news and viewsnews and views
volume 43 | number 3 | mArCH 2011 | nature genetics
oncogenes. These data, in combination with
additional genetic studies, strongly suggest that
Jnk1 and Jnk2 perform redundant functions
downstream of Mkk7 in phosphorylating p53 in
response to oncogenic stress (Fig. 1). An inter-
esting question that emerges from these studies
is whether a specific isoform of Mkk7 is respon-
sible for the transmission of the oncogenic stress
signal to Jnk and p53. Map2k7, which encodes
Mkk7, is alternatively spliced to produce six dis-
tinct kinase isoforms, each with differing kinase
activity and affinity for Jnk6. Moreover, Mkk7
is activated through phosphorylation of its
activation loop by any of a number of Map3ks
(Fig. 1). The upstream kinases that are respon-
sive to oncogenic stress remain unknown.
Pathways to oncogenic stress
The normal cellular response to oncogenic
stress requires the tumor suppressor protein
p53. Nevertheless, the mechanisms linking
oncogene activation to p53 induction have
remained controversial. Evidence from stud-
ies of early-stage human tumors and animal
models suggests that oncogene-induced repli-
cation stress activates a DNA damage response
(DDR), which in turn activates p53 (Fig. 1)7–9.
The implication of these studies is that p53-
dependent tumor suppression in response to
oncogenic stress acts through the DDR. An
alternative view, also supported by compel-
ling evidence, suggests that the acute response
to DNA damage is not the critical tumor
suppressive function of p53. Elegant studies
performed by Christoporou et al. have shown
that only delayed activation of p53 (several days
after acute DNA damage) can protect mice from
later development of radiation-induced cancer10.
Although the tumor suppressor function of p53
is dependent on the upstream factor Arf, the
stabilization of p53 downstream of the DDR is
Arf independent11,12. How do the findings by
Schramek et al. fit into this picture? Their study
shows that Mkk7 is required for activation of
p53 in response to oncogenic stress and also in
response to DNA damage4. It is possible, how-
ever, that oncogenic stress and DNA damage
feed into p53 independently (Fig. 1). Indeed,
the authors observed that acute K-Ras activa-
tion in vitro was not sufficient to activate the
DDR, whereas it did result in upregulation of
p53 target genes. It would be of interest to deter-
mine whether or not Arf functions to regulate
the stabilization of p53 downstream of Mkk7
and Jnk in response to oncogenic stress.
OIS-induced activation of Mkk7
How does a cell sense oncogenic stress and sub-
sequently activate Mkk7? One possibility, as
suggested by Schramek et al. is that Mkk7, or an
upstream Map3k, is directly activated by DNA
damage, similar to other kinases that phos-
phorylate p53 to enhance its stability (Fig. 1).
Alternately, Mkk7 could be activated in an
indirect way, for example, through a feedback
pathway. Indeed, Mkk7-Jnk signaling is highly
activated by cytokines, such as IL-1α and Tnfα
(ref. 13). Moreover, cells experiencing onco-
genic stress establish a cytokine milieu by tran-
scriptionally upregulating an inflammatory
gene expression program14,15. IL-1α is among
the cytokines that are dramatically induced by
oncogenic stress15. As a result, Mkk7 could
operate downstream in an autocrine-paracrine
cytokine signaling pathway that promotes OIS
(Fig. 1). Although further studies are required
to determine precisely how oncogenic stress
leads to p53 stabilization and subsequent cell
cycle arrest, it is clear that the Jnk signaling
pathway is a key player in this phenomenon.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
1. Serrano, M. et al. Cell 88, 593–602 (1997).
2. Michaloglou, C. et al. Nature 436, 720–724 (2005).
3. Dankort, D. et al. Nat. Genet. 41, 544–552 (2009).
4. Schramek, D. et al. Nat. Genet. 43, 212–219 (2011).
5. Bode, A.M. & Dong, Z. Mol. Carcinog. 46, 591–598
6. Tournier, C. et al. Mol. Cell. Biol. 19, 1569–1581
7. MacPherson, D. et al. EMBO J. 23, 3689–3699
8. Bartkova, J. et al. Nature 434, 864–870 (2005).
9. Chao, C. et al. EMBO J. 25, 2615–2622 (2006).
10. Christophorou, M.A. et al. Nature 443, 214–217
11. Efeyan, A. & Serrano, M. Cell Cycle 6, 1006–1010
12. Meek, D.W. Nat. Rev. Cancer 9, 714–723 (2009).
13. Tournier, C. et al. Genes Dev. 15, 1419–1426
14. Kuilman, T. et al. Cell 133, 1019–1031 (2008).
15. Coppé, J.P. et al. PLoS Biol. 6, 2853–2868 (2008).
treat a child with more or less intensive ther-
apy based on clinical parameters. The deter-
minants for risk categorization have expanded
from clinical observations, such as leukocyte
count at diagnosis, age at diagnosis and ethnic
background, to molecular analyses, such as
DNA index, T lymphocyte lineage, the pres-
ence of specific chromosomal translocations
in leukemic blasts and evidence of minimal
residual disease early in therapy. The studies
used by Yang et al.2 to investigate the role of
genetic ancestry on treatment outcome were
designed to evaluate response to late inten-
sification of therapy. Notably, Yang et al.2
found that one additional round of therapy
designated as Hispanic have a higher risk for
relapse of ALL1. On page 237 of this issue2,
Mary Relling and colleagues use genome-wide
SNP genotypes and principal component ana-
lysis3 to explore the effects of genetic ancestry
on the risk of ALL relapse (Fig. 1). They report
that children estimated to have 10% or more
Native American ancestry have a higher rate of
relapse than children from other backgrounds,
with an estimated effect roughly similar to that
of self-reported Hispanic ancestry.
Admixture mapping and beyond
Today, pediatric oncologists must balance
efficacy and toxicity in deciding whether to
Over the last 40 years, there has been a steady
improvement in survival rates for children with
acute lymphoblastic leukemia (ALL), the most
common form of pediatric cancer. Though it is
viewed as a ‘curable’ cancer by some, this is not
true for all children, as about 20% of affected
children will suffer relapse and eventually die
from this condition. For several decades, pedi-
atric oncologists have recognized that children
a twist on admixture mapping
Stephen J Chanock
a new study uses genome-wide snP genotypes to identify a subset of children undergoing therapy for acute lymphoblastic
leukemia that are at increased risk for relapse. Borrowing from the classical approach of admixture mapping, the work
shows how genome-wide assessment of genetic ancestry can be used as a biomarker for disease outcome.
Stephen J. Chanock is at the Laboratory of
Translational Genomics, Division of Cancer
Epidemiology and Genetics, National Cancer
Institute, Bethesda, Maryland, USA.
© 2011 Nature America, Inc. All rights reserved.
news and views
nature genetics | volume 43 | number 3 | mArCH 2011
Although this study illustrates a possible
twist on the use of genome-wide SNP geno-
typing, more work is clearly needed to trans-
late these findings into short-term clinical use.
Ultimately, the mapping and functional study of
regions influencing risk of relapse could result
in a series of highly specific tests. If this is real-
ized, the broad-based assessment of ancestry
could become inconsequential. Nonetheless,
this study is a step in the right direction.
COMPETING FINANCIAL INTERESTS
The author declares no competing financial interests.
1. Kadan-Lottick, N.S. et al. J. Am. Med. Assoc. 290,
2. Yang, J.J. et al. Nat. Genet. 43, 237–241 (2011).
3. Price, A.L. et al. Nat. Genet. 38, 904–909 (2006).
4. Rife, D.C. Am. J. Hum. Genet. 6, 26–33 (1954).
5. Smith, M.W. et al. Am. J. Hum. Genet. 74, 1001–1013
6. Tian, C. et al. Am. J. Hum. Genet. 79, 640–649
7. Price, A.L. et al. Am. J. Hum. Genet. 80, 1024–1036
8. Mao, X. et al. Am. J. Hum. Genet. 80, 1171–1178
9. Winkler, C.A. et al. Annu. Rev. Genomics Hum. Genet.
11, 65–89 (2010).
10. Treviño, L.R. et al. Nat. Genet. 41, 1001–1005
11. Chanock, S.J. et al. Nature 447, 655–660 (2007).
12. Price, A.L. et al. PLoS Genet. 5, e1000519 (2009).
13. Tang, H. et al. Am. J. Hum. Genet. 81, 626–633
14. Zaitlen, N. et al. Am. J. Hum. Genet. 86, 23–33 (2010).
15. Naqvi, M. et al. J. Asthma 44, 639–648 (2007).
risk, further work is required to conclusively
establish the significance of local, informa-
tive markers before turning to the underlying
biology of the causal variants11. Moreover, it is
optimal to examine control sets comprised of
individuals of comparable ancestry to exclude
candidate regions falsely identified by chance
or by incompletely controlling for recent
evolutionary history12,13. Still, independent
validation is daunting because of a paucity of
admixed children undergoing ALL therapy. By
leveraging the genetic variability across popula-
tions, follow-up studies should be able to use
the distinct differences between ancestries
to localize actual regions influencing risk of
relapse and to more efficiently map the variants
responsible for the signal(s)14. To pursue candi-
date regions, collaborative networks are needed
to pool studies, particularly to attain sufficient
numbers of underrepresented groups, such as
individuals with Native American ancestry.
Indeed, there should be a concerted effort
to conduct additional studies across Latin
America. In many ways, this challenge is simi-
lar to the issues facing the study of less common
diseases or disease subtypes, such as estrogen
receptor–negative breast cancer or ethnicity-
specific response to therapy (for example,
asthma in Latino/Hispanic populations)15.
Because often we do not fully understand
the biological basis of a biomarker before
applying it in the clinical venue, it makes sense
to proceed cautiously and determine whether
genome-wide SNP arrays, or perhaps a well-
chosen set of ancestry-informative markers,
can serve as a clinical test with greater speci-
ficity than self-reported ancestry. Certainly,
we have to ask if the findings of Yang et al.2
can be replicated in an independent set of
studies, an important benchmark in translat-
ing promising results into clinical practice.
Lastly, how will we educate health providers
to wisely interpret and convey the results of
broad ancestry-based genetic analyses to
families? This latter point is especially sensi-
tive and may stretch the limits of social toler-
ance. However, when faced with life and death
choices, such as the treatment options for chil-
dren with ALL, particular care will need to be
exercised in communicating the interpretation
of this type of ‘genome-wide test’.
Figure 1 Principal component analysis of genome-
wide SNP genotyping data can distinguish an
admixed population from other populations. The
admixed population is shown in green. The two
ancestral populations contributing to the admixed
population are shown in red and blue. A more
distantly related population is shown in black.
mitigated the reported risk associated with
Native American ancestry.
Since the 1950s, when Rife proposed using
linkage disequilibrium caused by recent popu-
lation admixture to map traits, geneticists have
sought to develop tools to find genetic loci that
contribute to differences in diseases or traits
that vary in prevalence between ancestral
populations4. Comprehensive annotation of
common genetic variants across the genome
has led to the generation of admixture maps,
specifically for African Americans and Latino/
Hispanics5–8. Mapping by admixture linkage
disequilibrium (MALD) has been success-
ful in discovering risk loci for a few diseases,
but the results thus far do not fully explain
the observed differences in disease preva-
lence between populations. Although there
are notable regions identified by MALD
that include body mass index, hypertension,
end-stage renal disease, multiple sclerosis
and prostate cancer, they have been primar-
ily reported in African Americans9. Mapping
loci by admixture in Latino/Hispanics has had
limited success, despite strong evidence for an
increased prevalence for common conditions
such as type 2 diabetes and hypertension in
Some of the children analyzed in the study
by Yang et al.2 overlap with those analyzed in
a previous genome-wide association study that
uncovered several regions associated with risk
for pediatric ALL10. In the current study, the
authors turned admixture mapping on its side
and entered different waters, analyzing ances-
try as a genome-wide biomarker. The findings
are perhaps fortuitous because it is likely that
they set out, as is the case in a genome-wide
association study, to find a handful of regions
that could explain the observed differences
in response across the study. Instead, they
reverted to looking at the big picture and
identified Native American genetic ancestry
as a possible new clinical characteristic. This
is a lesson to us all to keep looking at our data
in new ways.
Although the study by Yang et al.2 points to
a few candidate regions that might contribute
to the ancestry-related differences in relapse
© 2011 Nature America, Inc. All rights reserved.