H2S protects hippocampal neurons against hypoxia-reoxygenation injury by promoting RhoA phosphorylation at Ser188

Inhibition of RhoA-ROCK pathway is involved in the H2S-induced cerebral vasodilatation and H2S-mediated protection on endothelial cells against oxygen-glucose deprivation/reoxygenation injury. However, the inhibitory mechanism of H2S on RhoA-ROCK pathway is still unclear. The aim of this study was to investigate the target and mechanism of H2S in inhibition of RhoA/ROCK. GST-RhoAwild and GST-RhoAS188A proteins were constructed and expressed, and were used for phosphorylation assay in vitro. Recombinant RhoAwild-pEGFP-N1 and RhoAS188A-pEGFP-N1 plasmids were constructed and transfected into primary hippocampal nerve cells (HNCs) to evaluate the neuroprotective mechanism of endothelial H2S by using transwell co-culture system with endothelial cells from cystathionine-γ-lyase knockout (CSE−/−) mice and 3-mercaptopyruvate sulfurtransferase knockout (3-MST−/−) rats, respectively. We found that NaHS, exogenous H2S donor, promoted RhoA phosphorylation at Ser188 in the presence of cGMP-dependent protein kinase 1 (PKG1) in vitro. Besides, both exogenous and endothelial H2S facilitated the RhoA phosphorylation at Ser188 in HNCs, which induced the reduction of RhoA activity and membrane transposition, as well as ROCK2 activity and expression. To further investigate the role of endothelial H2S on RhoA phosphorylation, we detected H2S release from ECs of CSE+/+ and CSE−/− mice, and 3-MST+/+ and 3-MST−/− rats, respectively, and found that H2S produced by ECs in the culture medium is mainly catalyzed by CSE synthase. Moreover, we revealed that both endothelial H2S, mainly catalyzed by CSE, and exogenous H2S protected the HNCs against hypoxia-reoxygenation injury via phosphorylating RhoA at Ser188.


Introduction
Ischemic stroke is one of the main factors responsible for morbidity and death 1 . The restoration of blood flow is a very important way for the treatment of ischemic injury. However, the restoration of blood flow will cause reperfusion injury. Cerebral ischemia-reperfusion (I/R) can cause neuronal damage in varying degrees, including inflammation, necrosis, and apoptosis 2 . Cerebral I/R injury is particularly prone to induce apoptosis of vertebral neurons in the hippocampus area 3,4 . Therefore, the protection of hippocampal nerve cells (HNCs) is extremely important in the treatment of I/R injury. Previous researchers reported that the brain neurovascular unit (NVU) containing endothelial cells (ECs), HNCs, pericytes, astrocytes, and other cells regulates the metabolic homeostasis of the central nervous system and plays an important role in the neuroprotection and repair of the damaged tissue induced by ischemic stroke [5][6][7][8] . It must be pointed out that stabilization of EC function has a potential neuroprotective effect on the cerebral ischemic jury 9 .
Hydrogen sulfide (H 2 S) is a ubiquitous second messenger molecule. Endogenous H 2 S is mainly produced from L-cysteine by cystathionine-γ-lyase (CSE) and cystathionine-β-synthase, and from β-mercaptopyruvic acid catalyzed by 3-mercaptopyruvate sulfurtransferase  in the mitochondria. In vasculature, endogenous H 2 S is mainly produced by CSE and 3-MST [10][11][12][13] . H 2 S has many important functions, including modifying neuronal conduction, regulating vascular tone, protecting tissues from oxidative stress, anti-oxidation, anti-apoptosis, and anti-inflammatory 14 . NaHS, an exogenous H 2 S donor, can also alleviate the neurotoxin acrylonitrile-induced neuronal damage in rats 15,16 and inhibit rat retinal ganglion cell apoptosis 17 . Our previous studies have demonstrated that CSE-produced H 2 S has a protective effect on cerebral I/R injury in mice via inhibition of the RhoA/Rho-kinase (ROCK) pathway 18 .
RhoA, belonging to the small GTPases family, is widely distributed in the cytoplasm and is activated when it combines with GTP, and translates from the cytoplasm to the cell membrane. RhoA controls the activity of copious downstream effectors, such as ROCK 19 . ROCK contains two isoforms ROCK 1 and ROCK 2 . ROCK 2 is regarded as being responsible for the neuronal death and axon degeneration and regeneration after ischemic stroke 20 . Thus, the RhoA/ROCK pathway is closely related to the growth of neurons and axons 21,22 . Besides, emerging evidences indicated that the RhoA-ROCK pathway is involved in the pathological process of the central nervous system diseases. For instance, ischemic brain injury induces RhoA activation and upregulation of ROCK 2 expression 23 , which leads to the neuronal cell death 24 , and inhibition of ROCK protects rats against hippocampal damage induced by cerebral I/R 25 . Although it has been confirmed that increase of H 2 S release has a protective effect on nerve cells against cerebral I/R injury via inhibition of the RhoA/ROCK pathway 18 , the target and mechanism of H 2 S in inhibition of RhoA/ROCK pathway is unknown and there is no research report on the relationship between H 2 S in the cerebral vascular ECs and RhoA-ROCK pathway in nerve cells.
The phosphorylation of RhoA at the Ser188 site can inhibit its activity and mediate its translocation from the cytoplasm to the cell membrane 26 . On the contrary, decreased phosphorylation of RhoA Ser188 in the hippocampal neuronal cells induces RhoA activation 27 . It has been reported that the phosphorylation of RhoA Ser188 mediated by cAMP or cGMP kinase can be regulated by many factors such as ste20-related kinase (SLK), angiotensin II type 2 receptor, and nitric oxide (NO) 28,29 . In view of the fact that H 2 S inhibits the activity of RhoA, we speculate that H 2 S may regulate the phosphorylation of RhoA at Ser188 such as NO, thereby protecting the brain against hypoxia-reoxygenation (H/R) injury.
In the present study, we first constructed and expressed wild-type GST-RhoA wild and mutant GST-RhoA S188A proteins in vitro, and carried out in vitro phosphorylation and autoradiography tests to explore the effect of exogenous H 2 S on RhoA phosphorylation. Next, we constructed and transfected the recombinant eukaryotic plasmids (RhoA wild -pEGFP-N1 and RhoA S188A -pEGFP-N1) into HNCs, to further analyse the effect of exogenous and endogenous H 2 S on the phosphorylation of RhoA at the Ser188, and the associated neuroprotective effects.

Effect of H 2 S on RhoA Ser188 phosphorylation in vitro
To determine the effect of H 2 S on Ser188 phosphorylation of RhoA, GST-RhoA wild and GST-RhoA S188A proteins expressed in Escherichia coli were extracted and purified ( Supplementary Fig. S1A-D, I-K), and were used for in vitro phosphorylation assay.
As shown in Fig. 1A, B, results of autoradiography showed that H 2 S donor NaHS (100 μmol/L) significantly promoted the phosphorylation of GST-RhoA wild protein in the presence of protein kinase 1 (PKG1), but the promoted phosphorylation was significantly attenuated by PKG1 inhibitor DT-2 (2 μmol/L). However, Fig. 1C showed that there was no phosphorylation of GST-RhoA S188A protein. These results indicated that H 2 S could promote phosphorylation of RhoA and Ser188 is required for the H 2 S-mediated RhoA phosphorylation.

Ser188 mediated H 2 S-induced RhoA phosphorylation in rat hippocampal neuron
Effect of H 2 S on the RhoA phosphorylation and role of Ser188 in the RhoA phosphorylation were confirmed in cultured primary rat hippocampal neurons. As shown in Supplementary Fig. S2A-F, the green fluorescence caused by microtubule-associated protein 2 (MAP-2, a specific marker of neuron) antibody was presented in the cytoplasm of the cultured cells, suggesting that the cultured rat hippocampal cells were neurons.
The green fluorescent protein (GFP)-labeled wild-type RhoA (GFP-RhoA wild ) and its mutant (GFP-RhoA S188A ) eukaryotic expression plasmids pEGFP-N1 (Supplementry Fig. S1E-H) were respectively transfected into rat hippocampal neurons by using transient electrotransfection. Results showed that the transfection efficiency was significantly higher at 48 h after transfection than that at 24 h or at 72 h ( Fig. 2A). Therefore, neurons transfected with either GFP-RhoA wild or GFP-RhoA S188A plasmids for 48 h were used for further experiment.
By using western blotting assay, untagged (RhoA), phosphorylated untagged RhoA (p-RhoA), GFP-RhoA wild , phosphorylated GFP-RhoA wild (p-GFP-RhoA wild ), GFP-RhoA S188A , and phosphorylated GFP-RhoA S188A (p-GFP-RhoA S188A ) were detected. As shown in Fig. 2B-D, treatment of NaHS at 100 and 200 µmol/L for 1 h promoted phosphorylation of RhoA in the neurons transfected with empty or GFP-RhoA wild , or GFP-RhoA S188A plasmids, as well as phosphorylation of GFP-RhoA wild in the GFP-RhoA wild plasmid-transfected neurons, p-RhoA/ RhoA ratio, or p-GFP-RhoA wild /GFP-RhoA wild ratio increased markedly compared with those in the control group. However, NaHS had no effect on phosphorylation of GFP-RhoA S188A ; p-GFP-RhoA S188A /GFP-RhoA S188A ratio in the GFP-RhoA S188A plasmid-transfected neuron did not significantly change in comparison with that in the control group. These results suggested that H 2 S could induce RhoA phosphorylation at Ser188 residue in rat hippocampal neurons.

Role of Ser188 in H 2 S-induced RhoA translocation and inactivation in the neurons
Accumulating evidences have shown that when RhoA changes from an active state to an inactive state, it will translocate from the plasma membrane to the cytosol 27 . As shown in Fig. 3A-F, the results of western blotting showed that in primary rat hippocampal neurons transfected with either empty or GFP-RhoA wild , or GFP-RhoA S188A plasmids, treatment of 100 and 200 µmol/L NaHS for 1 h caused significant decreases of the locations of RhoA and GFP-RhoA wild in the plasma membrane, but promoted their location in the cytosol. However, 100 and 200 µmol/L NaHS had no obvious effect on location of GFP-RhoA S188A in the plasma membrane and the cytosol. These results indicated that the Ser188 of RhoA is required for the H 2 S-caused translocation of RhoA from the plasma membrane to cytosol in the neuron. Figure 3G-I showed that 100 and 200 µmol/L NaHS could significantly inhibit the RhoA activity in the neuron transfected with either empty plasmids or GFP-RhoA wild plasmids, but this inhibitory effect was significantly decreased in the GFP-RhoA S188A plasmid-transfected neurons, which was indicated by no inhibition of 100 µmol/L NaHS on RhoA activity. These results suggested that H 2 S could inhibit RhoA activity through Ser188.

Role of RhoA Ser188 in H 2 S-induced inhibition of ROCK 2 protein expression and its activity
As shown in Fig. 4, NaHS 100 and 200 µmol/L significantly inhibited the ROCK 2 protein expression and its activity in rat hippocampal neurons transfected with either empty plasmids or GFP-RhoA wild plasmids. However, in GFP-RhoA S188A plasmid-transfected neurons, 100 µmol/L NaHS did not have this inhibitory effect and only 200 µmol/L NaHS had a significant inhibition. These results indicated that Ser188 of RhoA is associated with the H 2 S-induced inhibition of ROCK 2 protein expression and its activity in the neurons.

RhoA Ser188 mediated in protection of H 2 S on H/R injury in rat hippocampal neurons
As shown in Fig. 5, H/R injury in rat hippocampal neurons transfected with either empty plasmids or GFP-RhoA wild plasmids, or GFP-RhoA S188A plasmids, was indicated by the decreased cell viability and the increased activities of lactate dehydrogenase (LDH) and nerve-specific enolase (NSE) in the cultured supernatant. In the empty plasmid-or GFP-RhoA wild plasmidtransfected neurons, both 100 and 200 μmol/L NaHS significantly inhibited the decreased cell viability and the increased activities of LDH and NSE. In addition, 50 µmol/L NaHS also inhibited the decreased cell viability in the GFP-RhoA wild plasmid-transfected neurons. However, inhibitory effects of NaHS were markedly reduced in GFP-RhoA S188A plasmid-transfected neurons, only up to 200 µmol/L; NaHS significantly inhibited the decreased cell viability and the increased  activities of LDH and NSE. These results indicated that RhoA Ser188 mediated the protective effect of H 2 S on H/R injury.
Effect of endothelial-derived H 2 S on RhoA phosphorylation, activity, and translocation in the neurons By using co-culture of primary cerebrovascular ECs with rat hippocampal neurons, effects of endothelialderived H 2 S on phosphorylation, activity, and translocation of RhoA in the neurons were studied. Due to the fact that CSE and 3-MST are the main enzymes responsible for the synthesis of H 2 S in vascular ECs 10-13 , genetic identification of CSE −/− and CSE +/+ mice, as well as 3-MST +/+ and 3-MST −/− rats was undertaken (Supplementary Fig. S3A, B). Cerebrovascular ECs isolated from CSE wild-type and knockout mice or 3-MST wild-type and knockout rats were identified by factor VIII (Supplementary Fig. S2G-L) and were used in co-culture, and 1 × 10 −6 mol/L acetylcholine (ACh) was used to stimulate H 2 S release from ECs.
As shown in Supplementary Fig. S4A-D, western blotting results showed that p-RhoA/total RhoA ratio and RhoA in the cytosol were significantly reduced, whereas RhoA activity and RhoA location in the plasma membrane were markedly increased in the neurons cocultured with CSE −/− ECs compared with those in the neurons co-cultured with CSE +/+ ECs. However, endothelial 3-MST knockout did not affect p-RhoA/ total RhoA ratio, RhoA activity, and RhoA distribution in the neurons co-cultured with ECs. These results suggested that endothelial-derived H 2 S could induce phosphorylation of RhoA, inhibited its activity, and caused translocation of RhoA from the plasma membrane to the cytosol in rat hippocampal neurons, and perhaps CSE-produced H 2 S exerted a predominant effect. showing the effect of NaHS on membrane fraction of RhoA in HNCs transfected with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids, respectively. E-cadherin was used as a loading control. (D-F) Representative immunoblot images (left) and summary data (right) showing the effect of NaHS on cytosolic fraction of RhoA in HNCs transfected with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids, respectively. β-Actin was used as a loading control. Here, 48 kDa refers to GFP-RhoA; 22 kDa refers to RhoA. Data are shown as the mean ± SEM; n = 3. *P < 0.05, **P < 0.01 or # P < 0.05, ## P < 0.01 vs. the control. (G-I) NaHS inhibits RhoA activity in HNCs transfected with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids, respectively. Data are shown as the mean ± SEM; n = 3. *P < 0.05 vs. the control.
Effect of endothelial-derived H 2 S on ROCK 2 protein expression and activity in the neurons As shown in Supplementary Fig. S5A, B, both ROCK 2 protein expression and activity significantly increased in neurons co-cultured with CSE −/− ECs compared with those in the neurons co-cultured with CSE +/+ ECs. However, endothelial 3-MST knockout in ECs (3-MST −/− ECs) had no effect on ROCK 2 protein expression and activity in the co-cultured neurons. The results indicated that endothelial CSE-produced H 2 S could significantly inhibit ROCK 2 expression and activity in the neurons .
Endothelial CSE-produced H 2 S in co-culture model of ECs and the neuron Supplementary Fig. S5C showed that H 2 S content in the culture supernatant significant lowered in co-culture model of CSE −/− ECs and neurons than that in coculture model of CSE +/+ ECs and neurons. However, 3-MST knockout in ECs had no effect on H 2 S content in the culture supernatant of co-culture model. The results suggested that endothelial-derived H 2 S in the culture supernatant was mainly produced by CSE rather than 3-MST.

Role of Ser188 in CSE-produced H 2 S-induced RhoA phosphorylation, translocation, and ROCK 2 protein expression in rat neurons
In the co-culture of rat hippocampal neurons and CSE +/+ ECs or CSE −/− ECs, the neurons were transfected with empty plasmids or GFP-RhoA wild plasmids, or GFP-RhoA S188A plasmids.
As shown in Fig. 6A, p-RhoA/RhoA ratio in empty or GFP-RhoA wild , or GFP-RhoA S188A plasmid-transfected neurons co-cultured with CSE −/− ECs was markedly lowered than that in neurons co-cultured with CSE +/+ ECs. Similarly, co-culture with CSE +/+ ECs has enhanced effect on the p-GFP-RhoA wild /GFP-RhoA wild ratio in the neurons transfected with GFP-RhoA wild plasmids, but had no effect on the ratio of p-GFP-RhoA S188A /GFP-RhoA S188A in the neurons transfected showing the effect of NaHS on ROCK 2 expression in HNCs transfected with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids, respectively. β-Actin was used as a loading control. (D-F) NaHS inhibits ROCK 2 activity in HNCs transfected with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids, respectively. Data are shown as the mean ± SEM; n = 3. *P < 0.05, **P < 0.01 vs. the control group.
with GFP-RhoA S188A plasmids. The results suggested that Ser188 mediated RhoA phosphorylation in rat hippocampal neurons induced by endothelial CSE-produced H 2 S. Besides, Fig. 6A also showed the weak band of p-GFP-RhoA (48 kDa) from neurons transfected with GFP-RhoA S188A and co-cultured with CSE +/+ ECs, suggesting the H 2 S-induced RhoA phosphorylation in different residue at a low level. Figure 6B-D showed that knockout of CSE in ECs significantly decreased RhoA location in the cytosol, increased RhoA distribution in the membrane, and ROCK 2 protein expression in the co-cultured neurons transfected with empty or GFP-RhoA wild , or GFP-RhoA S188A plasmids. In the co-cultured neurons transfected with GFP-RhoA wild plasmids, the knockout of CSE induced a reduced GFP-RhoA wild in the cytosol and increased GFP-RhoA wild in membrane and ROCK 2 protein expression. However, in the co-cultured neurons transfected with GFP-RhoA S188A plasmids, the knockout of CSE did not affect the distribution of GFP-RhoA S188A in the membrane and cytosol, as well as ROCK 2 protein expression. The results suggested that endothelial CSEproduced H 2 S could induce translocation of RhoA from the membrane to cytosol and inhibition of ROCK 2 protein Effects of NaHS on cell viability of HNCs transfected respectively with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids following H/R. (D-F) Effects of NaHS on lactate dehydrogenase (LDH) release from H/R injury HNCs transfected with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids, respectively. (G-I) Changes of release of nerve-specific enolase (NSE) from H/R injury HNCs transfected with empty, GFP-RhoA wild , and GFP-RhoA S188A plasmids, respectively. Data are shown as the mean ± SEM; n = 6. *P < 0.05, **P < 0.01. expression in rat hippocampal neurons, and these effects are associated with Ser188 of RhoA.

Role of Ser188 in protection of endothelial CSE-produced H 2 S against H/R injury in the neurons
As shown in Fig. 7A-C, in CSE −/− EC-co-cultured rat hippocampal neurons transfected with either empty plasmids or GFP-RhoA wild plasmids, or GFP-RhoA S188A plasmids, 8 h of hypoxia followed by 6 h of reoxygenation induced a significant H/R injury indicated by decreased cell viability and increased activities of LDH and NSE in cultured supernatant compared with those in the sham group. H/R-induced cell injury was markedly decreased in CSE +/+ EC-co-cultured neurons transfected with either empty or GFP-RhoA wild plasmids. However, co-cultured neurons transfected with GFP-RhoA S188A plasmids had no effect on the H/R-induced reduction of cell viability and increases of LDH and NSE activities (Fig. 7D, E). The activity of RhoA and ROCK 2 in the neurons were significantly increased following H/R injury. CSE +/+ ECs significantly decreased RhoA and ROCK 2 activities in the co-cultured neurons transfected with empty or GFP-RhoA wild plasmids, but had no effect in the co-cultured neurons transfected with GFP-RhoA S188A plasmids. The results indicated that endothelial CSE-produced H 2 S could inhibit the activity of RhoA and ROCK 2 in the neurons, and this effect is related to RhoA Ser188.
Apoptosis is also a critical index for H/R injury. Hoechst staining and Annexin V/propidium iodide (PI) staining were respectively used for the detection of apoptotic cells in the present study. Figure 8A-F showed that H/R injury induced significant increases of apoptotic cells and intracellular free Ca 2+ fluorescence intensity in CSE −/− EC-co-cultured neurons transfected with either empty plasmids or GFP-RhoA wild plasmids, or GFP-RhoA S188A plasmids. CSE +/+ ECs could significantly decrease H/R injury-induced increase of apoptotic cells and intracellular free Ca 2+ fluorescence intensity in co-cultured neurons transfected with empty or GFP-RhoA wild plasmids, but had no effect in co-cultured neurons transfected with GFP-RhoA S188A plasmids. These results suggested that endothelial CSE-produced H 2 S could protect the neuron from H/R injury via inhibiting the increase of intracellular free Ca 2+ and the effects were mediated by RhoA Ser188.

Discussion
In view of the condition that the phosphorylation of RhoA requires cAMP-dependent protein kinase A or cGMP-dependent PKG 30 , we used PKG1 as the kinase to catalyze the phosphorylation of RhoA at Ser188 in vitro. Our results showed that RhoA is phosphorylated in the presence of PKG1, which is consistent with the previous results that the gas signal molecule NO can phosphorylate RhoA at Ser188 31,32 . H 2 S is considered to be the third gas signal molecule and its physiological functions are receiving increasing attentions 33 . The pharmacological effects of H 2 S produced by NaHS are now considered reliable 34 . Our in vitro phosphorylation and autoradiography results showed that exogenous H 2 S could also promote the phosphorylation of RhoA at Ser188 mediated by PKG1, which was blocked by PKG1 inhibitors. In addition, the eukaryotic recombinant plasmids containing expressed genes of GFP-RhoA wild and GFP-RhoA S188A were respectively transferred into primary hippocampal neurons, to further confirm the effect of H 2 S. The results showed that NaHS increased the phosphorylated protein of RhoA wild , but not RhoA S188A one in neurons by using western blotting. These findings revealed that H 2 S promoted phosphorylation of RhoA at Ser188; moreover, we also demonstrated that endothelial H 2 S, mainly produced by CSE, regulates RhoA phosphorylation at Ser188.
It is widely recognized that phosphorylation of RhoA at Ser188 directly inhibits RhoA activity 35 , but phosphorylation at Ser88 increases its activity 36 . In the present study, we demonstrated that RhoA activity is significantly inhibited in HNCs after phosphorylation at Ser188, which is consistent with previous results 37 . As we all know, the activated RhoA is mainly distributed in the cell membrane and non-activated RhoA is mainly distributed in the cytoplasm 38 , and NO increases the distribution of RhoA in the cytoplasm and reduces its location in the cell membrane, by promoting the phosphorylation of RhoA at Ser188. Besides, prostaglandin E1 may prevent RhoA translocation to the membrane by promoting the phosphorylation of RhoA at Ser188 on the platelet actin cytoskeleton 31,39 . Furthermore, in vascular smooth muscles, PKG-dependent inactive RhoA accumulates in the cytoplasm 40 . Our results also showed that the distribution of RhoA in neuronal membranes is significantly reduced and its distribution in the cytoplasm is significantly increased after RhoA phosphorylation induced by H 2 S. Previous researches have reported that RhoA activity can not only affect its own translocation but also regulate the activity and expression of its downstream signaling molecule ROCK 2 41-43 . Phosphorylation of RhoA at Ser188 prevents the activation of RhoA and reduces the affinity between RhoA with ROCK 44,45 . In this study, we also found that both ROCK 2 activity and expression were significantly inhibited in primary hippocampal neurons, when RhoA was phosphorylated in Ser188.
The phosphorylation and dephosphorylation of proteins play crucial roles in the regulation of synaptic function under physiological and pathophysiological conditions 46 . For example, phosphorylation of RhoA at Ser188 mediated by protein kinase A prevents it from binding to ROCK and then promotes neuronal differentiation and synaptic proliferation 47 . Furthermore, phosphorylation of RhoA is involved in the protection of ischemic damage of the heart, brain, etc. 48 . Accumulating studies have shown that ischemic brain injury induces a significant decrease of nerve cell viability and increases of LDH and NSE leakages. Hence, similar to cell viability, LDH and NSE leakages have also been applied to assess neuron injury 49,50 . β-Actin was used as a loading control. RhoA expression in membrane fraction (C) and cytosolic fraction (D) of HNCs, which were transfected with empty, GFP-RhoA wild , or GFP-RhoA S188A plasmids, and co-cultured with endothelial cells derived from CSE −/− or CSE +/+ mice. Here, 48 kDa and 22 kDa refer to GFP-RhoA and RhoA, respectively. Data are shown as the mean ± SEM; n = 3. *P < 0.05, **P < 0.01.
Other researchers reported that H 2 S is predicted to have physiological functions in the brain 51 . We found that H/R induces significant increases of LDH and NSE leakages, and reduction of neuronal viability, which could be inhibited by NaHS. However, the inhibitory effects of NaHS were markedly reduced in GFP-RhoA S188A plasmid-transfected neurons. These data indicated that exogenous H 2 S reduces hippocampal neuronal damage induced by H/R, by promoting RhoA phosphorylation at Ser188.
As aforementioned, the homeostasis of the central nervous system is maintained by the brain NVU 52 . Therefore, the co-culture model of ECs and nerve cells was chosen to further demonstrate the role of endogenous H 2 S released from vascular ECs on the phosphorylation of RhoA at Ser188 in HNCs, membrane translocation of RhoA, ROCK 2 expression, and activity. Moreover, the effect of endothelial H 2 S-induced RhoA phosphorylation on the damage of HNCs induced by hypoxia and reoxygenation was assessed. Our data revealed that H 2 S released from ECs could protect nerve cells against hypoxia and reoxygenation injury via promoting the RhoA phosphorylation at Ser188, which was supported by the results that CSE +/+ ECs could protect the neurons transfected with GFP-RhoA wild but not GFP-RhoA S188A against the hypoxia and reoxygenation injury, and inhibit the increase of intraneuronal Ca 2+ in co-culture mode. These findings were consistent with previous research that H 2 S reduces the calcium ions in cells and protects myoblasts from apoptosis and oxidative stress induced by Golgi stress 53 . On the contrary, cerebral I/R-induced calcium overload in nerve cells is the promoter of nerve cell apoptosis 54 . Thus, our results indicated that the reduction of calcium ion concentration in cells may be involved in the RhoA phosphorylation at Ser188-induced neuroprotection of endothelial H 2 S. In a word, we demonstrated for the first time that both endothelial-derived H 2 S and exogenous H 2 S protect nerve cells from hypoxia and reoxygenation damage by promoting RhoA phosphorylation at Ser188.
Although endothelial H 2 S is mainly generated by CSE and 3-MST, our results showed that ACh-stimulated release of H 2 S from vascular ECs is mainly from CSE. The reason of CSE-and 3-MST-caused different amounts of H 2 S release is worthy of further exploration.
In summary, we demonstrated that exogenous and endothelial H 2 S promote RhoA phosphorylation at Ser188 for the first time. Our results also indicated that the RhoA phosphorylation at Ser188 contributes to inhibition of RhoA activity and membrane transposition, and reduction of ROCK 2 activity and expression. Furthermore, we found that exogenous and endothelial H 2 S mitigates H/R injury of neurons via accelerating RhoA phosphorylation at Ser188. These findings provide potential targets for treating or preventing cerebral I/R injury.

Fig. 7 H 2 S produced by CSE in endothelial cells mitigates H/R injury and regulates the activity of RhoA and ROCK 2 in HNCs. (A-C)
Effects of CSE-produced H 2 S from endothelial cells on cell viability, release of LDH, and NSE from HNCs transfected with empty, GFP-RhoA wild , or GFP-RhoA S188A plasmids under H/R conditions. Effects of endothelial CSE-produced H 2 S on RhoA activity (D) and ROCK 2 activity (E) in HNCs transfected with empty, GFP-RhoA wild , or GFP-RhoA S188A plasmids under H/R conditions. Data are shown as the mean ± SEM; n = 3. *P < 0.05, **P < 0.01.

Animal
Wild-type (CSE +/+ ) and CSE-knockout (CSE −/− ) C57BL/6J mice, and wild-type (3-MST +/+ ) and 3-MSTknockout (3-MST −/− ) Sprague Dawley (SD) rats were obtained from Shanghai Biological Model Biotechnology, Co., Ltd. The animals were housed in the Animal Center of Anhui Medical University, with free access to food and water, and controlled temperature at 22 ± 2°C. All the efforts were undertaken to minimize the pain or discomfort of the animals; all experimental procedures were carried out in accordance with the operating procedures approved by the Ethical Review Committee of Anhui Medical University, which comply with the guidelines for the care and use of laboratory animals of the National Institutes of Health (NIH publication number 85-23, revised in 2011).
Expression the recombinant RhoA and RhoA S188A The glutathione-S-transferase (GST)-tag RhoA wild -pGEX-6p1 and RhoA S188A -pGEX-6p1 plasmids were constructed by Gene Create Biological Engineering, Co. (Wuhan, China) from rat genome. RhoA S188A is that Ser188 of RhoA was mutated to an Ala at the mRNA level. The plasmids were expressed in E. coli and purified as previously described 26 . Two prokaryotic expression plasmids were transferred into BL-21 E. coli, respectively, and then the expressions of GST-RhoA wild and GST-RhoA S188A proteins were induced by isopropylthio-β-D-galactoside. The collected protein was sonicated and purified by GST-Sepharose 4B affinity chromatography (GST-Sepharose FF). We used the bicinchoninic acid method to quantitatively analyze the purified protein. The eluted GST-RhoA wild and GST-RhoA S188A proteins were used for in vitro phosphorylation assay.

In vitro phosphorylation assay
In vitro phosphorylation assay was performed as previously described 26 . Briefly, the kinase buffer containing 50 mM Tris, 10 mM MgCl 2 , 1 mM dithiothreitol, and 100 μM ATP was prepared. Then, 20 μg of RhoA wild or RhoA S188A , 10 μCiof [γ-32 P]-ATP, 100 ng kinase and 100 μM NaHS were sequentially added to the kinase buffer and the mixture was then incubated in a shaker at 30°C for 30 min. After that, the mixture was separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and stained by using Coomassie Brilliant Blue. The separated proteins were incubated with X-ray film for 72 h and displayed by autoradiography.

EC culture
Primary rat brain vascular ECs were respectively isolated from 6 to 8 weeks old CSE −/− and CSE +/+ mice, and 3-MST −/− and 3-MST +/+ rats, as previously described 55 . In short, rats were anesthetized by isoflurane. The intact brains were carefully isolated after cervical dislocation and placed in a culture dish with sterile phosphate buffered saline (PBS). The rat basilar artery was carefully separated and transferred to eppendorf tubes containing 0.1% type I collagenase and were quickly cut into pieces; vascular fragments were digested for 30 min at 37°C. The mixture was centrifuged at 1000 r.p.m. for 10 min to obtain the cell pellet, which was resuspended in ECM medium (5% fetal bovine serum + 1% EC growth supplement + 1% penicillin-streptomycin). According to the principle of differential adhesion, the medium was changed after 1 h to obtain vascular ECs. The cells were identified by immunofluorescence staining for VIII.

Neuronal culture
Primary rat hippocampal neurons were isolated from newborn SD rats. Briefly, newborn rats were soaked in (see figure on previous page) Fig. 8 Role of Ser188 in the protection of endothelial CSE-produced H 2 S against H/R injury in the neurons. Hoechst 33258 staining, 400 μm; Fluo-8 staining, 200 μm. (A) Hoechst staining of apoptotic nucleus in HNCs transfected with empty, GFP-RhoA wild , or GFP-RhoA S188A plasmids under hypoxia-reoxygenation (H/R) condition. Representative immunoblot images (C) and summary data (B, D) of cell apoptosis in HNCs transfected with empty, GFP-RhoA wild , or GFP-RhoA S188A plasmids under H/R condition. Representative immunoblot images (E) and summary data (F) of calcium concentration in HNCs transfected with empty, GFP-RhoA wild , or GFP-RhoA S188A plasmids under H/R condition. Data are shown as the mean ± SEM; n = 3. *P < 0.05, **P < 0.01. 75% alcohol for surface disinfection under anesthesia, the brain was completely isolated after killing, and then the hippocampus tissue was separated and cut into pieces in 0.125% trypsin solution. After digestion for 20 min at 37°C with being gently shaken every 5 min, the equal volume of complete medium was added to stop the digestion. Cells were collected after filtration and centrifugation, and Dulbecco's modified Eagle complete medium was added to resuspend the culture cells in a sixwell plate precoated with polylysine. The cells were incubated at 37°C in 5% CO 2 incubator. Twenty-four hours later, the medium was replaced with neurobasal serum-free medium (neurobasal + 2% B27 + 0.5 mmol/L L-glutamine + 1% penicillin-streptomycin); 48 h later, 20 μM cytarabine was used to inhibit the growth of glial cells. Immunofluorescence staining was performed to identify HNCs with neuronal MAP-2.
Electroporation of eukaryotic expression plasmids of recombinant RhoA wild and RhoA S188A into cultured neurons The GFP-tag eukaryotic expression plasmids containing RhoA wild -and RhoA S188A -expressed genes were constructed by Gene Create Biological Engineering, Co. (Wuhan, China) from rat genome. The two recombinant plasmids were designated as RhoA wild -pEGFP-N1 and RhoA S188A -pEGFP-N1, respectively. According to previous research 56 ,~5 × 10 6 HNCs were mixed with 5 μg RhoA wild -pEGFP-N1 or RhoA S188A -pEGFP-N1 and 70~80 μL Opti-MEM. The mixed solution was transferred to the electroporation cuvette and electroporation was performed by using CUY21EDIT II super multi-pulse in vivo/cell electroporation instrument (BEX, Tokyo, Japan) with program of perforation pulse voltage 350 V, pulse drive voltage 20 V, pulse time 10 ms, and cycle number 20 times. Fluorescence microscope was used to detect transfection efficiency.

H/R protocol
HNCs were cultured with sugar-free medium in a threegas incubator (1% oxygen, 5% nitrogen, 94% carbon dioxide, 37°C) for 8 h, then the cells were replaced with normal culture medium and transferred to the incubator under normal condition for 6 h. The control group cells are maintained under normal condition.

Determination of RhoA and ROCK 2 activities
Determination of RhoA and ROCK 2 activities was performed according to the manufacturer's instructions and our previous study 18 . Briefly, the lysate of HNCs was collected after cell sonication and centrifugation to measure RhoA and ROCK 2 activities by G-LISA activation assay. Optical density (OD) value of the sample at 450 nm was recorded by using colorimetric assay.

Determination of cell viability and LDH
Cell viability and LDH activity in cell culture supernatant were determined according to the manufacturer's instructions. Briefly, cultured cells were incubated with 10 μl cell counting kit-8 solution for 1 h. Cell activity was detected at 450 nm by a microplate reader. The cell culture medium was collected to detect the LDH activity by using spectrophotometry at 450 nm and the result was expressed as U/L.

Determination of NSE activity
NSE activity in cell-cultured supernatant was assessed according to the manufacturer's instructions. Supernatant of cell culture was collected and transferred to 96-well plates to detect the NSE activity by using rat NSE enzymelinked immunoassay kit. The OD value at 450 nm was recorded by a microplate reader and the results were expressed as μg/L.

Flow cytometry assay for the cell apoptosis
Annexin V-Alexa Fluor 647/PI cell apoptosis detection kit was used to detect cell apoptosis in compliance with the manufacturer's instructions. The cells were collected and washed with PBS, and then resuspended in binding buffer. After that, cells were incubated in PI and fluorescein isothiocyanate-Annexin V for 30 min. The percentage of apoptotic cells was analyzed on flow cytometry. Each test was repeated three times.

Measurement of H 2 S concentration
H 2 S concentration in cell culture medium of cultured ECs was measured according to our previous report 55 . The cells were pretreated with ACh for 30 min at 37°C. The cell culture medium was collected and incubated with 1% zinc acetate, 10% trichloroacetic acid, 20 µM N,N-dimethyl-pphenylenediamine sulfate, and 30 µM FeCl 3 for 25 min. The absorbance of the mixed solution was detected at 665 nm by a microplate reader. The H 2 S concentration was quantified according to the NaHS standard curve, which represents the amount of H 2 S released by ECs.

Co-culture of cerebrovascular ECs with HNCs
Cell co-culture technique were in compliance with previous research with some modification 9 . Transwell coculture system (Millicell small suspension cell culture insert, 0.4 µm polyester membrane, 6 wells; Corning, USA) was used to co-culture cerebrovascular ECs with HNCs for 24 h. The cells were pretreated with NaHS or ACh for 1 h before hypoxia to stimulate H 2 S release.

Western blotting
The cells were lysated in radioimmunoprecipitation assay buffer, which consisted of Tris-HCl 50 mM, NaCl 150 mM, 1% sodium deoxycholate, 0.1% SDS, sodium fluoride, 1% Trition X-100, and EDTA. The total proteins were separated by 12% SDS-PAGE and then transferred to polyvinylidene difluoride membranes. The bands were visualized by using an enhanced chemiluminescence kit. Membrane protein and cytoplasmic protein extraction are performed according to the instructions in the kit. Briefly, the homogenate is centrifuged at a low speed to remove unbroken cells and nuclei, and the supernatant obtained by high-speed centrifugation contains cytoplasmic proteins. After the precipitate is dissolved, membrane proteins are obtained.

Statistical analysis
All data in this study are expressed as mean ± SEM. The data analysis was carried out by a blinded investigator, who did not know which group the samples came from. Statistical analyses were performed to identify the normal distribution and homogeneity of variance of data by oneway analysis of variance, followed by Student's t-tests to determine differences between groups. P-values < 0.05 were considered to be statistically significant.