news and views
volume 44 | number 6 | june 2012 | nature genetics
castration-resistant metastatic tumors, suggest-
ing that dysregulation of the androgen recep-
tor pathway could be an early event in prostate
tumorigenesis. Genes for other transcriptional
regulators, such as the transcriptional initiation
component MED12, are also mutated in both
stages of disease.
ETS fusion–negative tumors
Over half of prostate cancers bear transloca-
tions involving the genes of ETS family tran-
scription factors, such as ERG and ETV1 (refs.
8,9). This has led to the idea that ETS fusion–
positive tumors form a distinct subgroup6, but
the molecular basis of disease in ETS fusion–
negative cancers is less clear. Some ETS fusion–
negative tumors have translocations in genes
in the Raf kinase pathway10 or have outlier
expression of SPINK1 (ref. 11), but the genetic
mechanisms remain unexplained in many of
these cancers. Barbieri et al. find that muta-
tions in SPOP, present in 6–15% of primary
tumors, are mutually exclusive from ETS
translocations and likely explain a significant
fraction of ETS fusion–negative cancers. SPOP
encodes an E3 ubiquitin ligase component12
and is also mutated in colorectal cancer in
sequences encoding the distinct BTB dimeriza-
tion domain. SPOP mutations in prostate
cancer exclusively affect the substrate-binding
MATH domain, implicating SPOP as a tumor
suppressor, as the mutated protein binds more
weakly to substrate in vitro. However, SPOP
copy-number loss is rarely, if ever, observed
in prostate cancer, raising the possibility that
these mutations may confer de novo gain of
function. Alteration in the CHD1 gene encod-
ing a chromatin-remodeling factor defines
another ETS fusion–negative class of prostate
cancer2, as this gene was previously found to
be mutated or rearranged in three out of seven
prostate cancer whole genomes sequenced3.
Collectively, tumors with mutated SPOP and
CHD1 account for a substantial fraction of ETS
fusion–negative prostate cancers.
With a fairly complete list of prostate cancer
genome alterations spanning the full spec-
trum of disease now in hand, it is time to ask
the critical question of whether this informa-
tion can guide a more precise approach to the
diagnosis and treatment of primary disease.
Earlier data comparing copy-number altera-
tions in primary and metastatic disease found
that late-stage tumors resemble a subset of
highly altered primary tumors and that altered
copy number in primary disease was associated
with a greater risk of relapse5. Comparison of
copy-number alterations observed in primary
versus castration-resistant disease by Barbieri
et al. and Grasso et al., respectively, supports the
observation of increased copy-number alteration
in prostate cancer, albeit without clinical data to
conclude association with outcome (Fig. 1). The
whole-exome mutation data from these new
reports provide additional variables that should
be examined in future studies of prognosis,
together with mRNA expression profiles recently
reported to correlate with outcome7,13. Future
work may allow patients with high-risk disease
to be treated more aggressively, perhaps with
novel agents targeted at relevant driver lesions,
whereas patients with low-risk disease might be
watched for signs of progression without treat-
ment. The wider lens provided by these reports
therefore offers the hope of tailoring treatment
of prostate cancer on the basis of genomic risk
and the presence of driver lesions.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
1. Barbieri, C.E. et al. Nat. Genet. 44, 685–689 (2012).
2. Grasso, C.S. et al. Nature published online,
doi:10.1038/nature11125 (20 May 2012).
3. Berger, M.F. et al. Nature 470, 214–220 (2011).
4. Robbins, C.M. et al. Genome Res. 21, 47–55 (2011).
5. Taylor, B.S. et al. Cancer Cell 18, 11–22 (2010).
6. Demichelis, F. et al. Genes Chromosom. Cancer 48,
7. Cuzick, J. et al. Lancet Oncol. 12, 245–255 (2011).
8. Tomlins, S.A. et al. Nature 448, 595–599 (2007).
9. Shen, M.M. & Abate-Shen, C. Genes Dev. 24, 1967–
10. Palanisamy, N. et al. Nat. Med. 16, 793–798 (2010).
11. Tomlins, S.A. et al. Cancer Cell 13, 519–528 (2008).
12. Furukawa, M., He, Y.J., Borchers, C. & Xiong, Y. Nat.
Cell Biol. 5, 1001–1007 (2003).
13. Cuzick, J. et al. Br. J. Cancer 106, 1095–1099
et al.3 in this issue report that our DNA changes
in subpopulations of cells over time and that
these changes may predict the subsequent
development of cancer.
Over 100 years ago, Wilhelm Roux and
August Weismann independently asserted
that the particles of heredity are differentially
apportioned during embryonic divisions
to give rise to genetically different cells that
have unique roles in development, an idea
they called mosaicism4,5. Later studies largely
rejected this hypothesis, showing that somatic
tissues arise from the differential use of genes
that are shared across all cells. With notable
exceptions of specialized processes such as the
V(D)J recombination that occurs in certain
immune cells6, examples of mosaicism have
largely been limited to congenital genetic ill-
ness7 and recognized cancer8,9.
What do the trillions of cells comprising an
individual human have in common? Aside
from the commensal organisms that colonize
the body, the prevailing answer to this ques-
tion has been the human genome. In normal
parlance, as well as in the scientific literature1,
we speak of a genome sequence as belonging
to a given person and existing in all of that
individual’s cells. But how accurate is this
notion? Studies by Laurie et al.2 and Jacobs
exploring the variation within
Evan Z Macosko & Steven A McCarroll
we usually think of an individual’s cells as sharing the same genome. Challenging this notion, two new studies show
that somatic mosaicism is common and can be an early herald of cancer.
Evan Z. Macosko and Steven A. McCarroll
are in the Department of Genetics at Harvard
Medical School, Boston, Massachusetts, USA,
and in the Program in Medical and
Population Genetics at the Broad Institute
of MIT and Harvard, Cambridge,
More common than we think
Recently, however, several studies have
hinted that somatic mosaicism is more wide-
spread than previously thought. One study of
human embryos fertilized in vitro found that
70% incurred segmental imbalances post-
meiotically10. Another study detected mosa-
icism in 1.7% of blood and buccal genomic
DNA samples11. And earlier this year, somatic
variation in blood samples from a twin cohort
identified copy-number differences in 3% of
twin pairs12. These results suggest that detect-
able clonal mosaicism occurs in a measurable
proportion of adults over 50 years of age.
Just how pervasive is clonal mosaicism in the
general population? In at least one tissue, the
blood, the collection of samples for genome-
wide association studies (GWAS) enables sur-
veys of genomic mosaicism in large population
© 2012 Nature America, Inc. All rights reserved.