Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

The small GTPases of the Rho family RhoA, Rac1, and Cdc42 are molecular switches that organize the actin cytoskeleton (for review, see Van Aelst and D'Souza-Schorey, 1997; Hall, 1998; Luo, 2000; Ridley, 2000). In neural cells, Rac1 and Cdc42 induce neurite formation, whereas RhoA causes their retraction via stress fibers (Nobes and Hall, 1995, 1999; Tigyi et al., 1996; Amano et al., 1997; Santos et al., 1997; Bito et al., 2000). Only in its GTP binding state can RhoA interact with its effectors, which include ROCK-I (p160ROCK), ROCK-II (ROK-Rho-kinase), protein kinase N (PRK1/2-PKN), citron, citron kinase, mDia1, mDia2, rhophilin, and rhotekin (Leung et al., 1995; Ishizaki et al., 1996; Matsui et al., 1996; Nakagawa et al., 1996; Van Aelst and D'Souza-Schorey, 1997; Hall, 1998; Kaibuchi et al., 1999). ROCKs increase the phosphorylation state of the myosin light chain (Kimura et al., 1996; Amano et al., 1997), which enhances the tension of stress fibers (Fukata et al., 2001). The pharmacological agent Y-27632 inhibits both ROCKs with an IC₅₀ value in the submicromolar range (Uehata et al., 1997; Ishizaki et al., 2000). Several serine/threonine kinase inhibitors recently have been reported to inhibit ROCK-II when tested in a cell-free assay (Davies et al., 2000). Also, H89, previously considered to be a selective inhibitor of protein kinase A (PKA), has been shown to inhibit ROCK-II with an IC₅₀ value of 270 nM (Chijiwa et al., 1990; Davies et al., 2000).

This finding is of special importance because PKA and the two ROCKs have been reported to have opposite effects on neurite formation. Whereas activated ROCKs can prevent or retract such extensions (Leprince et al., 1996; Tigyi et al., 1996; Bito et al., 2000), activated PKA can facilitate neurite formation by inhibiting RhoA (Dong et al., 1998). Thus, the inhibitory effect of H89 on ROCK-II or even both ROCKs may mask the suppression of PKA activity. Consequently, H89 may facilitate neurite formation, although retraction is expected after inhibition of PKA. In the present study, we have characterized the effects of H89 on the cytoskeleton in the neuroblastoma × glioma cell line NG 108-15, which can form neurite-like extensions (Gerber et al., 1978; Schecter, 1983).

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. The cytotoxic necrotizing factor 1 (CNF1)-glutathione S-transferase fusion protein was produced in Escherichia coli and was purified as described previously (Schmidt et al., 1997). Y-27632 was a generous gift form Yoshitomi Pharmaceutical Industries (Saitama, Japan). H89 and forskolin were bought from Calbiochem (Bad Soden, Germany) and Tocris (Köln, Germany), respectively.

Cell Culture. Neuroblastoma × glioma hybrid cells NG 108-15 were cultured at 37°C with 8.6% CO₂ in Dulbecco's modified Eagle medium with 4.5 g/l glucose (DMEM, PAN, Aidenbach, Germany), supplemented with 10% fetal calf serum (FCS) (Biochrom, Berlin Germany) and 100 IU/ml penicillin/100 µg/ml streptomycin. To induce neurite extensions, cells were cultured in Neurobasal medium without serum (see also figure legends for respective time courses). The medium was supplemented with B27, i.e., a mixture of substituents that support neuronal differentiation (Brewer and Cotman, 1989). Neurobasal medium and B27 were bought from Invitrogen (Carlsbad, CA).

Immunocytochemistry. Cells were fixed with 4% paraformaldehyde for 20 min, washed with PBS, and permeabilized with 0.1% (v/v) Triton X-100. Normal goat serum and normal donkey serum only for ROCK-II were used to block unspecific reactions. Thereafter, cells were incubated with one of the following primary antibodies: monoclonal mouse anti--tubulin-III antibody (Sigma, Deisenhofen, Germany); monoclonal mouse anti MAP-2a,b antibody (Roche Molecular Biochemicals, Mannheim, Germany); polyclonal rabbit anti-ROCK-I antibody (Santa Cruz Biotechnology, Heidelberg, Germany); and polyclonal goat anti-ROCK-II antibody (Santa Cruz Biotechnology). The resulting -tubulin-III and MAP-2 immune complexes were visualized with a Cy^Tm 3-conjugated F(ab')₂ fragment of goat anti-mouse IgG (Dianova, Hamburg, Germany). A Cy^Tm 3-conjugated F(ab')₂ fragment of goat anti-rabbit was used to detect ROCK-I and a Cy^Tm 3-conjugated F(ab')₂ fragment of donkey anti-goat was used to detect ROCK-II (Dianova). For actin staining, cells were fixed with 4% paraformaldehyde for 20 min, washed with PBS, and permeabilized with 0.1% (v/v) Triton X-100. Afterward, the cells were incubated with tetramethylrhodamine B isothiocyanate-conjugated phalloidine (Biozol, München, Germany) and washed again with PBS.

Confocal Image Analysis. Cells were imaged using a Bio-Rad (Hercules, CA) MRC 1024 (version 3.2) confocal system with a krypton-argon laser and an Axiovert 135TV microscope (Zeiss, Oberkochen, Germany). Cy^Tm 3 fluorescence was measured using an excitation wavelength of 554 nm and an emission filter set at 576. A 40× water objective lens was used. Images were obtained using laser sharp 2.1T software and processed using Photopaint (Corel Corporation, Ottawa, Ontario, Canada).

Western Blot Analysis of ROCK-I and ROCK-II. NG108-15 cells were grown for 4 h in Neurobasal medium supplemented with B27. They were lysed on ice for 15 min in 1 ml of lysis buffer (20 mM Tris-HCl [pH 7.4], 137 mM NaCl, 10% glycerol, 1% Triton X-100, 2 mM EDTA, 50 mM sodiumglycerophosphate, 20 mM sodium pyrophosphate, 0.5 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 5 mM benzamidine, 1 mM sodium orthovandate, 5 µg of aprotinin, and 5 µg of leupeptin per milliliter). The lysates were collected with a rubber policeman, sonicated, and centrifuged (10 min, 2000g, 4°C). Rat brain was homogenized in lysis buffer, sonicated, and centrifuged. After determination of the protein concentrations of the supernatants (Bradford, 1976), aliquots containing 100 µg of protein were separated on SDS-7.5% polyacrylamide gels and transferred to a nitrocellulose filter. Western blot immunoanalysis was performed using a polyclonal rabbit anti-ROCK-I (H-85) antibody or a polyclonal goat anti-ROCK-II (C-20) antibody (Santa Cruz Biotechnology). Specific immunoreactivity was detected by using an ECL kit (Amersham International, Uppsala, Sweden).

Preparation of Transfection Vectors and Cell Transfection. The coding regions of the constitutively active RhoA (RhoAV14), the dominant negative RhoA (RhoAN19), and the wild-type RhoA (RhoAWT) genes were excised from the plasmid GEX with BamHI/EcoRI and inserted in frame into BglII/EcoRI sites of pEGFP-Cl (CLONTECH, Heidelberg, Germany).

Evaluation of Cell Morphology and Statistics. Cells bearing extensions were examined using the 20-fold magnification of a Zeiss Axiophot microscope. At least 15 cells per view were counted in 10 areas. The length of the extensions was determined using the confocal laser microscope and Laser sharp 2.1T software. For statistical analysis Kruskal-Wallis and Mann-Whitney-U tests were used. If samples showed normal distribution, analysis of variance was combined with Scheffe's test.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

When incubated in DMEM medium with 10% FCS, NG 108-15 cells were polymorphic and had no extensions (data not shown). Incubation in serum-free Neurobasal medium plus B27 induced extensions in a time-dependent manner. After 8 h, approximately 20% of the cells had extensions that were 50 µm (Fig. 1, A and M). After 26 h, approximately 60% of the cells showed extensions that were 50 µm (e.g., Fig. 3A). Both incubation parameters were used for the subsequent experiments.

View larger version (54K): [in this window] [in a new window]   Fig. 1.   Y-27632 and H89 change the morphology and the distribution of cytoskeletal proteins in NG 108-15 cells. Cells were incubated in DMEM medium containing 10% FCS. One day after the last passage, they were incubated in serum-free Neurobasal medium plus the supplement B27 for 4 h before they were treated with 5 µM Y-27632 (B, E, H, and K) or 10 µM H89 (C, F, I, and L) for another 4 h; controls are shown in A, D, G, and J. After fixation, cells were analyzed with phase contrast (A-C) or stained for the cytoskeletal proteins actin (D-F), -tubulin (G-I), and MAP-2 (J-L). Bar ~ 50 µm. M, concentration response curve of the effects of 0.1, 0.5, 1, 5, 10, and 30 µM Y-27632 (squares) or H89 (rhombes) on the production of extensions  50 µm. The number of extension-bearing cells is expressed as the percentage of total number of cells analyzed. Broken line, percentage of control cells. Mean ± S.E.M.; n  150.

Y-27632 and H89 Induce Extensions in NG 108-15 Cells. Because inactivation of RhoA and of ROCKs has been shown to facilitate neurite formation, we first compared the effects of Y-27632 and H89 on the extensions of differentiating cells. The ROCKs inhibitor Y-27632 (5 µM) caused pronounced changes in NG 108-15 cell morphology, when it was added during the last 4 h of the 8-h differentiation period. Approximately 80% of the cells developed one to three extensions that were  50 µm (Fig. 1, B and M). Whereas staining of filamentous actin with phalloidine showed numerous filopodia in controls (Fig. 1D), cells treated with Y-27632 displayed actin structures similar to growth cones (Fig. 1E). Also the neuron-specific proteins -tubulin III and MAP-2 were expressed by the NG 108-15 cells. Immunoreactive material was uniformly dispersed in the controls (Fig. 1, G and J). Treatment with Y-27632 produced densely filled extensions (Fig. 1, H and K). The effects of 10 µM H89 on the morphology of the NG 108-15 cells corresponded to those of Y-27632 (Fig. 1, C, F, I, and L). H89 and Y-27632 produced extensions in approximately 80% of the cells at a concentration of 1 µM (Fig. 1M). At this concentration, Y-27632 has been reported to inhibit the RhoA effectors ROCK-I, ROCK-II, and PKN but not PKA (Uehata et al., 1997; Davies et al., 2000), whereas H89 has no effect on PKN but inhibits PKA and ROCK-II (Davies et al., 2000). Taken together, these results suggested that Y-27632 and H89 induced extensions by inhibiting one or both ROCKs.

View larger version (67K): [in this window] [in a new window]   Fig. 2.   NG 108-15 cells express both ROCKs. Upper panel shows Western blot analysis: 100 µg of soluble proteins obtained from lysates of NG 108-15 cells or rat brain were separated on a SDS-7.5% polyacrylamide gel and probed with specific polyclonal antibodies against ROCK-I and ROCK-II. Lower panel, immunocytochemical evidence for the presence of the ROCKs in NG 108-15 cells. Bar ~50 µm

Y-27632 and H89 Do Not Prevent the Retraction of Extensions Caused by Constitutively Active RhoAV14 and Wild-Type RhoA. Activated RhoA can retract cell extensions (Nobes and Hall, 1995; Tigyi et al., 1996; Amano et al., 1997; Santos et al., 1997; Schmidt et al., 1997; Nobes and Hall, 1999). To test whether H89 inhibited morphological effects of RhoA, we used NG 108-15 cells that transiently expressed a fusion protein consisting of enhanced green fluorescent protein (EGFP) and RhoAV14 or overexpressed RhoAWT. As controls, NG 108-15 cells were used that had been transfected with the empty EGFP vector. The experiments were performed with fully differentiated cells which best showed retraction of extensions caused by RhoAV14 or RhoAWT. After 24 h in serum-free Neurobasal medium plus B27, approximately 57% of the EGFP expressing cells showed extensions. Addition of 5 µM Y-27632 or 10 µM H89 enhanced the number of cells with extensions to approximately 80% (Figs. 3, A, B, and C, and 4). In addition, both agents elongated the extensions from 87 µm in controls to approximately 140 µm (Table 1). Cells that expressed RhoAV14 had small polymorphic cell bodies without extensions (Figs. 3D and 4). Tightly bundled stress fibers were also observed (Fig. 3G). Treatment with 5 µM Y-27632 or with 10 µM H89 did not produce any extensions in these cells (Figs. 3, E, F, H, and I, and 4). Y-27632 was also ineffective, when used at concentrations of 10, 30, and 50 µM (data not shown). However, both inhibitors enlarged the cell bodies (Fig. 3, D-F) and strongly reduced the density of stress fibers (Fig. 3, G-I) indicating that they indeed inhibited the relevant ROCKs in these cells. However, inactivation of ROCKs was not sufficient to overcome the retraction of extensions caused by RhoAV14. Two explanations seemed possible for this lack of activity. The mutated amino acid may have altered the effector loop of RhoAV14 so that effectors were activated, which caused neurite retraction independent of ROCKs. Alternatively, the overexpression of an active GTPase per se may have activated such effectors. To test the latter hypothesis, we overexpressed wild-type RhoA in the NG 108-15 cells. Compared with EGFP controls, these cells displayed fewer extensions, although they had more extensions than cells transfected with RhoAV14 (Figs. 3J and 4). Also cells expressing the RhoAWT did not respond to Y-27632 or H89 with increased neurite formation (Figs. 3, K and L, and 4). This finding indicated that high intracellular concentrations of active RhoA induced the retraction of extensions independent of ROCKs by recruiting other effectors.

View larger version (60K): [in this window] [in a new window]   Fig. 3.   Y-27632 (5 µM) and H89 (10 µM) do not prevent the retraction of extensions caused by transfection of NG 108-15 cells with constitutively active RhoAV14 or RhoAWT. One day after the last passage, NG 108-15 cells were incubated for 4 h in serum-free Neurobasal medium plus B27. Then cells were transfected for 2.5 h with a plasmid coding for EGFP alone (A-C) or with a plasmid coding for EGFP/RhoAV14 (D-I) or EGFP/RhoAWT (J-L). After a 16-h equilibration period, Y-27632 and H89 were added for another 4 h. After fixation, cells were stained for filamentous actin. EGFP fluorescence is shown in A through F and J through L; actin staining is shown in G through I. Cells treated with Y-27632 or H89 are shown in B, E, H, and K and C, F, I, and L, respectively. G through I show cells of D through F at higher magnification. Bar ~50 µm.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Quantification of the experiment shown in Fig. 3. Cells with extensions 50 µm are shown as percent of total number of cells analyzed. Mean ± S.E.M.; n  150, a shows significant difference (P < 0.05) to EGFP controls; b shows significant difference (P < 0.05) to EGFP cells treated with Y-27632 and H89, respectively.

View this table: [in this window] [in a new window]   TABLE 1 Effect of 10 µM H89 and 5 µM Y-27632 on formation of cell extensions induced by 10 µM forskolin Cells with extensions 50 µm and cells with branched extensions are shown as percentage of total number of cells analyzed. Mean ± S.E.M.; n  150. Length of longest extension is given in micrometers; n = 48.

Y-27632 and H89 Block the CNF1-Induced Retraction of Extensions. Another way to activate RhoA is to treat the cells with CNF1 of E. coli. This toxin deamidates RhoA at Gln-63 and thereby prevents the inactivation of the GTPase (Flatau et al., 1997; Schmidt et al., 1997; Barth et al., 1999). Therefore, we treated NG 108-15 cells with CNF1 to study whether H89 and Y-27632 affected the resultant morphological effects. Approximately 60% of the NG 108-15 cells transfected with the empty EGFP vector showed extensions when they were preincubated for 24 h in serum-free Neurobasal medium plus B27 (Fig. 5J; see also previous experiment). Treatment with CNF1 (100 ng/ml) for 4.5 h induced large polymorphic cell bodies. Now, only 11.8% of the cells had extensions (Fig. 5, A and J). However, 5 µM Y-27632 or 10 µM H89 added to the incubation medium 30 min before CNF1 blocked the CNF1 induced retraction of extensions, more than 60% of the CNF1-treated cells still showed extensions (Fig. 5, B, C, and J).

View larger version (66K): [in this window] [in a new window]   Fig. 5.   Retraction of cell extensions caused by CNF1 (100 ng/ml) is prevented by 5 µM Y-27632 or 10 µM H89 and by transfection of cells with RhoAN19. NG 108-15 cells were transfected with plasmids coding for EGFP alone (A-C) or for EGFP/RhoAN19 (D-I). EGFP immunofluorescence is shown in A through I. Sixteen hours after transfection, cells were treated with Y-27632 (B, E, and H) or H89 (C, F, and I) for 4.5 h. CNF1 was given 0.5 h after the drugs for 4 h (A-C and G-I). Bar ~50 µm. Quantification of the effects of CNF1 on cell extensions is shown in J. Cells with extensions  50 µm are shown as percent of total number of cells analyzed. Cells transfected with EGFP alone (open columns); cells transfected with EGFP/RhoAN19 (striped columns). Mean ± S.E.M.; n  150; a shows significant difference (P < 0.05) to EGFP controls; b shows significant difference (P < 0.05) to EGFP cells treated with CNF1.

PKA and the Formation of Extensions in NG 108-15 Cells. Because these data did not exclude the possibility that H89 exerted an additional effect via PKA, we studied the actions of a selective PKA inhibitor on the morphological changes induced by CNF1. The PKI that selectively blocks the catalytic site of PKA (Day et al., 1989) was used for these experiments. NG 108-15 cells were transfected with expression plasmids coding for EGFP alone or an EGFP/PKI fusion protein. The resulting cells had similar morphologies (Fig. 6, A, D, and G). In both cell types, CNF1 (100 ng/ml) retracted the extensions (Fig. 6, B, E, and G).

View larger version (64K): [in this window] [in a new window]   Fig. 6.   The protein kinase A inhibitor PKI does not change the effect of CNF1 on cell morphology but prevents that of forskolin. Cells were transfected with plasmids coding for EGFP alone (A-C) or for an EGFP/PKI fusion protein (D-F). Sixteen hours later, cells were treated for 4 h with vehicle (A and D), CNF1 (100 ng/ml) (B and E) or forskolin (5 µM) (C and F). Bar ~50 µm. Quantitative analysis of cells showing extensions after treatment with CNF1 or with forskolin is given in G. Cells with extensions 50 µm are shown as percent of total number of cells analyzed. Mean ± S.E.M.; n  150, a shows significant difference (P < 0.05) to EGFP; b shows significant difference (P < 0.05) to PKI; c shows significant difference (P < 0.05) to EGFP + forskolin.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

In the present study, we tested whether the established PKA inhibitor H89 also inhibited morphologically relevant ROCKs. By comparing its effect on neurite formation in NG 108-15 cells with that of Y-27632, which inhibits ROCKs but not PKA (Uehata et al., 1997; Ishizaki et al., 2000), we found that both agents indeed exerted nearly identical effects. In undifferentiated cells, they produced neurite-like extensions and caused similar changes in the distribution of the cytoskeletal proteins F-actin, neurotubulin, and MAP-2. In differentiated cells, which already have long extensions, their effect on neurite formation was less pronounced. However, they blocked the retraction caused by the E. coli toxin CNF1, which activates RhoA. Experiments with dominant negative RhoA confirmed that CNF1 indeed caused the neurite retraction by acting via RhoA. They suggested that Y-27632 and H89 indeed inhibited the effect of CNF1 by acting on the relevant ROCKs. ROCK-I and ROCK-II have been found in central nervous system or cell lines derived from it and both isoforms induced neurite retraction (Nakagawa et al., 1996; Hirose et al., 1998; Katoh et al., 1998). In NG 108-15 cells we found both isoforms. This finding makes it possible that H89 does not only act on ROCK-II as has been reported before (Davies et al., 2000), but also on ROCK-I. In these experiments, we obtained no evidence that H89 exerted its action by inhibiting PKA. Use of PKI, a selective inhibitor peptide of PKA, corroborated this conclusion. PKI did not affect neurite formation, when transfected into NG 108-15 cells. PKI also did not alter the retraction caused by CNF1. Thus, our results confirmed and extended biochemical data that H89 can block ROCK(s) (Davies et al., 2000). In view of a previous report that H89 can antagonize -adrenoceptors (Penn et al., 1999), the present data further limit its usefulness.

Although these results suggested that ROCKs mediated the retraction of neurites in the NG 108-15 cells, we observed that RhoA can exert such an effect independent of ROCKs. Thus, neurite retraction caused by constitutively active RhoA was neither blocked by Y-27632 nor by H89. Moreover, both agents were not able to prevent the retraction caused by overexpression of wild-type RhoA. The formation of stress fibers induced by RhoA depends on the presence of active ROCKs (Nakano et al., 1999). Because Y-27632 and H89 prevented the formation of stress fibers by constitutively active RhoA, both agents effectively inhibited the ROCKs. This finding indicated that the neurite retraction was independent of stress fibers and that additional effectors were involved. mDia has been described as a RhoA effector that acts on actin polymerization via profilin (Watanabe et al., 1997). Both ROCK and mDia seem to be necessary for the formation of regularly organized stress fibers (Nakano et al., 1999). Whether mDia can act independently of RhoA and thereby cause neurite retraction is unknown and has to be tested in future experiments. Transient transfections with expression vectors can result in high concentrations of the respective protein. Therefore, it is possible that the RhoA proteins produced by the NG 108-15 cells in large and unphysiological amounts recruited and activated other RhoA effectors or even nonRhoA effectors which then caused neurite retraction. This speculation is in agreement with our finding that Y-27632 and H89 prevented the neurite retraction induced by activation of endogenous RhoA with CNF1.

Neurites are induced or elongated if a respective driving force prevails over the retraction caused by actin filaments. This driving force is probably provided by Rac1 and Cdc42, i.e., other members of the Rho family of GTPases (Nobes and Hall, 1995; Tigyi et al., 1996; Amano et al., 1997; Santos et al., 1997; Nobes and Hall, 1999; Bito et al., 2000). The strength of the driving force may depend on the cell type and on factors in the incubation medium that activate or inhibit the GTPases. Such agents may explain why we observed neurite elongation induced by Y-27632 or H89 in NG 108-15 cells maintained in Neurobasal medium, whereas such an effect was less or not at all present in other studies (Chijiwa et al., 1990; Hansen et al., 2000).

Activation of PKA initiates neurite formation and guides axonal pathfinding (Heidemann et al., 1985; Chijiwa et al., 1990; Ming et al., 1997; Song et al., 1997; Shimomura et al., 1998; Wang and Murphy, 1998; Hansen et al., 2000). PKA phosphorylates RhoA at serine 188 (Lang et al., 1996) and thereby reduces the binding and activation of ROCKs (Dong et al., 1998) leading to reduced tension of actin fibers and thus to neurite outgrowth. H89 decreases the activity of PKA and thus disinhibits RhoA, whereas it can directly inhibit ROCKs. Therefore, we also tested which of both possible effects of H89 prevails. Because ROCKs are downstream of RhoA, it was to be expected that their inhibition will dominate the resultant morphology. This was indeed observed when forskolin and H89 were tested together. In conclusion, the present results suggest H89 may not be used as an antagonist of PKA in systems in which RhoA and its effector ROCK play a role.