HMDB ID input pane
Node attributes
Network view
Edge attributes
Supplemental Figure 1. Cytoscape App MetBridge Generator. The left pane is Cytoscape App
MetBridge Generator (code name: rsMetabPPI). User can input HMDB numbers to search for
subnetwork involving the given metabolites. Upper-right pane shows MetBridgeDKD network. Red,
blue and pink nodes represent metabolites, enzymes, and bridge proteins, respectively. Pink, red
and blue edges represent metabolism, manually curated metabolism and PPI respectively.
Selected nodes and edges are colored in yellow and red, respectively, and also marked by the
red arrow. The detailed information on the selected nodes (node attributes) is displayed in the
lower pane. The detail on the selected edges (edge attributes) is displayed in the lower part of
the figure.
A
B
catabolic process
metabolic process
lipid catabolic process
intracellular
non-membrane-bounded organelle
biological_ process
small molecule catabolic process
sarcomere
cellular catabolic process
cellular process
myofibril
non-membrane-bounded organelle
membrane-bounded organelle
small molecule metabolic process
cellular metabolic process
contractile fiber
organelle
nitrogen compound metabolic
process
primary metabolic process
cellular lipid catabolic process
cell part
contractile fiber part
intracellular organelle
cellular nitrogen compound
metabolic process
lipid oxidation
cellular amino acid and derivative
metabolic process
organic acid metabolic process
lipid modification
fatty acid oxidation
cellular lipid metabolic process
fatty acid catabolic process
microbody
organelle part
amine metabolic process
cellular_component
intracellular part
organic acid catabolic process
peroxisome
cytoplasm
membrane-enclosed lumen
mitochondrion
intracellular organelle part
amine catabolic process
cellular amino acid metabolic
process
fatty acid beta-oxidation
macromolecular complex
carboxylic acid catabolic process
organelle lumen
cellular amino acid catabolic
process
fatty acid beta-oxidation using
acyl-CoA oxidase
mitochondrial lumen
mitochondrial part
carboxylic acid metabolic process
intracellular organelle lumen protein complexmitochondrial matrix
fatty acid alpha-oxidation
branched chain family amino acid
catabolic process
fatty acid metabolic process
fatty acid beta-oxidation
multienzyme complex
branched chain family amino acid
monocarboxylic acid metabolic
metabolic process
process
citrate metabolic process
Enzymes
5.00E-2
p-value
D
protein ubiquitination
modification-dependent protein
catabolic process
proteolysis involved in cellular
protein catabolic process
post-translational protein
modification
ubiquitin-dependent protein
catabolic process
protein modification by small
proteolysis
protein conjugation
modification-dependent
macromolecule catabolic process
cellular protein catabolic process
protein catabolic process
protein modification process
cell
integral to membrane
membrane part
G1/S transition of mitotic cell cycle
regulation of programmed cell death
regulation of molecular function
cellular component organization
cytoskeletal part
regulation of hydrolase activity
cytoplasm
non-membrane-bounded organelle
regulation of caspase activity
mitochondrial
intermembrane space
organelle
cytoskeleton
signaling regulation of catalytic activity
enzyme linked receptor protein
signaling pathway
signaling pathway
organelle envelope lumen
organelle envelope
intracellular
non-membrane-bounded organelle
regulation of cell death
intracellular organelle part
intracellular organelle
intermediate filament cytoskeleton
regulation of apoptosis
organelle organization
transmembrane receptor protein
tyrosine kinase signaling pathway
membrane-enclosed lumen
microtubule cytoskeleton
positive regulation of cell death
biological_process
cell surface receptor linked
signaling pathway
protein complex
keratin filament
regulation of cellular process
cellular process
epidermal growth factor receptor
signaling pathway
cellular_component
macromolecular complex
intracellular part
microtubule
intermediate filament
biological regulation
mitotic cell cycle
mitochondrion organization
ubiquitin ligase complex
positive regulation of cellular
process
cellular metabolic process
metabolic process
cell cycle
envelope
cullin-RING ubiquitin ligase
complex
intracellular
organelle part
positive regulation of biological
process
catabolic processprimary metabolic process
regulation of biological process
interphase
cell part
pore complex
cellular component movement
cellular macromolecule metabolic
process
macromolecule metabolic process
cellular catabolic process
cell cycle phasecell cycle process
< 5.00E-7
membrane
intrinsic to membrane
cellular protein metabolic process
cellular macromolecule catabolic
process
protein metabolic process protein modification by small
protein conjugation or removal
macromolecule catabolic process
macromolecule modification
interphase of mitotic cell cycle
cytoplasmic part
intracellular
cellular ketone metabolic process
cellular amine metabolic process
oxoacid metabolic process
C
intracellular membrane-bounded
organelle
cell
lipid metabolic process
cytoplasmic part
mitochondrial envelope
mitochondrial part
membrane-bounded organelle
intracellular membrane-bounded
organelle
mitochondrion
regulation of endopeptidase activity
regulation of peptidase activity
Bridge proteins
Supplemental Figure 2. GO terms enriched in enzymes and bridge proteins in the network
connecting thirteen metabolites. Each node represents GO term. Each arrow represents
hierarchical relationship between the terms. Significantly enriched terms in the hierarchy were
colored depending on their significance. GO biological processes and cellular components
enriched in the enzymes (A and B, respectively) and those enriched in the bridge proteins
(C and D, respectively) are shown.
Control
MDM2
db/m
db/db
Supplemental Figure 3: Representative image of immunohistochemistry of MDM2 in kidney
tissue of db/m and db/db mice. There were no clear differences in the tissue expression level
of MDM2 between db/m and db/db mice. n = 3 per group. Magnification: 40x.
Nephrin
A
0.00008
0.00015
0.00010
0.00002
0.00005
Podocin
0.0004
0.00000
Nut
Plac
D
*
0.000025
*
0.0190
0.000010
0.0001
E
CD2AP
0.000015
0.0002
0.0000
Nut
Plac
0.000020
0.0381
0.0003
*
0.0422
0.0095
0.00004
0.00000
C
0.00020
*
0.00006
Podocalyxin
B
0.000005
Synaptopodin
0.0010
0.000000
Nut
Plac
*
0.0190
F
WT1
0.0008
0.0006
0.0006
0.0004
0.0004
0.0002
0.0002
0.0000
0.0000
Nut
*
0.0010
0.0008
Plac
Nut
Plac
0.0422
Plac
Nut
Supplemental Figure 4. Expression analyses of marker genes for diabetic nephropathy.
Podocyte marker gene expression levels for control mice treated with placebo (Plac) and
Nutlin-3a-treated diabetic mice (Nut) are shown in A-F. Benjamini-Hochberg corrected
p-values were below 0.05 for all the genes.
MCCC2
p = 0.444
8.5
Tubules
p = 0.195
p = 0.00749 *
6.0
5.5
6.5
6.0
7.0
6.5
7.5
7.0
8.0
log 2 (mRNA intensity)
p = 0.00584 *
7.5
log 2 (mRNA intensity)
8.0
Glomeruli
LD
ERCB Pima
LD
ERCB Pima
Supplemental Figure 5: Expression analyses of methylcrotonoyl-CoA carboxylase 2 (MCCC2)
in tissue biopsy of human kidney. Gene expression levels of MCCC2 in glomeruli and tubules
of living donors (LD) were compared with those of ERCB and Native Americans (Pima).
Supplemental methods
Assessment of hub bridge proteins
We attempted to extract “bridge” proteins which have significant number of interactions to the
enzymes regulating our thirteen metabolites based on a simple statistical framework. First, we
classified the set of KEGG enzymes in our MetBridge network (Z) into (A1) those which are
associated with at least one of given set of metabolites m ∈ M (M is the set of our thirteen
metabolites in this case) or (A2) those which are not associated with any one ofm ∈ M (z ∈ A1
⋁ z ∈ A2, A1 ∩ A2={ } ). The enzymes which directly interact with those which are associated
with one of our thirteen metabolites were also classified asA1 in order to account the possibility
that such interactions among a pair of enzymes may constitute an enzymatic complex or a
functional unit of the metabolic reaction. We searched for bridge proteins (P) which interact with
at least one enzyme in A1. Then for each protein p ∈ P, we classified enzymes z ∈ Z into
(B1(p)) those which interact with the protein and (B2(p)) those which do not (z ∈ B1(p) ⋁z ∈
B2(p), B1(p) ∩ B2(p) = { }). Thus, each enzyme z ∈ Z can be classified asA1 or A2, and also
B1(p) or B2(p) for a given protein p. Subsequently we tested enrichment of enzymes which are
categorized both in A1 and B1(p) using hypergeometric test; Let rA1 and rB1(p) be the actual
probability of z classified as A1 and B1(p) respectively. Let rA1B1(p) be the actual rate of z
classified as both A1 and B1(p). Then, the hypotheses to test are H0: rA1B1(p) = rA1 ⋅ rB1(p) and H1:
rA1B1(p) > rA1 ⋅ rB1(p), where rA1, rB1(p) and rA1B1(p) are estimated by 𝑟̂𝐴1 =
and 𝑟̂𝐴1𝐵1(𝑝) =
|𝐴1 ∩𝐵1 (𝑝)|
|𝑍|
|𝐴1 |
, 𝑟̂𝐵1(𝑝) =
|𝑍|
|𝐵1 (𝑝)|
|𝑍|
respectively. This was to test whether a given bridge candidate protein
has significantly more interactions to the enzymes regulating our thirteen metabolites compared
to other KEGG enzymes. The number of calculated p-values were identical to the number of
bridge proteins. These p-values were corrected using Benjamini–Hochberg procedure.
Calculating expected number of PPIs among MetBridgeDKD network
The whole MetBridge network contained 58 703 edges among 3 387 proteins. Thus, under the
assumption that two different proteins randomly picked from MetBridge interact at probability p,
the estimated probability 𝑝̂ indicating an interaction between randomly picked two proteins was
3387
) ≅ 0.0102. We tested whether the probability p13 of two proteins randomly
2
58703/ (
picked from MetBridgeDKD is equal to p (H0: p13 = p, H1: p13 > p). Since MetBridgeDKD contained
291
3387
) × 58703/ (
)≅
2
2
291 proteins, the expected number of edges were calculated to be (
431.967.