Whole mitochondrial genome sequencing of Malaysian patients with cardiomyopathy

Introduction

Cardiomyopathy (CMP) is commonly known to affect the heart muscle and its ability to pump blood around the body. In the early stages of CMP, patients are commonly asymptomatic. However, with disease progression, symptoms of heart failure such as exertional dyspnoea, fatigue and pedal oedema are often presented. CMP is one of the common causes of heart failure (Dokainish et al., 2017). There are several subtypes of CMP, including hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), restrictive cardiomyopathy (RCM), and left ventricular non-compaction cardiomyopathy (LVNC) (Elliott et al., 2008). HCM and DCM are the two most common subtypes of CMP. HCM is characterised by left ventricular hypertrophy due to cardiomyocyte hypertrophy, whereas DCM is characterised by enlarged cardiac chambers and systolic dysfunction. The estimated prevalence of HCM in the general population was 1 in 500 to 200 individuals (Semsarian et al., 2015), while the prevalence of DCM was 1 in 250 individuals (Hershberger, Hedges & Morales, 2013). In fact, the prevalence may be higher as the estimations made were probably conservative (McKenna & Judge, 2021). Nevertheless, the exact aetiology of CMP remains inconclusive. It is generally accepted that environmental and genetic factors play vital roles in the CMP onset. Approximately 20–60% of CMP cases were reported as familial (Marian & Braunwald, 2017; Sweet, Taylor & Mestroni, 2015). Mitochondrial inheritance was known to contribute to the onset of CMP.

Mitochondria are the powerhouses of cells because of their ability to generate adenosine triphosphate (ATP) molecules through oxidative phosphorylation (OXPHOS) carried out by electron transport chain (ETC) complexes. Mitochondria are responsible for cell death regulation by releasing cytochrome c to initiate necrosis and apoptosis under stress conditions (Bock & Tait, 2020). Cardiomyocytes are among the high energy demand cells necessitated by the significant workload of blood pumping. Mitochondrial dysfunction can result in inadequate ATP production, leaving the energy demand of cardiomyocytes unsatisfied and ultimately leading to CMP (Ramaccini et al., 2020) or the worst, heart failure (Chistiakov et al., 2018). The mitochondrion is one of the sites where reactive oxygen species (ROS) are generated as a by-product of OXPHOS. Defective mitochondria also increase ROS production, triggering the opening of mitochondrial permeability transition pores and stimulating the release of cytochrome c that activates cardiac cell death (Zorov et al., 2000). The destruction of cardiomyocytes can cause CMP by deteriorating heart pumping ability.

Mitochondrial DNA (mtDNA) is a small circular double-stranded DNA with 16,569 bp that can only be transmitted maternally. It contains 37 genes that encode two ribosomal RNAs (rRNAs), 22 transfer RNAs (tRNAs), and 13 protein complexes that are involved in ETC (Andrews et al., 1999). The only non-coding region in mtDNA is the D-loop that is located at positions 16,024 to 576, serving as the origin of the transcription and promoter region.

Mutations in mtDNA were reported in both HCM and DCM patients. The m.3243A>G mutation in MT-TL1 gene is among the most common mitochondrial tRNA mutations reported in CMP (Hollingsworth et al., 2012). Studies have reported m.1555A>G mutation in the MT-RNR1 gene which encodes 12S rRNA in a patient with maternally inherited CMP (Santorelli et al., 1999). Mutations in protein-coding genes such as MT-ND1 and MT-CYB may directly impact the OXPHOS as these genes are vital to encode subunits of complexes involved in ETC. The m.3395A>G mutation in MT-ND1 gene (Alila et al., 2016) and m.15482T>C mutation in MT-CYB (Hagen et al., 2013a) were reported in HCM and DCM patients, respectively. However, mitochondrial haplogroups were also frequently reported to be associated with CMP. A study in Spain reported that individuals with haplogroup H had a higher risk of developing DCM (Fernandez-Caggiano et al., 2013). Another study conducted in Denmark identified haplogroups H and HV as the risk factors for HCM development. In contrast, individuals with haplogroups J and UK had a lower risk to develop HCM (Hagen et al., 2013b).

Most studies on mitochondrial mutations in CMP patients were carried out in Western countries, rarely in Asia, and none in Malaysia. Therefore, this study is important as the first study on the full mutation spectrum of the mitochondrial genome in Malaysian CMP patients. The findings in the present study are irrefutably fundamental to fill the gap of knowledge for the mtDNA spectrum in Malaysian CMP patients.

Materials and Methods

Sequence analysis

FASTQ files were input to Illumina DRAGEN FastQC + MultiQC for initial quality checking of the sequencing data. FastQC generated the overview of quality parameters such as input reads quality score, GC content, sequence length distribution, and duplication levels, while MultiQC allowed summary and visualisation of results across all samples. The reference used was Hg38 Alt-Aware, with HLAs. Next, the FASTQ files were input to Illumina mtDNA Variant Processor. Adapters were trimmed and reads with short amplicon were removed. BWA-MEM algorithm in Burrows–Wheeler Alignment Tool was used to align and map the sequence read to the revised Cambridge Reference Sequence (rCRS) (Li, 2013). Only reads with Q-score >30, analysis threshold >10%, interpretation threshold >25% and read count >10 were included for variant calling. The resulting VCF files were then input into Illumina mtDNA Variant Analyzer for visualisation of coverage and variant points position in comparison to rCRS. Variants were annotated with BaseSpace Variant Interpreter (v2.11), and Universal mtDNA Variant Converter and One Stop Annotation (mvTool v.6) in MSeqDR (Shen et al., 2018). The variants were compared and cross-checked with reliable databases that are still being maintained and constantly updated such as MITOMAP, Genome Aggregation Database (gnomAD), HmtDB, HmtVar, MSeqDR and ClinVar for their clinical significance. MITOMAP and gnomAD were referred for the allele frequencies of variants in various populations. Haplogroup assignment was conducted with HaploGrep 2 to determine the haplogroup of each subject based on their mitochondrial sequence (Weissensteiner et al., 2016). Circa (http://omgenomics.com/circa/) was used to generate the visualisation of variants distribution across the whole mitochondrial genome. The overview of the whole mitochondrial genome sequencing and analysis in this study is shown in Fig. 1.

Results

Novel variants

Variants that were not reported in MITOMAP and/or gnomAD were considered as novel variants in this study. A total of 18 novel variants were detected in the CMP patients (Table 2; Fig. 2), including 10 (55.6%) non-synonymous, six (33.3%) in tRNA and two (11.1%) in rRNA. The non-synonymous variants were: two heteroplasmic variants in MT-ND1 (m.3511A>C and m.3577A>C), one heteroplasmic variant in MT-CO1 (m.7416T>C), one heteroplasmic variant in MT-ATP6 (m.8573G>C), one homoplasmic variant in MT-CO3 (m.9318C>A), one homoplasmic variant in MT-ND3 (m.10144G>T), two homoplasmic variants in MT-ND4 (m.11916T>A and m.11918T>G) and two homoplasmic variants in MT-ND5 (m.12595A>G and m.12663C>A). For rRNA, one homoplasmic variant was found in MT-RNR1 (m.1008C>T), while the remaining five variants were found heteroplasmic in MT-RNR2 (m.1731A>G, m.2465T>A, m.2473A>C, m.2475T>C, m.2479C>A). The variants found in tRNA were m.4313T>A in MT-TI and m.5551C>T in MT-TW; both in the homoplasmic state. The in-silico analysis of pathogenicity for the novel variants is illustrated in a heatmap (Fig. 3; Table S7). Interestingly, three variants (m.8573G>C, m.11916T>A and m.11918T>G) were predicted to have a strong negative impact on the protein function by most of the in-silico predictors; thus, their effects on the onset of CMP could not be neglected.

Table 2:

List of novel mitochondrial variants detected in Malaysian CMP patients.

Variant	Gene	Complex	AA change	CMP patients
				Frequency, n (%)			Haplogroup
				DCM (N = 87)	HCM (N = 58)	Total (N = 145)
m.1108C>T	MT-RNR1	–	–	1 (1.1)	0 (0.0)	1 (0.7)	M12a2
m.1731A>G	MT-RNR2	–	–	0 (0.0)	1 (1.7)	1 (0.7)	D4
m.2465T>A	MT-RNR2	–	–	3 (3.4)	1 (1.7)	4 (2.8)	R8a1, M*, U1a1c1d
m.2473A>C	MT-RNR2	–	–	0 (0.0)	1 (1.7)	1 (0.7)	F1f
m.2475T>C	MT-RNR2	–	–	1 (1.1)	0 (0.0)	1 (0.7)	M71a1a
m.2479C>A	MT-RNR2	–	–	1 (1.1)	0 (0.0)	1 (0.7)	F1e3
m.3511A>C	MT-ND1	I	Thr → Pro	1 (1.1)	0 (0.0)	1 (0.7)	B4c1b2a2
m.3577A>C	MT-ND1	I	Met → Leu	1 (1.1)	0 (0.0)	1 (0.7)	M18
m.4313T>A	MT-TI	–	–	1 (1.1)	0 (0.0)	1 (0.7)	D5b4
m.5551C>T	MT-TW	–	–	1 (1.1)	0 (0.0)	1 (0.7)	F1a3b
m.7416T>C	MT-CO1	IV	Phe → Leu	1 (1.1)	0 (0.0)	1 (0.7)	B5a1a
m.8573G>C	MT-ATP6	V	Gly → Ala	3 (3.4)	2 (3.4)	5 (3.4)	M8a2, B5a1
m.9318C>A	MT-CO3	IV	His → Asn	1 (1.1)	0 (0.0)	1 (0.7)	M3a1
m.10144G>T	MT-ND3	I	Gly → Val	1 (1.1)	0 (0.0)	1 (0.7)	M53
m.11916T>A	MT-ND4	I	Phe → Tyr	1 (1.1)	0 (0.0)	1 (0.7)	R30b2a
m.11918T>G	MT-ND4	I	Ser → Ala	1 (1.1)	0 (0.0)	1 (0.7)	F2b1
m.12595A>G	MT-ND5	I	Met → Val	0 (0.0)	1 (1.7)	1 (0.7)	D4h3
m.12663C>A	MT-ND5	I	Asn → Lys	0 (0.0)	1 (1.7)	1 (0.7)	B4a1a

DOI: 10.7717/peerj.13265/table-2

Note:

AA, amino acid; variant(s) in bold is highly potential to be pathogenic.

Discussion

In recent years, numerous mtDNA variants have been identified and reported to be associated with CMP (Govindaraj et al., 2014; Kargaran et al., 2020). However, only limited number of studies were carried out in Asia, and none was conducted in Malaysia. This is the first whole mitochondrial genome sequencing study to elucidate the role of mtDNA variants in Malaysian CMP patients. The interpretation of the pathogenicity of mtDNA variants can be challenging due to conflicting reports from different laboratories. Guidelines developed by The American College of Medical Genetics (ACMG) and the Association of Molecular Pathology (AMP) are essential to enhance consistency in classifications of variants (Richards et al., 2015). Recently, more precise and specific guidelines for mtDNA variants interpretation and specification of ACMG/AMP standards were developed (McCormick et al., 2020), suggesting the utilisation of comprehensive databases such as MITOMAP, HmtDB, HmtVar, MSeqDR and ClinVar for mtDNA variant assessment. In favour of concise result reporting, our study utilised the suggested databases during data analysis for variants detected in our CMP patient cohort.

In the present study, 18 novel variants were observed, of which m.8573G>C, m.11916T>A and m.11918T>G variants were predicted to be potentially pathogenic via in-silico analysis. These three variants were not reported in gnomAD. Variant m.8573G>A, which was at the same position as one of the novel variants (m.8573G>C) in this study, was reported to be benign in ClinVar (Ganetzky et al., 2019). However, the novel variant m.8573G>C found in this study was a transversion mutation, which has more a detrimental effect due to the change of ring structure in nucleic acid compared to transition mutation in m.8573G>A (Lyons & Lauring, 2017). Variant m.8573G>C caused a substitution of glycine to alanine at a highly conserved region (CI = 93.3%) in MT-ATP6. Although glycine and alanine are both non-polar, the replacement of glycine at the conserved region could alter the protein structure. Glycine is the only amino acid with hydrogen but not carbon as its side chain, making it to be extra versatile as it can reside in difficult part of protein, such as beta turn and loop in secondary structure (Betts & Russell, 2003). Five patients (three DCM and two HCM) in this study harboured heteroplasmic m.8573G>C variant, with loads ranging from 6.75% to 12.66%. Most disease-associated pathogenic mtDNA variants tend to exist as heteroplasmic (Ye et al., 2014). Interestingly, four patients had the same haplogroup, which was B5a1, whereas one patient had a different haplogroup, which was M8a2, suggesting the pathogenicity of this variant regardless of haplogroup background.

On the other hand, a DCM patient was observed to harbour m.11916T>A variant, which caused a change from phenylalanine to tyrosine at 93.3% conserved region in MT-ND4. Although phenylalanine and tyrosine are both aromatic and non-polar, tyrosine has an extra –OH in the phenyl ring which would enhance the solubility. The change in hydrophilicity would alter the structure of the protein. Another variant from MT-ND4, m.11918T>G was observed in a DCM patient. This variant caused a change of serine to alanine. Serine is a small, polar amino acid that often acts as an active site due to the presence of R-group. In contrast alanine is non-polar, hydrophobic and with non-reactive side chain. Although the alteration occurred at a region conserved only in mammals and insect (CI = 80.0%), the substitution of amino acid with completely different properties could impact the folding and function of protein, leading to the malfunction of Complex I and reduced energy production that caused CMP.

According to McCormick et al. (2020), a variant with allele frequency <0.00002 fulfils ACMG pathogenicity code PM2 and coding variant that is predicted to have a deleterious effect by APOGEE (score >0.5 = pathogenic) fulfils ACMG pathogenicity code PP3. All the three variants mentioned above fulfilled the criteria for PM2 and PP3. At present, these variants could only be classified as uncertain significance because they did not meet the criteria for likely pathogenic and pathogenic classification. However, in view of remarkable results in the in-silico analysis and the fact that these variants are located in highly conserved regions, functional studies to elucidate the structural and biochemical effect of these variants are needed to denote their negative consequences to the mitochondrial function.

This study also revealed 17 variants that were associated with CMP and/or mitochondrial diseases. Among the 17 variants, two variants, m.1555A>G and m.11778G>A, were reported with confirmed pathogenicity. In this study, there was one HCM patient with haplogroup E1a2 who was observed to harbour the homoplasmic m.1555A>G variant. This variant was frequently reported in sensorineural hearing loss (SNHL) (Jiang et al., 2015; Mutai et al., 2017). Besides, this variant also altered the secondary protein structure of 12S ribosomal RNA, creating a binding site for aminoglycoside and causing ototoxicity (Prezant et al., 1993). Furthermore, the rate of protein synthesis was also disrupted due to the alteration in protein conformation and translation (Hobbie et al., 2008). Biochemical test also showed that cell harbouring this variant had decreased ATP production and increased ROS production (Wrzesniok et al., 2013). All the biochemical defects above could impact the protein synthesis needed for OXPHOS in mitochondria and lead to energy deficiency in cardiomyocytes. This variant was reported heteroplasmic in different type of tissues in a woman with maternally inherited CMP (Santorelli et al., 1999). Biochemical studies from homogenates of muscle and skin also revealed a decrease in complex activities in the respiratory chain. Furthermore, the m.1555A>G variant has been associated with a paediatric patient with SNHL and DCM (Skou et al., 2014). On the other hand, another paediatric patient with DCM was also reported to harbour the m.1555A>G variant in homoplasmic state. The proband’s sister and brother died at their infancies due to DCM as well (Alila-Fersi et al., 2017).

Another pathogenic variant with confirmed status observed in this study was m.11778G>A, which was reported in 90% of Leber’s hereditary optic neuropathy (LHON) patients of Asian descent (Jia et al., 2006). LHON is an ophthalmologic disease that also often manifests systemic abnormalities such as CMP (Finsterer & Zarrouk-Mahjoub, 2016). This missense variant resulted in the substitution of arginine to histidine, which share the same properties as a basic polar amino acid. However, the change occurred in a 100% conserved region of MT-ND4; therefore, the impact should not be underestimated.

Cybrid studies had revealed that m.11778G>A caused reduction in maximal respiration rate and Complex I activity (Brown et al., 2000). Besides, cybrids with m.11778G>A also showed increased ROS levels and apoptosis rate (Giordano et al., 2011). A study from Japan reported that a woman with LHON and CMP harboured m.11778G>A (Watanabe, Odaka & Hirata, 2009). Strikingly, the DCM patient in our study who harboured the m.11778G>A variant was also found to carry the only heteroplasmic variant, m.9804G>A, among the 17 disease-associated variants. Although the pathogenicity of m.9804G>A was uncertain due to conflicting evidence, in-silico analysis from PROVEAN, MutPred, Mitoclass.1, CADD, PhD-SNP and APOGEE predicted it to be pathogenic and have deleterious effect. Furthermore, m.9804G>A is located in a highly conserved region (CI = 93.3%) of MT-CO3. It was plausible that both variants can act synergically to cause malfunction of Complex I (MT-ND4) and IV (MT-CO3) and disrupt the ETC, which then reduced the ATP production and elevated the ROS level. Energy deficiency and apoptosis are detrimental to cardiomyocytes and may result in CMP.

Among the 17 variants, the only tRNA variant detected in a HCM patient in the study, m.4316A>G, was predicted to be possibly benign by MitoTIP. However, it is located at A58 of the T-loop in tRNA^Ile encoded by MT-TI, which is also highly conserved among 15 species (CI = 93.3%). According to the Mamit-tRNA database, the adenine is conserved in 31 out of 34 Euarchontoglires organisms. Besides, m.4316A>G was reported in a patient with HCM and profound hearing loss (Chamkha et al., 2011). It is also located next to m.4317A>G, which was first described in fatal infantile CMP (Tanaka et al., 1990), followed by multiple mitochondrial diseases such as deafness and myopathy (Zheng et al., 2020). Moreover, variant m.4317A>G was depicted to alter the structure by rearrangement of T-arm in tRNA^Ile. Study has shown the alternation affects the aminoacylation and thus causing impaired tRNA conformation and stability (Meng et al., 2018). The same study also demonstrated the synergy effect of m.4317A>G with m.1555A>G in deafness, whereby the cell lines carrying both variants exhibited significant deterioration of mitochondrial function than cell lines carrying only m.1555A>G variant (Meng et al., 2018). All these outcomes suggest the potential pathogenicity of m.4316A>G in CMP.

The m.9025G>A variant identified in this study is localised at a conserved region among 15 species with CI = 100%. The variant caused the change of non-polar glycine to polar serine at position 167. The alteration also introduced changes in the interaction between neighbouring residues whereby serine forms polar contacts with isoleucine at 164 and methionine at 171. On top of that, glycine at position 167 plays its role as a residue in the trans-membrane functional domain of ATP synthase membrane subunit 6 protein. The variant was detected in members of a family with HCM (Kargaran et al., 2020). Besides CMP, the variant was also reported in patients with mitochondrial disorder such as Leigh syndrome, motor neuropathy and 3-methylglutaconic aciduria type IV (López-Gallardo et al., 2014). The significance of m.9025G>A variant in pathogenesis of CMP was further supported by a functional study which observed a decreased baseline respiration and increased ROS production, thereby promoting apoptosis (Ganetzky et al., 2019). Besides, m.9025G>A was predicted to be ‘deleterious’, ‘pathogenic’ and ‘damaging’ by 11 in-silico predictors in this study. All these findings suggest that the change of amino acid with different properties by m.9025G>A would impair the active binding site of a protein and hinder the OXPHOS that is responsible for ATP production, followed by insufficient energy supply to the cardiomyocytes and consequently leading to CMP (Heidari et al., 2020).

Despite observing novel and CMP and/or mitochondrial disease-associated variants among the Malaysian CMP patients, several limitations in this study warrant careful interpretation of the findings. First, the control cohort was not included in the study. Healthy age and ethnically matched controls are vital as a comparison group to exclude the founder variant in certain populations during the analysis. Second, the mitochondrial genome spectrum in family members of study subjects was not evaluated; hence, screening of segregation and penetrance of variant cannot be performed. Third, genetic analysis of nuclear-encoded genes such as sarcomeric genes was absent, thus the potential effect of nuclear mutation among the patients is not known.

Conclusion

In summary, the whole mitochondrial genome sequencing of CMP patients in Malaysia revealed a total of 1,077 variants. After excluding ancestral variants, haplogroup markers and variants located in the control region, there were 512 variants remaining for further analysis. Amongst these variants, 18 were novel and 17 were CMPs and/or mitochondrial disease-associated variants. Three novel variants (m.8573G>C, m.11916T>A and m.11918T>G) were potentially pathogenic, as in-silico analysis predicted them to have deleterious effects. Two confirmed pathogenic variants (m.1555A>G and m.11778G>A) were also observed among the CMP patients in this study. Although the results are preliminary and the functional significance of the variants is yet to be elucidated, the findings of this study can provide fundamental insight into the distribution of mitochondrial genome variants in Malaysian CMP patients.

Supplemental Information