Mutant Presenilin 1 Dysregulates Exosomal Proteome Cargo Produced by Human-Induced Pluripotent Stem Cell Neurons.

The accumulation and propagation of hyperphosphorylated tau (p-Tau) is a neuropathological hallmark occurring with neurodegeneration of Alzheimer's disease (AD). Extracellular vesicles, exosomes, have been shown to initiate tau propagation in the brain. Notably, exosomes from human-induced pluripotent stem cell (iPSC) neurons expressing the AD familial A246E mutant form of presenilin 1 (mPS1) are capable of inducing tau deposits in the mouse brain after in vivo injection. To gain insights into the exosome proteome cargo that participates in propagating tau pathology, this study conducted proteomic analysis of exosomes produced by human iPSC neurons expressing A246E mPS1. Significantly, mPS1 altered the profile of exosome cargo proteins to result in (1) proteins present only in mPS1 exosomes and not in controls, (2) the absence of proteins in the mPS1 exosomes which were present only in controls, and (3) shared proteins which were upregulated or downregulated in the mPS1 exosomes compared to controls. These results show that mPS1 dysregulates the proteome cargo of exosomes to result in the acquisition of proteins involved in the extracellular matrix and protease functions, deletion of proteins involved in RNA and protein translation systems along with proteasome and related functions, combined with the upregulation and downregulation of shared proteins, including the upregulation of amyloid precursor protein. Notably, mPS1 neuron-derived exosomes displayed altered profiles of protein phosphatases and kinases involved in regulating the status of p-tau. The dysregulation of exosome cargo proteins by mPS1 may be associated with the ability of mPS1 neuron-derived exosomes to propagate tau pathology.

Introduction

Alzheimer’s disease (AD) is a neurodegenerative disorder resulting in deficits in cognitive function. AD brain neuropathology displays progressive neuronal loss and severe neurodegeneration, with the accumulation of amyloid plaque deposits and neurofibrillary tangles (NFTs). NFTs result from the accumulation of hyperphosphorylated tau (p-Tau)[1−5] which lacks the ability to interact with microtubules, leading to detrimental effects on neuronal synaptic functions. Tau oligomers impair long-term potentiation and result in memory loss.[6] Tau undergoes cell–cell propagation in the brain cortex and hippocampus,[7−10] which leads to memory deficits and synaptic impairment.[11−15]

Recent evidence demonstrates that exosomes are involved in tau propagation.[16−19] Exosomes are released from neurons, glia, and many cell types and are characterized as extracellular vesicles of the endosomal origin[20−23] for the removal of cellular components. Exosomes participate in the transcellular shuttling of exosome cargo molecules consisting of proteins, RNAs, lipids, and metabolites.[24] Tau is present in exosomes from the cerebrospinal fluid of AD patients.[25] Studies of neuron-derived exosomes (NDEs) isolated from the plasma of AD patients demonstrate that p-Tau levels in NDE predict conversion from mild cognitive impairment to dementia of AD.[17] Significantly, the intracranial injection of AD patient-derived plasma NDEs into the mouse brain results in AD-like tau neuropathology. Furthermore, the inhibition of the exosome synthesis retards tau propagation in the mouse brain involving microglia.[16] These findings show that exosomes can mediate transcellular spreading of tau in the brain.

Familial AD (FAD) represents genetic forms of AD and sporadic AD represents AD patients with no known genetic mutation(s).[26−28] Gene mutations of FAD have provided much insight into molecular mechanisms of AD. These FAD gene mutations consist primarily of mutations of presenilin 1 (PS1), which represent most FAD, together with the mutations of presenilin 2 (PS2) and the amyloid precursor protein (APP).[29−31] Such FAD mutations result in progressive cognitive dysfunction and brain neuropathology consistent with AD.

To gain an understanding of the role of a FAD mutant PS1 (mPS1) in exosome-mediated tau neuropathology, we previously examined the role of the A246E mPS1[32,33] in regulating exosomal p-Tau produced by mPS1 patient-derived induced pluripotent stem cell (iPSC) neurons.[34] The mPS1 iPSC neurons displayed elevated p-Tau and secreted exosomes containing p-Tau. The intracranial injection of these exosomes into the mouse brain resulted in aggregated p-Tau in the hippocampus.[34] These results show that the A246E mPS1 participates in the exosome-mediated transmission of tau pathology.

These findings lead to the important question: What is the composition of the protein cargo of mPS1 NDEs? We hypothesized that the A246E mPS1 mutation may dysregulate the protein cargo of exosomes. To test this hypothesis, we performed a study to define the proteome of exosomes produced by patient-derived iPSC neurons with or without the mPS1 A246E mutation, which is achieved by nano-liquid chromatography with tandem mass spectrometry (LC–MS/MS)-based proteomics.

Significantly, our results demonstrate that mPS1 dysregulated the exosome cargo proteins. Global proteomic data were acquired by LC–MS/MS tandem mass spectrometry with label-free quantification and assessed for gene ontology (GO), STRING protein networks, brain functions, and AD mechanisms. Results showed that mPS1 alters the profile of exosome cargo proteins to result in (1) proteins present only in mPS1 exosomes, and not in controls, (2) the absence of proteins in the mPS1 exosomes, which were present only in controls, and (3) shared proteins which were upregulated or downregulated in the mPS1 exosomes compared to controls. These findings show that the A246E mutation of PS1 dysregulates the protein cargo of exosomes generated by human iPSC neurons.

Results

Mutant PS1 Exosomes Induce Tau Deposits in the Mouse Brain

Deposits of accumulated tau in the brain is a characteristic of AD pathology and neurodegeneration. We found that when exosomes produced from iPSC neurons expressing the mutant A246E presenilin (mPS1) were injected into the mouse brain, tau pathology resulted which was observed as deposits of tau in the brain hippocampus (Figure S1). In contrast, control exosomes (from wild-type iPSC neurons) injected into the mouse brain had no effect.[34]

The functional difference of the mPS1 exosomes to induce tau deposits, compared to control exosomes, led to the purpose of this study to assess the proteome cargoes of mPS1 exosomes compared to the control exosomes.

Workflow Strategy to Assess the Protein Cargo of Exosomes Produced by mPS1 iPSC Neurons Compared to Controls

The proteomic strategy for analyses of exosomes generated by mPS1 and control iPSC neurons is conducted, as shown in Figure , using label-free quantitative mass spectrometry and bioinformatics tools for data analyses. This study was conducted by (a) isolation of exosomes from mPS1 iPSC neurons and from control iPSC neurons, with the confirmation of neuronal markers (Table S1), (b) nano-LC–MS/MS tandem mass spectrometry of tryptic digests of exosomes, and (c) bioinformatics to assess biological processes by GO and protein interaction networks by STRING-db.

Figure 1

Proteomic study design of exosomes generated by mPS1 iPSC neurons and by control iPSC neurons. (a) mPS1 human neurons and control wild-type PS1 human neurons for exosome isolation. The iPSC neurons were derived (by a reprograming protocol) from a patient with A246E mPS1 and from a healthy control patient with wild-type PS1, as we have reported earlier.[19,34−37] It is known that reprograming of the fibroblast from biopsies into pluripotent stem cells and differentiation into neurons erases the aging phenotype and generates “age-equivalent” iPSC neurons.[38,39] The neuronal phenotype of the iPSC neurons has been confirmed by the presence of multiple neuronal markers representing synaptic neurotransmission. Exosomes generated by the neurons were released into media (three biological replicates) which was collected for exosome isolation. (b) Nano-LC–MS/MS tandem mass spectrometry of exosomes. Proteins of exosomes were collected by MeOH precipitation, digested with trypsin/LysC by peptide solid-phase extraction (SPE),[40] and subjected to nano-LC–MS/MS tandem mass spectrometry.[41] (c) Bioinformatics of proteomic data. MS/MS mass spectrometry data were analyzed for the peptide spectrum matching and protein identification, combined with quantification, by PEAKS (v. 8.5) software.[42,43] Proteomic data were analyzed for biological systems by GO[44,45] and for protein interaction networks by STRING-db.[46−48]

Exosomes Secreted from mPS1 and Control Human iPSC Neurons

Exosomes isolated from mPS1 and control iPSC neurons were subjected to nanoparticle tracking analysis (NTA) to assess the distribution of exosome particle sizes (Figure S2). The mPS1 NDEs had a peak of particles with a diameter of ∼160 nm (∼100–300 nm). The control exosomes had a peak of particles with a diameter of ∼170 nm (∼100–300 nm). These vesicle diameters fell within the reported ranges of exosome diameters of approximately 50–150 nm.[49−51] These vesicles are enriched for exosome components including CD63 and CD81, as shown in the proteomic data for these exosomes (see Supporting Information Data S1). These markers indicate exosomes derived from endosomes.[20−23]

Protein Counts for Proteomic Data of mPS1 and Control Exosomes

Nano-LC–MS/MS-based proteomics identified 1117 total proteins from exosomes secreted by mPS1 and control iPSC neurons (Figure ). For the mPS1 and control exosomes, 842 and 1008 proteins were identified, respectively. Proteins uniquely present in only the mPS1 exosomes numbered 109 and proteins present in only the control exosomes numbered 275. The mPS1 and control exosomes shared 733 proteins present in both groups.

Figure 2

Protein identification counts of mPS1 and control exosome proteomic data. (a) Protein counts. The numbers of proteins identified in mPS1 and control exosomes are shown, including proteins present only in mPS1 or control exosomes, as well as shared proteins. (b) Venn diagram of unique and shared proteins of mPS1 and control exosomes. Proteins present only in mPS1 or control exosomes, or shared in both types of exosomes are shown.

Proteins Present Only in mPS1 Exosomes

The mPS1 exosomes contained 109 unique proteins (Figure , and listed in Supporting Information Data S1), which were absent in the control exosomes. GO analyses[45] revealed a significant enrichment of the mPS1 only proteins in biological pathways of the extracellular matrix (ECM) and structural organization, cell adhesion, development, and multicellular processes (Figure ). The mPS1 only proteins were also enriched in the molecular binding of glycosaminoglycans, ECM, calcium, as well as endopeptidase activity. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses[52] indicated significant enrichment in protein homeostasis, ECM receptor interaction, glycosaminoglycan degradation, as well as protein degradation and processing in the lysosome and endoplasmic reticulum (ER), respectively. These findings suggest that proteins present in only mPS1 exosomes participate in ECM binding and endopeptidase functions, which involve protein processing by lysosomes and ER.

Figure 3

GO analyses of proteins present only in mPS1 exosomes. GO analyses of proteins found only in mPS1 exosomes indicate involvement in (a) biological process pathways, (b) molecular function pathways, and (c) KEGG pathway. GO enrichment is significant with FDR <1%.

Protein network analyses of proteins present in only the mPS1 exosomes was evaluated by STRING.[46,47] The analysis showed that 38 out of the 109 mPS1 only proteins are in significant protein networks as nodes, as illustrated in Figure . The predicted protein interactions include functions of the extracellular structure and ECM and sulfur binding (Figure ).

Figure 4

Protein network analyses of proteins present only in mPS1 exosomes. STRING-db protein interaction analyses indicate that 38 proteins (out of the 109 proteins identified only in mPS1 exosomes) are enriched for known protein–protein interactions. Interactions utilized scores set to high confidence (0.7) that predicted links existing among proteins.

The top network hub proteins with the largest number of interactors displayed three to seven interactions for each hub (Table ). The interactors for each hub protein are listed in Supporting Information Data S1. Among these top 11 hub proteins, a large portion represented collagen genes of COL8A1, COL5A3, COL6A2, COL6A3, and COL14A1, combined with proteases involved in procollagen processing [PCOLCE and tolloid-like 1 (TLL1)] and collagen fibril assembly [decorin (DCN)]. Collagen proteins are secreted from cells and deposited into the ECM as structural proteins and signaling molecules.[53] COL8A1 is a network-forming collagen involved in vessel wall integrity.[53] COL5A3 is a fibril-forming collagen.[53] COL6A2 is a beaded filament[53] and has been found as a marker in cerebral amyloid angiopathy.[54] COL6A3 is also a beaded filament[53] and has roles in dystonia, alcohol dependence, and congenital muscular dystrophy[55−57] COL14A1 is a fibril-associated collagen.[53] The TLL1 metalloendopeptidase potentiates a procollagen processing protease known as bone morphogenetic protein.[53,58] The procollagen C-endopeptidase enhancer 1 protein (PCOLCE) promotes procollagen C-protease activity that cleaves type I procollagen C-propeptide.[59] DCN participates in the collagen fibril assembly and has a stimulatory effect in autophagy and inflammation.[60,61] Serine protease 23 (PRSS23) is a vascular protease.[62]

Table 1

Hub Proteins of Networks Present Only in mPS1 Exosomesa

gene name	protein description	# nodes	functions in the brain, AD, related
COL5A3	collagen type V α-3 chain	6	fibril-forming collagen[53]
COL6A2	collagen type VI α-2 chain	3	beaded filament collagen[53]
			marker present in cerebral amyloid angiopathy[54]
COL6A3	collagen type VI α-3 chain	5	beaded filament collagen[53]
			related to dystonia, alcohol dependence, and congenital muscular dystrophy[55−57]
COL8A1	collagen type VIII α-1 chain	7	network-forming collagen[53]
			involved in vessel wall integrity[53]
COL14A1	collagen type XIV α-1 chain	4	fibril-associated collagen with interrupted triple helices[53]
DCN	decorin	4	role in collagen fibril assembly[60]
			stimulatory effect in autophagy and inflammation[61]
TLL1	tolloid-like 1, metalloendopeptidase	4	metalloprotease that cleaves fibrillar procollagens I, II, III, V, and VI[58]
			role in neurogenesis[58]
PCOLCE	procollagen endopeptidase enhancer 1	4	enhances procollagen C-proteinase activity that cleaves type I procollagen C-propeptide[59]
PRSS23	serine protease 23	4	vascular protease[62]
IGFBP7	insulin-like growth factor-binding protein 7	3	attenuates function of ILPs[63]
			upregulated in human AD brain[63]
			inhibition of IGFBP7 in AD mice restores memory impairment and associative learning behavior[63]
			biomarker for AD[63]
APOA2	apolipoprotein A2	3	role in lipid metabolism[64]
			associated with lifespan and cognitive function[64]

Proteins identified only in the mutant PS1 exosomes are listed by gene name, number of interacting proteins terms “nodes”, and description of the protein function. All proteins were identified with FDR less than 1% (see Experimental Procedures).

Several of the hub proteins have been shown to participate in AD-related functions of cognition and learning. IGFBP7, insulin-like growth factor-binding protein 7, a protein that attenuates the function of insulin-like peptides (ILPs), is upregulated in the brains of AD patients and in a mouse model of AD (APP/S1-21 mice containing the Thy1 promoter-driven APP KM670/671NL and Thy1 promoter driven PS1 L166P transgenes);[63] the expression of IGFBP7 in mice results in impaired memory and significantly, the inhibition of IGFBP7 restores memory deficits. These findings support a role for IGFBP7 in AD-related memory loss, involving IGFBP7 attenuation of ILP function. With respect to APOA2, a protein involved in lipid metabolism, this protein is associated with cognitive function and lifespan.[64]

The mPS1 exosomes also contain ADAMTS1, ADAMTS7, and ADAMTS12 (Figure ) which are not present in control exosomes. These three proteins are members of the ADAMTS protease family of zinc metallopeptidases, which are secreted and act on ECM components.[65,66] These components have been found to be associated with AD and related human diseases in genome-wide association studies (GWAS). Notably, ADAMTS1 has been found to be associated with the risk for late-onset AD in GWAS evaluations.[67] ADAMTS1 is also linked to degenerative intervertebral disc disease.[68] ADAMTS12 has been found to be associated with cerebral vascular aneurysm and pediatric stroke.[69] ADAMTS7 is linked to atherosclerosis and arterial disease.[67,70,71]

Among other proteins present only in mPS1 exosomes, several are involved in ECM functions. TGFBI (transforming growth factor-β-induced protein ig-h3) is an abundant protein based on its quantification (Figure , and Supporting Information Data S1). TGFBI is a secreted ECM protein that binds to type I, II, and IV collagens in cell–collagen interactions.[72−74] Type I and IV collagens are both present in mPS1 and control exosomes (see Supporting Information Data S1).

Figure 5

Abundance of quantifiable proteins present only in mPS1 exosomes. Quantifiable proteins are assessed for their abundance in the mPS1 exosomes by bar graphs and indicated by intensity areas from mass spectrometry data (conducted as described in the Experimental Procedures section).

The EXT2 protein, exostosin-2, is also present in only the mPS1 exosomes (Figure , and Supporting Information Data S1). EXT2 is a glycosyltransferase involved in the biosynthesis of heparin sulfate found at cell surfaces and in the ECM, participating in cell–matrix interactions.[75] Another protein of similar abundance is POSTN, which is a secreted ECM protein involved in cell adhesion and tissue regeneration;[76] POSTN participates in epithelial/mesenchymal interactions which are important in inflammation through the activation of NF-κB.

Overall, distinct proteins present in only the mPS1 exosomes (and not in control exosomes) possess ECM functions which include collagen isoforms and proteases for procollagen processing, ADAMTS zinc metallopeptidases which act on ECM substrates, and abundant TGFBI, EXT2, and POSTN which participate in ECM mechanisms. These findings suggest that the mPS1 exosomes uniquely contain protein cargo components involved in the ECM structure, function, and regulation.

Proteins Present Only in Control Exosomes and Absent in mPS1 Exosomes

Proteomic analyses of the control exosomes identified 275 unique proteins (Figure , and listed in Supporting Information Data S1), which were absent in the mPS1 exosomes. GO analyses indicated significant enrichment of the control only proteins in biological pathways of the cellular component organization and biogenesis in cellular processes, protein localization, and translational initiation (Figure ). The control only proteins were also enriched in the molecular binding functions for protein, RNA, ribosomes, chromatin, and cell adhesion. Control only proteins were also enriched in proteasome activity. KEGG analyses indicated significant enrichment in ribosome, proteasome, and amino acid biosynthesis pathways. These results suggest that proteins present in only the control exosomes, and absent in the mPS1 exosomes, function in cellular protein localization and translational mechanisms, combined with binding functions of diverse molecules consisting of proteins, RNA, ribosomes, chromatin, as well as cell adhesion.

Figure 6

GO analyses of proteins present only in control exosomes. GO analyses of proteins present only in control exosomes indicate involvement in (a) biological process pathways, (b) molecular function pathways, and (c) KEGG pathway. GO enrichment is significant with FDR at <1%.

Protein network analyses of proteins present in only the control exosomes were assessed by STRING for predictions of protein–protein interaction networks. STRING analyses showed that 141 out of the 275 control only proteins are associated with significant protein networks, as illustrated in Figure . The top network hub proteins displayed 10 to 26 protein interactions for each hub (Table ). The interacting proteins for the top hub proteins are listed in Supporting Information Data S1. The hub proteins with high numbers of protein interactions (Table ) represent clustered networks of ribosomal proteins, protein phosphatases, translation initiation factors, RNA binding proteins for RNA processing and regulation, proteasome functions, and others (Figure ). The presence of these hubs in only control exosome networks highlights the absence of these protein functions in the mPS1 exosomes.

Figure 7

Protein network analyses of proteins present only in control exosomes. STRING-db network analyses indicate that 141 proteins (out of the 275 proteins present in only control exosomes) were enriched for known protein–protein interactions. Interactions utilized scores set to high confidence (0.7) to predict protein networks.

Table 2

Hub Proteins of Networks Present Only in Control Exosomesa

network cluster	gene name	protein description	# nodes	reported functions in AD or neurodegeneration
ribosomes	RPL4	60S ribosomal protein L4	22	None
	RPL7	60S ribosomal protein L7	17	forms aggregates with tau pathology[119]
	RPL12	60S ribosomal protein L12	15	associated with AD transcriptional changes[120]
	RPL13A	6S ribosomal protein L13a	11	None
	RPL18A	60S ribosomal protein L18a	26	None
	R	ribosomal protein L19	24	decreased synthesis in p-tau neurons[121]
	RPL27A	60S ribosomal protein L27a	16	None
	RPS2	40S ribosomal protein S2	16	None
translation initiation	ElF2S3	eukaryotic translation initiation factor 2 subunit 3	17	None
	ElF3B	eukaryotic translation initiation factor 3 subunit B	10	None
	EIF3J	eukaryotic translation initiation factor 3 subunit J	15	None
phosphatase	PPP2R2A	serine/threonine-protein phosphatase 2A 55 kDa regulatory subunit B α	17	dephosphorylation of p-tau[92]
RNA binding and processing	SRSF1	serine/arginine-rich-splicing factor 1	17	participates in binding to tau pre-mRNA[117]
	SRSF7	serine/arginine-rich-splicing factor 7	10	role in regulation of tau RNA alternative splicing[118]
	FUS	RNA-binding protein FUS	15	mutations in FUS are linked to familial ALS and FTLD[96−99,105,106]
	HNRNPU	heterogeneous nuclear ribonucleoprotein U	15	None
	PA2G4/EBP1	proliferation-assoc. protein 2G4	14	None
	DHX9	ATP-dependent RNA helicase	12	None
	NUDT21	cleavage and polyadenylation specificity factor subunit 5	12	None
proteasome	PSMC1	26S proteasome regulatory subunit 4	12	None
	PSMC4	26S proteasome regulatory subunit 68	14	dysregulated in AD and accumulates in Lewy bodies in PD[122]
	PSMD11	26S proteasome non-ATPase regulatory subunit 11	13	None

Proteins identified only in the control PS1 exosomes are listed by gene name, number of interacting proteins terms “nodes”, and description of protein function. All proteins were identified with FDR less than 1% (see Experimental Procedures).

Among proteins present in only the control exosomes, ribosome binding proteins comprise the largest group of hub proteins each having large numbers of interacting proteins (11–26 interactions for each hub). Such hub proteins consist of the 60S ribosomal proteins RPL4,[77] RPL7,[78,79] RPL12,[80] RPL13A[81] RPL18A[82] RPL19,[83] and RPL27A,[84] combined with the 40S ribosomal protein RPS2.[85] This ribosomal cluster of protein hubs (shown in Figure ) includes the translation initiation factors EIF2S3, EIF3B, and EIF3J[86−88] as hub proteins (Table ). This cluster of hubs for ribosomal and initiation factors participates in protein translation. The presence of these ribosomal proteins in only control exosomes suggests that the mPS1 exosomes may be compromised in protein translation functions.

Of notable interest is that only the control exosomes contain the hub protein PPP2R2A, serine/threonine protein phosphatase 2A regulatory subunit B α, a subunit of PP2A (Table ), which is the major tau phosphatase.[89−91] PP2A is a heterotrimeric complex composed of catalytic, regulatory, and scaffolding subunits.[89,90] The PPP2R2 regulatory subunit of PP2A targets p-Tau as a substrate for PP2A.[92] Along with PPP2R2A, the control exosomes (not mPS1 exosomes) contain the three phosphatase catalytic subunits of PP1α, PP1β, and PP2B (calcineurin) (Supporting Information Data S1). The exclusion of these phosphatase subunits from the mPS1 exosomes, compared to their presence in the control exosomes, suggests that the absence of phosphatase components may promote p-Tau and its neuropathology, as previously observed with mPS1 exosome injections into the mouse brain.[36]

The control exosomes also uniquely contain hub proteins of networks for RNA binding and processing proteins (Table ), which are absent in the mPS1 exosomes. These RNA-modulating hub proteins consist of SRSF1[93] and SRSF7[94] pre-messenger RNA (mRNA) splicing factors, the HNRNPL[95] component of ribonucleoprotein complexes of heterogeneous nuclear RNA in the nucleus, the RNA-binding proteins FUS[96−101] and PA2G (EBP1),[102] the ATP-dependent RNA helicase DHX9,[103] and the NUDT21[104] factor for 3′-RNA cleavage and polyadenylation.

Notably, the FUS RNA binding protein is significantly elevated in human brains of AD and frontotemporal lobular degeneration (FTLD) subjects.[96] Mutant forms of FUS are linked to amyotrophic lateral sclerosis (ALS) and FTLD.[97−99,105,106] The RNA-binding ability of mutant FUS is necessary for neurodegeneration.[97] Mutant FUS results in the defective RNA metabolism[99,100] and suppresses axonal protein synthesis.[99] FUS is normally present in nuclei[96] but mutant FUS accumulates in the cytoplasm of ALS and FTLD human brains[97,99−101] and results in an increase of toxicity involving synaptic dysfunction and behavioral deficits in motor and cognitive functions.[99−101]

The control exosomes contain components of the proteasome complex consisting of a cluster of hub proteins consisting of PSMC1, PSMC4, and PMSD11 regulatory subunits of the 26S proteasome[107−109] (Table ). This cluster interacts with COPS2, COPS3, and COPS7B[110−112] (Figure ), components of the COP9 signalosome complex,[113,114] which is involved in the ubiquitin conjugation pathway through the regulation of E3 ligase complexes for the regulation of ubiquitin-targeted proteasome degradation of proteins.[113,115,116]

The absence of these protein network hubs in the PS1 exosomes, compared to their presence in only the control exosomes (Table ), demonstrate that mPS1 exosomes display alterations in ribosomal proteins, serine/threonine-protein phosphatase, translation initiation factors, RNA binding proteins, and proteasome functions. Furthermore, several of the proteins present only in control exosomes are associated with tau with respect to binding to tau pre-mRNA by SRSF1,[117] the regulation of tau RNA alternative splicing by SRSF7,[118] RPL7 formation of aggregates with tau pathology,[119] association of RPL12 with AD transcriptional changes,[120] decreased synthesis of RPL19 in neurons containing p-Tau,[121] dysregulation of PSMC4 in AD,[122] and de-phosphorylation of tau by PPP2R2A.[92]

Proteins Shared by mPS1 and Control Exosomes: Upregulation and Downregulation by mPS1

Proteomic data showed that the shared proteins present in both the mPS1 and control exosomes consisted of 733 proteins (Figure , and listed in Supporting Information Data S1). GO analyses indicate that mPS1 and control exosomes share proteins functioning in biological processes of vesicle-mediated transport and localization, exocytosis and secretion, combined with biogenesis and organization of cellular components (Figure ). Molecular functions of the shared proteins consist of binding proteins, anions, carbohydrates, protein complexes, signaling, GTP, and small molecules (Figure ). KEGG pathway analyses indicate protein systems for endocytosis, ECM–receptor interaction, actin regulation, proteasome, and signaling systems.

Figure 8

GO analyses of proteins shared by mPS1 and control exosomes. GO analyses of proteins identified in mPS1 and control exosomes indicate involvement in (a) biological process pathways, (b) molecular function pathways, and (c) KEGG pathway. GO enrichment is significant with FDR <1%.

The quantitated shared proteins were assessed for the upregulation or downregulation in mPS1 compared to control exosomes by heat maps (Figure ). We also assessed proteins with log2(mPS1/control) ratios of ≥2, or ratios ≤−2, in another heat map illustration (Figure S3) that shows the high portion of proteins that were increased by 4- to 140-fold, or decreased by 75 to 99% in mPS1 exosomes compared to controls. Clearly, the substantial upregulation and downregulation of proteins occurred in mPS1 exosomes compared to control.

Figure 9

Upregulation and downregulation of shared proteins in mPS1 and control exosomes. Heat map illustration was conducted for the quantifiable proteins assessed by log2(mPS1/control) values (significance of p < 0.05). Quantified proteins shared between mPS1 and control exosomes were evaluated by comparing log2(mPS1/control), shown in heat maps.

Upregulated Proteins

The most highly upregulated proteins in the mPS1 exosomes compared to controls consisted of HSP90B1, AEBP1, and ALB. HSP90B1 (GRP94), heat shock protein 90 β family member 1, is a molecular chaperone that functions in the processing and transport of secreted proteins;[123,124] brain levels of this chaperone are increased in a mouse model of AD and may participate in Aβ clearance.[125] AEBP1, AE binding protein, is a carboxypeptidase-like protein involved in collagen metabolism.[126] ALB, albumin, functions as a carrier protein for small molecules and for stabilizing extracellular fluid, such as blood.[127] STRING analyses of upregulated proteins in mPS1 compared to control exosomes, revealed APP which yields β-amyloid, as a hub protein interacting with HSP90B1, ALB, TF (transferrin), A2M (α2-macroglobulin protease inhibitor), TGFB1 (transforming growth factor β 1), as well as glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (Figure a).

Figure 10

Network analyses of upregulated and downregulated proteins in mPS1 compared to control exosomes. (a) Upregulated proteins in mPS1. STRING-db analyses indicate that 17 proteins (out of the 46 upregulated shared proteins) were enriched with respect to known protein–protein interactions. (b) Downregulated proteins in mPS1. STRING-db analyses show that 17 proteins (out of the 39 downregulated shared proteins) were enriched with respect to known protein–protein interactions.

Specific functions of the upregulated proteins and their relationships to tau and APP/β-amyloid of AD are provided in Table S2.[89,128−243] Biological functions of upregulated proteins consisting of trafficking and cell morphology (AEBP1, COL11A1, FAT4, GDI1, GPC6, GSN, SPTAN1, THSD7A, TLN1, TUBB4A, VIM, and WDR1), biochemical and binding (ALB, DECR1, ENO1, GAPDH, GPC6, MDH1, PHGDH, SHBG, SRI, TF, and UGP2), chaperone protein folding (CCT4, HSP90B1, PDIA4, and STIP1), proteases and protease inhibitors (A2M, C1R, PRCP, and THSD7A), synaptic function (CRMP1, NTPX2, and PLXNB1), development and growth (CCDC80, IGSF8, and TGFBI), transcriptional and nuclear regulation (DIP2B, SUPT16H, and TNPO1), cell signaling (ARL3), RNA features (HNRNPK), protein translation (EIF3B), and immune response (LGALS3BP) are listed in order of the largest to lowest number of proteins in these categories. Among these upregulated functions, several proteins are associated with p-Tau and APP/β-amyloid of AD. PPP2R1A functions as a PP2A tau phosphatase[128,129] and, thus, the upregulation of PPP2R1A may increase p-Tau. The GAPDH protein has been shown to bind to tau.[130,131] Several proteins have been suggested as possible biomarkers of p-Tau or aggregated tau in NFTs which consist of A2M,[132,133] AEBP1,[134] MDH1,[135] and PLXNB.[136] Furthermore, APP and associated proteins with relationships with APP and β-amyloid were upregulated, consisting of A2M,[132,133] GAPDH,[130,131] GSN,[137,138] LGALS3BP,[139] PLXNB1,[136] and VIM.[140]

Downregulated Proteins

The most downregulated proteins in mPS1 compared to control exosomes consisted of NUMA1 (nuclear mitotic apparatus protein 1), COL4A2 (collagen type IV α 2 chain), PYGB (glycogen phosphorylase B), SPON1 (spondin 1), CTNNB1 (catenin β 1), AHSG (α 2-HS glycoprotein), and KIF5B (kinesin family member 5B) (see heat map in Figure ). STRING network analyses of the downregulated proteins illustrated known interactions of COL1A2 and COL4A2 with PLOD1 (Figure b). Among the downregulated proteins, several participate in mechanisms of AD; SPON1 interacts with APOE and APP;[244,245] CTNNB1 interacts with presenilins[246] and participates in apoptosis;[247] and KIF5B kinesin participates in the axonal transport of APP.[248]

Specific functions of the downregulated proteins are provided in Table S3.[166,249−340] Biological functions of downregulated proteins consist of trafficking and cell morphology (AHSG, AP2A2, COL1A2, COL4A2, DCN, FAT1, KIF5B, LLGL1, NID1, NUMA1, SEPT2, and TPM4) biochemical and binding activities (ALDOA, ANXA5, ASS1, ATP1A2, CKB, CLIC1, LOXL3, PLOD1, PSAP, PYGB, SLC9A3R1, and SPOCK1), synaptic regulation (CTTNNA2 and SPON1), chaperone protein folding (HSP90B1 and HSPD1), development and growth (CTTNNA2, SPON1, and YES1), transcriptional nuclear regulation (PARP1, SSBP1, and XRCC5), and cell signaling (CTTNNB1, GPC1, and MFGE8).

Exosome Phosphatases and Kinases Associated with Tau Phosphorylation

Proteomic data from mPS1 and control exosomes were evaluated for phosphatase and kinase components known to participate in de-phosphorylation and phosphorylation of tau, respectively.[89,341,342] Several protein phosphatase components were absent in the mPS1 exosomes and present in only the control exosomes; these components consist of the phosphatase catalytic subunits PPP1CA, PPP1CB, and PPP3CA [serine/threonine protein phosphatase PP1-α catalytic subunit, serine/threonine protein phosphatase PP1-β catalytic subunit, and phosphatase PP2B catalytic subunit (calcineurin), respectively][89,341−343] (Table ), combined with the regulatory phosphatase subunit of PPP2R2A (serine/threonine protein phosphatase 2A 55 kDa regulatory subunit B α isoform).[89,342−345] In addition, PPP2R1A (serine/threonine-protein phosphatase 2A 65 kDa regulatory subunit A α) is upregulated in mPS1 exosomes compared to the control (Table ). Roles for these phosphatase components for de-phosphorylation of tau have been reported.[89,342−345] The absence of numerous protein phosphatase components may be consistent with facilitation of p-Tau formation by mPS1 exosomes.[36]

Table 3

Tau Phosphatases and Kinases in mPS1 and Control Exosomes

		mPS1 or control exosomes
gene name	description	only mPS1	only control	shared, log2 of mPS1/control	role in tau phosphorylation	references
Phosphatases
PPP1CA	serine/threonine protein phosphatase PP1-α catalytic subunit		+		PP1 α catalytic subunit contributes to de-phosphorylation of tau	(89,342,343)
PPP1CB	serine/threonine protein phosphatase PP1-β catalytic subunit		+		PP1 β catalytic subunit contributes to de-phosphorylation of tau	(89,342,343)
PPP3CA (calcineurin)	serine/threonine-protein phosphatase PP2B catalytic subunit		+		calcineurin de-phosphorylates pS262 and pS396 on tau, which are both found in parahelical filaments	(89,342,343)
PPP2R2A	serine/threonine protein phosphatase 2A 55 kDa regulatory subunit B α		+		B55α regulatory subunit of PP2A recognizes p-tau substrate for PP2A de-phosphorylation	(89,92,342,345)
PPP2R1A	serine/threonine-protein phosphatase 2A 65 kDa regulatory subunit A α			+, 1.670	this subunit of PP2A is required for tau (and other substrate) de-phosphorylation	(89,92,342,345)
Kinases
PRKDC	DNA-dependent protein kinase catalytic subunit		+		DNA-PK catalytic subunit phosphorylates tau in vitro	(347)
CSNK2B	casein kinase II subunit β		+		CK2 phosphorylates SET to induce PP2A inhibition resulting in hyper-phosphorylation of tau; CK2 is overactive in AD	(349)
CDK1	cyclin-dependent kinase 1		+		CDK has high affinity for phosphorylation of SP motifs of tau; CDK1 also inhibits PP2A via SET	(346, 348)
FYN	tyrosine protein kinase Fyn			+, 0.131	Fyn phosphorylates tau and also inhibits the main tau phosphatase PP2A by phosphorylation	(341,350,351)
MAPK3 (ERK1)	mitogen-activated protein kinase 3			+, −0.156	Erk1 is involved in abnormal tau phosphorylation at sites identified in AD brains	(341,352)
MAPK1 (ERK2)	mitogen-activated protein kinase 1			+, −0.768	Erk2 can phosphorylate tau at 15 sites in vitro and its increased activity can contribute to tauopathy	(341,352,353)
SRC	proto-oncogene tyrosine-protein kinase Src			+, na	Src phosphorylates tau present in NFTs	(350)

Several protein kinases with roles in the phosphorylation of tau were identified in only control exosomes, consisting of PRKDC, CSKN2, and CDK1 (DNA-dependent protein kinase catalytic subunit, casein kinase II subunit β, and cyclin-dependent kinase 1, respectively)[341,346−349] (Table ). Among the proteins present in both mPS1 and control exosomes, FYN was moderately upregulated in mPS1 compared to control exosomes (Table ). FYN has dual functions for the direct phosphorylation of tau and also phosphorylation of PP2A, which inhibits the PP2A-mediated de-phosphorylation of p-Tau.[341,350,351] The MAPK3 and MAPK1 (Erk1 and Erk2 protein kinases involved in tau phosphorylation)[341,352,353] are moderately downregulated in mPS1 compared to control exosomes (Table ). The SRC kinase (proto-oncogene tyrosine-protein kinase Src) was present in both mPS1 and control exosomes; SRC participates in the phosphorylation of tau.[350] These findings suggest that the balance of several kinases is modified in the mPS1 exosomes compared to the control exosomes.

Discussion

The exosome cargo produced by patient-derived iPSC neurons with the presenilin mutation A246E (mPS1) was investigated by proteomics and protein network analyses and compared to exosomes generated by normal human iPSC neurons derived from a nondemented control patient. These studies were designed to gain insights into the protein cargo of mPS1 exosomes which induce the propagation of tau pathology in the mouse brain.[34] Significantly, findings showed that mPS1 alters the profile of exosome cargo proteins to result in (1) proteins present only in mPS1 exosomes and not in controls, (2) the absence of proteins in the mPS1 exosomes which were present only in controls, and (3) shared proteins which were upregulated or downregulated in the mPS1 exosomes compared to controls (Figure ). These data show that mPS1 dysregulates the proteome cargo of exosomes which result in the acquisition of proteins involved in ECM functions, deletion of proteins involved in RNA, and protein translation systems along with protease and related functions, combined with the upregulation and downregulation of shared proteins (Figure ). Notably, mPS1 exosomes display dysregulation of protein phosphatases and kinases known to be involved in regulating the phosphorylation status of tau. The altered profile of exosome cargo proteins by mPS1 may be associated with the ability of the mPS1 exosomes to propagate p-Tau neuropathology in the mouse brain.[34]

Figure 11

Mutant PS1 dysregulates exosome cargo through the acquisition and loss of proteins, combined with the upregulation and downregulation of proteins shared with control exosomes. The A246E mPS1 dysregulates the proteome cargo of exosomes generated by patient-derived iPSC neurons. Proteomic data demonstrated that mPS1 exosomes contain (1) proteins present only in mPS1 exosomes (pink circles) and not in controls, which included ECM collagen, proteases, and related components, (2) loss of proteins in the mPS1 exosomes (gray circles) which were present only in controls (gray circles), consisting of phosphatases, ribosomal, and protein translation proteins, RNA binding and processing proteins, proteasomes, and related, and (3) shared proteins (blue circles) for exocytotic, secretory, and related functions, many of which were upregulated or downregulated in the mPS1 exosomes compared to controls. The mPS1 exosomes induce tau pathology in the mouse brain after the in vivo injection,[34] but the control exosomes have no effect.

Tau undergoes phosphorylation and de-phosphorylation by protein kinases and phosphatases, respectively, to regulate the amount of p-Tau in AD and healthy brains. Hyperphosphorylation of tau occurs in AD and results in disrupted microtubule stability and deficits in axonal and synaptic functions.[354,355] Because the mPS1 exosomes result in the propagation of p-Tau in the mouse brain,[34] evaluation of mPS1 exosomes for regulators of phosphorylation was conducted. The most notable finding was that only the mPS1 exosomes lacked many phosphatase components known to participate in the de-phosphorylation of p-Tau that were present in control exosomes. The PPP2R2A regulatory subunit of protein phosphatase 2A was exclusively found in the control exosomes and not in mPS1 exosomes. PPP2R2A is a regulatory subunit that specifically targets p-Tau as the substrate for de-phosphorylation. The regulatory subunit with the catalytic and scaffolding subunits comprise the heterotrimeric phosphatase complex. There are numerous regulatory subunits for substrate recognition; therefore, it is significant that the tau-targeting subunit was present in only the control exosomes. In addition, the phosphatase catalytic subunits PPP1CA, PPP1CB, and PPP3CA were also exclusively found in the control exosomes and not in mPS1. The distinct loss of PPP2R2A by mPS1 exosomes for targeting tau as the substrate, combined with the loss of several catalytic subunits of protein phosphatases, may facilitate p-Tau involved in AD pathology.

With respect to protein kinases known to participate in phosphorylation of tau,[89] the PRKDC, CSKN2, and CDK1 kinases were identified in only the control exosomes and not the mPS1 exosomes. The tyrosine kinase FYN was moderately upregulated in mPS1 exosomes compared to controls; FYN has dual functions which include direct phosphorylation of tau and inhibiting PP2A the de-phosphorylation of p-Tau.[341,350,351] The MAPK3 and MAPK1 kinases were moderately downregulated in the mPS1 exosomes compared to controls. These findings suggest the dysregulation of the balance of several tau protein kinases in the mPS1 compared to control exosomes.

The mPS1 exosomes acquire distinct proteins which were absent in the control exosomes. Unique proteins present in only the mPS1 exosomes possess ECM functions consisting of collagen isoforms (collagen types V, VI, VIII, and XIV) and proteases (DCN protease and TLL1 metalloendopeptidase) for procollagen processing, ADAMTS zinc metallopeptidases which act on ECM substrates, and abundant TGFBI, EXT2, and POSTN which participate in ECM mechanisms. These findings suggest that the mPS1 exosomes uniquely contain protein cargo components involved in the ECM structure and function. Such ECM functions may participate in mPS1 exosome-mediated propagation of tau pathology in the brain.

Losses of proteins in mPS1 exosomes were observed because such proteins were present only in control exosomes. The control exosomes contain clusters of protein networks which function in ribosomal and initiation factors in protein translation systems, RNA binding and RNA processing systems, and proteasome components for protein degradation. The absence of these functional systems in the mPS1 exosomes suggests their lack of biosynthetic capacity for RNA processing and protein translation. Furthermore, the lack of proteasome components in the mPS1 exosomes could possibly contribute to the accumulation of misfolded proteins in tau pathology via the proteasome regulation of the misfolded protein response. Clearly, the mPS1 exosomes lack protein components integral to RNA-based protein translation combined with protein homeostasis. These findings demonstrate that mPS1 redirected the routing of approximately one-fourth of the normal exosomal proteins in a manner such that it prevents packaging into the mPS1 exosomes.

With respect to proteins that were absent in mPS1 exosomes compared to control exosomes, or absent in control exosomes compared to mPS1 exosomes, it is realized that such data are dependent on the mass spectrometry detection limits and dynamic range for data acquisition. Therefore, the “absence” of proteins is based on the mass spectrometry method. It is possible that proteins may be present at levels below the quantitative detection limit of the mass spectrometry method.

The proteomic data also show that a large portion of mPS1 and control exosome cargoes are common to both exosome types. These shared proteins represent fundamental exosome activities consisting of vesicle-mediated transport, exocytosis, and secretion processes, which involve molecular protein binding functions and enzymes. Among the shared proteins, mPS1 exosomes displayed a significant upregulation and downregulation of protein components (Figure ). It is of interest that the APP was upregulated in mPS1 exosomes. APP serves as substrate of PS1, which is the aspartyl protease subunit of the γ-secretase complex which catalyzes processing of APP to generate Aβ peptides.[222] Thus, mPS1 elevates its APP substrate in exosomes. It will be of interest in future studies to examine whether mPS1 exosomes induce Aβ neuropathology combined with the propagation of tau pathology in the brain.

Along with APP, the mPS1 exosomes (compared to control exosomes) also display the strong upregulation of HSP90B1, AEBP1, and ALB proteins. Brain levels of the HSP90B1 chaperone are increased in a mouse model of AD and may participate in Aβ clearance.[243] The strong downregulation of proteins also occurs in the mPS1 exosomes (compared to controls) which consist of NUMA1 (nuclear mitotic apparatus protein 1), COL4A2 (collagen type IV α 2 chain), PYGB (glycogen phosphorylase B), SPON1 (spondin 1), CTNNB1 (catenin β 1), AHSG (α 2-HS glycoprotein), and KIF5B (kinesin family member 5B). Among these downregulated proteins, several participate in AD, consisting of SPON1 interaction with APOE and APP,[244,245] CTNNB1 interacts with presenilins[246] and participates in apoptosis,[247] and KIF5B participates in the axonal transport of APP.[264,265]

The A246E mPS1 is one of the numerous mutations reported for PSEN1 in AD.[32,33,356,357] The results of this study demonstrating the mPS1 dysregulation of exosome proteins lead to the question of whether other PSEN1 mutations of AD might dysregulate exosome cargo molecules. PSEN1 mutations are the most common cause of early onset FAD. Mutant presenilins comprise the majority of the 150 presenilin mutations, which include about a dozen mutations of the homologous PS2 (PSEN2). It will be of interest to gain further insights into possible relationships of numerous presenilin mutations and exosome cargoes for the propagation of tau and related pathology in AD.

Experimental Procedures

Experimental Design

The workflow for proteomic analyses of exosomes isolated from human iPSC neurons expressing the A246E mPS1 or from control human iPSC neurons is shown in Figure . Human iPSC neurons were derived from the respective patient or healthy control tissue biopsies.[34,35] Each mPS1 and control group consisted of three biological replicates of neuronal cultures (n = 3) which allowed statistical evaluations (by Student’s t-test, significance of p < 0.05). Exosomes were isolated from the media of neuronal cell cultures of the mPS1 and control groups using ExoQuick-TC (System Biosciences, Palo Alto, CA) (Figure a). Exosome proteins were digested with trypsin/LysC and subjected to nano-LC–MS/MS tandem mass spectrometry analysis on a Dionex UltiMate 3000 nano LC and an Orbitrap Q-Exactive mass spectrometer (Thermo Fisher Scientific, Carlsbad, CA) (Figure b). Samples were injected twice (in randomized order) into the nano-LC–MS/MS system for global proteomic analyses. Bioinformatic analyses of MS1 and MS2 data for peptide and protein identification and label-free quantitation (LFQ) used PEAKS Studio 8.5 (Bioinformatics Solutions Inc., Waterloo ON, Canada). PEAKS searched the human protein sequence database (UniprotKB/SwissProt 2018_2 with 71,783 entries) for peptide spectrum matches and protein identification with LFQ (Figure c).

The criteria for the inclusion of an identified or quantifiable protein in a biological replicate sample required that the protein was identified in at least one of the two technical replicates per biological sample. The criteria for the inclusion of an identified or quantifiable protein in either the mPS1 or control groups required that the protein was identified in at least two out of the three biological replicates per group. Quantifiable data of the mPS1 and control groups were compared by Student’s t-test.

Human iPSC Neuronal Cultures: mPS1 and Control

Human iPSC neuronal cells were prepared from a control patient biopsy (non-demented) and from a patient possessing the A246E mutation in the PSEN1 gene, using our published protocol.[19,34,35] The control iPSC neurons were derived from a biopsy of a non-demented male subject aged 86 years[36] and the mPS1 neurons were derived from a biopsy of a male subject, age d56 years, harboring the A246E mPS1 with AD onset at about 50 years;[37] all subject data have been deidentified. It is known that reprograming of the fibroblast from biopsies into pluripotent stem cells and differentiation into neurons erases the aging phenotype and generates “age-equivalent” iPSC neurons.[38,39] Therefore, the mPS1 and control iPSC neurons prepared for this study are “age-equivalent.”

For the culture of iPSC neurons, neural stem cells plated at a density of 1.5 × 105 cells/cm2 on Matrigel-coated (70 μg/mL, BD Bioscience) dishes and were grown to ∼80% confluence. Neuronal differentiation was initiated by the removal of basic fibroblast growth factor (bFGF, Biopioneer) from the media (DMEM–F-12, 1% N-2, 2% B-27, Pen–Strep, 20 ng/mL bFGF). Differentiation into neurons was indicated by neuronal markers of synaptic neurotransmission (shown in Table S1). Conditioned culture media was collected from cells at 3–4 day intervals. Cell viability in cultures was typically greater than 90–95% viable.

Exosome Isolation and NTA

The isolation of exosomes from cell culture medium used ExoQuick-TC (System Biosciences, Inc.) according to the provided protocol. The medium was incubated with ExoQuick-TC with rotation overnight at 4 °C, then centrifuged at 1500g for 30 min at 4 °C. The pellet was placed in phosphate-buffered saline with EDTA-free protease and a phosphatase inhibitor cocktail and stored at −70 °C. Protein levels of exosome preparations were measured using the bicinchoninic acid protein assay kit (Pierce Biotechnology). Exosomes (10 μg) were evaluated for the size distribution and concentration by NTA with a NanoSight LM10 instrument.

Trypsin/LysC Digestion and LC–MS/MS

Proteins of exosome samples (100 μg each) were precipitated in 90% ice-cold methanol on ice for 15 min, and then centrifuged for 30 min at 14,000g (4 °C). The resulting protein pellet was dried in a vacuum centrifuge, resuspended in urea buffer (8 M urea, 50 mM Tris–HCl, pH 8), and sonicated. For reduction, dithiothreitol (DTT, 100 mM stock) was added to obtain 5 mM DTT and samples were incubated at 55 °C for 45 min, and cooled at room temperature (RT) for 5 min. For cysteine alkylation, iodoacetamide (IAA, 200 mM stock in 50 mM Tris–HCl, pH 8) was added to obtain 15 mM IAA, incubated in the dark at RT for 30 min, and quenched by the addition of DTT to 5 μM. To lower the urea to less than 1 M, samples were diluted with 50 mM Tris–HCl, pH 8. Trypsin/LysC (Promega) was added to each sample at a ratio of 50:1 protein/trypsin (w/w) and incubated at RT for 18–24 h, and quenched by the addition of trifluoroacetic acid (TFA) to less than 0.5%. Samples were stored at −70 °C.

Tryptic peptides were purified and desalted on C18 stage tip SPE using Empore C18 wafers (3M), as reported in the field.[40] The stage tip was washed with acetonitrile (ACN) and equilibrated with 0.1% TFA. Samples were loaded, washed with 0.1% TFA, eluted with 50% ACN/0.1% TFA, dried in a vacuum centrifuge, resuspended in water and sonicated, and peptide levels were determined by the total peptide assay kit (Thermo Fisher). Samples were dried in a SpeedVac and stored at −70 °C.

LC–MS/MS Tandem Mass Spectrometry

LC–MS/MS was performed on a Dionex UltiMate 3000 nano-LC and an Orbitrap Q-Exactive mass spectrometer (Thermo Fisher Scientific). Peptide samples were re-suspended in 2% ACN, 0.1% TFA to 0.6 μg/μL peptide concentration. Each sample was injected twice (2.5 μg per injection) onto the nanoLC column (75 μm inner diameter, 360 μm outer diameter, and 25 cm length) packed with the BEH C18 (1.7 μm diameter) solid-phase material and heated to 65 °C with a column heater.[41] LC used a flow rate of 0.3 μL/min using a 120 min linear gradient of 5 to 25% ACN/0.1% formic acid, followed by a 5 min linear gradient of 25 to 95% ACN/0.1% formic acid. MS and MS/MS spectra were obtained in positive ion data-dependent mode. MS1 was acquired in the profile mode with a 3 × 106 automatic gain control target, 100 ms maximum injection time, a 310–1250 m/z window, and 70,000 resolution (at m/z 200). MS2 was acquired in the centroid mode with 1 × 105 AGC target, 50 ms maximum time for injection, 2 × 103 minimum precursor intensity, 35 s per 10 ppm dynamic exclusion, 17,500 resolution (at m/z 200), a first mass of m/z 150, and HCD collision energy of 28. A LC–MS/MS report is provided in Supporting Information S1. LC–MS/MS files are available at www.proteomexchange.org with the dataset identifier PXD019424, or at www.massive.ucsd.edu with dataset identifier MSV000085478 or ID = 229a900a721d40ada5e6fc806241ba2b.

Protein Identification

MS and MS/MS data files were queried by PEAKS (v. 8.5) bioinformatic software[42] for peptide identification and label-free quantitation (LFQ) analyses (next section). The data files were searched against the UniprotKB/SwissProt human protein sequence database (release 2018_02) having 71,783 entries. Peptide identifications included searching of a decoy-fusion spectrum library of human proteins, generated by PEAKS v. 8.5 from the human protein database of UniprotKB/SwissProt. PEAKS parameters for protein and peptide identification consisted of trypsin (cleavages at Arg and Lys, and two missed or nonspecific cleavages allowed), carbomidomethylation on Cys, oxidation of Met, pGlu, and N-terminal acetylation, and phosphorylation were included in the search parameters. Precursor mass error tolerance was 25 ppm, mass tolerance for the fragment ion was 0.01 Da, and threshold peptide scores of −log10P ≥ 32. The threshold score was <1% FDR (false discovery rate), equivalent to −log10P > 20. The report of the PEAKS analyses is in Supporting Information S2. The threshold score for protein identification was −log10P ≥ 55, equivalent to 1% FDR. The Supporting Information provides assigned peptide sequences and protein identifications (Supporting Information Data S1). Supporting Information Data S1 (master table of data) summarizes the identified and quantified proteins in mPS1 and control exosomes. Single peptides passing the criteria for protein identification are provided with MS/MS spectra in Supporting Information S3.

Protein Quantification

Label-free quantitation (LFQ) of proteins was assessed by PEAKS (v. 8.5) (Supporting Information Data S1). Extracted ion chromatographs of MS2 peaks were converted to the area under the curve, and the peak areas of MS2 of each peptide spectrum were summed to determine protein relative abundances. Spectra were filtered for quality parameters prior to LFQ, consisting of peptide quality of >0.3, abundance of 1 × 104, and present in 25% or more of technical replicates. Replicates were compared for retention time and isotope pattern for inclusion in quantitation analysis. The normalization of technical variations used the LOESS-G application.[43]

Analytical replicate reliability was restricted by −log10P and quality assessed as 1/log(σ), where σ is the variance between technical runs; −log10P > 20 represents 1% FDR. The imputation of the quantitative area with a value of 0 was achieved with a value representing the lower 5% of values within standard deviation (SD) of |1|. Multiple isoforms within a protein group were inspected to assure that the isoforms were assigned the same quantitation data. Biological replicate values for protein quantifications of mPS1 and control groups were averaged and SD computed, using Student’s t-test to assess the significance (p < 0.05) of mPS1 compared to the control for significantly regulated proteins.

The PEAKS bioinformatic data analyses are provided in the master table of data (Supporting Information Data S1), which identifies the proteins present only in mPS1 exosomes, only in control exosomes, and proteins shared by the two groups. The Supporting Information Data S1 master table contains the lists of identified proteins, quantifiable proteins with their summed peak area values, and details of protein properties including % coverage and number of peptides per protein group.

GO and STRING-db Network Analyses

Identified and quantified proteins in experimental groups were evaluated for GO systems and protein–protein interactions using STRING-db (https://string-db.org/).

GO analyses indicated the significant enrichment of exosome protein groups to GO terms with FDR <1% using Benjamini–Hochberg.[44,45] FDR was assessed by hypergeometric testing, a probability distribution that assesses the significance of having hits within experimental gene sets compared to total genes in the GO pathway.

Protein networks were assessed by STRING (version 11.0)[46] (www.string-db.org). STRING utilizes a database of protein interactions data sets (DIP, BioGRID, HPRD, IntAct, MINT, and PDB). Significant protein–protein network enrichment was assessed by a probability p-value to indicate whether an experimental group of proteins have more interactions compared to interactions expected from a randomly selected protein group of the same size at a high confidence score of 0.7.[47]

Heat Maps of Significantly Upregulated and Downregulated Proteins in mPS1 and Control Exosomes

Quantifiable proteins shared by the mPS1 and control groups were analyzed by log2(mPS1/control) ratios and shown in heat maps for significant differences between mPS1 and control (p < 0.05). Power analysis of quantitative data were conducted using the pwr.2p.test function of the pwr package in R, with n = 3, significance set to 0.05 (p < 0.05), and power of 0.8 or greater. Heat maps were generated with the heat map function in R studio (https://www.rstudio.com/products/rstudio/).[48]

Data Availability

LC–MS/MS files are available at www.proteomexchange.org for identifier number PXD019424, or at www.massive.ucsd.edu under the identifier number MSV000085478 or ID = 229a900a721d40ada5e6fc806241ba2b.