Thinking outside the nucleus: mitochondrial DNA copy number in health and disease

Abstract

Mitochondrial DNA copy number (mtDNA-CN) is a biomarker of mitochondrial function and levels of mtDNA-CN have been reproducibly associated with overall mortality and a number of age-related diseases, including cardiovascular disease, chronic kidney disease, and cancer. Recent advancements in techniques for estimating mtDNA-CN, in particular the use of DNA microarrays and next-generation sequencing data, have led to the comprehensive assessment of mtDNA-CN across these and other diseases and traits. The importance of mtDNA-CN measures to disease and these advancing technologies suggest the potential for mtDNA-CN to be a useful biomarker in the clinic. While the exact mechanism(s) underlying the association of mtDNA-CN with disease remain to be elucidated, we review the existing literature which supports roles for inflammatory dynamics, immune function and alterations to cell signaling as consequences of variation in mtDNA-CN. We propose that future studies should focus on characterizing longitudinal, cell-type and cross-tissue profiles of mtDNA-CN as well as improving methods for measuring mtDNA-CN which will expand the potential for its use as a clinical biomarker.

Introduction

The mitochondrion is a double membraned organelle that produces over 90% of the cell’s chemical energy via respiration. Mitochondria have essential roles in the production of adenosine triphosphate (ATP), in the regulation of bioenergetic processing, in multiple homeostatic, apoptotic and signaling pathways, (Wallace, 1992) and in the biosynthesis of macromolecules, such as nucleotides, lipids, heme and iron-sulfur clusters (Vakifahmetoglu-Norberg et al., 2017).

Mitochondrial dysfunction, generally characterized as a loss of efficiency in oxidative phosphorylation, is a hallmark of aging and a variety of chronic diseases (Nicolson, 2014). Mitochondrial dysfunction results in inefficient cellular energy production and in increased levels of reactive oxygen species (ROS) which may damage lipids, proteins and nucleic acids (Guyatt et al., 2017). Mitochondrial dysfunction also affects the expression of nuclear genes involved in metabolism, growth, differentiation, and apoptosis (Pello et al., 2008). All these changes may explain the contribution of mitochondrial dysfunction to chronic and complex human diseases.

A major limitation to the routine evaluation of mitochondrial dysfunction in clinical practice is the lack of reliable measures of mitochondrial dysfunction available for clinical use. Mitochondrial DNA copy number (mtDNA-CN) is a promising biomarker of mitochondrial dysfunction that has the potential to become widely available in clinical practice. Other measures of mitochondrial dysfunction, including cell culture-based methods such as the Seahorse XF Cell Mito Stress Test (Agilent, USA), are optimized in vitro, do not make use of pre-existing datasets and cannot be scaled-up for widespread use. In this review, we describe the rationale for using mtDNA-CN as a biomarker of mitochondrial dysfunction, the issues associated with its measurement and interpretation in clinical and population studies, and the evidence linking variation in mtDNA-CN to human health disorders, including cardiovascular disease (Ashar et al., 2017; Dai et al., 2012), chronic kidney disease (Tin et al., 2016), cancer (Taylor and Turnbull, 2005), and all-cause mortality (Ashar et al., 2015; Mengel-From et al., 2014). mtDNA-CN is also modified in patients with primary mitochondrial disease, often due to mutations in genes involved in mtDNA integrity (Grady et al., 2018). Increases in mtDNA-CN in primary mitochondrial diseases may represent a compensatory response to attenuate the effect of the primary mtDNA mutation (Bianco et al., 2018). The identity and mechanisms behind mtDNA-CN variation in primary mitochondrial diseases are not discussed in this review.

The mitochondrial genome

The mitochondrial genome (mtDNA) is a circular, intron-free, double-stranded, haploid DNA strand 16.5 kb in length that encodes 37 genes (Raffoul et al., 2012). The organization of the mammalian mitochondrial genome is highly conserved (Clayton, 1992). The human mtDNA encodes 13 essential genes involved in the oxidative phosphorylation system, while the remaining genes play essential roles in assembling amino acids into functional proteins.

The growth, movement, division and fusion of mitochondria is highly regulated (Yamada et al., 2016). Division of mitochondria is important for proper organelle functioning, distribution and turnover (Yamada et al., 2016). The number of mitochondria in each cell changes due to fission and fusion of mitochondria, two processes that play a key role in maintaining mitochondrial function. Since the total number of mitochondria in each cell is difficult to determine and does not fully reflect functional capacity, we must rely on proxy measures such as mtDNA-CN.

Replication of mtDNA is regulated by the D-loop, (Nicholls and Minczuk, 2014) a non-coding region within mtDNA, and occurs independently from nuclear DNA replication, although many factors associated with mtDNA replication are encoded by nuclear DNA (Bogenhagen and Clayton, 1977). In contrast to the nuclear genome, which contains only two copies per cell, the mitochondrial genome is present in multiple copies per cell (from 100 to 10,000), depending on cell type (Wai et al., 2010). High ATP requirement cells such as heart and skeletal muscle cells have as high as ~7,000 mtDNA copies per cell, while low energy requirement cells such as spleen and liver cells have as low as ~100 copies per cell (Chabi et al., 2003; Kelly et al., 2012; Miller et al., 2003).

mtDNA-CN as a biomarker of mitochondrial function

Levels of mtDNA-CN are directly correlated with energy reserves, oxidative stress, and changes in mitochondrial membrane potential (Guha and Avadhani, 2013). The relationship of mtDNA-CN to mitochondrial function was evidenced in knockdown cell models of reduced mtDNA-CN, which resulted in reduced expression of vital complex proteins, altered cellular morphology, and lower respiratory enzyme activity (Jeng et al., 2008). In these models, mitochondrial function was rescued after restoring mtDNA-CN to wild-type levels. mtDNA-CN levels are dependent on the maintenance of mitochondrial genome stability via proper mitochondrial translation, as evidenced by reduction of mtDNA-CN due to dysfunction in the synthesis of mitochondrial ribosomal protein components (Zhang and Singh, 2014). Further, although mitochondrial fission is coordinated with DNA replication, it does not seem to influence mtDNA-CN levels (Ishihara et al., 2009). In contrast, fusion is critical for mtDNA-CN maintenance (Chen et al., 2010; Y. Chen et al., 2011; Silva Ramos et al., 2019). Taken together, these observations highlight that reduced mtDNA-CN serves as a useful biomarker of mitochondrial dysfunction that can be measured in clinical and population studies.

Measurement of mtDNA-CN

mtDNA-CN levels can be readily measured in extracted DNA from peripheral blood or other tissues, although in practice most measurements are performed in peripheral blood due to ease of accessibility. Measurements of mtDNA-CN in blood have been shown to be relevant to a variety of chronic diseases (Ashar et al., 2017, 2015; Chen et al., 2014; Tin et al., 2016; Yiyi Zhang et al., 2017). However, when available, mtDNA-CN should be measured directly in the tissue of interest.

Until recently, quantitative PCR (qPCR) has been widely considered the gold standard for measuring mtDNA-CN. The standard protocol consists of a multiplexed Taqman assay in which the number of copies of a mitochondrial gene is compared to the number of copies of a nuclear gene to achieve a relative measure of mtDNA-CN (Ashar et al., 2015). Numerous target genes have been used successfully including, but not limited to, MT-ND1, MT-ND4, MT-CYB, and MT-TL1 from the mitochondrial genome, and B2M, RPLPO, ACTB, and RPPH1 for the nuclear genome (Ashar et al., 2015; Bai and Wong, 2005; Dimmock et al., 2010; Gu et al., 2013; Jędrak et al., 2017; Kamfar et al., 2016; Knez et al., 2016). Recently, digital PCR (dPCR) has emerged as an additional method for calculating mtDNA-CN particularly due to its ability to quantify absolute copy number by determining the ratio of the number of positive mitochondrial probe copies to the number of nuclear probe copies (Li et al., 2018; Memon et al., 2017; O’Hara et al., 2019; Ye et al., 2017). mtDNA-CN levels can also be measured from pre-existing genotyping microarray, whole exome, and whole genome sequence data (Table 1) (Ding et al., 2015; MitoPipeline, 2018; Qian et al., 2017).

Table 1.

mtDNA-CN estimation methods

Platform	qPCR	digitalPCR	Microarray	Sequencing
mtDNA-CN Measurement	Ct mitochondrial probe - Ct nuclear probe	Ratio of number of positive mitochondrial probe copies/uL to the number of nuclear probe copies/uL	Median mitochondrial probe intensity adjusted for PCs generated from nuclear probes	(Mitochondrial DNA coverage divided by autosomal coverage)* 2
Absolute measure of mtDNA-CN	No	Yes	No	No
Number of mtDNA loci interrogated	1–2	1–2	~25–150	Whole Mitochondrial Genome
Cost	$	$$	$$$$	$$$$$
Detect mtDNA deletions	Yes, major deletions with additional probe	Yes, major deletions with additional probe	No	Yes
Detect Single Base-Pair Heteroplasmy	No	No	No	Yes
Haplogroup determination	No	No	No	Yes

PCs: Principal Components; Ct: Cycle threshold; qPCR: Quantitative PCR; mtDNA: Mitochondrial DNA; mtDNA-CN: Mitochondrial DNA Copy Number.

Further, the DNA extraction method affects the accuracy of mtDNA-CN measurements. Specifically, organic solvent extraction (phenol-chloroform-isoamyl alcohol) provides a more consistent and accurate mtDNA-CN estimation compared to silica-based methods, (Guo et al., 2009) likely because column kit parameters are typically optimized for the isolation of DNA fragments ≥ 50 Kb leading to inadequate isolation of the smaller mtDNA genome (Nacheva et al., 2017). One solution to reduce biases introduced by DNA extraction may be to use direct measurements that avoid traditional DNA extraction, such as measuring mtDNA-CN from direct cell lysis, an approach that outperforms both organic solvent extraction and a column-based method (Longchamps et al., 2020). This method involves pelleting cells and removing the supernatant followed by agitation and heating in solution to disrupt the pellet and subsequent centrifugation to pellet out any insoluble inhibitors (Longchamps et al., 2020). Further, direct cell lysis provides a more cost-effective approach and is faster than traditional DNA extraction, although its use limits the utility of the sample for downstream applications.

While qPCR, dPCR, genotyping microarray, and sequence data reliably quantify mtDNA-CN, there are two major challenges to the widespread use of these measurements in clinical practice. First, other than dPCR, these methods do not produce absolute measures of mtDNA-CN, but relative measures with respect to the number of nuclear DNA copies in the sample. Second, these measurements are affected by batch effects which must be adjusted for prior to data interpretation. Because mtDNA-CN values are often relative values and there is between-batch variability, mtDNA-CN values cannot be directly compared across studies or when assays are performed over time in the same individuals. In addition, since the number of mitochondria varies by cell type, mtDNA-CN levels will depend on the cell composition of the sample. For peripheral blood, mtDNA-CN are positively correlated with platelet count and inversely correlated with white blood cell (WBC) count (Ashar et al., 2015; Hurtado-Roca et al., 2016; Knez et al., 2016; Tin et al., 2016). As a consequence, mtDNA-CN measurements need to be standardized by sex, age, platelet count and WBC count.

In addition to tissue-specificity, average mtDNA-CN levels in blood decrease with age after approximately age 50 and are on average higher in women relative to men (Ashar et al., 2015; Mengel-From et al., 2014). Although little is known about the precise mechanisms that lead to variation in mtDNA-CN, genetic and environmental contributions are hypothesized to interact to determine the number of mitochondria in cells. In a genome-wide association study (GWAS) of 10,442 Chinese women, two loci were associated with mtDNA-CN levels, one within the 3’ UTR of mitochondrial transcription factor A (TFAM) and the other within intron 1 of cyclin dependent kinase 6 (CDK6) (Cai et al., 2015). In addition to genetic factors, various chemicals and drugs are known to affect mtDNA-CN levels (Young, 2017).

Detecting large heteroplasmic deletions and single base-pair heteroplasmies

Mitochondrial heteroplasmy reflects the coexistence of multiple mitochondrial haplotypes in a single organism. The technical ease of multiplexing in qPCR and digital PCR has led to the development of several techniques which enable the simultaneous detection of large heteroplasmic deletions and mtDNA-CN quantification in a single assay (Bai and Wong, 2005; Phillips et al., 2014; Rygiel et al., 2015). Similarly, mtDNA deletion detection methods for next generation sequence data have been packaged into several useful bioinformatics pipelines (Bosworth et al., 2017; Goudenège et al., 2019). Low frequency single base-pair heteroplasmies have traditionally been difficult to detect due to the coverage necessary to eliminate the contribution of sequencing error. However, with sufficient coverage several pipelines such as mitoCaller and mtDNA-Server exist which enable users to detect heteroplasmies contributing to as little as 1% of mtDNA (Ding et al., 2015; Weissensteiner et al., 2016).

Molecular consequences of variation in mtDNA-CN

In vitro reductions in mtDNA-CN are associated with downregulation of mitochondrial transcription and decreased levels of proteins involved in oxidative phosphorylation, specifically ND1, CYTB and COX-I (Jeng et al., 2008). Furthermore, natural variation in mtDNA-CN in cancer tissues is highly correlated with expression levels of genes involved in cellular respiration and mitochondrial function (Reznik et al., 2016). The reduction in respiratory capacity induced by reduced mtDNA-CN levels may have important pathogenic consequences.

Several theories involving immune dysfunction, energy demands, inflammation and altered cell signaling may explain the mechanisms linking mitochondrial dysfunction in whole blood to chronic disease (Figure 1). Mitochondrial-mediated changes in oxidative capacity may lead to chronic inflammation via macrophage polarization. Macrophages can change sub-types based on the surrounding microenvironment, with TNF-α and IFN-γ signaling pro-inflammatory M1 macrophages and TGF-β and IL-10 signaling anti-inflammatory M2 macrophages (Martinez et al., 2008). Interestingly, M1 macrophages generate ATP through glycolysis while M2 macrophages use oxidative phosphorylation (Ravi et al., 2014; Rodríguez-Prados et al., 2010; Vats et al., 2006). In the presence of reduced mtDNA-CN, mitochondrial dysfunction may be the consequence of an insufficient number of mtDNA molecules to encode for proteins of the cellular respiration pathway. As a result, macrophages may no longer be able to switch to their anti-inflammatory M2 subtype, leading to a reduction in anti-inflammatory mediators that would assist in tissue repair and the resolution of inflammation.

Figure 1.

Summary of proposed relationship between mtDNA-CN and disease.

Lending further support to the macrophage hypothesis, many of the diseases associated with mtDNA-CN involve immune dysfunction. Cardiovascular disease (CVD), and more specifically the atherosclerotic processes which leads to CVD, involves a chronic inflammatory processes from initiation through progression and eventual thrombotic events (Libby et al., 2002). Similar to CVD, a chronic pro-inflammatory state directly contributes to morbidity and mortality in chronic kidney disease and liver disease (Czaja, 2014; Silverstein, 2009). Furthermore, immune dysfunction and microglial activation have been shown to be signature landmarks of neurodegenerative disorders such as Alzheimer’s disease and Parkinson’s disease (Amor et al., 2010; Morgan et al., 2012; Rodriguez and Kern, 2011).

Mitochondrial dysfunction may also affect nuclear gene expression and methylation patterns (Friis et al., 2014; Picard et al., 2014; Srinivasan et al., 2016). Mitochondrial retrograde signaling (signaling from the mitochondrion to the nucleus) affects nuclear gene expression through disruption of the mitochondrial membrane potential which in turn leads to global changes in nuclear gene expression (Guantes et al., 2015; Guha et al., 2016; Muir et al., 2016) via chromatin activation, regulation of transcriptional activity, alternative splicing and regulation of protein synthesis (Guantes et al., 2015). The impact of mtDNA-CN levels seem to be particularly strong for alternative splicing, modulating both the abundance and type of mRNAs through direct impact on RNA Pol II kinetics (Guantes et al., 2015). In the presence of a reduced number of mitochondria, the cell enters a state of energy crisis and experiences changes in nuclear gene expression that may increase the risk of chronic disease, cancer and/or premature aging (Guantes et al., 2015).

Finally, reduced mtDNA-CN levels has been linked to increased oxidative stress via increased production of ROS. In CVD, mtDNA-CN was associated with the activation of proatherogenic genes, atherogenesis, plaque instability, thrombosis via LDL oxidation, and altered electrophysiology (Aggarwal and Makielski, 2013; Berliner and Heinecke, 1996; Erusalimsky, 2009; Jain et al., 2013; Liu et al., 2010; Lo et al., 2005; Sánchez-Santos et al., 2018; Yu and Bennett, 2014). Mitochondrial-mediated oxidative stress and resulting hepatocyte apoptosis along with impaired fatty acid oxidation have long been implicated in liver disease (Cichoż-Lach and Michalak, 2014; Sookoian et al., 2010). Furthermore, mitochondrial-mediated oxidative stress is also involved in aging-related diseases and neurodegeneration, such as promoting the production of amyloid-β peptide (Aβ) in Alzheimer’s disease (AD) (Gandhi and Abramov, 2012; Kim et al., 2015). In fact, mitochondrial dysfunction has been established as a hallmark of the aging process (Tidwell et al., 2017) and lifespan studies in C.elegans suggest that mitochondrial-nuclear crosstalk is a major determinant of lifespan (Cristina et al., 2009).

mtDNA Copy Number and the Risk of Chronic Disease

Changes in mtDNA-CN precede disease onset, implicating mtDNA regulation in the development of a variety of chronic diseases (Table 2).

Table 2.

Studies reporting association with mtDNA-CN and risk of chronic disease

Study	Setting	Study Design	Sample Size	Tissue	Method for mtdNA estimation	Endpoint	Results
Cardiovascular
Ashar et al (2017)(Asharet al., 2017)	Prospective	Population-based	21,870(11,153 from ARIC, 4,830 from CHS, and 5,887 from MESA)	Buffy coat, peripheral leukocytes	ARIC/MESA: Array-based; CHS: qPCR	Coronary heart disease (CHD)	↓ Decreased mtDNA-CN is associated with CHD(OR=1.29, 95% Cl, 1.24–1.33)
Chen et al (2014)(Chen et al., 2014)	Retrospective	Case-control	378 CHD cases, 378 controls	Peripheral leukocytes	qPCR		↓ Decreased mtDNA-CN is associated with CHD (OR=2.38, 95% Cl, 1.33–4.69)
Ashar et al (2017)(Asharet al., 2017)	Prospective	Population-based	21,870(11,153 from ARIC, 4,830 from CHS, and 5,887 from MESA)	Buffy coat, peripheral leukocytes	ARIC/MESA: Array-based; CHS: qPCR	Stroke	↓ Decreased mtDNA-CN is associated with Stroke (OR= 1.11, 95% Cl, 1.06–1.16)
Zhang et al (2017)(Yiyi Zhang et al., 2017)	Prospective	Population-based	11,093 participants	Buffy coat	Array-based	Sudden cardiac death (SCD)	↓ Decreased mtDNA-CN is associated with SCD (HR: 2.24, 95% Cl, 1.58–3.19; P-trend <0.001)
Kidney
Tin et al (2016)(Tinetal., 2016)	Prospective	Population-based	9,058 participants	Buffy coat	Array-based	Chronic kidney disease (CKD)	↓ Decreased mtDNA-CN is associated with CKD (HR=0.65, 95% Cl, 0.56 to 0.75; PO.001)
Zhang et al (2017)(Yuheng Zhang et al., 2017)	Retrospective	Case-control	109ESRD cases, 112 controls	Peripheral leukocytes	qPCR	End stage renal disease (ESRD)	1 Decreased mtDNA-CN is associated with ESRD (2.85 ±3.01 versus 5.94 ± 5.63; P<0.001)
Zhang etal (2017)(Yuheng Zhang et al., 2017)	Retrospective	Case-control	56 ESRD cases, 58 controls	Plasma	qPCR		↑ Increased mtDNA-CN is associated with ESRD (6.97 ±1.16 versus 5.50 ± 0.93, P<0.001)
Liver Disease
Sookoian et al (2010)(Sookoian etal., 2010)	Retrospective	Case-control	63 NAFLD cases, 11 controls	Liver biopsy	qPCR	Non-alcoholic fatty liver disease (NAFLD)	↓ Decreased mtDNA-CN is associated with NAFLD (PO.01)
Pirola et al (2015)(Pirola et al., 2015)	Retrospective	Case-control	67 NAFLD cases, 23 controls	Liver biopsy	qPCR		↓ Patients with NAFLD have a significantly lower liver mtDNA copy number (62.5 ±41.0), in comparison with controls (103.5 ±88.0)
Neurodegenerative Disorders
Pyle et al (2016)(Pyle et al., 2016)	Retrospective	Case-control	363 peripheral blood patient samples (262 controls), 151 substantia nigra pars compacta (SNpc) patient samples (33 controls)	Peripheral leukocytes, substantia nigra pars compacta	qPCR	Parkinson’s Disease (PD)	↓ Decreased mtDNA-CN is associated with PD (Blood: P=2.86×10–4; Substantia nigra: P=4.Oxio-3)
Wei et al (2017)(W. Wei et al., 2017)	Retrospective	Case-control	282 cases, 351 controls	Cerebellum, cerebral cortex	Exome sequencing	Alzheimer’s Disease (AD)	↓ Decreased mtDNA-CN is associated with AD(P=2.85×10–7)
Wei et al (2017)(W. Wei et al.,2017)	Retrospective	Case-control	181 cases, 351 controls	Cerebellum, cerebral cortex	Exome sequencing	Creutzfeldt-Jakob Disease (CJD)	↓ Decreased mtDNA-CN is associated with CJD(P=3.34><10–7)
Cancer Risk
Xing et al (2008)(Xing et al., 2008)	Retrospective	Case-control	260 cases, 281 controls	Peripheral leukocytes	qPCR	Kidney Cancer	↓ Decreased mtDNA-CN is associated with Kidney Cancer (OR =
							1.53, 95% Cl, 1.07 to 2.19)
Thyagarajan et al (2012)(Thyagaraj anetal., 2012)	Retrospective	Case-control	422 cases, 874 controls	Peripheral blood	qPCR	Colorectal Cancer	↓⁄↑ Compared to 2nd quartile, OR (95% Cl) for subjects in the lowest and highest quartiles of relative mtDNA copy numbers were 1.81 (1.13–2.89) and 3.40 (2.15–5.36), respectively (Pcurvilinearity<0.0001)
Lynch etal (2011)(Lynchet al., 2011)	Retrospective	Case-control	203 cases, 656 controls	Whole blood	qPCR	Pancreatic Cancer	↑ Increased mtDNA-CN is associated with pancreatic cancer (OR=1.14, 95% Cl, 1.06–1.23)

OR: Odds Ratio; CI: Confidence Interval; HR: Hazard Ratio; qPCR: Quantitative PCR; mtDNA: Mitochondrial DNA; mtDNA-CN: Mitochondrial DNA Copy Number; ARIC: Atherosclerosis Risk in Communities Study; MESA: Multi-Ethnic Study of Atherosclerosis; CHS: Cardiovascular Health Study; CHD: Coronary heart disease; SCD: Sudden Cardiac Death; CKD: Chronic Kidney Disease; ESRD: End stage renal disease; NAFLD: Non-alcoholic fatty liver disease; PD: Parkinson’s Disease; AD: Alzheimer’s Disease; CJD: Creutzfeldt-Jakob Disease.

Cardiovascular disease

In a large pooled analysis of three major cohorts, mtDNA-CN measured from buffy coat was inversely associated with prevalent and incident CVD, coronary heart disease (CHD), stroke, and sudden cardiac death (SCD), independent of traditional risk factors (3665, 2460, 1583, and 361 incident events for CVD, CHD, stroke, and SCD, respectively) (Ashar et al., 2017; Yiyi Zhang et al., 2017). In these cohorts, adding mtDNA-CN to the ACC/AHA risk evaluation criteria for primary prevention significantly improved risk reclassification for CVD (Ashar et al., 2017), and it improved the sensitivity and specificity for the ACC/AHA recommendations on initiating statin therapy for primary prevention of atherosclerotic cardiovascular disease (continuous net reclassification index, 0.194; 95% CI, 0.130–0.258; P< 0.001) (Ashar et al., 2017). In addition, mtDNA-CN measured from leukocytes was associated with the presence of coronary atherosclerotic plaque in a small case-control study (378 CHD cases, 378 controls) (Chen et al., 2014) and mtDNA-CN from peripheral blood was inversely associated with an angiographic severity score in coronary atherosclerosis in cross-sectional analysis (Liu et al., 2017; Neeland et al., 2012).

Kidney Disease

In the Atherosclerosis Risk in Communities (ARIC) cohort, higher mtDNA-CN measured from buffy coat was associated with a reduced risk of incident chronic kidney disease (CKD) after adjusting for traditional CKD risk factors, such as prevalent diabetes, hypertension, and C-reactive protein levels (Tin et al., 2016). Others have found similar associations with kidney disease, including a recent study which identified decreased copy number measured from peripheral blood in patients with end stage renal disease (ESRD) (Yuheng Zhang et al., 2017). Interestingly, mtDNA-CN measured from serum was increased in patients with ESRD and in the presence of markers of renal injury (Eirin et al., 2016; Yuheng Zhang et al., 2017). Further, decreased mtDNA is observed in urinary supernatant compared to intra-renal samples which suggests different mechanisms are at play in these tissues (P. Z. Wei et al., 2017). One possible explanation for this observation may be that cell-free mtDNA measured from serum or urine could reflect higher rates of apoptosis rather than mitochondrial dysfunction, thus leading to elevated mtDNA-CN associations with disease. Further, in a large independent CKD cohort, decreased mtDNA-CN was significantly associated with infection-related deaths (Fazzini et al., 2019) which lends additional support to recent findings linking HIV disease severity with decreased mtDNA-CN (Sun et al., 2019).

Liver Disease

mtDNA-CN has been shown to be associated with several classes of liver disease, including non-alcoholic fatty liver disease (NAFLD). In NALFD, mtDNA-CN is reduced in liver tissue of patients as compared to controls (Pirola et al., 2015; Sookoian et al., 2010). Additionally, the mtDNA/nDNA ratio has been found to be inversely correlated with insulin resistance, serum fasting glucose and plasma fasting insulin (Sookoian et al., 2010).

Neurodegenerative Disorders

Cerebellar neuron dysfunction has been associated with impaired mitochondrial structure and bioenergetic function (Bartesaghi et al., 2010). Patients with polymerase gamma encephalopathy display dopaminergic substanta nigral neurons with significantly lower mtDNA-CN and an overall loss of dopaminergic neurons in the pars compacta region (Tzoulis et al., 2013). The link to neurodegeneration appears independent of rare mitochondrial DNA depletion syndromes as independent analysis found significantly lower mtDNA-CN in substantia nigra pars compacta tissue and peripheral blood samples in patients with Parkinson’s disease (PD) (Pyle et al., 2016). Interestingly, the reduction of copy number appears to be brain region specific as no reduction in mtDNA-CN was observed in the frontal cortex (Pyle et al., 2016). Although recent work did not replicate the previous findings for PD, reduced mtDNA-CN was observed in AD and Creutzfeldt-Jakob Disease (W. Wei et al., 2017).

Cancer

A number of case-control studies which measured mtDNA-CN from blood between cancer patients and normal healthy controls have identified associations between mtDNA-CN levels and increased risk of cancer. Specifically, one such study found an association between decreased mtDNA-CN in peripheral leukocytes and renal cell carcinoma (Xing et al., 2008). Further, pancreatic cancer risk correlates with increased mtDNA-CN levels in whole blood (Lynch et al., 2011). Additionally, colorectal cancer risk has been shown to be associated with both high levels and low levels of mtDNA-CN in peripheral blood (Thyagarajan et al., 2012). Interestingly, increased mtDNA-CN in colorectal cancer correlates with the presence of the common 4977 mitochondrial deletion which spans five tRNA genes and seven protein-coding genes (T. Chen et al., 2011).

Differences in mtDNA Copy Number between Cancer and Normal Tissue

A hallmark characteristic of cancer is the presence of defective mitochondria in cancer cells (Reznik et al., 2016), including mutations, deletions and copy number variation of mtDNA. In addition to studies which have assessed the contribution of mtDNA-CN to the risk of cancer, mtDNA-CN differences have also been found between cancer and normal tissue for a variety of cancer types (Table 3). Specifically, depletion of mtDNA relative to matched normal tissue is particularly common in bladder, breast, and kidney cancers (Reznik et al., 2016). In addition, there is an association between the incidence of several somatic alterations, such as IDH1 mutations in gliomas, and mtDNA content (Reznik et al., 2016). Further, in some cancer types, mtDNA content is correlated with the expression of respiratory genes and negatively correlated with the expression of immune response and cell cycle genes (Reznik et al., 2016). These observations suggest that mtDNA-CN measurement may be useful as a clinical marker of tumor identification, progression, and treatment.

Table 3.

Studies reporting differences in mtDNA-CN between cancer and normal tissue

Study	Setting	Study Design	Sample Size	Tissue	Method for mtdNA estimation	Endpoint	Results
Reznik et al (2016)(Rezni ketal., 2016)	Retrospective	Tumor-normal	45 matched samples	Bladder tissue	Whole exome/ whole genome	Bladder	↓ Decreased mtDNA-CN in bladder tumor tissue (BH-corrected Mann-Whitney p-value <0.05)
Reznik et al (2016)(Rezni ketal., 2016)	Retrospective	Tumor-normal	81 matched samples	Breast tissue	Whole exome/ whole genome	Breast	↓ Decreased mtDNA-CN in breast tumor tissue (BH-corrected Mann-Whitney p-value <0.05)
Yu et al (2007)(Yu et al., 2007)	Retrospective	Tumor-normal	59 matched samples		qPCR		↓ Decreased mtDNA-CN in breast tumor tissue (P=0.001)
Reznik et al (2016)(Rezni ketal., 2016)	Retrospective	Tumor-normal	Clear-cell: 249 matched samples; Papillary: 55 matched samples	Kidney tissue	Whole exome/ whole genome	Kidney	↓ Decreased mtDNA-CN in kidney tumor tissue (BH-corrected Mann-Whitney p-value <0.05)
Yuan et al (2020)(Yuan etal.,2020)	Retrospective	Tumor-normal	27 matched samples		Whole genome		↓ Decreased mtDNA-CN in renal clear-cell carcinoma tissue (two-sided Wilcoxon signed-rank p-value <0.001)
Reznik et al (2016)(Rezni ketal., 2016)	Retrospective	Tumor-normal	25 matched samples	Esophegeal tissue	Whole exome/ whole genome	Esophageal	↓ Decreased mtDNA-CN in esophageal tumor tissue (BH-corrected Mann-Whitney p-value <0.05)
Reznik et al (2016)(Rezni ketal., 2016)	Retrospective	Tumor-normal	92 matched samples	Squamous cell tissue	Whole exome/ whole genome	Head/neck squamous cell	↓ Decreased mtDNA-CN in squamous cell tumor tissue (BH-corrected Mann-Whitney p-value <0.05)
Reznik et al (2016)(Rezni ketal.,2016)	Retrospective	Tumor-normal	83 matched samples	Liver tissue	Whole exome/ whole genome	Liver	↓ Decreased mtDNA-CN in liver tumor tissue (BH-corrected Mann-Whitney p-value <0.05)
Yuan et al (2020)(Yuan etal.,2020)	Retrospective	Tumor-normal	29 matched samples		Whole genome		↓ Decreased mtDNA-CN in liver tumor tissue (two-sided Wilcoxon signed-rank p-value <0.01)
Yuan et al (2020)(Yuan etal.,2020)	Retrospective	Tumor-normal	25 matched samples	Myeloid tissue	Whole genome	Myeloproliferative neoplasm	↓ Decreased mtDNA-CN in tumor tissue (two-sided Wilcoxon signed-rank p-value O.001)
Reznik et al (2016)(Rezni ketal.,2016)	Retrospective	Tumor-normal	137 matched samples	Lung tissue	Whole exome/ whole genome	Lung	↓ Decreased mtDNA-CN in lung adenocarcinoma tumor tissue (BH-corrected Mann-Whitney p-value <0.05)
Yuan et al (2020)(Yuan etal.,2020)	Retrospective	Tumor-normal	25 matched samples		Whole genome		↓ Incresed mtDNA-CN in lung squamous cell carcinoma tissue (two-sided Wilcoxon signed-rank p-value <0.01)
Savagner et al (2001)(Savag ner et al., 2001)	Retrospective	Tumor-normal	22 matched samples	Thyroid tissue	Southern blot analysis of rRNA	Thyroid	↓ Increased mtDNA-CN in thyroid tumor tissue (PO.001)
Mizumachi et al (2008)(Mizu machi et al., 2008)	Retrospective	Tumor-normal	9 cell lines	Prostate cancer cell line (LNCaP)	qPCR	Prostate	↓ mtDNA was increased in 7 of 9 prostate cancer lines compared to adjacent cells
Yuan et al (2020)(Yuan etal.,2020)	Retrospective	Tumor-normal	81 matched samples	Lymph tissue	Whole genome	Chronic Lymphocytic Leukemia	↓ Increased mtDNA-CN in turmor tissue (two-sided Wilcoxon signed-rank p-value <0.001)
Yuan et al (2020)(Yuan etal.,2020)	Retrospective	Tumor-normal	111 matched samples	Pancreatic tissue	Whole genome	Pancreatic adenocarcinoma	↓ Increased mtDNA-CN in pancreatic tissue (two-sided Wilcoxon signed-rank p-value <0.001)

qPCR: Quantitative PCR; mtDNA: Mitochondrial DNA; mtDNA-CN: Mitochondrial DNA Copy Number; BH: Benjamini-Hochberg; rRNA: ribosomal RNA.

Mutations of the mitochondrial genome have been associated with the formation, growth and metastasis of tumor cells in multiple types of cancer (Ju et al., 2014) (Table 3). In the most comprehensive comparison of mtDNA content across 15 cancer types with normal adjacent tissue (Reznik et al., 2016), six cancer types displayed decreased mtDNA-CN (bladder, breast, esophageal, head/neck squamous cell, kidney, and liver), one cancer type showed increased mtDNA-CN (lung adenocarcinoma), and six cancer types displayed no difference (colorectal, pancreatic, prostate, gastric, thyroid, and uterine) (Reznik et al., 2016). Other studies have confirmed decreased mtDNA-CN in breast cancer (Yu et al., 2007). Further, recent molecular characterization of mitochondrial genomes in human cancers showed increased mtDNA-CN in chronic lymphocytic leukemia, lung squamous cell carcinoma and pancreatic adenocarcinoma, and decreased mtDNA-CN in kidney clear cell carcinoma, liver cancer and myeloproliferative neoplasm (Yuan et al., 2020). Thyroid and prostate cancer cells have been found by other groups to be enriched for mitochondrial content, with prostate cancer showing an increase of up to 78% (Mizumachi et al., 2008; Savagner et al., 2001). This increase has been suggested to be the consequence of cell compensation for defective oxidative phosphorylation, leading to lower ATP production and increased risk for cancer (Hertweck and Dasgupta, 2017; Mizumachi et al., 2008).

Across five cancer types, statistically significant associations have been identified between mtDNA-CN and patient survival and across three cancer types higher mtDNA content has been associated with increased survival (Reznik et al., 2016). Interestingly, it has been suggested that tumor cells may increase mtDNA-CN as a self-protective mechanism to prevent apoptosis, given that reduced mtDNA-CN increases ROS levels in tumor cells, tumor cell sensitivity to chemotherapeutic drugs, and the rate of apoptosis. Thus, mtDNA-CN variation may be a possible therapeutic target for treatment of tumors in a clinical setting (Mei et al., 2015). However, it is plausible that mitochondrial biogenesis is upregulated in cancers as a compensatory mechanism only (Gasparre et al., 2011) and therefore further research is necessary to elucidate the precise mechanism underlying the association of mtDNA-CN and cancer etiology/progression.

Conclusion

In understanding the etiology of complex diseases, it has become clear that the complete knowledge of nuclear DNA sequence is not enough to fully predict complex disease risk. As discussed here, an emerging body of evidence supports roles for mtDNA in the complex underpinnings of a variety of diseases, including a number of cancers and aging-related disorders. A common link in these studies include anti-inflammatory pathways. These mechanisms will be further elucidated as our ability to measure mtDNA-CN from sequencing and microarray technologies expands. As studies increase in power and functional assessment of mechanisms underlying the effect of mtDNA on mitochondrial function and gene expression improve, our understanding of variation in mtDNA-CN as cause or consequence of disease development will rapidly improve.

mtDNA-CN is an especially attractive biomarker because its measurement in blood is both non-invasive and relatively cost-friendly to obtain. The proposed utility of mtDNA-CN as a biomarker for disease has been suggested by the observation that mtDNA content can differentiate healthy controls from patients with cancer and other diseases (Memon et al., 2017). In addition, mtDNA-CN has been shown to be relevant for risk reclassification for cardiovascular disease (Ashar et al., 2017). Currently, these applications are limited by several analytical factors affecting the accurate and reproducible quantification of mtDNA-CN, as discussed in this review. The recent confirmation that human mtDNA is methylated adds yet another level of complexity to the crosstalk between the nucleus and mitochondrion and its control (Ghosh et al., 2014). We close by suggesting that improved detection techniques for mtDNA-CN as well as greater understanding of the mechanisms underlying individual, cell-type, and tissue-specific variation in mtDNA-CN are essential to determining the direct pathological, therapeutic and/or clinical relevance of this relatively cost-effective and easily measured biomarker.

Highlights.

• mtDNA-CN, a proxy for mitochondrial function, is associated with a number of age-related diseases and cancers

• mtDNA-CN can be measured using a variety of technologies, including pre-existing microarray and sequencing data

• Mitochondrial dysfunction may lead to alterations to gene expression and subsequent disease outcomes via a number of relevant pathways

• mtDNA-CN shows promise as a potential biomarker for chronic disease