EXAMPLE WORK

Introduction

Mitochondria-are-the-main-source-of-ROS

Coronary artery disease (CAD) remains the leading cause of death (see WHO http:// www.who.int/en/). Hypercholesterolemia, its major risk factor, has been associated with exacerbated production of reactive oxygen species (ROS) [1, 2], which significantly contributes to the initiation and development of atherosclerotic lesions [3]. Mitochondria are the main intracellular source of ROS and also the main targets, when ROS are produced exceedingly [4]. Here, we hypothesized that one way in which hypercholesterolemia mith increase ROS is by hampering mitochondrial function, which would be likely reflected by changes in mitochondrial transcription patterns.

The-Reversa-Mice-Model

To address this question, we used the so called Reversa mouse model with human-like hypercholesterolemia. In these mice, LDL receptor-deficiency and apo-B100 knock-in (Ldlr−/−Apob100/100) results in large numbers of small lipoproteins, leading to spontaneous and rapid atherosclerosis development even on a chow diet. Whereas an additional genetic switch allows to inhibit hepatic synthesis of lipoproteins (Mttpflox/floxMx1-Cre) and abrupt plasma cholesterol-lowering [6]. Previously, Skogsberg et. al. have thoroughly investigated different stages of atherosclerotic lesion development and their response to plasma cholesterol-lowering in the Reversa mouse model and generated transcriptome data from the whole-tissue homogenates of their aortic arches, at each time point [7, 8].

Data and Methods

Workflow-Network-Analysis-Reveals-a-Causal-Role-of-Mitochondrial-Gene-Activity-in-Atherosclerotic-Lesion-Formation

Here, we employed a three-step systems-level network approach:

1) First, we performed differential expression analyses of the consecutive time points to identify mitochondrial signature genes. We then analyzed transcriptome profiles in terms of co-expressed sets of genes (i.e., modules), which we ranked by their overrepresentations of mitochondrial signature genes, relevance to changes in atherosclerotic lesions size over time and enrichment with coronary artery disease (CAD) risk from human genome-wide association studies (GWAS)[9].

2) Second, we explored the top scoring modules by functional enrichment analysis as well as transcription factor binding sites (TFBS) prediction, to facilitate their biological interpretation.

3) Finally, we performed key regulatory gene selection, by ranking mitochondrial signature genes from the top scored modules, considering their connectedness within the module (’hub’ genes), gene relevance to lesion expansion, as well as their involvement in ROS homeostasis pathways and transcriptional regulation.

Results

Identification of mitochondrial signature genes

During hypercholesterolemia, we observed a massive down-regulation of mitochondrial genes, specifically at the time of rapid atherosclerotic lesion expansion and foam cell formation, i.e. between 30 and 40 weeks of age.

Interestingly, both phenomena - down-regulation of mitochondrial genes and lesion expansion - were largely reversible by genetically switching mice to normocholesterolemia.

Identification-of-mitochondrial-signature-genes

Gene co-expression network module identification

Gene co-expression network module identification

We performed weighted gene co-expression network analysis WGCNA
to identify tightly correlated sets of genes (i.e., modules). In order to prioritize modules by their relevance to atherosclerotic lesion progression and mitochondrial gene representations, we scored them (+1 or 0) in five different categories by: (i) correlation to changes in atherosclerotic lesion size over time; (ii) CAD risk enrichment (causality) as identified using human GWAS data [9]; (iii-v) the overlaps with expressed mitochondrial genes (n=1,240), mitochondrial lesion expansion (n=352) and mitochondrial cholesterol-responsive (n=63) signature genes.

Mitochondria-relevant module prioritization by scoring

Two modules - green and turquoise - scored positive in all five categories. Both modules negatively correlated with changes in atherosclerotic lesion size over time, displayed significant enrichment for CAD/atherosclerosis risk across multiple tissues, and were significantly enriched with mitochondrial genes, in particular, mitochondrial lesion expansion and cholesterol-responsive signature genes. In fact, the modules were highly related to each other and part of the same meta-module. Taken together, these observations suggest that the two mitochondria-related modules might be important to protect against atherosclerosis, hence our further analysis focused on understanding their genes and pathways.

Mitochondria-relevant module prioritization by scoring

Exploring the top scoring modules

Exploring-the-Top-Scoring-Mitochondrial-Modules

In both modules, functional enrichment analysis suggested that mitochondrial biogenesis might be affected, as we observed enrichment of several key mitochondrial pathways, including mitochondrial fatty acid β-oxidation, TCA cycle, oxidative phosphorylation and electron transport chain, as well as the 55S, 39S and 28S mitochondrial ribosomal subunits. We also observed enrichment of genes involved in homeostasis pathways of four common ROS (hydroxyl radicals, hydrogen peroxide, superoxide and peroxynitrite).

Finally, both modules were also enriched for the transcription factor estrogen related receptor ERR-α (ESRRA) binding sites. Together, ERR-α/PGC1 are known as master regulators of mitochondrial biogenesis and antioxidant defenses.

Ranking top scoring module genes to identify mitochondrial key regulators

Green_Modulte_Top_Scoring_Genes

In the third part of our analysis, we utilized the information obtained from the first two parts and performed candidate gene selection by scoring (+1 or 0). In both modules, there were no genes that would score positively in all nine categories. In fact, the PGC1 factors where among the highest scoring genes in the respective modules: PPARGC1A (PGC1-β) in the green module obtained a total score of +8, whereas PPARGC1B (PGC1-β) in the turquoise module scored +7.

Turquoise_Modulte_Top_Scoring_Genes

Hypothesis

Here, we summarize our findings in a hypothetical chain of events linking hypercholesterolemia to changes in mitochondrial transcriptome and subsequent atherosclerotic lesion formation. Based on previous studies, it is known that mitochondria are the main intracellular source of ROS [10] and that hypercholesterolemia progressively increases mitochondrial ROS production in the arterial wall [1, 2]. What is also known, is that mitochondria themselves are targets of excessive ROS production and/or prolonged oxidative stress, leading to mitochondrial damage and loss-of-function. This happens, in part, by decreasing their antioxidative capacity and diminishing important quality control mechanisms, such as mitochondrial biogenesis, which is essential to replenish the damaged components [10]. Decrease in mitochondrial numbers and/or mitochondrial biogenesis would then manifest itself as massive changes (decrease) in mitochondrial transcriptome [7], observed in this study. Consequently, accumulation of dysfunctional mitochondria may further increase ROS production, initiating a vicious cycle where ROS produced more ROS, triggering inflammatory responses and cell death in the arterial wall, and ultimately contributing to lesion development [5].

Hypothesis-Mitochondrial-ROS-Hypercholesterolemia

Conclusion

In conclusion, our systems-level network analysis, linking time-resolved transcriptome data in atherosclerosis-prone hypercholesterolemic mice with CAD-associated eSNPs from human GWAS provides strong evidence that arterial wall responds to hypercholesterolemia by a coordinated down-regulation of mitochondrial genes. Moreover, these genes are tightly connected in co-expression network modules related to mitochondrial biogenesis and antioxidant responses, possibly regulated by ERR-α/PGC1, and supported as causal for CAD/atherosclerosis. Our study further supports the high predictive power of network strategies in identifying key regulatory genes and contributes to better understanding of the molecular processes leading to altered mitochondrial ROS metabolism in atherosclerosis, which could lead to more effective therapeutic strategies.

References


[1]  Krause, S. et al. Increased generation of reactive oxygen species in mononuclear blood cells from hypercholesterolemic patients. Thromb Res 71, 237–240 (1993).
[2] Oliveira, H. C. F. et al. Oxidative stress in atherosclerosis-prone mouse is due to low antioxidant capacity of mitochondria. FASEB J 19, 278–280 (2005). URL http://dx. doi.org/10.1096/fj.04-2095fje.
[3] Madamanchi, N. R. & Runge, M. S. Mitochondrial dysfunction in atherosclerosis. Circ Res 100, 460–473 (2007). URL http://dx.doi.org/10.1161/01.RES.0000258450. 44413.96.
[4]  Kotiadis, V. N., Duchen, M. R. & Osellame, L. D. Mitochondrial quality control and com- munications with the nucleus are important in maintaining mitochondrial function and cell health. Biochim Biophys Acta 1840, 1254–1265 (2014). URL http://dx.doi.org/10. 1016/j.bbagen.2013.10.041.
[5] Gutierrez, J., Ballinger, S. W., Darley-Usmar, V. M. & Landar, A. Free radicals, mitochon- dria, and oxidized lipids: the emerging role in signal transduction in vascular cells. Circ Res 99, 924–932 (2006). URL http://dx.doi.org/10.1161/01.RES.0000248212. 86638.e9.
[6]  Véniant, M. M. et al. Defining the atherogenicity of large and small lipoproteins containing apolipoprotein b100. J Clin Invest 106, 1501–1510 (2000). URL http://dx.doi.org/ 10.1172/JCI10695.
[7] Skogsberg, J. et al. Transcriptional profiling uncovers a network of cholesterol-responsive atherosclerosis target genes. PLoS Genet 4, e1000036 (2008). URL http://dx.doi. org/10.1371/journal.pgen.1000036.
[8]  Björkegren, J. L. M. et al. Plasma cholesterol-induced lesion networks activated before regression of early, mature, and advanced atherosclerosis. PLoS Genet 10, e1004201 (2014). URL http://dx.doi.org/10.1371/journal.pgen.1004201.
[9]  CARDIoGRAMplusC4D Consortium. Deloukas, P. et al. Large-scale association analysis identifies new risk loci for coronary artery disease. Nat Genet 45, 25–33 (2013). URL http://dx.doi.org/10.1038/ng.2480.
[10] Ekstrand, M. et al. Imaging of intracellular and extracellular ros levels in atherosclerotic mouse aortas ex vivo: Effects of lipid lowering by diet or atorvastatin. PLoS One 10, e0130898 (2015). URL http://dx.doi.org/10.1371/journal.pone.0130898.