Response to: “Merit of integrating in situ transcriptomics and anatomical information for cell annotation and lineage construction in single-cell analyses of Populus”

Shaoming Liang, Yiling Li, Yang Chen, Heng Huang, Sijia Li, Yuanzhong Jiang, Tao MaGenome 2024 · DOI · OpenAlex

Abstract

This study integrates five independent single-cell RNA sequencing (scRNA-seq) datasets of Populus to address inconsistencies in wood formation lineage mapping. By harmonizing data from 62,256 cells, the authors identify conserved marker genes, propose a "cambium region" definition to account for marker plasticity, and highlight technical biases inherent in protoplast-based sequencing of woody tissues.

Introduction

In recent years, single-cell transcriptomics has been employed to map the developmental lineage of wood formation in Populus (poplar) trees. However, high-profile studies have often reached conflicting conclusions regarding the specific genes that identify cell types and the transitions from stem cells to mature wood. These discrepancies are likely driven by biological variation across species and growth stages, as well as the technical noise associated with extracting RNA from thick-walled woody cells. This paper serves as a meta-analysis, integrating disparate datasets to find a consensus and identify the root causes of previous misalignments.

Methods

The authors performed a large-scale data harmonization of raw scRNA-seq data from five major studies (Li et al. 2021, Chen et al. 2021, Xie et al. 2022, Tung et al. 2023, and Li et al. 2023). One study (Du et al. 2023) was excluded due to a lack of necessary spatial information. Expression matrices were integrated based on 1:1 orthologous relationships to match equivalent genes across different poplar species. Using standard bioinformatic pipelines, 62,256 cells were grouped into 21 distinct clusters. To validate these "virtual" clusters against physical tissue, the authors compared them with Laser Capture Microdissection (LCM) data, which provides spatial context by sequencing physically isolated tissue areas.

Key Results and Insights

The resulting integrated atlas, now available as a unified resource on Zenodo, comprises 62,256 cells with a median of 1,730 expressed genes per cell. A primary finding is the significant "plasticity" of marker genes; while genes like XCP1 and LAC17 are consistent across all samples, others such as WPP2 and Expansin A6 showed no specific pattern in the integrated dataset. This suggests that cell identification based on a limited set of genes is unreliable.

Furthermore, the study revealed that classic cambium markers like PXY and WOX4 are expressed across a wide range of clusters. Consequently, the authors recommend defining a "cambium region" or "zone" rather than attempting to pinpoint a single cambium cell type until more specific markers are identified. Correlation analysis also showed that LCM-collected vessel cells often correlate with multiple transcriptomic clusters, indicating that LCM may capture mixtures of cells at various developmental stages.

Integrated scRNA-seq atlas and marker gene expression across five Populus datasets. (A) UMAP plots of integrated and individual datasets. (B) Cluster proportions and dot plots showing expression of marker genes for xylem, cambium, and ray regions. (C) Correlation of clusters with LCM data.

Discussion and Future Directions

A major limitation identified in current wood scRNA-seq workflows is "protoplast bias." Because wood cells have varying wall thicknesses, the process of removing cell walls likely biases data toward certain cell types, distorting the true composition of the stem. The authors suggest that single-nuclei RNA-seq (snRNA-seq)—which avoids the need for total cell wall dissolution—would provide a more accurate evaluation. Future work must focus on identifying more "conserved" markers that remain stable across different species and developmental stages to better distinguish between similar cell types like fibers and vessels.

Summary updated

BibTeX
@inproceedings{liang2024response,
  author = {Shaoming Liang and Yiling Li and Yang Chen and Heng Huang and Sijia Li and Yuanzhong Jiang and Tao Ma},
  title = {{Response to: “Merit of integrating in situ transcriptomics and anatomical information for cell annotation and lineage construction in single-cell analyses of Populus”}},
  booktitle = {Genome},
  year = {2024},
  doi = {10.1186/s13059-024-03228-4},
}

Loading related content...

Figure