Article pubs.acs.org/jpr
A Shotgun Proteomics Approach Reveals a New Toxic Role for Alzheimer’s Disease Aβ Peptide: Spliceosome Impairment Domenico Nuzzo,† Luigi Inguglia,‡,§ Jessica Walters,† Pasquale Picone,† and Marta Di Carlo*,† †
Istituto di Biomedicina ed Immunologia Molecolare “A. Monroy” (IBIM), Via Ugo La Malfa 153, 90146 Palermo, Italy Istituto di Biofisica (IBF), Via Ugo La Malfa 153, 90146 Palermo, Italy § Euro-Mediterranean Institute of Science and Technology, 90146 Palermo, Italy ‡
ABSTRACT: Proteomic changes have been described in many neurodegenerative diseases, including Alzheimer’s disease (AD). However, the early events in the onset of the pathology are yet to be fully elucidated. A cell model system in which LAN5 neuroblastoma cells were incubated for a short time with a recombinant form of Aβ42 was utilized. Proteins extracted from these cells were subjected to shotgun proteomics analysis by LTQ-Orbitrap-MS followed by label-free quantitation. By bioinformatics tools we found that the most significant of those found to be up-regulated were related to cytoskeletal dynamics (Rho related) and membrane-related processes. The most significant of the down-regulated proteins were hnRNP-related. In particular, hnRNPs involved in ribosomal biogenesis and in splicing were down-regulated. The latter of these processes stood out as it was highlighted ubiquitously and with the highest significance in the results of every analysis. Furthermore, our findings revealed down-regulation at every stage of the splicing process through down-regulation of every subunit of the spliceosome. Dysregulation of the spliceosome was also confirmed using a Western blot. In conclusion, these data suggest dysregulation of the proteins and processes identified as early events in pathogenesis of AD following Aβ accumulation. KEYWORDS: shotgun proteomics, Alzheimer’s disease, early events in AD, spliceosome
■
In AD, studies of amyloid and tau3,4 have provided extensive knowledge concerning pathogenic mechanisms such as oxidative stress, mitochondrial dysfunction, inflammation, and apoptosis;5,6 however, the underlying etiology of the disease remains incompletely understood. It is believed that the pathophysiological process of AD begins with a long “preclinical” phase many years before the diagnosis of AD dementia.8 Therapeutic intervention at this stage of the disease could be vastly beneficial; however, an understanding of the link between the initiation of the pathological cascade and the disease’s biomarkers is necessary to actualize this possibility. One key obstacle in this research is differentiating between patients with age-related cognitive impairment and those with preclinical AD.8 In one study on patients with mild cognitive impairment (MCI), this was only found to convert into dementia in 23% of cases.9 AD progression can be determined according to the presence and levels of Tau neurofibrillary tangles and Aβ amyloid deposits in well-defined brain areas. On this basis, AD is classified in six Braak stages (I−VI).1−10 Braak stages I and II are characterized by the presence of only a few neurofibrillary changes in the transentorhinal region of the brain, and no sign of cognitive decline is associated. However,
INTRODUCTION
Alzheimer’s disease (AD) is the most common form of dementia, and its prevalence increases exponentially with age. AD patients gradually lose cognitive function, control over their sense of orientation, their emotions, and other aspects of behavior. Thirty-five million people are now considered to be affected by AD, and this number is expected to double in the next few decades. Consequently, AD is now considered a very serious public health problem due to the direct and indirect costs of caring for the sufferers of this disease. Pathohistologically, AD is characterized by the presence of extracellular deposits of beta amyloid protein (Aβ) in diffuse and neuritic plaques, neurofibrillary changes leading to intracellular deposition of hyperphosphorylated tau protein in neurofibrillary tangles, dystrophic neurites, and neuropil threads.1 Amyloid plaques are produced by pathological deposition of amyloid-β peptide (Aβ). This small protein is derived from the sequential proteolytic cleavage of amyloid precursor protein (APP) by β-site APP cleaving enzyme 1 (BACE1) and γsecretase, a multisubunit protease complex comprised of proteins such as presenilin 1 and 2 (PS1, PS2).2,3 To date, therapies targeting these classical markers have been unsatisfactory, and only palliative treatment is currently available for patients. Advances in understanding of AD pathogenesis are needed to develop effective therapies. © 2017 American Chemical Society
Received: October 24, 2016 Published: February 3, 2017 1526
DOI: 10.1021/acs.jproteome.6b00925 J. Proteome Res. 2017, 16, 1526−1541
Article
Journal of Proteome Research
blue stained.15 Here, a mixture of Aβ42 small aggregates of about 25 kDa as the larger species is referred as oligomers.
in these initial stages, the cellular dysfunctions that progressively lead to dementia begin. Accumulation of Aβ is believed to be the earliest pathohistological feature of AD.4−11 As such, a greater understanding of the immediate and direct effect of the accumulation of this protein on neuronal cells may shed light on the mechanisms involved in the early, preclinical stage of the disease. Furthermore, studies in vitro have demonstrated that by using a recombinant Aβ42 it is possible to reproduce the molecular and cellular dysfunctions leading to altered signaling cascade, oxidative stress, inflammation, mitochondrial dysfunction, and apoptosis, occurring during AD onset and progression.12−16 Co-administration of antioxidants and antiinflammatory molecules inhibits all cellular dysfunction induced by the recombinant Aβ42 and prevents neurodegeneration.17,18 Similarly, administration of a recombinant Aβ42 on an in vivo simple model system reproduces cellular dysfunctions occurring during degenerative processes which, as in the previous model, can be inhibited by using antioxidants and antiinflammatory molecules.12−19 Thus, establishing the right experimental conditions, it is possible to mimic the dysfunction occurring during the different AD stages. Development of proteomics as an area of research over the past two decades has allowed comparison of proteomic profiles in disease and normal conditions, and consequently, potential identification of novel disease biomarkers and therapeutic targets. A proteomic approach has been successfully applied to identify potential biomarkers to discriminate AD, PD patients, and control subjects both in serum and cerebrospinal fluid (CSF), for example.20 The aim of this study was to identify potential biomarkers involved in early AD. For this purpose, we employed a shotgun proteomic approach using a LTQ-Orbitrap mass spectrometer (LC−MS/MS). This approach allowed us to compare the proteome of LAN5 neuroblastoma cells, slightly exposed to a recombinant Aβ42, with control cells. Bioinformatic analysis of the mass spectrometry data highlighted dysregulation of various processes and pathways in the treated cells. Gene ontology (GO) analysis suggested that processes related to RNA metabolism, chromosomal structure, and cytoskeletal structure, Rho-associated and membrane-associated processes, in particular, were modulated by Aβ. Pathway enrichment analysis revealed that of the pathways these processes were involved in, that of the spliceosome was the most significantly altered. RNA splicing is an essential cellular process for converting precursor messenger ribonucleic acid (pre-mRNA) into mature mRNA for use in protein translation. RNA processing allows multiple protein isoforms to be produced from the same transcript, thus increasing genetic heterogeneity. Changes in alternative splicing have been associated with neurodegeneration, including in AD.21 Our findings suggest that spliceosomal dysregulation may play a key role in the toxic effect of Aβ in the early pathogenesis of AD.
■
Cell Cultures and Treatments
The neuroblastoma LAN5 cell line was used as a cellular model. This cell line exhibits neuronal characteristics, including expression of neurofilaments and display of short neurites. Cells were cultured with RPMI 1640 medium (CELBIO) supplemented with 10% fetal bovine serum (FBS) (GIBCO), 2 mM L-glutamine (SIGMA), and 1% antibiotics (50 mg/mL−1 penicillin and 50 mg/mL−1 streptomycin) (SIGMA). Cells were maintained in a humidified 5% CO2 atmosphere at 37 ± 0.1 °C. To determine the Aβ42 concentration to be utilized in proteomic experiment, LAN5 (5 × 106/ml) cells were incubated with different concentrations (20, 40, 60 μM) of Aβ42 oligomers for 1 h. After these treatments LAN5 cells were submitted both to MTS assay and morphological analysis by microscopic inspection (Zeiss Axio Scope with a camera Axiocam). On the basis of these results the concentration of 40 μM was chosen for the preparative experiments. For Western blot experiments LAN5 cells were treated with Aβ42 at 20, 40, and 60 μM for 1 h or with 20 and 40 μM for 15, 60, and 120 min, or with H2O2 at 4 mM or doxorubicina 1 μM for 15, 60, and 120 min. Doxorubicin and H2O2 were utilized at concentrations according to Albert et al.22 and Picone et al.23 Determination of Cell Viability
Cell viability was measured by MTS assay (Promega Italia, S.r.l., Milan, Italy). MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium] was utilized according to the manufacturer’s instructions. After cell treatments, 20 μL of the MTS solution was added to each well, and the incubation was continued for 4 h at 37 °C, 5% CO2. The absorbance was read at 490 nm on the Microplate reader WallacVictor 2 1420 Multilabel Counter (PerkinElmer, Inc., Monza, Italy). Results were expressed as the percentage MTS reduction to the control cells. Total Protein Extraction and Western Blotting
Total proteins were prepared by dissolution in solubilizing buffer (50 mM Tris−HCl pH 7.4, 150 mM NaCl, 0.5% Triton X-100, 2 mM PMSF, 1 mM DTT, 0.1% SDS with protease inhibitor, Amersham, and phosphatase inhibitor cocktail II, Sigma), LAN5 cells were untreated or treated as described above. Protein samples (50 μg) were resolved by 10% SDSPAGE gel and transferred onto nitrocellulose filters for Western blotting using anti-GSK3α/β (1:1000), antiphospho-GSK3β (1:1000), anti-SmB/B′/N (1:500), anti-PRP6 (1:500), antiMAPK1 (1:1000) purchased from Santa Cruz, anti-β-actin (1:10000) purchased from Sigma. Secondary antibodies conjugated to horseradish peroxidase (1:2000) purchased from Cell Signaling were detected using the NOVEX ECL HRP chemiluminescence kit (cat. no. WP20005, Invitrogen) according to the manufacturer’s instructions. Band intensities were analyzed with a gel documentation system (BioRad), and expression was normalized with β-actin expression. The protein levels were densitometrically quantified and expressed as percentage relative to controls. The Western blot was replicated three times using proteins extracted from the two different experiments described above.
EXPERIMENTAL PROCEDURES
Preparation and Characterization of Aβ42 Oligomers
The recombinant Aβ42 (Aβ42) was produced, purified, and prepared in oligomeric form according to refs 12−14. Briefly, Aβ42, in monomeric form, was dissolved in 0.01 M Tris−HCl at pH 7.2. After 60 min of incubation at 37 °C, aliquots were characterized by dynamic light scattering (DLS) and successively loaded on a nondenaturing PAGE and Coomassie
Experimental Design
LAN5 cells were cultured in four separate flasks. Two flasks were utilized for Aβ42 (40 μM) treatment, and two flasks were untreated. Each flask was utilized for two technical replicates. 1527
DOI: 10.1021/acs.jproteome.6b00925 J. Proteome Res. 2017, 16, 1526−1541
Article
Journal of Proteome Research
ments. Full scans were performed in the Orbitrap with resolution of 30000 at 400 m/z, and the automatic gain control was set to 1000000 ions and the lock mass option was enabled on a protonated polydimethylcyclosiloxane background ion ((Si(CH3)2O)6; m/z = 445.120025) as internal recalibration for accurate mass measurements.25 Peptide ions were selected as the 10 most intense peaks (Top 10) of the previous scan. The signal threshold for triggering an MS/MS event was set to 500 counts. Higher energy collisional dissociation (HCD), performed at the far side of the C-trap, was chosen as the fragmentation method by applying a 40% value for normalized collision energy, an isolation width of m/z 3.0, a q value of 0.25, and an activation time of 0.1 ms. Nitrogen was used as the collision gas.
Western blotting of a further set of experiments has validated the expression of some proteins. The data set containing identification and quantification data has been deposited in PRIDE/ProteomeXchange24 with the data set identifier PXD002842 and Project DOI:10.6019/ PXD002842. Preparation of Protein Extracts for Mass Spectrometry
Protein preparations were incubated with 1% SDS in 100 mM Tris−HCl (pH 8.8), 100 mM DTT, for 5 min at 95 °C and 500 rpm using a Thermomixer comfort (Eppendorf, Hamburg, Germany). Extracts were centrifuged, and the proteincontaining supernatants were subjected to 1-DE separation through 4−20% TGX gels (Bio-Rad) to check the quality of the protein extracts. Samples were then quantified using the BCA quantification kit (Pierce, Rockford, IL, now part of Thermo Scientific).
Data Analysis
Proteome Discoverer platform (version 1.3; Thermo Scientific, Bremen, Germany), interfaced with an in-house Mascot server (version 2.3, Matrix Science, London, UK), was used for data parsing and protein identification, according to the following criteria: Homo sapiens, Database UniProtKB/Swiss-Prot (release 2012_05), the number of entries was 20272, enzyme trypsin, maximum missed cleavage sites 2, taxonomy Homo sapiens, precursor mass tolerance 10 ppm, fragment mass tolerance 0.02 Da, cysteine carbamidomethylation as static modification, N-terminal glutamine conversion to pyroglutammic acid and methionine oxidation as dynamic modifications. The Percolator algorithm was used for peptide validation (peptide confidence: q value 0.5) Following Stimulation with Aβ42 That Are Present in the KEGG Alzheimer’s Pathwaya gene name
uniprot accession
APH1A
Q96BI3
ITPR1
Q14643
ATP2A2
P16615
ERK1/ ERK2
P27361
NDUFB7
P17568
a
function (STRING)
p value
fold change
essential subunit of the γ-secretase complex, an endoprotease complex that catalyzes the intramembrane cleavage of integral proteins such as Notch receptors and APP. It probably represents a stabilizing cofactor for the presenilin homodimer that promotes the formation of a stable complex (265 aa) intracellular channel that mediates calcium release from the endoplasmic reticulum following stimulation by inositol 1,4,5trisphosphate (2710 aa) this magnesium-dependent enzyme catalyzes the hydrolysis of ATP coupled with the translocation of calcium from the cytosol to the sarcoplasmic reticulum lumen. Isoform SERCA2A is involved in the regulation of the contraction/relaxation cycle (1042 aa) (MAPK1 only) involved in both the initiation and regulation of meiosis, mitosis, and postmitotic functions in differentiated cells by phosphorylating a number of transcription factors such as ELK1; phosphorylates EIF4EBP1 required for initiation of translation; phosphorylates MAP2; phosphorylates SPZ1 (by similarity); phosphorylates heat shock factor protein 4 (HSF4) and ARHGEF2 (360 aa) accessory subunit of the mitochondrial membrane respiratory chain NADH dehydrogenase (complex I), which is believed not to be involved in catalysis; complex I functions in the transfer of electrons from NADH to the respiratory chain; the immediate electron acceptor for the enzyme is believed to be ubiquinone (137 aa)
0.04
0.6
0.05
0.8
0.004
0.6
0.009
1.5
0.009
1.3
Description of protein function from KEGG.
Ontology” (REViGO)36 was used to summarize the functional information on significant GO terms by multidimensional scaling. Pathway enrichment analysis was then performed using “Kyoto Encyclopedia of Genes and Genomes” (KEGG)37,38 to determine the pathways in which the significantly up- and down-expressed proteins were implicated.
■
performed to determine the effect of Aβ42 on cell viability (Figure 1B), and decreased viability in a dose-dependent manner, relative to the control, was confirmed. Moreover, to ascertain that a process contributing to Alzheimer’s pathogenesis was initiated, we analyzed the expression of glycogen synthase kinase-type 3 (GSK3β), a protein that is activated by Aβ and induces phosphorylation of tau.39 By Western blot we detected an increase of level of expression both of GSK3β and phospho-GSK3β in a dose-dependent manner. However, given the reduction in viability in the presence of Aβ42, we chose to utilize a concentration of 40 μM for 1 h for preparative analysis by shotgun proteomic, and the experimental design is shown in Figure 1E. Proteins extracted from these and control cells were then subjected to mass spectrometry analysis. In total, 3721 proteins were identified from the control and treated cell samples. Among these proteins, 247 were found to be immediately modulated following exposure to Aβ42, and as expected, a
RESULTS
Detection of Gene Expression Changes in AD Early Events
In order to identify the immediate effect of Aβ42 on the proteome of neuronal cells and consequent alteration of processes and pathways as a step toward understanding the initial stages of AD pathology and potential biomarkers, LAN5 cells were incubated with different concentrations of Aβ42 for a short time (1 h). Microscopic inspection revealed a various degrees of degeneration and morphological changes resulting in a reduction of the cellular body, loss of neurites, and diminution of cell number (Figure 1A). An MTS assay was 1529
DOI: 10.1021/acs.jproteome.6b00925 J. Proteome Res. 2017, 16, 1526−1541
Article
Journal of Proteome Research
Figure 2. (A) Volcano plot of protein expression levels of Aβ treated LAN5 cells relative to control. Y axis indicates −log10 (p value), and X axis indicates log2 (fold change). Points significantly (p < 0.05) up- or down-expressed (log2-fold change 0.5) are displayed in purple. (B, C) Biochemical validation of MAPK1 up-regulation and PRP6 down-regulation. (B) Western blot of proteins extracted from LAN5 cells treated with Aβ42 and incubated with anti-MAPK1 and anti-PRP6. (C) Uniformity of gel loading was confirmed with β-actin as standard. Quantification of immunoreactivity was performed using densitometric analysis. *p < 0.01 versus the control group.
play a part in the up-regulation of proteins within these networks (Figure 3B). Among proteins found to be significantly down-regulated following Aβ42 treatment, two protein−protein interaction networks were found to be present. Network C1 contains proteins related to rRNA metabolism (NOLC1, DKC1, DDX21, RRP12, WDR36, NOP2, NOP58, DDX18, EBNAIBP2, and PES). The nodes of the network C2 consist mainly of proteins related to RNA metabolism, and many with the splicing process in particular (PRPF6, HNRNPs, SNRNP70, DDX23, HNRNPR, HNRNPF, HNRNPH3, HNRNPA1, HNRNPA2B1, and HNRPL) (Figure 3C).
number of proteins known to play a role in the pathology of AD were found (Table 1). These proteins are involved in processes including regulation of intracellular calcium levels (ITPR1, ATP2A2), mitochondrial respiration (NDUFB7), processing of APP by the γ-secretase complex (APH1A), and inflammation and tau processing (ERK1/ERK2). Dysregulation of these processes are well established to be key events in AD’s pathology,40−42 and our results suggest that this modulation occurs as an immediate result of high levels of Aβ42, and as such, in the earliest stages of the disease. Furthermore, a volcano plot was built from the mass spectrometry data in order to visualize significantly modulated proteins (p < 0. 05 and log2fold change 0.5). A total of 159 proteins were found to be up-regulated and 88 to be down-regulated relative to the control, and these were selected for further analysis (Figure 2A). To confirm the mass spectrometry finding two proteins, from the up-/down-regulated list, were analyzed by Western blot experiment. In agreement with bioinformatics results, MAPK1 was up-regulated and PRP6 was downregulated (Figure 2B,C).
Functional Annotation of Early AD Profile Clusters
Given the presence of clearly defined networks of interaction in the STRING analysis of our data, we hypothesized that we may be able to identify modulation of discrete processes implicated in the immediate effect Aβ42 exposure on neuronal cells. A GO approach was employed to explore this possibility by use of the bioinformatics tools DAVID and REViGO. First, DAVID was used to analyze the modulated GO terms associated with the Uniprot accession number of the under and overexpressed proteins. Of the GO terms that were found to be significantly associated (p < 0.05) with the proteins submitted, the 11 most significant were selected and presented graphically for each of the three domains: biological process, cellular component, and molecular function (Figure 4,5,6). The GO terms with which the significantly modulated proteins were annotated were submitted to REViGO in order to reduce redundant terms and present them in a multidimensional semantic matrix spaced according to their similarity. Size of circles is proportional to “Log Size” which indicates the percentage of genes annotated with the term. Thus, a larger bubble indicates a process with which a greater number of GO terms have been annotated and vice versa. The tool REViGO demonstrated a greater degree of similarity through a greater degree of clusterization in the semantic matrix containing GO terms associated with downregulated proteins (Figure 4B) than of that which was produced using the GO terms associated with the up-regulated proteins (Figure 4D). The GO terms of the clusters include biological processes such as RNA splicing, RNA processing, and chromosome organization. These findings are reflected in
Physical and Functional Association of Proteins
The bioinformatics tool STRING was used to visualize known and predicted protein−protein interaction networks between the proteins identified as modulated (Figure 3A). As the several interaction networks found were divided according to up- and down-regulation, the STRING analysis was performed on the lists divided according to ± fold change. We selected the k-means algorithm as it is unsupervised clustering algorithm based on adjacent matrix, which groups molecules based on prespecified criteria. We divided the network into seven clusters for total modulated proteins and three for up- or down-regulated proteins because these values are the best to show nonoverlapping clusters. Two networks of protein−protein interactions were seen to be present among proteins found to be significantly up-regulated following Aβ42 treatment. Network B1 (RELA, PRKACB, GNG10, MAPK1, MAPK3, YWHAQ, ARF6, DUT and GNAZ) and B2 (CDC42, ROCK2, SEPT2, EXOC5, ANXA2, and CYF1P1) were linked primarily by interactions between ROCK2 and CDC42 of network B2, and various proteins in network B1, in which map kinases 1 and 3 (MAPK1 and MAPK3) were central. This suggests that GTPase activated Rho-associated signaling may 1530
DOI: 10.1021/acs.jproteome.6b00925 J. Proteome Res. 2017, 16, 1526−1541
Article
Journal of Proteome Research
Figure 3. continued
1531
DOI: 10.1021/acs.jproteome.6b00925 J. Proteome Res. 2017, 16, 1526−1541
Article
Journal of Proteome Research
Figure 3. continued
1532
DOI: 10.1021/acs.jproteome.6b00925 J. Proteome Res. 2017, 16, 1526−1541
Article
Journal of Proteome Research
Figure 3. Known and predicted protein−protein interaction networks between up-regulated and down-regulated proteins. (A) STRING analysis of proteins significantly (p < 0.05) modulated (log2-fold change 0.5) clustered according to k-means (“7”). (B) STRING analysis of proteins significantly (p < 0.05) up-regulated (log2-fold change >0.5) clustered according to k-means (“3”). (C) STRING analysis of proteins significantly (p < 0.05) down-regulated (log2-fold change
