TGFβ1 Induces Axonal Outgrowth via ALK5/PKA/SMURF1-Mediated Degradation of RhoA and Stabilization of PAR6

Visual Abstract


Introduction
Interruption of the brain's blood supply, e.g., in case of ischemic stroke, can result in life-long disability because of the loss of neurons (Murphy and Corbett, 2009;Cramer, 2018). Most available therapeutics for stroke patients target the acute phase in the hope to protect neurons from the ischemic damage, while rehabilitation to date remains the only treatment option for chronic stroke patients (Krakauer et al., 2012;Zeiler and Krakauer, 2013;Wahl and Schwab, 2014). Both in human stroke patients as well as animal models of stroke, the spontaneous recovery observed in the weeks after the incident has been attributed in parts to structural plasticity of healthy cortical neurons and connections, e.g., horizontal neurons in the peri-infarct region or the contralateral corticospinal neurons (CSNs) in case of large strokes (Carmichael et al., 2017). In this regard, a better understanding of the molecular cues inducing, guiding and maintaining this compensatory sprouting and rewiring response is pivotal to develop novel therapeutic approaches to enhance recovery after stroke. Inflammation-derived cytokines and locally released growth factors may have a beneficial effect on the repair mechanisms by directly promoting axonal regeneration (Benowitz and Popovich, 2011;Vidal et al., 2013). In the cortex surrounding a focal stroke lesion the upregulated cytokine growth differentiation factor 10 (GDF10) was reported to enhance structural plasticity and motor recovery . GDF10 is a member of the highly evolutionarily conserved transforming growth factor (TGF) superfamily, which also includes the TGFb and bone morphogenic protein (BMP) cytokine families (Weiss and Attisano, 2013;Zhang et al., 2017). A recent study highlighted the involvement of TGFb 1 signaling in neurogenesis and axonal regeneration in the peri-infarct cortex by viral knockdown of the co-receptor ALK5, leading to decreased levels of the growth associated protein GAP43 within the first two weeks after stroke, concomitant with a decrease in functional recovery (Zhang et al., 2019). A recently published transcriptomic screen showed an upregulation of TGFb 1 in the stroke-denervated cervical hemicord of adult mice at 28 d after a large cortical stroke. This finding suggests that TGFb 1 could be a mediator of the observed compensatory sprouting of the intact-side corticospinal tract (CST) and axon elongation/arborization in the spinal cord (Kaiser et al., 2019).
In the present study, we show that TGFb 1 is transcriptionally upregulated in the region of the axotomized cervical CST and in the premotor layers of the CST-deprived spinal gray matter as early as 4 d postinjury (dpi). Using a simple but highly reproducible in vitro assay, we demonstrate that TGFb 1 treatment rescues neurite outgrowth in the growth inhibitory environment of crude spinal cord extract (SCE) through the canonical TGFb 1/ALK5 signaling axis. Further hypothesis driven pharmacological blockade studies suggest an underlying signaling mechanism involving a PKA-mediated phosphorylation of the E3 ubiquitin ligase SMURF1, switching its substrate preference from PAR6 to Ras homolog A (RhoA). In conclusion, we show that besides canonical transcriptional changes associated with TGFb 1, it also enhances neurite outgrowth by downregulating RhoA, the downstream signaling mediator of many CNS-associated growth inhibitory molecules. Thereby TGFb 1 has the potential to prime the neuronal cytoskeleton into a growth permissive state despite of the inhibitory CNS environment.

Animals
All animal experiments were performed with the approval of and in strict accordance with the guidelines of the Zurich Cantonal Veterinary Office. A total of n = 21 adult C57BL/6J mice (two to three months, 20-28 g, female, Charles River Laboratories) were used in this study. Only one sex was used to minimize lesion size variability and animal numbers. Animals were housed in groups of four to five under a constant 12/12 h light/dark cycle with food and water ad libitum.

Photothrombotic stroke
For all surgeries, mice were initially anesthetized using 3-4% isoflurane, transferred to a stereotactic frame (Kopf Instruments) and kept at 1-2% isoflurane throughout the surgery. Body temperature was maintained at 37°C on a heating pad. All animals received a unilateral photothrombotic stroke to lesion the right side sensorimotor cortex as previously described (Watson et al., 1985;Kaiser et al., 2019). Briefly, the skull was exposed by a midline incision of the scalp. An opaque template with a defined opening (3 Â 5 mm) was aligned to the midline over the right motor and premotor cortex (À2 to 13 mm A/P, 0-3 mm M/L related to bregma). Five minutes after intraperitoneal injection of 0.1 ml Rose Bengal (10 mg/ml in 0.9% NaCl; Sigma-Aldrich), the skull was illuminated for 10.5 min with a cold light source (Olympus, KL1500LDC, 150 W, 3000 K) placed firmly on top of the skull. Control animals were given a sham operation without illumination of the skull. Postoperative care included recovery on a heating mat, sustained analgesia provided via drinking water (Novalgin, 2 mg/ml with 5% sucrose) and antibiotic treatment where necessary for 3 d.

Behavioral testing
Behavioral tests were performed before surgery (baseline) as well as 4, 7, 14, and 28 d after photothrombotic stroke (dpi) of the right motor cortex. Poststroke impairment and recovery of forelimb function was assessed using the cylinder test. Forelimb paw touches to the cylinder wall during spontaneous rearing behavior were recorded for 20 min or 30 rears in total (n = 12). Paw dragging of the affected limb was scored as the percentage of paw drags divided by total number of paw touches (Roome and Vanderluit, 2015). The horizontal ladder walk test was used as an additional, more sensitive assessment of skilled limb placement (Metz and Whishaw, 2009). Mice were habituated to the ladder run 1 d before test recordings; no further training was conducted. Three trials on a 40-cm-long ladder with irregularly spaced rungs of 1-to 2cm distance were recorded on each testing day (n = 12). The number of foot errors was measured as the number of total misplacements (slips) divided by the total number of steps taken. All recorded steps were analyzed and no videos were excluded for the analyses to avoid bias toward stroked groups (n = 3 videos per animal per test day).

Perfusion fixation and tissue processing
All mice were terminally anesthetized with 3-5% isoflurane followed by injection of pentobarbital (300 mg/kg body weight, i.p.; Streuli Pharma AG). Animals were transcardially perfused with ice-cold Ringer's solution [containing 10 5 IU/l heparin (Roche) and 0.25% NaNO 2 ]. Brains and spinal cords were quickly dissected and snap-frozen on dry ice to preserve RNA quality. The brains were immersed in 4% PFA overnight before being transferred to 30% sucrose in phosphate buffer (PB) for cryoprotection. Samples were blinded from the point of tissue harvesting.

Analysis of lesion completeness
For the accurate analysis of lesion size, brain crosssections (40 mm) were stained on-slide with cresyl violet solution for 1 min, dehydrated in a series of increasing ethanol concentrations and washed in Xylol before coverslipping with Eukitt (Sigma-Aldrich). Brain sections at four defined landmarks (1.98, 0.98, À0.22, and À1.34 mm, in relation to bregma) were analyzed for stroke volume and depth of the cortical lesion. Average cortical stroke depth was calculated as the average of lesion depth across these four landmarks.

In situ hybridization
Sections (16 mm) of fresh frozen tissue were cut on a cryostat at À20°C and stored at À80°C. In situ gene expression was assessed using the RNAScope protocol (Advanced Cell Diagnostics, RNAscope Fluorescent Multiplex Assay) according to the manufacturer's protocol. Briefly, slices were fixed in 4% PFA for 20 min before being hybridized to the probe (RNAScope: TGF-b 1, catalog #407751), which was further amplified using the branched DNA amplification methods. Sections were counterstained with DAPI and digitalized (Zeiss, Axio Scan.Z1, 200Â). For analysis using ImageJ/FIJI, only slices with tears or folds were excluded; n = 3 sections per animal per group and spinal levels C5 and C6 were randomly selected. Quantification of mRNA expression was expressed as the percentage of white signal over black background in defined regions (CST in the dorsal funiculus and intermediate gray matter (iGM) of the denervated hemicord) after applying a threshold that was manually defined for five randomly selected images of the dataset and averaged.

Neurite outgrowth assay
Candidate factors were tested using a previously described neurite outgrowth assay (Maibach et al., 2020). Briefly, N1E-115 mouse neuroblastoma cells were plated at a density of 10,000 cells/cm 2 in Neurobasal medium supplemented with 2% L-glutamine and 1% PenStrep to induce neuron-like differentiation. After 24 h of differentiation the cells were supplemented with crude adult rat spinal cord CHAPS extract, and candidate factors or pharmacological agents were added (Maibach et al., 2020). The following proteins and molecules were used at the indicated concentrations: 1 ng/ml TGFb 1 (R&D Systems), 1 nM TEW7197 (Selleckchem, S7530), 1 nM ML347 (Selleckchem, S7148), 10 nM SIS3 (Selleckchem, S7959), 10 nM A01 (Merck, SML1404), 10 nM Ht31 (Tocris Biotechne, 6286), and 10 nM ATM (Merck, 157201). After an additional 24-h outgrowth phase, the cells were fixed and counterstained with Coomassie solution (0.25% Coomassie Brilliant Blue R250, 50% MeOH, 10% HoAC). Images were acquired randomly over the wells at 10Â magnification. Mean neurite outgrowth per cell was quantified in ImageJ by applying a grid to the pictures and counting intersections of neurites with the grid lines and relating this number to the total number of cell bodies in the corresponding well (Rønn et al., 2000). Experiments were conducted in five biological replicates with three technical replicates per condition.

Immunoblotting
Cells were washed twice in PBS on ice and lysed in RIPA buffer (150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, and 50 mM Tris; pH 8) containing 2Â HALT phosphatase inhibitor cocktail and 5 mM EDTA. The lysates were incubated on ice for 30 min and centrifuged at 13,000 Â g for 15 min at 4°C. The supernatants were collected and stored at -80°C.
The samples were prepared in Laemmli buffer (Bio-Rad) supplemented with 10% b MEtOH and denatured at 90°C for 3 min. The samples were separated on pre-cast 4-15% Mini PROTEAN R TXG TM gels (Bio-Rad) at 250 V in Tris-glycine running buffer (25 mM Tris, 192 mM glycine, and 0.1% SDS; pH 8.3). Proteins were transferred onto a 0.45 mm PVDF membrane in Tris-glycine transfer buffer (25 mM Tris, 19 2 mM glycine, and 20% MeOH) for 90 min with a constant current of 300 mA. Subsequently, membranes were blocked for 1 h with 5% BSA (Sigma) in TBS-T (10 mM Tris, 150 mM NaCl, and 0.01% Tween 20; pH 7.5) and probed with primary antibodies overnight at 4°C. The membranes were washed three times in TBS-T, probed with secondary HRPcoupled antibodies (Thermo Fisher Scientific) at a concentration of 0.05-0.1 mg/ml for 1 h at room temperature (RT) and washed again three times in TBS-T. Detection was performed using SuperSignal West PICO (Thermo Scientific) or WesternBright Sirius (Advansta) chemiluminescent substrates, and images were acquired on the Gel Doc imager (Bio-Rad). Densitometry analysis was performed with ImageJ/FIJI software, and values were normalized to the housekeeping gene GAPDH or total protein and the mean value of the corresponding control group. For Western blotting, the primary antibodies are summarized in Table 1.

RhoA assessment in growth cones
Cells were fixed for 15 min in 4% PFA and subsequently blocked and permeabilized in blocking buffer (0.1% Triton X-100 and 10% BSA in PBS) for 1 h. Primary antibody [mouse anti-RhoA (Santa Cruz, SC-418), 1:500] was applied in 1% BSA in PBS and incubated over night at 4°C. The samples were washed three times in PBS, followed by incubation with secondary antibodies (1 mg/ml antimouse Cy3-coupled antibodies; Invitrogen, A10521, A10520) in 1% BSA in PBS for 1 h at RT. After washing three times in PBS, cells were counterstained with 1:100 Phalloidin Alexa Fluor 488 (Invitrogen, A12379) to visualize the actin cytoskeleton and DAPI (50 nM, Invitrogen, D3571) as a nuclear counterstain. The samples were coverslipped and imaged at 200Â. For image acquisition, exposure times were kept constant and below gray scale saturation. For immunofluorescence normalization in ImageJ/FIJI, the signal in the phalloidin channel was thresholded and an area mask was created around the fluorescent object. This mask was then applied onto the RhoA pictures and the total pixel intensity within the area was measured. This value was normalized to the area of the growth cone or cell body. Similarly, a nuclear counter stain was used to define the area of the nucleus.

Statistical analysis
Statistical analysis was performed with Prism 7.0 (GraphPad Software). To detect differences between groups and within groups over time, two-way ANOVA followed by Dunnett's multiple-comparisons (MC) was used. Other symbols might be used to indicate two comparisons in one graph. In bar graphs, all data are plotted as mean standard error of the mean (6SEM), while data were normalized to solvent controls in the neurite outgrowth assays. Dots in the behavioral graphs represent individual animals. For all neurite outgrowth assays, n = 5 biological replicates with n = 3

Results
TGFb1 is endogenously upregulated at early time points in the stroke denervated hemicord In mice, a significant degree of spontaneous re-innervation of the stroke denervated cervical spinal cord by sprouting of contralesional CSNs can be detected 28 d after large lesions to the motor cortex (Ueno et al., 2012;Bachmann et al., 2014;Kaiser et al., 2019). At this late time point after stroke, TGFb 1 expression was found to be upregulated in the cortex (Lehrmann et al., 1995;Knuckey et al., 1996;Ata et al., 1999;Zhu et al., 2001;Doyle et al., 2010) as well as in the stroke-denervated hemicord (Kaiser et al., 2019). However, sprouting of contralateral cortical axons may be initiated much earlier in the stroke-denervated hemicord. To address whether TGFb 1 could serve as a trigger and modulator of structural plasticity in the stroke-denervated hemicord, we evaluated the expression pattern of TGFb 1 within the dorsal funiculus and gray matter at selected, early time points after stroke (2, 4, 7, and 28 dpi; Fig. 1A,B). We ensured that stroke lesions were consistent in size and location at all time points. Strokes were primarily localized to the sensory-motor cortex including the premotor regions and successfully ablated all cortical layers with little to no impact on the corpus callosum or deeper structures (Fig.  1C) .This stroke model induced a behavioral deficit in the forelimbs which recovered partially over the course of four weeks as assessed by paw dragging and foot faults in the horizontal ladder task (Fig. 1D). In situ hybridization for TGFb 1 in the stroke-denervated spinal cord showed an increase of TGFb 1 mRNA at 4 dpi in both analyzed regions, the CST domain of the dorsal funiculus as well as the iGM (Fig. 1E,F). This increased expression is transient Figure 1. TGFb 1 is upregulated in the stroke-denervated cervical spinal cord at early time points, transiently in the iGM and persistently in the degenerating CST in the dorsal funiculus. A, B, Experimental study timeline: adult female C57BL/6 mice received a large unilateral cortical stroke (sham: n = 6; 2 dpi: n = 3; 4 dpi: n = 3; 7 dpi: n = 4; 28 dpi: n = 5; total: n = 21). All animals were tested behaviorally and euthanized at 2, 4, 7, or 28 dpi, respectively. C, Average cortical stroke depth showed no significant differences among the experimental groups with strokes reaching into deep cortical layers to the corpus callosum (100%). Estimated location of CSNs in Layer V is indicated by a red dotted line. The sensorimotor area was specifically injured (Nissl staining at 7 dpi, representative picture). Scale bar = 500 mm. D, Behavioral analysis for paw drags in the cylinder test and number of missteps on the irregular horizontal ladder shows that stroke induction resulted in a deficit in motor behavior with a subsequent, partial functional recovery within 28 dpi. # is used for statistical comparison to baseline levels, p for comparison to 4 dpi. E, Representative pictures of TGFb 1 mRNA expression in the degenerating CST and iGM of the spinal levels C5/6 in sham animals as well as in mice at 2, 4, 7, and 28 dpi. Scale bar = 50 mm. F, Quantification of the percentage of area signal in the CST shows an upregulation of TGFb 1 mRNA starting at 4 dpi and persisting until 28 dpi. In the iGM, TGFb 1 mRNA is transiently upregulated at 2 and 4 dpi. pp , 0.05, ppp , 0.01, pppp , 0.001.
Research Article: New Research in the iGM and persists up to 28 dpi in the CST, suggesting that TGFb 1 is present at time points when growth initiation and axon elongation and arborization of contralesional CSNs and other tracts occur.

TGFb1 rescues neurite outgrowth in inhibitory environment via canonical ALK5/SMAD3 signaling
We first investigated the potential of TGFb 1 to rescue neurite outgrowth in a growth inhibitory environment. To model the inhibitory in vivo CNS environment, we used our previously established neurite outgrowth assay (Maibach et al., 2020). In this assay, application of crude spinal cord CHAPS extract (SCE) to the differentiated neuron-like cell line N1E-115 over a 24-h process formation phase, inhibits outgrowth in a dose-dependent manner, without affecting cell survival. As TGFb 1 was shown to increase axon length in neurons, including primary cortical neurons (Abe et al., 1996;Knöferle et al., 2010;, we first established an effect of TGFb 1 in our neurite outgrowth model. We found that already nano molar concentrations of TGFb 1 were able to restore neurite outgrowth of IC50 SCE-treated N1E-115 cells to ;80% ( Fig. 2A,B).
We investigated the specific receptor complex through which the neurite outgrowth promoting TGFb 1 signaling might be transduced (Fig. 2C). While pharmacological inhibition of ALK1 had no effect on the TGFb 1-mediated rescue of neurite outgrowth in the inhibitory SCE treatment conditions, inhibition of ALK5 prevented the TGFb 1-mediated rescue (Fig. 2D,E). This suggests that the TGFb 1 signal is propagated via the canonical TGFb R2/ALK5 signaling receptor complex. Phospho-profiling of the ALK5 downstream effector SMAD3 revealed an increased phosphorylation/activation of SMAD3 in the TGFb 1treated conditions (Fig. 2F,G). Accordingly, pharmacological inhibition of SMAD3 prevented the TGFb 1mediated rescue effect (Fig. 2H). Taken together, these results demonstrate that TGFb 1 signals via the canonical TGFb R2/ALK5/SMAD3 axis to rescue neurite outgrowth under CNS growth inhibitory conditions.

TGFb1 induces SMURF1-mediated downregulation of RhoA and stabilization of PAR6
We did not observe an increase in neurite outgrowth in the absence of SCE after 24 h of TGFb 1 treatment alone ( Fig. 1B) with the extremely low TGFb 1 concentration used in this study. This lack of a general growth promoting effect suggests that TGFb 1 signaling specifically interferes with and cancels the SCE-induced growth inhibitory signaling cascades. We therefore hypothesized that TGFb 1 could negatively influence the RhoA/Rho-associated, coiled-coil containing protein kinase (ROCK) pathway, a key downstream signaling effector of many growth inhibitory molecules (Thiede-Stan and Schwab, 2015). One possible mechanism for such crosstalk could be the SMURF1-mediated RhoA ubiquitination which targets RhoA for degradation and thereby enhances neurite outgrowth (Wang et al., 2003;Smith et al., 2005;Sahai et al., 2007;Vohra et al., 2007;Narimatsu et al., 2009). The substrate preference of SMURF1 is modulated by PKA-dependent phosphorylation, which increases its binding affinity for RhoA relative to PAR6 (Cheng et al., 2011). In parallel to enhanced RhoA degradation, the membrane associated PAR6/PKC complex is stabilized. Both PKA activation (via association with SMAD3) and induction of a PAR6/ protein kinase C (PKC) complex at the membrane have been reported as downstream effectors of TGFb 1 signaling in endothelial cells (Ozdamar et al., 2005;Wang et al., 1998;Yang et al., 2013), making it a possible mechanism for the observed rescue of neurite outgrowth by TGFb 1.
To test this hypothesis, we inhibited selected key molecules of the proposed non-canonical signaling cascade (Fig. 3A). Inhibition of SMURF1 by the blocker A01 (Cao et al., 2014) prevented the TGFb 1 elicited rescue of neurite outgrowth from SCE-induced growth inhibition (Fig. 3B). Likewise, inhibition of substrate recruiting PKA scaffolding protein A kinase anchoring protein (AKAP) by Ht31 (Kennedy and Scott, 2015) as well as pharmacological inhibition of PKC and PAR6 interaction ATM (Butler et al., 2015;Erdogan et al., 2006;Stallings-Mann et al., 2006) resulted in the loss of the TGFb 1mediated rescue effect (Fig. 3C,D). Furthermore, TGFb 1 co-treatment with SCE resulted in a significant downregulation of RhoA levels both globally (Fig. 3E,F) and locally in growth cones (Fig. 3H,I). This reduction of RhoA was associated with increased PAR6 levels (Fig.  3E,G), supporting the hypothesis of a TGFb 1 induced switch in SMURF1 substrate preference leading to lower levels of RhoA and higher levels of PAR6 as an underlying mechanism for the observed neurite outgrowth recovery.

Discussion
We observed an upregulation of TGFb 1 mRNA at 4 dpi in the stroke-affected CST and the iGM (laminae 5-7) in the cervical spinal cord after a large stroke to the motor cortex, a time point when compensatory axonal sprouting of the CST may be triggered. This increased expression persisted in the CST area in the dorsal funiculus up to 28 dpi. In vitro neurite outgrowth analysis showed that TGFb 1 can rescue neurite outgrowth through the canonical ALK5/SMAD3 signaling axis in a growth inhibitory environment. Downstream of the TGFb R2/ALK5 receptor, TGFb 1 led to the accumulation of PAR6, and negatively impacted on the RhoA/ROCK pathway by reducing the SCE-induced accumulation of RhoA. We propose an underlying signaling mechanism involving PKA-dependent SMURF1 regulation, leading to a substrate preference switch of SMURF1 from PAR6 to RhoA, thereby decreasing RhoA and increasing the growth enhancing state of the cytoskeleton. Such a mechanism could be the basis on which TGFb 1 mediates compensatory sprouting of reinnervating CSN axons in the stroke-denervated spinal cord.
Following stroke, TGFb 1 has been reported to exert neuroprotective effects. Administration of TGFb 1 led to reduced infarct sizes (Gross et al., 1993;Prehn et al., 1993), while blocking TGFb 1 signaling led to increased ischemic damage (Ruocco et al., 1999). Direct neuronal or glial effects could not be distinguished from anti-inflammatory effects, however. Evidence for TGFb 1 as a regulator of axonal growth is sparse and contradictory. Several in vitro studies correlated TGFb 1 treatment with axonal growth promotion in several subtypes of neurons (Abe et al., 1996;Stegmüller et al., 2008;Knöferle et al., 2010;Walshe et al., 2011;Li et al., 2015), while others reported an inhibitory or no effect at all (Ho et al., 2000;Do et al., 2013;Jaskova et al., 2014). These varying effects may be driven by context-specific and cell-specific signaling of TGFb (Massagué, 2012); our present finding that TGFb 1 alone did not affect neurite outgrowth, but reversed the growth inhibitory effect of a crude spinal cord protein extract are in line with such a concept. The importance of studying the effects of TGFb 1 on axonal growth in vivo is thus evident, however, interpretation of the results may be complex owing to the pleiotropic roles of TGFb in the control of cell proliferation, differentiation, wound healing and immune system (Kang et al., 2009;Morikawa et al., 2016).
Previous studies have reported indirect effects of TGFb 1 on modulation of axonal growth , e.g., by induction of astrocyte proliferation (Kohta et al., 2009). Endogenous TGFb 1 may also serve as a trophic factor as TGFb 1 deficiency leads to an exacerbated neuronal cell death after facial nerve axotomy (Makwana et al., 2007). TGFb 1 can be secreted by astrocytes and by immune cells, thereby often functioning as an immunosuppressant (Vidal et al., 2013;Liddelow et al., 2017;Tripathi et al., 2017). Recently, TGFb 1/ALK5 signaling was associated with increased dendritic plasticity of cortical neurons, axonal sprouting of corticorubral projections and increased motor recovery after ischemic reperfusion injury (Zhang et al., 2019).
A previous study reported an upregulation of TGFb 1 mRNA in the stroke-denervated spinal cord at 28 dpi (Kaiser et al., 2019), suggesting that TGFb 1 might serve as a local tissue-derived pro-regenerative cue for re-innervating axons after cortical stroke. Here, we observed an increase in TGFb 1 mRNA as early as 4 dpi. While we found a transient increased expression of TGFb 1 in the iGM, this increase persisted in the dorsal funiculus in the CST area. Interestingly, this temporal and spatial expression profile matches the previously described activation of microglia and infiltration of macrophages (Kaiser et al., 2019), suggesting that these cells may be the predominant source of TGFb 1 in the stroke-denervated spinal cord. This interpretation is in line with a previous finding of increased TGFb 1 expression in microglia/macrophages in the stroke penumbra early after middle cerebral artery occlusion (Doyle et al., 2010).
We addressed the potential downstream pathways of TGFb 1 leading to increased neurite outgrowth under growth inhibitory conditions. We conducted our studies using aa in vitro model the neuroblastoma-derived cell line N1E-115 treated with crude SCE (Maibach et al., 2020). While this model lacks the direct translational impact of primary cell cultures, it ensures a high reproducibility between studies that primary cultures fail to provide owing to an often large batch-to-batch variability and contamination with non-neuronal cells. High sensitivity and reproducibility are a crucial prerequisite for a routine assay that is used for therapeutic compound screening. By using the well characterized neuron-like N1E cells, we were also able to avoid non-neuronal cell contamination by, e.g., astrocytes, which might serve as a primary source for TGFb 1 and thus interfere with our results in this study. An additional advantage of this assay is the ability to distinguish effects of the treatment on neurite outgrowth directly versus neuronal cell death as counted by number of cells. Treatment of N1E cells with SCE did not lead to an increase in cell death, but to a stunted neurite outgrowth with cells portraying a "panning" like phenotype. Co-treatment with TGFb 1 led to a robust increase in neurite outgrowth under these inhibitory conditions. Moreover, the neurite growth inhibitory effects of SCE in the N1E cultures were highly comparable to those previously published in primary cortical or hippocampal neurons, and the TGFb 1-mediated rescue of neurite outgrowth from SCE-induced growth inhibition in our in vitro model of N1E-115 cells was dependent on ALK5 and SMAD3, as was shown to be the case for cortical neurons in vivo (Zhang et al., 2019).
We then went on to pharmacologically block selected downstream targets of TGFb 1/ALK5 signaling to identify potential effects on neurite outgrowth in this model. By applying the candidate factors in a well-defined time frame after initial neurite outgrowth and in combination with inhibitory SCE, we were able to address post-translational signaling mechanisms, including phosphorylation and ubiquitination. While these studies allow us to address the direct effects of pharmacological interventions, we did not address effects on a transcriptional level in this study.
Application of SCE was shown to increase RhoA levels both in the cell body as well as in the growth cone. RhoA activity is associated with neurite retraction in response to growth cone collapsing agents (Feltrin and Pertz, 2012;Jeon et al., 2012). Pharmacological blockade of RhoA or the downstream effector Rho-kinase (ROCK) abolished the inhibitory effects of SCE, specifically also of its active ingredients Nogo-A and MAG (Niederöst et al., 2002). TGFb 1 treatment was associated with a downregulation of RhoA back to baseline levels (control group without SCE treatment), indicating that TGFb 1 signaling might have led to a decreased response to the inhibitory molecules of SCE such as Nogo-A or MAG.
We further saw an accumulation of PAR6 protein in the TGFb 1-treated N1E-115 cells, suggesting increased protein stability or synthesis resulting from the TGFb 1 treatment. Previous studies have described TGFb 1-mediated regulation of proteasomal processes in neuronal cells (Tadlock et al., 2003;Knöferle et al., 2010). Along these lines, we found that the TGFb 1-mediated rescue of neurite outgrowth depended on the E3 ubiquitin ligase SMURF1 as well as PKC/PAR6 interaction. In various cell types, TGFb 1 signaling induces cellular polarity by recruitment and activation of the PAR6/PKC complex to TGFb R2 (Ozdamar et al., 2005). Other studies suggested that this complex relocates SMURF1 to the membrane, where it induces RhoA degradation (Wang et al., 2003;Sahai et al., 2007;Narimatsu et al., 2009) and neurite outgrowth (Bryan et al., 2005;Vohra et al., 2007).
Notably, in our study, the TGFb 1 induced neurite outgrowth was also dependent on PKA activation. Previous studies have shown that PKA is a downstream target of TGFb 1 and is associated with SMAD proteins (Wang et al., 1998;Yang et al., 2013). PKA-mediated phosphorylation of the ubiquitin ligase SMURF1 in the RhoA interaction domain was shown to increase SMURF1 affinity for RhoA relative to PAR6, thereby contributing to axon formation in vitro and neuronal polarization in vivo (Cheng et al., 2011). It is possible that TGFb 1 signaling may also influence additional Rho-like GTPases such as Rac and Cdc42 and effectively regulate neurite outgrowth through additional signaling pathways (Zhang, 2009). Based on the present data, we propose that TGFb 1 enhances axonal outgrowth via stabilization of PAR6 and degradation of RhoA via a PKA/SMURF1-dependent mechanism. These results can guide future experiments to analyze these mechanisms in vivo and in particular in CSNs after a cortical stroke.
Endogenous TGFb 1 levels were shown to be increased in early phases after stroke, within the critical period of plasticity (Doyle et al., 2010). While the present and other studies suggest that prolonged treatment with TGFb 1 or a downstream effector of the TGFb 1/ALK5 signaling pathway may be used to enhance plasticity and extend its critical window after stroke Zhang et al., 2019), a systemic delivery of TGFb 1 would bring about unwanted negative side-effects owing to the plethora of functions of TGFb 1 as a inflammation related molecule (Akhurst and Hata, 2012). It is thus necessary to understand the specific signaling mechanisms by which TGFb 1 mediates structural plasticity and axonal outgrowth, both in vitro and in vivo, to develop a successful therapeutic agent.
The present study investigates a potential signaling mechanism through which TGFb 1 induces neurite outgrowth in a growth inhibitory environment in vitro and possibly compensatory axonal sprouting in vivo. This signaling pathway includes the canonical receptor components ALK5 and SMAD3 and a downstream activation of the cytoskeleton via PKA/SMURF1 induced downregulation of RhoA and upregulation of PAR6.