The translatability of pain across species

Abstract

The poor translational record of pain research has suggested to some observers that species differences in pain biology might be to blame. In this review, I consider the evidence for species similarity and differences in the pain research literature. Impressive feats of translation have been demonstrated in relation to certain genetic effects, social modulation of pain and pain memory. The degree to which pain biology in rodents predicts pain biology in humans has important implications both for evolutionary accounts of pain, but also the success of analgesic drug development going forward.

This article is part of the Theo Murphy meeting issue ‘Evolution of mechanisms and behaviour important for pain’.

1. Poor translation in pain research and the translational assumption

The field of pain has suffered through several decades with almost no regulatory approvals of new molecular entities for the treatment of chronic pain (with the curious exception of migraine), commonly interpreted as a failure of preclinical-to-clinical translation. In the most recent of a number of critical reviews essentially asking, ‘what went wrong?’, Yezierski & Hansson [1] point to the clinical trial failure of no less than 14 therapeutic targets identified by preclinical research. A fair accounting of this unenviable record would, however, note that: (i) not all of these targets were abandoned because of lack of efficacy in humans (some failed because of poor tolerability or the presence of rare but serious side effects), (ii) some of the targets may not actually be permanently abandoned, (iii) in very few cases were drug exposure levels actually determined, and (iv) decisions surrounding the design of many of these trials may have been based more on marketing considerations than scientific ones. Neither is this problem unique to pain research; one can easily point to even more egregious numbers of failed clinical trials of treatments for Alzheimer's disease [2], stroke [3] and sepsis [4], for example.

Despite there being no paucity of alternative explanations, many have placed the blame for poor translation onto preclinical research. In general, there have been three separate complaints. Some have questioned the integrity of preclinical research itself, suggesting we are in the midst of a ‘replication crisis’ [5] brought on by poor incentives, poor scientific reporting, and sloppy and/or inappropriate statistical practices. This, of course, is a problem of internal validity. Many have also pointed to the existence of poor external validity: the poor fit between the features of commonly used animal models and the clinical reality of the diseases they are supposed to model. I have previously reviewed this mismatch in depth [6–8]. Suffice it to say that pain researchers have made, in retrospect, some arguably ill-advised choices regarding their choice of subject sex (male), genotype (inbred C57BL/6), age (young), time of testing following injury (soon after), method of inducing pain (artificial) and outcome measures (evoked and reflexive). Finally, some have suggested that species differences in biology between humans and the rodents we almost unanimously rely on as animal models [8] have been underestimated.

This possibility is the one with the most relevance to the question of the evolutionary conservation (or lack thereof) of pain among mammalian species. Modern pain research is built on a ‘translational assumption’, broadly speaking that mouse = rat = human. This assumption seems especially likely to be valid for pain, given its evolutionary maturity. As described elsewhere in this Special Issue, pain, or at least pain-related behaviours, are phylogenetically ancient and widespread [9–11]. No one takes the translational assumption literally, of course; there will always be species differences, especially of a quantitative nature. Any preclinical scientist who doubts that the biology of pain is qualitatively similar in rodents and humans should not be a preclinical scientist at all. If major qualitative differences exist in pain mechanisms between species, a comprehensive and causal understanding of pain biology might not ever be possible. But what evidence has actually been gathered on this all-important question?

2. Species differences in pain biology

Although humans and rodents diverged 65–75 Ma [12], only 300 genes appear to be unique to one or the other species [13]. This might mitigate against discovering major species differences in all but the most modern evolutionary traits. The very existence of clinical trials that failed because of lack of efficacy is often taken as prima facie evidence for species differences in biology, but as described above, this would be a logical mistake. If a species difference in drug efficacy indeed exists, it is much more likely to be a quantitative difference related to pharmacokinetics (PK; e.g. differing penetration or metabolism) or pharmacodynamics (PD; e.g. differing receptor levels) than a true qualitative difference related to the suitability of the drug target itself (i.e. protein X is involved in pain processing in rodents but not humans). This being said, it is notable that the most celebrated analgesic development failure—neurokinin-1 (NK1) antagonists—failed to show clinical efficacy against pain even though PK/PD issues are not apparent, as evidenced by the clinical use of the NK1 antagonist, aprepitant, as an anti-emetic [14].

The paucity of known species differences in pain between rodents and humans is probably owing to the limited opportunities for direct comparisons between them. Very few pain-relevant tissues are accessible in live donor humans, including biopsied skin, and rarely, dorsal root ganglia (DRG; from traumatic avulsions, or patients undergoing tumour resection or spinal reconstruction surgery, for example), but excluding central nervous system tissue. Electrophysiological data obtained from human DRG are broadly similar to that observed in the rodent literature (see [15]), as is DRG neurotrophin receptor expression [16], although direct comparisons are problematic because of the necessarily heterogeneous nature of the human clinical cohorts. Much of what we know about pain mechanisms in humans derives from imaging studies, but functional imaging in unanaesthetized laboratory animals remains extremely rare [17]. Microneurography—minimally invasive recording of intact peripheral nerve fibres in vivo—can be performed in both humans and rodents, but is technically difficult and even more rarely attempted in the latter [18]. The few extant studies have observed similar conductance velocities in rodents and humans [19,20].

One interesting and rare example of a multi-species comparison is in the expression of the SCN9A gene encoding the Na_v1.7 sodium channel [21], a study undertaken to explain why humans with loss-of-function mutations of this gene are generally healthy (other than their complete pain insensitivity) [22], whereas mice with global null mutations of the same gene were reported to die shortly after birth ([23], but see [24]). Using in situ hybridization and immunohistochemistry, it was shown that adrenal and pituitary glands of rodents, but not primates, express Na_v1.7 channels [21]. Other noted examples of pain-relevant species differences include: (i) the insensitivity of certain small mammals (Octodon degus and Ochotona rufescens) to morphine analgesia [25,26], (ii) the insensitivity of various bird species to transient receptor potential, V1 (TRPV1) agonists, like capsaicin [27], (iii) the insensitivity to acid pain of the naked mole-rat [28], (iv) species-specific activation or blockade of the transient receptor potential, A1 (TRPA1) channel by many ligands [29], and (v) different electrical properties of rodent versus human Na_v1.8 channels [30]. In an attempt to explain the clinical trial failure of the glial modulating drug, propentofylline, it was demonstrated that lipopolysaccharide treatment induced nitrite release in rat but not human glial cells [31]. In fact, there are many documented mouse–human species differences in immune system parameters [32,33].

What about direct comparisons between rats (representing 81% of preclinical pain research in 2008; [8]) and mice (representing 14% of preclinical pain research in 2008; [8])? There are, in fact, very few published head-to-head comparisons, and all should be treated with caution, as it is difficult to distinguish a true species difference from a difference between one strain of rat versus one strain of mouse. For example, a study of thermal pain sensitivity in three rat strains and three mouse strains revealed increased sensitivity in female Long Evans rats and Swiss Webster mice compared to males, decreased sensitivity in female Sprague Dawley rats compared to males, and no sex differences in Wistar Kyoto rats, CD-1 or ND4 mice [34]. Comparisons of one particular rat strain to one particular mouse strain would lead to all possible conclusions regarding ‘species differences’ in the sex-dependence of pain sensitivity. Table 1 lists available findings of qualitative rat-versus-mouse differences relevant to pain, either in head-to-head studies or where the authors explicitly referred to a finding in the species currently under study as contrasting with the literature obtained using the other species. One would expect that many more mouse/rat species differences in pain are known to pharmaceutical company personnel, but never published in the scientific literature.

Table 1.

Published qualitative rat versus mouse species differences of relevance to pain.

difference	rat	mouse	reference
effect of isolation on morphine analgesia	decreases	increases	[35]
behaviour after cyclophosphamide cystitis	+abdominal crises	−abdominal crises	[36]
number of lumbar vertebrae	6	variable (5–6)	[37]
genes regulated by chronic pain (microarray)	43 genes (see main text)	10 genes	[38]
secretagogin expression in spinal cord	no colocalization	colocalized with CGRPc	[39]
TRPV1a expression in nociceptors	peptidergic only	peptidergic + IB4+b	(cf. [40])
traumatic brain injury-induced allodynia	unrelated to stress	dependent on stress	[41]

^aTransient receptor potential, V1.

^bIsolectin B4.

^cCalcitonin gene-related polypeptide.

One of these findings is especially worrisome. LaCroix-Fralish et al. [38] performed a meta-analysis on all 20 microarray (the antecedent of RNAseq technology) gene expression profiling studies of chronic neuropathic or inflammatory pain published as of 2009. A total of 79 genes were found to be significantly upregulated or downregulated in more studies than would be predicted by chance. Quantitative-polymerase chain reaction experiments were subsequently performed in an attempt to independently confirm all 79 genes. Forty-three genes were confirmed in the rat, not unexpectedly, because 18 out of 20 microarray studies leading to the generation of the list of 79 genes were originally performed in that species. However, using identically treated tissue from the mouse, only 10 of 79 genes were confirmed to be regulated. This finding strongly suggests that a different set of genes is regulated by chronic pain in the mouse compared to the rat, which would itself be indicative of broadly species-specific pathophysiology of chronic pain.

3. Translational pain experiments

My laboratory has, over the past few years, attempted to perform wholly translational pain studies in which essentially the same experiment is performed in mice and humans and reported contemporaneously. Three such projects are particularly noteworthy.

Differences among mouse strains in their sensitivity to various modalities of pain are ubiquitous and robust [42]. One of the most reliable of these is the difference between the A/J and C57BL/6 J strains, which are resistant and sensitive, respectively, to inflammatory noxious stimuli causing tonic (on the order of minutes) nocifensive behaviours such as formalin (late-phase) and capsaicin licking [43–45]. Using an F₂ intercross between mice of these strains, we localized the genomic region containing a gene responsible for a large portion of this strain-dependent response to distal mouse chromosome 10 [44]. The region in question contained hundreds of genes, however, and various ‘positional cloning’ techniques were employed until the responsible gene was finally identified as Avpr1a, coding for the mouse vasopressin-1A receptor [45]. Null mutant mice lacking vasopressin-1A receptors displayed 50% more licking behaviour after capsaicin injection than wild-type mice, leading to the hypothesis that human AVPR1A variants might similarly affect capsaicin pain ratings in our species. A subsequent comparison of capsaicin pain visual analogue scores of humans with various alleles at the most common AVPR1A single nucleotide polymorphism revealed no differences among groups. But a more comprehensive analysis of the data revealed the existence of a highly significant three-way interaction between the AVPR1A gene, participant sex, and participant stress ratings immediately prior to capsaicin cream application, such that AVRP1A genotype affected pain levels only in men experiencing high levels of stress [45]. Armed with this knowledge, we revisited the mouse data, which revealed that the strength of the genetic association was indeed far larger in male versus female mice. In addition, new experiments were performed in A/J and C57BL/6 J mice that were either extensively habituated to the testing environment (and thus under low stress at the time of pain testing) or not (and thus under high stress, as in the original experiments). We found that a strain difference was only observed in non-habituated mice [45], suggesting that the vasopressin-1A gene wasn't a ‘pain gene’ at all, but rather a ‘stress-induced analgesia gene,’ in both mice and humans. Thus, in this series of experiments, a mouse finding inspired a human study that revealed subtleties that were then directly testable in new mouse experiments. Very few such examples of mouse-to-human-to-mouse translations have ever been reported, but the paradigm is certainly a powerful one.

Empathy is often held up as a human-specific ability, inaccessible to other animals. In fact, there have been scientific and anecdotal reports to the contrary, but these have been largely dismissed as anthropomorphizing (see [40]). In 2006, we provided comprehensive evidence that mice were capable of a rudimentary form of empathy [46], albeit one that all more complex forms are probably dependent on: emotional contagion. These experiments demonstrated that mice displayed higher levels of pain behaviour when tested alongside familiar (but not stranger) conspecifics, and featured synchronization of both level and timing of that behaviour within the dyad. This finding has been replicated and extended in a number of ways, and the study of empathy-like processes in rodents is now an active subfield of social/cognitive neuroscience (see [47–50]). Although empathy has and continues to be widely studied in humans, direct evidence of emotional contagion of pain in our species had not been explicitly demonstrated. We set out initially to investigate why emotional contagion of pain did not also occur between stranger mice, and determined using a pharmacological approach that it was blocked by stress among both members of the dyad [51]. Indeed, we had previously observed—unsurprisingly in retrospect—that placing stranger males together in a confined space increases stress levels as measured via deposition of faecal boli or plasma corticosterone [52]. A human version of the mouse dyadic pain testing paradigm was designed whereby participants were tested for cold pressor (30 s in 4°C water) pain alone, and also directly across from either a friend or stranger. Emotional contagion, like in mice, was operationalized as higher pain sensitivity in the dyadic versus the isolated condition. We found that either pharmacological (metyrapone administration, as in mice) or psychological (a shared videogame experience) stress reduction was sufficient to allow the emergence of emotional contagion of pain between strangers [51]. This study showed that not only can a phenomenon be convincingly demonstrated in mice and humans simultaneously, but that the operating principles underlying that phenomenon (i.e. emotional contagion being dependent on low-stress levels) were similar in both species.

An increasing number of prominent pain researchers are starting to conceptualize chronic pain in humans as a learning and memory phenomenon, designed to protect against threat [53,54], and in rodents evidence of the overlap of biological mechanisms subserving memory and pain continues to mount [55,56]. Despite this, direct evidence of context-dependent pain hypersensitivity has been decidedly lacking. We devised an extremely simple protocol in the mouse, involving only a single pairing of unconditioned stimulus (UCS; intraperitoneal acetic acid) with conditioned stimulus (CS; the testing environment itself), and observed reliable context-dependent increases in sensitivity to acute thermal pain (the conditioned response; CR) [57]. That this phenomenon would exist was not in the least surprising; what was wholly unexpected was our observation that it seemed to occur only in male mice, owing to testosterone-dependent increases in stress produced by the conditioning. The human paradigm we devised to parallel the mouse study was remarkably similar, with the same CS and CR as in mice (albeit applied to a different body part); in humans the UCS was a 20 min application of the submaximal effort tourniquet test [58]. As in mice, male-specific context-dependent pain hypersensitivity was observed, and as in mice the sex difference was owing to the fact that only males became stressed when the CS was reintroduced [57].

4. Future directions and implications for evolutionary hypotheses

I believe that pain research would be well served by more examples of translational pain studies, where ‘translational’ is defined as similar experiments performed on widely varying mammalian species, one of them (where possible) being Homo sapiens. Insight gained from humans can focus and prioritize rodent experiments, and rodent experiments are essential at separating causation from mere correlation. Although the current vogue in analgesic drug development is to rely on human genetics for proof-of-principle, the number of analgesic targets that can be validated this way is likely to be extremely limited. In any case, a true causal and mechanistic understanding of the target's true role in pain pathophysiology can only come from experiments performed in animal models [7]. The more examples of successful translation that are documented, the more confidence we are likely to have in the success of clinical interventions derived from and/or supported by animal studies.

It is unclear, however, as to whether a comparison of successful translations (mice≈humans) to species differences (mice≠humans) will have much bearing on the broader questions of the phylogenetic continuity of pain biology, or the appropriateness of rodents as model organisms for human pain pathophysiology. For that, direct experimental evidence on questions of pain adaptations will need to be compiled. Such evidence is currently in extremely short supply, especially from rats and mice representing the vast majority of non-human subjects of pain research [8].

Data accessibility

This article has no additional data.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.