The Rho Kinase Pathway Regulates Mouse Adult Neural Precursor Cell Migration§


  • Author contributions: S.Y.L.: conception and design, collection and assembly of data, data analysis and interpretation, manuscript writing; C.H.F. and A.T.: collection and assembly of data, data analysis and interpretation; K.J.D.: collection of data; A.M.T.: conception and design, financial support, data analysis and interpretation, manuscript writing, final approval of manuscript.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS December 9, 2010.


Adult neural precursor cells (NPCs) in the subventricular zone (SVZ) normally migrate via the rostral migratory stream (RMS) to the olfactory bulb (OB). Following neural injury, they also migrate to the site of damage. This study investigated the role of Rho-dependent kinase (ROCK) on the migration of NPCs in vitro and in vivo. In vitro, using neurospheres or SVZ explants, inhibition of ROCK using Y27632 promoted cell body elongation, process protrusion, and migration, while inhibiting NPC chain formation. It had no effect on proliferation, apoptosis, or differentiation. Both isoforms of ROCK were involved. Using siRNA, knockdown of both ROCK1 and ROCK2 was required to promote NPC migration and morphological changes; knockdown of ROCK2 alone was partially effective, with little/no effect of knockdown of ROCK1 alone. In vivo, infusion of Y27632 plus Bromodeoxyuridine (BrdU) into the lateral ventricle for 1 week reduced the number of BrdU-labeled NPCs in the OB compared with BrdU infusion alone. However, ROCK inhibition did not affect the tangential-to-radial switch of NPC migration, as labeled cells were present in all OB layers. The decrease in NPC number at the OB was not attributed to a decrease in NPCs at the SVZ. However, ROCK inhibition decreased the density of BrdU-labeled cells in the RMS and increased the distribution of these cells to ectopic brain regions, such as the accessory olfactory nucleus, where the majority differentiated into neurons. These findings suggest that ROCK signaling regulates NPC migration via regulation of cell-cell contact and chain migration. STEM CELLS 2011;29:332–343


Adult neural precursor cells (NPCs) are predominantly located in the subventricular zone (SVZ) of the lateral ventricles or in the subgranular zone of the dentate gyrus. The SVZ NPCs produce neuroblasts that normally migrate along the rostral migratory stream (RMS) and integrate into the olfactory bulb (OB) but following central nervous system (CNS) damage NPCs can also migrate from the SVZ to the site of damage [1, 2].

For appropriate migration to the correct site, cell adhesion/traction, competency to migrate, and ability to respond to directional signals is required. For NPCs, molecules such as integrins [3, 4] and cell adhesion molecules, such as the NPC marker, polysialated neural cell adhesion molecule (PSA-NCAM) [5–11], are involved in providing the framework for migration, whereas a number of growth factors, such as epidermal growth factor (EGF) confer the capacity for migration [12, 13]. These factors make the cells competent to migrate and may be permissive factors allowing NPCs to respond to directional cues. Directional migration cues, which are relevant following neural injury, include members of the chemokine family, such as CXCL12/CXCR4 [14–16]. Axon guidance molecules, such as Slits [17, 18] and other factors [19, 20], also play a role in the directed NPC migration along the RMS, along with chemoattractive agents secreted from the OB such as sonic hedgehog [21] or secreted or membrane-bound factors on the astrocytic glial tube that surrounds the NPCs in the RMS [22, 23].

NPC intrinsic molecules, such as PSA-NCAM also regulate how NPCs respond to external cues. Removal of PSA from NCAM molecules resulted in dispersion of chains into individual cells, suggesting that PSA is required to maintain the chains of neuroblasts within the RMS [24]. PSA acts like a repulsive agent to lower cell adhesion of NCAM and creates a permissive environment for cell translocation. The absence of PSA enhances cell adhesions via various molecules such as NCAM, L1, integrins, and cadherins [25–28].

Therefore, there are many external influences that modulate the ability of a NPC to migrate to the correct site, under normal physiological conditions or after neural injury. However, at the level of the individual NPC, these signals must be integrated to produce the appropriate response. Candidates for such regulatory roles include members of the Rho-GTPase family of molecules, such as Rho and Rac1, which translate external signals into alterations in cytoskeletal reorganization and hence have the ability to influence migration. A large body of literature has characterized the cellular mechanisms and signal transduction pathways of Rho, Rho-dependent kinase (ROCK) and related regulatory members, such as Rac and PI3K, on cellular motility and chemotaxis in various cell types [29–36].

We hypothesized that targeting these pathways would alter NPC migration under basal conditions and could be attractive therapeutic targets following neural injury to increase numbers of NPCs at the injury site to aid repair. There are specific inhibitors of Rho and Rac1 pathways, some of which are currently in clinical use. Inhibitors of ROCK, the major downstream mediator of Rho-GTPase activation, such as Y27632 and HA1077 (Fasudil), have shown promise as promoters of neural repair. HA1077 is in clinical use for managing cerebral vasospasm associated with subarachnoid hemorrhage and is under clinical trial for acute ischemic stroke [37], among other conditions. Application of Y27632 has also been investigated for promotion of neurite outgrowth and axonal regeneration [38–45] and inactivation of Rho or ROCK signaling promoted functional recovery and/or axonal regeneration after neural injuries [46–48].

Here, we have examined the roles of Rho-GTPases on adult mouse SVZ-derived NPC migration using three models. In vitro, preliminary screening and quantitation of the effects of Rho-GTPase inhibitors was performed on neurospheres, which showed that inhibition of the Rho pathway markedly enhanced, and inhibition of the Rac1 pathway markedly inhibited NPC migration. In terms of potential therapeutic usefulness, inhibition of the Rho pathway was chosen for further analysis. Enhanced migration in the presence of Y27632 was first confirmed in vitro using SVZ explants and then tested in vivo. Infusion of Y27632 into the lateral ventricle had no effect on numbers of NPCs in the SVZ, but decreased numbers of newborn cells in the OB and decreased NPC density in the RMS, with increased numbers of newborn neurons in ectopic sites, such as accessory olfactory nucleus (AON).



All use of experimental animals was approved by the Animal Experimentation Ethics Committee of the University of Melbourne. All procedures were conducted in strict accordance with the National Health and Medical Research Council of Australia guidelines.

Rho-GTPase Inhibitors

The ROCK inhibitors Y27632 (Sigma, Sydney, Australia, and HA1077 (Upstate [Millipore], NSW, Australia, and the Rac1 inhibitor (Merck, Kilsyth, Victoria, were dissolved in water and the PI3K inhibitor LY294002 (Merck) in DMSO (Sigma) dimethyl sulfoxide (DMSO) (0.1%) was used as a control for LY294002 addition. The inhibitors were used at the concentrations as indicated in the results.

Neurosphere Assays

Neurosphere cultures were established from the anterior SVZ of 6- to 8-week-old male C57BL/6 mice (Biomedical Animal Facility, The University of Melbourne) and cultured in complete neurosphere medium containing 20 ng/ml EGF (PeproTech, Australian Laboratory Services, Sydney, Australia, and 10 ng/ml fibroblast growth factor-2 (FGF2) (PeproTech), as previously described [49]. The cultures were used for experiments at 5–6 days in vitro (DIV) between passages 1–5.

Neurospheres were plated on fibronectin (Chemicon, Temecula, California, in complete neurosphere medium/EGF/FGF2 and treated with inhibitors for migration, proliferation, differentiation, and apoptosis analyses. Neurospheres that were >100 μm or <80 μm in diameter were excluded from the assays. A migrating neurosphere was defined as when the neurosphere was completely dispersed (indistinguishable center of the sphere) at 24 hours. Migration was assessed by (a) the extent of cell migration (area of migration), (b) the extent of cell distribution (minimum distance of closest cells measured from the nucleus), and (c) the density of cell distribution (number of neighboring cells within 20-μm radius of each cell). The measurement of area of migration was performed using ImageJ (Wayne Rasband, NIH) and the macro “Hull and Circle” (NIH) written by Audre Karperien, Charles Sturt University, Australia and Thomas R. Roy, University of Alberta, Canada. A second macro, written by Tet Woo Lee, University of Auckland, New Zealand. (e-mail: and based on the Euclidean Distance Algorithm, was used with ImageJ (NIH) for the minimum distance and cell density analyses. Data were collected from three independent experiments, with a minimum of 35 neurospheres per treatment group for each migration experiment. For proliferation and apoptosis assays, the percent of Ki67- or cleaved caspase-3-positive cells was determined from 10 nonoverlapping fields, with at least triplicate wells per condition from three independent experiments.

siRNA Knockdown of ROCK

ROCK1 and ROCK2 expression was downregulated in neurospheres using siRNA. siRNAs targeting ROCK1 or ROCK2 were purchased from Invitrogen: (Melbourne, Australia, Rock1 Stealth Select RNAi 3 siRNA Set (MSS208676; MSS208677; MSS276868) and Rock2 Stealth Select RNAi 3 siRNA Set (MSS208679; MSS208680; MSS 208681). These were used as pooled 40 nM sets: ROCK1 or ROCK2 sets alone (each individual siRNA at 13.3 nM) or ROCK1 and ROCK2 sets combined (each individual siRNA at 6.6 nM). The negative control siRNA (Stealth RNAi siRNA Negative Control Med GC; Invitrogen) was also used at 40 nM. Additionally, all transfections included 40 nM BLOCKiT Alexa Fluor Red Fluorescent Control RNA (Invitrogen) to allow detection of transfected cells. Neurospheres were dissociated as for passaging, then transfected using Amaxa nucleofection with mouse NSC nucleofector solution, essentially according to the manufacturer's instructions (Lonza, Switzerland,, with modifications to increase cell survival and transfection efficiency [50]. Cells were grown into neurospheres, plated onto fibronectin-coated 24-well plates, and assessed in the migration assay, as above or RNA was extracted.

SVZ Explant Cultures

Cultures of SVZ explants were established from postnatal 3-day-old C57BL/6 pups. SVZ tissue at 2.10- to 2.20-mm Bregma was dissected, placed in ice-cold undiluted growth factor reduced Matrigel (Becton-Dickinson, North Ryde, Australia,, and overlaid with Neurobasal medium containing B27 supplement (Invitrogen) and 0.5 mM glutamine, with or without 5 μM or 50 μM Y27632 for 24 hours. The patterns of explant migration were categorized into (a) chain migration, which was defined by the formation of cell chains consisting of three or more cells with contacting cell bodies exiting the explants or (b) dispersed migration, which was defined by dissociated cells migrating individually from the explants. The pattern of migration was expressed as a percent of total number of explants exhibiting each type of migration pattern from three independent experiments, with a minimum of triplicate wells per treatment group for each experiment.

Lateral Ventricle Infusion

Brain infusion cannulae and Alzet osmotic pumps (Model 1007D, 0.5 μl/hr, 7-day duration; BioScientific, NSW, Australia, were used. Control mice were infused with bromodeoxyuridine (BrdU; 10 mg/ml) and treated mice were infused with BrdU plus Y27632 (100 μM). Adult mice were anesthetized with an intraperitoneal injection of 120 mg/kg Ketamil (Ketamine Hydrochloride, Troy Laboratories, Smithfield, NSW, and 16 mg/kg Xylazil (Xylazine Hydrochloride, Troy Laboratories). Using aseptic techniques, cannulae were inserted into the right lateral ventricle (0.0-mm Bregma, 0.8-mm medial-lateral, 2.5-mm deep). Animals were given analgesia (Metacam, Boehringer Ingelheim, North Ryde, Australia, in drinking water for 48 hours postsurgery. The infusion assembly was left in place for 7 days or 1 month, and then the brains were removed for analysis.

Tissue Preparation

Mice were anesthetized with 100 mg/kg of sodium pentobarbitone (Lethabarb, Virbac, Milperra, Australia, and perfused transcardially. Brains were removed and postfixed in 4% paraformaldehyde overnight then cyroprotected in sucrose. Whole brains were frozen in Tissue-Tek optimal cutting temperature compound (OCT; Sakura-Finetek, The Netherlands,, and 10-μm coronal sections were collected at the level of OB (3.5- to 4.0-mm Bregma), RMS (2.5- to 3.0-mm Bregma), anterior SVZ (1.0 to −0.5 mm Bregma), and the infusion site (0.2 to −0.2 mm Bregma) for analysis. Coordinates according to [51].


For fluorescence immunocytochemistry, neurospheres were paraformaldehyde-fixed and blocked with 5% v/v goat serum alone or with 0.2% v/v Triton-X 100 (Sigma) in phosphate-buffered saline (PBS). Neurospheres were incubated with rabbit anti-Ki67 (Lab vision, Fremont, CA,, rabbit-cleaved caspase-3 (Cell Signaling, Arundel, Australia,, mouse anti-βIII-tubulin (Promega, Madison WI,, rabbit antiglial fibrillary acidic protein (GFAP) (Dako, Carpentaria, CA, or rabbit anti-ROCK1 (Santa Cruz, CA, overnight and then goat anti-rabbit Cy2 or anti-mouse Cy3 (Jackson ImmunoResearch, West Grove, PA,, or donkey anti-rabbit Alexa488 (Invitrogen) for 1-hour. For ROCK2 immunostaining with goat anti-ROCK2 (Santa Cruz) and donkey anti-goat Cy2 (Jackson Immunoresearch), neurospheres were fixed in methanol. All cultures were counterstained with 4′,6-diamidino-2-phenylindole (DAPI).

For peroxidase immunohistochemistry, cryosections were postfixed for 10 minutes, endogenous peroxidases inhibited in 0.3% v/v H2O2 in methanol for 30 minutes, the DNA denatured by 20 minutes in 2 N HCl at 37°C. Sections were blocked with 5% v/v donkey serum (Invitrogen) and 0.2% v/v Triton-X 100 in PBS and incubated with sheep anti-BrdU overnight (Exalpha Biologicals, MA, followed by biotinylated rabbit anti-sheep (Vector Laboratories, Burlingame, CA) for 1 hour. Sections were incubated with ABC solution (Vector Laboratories, Burlingame, CA, for 30 minutes, then washed and developed in diaminobenzidine liquid chromogen (Dako, Carpentaria, California). Sections were coverslipped in DPX (Merck). For fluorescent BrdU immunohistochemistry, the H2O2 incubation step was omitted and donkey anti-sheep Cy3 or Alexa488 (Jackson ImmunoResearch) was used, with rabbit anti-GFAP (Dako), mouse anti-NeuN (Millipore) or rat anti-CD11b (Millipore), and donkey anti-rabbit Alexa488 (Molecular Probes, Invitrogen), goat anti-mouse Cy3, or anti-rat Cy3 (Jackson ImmunoResearch).

BrdU+ Cell Number Analysis

The number of BrdU+ cells was determined from cryosections at least 100-μm apart, at the level of OB, RMS, anterior SVZ, and the infusion site. BrdU+ cell counts were performed in defined regions and the area of each region was measured. Results were expressed as cells per millimeter square. At the OB, different regions (granular, mitral, external plexiform, and periglomerular cell layers) were analyzed separately. The SVZ was defined as the entire lateral wall of the lateral ventricle, including the dorsal-lateral segment. Analysis of the RMS included determining the density of BrdU+ cells and the number of neighboring cells within 20 μm of each individual cell within one 250 × 250-μm region of the RMS in each section (n = 8 sections per animal). At the level of the RMS, the number of BrdU+ cells in the AON and anterior cortex was also determined (n ≥ 30 sections per animal). For analysis at 1 week, four control and five treated animals were examined. For analysis at 1 month, three control and two treated animals examined and the investigator was blind to treatment. Area measurements were performed using AxioVision v3.1 image analysis program (Zeiss, North Ryde, Australia,

Reverse Transcription Polymerase Chain Reaction

Total RNA was extracted from neurospheres using a total RNA purification kit according to the manufacturer's instructions (QIAGEN, Doncaster, Australia, Total RNA was reverse transcribed into cDNA using SuperScript III first-strand synthesis system (Invitrogen) and polymerase chain reaction (PCR) was performed using ROCK1 forward 5′ATGTCGAC TGGGGACAGTTTTG3′ and reverse 5′CATCACCGCCTTGGG ATTTTAA3′ primers, ROCK2 forward 5′GATGGCTTAAAT TCCTTGGTCC3′ and reverse 5′GAGCTGCCGTCTCTCTTATG TT3′ primers, β-actin forward 5′CTGAAGTACCCCATTGAAC ATGGC3′ and reverse 5′CAGAGCAGTAATCTCCTTCTGCAT3′ primers and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) forward 5′GGTGAAGGTCGGTGTGAACG3′ and reverse 5′TTGGCTCAACCCTTCAAGTGG3′ primers. The annealing temperature for PCR was 58°C and performed for 25 cycles. Gels were imaged and digitized using gel documentation equipment (Fuji LAS3000, Fujifilm, Sydney, Australia, band intensity calculated using Image J (NIH) software and normalized to actin or GAPDH.


All data were presented as mean ± SEM and statistical analyses were performed using GraphPad Prism v4.03 (GraphPad Software, Inc., San Diego, CA, with p < .05 considered statistically significant. For in vitro experiments, comparisons between groups were analyzed using analysis of variance (ANOVA), with Dunnett's test for migration assays and Tukey's test for proliferation, apoptosis, and explant assays. For in vivo experiments, data were analyzed using the unpaired t test.


ROCK, Rac1, and PI3K Inhibitors Affected the Morphology of NPCs In Vitro

Neurospheres were plated onto fibronectin with ROCK, Rac1, and PI3K inhibitors. Untreated NPCs had lamellipodia and migrated as sheets of interconnected cells (Fig. 1A). Treatment with Y27632 for ROCK inhibition promoted an elongated, mostly bipolar shape (Fig. 1B), with “growth cone-like” lamellipodial formations restricted to the tips of the leading edges (Fig. 1B′). ROCK inhibition significantly reduced the membrane contact between the migratory NPCs and promoted single-cell migration. In contrast, few, if any, Rac1 inhibitor-treated NPCs migrated away from the neurospheres (Fig. 1C). Rac1 inhibition reduced lamellipodial formation and maintained high cell-cell contact. Treatment with LY294002 to inhibit PI3K also reduced lamellipodial formation (Fig. 1D), but did not interfere with NPC migration.

Figure 1.

Effects of Rho-GTPase inhibitors on neural precursor cell (NPC) morphology, proliferation, and survival. (A): Untreated NPCs migrated as interconnected cells with lamellipodia. (B): Y27632-treated NPCs showed thinning of cell bodies and decreased lamellipodia, with “growth cone-like” protrusions at the leading tips of the cells (arrows in [B′]). (C): Rac1 inhibitor-treated NPCs showed reduced membrane protrusion and did not migrate away from the neurosphere. (D): LY294002-treated NPCs showed predominantly rounded cell bodies, with limited lamellipodia. DAPI-stained nuclei of (E) control, (F) Y27632, (G) Rac inhibitor, and (H) LY294002-treated cells. Scale bar = 100 μm (A–D); 25 μm (B′); 200 μm (E–H). There were no differences in (I) number of cells per neurosphere, (J) proliferation, and (K) apoptosis. Mean ± SEM of n = 3 independent experiments. Abbreviations: DMSO, dimethyl sulfoxide; DAPI, 4′,6-diamidino-2-phenylindole.

The morphological changes were reflected in the “closeness” of NPC association in the plated neurospheres, as assessed by DAPI labeling of nuclei (Fig. 1E–1H). ROCK inhibition showed the loosest NPC association (Fig. 1F) compared with control neurospheres (Fig. 1E), whereas Rac1 inhibitor treated NPCs were closely associated (Fig. 1G) and LY294002 had little effect (Fig. 1H). There were no significant differences in the size or density of the neurospheres between the treatment groups (Fig. 1I). In addition, none of the treatments significantly affected NPC proliferation (Fig. 1J) or apoptosis (Fig. 1K) under these conditions. Additionally, they did not affect NPC differentiation into βIII-tubulin expressing neurons or GFAP-expressing astrocytes (data not shown). This suggested that the varying degrees of “closeness” were likely a reflection of effects on cell-to-cell association and/or migration.

ROCK Inhibition Increased and Rac1 Inhibition Decreased NPC Migration In Vitro

To examine the degree of cell association, the area of the migrated cells within a neurosphere, the density of NPCs within a 20-μm radius of each cell, and the minimum distance between cells in each neurosphere were examined by automated analysis of DAPI stained nuclei. In a dose-dependent manner, Y27632 increased the area of NPC spread (Fig. 2A, ANOVA p < .0001), decreased the number of neighboring NPCs within a 20-μm radius (Fig. 2A′, ANOVA p < .0001) and increased the minimum distance between the NPCs (Fig. 2A″, ANOVA p < .0001). In contrast, Rac1 inhibitor treatment decreased the area covered (Fig. 2B, ANOVA p < .001), increased cell density at the highest concentrations (Fig. 2B′, ANOVA p = .06 overall and p < .05 for 50-μM Rac1 inhibitor by post hoc test), and decreased the minimum distance between NPCs (Fig. 2B″, ANOVA p = .01). LY294002 treatment did not significantly change the area of spread (Fig. 2C, ANOVA p = .61), cell density (Fig. 2C′, ANOVA p = 0. 06), or minimum distance between cells (Fig. 2C″, ANOVA p = .66). ROCK inhibition therefore significantly increased NPC migration, Rac1 inhibition significantly decreased NPC migration, and PI3K inhibition did not affect NPC migration under any of the criteria that were used for these migration analyses. Addition of LY294002 did, however, effectively downregulate the PI3K pathway (Supporting Information Fig. 1), whereas Y27632 decreased phosphorylation of myosin light chain II (Supporting Information Fig. 2), a major regulator of the cytoskeleton downstream of ROCK signaling.

Figure 2.

The effect of ROCK, Rac1, and PI3K inhibition on NPC migration. The extent of NPC migration was assessed by analysis of the spread of DAPI-stained nuclei. (A, B, C): The area covered by the migrating cells, (A′, B′, C′) the number of neighboring NPCs with a 20-μm radius of each cell (cell density), and (A″, B″, C″) the minimum distance between neighboring cells. Results were normalized as a percentage of control untreated values and show the mean ± SEM of n = 3 independent experiments. *, p < .05; **, p < .01; ***, p < .001. Abbreviations: DMSO, dimethyl sulfoxide; DAPI, 4′,6-diamidino-2-phenylindole; NPCs, neural precursor cells; Unt, untreated.

Effect of ROCK Inhibition on NPC Morphology Was Reversible and Similar Regardless of Substrate or ROCK Inhibitor Used

Reverse transcription polymerase chain reaction (RT-PCR) was used to determine that NPCs expressed both ROCK1 and ROCK2 isoforms (Fig. 3A) and the correct identities of the PCR expression products were confirmed by DNA sequencing (data not shown). To confirm that the effects of Y27632 were through ROCK inhibition, the effect of another widely used ROCK inhibitor, HA1077 (10–100 μM), was tested and produced similar effects to Y27632 (Fig. 3B). The effect on NPC morphology in the presence of ROCK inhibition was independent of extracellular matrix (ECM) substrate used, with Y27632 inducing an elongated and generally bipolar morphology on both fibronectin and laminin. In untreated conditions, whole neurospheres plated on fibronectin formed a monolayer within 24 hours (Fig. 3C), but by 72 hours the cells formed chain-like structures (Fig. 3E) and this was blocked by ROCK inhibition. Addition of Y27632 to the NPC chains at 72 hours promoted chain dissociation and cell dispersal, whereas removal of Y27632 from separated cells at 72 hours led to formation of NPC chains (Fig. 3E, 3F). Although the NPCs did not form chains on the laminin substrate, the effect of ROCK inhibition on NPC morphology on laminin was also reversible (Fig. 3G, 3H).

Figure 3.

Neural precursor cells (NPCs) express both isoforms of ROCK and ROCK inhibition prevented chain formation of neurosphere-derived NPCs. (A): NPCs expressed ROCK1 and 2 isoforms by reverse transcription polymerase chain reaction. (B) Similar to Y27632, the ROCK inhibitor HA1077 (25 μM) reduced lamellipodial formation and promoted cell body elongation. (C): Untreated NPCs migrated as connected cells and (D) Y27632 (25 μM)-treated NPCs as single cells on fibronectin at 24 hours. (E): By 72 hours, untreated NPCs formed chains. (F): ROCK inhibition (Y27632) prevented chain formation. Introduction of Y27632 after chain formation at 72 hours disassociated the chains, whereas removal of ROCK inhibition after 72 hours of Y27632 treatment re-established the chains. (G): Untreated NPCs did not form NPC chains on laminin but (H) ROCK inhibition induced membrane thinning and cell body elongation. Scale bar = 50 μm (B); 100 μm (C–H). Abbreviation: ROCK, Rho-dependent kinase.

Both ROCK1 and ROCK2 Regulate NPC Morphology and Migration

As NPCs expressed both isoforms of ROCK and the inhibitors used blocked both isoforms, siRNA knockdown was used to examine the relative contributions of ROCK1 and ROCK2 on NPC morphology and migration. Negative control, ROCK1, ROCK2, or ROCK1 and ROCK2 siRNAs were transfected into NPCs, along with a red fluorescent RNA to allow detection of transfected cells. Transfected cells were grown into neurospheres, then plated onto fibronectin for 24 hours. Virtually all cells in neurospheres that adhered to the fibronectin showed red fluorescence, indicating very high transfection efficiency (Fig. 4B, 4F, 4J, 4N). Immunostaining for ROCK1 (Fig. 4C, 4G, 4K, 4O) or ROCK2 (Fig. 4D, 4H, 4L, 4P) or RT-PCR (Fig. 4Q) indicated that ROCK1 and ROCK2 expression was decreased in the presence of ROCK1 and ROCK2 siRNA, respectively. By densitometric analysis, ROCK1 RNA was decreased to approximately 20% of control levels with ROCK1 siRNA alone and to 60% when combined with ROCK2 siRNA. ROCK2 RNA was decreased to 30% of control levels with ROCK2 siRNA alone and to 50% when combined with ROCK1 siRNA. NPCs transfected with ROCK1 siRNA were flat, showed cell-cell contact and limited migration, similar to control transfected cells (Fig. 4A–4H, 4R). NPCs transfected with ROCK2 RNA showed significantly increased migration, with a mix of cell morphologies ranging from the same as control cells to cells with an elongated morphology (Fig. 4I–4L, 4R). Cells transfected with both ROCK1 and ROCK2 siRNAs showed a morphology that was similar to that observed with Y27632; the cells had an elongated bipolar morphology, little cell-cell contact and a further increased area of migration (Fig. 4M–4P, 4R).

Figure 4.

Both isoforms of ROCK regulate neural precursor cell (NPC) morphology and migration, but ROCK2 is more potent than ROCK1. siRNA was used to downregulate ROCK1 and/or ROCK2 expression. Transfection efficiency was assessed in live cells (A, E, I, M) by uptake of a red fluorescent RNA (B, F, J, N), indicating that the majority of NPCs in the plated neurospheres were successfully transfected. Successful downregulation of ROCK1 and ROCK2 RNA and protein expression was observed by reverse transcription polymerase chain reaction (Q) and immunostaining for ROCK1 (C, G, K, O) and ROCK2 (D, H, L, P), respectively. ROCK1 downregulation alone had no effect on cell morphology or migration (E–H, R) compared with negative control siRNA (A–D, R). ROCK2 downregulation induced an elongated morphology in some cells and enhanced the area of migration (I–L, R), whereas combined downregulation of ROCK1 and ROCK2 induced a bipolar morphology in most cells and further increased the area of migration (M–P, R). (R): The area of migration is expressed as area per cell number to compensate for variability in neurosphere size and shows the mean ± SEM of n = 3 individual experiments, from three neurosphere lines. **, p < .01; ***, p < .001. Scale bar = 50 μm (A–P). Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; ROCK, rho-dependent kinase.

ROCK Inhibition Decreased Chain Migration from SVZ Explants

ROCK inhibition reduced neurosphere-derived NPC chain formation, however, the cells had been passaged and may not reflect chain formation from primary NPCs. Therefore, we examined the effect of ROCK inhibition on migration of NPCs from SVZ explants as a better in vitro model of NPC chain migration. Approximately 50% of the explants contained cells that migrated into the Matrigel and they did so either as chains or as individual cells (Fig. 5A–5E). Under basal conditions, the percentage of explants that showed either mode of migration was similar (Fig. 5E). ROCK inhibition with Y27632 (5 μm and 50 μM) inhibited chain migration and promoted a dispersed pattern of cell migration (Fig. 5E).

Figure 5.

ROCK inhibition prevented chain migration of subventricular zone (SVZ) explant-derived cells. Representative examples of untreated (A, B) and Y27632 (5 μM and 50 μM)-treated (C, D) SVZ explants cultured in Matrigel. Cells of both untreated and treated explants exhibited two patterns of migration, (A, A′, C) chain migration and (B, B′, D) dissociated migration. (E): Untreated explants equivalently showed either mode of migration. Y27632 decreased the percentage of explants showing chain migration and increased the percentage showing dissociated migration. Scale bar = 100 μm (A–D); 25 μm (A′, B′). Mean ± SEM of n = 3 independent experiments. **, p < .01; ***, p < .001.

ROCK Inhibition Altered SVZ-Derived NPC Migration In Vivo

To determine whether ROCK inhibition altered NPC migration in vivo, Y27632 and BrdU were infused into the lateral ventricle of adult mice for 1 week and brains were examined at 1 week and 1 month. Intrathecal coadministration of BrdU enabled us to infer that the BrdU-labeled NPCs and their progeny had also been exposed to Y27632 or vehicle control (saline). Cryosections were collected from four regions of the brain: from the center of the OB, the forebrain including the RMS, the anterior SVZ, and at the site of infusion (injury site).

ROCK Inhibition Increased the Area of the SVZ

Y27632-treated animals showed a slightly increased SVZ area at 1 week, especially apparent at the dorsal lateral segment of the lateral ventricle (Fig. 6A–6C). This was not due to an increase in the number of BrdU+ cells within the SVZ. BrdU+ NPCs of the treated SVZ (Fig. 6B) appeared less densely packed than the BrdU+ NPCs in the SVZ of the control animals (Fig. 6A), however, analysis of the number of cells per SVZ section showed no significant difference between the control and the treated animals (Fig. 6D), although there was a trend to decreased density (Fig. 6E). These results suggest that ROCK inhibition did not significantly alter the number of BrdU+ NPCs within the SVZ.

Figure 6.

ROCK inhibition decreased neural precursor cell (NPC) migration to the olfactory bulb. (A, B): Mice were infused with BrdU ± Y27632 into the right lateral ventricle for 7 days and then brains were immunostained for BrdU labeling. Treated SVZ showed less tightly packed BrdU+ NPCs, especially within the dorsolateral corner (dotted lines in [A, B]). (C): ROCK inhibition modestly increased the area of SVZ but (D) did not affect the number of BrdU+ NPCs per section or (E) cell density in the SVZ. Scale bar = 50 μm (A, B). (F): In the olfactory bulb, ROCK inhibition (G) showed a trend to decrease numbers of newborn NPCs at the RMS-OB and (H) decreased numbers of NPCs in the granule, mitral, and EPL layers. Mean ± SEM of five treated mice and four control mice; *, p < .05; **, p < .01 comparing control and treated in each region. Abbreviations: BrdU, bromodeoxyuridine; EPL, external plexiform; Gr, granule; GL, glomerular cell layer; Mi, mitral; RMS-OB, rostral migratory stream region of the olfactory bulb; SVZ, subventricular zone.

ROCK Inhibition Decreased Numbers of Newborn NPCs in the OB

The number of BrdU+ NPCs was counted at the OB-end of the RMS (RMS-OB) and the granule, mitral, external plexiform, and glomerular cell layers of the OB at 1 week (Fig. 6F–6H). ROCK inhibition showed a trend of decreased NPC density at the RMS-OB compared with the untreated animals (Fig. 6G). Within the OB proper, in the Y27632-infused animals there were significant decreases in the density of newborn NPCs, particularly in the granule, mitral, and external plexiform layers (Fig. 6H).

ROCK Inhibition Decreased Cell Density in the RMS

After 1 week, BrdU+ NPCs within the RMS of the treated animals appeared less densely packed compared with the control RMS and numerous BrdU+ cells were scattered beyond the boundary of the treated RMS (Fig. 7B). Measurement of the RMS indicated that ROCK inhibition did not change the transverse area (Fig. 7C), however, the cell density in the RMS of treated mice was decreased compared with control (Fig. 7D), as was the number of neighboring NPCs within 20 μm of a given cell (Fig. 7E).

Figure 7.

Rho-dependent kinase (ROCK) inhibition decreased neural precursor cell (NPC) density in the RMS and induced ectopic migration. Mice were infused with BrdU ± Y27632 into the right lateral ventricle as earlier. (A): The vertical line on the schematic indicates the level of RMS collected for analysis and the coronal view indicates the forebrain regions that were assessed. (B): BrdU+ NPCs in the treated RMS appeared more dispersed and frequently showed scattered precursors beyond the boundary of the RMS. Dotted lines indicate the region defined as RMS. (C): The area of the RMS was the same in control and treated mice. (D): ROCK inhibition decreased cell density in the treated RMS, and (E) the number of neighboring NPCs within 20 μm of a given cell. (F): ROCK inhibition induced an increase in the number of BrdU+ cells in regions outside of the RMS, such as the anterior cortex. (G): Control NPCs in the RMS expressed PSA-NCAM (green); DAPI (blue), and (H) NPCs of Y27632-infused animals expressed PSA-NCAM at and adjacent to the RMS. PSA-NCAM expression was also detected on isolated cells in (I) the anterior cortex and (J) at the AON. (K): Ectopic migration into the AON was more obvious at 1 month, with an increase in total BrdU+ cell numbers and NeuN+ neurons. (L): The majority of the BrdU+ cells were NeuN+ neurons and the percentage was increased in treated animals. (M): Representative image showing BrdU and NeuN colabeling in the AON of a treated mouse. Mean ± SEM of five treated mice and four control mice at 1 week and two treated and three control mice at 1 month. *, p < .05. Scale bar = 50 μm (B, I, J); 200 μm (G, H); 100 μm (M). Abbreviations: AON, accessory olfactory nucleus; BrdU, bromodeoxyuridine; RMS, rostral migratory stream.

Increased Ectopic Migration Outside the RMS

Newborn NPCs were frequently seen detached and leading away from the core of the treated RMS (Fig. 7B). At 1 week, ROCK inhibition appeared to enhance ectopic migration of NPCs outside of the RMS to areas such as the AON and the anterior cortex, with increased numbers of BrdU+ cells (Fig. 7F). Immunostaining for PSA-NCAM was observed as expected within the tight cluster of migrating precursors of the RMS in the control and treated animals (Fig. 7G, 7H). Treated animals frequently showed PSA-NCAM+ cells scattered beyond the boundary of the RMS, in addition to the migrating cells of the RMS. Ectopic cells in the anterior cortex (Fig. 7I) and the AON (Fig. 7J) of both the control and treated animals also showed expression of PSA-NCAM. This suggested that the BrdU+ cells observed outside of the RMS were possibly NPCs that had migrated from the RMS or SVZ. The comparative increase in number of BrdU-labeled cells was also observed in the AON at 1 month. The majority of labeled cells were NeuN+ neurons, with a modest increase in the percentage of NeuN colabeled cells in treated animals (Fig. 7K–7M). However, in the anterior cortex, very few BrdU+ cells remained at 1 month (<1 cell per section) and even fewer were NeuN+ (<1%).

To determine whether ROCK inhibition enhanced recruitment of newborn NPCs to the injury induced by insertion of the infusion cannula, the number of BrdU+ cells at the margin and the periphery of the injury site was examined at 1 week. The injury site had many BrdU+ cells, whereas the contralateral side showed few BrdU+ cells in the cortex (Supporting Information Fig. 3), with no significant difference between control and treated animals (Supporting Information Fig. 3B). There was an increased percentage of proliferative astrocytes at the margin but not the periphery of the injury site of the treated mice, with no difference in the percentage of CD11b+ macrophages (Supporting Information Fig. 3B, 3C–3E).


This study examined the effects of Rho-GTPase inhibition on NPC migration. The inhibition of ROCK increased neurosphere-derived NPC migration, whereas Rac1 inhibition decreased and PI3K inhibition had no effect on migration. Given the potential for ROCK inhibition to thus increase migration of adult NPCs in a therapeutic setting, the effect of ROCK inhibition on NPC biology was then characterized in more detail using neurospheres, SVZ explants and in vivo analyses. In each of these model systems, ROCK signaling played a significant role in NPC-NPC contact; inhibition of ROCK prevented neurosphere-derived NPC chain formation and SVZ explant chain migration, as well as NPC-NPC association in vivo. Unlike for neurospheres, which exhibited general increased migration in response to ROCK inhibition, infusion of Y27632 into the lateral ventricles decreased migration of NPCs to the OB. However, BrdU+ cell numbers were increased in ectopic sites, such as anterior cortex at 1 week and AON by 1 month, suggesting that, similar to the in vitro results, migration was enhanced but there was a partial loss of directed migration.

Regulation of NPC Morphology and Migration by Rho GTPase Signaling

ROCK, Rac1, and PI3K differentially affected NPC migration. In vitro, the inhibition of Rac1 and PI3K affected lamellipodial protrusion of NPCs, indicating highly conserved roles for these molecules in modulating the actin cytoskeleton at the cell front that is required for environmental sensing. The inhibition of PI3K limited the production of lamellipodia but did not affect migration, indicating that the PI3K pathway was not essential for NPC motility. ROCK inhibition of NPCs also influenced the mode of cell migration, changing the way the NPCs interacted, promoting membrane retraction, and loss of contact with neighboring cells. As a result, the treated NPCs predominantly migrated in single-cell pattern. This was the case using two chemical inhibitors of ROCK, Y27632 and HA1077. We demonstrated that adult NPCs express both isoforms of ROCK, namely ROCK1 and ROCK2. These isoforms share 65% protein homology and especially high conservation at the kinase-binding domain (92% homology) [52]. Because of this high-sequence homology, development of an exclusively ROCK1 or ROCK2 inhibitor has proven to be difficult and most of the inhibitors currently available target both isoforms, therefore, the relative contributions of each isoform is difficult to determine. Using siRNA-mediated knockdown of ROCK1 and/or ROCK2 expression, we showed that both isoforms are involved in regulating NPC migration and morphology, as decreased expression of both was required to reproduce the effect of Y27632. However, unlike ROCK1, knockdown of ROCK2 was partially effective, suggesting that it may be the dominant isoform in NPCs and can compensate for decreased expression of ROCK1.

The change to a single-cell migration mode in ROCK inhibited NPCs is of importance as it potentially alters signal transduction between NPCs, which ultimately influences the response of the NPCs to a variety of factors. In vivo, following Y27632 infusion into the lateral ventricle, the ensuing decrease in NPC-NPC association appeared to negatively impact on the basic navigational capability of the NPCs to the OB, resulting in a partial loss of directed migration. Physiologically, NPCs in the SVZ/RMS migrate in a closely associated chain formation, ensheathed by networks of glial fibers [7, 9, 10, 53]. Close association among NPCs or between NPCs and glial cells within the RMS is believed to be important to maintain chain migration, as disruption to the cell-cell interactions disrupts NPC chain migration. Several classes of adhesion and guidance molecules, such as integrins and laminin, have been identified, which are important for NPC chain migration [3, 54, 55]. Furthermore, these molecules signal through downstream Rho/ROCK effectors [56–59]. A collective consensus is that multiple receptor-induced signals act synergistically to mediate the chain migration of NPCs in the RMS. The current work indicates that the Rho pathway is a key player in modulating this migration, as inhibition of ROCK induced chain dissociation and redistribution of NPCs to other brain regions, such as the AON and anterior cortex.

Regulation of NPC Survival and Differentiation by Rho-GTPases

We found no effect on survival or differentiation of NPCs in vitro; however, there was an increased percentage of BrdU+ neurons in the AON, indicating a possible survival or differentiation advantage in vivo. The percentage of astrocytes at the injury site was increased at 1 week, but given the short time frame for differentiation, this is most likely due to increased proliferation of local reactive cortical astrocytes [60], proliferation of which is enhanced by ROCK inhibition [61]. In other neural stem cell systems, ROCK inhibition has been shown to affect survival and differentiation. Under conditions that induce apoptosis, ROCK inhibition can promote NPC survival, such as following transplantation of embryonic stem (ES) cell-derived NPCs [62]. In human ES cell-derived neural stem cells, it also promotes differentiation into neural crest-like cells [63] and blocks lysophosphatidic acid (LPA)–induced inhibition of neuronal differentiation, with no effect on neuronal differentiation by itself [64]. Recently, systemic infusion of HA1077 (Fasudil), another ROCK inhibitor, was reported to promote proliferation and neurogenesis of SVZ NPCs following hypoxia/reperfusion in mice [65], however, whether or not Fasudil had an effect on NPCs in noninjured animals was not reported.

Regulation of NPC Migration to Sites of Neural Damage

Studies examining progressive neurodegenerative diseases and models of Parkinson's, Huntington's, and Alzheimer's diseases have indicated that SVZ-derived NPCs can be activated and mobilized, although to a limited degree, to the site of damage [66–71], as also reported for stroke and traumatic brain injury. Ischemic stroke induced neurogenesis in the rodent SVZ [1, 72, 73] and stimulated newborn precursor migration to the peri-infarct regions [74, 75]. This migration may be induced by SDF-1/CXCL4, which plays a role in directing newborn neurons toward an ischemic lesion [76] and may be enhanced by release of mitogenic factors, such as EGF [77, 78].

Enhancement of migration of NPCs to damaged neural tissue would be hoped to improve repair. However, although inhibition of ROCK promoted NPC migration in vitro, this was not in a directed fashion and indeed in vivo, this led to a decrease in the normal physiological migration of NPCs to the OB. This is probably not because migration was decreased per se but that the normal, directed migration along the RMS was disrupted, causing some NPCs to migrate to ectopic sites.

Use of ROCK Inhibitors for Treatment of Neural Damage

ROCK inhibitors are currently in use or in human trials for a variety of CNS and non-CNS conditions [37, 79–88]. Application of ROCK inhibitors in vivo may affect NPC cell-cell association and promote aberrant NPC migration, leading to altered NPC signal transduction. The current findings suggest that inhibition of ROCK may interfere with physiological NPC homing mechanisms, as reflected by the increased redistribution of newborn NPCs to brain regions other than the physiological integration site. This phenomenon may be a secondary result of the disruption of NPC-NPC association. Overall, our findings indicate that the use of ROCK inhibition as a clinical tool may have unexpected impact on NPCs and warrants further study to fully elucidate potential beneficial and detrimental effects.


This work was supported by National Health and Medical Research Council of Australia Project grant (#454384) and Fellowship to A.M.T. (#350226).


The authors indicate no potential conflicts of interest.