SCIENCE sciencemag.orgGRAPHIC: MELISSA THOMAS BAUM/
SCIENCE
copy number alterations found in urothelial
cancers. Lawson et al. detected copy number
alterations in only 28% of exomes, whereas
the average bladder cancer has 200 per
exome. Thus, genome instability occurs late
in the malignant transformation of the uro-
thelium, a pattern also seen in skin, esopha-
geal, and colon transformation (1, 3, 5).
Slowing the accumulation of mutations in
normal tissue might slow aging and reduce
the risk of cancer, and therefore it is important
to understand the causes of these mutations.
The two studies analyzed their respective
sequencing data for “mutational signatures”
that might point to these causes. Mutational
signatures are the patterns of single base mu-
tations within distinct sequences of preced-
ing and following bases that can distinguish
various mutational processes. Both studies
identified the mutational signature caused by
APOBEC cytidine deaminase activity in more
than half of the individuals studied, although
it was not found in every sample from those
individuals. This signature is almost always
found in bladder cancer and often in cancer
of the ureter; it is also found in many other
cancer types ( 11 ). The two studies show that
increased APOBEC activity seems to accom-
pany malignant transformation in urothelial
cancer. The mechanisms responsible for thisactivation, and whether it is a cause or an ef-
fect of transformation, are unknown.
Lawson et al. sequenced the entire ge-
nomes of a subset of the urothelial samples,
which enabled them to uncover previously
unknown mutational signatures. One of these
signatures correlated with tobacco smoking.
This may solve the puzzle that no specific
smoking-associated mutational signature
has been identified in bladder cancer, even
though tobacco smoking is a well-established
risk factor for bladder cancer and causes a
highly recognizable signature in lung cancer.
The smoking signature in lung cancer bears
no resemblance to the smoking-associated
signature identified by Lawson et al., and the
mechanisms by which tobacco might cause it
are unknown. Li et al. did not detect this sig-
nature, possibly because they analyzed whole
exomes rather than whole genomes.
A prominent difference between the mu-
tational signatures detected in the two stud-
ies is the presence of the signature caused by
aristolochic acid (AA) in more than half of
the ureter samples and about a third of the
bladder samples examined by Li et al. AA is
a kidney toxin and carcinogen that occurs
naturally in some herbs used as medicine. AA
exposure is widespread in East Asia, includ-
ing China (the location of patients studiedby Li et al.), but very limited in Europe (the
location of individuals studied by Lawson et
al.) (11–13). AA mutagenesis was previously
detected in a few normal urothelial samples
( 13 , 14 ), and Li et al. now show that AA muta-
genesis is widespread in normal urothelium.
This suggests that it may be possible to re-
liably assess the AA mutational signature in
cells or DNA shed from normal urothelium
and, using this signature, noninvasively as-
sess previous AA exposure ( 13 ). A noninvasive
test for AA exposure would offer substantial
benefits to research into the epidemiology of
AA-associated disease and to secondary pre-
vention of cancer and kidney failure in AA-
exposed individuals.
An unexpected finding was the strong dif-
ference in driver mutation preferences be-
tween individuals, with, for example, one per-
son having multiple independent mutations
in KDM6A and few in ARID1A, and another
person with the opposite pattern. It will be
interesting to see how general this phenom-
enon is and whether driver preferences can
predict cancer risk or, very speculatively,
suggest prophylactic therapies. Each study
explored mutations in a different dimen-
sion. One studied one or two samples from
a broad swath of individuals, and the other
was a deep study of many samples from a few
individuals. To understand the implications
of differences in preferences of driver gene
mutations between individuals, a study that
is simultaneously broad and deep is required.
Nevertheless, the shared message of Li et al.
and Lawson et al. is that histologically nor-
mal urothelium contains many clones that
are only a few steps away from turning malig-
nant but that rarely do. Studies using mouse
models of the development of urothelial
cancers, similar to a recent study of the de-
velopment of esophageal cancer ( 15 ), would
further delineate the mechanisms by which
mutations in the normal human urothelium
sometimes give rise to cancer. jREFERENCES AND NOTES- I. Martincorena et al., Science 348 , 880 (2015).
- M. A. Lodato et al., Science 359 , 555 (2018).
- I. Martincorena et al., Science 362 , 911 (2018).
- S. F. Brunner et al., Nature 574 , 538 (2019).
- H. Lee-Six et al., Nature 574 , 532 (2019).
- A. Yokoyama et al., Nature 565 , 312 (2019).
- L. M o o re et al., Nature 580 , 640 (2020).
- K. Yoshida et al., Nature 578 , 266 (2020).
- A. R. J. Lawson et al., Science 370 , 75 (2020).
- R. Li et al., Science 370 , 82 (2020).
- L. B. Alexandrov et al., Nature 578 , 94 (2020).
- A. W. T. Ng et al., Sci. Transl. Med. 9 , eaan6446 (2017).
- H. Lu et al., Theranostics 10 , 4323 (2020).
- Y. D u et al., E u r. U ro l. 71 , 841 (2017).
- B. Colom et al., Nat. Genet. 52 , 604 (2020).
ACKNOWLEDGMENTS
This work was supported by Singapore National
Medical Research Council grant MOH-000032/MOH-
CIRG18may-0004 and by the Singapore Ministry of Health
via the Duke-NUS Signature Research Programmes. I thank
A. Boot for comments.
10.1126/science.abe0955BladderCancerAA-containing herbsSmokingEndogenous ExogenousUreterKidneySection of urothelium from
the bladder or ureterDetect genes that drive clonal expansionSequence and detect somatic mutationsMutational signatures suggest causes of mutationsClonal expansion
in normal urothelium:
KMT2D and KDM6A
mutationsMalignant transformation
triggered by mutations in other
genes (e.g., TP53, PIK3CA, FGFR3,
RB1) or other factorsAA, aristolochic acid; FGFR3, fbroblast growth factor receptor 3;
KDM6A, lysine-specifc demethylase 6A;
KMT2D, histone-lysine N-methyltransferase 2D;
PIK3CA, phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit a;
RB1, RB transcriptional corepressor 1; TP53, tumor protein 53.APOBECs2 OCTOBER 2020 • VOL 370 ISSUE 6512 35Mutations in the bladder and ureter linings
Cells in the normal urothelium that lines the ureter and bladder accumulate mutations caused by endogenous
mutagenic processes (e.g., APOBEC cytidine deaminases) or by exogenous mutagens (e.g., AA or tobacco
smoke). Some mutations (e.g., in the chromatin modifying genes KMT2D and KDM6A) confer competitive
advantages that drive cells to colonize larger regions of the urothelium. Additional mutations (e.g., in the genes
TP53, PIK3CA, FGFR3, or RB1) and perhaps other changes are needed to trigger malignant transformation.