The widely adopted model of ongoing
speciation-with-gene-flow for M and S (14) pos-
its that frequent hybridization leads to M-S ge-
nome homogenization in all except a few small
regions near centromeres (“speciation islands”),
contribute to differential fitness (i.e., ecological
and reproductive isolation). Detection of much
more widespread genomic divergence based on
genotyping (25) and whole genome sequencing
and the process of speciation more advanced, than
tification of genetic changes instrumental and not
divergence will be more difficult than initially
independently assembled M and S genomes and a
SNP genotyping array (25) are now available for
detecting morphologically cryptic vector subdivi-
sions, probing their molecular basis, and ultimately
developing innovative malaria interventions.
References and Notes
1. Y. F. Li, J. C. Costello, A. K. Holloway, M. W. Hahn,
Evolution 62, 2984 (2008).
2. H. D. Rundle, P. Nosil, Ecol. Lett. 8, 336 (2005).
3. A. della Torre et al., Insect Mol. Biol. 10, 9 (2001).
4. A. della Torre, Z. Tu, V. Petrarca, Insect Biochem.
Mol. Biol. 35, 755 (2005).
5. A. Diabatéet al., Proc. Biol. Sci. 276, 4215 (2009).
6. A. Diabatéet al., J. Med. Entomol. 43, 480 (2006).
7. F. Tripet et al., Mol. Ecol. 10, 1725 (2001).
8. B. Caputo et al., Malar. J. 7, 182 (2008).
9. E. Oliveira et al., J. Med. Entomol. 45, 1057 (2008).
10. C. Costantini et al., BMC Ecol. 9, 16 (2009).
11. A. Diabaté, R. K. Dabire, N. Millogo, T. Lehmann, J. Med.
Entomol. 44, 60 (2007).
12. T. Lehmann, A. Diabaté, Infect. Genet. Evol. 8, 737 (2008).
13. P. Nosil, D. J. Funk, D. Ortiz-Barrientos, Mol. Ecol. 18,
14. T. L. Turner, M. W. Hahn, S. V. Nuzhdin, PLoS Biol. 3,
15. B. J. White, C. Cheng, F. Simard, C. Costantini,
N. J. Besansky, Mol. Ecol. 19, 925 (2010).
16. M. Carneiro, N. Ferrand, M. W. Nachman, Genetics 181,
17. J. L. Feder, P. Nosil, Evolution 64, 1729 (2010).
18. M. Slatkin, Science 236, 787 (1987).
19. Materials and methods are available as supporting
material on Science Online.
20. R. A. Holt et al., Science 298, 129 (2002).
21. B. S. Weir, L. R. Cardon, A. D. Anderson, D. M. Nielsen,
W. G. Hill, Genome Res. 15, 1468 (2005).
22. J. M. Akey et al., Proc. Natl. Acad. Sci. U.S.A. 107,
23. W. Du et al., Insect Mol. Biol. 14, 179 (2005).
24. J. E. Mehren, Curr. Biol. 17, R240 (2007).
25. D. E. Neafsey et al., Science 330, 514 (2010).
26. We thank W. M. Gelbart, an early advocate of this project,
and J. L. Feder, P. Nosil, and M. W. Hahn for critical review.
P. Howell of MR4 provided mosquitoes. Funding for genome
sequencing of M (U54-HG00379) and S (U54-HG03068)
was provided by the National Human Genome Research
Institute. N.J.B. was supported by NIH (RO1 AI63508 and
AI076584). M.K.N.L. was supported by Biotechnology
and Biological Sciences Research Council research grant
BB/E002641/1 to G.K.C. Sequence data are deposited with
GenBank (accessions ABKP00000000 and ABKQ00000000).
Supporting Online Material
Materials and Methods
Figs. S1 to S8
Tables S1 to S5
28 July 2010; accepted 9 September 2010
SNP Genotyping Defines Complex
Gene-Flow Boundaries Among
African Malaria Vector Mosquitoes
D. E. Neafsey,1* M. K. N. Lawniczak,2* D. J. Park,1S. N. Redmond,2M. B. Coulibaly,3S. F. Traoré,3
N. Sagnon,4C. Costantini,5,6C. Johnson,1R. C. Wiegand,1F. H. Collins,7E. S. Lander,1
D. F. Wirth,1,8F. C. Kafatos,2N. J. Besansky,7G. K. Christophides,2M. A. T. Muskavitch1,8,9†
that promote malaria transmission and complicate vector control efforts. A high-density, genome-wide
mosquito SNP-genotyping array allowed mapping of genomic differentiation between populations and
species that exhibit varying levels of reproductive isolation. Regions near centromeres or within
polymorphic inversions exhibited the greatest genetic divergence, but divergence was also observed
elsewhere in the genomes. Signals of natural selection within populations were overrepresented
among genomic regions that are differentiated between populations, implying that differentiation is
often driven by population-specific selective events. Complex genomic differentiation among speciating
vector mosquito populations implies that tools for genome-wide monitoring of population structure
will prove useful for the advancement of malaria eradication.
morbidity and mortality are greatest. Population
nopheles gambiae is the primary vector
of human malaria in sub-Saharan Africa,
where annual burdens of malaria-induced
subdivision within A. gambiae is pervasive but
in the past. A. gambiae is composed of at least
two morphologically identical incipient species
known as the M and S molecularformsbasedon
fixed ribosomal DNA sequence differences (1).
The M and S forms are further divided by in-
forms, including Mopti (molecular form M),
Savanna (molecular form S), and Bamako (mo-
lecular form S), each of which we examine here,
and each of which has specialized for different
breeding sites (2, 3). Furthermore, A. gambiae
belongs to a species complex of seven recently
diverged, morphologically identical sibling taxa,
which we also examine here. Population sub-
division can increase disease transmission inten-
sity and duration, as new mosquito populations
evolve to exploit changing habitats and varied
seasonal conditions. Vector control efforts can be
complicated by population subdivision, because
populations vary for traits on which interventions
depend, such as indoor feeding behavior (4, 5)
and insecticide susceptibility (6).
Genes underlying epidemiologically relevant
phenotypic diversification among vector popula-
tions must reside within genomic regions that are
differentiated among those populations. Most
previous efforts to detect genetic differentiation
between mosquito populations have been unable
to localize differentiated regions, even when pop-
ulation divergence has been detected [for in-
stance, between S and Bamako (7)] or lacked
resolution to map all but the most highly differ-
entiated regions [for example, between M and S
(8, 9)]. High-resolution mapping of genomic
regions differentiated between vector populations
will advance our understanding of phenotypic
diversification. Furthermore, ongoing assessment
of gene flow among vector populations is essen-
tial for implementation of control measures de-
signed for natural genetic variants [for instance,
transgenic variants (11) within mosquito popula-
tions, as we strive yet again to eradicate malaria.
We used a customized Affymetrix single-
nucleotide polymorphism (SNP) genotyping ar-
ray to analyze 400,000 SNPs identified through
sequencing of the M and S incipient species of
A. gambiae (12). We hybridized individual arrays
females from the three known sympatric A. gam-
biae populations in Mali (M, S, and Bamako)
We then hybridized DNA pooled from the same
Center, Bamako, Mali.4Centre National de Recherche et For-
de Recherche pour le Développement, Unité de Recherche
la Lutte contre les Endémies en Afrique Centrale, Yaounde,
Cameroon.7University of Notre Dame, Notre Dame, IN 46556,
9Boston College, Chestnut Hill, MA 02467, USA.
*These authors contributed equally to this work.
†To whom correspondence should be addressed. E-mail:
22 OCTOBER 2010VOL 330
on October 26, 2010
degree to which quantitative differences in allele
frequencies could be assessed with the use of
pooled DNA. We also hybridized a pool of DNA
from 20 field-collected individuals of the sister
species A. arabiensis. Results obtained from
correlated (Pearson’s correlation coefficient r2=
0.96 for M, S, and Bamako comparisons) (fig.
S1), indicating that the majority of SNPs assayed
on the array yield useful quantitative information
regarding divergence in allele frequencies be-
tween pooled samples.
Pooled hybridization data revealed that the
greatest differentiation between the recently sub-
cluster of inversions on chromosomal arm 2R
(Fig. 1A). This pattern is concordant with models
of speciation in the face of ongoing gene flow,
which predict that early in the speciation process,
divergence will be localized to regions of low
recombination, such as inversions (16–20). In
partially reproductively isolated populations like
S and Bamako, these divergent genomic regions
are most likely to contain genes (table S2) di-
rectly responsible for differential niche adapta-
tion and reproductive isolation, whereas ongoing
gene flow should homogenize the remainder of
the genome (21–23).
The M and S mosquito populations in Mali
exhibit divergence that is much greater and more
S and Bamako (Fig. 1B). This might be expected
given the broader geographic ranges of M and S
relative to Bamako and their presumed longer
divergence time (2). We found that all pericentro-
between M and S (fig. S2), in accordance with
previous observations (8, 9, 24). However, we
unexpectedly detected shorter regions of substan-
tial differentiation at various distances from
centromeres along each chromosome. The exis-
tence of extensive divergence within nonpericen-
tromeric regions suggests that realized gene flow
between these two incipient species is low, de-
findings, which we obtained with the use of DNA
patterns observed in the sequencing-based SNP
analysis of M and S mosquito colonies (12).
colony-derived M mosquitoes from Cameroon to
between M and S is geographically restricted to
between M populations from Mali and Cam-
eroon than between S populations from these
locations, and it has been speculated that another
incipient speciation event may be occurring with-
in M (26). However, with the exception of the
2La inversion, we find extremely similar patterns
of differentiation between S and M, regardless of
the geographic origin of the M population that
was analyzed (Fig. 1C and fig. S3). This finding
suggests that the genomic regions differentiated
West and Central Africa and may harbor the
genes facilitating niche differentiation as well as
pre- and postmating isolation between these taxa.
However, the great extent of genomic divergence
the earliest stages of the M and S speciation pro-
cess will prove challenging.
species A. arabiensis, between which hybridiza-
tion can occur in nature, although it yields sterile
males (27). Because SNPs assayed on the array
are segregating in A. gambiae but may not be seg-
regating in A. arabiensis, we could not compare
the overall magnitude of genomic divergence be-
tween these taxa with the divergence between
forms of A. gambiae. To avoid bias, we undertook
that were found to exhibit similar allelic intensity
ratios in the M and S pools. This assay set was
sufficient to indicate that the profile of relative dif-
is less heterogeneous than that in the M versus S
comparison (Fig. 1D), even as it echoes some of
the same highly divergent regions. Chromosomes
differentiation, similar to the pattern we observed
between the M and S forms of A. gambiae, with
additional differentiation across the entire X chro-
mosome, presumably due to the large Xag inver-
inversion fixed in A. arabiensis relative to the
ancestral X arrangement (28, 29).
Although particular inversion arrangements
are not exclusive to any of the A. gambiae pop-
ulations that we profiled, these genomic regions
clearly harbor an excess of differentiation be-
tween populations compared with other regions
of the genome (Figs. 1 and 2). Inversions may be
hotspots for differentiation, even when main-
tained at similar frequencies in different popu-
lations, if recombination suppression facilitates
functional divergence of the inverted and wild-
within all three forms of A. gambiae is extremely
low, extending no more than a few thousand base
pairs (fig.S4). Therefore,groups of loci that reside
within regions of lower recombination in the
A. gambiae genome would be more likely to es-
tablish consistent patterns of cosegregation.
It is important to distinguish a difference in
inversion frequency between populations versus
differentiation of alternative inversion arrange-
analysis (PCA) of SNP genotypes within inver-
sion boundaries indicates that, although the S and
Bamako populations harbor different frequencies
of the 2Rj, 2Rb, 2Rc, and 2Ru inversions (table
S3), the b arrangement of 2Rb is divergent be-
S and Bamako, it is differentiating independently
within each population. Similarly, both arrange-
ments of 2Rb, as well as the uninverted arrange-
M and S (Fig. 2 and table S3). However, the in-
verted 2La arrangement is an exception to this
pattern: M, S, and Bamako mosquitoes homozy-
gous for the 2La inversion exhibit much less di-
vergence between the 2La breakpoints than is
observed in the same three populations for all
other inversions (Fig. 2A). The close clustering of
Fig. 1. Relativelocaldiver-
comparisons of mosquito
z scores (standard devia-
tions) and scaled so that 0
reflects the modal diver-
ference in allelic intensity
ratios measured over adja-
cent 50 SNP stepping win-
dows. The colored regions
labeled with letters repre-
of A. gambiae from Mali.
(B) Divergence between A.
gambiae M-form mosqui-
from Mali. (C) Divergence
between M-form mosqui-
toes from Cameroon and
S-form mosquitoes from
Mali. (D) Divergence be-
tween A. arabiensis from
Burkina Faso and A. gam-
biae from Mali.
Relative Local Divergence (Z score; mode = 0)
Mali S vs. Mali Bamako
2R 2L3R 3LX
b c u
Mali M vs. Mali S
2R 2L 3R3LX
Cameroon M vs. Mali S
A. gambiae vs.
VOL 330 22 OCTOBER 2010
on October 26, 2010
arrangement) with individuals homozygous for
S, and Bamako) supports earlier hypotheses re-
garding introgression between species within this
region (30). Indeed, the region within the 2La
inversion breakpoints shows divergence between
A. gambiae and A. arabiensis that is lower than
light the degree of similarity within each of these
partially isolated populations. With the exception
of 2Rj, no inversion is diagnostic of a particular
population in our sample. However, the consist-
ently independent clustering of M, S, and Bamako
mosquitoes by PCA across all inversions except
2La affirms the legitimacy and genetic distinc-
tiveness of these groups.
We next examined the data for signals of nat-
ural selection. The genomic regions exhibiting
greatest divergence (Fst> 0.6) between M and S
exhibit significantly reduced polymorphism in
P = 1.14 × 10–47(1.14E-47); S: one-tailed t test,
P = 1.88E-120], as might be expected if differ-
by polymorphism-eliminating selective sweeps
(31). To explore selection more deeply, we anal-
yzed SNP calls from individual hybridizations of
M, S, and Bamako mosquitoes with the use of
uates the likelihood of a sweep within a particular
genomic region, given the allele frequency spec-
trum of local SNPs. Several genomic regions ap-
pear to have experienced recent sweeps within
each of the three forms (Fig. 3). The pericentro-
meric regions of all three chromosomes exhibit
the strongest signals of selective sweeps for M
and S, suggesting that the extensive divergence
observed in these regions has been driven by
selection (Fig. 3). Indeed, the degree of con-
cordance between the profiles of selection and
differentiation for M and S [chi-squared test, P=
2.4E-104 (14)] implies a causal role for selection
within some genomic regions where differentia-
tion is observed. Additionally, the fact that dif-
ferent regions of the genomes of M, S, and
Bamako show evidence of selective sweeps sug-
gests that these populations are experiencing dif-
ferent selective pressures that shape genetic
variation independently (Fig. 3).
Analysis for functional enrichment (15)
among 68 genes found in candidate sweep re-
gions identified two interesting categories of
genes significantly overrepresented after correc-
tion for multiple testing: (i) multicellular organis-
mal development (P = 9.1E-4; five genes), and (ii)
serine-type endopeptidase activity (P = 2.6E-2;
nine genes, five of which occur in a pericentro-
meric cluster on 3L). Of the five genes annotated
as being involved in development, three encode
homeodomain-containing transcriptional regula-
tors (AGAP004659, sweep in S; AGAP004660,
sweep in S; AGAP004696, sweep in Bamako),
one encodes a member of the Hedgehog signal-
ing pathway (AGAP004637, sweep in S), and one
encodes a member of the Wnt signaling pathway
lie ecological niche differentiation and/or repro-
ductive isolation mechanisms that reinforce the
The gene encoding CPF3, a cuticular protein spec-
a pericentromeric sweep region in S on chromo-
some 2L. CPF3 is the gene exhibiting the most
significant difference in expression between M
upon mating (34). These combined observations
motivate further investigation of CPF3 and its
Table S2 presentsafulllistofthe536genesfound
in differentiated and/or sweep regions. Among
are significantly overrepresented (X total = 173;
chi-squared test; P < 2.2E-16).
Our findings demonstrate the power of high-
resolution SNP arrays for mapping genetic di-
vergence among vector mosquito taxa within the
differentiation between distinct populations and
selective sweeps within populations is valuable
for identifying and monitoring alleles that me-
diate traits critical for malaria transmission and
vector control. The differentiated genomic re-
harbor genes (table S2) of epidemiological im-
portancefor disease transmission, including loci
influencing reproduction, longevity, insecticide
resistance, aridity tolerance, larval habitat, and
other traits that differ among mosquito popula-
Fig. 2. PCA plots of the 2La (A), 2Rj (B), 2Rb
(C), 2Rc (D), and 2Ru (E) inversion regions.
Circled regions indicate groups of mosquitoes
homozygous (ii) or heterozygous (i+) for the in-
version or homozygous for the wild-type ar-
22 OCTOBER 2010VOL 330
on October 26, 2010
References and Notes Download full-text
1. A. della Torre et al., Insect Mol. Biol. 10, 9 (2001).
2. T. Lehmann, A. Diabate, Infect. Genet. Evol. 8, 737
3. N. C. Manoukis et al., Proc. Natl. Acad. Sci. U.S.A.
105, 2940 (2008).
4. W. Takken, Trop. Med. Int. Health 7, 1022 (2002).
5. H. D. Trung et al., Trop. Med. Int. Health 10, 251 (2005).
6. C. Fanello et al., Insect Mol. Biol. 12, 241 (2003).
7. M. A. Slotman et al., Am. J. Trop. Med. Hyg. 74, 641 (2006).
8. T. L. Turner, M. W. Hahn, S. V. Nuzhdin, PLoS Biol. 3,
Mol. Ecol. 19, 925 (2010).
10. A. Enayati, J. Hemingway, Annu. Rev. Entomol. 55,
11. V. Koufopanou, M. R. Goddard, A. Burt, Mol. Biol. Evol.
19, 239 (2002).
12. M. K. N. Lawniczak et al., Science 330, 512 (2010).
13. M. Coluzzi, A. Sabatini, A. della Torre, M. A. Di Deco,
V. Petrarca, Science 298, 1415 (2002); published online
3 October 2002 (10.1126/science.1077769).
14. A. della Torre, Z. Tu, V. Petrarca, Insect Biochem. Mol. Biol.
35, 755 (2005).
15. Materials and methods are available as supporting
material on Science Online.
16. M. Coluzzi, Prog. Clin. Biol. Res. 96, 143 (1982).
17. M. A. Noor, K. L. Grams, L. A. Bertucci, J. Reiland,
Proc. Natl. Acad. Sci. U.S.A. 98, 12084 (2001).
18. L. H. Rieseberg, Trends Ecol. Evol. 16, 351 (2001).
19. A. Navarro, N. H. Barton, Evolution 57, 447 (2003).
20. R. K. Butlin, Mol. Ecol. 14, 2621 (2005).
21. C. I. Wu, C. T. Ting, Nat. Rev. Genet. 5, 114 (2004).
22. C. A. Machado, R. M. Kliman, J. A. Markert, J. Hey,
Mol. Biol. Evol. 19, 472 (2002).
23. S. Via, Proc. Natl. Acad. Sci. U.S.A. 106 (suppl. 1),
24. A. D. Stump et al., Proc. Natl. Acad. Sci. U.S.A. 102,
25. F. Tripet et al., Mol. Ecol. 10, 1725 (2001).
26. M. A. Slotman et al., Mol. Ecol. 16, 639 (2007).
27. C. J. Curtis, in Recent Developments in the Genetics of
Disease Vectors, W. M. Steiner, Ed. (Stripes Publishing,
Champaign, IL, 1982), pp. 290–312.
28. N. J. Besansky et al., Proc. Natl. Acad. Sci. U.S.A. 100,
29. M. Slotman, A. Della Torre, J. R. Powell, Genetics 167,
30. B. J. White et al., PLoS Genet. 3, e217 (2007).
31. J. M. Smith, J. Haigh, Genet. Res. 23, 23 (1974).
32. R. Nielsen et al., Genome Res. 15, 1566 (2005).
33. B. J. Cassone et al., Mol. Ecol. 17, 2491 (2008).
34. D. W. Rogers et al., Proc. Natl. Acad. Sci. U.S.A. 105,
35. This work was supported by Burroughs Wellcome Fund
Request 1008238, the Broad Institute Director’s Fund, the
Wellcome Trust Programme grant 077229/Z/05/Z to F.C.K.
and G.K.C., National Human Genome Research Institute
support to E.S.L., the Harvard School of Public Health
Department of Immunology and Infectious Diseases, and the
DeLuca Professorship from Boston College to M.A.T.M. In
addition, M.K.N.L. was supported by a Biotechnology and
Biological Sciences Research Council research grant
Information Short Read Archive with accession numbers
SRX005397 to SRX005403. SNPs have been submitted to
dbSNP Build B133 (ss/rs numbers in progress) and,
meanwhile, are listed in the supporting online
material. ArrayExpress accession number: A-AFFY-167.
Supporting Online Material
Materials and Methods
Figs. S1 to S5
Tables S1 to S3
dbSNP Accession Numbers
1 June 2010; accepted 9 September 2010
ATM Activation by Oxidative Stress
Zhi Guo,1Sergei Kozlov,2Martin F. Lavin,2Maria D. Person,3Tanya T. Paull1*
The ataxia-telangiectasia mutated (ATM) protein kinase is activated by DNA double-strand breaks
(DSBs) through the Mre11-Rad50-Nbs1 (MRN) DNA repair complex and orchestrates signaling
cascades that initiate the DNA damage response. Cells lacking ATM are also hypersensitive to
insults other than DSBs, particularly oxidative stress. We show that oxidation of ATM directly
induces ATM activation in the absence of DNA DSBs and the MRN complex. The oxidized form
of ATM is a disulfide–cross-linked dimer, and mutation of a critical cysteine residue involved
in disulfide bond formation specifically blocked activation through the oxidation pathway.
Identification of this pathway explains observations of ATM activation under conditions of oxidative
stress and shows that ATM is an important sensor of reactive oxygen species in human cells.
mature aging, and a high incidence of lymphoma
(1). These defects may be connected through dys-
indicated by observations that mammalian cells
atients with ataxia-telangiectasia (A-T)
lack functional A-T mutated (ATM) pro-
and hypersensitivity to agents that induce oxida-
tive stress (2). Lymphoma incidence and the loss
of hematopoietic stem cells that occurs in mice
lacking ATM can be suppressed by antioxidants
(3–5), indicating an important role for ATM in
regulating cellular defenses against redox stress.
ATM is activated in response to changes in in-
clear what the initiating event is in these cases or
how this relates to Mre11-Rad50-Nbs1 (MRN)–
mediated ATM activation that is dependent on
DNA double-strand breaks (DSBs).
To address these questions, we used primary
human fibroblasts and induced oxidative stress
with H2O2or DSBs with bleomycin (11). Auto-
phosphorylation of ATM on Ser1981, phospho-
rylation of the tumor suppressor p53 on Ser15,
and phosphorylation of the protein kinase Chk2
on Thr68all occurred in response to H2O2or to
bleomycin treatment (Fig. 1, A to C). Phospho-
1Howard Hughes Medical Institute, Department of Molecular
Genetics and Microbiology, and Institute for Cellular and Mo-
lecularBiology (ICMB), UniversityofTexas atAustin,Austin, TX
78712, USA.2Radiation Biology and Oncology Laboratory,
Queensland Institute of Medical Research, and School of
3ICMB Analytical Instrumentation Facility Core, College of Phar-
macy, University of Texas at Austin, Austin, TX 78712, USA.
*To whom correspondence should be addressed. E-mail:
Fig. 3. Profiles of ge-
nomic regions subject to
recent selective sweeps
ue of a selective sweep
icant signals of selection
correction) are indicated
−log(P value) Sweep
VOL 33022 OCTOBER 2010
on October 26, 2010