Proteomic Analysis of the Myocardium in Hypertrophic Obstructive Cardiomyopathy

Hypertrophic cardiomyopathy (HCM) is a common inherited disease caused mainly by dominantly inherited mutations in genes encoding cardiac sarcomere proteins.¹ The disease phenotype is characteristically diverse, ranging from no clinically detectable abnormality to extreme hypertrophy, and death from ventricular arrhythmia, heart failure, and stroke. The biological effects of individual genetic variants only partly explain this complex clinical spectrum and much still needs to be learnt about the downstream pathways that lead to disease development and progression.

Global proteome studies have provided mechanistic insights into many human cardiovascular diseases, including heart failure,² aortic aneurysm,³ atherosclerosis,⁴ atrial fibrillation,⁵ and inflammatory dilated cardiomyopathy,⁶ but there have been few proteomic studies in HCM. In this study, we perform comparative whole tissue proteomics using myocardial tissue removed at the time of septal myectomy. Our aim was to identify dysregulated proteins that provide insight into disease pathogenesis.

Methods

The National Research Ethics Committee approved the study (REC Ref no. 04/0035 and REC 11/LO/0913) and written informed consent was obtained from all patients before surgery. Detailed methods are available in Methods in the Data Supplement. The data, analytic methods, and study materials will be made available to other researchers for purposes of reproducing the results or replicating the procedure. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository.

Results

Study Cohort

The clinical characteristics of patients included in the phase 2 study are shown in Table. The demographics and genetic characteristics of patients included in phase 1 and 2 studies are shown in Table I in the Data Supplement. Genetic testing was performed in 47 of the 52 HCM patients. Patients (n=6) with a variant of uncertain significance were excluded from the subgroup analysis by genotype in the phase 2 study. Two patients included in the phase 1 study had genetic variant in MYBPC3 initially classified as pathogenic downgraded to a variant of uncertain significance at the time of final analysis.

Identification and Classification of Proteins

A total of 1586 proteins were identified across all heart samples, 722 of which were detectable across both disease and control groups (Table IIA in the Data Supplement: protein list [see excel file]). After excluding proteins with <1 unique peptide, the core proteome consisted of 531 proteins (Table IIB in the Data Supplement; see excel file). Principal component analysis demonstrated that the control samples were clustered with separation from the HCM samples, indicating the HCM tissue had a distinctly different proteome (Figure 1).

• Download figure
• Download PowerPoint

Proteomic Differential Expression Analysis

A statistically significant difference (P<0.05) was found between HCM and controls for 151 proteins; 56 proteins were upregulated and 95 downregulated in HCM compared with controls. The fold change was >1.3 for 125 proteins and >1.5 for 79 proteins (Table IIB in the Data Supplement; see excel file).

Gene Ontology and Pathway Analysis

The core proteome data set was submitted for gene ontology and IPA pathway analysis. Functional annotation clustering using PANTHER-GO (Protein Analysis Through Evolutionary Relationships-Gene Ontology) Biological process analysis showed that many proteins identified in the phase 1 study were associated with muscle contraction and metabolism (Table IIC in the Data Supplement: Gene ontology [see excel file]). Proteins involved in actin cytoskeletal signaling, calcium signaling, and ILK (integrin-linked kinase) signaling pathways were upregulated, whereas proteins involved in oxidative phosphorylation and mitochondrial dysfunction pathways were downregulated (Figure 2; Table IID in the Data Supplement: Canonical pathways [see excel file]). Upregulated biological functions specifically associated with the heart included cardiac hypertrophy, arrhythmia, dilatation, and heart failure (Table IIE in the Data Supplement: Biological functions [see excel file]).

• Download figure
• Download PowerPoint

Validation of Expression Proteomic Data Using a Targeted Multiple Reaction Monitoring-Mass Spectrometry Assay

The 33 proteins included in the validation assay are listed in Table III in the Data Supplement. The direction, fold change, and significance level across the pilot, phase 1 and phase 2 studies are shown. Using quantitative mass spectrometry/mass spectrometry 8 proteins were found to be significantly altered between the HCM and control groups. The multiple reaction monitoring transitions are shown in Table IV in the Data Supplement. ACTS (actin, α-skeletal muscle; P68133), FHL1 (four and a half LIM domains protein 1; Q13642), DES (desmin; P17661), CA3 (carbonic anhydrase 3; P07451), and LUM (lumican; P51884) were elevated in HCM versus control (P<0.01). Fructose-bisphosphate ALDOA (aldolase A; P04075), CKM (creatine kinase M-type; P06732), and PGAM2 (phosphoglycerate mutase 2; P15259 and PGAM1, P18669) were reduced in HCM versus control (P<0.01).

Structural Proteins

The percentage of cardiac α-actin (expressed as a percentage of total skeletal and cardiac actin) was lower in HCM (54±3%) than both controls (65±7%; P<0.05) and aortic stenosis (AS; 61±8%; P=0.016). FHL1 was higher in HCM (326±113pmol/L) compared with controls (111±90, P=0.0002) and AS (154±139, P=0.0031). DES was higher in HCM (211±83 pmol/L) compared with controls (128±36, P=0.011) and AS samples (119±80, P=0.0051; Figure 3).

• Download figure
• Download PowerPoint

Metabolic Proteins

CKM-type was higher in controls (738±154 pmol/L) than HCM (458±109, P=0.0002) and AS (476±176, P=0.0175). Fructose-bisphosphate ALDO was lower in HCM (154±44 pmol/L) than controls (201±62, P=0.0425), but there was no difference with AS samples. The peptide chosen for validation of PGAM exists in both the b-type (PGAM1) and m-type (PGAM2) subunits selected because the mb-isozyme predominates in cardiac muscle.

Lumican

The discovery of significantly elevated levels of lumican was a novel finding and considered to be relevant to disease mechanisms. Therefore, 2 peptides were chosen for inclusion in the phase 2 study. There was a linear correlation between both peptides (r²=0.96, P<0.01) indicating they are good markers for whole protein quantitation (Figure 4A). Lumican concentration was higher in HCM (283±176 pmol/L) than controls (133±25, P=0.003) and lower than AS (544±311, P=0.026; Figure 4B). Lumican concentration correlated with left atrial area on echocardiogram (ρ 0.34 [95% CI, 0.06–0.57]; P=0.015; Figure IV in the Data Supplement), but there was no correlation with age, sex, body mass index, glomerular filtration rate, CRP (C-reactive protein), NT-proBNP (N-terminal pro-B-type natriuretic peptide), left ventricular wall thickness, or cardiopulmonary bypass time. Lumican concentration was higher in 16 patients with a pathogenic sarcomere mutation (366±230 pmol/L) than 24 patients (243±124, P=0.041) with no pathogenic variant identified. Of the patients that underwent cardiac magnetic resonance imaging, lumican was higher in those with late gadolinium enhancement in the septum (n=20, 317±151 pmol/L) compared with those without late gadolinium enhancement (n=12, 208±121 pmol/l, P=0.032; Figure 4D).

• Download figure
• Download PowerPoint

Carbonic Anhydrase 3

CA3 concentration was higher in HCM (108±129 pmol/L) than controls (21±7, P=0.003), but there was no significant difference between AS and HCM (Figure 5A). There was a positive correlation between % skeletal actin and CA3 (r=0.271, P=0.04) and levels of FHL1 and CA3 (r=0.352, P=0.07). There was no difference between patients with and without a pathogenic sarcomere mutation (Figure 5B). To be certain this protein was myocardial in origin, the increased expression of CA3 was confirmed by immunohistochemistry where it was found to be localized to the cytoplasm of the cardiomyocytes with considerable heterogeneity in expression both between samples and within the same sample (Figure 5C). In patients with HCM (n=51) and AS (n=7), CA3 levels correlated with serum NT-proBNP (ρ=0.33 [95% CI, 0.07–0.54], P=0.013, n=58) and left atrial area (ρ=0.30 [95% CI, 0.04–0.52], P=0.023, n=58). There was no correlation with age or time on cardiopulmonary bypass, both potential confounders of underlying myocardial metabolic and acid-base status. No significant difference in CA3 was found in patients with late gadolinium enhancement on cardiac magnetic resonance imaging compared with those without.

• Download figure
• Download PowerPoint

Discussion

This is the first unbiased label-free comparative whole tissue proteomic characterization of myocardium from patients with HCM. Several observations are compatible with the known pathophysiology of the disease, while others provide new insights into cellular processes that might be involved in cardiac dysfunction. Although there are limitations to studying whole heart homogenate, this was an exploratory study to generate testable hypotheses. To date, there has only been one global proteomic study in a murine model of HCM expressing a TNNI3 gene mutation (Gly203Ser) which showed >2-fold changes in 34 proteins involved in energy production, calcium handling, and cardiomyocyte structure.⁷ There have been several small transcriptomic studies, which corroborate many of our findings, including increased expression of markers of secondary hypertrophy and changes in proteins associated with energy metabolism, the ECM (extracellular matrix) and inflammation.^8–10

Altered Myocardial Metabolism

The normal myocardium depends on oxygen for high-energy phosphate production via oxidative phosphorylation.¹¹ A switch in energy substrate preference from carbohydrates to fatty acids is a hallmark of the transition from fetal to adult cardiac metabolism.¹² We observed a mixed pattern of changes in glycolytic enzymes, but since several of these proteins catalyze bidirectional reactions, it was not possible to ascertain whether glycolysis or gluconeogenesis was upregulated or downregulated (functional assays would be necessary to confirm this).¹³ Enzymes responsible for the metabolism of medium and long-chain fatty acids, including long-chain fatty acid coenzyme A (CoA) ligase 1, very long-chain specific acyl-CoA dehydrogenase, and medium chain specific acyl CoA dehydrogenase were all downregulated. The mitochondrial trifunctional enzyme subunit β, which catalyzes the final step of β-oxidation, was also reduced. Overall these findings suggest that β-oxidation is downregulated in HCM. The reduction of creatine kinase observed in both HCM and AS (Figure 3) is consistent with previous reports of impaired high-energy phosphate metabolism in these diseases.^14,15 In the phase 1 study, there was also a significant reduction of ADP ATP translocase 1 in HCM compared with controls and many of the subunits of ATP synthase (File II in the Data Supplement) which supports the notion of an overall decrease in energy flux within the cell as downregulation of cytoplasmic creatine kinase causes a decline in ATP transfer into the mitochondria and thus a reduction in energy delivered to the myofibrils.¹⁶

Mitochondria generate energy in the form of ATP to support intracellular processes, including contraction of myofilaments.¹⁷ Mitochondrial dysfunction is a recognized feature of many diseases, including HCM, but it is unclear whether it is a primary or secondary phenomenon.¹⁸ Pathway analysis found oxidative phosphorylation was downregulated (Figure 2) with decreased expression of multiple protein subunits from complex I, III, IV, and V in HCM compared with normal controls. This finding is in keeping with evidence from animal and human studies suggesting that HCM is characterized by a reduction in the concentration of high-energy phosphates in the myocardium^19,20 and with the hypothesis that energy deficiency is a fundamental driver of the HCM phenotype.²¹

Citrate synthase activity is frequently used as a measure of mitochondrial enrichment.²² We found no difference in expression of citrate synthase (O75390) in HCM compared with controls (File II in the Data Supplement) implying that the observed changes in the electron transport chain were not because of fewer mitochondria. As there was no difference in expression of complex II, which is encoded entirely by nDNA, another possible hypothesis is that perturbations in mitochondrial DNA (mtDNA) contribute to the reduced expression of protein subunits making up the electron transport chain. This idea has been considered in patients with congenital heart disease where mtDNA depletion has been attributed to reduced mtDNA replication in hypertrophied and failing right ventricles.²³

Novel Findings

Four and a Half LIM Domains Protein 1

FHL1 is strongly expressed in the developing outflow tract and to a lesser extent in myocardium. In mouse models of dilated and HCM, cardiac ventricular expression of FHL1, but not of related proteins FHL2 or FHL3, is upregulated.²⁴ The FHL1 protein has also been shown to be significantly upregulated in human and animal models of pulmonary hypertension where its expression is dependent on hypoxia-inducible transcription factor.²⁵ FHL1 interacts with MyBP-C (myosin-binding protein C) contributing to sarcomere assembly²⁶and also Talin-1 although the function of this interaction is not understood.²⁵ A recent transcriptomic study suggested FHL1 may be a genetic modifier in HCM,²⁷ but the mechanism and role of increased FHL1 expression require further investigation.

Lumican

Lumican, an SLRP (small leucine-rich proteoglycan) was elevated in HCM and AS samples and correlated with myocardial fibrosis determined noninvasively using cardiac magnetic resonance imaging. SLRPs regulate the kinetics, assembly, spatial organization, and stabilization of collagen fibrils and influence tissue fibrosis through interaction with TGF (transforming growth factor)-β and possibly via proinflammatory pathways.²⁸ Lumican mRNA has previously been identified in human heart and seems to be significantly upregulated in HCM and dilated cardiomyopathy compared with controls.²⁹ In a mouse model, left ventricular lumican protein levels were shown to increase after aortic banding and fell after debanding implying that production can be altered in response to external stimuli.³⁰ Stimulation of cardiac fibroblasts with recombinant glycosylated lumican, causes an increase in type I collagen, a decrease in activity of the collagen-degrading enzyme MMP (matrix metalloproteinase)-9, and an increase in the phosphorylation of fibrosis-inducing SMAD3.³¹ The stimulus to increased lumican production in HCM is unclear, but the direct link with molecules known to be important for cardiac remodeling and fibrosis and mechanical stress is of relevance to HCM where wall stress is increased by left ventricular outflow obstruction, asymmetrical hypertrophy, and cardiomyocyte disarray. Furthermore, the correlation we found with sarcomere mutations and late gadolinium enhancement suggests it may be important in disease pathogenesis.

Carbonic Anhydrase 3

The presence of CA in heart muscle is well established with cardiac ventricular myocytes expressing different isoforms in the cytosol (CA2), mitochondria (CA5), sarcolemma (CA4, CA14), and sarcoplasmic reticulum membrane (CA4, CA9, and CA14).³² However, the CA3 isoform has previously been reported at low levels in the myocardium.^33,34 In this study, we have demonstrated by 2 methods that CA3 is overexpressed in HCM but whether this is a response to cardiac hypertrophy in general remains to be determined. The positive correlation with heart failure markers (NT-proBNP and left atrial size) and other proteins (skeletal actin and FHL1) known to be upregulated in pathological cardiac hypertrophy and heart failure suggests this isoform may not be disease specific. In the heart, cellular or mitochondrial acidosis can impair contractility by reducing calcium sensitivity and compromising energetics. By transporting carbon dioxide out of cells into nearby capillaries acidosis is prevented and intracellular pH maintained. CA3 has not been specifically studied in human hearts, but there is emerging evidence that isoforms CA2 and CA4 are activated in both hypertrophic and dilated failing human hearts.³⁵ Carbonic anhydrase inhibition has been shown to prevent and revert cardiomyocyte hypertrophy in cultured myocytes.³⁶ Further studies are required to understand whether overexpression of CA is a protective response or whether CA activity could or should be pharmacologically modulated.

Limitations

As drug refractory patients with outflow tract obstruction requiring surgery, represent an advanced stage of HCM we cannot extrapolate our findings to early disease. The control samples were obtained from unused donor hearts, some of whom suffered catastrophic neurological injury which could have altered the tissue proteome. Nonetheless, the inclusion of AS samples as a positive control group in the phase 2 study gave further insight into whether novel findings might be disease specific or related to general downstream effects observed in hypertrophy and heart failure.

Global proteomic studies of human heart tissue are complicated by factors, such as disease state, tissue and genetic heterogeneity, medical comorbidities, and drug therapy. Approaches to overcome these problems include laser capture microdissection, use of animal models and induced pluripotent stem cell lines. To date, none of these studies have been undertaken in HCM. The whole tissue homogenate approach we used is similar to the approach that has been used in the 2-dimensional gel electrophoresis myocardial protein databases and comparative dilated cardiomyopathy studies.³⁷ Although multiple subcellular fractionation steps would have enhanced organelle isolation, it would still be difficult to reliably identify the cell of origin (eg, myocyte, fibroblast, or endothelial cell). We found no difference in expression of citrate synthase implying that changes were not just because of fewer mitochondria. After protein extraction and centrifugation, the insoluble pellet (containing extracellular proteins) was retained for further work. While we acknowledge the inherent limitations, we think this study still serves as an important basis to examine the HCM proteome

Although pathway analysis tools offer an insight into the underlying biology, there are inherent limitations in using a list of differentially expressed proteins (or genes) to make firm conclusions about disease mechanisms. As further experimental evidence is generated, the annotation knowledge base will grow, and it is likely pathway and network analysis will provide a better understanding of the downstream effects of altered proteins. While the increased or decreased expression of a protein does not necessarily determine whether pathways are upregulated or downregulated, the similarity between our results and previous transcriptomic studies provides supporting evidence at the protein level for an altered pattern of metabolic and structural proteins in HCM.

Conclusions

The myocardial proteome of HCM provides supporting evidence for dysregulation of metabolic and structural proteins, as well as proteins associated with muscle contraction, calcium regulation, and oxidative stress. The discovery that lumican is raised in HCM hearts provides insight into the myocardial fibrosis that characterizes this disease.

Acknowledgments

All data and metadata underlying the findings reported in this article will be deposited in the Proteomics Identifications (PRIDE) database public repository.

Footnotes