CORDIS - EU research results

Multi-level analysis of the evolution of cooperative behaviour in social insects

Final Report Summary - EVOCOOP (Multi-level analysis of the evolution of cooperative behaviour in social insects)

The research project “Multi-level analysis of the evolution of cooperative behaviour in social insects” aimed to characterize the deep regulatory mechanisms of a complex behaviour that is at the basis for the evolution of sociality. Cooperation is widespread in many animal groups (including humans) and it has been well characterized in many systems in terms of the behavioural patterns that it involves. However, it is still unclear how this complex behaviour is regulated at the level of brain functions, what genes are the major effectors and how they are regulated. Furthermore, from an evolutionary point of view, it is of great interest to understand whether cooperative behaviour is regulated by similar effecters and mechanisms across multiple organisms that evolved sociality at different levels of complexity. This project pertains two social insects that are characterized by significantly different levels of sociality: the primitively eusocial paper wasp Polistes dominula and the highly eusocial fire ant Solenopsis invicta. Both these insect typically display cooperative behaviour at different moment of their colony life cycle and among different colony members. Particularly relevant for this project are colony founding, when multiple queen foundresses might join the same nest to start a new colony together, and later, when the colony reaches the mature stage and multiple functional queens coexist. Interestingly in both paper wasps and fire ants, cooperation during colony founding is not characterized by linear and peaceful patterns of interaction among nestmates, but it is interspersed by subtle dominance hierarchies, reproductive competition and aggression. This eventually result in open conflict by the end of the founding stage.
A first part of the project was dedicated to the analysis of cooperation and aggression in young colonies of P. dominula. Cooperation and aggression can be seen as two opposite sides of the same process, and it is of great interest to understand whether they are regulated by similar molecular mechanisms. We adopted a candidate gene approach to select a group of genes that were associated with cooperative and aggressive behaviour in previous microarray studies performed in paper wasps and fire ants. We then characterized the levels of expression for these genes in four groups of wasps: two groups were characterized by the physiology of a dominant wasp and more aggressive behaviour, two were subordinate wasps displaying more cooperative behaviour. This analysis revealed that one gene, vitellogenin, showed consistent association between dominant phenotype/aggressive behaviour and high levels of expression in the brain. We selected vitellogenin for an approach of RNA interference (RNAi), where it is possible to knock out the expression of a gene in order to see whether the associated phenotype is modified. In this study, we measured aggressive behaviour as the phenotype of interest.
The second part of the project focused on colony founding in S. invicta. This is a highly plastic process, where the same queen can opt for starting a colony alone or joining other queens in a communal nest. We used a powerful approach of RNA sequencing to compare the levels of expression for the whole fire ant gene set in founding queens that adopted the two modalities. We quantified gene expression immediately after colony founding, when multiple foundresses cooperate within the same nest, and also at the end of the founding process, when interactions among nestmates are characterized by open conflict. This approach enabled us to investigate what molecular mechanisms play a major role in the setup of cooperation during colony founding and to understand whether the same regulators are responsible for the later transition to conflict. For associations of multiple foundresses, we also looked at different group sizes and this will help us understanding how the regulation of cooperative/conflictual interactions are shaped at the level of the brain by the social environment.

Cooperation is a prime example of a complex social behaviour that is widespread in nature and occurs across multiple taxa. Normally, cooperative behaviour is performed among related individuals, as expected under kin selection (Queller and Strassmann 1998), with the effort spent by each individual in performing the behaviour being paid back by an increase in fitness. Advantages can be direct, through individual reproduction, or indirect, through reproduction of highly related individuals (inclusive fitness, Bourke 2011). One of the best examples of cooperative behaviour occurs in social insect colonies, where usually only one or a few individuals reproduce while the majority of the other colony members gain indirect fitness by enhancing the reproduction of their relatives (Wilson 1971). This type of cooperation in social insects is often identified as altruism and it is one of the major achievements of the evolution of sociality in the animal kingdom. Despite the great interest that cooperation has always triggered, this behaviour has been mainly investigated in its highest expression. For example, in social insects, much attention has been paid to the extreme derivation of social behaviour, i.e. altruism and division of labour among related individuals, and these have been mainly studied in highly advanced eusocial systems such as the honeybee. Very little is known about the molecular mechanisms regulating cooperative behaviour in non-model more primitive social systems and among unrelated individuals.
Cooperation in insect societies is not limited to mature colonies, and a key expression of such behaviour can occur during colony founding, a central step in the dispersal and survival of the offspring (Brown and Bonhoeffer 2003). Cooperative colony founding enables young queens to speed up the process of forming and establishing a new colony in order to be more successful against enemies (predators, parasites and pathogens) and competitors (other colonies of conspecifics) occurring in the same area (Brown 2000). Interestingly, in this case the cooperative behaviour is frequently performed among unrelated individuals (Queller et al. 2000). Unrelated group members have been reported in kingfishers, manakins, mongooses, halictine bees, paper wasps and ants. Understanding such cooperative behaviour is challenging, because stability in groups of unrelated individuals requires that all members should have a probability of receiving direct benefits, a condition which should have a great impact on the nature of their cooperation (Bernasconi and Strassmann 1999). In fact, within groups of unrelated individuals initial cooperation rapidly becomes a more complex network of behaviours, including aggression, dominance hierarchies and reproductive conflict (Leadbeater et al. 2011). This transition can occur very soon after the formation of the association or later, when the association is already established. The repeated evolution of cooperative behaviour in both simple and complex insect societies, at distinct life stages (multiple queens during colony-founding vs. mature multi-queened colonies) provides the opportunity to dissect the interaction between intrinsic and extrinsic factors governing this puzzling social behaviour (Manfredini et al. 2013).
Despite their apparently opposing nature, cooperation and aggression can be seen as two sides of the same coin, two facets of social behaviour. It is therefore of great interest to understand whether these two behaviours are regulated by different molecular mechanisms or are related at a mechanistic level. This is now possible thanks to the release of the genome for many new social species and thanks to the development of cutting-edge molecular approaches capable to link genes and behaviour. One of these is the sociogenomic approach i.e. the understanding of social life in molecular terms, that has emerged as the new paradigm for understanding complex behaviour (Grozinger and Robinson 2010, Johnson and Linksvayer 2010). Pioneering studies in the honeybee Apis mellifera have revealed the genetic and molecular basis underlying social behaviour (Zayed and Robinson 2012). Behaviour results from the interaction of multiple intrinsic and extrinsic factors, generally organised in a nonlinear and often unpredictable fashion (Hofmann 2003). Consequently, a key question asked in sociogenomics, and in studies of complex behaviour in general, is what is the relative importance of intrinsic factors (genetics, physiology or phenotype of an organism) versus the external environment, i.e. social interactions, in determining the expression of complex social behaviours (Robinson et al. 2008, Toth et al. 2007). Sociogenomics can now rely on new powerful techniques to describe gene functioning in great details. These include microarrays or RNA sequencing, to characterize patterns of expression for the whole gene-set of an organism and RNA interference (RNAi) a cutting-edge technique in molecular ecology that moves the focus from correlation to causation, thus providing evidence that the gene of interest is directly responsible for producing the observed behaviour.

The aim of this project was to characterize in a comparative fashion the molecular regulation of cooperation and aggression in two social insects that display different levels of sociality: the primitively eusocial paper wasp Polistes dominula and the highly eusocial fire ant Solenopsis invicta. Both these insect typically display cooperative behaviour at different moment of their colony life cycle and among different colony members. Polistes paper wasps produce small annual societies with morphologically similar individuals playing different behavioural roles. Individual or multiple foundresses establish nests in the spring (haplometrosis and pleometrosis, respectively), rearing workers that eventually take over colony tasks such as nest construction, defence, brood rearing, and foraging (Reeve 1991). In pleometrotic associations, while there is a dominant egg-laying queen across the life cycle of the colony, cofoundresses coexist with the functional queen, manage to lay eggs while the queen is present and can replace her when she dies. Social harmony within the colony is maintained through dominance hierarchies based on aggressive interactions. In late summer, colonies rear males and gynes (future reproductives); the gynes overwinter and found new nests in the spring. Polistes, in particular Polistes dominula, have long been a preeminent model system for the study of fundamental questions in behaviour and sociobiology (Turillazzi and West-Eberhard 1996, Pardi 1948). With the development of genomic and transcriptomic tools for these wasps (Toth et al. 2010, Liebert et al. 2008), they have become a powerful system for sociogenomic studies of cooperative behaviour in relatively simple societies.
Solenopsis invicta fire ant live in large perennial colonies characterized by clear dimorphism between reproductive individuals (female sexuals and males) and the worker caste. Organization of colony life may follow two distinct social forms: monogyny (a single functional queen) and polygyny (multiple functional queens). In polygyne colonies all queens are egg-layers and though there are no evident dominance hierarchies egg-laying rate can differ from queen to queen thanks to mechanisms of reciprocal reproductive inhibition via pheromones. Colony founding occurs in polygyne colonies through budding from the mother colony, when a young queen leaves with a group of workers and establishes a new nest. These colonies can later accept additional reproductive queens. In the monogyne social form the scenario is different: sexually mature queens engage in mating flights and they disperse far from the mother colony. Newly mated queens (NMQs) can establish a new nest individually (haplometrosis) or in groups (pleometrosis, Tschinkel and Howard 1983). Since mature monogyne colonies only tolerate a single functional queen, the fate of pleometrotic associations is inevitably to disaggregate after the emergence of the first workers: all queens but one leave the colony or are executed (Balas and Adams 1996). Fire ants have been largely studied in the past for aspects related to the basic mechanisms of colony life and structure (Tschinkel 2006). Recently, S. invicta has emerged as a model to investigate the genetic basis for social organization, since the existence of the two social forms is directly associated with two variants of a Y-like social chromosome (Wang et al. 2013). This discovery, together with the sequencing of the fire ant genome (Wurm et al. 2011), makes S. invicta an ideal candidate to study the interaction between genome and social environment.

In the first part of the project we aimed to characterize the molecular regulation of cooperative and aggressive behaviour in the brain of P. dominula foundresses. We addressed the following questions:

- What genes are associated with cooperation and aggression during colony founding?
- What are the best predictors of these complex behaviour?
- Can we prove a causal link between gene functioning and behaviour of interest?

We addressed these questions in incipient colonies sampled in the field and reared in the lab. We performed morpho-physiological measures and behavioural assays in the lab, to identify dominant (more aggressive) and subordinate (more cooperative) phenotypes. We obtained quantification of gene expression with real-time PCR (QPCR) and we manipulated the expression level of genes of interest with RNAi.
The second part of the project was dedicated to identifying the molecular mechanisms that regulate cooperative colony founding in S. invicta. For this purpose, we used an RNAseq approach to characterize global patterns of gene expression in queen foundresses. We focused on two key moments of the founding process and we addressed the following questions:

- How is cooperative breeding initiated in fire ants?
- What are the genes involved in the regulation of this complex social behaviour?

- How is the brain modelled by cooperative breeding in fire ants?
- What genes respond to the different social environment that founding queens experience?
- How are these genes affected by the different social complexity associated with each founding syndrome?

PART 1: characterization of the patterns of expression of genes responsible for cooperative and aggressive behaviour in paper wasps
The first step to accomplish this part of the project consisted of producing a list of candidate genes potentially involved in cooperative/aggressive behaviour in P. dominula. For this purpose, we considered two previous microarray studies where the authors characterized the global patterns of gene expression among co-founding queens in two different social insects. The first study identified genes that were differentially regulated between dominant and subordinate co-foundresses (and workers) in another paper wasp, Polistes metricus (Toth et al. 2014). The second study was performed in the fire ant S. invicta and looked at genes that differed between single (haplometrotic) and multiple (pleometrotic) foundresses, and also between winners and losers in pleometrotic associations (Manfredini et al. 2013). We took the lists of significantly differentially expressed genes (DEGs) and we performed an overlap analysis to isolate those genes that were differentially regulated in both studies. This group included 17 genes. By looking at every possible source of annotation that was available for these genes (e.g. published studies and GeneBank) not only in wasps and ants but also other organisms, we reduced the list to 12 genes that were sufficiently characterized. Finally, from these we chose 6 genes that showed the most interesting patterns of expression for the purpose of our study. These genes displayed the following patterns of expression:

- For the paper wasp study:
1a) candidate genes for cooperation: genes that were consistently up-regulated in subordinate foundresses and subordinate workers
1b) candidate genes for aggression: genes that were consistently up-regulated in dominant foundresses and dominant workers

- For the fire ant study:
2a) candidate genes for cooperation: genes that were consistently up-regulated in pleometrotic queens (compared to haplometrotic) during the cooperative phase and down-regulated in pleometrotic winners during the conflict phase
2b) candidate genes for aggression: genes that were consistently down-regulated in pleometrotic queens (compared to haplometrotic) during the cooperative phase and up-regulated in pleometrotic winners during the conflict phase
Here are the 6 candidate genes that were selected together with information on some of the biological processes that they regulate:

3 genes for cooperation
¬ alpha-coatomer protein: regulation of lipid storage; phagocytosis
¬ Inositol-3-phosphate synthase: circadian rhythm
¬ Rasputin: ovarian follicular epithelium; positive regulation of gene expression; compound eye photoreceptor fate commitment; Ras protein signal transduction

3 genes for aggression
¬ Beadex: response to cocaine; locomotor rhythm
¬ vitellogenin: synthesis and release of the yolk protein; reproductive development; behavioural maturation in honey bee workers
¬ Octopamin β: neuromodulator responsible for increased aggression in ant queens

We identified the P. dominula orthologues for these genes and we designed specific primers to perform QPCR on RNA samples from brains of paper wasp specimens.
Experimental wasps came from P. dominula incipient colonies that were collected in the field in the summer of 2015 in mainly three locations: the campus of the Iowa State University, downtown Ames (Iowa) and the campus of the University of Minnesota in Minneapolis/St. Paul. We sampled a total of 73 haplometrotic colonies and only 10 pleometrotic colonies (min. 2 and max. 5 co-foundresses on the same nest). These colonies were housed in the insect rearing facilities at ISU and maintained following protocols established in the Toth Lab. We monitored pleometrotic colonies for a period of 2 weeks to identify behavioural patterns of co-foundresses that helped us recognizing dominant/subordinate hierarchies. Foundresses that spent most of the time on the nest were provisionally labelled as dominant while those mostly engaged in foraging activities were labelled as subordinates (as there is no clear external morphological difference between dominant and subordinate co-foundresses). These labels were confirmed after scoring a series of morpho-physiological measures that were taken after killing wasp co-foundresses in liquid nitrogen and dissecting their abdomens. These measures were:
• egg development (score 1 to 3)
• ovary size (small, intermediate, large)
• amount of fat bodies (score 1 to 3)
• gaster size (length, width, combination of the two)

Ideally, a dominant foundress was the wasp on the nest with the highest scores for all measures. In total we processed 33 wasps and from these we selected the best 7 best dominant and 7 best subordinate from the same 7 colonies. These were the first 2 groups of our QPCR experiment. The other 2 groups were monogyne queens and newly emerged workers from monogyne colonies. Queens in monogyne colonies were labelled before the emergence of workers and their status was confirmed by observing ovary development after dissection. Hence, the design for the QPCR experiment included 4 groups of wasps, i.e. 2 dominant (less cooperative) and 2 subordinate (more cooperative) composed as follows:

1. dominant A = dominant queens from pleometrotic couples (N=7)
2. dominant B = monogyne queens (N=7)
3. subordinate A = subordinate queens from pleometrotic couples (N=7)
4. subordinate B = workers from monogyne colonies (N=7)

We dissected brains out of the head capsule in dry ice to prevent RNA degradation. We then isolated total RNA and performed cDNA synthesis that was used to perform QPCR on the 6 candidate genes + 2 control housekeeping genes (actin and elongation factor 1). Of the 6 genes that were checked, only vitellogenin showed a significant and consistent pattern of up-regulation in dominant (more aggressive) wasps (i.e. dominant queens from pleometrotic couples and monogyne queens) and down-regulation in the other 2 groups. Therefore vitellogenin was selected as the target gene for the RNAi approach

PART 2: silencing of vitellogenin by means of RNAi to revert aggressive behaviour in wasp foundresses
In this part of the project we aimed at knocking down the expression of vitellogenin in the head of P. dominula foundresses and to cause a switch in behavioural patterns from more aggressive to more cooperative (less aggressive). While this approach has been applied to other organisms (including honey bees) this was the first time that it was attempted in paper wasps. For this experiment we used queens from small monogyne colonies that were part of the pool of colonies sampled in Summer 2015 (see above). In a first pilot trial, we tested 2 different doses of the molecular construct to silence vitellogenin (vg-siRNA), we scored aggressive behaviour in a small sub-sample of wasps and we measured level of expression of vitellogenin in the brain of all specimens. In a second trial we performed a similar experiment with a higher dose of the construct but we scored aggressive behaviour only before RNAi treatment.
We designed 3 vg-siRNA construct targeting 3 different exons of the vitellogenin mRNA. These constructs were then mixed to obtain a single solution with equal quantities of each construct. We also obtained an EGFP control target (aspecific for the wasp mRNA). Both solutions were combined with lipofectamine, a medium to deliver the mix to the brain cells. We delivered 1ul of the vg-siRNA construct with a microcapillary connected to a microinjector device. The tip of the capillary was inserted through a small opening that was created with a thin scalpel in the back of the head capsule, in correspondence of the medium ocellus. The solution was delivered very slowly, to prevent any oozing out and consequentially loss of the construct. Wasps were ice-anesthetized for 30 minutes before and after the procedure, and they were kept on ice during the treatment as well.

Pilot trial. For this experiment we treated 22 queens that were assigned to one of the three following experimental groups:
a. vitellogenin high dose: 3 ng/ul (N=8)
b. vitellogenin low dose: 0.7 ng/ul (N=7)
c. EGFP control (N=7)

Two queens died during manipulation and therefore we were left with 20 queens to process. These queens were returned to their colony and observed for 3 days to record their position with respect to the nest. On the third day, we recorded behavioural videos for only 6 wasps (2 per group). Thereafter all 20 queens were flash-frozen in liquid nitrogen. We dissected ovaries of these queens: apart from one, all samples revealed mature eggs in the ovaries independently of the treatment, confirming that they all were the functional queen of their colony. We then dissected brain and isolated total RNA to perform cDNA synthesis and QPCR reactions.

Second trial. For this experiment we used 17 queens that were allocated as follows:
i. vitellogenin treatment: 60 ng/ul – i.e. 20X higher than vitellogenin high pilot (N=10)
ii. EGFP control (N=7)

Prior to treatment, 6 queens were filmed (3 vitellogenin and 3 EGFP). Conditions for the RNAi treatment were the same as above. After 3 days, queens were flash-frozen in liquid nitrogen and stored in a -80°C freezer.

Behavioural assays. Aggression tests were performed within a glass Petri dish (radius 10.5 cm) and videotaped. Each experimental wasp was presented with an unfamiliar wasp of a similar rank and they were allowed to interact for 10 minutes. The following behaviour were recorded (with score) by a student blind to the experimental setup:
• non-aggressive contact (0) + duration
• antennal box (1)
• dart (2)
• dart with mandible open (3)
• bite (4)
• attempted mount (4)
• grapple (5)
• mount (5)

Thereafter, an aggression index was calculated between paired wasps (based on the index used in (Sheehan and Tibbetts 2008) that accounts for the number and intensity of each aggressive behaviour. This aggression index was calculated by assigning behaviours a score and dividing the summed aggression score by the total number of contacts (aggressive or non-aggressive).
The pilot trial revealed no significant differences in the expression levels of vitellogenin in treatments vs. control. Also aggression scores did not indicate that the behavioural patterns were modified because of the RNAi treatment. Samples from the second trial have not been processed yet and therefore we are unable to present these results at the moment.

PART 3: characterization of the molecular mechanisms regulating cooperative colony in fire ants
Newly mated queens of Solenopsis invicta were sampled on May 4th 2014 in a parking lot in Gainesville (Florida) immediately after a big mating flight. This area is densely populated by monogyne colonies and the genotyping of samples collected in previous years confirmed that queens collected in this area belong to the monogyne social form. Queens were individually collected with forceps directly from the tarmac and transferred to a small plastic cup. All these queens were wingless, hence they had spent several minutes up to 2 hours on the tarmac looking for a suitable nest location. In fact, within 2 hours all queens have usually disappeared from above ground in field observations (Tschinkel 2006). A set of 34 queens was immediately frozen on dry ice as TIME 0 sample.
After a set of plastic cups (12 total) was completed, the queens were released in large trays containing nesting chambers where fire ant queens usually build their colony in lab conditions. As the mode of colony founding in the field is density dependent, we used two different setups to promote spontaneous formation of haplometrotic and pleometrotic nests. We used lower density to promote haplometrosis: we released 24 queens in a large tray containing 24 nesting chambers (7 trays total). We used higher density to promote pleometrotic associations instead: here 48 queens were released in a smaller tray containing only 14 nesting chambers (7 trays total). Ultimately, the proportion of groups that were haplometrotic or pleometrotic did not differ across the low and high density groups.
All 14 trays were transported to an environmental chamber were queens were reared in standard claustral conditions (no food, no water, in the dark). For the first 2 days, nesting chambers were left open so that queens could freely move from one chamber to another (mimicking what normally happens in the field). We recorded the numbers of haplometrotic queens and pleometrotic groups for both days. At the end of DAY 2, we transferred each nesting chamber to a separate pencil box so that queens were no longer allowed to move across nests. We kept queens in these conditions until the emergence of workers (1 month approximately). Thereafter incipient colonies were provided with water, sugar water and frozen crickets ad libitum and housed in the lab in standard conditions of temperature and humidity.
Prior to allocating queens to experimental groups for RNAseq we dissected abdomens to check the spermatheca for mating status and to look at ovary development. Only queens that were mated and had fully developed ovaries were considered for this study. For these queens, we dissected brains out of the head capsule on dry ice and we isolated total RNA from brain tissue for RNAseq. We performed two different RNAseq experiments using selected queens.
For this experiment we analysed queens immediately after mating flights and queens that spontaneously chose to found a new colony alone (haplometrosis) and in group (pleometrosis):

1) newly mated queens sampled at TIME 0 A1
2) haplometrotic queens sampled at DAY 3 A2
3) pleometrotic queens sampled at DAY 3 from large pleometrotic groups (>12 queens) A3

We compared the same group of newly mated queens from “A” to queens that were exposed for 3 weeks to 3 different social environments: social isolation (haplometrosis), small social groups and large social groups. In this case, haplometrotic queens were derived from initial pleometrotic associations (from groups of 7-11 queens) to eliminate any possible pre-existing conditions associated with spontaneous haplometrosis that could make this group different from the other two. All queens were sampled before the emergence of workers:

1) newly mated queens at TIME 0 B1
2) haplometrotic queens at DAY 25 B2
3) pleometrotic queens at DAY 25 from small groups (2-6 queens) B3
4) pleometrotic queens at DAY 25 from large group (7-21) B4

Though for a couple of large groups the difference with small groups at day 25 is not great (7-8 queens vs. 5-6), they were reared in conditions of higher social complexity for long time (16-17 queens initially vs. 5-6) and therefore this should be the major cause for shaping their brain neural structure.

Brain RNA samples from 44 S. invicta foundresses were sent to the sequencing facility for further processing. Only 33 samples survived quality control and were suitable for library preparation and sequencing. These samples were distributed as follows:

A) A/B1 = 6 queens A2 = 6 queens A3 = 6 queens
B) A/B1 = 6 queens B2 = 5 queens B3 = 5 queens B4 = 5 queens

cDNA libraries were prepared with the TruSEQ stranded RNA-Seq kit and they were loaded on 2 lanes of a Illumina HiSeq 2500 sequencer. This produced an output of 470 million reads each lane (paired-end, 125bp per read) that corresponded approximately to 30 million reads per specimen.
We are currently analysing the RNAseq data hence we are unable to provide a description of the results for this part of the project.

PART 4: analysis of RNAseq data
As part of this project, we developed a pipeline for the analysis of RNAseq data for studies in the field of behavioural ecology. We validated this pipeline in two studies that were not part of the initial plan of this fellowship but addressed similar questions. One study, recently published in BMC Genomics (Manfredini et al. 2015) investigated the changes in brain gene expression in honey bees associated with natural mating and exposure to CO2 (a procedure used to perform artificial insemination of honey bee queens). In the second study, currently in preparation for submission, we characterized the patterns of gene expression in the brains of bumblebee queens that successfully mated and became reproductively mature compared to queens that failed to achieve these important life history transitions.
Below is a summary of the key steps for the analysis of RNAseq data that we included in our pipeline and will be used for the analysis of data from the fire ant experiments.

1) Analysis of read counts
- Quality control with FastQC
- Trimming adapters, low quality bases and discarding short reads with TRIMMOMATIC
- Filtering out ribosomal RNA with SORTME_RNA
- Alignment of the clean reads to the genome with TopHat for Illumina
- Extrapolating read counts from alignments with SAMtools idxstats

2) Analysis of differential gene expression
- Global analyses of gene expression with GLM and planned linear contrasts
- Visualization of global patterns of gene expression with Hierarchical Clustering (heatmap), Principal Component analysis and K-means clustering
- Pairwise comparisons between groups of interest with edgeR or DESeq in R

3) Identification of the biological functions associated with differentially expressed genes
- Inferring annotations for genes of interest with BLAST
- Obtaining enrichment for GO terms and KEGG pathways with DAVID, or topGO and GOseq in R
- Visualization of GO terms with REVIGO, cateGOrizer and amigo

4) Comparative studies across species
- Overlapping lists of genes or GO terms with Venny
- Calculating significant overlaps with hypergeometric tests and Fisher’s exact test

5) Analysis of gene networks
- Identifying modules of co-expressed genes with WGCNA in R
- Visualizing gene networks and “hub” genes with VisANT

The analysis of the RNAseq data from the fire ant experiments will largely increase our knowledge of how cooperative behaviour is regulated in social systems. The powerful platform that we used and the large number of biological replicates across different social groups and time points will enable us to detect with great depth even small changes that occur at the level of brain gene expression. These data will be among the first evidence of how complex social behaviours are regulated at the molecular level in the insect brain. We will not only provide names of potential candidate genes that are master regulator of behavioural performances, but we will also analyse how brain genes are organized in functional networks. Thereafter, we will combine evidence from these experiments with the output from the work on paper wasps, where we looked at key regulators of cooperation and aggression in P. dominula foundresses. This comparative approach will be fundamental to determine whether similar key genes or the same genetic toolkit are responsible for regulating complex social behaviours across different social insects. These findings will be of great interest not only for the social insect community but also for all scientists interested in social behaviour, from both an evolutionary and functional perspective. Furthermore, the RNAi approach, if successful, will provide a unique tool to prove the causal link between a specific gene and the behaviour it regulates and, from a more applied perspective, it will potentially open new avenues of biological control for invasive paper wasps (and potentially for other social insects as well).
The impact of this research project will be vectored by at least two scientific publications in peer-reviewed journals that are at the top in the field of molecular and behavioural ecology. Another form of important impact will be the diffusion of the methodological approaches that we developed during the execution of the project. We have applied for the first time the RNAi approach to paper wasps to target gene expression in the brain, and this protocol is currently being used in the Toth Lab to follow-up on our work. So far, a similar approach for insects has been used only in honey bees. Hence our contribution will help to spread the technique to other insect models (social and non-social). Furthermore, we have established a protocol to dissect brains of fire ant queens and isolate RNA samples that are suitable for RNAseq. Even though a similar goal has been previously achieved in other labs, our accomplishment is still remarkable and will be of great use for all scientists at Royal Holloway and beyond. Finally, thanks to extensive practice with two previous unrelated projects, we have developed a custom pipeline to analyse RNAseq data that we are now using to analyse the fire ant data. This protocol resulted from the interaction with different research groups, including the Paccanaro Lab in the Centre for Systems and Synthetic Biology at Royal Holloway, thus will be of great value for all scientists interested in performing similar analyses.


M. T. Balas, E. S. Adams, Behavioral Ecology and Sociobiology 38, 391 (1996).
G. Bernasconi, J. E. Strassmann, Trends in Ecology & Evolution 14, 477 (1999).
A. F. G. Bourke, Proceedings of the Royal Society B: Biological Sciences 278, 3313 (2011).
M. J. F. Brown, Trends in Ecology and Evolution, Vol. 15, No. 3, 03, p. 116-116 (2000).
M. J. F. Brown, S. Bonhoeffer, Evolutionary Ecology Research, Vol. 5, No. 2, 02, p. 305-313 (2003).
P. G. Ferreira et al., Genome Biology 14, R20 (2013).
C. M. Grozinger, G. E. Robinson, in Enc An Behav, M. Breed, J. Moore, Eds. (Elsevier, Oxford, 2010).
H. A. Hofmann, Journal of Neurobiology 54, 272 (2003).
B. R. Johnson, T. A. Linksvayer, The Quarterly Review of Biology 85, 57 (2010).
E. Leadbeater, et al., Science 333, 874 (2011).
A. Liebert et al., Journal of Insect Behavior 21, 72 (2008).
F. Manfredini, et al., PLoS Genetics 9(8): e1003633 (2013).
F. Manfredini, et al., BMC genomics 16.1 (2015).
L. Pardi, Physiological Zoology 21, 1 (1948).
D. C. Queller et al., Nature 405, 784 (2000).
D. C. Queller, J. E. Strassmann, BioScience 48, 165–175(1998).
H. K. Reeve, in The Social Biology of the Wasps (Cornell University Press, Ithaca, New York, 1991).
G. E. Robinson, R. D. Fernald, D. F. Clayton, Science 322, 896 (2008).
M. J. Sheehan, E. A. Tibbetts, Current Biology 18.18 (2008).
A. L. Toth et al., Science 318, 441 (2007).
A. L. Toth et al., Proceedings of the Royal Society B-Biological Sciences 277, 2139 (2010).
A. L. Toth et al., BMC genomics 15.1 (2014).
S. Turillazzi, M. J. West-Eberhard, Natural history and evolution of paper-wasps (Oxford, 1996).
W. R. Tschinkel, The Fire Ants. (Harvard University Press, Cambridge, MA, 2006).
W. Tschinkel, D. Howard, Behavioral Ecology and Sociobiology 12, 103 (1983).
J. Wang et al., Nature 00, 000 (2013).
E. O. Wilson, The insect societies. (1971).
Y. Wurm et al., PNAS 108, 5679 (2011).
A. Zayed, G. E. Robinson, Annual Review of Genetics 46, 591 (2012).