Molecular and spatial transcriptomic classification of midbrain dopamine neurons and their alterations in a LRRK2^G2019S model of Parkinson’s disease

Reviewer #1 (Public review):

Summary:

Dopamine neurons contribute to motivated and motor behaviors in many ways, and ample recent evidence has suggested that distinct dopamine neuron subclasses support discrete behavioral and circuit functions. Prior studies have subdivided dopamine neurons by spatial localization, gene expression patterns, and physiological properties. However, many of these studies were bound by previous technical limitations that made comprehensive subclassification efforts difficult or impossible. The main goal of this manuscript was to characterize and further define dopamine neuron heterogeneity in the ventral midbrain. The study uses cutting-edge single nucleus RNA-seq (on the 10X Genomics platform) and spatial transcriptomics (on the MERFISH platform) to define dopamine neuron heterogeneity with unprecedented resolution. The result is a convincing and comprehensive subclassification of dopamine neurons into three main families, each with major branches and subtypes. In addition, the study reports comparisons between wild-type mice and mice that harbor a G2019S mutation in the Lrrk2 gene, which models a common cause of autosomally dominant Parkinson's Disease in humans. These results, while less robust due to the nature of the group comparisons, nevertheless identify vulnerability within specific dopamine neuron subpopulations. This vulnerability may contribute unique risk of dopamine neuron loss in the context of Parkinson's disease. Overall, the study is careful and rigorous and provides a critical resource for the rapidly evolving knowledge of dopamine neuron subtypes.

Strengths:

(1) The creation of a public-facing app where the snRNA-seq data can be investigated by anyone is a major strength.

(2) The manuscript includes careful comparisons to prior datasets that have sought to explore dopamine neuron heterogeneity. The result is a useful synthesis of new findings with previously published work, which is helpful for moving the field forward in this area.

(3) The integration of snRNA-seq with MERFISH results is particularly strong and enables insight not only into subclassification but also into how this relates to spatial localization. The careful neuroanatomy reveals important distinctions between Sox6, Calb1, and Gad2 positive dopamine neuron families, with some degree of spatial overlap.

Weaknesses:

(1) Important details about the nature of DEG comparisons between the wild type and the Lrrk2 G2019S model are missing.

(2) Some aspects of the integration between snRNA-seq and MERFISH data are not clear, and many MERFISH-identified cells do not appear to have a high-confidence cluster transfer into the snRNA-seq data space. Imputation is used to overcome some issues with the MERFISH dataset, but it is not clear that this is appropriate.

Reviewer #2 (Public review):

Gaertner and colleagues present a study examining the transcriptomic diversity and spatial location of dopaminergic neurons from mice and examine the changes in gene expression resulting from knock-in of the Parkinson's LRRK G2019S risk variant. Overall, I found the manuscript presented their study very clearly, well written with very clear figures for the most part. I am not an expert on mouse neuroanatomy but found their classification reasonably well justified and the spatial orientation of dopaminergic neurons within the mouse brain informative and clear. While trends were clear and well presented, the apparent spatial heterogeneity suggests that knowledge of the functional connections and roles of these neurons will be required to better interpret the results presented, but nonetheless their findings exposed significant detail that is required for further understanding.

The study of the transcriptional effects of the LRRK2 KI was also informative and clearly framed in terms of a focused analysis on the effects of the KI only on dopaminergic neurons. However, I think there are issues here in both methodology, narrative, and clarity.

(1) In the GO pathway analyses (both GSEA and DEG GO), I did not see a correction applied to the gene background considered. The study focusses on dopaminergic neurons and thus the gene background should be restricted to genes expressed in dopaminergic neurons, rather than all genes in the mouse genome. The problem arises that if we randomly sample genes from dopaminergic neurons instead of the whole genome, we are predisposed to sampling genes enriched in relevant cell-type-specific roles (and their relevant GO terms) and correspondingly depleted in genes enriched in functions not associated with this cell type. Thus, I am unsure whether the results presented in Figures 8 and 9 may be more likely to be obtained just by randomly sampling genes from a dopaminergic neuron. The background should be limited and these functional analyses rerun.

(2) In the scRDS results, I am unsure what is significant and what isn't. The authors refer to relative measures in the text ("highest") but I do not know whether these differences are significant nor whether any associations are significantly unexpected. Can the x-axis of scRDS results presented in Figure 9 H and I be replaced with a corrected p-value instead of the scRDS score?

(3) The results discussed at the bottom of page 13 state that 48.82% of the proteins encoded by the Calb1 DEGs have pre-synaptic localisations as opposed to 45.83% of the SOX6 DEGs, which does not support the statement that "greater proportions of DEGs are associated with presynaptic locations in cells from vulnerable DA neurons (Sox6 family, [and in particular,Sox6^tafa1]), compared to less vulnerable ones (Calb1 family)".

(4) While an interest in the Sox6^tafa1 subtype is explained through their expression of Anxa1 denoting a previously identified subtype associated with locomotory behaviours, it was unclear to me how to interpret the functional associations made to DEGs in this subtype taken out of context of other subtypes. Given all the other subtypes, it is not possible to ascertain how specific and thus how interesting these results are unless other subtypes are analysed in the same way and this Sox6^tafa1 subtype is demonstrated as unusual given results from other subtypes.

(5) On p12, the authors highlight Mir124a-1hg that encodes miR-124. This is upregulated in Figure 8D but the authors note this has been to be downregulated in PD patients and some PD mouse models. Can the authors comment on the directional difference?

(6) Lastly, can the authors comment on the selection of a LogFC cut-off of 0.15 for their DEG selection? I couldn't see this explained (apologies if I missed it).