due mainly to two samples, P4U1 and P7U6,
while in the non-AA-associated distribution,
the peak corresponding to larger clone sizes
was almost exclusively due to sample P65U.
This observation demonstrates that drastic
clonal expansions have occurred in some
individual MNU tissues. One possible expla-
nation for this observation is that urothelial
cells from the same clone have acquired more
than one driver mutation, which confers pro-
liferative and competing advantages. To test
this hypothesis, we deduced the co-occurrences
of driver mutations in those clones using the
pigeonhole principle ( 37 ). Indeed, more than
one driver mutation was nested in single clones
in each of the above three samples (P4U1, P7U6,
and P65U) (Fig. 3D). For example, mutations
inKDM6A,TP53,KMT2D, andARID1Awere
simultaneously acquired by a single clone in P4U1
(Fig. 3D). Compared with non-AA-associated
samples, AA-associated MNU had a significantly
86 2 OCTOBER 2020•VOL 370 ISSUE 6512 sciencemag.org SCIENCE
Mela.
LUAD
Col.
KCC
Breast
AML
MNU
(non-AA)
MNU(AA) UCC
SBS10
0.01
0.1
10
100
1
Mutation burden (muts/Mb)
0
10
(^20) 5’-TpCpT-3’
5’-TpCpG-3’
5’-TpTpT-3’
- Cosine similarity = 0.86(P65U)
0
10
20
- 5’-TpCpT-3’
5’-TpCpG-3’
5’-TpTpT-3’
COSMIC signature 10
C>A C>G C>T T>A T>C T>G
(SBS10a&10b)
Number of mutations 0
500
1000
1500
2000
0.0 0.4 0.8 1.2 1.6 2.0
0
500
1000
1500
0.00.40.81.21.62.0
0
100
200
300
400
500
0.00.40.81.21.62.0
AA-associated
Non-AA
Mutant clone size (mm²)
AB
% Mutations
C
0
20
40
60
80
100
KMT2D
KMT2D
TP53
KMT2D
KDM6A
ARID1A
ARID1A
0
20
40
60
80
100
FAT1
STAG2
KDM6A
CREBBP
ATM
ATM
ATM
KMT2D
FAT1
KMT2C
ATM
FAT1
CREBBP
ATM
TP53
0
20
40
60
80
100
% Mutant cells
Mutant clone size (mm
2 )
P4U1 P7U6 P65U
Nonsyn. Syn.
Non-AAMNUOther AAMNU
P65U
(SBS10)
P4U1
(AA)
P7U6
(AA)
DE
0.250.75 1.25 0.250.500.751.001.25
0.01
1
10
100
0.1
Mutation burden (muts/Mb)
Morphologically normal urothelium
0.89
(Sig10)
Mean clone size (mm²)
AA
Other
Signature
0.61 1.06
0.34 0.43
1.34
0.31 0.33 0.47
0.51
0.87 0.95
0.670.53
0.28 0.34
0.26
0.28
0.31
0.48
0.39
0.260.24
0.27 0.35
0.29
0.40 0.410.37
0.91
0.28
F
p<0.001
í
0.25 í í í í í í
0.50
0.75
Passengers
KMT2DTP53KDM6A
CREBBP KMT2C
Mutant cell fraction (MCF)
Mean MCF
(passengers)
Syn.
Drivers
Mean size
with CI95%
í
G
ATMFAT1
KMT2D KDM6A TP53
0
50
100
150
200
250
dN/dS ratios (q<0.05)
Missense
Nonsense+splice
Indels
H
0
5
10
KMT2D
TP53
KDM6A
CREBBP
KMT2C
Mut_no
0.25
0.50
0.75
Proportion
C:G>A:TC:G>G:CC:G>T:AT:A>A:TT:A>C:GT:A>G:C
Indels
I
FAT1
ATM
Fig. 3. Mutational burden and mutant clone expansion in MNU.
(A) Comparison of mutational burdens (muts/Mb, mutations per megabase) in
MNU (both AA-associated and non-AA-associated) and UCC samples. Median
mutational burdens of six other cancer types are indicated by dashed lines. AML,
acute myeloid leukemia; KCC, kidney clear cell carcinoma; Col., colorectal
carcinoma; LUAD, lung adenocarcinoma; Mela., melanoma. SBS10 resembles the
COSMIC mutational signature 10 (SBS10a and 10b). (B) Comparison of
mutational contexts of sample P65U and COSMIC SBS10a and 10b.
Representative 3-bp mutational contexts are labeled. (C) Distributions of mutant
clone sizes of mutations in AA-associated and non-AA-associated MNU samples.
Theyaxis represents the number of mutations. (D) Bar plots displaying the co-
occurrence of driver mutations in the same clones deduced on the basis of the
pigeonhole principle. (E) Comparison of clone sizes among different MNU
samples. Nonsyn., nonsynonymous; Syn., synonymous. (F) Mutational burdens
and average mutant clone sizes in MNU samples. The average mutant clone sizes
of AA-associated samples are labeled. (G) Comparison of clone sizes between
putative driver and passenger mutations. Wilcoxon rank-sum test was used. P<
0.05, P< 0.01, P< 0.001. CI95%, 95% confidence interval. (H) dN/dS
ratios for the three genes under significant positive selection in MNU.
(I) Mutational spectra of putative driver mutations. Mut_no, mutation number.
RESEARCH | RESEARCH ARTICLES
- 5’-TpCpT-3’