Cytoplasmic hybrid (cybrid) cell lines as a practical model for mitochondriopathies

Abstract

Cytoplasmic hybrid (cybrid) cell lines can incorporate human subject mitochondria and perpetuate its mitochondrial DNA (mtDNA)-encoded components. Since the nuclear background of different cybrid lines can be kept constant, this technique allows investigators to study the influence of mtDNA on cell function. Prior use of cybrids has elucidated the contribution of mtDNA to a variety of biochemical parameters, including electron transport chain activities, bioenergetic fluxes, and free radical production. While the interpretation of data generated from cybrid cell lines has technical limitations, cybrids have contributed valuable insight into the relationship between mtDNA and phenotype alterations. This review discusses the creation of the cybrid technique and subsequent data obtained from cybrid applications.

Highlights

• •The cytoplasmic hybrid (cybrid) model can be used to determine mitochondrial DNA (mtDNA) contributions to phenotypic alterations.
• •Cybrids are used to study mitochondriopathies such as Parkinson’s disease and Alzheimer’s disease.
• •mtDNA heteroplasmy threshold and nuclear DNA-mtDNA compatibility can be determined using cybrid models.

Graphical Abstract

Mitochondria are an essential cellular reticulum driving the molecular reduction of oxygen into the energy molecule ATP. The presence of a separate mitochondrial genome was discovered in 1963 [1,2]. The mitochondrial genome is comprised of circular double-stranded DNA containing 16,569 base pairs. While not all components of the electron transport chain (ETC) are encoded by the mitochondrial genome, its integrity is vastly important for ETC function. The mitochondrial genome encodes 11 components of the ETC and two subunits of ATP synthase (Fig. 1). In addition, the mitochondrial genome contains 24 tRNA/rRNA genes, which are required for translation of mitochondrial proteins (Fig. 1). The importance of mtDNA is highlighted by findings of mtDNA mutations and mitochondrial dysfunction in diseases associated with aging [3–5].

Fig. 1.

The mitochondrial genome. This figure depicts the circular mitochondrial genome. Structural genes are shown in yellow. rRNA genes are shown in blue, while tRNA genes are shown in black. The regulatory region, D-loop, is shown in red. CO = cytochrome oxidase, ND = NADH dehydrogenase.

The cybrid technique

Rho zero cells

To facilitate cybrid generation, cells devoid of mtDNA are commonly utilized. These cells are termed ρ0 cell lines because prior to identification of the mitochondrial genome the DNA found in the cytoplasm of cells was termed “ρ-DNA” [6]. The ability to deplete mtDNA was first discovered in yeast models where natural mtDNA depletion occurred under conditions in which glycolysis was favored over mitochondrial respiration. This led to artificial techniques to deplete cells of mtDNA, the oldest of which is the use of the DNA-intercalating agent ethidium bromide (EtBr). EtBr, a positively charged aromatic compound, is attracted to negatively charged DNA molecules located within negatively charged mitochondrial matrices. Intercalation of EtBr into DNA results in the failure of DNA replication by DNA polymerase. The use of EtBr in yeast cells culminated in partial and complete mtDNA depletion [7–9].

When extended to vertebrate cells, the development of mtDNA depletion protocols proved more difficult. In one reported early attempt, while treatment with EtBr did result in almost complete depletion of mtDNA from the VA₂B human cell line, the high concentrations of EtBr that were used were toxic. Additionally, immediately following removal of EtBr, mtDNA was replenished [10].

The first successful vertebrate ρ0 cell lines were made in chicken embryo fibroblast cells using EtBr. The investigators determined the resulting cells were auxotrophic for the pyrimidine nucleoside uridine [11]. The basis for this can be explained by the function of dihydroorotate dehydrogenase. This mitochondrial enzyme is required for the synthesis of pyrimidines and relies upon the ETC to function [12]. Therefore, depletion of mtDNA and subsequent ETC failure will disrupt pyrimidine synthesis. Supplementing mtDNA- depleted cells with uridine bypasses the ETC-dependent, dihydroorotate dehydrogenase-catalyzed step in the pyrimidine synthesis pathway. An avian ρ0 cell line was subsequently generated using the same protocol, reducing mtDNA copy number from 300 per cell to undetectable levels [13]. It is important to emphasize, though, that ρ0 cells retain mitochondria [14]. These mitochondria are not respiration-competent, but they do retain a membrane potential, presumably by operating their ATP synthase in reverse. Although their structure is altered they undergo proliferation and receive nuclear-encoded mitochondrial proteins. These respiration-incompetent mitochondria have been colloquially referred to by some authors as “mitoids” [15].

In 1989, the human osteosarcoma 143B cell line was depleted of mtDNA. The ρ0 clones (143B101 and 143B106) were auxotrophic for both uridine and pyruvate [16]. The necessity for pyruvate supplementation was hypothesized to be required for the maintenance of cellular redox status [17]. Cells devoid of mtDNA rely upon glycolysis for ATP production. Under physiological conditions, glycolysis increases NADH levels, while mitochondrial respiration produces NAD⁺. Therefore, the absence of mitochondrial respiration significantly increases the NADH: NAD⁺ ratio, interfering with glycolytic capacity. Supplementation with pyruvate leads to the formation of lactate, which generates NAD⁺, thus allowing glycolytic function and ATP production. Additional cell lines have been depleted of mtDNA over the course of time. Table 1 provides a list of currently reported ρ0 cell lines.

Table 1.

Reported ρ0 cell lines.

ρ0 cell line	Reference
143B osteosarcoma	[16]
HeLa cervical carcinoma	[18]
A549 lung carcinoma	[19]
Lymphoblastoid, Wal2A	[20,21]
SH-SY5Y neuroblastoma	[22]
NT2 teratocarcinoma	[23,24]
U251 glioma	[25]
Fibroblast/transformed fibroblast	[26]
LL/2-m21 mouse	[27]
T-cell lymphoblastic leukemia, molt-4	[28]

To understand the mechanisms of mtDNA depletion, it is important to understand that the depletion of mtDNA occurs in two steps. First is the cessation of mtDNA replication. This step does not eliminate existing mtDNA molecules, but terminates their ability to replicate the mtDNA. As the cells divide, the mtDNA is divided among the new cells, leading to continuous dilution of the mtDNA pool. This creates a cell population in which some cells contain mtDNA while other cells lack the mtDNA—and are now “ρ0”. The second step begins at this point, in which ρ0 cells are isolated and expanded, or mtDNA depletion continues until all mtDNA is degraded.

Further development of techniques to deplete cells of mtDNA have moved away from the use of EtBr. The mitochondrial DNA polymerase γ inhibitor, ditercalinium, or expression of a dominant negative mitochondrial DNA polymerase γ construct have been used successfully to create ρ0 cell lines [29,30]. Exposure to dideoxynucleoside analogues, which interfere with mtDNA replication, leads to mtDNA depletion myopathy and is another technique for creating ρ0 cell lines [31,32]. Rhodamine 6-G, first noted for its ability to interfere with mitochondrial ETC function, is an additional tool for the generation ρ0 cell lines [33–35]. More recent techniques to deplete mtDNA have used the expression of mitochondrial targeted EcoRI, which is the focus of a biotech company, Rho Zero Technologies [36,37].

The generation of ρ0 cell lines requires techniques to verify depletion of the mtDNA. Several applications have been reported describing how to document ρ0 status. Techniques include southern blotting, PCR of mtDNA, competitive PCR to ratio mtDNA and nuclear pseudogenes, measurement of oxygen consumption and cytochrome c oxidase activity, pico green staining, and uridine/pyruvate auxotrophy [18,19,21–26]. The majority of these techniques depend upon the demonstration of a negative result. Most of these techniques, therefore, are limited by the sensitivity of the mtDNA detection protocol. Testing for uridine/pyruvate auxotrophy, however, is a practical approach which does not rely upon a negative result. A more recently developed technique, competitive PCR, is more labor intensive but also does not rely upon demonstration of a negative result. The competitive PCR approach provides a means to statistically analyze the probability of mtDNA depletion and that ρ0 status has been achieved [25].

mtDNA Transfer

During the development of mtDNA depletion techniques, methods for transferring mtDNA into cells were simultaneously being established. The first attempt, in 1972, led to the creation of a heterokaryon hybrid cell, which used the Sendai virus to fuse cells together [38]. Two years later, a separate technique was carried out using mixtures of nucleated cells and cytoplasts (non-nucleated cells) [39]. Bunn et al. called the resulting cells “cybrid” cells, distinguishing them from the term “hybrid”, which implies the mixture of two nucleated cells.

The cybrid technique was first utilized in 1975, to determine if chloramphenicol resistance was mediated by the cytoplasm in a human cell line [40]. Cytoplasts were generated from a chloramphenicol resistant HeLa cell line fused with a chloramphenicol susceptible HeLa cell line that was also thymidine kinase negative (TK⁻). The difference in TK content allowed for a selection process to eliminate intact chloramphenicol resistant cells which may have failed the enucleation procedure. This is important because both HeLa cell lines perpetuate, and contamination with intact chloramphenicol-resistant cells could lead to confounded results. It was ultimately concluded that the cybrid cells were chloramphenicol resistant, and that this characteristic was dependent upon the cytoplasm of the cell. Fig. 2 depicts this type of cybrid generation strategy, in which mitochondria are transferred from one nuclear partner to another, even though the recipient nucleated cell is not a ρ0 cell.

Fig. 2.

Cybrid generation techniques. Cybrids are produced by combining cytoplasm from nucleated cells with non-nucleated cells or cytoplasts. The nucleated cell can be unaltered, or could have undergone depletion of endogenous mtDNA before cytoplasmic mixing. In either case, the goal is to populate the nucleated cell with mtDNA from the non-nucleated cell.

A separate cybrid technique—first reported in 1989—fuses enucleated cytoplasts with ρ0 cells. Cytoplasmic transfer to the ρ0 cell line was mediated through the use of polyethylene glycol, as well as direct injection of cytoplast mitochondria [16]. Repopulation of the ρ0 cell line with mtDNA was depicted through loss of uridine/pyruvate auxotrophy and restoration of oxidative phosphorylation biochemical measurements. Fig. 2 shows a schematic for this cybrid generation strategy. To further simplify the cybrid technique, Chomyn et al. [41] used a procedure in which platelets were employed as cytoplasts. Platelets do not contain a nucleus, therefore pre-fusion enucleation of the cytoplasmic donor cell is not required.

Two separate studies subsequently reported that synaptosomes from brain tissue could also serve as cytoplasts for mtDNA fusion [42,43]. However, the use of synaptosomes has overall proved challenging, because the efficiency of mtDNA transfer is low when fresh brain extracts are used from mice and dismal when post-mortem human brain tissue are used [44].

The use of cybrids to study mitochondrial genetics and mtDNA-nuclear DNA compatibility

Early cybrid studies used fibroblast mitochondria which contained mtDNA that harbored known pathogenic mutations. The major goal of these initial studies was to understand the pathological mechanisms underlying diseases associated with mtDNA mutations. To better understand the power of cybrids for this application, it is worth reviewing some basic mitochondrial genetic tenets.

Cells contain numerous copies of mtDNA. Homoplasmy defines a state in which the mtDNA sequence of different mtDNA molecules within a cell is always the same. Homoplasmy is the mitochondrial equivalent of nuclear homozygosity. However, it is reasonably common for mtDNA molecules within a cell, tissue, or organism to exhibit sequence heterogeneity. This state is called heteroplasmy. Heteroplasmy can be considered to loosely correspond to nuclear heterozygosity, but in reality it is far more complex. Because mtDNA is not a binary system like nuclear DNA, different ratios of wild-type and mutant mtDNA molecules can reside within a single mitochondrion or a single cell. Therefore, a heteroplasmic mutation can be present in varying degrees, and in either low or high abundance.

Threshold refers to the mtDNA mutational load that is required within a cell to cause a biochemical or phenotypic consequence. It is reasonable to hypothesize that the more pathogenic the mtDNA mutation, the lower abundance is necessary to reach a threshold that changes mitochondrial and, therefore, cell physiology. A schematic depicting the use of cybrids to study mtDNA heteroplasmy is shown in Fig. 3A.

Fig. 3.

Applications of the cybrid model. Cybrids can be used to (A) evaluate heteroplasmy-threshold corrections. It is unknown whether heteroplasmic mtDNA changes are found at the cellular level or the level of each mitochondrion. Such that, a single mitochondria could contain both wild-type and mutated mtDNA, or a cell could contain a population of mitochondria which have wild-type mtDNA and a population of mitochondria which have mutant mtDNA. Beyond that, a particular cell could contain all wild-type mtDNA, while a separate cell contains all mutant mtDNA. (B) Test mtDNA–nuclear DNA compatibility and (C) screen for mtDNA-determined functional differences between cell lines containing mtDNA from different sources.

The A3243G mtDNA mutation associated with the mitochondrial encephalopathy, lactic acidosis, and stroke-like episodes syndrome (MELAS) was studied using the cybrid model. mtDNA nucleotide 3243 is located within one of the two mtDNA genes that encodes a leucine tRNA (tRNA Leu^UUR). This mutation does not produce an aberrant protein because the gene product is synthetic and not structural in nature. MELAS cybrids were made using osteosarcoma ρ0 cell lines, and found that a high mutational burden of 85% was required before phenotypic consequences were observed [45–48]. The mechanism leading to biochemical changes with this mtDNA mutation is related to translation deficiency of mtDNA-encoded structural proteins [45,49]. Following these initial studies, it became apparent that the nuclear background of the ρ0 cell used influenced the mitochondrial genotype–phenotype relationship. In the osteosarcoma MELAS cybrid model, a 90% mutational burden resulted in a 30% decrease in cytochrome oxidase activity, while an A549 lung carcinoma MELAS cybrid model required only a 55% mutational burden to observe a 50% reduction in cytochrome oxidase activity [50]. Both cybrid models required substantial mtDNA mutation load to alter biochemical endpoints compared to those required in human muscle [51,52]. Finally, nuclear differences in mutational drift were observed over time. For example, one nuclear background led to drift away from the mutant mtDNA, while a separate nuclear background led to drift toward the mutant mtDNA [50,53,54].

Other cybrid studies have examined mtDNA transfer from people with Leber’s hereditary optic neuropathy (LHON). LHON is associated with mutations in mtDNA structural genes, unlike the synthetic gene mutations observed in MELAS. No single mutation leads to the LHON phenotype, as several distinct mutations have been reported. All mutations that lead to the LHON phenotype are found in structural genes, and mutations within the NADH dehydrogenase (ND) genes, which encode for Complex I subunits, are most common. The most frequently reported mutation is G11778A, which resides in the ND4 gene. LHON cybrid data suggests that biochemical consequences depend upon the mutation and nuclear background of the ρ0 cell line that is utilized. For example, LHON mutations that show no Complex I defects in one nuclear background can show Complex I defects in a separate nuclear background. The only consistent parameters among various LHON cybrids are oxygen consumption deficits and a reduction in Complex I-dependent ATP synthesis. All phenotypic differences required a high mutational load, which is consistent with what is observed in the clinical setting, where people afflicted with LHON are homoplasmic or the mtDNA mutation is present at a high copy number [21,55–61].

LHON cybrid studies using an osteosarcoma nuclear background more recently reported that mitochondrial haplogroup status is also an important determining factor for the pathogenic potential of different LHON mutations [62–64]. Mitochondrial haplogroups are determined by ancestral hereditary. Mitochondrial haplogroup J and Uk are found at higher levels in persons afflicted with LHON, however haplogroup H is under represented in the incidence of LHON. Cybrids from different human mitochondrial haplogroups were generated using platelets fused on the osteosarcoma nuclear background. Haplogroup Uk and J have decreased mtDNA, mitochondrial RNA and mitochondrial protein translation products compared to haplogroup H. In addition, haplogroups Uk and J have lower oxygen consumption, ATP production, and mitochondrial membrane potential when compared to haplogroup H. In particular, LHON osteosarcoma cybrids had accumulations of low-molecular weight sub-complexes, particularly from Complex I structural proteins. However, polymorphisms in mtDNA, or specific mitochondrial haplogroups, influenced the assembly rates and stability of Complex I, III, and IV [62]. Overall, cybrid studies have contributed and will continue to be important to the understanding of how haplogroups influence multifactorial diseases, such as LHON.

Further LHON cybrid studies used the teratocarcinoma NT2 cell line as a nuclear background donor [35]. From a modeling perspective, the use of this ρ0 cell line to generate LHON cybrids may be argued to be more closely related to the disease itself, since NT2 cells display “neuronal-like” features and can be differentiated into neurons using retinoic acid [65]. It was found that undifferentiated LHON NT2 cybrids did not have altered biochemical parameters, but upon differentiation parametric differences could be observed. The phenotypic differences observed in the LHON NT2 cybrids were diminished efficiency in the differentiation protocol and production of higher levels of reactive oxygen species (ROS) when compared to control NT2 cybrids. An additional study using LHON NT2 cybrids depicted these cells were more sensitive to H₂O₂, displayed no change in superoxide dismutase (SOD) activity, but had decreased glutathione (GSH) levels [66]. However, these data again depict the importance of nuclear background, as similar LHON cybrids generated on an osteosarcoma background showed increased ROS production, decreased SOD activity, but no reduction in GSH levels [67].

Other diseases with known mtDNA mutations have been studied using the cybrid technique. These include myoclonic epilepsy and ragged red fiber disease (MERRF), Leigh’s disease due to mtDNA mutation/neuropathy, ataxia, retinitis pigmentosa (NARP), and Kearns–Sayre Syndrome [41,68–73]. Kearns–Sayre Syndrome involves deletions of mtDNA, and cybrid work has resulted in notable findings [72]. One important finding is that large mtDNA deletions result in mitochondrial respiration failure, but mtDNA duplications do not. In addition, osteosarcoma cybrid cell lines, it appears, prefer to maintain a stable mtDNA total amount, as opposed to maintaining a constant mtDNA copy number. When osteosarcoma cells are predominately populated with mtDNA that contains large deletions, mtDNA copy number increases to compensate, which ultimately results in an amount of total mtDNA that is equivalent, in mass to that of cells without large deletions.

Studies examining mtDNA-nuclear DNA compatibility have also utilized the cybrid approach. These studies generated xenomitochondrial cell lines, where cytoplasts from primate fibroblasts were fused with a human osteosarcoma ρ0 cell line [74,75]. This mtDNA transfer gave rise to viable cybrid cell lines when the transferred mtDNA was derived from common chimpanzee, pigmy chimpanzee or gorilla fibroblasts. Orangutan, Old-world monkey, New-world monkey, and Lemur mtDNA were not compatible with the human osteosarcoma nuclear background, as no viable cybrid cells resulted. Cybrids expressing common chimpanzee mtDNA displayed a 20% lower oxygen consumption rate, while cybrids expressing pigmy chimpanzee mtDNA had 34% reduced oxygen consumption, and cybrids expressing gorilla mtDNA had 27% less oxygen consumption. No deficits were found for any of the xenomitochondrial cybrids with respect to Complex II, III, or IV activities. However, cybrids expressing common chimpanzee mtDNA showed a 41% reduction in Complex I activity, while cybrids expressing pigmy chimpanzee mtDNA had a 43% reduction in Complex I activity, and cybrids expressing gorilla mtDNA had 45% lower complex I activity. These are findings of interest because there are 159 amino acid differences between human and common chimpanzee mtDNA coding, 155 amino acid differences between human and pigmy chimpanzee mtDNA coding, 195 amino acid differences between human and gorilla mtDNA coding, and 411 amino acid differences between human and orangutan mtDNA coding. The conclusions drawn from this study are that primates which evolved longer than approx. 18 million years ago cannot function to replace human mtDNA. Fig. 3B shows a schematic of this cybrid application.

The use of cybrids as a Model for Mitochondriopathy-Associated sporadic diseases

Cybrid models have also been applied to test mitochondrial and cell physiology alterations arising from un-sequenced mtDNA. Here, mtDNA is transferred from an individual with a particular condition or disease, as shown in Fig. 3C. The resulting cybrid cell lines are expanded and used to measure biochemical and molecular endpoints, and compared to cybrid cell lines derived from control patients. Due to the nuclear background and environment (i.e. cell culture parameters) being constant, the differences observed in measured indices are assumed to reflect differences in mtDNA [76]. However, several criticisms of this cybrid application have arisen.

Most controversy surrounding this cybrid application is due to its assumption that the functional differences in cybrid cell lines are due to differences in mtDNA. Because the cybrid technique does not solely transfer isolated mtDNA, this assumption of differences being solely due to mtDNA is a leap of faith in some aspects. At a minimum, whole mitochondrial transfer occurs from donor platelets or cytoplasts. However, during the time it takes to select for cells repopulated with mtDNA, the cybrids pass through multiple cell division cycles, during which any possible non-perpetuating materials that were transferred are repeatedly diluted. For a case in which a doubling time of 24 h is expected, several weeks of cybrid selection and expansion would result in over a billion-fold dilution of any non-perpetuating material. Therefore, theoretically the only transferred cytoplast or platelet component that is self-perpetuating is mtDNA.

Another common criticism of this cybrid application is verification of results. Presumably, if a goal of using cybrid models is to implicate the presence of mtDNA differences in structural, synthetic, or regulatory genes (such as the D-loop), then actual nucleotide changes should be demonstrable. While this notion is conceptually correct, it is important to note that correlating specific nucleotide deviations to specific phenotype changes is challenging because mtDNA sequences between individuals tend to vary substantially. The question, therefore, is often not whether a particular mtDNA donor has sequence variations, but whether deviations that can be detected are functionally relevant. This differs from studies of Mendelian diseases, in which the functional relevance of a particular DNA sequence is generally resolved by tracking the suspect variation through family pedigrees; addressing whether the sequence variation is exclusive to a particular phenotype is relatively straightforward. However, the mitochondrial genome is not inherited through Mendelian patterns. mtDNA is normally inherited maternally, although rare cases of mutant mtDNA inheritance via sperm have been reported [77]. While some mtDNA-associated diseases do show recognizable maternal inheritance patterns [78], it is increasingly suspected that a majority of the time mtDNA influences disease risk or development in a primarily sporadic or pseudo-sporadic fashion [79,80]. Overall, this limits the ability to associate mtDNA sequence variations with a particular disease state.

When it comes to establishing mtDNA sequence-phenotype associations, though, the greatest limiting factor may turn out to be mtDNA heteroplasmy. The traditional Sanger-based sequencing approach is not designed to detect low abundance heteroplasmic mutations; if the mutational load is less than 20–30%, the ability to detect the mtDNA change is unreliable [81,82]. Further, recent data indicate mtDNA contains large numbers of low percentage, “microheteroplasmic” sequence variations [3,83–86]. It has been postulated that these microheteroplasmic variations are present at 1–2% abundance [87]. Microheteroplasmic variations within tissue also exist, and the presence of “compound microheteroplasmy”, once postulated [88], has since been verified using next generation sequencing [86].

In addition to questions over the functional significance of mtDNA sequence deviations, the origin of especially heteroplasmic sequencing deviations may be difficult to prove. mtDNA sequence mutations that are anatomically limited within an individual may arise from inherited mutations via mitotic segregation (the mtDNA correlate of mosaicism), or represent somatically acquired mutations. Therefore, we will consider cybrid studies that depict biochemical differences between mtDNA donors as simply supporting the notion that functionally relevant mtDNA differences exist between the mtDNA donors, without concluding whether those differences are inherited or acquired.

Cybrid models have been used to screen for mtDNA aberrancy in several neurodegenerative diseases, particularly Parkinson’s disease (PD) and Alzheimer’s disease (AD). When the first cybrid studies of PD and AD were performed, it was already known that people afflicted with these diseases had specific ETC deficits [89,90]. Many argue that mitochondrial defects are secondary pathologies and/or irrelevant to the phenotype of these diseases. However, the fact remains that mitochondrial function within these diseases is not normal. This has led to these diseases being classified as “neurodegenerative mitochondriopathies” [91].

It has been repeatedly shown that Complex IV activity is reduced in AD subjects, and that Complex I activity is reduced in PD patients [76,92–95]. These deficits are not only found in the brains of those afflicted with AD and PD, but also systemically in platelet mitochondria. The sporadic and systemic nature of these diseases makes the cybrid model a good application for study [96]. The cybrid technique seemed a reasonable approach to address the question of whether mtDNA accounted for at least part of the lesions observed in AD and PD.

The first PD cybrid study used the human SH-SY5Y neuroblastoma (“neuron-like”) ρ0 cell line as the nuclear background. Platelets from 24 PD subjects and 28 age-matched control subjects provided mtDNA [15]. ETC Vmax measurements were taken from post-nuclear mitochondria enriched fractions from each cell line, 52 in total. Absolute Vmax levels were corrected for total protein content of mitochondrial-enriched fractions. The Complex I Vmax was decreased by 20% in the PD group compared to the control group. No difference was observed in Complex IV Vmax measurements between groups. This indicated mitochondrial repopulation was equivalent between PD and control groups. This study supports the notion that Complex I deficits observed in sporadic PD patients are at least partially determined by mtDNA.

Deficits in Complex I activity in PD cybrids compared to control cybrids were verified by four independent studies. One study used the SH-SY5Y ρ0 cell line as the nuclear background, while the other studies used the A549 lung carcinoma ρ0 cell line or the NT2 teratocarcinoma cell line, respectively [97–100]. A separate study was completed in which cybrids were generated from a large family in which multiple members had PD [78]. The pattern of PD inheritance of this family was consistent with maternal transmission. Cybrids generated from family members afflicted with PD had decreased Complex I activity when compared to cybrids generated from family members not afflicted with PD. Furthermore, family members who had strict maternal inheritance, but were younger than the age of onset for PD within this family, had lower mean Complex I activity than their paternally-descended cousins. In contrast, cybrids containing platelet mtDNA from members of the Contursi kindred did not show reduced Complex I activity [101]. In the Contursi kindred PD arises due to an autosomal dominant mutation in the α-synuclein gene. This inheritance pattern implies that the genetic etiology differs between those with nuclear gene mutations and those with the more common sporadic form or forms. mtDNA would be more likely to contribute to risk or onset in the sporadic cases.

PD cybrids have been analyzed for a variety of biochemical and molecular parameters. Increased ROS production and upregulation of antioxidant enzyme activity have been observed [15,99,102]. PD cybrids are more sensitive to H₂O₂ exposure and have decreased GSH content [103]. Bcl-2 and Bcl-XL levels are upregulated, NFkB is activated, there is enhanced p38 and JNK activity, and PARP cleavage is increased [100,104–106]. These findings are consistent with increased oxidative stress, as signaling pathways involved in both cell death and survival stress responses are enhanced in PD cybrids. Furthermore, mitochondrial calcium levels are reduced [107]. Maximum respiration capacity was lowered in PD cybrids, and enhanced proton leak was apparent [100]. Mitochondrial membrane potentials were relatively depolarized [99]. ATP content was decreased and PD cybrid cell lines released more LDH [99]. PD cybrids are more susceptible to MPP+ induced cell death [15,100]. The ultrastructure of mitochondria is altered and PD cybrids develop synuclein aggregation indicative of the Lewy bodies observed in the substantia nigra of PD subjects [78,108–110].

One study isolated clones from PD cybrids which developed Lewy Bodies, leading to the subcloning of three PD cybrid cell lines [110]. This approach yielded inconsistent results, as each subcloned PD cybrid exhibited different phenotypes. For example, in one clone mitochondrial respiration was decreased, in the second clone mitochondrial respiration was increased, while in the third clone mitochondrial respiration went unchanged. However, the data did correspond to the nucleoid content of each clone.

A study of PD cybrids generated on the HeLa cervical carcinoma ρ0 nuclear background did not find any reduction in Complex I activity, n = 10 [111]. It is important to note this study, and it is fair to assume data obtained from it are accurate. However, the results from this study have little bearing on the numerous SH-SY5Y and A549 ρ0 based PD cybrid studies, which have elucidated biochemical differences between control and PD cybrid cell lines.

In 1997, Davis et al. reported that, as a group, cybrid cell lines generated from AD subjects had a lower Complex IV activity than cybrid cell lines generated from a demographically matched group of control subjects [112]. This report was subsequently retracted due to an interpretive error involving non-cybrid data that this manuscript also contained [113]. The cybrids in this study were generated on an SH-SY5Y neuroblastoma ρ0 nuclear background. Complex IV Vmax measurements were completed using post-nuclear, mitochondrial enriched fractions from AD 45 and 20 control cybrid cell lines. The AD cybrid cell lines had a 20% reduction in Complex IV activity when compared to the control subjects. Complex I Vmax was equivalent between the two groups, suggesting that mitochondrial repopulation was comparable.

These initial AD cybrid data were recapitulated in subsequent studies. In AD 76 and 111 control cybrid lines also generated against an SH-SY5Y nuclear background, a 15% reduction in AD cybrid Complex IV activity was observed [114]. Smaller AD cybrid studies that used the SH-SY5Y nuclear background have also reported reduced Complex IV activity [107,115–117]. Two separate studies using the NT2 ρ0 cell line to generate AD cybrids also found decreased Complex IV activity. The first study compared AD 15 cybrid lines to 9 control cybrid lines, and reported a 16% decrease in Complex IV activity [23]. The second study compared AD 6 cybrid lines to 5 control cybrid lines, and reported a 22% decrease in Complex IV activity [118].

Cybrids generated from AD patients on both the SH-SY5Y and NT2 nuclear background show biochemical and molecular changes. Increased ROS production was demonstrated using fluorescence-based peroxide measurements [23,112,118]. AD cybrids generated on the NT2 nuclear background displayed increased antioxidant enzyme activities, protein carbonyl markers, and lipid peroxidation markers [23,118]. AD cybrids on both the NT2 and SH-SY5Y nuclear backgrounds showed activation of MAPK, Akt, and NFkB with increased caspase-3 activation and cytochrome c cytoplasmic levels [118–121]. These findings are consistent with pathways associated with cell death and stress responses. AD cybrids generated on the SH-SY5Y nuclear background had decreased basal heat shock factor I (HSFI) binding activity, increased phosphoinositide signaling, and low AP1 binding [122,123].

Decreased ATP levels (NT2 and SH-SY5Y AD cybrids), reduced mitochondrial movement in neurites (SH-SY5Y AD cybrids), reduced mitochondrial calcium (SH-SY5Y AD cybrids), reduced mitochondrial membrane potential (both SH-SY5Y and NT2 AD cybrids), and reduced mitochondrial “flickering” (SH-SY5Y AD cybrids) have been demonstrated [115,117–119,124–126]. AD cybrids generated on the NT2 nuclear background were more susceptible to cell death induced by β-amyloid exposure [118]. Mitochondrial ultrastructure is affected in SH-SY5Y AD cybrid cell lines, and over time increased mitochondrial proliferation is observed [108,116]. Finally, both intracellular and extracellular β-amyloid levels were increased in SH-SY5Y AD cybrid cell lines [119]. Dense extracellular aggregates which stained positive for Aβ_1–40 and Aβ_1–42 were also observed in the same study.

A recently published, detailed examination of AD cybrid cell lines also reported bioenergetic changes. In this study, 8 AD, 7 mild-cognitive impairment (MCI, frequently a precursor AD state), and 7 age-matched control cybrid cell lines were generated using platelets from human subjects and the SH-SY5Y ρ0 nuclear background [127]. Complex IV activity was significantly decreased when normalized to cytochrome oxidase subunit 4 isoform 1 (COX4I1) protein in the AD group. Both the MCI and AD cybrids displayed less glycolytic activity, while the MCI cybrids had decreased basal mitochondrial oxygen consumption and the AD cybrids showed an increased respiratory leak rate. Both AD and MCI cybrid cell lines had decreased NAD⁺/NADH ratios and the AD cybrids showed increased ADP/ATP ratios. Expression of proteins involved in mitochondrial biogenesis and bioenergetic fluxes were found to be altered. In both the MCI and AD cybrid cell lines, HIF1α protein was decreased, while its mRNA levels were elevated. In contrast, PGC1-α expression was reduced in both MCI and AD cybrids. Sirt1 cellular localization was altered, and phosphorylated Sirt1 protein was decreased in AD cybrids. The AD cybrid cell lines displayed upregulated expression of phosphorylated AMPK, phosphorylated p38, Bax, DRP1, TFAM, cytochrome c, and Complex IV subunits, and mtDNA levels were increased. mTOR expression was reduced and phosphorylated DRP1 levels were decreased in the AD cybrid cell lines. Overall, bioenergetic fluxes, mitochondrial mass, mitochondrial morphology dynamics, and cell signaling pathways were altered in the MCI and AD cybrid cell lines.

An additional study independently examined the same set of AD and control cybrid cell lines [128]. For the AD cybrid cell lines, a reduction in Complex I, Complex III, and Complex IV activities, as well as reduced ATP levels, was reported. ROS production, phosphorylated ERK, and mitochondrial DLP-1 levels were also increased in AD cybrids. Mitochondrial density and length were decreased.

A separate study generated AD, MCI, and age-matched control cybrid cell lines from human platelets and the NT2 ρ0 nuclear background [117]. It was found that both AD and MCI cybrid cell lines had decreased Complex IV activity and mitochondrial membrane potential. Increased lipid peroxidation, superoxide production, and protein carbonyls were observed in both AD and MCI cybrids.

One published study of AD cybrids generated on the HeLa nuclear background is notable for not detecting differences between AD and control cybrid line Complex IV activities [44]. Results from this study are confounded by the fact that mtDNA transfer from platelet, fibroblast, and synaptosomes were attempted, and the sources of mtDNA were not matched between AD and control cybrid cell lines. When AD and control cybrids generated from platelet mtDNA transfer are compared, data from only three cell lines from each group were provided. It is reasonable to take into account this study’s small sample size when considering its conclusions. Regardless, even if this report’s preliminary findings were to be confirmed in an adequately powered HeLa-based cybrid study, the relevance of this to the numerous positive SH-SY5Y and NT2-based AD cybrid studies would be unclear.

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease that has autosomal dominant and sporadic etiology. Sporadic ALS accounts for approximately 90% of ALS cases and has a later age of onset. Based on the idea that mtDNA could act as a late onset, sporadic neurodegenerative disease risk factor, the cybrid model was used to study mtDNA in ALS [129,130]. Platelets from ALS patients were used as the mtDNA donor and SH-SY5Y neuroblastoma ρ0 cells provided the nuclear background. Eleven ALS cybrid cell lines and 12 age-matched control cybrid cell lines were generated. Relative to the control cybrid lines, post-nuclear mitochondrial enriched fractions from the ALS cybrid lines displayed trends for reductions in Complex III and Complex IV activities, and a statistically significant reduction in Complex I activity [131]. A subsequent study, interestingly, reported reduced Complex I activity in muscle tissue from ALS patients [132]. Although ROS production in the form of cytoplasmic peroxide levels was not increased in ALS cybrids, an increase in antioxidant enzyme activities was apparent [129]. ALS cybrid mitochondria also demonstrated reduced mitochondrial calcium concentrations.

A second ALS cybrid study used osteosarcoma ρ0 cells and platelets as the mitochondrial donor source [133]. This study generated 13 ALS cybrids and 10 control cybrids. While no difference in mitochondrial respiration was reported, this study did not directly assay the Complex I Vmax (the parameter demonstrated to be different in the Swerdlow et al. study [129]). Similarly, no change in ROS production in the form of cytoplasmic peroxide was noted; antioxidant enzyme levels were not reported. It is difficult to generalize the findings of these studies because there is a lack of overlap between experimental parameters, and cybrids were generated using different nuclear backgrounds.

Another late onset, sporadic neurodegenerative disease in which the cybrid model has been used is progressive supranuclear palsy (PSP). As early as 1994, PSP subjects were found to have reduced muscle oxidative metabolism [134]. In the initial PSP cybrid study, lines were generated using the SH-SY5Y nuclear background and platelet mitochondria from 15 PSP and 17 age-matched control subjects [135]. Compared to control cybrid lines, in the PSP cybrid lines the Complex I Vmax was reduced by 12.4%, while Complex IV activities were equivalent between the groups. Antioxidant enzyme activities were elevated (presumably upregulated) in the PSP cybrid group. Other investigators have used the same cybrid lines and have reported reductions in oxidative respiration, activity of Complexes I + III, and ATP levels in the PSP cybrids. Aconitase activity is reduced, suggesting increased mitochondrial oxidative stress. Exposure to ETC inhibitors also caused greater mitochondrial membrane depolarization in the PSP cybrids when compared to control cybrids. Treatment of cybrids with a Complex I inhibitor, rotenone, caused a more robust increase in PSP cybrid malondialdehyde (MDA) and hydroxynonenal (HNE) production than it did in control cybrid lines. Increased MDA and HNE in the PSP lines are consistent with a greater degree of rotenone-induced oxidative stress. On the other hand, MDA and HNE differences were not observed after treatment with malonate (a Complex II inhibitor) or cyanide (a Complex IV inhibitor) [136,137].

The cybrid model was used to study Huntington’s disease (HD), a strictly Mendelian, autosomal dominant neurodegenerative disease [138]. While mitochondrial function is perturbed in HD, thereby qualifying it as a mitochondriopathy, based on its genetics there is no compelling reason to suspect the presence of functionally meaningful, disease-specific mtDNA variations (at least not from the perspective of inherited mtDNA sequence deviations). Consistent with this prediction, while direct measurements of HD subject platelets show decreased Complex I activity [139], HD cybrid cell lines do not [138,140,141]. On the other hand, Fereira et al. did report HD cybrids showed increased glycolysis, increased ATP production, increased peroxide production, and decreased mitochondrial NADH:NAD⁺ ratios [140,141]. If correct, the presence of a mutant Huntingtin gene may, therefore, give rise to somatic yet perpetually-sustained mtDNA modifications.

Finally, the cybrid model has also been used to study mtDNA contributions to aging. In an initial study, fibroblast mtDNA was used to generate cybrids from 21 individuals on an osteosarcoma nuclear background [142]. The cybrid lines generated from these 21 individuals were subsequently sub-cloned, ultimately giving rise to a total of 356 uniquely expanded cybrid cell lines. Among these lines an inverse correlation between mitochondrial respiration and the age of the mtDNA donor was observed; the older the mtDNA donor, the lower the respiration capacity of the resulting cybrid line. In contrast to this, though, a separate, much smaller study using HeLa ρ0 cells as the nuclear background did not report correlations between mtDNA donor age and respiration [143]. In this negative study, two cybrid cells lines were generated. In one line, mtDNA was obtained from fibroblasts from a 97-year old, and in the other mtDNA was obtained from fibroblasts from a fetus. This study also used a clonal approach, and a total of five clones were analyzed. Complex IV activity constituted the experimental endpoint, and among these five clones comparable Complex IV activities were observed.

Aging cybrid clones were also generated on an osteosarcoma nuclear background from a 100 year-old donor bearing a T414G mutation. This particular mutation, which is quantitatively associated with aging, is located within the mtDNA control region [144]. No differences in oxygen consumption or ETC activities were observed. However, two different mutations in the 16S rRNA gene (T1843C and A1940G) did associate with reduced oxygen consumption and reduced ETC activities. Lastly, another study that used mouse muscle and synaptosomes to generate cybrid cell lines on a mouse LL/2-m21 nuclear background [27] found differences in cybrid lines derived from young and old mtDNA donors. When muscle cytoplasts were used, cybrids containing mtDNA from the old mice showed decreased Complex I and Complex III activities, less ATP production, and reduced oxygen consumption (but no change in Complex IV activity). When synaptosomes were used, cybrids containing mtDNA form the old mice showed decreased Complex I activity and lowered ATP levels.

Methodological considerations

It is necessary to clarify the point that cybrid methodology varies significantly between individual studies, because these methodological variations may potentially account for and/or contribute to conflicting findings. This section discusses these methodological disparities.

Overall, the nuclear background of the ρ0 cell line used to generate cybrid cell lines influences study outcomes [50,59,71,145,146]. This finding is not surprising because mitochondrial physiology varies among tissue types and nuclear DNA largely determines these variations [147,148]. There is no guarantee that mitochondria obtained from a common source will similarly function within the context of different nuclear backgrounds, or that mitochondria obtained from different sources will consistently function in similar patterns when expressed within the context of different nuclear backgrounds.

The majority of ρ0 cell lines generated for use in cybrid models are tumor cell lines. Tumor cell lines exhibit aneuploidy, and therefore do not contain a normal chromosome complement. Thus, pieces of chromosomes may be missing or translocated, or additional chromosomes may be present. Aneuploidy varies significantly between different ρ0 cell lines [149–153]. While it has not been resolved exactly how aneuploidy impacts mitochondrial function, it could potentially affect nuclear DNA-mtDNA ETC subunit stoichiometry. This could cause some ρ0 cell lines to become more or less sensitive in measurements of mtDNA genotype–phenotype correlations. Moreover, it is important to consider that these tumor cells lines are anaerobic at baseline. Whether or not the aerobic versus anaerobic nature of the host cell influences mtDNA genotype–phenotype correlations should be taken into account. Data regarding this are limited, yet it seems reasonable to assume cybrid models using ρ0 cell lines derived from more aerobic tumor lines may yield inconsistent results when compared with cybrid studies in which ρ0 cells derive from more anaerobic tumor lines.

The majority of cybrid studies have utilized parental cell lines depleted of mtDNA by exposure to EtBr. However, new techniques for depleting mtDNA are becoming commercially available. Therefore, in future cybrid studies it will be imperative to take into account the method by which the parental cell lines were depleted of mtDNA. The source of the mtDNA donation, pre-fusion handling of the mtDNA-donating material, the post-fusion selection process, and variability in cell culture methods can all potentially influence findings. Finally, assay methodology, assay selection and sample sizes are all potential variables that need to be considered.

Reported findings using the cybrid model were obtained from studies that used isolated cybrid clones and mixed populations of cybrid clones. When a particular assessment requires a homogeneous population of cybrid cells, the preparation of clonal cybrids is a more appropriate choice. Examples of such studies include those examining heteroplasmy-threshold relationships of known mtDNA mutations. When cybrids are used to screen for functionally relevant mtDNA differences between mtDNA from different mtDNA donors, or when individually expanded clones from a cybrid line are biochemically assayed, the mean value obtained from the independent assay approximates the value obtained when the same cybrid is assayed using a mixed culture[114]. In conclusion, when using the cybrid model to screen for mtDNA genotype–biochemical phenotype correlations from unsequenced mtDNA, the use of mixed cybrid populations would seem to be the more logical approach.

The “best” endpoint to show functional differences in mtDNA-related cybrid lines is unclear and not universal. However, data suggests that radical production is a useful biochemical measure [154]. In clinical manifestations associated with oxidative stress, the cybrid model can aid in addressing the origin of free radical generation. If cybrid cell lines expressing mtDNA from a group of subjects perpetuate oxidative stress, this suggests mitochondria, and very likely mtDNA, contribute to that oxidative stress.

2. Conclusions

Data generated using cybrid models have enabled investigators to address questions related to mtDNA-induced mitochondrial biology. Furthermore, cybrid studies have provided important insight into how altered mitochondrial function influences cell physiology. Future investigators should find the cybrid model useful when interrogating nuclear–mitochondrial interactions. For studies investigating mtDNA heteroplasmy-threshold analysis, the cybrid model provides straightforward data that appropriately illustrate fundamental mitochondrial genetic principles. However, generalizing data to non-artificial biological models must proceed with caution, because critical differences exist between experimentally generated cybrid cell lines and the actual systems they serve to model.

Functional assessments of mitochondria with un-sequenced mtDNA show mitochondria from different sources are functionally unique. For several sporadic neurodegenerative diseases that do not show Mendelian inheritance, and where mitochondrial dysfunction occurs, mitochondrial transfer from individuals afflicted with those diseases can generate cybrid cell lines which recapitulate and perpetuate the same biochemical defects. The most straightforward interpretation for these findings is that mtDNA at least partly drives functional alterations observed in both the disease state and in cybrid cell lines prepared via the transfer of mitochondria obtained from the human subjects. While this interpretation has not been conclusively confirmed, data to date do not support an alternative explanation.

While cybrids remain a valuable technique and will continue to inform mtDNA-oriented studies, it is worth noting that animal-based models are also increasingly being used to elucidate mtDNA genotype–phenotype correlations. In one example, female progeny were backcrossed repeatedly with their original paternal strain [155]. Since mtDNA is maternally inherited, the mouse line that resulted was one in which the nuclear background was almost exclusively from the paternal strain, while the mtDNA was entirely from the maternal strain. These mouse studies, which use the same nuclear background with varying mtDNA show differences in oxidative phosphorylation phenotypes, which may be relevant to multifactorial diseases [63,156–158]. Other mouse models include those that express mutated versions of genes used in mtDNA replication and maintenance [159–161]. Mitochondrial DNA transgenic mice have also been generated, and review articles describing these advances have been published [162–165]. Finally, recent advances that allow for the transfer of mitochondria between different mouse strains will very likely advance our understanding of nuclear DNA–mtDNA functional relationships, as well as the impact of mtDNA variation on cell physiology [166].