LETTERS

Very Late Onset in Ataxia Oculomotor Apraxia

Type I
Chiara Criscuolo, MD.1,2 Pietro Mancini, PhD,1
Valeria Menchise, PhD,3,4 Francesco Saccà, MD,1
Giuseppe De Michele, MD,1 Sandro Banfi, PhD,2 and
Alessandro Filla, MD.1

Missense and truncating mutation in APTX gene are respon-

sible of ataxia with oculomotor apraxia type 1 (AOA1) an
autosomal recessive disorder characterized by early onset,
progressive ataxia, peripheral neuropathy, oculomotor
apraxia, and cerebellar atrophy.1,2 We have already reported
two patients with onset at 25 and 29 years suggesting that
AOA1 should be considered in patients with onset after 25
years.3 Here, we report a patient with a very late onset and a
new mutation further expanding the clinical and molecular
spectrum of the disease.
The coding sequence of APTX was analyzed using dena-
turing high-performance liquid chromatography in 15 index
cases selected among 90 sporadic late-onset non–Friedreich Fig. Superposition of the overall fold of the homology models
ataxia (FRDA) patients. The inclusion criteria were onset be- built for H166-F282 aprataxin and its L223P mutant with
tween 25 and 50 years, sporadic occurrence, and clinical 3D-PSSM software.5 ␣-Helices and ␤-strands are represented in
and/or neurophysiological signs of peripheral neuropathy. gray and black, respectively. The alignment between H166-
APTX analysis showed a new homozygous 6668T3 C mu- F282 sequence and the template (protein kinase C inhibitor–1;
tation (exon 5) in one patient, resulting in the substitution pdb code ⫽ 1kpf) shows short deletions only in loop regions.
of proline for leucine at codon 223 (L223P). Onset of the
disease was at age 40 years with ataxia followed by dysarthria. Departments of 1Neurological Sciences, Federico II University;
Neurological examination showed mild gait and limb ataxia, 2
Telethon Institute of Genetics and Medicine; 3Istituto di
slurred speech, dysphagia, normal ocular movements, tongue Biostrutture e Bioimmagini CNR, Naples; and 4Bioindustry
and limb fasciculations, brisk jaw jerk, areflexia, and de- Park Canavese, Colleretto Giacosa, Italy
creased vibration sense at the external malleoli. Normal se-
rum creatine kinase and albumin, mild familial hypercholes-
References
terolemia, and hypertrigliceridemia were observed.
1. Moreira MC, Barbot C, Tachi N, et al. The gene mutated in
Motor (47m/sec in the elbow-wrist segment; lower limit ataxia-ocular apraxia 1 encodes the new HIT/Zn-finger protein
54) and sensory (43.3 m/sec in the digit III–wrist segment; aprataxin. Nat Genet 2001;29:189 –193.
lower limit 51) conduction velocities were slightly decreased at 2. Date H, Onodera O, Tanaka H, et al. Early-onset ataxia with ocular
the median nerve, and sensory action potential was markedly motor apraxia and hypoalbuminemia is caused by mutations in a
reduced at wrist (1.8 ␮V; lower limit 6). Electromyogram new HIT superfamily gene. Nat Genet 2001;29:184 –188.
showed a neurogenic pattern in the anterior tibial muscle. 3. Criscuolo C, Mancini P, Saccà F, et al. Ataxia with oculomotor
Brain magnetic resonance imaging showed cerebellar atrophy. apraxia type 1 in southern Italy: late onset and variable pheno-
After 20 years of disease, the patient walks without supports type. Neurology 2004;63:2173–2175.
and is still working, and only speech is significantly worsened. 4. Amouri R, Moreira MC, Zouari M, et al. Aprataxin gene muta-
tions in Tunisian families. Neurology 2004;63:928 –929.
These findings suggest that the new mutation is associated
5. Fischer D, Barret C, Bryson K, et al. CAFASP-1: critical assess-
with a peculiarly mild phenotype. Homology modeling calcu- ment of fully automated structure prediction methods. Proteins
lations performed for aprataxin and its L223P mutant showed 1999;(suppl 3):209 –217.
that H166-F282 sequence is compatible with a HIT-like do-
main for both proteins (Fig). However, the P223L substitu- DOI: 10.1002/ana.20463
tion falls in the middle of an ␣-helix, and proline is known to
have a low propensity for an ␣-helix formation. Thus, such a High Endogenous Cannabinoid Levels in the
point mutation may lead to some destabilization of the protein Cerebrospinal Fluid of Untreated Parkinson’s
fold. L223P is not present in the healthy brother, is conserved Disease Patients
in human (Gene Bank accession number AY040777) and
Antonio Pisani, MD,1 Filomena Fezza, PhD,2
mouse (NM175073), and was absent in 150 southern Italian Salvatore Galati, MD,1 Natalia Battista, PhD,3
controls. Simone Napolitano, MD,1 Alessandro Finazzi-Agrò, MD,2
We emphasize the potential pitfall when selecting only pa- Giorgio Bernardi, MD,1 Livia Brusa, MD, PhD,1
tients with early onset for APTX screening. The identifica- Mariangela Pierantozzi, MD,1 Paolo Stanzione, MD,1
tion of patients with preserved knee jerks suggests also that and Mauro Maccarrone, MD, PhD3,4
neuropathy is inconstant.4 To date, mutations have been
found only in the last three APTX exons. Thus, the screening Endocannabinoids are a class of bioactive lipids responsible
of such exons could be useful in sporadic or recessive patients for the activation of type 1 (CB1) and type 2 (CB2) can-
with cerebellar atrophy, after exclusion of FRDA. nabinoid receptors. A close functional interaction between

© 2005 American Neurological Association 777

Published by Wiley-Liss, Inc., through Wiley Subscription Services
dopaminergic transmission and the endocannabinoid sys-
tem has been proposed in motor function,1 and indeed
CB1 receptors are densely expressed in the striatum, where
this interaction is thought to occur.1,2 Recently anandam-
ide (N-arachidonoylethanolamine), the best characterized
endocannabinoid, has been measured in human cerebrospi-
nal fluid (CSF) of treated psychotic patients.2 In a rat
model of Parkinson’s disease (PD), an increase in anand-
amide levels accompanies dopamine cell loss induced by in-
jections of 6-hydroxy-dopamine (6-OHDA).3 To verify the
hypothesis of a functional interaction between dopaminer-
gic transmission and endocannabinoids in humans, we en-
rolled 16 patients affected by clinically defined PD in this
study. Seven patients met the criteria for de novo PD. The
remaining nine were patients with early-mild or advanced
PD who underwent a complete drug washout before CSF
examination, to assess disease progression independently of
this study. The control group consisted of 14 age-matched
patients who underwent lumbar puncture as part of the di-
agnostic process. To evaluate endogenous anandamide lev-
els, we extracted lipids from 1ml CSF as reported,3 and the
organic phase was dried under nitrogen. In control subjects,
the average concentration of CSF anandamide was 5.23 ⫾
1.81pmol/ml (Fig. 1). Conversely, in PD patients anand-
amide levels were more than doubled (Fig; 10.7 ⫾
3.2pmol/ml; p ⬍ 0.001). It was noteworthy that anandam-
ide levels in PD patients were independent of the stage of
disease, Unified Parkinson’s Disease Rating Scale score, or
Fig. Endogenous levels of anandamide (AEA) in the cerebrospi-
their belonging to the de novo group or to the group of
nal fluid (CSF) of control and Parkinson’s disease (PD) pa-
patients undergoing drug washout. tients. The top panel shows the levels of AEA in each subject,
Our findings represent the first evidence ever of abnor- expressed as picomoles per milliliter of CSF. The bottom panel
mally high levels of the endogenous cannabinoid anandam- summarizes the data of each group, reported as means ⫾ stan-
ide in untreated PD patients. The notion that endocan- dard deviations (vertical bars). Statistical analysis was per-
nabinoids are hyperactive in the striatum in the course of formed by the nonparametric Mann–Whitney U test, elaborat-
PD is supported by several lines of experimental evidence. ing experimental data by means of the InStat 3 program
Increased anandamide levels were reported in rat models of (GraphPAD Software for Science, San Diego, CA). Statistical
PD.3,4 Accordingly, CB1 receptors were found at higher significance was set at p value less than 0.001. The Unified
levels in postmortem PD striata and in L-dopa–treated Parkinson’s Disease Rating Scale (UPDRS), motor section III
MPTP-treated monkeys.5 Of interest, anandamide was score, was conducted by an expert neurologist. All subjects gave
shown to decrease the hyperactivity of glutamatergic corti- their written informed consent to the procedures, which had
costriatal pathways in 6-OHDA–denervated rats and to been approved previously by the local ethics committee. CSF
stimulate dopamine release.2,3 Together, these observations samples were collected and routine analyses were performed, in-
suggest that the increase of anandamide might reflect a cluding CSF/serum albumin and IgG quotients, and determina-
compensatory mechanism occurring in the striatum of PD tion of oligoclonal bands. All CSF samples were normal, as
patients, aimed at normalizing chronic dopamine depletion, were routine blood analyses. To evaluate endogenous anandam-
ide levels, we extracted lipids from 1ml CSF as reported,3 and
thus extending for the first time to humans previous data
the organic phase was dried under nitrogen. Dry pellet was re-
on animal models of PD. Understanding the functional in-
suspended in 20␮l of methanol and was analyzed by reverse-
terplay between dopamine and the endocannabinoid system phase high-performance liquid chromatography after derivatiza-
opens new perspectives for the treatment of PD. tion with 4-(N-chloroformylmytlmethyl-N-methyl)-amino-7-
N,N-dimethylaminosulphonyl-2,1,3-benzoxa-diazole (Tokyo
Kasei Kogyo Co., Tokyo, Japan); CSF anandamide levels are
expressed as pmol/ml.
This study was supported by Ministry Education, University and
Research, (Confinaziamento 2003, M.M.; 2004, G.B.; Fondo Inves-
timenti Ricerca di Base and Fondo Integrativo Speciale Ricerca,
A.P.), Ministero della Salute (Progetto di Ricerca Finalizzata Fonda-
zione Santa Lucia 2004, A.F.-A.), and Ministero della Salute-
Regione Lazio (G.B., A.P.).

778 Annals of Neurology Vol 57 No 5 May 2005

1
Neurology Clinic, Department of Neuroscience, University of Table. The Association between Serum HDL Levels and
Rome “Tor Vergata” and Santa Lucia Foundation “Istituto di Hippocampal Volume on MRI in the Rotterdam Scan Study
Ricerca e Cura a Carattere Scientifico (IRCCS); 2Department
of Experimental Medicine and Biochemical Sciences, Hippocampal Volume
University of Rome “Tor Vergata”; 3Department of
Regression Coefficient
Biochemical Sciences, University of Teramo and Center of (95% CI)a p
Experimental Neurobiology C. Mondino, Mondino-Tor
Vergata-Santa Lucia, IRCSS, Rome, Italy. HDL level 5 years before MRI
Total (n ⫽ 503) ⫺0.07 (⫺0.31 to 0.17) 0.56
References Age 75–85 yr (n ⫽ 165) ⫺0.06 (⫺0.45 to 0.33) 0.76
1. Piomelli D. The molecular logic of endocannabinoid signalling. HDL level at time of MRI
Nat Rev Neurosci 2003;4:873– 884. Total (n ⫽ 503) ⫺0.09 (⫺0.31 to 0.13) 0.43
2. Giuffrida A, Leweke FM, Gerth CW, et al. Cerebrospinal anan- Age 75–85 yr (n ⫽ 165) ⫺0.20 (⫺0.58 to 0.18) 0.30
damide levels are elevated in acute schizophrenia and are in- a
versely correlated with psychotic symptoms. Neuropsychophar- Regression coefficients from multiple linear regression analyses were
adjusted for age, sex, and lipid-lowering drugs.
macology 2004;29:2108 –2114.
3. Gubellini P, Picconi B, Bari M, et al. Experimental parkinsonism HDL ⫽ high-density lipoprotein; MRI ⫽ magnetic resonance im-
alters endocannabinoid degradation: implications for striatal glu- aging; CI ⫽ confidence interval.
tamatergic transmission. J Neurosci 2002;22:6900 – 6907.
4. Di Marzo V, Hill MP, Bisogno T, et al. Enhanced levels of en-
dogenous cannabinoids in the globus pallidus are associated with
a reduction in movement in an animal model of Parkinson’s dis-
ease. FASEB J 2000;14:1432–1438. merely reflect frailty and poor health status.3 Cholesterol lev-
5. Hurley MJ, Mash DC, Jenner P. Expression of cannabinoid CB1 els decrease with more advanced Alzheimer’s disease.4 Wolf’s
receptor mRNA in basal ganglia of normal and parkinsonian hu- main analyses were based on the total group of 86 subjects
man brain. J Neural Transm 2003;110:1279 –1288. and may have been affected by the inclusion of persons with
Alzheimer’s disease. They still found an association between
DOI: 10.1002/ana.20462 low HDL and small hippocampal volumes after exclusion of
Alzheimer’s disease patients. However, persons with mild
Serum Lipids and Hippocampal Volume: The cognitive impairment may have early Alzheimer’s disease and
Link to Alzheimer’s Disease? be more frail and have lower HDL levels and smaller hip-
Tom den Heijer, MD, PhD,1,2 pocampal volumes than nondemented elderly. It would be of
Albert Hofman, MD, PhD,1 interest to examine associations for nondemented subjects
Peter J. Koudstaal, MD, PhD,2 separately.
and Monique M. B. Breteler, MD, PhD1 In our study, almost 9% of subjects used lipid-lowering
drugs at time of MRI. Exclusion of those persons did not
Wolf and colleagues recently reported a positive association change our results. Wolf and colleagues provide no informa-
between high-density lipoprotein (HDL) cholesterol levels tion on drug use, and we wonder whether drug use was sim-
and hippocampal volume on magnetic resonance imaging ilar among the three cognitive groups.
(MRI).1 They suggest that high HDL levels could protect In conclusion, in our larger sample of nondemented el-
against hippocampal atrophy and Alzheimer’s disease. In derly we found no support for the suggestion by Wolf and
the Rotterdam Scan Study, a population-based MRI study, colleagues that increasing HDL levels may reduce hippocam-
we sought to replicate their findings. The protocol of the pal atrophy and the risk of Alzheimer’s disease.
Rotterdam Scan Study has been described elsewhere.2 In
511 nondemented subjects (aged 60 –90 years; mean, 72
This study was supported by grants from the Netherlands Organi-
years; 52% women), we measured hippocampal volumes on
sation for Scientific Research (NWO 904-61-096) and the Health
MRI. Nonfasting serum HDL levels taken on average 5 Research and Development Council (Z0N, A.H., M.M.B.B.).
years before MRI (range, 3–7 years; mean HDL,
1.3mmol/L; standard deviation [SD], 0.3) and at time of
MRI (mean HDL, 1.3mmol/L; SD 0.4) were available for Departments of 1Epidemiology and Biostatistics and
2
503 subjects. HDL cholesterol levels were not related to Neurology, Erasmus MC, Rotterdam, the Netherlands
hippocampal volume (Table), neither in the total group nor
when we restricted the analyses to subjects in the age range References
that Wolf and colleagues studied (75– 85 years). Subgroup 1. Wolf H, Hensel A, Arendt T, et al. Serum lipids and hippocam-
analyses showed no differences between men and women or pal volume: the link to Alzheimer’s disease? Ann Neurol 2004;
between APOE ε4 carriers and noncarriers. 56:745–749.
2. den Heijer T, Vermeer SE, Clarke R, et al. Homocysteine and
Wolf’s study included nondemented elderly (n ⫽ 26),
brain atrophy on MRI of non-demented elderly. Brain 2003;
mild cognitively impaired (n ⫽ 35), and dementia patients 126:170 –175.
(n ⫽ 25). Dementia patients had the lowest HDL levels and 3. Volpato S, Leveille SG, Corti MC, et al. The value of serum
the smallest hippocampal volumes, which led the authors to albumin and high-density lipoprotein cholesterol in defining
suggest that low HDL is a risk factor for Alzheimer pathol- mortality risk in older persons with low serum cholesterol. J Am
ogy. An alternative explanation is that low HDL levels Geriatr Soc 2001;49:1142–1147.

Annals of Neurology Vol 57 No 5 May 2005 779