Discussion
RV adaptation to pressure overload remains the main prognostic factor in patients with chronic PH [
10], albeit poorly understood. Our analysis, including imaging, proteomics, and metabolomics studies in pigs (Fig.
5), produced a number of key findings. (1) RV adverse remodeling and dysfunction (assessed by CMR) was greater in the presence of PH (M1 and M2) than in animals with isolated RV pressure overload (M3), even though pigs in the M3 group had larger increases in RV systolic pressure and more pronounced RV hypertrophy. (2) Unlike the isolated pressure overload model, the PH models were associated with deficiency in metabolites such as arginine, histidine, taurine, and purines, together with altered abundance of proteins related to the coagulation pathways and complement system activation. (3) All three models of RV pressure overload showed systemic increases in lipid compounds although of different category. (4) In the integrative analysis, the parameters most strongly associated with maladaptive RV hypertrophy in PH were deficiencies in arginine, histidine, taurine, and purine metabolic pathways, complement system activation, and increased circulating FAs.
Although previous experimental studies have investigated potential mechanisms of RV failure in PH [
1], these were performed in small animal models, and thus hemodynamic evaluation of PH severity was limited to RV systolic pressure; moreover CMR analysis was very seldom included. Other drawbacks of small animal studies are the induction of PH by infusion of toxins such as SU5416 which could modify signaling pathways in the RV, and the acquisition of results after a short follow-up. Finally, previous studies have generally focused on the RV myocardium, and have not investigated potential mediators leading to myocardial injury, which may have an origin in the pulmonary vasculature.
Our metabolomic analysis showed that PH was associated with significant changes in energy metabolism and increased oxidative stress. In line with prior findings in PAH patient samples [
16,
30,
32], we found decreased plasma arginine in both PH models, correlating with maladaptive RV hypertrophy in vivo. Decreased plasma arginine is an indicator of increased arginase activity, which leads to competition with nitric oxide (NO) synthase and hence reduced NO [
25]. Indeed, previous studies have shown an inverse association between plasma arginine and PAP [
19], and also prognosis [
16,
32]. The two PH models also showed reduced myocardial expression of arginine
N-methyltransferases, and M1 animals additional upregulation in lung parenchyma NOX5. These findings are in line with patient data showing PRMT1 downregulation and NOX5 overexpression in heart failure [
22]. Moreover, PRMT1 ablation in mouse cardiomyocytes causes myocardial dysfunction [
22].
The PH models also had altered histidine metabolism, indexed by significantly decreased plasma levels of anserine, 1-methyl-histidine, carnosine, and histidine. Decreased plasma histidine is a prognostic indicator in patients with heart failure [
4], likely due to the shift from FA to glucose utilization in response to the histidine deficiency.
Animals with PH induced by aorto-pulmonary shunting (M2) showed a particularly pronounced effect on taurine metabolism, evidenced by reduced plasma levels of hypotaurine, taurine, alanine, and pyruvic acid. Taurine has an antioxidant effect mainly attributed to its mitochondrial protective effect, and low plasma taurine has been reported in patients with diabetic cardiomyopathy [
6] and linked to cardiac remodeling in experimental models of taurine deficiency [
18] Nevertheless, the most altered pathway in the aorto-pulmonary shunting PH model was the purine pathway, with marked decreases in 8 metabolites, including inosine, hypoxanthine, xanthine, guanine, and uric acid. These metabolites have been inversely linked to PAH and hemodynamic severity [
13,
30] and have been proposed as prognostic indicators in PAH [
17]. The more pronounced changes to taurine and purine metabolism in M2 than in M1 likely reflects a greater involvement of the pulmonary vasculature. The integrative analysis revealed that the taurine and purine pathway deficiencies were related to PH severity and RV maladaptive hypertrophy, suggesting that alterations to these pathways might be important in maladaptive RV remodeling associated with PH.
Lipidomics has seldom been used to investigate PH, and our study provides interesting data in this area. Plasma lipid concentrations were significantly increased in all three RV pressure overload models, thus suggesting an association with RV hypertrophy independently of PAP. However, the type of lipid affected differed between the models, with PCs and TGs increased in M3, FAs and SMs in M2, and only a few glycosphingolipids in M1. Interestingly, the only lipids showing a significant correlation with PH severity and maladaptive RV hypertrophy were free FAs. In PH, FA accumulation has been linked to the disruption of β-oxidation in mitochondria and peroxisomes [
5] and the resulting metabolic switch from lipid oxidation to glycolysis and lactate-producing anerobic metabolism, causing a gradual export of lipids into the bloodstream and increasing circulating free FA concentrations [
3]. Impaired myocardial β-oxidation and increased glucose utilization result in less efficient ATP production, decreased contractile performance, and progression toward RV failure [
31]. Chronic RV failure is frequently associated with cachexia, particularly in patients with concomitant LV heart failure, which aggravates prognosis. The large differences in weight gain between animals with PH (M1 and M2) compared to animals with isolated pressure overload (M3), together with the differences in metabolism, open new opportunities to understand the genesis and differences in the establishment of cachexia among patients.
The proteomics analysis revealed a significant overrepresentation of biologic processes related to blood coagulation, complement activation, and lipid metabolism, suggesting these as functional proteomic footprints of PH-derived RV failure in plasma. Whereas changes in complement activation were observed in both PH models, only the postcapillary PH model showed evidence of an enhanced procoagulant state, which might reflect greater venous stagnation in postcapillary PH. Complement activation correlated significantly with PH hemodynamic severity and maladaptive RV hypertrophy in the integrated interaction network, thus suggesting that inflammation plays an important role in RV dysfunction secondary to PH. This inflammation and oxidative stress in PH may have different sources. Systemic increase in free FAs, observed in M2, especially arachidonic acid, are precursors of pro-inflammatory and pro-oxidant eicosanoids, such as prostaglandins, leukotrienes, and thromboxanes [
11]. On the other hand, the decreased arginine and impaired NOS activity may reduce NO production and increase superoxide generation [
29]. Remarkably, the model of pure RV pressure overload, despite developing a much more pronounced RV hypertrophy, maintained normal RV function and showed little of the alteration of pathologic biologic processes seen in the presence of PH. Instead, M3 animals presented an adaptive RV hypertrophy with markers that have been reported as cardioprotective, such as increased expression of C5aR1 [
28]. CILP1 and NT-proBNP, peptides previously linked to maladaptive RV remodeling [
12] were significantly increased in M1 and M2, respectively, which confirmed the higher specificity of CILP1 over NT-proBNP for RV versus LV involvement [
12,
27]. However, we only found a modest correlation between NT-proBNP levels and RV ejection fraction but not for CILP1. This is possibly because all animals had a RV ejection fraction higher than 42% and CILP1 is a marker of more advanced RV remodeling.
Our study is limited by the sample size, which might have rendered the analysis underpowered to assess differences in some parameters. Moreover, proteomics analyses were performed on a smaller cohort than metabolomics. The reason was that proteomics methodology (based on isobaric labeling), can minimize technical variability and allows very precise quantification, while metabolomics is undertaken in label-free mode, with independent runs for each sample. In addition, since all proteomics and metabolomics analyses were performed on plasma samples, it is difficult to determine if they reflect metabolomic changes originating in the lung or RV. However, this limitation is to some degree overcome by the use of the pure RV pressure-overload model, without PH, and the targeted molecular biologic analysis of RV and lung parenchyma samples. Finally, our study was performed in male castrated animals which may limit generalization of these results.
In conclusion, our study integrating imaging and omics in large-animal experimental models demonstrates that, beyond pressure overload, metabolic alterations play a relevant role in RV dysfunction in PH. These findings open up new research avenues for the identification of novel therapeutic targets.