Multiparametric assays generate biological activity profiles that provide valuable insight into complex disease models. The use of multiple assay measurements in RNA interference (RNAi) high-throughput screening (HTS) provides biological signatures produced by knocking down individual genes. This strategy has been applied to a genome-wide high-throughput small interfering RNA (siRNA) screen measuring proteolysis of β-amyloid precursor protein (APP) into amyloid β peptides, a critical step in the pathogenesis of Alzheimer's disease. The assay measures amounts of secreted Aβ40, Aβ42, sAPPα, and sAPPβ from HEK 293 cells stably expressing an optimized APP construct following siRNA transfection. The effect of each siRNA on the four different APP products was simultaneously measured in order to identify human genes that regulate the amyloidogenic processing of APP. Genes with BACE-like and γ-secretase-like activity profiles were identified for further biological characterization.

Introduction

RNA interference (RNAi) has been used to test the effect of knocking down individual genes in cell division (1), apoptosis (2), endocytosis (3), and Mycobacterium infection (4). Adaptation of RNAi technology to the high-throughput screening (HTS) environment has given scientists the ability to characterize gene function via gene-specific knockdown on a genome-wide scale in order to identify new therapeutic targets. Fully automated robotic protocols have been developed to create reproducible small interfering RNA (siRNA) transfection results in 384-well plate format. This allows for a rapid, genome-wide test of genes with functional relevance to virtually any cellular process that can be modeled in a transfectable cell line.

Here, we report on a siRNA screen of 15,200 genes for effects on the processing of amyloid precursor protein (APP) into amyloid peptides to identify potential therapeutic targets for the treatment of Alzheimer's disease. Alzheimer's disease affects as many as 33% of those over age 85 (5) and is characterized by two neuropathologies: (i) senile plaques, made up primarily of the amyloid β (Aβ) peptide (6,7), and (ii) neurofibrillary tangles, composed of hyperphosphorylated tau protein (8). Aβ peptides are generated from APP by cleavage at the β-secretase and γ-secretase sites. Cleavage at the β-site is achieved by the aspartyl protease BACE1 (9), while γ-site cleavage is carried out by the γ-secretase complex composed of presenilins-1 or -2 (PSEN1 or PSEN2), nicastrin (NCSTN), Aph-1a or -1b (APH1A or APH1B), and Pen-2 (PSENEN) (10). Cleavage at the γ-secretase site primarily yields a 40-amino acid form of Aβ (Aβ40), along with a 42-amino acid form (Aβ42), which is the predominant form found in senile plaques. Genetic evidence suggests that Aβ42 plays a causal role in Alzheimer's disease (11). Genes that affect production of Aβ42 are therefore of interest as drug targets for the treatment of Alzheimer's disease.

Screens consisting of a single readout may not be reflective of in vivo models for a particular disease, because they provide an oversimplified, one-dimensional view of the biology studied. It has been shown that signature-based screens simultaneously measuring multiple end points may provide results that correlate better with in vivo outcomes (12). Therefore, a series of intricate HTS transfection and detection protocols were developed to produce a set of multiparametric data that identified genes involved in the processing of APP or secretion and clearance of its products. We found known APP processing genes to have a characteristic activity profile when examined on a multi-analyte panel (Aβ40, Aβ42, sAPPα, and sAPPβ) and used this functional signature to rapidly separate false positives from hits truly affecting APP processing. Biological process analysis of hits obtained from the screen showed enrichment for genes associated with notch receptor processing, amyloid precursor protein catabolism, and β-amyloid metabolism. Many of the resultant genes could potentially be pursued as new therapeutic targets to prevent onset of Alzheimer's disease (13).

Materials and Methods

Cell Line

Human embryonic kidney (HEK) 293T cells (HEK NFEV WTα #7) were stably transfected with APP containing an enhanced β-secretase cleavage site with a mutated BACE1 cleavage site (NFEV) and a wild-type α-secretase site (14). Cells were cultured in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS), 1% penicillin-streptomycin (all from Invitrogen, Carlsbad, CA, USA), and 2 µg/mL puromycin (Sigma, St. Louis, MO, USA).

Control siRNA

Biologically relevant control sequences (sense listed) included on each plate were BACE1 (pool of three siRNAs) 5′-CCUUCGUUUGCCCAAGAAA-3′, 5′-GGCAGUCUCUGGUAUACAC-3′, and 5′-CUGUCGGAGGGAGCA-UGAU-3′; PSEN1 (pool of three siRNAs) 5′-CUGUUUCGUAGCCAUAUUA-3′, 5′-CACCCUGAGCCAUUAUCUA-3′, and 5′-CCUGGUGGUUCUGUAU-AAA-3′; PSENEN (single siRNA) 5′-CCACGUUCUCUGCUGACAU-3′; and NCSTN (single siRNA) 5′-GUGCCAGGAUCCAAGUAAA-3′. As a negative control, we used nonspecific control duplex #1 (Dharmacon, Lafayette, CO, USA).

siRNA Library

A total of 15,200 siRNA pools (three single siRNA sequences per pool) representing a corresponding number of genes were tested. This genome-scale screen of the Merck human siRNA library was carried out on 21-mer oligonucleotides designed by Rosetta Inpharmatics (Seattle, WA, USA) and synthesized by Proligo (Boulder, CO, USA). Rosetta's siRNA design algorithm specifically targets all GenBank classified isoforms of each gene target. Given an estimate of 1.5 transcripts per gene (15) in the human genome, the siRNA library targets an estimated 22,800 unique mRNA transcripts. Transfection protocols can be found in (Table 1).