Dysregulation of the De Novo Proteome Accompanies Pathology

bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 Dysregulation of the de novo proteome accompanies pathology

2 progression in the APP/PS1 mouse model of Alzheimer’s disease

4 Megan K. Elder1, Hediye Erdjument-Bromage2,3, Mauricio M. Oliveira1,

5 Maggie Mamcarz1, Thomas A. Neubert2,3, Eric Klann1

7 1. Center for Neural Science, New York University, New York, NY 10003

8 2. Department of Cell Biology, New York University School of Medicine, New York, NY 10016

9 3. Kimmel Center for Biology and Medicine at the Skirball Institute, New York University

10 School of Medicine, New York, NY 10016

12 Correspondence should be addressed to Eric Klann at [email protected]

1 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

17 Abstract

18 Alzheimer’s disease (AD) is an age-related neurodegenerative disorder, but neuropathological

19 changes in AD begin years before memory impairment. Investigation of the early molecular

20 abnormalities in AD might offer innovative opportunities to target memory impairment prior to its

21 onset. Decreased protein synthesis plays a fundamental role in AD, yet the consequences for

22 cellular function remain unknown. We hypothesize that alterations in the de novo proteome drive

23 early metabolic changes in the hippocampus that persist throughout AD progression. Using a

24 combinatorial amino acid tagging approach to selectively label newly synthesized proteins, we

25 found that the de novo proteome is disturbed in APP/PS1 mice prior to pathology development,

26 affecting the synthesis of multiple components of synaptic, lysosomal and mitochondrial

27 pathways. Furthermore, the synthesis of large clusters of ribosomal subunits were affected

28 throughout neuropathology development in these mice. Our data suggest that large-scale

29 changes in protein synthesis could underlie cellular dysfunction in AD.

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

42 Introduction

43 Alzheimer’s disease (AD) is an age-related neurodegenerative condition characterized by

44 progressive and devastating cognitive impairment. AD is classically characterized by extensive

45 deposition of amyloid beta (Aβ) plaques and intracellular inclusions of hyperphosphorylated tau,

46 in the form of neurofibrillary tangles (Alzheimer, 1907). In addition to the deposition of plaques

47 and tangles, AD is characterized by extensive atrophy of the brain, which follows a prescribed

48 trajectory throughout disease progression. Initially atrophy is restricted to the hippocampi and

49 medial temporal lobes, even before the appearance of symptoms (Scahill et al., 2002). The

50 cause of this neuronal atrophy has not yet been elucidated, but it is likely that pathological

51 alterations in protein synthesis are a contributing factor (Hernández-Ortega et al., 2016). In a

52 healthy biological system, proteins are made and degraded in a controlled process termed

53 protein homeostasis (proteostasis). However, the intrinsic and extrinsic stressors that

54 accumulate with age and disease can challenge the delicate balance between protein synthesis

55 and degradation (Kaushik and Cuervo, 2015).

56 Removal of misfolded or otherwise damaged proteins is essential to avoid toxicity and is

57 regulated by a variety of proteolytic systems, whose dysfunction has been implicated in AD

58 pathophysiology (Ciechanover and Kwon, 2015). Proteasomal activity is perturbed in vulnerable

59 cortical areas and the hippocampus in late stage AD (Keller et al., 2000), whereas proteolysis

60 has been shown to be impaired in very mild stages of the disease, suggesting that proteostasis

61 disruption might occur prior to the development of symptoms (Mawuenyega et al., 2010).

62 On the opposite side of the proteostatic balance is protein synthesis. Following

63 transcription, mRNA is translated to a polypeptide chain via ribosomal activity, a process that in

64 neurons occurs both somatically and locally in dendrites and axons (Holt et al., 2019). This de

65 novo protein synthesis is a dynamic process, as it is necessary for basal neuronal function,

66 responding to stimuli that induce long-lasting plasticity, and for memory consolidation (Costa-

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

67 Mattioli et al., 2009; Richter and Klann, 2009). In AD, multiple studies have observed

68 dysregulation of translation throughout the disease process, as indicated by changes in p70 S6

69 kinase 1, eukaryotic initiation factors 2 alpha (eIF2α) and 4E (eIF4E) phosphorylation (An et al.,

70 2003; Chang et al., 2002; Ferrer, 2002; Li et al., 2004; Lourenco et al., 2013; Ma et al., 2013),

71 as well as by decreased ribosomal function (Ding et al., 2005; Hernández-Ortega et al., 2016;

72 Langstrom et al., 1989; Sajdel-Sulkowska and Marotta, 1984). Curiously, altered expression of

73 ribosomal RNAs and ribosomal protein-coding mRNAs in the hippocampus appear even prior to

74 the onset of symptoms and the development of hallmark AD pathologies (Ding et al., 2005;

75 Hernández-Ortega et al., 2016). Furthermore, ribosomal dysfunction is specific to cortical areas

76 that show the greatest atrophy in later stages of the disease, including the inferior parietal lobule

77 and superior middle temporal gyri (Ding et al., 2005).

78 Although alterations in the level of de novo protein synthesis in AD have been implied for

79 several years, the mechanistic details involved in this dysregulation remain largely unknown

80 (Buffington et al., 2014). When examined in APP/PS1 mutant mice, which express familial AD-

81 associated mutations in the amyloid precursor protein (APPswe) and presenilin-1 genes (PS1ΔE9)

82 (Ma et al., 2013), an approximately 30% decrease in de novo protein synthesis was observed in

83 aged symptomatic APP/PS1 mice compared to wild-type littermates (Ma et al., 2013).

84 Elucidating the identity of these dysregulated proteins has become an important focus of

85 research, and two independent studies of transgenic mice that experimentally model different

86 aspects of AD pathology have used in vivo metabolic labelling to isolate de novo synthesized

87 proteins for subsequent mass spectrometry analysis (Evans et al., 2019; Ma et al., 2020). The

88 use of a less complex system, such as live isolated hippocampal slices, offers a different

89 resolution and a narrower timeframe for this snapshot of the de novo proteome.

90 Herein, we used BONLAC, a combinatorial approach of stable isotope labelling by amino

91 acid tagging (SILAC) and biorthogonal noncanonical amino acid tagging (BONCAT), and mass

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

92 spectrometry to compare de novo protein synthesis in acute hippocampal slices from

93 asymptomatic and symptomatic APP/PS1 mutant mice for the identification and measurement

94 of relative abundance of de novo synthesized proteins (Bowling et al., 2019; Bowling et al., 2016;

95 Dieterich et al., 2006; Eichelbaum et al., 2012; Evans et al., 2019; Ong et al., 2002; Zhang et al.,

96 2014). Here, we show for the first time a significant dysregulation of proteostasis in APP/PS1

97 mutant mice even before symptom onset, with significant alterations in networks of proteins

98 involved in both protein degradation and synthesis. Further, in aged APP/PS1 mice that display

99 memory impairments (Volianskis et al., 2010), we observed dysregulation of both lysosomal and

100 mitochondrial proteins, as well as components of the ribosome. Together, these findings suggest

101 that alterations in the synthesis of proteins involved in a variety of critical cellular pathways are

102 impaired early in the AD process and likely underlie the deterioration of neuronal function,

103 leading to loss of memory.

104

105

106

107

108

109

110

111

112

113

114

115

116

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

117 Results

118 Aging-related downregulation of hippocampal protein synthesis in APP/PS1 mutant mice

119 Previous studies using puromycin and azidohomoalanine (AHA) incorporation to label newly

120 synthesized proteins in mice modeling aspects of AD pathology demonstrated a reduction in de

121 novo protein synthesis (Beckelman et al., 2019; Evans et al., 2019; Ma et al., 2013; Ma et al.,

122 2020), and the translation efficiency of polyribosomes isolated from the mild cognitive

123 impairment (MCI) and end-stage AD brain is reduced by >60% in affected cortical areas (Ding

124 et al., 2005). We first confirmed that histopathological changes are observed in the brain of

125 asymptomatic vs symptomatic APP/PS1 mice by performing immunohistochemistry against Ab

126 in brain sections including the dorsal hippocampus. As expected, we found a robust increase in

127 the number and size of amyloid deposits in the brains of symptomatic APP/PS1 mice when

128 compared to asymptomatic APP/PS1 mice (Figure 1A-C). Using BONCAT labeling, we observed

129 a ~20% decrease in the level of de novo protein synthesis in symptomatic (12+ months old)

130 APP/PS1 mice compared to asymptomatic APP/PS1 mice (3-5 months-old; Figure 1D).

131 Collectively, these findings support and extend previous work conducted in our laboratory, which

132 highlighted a significant decrease in de novo protein synthesis between symptomatic wild type

133 and APP/PS1 mice using SUnSET, another non-radioactive method for tagging and monitoring

134 new protein synthesis (Ma et al., 2013). Therefore, using BONLAC labeling we proceeded to

135 identify changes in the de novo proteome in the APP/PS1 mutant mice before and after symptom

136 onset (Figure 1E).

137

138 Large-scale proteomic dysfunction is observed early in pathology development in the

139 hippocampus of APP/PS1 mice

140 In order to identify candidate proteins of interest, we used BONLAC and a previously published

141 multi-layered analysis based in coincidence detection (Bowling et al., 2019; Bowling et al., 2016).

6 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

142 In the hippocampi of asymptomatic mice (3-5 months-old), 2510 de novo synthesized proteins

143 were measured in at least one sample. 1826 proteins were measured across the majority of

144 samples (Supplementary Figure 1), and of these 1826, 180 were differentially regulated in the

145 APP/PS1 mice (Figure 2A; Supplementary Figure 2). The majority of differentially regulated

146 proteins were downregulated (fold change ≤ 0.8; 107 or 5.9% of total proteins), whereas 73

147 proteins were upregulated (fold change ≥ 1.2; 4.0% of total proteins; Figure 2B) in the APP/PS1

148 mice. When the biological relevance of these changes were interrogated using DAVID and

149 StringDB, the GO and KEGG pathways that were highlighted included the cellular process

150 ‘Alzheimer’s disease’, as well as proteostasis-related components, gathered under the terms

151 ‘proteasome’ and ‘ribosome’ (Figure 2C-D). These findings indicate that even in young APP/PS1

152 mice that do not show display memory deficits, processes previously linked to AD and critical to

153 cell functioning are disturbed.

154

155 Alterations in the de novo proteome correspond with pathologies in symptomatic

156 APP/PS1 mice

157 Following these investigations in asymptomatic APP/PS1 mice, we investigated the impact of

158 aging and the development of AD-like pathology on the de novo proteome. As expected from

159 the decreased rate of translation apparent in AHA incorporation levels (Figure 1D), fewer newly

160 synthesized proteins were observed in APP/PS1 mice aged ≥ 12 months compared to younger

161 mice. In the hippocampi of symptomatic mice, 2065 proteins were measured at least once

162 (Supplementary Figure 3), 855 proteins were measured across the majority of samples, and 113

163 proteins were consistently altered in the APP/PS1 mice (Figure 3A; Supplementary Figure 4).

164 Proteins whose synthesis was reduced by 20% compared to age-matched wild-type mice made

165 up 2.7% of the consistently detected proteins (23 proteins), while BONLAC detected 90 proteins

166 with increased de novo synthesis in the symptomatic APP/PS1 mice (10.6%; Figure 3B).

7 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

167 Through GO analysis via DAVID and visualization using StringDB, we observed several

168 functional clusters (Figure 3C), with the top hits including ‘Alzheimer’s disease’, ‘ribosome’, and

169 ‘lysosome’ (Figure 3D). Together, these findings indicate that the dysregulation of core cellular

170 processes relies at least partially on impaired protein synthesis, and that changes in de novo

171 proteome happen even before full pathology onset.

172

173 Dysregulation of the synthesis of BONLAC-identified candidates is reflected in protein

174 expression

175 Next, to understand whether these changes in de novo protein synthesis corresponded to altered

176 protein expression, we validated candidates from several enriched clusters using western

177 blotting, comparing changes in expression between APP/PS1 and wild-type littermates. In order

178 to validate differences from the same group used for proteomic studies, we reserved BONLAC-

179 labeled acute hippocampal slices from animals prior to sample preparation for mass

180 spectrometry analysis, and instead lysed the tissue for western blot analysis (Figure 4A).

181 Alongside APP, which was selected as a positive control (Radde et al., 2006), other AD-

182 associated proteins known to be involved in lysosome-mediated protein degradation (cathepsin-

183 d; ctsd (Cataldo et al., 1995; Hamazaki, 1996)) and synaptic plasticity (neuromodulin; GAP-43

184 (Bogdanovic et al., 2000; de la Monte et al., 1995)) were probed. Samples were also probed for

185 the expression of the large 60S ribosomal proteins Rpl13 and Rpl18. As shown in Figure 4B and

186 Figure 4C, the de novo synthesis and the total expression of APP were increased in both age

187 groups of APP/PS1 mice, as expected. Neither the synthesis nor the expression of Ctsd protein

188 were altered in asymptomatic APP/PS1 mice compared to wild-type littermates. However in

189 aged, symptomatic mice, a similar trend was found in both BONLAC and total protein levels. In

190 contrast to younger mice, the older APP/PS1 mice also showed significantly reduced synthesis

8 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

191 and total expression of the synaptic protein GAP-43, which corresponds with a known decline in

192 synaptic density at this age (Alonso-Nanclares et al., 2013).

193 The large 60S ribosomal subunit protein Rpl13 also was examined after BONLAC

194 labelling indicated synthesis of this protein was decreased in APP/PS1 mutant mice both before

195 and after the appearance of symptoms. Western blot analysis confirmed that there was a non-

196 significant trend in the reduction in expression of this ribosomal protein in both groups of

197 APP/PS1 mice. The expression of an additional ribosomal protein, Rpl18, was investigated as

198 BONLAC labelling indicated the synthesis of this protein varied throughout the development of

199 pathology in the APP/PS1 mice. In asymptomatic mice, both newly synthesized and total

200 expression of Rpl18 were decreased at trend level relative to wild-type mice, whereas both

201 measures were increased in aged symptomatic APP/PS1 mice. Next, we performed a linear

202 regression analysis between the BONLAC and western blot fold changes (APP/PS1 vs. wild

203 type) to evaluate whether a linearity existed between the values. Linear regression analysis

204 highlighted a significant correlation between detection of the candidate protein in the de novo

205 analysis vs. candidate protein expression in the total protein fraction (Figure 4D). Together,

206 these results indicate that altered synthesis of proteins detected with BONLAC are correlated

207 with changes in global protein expression.

208

209 Cluster analysis of commonly synthesized proteins in both asymptomatic and

210 symptomatic APP/PS1 mice highlights key AD pathways

211 Following the corroboration of BONLAC candidates by western blotting, we further investigated

212 the overall patterns of protein synthesis regulation observed in the APP/PS1 mice. Although

213 examination of significant changes (as detected by C-score analysis) reveal statistically relevant

214 alteration of the de novo proteome, it is noteworthy that smaller fluctuations in expression (fold

215 change < 20%: labelling ratio > 0.8 or < 1.2) can also provide information as to whether a specific

9 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

216 cell process is altered. Therefore, we compared all 791 newly synthesized proteins identified in

217 both asymptomatic and symptomatic APP/PS1 mice and their wild-type littermates. To

218 investigate whether any biological pathways were differentially affected, we conducted

219 hierarchical clustering analysis using Cytoscape. Several key clusters were observed (Figure 5,

220 Supplementary Table 1), closely associated with the KEGG pathways ‘Alzheimer’s disease’

221 (Figure 5 Cluster 5; Supplementary Figure 9) and ‘protein processing in the endoplasmic

222 reticulum’ (Figure 5 Cluster 1; Supplementary Figure 5). The GO terms ‘synapse’ (Figure 5

223 Cluster 2; Supplementary Figure 6), and ‘synaptic vesicle cycle (Figure 5 Cluster 3;

224 Supplementary Figure 7) were enriched in this analysis, as were the GO terms ‘axo-dendritic

225 transport’ (Figure 5 Cluster 4; Supplementary Figure 8), ‘myelin sheath’ (Figure 5 Cluster 5;

226 Supplementary Figure 9), ‘mitochondrial part’ and ‘electron transfer activity’ (Figure 5 Cluster 6;

227 Supplementary Figure 10).

228

229 Dysregulation of ribosomal protein synthesis is a common feature in APP/PS1 mice

230 throughout the development of symptoms

231 As described above, C-score detection of candidates generated a list of proteins that were highly

232 dysregulated in the APP/PS1 hippocampus (+/- 20% compared to wild-type). When these

233 dysregulated candidate proteins were compared between age groups, only 31 proteins were

234 significantly dysregulated in both the asymptomatic and symptomatic APP/PS1 mice relative to

235 age-matched wild-type littermates (Figure 6A). The majority of these proteins showed either an

236 increased level of synthesis at both ages (11/31) or reversed expression (decreased synthesis

237 in asymptomatic APP/PS1 mice and increased synthesis in symptomatic mice, or vice versa;

238 15/31) in comparison to wild-type mice. Only five proteins showed consistently decreased

239 synthesis in the APP/PS1 mice, regardless of age (Figure 6B). Identification of the proteins

240 consistently dysregulated in the APP/PS1 mice highlighted a significant cluster of ribosomal

10 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

241 proteins, three of which are components of the small 40S ribosomal subunit, and six of which

242 are components of the large 60S ribosomal subunit (Figure 4C). Taken as a whole, these results

243 support the notion that alterations in the expression of ribosomal proteins is an early feature of

244 AD pathology, and may well play a role in protein synthesis dysfunction through the course of

245 disease.

246

247

248

249

250

251

252

253

254

255

256

257

258

259

260

261

262

263

264

265

11 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

266 Discussion

267 Dysregulated proteostasis underlies many of the hallmark pathologies of AD. Both the deposition

268 of Aβ in the form of plaques and accumulation of hyperphosphorylated tau-containing tangles

269 suggest impaired protein degradation systems, and indeed, dysfunctional proteolysis has been

270 observed in AD brains (Ciechanover and Kwon, 2015). Further, translation is impaired

271 throughout the AD process [10-17] and de novo protein synthesis is decreased in mouse lines

272 that model various aspects of AD pathology [14, 23, 62]. The identity of the proteins that are

273 differentially synthesized before and after AD symptom onset in the human brain are unknown.

274 Interestingly however, ribosomal function was shown to be impaired in the brains of individuals

275 diagnosed with mild cognitive impairment (Ding et al., 2005).

276 Recently, the de novo proteome of mouse lines that model aspects of AD have become

277 a focus of study. Decreased de novo protein synthesis was observed in neurons with high levels

278 of hyperphosphorylated tau in the K369I (K3) and rTg4510 transgenic mouse models of

279 tauopathy and neurodegeneration (Evans et al., 2019). Moreover, and consistent with our

280 findings, investigations using in vivo metabolic labelling has highlighted significant dysregulation

281 of the synthesis of proteins involved in vesicle transportation and mitochondrial functioning prior

282 to symptom onset in APP/PS1 mice (Ma et al., 2020). This study also investigated changes in

283 the proteome both pre- and post-symptom onset, but the young, asymptomatic mice selected

284 for the study (2 months-old) do not yet show hallmark Aβ deposition in the hippocampus, which

285 begins at 3-4 months of age (Radde et al., 2006). Moreover, although impairments in

286 hippocampal LTP and cognitive ability are apparent by 9 months of age (Gengler et al., 2010;

287 Radde et al., 2006; Serneels et al., 2009), the age selected for the symptomatic group in this

288 work, several studies have indicated that older mice (>12 months old) possess greater deficits

289 in spatial learning and memory (Gruart et al., 2008; Park et al., 2006; Trinchese et al., 2004; Xu

290 et al., 2013).

12 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

291 In the present study, we used BONLAC labelling in acute hippocampal slices from ~4

292 month-old asymptomatic and ~12 month-old symptomatic APP/PS1 mice and their wild-type

293 littermates to isolate and identify alterations in the de novo hippocampal proteome. The use of

294 animals at these ages both supports the recent in vivo work described above (Ma et al., 2020),

295 and extends the coverage of the de novo proteome investigations throughout pathology

296 development in this model. Decreased synthesis was observed in aged APP/PS1 mice, as

297 represented by lower AHA incorporation and fewer proteins identified in the mass spectrometry

298 screen, as anticipated from prior work indicating an age-dependent decrease in translation in

299 this AD model (Ma et al., 2013). At both timepoints, the vast majority (>85%) of measured

300 proteins were synthesized consistently regardless of genotype, further illustrating the tight

301 regulation of the proteome required for cell functioning. However, defined patterns of

302 dysregulation were observed in APP/PS1 mice relative to unaffected littermates.

303 In young APP/PS1 mice that do not yet show symptoms, we observed a specific pattern

304 of dysregulated protein synthesis, resulting in predominately decreased translation of proteins

305 involved in a series of pathways critical to cell functioning. Of note, proteins involved in both

306 protein degradation (via GO term ‘proteasome’) and protein synthesis (via GO term ‘ribosome’)

307 showed dysregulated synthesis in the asymptomatic APP/PS1 hippocampus relative to wild-type

308 littermates. In fact, the dysfunctional production of components of the proteasome could have a

309 negative effect on ribosomal function and general protein synthesis, as an interplay between

310 impaired proteasomal function and translation has previously been described (Ding et al., 2005).

311 Dysregulated synthesis of synaptic proteins also were highlighted at this time point, which could

312 underlie the decreased synaptic density previously observed in 4 month-old APP/PS1 mice

313 (Smith et al., 2009), as well as the impaired synaptic transmission observed in APP/PS1 mice

314 at 5 months of age (Zhang et al., 2005). Thus, our findings may represent a critical window in

13 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

315 which modulation of translation limits synaptic changes that result in memory deficits later in

316 disease progression.

317 When de novo protein synthesis was examined in 12+ month-old APP/PS1 mice that

318 display pronounced AD-like pathology and memory impairment, a large number of mitochondrial

319 proteins previously linked to AD were dysregulated. Mitochondrial structure and function is

320 known to be compromised in human patients (Baloyannis, 2006; Hirai et al., 2001), but whether

321 these deficits contribute to the disease or whether they are a biproduct of disease progression

322 continues to be debated (Swerdlow, 2018). Alongside mitochondria, lysosomal protein synthesis

323 was dysregulated in the symptomatic APP/PS1 mice. Increasing evidence suggests that

324 lysosomal biogenesis is initially upregulated in early stages of AD (Cataldo et al., 1995), before

325 progressive dysfunction throughout the disease process results in the accumulation of enlarged

326 lysosomes, which fail to effectively degrade their contents (Nixon, 2017).

327 A final significant pathway that was observed in the analysis of dysregulated proteins

328 following symptom onset was constituents of the cytosolic ribosome, including eight ribosomal

329 proteins and nucleolar protein of 40 kDa (Zcchc17). This zinc ion binding protein has recently

330 been identified as a modulator of synaptic gene expression in AD (Tomljanovic et al., 2018).

331 Moreover, Zcchc17 is thought to be involved in ribosome biogenesis and maturation (Chang et

332 al., 2003). Two protein components of the ribosome that were differentially regulated in the

333 mature APP/PS1 tissue, Rpl13 (decreased BONLAC labelling compared to wild-type) and Rpl18

334 (upregulated in APP/PS1 mice) were selected for validation analyses in hippocampal lysates.

335 The total expression of both proteins reflected the level of synthesis as detected in the mass

336 spectrometry screen, indicating that these are biologically relevant changes that require further

337 investigation. Rpl13 has recently been described as a ‘core’ ribosomal protein included in all

338 actively translating ribosomes (Shi et al., 2017), and yet has also been shown to be dysregulated

339 at the gene expression level in hippocampal tissue from severe AD patients (Kong et al., 2009).

14 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

340 Although no study has yet linked alterations in Rpl18 expression with AD pathology, recent

341 analysis has indicated dysregulation of Rpl18 gene expression occurs early in the AD process

342 (Martínez-Ballesteros et al., 2017).

343 In order to determine the full impact of APP/PS1 mutations on protein synthesis in these

344 mice, we examined the biological pathways that exhibited changes across pathology

345 development as detected by hierarchical cluster analysis of all proteins detected in both age

346 groups. This list was not limited to proteins identified by C-score analysis as significantly

347 dysregulated, and included proteins detected in the majority of samples which showed a fold

348 change of <20%. Several functional clusters emerged, indicating pathways specific to stage of

349 pathology development. Two clusters were specifically upregulated in the symptomatic APP/PS1

350 mice, those associated with the GO terms ‘protein processing in the endoplasmic reticulum (ER)’

351 and ‘mitochondrial part: electron transfer activity’. Increased synthesis of proteins, including

352 Stub1/CHIP, Cat and Hsph1, which have previously been implicated in the AD process (Habib

353 et al., 2010; Poirier et al., 2019; Singh and Pati, 2015), indicate that the unfolded protein

354 response (UPR), the physiological response to the ER stress, might be chronically activated

355 later in disease. This finding is consistent with our previous studies showing that reducing the

356 expression of PERK, an eIF2a kinase that triggers downregulation of protein synthesis in

357 response to UPR activation, could restore hippocampal plasticity and memory deficits in

358 symptomatic APP/PS1 mice (Ma et al., 2013).

359 Mitochondrial dysfunction is known to be an early event in both APP/PS1 mice (Chen et

360 al., 2019) and human AD brain pathology (Hauptmann et al., 2009; Moreira et al., 2007;

361 Nunomura et al., 2001), and is believed to play a role in the synaptic loss that occurs early in the

362 disease process (Du et al., 2012). Downregulated synthesis of mitochondrial proteins,

363 specifically the oxidative phosphorylation-associated proteins Ndufs5, Ndufa12, Cox4i1 and

364 Cox6b in the asymptomatic APP/PS1 mice could underlie the decreased electron transport chain

15 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

365 capacity observed in this and other models of AD-like pathology (Bo et al., 2014; David et al.,

366 2005).

367 Functional clusters that exhibited downregulation in the aged APP/PS1 mice compared

368 to the asymptomatic cohort were also observed. The first cluster was associated with the GO

369 term ‘synapse’, which may be a function of the decreased spine density previously observed in

370 symptomatic APP/PS1 mice (Smith et al., 2009). Proteins found in this hub included those

371 involved in glutamatergic signaling, known to be impaired in both the human AD brain and

372 APP/PS1 mice (Minkeviciene et al., 2008). In addition, a cluster corresponding to ‘glycolysis and

373 gluconeogenesis’ was decreased in the aged APP/PS1 hippocampal de novo proteome.

374 Decreased glucose metabolism has been observed in the human AD brain using radioactive

375 glucose labelling and PET scanning, and correlates well with AD-associated pathologies

376 (Landau and Frosch, 2014; Markesbery, 2010; Martins et al., 2018).

377 Although the cluster analysis described above examined general trends, we also

378 conducted focused analysis using C-score-mediated detection of dramatic changes in de novo

379 protein synthesis, where candidate proteins were selected for further analysis if they showed a

380 fold-change of >20% in either direction in APP/PS1 mice vs. wild-type littermates. Comparison

381 of proteins selected in this manner in asymptomatic and symptomatic mice revealed that

382 although few proteins were dysregulated in both groups, one third of these proteins were

383 identified as ribosomal subunits. The ribosome is composed of four ribosomal RNA molecules

384 and 80 ribosomal proteins (Yoshihama et al., 2002), which together form the small and large

385 ribosomal subunits that function together to translate mRNAs into proteins. Recent examination

386 of de novo hippocampal protein synthesis following prolonged in vivo metabolic labelling

387 revealed dysregulated synthesis of Rps3a (Ma et al., 2020). Moreover, dysregulated expression

388 of genes encoding several of the ribosomal proteins observed in our study (Rpl7, Rps6, Rps17

389 and Rps26) has previously been described in the AD brain (Hernández-Ortega et al., 2016).

16 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

390 Notably, decreased synthesis of ribosomal proteins is a common feature in both amyloid and

391 tauopathy mouse models, as a recent de novo proteomic study using K3 and rTg4510 models

392 of tauopathy revealed dysfunctional translation of these proteins (Evans et al., 2019).

393 The protein content of functional ribosomes was long thought to be largely homogenous,

394 but recent studies have indicated sub-stoichiometric inclusion of ribosomal proteins in actively-

395 translating ribosomes in mammalian cells (Shi et al., 2017; Slavov et al., 2015). In neurons,

396 active remodeling of ribosomal protein content has been observed in response to axonal

397 stimulation (Shigeoka et al., 2019). Notably, ribosomal protein gene expression is regulated

398 throughout the aging process in the brain (Ximerakis et al., 2019), indicating that the dynamic

399 regulation of ribosomal components may play a role in cellular adaption to aging. Further,

400 deficiency in particular ribosomal proteins results in hypersensitivity to deficits in the protein

401 degradation system (McIntosh et al., 2011). Importantly, the protein constituents of the actively

402 translating ribosome appear to confer a level of preference for which mRNAs are translated,

403 suggesting a further level of translational control (Shi et al., 2017). Together with the

404 aforementioned studies, our findings make a strong case for further detailed investigations into

405 the dysregulation of ribosomal protein synthesis and its subsequent consequences in AD.

406 One important consideration when using BONLAC labeling is that the primary output is

407 the ratio of medium to heavy isotope labelling, which is used to generate the fold-changes that

408 permit relative quantitation. A caveat to using this system in AD (in which translation is known to

409 be impacted) is that if a protein is not synthesized at all during the labeling window, or synthesis

410 levels are below detection in one sample within the pair, a ratio will not be generated. As we

411 have observed, both here (Figure 1A) and previously (Ma et al., 2013; Ma et al., 2020), a

412 significant decline in gross de novo protein synthesis in the APP/PS1 mouse (Ma et al., 2013;

413 Ma et al., 2020), this inherent bias may have led to an exaggerated detection of upregulated vs.

414 downregulated proteins in the symptomatic APP/PS1 hippocampal samples. Bioinformatic

17 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

415 comparison of the de novo hippocampal proteome generated in this, and recent in vivo analysis

416 (Ma et al., 2020), with the basal proteome could reveal proteins that were not synthesized over

417 the labelling period and were thus precluded from BONLAC ratio generation. Further by

418 combining BONLAC labelling and TMT sample multiplexing, the number of missing peptide

419 quantification values in each sample is greatly reduced, providing deeper coverage of the de

420 novo proteome (Zecha et al., 2019). The incorporation of this novel approach into subsequent

421 studies promises to reveal new insights into the cellular processes that accompany pathology

422 progression in the APP/PS1 mouse.

423 In summary, here we have illustrated a compelling picture of dysregulation of the de novo

424 proteome in the APP/PS1 mouse model of AD-like amyloidy, both before and after the

425 appearance of symptoms. Following validation of several candidate proteins to support the

426 relevance of the changes observed in the mass spectrometry screen, we conducted robust

427 bioinformatic analysis of the newly made proteins. We observed dysregulated synthesis of

428 proteins involved in several pathways known to be altered throughout the course of AD, including

429 those involved in synaptic, mitochondrial and lysosomal functions, but we also observed

430 significant disturbance of protein components of the ribosome. In light of recent research that

431 indicates inclusion or exclusion of specific ribosomal subunits bestows selectivity to mRNA

432 translation, the significant up- and down-regulation of ribosomal protein synthesis identified here,

433 even prior to the development of pathology, may underly the progressive deterioration of cellular

434 function and memory observed in this AD model, and in individuals with AD.

435

436

437

438

439

18 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

440 Materials and Methods

441 Animals

442 All procedure involving animals were performed in accordance with protocols approved by the

443 New York University Animal Welfare Committee and followed the National Institutes of Health

444 (NIH) Guide for the Care and Use of Laboratory Animals. All mice were house in the New York

445 University animal facility. Mice of both sexes were used. APP/PS1 transgenic mice (B6;C3-

446 Tg(APPswe, PSEN1/dE9)85Dbo/Mmjax) and their wild-type littermates were bred and

447 maintained on C57-BL6 and B6.C3 (Jackson Labs) backgrounds. Mice were housed with their

448 littermates in groups of two to three animals per cage and kept on a 12-hour regular light/dark

449 cycle, with food and water provided ad libitum. All genotypes were verified by polymerase chain

450 reaction. Mice were used at an age of either 3-5 months or 12-15 months.

451

452 Immunohistochemistry

453 Mice were deeply anesthetized with ketamine (150 mg/kg) and transcardially perfused with 0.1M

454 PBS followed by 4% PFA before brains were removed and post-fixed for 48 hours. 40 μm free-

455 floating coronal sections containing the hippocampus were cut using a Leica vibratome,

456 collected and stored at 4°C in 0.01% sodium azide until use. Sections were permeabilized in

457 0.5% Triton X-100 (15 min) prior to blocking (5% normal goat serum, 0.1% Triton X-100; 1 h).

458 Slices were incubated overnight at 4°C in rabbit anti-amyloid beta antibody (1:200; clone 6E10,

459 Enzo Life Sciences), followed by Alexa-488-labelled goat anti-rabbit secondary antibody (1:500;

460 RT, 1.5 h). Slices were mounted using ProLong Gold Antifade Mountant with DAPI. Tile-scan z

461 stack images (10-15 optical sections depending on slice thickness) were collected using an SP8

462 confocal microscope (Leica) with a 20X magnification lens and Leica LASX software, with laser

463 intensity, smart gain and offset maintained throughout the experiment. Images were processed

464 using ImageJ 2.0.0 using the Bio-Formats importer plug-in. Plaque number was manually

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

465 quantified using the Cell Counter plugin (n = 1-2 animals/group, 1 dorsal hippocampal

466 slice/animal).

467

468 AHA and SILAC Dose

469 AHA and SILAC labels were used as previously described (Bowling et al., 2016). AHA was

470 purchased from Click Chemistry Tools, AZ, USA, and SILAC amino acids (13C6-15N2-lysine,

471 13C6-15N4-arginine (Lys8/Arg10) and D4-lysine/13C6-arginine (Lys4/Arg6)) were obtained

472 from Cambridge Isotope Laboratories, MA, USA and used at previously described

473 concentrations (Bowling et al., 2016; Zhang et al., 2014). Assignment of SILAC labels to

474 experimental conditions (wild-type or APP/PS1) was alternated between biological replicates to

475 ensure that results were not biased by labelling.

476

477 Acute Hippocampal Slice Preparation and Incubation

478 400 μm transverse hippocampal slices were obtained as previously described (Chévere-Torres

479 et al., 2012; Kaphzan et al., 2013; Santini et al., 2015). Slices were recovered and incubated in

480 ACSF for 20 min at room temperature, then at 32°C for 45 min. AHA (1 mM) and SILAC amino

481 acids were added to the ACSF, and slices were incubated for 5 hrs. Following the labelling

482 period, slices were immediately flash frozen for mass spectrometry and stored at - 80 °C until

483 use.

484

485 BONLAC Sample Preparation for Mass Spectrometry.

486 Samples were prepared as previously described (Bowling et al., 2016) using the Click-IT Protein

487 Enrichment Kit (ThermoFisher Scientific). BONLAC was carried out with a minimum of five runs

488 per condition, with each run examining hippocampal slices from one wild type and one age-

489 matched APP/PS1 animal (n = 5-7 biological replicates made up of 1 APP/PS1 and 1 wild type;

20 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

490 5-7 animals of each age per genotype were used in total). Briefly, following labeling, equal

491 weights of tissue from age-matched pairs of wild type and APP/PS1 animals were lysed together

492 in buffer containing 8 M urea, 200 mM Tris pH 8, 4% CHAPS, 1 M NaCl and protease inhibitor

493 cocktail (cOmplete, Mini, Roche; two tablets per 10 mL of lysis buffer). The lysate was sonicated

494 before AHA-labeled proteins were covalently coupled to alkyne-tagged agarose beads using

495 reagents provided in the kit. Beads were washed repeatedly with SDS (1% SDS, 100 mM Tris

496 pH 8, 250 mM NaCl and 5 mM EDTA) and alkyne-bound proteins were reduced with DTT at 70

497 °C before alkylation with iodoacetamide protected from light at room temperature. Beads were

498 then washed sequentially to remove non-specifically bound proteins with 100 column volume

499 SDS wash buffer, 8 M urea, and finally with 20% acetonitrile. Bound proteins were digested on-

500 resin with trypsin (Trypsin Gold, Mass spectrometry grade, Promega) at 37 °C overnight in 25

501 mM ammonium bicarbonate, and the resulting tryptic peptides were desalted using hand-packed

502 StageTips (Rappsilber et al., 2007). Desalted peptides were dried to a small droplet, under

503 vacuum, in a SpeedVac.

504

505 Liquid Chromatography and Mass Spectrometry

506 LC-MS was conducted using a Thermo Scientific EASY-nLC 1000 coupled to a Q Exactive, High

507 Field mass spectrometer (ThermoFisher Scientific) equipped with a nanoelectrospray ionization

508 source. Peptide separation was achieved with a self-packed ReproSil-Pur C18 AQ 3μ reverse

509 phase column (Dr.Maisch GmbH, Germany, 75 μm inner ID, ~25cm long). Peptides were eluted

510 via a gradient of 3 - 40% acetonitrile in 0.1% formic acid over 120 min at a flow rate of 250 nL/min

511 at 45 °C, maintained with a column heater (Sonation GmbH, Germany). The mass spectrometer

512 was operated in data-dependent mode with survey scans acquired at a resolution of 120,000 at

513 m/z 400. Up to the top 15 most abundant precursors from the survey scan were selected with

514 an isolation window of 1.6 Thomsons and fragmented by higher energy collisional dissociation

21 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

515 with normalized collision energies of 27. The maximum ion injection times for the survey scan

516 and the MS/MS scans were 60 ms, and the ion target value for both scan modes were set to

517 1,000,000.

518

519 Protein Identification and Quantitation

520 Raw files obtained from mass spectrometry runs were processed using the MaxQuant

521 computational proteomics platform (Version 1.5.5.1(Cox and Mann, 2008)) for peptide

522 identification and quantitation. Fragmentation spectra were searched against the Uniprot mouse

523 protein database (downloaded on 12/20/2017, containing 16,950 non-redundant protein entries,

524 combined with 262 common contaminants), allowing for up to two missed tryptic cleavages.

525 Cysteine carbamidomethylation was set as a fixed modification, and methionine oxidation,

526 acetylation of protein N-terminal, D-4 lysine, 13C6-arginine 13C6-15N2-lysine and 13C6-15N4-

527 arginine were used as variable modifications for the database search. The mass tolerances were

528 set to 7 ppm for precursor, and 10 ppm for fragment respectively. A false discovery rate (FDR)

529 of 1% was applied to both peptide and protein identifications.

530

531 Data Availability

532 The raw mass spectrometry data generated during this study are available at MassIVE (Center

533 for Computational Mass Spectrometry, University of California, San Diego) with the accession

534 number (ftp://massive.ucsd.edu/MSV000085962/). Note to reviewers: data can be accessed

535 prior to publication using the following details: User: MSV000085962; Password: a

536

537 Computational Processing of BONLAC MS data

538 MaxQuant-normalized H/M ratios (heavy vs. medium isotopes) were used for quantitative

539 analysis. Ratios were inverted for experiments where isotopic labelling was reversed. A custom-

22 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

540 made R script was used to select protein candidates where the average ratio was above or

541 below an arbitrary threshold of 20% (>0.8 or <1.2). Only proteins that were measured in the

542 majority of samples and were consistently up- or down-regulated following manual examination

543 of the dataset were selected. This method allows for non-biased selection of up- or down-

544 regulated proteins, taking variation between replicates into account.

545

546 Protein Identity and Interaction Interpretation

547 Gene Ontology analysis was performed using the Database for Annotation, Visualization and

548 Integrated Discovery (DAVID version 6.8) (Sherman and Lempicki, 2009). Candidate proteins

549 selected according to the methods outlined above were added to the software as a ‘gene list’,

550 and the background was considered as all proteins previously measured in hippocampal brain

551 slices (Bowling et al., 2016). To depict the function clustering of the data described by DAVID,

552 the STRING database (version 11.0) was used. This online resource contains both known and

553 predicted protein-protein interactions (Szklarczyk et al., 2019). Clusters of proteins of interest

554 were highlighted for visualization, and FDR of GO term was noted.

555 For hierarchical clustering of datasets, Cytoscape (Version 3.71) was used. Datasets

556 were uploaded to the program as lists of proteins, and STRING enrichment was conducted

557 against the mouse genome. Fold change (average or individual APP/PS1:wild type samples, as

558 appropriate) were uploaded against the relevant network. Hierarchical clusters were generated

559 using the clusterMaker plugin, with pairwise average linkage and Euclidean distance metric.

560

561 Western Blot Validation

562 All western blotting was carried out as previously described (Sharma et al., 2010). Briefly,

563 hippocampal tissue reserved from samples prior to mass spectrometry processing was

564 homogenized in lysis buffer protease and phosphatase inhibitors. Protein concentration was

23 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

565 measured using BCA assay (Pierce). Aliquots of protein (20-40 μg) were separated using Bolt

566 Bis-Tris gels (4-12%; Thermo Scientific) and transferred to a nitrocellulose membrane which was

567 probed using appropriate primary and secondary antibody pairs. Membranes were washed and

568 probed for total protein using MemCode Reversible Protein Stain (Thermo Scientific), before

569 antibody signal was detected using chemiluminescence (GE Healthcare). Band density values

570 were normalized to total protein as detected using MemCode. Mean band densities for samples

571 from APP/PS1 mouse samples were normalized to corresponding samples from wild type

572 animals.

573

574 Antibodies

575 The following antibodies were used in this study: rabbit anti-APP monoclonal antibody (1:1000;

576 ab32136, Abcam), mouse anti-cathepsin-d antibody (1:500; ab6313, Abcam), rabbit anti-GAP-

577 43 polyclonal antibody (1:500; ab5220, Sigma), anti-Rpl13 antibody (1:1000; PA5-41715,

578 Thermofisher), and anti-Rpl18 antibody (1:5000; ab241988, Abcam). Secondary antibodies

579 were either goat anti-mouse IgG HRP or goat anti-mouse IgG HRP (Promega; 1:10,000)

580 respectively.

581

582 Acknowledgments

583 The authors acknowledge Dr. Francesco Longo for his assistance with hippocampal slice

584 preparation and Dr. Heather Bowling for her contributions to the experimental design. This work

585 was supported by NIH grants S10OD023659 and NS050276 to TN, and

586 NS047384 and NS034007 to EK. We thank the NYU Mass Spectrometry Core for Neuroscience

587 at the NYU School of Medicine for computing resources, bioinformatics support, and assistance

588 with data analysis and interpretation.

589

24 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

590 Competing interests

591 The authors declare no competing interests.

592

593 References

594 Alonso-Nanclares, L., Merino-Serrais, P., Gonzalez, S., and DeFelipe, J. (2013). Synaptic changes in the 595 dentate gyrus of APP/PS1 transgenic mice revealed by electron microscopy. J Neuropathol Exp Neurol 596 72, 386-395. 597 598 Alzheimer, A. (1907). Uber eine eigenartige Erkrankung der Hirnrinde. Zentralbl Nervenh Psych 18, 177- 599 179. 600 601 An, W.-L., Cowburn, R.F., Li, L., Braak, H., Alafuzoff, I., Iqbal, K., Iqbal, I.-G., Winblad, B., and Pei, J.-J. 602 (2003). Up-regulation of phosphorylated/activated p70 S6 kinase and its relationship to neurofibrillary 603 pathology in Alzheimer's disease. The American journal of pathology 163, 591-607. 604 605 Baloyannis, S.J. (2006). Mitochondrial alterations in Alzheimer's disease. Journal of Alzheimer's disease 606 9, 119-126. 607 608 Beckelman, B.C., Yang, W., Kasica, N.P., Zimmermann, H.R., Zhou, X., Keene, C.D., Ryazanov, A.G., and 609 Ma, T. (2019). Genetic reduction of eEF2 kinase alleviates pathophysiology in Alzheimer's disease model 610 mice. J Clin Invest 129, 820-833. 611 612 Bo, H., Kang, W., Jiang, N., Wang, X., Zhang, Y., and Ji, L.L. (2014). Exercise-induced neuroprotection of 613 hippocampus in APP/PS1 transgenic mice via upregulation of mitochondrial 8-oxoguanine DNA 614 glycosylase. Oxidative medicine and cellular longevity 2014. 615 616 Bogdanovic, N., Davidsson, P., Volkmann, I., Winblad, B., and Blennow, K. (2000). Growth-associated 617 protein GAP-43 in the frontal cortex and in the hippocampus in Alzheimer's disease: an 618 immunohistochemical and quantitative study. Journal of neural transmission 107, 463-478. 619 620 Bowling, H., Bhattacharya, A., Zhang, G., Alam, D., Lebowitz, J.Z., Bohm-Levine, N., Lin, D., Singha, P., 621 Mamcarz, M., Puckett, R., et al. (2019). Altered steady state and activity-dependent de novo protein 622 expression in fragile X syndrome. Nature Communications 10, 1710. 623 624 Bowling, H., Bhattacharya, A., Zhang, G., Lebowitz, J.Z., Alam, D., Smith, P.T., Kirshenbaum, K., Neubert, 625 T.A., Vogel, C., and Chao, M.V. (2016). BONLAC: A combinatorial proteomic technique to measure 626 stimulus-induced translational profiles in brain slices. Neuropharmacology 100, 76-89. 627 628 Buffington, S.A., Huang, W., and Costa-Mattioli, M. (2014). Translational control in synaptic plasticity and 629 cognitive dysfunction. Annual review of neuroscience 37, 17-38. 630

25 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

631 Cataldo, A.M., Barnett, J.L., Berman, S.A., Li, J., Quarless, S., Bursztajn, S., Lippa, C., and Nixon, R.A. 632 (1995). Gene expression and cellular content of cathepsin D in Alzheimer's disease brain: evidence for 633 early up-regulation of the endosomal-lysosomal system. Neuron 14, 671-680. 634 635 Chang, R.C., Wong, A.K., Ng, H.-K., and Hugon, J. (2002). Phosphorylation of eukaryotic initiation factor- 636 2α (eIF2α) is associated with neuronal degeneration in Alzheimer's disease. Neuroreport 13, 2429-2432. 637 638 Chang, W.-L., Lee, D.-C., Leu, S., Huang, Y.-M., Lu, M.-C., and Ouyang, P. (2003). Molecular 639 characterization of a novel nucleolar protein, pNO40. Biochemical and Biophysical Research 640 Communications 307, 569-577. 641 642 Chen, L., Xu, S., Wu, T., Shao, Y., Luo, L., Zhou, L., Ou, S., Tang, H., Huang, W., Guo, K., et al. (2019). Studies 643 on APP metabolism related to age-associated mitochondrial dysfunction in APP/PS1 transgenic mice. 644 Aging (Albany NY) 11, 10242-10251. 645 646 Chévere-Torres, I., Kaphzan, H., Bhattacharya, A., Kang, A., Maki, J.M., Gambello, M.J., Arbiser, J.L., 647 Santini, E., and Klann, E. (2012). Metabotropic glutamate receptor-dependent long-term depression is 648 impaired due to elevated ERK signaling in the ΔRG mouse model of tuberous sclerosis complex. 649 Neurobiology of disease 45, 1101-1110. 650 651 Ciechanover, A., and Kwon, Y.T. (2015). Degradation of misfolded proteins in neurodegenerative 652 diseases: therapeutic targets and strategies. Experimental & Molecular Medicine 47, e147-e147. 653 654 Costa-Mattioli, M., Sossin, W.S., Klann, E., and Sonenberg, N. (2009). Translational control of long-lasting 655 synaptic plasticity and memory. Neuron 61, 10-26. 656 657 Cox, J., and Mann, M. (2008). MaxQuant enables high peptide identification rates, individualized ppb- 658 range mass accuracies and proteome-wide protein quantification. Nature biotechnology 26, 1367-1372. 659 660 David, D.C., Hauptmann, S., Scherping, I., Schuessel, K., Keil, U., Rizzu, P., Ravid, R., Dröse, S., Brandt, U., 661 and Müller, W.E. (2005). Proteomic and functional analyses reveal a mitochondrial dysfunction in P301L 662 tau transgenic mice. Journal of Biological Chemistry 280, 23802-23814. 663 664 de la Monte, S.M., Ng, S.C., and Hsu, D.W. (1995). Aberrant GAP-43 gene expression in Alzheimer's 665 disease. The American journal of pathology 147, 934-946. 666 667 Dieterich, D.C., Link, A.J., Graumann, J., Tirrell, D.A., and Schuman, E.M. (2006). Selective identification 668 of newly synthesized proteins in mammalian cells using bioorthogonal noncanonical amino acid tagging 669 (BONCAT). Proceedings of the National Academy of Sciences 103, 9482. 670 671 Ding, Q., Markesbery, W.R., Chen, Q., Li, F., and Keller, J.N. (2005). Ribosome dysfunction is an early 672 event in Alzheimer's disease. Journal of Neuroscience 25, 9171-9175. 673 674 Du, H., Guo, L., and Yan, S.S. (2012). Synaptic mitochondrial pathology in Alzheimer's disease. 675 Antioxidants & redox signaling 16, 1467-1475.

26 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

676 Eichelbaum, K., Winter, M., Berriel Diaz, M., Herzig, S., and Krijgsveld, J. (2012). Selective enrichment of 677 newly synthesized proteins for quantitative secretome analysis. Nat Biotechnol 30, 984-990. 678 679 Evans, H.T., Benetatos, J., van Roijen, M., Bodea, L.G., and Götz, J. (2019). Decreased synthesis of 680 ribosomal proteins in tauopathy revealed by non-canonical amino acid labelling. The EMBO journal 38, 681 e101174. 682 683 Ferrer, I. (2002). Differential expression of phosphorylated translation initiation factor 2 alpha in 684 Alzheimer's disease and Creutzfeldt–Jakob's disease. Neuropathology and applied neurobiology 28, 441- 685 451. 686 687 Gengler, S., Hamilton, A., and Hölscher, C. (2010). Synaptic plasticity in the hippocampus of a APP/PS1 688 mouse model of Alzheimer's disease is impaired in old but not young mice. PLoS One 5, e9764-e9764. 689 690 Gruart, A., López-Ramos, J.C., Muñoz, M.D., and Delgado-García, J.M. (2008). Aged wild-type and APP, 691 PS1, and APP + PS1 mice present similar deficits in associative learning and synaptic plasticity 692 independent of amyloid load. Neurobiol Dis 30, 439-450. 693 694 Habib, L.K., Lee, M.T.C., and Yang, J. (2010). Inhibitors of catalase-amyloid interactions protect cells from 695 beta-amyloid-induced oxidative stress and toxicity. J Biol Chem 285, 38933-38943. 696 697 Hamazaki, H. (1996). Cathepsin D is involved in the clearance of Alzheimer's β-amyloid protein. FEBS 698 letters 396, 139-142. 699 700 Hauptmann, S., Scherping, I., Dröse, S., Brandt, U., Schulz, K., Jendrach, M., Leuner, K., Eckert, A., and 701 Müller, W. (2009). Mitochondrial dysfunction: an early event in Alzheimer pathology accumulates with 702 age in AD transgenic mice. Neurobiology of aging 30, 1574-1586. 703 704 Hernández-Ortega, K., Garcia-Esparcia, P., Gil, L., Lucas, J.J., and Ferrer, I. (2016). Altered machinery of 705 protein synthesis in Alzheimer's: from the nucleolus to the ribosome. brain pathology 26, 593-605. 706 707 Hirai, K., Aliev, G., Nunomura, A., Fujioka, H., Russell, R.L., Atwood, C.S., Johnson, A.B., Kress, Y., Vinters, 708 H.V., Tabaton, M., et al. (2001). Mitochondrial abnormalities in Alzheimer's disease. The Journal of 709 neuroscience : the official journal of the Society for Neuroscience 21, 3017-3023. 710 711 Holt, C.E., Martin, K.C., and Schuman, E.M. (2019). Local translation in neurons: visualization and 712 function. Nature Structural & Molecular Biology 26, 557-566. 713 714 Kaphzan, H., Buffington, S.A., Ramaraj, A.B., Lingrel, J.B., Rasband, M.N., Santini, E., and Klann, E. (2013). 715 Genetic reduction of the α1 subunit of Na/K-ATPase corrects multiple hippocampal phenotypes in 716 Angelman syndrome. Cell reports 4, 405-412. 717 718 Kaushik, S., and Cuervo, A.M. (2015). Proteostasis and aging. Nature Medicine 21, 1406-1415. 719 Keller, J.N., Hanni, K.B., and Markesbery, W.R. (2000). Impaired proteasome function in Alzheimer's 720 disease. Journal of neurochemistry 75, 436-439.

27 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

721 Kong, W., Mou, X., Liu, Q., Chen, Z., Vanderburg, C.R., Rogers, J.T., and Huang, X. (2009). Independent 722 component analysis of Alzheimer's DNA microarray gene expression data. Molecular Neurodegeneration 723 4, 5. 724 725 Landau, S.M., and Frosch, M.P. (2014). Tracking the earliest pathologic changes in Alzheimer disease 726 (AAN Enterprises). 727 728 Langstrom, N., Anderson, J., Lindroos, H., Winbland, B., and Wallace, W. (1989). Alzheimer's disease- 729 associated reduction of polysomal mRNA translation. Molecular Brain Research 5, 259-269. 730 731 Li, X., An, W.-L., Alafuzoff, I., Soininen, H., Winblad, B., and Pei, J.-J. (2004). Phosphorylated eukaryotic 732 translation factor 4E is elevated in Alzheimer brain. Neuroreport 15, 2237-2240. 733 734 Lourenco, M.V., Clarke, J.R., Frozza, R.L., Bomfim, T.R., Forny-Germano, L., Batista, A.F., Sathler, L.B., 735 Brito-Moreira, J., Amaral, O.B., and Silva, C.A. (2013). TNF-α mediates PKR-dependent memory 736 impairment and brain IRS-1 inhibition induced by Alzheimer’s β-amyloid oligomers in mice and monkeys. 737 Cell metabolism 18, 831-843. 738 739 Ma, T., Trinh, M.A., Wexler, A.J., Bourbon, C., Gatti, E., Pierre, P., Cavener, D.R., and Klann, E. (2013). 740 Suppression of eIF2α kinases alleviates Alzheimer's disease–related plasticity and memory deficits. 741 Nature neuroscience 16, 1299. 742 743 Ma, Y., McClatchy, D.B., Martínez-Bartolomé, S., Bamberger, C., and Yates, J.R. (2020). A Temporal 744 Quantitative Profiling of Newly Synthesized Proteins during Aβ Accumulation. bioRxiv, 745 2020.2001.2014.906560. 746 747 Markesbery, W.R. (2010). Neuropathologic alterations in mild cognitive impairment: a review. Journal of 748 Alzheimer's Disease 19, 221-228. 749 750 Martínez-Ballesteros, M., García-Heredia, J.M., Nepomuceno-Chamorro, I.A., and Riquelme-Santos, J.C. 751 (2017). Machine learning techniques to discover genes with potential prognosis role in Alzheimer’s 752 disease using different biological sources. Information Fusion 36, 114-129. 753 754 Martins, R.N., Villemagne, V., Sohrabi, H.R., Chatterjee, P., Shah, T.M., Verdile, G., Fraser, P., Taddei, K., 755 Gupta, V.B., and Rainey-Smith, S.R. (2018). Alzheimer’s disease: a journey from amyloid peptides and 756 oxidative stress, to biomarker technologies and disease prevention strategies—gains from AIBL and DIAN 757 cohort studies. Journal of Alzheimer's Disease 62, 965-992. 758 759 Mawuenyega, K.G., Sigurdson, W., Ovod, V., Munsell, L., Kasten, T., Morris, J.C., Yarasheski, K.E., and 760 Bateman, R.J. (2010). Decreased clearance of CNS β-amyloid in Alzheimer’s disease. Science 330, 1774- 761 1774. 762 763 McIntosh, K.B., Bhattacharya, A., Willis, I.M., and Warner, J.R. (2011). Eukaryotic cells producing 764 ribosomes deficient in Rpl1 are hypersensitive to defects in the ubiquitin-proteasome system. PLoS One 765 6, e23579.

28 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

766 Minkeviciene, R., Ihalainen, J., Malm, T., Matilainen, O., Keksa-Goldsteine, V., Goldsteins, G., Iivonen, H., 767 Leguit, N., Glennon, J., Koistinaho, J., et al. (2008). Age-related decrease in stimulated glutamate release 768 and vesicular glutamate transporters in APP/PS1 transgenic and wild-type mice. J Neurochem 105, 584- 769 594. 770 771 Moreira, P.I., Santos, M.S., and Oliveira, C.R. (2007). Alzheimer's disease: a lesson from mitochondrial 772 dysfunction. Antioxidants & redox signaling 9, 1621-1630. 773 774 Nixon, R.A. (2017). Amyloid precursor protein and endosomal-lysosomal dysfunction in Alzheimer's 775 disease: inseparable partners in a multifactorial disease. The FASEB Journal 31, 2729-2743. 776 777 Nunomura, A., Perry, G., Aliev, G., Hirai, K., Takeda, A., Balraj, E.K., Jones, P.K., Ghanbari, H., Wataya, T., 778 and Shimohama, S. (2001). Oxidative damage is the earliest event in Alzheimer disease. Journal of 779 Neuropathology & Experimental Neurology 60, 759-767. 780 781 Ong, S.E., Blagoev, B., Kratchmarova, I., Kristensen, D.B., Steen, H., Pandey, A., and Mann, M. (2002). 782 Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to 783 expression proteomics. Mol Cell Proteomics 1, 376-386. 784 785 Park, J.H., Widi, G.A., Gimbel, D.A., Harel, N.Y., Lee, D.H.S., and Strittmatter, S.M. (2006). Subcutaneous 786 Nogo Receptor Removes Brain Amyloid-β and Improves Spatial Memory in Alzheimer's Transgenic Mice. 787 The Journal of Neuroscience 26, 13279-13286. 788 789 Poirier, Y., Grimm, A., Schmitt, K., and Eckert, A. (2019). Link between the unfolded protein response and 790 dysregulation of mitochondrial bioenergetics in Alzheimer’s disease. Cellular and Molecular Life Sciences 791 76, 1419-1431. 792 793 Radde, R., Bolmont, T., Kaeser, S.A., Coomaraswamy, J., Lindau, D., Stoltze, L., Calhoun, M.E., Jäggi, F., 794 Wolburg, H., Gengler, S., et al. (2006). Abeta42-driven cerebral amyloidosis in transgenic mice reveals 795 early and robust pathology. EMBO Rep 7, 940-946. 796 797 Rappsilber, J., Mann, M., and Ishihama, Y. (2007). Protocol for micro-purification, enrichment, pre- 798 fractionation and storage of peptides for proteomics using StageTips. Nature protocols 2, 1896. 799 800 Richter, J.D., and Klann, E. (2009). Making synaptic plasticity and memory last: mechanisms of 801 translational regulation. Genes & development 23, 1-11. 802 803 Sajdel-Sulkowska, E.M., and Marotta, C.A. (1984). Alzheimer's disease brain: alterations in RNA levels 804 and in a ribonuclease-inhibitor complex. Science 225, 947-949. 805 806 Santini, E., Turner, K.L., Ramaraj, A.B., Murphy, M.P., Klann, E., and Kaphzan, H. (2015). Mitochondrial 807 superoxide contributes to hippocampal synaptic dysfunction and memory deficits in Angelman 808 syndrome model mice. Journal of Neuroscience 35, 16213-16220. 809

29 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

810 Scahill, R.I., Schott, J.M., Stevens, J.M., Rossor, M.N., and Fox, N.C. (2002). Mapping the evolution of 811 regional atrophy in Alzheimer's disease: unbiased analysis of fluid-registered serial MRI. Proc Natl Acad 812 Sci U S A 99, 4703-4707. 813 814 Serneels, L., Van Biervliet, J., Craessaerts, K., Dejaegere, T., Horré, K., Van Houtvin, T., Esselmann, H., 815 Paul, S., Schäfer, M.K., Berezovska, O., et al. (2009). gamma-Secretase heterogeneity in the Aph1 816 subunit: relevance for Alzheimer's disease. Science (New York, NY) 324, 639-642. 817 818 Sharma, A., Hoeffer, C.A., Takayasu, Y., Miyawaki, T., McBride, S.M., Klann, E., and Zukin, R.S. (2010). 819 Dysregulation of mTOR signaling in fragile X syndrome. Journal of Neuroscience 30, 694-702. 820 821 Sherman, B.T., and Lempicki, R.A. (2009). Systematic and integrative analysis of large gene lists using 822 DAVID bioinformatics resources. Nature protocols 4, 44-57. 823 824 Shi, Z., Fujii, K., Kovary, K.M., Genuth, N.R., Röst, H.L., Teruel, M.N., and Barna, M. (2017). Heterogeneous 825 ribosomes preferentially translate distinct subpools of mRNAs genome-wide. Molecular cell 67, 71-83. 826 e77. 827 828 Shigeoka, T., Koppers, M., Wong, H.H.-W., Lin, J.Q., Cagnetta, R., Dwivedy, A., de Freitas Nascimento, J., 829 van Tartwijk, F.W., Ströhl, F., Cioni, J.-M., et al. (2019). On-Site Ribosome Remodeling by Locally 830 Synthesized Ribosomal Proteins in Axons. Cell Reports 29, 3605-3619.e3610. 831 832 Singh, A.K., and Pati, U. (2015). CHIP stabilizes amyloid precursor protein via proteasomal degradation 833 and p53-mediated trans-repression of β-secretase. Aging Cell 14, 595-604. 834 835 Slavov, N., Semrau, S., Airoldi, E., Budnik, B., and van Oudenaarden, A. (2015). Differential Stoichiometry 836 among Core Ribosomal Proteins. Cell Reports 13, 865-873. 837 838 Smith, D.L., Pozueta, J., Gong, B., Arancio, O., and Shelanski, M. (2009). Reversal of long-term dendritic 839 spine alterations in Alzheimer disease models. Proceedings of the National Academy of Sciences 106, 840 16877. 841 842 Swerdlow, R.H. (2018). Mitochondria and Mitochondrial Cascades in Alzheimer's Disease. J Alzheimers 843 Dis 62, 1403-1416. 844 845 Szklarczyk, D., Gable, A.L., Lyon, D., Junge, A., Wyder, S., Huerta-Cepas, J., Simonovic, M., Doncheva, N.T., 846 Morris, J.H., Bork, P., et al. (2019). STRING v11: protein-protein association networks with increased 847 coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res 47, 848 D607-d613. 849 850 Tomljanovic, Z., Patel, M., Shin, W., Califano, A., and Teich, A.F. (2018). ZCCHC17 is a master regulator of 851 synaptic gene expression in Alzheimer's disease. Bioinformatics 34, 367-371. 852 853 Trinchese, F., Liu, S., Battaglia, F., Walter, S., Mathews, P.M., and Arancio, O. (2004). Progressive age- 854 related development of Alzheimer-like pathology in APP/PS1 mice. Annals of Neurology: Official Journal 855 of the American Neurological Association and the Child Neurology Society 55, 801-814.

30 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

856 Volianskis, A., Køstner, R., Mølgaard, M., Hass, S., and Jensen, M.S. (2010). Episodic memory deficits are 857 not related to altered glutamatergic synaptic transmission and plasticity in the CA1 hippocampus of the 858 APPswe/PS1δE9-deleted transgenic mice model of ß-amyloidosis. Neurobiol Aging 31, 1173-1187. 859 860 Ximerakis, M., Lipnick, S.L., Innes, B.T., Simmons, S.K., Adiconis, X., Dionne, D., Mayweather, B.A., 861 Nguyen, L., Niziolek, Z., and Ozek, C. (2019). Single-cell transcriptomic profiling of the aging mouse brain. 862 Nature neuroscience 22, 1696-1708. 863 864 Xu, Z.-Q., Zhang, L.-Q., Wang, Q., Marshall, C., Xiao, N., Gao, J.-Y., Wu, T., Ding, J., Hu, G., and Xiao, M. 865 (2013). Aerobic exercise combined with antioxidative treatment does not counteract moderate- or mid- 866 stage Alzheimer-like pathophysiology of APP/PS1 mice. CNS Neurosci Ther 19, 795-803. 867 868 Yoshihama, M., Uechi, T., Asakawa, S., Kawasaki, K., Kato, S., Higa, S., Maeda, N., Minoshima, S., Tanaka, 869 T., Shimizu, N., et al. (2002). The human ribosomal protein genes: sequencing and comparative analysis 870 of 73 genes. Genome Res 12, 379-390. 871 872 Zecha, J., Satpathy, S., Kanashova, T., Avanessian, S.C., Kane, M.H., Clauser, K.R., Mertins, P., Carr, S.A., 873 and Kuster, B. (2019). TMT labeling for the masses: a robust and cost-efficient, in-solution labeling 874 approach. Molecular & Cellular Proteomics 18, 1468-1478. 875 876 Zhang, G., Bowling, H., Hom, N., Kirshenbaum, K., Klann, E., Chao, M.V., and Neubert, T.A. (2014). In- 877 depth quantitative proteomic analysis of de novo protein synthesis induced by brain-derived 878 neurotrophic factor. Journal of proteome research 13, 5707-5714. 879 880 Zhang, H., Gong, B., Liu, S., Fa, M., Ninan, I., Staniszewski, A., and Arancio, O. (2005). Synaptic fatigue is 881 more pronounced in the APP/PS1 transgenic mouse model of Alzheimer's disease. Current Alzheimer 882 Research 2, 137-140. 883 884

31 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

885 Figures 886

887 888 Figure 1. BONLAC-mediated labelling of de novo protein synthesis in the hippocampus of young and aged 889 wild-type and APP/PS1 mice. (A) Representative immunofluorescent images showing the progressive deposition 890 of amyloid beta plaques in the hippocampus of symptomatic 12 month-old APP/PS1 mice, but not asymptomatic 4 891 month-old APP/PS1 or 12 month-old wild-type (WT) mice (Blue = DAPI, Green = Aβ). (B) Quantification of the 892 number of Aβ plaques in the hippocampus of asymptomatic APP/PS1 mice, 12 month-old WT and symptomatic 12 893 month-old APP/PS1 mice. (C) Quantification of the average area of the Aβ plaques observed in these groups (n = 894 1-2 mice per group for (B) and (C)). (D) De novo synthesized protein in 3-5 month-old asymptomatic vs. 12+ month- 895 old symptomatic APP/PS1 mouse hippocampal slices as detected via AHA labelling, followed by biotin-alkyne click 896 reaction and western blot (normalized to total protein (as determined via MemCode staining), expressed relative to 897 average asymptomatic biotin signal; n = 5 mice/condition; statistical significance calculated using unpaired two 898 tailed t tests: p = 0.0005). (E) Schematic showing BONLAC workflow for acute hippocampal slices from WT and 899 APP/PS1 mice. Following labelling, slices from one APP/PS1 and one age-matched WT mouse are pooled for 900 subsequent processing. 901

32 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

A Protein detected in majority of samples B 35

Candidate 33

All proteins detected; Down: 107 (5.9%) 31 Up: 73 (4.0%)

29 proteins Intensity

(C-sc

ored)

Unchanged: 1646 (90.1%) 21 - 1.5 - 1 - 0.5 0 0.5 1 1.5 Fold Change (APP-PS1 / WT) C D Proteasome (FDR 2.7E-4) Psma2 Psmb1 Psmb6 Psma7 Psmb2 Psmb7

Ribosome (FDR 8.6E-4) Rpl3 Rpl34 Rps26 Rpl19 Rps17 Rps30 Rpl31 Rps23

Alzheimer’s disease (FDR 4.8E-1) APP Ndufs7 Ndufv3 Ndufb3 Ndufs6 Uqcrq

Proteasome Ribosome Synapse 902 903 Figure 2. Steady state proteome differs in wild-type and asymptomatic APP/PS1 mouse hippocampus. (A) 904 Fold change vs. intensity distribution plot of all newly made proteins detected in wild-type (WT) vs. 3-5 month-old 905 APP/PS1 hippocampus (light grey), with proteins that were detected in majority of samples (>3 out of 5 samples; 906 grey) and dysregulated candidate proteins identified by C-score screen (upregulated ≥ 20% in red, downregulated 907 ≤ 20% in blue). (B) Doughnut plot showing the proportion of detected proteins that were downregulated (fold change 908 < 0.8, blue), upregulated (fold change > 1.2, red) or unchanged in the majority of asymptomatic APP/PS1 mice 909 compared to WT littermates. (C) String diagram depicting enriched gene ontology groups. (D) Top functional 910 clusters in 3-5 month-old APP/PS1 mice compared to WT using DAVID. Red = upregulated proteins. Blue = 911 downregulated proteins. 912 913 914

33 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

A B 33 Protein detected in majority of samples 31 Candidate Down: 23 (2.7%)

29 All proteins detected; Up: 90 (10.6%)

proteins 25

23 Intensity

(C-sc 21

19 ored) Unchanged: 742 (86.8%)

15 -1.5 -1 -0.5 0 0.5 1 1.5 Log Fold Change (APP-PS1 / WT) C D Alzheimer’s disease (FDR 1.7E-4) Apoe Atp5d Grin1 Ndufv1 APP Atp5o Ndufs7 Uqcrc1 Alzheimer’s disease

Ribosome (FDR 9.14E-9) Ribosome Hba-1 Rpl27 Rpl37 Zcchc17 Rpl18 Rpl29 Rpl8 Lysosome Rpl23a Rpl36a Rps4x Mitochondria Lysosome (FDR 4.04E-6) Anxa2 Cat Ctsd GFAP Apoe Chga Ctss Sqstm1 Atp6v0d1 Ctsb Gaa Vps35

915 916 Figure 3. Analysis of altered de novo proteome highlights impaired proteostasis in symptomatic APP/PS1 917 mice. (A) Fold change vs. intensity distribution plot showing all proteins detected in 12+ month-old APP/PS1 mouse 918 hippocampal slices vs. wild-type (WT) littermates using BONLAC (light grey). Dark grey overlay depicts proteins 919 that were consistently detected in the majority of samples (>4/7), and candidate proteins identified by C-score 920 screen as upregulated (fold change ≥ 1.2) shown in red, while downregulated proteins (≤ 0.8) are labelled blue. (B) 921 Doughnut plot indicating the majority of proteins detected in symptomatic mice are not altered in APP/PS1 mice 922 (742; 86.8% of proteins detected). 23 proteins are downregulated compared to WT mice (2.7% of proteins show a 923 fold change >0.8, blue) and 90 proteins are upregulated in APP/PS1 mice (10.6% of proteins show a fold change 924 <1.2, red). (C) String diagram showing enriched gene ontology networks. (D) DAVID-identified functional clusters 925 in 12+ month-old APP/PS1 mice. Red = upregulated proteins. Blue = downregulated proteins. 926

34 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

927 928 Figure 4. Candidate proteins identified with BONLAC screen of de novo proteome exhibit altered expression 929 levels in APP/PS1 mice. (A) Schematic of workflow for slice allocation for mass spectrometry and western blot 930 protein validation. (B) Western blot quantitation of candidate proteins selected from BONLAC screen in 931 asymptomatic (white bars) and symptomatic (teal bars) mice, normalized to total protein (as assessed by 932 MemCode) and expressed relative to wild-type (WT; n = 5-8 mice); representative western blots showing selected 933 protein levels in WT and APP/PS1 mouse hippocampal lysates (asymptomatic: black box; symptomatic: teal box). 934 (C) Comparison table showing average fold change in de novo synthesis of candidate proteins in asymptomatic 935 and symptomatic APP/PS1 vs WT littermates as identified by BONLAC (asymptomatic n=5; symptomatic n=7) 936 compared to change in total expression quantified by western blot. Vibrant color indicates value reached cutoff (+/- 937 20%; >0.8 or < 1.2); pale color indicates near threshold (within 5%); white indicates no change. Statistical 938 significance calculated using unpaired two tailed t tests; * = p ≤ 0.05, *** p ≤ 0.001, **** p ≤ 0.0001. (D) Linear 939 regression showing correlation between average de novo synthesized protein fold-change (as determined by 940 BONLAC) vs. average total protein fold-change (as determined via western blot) of candidate proteins in the 941 APP/PS1 vs. WT hippocampus before (white circles) and after (teal circles) pathology development (r 2 value = 942 0.8376, figure shows mean +/- SEM (vertical error bars: BONLAC; horizontal error bars: western blot); p = 0.0002). 943

35 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Pathology development

Asymptomatic Symptomatic Cluster 1

Cluster 2

Cluster 3

Cluster 4

Cluster 5

Cluster 6

-0.7 0 0.7 944 Log average fold change 945 Figure 5. The biological pathways affected by dysregulated hippocampal protein synthesis vary throughout 946 pathological progression in APP/PS1 mice. Hierarchical clustering heatmap reveals similarities and divergence 947 in the mean-normalized log protein fold change of the 791 proteins detected in both asymptomatic (~4 months-old) 948 and symptomatic (12+ months-old) APP/PS1 mice relative to wild-type (WT) littermates. GO analysis evidenced 6 949 major clusters being significantly modified, which are: 1) protein processing in the endoplasmic reticulum (FDR 950 3.2E-04); 2) synapse (FDR 5.6E-07); 3) synaptic vesicle cycle (FDR 2.77E-06), membrane trafficking (FDR 6.24E- 951 11) and synapse (FDR 4.82E-29); 4) axo-dendritic transport (FDR 4.86E-08) and glycolysis/gluconeogenesis 952 (2.56E-05); 5) myelin sheath (5.17E-18) and Alzheimer’s disease (8.16E-07); and 6) mitochondrial part (FDR 2.54E- 953 08) and electron transfer activity (FDR 4.02E-05). FDR generated by StringDB algorithms. Red = upregulated 954 proteins. Blue = downregulated proteins. White = unchanged protein expression.

36 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

955 956 Figure 6. Steady state proteome is predominantly distinct throughout the development of symptoms. (A) 957 Minimal overlap is observed between proteins that are dysregulated (as detected by C-score rank algorithm) in the 958 hippocampi of 3-5 month-old asymptomatic vs. 12+ month-old APP/PS1 mice. (B) Hierarchical clustering of log fold 959 changes reveals the majority of proteins which are dysregulated in both asymptomatic and symptomatic APP/PS1 960 mice compared to WT littermates do not show the same trend. Red = upregulated proteins. Blue = downregulated 961 proteins. White = unchanged protein expression. (C) String diagram revealing an enrichment of proteins in the GO 962 network ‘Cytosolic ribosome’ in APP/PS1 throughout the disease process. White arrows = direction of regulation in 963 asymptomatic mice. Teal arrows = direction of regulation in symptomatic mice. 964 965

37 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

966 Supplementary Figures 967

-1 0 1 968 Average log fold change 969 Supplementary Figure 1. All proteins detected in the asymptomatic APP/PS1 hippocampus as detected by 970 BONLAC. Hierarchical clustering-generated heatmap showing all protein fold-change ratios identified in at least 971 one sample from the BONLAC screen. Red indicates proteins who show higher levels of de novo synthesis in ~4 972 month-old APP/PS1 mice compared to wild-type (WT) littermates, while downregulated proteins are shown in blue. 973 White = no change. Black = MaxQuant ratio not calculated. 974 975

38 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Cluster 1

PPI enrichment p value: 4.54E-4 GO: proteasome core complex; FDR 8.16E-05

Cluster 2

PPI enrichment p value: 9.96E-4 GO: RNA binding (FDR 3.15E-05) GO: ribonucleoprotein complex (FDR 1.2E-04) 976 -1 0 1 977 Supplementary Figure 2. Dysregulated protein candidates in the asymptomatic APP/PS1 hippocampus as 978 detected by BONLAC. Hierarchical clustered heatmap showing all candidate proteins identified from the BONLAC 979 screen of ~4 months-old APP/PS1 vs. wild-type (WT) ranked by automated C-score. Protein candidates were 980 identified by R script as showing an average fold change across all samples of ± 20% (<0.8 or >1.2). Manual 981 validation confirmed candidates were detected in the majority of samples (>3/5), and that majority of samples 982 showed the same trend (>50% either <0.79 or >1.19). Nodes in blue indicate downregulation, while red indicates 983 increased de novo synthesis in the asymptomatic APP/PS1 mice compared to WT littermates. Protein clusters are 984 represented visually by String diagrams (right), with key pathways highlighted. White = no change. Black = 985 MaxQuant ratio not calculated. 986 987

39 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Cluster 1 PPI enrichment p value: <1.06E-16 GO: Citrate cycle (TCA cycle; FDR 2.11E-09) GO: Gap junction (FDR 7.02 E-09) GO: Mitochondrion (FDR 1.74 E-08)

Cluster 2

PPI enrichment p value: <1.06E-16 GO: Ribonucleoprotein complex ( FDR 7.48E-12) -1 0 1 GO: Synapse (FDR 7.93 E-09) 988 989 Supplementary Figure 3. All proteins detected in the symptomatic APP/PS1 hippocampus as detected by 990 BONLAC. Hierarchical clustering heatmap showing all protein fold-change ratios identified in the majority of 991 samples from the BONLAC screen. Red indicates proteins which show higher levels of de novo synthesis in 12+ 992 month-old APP/PS1 mice compared to WT littermates (APP is highlighted in top left), while downregulated proteins 993 are shown in blue. String diagram of selected cluster is shown on right. White = no change. Black = MaxQuant ratio 994 not calculated. 995

40 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

PPI enrichment p value: 4.8E-10 GO: lysosome (FDR 2.22E-07) GO: immune system (FDR 4.01E-08) GO: cytosolic ribosome (FDR 2.4E-04)

Cluster 1

996 -1 0 1 997 Supplementary Figure 4. Dysregulated protein candidates in the hippocampus of symptomatic APP/PS1 998 mice as detected by BONLAC. Hierarchical clustered heatmap of candidate proteins identified from the BONLAC 999 screen ranked by automated C-score. Protein candidates were identified by customized R script as showing an 1000 average fold change across all samples of ± 20% (<0.8 or >1.2). Manual validation confirmed candidates were 1001 detected in the majority of samples (>4/7), and that majority of samples showed the same trend (>50% either <0.79 1002 or >1.19). String diagram representing main clusters shown on right, with key pathways highlighted. Black cells 1003 indicate absence of ratio. Red = upregulated proteins. Blue = downregulated proteins. White = no change. Black = 1004 MaxQuant ratio not calculated. 1005

41 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1006 1007 Supplementary Figure 5. Visual depiction of proteins in Hierarchical Cluster 1 (from Figure 5). String diagram 1008 showing biological networks between the proteins identified in Cluster 1. Figure generated by StringDb; Red node 1009 = GO: Protein processing in the endoplasmic reticulum; FDR:3.2E-4.

42 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1010 1011 Supplementary Figure 6. Visual depiction of proteins in Hierarchical Cluster 2 (from Figure 5). String diagram 1012 showing biological networks between the proteins identified in Cluster 2. Figure generated by StringDb; Red node 1013 = GO: Synapse; FDR:5.6E-7. 1014

43 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1015 1016 Supplementary Figure 7. Visual depiction of proteins in Hierarchical Cluster 3 (from Figure 5). String diagram 1017 showing biological networks between the proteins identified in Cluster 3. Figure generated by StringDb; Red node 1018 = GO: synaptic vesicle cycle (FDR: 2.77E-6); Green node = GO: synapse (FDR: 4.82E-29); Blue node = GO: 1019 membrane trafficking (FDR: 6.24E-11). 1020 1021 1022

44 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1023 1024 Supplementary Figure 8. Visual depiction of proteins in Hierarchical Cluster 4 (from Figure 5). String diagram 1025 showing biological networks between the proteins identified in Cluster 4. Figure generated by StringDb; Red node 1026 = GO: Axo-dendritic transport (FDR: 4.86E-8); Blue = GO: Glycolysis/Gluconeogenesis (FDR: 2.56E-5). 1027

45 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1028

1029 1030 Supplementary Figure 9. Visual depiction of proteins in Hierarchical Cluster 5 (from Figure 5). String diagram 1031 showing biological networks between the proteins identified in Cluster 5. Figure generated by StringDb; Red node 1032 = GO: myelin sheath (FDR: 5.17E-18); blue node = GO: Alzheimer’s disease (FDR: 8.16E-07). 1033 1034 1035 1036

46 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1037 1038 Supplementary Figure 10. Visual depiction of proteins in Hierarchical Cluster 6 (from Figure 5). String 1039 diagram showing biological networks between the proteins identified in Cluster 6. Figure generated by StringDb; 1040 Blue node = GO: mitochondrial part (FDR: 4.22E-07); Red node = GO: electron transfer activity (FDR: 2.88E-05). 1041

47 bioRxiv preprint doi: https://doi.org/10.1101/2020.08.21.260786; this version posted August 22, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1042 Supplementary Table 1043

1044 FDR: 3.2E-4 GO: protein processing in ER 6.68E-05 p-value: enrichment PPI Adamts4 Slc25a12 1 Cluster Ndufv1 Ap2m1 Atp1b2 Uqcrc1 Col1a2 Vcam1 Dnajb1 Arglu1 Dpysl5 Cand1 Anxa2 Hsph1 Anxa5 Rbm3 Hyou1 Rpl29 Stub1 Sarnp Aldh2 Pdia6 Nfasc Apoe Srsf4 Pdhb B2m Rpl8 Ctsd Ctss Plek Cat Cytoscapeand StringDb. wild and APP/PS1 symptomatic in depicted SupplementaryTable1. FDR: 5.6E-7 GO: synapse 6.34E-09 p-value: enrichment PPI Cluster 2 Cluster Tax1bp1 Slc4a10 Arhgef7 Nckap1 Phactr1 Pgam1 Impact Akr1a1 Trim28 Fkbp1a Dzank1 Phyhip Usp14 Nudt3 Rph3a Soga3 Ewsr1 Gspt1 Agfg2 Amph Gsta4 Prdx5 Clip2 Ddx6 Gem Qdpr Mkl2 Snca Sncb Tpi1 Abr Ddt iue 5 Figure Ppp1r10 Maged1 Morf4l2 Map2k1 Zcchc18 Arpc1a Trim32 Uqcrc2 Sumo2 Camkv Irf2bpl Mef2d Dnajc6 Ube2n Pebp1 Hook3 Icam5 Pcmt1 Srcin1 Cstf2t Mdh1 Gmps Gad1 Park7 Calb1 Cplx1 Got1 Bcan Prepl Ldhb Psd3 Cbr1 FDR: 2.77E-6 Go: synaptic vesicle cycle 1.0e-16< p-value: enrichment PPI Cluster 3 Cluster Dync1h1 Camk2a Ctnnb1 Ctnna1 Csnk2b Tardbp Ctnna2 Dlgap1 Ppp1cb Nedd4 Lrpap1 Psmd2 and Ap2b1 Luc7l2 Rab6a Tpm3 Tceb1 Aplp1 Sept7 Sort1 Psip1 Spin1 Bag6 Stip1 Add2 Edil3 Scg2 Sfpq Cct3 Cnp Fos Sf1

Prote upeetr Fgrs 5 Figures Supplementary Hist1h4a Smarca4 C1qtnf4 Ppp1r2 Nap1l1 Ndufv2 Dnajb4 Hnrnpl G3bp2 Mat2a Clstn3 Epha4 Clstn1 Atxn2l Cyfip2 Apbb1 Snrpn Trim9 Cadps Nono Vsnl1 Gnl3l Idh3a Khsrp Hpca Egr1 Fis1 Tcf4 Cpe Jup Nsf Cs in identitiescontainedin biologicalwithinkey pathwaysasdetected byhierarchical the clustering - ye os hippoc mouse type Hnrnpul2 Tsc22d1 Mapre3 Camk2b Snap47 Pacsin1 Prkar1b Map1b Rbm42 Sptan1 Serbp1 Hnrnpk Sptbn1 Ppp2ca Psma1 Sh3gl2 Coro1c Ddx17 Hspa9 Arpc3 Cntn1 Dctn2 Csde1 Sept5 Tceb2 Prdx6 Nell2 Psap Vim Calu Cct4 Ogt FDR: 6.24E-11 GO: membrane trafficking Hsp90ab1 Map7d1 Chmp4b Tpd52l2 Rad23b Hp1bp3 Psmc5 Gnao1 Dnajc7 Anxa6 Hspa2 Rab5a Hspa4 Ap2a1 Pdha1 Dnm1 Capzb Lasp1 Sf3b1 Snw1 Actr2 Tollip Rpsa Add3 Crym Clip3 Ddx5 Nptn Clic4 Cct8 Flnb Eif5 Gatad2b Syngap1 Hnrnph1 Snap25 Ndufa4 Prkar1a Rps27a Map1a Sphkap Nedd4l Ywhab Cx3cl1 Spon1 Rps19 Ndrg3 Basp1 Matr3 Sugt1 Atp5b Dhrs1 Efhd2 Ctbp1 Cdkl5 Napb Cct6a Cap1 Dhx9 Dlg3 Dlg4 Cct2 Pld3 Son mu ( = 5 = (n ampus Gabarapl2 Atp6v1g2 - Snap91 Atp1b1 Atp2b1 Ndufs1 Eef1a1 Atp5a1 Zfand5 Dnaja2 Ywhah Sorbs2 Akap8l Psma4 Stmn2 Nme1 Eif4g1 Dcaf7 Gnai1 Gtf2b Elavl3 Dctn1 Rpl36 Stau1 Lmna Ptk2b Atp5j Napg 10. 10. Pfkm Fth1 Cct5 Cfl1 BONLAC FDR: 4.82E-29 GO: synapse Hnrnpa3 Timm13 Caprin1 Akap12 Psmc3 Tagln2 Myo5a Clstn2 Tspyl4 Usp9x Fubp1 Zmiz1 Sept7 Mdh2 Pspc1 Elavl4 Wdr1 Prkcg Txnl1 Gas7 Upf1 Maff Bin1 Gdi2 Eef2 Rtn1 Glg1 Ppia Vcp Cltc Hk1 Esd Atp6v1e1 Atp6v1c1 - Chmp2a Slc25a4 Tuba4a Ndufa8 Ociad1 Baiap2 Ppp3cb Hnrnpu Rbfox1 Hnrnpd Kcnab2 Ubqln2 Stmn3 Stxbp1 Nrcam Dpysl3 Eif4a1 Sap18 Cndp2 U2af2 Bcas1 Arpc2 mc/eoyegop. P ad D a dtce via detected as FDR and PPI mice/genotype/group). 7 Stx12 Chgb Aak1 Celf4 Cyc1 Mbp Cct7 - Ik eetd rtis bevd n oh smtmtc and asymptomatic both in observed proteins detected Hnrnpa1 Sparcl1 Psmc2 Taldo1 G3bp1 Ndrg1 Sept3 Puf60 Srsf3 Celf2 Ank2 FDR: 3.02E-5 GO: glycolysis and gluconeogenesis 1.0e-16< p-value: enrichment PPI Atp6v1b2 4 Cluster Atp6v1d Atp6v1a Camk2g Anp32a Atxn10 Dpysl2 Dnm1l Ahcyl1 Actr1b Cend1 Ap3b2 Ap2a2 Ckap5 Cdc37 Atxn2 Aldoa Acot7 Dmtn Cplx2 Ddx1 Cnn3 Cnbp Add1 Dlg2 Abi2 Aars Cttn Araf Bsn Ak5 Ckb Hspa12a Hnrnph2 Gprasp2 Gprasp1 Khdrbs3 Khdrbs1 Homer1 Gpm6a Hmgb1 Eef1a2 L1cam Hspa4l Gapdh Iqsec1 Eif4g3 Gmfb Glod4 Fscn1 Hint1 Gphn Gpd2 Fosl2 Kif5c Kif1a Eno2 Eno1 Eif4h Ehd1 Gdi1 Dstn Klc1 Hgs Map1lc3b Pafah1b2 Pafah1b1 Ppp2r1a Map7d2 Ppp3ca Ppme1 Ptges3 Mapk1 Rbbp7 Ndrg4 Mbnl2 Rnf14 Pcbp2 Pcbp1 Map6 Map2 Nudc Ncdn Prnp Pgk1 Pfn2 Pdxk Pak1 Mlf2 Ldha Ppid Pls3 Pkm Pfkp Ptn Mif Ube2d3 Ubqln1 Spock2 Snrpd3 Ywhae Tagln3 Sucla2 Wbp2 Uchl1 Tcea1 Rufy3 Uba1 Tufm Synj1 Usp5 Tppp Sub1 Rnh1 Tpt1 Tbcb Syn2 Syn1 Snx1 Sgta Sbds Sik3 Tnr FDR: 5.17e-18 GO: myelin sheath 1.0e-16< p-value: enrichment PPI Atp6v0d1 Adamts1 Slc25a22 Slc9a3r1 5 Cluster Hnrnpa0 Hmox1 Pabpc1 Psmc6 Psmc1 Pa2g4 Fyttd1 Timp2 Hspa8 Kpnb1 Ckmt1 Atp5o Uqcrb Cox6c Nptxr Sparc Kif5b Vapa Aco2 Sdha Grb2 Chp1 Ifrd1 Ctsb Fxr1 Ctsl Ncl Tkt Fam168a Aldh5a1 Timm10 Tubb4b Ndufa7 Syncrip Coro1a Dnaja1 Hpcal4 Adrm1 Mfge8 Rap1b Phgdh Asrgl1 Vps35 Olfm1 Acat1 Rps4x Ptbp2 Idh3g Grin1 Srsf7 Nrgn Zeb2 Rhob Igsf8 Rtn3 Junb Plp1 Plat Syp Ezr FDR:8.16e-07 GO: Alzheimer's disease Hsp90aa1 Sqstm1 Impdh2 Tubb4a Rpl23a Atp1a2 Ndufs8 Nufip2 Numbl Psmd4 Hspa5 Eif2s2 Npm1 Hspe1 Cpne6 Synpo Rpl27 Ncald Rplp2 Cox5b Aplp2 Prkce Srsf2 Naca Rtn4 Safb Sirt2 Syt1 Idh1 Glul Erc1 Oat Hnrnpa2b1 Hnrnpab Camk2d Dynlrb1 Rbm39 Slc3a2 Rps13 Lancl1 Tra2b Ccsap Clint1 Atp5h Gng2 Prdx1 Syt11 Stx1b Srsf1 H1f0 Oxr1 Gaa Yars Dsp Clta Gls Jun Sri