Dissecting the Roles of the Nuclear and Mitochondrial Genomes in a Mouse Model of Autoimmune Diabetes

The mitochondrion is unique in being the only cellular organelle outside of the nucleus that carries its own DNA, which is passed on exclusively from mother to child. The mitochondrial genome in humans encodes for 13 essential polypeptides, two rRNAs, and 22 tRNAs (1). The 13 polypeptides are all subunits of the respiratory chain responsible for generating cellular ATP. The remaining proteins of the respiratory chain, along with proteins required for housekeeping functions, are encoded by nuclear DNA (nDNA). For this reason, mitochondrial function is influenced by both the mitochondrial and nuclear genomes.

Pathogenic mtDNA mutations cause many primary mitochondrial diseases (2). Mitochondrial dysfunction has also been implicated in aging as well as in more common diseases, such as cancer and diabetes. Patients with primary mitochondrial disease caused by mutations in mtDNA tend to have an elevated risk of diabetes (e.g., mtDNA mutation m.3243A>G is associated with diabetes). The number of asymptomatic carriers of mutated mtDNA variants is estimated to be one in 200 adults, whereas one in 5,000 adults is estimated to be affected by mitochondrial disease (3). Thus, the chances of any given mtDNA mutation exhibiting a disease phenotype are often unpredictable.

To add to this unpredictability, not all mitochondria within a cell possess uniform mtDNA. This presence of two or more mtDNA genotypes within a cell is termed heteroplasmy. Changes in the heteroplasmy of a pathogenic variant from mother to child, or even between tissues within the same individual, are one of the deciding factors for disease occurrence (4,5). Moreover, the relative copy number of mtDNA may also vary dramatically between healthy and affected individuals or between different tissues. Experimental evidence has shown that removal of the mitochondrial transcription factor A gene (Tfam) specifically from β-cells leads to a lower mtDNA copy number, compromised glucose-induced insulin responses, and impaired insulin secretion (6). Several studies of patients with mitochondrial diseases (e.g., MELAS and myoclonic epilepsy with ragged-red fibers) have shown that the mtDNA copy number correlates with the severity of disease and has also been reported to affect the penetrance of diseases (7,8). Therefore, copy number variation in the context of mitochondrial diseases can provide information about the underlying genetic cause of the disease, the potential for it to be inherited, and the potential impact on mitochondrial and cellular function.

The generation of ATP by mitochondria within the pancreatic β-cell is a vital step in glucose-stimulated insulin secretion. Substrate oxidation within mitochondria generates ATP, and it is the increase in the ATP-to-ADP ratio within the β-cell that is responsible for the closure of the K⁺ ATP channel, the depolarization of the cell membrane, and the secretion of insulin (9). Type 1 diabetes is an autoimmune disease. The body’s T cells attack and destroy the insulin-producing β-cells found in the pancreatic islets. The resulting lack of insulin then must be compensated for with insulin injections or pumps, making life progressively more difficult and leading to other complications (10).

The NOD/ShiLtJ mouse line has been used as a polygenic mouse disease model of type 1 diabetes. Females from this background start developing diabetes at age ∼12 weeks, whereas diabetes in males is delayed and less frequent (11,12). Because mitochondrial metabolic activity seems to serve a vital role in controlling T-cell response and the progression of autoimmune diseases (13), we set out to examine the role of mitochondrial function in the diabetic phenotype of the NOD/ShiLtJ mice and further explore the pattern of diabetes onset and inheritance by dissecting the contributions of the nuclear and mitochondrial genomes using pronuclear transfer technology (14).

In this article, mitochondrial–nuclear exchange (MNX) mice were generated to help explore two important questions in mitochondrial medicine: 1) the dynamic process involving mtDNA heteroplasmy and its inheritance pattern between mother and offspring and 2) the relative contribution of the nuclear versus mitochondrial genome to diabetes. To this end, NOD/ShiLtJ and nondiabetic C57BL/6J mice were selected to undergo MNX. The resulting chimeric mouse strain allowed us to unravel the underlying causes by testing the effect of different mitochondrial genomes in the presence of the same nuclear genomic background. At the same time, heteroplasmic mice obtained during the creation of the MNX mice allowed us to simultaneously investigate for signs of possible directional selection favoring one mtDNA haplotype over the other.

MNX mice were generated via pronuclear transfer as described previously (15), with some modifications. Recipient and donor zygotes were generated for both the NOD/ShiLtJ and C57BL/6J backgrounds. All mice were ordered from The Jackson Laboratory (Bar Harbor, ME), and all animal procedures were reviewed and approved by the Cincinnati Children’s Hospital and University at Buffalo Institutional Animal Care and Use Committees.

In brief, females were first superovulated (16) and then mated with males of the same genotype. Fertilized zygotes were harvested the morning after fertilization, and the cumulus cells were removed using 3-min incubation with 0.1% hyaluronidase.

Pronuclei were first removed from the recipient zygotes. Donor pronuclei from the opposite strain were then isolated and transferred into the enucleated recipient zygotes. All zygote manipulations and injections were performed using an Olympus IX73 inverted microscope (Olympus Corporation) with a 37°C heated staged micromanipulator (Narishige International USA, Inc.) and a Piezo-Drive microinjector. The pronuclear transfer zygotes were washed three times and then cultured in Cleavage Medium (Cook Medical, USA) at 37°C with 5% CO₂.

The resulting zygotes were transplanted into a pseudopregnant female recipient mouse, resulting in four viable offspring. Two of the offspring were male C57BL/6J^NOD/ShiLtJ mice, which were not suitable for establishing an MNX line. The other two offspring were female NOD/ShiLtJ^C57BL/6J mice. One of these offspring (no. 9074 in Fig. 1) died prematurely as a result of apparent dehydration, but the remaining NOD/ShiLtJ^C57BL/6J mouse (no. 9075 in Fig. 1) was able to successfully establish an MNX line through mating with a NOD/ShiLtJ male.

Regular weighing and biweekly glucose testing of the mice were carried out using the Accu-Chek Guide meter and glucose strips (Roche Diabetes Care, Inc.). Readings of ≥250 mg/dL defined diabetes onset. For genotyping, ear/tail clips were collected and genomic DNA extracted for PCR amplification and Sanger sequencing. The following primer set was used: forward, 5′-ACTTCACCATCCTCCAAGC-3′, and reverse, 5′-GAGAGCGAAATATAAGTGTCCC-3′. The reaction conditions were as follows: 95°C for 2 min, followed by 40 cycles of 94°C for 30 s, 58°C for 30 s, and 72°C for 30 s. A final extension step was then performed at 72°C for 4 min.

Complex activity assays for complexes I and IV were run using specific kits from Abcam. Liver tissue was placed in a 1.5-mL isolation buffer (225 mmol/L mannitol, 75 mmol/L sucrose, 5 mmol/L HEPES, and 1 mmol/L EGTA [pH 7.4] when ice cold) and minced into very small pieces, after which 5.5 mL of the isolation buffer was added and transferred to a Dounce homogenizer. After complete homogenization, it was placed in 15-mL falcon tubes and centrifuged at 2,000g for 5 min. The supernatant was transferred to a fresh tube and centrifuged at 14,000g for 10 min at 4°C. The supernatant was then discarded, and the pellet was resuspended in 7 mL of isolation buffer. The supernatant was saved and centrifuged again at 14,000g for 10 min. The pellet obtained was then suspended in 300 μL of assay buffer (125 mmol/L, 20 mmol/L HEPES, 2 mmol/L MgCl₂, 2 mmol/L KH₂PO₄, and 40 μmol/L EGTA [pH 7.2]). Protein estimation was carried out using the Pierce BCA Protein Assay Kit (cat. no. 23225; Thermo Fisher Scientific).

Tissue DNA was extracted using the DNeasy Blood and Tissue Kit (cat. no. 69504; Qiagen). DNA was then used to perform quantitative PCR using the QuantiTect SYBR Green PCR Kit (cat. no. 204143; Thermo Fisher Scientific). The primer sets used were as follows: tRNA-Val-forward, 5′-CTAGAAACCCCGAAACCAAA-3′; tRNA-Val-reverse, 5′-CCAGCTATCACCAAGCTCGT-3′; B2M-forward, 5′-ATGGGAAGCCGAACATACTG-3′; and B2M-reverse, 5′-CAGTCTCAGTGGGGGTGAAT-3′. The reactions were performed as follows: 95°C for 10 min, followed by 45 cycles of 95°C for 30 s, 58°C for 30 s, 72°C for 45 s, and 72°C for 10 min.

Mouse pancreatic tissues were first fixed in 4% paraformaldehyde overnight. Frozen tissue blocks were then made, and tissue sectioning was performed on a Leica CM1800 cryostat. Slides were blocked using blocking buffer (10% goat serum, 50 μL Triton X-100, and 9 mL PBS) for 1 h at room temperature and then washed for 5 min each in PBS. The slides were stained overnight at 4°C with primary antibody solutions rabbit monoclonal antibody insulin (cat. no. 3014S; Cell Signaling Technology) and rabbit anti-mouse CD3 monoclonal antibody (cat. no. 14-0032-82; eBioscience) at 1:500 and 1:100 dilutions, respectively. The next day, slides were washed using PBS three times for 5 min each and then stained with goat anti-rabbit immunoglobulin G (H+L) cross-adsorbed secondary antibody, Alexa Fluor 488, and goat anti-rat immunoglobulin G (H+L) cross-adsorbed secondary antibody, Alexa Fluor 568, both diluted to 1:500. Slides were incubated in the dark at room temperature for 1 h and then washed three times with PBS. Slides were then mounted in Vectashield (Vector Laboratories, Newark, CA) and imaged on an Andor Dragonfly spinning disk confocal microscope (Andor Technology, Belfast, Northern Ireland).

Fresh pancreatic tissues originating from 14-week-old NOD/ShiLtJ, C57BL/6J, and MNX mice were immersed in a 10% formalin solution and fixed for 24 h. The fixed tissues were embedded in paraffin and sectioned. Hematoxylin-eosin staining was performed, and images of the sections were scored blindly by two investigators according to published criteria (17).

Mouse embryonic fibroblasts were seeded into an XFe96 cell culture plate (Agilent Technologies) at a density of 1.0 × 10⁴ per well, and XF calibrant solution was added in advance. After 24 h, 1 μmol/L oligomycin A, 2 μmol/L FCCP, and 500 nmol/L rotenone/1 μmol/L antimycin A from the Seahorse XF Cell Mito Stress Test Kit (Agilent Technologies) were added to the sensor cartridge (Agilent Technologies). The culture medium in the cell culture plate was then removed, and fresh prewarmed test medium was added. After this, the cell culture plate was incubated in a CO₂-free incubator at 37°C for 1 h. The sensor cartridge was then placed in the Seahorse XFe96 Analyzer (Agilent Technologies) for equilibration before the culture plate was placed in the device for analysis.

ATP production was measured from mouse embryonic fibroblasts using the ATP Bioluminescence CLS II Kit (Roche, Basel, Switzerland) according to the manufacturer’s instructions. A BioTek Synergy H1 plate reader was used to read the luminescence values.

The data sets generated and/or analyzed during the current study are available from authors J.S. or T.H. upon reasonable request.

The mouse lines NOD/ShiLtJ and C57BL/6J were chosen to undergo pronuclear transfer. The resulting embryos after pronuclear transfer were transplanted into a pseudopregnant mouse (Fig. 1). The surviving female progeny (no. 9075 in Fig. 1) was then crossed with a NOD wild-type male to produce F1 pups of varying levels of heteroplasmy for the C57BL/6J haplotype.

The C57BL/6J strain was chosen for this study because it differs from NOD/ShiLtJ at only three specific mtDNA positions (Fig. 2A). The specific genes affected by these variants are predicted to affect the translational process as well as the functioning of individual respiratory complexes (Fig. 2A). The mtDNA variant m.9461T>C, which affects the ND3 gene that codes for a protein subunit of NADH (i.e., ubiquinone oxidoreductase, also known as respiratory complex I), would be predicted to alter that activity of complex I. Similarly, the m.9348G>A and m.9821insAA variants would be expected to affect respiratory complex IV or overall mitochondrial protein biosynthesis, respectively.

Next-generation sequencing (NGS) was first used to sequence tissues samples from the mice to determine their mtDNA composition (Fig. 2B) (18). In later generations, quantification based on the peak heights of each variant using Sanger sequencing quickly became the method of choice, because it provided similar results to NGS using a simpler and faster protocol (Fig. 2B). For calculating the heteroplasmy levels from Sanger sequencing, the peak heights were measured directly from chromatograms using ImageJ, and the relative frequency for each variant at a given position was calculated using the following formula: frequency of variant X = X peak/(X peak + Y peak).

The fidelity of transmission from mother to offspring for a heteroplasmic mtDNA variant is a unique and critical aspect of mtDNA-based diseases. Heteroplasmy levels often vary between tissues from the same animal depending on the organ affected and the type of disease. Variations in heteroplasmy can also occur from mother to offspring as a result of the mitochondrial bottleneck that occurs during oogenesis (19,20). Our own data bear this out, with large changes in heteroplasmy often observed between tissues in the same animal (Fig. 3A–C) and across generations (Fig. 3D–F). However, when tissue-to-tissue heteroplasmy shifts are averaged across animals, the observed shifts are statistically insignificant (Fig. 3B). Furthermore, despite a few exceptions, the progeny in our data set showed heteroplasmy levels that were generally close to those of their mother. Furthermore, across 167 progeny, there was an average change of only ∼0.17% toward the NOD haplotype, which was found to be statistically insignificant (Fig. 3E). This suggests that there is no significant selection favoring either haplotype in either the germ line or somatic tissue.

To compare the course of diabetes between the control and MNX mice, regular monitoring of body weight, blood glucose levels, and survival was carried out (Fig. 4). Overall, the NOD control mice were the first to become diabetic and eventually die. In contrast, while the MNX mice would still develop diabetes and eventually die as well, their disease course was noticeably delayed, implying that mtDNA plays some role in disease progression. C57BL/6J mice were, as expected, not prone to the disease at all. Together, these results imply a positive effect of the NOD/ShiLtJ haplotype on diabetes onset and overall survival.

In order to establish a molecular basis for the improved phenotype observed in the MNX mice, we next tested the mitochondrial function of mouse embryonic fibroblast cells derived from the MNX, NOD control, and C57BL/6J backgrounds. Seahorse analysis showed statistically significant improvements in basal respiration, maximal respiration, spare respiratory capacity, and ATP-linked respiration for the NOD/ShiLtJ^C57BL/6J MNX mice relative to NOD control mice, although this improvement did not quite reach the level seen in nondiabetic C57BL/6J cells (Fig. 5A). A similar level of improvement was also observed for the MNX fibroblasts in ATP production assays as compared with NOD control cells (Fig. 5B), further suggesting that mitochondrial function is improved in the MNX cells.

On the basis of these results, we focused next on narrowing down the underlying mitochondrial functions behind the metabolic and phenotypic improvements in the MNX mice. Because two of the three variants that differ between C57 and NOD occur in components of complexes I and IV, we hypothesized that alterations in the activity of one of these complexes could be the underlying cause for the differences between the MNX and NOD control mice.

On the basis of this hypothesis, complex I and IV activity assays were performed on tissues from NOD control and MNX mice. The results showed a significantly higher level of complex I activity in liver mitochondrial extracts from NOD control mice relative to MNX mice (Fig. 6A), whereas complex IV activity was unchanged between the two strains (Fig. 6B). This suggests that a reduction in complex I activity (and the associated variant at mtDNA position 9,461) may be the underlying cause for the delayed diabetic phenotype in the MNX mice.

To investigate the possible role of mtDNA copy number in the differences between the MNX and NOD control mice, a quantitative PCR analysis was performed. However, no significant differences in mtDNA copy number were observed between the MNX and NOD control mice for any tissue tested (Fig. 7), indicating that the differences between these strains are not due to differences in the overall amount of mtDNA or mitochondrial biomass.

To further confirm that there was indeed a difference in the rate of disease occurrence and progression between the NOD control and MNX mice, histologic studies were performed to provide direct validation of pathologic changes in the pancreas (Fig. 8). To do this, we first carried out an insulitis scoring analysis of hematoxylin-eosin–stained pancreatic tissues (Fig. 8A–C) to characterize the level of lymphocyte infiltration into the islets. This analysis showed a statically significant reduction in the leukocyte infiltration of the MNX tissues as compared with the NOD control tissues (Fig. 8B and C), in agreement with the delayed onset of diabetes and improved survival in the MNX mice. As an extension of these results, pancreatic cryotissue sections from age-matched females were also obtained and stained for insulin and CD3. The NOD control tissue slides showed less insulin staining and fewer pancreatic islets overall, which were also highly infiltrated by CD3⁺ T cells, in comparison with the MNX tissues (Fig. 8D and E). Statistical analysis using ImageJ was also performed on the confocal images (21) to calculate the area covered by intact islet cells and CD3 cells. Interestingly, although the pancreatic cells were clearly affected by diabetes in both groups, both islet cell loss and T-cell infiltration were noticeably reduced in the MNX mice (Fig. 8F and G), in line with the improved survival and delayed disease course observed for the MNX mice.

The impact of the mitochondrial genome on common diseases like type 1 diabetes is not yet fully understood. This study was designed to explore the impact of both the nuclear and mitochondrial genomes on disease onset, progression, and survival. To do so, mice were created using MNX, the nuclear genome of which belonged to the diabetes-prone NOD/ShiLtJ genetic background but the mitochondrial genome belonged to the nondiabetic C57BL/6J background. These genetically unique mice allowed us to explore several important questions related to mitochondrial inheritance and the role of mtDNA in diabetes.

First, these mice represented a chance to conduct an animal study of the genetics and transmission dynamics related to mitochondrial replacement therapy. Mitochondrial replacement therapy was successfully carried out in a 38-year-old woman asymptomatic for Leigh syndrome (mtDNA 8993T>G) with a history of four pregnancy losses (22). Metaphase II spindle transfer was used in this case to reduce mutational mtDNA load in the offspring to as low as 2.36–9.23% in different tissues. Long-term follow-up to date has not recorded any signs of disease occurrence. This case was a crucial building block for our study. Pronuclear transfer was used for our study to examine the dynamics involved with disease transmission in terms of heteroplasmy. Most offspring were closer to the heteroplasmy of the mother, but a few were noted to have scattered heteroplasmy percentages. This phenomenon of shifts in heteroplasmy from mother to child seems to be the result of a genetic bottleneck in the germ line, wherein only a subset of mtDNA molecules replicate during a key phase of oocyte development. This can often lead to dramatic and unpredictable changes in heteroplasmy frequencies across generations (19,20). Although a wide range of shifts were observed, upon statistical testing no significant bias toward either haplotype was observed. This suggests that, at least at the level of the germ line, neither mtDNA haplotype possesses a replicative or selective advantage over the other.

Beyond the inheritance patterns observed in the MNX mice, phenotypic readouts such as weight, blood glucose, and survival all showed that the MNX mice had delayed onset of diabetes and prolonged survival relative to the unaltered NOD control mice. Immunofluorescence staining of pancreatic tissues from the two groups also suggested that the disease was at a more advanced stage in NOD mice as compared with MNX mice. Specifically, there was a decrease in the number of intact islet cells but an increase in CD3⁺ T cells in the NOD mice at the same age. To help understand this delayed disease progression, enzymatic activity assays for respiratory complexes I and IV were carried out to see if any of the three different mtDNA variants showed a significant impact on disease onset or progression. The m.9461T>C variant, which is a synonymous substitution (AUU to AUC) in the start codon of the ND3 gene, did seem to affect activity levels for complex I of the electron transport chain. Specifically, an increase in complex I activity in the NOD control mice was seen in comparison with the MNX mice. This suggests that this specific mtDNA variant may be responsible for the accelerated disease onset and progression in NOD mice. On the other hand, no significant differences in mtDNA copy number were observed between the tissues of NOD and MNX mice, ruling out mitochondrial biomass or the overall amount of mtDNA as a factor in the NOD disease progression.

This study was a substantial stepping stone in terms of the interplay between the nuclear and mitochondrial genomes in a commonly occurring disease. At least one previous study has examined the role of a separate mtDNA mutation in a complex I subunit, mt-Nd2^a, in protecting against the onset of autoimmune diabetes in the NOD background. Interestingly, although the mt-Nd2^a variant ultimately failed to protect NOD mice from developing diabetes, it did prevent them from acquiring diabetes after adoptive transfer of diabetogenic BDC2.5 CD4⁺ T-cell clones (23). Therefore, there would still seem to be a protective trend present with this separate complex I variant, consistent with a key role for complex I in the progression of autoimmune diabetes. This possibility that complex I dysfunction and/or hyperactivity may underpin the NOD diabetic phenotype can be further elaborated in future experiments in other mouse models of type 1 diabetes, such as streptozotocin-treated mice. Another option may be to use mtDNA-based editing enzymes to alter specific nucleotides in the mitochondrial genome of the original NOD/ShiLtJ strain. Recent work with DddA-derived cytosine base editors has shown that these enzymes can be used to create targeted C-to-T base conversions in mtDNA (24).

Ultimately, we believe this study will enable the development of possible remedies for type 1 diabetes, as well as mitochondrial diseases as a whole. Studies in mouse models of experimental autoimmune encephalomyelitis have shown that drugs that inhibit glycolysis and force a shift toward oxidative phosphorylation, such as 2-deoxy-d-glucose and dichloroacetate, can reduce T-cell activation and suppress the autoimmune response (25,26). Future experiments in our NOD pronuclear transfer mice may provide additional insights into the value of such interventions in an animal model of type 1 diabetes. For example, metformin is an oral antihyperglycemic agent commonly used to treat type 2 diabetes. It is known to increase insulin sensitivity in the liver, improve overall glycemic control, and reduce metabolic syndrome in prediabetic individuals. However, it has only recently been tested for efficacy against type 1 diabetes (27,28). Moreover, the hyperactivity of complex I seen in unaltered NOD mice would be expected to cause increased reactive oxygen species (ROS) production and oxidative stress, leading to the dysfunctional ATP production and increased T-cell infiltration that were observed in our data. In this light, it is interesting to note that metformin has been previously shown to alleviate salivary gland inflammation in NOD mice, although the effect of the drug on diabetes onset was not described (29). Other studies in diabetic mouse models have observed the same combination of complex I hyperactivity, increased ROS, and decreased ATP production and have also suggested that repression of complex I hyperactivity and/or ROS originating from complex I could prevent or delay the occurrence of diabetes (27,30). The results suggest that future studies should keep electron transport chain energy in mind as a possible contributing factor in the pathology of type 1 diabetes.

Acknowledgments. The authors thank Dr. Yueh-Chiang Hu and the Cincinnati Children’s Hospital Medical Center Transgenic Animal and Genome Editing Core Facility for generously allowing us to use their equipment for performing the pronuclear transfer protocol to generate the MNX mice. The authors also thank the Optical Imaging and Analysis Facility, School of Dental Medicine, State University of New York at Buffalo, where the confocal images in this study were acquired.

Funding. W.Z. received funding from the National Natural Science Foundation of China (grants U20A20350 and 82001635) and the National Key R&D Program of China (grant 2021YFC2700901).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. W.Z. designed and generated the NOD/ShiLtJ^C57BL/6J MNX mice. W.Z., J.C., T.Y., W.L., J.V., and J.S. performed the subsequent experiments, data collection, and statistical analysis. J.C. wrote the first version of the manuscript. J.S. and T.H. designed the study. All authors were involved in the subsequent writing and revising of the manuscript and approved the final version. T.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.