In Vivo Complementation of Complex I by the Yeast Ndi1 Enzyme

Recent studies suggest that dysfunction of the NADH-quinone oxidoreductase (complex I) is associated with a number of human diseases, including neurodegenerative disorders such as Parkinson disease. We have shown previously that the single subunit rotenone-insensitive NADH-quinone oxidoreductase (Ndi1) of Saccharomyces cerevisiae mitochondria can restore NADH oxidation in complex I-deficient mammalian cells. The Ndi1 enzyme is insensitive to complex I inhibitors such as rotenone and 1-methyl-4-phenylpyridinium ion, known as a metabolite of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP). To test the possible use of the NDI1 gene as a therapeutic agent in vivo, we chose a mouse model of Parkinson disease. The NDI1-recombinant adeno-associated virus particles (rAAV-NDI1) were injected unilaterally into the substantia nigra of mice. The animals were then subjected to treatment with MPTP. The degree of neurodegeneration in the nigrostriatal system was assessed immunohistochemically through the analysis of tyrosine hydroxylase and glial fibrillary acidic protein. It was evident that the substantia nigra neurons on the side used for injection of rAAV-NDI1 retained a high level of tyrosine hydroxylase-positive cells, and the ipsilateral striatum exhibited significantly less denervation than the contralateral striatum. Furthermore, striatal concentrations of dopamine and its metabolites in the hemisphere that received rAAV-NDI1 were substantially higher than those of the untreated hemisphere, reaching more than 50% of the normal levels. These results indicate that the expressed Ndi1 protein elicits resistance to MPTP-induced neuronal injury. The present study is the first successful demonstration of complementation of complex I by the Ndi1 enzyme in animals.

The mammalian proton-translocating NADH-quinone oxidoreductase (complex I) 4 is located in the inner mitochondrial membranes, is composed of 46 different subunits, and bears one FMN and eight ironsulfur clusters as cofactors (1)(2)(3). It has been known for many years that the structural and functional defects of this enzyme complex are involved in a number of human mitochondrial diseases (4 -6). Therefore, it is expected that a strategy to reestablish the function of complex I would lead to the treatment of human diseases caused by the defects of this enzyme complex. The respiratory chain of certain organisms in bacteria and fungi, but not in mammals, lacks the complex I-type enzyme. Instead, the functionality of complex I, namely oxidation of NADH and reduction of quinone, is performed by structurally simpler alternatives (collectively called NDH-2). Therefore, a simple question was asked: can the NDH-2-type enzyme be implemented in mammalian mitochondria, and can it supplement malfunctioning complex I? As a therapeutic method that works irrespective of the cause of complex I deficiencies, we have proposed using the Ndi1 protein, which is one of the NDH-2-type enzymes, found in the mitochondria of Saccharomyces cerevisiae (7)(8)(9)(10)(11)(12). The Ndi1 enzyme is composed of a single polypeptide of 53 kDa and contains noncovalently bound FAD and no iron-sulfur clusters. This enzyme is resistant to complex I inhibitors like rotenone and 1-methyl-4-phenylpyridium ion (MPP ϩ ) (9). The gene encoding the Ndi1 protein (NDI1) has been cloned and sequenced (13,14). By using a recombinant adeno-associated virus carrying the NDI1 gene that encodes the Ndi1 protein (rAAV-NDI1), we were able to express the Ndi1 protein in cultured mammalian cells. The expressed Ndi1 protein was correctly imported to the mitochondria of host cells with its own leading sequence (10,12). Recently, we have shown that the expressed Ndi1 enzyme is functionally active and is able to restore NADH oxidation in the mitochondria of complex I-deficient cells (8,11). In addition, NDI1-transduced cells can grow in the culture containing various complex I inhibitors (12).
A number of recent studies have indicated a reduced activity of complex I in Parkinson disease (PD) both in animal models and patients (15)(16)(17)(18). In fact, it has been suggested that environmental toxins known to inhibit complex I may be the primary cause of PD. Therefore, to further test the idea of using Ndi1 as a remedy for complex I defects (19), we chose an animal model of PD to carry out in vivo experiments. Among the animal models available to date, mouse models involving the neurotoxin MPTP have been widely used. It is hypothesized that MPTP is oxidized to MPP ϩ , which is then transported into dopaminergic neurons by the dopamine transporter. MPP ϩ is accumulated in the mitochondrial matrix and inhibits complex I. The damaged complex I appears to generate reactive oxygen species, which may induce death of neurons. Furthermore, in these models the activity of complex I is reported to be lowered (20), and therefore they are suitable as our test system. In this paper, we report that the Ndi1 protein expressed in the substantia nigra (SN) of mouse brain offered protection against the neurodegeneration caused by MPTP administration.
Animals-Eight-week-old male (22-25 g) C57BL/6 mice (obtained from our in-house breeding colony) were housed three to four per cage in temperature-controlled rooms under a 12-h light/dark cycle with free access to food and water. The housing and treatment of the animals were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and all procedures were approved by the Institutional Animal Care and Use Committee at The Scripps Research Institute.
Injection of AAV Vectors-For each surgical procedure described herein, mice received ketamine (50 mg/kg, intraperitoneally) before surgery. Anesthesia was induced with 3% isoflurane in O 2 flow, and mice were placed into a stereotaxic frame (Kopf Instruments, Tujunga, CA). Anesthesia was then maintained with 1% isoflurane in O 2 through a mouth tip fixed to the stereotaxic frame. All injections were made using a 5-l Hamilton microsyringe with a 30-gauge beveled needle. The anterior-posterior and medial-lateral stereotaxic coordinates for the striatum and the SN were calculated from bregma, and the dorso-ventral coordinates were calculated from the dural surface. A burr hole was drilled in the skull over the target site, and the microsyringe was inserted vertically into the SN at the following coordinates: anterior-posterior Ϫ3.3 mm, medial-lateral 1.5 mm, dorso-ventral Ϫ3.9 mm; and anteriorposterior Ϫ3.3 mm, medial-lateral 1.0 mm, dorso-ventral Ϫ4.1 mm. Mice received stereotaxic injections of rAAV-NDI1 (1 ϫ 10 11 IU/ml) or rAAV-GFP (1 ϫ 10 11 IU/ml) suspended in PBS containing 0.1% fluorescent microspheres (v/v; Polysciences Inc., Warrington, PA) into the SN (2 l/site, 4 l total) at a rate of 0.5-1.0 l/min. The microsyringe was left in place for an additional 5 min before retracting slowly. Animals that were used as controls received the same volume of solution, which contained 0.1% fluorescent microspheres (v/v) in PBS.
MPTP Treatment-MPTP handling and safety measures were in accordance with the Chemical Hygiene Plan developed at The Scripps Research Institute. Approximately 4 -5 months after the rAAV-NDI1 injection in the SN, mice were subjected to MPTP treatment. We followed the method reported by Sonsalla and Heikkila (22) for an acute MPTP mouse model. MPTP (0.1 ml in PBS; Sigma) was administered intraperitoneally at a dose of 15 mg/kg of body weight. A total of four injections were performed at 2-h intervals with a Ͼ90% survival rate. When the dose of MPTP was increased to 20 mg/kg, only 30 -50% of MPTP-treated mice survived as described in the literature (22,23). Lower doses tended to be less effective with regards to DA loss in the striatum. Therefore, a dose of 15 mg/kg seems to be most appropriate to obtain the desired effects by MPTP. Each group consisted of 12 mice to allow for statistically reliable analysis between groups. Seven days after the MPTP administration, brain tissue was collected and subjected to immunohistochemical analysis for the assessment of neuronal degeneration.
Immunohistochemical Analysis-For histological studies, mice were perfused with PBS followed by cold 4% (w/v) paraformaldehyde in PBS (pH 7.4), and brains were post-fixed for 1 h in the same buffer at 4°C. 30-m coronal sections were cut using a cryostat and were collected onto slides and stored at Ϫ20°C. Immunohistochemistry using antibodies against Ndi1 (1:250; prepared in our laboratory), tyrosine hydroxylase (TH) (1:500; Calbiochem), and glial fibrillary acidic protein (GFAP) (1:250; Sigma) was carried out on slide sections as follows. After fixation with 4% paraformaldehyde/PBS at room temperature for 20 min, the sections were washed three times with PBS and incubated in 3% hydrogen peroxide (H 2 O 2 ) for 30 min to quench endogenous peroxidases followed by permeabilization and blocking for nonspecific binding with 10% normal goat serum, 5% horse serum, and 0.1% Triton X-100/PBS at room temperature for 1 h. Sections were then incubated in primary antibodies overnight at 4°C. For TH and GFAP, sections were incubated for 1 h with biotinylated secondary antibody (1:200; Jackson ImmunoResearch Laboratories) followed by three washes for 10 min each with PBS at room temperature. Secondary antibody was visualized using the ABC Elite kit (Vector Laboratories, Burlingame, CA) and 3,3Ј-diaminobenzidine tetrachloride (Sigma). In all staining procedures, deletion of the primary antibody served as a control. The immunostained sections were examined using an epifluorescence or bright-field microscope. Ndi1 protein staining was done using the tyramide signal amplification-direct procedure following the manufacturer's instructions (PerkinElmer Life Sciences). Briefly, the sections were incubated with the primary antibodies as described above and then incubated with horseradish peroxidase-conjugated goat anti-rabbit IgG (1:1000; Calbiochem) at room temperature for 2 h. The sections were then washed three times for 10 min each with PBS at room temperature and incubated with the fluorophore tyramide (1:100 dilution with amplification buffer) for 7 min at room temperature in a dark, moist chamber followed by three washes for 10 min each in PBS at room temperature and examined using a fluorescence microscope. To stain nucleic acid, the sections were treated with TO-PRO-3 iodide (1:750 dilution with amplification buffer) for 10 min at room temperature and washed three times for 5 min each in PBS.
NADH Activity Staining-Histochemical staining for NADH dehydrogenase activity was based on the NADH-tetrazolium reductase reaction (24). Tissue sections were incubated for 10 min at room temperature in NADH-tetrazolium reductase reaction solution (0.2 M Tris-HCl, pH 7.4, 1.5 mM NADH, and 1.5 mM nitro blue tetrazolium) and then washed three times with deionized water.
HPLC Analysis-Striatal DA and its metabolite levels were determined by HPLC. The mice were sacrificed, and brains were quickly removed and placed in brain matrix (RBM 2000C; Activational Systems Inc.) on dry ice and sliced with a razor blade to 1-2-mm sections. Striatal regions from each side of the brain were isolated separately and weighed. Each sample was briefly sonicated in 4 volumes of ice-cold 0.4 M perchloric acid containing 0.15% sodium metabisulfite and 0.05% disodium EDTA and deproteinized by centrifugation at 13,000 ϫ g for 20 min at 4°C. Aromatic amines and their metabolites were separated using ion-paired reversed phase HPLC coupled with electrochemical detection according to Wagner et al. (25). Samples (10 l) kept at 4°C were injected into an HPLC system equipped with an A-314 C 18 column (5 m, 300 ϫ 6.0 mm; YMC). The flow rate was 0.55 ml/min through a pump (LC10AD; Shimadzu) at a pressure of 100 -110 kg/cm 2 at 30°C. The mobile phase was composed of 75 mM sodium phosphate, 2.78 mM sodium octyl sulfate, 33 mM triethylamine, and 0.1 mM EDTA with the final pH adjusted to 3.43 prior to adding 25% (v/v) methanol. The signal was detected on a graphite carbon working electrode set at ϩ0.75 V (against the Ag/AgCl reference) (ECD-300; Eicom). The data were collected through an EPC-500 processor (Eicom), and peak areas were calculated using the PowerChrom software and quantified from a standard curve. Linearity was maintained from 10 fmol to 100 pmol for standards.
Materials-MPTP, 3,3Ј-diaminobenzidine tetrachloride, NADH, and Triton X-100 were purchased from Sigma; anti-TH and horseradish peroxidase were from Calbiochem; 4Ј,6Ј-diamidino-2-phenylindolecontaining mounting medium and ABC Elite kit were from Vector Laboratories; fluorescent microspheres were from Polysciences Inc.; the tyramide signal amplification kit was from PerkinElmer Life Sciences; nitro blue tetrazolium chloride monohydrate was from Acros Organics (Morris Plains, NJ); Permount mounting medium was from Fisher. The OCT compound was from Sakura (Torrance, CA).

Expression of the GFP in Mouse
Brains-For a therapeutic agent to be useful, it is necessary to deliver the agent to the dopaminergic neurons in the SN. To test the efficacy of the virus in mouse brains, we used rAAV particles carrying the GFP gene (rAAV-GFP). The rAAV-GFP particles were injected unilaterally into the SN region of mice. After 6 -8 weeks, a good level of GFP fluorescence was observed in the SN of the side of the brain where rAAV-GFP had been injected (Fig. 1). Furthermore, sections at the level of striatum also exhibited GFP fluorescence on the ipsilateral side but not on the contralateral side, indicating that the expressed protein was transported anterogradely to nerve terminals. It is therefore likely that the rAAV particles that were constructed could efficiently deliver the transgene in dopaminergic neurons of the mouse SN.
Functional Expression of the Ndi1 Protein in the Mouse SN-The unilateral injection of rAAV-NDI1 particles into the SN of the brain was carried out using the same method as for rAAV-GFP (see above). As shown in Fig. 2, the Ndi1 protein was clearly detected in the SN. It was determined that the injection of rAAV-NDI1 into a slightly dorsal and posterior region of the SN results in a more effective expression of Ndi1 in dopaminergic neurons than a direct injection into the SN. To ascertain that the expressed Ndi1 is functioning, the NADH dehydrogenase activity assay was carried out on an adjacent section using NADH and tetrazolium dye as the substrates. A number of blue deposits were clearly seen in the areas where the Ndi1 expression was distinct (Fig. 2). The opposite side gave a low level of color development that was presumably ascribed to complex I. These results strongly suggest that the Ndi1 enzyme expressed in the mouse SN is biologically active. The expression of the Ndi1 protein in mouse brain was observed for at least 7 months, which was expected because of the long term nature of AAVbased gene delivery. One potential concern may arise from the fact that Ndi1 substitution reduces the P/O ratio from 3 to 2 because the Ndi1 enzyme is not a proton pump (9). However, it should be noted that the animals that received rAAV-NDI1 did not show apparent health problems.
MPTP Treatment and Immunohistochemical Assessment of Neurodegeneration-After having established experimental conditions for the Ndi1 expression in the SN, experiments were carried out that critically tested whether Ndi1 had protective effects against MPTP-induced neurodegeneration. The conditions for MPTP treatment were worked out by following the method reported in the literature (22). It turned out that four injections of MPTP at a dose of 15 mg/kg were most appropriate to achieve the desired level of degeneration in a reproducible manner keeping the survival rate high. Brain sections at the level of the SN and of the striatum were collected, and dopaminergic cells were visualized immunohistochemically using the antibody against TH. Typical results of the TH staining are presented in Fig. 3A. As expected, MPTPtreated mice showed a significant loss of the TH immunoreactivity in the SN and denervation in the striatum. The decreased level of striatal TH was retained for at least a week. Similar results have been reported by Sugama et al. (26) and Przedborski and colleagues (27, 28) by using the same MPTP injection procedure (15 mg/kg/dose and a total of four injections at 2-h intervals).
In the following experiment, 13 2-month-old mice received unilateral injections of the rAAV-NDI1 particles in the left SN region. About 5-6 months after the rAAV-NDI1 injection, MPTP was administered to mice as described above. Brains were collected 7 days after the MPTP treatment and were evaluated for the degree of neurodegeneration. As shown in Fig. 3B, the SN of the side where rAAV-NDI1 had been injected exhibited a higher number of TH-positive neurons with clearly visible processes. In contrast, the TH reactivity of the untreated side greatly diminished because of MPTP treatment. Moreover, TH staining of the ipsilateral striatum from the same brain was significantly stronger than that of the contralateral striatum. In addition to the TH reactivity, we examined the tissue for GFAP-immunopositive cells. GFAP is another marker commonly used for assessing central nervous system injury. In contrast to TH, staining of the GFAP in the striatum was denser on the contralateral side than the ipsilateral side, indicating less damage of neuronal terminals in the hemisphere that had received rAAV-NDI1. These results strongly suggest that the NDI1 gene delivered to the SN confers protection against nigral degeneration by compensating for complex I malfunction.
Because the rAAV-NDI1 injection was performed unilaterally, the untreated side served as a control. This allows direct comparison between the two sides within each mouse, eliminating possible deviations that may arise from variations among the individuals in the effectiveness of MPTP treatment and virus injection. Of the 13 mice tested, 8 mice clearly exhibited the protective effect on the side of the rAAV-NDI1 injection and 2 mice to a lesser extent. Three mice showed the same degree of degeneration on both sides, but none had more neuronal damage on the rAAV-NDI1-treated side. Images of striatal sections from mice that had a positive response were subjected to statistical analysis for staining intensity (Fig. 3C). In both TH-and GFAP-stained sections, a clear difference was observed between the ipsilateral side and the contralateral side. The same images were further analyzed by image processing that allows visualization of the difference. Fig. 4A shows example images from four representative mice. As expected, the ipsilateral side exhibited a distinctively higher TH density than the opposite side. The degree of difference in the density between the two sides varied slightly among the mice presumably because of some variation in the expression of Ndi1 in the SN. Images on the right-hand side of Fig.  4A represent GFAP staining of the striatal sections from the same brain as that shown for TH staining on the left. The contralateral side consistently exhibited a higher degree of damage. Again, statistical analysis confirmed the protective effect of the Ndi1 enzyme on the ipsilateral striatum (Fig. 4B).
MPTP Treatment and Levels of DA and Its Derivatives in the Striatum-In the animal model employed, MPTP administration can effectively cause rapid and drastic depletion of striatal DA. To confirm the immunochemical data, we have determined the concentrations of DA and its metabolites in the striatum. The procedures for rAAV-NDI1 injections and MPTP treatment were the same as those described for the histochemistry experiments except that the striata from the two hemispheres were sampled separately. As shown in Fig. 5, the concentration of DA in the striatum of the MPTPtreated mice was Ͻ10% of that of the control mice. Similarly, both 3,4-dihydroxyphenylacetic acid and homovanillic acid significantly diminished in the MPTP-treated group. In the group of mice that had received rAAV-NDI1, there was a clear difference between the two hemispheres. All values of the ipsilateral side were 2-4 times greater than those of the contralateral side and reached ϳ60% of the normal level. The observed protective effect by rAAV-NDI1 is in total agreement with the immunochemical results.

DISCUSSION
Gene replacement strategies have been explored for many years as an approach to correcting mtDNA-related disorders (29,30). The first suc-  (22), which describes an acute MPTP mouse model of PD. Brain samples were examined 7 days after MPTP administration as detailed under "Experimental Procedures." A substantial reduction of TH reactivity in the SN and striatum was obvious in the MPTP-treated mice which is in agreement with the literature. B, assessment of neurodegeneration in the rAAV-NDI1-injected mouse. The rAAV-NDI1 particles (1.0 ϫ 10 11 IU/ml, 2 l/site, 4 l total) were injected unilaterally into the left SN. Six months later, mice were subjected to MPTP treatment as in A. Brain sections at the level of SN and striatum were immunohistochemically stained for TH. Staining of the striatum for GFAP, a marker of central nervous system injury, was also carried out. Protection of TH-positive cells by rAAV-NDI1 is clearly seen in the SN section. In addition, the striatal section exhibited a higher TH level on the ipsilateral side than on the contralateral side. Similarly, the ipsilateral striatum showed less damage as revealed by GFAP staining. C, comparison by statistical analysis of staining density of the striatum between the ipsilateral side and the contralateral side in animals that received rAAV-NDI1 in the left SN prior to MPTP administration. For each TH and GFAP staining, a total of 28 images of striatal sections as exemplified in B were collected, and the mean density of the striatum region was evaluated using ImageJ software (37). The density value ranged from 0 (white) to 255 (black). Results given are means Ϯ S.D. ipsi, ipsilateral side; contra, contralateral side; *, p Ͻ 0.001. MAY 19, 2006 • VOLUME 281 • NUMBER 20 cessful work reports using a yeast system in which subunit 8 of the ATP synthase is replaced by an artificial nuclear gene designed to code for this subunit (31). More recently, allotopic expression is achieved in human cell lines for two mitochondrially encoded subunits, which are able to partially correct respiratory defects caused by pathogenic mtDNA mutations (32,33). Our earlier experiments for restoring NADH oxidation by the NDI1 gene were performed using complex I-deficient cell lines irrespective of defects of mtDNA-and nuclear DNA-encoded subunits. These achievements were limited to tissue culture cell lines or yeast. The present study is the first realization of such attempts using animals. The Ndi1 enzyme expressed in dopaminergic neurons of mouse brains elicits resistance to neuronal injury caused by MPTP, clearly indicating the complementation of complex I by the Ndi1 protein. Use of the Ndi1 protein has advantages over the allotopic strategy. First, the Ndi1 polypeptide is not as extremely hydrophobic as most of the mtDNA-encoded subunits, which may alleviate the problem of aggregation when it is expressed in the cytoplasm. Second, because Ndi1 complements functionality of complex I as a whole, issues such as proper integration of the expressed subunit and displacement of the defective subunit are not relevant. In addition, appropriate animal models of defects of complex I subunits are not available at the present time. The MPTP treatment, on the other hand, provided a good model in which to test the Ndi1 strategy.

In Vivo Complementation of Complex I
Although PD is not recognized as a mitochondrial disorder, reduced activity of complex I is no doubt involved in the early stage of the cellular events that would eventually trigger neuronal cell death. Consequences of complex I inhibition include bioenergetic disturbance such as a lowered level of cellular ATP, or more likely, an elevated production of reactive oxygen species. The principle of our strategy is to bypass complex I and to reestablish electron transfer from NADH to quinone by introducing the Ndi1 enzyme. It is apparent that this approach is effective in the MPTP mouse model tested, and the findings establish a key role of complex I in the neurodegenerative process.
The protective effect reported here is obviously not optimal. First, we deliberately carried out unilateral injections of rAAV-NDI1 to obtain unequivocal assessment in each individual animal. We expect improved protection with bilateral injections, which are currently being assessed. Second, it may be necessary to refine the conditions for NDI1 delivery in the SN. Factors such as infectious titer of viral particles can be adjusted for better efficacy, and/or the use of AAV of different serotype (i.e. type 5) might boost the level of transgene expression (34). Third, we have thus far employed an acute model of PD, and the next step could be to examine chronic animal models that are currently available for mouse and rat (17,23,35).
It would be of interest to cause the degeneration in animals first and then inject the rescue gene. This set of experiments should mimic the real situation in which the protective treatment can only be applied after the damage to the nigrostriatal system has begun. At the present time,  Fig. 3C were processed through the threshold tool of the ImageJ software, and pixels with brightness values between a lower and a higher threshold are displayed in red. The upper threshold was selected in such a way that the difference between the two hemispheres became most distinct. The lower threshold was set to eliminate high levels of background that were mostly because of tearing in the brain sections. Pixels below the lower threshold are displayed in black on the images. A, sections from four representative mice. The ipsilateral side and the contralateral side of the striatum are marked by filled and open arrows, respectively. ipsi, ipsilateral side; contra, contralateral side. B, statistical analysis based on the processed images. The area in red in the striatum was used as the index for the staining intensity. Results given are means Ϯ S.D. *, p Ͻ 0.001.  Fig. 3. Striatal regions from each side of the brain were isolated separately, and DA and its metabolite levels were determined by HPLC. As expected, the striatal contents of DA and its derivatives in the MPTP-treated mice (gray) were greatly diminished compared with the control mice (black). In the group of mice that received rAAV-NDI1 injections in the SN prior to MPTP treatment, the ipsilateral striatum exhibited markedly higher values than the contralateral side. The values of 100% for DA, 3,4-dihydroxyphenylacetic acid (DOPAC), and homovanillic acid (HVA, 4-hydroxy-3-methoxyphenylacetic acid) are 66.1, 6.82, and 3.86 pmol/mg tissue. Results given are mean Ϯ S.D. (n ϭ 6). *, p Ͻ 0.005. rodent models that resemble human PD more in terms of slow development are not available. However, in monkeys, parkinsonian features become manifest upon a chronic administration of MPTP over the period of several months and can be maintained for 2 months (36). This model may provide us with a long term, progressive state of neuronal deficiencies in which to test the potency of rAAV-NDI1.
The single subunit Ndi1 is capable of substituting for the mitochondrial complex I and thus might become a versatile tool in broader applications. The approach employed in this study may open a new avenue toward understanding disorders in which complex I is involved.