Analytic Summary of a Peer-Reviewed IMRAD Article

Article: Combined single-cell and spatial transcriptomics reveal the molecular, cellular and spatial bone marrow niche organization

Authors: Chiara Baccin, Jude Al-Sabah, Lars Velten, Patrick M. Helbling, Florian Grünschläger, Pablo Hernández-Malmierca, César Nombela-Arrieta, Lars M. Steinmetz, Andreas Trumpp, and Simon Haas

Published: Nature Cell Biology, January 2020

Introduction

The introduction begins with an explanation of what bone marrow (BM) niches are (specialized microenvironments of cells that regulate different processes in bone marrow), and how they have previously been studied (using single markers). Once this background is introduced, the authors move to the limitations of these approaches, which include inadequate differentiation of heterogeneous populations of cells as a result of imprecise cellular markers. Finally, the introduction ends with an explanation of how the authors have overcome these limitations in the current study. In this case, the authors combine single-cell RNA sequencing (scRNAseq) with computational spatial analysis to understand how bone marrow cells are organized into niches with specific defining factors.

Results

The first thing the authors investigated was the ability of scRNAseq to accurately characterize bone marrow cells types. They used droplet-based scRNAseq on 7,497 cells, which were divided into 32 clusters of different cell types and stages of differentiation. This method was tested over a variety of bone marrow preparations and it was found that more extensive digestion of the bone marrow resulted in more abundant cell identification. The scRNAseq not only verified this method for the identification of most known BM cell types but also separated two previously unknown populations.

Once the identity of cells in bone marrow was determined, the authors moved on to integrating spatial information using a combination of laser-capture microdissection and sequencing (LCM-seq). This approach overcomes the limitations of low-quality input RNA and allowed for the localization of cell clusters to distinct niches. In order to validate the cell cluster localization, the authors used previously published cell marker combinations to tag specific clusters and immunofluorescent staining on bone sections to visualize the locations of the distinct clusters.

Next, the authors developed an algorithm called RNA-Magnet, which uses the expression patterns of cell-surface receptors and their binding partners to predict potential physical interactions between single cells. They found that RNA-Magnet predicted higher adhesiveness of clusters to niches where they were more localized. Overall, this showed that RNA-Magnet can predict spatial information from single-cell RNA sequencing data.

Once the authors established that different cell types are preferentially found in certain locations, they used RNA-Magnet to visualize the cell-to-cell communication within the bone marrow. They found that cells primarily communicate within their own groups, and it is the combination of signaling from different niches that drive biological processes.

Much of the results in this paper show the proportions of cells in different locations and are represented by scatter plots, bar graphs, and spatial clustering graphs. There is not as much statistical analysis, except to show that the number of cells in a given location is significantly greater than other cells in that same location.

Discussion

In the discussion section, the authors summarize the general methodology and findings, reiterating that they combined single-cell data with spatially resolved transcriptomics data to map all of the cell types in bone marrow and assign them to specific niches. Two new subsets of cells were identified to be separate from previously identified cell types. Finally, the authors discuss how the data supports a new understanding of cellular niches wherein different cell types attach to cellular scaffolding in specific areas and communicate largely within their cell groups. Looking into the future, the authors postulate that this technology could be applied to other tissue and organ systems, as well as to understanding the role niches play in differentiating stem cells.

Methods

The methods described in this paper can be divided into three broad categories; mouse type and tissue harvesting, sequencing (including scRNAseq using 10x Genomics, FACS-indexed scRNAseq, LCM-seq, and bulk RNA-seq), and data analysis and visualization. Analysis methods describe how they thresholded cell numbers to ensure they were getting accurate data that did not include fragmented or doubled cells. The methods also list all of the packages used in the scatter plots and clustering plots, as well as in any other data visualization. The data and analysis code are available to view or download freely, along with vignettes for recreating some of the analysis steps.

Article Analysis

Is the topic of the paper somewhat original?

The topic of this paper is original in multiple ways, as there had not been another group to tackle the spatial organization of cell niches in bone marrow before this paper. The authors used methods that had not been previously used together for the analysis of bone marrow. Namely, they created a new program, RNA-Magnet, specifically to achieve this. While this part of the research is very original, the paper still draws heavily on previous research, especially in the ways that the authors identify and cluster cell types. Cell sorting and clustering were done using methods that are very common in molecular biology; the synthesis of the information gathered from that was the original part of this research. RNA-Magnet was developed to predict the location of different cell populations based on the specific expression of cell adhesion molecules and the way cells interact with each other through those molecules.

Do the authors have a solid track record?

Andreas Trumpp is the Head of Division "Stem Cells and Cancer" and Managing Director of HI-STEM, the Heidelberg Institute for Stem Cell Technology and Experimental Medicine. One of the group leaders, Simon Haas, directs a lab specializing in Systems Hematology, Stem Cells and Precision Medicine. The other authors on this paper are postdocs and graduate students in these or related groups at the European Molecular Biology Laboratory and Stanford University. Dr. Trumpp has an h-index of 73, and Dr. Haas has an h-index of 16. The h-index is a measure of productivity and impact. A higher h-index indicates a higher number of papers that have been highly cited. Clearly, Dr. Trumpp is very influential in his field and while Dr. Haas has a much lower h-index, he is much earlier in his career and seems to be on an upward trajectory in terms of high-impact publications. These authors have a very solid track record, and it is likely that their graduate students and postdocs do as well, though with fewer papers it is more difficult to find that for certain.

What was the aim of the study? What hypothesis did the researchers test? Are the conclusions reached (assuming they are valid) important to you and others (explain)?

The aim of this study was to investigate the molecular, cellular, and spatial organization of bone marrow (BM) cells. The organization of bone marrow has not been well understood and different ideas have circulated in the field. This study is the first step towards creating consensus about the roles of different cell types in BM. I believe that the understanding generated by this study is very important, not necessarily to the public but to the scientific community. Personally, the methods used in this study are of great interest to me and the results that they were able to obtain are important in that they validate the methods.

Were enough data obtained to reach valid conclusions?

The best ways to ensure statistical significance are to have a large dataset or do multiple repetitions. With a dataset of 7,497 cells, I believe that the first condition of a large dataset is met, thus making it more likely that the conclusions were valid. More than that, conclusions were validated in multiple ways, using methods that are widely accepted in this field.

Do the Results section and the Methods section match?

The Results section and the Methods section do match. All of the processes used in this paper were carefully detailed in the Methods section and the data and code were also available through links in the Methods section. All of the parameters set for the experiments were described, including the parameters used in the analysis programs. For example, the authors write that “Clustering was performed using the default method from the Seurat package, with the resolution parameter set to five.” This is very clear and makes the results in this paper easily reproducible because anyone can follow exactly what the authors did, including using the data the authors generated.

Have the authors discussed possible limitations of the study?

The authors addressed possible limitations of the study mainly by predicting how this technology may be used in the future to create an even deeper understanding of bone marrow niches and cellular organization. Specifically, the authors write that “In the future, it will be of interest to investigate whether such extrinsic, niche-driven variations determine early fate decisions of hematopoietic stem cells.” While this study showed that niche-specific variation certainly does exist in bone marrow and the niches can influence the “behavior” of the cells that exist in those niches, the research did not necessarily determine a causal relationship. A more thorough investigation into the influence of niche environments on cell-specific gene expression is needed to show how the niches regulate hematopoietic activities.

Do the study’s findings have practical importance, regardless of whether they have statistical significance?

The study’s findings do have practical importance, for the scientific and medical communities especially. Scientifically, this study reveals new basic information, which is the goal of any scientific inquiry, and also creates a framework for other groups to follow to understand the organization of any other tissue or organ. Medically, the conclusions of this research are important because understanding BM is the first step towards creating therapies that treat BM diseases.