Expression of Rattus norvegicus mtDNA inMus musculus Cells Results in Multiple Respiratory Chain Defects*

The production of in vitro andin vivo models of mitochondrial DNA (mtDNA) defects is currently limited by a lack of characterized mouse cell mtDNA mutants that may be expected to model human mitochondrial diseases. Here we describe the creation of transmitochondrial mouse (Mus musculus) cells repopulated with mtDNA from different murid species (xenomitochondrial cybrids). The closely related Mus spretus mtDNA is readily maintained when introduced into M. musculus mtDNA-less (ρ0) cells, and the resulting cybrids have normal oxidative phosphorylation (OXPHOS). When the more distantly related Rattus norvegicus mtDNA is transferred to the mouse nuclear background the mtDNA is replicated, transcribed, and translated efficiently. However, function of several OXPHOS complexes that depend on the coordinated assembly of nuclear and mtDNA-encoded proteins is impaired. Complex I activity in the Rattusxenocybrid was 46% of the control mean; complex III was 37%, and complex IV was 78%. These defects combined to restrict maximal respiration to 12–31% of the control and M. spretusxenocybrids, as measured polarographically using isolated cybrid mitochondria. These defects are distinct to those previously reported for human/primate xenocybrids. It should be possible to produce other mouse xenocybrid constructs with less severe OXPHOS phenotypes, to model human mtDNA diseases.

Mitochondrial DNA (mtDNA) 1 is a 16.5-kilobase pair genome unique to mitochondria, which encodes 13 essential protein subunits of the multimeric OXPHOS complexes I, III, IV, and V. Complex II is entirely nuclear gene-encoded. Unlike nuclear genes the mtDNA exists in thousands of copies per cell, is replicated throughout life, and in most animals including humans is maternally inherited. Transcription of mtDNA is linked to replication, and the 13 proteins are translated on mitochondrial ribosomes and assembled with around 60 nuclear gene products to produce the functional holocomplexes (for recent reviews see Refs. 1 and 2).
The five OXPHOS complexes act as a redox pathway, trans-ferring electrons from NADH and FADH 2 to molecular oxygen. The passage of electrons is linked to proton efflux from the matrix through complexes I, III, and IV across the mitochondrial inner membrane. The proton gradient in turn is the source of power for the H ϩ -ATP synthase (2). OXPHOS produces the majority of the ATP of the cell and is also a major source of reactive oxygen species production. In efficiently respiring mitochondria around 2% of oxygen consumed is directly converted into reactive oxygen species, and this amount is increased when the respiratory chain is inhibited in in vitro and in vivo studies (3). Genetic defects of OXPHOS may therefore result from nuclear or mtDNA gene mutations. Since the first descriptions of disease-related mtDNA mutations appeared just over 10 years ago (4,5), the catalogue of such mutations has grown to over 50 (see Refs. 6 and 7 and MITOMAP). Clincal presentations of mtDNA diseases are extremely variable, although a few syndromes are associated with specific mtDNA defects. Presentations range from severe and progressive neonatal or childhood neuromuscular disease, to milder adult disease including muscle weakness, cardiac disease, diabetes, vision, and hearing loss (see Refs. 6 -8).
Despite these genetic advances, our understanding of pathogenesis in mitochondrial diseases remains poor, with little progress made on the underlying causes of clinical heterogeneity and progression of symptoms, which in turn has inhibited development of directed therapies. The creation of in vitro and in vivo mouse models of mtDNA diseases is therefore of great current interest. However the unique features of mitochondrial genetics including the high cellular mtDNA copy number and the lack of demonstrated mechanisms of recombination have presented technical barriers to producing mtDNA "knockout" mice.
Several groups have reported different approaches aimed at directly manipulating the mitochondrial genotype of oocytes or zygotes or using direct mitochondrial injection into oocytes (see Ref. 9). Two other groups (10,11) have reported production of transmitochondrial mice using an embryonic stem (ES) cell approach and one of very few characterized cultured mouse cell mtDNA mutants (a rRNA mutation resulting in resistance to chloramphenicol). One of these groups recently reported successful germ line transmission of this mtDNA mutation, although from limited data presented to date the phenotype appears to be post-natal lethal (12).
If the ES cell approach to producing transmitochondrial mice is validated, the major limitation to producing models is now the lack of available mtDNA mutants in cultured mouse cells. Another approach to producing OXPHOS defects in mouse cells was suggested by Kenyon and Moraes (13) who established that human mtDNA-less ( 0 ) cells could be repopulated with closely related primate mtDNAs. The resulting "xenomitochondrial cybrids" could replicate and transcribe the foreign mtD-NAs and showed defective respiratory chain complex I with preserved function of the other complexes (14).
Here we report the introduction of mtDNAs from different murid species into mouse cells to produce mouse xenomitochondrial cybrids. We have introduced closely related (Mus spretus) and more distantly related (Rattus norvegicus) murid mtDNAs into a Mus musculus 0 cell line, which in the latter case results in multiple respiratory chain defects similar to some human mtDNA diseases.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture Conditions-All cells were grown in RPMI 1640 medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Life Technologies, Inc.) at 37°C and 5% CO 2 . Primary fibroblast lines were created from a 2-day-old laboratory mouse (M. musculus, CBA/C57Bl6 cross) and 5-week-old M. spretus (Spret/Ei, a gift from Dr. Simon Foote, Walter & Eliza Hall Institute). Skin from the belly of euthanized mice was washed with phosphate-buffered saline with 2ϫ penicillin/streptomycin (Life Technologies, Inc.) and incubated with 0.25% trypsin (Life Technologies, Inc.) in RPMI medium for 4 h at 4°C, incubated at 37°C for 20 min, then passed through a tissue sieve using a glass pestle, and plated in complete medium in 75-cm 2 tissue culture flasks (Nunc, Denmark). Media were replaced twice weekly, and after 2-3 weeks the primary cultures began to grow vigorously, and aliquots were viably frozen.
Two independent mouse 0 cell clones were used, designated LMEB3 and LMEB5. These were produced from the parental line LMTK Ϫ by exposure to ethidium bromide and are described in detail elsewhere (15). These mtDNA-less cells are auxotrophic for uridine and pyruvate (see Ref. 16), and the medium was supplemented with glucose to 4.5 mg/ml, uridine, 50 g/ml, and pyruvate, 1 mM (RPMI/GUP medium).
Production of Transmitochondrial Cybrids-Cybrids were produced by enucleation of mitochondrial donor cells and fusion of the cytoplasts with mouse 0 cells followed by selection for respiratory-competent transformants as described in detail previously for human cells (17). Cells used as mitochondrial donors included the mouse primary fibroblast line, the M. spretus primary fibroblast line, the Rattus lines NRK52E and RN1T, and the hamster lines Chinese hamster ovary cells and Syrian hamster kidney. Briefly, 5 ϫ 10 6 cells were enucleated by centrifugation at in an isopycnic Percoll (Amersham Pharmacia Biotech) gradient in the presence of 10 g/ml cytochalasin B (Sigma). The cytoplast/karyoplast mixture was combined with 2 ϫ 10 6 mouse 0 cells and then centrifuged at 20,000 ϫ g for 10 min; the supernatant was aspirated and the pellet overlaid with polyethylene glycol (Sigma) for 1 min. The polyethylene glycol was then removed, and the cells were gently resuspended in complete medium and plated at 10 4 and 10 5 cells per 100-mm dish in RPMI-GUP medium. After 24 h the medium was replaced with select medium as follows: RPMI supplemented with 5% dialyzed fetal bovine serum (Life Technologies, Inc.) and 50 g/ml bromodeoxyuridine (Sigma). In this medium lacking uridine and pyruvate the 0 cells cannot grow, and the bromodeoxyuridine kills surviving TK ϩ mitochondrial donor cells so that only LMTK Ϫ cybrids can survive. After 7-14 days cybrid clones were isolated using cloning cylinders, expanded, and viably frozen. Three independent clones were frozen from each successful fusion experiment, and cybrids were used in experiments from passage 5 through passage 10.
To confirm the nuclear origin of the cybrids, karyotyping of a control cybrid and a Rattus xenocybrid clone was performed. Cells were grown in RPMI/GUP medium in a 25-cm 2 flask. Cells were treated with 50 ng/ml demecolcine (Sigma) for 5 h, harvested, resuspended in 10 ml of 75 mM KCl, and incubated at room temperature for 12 min. Cells were pelleted before being resuspended in 5 ml of fresh methanol/acetic acid (3:1). Cells were fixed overnight at 4°C, pelleted, washed once in fixative, and then resuspended to give a pale milky solution. Cells were then dropped onto wet, acid-washed slides from a distance of 70 cm, allowed to dry, and stained with Giemsa (Sigma).
mtDNA Genotyping by DNA Sequencing-A mtDNA D-loop fragment was generated by PCR from genomic DNA isolated from cybrid clones. Primers were chosen to enable amplification from both mouse (18) and R. norvegicus (19) mtDNA. The mouse amplicon was 461 base pairs in length, whereas the rat amplicon was 496 base pairs due to intervening sequence insertions compared with the mouse mtDNA (18,19). The forward primer sequence was 5Ј-ctc aac ata gcc gtc aag gc-3Ј representing nucleotides 15934 -15953 (18), and the reverse primer was 5Ј-acc aaa cct ttg tgt tta tgg g-3Ј representing nucleotides 59 -80. PCR was performed using 1.0 unit of Taq polymerase (Life Technologies, Inc.), 20 ng of DNA, and 35 cycles of 94°C for 30 s, 55°C for 1 min, and 72°C for 1 min.
Sequencing was performed using 60 ng of purified amplicon, with a Thermo Sequenase II dye terminator cycle sequencing kit (Amersham Pharmacia Biotech). Sequencing reactions were visualized on an ABI 377 automated sequencer.
Southern Blot Determination of mtDNA-The mouse and rat D-loop PCR amplicons described above were used to probe mtDNA, whereas a PCR fragment of mouse 18 S rRNA gene was used as a nuclear probe. This PCR amplicon was 501 base pairs in length; the forward primer sequence was 5Ј-ctg tgg taa ttc tag agc taa tac atg ccg-3Ј, representing positions 151-180 of the mouse 18 S rRNA gene (GenBank TM accession number X00686) The reverse primer sequence was 5Ј-tat acg cta ttg gag ctg gaa tta cc-3Ј, representing nucleotides 626 -651. Probes were labeled with [ 32 P]dCTP using a RediPrime II random prime labeling kit (Amersham Pharmacia Biotech). For each lane of a 0.85% agarose gel, 5 g of genomic DNA was loaded after restriction with SacI (Life Technologies, Inc.), which has a single restriction site in mouse and rat mtDNA. The transfer, probe annealing, and washing was done using standard conditions.
Mitochondrial Protein Translation-Mitochondrial translation products were labeled with [ 35 S]methionine as described previously (20). Approximately 10 5 cells in single wells of a 24-well tissue culture dish (Nunc) were washed with methionine-free RPMI before adding 0.5 ml of the same medium containing 0.1 mg/ml cycloheximide and incubating the cells at 37°C, 5% CO 2 for 15 min. After this preincubation 50 Ci of [ 35 S]methionine (ICN) was added to each well, and the cells returned to the incubator for 2 h. Cold methionine (0.1 mM, Sigma) was then added, and the cells were incubated a further 30 min. Cells were immediately harvested, centrifuged, and frozen at Ϫ80°C.
Cell pellets were prepared for electrophoresis by thawing and dilution in 50 mM Tris, pH 6.8, containing 0.1% SDS and 1 mM ␤-mercaptoethanol. The cell lysates were sonicated with 3 pulses at setting 5 using a Tosco sonicator (MSE, Melbourne, Australia). Electrophoresis of labeled proteins was performed using 15% acrylamide (1:36 bisacrylamide) mini-gels (CBS Scientific, Del Mar, CA) with 10 g of cellular protein per lane. Gels were electrophoresed at 100 V for 30 min, fixed in methanol/acetic acid, treated with "Amplify" (Amersham Pharmacia Biotech), and dried for autoradiography.
Lactate Measurement-Lactate was measured in media using a commercial kit (Sigma). Cybrids were grown to confluence in 24-well culture dishes, and the media were replaced with 0.5 ml of fresh media after two rinses with 1 ml of media, and the cells were incubated for 10 h. Samples of media were removed at 2-h intervals and tested immediately for lactate concentration.
Mitochondrial Isolation and Polarographic Analysis-Mitochondrial isolation and polarography using freshly isolated mitochondria was performed as described previously in detail for human cell mitochondria (17). Control and xenomitochondrial cybrids (xenocybrids) were grown in paired cultures so that a control was always processed alongside the experimental xenocybrid culture. Cultures were expanded to around 10 9 cells by seeding 2 ϫ 10 7 cells into roller bottles (Corning) in 200 ml of complete medium, expanded after 4 or 5 days to 1000 ml, and then harvested after another 4 days.
Mitochondria were isolated by digitonin lysis and homogenization, followed by differential centrifugation (17). The mitochondria were resuspended to around 20 mg/ml protein in isolation buffer and used immediately for polarographic analysis. Protein was measured according to the Lowry method. Aliquots were also frozen at Ϫ80°C for enzymological analysis.
Polarography was performed using an Instech (Plymouth Meeting, PA) microelectrode and chamber coupled with a magnetic stirrer and an Instech model 203 dual oxygen electrode amplifier with output to a chart recorder. Each mitochondrial isolate was tested for respiratory capacity using pyruvate plus malate, glutamate plus malate, and succinate as substrates. Two runs were performed with each substrate, and in each run two additions of ADP were made. Conditions and experimental details were as described previously (17).
OXPHOS Enzymology-Complex I (NADH:ubiquinone oxidoreductase, EC 1.6.5.3), complex II (succinate:ubiquinone oxidoreductase, EC 1.3.5.1), complex II ϩ III (succinate:cytochrome c oxidoreductase), and complex IV (ferrocytochrome-c:oxygen oxidoreductase, or cytochrome oxidase, EC 1.9.3.1) activities were measured spectrophotometrically, essentially as described previously (17), with the following modifica-tions. A single beam spectrophotometer was used for all assays (Cintra 10, GBC Melbourne) instead of a dual-beam machine. For complex I, the oxidation of NADH was followed at 340 nm (21) using ubiquinone-1 (Sigma) as electron acceptor. For the complex II, II ϩ III, and IV assays, conditions were as described previously (17) except the single-beam machine was used. For each assay, each mitochondrial sample was tested in triplicate.  Fig. 1). Both the control cybrids and M. spretus xenomitochondrial cybrids were similar in appearance and growth characteristics to each other and to the parental LMTK Ϫ cell line. Fusions using enucleated Rattus NRK52E cells with LMEB5 and enucleated Rattus RN1T cells with LMEB3 also produced cybrids at high frequencies, around one per 10 4 0 cells used. However both these sets of cybrid clones grew more slowly, showed a morphology intermediate between the control cybrids and the 0 cells, and acidified the media much more quickly than the control cybrids. The Mus/Rattus divergence is dated at around 10 million years ago (Fig. 1). Table I summarizes the amino acid sequence differences between the M. musculus and R. norvegicus mtDNA-encoded proteins.

Production of Transmitochondrial Cybrids-Control
Fusions of enucleated hamster cells with LMEB3 failed to produce any cybrids in two experiments, one using Chinese hamster (C. grigaeus) cells and the other using Syrian golden hamster (M. aureus) cells. The Muridae subfamily Cricetinae, which includes the hamsters, diverged from the Murinae subfamily around 16 million years ago (Fig. 1).
Karyotyping of a control (M. musculus) cybrid and a Rattus xenocybrid revealed the same number of chromosomes, with a mean of 40 Ϯ 2.2 (n ϭ 4 counts) for the control, and 40 Ϯ 1.5 (n ϭ 6 counts) for the xenocybrid.
mtDNA Genotyping of Cybrids by DNA Sequencing-For the M. spretus and R. xenocybrids, mitochondrial transfer was verified by direct sequencing of a PCR fragment of the mtDNA D-loop region. The M. spretus xenocybrid showed 81% identity with the published M. musculus sequence (18) for the region sequenced (GenBank TM accession number AF287305). The Rattus xenocybrids produced using NRK52E cells as mitochondrial donors showed only 62% identity to M. musculus for the region sequenced and in agreement with the published R. nor-vegicus sequence (19). The cybrids made from RN1T cells showed a single nucleotide difference to the R. norvegicus sequence, an insertion of a C in a 8-C tract at nucleotide pair 16075. This cell line is listed by ATCC as Rattus rattus, but this almost perfect identity with the R. norvegicus sequence indicates it is also derived from R. norvegicus, as R. rattus would be expected to show many more sequence differences in this highly polymorphic D-loop region, based on molecular studies of these two species (25).
Southern Blot Determination of mtDNA in the Rattus Xenocybrid-Southern blot of control cybrid and Rattus xenocybrid DNA after restriction with SacI and probing with a mouse and rat mtDNA D-loop PCR amplicon showed similar mtDNA hybridization signals. SacI has a single restriction site in both mouse and rat mtDNA, resulting in a single band seen at 16.5 kilobase pairs. The blot showed that similar levels of mtDNA are maintained in the xenocybrid compared with the control.
Mitochondrial Protein Translation in the Rattus Xenocybrid-Mitochondrial translation products labeled in the presence of cycloheximide are shown after electrophoresis in Fig. 2. Most of the 13 proteins are identifiable by mobility together with characteristic intensity of labeling, as shown at left in Fig.  2. The profiles of the control cybrid and Rattus xenocybrid are similar overall. One mobility polymorphism between the M. musculus and Rattus products is evident, possibly the ND2 gene product NAD2, which was also seen in the parental NRK52E cells (not shown). Two proteins appear to be produced in increased amounts, possibly NAD4 and ATP6, while two others, likely NAD5 and cytochrome b, appear to be decreased. The same pattern was observed in two independent labeling experiments. The similarity of the profiles shows that transcription and translation of the foreign mtDNA is not markedly impaired.
Lactate Measurement- Fig. 3 shows the lactate production from the control and M. spretus cybrids, the Rattus cybrid, and the 0 cell line. A respiratory defect is signaled in the Rattus xenocybrid by the 10-fold greater lactate production compared with the control cybrid, whereas the M. spretus xenocybrid shows similar low lactate production to the control (Fig. 3). The 0 cell line produced 2-fold greater amounts of lactate than the Rattus xenocybrid, indicating that the cybrid was able to produce some ATP oxidatively. Fig. 4 shows typical polarograph traces of site I (glutamate ϩ malate) and site II (succinate) oxidation in the control and Rattus xenocybrid mitochondria. The control traces (A and B) show good coupling to ADP stimulation and high respiration rates and ADP/O ratios. In the Rattus xenocybrid (C and D),   NAD1  90  318  315  32  NAD2  74  345  345  88  COX I  97  514  514  14  COX II  99  227  227  3  ATP8  79  67  67  14  ATP6  95  226  226  12  COX III  97  261  261  9  NAD3  87  115  114  15  NAD4L  86  98  97  14  NAD4  87  459  459  59  NAD5  78  610  607  133  NAD6  80  172  172  34  Cytochrome b  93  380  381  26 site I-linked respiration is impaired but remains coupled to ADP availability, whereas site II-linked respiration shows a more severe defect, with little stimulation by added ADP. Comparison of the mean specific respiratory rates from three independent experiments (Table II) revealed that the M. spretus cybrid showed similar respiratory capacity to the control, whereas the Rattus cybrid state III (ADP-stimulated) rates were significantly lower than the control. This was true for site I substrates (31% mean control rates) and succinate (12% mean control rate). The ADP/O ratios obtained with site I substrates were also significantly lower in the Rattus cybrid (71% controls, Table II). These results are consistent with defects in either or both complex III or IV limiting respiratory chain electron flow, possibly combined with a complex I defect.

Polarographic Measurement of OXPHOS in Xenocybrids-
OXPHOS Enzymology- Table III summarizes the mean activity of OXPHOS complexes I, II, II ϩ III, and IV measured in triplicate using three independent mitochondrial isolates from each of the control, M. spretus and Rattus xenocybrids. No significant differences were found for the M. spretus xenocybrid compared with the control. The Rattus xenocybrid showed normal complex II activity with varying defects of the other complexes. Complex I (rotenone-sensitive activity) was significantly decreased to 46% of the control levels, whereas complex IV showed a partial but significant defect with 78% control activity (Table III). The most striking defect was indicated by the II ϩ III assay result, showing only 37% of the control activity (Table III). Together with the preserved complex II activity, this result suggests a severe complex III defect since this linked assay is normally rate-limited by complex II (26). DISCUSSION By introducing R. norvegicus mtDNA into a mouse mtDNAless cell line, we have produced a new mouse in vitro model of a severe mtDNA OXPHOS defect. Only a handful of mtDNA mouse cell mutants have been described, the best characterized being a point mutation in the 16 S rRNA gene resulting in resistance of the mitochondrial ribosome to chloramphenicol (CAP R ) (27) which results in multiple respiratory chain defects (10,28).
The lack of methods to produce targeted mtDNA knockouts has limited attempts to produce mtDNA mutant mice. A transfection approach using a DNA construct linked to a mitochondrial target peptide leader sequence is promising but has not yet succeeded in transforming mitochondria in cultured cells (29,30). Kenyon and Moraes (13) pioneered the approach of xenomitochondrial cybrids by showing that only the most closely related primate mtDNAs could be maintained in human 0 cells. These human/primate xenomitochondrial cybrids exhibited defects in respiratory chain complex I (14) and an increased sensitivity to programmed cell death when treated with complex I inhibitors (31).
The finding by Moraes et al. (32) that human cells preferentially replicate even defective human mtDNA over introduced normal primate mtDNAs showed that mtDNA determinants for trans-acting factors may be more important than OXPHOS functionality.
In the present studies we found that there is also a limit on mtDNA compatibility as indicated by the inability of the mouse 0 cells to maintain hamster mtDNA. Divergence of hamsters (Muridae subfamily Cicetidae) from the Murinae subfamily including mice and rats was around 16 million years ago ( Fig.  1 (22)). The Mus/Rattus divergence was around 10 million years ago ( Fig. 1 (23)) and may approximate the limit of compatibility in this system. The resulting xenocybrids exhibited a severe OXPHOS impairment (Table II and III and Figs. 3 and  4) despite showing preserved mtDNA replication and translation (Fig. 3). This shows the defects are consequent to the mismatches between the mouse nuclear OXPHOS subunits and the Rattus mtDNA subunits (Table I; see also accompanying article (33)). A marked defect of complex I was seen in the Rattus xenocybrids (Table III) but unlike the human/primate xenocybrids there was also a striking defect of complex III together with a partial defect of complex IV (Table III). This shows that predictions from the human/primate system do not translate directly into the murid system. The number of amino acid substitutions range from only 3 in the highly conserved COX-2 protein, to 133 in NAD5 (Table I), making it difficult to predict which changes are functionally important. The suggestion of a severe complex III defect is particularly interesting since cytochrome b is the only mtDNAencoded subunit of this complex. Cytochrome b shows 26 amino acid differences between M. musculus and R. norvegicus (Table  I). The polarographic results (Table II and Fig. 4) show that succinate-linked respiration is more severely impaired than site 1-linked respiration, despite the latter requiring the same electron transfer pathway through complexes III and IV. This All mitochondrial proteins appear to be produced in the xenocybrid, although ratios of some appear to differ in this long pulse (2 h) labeling experiment, as indicated by the arrows. The asterisk indicates a mitochondrial translation product showing a mobility polymorphism, probably ND2. For each lane, 10 g of cellular protein was loaded, and the 15% polyacrylamide mini-gel was electrophoresed under reducing conditions. CAP indicates the labeling of control cybrid cells in the presence of chloramphenicol in addition to cycloheximide.
FIG. 3. Lactate production in xenocybrids and 0 cells. A severe OXPHOS defect is signaled in the Rattus xenocybrid (f) by a 10-fold increased lactate production compared with the control and M. spretus cybrids (q). The LMEB3 0 cell line (OE), without oxidative ATP production, shows a further 2-fold increase, showing that the Rattus xenocybrid is producing some ATP aerobically. could result from substrate inhibition exacerbating the complex III defect, as there will be a greater steady state ubiquinol concentration during succinate oxidation compared with site 1-linked substrate oxidation due to the higher specific activity of complex II over complex I.
M. spretus and M. musculus diverged approximately 2-3 million years ago (Fig. 1). Although the present studies show no differences in OXPHOS capacity for the M. spretus xenocybrids, further studies may show subtle but important defects in this model. The levels of M. spretus mtDNA were too low in the mice produced by Irwin et al. (9) to conclude there were no pathogenic effects, and creation of this mouse would be an important proof of the principle from which to proceed to make other xenomitochondrial mice using the ES cell approach.
By making further xenomitochondrial constructs with species intermediate between M. spretus and Rattus, it should be possible to produce cybrids with intermediate OXPHOS phenotypes. By comparative studies of OXPHOS and mtDNA sequences from the species used, it should be possible to gain important new insights into the nuclear-mtDNA-encoded subunit interactions, in addition to the potential value of this system in producing mouse models of human mtDNA diseases. Evolution has provided a rich source of options. The Muridae is the most successful family of the order Rodentia, having the greatest number of extant species of any mammalian family (22,24). There are literally dozens of potential constructs that can be produced. The Gerbillinae subfamily of Muridae, the gerbils, are intermediate in divergence to Rattus and the hamsters (Fig. 1) and may produce viable xenocybrids. We would predict these constructs to produce severe OXPHOS defects if the mtDNAs could be replicated.