Original research
Using MutPred derived mtDNA load scores to evaluate mtDNA variation in hypertension and diabetes in a two-population cohort: The SABPA study

https://doi.org/10.1016/j.jgg.2016.12.003Get rights and content

Abstract

Mitochondrial DNA (mtDNA) variation has been implicated in many common complex diseases, but inconsistent and contradicting results are common. Here we introduce a novel mutational load hypothesis, which also considers the collective effect of mainly rare variants, utilising the MutPred Program. We apply this new methodology to investigate the possible role of mtDNA in two cardiovascular disease (CVD) phenotypes (hypertension and hyperglycaemia), within a two-population cohort (n = 363; mean age 45 ± 9 yrs). Very few studies have looked at African mtDNA variation in the context of complex disease, and none using complete sequence data in a well-phenotyped cohort. As such, our study will also extend our knowledge of African mtDNA variation, with complete sequences of Southern Africans being especially under-represented. The cohort showed prevalence rates for hypertension (58.6%) and prediabetes (44.8%). We could not identify a statistically significant role for mtDNA variation in association with hypertension or hyperglycaemia in our cohort. However, we are of the opinion that the method described will find wide application in the field, being especially useful for cohorts from multiple locations or with a variety of mtDNA lineages, where the traditional haplogroup association method has been particularly likely to generate spurious results in the context of association with common complex disease.

Introduction

Cardiovascular disease (CVD) is an umbrella term that encompasses several distinctive disease phenotypes incisive of myocardial infarction, stroke, congenital heart disease and risk factors such as hypertension, and diabetes type 2 (Mensah, 2013). Discrepancies between different population groups exist in the onset, development and outcome of CVDs (Okin et al., 2011, Moran et al., 2013, Owolabi et al., 2014). Although environmental and lifestyle factors play a role in the risk of developing CVD, genetic factors are likely to account for some of the observed discrepancies in CVD onset, and particularly progression/outcome of disease (Achilli et al., 2011). Nuclear DNA risk factors have not been able to account for all the observed inconsistencies among different population groups (Kaufman et al., 2015). Thus, other sources of variation, such as epigenetics and mitochondrial DNA (mtDNA) variation might account for some of the missing heritability in CVDs.

mtDNA mutations are a common cause of inherited disease (Gorman et al., 2015). Both diabetes and cardiomyopathy are frequent symptoms in primary mitochondrial disease (Taylor et al., 2003, Yarham et al., 2010), especially in patients with the mtDNA m.3243A>G mutation (Hollingsworth et al., 2012). While clinically manifesting mtDNA mutations are a recognised cause of human disease, many studies have suggested a role for common mtDNA variants (Hernstadt and Howell, 2004, Wallace, 2010) and the combined effect of rare population variants (Elson et al., 2006) in common complex disease. For CVDs specifically, a more recent Framingham Heart sub-study showed significant associations between several population as well as rare variants, and variation in blood pressure and fasting blood glucose levels (Liu et al., 2012). Cardena et al., 2014, Cardena et al., 2016 found the hypertensive phenotype to be more prevalent among heart failure patients with an African mitochondrial haplogroup than among those with European haplogroups, in an admixed Brazilian cohort.

Human mtDNA codes for 13 essential polypeptide components of the mitochondrial oxidative phosphorylation (OXPHOS) system. mtDNA undergoes strict maternal inheritance, resulting in the absence of bi-parental recombination (Elson et al., 2001), and has a high mutation rate (Tuppen et al., 2010). As such, the evolution of mtDNA is characterised by the emergence of distinct lineages (or haplogroups) (Hernstadt et al., 2002). This results in high levels of mtDNA variation at the population level despite its rather small size, which is also illustrated by the large number of sub-haplogroups (van Oven and Kayser, 2009). Africa (haplogroups L0‒L6) has the highest levels of nuclear and mitochondrial genetic diversity (Salas et al., 2002); this diversity was reduced by population bottlenecks in groups migrating from Africa (Manica et al., 2007). However, variation in African populations, especially in relation to disease, is still under studied when compared to variation in super clade N (Cavadas et al., 2015, Gurdasani et al., 2015). Only about 12% of mtDNA sequences on GenBank are of African lineages. In terms of mtDNA ancestry, the African haplogroup L3 incorporates super clades M and N, which encompass all the European and Asian haplogroups (Hernstadt et al., 2002, van Oven and Kayser, 2009, Rosa and Brehem, 2011).

Given the unique inheritance pattern of mtDNA, it is worth considering the hypothesis that might link mtDNA population variation to common complex disease. Firstly, studies showing significant associations of several disease phenotypes with specific mtDNA haplogroups have been published, suggesting that one or more common population variants may modify risk or outcome of disease. Secondly, the high mutation rate of mtDNA frequently results in the same variant being present more than once on the phylogeny (Hernstadt et al., 2002). Showing that such a variant modifies risk or alters the course of disease (Yu et al., 2008) in two different haplogroups or global populations, would be excellent evidence of its role in disease. The third possibility is that rare mtDNA mutations might have an effect, either at the individual level with there being synergist effects of multiple rare variants in patients, or rare variants might just be seen more frequently in the patient group.

The first hypothesis mentioned above, known as the haplogroup association hypothesis, has to date been the classic approach when considering the role of mtDNA variation in common disease. These studies have been controversial due to their low repeatability; often the association is not detected in a second cohort, or an association with a different haplogroup might be uncovered (Salas and Elson, 2015). Some studies have taken a two-cohort approach in an attempt to address these problems (Elson et al., 2006, Chinnery et al., 2010). Even so, this approach has proved unsuitable for many studies with cohorts that do not have the large numbers required for well-powered haplogroup association studies (Samuels et al., 2006).

In this study, we took an alternative approach, using MutPred pathogenicity scores to derive “mutational loads”. This is a new version of the “mutational load” hypothesis proposed by Elson et al. (2006), which looked at the frequency of mildly deleterious (rare) variants in patient and control groups, to test the possibility of a cumulative effect of these variants. After calculating “MutPred mutational loads” by summing the MutPred scores generated for each of the non-synonymous variants on an individual's mtDNA, we adjusted these for the position of the sequence in the phylogeny, ultimately calculating “MutPred adjusted loads”. Many different tools for predicting pathogenicity of mtDNA variants exist, but in a comparative study by Thusberg et al. (2011), MutPred and SNPs&GO outperformed all other methods, which included PolyPhen2, SIFT and SNAP. The MutPred program has also been widely validated in the context of mtDNA variation (Pereira et al., 2011). The MutPred algorithm incorporates elements of the SIFT algorithm, and assigns a pathogenicity score between 0 and 1, with zero being a benign substitution. A pathogenic score above 0.5 can be considered an “actionable hypothesis”, while a score above 0.75 can be considered a “confident hypothesis” (Li et al., 2009). As there is selection against mildly deleterious variants at the population level (Elson et al., 2004, Soares et al., 2013), variants with MutPred scores above the “actionable hypothesis” threshold (0.5) are less likely to define major haplogroups; rather, they are more likely to be rare and seen on very recent branches (twigs) of the phylogeny (Pereira et al., 2011). Because the number of low-scoring, low-impact, common variants differs greatly among different population groups, their inclusion in the calculation of mutational loads could introduce noise that is unlikely to have phenotypic impact. By excluding variants with a pathogenicity score below 0.5, and thus most common population variants, before calculating MutPred adjusted loads, we aim to highlight the impact of rare variants while reducing the effect of population stratification.

Although mtDNA variation involvement in CVDs has previously been investigated (Chinnery et al., 2010), these studies have been predominantly in Caucasian European populations. Very few such studies have been published on African populations or those of African descent; most were on small sized cohorts, and often focused on specific previously reported variants (Khogali et al., 2001, Robinson et al., 2004, Ameh et al., 2011). To assume that findings from studies in European populations can easily be extrapolated to be less investigated, genetically diverse African population groups would be short sighted (van der Westhuizen et al., 2015), as even clinically proven disease causing mutations as well as the underlying spectrum of mutations are known to have differing impacts in Africans (van der Walt et al., 2012, van der Westhuizen et al., 2015).

In this study, we used the two-population Sympathetic Activity and Ambulatory Blood Pressure in Africans (SABPA) cohort (Malan et al., 2015) (Table 1). This is a South African cohort which consisted of 409 Black and Caucasian South African participants from the same geographical area (Rosa and Brehem, 2011, Salas and Elson, 2015). Although the participants were matched for gender, age and socio-economic status, a significantly higher percentage of Black participants, compared to Caucasians participants, with optimal blood pressure at the start of the study, developed hypertension within five years (Schutte et al., 2012, Hamer et al., 2015). A wide range of clinical and phenotypical analysis data are available for this cohort, including the golden standard 24 h ambulatory blood pressure monitoring (24 h ABPM) measurements and a measurement of hyperglycaemia (HbA1c, glycated haemoglobin), which are routinely used to identify hypertension and diabetes respectively in a clinical setup.

Using this unique cohort, we aim to determine if mtDNA variation, using the MutPred adjusted load as defining parameter, is different in those with hypertension or hyperglycaemia when compared to those without. As such, we present a new and updated method for attempting to associate mtDNA variation with a complex trait. This method would be applicable to all mtDNA association studies and be less affected by population stratification.

Section snippets

Results

After some samples fall-out due to technicalities such as sample unavailability and insufficient DNA extraction, 194 participants with macro-haplogroup L (haplogroups L0‒L4) and 169 participants with macro-haplogroup MN (haplogroups M, N, R, B, H, I, J, K, T, U and W) were used in this study. In Table 1, some of the most important phenotypical measurements (age, body mass index, blood glucose levels and blood pressure) are summarised. Immediately, large differences in both systolic and

Discussion

From the landmark Framingham Heart Studies (Mahmood et al., 2014), several risk factors have been identified as significant predictors of hypertension. In the SABPA cohort also, these factors, which included gender and population group, significantly predicted blood pressure (Hamer et al., 2015, Malan et al., 2015). Disparities in CVD onset and development between different population groups are often linked to gross socio-economic inequality (Kaufman et al., 2015). However, it is important to

Cohort recruitment and sample collection

We included participants of the SABPA cohort in this study. Details on participant recruitment, and sample and data collection have been published elsewhere (Hamer et al., 2015, Malan et al., 2015). Inclusion criteria for the SABPA study were urban Black and Caucasian male and female teachers from South Africa (n = 409), with similar socio-economic status, aged 20–65 years, from the North-West Province (Malan et al., 2015). Participants were enrolled in the project in 2008–2009, having their

Acknowledgments

We acknowledge the Faculty of Natural Sciences of the North-West University for contributing to funding and Thermo Fisher South Africa for providing additional technical resources to this study. The present study was partially funded by the South African National Research Foundation; Medical Research Council; ROCHE Diagnostics; North-West University, South Africa; as well as the Metabolic Syndrome Institute, France. We would also like to acknowledge funding support from the Royal Society and

References (53)

  • A. Moran et al.

    The epidemiology of cardiovascular diseases in Sub-Saharan Africa: the global burden of diseases, injuries and risk factors 2010 Study

    Prog. Cardiovasc. Dis.

    (2013)
  • L. Pereira et al.

    Comparing phylogeny and the predicted pathogenicity of protein variations reveals equal purifying selection across the global human mtDNA diversity

    Am. J. Hum. Genet.

    (2011)
  • T. Polzin et al.

    On Steiner trees and minimum spanning trees in hypergraphs

    Oper. Res. Lett.

    (2003)
  • A. Salas et al.

    The making of the African mtDNA landscape

    Am. J. Hum. Genet.

    (2002)
  • A. Salas et al.

    Mitochondrial DNA as a risk factor for false positives in case-control association studies

    J. Genet. Genomics

    (2015)
  • D.C. Samuels et al.

    The power to detect disease associated with mitochondrial DNA haplogroups

    Am. J. Hum. Genet.

    (2006)
  • R.W. Taylor et al.

    A homoplasmic mitochondrial transfer ribonucleic acid mutation as a cause of maternally inherited hypertrophic cardiomyopathy

    J. Am. Coll. Cardiol.

    (2003)
  • H.A. Tuppen et al.

    Mitochondrial DNA mutations and human disease

    Biochim. Biophys. Acta

    (2010)
  • A. Achilli et al.

    Mitochondrial DNA backgrounds might modulate diabetes complications rather than T2DM as a whole

    PLoS One

    (2011)
  • J. Ameh et al.

    The search for mitochondrial tRNALeu(UUR) A3243G mutation among type 2 diabetes mellitus patients in the Nigerian population

    Afr. J. Biotechnol.

    (2011)
  • American Diabetes Association

    Standards of medical care in diabetes

    Diabetes Care

    (2010)
  • R.M. Andrews et al.

    Reanalysis and revision of the Cambridge reference sequence for human mitochondrial DNA

    Nat. Genet.

    (1999)
  • H.-J. Bandelt et al.

    Median-joining networks for inferring intraspecific phylogenies

    Mol. Biol. Evol.

    (1999)
  • H.-J. Bandelt et al.

    Mitochondrial portraits of human populations

    Genetics

    (1995)
  • M.M. Cardena et al.

    Mitochondrial and genomic ancestry are associated with etiology of heart failure in Brazilian patients

    J. Hum. Hypertens.

    (2016)
  • B. Cavadas et al.

    Fine time scaling of purifying selection on human nonsynonymous mtDNA mutations based on the worldwide population tree and mother-child pairs

    Hum. Mutat.

    (2015)
  • Cited by (0)

    1

    These authors contributed equally to this work.

    View full text