Introduction
Astrocytes make important contributions to the maintenance of the function of the mammalian central nervous system (CNS)—not only are they the most populous cells, but they play major roles in the maintenance of CNS health through their involvement in energetics, L-glutamate (Glu) homeostasis, anti-oxidant activity and release of trophic factors and gliotransmitters (Ridet et al., 1997; Maragakis and Rothstein, 2006; Parpura et al., 2012). Moreover astrocytes are well documented to be plastic cells that change their morphology, and hence biology, in response to alterations in the extracellular milieu, which may elicit short- and/or long-term responses. These morphological changes occur in normal and pathological brain tissue, and there is an ever expanding literature that astrocytes exist in diverse phenotypes across a continuum exhibiting pro-survival (“cytotrophic”) and destructive (“cytotoxic”) components (McMillian et al., 1994; Panickar and Norenberg, 2005; Sofroniew and Vinters, 2010). Astrocytes are historically considered to contribute to brain pathologies in a “secondary” mode during what has been termed reactive gliosis (Ridet et al., 1997; Maragakis and Rothstein, 2006), but there is now a solid body of growing evidence supporting their primary role in non-cell autonomous injury, where they secrete toxic entities and/or contribute to proteinopathies, driving disease progression in various neuropathologies (Lobsiger and Cleveland, 2007; Ilieva et al., 2009; Burda and Sofroniew, 2014). Astrocytes may contribute to non-cell autonomous injury in motor neurone disease (MND; amyotrophic lateral sclerosis) (Pirooznia et al., 2014) and other neurodegenerative conditions (Di Malta et al., 2012). Whilst the glial scar has long been considered a genuine target for drug development (Mueller et al., 2009), and especially the application of inhibitors of Rho kinase (ROCK) (Mueller et al., 2005), this view is simplistic given advances in our recent knowledge since astrocytes are much more than an inflammatory cell displaying adaptive plasticity in the functioning CNS. Thus many aspects of astrocyte biology offer options as attractive targets to improve their brain health and hence to effect a resultant improvement on synaptic function (Vargas and Johnson, 2010).
Our earlier research demonstrated the association between astrocytic morphology and a number of important aspects of astrocytic function, particularly the abundance and activity of Glu transporters (excitatory amino acid transporters, EAATs). We observed that altering the morphology of astrocytes, using cyclic AMP analogs or ROCK inhibitors, also increased Glu uptake (Lau et al., 2005), elevated transporter Vmax with an approximate doubling of EAAT2 expression at the cell surface and a smaller increase in EAAT1 expression as quantified by biotinylation and immunoblotting (Lau et al., 2011). Similar changes in EAAT activity or abundance were also observed in other treatments altering astrocytic morhpoholgy (Zagami et al., 2005, 2009; O'Shea et al., 2006). We concluded that ROCK inhibitor-induced elevations in Glu transporter function may contribute to their beneficial actions in brain pathologies, since enhanced EAAT activity is likely to be beneficial in CNS injury where excitotoxicity is a common mechanism effecting neurodegeneration (Beart and O'Shea, 2007; Sheldon and Robinson, 2007). Later work led to the hypothesis that changes to the astrocytic cytoskeleton induced by Rho kinase inhibitors were accompanied by the adoption of a “healthy” phenotype. We defined this pro-survival, healthy phenotype as possessing elevated expression of EAAT2, BDNF and key anti-oxidant genes. A shift in the F-/G-actin ratio in favor of G-actin, indicating a reduction in actin stress fibers and alterations to cytoskeletal signaling mechanisms (Kuhn et al., 2000), was also considered integral to this “healthy” phenotype. Much more is known about ROCK inhibitors and their ameliorative actions on destructive (“cytotoxic”) glial scarring (Mueller et al., 2005), but our findings reveal diverse “healthy” effects on the astrocyte transcriptome likely to be beneficial in brain injury.
Our continued interest in the relationship between astrocyte morphology and biology led us to apply tissue engineering (Teo et al., 2006) to astrocytes. We found that 3D poly-ε-caprolactone (PCL) scaffolds altered astrocytic responses in vivo in a model of traumatic brain injury (Nisbet et al., 2009). Here our hypothesis was promotion by the bioscaffold of a cytotrophic astrocytic phenotype, so when considered with a likely role for the extracellular matrix (ECM) (Lau et al., 2012), we speculated about links to Rho GTPases, perhaps involving the actin cytoskeleton. In primary culture, astrocytes on 3D PCL scaffolds displayed reduced cytoskeletal stress as confirmed by decreased expression of GFAP and increased G-actin (Lau et al., 2014), and, when maintained over an extended periods, possessed an extensively arborized, stellate morphology. These astrocytes showed a gene expression profile strikingly similar to that of 2D astrocytes treated with Fasudil, with up-regulation of genes for EAAT2, BDNF and anti-oxidant enzymes (Lau et al., 2014). Since 2D astrocytes treated with Rho kinase inhibitors also adopt a stellate shape, our astrocyte transcriptome (Lau et al., 2012) is likely to contain insights into previously unsuspected mechanisms given new literature on this class of molecules.
In this study, we sought to place our findings in their contemporary context (Parpura et al., 2012; Burda and Sofroniew, 2014), by further interrogating our transcriptome after Fasudil treatment through mining this astrocytic database to reveal previously unexplored biological themes. Secondly, given our success with 3D bioengineered astrocytes, we undertook additional analyses on the possible combined benefits of Rho kinase inhibitors in our 3D culture model. Together these data provide further evidence that ROCK inhibitors produce physiologically beneficial responses in astrocyte biology which are likely to be beneficial in the management of inflammation in diverse neuropathologies.
Materials and methods
Animals
C57BL/6 mice were obtained from the Florey Neuroscience Institutes (Melbourne, VIC, Australia). All experiments receive ethical approval from the Florey Neuroscience Institutes Animal Experimentation Ethics Committee (ethics approval number 07-061). Experiments were performed in accordance with the Prevention of Cruelty to Animals Act 1986 under the guidelines of the National Health and Medical Research Council Code for the Care and Use of Animals for Experimental Purposes in Australia.
Bioengineering, cytochemistry and neurochemical assays
Secondary astrocytic cultures were established from forebrain of postnatal d1.5 mice as described previously (Lau et al., 2011).
Briefly, forebrains were dissected in ice-cold solution (HBSS, Hanks balanced salt salution: 137 mM NaCl, 5.37 mM KCl, 4.1 mM NaHCO3, 0.44 mM KH2PO4, 0.13 mM Na2HPO4, 10 mM HEPES, 1 mM sodium pyruvate, 13 mM D(+)glucose, 0.01 g/L phenol red), containing 3 mg/ml bovine serum albumin (BSA) and 1.2 mM MgSO4, pH7.4). Cells were dissociated, centrifuged, and the pellet resuspended in astrocytic medium (AM: DMEM, Dulbecco's modified eagle medium, 10% FBS, 100 U/ml penicillin/streptomycin, 0.25% (v/v) Fungizone™), preheated to 36.5°C at a volume of 5 ml per brain and plated at 10 ml per 75 cm2 flask. Cells were maintained in a humidified incubator supplied with 5% CO2 at 36.5°C and complete medium changes were carried out twice weekly.
After 10 days in vitro (div), when a confluent layer had formed, the cells were shaken overnight (180 rpm) and rinsed in fresh medium to remove non-astrocytic cells. Astrocytes were subsequently detached using 5 mM EDTA (10 min at 37°C) and seeded on 96-well plates, random or aligned PCL scaffolds in 96-well plates (all 8 × 103 cells/well) or on 13 mm glass coverslips (conventional 2D controls) in 24-well plates (2 × 104 cells/well). 3D fibrous scaffolds were engineered from unfunctionalized PCL using electrospinning (Nisbet et al., 2009); thicknesses were approximately 250 and 150 μm for random and aligned scaffolds, respectively. Astrocytes were treated 8 div later with vehicle, N6,2′-O-dibutyryladenosine 3′,5′-cyclic monophosphate (dbcAMP,1 mM), or Rho kinase inhibitors Y27632 (30 μM) or Fasudil (100 μM) for a further 72 h when biochemical and morphological analyses were undertaken.
Cytochemistry for GFAP, F-actin and G-actin has been described previously (Lau et al., 2011). For immunocytochemistry, cells were washed with phosphate buffered saline (PBS:137 mM NaCl, 0.5 M Na2HPO4, 0.5 M NaH2PO4, pH 7.4) and fixed in 4% paraformaldehyde (PFA) in PBS for 10 min, followed by three washes with Tris buffered saline (TBS: 50 mM Tris–HCl, 1.5% NaCl, pH 7.6). Non-specific binding was blocked with 10% normal goat serum/normal donkey serum (NGS/NDS) in TBS containing 0.3% Triton X-100. Cells were then incubated with primary antibodies against GFAP (1:1000; Chemicon) or AHNAK (1:500; Molecular Probes), a marker of enlargeosome activity (Racchetti et al., 2012), at 4°C overnight on a rocker platform. Cells were then washed and incubated with secondary antibodies (anti-rabbit Alexa Fluor® 488 for GFAP 1:500; anti-mouse Alexa Fluor®568 for AHNAK 1:500; Molecular Probes) and Hoechst 33342 (1:500 dilution) diluted in 2% (v/v) NGS or NDS in PBS containing 0.3% (v/v) Triton X-100 for 3 h at room temperature. Cells were again washed with PBS three times at room temperature. Coverslips and scaffolds were then mounted on glass microscope slides using Dako fluorescence mounting medium and left to dry in the dark overnight. Both coverslips and scaffolds were stored at 4°C until examined by microscopy.
For concurrent labeling of F- and G-actin, cells were washed rapidly with PBS twice by vacuum aspiration and incubated in stabilizing solution (10 mM Tris base, 0.15 M NaCl, 0.01% Triton X-100, 2 mM MgCl2, 0.2 mM DTT (Bio Vectra, Canada), 10% glycerol) for 1 min at 4°C. Cells were then washed rapidly with chilled PBS (4°C) twice and fixed in 4% (v/v) PFA in PBS for 15 min at 4°C. Cells were washed in PBS twice at room temperature and excess PFA was quenched by adding 50 mM NH4Cl in PBS for 15 min at room temperature. Cells were permeabilized in 0.5% (v/v) Triton X-100 in PBS for 5 min and incubated in blocking solution (2% (v/v) BSA, 0.1% (v/v) Triton X-100 in PBS) for 15 min at room temperature. Following another two washes with PBS, cells were incubated in dye solution (DNaseI-Alexa Fluor 488 1:250; Molecular Probes; rhodamine-phalloidin 1:125 in blocking solution; BDH) for 30 min in the dark. Cells were then washed in PBS three times and mounted on glass microscope slides using Dako fluorescence mounting medium and left to dry in the dark. Cells on scaffolds were mounted by placing the scaffolds with the cells on top, mounted with Dako fluorescence mounting medium and glass coverslips (13 mm round; Menzel-Glaser). Slides were stored at 4°C until required for imaging.
Methods for measurement of cellular viability [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT; an index of mitochondrial function) and lactate dehydrogenase assays] have been published (Lau et al., 2011). After treatmentd, MTT was added to the wells give a final concentration of 0.5 mg/ml, incubated with the cells at 36.5°C with 5% CO2 for 30 min. Media were aspirated and 300 μl of dimethyl sulfoxide (DMSO) was added into each well to dissolve the formazan product. The absorbance was subsequently measured at 570 nm using a Bio-Rad Benchmark Plus microplate spectrophometer. Lactate dehydrogenase (LDH) assay was carried out using a commercially available kit (Roche). Medium (50 μl) was collected from each well and placed in a 96-well plate. The samples were incubated with the reaction mixture (Cytotoxicity Detection Kit from Roche) according to the manufacturer's protocol and left in the dark for 30 min. The absorbance of the sample mixture was determined at 490 nm using a Bio-Rad Benchmark Plus microplate spectrophometer. Data from these experiments were analyzed using Two-Way repeated-measures ANOVA with Bonferroni's post-hoc test using Graphpad Prism software (Version 6).
To examine changes in relative abundance of F-actin and G-actin, the G-actin image was “subtracted” from the corresponding F-actin image after both images were converted to gray-scale, and integrated optical density was measured using ImageJ (NIH: version 1.37). Image analysis used data from 4 images/well from 2 wells/culture over 3 independent cultures. The average value for all fields subjected to the same treatment in an individual experiment was analyzed as a single data point. All images for each form of actin were obtained using the same exposure settings. Statistical comparisons were made using Two-Way repeated-measures ANOVA with Bonferroni's post-hoc test using Graphpad Prism software (Version 6).
Microarray analyses
Full details have been given previously (Lau et al., 2012), where we validated microarray data by quantitative RT-PCR. Differentially expressed genes between control and Fasudil-treated samples at each time point were then filtered to include only those passing a stringent false cut off of 0.05. Data have been previously deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE25829 (Lau et al., 2012).
Discussion
Given astrocytes are a plastic glial population existing in various morphologies and displaying diverse biologies, major and wide-ranging effects would be expected upon manipulation of the ROCK/Rho system, a key determinant via actin of cellular survival, migration and proliferation (Riento and Ridley, 2003). Whilst our initial bioinformatic analyses of transcriptomic changes induced by Fasudil were directed at major biological processes regulating cytoskeletal reorganization, we also found significant changes in expression of a diverse group of genes associated with astrocyte function, and suggested that overall ROCK inhibitors would produce “healthy and physiologically beneficial responses in astrocyte biology” (Lau et al., 2012). We thus postulated the existence of a pro-survival, cytotrophic phenotype wherein the essential criteria were a preponderance of G-actin and elevated expression of EAAT2, BDNF and key anti-oxidant genes. Reappraisal of our bioinformatic data revealed diverse additional effects on the astrocyte transcriptome likely to be beneficial in brain injury. Gene expression profiles of motor and autophagic cellular cascades and inflammatory/angiogenic responses were all inhibited favoring adoption of what might be considered “healthy,” anti-migratory phenotype. These types of changes generally have not been a focus in astrocytes, although often documented in neurones, but there is the beginnings of an new literature describing roles for Rho GTPases or actions of ROCK inhibitors in the biological processes reported here (vide supra). This concept of an anti-migratory phenotype (Cárdenas et al., 2014) is an interesting one for in vitro ROCK inhibitors produced very rapid stellation of astrocytes (approximately 15 min with most changes complete by 3–6 hr), a morphology which in uninjured brain is considered to reflect non-migratory properties, and here may indicate Fasudil produces a normal cytotrophic astrocyte. This state may resemble cytotrophic components of minimal, self-resolving hypertrophy often stated to occur in minor trauma/injury wherein there is re-establishment of a healthy physiological phenotype (Balasingam and Yong, 1996; Sofroniew, 2009; Burda and Sofroniew, 2014). Earlier work has documented the ability of ROCK inhibitors to produce extension of GFAP-positive, presumed astrocytic processes in vitro and in vivo models of nerve crush (Sagawa et al., 2007; Ichikawa et al., 2008). Extensive process formation has also been noted with ROCK inhibitors in wound healing models using astrocytes (Holtje et al., 2005). Such morphological rearrangements as discussed here are underpinned by extensive remodeling of the actin cytoskeleton, particularly lamellipodia and filopodia (Le Clainche and Carlier, 2008; Mattila and Lappalainen, 2008).
Tissue engineering in combination with materials science has been also been employed to manipulate astrocyte biology. We have previously reported in an in vivo model of traumatic brain injury (Nisbet et al., 2009) that, in the presence of PCL scaffolds, astrocytes provided signals encouraging neuritic infiltration of injured tissue, with findings favoring the existence of early cytotoxic and late cytotrophic components of astrogliosis. Others have reported variable data with astrocytes in tissue engineering and the literature suggests that the type of engineering approach employed greatly influences the nature of the outcome, and specifically the temporal contributions of cytotrophic vs. cytotoxic astrogliosis (Iannotti et al., 2003; Wong et al., 2007; Nisbet et al., 2010). Nevertheless our success in vivo encouraged us to pursue mechanistic studies with tissue engineering of astrocytes on PCL bioscaffolds where we found in 3D extensive process formation, stellation and adoption of a cytotropic phenotype resembling that found in 2D astrocytes treated with ROCK inhibitors (vide supra). There have been remarkably few successes where bioengineering strategies in vitro have led to the establishment of viable astrocytes on bioscaffolds (Puschmann et al., 2013; Zuidema et al., 2014), but our use of secondary astrocytes allowed the long term maintenance of mature cells on PCL bioscaffolds (Lau et al., 2014). Given there was a body of evidence for beneficial effects of ROCK inhibitors in models of head and spinal trauma (Raad et al., 2012; Watzlawick et al., 2014) we extended our study to explore whether the inclusion of Fasudil and Y27632 would provide further benefits in our 3D astrocyte model. Here we found that both ROCK inhibitors produced additional GFAP-positive processes relative to PCL scaffolds alone, and there seemed a bonus of a further shift to an even greater preponderance of G-actin relative to F-actin. AHNAK, a marker of enlargeosome activity and migration (Racchetti et al., 2012), proved very suitable for immunostaining of fine astrocytic processes, being partially co-localized with GFAP, but revealing much more of the astrocyte arbor than GFAP, an effect which was particularly obvious on aligned bioscaffolds. Overall these data are consistent with ROCK inhibitors providing further beneficial effects over and above PCL scaffolds alone. Preliminary evidence from Western immunoblotting suggested ROCK inhibitors reduced GFAP expression relative to dbcAMP (Supplementary Figure 1), whilst patterns of AHNAK expression were generally consistent with immunocytochemistry. In ongoing work examining Glu transporter activity, we confirmed our previously reported elevation of uptake in 2D astrocytes by Fasudil and Y27632 (Lau et al., 2011), and found EAAT activity appeared to be elevated 2–4 fold in astrocytes maintained on bioscaffolds (data not shown). Whilst we need to undertake further experiments to document fully the phenotype of the astrocytes found here, it does seem that the phenotype may be shifted even further toward the direction of cytotrophic astrogliosis. Our findings here, demonstrating that Fasudil and Y27632 under certain conditions either increased cell viability or decreased cell damage compared with control conditions (Figure 4), provide further evidence for the potential benefits of ROCK inhibitors via direct effects on astrocytes, which may contribute to the beneficial outcomes of these treatments in brain injury. Taken together there would seem to be a case for combining Rho kinase inhibition with tissue engineering in models of traumatic brain and spinal cord injury.
ROCK inhibitors, and especially Fasudil, have been examined in various injury models where astrocytes may contribute to the pathology, perhaps via cytotoxic inflammation, and/or by compromising recovery from trauma/neurodegeneration. Glia, including astrocytes, are known to contribute to the neuropathology of MND through non-cell autonomous mechanisms (Vargas and Johnson, 2010; Pirooznia et al., 2014). MND is a rapidly advancing degenerative condition where suitable new therapeutic strategies are badly needed (Turner and Talbot, 2008), so the pro-survival response produced by Fasudil in the SOD1 mouse model of MND is an impressive advance (Takata et al., 2013; Tonges et al., 2014) since the rapid progression of disease in this model generally does not respond to interventions (Turner and Talbot, 2008). Attentuation of astroglial pathology was noted early after Fasudil treatment whereas beneficial effects were noted at most stages on microglial mediated inflammation (Tonges et al., 2014). Rho kinase inhibitors may also be useful in other neurodegenerative conditions where astrocytes contribute to inflammation, perhaps by non-cell autonomous mechanisms. For example in Alzheimer's disease, where astrocytic mechanisms are linked to disease risk factors and where their contribution to synaptic signaling is compromised by amyloid-β peptide (Talantova et al., 2013), Fasudil may have therapeutic potential as it suppressed the inflammation in rodent hippocampus induced by amyloid-β peptide (Song et al., 2013). In experimental autoimmune encephalomyelitis, an animal model of multiple sclerosis, Fasudil reduced inflammation and demyelination. Early or late administration of Fasudil exerted beneficial effects producing a shift in macrophage function from the cytotoxic M1 to the anti-inflammatory or reparative M2 phenotype in spinal cord and spleen (Liu et al., 2013). Subsequently the same laboratory suggested that at least part of this action of Fasudil was via a diminution of cytotoxic astrogliosis and less infiltration of inflammatory cells across the blood brain barrier (Guo et al., 2014). Interestingly, astrocyte phenotype is now recognized to determine the outcome of CNS repair and myelination, and components of astrocyte biology likely represent valid targets to enhance lesion repair in multiple sclerosis (Barnett and Linington, 2013).
Given the seminal role astrocytes play in synaptic transmission and maintenance of brain function generally, their diverse biology offers many options for potentially “druggable” targets—beneficial shifts to cytotrophic phenotypes would improve their overall health and ameliorate cytotoxic inflammation in brain pathologies, and thus conceivably allow maintenance of appropriate synaptic function (Vargas and Johnson, 2010). Another often not discussed aspect of astrocyte biology is their multiple “morphological” interfaces, not only via the communication of astrocytic tight junctions, but also with different cellular populations viz. neurones, oligodendrocytes, blood vessels, blood brain barrier and microglia (Volterra and Meldolesi, 2005), including via the quad-partite synapse (Schafer et al., 2013). Thus astrocytes are seminally placed from an organizational perspective to orchestrate the biology of these different CNS populations by integrating synaptic and non-synaptic signaling. The recent work with ROCK inhibitors showing cytotrophic changes in astrocyte function in disease models of MND (Tonges et al., 2014) and multiple sclerosis (Guo et al., 2014) are particularly encouraging. Indeed, we speculate that Fasudil-induced changes in astrocytic phentotype exert beneficial effects which should be taken in a similar context to the much popularized “healthy” shift in macrophage function (now extended to microglia) from the cytotoxic M1 to the reparative M2 phenotype—and which is produced by Fasudil in spinal cord in experimental autoimmune encephalomyelitis (Liu et al., 2013). Meta-analysis of data from experimental studies of spinal cord injury evaluating Rho A/ROCK blockade found significant overall improvement of locomotor function. Overall a possible role in inflammatory events was noted and the strategy was considered a plausible one for management of human spinal cord injury (Watzlawick et al., 2014). Certainly it is clear with advances in the design of increasingly effective ROCK inhibitors (Guan et al., 2013), and with new developments in tissue engineering and drug delivery via nanoparticles that ROCK inhibitors alone or in concert with these new technologies are likely to be widely applicable to management of inflammation in neurodegenerative conditions.
Conflict of interest statement
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.