This is the first of a four-part report on selected talks from the 10th IASP Research Symposium, The Genetics of Pain: Science, Medicine and Drug Development. See also Part 2, Part 3, and Part 4, or download a PDF of the entire report.
Pain researchers convened on 7-9 February 2012 for a first-of-its-kind symposium dedicated to the genetics of pain. The meeting, sponsored by the International Association for the Study of Pain (IASP) and organized by the Genetics and Pain Special Interest Group, drew 120 registrants and 20 faculty to Miami Beach, Florida, US, for talks that ran the gamut from gene discovery in fruit flies to clinical development of ion channel blockers.
The meeting signaled a coming of age for the field of pain genetics. In his talk, Clifford Woolf, Children’s Hospital Boston, US, said that he is not a geneticist, but he joined the pain genetics train looking for research tools. “And it has been a wonderful ride—sometimes on a ghost train, sometimes stalled in a siding, and sometimes running out of control and almost derailed.” The meeting, he said, showed that the study of the genetics of pain “has become mainstream in neurobiology, and that’s exciting.”
Of mice (and rats) and men (and women)
The opening session, on preclinical studies of pain genetics, highlighted the central place of rodents in the discovery of pain genes, and in understanding the interplay of genes and environment. The talks showed the power of a translational approach that moves from animals to humans and back again to uncover candidate pain genes, elucidate mechanisms, and validate targets.
The first speaker of the conference, Marshall Devor, Hebrew University, Jerusalem, Israel, laid the groundwork for talks to come. Discovering genes that influence individual variation in pain sensitivity or risk for chronic pain serves two purposes, he said. First, it opens doors to understanding individual patients: making a diagnosis, illuminating a prognosis, predicting drug responses, or even comforting patients by telling them, “It's not your fault.” Second, gene discovery and, in particular, unbiased genomewide methods will benefit many patients by uncovering novel pain mechanisms and potential new treatment targets. The real power, as Devor sees it, is the “promise for finding pathways in the physiology of pain that we never dreamed of before.”
The study of monogenetic pain diseases—rare occurrences like familial migraine, congenital insensitivity to pain, or congenital pain syndromes—has illuminated important players in pain pathways. But to identify genes that contribute to more common pain conditions, other approaches are needed. Linkage analysis is used when researchers have access to family members with and without pain. When it is impossible to do family studies (in post-operative pain, e.g., where every family member would have to have had the same operation), the alternative is association studies. Here, groups of unrelated people are compared on a case-control basis to identify genetic variants that are distributed unevenly in those with pain and those without. Association studies test either a limited number of pre-selected genes (a candidate gene approach) or all variants in an unbiased screen (a genomewide approach).
Devor said he believes genomewide association studies (GWAS) in humans are “the path to real discovery,” but such investigations are expensive, and so far very few have been funded for pain. He made the case that doing genomewide scans in animals is a way to “jumpstart” human studies, enabling the ultimate identification of human genes at a much lower cost. Identifying mouse strains that vary in the phenotype of choice (or making such lines by selective breeding) allows for the mapping of a phenotype to one or more quantitative trait loci (QTL). This can lead to the identification of the responsible pain gene or genes, which can then be confirmed in humans.
An example of this is the discovery that a variant of the CACNG2 gene is a risk factor for chronic pain after breast surgery (Nissenbaum et al., 2010). More than a decade ago, Devor collaborated with Jeffrey Mogil, McGill University, Montreal, Canada, and others to phenotype a panel of 12 common mouse strains in multiple pain tests, including autotomy after nerve injury, a model of neuropathic pain (Mogil et al., 1999). Further breeding and linkage analysis, and the use of inbred recombinant strains, led to the identification of a locus on chromosome 15 (Seltzer et al., 2001; Devor et al., 2005), which was ultimately narrowed to 155 candidate genes. Using expression data, functional annotation, and single nucleotide polymorphism (SNP) association, Devor and colleagues zeroed in on a single candidate, CACNG2. As the group reported in 2010, when they then looked at variants in the corresponding human gene, they found a haplotype of three SNPs that was associated with an increased risk of pain after mastectomy in a sample of 549 Israeli women, some of whom had pain and some of whom did not.
“If this can be replicated, it’s useful,” Devor said. Possibly, if a woman is genetically more likely to develop pain, that information may be part of a decision on what type of surgery to have, for example.
CACNG2 encodes the voltage-dependent calcium channel gamma subunit 2 (also known as stargazin), a protein which both regulates the trafficking of AMPA-type glutamate receptors and controls the excitability of neurons. The gene has been implicated in epilepsy, indicating a “deep connection between epilepsy and neuropathic pain having to do with excitability of neural networks,” Devor said.
High-throughput phenotyping in a new mouse model
Ze’ev Seltzer, University of Toronto, Canada, described an ambitious translational project involving high-throughput phenotyping in a mouse model of neuropathic spreading pain, and a large cohort of traumatic limb amputees he has assembled for genetic analysis of post-amputation neuropathic pain.
Seltzer described a new mouse model of unilateral damage to the infraorbital nerve (ION). In mice, as in other mammals and humans, the ION innervates the face from just below the eyebrow to the upper lip, and from the side of the nose to the cheek. In rodents, the ION also innervates the vibrissal pad that anchors the whiskers. Large parts of the rodent brain are dedicated to processing sensory input from the whiskers, and many labs have used ION injury in rats and mice as a model of neural plasticity, but not to study pain.
In new work, Seltzer and collaborators showed that they could use ION injury in mice to model the “extra-territorial” spread of neuropathic pain, a problem that occurs in many chronic pain patients where pain spreads beyond the field innervated by an injured nerve to surrounding regions. If they cut the ION, sensation was lost in the innervation area, including the whisker pad, but then over weeks some strains of mice developed increased pain sensitivity (allodynia and hyperalgesia) to mechanical and heat stimuli in the ears, contralateral whisker pad, forehead, nose, all four paws, and even the tail, “practically all over their body,” Seltzer said.
To investigate the genetic basis of pain spread, the researchers compared the levels of allodynia and hyperalgesia in male and female mice from three different genetic backgrounds (A/J, C57BL/6J, and DBA/2J strains). The highest contrast in the extent of spread of mechanical and heat allodynia and hyperalgesia was found between A/J and C57BL/6J mice of both sexes. Based on these differences, Seltzer’s group selected the AXB-BXA panel of 23 recombinant inbred lines, descendants of the two contrasting parental strains, for a detailed genetic mapping study. With publicly available genomic data on the strains, and Internet-based mapping software (WebQTL), Seltzer said, the task of associating each trait with a specific region of the genome becomes a simple statistical analysis.
However, it takes more than that to identify and validate the causative genes in each region. To narrow the list of candidate genes, Seltzer and colleagues at the University of Zhejiang, China, phenotyped an additional 15 strains of fully haplotype-mapped mice (the Peltz HapMap panel; see Zheng et al., 2012). The strength of this approach is that it fine-maps traits to chromosomal regions orders of magnitude smaller than the AXB-BXA panel. Combining the two assays is expected to decrease the candidate gene list to a manageable few.
What can they expect for results from these kinds of studies? Seltzer gave one example: the response of naïve animals to noxious heat, where his group sees a strong sex effect in some lines but not others. Mapping of the trait yielded five loci, including a major one on chromosome 7 that includes the gene for a metabotropic glutamate receptor. More work is needed, he said, to validate this gene.
The Toronto and Zhejiang teams have just finished phenotyping pain spread in approximately 1,600 mice from 40 strains or lines after ION injury or sham surgery. To help with the task, Seltzer and colleagues developed high-throughput avoidance tests for thermal or mechanical pain. They built a nine-chamber testing apparatus, where each chamber floor has a 30°C section and a 40°C section. Naïve animals show no temperature preference, but the higher temperature is painful to animals with neuropathic injuries. To test for mechanical allodynia, the chambers are fitted with smooth and spiked floor sections, and allodynic animals avoid the spiked surface. Video tracking is used to analyze animal behavior, and the set-up allows testing of nine animals in parallel in three to five minutes.
Moving to people
While mice can be marshaled by the hundreds or thousands for genetic studies, one of the biggest challenges in human pain genetics is achieving large, well-characterized cohorts to attain sufficient statistical power in association studies. To address that issue, Seltzer has been recruiting thousands of subjects for a genetic study of post-traumatic neuropathic pain in an unusual locale: Cambodia. He and his colleagues have enrolled 5,500 amputees who sustained injuries from land mines and unexploded munitions that still litter the country 20 years after the end of the last war. Approximately one in 230 Cambodians is an amputee (the highest rate in the world), and there are an estimated 60,000 amputees in the country. In Cambodia, one of the poorest countries in the world, the injured receive no first aid, no acute pain control, and cannot afford adequate treatment for chronic pain. However, the Cambodian government, with the help of foreign organizations, offers free prostheses and rehabilitation to land-mine victims. Seltzer’s collaborators have been working in the clinics where the replacement limbs are to recruit study volunteers and collect their DNA. The phenotyping consists of a three-hour interview that gathers demographic data, medical history, information on prior pain, amputation scenario, pain type, disability, and psychological state, as well as quantitative sensory testing (pin prick, cold pressor test, and thermal grill illusion, among other data). All the data, along with automated measurement of heart rate, blood pressure, and glucose levels, are beamed back to Toronto by satellite. In addition to the amputees, most of whom are male, Seltzer and colleagues have also collected DNA from 500 controls, plus 500 male family members of the affected group.
To achieve even larger sample sizes for genetic studies of neuropathic pain, Seltzer and collaborators have set up a consortium which they call the Tactical Union of Research Networks in Pain Genetics (TURN-PAGE). By combining seven cohorts, including amputees and post-surgical patients, the consortium now covers just over 10,000 subjects, by far the largest cohort assembled for pain genetics. The consortium includes amputees from Cambodia, Germany (collected by Herta Flor and coworkers at the Heidelberg University, Mannheim, Germany), and Israel (collected by Seltzer and his colleagues at Sheba and Beit Levinstein Medical Centers, Israel), five cohorts of post-surgical neuropathic pain, including two cohorts of women post-mastectomy (collected in the US by Inna Belfer and her colleagues at the University of Pittsburgh, and in Israel by Seltzer and colleagues at Hadassah and Sheba Medical Centers), a cohort of post-cardiac surgery patients (collected by Joel Katz, Hance Clarke, and Seltzer at Toronto General Hospital, Canada), and a cohort of neuropathic pain patients (the Canadian Multicenter Neuropathic Pain Database (NePDAT) Registry collected by Dwight Moulin, Western University, Ontario, Canada, and others). The consortium also collaborates with deCODE Genetics, Iceland, which has access to a large group of various chronic pain conditions. The total number of subjects is expected to increase to 14,000 by the end of 2013.
The next challenge is to find the money to do the genetic studies. “Each item on the genotyping menu has a price tag,” Seltzer said. The consortium would like to do a genomewide association study for common variants, as well as look for rare trait-associated loci. All of these studies are costly, and Seltzer said he is now looking for support from pharmaceutical companies and from the governmental funding agencies in the US and Canada. The consortium has teamed up with Beijing Genetics Institute, the largest genotyping facility in the world, to do exome sequencing in several hundred amputees, comparing those with the highest levels of pain and pain-free subjects. Also in the works is a screening with a commercially available exomics array.
It’s complicated
How are we doing overall in pain genetics? Good in some ways, said Jeffrey Mogil, McGill University, Montreal, Canada. His Pain Gene Database currently lists data on 358 knockout mice with a pain phenotype, drawn from 839 publications. “What percentage of those genes will end up contributing to individual differences is anyone’s guess,” he said, but he believes it will be a pretty large one. In humans, association studies have exploded since 2005, implicating approximately 120 genes. But, he pointed out, many associations have proven hard to replicate.
The problem, Mogil suggested, is that “we have been too simpleminded.” Chronic pain is the result of many interactions, and it is known that genes interact with other genes and with environmental factors. Unaccounted-for interactions may explain why association studies don’t replicate more often than they do now. However, Mogil said, “If we position ourselves to look at these interactions, we’re going to find them more often than not.” That means collecting as many data as possible to increase the chances of identifying hidden complexities.
As evidence, he presented his recent work on vasopressin and stress-induced analgesia (see PRF related news story on Mogil et al., 2011). Mogil and colleagues first identified an association of the vasopressin receptor gene Avpr1a with inflammatory pain sensitivity in mice. But, following up on the result in humans, they unearthed a three-way interaction among the receptor gene variants, stress, and sex. All three factors determine whether mice, or humans, display an analgesic response to administered vasopressin. The study revealed that vasopressin mediates stress-induced analgesia, but only in males. Along the way, it served up a cautionary tale for gene hunters: The original finding, an association between the receptor gene and inflammatory pain in mice, was only detected because the animals were stressed during the testing procedure. Animals that were acclimated to the testing scenario beforehand did not show the association.
“I don’t think we can get away from the complexity, because it’s part of the system,” Mogil said.
Mice made for the job
A pair of talks in later sessions expanded on the utility of mouse genetics. In the years since Mogil’s original analysis of strain differences in pain phenotypes, the development of panels of recombinant inbred strains and advances in sequencing have given researchers ever more powerful genomewide tools to reveal the genetic underpinnings of pain.
William Lariviere, University of Pittsburgh, US, described the use of recombinant inbred strains as genetic tools. These strains serve as useful reference populations, because their genotype, phenotype, and expression profiles are stable over time and data can be pooled and shared among different labs. He has utilized the BXD recombinant inbred panel (a set of strains derived from crosses between C57 Black 6/6J and DBA/2J strains) to identify QTLs for mechanical sensitivity and nociception or acetic acid-induced abdominal extensions (a model of inflammatory pain). In each case, combining linkage mapping with publicly available expression data has helped to narrow down the candidate gene list (e.g., see Nair et al., 2011). Lariviere is now also studying the genetic determinants of the response to melittin in honeybee venom, an alternative therapy used to treat arthritis (reviewed in Chen and Lariviere, 2010). He reported that some mouse strains are hypersensitive to the painful effects of melittin, a trait that maps to an interval on chromosome 18 containing only five known genes. Other strains are hyposensitive, and that trait maps to a two-gene interval on chromosome 19. For the hyposensitivity trait, expression covariance data implicate the gene Atrnl1, which encodes a protein that interacts with the melanocortin 4 receptor, but whose function is unclear and has never been implicated in pain before.
“We are just at the beginning of exploiting these models,” Lariviere said. At the same time, he said, the BXD model is "going out of style." Because the panel results from the cross of just two strains, there is limited genetic diversity. However, a new model is nearing completion. The collaborative cross is the product of breeding between eight founder strains, of which three are derived from wild-type mice. The resulting strains together will represent greater than 90 percent of mouse genetic diversity and offer higher-precision genetic mapping. Results using the new strains in tests of thermal sensitivity (hot plate latency) reveal a locus on chromosome 5 containing six genes, and for mechanical nociception (tail clip test), a locus on chromosome 2 containing only five known genes (Philip et al., 2011). For new genetic studies, Lariviere said, this is the model researchers should be proposing in their grant applications, unless they can justify needing all the additional supporting information already available for the BXD strains.
Roy Levitt, University of Miami Miller School of Medicine, US, has tapped into another mouse resource to identify genes that contribute to the risk of persistent pain after surgery. The mouse HapMap project has genotyped 94 strains of mice and identified eight million SNPs. To look for genetic variants associated with pain after surgery, Levitt and colleagues measured thermal withdrawal in 16 inbred strains of mice (using the Hargreaves test) at baseline, 1, 7, 14, and 21 days after chronic constriction injury of the sciatic nerve. They plotted the withdrawal latency over time, and used the area under the curve to calculate a persistent pain index that they showed was heritable. Levitt presented unpublished data on haplotype association mapping using three-SNP windows that revealed loci on chromosome 5 (covering five genes) and 12 (containing one gene) associated with persistent post-surgical pain.
The gene on chromosome 12 encodes Nova-1 (neuro-oncological ventral antigen-1), a tissue-specific RNA-binding protein that regulates alternative splicing in the central nervous system. Nova-1’s targets include many proteins in the inhibitory synapse (Ule et al., 2005). Nova-1 protein regulates the splicing of kinases and phosphatases, and the phosphodomains in many proteins (Zhang et al., 2010). Therefore, the protein has the capacity to affect overall phosphorylation patterns in the central nervous system.
Levitt and colleagues looked at Nova-1 protein and activity in dorsal root ganglia associated with injury. They found that the protein is expressed in nociceptors, and that its activity differs among mouse strains. A comprehensive RNA profiling revealed statistically significant differences in the splicing patterns of Nova-1 targets among strains, which might account for differences in susceptibility to post-operative pain.
Is Nova-1 associated with pain in people? Levitt found preliminary evidence that says yes. In three cohorts (osteoarthritis, pain one year after hernia surgery, and post-herpetic neuralgia), there was a statistically significant association between one or more Nova1 SNPs and pain. No association was seen in a sciatica group.
Levitt says the next steps are to identify SNPs that explain the varied function of the gene and understand how they affect persistent pain susceptibility in animal models and in human populations. He added that, down the road, this could lead to diagnostics that help determine who is at risk for persistent post-operative pain, and better interventions.
