously (Lehrach et al., 1977; Malosio et al., 1991). Hybridizations and washes were performed under the same high stringency conditions used for the screening of the libraries. Hybridizations took place in hybridization buffer supplemented with32P-labeled probes (0.5–1.0 × 106cpm/ml) for 12–16 hr at 65°C. After high stringency washes (0.2 × SSC at 65°C), x-ray films were exposed for 3–7 d to the hybridized filters. RNA blots were reprobed with an 18 S ribosomal RNA probe. Quantitations of hybridized bands were performed by computer densitometry (Molecular Dynamics, Sunnyvale, CA). The values for the different GTPase transcripts were normalized to the corresponding values obtained for the 18 S ribosomal RNA.
In situ hybridization. The 370- to 400-bp-long cDNAs obtained by PCR, coding for the different Rho proteins, were used for the preparation of sense and antisense RNA probes to be used forin situ hybridizations. The specificity of hybridization of these probes had been assessed previously by Northern blotting. After linearization of the pBluescript plasmids with the appropriate restriction enzyme, T3 and T7 polymerases were used to generate high specificity 35S-labeled riboprobes by incorporation of [α35S]rUTP (Amersham, Arlington Heights, IL) into the transcribed RNA (RNA transcription kit, Stratagene, La Jolla, CA). The DNA template was removed by digestion with DNase I and the labeled RNA probe was purified through a Bio-spin column (Boehringer Mannheim). The purified probe was supplemented with 20 mm DTT, and an aliquot was counted in scintillation fluid.
For in situ hybridization, paraffin sections of chick embryos were dewaxed, deproteinated, and post-fixed in 4% paraformaldehyde (Rugarli et al., 1993). The 35S-labeled riboprobes were diluted at 4.2–6.3 × 107cpm/ml (∼60 ng/ml) in hybridization buffer containing 50% (v/v) formamide, 0.3 m NaCl, 10 mm Tris-HCl, pH 7.6, 10 mm NaH2PO4, pH 6.8, 5 mm EDTA, pH 8.0, 0.2% (w/v) Ficoll 400, 0.2% (w/v) polyvinylpyrrolidone, 10% (w/v) dextran sulfate, 50 mmDTT, 0.5 mg/ml poly ribo A, and 50 μg/ml yeast tRNA, and added to the sections. Sections were incubated overnight at 55°C in a humidified chamber. Stringency washes at 64°C included several washes in 2× SSC, 50% formamide, 20 mm β-mercaptoethanol, in 4× SSC, 20 mm Tris-HCl, pH 7.6, 1 mm EDTA, and 30 min incubation with RNase A (10 μg/ml). Slides were air-dried and exposed to x-ray films for 3–6 d. Subsequently, slides were dipped into Kodak NTB-2 emulsion and exposed at 4°C for 15–21 d. Sections hybridized to sense probes for the different genes were processed in parallel and used as controls for nonspecific hybridization. After development, the slides were counterstained with Hoechst 33258 and analyzed by dark-field illumination and by UV fluorescence.
Cell culture. Cultures enriched in retinal neurons and in retinal glia (Biscardi et al., 1993) were prepared from E7 chick retinas. Neural retinas were dissected and trypsinized, and cultures of retinal neurons were obtained under serum-free conditions as described (de Curtis et al., 1991). After 18 hr in culture, neuronal cells were used to prepare total RNA as described above. For glial cells, cells from trypsinized retinas were cultured in DMEM with 5% fetal calf serum. Confluent monolayers were transferred to new culture dishes to dilute neurons; remaining neurons were washed off the glial monolayers. Neuron-free monolayers were then used for total RNA preparation as described above.
Expression of cRac1B in retinal cells. The full-length cDNA for cRac1B was subcloned into pcDNA-I-Amp vector (Invitrogen, Carlsbad, CA) containing a sequence including the YDVPDYA amino acids of the influenza hemagglutinin (HA), and the pcDNA-I-HA-Rac1B plasmid obtained was used for transfections of primary retinal cells.
For transfections, we used a protocol modified from Boussif et al. (1995). Approximately 300,000 retinal cells obtained from E6 chick neural retinas were plated in each 1.5-cm-diameter well containing a glass coverslip coated with 200 μg/ml poly-d-lysine and 40 μg/ml laminin. Cells were cultured overnight at 37°C, 5% CO2 as described (de Curtis et al., 1991), to induce neurite extension. Cells were then incubated with 200 μl/well of 150 mm NaCl containing 150 nmol of polyethylenimine (PEI 50 kDa; Sigma) and 5 μg of pcDNA-I-HA-Rac1B plasmid in 0.5 ml of transfection medium [50% retinal growth medium (RGM), 50% DMEM, and 5% fetal calf serum]. After 3 hr of culture, the medium was replaced with serum-free RGM, and the cells were cultured for an additional 24 hr. Cells were then fixed with paraformaldehyde or with cold (−20°C) methanol and processed for indirect immunofluorescence as described byCattelino et al. (1995). Cells were incubated for 1 hr at room temperature with the following primary antibodies: a monoclonal antibody against the HA-tag, a polyclonal antibody against the 200 kDa neurofilament protein (Sigma), and a polyclonal antibody against the extracellular portion of the integrin α6 subunit (de Curtis and Reichardt, 1993). Cells were subsequently incubated for 30 min with TRITC-conjugated sheep anti-mouse IgG together with FITC-conjugated sheep anti-rabbit IgG (Boehringer Mannheim) and observed using a Zeiss-Axiophot microscope.
Cloning of five Rho family members expressed in chicken embryonic neural retina
Our first aim has been the identification of members of the Rho family of GTPases expressed in developing neurons. For this purpose we have used RT-PCR using two sets of degenerate oligonucleotides to amplify fragments of transcripts of Rho family genes from RNA prepared from developing chick retinas, which have then been used to isolate cDNA clones from λgt10 cDNA libraries.
Fragments of Rho family cDNAs were amplified by PCR from cDNAs prepared from E6 chick neural retina mRNAs. The PCR reactions were performed in the presence of either one of two sets of degenerated oligonucleotides: the oligonucleotides RhoA-1 and RhoA-2 corresponding to the FSKD(Q/E)FP and MKQEPV(K/R) peptides, specific for the human RhoA, B, and C proteins, and the oligonucleotides RhoF-1 and RhoF-2 corresponding to the KTCLLI and Y(L/M/V)ECSA peptides, specific for all known human Rho family members. Restriction analysis and sequencing of ∼100 cDNA fragments obtained by PCR identified five different DNA sequences encoding proteins with a high degree of homology to known human Rho family members. The five different PCR fragments were used to screen two λgt10 cDNA libraries, one from E10 chick embryo and one from E13 chick brain. In this way, several λ phage clones were found that contain coding regions corresponding to the five identified chicken Rho family genes. Sequence analysis of the isolated clones (Fig.1), and the comparison with the sequences of known human Rho proteins (Fig.2 B,C), allowed us to identify open reading frames coding for the predicted full-length polypeptides of four of the five genes. Several unsuccessful attempts were made to isolate from two available chick cDNA libraries a full-length clone for a fifth cDNA, for which no 5′ terminal sequence could be found (Fig. 2 A, cRhoC).
Comparison at the nucleotide and polypeptide levels of the five chicken genes with the human Rho and Rac sequences (Fig.2 B,C) revealed that three chicken Rho and two chicken Rac homologs had been isolated. We propose to name the five chicken genes cRhoA, cRhoB, cRhoC,cRac1A, and cRac1B. At the amino acid level, cRhoA is 100% identical to human RhoA, whereas cRhoB and cRhoC show 97.5% and 95.4% identity, respectively, to their human counterparts. For cRhoC, comparison with the respective human gene indicated that the sequence corresponding to the 19 amino terminal amino acid residues is missing. Interestingly, both cRac1Aand cRac1B show the highest degree of identity with the human Rac1 sequence (88.5% and 80.7%, respectively). Comparison of the polypeptide sequences derived from the chick clones with the human sequences (Fig. 2 C) confirmed that the cRac1A polypeptide is 100% identical to the human Rac1 protein, whereas the cRac1B polypeptide is 93.7% identical to human Rac1, and only 89% identical to human Rac2. In particular, the C-terminal portion of the cRac1B polypeptide sequence showed a much higher degree of identity for human Rac1 than for human Rac2 (not shown). We propose thatcRac1B represents a new Rac gene.
Developmental regulation of expression of GTPases in the chicken retina
To characterize in more detail the expression of the five identified Rho family transcripts during retinal development, we have analyzed their expression by Northern blot analysis (Fig.3). Filters with total RNA prepared from neural retinas isolated from different developmental stages (E6, E8, E10, and E12) were probed with random-primed 32P-labeled cDNAs. Distinct RNA hybridization patterns were obtained with each probe. A single band corresponding to 2.4 and 1.65 kb transcripts was detected for cRhoB and cRac1B, respectively (Fig.3 B,E), whereas two different bands were detected forcRhoA, cRhoC, and cRac1A (Fig. 3,A, C, and D, respectively). Two different RNA blots were probed for each transcript and quantitated by densitometric scanning (Fig. 4). The data presented in Figure 4 were obtained after normalizing the value for each transcript with the corresponding value obtained after hybridization for the 18 S RNA (not shown). For cRhoA, cRhoC, andcRac1A, quantitations at each developmental stage represent the sum of the two transcripts (Fig. 3 A,C,D). The results show that the expression of the five transcripts is regulated differently during maturation of the retina. In particular,cRhoB and cRac1B transcripts were upregulated during retinal development, whereas the other transcripts were downregulated (Fig. 4).
For the detection and quantitation of the cRhoC transcripts, the blots had to be exposed for autoradiography 10 times longer compared with those for the other transcripts (Fig. 3 C), suggesting that this gene is not as abundantly expressed as the other Rho family genes in the developing neural retina; on the other hand, we found that cRhoC was highly expressed in non-neuronal chick cells (data not shown).
Expression of cRac1B mRNA in neurons and glia from developing neural retina
To check whether the newly identified, neural-specific cRac1B GTP-binding protein was present also in non-neuronal cells of the CNS, we prepared cultures enriched either in neurons or in glial cells from E7 neural retinas, as described in Materials and Methods. Northern blot analysis from gels loaded with the same amount of total RNA isolated from the two different cell preparations showed that similar amounts of the 1.65 kb cRac1B transcript were present in glial cells and neurons at this stage of development (Fig.5, lanes 1 and 2, respectively).
Differential distribution of Rho proteins in developing chicken embryos
The differential expression of the identified Rho family transcripts during retinal development encouraged us to further characterize their expression in the developing chick nervous system byin situ hybridization. Because cRhoC was poorly expressed in the neural retina compared with the other four genes, we limited the distribution studies to the four abundantly expressedcRhoA, cRhoB, cRac1A, and cRac1B genes. Sections obtained from E6.5 and E8.5 chicken embryos were analyzed. The overall pattern of expression at E6.5 showed clear differences among the transcripts (Fig. 6). Those forcRhoA (not shown) and cRac1A (Fig.6 B) were quite homogeneously distributed throughout the embryo, whereas the distribution of cRhoB (Fig.6 A) and cRac1B (Fig. 6 C) transcripts was more restricted. In particular, cRac1Btranscript seemed concentrated in the developing nervous system, including the retina, the tectum, the spinal cord, the dorsal root ganglia (DRGs), and the trigeminal ganglion (Fig. 6 C). The same structures were labeled also by cRhoB,cRac1A (Fig. 6, A and B, respectively), and cRhoA (not shown) antisense probes. At higher magnification, DRGs labeling for cRhoB (Fig.6 D) appeared concentrated on the more dorsal half of the structures, quite different from the distributions ofcRac1A (Fig. 6 F), cRac1B, andcRhoA (not shown), which were homogeneous throughout the ganglia. Differences were also observed in the labeling of the developing tectum. In fact, although the overall distribution of thecRac1A and cRac1B transcripts in this structure appeared homogeneous (Fig. 6, B and C, respectively), at higher magnification cRhoB staining was stronger in the external layer, presumably corresponding to postmitotic neurons derived from the neuroepithelium (Fig.6 H).
Similar to what we observed at E6.5, in E8.5 embryos the patterns of distribution of the transcripts for cRhoA andcRac1A were more homogeneous than those for cRhoBand cRac1B (not shown). In contrast, the cRac1Btranscript was highly concentrated in the developing nervous system (not shown), whereas the distribution of cRhoB, although somewhat more widespread in comparison with cRac1B, showed several interesting features within different structures of the developing nervous system. Figure7 A shows a low-power magnification of a parasagittal section of an E8.5 chicken head, incubated with an antisense probe for cRhoB. Several structures of the developing nervous system, including a very bright area below the eye corresponding to the trigeminal ganglion, express high levels of the transcript. Higher magnification of the developing cerebellum at E8.5 showed that cRhoB was highly concentrated in the presumptive Purkinje cell layer (arrowheads, Fig.7 B), whereas cRhoA was distributed homogeneously throughout the entire region (Fig. 7 D). The distribution of both cRac1A and cRac1B transcripts was similar to that of cRhoA, although the signal was not as strong (not shown). Higher magnification of the eye region showed a clear concentration of the cRhoB transcript in the retinal ganglion cell (RGC) layer of E8.5 retinas (arrows, Fig.8 C). In contrast,cRhoA was distributed homogeneously within the neural retina (Fig. 8 A). The distribution of cRac1B was similar to that of cRhoB, although weaker (not shown), whereas the expression pattern of cRac1A was similar to that of cRhoA (not shown). At E6.5, the distribution of the different transcripts in the neural retina was similar to that observed in E8.5 embryos, although the concentration of the cRhoBtranscript in the RGC layer was not as distinct, probably because of the presence of a less defined ganglion cell layer at this stage (not shown).
The distribution of the transcripts in the spinal cord was analyzed in transversal/oblique sections from E6.5 and E8.5 embryos. Interestingly, three different patterns were revealed by in situ analysis. At E6.5, the expression of cRac1B was quite homogeneous throughout the section of the spinal chord (Fig.9 D). In contrast,cRhoA and cRac1A (Fig. 9, A andC, respectively) were concentrated around the ventricular zone, where proliferation is occurring, and in the ventral area of the spinal chord, where motor neurons are located. Finally, a third pattern was observed for cRhoB (Fig. 9 B), which was highly expressed in the ventral portion of the spinal cord, including the floor plate and the area with motor neurons. At this stage, DRGs visible on the side of the spinal cord were positive for all four tested GTP-binding protein mRNAs. At E8.5 the pattern of distribution of cRhoA and cRac1A was similar to that observed at E6.5, although the differences in the intensity of the signal among distinct areas were not as clear (not shown); a stronger, still homogeneous signal was found for cRac1B (Fig.9 F), whereas the distribution of the cRhoBtranscript seemed more restricted than in E6.5 spinal cord and was localized to the motor neuron region and the floor plate (Fig.9 E).
Distribution of the cRac1B polypeptide in retinal neurons
The distribution of the cRac1B protein in retinal neurons was studied by expressing an epitope-tagged form of the protein. Cultured primary retinal neurons were transiently transfected with the pcDNA-I-HA-Rac1B vector containing the sequence encoding for an HA-tagged cRac1B protein. We obtained the best transfection efficiencies by using PEI 50 kDa on E6 retinal cells that had been cultured for ∼12 hr before treatment. Immunofluorescence with an anti-HA antibody showed that cRac1B was homogeneously distributed in retinal cells (Fig. 10). In particular, cRac1B was also visible along neurites and in all their protrusions. Double immunofluorescence staining was used to identify transfected neurons expressing a 200 kDa neurofilament polypeptide (Fig.10 A,B). Neurofilament-negative cells expressing the HA-Rac1B construct were also present in culture (not shown). The distribution of the epitope-tagged cRac1B in neurons showed a pattern similar to that of the integrin α6 subunit, which is expressed on the surface of these cells (Fig. 10 C,D), suggesting a possible association of the Rac1B polypeptide with the plasma membrane.
Five major conclusions can be drawn from the data presented in this paper. First, developing neural retinal cells express mRNAs coding for at least five components of the Rho family of GTPases: three coding for Rho proteins and two coding for Rac proteins. Second, the comparison of the cDNAs with the human homologs indicates that one of the Rac proteins represents a novel Rac gene. Third, the levels of expression of the five transcripts are differentially regulated during the development of the retina. Fourth, four of the identified transcripts show distinct patterns of distribution in developing chick embryos, with particularly high levels of expression of all the transcripts, and prominent localization of the newly identified cRac1B gene in the developing nervous system. Finally, the expression of an epitope-tagged form of cRac1B in primary retinal neurons reveals a homogeneous distribution of the polypeptide in the cell body and along neurites. These results demonstrate for the first time that Rho family GTPases are highly expressed in the developing CNS and peripheral nervous system, suggesting that these GTPases play an important role during the development of the vertebrate nervous system.
Several extracellular cues, including extracellular matrix glycoproteins, can induce dramatic morphological changes in the developing neurons, which result in the formation of neurites (Sanes, 1989; de Curtis, 1991; Reichardt and Tomaselli, 1991). We have shown previously that the dramatic effects of laminin on neurite extension from retinal neurons in culture are mediated by the α6β1 integrin laminin receptor (de Curtis and Reichardt, 1993). The molecular mechanisms underlying these processes remain poorly understood. Recent studies in non-neuronal cells have shown that Rho family GTPases regulate the formation of actin-based structures such as filopodia, lamellipodia, and stress fibers (Nobes and Hall, 1995). Although stress fibers are not found in growth cones, filopodia and lamellipodia are actin-dependent processes also involved in growth cone navigation. Moreover, recent data postulate the involvement of Rho family GTPases in the regulation of actin-mediated growth cone migration (Jalink et al., 1994; Postma et al., 1996). In addition to their role in cytoskeletal reorganization, Rho family GTPases have been involved in the regulation of the activity of transcription factors and in membrane traffic (Ridley, 1996). Furthermore, a number of possible effectors for these GTPases have been identified recently (for review, see Ridley, 1996).
With the aim of studying the role of these GTPases during the development of the neuronal phenotype, we have looked for cDNA clones of Rho family GTPases expressed in primary neurons. We have used E6 retinas as the source of mRNA for this study, because this is the stage at which cultured retinal neurons respond to laminin by extending neurites. Three of the identified small GTP-binding proteins expressed by neural retinal cells correspond to the chick homologs of the already known human RhoA, RhoB, and RhoC genes (Madaule and Axel, 1985). Two other cDNAs were related toRac genes and predicted two different Rac proteins, one showing complete identity (cRac1A) and the other showing a high degree of identity (cRac1B) to human Rac1. We think that the cRac1B protein does not correspond to the chick homolog of human Rac2, because cRac1B shows a higher degree of identity to the human Rac1 than to human Rac2 at both the nucleotide and protein level. Furthermore, although theRac1 gene is known to be expressed in various tissues and cell lines, the expression of Rac2 is restricted to cells of the hemopoietic lineage (Didsbury et al., 1989; Shirsat et al., 1990;Moll et al., 1991).
The temporal expression of the transcripts coding for the different identified chicken GTPases during the development of the neural retina has been investigated. Our data show that the five transcripts are differentially regulated between E6 and E12; in fact, although the expression of cRhoA decreased after E8 and that of cRac1A and cRhoC decreased after E10, the expression of cRac1B and cRhoB showed an increase between E6 and E10 and a decrease afterward. Between E6 and E12, neural retinal cells migrate and organize into the different layers that are recognizable in the mature retina. Furthermore, the RGCs, which form at E6 the only clearly identifiable neuronal layer of the retina, do actively extend their axons toward their target, the optic tectum. The first axons of the RGC layer reach the optic tectum at E6, and by E12 all of them have reached the target. Extracellular matrix components are expressed along virtually the entire embryonic retinotectal pathway (McLoon, 1984; Adler et al., 1985; Cohen et al., 1987; Halfter and Fua, 1987; McLoon et al., 1988; Bartsch et al., 1995). In the optic stalk, laminin expression is transient and correlates with the ability of RGCs to use laminin as a substrate (Cohen et al., 1987, 1989). Similarly, expression of tenascin in the tectum at the time of innervation by RGC axons has been correlated with the capacity of these neurons to extend neurites on tenascin in culture (Bartsch et al., 1995). Interestingly,in situ hybridization revealed accumulation ofcRhoB transcript in the RGC layer that was particularly evident at E8.5, and also of cRac1B, although the accumulation was less dramatic, whereas cRhoA andcRac1A were homogeneously distributed in the whole neural retina. Moreover, the expression of all studied Rho family GTPases was decreased by E12, when retinal layers have formed and all RGC axons have reached the optic tectum. Because Rho and Rac proteins have been implicated in the organization of the actin cytoskeleton, one hypothesis is that the observed expression of these proteins in RGCs may be required for the process of neuritogenesis, which occurs at this time of development.
Another aim of this study was the analysis of the distribution of the identified GTPases in the developing chick embryo. A striking result from this study is the observation that cRhoA,cRhoB, cRac1A, and cRac1B are strongly expressed in the developing nervous system. In fact, in situhybridization on sections from E6.5 and E8.5 embryos showed strong labeling of both CNS and peripheral nervous system. At both stages, DRGs showed high levels of expression of the transcripts, and a more detailed analysis showed differences in the pattern of expression that were particularly evident for cRhoB and cRac1A. The observed strong cRhoB expression in trigeminal ganglia may be correlated with the innervation of the target by the axons of the trigeminal sensory neurons that is actively occurring at this stage (Windle and Austin, 1936; Moody et al., 1989). Clear differences were detected in the localization of the different GTPase transcripts within the spinal cord. The specific localization of cRhoBtranscripts in layers of the developing CNS, in contrast to the homogeneous distribution of the transcripts of other GTPases within the same structures, also suggests specific and different functions of distinct members of the Rho family during neuronal development. Such a conclusion is corroborated by recent studies in invertebrates that have shown Caenorhabditis elegans RhoA to be expressed at highest levels during embryogenesis and particularly enriched in the pharyngeal nerve ring and at the tip of the head containing chemosensory and mechanosensory neurons (Chen and Lim, 1994). Furthermore, theDrosophila DRac1 and DCdc42, are also highly expressed in the nervous system, where they are involved in axonal outgrowth (Luo et al., 1994). Interestingly, we found that all four genes analyzed in this paper are expressed in the chicken developing cerebellum and that cRhoB is concentrated in the presumptive Purkinje cell layer. This might correlate with the recent observation that perturbation of Rac1 activity in mice Purkinje cells leads to modifications of the axonal and dendritic structures of these cells (Luo et al., 1996).
Expression of an epitope-tagged cRac1B has allowed the analysis of the distribution of this new neural-specific Rac in primary neurons. The cellular localization of cRac1B is similar to that of the integrin α6 subunit, a known plasma membrane component, suggesting a possible association of cRac1B with the plasma membrane of neurons, although further work is required to prove association of this protein with the plasma membrane. Like the other members of the family, cRac1B has a C-terminal motif that can be isoprenylated and could account for its possible association to the plasma membrane. In particular, cRac1B is uniformly expressed along actin-rich neurites and their protrusions. This localization could correspond to a prerequisite for the rapid reorganization of the actin cytoskeleton during filopodia extension, a process required for neurite extension or neurite branching, and future work will be aimed at exploring this issue.
In conclusion, the results presented in this paper have shown for the first time that various members of the Rho family of small GTP-binding proteins are differentially and specifically expressed in the CNS and peripheral nervous system of chicken embryos in concomitance with complex events of neuronal differentiation. In view of the widely accepted role of these proteins in multiple aspects of cell physiology, these observations strongly support an important role for Rho family GTPases in the acquisition of the mature neuronal phenotype.
This work was supported by Telethon–Italy (Grant No. 791). M.L.M. was supported by a postdoctoral fellowship from the University of Milano. We are grateful to Dr. Elena Rugarli (TIGEM, Milano, Italy) and Dr. Elena Zanaria (University of Pavia) for providing chick embryo sections for in situ hybridization; Dr. C. Nottenburg (Fred Hutchinson Cancer Research Center, Seattle, WA) for the E10 chick cDNA library; Dr. B. Ranscht (La Jolla Cancer Research Foundation, La Jolla, CA) for the E13 chick brain cDNA library; and Dr. M. A. Impagnatiello for the pcDNA-I-Amp-HA plasmid. We also thank Dr. R. M. Alvarado-Mallart and Dr. E. Gallego (Institut National de la Santé et de la Recherche Médicale U-106, Paris, France) for their help in the interpretation of the results from the in situ hybridization studies, and Dr. Edoardo Boncinelli and Dr. Jacopo Meldolesi for critical reading of this manuscript.
Correspondence should be addressed to Ivan de Curtis, Cell Adhesion Unit, Department of Biological and Technological Research (DIBIT), San Raffaele Scientific Institute, via Olgettina 58, 20132 Milano, Italy.