Molecular Psychiatry (2004) 9, 1122–1128 & 2004 Nature Publishing Group All rights reserved 1359-4184/04 $30.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE Endothelin-converting enzyme-1 is expressed in human cerebral cortex and protects against Alzheimer’s disease B Funalot1, T Ouimet1, A Claperon1, C Fallet2, A Delacourte3, J Epelbaum4, T Subkowski5,NLe´onard2, V Codron6, J-P David3,7, P Amouyel6, J-C Schwartz1 and N Helbecque6 1Institut National de la Sante´ et de la Recherche Me´dicale (INSERM) Unit 573, Paris, France; 2Department of Neuropathology, Hoˆpital Sainte-Anne, Paris, France; 3INSERM Unit 422, Lille, France; 4INSERM Unit 549, Paris, France; 5BASF Main Laboratories, Ludwigshafen, Germany; 6INSERM Unit 508, Institut Pasteur, Lille, France; 7Department of Gerontology, Hoˆpital Emile-Roux, Limeil-Bre´vannes, France

Cerebral accumulation of b-amyloid peptide (Ab) is a central event in the pathogenesis of Alzheimer’s disease (AD). Endothelin-converting enzyme-1 (ECE-1) is a candidate Ab-degrading enzyme in brain, but its involvement in AD pathogenesis was never assessed. We first performed brain immunocytochemistry, using a monoclonal anti-ECE-1 antibody, and observed neuronal ECE-1 expression in various cortical regions of nondemented subjects. In the hippocampus, ECE- 1 immunoreactivity showed a stereotypical pattern inversely correlated with susceptibility to Ab deposition, further suggesting a physiological role in Ab clearance. In order to undertake a genetic association study, we identified a functional genetic variant (ECE1B C-338A) located in a regulatory region of the ECE1 gene. We showed that the A allele is associated with increased transcriptional activity in promoter–reporter gene assays and with increased ECE-1 mRNA expression in human neocortex. In a case–control study involving 401 patients with late-onset AD and 461 aged controls, we found that homozygous carriers of the A allele had a reduced risk of AD (OR ¼ 0.47, 95% CI 0.25–0.88). This finding was strengthened by the analysis of two other genetic variants of the ECE1 gene, which showed that the genetic association is extended over at least 13 kilobases of the gene sequence. Our results suggest that ECE-1 expression in brain may be critical for cortical Ab clearance and offer new potential targets for therapeutic interventions in AD. Molecular Psychiatry (2004) 9, 1122–1128. doi:10.1038/sj.mp.4001584 Published online 31 August 2004 Keywords: endothelin-converting enzyme; Alzheimer’s disease; late-onset Alzheimer’s disease; b-amyloid peptide; clearance; zinc metalloprotease; immunohistochemistry; genetic polymorph- ism; genetic association study

Several lines of evidence support an early pathogenic 3.4.24.11), endothelin-converting enzyme-1 (ECE-1, role for the b-amyloid peptide (Ab) in Alzheimer’s EC 3.4.24.71), endothelin-converting enzyme-2 (ECE- disease (AD).1–3 Ab is a physiological peptide con- 2) and insulysin (EC 3.4.24.56), have been shown to tinuously synthesized in brain, produced by cleavage influence endogenous Ab levels in the brains of their of the amyloid precursor protein (APP). In familial respective knockout mice.11–14 Neprilysin, ECE-1 and cases of early-onset AD, various mutations of the ECE-2 are type II integral membrane peptidases, genes encoding APP and presenilins have been classified as members of the M13 family,15 whereas shown to increase Ab production.3,4 By contrast, the insulysin is a cytosolic enzyme belonging to the M16 mechanisms of cortical Ab accumulation in late-onset family.16 The four enzymes are suspected to have a AD (LOAD) patients appear to be more complex and physiological role in Ab clearance in human brain, may involve increased expression and activity of the and therefore to protect against development of AD, b-site APP-cleaving enzyme (BACE-1),5–7 increased but direct evidence of their involvement in AD aggregation of Ab fibrils or decreased Ab clearance.8 pathogenesis is lacking. Neprilysin immunoreactivity Several proteases were shown to hydrolyze Ab in appears to be weak or undetectable in human cerebral vitro or in cell-based assays, and are likely candidates cortex.17,18 By contrast, insulysin immunoreactivity is for a role in Ab clearance in brain.9,10 Among these, marked in hippocampal and neocortical neurons.19,20 only four zinc metalloproteases: neprilysin (EC Several genetic association studies tested for an association between neprilysin or insulysin gene variants and AD, but gave controversial results.21–25 Correspondence: Dr B Funalot, MD, INSERM Unit 573, Centre To date, the expression of ECE-1 in human brain had Paul-Broca, 2ter rue d’Ale´sia, 75014 Paris, France. never been studied, and its potential role in the E-mail: [email protected] Received 16 September 2003; revised 05 May 2004; accepted 19 pathogenesis of AD had never been assessed. We May 2004 have addressed this question by a combination of ECE-1 and Alzheimer’s disease B Funalot et al 1123 expression and genetic studies, and provide evidence gene constructs were transiently transfected in the for a protective role of ECE-1 against AD in the aging ECV304 cell line. The pcDNA3.1/HisB/LacZ plasmid brain. (Invitrogen), containing the b-galactosidase gene under the control of a CMV promoter, was used as an internal standard. SEAP activity was measured in a Materials and methods plate fluorometer (Dynatech) using 4-methylumbelli- Brain immunocytochemistry feryl phosphate as a substrate, and normalized with Brains from eight nondemented subjects were stu- respect to b-galactosidase activity. Results represent died. Ages at death were 38, 56, 82, 84, 85, 86, 92 and the mean of at least six transfection experiments for 99 years. Brain specimens were fixed in 10% each construct. formaldehyde and embedded in paraffin. ECE-1 immunocytochemistry (ICC) was performed using Relative quantitative RT-PCR mouse monoclonal antibodies directed against ECE- We could obtain neocortical samples from seven aged 1 common C-terminal sequence (B61/104, diluted nondemented subjects. All of them were hospitalized 1 : 50, and E15/6, diluted 1 : 20). It was previously for various disorders and died in public hospitals. shown that both antibodies are specific of ECE-1.26 Samples of Brodmann area 9 (prefrontal cortex) were However, we did not test these antibodies against the deep-frozen at autopsy (excluding patients with post- ECE-2 protein, which has B60% identity with ECE-1. mortem delay superior to 24 h) and subsequently Sections were microwaved in citrate buffer, followed stored at À801C. Total RNA was extracted from frozen by endogenous peroxydases inactivation (3% H2O2) samples using the TRIzol method (Invitrogen), fol- and treatment with the blocking agent from the lowing the manufacturer’s instructions. In all, 1 mgof universal HRP immunostaining kit (Immunotech). total RNA was reverse-transcripted using random Primary antibody was identified with a biotinylated hexamers and AMV-reverse transcriptase (Finn- secondary antibody and the avidin–biotin complex, zymes), in a volume of 25 ml, following the manufac- and visualized with 3,3-diaminobenzidine. Immunos- turer’s protocol. A volume of 1 ml of this reaction was tained paraffin sections were counterstained with used in all subsequent PCRs. Amplification of ECE-1 hemalun. The level of ECE-1 staining in the various cDNA was performed using the following primers: regions was assessed as low ( þ ), intermediate ( þþ) ECE-1F: 50-AGT ATG ACA AGG ACG GGA ACC-30 or high ( þþþ). and ECE-1R: 50-CTT ACC AGA CTT CGC ACT TGT G- 30 (resulting in a 459 bp product). Relative quantita- Polymorphism screening tive duplex PCRs were performed in a volume of Genomic DNA was extracted from peripheral whole 25 ml, using 18S RNA as an internal standard.29 Each blood using standard techniques. For each of the three reaction contained ECE-1-specific primers and Quan- promoter regions of the ECE1 gene, B1000 base pairs tumRNA Universal 18S primers (Ambion, resulting in (bp) of genomic DNA were amplified by polymerase a 315 bp product) in the same tube, following the chain reaction (PCR) in 30 healthy individuals, using manufacturer’s instructions. We tested different 18S AmpliTaq Gold (Perkin-Elmer), as described pre- primer-to-competimer ratios and finally used ratios of viously (ECE-1a, ECE-1c),27 or GC-RICH PCR System 2 : 8 for ECE-1 relative quantification. PCR was (Roche Diagnostics), following manufacturer’s in- performed using AmpliTaq Gold (Perkin-Elmer), with structions (ECE-1b-ECE-1d). PCR products were 9 min of denaturation at 951C, followed by 28 cycles screened for sequence variations by direct sequencing comprising 30 s at 941C, 30 s at 621C, and 30 s at 721C on a Li-Cor 4000L automated DNA sequencer, as (the optimal number of cycles had been determined as described previously.27 Wild-type and variant se- recommended). PCR products were loaded onto a 2% quences were analyzed for cis-elements using the agarose gel, stained with ethidium bromide, and computer program MatInspector (http://transfac. visualized by UV transillumination, using a CCD gbf.de/TRANSFAC).28 video camera. Raw images were directly saved as .tif files, and densitometric analysis was performed for Reporter gene assays each sample using the NIH Image software. For each All nucleotide positions (np) are given with respect to subject, relative quantitative RT-PCR were performed the ECE-1b translation start site. Genomic DNAs from and analyzed in triplicate, and the mean of the three two subjects (one homozygous for the A allele and the experiments was used in subsequent tests. other for the C allele of the ECE1B C-338A poly- morphism) were used for promoter activity assays. Genetic association study Genomic DNA regions between np À823 and À2, and French Caucasian patients affected by LOAD were between À823 and þ 258 (encompassing the entire recruited in hospitals. All were thoroughly investi- ECE-1d-specific promoter sequence) were subcloned gated. Clinical diagnosis of probable AD was made into the promoterless pSEAP2-Basic vector (Great according to the DSM III R and NINCDS-ADRDA EscAPe SEAP Reporter System 3, Clontech) upstream criteria. None of the subjects had a family history of from the SEAP (Secreted Alkaline Phosphatase) gene. AD consistent with Mendelian inheritance. Caucasian Sequencing of the vectors confirmed that the C/A controls of similar ages were recruited in France and variation was the only sequence change. The four defined as subjects without DSM III R dementia

Molecular Psychiatry ECE-1 and Alzheimer’s disease B Funalot et al 1124 criteria and with integrity of their cognitive functions. and CA4 fields (Figure 1b and c). Negative controls The study was approved by the ethical committee of showed no staining. A similar pattern of immunos- the institutions and written consent was obtained taining was observed when using the B61/104 or the from all patients or from their relatives. E15/6 monoclonal antibodies (Supplementary Figure 1) ECE1B C-338A and ECE1A G-377A genotypes were which strongly supports that this staining is specific determined by PCR-RFLP techniques. Amplification of brain ECE-1. However, we cannot completely rule primers were: 50-tag ggt tat agg aga ggg ctc agg-30 and out that both antibodies could crossreact with ECE-2 50-aag tat cag gaa ggt gcc ctc aat-30 for ECE1B C-338A, (which has 60% homology with ECE-1). The pattern 50-AGG AGG CCT GCC ACA CCC TGC-30 and 50-ATG of expression of ECE-1 across the hippocampus CAA CAC GAA GGC ACC TAC-30 for ECE1A G-377A. appears to be inversely correlated with susceptibility ECE1B amplicons were digested with the restriction to Ab deposition: it was previously shown that Ab endonuclease Tsp509I (New England Biolabs), the deposition first occurs in the subiculum and CA1 446 bp PCR product containing either one (À338C) or field, whereas the CA2–CA4 fields are much more two (À338A) restriction site(s) for Tsp509I. ECE1A resistant to Ab deposits.30 These different suscept- amplicons were digested with the restriction endo- ibilities appear to be associated with striking differ- nuclease NlaIII (New England Biolabs), the 947 bp ences in ECE-1 expression in the hippocampal fields. PCR product containing either one (À377G) or two (À377A) restriction site(s) for NlaIII. Digestion pro- Identification of a functional polymorphism in the ducts were loaded onto submarine agarose gels and ECE-1 gene visualized by ethidium bromide staining. In order to explore the hypothesis that the levels of ECE1B A-2460G genotype was determined by the ECE-1 expression could be genetically controlled and TaqMan allelic discrimination method on an ABI influence the risk of developing AD, we screened the Prism 7000 Sequence Detection System (Applied ECE1 gene for common polymorphisms. The ECE-1 Biosystems). Sequences of amplification primers protein exists in four isoforms (ECE-1a, ECE-1b, were: 50-GCC GGG TGA ATC CCT TAC AC-30 and 50- ECE-1c, and ECE-1d), which only differ by their TTT TGA AAA AGA AGT TCA GTT CAC AGA ACA- N-terminal amino-acid tails.31–33 The different iso- 30, sequences of hybridization probes were: 50-ACC forms result from the existence of four alternative CAC ATG CCC CG-30 (allele A) and 50-CCC ACG TGC promoters in the ECE1 gene, each core promoter CCC G-30 (allele G). driving the transcription of one isoform (Figure 2a). By direct gene sequencing, we identified a single- Statistics nucleotide polymorphism (C/A substitution) in the Differences in mean alkaline phosphatase activity promoter region controlling ECE-1b-ECE-1d isoform (promoter–reporter gene study) were assessed by transcription (Figure 2a). Sequence analysis with the two-tailed Student’s t-test. Differences in neocortical MatInspector program, using the TRANSFAC data- ECE-1 mRNA expression levels between genotype base of transcription factor-binding sites,28 showed groups were assessed by one-tailed Student’s t-test. In that presence of the variant (A allele) created a the genetic association study, genotypic frequencies potential binding site for E2F transcription factors. were compared by using Pearson’s w2 test. Multi- Recently, the same genetic variant was independently variate logistic regression models were used to identified by others as an E2F-binding site.34 To explore the effect of the ECE1B C-338A and ECE1A confirm the functionality of this polymorphism, the G-377A variants on AD risk. Statistical interactions influence of the C/A substitution on transcriptional between genotype and covariates (age, sex, ApoE activity was assessed using promoter–reporter gene genotype) were systematically explored. A P ¼ 0.05 experiments. We tested two promoter sequences of level of statistical significance was used in all tests. different sizes: the shorter one was limited to the 822 bp upstream from the ECE1-b first codon, whereas the larger one (1081 bp) also contained the ECE-1d Results core promoter (Figure 2a). These promoter sequences ECE-1 expression in human cerebral cortex were placed upstream from an alkaline phosphatase We first performed immunocytochemical studies on reporter gene, and the resulting constructs were various brain regions obtained from nondemented transfected in ECV304 cells, which constitutively subjects and observed a neuronal ECE-1 immunor- express ECE-1b and ECE-1d isoforms.31 As shown in eactivity in the hippocampus as well as in various Figure 2b, presence of the A allele in the ECE-1b neocortical regions (Figure 1). In the neocortex, ECE-1 promoter alone was responsible for a 30% increase in immunoreactivity was observed in frontal, parietal, reporter gene expression (Po0.001). When studying temporal and occipital regions ( þ / þþ), mainly in the larger genomic sequence containing the ECE-1d pyramidal neurons of layers III and V (Figure 1a). In core promoter, presence of the A allele resulted in a the hippocampus, the subregional pattern of expres- much higher increase (300%) in reporter gene sion of ECE-1 was similar for all subjects, showing a expression (Po0.001). Finally, we tested whether weak immunoreactivity ( þ ) in the subiculum and carrying the A allele increased ECE1 gene transcrip- CA1 field, contrasting with a marked staining tion in human prefrontal neocortex (Brodmann area ( þþþ) of the pyramidal cells of the CA2, CA3, 9). As shown in Figure 2c, we found a two-fold higher

Molecular Psychiatry ECE-1 and Alzheimer’s disease B Funalot et al 1125

Figure 1 ECE-1 immunostaining in the human cerebral cortex; bar ¼ 100 mm. (a) Expression of ECE-1 in the neocortex of nondemented subjects. ECE-1 immunoreactivity was observed in frontal, parietal, temporal, and occipital regions, mainly in pyramidal neurons of layers III and V (A9: frontal; A39: parietal; A22: temporal; A18: occipital). (b) ECE-1 expression in the different hippocampal fields of a 38-year-old subject. Neuronal expression is weak in the subiculum and CA1, and much higher in fields CA2–CA4. (c) ECE-1 expression in the different hippocampal fields of a nondemented 84-year-old subject with no extracellular amyloid deposits in the hippocampus. The same pattern of neuronal ECE-1 expression is observed. In addition, other small-sized structures showing ECE-1 immunoreactivity are visible in the vicinity of stained neurons. Immunostaining of an adjacent slide with an anti-CD68 monoclonal antibody confirmed that these additional ECE-1-positive structures are microglial cells.

level of ECE-1 mRNA expression in carriers (three AC association study comparing 401 subjects with spora- subjects) as compared to noncarriers (four CC sub- dic LOAD and 461 age- and sex-matched controls. jects; P ¼ 0.03). An obvious limitation of these last Results of the genotyping are summarized in Table 1. results is related to the small sample size, which will The frequency of AA homozygotes was reduced in the need to be increased in further studies. A concomitant LOAD group (P ¼ 0.04), whereas the frequency of A study of ECE-1 protein and activity will also be of allele carriers (dominant model) was not significantly interest. different (P ¼ 0.20). After adjustment for age, sex, and ApoE status, the protective effect of AA homozygos- ECE1 polymorphisms and protection against AD ity, as expressed by the odds ratio (OR), was 0.47 We next tested the hypothesis of a protective effect of (95% CI 0.25–0.88). No statistical interaction between the ECE1B C-338A variant against LOAD in a genetic ECE-1 and ApoE genotypes was detected, suggesting a

Molecular Psychiatry ECE-1 and Alzheimer’s disease B Funalot et al 1126

Figure 2 Identification of a functional genetic variant (ECE1B C-338A) in the ECE1 gene. (a) Organization of the ECE1 promoter regions. Black ellipses: isoform-specific core promoters; open boxes: isoform-specific first exons (the 16 following exons are common to all isoforms); dotted lines below indicate promoter sequences used in promoter–reporter gene constructs. The genetic polymorphism (open arrowhead) is located 338 bp upstream from the ECE-1b translation start site. (b) Promoter–reporter gene experiments show a functional effect of the variant in two different promoter–reporter gene constructions. Presence of the A allele of the polymorphism in the constructs resulted in increased reporter gene expression when compared with constructs containing the C allele. A 30% increase (Po0.001) was observed with the 822 bp promoter sequence, and a four-fold increase with the larger promoter sequence (Po0.001). (c) Comparison of ECE1 gene expression levels in the prefrontal neocortex of carriers (three AC heterozygotes) vs noncarriers (four CC homozygotes) of the genetic variant.

homogeneous effect of the ECE-1 gene variant in LD between ECE1B C-338A and ECE1A G-377A subjects with or without ApoE4 allele(s). To strength- (D0 ¼ 0.824) and a reduced frequency of A allele en this genetic association, we searched for other carriers in the patients’ group (P ¼ 0.03). The combi- variants of the ECE1 gene that would influence the nation of AA homozygosity at the ECE1B C-338A risk of AD. We identified an A/G polymorphism polymorphism and presence of at least one A allele at in public databases (rs213048 in dbSNP: http:// the ECE1A G-377A polymorphism was associated www.ncbi.nlm.nih.gov/entrez/query.fcgi?db ¼ Snp), lo- with the maximal risk reduction (OR ¼ 0.39, 95% CI cated 2302 bp upstream from the ECE1B C-338A 0.16–0.95 after adjustment), but it should be kept in variant (ECE1B A-2640G). Genotyping of this poly- mind that this last result relies on a small number of morphism in 185 control individuals showed that controls (n ¼ 22) and patients (n ¼ 9). ECE1B A-2640G and ECE1B C-338A are in perfect linkage disequilibrium, having the same allelic frequencies (frequency of the G allele of ECE1B Discussion A-2640G ¼ frequency of the A allele of ECE1B C-338A ¼ 0.29) and genotypic distributions. Compar- Ab is continuously synthesized in all human tissues, ison with the chimpanzee genomic sequence (http:// but in the absence of a genetic disease deposition of www.ensembl.org/Pan_troglodytes/) indicated that aggregated Ab in brain only occurs in aged subjects. In the À2640G/À338A association represents the ances- situ Ab degradation by endopeptidases might play a tral haplotype, so that genotyping the ECE1B C-338A major role in Ab clearance,9,10 but proteases respon- variant appears equivalent to determining the pre- sible for this degradation in human brain still have to sence or absence of the ancestral haplotype at this be characterized. Several candidates have been locus. By direct gene sequencing, we identified suggested, among which are angiotensin-converting another genetic variant (ECE1A G-377A), located enzyme (ACE), some matrix metallopeptidases, plas- B11 kilobases (kb) downstream from the ECE1B C- min, thrombin, and other proteases.9,10 To date, only 338A variant, in the ECE-1a-specific promoter region. neprilysin, ECE-1, ECE-2, and insulysin have been Sequence analysis with the MatInspector program shown to influence brain Ab levels when examined in indicated no change in transcription factor-binding knockout mice.11–14,35,36 The pattern of expression of sites. ECE1A G-377A genotypes in the patients and ECE-1 observed in human neocortex and hippocam- controls are showed in Table 1. We observed a strong pus (Figure 1) further suggests that this enzyme is

Molecular Psychiatry ECE-1 and Alzheimer’s disease B Funalot et al 1127 Table 1 Genotype numbers and frequencies of the ECE1B C-338A and ECE1A G-377A polymorphisms

ECE1B C-338A genotype ECE1A G-377A genotype

CC AC AA GG AG AA

Controls (n ¼ 461) 241 (52.3%) 182 (39.5%) 38 (8.2%) 366 (79.4%) 88 (19.1%) 7 (1.5%) LOAD (n ¼ 401) 227 (56.6%) 155 (38.7%) 19 (4.7%) 341 (85.0%) 55 (13.7%) 5 (1.2%)

LOAD: late-onset Alzheimer’s disease. involved in physiological degradation of Ab in Finally, our findings also offer new therapeutic human cerebral cortex throughout life. options in AD. Indeed, the results of the genetic We showed that homozygosity for a regulatory association study suggest that a limited increase in polymorphism in the ECE1 gene (ECE1B C-338A) ECE-1 activity could prevent, or delay, up to half of could protect against LOAD. A threshold effect could LOAD cases. One means of enhancing ECE-1 activity explain why AA subjects are protected against LOAD, would be an upregulation of the gene transcription, whereas AC subjects are not: the A allele being notably via the E2F pathway. Alternatively, the use of associated with increased ECE1 promoter activity, AA allosteric activators of the enzyme or even gene homozygotes are expected to have the highest levels therapy may become realistic in a next future. All of ECE-1 expression. However, we could not obtain these opportunities make ECE-1 a new potential target direct evidence of this in our expression study: we for AD therapeutic intervention. found increased ECE-1 mRNA expression in the brain of AC heterozygotes, but unfortunately we had no brain tissues from AA homozygotes for our experi- Acknowledgements ments (this genotype being the rarest in Caucasian We thank Dr Suzanne Trottier and Dr Pierre Sokoloff populations). for their help in the preparation of the manuscript. Our findings were strengthened by the study of two other variants of the ECE1 gene, showing that the association with AD risk extends over at least 13 kb of References the gene sequence. It will now be important to assess 1 Naslund J, Haroutunian V, Mohs R, Davis KL, Davies P, Greengard the effect of ECE1 genetic variation in other popula- P et al. Correlation between elevated levels of amyloid tions, in particular with different genetic back- beta-peptide in the brain and cognitive decline. JAMA 2000; 283: grounds. Hence, our findings offer the opportunity 1571–1577. to evaluate a novel genetic locus as a modifier of 2 Delacourte A, Sergeant N, Champain D, Wattez A, Maurage CA, Lebert F et al. Nonoverlapping but synergetic tau and APP LOAD risk. Multiple genes are probably involved in pathologies in sporadic Alzheimer’s disease. Neurology 2002; 59: the genetic predisposition to LOAD, and identifica- 398–407. tion of an increasing number of these genes should 3 Hardy J, Selkoe DJ. The amyloid hypothesis of Alzheimer’s markedly increase prediction of disease.37 disease: progress and problems on the road to therapeutics. Science 2002; 297: 353–356. Beyond the fact that variation in the ECE1 gene may 4 Tanzi RE, Bertram L. New frontiers in Alzheimer’s disease modulate LOAD genetic risk, the results of our genetic genetics. Neuron 2001; 32: 181–184. association study can also be viewed as the first 5 Holsinger RM, McLean CA, Beyreuther K, Masters CL, Evin G. functional evidence of the involvement of ECE-1 in Increased expression of the amyloid precursor beta-secretase in LOAD pathogenesis. The resulting ORs of 0.4–0.5 Alzheimer’s disease. Ann Neurol 2002; 51: 783–786. 6 Fukumoto H, Cheung BS, Hyman BT, Irizarry MC. Beta-secretase indicate a protection of magnitude similar to that protein and activity are increased in the neocortex in Alzheimer 38 conferred by the ApoE2 allele. Interestingly, ApoE disease. Arch Neurol 2002; 59: 1381–1389. genotype also seems to influence Ab clearance.8,9 7 Yang LB, Lindholm K, Yan R, Citron M, Xia W, Yang XL et al. Altogether, these results reinforce the hypothesis that Elevated beta-secretase expression and enzymatic activity detected b in sporadic Alzheimer disease. Nat Med 2003; 9: 3–4. variations in A clearance influence the risk of LOAD. 8 Selkoe DJ. Deciphering the genesis and fate of amyloid beta- At a subcellular level, it is noteworthy that the two protein yields novel therapies for Alzheimer disease. J Clin Invest ECE-1 isoforms for which we demonstrate a transcrip- 2002; 110: 1375–1381. tional control by the ECE1B C-338A variant (ECE-1b 9 Selkoe DJ. Clearing the brain’s amyloid cobwebs. Neuron 2001; 32: and ECE-1d, Figure 2b) are targeted to the endosomal 177–180. 39 10 Carson JA, Turner AJ. Beta-amyloid catabolism: roles for neprilysin system. BACE-1 is also targeted to the endosomal (NEP) and other metallopeptidases? J Neurochem 2002; 81: 1–8. 40 system, and the results of several groups support an 11 Iwata N, Tsubuki S, Takaki Y, Shirotani K, Lu B, Gerard NP et al. important role for this intracellular compartment in Metabolic regulation of brain Abeta by neprilysin. Science 2001; Ab production.41–43 In addition, endocytic pathway 292: 1550–1552. b 12 Eckman EA, Watson M, Marlow L, Sambamurti K, Eckman CB. abnormalities have been shown to precede A Alzheimer’s disease beta-amyloid peptide is increased in mice deposition in the brains of patients affected by deficient in endothelin-converting enzyme. J Biol Chem 2003; 278: sporadic AD.44 2081–2084.

Molecular Psychiatry ECE-1 and Alzheimer’s disease B Funalot et al 1128 13 Farris W, Mansourian S, Chang Y, Lindsley L, Eckman EA, Frosch invasive human melanoma cell subpopulations. Anal Biochem MP et al. Insulin-degrading enzyme regulates the levels of insulin, 2001; 295: 17–21. amyloid beta-protein, and the beta-amyloid precursor protein 30 Thal DR, Rub U, Schultz C, Sassin I, Ghebremedhin E, Del Tredici intracellular domain in vivo. Proc Natl Acad Sci USA 2003; 100: K et al. Sequence of Abeta-protein deposition in the human medial 4162–4167. temporal lobe. J Neuropathol Exp Neurol 2000; 59: 733–748. 14 Miller BC, Eckman EA, Sambamurti K, Dobbs N, Chow KM, 31 Valdenaire O, Lepailleur-Enouf D, Egidy G, Thouard A, Barret A, Eckman CB et al. Amyloid-beta peptide levels in brain are Vranckx R et al. A fourth isoform of endothelin-converting enzyme inversely correlated with insulysin activity levels in vivo. Proc (ECE-1) is generated from an additional promoter molecular Natl Acad Sci USA 2003; 100: 6221–6226. cloning and characterization. Eur J Biochem 1999; 264: 341–349. 15 Turner AJ, Tanzawa K. Mammalian membrane metallopeptidases: 32 Valdenaire O, Rohrbacher E, Mattei MG. Organization of the gene NEP, ECE, KELL, and PEX. FASEB J 1997; 11: 355–364. encoding the human endothelin-converting enzyme (ECE-1). J Biol 16 Akiyama H, Shii K, Yokono K, Yonezawa K, Sato S, Watanabe K Chem 1995; 270: 29794–29798. et al. Cellular localization of insulin-degrading enzyme in rat liver 33 Schweizer A, Valdenaire O, Nelbock P, Deuschle U, Dumas Milne using monoclonal antibodies specific for this enzyme. Biochem Edwards JB, Stumpf JG et al. Human endothelin-converting Biophys Res Commun 1988; 155: 914–922. enzyme (ECE-1): three isoforms with distinct subcellular localiza- 17 Akiyama H, Kondo H, Ikeda K, Kato M, McGeer PL. Immunohis- tions. Biochem J 1997; 328: 871–877. tochemical localization of neprilysin in the human cerebral cortex: 34 Funke-Kaiser H, Reichenberger F, Kopke K, Herrmann SM, Pfeifer inverse association with vulnerability to amyloid beta-protein J, Orzechowski HD et al. Differential binding of transcription (Abeta) deposition. Brain Res 2001; 902: 277–281. factor E2F-2 to the endothelin-converting enzyme-1b promoter 18 Carpentier M, Robitaille Y, DesGroseillers L, Boileau G, Marcin- affects blood pressure regulation. Hum Mol Genet 2003; 12: kiewicz M. Declining expression of neprilysin in Alzheimer 423–433. disease vasculature: possible involvement in cerebral amyloid 35 Shirotani K, Tsubuki S, Iwata N, Takaki Y, Harigaya W, Maruyama angiopathy. J Neuropathol Exp Neurol 2002; 61: 849–856. K et al. Neprilysin degrades both amyloid beta peptides 1–40 and 19 Bernstein HG, Ansorge S, Riederer P, Reiser M, Frolich L, Bogerts 1–42 most rapidly and efficiently among thiorphan- and phos- B. Insulin-degrading enzyme in the Alzheimer’s disease brain: phoramidon-sensitive endopeptidases. J Biol Chem 2001; 276: prominent localization in neurons and senile plaques. Neurosci 21895–21901. Lett 1999; 263: 161–164. 36 Eckman EA, Reed DK, Eckman CB. Degradation of the Alzheimer’s 20 Cook DG, Leverenz JB, McMillan PJ, Kulstad JJ, Ericksen S, amyloid beta peptide by endothelin-converting enzyme. J Biol Roth RA et al. Reduced hippocampal insulin-degrading enzyme in Chem 2001; 276: 24540–24548. late-onset Alzheimer’s disease is associated with the apolipopro- 37 Yang Q, Khoury MJ, Botto L, Friedman JM, Flanders WD. tein E-epsilon4 allele. Am J Pathol 2003; 162: 313–319. Improving the prediction of complex diseases by testing for 21 Sodeyama N, Mizusawa H, Yamada M, Itoh Y, Otomo E, multiple disease-susceptibility genes. Am J Hum Genet 2003; 72: Matsushita M. Lack of association of neprilysin polymorphism 636–649. with Alzheimer’s disease and Alzheimer’s disease-type neuro- 38 Farrer LA, Cupples LA, Haines JL, Hyman B, Kukull WA, Mayeux pathological changes. J Neurol Neurosurg Psychiatry 2001; 71: R et al. Effects of age, sex, and ethnicity on the association between 817–818. apolipoprotein E genotype and Alzheimer disease. A meta- 22 Oda M, Morino H, Maruyama H, Terasawa H, Izumi Y, Torii T et al. analysis. APOE and Alzheimer Disease Meta Analysis Consor- Dinucleotide repeat polymorphisms in the neprilysin gene are not tium. JAMA 1997; 278: 1349–1356. associated with sporadic Alzheimer’s disease. Neurosci Lett 2002; 39 Muller L, Barret A, Etienne E, Meidan R, Valdenaire O, Corvol P 320: 105–107. et al. Heterodimerization of endothelin-converting enzyme-1 23 Clarimon J, Munoz FJ, Boada M, Tarraga L, Sunyer J, Bertranpetit J isoforms regulates the subcellular distribution of this metallopro- et al. Possible increased risk for Alzheimer’s disease associated tease. J Biol Chem 2003; 278: 545–555. with neprilysin gene. J Neural Transm 2003; 110: 651–657. 40 Huse JT, Pijak DS, Leslie GJ, Lee VM, Doms RW. Maturation and 24 Boussaha M, Hannequin D, Verpillat P, Brice A, Frebourg T, endosomal targeting of beta-site amyloid precursor protein-cleav- Campion D. Polymorphisms of insulin-degrading enzyme gene are ing enzyme. The Alzheimer’s disease beta-secretase. J Biol Chem not associated with Alzheimer’s disease. Neurosci Lett 2002; 329: 2000; 275: 33729–33737. 121–123. 41 Perez RG, Soriano S, Hayes JD, Ostaszewski B, Xia W, Selkoe DJ 25 Prince JA, Feuk L, Gu HF, Johansson B, Gatz M, Blennow K et al. et al. Mutagenesis identifies new signals for beta-amyloid Genetic variation in a haplotype block spanning IDE influences precursor protein endocytosis, turnover, and the generation of Alzheimer disease. Hum Mutat 2003; 22: 363–371. secreted fragments, including Abeta42. J Biol Chem 1999; 274: 26 Subkowski T, Hillen H, Kroger B, Schmidt M. Monoclonal 18851–18856. antibodies against human endothelin-converting enzyme-1. 42 Yamazaki T, Chang TY, Haass C, Ihara Y. Accumulation J Immunoassay 1998; 19: 75–93. and aggregation of amyloid beta-protein in late endosomes of 27 Ito C, Morisset S, Krebs MO, Olie JP, Loo H, Poirier MF et al. Niemann-pick type C cells. J Biol Chem 2001; 276: 4454–4460. Histamine H2 receptor gene variants: lack of association with 43 Mathews PM, Jiang Y, Schmidt SD, Grbovic OM, Mercken M, schizophrenia. Mol Psychiatry 2000; 5: 159–164. Nixon RA. Calpain activity regulates the cell surface distribution 28 Quandt K, Frech K, Karas H, Wingender E, Werner T. MatInd and of amyloid precursor protein. Inhibition of clapains enhances MatInspector: new fast and versatile tools for detection of endosomal generation of beta-cleaved C-terminal APP fragments. consensus matches in nucleotide sequence data. Nucleic Acids J Biol Chem 2002; 277: 36415–36424. Res 1995; 23: 4878–4884. 44 Cataldo AM, Peterhoff CM, Troncoso JC, Gomez-Isla T, Hyman BT, 29 Goidin D, Mamessier A, Staquet MJ, Schmitt D, Berthier-Vergnes Nixon RA. Endocytic pathway abnormalities precede amyloid beta O. Ribosomal 18S RNA prevails over glyceraldehyde-3-phosphate deposition in sporadic Alzheimer’s disease and Down syndrome: dehydrogenase and beta-actin genes as internal standard for differential effects of APOE genotype and presenilin mutations. quantitative comparison of mRNA levels in invasive and non- Am J Pathol 2000; 157: 277–286.

Supplementary Information accompanies the paper on Molecular Psychiatry website ( http://www.nature.com/mp)

Molecular Psychiatry