Transplantation of hPSC-derived pericyte-like cells promotes functional recovery in ischemic stroke mice

Pericytes play essential roles in blood–brain barrier (BBB) integrity and dysfunction or degeneration of pericytes is implicated in a set of neurological disorders although the underlying mechanism remains largely unknown. However, the scarcity of material sources hinders the application of BBB models in vitro for pathophysiological studies. Additionally, whether pericytes can be used to treat neurological disorders remains to be elucidated. Here, we generate pericyte-like cells (PCs) from human pluripotent stem cells (hPSCs) through the intermediate stage of the cranial neural crest (CNC) and reveal that the cranial neural crest-derived pericyte-like cells (hPSC-CNC PCs) express typical pericyte markers including PDGFRβ, CD146, NG2, CD13, Caldesmon, and Vimentin, and display distinct contractile properties, vasculogenic potential and endothelial barrier function. More importantly, when transplanted into a murine model of transient middle cerebral artery occlusion (tMCAO) with BBB disruption, hPSC-CNC PCs efficiently promote neurological functional recovery in tMCAO mice by reconstructing the BBB integrity and preventing of neuronal apoptosis. Our results indicate that hPSC-CNC PCs may represent an ideal cell source for the treatment of BBB dysfunction-related disorders and help to model the human BBB in vitro for the study of the pathogenesis of such neurological diseases.

pericytes came to the major conclusion that pericytes suppress transcytosis within the brain endothelium (Daneman et al 2010., Armulik et al., 2010. Further studies have substantiated these findings (Ben-Zvi Nature 2014). Assays to test the ability of the iPSC-CNC derived pericyte-like cells in suppressing transcytosis in HBMECs would be preferred. Such assays were performed by Stebbins and Gastfriend et al. In addition, quantification of frayed TJs is needed.
- Fig. 7. Please include gross coronal sections of brains with quantification of infarct zone following MCAO with and without the I.V. infusion of HBVPs and iPSC-CNC derived pericyte-like cells. While the improvement in neurological deficits and neuronal death are convincing it would be ideal to see the gross pathological features following these experiments.
-NCSC-PCs have no added benefit beyond HBVP. The injection of another cell types that fails to provide protection would be useful to show something unique about pericytes. In general, there lacks mechanistic detail on how either of these cells contribute to increased BBB properties. Is it conveyed by release of a molecule, or do pericytes need to integrate into the vascular wall in order for these properties to be conferred? It is possible that pericytes are supplying trophic factors such as pleiotropin (Nikolakopoulou et al., 2019), this could be considered as a protective mechanism.
-Functional tests of the BBB are needed in the MCAO experiments, brain edema/water content is not sufficient. Typically, release of endogenous blood-borne molecules are quantified (i.e. albumin, fibrinogen), and exogenous dyes of varying molecular weight are used (i.e. fluorescent-dextrans).
- Figure 8B,C: TJ protein expression and organization is difficult to discern in cross sections of vessels. Images showing the longitudinal orientation of vessels will provide more information in expression and organization and the consistency along the vasculature. IHC for TJ proteins are generally clean where localization is distinctly between endothelial membranes. The quality of TJ staining in the figure is poor. Perhaps better antibodies are needed to provide a clearer depiction of TJ protein expression and localization. It seems that the focus was on larger diameter vessels where the BBB properties are not as profound, therefore a capillary focus is ideal. Anatomical locations are also needed for all images taken within the brain and their relation to the penumbra or infarct core. Please provide imaging approach in methods, the thickness of sections, maximum projections etc.
-Does I.V. infusion of HBVP and iPSC-CNC derived pericyte-like cells suppress transcytosis within the brain endothelium following MCAO? - Fig. S8. Data showing perivascular localization of iPSC-CNC derived pericyte-like cells following I.V. infusion is not convincing. A couple incidental images are shown. Pdgfrb staining could be improved and images similar to ones found in Supplement Fig 9 would be great to see with endothelial-staining as well. Please quantify the amount of vessels with homed-pericytes. Where is this happening in relation to the infarct core? What kind of vessels? These are important factors to these studies and therefore needs to be included in a main figure.
Reviewer #2 (Remarks to the Author): This manuscript by Li and colleagues describes the differentiation of human pluripotent stem cells to pericyte-like cells through a neural crest intermediate. The authors use a variety of in vitro tests to demonstrate the pericyte character of their cells, all of which are reasonably well done and convincing. However, these results have been previously demonstrated in other studies (which the authors appropriately note). The authors then use a middle cerebral artery occlusion model and intravenous injection of the hPSC-derived pericytes to study therapeutic potential of these cells. These results are somewhat less convincing due to a lack of proper controls and a limited battery of tests. Overall, these limitations mute the reviewer's enthusiasm for this work.
Major concerns: 1) None of the in vitro work is novel per say. The authors correctly cite two papers that came out earlier this year demonstrating pericyte derivation from hPSCs through a neural crest intermediate. I can appreciate that the authors' work was likely in process when these other papers were published. However, given the profile of Nature Communications, I would be surprised to see a publication that confirms past results. This would be a decision for the editor, in my opinion.
2) The in vivo work lacks a proper cohort of experiments, but more importantly, lacks proper controls. Many previous reports have examined the beneficial effects of mesenchymal stem cells on stroke recovery, and as the authors properly note, pericytes have mesenchymal stem cell-like activity. However, all of their in vivo experiments use a vehicle control as the negative control, rather than a proper control cell type -such as hPSC-derived mesenchymal cells or hPSC-derived pericytes that are not "brain-specific". I suspect the effects that they observe are just general benefits of a cell that can suppress inflammation, rather than effects that are specific to brain pericytes. This outcome would significantly dampen the novelty of this work.
Minor concerns related to experiments: 1) In vitro characterization of BBB properties -the authors solely look at how pericytes improve tight junction-related properties in BMECs. Pericytes are known to serve other purposes, for example suppression of nonspecific transcytosis. The authors should demonstrate these properties as well. It is also unclear how the authors quantified "frayed" tight junctions. Their observations are very subjective. There are automated ImageJ and Matlab plugins that can be used to assess tight junction continuity quantitatively.
2) Figure 7B-D -a more comprehensive battery of assays for stroke recovery would include at least a rotarod test.
3) Fig 7E -the authors should show an example image of water content for each condition. Fig 8 -these images again seem subjective. The images in panel A for PBS could just be a slice where fewer vessels are in plane. It is also difficult to see differences in panels B and C. It is unclear why the authors didn't do a vessel isolation and a more quantitative assay like western blot.

4)
Reviewer #3 (Remarks to the Author): Comments on NCOMMS-19-24655 Pericytes are perivascular cells that grow adjacent to capillary vessels and play important roles in blood brain barrier integrity and cerebrovascular function. Here the authors describe a protocol for differentiating pericytes from human pluripotent stem cells through a neural crest intermediate stage.
The differentiated pericytes expressed markers for Pdgfrb, NG2, CD146, caldesmon, and vimentin, determined by immunostaining. An in vitro tube formation assay with endothelial cells indicated that derived pericytes could form tube-like structures, and a gel contraction assay indicated that the derived pericytes could contract the gel in which they were imbedded. The authors then briefly present experiments using their derived pericytes as a treatment after transient middle cerebral artery occlusion (tMCAO). Derived pericytes injected intravenously into mice after tMCAO improved neurological and behavior scores and reduced neuronal death, and improved tight junction protein fluorescence profiles in large vessels. In several assays and experiments in vitro and in vivo, commercially available human embryonic brain vascular pericytes were used for comparison with the derived pericytes.
Overall it is not clear what novelty the authors pericyte differentiation method brings to the field. There are now two published methods detailing pericyte differentiation from stem cell sources. Characterization of resulting pericytes described in this manuscript is minimal and entirely in vitro. Furthermore, characterization and mechanism of effects of iPSC-derived pericytes on stroke outcome is inadequate. Many concerns exist: 2. The characterization of NCSC-PCs is minimal in vitro staining. It would be beneficial to perform western blots to compare levels of known pericyte markers between NCSC-PCs and primary pericytes.
3. How are the pericytes injected into the blood stream crossing the BBB? What is the mechanism? Cells do not normally spontaneously cross the BBB in general.
4. Related to above, how was clumping of the infused cells controlled? How was the infusion of NCSC-PCs and HBVPs performed? What is the estimated survival-percentage of the pericyte cells inside the mouse body relative to the number injected? 5. Some pericytes express SMA. Detection of SMA expression in pericytes may be dependent on staining techniques (PMID:29561727).
7. In addition to the gel assay in Figure 6 A-B, it would be more convincing to demonstrate hiPSCderived pericyte contractility at the single cell level.
8. The authors investigate the pluripotent capacity of NCSC-PCs and HBVPs and found that their cells could be induced to differentiate to osteocytes, adipocytes, or chondrocytes. This seems more typical of mesodermal derived pericytes to differentiate into peripheral cells. However, brain pericytes have been demonstrated to differentiate into other brain cell types which is perhaps more interesting and relevant to NCSC-PCs (CNS pericytes) (Curr Pharm Des. 2008;14(16):1581-93, Stem Cells. 201533(6):1962-74. doi: 10.1002/stem.1977, Exp Gerontol. 2018112:30-37. doi: 10.1016/j.exger.2018 9. For the in vitro BBB models, the TEER values are very low for endothelial monolayers. Have the authors considered using primary endothelial cultures?
10. Figure 8 shows staining on large vessels (over 25 um diameter). How are pericytes, which are normally present on capillaries, contribute to such large vessels? The authors should focus on studying capillaries where the pericytes normally exist.
11. In the stroke experiments, were there differences between the stroke groups in terms of lesion volumes?
12. Since the iPSC-derived pericytes are injected into the vasculature after stroke, how are they getting to the infarct region, penumabra area, and other peri-infract areas, where blood flow likely does not exist, is low or greatly reduced?
13. The authors do not show adequate evidence of BBB leakage reduction after pericyte injection. BBB leakiness should be measured by staining for extravasated molecules. Reduced water content might just reflect changes in brain edema, and does not necessarily reflect vasogenic edema or BBB changes.
14. It is unclear whether changes in Glut1 after stroke with pericyte treatment are due to differences in microvascular density or Glut1 levels in the vasculature. This should be evaluated more carefully, using microvessel western blotting for example.
15. How was the rejection controlled in C57 mice? Immunosuppression?
16. Figure 8 Should start with a staining and quantification for pericyte coverage. Supplementary figure S8 stainings are not convincing, it is difficult to see cell morphology for example.
17. How certain are the authors that the glial cells observed in their differentiated cultures are "schwann" cells? The markers used (S100b and GFAP) are typically used as astrocyte markers, and astrocytes are commonly observed in neuronal iPSC cultures.
18. Are all HBVP cells used for comparison studies sourced from ScienCell? Pericytes from this source are human embryonic brain pericytes, not adult pericytes. This should be made clear in the text and considered carefully when interpreting results.
19. The authors spend a significant part of their discussion (most of page 23) describing pericyte models and treatments in non-brain tissue. Brain pericytes are well known to have somewhat different properties than peripheral tissues.

Response to reviewers
The reviewers raised a number of constructive criticisms and suggestions. To fully address them, we have performed additional experiments as well as implementing considerable changes to the manuscript. As a result, we believe the manuscript is much stronger. We wish to take this opportunity to thank the reviewers for their valuable input. Below, we summarize the reviewers' comments, and describe point-by-point how we have addressed them.

Reviewer #1
In the manuscript entitled "Transplantation of cranial neural crest-derived pericytes from human pluripotent stem cells promotes functional recovery in a murine model of ischemic stroke" by Sun, Gong, Huang et al., explored the protective potential of cranial neural crest-derived pericytes (from human iPSCs) following ischemic stroke (MCAO). This study shows that their protocol of differentiating iPSCs into cranial neural crest (CNC) stem cells into is robust.
They show that they can further differentiate them into pericyte-like cells (expressing the canonical pericyte markers Pdgfrb and NG2) using PDGFBB and bFGF. They show that the iPSC-CNC-derived pericyte-like cells participate in vascular growth and promote BBB properties in human brain microvascular endothelial cells (HBMECs). They show that I.V. infusion of iPSC-CNC-derived pericyte-like cells improve neurological/behavioral defects, neuronal death, and brain edema following MCAO-induced ischemic stroke. The protective effects elicited by the iPSC-CNC-derived pericyte-like cells may be due to improvements in the BBB properties in the brain vasculature.
While this paper does present a novel therapeutic idea in using iPSC-CNC derived pericyte-like cells to improve the pathological events of ischemic stroke, a very similar method in producing pericyte-like cells from iPSCs through a CNC differentiation protocol this has been previously published (Stebbins and Gastfriend et al., Mar 2019). However if more thorough investigations can be provided to show how the iPSC-CNC derived pericyte like cells improve the BBB and attenuate neuronal damage then the novelty of these studies may be heightened.

Major comments:
Point 1: More thorough characterization is required to define the "pericyte-like" cells in addition to Pdgfrb and NG2. Vimentin and Caldesmon are also highly expressed by vSMCs and fibroblasts. Expression of Anpep, Abcc9, kcnj8 would be ideal to see as well. Expressional analysis of other neural-crest derived and fibrocyte markers would be ideal such as Pdgfr-alpha, Col1a1/1a2, Tbx18 to understand how related these pericyte-like cells are to pericytes. It would be best to change phrasing in the title and throughout the text to refer to these cells as pericyte-like.
Response: As suggested, we refer to these cells as pericyte-like cells throughout the text and performed additional assays to thoroughly characterize the properties of human pluripotent stem cells-derived cranial neural crest pericyte-like cells (hPSC-CNC PCs). Using fluorescence activated cell sorting (FACS) analysis, we identified that PDGFRα, CD248, and NG2 were homogeneously expressed in hPSC-CNC PCs and HBVPs. NOTCH3, an important regulators of brain vascular integrity and pericyte expansion (Development. 2014,141:307-17), could also be detected in at least 50% of hPSC-CNC PCs and HBVPs. This data was added as Fig. S5a.
Accordingly, we compared the expression of pericyte or fibroblast-related genes at transcriptional level by qRT-PCR and found hPSC-CNC PCs and HBVPs expressed ABCC9, DLK1, KCNJ8, ANPEP (CD13), IFITM1, TBX18, PDGFRA, COL1A1, COL1A2, and LUM at similar or different levels, which was in accordance with our RNA-Seq data (Fig. S8d). This data was added as Fig.   S5b. In addition, we also identified hPSC-CNC PCs expressed PDGFRβ, Vimentin COL1A1, CD13, and IFITM1 by western blotting. Both of these data thoroughly define the characteristics of hPSC-CNC PCs cells. The corresponding data was presented as Fig. S5c.

Point 2:
The rationale for using human vascular brain pericytes (HVBPs) is not clearespecially when the pericyte transplantation experiments are ultimately performed in mice. More info is needed on how the HVBPs were isolated and whether they are freshly isolated or primary cell line.

Response:
We are sorry for the unclear description. In this study, human vascular brain pericytes (HVBPs) are served as a positive control of hPSC-CNC PCs. HBVPs were purchased from ScienCell Research Laboratories (Carlsbad, CA, USA; Catalog #1200), which were isolated from human brain, cryopreserved at passage one after purification, and could be expanded for 15 population doublings, according to the manufacturer's instructions. These cells have been widely used to study the formation and functionality of the blood-brain barrier (J Cell Mol Med. 2016,20:980-986;Acta Neuropathol.2016, 131:753-773;Int J Biol Sci. 2016,12:87-99). The corresponding information was added in Methods section.
Point 3: Please rephrase the "tube assays" to cord assay (Stebbins and Gastfriend et al.,). This phrasing presumes that the HBMECs co-cultured with HBVPs or iPSC-CNC pericyte-like cells are forming tubes however they are not known to lumenize under these conditions. Alternatively, data confirming lumenization could be provided.
Response: As suggested, we rephrase the "tube assays" to "cord assay" both in Results and Methods sections.
Point 4: Fig. 6A,B. The contractility assay is not typical and is testing tension and adhesion. A cell type that does not exhibit these behaviors would be needed to show this is unique to the HBVPs and iPSC-CNC pericyte-like cells.
Perhaps there is a better test for this and the authors can include a positive control/agonist to stimulate pericyte contraction in addition to adding a negative control cell type. In the absence of alpha-SMA expression, information on the mechanism of contractility would be appreciated.
Response: This is a good point. To address this concern, we performed additional experiments and used human bone marrow-derived mesenchymal Response: This is a good suggestion. As suggested, we performed additional experiments to determine the BBB integrity in vivo through Evans blue assay, measurement of fibrinogen/albumin deposits and fluorescent-dextran (70 kDa dextran-TRITC) assay. Compared to the PBS group, infusion of hPSC-CNC PCs or HBVPs significantly decreased the leakage of Evans blue dye (p＜ 0.05). However, treatment with HDFs did not improve the blood-brain barrier function efficiently. Similar results were obtained in fluorescent-dextran assay and measurement of fibrinogen/albumin deposits. Increased leakage of the endogenous plasma protein fibrinogen and albumin can be detected in PBS and HDFs groups, while treatment with HBVPs and hPSC-CNC PCs remarkably decreased the fibrinogen and albumin deposits in the brain (p＜ 0.001). Moreover, we discovered that more dextran-TRITC passed through the BBB in PBS and HDFs groups than that in HBVPs and hPSC-CNC PCs groups (p＜0.05). These data suggest that transplanted HBVPs or hPSC-CNC PCs could efficiently restore the integrity of BBB. The corresponding data was added as Fig. 7d, e. Accordingly, we added more detailed description of the anatomical locations and imaging approach of samples. All the samples were collected from the penumbra area, sliced into 100 µm-thick sections and then used for histochemical staining. Immunofluorescent images were acquired using the LSM800 confocal microscope (Zeiss) and Dragonfly high speed confocal microscopy (ANDOR, Oxford Instruments). Quantitative image analysis, including NG2 + pericyte coverage, fluorescent density analysis of extravascular dextran-TRITC, fibrinogen and albumin deposits, were performed from maximum projections of 10-µm-thick Z-stack images and conducted by the investigators blinded to the treatment groups using ImageJ software (NIH). We provide a detailed description of the experimental protocol in Methods section. Response: We are sorry for the inadequate evidence to identify the perivascular localization of the transplanted hPSC-CNC PCs. To address the concern of reviewer, we performed additional experiments to determine the number and location of homed-hPSC-CNC PCs and the amount of vessels with these homed-pericytes. FACS analysis was applied to test the homing ability of transplanted hPSC-CNC PCs to the BBB, and we found that the number of the homed-hPSC-CNC PCs (DsRedE2 + cells) increased gradually after cell transplantation (~1% of total cells on day 7). Fluorescence microscopy indicated that most of the DsRedE2 + cells located at the capillaries in penumbra area. We also used fluorescent lectin-Dylight488 to label blood vessels to identify the anatomical relationship between endothelial cells and the transplanted hPSC-CNC PCs. The results revealed that the amount of vessels (green) covered by homed-hPSC-CNC PCs (red) was about 10% 1 day post cell infusion, while about 30% of the capillaries in penumbra area were surrounded by DsRedE2 + hPSC-CNC PCs 7 days post-treatment.
Furthermore, immunofluorescence staining demonstrated that most of the DsRedE2 + homed-hPSC-CNC PCs still expressed the pericyte-specific markers, PDGFRβ and NG2, and localized near the lectin-Dylight488 + endothelial cells. These results confirmed the perivascular localization of the transplanted hPSC-CNC PCs. The corresponding data was added as Fig. 9a-d; S9b.

Reviewer #2
This manuscript by Li and colleagues describes the differentiation of human pluripotent stem cells to pericyte-like cells through a neural crest intermediate. Overall, these limitations mute the reviewer's enthusiasm for this work. Compared to this study, our method could efficiently obtain putative cranial neural crest stem cells within a relatively short period (6 days). Faal T et al.
also reported that 5-day treatment with CHIR99021-containing serum-free medium could efficiently induce hPSCs differentiate into p75 + HNK1 + NCSCs (the quantitative data of the percentage of p75 + HNK1 + cells was not provided), and the resulting cells without purification were then passaged and directly used for pericyte differentiation (Stem cell reports 2019,12:451-460).
Therefore, pericyte-like cells from other developmental origin (e.g., mesoderm) may exist in this study.
Both the above two studies used the p75 + HNK1 + NCSCs for deriving pericyte-like cells. Nonetheless, previous studies demonstrated there were two populations in p75 + cells derived from human pluripotent stem cells, only p75 bright HNK1 + cells with high levels of AP2α were authentic neural crest stem cells, while p75 dim HNK1 + cells expressing high levels of PAX6 were actually neural stem cells (Proc Natl Acad Sci USA. 2011,108:19240-5. Stem Cell Rep 2017,9:1043-1052. We also observed similar gene expression pattern in p75 bright HNK1 + (PAX6 low AP2α high ) cells and p75 dim HNK1 + (PAX6 high AP2α low ) cells in our differentiation system (see the following figure). Then we isolated this subpopulation of p75 bright HNK1 + cells that expressed a set of cranial markers (HOXA1, HOXB1, LHX5 and OTX2; Fig. 1d) for pericyte differentiation. Moreover, we showed that hPSC-CNC PCs in our protocol did not express the posterior HOX genes (HOX10-12), which were highly expressed by mesoderm progenitor-derived pericyte-like cells from hPSCs (hPSC-MP PCs), although hPSC-CNC PCs and hPSC-MP PCs shared similar expression pattern of typical pericyte markers (Fig. S7a, b). Taken together, our protocol represents a new, fast, and efficient method for derivation of cranial pericyte-like cells from hPSCs through distinctive developmental pathway, which may provide new clues for the derivation and application of iPSC-CNC derived pericyte-like cells. However, all of their in vivo experiments use a vehicle control as the negative control, rather than a proper control cell typesuch as hPSC-derived mesenchymal cells or hPSC-derived pericytes that are not "brain-specific". I suspect the effects that they observe are just general benefits of a cell that can suppress inflammation, rather than effects that are specific to brain pericytes.
This outcome would significantly dampen the novelty of this work. Moreover, we performed additional experiments to quantify the frayed TJs in HBMECs with or without pericyte-like cells according to the previous study (Sci Adv. 2019,5:eaau7375.), and found that the frayed ZO-1 tight junctions were greatly reduced after pericyte-like cell coculture when compared to HBMECs monoculture. The corresponding data was added as Fig. 5h. Figure 7B-Da more comprehensive battery of assays for stroke recovery would include at least a rotarod test.

Point 2:
Response: As suggested, we performed rotarod test to evaluate locomotor coordination and motor balance after tMCAO, besides the previous corner test and adhesive removal test. Latency to fall from the rod was recorded for each group. The results revealed that the rotarod latency (about 50 s) did not differ between all the groups on day 1 post cell transplantation. However, we found that the latency time was markedly prolonged on day 3 and day 7 after cell injection in hPSC-CNC PCs (up to 150 sec on day 7; p＜0.001) and HBVPs groups (up to 100 s on day 7; p＜0.01) when compared to the PBS and HDFs groups (about 50 s). The corresponding data was added as Fig. 6c. Fig 7E -the authors should show an example image of water content for each condition.

Point 3:
Response: As suggested, we provided representative images of water content (together with Evans blue staining) for each group. We observed that cerebral edema was apparent in the PBS and HDFs groups, whereas transplantation of hPSC-CNC PCs and HBVPs noticeably reduced the cerebral edema.
Quantification of water content also showed markedly reduction of cerebral edema in hPSC-CNC PCs and HBVPs groups when compared to PBS and HDFs groups (p＜0.05), which was in accordance with the corresponding brain swelling phenotype in different groups. The corresponding data was added as In addition, we performed vessel isolation from brain samples for the 22 detection TJ proteins using western blotting as suggested. The results exhibited that the protein levels of GLUT1, ZO-1, and Occludin were sharply downregulated in PBS group and were significantly elevated after transplantation of hPSC-CNC PCs or HBVPs, whereas HDFs treatment did not affect the expression of TJ proteins. These results were in line with quantitative data from immunostaining of TJ proteins. The corresponding data was added as Fig. 8c.

Reviewer #3
Pericytes are perivascular cells that grow adjacent to capillary vessels and play important roles in blood brain barrier integrity and cerebrovascular function. Here the authors describe a protocol for differentiating pericytes from days yielded ~90% p75 + /HNK1 + NCSCs, which were further purified for pericyte differentiation (Sci Adv. 2019,5:eaau7375.). Compared to this study, our method could efficiently obtain putative cranial neural crest stem cells within a relatively short period (6 days). Faal T et al. also reported that 5-day treatment with CHIR99021-containing serum-free medium could efficiently induce hPSCs to differentiate into p75 + HNK1 + NCSCs (the quantitative data of the percentage of p75 + HNK1 + cells was not provided), and the resulting cells without purification were then passaged and directly used for pericyte differentiation (Stem cell reports 2019,12:451-460). Therefore, pericyte-like cells from other developmental origin (e.g., mesoderm) may exist in this study.
Both the above two studies used the p75 + HNK1 + NCSCs for deriving the pericyte-like cells. Nonetheless, previous studies demonstrated there were two populations in p75 + cells, only p75 bright HNK1 + cells with high levels of AP2α were authentic neural crest stem cells, while p75 dim HNK1 + cells expressing high levels of PAX6 were actually neural stem cells (Proc Natl Acad Sci USA. 2011,108:19240-5. Stem Cell Rep. 2017,9:1043-1052. We also observed similar gene expression pattern in p75 bright HNK1 + (PAX6 low AP2α high ) cells and p75 dim HNK1 + (PAX6 high AP2α low ) cells in our differentiation system (see the following figure). Then we isolated this subpopulation of p75 bright HNK1 + cells that expressed a set of cranial markers (HOXA1, HOXB1, LHX5 and OTX2;  showed that pericyte-like cells of both origins highly expressed CD13, NG2, CD146, CD29, CD44, CD73, and CD105. To discriminate the difference between the mesodermal-and neural crest-derived pericyte-like cells, we used qPCR to detect homeodomain transcription factors (HOX genes) that determining the anterior-posterior identity of body segments (Development. 2005(Development. ,132:2931. Interestingly, the results unveiled a different HOX gene expression pattern between these two types of pericytes. We found that mRNA level of some of the posterior HOX genes (HOX10-12) was highly up-regulated in human mesodermal progenitor cell-derived pericyte-like cells (hPSC-MP PCs), rather than those of hPSC-CNC PCs, which was in line with the RNA-Seq data (Fig. S7b). The above evidence suggests that hPSC-MP PCs and hPSC-CNC PCs represent distinct developmental origins in spite of similar phenotype marker expression pattern, which may supply valuable insights to uncover the underlying mechanism why they possess different therapeutic capacity in tMCAO model. The corresponding information was added in Results and Discussion section. Accordingly, we compared the expression of pericyte or fibroblast-related genes at transcriptional level by qRT-PCR and found hPSC-CNC PCs and HBVPs expressed ABCC9, DLK1, KCNJ8, ANPEP (CD13), IFITM1, TBX18, PDGFRA, COL1A1, COL1A2, and LUM at similar or different levels, which was in accordance with our RNA-Seq data (Fig. S8d). This data was added as Fig.   S5b. In addition, we also identified hPSC-CNC PCs expressed PDGFRβ, Vimentin, COL1A1, CD13, and IFITM1 by western blotting. Both of these data thoroughly define the characteristics of neural crest stem cell-derived pericyte-like cells. We added these points to the Results section. The corresponding data was presented as Fig. S5c.
Point 3: How are the pericytes injected into the blood stream crossing the BBB?
What is the mechanism? Cells do not normally spontaneously cross the BBB in general.  S9a). We speculate that there may be at least three approaches contributing to the homing of hPSC-CNC PCs to the damaged BBB. First, in tMCAO mice, the reperfusion stage after vessel occlusion may allow the hPSC-CNC PCs to reach around the damaged site in the brain through the blood flow. Second, the ischemia-reperfusion injury would lead to the disruption of BBB structure and increase the BBB permeability as early as 6 hrs post-stroke (Nat Commun. 2016,7:10523;Neuron. 2014,82:603-17), which could result in pericyte-like cells crossing the BBB. Furthermore, the PDGF-B/PDGFRβ interaction may also play a pivotal role during the homing of pericyte-like cells. It was reported that PDGF-B and PDGFRβ were essential factors for the recruitment of pericytes to brain capillaries during embryonic development (Science. 1997,277:242-5;Development. 1999,126:3047-55), and the expression of PDGF-B was found to be conspicuously elevated from 3 h to 1 week in MCAO model (Brain Res Mol Brain Res. 2003,113:44-51). In addition, PDGFRβ was highly expressed in hPSC-CNC PCs as detected by qPCR, immunostaining and western blotting (Fig. 2c, d; S5c). Therefore, the injected hPSC-CNC PCs could be recruited to the penumbra area and help to reconstruct the BBB and prevent the apoptosis of neurons.

Actually, our results indicated that about ~1% of injected cells (on day 7)
were recruited and most of the homed hPSC-CNC PCs (DsRedE2 + cells) located at the capillaries in penumbra area. We also identified about 30% of the capillaries in penumbra area were surrounded by DsRedE2 + hPSC-CNC PCs 7 days post-treatment. Furthermore, immunofluorescence staining demonstrated that most of the DsRedE2 + homed-hPSC-CNC PCs still expressed PDGFRβ and NG2, and localized near the lectin-Dylight488 + endothelial cells (Fig. 9a-d; S9b). We revised the discussion to make this point more clear.

Point 4: Related to above, how was clumping of the infused cells controlled?
How was the infusion of NCSC-PCs and HBVPs performed? What is the estimated survival-percentage of the pericyte cells inside the mouse body relative to the number injected?

Response:
We added description of the cell infusion assay to the Methods section. In brief, confluent cells were trypsinized, triturated repeatedly and passed through a 40µm cell strainer to enrich for single cells. Cells were counted and visualized on a hemacytometer. If cell clumps were observed, cells were passed through a 25-gauge needle 10 times. Then 1×10 6 single cell suspension for each mouse were injected via caudal vein. The corresponding information was described in Methods section.
To address the reviewer's second concern, we performed additional experiments to detect the survival-percentage of transplanted cells homing to the damaged sites by FACS analysis, and discovered that the percentage of the homed-hPSC-CNC PCs (DsRedE2 + cells) increased gradually after cell transplantation (~1% of total cells in the ipsilateral hemisphere on day 7). The corresponding data was added as Fig. 9a.  10.1038/srep35108), in hiPSCs?
Response: As suggested, we performed qPCR and western blotting to detect the expression of interferon-induced transmembrane protein 1 (IFITM1) in pericyte-like cells. The result showed that mRNA and protein of IFITM1 was strongly expressed in hiPSC-CNC PCs (Fig. S5b, c). Previous study also 32 reported that IFITM1 was expressed in human embryonic stem cells (hESCs) (FEBS Open Bio. 2017,7:1102-1110, which is in accordance with our RNA-Seq data (the TPM value of IFITM1 in hiPSC is 662). Then qPCR was also carried to compare the mRNA level of IFITM1 between hiPSCs and hPSC-CNC PCs, and we found that hPSC-CNC PCs had significantly higher level (3~9 folds) of IFITM1 transcripts than that in hiPSCs (see the following figure).

Point 7:
In addition to the gel assay in Figure 6 A serum-containing medium for 2 weeks. We infer that most, if not all, the pericyte-like cells (with mesenchymal phenotype) might lose the multipotency of NCSCs. Therefore, as suggested by the reviewer, we perform additional experiments to investigate whether hPSC-CNC PCs and commercial HBVPs have similar multilineage-differentiation characteristics under protocols described in the above reports. We found that less than 1‰ hPSC-CNC PCs could be differentiated into Tuj1 + neurons and S100B + astrocytes. HBVPs also showed the similar tendency (see the following figure). These results indicate that a very small population of the hPSC-CNC PCs retained the neural lineages differentiation potential in vitro, although such behavior was not found in transplanted cells in our study. Response: In our study, the HBMECs were primary cells and obtained from ScienCell Research Laboratories (which were isolated from human brain and cryopreserved at passage one; Catalog #1000) and used for TEER assay at passage 3-4. Here, the TEER values for HBMECs monolayers are about 20-50 Ωcm 2 , which is similar to the results of previous reports using the same cell source or other primary endothelial cell lines (Front Cell Neurosci. 2019.13:230;MethodsX. 2015,3:25-34;Handbook of Neurochemistry and Molecular Neurobiology, 3rd ed., 2007, 29-55). However, some studies also reported significant higher TEER values when using hPSC-derived BMECs (Stem Cell Reports. 2019,12:451-460;Sci Adv. 2019,5: eaau7375;Stem Cell Rep. 2017,8:894-906) or primary HBMECs from ScienCell Research Laboratories (Cell Rep. 2019,26:1598-1613J Neurosci Methods. 2013,212:173-9;BMC Neurol. 2019,19:289). Therefore, we speculate the discrepancy in TEER values may be due to the difference in cell sources/batches, TEER assay protocol, or measurement equipment used in these studies. Furthermore, when co-cultured with CNC PCs or HBVPs, significantly increased TEER value in HBMECs was observed from day 2 to day 4. Also, the immunostaining revealed that HBMECs co-cultured with CNC PCs or HBVPs had notably fewer frayed ZO-1 tight junctions and decreased the transcytosis in HBMECs compared to HBMECs monoculture (Fig. 5e-j). The above evidence demonstrates that the CNC PCs could efficiently improve the barrier function of HBMECs and we believe that our findings are convincing.
How are pericytes, which are normally present on capillaries, contribute to such large vessels? The authors should focus on studying capillaries where the pericytes normally exist.

Response:
We would like to clarify that Figure 8 displayed the co-staining of TJ proteins with endothelial marker CD31 or astrocyte marker GFAP in brain samples from different groups, which showed that the expression of TJ proteins was significantly restored in these large vessels, rather than showing the localization of the transplanted hPSC-CNC PCs.  1997,277:242-5;Development. 1999,126:3047-55), and the expression of PDGF-B was found to be conspicuously elevated from 3 h to 1 week in MCAO model (Brain Res Mol Brain Res. 2003,113:44-51). In addition, PDGFRβ was highly expressed in hPSC-CNC PCs as detected by qPCR, immunostaining and western blotting (Fig. 2c, d; S5c). Therefore, the injected hPSC-CNC PCs could be recruited to the penumbra area and help to reconstruct the BBB and prevent the apoptosis of neurons.
Actually, our results indicated that about ~1% of injected cells (on day 7) were recruited and most of the homed DsRedE2 + cells located at the capillaries in penumbra area. We also identified about 30% of the capillaries in penumbra area were surrounded by DsRedE2 + hPSC-CNC PCs 7 days post-treatment. Furthermore, immunofluorescence staining demonstrated that most of the DsRedE2 + homed-hPSC-CNC PCs still expressed PDGFRβ and NG2, and localized near the lectin-Dylight488 + endothelial cells (Fig. 9a-d;   S9b). We revised the discussion to make this point more clear. Point 17: How certain are the authors that the glial cells observed in their differentiated cultures are "schwann" cells? The markers used (S100b and GFAP) are typically used as astrocyte markers, and astrocytes are commonly observed in neuronal iPSC cultures.
Response: This is a good point. It is known that S100b and GFAP are typical markers of astrocytes in central nervous system (Acta Neuropathol. 2010,119: 7-35), while they are also widely used for the characterization of schwann cells (Nat Biotechnol. 2007,25:1468Am J Stem Cell 2013,2:119-131;Stem Cells Transl Med. 2012,1:266-78.). In our study, however, these S100b and GFAP double positive cells are derived from p75 high HNK1 + neural crest stem cells, which usually generate cells of peripheral nervous system (Nat Biotechnol. 2009,27:275-80;Nat Biotechnol. 2007,25:1468. Moreover, we performed additional experiments and found that most of the GFAP + glial cells co-expressed another Schwann cell marker, myelin basic protein (MBP), which is not present in astrocytes (J Neurosci. 1986(J Neurosci. ,6:1925(J Neurosci. -1933 Response: This is a good suggestion. It was reported that brain pericytes had some different characteristics in contrast to the peripheral counterparts. First, brain pericytes help to seal the BBB endothelium by regulating the expression of endothelial TJ proteins, which results in low rate of bulk-flow transcytosis. However, capillary endothelium in peripheral organs is leaky (Nat Neurosci. 2016,19:771-83;Physiol Rev. 2019,99:21-78.). We did discover that our hPSC-CNC PCs could strengthen the BBB integrity by increasing the expression level or inhibiting the disruption of TJ proteins and suppressing transcytosis in BBB endothelium both in vitro and in vivo. Also, Vanlandewijck M et al. (Nature. 2018,554:475-480.) revealed that CD13 was not expressed by lung pericytes, although these peripheral pericytes shared canonical pericyte markers Pdgfrb, Cspg4, and Des with brain pericytes. Moreover, they found that many members of different types of transporters were abundant in brain pericytes but low or absent in lung pericytes. In our study, we demonstrated that almost all of the hPSC-CNC PCs and HBVPs were positive for CD13 by FACS analysis. We further discovered that hPSC-CNC PCs highly expressed some of the solute carrier (SLC) transporters (SLC1A5, SLC2A10, SLC16A1, and others), ATP-binding cassette (ABC) transporters (ATP1B3, ATP13A1, ATP2C1, and others), and ATPase (ATP) family members (ABCA1, ABCA2, ABCA7, and others), as illustrated by RNA-Seq data (Fig. S8d). The above evidence suggests that the hPSC-CNC PCs closely resemble their in vivo counterparts in the phenotype and functional properties.
More importantly, the dysfunction of brain pericytes has been implicated in various neurodegenerative diseases. Nikolakopoulou AM et al. showed that pericyte loss led to a rapid neuron loss in the brain due to loss of As suggested, we repackaged this part and the above information was added in Discussion section.
Point 20: Important stroke methodology details are missing. What is the mortality rate of stroke surgery, exclusion criteria for these experiments, and how many mice were excluded? Was laser Doppler or a similar method used to confirm successful occlusion?
Response: As suggested, we provide a detailed description of the experimental protocol in Methods section. The mortality rate of stroke surgery in our study was about 22.2% (52/234) on day 3 and 30.4% (28/92) on day 7.
The inclusion criteria were: 1) regional cerebral blood flow decreases ＞70% during occlusion as detected by laser Doppler flowmetry; 2) neurological deficits 3 hour after tMCAO with neurological score between 2~3 points (determined as described in Stem cells 2018Stem cells , 36: 1295Stem cells -1310. Overall there were 58 mice were excluded (49 for death and 9 for neurological scores).
Laser Doppler flowmetry was used to confirm the ischemic stage and reperfusion stage. The corresponding data was added as Fig. S9a.

Point 21:
The title of the manuscript does not match very well the data presented. The stoke data presented is minimal.
Response: As suggested, in the revised manuscript, we pay particular attention to the in vivo function and therapeutic potential of cranial neural crest-derived pericyte-like cells from human pluripotent stem cells for treating the murine model of ischemic stroke. We believe it will match the title quite well.