Statistical analysis

Statistical analyses were carried out using Sigmaplot (Systat Software Inc, San Jose, CA, USA). Data were analyzed by Student’s t test or one-way analysis of variance (ANOVA). Significant differences were further evaluated using Tukey’s honest significant difference post-hoc test. A p-value ˂ 0.05 was considered statistically significant. All data are expressed as mean ± S.E.

RESULTS

Part A. Chronic stroke animal models and therapeutic effects of MSC/Ngn1 cells in chronic stroke.

1. Characterization of chronic stroke animal models.

To generate a chronic stroke model of rat and mice, the intraluminal transient middle cerebral artery occlusion method was carried out in rats and mice (Fig 1A). Twenty four hours after reperfusion, TTC staining showed that both rat and mouse stroke models developed clear infarction (Non-stained area) in the ipsilateral striatum and dorsolateral cortex verifying proper occlusion of the Middle cerebral artery (Fig 1B) We also assessed the structural integrity of the ischemic brain over the same 28-day period using MRI. The hypointense areas in the ipsilateral hemisphere were evident at the acute phase of both rat and mouse models. In the chronic phase, rat MCAo showed an infarct cavity as shown by hypointense areas in the ipsilateral hemisphere (Fig 2A). However, In a mouse model of chronic stroke, MRI did not show any hypointense areas suggesting that the infarct cavity was not formed in a chronic stroke mouse model (Fig 2B). Sham-operated animals did not show any hypointense areas in the brain. Further, we evaluated the tissue integrity of chronic stroke animal models by Hematoxylin & Eosin and Cresyl-violet staining at 1 month after ischemia-reperfusion injury and compared with the histology of acute stroke i.e.,

3 days after reperfusion and sham-operated animals (Fig 2 C, D). Hematoxylin

& Eosin and Cresyl violet staining of coronal brain sections at the level of striatum grossly depicts the edematous ipsilateral hemisphere with the faintly stained ischemic area in acute stroke while atrophied ipsilateral hemisphere in chronic stroke. The tissue integrity of the chronic stroke rat brain was worse than a chronic stroke mouse brain. Furthermore, the presence of cells bearing pycnotic nuclei and vacuolated cytoplasm were readily present in the ischemic area of the acute stroke brain. In the chronic stroke brain, cells bearing non-neuronal morphology occupied the corresponding ischemic core region.

Fig 1. Induction of middle cerebral artery occlusion (MCAo) Rat and mouse model.

A. Transient cerebral ischemia was induced by inserting silicon-coated nylon suture (4-0 for rats and 6-0 for mice) via the right external carotid artery (ECA) through the internal carotid artery (ICA) to block middle cerebral artery (MCA).

B. Infarction resulted from the occlusion of MCA was evaluated by TTC staining 24 hours after reperfusion.

Fig 2. A general overview of a brain tissue structure of chronic stroke animal models. The changes in brain tissue integrity were monitored by MRI, H&E staining and cresyl violet staining from acute (day 3) to chronic phase of stroke (Day 30) in both rat and mouse model of stroke. MCA occlusion resulted in infarction of striatum and dorsolateral cortex of both rat and mouse brain in the acute phase of the stroke. In the chronic phase, rat MCAo resulted in the infarct cavity while chronic mouse MCAo resulted in atrophied ipsilateral brain hemisphere (A, B). Hematoxylin & Eosin and Cresyl violet staining showed significant loss of neural cells in acute stroke. Cells with Pycnotic nuclei and vacuolated cytoplasm are evident in ischemic core and penumbra regions. In the chronic phase, the infiltration of glial cells (non-neuronal morphology) is evident in ischemic core and penumbra regions of the ischemic brain (C, D).

Immunostaining against the neuronal marker, NeuN, and astroglial marker, GFAP showed large NeuN and GFAP non-reactive areas corresponding to ischemic region suggestive of on-going cell death mechanisms in acute stroke brain. In the chronic phase, there were sparsely distributed NeuN+ Neurons in the peri-infarct region while GFAP+ cells made a glial scar delineating ischemic and normal viable tissue. Iba1+ microglia/macrophages were readily observed in the ischemic region of acute stroke brain and persisted in the chronic phase of the stroke (Fig 3 A, B). In the chronic phase of ischemic stroke, there is the presence of growth inhibitory glial scar forming a fine border between ischemic core and normal viable brain tissue (4 A). This glial scar primarily consists of reactive astrocytes, activated macrophages/microglia, and extracellular matrix molecules, predominantly chondroitin sulfate proteoglycans (CSPGs). We found that, in the brain of chronic stroke brain, GFAP positive astrocytes became hypertrophied and elongated their processes from penumbra into the infarct core (4 B). These astrocytes strongly upregulated GFAP protein, a hallmark of astrogliosis responding to ischemic stroke. We were also able to observe that the pronounced change in astrocytic morphology and GFAP expression in reactive astrocytes was also accompanied by the upregulation of intermediate filament protein, Nestin. Double immune-histochemical studies showed that GFAP-positive reactive astrocytes in the glial scar expressed a high amount of extracellular matrix-like Neurocan and CSPGs, but did not overlap with ED1

positive activated microglia/macrophages. GFAP and ED1 positive cells along with extracellular matrix molecules neurocan and CSPGs formed a layer around the ischemic lesion, suggesting the presence of inhibitory glial scar in the chronic stroke rat brain. The glial scar is positioned in such a way that it forms a border around the injury site and acts as a neuroprotective barrier to evading inflammatory cells.

Fig. 3. Temporal changes in neural cells in acute (Day 3) and chronic stroke (Day 30). Immunostaining showed large NeuN and GFAP non-reactive areas corresponding to ischemic region suggestive of on-going cell death mechanisms in acute stroke brain. Iba1 immunostaining revealed activation of resident microglia and infiltration of macrophages into the ischemic brain in the acute phase of the stroke. In the chronic phase, there were sparsely distributed NeuN+

Neurons in the peri-infarct region while GFAP+ cells made a dense glial scar delineating ischemic and normal viable brain tissue. Iba1+

microglia/macrophages with activated morphology densely populated in the ischemic brain in the chronic phase of the stroke.

Fig. 4.Characterization of Glial Scar in chronic stroke. A. Low magnification image of a chronic stroke brain stained with GFAP & MAP2 and Region of interest for characterizing glial Scar. B. In the chronic phase of ischemic stroke, there is the presence of growth inhibitory glial scar forming a fine border between ischemic core and normal viable brain tissue. This glial scar primarily consists of reactive astrocytes (GFAP/Nestin+), activated macrophages/microglia (Iba1/ED1+) and extracellular matrix molecules, predominantly chondroitin sulfate proteoglycans (CSPGs). Note GFAP positive astrocytes became hypertrophied and elongated their processes from penumbra

into the infarct core.

Overall, the MCAo mice exhibited poor performance in behavioral tests during the first week after ischemia, but their performance improved spontaneously over the time. The MCAo animals consistently showed higher NSS score compared to the sham group through the testing period (Fig. 5A) In corner tests, normal or sham-operated mice displayed similar tendency in right and left turning behavior when the animals reached the corner. However, the MCAo animals showed the right-sided bias in turning back since the animals had the paresis in the left side (Fig. 5 B). The laterality index, the indicator of right turning preference, was significantly increased in MCAo animals (p<0.05).

Within the span of 28 days the right turning bias in the ischemic group was partially reduced owing to spontaneous improvement in their sensory-motor function on the left side of the body. The sham-operated animals turned both the directions with almost equal probability, over the entire period of behavioral testing. In the pole test before the surgery, when the mice were placed on the top of the pole facing their head upward, mice in both the sham and MCAo groups could turn their head completely vertically downward within baseline value of 1 second and reached the floor within 4 sec, when they were placed on the top of the pole facing their head upward.Ttotal, the time taken to reach the floor, was significantly increased after MCAo (Fig. 5C). Although the MCAo mice showed profound improvement in the motor function (bradykinesia) over the time, the

sham- operated animals always took lesser time to reach the floor over the testing span of 1 month. In the rotarod test, the MCAo mice remained on the rotarod for lesser time than did the sham-operated mice for the first week, but the difference became non-significant between the groups after a week (Fig. 5D)

Fig. 5. Behavior assessment in chronic stroke animal models. Behavior tests were performed on acute i.e, day 3 and chronic i.e, day 30 after the MCAo surgery. All data are presented as means±S.E. Both Rat and mouse MCAo animals showed severe behavioral deficits in neurological severity scores and corner test compared to sham operated animals in acute as well as chronic phase.

Rotarod and adhesive removal test were sensitive to detect behavioral deficits in chronic stroke rat models. Rotarod test was not sensitive to detect behavioral deficits in chronic phase of chronic stroke mouse model. The latency to reach the floor from the top of the pole was increased in the mouse MCAo animals compared to the sham group. (t-test, *p<0.05, ** p<0.01)

2. Therapeutic effects of MSC/Ngn1 cells in chronic stroke

We previously reported that Ngn1 overexpressing MSCs exerted higher therapeutic effects than naïve MSCs in acute stroke animal models. To assess the role of these MSC/Ngn1 cells in stroke recovery in later phases of stroke progression, cells were transplanted into the brain of Sub-acute (Day 15) and chronic stroke (1 month) rats. Analysis of relative neurological severity score at 8 weeks after stroke revealed that neither MSC/Ngn1 nor MSC cells exert beneficial effects on stroke recovery when treated at the chronic phase (Fig 6).

However, when treated at the acute or sub-acute phase of stroke, both MSC/Ngn1 and MSC cells reduced relative neurological severity compared to PBS vehicle with visibly higher recovery in MSC/Ngn1 group.

Fig 6. Ineffectiveness of MSC/Ngn1 cells in chronic stroke A. To test the therapeutic efficacy of these Ngn1 over-expressing MSCs (MSC/Ngn1) in stroke recovery in later phases of stroke progression, cells were transplanted into the ipsilateral striatum of acute (Day 3), Sub-acute (Day 15) and chronic stroke (1 month) rats. B. Analysis of relative neurological severity score at 2 months after stroke revealed that neither MSC/Ngn1 nor MSC cells exert beneficial effects on stroke recovery when transplanted at the chronic phase. However, when transplanted at an acute or sub-acute phase of stroke, both MSC/Ngn1 and MSC cells reduced relative neurological severity compared to PBS vehicle with visibly higher recovery in MSC/Ngn1 group.

- Contributed by Dr. YSW

Part B: Adenoviral mediated therapeutic genes transfer in MSCs.

1. MSC transduction with adenoviral vector expressing GFP.

To investigate adenovirus transduction efficiency, MSCs were transduced with Ad-GFP at various MOIs for 2 h and then examined at 48 h post-infection by fluorescence microscopy and flow cytometry. Results demonstrate a linear increase in the frequency of GFP-positive cells with increasing MOI with no GFP signal detected at 0 MOI and about 80% GFP positive cells at 200 MOI. The histogram (Fig. 7A) and bar diagram (Fig. 7C) show the frequency of GFP-positive cells at the indicated transduction conditions.

2. Transgene expression and growth kinetics of Ad-GFP-transduced MSCs To assess the effect of adenoviral transduction on the proliferation of cells, Ad-GFP-transduced MSCs were cultured under standard conditions for 2 passages. Cell counting studies revealed the linear decrease in the total cell numbers with increasing MOI after 12 days in culture (Figs. 8A and 8B). The number of GFP-positive cells and mean fluorescence intensity were also significantly decreased in subsequent passage (Figs. 8C and 8D). MSCs retained a normal fibroblastic morphology at 2 days post-transduction that gradually changed to a flattened, oval, and enlarged shape, mostly in GFP-positive cells (arrows in Fig. 8A). Senescence-associated beta-galactosidase (SA-beta-gal) staining confirmed that these morphological changes were the result of cell senescence (Fig. 8F).

Fig. 7. Adenoviral gene transduction of mesenchymal stem cells (MSCs).

(A) Fluorescence-activated cellsorting analysis of MSCs transduced with GFPexpressing adenovirus (1-200 multiplicity of infection [MOI]) and untransduced controls. (B) Microscopy images of GFP expression in MSCs transduced at 200 MOI. Scale bar = 50 µm. (C) The percentage of GFP positive cells with increasing MOI.

Fig. 8. Transgene expression in Ad-GFP MSCs. (A) Representative fluorescence images of transduced cells at 2 and 12 days post-transduction.

GFP-positive cells display a larger cell size with flattened morphology (arrows).

Scale bars = 50 µm (B-D) The total number of cells, frequency of GFP-positive cells, and mean GFP intensity were measured using flow cytometry. (E) Senescence-associated β- galactosidase activity was examined in transduced MSCs (MOI 50) and cultured for one passage. Arrows indicated positive X-gal staining. Scale bars = 50 µm

3. Cell surface marker expression and multi-lineage differentiation

To evaluate the effects of adenoviral transduction on the MSC phenotype, cell surface marker expression and multilineage differentiation potential were examined. As shown in Fig. 9, positive cells and GFP-negative cells displayed similar phenotype profiles. Ad-GFP transduced MSCs were positive for CD105, CD29, CD73 and CD90 while beingnegative for HLA-DR, CD34. The population of surfacemarker and GFP double positive was similar to transduction efficiency of MSCs by Ad-GFP at 50 MOI. We also assessed the effect of adenoviral transduction on MSC differentiation into osteocytes, adipocytes, and chon- drocytes. As expected, osteocytic cultures exhibited mineralization nodules over the MSC monolayer that stained positively with Alizarin S, confirming successful osteocyte differentiation (Figs.

10A and 10B). Similarly, GFP immunohistochemistry and Alcian blue staining revealed metachromatic extracellular matrix deposition in the GFP-positive chondrogenic spheroids. (Figs. 10C and 10D). The chondrogenic spheroids in both the MSCs and Ad-GFP MSCs group were similarly sized and produced extracellular matrix to a similar extent suggesting that Ad-GFP MSCs and control MSCs could still differentiate into chondrogenic- and osteogenic-lineages. In contrast, the majority of GFP-positive cells lacked the capacity for adipogenic differentiation as shown by the absence of lipid vacuoles, which were readily apparent in GFP-negative and untransduced counterparts (Figs.

10E-10H).

Fig. 9. Surface antigen expression profiles of Ad-GFP MSCs. MSCs transduced with GFP-expressing adenovirus (MOI at 50) were examined for MSC surface marker expression at 2 days post-transduction. Isotype antibodies served as negative control.

Fig. 10. Mesodermal differentiation potential of Ad-GFP MSCs. (A, B) Osteogenic differentiation was assessed by alizarin red S staining to visualize calcium deposits. (C, D) Chondrogenic differentiation was assessed by Alcian blue and immunohistochemical staining to detect glycosaminoglycans and GFP expression, respectively. Dark brown colored cells in (D) indicate GFP-positive cells. (E, F) Adipogenesis was monitored for intracellular lipid accumulation by microscopy and (G, H) oil-red O staining. All photomicrographs are the merges of GFP and bright field images. Scale bars = 50 µm.

4. Long-term transgene expression under growth-restricted conditions To examine long-term GFP expression stability, transduced MSCs were cultured for 30 days without passaging. Contrary to proliferating cultures, Ad-GFP MSCs subjected to growthrestrictive conditions showed stable Ad-GFP expression for at least 1 month in culture (Fig.11). However, long-term cultured MSCs acquired a senescent morphology with enlarged cell size (Fig.11, arrows).

5. Generation of MSC and MSC/Ngn1 cells overexpressing HGF.

The inability of MSC/Ngn1 cells in the treatment of chronic stroke impelled us to explore proper disease-modifying therapeutic candidate, Hepatocyte growth factor (HGF) which has been known to possess potent angiogenic, neurogenic, synaptogenic and anti-fibrotic properties in several pre-clinical and pre-clinical studies. We over-expressed HGF into MSC and MSC/Ngn1 cells by using an Adenoviral vector (Serotype 5). Assessment of HGF transgene 48 hours after transduction by RT-PCR, western blot, and immuno-cytochemical analysis shows that MSCs and MSC/Ngn1 cells transduced with Adeno-HGF expressed HGF at mRNA and protein level. The expression of HGF by MSC and MSC/Ngn1 cells was not detected by western blotting (Fig 12 C-F). The expression of Neurogenin 1, Ngn1 was confirmed in MSC/Ngn1 and MSC/Ngn1+HGF cells by RT-qPCR (Fig 12 A, B).

Fig. 11. Transgene stability in Ad-GFP MSCs under non-dividing culture conditions. Long-term transgene stability was monitored after 30 days of culturing under conditions unfavorable to cell division (i.e, high-density plating with limited passaging). GFP expression was then assessed by fluorescence microscopy. Arrows indicate representative cells.

Fig 12. Generation of MSC and MSC/Ngn1 cells overexpressing Hepatocyte growth factor, HGF. HGF was overexpressed into MSC and MSC/Ngn1 cells by using an Adenoviral vector encoding human HGF. Assessment of Ngn1 and HGF transgenes in MSC, MSC/Ngn1, MSC+HGF, and MSC/Ngn1+HGF cells were analyzed 48 hours after transduction by RT-PCR, western blotting and immuno-cytochemical analysis. MSC/Ngn1 and MSC/Ngn1+HGF cells expressed a higher level of Ngn1 gene compared to MSC and MSC+HGF cells (A, B). Similarly, MSC+HGF and MSC/Ngn1+HGF cells expressed a higher level of HGF compared to MSCs and MSC/Ngn1 cells (C, D). The Over-expression of HGF was assessed by western blot and immuno-cytochemical methods (E, F)

6. Cell surface marker expression and multilineage differentiation of Adeno-HGF transduced MSC and MSC/Ngn1 cells

To evaluate the effects of adenoviral transduction on the MSC phenotype, cell surface marker expression was examined. As shown in Fig. 13, Adeno-HGF transduced MSCs and MSC/Ngn1 cells displayed similar phenotype profiles. Ad-HGF transduced MSCs and MSC/Ngn1 cells were positive for CD105 and CD90 while being negative for HLA-DR, CD34, and CD45 at similar proportion to Adeno-HGF non-transduced cells. Similar to our previous study, MSCs expressing Ngn1 did not differentiate into osteo and adipogenic lineage. However, Adeno-HGF transduction did not alter the osteogenic differentiation of MSCs (Fig 14). Both osteocytic cultures of MSC and MSC+HGF cells exhibited mineralization nodules over the MSC monolayer that stained positively with Alizarin S, confirming successful osteocyte differentiation (Figs. 14A). Similarly, to the results from Adeno-GFP transduced MSCs the majority of Adeno-HGF transduced MSCs lacked the capacity for adipogenic differentiation as shown by the absence of lipid vacuoles, which were readily apparent in untransduced counterparts (Fig 14B).

Fig 13. Surface antigen expression profiles of Adeo-HGF transduced MSCs and MSC/Ngn1 cells. MSCs and MSC/Ngn1 cells transduced with HGF-expressing adenovirus were examined for MSC specific surface marker expression at 2 days post-transduction. Isotype antibodies served as a negative control.

Fig. 14. Mesodermal differentiation potential of Ad-HGF transduced MSCs and MSC/Ngn1 cells. (A) Osteogenic differentiation was assessed by alizarin red S staining to visualize calcium deposits (B) Adipogenesis was monitored for intracellular lipid accumulation by oil-red O staining. Negative control included cells growth in non-differentiation media. i.e., Fresh MSC growth media.

Part C: Therapeutic effects of HGF overexpressing MSC/Ngn1 cells in chronic stroke.

1. Improved functional recovery and tissue integrity with MSCs/Ngn1+HGF

Data described above (Fig. 6) demonstrated that both MSC/Ngn1 or MSC cells did not exert therapeutic benefits in the chronic phase of ischemic injury. Thus, we next investigated whether over-expression of HGF in MSC or MSC/Ngn1 cells could improve behavioral functions in the chronic stroke brain.

All the ischemic animals recovered spontaneously in behavior from day 2 to day 30 when animals had stable residual behavior. There was no significant difference in the rota-rod performance and adhesive removal time among/between all the groups at 1 month after ischemic injury. At 1 week to 8

All the ischemic animals recovered spontaneously in behavior from day 2 to day 30 when animals had stable residual behavior. There was no significant difference in the rota-rod performance and adhesive removal time among/between all the groups at 1 month after ischemic injury. At 1 week to 8