Fatal breathing dysfunction in a mouse model of Leigh syndrome

Leigh syndrome (LS) is a subacute necrotizing encephalomyelopathy with gliosis in several brain regions that usually results in infantile death. Loss of murine Ndufs4, which encodes NADH dehydrogenase (ubiquinone) iron-sulfur protein 4, results in compromised activity of mitochondrial complex I as well as progressive neurodegenerative and behavioral changes that resemble LS. Here, we report the development of breathing abnormalities in a murine model of LS. Magnetic resonance imaging revealed hyperintense bilateral lesions in the dorsal brain stem vestibular nucleus (VN) and cerebellum of severely affected mice. The mutant mice manifested a progressive increase in apnea and had aberrant responses to hypoxia. Electrophysiological recordings within the ventral brain stem pre-Bötzinger respiratory complex were also abnormal. Selective inactivation of Ndufs4 in the VN, one of the principle sites of gliosis, also led to breathing abnormalities and premature death. Conversely, Ndufs4 restoration in the VN corrected breathing deficits and prolonged the life span of knockout mice. These data demonstrate that mitochondrial dysfunction within the VN results in aberrant regulation of respiration and contributes to the lethality of Ndufs4-knockout mice.

Leigh syndrome (LS), a progressive neurodegenerative disorder (1), is the most common infantile mitochondrial disorder, affecting 1 in 40,000 live births (2). MRI reveals bilateral hyperintense lesions in many brain regions of LS patients (3, 4) that correlate with gliosis, demyelination, capillary proliferation, and/or necrosis (5). Behavioral symptoms of LS patients vary widely, but can include developmental retardation, hypotonia, ataxia, dystonia, optic atrophy, hearing impairment, breathing abnormalities, dysarthria, swallowing difficulties, and failure to thrive (3, 6, 7). Respiratory arrest, the most common cause of death for LS patients, is observed in up to 75% of cases (8, 9). The necessity for mechanical ventilation or the occurrence of sudden death of LS patients is associated with brain stem lesions (3, 10). However, case report data are usually scarce, and the mechanisms involved in the respiratory apnea in LS cases are unclear. The lack of a genetic model to study disease progression and the cause of death has impeded a rigorous investigation.

The most prevalent mitochondrial disorders affect complex I (11–13), and complex I deficiencies are detected in many LS patients (4, 14, 15). Mutations in the nuclear gene encoding in NDUFS4 (NADH dehydrogenase [ubiquinone] iron-sulfur protein 4), one of the 45 subunits of complex I, have been described as causing LS in humans (16–20). Human LS patients harboring NDUFS4 mutations develop brain stem and basal ganglia lesions and die from cardiorespiratory failure at an early age, with marked apneic episodes in the majority of cases along with a minor presence of hypertrophic cardiomyopathy (21). We developed a line of mice in which the Ndufs4 gene can be inactivated by Cre recombinase (22). Here, we delineate the cause of death of Ndufs4-deficient mice and develop a gene therapy strategy.

KO mice develop symptoms similar to those of humans with NDUFS4 mutations. Mice with both alleles of the Ndufs4 gene inactivated in all cells or just in the CNS manifest a fatal progressive encephalopathy and behavioral characteristics similar to those of people with LS, especially those with NDUFS4 mutations (21–23). The similarities include failure to thrive, growth retardation, ataxia, hypotonia, visual problems, and breathing irregularities (Table 1). Brain lesions detected by imaging and histology are also similar (24).

Table 1

Clinical observations in LS and Ndufs4-KO mice

Brain lesions in KO mice are detected by MRI and gliosis. T2-weighted MRI reveals edematous regions as a hyperintense signal that is a primary diagnostic sign for LS. Of 5 KO mice examined, 2 mid-stage and 2 late-stage mice (as defined in the methods section) had hyperintense lesions in the deep cerebellar fastigial nucleus (FN), cerebellar lobes (predominantly posterior lobes such as VIII, IX, X, but extending to anterior lobes in late-stage animals), the external plexiform layer of the olfactory bulb (OB), and the dorsal medulla in either sagittal or coronal views that are not seen in control mice (Figure 1A). Lesions in the dorsal medulla appear to be in the vestibular nucleus (VN) (Figure 1B). Gliosis develops progressively in these same brain regions, as revealed by immunostaining with antibodies that detect glial reactivity. For example, immunostaining for glial markers reveals extensive bilateral neuroinflammation in the VN of KO mice (Figure 1C). Neurons with cytoplasmic vacuoles and aberrant mitochondria containing condensed cristae as well as microglia engorged with cytoplasmic remnants were observed at the ultrastructural level in the same brain regions (23). The severe encephalopathy leads to a shortened life span with greater than 90% mortality by P50 (Figure 1D).

Figure 1

Alterations in MRI and histopathology in KO mice. (A) T2-weighted in vivo MRI of control and late-stage KO mouse. Sagittal MRIs (top panels) of control (CT) mouse and late-stage KO mouse show hyperintense regions (white areas) in lateral ventricles of control mouse, whereas KO mouse shows distinct lesions in exterior plexus of OB and dorsal surface of brain stem with extensive lesions of cerebellum (white arrows) together with a hypointense alteration in the midbrain (asterisk). Coronal imaging of control and KO mice shows only moderate T2-weighted MRI intensity in fourth ventricular region of control mouse whereas a mid-stage KO mouse (bottom, right) had significant lesions of the dorsal medulla (VN, white arrow) and tenth cerebellar lobe (black arrow). (B) Diagram of the VN according to Franklin and Paxinos (60). VN is circled. 4v, fourth ventricle. (C) Staining for Iba1 (microglial cell marker) and GFAP (astrocyte marker) in the VN of control (top left) and KO (top right) mice shows presence of symmetrical lesions filled with abundant microglial cells and surrounded by GFAP-positive reactive astrocytes in KO compared with control mice. Scale bar: 1,000 μm. (Bottom row) Close-up images of the VN (indicated by asterisks in C) show increased GFAP and Iba1 intensity and morphological changes in glial cells in KO mice. Scale bar: 250 μm. (D) Survival curve of KO mice (n = 101).

KO mice have abnormal respiratory responses. KO mice are small and manifest hypothermia, ataxia, lethargy, and sensory abnormalities (22, 23). The cerebellar and VN lesions could easily account for the ataxia, but the underlying cause of the other symptoms and death remained enigmatic until we observed intermittent breathing irregularities. KO mice lose most of their hair at approximately P20, which facilitates observing breathing irregularities that include hypo- or hyperventilation and gasping (Supplemental Video 1; supplemental material available online with this article; doi: 10.1172/JCI62923DS1). The appearance of gasping-like episodes was commonly observed under more stressful conditions, for example, when mice were handled.

To delineate respiratory responses of KO mice, we used barometric, whole-body, flow plethysmography to record the breathing of unanesthetized, freely moving KO and control mice (Figure 2A). Under normoxic conditions (20.9% O₂), neonate (P8–P12) and older (>P30) KO mice had normal respiratory frequency (fR); however, they had significantly higher tidal volume (V_T) and minute ventilation (V_E = fR × V_T) compared with control mice of the same age (Figure 2, B and C). KO mice had intermittent episodes of apnea (cycle time [T_TOT] exceeding twice the length of the mean) and periods during which fR was very irregular, as measured by the irregularity score (IS) (24). These abnormalities increased with age (r² = 0.50; P = 0.02 for the number of apneas and for IS, r² = 0.64; P < 0.01; Figure 2, D and E). In response to hypoxia (5 minutes at 10% O₂), there was an initial augmentation of breathing followed by a depression. Both neonate and older KO mice had equivalent augmentation of breathing, but much more severe depression in V_T and V_E compared with control mice (Figure 2F). Although variable, some KO mice also had an abnormal response to excess CO₂ (hypercapnia; 5 minutes at 5% CO₂). Six out of 13 (neonate and adult) KO mice failed to show the hyperventilation response that was observed in all 6 control mice. Interestingly, the remaining 7 KO mice had an exaggerated response compared with control mice (Supplemental Table 1).

Figure 2

Abnormal breathing patterns of KO mice. (A) A 30-second breathing record of freely moving, 46-day-old mice recorded by barometric, whole-body, flow plethysmography reveals breathing irregularity and apneas in the KO mouse (bottom trace) compared with the control littermate (top trace). Upward deflections in traces represent inspiratory movements. (B and C) Breathing parameters of neonate (B) and adult (C) control and KO mice. The V_T and V_E are significantly higher in KO mice than in control mice. *P < 0.05; **P < 0.01, Student’s t test. (D and E) Development of apneas (D) and irregularity of the breathing frequency (E) of control and KO mice were significantly correlated with age. (F) Changes in breathing of neonate control and KO mice exposed to 5 minutes of hypoxia (10% O₂; gray area). The respiratory depression of KO mice observed at the fifth minute of exposure was significantly greater than that of control mice, *P < 0.01, 2-way ANOVA, Bonferroni’s post-hoc test. Data are shown as mean ± SEM.

We also measured breathing rate, heart rate, and arterial oxygen saturation in control and KO mice at various stages of disease progression by optical spectrometry in restrained animals. As with the plethysmography, breathing rate was variable, but with this method it was significantly lower in KO mice than in control mice (∼150 vs. ∼220 breaths/min, KO vs. control; P < 0.001). This discrepancy between methods may be explained by the stress induced by restraint. Furthermore, the reduction in breathing rate was more evident in later stages of the disease (Figure 3, A and B, P < 0.05). Likewise, heart rate for KO mice was lower than for control mice (∼500 vs. ∼650 beats/min, KO vs. control; P < 0.01), especially in late stages of the disease (Figure 3, C and D, P < 0.05). Consistent with the lower breathing and heart rates, the percentage of saturation of arterial blood in late-stage KO mice was often less than 99%, which was not observed in control mice (∼95% vs. ∼99%, KO vs. control, P < 0.05; Figure 3, E and F).

Figure 3

Reduced breathing, heart rate, and oxygen saturation in KO mice. Measurements were made using a pulse oximeter attached to the bare necks of control (n = 7) and KO mice at various stages of disease progression (total n = 11; n = 2, 3, and 6 mice, early, mid-stage, and late-stage respectively) as described in Methods. For KO mice, there were often significant fluctuations of the values over time; average rates during the measurement period are plotted. (A) Breathing rate was lower in mid- and late-stage KO mice. *P < 0.05, 1-way ANOVA, Tukey’s post-test. (B) Representative traces of control and early- and late-stage KO mice (C) Heart rate at all stages of disease progression was significantly less than that of controls. *P < 0.05, 1-way ANOVA, Tukey’s post-test. (D) Representative traces of control and early- and late-stage KO mice. (E) Arterial blood oxygen saturation became variable and significantly lower in late-stage KO mice. (F) Representative traces of control and early- and late-stage KO mice. Data points represent individual mice. Data are shown as mean ± SEM.

Abnormal responses of the pre-Bötzinger complex in KO mice. The pre-Bötzinger complex (preBötC) in the ventral medulla is critical for respiratory rhythm generation (25, 26) and may play a role in the hypoxic response (27, 28). Extracellular recordings of brain stem slices containing the preBötC were obtained from P8–P10 KO mice and their control littermates. Under control conditions (95% O₂, 5% CO₂), these slices revealed no significant differences in frequency and regularity of bursting activity. In response to hypoxia (95% N₂, 5% CO₂), the isolated preBötC generates an initial augmentation phase characterized by increased bursting frequency followed by a depression during which bursting frequency returns to, or falls below, baseline activity. During this response, the inspiratory burst discharge switches from “fictive eupnea” to “fictive gasping” (29). The initial augmentation phase in slices from KO mice did not differ from that of control littermates (Figure 4, A–F), but during depression, the amplitude of fictive gasping was significantly decreased in KO mice compared with controls (Figure 4, A–F). The frequency of fictive gasping was not statistically different between control and KO mice, but while 100% of slices from control mice continued fictive gasping throughout the 10 minutes of hypoxia, 30% of slices from KO mice stopped fictive gasping.

Figure 4

Abnormal extra- and intracellular recordings of PreBötC of KO mice. (A–E) Simultaneous intracellular whole cell (lower trace, whole-cell voltage; Vm) and multi-unit population recordings (upper trace, integrated and rectified multi-unit recording; VRG) containing PreBötC from brain stem slices of KO and control mice in response to hypoxic conditions (95% N₂/5% CO₂, A). Intracellular recordings during fictive eupnea revealed a lower number of APs/burst (B) and a lower AP frequency (D) of KO mice (n = 8) compared with control mice (n = 5). In hypoxic conditions, control mice decreased to 56.96% ± 47.04% of baseline (C, n = 5), while KO mice (E, n = 8) stopped firing APs by the end of the 10-minute hypoxic exposure. (F) Population recordings of KO mice (n = 10) showed significantly (P < 0.05) reduced fictive gasping compared with control mice (n = 10). (G) Resting potential of inspiratory cells was not significantly different in slices from KO mice compared with control mice. Data points represent individual cells. (H) Example of a drive potential (upper trace, whole cell voltage; Vm) during a population burst (lower trace, integrated and rectified multi-unit recording of the VRG). (I) The depolarization in phase with the network burst was significantly reduced in slices from KO mice (5.51 ± 0.59 mV, n = 7) when compared with the control mice (10.13 ± 1.30 mV, n = 5, P = 0.005) under normoxic conditions. In addition, the KO mice had a severe reduction of the underlying depolarization when challenged by hypoxia (from 5.51 ± 0.59 to 0.52 ± 1.36 mV, n = 7, P = 0.0012), while the control mice did not show a significant reduction (10.13 ± 1.30 mV to 6.09 ± 6.43 mV). *P < 0.01. Data are shown as mean ± SEM.

Extracellularly recorded population activity does not provide insight into how the amplitude was determined. Hence, we characterized the hypoxic response using intracellular recordings of preBötC inspiratory neurons in slices from KO and control mice. The cellular firing in phase with the extracellular burst was measured, and the number and frequency of action potentials per burst (APs/burst) was quantified. Under control conditions, the resting potential was identical between control and KO mice; however, the depolarization in phase with the network burst was significantly reduced in KO mice (5.51 ± 0.59 mV, n = 7) compared with controls (10.13 ± 1.30 mV, n = 5, p<0.01, Figure 4, G–I). AP/burst and firing frequency of KO mice were significantly lower than in control littermates (Figure 4, B and D). Upon transition from control to hypoxic conditions, the changes in resting potential in control and KO mice were similar. However, KO mice had a severe reduction of the underlying depolarization when challenged by hypoxia (from 5.51 ± 0.59 to 0.52 ± 1.36 mV; n = 7, P < 0.01), while control mice did not show a significant reduction (10.13 ± 1.30 to 6.09 ± 6.43 mV, Figure 4, G–I). The AP/burst of KO mice dropped dramatically, and 2 out of 7 of the slices stopped gasping after more than 5 minutes of hypoxia, while in control slices, a relatively mild reduction of AP/burst was observed, and all of the control slices maintained gasping (Figure 4, C and E). Interestingly, the lack of gasping observed in KO slices in hypoxic conditions was restored after bath application of tolbutamide, an ATP-sensitive potassium channel (K_ATP) inhibitor (Supplemental Figure 1), suggesting an intrinsic deficit in the physiology of preBötC neurons, which may contribute to the breathing alterations in KO mice.

Although respiratory impairment is one of the clinical signs of KO mice, microglial activation was rarely observed in any of the classical respiratory centers, such as the medial parabrachial nucleus, nucleus of the solitary tract, or preBötC. However, in late-stage KO mice, microglial activation was observed around serotonergic cell bodies in the medullary raphe (raphe pallidus and obscurus), a brain region important for respiratory control (30). In some cases, microglial activation extended to their projections, accompanied by mild microgliosis in the preBötC that was identified by neurokinin 1 receptor immunolabeling (Supplemental Figure 2, A–D), suggesting that the alterations in these neurons was not sufficient to elicit a severe secondary inflammatory response. No degenerative changes were detected in the chemoreceptive Phox2B-immunoreactive neurons of the retrotrapezoid nucleus (31, 32).

Selective inactivation of Ndusf4 in the VN induces neurodegeneration and breathing abnormalities and increases mortality. The preBötC is essential for breathing in vivo (25), and it is influenced by a large number of neural inputs including those from the VN (see Discussion). Because the VN is a prominent site of gliosis and neurodegeneration in KO mice, we selectively inactivated the Ndufs4 gene only in that nucleus by bilaterally injecting adult Ndufs4^lox/lox mice with adenoassociated virus (type 1) expressing Cre-GFP (AAV1-Cre-GFP) to generate AAV-VN–KO mice. As a control, Ndufs4^lox/lox mice were injected with an AAV1 encoding a nonfunctional Cre (AAV1-CreΔ-GFP) to generate AAV-VN–CT mice. Extensive GFP expression was observed within the VN after bilateral viral transduction (Figure 5A). Furthermore, AAV-VN–KO mice developed extensive microglial activation 1 month after surgery, assessed by staining of the Iba1 marker (Figure 5A). Iba1 staining was dependent upon loss of Ndufs4 expression because no remarkable microglial activation was observed in the VN of AAV-VN–CT mice (Figure 5A).

Figure 5

Histopathological and behavioral alterations after Ndufs4 inactivation in the VN. (A) (Left panel) Representative image of the degree of transduction (GFP-positive cells) after vector delivery in the VN. (Right panels) Iba1 staining (red) after injection of a nonfunctional Cre-expressing (CREΔ-GFP, AAV-VN–CT mice, top right) or a functional Cre-expressing (CRE-GFP, AAV-VN–KO mice, bottom right) AAV1 vector in the VN of Ndufs4^lox/lox mice. Scale bars: 500 μm (left panel); 250 μm (right panels). (B) Survival curve in AAV-VN–CT (n = 17) and AAV-VN–KO mice (n = 21) after VN injection. *P < 0.05, Gehan-Breslow-Wilcoxon Test versus AAV-VN–CT. (C) Percentage of body weight change after injection in AAV-VN–CT (n = 10) and AAV-VN–KO (n = 15) mice. Day 0 is considered baseline (BL) (100%). ***P < 0.001, genotype, 2-way ANOVA. (D) Change in rotarod performance after injection in AAV-VN–CT (n = 6) and AAV-VN–KO (n = 7) mice. *P < 0.05; **P < 0.01, 2-way ANOVA, Bonferroni’s post-hoc test versus AAV-VN–CT. (E) Changes in breathing of AAV-VN–CT (n = 14) and AAV-VN–KO (n = 6) mice exposed to 5-minute hypoxia (10% O₂; gray area). (F) Changes in breathing of AAV-VN–CT (n = 14) and AAV-VN–KO (n = 6) mice exposed to 5-minute hypercapnia (5% CO₂; gray area). *P < 0.05; **P < 0.01 versus AAV-VN–CT; 2-way ANOVA, Bonferroni’s post-hoc test. Data are shown as mean ± SEM.

Inactivation of Ndufs4 bilaterally within the VN led to weight loss (Figure 5C), motor coordination deficits as assessed by rotarod testing (Figure 5D), and increased mortality (Figure 5B). As the behavior of the mice deteriorated, we observed occasional gasping by the AAV-VN–KO mice. Plethysmography revealed that they had a normal response to hypoxia (5 minutes with 10% O₂) (Figure 5E), but a blunted hypercapnic (5 minutes with 5% CO₂) ventilatory response (Figure 5F). Moreover, AAV-VN–KO mice had more irregular breathing (IS = 21.2 ± 2.30 AAV-VN–CT mice, n = 14 vs. IS = 36.6 ± 6.10 AAV-VN–KO mice, n = 7; P < 0.05). Injection of the AAV1-Cre-GFP into the VN of conditional tdTomato reporter (line Ai14) mice (33) did not produce any of the symptoms observed when the virus was injected into Ndufs4^lox/lox mice (not shown). Analysis of the Cre-induced tdTomato fluorescence from the reporter gene confirmed that transductions of cells was restricted to the VN and that this nucleus sends projections to the ventral medulla in the vicinity of the preBötC (Supplemental Figure 3) as described before (34).

Viral restoration of Ndufs4 in the VN attenuates respiratory abnormalities and ameliorates the fatal phenotype of KO mice. Because loss of Ndufs4 function in the VN contributes to the motor phenotype and breathing abnormalities of KO mice, we asked whether reexpression of Ndufs4 in the VN was sufficient to ameliorate disease progression in KO mice. An AAV1 construct encoding Ndufs4-IRES-GFP was delivered bilaterally into the VN of KO mice at P21 as a viral rescue strategy (referred to as AAV-VN-VR mice). At P21, the mice were small (<15 g), presymptomatic (22, 23), and more sensitive to anesthesia, making viral injections challenging. Correct targeting was confirmed by extensive GFP expression within the VN (Figure 6A). In accordance with our previous work (23), histopathological analysis of the VN of noninjected mice showed severe microglial activation and tissue lesions (Figure 6B) that were markedly reduced in the injected mice (Figure 6C), suggesting that the cell death observed in the VN of noninjected KO mice is due to the lack of Ndufs4 in that nucleus. The viral rescue delayed clinical progression of the disease (Figure 6F, P < 0.05) and significantly increased the life span of AAV-VN-VR mice compared with KO mice (Figure 6G; median life span: 53 days for KO vs. 69 days for AAV-VN-VR; P < 0.01). Interestingly, the longer-lived AAV-VN-VR mice, which had reduced gliosis in the VN, developed lesions in the striatum (Figure 6, D and E). Striatal lesions are common in people with LS and other mitochondrial diseases, but had not been seen before in late-stage (P50) KO mice (23), suggesting that the prolonged life span observed in the virally rescued mice reveals other brain regions where gliosis develops more slowly.

Figure 6

Ndufs4 reexpression in the VN of KO mice improves breathing and ameliorates disease progression. (A) Representative image of the degree of transduction (GFP-positive cells) after vector delivery in the VN of KO mice (AAV-VN-VR mice). (B–E) Iba1 staining (red) in the VN (B and C) or striatum (D and E) of KO or AAV-VN-VR mice. Asterisk denotes presence of lesions. Cb, cerebellum. Scale bars: 500 μm (A); 250 μm (B–E). (F) Clinical score of the disease progression in KO (n = 9) and AAV-VN-VR (n = 7) mice as described (24). *P < 0.05 genotype, 2-way ANOVA. F(1,121) = 4.18. (G) Survival curve in KO (n = 16) and AAV-VN-VR (n = 7) mice. **P < 0.01, Gehan-Breslow-Wilcoxon test versus KO mice. (H) Changes in breathing of AAV-VN–CT (n = 14), AAV-VN-VR (n = 3), and KO (n = 10) mice exposed to 5-minute hypoxia (10% O₂; gray area). *P < 0.05 versus control mice, 2-way ANOVA, Bonferroni’s post-hoc test. (I) Changes in breathing of AAV-VN–CT (n = 14), AAV-VN-VR (n = 3), and nonresponding KO (n = 5) mice exposed to 5-minute hypercapnia (5% CO₂; gray area). *P < 0.05 versus control mice, 2-way ANOVA, Bonferroni’s post-hoc test. Data are shown as the mean ± SEM.

To ascertain whether improved respiratory response may account for the reduced mortality in the AAV-VN-VR mice, we measured the response of late-stage virally rescued KO mice and controls (control mice injected with AAV-CreΔ-GFP) by plethysmography. AAV-VN-VR mice showed a breathing response to hypoxia comparable to that of control mice during the early (third minute after of hypoxia) phase of the response. However, this was lost at later time points, with the AAV-VN-VR mice showing a response parallel to that of KO mice (Figure 6H and Supplemental Figure 4, A and B). Restoring expression of Ndufs4 in the VN also normalized the hypercapnic ventilatory response of the AAV-VN-VR mice compared with virally injected control mice, in contrast with that of hyperresponding or nonresponding KO mice (Figure 6I and Supplemental Figure 4, C and D). Furthermore, the breathing of the AAV-VN-VR mice recovered its regularity (IS = 21.2 ± 2.30 for control mice, n = 14; IS = 25.7 ± 4.43, for AAV-VN-VR, n = 3). These results show that restoring Ndufs4 function to the VN is sufficient to partially correct the early breathing deficits in KO mice and ameliorate the fatal disease progression of KO mice.

A primary focus of the present study was to elucidate the probable cause of death of KO mice. We started with knowledge of the brain regions that showed the most gliosis and neurodegeneration, as revealed by histopathology and MRI, but the observation of gasping behavior of KO mice provided the first indication of abnormal breathing behaviors. We noted previously (23) that KO mice resemble humans with LS with regard to a large range of phenotypes and we now extend that to breathing. Respiratory abnormalities affect many LS patients (8, 9, 35, 36) and virtually all patients with mutations in the NDUFS4 gene (16–21, 37); however, the link between mitochondrial dysfunction and breathing could not be investigated without an animal model.

We propose that lesions within the VN and FN as well as compromised preBötC function in KO mice result in dysregulation of the central respiratory network and ultimately respiratory failure. Breathing rates determined by optical spectrometry were variable, but older KO mice displayed a reduced breathing rate, especially under stress, along with lower heart rates and incomplete oxygenation of arterial blood compared with control mice. Both neonate and adult KO mice had higher ventilation and higher V_T compared with control mice, as measured by plethysmography, which may be a compensatory response. Patients with LS also display hyperventilation and hypocapnia, resulting in respiratory alkalosis (38). Breathing by KO mice became more irregular and was interrupted by frequent apneas as they grew older. Apneas were a common response to handling. KO mice had abnormal responses when exposed to low levels of O₂ or high levels of CO₂. Although changes in oxygen and pH are thought to be primarily detected in the arterial blood by the carotid bodies (39), the CNS is also sensitive to changes in central levels of O₂ and CO₂ (27, 31, 40, 41). In the absence of carotid bodies, hypoxia still triggers a ventilatory response (42). Furthermore, mice lacking Ndufs4 function only in the CNS have identical breathing defects; therefore, central rather than peripheral mechanisms are primarily involved. Indeed, brain stem slices containing the preBötC exhibited severe abnormalities in inspiratory bursting activity. Under normoxic conditions, inspiratory neurons of KO mice generated fewer APs per burst, which was associated with a lower firing frequency rate compared with that of control mice. These results suggest that the increased V_T and V_E in the intact animal is a compensatory mechanism to a decreased central respiratory drive. When challenged with a 10-minute exposure to hypoxia, the amplitude of fictive gasping in KO mice was reduced compared with that in controls, although the initial augmentation phase was identical. Hence, the central respiratory network initially compensates for decreased levels of O₂, but fails when decreased levels are maintained for a long time, which may explain why KO mice generate more apneas with increasing age. Increased apneas and blunted ventilatory responses to hypoxia and hypercapnia may cause fatal respiratory arrest, as suggested for other pathologies (43, 44). In agreement with our results, all described NDUFS4-LS patients presented apneic episodes and died of cardiorespiratory failure (21). KO mice, however, also presented altered heart rate, especially at older ages, which could contribute to the death of the animals. However, the neural origin of KO pathology has been established previously, and deletion of Ndufs4 in cardiomyocytes and skeletal muscle do not cause any overt clinical phenotype in mice (45), ruling out a primary cardiac defect. Although deficits in the peripheral nervous system inputs to the heart or heart conductivity were not tested in our model, they have not been consistently reported in LS patients (7, 21).

ATP is involved in the maintenance of gasping during the depression phase (46). Furthermore, during hypoxia, ATP-sensitive potassium channels have been reported to be activated in inspiratory neurons of the preBötC (47); therefore, reduced ATP generation may play a major role in this response. Interestingly, mice lacking ATP-sensitive potassium channel 6.2 showed an impaired maintenance of gasping during severe hypoxia (48). We hypothesize that KO mice produce less ATP in energy-demanding cells because of complex I dysfunction. Therefore, during hypoxia, when the ability to produce ATP is compromised, the ATP levels of cells in KO mice may drop dramatically compared with those in control mice, leading to an impaired gasping response which may lead to opening of K_ATP channels. Accordingly, bath application of a K_ATP channel antagonist restored fictive gasping during hypoxic events in PreBötC slices of KO mice. Further work is warranted to confirm this hypothesis.

Although our results indicate that intrinsic properties of preBötC are compromised in KO mice, the mild gliosis within this complex suggests that regulatory inputs to the preBötC play an important role in respiratory dysfunction and death. The cerebellum, FN, and VN are responsible for maintaining balance as well as modulation of breathing, coordination of breathing and swallowing, and adjustment of breathing to postural changes (49). The importance of the cerebellum and brain stem for regulation of automatic ventilation is also manifest in sleep apneas (50–52). These structures also mediate autonomic breathing dysfunction in sudden infant death syndrome and other sleep disorder breathing and cardiovascular syndromes (51, 53). Electrical stimulation of VN neurons modulates firing properties of neurons in the preBötC (34, 54). Consequently, we hypothesize that the gradual demise of the VN network contributes to both the ataxia and respiratory abnormalities observed in KO mice.

Inactivation of Ndufs4 selectively in the VN leads to abnormal control of breathing, and viral restoration of Ndufs4 expression in the VN ameliorates the breathing deficits and mortality of KO mice. Thus, Ndufs4 function in the VN is critical for maintenance of normal balance and ventilation in response to hypercapnia. KO mice with Ndufs4 restored only in the VN probably still have defects in other regulatory inputs to the preBötC as well as the preBötC itself. Furthermore, our viral injections were in the medial and lateral parts of the VN because they show the most gliosis (23); however, other VN subdivisions may also be involved in breathing (55). Other neighboring structures in the dorsal medulla or axons coursing through this area may be affected by inflammation and edema. For example, the nucleus tractus solitarius is involved in breathing regulation (56), and it (along with the VN) projects to the parabrachial nucleus. Lesions of the parabrachial nucleus promote motion sickness, nausea, and anorexia, which are common conditions associated with breathing irregularities and are often observed in LS patients (7).

In summary, our studies validate the usefulness of KO mice as a model for studying the physiological consequences of mitochondrial dysfunction in the brain and underscore the importance of the VN in disease progression.

View Supplemental data

View Supplemental video 1

This work was supported by a grant from the NW Mitochondrial Guild (to R.D. Palmiter). A. Quintana was a recipient of a Ministerio de Ciencia e Innovación (Spain) (MICINN) Postdoctoral Mobility Program fellowship. The authors thank Jennifer Stone, Raj Kapur, and Elisenda Sanz for their help in this research and Glenda Froelick for assistance with histology.

Address correspondence to: Richard Palmiter, HHMI Box 357370, University of Washington, Seattle, Washington 98195, USA. Phone: 206.543.6064; Fax: 206.543.0858; E-mail: Palmiter@uw.edu.

Conflict of interest: The authors have declared that no conflict of interest exists.

Reference information: J Clin Invest. 2012;122(7):2359–2368. doi:10.1172/JCI62923.

1. Leigh D. Subacute necrotizing encephalomyelopathy in an infant. J Neurol Neurosurg Psychiatry. 1951;14(3):216–221.
   View this article via: PubMed CrossRef Google Scholar
2. Rahman S, et al. Leigh syndrome: clinical features and biochemical and DNA abnormalities. Ann Neurol. 1996;39(3):343–351.
   View this article via: PubMed CrossRef Google Scholar
3. Arii J, Tanabe Y. Leigh syndrome: serial MR imaging and clinical follow-up. AJNR Am J Neuroradiol. 2000;21(8):1502–1509.
   View this article via: PubMed Google Scholar
4. Loeffen J, et al. The first nuclear-encoded complex I mutation in a patient with Leigh syndrome. Am J Hum Genet. 1998;63(6):1598–1608.
   View this article via: PubMed CrossRef Google Scholar
5. Cavanagh JB, Harding BN. Pathogenic factors underlying the lesions in Leigh’s disease. Tissue responses to cellular energy deprivation and their clinico-pathological consequences. Brain. 1994;117(pt 6):1357–1376.
   View this article via: PubMed CrossRef Google Scholar
6. Piao YS, Tang GC, Yang H, Lu DH. Clinico-neuropathological study of a Chinese case of familial adult Leigh syndrome. Neuropathology. 2006;26(3):218–221.
   View this article via: PubMed CrossRef Google Scholar
7. Finsterer J. Leigh and Leigh-like syndrome in children and adults. Pediatr Neurol. 2008;39(4):223–235.
   View this article via: PubMed CrossRef Google Scholar
8. Yasaki E, et al. Characteristics of breathing abnormality in Leigh and its overlap syndromes. Neuropediatrics. 2001;32(6):299–306.
   View this article via: PubMed Google Scholar
9. Zhang Y, et al. Clinical and molecular survey in 124 Chinese patients with Leigh or Leigh-like syndrome. J Inherit Metab Dis. 2007;30(2):265.
   View this article via: PubMed CrossRef Google Scholar
10. Kartikasalwah A, Lh N. Leigh syndrome: MRI findings in two children. Biomed Imaging Interv J. 2010;6(1):e6.
    View this article via: PubMed Google Scholar
11. Benit P, et al. Genotyping microsatellite DNA markers at putative disease loci in inbred/multiplex families with respiratory chain complex I deficiency allows rapid identification of a novel nonsense mutation (IVS1nt -1) in the NDUFS4 gene in Leigh syndrome. Hum Genet. 2003;112(5–6):563–566.
    View this article via: PubMed Google Scholar
12. Loeffen JL, et al. Isolated complex I deficiency in children: clinical, biochemical and genetic aspects. Hum Mutat. 2000;15(2):123–134.
    View this article via: PubMed CrossRef Google Scholar
13. Petruzzella V, Papa S. Mutations in human nuclear genes encoding for subunits of mitochondrial respiratory complex I: the NDUFS4 gene. Gene. 2002;286(1):149–154.
    View this article via: PubMed CrossRef Google Scholar
14. Lebon S, et al. Recurrent de novo mitochondrial DNA mutations in respiratory chain deficiency. J Med Genet. 2003;40(12):896–899.
    View this article via: PubMed CrossRef Google Scholar
15. Malfatti E, et al. Novel mutations of ND genes in complex I deficiency associated with mitochondrial encephalopathy. Brain. 2007;130(pt 7):1894–1904.
    View this article via: PubMed CrossRef Google Scholar
16. Anderson SL, et al. A novel mutation in NDUFS4 causes Leigh syndrome in an Ashkenazi Jewish family. J Inherit Metab Dis. 2008;31 suppl 2: :S461–S467.
    View this article via: PubMed CrossRef Google Scholar
17. Budde SM, et al. Clinical heterogeneity in patients with mutations in the NDUFS4 gene of mitochondrial complex I. J Inherit Metab Dis. 2003;26(8):813–815.
    View this article via: PubMed CrossRef Google Scholar
18. Leshinsky-Silver E, et al. NDUFS4 mutations cause Leigh syndrome with predominant brainstem involvement. Mol Genet Metab. 2009;97(3):185–189.
    View this article via: PubMed CrossRef Google Scholar
19. Petruzzella V, et al. A nonsense mutation in the NDUFS4 gene encoding the 18 kDa (AQDQ) subunit of complex I abolishes assembly and activity of the complex in a patient with Leigh-like syndrome. Hum Mol Genet. 2001;10(5):529–535.
    View this article via: PubMed CrossRef Google Scholar
20. van den Heuvel L, et al. Demonstration of a new pathogenic mutation in human complex I deficiency: a 5-bp duplication in the nuclear gene encoding the 18-kD (AQDQ) subunit. Am J Hum Genet. 1998;62(2):262–268.
    View this article via: PubMed CrossRef Google Scholar
21. Assouline Z, et al. A constant and similar assembly defect of mitochondrial respiratory chain complex I allows rapid identification of NDUFS4 mutations in patients with Leigh syndrome. Biochim Biophys Acta. 2012;1822(6):1062–1069.
    View this article via: PubMed Google Scholar
22. Kruse SE, Watt WC, Marcinek DJ, Kapur RP, Schenkman KA, Palmiter RD. Mice with mitochondrial complex I deficiency develop a fatal encephalomyopathy. Cell Metab. 2008;7(4):312–320.
    View this article via: PubMed CrossRef Google Scholar
23. Quintana A, Kruse SE, Kapur RP, Sanz E, Palmiter RD. Complex I deficiency due to loss of Ndufs4 in the brain results in progressive encephalopathy resembling Leigh syndrome. Proc Natl Acad Sci U S A. 2010;107(24):10996–11001.
    View this article via: PubMed CrossRef Google Scholar
24. Telgkamp P, Cao YQ, Basbaum AI, Ramirez JM. Long-term deprivation of substance P in PPT-A mutant mice alters the anoxic response of the isolated respiratory network. J Neurophysiol. 2002;88(1):206–213.
    View this article via: PubMed Google Scholar
25. Smith JC, Ellenberger HH, Ballanyi K, Richter DW, Feldman JL. Pre-Botzinger complex: a brainstem region that may generate respiratory rhythm in mammals. Science. 1991;254(5032):726–729.
    View this article via: PubMed CrossRef Google Scholar
26. Gray PA, Janczewski WA, Mellen N, McCrimmon DR, Feldman JL. Normal breathing requires preBotzinger complex neurokinin-1 receptor-expressing neurons. Nat Neurosci. 2001;4(9):927–930.
    View this article via: PubMed CrossRef Google Scholar
27. Hill AA, Garcia AJ 3rd, Zanella S, Upadhyaya R, Ramirez JM. Graded reductions in oxygenation evoke graded reconfiguration of the isolated respiratory network. J Neurophysiol. 2011;105(2):625–639.
    View this article via: PubMed CrossRef Google Scholar
28. Solomon IC, Edelman NH, Neubauer JA. Pre-Botzinger complex functions as a central hypoxia chemosensor for respiration in vivo. J Neurophysiol. 2000;83(5):2854–2868.
    View this article via: PubMed Google Scholar
29. Lieske SP, Thoby-Brisson M, Telgkamp P, Ramirez JM. Reconfiguration of the neural network controlling multiple breathing patterns: eupnea, sighs and gasps [see comment]. Nat Neurosci. 2000;3(6):600–607.
    View this article via: PubMed CrossRef Google Scholar
30. Corcoran AE, et al. Medullary serotonin neurons and central CO2 chemoreception. Respir Physiol Neurobiol. 2009;168(1–2):49–58.
    View this article via: PubMed CrossRef Google Scholar
31. Abbott SB, et al. Photostimulation of retrotrapezoid nucleus phox2b-expressing neurons in vivo produces long-lasting activation of breathing in rats. J Neurosci. 2009;29(18):5806–5819.
    View this article via: PubMed CrossRef Google Scholar
32. Guyenet PG, Bayliss DA, Stornetta RL, Fortuna MG, Abbott SB, DePuy SD. Retrotrapezoid nucleus, respiratory chemosensitivity and breathing automaticity. Respir Physiol Neurobiol. 2009;168(1–2):59–68.
    View this article via: PubMed CrossRef Google Scholar
33. Madisen L, et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat Neurosci. 2010;13(1):133–140.
    View this article via: PubMed CrossRef Google Scholar
34. Holstein GR, Friedrich VL Jr, Kang T, Kukielka E, Martinelli GP. Direct projections from the caudal vestibular nuclei to the ventrolateral medulla in the rat. Neuroscience. 2011;175:104–117.
    View this article via: PubMed CrossRef Google Scholar
35. Naess K, et al. MtDNA mutations are a common cause of severe disease phenotypes in children with Leigh syndrome. Biochim Biophys Acta. 2009;1787(5):484–490.
    View this article via: PubMed Google Scholar
36. Pincus JH. Subacute necrotizing encephalomyelopathy (Leigh’s disease): a consideration of clinical features and etiology. Dev Med Child Neurol. 1972;14(1):87–101.
    View this article via: PubMed Google Scholar
37. Papa S, et al. Mutation in the NDUFS4 gene of complex I abolishes cAMP-dependent activation of the complex in a child with fatal neurological syndrome. FEBS Lett. 2001;489(2–3):259–262.
    View this article via: PubMed CrossRef Google Scholar
38. Pronicka E, et al. Compulsory hyperventilation and hypocapnia of patients with Leigh syndrome associated with SURF1 gene mutations as a cause of low serum bicarbonates. J Inherit Metab Dis. 2001;24(7):707–714.
    View this article via: PubMed CrossRef Google Scholar
39. Heymans C, Bouckaert JJ. Sinus caroticus and respiratory reflexes: I. Cerebral blood flow and respiration. Adrenaline apnoea. J Physiol. 1930;69(2):254–266.
    View this article via: PubMed Google Scholar
40. Solomon IC. Ionotropic excitatory amino acid receptors in pre-Botzinger complex play a modulatory role in hypoxia-induced gasping in vivo. J Appl Physiol. 2004;96(5):1643–1650.
    View this article via: PubMed CrossRef Google Scholar
41. Telgkamp P, Ramirez JM. Differential responses of respiratory nuclei to anoxia in rhythmic brain stem slices of mice. J Neurophysiol. 1999;82(5):2163–2170.
    View this article via: PubMed Google Scholar
42. Curran AK, Rodman JR, Eastwood PR, Henderson KS, Dempsey JA, Smith CA. Ventilatory responses to specific CNS hypoxia in sleeping dogs. J Appl Physiol. 2000;88(5):1840–1852.
    View this article via: PubMed Google Scholar
43. Dubreuil V, et al. A human mutation in Phox2b causes lack of CO2 chemosensitivity, fatal central apnea, and specific loss of parafacial neurons. Proc Natl Acad Sci U S A. 2008;105(3):1067–1072.
    View this article via: PubMed CrossRef Google Scholar
44. McKay LC, Janczewski WA, Feldman JL. Sleep-disordered breathing after targeted ablation of preBotzinger complex neurons. Nat Neurosci. 2005;8(9):1142–1144.
    View this article via: PubMed CrossRef Google Scholar
45. Sterky FH, et al. Altered dopamine metabolism and increased vulnerability to MPTP in mice with partial deficiency of mitochondrial complex I in dopamine neurons. Hum Mol Genet. 2012;21(5):1078–1089.
    View this article via: PubMed CrossRef Google Scholar
46. Gourine AV, Llaudet E, Dale N, Spyer KM. Release of ATP in the ventral medulla during hypoxia in rats: role in hypoxic ventilatory response. J Neurosci. 2005;25(5):1211–1218.
    View this article via: PubMed CrossRef Google Scholar
47. Mironov SL, Langohr K, Haller M, Richter DW. Hypoxia activates ATP-dependent potassium channels in inspiratory neurones of neonatal mice. J Physiol. 1998;509(pt 3):755–766.
    View this article via: PubMed CrossRef Google Scholar
48. Miyake A, Yamada K, Kosaka T, Miki T, Seino S, Inagaki N. Disruption of Kir6.2-containing ATP-sensitive potassium channels impairs maintenance of hypoxic gasping in mice. Eur J Neurosci. 2007;25(8):2349–2363.
    View this article via: PubMed CrossRef Google Scholar
49. Yates BJ, Bronstein AM. The effects of vestibular system lesions on autonomic regulation: observations, mechanisms, and clinical implications. J Vestib Res. 2005;15(3):119–129.
    View this article via: PubMed Google Scholar
50. Kumar R, Macey PM, Woo MA, Alger JR, Harper RM. Diffusion tensor imaging demonstrates brainstem and cerebellar abnormalities in congenital central hypoventilation syndrome. Pediatr Res. 2008;64(3):275–280.
    View this article via: PubMed CrossRef Google Scholar
51. Krous HF, Chadwick AE, Haas E, Masoumi H, Stanley C. Sudden infant death while awake. Forensic Sci Med Pathol. 2008;4(1):40–46.
    View this article via: PubMed CrossRef Google Scholar
52. Harper RM. Sudden infant death syndrome: a failure of compensatory cerebellar mechanisms? Pediatr Res. 2000;48(2):140–142.
    View this article via: PubMed CrossRef Google Scholar
53. Greenough G, Sateia M, Fadul CE. Obstructive sleep apnea syndrome in a patient with medulloblastoma. Neuro Oncol. 1999;1(4):289–291.
    View this article via: PubMed Google Scholar
54. Zheng Y, Umezaki T, Nakazawa K, Miller AD. Role of pre-inspiratory neurons in vestibular and laryngeal reflexes and in swallowing and vomiting. Neurosci Lett. 1997;225(3):161–164.
    View this article via: PubMed CrossRef Google Scholar
55. Xu F, Zhuang J, Zhou TR, Gibson T, Frazier DT. Activation of different vestibular subnuclei evokes differential respiratory and pressor responses in the rat. J Physiol. 2002;544(pt 1):211–223.
    View this article via: PubMed CrossRef Google Scholar
56. Conrad SC, Nichols NL, Ritucci NA, Dean JB, Putnam RW. Development of chemosensitivity in neurons from the nucleus tractus solitarii (NTS) of neonatal rats. Respir Physiol Neurobiol. 2009;166(1):4–12.
    View this article via: PubMed CrossRef Google Scholar
57. Fadok JP, Dickerson TM, Palmiter RD. Dopamine is necessary for cue-dependent fear conditioning. J Neurosci. 2009;29(36):11089–11097.
    View this article via: PubMed CrossRef Google Scholar
58. Bruchas MR, et al. Selective p38alpha MAPK deletion in serotonergic neurons produces stress resilience in models of depression and addiction. Neuron. 2011;71(3):498–511.
    View this article via: PubMed CrossRef Google Scholar
59. Viemari JC, et al. Mecp2 deficiency disrupts norepinephrine and respiratory systems in mice. J Neurosci. 2005;25(50):11521–11530.
    View this article via: PubMed CrossRef Google Scholar
60. Franklin KB, Paxinos G. The Mouse Brain in Stereotaxic Coordinates. 3rd ed. New York, New York, USA: Academic Press; 2008.