Chronic pain represents a significant public health concern and current therapies do not adequately meet patient needs. With the advent of genomic technologies, pain researchers have begun to identify key loci associating with human pain diseases, and these data are instructing the development of next generation analgesics. Although human genetics efforts have been effective, complementary approaches, including functional genomics, may provide additional insight into pain genetics and help identify additional new drug targets. In the present review, we discuss the use of fruit fly systems biology combined with mammalian genetics to accelerate the discovery of novel pain genes and candidate drug targets.
Pain affects millions of people worldwide, producing an enormous burden on an individual's quality of life. Pain perception (or nociception in model organisms) is critical for survival in the face of environmental danger and is under stringent evolutionary pressure. Studies with experimental animal models of pain have revealed significant contribution of genetic factors to variability in pain sensitivity and, when combined with genetic tools, have shed insight into the molecular nature of pain perception. Additional lines of evidence come from human genetics; with twin studies reporting approximately 50% of pain sensitivity as heritable [1-4]. Furthermore, human association studies have also reported racial, ethnic and individual differences in acute and chronic pain [5-11]. For example African Americans and non-Causasian Hispanics often report more pain sensitivity compared to Causacians [3, 8, 9, 12, 13] and females report greater pain than males [3, 10, 11, 14-16].
Efforts to associate pain sensitivity in humans with specific genetic loci have identified a handful of pain gene candidates, including catechol-O-methyltransferase (COMT) , GTP cyclohydrolase (GCH1) , mu opiod receptor (OPRM1) , an ionotropic ATP-gated receptor (P2X7R) , a serotonin transporter (SLC6A4)  and multiple loci of human leukocyte antigen (HLA) . However, as with other complex traits, these identified polymorphisms explain less than 5% of the genetics involved in chronic pain susceptibility . These data indicate that pain variability is influenced by genetics, and pinpointing these genetic determinants, as well as population-wide genetic variability in these elements, may inform future general or more patient-specific therapies.
The human, mouse and fly genomes have been sequenced [24-26]; however, until recently most genes had not been functionally annotated in vivo, and genomic approaches to investigate tissue or organ-specific function were not possible. In the present review, we focus on presenting our current functional genomics strategies to identify and validate the genes and pathways involved in pain perception (or nociception in model organisms). This includes the first ever full genome neural-specific functional annotation of any behaviour in any species, combined with mouse and human genetics and genomics approaches to validate novel candidate pain genes. These strategies aim to advance our understanding of the genetics of pain or nociception, and to aid in the development of novel classes of analgesics.
Drosophila as a suitable model for pain research
Drosophila melanogaster is a powerful tool for functional genomics research approaches. Maintaining a large fruit fly colony is much cheaper than vertebrate model organisms, and fruit fly breeding is also much faster as a result of a short 10-day life cycle. Most importantly, fruit flies show a high degree of homo-logy with humans at the organ and gene level, with flies sharing functional counterparts for most organ systems [27-30] and the fly genome contains approximately 60% of human disease genes [29-31]. Fruit flies are particularly useful for nociception because ethical concerns are a major issue in vertebrate pain research.
Over approximately the last 100 years, the Drosophila research community has developed a powerful genetic toolbox that allows researchers to collect vast amounts of in vivo functional information in relatively short periods of time . Significant recent advances in fruit fly technology involved massive efforts to generate full genome transgenic RNA interference fly libraries [33, 34]. This unprecedented technology now allows a researcher to target any gene in the genome in a temporal and tissue-specific manner, circumventing lethality, and enabling large-scale functional genomic screening for cell, organ or system autonomous functions [33-41]. This tool is especially useful for investigating the genomics of innate behaviour because behaviour requires an in vivo model. Remarkably, flies and humans exhibit many similar behavioural patterns, including courtship, aggression, circadian rhythm and sleep, learning and memory, and nociceptive avoidance behaviour [32, 42, 43].
Anatomically and functionally, Drosophila nociceptors resemble vertebrate nociceptors. For example, both exhibit characteristic naked-nerve endings and use ion channels of the transient receptor potential channel (TRP) family to generate neural/behavioural responses to noxious stimulation . Moreover, these nociceptors exhibit sensitization following injury or exposure to inflammatory mediators [44, 45]. These features permit pain researchers to combine behavioural, physiological and molecular genetic methods in multiple species to help guide research . More recently, large-scale genomic approaches have identified multiple genes and pathways that appear to be components of a conserved genetic architecture of nociception in flies to pain perception in humans [35, 46]. Despite these similarities, pain and nociception are multifactorial, complex processes, and many differences between human and Drosophila nociceptors exist. As a result, Drosophila nociception may only be able to model certain aspects of mammalian pain. Therefore, prudent interpretation of the genes and pathways identified in Drosophila nociception is necessary, and validation experiments using mammalian pain models are required to confirm a conserved role for the gene or pathway of interest.
Global screen for identifying nociception genes in Drosophila
A number of fly nociception assay systems have been developed to investigate mechanisms involved in thermal, chemical and mechanical nociception [35, 43-45, 47-52]. Detailed description of these assays can be found in recent reviews [53, 54]. Research using these systems has led to the identification of a limited number of fly nociception or conserved pain genes. These include the TRP channels NOMPC, painless  and TRPA1 , the sodium channel pickpocket , a novel class of conserved mechanoreceptors (DmPiezo; the Drosophila homologue of mammalian Piezo) [56, 57] and a Drosophila homologue of tumour necrosis factor (Eiger) and its receptor, Wengen [44, 45]. Given the fact that most human disease genes are conserved in the flies , there are likely many novel conserved pain genes remaining to be identified using fruit fly nociception models.
To accelerate pain gene discovery, we have performed a full genome functional and neural-specific assessment of noxious heat avoidance behaviour in the fruit fly. In Drosophila adults, assays to assess the loss of thermal nociceptive behaviour (analgesia) measure the percentage of flies that show failure to avoid a noxious thermal surface when given a choice between a noxious (46 °C) and innocuous (31 °C) surface  (Fig. 1A). Assessment of the thermal nociceptive response in Drosophila larvae is achieved by measuring latency time for a stereotypical rolling response after a fly larvae is touched by a heat probe  (Fig. 1B). Depending on the latency time, animals can be identified as exhibiting analgesia (decreased nociception), hyperalgesia (increased nociception) and allodynia (nociception to innocuous stimuli). These fly nociception assays are relatively high throughput and, when combined with whole-genome functional annotation, RNA interference has revealed hundreds of new conserved genes implicated in various levels of the heat nociceptive response .
A novel functional genomics approach to identify pain genes by combining fly and human systems
Genome-wide association studies (GWAS) have revolutionized biomedical research and have helped identify hundreds of genetic variants that play a role in various complex diseases . Major association efforts are currently underway for acute pain and susceptibility to various chronic pain conditions. Genetic susceptibility to migraine was the first pain-related GWAS reported , possibly because this condition is relatively common (approximately 10% of the population) and patients with this condition report pain as the major symptom , making it more amenable to large-scale genetic association studies. Several genetic variants associated with migraine have been reported. These include the lipoprotein receptor LRP1 suggested to interact with glutamate receptors [6, 59]; TRPM8 known to encode for a cold and pain sensor ; PRDM16 with unknown function, and a variant rs1835740 located between MTDH and PGCP, which are two genes involved in glutamate homeostasis . Additionally, a GWAS meta-analysis study of chronic widespread pain has uncovered that the variant rs13361160, located between chapronin containing TCP1-complex-5 gene (CCT5) and FAM173B, was associated with a 30% higher risk of chronic widespread pain . Another recent study using a combination of a heat sensitivity test and exome sequencing on a normal population showed that heat pain sensitivity is strongly associated with variation in GZMM, a gene encoding a serine protease called granzyme M, and moderately associated with several other genes, including PDHA2 encoding pyruvate dehydrogenase α2, DLD encoding dihydrolipoamide dehydrogenases and MYPN encoding myopalladin .
One major challenge for genome-wide human genomics approaches is the requirement for massive (often 100 000 participants) patient cohorts. For pain, these cohorts would take years or decades to organize, cost millions of dollars to genotype and, as a result of the inherently subjective aspects of pain perception and assessment, may exhibit a high degree of variability. Moreover, even with a large cohort and standardized phenotyping, unguided genome-wide association efforts require large statistical corrections for multiple comparisons and, in the end, usually only attribute significance to loci explaining only 5–10% of the total genetic contribution for a given disease or complex trait .
Providing functional a priori knowledge for a given gene in a specific trait or disease can help geneticists target evaluation of candidate gene association data to a subset of loci and, accordingly, increase power for targeted association efforts and improve the likelihood of identifying significant loci. We have applied this reasoning to pain, employing functional genomics strategies to evaluate conserved nociception genes in model organisms and, in this way, provide a priori knowledge to instruct targeted association by human geneticists [35, 36]. For pain, the identification of candidate conserved nociception genes was achieved by targeting human association analysis to genes that are functionally relevant in our systems, and then confirming significant results in multiple available human cohorts. If we can show that a gene is involved in nociception in flies and pain in humans, we can then focus on mechanistic studies using pain phenotyping assays using well established pain paradigms (e.g. transgenic mouse models) (Fig. 2).
Identification of pathways regulating nociception
Our functional annotation of the Drosophila genome uncovered over 500 genes involved in thermal nociception, 399 of which have human orthologues and a number of which are already established mammalian pain genes . In addition, several genes from this Drosophila screen have also been isolated in human association studies for pain. These consist of genes associated with chronic pain including OPRM1 encoding the mu opioid receptor  and ADRB2 encoding β-2 adrenergic receptor, a major target for epinephrine . Our Drosophila screen for heat nociception also identified fly orthologues for genes associated with heat pain sensitivity in humans such as PDHA2 (pyruvate dehydrogenase α2), DLD (dihydrolipoamide dehydrogenases), and MYPN (myopalladin) .
Because many common molecular signalling pathways exist in both Drosophila and humans, we performed functional enrichment analysis of mammalian orthologues of these fly nociception genes to predict molecular pathways involved in thermal nociception. Analysis of the 399 human orthologues and their first-degree binding partners (based on data from yeast-2-hybrid screens) using KEGG pathways and Broad Institute C2 gene sets allowed the construction of a comprehensive conserved network map for thermal nociception. Interestingly, this network map of nociception highlights a number of signalling pathways, including Ca2+ signalling, phosphatidylinositol signalling, protein kinase C, nuclear factor kappa-light-chain-enhancer of activated B cells, cAMP response element-binding protein nitric oxide synthase/inducible nitric oxide synthase, arachidonic acid, G protein-coupled receptor signalling and long-term potentiation. Our recent efforts have focused on further investigation of specific Ca2+ signalling molecules and phospholipid signalling pathways in multiple animal models of pain research and in patient cohorts.
Ca2+ signalling components as conserved components of nociception
Combining pain mediating genes found in a Drosophila full genome functional screen with targeted human associations for pain has allowed us to start defining the conserved core components of pain signalling. Chronic pain assessment was previously performed on a cohort of adult patients following surgery  who were monitored for chronic pain over 2 years, and DNA samples from these patients can be used for targeted genotyping . Interestingly, many conserved genes annotated for Ca2+ signalling were found to also show some level of association with chronic pain (I. Belfer, unpublished data). This suggests that the nociception genetic architecture is conserved from flies to humans, and our approach may yield novel insight into pain mechanisms. Although Ca2+ signalling has been implicated in nociception long ago , the results are encouraging and support our overall methodology. The conserved Ca2+ signalling genes implicated in pain perception from our approach include rdgC, CG5890, TRPA1, cpn, CadN, Caki, stj and CaMKII. So far, further mechanistic validation of this gene set for a role in nociception has been limited to TRPA1 and stj; however, a potential orthologue of CG5890 was previously shown to be a multimodal pain gene acting at the spinal gate .
The Drosophila TRP-family ion channel TRPA1 responds to subnoxious warmth and reactive electrophiles such as mustard oil, wasabi, cigarette smoke and tear gas [48, 67-69]. Additionally, studies in mouse models have shown that TPRA1 is specifically required for inflammatory heat nociception  and, more recently, a human TRPA1 gain-of-function mutation was linked to familial episodic pain syndrome . In flies, a role for TRPA1 in acute heat  and chemical  nociception has now been established, and dTRPA1 mutant flies show reduced thermal nociceptive responses in both adult and larval Drosophila nociception assays. In addition, dTRPA1 was shown to be required for Hh (hedgehog) signalling-induced hyperalgesia in Drosophila larvae following UV-induced tissue damage . Thus, despite some differences in nociceptive modalities, a role for TRPA1 in nociception or pain appears to be conserved across phyla.
Conserved functional genomics strategies have established that the Drosophila gene straight-jacket (stj, a subunit of multiple calcium channels) and its mammalian orthologue α2δ3, also play a role in invertebrate nociception and vertebrate pain sensation. Prior to this, stj was appreciated for a role in synaptic function at the neuromuscular junction . Compared to controls, stj mutant larvae show slower responses to heat stimuli. Defective responses were also observed in stj mutants in the adult fly nociception assay  (G.G. Neely, unpublished data). When the mammalian orthologue (α2δ3) was knocked-out in mice, acute and post-inflammatory thermal analgesia was observed. Furthermore, targeted genotyping in human pain cohorts has highlighted single nucleotide polymorphisms at the α2δ3 locus that associate with altered acute heat pain insensitivity or chronic post-surgical back pain. Importantly, the characterization of this gene has led to the possible identification of a second central pain gate in the mammalian thalamus, with α2δ3 as the first gene identified to be involved in processing of thermal pain at this site. Most interestingly, loss of α2δ3 resulted in a sensory cross-activation phenotype (reminiscent of synaesthesia in humans) and this is the first genetic insight into this process in any species .
Phosphoinositide signalling as a mediator of pain
In addition to Ca2+ signalling, our pathway analysis also highlighted phosphoinositide signalling as a component of the conserved nociception network . Phosphoinositide signalling involves the generation of phosphatidylinositol 4,5-bisphosphate (PIP2) via phosphatidylinositol-4-phosphate 5-kinase (PIP5K), which is further converted to phosphatidylinositol (3,4,5)-trisphosphate via phosphoinositide 3-kinase (PI3K). Although a role for phosphoinositide signalling in pain perception was first established over a decade ago , the precise role remains controversial. For example, PIP2 was reported to enhance the activity of TRPV1 , whereas other studies reported that PIP2 inhibited the activity of TRPV1 [73, 75]. Therefore, the specific signalling molecules and downstream effectors involved in this pathway, and their contribution to pain regulation, need to be further clarified. Targeted genotyping of human pain patients showed some disease association with PIP5Kα and PI3Kγ (I. Belfer, unpublished data); therefore, we focused further characterization on these factors. Compared to littermate controls, PIP5Kα mutant mice were found to exhibit a significant hyper-responsiveness to radiant heat and contact heat. Because some noxious heat stimuli are transmitted through TRPV1, we also investigated whether PIP5Kα mutant mice exhibit exaggerated TRPV1 responses. Indeed, it was observed that PIP5Kα mutant mice display heightened reactivity toward the TRPV1 agonist capsaicin. Based on our fly and human data, we also characterized PI3Kγ mutant mice (p110γ). PI3Kγ is the only G-protein coupled PI3K and is expressed in TRPV1-positive peripheral sensory neurones. Similar to PIP5Kα mutant mice, PI3Kγ mutant mice also exhibited a thermal hyperalgesia phenotype and exaggerated behavioural response to capsaicin. PI3Kγ has been shown to act through kinase dependent and kinase independent mechanisms [76, 77], and PI3Kγ regulation of pain sensitivity is through its kinase-dependent effects as PI3Kγ kinase-dead knock-in mice still exhibit a thermal hyperalgesic phenotype. Mechanistically, lipid signalling was shown to regulate heat responses in the dorsal root ganglion (DRG), with PI3Kγ−/− DRG neurones showing substantially stronger temperature dependence (Q10) of ion channel conductivity compared to wild-type littermates. Moreover, consistent with the observation of enhanced response to capsaicin in PI3Kγ−/− mice in vivo, isolated PI3Kγ−/− DRG neurones were found to exhibit heightened sensitivity to capsaicin . Taken together, these data highlight that phosphatidylinositol signalling negatively regulates heat pain perception and TRPV1 reactivity in vivo .
Pain is a complex trait governed by genetic variability, genetic networks and gene–environment interactions. One major goal in pain research is to identify the mechanisms and genetics involved in susceptibility and resistance to various pain diseases, with the goal of developing novel painkillers to manage this disease effectively. Efforts to determine the genetic basis of pain disease using candidate gene approaches have been successful, although much of the heritability of pain disease remains to be identified . Although human pain cohorts have been described, GWAS results for acute and chronic pain are not yet available. For many complex traits, including subjective behavioural indications such as pain, it may be difficult and expensive to assemble large standardized cohorts that are adequately powered, and developing other complimentary methods to identify new human disease mechanisms should be considered. One effective method for accelerating disease gene discovery is via a conserved functional genomics strategy to provide a priori disease relevance from model organisms, followed by assessment of disease participation through targeted human associations. For disease gene candidates that meet both criteria, further investigation in rodent models, or pharmacological intervention when possible, may then be warranted. In the present review, we describe our application of this strategy to assess the conserved genomics of nociception to pain perception, which has allowed us to highlight multiple new mammalian pain genes and pathways. An approach using functional genomics to ‘scan’ animal genomes for candidate disease genes, combined with more targeted association efforts in human cohorts, as well as validation experiments in mammals, not only serves as a complementary approach to large-scale association studies, but also may identify bona fide disease genes with true therapeutic potential that may never achieve significance if investigated solely via GWAS-based approaches.