Author + information
- Alejandro Santos-Lozano, PhD,
- Carmen Fiuza-Luces, PhD∗ (, )@hospital12Oct@UEuropea,
- David Fernández-Moreno, MSc,
- Francisco Llavero, PhD,
- Joaquín Arenas, PhD,
- Juan Antonio López, PhD,
- Jesús Vázquez, PhD,
- Pilar Escribano-Subías, MD, PhD,
- José L. Zugaza, PhD and
- Alejandro Lucia, MD, PhD
- ↵∗Research Institute of Hospital “12 de Octubre” (‘i+12’), Avenida de Cordoba, s/n, 28041 Madrid, Spain
An adjuvant therapy for pulmonary arterial hypertension (PAH), a condition characterized by complex vascular remodeling ultimately culminating in occlusive arteriopathy and right heart failure, is regular exercise (1). Yet, whether/how exercise affects PAH pathobiology is unknown. Systems biology can help reveal the protein networks involved in disease conditions and how they are affected by treatment.
Using this approach, we studied the plasma proteome of a subgroup of PAH patients participating in a randomized controlled trial (RCT) where the intervention group performed an 8-week exercise program, which significantly improved a PAH-mortality predictor: cardiorespiratory fitness (vs. no changes in inactive control subjects) (1). To minimize the risk of bias given the multifactorial etiology of PAH, resting samples for proteomic analyses were collected before and after the RCT only in those participants (n = 9 [5 from intervention group, 1 male, age 35 to 53 years] and 4 control subjects [1 male; age 40 to 56 years]) presenting with homogeneous pathophysiology (idiopathic, hereditary, or connective tissue disease-associated PAH) and treatment (oral anticoagulants + phosphodiesterase-5 inhibitors + endothelin-receptor antagonists).
We determined protein expression with tandem-mass tag-based quantitative proteomics (2), and then used 2 statistical approaches to identify relevant proteins potentially associated with exercise-training effects on the pathobiology of PAH: Wilcoxon’s test, to identify proteins that were differentially expressed as a result of the 8-week exercise training intervention (vs. the control group); and threshold-based cross-validation methods, to identify proteins with the best between-group classifier potential (with “good classifiers” allowing assignment of a given plasma sample to 1 of the 2 patient groups, exercise or control [p value of accuracy cross-validation <0.05, applying leave-one-out analysis]). For the second approach, we employed a data-mining strategy developed by Anaxomics Biotech using artificial intelligence-based methods.
From the 1,251 proteins identified, only 25 showed changes in their expression after the training program that were potentially associated with the pathobiology of PAH: 10 differentially expressed, 12 good classifiers, and 3 accomplishing both criteria. To assess their biological role in PAH, we performed 2 different systems biology-based analyses. We initially studied the relationship of the identified proteins with those known to be relevant in PAH (“effector proteins”) using protein-protein networks generated on the basis of a molecular definition of PAH that considered 5 main disease “effector motifs” (including 81 effector proteins in total): 1) pulmonary vascular remodeling; 2) inflammation; 3) vasoconstriction; 4) vascular extracellular matrix remodeling; and 5) endothelial-to-mesenchymal transition. Public databases were consulted for PAH molecular characterization and protein-network generation (2). The direct physical/functional interactions between the differentially expressed proteins and PAH effectors were evaluated, unveiling protein–protein links through which the 25 identified proteins could play a role in the pathobiology of PAH.
The potential molecular relationships between the identified proteins and PAH pathobiology were evaluated using artificial neural networks (ANN), through the application of “Therapeutic Performance Mapping System” technology, which applies supervised machine learning methods based on human protein functional networks to infer clinical and protein-level knowledge. We used ANN to determine relationships between proteins and clinical elements of the network, assigning a 0 to 100 ANN score to each protein according to its functional link to PAH (globally or for each effector motif): strong (>76, p < 0.05), medium-strong (40 to 76, p = 0.05 to 0.25) or weak (<40, p > 0.25).
The best candidates were cathepsin-D, NCAM1, neuropilin-1, profilin-1, and SPARC-like protein-1 (SPARCL1) (Table 1). Next, we studied the effects of acute exercise on these proteins in all patients (n = 19) in the RCT intervention group by measuring their plasma concentration before and after a typical RCT session (cycle-ergometer + resistance exercise) using enzyme-linked immunosorbent assays. Only neuropilin-1 concentration changed significantly (+6.7%, 3 replications average; Wilcoxon p < 0.034), a finding that was replicated in another cohort (n = 9 healthy individuals [6 female, age 40 to 55 years]) (+15.5%, 3 replications average; p < 0.029).
These latter results are in agreement with previous research showing increased skeletal-muscle neuropilin-1 mRNA levels following acute resistance exercise (3). This glycoprotein is involved in systemic/lung vascular development, and loss of semaphorin-3/neuropilin-1 signaling is associated with hypertensive changes in fetal lung arteriolar walls (4). Further research might determine whether neuropilin-1 acts, as our preliminary results suggest, as a muscle-derived factor (i.e., myokine) with potential beneficial effects at the pulmonary vessel level (by attenuating vascular remodeling).
It remains to be established whether the benefits of regular exercise on PAH are mediated by chronic effects on protein networks (e.g., NCAM1 down-regulation [Table 1]) and/or cumulative acute effects (e.g., neuropilin-1 increases).
Please note: Prof. Fiuza-Luces is supported by Miguel Servet contract (CP18/00034) by the Institute of Health Carlos III (ISCIII). Research by Profs. Arenas and Zugaza and Dr. Lucia is funded by FIS, Fondo de Investigaciones Sanitarias (grant numbers PI14/00903, PI1702052, PI18/00207, PI15/00558 and PI18/00139) and FEDER funds from the European Union. Profs. López's and Vázquez's research was supported by competitive grants from the Spanish Ministry of Economy and Competitiveness (MINECO) (BIO2015-67580-P), the Fundació MaratóTV3 (grant 122/C/2015), and “la Caixa” Banking Foundation (project code HR17-00247). Dr. Lucia and Prof. Vázquez acknowledge Universidad Europea de Madrid (Faculty of Sports Sciences) and CIBER de Enfermedades Cardiovasculares (CIBERCV), respectively. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- 2019 American College of Cardiology Foundation
- Gonzalez-Saiz L.,
- Fiuza-Luces C.,
- Sanchis-Gomar F.,
- et al.
- Fiuza-Luces C.,
- Santos-Lozano A.,
- Llavero F.,
- et al.