GepLiver: an integrative liver expression atlas spanning developmental stages and liver disease phases

Abstract

Chronic liver diseases usually developed through stepwise pathological transitions under the persistent risk factors. The molecular changes during liver transitions are pivotal to improve liver diagnostics and therapeutics yet still remain elusive. Cumulative large-scale liver transcriptomic studies have been revealing molecular landscape of various liver conditions at bulk and single-cell resolution, however, neither single experiment nor databases enabled thorough investigations of transcriptomic dynamics along the progression of liver diseases. Here we establish GepLiver, a longitudinal and multidimensional liver expression atlas integrating expression profiles of 2469 human bulk tissues, 492 mouse samples, 409,775 single cells from 347 human samples and 27 liver cell lines spanning 16 liver phenotypes with uniformed processing and annotating methods. Using GepLiver, we have demonstrated dynamic changes of gene expression, cell abundance and crosstalk harboring meaningful biological associations. GepLiver can be applied to explore the evolving expression patterns and transcriptomic features for genes and cell types respectively among liver phenotypes, assisting the investigation of liver transcriptomic dynamics and informing biomarkers and targets for liver diseases.

Identification of a chromatin-bound ERRα interactome network in mouse liver

“…Remarkably, further analysis of the top 10 KEGG pathways across comparisons revealed OXPHOS components as key co-occupied targets underscoring the top 4 most significant pathways based on mean significance, namely NAFLD (non-alcoholic fatty liver disease), OXPHOS, thermogenesis, and diabetic cardiomyopathy (Figure 4B). Interrogation of liver mouse and human expression profiles from a recently developed integrated liver expression atlas GepLiver [26] revealed similar correlation relationships between ESRRA expression and its 9 examined coregulators in mouse models and human patients with NAFLD (Figure 4C). Notably, negative correlations were found between ESRRA and its known transcriptional corepressors NCOR1 and PROX1 .…”

“…Intriguingly, while PGC-1α is regarded as the main coactivator of ERRα, only PGC-1β was found in our RIME study. Examination of gene expression profiles in GepLiver [26] revealed a 1.8-fold higher median expression level of the PGC-1β vs PGC-1α encoding genes in the normal mouse liver. This is in stark contrast to the nearly 11-fold higher PGC-1α vs PGC-1β median levels observed in the normal human liver.…”

“…Normalized mouse and human expression data (transcripts per million, TPM) of normal liver and NAFLD were downloaded from GepLiver [26] to measure using Spearman correlation the strength and degree in correlation between Esrra mRNA levels and the expression of a subset of its identified RIME interactors. Mouse (normal, n=94, NAFLD, n=125) and human (normal, n=362; NADLD, n=503).…”

Identification of a chromatin-bound ERRα interactome network in mouse liver

“…3 , revealing the existence of transcription factor regulatory modules. Interrogation of liver mouse and human expression profiles from a recently developed integrated liver expression atlas GepLiver [ 27 ] revealed similar correlation relationships between ESRRA expression and its 9 examined coregulators in mouse models and human patients with NAFLD ( Figure 4 C). Notably, negative correlations were found between ESRRA and its known transcriptional corepressors NCOR1 and PROX1 .…”

“…Normalized mouse and human expression data (transcripts per million, TPM) of normal liver and NAFLD were downloaded from GepLiver [ 27 ] to measure using Spearman correlation the strength and degree in correlation between Esrra mRNA levels and the expression of a subset of its identified RIME interactors. Mouse (normal, n = 94, NAFLD, n = 125) and human (normal, n = 362; NAFLD, n = 503).…”