Combined treatment of sodium ferulate, n‐butylidenephthalide, and ADSCs rehabilitates neurovascular unit in rats after photothrombotic stroke

Abstract The remodelling of structural and functional neurovascular unit (NVU) becomes a central therapeutic strategy after cerebral ischaemic stroke. In the present study, we investigated the effect of combined therapy of sodium ferulate (SF), n‐butylidenephthalide (BP) and adipose‐derived stromal cells (ADSCs) to ameliorate the injured NVU in the photochemically induced thrombotic stroke in rats. After solely or combined treatment, the neovascularization, activation of astrocytes, neurogenesis, expressions of vascular endothelial growth factor (VEGF) and claudin‐5 were assessed by immunohistochemical or immunofluorescence staining. In order to uncover the underlying mechanism of therapeutic effect, signalling of protein kinase B/mammalian target of rapamycin (AKT/mTOR), extracellular signal‐regulated kinase 1/2 (ERK1/2), and Notch1 in infarct zone were analysed by western blot. 18F‐2‐deoxy‐glucose/positron emission tomography, magnetic resonance imaging, Evans blue staining were employed to evaluate the glucose metabolism, cerebral blood flow (CBF), and brain‐blood barrier (BBB) permeability, respectively. The results showed that combined treatment increased the neovascularization, neurogenesis, and VEGF secretion, modulated the astrocyte activation, enhanced the regional CBF, and glucose metabolism, as well as reduced BBB permeability and promoted claudin‐5 expression, indicating the restoration of structure and function of NVU. The activation of ERK1/2 and Notch1 pathways and inhibition of AKT/mTOR pathway might be involved in the therapeutic mechanism. In summary, we have demonstrated that combined ADSCs with SF and BP, targeting the NVU remodelling, is a potential treatment for ischaemic stroke. These results may provide valuable information for developing future combined cellular and pharmacological therapeutic strategy for ischaemic stroke.


| INTRODUCTION
Although neuroprotective agents are demonstrated to be effective in functional recovery of stroke in animal model, most clinical trials showed uniformly negative results. 1 It suggests that targeting only neuronal cells is not enough for stroke therapy and neuron-astrocytecapillary interaction should be taken into consideration. 2 In 2002, Report of the Stroke Progress Review Group updated the concept of neurovascular unit (NVU) which emphasized the complexity of interactions between neurons, vascular cells and glia in the brain. Since then, protection and regeneration of NVU has been a new focus for stroke treatment. 3 Studies of NVU describe that upon brain injury, astrocytes regulate the neurovascular coupling by promoting angiogenesis and improve the regional cerebral blood flow (rCBF). In addition, the newly formed vessels secrete neurovascular trophic factors and chemokines in the microenvironment of injured brain that induces the migration of adjacent neural stem cells and subsequent integration in the parenchyma of ischaemic region. Coordination of neovasculature and astrocytes also regulate the synaptogenesis and axonal sprouting. 4,5 Therefore, NVU is now the important therapeutic target for postischaemic stroke treatment and management.
In last two decades, numerous reports have shown that mesenchymal stem cells (MSCs) are a promising therapeutic regimen of human diseases including cerebral ischaemia. Administration of adipose-derived stromal cells (ADSCs), a type of MSCs, decreased the infarct volume, inflammation, cell apoptosis, and improved the poststroke sensorimotor dysfunction. Furthermore, stereotactically injected ADSCs were found to differentiate into neurons which replaced necrotic cells, and increased neuronal survival. 6,7 In a study of comparing the therapeutic effect of MSCs from adipose tissue and bone marrow (BM) on ischaemic stroke, ADSCs had distinctly higher proliferation, differentiation and secretory ability, and their efficacies of reducing the infarct size and brain oedema, improving the motor function were also superior to BM-derived stem cells. 8 In recent years, ADSCs have received great attention because of their nonethical controversy compared to embryonic stem cells, and minimally invasive collection procedure compared to BM-MSCs. These advantages have made ADSCs a more appropriate and important source of MSCs in the development of cell therapy for human diseases including stroke.
Combined pharmaceutical treatment with MSCs for ischaemic stroke is a potential and feasible approach. More effective and synergistic therapeutic outcome for ischaemic stroke by combined MSCs and pharmaceutical treatment through promotion of stem cell migration and survival, anti-apoptosis, endogenous stem cell proliferation, neurotrophic factor secretion, and angiogenesis have been observed. 9 Among the drugs used in the combined treatment with stem cells, several Chinese medicines and their active compounds showed remarkable therapeutic results, such as to induce differentiation of MSCs into neural-like cells, maintain the pluripotency of embryonic stem cells and enhance the efficacy of induced pluripotent stem cell generation. [10][11][12] Our previous study also showed that sodium ferulate (SF) and n-butylidenephthalide (BP), two active components extracted from Radix Angelica sinensis, were able to enhance angiogenesis, neurogenesis and reduce infarction volume in the rat middle cerebral artery occlusion (MCAo) model when used in combination with MSCs. 13 However, the therapeutic effect of combined treatment with ADSCs, SF, and BP on the photothrombotic stroke (PTS) model is still unknown. In this study, with the aid of multimodal imaging techniques (eg, magnetic resonance imaging, 18 F-2-deoxyglucose positron emission tomography and fluorescent optical imaging), we sought to know the efficacy of the combined therapy on restoring the structure and function of NVU in infarct region after stroke.

| ADSCs isolation and identification
ADSCs were isolated from adipose tissue of six-week-old C57BL/6L mice. Briefly, adipose tissue was carefully excised from abdominal cavity and immediately digested by 0.1% collagenase type IV mixed in alphaminimum essential medium (αMEM) containing 10% foetal bovine serum (FBS, Hyclone) at 37°C for 1 hour. The digested tissue was filtered using a 45 μm nylon filter mesh (BD Falcon) and centrifuged at 1500 g for 10 minutes. After removal of the supernatant, the pellet was resuspended in αMEM supplemented with 20% FBS and seeded in a 6cm tissue culture dish. The culture medium was refreshed every 2 days, and cells were passaged at 80%-90% confluence.
To identify the isolation of a true ADSC population, we evaluate the adipogenic and osteogenic differentiation potential of ADSCs.

| Introduction of luciferase gene in ADSCs
Codon-optimized reporter gene encoding firefly luciferase (Luc) was cloned in a lentiviral expressing vector. Lentiviral particles were ZHAO ET AL. | 127 produced by cotransfection with plasmid pRSV-Rev, pMDLg-PRRE, and pHEF-VSVG into 293T packaging cells. 24 hours later, the medium was replaced by fresh medium containing 1% bovine serum albumin (BSA). Virus-containing medium was collected 24 hours later and filtered with 22 μm filter. ADSCs were infected with lentiviral particles followed by puromycin selection (2 mg/mL) to establish stably transduced cells. The Luc expression in ADSCs was evaluated by incubating the colonies with luciferin solution and imaged with IVIS 50 (Perkin Elma, UK). Image parameter was set as following: Exposure Time = Auto, Binning = medium, f/stop = 1, FOV = 12. Quantification was performed using living image software 3.2 (IVIS Imaging System, Perkin Elma, UK).

| Photochemically induced stroke model
The details of experimental method were described in our previous study. 14 Briefly, 7-week old male Sprague-Dawley (SD) rats under anesthetization was illuminated by a laser beam with 532 nm wavelength at the middle of the craniotomic window which was made over the somatosensory cortex with the center at the coordinate of 1 mm rostrally from the bregma and 3.5 mm lateral to the midline for 30 minutes upon slow injection of rose bengal (20 mg/kg body weight) through tail vein. All rats after induced stroke were able to survive until they were killed at the end-point in this study. Before and after thrombosis was induced by photochemical method, a moorFLPI-2 Full-Field Laser Perfusion Imager (Moor Instruments, Axminster, UK) was placed at the center of cranial window where the laser beam illuminated in order to examine the blockage of rCBF owing to cerebral vascular embolization. The laser speckle images were acquired with 25-Hz sampling frequency, 1 frame/s, 580 × 752 pixels resolution, and zoom size of 5.6 mm × 7.5 mm.

| Experimental groups and therapeutic interventions
Rats with induced photochemical stroke were randomly divided into four groups: PTS, SF + BP, ADSC, and ADSC + SF + BP. Four hours after stroke induction, 20 μL phosphate buffer saline (PBS) was injected into the margin of laser illuminated area in rats of group PTS and SF + BP. In the rats of group ADSC and ADSC + SF + BP, 5 × 10 5 ADSCs (in 20 μL PBS) were injected into the same place.
After stem cells or PBS injection, SF (60 mg/kg) was daily administrated in rats of group SF + BP and ADSC + SF + BP for consecutive 14 days via intraperitoneal injection. BP (10 mg/kg) was subcutaneously injected in the same groups once a day for 3 days.
For control treatment in rats of group PTS and ADSC, PBS was injected following the time-points of SF and BP delivery.

| Western blot
Protein samples (n = 3) were, respectively, extracted from the infarct tissue at day 7 and ipsilateral hemispheres at day 14 after stroke induction. Western blot was performed according to previous protocol. 13  secondary antibodies (1:5000, LI-COR Biosciences) at room temperature for 1 hour. Chemiluminescent signal was imaged and quantified using the Odyssey Infrared Imager (LI-COR).

| Perfusion-weighted imaging by MRI
Perfusion-weighted imaging (PWI) was performed with the procedure described in our previous report. 14 BioSpec-70/30 7T system (Bruker, Ettlingen, Germany) with a birdcage head-coil of 75 mm inner diameter for radio frequency (RF) transmission and a 20 mm diameter surface coil for reception were used. Prior to imaging, rats (n = 6) under anaesthesia (performed by initial inhalation of 4% isoflurane for 3 minutes and maintained with 2% isoflurane in a mixture of 20% oxygen and 80% air) were placed in the stereotaxic holder of MRI machine equipped with a heating system to maintain body temperature and a pressure detector to monitor respiration.

| Evans blue dyeing and IVIS detection
Disruption of blood-brain barrier (BBB) after photothrombotic stroke was evaluated by in vivo Evans blue (EB) fluorescent imaging. 14 Briefly, 2 hours after EB dye (Sigma, 10 mg/mL in saline, 2.5 mL/kg rat weight) was injected via tail vein, animals in each group (n = 6) at day 3 and 7 were killled with an overdose of pentobarbital injection (RMB, Animal Health Ltd., UK). The brains were carefully removed from the skulls immediately after death. To detect the presence of EB, the intact brain was imaged using IVIS 50 (PerkinElmer, UK) with following steps and image acquisition settings: the brains were placed at the center of imaging field, and images were acquired for 2 seconds using Cy5.5 band pass filter channel (excitation/emission wavelength: 615~665 nm/695~770 nm). ROI selection and quantification were performed using living image software 3.2 (IVIS Imaging System, Perkin Elma, UK).

| Statistical analysis
All results were expressed as means ± SD. One-way ANOVA was For tracking of ADSCs after injection into animals, Luc gene was introduced into ADSCs by lentiviral transduction followed by stable cell selection. Expression of Luc in ADSCs was imaged using an IVIS system. As shown in Figure 1D, expression of Luc in ADSCs was observed thru detecting the luminescence upon addition of luciferin.
The intensity of luminescent signal in cells positively correlated with cell number (R 2 = 0.9977).

| ADSC transplantation and tracking ex vivo
For ADSC treatment, cells were intracerebrally injected at the margin of ischaemic region after PTS. At day 7 and 14 after transplantation, Luc expression of ADSCs was detected in the brain of rats in group ADSC and ADSC + SF + BP ( Figure 3A). Notably, the expression of Luc in group ADSC + SF + BP was higher than in group ADSC at both day 7 and 14, and also was significantly higher at day 14 than that at day 7 in the same group ( Figure 3B, P < 0.01). This observation suggested that the microenvironment where the ADSCs were injected in the group treated with SF + BP might favour the survival of ADSCs, and even promote their proliferation.

| Combined treatment of ADSCs, SF, and BF promote neovascularization in infarct lesion
After treatment, the neovascularization was evaluated by detecting the expression of vWF and α-SMA. As shown in Figure 4A Figure 4D). These results indicated that combined treatment of ADSC + SF + BP had better effect on vascularization including capillary/vascular density and diameter, as compared to treatment with ADSCs or SF + BP alone.
The expression of angiogenic factor VEGF was also evaluated, and the result showed that more cells with positive staining (brown) were presented in group ADSC and ADSC + SF + BP than in group SF + BP and PTS ( Figure 4E). Quantitative result indicated that the expression of VEGF in ADSC group were 6-fold and threefold higher than that in group PTS (P < 0.01) and SF + BP group (P < 0.01), respectively. The expression of VEGF in group ADSC + SF + BP were 2.3-fold higher than that in group ADSC (P < 0.01, Figure 4F).
These data showed that combined treatment of ADSCs and SF + BP effectively enhanced the expression of VEGF.

| Combined treatment of ADSCs, SF, and BF modulate the activation of astrocytes and promote neurogenesis
To know what the effect of ADSCs or SF + BP on astrocytes upon treatment of stroke, the expression of GFAP was detected by immunostaining. In group PTS, only few scattered GFAP + cells were observed locating at the peri-infarct zone, and even less number of cells were found in group SF + BP ( Figure 5A). In contrast, obvious high number of GFAP + cells was noted in group ADSC. Interestingly, in group ADSC + SF + BP, these GFAP + cells mostly resided in the infarct region, instead of in the peri-infarct region. Quantitative analysis showed that the number of GFAP + cells in peri-infarction zone in ADSC group was 5.6-fold higher than in group PTS (P < 0.01), but was 1.7-fold lower in group SF + BP group than in group PTS (P < 0.05, Figure 5B). These results suggested an inhibitory action on astrocyte activation by SF + BP treatment and an activation effect by ADSCs. Combined treatment by ADSC + SF + BP promoted the migration of active astrocytes into infarct region.
In order to assess whether combined treatment promoted the neurogenesis, immunohistochemical staining for Tuj1 and western blotting of DCX were performed. As shown in Figure 5A, 85.53 ± 7.51/mm 2 , P < 0.01, Figure 5C). Number of Tuj1 + cells in group ADSC + SF + BP was 1.4-fold higher than group ADSC and 2.7-fold higher than group SF + BP group (P < 0.01). Quantitative analysis of western blot showed that the expression of DCX in group ADSC + SF + BP at day 14 was the highest compared with those in other three groups ( Figure 5D). The results suggested that there was a remarkable promotion of neurogenesis by combined treatment of ADSC + SF + BP.

| The improvements of cerebral blood flow and glucose metabolism
To monitor the evolutional CBF change in photothrombotic stroke, PWI was performed on rats at day 1, 3, 7, and 14 after stroke induction ( Figure 6A). Quantitative analysis indicated that the rCBF ratio at the infarct region in group ADSC and ADSC + SF + BP increased rapidly at day 3, with the value 2.4-fold and 3.3-fold higher than that in group PTS, respectively. At day 14, the ratios in both group increased and were 2.9-fold and 3.9-fold higher than that in group SF + BP (P < 0.01). Furthermore, at day 3 and day 14, rCBF ratio in group ADSC + SF + BP was higher than that in group ADSC (1.4fold at day 3, and 1.4 -fold at day 14, P < 0.01, Figure 6B). This result strongly suggested that combined treatment of ADSC and SF + BP significantly restored CBF in the infarct region after photothrombotic stroke. 18 F-FDG PET imaging was performed for the investigation of metabolic change in each group at day 1, 3, 7, and 14 after stroke induction. As shown in Figure 6C, 18  DUR in each group at day 14 were 2.0-fold, 2.7-fold, 3.1-fold and 5.1-fold lower as compared with that at day 1 (P < 0.05, Figure 6D).
Although the improvement of glucose metabolism in the infarct region was noted over time in each group, DUR in group ADSC + SF + BP was significantly improved than that in PTS group at day 7 (P = 0.034) and 14 (P = 0.038), showing that the combined treatment of ADSC + SF + BP effectively improved the glucose metabolism in infarct region.

| Combined treatment of ADSCs, SF, and BF recovered BBB disruption
Our pervious study indicated that the leakage of BBB gradually increased at day 3 and began to restore at day 7 in rat PTS model. 14 In this study, we sought to evaluate the restoration of BBB disruption after treatment. As shown in Figure 7A, Evans blue leakage was renovated, with greatest amelioration seen in group SF + BP. Treatment of ADSCs resulted in the highest BBB leakage even compared to group PTS at day 3. Quantitative analysis indicated that the fluorescence signal in group SF + BP was 1.8-fold lower than that in group PTS at day 3 (P < 0.01), and signal in group ADSC was 1.3fold higher as compared with group PTS at day 3 (P < 0.01), suggesting that SF + BP treatment repaired the BBB leakage. At day 7, signal in group ADSC was still higher than in group SF + BP and ADSC + SF + BP, although there was no statistical difference (P > 0.05). However, at day 3 or 7, fluorescence signal in group ADSC + SF + BP was significantly lower than that in group ADSC group ( Figure 7B). Additionally, as shown in Figure 7C Quantitative analysis showed that ADSC + SF + BP treatment restored rCBF more rapidly B, and enhanced glucose metabolism more efficiently (C) as compared with other three groups. Data were expressed as means ± SD. *P < 0.05, **P < 0.01, compared with group PTS; ## P < 0.01, compared with SF + BP group, && P < 0.01, compared with ADSC group, £ P < 0.05, ££ P < 0.01, compared with day 1, $ P < 0.05, $$ P < 0.01, compared with day 3. N = 6 ZHAO ET AL. F I G U R E 7 Evaluational change of blood-brain-barrier integrity poststroke. A, Representative images of EB staining and fluorescence imaging of rat brain after stroke in different treatment groups at day 3 and 7. B, Quantitative result of EB fluorescence showed that SF + BP treatment remarkably ameliorated BBB leakage while ADSC treatment had adverse effect at day 3. However, combined treatment of ADSC + SF + BP effectively ameliorated BBB leakage at day 3-7. (C) and (D) Representative images of claudin-5 by immunofluorescence staining were presented and quantitative analysis showed the expression of claudin-5 in SF + BP group was the most significant. Data were expressed as means ± SD. *P < 0.05, **P < 0.01, compared with PTS group, ## P < 0.01, compared with SF + BP group; && P < 0.01, compared with ADSC group; $$ P < 0.01, compared with day 3. N = 6. Scale bar: 100 μm 3.8 | Regulations of AKT/mTOR, ERK1/2 and

Notch1 signalling pathways
In order to explore the corresponding mechanism associated with NVU remodelling after combined treatment, AKT/mTOR, ERK1/2 and Notch1 pathway were analysed in the infarct region by western blot.
As shown in Figure 8, phosphorylated AKT and mTOR in group It is now well accepted that upon acute ischaemic stroke, a rapid NVU injury is triggered. To restore NVU, reestablishment of vasculature in the infarct region is the major task upon treatment. A rebuilt vessel network will ensure the rCBF recovery, delivery of oxygen and nutrients, migration of stem/progenitor cells as well as removal of necrotic debris. 12 28 In our study, combination treatment did not exasperate BBB integrity, despite that VEGF expression was the highest among all the treatment groups.
Recent report indicated that VEGF was able to bind by astrocytederived Pentraxin 3, and subsequently decrease VEGF-induced endothelial permeability in vitro. 29 Some preclinical studies showed that BBB disruption poststroke did not totally attribute to tight junction disassembly, at least functional endothelial alterations and endothelial damage involved in the injury process. 30,31 Reeson et al 32 demonstrated that VEGF-associated BBB breakdown after stroke may be attributed to endothelial transcytosis rather than tight junctions in diabetic mice, but in nondiabetic mice, VEGF receptor 2 inhibitor increases BBB leakage. The studies reflect the multifaceted characteristics of VEGF on BBB integrity, so the precise mechanisms need to be further elucidated in the future.
Signalling of AKT/mTOR, ERK1/2 and Notch1 was reported to be at least partly involved in the BBB leakage, angiogenesis, neurogenesis, and reactive astrocytes upon ischaemia stroke. 13,33,34 By suppression of Akt phosphorylation using electroacupuncture pretreatment in ischaemic cortices, BBB permeability together with brain oedema were able to be effectively reduced. 35 Also, inhibition of mTOR by rapamycin increased the expession of tight junction protein zonula occludens-1 in brain microvascular endothelial cells and attenuated EB extravasation in rat ischaemic hemisphere. 36 In the present study, the inactivation of AKT/mTOR in the infarcted region by combination treatment was correlated with attenuation of BBB permeability. Notably, in our results, increased VEGF expression and angio-neurogenesis after combined ADSC + SF + BP treatment may be associated with the activation of the ERK1/2 and Notch signalling. It was reported that activation of ERK1/2 cascade not only protected cortical neurons upon ischaemic injury, but also enhanced VEGF expression and angiogenic effect poststroke. 37,38 Activation of Notch signalling in human umbilical cord MSCs after cocultured with oxygen glucose deprivation-induced neurons increased VEGF secretion and the ability of capillary-like tube formation, as well as contributed to pericyte attachment and endothelial cells survival 39,40 ; simultaneously, activating signalling of Notch1 facilitated the subventricular zone neurogenesis after focal ischaemia. 41 Furthermore, our study showed that SF + BP treatment significantly suppressed the reactive astrocytes, which correlated with its down-regulated Notch1 signalling resulted in reduction in reactive astrocyte proliferation after ischaemic stroke. 42 In summary, this is the first study demonstrating that combined treatment of SF + BP and ADSCs was able to effectively ameliorate the structure and function of the NVU in the infarct zone induced by photochemical stroke, and the therapeutic mechanism might associate with the regulations of AKT/mTOR, ERK1/2, and Notch1 pathways. These results may provide valuable information for developing future combined cellular and pharmacological therapeutic strategy for ischaemic stroke, especially targeting the NVU remodelling.