Single-Cell Transcriptomics and In Vitro Lineage Tracing Reveals Differential Susceptibility of Human iPSC-Derived Midbrain Dopaminergic Neurons in a Cellular Model of Parkinson’s Disease

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The work described by Cardo et al., described the generation using CRISPR/CAS9 of new tool cell lines expressing BFP under the control of LMX1A, a mDA progenitor specification master gene. They show using ICC, FACS and scRNAseq that the sorted cells mainly produce mDA neurons and can be used to study the mechanisms of diseases specifically affected this population of neurons. To support their demonstration, authors used a MPP phenotypic models of Parkinson’s disease and analysed the specificity of MPP toxicity in BFP positive or negative cells. 

To be amenable to publication, the authors should more clearly demonstrate the value of their sorting sytem but this can be simply done by adding additional characterisation of the model/system as well as a careful interpretation and rewriting of some part of the study.

METHODS

The protocol described in this section to produce mDA cells is not clear and there is some discrepancy with what is proposed in figure 2 (for explemple, sorted cells are exposed to TGFB3 cocktail but does it concern also not sorted culture? This is not of the graph figure 2). Text or figure have to be corrected to be consistent in the two parts and protocol made clearer for non-experts ready. Keep in mind that the reader should be able to replicate the data.

 

RESULTS:

Generation of human iPSC LMX1A-Cre/AAVS-BFP lines (Figure 1):

The quality of the cell lines is well performed however, it is not clear if author check that no off –target edition occurred (5 top predicted genes are sufficient). This is gols standard validation for CRISPR lines and this should be done.

 

Immunocytochemistry/FACS characterisation of the edited lines (Figures 2/3):

- Authors need to quantify by ICC or FACS (or both) the direct co-expression (co-localization) of LMX1A and BFP. It has to be clear that BFP is only expressed in LMX1A positive cells. This is not demonstrated clearly  in figure 2 (conclusion line 233- 234 is not true or supported by any data) and questionable in figure 3A and B where, in not sorted cultures, 80% of the cells are positive for BFP but only 60% for LMX1A. This is key that only LMX1A positive cells express BFP for the tool to be used for lineage tracing and for sorting.

- More generally: FOXA2 is a marker of VENTRAL midbrain (should be corrected line 233), PAX6 is OK for dorsal forebrain but not for lateral midbrain. Lateral midbrain should be investigated by labelling NKX6.1. The authors should also label and quantify ventral forebrain cells (DLX2 and GSX2) and ventral hindbrain (GBX2) that can also contaminate the culture.

This will help demonstrated that there is a value to sort the culture. If the culture is only composed of ventral midbrain LMX1A positive cells, there is no point or additional value to sort the cells.

- The authors need to go deeper in the characterisation of TH neurons. GIRK2 is a marker only for substancia nigra neurons. However, the RNAseq data suggest that the main identity of the neurons produced from LMX1A sorted cells is mainly form the ventral tegmental area (Calbindin positive, clusters C1 and C2). ICC for Calbindin has to be perform and the proportion of GIRK2 (SN) vs Calbindin (VTA) in the TH (DA neurons) population needs to be carefully done to conciliate this discrepancy. This is important to interpret the value of the sorting in the context of PD that mainly affect SN neurons (VTA is spared).

scRNA seq (Figure 4)

Clusters obtained from sorted cells should be compared to the clusters present in not sorted culture (should be extrapolate from scRNAseq of figure 5: not sorted, not MPP treated cells), to show and discuss the value of sorting. Did sorting ended up with less, more pure clusters compared to not sorted cultures.

Selective vulnerability to MPP (Figure 5)

Figure 5B-C: The authors only proposed indirect evidences of the selective vulnerability of BFP positive cells. Not sorted culture should be treated with MPP then BFP cells have to be sorted and percentage of viable cells, Annexin cells and IP cells quantify in sorted and not sorted cells. As an alternative, Annexin and PI can be counted in BFP vs non BFP positive cells after MPP treatment (co-localisation can be done by FACS). Additionally, authors should performed a Caspase -3 labelling and/or TUNEL staining by ICC and co-stained with BFP in one hand, and TH in another hand, to firmly affirmed the specificity of apoptosis/cell death induction in BFP positive cells.

This is particularly important since results of the scRNAseq do not really support specificity as presented in figure 5 D to I. This part is particularly hard to understand and to interprete. Authors should extant the explanations in the text and add labels explaining the different treatments regimen directly on the graph/illustration. The figure legend need to be extend to made understanding of the data easier.

GENERAL:

Please indicated in each legend of figures (ICC and FACS) the number of differentiation but ALSO the number of technical replicates per differentiation (i.e. number of well or slides analysed) and the minimum number of cells counted per replicates.

DISCUSSION :

The discussion is too superficial. The authors need to discuss more deeply their results, findings and their implication in a broadest context.

More particularly the authors must discuss:

- How their tool, based on BFP, is comparable to others LMX1A sorting systems described for pluripotent stem cells and in which extend it is different. What are the advantage of this technology? (see for comparison Samat B, Nat comm, 2016; de Luzy IR, J.Neurosciences, 2019; Yoo JE, NPJ Parkinson Dis., 2021)

- Why choosing LMX1A rather than other progenitor markers (like En-1 for example) latter markers such as PITX3 or even TH?

- Discuss deeper the results of the 2 figures reporting RNAseq that remain too superficial. What are the new demonstrations and the author conclusions?

- What would be the next steps using the cell lines to investigate mechanisms of PD in the MPP phenocopy model and/or in genetic models?

In general, the discussion should support the author’s conclusions and demonstrated the value of this new model.

 

 

Author Response

We thank the reviewer for their constructive comments which helped the improvement of the MS.

The protocol described in this section to produce mDA cells is not clear and there is some discrepancy with what is proposed in figure 2 (for example, sorted cells are exposed to TGFB3 cocktail but does it concern also not sorted culture? This is not of the graph figure 2). Text or figure have to be corrected to be consistent in the two parts and protocol made clearer for non-experts ready. Keep in mind that the reader should be able to replicate the data.

We thank the reviewer for pointing this out. The schematics in figure 2A is a correct outline of the protocol. We however enhanced media supplement for cell sorting and the 1^st week of post-sort culturing in order to maximise cell survival. We have revised the relevant part of the Method highlighting the slight modification in media supplements for all post sort cell fractions and non-sorted sister control cultures.

The quality of the cell lines is well performed however, it is not clear if author check that no off –target edition occurred (5 top predicted genes are sufficient). This is gold standard validation for CRISPR lines and this should be done.

We used Gt-scan to predict the gRNAs off-targets (https://gt-scan.csiro.au/submit/). This analysis showed that the gRNAs used in this study present no exact match, nor 1 or 2 mismatches and hence predicted no potential off-targets. This test was verified using other tools (Deskgen and Atum). Using genomic PCR method, we didn’t detect any off targets in previous 8 independent genome editing projects where 1 or 2 mismatches were present, we therefore didn’t carry out loci-specific PCR test.

However, as routine, we performed genome wide screen for potential inadvertent genomic event occurred during CRISPR/Cas9 genome editing using the Illumina Infinium Global Screening Array v2.0. Comparing to parental Kolf2 cells, no additional copy number variance or single nucleotide polymorphisms were detected. The method and screening outcome is provided in the supplemental file.

Authors need to quantify by ICC or FACS (or both) the direct co-expression (co-localization) of LMX1A and BFP. It has to be clear that BFP is only expressed in LMX1A positive cells. This is not demonstrated clearly in figure 2 (conclusion line 233- 234 is not true or supported by any data) and questionable in figure 3A and B where, in not sorted cultures, 80% of the cells are positive for BFP but only 60% for LMX1A. This is key that only LMX1A positive cells express BFP for the tool to be used for lineage tracing and for sorting.

We fully appreciate the importance of this point. Unfortunately, the only LMX1A and BFP antibodies worked in our hands are made in rabbits so not compatible for double staining.  We however have LMX1A antibody staining data performed one day after sorting (line L25).  95.40±0.15% of BFP⁺ cells were stained positive to LMX1A one day after FACS isolation at day 18. This information is now included in the revised figure S1D-E.

More generally: FOXA2 is a marker of VENTRAL midbrain (should be corrected line 233), PAX6 is OK for dorsal forebrain but not for lateral midbrain. Lateral midbrain should be investigated by labelling NKX6.1. The authors should also label and quantify ventral forebrain cells (DLX2 and GSX2) and ventral hindbrain (GBX2) that can also contaminate the culture. This will help demonstrated that there is a value to sort the culture. If the culture is only composed of ventral midbrain LMX1A positive cells, there is no point or additional value to sort the cells.

The text description relating FOXA2 and PAX6 has been reworked in the revised MS and highlighted. We did not see expression of DLX2 and GSX2 in our scRNAseq data, suggesting a lack of ventral forebrain cells. However, GBX2, NKX6-1 and PAX6 expression was detected at various stages, indicating the presence of lateral midbrain and hindbrain cells.

The authors need to go deeper in the characterisation of TH neurons. GIRK2 is a marker only for substancia nigra neurons. However, the RNAseq data suggest that the main identity of the neurons produced from LMX1A sorted cells is mainly form the ventral tegmental area (Calbindin positive, clusters C1 and C2). ICC for Calbindin has to be perform and the proportion of GIRK2 (SN) vs Calbindin (VTA) in the TH (DA neurons) population needs to be carefully done to conciliate this discrepancy. This is important to interpret the value of the sorting in the context of PD that mainly affect SN neurons (VTA is spared).

We fully agree with the reviewer regarding the importance of producing PD relevant mDAs (ie. SNc-like DA). We do have GIRK2⁺ (gene ID: KCNJ6) as indicated in figure 4D and figure 5G. We also detected SOX6, a transcription factor preferentially expressed in the developing SNc (PMIDs: 25127144, 34678205, 34758317). In light of the observed differential vulnerability of BFP⁺ neurons to MPP⁺, our data would be consistent with the view that SNc-like DA neurons or their immature precursors are produced in our cultures.

We understood however that GIRK2 expression is not specific to SNc in the fetal midbrain. Several published hPSC literatures inferring the production of SNc DA neurons draw evidence from analysis of transplanted hPSC derivatives that have been matured in the host several months.  Moreover, single cell transcriptomics analysis (PMID: 27716510) indicated that cell clusters defined as SNc and VTA neurons in the adult brain are not yet present in the human fetal ventral midbrain, nor in DA induced hPSC cultures.  

Clusters obtained from sorted cells should be compared to the clusters present in not sorted culture (should be extrapolate from scRNAseq of figure 5: not sorted, not MPP treated cells), to show and discuss the value of sorting. Did sorting ended up with less, more pure clusters compared to not sorted cultures.

The suggested re-analysis revealed similar cluster distribution between the unsorted and (d18) sorted BFP⁺ samples. The lack of difference could be attributed to the high mDA content in our unsorted cultures. The relatively smaller number of cells being analysed using the ICELL8 platform (as compared to the 10x) may also limits the detection of population heterogeneity. In keeping with this possibility, the B0 neuronal cluster, which is composed of mostly derivatives of sorted BFP^- fraction, was found to exhibit gene expression traits not expected for mDA neurons (eg. SST, ISL1, SLC17A6, and SLC17A7) and show greatest numbers of DEGs to B1 (B0 vs B1, 249 DEGs) and B3 (B0 vs B3, 214 DEGs). In contrast, only 12 DEGs were detected between B1 and B3 DA neurons. This finding suggests that BFP^- neural progenitors primarily give rise to non-DA or mis-specified DA neurons, hence provides a rational for sorting.

Figure 5B-C: The authors only proposed indirect evidences of the selective vulnerability of BFP positive cells. Not sorted culture should be treated with MPP then BFP cells have to be sorted and percentage of viable cells, Annexin cells and IP cells quantify in sorted and not sorted cells. As an alternative, Annexin and PI can be counted in BFP vs non BFP positive cells after MPP treatment (co-localisation can be done by FACS). Additionally, authors should performed a Caspase -3 labelling and/or TUNEL staining by ICC and co-stained with BFP in one hand, and TH in another hand, to firmly affirmed the specificity of apoptosis/cell death induction in BFP positive cells.  This is particularly important since results of the scRNAseq do not really support specificity as presented in figure 5 D to I. This part is particularly hard to understand and to interprete. Authors should extant the explanations in the text and add labels explaining the different treatments regimen directly on the graph/illustration. The figure legend need to be extend to made understanding of the data easier.

In addition to the experiment presented in the paper, we did perform the Annexin V assay using the work flow as suggested by the reviewer. However, similar level of live/dead cells was observed between the sorted BFP⁺ and BFP^- fractions in both L25 and L35 lines. Our interpretation was that MPP⁺ sensitive (dead) cells were selected out during the process of sorting while the harvested remaining BFP+ cells had similar response to BFP- fraction. This outcome is inconclusive and doesn’t contract to the findings made with unsorted cultures undergoing Annexin V assay straight after MPP+ treatment.

We agree that Caspase 3 or TUNEL staining in conjunction with BFP and TH would provide additional support for the selective death of BFP⁺ cells. This experiment is outside the scope of the revision time frame however.

As suggested by the reviewer, We revised the MPP+/scRNAseq part of text and associated figure legend to highlight the difference between B3 (DA) and B0 (non-DA) clusters, which serve as additional evidence on selective vulnerability of BFP+ cells to MPP⁺.

Please indicated in each legend of figures (ICC and FACS) the number of differentiation but ALSO the number of technical replicates per differentiation (i.e. number of well or slides analysed) and the minimum number of cells counted per replicates.

This information was provided in the Method part of the original MS (under Immunofluorescence). We have however also included this information in the revised figure legend as requested by the reviewer.

The discussion is too superficial. ----

We thank the review for their constructive suggestions. We have included new discussions that we believe covers all aspects requested by the reviewer.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The authors used a safe harbour insertion of a markers Blue Fluorescent Protein (BPF) using LMX1A-CRE and AAVS1-BFP (to activate and maintain BFP expression once the LMX1A dopaminergic lineage marker gene is expressed).  This allows them to trace dopaminergic neurons (DANs) in iPSC derived cultures. The logic for selecting LMX1A is clear, it is a well know marker of DAN lineage.  The authors select two heterozygous lines with a CRE knockin at the end of the LMX1A protein to drive the expression of BPF.  They show these lines produce normal iPSC and DANs and that they can enrich BPF positive cells by FACS. They characterize a time course of DANs using scRNAseq.  Then they look at the sensitivity of the PD inducing toxin MPTP and perform scRNAseq on MPP treated and untreated DANs.  I commend the authors for a great feat of work.  The work is well done, however there are a lot of issue with the rationale for experiments and the description of results. I find the MPP and scRNAseq experiments very interesting, but I am left wondering why the generation of the BFP reporter line was necessary as these experiments could have been done in wildtype cells.

Major issues:

 

1.     Clarity of the description of the study

a.     The point of the study doesn’t come across clearly in the introduction.  Are the authors generating this line simply to be able to FACs sort early DANs or progenitor cells that will become DANs?

b.     The specifics of the method of generating the line are well described.  However, the general concept is not introduced.  The reader is expected to know that CRE expression will activate the BPF expression and that AAVS1 locus is a safe harbour site.  Which as I understand means is will not be silenced by the cell’s homeostatic mechanisms overtime.  These concepts should be described.  

c.     KOLF2 iPSC line is not introduced.  Is this a well-established line published before?  Provide references.

d.     A timeline schematic of the DAN differentiation should be shown for clarity.

 

2.     Lack of BFP positive cells in homozygous knock in.  The authors point out that in CRE+/+ lines they get no BFP expression and propose that this is due to deleterious effect of the CRE knock in.  Are we to assume that in the CRE-/+ the negative effect is lost.  What has happened to LMX1A expression?  Is it replaced with CRE? CRE is not usually toxic to cells.  The authors should clarify what they propose is happening.  Additionally, levels of CRE and LMX1A protein should be examined. By the schematic it appears the last exon of LMX1A is replace with CRE.  This should also be better described.

3.     The supplemental figures were not included.

4.     The github link is not complete and so the code and data are not available for review. The data was also not made available for review.

5.     It seems odd that the FACS sorting BFP cells only increase the BPF+ population from ~55% to ~80%.  How do the authors explain so many BFP- cells going through their gating strategy?

6.     Single cell sequencing characterization analysis:

a.     It would be clearly if the author annotated their clusters from Figure 4 A before showing the number of cells from each time point in each cluster group in C.  As they have done in D.

b.     It would be easier to understand the DGE contrast if the younger age was list before the older age in E.

c.     The comparisons in G and H are pointless to me.  What do the authors want me to learn here?  PCA analysis is a dimensional reduction, and the clusters would be expected to segregate in PC space, but this doesn’t inform us about their characteristics.

d.     In panel I the velocity analysis is very interesting.  The authors should add labels to the UMAP for the cell types (Prog, DA, Astrocytes).

e.     The overlay density plots of the velocity output are not well described.  Are we looking at a selected set of gene expression?  A gene module?  Some kind of velocity score?  It would be more informative to show expression of DA markers or modules of markers over pseudo time.

f.      Also, it is not clear if the scRNAseq was done on the BFP sorted population of the whole population.

g.     In the MPP treatment experiments the authors show a decrease in BFP positive cells.  Are we to assume this means DANs are being selectively kill?  The author’s interpretation is not explained.

7.     scRNAseq MPP treatment:

a.     It would be useful to annotate the clusters in Figure 5F and H.

b.     Here the authors should analyze this data to identify characteristics of the cells lost or dying from the MPP treatment.  What are the characteristics of the surviving cells?  If DANs persist how are these different from the absent population?  They can compare to the cells in the basal condition. The authors have compared by each cluster the GO terms effected and apoptosis and mitochondrial health pathways affected, which is not very revealing.

 

Minor:

1.     Figure 1 the legend has incorrect matching for the lettering in the figure and the panel descriptions.  1C is 1C and 1D, 1D is 1E.  The description for the pie charts D is not sufficient.  Figure 1E is not described in the legend.

2.     Figure 2 the statistics are not indicated on the plots. Also, the number of images used and if n=field of view, well or experimental replicated is not indicated.

 

Comments on the Quality of English Language

The figure legends are difficult to understand.  I think the clarity of text is impaired by the language. 

Author Response

We thank the reviewer for their constructive comments which helped the improvement of the MS.

 

1. Clarity of the description of the study
2. The point of the study doesn’t come across clearly in the introduction.  Are the authors generating this line simply to be able to FACs sort early DANs or progenitor cells that will become DANs?

The ability of isolating pure mDA progenitors for downstream interrogation during their differentiation to mDA neurons was indeed a strong motive. We are also interested in using this tool to study other progenies of LMX1A progenitors, for instance, glia cells. We have revised the introduction and discussion to highlight the rationals.

1. The specifics of the method of generating the line are well described.  However, the general concept is not introduced.  The reader is expected to know that CRE expression will activate the BPF expression and that AAVS1 locus is a safe harbour site.  Which as I understand means is will not be silenced by the cell’s homeostatic mechanisms overtime.  These concepts should be described.  

We thank the reviewer for raising this point. We have provided explanation on design concept in the revised MS text.

1. KOLF2 iPSC line is not introduced.  Is this a well-established line published before?  Provide references.

The KOLF2 iPSC line was generated by the Sanger Institute (https://hpscreg.eu/cell-line/WTSIi018-B). This line has been rigorously characterised with all information (eg. whole genome sequencing data) publicly accessible. Hence it serves as excellent reference iPSC line for research. We have included the website link (which include several publications about the line) in the revised MS section when the line was first introduced.

1. A timeline schematic of the DAN differentiation should be shown for clarity.

This information is now added in figure 2A.

 

2. Lack of BFP positive cells in homozygous knock in.  The authors point out that in CRE+/+ lines they get no BFP expression and propose that this is due to deleterious effect of the CRE knock in.  Are we to assume that in the CRE-/+ the negative effect is lost.  What has happened to LMX1A expression?  Is it replaced with CRE? CRE is not usually toxic to cells.  The authors should clarify what they propose is happening.  Additionally, levels of CRE and LMX1A protein should be examined. By the schematic it appears the last exon of LMX1A is replace with CRE.  This should also be better described.

As shown in figure 1A, the Cre coding sequence was inserted in frame immediately before the LMX1A stop codon downstream of P2A. Hence LMX1A expression is not predicted to be affected. While CRE is not usually toxic to cells, rare toxic effect can happen (PMID: 37692772). Since the heterozygous Cre line activates BFP expression and that we observed similar mDA differentiation efficiency, we took a pragmatic approach to focus our resource on progressing the planned work with the heterozygous lines. In responding the reviewer’s comment, we have edited the relevant MS text highlighting targeting design.

3. The supplemental figures were not included.

A single pdf file containing supplemental methods, supplemental figures tables was submitted to the system. We regret to hear that the reviewer couldn’t view this material.

 

4. The github link is not complete and so the code and data are not available for review. The data was also not made available for review.

We apologies for the inconvenience. Raw data access link was not required for submission. The deposition is now completed, and GEO accession number provided in the revised MS. Code is now available in GitHub (https://github.com/jmonzon87/LMX1A_iPSC_DA).

 

5. It seems odd that the FACS sorting BFP cells only increase the BPF+ population from ~55% to ~80%.  How do the authors explain so many BFP- cells going through their gating strategy?

As shown in figure S1C, we applied a high stringent gating strategy for sorting BFP⁺ cells, which were well separated from the BFP^- population. Post-sort flow cytometry monitoring of sorted BFP⁺ cells in independent studies confirmed the purity of sorted cells.

The lower BFP% referred were data obtained from BFP antibody staining. It is likely due to 1) low sensitivity of the antibody, 2) parameter setting criteria for the automated counting in CellProfiler so the number of BFP⁺ cells was underestimated.

6. Single cell sequencing characterization analysis:
7. It would be clearly if the author annotated their clusters from Figure 4 A before showing the number of cells from each time point in each cluster group in C.  As they have done in D.
8. It would be easier to understand the DGE contrast if the younger age was list before the older age in E.

We have made the edit as suggested by the reviewer.

 

1. The comparisons in G and H are pointless to me.  What do the authors want me to learn here?  PCA analysis is a dimensional reduction, and the clusters would be expected to segregate in PC space, but this doesn’t inform us about their characteristics.

The panels were meant to illustrate a similar temporal progression between human fetal midbrain and hiPSC-derived mDA cells. It was unfortunate that the panel concerning human fetal data was presented in figure S3 that the reviewer couldn’t access, making it hard to understand the main figure. We have now placed the human fetal panel in the revised figure 4G.

 

1. In panel I the velocity analysis is very interesting.  The authors should add labels to the UMAP for the cell types (Prog, DA, Astrocytes).

Added as requested.

1. The overlay density plots of the velocity output are not well described.  Are we looking at a selected set of gene expression?  A gene module?  Some kind of velocity score?  It would be more informative to show expression of DA markers or modules of markers over pseudo time.

We thank the reviewer for pointing this out. We have modified the results section and the figure legends to make this point clear. RNA velocity was calculated on the top 5000 most variable genes (information given on the supplementary material). We have additionally included a new plot in the supplementary material showing the average expression of dopaminergic markers along the pseudotime based on RNA velocity.

1. Also, it is not clear if the scRNAseq was done on the BFP sorted population of the whole population.

The first scRNAseq experiment (data shown in figure 4) was performed with the whole population at days 21, 30, 45 and 65. The second experiments was performed with three types of day 45 cultures: BFP⁺ fraction sorted at day 18, BFP^- fraction sorted at day 18, unsorted sister cultures grown in the same condition. We have gone through the revised MS making sure these information are visible.

1. In the MPP treatment experiments the authors show a decrease in BFP positive cells.  Are we to assume this means DANs are being selectively kill?  The author’s interpretation is not explained.

We have edited this part of the MS to improve clarity.

7. scRNAseq MPP treatment:
8. It would be useful to annotate the clusters in Figure 5F and H.

Annotation added as suggested.

1. Here the authors should analyze this data to identify characteristics of the cells lost or dying from the MPP treatment.  What are the characteristics of the surviving cells?  If DANs persist how are these different from the absent population?  They can compare to the cells in the basal condition. The authors have compared by each cluster the GO terms effected and apoptosis and mitochondrial health pathways affected, which is not very revealing.

We fully recognise the importance of the questions raised by the reviewer and had attempted to address some of these previously with little clear outcome.

The B3 population (mostly derived from d18 BFP positive sorted progenitors) show the largest reduction after MPP treatment, despite exhibiting few differences (n = 12 DEGs) to B1 (derivatives of both BFP sorted positive and non-sorted d18 progenitors). We thus during the short revision period focused on the B3 population. 78 DEGs were detected between the basal B3 and predicted B3 in the MPP⁺ condition. However, it is challenging to discern whether the DGEs detected were driven by cell identity or from their response to MPP⁺. We have in the revised MS discussed the limitations and potential approach to address this question in future studies.

 

Minor:

1. Figure 1 the legend has incorrect matching for the lettering in the figure and the panel descriptions.  1C is 1C and 1D, 1D is 1E.  The description for the pie charts D is not sufficient.  Figure 1E is not described in the legend.

Our apologies for the error. The figure legend is now updated.

2. Figure 2 the statistics are not indicated on the plots. Also, the number of images used and if n=field of view, well or experimental replicated is not indicated.

We use the same workflow for data collection and analysis concerning immunostaining. The information regarding replicate numbers was provided in the Immunofluorescence part of the Method.

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

no suggestions

Author Response

No further comments raised.

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have done an excellent job of addressing my previous comment. After reading the revised manuscript I have only a few minor concerns.

 

In comparison of cell type proportions with and without MPP in NS, BFP negative and BFP positive populations they give one CHI square statistic and say that the BFP+ DA neurons decrease with MPP treatment.  In Figure 5H there seems to be an increase in sorted BFP+ DA neurons in B1 cluster and a decrease in unsorted DA neurons. The description of the results doesn’t match the stack bar chart. Both unsorted and sorted BFP+ DA neurons decrease in cluster B3 as described. The text should be adjusted to reflect the plot in Figure 5H, or if the text is correct the data should be visualized in a different manner. Instead of a CHI square test the authors should perform permutation tests changes in proportions between basal and MPP treatment, grouped by cluster.  Here the authors will test if the decrease in the proportion of DA neurons in B3 is significant. Here is the type of test I mean (https://github.com/rpolicastro/scProportionTest).

 

The authors have cited the source of the KOLF2 line in the methods.  I think it would be good to site manuscript describing the generation of this line or the first publication using line in when it first appears on line 70 in the manuscript.

 

Comments on the Quality of English Language

There are many grammatical errors, for example line 54 “For example, tyrosine hydroxylase (TH)-based reporter system has been employed…” This should either be “a tyrosine hydroxylase (TH)-based… “ or  “TH-based reporter systems …”. There are many similar mistakes throughout.

 

line 41 – missing article  “these in vitro produce”

Line 44 – midbrain dopaminergic neurons should be mDA as it was abbreviated above.

Lines 46-52 – could be adjusted to read more smoothly and clearly.

 

I have not pointed out all the errors throughout the manuscript. There are also places where there are not mistakes but ease of reading could be improved. I recommended the manuscript be edited thoroughly. 

Author Response

We thank the reviewer for their helpful comments. Please see the point-by-point response below.

 

In comparison of cell type proportions with and without MPP in NS, BFP negative and BFP positive populations they give one CHI square statistic and say that the BFP+ DA neurons decrease with MPP treatment.  In Figure 5H there seems to be an increase in sorted BFP+ DA neurons in B1 cluster and a decrease in unsorted DA neurons. The description of the results doesn’t match the stack bar chart. Both unsorted and sorted BFP+ DA neurons decrease in cluster B3 as described. The text should be adjusted to reflect the plot in Figure 5H, or if the text is correct the data should be visualized in a different manner. Instead of a CHI square test the authors should perform permutation tests changes in proportions between basal and MPP treatment, grouped by cluster.  Here the authors will test if the decrease in the proportion of DA neurons in B3 is significant. Here is the type of test I mean (https://github.com/rpolicastro/scProportionTest).

 

We have performed the permutation test suggested by the reviewer and included the results of such test as a new supplementary figure (Figure S5). From the neuronal populations B3 is indeed significantly reduced in the MPP⁺ condition. The text is updated accordingly.

 

The authors have cited the source of the KOLF2 line in the methods.  I think it would be good to site manuscript describing the generation of this line or the first publication using line in when it first appears on line 70 in the manuscript.

A reference for the KOLF2 line is now added (No. 20).

There are many grammatical errors….

The errors listed by the reviewer has been corrected. Furthermore, the manuscript has been checked and edited by our native English speaker co-author.