none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10A27106_B1_Br3854
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10A27106_C1_Br3873
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10A27106_D1_Br3880
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10T31036_B1_Br3854
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10T31036_C1_Br3873
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10T31036_D1_Br3880
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10A27004_D1_Br3880
0.0
2.5
5.0
7.5
10.0
V10A27106_A1_Br387402%3
0
1
2
3
4
5
V10T31036_A1_Br387402%3
0
1
2
3
V10A27004_A1_Br387402%3
0
50
100
150
0
10
20
30
V10A27106_A1_Br3874 SNAP25
0
10
20
30
10T31036_A1_Br3874 SNAP25 10A27004_A1_Br3874 SNAP25
Spatial Transcriptomics Analysis of Aβ-tau Synergy in the Inferior Temporal Cortex of
the Human Brain in Alzheimer’s Disease
SH Kwon1, M Tippani1, S Parthiban2, AB Spangler1, HR Divecha1, KD Montgomery1, C Bruce3, S Williams3, M Mak3, G Yu3, J Avalos-Gracia3, TM Hyde1, SC
Page1, SC Hicks2, K Martinowich1, KR Maynard1,*, L Collado-Torres1,* *:
[email protected] [email protected]
1: Lieber Institute for Brain Development, Baltimore, MD. 2: Department of Biostatistics, JHBPSH. 3: 10x Genomics, Inc. Pleasanton, CA.
INTRODUCTION
VISIUM-IF AD STUDY DESIGN
PATHOLOGY DATA FROM VISIUM-IF
CONCLUSIONS
REFERENCES
IDENTIFYING WHITE/GRAY MATTER
Background: Alzheimer’s disease (AD) is a neurodegenerative disorder characterized by senile
plaques of β-amyloid (Aβ) and neurofibrillary tangles of phosphorylated tau (pTau).
Accumulating evidence suggests that Aβ and tau act in concert to orchestrate complex
pathobiological events underlying AD, but their additive effect remains poorly understood. We
investigated Aβ-tau synergy at the level of gene expression in the inferior temporal cortex
(ITC). 10x Genomics Visium ImmunoFluorescence (Visium-IF) is a new platform in which IF
staining is paired with on-slide cDNA synthesis using spatially-barcoded arrays.
Methods: We employed Visium-IF to generate a proteomic-based, spatially-resolved,
transcriptomic-scale map of the human ITC in AD. ITC brain blocks were dissected from 3 brain
donors with late-onset AD (Amyloid C/Braak III) and 1 age-matched neurotypical donor. We
validated laminar patterning in each block using RNAscope in situ hybridization with Layer I
and VI marker genes (RELN and NR4A2). Cryosections from each block were collected in
triplicate on Visium arrays and immunostained for Aβ, pTau, MAP2 and GFAP to visualize
plaques, tangles, neurites, and astroglial processes. Each section was scanned using
multispectral imaging and sequenced: whole-genome and targeted gene expression using the
10X Genomics human neuroscience panel.
Results: After quality-control, we constructed SpatialExperiment R objects with 38,115 spots
across 27,853 genes, one for each expression panel. We segmented the IF images using
VistoSeg and used these metrics to categorize spots into seven groups: 1) no pathology, 2) Aβ
only, 3) pTau only, 4) Aβ/pTau, 5) Aβ adjacent, 6) pTau adjacent, 7) Aβ/pTtau adjacent. We
identified spatially-aware gene expression k clusters using BayesSpace across all samples, with
a k range from 2 to 28. Using limma we identified differentially expressed genes between the
pathology groups in the gray matter of the cortex only. We further spatially-registered the
BayesSpace clusters to the cortex layers and created an interactive website using spatialLIBD.
PATHOLOGY SPOTS LABELING
GENES ASSOCIATED WITH AD PATHOLOGY
• Our experiments and data analysis strategies can enable us to identify
molecular, cellular, and morphological associations with spatially-
localized Aβ and tau pathology. This project motivated the development
of software for Visium-IF analysis.
• Spatial clusters between whole genome and targeted sequencing panel
are highly concordant.
Join us! https://www.libd.org/careers/
• Research Associate: masters or undergrad with 3 years of experience
• Staff Scientist: PhD or masters with 5 years of experience
https://lcolladotor.github.io/bioc_team_ds/
Questions? @lcolladotor or
[email protected]
• VisiumIF https://www.10xgenomics.com/products/spatial-proteogenomics
• Targeted sequencing panel https://www.10xgenomics.com/products/targeted-gene-
expression
• SpatialExperiment https://doi.org/10.1093/bioinformatics/btac299
• VistoSeg https://doi.org/10.1101/2021.08.04.452489
• BayesSpace https://doi.org/10.1038/s41587-021-00935-2
• limma https://doi.org/10.1093/nar/gkv007
• DLPFC Visium data and layer registration https://doi.org/10.1038/s41593-020-00787-0
• spatialLIBD https://doi.org/10.1101/2021.04.29.440149
VISIUM IMMUNOFLUORESCENCE (IF)
ACKNOWLEDGEMENTS
Sang Ho Kwon
@sanghokwon17
Madhavi Tippani
@MadhaviTippani
% pTau/spot % Abeta/spot BayesSpace k=28
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10A27106_A1_Br3874
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10T31036_A1_Br3874
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
V10A27004_A1_Br3874
PATHOLOGY DIFFERENCES IN GRAY MATTER
P1
P7
Pathology
Pathology Groups
Sowmya Parthiban
@sowmyapartybun
Case AgeDeath Race RIN Braak CERAD
Neurotypical
Br3874
73 EUR/CAUC 7.2 B2 C0
AD #1
Br3854
65 EUR/CAUC 7.0 B3 C3
AD #2
Br3873
88 EUR/CAUC 7.2 B3 C3
AD #3
Br3880
90 EUR/CAUC 7.1 B3 C3
−1
−0.9
−0.8
−0.7
−0.6
−0.5
−0.4
−0.3
−0.2
−0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
WM
L6
L5
L4
L3
L2
L1
11
2
17
13
28
25
27
10
20
21
15
5
6
7
1
16
9
24
3
4
8
14
12
26
DLPFC Visium Data
Maynard et al 2021
DLPFC LAYERS REGISTRATION
Whole genome
Targeted sequencing
640
155
49
6
55
17
709
15
438
43
156
309
7
709 104
581
11
1193
49
2
3 1
12
277
631
73
73
57
283
1776
21
666
68
148
3
301
20
376 46
30
10
44
14
1757
37
831
67
132
4
137
19
1448
102
279
21
883
73
4
11
6
219
524
15
36
17
141
2450
54
712
136
117
1
121 31
205
899
45
39
35
141
2112
22
556
101
173
1
216
13
S1_B1_3854 S1_C1_3873 S1_D1_3880 S2_B1_3854 S2_C1_3873 S2_D1_3880 S3_D1_3880
0.5 1.0 1.5 2.0 2.5
0.5 1.0 1.5 2.0 2.5
0.5 1.0 1.5 2.0 2.5
0.5 1.0 1.5 2.0 2.5
0.5 1.0 1.5 2.0 2.5
0.5 1.0 1.5 2.0 2.5
0.5 1.0 1.5 2.0 2.5
0%
25%
50%
75%
100%
Percentage
path_groups
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
BayesSpace k = 2 1: gray matter 2: white matter
−1
−0.9
−0.8
−0.7
−0.6
−0.5
−0.4
−0.3
−0.2
−0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
WM
L6
L5
L4
L3
L2
L1
1
2
k =2
1: GM, 2: WM
BayesSpace clusters
1
2
V10A27106_A1_Br3874
1
2
V10T31036_A1_Br3874
1
2
V10A27004_A1_Br3874
1
2
V10A27106_B1_Br3854
1
2
V10A27106_C1_Br3873
1
2
V10A27106_D1_Br3880
1
2
V10T31036_B1_Br3854
1
2
V10T31036_C1_Br3873
1
2
V10T31036_D1_Br3880
1
2
V10A27004_D1_Br3880
Prop IF/Spot
Prop pTau/spot Prop Aβ/spot
−20
−10
0
10
20
−20 0 20 40 60 80
runPCA 01 (42%)
runPCA 02 (10%)
path_groups
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
−20
−10
0
10
20
−20 0 20 40 60 80
runPCA 01 (42%)
runPCA 02 (10%)
sample_id
V10A27004_D1_Br3880
V10A27106_B1_Br3854
V10A27106_C1_Br3873
V10A27106_D1_Br3880
V10T31036_B1_Br3854
V10T31036_C1_Br3873
V10T31036_D1_Br3880
−20
0
20
40
−25 0 25 50
runPCA 01 (33%)
runPCA 02 (17%)
path_groups
none
Ab+
next_Ab+
pT+
next_pT+
both
next_both
−20
0
20
40
−25 0 25 50
runPCA 01 (33%)
runPCA 02 (17%)
sample_id
V10A27004_D1_Br3880
V10A27106_B1_Br3854
V10A27106_C1_Br3873
V10A27106_D1_Br3880
V10T31036_B1_Br3854
V10T31036_C1_Br3873
V10T31036_D1_Br3880
p<0.05 in
targeted
sequencing
panel