Inter-neuronal variability in human nociceptor electrophysiology: experimentally-driven computational study of response to drugs and channelopathies
Lead Research Organisation:
University of Oxford
Department Name: Computer Science
Abstract
Chronic pain affects millions of people worldwide, but currently lacks effective treatments. A potential target for new therapies to relieve chronic pain are the millions of specialised sensory neurons in our bodies, called nociceptors, that detect harmful stimuli and transmit this information to the brain, where it can be perceived as pain. Nociceptors normally provide us with a protective sense of pain that helps us to avoid harm. Recently, patients with chronic neuropathic pain (unwanted pain due to damage to the nervous system) and with complete insensitivity to pain have had their pain states linked to genetic mutations in specific ion channels that are found most commonly in nociceptors. These ion channels are proteins found in the outer membranes of nerve cells, that open and close to control the flow of electrically charged ions, such as sodium and potassium, into and out of the cell. Each nociceptor contains many different ion channels, and the combined flow of electrical current through these channels determines how it responds when we encounter both harmless and harmful stimuli. These findings have shown that there are particular ion channels that play a vital role in determining which stimuli cause each of us to experience pain.
Studies into how these ion channel mutations affect pain signalling, and the search for new drugs that could treat chronic pain by targeting nociceptor ion channels are ongoing. Currently these studies use nociceptors from animals, particularly rats and mice. This is because human nociceptors, donated to research by organ donors, are scarce and we have limited data from them. Unfortunately there is not a perfect match between animal and human nociceptors and ion channels. Therefore findings made using animal models may not translate to human biology. Nociceptors themselves are also highly varied, as they are divided up into different sub-types, sensitive to different stimuli such as heat or cold, and show differences in behaviour between individuals.
To address these problems, the goal of this project is to develop computer models of human nociceptors that are representative of the wide range of variability we see between nociceptor sub-types, and between nociceptors from different individuals. Computational models of nociceptor electrical activity mathematically describe how the different ion channels in a human nociceptor open and close, and how this affects the signalling properties of the nociceptor. The equations in these models are too complex to solve by hand, and must be solved on computers. Usually these models only describe the average behaviour of nociceptors, but instead we have developed a method to construct population of nociceptor models, consisting of thousands of different models, where every model produces behaviours that are within the range of biological variability observed in experiments, but where every model has a slightly different makeup, such as having different densities of each type of ion channel in its membrane. Therefore, each model behaves differently, and responds differently to simulated application of drugs or the insertion of an ion channel mutation.
Mutations and other factors such as nerve injury and inflammation can make nociceptors hyper excitable. This means they can fire signals for longer, or at lower thresholds, leading to unwanted, non-protective pain. Drugs that block particular ion channels could return these nociceptors back to normal excitability, but effective therapies of this kind have yet to be developed. Current findings suggest individual drugs alone may not be sufficient to achieve this, and combinations of drugs might be more effective. We will use computer simulations to screen a wide range of different drug combinations to predict which combinations restore normal excitability to nociceptors exposed to inflammatory agents, and to nociceptors with ion channel mutations linked to the development of neuropathic pain.
Studies into how these ion channel mutations affect pain signalling, and the search for new drugs that could treat chronic pain by targeting nociceptor ion channels are ongoing. Currently these studies use nociceptors from animals, particularly rats and mice. This is because human nociceptors, donated to research by organ donors, are scarce and we have limited data from them. Unfortunately there is not a perfect match between animal and human nociceptors and ion channels. Therefore findings made using animal models may not translate to human biology. Nociceptors themselves are also highly varied, as they are divided up into different sub-types, sensitive to different stimuli such as heat or cold, and show differences in behaviour between individuals.
To address these problems, the goal of this project is to develop computer models of human nociceptors that are representative of the wide range of variability we see between nociceptor sub-types, and between nociceptors from different individuals. Computational models of nociceptor electrical activity mathematically describe how the different ion channels in a human nociceptor open and close, and how this affects the signalling properties of the nociceptor. The equations in these models are too complex to solve by hand, and must be solved on computers. Usually these models only describe the average behaviour of nociceptors, but instead we have developed a method to construct population of nociceptor models, consisting of thousands of different models, where every model produces behaviours that are within the range of biological variability observed in experiments, but where every model has a slightly different makeup, such as having different densities of each type of ion channel in its membrane. Therefore, each model behaves differently, and responds differently to simulated application of drugs or the insertion of an ion channel mutation.
Mutations and other factors such as nerve injury and inflammation can make nociceptors hyper excitable. This means they can fire signals for longer, or at lower thresholds, leading to unwanted, non-protective pain. Drugs that block particular ion channels could return these nociceptors back to normal excitability, but effective therapies of this kind have yet to be developed. Current findings suggest individual drugs alone may not be sufficient to achieve this, and combinations of drugs might be more effective. We will use computer simulations to screen a wide range of different drug combinations to predict which combinations restore normal excitability to nociceptors exposed to inflammatory agents, and to nociceptors with ion channel mutations linked to the development of neuropathic pain.
Technical Summary
Animal models of dorsal root ganglion (DRG) neurons are widely used in pain research as in vitro models of human nociception, due to a lack of human-specific alternatives. However, these models do not capture human-specific electrophysiology, including differences in ion channel function, and do not address significant inter-neuronal variability, e.g. differences in ion channel expression and action potential morphology between DRG neuron sub-types. This heterogeneity is difficult to address through experiments alone but can result in variable responses to therapies and disease.
We have developed a method for integrating biological variability with in silico modelling, using experimentally-calibrated populations of models, and have used this approach extensively in cardiac electrophysiology. We propose integrating new recordings of human DRG neuron electrophysiology, provided by our collaboration with Anabios Corporation, with our methodology to construct and validate populations of in silico human DRG models that include inter-neuronal electrophysiological variability at two levels: variability between individual neurons, and variability between neuronal sub-types as determined by sensitivity to noxious stimuli (e.g. heat, irritating chemicals or cold).
Mutations in specific sodium channel isoforms (e.g. Nav 1.7, 1.8 and 1.9) can lead to chronic pain and insensitivity to pain. This finding has spurred development of selective blockers of these channels. However, the effects of blocking these channels and the underlying mechanisms of these mutations are not fully understood. We propose using populations of human DRG models to improve on and replace existing animal DRG experiments that are used to predict the range of responses to ion channel blocking drugs, and to understand the mechanisms that link mutations in individual ion channels with their effects on DRG neuron excitability.
We have developed a method for integrating biological variability with in silico modelling, using experimentally-calibrated populations of models, and have used this approach extensively in cardiac electrophysiology. We propose integrating new recordings of human DRG neuron electrophysiology, provided by our collaboration with Anabios Corporation, with our methodology to construct and validate populations of in silico human DRG models that include inter-neuronal electrophysiological variability at two levels: variability between individual neurons, and variability between neuronal sub-types as determined by sensitivity to noxious stimuli (e.g. heat, irritating chemicals or cold).
Mutations in specific sodium channel isoforms (e.g. Nav 1.7, 1.8 and 1.9) can lead to chronic pain and insensitivity to pain. This finding has spurred development of selective blockers of these channels. However, the effects of blocking these channels and the underlying mechanisms of these mutations are not fully understood. We propose using populations of human DRG models to improve on and replace existing animal DRG experiments that are used to predict the range of responses to ion channel blocking drugs, and to understand the mechanisms that link mutations in individual ion channels with their effects on DRG neuron excitability.
Planned Impact
Rat and mouse dorsal root ganglion (DRG) neurons are the predominant in vitro animal models used in pain research and preclinical drug development, due to the limited availability of human DRGs. We propose development of in silico populations of neuronal models as a replacement for in vitro experiments used for two roles:
1) To determine drug-induced changes in human DRG excitability.
2) To predict and understand the effect on human DRG excitability caused by ion channel mutations and associated mechanisms of the mutations.
We searched PubMed to estimate the use of rodents in these classes of experiments.
Our base search query was: (mouse OR mice OR rat OR rodent) AND (nociceptor OR DRG OR dorsal root ganglion) AND pain AND (ion channel OR ion channels OR Nav) with the addition of:
1. AND (drug block OR channel block OR pharmacological block OR current inhibition OR current block)
2. AND (mutation OR channelopathy)
The base search produced 3669 papers published over the last 5 years, indicating wide use of these models. Queries 1 and 2 produced 79 and 39 non-review papers published over the last 5 years respectively, for a total of 118 papers. We analysed the first 20 papers for each query to determine average numbers of animals used for DRG studies only. Mean animals used for DRG studies in the 9 papers where number of animals used was reported was 64, with a range of 10 to 181. Therefore annual animal use in these experiments is estimated at 1500 animals, while total animal use in all DRG experiments is estimated at 45,000 animals. These numbers does not include animals use in industry, which is likely to be substantial given the current interest in developing selective sodium channel blockers targeting human DRGs. 31 of 40 papers did not report the numbers of animals used. Numbers of observations were usually numbers of neurons, without information on numbers of animals used to obtain the neurons.
We believe adoption of our models could allow replacement of 50% of current studies, replacing the use of 750 animals annually. In the long term, adoption of our models in 10% of total DRG studies in pain could replace use of 4,500 animals annually. In addition, this project could also facilitate replacement in the following ways:
1) As part of a combined approach with other non-animal methods in pain research. Human DRGs, stem cell-derived DRGs, and in silico modelling incorporating inter-neuronal variability could provide a human-specific, animal-free platform for research. Here, in silico models would be used to develop hypotheses and provide mechanistic insight, iterating between modelling and experiment.
2) Through interest from industry and regulatory agencies, as has been the case with our work in cardiac electrophysiology, facilitated by our Virtual Assay software which allows use of the methodology without modelling and programming expertise. Virtual Assay supports human cardiac modelling for prediction of drug effects, and is extensible to other areas of biology.
For dissemination of models, software and the skills required to use them our plan is the following:
1) Models and software to run simulations and analyse results will be freely available and open source. We will use the programming language Python, and the NEURON simulation software, which has been used in over 1450 publications. Both are freely available and widely used in scientific computing.
2) Provide online video and text tutorials to explain how to use the models, run simulations, analyse results, and replicate published figures, aimed at a scientific audience without previous computational expertise.
3) Organise a workshop to disseminate the models and raise awareness in the pain research community and in industry. We have previously organised an inter-sectorial workshop in cardiac electrophysiology to increase awareness of our work in this manner.
4) Integrate models into Virtual Assay software following interest from industry.
1) To determine drug-induced changes in human DRG excitability.
2) To predict and understand the effect on human DRG excitability caused by ion channel mutations and associated mechanisms of the mutations.
We searched PubMed to estimate the use of rodents in these classes of experiments.
Our base search query was: (mouse OR mice OR rat OR rodent) AND (nociceptor OR DRG OR dorsal root ganglion) AND pain AND (ion channel OR ion channels OR Nav) with the addition of:
1. AND (drug block OR channel block OR pharmacological block OR current inhibition OR current block)
2. AND (mutation OR channelopathy)
The base search produced 3669 papers published over the last 5 years, indicating wide use of these models. Queries 1 and 2 produced 79 and 39 non-review papers published over the last 5 years respectively, for a total of 118 papers. We analysed the first 20 papers for each query to determine average numbers of animals used for DRG studies only. Mean animals used for DRG studies in the 9 papers where number of animals used was reported was 64, with a range of 10 to 181. Therefore annual animal use in these experiments is estimated at 1500 animals, while total animal use in all DRG experiments is estimated at 45,000 animals. These numbers does not include animals use in industry, which is likely to be substantial given the current interest in developing selective sodium channel blockers targeting human DRGs. 31 of 40 papers did not report the numbers of animals used. Numbers of observations were usually numbers of neurons, without information on numbers of animals used to obtain the neurons.
We believe adoption of our models could allow replacement of 50% of current studies, replacing the use of 750 animals annually. In the long term, adoption of our models in 10% of total DRG studies in pain could replace use of 4,500 animals annually. In addition, this project could also facilitate replacement in the following ways:
1) As part of a combined approach with other non-animal methods in pain research. Human DRGs, stem cell-derived DRGs, and in silico modelling incorporating inter-neuronal variability could provide a human-specific, animal-free platform for research. Here, in silico models would be used to develop hypotheses and provide mechanistic insight, iterating between modelling and experiment.
2) Through interest from industry and regulatory agencies, as has been the case with our work in cardiac electrophysiology, facilitated by our Virtual Assay software which allows use of the methodology without modelling and programming expertise. Virtual Assay supports human cardiac modelling for prediction of drug effects, and is extensible to other areas of biology.
For dissemination of models, software and the skills required to use them our plan is the following:
1) Models and software to run simulations and analyse results will be freely available and open source. We will use the programming language Python, and the NEURON simulation software, which has been used in over 1450 publications. Both are freely available and widely used in scientific computing.
2) Provide online video and text tutorials to explain how to use the models, run simulations, analyse results, and replicate published figures, aimed at a scientific audience without previous computational expertise.
3) Organise a workshop to disseminate the models and raise awareness in the pain research community and in industry. We have previously organised an inter-sectorial workshop in cardiac electrophysiology to increase awareness of our work in this manner.
4) Integrate models into Virtual Assay software following interest from industry.