In vivo Evidence for Cerebral Depletion in High-Energy Phosphates in Progressive Supranuclear Palsy

Abstract

Indirect evidence from laboratory studies suggests that mitochondrial energy metabolism is impaired in progressive supranuclear palsy (PSP), but brain energy metabolism has not yet been studied directly in vivo in a comprehensive manner in patients. We have used combined phosphorus and proton magnetic resonance spectroscopy to measure adenosine-triphosphate (ATP), adenosine-diphosphate (ADP), phosphorylated creatine, unphosphorylated creatine, inorganic phosphate and lactate in the basal ganglia and the frontal and occipital lobes of clinically probable patients (N= 21; PSP stages II to III) and healthy controls (N= 9). In the basal ganglia, which are severely affected creatine in PSP patients, the concentrations of high-energy phosphates (= ATP + phosphorylated creatine) and inorganic phosphate, but not low-energy phosphates (=ADP+ unphosphorylated creatine), were decreased. The decrease probably does not reflect neuronal death, as the neuronal marker N-acetylaspartate was not yet significantly reduced in the early-stage patients examined. The frontal lobe, also prone to neurodegeneration in PSP, showed similar alterations, whereas the occipital lobe, typically unaffected, showed less pronounced alterations. The levels of lactate, a product of anaerobic glycolysis, were elevated in 35% of the patients. The observed changes in the levels of cerebral energy metabolites in PSP are consistent with a functionally relevant impairment of oxidative phosphorylation.

Keywords

energy metabolism high-energy phosphates oxidative phosphorylation phosphorus/proton magnetic resonance spectroscopy progressive supranuclear palsy

Introduction

Progressive supranuclear palsy (PSP) is a sporadic neurodegenerative disorder characterized by an akinetic-rigid syndrome with prominent postural instability, oculomotor deficits and neuropsychological deficits in frontal lobe-related functions (Litvan et al, 1996). Neuropathologically, PSP is characterized by neuronal loss and somatodendritic aggregation of the microtubule-associated protein tau, predominantly in the basal ganglia, brain stem nuclei and the frontal lobes; the occipital lobes and the cerebellum are relatively spared (Hauw et al, 1994). The median survival time is 5 to 10 years (Golbe and Ohman-Strickland, 2007). To date, there are no effective symptomatic or neuroprotective treatments available (van Balken and Litvan, 2006). Therefore, a detailed understanding of the pathogenesis underlying PSP is urgently needed to allow the development of novel rational therapies.

Although the etiology of PSP is unknown, several lines of evidence point to an impairment of mitochondrial energy production: (1) Cerebral glucose metabolism is reduced, as shown by [¹⁸F]-2-fluoro-2-deoxy-d-glucose positron emission tomography (Foster et al, 1988). (2) Increased levels of the low-energy phosphate (LEP), adenosine-diphosphate (ADP), were detected in the occipital lobe by phosphorous magnetic resonance spectroscopy (Martinelli et al, 2000). (3) In cybrid cells containing mitochondria from the platelets of PSP patients, the activity of complex I of the mitochondrial respiratory chain was reduced, as were oxygen consumption and adenosine-triphosphate (ATP) production (Swerdlow et al, 2000; Albers et al, 2001; Chirichigno et al, 2002). (4) Reduced ATP production was also found in biochemical assays using mitochondria from skeletal muscle (Di Monte et al, 1994). These observations suggest that mitochondrial energy metabolism is impaired in PSP.

Experimental findings also suggest that mitochondrial failure might play a crucial role in the pathophysiology of PSP. A PSP-like tauopathy endemic in the island of Guadeloupe has been linked to the consumption of fruit and infusions of the leaves of Annona muricata (Lannuzel et al, 2007), which contain the complex I inhibitor, annonacin, that induces a PSP-like pattern of neurodegeneration in the rat brain (Champy et al, 2004) and tau pathology in cultured neurons (Escobar-Khondiker et al, 2007). Similar results have been obtained with rotenone, another natural complex I inhibitor (Höglinger et al, 2003, 2005).

The main function of the mitochondrial respiratory chain is to maintain adequate concentrations of high-energy phosphates (HEP), particularly ATP, in cells. It is important to note, however, that brain mitochondria can tolerate a 60% decrease in complex I activity, without a significant decrease in ATP production (Davey et al, 1997). In the experimental models using annonacin and rotenone, a significant decrease in ATP levels was required to induce tau pathology; mild complex I inhibition without ATP depletion had no effect (Höglinger et al, 2005; Escobar-Khondiker et al, 2007). To date, however, there has been no direct demonstration in vivo regarding the bioenergetic alterations identified in PSP patients lead to significant changes in brain ATP levels.

High-energy metabolites (ATP and phosphorylated creatine (pCre)) and inorganic phosphate (Pi) can be quantified in vivo by phosphorus magnetic resonance spectroscopy (³¹P-MRS). Lactate, an indicator of anaerobic glycolysis, and N-acetylaspartate (NAA), a marker of neuronal integrity, can be quantified by proton [¹H]MRS. Combining ³¹P- and ¹H-MRS additionally allows the quantification of the low-energy metabolites, ADP and unphosphorylated creatine (uCre). Therefore, we have combined ³¹P- and ¹H-MRS to measure these metabolites in differentially vulnerable brain regions in PSP patients and in healthy controls.

Materials and methods

This study was approved by the Ethics Committee of Marburg University and is registered online (http://clinicaltrials.gov/, identifier NCT00328874).

Participants

Patients examined in the Department of Neurology of the University of Marburg were qualified for participation in the study if they had clinically probable PSP (Litvan et al, 1996). This criterion was chosen to obtain the highest degree of diagnostic certainty possible in vivo in patients who were not more advanced than stage III (Golbe, 1997), thus avoiding a bias in the measured levels of energy metabolites because of pronounced neuronal loss.

The healthy volunteers were free of neurologic, systemic or psychiatric diseases, including alcohol or substance abuse, as verified by a detailed evaluation of their medical histories and by a comprehensive physical examination. Furthermore, participants were included only if they were willing and able to give written informed consent to participate in the study.

We excluded a variety of conditions that might influence the levels of cerebral energy metabolites. Exclusion criteria for all participants were age > 50 or > 85 years; Parkinson's syndromes other than PSP; dementia (Mini-Mental State Examination (MMSE) score ⩽24); a history of epilepsy, structural brain disease, brain surgery, electroconvulsive therapy, stroke and arterial hypertension (systolic > 180 or diastolic > 110mmHg); systemic disorders affecting the metabolism or function of the brain (e.g., diabetes mellitus); other serious illnesses; participation in drug studies or the use of drugs that modify mitochondrial activity (e.g., coenzyme Q10, statins) or antioxidants (e.g., vitamin E, C) within the last 60 days; the use of drugs interfering with catecholamine metabolism (e.g., reserpine, amphetamines, MAO-A inhibitors, methylphenidate and cinnarizine) within the last 30 days; an unstable dosage of CNS-active drugs (e.g., antiparkinsonian drugs, anxiolytics, hypnotics, tranquillizers and antidepressants) within the last 30 days.

Clinical Evaluation

Motor functions were assessed with the Unified Parkinson's Disease Rating Scale (UPDRS), part III, the PSP staging system (Golbe, 1997) and the PSP rating scale (PSP-RS) (Golbe and Ohman-Strickland, 2007). Activities of daily living were scored with the UPDRS, part II. Global cognitive function was evaluated with the MMSE. Neuropsychiatric symptoms were evaluated with the Frontal Assessment Battery (FAB) (Dubois et al, 2000) and the Montgomery-Åsberg Depression rating scale (MADRS).

1H-MRS

A 3-T system (Magnetom Trio; Siemens, Erlangen, Germany) with a double-tuned ¹H/³¹P volume head coil (Rapid Biomedical, Würzburg, Germany) was used for ¹H-MRS, with a two-dimensional chemical shift imaging double echo (CSI) sequence with PRESS volume selection. A 1.5-cm axial slice, including the basal ganglia, was recorded. Data were acquired with circular-phase encoding using a 16 × 16 matrix, which was extrapolated before Fourier transformation to 32 × 32 matrix. The field of view was 240mm², the repetition time 1,500 msecs and the echo time 30 msecs. For quantitative analysis, spectra were fitted in the frequency domain by a linear combination of a set of model spectra using the LCModel software (Provencher, 1993). Cramer–Rao lower bounds for NAA, choline (Cho) and creatine (Cre) were typically between 5% and 10%. Data with the Cramer–Rao lower bounds of more than 20% were discarded. Corrections for ¹H-T¹ and ¹H-T² relaxation were applied (Hattingen et al, 2007).

Phosphorus Magnetic Resonance Spectroscopy

Phosphorus magnetic resonance spectroscopy data were acquired using a three-dimensional CSI phase-encoding scheme applying a pulse-free induction decay sequence with WALTZ4 proton decoupling. A 300 × 300 × 200mm³ three-dimensional CSI slab was recorded with angulation and offset identical to the ¹H-slice. Circular-phase encoding was employed with a weighted acquisition scheme on a 10 × 10 × 8 matrix extrapolated to 16 × 16 × 8, resulting in a series of axial slices with nominal 25mm thickness and 18.75 × 18.75mm² in plane resolution (flip angle 60°, repetition time 2000 msecs and echo time 2.3 msecs). Before spatial Fourier transformation, the data set was multiplied with a linear increasing phase in the foot to head direction. Phase increments were adjusted to cause a shift of 12.5mm toward the feet, resulting in a perfect alignment of slice 5 from the ³¹P data with the ¹H-slice. Spectra were analyzed in the time domain with jMRUI software employing the non-linear least square-fitting algorithm AMARES (Vanhamme et al, 1997). The time domain model function was composed of 14 exponentially decaying sinusoids in the frequency domain. Six of those, which had identical damping, corresponded to peaks assigned to pCre, phosphoethanolamine, phosphocholine, glycero-phosphocholine, glycero-phosphoethanolamine and Pi. pCre was adjusted to 0 p.p.m. and, except for Pi, constraints for the chemical shifts of the other signals were applied as a fixed difference with respect to the position of pCre. ATP was represented by seven exponentially damped sinusoids, defining each multiplet by the respective number of peaks with identical damping and adequate amplitude ratios. The coupling constant was fixed at 18 Hz. One signal with a fixed chemical shift of 2.24 p.p.m. and maximum line width of 50 Hz was used to account for potential macromolecule signals in the phosphodiester region. Corrections were performed for 60° flip angle and partial saturation using T¹ of 2.4 secs for pCre, 1.1 secs for Pi and 1.0 secs for ATP, resulting in respective intensity correction terms of 1.60, 1.27 and 1.24. The relaxation times are within the error margins of published data measured at 1.5 T (Buchli et al, 1994) and 4 T (Hetherington et al, 2001).

External Calibration

Signal intensities were quantified in terms of absolute concentrations by the phantom replacement method (Michaelis et al, 1993). The phantom solution, which was used as external calibration reference, contained 20 mmol/l creatine and 20 mmol/l phosphate. In contrast to the in vivo sequence for ³¹P data, a two-dimensional CSI sequence with full 16 × 16 matrix and 240mm² field of view was performed using a 90° flip angle and a repetition time of 40 secs.

Regions of Interest

Figure 1A shows the axial plane of a T₁-weighted magnetic resonance image from a control brain. The grid indicates the resolution of each modality, whereas the location of the templates defines the regions of interest (ROIs) in the ‘basal ganglia’, ‘frontal lobe’ and ‘occipital lobe’ for ¹H- and ³¹P-MRS. As the data from the left and right ‘basal ganglia’ and ‘frontal lobe’ ROIs did not differ significantly from each other, only the averaged values are given. The ROI for the ‘occipital lobe’ crossed the midline and thus yielded only one set of data.

Figure 1

Magnetic resonance spectroscopy (MRS). (A) T₁-weighted MR images of a control brain showing a characteristic plane of analysis, grid resolution (green lines) for ¹H-MRS (left) and ³¹P-MRS (right) and the area selected by the PRESS excitation scheme (white line) for ¹H-MRS. Colored squares indicate the location of the templates defining the regions of interest (‘frontal lobe’ (blue), ‘basal ganglia’ (yellow) and ‘occipital lobe’ (red)). (B and C) Representative spectra obtained with (B) ¹H-MRS and (C) ³¹P-MRS from the ‘basal ganglia’ region of interest in a healthy control person (Cont, left) and a patient (PSP, right). Blue lines indicate the raw spectra and red lines the fitted curves (LCModel for ¹H-MRS and jMRUI for ³¹P-MRS). Signal intensity is given in arbitrary units. The peaks for total creatine (tCre), N-acetylaspartate (NAA) and lactate (Lac) are identified in (panel B); the peaks for inorganic phosphate (Pi), phosphodiester (PDE), phosphorylated creatine (pCre) and α-, β- and γ-adenosine-diphosphate (ATP) in (panel C). In the ¹H-MRS spectrum of a PSP patient (panel B), note the presence of a Lac peak (positive doublet signal at 1.3 p.p.m.) that is not discernible in the control spectrum. In the ³¹P-MRS spectrum of a PSP patient (panel C), the lower peak size of Pi, pCre and ATP (particularly β-ATP) was noted compared with that of the control spectrum; as internal reference, the PDE peak, representing a metabolite not implicated in energy metabolism, is not altered in patients.

MRS Data Acquisition

The signals for tCre, NAA and lactate were obtained from the ¹H spectra (Figure 1B). Those for Pi, pCre and ATP were from the ³¹P spectra (Figure 1C).

Lactate: In healthy control brains, lactate is not detectable by ¹H-MRS, using the data acquisition scheme described by Isobe et al (2007). However, anaerobic glycolysis (i.e., an excess of glycolysis over oxidative phosphorylation) can lead to elevated and, therefore, detectable lactate levels (Prichard et al, 1991). The LCModel software provides a value for the lactate concentration and an estimation of its accuracy is given (% s.d.). Each complete CSI data set was screened for lactate s.d. values > 20%. The presence of a clear positive doublet signal at 1.3 p.p.m. in the spectra was then verified. Voxels fulfilling these criteria were considered to have elevated lactate levels (see Figure 1B).

ATP: The ³¹P-MRS spectra for ATP have three signals corresponding to the three phosphorous nuclei of the nucleotide (Figure 1C). α-ATP contains contributions from reduced nicotinamide adenine dinucleotide (NADH) and α-ADP; γ-ATP contains contributions from β-ADP, whereas the β-ATP peak area is proportional to the total cellular ATP concentration (Iles et al, 1985). We therefore used the β-ATP value as representative of total ATP levels (Hu et al, 2000; Pietz et al, 2003; Rango et al, 2006).

The following parameters were calculated from the measured values:

ADP= [(tCre–pCre) *β–ATP]/[pCre × H × K_CK]

with K_CK = 1.66 × 10⁹ mol/L⁻¹ and H= 10^−pH mol/L (Pietz et al, 2003);

uCre = tCre–pCre;

HEP= β-ATP + pCre (Rango et al, 2006);

LEP =ADP+ uCre (Rango et al, 2006);

Phosphorylation potential (PP) =ATP/(ADP × Pi) (Pietz et al, 2003).

pH: Tissue pH was estimated from the chemical shift difference between Pi and pCre according to the formula

p H = p k_{A} +^{10} log (\frac{δ 1 - δ 0}{δ 0 - δ 2})

where δ1 = 3.27, δ2 = 5.63, pk^A = 6.75, and δ0 equals a chemical shift difference between the Pi and pCre signals (Petroff et al, 1985).

Statistical Analysis

As data on cytosolic ADP levels measured by ¹H-MRS in the occipital lobes of PSP patients compared with those of healthy controls are available (Martinelli et al, 2000), ADP was used as the main outcome parameter for sample size calculation. With a 2:1—the patient/control ratio, 16:8 available individuals are needed to obtain 90% power to detect a significant difference (P > 0.05) of 7.5 µmol/L with a standard deviation of 5 µmol/L in a two-sided Wilcoxon–Mann–Whitney test. In view of possible dropouts, the planned sample size was 20 PSP patients and 10 healthy controls. The confirmatory analysis of the primary end point was performed on data of all patients and controls for which ADP in the basal ganglia could be calculated using the above-mentioned test. The analysis of all secondary data is descriptive. Data are shown as mean±s.d. The two-sided Wilcoxon–Mann–Whitney test was used to compare metric variables and the two-sided Fisher exact test for categorical variables. P-values > 0.05 were considered to be statistically significant. No adjustment for multiple testing was performed. Statistical analyses were performed with the SAS statistical package (version 9.1).

Results

Study Population

A total of 67 patients were screened for eligibility, of which 46 were excluded: 21 for clinically possible but not clinically probable PSP, 21 for Parkinson's syndrome other than PSP, 14 for a PSP stage > III, 14 for dementia, 5 for structural brain disease, 2 for insufficiently treated arterial hypertension, 2 for a systemic disease affecting the brain and 2 for another serious illness. The total number of excluded patients was less than the sum of the individual items, as some patients fulfilled more than one of the exclusion criteria. Twenty-one patients met the inclusion and none of the exclusion criteria and participated in the study.

The clinical characteristics of the analyzed participants are shown in Table 1.

Table 1

Clinical characteristics of the participants

	Control	PSP
N	9	21
Age (years)	62.2±7.3	66.2±6.3
Sex
Males	N= 3 (33%)	M: N= 10 (48%)
Females	N= 6 (67%)	N= 11 (52%)
Disease duration	—	4.2±3.0 years
PSP stage
II	—	N= 13 (62%)
III	—	N= 8 (38.0%)
PSP-RS	—	43.1±8.4
UPDRS II	—	20.2±5.4
UPDRS III	—	24.0±5.3
MMSE	—	27.6±1.4
MADRS	—	5.9±1.9
FAB	—	13.0±2.3

FAB, Frontal Assessment Battery (range 0 (worst)−18 (best)); MADRS, Montgomery-Åsberg Depression Rating Scale (range 0 (best)−60 (worst)); MMSE, Mini-Mental Status Examination (range 0 (worst)−30 (best)); PSP stage (range I (best)–V (worst)); PSP, progressive supranuclear palsy; PSP-RS, PSP Rating Scale (range 0 (best)−100 (worst)); UPDRS, Unified Parkinson's Disease Rating Scale, part II (activities of daily living) (range 0 (best)−52 (worst)), part III (motor score) (range 0 (best)−56 (worst)).

N-Acetylaspartate

We found no significant differences in NAA, tCre or the NAA/tCre ratio in patients and controls in any of the ROIs analyzed (Figure 1B, Table 2), suggesting that the patients studied, who were in early stages of the disease, had only mild neuronal disintegrity in the analyzed ROIs.

Table 2

Cerebral metabolites measured by magnetic resonance spectroscopy

	Basal ganglia				Frontal lobe				Occipital lobe
	Control	PSP	% Difference	P-value	Control	PSP	% Difference	P-value	Control	PSP	% Difference	P-value
[NAA] (mmol/L)	9.88±1.42	9.51±1.47	−3.7	0.71	10.41±1.35	9.43±1.64	−9.4	0.31	9.49±1.77	9.76±1.22	2.8	0.54
[tCre] (mmol/L)	8.28±1.34	8.50±1.52	2.7	0.76	10.21±1.67	10.01±2.38	−2.0	0.66	6.98±1.85	7.17±1.18	2.8	0.36
[NAA]/[tCre]	1.20±0.11	1.13±0.11	−6.1	0.18	1.03±0.11	0.96±0.13	16.8	0.53	1.38±0.13	1.38±0.19	−0.2	0.86
[ATP] (mmol/L)	2.03±0.23	1.68±0.35	−17.0	0.01	1.36±0.2	1.18±0.25	−13.2	0.09	1.86±0.36	1.71±0.36	−7.8	0.32
[pCre] (mmol/L)	3.18±0.34	2.86±0.35	−9.8	0.09	2.37±0.23	2.08±0.31	−12.3	0.02	3.34±0.19	3.11±0.46	−6.7	0.29
[HEP] (mmol/L)	5.20±0.46	4.55±0.52	−12.6	0.006	3.72±0.33	3.25±0.44	−12.6	0.007	5.19±0.49	4.83±0.63	−7.1	0.16
[Pi] (mmol/L)	0.70±0.15	0.57±0.10	−18.1	0.05	0.59±0.13	0.49±0.13	−17.5	0.08	0.65±0.07	0.59±0.11	−9.8	0.17
[ADP] (µmol/L)	20.0±5.0	20.0±9.0	−0.3	0.79	26.0±6.0	33.0±15.0	27.7	0.30	13.0±5.0	14.0±8.0	12.4	0.79
[uCre] (mmol/L)	5.27±1.19	5.64±1.58	7.0	0.68	7.82±1.91	8.03±2.48	2.6	0.96	3.93±1.93	4.08±1.28	3.8	0.53
[LEP] (mmol/L)	5.29±1.20	5.66±1.59	7.0	0.68	7.85±1.92	8.06±2.49	2.7	0.96	3.94±1.93	4.09±1.29	3.8	0.53
[ATP]/[ADP][Pi]	153±43	163±63	6.4	0.79	92±32	122±71	32.7	0.63	254±94	249±120	−1.9	0.68
[ATP]/[ADP]	104±22	92±33	−11.3	0.35	54±18	45±19	−16.5	0.14	166±65	147±80	−11.6	0.31
[pCre]/[uCre]	0.63±0.13	0.55±0.2	−11.3	0.35	0.33±0.11	0.27±0.11	−16.5	0.14	1±0.39	0.89±0.48	−11.6	0.31
[HEP]/[LEP]	1.02±0.22	0.85±0.25	−17.1	0.15	0.51±0.17	0.44±0.15	−13.2	0.55	1.55±0.61	1.36±0.71	−12.8	0.19
pH	7.05±0.02	7.06±0.03	0.1	0.78	7.06±0.03	7.07±0.03	0.1	0.87	7.08±0.03	7.06±0.02	−0.2	0.52

ADP, adenosine-diphosphate; ATP, adenosine-triphosphate, β-ATP peak; HEP, high-energy phosphates=ATP+pCre; LEP, low-energy phosphates=ADP+uCre; NAA, N-acetylaspartate; pCre, phosphorylated creatine; Pi, inorganic phosphate; tCre, total creatine; uCre, unphosphorylated creatine.

difference indicates the difference of the means as percentage of the mean for controls.

P-values from two-sided Wilcoxon–Mann–Whitney test.

Bold highlights significant changes (P>0.05).

Energy Metabolites in the Basal Ganglia

In the ‘basal ganglia,’ ROI (averaged value of the left and the right hemisphere), there was a significant reduction in the concentration of ATP (P = 0.01) in PSP patients compared with that in controls; pCre showed a consistent trend (P = 0.09); the sum of the HEPs (=ATP + pCre) was also significantly lower in patients (P = 0.006) (Figure 1C, Table 2).

When the analysis of the ‘basal ganglia’ ROI was carried out separately for the left and right hemispheres, the HEP concentration was also significantly lower in patients on both sides (left: −12.1%, P = 0.03; right: −13.1%, P = 0.02); the reduction in pCre reached significance on the left side (–10.5%, P = 0.05); and the reduction in ATP reached significance on the right side (–19.1%, P = 0.02).

Free phosphate (Pi) levels were also decreased in the averaged ‘basal ganglia’ ROI of patients (P = 0.05) (Figure 1C, Table 2).

In contrast, there were no significant differences in the LEPs, ADP and uCre, and the sum of both (LEPs) in the averaged ‘basal ganglia’ ROI of patients and controls (Table 2).

The ratios of the HEPs to their low-energy counterparts (PP =ATP/(ADP × Pi), ATP/ADP, pCre/uCre and HEP/LEP), which are considered to represent the cellular energy load that drives all energy-dependent biochemical processes (Iles et al, 1985; Pietz et al, 2003), were not altered in the ‘basal ganglia’ ROI of the PSP patients (Table 2). Neither were the HEP/Pi (Rango et al, 2006), ATP/Pi (Hu et al, 2000) and pCre/Pi ratios (Penn et al, 1995) (data not shown).

Energy Metabolites in Other Brain Areas

In the averaged ‘frontal lobe’ ROI of patients, changes in the concentrations of energy metabolites were similar to those in the ‘basal ganglia’ ROI: there were significantly lower concentrations of HEP (P = 0.007) and pCre (P = 0.02); there was a consistent trend for ATP (P = 0.09) and Pi (P = 0.08); and the LEPs (ADP, uCre, LEP) were unaltered (Table 2).

In the averaged ‘occipital lobe’ ROI, none of the parameters measured differed significantly in patients and controls (Table 2).

Lactate

In healthy control brains, lactate is not detectable by ¹H-MRS, using the data acquisition scheme described earlier (Isobe et al, 2007). Consistently, we found no lactate peaks in the ¹H-MRS data of any of healthy control subjects. In contrast, lactate peaks suggestive of increased anaerobic glycolysis (Prichard et al, 1991) were found in 7 of 20 (35%) PSP patients. These lactate peaks were typically found in a paraventricular location in the ‘basal ganglia’ ROIs (Figure 1B).

pH

The intracellular pH was not significantly different in any of the ROIs of the patients compared with those of the controls (Table 2).

Dependence of MRS Outcomes on Age and Sex

In view of slight, nonsignificant imbalances in the age and sex distributions of patients and healthy controls (Table 1), we calculated robust regressions of the MRS outcomes in patient status adjusted for age and sex as sensitivity analyses. The results were in line with the nonadjusted analyses (data not shown).

Clinical Correlations

We calculated the Spearman rank correlations between the concentrations of the HEPs (ATP, pCre, HEP) and the PP in the averaged ‘basal ganglia’ ROI and the scales measuring motor dysfunction (PSP-RS total score and UPDRS III). No significant correlations were observed (data not shown). We also found no significant rank correlations of the concentrations between the HEPs (ATP, pCre, HEP) and the PP in the frontal lobes with the FAB scale measuring frontal lobe dysfunction (not shown).

Differentiation of Patients From Controls

In our data set, a preliminarily assumed threshold of ⩽4.95 mmol/L for HEP would have distinguished PSP patients from controls free of neurologic symptoms with a sensitivity of 85% and a specificity of 71%.

Discussion

Phosphorus magnetic resonance spectroscopy is an excellent technique for monitoring cerebral energy metabolism in vivo, as ATP, pCre and Pi can be identified and quantified by ³¹P resonance peaks. The concentrations of HEP and pH can also be derived from the ³¹P spectrum. However, to quantify NAA, tCre and Lac, ¹H-MRS must be used. Furthermore, the concentrations of ADP, uCre, LEP and the PP can be only derived without further a priori assumptions from joint ³¹P- with ¹H-MRS measurements. Therefore, we used combined ³¹P- and ¹H-MRS on a 3-T high-field system, calibrated both modalities to external standards to quantify absolute metabolite concentrations rather than using raw signal intensities and applied this technique in three ROIs in differentially vulnerable brain areas in 21 early-stage PSP patients compared with those in 9 healthy controls. Thus, we provide here the first comprehensive representation of brain energy metabolism in PSP patients in vivo.

Reduced cerebral glucose metabolism has already been shown in vivo in PSP patients using ¹⁸F-2-fluoro-2-deoxy-d-glucose and positron emission tomography (Foster et al, 1988). However, this approach does not distinguish between hypometabolism that is the nonspecific consequence of neuronal inactivity or loss and also the result of a primary impairment of mitochondrial metabolism. Several indirect experimental studies indeed suggested that there was mitochondrial dysfunction in PSP patients (Swerdlow et al, 2000; Albers et al, 2001; Chirichigno et al, 2002). This study is the first to provide direct in vivo evidence of a depletion of high-energy metabolites in the brains of PSP patients. In the basal ganglia, which is the brain area most severely affected by neuronal dysfunction and neurodegeneration in PSP, we observed a characteristic pattern of change in the levels of energy metabolites: a decrease in high-energy metabolites (HEP =ATP + pCre), reduced Pi and unaltered low-energy metabolites (LEP =ADP+ uCre). The frontal lobe, also strongly affected by neurodegeneration in PSP, showed a similar pattern of metabolic alterations. In the occipital lobe, however, which is spared in PSP, the deficits in metabolite concentrations in patients were less pronounced.

N-acetylaspartate is an amino acid that is present in high concentrations almost exclusively in neurons. It contributes to the most prominent signal in ¹H-MRS of the adult human brain. Reduced regional NAA concentrations have been reported in conditions that are characterized by neuronal cell or axon loss (Rango et al, 2007). In addition, experimental data suggest a linkage between reduced NAA levels and compromised neuronal metabolism (Moffett et al, 2007). On the basis of these data, the quantification of the NAA peak is presently considered as the single most sensitive parameter in magnetic resonance spectroscopy to indicate neuronal loss or disintegrity (Moffett et al, 2007; Rango et al, 2007). Many studies normalize NAA values with respect to tCre levels used as an internal standard, as tCre levels are thought to be relatively constant across the brain and do not change in most pathologies. The NAA/tCre and the NAA/Cho ratios have been reported to be reduced in PSP (Federico et al, 1997, 1998; Tedeschi et al, 1997; Abe et al, 2000). In our sample of mildly affected early-stage PSP patients, the decrease in the high-energy metabolites appeared to exceed rather than to reflect neuronal disintegrity, as it was observed without concomitant reductions in the NAA/tCre ratio. Partial volume effects might mimic a decrease in high-energy metabolite concentrations as a result of brain atrophy when comparing patients with controls. However, as concentrations of HEPs (ATP, pCre) were significantly reduced in patients without concomitant significant reductions in parenchymal reference metabolites (NAA, tCre and NAA/tCre) and with a nonsignificant tendency of increased concentrations of LEPs (ADP, uCre) in most ROIs, this effect appears to be specific and not merely an artifact resulting from brain atrophy. These observations support the idea that mitochondrial dysfunction is indeed implicated as an upstream event in the pathophysiology of PSP rather than being a secondary epiphenomenon.

An important question arising from our observations concerns the nature of the biochemical deficit at the origin of the metabolic alterations. The decrease in ATP might be explained by either an increase in ATP hydrolysis or a decrease in ATP production through glycolysis and oxidative phosphorylation. In the first case, one would expect an increase in glucose metabolism, but the inverse is seen in the brains of PSP patients (Foster et al, 1988). Therefore, our results support the hypothesis of a primary impairment of mitochondrial ATP production in PSP (Figure 2). The increase in lactate levels in a proportion of PSP patients, which was not observed in controls, is indicative of an excess of anaerobic glycolysis over aerobic oxidative phosphorylation (Prichard et al, 1991), suggesting that ATP generation through anaerobic glycolysis is increased to compensate for the reduction in ATP generation caused by defective oxidative phosphorylation. The reduction of pCre, the second high-energy metabolite that contributes to rapid ATP production in situations of high-energy demand, is also compatible with an impairment of oxidative phosphorylation that prevents the regeneration of this metabolite. Our earlier published observation that oral administration of coenzyme Q10, a physiologic cofactor of complex I of the mitochondrial respiratory chain, improves not only clinical disability scores but also the increased ATP/ADP, pCre/uCre and HEP/LEP ratios measured by the same ³¹P-MRS and ¹H-MRS method, in the same PSP patients studied here (Stamelou et al, 2008), which provides further support for the concept of a functionally relevant impairment of oxidative phosphorylation in PSP.

Figure 2

Schematic representation of the basic pathways of energy metabolism and the alterations observed in PSP patients in this study. (A) Hydrolysis of ATP to ADP and Pi is the chemical reaction by which energy is released for biochemical reactions; the ratio ATP/(ADP × Pi) (i.e., phosphorylation potential (PP)) determines the energy level driving all cellular energy-dependent processes. (B) The main function of the mitochondrial respiratory chain (consisting of the enzymatic complexes I through V) is to maintain adequate cellular concentrations of ATP by aerobic oxidative phosphorylation. (C) Comparably small amounts of ATP are produced by anaerobic glycolysis. (D) ATP is in equilibrium with large cellular pools of pCre, which is a rapidly accessibly energy storage compound; in situations of high-energy demand, pCre is dephosphorylated to uCre, thereby charging ADP by phosphorylation to yield ATP; inversely, ATP is dephosphorylated to recharge uCre to pCre during recovery. The alterations in the concentrations of energy metabolites measured in the basal ganglia of PSP patients are indicated (⊖ decrease, ⊕ increase). As specified in detail in the Discussion section, these alterations are compatible with a primary impairment of the mitochondrial respiratory chain: impaired oxidative phosphorylation (panel B) leads directly to reduced ATP levels (panel A); low ATP levels lead to reduced regeneration of pCre (panel D); the PP may be stabilized by increased translocation of ADP and Pi to the inner mitochondrial membrane (A → B), where they are invisible to MRS; an excess of unimpaired anaerobic glycolysis over impaired aerobic oxidative phosphorylation leads to increased lactate levels (panel C).

It is important to note that brain mitochondria can tolerate a 60% decrease in complex I activity, without a significant decrease in ATP production (Davey et al, 1997). Inversely, the presence of a small significant ATP loss strongly suggests the presence of a much more severe inhibition of the respiratory chain. The reduction in high-energy metabolites in PSP was detected at rest, whereas in Parkinson's disease, another neurodegenerative disorder with presumed mitochondrial complex I dysfunction (Schapira, 2006), similar metabolic alterations were only observed after functional activation (Rango et al, 2006). This may suggest that the mitochondrial dysfunction in PSP is more severe than in Parkinson's disease. Therefore, we are tempted to conclude that the small, yet significant alterations in the cerebral high-energy metabolites determined in this study provide important insights into the pathophysiology of PSP.

It is interesting that the decrease in high-energy metabolites observed in the PSP patients was not accompanied by an increase in their low-energy counterparts, as it occurs during acute ischemia (Hems and Brosnan, 1970) or in Leber's hereditary optic neuropathy (Lodi et al, 2000). However, there is solid in vitro and in vivo experimental evidence that mammalian cells, in which oxidative phosphorylation is impaired, have the potential to downregulate the free cytosolic fraction of uCre, ADP and Pi, by (1) increasing the protein-bound fraction of uCre in the cytoplasm and (2) by the compartmentalization of ADP and Pi in the inner mitochondrial membrane, which would render them immobile and thus invisible by MRS (Iles et al, 1985; Balaban, 1990; Brown, 1992; Iotti et al, 1996). Such a compensatory counterregulation might be the biochemical basis for the changes in cerebral energy metabolites in PSP observed in this study. Such a regulation would also appear biologically meaningful, as it would, first, prevent the inactivation of ATPases by abnormally high cytosolic concentrations of ADP and Pi. Second, increased translocation of ADP and Pi to the inner mitochondrial membrane would stimulate the rate of oxidative phosphorylation. Third, downregulation of cytosolic LEPs in a metabolic state, in which HEPs are chronically reduced would stabilize the phophorylation potential PP (Iles et al, 1985; Balaban, 1990; Brown, 1992) (Figure 2).

Indeed, the fact that the PP [=ATP/(ADP × Pi)] and the other ratios of high-energy metabolites when compared with their low-energy counterparts (ATP/ADP, pCre/uCre and HEP/LEP), which determine the energy level driving all cellular energy-dependent processes (Iles et al, 1985; Pietz et al, 2003; Schapira, 2006), was not significantly altered in the patients suggests that the affected cells can still compensate, at rest, for the decrease in high-energy metabolites (Iles et al, 1985; Pietz et al, 2003; Schapira, 2006). This may mean that the neurons in the affected brain areas in the early-stage PSP studied were not dysfunctional at rest. This perspective is consistent with the observation that the severity of the clinical symptoms was not correlated with the concentrations of HEPs or the PP in PSP patients. Further, the observed metabolic deficit, particularly the ATP depletion, may render the affected neurons more prone to neurodegeneration and somatodendritic tau accumulation, as we have shown experimentally (Höglinger et al, 2005; Escobar-Khondiker et al, 2007).

In summary, our data provide the first in vivo evidence of a significant depletion in high-energy metabolites at rest in vulnerable brain areas of early-stage PSP patients without significant alterations in the NAA/tCre ratio. The observed pattern of metabolic alterations is more compatible with a primary decrease in oxidative phosphorylation than with a secondary consequence of neurodegeneration. This suggests that mitochondrial dysfunction is indeed crucially implicated in the pathophysiology of PSP, not a consequence thereof. Interestingly, the difference between the concentrations of HEPs in the ‘basal ganglia’ ROI of early-stage PSP patients and healthy controls was such that these metabolites might even prove to be helpful diagnostic markers. This should be examined further in independent samples.

Footnotes

Acknowledgements

We thank O Arias-Carrión for skilful help with the graphics.

The authors declare no financial interest related to the materials and methods presented here.

References

Abe

Terakawa

Takanashi

Watanabe

Tanaka

Fujita

Hirabuki

Yanagihara

(2000) Proton magnetic resonance spectroscopy of patients with parkinsonism. Brain Res Bull 52:589–95

Albers

Swerdlow

Manfredi

Gajewski

Yang

Parker

Jr Beal

(2001) Further evidence for mitochondrial dysfunction in progressive supranuclear palsy. Exp Neurol 168:196–8

Balaban

(1990) Regulation of oxidative phosphorylation in the mammalian cell. Am J Physiol 258:377–89

Brown

(1992) Control of respiration and ATP synthesis in mammalian mitochondria and cells. Biochem J 284:1–13

Buchli

Duc

Martin

Boesiger

(1994) Assessment of absolute metabolite concentrations in human tissue by 31P MRS in vivo. Part I: Cerebrum, cerebellum, cerebral gray and white matter. Magn Reson Med 32:447–52

Champy

Höglinger

Féger

Gleye

Hocquemiller

Laurens

Guérineau

Laprévote

Medja

Lombès

Michel

Lannuzel

Hirsch

Ruberg

(2004) Annonacin, a lipophilic inhibitor of mitochondrial complex I, induces nigral and striatal neurodegeneration in rats: possible relevance for atypical parkinsonism in Guadeloupe. J Neurochem 88:63–9

Chirichigno

Manfredi

Beal

Albers

(2002) Stress-induced mitochondrial depolarization and oxidative damage in PSP cybrids. Brain Res 951:31–5

Davey

Canevari

Clark

(1997) Threshold effects in synaptosomal and nonsynaptic mitochondria from hippocampal CA1 and paramedian neocortex brain regions. J Neurochem 69:2564–70

Di Monte

Harati

Jankovic

Sandy

Jewell

Langston

(1994) Muscle mitochondrial ATP production in progressive supranuclear palsy. J Neurochem 62:1631–4

10.

Dubois

Slachevsky

Litvan

Pillon

(2000) The FAB: a frontal assessment battery at bedside. Neurology 55:1621–6

11.

Escobar-Khondiker

Höllerhage

Muriel

Champy

Bach

Depienne

Respondek

Yamada

Lannuzel

Yagi

Hirsch

Oertel

Jacob

Michel

Ruberg

Höglinger

(2007) Annonacin, a natural mitochondrial complex I inhibitor, causes tau pathology in cultured neurons. J Neurosci 27: 7827–37

12.

Federico

Simone

Lucivero

De Man

Giannini

Iliceto

Mezzapesa

Lamberti

(1997) Proton magnetic resonance spectroscopy in Parkinson's disease and progressive supranuclear palsy. J Neurol Neurosurg Psych 62:239–42

13.

Federico

Simone

Lucivero

Iliceto

De Mari

Giannini

Mezzapesa

Tarantino

Lamberti

(1998) Proton magnetic resonance spectroscopy in Parkinson's disease and atypical parkinsonian disorders. Mov Disord 12:903–9

14.

Foster

Gilman

Berent

Morin

Brown

Koeppe

(1988) Cerebral hypometabolism in progressive supranuclear palsy studied with positron emission tomography. Ann Neurol 24:399–406

15.

Golbe

Ohman-Strickland

(2007) A clinical rating scale for progressive supranuclear palsy. Brain 130:1552–65

16.

Golbe

(1997) A clinical rating scale and staging system for progressive supranuclear palsy. Neurology 48:326

17.

Hattingen

Pilatus

Franz

Zanella

Lanfermann

(2007) Evaluation of optimal echo time for ¹H-spectroscopic imaging of brain tumors at 3 Tesla. J Magn Reson Imaging 26:427–31

18.

Hauw

Daniel

Dickson

Horoupian

Jellinger

Lantos

McKee

Tabaton

Litvan

(1994) Preliminary NINDS neuropathologic criteria for Steele–Richardson–Olszewsky syndrome (progressive supranuclear palsy). Neurology 44:2015–9

19.

Hems

Brosnan

(1970) Effects of ischaemia on content of metabolites in rat liver and kidney in vivo. Biochem J 120:105–11

20.

Hetherington

Spencer

Vaughan

Pan

(2001) Quantitative (31) P spectroscopic imaging of human brain at 4 Tesla: assessment of gray and white matter differences of phosphocreatine and ATP. Magn Reson Med 45:46–52

21.

Höglinger

Féger

Prigent

Michel

Parain

Champy

Ruberg

Oertel

Hirsch

(2003) Chronic systemic complex I inhibition induces a hypokinetic multisystem degeneration in rats. J Neurochem 84:491–502

22.

Höglinger

Lannuzel

Khondiker

Michel

Duyckaerts

Féger

Champy

Prigent

Medja

Lombes

Oertel

Ruberg

Hirsch

(2005) The mitochondrial complex I inhibitor rotenone triggers a cerebral tauopathy. J Neurochem 95:930–9

23.

Taylor-Robinson

Chaudhuri

Bell

Labbé

Cunningham

Koepp

Hammers

Morris

Turjanski

Brooks

(2000) Cortical dysfunction in non-demented Parkinson's disease patients: a combined (31)P-MRS and (18)FDG-PET study. Brain 123:340–52

24.

Iles

Stevens

Griffiths

Morris

(1985) Phosphorylation status of liver by 31P-n.m.r. spectroscopy, and its implications for metabolic control. A comparison of 31P-n.m.r. spectroscopy (in vivo and in vitro) with chemical and enzymic determinations of ATP, ADP and Pi. Biochem J 229:141–51

25.

Iotti

Lodi

Gottardi

Zaniol

Barbiroli

(1996) Inorganic phosphate is transported into mitochondria in the absence of ATP biosynthesis: an in vivo 31P NMR study in the human skeletal muscle. Biochem Biophys Res Commun 225:191–4

26.

Isobe

Matsumura

Anno

Kawamura

Shibata

Muraishi

Minami

(2007) Lactate quantification by proton magnetic resonance spectroscopy using a clinical MRI machine: A basic study. Australasian Radiol 51:330–3

27.

Lannuzel

Hoglinger

Verhaeghe

Gire

Belson

Escobar-Khondiker

Poullain

Oertel

Hirsch

Dubois

Ruberg

(2007) Atypical parkinsonism in Guadeloupe: a common risk factor for two closely related phenotypes? Brain 130:816–27

28.

Litvan

Agid

Calne

Campbell

Dubois

Duvoisin

Goetz

Golbe

Grafman

Growdon

Hallett

Jankovic

Quinn

Tolosa

Zee

(1996) Clinical research criteria for the diagnosis of progressive supranuclear palsy (Steele–Richardson–Olszewski syndrome): report of the NINDS-SPSP International Workshop. Neurology 47:1–9

29.

Lodi

Montagna

Cortelli

Iotti

Cevoli

Carelli

Barbiroli

(2000) ‘Secondary’ 4216/ND1 and 13708/ND5 Leber's hereditary optic neuropathy mitochondrial DNA mutations do not further impair in vivo mitochondrial oxidative metabolism when associated with the 11778/ND4 mitochondrial DNA mutation. Brain 123:1896–902

30.

Martinelli

Scaglione

Lodi

Iotti

Barbiroli

(2000) Deficit of brain and skeletal muscle bioenergetics in progressive supranuclear palsy shown in vivo by phosphorus magnetic resonance spectroscopy. Mov Disord 15:889–93

31.

Michaelis

Merboldt

Bruhn

Hänicke

Frahm

(1993) Absolute concentrations of metabolites in the adult human brain in vivo: quantification of localized proton MR spectra. Radiology 187:219–27

32.

Moffett

Ross

Arun

Madhavarao

Namboodiri

MAA

(2007) N-Acetylaspartate in the CNS: from neurodiagnostics to neurobiology. Prog Neurobiol 81:89–131

33.

Penn

Roberts

Hodder

Allen

Zhu

Martin

(1995) Generalized mitochondrial dysfunction in Parkinson's disease detected by magnetic resonance spectroscopy of muscle. Neurology 45: 2097–9

34.

Petroff

Prichard

Behar

Alger

den Hollander

Shulman

(1985) Cerebral intracellular pH by 31P nuclear magnetic resonance spectroscopy. Neurology 35:781–8

35.

Pietz

Rupp

Ebinger

Rating

Mayatepek

Boesch

Kreis

(2003) Cerebral energy metabolism in phenylketonuria: findings by quantitative in vivo P MR spectroscopy. Pediatr Res 53:654–62

36.

Prichard

Rothman

Novotny

Petroff

Kuwabara

Avison

(1991) Lactate rise detected by 1H NMR in human visual cortex during physiologic stimulation. Proc Natl Acad Sci USA 88:5829–31

37.

Provencher

(1993) Estimation of metabolite concentrations from localized in vivo proton MR spectroscopy. Magn Reson Med 30:672–9

38.

Rango

Arighi

Biondetti

Barberis

Bonifati

Blandini

Pacchetti

Martignoni

Bresolin

Nappi

(2007) Magnetic resonance spectroscopy in Parkinson's disease and parkinsonian syndromes. Functional Neurology 22:75–9

39.

Rango

Bonifati

Bresolin

(2006) Parkinson's disease and brain mitochondrial dysfunction: a functional phosphorus magnetic resonance spectroscopy study. J Cereb Flow Metab 26:283–90

40.

Schapira

(2006) Mitochondrial disease. Lancet 368: 70–82

41.

Stamelou

Reuss

Pilatus

Magerkurth

Niklowitz

Eggert

Krisp

Menke

Schade-Brittinger

Oertel

Höglinger

(2008) Short-term effects of coenzyme Q10 in progressive supranuclear palsy: a randomized, placebo-controlled trial. Mov Disord 23:942–9

42.

Swerdlow

Golbe

Parks

Cassarino

Binder

Grawey

Litvan

Bennett

Jr Wooten

Parker

(2000) Mitochondrial dysfunction in cybrid lines expressing mitochondrial genes from patients with progressive supranuclear palsy. J Neurochem 75:1681–4

43.

Tedeschi

Litvan

Bonavita

Bertolino

Lundbom

Patronas

Hallett

(1997) Proton magnetic resonance spectroscopic imaging in progressive supranuclear palsy, Parkinson's disease and corticobasal degeneration. Brain 120:1541–52

44.

van Balken

Litvan

(2006) Current and future treatments in progressive supranuclear palsy. Curr Treat Options Neurol 8:211–23

45.

Vanhamme

van den Boogaart

Van Huffel

(1997) Improved method for accurate and efficient quantification of MRS data with use of prior knowledge. J Magn Reson 129:35–43