Role of Rho GTPase in astrocyte morphology and migratory response during in vitro wound healing


Address correspondence and reprint requests to Dr Markus Höltje, Charité-Universitätsmedizin Berlin, Centrum für Anatomie, AG Funktionelle Zellbiologie, Philippstrasse 12, D-10115 Berlin, Germany. E-mail:


Small Rho GTPases are key regulators of the cytoskeleton in a great variety of cells. Rho function mediates morphological changes as well as locomotor activity. Using astrocyte cultures established from neonatal mice we investigated the role of Rho in process formation during astrocyte stellation. Using a scratch-wound model, we examined the impact of Rho on a variety of morphological and functional variables such as stellation and migratory activity during wound healing. C3 proteins are widely used to study cellular Rho functions. In addition, C3 derived from Clostridium botulinum (C3bot) is considered selectively to promote neuronal regeneration. Because the latter requires a balanced activity of neurones and glial cells, the effects of C3 protein on glial cells such as astrocytes have to be considered carefully. Low nanomolar concentrations of C3 proteins significantly promoted process outgrowth and increased process branching. Besides enzymatic inactivation of Rho by ADP-ribosylation, changes in protein levels of the various Rho GTPases may also contribute to the observed effects. Furthermore, incubation of scratch-wounded astrocyte cultures with C3bot accelerated wound healing. By inhibiting the Rho downstream effector ROCK with the selective inhibitor Y27632 we were able to demonstrate that the accelerated wound closure resulted from both enhanced polarized process formation and increased migratory activity of astrocytes into the lesion site. These results suggest that Rho negatively regulates astrocytic process growth and migratory responses after injury and that its inactivation by C3bot in nanomolar concentrations promotes astrocyte migration.

Abbreviations used

bovine serum albumin




Dulbecco's modified Eagle's medium


glial fibrillary acidic protein


isolectin B4


lactate dehydrogenase


phosphate-buffered saline


Rho-associated kinase


sodium dodecyl sulfate–polyacrylamide gel electrophoresis


[4-[3-(4-iodophenyl)-2-(4-nitrophenyl)-2H-5-tetrazolio]-1,3-benzene disulphonate

Astrocytes can adopt a multitude of different morphologies depending on the location in the CNS and possible interactions with other cell types. During physiological brain function, astrocytes undergo prominent morphological changes, i.e. when interacting with neurones (Barres and Barde 2000). Among them is an increase in the number of astrocytic processes associated with synaptogenesis during motor learning in the cerebellum (Anderson et al. 1994). In line with a role in regulating cerebral blood flow and the blood–brain barrier, changes of astrocyte morphology upon interaction with endothelial cells were reported (Mi et al. 2001). Along with changes in the morphology of astrocytes come alterations in physiological parameters, such as the modulation of gating properties of a Cl channel in neocortical astrocytes (Lascola et al. 1998).

Astrocytes also play a pivotal role in scar formation following traumatic brain injury. In response to chemical or pathological tissue damage, astrocytes acquire characteristic functional and morphological features referred to as reactive gliosis (Eddleston and Mucke 1993; Ridet et al. 1997). Concomitant with hyperplasia and the formation of hypertrophic processes are increases in the levels of glial fibrillary acidic protein (GFAP), which is the prototypical biochemical change, and in the actin-associated protein palladin along the edge of the lesion (Boukhelifa et al. 2003). Formation of the glial scar is supposed to be the major obstacle to neuronal regeneration (Stichel and Müller 1998; Nieto-Sampedro 1999; David and Lacroix 2003).

Small GTPases of the Rho family are key regulators of the actin cytoskeleton in many neuronal cells (Hall 1998; Luo 2000). In cultured astrocytes inactivation of RhoA by Clostridium botulinum C3 transferase (C3bot) induces irreversible cytoskeletal changes to a stellate morphology, resembling that observed in vivo (Ramakers and Moolenaar 1998). On the other hand, activation of RhoA by the phospholipid lysophosphatidic acid leads to the assembly of F-actin during stress fibre formation and focal adhesions (Ridley and Hall 1992). In our study we wanted to elucidate in more detail the effect of Rho inactivation on astrocyte morphology. Another important consideration was the extent of C3bot effects on astrocyte properties. C3bot is considered as a promising tool in neuroregeneration as it has been reported to promote axonal growth both in its enzymatically active (Jin and Strittmatter 1997; Lehmann et al. 1999; Ahnert-Hilger et al. 2004; Bertrand et al. 2005) and inactive (Ahnert-Hilger et al. 2004) forms. Axonal regeneration, however, depends on a well balanced activity of both neuronal and glial cells; the effects on the latter may sometimes contradict positive neuroregenerative effects. In the present study we applied various types of C3 protein to primary cultures of mouse astrocytes. To compare the morphological changes obtained with the different C3 isotypes we performed morphometric measurements of process-bearing astrocytes. In addition, we studied the influence of Rho activity on mechanisms underlying processes involved in astrocytic wound closure after in vitro wounding using a scratch-wound assay (Yu et al. 1993; Matyash et al. 2001). Re-population of the scratched area after inhibition of Rho signalling either by inactivation of Rho with C3bot, or by blocking the Rho downstream target kinase ROCK with the selective inhibitor Y27632, was investigated with special emphasis on process formation, locomotion (migration) and proliferation.

Experimental procedures


Phosphatidylcholine was obtained from Lipoid GmbH (Ludwigshafen, Germany). Cholesterol and ROCK inhibitor Y27632 were purchased from Sigma Chemical Co. (St Loius, MO, USA). Clodronate was a gift from Roche Diagnostics GmbH (Mannheim, Germany).


A polyclonal antiserum against GFAP used to label astrocytes was purchased from Sigma. An affinity-purified mouse monoclonal antibody against RhoA protein was obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). Rac1 and Cdc42 monoclonal antibodies were purchased from BD Transduction Laboratories (NY, USA). Lectin labelling of microglial cells was performed using Griffonia simplicifolia lectin ILB4 directly coupled to FITC (ILB4-FITC) obtained from Vector Laboratories (Burlingame, CA, USA). Oregon Green 488-labelled goat anti-rabbit (Molecular Probes, Eugene, OR, USA), Texas Red-labelled goat anti-rabbit and Cy3-labelled goat anti-rabbit (both from Jackson Immuno Research Laboratories, West Grove, PA, USA) secondary antibodies were used.

Recombinant C3 proteins

The gene of C. botulinum C3 (accession no. X59039) or enzymatically inactive mutants of C3 harbouring glutamine (Q) or alanine (A) instead of glutamate (E) at position 174 (C3botE174A, C3botE174Q) and the gene of C. limosum C3 transferase (accession No. X872155), were cloned into pGEX-2T expression vector. After expression in Escherichia coli, GST-C3 fusion proteins were affinity purified on glutathione-Sepharose followed by thrombin cleavage to obtain C3 proteins. Following cleavage, thrombin activity was inhibited by incubation with benzamidine sepharose beads. The gene of Staphylococcus aureus transferase strain HMI (accession no. AJ277173) was cloned into pQE30. After expression in E. coli, His-tagged transferases were purified on nickel columns.

Astrocyte cultures

Primary astrocyte cultures were prepared from NMRI mouse brains at postnatal day 2 or 3. Briefly, meninges were removed and brains were mechanically dissociated in Hank's buffered salt solution using fire-polished Pasteur pipettes and centrifuged at 300 g for 3 min. Cells were re-dissociated in the same solution and the procedure was repeated twice with decreasing pipette tip diameters. Cells were first seeded on to six-well plates (3.5 cm diameter/well) pretreated with poly-l-lysine [100 µg/mL in phosphate-buffered saline (PBS)]. Astrocytes were incubated at 5% CO2 in Dulbecco's modified Eagle's medium (DMEM), supplemented with 10% fetal calf serum, 100 U/mL penicillin/streptomycin and 2 mm l-glutamine. At 4 days in vitro microglial cells were detached from the astrocyte monolayer by repeated shaking off. After 5 days in culture with two changes of medium, cells were split and subcultured in 24-well plates at a densitiy of 4 × 104 cells/well on to glass coverslips pretreated with poly-l-lysine.

Treatment with C3 protein isoforms

Astrocyte cultures were incubated for the time periods indicated with the appropriate concentrations of C3 protein added to serum-free DMEM, supplemented with penicillin/streptomycin and glutamine.

Treatment with liposomes

To further purify the astrocyte cultures we used the method of clodronate-mediated killing of macrophages/microglia. In brief, suspensions of multilamellar liposomes containing PBS (as control) and dichloromethylene diphosphonate (clodronate) were generated according to established procedures (Van Rooijen and Sanders 1994). For treatment with liposomes, astrocytes were seeded on glass coverslips as described above. After reaching a subconfluent stage, cultures were incubated with either 10% PBS–liposomes or 10% clodronate–liposomes (diluted in culture medium). Untreated cultures served as controls. After 24 h cultures were washed three times to remove the liposomes and were grown for another 72 h under normal conditions to recover. Cells were then incubated with C3bot (20 nm) for 24 h in serum-free medium. Afterwards cultures were stained with DAPI to determine the total amount of cells, fixed and processed for immunostaining. The purity of the cultures was determined by fluorescence labelling with ILB4; this confirmed the lack of microglia in clodronate-treated compared with control and PBS-treated cultures.


Cells were washed twice times with PBS and fixed for 15 min with 4% paraformaldehyde dissolved in PBS. Fixed cells were incubated for 1 h in blocking solution containing 0.1% Triton X-100, 2% bovine serum albumin (BSA) and 5% normal goat serum (in PBS). Incubation with the primary antisera was carried out overnight at 4°C in blocking solution. Following two washing steps, each for 10 min with PBS, fluorescent dye-conjugated secondary antibodies (dissolved in 2% BSA/PBS) were applied for 1 h at room temperature (20°C). In case of ILB4-FITC staining, the cells were incubated for 45 min at room temperature with the coupled lectin. Cells were washed again with PBS and mounted with IMMU-MOUNT (Thermo Electron, Pittsburgh, PA, USA). Micrographs were taken with a Leica DC camera under a Leica DMRB microscope (Leica, Wetzlar, Germany).

Morphometric analysis

Morphometric analysis of astrocytes was performed using Neurolucida software (Micro Brightfield, Williston, VT, USA). We analysed the overall process length of individual astrocytes including all branches (given in micrometres). Furthermore, we analysed the number of process branches regardless of the order of single branches. For more detail, see Fig. 1(b). Data were acquired from at least three independent preparations (cultures obtained from littermates). Usually, data from a single culture with 25–30 cells measured for each condition are presented.

Figure 1.

Induction of astrocyte stellation by C3bot: concentration- and time-dependent effects of C3bot on cultivated astrocytes. (a) After cultivation with fetal calf serum-containing medium, cells were treated for 3 days with or without 20 nm C3bot in the absence of serum. Astrocytes were stained for the intermediate filament protein GFAP. Incubation with C3bot induced a shape change from a predominantly polygonal cell form to stellate morphology in a high proportion of astrocytes. Values are mean ± SD. Scale bar 50 µm. (b) For morphometric analysis one of the two processes originating from a bifurcation was defined as a branch, irrespective of any morphological features of a given process. The figure shows an astrocyte with two branches (originating from one process). A cell with one branch, each originating from two different processes would be placed the same category. (c) Cultivated astrocytes were incubated for 3 days under serum-free conditions with C3bot concentrations ranging between 0.2 and 50 nm. Fixed cells were stained for GFAP. Cells with stellate morphology were selected randomly and subjected to morphometric analysis for measurement of total process length and number of process branches per cell. Values are mean ± SD. Toxin at all concentrations had a significant effect on length and branches (p < 0.01; Student's t-test). (d) Cultivated astrocytes were incubated under serum-free conditions with 10 nm C3bot for the indicated time periods. Fixed cells were stained for GFAP and subjected to morphometric analysis. Values are mean ± SD. Statistical signficance (p < 0.01 for length and branches) of the effects obtained with the different incubation times (12 h, 24 h, 48 h, 72 h) was confirmed using student's t-test.

In vitro scratch-wound assay

Confluent astrocyte monolayers grown on glass coverslips were used for scratch assays. Three parallel wounds of either 300 µm or 500 µm in width respectively were scratched with the blunted tips of sterile scalpel blades. Cells were fixed 0, 12, 24 or 48 h after wounding, stained and photomicrographs were taken. Individual images (7–8 per scratch wound) taken with the 10 × magnification lens documented the re-population of the lesion site over the whole wound and were analysed for their cell-free area (given in mm2) using Scion Image software Scion Corp., Frederick, MD, USA. In order to count cell nuclei within the lesion site, squares of 200 × 200 µm (seven per individual image) along the midline of the wound were analysed for DAPI-stained nuclei. Data were acquired from at least three independent preparations.

Sequential ADP-ribosylation

Primary astrocyte cultures were treated with different C3bot concentrations for the indicated time periods. Following removal of the medium, cells were rinsed once with PBS and frozen at − 80°C. The frozen cultures were resuspended in buffer consisting of 20 mm HEPES pH 7.4, 100 mm NaCl and 2 mm MgCl2. Cell lysis was performed by sonification (Bandelin sonicator; Bandelin electronics, Berlin, Germany) at 4°C. After centrifugation for 10 min at 2000 g, the postnuclear supernatant was subjected to in vitro ADP-ribosylation using C3bot or C3stau2. Some 25 µg astrocyte protein was incubated with either C3bot (10 ng/µL) or C3stau2 (10 ng/µL) in a total volume of 25 µL in the presence of 50 mm HEPES buffer pH 7.3, supplemented with 5 mm MgCl2, 2.5 mm dithiothreitol, 10 mm thymidine, 2.5 µm NAD and 1 µCi (0.05 µm final concentration) [adenylate-32P]NAD at 37°C for 15 min. After addition of Laemmli sample buffer, samples were boiled for 10 min at 95°C and subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) on 12.5% gels. Gels were stained with Coomassie Brilliant Blue R-250, dried, and further analysed by phosphorimaging (Cyclone Packard; Perkin-Elmer, Boston, MA, USA).

Measurement of proliferation

To assess astrocytic proliferation, metabolic activity was measured as a correlate. Ten to fourteen-day-old astrocytes were trypsinized and plated in 96-well plates with 5 × 104 cells/well and left for 24 h to recover. They were then incubated for 24 h with or without C3 toxin (20 nm) in the absence of fetal calf serum. L929 (supernatant from a fibroblast cell line, which is known to stimulate proliferation; diluted 1 : 1 in DMEM) served as a positive control. The metabolic activity was determined using WST-1 reagent [4-[3-(4-iodophenyl)-2-(4-nitrophenyl)-2H-5-tetrazolio]-1,3-benzene disulphonate; Roche Diagnostics GmbH] according to the manufacturer's instructions. This assay is based on the enzymatic cleavage of WST tetrazolium salt to formazan by the succinate-tetrazolium reductase system of the respiratory chain of intact mitochondria. The colour reaction was measured after 3 h in a microplate reader (1420 Victor; Wallac Oy, Turku, Finland) at 450 nm and 620 nm as a reference wavelength. WST values of control cultures were set as 100%.

Viability test

To determine astrocyte survival after treatment with C3 protein the lactate dehydrogenase (LDH) assay was used (Roche Diagnostics GmbH). LDH is an intracellular cytosolic enzyme that can be found in the supernatant only if membrane integrity is impaired. Briefly, the supernatants from C3bot-treated astrocytes were collected after 24 h of incubation and the amount of LDH released into the extracellular medium was determined according to the manufacturer's instructions. The colour reaction was measured in a microplate reader (1420 Victor) at 450 nm and 620 nm as a reference wavelength. LDH values of control cultures were set as 100%.


Effects of C3 proteins on astrocyte morphology

Bacterial C3 proteins specifically ADP-ribosylate small GTPases of the Rho subfamily (RhoA, B and C), rendering them inactive (Boquet et al. 1998; Just et al. 2001; Aktories and Just 2005). All members of the C3 protein family share the same mechanism of action. Using the ubiquitous NAD + as a co-substrate, the ADP-ribose moiety becomes N-glycosydically linked to Asn-41 within the effector region of Rho, resulting in stabilization of the inactive form of the GTPase (Genth et al. 2003).

First, we analysed the effects of C3bot proteins on astrocytes to elucidate Rho GTPase-mediated changes in their morphology. Cultivated astrocytes were incubated for 3 days with a final concentration of 20 nm C3bot under serum-free conditions. After removal of the medium, cells were fixed, stained for the intermediate filament protein GFAP, and the proportion of astrocytes that exhibited a clear stellate morphology was calculated (Fig. 1a). The predominant shape observed in primary astrocytes cultured in serum-containing medium is a flattened, polygonal cell type (Raff et al. 1983), depending on intracellular pH (Cechin et al. 2002; Gottfried et al. 2003). In serum-free cultures astrocytes can change from the fibroblast-like shape to a process-bearing morphology (Safavi-Abbasi et al. 2001). Following treatment with C3bot, the proportion of process-bearing stellate astrocytes increased from about 5% in control cultures to 82% in cultures incubated with C3bot. The observed morphological changes were not only based on the retraction of the cytoplasm between processes terminating at the original boundaries of the cell. Rather, the thin fibre-like processes must have exceeded these boundaries because the stellate astrocytes were considerably larger than the typical polygonal cells.

Prompted by the finding that the cultivated astrocytes reacted to protein concentrations within the low nanomolar range, we performed dose–response experiments with C3bot concentrations ranging from 0.2 nm to 50 nm. In order to evaluate the effects obtained with the various C3 concentrations, we subjected individual cells to a morphometric analysis. Process-bearing astrocytes were chosen randomly, and the overall process length per cell and the total number of branches were measured (Figs 1b and c). This analysis revealed that even picomolar concentrations, i.e. 0.2 nm C3bot, enhanced process outgrowth accompanied by a more extensive branching pattern. This effect saturated at a concentration of between 5 and 25 nm, with a fourfold increase in both process length and quantity of branches.

We also studied the time course of the C3bot effects (Fig. 1d). We incubated primary astrocytes with 10 nm C3bot for increasing time periods, and then randomly selected process-bearing astrocytes for morphometric analysis 0, 12, 24, 48 and 72 h after C3bot application. After 12 h a significant increase in process length and an even more pronounced increase in process branching was observed. After 72 h the total length of astrocyte processes was increased to about 800 µm in C3bot-treated cells, compared with about 200 µm for untreated cells. Furthermore, astrocytes incubated with C3bot developed five times more branches than control cells (9.15 vs. 1.8). A slight increase in process length and branching was also observed for control cells during the observation period of 72 h. This might be attributed to a general promoting effect of serum starvation on astrocyte process formation. Taken together, the time dependence of C3bot effects is well in line with the data obtained from the dose–response experiments regarding maximal length and degree of arborization.

To address the question of whether incubation with C3bot alters protein levels of RhoA and the related GTPases Rac1 and Cdc42, we performed western blot studies. First, we assessed the protein content in cultures treated with different C3bot concentrations for 72 h. The RhoA signal decreased with increasing C3bot concentrations (5, 20 and 80 nm) to a minimum of 50% of that observed under control conditions (Figs 2a–c). Conversely, the cellular concentration of Rac1 and Cdc42 was not diminished after C3bot application. Rather, at low concentrations (5 and 20 nm respectively) protein levels were slightly higher than those of controls for Rac1 and increased by almost 100% (at 5 nm C3bot) for Cdc42. At 80 nm C3bot, protein levels of both Rac1 and Cdc42 equalled control levels. As an internal control, α-tubulin content was visualized.

Figure 2.

Concentration- and time-dependent effects of C3bot on small GTPase protein expression and ADP-ribosylation levels of Rho in cultured astrocytes. (a–c) Astrocytes were incubated for 72 h with C3bot at the concentrations indicated. Proteins were separated by SDS–PAGE and subjected to western blot analysis. The protein pattern shown in (a) was taken from a representative experiment. Equal amounts of total protein (30 µg) were applied per lane, as confirmed by coomassie staining (b). Densitometric signal analysis of RhoA, Rac1, Cdc42 and α-tubulin as internal standard was performed from three independent experiments using Laboratory-Image software (c). Data obtained for RhoA expression were statistically significant at 20 and 80 nm C3bot, whereas for Rac1 and Cdc42 effects were significant at 5 and 20 nm respectively (p < 0.05; Student's t-test). (d) Under the same experimental conditions, cells were homogenized after incubation and the postnuclear supernatant was ADP-ribosylated by C3bot in the presence of [32P]NAD. The samples were separated by SDS–PAGE and analysed by phosphorimaging. In this assay, a decrease in incorporation of radioactivity in the Rho band reflects the ADP-ribosylation in the intact cell. The bar chart shows the quantification of phosphorimager data. Effects were significant (p < 0.05) at 20 and 80 nm C3bot. The image beneath was taken from a representative experiment. (e, f) Astrocytes were incubated with 20 nm C3bot for the indicated time periods. The bar chart in (e) illustrates the densitometric analysis of the RhoA western blot signals, whereas that in (f) represents the [32P]ADP-ribosylation of Rho. The observed effects were statistically significant (p < 0.05), except for RhoA protein expression at 12 h. Images were taken from representative experiments. In general, data presented in this figure represent mean ± SD.

To confirm internalization of C3bot even at nanomolar concentrations and to estimate the amount of ADP-ribosylation of Rho by C3bot during the 72 h incubation period, we performed a C3-catalysed [32P]ADP-ribosylation reaction of cell lysates (sequential ADP-ribosylation; Fig. 2d). A decrease in incorporation of radioactivity therefore reflects the amount of ADP-ribosylation in intact cells. Quantification of the signals obtained revealed a concentration-dependent increase in ADP-ribosylation of Rho proteins in intact cells to almost 50% at 80 nm C3bot. It should be noted that this reflects the ADP-ribosylation level of total Rho, including RhoA, B and C. To gain further insight into the time course of RhoA expression and ADP-ribosylation levels, we applied C3bot at 20 nm for 12, 24 and 72 h (Figs 2e and f). Western blot analysis documented a decreased RhoA content with prolonged incubation times, to a minimum of 47% at 72 h compared with control conditions. The data obtained from ADP-ribosylation experiments showed that there was a significant amount of ADP-ribosylated Rho at all time points, ranging between 65 and 50% in intact cells. The data obtained with 20 nm C3bot at 72 h in this time course experiment support the results of the concentration-dependence experiments carried out under the same conditions.

Recently, we showed that C3bot, besides its well known ADP-ribosylating mechanism, has an additional neurotrophic effect on axonal growth and branching of hippocampal neurones (Ahnert-Hilger et al. 2004) completely independent of its enzymatic activity. To test whether the observed effects on cultured astrocytes depend exclusively on Rho-mediated pathways, we incubated our cultures for 3 days with the enzymatically dead C3botE174A or C3botE174Q. An exchange of the catalytic active glutamate in position 174 to alanine and glutamine respectively results in a complete loss of ADP-ribosylating activity (Aktories et al. 2000). No morphological changes were observed when astrocytes were incubated with 100 nm of the mutant C3bot proteins (Fig. 3a). C3bot, applied as positive control, exhibited the expected effects on process formation.

Figure 3.

Morphological changes to stellate morphology depend strictly on Rho inactivation. (a) Astrocytes were incubated for 3 days with serum-free medium containing no additives or 20 nm C3bot, 100 nm C3botE174A or 100 nm C3botE174Q. Cells were fixed, stained for GFAP and subjected to morphometric analysis. In contrast to the enhanced process growth and branching observed after incubation with enzymatically active C3bot, the two mutant toxins with no enzymatic activity exhibited no morphological effects. (b) In a similar experiment, we incubated astrocyte cultures for 3 days with 100 nm C3lim or C3stau2; 20 nm C3bot was used as a positive control. Both C3lim and C3stau2 exhibited positive effects, especially on process growth; however, the effects were smaller than those produced with with C3bot. Data are mean ± SD. Statistical significance (p < 0.01 for length and branches) of the effects obtained with all enzymatically active toxin concentrations was confirmed using student's t-test.

As C3 protein from C. botulinum, but neither the related C3 transferases C3lim from C. limosum nor C3stau2 from S. aureus, exhibited Rho-independent neurotrophic effects on hippocampal neurones (Ahnert-Hilger et al. 2004), we also tested the effects of C3lim and C3stau2 for effects on astrocyte morphology (Fig. 3b). Application of 100 nm C3lim or C3stau2 (20 nm C3bot as positive control) significantly enhanced both process growth and branching. However, the effects were not as pronounced as those observed with C3bot.

Astrocyte cultures also contain other brain cells, such as microglia responsible for the innate immune response of the brain. Usually, a low percentage (3–5%) of the cells in primary astrocyte culture represent microglial cells, even after repeated shaking off (Meucci et al. 1998; Matyash et al. 2001). To exclude that the observed morphological changes induced by low nanomolar concentrations of C3bot were due to the release of microglial mediators, i.e. cytokines, we created microglia-free astrocyte cultures by treatment with clodronate–liposomes. Liposomes are ingested by microglial cells (or macrophages in general) followed by intracellular release and accumulation of clodronate which induces cell death. After incubation with 10% clodronate–liposomes and a subsequent recovery period, cells were treated with 20 nm C3bot for 24 h (Fig. 4). Depletion of microglial cells was confirmed by ILB4 staining. Microglia-depleted astrocyte cultures incubated with C3bot showed a shape change towards the stellate morphology, indicating a direct effect of C3bot on cultivated astrocytes. As expected, control experiments performed with PBS–liposomes and clodronate–liposomes without subsequent C3bot incubation revealed no effect on astrocyte morphology (data not shown).

Figure 4.

C3bot induces stellation in microglia-depleted astrocyte cultures. Subconfluent cells incubated with serum-containing medium were incubated for 24 h with either 10% clodronate-liposomes to kill microglial cells or no additives. After exchange of the medium and a subsequent recovery period of 3 days, the serum was removed and cells were incubated with or without 20 nm C3bot for 24 h. Astrocyte morphology was visualized using phase-contrast (Pc) microscopy or GFAP staining. Killing of microglial cells after treatment with clodronate-liposomes was documented by immunostaining against ILB4. DAPI staining was used to determine cell numbers. Treatment with clodronate-liposomes resulted in a complete loss of microglial cells but did not prevent stellation of astrocytes induced by C3bot. Scale bar 50 µm.

C3bot enhances re-population after scratch lesion

In order to investigate the effect of Rho inactivation on cellular mechanisms underlying astrocyte responses to tissue-damaging stimuli, we used an in vitro scratch-wound model. Scratch-wound assays have been used to study different aspects of repair processes during scar formation along the lesion site (Nobes and Hall 1999; Környei et al. 2000). Among these are changes in protein expression, cell morphology and locomotion.

To address these issues, we generated scratch wounds in confluent monolayers of astrocytes by using blunted scalpel blades. Immediately after scratching we incubated the cultures with 20 nm C3bot and compared the time course of re-population of the denuded area with that of scratched control cultures (Fig. 5). After 24 h the initial cell-free area (about 300 µm in width) in C3bot-treated cultures was closed to about 60%, whereas under control conditions astrocytes had only re-populated about 30% of the wound (Figs 5a and b). Many elongated or hypertrophic (Yu et al. 1993) processes with spread morphology could be recognized within the scratched area under either condition, but were more pronounced in cultures treated with C3bot. In a similar experiment, we generated a scratch of about 500 µm in width to follow wound closure over a longer distance and time course. We documented the wound closure qualitatively up to 48 h (Fig. 5c). The accelerated wound healing of C3bot-treated cells compared with control cells was again clearly detectable, being most pronounced at 48 h.

Figure 5.

Astrocytes incubated with C3bot exhibit a faster re-population after in vitro wounding. (a) Confluent monolayers of astrocytes cultivated under control conditions or incubated with 20 nm C3bot were scratched with a blunted scalpel tip to generate a cell-free wound area of about 300 µm in width. Cells were fixed immediately and 24 h after wounding, and stained for GFAP. After 24 h C3bot-treated cells had re-populated the wound to a greater extent than control cells. Scale bar 300 μm. (b) Higher-magnification photomicrographs show hypertrophic cell processes along the margins of the scratch towards the denuded area after 24 h. Quantification of the initial cell-free area and the remaining cell-free area after 24 h revealed a faster re-population by the C3bot-treated cells. Values are mean ± SD. Statistical significance (p < 0.01) of the effects obtained was confirmed using student's t-test. Scale bar 300 μm. (c) Generation of a wider (500 µm) scratch wound. Images show the cell-free area (in black) immediately after wounding as well as 12, 24 and 48 h after creating the lesion. As seen with the 300-µm scratches, astrocytes incubated with C3bot exhibited faster wound closure than control cultures. Scale bar 500 µm. (d) Astrocyte cultures incubated for 24 h with 20 nm C3bot were subjected to WST-1 and LDH assays. The extinction measured in these colorimetric assays correlates with the amount of metabolically active cells (WST-1) or the amount of LDH liberated from dead or lysed cells and can therefore be used to indicate alterations in the proliferation rate after C3bot application and possible cytotoxicity of C3bot respectively. No significant changes in either proliferation or cell lysis after C3bot treatment were observed. Values are mean ± SD. Effects were not significant (p > 0.05) according to student's t-test.

To investigate whether astrocyte proliferation has an impact on the enhanced re-population after C3 treatment, we measured the metabolic activity as a correlate using the WST-1 assay (Fig. 5d). There were no significant differences between control and C3bot conditions, indicating that astrocyte proliferation was not altered after treatment with C3bot. In addition, we used a LDH assay to test whether incubation with C3bot had negative effects on membrane integrity. However, the test revealed no significant increase in cell damage after C3bot application.

Rho inactivation leads to enhanced cell migration and process formation

We were now interested to determine whether the accelerated wound closure resulted from an enhanced rate of polarized process formation after Rho inactivation or also resulted from an increased migratory response of the astrocytes after wounding. To this end, we investigated the role of a major downstream target of RhoA. As in previous scratch experiments, we generated wounds that were 300 µm in width but, instead of C3 protein, we applied the selective Rho kinase inhibitor Y27632 to the cultures at 20 µm. Y27632 has been shown to induce stellation in cultured astrocytes (Abe and Misawa 2003) and to be involved in the closure of wounds by embryonic fibroblasts (Nobes and Hall 1999). Cells were fixed at 0, 12, 24 or 48 h after scratching. In addition to GFAP staining, we performed DAPI staining to localize the nuclei of the cells covering the wound area. Similar to the effects seen with C3bot, accelerated scar formation was observed in cultures treated with Y27632 (Fig. 6a). The photomicrographs taken at higher magnification (Fig. 6b) clearly illustrate enhanced process elongation towards the wound site by 12 h after scratching in cultures incubated with Y27632 compared with control cells. In addition, DAPI staining revealed an increased dispersion of nuclei from the wound edges into the cell-free area in cultures treated with Y27632 compared with control cultures (Figs 6a and b). After 48 h the density of the nuclei within the wound area almost equalled that close to the wound. Conversely, in untreated control cultures a clear gap devoid of nuclei was still present at the same time point. Quantification of the remaining cell-free area and counting the cell nuclei revealed that re-population of the scratch was strongly promoted in cells treated with the Rho kinase inhibitor at all time points (Fig. 6c). This might reflect the enhanced migratory activity of Y27632-treated astrocytes. However, comparable data were obtained when using C3bot protein (not shown).

Figure 6.

Inhibition of Rho signalling by Y27632 enhances process formation and migration of astrocytes after in vitro wounding. In a similar experiment to that presented in Fig. 5, we incubated astrocyte monolayers wounded with scratches (300 µm in width) with the Rho kinase inhibitor Y27632 at a final concentration of 20 µm. Cells were fixed 0, 12, 24 or 48 h after wounding and double-stained for GFAP and the incorporation of DAPI to localize cell nuclei. (a) The cell-free area was re-populated more quickly by astrocytes incubated with Y27632 than in control cultures. DAPI staining revealed that part of the effect was due to faster migration of cells into the wound. (b) Higher-magnification images (GFAP and DAPI merged) showed enhanced process elongation directed towards the lesion site along the wound margins in Y27632-treated cultures. (c) Quantification of the initial cell-free area, the remaining cell-free area and DAPI-stained cell nuclei 12, 24 and 48 h after wounding. Values are mean ± SD. Statistical significance (p < 0.01) of the effects obtained was confirmed using student's t-test. Scale bars 300 µm.

Thus, impairment of Rho signalling resulted in both an enhanced process extension towards the scratch and an increased migratory activity of astrocytes contributing to a faster response to in vitro injury.


The present study demonstrated the central role of Rho in the promotion of astrocytic process growth and branching. It also revealed a crucial function of Rho in mechanisms underlying polarized process extension and migration, but not proliferation, during astrocytic wound healing.

Prompted by our recent findings that even low nanomolar concentrations of C3 proteins exhibited effects on the regulation of axonal and dendritic growth and branching of hippocampal neurones that were, in the case of C3bot, independent of enzymatic activity (Ahnert-Hilger et al. 2004) we applied C3 protein preparations in the low nanomolar range to our cultures. We detected promoting effects on astrocytic process growth and branching that were already discernible at the lowest concentration of 0.2 nm C3bot, whereas C3lim and C3stau2 were less effective.

In contrast to the Rho-independent neurotrophic properties of C3bot observed on hippocampal neurones, the effects on astrocytes were strictly Rho dependent, as revealed by the application of the mutant enzymatically deficient C3bot proteins.

Evidence for the central role of Rho in the reorganization of the actin cytoskeleton during stellation comes from approaches using C3 protein, which prevents lysophosphatidic acid- or thrombin-induced reversal of stellation (Suidan et al. 1997). Our in vitro study for the first time clearly showed by quantitative measurement that C3bot-induced stellation of astrocytes occurs not only as a result of retraction of the cytoplasm to the perinuclear area as described for cyclic AMP-raising agents (Ramakers and Moolenaar 1998). Rather, substantial process growth, accompanied by an increased branching that far exceeded the original cell boundaries, was detected. To what extent mechanisms such as membrane stretching or cavitation contribute to this morphological change remains unknown.

In previous studies, relatively high C3 protein concentrations, usually ranging between 1 and 10 µm, were used to inhibit Rho activity (Ramakers and Moolenaar 1998; Nobes and Hall 1999). Our results may therefore point to the existence of a high-affinity binding site on astrocyte membranes and challenge the current opinion that C3 transferases are only taken up by rather unspecific pinocytosis when present at micromolar concentrations (Boquet et al. 1998). Although we were unable to detect the exact levels of RhoA ADP-ribosylation, as our measurements did not allow distinction between RhoA, B and C, the data point to a significant amount of internalization of C3bot. This was most evident after 12 h of incubation with 20 nm C3bot, when the rhoA content was unchanged but the incorporation of radioactivity was reduced to about 60%. The slightly increased incorporation of radioactivity with time might be due to an altered (increased) expression level of other C3bot targets such as rhoB or C, which we did not examine.

An interesting observation was that incubation with increasing C3bot concentrations or prolonged incubation times resulted in a decrease in RhoA levels. It appeared that at 20 nm C3bot hardly any ADP-ribosylated RhoA was left after 72 h of incubation. It is tempting to speculate that ADP-ribosylated RhoA undergoes degradation by the proteasome. At 1 µm C3bot, remaining RhoA protein content was reduced to 25% of that under control conditions (data not shown). Possible ubiquitination of RhoA is currently under investigation. However, other factors that alter protein expression itself, such as inhibition of transcription factors or enhanced mRNA instability, cannot yet be ruled out. At the same time, protein levels of Rac1 and, more pronounced, of Cdc42 were enhanced at lower C3bot concentrations and returned to control levels at 80 nm C3bot. This indicates an additional, presumably indirect, concentration-dependent effect of C3bot on Rac1 and Cdc42. A large body of evidence has emerged suggesting that there is extensive cross-talk between the three GTPases (Yamaguchi et al. 2001; Nimnual et al. 2003; Yuan et al. 2003), altering the activation status of the respective GTPases. The total amount of protein, however, was not affected in these studies. The mechanism by which C3bot exerts these effects remains to be elucidated in further detail.

By using scratch-wound assays we were able to show that inhibition of Rho by either direct inhibition with C3bot or by blocking downstream targets such as ROCK leads to enhanced wound closure in astrocyte monolayers. Along with the increased formation of elongated and spread processes, we demonstrated enhanced migratory activity of astrocytes treated with Y27632 after injury. Incubation of astrocyte cultures with C3bot, however, did not further alter proliferation rates in our cultures, indicating that only morphological/locomotory responses were affected by the inhibition of Rho signalling.

The role of Rho during astrocyte wound healing may be contrasted with Rho-dependent mechanisms in other cell types during repair processes. In general, Rho appears to be responsible for the formation of cell adhesions (Ridley and Hall 1992) and actin–myosin-based contractions of the cell body and at the rear (Raftopoulou and Hall 2004). As demonstrated in an epithelial cell line of rat small intestine (Santos et al. 1997), embryonic chick epithelial cells (Brock et al. 1996) and human endothelial cells (Aepfelbacher et al. 1997), activation of Rho seems to be a requirement during the early phase of mucosal or vessel reconstitution or the healing of incisional wing bud wounds. Brock et al. (1996), however, showed that the leading cells were drawn forward by a purse-string-like contraction of an actin cable running at the margin of the whole wound, a cytoskeletal machinery absent in astrocytes.

Applying microinjection techniques to scratched embryonic fibroblast cells, it was shown that constitutively active RhoAV14 significantly inhibited wound closure (Nobes and Hall 1999). In the same study the authors were able to demonstrate that incubation with Y27632 promoted wound healing, whereas the large amounts of C3 toxin injected had the opposite effect. As fibroblasts contribute only to a minor extent, if at all, to primary scar formation within the CNS, data obtained from these cells can not substitute for data obtained from reactive astrocytes. As shown by others (Narumiya et al. 1997; Amano et al. 2000) and in the present paper, inhibition of stress fibre and focal adhesion formation with C3bot or, more directly, Y27632 seems to improve astrocyte long-distance migration during wound healing as migration was accelerated.

It would appear that the mechanisms underlying astrocyte migration do not rely exclusively on Rho pathways. In scratch-wounded astrocytes, the activity of integrin-mediated Cdc42 activity is crucial for the establishment of cell polarity by the microtubule organizing centre (Etienne-Manneville and Hall 2001). The authors also demonstrated that, at the same time, Rac activity is required for protrusion formation. Hence, enhancement of Rac1 and Cdc42 levels, as shown in the present paper, may contribute to the faster wound closure after incubation with C3bot.

The results presented here demonstrate some limitations in the use of C3 protein or ROCK inhibitors to foster axon regeneration on growth-inhibitory substrates (Borisoff et al. 2003; Monnier et al. 2003; Fischer et al. 2004; Bertrand et al. 2005). Enhanced axon outgrowth as shown in these studies may only be observed when no astrocytes are present to form a glial scar (or astrocytes are not affected by C3bot, i.e when using intracellular application techniques). In this respect, the enzymatically deficient C3bot forms are more favourable as their effects are restricted to neuronal cells (Ahnert-Hilger et al. 2004) without affecting astrocytes (this paper).


The authors are indebted to Ursel Tofoté, Birgit Metze and Sandra Hagemann for expert technical assistance. This work was supported by the Deutsche Forschungsgemeinschaft (AH 67/4-1 and JU 231/4-1).