INSIGHTS | PERSPECTIVES
sciencemag.org SCIENCEThe structural mechanisms underlying this
phenomenon are unknown and may in-
volve conformational selection of distinct
PrPSc species. The conformational selection
model predicates the coexistence of multi-
ple conformers within a single infected or-
ganism, some of which may replicate more
efficiently in their host under certain envi-
ronmental circumstances. The incubation
time of prion infections can vary immensely
between different strains, and the delay in
the onset of the pathology might reflect the
time needed for such selection to occur.
PrPSc conformer heterogeneity may also
underlie the barriers that control interspecies
prion transmission, the strength of which is
variable and depends both on host factors
and on prion strains. Although prion propa-
gation from cows to humans results in vari-
ant CJD, sheep prions appear to be largely
innocuous to humans. This species barrier
relies both on the structural diversity of the
PrPSc contained in the inoculum and the PrPC
of the host, which cannot always interact
with the misfolded conformer efficiently.
The ideas promulgated by Prusiner
have undergone a marked metamorphosis.
Templated nucleation of protein aggregates
is now known to underlie not only diseases
but also many physiological processes, some
of which bear little resemblance to the origi-
nal set of diseases that attracted Prusiner’s
attention. Notably, the structural predictions
of the prion model were verified for several
prionoids but not for prions. As such, many
of the questions raised by Prusiner in 1982—
prion structure, mechanism of replication,
and drivers of toxicity—are still open. Based
on historical evidence, addressing these
questions in the prion arena may, once again,
provide answers that will also apply to more
prevalent neurodegenerative diseases. j
REFERENCES AND NOTES
- S. B. Prusiner, Science 216 , 136 (1982).
- J. S. Griffith, Nature 215 , 1043 (1967).
- M. Prinz et al., Nature 425 , 957 (2003).
- H. Büeler et al., Cell 73 , 1339 (1993).
- G. P. Saborio, B. Permanne, C. Soto, Nature 411 , 810
(2001). - R. Hervas et al., Science 367 , 1230 (2020).
- J.-Y. Li et al., Nat. Med. 14 , 501 (2008).
- K. C. Luk et al., Proc. Natl. Acad. Sci. U.S.A. 106 , 20051
(2009). - F. Clavaguera et al., Nat. Cell Biol. 11 , 909 (2009).
- A. Küffer et al., Nature 536 , 464 (2016).
- G. Legname et al., Science 305 , 673 (2004).
- G. Spagnolli et al., PLOS Pathog. 15 , e1007864 (2019).
- E. Vázquez-Fernández et al., PLOS Pathog. 12 , e1005835
(2016). - S. Brandner et al., Nature 379 , 339 (1996).
- J. Li, S. Browning, S. P. Mahal, A. M. Oelschlegel, C.
Weissmann, Science 327 , 869 (2010).
ACKNOWLEDGMENTS
The authors thank K. Frontzek for the micrograph of a CJD
brain and G. Spagnolli for the PrPSc molecular model. The
Aguzzi lab is supported by the Nomis Foundation and the
Swiss National Science Foundation.
10.1126/science.abb8577
CANCERMutational selection in
normal urothelium
Mutations in normal tissue point to causes of
DNA damage and set the stage for cancer
By Steven G. RozenC
ells have elaborate machinery to pre-
serve the integrity of their genomes,
which nevertheless relentlessly gather
new mutations over time. Recent
technical advances have enabled
high-resolution delineation of these
accumulated mutations and their spatial or-
ganization in tissues ( 1 – 8 ). It is now possible
to deduce which mutations in which genes
allowed cells to outcompete their neighbors
and colonize nearby regions of normal tissue
(clonal expansion). One can also sometimes
ascertain which endogenous mutational pro-
cesses or external mutagens caused these so-
matic mutations. These recent findings have
profound implications for understanding
aging and the early stages of cancer initia-
tion. On pages 75 and 82 of this issue, Lawson
et al. ( 9 ) and Li et al. ( 10 ), respectively, delve
into the somatic mutations lurking in the
normal urothelium—the lining of the blad-
der and ureter—and relate them to cancers
in these tissues.
Although the findings of the two studies
are broadly consistent, substantial method-
ological differences account for some diver-
gence. Lawson et al. studied the urothelium
of the bladder, primarily in organ donors
from Europe, whereas Li et al. studied the
nonmalignant urothelium of both the blad-
der and the ureter in cancer patients from
China. Additionally, Lawson et al. studied
an average of ~100 tiny (~0.01 mm^2 ) urothe-
lial samples each from 15 organ donors and
5 cancer patients; by contrast, Li et al. stud-
ied an average of 1.3 large (~2 mm^2 ) samples
of normal tissue from 120 cancer patients.
The studies also differed in their sequencing
approaches. Lawson et al. sequenced a mix
of whole genomes, whole exomes, and a tar-
geted panel of 321 cancer-associated genes,
whereas Li et al. sequenced whole exomes.
Despite the methodological differences,
both studies analyzed somatic mutations in
normal urothelium to show that clonal ex-
pansion occurred. Both studies found that aselective advantage (indicated by high preva-
lence of a clone) usually stemmed from muta-
tions in genes encoding proteins involved in
histone modification and chromatin remod-
eling. Within these genes, truncating, and
presumably inactivating, mutations were
often strongly selected. The affected genes
prominently included KMT2D (histone-lysine
N-methyltransferase 2D) and KDM6A (lysine-
specific demethylase 6A). These chromatin
remodeling genes are also mutated in many
cancers of the urothelium, suggesting that
the nonmalignant expanded clones driven
by these genes regularly, but not frequently,
become malignant. This contrasts with the
esophagus, in which some genes that often
drive clonal expansion in the normal tissue
only rarely act as drivers in cancers (3, 6).
By examining many samples per indi-
vidual, Lawson et al. found that, in some
individuals, the urothelium showed strong
selective preference for mutations in partic-
ular genes. For example, one person had 35
distinct KDM6A mutations distributed over
multiple clones and 2 different ARID1A (AT-
rich interactive domain-containing protein
1A, also involved in chromatin remodeling)
mutations, whereas another person had 4
different KDM6A mutations and 20 different
ARID1A mutations distributed over multiple
clones. It was not possible to make such ob-
servations with the study design of Li et al.,
but with a larger sample of individuals, Li et
al. may have been better able to assess the
prevalence of mutations in different genes.
Both studies found that, with the excep-
tion of the chromatin remodeling genes,
most other genes that are commonly mu-
tated in urothelial cancers were rarely mu-
tated in normal urothelium. These include
the well-established cancer driver genes
PIK3CA (phosphatidylinositol 4,5-bisphos-
phate 3-kinase catalytic subunit a), FGFR3
(fibroblast growth factor receptor 3), and
RB1 (RB transcriptional corepressor 1). Thus,
mutations in these driver genes might be
later events that finally trigger malignant
transformation of cells harboring mutations
that drive clonal expansion (see the figure).
Both studies also found few large-scale copy
number alterations in the normal urothe-
lium, in contrast to the abundant large-scaleCentre for Computational Biology and Programme in
Cancer and Stem Cell Biology, Duke–National University of
Singapore (NUS) Medical School, 169857 Singapore. Email:
[email protected]34 2 OCTOBER 2020 • VOL 370 ISSUE 6512