Supplementary Data
Expression levels during development and adulthood (extended version)
Pure FTD-genes
MAPT
Data from HBA showed high expression rates for MAPT in all assessed brain tissues (Supplementary
Figure [SF] 1a); except for the cerebellum that exhibited a constant pattern, transcripts levels were
increasing during the prenatal phase (from 8-11 log2 intensity levels) and then remained constant after
birth and during aging for all tissues (10<log2<12 intensity levels). This tendency was confirmed by
transcription data from Braineac, where brain areas such as FCTX, TCTX and OCTX, together with
HIPP, THAL and CRBL exhibited highest expression rates (log2≤8.5; SF 1b). In Braineac the
expression levels ranged from a maximum of 8.5 (FCTX) to a minimum of 7.9 (WHMT) log 2 intensity
levels, with a 1.5 increase fold between the maximum and the minimum. These data suggest that
MAPT transcripts are elevated across brain areas (especially FCTX, TCTX, OCTX, HIPP, THAL and
CRBL) and that MAPT is important throughout the lifespan of an individual. The FCTX and TCTX are
the main brain areas affected in FTD associated with MAPT variability, and THAL and HIPP may also
be affected; conversely the OCTX and CRBL are generally not affected in FTD, whereas the WHMT,
which has recently shown to be affected in FTD [1], exhibits the lowest transcript levels.
GRN
Data from HBA showed high expression rates for GRN in all assessed brain tissues, with 8<log2<10
intensity levels throughout all life stages (SF 2a). In Braineac the brain regions showing highest
expression (log2=7.5) were SNIG, WHMT and MEDU, followed by the HIPP, THAL, FCTX, OCTX and
TCTX (that in Braineac had all similar expression rates [log2~7.4]), whereas there was a substantial
decrease in the CRBL (log2~6.8; SF 2b) with a 1.6 increase fold between the maximum and the
minimum levels. Also in this case, data from both datasets clearly suggest that GRN appears to be an
important factor throughout life. Variability in GRN has been associated to some extent with TDP-43
pathology; this signature is identified mainly in the FCTX, subcortical regions and, more rarely, in the
HIPP. In all of these brain areas expression rates for GRN were similar (log2~7.4).
CHMP2B
Data from HBA for CHMP2B showed high expression that remained almost constant (log2~9) in all
brain tissues during all life stages (SF 3a). Only the CRBL revealed a gradual decrease form 10 in the
pre-natal phase to log2<8 in the late stages of life. In Braineac the brain regions showing highest
expression (log2~6.6) were WHMT, MEDU, and SNIG, followed by the THAL, HIPP, TCTX, FCTX and
OCTX (log2~6.2), whereas there was a substantial decrease in the CRBL (log2~5.4) with a 2.4
1
increase fold between the maximum and the minimum levels (SF 3b). CHMP2B variability has been
associated with the FTLD-UPS type of pathology, which is mainly affecting the FCTX and other
cortical areas and the dentate gyrus which are all brain areas that exhibited similar expression rates
for CHMP2B (6<log2<6.2).
RAB38
Data from HBA for RAB38 showed low expression levels in all brain tissues with a decrease during the
pre-natal stage from log2~6 to below 5, level that then remains constant throughout late pre-natal
stages as well as during infancy and adulthood (SF 4a). Data from Braineac showed that expression
levels for RAB38 were low (4<log2<4.5) in all brain tissues; specifically, expression was highest in
WHMT and lowest in CRBL (with a fold change of 1.2; SF 4b); considering that RAB38 was
associated with the bvFTD subtype [2] a closer look at the FCTX revealed intensity levels log2~4.4. ;
RAB38 showed rather lower expression rates, potentially implying that RAB38 is a sensitive cellular
marker thus, given this peculiar threshold, a sudden change in expression levels might significantly
impact cellular homeostasis.
CTSC
Data from HBA for CTSC showed high expression levels in all assessed tissues during early
development stages and then a decrease in the late pre-natal stage; after birth the levels are
moderate and remain constant during aging with highest levels in the THAL (7<log 2<8) and lowest in
the CRBL (log2~6; SF 5a). Data from Braineac also showed highest expression in the THAL (log2~6.8)
with a notable 3.8 increase fold compared to the CRBL where expression is lowest (log 2<4.8; SF 5b);
considering that CTSC was associated with the bvFTD subtype [2] a closer look at the FCTX revealed
moderate intensity levels (log2<5.6) throughout life suggesting that CTSC is an important factor at
every age.
BTNL2
Data from HBA for BTNL2 showed low to moderate (5<log2<6) expression levels throughout
development and aging (SF 6a). Data from Braineac also showed that BTNL2 had overall moderate
and conserved expression levels across all tissues (log2~5.6; SF 6b), with a 1.2 increase fold between
SNIG (log2~5.7) and CRBL (log2~5.5). BTNL2 was associated with the entire FTD spectrum (bvFTD,
PPA and FTD-MND), thus a closer look at all brain areas affected in FTD (FCTX, TCTX, THAL,
WHTM and HIPP) showed moderate expression levels (log2~5.6) throughout life suggesting that
BTNL2 is an important element at every age.
HLA-DRA
Data from HBA for HLA-DRA showed relatively low (log2~4) expression levels during development
(pre-natal stages) and in the early stages of life (log2~5), whereas expression increased in later stages
2
of adulthood reaching peaks of log2~8–9 for all tissues, besides the CRBL that showed intensity levels
of log2~7 (SF 7a). Braineac showed highest expression rate in MEDU (immediately followed by SNIG
and WHMT; log2~7) and lowest in CRBL (log2~5), with a fold change of 3.8 (SF 7b). HLA-DRA was
associated with the entire FTD spectrum (bvFTD, PPA and FTD-MND), thus a closer look at all brain
areas affected in FTD (FCTX, TCTX, THAL, WHTM and HIPP) showed intensity levels varying
between log2~5.7 (FCTX and TCTX), ~6.1 (THAL), ~6.5 (HIPP) and ~7 (WHTM) suggesting that HLADRA is an important element in all brain tissues that are vulnerable in FTD.
HLA-DRB5
Data from HBA for HLA-DRB5 revealed low expression levels at all stages (pre-natal, during infancy
and adulthood) with transcript intensity levels of log2~4 in all assesse brain tissues (SF 8); of note,
data from Braineac were not available for this gene. These data suggest that HLA-DRB5 has in
general low expression rates, thus changes in expression levels might be dependent on stimuli and/or
signalling involving immune response.
TMEM106B
Data from HBA for TMEM106B showed high (log2~10) expression levels throughout life in all tissues
(SF 9a); data from Braineac showed robust and consistent expression (log2≤7.5) across all tissues,
with a minimal fold change of 1.1 between WHMT (highest expression rate) and SNIG (lowest
expression rate) (SF 9b). TMEM106B is mainly associated with FTLD-TDP and GRN variability and
the main affected areas (FCTX, sub-cortex and HIPP) revealed robust constitutive expression levels
(log2≤7.5).
Spectrum FTD-genes
Data from HBA revealed that all genes (C9orf72, VCP, SQSTM1, UBQLN2, OPTN, TDP-43 and FUS)
had moderate to high expression rates across all tissues during development and aging (SF 10-16)
and it was noteworthy that C9orf72, TDP-43 and FUS had highest expression levels in CRBL
(8<log2<9; 9<log2<10; 10<log2<11, respectively; SF 10, 15-16a) in adulthood. These data were
reflected in the Braineac dataset where expression rates for these three genes were highest in the
CRBL (log2~7.4; log2=7.5; log2~7.9, respectively; SF 10, 15-16b). For the other genes (VCP,
SQSTM1, UBQLN2 and OPTN) data from HB Atlas revealed overall robust expression levels in all
tissues (10<log2<11; 7<log2<9; 10<log2<11; 7<log2<9, respectively) and this trend was confirmed in
Braineac; of note, expression levels across tissues were homogeneous for VCP (7.7<log2<7.9 with a
fold change of 1.2 between MEDU [highest expression levels] and CRBL [lowest expression levels])
and SQSTM1 (7.3<log2<7.5 with a fold change of 1.2 between SNIG [highest expression levels] and
PUTM [lowest expression levels]) (SF 11-14a). Conversely, there was a slight slope in the expression
levels across tissues in Braineac for UBQLN2 (7.9<log2<8.4) and OPTN (6.3<log2<6.9): respectively,
3
highest expression was seen in FCTX and TCTX with a 1.5 increase fold compared to WHMT
(UBQLN2) and in TCTX and FCTX with a 1.6 increase fold comparatively to THAL (OPTN) (SF 1114b).
Functional annotation of additional modules in FCTX and TCTX
CHMP2B – Darkolivegreen module – FCTX
No significant BPs neither MFs could be identified for this module, whereas for the CCs only three
terms were evident indicating the “cytoplasm” (p=3.76x10-4) and elements such as “bounding
membrane of organelle” (p=6.81x10-3; Supplementary Table [ST] 3).
Pathway analysis revealed 3 significant terms (p<0.05): among these were the “peroxisome”
(p=1.21x10-2) and “peroxisomal lipid metabolism” (p=1.8x10-2; Table 3).
Interestingly, enrichment analysis for the WGCNA Brain list provided evidence for overrepresentation
of oligodendrocyte markers: “turquoise M9 – Oligodendrocyte” (CTX) (p=1.13x10-9), “blue M2 –
Oligodendrocytes” (HumanMeta) (p=2.49x10-8), and “Oligodendrocyte probable” (Cahoy) (p=5.19x106
).
FUS and TDP-43 – Lightcyan, blue and turquoise modules – FCTX
Functional annotation analysis for FUS (ID3656904) in the lightcyan module showed 18 BPs, 3 CCs
and 4 MFs (p<0.05) (ST 12).
The BPs for this module revealed a number of precise terms: “RNA metabolic process” (p=9.49x10-8),
“RNA processing” (p=5.61x10-7), “gene expression” (p=8.38x10-7), “mRNA processing” (p=2.1x10-5)
and “RNA splicing” (p=2.58x10-3). The CCs clearly indicated the “nucleus” (p=8.65x10-7). Finally, for
the MFs “nucleic acid binding” was the most significant term (p=5.63x10-5).
After pathway analysis we found 2 significant (p<0.05) terms: “gene expression” (p=1.57x10-2) and
“spliceosome” (p=2.52x10-2; Table 3).
The module was enriched for different GO terms associated with “metabolic process and RNA
processing”, whereas the only Brain list term found enriched was with “turquoise M14 Nucleus”
(p=4.13E-05) from Miller et al. (2010). Similar results were obtained for TCTX.
Then, considering FUS identified by the two transcripts (ID3656950 and 3656954) and clustering in
the blue module, we only found 1 term, “DNA binding” (p=8.82x10-4; Table 3), for the MFs.
Pathway analysis indicated 3 significant terms (p<0.05), including “generic transcription pathway”
(p=1.45x10-11) and “gene expression” (p=2.46x10-4; Table 3).
4
Finally, TDP-43 that clustered in the turquoise module revealed 17 BPs, 11 CCs and 12 MFs (p<0.05;
ST 13).
Among the most significant BPs we found various terms, including “sensory perception of chemical
stimulus” (p=4.76x10-8) and “sensory perception” (p=2.47x10-7). Among the CCs several terms
indicated the “extracellular space” (p=8.68x10-10), including the “extracellular matrix” (p=9.57x10-7);
finally, the MFs showed “G-protein coupled receptor activity” (p=4.79x10-11) and “transmembrane
signaling receptor activity” (p=4.89x10-10) among other signal transduction terms.
Pathway analysis revealed up to 15 significant (p<0.05) terms. These were variable and difficult to
interpret for association with brain disorder.
Of note, both the turquoise and blue modules include a high number of co-expressed genes (4759 and
3329, respectively); this feature creates a noisy environment, which, to some extent, represents a
limitation in our analysis of FUS and TDP-43.
MAPT – Lightyellow module MAPT – TCTX
The functional annotation analysis for this module revealed up to 24 BPs, 11 CCs and 7 MFs with
p<0.05 (ST 6).
The most significant term pointed strictly towards transcription processes through “transcription from
RNA polymerase II promoter” (p=1.23x10-4), which was further supported by “transcription DNAtemplated” (p=2.1x10-2) and “RNA metabolic process” (p=3.96x10-2). When assessing the CCs, two
terms were highly significant indicating the “nucleus” (p=3.69x10-5) and the “nuclear lumen”
(p=8.95x10-4). Finally, the MFs highlighted one highly significant term: “protein binding” (p=2.93x10-4).
In addition other interesting terms here were “transcription factor”, “nucleic acid” and “poly(A) RNA
binding” (p=1.42x10-2, 1.63x10-2 and 2.98x10-2).
Pathway analysis through gProfiler indicated 6 terms (p<0.05): the most sensible and in line with our
functional annotation analysis was “RNA Polymerase II Transcription Elongation” (p=4.21x10-2; Table
4).
Again, significant overlap “green M10 – Glutamatergic Synaptic Function” (CTX) (p=3.79E-05) was
observed.
GRN – Cyan module – TCTX
No significant BPs neither MFs could be identified, whereas there were up to 5 significant CCs
(p<0.05; ST 8). These indicated general terms such as “Intracellular part” (p=5.64x10-3) and
“membrane-bounded organelle” (p=3.27x10-2).
Pathways analysis revealed 7 significant terms associated with this module (p<0.05); of note were
“signaling by Wnt” (p=7.61x10-3) and “lysosome” (p=1.21x10-2; Table 4).
5
Also the cyan module was enriched for green M10 GlutamatergicSynapticFunction CTX (9.89E-08),
and also for green M5 Mitochondria HumanMeta (p=1.74E-05).
CHMP2B – Green module – TCTX
After functional annotation analysis there were 37 significant BPs, 29 CCs and 8 MFs (p<0.05; ST 8).
“Intracellular transport” (p=8.91x10-8), including “protein transport” (p=1.76x10-7) and “protein
localization” (p=1.97x10-5) were among the most significant BPs. Other interesting terms were
“cytoplasmic transport” (p=1.31x10-6), “mitochondrion organization” (p=3.24x10-5), “ribosome
biogenesis” (p=4.9x10-4) and “RNA processing” (p=5.87x10-4), particularly indicating “ncRNA metabolic
process” (p=1.35x10-3) and “rRNA metabolic process” (p=8.27x10-3). The most significant CC was
“intracellular organelle part” (p=3.14x10-13) that was further specified by “cytoplasm” (p=3.34x10-7) as
well as “nuclear lumen” (p=4.3x10-7). In addition there were also terms referring to the “mitochondrion”
(p=3.83x10-3), and “catalytic” and “ribonucleoprotein complex” activity (p=7x10-4 and 1.6x10-2). Finally,
the most significant MFs pointed to “poly(A) RNA binding” (p=3.27x10-9) and “catalytic activity”
(p=2.67x10-4), as well as “transferase” and “protein transporter activity” (p=2.57x10-3 and 4.34x10-3).
Pathway analysis showed up to 9 significant terms (p<0.05). Interestingly, the most significant
indicated “antigen processing: ubiquitination and proteasome degradation” (p=2.3x10-5), which was
supported by two further pathways such as “class I MHC mediated antigen processing and
presentation” (p=1.17x10-4) and “ubiquitin mediated proteolysis” (p=2.24x10-3). In addition, also
“chromatin organization” and “chromatin modifying enzymes” (p=2.4x10-4, both) were evident (Table
3).
Using the Brain list, we obtained among the top results a significant overlap with the “blue M16 –
Neuron” module (CTX) (p=8.75E-36), whereas no other significant term was pointing to
oligodendrocytes. Looking at the cross-tabulation of module in FCTX vs TCTX, we observed that the
darkolivegreen module in FCTX was split into a darkolivegreen module and green module in TCTX,
with the darkolivegreen module maintaining the evidence for oligodendrocytes signature (“blue M2 –
Oligodendrocytes”, p=1.75E-06). Given that FCTX is the main brain area of interest, and also that
CHMP2B showed a higher MM in the FCTX module, we consider more appropriate the
oligodendrocytes signature for CHMP2B as a result of this study.
OPTN – grey60 module – TCTX
This module only revealed 2 BPs and 3 CCs, whilst no MFs (p<0.05; not shown).
The BPs indicated “synaptic transmission” (p=5.37x10-6) and, more generally “cell-cell signaling”
(p=8.83x10-6), whereas the CCs showed “membrane part” (p=1.26x10-2) and “cell periphery”
(p=2.09x10-2).
6
Pathway analysis highlighted 7 potential pathways (p<0.05). These were showing elements active in
the brain such as “Neuroactive ligand-receptor interaction” (p=3.64x10-3).
FUS and TDP-43 – midnightblue, pink and magenta modules – TCTX
The FUS transcript with ID3656904 clustered in the midnightblue module (n=268); 73 BPs, 29 CCs
and 10 MFs reached significance (p<0.05; ST 15).
The BPs indicated clearly “RNA metabolic process” (p=7.49x10-12), “gene expression” (p=1.2x10-11)
and “mRNA processing” (p=4.15x10-8). Of note, also “RNA splicing” (p=5.04x10-3) was shown. Among
the CCs we saw that the major compartment was the nucleus (p=3.68x10-12) and for the MFs the main
terms were “RNA binding” (p=3.08x10-12) and “poly(A) RNA binding” (p=5.15x10-12). In addition, of
interest was also “transcription factor binding transcription factor activity” (p=1.16x10-2).
Pathway analysis identified 12 potential pathways (p<0.05). “Gene Expression” (p=8.33x10-5) was the
most significant followed by “Processing of Capped Intron-Containing Pre-mRNA” (p=1.3x10-2), clearly
relating to the metabolism of the RNA (Table 3).
The FUS transcript with ID3656950 clustered in the pink module (n=2979); 13 BPs and 2 MFs were
significant (p<0.05; not shown), whereas there were no CCs.
The BPs in which FUS was involved clearly indicated “nucleic acid metabolic process” (p=7.5x10-8)
and that was further supported by the MFs “DNA binding” (p=1.3x10-7) and “nucleic acid binding”
(p=2.32x10-4).
The pathway analysis just revealed 3 significant pathways (p<0.05) for which the most statistically
significant were “Generic Transcription Pathway” (p=5.68x10-11) and “Gene Expression” (p=1.84x10-5;
Table 3).
Finally, the FUS transcript with ID3656954 and TDP-43 co-clustered within the magenta module
(n=1000). Here 16 BPs, 2 CCs and 16 MFs were significant (p<0.05; ST 16). FUS and TDP-43 were
not among the genes supporting the GO terms in neither category.
For the BPs we noted that “inorganic cation transmembrane transport” (p=9.53x10-6) and “ion
transmembrane transport” (p=9.55x10-6) were the most significant; in addition, we also identified
“synaptic transmission” (p=1.09x10-3) and “potassium ion transmembrane transport” (p=4x10-3). The
CCs further supported the previous category revealing “voltage-gated potassium channel complex”
(p=2.86x10-2) and “potassium channel complex” (p=3.47x10-2). Finally the MFs showed “cation
transmembrane transporter activity” (p=3.72x10-5) and “potassium channel activity” (p=1.82x10-2)
among others providing consistency to this module. Pathway analysis revealed 3 significant potential
pathways (p<0.05). These confirmed the functional annotation data revealing “Neuronal System”
(p=3.15x10-6), “Potassium Channels” (p=5.85x10-3) and “Voltage gated Potassium channels”
(p=9.81x10-3).
7
Brain areas other than frontal and temporal cortex
Here we report relevant modules in brain areas other than frontal cortex and temporal cortex. All
statistics and modules’ functional annotations described hereafter are summarized in ST 18 and 19,
respectively.
Putamen
MAPT and GRN co-clustered in the purple module. None of the transcripts was a hub and their MM
values were > 0.5; this module was enriched for transcription-related processes. CHMP2B and
TMEM106B were found in the green module: TMEM106B was a hub with high MM values, whilst
CHMP2B among the 25% most interactive genes. This module was associated with cytoplasmic
protein catabolism and transport. HLA-DRA and CTSC were together in the royalblue module: neither
transcript was a hub while HLA-DRA showed high MM value. This module indicated strong association
with immune system processes. UBQLN2 and C9orf72 were found together in the brown module: here
UBQLN2 was a hub with high MM value, whilst C9orf72 had weak statistics. This module associated
with vesicle trafficking, proteolysis and protein catabolism in the synapse.
Thalamus
VCP and C9orf72 co-clustered in the lightyellow module. VCP was a hub and C9orf72 was among the
15% most interconnected genes; both exhibited elevated MM values. This module indicated protein
catabolic process through proteolysis and biology of the proteasome, and RNA processing or gene
expression related processes.
Hippocampus
CHMP2B and TMEM106B were found in the greenyellow module. TMEM106B was a hub while both
transcripts had MM values > 0.6. This module indicated implication of RNA metabolism, particularly,
mRNA splicing. We found HLA-DRA and CTSC together in the darkgrey module. Neither transcript
was a hub, whilst both showed high MM values. This module revealed association with immune
system processes. C9orf72 and OPTN co-clustered in the grey60 module: both had MM values > 0.5
8
but none was a hub. Functional annotation here indicated gene expression and RNA metabolism.
VCP and UBQLN2 were together in the black module: both had high MM values and were almost
hubs. This module indicated general protein catabolic processes.
White matter
HLA-DRA and CTSC co-clustered in the black module. CTSC was a hub and HLA-DRA was among
the 21% most interconnected transcripts; both transcripts exhibited high MM values. This module
associated with immune system processes. TMEM106B was a hub with high MM value in the tan
module that indicated general protein metabolism and transport. VCP was a hub with high MM value
in the darkturquoise module that indicated association with the biology of the ER-membrane network.
C9orf72 and UBQLN2 co-clustered in the royalblue module: both had MM values > 0.5, but neither
was a hub. Functional annotation pointed to protein modification and organization of the
mitochondrion.
Cerebellum
C9orf72 and OPTN were found in the purple module: neither was a hub and MM values were barely >
0.5. Overall, this module indicated protein catabolism.
Medulla
C9orf72 and VCP co-clustered in the grey60 module. Both had MM values > 0.5, but none was a hub.
This module indicated ubiquitin-dependent protein catabolic process and proteasome complex activity.
9
Supplementary Figures
Supplementary Figure 1
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
10
Supplementary Figure 2
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
11
Supplementary Figure 3
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
12
Supplementary Figure 4
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
13
Supplementary Figure 5
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
14
Supplementary Figure 6
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
15
Supplementary Figure 7
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
16
Supplementary Figure 8
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
17
Supplementary Figure 9
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
18
Supplementary Figure 10
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
19
Supplementary Figure 11
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
20
Supplementary Figure 12
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
21
Supplementary Figure 13
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
22
Supplementary Figure 14
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
23
Supplementary Figure 15
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
24
Supplementary Figure 16
a)
days
500
2000
10000
30000
years
1.37
5.48
27.40
82.19
b)
25
Supplementary Figure 17
Number of genes that fall into modules in FCTX (rows) versus modules in TCTX (columns) (the
module colours are not matched). Module overlap is quantified using Fisher's exact test to assign a
significance level to each module overlap, which is displayed by a colour scale based on log10(p).
Genes included in FCTX and TCXT modules were moderately shared: the black module in FCTX
(comprising MAPT and GRN) showed 147 transcripts (including MAPT) overlapping with the
lightyellow and 127 (including GRN) with the cyan modules in TCTX. The darkred module (containing
HLA-DRA and CTSC) had 123 transcripts (including HLA-DRA and CTSC) shared with the lightcyan
module in TCTX. The red module (comprising TMEM106B) had 64 transcripts (including TMEM106B)
overlapping with the darkturquoise and 436 shared with the green modules in TCTX. The purple
module (containing C9orf72, VCP, UBQLN2 and OPTN) showed 577 transcripts (including C9orf72,
VCP and UBQLN2) overlapping with the purple module in TCTX.
26
Composite Z-summary statistics for each module in FCTX and TCTX. a) The modules identified in
FCTX showed all elevated Z-summary score (>10); this was indicative of moderate to high preservation of each
module across tissues. Only the darkolivegreen module was not preserved in PUTM and SNIG. b) The modules
identified in TCTX showed, for the most, elevated Z-summary score (>10); only the grey60 module was not
preserved, and the magenta module had low Z-summary score (<10) exclusively in the PUTM.
a
FCTX modules
size
TCTX PUTM THAL
HIPP
WHMT
CRBL MEDU
SNIG
OCTX
mean
black
791
25.93
22.66
16.81
23.26
17.3
22.41
19.68
20.08
32.75
22.32
blue
3329
31.86
26.18
22.97
28.45
30.74
25.37
18.41
27.64
27.46
26.56
darkolivegreen
63
16.81
8.33
11.53
18.37
15.44
15.03
18.5
8.74
18.74
14.61
darkred
141
31.84
23.5
29.53
38.6
40.86
31.95
38.53
32.3
31.31
33.16
lightcyan
209
21.17
17.59
14.78
25.76
28.27
28.72
22.9
11.03
31.67
22.43
purple
1559
40.62
25.63
37.56
43.33
27.84
32.01
29.85
44.4
48.18
36.6
red
925
28.21
21.47
19.47
23.39
14.43
26.81
23.33
15.41
33.57
22.9
turquoise
4759
34.21
42.31
30.72
34.47
34.13
39.46
31.47
30.6
39.24
35.18
CRBL MEDU
SNIG
OCTX
mean
b
TCTX
modules
cyan
size
FCTX PUTM THAL
HIPP
WHMT
276
25.56
15.31
18.84
17.19
14.43
16.43
15.81
20.36
23.7
18.63
darkturquoise
164
27.45
16.05
14.13
16.06
17.49
27.34
14.4
13.06
29.61
19.51
green
1439
38.23
26.38
25.43
29.92
24.7
33.96
27.36
24.82
45.3
30.68
grey60
220
11.47
10.16
4.49
13
8.58
6.22
5.61
2.25
24.42
9.58
lightcyan
250
31.33
18.65
25.8
36.8
39.21
28.51
37.67
32.87
28.81
31.07
lightyellow
210
30.88
15.22
11.88
14.5
14.19
13.4
11.62
13.28
26.13
16.79
magenta
1000
23.45
1.41
15.97
17.15
14.03
11.25
20.24
13.57
26.21
15.92
midnightblue
268
19.63
14.01
11.77
18.5
14.1
29.73
11.87
12.01
29.85
17.94
pink
2979
32.19
29.44
23.9
26.83
31.42
23.1
20.53
25.17
30.94
27.06
purple
830
40.39
32.11
37.78
48.99
24.9
34.36
29.73
44.44
58.59
39.03
27
References for the pre-defined brain lists (useBrainLists=TRUE, in the userListEnrichment WGCNA
function) as described in the manuscript.
Category
References
ABA
Cell type markers from: Lein ES, et al. (2007) Genome-wide atlas of gene expression in the adult
mouse brain. Nature 445:168-176.
ADvsCT_inCA1
Lists of genes found to be increasing or decreasing with Alzheimer's disease in 3 studies:
Blalock E, Geddes J, Chen K, Porter N, Markesbery W, Landfield P (2004) Incipient
Alzheimer's disease: microarray correlation analyses reveal major transcriptional and tumor
suppressor responses. PNAS 101:2173-2178.
o Colangelo V, Schurr J, Ball M, Pelaez R, Bazan N, Lukiw W (2002) Gene expression profiling
of 12633 genes in Alzheimer hippocampal CA1: transcription and neurotrophic factor downregulation and up-regulation of apoptotic and pro-inflammatory signaling. J Neurosci Res
70:462-473.
o Liang WS, et al (2008) Altered neuronal gene expression in brain regions differentially
affected by Alzheimer's disease: a reference data set. Physiological genomics 33:240-56.
Postsynaptic Density Proteins from: Bayes A, et al. (2011) Characterization of the proteome,
diseases and evolution of the human postsynaptic density. Nat Neurosci. 14(1):19-21.
o
Bayes
Blalock_AD
Modules from a network using the data from: Blalock E, Geddes J, Chen K, Porter N, Markesbery
W, Landfield P (2004) Incipient Alzheimer's disease: microarray correlation analyses reveal major
transcriptional and tumor suppressor responses. PNAS 101:2173-2178.
CA1vsCA3
Lists of genes enriched in CA1 and CA3 relative to other each and to other areas of the brain,
from several studies:
o
o
o
o
o
Ginsberg SD, Che S (2005) Expression profile analysis within the human hippocampus:
comparison of CA1 and CA3 pyramidal neurons. J Comp Neurol 487:107-118.
Lein E, Zhao X, Gage F (2004) Defining a molecular atlas of the hippocampus using DNA
microarrays and high-throughput in situ hybridization. J Neurosci 24:3879-3889.
Newrzella D, et al (2007) The functional genome of CA1 and CA3 neurons under native
conditions and in response to ischemia. BMC Genomics 8:370. 4. Torres
Torres-Munoz JE, Van Waveren C, Keegan MG, Bookman RJ, Petito CK (2004) Gene
expression profiles in microdissected neurons from human hippocampal subregions. Brain
Res Mol Brain Res 127:105-114.
In either Ginsberg or Lein or Torres list.
Cahoy
Definite (10+ fold) and probable (1.5+ fold) enrichment from: Cahoy JD, et al. (2008) A
transcriptome database for astrocytes, neurons, and oligodendrocytes: A new resource for
understanding brain development and function. J Neurosci 28:264-278.
CTX
Modules from the CTX (cortex) network from: Oldham MC, et al. (2008) Functional organization of
the transcriptome in human brain. Nat Neurosci 11:1271-1282.
DiseaseGenes
Probable (C or better rating as of 16 Mar 2011) and possible (all genes in database as of ~2008)
genetics-based disease genes from: http://www.alzforum.org/
EarlyAD
Genes whose expression is related to cognitive markers of early Alzheimer's disease vs. nondemented controls with AD pathology, from: Parachikova, A., et al (2007) Inflammatory changes
parallel the early stages of Alzheimer disease. Neurobiology of Aging 28:1821-1833.
HumanChimp
Modules showing region-specificity in both human and chimp from: Oldham MC, Horvath S,
Geschwind DH (2006) Conservation and evolution of gene coexpression networks in human and
chimpanzee brains. Proc Natl Acad Sci USA 103: 17973-17978.
HumanMeta
Modules from the human network from: Miller J, Horvath S, Geschwind D (2010) Divergence of
human and mouse brain transcriptome highlights Alzheimer disease pathways. Proc Natl Acad Sci
107:12698-12703.
28
JAXdiseaseGene
Genes where mutations in mouse and/or human are known to cause any disease. WARNING: this
list represents an oversimplification of data! This list was created from the Jackson Laboratory:
Bult CJ, Eppig JT, Kadin JA, Richardson JE, Blake JA; Mouse Genome Database Group (2008)
The Mouse Genome Database (MGD): Mouse biology and model systems. Nucleic Acids Res 36
(database issue):D724-D728.
Lu_Aging
Modules from a network using the data from: Lu T, Pan Y, Kao S-Y, Li C, Kohane I, Chan J,
Yankner B (2004) Gene regulation and DNA damage in the ageing human brain. Nature 429:883891.
MicroglialMarkers
Markers for microglia and macrophages from several studies:
o
o
o
o
Gan L, et al. (2004) Identification of cathepsin B as a mediator of neuronal death induced by
Abeta-activated microglial cells using a functional genomics approach. J Biol Chem
279:5565-5572.
Albright AV, Gonzalez-Scarano F (2004) Microarray analysis of activated mixed glial
(microglia) and monocyte-derived macrophage gene expression. J Neuroimmunol 157:27-38.
Ait-Ghezala G, Mathura VS, Laporte V, Quadros A, Paris D, Patel N, et al. Genomic
regulation after CD40 stimulation in microglia: relevance to Alzheimer's disease. Brain Res
Mol Brain Res 2005;140(1-2):73-85.
Thomas, DM, Francescutti-Verbeem, DM, Kuhn, DM (2006) Gene expression profile of
activated microglia under conditions associated with dopamine neuronal damage. The
FASEB Journal 20:515-517.
MitochondrialType
Mitochondrial genes from the somatic vs. synaptic fraction of mouse cells from: Winden KD, et al.
(2009) The organization of the transcriptional network in specific neuronal classes. Mol Syst Biol
5:291.
MO
Markers for many different things provided to my by Mike Oldham. These were originally from
several sources:
Genetics-based disease genes in two or more studies from http://www.alzforum.org/
(compiled by Mike Oldham).
o Bachoo, R.M. et al. (2004) Molecular diversity of astrocytes with implications for neurological
disorders. PNAS 101, 8384-8389.
o Foster, LJ, de Hoog, CL, Zhang, Y, Zhang, Y, Xie, X, Mootha, VK, Mann, M. (2006) A
Mammalian Organelle Map by Protein Correlation Profiling. Cell 125(1): 187-199.
o Morciano, M. et al. Immunoisolation of two synaptic vesicle pools from synaptosomes: a
proteomics analysis. J. Neurochem. 95, 1732-1745 (2005).
o Sugino, K. et al. Molecular taxonomy of major neuronal classes in the adult mouse forebrain.
Nat. Neurosci. 9, 99-107 (2006).
Modules from the mouse network from: Miller J, Horvath S, Geschwind D (2010) Divergence of
human and mouse brain transcriptome highlights Alzheimer disease pathways. Proc Natl Acad Sci
107:12698-12703.
o
MouseMeta
Sugino/Winden
Conservative list of genes in modules from the network from: Winden K, Oldham M, Mirnics K,
Ebert P, Swan C, Levitt P, Rubenstein J, Horvath S, Geschwind D (2009). The organization of the
transcriptional network in specific neuronal classes. Molecular systems biology 5. NOTE: Original
data came from this neuronal-cell-type-selection experiment in mouse: Sugino K, Hempel C, Miller
M, Hattox A, Shapiro P, Wu C, Huang J, Nelson S (2006). Molecular taxonomy of major neuronal
classes in the adult mouse forebrain. Nat Neurosci 9:99-107
Voineagu
Several Autism-related gene categories from: Voineagu I, Wang X, Johnston P, Lowe JK, Tian Y,
Horvath S, Mill J, Cantor RM, Blencowe BJ, Geschwind DH. (2011). Transcriptomic analysis of
autistic brain reveals convergent molecular pathology. Nature 474(7351):380-4
29
References
1.
2.
Lam, B.Y., et al., Longitudinal white matter changes in frontotemporal dementia subtypes. Hum Brain
Mapp, 2014. 35(7): p. 3547-57.
Ferrari, R., et al., Frontotemporal dementia and its subtypes: a genome-wide association study. Lancet
Neurol, 2014. 13(7): p. 686-99.
30