A deep neural network to assess spontaneous pain from mouse facial expressions.

Grimace scales quantify characteristic facial expressions associated with spontaneous pain in rodents and other mammals. However, these scales have not been widely adopted largely because of the time and effort required for highly trained humans to manually score the images. Convoluted neural networks were recently developed that distinguish individual humans and objects in images. Here, we trained one of these networks, the InceptionV3 convolutional neural net, with a large set of human-scored mouse images. Output consists of a binary pain/no-pain assessment and a confidence score. Our automated Mouse Grimace Scale integrates these two outputs and is highly accurate (94%) at assessing the presence of pain in mice across different experimental assays. In addition, we used a novel set of "pain" and "no pain" images to show that automated Mouse Grimace Scale scores are highly correlated with human scores (Pearson's r = 0.75). Moreover, the automated Mouse Grimace Scale classified a greater proportion of images as "pain" following laparotomy surgery when compared to animals receiving a sham surgery or a post-surgical analgesic. Together, these findings suggest that the automated Mouse Grimace Scale can eliminate the need for tedious human scoring of images and provide an objective and rapid way to quantify spontaneous pain and pain relief in mice.

Olfactory exposure to males, including men, causes stress and related analgesia in rodents.

We found that exposure of mice and rats to male but not female experimenters produces pain inhibition. Male-related stimuli induced a robust physiological stress response that results in stress-induced analgesia. This effect could be replicated with T-shirts worn by men, bedding material from gonadally intact and unfamiliar male mammals, and presentation of compounds secreted from the human axilla. Experimenter sex can thus affect apparent baseline responses in behavioral testing.

Hoxb8 intersection defines a role for Lmx1b in excitatory dorsal horn neuron development, spinofugal connectivity, and nociception.

Spinal cord neurons respond to peripheral noxious stimuli and relay this information to higher brain centers, but the molecules controlling the assembly of such pathways are poorly known. In this study, we use the intersection of Lmx1b and Hoxb8::Cre expression in the spinal cord to genetically define nociceptive circuits. Specifically, we show that Lmx1b, previously shown to be expressed in glutamatergic dorsal horn neurons and critical for dorsal horn development, is expressed in nociceptive dorsal horn neurons and that its deletion results in the specific loss of excitatory dorsal horn neurons by apoptosis, without any effect on inhibitory neuron numbers. To assess the behavioral consequences of Lmx1b deletion in the spinal cord, we used the brain-sparing driver Hoxb8::Cre. We show that such a deletion of Lmxb1 leads to a robust reduction in sensitivity to mechanical and thermal noxious stimulation. Furthermore, such conditional mutant mice show a loss of a subpopulation of glutamatergic dorsal horn neurons, abnormal sensory afferent innervations, and reduced spinofugal innervation of the parabrachial nucleus and the periaqueductal gray, important nociceptive structures. Together, our results demonstrate an important role for the intersection of Lmx1b and Hoxb8::cre expression in the development of nociceptive dorsal horn circuits critical for mechanical and thermal pain processing.

Development of PainFace software to simplify, standardize, and scale up mouse grimace analyses.

ABSTRACT: Facial grimacing is used to quantify spontaneous pain in mice and other mammals, but scoring relies on humans with different levels of proficiency. Here, we developed a cloud-based software platform called PainFace (http://painface.net) that uses machine learning to detect 4 facial action units of the mouse grimace scale (orbitals, nose, ears, whiskers) and score facial grimaces of black-coated C57BL/6 male and female mice on a 0 to 8 scale. Platform accuracy was validated in 2 different laboratories, with 3 conditions that evoke grimacing-laparotomy surgery, bilateral hindpaw injection of carrageenan, and intraplantar injection of formalin. PainFace can generate up to 1 grimace score per second from a standard 30 frames/s video, making it possible to quantify facial grimacing over time, and operates at a speed that scales with computing power. By analyzing the frequency distribution of grimace scores, we found that mice spent 7x more time in a "high grimace" state following laparotomy surgery relative to sham surgery controls. Our study shows that PainFace reproducibly quantifies facial grimaces indicative of nonevoked spontaneous pain and enables laboratories to standardize and scale-up facial grimace analyses.

Development of PainFace software to simplify, standardize, and scale up mouse grimace analyses.

ABSTRACT: Facial grimacing is used to quantify spontaneous pain in mice and other mammals, but scoring relies on humans with different levels of proficiency. Here, we developed a cloud-based software platform called PainFace ( http://painface.net ) that uses machine learning to detect 4 facial action units of the mouse grimace scale (orbitals, nose, ears, whiskers) and score facial grimaces of black-coated C57BL/6 male and female mice on a 0 to 8 scale. Platform accuracy was validated in 2 different laboratories, with 3 conditions that evoke grimacing-laparotomy surgery, bilateral hindpaw injection of carrageenan, and intraplantar injection of formalin. PainFace can generate up to 1 grimace score per second from a standard 30 frames/s video, making it possible to quantify facial grimacing over time, and operates at a speed that scales with computing power. By analyzing the frequency distribution of grimace scores, we found that mice spent 7x more time in a "high grimace" state following laparotomy surgery relative to sham surgery controls. Our study shows that PainFace reproducibly quantifies facial grimaces indicative of nonevoked spontaneous pain and enables laboratories to standardize and scale-up facial grimace analyses.

Loss of neuronal potassium/chloride cotransporter 3 (KCC3) is responsible for the degenerative phenotype in a conditional mouse model of hereditary motor and sensory neuropathy associated with agenesis of the corpus callosum.

Disruption of the potassium/chloride cotransporter 3 (KCC3), encoded by the SLC12A6 gene, causes hereditary motor and sensory neuropathy associated with agenesis of the corpus callosum (HMSN/ACC), a neurodevelopmental and neurodegenerative disorder affecting both the peripheral nervous system and CNS. However, the precise role of KCC3 in the maintenance of ion homeostasis in the nervous system and the pathogenic mechanisms leading to HMSN/ACC remain unclear. We established two Slc12a6 Cre/LoxP transgenic mouse lines expressing C-terminal truncated KCC3 in either a neuron-specific or ubiquitous fashion. Our results suggest that neuronal KCC3 expression is crucial for axon volume control. We also demonstrate that the neuropathic features of HMSN/ACC are predominantly due to a neuronal KCC3 deficit, while the auditory impairment is due to loss of non-neuronal KCC3 expression. Furthermore, we demonstrate that KCC3 plays an essential role in inflammatory pain pathways. Finally, we observed hypoplasia of the corpus callosum in both mouse mutants and a marked decrease in axonal tracts serving the auditory cortex in only the general deletion mutant. Together, these results establish KCC3 as an important player in both central and peripheral nervous system maintenance.

Coding of facial expressions of pain in the laboratory mouse.

Facial expression is widely used as a measure of pain in infants; whether nonhuman animals display such pain expressions has never been systematically assessed. We developed the mouse grimace scale (MGS), a standardized behavioral coding system with high accuracy and reliability; assays involving noxious stimuli of moderate duration are accompanied by facial expressions of pain. This measure of spontaneously emitted pain may provide insight into the subjective pain experience of mice.

Monoaminergic mediation of hyperalgesic and analgesic descending control of nociception in mice.

ABSTRACT: Descending control of nociception (DCN; also known as conditioned pain modulation [CPM], the behavioral correlate of diffuse noxious inhibitory controls) is the phenomenon whereby pain inhibits pain in another part of the body and is the subject of increasing study because it may represent a biomarker of chronic pain. We recently discovered that pain modulation on the application of a DCN paradigm involving low-intensity test stimuli occurs in the direction of hyperalgesia in healthy mice and rats, whereas the use of high-intensity stimuli produces analgesia. To elucidate the physiological mechanisms underlying hyperalgesic DCN, we administered agonists and antagonists of norepinephrine (NE) and serotonin (5-HT) receptors, key neurochemical players in the production of analgesic DCN. We find that 3 different monoamine reuptake inhibitors-the NE-selective reboxetine, the 5-HT-selective fluoxetine, and the dual NE/5-HT agonist duloxetine-all abolish hyperalgesic DCN when administered into the spinal cord (but not systemically), with no effect on heat or mechanical pain sensitivity. The reversal by reboxetine of hyperalgesic DCN is mediated by α 2 -adrenergic receptors (ie, blocked by atipamezole), and the fluoxetine reversal is mediated by 5-HT 7 receptors (ie, blocked by SB269970). By contrast, analgesic DCN was found to be reversed by atipamezole and SB269970 themselves, with no effect of reboxetine or fluoxetine. Thus, hyperalgesic DCN seems to be the neurochemical opposite to analgesic DCN. These data further validate and help elucidate a preclinical paradigm that mimics dysfunctional CPM and thus may form the basis of translational experiments that aim to reveal preventative pharmacological strategies for individuals predisposed to persistent pain.

Sex-specific effects of neuropathic pain on long-term pain behavior and mortality in mice.

ABSTRACT: Human epidemiological studies suggest that chronic pain can increase mortality risk. We investigated whether this was true in mice so that underlying mechanisms might be identified. At 10 weeks of age, C57BL/6 mice of both sexes received sham or spared nerve injury (SNI) surgery producing neuropathic pain. Mice were weighed monthly, tested behaviorally for mechanical and cold sensitivity and guarding behavior every 3 months postsurgery, and otherwise left undisturbed in their cages until death by natural causes. Evidence of pain over the lifespan displayed a strikingly sex-specific pattern. Male mice displayed largely stable mechanical and cold hypersensitivity and guarding at 6 to 30 months post-SNI. By contrast, female mice displayed a biphasic temporal pattern of mechanical hypersensitivity and guarding behavior, with a complete resolution of SNI-induced pain behavior at 6 to 9 months post-SNI followed by the return of pain thereafter. Mouse lifespan was not significantly altered by SNI in either sex nor was frailty as assessed by cage inspection in the last 6 months of life. However, in male mice with SNI, we observe a significant correlation between average lifetime mechanical hypersensitivity and lifespan, such that death occurred sooner in male mice exhibiting more evidence of chronic pain. This relationship was not observed in female SNI mice nor in sham-operated mice of either sex. This experiment is the first to investigate pain behavior over an entire adult lifetime and suggests that biology of relevance to human chronic pain is being ignored by the very short timespans of most extant preclinical pain research.

Spinal cord Toll-like receptor 4 mediates inflammatory and neuropathic hypersensitivity in male but not female mice.

The innate immune system is increasingly appreciated to play an important role in the mediation of chronic pain, and one molecule implicated in this process is the Toll-like receptor 4 (TLR4). Here, using pharmacological and genetic manipulations, we found that activating TLR4 in the spinal cord, with the agonist lipopolysaccharide (LPS), causes robust mechanical allodynia but only in male mice. Spinal LPS had no pain-producing effect in female mice. TLR4 also has a sex-specific role in inflammatory (complete Freund's adjuvant) and neuropathic (spared nerve injury) pain: pain behaviors were TLR4 dependent in males but TLR4 independent in females. The sex differences appear to be specific to the spinal cord, as LPS administered to the brain or the hindpaw produces equivalent allodynia in both sexes, and specific to pain, as intrathecal LPS produces equivalent hypothermia in both sexes. The involvement of TLR4 in pain behaviors in male mice is dependent on testosterone, as shown by gonadectomy and hormone replacement. We found no sex differences in spinal Tlr4 gene expression at baseline or after LPS, suggesting the existence of parallel spinal pain-processing circuitry in female mice not involving TLR4.

Olfactory exposure to late-pregnant and lactating mice causes stress-induced analgesia in male mice.

In an attempt to improve reproducibility, more attention is being paid to potential sources of stress in the laboratory environment. Here, we report that the mere proximity of pregnant or lactating female mice causes olfactory-mediated stress-induced analgesia, to a variety of noxious stimuli, in gonadally intact male mice. We show that exposure to volatile compounds released in the urine of pregnant and lactating female mice can themselves produce stress and associated pain inhibition. This phenomenon, a novel form of female-to-male chemosignaling, is mediated by female scent marking of urinary volatiles, such as n-pentyl-acetate, and likely signals potential maternal aggression aimed at defending against infanticide by stranger males.

Long-term male-specific chronic pain via telomere- and p53­mediated spinal cord cellular senescence.

Mice with experimental nerve damage can display long­lasting neuropathic pain behavior. We show here that 4 months and later after nerve injury, male but not female mice displayed telomere length (TL) reduction and p53­mediated cellular senescence in the spinal cord, resulting in maintenance of pain and associated with decreased lifespan. Nerve injury increased the number of p53­positive spinal cord neurons, astrocytes, and microglia, but only in microglia was the increase male­specific, matching a robust sex specificity of TL reduction in this cell type, which has been previously implicated in male­specific pain processing. Pain hypersensitivity was reversed by repeated intrathecal administration of a p53­specific senolytic peptide, only in male mice and only many months after injury. Analysis of UK Biobank data revealed sex-specific relevance of this pathway in humans, featuring male­specific genetic association of the human p53 locus (TP53) with chronic pain and a male-specific effect of chronic pain on mortality. Our findings demonstrate the existence of a biological mechanism maintaining pain behavior, at least in males, occurring much later than the time span of virtually all extant preclinical studies.

The Rat Grimace Scale: a partially automated method for quantifying pain in the laboratory rat via facial expressions.

We recently demonstrated the utility of quantifying spontaneous pain in mice via the blinded coding of facial expressions. As the majority of preclinical pain research is in fact performed in the laboratory rat, we attempted to modify the scale for use in this species. We present herein the Rat Grimace Scale, and show its reliability, accuracy, and ability to quantify the time course of spontaneous pain in the intraplantar complete Freund's adjuvant, intraarticular kaolin-carrageenan, and laparotomy (post-operative pain) assays. The scale's ability to demonstrate the dose-dependent analgesic efficacy of morphine is also shown. In addition, we have developed software, Rodent Face Finder®, which successfully automates the most labor-intensive step in the process. Given the known mechanistic dissociations between spontaneous and evoked pain, and the primacy of the former as a clinical problem, we believe that widespread adoption of spontaneous pain measures such as the Rat Grimace Scale might lead to more successful translation of basic science findings into clinical application.

Mast cell stabilizer ketotifen fumarate reverses inflammatory but not neuropathic-induced mechanical pain in mice.

INTRODUCTION: Mast cell (MC) activation could establish a positive feedback loop that perpetuates inflammation and maintains pain. Stabilizing MCs with ketotifen fumarate (KF) may disrupt this loop and relieve pain. OBJECTIVE: We aimed to test the effect of treatment with KF in pain assays in mice and in a case series of patients with chronic widespread pain. METHODS: The analgesic effect of KF was tested in CD-1 mice injected with formalin, complete Freund's adjuvant, or subjected to spared nerve injury. In addition, wild-type (C57BL/6) and MC-deficient (C57BL/6-Kit W-sh/W-sh) mice were injected with formalin or complete Freund's adjuvant and treated with KF. Patients with chronic widespread pain (n = 5; age: 13-16 years) who failed to respond to standard of care participated in a 16-week treatment trial with KF (6 mg/d). Ketotifen fumarate's therapeutic effect was evaluated using the patient global impression of change. RESULTS: In the mouse experiments, KF produced dose- and MC-dependent analgesic effects against mechanical allodynia in the acute and chronic inflammatory pain but not neuropathic pain assays. In the patient case series, 4 patients reported that activity limitations, symptoms, emotions, and overall quality of life related to their pain condition were "better" or "a great deal better" since beginning treatment with KF. This was accompanied by improvements in pain comorbid symptoms. CONCLUSION: Treatment with KF is capable of reducing established inflammatory-induced mechanical nociception in an MC-dependent manner in mice, and it may be beneficial for the treatment of chronic pain conditions.

The influence of aging and duration of nerve injury on the antiallodynic efficacy of analgesics in laboratory mice.

INTRODUCTION: Increasing attention is being paid to the effects of organismic factors like age on pain sensitivity. However, very little data exist on this topic using modern algesiometric assays and measures in laboratory rodents. OBJECTIVES: We investigated the effect of age and duration of nerve injury on baseline mechanical thresholds, neuropathic allodynia, and the antiallodynic and analgesic efficacy of 4 systemically administered analgesics: amitriptyline, diclofenac, morphine, and pregabalin. METHODS: Mice of both sexes and 3 conditions were compared: Young-Young, in which baseline testing (von Frey thresholds), the injury producing neuropathic pain (spared nerve injury [SNI]) and subsequent drug testing occurred while mice were young (8-10 weeks); Young-Old, in which mice received the nerve injury while young but were tested for drug efficacy over 10 months later; and Old-Old, in which both the nerve injury and drug testing occurred at approximately 1 year of age. RESULTS: Old-Old mice were found to display higher baseline mechanical sensitivity than other groups. No group differences were seen in SNI-induced allodynia in males; female Young-Old mice were found to display greatly reduced allodynia. With respect to drug efficacy, no differences among conditions were observed for amitriptyline, diclofenac, or morphine. For pregabalin, however, Young-Old mice displayed significantly reduced antiallodynia, and the drug was completely ineffective in Old-Old mice. CONCLUSION: Novel findings include the apparent remission of SNI-induced allodynia in female mice 10 months after injury and reduced pregabalin antiallodynic effects produced by both the passage of time after nerve injury and aging.

Recruitment of Spinoparabrachial Neurons by Dorsal Horn Calretinin Neurons.

The dorsal horn of the spinal cord is the first integration site of somatosensory inputs from the periphery. In the superficial layers of the dorsal horn, nociceptive inputs are processed by a complex network of excitatory and inhibitory interneurons whose function and connectivity remain poorly understood. We examined the role of calretinin-expressing interneurons (CR neurons) in such processing and show that they receive direct inputs from nociceptive fibers and polysynaptic inputs from touch-sensitive Aß fibers. Their activation by chemogenetic or optogenetic stimulation produces mechanical allodynia and nocifensive responses. Furthermore, they monosynaptically engage spinoparabrachial (SPb) neurons in lamina I, suggesting CR neurons modulate one of the major ascending pain pathways of the dorsal horn. In conclusion, we propose a neuronal pathway in which CR neurons are positioned at the junction between nociceptive and innocuous circuits and directly control SPb neurons in lamina I.

Increased pain sensitivity and decreased opioid analgesia in T-cell-deficient mice and implications for sex differences.

The processing of pain in the central nervous system is now known to have an important immune component, including T cells of the adaptive immune system. T cells have been shown to release endogenous opioids, and although it is well known that opioids have effects on T-cell populations, very little attention has been given to the converse: how T cells may affect opioid regulation. We find here that, in addition to displaying significantly increased baseline pain sensitivity across various pain modalities, T-cell-deficient mice (CD-1 nude, Rag1 null mutant, and Cd4 null mutant) exhibit pronounced deficiencies in morphine inhibition of thermal or inflammatory pain. Nude mice are also deficient in endogenous opioid-mediated analgesia, exhibiting no stress-induced analgesia from restraint. The relevant T-cell subpopulation seems to be CD4 T cells because adoptive transfer of them but not CD8 cells into nude mice rescues both the pain and morphine analgesia phenotypes. As previously reported, we also observe a sex difference in CD-1 mice, with females requiring 2- to 3-fold more morphine than males to produce equal analgesia. Nude mice display no sex differences in morphine analgesia, and the sex difference is restored in nude mice of either sex receiving CD4 T cells from CD-1 donor male or female mice. These results suggest that CD4 T cells play an as yet unappreciated role in opioid analgesia and may be a driver of sex differences therein.

Modulation of social behavior and dominance status by chronic pain in mice.

The potential influence of pain on social behavior in laboratory animals has rarely been evaluated. Using a new assay of social behavior, the tube co-occupancy test (TCOT), we assess propinquity-the tendency to maintain close physical proximity-in mice exposed to pain using subcutaneous zymosan or spared nerve injury as noxious stimuli. Our previous experience with the TCOT showed that outbred mouse sibling dyads show higher levels of tube co-occupancy than stranger dyads. We find here that long-lasting pain from spared nerve injury given to both mice in the dyad abolishes this effect of familiarity, such that strangers also display high levels of propinquity. We performed a separate experiment to assess the effect on dominance behavior of nerve injury to one or both mice of a dyad in which relative dominance status had been previously established via the confrontation tube test. We find that neuropathic pain given only to the dominant mouse reverses the relationship in male but not female mice, such that the previously subordinate mouse becomes dominant. These observations bolster the scant but growing evidence that pain can robustly affect social behavior in animals.

Conditioned pain modulation in rodents can feature hyperalgesia or hypoalgesia depending on test stimulus intensity.

The counterirritation phenomenon known as conditioned pain modulation, or diffuse noxious inhibitory control in animals, is of increasing interest due to its utility in predicting chronic pain and treatment response. It features considerable interindividual variability, with large subsets of pain patients and even normal volunteers exhibiting hyperalgesia rather than hypoalgesia during or immediately after receiving a conditioning stimulus. We observed that mice undergoing tonic inflammatory pain in the abdominal cavity (the conditioning stimulus) display hyperalgesia, not hypoalgesia, to noxious thermal stimulation (the test stimulus) applied to the hindpaw. In a series of parametric studies, we show that this hyperalgesia can be reliably observed using multiple conditioning stimuli (acetic acid and orofacial formalin), test stimuli (hindpaw and forepaw-withdrawal, tail-withdrawal, hot-plate, and von Frey tests) and genotypes (CD-1, DBA/2, and C57BL/6 mice and Sprague-Dawley rats). Although the magnitude of the hyperalgesia is dependent on the intensity of the conditioning stimulus, we find that the direction of effect is dependent on the effective test stimulus intensity, with lower-intensity stimuli leading to hyperalgesia and higher-intensity stimuli leading to hypoalgesia.

Effect of Human Genetic Variability on Gene Expression in Dorsal Root Ganglia and Association with Pain Phenotypes.

Dorsal root ganglia (DRG) relay sensory information to the brain, giving rise to the perception of pain, disorders of which are prevalent and burdensome. Here, we mapped expression quantitative trait loci (eQTLs) in a collection of human DRGs. DRG eQTLs were enriched within untranslated regions of coding genes of low abundance, with some overlapping with other brain regions and blood cell cis-eQTLs. We confirm functionality of identified eQTLs through their significant enrichment within open chromatin and highly deleterious SNPs, particularly at the exon level, suggesting substantial contribution of eQTLs to alternative splicing regulation. We illustrate pain-related genetic association results explained by DRG eQTLs, with the strongest evidence for contribution of the human leukocyte antigen (HLA) locus, confirmed using a mouse inflammatory pain model. Finally, we show that DRG eQTLs are found among hits in numerous genome-wide association studies, suggesting that this dataset will help address pain components of non-pain disorders.