Characterizing the Heterogeneity of Liver Cell Populations Under a NASH-Related Hepatotoxicant Using Single-Nuclei RNA Sequencing

Single-cell transcriptomic technologies constitute nowadays a powerful and robust tool to decode tissue heterogeneity and study gene expression at the individual cell level, constituting an essential approach to unravel cell-specific transcriptomic changes under pathological conditions as well as in response to exogenous agents, therefore paving the path to unravel the role of specific cell populations, mechanisms, and molecular events. Although single-cell RNA sequencing (scRNA-seq) provides robust and reliable data, some bias might be added due to cell isolation–induced transcriptional changes, while some cell types are more difficult to collect, being more vulnerable during the tissue dissociation process and therefore less represented in the final dataset. Furthermore, scRNA-seq requires freshly isolated cells, representing an important technical drawback. Recovery of RNA from isolated nuclei obtained from frozen tissue has already been described, presenting enough quality and sufficient amount to conduct transcriptomic profiling studies.¹ Therefore, single-nuclei RNA-seq (snRNA-seq) is now emerging as a valuable alternative method to measure single-cell transcriptomic profiles using isolated nuclei.

Characterizing the cell-specific effects of hepatotoxicants has been challenging, and most of the data have been obtained by bulk RNA-seq data from total tissue. The environmental toxin and aryl hydrocarbon receptor agonist TCDD (2,3,7,8-tetrachlorodibenzo-p-dioxin) is known to promote steatosis and progression to nonalcoholic steatohepatitis and fibrosis in mice and is regarded as a potential contributing factor in the etiology of metabolic liver diseases although its cell-specific effects are yet to be unveiled. In this issue of Cellular and Molecular Gastroenterology and Hepatology, Nault et al² used snRNA-seq transcriptomic profiling of frozen liver samples in order to determine TCDD-induced cell-specific alterations in mouse liver. A total of 16,015 nuclei were isolated from livers of control and TCDD-administered mice, allowing the identification of 19,907 gene transcripts. After integration and clustering of nuclear transcriptomic profiles, 11 different clusters were observed, from which the “neutrophil cluster” was exclusively found in the TCDD group. Furthermore, TCDD altered the relative proportion of cell-specific nuclei clusters, increasing the frequency of inflammation-related clusters (e.g., macrophages, B cells, T cells), being in accordance with the reported proinflammatory effect of TCDD. A decrease in the relative proportion of hepatocytes (midcentral, portal, and midportal) was observed in TCDD samples, apart from central hepatocytes, which were shown to present a slight increase. By comparing these changes with differential gene expression in bulk RNA-seq tissue from this same model, the authors confirmed the impact of TCDD on specific cell population shifts. In addition, cholangiocytes presented the lowest number of differentially expressed genes (n = 122), while the expression of 7625 genes was altered in midcentral hepatocytes. On the one hand, Gene Ontology analysis revealed that RAS signaling and related pathways were mostly enriched in nonparenchymal cells after TCDD administration. On the other hand, metabolic processes were essentially associated with hepatocytes, mainly including a repression of cholesterol and triglyceride synthesis and an increase in xenobiotic detoxification pathways. Finally, analysis of macrophage nuclei allowed the identification of 5 distinct clusters, and importantly, a Kupffer cell subtype characterized by high expression of Gpnmb was identified solely in TCDD livers, mirroring previous observations from a diet-induced nonalcoholic steatohepatitis mouse model.

Up to date, no snRNA-seq studies on liver tissue have yet been published, therefore highlighting the novelty of this work and the usefulness of this approach for stored frozen samples from patients with liver diseases. Overall, performing a RNA-seq transcriptomic analysis of single nuclei isolated from frozen liver tissue allowed the identification of distinct hepatic cells, cell population shifts, and abundances, and discriminated specific pathways related to TCDD, which were further corroborated in bulk RNA-seq tissue data. The approach used by Nault et al to evaluate the hepatic effects of this hepatotoxicants present some advantages over single-cell analysis. First, using frozen samples to conduct this type of study (over the need to use freshly isolated cells for scRNA-seq) may reduce the number of animals in preclinical drug or chemical toxicity assessments by the use of stored frozen samples. Second, nuclei isolation is easily performed (even from tissues and organs), thus avoiding the complex and time-consuming steps of single-cell analysis and the obtained results may be comparable to the ones resulting from single-cell approaches. In fact, snRNA-seq was shown to provide high concordance rate,^3,4 or even improved⁵ transcriptome profiling when compared with scRNA-seq, allowing the identification of gene expression profiles associated with specific diseases. Still, considering that a high proportion of the RNAs are found in the cytoplasm, some information might be lost when analyzing single nuclei. In this regard, only ∼67% of the protein coding genes were detected in single nuclei,^1,4 raising some concerns in the impact of these procedures. Furthermore, when comparing snRNA-seq and scRNA-seq data, some cell populations were underrepresented in the former.⁶ Moreover, snRNA-seq libraries are often subject to contamination with surrounding RNA, which might introduce bias to the data. Employing debris identification using expectation maximization⁷ in order to quantify the eventual contamination of samples as well as isolating nuclei following a novel rapid enzyme and detergent-free column-based nuclei isolation method⁸ will certainly increase the robustness of this approach. Additionally, the cell-specific spatial organization is not being taken into account in this type of approaches and need further improvements in the future. Finally, considering all the strengths and limitations (Figure 1), the findings reported in this study should be corroborated by scRNA-seq to confirm the specific liver cell alterations.

Figure 1.

Strengths, limitations and future perspectives of liver snRNA-seq upon TCDD exposure. Created with BioRender.com. DIEM, debris identification using expectation maximization; NASH, nonalcoholic steatohepatitis.

Detecting changes in cell abundances and population shifts is of extremely importance when comparing different disease stages (eg, normal vs premalignant vs tumor tissues) and should be taken into consideration in future analysis. Although constituting a major challenge nowadays, integrating and combining diverse omics-based approaches at the single-cell level might be the way to clearly understand TCDD (or other hepatotoxicants) exposure consequences. Overall, snRNA-seq might provide novel insights into nuclear-specific regulatory mechanisms and will specifically allow the differential identification of transcripts usually found in the nucleus, such as long noncoding RNAs.