Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2014 Nov 7;9(11):e112193.
doi: 10.1371/journal.pone.0112193. eCollection 2014.

Systems Level Analysis and Identification of Pathways and Networks Associated With Liver Fibrosis

Affiliations
Free PMC article

Systems Level Analysis and Identification of Pathways and Networks Associated With Liver Fibrosis

Mohamed Diwan M AbdulHameed et al. PLoS One. .
Free PMC article

Abstract

Toxic liver injury causes necrosis and fibrosis, which may lead to cirrhosis and liver failure. Despite recent progress in understanding the mechanism of liver fibrosis, our knowledge of the molecular-level details of this disease is still incomplete. The elucidation of networks and pathways associated with liver fibrosis can provide insight into the underlying molecular mechanisms of the disease, as well as identify potential diagnostic or prognostic biomarkers. Towards this end, we analyzed rat gene expression data from a range of chemical exposures that produced observable periportal liver fibrosis as documented in DrugMatrix, a publicly available toxicogenomics database. We identified genes relevant to liver fibrosis using standard differential expression and co-expression analyses, and then used these genes in pathway enrichment and protein-protein interaction (PPI) network analyses. We identified a PPI network module associated with liver fibrosis that includes known liver fibrosis-relevant genes, such as tissue inhibitor of metalloproteinase-1, galectin-3, connective tissue growth factor, and lipocalin-2. We also identified several new genes, such as perilipin-3, legumain, and myocilin, which were associated with liver fibrosis. We further analyzed the expression pattern of the genes in the PPI network module across a wide range of 640 chemical exposure conditions in DrugMatrix and identified early indications of liver fibrosis for carbon tetrachloride and lipopolysaccharide exposures. Although it is well known that carbon tetrachloride and lipopolysaccharide can cause liver fibrosis, our network analysis was able to link these compounds to potential fibrotic damage before histopathological changes associated with liver fibrosis appeared. These results demonstrated that our approach is capable of identifying early-stage indicators of liver fibrosis and underscore its potential to aid in predictive toxicity, biomarker identification, and to generally identify disease-relevant pathways.

Conflict of interest statement

Competing Interests: The authors have declared that no competing interests exist.

Figures

Figure 1
Figure 1. Workflow used in this study to identify pathways and networks associated with liver fibrosis.
Figure 2
Figure 2. Number of fibrosis-relevant genes from differential and co-expression analysis.
Number of genes in the liver fibrosis-relevant differentially expressed gene list and liver fibrosis-relevant co-expressed gene list and the overlap between them.
Figure 3
Figure 3. Genes that mapped to the enriched Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways.
The average log2 fold-change ratio across chemical exposures that produced periportal liver fibrosis was used as the gene expression value. Genes with average log2 fold-change ratios >0.6 are colored in red. Genes with average log2 fold-change ratios <−0.6 are colored in green. Genes whose average log2 fold-change ratios are between 0.6 and −0.6 are colored in grey.
Figure 4
Figure 4. Statistical significance analysis of network module M5.
A) The comparison of number of nodes in M5 to that from random sampling analysis. B) The comparison of number of edges in M5 to that from random sampling analysis. C) The comparison of number of nodes in M5 to the number present in shuffled protein-protein interaction (PPI) networks. D) The comparison of number of edges in M5 to the number present in shuffled PPI networks.
Figure 5
Figure 5. Liver fibrosis-relevant network module M5.
Proteins encoded by genes with average log2 fold-change ratios>0.6 are colored in red. Proteins encoded by genes with average log2 fold-change ratios <−0.6 are colored in green. Proteins encoded by genes with average log2fold-change ratios between 0.6 and −0.6<−0.6 are colored in grey. Proteins without corresponding gene expression data are shown as white circles.
Figure 6
Figure 6. Myocilin interaction network.
First neighbors of myocilin (MYOC) in the entire high-confidence human protein-protein interaction (PPI) network.
Figure 7
Figure 7. Activation of genes encoding proteins in liver fibrosis-relevant network module M5 at different time points.
Genes encoding proteins with average log2 fold-change ratios>0.6 are colored in red. Genes encoding proteins with average log2 fold-change ratios <−0.6 are colored in green. Genes encoding proteins with average log2fold-change ratios between 0.6 and −0.6<−0.6 are colored in grey. A) Activation at 0.25-day exposure. The mapped expression profile is the average log2 ratio in 1-naphthyl isothiocyanate 30 mg/kg and 60 mg/kg, at 0.25-day exposure. B) Activation at 1 day of exposure. The mapped expression profile is the average log2 ratio in 1-naphthyl isothiocyanate 30 mg/kg and 60 mg/kg, and 4,4'-Methylenedianiline 81 mg/kg, at 1 day of exposure. C) Activation at 3 days of exposure. The mapped expression profile is the average log2 ratio in 1-naphthyl isothiocyanate 30 mg/kg and 60 mg/kg, and 4,4'-Methylenedianiline 81 mg/kg, at 3 days of exposure. D) Activation at >3 days of exposure. The mapped expression profile is the average log2 ratio across chemical exposures that produced liver fibrosis.
Figure 8
Figure 8. Analysis of genes in liver fibrosis-relevant network module M5.
A) Hierarchical clustering of 640 chemical exposures using genes in liver fibrosis-relevant network module M5. The conditions that clustered with four liver fibrosis-producing conditions are highlighted and listed. Genes with Z-scores>2 are colored in red. Genes with Z-scores <−2 are colored in green. Genes with Z-scores between 2 and -2 are colored in yellow. NA in the table represents that histopathological data was not available for that chemical exposure condition. B) Average Z-scores across the genes in module M5 for each of the 640 chemical exposure conditions.
Figure 9
Figure 9. Validation with external datasets.
M5 gene expression compared with external datasets. A) GSE13747 represents liver fibrosis produced by bile duct ligation. B) GSE6929 represents sunitinib (SU11248) treatment in liver cirrhosis.

Similar articles

See all similar articles

Cited by 15 articles

See all "Cited by" articles

References

    1. Malhi H, Gores GJ (2008) Cellular and molecular mechanisms of liver injury. Gastroenterology 134: 1641–1654. - PMC - PubMed
    1. Sebastiani G (2012) Serum biomarkers for the non-invasive diagnosis of liver fibrosis: the importance of being validated. Clin Chem Lab Med 50: 595–597. - PubMed
    1. Brenner DA (2009) Molecular pathogenesis of liver fibrosis. Trans Am Clin Climatol Assoc 120: 361–368. - PMC - PubMed
    1. Bataller R, Brenner DA (2005) Liver fibrosis. J Clin Invest 115: 209–218. - PMC - PubMed
    1. Baranova A, Lal P, Birerdinc A, Younossi ZM (2011) Non-invasive markers for hepatic fibrosis. BMC Gastroenterology 11: 91. - PMC - PubMed

Publication types

MeSH terms

Substances

Grant support

The authors were supported by the Military Operational Medicine Research Program and the U.S. Army's Network Science Initiative, U.S. Army Medical Research and Materiel Command (USAMRMC, http://mrmc.amedd.army.mil), Ft. Detrick, MD. This research was supported in part by an appointment to the Postgraduate Research Participation Program at the U.S. Army Center for Environmental Health Research (USACEHR, http://usacehr.amedd.army.mil) administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and USAMRMC. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Feedback