Background
Genetic, epidemiologic, and biochemical evidence suggests that apolipoprotein E, low-density lipoprotein receptors, and lipid metabolism play important roles in sporadic Alzheimer disease (AD).
Objective
To identify novel candidate genes associated with sporadic AD.
Design
We performed an unbiased microarray screen for genes differentially expressed in lymphoblasts of patients with sporadic AD and prioritized 1 gene product for further characterization in AD brain.
Setting
Emory University, Atlanta, Ga.
Subjects
Cell lines were used from 14 patients with AD and 9 normal human control subjects.
Results
Six genes were differentially expressed in lymphoblasts of 2 independent groups of patients with probable AD and autopsy-proven AD. We hypothesized that 1 of the genes, termed low-density lipoprotein receptor relative with 11 binding repeats (LR11) (reduced 1.8- and 2.5-fold in AD lymphoblasts vs controls), might be associated with sporadic AD on the basis of its function as neuronal apolipoprotein E receptor. We found dramatic and consistent loss of immunocytochemical staining for LR11 in histologically normal-appearing neurons in AD brains. This reduction of LR11 protein was confirmed by quantitative Western blotting (P = .01).
Conclusions
There is loss of the microarray-derived candidate, LR11, in neurons of AD brains. This study shows that microarray analysis of widely available lymphoblasts derived from patients with AD holds promise as a primary screen for candidate genes associated with AD.
Alzheimer disease (AD) is the most important cause of aging-related dementia, and its prevalence has risen dramatically with increases in the oldest segments of the population. While the mechanisms underlying the disease remain poorly understood, deposition of senile plaques composed of fibrillar aggregates of amyloid-β (Aβ) peptide is believed to play an important pathogenic role.1,2 Converging lines of evidence also implicate apolipoprotein E (ApoE), low-density lipoprotein (LDL) receptors, and lipid metabolism in AD pathogenesis. Epidemiologic data link serum cholesterol level, dietary fatty acids, and exposure to certain lipid-lowering drugs to AD risk.3-5 Apolipoprotein E is the most abundant apolipoprotein expressed in brain, and a common polymorphism in the APOE gene represents the major genetic risk factor for sporadic, late-onset cases of AD.6 Other genetic association studies implicate members of the LDL receptor family, which bind ApoE, in AD pathogenesis.7,8 The mechanism underlying these associations is unclear. Apolipoprotein E binds Aβ peptide9 and may modulate Aβ fibrillization and amyloid deposition, or promote internalization of Aβ through the LDL receptor–related protein (see reviews10,11).
Previous studies have shown abnormal biochemical responses in extraneural tissues in sporadic AD, including platelets12 and lymphoblasts.13-15 In addition, most genes implicated in AD are ubiquitously expressed,12,16,17 and skin fibroblasts from individuals carrying a familial AD mutation secrete excessive amounts of Aβ peptide.18 Therefore, physiologically relevant alterations in AD might be reflected in shared gene expression changes in neural and extraneural tis sues. In this study, we used complementary DNA microarrays to screen for genes differentially expressed in lymphoblasts from patients with probable or definite AD. This novel strategy identified changes in 6 transcripts, including the lipoprotein receptor LR11 (LDL receptor relative with 11 binding repeats).19,20 On the basis of its function as a neuronal ApoE receptor and its expression in the brain, we hypothesized that LR11 might play a role in sporadic AD. To test this hypothesis, we examined LR11 protein expression in AD brains.
Lymphoblast lines were obtained from healthy elderly control subjects and patients with AD who were all well characterized via annual assessments in the Alzheimer’s Disease Center at Emory University, Atlanta, Ga. Informed consent was obtained in accordance with the regulations of the institutional review board at Emory University. The diagnosis of probable AD was made according to criteria of the National Institute of Neurological Disorders and Stroke and consensus of 2 experienced clinicians. (Multiple clinicians, including A.I.L. and J.J.L., participated in establishing the consensus diagnosis.) The pathologic diagnosis of definite AD was made by a neuropathologist according to criteria of the Consortium to Establish a Registry for Alzheimer’s Disease.21 (Multiple neuropathologists, including M.G., were involved.) Cell lines were used from a total of 14 patients with AD and 9 normal human control subjects (Table 1).
Patient lymphocytes were immortalized by the Neitzel method22 in the Emory General Clinical Research Center. In brief, the lymphocytes were removed in the buffy coat layer after gradient separation (Histopaque-1077; Sigma-Aldrich Corp, St Louis, Mo). The lymphocytes were then incubated with Epstein-Barr virus (B95-8, American Type Culture Collection, Manassas, Va) in transforming medium consisting of RPMI-1640 (Gibco-BRL, Gaithersburg, Md), 20% heat-inactivated fetal bovine serum (Gibco-BRL), 2-μg/mL cyclosporine (Sandoz, Inc, Princeton, NJ), 2mM l-glutamine, and penicillin-streptomycin. After transformation, the lymphoblasts were maintained in RPMI-1640 medium supplemented with 15% heat-inactivated fetal bovine serum and 110-μg/mL sodium pyruvate, and stocks of cells were stored in liquid nitrogen. The cell lines were thawed and cultured in RPMI-1640 medium supplemented with 15% heat-inactivated fetal bovine serum to a confluency of 80%.
Complementary dna microarray screen
Lymphoblast lines from patients with AD and controls were matched for age, sex, and race. As expected, a higher APOE ε4 allele frequency was present in the patients with AD than controls, but this was matched as closely as possible. Two experiments were performed with the use of lymphoblast messenger RNA (mRNA) from patients with clinically diagnosed probable AD, and a group with autopsy-confirmed definite AD. For experiment 1, cell lines from 7 patients with probable AD plus 1 patient with definite AD were compared with those from 8 controls; in experiment 2, 6 definite AD samples were compared with 6 controls. Total RNA was purified from each cell line by means of a kit (RNeasy Mini kit; Qiagen, Valencia, Calif) and pooled for controls and patients with AD in each experiment. The mRNA was isolated according to a batch protocol (Oligotex; Qiagen) and quantified by spectrophotometry. Complementary DNA probe synthesis, hybridization with human UniGEM V complementary DNA microarrays, and signal analysis were conducted by Incyte Genomics (St Louis, Mo) as described.23 Transcript abundance for 7270 genes was assessed in experiment 1 and for 9374 genes in experiment 2. Because false-positive results are particularly high for low-intensity genes, a selective intensity filter (absolute fluorescence intensities ≥800) was applied to exclude genes with low hybridization signal intensities. Then, genes with an AD-to-control fluorescence intensity ratio (fold change) of 1.8 or greater were considered significant according to standard recommendations24 based on reproducibility data generated by Incyte Genomics, indicating that the level of detectable differential expression is 1.8-fold for UniGEM arrays.
In experiment 1, quantification of mRNA was additionally confirmed by Northern blot hybridization with glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Northern hybridization analysis was performed with the use of formaldehyde to denature 500 ng of polyA RNA, followed by electrophoresis and transfer to a hybridization transfer membrane (GeneScreen; NEN Research Products, Boston, Mass). An in vitro transcription kit (MEGAscript; Ambion, Inc, Austin, Tex) was used for [α phosphorous P 32]2′-deoxyuridine 5′-triphosphate ([α 32P]dUTP) labeling of RNA probes. The intensity of bands was quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, Calif).
Tissue samples from frontal cortex of 5 human controls and 6 patients with AD were thawed and homogenized in Tris-EDTA, pH 7.4, plus protease inhibitors (Complete Protease Inhibitors; Hoffman-La Roche, Inc, Nutley, NJ). Protein concentrations were measured with a protein assay kit (BCA; Pierce, Rockford, Ill). Samples were separated across a 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis gel and transferred overnight to polyvinylidene fluoride membranes (Immobilon-P; Millipore, Billerica, Mass). Blots were blocked in 5% nonfat milk–Tris-buffered saline at room temperature for 30 minutes, then probed overnight at 4°C with a polyclonal antibody to LR11 carboxyl terminus, whose specificity has been demonstrated previously,25 or a monoclonal antibody to the VPS10 domain.26 Blots were rinsed and incubated for 1 hour at room temperature with secondary antibodies conjugated to horseradish peroxidase (Pierce) and visualized by chemiluminescence (Renaissance; Perkin-Elmer, Boston, Mass). Images were captured and band intensities measured by means of a Kodak Image Station 440cf (Eastman Kodak, New Haven, Conn) and software (Perkin-Elmer). Band intensities were normalized for protein loading by reprobing blots with anti–14-3-3 monoclonal antibody (C.J.H., A.I.L., unpublished data, 2000).
Blocks of frontal cortex from 13 patients with AD and 7 controls were fixed for 24 to 48 hours in 4% paraformaldehyde, then embedded in paraffin or cryoprotected in 30% sucrose and frozen. Paraffin-embedded blocks were cut into 8-μm sections, deparaffinized, and pretreated with pepsin (Biomeda, Foster City, Calif). Frozen blocks were cut into 50-μm sections. Sections were then treated with hydrogen peroxide, washed in Tris buffer, blocked with normal serum, and incubated with anti-LR11 antibodies overnight at 4°C. On day 2, sections were incubated with biotinylated secondary antibody followed by avidin-biotin-peroxidase complex (Vector Elite ABC kit; Vector Laboratories, Burlingame, Calif). Color development was carried out with 3,3′-diaminobenzidine. Control sections incubated without primary antibody showed negligible staining.
Primary microarray screen in lymphoblasts of patients with probable ad
To screen for genes differentially expressed in lymphoblasts of patients with AD, we performed 2 independent microarray experiments. In the first experimental set, immortalized lymphoblasts from 7 patients with a clinical diagnosis of probable AD and 1 patient with autopsy-confirmed definite AD were compared with lymphoblasts from a group of 8 age-, sex-, and race-matched control lymphoblast lines (Table 1). Of 7270 genes analyzed, expression of 15 mRNAs was decreased by 1.8-fold or greater, expression of 3 was increased by 1.8-fold or greater, and 7252 mRNAs were below significance threshold (a complete list of differentially expressed genes is available from the authors). To technically confirm the initial microarray results with an independent method, mRNA levels of selected genes were assessed by Northern hybridization in the probable AD and control groups. Hybridization signals for Bcl2-related protein A1 and interferon-γ receptor mRNA were decreased, while hybridization signal for IgG3 mRNA was increased (Figure 1). These results were consistent with the fold changes observed for each of these genes by microarray hybridization (−2.1, −2.4, and +3.3, respectively). Hybridization signal for the housekeeping gene GAPDH was unchanged.
Secondary microarray screen in lymphoblasts of patients with definite ad
We sought to validate our initial results and narrow the list of candidate genes by analyzing samples from an independent group of patients with autopsy-confirmed diagnosis of AD. In this second experiment, lymphoblast mRNA from 6 patients with definite AD was compared with lymphoblast mRNA from 6 age-, sex-, and race-matched normal controls (Table 1). In the definite AD group, of 9374 genes analyzed, mRNA expression of 108 was decreased by 1.8-fold or greater, expression of 7 was increased by 1.8-fold or greater, and 9259 mRNAs were below significance threshold (a complete list of differentially expressed genes identified in the microarray screens is available from the authors). Down-regulation of 5 genes and up-regulation of 1 gene was confirmed in both microarray experiments (Table 2). One of the consistently down-regulated genes, LR11, is an ApoE recep tor that is predominantly expressed in brain and possesses structural and functional homologies to LDL receptor–related protein, a receptor etiologically linked to AD. On the basis of these considerations, we selected LR11 for further examination in control and AD brains.
LR11 PROTEIN EXPRESSION IN HUMAN BRAIN
To establish the potential biological relevance of changes in LR11 gene expression for AD, we examined LR11 in control and AD brains at the level of protein expression. Immunohistochemistry of 13 AD brains and 7 controls showed a remarkable reduction in LR11 expression in AD (Figure 2A and B). In control brains, pyramidal neurons in the frontal cortex showed strongly labeled small cytoplasmic puncta throughout the cell body and the proximal dendrites (Figure 2C). In striking contrast, there was dramatic loss of LR11 staining in pyramidal neurons in AD frontal cortex (Figure 2D). The difference between control and AD brains was remarkably consistent, and marked loss of LR11 staining in pyramidal neurons was found in each of the AD cases examined. In addition to neurons, punctate LR11 staining was also found in glial cells. However, unlike pyramidal neurons, glial staining in frontal cortex was preserved in AD brains (Figure 2E and F). The loss of LR11 immunoreactivity was not due to neuronal loss in the AD brains. Paraffin-embedded sections counterstained with hematoxylin demonstrated strong LR11 staining in dentate granule neurons in controls (Figure 2G), but hematoxylin-stained hippocampal neurons in AD brains were nearly devoid of LR11 immunoreactivity (Figure 2H).
Western blots of total frontal cortex homogenates of 5 human controls and 6 patients with AD confirmed decreased LR11 protein levels in AD brain. Immunoblots from control and AD brains showed a specific LR11 band migrating at approximately 250 kDa, as previously reported.19,20 Quantification of relative band intensity showed a 25% reduction in LR11 in AD frontal cortex vs controls (P = .01) (Figure 3). Levels of an unrelated, abundantly expressed antigen, 14-3-3, were similar in AD and control brains. Compared with the drastic decrease in neuronal LR11 staining by immunocytochemistry, Western blotting indicated relatively modest reduction in LR11 band intensity in total cortical homogenates. This difference may reflect the contribution of glial LR11, which is retained in AD cortex (Figure 2E and F).
Our findings suggest a novel link between LR11 and AD; to our knowledge, this report is the first to identify a candidate disease-associated gene in an unbiased microarray screen of blood from patients with AD. Peripheral cells express genes associated with AD12,16,17 and model some processes involved in pathological changes of AD brains.27,28 The structure and function of LR11 as a mosaic ApoE receptor lends biological plausibility to the microarray results, and examination of LR11 in brain strongly supports the hypothesis that it plays a role in AD. In agreement with previous studies,29 we detected LR11 in widespread populations of neurons in neocortex, limbic cortex, and cerebellum. In AD brains, LR11 immunoreactivity was lost with remarkable consistency. Moreover, LR11 staining was decreased specifically in neurons, but staining was preserved in glia. The loss of immunoreactivity was not simply due to cell loss, as hematoxylin counterstaining showed otherwise healthy-appearing neurons (Figure 2). Most of the AD brains in this study were from patients with late-stage disease. Additional studies of patients with mild AD and mild cognitive impairment will be helpful in determining whether LR11 plays a role in early stages of disease development.
The unique multidomain structure of LR11 suggests potential roles as a cell-surface lipoprotein receptor and as an intracellular sorting receptor. There is a cluster of extracellular ligand binding repeats and a cytoplasmic internalization sequence that are present in all endocytosis competent lipoprotein receptors.30 In addition, LR11 contains a VPS10 homology domain near the amino terminus and a Golgi-localized, gamma-ear-homology domain, adenosine diphosphate–ribosylation (ARF)–binding protein (GGA) binding domain in the cytoplasmic tail.19,20,31 The VPS10 domains are involved in trafficking from the Golgi to the vacuole in yeast,32 and GGAs have been shown to mediate trafficking between Golgi and the endosomal-lysosomal system.33 Given its structural features, LR11 seems to be ideally positioned to interact with AD-associated proteins, and additional studies suggest that LR11 expression may influence levels of Aβ (K.O., unpublished data, 2003).
This exploratory study suggests that gene expression analysis of widely available lymphoblasts derived from patients with AD holds promise as a primary screen for candidate genes associated with AD. Our current studies, using this approach, identified the brain ApoE receptor, LR11, as an intriguing candidate molecule for sporadic AD. Further studies using larger sample sizes and refined microarray and bioinformatics procedures coupled with mechanistic validation of candidates are warranted.
Correspondence: Allan I. Levey, MD, PhD, and James J. Lah, MD, PhD, Center for Neurodegenerative Disease, 615 Michael St, Suite 505, Atlanta, GA 30322 (alevey@emory.edu; jlah@emory.edu).
Accepted for Publication: November 19, 2003.
Author Contributions:Study concept and design: Scherzer, Levey, Lah, and Offe. Acquisition of data: Scherzer, Gearing, Rees, Bujo, Lah, Offe, Fang, and Heilman. Analysis and interpretation of data: Scherzer, Gearing, Schaller, Levey, Lah, and Offe. Drafting of the manuscript: Scherzer, Bujo, Levey, Lah, and Fang. Critical revision of the manuscript for important intellectual content: Scherzer, Gearing, Rees, Schaller, Levey, Lah, Offe, and Heilman. Obtained funding: Levey. Administrative, technical, and material support: Gearing, Rees, Bujo, Levey, Lah, Fang, and Heilman. Study supervision: Levey and Lah.
Funding/Support: This study was supported in part by National Institutes of Health grants P30AG10130 from the National Institute on Aging and M01RR000039 from the National Center for Research Resources, and by the Rotary CART (Coins for Alzheimer's Research Trust) fund (to Dr Levey).
This article was corrected on 2/6/2007, prior to publication of the correction in print.
1.Selkoe
DJ Toward a comprehensive theory for Alzheimer’s disease: hypothesis: Alzheimer’s disease is caused by the cerebral accumulation and cytotoxicity of amyloid β-protein.
Ann N Y Acad Sci 2000;92417- 25
PubMedGoogle ScholarCrossref 2.Selkoe
DJ Alzheimer’s disease: genes, proteins, and therapy.
Physiol Rev 2001;81741- 766
PubMedGoogle Scholar 4.Wolozin
BKellman
WRuosseau
PCelesia
GGSiegel
G Decreased prevalence of Alzheimer disease associated with 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors.
Arch Neurol 2000;571439- 1443
PubMedGoogle ScholarCrossref 5.Rockwood
KKirkland
SHogan
DB
et al. Use of lipid-lowering agents, indication bias, and the risk of dementia in community-dwelling elderly people.
Arch Neurol 2002;59223- 227
PubMedGoogle ScholarCrossref 6.Corder
EHSaunders
AMStrittmatter
WJ
et al. Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families.
Science 1993;261921- 923
PubMedGoogle ScholarCrossref 7.Kang
DESaitoh
TChen
X
et al. Genetic association of the low-density lipoprotein receptor–related protein gene (LRP), an apolipoprotein E receptor, with late-onset Alzheimer’s disease.
Neurology 1997;4956- 61
PubMedGoogle ScholarCrossref 8.Okuizumi
KOnodera
ONamba
Y
et al. Genetic association of the very low density lipoprotein (VLDL) receptor gene with sporadic Alzheimer’s disease.
Nat Genet 1995;11207- 209
PubMedGoogle ScholarCrossref 9.Strittmatter
WJWeisgraber
KHHuang
DY
et al. Binding of human apolipoprotein E to synthetic amyloid β peptide: isoform-specific effects and implications for late-onset Alzheimer disease.
Proc Natl Acad Sci U S A 1993;908098- 8102
PubMedGoogle ScholarCrossref 10.Brendza
RPBales
KRPaul
SMHoltzman
DM Role of apoE/Aβ interactions in Alzheimer’s disease: insights from transgenic mouse models.
Mol Psychiatry 2002;7132- 135
PubMedGoogle ScholarCrossref 12.Di Luca
MColciaghi
FPastorino
LBorroni
BPadovani
ACattabeni
F Platelets as a peripheral district where to study pathogenetic mechanisms of Alzheimer disease: the case of amyloid precursor protein.
Eur J Pharmacol 2000;405277- 283
PubMedGoogle ScholarCrossref 13.Ibarreta
DParrilla
RAyuso
MS Altered Ca2+ homeostasis in lymphoblasts from patients with late-onset Alzheimer disease.
Alzheimer Dis Assoc Disord 1997;11220- 227
PubMedGoogle Scholar 14.Ibarreta
DUrcelay
EParrilla
RAyuso
MS Distinct pH homeostatic features in lymphoblasts from Alzheimer’s disease patients.
Ann Neurol 1998;44216- 222
PubMedGoogle ScholarCrossref 15.Cecchi
CLatorraca
SSorbi
S
et al. Glutathione level is altered in lymphoblasts from patients with familial Alzheimer’s disease.
Neurosci Lett 1999;275152- 154
PubMedGoogle ScholarCrossref 16.Li
QXFuller
SJBeyreuther
KMasters
CL The amyloid precursor protein of Alzheimer disease in human brain and blood.
J Leukoc Biol 1999;66567- 574
PubMedGoogle Scholar 17.Schlossmacher
MGOstaszewski
BLHecker
LI
et al. Detection of distinct isoform patterns of the β-amyloid precursor protein in human platelets and lymphocytes.
Neurobiol Aging 1992;13421- 434
PubMedGoogle ScholarCrossref 18.Citron
MVigo-Pelfrey
CTeplow
DB
et al. Excessive production of amyloid β-protein by peripheral cells of symptomatic and presymptomatic patients carrying the Swedish familial Alzheimer disease mutation.
Proc Natl Acad Sci U S A 1994;9111993- 11997
PubMedGoogle ScholarCrossref 19.Yamazaki
HBujo
HKusunoki
J
et al. Elements of neural adhesion molecules and a yeast vacuolar protein sorting receptor are present in a novel mammalian low density lipoprotein receptor family member.
J Biol Chem 1996;27124761- 24768
PubMedGoogle ScholarCrossref 20.Jacobsen
LMadsen
PMoestrup
SK
et al. Molecular characterization of a novel human hybrid-type receptor that binds the α2-macroglobulin receptor–associated protein.
J Biol Chem 1996;27131379- 31383
PubMedGoogle ScholarCrossref 21.Mirra
SSHeyman
AMcKeel
D
et al. The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD), part II: standardization of the neuropathologic assessment of Alzheimer’s disease.
Neurology 1991;41479- 486
PubMedGoogle ScholarCrossref 23.Bandman
OColeman
RTLoring
JFSeilhamer
JJCocks
BG Complexity of inflammatory responses in endothelial cells and vascular smooth muscle cells determined by microarray analysis.
Ann N Y Acad Sci 2002;97577- 90
PubMedGoogle ScholarCrossref 24.Young
ANAmin
MBMoreno
CS
et al. Expression profiling of renal epithelial neoplasms: a method for tumor classification and discovery of diagnostic molecular markers.
Am J Pathol 2001;1581639- 1651
PubMedGoogle ScholarCrossref 25.Hampe
WRiedel
IBLintzel
JBader
COFranke
ISchaller
HC Ectodomain shedding, translocation and synthesis of SorLA are stimulated by its ligand head activator.
J Cell Sci 2000;1134475- 4485
PubMedGoogle Scholar 26.Hirayama
SBujo
HYamazaki
H
et al. Differential expression of LR11 during proliferation and differentiation of cultured neuroblastoma cells.
Biochem Biophys Res Commun 2000;275365- 373
PubMedGoogle ScholarCrossref 27.Govoni
SGasparini
LRacchi
MTrabucchi
M Peripheral cells as an investigational tool for Alzheimer’s disease.
Life Sci 1996;59461- 468
PubMedGoogle ScholarCrossref 28.Eckert
ASchindowski
KLeutner
S
et al. Alzheimer’s disease–like alterations in peripheral cells from presenilin-1 transgenic mice.
Neurobiol Dis 2001;8331- 342
PubMedGoogle ScholarCrossref 29.Motoi
YAizawa
THaga
SNakamura
SNamba
YIkeda
K Neuronal localization of a novel mosaic apolipoprotein E receptor, LR11, in rat and human brain.
Brain Res 1999;833209- 215
PubMedGoogle ScholarCrossref 30.Herz
JBeffert
U Apolipoprotein E receptors: linking brain development and Alzheimer’s disease.
Nat Rev Neurosci 2000;151- 58
PubMedGoogle ScholarCrossref 31.Jacobsen
LMadsen
PNielsen
MS
et al. The sorLA cytoplasmic domain interacts with GGA1 and -2 and defines minimum requirements for GGA binding.
FEBS Lett 2002;511155- 158
PubMedGoogle ScholarCrossref 32.Marcusson
EGHorazdovsky
BFCereghino
JLGharakhanian
EEmr
SD The sorting receptor for yeast vacuolar carboxypeptidase Y is encoded by the VPS10 gene.
Cell 1994;77579- 586
PubMedGoogle ScholarCrossref 33.Puertollano
RAguilar
RCGorshkova
ICrouch
RJBonifacino
JS Sorting of mannose 6-phosphate receptors mediated by the GGAs.
Science 2001;2921712- 1716
PubMedGoogle ScholarCrossref