Synaptic Involvement of the Human Amygdala in Parkinson’s Disease

α-Synuclein, a protein mostly present in presynaptic terminals, accumulates neuropathologically in Parkinson’s disease in a 6-stage sequence and propagates in the nervous system in a prion-like manner through neurons and glia. In stage 3, the substantia nigra are affected, provoking motor symptoms and the amygdaloid complex, leading to different nonmotor symptoms; from here, synucleinopathy spreads to the temporal cortex and beyond. The expected increase in Parkinson's disease incidence accelerates the need for detection biomarkers; however, the heterogeneity of this disease, including pathological aggregates and pathophysiological pathways, poses a challenge in the search for new therapeutic targets and biomarkers. Proteomic analyses are lacking, and the literature regarding synucleinopathy, neural and glial involvement, and volume of the human amygdaloid complex is controversial. Therefore, the present study combines both proteomic and stereological probes. Data-independent acquisition-parallel accumulation of serial fragmentation proteomic analysis revealed a remarkable proteomic impact, especially at the synaptic level in the human amygdaloid complex in Parkinson’s disease. Among the 199 differentially expressed proteins, guanine nucleotide-binding protein G(i) subunit alpha-1 (GNAI1), elongation factor 1-alpha 1 (EEF1A1), myelin proteolipid protein (PLP1), neuroplastin (NPTN), 14-3-3 protein eta (YWHAH), gene associated with retinoic and interferon-induced mortality 19 protein (GRIM19), and orosomucoid-2 (ORM2) stand out as potential biomarkers in Parkinson’s disease. Stereological analysis, however, did not reveal alterations regarding synucleinopathy, neural or glial populations, or volume changes. To our knowledge, this is the first proteomic study of the human amygdaloid complex in Parkinson’s disease, and it identified possible biomarkers of the disease. Lewy pathology could not be sufficient to cause neurodegeneration or alteration of microglial and astroglial populations in the human amygdaloid complex in Parkinson’s disease. Nevertheless, damage at the proteomic level is manifest, showing up significant synaptic involvement.


Synaptic Involvement of the Human Amygdala in Parkinson's Disease
Sandra Villar-Conde 1,2 , Veronica Astillero-Lopez 1,2 , Melania Gonzalez-Rodriguez 1,2 , Daniel Saiz-Sanchez 1,2 , Alino Martinez-Marcos 1,2,* , Isabel Ubeda-Banon 1,2,* , and Alicia Flores-Cuadrado 1,2   α-Synuclein, a protein mostly present in presynaptic terminals, accumulates neuropathologically in Parkinson's disease in a 6-stage sequence and propagates in the nervous system in a prion-like manner through neurons and glia.In stage 3, the substantia nigra are affected, provoking motor symptoms and the amygdaloid complex, leading to different nonmotor symptoms; from here, synucleinopathy spreads to the temporal cortex and beyond.The expected increase in Parkinson's disease incidence accelerates the need for detection biomarkers; however, the heterogeneity of this disease, including pathological aggregates and pathophysiological pathways, poses a challenge in the search for new therapeutic targets and biomarkers.Proteomic analyses are lacking, and the literature regarding synucleinopathy, neural and glial involvement, and volume of the human amygdaloid complex is controversial.Therefore, the present study combines both proteomic and stereological probes.Data-independent acquisition-parallel accumulation of serial fragmentation proteomic analysis revealed a remarkable proteomic impact, especially at the synaptic level in the human amygdaloid complex in Parkinson's disease.Among the 199 differentially expressed proteins, guanine nucleotide-binding protein G(i) subunit alpha-1 (GNAI1), elongation factor 1-alpha 1 (EEF1A1), myelin proteolipid protein (PLP1), neuroplastin (NPTN), 14-3-3 protein eta (YWHAH), gene associated with retinoic and interferon-induced mortality 19 protein (GRIM19), and orosomucoid-2 (ORM2) stand out as potential biomarkers in Parkinson's disease.Stereological analysis, however, did not reveal alterations regarding synucleinopathy, neural or glial populations, or volume changes.To our knowledge, this is the first proteomic study of the human amygdaloid complex in Parkinson's disease, and it identified possible biomarkers of the disease.Lewy pathology could not be sufficient to cause neurodegeneration or alteration of microglial and astroglial populations in the human amygdaloid complex in Parkinson's disease.
Nevertheless, damage at the proteomic level is manifest, showing up significant synaptic involvement.
Parkinson's disease (PD) is the second most common neurodegenerative disease and is characterized by nigrostriatal denervation caused by the death of dopaminergic neurons in the substantia nigra, provoking motor symptoms that lead to clinical diagnosis (bradykinesia, rigidity, resting tremor, and gait alterations).In parallel, and often decades before, pathological involvement of many other structures leads to prodromal, nonmotor symptoms (hyposmia, sleep disorders, neuropsychiatric features, autonomic dysfunction, mild cognitive impairment and pain, and cognitive disturbances) (1)(2)(3).This disease is included in synucleinopathies due to the involvement of misfolding and aggregation of α-synuclein (α-syn).This proteinopathy is cumulative and predictable allowing staging and pathological diagnosis (4).Interestingly, stage 3 is characterized not only by the involvement of the substantia nigra causing motor symptoms but also by the involvement of the amygdaloid complex (AC), which provokes hyposmia (5), dysautonomia, apathy, anxiety, and depression (6,7) and facilitates synucleinopathy progression to the temporal cortex and beyond causing dementia (8,9).
α-Synucleinopathy is characterized by aggregates constituting Lewy bodies (LBs) and neurites (LNs) (10).This protein (α-syn), encoded by the SNCA gene, is highly expressed throughout the brain and is particularly enriched in the presynaptic terminals of neurons.In physiological conditions, α-syn is a synaptic vesicle-bound multimer, whereas in diseased conditions, it is able to modify its conformation forming deposits (11).Furthermore, these misfolded proteins can induce naïve proteins to misfold (seeding) and propagate (spreading) (12) through the nervous system in a prion-like manner through synaptic connections (13).Although spreading mainly occurs through neurons, it has been proposed that α-syn toxicity includes neuroinflammation, where α-syn could be a primary trigger for this phenomenon (13).The earliest studies revealed the presence of astrocytes loaded with α-syn aggregates in the AC (14,15) and microglia activation in PD (16).Whether astroglia and microglia facilitate the clearance of α-syn or, conversely, facilitate its spread (17)(18)(19)(20) remains unclear.However, the implications for the glial population are not known.
The expected increase in the incidence of PD estimated for the coming years (21) raises the urgency to find treatments that stop or delay the disease.However, the heterogeneity of PD, including clinical phenotypes, genetic predispositions, pathological aggregates, and pathophysiological pathways, poses a challenge in the search for new therapeutic targets and biomarkers (22).Therefore, it would be interesting to carry out proteomic analysis on potential proteins that interact with α-syn or synaptic proteins to try to identify pathophysiological pathways and potential biomarkers.Available proteomic studies have focused on the substantia nigra, but essential hubs, such as the AC, have not been investigated (23).
On the other hand, examination of neurodegeneration using specific neural markers has not been carried out, and the glial population has not been stereologically quantified.The AC, as mentioned above, is preferentially involved in pathology.It is a heterogeneous and intricate structure with multiple cortical and subcortical connections (24).Historically, a differential distribution of LBs and LNs has been qualitatively described in this nuclear complex, with the accessory cortical and central nucleus being the most involved in pathology (25,26), although quantitative data have not confirmed these observations (26).This apparent differential affectation could be explained according to the prion-like hypothesis (13).The cortical nucleus has projections with the olfactory regions, the central nucleus has connections with the autonomic nuclei of the brainstem, and the basal and lateral nuclei have connections with the limbic system and neocortex (27).The greater presence of LBs in the cortical nucleus was shown to be related to its neurodegeneration and loss of volume in a unique stereological study published on this topic (28).Hence, the possible volume reduction of the human AC with PD could be a biomarker.However, the results of volume quantification in patient (29)(30)(31)(32) studies are highly contradictory.Postmortem estimation using the stereological Cavalieri method could provide a more precise approximation since it makes it possible to delimit the different nuclei of the AC.
For all these reasons, the present study consists of a complementary proteomic and stereological analysis of the human AC.Proteomic analysis is used to identify the main protein alterations of the human AC in PD related to α-syn, neuroinflammation, and synapses that can act as biochemical biomarkers of the disease.Stereological analysis is used to analyze the fraction of the area occupied by α-syn in the main nuclei of the AC in PD and its impact on specifically identified neuronal, microglial, and astroglial populations, as well as possible volumetric changes of its anatomical structure that can be used as imaging biomarkers.Therefore, this report constitutes the first proteomic and stereological study analyzing neuronal and glial populations using specific markers in the human AC in PD.

Experimental Design and Statistical Rationale
All experiments were performed in accordance with the Declaration of Helsinki principles and the Clinical Research Ethics Committee of the University Hospital of Ciudad Real (PID2019-108659RB-I00).
Thirty-six postmortem amygdala from PD and non-PD (NPD) patients were used: 16 frozen unfixed cases (n = 8 PD, n = 8 NPD) for proteomic analysis (Table 1) and 20 fixed cases (n = 10 PD, n = 10 NPD) for stereological analysis (Table 2).First, data-independent acquisition-parallel accumulation serial fragmentation (dia-PASEF) was carried out by the Instituto Maimonides de Investigaci ón Biom édica de C órdoba IMIBIC Proteomic Facility.Second, for the bioinformatic study, different tools were used such as Perseus, Biological General Repository for Interaction Datasets (BioGRID), Synaptic Gene Ontologies (SynGO), and Ingenuity (IPA), to identify proteins, interactors with α-syn, differentially expressed proteins (DEPs) in synapses and their role, and biofunctional pathways and protein-protein interactions, respectively.After that, some DEPs were validated by immunofluorescence (fixed samples) and Western blot (frozen samples).
The stereology data were analyzed by Stereo Investigator (https:// www.mbfbioscience.com/products/stereo-investigator)software, including the area fraction fractionator (fraction of area occupied by α-syn), optical fractionator (density of neurons, microglia, and astroglia), and Cavalieri method (AC and nuclei volume).Comprehensive information on criteria sample selection, processing, analysis, and statistics are provided in the corresponding figures, tables, and main text.Samples were selected based on these criteria.Cases neuropathologically diagnosed with PD at Braak α-syn stage 4 to 6 were included in the PD group, while those with no clinical or neuropathological evidence of PD were grouped as NPD.Cases with other neurodegenerative disorders or pathologies that could interfere with the goal of the study, such as small vessel disease, were excluded.However, cases with early-stage Alzheimer's disease (Braak tau) were included (Tables 1 and 2).Only cases that were matched in age and brain weight were chosen, preventing these factors from influencing the results.Therefore, there were no statistically significant differences in age and brain weight between the PD and NPD groups in either the stereological (age: PD 76 ± 2; NPD 66 ± 5) (brain weight: PD 1260 ± 119.0; NPD 1263 ± 169.8) or the proteomic analysis (age: PD 70 ± 8; NPD 72 ± 5) (brain weight: PD 1299 ± 114.8; NPD 1257 ± 156.4).Finally, the distance to bregma was established to range from 4.0 to 10.7 mm for all AC specimens.All potential cases that fit the predetermined criteria were included in the investigation.However, the number of samples that could be collected from tissue biobanks limited the sample size.
Samples were precipitated using methanol/chloroform, and the pellet was dissolved in 100 μl of RapiGest SF Surfactant (Waters).Total protein was measured using the Qubit Protein Assay Kit (Thermo Fisher Scientific), and 25 μg of protein was digested using the iST kit (PreOmics).Peptides were diluted using LC-MS H 2 O 0.1% (v/v) formic acid to 10 ng/μl.Two hundred nanograms of peptides were loaded onto Evotips (Evosep) for purification.Pierce HeLa tryptic Digest Standard (Thermo Fisher Scientific) was also loaded for quality control.
Liquid Chromatography-Tandem Mass Spectrometry -Liquid Chromatography-Tandem Mass Spectrometry was carried out using an Evosep One LC system (Evosep) coupled to a TIMS Q-TOF instrument (timsTOF Pro, Bruker Daltonics) via a nanoelectrospray ion source (Captive Spray Source, Bruker Daltonics).Samples were separated with the predefined 60 sample per day method (21 min gradient time, 200 ng peptides) on a reverse-phase analytical column (8 cm × 150 μm, C18 1.5 μm (EV1109)) heated to 40 • C. The mobile phases were H 2 O and acetonitrile buffered with 0.1% formic acid (LC-MS grade, Thermo Fisher Scientific).
To build the MS/MS spectral libraries, a pool of equal mixtures of the original samples were analyzed using data-dependent acquisition-parallel accumulation serial fragmentation (dda-PASEF) (identified proteins in the spectral library are listed in Supplementary Data 1A), and the individual samples were analyzed using dia-PASEF.
The acquisition modes recorded spectra from 475 to 1000 m/z.Ion mobility resolution was calibrated with three Agilent ESI-L Tuning Mix ions (mass spectra, ion mobility: 622.0289 m/z, 0.9848 V cm −2 ; 922.0097 m/z, 1.1895 V cm −2 ; 1221.9906m/z, 1.3820 V cm −2 ) and set to 0.85 to 1.30 V cm −2 .dda-PASEF consisted of four PASEF MS/MS.The accumulation of ions and the ramp times parallel to the scans were 100 ms, with a total cycle time of 0.53 s.Any individually charged ions were excluded from MS/MS analysis.Moreover, precursors that reached the target value of 20,000 were excluded for 0.4 min.A Q window of ±2 Da for m/z <700 and ±3 Da for m/z >800 was used for isolating precursors.The 12 most intense ions were fragmented.For dia-PASEF, 21 nonoverlapping windows of 25 Da were set, resulting in a total cycle time of 0.95 s.The collision energy decreased linearly from 45 eV at an ion mobility of 1.30 Vs cm −2 to 27 eV at 0.85 Vs cm −2 in both the dda-PASEF and dia-PASEF methods.
Protein Identification -FragPipe (17.1), which includes MSFragger (3.4) and EasyPQP (0.1.25)components, was used to build spectral libraries.Peptide identification was performed using MSFragger.Databases of Homo sapiens protein sequences (UP000005640) from UniProt (reviewed sequences only; Apr 2021) and common contaminating proteins, which contained 20,382 total sequences, were used.Inverted protein sequences were added to the original databases.The initial mass tolerance was set at 20 ppm for precursor and fragment ions.Trypsin was set as described above with a maximum of two missed cleavages.Methionine oxidation and N-terminal acetylation were established as variable modifications, and carbamidomethylation was established as a fixed modification.A peptide length range of 7 to 50 amino acids and a peptide mass range of 500 to 5000 Da were set.The maximum number of variable modifications per peptide was fixed at 3. PeptideProphet was used to calculate the probability of correct identification of peptides for spectrum matching and to assemble peptides into proteins.Philosopher Filter was used to assign each identified peptide as a razor peptide to a single protein or protein group that had the greatest peptide evidence.The false discovery rate (FDR) was set to 1% for peptide spectrum match or ion/peptide and protein identification.EasyPQP was used for aligning peptides to a common indexed retention time scale and peptide ion mobility to that from one of the reference runs automatically selected.The final spectral library was filtered at a 1% FDR at the peptide and protein levels.
DIA-NN 1.8 (https://github.com/vdemichev/DiaNN/releases/tag/1.8) was used for dia-PASEF analysis and operated with maximum mass tolerances set to 15 ppm.The samples were analyzed with run-to-run pairing (match between ranks) enabled.Protein inference in DIA-NN was configured to use the assembled proteins in the spectral library."Protein.Group" column in the DIA-NN report was used to identify the protein group, and PG.MaxLFQ label-free quantification was used to obtain the normalized amount.Tables generated by Philosopher were used to identify proteotypic peptides.The quantification mode was set to "Any LC (high precision)".All other settings were left as default.The DIA-NN output was filtered at a q value <1% for precursors and proteins.
The FDR validation was filtered to include only unmodified peptides or peptides with carbamidomethylated cysteines, oxidized methionines, or excised N-termini of methionines.The library was screened for precursors/proteins with a 2 to 4 charge range and a 100.0 to 1700.0 m/z mass range.
As mentioned above, dia-PASEF proteomics analysis was carried out at the Instituto Maimonides de Investigaci ón Biom édica de C órdoba IMIBIC Proteomic Facility.The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org)via the PRIDE partner repository with the dataset identifier PXD043503 (Username: reviewer_pxd043503@ebi.ac.uk;Password: eHWcHRb6).
Data analysis -Perseus (1.6.15.0) was used to analyze the identified proteins.After log2 transformation, data were normalized using the width adjustment method.Proteins with one razor peptide were removed, and missing values were imputed (75% minimum valid in each group).An unpaired two-tailed t test (p-value <0.05) was employed to estimate significant differences.Proteins with a fold change ≥1.3 and ≤0.77 were considered upregulated and downregulated, respectively.Proteins that interact with α-syn were obtained by the BioGRID (https://thebiogrid.org) (4.4.211).The Human Body Fluid Proteome (https://bmbl.bmi.osumc.edu/HBFP)database was used to test whether DEPs that are α-syn interactors were found in human body fluids.Lists of proteins enriched in neurons, microglia, and astroglia were taken from published experimental data (36).SynGO (https://syngoportal.org) (20210225) was employed to determine which DEPs are in the synapse and what role they play in it.IPA (http://www.ingenuity.com) was performed for the analysis of biofunctional pathways and protein-protein interaction networks of DEPs and MAP2 (neuronal protein), AIF1 (microglial protein), GFAP (astroglial protein), and SNCA (α-syn protein).
Immunofluorescence and Western Blotting for Proteomic Validation -Immunofluorescence and Western blot analysis were carried out as previously published (33)(34)(35).
For quantitative validation, Western blot analysis was performed (Table 1).Protein concentration was measured with the bicinchoninic acid protein determination kit (Sigma-Aldrich) and a Multiskan FC microplate photometer (Thermo Fisher Scientific).Twenty micrograms of protein per sample was loaded onto 10% polyacrylamide gels and separated by SDS-PAGE.Transfer to polyvinylidene difluoride membranes was performed with a Trans-blot Turbo transfer system (Bio-Rad).Membranes were blocked with skim milk and incubated with primary antibodies at 4 • C and with secondary antibodies at RT (Table 3).After incubation with enhanced chemiluminescence reagents (Thermo Fisher Scientific), the band intensities were revealed with SyngeneG:BOX (GeneSys software: https://www.syngene.com/software/genesys-rapid-gel-image-capture/) and analyzed with ImageJ (https://imagej.net/ij/)(33,35).GraphPad Prism 6.01 (https:// www.graphpad.com/)was used for statistical analyses.Outliers identified by the ROUT method were removed.Shapiro-Wilk tests (n < 30) were performed to analyze the normality of the sample (p-value >0.05).Statistical comparisons were performed using the unpaired Mann-Whitney test.Data are presented as the mean ± SD.Significant differences were considered when p-value <0.05.

Stereological Analysis
Immunohistochemical Procedures -Tissue processing and immu- nohistochemical staining were carried out following the procedures previously described in our laboratory (34) (Table 2).Briefly, tissue was postfixed in 4% paraformaldehyde and cryoprotected with 2% dimethyl sulfoxide solution and 10% glycerol and, subsequently, with 2% dimethyl sulfoxide solution and 20% glycerol.Samples were cut with a freezing microtome into 50 μm coronal sections.Sections were collected in series: one was Nissl-stained, and the remaining sections were used for immunohistochemical assays.To unmask epitopes, sections were boiled in citrate buffer under pressure.H 2 O 2 (1%) inactivated the activity of endogenous peroxidases.Sections were blocked and incubated at 4 • C with shaking with primary antibodies (Table 3) and subsequently with the secondary biotinylated antibody at RT (Table 3).Avidin-biotin complex (ABC standard, Vectastain) was applied and reacted with 0.025% 3,3 ′ -diaminobenzidine and 0.1% H 2 O 2 .Sections were mounted on slides and covered with DPX (Sigma-Aldrich).
The area fraction occupied by α-syn was analyzed using the area fraction fractionator method (Plan Apochromat 20x/0.5,Ref. 420650-9901).This analysis estimates the percentage of area occupied by α-syn.The optical fractionator method was applied to analyze the density (estimated population using the mean section thickness/ measured volume) of neurons (MAP2), microglia (Iba-1), and astroglia (GFAP) (Plan Apochromatic, 63x/1.4,Oil lens, Ref. 420782-9900).The number of cells counted in the study area was divided by the estimated volume of the region.These quantifications were performed under single-blind conditions.The volume was estimated with a Cavalieri probe using Nissl staining (Plan-Neofluar 1x/0.025,Ref. 420300-9900).More details about these stereological methods were previously described (34,38).
Statistical Analysis -GraphPad Prism (6.01) was used for statistical    To analyze the correlation between the area fraction occupied by α-syn and the duration of PD, Spearman's test was employed.All data are presented as the mean ± SD.Significant differences were considered when p-value <0.05.

Proteomic Analysis
Dia-PASEF proteomic analysis identified 2202 proteins (Supplementary Data 1B), of which 2167 proteins with at least two unique peptides were quantified (Supplementary Data 1C).To facilitate the interpretation and follow-up of the proteomic results, the reader is encouraged to view Figure 1 in conjunction with Figure 2. Statistically significant proteins in both the PD and NPD groups showed different expression patterns (Fig. 2A, Supplementary Data 1D).A total of 199 proteins were DEPs in the ACs of humans with PD (94 upregulated and 105 downregulated) (Figs.1A and 2B, Supplementary Data 1E).Of the DEPs, 25 (16 upregulated and nine downregulated) were identified as SNCA interactors (Figs.1B and 2C, Supplementary Data 1F) and all were found in biofluids such as plasma, serum, cerebrospinal fluid, saliva, urine, etc.Of the total proteins that are composed of the annotated interactome of the SNCA (562 proteins), 35.7% (201 proteins) were quantified in this study.While 12.43% of the quantified proteins that interact with SNCA are DEPs (25 of the 201), only 8.85% of the noninteractome-quantified proteins are differentially regulated (174 of the 1966) (Fig. 1B).Regarding cellular populations, Venn diagram comparisons of the list of 199 DEPs with the published lists of proteins enriched in neurons, microglia, and astrocytes (36) revealed that three DEPs-one upregulated and two downregulatedwere enriched in neurons (Fig. 2D), 11 (four upregulated and seven downregulated) in microglia (Fig. 2E), and 17 (seven upregulated and 10 downregulated) in astrocytes (Fig. 2F).Moreover, one of these typical microglial proteins and two typical astrocytic proteins are part of the α-syn interactome (Fig. 2, E and F).Studying the role of synaptic proteins in PD is crucial given the synaptic location of α-syn.In SynGO, 22.75% (493/2167) of the proteins identified in the proteomic study had annotations for cellular components and biological processes.Of them, 9.33% (46/493) were differentially expressed in the AC in PD (Fig. 1C, Supplementary Data 1G).In addition, upregulated and downregulated DEPs activated or decreased biofunctions in relation to synapses (Fig. 2G, Supplementary Data 1H).The protein-protein interaction network showed that the pathological protein SNCA has several direct and indirect interactions with many of the DEPs and with the neuronal (MAP2), microglial (AIF1, synonymous with Iba-1), and astroglial (GFAP) proteins analyzed by stereology (Fig. 2H, Supplementary Data 2).
In addition, the potential localization of DEPs in neurons, microglia, and astroglia was analyzed.EEF1A1 labeled neurons immunoreactive for MAP2 (Fig. 4, A-C).PLP1 labeled the immunoreactive axons for PAN (Fig. 4, D-F).YWHAH localized especially to the nucleus of neurons immunospecific for MAP2 (Fig. 4, G-I).ORM2 rarely colocalized with GFAP-labeled astroglial cells (Fig. 4, J-L).NPTN and GNAI1 were distributed homogeneously throughout the tissue without localization preferences (data not shown).GRIM19 could not be studied together with other antibodies.

Stereological Analysis
α-Synuclein Area Fraction -The area fraction occupied by α-syn in the Ba, Co, and Ce nuclei of the AC did not show significant differences in the PD group (Fig. 5, see Supplemental Table S1 for stereological data).
Interestingly, the duration of PD correlates positively with the values of the area fraction occupied by α-syn in each nucleus of the AC (Supplemental Fig. S1).
Cell Populations and Volume Estimation -MAP2-, Iba-1-, and GFAP-positive cell density remained stable in the AC and the Ba, Co, and Ce nuclei in the PD and NPD groups (Fig. 6, see Supplemental Tables S2-S4 for stereological data).The volume of the total AC and its nuclei remained unchanged between the PD and NPD groups (Fig. 7, see Supplemental Table S5 for stereological data).

DISCUSSION
As stated, α-syn is the proteinopathy associated with PD constituting Lewy pathology (10), allowing neuropathological staging (4).Lewy pathology progresses in the brain in a prionlike manner (39) involving many prodromal structures until stage 3, when the substantia nigra, which provokes the motor diagnosed symptoms, and the AC are involved (10).The AC is an essential hub in synucleinopathy progression (8); therefore, proteomic analysis has focused on α-syn interactors and synaptic involvement.In parallel, α-syn distribution, neural (40) and glial (41) involvement, and AC volume have been discussed in previous controversial literature.
Proteomic alterations could provide clues as to how the biofunctions of cell populations are altered in PD.To date, most proteomic studies in PD have focused on the substantia nigra (23).However, our study is the first to analyze the proteome of the AC of patients with PD and revealed significant damage at the synaptic level (Figs. 1 and 2).α-Syn is implicated in synapse damage in PD, its interaction with mitochondrial complex I, and its role in synaptic function are areas of ongoing research to better understand the mechanisms underlying PD pathogenesis (42).Proteins with a large fold change and a relationship of interest with cell populations (neurons, microglia, and astroglia) or with the pathological α-syn protein and its potential role in PD were validated as DEPs, which included a total of five upregulated DEPs (GNAI1, EEF1A1, PLP1, NPTN, and YWHAH) and two downregulated DEPs (GRIM19 and ORM2) (Figs. 1 and 3).
The GNAI gene has previously been proposed as a potential target in the diagnosis and treatment of PD and has been reported to be especially upregulated in the striatum and hippocampus of a PD murine model (43).It is a regulator of adenylate cyclase, which plays an important role in the cholinergic synaptic pathway (43) (Fig. 8).Under physiological conditions, the cholinergic pathway promotes the survival of neurons through mitochondrial dysfunction, oxidative stress,  and neuroinflammation and controls dopaminergic activity (44).Therefore, this pathway is notably relevant in PD, and its dysfunction contributes not only to cognitive but also to other nonmotor features and motor impairments (45).The striatum, which is a critical node for balancing dopamine signaling and regulating movement, projects cholinergic projections toward the AC (44).Acetylcholine is a two-unit neurotransmitter of this pathway, linking to two kinds of acetylcholine receptors, G-coupled muscarinic acetylcholine receptors, and ionotropic nicotinic receptors (46).In fact, several studies suggest the clinical usefulness of antimuscarinic drugs for treating PD motor symptoms or inhibitors of the enzyme acetylcholinesterase for the treatment of dementia (46).According to previous studies, GNAI protein was also upregulated in the AC of patients with PD (Fig. 3, A, B, O and P), which may counteract the cholinergic imbalance.
Although the mechanism by which EEF1A1 contributes to PD is not fully understood, it has been proposed as a gene of interest to understand PD given its important role in the maintenance of synaptic plasticity and in the control of the heat shock response (47) (Fig. 8).Our study revealed that EEF1A1 is upregulated in the AC of patients with PD.Qualitative validation showed such a reduction (Fig. 3, C and D), even though quantitative Western blot analysis (Fig. 3, O and P) did not show significant differences between NPD and PD.Our result is consistent with a study in cell cultures demonstrating an increase in EEF1A1 expression after cell exposure to 1-methyl-4-phenylpyridinium (48).According to the data presented, it seems that EEF1A1 could prevent protein, neurite and synapse disruption, and, ultimately, neuronal death (Fig. 8).This agrees with the lack of neurodegeneration reported by stereological results (Fig. 6, A-C).Indeed, colocalization analyses showed that EEF1A1 labeled neurons immunoreactive for MAP2 (Fig. 4, A-C).EEF1A1 could be a potential biomarker of PD, as it has also been shown to be upregulated in platelets from PD patients (49).
Another protein of diagnostic and therapeutic interest is PLP1 (50).PLP1 participates in sphingolipid and glutathione metabolism and in the maintenance of myelin in the central nervous system (50).Homeostasis of membrane sphingolipids in neurons and myelin is critical to inhibit loss of synaptic plasticity, cell death, and neurodegeneration (51).Under physiological conditions, acid sphingomyelinase and acid ceramidase are enzymes involved in the reduction of α-syn production and aggregation.This metabolism is dysregulated in PD, inducing, among others, the transition of the monomeric form of α-syn to an oligomeric toxic form, even the appearance of LBs (51).In addition, dysfunctional sphingolipids can disrupt the autophagy process, which is responsible for the clearance of misfolded proteins (α-syn) and damaged organelles (52).In our study, PLP1 was upregulated (Fig. 3, E,  F, O and P) and colocalized with the axonal marker PAN (Fig. 4, D-F).Most likely, PLP1 is upregulated in PD in order to keep myelin in a nonpathological state that allows rapid longdistance synaptic transmission (Supplemental Fig. S2).In fact, LNs and LBs particularly affect neurons with unmyelinated axons (53), so PLP1 overexpression could be a mechanism to protect against Lewy pathology (Fig. 8).In agreement with our results, a previous study indicated an increase in PLP1 transcripts in the frontal cortex of both mouse and human PD (54).However, in a murine model of PD, the PLP1 protein was downregulated in blood samples (55).
NPTN is part of the presynapse and postsynapse and performs important functions, including the induction of neurite outgrowth and the regulation of synaptic structure and function, synaptic plasticity, and neuronal energy supply (56).Our results indicate NPTN upregulation in the AC of patients with PD (Fig. 3, G, H, O and P), which is consistent with previous studies (Fig. 8) (57,58).Data in rat glial cells have indicated that endoplasmic reticulum stress causes NPTN upregulation, activating the inflammatory response.However, this report also showed that mesencephalic astrocyte-derived neurotrophic factor, an NPTN ligand identified as a protector of dopaminergic neurons, is also upregulated, inhibiting inflammation and preventing cell death (58).
YWHAH, or 14-3-3η, is a member of the 14-3-3 protein family.The 14-3-3 proteins are most abundantly expressed in the brain and are involved in cell signaling, growth, division, adhesion, differentiation, survival, apoptosis, and ion channel regulation (59).Furthermore, the YWHAB, YWHAH, and YWHAE isoforms have been reported to interact with α-syn, and colocalization between 14-3-3 proteins and LBs has been described (60).However, the YWHAH isoform has not been detected in LBs by immunohistochemical assay (61).A study in cell cultures of dopaminergic neurons showed that inducing oxidative stress downregulated the YWHAH gene (59).Our results indicate an increase in YWHAH protein in the AC of patients with PD, which was validated at a qualitative level (Fig. 3, I and J); however, at a quantitative level, it did not show a significant increase in PD cases (Fig. 3, O and P).In addition, colocalization analysis showed its position in the nucleus of MAP2-positive neurons (Fig. 4, G-I).We hypothesized that during PD, YWHAH is upregulated so that its chaperone activity maintains cell survival and controls cell death.YWHAH Mitochondrial dysfunction is an inherent mechanism of PD, contributing in different ways: leading to impaired energy metabolism and consequently to an increase in reactive oxygen species, impairing mitochondrial quality control, and impairing calcium homeostasis, contributing to the loss of dopaminergic neurons in PD (62).Deficiency of GRIM19, a subunit of respiratory electron chain complex 1, has been associated with lower rates of mitochondrial DNA transcription and replication in the substantia nigra in PD (63).The present study confirms the downregulation of GRIM19 in the affected AC during PD.The qualitative validation by immunofluorescence showed a clear reduction (Fig. 3, K and L) that was not significant in the quantitative validation by Western blot (Fig. 3, O and P).Therefore, GRIM19 deficiency in the AC of patients with PD could contribute to mitochondrial dysfunction and consequent oxidative stress (Fig. 8).According to this result, α-syn also interacts with the subunit of respiratory electron chain complex 1 and induces its inhibition (42,62), playing an important role in the pathogenesis of PD.
Another characteristic process of PD is microglia-mediated neuroinflammation, which is a complex network of interactions comprising immune and nonimmune cells, adjunct to mediators of the immune response, contributing to the progression of this disease (64).Among these immunomodulators, ORM2, which is preferentially expressed in the brain and predominantly synthesized by astrocytes, plays an important role (65).However, our study did not show an exacerbated colocalization between ORM2 and GFAP (Fig. 4, J-L), which could be explained by ORM2 secretion.In addition, the present study showed that ORM2 is downregulated in the AC of patients with PD (Fig. 3, M-P).It is possible that astrocytes decrease ORM2 production during PD to stimulate the activity of microglia already exhausted from coping with LBs and LNs (Fig. 8).
Therefore, α-syn can induce toxicity (cell death) in different contexts, namely the reduction of the GRIM19 protein together with blocking complex I in the mitochondria by α-syn contributes to increase reactive oxygen species production and decrease ATP synthesis.In addition, neuroinflammation may potentiate α-syn dysfunction, driving chronic progression of neurodegeneration by inducing astrocytes and microglia reaction, enhanced by ORM2 reduction.However, seeding and spreading, via endocytosis to neighboring neurons, could be slowed down by NPTN, EEF1A1, and YWHAH, which could preserve the homeostasis of the neuron, preventing the neurites disruption and aberrant vesicle trafficking.Other protective mechanisms could be the preservation of myelin (PLP1) and the cholinergic synapse (GNAI) (Fig. 8).
On the other hand, these proteins have a key characteristic that makes them interesting for further research as biomarkers.They have been detected in human body fluids such as cerebrospinal fluid, in the case of GNAI1, EEF1A1, PLP1, NPTN, YWHAH, and ORM2, or urine, in the case of GRIM19, which would facilitate their use as potential biomarkers of interest in PD.
The results of the proteomic analysis indicate that the pathology of AC with PD causes proteomic alterations at the synaptic level.This synaptic damage could go further and lead to alterations in the cell populations that comprise them.To determine the effect that the pathology has on the neuronal and glial populations and on the volume of the AC and its nuclear groups, a stereological study was carried out.
According to the literature, LN and LB inclusions are differentially distributed among amygdaloid nuclei (25,28).A semiquantitative analysis showed that the Co and Ce nuclei were affected earlier and more severely than the basal nucleus (25).The prion-like hypothesis allows us to explain these nuclear differences (66).However, our results indicate that the fraction of area occupied by α-syn does not differ between AC nuclei (Fig. 5), which is in consonance with a previous study by our group (26).The advanced stages of the cases used could explain this pathological homogeneity, as the entire AC is seriously affected in advanced stages of PD.Additionally, there is a positive correlation between the duration of PD and the area fraction occupied by α-syn (Supplemental Fig. S1).
Another semiquantitative study indicated that the percentage of neurons (marked with Nissl staining) with LBs is higher in the cortical and basolateral nuclei (28) and related with the death of the neuronal population (28).Nevertheless, our results revealed that the neuronal population of the AC and its nuclei was not altered in PD (Fig. 6, A-C).In addition to stereological techniques, the use of neuron-specific labeling minimizes counting errors compared to Nissl staining (67).As previously suggested, some cells can survive LN and LB impairment for a long time but probably cease functioning long before dying (53).
The presence of astrocytes with α-syn aggregates in the AC of patients with PD has been described (14,15,41) as well as the activation of microglia in PD.However, this is the first study to analyze microglial (Fig. 6, D-F) and astroglial (Fig. 6, G-I) density in the AC of patients with PD, reveling that remain stable in the AC and its nuclei.Whether microglia and astroglia favor the spread of Lewy pathology or, in contrast, eliminate or retain LNs and LBs, avoiding neuronal involvement is still unknown (68)(69)(70)(71).
Potential changes in cell populations (gray matter) and/or their connections (white matter) can alter the volume measurable by imaging techniques, which can be used to analyze the total volume of the AC, but stereological approaches are more costly and allow the delimitation of AC nuclear complexes.A single stereological study analyzed the AC volume of patients with PD and found AC atrophy due to volume reduction in the corticomedial complex, in which higher neuronal loss has been reported (28).Our results indicated that the volumes of the AC and its nuclei did not vary between PD and NPD cases (Fig. 7).We consider that although the damage caused by LNs and LBs is not sufficient to lead to the loss of gray and/or white matter, it could damage cell functionality, as revealed by proteomic data.

CONCLUSIONS
To our knowledge, this is the first proteomic study of the AC of patients with PD, and it identified possible biomarkers of the disease.Considering the data reported here, it seems clear that Lewy pathology is not sufficient to cause neurodegeneration or alteration of microglial and astroglial populations in the AC of humans with PD.However, damage at the proteomic level is evident, highlighting significant synaptic involvement.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE
All experiments were performed in accordance with the Clinical Research Ethics Committee of the University Hospital of Ciudad Real (PID2019-108659RB-I00).
Other data generated or analyzed during this study are included in this published article and its supplementary information files are available from the corresponding author upon reasonable request.
DIANN does not generate annotated spectrum files but.tsv files, which are results files.Alphamap (https://github.com/MannLabs/alphamap) allows you to visualize the sequences and peptides identified by DIANN for each identified protein.AlphaViz (https://github.com/MannLabs/alphaviz) is an automated visualization pipeline to link these identifications with the original raw data and easily assess their individual quality or the overall quality whole samples.
Supplemental data -This article contains supplemental data.

FIG. 1 . 9 FIG. 2 .
FIG. 1. Summary of proteomic results.The diagram shows the different bioinformatics analyses carried out, differentially expressed proteins (DEPs) obtained, and those that have passed the criteria established for their validation.A, Dia-PASEF proteomic analysis identified 2202 proteins, of which 2167 were quantified.One hundred ninety-nine differentially expressed proteins (94 upregulated and 105 downregulated) were obtained using a p-value <0.05 and a fold change ≥1.3 and ≤0.7.B, of the total proteins that are composed of the annotated interactome of the SNCA (562 proteins), 35.7% (201/562) were quantified in this study, while 12.43% of the quantified proteins that interact with SNCA are DEPs (25/201).C, 22.7% (493/2167) of the quantified proteins comprised synaptic cellular components (CC) and synaptic biological processes (BP), and 9.3% (46/493) of these were DEPs.D, five upregulated (red) DEPs (GNAI1, EEF1A1, PLP1, NPTN, YWHAY) and two downregulated (green) DEPs (GRIM19, ORM2) were selected for the study according to the criteria established for it.The few matching DEPs of the annotated interactome of the SNCA and comprised synaptic CC and synaptic BP did not meet the criteria for selection.dia-PASEF, data-independent acquisition-parallel accumulation serial fragmentation.

FIG. 4 .
FIG. 4. Localization of differentially expressed proteins in stereological cell types.EEF1A1 (A and C) labels neurons as does the neuronalspecific antibody MAP2 (B and C).PLP1 (D and F) colocalizes with the axonal marker PAN (E and F).YWHAH (G and I) is especially localized in the nucleus of MAP2-tagged neurons (H and I).ORM2 (J and L) barely colocalizes with GFAP (K and L).Dashed lines delineate neuronal somas.Arrows point to colocalizations.Scale bar represents 20 μm.

TABLE 1
Demographic, clinicopathological features, and assay of the cases used for proteomic analysis

TABLE 3
Primary and secondary antibodies used

TABLE 3 -
Continued using unpaired two-tailed t tests to estimate significant differences in age, brain weight, volume, and MAP2, Iba-1, and GFAP densities.The Kruskal-Wallis test was used to calculate the significance of the area fraction occupied by α-syn. performed