Decreased Catalytic Activity of the Insulin-degrading Enzyme in Chromosome 10-Linked Alzheimer Disease Families*

Insulin-degrading enzyme (IDE) is a zinc metalloprotease that degrades the amyloid β-peptide, the key component of Alzheimer disease (AD)-associated senile plaques. We have previously reported evidence for genetic linkage and association of AD on chromosome 10q23–24 in the region harboring the IDE gene. Here we have presented the first functional assessment of IDE in AD families showing the strongest evidence of the genetic linkage. We have examined the catalytic activity and expression of IDE in lymphoblast samples from 12 affected and unaffected members of three chromosome 10-linked AD pedigrees in the National Institute of Mental Health AD Genetics Initiative family sample. We have shown that the catalytic activity of cytosolic IDE to degrade insulin is reduced in affected versus unaffected subjects of these families. Further, we have shown the decrease in activity is not due to reduced IDE expression, suggesting the possible defects in IDE function in these AD families. In attempts to find potential mutations in the IDE gene in these families, we have found no coding region substitutions or alterations in splicing of the canonical exons and exon 15b of IDE. We have also found that total IDE mRNA levels are not significantly different in sporadic AD versus age-matched control brains. Collectively, our data suggest that the genetic linkage of AD in this set of chromosome 10-linked AD families may be the result of systemic defects in IDE activity in the absence of altered IDE expression, further supporting a role for IDE in AD pathogenesis.

Insulin-degrading enzyme (IDE) is a zinc metalloprotease that degrades the amyloid ␤-peptide, the key component of Alzheimer disease (AD)-associated senile plaques. We have previously reported evidence for genetic linkage and association of AD on chromosome 10q23-24 in the region harboring the IDE gene. Here we have presented the first functional assessment of IDE in AD families showing the strongest evidence of the genetic linkage. We have examined the catalytic activity and expression of IDE in lymphoblast samples from 12 affected and unaffected members of three chromosome 10-linked AD pedigrees in the National Institute of Mental Health AD Genetics Initiative family sample. We have shown that the catalytic activity of cytosolic IDE to degrade insulin is reduced in affected versus unaffected subjects of these families. Further, we have shown the decrease in activity is not due to reduced IDE expression, suggesting the possible defects in IDE function in these AD families. In attempts to find potential mutations in the IDE gene in these families, we have found no coding region substitutions or alterations in splicing of the canonical exons and exon 15b of IDE. We have also found that total IDE mRNA levels are not significantly different in sporadic AD versus age-matched control brains. Collectively, our data suggest that the genetic linkage of AD in this set of chromosome 10-linked AD families may be the result of systemic defects in IDE activity in the absence of altered IDE expression, further supporting a role for IDE in AD pathogenesis.
Amyloid ␤-protein (A␤) 2 is the primary component of senile plaques, a pathological hallmark in the brains of patients with Alzheimer disease (AD). Elevated levels of cerebral A␤ have also been observed in AD patients (1,2), implicating excessive accumulation of A␤ as a key pathogenic event in AD. Unlike early onset autosomal dominant AD, the vast majority of AD cases do not show any clear evidence of Mendelian transmission and predominantly present with late onset AD (LOAD) (onset age Ͼ65). However, there is evidence that genetic factors play a significant role in modifying the disease risk/age of onset in the majority of LOAD cases (3,4). To date, only the ⑀4 allele of the apolipoprotein E gene (APOE) has been firmly established as a LOAD genetic risk factor and has been proposed to be involved in A␤ clearance (5). Cerebral A␤ accumulation has been proposed to greatly influence the age of onset of LOAD and is determined by the amount of A␤ generated versus the amount that is degraded and exported from the brain over one's lifetime (6,7).
Several proteases have been identified to degrade A␤, including neprilysin, plasmin, endothelin-converting enzyme-1 as well as insulin-degrading enzyme (IDE) (EC 3.4.24.56) (1). IDE, also called insulysin, is a zinc metalloprotease that cleaves small polypeptides, many of which share amyloid fibril-forming ability, including insulin, atrial naturetic peptide, amylin, calcitonin, and A␤ (8,9). IDE is a major protease to degrade soluble, monomeric A␤ (10,11) and is localized in the cytoplasm as well as on the cell surface and within mitochondria (11,12). Recent studies in animal models have demonstrated that knock out of IDE leads to elevated cerebral A␤ levels along with phenotypic characteristics of type 2 diabetes mellitus (13,14). Conversely, overexpression of IDE attenuates A␤ accumulation in transgenic AD mouse models (15). Finally, diabetes-inducing mutations in IDE in rat are associated with impaired neuronal A␤ catabolism, supporting co-morbidity of AD, and type 2 diabetes mellitus (16). Several studies have suggested genetic linkage and allelic association between LOAD and the IDE/KIF11 region on chromosome 10q in independent samples (17)(18)(19)(20). Our recent report on meta-analyses across all published studies (12 published reports for IDE to date) reveals a significant association of IDE with AD (21), confirming our original finding on the association and linkage of IDE to AD (17). However, neither pathogenic nor protective IDE gene mutations/variants have yet been validated in AD patients.
In an effort to localize novel AD genes, our group previously performed a full-genome linkage screen of the National Insti-tute of Mental Health AD Genetics Initiative family sample (22), which supported our earlier observation of genetic linkage in the chromosomal region near the IDE gene (17). In the current study, we chose three families showing the strongest evidence for linkage to IDE and systematically assessed the activity and expression of IDE in immortalized lymphoblast cell lines from these families. Potential alteration of IDE expression was also investigated in sporadic AD brains to confirm the result from the family-based lymphoblast samples.

EXPERIMENTAL PROCEDURES
Lymphoblastoid Cell Culture-Lymphoblastoid cell lines from affected and unaffected subjects of the three chromosome 10-linked families (total 12 samples) were obtained from The Rutgers University Cell and DNA Repository in New Jersey, and a control lymphoblast (GM7044) unrelated to the AD families was obtained from the Coriell Cell Repository. Among the affecteds, three subjects were male and two were female. The average age at death was 79.7 (Ϯ3.9 S.E.) years and the average age of onset was 72.1 (Ϯ3.7 S.E.) years. Of the unaffecteds, one subject was male and six were female, with the average age at death of 67.2 (Ϯ6.5 S.E.) years. Lymphoblasts were grown in RPMI 1640 medium (BioWhittaker) supplemented with 15% fetal bovine serum (Sigma). After inoculating the same number of living cells, exponentially growing lymphoblasts were collected for analysis.
Human Brain Samples-Temporal neocortical tissue from 24 AD brains and 11 control brains without any signs of a neurological disorder were included in the study. All AD subjects had been evaluated at the Memory Disorders Unit at Massachusetts General Hospital and met both the clinical (NINCDS-ADRDA) (23) and the neuropathological (CERAD, NIA/Reagan) (24,25) diagnostic criteria for AD. Nine of the non-neurological control brains had been autopsied at Massachusetts General Hospital, whereas two control brains were from the Harvard Brain Bank and University of Maryland, respectively. Of the AD cases, 58% were male and 42% female. The average age at death was 80.8 years (Ϯ1.6 S.E.) with a 15.6-h (Ϯ2.8 S.E.) post-mortal interval. Thirteen of the AD cases were APOE-⑀4-positive (⑀4 ϩ ). Among the control subjects, 50% were male and 50% female, with an average age at death of 82.4 years (Ϯ1.6 S.E.) years and with a 25.7-h (Ϯ5.6 S.E.) post-mortal interval. Two of the control subjects were ⑀4 ϩ .
In Vitro Insulin Degradation Assay-After collecting lymphoblast cells, membrane and cytosolic fractions were prepared by centrifugation of homogenized cell extracts at 100,000 ϫ g for 1 h as described by Huang et al. (26). Insulin-degrading activity was assayed as described by Song et al. (27). In brief, lymphoblast fractions (0.5-2.0 ml) were incubated with 50 nM tyrosine A 14 -labeled 125 I-insulin (Amersham Biosciences) in 100 mM potassium phosphate buffer (pH 7.3) at 37°C for 15 min (cytosol) or 120 min (membrane). Intact insulin was then removed by precipitation with 7.5% trichloroacetic acid. Following centrifugation, the degradation products of iodinated insulin, which remain in the supernatant, were counted on a ␥ counter. Blanks were run with the extract omitted and used for subtracting backgrounds. The degradation rate in pmol/ min/mg was calculated from the average of the counts/min val-ues in the trichloroacetic acid supernatant, the specific radioactivity of the iodinated insulin, and the amount of protein in the reaction. The assay was performed in duplicate in all families, whereas four independent measurements were performed for families II and III. The averages were calculated within and across the families.
In Vitro A␤ Degradation Assay-Lymphoblast samples were disrupted by incubation in hypotonic buffer (50 mM Tris-HCl, pH 7.4) and extrusion through a 22-gauge hypodermic needle three times. The post-nuclear homogenate was centrifuged at 20,000 ϫ g for 20 min to separate cytosolic (supernatant) and membrane (pellet) fractions, and protein concentration was determined by bicinchoninic acid assay (Pierce). Fluorometric quantification of A␤ degradation was performed using FA␤B (fluorescein-A␤-(1-40)-Lys(LC-biotin)) as described by Leissring et al. (28). Briefly, 5 mg of protein was incubated with 0.5 mM FA␤B for 90 min at 20°C in a degradation buffer (50 mM HEPES, 100 mM NaCl, 10 mM MgCl 2 , 0.05% bovine serum albumin, pH 7.4). The reactions were quenched by adding avidin to a final concentration of 0.5 mM, and fluorescence polarization (485 excitation, 535 emission) was determined on a PerkinElmer Life Sciences Victor2 multilabel plate reader. Reactions were performed in quadruplicate in the presence and absence of 10 M insulin, a relatively specific inhibitor of IDE activity, and the percentage of hydrolysis was calculated from standards containing no protease or excess recombinant IDE.
RNA Extraction-Total RNA was extracted from lymphoblast samples using the RNeasy kit (Qiagen) as described in the manufacturer's instructions. The concentration of the total RNA was determined using an NP-1000 spectrophotometer (Nanodrop). For brain samples, a 30 -50-mg piece of gray matter from the superior temporal sulcus was dissected under stringent RNase-free conditions from each brain in accordance with previously published methods (29). From each tissue, RNA was extracted with TRIzol reagent according to the manufacturer's instructions (Invitrogen). All samples were controlled for integrity of the 18 and 28 S ribosomal RNAs by microcapillary electrophoresis (not shown) (RNA 6000 Nano Assay, Agilent Technologies), and samples showing degradation were excluded from the study.
Reverse Transcription PCR (RT-PCR)-For the lymphoblast samples, first-strand cDNAs were synthesized from 5 g of the total RNA using 200 units of Superscript III reverse transcriptase (Invitrogen) and random hexamers as described in the manufacturer's instructions. To investigate alternatively spliced transcript variants, RT-PCR was carried out using the cDNAs as templates with TaqDNA polymerase (Qiagen) or Turbo Pfu DNA polymerase (Stratagene) and corresponding primers (Table 1). For the brain tissue samples, reverse transcription was carried out on 2 g of all total RNA samples to generate an equal number of cDNA copies using random hexamers and 200 units of Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions.
Quantitative RT-PCR-All quantitative PCR analyses were performed with iCycler TM (Bio-Rad). To determine IDE mRNA levels in the lymphoblast and brain samples, a 6-carboxyfluorescein-labeled TaqMan probe and a set of PCR primers (forward and reverse primers-TaqMan, Table. 1) were syn-thesized (Applied Biosystems), and quantitative PCR was performed on equal amounts of cDNA from each sample using TaqMan universal PCR master mix (Applied Biosystems). The IDE mRNA levels were measured in triplicates with four independent experiments and determined by the standard curve method. Relative IDE mRNA levels were calculated by normalization with glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or neuron-specific enolase (NSE, ENO2) mRNA levels, which were generated by quantitative PCR on the same cDNA samples using the SyBR Green I detection system (Applied Biosystems). Forward and reverse primers for GAPDH and NSE were designed with Primer Express TM software (PE Applied Biosystems) and synthesized (Qiagen). Each PCR product was sequenced to confirm its identity. For levels of exons 15a-and 15b-specific IDE mRNAs in the lymphoblast samples, a common forward primer residing in exon 14 (IDE 14F in Table 1) and either of the reverse primers specific for exon 15a or 15b were used for SyBR Green quantitative RT-PCR. The quantitative RT-PCR was done in triplicates with three independent experiments. Each exon-specific amplified DNA fragment was confirmed by sequencing. The pCR2.1 vectors (Invitrogen) harboring the same PCR fragments with exons 15a or 15b were utilized to generate standard curves. All of the PCR primer pairs for quantitative RT-PCR were located in different exons to avoid possible amplification of genomic DNA. Sequences of all of the primers and the probe are listed in Table 1.
Western Blot Analysis-Total cell extracts were obtained by resuspending lymphoblast cells in radioimmune precipitation assay buffer (10 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% cholic acid, 0.1% SDS, and 5 mM EDTA). The total lysates were separated on 4 -12% BisTris gel (Invitrogen), transferred to immunoblot polyvinylidene difluoride membrane (Bio-Rad), and hybridized with IDE polyclonal antibody (IDE-1, (11)) and actin monoclonal antibody (pan Ab-5, Neo-Markers). Quantification of the chemiluminescence signal was accomplished using the VersaDoc imaging system and the Quantity One quantification program (Bio-Rad). Relative IDE protein levels were calculated by normalization of IDE levels with actin signals, and averages from four independent experiments were taken.

IDE Activity in the Chromosome 10-Linked AD Families-
We first assessed IDE activity in the lymphoblast cell lines derived from 12 subjects of three AD families exhibiting particularly strong evidence of genetic linkage to chromosome 10 in the IDE gene region (17,22). Because insulin possesses a high affinity for IDE and is a major substrate of this peptidase, we employed a sensitive and reliable insulin degradation assay to assess IDE catalytic activity in membrane and cytosolic fractions prepared from lymphoblast samples (Fig. 1A). Peptidolytic activity of cytosolic IDE was significantly decreased in the affected versus unaffected subjects (Ͼ50%; p ϭ 0.001) after combining the members of all three families. Additionally, in one family (I), cytosolic activity of IDE was significantly lower in three affected versus three unaffected subjects (p ϭ 0.008). IDE activity was also decreased in the membrane fractions of the affected (versus unaffected) individuals; however, these differences did not reach statistical significance.
To confirm our finding of decreased cytosolic IDE activity particularly in families I and III, affecteds of which exhibited the most robust decreases in IDE activity in the insulin degradation assay, peptidolytic activity of IDE was next assayed using A␤ 1-40 as a substrate (Fig. 1B). IDE-mediated degradation of A␤ was decreased in affected versus unaffected subjects in both families. However, this difference did not reach statistical significance at least partly because of the less sensitive nature of this assay (versus the insulin-degrading assay) and lower affinity of IDE for A␤. Meanwhile, we could not obtain reliable data for the A␤-degrading ability of membrane IDE, possibly because of the limiting amount of IDE in the membrane fractions and decreased affinity of IDE for A␤.
IDE Expression in the Chromosome 10-Linked AD Families-We next assessed IDE expression by performing quantitative RT-PCR on cDNAs generated from lymphoblast cell lines from the same three AD families. The mRNA levels of IDE, normalized to GAPDH, were highly variable across samples and revealed no significant or consistent differences among affected versus unaffected subjects across the families (Fig. 2, A and B,  top). The families also exhibited differing trends for IDE mRNA levels in affected versus unaffected individuals; e.g. in family I, affected subjects had lower IDE mRNA levels than the unaffecteds (p ϭ 0.022), whereas an opposite trend was observed in family III (p ϭ 0.020).
Next, IDE protein levels were determined by quantitative Western blot analysis of total protein extract (Fig. 2, A and B,  bottom). Relatively high IDE expression in affected versus unaffected subjects of family III was confirmed at the protein level (p ϭ 0.018). However, the affected subjects in family I, who had relatively low IDE message levels, revealed no significant reductions in IDE protein levels. Overall, protein levels of IDE across the families did not exhibit significant or consistent differences according to disease status.
Levels of Alternative IDE Transcripts in the Chromosome 10-Linked AD Families-After establishing that no IDE coding mutations existed in the canonical exons of the IDE gene in these AD families, we next explored the possibility that altered mRNA splicing of IDE may account for the observed reduction in IDE activity. First, Northern blot analysis was performed with total RNA extracted from the lymphoblast samples. Two different IDE transcripts were detected (data not shown) that had similar sizes to those previously reported in rat, 3.6 and 5.9 kb (30). Northern blotting using PCR fragments covering three different IDE coding regions (Fig.  3A, (a), (b), and (d)) as probes, detected the two IDE mRNAs of indistinguishable sizes with no other additional bands (data not shown). Recently, Farris et al. (31) detected up to four different Northern blot bands of IDE in different tissues, two of which corresponded in size with the transcripts found in our lymphoblast samples. Based on the previously reported observations, we suspect that the two major IDE messages observed in these lymphoblasts are due to alternative polyadenylation sites. In any event, the two Northern blot bands were coordinately expressed, suggesting no significant differential regulation of the two transcripts among lymphoblasts from these families.
To assess alternative splicing of IDE, we performed RT-PCR on total RNA obtained from each lymphoblast sample using oligonucleotide primer pairs spanning the entire IDE coding region (Fig. 3A). As shown in Fig. 3B, no changes in sizes of PCR products were observed, suggesting no evidence for alternative splicing in the major transcript in these samples. Subsequently, fulllength IDE cDNA was produced by RT-PCR using high fidelity DNA polymerase (Pfu, Stratagene), and possible sequence changes in IDE mRNA were investigated more closely by sequencing. Neither exon swapping between similar-sized exons nor nucleotide sequence changes was observed in the major IDE transcript (data not shown).
Levels of the 15b-IDE Isoform Do Not Differ in the Chromosome 10-Linked Families-In addition to the 25 canonical exons, as many as nine additional putative exons in the IDE gene have been reported in GenBank TM . Exon 15b (15b), which is an additional exon in intron 15 and was originally identified in the human testis cDNA library, has been functionally characterized (31). The IDE isoform produced by the transcript in which 15b replaces canonical exon 15 (15a), was revealed to have less catalytic efficiency for insulin and A␤ versus the wild-type isoform. To test for levels of the 15b-containing IDE transcript in the lymphoblast samples, RT-PCR fragments containing the exon  To discriminate between the transcripts harboring 15a versus 15b, the same amounts of the amplified DNA fragments were digested with exon-specific restriction enzymes (Fig. 4A). As can be seen in Fig. 4B, small but significant amounts of the amplified PCR fragments were cleaved by DraI, the 15b-specific restriction enzyme, and left uncut by the 15aspecific digestion enzyme NheI in all of the lymphoblast samples. Complete digestion of the PCR fragments by the restriction enzymes was confirmed by digestion of same amount of corresponding PCR fragment amplified with IDE cDNA plasmid containing 15a as template (Fig. 4B, exon 15a). This result suggested that the IDE transcript containing 15b in place of 15a was expressed in the lymphoblast cell lines in the affected and the unaffected subjects from these families. Next, the levels of the 15a-and 15b-IDEs were measured by quantitative RT-PCR using exon-specific 3Ј primers for 15a and 15b and a common 5Ј primer residing in exon 14 (Fig. 4A). The ratio of 15b-IDE versus 15a-IDE was not significantly altered relative to IDE activities in the affected individuals (Fig. 4C).
IDE mRNA Levels in Sporadic AD Brain Samples-Given our observation of variable IDE expression in the lymphoblast samples from the chromosome 10-linked AD families (Fig. 2), we next set out to examine IDE expression in post-mortem brain samples from AD versus control subjects. As brain samples were not available from members of the chromosome 10-linked AD families, we assessed IDE mRNA levels in temporal cortical samples from 24 sporadic AD patients and 11 agematched controls. Levels of absolute IDE mRNA were highly variable among individual samples within each group, and no significant differences were detected between AD patients and control subjects (data not shown). We also measured the mRNA levels of two controls, GAPDH (a general housekeeping protein) and NSE (a neuronal-specific marker). NSE mRNA levels normalized to GAPDH mRNA levels (NSE/GAPDH) were reduced by 58.7% in the AD brains relative to the control brains (p ϭ 0.031), consistent with a substantial neuronal cell loss (Fig. 5A). Although mRNA levels of IDE normalized to GAPDH were not significantly higher in AD versus control   (solid arrows in A). Same amount of RT-PCR product from each sample was digested by either exon15a-or exon15b-specific restriction enzymes NheI and DraI, respectively. Digested RT-PCR products were separated on 2% agarose gel. The expected bands are indicated by arrowheads. An amplification product using a vector of IDE cDNA harboring exon 15a (exon15a) as a template served as a positive control for NheI (N) digestion and a negative control for DraI (D). C, exon-specific quantitative RT-PCR. 15a-and 15b-IDE mRNA levels were determined in lymphoblast samples of the AD family members using the same 5Ј primer (left solid arrow in A) and exon-specific 3Ј primers (dotted arrows in A). 15b/15a ratios of individual samples are presented as mean Ϯ S.D. of three independent determinations (left). Averages of the 15b/15a ratios within and across families were calculated (right).
brain samples (p ϭ 0.109) (Fig. 5B, left), IDE message levels normalized to NSE mRNA (IDE/NSE) were ϳ3-fold higher in AD versus control brain samples (p ϭ 0.010) (Fig. 5B, right). In addition, we found that the presence of the APOE-⑀4 allele in the AD subjects was not accompanied by statistically significant changes in IDE mRNA levels in this cohort of AD brains (data not shown).

DISCUSSION
This study represents the first functional assessment of the IDE gene, a significant candidate gene for LOAD (21), in AD families exhibiting genetic linkage to the IDE gene region of chromosome 10. We assessed IDE activity in lymphoblast cell lines from three chromosome 10-linked AD families and consistently observed reduced catalytic activity of IDE across affected individuals of all three AD families. Because IDE activity was decreased in both the cytosolic and the membrane fractions of the lymphoblast lines, the differences observed in affected versus unaffected subjects are unlikely to be the result of altered cellular compartmentalization of IDE. Reduced IDE activity in these AD lymphoblast lines was not due to decreased expression of IDE, because levels of IDE mRNA and protein did not differ between affected and unaffected members of these families. Taken together, these findings suggest that genetic linkage of these families to chromosome 10 may be driven by gene defects in IDE leading to dysfunction. Although the pathogenic defects in IDE remain unidentified, our results suggest that the deficit in activity is not due to altered gene expression.
Interestingly, in one family (III), the cytosolic activities and protein levels of IDE were inversely correlated, i.e. the lymphoblast samples from affected subjects with reduced IDE activity exhibited increased IDE expression (Figs. 1A and Fig. 2A). Thus, it is possible that the deficit in IDE activity in affected individuals of this family may lead to compensatory increases in IDE expression.
Although little is known about the regulation of IDE activity, chemical inhibitors known to abolish IDE catalytic activity include the Zn 2ϩ chelator 1,10-phenanthroline, the insulin-binding inhibitor bacitracin, the thiol-blocking agents, p-hydroxy-mercuribenzoate, and N-ethylmaleimide (32). In addition, two heat-stable IDE-interacting proteins have been shown to inhibit its insulin-degrading activity in mouse leukemic splenocytes (33,34). One of them was identified as ubiquitin, which, upon binding with IDE in a reversible and ATP-independent manner, exhibits a strong inhibitory effect at physiological concentrations. Ubiquitin and IDE are present in both neurofibrillary tangles and senile plaques (35)(36)(37)(38), raising the possibility that ubiquitin may regulate IDE activity in these pathological lesions. Allosteric properties of IDE activity have also been elucidated in which binding of certain peptide substrates to one subunit of IDE induces a conformational change that shifts the equilibrium to the more active dimer while also activating the adjacent subunit (39). The active site motif (HXXEH) and a cationic regulatory site both affect the allosteric properties. The cationic regulatory site binds anions, including nucleotide triphosphates (such as ATP), and alters the oligomerization state of the enzyme toward a monomer, which consequently changes the substrate specificity (27). Interestingly, some peptides, e.g. dynorphin B-9, activate IDE selectively toward A␤ cleavage but not toward insulin cleavage (39,40). In addition, IDE activity can be inhibited by phosphorylation (41), free fatty acids, and acyl-coenzyme A thioesters (42). Collectively, these previous findings indicate that IDE activity may be regulated by a complex combination of diverse regulatory factors. Thus, our observation of altered IDE activity in the lymphoblast lines from chromosome 10-linked AD families might also stem from, as of yet, unidentified gene defects impacting IDE regulatory mechanisms. Although no coding sequence changes were found in the major IDE transcript of the affected members of these three families, it is still possible that variants may exist in other, more elusive, alternatively transcribed coding regions of IDE. Along these lines, we investigated the potential contribution of 15b, the only alternatively transcribed exon to date shown to alter IDE activity. We found FIGURE 5. IDE mRNA levels in sporadic AD brain samples. The mRNA levels of IDE, GAPDH, and NSE were measured by quantitative RT-PCR using equal amounts of cDNAs extracted from temporal cortices of agematched control and AD brains. Relative mRNA levels were calculated by normalization with GAPDH and NSE mRNA levels. A, scatter plot of the relative NSE mRNA levels normalized to GAPDH (NSE/GAPDH). An average of each group is indicated. B, scatter plots of the relative IDE mRNA levels normalized to GAPDH (IDE/GAPDH, left) and NSE (IDE/NSE, right). p values were calculated by two-tailed Student's t test, and asterisks indicate statistically significant values.
neither coding DNA changes in the alternatively spliced 15b nor any evidence for changes in the levels of IDE transcript containing 15b in these AD families (Fig. 4C). Further analyses of the IDE gene will be necessary to determine the molecular mechanism underlying reduced IDE activity in these three chromosome 10-linked families.
Previous studies of IDE expression in AD brain have thus far produced inconsistent results. Although increased intracellular and neuronal immunostaining have been reported in AD brains (35), decreased levels of a carboxyl-terminal fragment of IDE were detected in cytosolic fractions of AD brains by Western blot analysis (43). IDE protein levels measured by enzymelinked immunosorbent assay were also reported to be increased in the cortical microvessels of AD patients by another group (44). Meanwhile, Cook et al. (45) report a statistically significant reduction in hippocampal IDE expression (protein and mRNA) in AD brains of APOE-⑀4 allele carriers as compared with brains of non-APOE-⑀4 AD patients and both APOE-⑀4positive and -negative non-demented controls. Using a similar number of brain samples as Cook et al. (45) but from temporal cortex, we did not detect a statistically significant decrease in the IDE expression for our cohort of AD brains (versus controls) when absolute IDE mRNA levels and the relative levels normalized to GAPDH were compared. Expression of GAPDH, which has previously been shown to be similar in AD and control brain (46), was also not significantly different in our AD versus control brain samples (data not shown), validating it as an appropriate control. Although our result would appear to be discrepant with the observations of Cook et al. (45), it should be pointed out that they had measured IDE levels by in situ hybridization and Western blot analysis, whereas we employed quantitative RT-PCR. Moreover, they had analyzed IDE levels in the hippocampus and observed regional differences (i.e. no decrease in CA-1), and we measured IDE levels in temporal cortical tissue. Other studies have also reported temporal and spatial differences in IDE steady-state expression levels, showing that, in the hippocampus, IDE protein levels diminish as a function of age, whereas cerebral cortical IDE expression is elevated (47). More in-depth investigation of local IDE expression and activity will be necessary for a better understanding of potentially pathogenic changes in IDE in different brain regions.
Additionally, we found that IDE expression was significantly increased in the AD brain samples when normalized to the neuronal specific message NSE. We also observed a significant decrease in NSE/GAPDH ratio, consistent with neuronal cell loss in AD. Because IDE is known to be expressed predominantly in neurons (11,35,45), one possible interpretation of these data is that IDE expression may be increased in certain neuronal populations in AD in temporal cortex, whereas the total number of neurons is decreased as a potential response to elevated A␤ levels. However, we cannot rule out the alternative possibility of increased IDE mRNA levels owing to activated astrocytes or microglia in AD. Cell type-specific IDE expression should be monitored to address this issue.
In summary, we have found that IDE activity is reduced in lymphoblast samples from affected versus unaffected family members of three chromosome10/IDE-linked AD families.
However, reduced IDE activity was not due to decreased IDE expression, suggesting the possibility of systemic functional defects in IDE in these families. Although sequencing of the major and the 15b-IDE transcripts did not reveal any coding region substitutions, further sequence analysis of alternative IDE transcripts may be warranted. Collectively, these data provide the first evidence of a systemic functional defect in IDE in AD patients from chromosome 10-linked families. Although these results provide further support for a role of IDE in AD pathogenesis, further investigation of this peptidase will be necessary to establish the mechanism by which IDE activity is altered in IDE-linked AD families.