Differences in Ab species in EOAD and DS
Amyloid plaques in DS and EOAD had similar amounts of total Ab, Ab40 and Ab42 (Figure 2A). The size of amyloid plaques was similar in DS and EOAD. However, amyloid plaques in DS had significantly higher amounts of both phosphorylated Ab and pyroglutamate Ab than EOAD cases (Figure 2B). Phosphorylated Ab immunoreactivity was observed both in plaques and in neurons in DS and EOAD. Two main types of intraneuronal staining were observed: staining consistent with presence in neurofibrillary tangles and neurons containing large puncta of phosphorylated Ab. Phosphorylated Ab was also observed in dystrophic neurites. While there were significantly increased levels of phosphorylated Ab in plaques in DS in comparison to EOAD, similar levels of intraneuronal phosphorylated Ab were observed in DS and EOAD. Pyroglutamate Ab was observed in amyloid plaques in both DS and EOAD. Significantly more pyroglutamate Ab was observed in DS in comparison to EOAD (Figure 2B).
Oligomers were visualized using the pan-oligomeric antibody TWF9, which is a conformational antibody that recognizes Ab oligomers in addition to other beta sheet containing oligomers [35]. Consistent with previous studies, TWF9 immunoreactivity was observed in neuronal soma. No immunoreactivity was observed within plaques. DS cases had significantly lower levels of TWF9 immunoreactivity in comparison to EOAD (Figure 2B).
Proteomic analysis of EOAD and DS amyloid plaques
Proteomic analysis of plaques and neighboring non-plaque tissue identified 2259 proteins (Supplementary Table 1). 85% of proteins (1915 proteins) were identified in both EOAD and DS samples, of which 1355 proteins were identified in all 20 samples, therefore confirming that our proteomic approach is a reliable way to quantify amyloid plaque proteins using microscopic amounts of formalin-fixed paraffin embedded human tissue samples. Proteins present in all 20 samples included major AD-associated proteins such as Ab, Tau and ApoE, therefore confirming the presence of these proteins both inside plaques and in surrounding non-plaque tissue.
Proteins enriched in plaques in both EOAD and DS
The main aim of this study was to identify proteins that were enriched in amyloid plaques in EOAD and DS in comparison to surrounding non-plaque tissue. 127 proteins were significantly enriched in amyloid plaques in either EOAD or DS (Supplementary Table 1). 48 proteins were consistently enriched in both DS and EOAD plaques (Table 2, Figure 3). Systematic literature searches revealed that 33/48 proteins have been previously confirmed as amyloid plaque proteins in late-onset AD, therefore validating our mass spectrometry approach and providing new evidence that similar proteins are enriched in amyloid plaques in different subtypes of AD (Table 2). In addition, we identified 15 proteins that were enriched in plaques in both EOAD and DS (Table 2) that were not previously known to be plaque associated proteins. Four of these proteins have been previously associated with either Ab or APP. Here, we provide the first evidence that these proteins are enriched in amyloid plaques. The remaining 11 proteins are amyloid plaque proteins that have not been previously associated with Ab, APP or amyloid plaques in any subtype of AD (Table 2).
As expected, Ab was highly enriched in plaques in comparison to the surrounding non-plaque tissue (12 and 7 fold enriched in EOAD and DS plaques respectively; Figure 3B). In contrast, while tau was abundant in both plaques and neighboring non-plaque tissue in DS and EOAD, there was no evidence of enrichment of tau in amyloid plaques. Examination of the abundance (overall intensity in plaques) of the 48 proteins enriched in both EOAD and DS showed that the most abundant proteins present were well-known plaque proteins (e.g. APP, ApoE, vimentin, clusterin, complement C3 and complement C4a; Figure 3C). We also observed a very high correlation in the total concentration of these proteins in plaques between EOAD and DS (Figure 3C). The most abundant novel plaque protein in both DS and EOAD was ezrin (EZR), which was one of the proteins selected for immunohistochemistry validation studies (Figure 3C).
Examination of the proteins that had the highest enrichment in plaques in both DS and EOAD included many proteins less studied in the AD field (Table 2; Supplementary Table 3). For example, COL25A1 was the most highly enriched protein in plaques in both EOAD and DS (104 and 113-fold enriched respectively). Other highly enriched plaque proteins in both EOAD and DS included MDK, NTN1, HTRA1, SMOC1 and OLFML3 (Figures 4A, 4B). The 48 proteins consistently enriched in plaques in both EOAD and DS also showed a highly significant degree of protein-protein interaction (p<1.0x10-16; Figure 3D) and were almost exclusively classified as either vesicle (enrichment FDR: 4.32x10-9) or extracellular proteins (enrichment FDR: 3.34x10-8). The enrichment of vesicle proteins was predominantly driven by endosome or lysosome proteins (Figure 3D; Supplementary Table 3). Synapse proteins were also particularly enriched (enrichment FDR: 1.90x10-3).
Table 2: 48 proteins consistently enriched in plaques in EOAD and DS. Proteins listed in order of fold change enrichment in EOAD; separated into previously confirmed plaque proteins, associated with Ab or APP, and novel. “Previously confirmed plaque proteins” were determined by published immunohistochemistry evidence of protein presence in plaque or by >1.5 fold enrichment in plaque in comparison to neighboring non-plaque tissue in late onset AD or preclinical AD [44]. Difference in AD tissue was determined by comparison with 33 previous proteomic studies of human AD brain tissue. “Mediates Ab pathology?” determined by literature searches for “Alzheimer’s disease and gene ID or protein name”. Protein was designated as mediating Ab pathology if altering protein expression in transgenic animal models or cell culture affected amyloid pathology.
Uniprot
|
Gene
|
Protein
|
Enrichment in EOAD plaques (fold change)
|
Enrichment in DS plaques (fold change)
|
Known plaque protein?
|
Difference in AD brain tissue
|
Mediates Ab pathology?
|
Previously Confirmed Plaque Proteins – Immunohistochemistry
|
Q9BXS0
|
COL25A1
|
Collagen alpha-1
|
104.3
|
113.1
|
Yes [74]
|
Increased
|
Increases pathology [6, 75]
|
O95631
|
NTN1
|
Netrin-1
|
34.9
|
58.7
|
Yes [60]
|
Increased
|
Decreases pathology [76]
|
P21741
|
MDK
|
Midkine
|
31.4
|
70.4
|
Yes [77]
|
Increased
|
Decreases pathology [78]
|
Q92743
|
HTRA1
|
Serine protease HTRA1
|
19.0
|
42.8
|
Yes [79]
|
Increased
|
Decreases pathology [80]
|
Q9H4F8
|
SMOC1
|
SPARC-related modular calcium-binding protein 1
|
12.9
|
58.8
|
Yes [60]
|
Increased
|
Unknown
|
P02649
|
APOE
|
Apolipoprotein E
|
10.4
|
17.2
|
Yes [81]
|
Increased
|
Increases pathology [82, 83]
|
Q14956
|
GPNMB
|
Transmembrane glycoprotein NMB
|
7.8
|
17.8
|
Yes (in plaque-associated microglia) [84]
|
Increased
|
Unknown
|
P0C0L4
|
C4A
|
Complement C4-A
|
7.5
|
10.1
|
Yes [85]
|
Increased
|
Unknown
|
P35052
|
GPC1
|
Glypican-1
|
7.5
|
8.5
|
Yes [86]
|
Decreased
|
Increases pathology [87]
|
P02743
|
APCS
|
Serum amyloid P-component
|
4.9
|
10.8
|
Yes [88]
|
Increased
|
Increases pathology [89]
|
Q9UIK5
|
TMEFF2
|
Tomoregulin-2
|
4.8
|
6.9
|
Yes [90]
|
n/a
|
Decreases pathology [91]
|
P02746
|
C1QB
|
Complement C1q subcomponent subunit B
|
3.3
|
4.3
|
Yes [92]
|
n/a
|
Increases pathology [93, 94]
|
P10909
|
CLU
|
Clusterin
|
3.2
|
4.0
|
Yes [95]
|
Increased
|
Increases pathology [7, 96]
|
Q00604
|
NDP
|
Norrin
|
2.9
|
4.6
|
Yes [79]
|
Increased
|
Unknown
|
P05067
|
APP
|
Amyloid-beta precursor protein
|
2.8
|
5.9
|
Yes [97]
|
Increased
|
Increases pathology [98]
|
P02747
|
C1QC
|
Complement C1q subcomponent subunit C
|
2.7
|
8.4
|
Yes [92]
|
Increased
|
Increases pathology [93, 94]
|
P01024
|
C3
|
Complement C3
|
2.5
|
2.9
|
Yes [92]
|
Increased
|
Increases pathology [94, 99, 100]
|
P41222
|
PTGDS
|
Prostaglandin-H2 D-isomerase
|
2.2
|
3.0
|
Yes [101]
|
Increased
|
Decreases pathology [101]
|
P26038
|
MSN
|
Moesin
|
2.1
|
2.6
|
Yes, in plaque-associated microglia [102]
|
Increased
|
Decreases pathology Darmellah et al., 2012)
|
P07093
|
SERPINE2
|
Glia-derived nexin
|
2.1
|
4.3
|
Yes [103]
|
Decreased
|
Increases pathology [104, 105]
|
Q9UBP4
|
DKK3
|
Dickkopf-related protein 3
|
2.1
|
1.8
|
Yes [106]
|
Increased
|
Decreases pathology [107]
|
Q8IV08
|
PLD3
|
Phospholipase D3
|
2.0
|
2.0
|
Yes [108]
|
n/a
|
Decreases pathology [109, 110]
|
O00468
|
AGRN
|
Agrin
|
1.9
|
2.9
|
Yes [111]
|
Increased
|
Decreases pathology [112]
|
Q07954
|
LRP1
|
Prolow-density lipoprotein receptor-related protein 1
|
1.8
|
2.1
|
Yes [113]
|
Increased
|
Inconsistent effects on pathology [114]
|
P08670
|
VIM
|
Vimentin
|
1.7
|
1.8
|
Yes, in surrounding astrocytes [115]
|
Increased
|
Increases pathology [116]
|
P16870
|
CPE
|
Carboxypeptidase E
|
1.6
|
2.1
|
Yes [117]
|
Increased
|
Unknown
|
Q15818
|
NPTX1
|
Neuronal pentraxin-1
|
1.6
|
1.7
|
Yes [118]
|
Increased
|
Increases pathology [119]
|
Previously Confirmed Plaque Protein - Proteomics
|
Q9NRN5
|
OLFML3
|
Olfactomedin-like protein 3
|
19.2
|
18.9
|
Yes [44]
|
Increased
|
Unknown
|
Q9HCB6
|
SPON1
|
Spondin-1
|
6.9
|
16.5
|
Yes [44]
|
n/a
|
Decreases pathology [120, 121]
|
O94985
|
CLSTN1
|
Calsyntenin-1
|
5.4
|
8.1
|
Yes [44]
|
Decreased
|
Increases pathology [122]
|
Q9ULB1
|
NRXN1
|
Neurexin-1
|
2.9
|
2.8
|
Yes [44]
|
Increased
|
Unknown
|
P51797
|
CLCN6
|
Chloride transport protein 6
|
2.8
|
9.7
|
Yes [44]
|
Increased
|
Unknown
|
Q9NVJ2
|
ARL8B
|
ADP-ribosylation factor-like protein 8B
|
2.2
|
2.9
|
Yes [44]
|
Increased
|
Decreases pathology [123]
|
Novel Plaque Proteins – Mechanistic Link with Aß or APP
|
O75110
|
ATP9A
|
Probable phospholipid-transporting ATPase IIA
|
1.8
|
2.3
|
No, but associated with Aß [124]
|
Increased
|
Increases pathology [124]
|
P15311
|
EZR
|
Ezrin
|
1.7
|
2.6
|
No, but associated with APP [125]
|
Increased
|
Decreases pathology [125]
|
O00299
|
CLIC1
|
Chloride intracellular channel protein 1
|
1.6
|
1.7
|
No, but associated with Aß [126]
|
Increased
|
Increases pathology [126]
|
O14773
|
TPP1
|
Tripeptidyl-peptidase 1
|
1.6
|
2.1
|
No, but associated with Aß [127]
|
Increased
|
Decreases pathology [127]
|
Novel Plaque Proteins- No Previous Association with Aß or APP
|
P51809
|
VAMP7
|
Vesicle-associated membrane protein 7
|
3.0
|
4.0
|
No
|
n/a
|
Unknown
|
Q9UNK0
|
STX8
|
Syntaxin-8
|
3.2
|
2.4
|
No
|
Increased
|
Unknown
|
Q5TH69
|
ARFGEF3
|
Brefeldin A-inhibited guanine nucleotide-exchange protein 3
|
3.2
|
5.2
|
No
|
Increased
|
Unknown
|
Q6IAA8
|
LAMTOR1
|
Ragulator complex protein LAMTOR1
|
2.6
|
2.9
|
No
|
n/a
|
Unknown
|
Q59EK9
|
RUNDC3A
|
RUN domain-containing protein 3A
|
2.3
|
5.6
|
No
|
n/a
|
Unknown
|
P40121
|
CAPG
|
Macrophage-capping protein
|
2.2
|
1.9
|
No
|
Increased
|
Unknown
|
Q9NQ79
|
CRTAC1
|
Cartilage acidic protein 1
|
2.1
|
2.2
|
No
|
n/a
|
Unknown
|
Q9P2S2
|
NRXN2
|
Neurexin-2
|
1.9
|
2.5
|
No
|
n/a
|
Unknown
|
Q99435
|
NELL2
|
Protein kinase C-binding protein NELL2
|
1.8
|
3.9
|
No
|
n/a
|
Unknown
|
Q9HB90
|
RRAGC
|
Ras-related GTP-binding protein C
|
1.9
|
2.2
|
No
|
n/a
|
Unknown
|
Q86Y82
|
STX12
|
Syntaxin-12
|
1.5
|
2.0
|
No
|
n/a
|
Unknown
|
Differences in plaque enriched proteins in EOAD and DS
Our results suggest that that major plaque enriched proteins in EOAD and DS were largely the same. The consistency of protein enrichment in plaques was even noted at an individual case level (Figure 4C, 4D). However, we were interested to determine whether there was evidence of plaque protein enrichment that was unique to either DS or EOAD beyond these common plaque-enriched proteins. 20 proteins were uniquely enriched in plaques in EOAD (Supplementary Table 4) and 59 proteins were uniquely enriched in plaques in DS (Supplementary Table 5). Pathway analysis of proteins that were uniquely enriched in plaques in either DS or EOAD showed that these proteins were also enriched in endosomal or lysosomal proteins, similar to the commonly enriched plaque proteins. These protein differences between DS and EOAD did not suggest the presence of unique disease mechanisms driving plaque development in DS or EOAD: pathway analysis showed that these proteins did not cluster to a particular functional pathway and the majority of proteins showed the same trend for enrichment in plaques in the other group. 80% (63/79 proteins) of proteins uniquely enriched in plaques in either EOAD or DS were still increased in plaques in the other subtype of AD, albeit at a level that did not meet our criteria for ‘enrichment in plaques’. Therefore, these results suggest that largely the same proteins are enriched in amyloid plaques in EOAD and DS.
We also directly compared plaque protein levels in DS and EOAD. For this analysis, plaque protein levels that were normalized against background protein levels for each individual case were used. 38 proteins were significantly different between DS and EOAD plaques after correction for background protein differences. 25 proteins were significantly higher and 13 proteins were significantly lower in DS plaques in comparison to EOAD plaques (Supplementary Table 7). Pathway analysis did not highlight enrichment of any cellular compartments or pathways for significantly different proteins in DS and EOAD plaques. Again, suggesting that plaque protein composition was largely the same in DS and EOAD.
We also examined if the triplication of chromosome 21 in DS resulted in any major differences in plaque associated proteins. Our proteomic analysis identified 22 proteins with genes on chromosome 21 (Supplementary Table 6). Of these, only three proteins were enriched in plaques in DS: APP, ITGB2 and COL18A1. APP was commonly enriched in plaques in both EOAD and DS. While ITGB2 and COL18A1 both had higher levels in plaques in comparison to non-plaques in EOAD, their level did not meet our criteria for designation as “enriched”. Therefore, our results suggest that the triplication of chromosome 21 is not necessarily associated with enrichment of those gene products in plaques, but rather may enhance the enrichment of selected proteins in plaques.
Validation: Comparison with Previous Proteomic Studies
Only one prior study has examined the proteome of amyloid plaques in comparison to surrounding non-plaque tissue [44]. This study identified proteins that were enriched in amyloid plaques in late-onset AD and preclinical AD. Despite the power of their dataset being limited by a small sample size (n=3 cases/group, pooled prior to mass spectrometry) and the different subtypes of AD analyzed in their study in comparison to ours, we were pleased to see that many of our plaque enriched proteins were validated in this previous study. 43 proteins were identified in both our study and enriched in late-onset AD plaques in Xiong et al. 26/43 commonly identified proteins were significantly enriched in either DS or EOAD plaques (Supplementary Table 1). The majority of the remaining proteins were also increased in plaques in our study, however they did not reach the criteria for significance in our study. All of the top 10 most highly enriched proteins in plaques in DS and EOAD in our study were also enriched in plaques in late onset AD (Figure 5).
Xiong et al. also identified 78 proteins that were enriched in plaques in preclinical AD. 53 of these proteins were also identified in our study, of which, 30 were enriched in DS or EOAD plaques in our study (Supplementary Table 1). The most notable protein that was not enriched in preclinical AD plaques was COL25A1, which was the most highly enriched protein in both DS and EOAD plaques in our study and was enriched in late-onset AD plaques in Xiong et al., (2019). This suggests that COL25A1 may only become enriched in plaques at a later stage of disease development. In contrast, the remaining top 10 most enriched proteins for both DS and EOAD were also enriched in plaques in preclinical AD (Figure 5), suggesting that plaques in preclinical AD largely contain the same proteins present in plaques at advanced stages of AD.
We also compared our data to Bai et al. (2020) [60] who identified 28 proteins that correlated with Ab abundance in human brain tissue throughout the progression of AD. 20 of these proteins were also identified in our study, of which 13 were significantly enriched in DS and/or EOAD plaques (Supplementary Table 1). The remaining 7 proteins were also increased in DS and/or EOAD plaques, however these did not reach our statistical stringency level to be considered a plaque-enriched protein.
The combined analysis of our data with these two previous studies identified 30 proteins that were consistently enriched in plaques or correlated with Ab in at least 3 analyses (Figure 5). This group of proteins represent a consistent amyloid plaque signature highlighting proteins that likely have an important role in amyloid plaques in addition to Ab. While the some of these proteins are well known plaque proteins (e.g. APP, ApoE, clusterin), the role of many of these proteins in AD is comparatively much less studied including 8 proteins that have only been discovered as an amyloid protein in proteomic studies (OLFML3, SPON1, CLSTN1, NRXN1, CLCN6, ARL8B, SYT11, SCIN). Combined, these comparisons with previous studies validates our findings and provides additional evidence that amyloid plaques are enriched in many proteins in addition to Ab, many of which are likely to be of pathological importance in AD and merit further investigation.
Validation: Immunohistochemistry
Fluorescent immunohistochemistry was used to validate the enrichment of four proteins in amyloid plaques. Ezrin (EZR) was selected as it was the most abundant novel plaque protein identified in our study. ARL8B was selected as a representative plaque-enriched lysosomal protein that had no prior immunohistochemistry evidence of presence in amyloid plaques. Moesin (MSN) and SMOC1 were selected as both have only one prior publication confirming their presence in plaques using immunohistochemistry, but no immunohistochemistry evidence of enrichment in plaques in EOAD or DS. Fluorescent immunohistochemistry confirmed that ezrin, moesin, ARL8B and SMOC1 were enriched in amyloid plaques in comparison to surrounding non-plaque tissue in DS, EOAD and late-onset sporadic AD. Moesin (Figure 6), Ezrin (Figure 7), and SMOC1 (Figure 8) strongly co-localized with Ab in amyloid plaques. Particularly intense immunoreactivity was observed in the aggregated core of dense-cored plaques for these proteins. Moesin was also observed in cells with a microglial morphology in both AD and control cases, consistent with a previous study that confirmed that moesin is a microglial protein [102].
SMOC1 strongly co-localized with amyloid fibrils only in a subset of amyloid plaques (Figure 8). The proportion of SMOC1 immunoreactive plaques in the hippocampus varied between subtypes of AD; SMOC1 was present in 58% amyloid plaques in DS in comparison to 47% of plaques in EOAD and 32% of plaques in late-onset AD (Figure 8A-B). This was consistent with our proteomic results that found a greater enrichment of SMOC1 in DS plaques in comparison to EOAD plaques. Both neuritic and diffuse plaques showed SMOC1 immunoreactivity (Figure 8C). Qualitatively, the proportion of SMOC1 immunoreactive plaques was higher in the hippocampus than in the neighboring cortex in all subtypes of AD. Interestingly, there was a large amount of colocalization of SMOC1 with plaques that also contained post-translationally modified Ab species (white arrows, Figure 8D). Minimal basal SMOC1 staining was observed in age-matched control cases, with the most consistent basal SMOC1 expression being in localized pockets of the choroid plexus.
ARL8B was also abundant in amyloid plaques in all subgroups (Figure 9). In contrast to SMOC1, the proportion of ARL8B immunoreactive plaques in the hippocampus was similar in DS and EOAD (77% and 79%, respectively; Figure 9A-B). However, a significantly lower proportion of plaques contained ARL8B in late-onset AD in comparison to EOAD (Figure 9A-B). Two distinct patterns of plaque-associated ARL8B staining were observed. In one subset of amyloid plaques, bright puncta of ARL8B were diffusely present throughout plaques (Figure 9C). In these plaques, ARL8B did not strongly colocalize with Ab. Instead, ARL8B was often observed in the regions of amyloid plaques that were not brightly stained for Ab (Figure 9C). Qualitatively, ARL8B colocalization in amyloid plaques was more commonly observed in the hippocampus than the cortex. Basal ARL8B staining in control hippocampal sections was observed in neuron soma throughout the cytoplasm and occasionally in primary processes (Figure 9C). Staining was particularly bright in hippocampal pyramidal neurons. Abundant ARL8B was also observed in granule cells in the dentate gyrus, in the choroid plexus, and in the nucleus of some cells in white matter. The same pattern of basal staining was observed in controls and all subtypes of AD. In the second subset of amyloid plaques, intense ARL8B immunoreactivity was observed in specific plaque-associated cells (Figure 9D). These cells were located at the periphery of plaques and had bright, punctate ARL8B throughout the cell cytoplasm and primary processes (Figure 9D) and had morphology consistent with reactive glia. Double fluorescent immunohistochemistry against ARL8B and MAP2, IBA1, or GFAP showed that these ARL8B positive plaque-associated cells were a subset of reactive plaque associated astrocytes.
We also validated the presence or absence of these four plaque protein in vascular amyloid pathology. MSN, EZR and SMOC1 immunoreactivity occasionally co-localized with CAA or in plaques which were in direct contact with blood vessels. However, ARL8B immunoreactivity was absent in vascular amyloid pathology, which is consistent with its weak direct colocalization with Ab in amyloid plaques (Figure 10).