How single-cell transcriptomics provides insight on hepatic responses to TCDD

Abstract

The prototypical aryl hydrocarbon receptor (AHR) ligand, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), has been a valuable model for investigating toxicant-associated fatty liver disease (TAFLD). TCDD induces dose-dependent hepatic lipid accumulation, followed by the development of inflammatory foci and eventual progression to fibrosis in mice. Previously, bulk approaches and in vitro examination of different cell types were relied upon to study the mechanisms underlying TCDD-induced liver pathologies. However, the advent of single-cell transcriptomic technologies, such as single-nuclei RNA sequencing (snRNAseq) and spatial transcriptomics (STx), has provided new insights into the responses of hepatic cell types to TCDD exposure. This review explores the application of these single-cell transcriptomic technologies and highlights their contributions towards unraveling the cell-specific mechanisms mediating the hepatic responses to TCDD.

1. Introduction

Environmental exposures are increasingly being linked to the development and progression of fatty liver disease (FLD), commonly referred to as toxicant-associated fatty liver disease (TAFLD) [1, 2]. To better understand the underlying molecular mechanisms involved in TAFLD, significant efforts have been directed towards investigating diverse environmental contaminants including the potent aryl hydrocarbon receptor (AHR) agonist 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) and related dioxin-like compounds [3–5]. For decades TCDD has been shown to induce hepatic lipid accumulation (steatosis), inflammation (steatohepatitis), and the deposition of extracellular matrix (fibrosis) in mice making it an important model for the investigation of TAFLD [4–6].

Toxicogenomic data including transcriptomics and metabolomics have played a critical role in increasing our understanding of TCDD-induced TAFLD [3, 7, 8]. While hepatocytes are clearly implicated as the most abundant liver cell type, other hepatic cell types such as immune cells and stellate cells are also involved in TAFLD progression. However, bulk analyses have limited ability to distinguish the intricate molecular events occurring within and between cell (sub)types of the liver in their endogenous environment (Figure 1). Innovative in vitro models such as spheroids and microphysiological systems address some of these limitations [9], but to accurately capture the interplay of systemic factors within the liver microarchitecture, single-cell transcriptomics is among the most effective approaches available today.

Figure 1.

Illustrative example of the advantages of using single-cell transcriptomics to investigate toxicant mediated liver responses. Single-cell transcriptomics can provide novel insight into the number and/or expression of cell-specific genes which are often indistinguishable using bulk transcriptomics. Moreover, rare cell types whose genes may not be detected in bulk transcriptomic analysis can be detected enabling investigation of the responses in unique (sub)types. Created in BioRender.com.

The emergence of single-cell transcriptomic technologies enables the investigation of individual cells and their organization through single-cell/nuclei RNA sequencing (sc/snRNAseq) and spatial transcriptomics (STx). Different single-cell transcriptomic platforms have been used to examine hepatic disease models from diet induced nonalcoholic fatty liver disease (NAFLD; now commonly referred to as metabolic dysfunction-associated fatty liver disease or MAFLD) [10] to hepatotoxicants including carbon tetrachloride (CCl₄) [11], acetaminophen (APAP) [12], and 1,4-bis-[2-(3,5,-dichloropyridyloxy)] benzene (TCPOBOP) [13]. These studies, as well as others have provided novel insight into the function and diversity of cell (sub)types in normal and diseased livers [14]. This review discusses the application of sc/snRNAseq and STx in the elucidation of underlying cell-specific mechanisms involved in TCDD-induced TAFLD, and how these technologies allow toxicologists to tackle novel questions about liver toxicity.

2. Cell-specific responses in TCDD-elicited TAFLD.

Bulk transcriptomic analyses capture an average of the multiple cell types of a tissue. Hepatocytes, the most common cell type, are better represented in these analyses compared to rarer cell types such as portal fibroblasts (Figure 1). Consequently, cell-specific responses can be masked or underappreciated. Moreover, differential expression in a bulk sample does not distinguish between changes due to altered cell proportions, altered expression within a specific cell type, or a combination of both factors [15]. With the ability to distinguish between distinct cell (sub)types using single-cell transcriptomic, it is becoming increasingly evident that changes in relative proportions and cell-specific gene expression are an important components of TCDD-elicited toxicities, not just in the liver, but also in other tissues such as the testes [15–17]. While some responses are common across many cell types cell types such as the induction of AHR battery genes [15–17], single-cell analyses has provided novel insights regarding the role of other cell types classically involved in FLDs (e.g., hepatocytes, immune cells, and hepatic stellate cells), as well as other cell types less commonly considered in the context of TAFLD (e.g., liver sinusoidal endothelial cells).

2.1. Hepatocytes – disruption of the liver microarchitecture

The liver is organized into lobules, polygonal units with a central vein in the middle and at the vertices a portal triad consisting of the portal vein, hepatic artery, and bile duct [18]. Although lobules are classically broken down into 3 broad zones from periportal (zone 1) to central (zone 3), single-cell transcriptomics has further refined the microarchitecture showing that lobules consist of several layers [19] and more precisely a functional continuum along the lobular axis [12, 17]. Hepatocytes have been extensively studied for their response to TCDD and other AHR ligands using both in vitro and in vivo models, though only a few studies have examined the zonated molecular responses [17, 20–22].

In TCDD-elicited TAFLD in mice, hepatocytes are the first to show phenotypic changes marked by lipid droplet accumulation in the periportal region [4, 17]. Conversely, pharmacokinetic modeling in rats predicted AHR activation and induction of target genes Cyp1a1 and Cyp1a2 initially in the centrilobular region [20]. Furthermore, computational modeling of zonated AHR activation by TCDD identified the disruption of Wnt/β-catenin signaling cascade and a negative feedback loop driven TCDD sequestration by Cyp1a2 [21]. STx and snRNAseq analyses validated these models. Using a dose-response study design both STx and snRNAseq showed initial induction of classical AHR battery genes, eventually being induced in the periportal region at higher doses. Notably, scRNAseq showed that nuclei responsive at lower doses were also those that more highly expressed the Ahr coming from the centrilobular region, consistent with previous characterization of Ahr zonation [23] and the STx profiling. It is unclear why lipids first accumulate in portal hepatocytes, but may reflect a zonal difference in AHR-mediated pro-oxidant and anti-oxidant responses and/or disruption of lipid uptake and metabolism which primarily occurs in the portal region [8, 24].

TCDD also elicited a loss of zonated mRNA and long-noncoding RNA (lncRNA) expression [17, 22], along with an overall loss of canonical hepatocyte intercellular signaling pathways such as growth hormones, thyroid signaling, and the coagulation cascade [22, 25]. Single-cell transcriptomic analysis of APAP injury and partial hepatectomy similarly identified a loss of zonated gene expression and Wnt/β-catenin signaling activation reflecting functional compensation to support liver repair and regeneration [26]. However, TCDD-elicited hepatotoxicity is species- and sex-specific [8, 27, 28] and scRNAseq has only been used on a limited number of models to date. Treatment with the constitutive androstane receptor (CAR) TCPOBOP has demonstrated sexually dimorphic zonal molecular patterning in differences in zonal responses [13] highlighting the need to investigate the role of zonation in species- and sex-specific responses to AHR agonists.

2.2. Inflammatory cells – infiltration of lipid-associated macrophage

Liver homeostasis is supported by diverse resident immune cells and the application of single-cell transcriptomics has further characterized the heterogeneity of hepatic immune cell (sub)types [29, 30]. Numerous studies have shown the immunomodulatory effects of TCDD in autoimmune hepatitis (AIH) models using concanavalin A (ConA) [31, 32] as well as a proinflammatory role in the liver [6, 33]. However, given the low level of these cells, their characterization requires enrichment for non-parenchymal cells (NPCs). In an AIH model, scRNAseq of isolated hepatic mononuclear cells showed that acute AHR activation by TCDD suppressed CD8+ T cell proliferation [31]. Trajectory analysis, where cells are ordered on a proliferation/differentiation continuum based on their transcriptomic profile, showed that TCDD and ConA co-treatment found fewer cells transitioning to Mki67 expressing CD8+ T Cells compared to ConA alone.

In contrast to the AIH model, development of inflammatory foci composed primarily of mononuclear cells and neutrophils are reported to be increased in TCDD-elicited TAFLD [33]. Development of Inflammatory foci likely involve multifactorial responses including the release of danger associated molecular patterns (DAMPs), microbiome cues, chemokine signaling, and direct AHR activation. Macrophage exhibit the largest cell population change and number of differentially expressed genes in snRNAseq analyses [15, 17]. Moreover, cell-cell signaling analysis identified CCL and TGFβ signaling between hepatocytes and inflammatory cells as well as within macrophages [22, 25]. The CCL member MCP-1 (Ccr2) is reported to drive early recruitment of F4/80⁺ (Adgre1; macrophage marker) cells to the liver [33]. Closer inspection of the macrophage population also revealed the infiltration of cells expressing Gpnmb, a transmembrane glycoprotein and putative NAFLD biomarker [34]. Gpnmb expressing cells were also enriched for Trem2 and Cd9 which identifies a macrophage subtype typically associated with lipids (lipid-associated macrophage; LAMs) in various tissues including atherosclerotic plaques, brain, adipose tissue, and other hepatic steatosis models [10, 30, 35]. The identification of LAMs in TCDD-elicited TAFLD suggests a common underlying mechanism between classical NAFLD/MAFLD and TAFLD.

2.3. Hepatic stellate cells – not just extracellular matrix remodelers

In a quiescent state hepatic stellate cells (HSCs) serve a homeostatic function, but following activation serve a critical role in extracellular matrix (ECM) remodeling and fibrosis. The involvement of HSCs in AHR-mediated fibrosis is widely reported, however, recent evidence suggests that the AHR can also suppress the activation of HSCs [36, 37]. Single-cell transcriptomics confirmed HSC-specific differential gene expression associated with extracellular matrix remodeling [22, 38], but did not find increased expression of activated myofibroblasts markers as reported in other scRNAseq studies [11]. HSCs are difficult to capture in sc/snRNAseq experiments [39], but STx analysis of TCDD treated livers identified induction of alpha-smooth muscle actin (Acta2 aka αSMA) consistent with widespread HSC activation [17].

HSCs also regulate the vasculature, liver microarchitecture, and immune function through cell-cell interactions with hepatocytes, liver sinusoidal endothelial cells (LSECs), and immune cells [40]. This is exemplified by Colony Stimulating Factor (CSF) and Platelet-Derived Growth Factor (PDGF) signaling between HSCs and macrophage, as well as vasculature endothelial growth factor (VEGF) and angiopoietin-like proteins (ANGPTL) between HSCs and LSECs [22, 25]. Interestingly, single-cell transcriptomics recently identified distinct central and portal HSC populations and showed activation of the central HSCs (CaHSCs) in a CCl₄ model of centrilobular fibrosis [41]. The Wnt/β-catenin signaling enhancer R-spondin 3 (Rspo3) was reported to be primarily expressed in CaHSCs [41], consistent with the role of β-catenin in maintaining central characteristics. RSPO3 levels have also been linked to liver disease severity [42], and induction in TCDD-elicited TAFLD appears to be primarily in HSCs [17] representing an additional factor disrupting lobular zonation. Collectively, accumulating evidence suggests HSCs are key signaling mediators in TAFLD beyond their role in ECM remodeling.

2.4. Liver sinusoidal endothelial cells – underappreciated cells in TAFLD.

LSECs play a critical role in filtering nutrients and waste products, monitoring foreign molecules, and maintaining liver homeostasis [43, 44]. Despite being the most abundant nonparenchymal cell type, the significance of LSECs in TCDD-elicited liver pathologies has received little attention. In other FLD models, LSECs are implicated as ‘gatekeepers’ through capillarization, the loss of transcellular pores called fenestration, which is typically reported early in disease pathogenesis [43]. Single-cell transcriptomics identified perturbed LSEC gene expression related to the vasculature and edema [17, 25] consistent with LSEC swelling, degeneration, and sinusoid stenosis following TCDD treatment of Rhesus monkeys [45]. While capillarization has not been examined in AHR-mediated TAFLD, other liver toxicants such as arsenic do elicit capillarization [46] suggesting it may be a common defense or disease mechanism of the liver. TCDD induced signaling factors that perturbed LSECs include VEGF and TGFβ which promote differentiation and capillarization of LSECs [44], as well as c-KIT (Cd117) which is reported to prevent age-mediated FLD through modulation of zonation [25, 47]. While transcription factor analysis found Sox17, associated with the proliferation of endothelial cells, to be specifically induced in LSECs [22, 48], it is unclear whether self-renewal/proliferation contributes to capillarization. Interestingly, network analysis previously identified Sox17 as a hub for gene regulation [49] implicating LSECs as key players in TCDD-elicited TAFLD.

3. Concluding Remarks

In addition to assessing the distinct roles of the liver cell types in TCDD-elicited hepatotoxicity, single-cell transcriptomics can also be used to examine the cellular microenvironment and microarchitecture at unprecedented resolution. This review outlines how these technologies have further defined the role of distinct cell types in TAFLD identifying common and novel mechanisms in TAFLD and other FLD models. Importantly, these studies not only refined our understanding of the role of hepatocyte zonation, but also identified contributions by other liver cell types that were not distinguishable in bulk transcriptomic analyses. For example, LSECs which were largely ignored in TAFLD due to technical limitations or were not generally considered to play a central role. Application of single-cell transcriptomics in hepatic toxicology is in its infancy presenting unique opportunities. Future studies contrasting how different cell populations respond to different toxicants, investigating spatial organization all cell types, and leveraging emerging machine learning models to characterize microenvironments has the potential to change the perspective of mechanistic toxicology towards a more cell-type centered approach (Figure 3).

Figure 3.

Overview of the potential applications of single-cell transcriptomic technologies in addressing toxicological questions at unprecedented resolution. Created in BioRender.com.

Figure 2.

Mechanistic insight leveraging novel insights of cell-specific differential expression and inferred cell-cell communication gained from the application of single-cell transcriptomic technologies. Initiated by the zonally-biases activation of the AHR in hepatocytes, perturbation in the balance of pro- and anti-oxidant processes as well as altered lipid import and export result in periportal steatosis which eventually progresses to the entire liver lobule. Cellular cues including DAMPs and cytokines contribute to the recruitment of inflammatory cells including LAMs producing the steatohepatitis phenotype. As disease progresses HSCs and LSECs contribute to changes in the cellular matrix and movement of molecules which may include capillarization of LSECs and vasculature remodeling signals from HSCs. A loss in growth factor signaling as well as aberrant production of WNT ligands such as RSPO3 results in a loss of zonation and functional compensation as the liver attempts to maintain normal functions and reestablish homeostasis. Collectively, TAFLD progression is mediated by the many cell types of the liver and their organization in the liver requiring novel approaches such as single-cell transcriptomics to unravel the interactions. Legends shows each cell type and its alternative state in parentheses. Created in BioRender.com.

Highlights:

• The liver microarchitecture plays a central role in toxicant-associated liver disease.

• Fatty liver diseases share many characteristics highlighting unifying mechanisms.

• Sinusoidal endothelial cells play an underappreciated role in TCDD toxicity.

• Single-cell transcriptomics further elucidates the underlying molecular mechanisms of liver disease.