Downloaded from genome.cshlp.

org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

The genome of the leaf-cutting ant Acromyrmex echinatior

suggests key adaptations to advanced social life and fungus
farming
Sanne Nygaard, Guojie Zhang, Morten Schiøtt, et al.

Genome Res. 2011 21: 1339-1348 originally published online June 30, 2011
Access the most recent version at doi:10.1101/gr.121392.111

Supplemental http://genome.cshlp.org/content/suppl/2011/05/27/gr.121392.111.DC1.html
Material
References This article cites 96 articles, 44 of which can be accessed free at:
http://genome.cshlp.org/content/21/8/1339.full.html#ref-list-1
Email alerting Receive free email alerts when new articles cite this article - sign up in the box at the
service top right corner of the article or click here

To subscribe to Genome Research go to:

http://genome.cshlp.org/subscriptions

Copyright © 2011 by Cold Spring Harbor Laboratory Press

 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

Research

The genome of the leaf-cutting ant Acromyrmex echinatior

suggests key adaptations to advanced social life
and fungus farming
Sanne Nygaard,1,9,11 Guojie Zhang,2,9 Morten Schiøtt,1,9 Cai Li,2,9 Yannick Wurm,3,4
Haofu Hu,2 Jiajian Zhou,2 Lu Ji,2 Feng Qiu,2 Morten Rasmussen,5 Hailin Pan,2
Frank Hauser,6 Anders Krogh,5,7,8 Cornelis J.P. Grimmelikhuijzen,6 Jun Wang,2,7,10,11
and Jacobus J. Boomsma1,10
1
Centre for Social Evolution, Department of Biology, University of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen,
Denmark; 2BGI-Shenzhen, Shenzhen 518083, China; 3Department of Ecology and Evolution, University of Lausanne, 1015 Lausanne,
Switzerland; 4Vital-IT Group, Swiss Institute of Bioinformatics, 1015 Lausanne, Switzerland; 5Centre for GeoGenetics, Natural History
Museum of Denmark, Øster Voldgade 5-7, 1350 Copenhagen, Denmark; 6Centre for Functional and Comparative Insect Genomics,
Department of Biology, University of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen, Denmark; 7Department of Biology,
University of Copenhagen, Copenhagen DK-2200, Denmark; 8Biotech Research and Innovation Center, University of Copenhagen,
Copenhagen DK-2200, Denmark

We present a high-quality (>1003 depth) Illumina genome sequence of the leaf-cutting ant Acromyrmex echinatior, a model
species for symbiosis and reproductive conflict studies. We compare this genome with three previously sequenced genomes
of ants from different subfamilies and focus our analyses on aspects of the genome likely to be associated with known
evolutionary changes. The first is the specialized fungal diet of A. echinatior, where we find gene loss in the ant’s arginine
synthesis pathway, loss of detoxification genes, and expansion of a group of peptidase proteins. One of these is a unique
ant-derived contribution to the fecal fluid, which otherwise consists of ‘‘garden manuring’’ fungal enzymes that are un-
affected by ant digestion. The second is multiple mating of queens and ejaculate competition, which may be associated with
a greatly expanded nardilysin-like peptidase gene family. The third is sex determination, where we could identify only
a single homolog of the feminizer gene. As other ants and the honeybee have duplications of this gene, we hypothesize that
this may partly explain the frequent production of diploid male larvae in A. echinatior. The fourth is the evolution of
eusociality, where we find a highly conserved ant-specific profile of neuropeptide genes that may be related to caste
determination. These first analyses of the A. echinatior genome indicate that considerable genetic changes are likely to have
accompanied the transition from hunter-gathering to agricultural food production 50 million years ago, and the transition
from single to multiple queen mating 10 million years ago.
[Supplemental material is available for this article.]

Active food production through farming is a landmark of human dependent on specific fungal symbionts for producing most of
cultural evolution (Diamond 2002), but has also evolved in a di- their food (Schultz et al. 2005; Aanen et al. 2009).
verse array of other organisms purely by natural selection (Mueller The fungus-growing ants have become an ecological and
et al. 2005; Boomsma 2011). Some of these represent forms of evolutionary model system of major importance. Just as in the
husbandry (e.g., Dictyostelium slime moulds) (Brock et al. 2011) or macrotermitine termites (Aanen et al. 2002), attine ant fungus
lack control over crop transmission (Littoraria snails) (Silliman and farming has a single origin ;50 MYA ago in South America
Newell 2003, and Stegastes damselfish) (Hata et al. 2010). Others, (Schultz and Brady 2008). The transition to active food pro-
such as the multiple lineages of ambrosia beetles have evolved duction was so successful that the tribe presently has at least 220
more complex fungus-farming systems, with extensive cotrans- species in 11 generally recognized genera, without any known
mission and mutual coadaptation between farmers and crops case of successful reversion to a hunter–gatherer lifestyle (Schultz
(Farrell et al. 2001). However, only the fungus-growing ants and and Brady 2008). The crown group of the Atta and Acromyrmex
termites have developed large-scale societies that are completely leaf-cutting ants in particular stands out as an impressive example
of social evolution. The common ancestor of these ants lived only
10 million years ago and achieved several major evolutionary
9 transitions roughly at the same time (Villesen et al. 2002), in-
These authors contributed equally to this work.
10
These authors contributed equally to this work. cluding: (1) large-scale use of live plant material to manure fungal
11
Corresponding authors. gardens; (2) extensive differentiation of worker castes to optimize
E-mail [email protected]. division of labor in foraging and conveyor-belt processing of
E-mail [email protected].
Article published online before print. Article, supplemental material, and pub- leaves, flowers, and fruits; (3) long-lived colonies with tens of
lication date are at http://www.genome.org/cgi/doi/10.1101/gr.121392.111. thousands of workers in Acromyrmex, and some millions in Atta;

21:1339–1348 Ó 2011 by Cold Spring Harbor Laboratory Press; ISSN 1088-9051/11; www.genome.org Genome Research 1339
www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

Nygaard et al.

(4) multiple mating of queens, most likely to ensure higher ge- when unrelated mycelia interact directly and when the fecal
netic diversity among workers, enabling more robust collective droplets that the ants use to manure their gardens come into
performance and higher resistance toward infectious disease contact with alien garden symbionts (Bot et al. 2001; Poulsen and
(Hughes and Boomsma 2004, 2006). Boomsma 2005; Ivens et al. 2009). (6) A substantial proportion of
Several draft genomes of ants have recently become available fungal enzymes and other proteins are not digested by the ants, but
(Bonasio et al. 2010; Wurm et al. 2011), addressing some of the deposited with the fecal droplets and mixed with the new leaf
fundamental genomic changes that accompanied the evolution of substrate in the growing top sections of fungus gardens (Schiøtt
eusocial colony life. These analyses generally use the genome of the et al. 2008, 2010), indicating that the obligate dependence on
honeybee, representing an independent hymenopteran lineage a fungal symbiont has induced prudent harvesting practices that
that evolved eusocial organization (Honeybee Genome Sequencing are likely to have left genomic signatures. (7) Haplodiploid sex
Consortium 2006), and three species of Nasonia parasitoid wasps determination normally implies that there are one or several sex-
(Werren et al. 2010) for comparison. With the present study we add determining loci of high heterozygosity (Hasselmann et al. 2008), so
another ant genome to the publicly available databases, providing that homozygote individuals meant to be females develop as ster-
some comparisons with the genome of the fire ant Solenopsis invicta ile diploid males. A. echinatior queens are known to produce many
(Wurm et al. 2011), which belongs to the same subfamily (Myrmi- diploid male larvae, but workers remove them early in development
cinae) as Acromyrmex, and with representatives of two other ant (Dijkstra and Boomsma 2007). This implies that detailed knowledge
subfamilies: Ponerinae and Formicinae (Bonasio et al. 2010). How- of sex-determination genes in this species, relative to social insects
ever, the major interest of sequencing the genome of Acromyrmex that do not suffer high potential fitness loads due to diploid male
echinatior is that this Panamanian ant species is a model system for production, will be rewarding. (8) Leaf-cutting ants have undergone
many key questions in evolutionary ecology, which will become major changes in behavior and physiology after they adopted fun-
accessible for genomic, transcriptomic, and proteomic studies now gus farming relative to their ancestors that have remained hunter–
that a reference genome is available. gatherers. Neuropeptides play essential roles in information pro-
Many aspects of the biology of A. echinatior have become cessing at all metabolic levels (Hauser et al. 2010), so that genomic
documented in recent years: (1) All queens mate with many males analyses of these key substances might provide important insights.
so that colonies are always chimaeras of approximately five to 15 The objective of this study was, therefore, to probe the A.
patrilines (Sumner et al. 2004). Competition between ejaculates echinatior genome for the key aspects of biological function out-
mediated by seminal fluid interactions has been demonstrated lined above. In particular, we provide in-depth analyses of de-
(Den Boer et al. 2010), providing interesting opportunities for toxification pathways in relation to homogeneous fungal food,
elucidating the molecular mechanisms involved in sperm com- arginine biosynthesis, peptidase gene family expansions, neuro-
petition. (2) Caste differentiation into small workers, large workers, endocrinology, and the sex-determining locus.
and gynes (future queens) is phenotypically plastic (Hughes et al.
2003; Hughes and Boomsma 2004), but some of the rare patrilines
Results and Discussion
cheat by being over-represented among the new queens that
a colony produces, either in association with over-representation The genome of A. echinatior (Fig. 1A) was obtained using the Illu-
among the small workers or the large workers (Hughes et al. 2003; mina HiSeq platform, which yielded 60.7 Gb of raw reads from
Hughes and Boomsma 2008). (3) The species is facultatively po- males of a single colony, and assembled with SOAPdenovo (Li et al.
lygynous and has a closely related inquiline social parasite, Acro- 2010), generating 300 Mb of assembled genome sequence (see
myrmex insinuator, which shares a direct common ancestor with Supplemental Tables S1, S2). The N50 scaffold length is 1.1 Mb,
its host (Schultz et al. 1998; Bekkevold and Boomsma 2000; longer than reported for other ant genomes (Supplemental Table
Sumner et al. 2003a,b, 2004). (4) Explicit studies on the heritability S3), with an average sequencing depth of 123 3 (Supplemental Fig.
of disease resistance have shown that the genetic diversity ema- S1), ensuring high accuracy at the nucleotide level.
nating from multiple queen-mating is likely to be adaptive Using flow cytometry, the total genome size of A. echinatior has
(Hughes and Boomsma 2004, 2006, 2008). (5) The fungal symbi- previously been estimated to be 335 Mb (Sirviö et al. 2006). Based
onts of this species are known to express incompatibilities both on k-mer coverage (Li et al. 2009a), we get a similar estimate of 313

Figure 1. The leafcutter ant A. echinatior and annotation of its protein-coding genes. (A) A winged male of the Panamanian leaf-cutting ant A. echinatior
in the fungus garden that is maintained by his major- and minor-worker sisters. (B) The total of 17,278 annotated protein-coding genes as obtained from
de novo predictions, GLEAN acceptance, homology (to C. floridanus, H. saltator, A. mellifera, N. vitripennis, D. melanogaster, C. elegans, or H. sapiens) and
transcriptome evidence. (Photo courtesy of David R. Nash Ó 2010.)

1340 Genome Research

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

The genome of an Acromyrmex leaf-cutting ant

Mb, suggesting that the assembled genome is 96% complete. Any lies to the assembly. The only GO category enriched in A.echinatior-
missing regions are likely to consist of repetitive sequences that specific genes was ‘‘transition metal ion binding’’ (Supplemental
cannot easily be assembled with current methods. The A. echinatior Table S7), whereas GO categories associated with CD27 receptor
genome is AT-rich, with a GC content of 46.9% in protein-coding binding, transferase activity, and nucleic acid binding were
exons, and an overall GC content of 33.7%. Both genome size and enriched in the Myrmicinae-specific gene families. In the gene
GC content were within the range reported for other ant genomes families where A. echinatior has lost homologs, enriched categories
(Supplemental Table S3; Bonasio et al. 2010; Wurm et al. 2011). A included spliceosome assembly and zinc ion binding (Supple-
total of 27.6% (82.6 Mb) of the sequenced genome was classified as mental Table S7).
repetitive sequence, based on both known and ab initio repeat li- Analysis of gene family expansion or contraction allows in-
braries (Supplemental Table S4), an estimate close to the recently ferences about the particular challenges that a species has faced or
published sequence data for the ponerine ant Harpegnathos saltator been released from over evolutionary time. We used the phylogenetic
(Bonasio et al. 2010), but lower than the more closely related gene family modeling pipeline CAFE (De Bie et al. 2006) to identify
S. invicta genome (Wurm et al. 2011) and higher than that of significantly (P < 0.001) expanded or contracted gene families in A.
Camponotus floridanus (Supplemental Table S3; Bonasio et al. 2010). echinatior and/or S. invicta myrmicine ants, relative to the other
Transcript data were obtained by Illumina sequencing of RNA eusocial hymenopteran genomes (Supplemental Fig. S3). Of the 5134
from a pooled sample of different castes and developmental stages. gene families that contained at least one sequence from each of these
We then annotated 17,278 protein-coding genes by using GLEAN five genomes, 10 were specifically expanded in A. echinatior, and
(Elsik et al. 2007) to integrate de novo gene predictions, tran- another 10 were contracted (Supplemental Table S8). Four families
scriptome evidence, and BLAST (Altschul et al. 1997) homology were found to be expanded in both representatives of the Myrmici-
information. Eighty-four percent of the annotated genes were nae, and three were contracted (Supplemental Table S9).
supported by transcript evidence, and 70% showed homology Families of olfactory receptors were identified among the
with known genes in other species (Fig. 1B; Methods). We manu- contracted gene families in both A. echinatior and S. invicta, al-
ally verified more than 200 of the gene models. though olfactory receptors have generally been expanded in ants
To assess the completeness of the annotation, we used the (Bonasio et al. 2010; Wurm et al. 2011). However, these genes are
CEGMA (Parra et al. 2007) set of 458 core eukaryotic genes. Almost difficult to annotate automatically, so more work will be needed to
all of these (449; 98%) were found in our gene set, again con- interpret these differences, as it is possible that other subfamilies of
firming the completeness of the genome. We furthermore anno- these receptors have expanded to compensate for losses elsewhere.
tated 316 predicted tRNA genes, 58 rRNA genes, 29 snRNAs, and 93 Some other expanded or contracted gene families were examined
miRNAs (Supplemental Table S5). The latter is close to the number in more detail, and will be discussed in the sections below.
of miRNAs reported in C. floridanus (96) (Bonasio et al. 2010), al-
though more extensive sequencing of short RNAs may increase Changes in detoxification pathways
this number further. BLAST searches revealed some contigs/scaf-
Insects harbor a range of enzymes to catalyze the detoxification
folds of bacterial origin (Supplemental Table S6). Most notably,
and neutralization of toxins that are ingested with food or other-
we saw evidence of two strains of Wolbachia endobacteria in A.
wise encountered in the environment. It has previously been
echinatior, consistent with previous findings (Van Borm et al. 2003).
reported that ants, in particular C. floridanus, show an expansion of
detoxification-related genes, which was attributed to their typical
Genomic analyses and comparisons across the ant phylogeny life style as generalized predators, aphid herders, and scavengers
(Bonasio et al. 2010). Since leafcutter ants rely on a symbiotic
To assess the functional changes that have occurred in A. echinatior
fungus as an almost exclusive food source, it would seem plausible
during its specialization to a fungus-farming life style, we com-
that they ingest fewer toxins, and thus need a smaller repertoire of
pared the predicted genes of A. echinatior with those of three
detoxification genes. We assessed whether any such adaptations
other ants: S. invicta, C. floridanus, and H. saltator, as well as to the
might have evolved by comparing the number of genes with pre-
more distantly related honeybee (Apis mellifera), parasitoid wasp
dicted protein domains corresponding to different central classes
(Nasonia vitripennis), and fruit fly (Drosophila melanogaster). For
of detoxification enzymes across the ants for which genomic in-
each genome, we assigned gene function (see Methods), followed
formation is available (Supplemental Table S10).
by gene clustering across all species. This generated 11,848 gene
We found a marked reduction in the number of cytochrome
families, 1995 of which contained exactly one sequence from ev-
P450 monooxygenase genes: A. echinatior only has 73 genes pre-
ery species. Fourfold degenerate codon positions were extracted
dicted to contain the CytP450 domain, while the other investigated
from these and used to construct a phylogenetic tree with PhyML
ants have 95–132. This reduction of CytP450 containing genes in A.
(Supplemental Fig. S2; Guindon et al. 2009), confirming that the
echinatior resembles the low number (60) observed in the honeybee,
two ants of the subfamily Myrmicinae (A. echinatior and S. invicta)
which also lives on a specialized diet of toxin-free food. Less-pro-
are most closely related, followed by C. floridanus and H. saltator, as
nounced reductions were found in the number of genes containing
expected from presently available phylogenies of all ants (Brady
the carboxy/cholinesterase and UDP-glucoronosyltransferase do-
et al. 2006; Moreau et al. 2006).
mains (Supplemental Table S10). In contrast to these reductions,
We calculated gene ontology (GO) enrichment in A. echinatior-
the number of gluthathione S-transferase genes appears slightly
specific genes, in myrmicine subfamily-specific gene families
higher in A. echinatior than in other ants or the honeybee.
(containing sequences from both A.echinatior and S. invicta, but
not from H. saltator or C. floridanus), and in gene families where the
homolog has been lost by A. echinatior (sequences from both S. Arginine metabolism
invicta and at least one of H. saltator and C. floridanus, but not Obligate mutualistic symbioses with a history of 50 million years
A. echinatior). For the latter group, we verified the absence of A. (Schultz and Brady 2008) can be expected to have evolved some
echinatior homologs by realigning sequences from the gene fami- degree of division of labor in the acquisition of essential amino

Genome Research 1341

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

Nygaard et al.

acids. This is particularly likely because Acromyrmex and Atta leaf- Nardilysin from rats has been shown to cut peptides at dibasic
cutter ants rear a highly derived and specialized fungal symbiont sites, suggesting a prohormone convertase function, i.e., cleaving
throughout their range (Mikheyev et al. 2008). Examples from a prohormone target into mature hormone(s) (Chesneau et al.
other symbioses showing similar adaptions are Euprymna squid 1994). Mammalian nardilysin is particularly abundant in the adult
supplying their symbiotic bioluminescent Vibrio bacteria with testes (Hospital et al. 1997; Fumagalli et al. 1998), and one of the
amino acids to support their growth (Graf and Ruby 1998) and two homologous genes in D. melanogaster is highly expressed
aphids relying on Buchnera endosymbiont to provide essential in the testes and the male accessory glands (www.flyatlas.org)
amino acids (Klasson and Andersson 2004). Camponotus ants are (Chintapalli et al. 2007). It is thus tempting to speculate that the
also known to rely on intracellular Blochmannia bacteria for the M16 expansion in A. echinatior has some connection to the high
production of essential amino acids (Feldhaar et al. 2007). degree of multiple queen mating and the ensuing ejaculate com-
We used BLAST (Altschul et al. 1997) to assign genes to petition after insemination, as the other ants used in our analysis
functional categories defined in the Kyoto Encyclopedia of Genes appear to have singly mated queens (Boomsma et al. 2009; Den
and Genomes (KEGG categories) (Kanehisa and Goto 2000) and Boer et al. 2010). The nardilysin gene group is not expanded in
examined the annotations to identify metabolic pathways where the honeybee, which also has multiple queen mating, but many
A. echinatior had an apparent loss-of-function compared with the other genes known to be involved in the sperm and accessory
other three ants for which genomes were available: S. invicta, C. gland secretion of Apis honeybees (Baer et al. 2009a,b) are also
floridanus, and H. saltator. We found that two genes involved in present in A. echinatior (Supplemental Tables S12, S13). This sug-
arginine biosynthesis appear to have been specifically lost in A. gests that many key elements of sperm transfer and sperm storage
echinatior: Argininosuccinate synthase (EC:6.3.4.5), which catalyzes are conserved and have been maintained in two hymenopteran
the conversion of aspartate and citruline into argininosuccinate, clades, where multiple queen mating evolved independently from
and argininosuccinate lyase (EC:4.3.2.1), which catalyzes the sub- single-mating ancestors, but that the mechanisms behind specific
sequent conversion of argininosuccinate to arginine and fumarate adaptions or ejaculate competition are unlikely to be the same
(Fig. 2). These losses are not due to missing sequence data in the (Boomsma 2009; Den Boer et al. 2010).
genome that we obtained, as we were able to identify the pseudo- A significant (P < 0.001), but less pronounced expansion was
gene for argininosuccinate synthase (Supplemental Table S11). This observed for a family of peptidase M14 proteins (Fig. 3B), where A.
is consistent with the evolution of a significant metabolic division echinatior has eight members, while the other ant species have four
of labor between the ants and their fungal crop, as functional ar- or five. All of these proteins belong to the CPA and CPB subfamily
ginine pathways appear to be present in the related Agaricales fungi of metallocarboxypeptidases that require metal ions (often zinc) as
Laccaria bicolor and Coprinopsis cinerea (KEGG, release 56.0). cofactors and cleave amino acids from the carboxy end of poly-
peptides (Reznik and Fricker 2001; Rodriguez de la Vega et al.
2007). Some of these enzymes work as digestive enzymes with
Expansion of peptidase gene families broad substrate specificity, while others activate or deactivate spe-
The most pronounced expansion was observed in a family of cific proteins by removing amino acids. D. melanogaster has 18
predicted peptidase M16 genes, where A. echinatior has 16 mem- genes that cluster in the same family, of which 13 are primarily
bers, while the other investigated Hymenoptera have two (S. expressed in the larval or adult digestive system, while the re-
invicta, H. saltator) or three (A. mellifera, C. floridanus, N. vitripennis). maining five have a broader expression, including high expression
The phylogenetic tree of this gene family (Fig. 3A) shows that the levels in the testes and spermatheca (Chintapalli et al. 2007,
genes cluster in two groups. One group contains genes that encode www.flyatlas.org; Tweedie et al. 2009, FlyBase.org). One of the A.
proteins similar to insulin degrading enzymes, and has only one to echinatior proteins (EGI65848) has been found in the fecal fluid of
two members in all insect genomes investigated, while the other the ants (M Schiøtt, unpubl.), which suggests a digestive role for
group contains genes encoding proteins similar to nardilysin, and these peptidases, either in the ant gut or in the fungus garden after
it is this group that is greatly expanded in A. echinatior. Peptidase defecation (Schiøtt et al. 2010).
M16 proteins have a conserved motif with the amino acid se-
quence HXXEH, which is believed to be the active site (Becker
and Roth 1992). However, most of the extra M16 proteins in A. Sex determination
echinatior seem to lack this motif, so it is possible that they no In ants and other Hymenoptera, males develop from unfertilized
longer function as active peptidases. eggs, whereas fertilized eggs develop into females, either workers or
queens. In honeybees, sex is determined
by the highly variable complementary sex
determiner (csd ) locus, which controls the
alternative splicing of the homologous
feminizer transcript, the actual effector of
sex determination (Hasselmann et al.
2008). The presence of two different csd
alleles therefore induces female de-
velopment, whereas a single copy or
two identical copies produce male phe-
notypes. Diploid males are occasionally
observed in ants, particularly under in-
Figure 2. Missing genes in the arginine biosynthesis pathway. The specific loss in A. echinatior of two
breeding (Kronauer et al. 2007), suggest-
genes that encode enzymes catalyzing two consecutive (final) steps in the biosynthesis of the amino acid
arginine. Enzymes are denoted by purple boxes with the EC numbers inside. Pale purple boxes with ing that a mechanism similar to csd also
dashed red borders indicate the two lost or pseudogenized genes. applies in many ants. Consistent with

1342 Genome Research

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

The genome of an Acromyrmex leaf-cutting ant

Figure 3. Peptidase expansions in the genome of A. echinatior. (A) Expansion of the M16 peptidase gene family with the insulin degrading enzyme,
present in one or two copies in all investigated insect genomes (below dotted line), and nardilysin genes (above dotted line). The A. echinatior genes in each
group are highlighted in yellow. Bootstrap support values >60% are given. (B) Expansion of the M14 peptidase gene family, with a dotted line separating
two subfamilies. The A. echinatior genes are highlighted in yellow. Bootstrap support values >60% are given. Phum was included to increase resolution of
this tree. Species (A. echinatior: Aech; H. saltator: Hsal; C. floridanus: Cflo; S. invicta: Sinv; A. mellifera: Amel; D. melanogaster: Dmel; N. vitripennis: Nvit;
Pediculus humanus: Phum) and GenBank ID are given for each sequence.

this notion, all ant genomes sequenced so far have produced two The fact that the sex-determination system of A. echinatior can
homologs of the feminizer gene (Bonasio et al. 2010; Wurm et al. apparently work with a single feminizer homolog seems to shed
2011). However, phylogenies suggest that the duplications in bees some doubt on whether the independent duplications of feminizer
and ants are independent (Bonasio et al. 2010; Wurm et al. 2011), in other ants are necessary for primary sex determination analo-
making it possible that the double feminizer homologs in ants and gous to the way in which the csd locus functions in the honeybee.
bees function differently. On the other hand, the high production of diploid male larvae in
In contrast to other ants, sequence searches revealed only A. echinatior (Dijkstra and Boomsma 2007) could be an effect of
a single homolog of feminizer in the A. echinatior genome (Sup- having only one feminizer homolog. Workers of A. echinatior cull
plemental Fig. S4). While it remains possible that another homo- almost all diploid males and presumably do so early in larval
log exists, it would be unlikely to have been missed with >1003 development, so that the potentially high cost of diploid male
coverage, and we found no indications that the single locus results production is almost always low in practice. Culling behavior
from a misassembly of two distinct genomic loci (Supplemental may thus have evolved to compensate for an inefficient sex-
Fig. S4). The inferred genome sequence also completely matches determination system.
the RNA-seq sequence as well as a previously obtained A. echinatior
fem cDNA sequence from another ant colony (data not shown),
making it unlikely that this genome region is not correctly as- Neuropeptides
sembled. The RNA-seq data furthermore indicate alternative Neuropeptides are short peptides that interact with membrane-
splicing (Supplemental Fig. S4), increasing the likelihood that this bound receptors, typically G protein-coupled receptors (Hauser
gene is a de facto feminizer homolog. et al. 2006, 2008, 2010) and steer important physiological

Genome Research 1343

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

Nygaard et al.

processes such as development, reproduction, feeding, and be- by gene expression studies that will be much more feasible with
havior. In insects there are 46 known neuropeptide genes, and the availability of a reference genome.
a recent study revealed that 20 of them (the core set) are consis-
tently found in all insects with a sequenced genome (Hauser et al.
2010; Supplemental Table S14). The remaining 26 neuropep-
Methods
tide genes (the variable set) constitute family-, genus-, or species- Biological material and DNA/RNA extraction
specific neuropeptide gene profiles (Hauser et al. 2010; Supple-
mental Table S15). Because these neuropeptide gene profiles are A monogynous, queen-right colony of A. echinatior (Ae372) was
collected in Gamboa, Panama in 2008, and maintained in the lab
likely to be related to variation in habitat, diet, or behavior across
on a diet of bramble leaves and rice at 25°C and 60%–70% RH.
insect clades (Hauser et al. 2010), we expected to find differences
Voucher specimens of this colony have been deposited at the
both between the four ant genomes and between ants and other
Zoological Museum, Copenhagen. Since male ants are derived
insects. However, we found that all four ant species share exactly
through arrhenotokous parthenogenesis, a single male contains
the same neuropeptide gene profile (Supplemental Tables S16, one haploid genome, whereas a pool of males from a monogynous
S17). This profile is unique for ants and differs from that of other colony represents the diploid genome of that colony’s mother
hymenopterans such as the honeybee and Nasonia wasps (Sup- queen.
plemental Table S18). These results indicate that the four ant spe- Short insert libraries (500 bp, 800 bp, and 2 kb) for Illumina
cies have a single common ancestor, consistent with other recent sequencing were made from extracted DNA of a single male of
molecular evidence that all ants form a monophyletic clade (Brady colony Ae372 using the Qiagen DNeasy Blood and Tissue Kit and
et al. 2006; Moreau et al. 2006), and that their neuroendocrinol- the protocol provided by the manufacturer. Long insert libraries (5,
ogy, despite their striking differences in habitat, feeding, and be- 10, and 20 kb) for Illumina sequencing were made from the DNA of
havior, has a very similar basic set-up. This remarkable conserva- 40 males from colony Ae372. This DNA was extracted by grinding
tion across a family of insects that diverged >100 MYA suggests the ants in 5 mL of lysis buffer (10 mM Tris-HCl, 400 mM NaCl, 2
a link with the early evolution of eusociality in ants. mM EDTA at pH 8) and adding 200 mL of 20% SDS, 300 mL of 5 mg/
Another surprise, when carrying out this comparative endo- mL proteinase K, and 10 mL of 100 mg/mL RNase A. After a 2-h
crine genomics analysis, was the absence of the RYamide gene in all incubation at 50°C, 5 mL of phenol/chloroform (1:1 pH 8) was
four ant genomes (Supplemental Table S16). The RYamide genes added and the sample was mixed on a slowly rotating wheel for 15
have only recently been discovered and were found to be present min. The sample was subsequently centrifuged at 3000g for 10
in all insects with a sequenced genome (the ‘‘core set’’) (Hauser min, after which the upper phase was transferred to a new tube and
mixed with 5 mL of phenol/chloroform (1:1 pH 8). After mixing
et al. 2010) and even in other arthropods, such as crustaceans and
for 15 min on a slowly rotating wheel, the sample was centrifuged
chelicerates (F Hauser, unpubl.). Their absence in ants is a unique
for another 10 min at 3000g, after which the upper phase was
joint feature and difficult to interpret at present. The biological
transferred to a new tube and mixed with 5 mL of chloroform/
action of insect RYamides is unknown, but mass spectrometry has
isoamyl alcohol (24:1). After another round of 15 min of mixing on
shown that they are abundant in the terminal ganglion of mos- a slowly rotating wheel, the sample was centrifuged again for 10
quitoes (Hauser et al. 2010), which, in general, innervates the min at 3000g, after which the upper phase was transferred to a new
sexual organs in insects. The absence of the RYamide neuropep- tube and mixed with 0.5 volume isopropanol. The tube was then
tides in ants might therefore be related to specific features of ant inverted 10–20 times before the precipitated DNA was collected
reproduction and possibly to the regulatory mechanisms that with a glass hook made of a pasteur pipette. The DNA was dissolved
evolved together with the differentiation between reproductive in 1 mL of TE buffer overnight.
and nonreproductive castes. Total RNA was extracted from eggs, larvae, pupae, workers,
males, and gynes from colony Ae372 by grinding the tissue in
liquid nitrogen and purifying the RNA with a Qiagen RNeasy mini
Concluding remarks kit using Qiagen buffer RLC instead of buffer RLT. A pooled sample
The genome sequence of A. echinatior shows that attine fungus- from all developmental stages was then generated to contain ap-
growing ants have great potential as model systems for studying proximately equimolar amounts of RNA from each stage.
symbiosis and reproductive conflict in the genomics era. The
present study only allowed us to explore some of the significant
DNA sequencing
changes that a history of 50 million years of obligate fungus
farming can be expected to have induced. All of these will have to For Illumina DNA sequencing, five paired-end sequencing libraries
be elaborated by detailed experimental work to clarify the ways in were constructed with insert sizes of ;500 bp, 800 bp, 2 kb, 5 kb,
and 10 kb. For small insert libraries, 5 mg of DNA was sheared to
which regulation of gene expression operates, a research agenda
fragments of 500–800 bp, end-repaired, A-tailed, and ligated to
that will have to involve the total complexity of the symbiotic
Illumina paired-end adapters (Illumina). The ligated fragments
syndrome of fungus farming (e.g., Sen et al. 2009; Poulsen and
were size selected at 500 and 800 bp on agarose gels and amplified
Currie 2010; Schiøtt et al. 2010; Suen et al. 2010).
by LM-PCR to yield the corresponding short insert libraries. For
The attractiveness of this model system is that it is an ecto- long insert size mate-pair library construction, 20–40 mg of geno-
symbiosis, in which hosts and symbionts can be separated and mic DNA was sheared to the desired insert size using nebulization
maintained in independent lab cultures for considerable periods of for 2 kb or HydroShear (Digilab) for 5 and 10 kb.
time, and where cross-fostering and other manipulation experi- The DNA fragments were end-repaired using biotinylated
ments are feasible (Armitage et al. 2011). These opportunities have nucleotide analogs (Illumina), size selected at 2, 5, and 10 kb. and
already been amply used, not only for A. echinatior (see the eight- circularized by intramolecular ligation. Circular DNA molecules
point summary above), but also for a number of other attine ant were sheared with Adaptive Focused Acoustic (Covaris) to an av-
model systems (e.g., Mueller et al. 2005; Schultz and Brady 2008; erage size of 500 bp. Biotinylated fragments were purified on
De Fine Licht et al. 2010). This kind of work can now be elaborated magnetic beads (Invitrogen), end-repaired, A-tailed, and ligated to

1344 Genome Research

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

The genome of an Acromyrmex leaf-cutting ant

Illumina paired-end adapters, size-selected again, and purified by solexa1.3-quals’’), which identifies exon–exon splice junctions.
LM-PCR. After their construction, the libraries were sequenced on Then Cufflinks (Trapnell et al. 2010) was used to assemble tran-
the Illumina HiSeq 2000, with read lengths of 90–100 bp. Raw scripts with junction information (parameters ‘‘-m 20 -s 10 -I
sequences were filtered for low quality, adapter sequence, paired- 10000’’).
end read overlap, and PCR duplicates. Sequence homology was assigned based on best TBLASTN
hits (E-value <1310 5) to gene sets from seven different species:
RNA sequencing C. floridanus (OGS v3.3), H. saltator (OGS v3.3), A .mellifera
(Amel_4.0), N. vitripennis (Nvit_1.0) downloaded from NCBI, D.
Total RNA was further purified using TRIzol (Invitrogen), and
melanogaster, C. elegans, and Homo sapiens downloaded from
poly(A) RNA was isolated with oligo-dT-coupled beads from 20 mg
Ensembl (release 56). This was followed by detailed protein–nu-
of total RNA. First-strand cDNA synthesis was performed with
cleotide alignment with Genewise (version 2.0) (Birney et al. 2004)
random hexamers and Superscript II reverse transcriptase (Invi-
to generate the gene structures.
trogen). The second strand was synthesized with E. coli DNA PolI
Two de novo prediction programs, Augustus (Stanke et al.
(Invitrogen). Double-stranded cDNA was purified with a Qiaquick
2006) and SNAP (Korf 2004), were used to predict genes, with pa-
PCR purification kit (Qiagen) and sheared with a nebulizer (Invi-
rameters trained on 500 randomly selected intact genes (full ORF
trogen) to 100–500-bp fragments. After end repair and addition of
found, including start and stop codon) from the homology-based
a 39dA overhang, the cDNA was ligated to Illumina PE adapter oligo
predictions. The evidence derived from homology-based (seven
mix, and size selected to 200 6 20-bp fragments by gel purification.
sets for seven species), de novo (two sets for two programs), and
After PCR amplification the libraries were sequenced using
expression (one set) were integrated by GLEAN (Elsik et al. 2007) to
Illumina HiSeq 2000 and the paired-end sequencing module. For
generate a consensus gene set.
smRNA-seq, we gel-purified 18–30 nt RNAs from the samples uti-
In addition to the GLEAN-predicted genes, we expanded the
lized for RNA-seq. Illumina 59 and 39 RNA adapters were sequen-
gene set by adding gene models based on homology only: The
tially ligated to the RNA fragments and the ligated products were
seven homology-based gene-prediction sets were merged into
size-selected on denaturing polyacrylamide gels. The adapter-linked
a union set. For overlapping gene models, the longest was se-
RNA was reverse transcribed with small RNA RT primers and am-
lected. These gene models were filtered by: (1) removing genes
plified with PCR using small RNA PCR primers 1 and 2 (Illumina).
containing premature stop codons inside the coding region; (2)
The libraries were sequenced with Illumina HiSeq 2000.
removing genes with more than two frameshifts; (3) removing
genes with a Genewise score <80. For Genewise incomplete pro-
Genome assembly, filtering, and repeat identification tein alignments, we extended the first exon upstream to find the
The genome was assembled using SOAPdenovo (Li et al. 2010) to start codon, and extended the last exon downstream to find the
first construct contigs based on the short insert libraries, then stop codon.
joining these to scaffolds using paired-end information, followed by The gene models from the two de novo methods were simi-
local reassembly of unresolved gap regions. The sequencing cover- larly merged, and the longest model was chosen in cases of overlap.
age of the assembled genome sequence was evaluated by mapping Gene models that did not pass the GLEAN criteria were still in-
the raw sequencing data back to the scaffolds using SOAPaligner cluded if their expression level exceeded 5% of the whole genome
(Li et al. 2009b), after which the coverage was calculated based on level. All gene models were subsequently filtered for transposable
the k-mer distribution as described in Li et al. (2009a). elements (TEs) by running InterProSscan (Zdobnov and Apweiler
The RepeatMasker program (version 3.2.6) (Smit et al. 1996) 2001) (see also below) while removing genes matching TE-related
was used to identify: (A) noninterspersed repeat sequences by us- protein domains. Manual checking of gene models was then per-
ing the ‘‘-noint’’ option, including Simple_repeat, Satellite, and formed using the Apollo Annotation editor (Lewis et al. 2002).
Low_complexity repeats; (B) known transposable elements from
the Repbase 15.02 Transposable Element (TE) library (Jurka et al. CEGMA validation, noncoding RNA annotation,
2005); (C) additional high and medium copy number repeats (>10 and functional annotation of protein-coding genes
copies) in the assembly, based on an ab initio repeat library con-
structed with RepeatScout (Price et al. 2005) with default param- The CEGMA set of 458 core eukaryotic genes (CEGs) was used to
eters. In addition, we predicted tandem repeats using TRF (Benson predict genes in A. echinatior using the CEGMA pipeline (Parra et al.
1999) with parameters set to ‘‘Match=2, Mismatch=7, Delta=7, 2007). Predicted CEGMA genes were considered present in our
PM=80, PI=10, Minscore=50, and MaxPeriod=12’’. gene set if they overlapped with existing gene models.
The assembled genome was BLAST searched against the NCBI We predicted tRNAs with tRNAscan-SE (Lowe and Eddy 1997)
nt database, and contigs that aligned to bacterial sequences with using default eukaryote parameters, and identified rRNAs by
>90% identity across >90% of their length were filtered out. All aligning invertebrate rRNA sequences (downloaded from NCBI,
predicted gene models were BLAST searched against the NCBI nr http://www.ncbi.nlm.nih.gov/) to the assembly using BLASTN
database, and contigs/scaffolds were removed if >75% percent of with an E-value of 1310 5. The snRNA genes were predicted using
their genes had highest similarity to bacterial sequences. The the Infernal software (Nawrocki et al. 2009) against the Rfam da-
remaining assembled genome was realigned against the removed tabase (Griffiths-Jones et al. 2005, release 9.1). To reduce the speed
contigs/scaffolds, and contigs/scaffolds with >80% similarity required for computation, a rough prefiltering was performed be-
across >80% of the sequence were removed. No sequences of fore running Infernal by BLASTN alignment of the assembly
fungal origin were identified. Bacterial scaffolds were reannotated against the Rfam sequence database with an E-value cut-off of 1.
using GeneMark.hmm-p (Lukashin and Borodovsky 1998) and We predicted miRNAs based on both homology and expres-
removed for all subsequent analyses. sion. A set of known miRNAs was constructed by merging the
previously identified miRNAs from H. saltator and C. floridanus
(Bonasio et al. 2010) with animal miRNAs from miRBase (Griffiths-
Transcript assembly and protein-coding gene annotation Jones et al. 2008, Release 16). A set of putative expressed miRNAs
RNA-seq reads were aligned to the genome using TopHat (Trapnell was obtained by extracting short reads (18–30 nt) from the RNA-
et al. 2009), (parameters ‘‘-p 4 -r 20–mate-std-dev 10 -I 10000– seq data, filtered for low-quality sequence.

Genome Research 1345

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

Nygaard et al.

Known and putative miRNAs were aligned to the genome (Guindon et al. 2009), run with the ‘‘-f e -d nt’’ parameters, a HKY85
assembly using Bowtie (Langmead et al. 2009), allowing one mis- model, and 1000 bootstraps. For the M14 Peptidase genes, protein
match. Hits that overlapped repeat, CDS, rRNA, tRNA, or snRNA sequences were aligned with MAFFT following the same procedure,
annotation were ignored. MIREAP (Qibin and Jiang 2008) (de- and PhyML was run directly on the protein alignment with ‘‘-f e -d
fault parameters) was then used to predict whether the hits were aa’’ parameters and 1000 bootstraps. Obtained trees were converted
contained in miRNA-like secondary structures. When predicted to PDF using FigTree (Rambaut 2006) and colors and labels were
miRNAs from the known and the expressed sets overlapped, the edited for clarity in Adobe Illustrator.
coordinates of the latter were used.
Protein domains and motifs were predicted for all genes by
Feminizer gene searches
running InterProScan (Zdobnov and Apweiler 2001, v4.5) using
known domains from Pfam, PRINTS, PROSITE, ProDom, and Search of six-frame translations of the A. echinatior genome was
SMART (Release 27). Gene Ontology (GO) (Ashburner et al. 2000) conducted with HMMER (http://hmmer.janelia.org/) v. 3.0, using
IDs for each gene were obtained from the corresponding InterPro several different HMM models made from alignments of: (1)
entry. KEGG function (Kanehisa and Goto 2000) was assigned by Transformer/Feminizer (Tra/Fems) protein sequences from many
best BLASTP hit (parameter ‘‘-p blastp -b 10000 -v 10000 -F F -e 1e-5’’) insect species (separately for 39 and 59 ends), (2) Hymenopteran
to all KEGG proteins (release 51). Tra/Fems, and (3) only ant Tra/Fems. These models consistently
yielded only a single genomic match.

Gene family assignment

Neuropeptide annotation
Protein-coding genes from five eusocial (A. echinatior, S. invicta,
C. floridanus, H. saltator, A. mellifera) and two other insects (N. Known invertebrate neuropeptide precursor and receptor se-
vitripennis, D. melanogaster) were clustered into gene families based quences were used in TBLASTN and BLASTP homology searches
on conjoined BLASTP alignments using the Treefam methodology against A. echinatior genomic scaffolds and annotated peptides
(Supplemental Fig. S3; Li et al. 2006). When more than one isoform (v.1.0) S. invicta (v.2.3), C. floridanus (v.3.3), and H. saltator (v.3.3).
was present from a given gene, we kept only the longest. After Gene structures and open reading frames were predicted using
clustering, we performed multiple alignments of protein sequences multiple web-based gene prediction programs, the CLC Main
for each gene family using MUSCLE (Edgar 2004) and reverse Workbench (www.clcbio.com), and by manual correction. Signal
translated the protein alignments to CDS alignments. peptides were predicted using SIGNALP (Emanuelsson et al. 2007),
and peptide processing was predicted using the ProP program
(Duckert et al. 2004).
Gene Ontology (GO) enrichment analyses
Fishers exact test was used to calculate the statistical significance of Data access
enrichment of GO categories. The P-values were adjusted for
multiple testing (false discovery rate), using a cutoff of 0.05 for This Whole Genome Shotgun project has been deposited at DDBJ/
adjusted P-values. To remove redundancy in the GO enrichment EMBL/GenBank under the accession no. AEVX00000000. The
results, only the lowest level was reported when GO categories version described in this study is the first version, AEVX01000000.
with parent–child relationships contained the same gene sets. Raw sequence data is available under the accession no. ERP000666.

Phylogenetic trees and gene family expansion/contraction Acknowledgments

To generate the overall phylogeny of the six Hymenopteran ge- We thank the Smithsonian Tropical Research Institute in Panama
nomes and D. melanogaster, fourfold degenerate (FFD) sites were for making facilities available to work on fungus-growing ants, and
extracted from MUSCLE alignments of 1995 single-copy gene the Autoridad Nacional del Ambiente (ANAM) of Panama for is-
families and merged into one alignment. The phylogeny was cal- suing collection and export permits. S.N., M.S., and J.J.B. were
culated using the PhyML (Guindon et al. 2009) implementation of supported by the Danish National Research Foundation; F.H. and
Maximum-Likelihood, with the HKY85 substitution model. The C.J.P.G. by the Danish Research Agency (FNU); and F.H., C.J.P.G.,
root of the tree was determined by minimizing the height of the A.K., and S.N. were supported by different grants from the Novo
whole tree with TreeBeST (Ponting 2007). Nordisk Foundation. Y.W. is supported by the Swiss National Sci-
To identify significantly expanded/contracted gene families, ence Foundation and an ERC grant to Laurent Keller. We thank
a linear time tree was estimated from the FFD alignment using Tom Gilbert, Laurent Keller, and Eske Willerslev for advice during
UPGMA (Sneath and Sokal 1973) and used with CAFE (De Bie et al. the initial phases of this project; David R. Nash for helpful com-
2006) to infer the significance of change in gene family size along ments on the manuscript and for providing photographs of the
each branch. This method takes into account the phylogenetic ants; and Xuehong Meng, Qiulin Yao, Fengming Sun, Yong Liu,
history, including rate and direction of change in gene family size. and Dongming Fang for assisting with the annotations. We thank
The phylogenetic tree of the M16 Peptidase genes was con- Ioannis Xenarios for access to the Vital-IT (http://www.vital-it.ch)
structed using Ruby/Bioruby (Aerts and Law 2009; Goto et al. 2010) Center for high-performance computing of the Swiss Institute of
scripts to organize the multiple alignment and tree construction. Bioinformatics, funded in part by the Integrated Computational
Predicted CDS sequences were first translated using transeq (Rice Genomics Resources of the Swiss Institute of Bioinformatics: RITA-
et al. 2000), and the resulting protein sequences were then aligned CT-2006-026204.
using MAFFT v6.717b (Katoh and Toh 2008) using the L-INS-i Authors’ contributions: S.N., M.R., and A.K. performed the
option for high accuracy. Subsequently, the protein sequences preliminary sequencing feasibility study. S.N., G.Z., M.S., A.K.,
were reverse-translated with PAL2NAL v13 (Suyama et al. 2006) to J.W., and J.J.B. designed the research. S.N., G.Z., M.S., C.L., Y.W.,
produce a codon-level multiple alignment, while eliminating badly F.Q., H.H., J.Z., L.J., H.P., F.H., and C.J.P.G. executed the research
aligning segments by Gblocks 0.91b (Talavera and Castresana and analyzed the data. S.N., M.S., and J.J.B. wrote the paper (with
2007). The codon alignment was then used as input to PhyML 3.0 input from F.H., C.L., G.Z., C.J.P.G., and Y.W.).

1346 Genome Research

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

The genome of an Acromyrmex leaf-cutting ant

References Farrell BD, Sequeira AS, O’Meara BC, Normark BB, Chung JH, Jordal BH.
2001. The evolution of agriculture in beetles (Curculionidae: Scolytinae
and Platypodinae). Evolution 55: 2011–2027.
Aanen DK, Eggleton P, Rouland-Lefevre C, Guldberg-Froslev T, Rosendahl
Feldhaar H, Straka J, Krischke M, Berthold K, Stoll S, Mueller MJ, Gross R.
S, Boomsma JJ. 2002. The evolution of fungus-growing termites and
2007. Nutritional upgrading for omnivorous carpenter ants by the
their mutualistic fungal symbionts. Proc Natl Acad Sci 99: 14887–
endosymbiont Blochmannia. BMC Biol 5: 48. doi: 10.1186/1741-7007-5-
14892.
48.
Aanen DK, de Fine Licht HH, Debets AJM, Kerstes NAG, Hoekstra RF,
Fumagalli P, Accarino M, Egeo A, Scartezzini P, Rappazzo G, Pizzuti A,
Boomsma JJ. 2009. High symbiont relatedness stabilizes mutualistic Avvantaggiato V, Simeone A, Arrigo G, Zuffardi O, et al. 1998. Human
cooperation in fungus-growing termites. Science 326: 1103–1106. NRD convertase: a highly conserved metalloendopeptidase expressed at
Aerts J, Law A. 2009. An introduction to scripting in Ruby for biologists. specific sites during development and in adult tissues. Genomics 47:
BMC Bioinformatics 10: 221. doi: 10.1186/1471-2105-10-221. 238–245.
Altschul SF, Madden TL, Schäffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ. Goto N, Prins P, Nakao M, Bonnal R, Aerts J, Katayama T. 2010. BioRuby:
1997. Gapped BLAST and PSI-BLAST: a new generation of protein Bioinformatics software for the Ruby programming language.
database search programs. Nucleic Acids Res 25: 3389–3402. Bioinformatics 26: 2617–2619.
Armitage S, Broch J, Fernández Marı́n H, Nash D, Boomsma J. 2011. Immune Graf J, Ruby EG. 1998. Host-derived amino acids support the proliferation of
defence in leaf-cutting ants: a cross-fostering approach. Evolution 65: symbiotic bacteria. Proc Natl Acad Sci 95: 1818–1822.
1791–1799. Griffiths-Jones S, Moxon S, Marshall M, Khanna A, Eddy SR, Bateman A.
Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, 2005. Rfam: annotating non-coding RNAs in complete genomes. Nucleic
Dolinski K, Dwight SS, Eppig JT, et al. 2000. Gene ontology: tool for the Acids Res 33: D121–D124.
unification of biology. The Gene Ontology Consortium. Nat Genet 25: Griffiths-Jones S, Saini HK, van Dongen S, Enright AJ. 2008. miRBase: tools
25–29. for microRNA genomics. Nucl Acids Res 36: D154–D158.
Baer B, Eubel H, Taylor NL, O’Toole N, Millar AH. 2009a. Insights into female Guindon S, Delsuc F, Dufayard JF, Gascuel O. 2009. Estimating maximum
sperm storage from the spermathecal fluid proteome of the honeybee likelihood phylogenies with PhyML. Methods Mol Biol 537: 113–137.
Apis mellifera. Genome Biol 10: R67. doi: 10.1186/gb-2009-10-6-r67. Hasselmann M, Gempe T, Schiøtt M, Nunes-Silva CG, Otte M, Beye M. 2008.
Baer B, Heazlewood JL, Taylor NL, Eubel H, Millar AH. 2009b. The seminal Evidence for the evolutionary nascence of a novel sex determination
fluid proteome of the honeybee Apis mellifera. Proteomics 9: 2085–2097. pathway in honeybees. Nature 454: 519–522.
Becker AB, Roth RA. 1992. An unusual active site identified in a family of Hata H, Watanabe K, Kato M. 2010. Geographic variation in the damselfish-
zinc metalloendopeptidases. Proc Natl Acad Sci 89: 3835–3839. red alga cultivation mutualism in the Indo-West Pacific. BMC Evol Biol
Bekkevold , Boomsma JJ. 2000. Evolutionary transition to a semelparous life 10: 185. doi: 10.1186/1471-2148-19-185.
history in the socially parasitic ant Acromyrmex insinuator. J Evol Biol Hauser F, Cazzamali G, Williamson M, Blenau W, Grimmelikhuijzen CJP.
13: 615–623. 2006. A review of neurohormone GPCRs present in the fruitfly
Benson G. 1999. Tandem repeats finder: a program to analyze DNA Drosophila melanogaster and the honey bee Apis mellifera. Prog Neurobiol
sequences. Nucleic Acids Res 27: 573–580. 80: 1–19.
Birney E, Clamp M, Durbin R. 2004. GeneWise and Genomewise. Genome Hauser F, Cazzamali G, Williamson M, Park Y, Li B, Tanaka Y, Predel R,
Res 14: 988–995. Neupert S, Schachtner J, Verleyen P, et al. 2008. A genome-wide
Bonasio R, Zhang G, Ye C, Mutti NS, Fang X, Qin N, Donahue G, Yang P, Li inventory of neurohormone GPCRs in the red flour beetle Tribolium
Q , Li C, et al. 2010. Genomic comparison of the ants Camponotus castaneum. Front Neuroendocrinol 29: 142–165.
floridanus and Harpegnathos saltator. Science 329: 1068–1071. Hauser F, Neupert S, Williamson M, Predel R, Tanaka Y, Grimmelikhuijzen
Boomsma JJ. 2009. Lifetime monogamy and the evolution of eusociality. CJP. 2010. Genomics and peptidomics of neuropeptides and protein
Phil Trans R Soc B 364: 3191–3207. hormones present in the parasitic wasp Nasonia vitripennis. J Proteome Res
Boomsma JJ. 2011. Farming writ small. Nature 469: 308–309. 9: 5296–5310.
Boomsma JJ, Kronauer DJC, Pedersen JS. 2009. Organization of insect Honeybee Genome Sequencing Consortium. 2006. Insights into social
societies: from genome to sociocomplexity (ed. J Gadau et al.), pp. 3–25. insects from the genome of the honeybee Apis mellifera. Nature 443:
Harvard University Press, Cambridge, MA. 931–949.
Bot AN, Rehner SA, Boomsma JJ. 2001. Partial incompatibility between ants Hospital V, Prat A, Joulie C, Chérif D, Day R, Cohen P. 1997. Human and rat
and symbiotic fungi in two sympatric species of Acromyrmex leaf-cutting testis express two mRNA species encoding variants of NRD convertase,
ants. Evolution 55: 1980–1991. a metalloendopeptidase of the insulinase family. Biochem J 327: 773–
Brady SG, Schultz TR, Fisher BL, Ward PS. 2006. Evaluating alternative 779.
hypotheses for the early evolution and diversification of ants. Proc Natl Hughes WOH, Boomsma JJ. 2004. Genetic diversity and disease resistance in
Acad Sci 103: 18172–18177. leaf-cutting ant societies. Evolution 58: 1251–1260.
Brock DA, Douglas TE, Queller DC, Strassmann JE. 2011. Primitive Hughes WOH, Boomsma JJ. 2006. Does genetic diversity hinder parasite
agriculture in a social amoeba. Nature 469: 393–396. evolution in social insect colonies? J Evol Biol 19: 132–143.
Chesneau V, Pierotti AR, Barré N, Créminon C, Tougard C, Cohen P. 1994. Hughes WOH, Boomsma JJ. 2008. Genetic royal cheats in leaf-cutting ant
Isolation and characterization of a dibasic selective societies. Proc Natl Acad Sci 105: 5150–5153.
metalloendopeptidase from rat testes that cleaves at the amino terminus Hughes WOH, Sumner S, Van Borm S, Boomsma JJ. 2003. Worker caste
of arginine residues. J Biol Chem 269: 2056–2061. polymorphism has a genetic basis in Acromyrmex leaf-cutting ants. Proc
Chintapalli VR, Wang J, Dow JAT. 2007. Using FlyAtlas to identify better Natl Acad Sci 100: 9394–9397.
Drosophila melanogaster models of human disease. Nat Genet 39: 715– Ivens ABF, Nash DR, Poulsen M, Boomsma JJ. 2009. Caste-specific symbiont
720. policing by workers of Acromyrmex fungus-growing ants. Behav Ecol 20:
De Bie T, Cristianini N, Demuth JP, Hahn MW. 2006. CAFE: a computational 378–384.
tool for the study of gene family evolution. Bioinformatics 22: 1269– Jurka J, Kapitonov VV, Pavlicek A, Klonowski P, Kohany O, Walichiewicz J.
1271. 2005. Repbase Update, a database of eukaryotic repetitive elements.
De Fine Licht HH, Schiøtt M, Mueller UG, Boomsma JJ. 2010. Evolutionary Cytogenet Genome Res 110: 462–467.
transitions in enzyme activity of ant fungus gardens. Evolution 64: Kanehisa M, Goto S. 2000. KEGG: Kyoto Encyclopedia of Genes and
2055–2069. Genomes. Nucleic Acids Res 28: 27–30.
Den Boer SPA, Baer B, Boomsma JJ. 2010. Seminal fluid mediates ejaculate Katoh K, Toh H. 2008. Recent developments in the MAFFT multiple
competition in social insects. Science 327: 1506–1509. sequence alignment program. Brief Bioinform 9: 286–298.
Diamond J. 2002. Evolution, consequences and future of plant and animal Klasson L, Andersson SG. 2004. Evolution of minimal-gene-sets in host-
domestication. Nature 418: 700–707. dependent bacteria. Trends Microbiol 12: 37–43.
Dijkstra MB, Boomsma JJ. 2007. The economy of worker reproduction in Korf I. 2004. Gene finding in novel genomes. BMC Bioinformatics 5: 59. doi:
Acromyrmex leafcutter ants. Anim Behav 74: 519–529. 10.1186/147-2105-5-59.
Duckert P, Brunak S, Blom N. 2004. Prediction of proprotein convertase Kronauer DJC, Johnson RA, Boomsma JJ. 2007. The evolution of multiple
cleavage sites. Protein Eng Des Sel 17: 107–112. mating in army ants. Evolution 61: 413–422.
Edgar RC. 2004. MUSCLE: multiple sequence alignment with high accuracy Langmead B, Trapnell C, Pop M, Salzberg S. 2009. Ultrafast and memory-
and high throughput. Nucleic Acids Res 32: 1792–1797. efficient alignment of short DNA sequences to the human genome.
Elsik CG, Mackey AJ, Reese JT, Milshina NV, Roos DS, Weinstock GM. 2007. Genome Biol 10: R25. doi: 10.1186/gb-2009-10-3-r25.
Creating a honey bee consensus gene set. Genome Biol 8: R13. doi: Lewis SE, Searle SM, Harris N, Gibson M, Lyer V, Richter J, Wiel C,
10.1186/gb-2007-8-1-r13. Bayraktaroglir L, Birney E, Crosby MA, et al. 2002. Apollo: a sequence
Emanuelsson O, Brunak S, von Heijne G, Nielsen H. 2007. Locating proteins annotation editor. Genome Biol 3: RESEARCH0082. doi: 10.1186/gb-
in the cell using TargetP, SignalP and related tools. Nat Protoc 2: 953–971. 2002-3-12-research0082.

Genome Research 1347

www.genome.org
 Downloaded from genome.cshlp.org on December 30, 2011 - Published by Cold Spring Harbor Laboratory Press

Nygaard et al.

Li H, Coghlan A, Ruan J, Coin LJ, Hériché J-K, Osmotherly L, Li R, Liu T, Sen R, Ishak HD, Estrada D, Dowd SE, Hong E, Mueller UG. 2009.
Zhang Z, Bolund L, et al. 2006. TreeFam: a curated database of Generalized antifungal activity and 454-screening of Pseudonocardia and
phylogenetic trees of animal gene families. Nucleic Acids Res 34: D572– Amycolatopsis bacteria in nests of fungus-growing ants. Proc Natl Acad Sci
D580. 106: 17805–17810.
Li R, Fan W, Tian G, Zhu H, He L, Cai J, Huang Q , Cai Q , Li B, Bai Y, et al. Silliman BR, Newell SY. 2003. Fungal farming in a snail. Proc Natl Acad Sci
2009a. The sequence and de novo assembly of the giant panda genome. 100: 15643–15648.
Nature 463: 311–317. Sirviö A, Gadau J, Rueppell O, Lamatsch D, Boomsma JJ, Pamilo P, Page RE Jr.
Li R, Yu C, Li Y, Lam T-W, Yiu S-M, Kristiansen K, Wang J. 2009b. SOAP2: an 2006. High recombination frequency creates genotypic diversity in
improved ultrafast tool for short read alignment. Bioinformatics 25: colonies of the leaf-cutting ant Acromyrmex echinatior. J Evol Biol 19:
1966–1967. 1475–1485.
Li R, Zhu H, Ruan J, Qian W, Fang X, Shi Z, Li Y, Li S, Shan G, Kristiansen K, Smit AFA, Hubley R, Green P. 1996. RepeatMasker Open-3.0. http://
et al. 2010. De novo assembly of human genomes with massively repeatmasker.org.
parallel short read sequencing. Genome Res 20: 265–272. Sneath PHA, Sokal RR. 1973. Numerical taxonomy: The principles and practice
Lowe TM, Eddy SR. 1997. tRNAscan-SE: a program for improved detection of of numerical classification. W.H. Freeman, San Francisco, CA.
transfer RNA genes in genomic sequence. Nucleic Acids Res 25: 955–964. Stanke M, Keller O, Gunduz I, Hayes A, Waack S, Morgenstern B. 2006.
Lukashin A, Borodovsky M. 1998. GeneMark.hmm: new solutions for gene AUGUSTUS: ab initio prediction of alternative transcripts. Nucleic Acids
finding. Nucleic Acids Res 26: 1107–1115. Res 34: W435–W439.
Mikheyev AS, Vo T, Mueller UG. 2008. Phylogeography of post-Pleistocene Suen G, Scott JJ, Aylward FO, Adams SM, Tringe SG, Pinto-Tomás AA, Foster
population expansion in a fungus-gardening ant and its microbial CE, Pauly M, Weimer PJ, Barry KW, et al. 2010. An insect herbivore
mutualists. Mol Ecol 17: 4480–4488. microbiome with high plant biomass-degrading capacity. PLoS Genet 6:
Moreau CS, Bell CD, Vila R, Archibald SB, Pierce NE. 2006. Phylogeny of the e1001129. doi: 10.1371/journal.pgen.1001129.
ants: diversification in the age of angiosperms. Science 312: 101–104. Sumner S, Hughes WOH, Boomsma JJ. 2003a. Evidence for differential
Mueller U, Gerardo N, Aanen D, Six D, Schultz T. 2005. The evolution of selection and potential adaptive evolution in the worker caste of an
agriculture in insects. Annu Rev Ecol Evol Syst 36: 563–595. inquiline social parasite. Behav Ecol Sociobiol 54: 256–263.
Nawrocki EP, Kolbe DL, Eddy SR. 2009. Infernal 1.0: inference of RNA Sumner S, Nash DR, Boomsma JJ. 2003b. The adaptive significance of
alignments. Bioinformatics 25: 1335–1337. inquiline parasite workers. Proc R Soc Lond B Biol Sci 270: 1315–
Parra G, Bradnam K, Korf I. 2007. CEGMA: a pipeline to accurately annotate 1322.
core genes in eukaryotic genomes. Bioinformatics 23: 1061–1067. Sumner S, Hughes WOH, Pedersen JS, Boomsma JJ. 2004. Ant parasite
Ponting C. 2007. TreeBeST: Tree building guided by Species Tree. http:// queens revert to mating singly. Nature 428: 35–36.
treesoft.sourceforge.net/treebest.shtml. Suyama M, Torrents D, Bork P 2006. PAL2NAL: robust conversion of protein
Poulsen M, Boomsma JJ. 2005. Mutualistic fungi control crop diversity in sequence alignments into the corresponding codon alignments. Nucleic
Acids Res 34: W609–W612.
fungus-growing ants. Science 307: 741–744.
Talavera G, Castresana J. 2007. Improvement of phylogenies after removing
Poulsen M, Currie CR. 2010. Symbiont interactions in a tripartite
divergent and ambiguously aligned blocks from protein sequence
mutualism: exploring the presence and impact of antagonism between
alignments. Syst Biol 56: 564–577.
two fungus-growing ant mutualists. PLoS ONE 5: e8748. doi: 10.1371/
Trapnell C, Pachter L, Salzberg SL. 2009. TopHat: discovering splice
journal.pone.0008748.
junctions with RNA-Seq. Bioinformatics 25: 1105–1111.
Price AL, Jones NC, Pevzner PA. 2005. De novo identification of repeat
Trapnell C, Williams BA, Pertea G, Mortazavi A, Kwan G, van Baren MJ,
families in large genomes. Bioinformatics 21: i351–i358.
Salzberg SL, Wold BJ, Pachter L. 2010. Transcript assembly and
Qibin L, Jiang W. 2008. MIREAP: MicroRNA Discovery By Deep Sequencing.
quantification by RNA-Seq reveals unannotated transcripts and
http://sourceforge.net/projects/mireap/. isoform switching during cell differentiation. Nat Biotechnol 28: 511–
Rambaut A. 2006. FigTree. http://tree.bio.ed.ac.uk/software/figtree/. 515.
Reznik SE, Fricker LD. 2001. Carboxypeptidases from A to Z: implications in Tweedie S, Ashburner M, Falls K, Leyland P, McQuilton P, Marygold S,
embryonic development and Wnt binding. Cell Mol Life Sci 58: 1790– Millburn G, Osumi-Sutherland D, Schroeder A, Seal R, et al. 2009.
1804. FlyBase: enhancing Drosophila Gene Ontology annotations. Nucleic
Rice P, Longden I, Bleasby A. 2000. EMBOSS: The European Molecular Acids Res 37: D555–D559.
Biology Open Software Suite. Trends Genet 16: 276–277. Van Borm S, Wenseleers T, Billen J, Boomsma JJ. 2003. Cloning and
Rodriguez de la Vega M, Sevilla RG, Hermoso A, Lorenzo J, Tanco S, Diez A, sequencing of wsp encoding gene fragments reveals a diversity of co-
Fricker LD, Bautista JM, Avilés FX. 2007. Nna1-like proteins are active infecting Wolbachia strains in Acromyrmex leafcutter ants. Mol Phylogenet
metallocarboxypeptidases of a new and diverse M14 subfamily. FASEB J Evol 26: 102–109.
21: 851–865. Villesen P, Murakami T, Schultz TR, Boomsma JJ. 2002. Identifying the
Schiøtt M, De Fine Licht HH, Lange L, Boomsma JJ. 2008. Towards transition between single and multiple mating of queens in fungus-
a molecular understanding of symbiont function: identification of growing ants. Proc R Soc Lond B Biol Sci 269: 1541–1548.
a fungal gene for the degradation of xylan in the fungus gardens of leaf- Werren JH, Richards S, Desjardins CA, Niehuis O, Gadau J, Colbourne
cutting ants. BMC Microbiol 8: 40. doi: 10.1186/1471-2180-8-40. JK, Group TNGW. 2010. Functional and evolutionary insights from
Schiøtt M, Rogowska-Wrzesinska A, Roepstorff P, Boomsma JJ. 2010. Leaf- the genomes of three parasitoid Nasonia species. Science 327: 343–
cutting ant fungi produce cell wall degrading pectinase complexes 348.
reminiscent of phytopathogenic fungi. BMC Biol 8: 156. doi: 10.1186/ Wurm Y, Wang J, Riba-Grognuz O, Corona M, Nygaard S, Hunt BG, Ingram
1741-7007-8156. KK, Falquet L, Nipitwattanaphon M, Gotzek D, et al. 2011. The
Schultz TR, Brady SG. 2008. Major evolutionary transitions in ant genome of the fire ant Solenopsis invicta. Proc Natl Acad Sci 108: 5679–
agriculture. Proc Natl Acad Sci 105: 5435–5440. 5684.
Schultz TR, Bekkevold D, Boomsma JJ. 1998. Acromyrmex insinuator new Zdobnov EM, Apweiler R. 2001. InterProScan - an integration platform for
species: an incipient social parasite of fungus-growing ants. Insectes Soc the signature-recognition methods in InterPro. Bioinformatics 17: 847–
45: 457–471. 848.
Schultz TR, Mueller UG, Currie CR, Rehner SA. 2005. Reciprocal
illumination: A comparison of agriculture in humans and in fungus-
growing ants. In Insect-fungal associations: Ecology and evolution (ed. FE
Vega, M Blackwell). Oxford University Press, New York. Received February 3, 2011; accepted in revised form May 25, 2011.

1348 Genome Research

www.genome.org