Maternally inherited diabetes and deafness (MIDD) is a monogenic mitochondrial disorder caused by a pathogenic variant in the MT-TL1 gene encoding a leucine transfer RNA. We propose a new hypothesis that explains how the MT-TL1 variant causes impaired glucose tolerance and diabetes in MIDD. We suggest that diabetes in MIDD primarily depends on a variable combination of insulin resistance and impaired β-cell function that seems more likely to occur in the presence of high skeletal muscle heteroplasmy and moderate β-cell heteroplasmy for m.3243A>G. The underlying genetic defect generates oxidative stress and disrupts the tricarboxylic acid cycle, leading to mTORC1 hyperactivity and modifying mitochondrial retrograde signaling. mTORC1 hyperactivity contributes to insulin resistance and β-cell dysfunction and to an increased load of the m.3243A>G phenotypic variant. Abnormal mitochondrial signaling affects the nuclear epigenome and influences MIDD phenotype. We highlight evidence that, despite being an apparent pathogenic factor, heteroplasmy in the blood and in tissues does not fully explain the phenotypic variability of this condition and that other factors, including mtDNA copy number, additional nuclear or mitochondrial variants, environmental factors, and metabolic characteristics of the patient, may contribute. A better understanding of the mechanisms leading to MIDD will help inform novel management strategies for this form of diabetes.
Maternally inherited diabetes and deafness (MIDD) is a mitochondrial disorder characterized primarily by hearing impairment and diabetes.
m.3243A>G, the most common phenotypic variant, causes a complex rewiring of the cell with discontinuous remodeling of both mitochondrial and nuclear genome expressions.
We propose that MIDD depends on a combination of insulin resistance and impaired β-cell function that occurs in the presence of high skeletal muscle heteroplasmy (approximately ≥60%) and more moderate cell heteroplasmy (∼25%–72%) for m.3243A>G.
Understanding the complex mechanisms of MIDD is necessary to develop disease-specific management guidelines that are presently lacking.
Introduction
Maternally inherited diabetes and deafness (MIDD) is a monogenic mitochondrial disorder caused by a pathogenic variant in the MT-TL1 gene encoding a leucine tRNA [tRNALeu(UUR)] (MIM 520000). Of the 20 causative variants of MIDD, m.3243A>G is the most prevalent, accounting for >80% of MIDD cases (1). Almost all published articles focus on the m.3243A>G variant.
The m.3243A>G variant is associated with several overlapping phenotypes that are described as distinctive syndromes, even though individuals can present with one syndrome and progress into another. The phenotype of MIDD is slowly progressive and heterogenous (2). The most severe phenotype, which likely reflects, at least partly, a higher level of heteroplasmy (3), comprises mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes (MELAS). It is found in ∼10% of clinically ascertained m.3243A>G affected carriers. Milder phenotypes include MIDD and chronic progressive external ophthalmoplegia.
The clinical characteristics of MIDD have previously been extensively described (2,4,5). The phenotype of MIDD is characterized primarily by progressive hearing impairment and diabetes inherited in a matrilinear manner, often over several generations, with a marked variability in the presentation. Additional symptoms may include myopathy, central neurological and psychiatric features, macular retinal dystrophy, gastrointestinal dysmotility (constipation, diarrhea, pseudo-obstruction), and, of importance for people with diabetes, congestive heart failure and focal segmental glomerulosclerosis (4,6). The clinical presentation of diabetes in MIDD is easily confused with that of type 1 or type 2 diabetes, contributing to the delayed diagnosis of mitochondrial disease as an underlying cause of diabetes. Therefore, MIDD in patients is mostly managed according to type 1 or type 2 diabetes guidelines.
The objective of this article is to review the underlying mechanisms, role of heteroplasmy, and presentation of diabetes as well as the characteristics of cardiovascular and renal diseases that are associated with MIDD. We then propose a model that explains how this alteration of MT-TL1 causes diabetes.
Epidemiology
The association between a mutation of the MT-TL1 gene and MIDD was first reported more than 40 years ago (7). Since then, investigators of population studies have found that ∼1 in 700 newborns had a m.3243A>G variant in umbilical cord blood with the heteroplasmic detection threshold set at 0.5% (8). In the UK Biobank study, whole-exome sequencing was performed in 179,862 clinically unselected participants. Diabetes was present in 16,386 participants and was caused by m.3243A>G in 0.14% (9). In contrast, a clinical diagnosis of m.3243A>G-related disease is made in only ∼1 in 28,000 individuals (10). This discrepancy suggests not only that many m.3243A>G carriers may be a- or paucisymptomatic (2) but also that the ubiquitousness of the clinical signs makes the recognition of the underlying diagnosis difficult for clinicians. Indeed, two reference centers independently found that m.3243A>G was the most common cause of syndromic diabetes (8%) among individuals with a confirmed molecular diagnosis of monogenic diabetes screened solely on the basis of diabetes (11,12). Finally, many case reports and case series have included Asian individuals (1) and a higher prevalence of the m.3243A>G variant has been reported among those with diabetes and end-stage renal disease of Japanese compared with European ancestry (4).
Clinical Presentation of MIDD
Due in part to a general lack of awareness of mitochondrial diseases and the marked variability of the phenotype, MIDD is commonly underdiagnosed in people with diabetes. It is hoped that the organization of an annual World Mitochondrial Disease Week (mitochondrialdiseaseweek.org/) will change this situation.
Best practice guidelines for genetic testing for mitochondrial disease have been published (13), but specific guidelines for the diagnosis of MIDD have not been proposed. MIDD usually presents progressively, although inaugural ketoacidosis has been reported (14). The Mitochondrial Medicine Society recommends measuring A1C at diagnosis of mitochondrial disease and thereafter every 1–2 years or as needed (15). The 2024 Standards of Care in Diabetes guidelines define prediabetes according to A1C 5.7%–6.4% (39–47 mmol/mol) and diabetes A1C >6.4% (48 mmol/mol) but do not mention mitochondrial diabetes (16).
Nevertheless, there are pronounced epidemiological and clinical differences among type 1 diabetes, type 2 diabetes, and MIDD (Table 1). The association of diabetes with hearing impairment, and a matrilinear inheritance (male and female siblings born to an affected mother carry the mutation), should alert the clinician of the possibility of MIDD. The penetrance of diabetes in MIDD is estimated to be >85%, with an average age at diagnosis of 37 years (range 11–68) (4). It is uncommon among the pediatric population (17).
Characteristics of MIDD and type 1 and type 2 diabetes
. | MIDD (5,14) . | Type 1 diabetes . | Type 2 diabetes . |
---|---|---|---|
Transmission | Matrilinear, monogenic (with genetic modifiers) | Polygenic | Polygenic |
Sex | M < F (1,14,19,99)* | M < F | M < F |
Age at diagnosis | 2nd–4th decade of life (71); 39% aged <35 years (14) | 58% aged <31 years | 80%–85% aged ≥40 years, <1% aged <20 years |
Ethnicity and race | All ethnic groups; described more often in Asian individuals (33) | More common in White people | All ethnic groups, more common in Hispanic, African American, Native American, and South Asian individuals |
BMI | Normal or low (1,14,19) | Similar to that in population without diabetes | Increased |
Onset | Progressive (<20% present acutely) | Acute | Progressive |
Autoimmunity | Absent or rare (69) | Very high (autoantibodies against GAD, IA2, and ZnT8; ICA; and IAA) | Low (100) |
Insulin resistance | Normal or increased (may reflect skeletal muscle heteroplasmy) | Similar to general population, may increase as diabetes progresses (101) | Common |
Acanthosis nigricans | Not observed | Rare | Common |
Comorbidities independent from glycemic control or obesity | Hearing impairment >75% (often precedes diabetes); other MIDD manifestations reflecting heteroplasmy of mutated MT-TL1 gene (4) | Autoimmune endocrine and nonendocrine conditions | None |
Cardiovascular and chronic kidney disease | Primarily a consequence of MIDD, with additional contribution of hyperglycemia (with genetic and environmental contributions) (20,102) | Primarily a consequence of hyperglycemia (with genetic and environmental contributions) | Primarily a consequence of hyperglycemia and obesity (with genetic and environmental contributions) |
Dyslipidemia | Similar to control participants without diabetes matched for BMI (80) | Similar to control participants with well-managed diabetes (103) | Present in 60%–70% of individuals (104) |
Blood pressure | Similar to control participants without diabetes matched for BMI (80) | Hypertension present in ∼30% of individuals (105) | Hypertension present in 50%–80% of individuals (105) |
. | MIDD (5,14) . | Type 1 diabetes . | Type 2 diabetes . |
---|---|---|---|
Transmission | Matrilinear, monogenic (with genetic modifiers) | Polygenic | Polygenic |
Sex | M < F (1,14,19,99)* | M < F | M < F |
Age at diagnosis | 2nd–4th decade of life (71); 39% aged <35 years (14) | 58% aged <31 years | 80%–85% aged ≥40 years, <1% aged <20 years |
Ethnicity and race | All ethnic groups; described more often in Asian individuals (33) | More common in White people | All ethnic groups, more common in Hispanic, African American, Native American, and South Asian individuals |
BMI | Normal or low (1,14,19) | Similar to that in population without diabetes | Increased |
Onset | Progressive (<20% present acutely) | Acute | Progressive |
Autoimmunity | Absent or rare (69) | Very high (autoantibodies against GAD, IA2, and ZnT8; ICA; and IAA) | Low (100) |
Insulin resistance | Normal or increased (may reflect skeletal muscle heteroplasmy) | Similar to general population, may increase as diabetes progresses (101) | Common |
Acanthosis nigricans | Not observed | Rare | Common |
Comorbidities independent from glycemic control or obesity | Hearing impairment >75% (often precedes diabetes); other MIDD manifestations reflecting heteroplasmy of mutated MT-TL1 gene (4) | Autoimmune endocrine and nonendocrine conditions | None |
Cardiovascular and chronic kidney disease | Primarily a consequence of MIDD, with additional contribution of hyperglycemia (with genetic and environmental contributions) (20,102) | Primarily a consequence of hyperglycemia (with genetic and environmental contributions) | Primarily a consequence of hyperglycemia and obesity (with genetic and environmental contributions) |
Dyslipidemia | Similar to control participants without diabetes matched for BMI (80) | Similar to control participants with well-managed diabetes (103) | Present in 60%–70% of individuals (104) |
Blood pressure | Similar to control participants without diabetes matched for BMI (80) | Hypertension present in ∼30% of individuals (105) | Hypertension present in 50%–80% of individuals (105) |
F, female; IAA, insulin autoantibody; ICA, islet cell antibodies; M, male.
*In the absence of population data, the male < female sex ratio observed in most published cohorts may reflect a selection bias (2).
MIDD and Cardiovascular Disease
In people with a confirmed m.3243A>G variant and clinical features of MIDD, cardiac disease was reported in 7 of 46 (15%) (18), 38 of 161 (23.6%) (1), and 20 of 48 (41.6%) (19), depending at least in part on methodology. The pathophysiology is complex and characterized primarily by left ventricular hypertrophy and cardiac failure (4). In asymptomatic people with confirmed m.3243A>G variant, cardiac MRI revealed increased ventricular mass index, increased left ventricular mass to end-diastolic volume index, myocardial thickening, increased torsion, and an increased torsion–to–endocardial strain ratio (reviewed in Finsterer and Zarrouk-Mahjoub [20]). Sudden death has been reported in people with MELAS and high blood and heart heteroplasmy levels for the m.3243A>G variant. Autopsy findings demonstrated myocardial thickening, subendocardial fibrosis, and patchy but prominent cytoplasmic vacuolation and enlargement of cardiomyocytes of the left ventricular myocardium. Conduction defects, ischemic heart disease, and hypertension have also been reported (reviewed in Finsterer and Zarrouk-Mahjoub [20]). Taken together, these data suggest that the primary cause of cardiac disease in MIDD reflects the underlying disorder that is a hallmark of mitochondrial disease: myocyte dysfunction secondary to a reduction in cardiac ATP synthesis. This contrasts with the cardiovascular disease observed in type 1 or type 2 diabetes where micro- and macrovascular diseases resulting from chronic hyperglycemia, dyslipidemia, hypertension, inflammation, or thrombosis in the complex context of genetic, epigenetic, and environmental factors play major roles.
MIDD and Kidney Disease
Among people with confirmed m.3243A>G variant and clinical features of MIDD, renal disease at various stages of severity was reported in 22 of 161 (13.7%) (1) and 37 of 84 (44.0%) (19). The most prevalent finding is focal segmental glomerular sclerosis. Renal disease may precede the diagnosis of either diabetes or deafness or even be the sole end-organ manifestation of the m.3243A>G variant (21). Tubulointerstitial nephropathy and cystic kidney disease have also been reported (4). The pathogenesis involves abnormal mitochondria in the smooth muscle or the glomerular epithelial cells, but the exact mechanism remains poorly understood (4). In one case report, two successive kidney biopsies showed first, mild mitochondrial degeneration in the distal tubular epithelial cells and, 20 years later, nephrosclerosis with interstitial fibrosis and arteriolar hyaline thickening (22). This also contrasts with the pathophysiology of kidney disease in diabetes, where the influence of chronic hyperglycemia and glomerular hyperfiltration leads to glomerular and tubulointerstitial inflammation, glomerulosclerosis, and tubulointerstitial fibrosis. Hypertension, commonly seen in type 2 diabetes and less frequently in type 1 diabetes, further worsens the evolution.
MT-TL1 and Cell Metabolism: In Vitro Data
Heteroplasmy
The human mitochondrial genome encodes 37 proteins: 13 protein subunits (all of which are involved in the oxidative phosphorylation process and code for subunits of ATP synthase [complex V], cytochrome c oxidase [complex IV], cytochrome b [complex III], and NAD dehydrogenase [complex I]), 2 rRNAs, and 22 tRNAs. The primary role of a tRNA is to deliver amino acids during protein translation and elongation. Among these 22 mitochondrial tRNA genes, two correspond to a leucine amino acid: MT-TL1, which is mutated in MIDD and encodes a leucine tRNA the anticodon of which will bind to RNA codon sequences UUA or UUG, and MT-TL2, which encodes a more abundant leucine tRNA that can bind to codons CUA, CUG, CUU, and CUC.
Each cell contains several hundreds or thousands of mitochondria, and each mitochondrion contains several copies of mtDNA. In MIDD, cells contain both mutated and nonmutated mtDNA, a characteristic known as heteroplasmy. Blood heteroplasmy decreases physiologically with aging and is modestly correlated with the presence of diabetes (19). Heteroplasmy is less dependent on age in tissues such as skeletal muscle, hair, and urinary epithelial cells (2,23). Urine, skeletal muscle, and age-corrected blood heteroplasmy correlate significantly with the severity of the disease as determined with the Newcastle Mitochondrial Disease Adult Scale (NMDAS) (23,24).
The intra- and interindividual degree of heteroplasmy varies tremendously, explaining at least in part the variability of MIDD biochemical and clinical expression, even within the same family (2). Limited data on human fetuses harboring the m.3243A>G variant indicate that mutant mtDNA molecules do not segregate much during embryogenesis and that levels of heteroplasmy are relatively uniform across fetal tissues (25,26). Over time, random and nonrandom tissue-specific replication, and segregation of mutant versus normal mitochondrial genomes, may result in varying heteroplasmy levels in different tissues of an affected adult. The mechanisms that mediate this tissue-specific segregation of heteroplasmy load remain poorly understood.
It has been proposed that the MIDD phenotype is associated with lower heteroplasmy for the m.3243A>G variant compared with the more severe MELAS phenotype (27). Indeed, compared with MELAS, MIDD tends to present later in life, and individuals with MIDD tend to have a higher body weight index, a better quality of life, and lower heteroplasmy in saliva and urinary epithelial cells. However, the correlation between the phenotype and the level of heteroplasmy can be modest and variable, not only in the blood (19) (where a broad range of heteroplasmy is observed across all cell types but is markedly reduced in T cells [28]) but also in tissues (29–31).
m.3243A>G Pathogenic Variant and Leucine Incorporation Into Mitochondrial Proteins
The mechanisms by which the pathogenic variant in the MT-TL1 gene impairs cell function have only been studied in vitro (32) and are complex and incompletely understood (33). m.3243A>G affects the structure and methylation of the tRNALeu (34). It causes a marked decrease in aminoacylation of the tRNA by the leucyl-tRNA synthase (35), an effect reversed by overexpression of the enzyme (36). In addition, it impairs the codon-anticodon recognition on the ribosome by preventing the physiological posttranscriptional modification of uridines by taurine at the anticodon wobble position of the leucine tRNA, affecting binding to UUG and, to a lesser extent, UUA (37,38). This results in a decrease in the incorporation of leucine into the mitochondrial proteins involved in the oxidative phosphorylation process and has a major impact on the stability of these proteins, leading to a severe reduction in functional respiratory chain complexes (39,40). As a result of cellular energy deficiency and downstream metabolic effects (41), MIDD/MELAS manifests itself within organs that are most metabolically active such as liver (42), heart, kidney, and brain (43) and/or have weak antioxidant defenses such as the endocrine pancreas (44), the cochlea (45), and the retina (46).
m.3243A>G Variant and Cellular Metabolism
One of the most intriguing aspects of MIDD is the paradox between the level of tissue heteroplasmy and the degree to which cellular metabolism is affected. In vitro experiments suggest that severe impairment of the mitochondrial function only takes place when the referent mtDNA allele is <5%. However, in vivo, severe phenotypes can also be observed with cells containing a much higher proportion of the referent mtDNA allele (47,48).
Two observations clarify this issue. First, using myoblasts homoplastic for the m.3243A>G variant from an individual with MELAS, Sasarman et al. (49) observed a marked contrast between a moderate decrease in overall mitochondrial protein synthesis and an almost complete lack of respiratory chain complexes. Both loss- and gain-of-function biochemical phenotypes were observed, potentially explaining why even mild heteroplasmy can cause a severe biochemical phenotype and why a specific threshold of heteroplasmy for the expression of the mutation cannot be established.
Second, in vitro studies, performed in human osteosarcoma–derived transmitochondrial cytosolic hybrid cells (cybrid cells) carrying increasing levels of heteroplasmy for m.3243A>G, identified discontinuous remodeling of the mitochondrial machinery (27,50–52). Modest levels of cellular heteroplasmy (<20%–30%) were associated with a five- to eightfold induction in the mutated MT-TL1, relatively normal glycolytic activity, preserved or enhanced maximum mitochondrial respiratory capacity (likely thanks to the referent allele mtDNA), increased ATP production (including under hyperglycemic conditions), and oxidative stress. In contrast, high cellular levels of heteroplasmy (50%–90%) were associated with a progressive increase in glycolysis and lactate and higher consumption of glucose but an overall decrease in glucose oxidation in the Krebs cycle and a decrease or only a transient increase in ATP production. Oxidative stress was present in the case of both low and high levels of heteroplasmy.
m.3243A>G Variant and Mitochondrial Metabolic Retrograde Signaling
Changes in m.3243A>G variant levels affect not only mtDNA but also nuclear DNA gene expression, suggesting that interactions between the nuclear and mitochondrial genomes play a role in MIDD phenotype (27,50). This mitochondrial metabolic retrograde signaling remains poorly understood but likely results from a variety of signals including changes in mitochondrial Ca2+ homeostasis and reactive oxygen species (ROS) and altered production of tricarboxylic acid cycle intermediates such as acetyl-CoA and α-ketoglutarate (50). Kopinski et al. (53) showed that changes in mitochondrial metabolites secondary to increasing severity of the m.3243A>G heteroplasmy caused discrete changes in the epigenomic state (including changes in histone acetylation and methylation) that correlate with altered gene transcription and associated clinical manifestations. Retrograde mitochondrial signaling may have positive effects, such as the induction of antioxidant gene expression by Nrf2 and ATF4 in response to the generation of excessive ROS (54). However, it can also have negative consequences, such as the induction of genes associated with neurodegenerative diseases at higher levels of heteroplasmy, potentially explaining the severe neurological phenotype of MELAS (50).
m.3243A>G Variant and the PI3K-Akt-mTORC1 Axis
The mammalian target of rapamycin (mTOR) is a protein kinase that controls cellular metabolism, autophagy, survival, proliferation, and migration, to maintain cellular homeostasis. This cascade includes mTORC1 and mTORC2 and integrates signals from growth factors, stress, energy status, oxygen, and amino acids. mTORC1 stimulates cell growth and metabolism, autophagy, and mitochondrial biogenesis, whereas mTORC2 regulates cell proliferation and survival (55). Activation of the mTORC1 pathway and disruption of the mTORC2 pathway induce insulin resistance (56).
Chung et al. (57) observed an increase in the activity of the PI3K-Akt-mTORC1 pathway that was strongly associated with redox imbalance, oxidative stress, and glucose dependence in fibroblasts from two individuals with a low (30.3%) and a high (86%) level of heteroplasmy for m.3243A>G. Increased activity of the PI3K-Akt-mTORC1 pathway was also observed in the skeletal muscle of an individual with m.3243A>G, suggesting that this signaling pathway is constitutively activated in tissues. Inhibition of the hyperactive PI3K-Akt-mTORC1 pathway not only improved mitochondrial function but also reduced mutant load, suggesting that activation of the PI3K-Akt-mTORC1 axis is a maladaptive response that helps sustain the mtDNA mutant load. Activation of the mTORC1 axis was not found in another mitochondrial disease where the pathogenic variant m.8993T>G affects the function of subunit a of the ATP synthase (ATP6) (58). This observation not only explains why different variants of the mitochondrial genome that affect oxidative phosphorylation may lead to different phenotypes but also that disease-specific approaches need to be developed.
The increase in the activity of the PI3K-Akt-mTORC1 pathway may play a key role in the relationship between heteroplasmy for m.3243A>G and diabetes. To our knowledge, in MIDD, hyperactivation of mTORC1 has not been formally confirmed in tissues other than muscle. In insulin-sensitive tissues, such as the skeletal muscle and adipose tissue, chronic excessive activity of the mTORC1 pathway inhibits insulin signaling and causes insulin resistance (56). In the β-cell, under physiological conditions, mTORC1 plays a positive role in the regulation of β-cell survival. However, when the β-cell undergoes metabolic stress, such as in type 2 diabetes, hyperactivation of mTORC1 contributes to β-cell failure (59).
Taken together, these data show that the m.3243A>G phenotypic variant leads to a complex rewiring of the cell that includes mitochondrial respiratory chain dysfunction, generation of ROS, discontinuous remodeling of the mitochondrial and nuclear genome expression, and overexpression of the mTORC1 pathway that may contribute to insulin resistance and β-cell failure.
MIDD: In Vivo Data
Both type 1 and type 2 diabetes result from a combination of genetic or environmental factors. Ultimately, diabetes develops in response to an absolute or relative deficiency in insulin secretion. In type 1 diabetes, hyperglycemia results primarily from the autoimmune destruction of the β-cells of the pancreas. In type 2 diabetes, hyperglycemia results from the combination of inadequate insulin secretion, excessive or inappropriate glucagon secretion, and resistance to insulin action. In this section, we review the role of insulin secretion, resistance to insulin action, and glucagon secretion in MIDD and propose a model to explain the development of MIDD.
Insulin Secretion in MIDD
The process of insulin synthesis and exocytosis by the β-cell involves sensing blood glucose elevations in response to a meal. Transport of glucose into the β-cell by GLUT2 requires energy and generates ROS. Generation of ROS is further increased in the presence of chronic hyperglycemia and glucose variability (60,61). After glucose phosphorylation by glucokinase, ATP is generated via glycolysis and the Krebs cycle in the mitochondria, leading to a closure of the ATP-sensitive potassium channels, membrane depolarization, opening of calcium channels, and exocytosis of insulin. The generation of ATP depends on the mitochondrial chain of oxidative phosphorylation, which is a major source of superoxide anion, a reactive molecule that can be converted into H2O2 by superoxide dismutase and then into H2O by enzymes such as catalase. The β-cell of the pancreas contains low levels of these antioxidative enzymes compared with highly metabolic tissues such as the liver (62–65). In vitro data have shown that the m.3243A>G variant could affect ATP production by the Krebs cycle (50).
Insulin and C-peptide response to intravenous administration of glucose were found to be decreased in five young individuals (four with diabetes and one with impaired glucose tolerance) from the same family with confirmed m.3243A>G. Unfortunately, there was no measure of skeletal muscle heteroplasmy and no assessment of insulin sensitivity (66). Other observations included progressively increasing fasting and 2-h postprandial blood glucose, detectable but low C-peptide concentrations at baseline and in response to glucagon, and inability to increase insulin secretion in response to various stimuli (67–69). Insulin response to arginine, an established secretagogue that promotes insulin secretion directly by depolarizing the β-cell membrane and increasing intracellular Ca2+ (70), tends to initially be preserved (66,71).
Postmortem evaluation of the pancreas was performed in eight individuals with a clinical diagnosis of MELAS and confirmation of a m.3243A>G variant in skeletal muscle and/or blood. Six had diabetes (pancreas heteroplasmy for m.3243A>G variant ranged from 25% to 72%), and two did not have diabetes (pancreas heteroplasmy for m.3243A>G variant was 45% in one and not detected in the other). Immunohistochemical studies of the pancreas demonstrated a reduction in total islet mass and in the number of β-cells. There was no evidence of insulitis or apoptosis (29). In another person with severe manifestations of MIDD and a confirmed m.3243A>G variant (stroke-like episodes, seizures, diabetes), postmortem analysis findings showed that pancreatic islets were small but that the percent heteroplasmy in the pancreas was relatively low (31%) compared with that in other tissues (all >58%) (72).
Glucagon Secretion in MIDD
In contrast to type 2 diabetes, there is no evidence for an excessive secretion of glucagon in MIDD. Glucagon secretion is normal before (73) and in the early phases of (66) diabetes. The absence of glucagon-producing α-cells in autopsy samples suggests a progressive loss of these cells as diabetes progresses (29). It could contribute to the difficulty of managing MIDD with insulin (74,75).
Resistance to Insulin Action in MIDD
Glucose transporters regulate tissue-specific glucose uptake and metabolism and are critical contributors to the control of whole-body glycemia. In adipose tissue and skeletal muscle, GLUT4 is the main transporter responsible for insulin-regulated glucose uptake. In the liver, >97% of glucose uptake is mediated through a different glucose transporter, GLUT2 (76). Insulin resistance is defined as a decrease in the response of peripheral tissues to increasing amounts of circulating insulin and contributes to elevated blood glucose levels.
In people with MIDD, oxidative stress may be a leading cause of insulin resistance (61), which is associated with the evolution from normal glucose tolerance to impaired glucose tolerance and overt diabetes. Chronic hyperglycemia further aggravates insulin resistance.
We have identified 15 reports with assessment of insulin resistance in individuals with or without hyperglycemia and harboring a m.3243A>G variant (Supplementary Table 1). Interpretation of the results is difficult due to the small size and lack of power of the studies (median no. of individuals 5 [range 1–23]), the variability of A1C, and the different techniques used to assess insulin resistance and because only 4 of 16 studies (17%) included measurement of skeletal muscle heteroplasmy. Decreased insulin sensitivity was observed in many but not all people with m.3243A>G (67,68,71,77,78). Whether insulin resistance develops before (79) or after (71) the onset of diabetes has been debated. Decreased insulin sensitivity (assessed through the insulin sensitivity index [BIGTT-SI] during oral glucose tolerance test and through HOMA of insulin resistance) was reported in 23 m.3243A>G carriers (8 with normal glucose tolerance, 18 with impaired glucose tolerance, and 3 with overt diabetes), suggesting that it may be an early event in the path to diabetes (80). In contrast, among 16 individuals (11 without diabetes and 5 with diabetes), insulin sensitivity (assessed through HOMA of insulin resistance) was lower in those with diabetes but not in those without diabetes (81).
Lindroos and colleagues (82,83) compared insulin secretion and insulin sensitivity in 15 individuals with confirmed epithelial and skeletal muscle heteroplasmy for m.3243A>G. Insulin sensitivity in the skeletal muscle and in the abdominal fat was decreased in all individuals irrespective of glucose homeostasis. In five individuals with normal or impaired glucose tolerance, insulin secretion was maintained, while in seven with previously diagnosed diabetes, insulin secretion was significantly decreased. In contrast, in the liver, fat content and insulin sensitivity were normal, including in individuals who had had diabetes for 6–14 years. The trend toward lower insulin sensitivity was inversely related to percent heteroplasmy in the skeletal muscle: 46% in those with normal glucose tolerance, 76% impaired glucose tolerance, and 80% diabetes. Impaired glucose tolerance or diabetes was present in all individuals with skeletal muscle heteroplasmy >66%.
In a cohort of 51 individuals with MIDD due to m.3243A>G, Jeppesen et al. (31) observed that skeletal muscle heteroplasmy of approximately ≥60% was associated with increasing resting plasma lactate, exercise intolerance, and diabetes or impaired glucose tolerance. None of the people with skeletal muscle m.3243A>G variant allele frequency <65% had overt diabetes, but normal or impaired glucose tolerance was found across the whole range of heteroplasmy observed in the study (range 2%–95%). There was no relationship between blood heteroplasmy and MIDD phenotype, consistent with the poor correlation between blood heteroplasmy and A1C previously reported (19). These data suggest that high skeletal muscle heteroplasmy is necessary but not sufficient for development of diabetes. Increased lactate concentrations, known to promote insulin resistance in skeletal muscle by suppressing glycolysis and impairing insulin signaling, may contribute to insulin resistance (84).
In a case report, one individual with MELAS and normal fasting blood glucose had a 50% decrease in insulin sensitivity and a 17% decrease in stimulated ATP synthesis in gastrocnemius muscle fibers (85).
Proposed Model
We propose that diabetes in MIDD primarily depends on a combination of insulin resistance and impaired β-cell function that is more likely to occur in the presence of higher skeletal muscle heteroplasmy (approximately ≥60%) and more moderate and variable β-cell heteroplasmy (∼25%–72%) for m.3243A>G (Fig. 1). Published data (Supplementary Table 1) suggest that insulin sensitivity is decreased in many people prior to any sign of glucose intolerance and is inversely correlated to skeletal muscle heteroplasmy for m.3243A>G.
Proposed model for the development of impaired glucose tolerance and diabetes in MIDD. The leucine tRNA shows the location of the m.3243A>G phenotypic variant and of the posttranscriptional modification of uridines (U) by taurine at the anticodon wobble position of the leucine tRNA; the combination of high skeletal muscle heteroplasmy and moderate β-cell heteroplasmy for m.3243A>G sends various signals that may cause 1) an increased in mTORC1 activity (leading to progressive insulin resistance and β-cell dysfunction and to increased variant load) and 2) mitochondrial retrograde signaling (leading to changes in the nuclear epigenome). Additional modifying factors affect the phenotype. IGT, impaired glucose tolerance; mTORC1, mammalian target of rapamycin complex 1; TCA, tricarboxylic acid cycle.
Proposed model for the development of impaired glucose tolerance and diabetes in MIDD. The leucine tRNA shows the location of the m.3243A>G phenotypic variant and of the posttranscriptional modification of uridines (U) by taurine at the anticodon wobble position of the leucine tRNA; the combination of high skeletal muscle heteroplasmy and moderate β-cell heteroplasmy for m.3243A>G sends various signals that may cause 1) an increased in mTORC1 activity (leading to progressive insulin resistance and β-cell dysfunction and to increased variant load) and 2) mitochondrial retrograde signaling (leading to changes in the nuclear epigenome). Additional modifying factors affect the phenotype. IGT, impaired glucose tolerance; mTORC1, mammalian target of rapamycin complex 1; TCA, tricarboxylic acid cycle.
The underlying genetic defect results in oxidative stress that initiates and exacerbates a vicious cycle at least partly through the mTORC1 cascade, leading to a further increase in insulin resistance, higher demand for insulin (86), progressive chronic hyperglycemia (87), and β-cell dysfunction (88). Glucose homeostasis remains preserved as long as the β-cell of the pancreas is able to secrete enough insulin to overcome decreased insulin sensitivity.
Whether the consequences of heteroplasmy for m.3243A>G on cell metabolism reviewed in this article also apply to other characteristics of MIDD, such as deafness, is unclear. For diabetes to develop, we propose that both peripheral insulin sensitivity and production of insulin by the β-cell need to be affected by the m.3243A>G phenotypic variant. In contrast, the development of hearing impairment (the severity of which correlates with muscle heteroplasmy (89), likely requires impaired cellular metabolism of only one tissue.
Despite being a clear pathogenic factor, heteroplasmy, not only in the blood but also in tissues, does not fully explain the phenotypic variability of MIDD (18,30), even with the variable tissular antioxidant defenses taken into account (29,62,64). Additional factors are increasingly recognized, including interaction between the mitochondrial and nuclear genomes (which is responsible for the synthesis of the majority of factors involved in mitochondrial function) (90), concomitant nuclear risk factors for type 2 diabetes (9), additional nuclear or mitochondrial variants (91), mtDNA copy number (23), age, sex, and environmental factors (33), decreased physical activity (which may reduce mtDNA copy number [92]), influence of the paternal genome (93), and metabolic characteristics of the person (recently shown to bear similarities between different phenotypes) (94). The extent to which they contribute to the development of diabetes remains poorly understood.
Conclusion
In this article, we propose a new hypothesis for the development of impaired glucose tolerance and diabetes in MIDD that suggests a major role for insulin resistance as an early event, in addition to suboptimal β-cell function. Presently, MIDD is mostly managed according to type 1 or type 2 diabetes guidelines. Our model suggests that these guidelines are not appropriate for MIDD. Type 1 diabetes diagnosis leads to the early initiation of insulin therapy, which may not be needed for several years in MIDD. Type 2 diabetes management focuses on the treatment of obesity, which is usually not an issue in MIDD, and on cardiorenal protection in the context of hyperglycemia and obesity.
A better understanding of the characteristics of MIDD may lead to the evidence-based development of a disease-specific algorithm for the management of MIDD. First, we need to gain a better understanding of the mechanisms leading to MIDD. Future studies designed to close existing gaps in knowledge should include the systematic determination of skeletal muscle heteroplasmy and of insulin sensitivity and a better assessment of the mitochondrial and nuclear genome, as well as of environmental characteristics of the people such as physical activity and diet. Second, new and existing noninsulin agents, including metformin (95), thiazolidinediones (96), sodium–glucose cotransporter 2 inhibitors (97), and glucagon-like peptide 1 (GLP-1) agonists (98), should be (re)investigated with a focus on their effect on oxidative stress and on the PI3K-Akt-mTORC1 axis at doses that are safe and do not cause undesirable weight loss.
This article contains supplementary material online at https://doi.org/10.2337/figshare.27728133.
Article Information
Acknowledgments. The authors thank Drs. Jean-Francois Lemay (University of Calgary) and Melanie Henderson (McGill University) for critical review of the manuscript.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.