Challenges and opportunities to bridge translational to clinical research for personalized mitochondrial medicine

Abstract

Mitochondrial disorders are a group of rare and heterogeneous genetic diseases characterized by dysfunctional mitochondria leading to deficient adenosine triphosphate synthesis and chronic energy deficit in patients. The majority of these patients exhibit a wide range of phenotypic manifestations targeting several organ systems, making their clinical diagnosis and management challenging. Bridging translational to clinical research is crucial for improving the early diagnosis and prognosis of these intractable mitochondrial disorders and for discovering novel therapeutic drug candidates and modalities. This review provides the current state of clinical testing in mitochondrial disorders, discusses the challenges and opportunities for converting basic discoveries into clinical settings, explores the most suited patient-centric approaches to harness the extraordinary heterogeneity among patients affected by the same primary mitochondrial disorder, and describes the current outlook of clinical trials.

Introduction

Mitochondria are energy powerhouses and central regulators of cell fate, redox balance, ion homeostasis, and metabolism [1]. As ubiquitous organelles, they act as central hubs for pathways involved in bioenergetic metabolism, neural development, synaptic activity, and connectivity. Therefore, mitochondrial dysfunction contributes to a broad spectrum of pathophysiological states beyond dysregulated adenosine triphosphate (ATP) synthesis, including dysregulation of cellular homeostasis, organellar communication, ion homeostasis, nutrient responses, stress responses, and the nuclear epigenome [2]; hence making mitochondrial dysfunction a linchpin for a myriad of diseases, including neurodevelopmental disorders, neurodegenerative and neuropsychiatric disorders, autism spectrum disorders (ASD), autoimmune neurological disorders, and aging [[3], [4], [5], [6], [7], [8], [9]].

This review focuses on primary mitochondrial diseases (PMDs), which are a heterogeneous group of rare genetic and metabolic disorders caused by pathogenic variants mapping in the nuclear and/or mitochondrial genome resulting in mitochondrial dysfunction and/or abnormal mitochondrial structure [[10], [11], [12], [13]]. Worldwide, PMDs collectively affect about 1 in 5000 live births, with individual PMD being ultra-rare [12]. More specifically, a cohort-based study in the North East of England with little consanguinity has provided the most comprehensive estimated prevalence of adult-onset PMDs with a prevalence of 2.6 cases or 9.6 cases per 100,000 due to a nuclear or mitochondrial pathogenic variant, respectively [12]. However, disease-based epidemiological studies are challenging and only provide with an estimated prevalence of PMDs due to a score of factors [11,12,23]: 1) the extensive clinical heterogeneity combined with the high variability of disease onset (childhood versus adult) and severity among patients harboring the same pathogenic variant; 2) access to a large study population of patients with a suspected or diagnosed PMD to minimize variability; 3) variations in the inclusion criteria being used in the diagnostic algorithms; 4) variation in prevalence due to genetic founders and consanguinity; 5) identity of the pathogenic variant; 6) the affected genome, mitochondrial or nuclear; 7) challenging clinical diagnosis due to a complex genotype-phenotype correlation often failing to result in an accurate diagnosis; 8) environmental modifiers, epigenetic factors, and haplotype groups, all contributing to the extraordinary heterogeneity of the clinical phenotypes; and 9) up to 80 distinct genes causing a specific PMD as in the case of Leigh syndrome.

While PMDs caused by nuclear pathogenic variants have a mendelian inheritance, PMDs due to mitochondrial pathogenic variants have a maternal inheritance [[14], [15], [16], [17]]. PMDs share a defective oxidative phosphorylation (OXPHOS) pathway responsible for mitochondrial ATP synthesis, which predominantly affects organs with high energy demands and a dependence on aerobic metabolism, such as the central and peripheral nervous system (CNS and PNS, respectively), musculosketal system, and cardiac system [17,18]. Thus, patients with a PMD exhibit a wide spectrum of clinical manifestations, including muscle weakness, developmental delays, neurological abnormalities, and metabolic disturbances [3,[19], [20], [21]]. Currently, patients with a progressive and multi-systemic PMD only have access to palliative therapies that fail to halt the progressive decline, resulting in significant disability, morbidity, and premature death [22]. The devastation wrought by PMDs underscores the urgency to address thus unmet medical need, to improve the meandering journey for clinical and genetic diagnostics along with assessment of patient-specific mitochondrial phenotypes endured by most patients, and to develop novel therapeutic modalities and drug candidates (Fig. 1).

Fig. 1.

General pipeline from translational to clinical research for primary mitochondrial disorders. The left column illustrates recruitment of family members suspected of an inherited primary mitochondrial disorder for a skin biopsy to derive dermal fibroblasts that are used for genetic testing and determination of the heteroplasmic levels for mitochondrial pathogenic variants. In the absence of animal models for mtDNA variants, patient-derived cells are used to determine: 1) mitochondrial signature via investigations of OXPHOS metabolism and mitochondrial homeostasis; 2) comprehensive omics-based investigations; 3) high-throughput drug screening to identify potential drug candidates and OXPHOS and omics-derived biomarkers; 4) Phase 1 of clinical trials to assess safety of drug candidates; 5) Phases II and III to assess efficacy of the tested drug candidate and to validate biomarkers and end-points; and 6) FDA approval of the drug product for therapeutic modality.

Challenges in clinical testing for primary mitochondrial disorders

Clinical heterogeneity and the long clinical diagnostic journey

As a group, primary mitochondrial disorders represent one of the most common of the inherited disorders of metabolism [18,23]. Their clinical presentation and severity can vary significantly among unrelated and related patients, making diagnosis and treatment challenging (Table 1). This variability complicates the clinical diagnosis, as it may not be immediately evident that a patient's symptoms are related to mitochondrial dysfunction or new symptoms or signs may develop over time.

Table 1.

Phenotypic manifestations associated with PMDs.

Central & Peripheral Neurological Symptoms	Visual and Hearing Symptoms	Skeletal and Cardiac Symptoms	Endocrine and Reproductive Symptoms	Gastro-Intestinal and Renal Symptoms
Encephalopathy	Visual loss	Myopathy	Diabetes mellitus	Gastroparesis
Epilepsy	Progressive external ophthalmoplegia	Exercise intolerance	Gestational diabetes	Gastro-intestinal dysmotility
Ataxia	Ptosis	Myoclonus	Lactic acidosis	Dysphagia
Stroke-like-episodes	Optic atrophy	Arrhythmia	Short stature	Cyclical vomiting
Migraine	Retinitis pigmentosa	Cardiomyopathy	Hypoparathyroidism	Gastroesophageal sphincter dysfunction
Cortical blindness	Sensorineural hearing loss	Cardiac conduction defects	Hypothyroidism
Dystonia			Adrenal insufficiency	Pancreatitis
Tremors			Hypogonadotropic hypogonadism	Hepatopathy
Parkinsonism			Infertility	Nephropathy
Development delay/regression				Fanconi syndrome
Cognitive impairment
Peripheral neuropathy

Canonical symptoms associated with specific PMDs are indicated in italics.

Even in the cases of PMDs with canonical symptoms, their clinical diagnosis remains challenging given that many unrelated and related patients do not exhibit all the canonical signs with the same chronology and may only present one of the clinical manifestations, such as stroke-like episodes with lactic acidosis linked to mitochondrial encephalomyopathy with lactic acidosis and stroke-like episodes (MELAS) [24], progressive external ophthalmoplegia associated with chronic progressive external ophthalmoplegia (CPEO) [25], and myoclonus linked to myoclonic epilepsy associated with ragged red fibers (MERRF) [26]. Another layer of complexity originates from the fact that many phenotypic manifestations of PMDs are also common in the general population, such as migraine, hearing loss, and diabetes mellitus, thereby failing to raise suspicion of PMD [27].

This clinical complexity is further accentuated among patients with a maternally inherited mitochondrial disease caused by a mitochondrial pathogenic variant, which exist under a state of heteroplasmy [28]. This stems from the fact that not all the multi-copies of the mitochondrial DNA (mtDNAs) harbor a specific pathogenic mitochondrial variant, which results in variable ratios of mutant and wildtype mtDNAs within a cell. Above a certain heteroplasmic threshold, patient's cells display an ATP deficit due to a deficient OXPHOS pathway, a response that is organ-specific based on their dependence on aerobic respiration and high energy demand [10,17]. As a result, the phenotypic expression of a maternally inherited mitochondrial disease is dictated by heteroplasmy, which may vary over time and among organs. Moreover, heteroplasmy of specific mitochondrial pathogenic variants can alter and dictate the time course of phenotypic manifestations among unrelated and related patients [28]. Patients with high heteroplasmic or near-homoplasmic pathogenic mitochondrial variants tend to exhibit an aggravated disease burden and progression, thereby easing the path making it more likely that a PMD diagnosis is considered and confirmed. Heteroplasmic levels of some mitochondrial pathogenic variants also impact the differential clinical presentation, as it is the case for the pathogenic m.8993T ​> ​G variant mapping in the mitochondrial MT-ATP6 gene. A heteroplasmy of m.8993T ​> ​G at 90 ​% or above results in maternally inherited Leigh syndrome (MILS), while a heteroplasmic level between 70 ​% and 90 ​% results in the progressive neurodegenerative disorder neuropathy ataxia and retinitis pigmentosa (NARP) [[29], [30], [31]].

Another level of complexity resides at the level of the nuclear genome given that it encodes the majority of mitochondrial proteins resulting in tissue-specificity of the mitochondrial proteome [[32], [33], [34]] and modulation of the penetrance of heteroplasmic mitochondrial pathogenic variants [28]. Thus, the nuclear genome acts as a major influencer of mitochondrial genome via the nucleo-mitochondria anterograde signaling, lending credence to the concept that the nuclear background of patients in part manipulate the wide spectrum of clinical phenotypes and age of onset. Congruent with this concept are several studies highlighting the nuclear background as a key modulator of heterogeneous phenotypic manifestations among patients: 1) a large study on a UK cohort of patients harboring the MELAS variant m.3243A ​> ​G [35]; 2) an anecdotal study on monozygotic twins harboring the m.3243A ​> ​G variant and almost identical clinical phenotype and age of onset [36]; and 3) our case study on an asymptomatic mother and a symptomatic daughter, both carrying the rare near-homoplasmic MELAS m.1630A ​> ​G, confirming the key contribution of the nuclear background in modulating the penetrance of a mitochondrial pathogenic variant and its clinical phenotype [37].

Since PMDs are ill-defined multi-systemic diseases with extensive phenotypic pleiotropy, their diagnosis by non-specialist clinicians is challenging, arduous, and time-consuming. Thus, most patients with a suspected PMD suffer from a protracted clinical diagnosis, which requires a specialist in mitochondrial disorders, resulting in a long diagnostic odyssey (Fig. 2). In addition, given the low prevalence of PMD, the number of patients with a specific PMD is usually small and dispersed over a broad geographical area, which consequently hampers the studies of natural history of these PMDs. These collective impediments have serious repercussions on choosing meaningful endpoints and assembling cohorts of patients with similar phenotype at similar stage of the disease to test candidate drugs in clinical trials.

Fig. 2.

Pipeline for clinical and genetic diagnoses of patients suspected of a primary mitochondrial disease. A schematic representation of a patient harboring a mixed mitochondrial population with healthy (blue) mitochondrial and diseased (red) mitochondria, which varies among different organs giving rise to heterogeneous neurological and non-neurological symptoms. Beneath is indicated the diagnostic workflow for intertwined and comprehensive clinical and genetic investigations.

Difficulties to characterize genetic heterogeneity

Confirming a mitochondrial disorder diagnosis often requires invasive procedures, such as muscle biopsies. These procedures are uncomfortable and carry risks, making them less desirable for both patients and clinicians. Due to their invasiveness, muscle biopsies are no longer recommended by the Mitochondrial Medicine Society for diagnostic purposes. Pediatric patients suspected of a mitochondrial disease undergo this surgical procedure under a general anesthesia, which often results in a metabolic crisis [38]. The recent advent of next-generation sequencing (NGS) for whole exome sequencing and whole genome sequencing [39,40] has curtailed the need for muscle biopsies for diagnostic purposes. However, this high-throughput genomic testing has emphasized the need of multidisciplinary approaches, involving geneticists, neurologists, and other specialists, to facilitate more accurate and timely diagnoses.

In the earliest years of mitochondrial medicine, diagnoses were made by methodologies in widespread use, well before commercial gene testing, biochemical lab analysis, muscle biopsy, histochemical analysis, electron microscopy, and enzymatic and polarographic analysis of the electron transport chain (ETC). Given that the common element of many mitochondrial diseases was found in abnormal muscle function, early studies used the ease of obtaining skeletal muscle for biochemical analysis [41,42]. An adaption of the Gomori trichrome staining technique then led to identification of mitochondrial proliferation in muscle by light microscopy, which correlated with the so-called ragged red fibers [43]. This initial diagnostic technique consists of a concomitant staining for cytochrome c oxidase (COX) and succinate dehydrogenase (SDH) in cross-sections of frozen muscle biopsies. While increased SDH staining indicates abnormal compensatory mitochondrial proliferation in the subsarcolemmal space, decreased COX staining illustrates mitochondrial dysfunction in COX-deficient muscle fibers [44].

In recent years, there have been notable developments in diagnostic approaches, such as tailored gene panels targeting critical known mitochondrial functions and more recently the use of NGS technologies that allow for comprehensive analysis of both nuclear and mitochondrial DNA [40,45,46]. This has revolutionized the accuracy and efficiency of genetic diagnosis, enabling healthcare professionals to identify a broader spectrum of mitochondrial disorders and associated pathogenic variants than ever anticipated. Given that PMDs exhibit extreme genetic heterogeneity with pathogenic nuclear variants mapping in about 1500 genes associated with mitochondrial functions [11], identifying a causative pathogenic variant among this vast genetic landscape is challenging, time-consuming and costly. Consequently, a majority of patients remain without a reliable genetic diagnosis.

Thus, laboratory diagnosis of mitochondrial disease depends on a multipronged analysis that can include biochemical, histopathologic, and genetic testing. Each of these modalities has complementary strengths and limitations in diagnostic utility. Despite remarkable technical progress, challenges persist in terms of the complexity of mitochondrial disorders, which can still lead to misdiagnosis or incomplete understanding of the underlying pathology. In essence, the current landscape for clinical and genetic testing of mitochondrial disorders reflects a field in transition, with increasing potential for precise diagnosis, but with continued room for improvement in understanding and addressing these complex conditions. The genotype-phenotype correlation is challenging due to the clinical and genetic heterogeneity, a challenge magnified by the multisystemic nature of mitochondrial disorders ranging from mild to severe phenotypes with a broad range of onset from infancy to adulthood, all of which add layers of complexity to reach an accurate diagnostics.

Limited biomarkers

Biomarkers for mitochondrial disorders are limited and often non-specific [[47], [48], [49]]. Traditional biomarkers, such as lactate, pyruvate, and creatine levels, are routinely measured in plasma and/or cerebrospinal fluid (CSF), but lack the specificity required for early and accurate diagnosis and are therefore only suggestive of a PMD [48]. Nevertheless, the analysis of mitochondrial metabolites, such as lactate and pyruvate in blood or cerebrospinal fluid, can provide valuable insights into mitochondrial energy production and dysfunction [50].

Lactate levels are not consistently elevated in blood or CSF of patients suspected of a PMD. For example, patients with Leber's hereditary optic neuropathy (LHON), Kearns-Sayre syndrome (KSS), or Leigh syndrome (LS) exhibit normal lactate blood levels [51], while patients with MELAS have elevated lactate levels in blood and CSF [52]. Since elevated blood lactate levels are also observed in unrelated pathologies, such as seizure and CNS infection, this biomarker by itself is not reliable for a definitive diagnosis of a PMD [53] or PMD disease status.

The plasma metabolite, pyruvate, is part of the diagnostic chemistry profile performed in clinical settings despite its instability and susceptibility to inadequate specimen collection thereby diminishing its reliability as a diagnostic biomarker [47]. However, the ratio of lactate to pyruvate in blood and/or CSF provides a valuable diagnostic biomarker with the caveat that it is limited to PMDs affecting the CNS.

Although elevated blood levels of alanine, glycine, and proline, are associated with PMDs caused by a defective OXPHOS pathway, their sensitivity and specificity remain ambiguous [47]. Furthermore, high blood levels of alanine are not specific to patients with a PMD, but also to patients with pyruvate metabolism disorders as a result of accumulation of alanine from the conversion of cytosolic pyruvate [54].

Creatine has emerged as a recent potential biomarker for PMDs based on its well-known link with mitochondrial bioenergetics [55]. However, by itself, it is not a sensitive marker for PMDs based on a 2013 study demonstrating that only 28 ​% of a cohort of 33 patients with different PMDs exhibit an elevated level of creatine [56]. The authors also reported a high percentage of false positives (43 ​%) and low sensitivity (60 ​%). Nevertheless, this biomarker in conjunction with other biomarkers could strengthen the suspected diagnosis of a PMD.

Recently, the fibroblast growth factor 21 (FGF-21) has shown to be a reliable first-line diagnostic biomarker for PMDs with myopathy [[57], [58], [59]]. The main caveat with this biomarker is the absence of altered serum levels in patients with a PMD predominantly affecting the nervous system, such as mitochondrial recessive ataxia syndrome (MIRAS), which exclude many PMDs leading to false negative diagnosis [57]. Thus, it appears that FGF-21 serum levels are a direct consequence of skeletal muscle pathology caused by a deficient OXPHOS metabolism. The growth differentiation factor 15 (GDF-15) is another reliable first-line diagnostic biomarker for PMDs with or without muscle pathology, as patients with different PMDs, such as MELAS, KSS, LS, and overlapping MELAS/LS, display elevated serum levels of GDF-15 [60]. A recent study from the epidemiological survey of the Emilia-Romagna Region in Italy (ER-MITO study) interrogated whether measuring both serum levels of FGF-21 and GFD-15 could improve the diagnostic significance for PMDS. Using a large cohort of patients with mixed primary mitochondrial pathology, the authors demonstrated the beneficial diagnostic value of concomitant increase in serum levels of FGF-21 and GDF-15 in patients with MELAS or MERRF, two PMDs associated with mitochondrial translation defects caused by mitochondrial pathogenic variants mapping in mitochondrial genes encoding mt-tRNAs [61].

Most recently, the measurement of serum levels of mitochondrial DNA (mtDNA) or its fragments, known as cell-free mtDNA (cf-mtDNA), has gained attention as a potential non-invasive biomarker for PMDs given its elevated levels with mitochondrial dysfunction in various diseases [62]. The 2020 ​ER-MITO study on MELAS patients has provided the seminal finding of a strong association between cf-mtDNA levels and occurrence of stroke-like-episodes, an indication of the progressive nature of this mitochondrial neurodegenerative disease [61]. Thus, further studies are warranted to decipher whether increased plasma levels of cf-mtDNA may expose the wave of neuronal loss occurring during stroke-like episodes and cause a plausible state of inflammation in the CNS.

In sum, ongoing research is focused on identifying novel and reliable biomarkers specific to PMDs to facilitate early diagnosis, stratification of patients with the same PMD, and monitoring of disease progression.

Limited knowledge of the pathophysiological mechanisms for PMDs

The lack of clinically relevant animal models for most PMDs stunts our current knowledge of underlying causes of PMDs and their pathophysiological molecular mechanisms. This is further compounded by the clinical heterogeneity among patients with a specific PMD in terms of disease epidemiology, manifestations, disease natural course, and progression. Not only this leads to a serious gap in clinical management of patients with PMDs, but it also has serious repercussion in the design of innovative therapeutic modalities and the discovery of novel drug candidates.

Engineering murine models for PMDs caused by pathogenic mitochondrial variants has been challenging owing to the multicopy nature of the mitochondrial genome, the limited tools to edit the mitochondrial genome, and the resistance of the mitochondrial genome to transgenic and directed genome editing technologies [17]. The challenging and time-consuming transmitochondrial techniques led to the successful generation of the first murine model mimicking LHON [Li et al., 2012]. This method used the murine heteroplasmic mitochondrial variant m.13997G ​> ​A, which is the equivalent of the human mitochondrial pathogenic variant, m.14600A ​> ​A, known to cause optic atrophy and cerebellar ataxia in patients [Malfatti et al., 2007]. In 2001, this transmitochondrial technique has successfully created the Mito-mouse harboring a 4696 base-pair deletion (ΔmtDNA4696), which includes six tRNA genes and seven structural genes [Nakada et al., 2001]. This Mito-mouse model shows mitochondrial dysfunction in several tissues, a phenotype that is lethal at the age of six months. Several additional Mito-mouse models were successfully created and fully detailed in recent reviews [66; Khotina et al. 2023]. Although this technique allows to reach a heteroplasmy above 50 ​%, none of these murine models do mimic key metabolic and phenotypic landmarks, such as ragged-red fibers in patients with MERRF [17; Khotina et al., 2023].

Although PMDs due to pathogenic nuclear variants are more numerous than those caused by pathogenic mitochondrial variants, the number of clinically relevant murine models for the former PMDs remains limited [[63], [64], [65]]. In light of a recent review dedicated to the current progress on engineering murine models for PMDs due to either mtDNA or nuclear pathogenic variants [66,], this review focuses on two murine knockout models for Leigh syndrome (LS) since it is the most common pediatric PMD, a fatal subacute necrotizing encephalomyopathy with an early onset and variable clinical presentation provoked by Complex I deficiency due to pathogenic nuclear variants [67,68]. The Ndufs4 knockout mouse (Ndufs4^−/−) is the most studied LS mouse model generated by disrupting the Ndufs4 gene encoding the mitochondrial Complex I NADH dehydrogenase-ubiquinone-FeS 4 (NDUFS4) [69], even though mutations in the Ndufs4 gene rarely cause LS in humans [70]. This mouse model is clinically relevant and therefore commonly used as a pre-clinical model to test innovative therapeutic modalities for LS and LS-like disorders, such as hypoxia [[71], [72], [73], [74], [75], [76]]. With the advent of CRISPR-base nuclear genome editing, we anticipate that additional models for PMDs caused by pathogenic nuclear variants will be created. However, it is worth emphasizing that murine models do not always result in the best model for human PMDs. This limitation is illustrated by the Surf1 knockout mouse (Surf1−/−), which exhibits little phenotypic manifestations of LS as observed in LS patient with pathogenic nuclear variant mapping in the Surfeit Locus Protein 1 (Surf1) gene [77]. Despite their limitations, murine animal models can still provide molecular and genetic clues on the pathogenic mechanisms of PMDS, thereby advancing our understanding of the molecular etiologies of PMDs.

Worth highlighting are the key contributions of three powerful genetic models, Drosophila melanogaster, Caenorhabditis elegans and Danio rerio (zebrafish) in mitochondrial biology, pathogenic mechanisms of PMDs, and the drug development process [66,[78], [79], [80]]. Although these animal models do not faithfully mimic the clinical symptoms associated with a specific PMD, they are easily amenable to genetic manipulations, which further our understanding of mitochondrial inheritance [81,82], discovery of novel disease alleles [83,84] and the molecular pathogenetic mechanisms linked to specific clinical symptoms in patients [[85], [86], [87]].

Generating murine models for PMDs due to pathogenic mitochondrial variants is further hindered by the limited molecular tools to reproducibly edit the mitochondrial genome and to mimic tissue-specific heteroplasmic levels and phenotypic expression of a specific pathogenic mitochondrial variant [88]. The CRISPR/Cas9 technology cannot be used to engineer murine models for maternally inherited PMDs, given the absence of a bone fide endogenous mechanism to import nucleic acids in the mitochondrial matrix [89,90]. The current mtDNA editing tools, such as restriction endonucleases, zinc finger nucleases (mitoZFNs), transcription activator-like effector nucleases (mitoTALENs) and mito-Tev-TALE, have many drawbacks to efficiently eliminate mutant mtDNA, thereby preventing their use in clinical setting [91]. Novel promising molecular tools, such as the protein-based editor Double-stranded DNA deaminase (DddA) [92] and mitochondrial-targeted meganucleases (mitoARCUS) [93] are being developed [94]. Thus, the technology of editing mtDNA is still years behind with lots of different types of hurdles to overcome before transitioning to clinical therapy.

Consequently, these hurdles have limited our knowledge on the molecular pathogenic mechanisms controlling clinical heterogeneity, multisystemic phenotypic manifestations, tissue specificity, age of onset, disease progression, the penetrance of pathogenic mitochondrial variants, and the impact of environmental modifiers on mitochondrial genotypes. These remaining challenges have a profound impact on patient care and clinical trials to test innovative therapeutic avenues.

Opportunities in clinical testing for primary mitochondrial disorders and personalized mitochondrial medicine

Patient-derived cellular paradigms

To circumvent the lack of appropriate animal models, patient-derived cells provide suitable pre-clinical paradigms to decipher the molecular pathogenic of PMDs, to design innovative and tailored therapeutic modalities, to assess the therapeutic potential of drug candidates, and to stratify patients into groups as low and effective responders to the drug candidate, based on their mitochondrial phenotypic recovery using multifunctional assays (Fig. 1). This stratification is particularly pertinent to patients with PMDs characterized by highly heterogeneous phenotypic heterogeneity. Among the many advantages specific to patient-derived cellular paradigms is the ability to investigate pathogenic nuclear and mitochondrial variants within the context of the patient's nuclear background in order to elucidate the genotype-phenotype correlation with the goal of understanding the extreme clinical heterogeneity among patients with the same pathogenic genotype. Several recent studies have revealed that the critical role of the nuclear genetic determinants plays to modulate the penetrance of pathogenic mitochondrial variants [[35], [36], [37],95]. Thus, patient-derived cellular paradigms are best suited to identify key genetic determinants for modulating disease severity and prognosis, congruent with the need to apply a personalized medicine-based approach for clinically heterogenous cohorts of patients with PMDs.

Although lymphocytes derived from peripheral blood samples and epithelial cells collected from patient's urine and cheek swabs, respectively, are readily available for diagnostic purposes, they have limited use for in-depth genetic analyses since they cannot be propagated to generate a repository of patient-derived cells for multi-omics and mitochondrial functional studies. Given the invasiveness of muscle biopsies with a risk of metabolic decompensation during general anesthesia, particularly in pediatric patients, the derived muscle cells are not ideal to carry out the follow-up multi-omics and functional analyses to discover clinically relevant genetic biomarkers for PMDs [96]. Furthermore, this patient-derived cellular paradigm is irrelevant in the context of PMDs without muscle pathology, such as in the case of primary mitochondrial hepatopathies with an OXPHOS deficiency for which a liver biopsy is advocated for accurate diagnosis [97].

In contrast, skin biopsies are not invasive and provide a versatile patient-derived cellular platform with unique advantages. The derived dermal fibroblasts are ideal for comprehensive multi-omics studies, mitochondrial functional analyses, and high-throughput screening of drug candidates for mitochondrial phenotypic rescue [17]. Patient-derived fibroblasts are also used to generate human induced pluripotent stem cells (iPSCs), which can be differentiated in somatic cell lineages clinically relevant to tissue-specific phenotypic manifestations exhibited by patients with PMD [98]. Given the lack of diverse and clinically relevant non-human models for PMDs, iPSCs are currently the most amenable patient-derived cellular system for lineage-specific high-throughput screening for drug discovery and multi-omics investigations toward personalized mitochondrial medicine. However, this patient-derived cellular paradigm specific for a PMD exhibits several limitations: 1) an expensive and lengthy process leading to the accumulation of deleterious genomic and phenotypic changes [99]; 2) requirement of a high-resolution DNA-fingerprinting to ensure valid disease models and corresponding isogenic controls [100]; 3) difficulty to consistently achieve a clinically relevant cellular phenotype for validating efficacy of targeted treatments or for high-throughput drug screening [101]; 4) limited number of differentiated and mature cell types with matching functional endpoints that are disease-relevant [102]; 5) Maintenance of high heteroplasmic levels within a phenotypic range of patient-specific mtDNA pathogenic variants during the differentiation process [103]; and 6) reprogramming into iPSCs can reset cellular age of lineage-specific derived cells, such as neurons [104], and/or induces mitochondrial rejuvenation [105], thereby no longer exhibiting the original pathophysiological characteristics of a PMD.

Reprogramming of patient-derived fibroblasts into iPSCs is most advantageous for patients with a PMD caused by pathogenic nuclear variants given that the patient's nuclear background contributes to the phenotypic expression and modulates the penetrance of these pathogenic variants in part as a consequence of secondary variants. The nuclear editing tool, CRISPR-Cas9, has been instrumental to generate several iPSC models for PMDs [98].

In contrast, the iPSC-based cellular platform exhibits several drawbacks and challenges in the case of PMD caused by a pathogenic mitochondrial variant, principally due to mtDNA heteroplasmy, which contributes to significant variability among iPSC lines and derived lineages as a result of heteroplasmic shift with a bias toward loss of mtDNA load [103]. More specifically, high heteroplasmic levels of the main pathogenic mitochondrial variant m.3243A ​> ​G known to cause MELAS induces neuronal death and inhibits cardiac lineage commitment [106]. However, mitochondrial bioenergetic analyses of specific iPSC-derived lineages harboring high heteroplasmic levels of the m.3243A ​> ​G, including retinal pigment epithelium cells, fibroblasts, and cortical neurons, reveal a deficient OXPHOS metabolism with decreased oxygen consumption rate suggesting partial recapitulation of the MELAS mitochondrial phenotype [107]. Thus, these encouraging initial studies on a handful of iPSC-generated specific lineages with the m.3243A ​> ​G remain incomplete with many fundamental unresolved questions due to the difficulty to harness the process of directed lineage restriction in the context of a broad heteroplasmic range. Similar studies must be extended to additional mitochondrial heteroplasmic variants.

Ultimately, iPSC-derived organoids will be an instrumental and complementary platform to elucidate the pathogenic mechanisms controlling the genotype-phenotype correlation for diverse maternally inherited PMDs with the chief objective of designing personalized therapeutic modalities to halt the progression of these intractable PMDs [108].

Multi-pronged investigations of the mitochondrial phenotype using patient-derived cells

The integration of functional assays with genetic analysis enhances the accuracy of diagnosis and enables a more comprehensive understanding of disease mechanisms. Functional mitochondrial investigations, such as complexome analyses, OXPHOS enzyme measurement and oxygen consumption rate, play a pivotal role in assessing the severity of a specific PMD by measuring the functional status of mitochondria directly in patients. They not only provide valuable insights into the amplitude of dysregulation of mitochondrial functions, but also are key to decipher the genotype-phenotype correlation. Thus, these comprehensive analyses aim at characterizing patients' mitochondrial signature and at tailoring palliative treatment regimen to abate the patient's phenotypic manifestations.

Pathogenic nuclear and mitochondrial variants causing PMDs map in genes encoding proteins with known roles in mitochondrial homeostasis and bioenergetics, including OXPHOS structural subunits and assembly factors, mtDNA replication and maintenance, mitochondrial transcription and translation, mitochondrial dynamics, mitochondrial biogenesis, mitophagy, and OXPHOS metabolism.

Using patient-derived cells from muscle biopsies and skin biopsies, assessment of OXPHOS enzymatic activities, also referred to as respiratory chain enzyme activities, can be indicative of specific PMDs. The most frequently observed isolated OXPHOS deficiencies linked to PMDs include those for Complex I (NADH:ubiquinone oxidoreductase), as in the case of LHON [109], and Complex IV (cytochrome c oxidase) predominantly causing LS [110], possibly due to their large number of mitochondrially encoded subunits. However, combined deficiencies of Complexes I, III, and IV are typically due to deficiency of mitochondrial DNA replication, RNA metabolism or translation [[111], [112], [113]].

OXPHOS analysis using the Seahorse Extracellular Flux Analyzer, is at the forefront of molecular diagnosis of PMDs using fibroblasts derived from skin biopsies performed on patients suspected of a PMD [31,37,114,115]. This analyzer concomitantly measures live oxygen consumption rate (OCR) and proton flux as extracellular acidification rate (ECAR) to quantify mitochondrial respiration via the OXPHOS pathway and glycolysis, respectively. More specifically, OCR is a key functional indicator of the mitochondrial ATP-linked respiration via the OXPHOS pathway. The patient's OXPHOS metabolism is interrogated using the mitochondrial stress test for real-time quantification of key mitochondrial bioenergetic parameters, such the basal respiration, the spare respiratory capacity, and the proton leak. While the basal respiration indicates the mitochondrial bioenergetic demand of patient-derived cells under basal conditions, the spare respiratory capacity is bioenergetic parameter for the reserve energy of patient-derived cells required to respond to a sudden high energy demand or to cellular or environmental stressors to advert an energy deficit. To validate the functional results from the Agilent Seahorse Mito Stress Test, the basal rate of ATP production from the two major pathways, OXPHOS and glycolysis, is quantified using the Agilent Seahorse Real-Time ATP rate assay [116]. Finally, the metabolic plasticity of patient-derived cells from OXPHOS to glycolysis is assessed using the Agilent Seahorse Glycolytic Rate Assay to quantify the glycolytic proton efflux from basal glycolysis in the absence and presence of OXPHOS inhibitors [116].

Thus, this technology allows to quantify in real-time the OXPHOS energy deficit and the deficient metabolic adaptability using patient-derived dermal fibroblasts and to confirm the patient's suspected mitochondrial etiology on a quick time scale and without any invasive procedures. Collectively, these real-time mitochondrial bioenergetic assays are well suited for unbiased high-throughput screening for repurposing drug, a strategy that has gained recent traction given the numerous challenges faced by patients, including a long-lasting diagnostic process, the absence of treatment, frequent hospitalizations over a patient's lifespan, the poor quality of life, significant disabilities, and premature death.

Multi-omics technologies and implications for patient-centered therapeutic modalities

Precision medicine holds promise for optimizing therapeutic outcomes in primary mitochondrial disorders, as currently patients are only offered a one-size-fits-all palliative approach. Analyzing the individual metabolism of each patient coupled with deciphering their pathophysiological mechanisms allows for personalized treatment approaches. An integrative multi-omics approach enables the discovery of patient-centered therapeutic targets based on their phenotypic, genetic, and functional profile. By integrating different single-omics entities, this approach has the potential to understand the pathogenesis of these PMDs with the objective to stratify patients based on a comprehensive analysis of their phenotypic, genomic, transcriptomic, proteomic, and metabolic signatures. However, this multi-omics approach toward personalized medicine remains at its infancy given that most studies are limited to two levels of omics [117,118].

The NGS technologies have revolutionized genetic diagnosis via unbiased approaches of whole exome sequencing, whole genome sequencing, and comprehensive sequencing of all the copies of the mitochondrial genome and have accelerated the discovery of novel disease genes to reduce the diagnostic odyssey for patients with PMDs [119]. However, this NGS-based molecular diagnostics in itself is insufficient to provide novel insight into the tissue-specific pathogenic mechanisms of PMDs to implement personalized patient care. Recent advances in transcriptomics have improved the molecular diagnostics of patients, especially those with non-canonical clinical phenotypes, by providing functional evidence of the pathogenicity of genome-wide variants of unknown significance [120]. RNA-sequencing (RNA-seq) provides an unbiased platform to expose aberrant transcripts as a result of non-coding variants located within the intronic sequence. In contrast with DNA-based investigations such as WES or WGS, RNA-seq allows for relative quantification of altered transcripts levels and the tissue-specific expression of transcripts due to non-coding variants mapping in regulatory regions dictating mRNA tissue-specific expression [121]. Thus, RNA-seq holds promises to assist in the molecular diagnosis of PMDs and to improve our understanding of pathogenic molecular mechanisms dictating the highly heterogeneous phenotypic manifestations, two key priorities in mitochondrial medicine [[122], [123], [124]]. However, the full potential of RNA-seq in mitochondrial is stunted by the limited sensitivity of predictive tools [125] and inadequate knowledge on computational approaches essential to decrease the inter-analysis barrier and to ensure clinical reproducibility [126,127]. These outstanding limitations might explain the lack of increased diagnostic yield from centers performing RNA-Seq despite the very promising potential of RNA-seq to detect novel variants. In most cases, RNA-seq has increased the diagnostic power by only 10 ​% and consequently has increased number of variants of unknown significance (VUS) with no novel methods to prioritize them [128,130].

Quantitative proteomics is instrumental for providing insight into the pathophysiological mechanisms of PMDS and adapting clinical care in a patient-centric modality. Our proof-of-concept study on a MELAS patient with recurrent stroke-like episode due to the pathogenic variant, m.14453G ​> ​A, illustrates the clinical relevance of global proteomics by exposing decreased levels of arginine due to downregulation of the key nodal enzyme, arginosuccinate synthase 1, for the arginine biosynthesis pathway [118]. Thus, in conjunction with WES and WGS, proteomics is not only be a valuable diagnostic tool to WES and WGS, but also provides a functional glimpse of pathogenic nuclear and mitochondrial variants to tailor the patient's cocktail of nutraceuticals for curtailing canonical symptoms and to ultimately develop novel patient-tailored mitochondrial drugs.

Metabolomics provides another useful systems biology-based platform to decipher emerging pathophysiological mechanisms of PMDs and to identify potential therapeutic targets for personalized mitochondrial medicine. In light of the low sensitivity and specificity of the current metabolites, such as lactate, amino acids, and organic acids, there is an urgent need to discover novel diagnostic biomarkers with more favorable specificity and sensitivity, an objective attainable with the advent of mass spectroscopy technology. Mass spectroscopy-based metabolomics is the most clinically relevant platform to characterize a patient's metabolic fingerprint on the basis that metabolites are the downstream results of endogenous genetic and protein regulations as well as exogenous influences, making metabolomics the closet to phenomics when compared to genomics, transcriptomics and proteomics [129,130]. Our proof-of-concept global metabolomics study on a MELAS patient harboring the rare MELAS variant, m.14453G ​> ​A, illustrates the clinically relevant link between disease phenotypes and metabolomics by revealing a deficit in the metabolite arginine validating the patient's recurrent stroke-like episode as a result of decreased nitric oxide availability [118].

In conclusion, an integrated systems biology-based platform currently offers the best approach to personalized medicine for patients with a specific PMD by initially identifying subgroups of patients via longitudinal phenomics, stratifying these patients by genomics and transcriptomics using the NGS technology, a partitioning that is further refined by linking proteomics with metabolomics. This arsenal of triangulated systems-based approaches complemented with patient-derived functional mitochondrial investigations has the potential to augment the predictive values of which therapeutical modality and drug fit best to specific subsets of patients and which ones are ineffective and/or unsafe.

Bridging the translational gap: the long road to patient-centered clinical trials for PMDs

Advances in understanding mitochondrial biology have led to the development of supportive therapies using a cocktail of nutraceuticals with the chief objectives to enhance the mitochondrial energy metabolism and to curtail the severity and progression of the clinical manifestations (Fig. 3) [131,132]. However, randomized double-blinded clinical trials have shown little therapeutic benefit to these nutraceuticals [22,133]. More specifically, these published clinical trials are small and mostly open-label. While coenzyme Q10 (CoQ10) and carnitine are the most frequently administered nutraceuticals, patients with PMDs often take additional nutraceuticals acting as OXPHOS enhancers, such as riboflavin, thiamine, biotin, niacin, and creatine, in conjunction with several antioxidants, such as lipoic acid, glutathione, vitamin A, and vitamin E [48,131].

Fig. 3.

Various nutraceutical-based interventions and drug candidates tested in clinical trials to stimulate mitochondrial biogenesis in patients with primary mitochondrial disorders. This schematic illustration only summarizes the most relevant types of interventions. A mitochondrion is shown in blue, while the nucleus is shown in purple. The cytoplasm is indicated in orange. The electron chain transfer is illustrated in green with the five multisubunit complexes of the OXPHOS pathway (CI, CII, CIII, CIV, and CV). The supplement L-carnitine is represented by a yellow circle with its role in the translocation of long chain fatty acids into the mitochondrial matrix to promote mitochondrial bioenergetics. The three electron carriers, coenzyme Q10 (CoQ10), idebenone and EPI-743 are illustrated by colored hexagons to promote electron transport between Complex I and Complex III and between Complex II and Complex III for ATP synthesis by Complex V (ATP synthase). The taurine supplement is indicated by a yellow circle and its action on the modification of the mitochondrial tRNA^Leu(UUR) is illustrated in the case of MELAS due to the pathogenic mitochondrial variant shown by an x letter in mutant mtDNA in red. Wild type (WT) mtDNA is shown in black. Patients with MELAS are prescribed oral or IV arginine (yellow circle) to produce nitric oxide (NO) via the enzyme nitric oxide synthase (NOS) for increasing vasodilation of small cerebral blood vessels to diminish the severity and frequency of stroke-like episode (SLE). Citrulline (green circle) is another supplement currently tested in a clinical trial on MELAS patients based on its conversion to arginine via the intermediate argininosuccinate (light blue circle). Omaveloxolone, currently tested in phase 2 clinical trial, is illustrated by a yellow circle and its action to boost the expression levels of the nuclear factor erythroid 2-related factor 2 (NRF-2) transcription factor by preventing its degradation is shown with an arrow and a positive sign. The drug candidate REN001, indicated by a blue circle, acts as an agonist to activate the peroxisomal proliferator activated receptor delta (PPARδ), resulting in increased expression of the peroxisomal proliferator activated receptor-gamma coactivator-1α (PGC-1α) to turn on the transcriptional program of mitochondrial biogenesis and bioenergetics.

This nutraceutical cocktail is often tailored for specific PMDs. MELAS patients exhibit a nitric oxide (NO) deficiency that provokes vasoconstriction leading to ischemia and hypoxemia and subsequently to stroke-like episode (SLE), a MELAS cardinal feature [134]. During the acute phase of SLE, MELAS patients exhibit low levels of L-arginine and consequently of NO, given that arginine is converted into NO by nitric oxide synthase. To restore normal NO levels, MELAS patients are prescribed oral L-arginine supplementation to decrease the frequency and severity of SLE, as well as to enhance the cerebral microvasculature dynamics (Fig. 3) [135]. During the acute phase of SLE, MELAS patients are administered intravenous arginine to induce endothelium-dependent vascular relaxation and to rapidly curtail this clinical symptom and its associated neural injury [136]. An ongoing clinical trial (NCT03952234) on MELAS patients assesses whether oral administration of citrulline could be a therapeutic modality by increasing NO availability to relax the vascular smooth muscles of small blood vessels resulting in their sustained patency [133,137,138]. Finally, the intravenous arginine therapy ameliorates acute metabolic strokes in patients diagnosed with other PMDs caused by pathogenic mitochondrial variants other than MELAS variant, mtDNA deletion, or pathogenic nuclear variants mapping in the FBXL4, POLG, NDUFS8 and SURF1 genes [139].

Patients with MELAS due to the pathogenic mitochondrial variant m.3443G ​> ​A, which causes a taurine modification defect in the mt-tRNA^Leu(UUR), are also prescribed oral administration of the nutraceutical taurine based on the significant decrease of SLE in patients enrolled in an open-label clinical trial (Fig. 3) [140].

Clinical trials were conducted in several countries on LHON patients with vision loss to test the efficacy of idebenone, a synthetic hydrosoluble analog of CoQ10, for reducing the impact and progression of vision loss after its onset due to impaired mitochondrial energy production in the retinal ganglion cells (Fig. 3) [141,142]. Results were encouraging but mixed in terms of stopping the worsening of visual loss [[143], [144], [145]]. However, retrospective analyses of these collective clinical trials have led to the recommendations for the first-line therapeutic treatment for non-chronic LHON patients (less than one year since the onset of visual loss) consisting of a daily dose of idebenone at 900 ​mg/day for at least one year [146]. Idebenone is not currently readily available in North America, as compared to in Europe where it was approved in 2007 by the European Medicines Agency for LHON and consequently received its designation as an orphan drug.

The experimental drug alpha-tocotrienol quinone (EPI-743), a potent anti-oxidant, is also used as another therapeutic option for LHON patients with recent visual loss occurring less than 4 months [147,148]. EPI-743 easily crosses the blood-brain barrier to act as an anti-oxidant in the central nervous system by targeting oxidoreductase enzymes and replenishing glutathione pools [149]. Several completed clinical trial tested EPI-743 as a therapeutic option for Leigh Syndrome (NCT02352896; NCT01721733), Pearson Syndrome (NCT02104336), or other PMDs (NCT01642056; NCT01370447) [150]. Currently, there is an ongoing phase 3 clinical trial (NCT05218655) to test vatiquinone (PTC-743), previously known as EPI-743, on patients genetically diagnosed with a PMD including Leigh Syndrome, Alpers Syndrome, MELAS, MERFF, and pontocerebellar hypoplasia type 6 (PCH6).

During the last decade, the landscape of clinical trials to test emerging small molecule drugs in patients genetically diagnosed with a PMD has been expanding. This positive outlook has been facilitated by the adoption of the 1983 Orphan Drug Act (ODA) that provides regulatory and financial incentives for augmenting investment on orphan drug discovery for rare diseases in the United States (https://www.fda.gov/industry/designating-orphan-product-drugs-and-biological-products/orphan-drug-act-relevant-excerpts). These clinical trials have non-subjective patient-centered primary outcomes with the objective to give rise to evidence-based guidelines. Since several excellent reviews on the pre-clinical mechanisms targeted by emerging pharmacological modalities have been published [[150], [151], [152], [153]], only the therapeutic approaches aimed at increasing the pool of healthy mitochondria to increase mitochondrial ATP synthesis by enhancing mitochondrial biogenesis or targeting mitophagy are emphasized.

A handful of clinical trials have investigated therapeutic candidates for enhancing mitochondrial biogenesis in patients with PMDs since pharmacological boosting of mitochondrial biogenesis is one of the main therapeutic avenues to overcome maladaptive mitochondrial biogenesis by manipulating the Peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1α) 5. Omaveloxolone, known to promote mitochondrial biogenesis by increasing the activity of the transcription factor NRF2 via inhibition of KEAP1 (Kelch-like ECH-associated protein 1), was tested in a phase 2 clinical trial in which 53 patients with mitochondrial myopathy were enrolled (NCT02255422). No clinical benefit was reported given that the primary outcome measure was not met [154]. The mitochondrial biogenesis enhancer REN001, which functions as a specific agonist of the peroxisome proliferator-activated receptor delta (PPARδ), was tested in a phase 1b clinical trial (NCT03862846) to evaluate the safety and tolerability in 23 patients with mitochondrial myopathy. Although the clinical trial is completed, no results are reported.

Targeting mitophagy via the mammalian target of rapamycin (mTOR) pathway provides a pharmacological avenue to elicit elimination of diseased mitochondria in patients with a PMD Based on the encouraging results of rapamycin in the murine model Ndufs4^−/− for LS [153], the rapamycin homolog, everolimus, was tested in a patient with LS who showed sustained benefit and in a patient with MELAS who failed to respond [154]. An open-label phase 2a clinical trial (NCT03747328) to test Sirolimus (ABI-009), a derivative of rapamycin, in patients genetically diagnosed with LS or LS-like was stopped due to withdrawal of the corresponding Investigational New Drug application sponsored by Aadi Bioscience, Inc (https://clinicaltrials.gov/study/NCT03747328).

Thus far, none of the currently tested drugs are tailored to a specific genetic defect, thereby limiting their efficacy. There is an urgent need to shrink the translational gap by deciphering the pathophysiological mechanisms specific to a PMD and its target(s) for rescuing the mitochondrial phenotype in order to ameliorate mitochondrial drug development and to customize therapeutic modalities.

Conclusive remarks and future perspective

Mitochondrial medicine has evolved from a research-based endeavor using the handful of murine models for a few PMDs to bridging translational research using patient-derived cellular paradigms with clinical research for improving genetic diagnosis, prognosis and mitochondrial drug development toward the design of innovative patient-centered therapeutic modalities. Advances in genomics, functional assays, biomarker discovery, and non-invasive diagnostics are promising avenues for overcoming the collective challenges associated with these complex PMDs and their maladaptive mechanisms of mitochondrial dysfunction. As our understanding of mitochondrial biology deepens and technology continues to progress, we can look forward to more precise and effective treatments for PMDs, ultimately improving the quality of life for affected individuals. This global approach in mitochondrial medicine is complemented by recent initiatives to develop non-invasive diagnostic tools, such as blood-based assays or imaging techniques, with the chief objective replace the current invasive procedures for improving patient comfort and compliance. Furthermore, progress in proteomics, metabolomics, transcriptomics, and algorithms for multi-omics integration have led to the identification of specific proteins, metabolites and transcripts that could serve as biomarkers for PMDs and for stratifying cohorts of patients harboring the same disease-causing nuclear or mitochondrial variants, but with different clinical manifestations and severity. Integrating these biomarkers into clinical practice holds promise for more accurate and timely diagnosis, as well as for monitoring disease progression and patient-centered treatment responses.

Author contribution

All three authors (AG, MU, and AC) contributed to the conception and design of the article. AC drafted the article. MU conceived and designed all the figures included in the article. AG conceived and wrote Table 1 of the article. AG, MU, and AC revised critically for important intellectual concept AND approved the final version to be published.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Anne Chiaramello (Corresponding author) and Andrea Gropman (co-author) reports financial support was provided by National Institutes of Health, National Center for Advancing Translational Sciences. Anne Chiaramello reports financial support was provided by U.S. Department of Defense.