Single-cell and single-nucleus RNA-seq uncovers shared and distinct axes of variation in dorsal LGN neurons in mice, non-human primates, and humans
Curation statements for this article:-
Curated by eLife
Summary: This manuscript provides a comparative analysis of the cell variety present in the dorsal lateral geniculate nucleus (dLGN) of mice, non-human primates, and humans using single-cell/single-nucleus RNA-sequencing (Smart-seq). The study identifies excitatory and inhibitory dLGN cell types in the three species and shows that the different subclasses of inhibitory neurons are relatively similar across species. In contrast, excitatory neurons appear to bear cross-species differences particularly between mouse and primates. The study provides an extensive description of the dLGN neurons, an important visual relay nucleus that has been so far poorly studied. As such, these data are very welcomed and will likely attract the interest of researchers working in visual function and beyond. The strong and creative bioinformatics analysis has uncovered interesting and subtle cross species links between different types of neurons.
Reviewer #1, Reviewer #2 and Reviewer #3 opted to reveal their name to the authors in the decision letter after review.
This article has been Reviewed by the following groups
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
- Evaluated articles (eLife)
Abstract
Abundant evidence supports the presence of at least three distinct types of thalamocortical (TC) neurons in the primate dorsal lateral geniculate nucleus (dLGN) of the thalamus, the brain region that conveys visual information from the retina to the primary visual cortex (V1). Different types of TC neurons in mice, humans, and macaques have distinct morphologies, distinct connectivity patterns, and convey different aspects of visual information to the cortex. To investigate the molecular underpinnings of these cell types, and how these relate to differences in dLGN between human, macaque, and mice, we profiled gene expression in single nuclei and cells using RNA-sequencing. These efforts identified four distinct types of TC neurons in the primate dLGN: magnocellular (M) neurons, parvocellular (P) neurons, and two types of koniocellular (K) neurons. Despite extensively documented morphological and physiological differences between M and P neurons, we identified few genes with significant differential expression between transcriptomic cell types corresponding to these two neuronal populations. Likewise, the dominant feature of TC neurons of the adult mouse dLGN is high transcriptomic similarity, with an axis of heterogeneity that aligns with core vs. shell portions of mouse dLGN. Together, these data show that transcriptomic differences between principal cell types in the mature mammalian dLGN are subtle relative to the observed differences in morphology and cortical projection targets. Finally, alignment of transcriptome profiles across species highlights expanded diversity of GABAergic neurons in primate versus mouse dLGN and homologous types of TC neurons in primates that are distinct from TC neurons in mouse.
Article activity feed
-
-
Reviewer #3:
In this manuscript, the authors utilize single-cell/single-nucleus RNA-sequencing to perform a comparative analysis of the cellular composition of the dorsal lateral geniculate nucleus (dLGN) in mice, non-human primates, and humans. This topic is important for a number of reasons, including (1) the dLGN is a critical center of visual processing about which we know relatively little; (2) the dLGN has emerged as a widely used experimental model of neural circuit development; and (3) in general, the integration of cross-species data at the transcriptomic level is important for identifying conserved mechanisms of brain function that may shed light upon neurological disease states. By employing a relatively deep RNA-sequencing approach (Smart-Seq) the authors identify major excitatory and inhibitory dLGN cell types within each …
Reviewer #3:
In this manuscript, the authors utilize single-cell/single-nucleus RNA-sequencing to perform a comparative analysis of the cellular composition of the dorsal lateral geniculate nucleus (dLGN) in mice, non-human primates, and humans. This topic is important for a number of reasons, including (1) the dLGN is a critical center of visual processing about which we know relatively little; (2) the dLGN has emerged as a widely used experimental model of neural circuit development; and (3) in general, the integration of cross-species data at the transcriptomic level is important for identifying conserved mechanisms of brain function that may shed light upon neurological disease states. By employing a relatively deep RNA-sequencing approach (Smart-Seq) the authors identify major excitatory and inhibitory dLGN cell types within each species. While the multiple inhibitory neuron subtypes were relatively similar across species, excitatory neurons displayed major differences particularly between mouse and both primate classes. The authors identified four major excitatory cell types in primate and human dLGN corresponding with known functional heterogeneity that places these neurons into magnocellular, parvocellular, and koniocellular populations. Interestingly, koniocellular neurons could be broken into two distinct subtypes. Somewhat surprisingly, the authors noted a lack of excitatory neuron diversity in the mouse, despite prior evidence suggesting these neurons can have different morphological and physiological features. Yet, although all excitatory neurons in the mouse clustered together, there were subtle differences in excitatory neurons in the mouse that aligned with different regions of mouse dLGN (shell vs core), suggesting that excitatory neuron heterogeneity may still exist along a more subtle continuum. Consistently, neurons in the shell region in mouse dLGN more strongly resembled koniocellular neurons in primates versus the core region, suggesting some level of conservation between excitatory neuron identity across species. While the study is largely descriptive, the authors are creative in their use of bioinformatics to uncover particularly interesting observations that the transcriptomic analysis yielded, and the paper is very interesting because of that. The major weakness of the paper is a paucity of robust FISH analyses to quantitatively validate the transcriptomic findings in all species. Overall, it is my opinion that this work is very important and that, at a broader level, it may help to define the relationship between transcriptomic cell type, functional/physiological cell type, and anatomical cell type within a brain region that is critical for visual function and that has emerged as a fascinating model of neural circuit development in the mouse.
Strengths:
The Smart-Seq transcriptomic technique chosen is appropriate to address the authors' questions.
The data were generated rigorously and subjected to an in-depth quality control pipeline prior to analysis. As a result, the quality of the transcriptomic data is high.
The paper includes a detailed, transparent description of the approach taken in the Results and Methods. The authors point out caveats and weaknesses - and how they were addressed - throughout the text.
The inclusion of tissue from thalamic nuclei surrounding the dLGN as a way to control for the unintentional inclusion of non-dLGN tissue in the experimental dissection was well-designed and effective.
Despite a couple of exceptions, the authors do an excellent job of placing their findings within the context of what is already known about dLGN cell types across different species, and how these cell types function differently in physiological, morphological, and anatomical terms.
The study is descriptive in nature but the authors do a nice job of laying out several interesting findings, such as the observation that GABAergic neurons are more conserved across species than relay neurons, with mouse neurons being particularly distinct. Another fascinating observation is that shell-located neurons in mouse dLGN are transcriptomically related to koniocellular neurons suggesting the possibility of a close relationship between thalamocortical connectivity and molecular identity across species.
Weaknesses:
The characterization of gene expression patterns through sequencing-based transcriptomics has emerged as a powerful tool for dissecting the brain, but it is important to couple such approaches with techniques like fluorescence in situ hybridization (FISH) to verify sequencing results in a histological context. While here the authors show 3 - 4 validations of mouse genes that seem to be restricted to or excluded from the shell versus the core dLGN regions (Figures 4G and S4E), the conclusions of the study would be better supported by a more extensive and rigorous analysis of cell-type-specific gene expression within all species described.
It is not entirely clear from the manuscript how the authors dissected the shell from the core region of the dLGN, given these regions are not as clearly distinct as the dLGN lamina in other species. One possibility would be to take advantage of the fact that the shell receives input from specific RGCs that can be targeted genetically by crossing a Cre driver to the TdTomato line, but I do not believe that that is what was done here. I also noted that the authors use ventral LGN (vLGN) as one of their controls for the precision of their micro-dissections, but given that the vLGN does not directly contact the dLGN, this had me wondering exactly how cleanly the shell and core regions of the mouse dLGN were isolated.
On lines 101 - 103, the authors state "...differentially expressed genes between donors were related to neuronal signaling and connectivity and not to metabolic or activity-dependent effects." Table S2 is cited, but the columns are not labeled such that a common reader could interpret them and confirm the statement in the text. Moreover, the text does not state how the authors made the determination that these differentially expressed genes are not related to "activity-dependent effects".
-
Reviewer #2:
The conclusion was quite surprising from their anatomical differences and connectivity to the cortex, however, implies different mechanisms underlie for species specific circuit organization.
The manuscript is well-organized and well-written with strong figures. I have only a few comments/suggestions to further improve the overall quality of this manuscript.
I understand obtaining human and NHP tissue is difficult and hard to perform numbers of ISH. Therefore, there is a database that provides additional information on gene expression in NHP LGN (https://gene-atlas.brainminds.riken.jp/). From this database, it is possible to obtain parvocellular specific and magnocellular specific gene expression by fine structure search, which may be worth comparing with the results in the current paper. Many researchers have realized that …
Reviewer #2:
The conclusion was quite surprising from their anatomical differences and connectivity to the cortex, however, implies different mechanisms underlie for species specific circuit organization.
The manuscript is well-organized and well-written with strong figures. I have only a few comments/suggestions to further improve the overall quality of this manuscript.
I understand obtaining human and NHP tissue is difficult and hard to perform numbers of ISH. Therefore, there is a database that provides additional information on gene expression in NHP LGN (https://gene-atlas.brainminds.riken.jp/). From this database, it is possible to obtain parvocellular specific and magnocellular specific gene expression by fine structure search, which may be worth comparing with the results in the current paper. Many researchers have realized that marmoset is one of the good animal models to understand human brain function and dysfunction, therefore, it is worth including marmoset for comparative analysis for community interest.
-
Reviewer #1:
In this manuscript, Bakken et al use single cell and single nucleus RNA-sequencing to conduct comparative analysis of dLGN in humans, macaques and mice. dLGN exhibits a dramatic reorganization and lamination in primates relative to mice. Other components of the visual system (retina, V1) have previously been explored with cross-species transcriptomic analyses to reveal species-specific or evolutionary modifications. How dLGN fits in this picture, and the extent to which differences amongst previously identified cell types can be discerned from transcriptomic data, is an important question.
The conclusions are supported by the data, but the paper could better motivate what the main questions or debates are.
Strengths:
The authors use highly sensitive SMART-seq v4 to collect and analyze thousands of cells from dLGN and some …
Reviewer #1:
In this manuscript, Bakken et al use single cell and single nucleus RNA-sequencing to conduct comparative analysis of dLGN in humans, macaques and mice. dLGN exhibits a dramatic reorganization and lamination in primates relative to mice. Other components of the visual system (retina, V1) have previously been explored with cross-species transcriptomic analyses to reveal species-specific or evolutionary modifications. How dLGN fits in this picture, and the extent to which differences amongst previously identified cell types can be discerned from transcriptomic data, is an important question.
The conclusions are supported by the data, but the paper could better motivate what the main questions or debates are.
Strengths:
The authors use highly sensitive SMART-seq v4 to collect and analyze thousands of cells from dLGN and some adjacent nuclei. The gene detection rate using this method is impressive, and the plate/strip-based workflow has distinct advantages in terms of lower ambient contamination and risk of doublets compared to microfluidics-based single cell platforms. Cells or nuclei are sorted to enrich for neurons, which are the main focus of this paper. Key results are validated by smFISH or by examining publicly available Allen Brain Atlas ISH data. By examining conservation and divergence of cell types and evolutionarily conserved thalamic nucleus that has nonetheless undergone dramatic anatomical reorganization, these data and analyses add to our understanding of how cell types evolve in mammalian brains. They also contribute nuance to the ongoing debate of the extent to which transcriptomic data alone can be used to identify and discriminate cell types that have been described using other methodologies.
Weaknesses:
The Introduction does a nice job of describing what is known about the anatomy and cell types of the dLGN in each species, but it is less obvious what the motivating cross-species question is. Similarly, the Discussion focuses on technical details but the take-away is not clear.
dLGN is collected from all species, but in some species (macaque, mouse), additional thalamic nuclei are also collected. These are useful for examining cell type correspondences across regions or shifts between species, but their inclusion in cross-species integrations can also distort results (e.g. with some integration approaches, inclusion of very different, dataset-specific cell types can distort integration of more similar types). Analyses could be done to better distinguish the evolutionary comparisons within dLGN itself vs. what is additionally learned from inclusion of extra-dLGN nuclei.
One major evolutionary difference can involve differences in cell type proportions. Some proportion results are described but mainly for individual species (some of which include extra-dLGN regions) rather than in the integrated maps, so they can't be compared across species. The FISH results could also be used to corroborate proportion changes when such data are available.
Parameters for clustering analysis (using CCA/Seurat) are not described. Often changes in parameters can change the clusters, and it would be important to know if species integration results robust across a range of parameters and inclusion of extra-dLGN regions.
Some expected genes (PVALB) are barely detected in the macaque neurons, raising the question of whether this is due to tissue or annotation/alignment quality.
-
Summary: This manuscript provides a comparative analysis of the cell variety present in the dorsal lateral geniculate nucleus (dLGN) of mice, non-human primates, and humans using single-cell/single-nucleus RNA-sequencing (Smart-seq). The study identifies excitatory and inhibitory dLGN cell types in the three species and shows that the different subclasses of inhibitory neurons are relatively similar across species. In contrast, excitatory neurons appear to bear cross-species differences particularly between mouse and primates. The study provides an extensive description of the dLGN neurons, an important visual relay nucleus that has been so far poorly studied. As such, these data are very welcomed and will likely attract the interest of researchers working in visual function and beyond. The strong and creative bioinformatics analysis …
Summary: This manuscript provides a comparative analysis of the cell variety present in the dorsal lateral geniculate nucleus (dLGN) of mice, non-human primates, and humans using single-cell/single-nucleus RNA-sequencing (Smart-seq). The study identifies excitatory and inhibitory dLGN cell types in the three species and shows that the different subclasses of inhibitory neurons are relatively similar across species. In contrast, excitatory neurons appear to bear cross-species differences particularly between mouse and primates. The study provides an extensive description of the dLGN neurons, an important visual relay nucleus that has been so far poorly studied. As such, these data are very welcomed and will likely attract the interest of researchers working in visual function and beyond. The strong and creative bioinformatics analysis has uncovered interesting and subtle cross species links between different types of neurons.
Reviewer #1, Reviewer #2 and Reviewer #3 opted to reveal their name to the authors in the decision letter after review.
-
