Comparative biology of pain: What invertebrates can tell us about how nociception works

Abstract

The inability to adequately treat chronic pain is a worldwide health care crisis. Pain has both an emotional and a sensory component, and this latter component, nociception, refers specifically to the detection of damaging or potentially damaging stimuli. Nociception represents a critical interaction between an animal and its environment and exhibits considerable evolutionary conservation across species. Using comparative approaches to understand the basic biology of nociception could promote the development of novel therapeutic strategies to treat pain, and studies of nociception in invertebrates can provide especially useful insights toward this goal. Both vertebrates and invertebrates exhibit segregated sensory pathways for nociceptive and nonnociceptive information, injury-induced sensitization to nociceptive and nonnociceptive stimuli, and even similar antinociceptive modulatory processes. In a number of invertebrate species, the central nervous system is understood in considerable detail, and it is often possible to record from and/or manipulate single identifiable neurons through either molecular genetic or physiological approaches. Invertebrates also provide an opportunity to study nociception in an ethologically relevant context that can provide novel insights into the nature of how injury-inducing stimuli produce persistent changes in behavior. Despite these advantages, invertebrates have been underutilized in nociception research. In this review, findings from invertebrate nociception studies are summarized, and proposals for how research using invertebrates can address questions about the fundamental mechanisms of nociception are presented.

all animals have three basic tasks critical for their survival [finding resources (food) to maintain themselves, successful reproduction, and avoiding/minimizing serious bodily harm]. The ability to detect stimuli that elicit damage to the body or the potential for such damage is defined as nociception by the International Association for the Study of Pain (IASP). How an animal responds to such stimuli and how nociceptive stimuli impact behaviors aimed at fulfilling the drives of feeding and reproduction have considerable adaptive significance, an idea supported in recent experiments by Crook et al. (2014). These adaptive benefits suggest that nociception across animal phyla involves conserved cellular and physiological processes that produce a common behavioral response to noxious stimuli (Walters 1994; Walters and Moroz 2009; Woolf and Walters 1991). Keeping these comparative/evolutionary considerations in mind can help identify fundamental conserved processes of nociception that may provide insights into the mechanisms of pathological pain conditions and how these conditions might be treated (Williams 2016).

Pathological pain conditions represent a major health care crisis and are estimated to affect as many as 100 million people at an annual cost of approximately $635 billion in the United States alone (Gaskin and Richard 2012; Institute of Medicine Committee on Advancing Pain Research and Education 2011). Despite its obvious importance, progress in developing new pain therapies has been disappointing (Mogil 2009). This has prompted, justifiably, calls for changes in how pain research is conducted, specifically, a move away from measures of evoked reflexes traditionally used to assess pain levels in animal subjects and toward more complex and ethologically relevant behavioral assays that more closely recapitulate the clinical features of pain conditions in humans (Mogil 2009). There is no disagreement on this point. However, another equally important barrier to progress in treating pain is the complexity of nociception itself, which involves a complex array of signaling mechanism in both the periphery and central nervous system (CNS) (Basbaum et al. 2009; Woolf and Ma 2007). In addition, spinal neural circuits involved in detecting, transmitting, and modulating nociception are remarkably complex themselves (Arcourt and Lechner 2015), and it is especially challenging to selectively activate nociceptive vs. nonnociceptive inputs to these circuits (Li and Bak 1976). Therefore, there is a need to better understand the biology of nociception in terms of the basic mechanisms that mediate nociceptive signaling, how nociception is modulated, and how nociception modulates other behaviors. This argues for a more comparative approach, using a greater range of animal species that can provide insights into the fundamental mechanisms of nociception.

Studies using invertebrate model systems can make an invaluable contribution to this comparative approach since invertebrates exhibit many of the basic features of nociception observed in mammals. These include the use of dedicated nociceptive and nonnociceptive afferents and the capacity for nociceptive stimuli to induce sensitization that can either make responses to subsequent nociceptive stimuli more intense (resembling hyperalgesia) or cause nonnociceptive stimuli to elicit a nocifensive response (resembling allodynia) (Babcock et al. 2011; McMackin et al. 2016; Walters 1987b; Walters et al. 2001). Furthermore, there is considerable similarity between invertebrates and vertebrates in the basic mechanisms by which bioelectric signals are produced and transmitted synaptically in the nervous system, i.e., the types of ion channels, neurotransmitters, and receptors that mediate nervous system function (Marder 2007). Selected invertebrates also provide advantages for linking cellular and physiological processes to nociception-related behaviors that cannot be easily replicated in many mammalian studies through either precise molecular genetic approaches, such as in Drosophila or Caenorhabditis elegans (examples include Hapiak et al. 2013; Kim et al. 2012a; and Tracey et al. 2003), or through the use of semi-intact preparations where it is possible to monitor both neurophysiological and behavioral changes, such as in Aplysia or Hirudo (examples include Illich and Walters 1997 and Yuan and Burrell 2013b).

This review will provide a brief background on nociception in invertebrates but will focus on research questions where invertebrates may be able to make a significant impact. Two organizational notes are worth mentioning. First, this review will emphasize the use of the term “nociception” as defined above and not the term “pain,” which the IASP defines as “an unpleasant sensory and emotional experience associated with actual or potential tissue damage”. There has been an extensive discussion in recent years as to whether invertebrates and nonmammalian vertebrates experience an emotional analog to pain observed in mammals (Crook and Walters 2011; Sneddon 2015; Walters 2016), and this issue will be discussed in more detail toward the end of this review. Second, there have already been excellent reviews of nociception research carried out in the noted ecdysozoans models Drosophila and C. elegans (Im and Galko 2012; Tobin and Bargmann 2004) so this review will emphasize research on lophotrochozoans species such as Aplysia (a marine mollusk) and Hirudo verbana (the medicinal leech, an annelid).

Nociception in Invertebrates

Vertebrate nociceptive and nonnociceptive somatosensory signals are carried by separate pathways from the periphery to the CNS (Perl 2007). Most nonnociceptive input is carried by Aβ-fibers that enter the spinal cord and then ascend to higher centers (e.g., nuclei in the brain stem and the thalamus) primarily via the dorsal column (Gardner and Johnston 2013). Most nociceptive input is carried by Aδ- and C-fibers, which synapse on second-order neurons in the superficial lamina of the spinal cord that then project to targets in the midbrain, thalamus, and cerebrum via the spinothalamic tract and other ascending pathways (Willis and Westlund 1997).

Invertebrate nervous systems exhibit similar segregation of nociceptive and nonnociceptive input. In Drosophila, there are four classes of multidendritic sensory neurons whose cell bodies are located in the periphery. The Class I–III multidendritic neurons do not elicit nocifensive responses, although they may modulate this behavior, and appear to carry proprioceptive and touch/pressure input (Tsubouchi et al. 2012). Class IV neurons are mechanical and thermal nociceptive neurons; they express proteins that are selectively involved in nociception, such as TRPA and pickpocket [a degenerin/epithelial Na⁺ channel (DEG/ENaC)], and their activation elicits nocifensive responses in flies (Guo et al. 2014; Hwang et al. 2007; Tracey et al. 2003; Zhong et al. 2010). Studies in the nematode C. elegans have also identified distinctive classes of nonnociceptive (ALM, AVM and PLM), polymodal nociceptive (ASH), and mechano-only nociceptive (PVD) neurons (Chalfie et al. 1985; Kaplan and Horvitz 1993; Tobin and Bargmann 2004). As in Drosophila, these nociceptive neurons express proteins associated with the detection and transmission of noxious stimuli, such as DEG/ENaC, TRPA1, and transient receptor potential vanilloid (TRPV) (Chatzigeorgiou et al. 2010; Colbert et al. 1997; Tobin et al. 2002; Wittenburg and Baumeister 1999).

The CNS of H. verbana (the medicinal leech) possess three distinct types of somatosensory neurons: the rapidly adapting touch-sensitive neurons (T cells), the slow-adapting pressure-sensitive neurons (P cells), and the high-threshold nociceptive (N) cells (Blackshaw et al. 1982; Nicholls and Baylor 1968). Similar to mammalian nociceptive neurons, there are distinct mechano-only and polymodal N cells in Hirudo, with the latter being sensitive to high-threshold mechanical stimulation, as well as noxious thermal and chemical stimuli (Pastor et al. 1996; Summers et al. 2014). Pharmacological evidence suggests that the polymodal, but not mechanical, N cell possesses both TRPV and TRPA1 channels (Pastor et al. 1996; Summers et al. 2014; Summers et al. 2015), but this has not been confirmed using molecular genetic approaches.

Distinct nociceptive and nonnociceptive afferents have also been observed in mollusks, including Aplysia and the squid (Loligo) (Crook et al. 2013; Fischer et al. 2011; Frost et al. 1997; Illich and Walters 1997). In Aplysia there is clear evidence that the LE and VC sensory neurons, which are located in the CNS and have been used in many classic learning and memory experiments (Kandel et al. 2014), are mechanical nociceptive afferents that express sensorin, a neuropeptide necessary for more persistent serotonin-mediated long-term facilitation (Hu et al. 2004; Hu et al. 2011; Illich and Walters 1997; Walters et al. 1983; Walters et al. 2004; Walters and Cohen 1997). Responses to noxious thermal or chemical stimuli have not yet been addressed. Low-threshold (nonnociceptive) mechanosensory input appears to be mediated by afferents whose cell bodies are located in the periphery and do not express sensorin (Walters and Cohen 1997; Xin et al. 1995). Studies by Walters and colleagues have uncovered considerable details about how injury-induced sensitization increases peripheral excitability and changes in receptive field sensitivity and size in Aplysia nociceptive afferents (Billy and Walters 1989a, 1989; Walters 1987a, 1987b; Weragoda et al. 2004; Weragoda and Walters 2007). These changes in Aplysia nociceptors are specific to the site of injury and exhibit many similarities with primary hyperalgesia in mammals (Sandkühler 2009). Therefore, these findings may provide insights into the basic mechanisms responsible for both acute and persistent forms of injury-induced sensitization. Activation of the LE and VC nociceptors is also capable of producing sensitization in Aplysia that is not site-specific and may represent a basic mechanism of secondary hyperalgesia (Barbas et al. 2003; Billy and Walters 1989b; Trudeau and Castellucci 1993; Weragoda and Walters 2007). There is also evidence from Aplysia and Drosophila studies demonstrating that injury can elicit peripheral inflammatory responses mediated by either host immune cells (Aplysia) or apoptotic epidermal cells (Drosophila) that contribute to sensitized responses to nociceptive and nonnociceptive stimuli (Babcock et al. 2009; Clatworthy et al. 1994; Clatworthy and Grose 1999; Farr et al. 1999).

Modulatory Neurotransmitters and Nociception

In the spinal cord, vertebrate nociceptive circuits use a variety of neuropeptides that contribute to either central sensitization of nociceptive pathways (e.g., substance P and bradykinin) or diminish nociceptive signaling (e.g., opioids) (Woolf 2011). Invertebrate nervous systems possess comparable neuropeptides, potentially including members of the tachykinin family (e.g., substance P) and opioid-like peptides (Miller-Pérez et al. 2008; Mills et al. 2016; Nusbaum and Blitz 2012; Salzet and Stefano 1997; Siviter et al. 2000; Stefano et al. 1997). Two tachykinin receptors have been identified in Drosophila that exhibit significant homology to vertebrate tachykinin receptors and respond to substance P (Li et al. 1991; Monnier et al. 1992). A gene encoding a tachykinin prohormone has also been identified in Drosophila that exhibits homology to the mammalian tachykinin prohormone gene that produces substance P (Siviter et al 2000). Recent molecular genetic studies by Galko and colleagues have found that this Drosophila tachykinin signaling pathway contributes to injury-induced sensitization of thermal nociceptors (type IV afferents) (Im and Galko 2012). Much of the evidence for endogenous opioid systems and their antinociceptive function in invertebrates is based on pharmacological or immunocytochemical approaches and requires further validation (Miller- Pérez et al. 2008; Salzet and Stefano 1997; Stefano et al. 1997). However, recent studies have found evidence for genes related to opioids and their receptors in C. elegans that play a role in modulating nociception and feeding (Cheong et al. 2015; Mills et al. 2016). Surprisingly, no genes encoding opioids or opioid receptors have been reported in Drosophila. Neurotrophins are another category of neuropeptide implicated in both peripheral and central forms of nociceptive sensitization in mammals (Coull et al. 2005; Mannion et al. 1999; Zhao et al. 2006; Zhou et al. 2008) that may have a comparable function in invertebrates. Specifically, neurotrophin-like peptides and receptors have been found to contribute to persistent synaptic facilitation in the same synapses found to mediate behavioral sensitization in Aplysia (Kassabov et al. 2013; Pu et al. 2014).

Modulation of vertebrate nociceptive circuits is also mediated by neurons in brain stem nuclei that have projections down to the spinal cord (Heinricher et al. 2009). These descending modulatory inputs are stimulated by both ascending nociceptive afferents from the spinal cord and from higher integrative brain regions that communicate changes in behavioral state (Gauriau and Bernard 2002; Heinricher et al. 2009). The transmitters released by these descending neurons (e.g., glutamate, norepinephrine, and serotonin) act on the spinal nociceptive circuits and can have both pro- and antinociceptive effects (Heinricher et al. 2009). Invertebrate nervous systems do possess comparable modulatory circuits, but this is not often recognized by the pain research community. Although not involved in nociception per se, especially detailed examples of circuits that incorporate ascending input that then activates descending modulatory neurons utilizing neuropeptides and biogenic amines have been found in studies of sensorimotor networks in crustaceans (Blitz and Nusbaum 2011; Bucher and Marder 2013; Stein 2009). In some cases, these modulatory neurons are found in distinct areas of the brain or segmental ganglia and project to other regions of the CNS, similar to descending supraspinal modulatory neurons (Hörner 1999; Sinakevitch et al. 2005). In other cases the modulatory neurons are distributed throughout the CNS, such as the segmental distribution of serotonergic and octopaminergic neurons in the Hirudo CNS (Crisp et al. 2002; Gilchrist et al. 1995; Macagno et al. 1987; Stuart et al. 1974) or in the thoracic and abdominal ganglia of insects (Hörner 1999). Nevertheless, the main features observed in supraspinal modulatory circuits of vertebrates are also observed in invertebrates, that is, afferents stimulate activity in modulatory neurons (e.g., serotonergic neurons) that feed back to act on the afferents themselves or on other neurons involved in producing the animal’s response to further nociceptive stimuli (Burrell and Sahley 2001; Ehrlich et al. 1992; Hawkins et al. 1993; Lockery and Kristan 1991; Szczupak and Kristan 1995; Velázquez-Ulloa et al. 2003). These modulatory pathways in invertebrates are also activated by integrative elements that communicate changes in behavioral state, analogous to the influence that cortical and subcortical regions can exert on brain stem descending modulatory circuits in vertebrates (Gaudry and Kristan 2009). Therefore, although the anatomical details may differ, both vertebrates and invertebrates use a similar organization in which afferent inputs stimulate modulatory neurons that then feed back to influence the activity of nociceptive circuits.

Activity-Dependent Plasticity in Nociceptive Circuits

Long-term potentiation (LTP) is an activity-dependent form of synaptic strengthening observed throughout the CNS and is thought to contribute to central sensitization of nociceptive circuits (Sandkühler 2009). NMDA receptor (NMDAR)-mediated LTP has been observed in C-fiber synapses following their repeated activation and appears similar to LTP observed in other regions of the mammalian CNS (Ikeda et al. 2006; Liu and Sandkühler 1997). However, there has been a question as to whether “classic” (meaning similar to what is observed in the hippocampus) NMDAR-dependent LTP contributes to nociceptive sensitization (Latremoliere and Woolf 2009; Latremoliere and Woolf 2010, although see comment by Sandkühler 2010). The main issue of this debate is that stimuli that elicit LTP in activated C-fiber pathways (a homosynaptic change) also increase signaling by nonactivated nociceptive afferents (a heterosynaptic change) that manifests as sensitization away from the site of injury, i.e., secondary hyperalgesia or allodynia (Latremoliere and Woolf 2009; Sandkühler 2009). This suggests a lack of synapse specificity in the primary somatosensory afferents when compared with NMDAR-dependent LTP in the hippocampus.

Studies using invertebrates may help to elucidate functional principles behind sensitization at the site of injury vs. sensitization away from the site of injury. In invertebrates noxious stimuli can elicit both site-specific sensitization and nonspecific sensitization that may correspond to primary and secondary hyperalgesia, respectively (Walters 1987b). That both activated and nonactivated afferent pathways are strengthened following noxious stimuli reflects the involvement of multiple parallel modulatory processes. Some of these processes may represent true homosynaptic changes that share mechanistic properties with NMDAR-mediated LTP. LTP has been reported in nociceptive synapses in Aplysia (Lin and Glanzman 1994; Murphy and Glanzman 1999) and in Hirudo nociceptive (unpublished observation) and nonnociceptive synapses (Burrell and Sahley 2004; Grey and Burrell 2010; Li and Burrell 2011). In both species, LTP is NMDAR mediated and exhibits synapse specificity. These synapse-specific forms of LTP may mediate sensitization at the site of injury while other heterosynaptic forms of modulation contribute to sensitization both at the site of injury and away from the site of injury. These heterosynaptic modulatory processes are likely mediated by a diverse array of peptidergic and aminergic transmitters resulting in increases in synaptic transmission or increases in excitability in the primary afferent inputs, the postsynaptic targets of these afferents, or interneurons between the afferents and their ultimate postsynaptic targets (Burrell and Sahley 2005; Hawkins et al. 1993; Trudeau and Castellucci 1993; Walters 1987b; Weragoda et al. 2004; Weragoda and Walters 2007).

There is a potential novel mechanism by which NMDAR-mediated LTP may be able to contribute to both sensitization at the site of injury and away from the site of injury that has been observed in the hippocampus and may also occur in nociceptive circuits. This process involves depression of inhibitory synapses (disinhibition), which is known to contribute to injury-induced mechanisms of sensitization of nociceptive and nonnociceptive pathways in the spinal cord (Lu et al. 2013; Torsney and MacDermott 2006). In the hippocampus, cannabinoid receptor-mediated depression of GABAergic signaling can produce disinhibition of glutamatergic synapses, permitting widespread NMDAR-mediated LTP (Puighermanal et al. 2009). The synapse specificity of LTP was not changed, but rather the threshold for producing LTP has been lowered to an extent that modest levels of activity can result in synaptic strengthening throughout the hippocampus. One could envision a similar process occurring in the spinal cord and invertebrate CNS due to either endocannabinoid- or opioid-mediated disinhibition (Lau and Vaughan 2014; Pernía-Andrade et al. 2009; Wang and Burrell 2016). Furthermore, such a mechanism would provide a potential link between disinhibition of spinal afferent inputs due to injury and persistent facilitation of nociceptive circuit output leading to chronic pain. These questions are especially amenable to study in invertebrates given that they have well-characterized nervous systems that lend themselves to detailed electrophysiological examination.

A recent study by Sandkühler and colleagues has uncovered yet another potential mechanism linking NMDAR-mediated LTP to the heterosynaptic spread of potentiation in nociceptive circuits. Induction of LTP in C-fiber synapses was found to stimulate the release of cytokines and d-serine by astrocytes and/or microglia, which produced heterosynaptic potentiation in nonactivated nociceptive synapses (Kronschläger et al. 2016). These findings add to a growing body of literature indicating that glia are active contributors to nociceptive sensitization (Ji et al. 2013; Walters 2014). Glial cells are also found in the invertebrate CNS that have comparable roles to vertebrate astrocytes and microglia in terms of support, neurodevelopment, behavioral plasticity, and responses to neuroinjury (Dahl and Muller 2014; Duan et al. 2005; Singhvi et al. 2016; Stout et al. 2014; Zwarts et al. 2015), so this is yet another area of research where invertebrates may be able to contribute.

The argument that NMDAR-mediated LTP contributes to nociceptive sensitization does not exclude the presence of other activity-dependent processes that are NMDAR independent. Studies in Aplysia nociceptive synapses have long shown the presence of activity-dependent increases in both synaptic transmission and excitability (Walters and Byrne 1983; Walters and Byrne 1985) that may contribute to site-specific sensitization/primary hyperalgesia (Billy and Walters 1989a; Hawkins et al. 1983; Walters 1987a, 1987b). At the cellular level increased nociceptive signaling is mediated by a combination of homosynaptic Ca²⁺-dependent processes and heterosynaptic neuromodulation that is mediated by serotonin (Abrams et al. 1991; Chitwood et al. 2001; Sutton and Carew 2000).

Many readers will recognize that these homosynaptic and heterosynaptic forms of plasticity in nociceptive synapses are already known to contribute to learning and memory. The idea that cellular mechanisms of learning and memory would also contribute to modulation nociception, i.e., injury-induced forms of sensitization represent a “nociceptive memory,” has been noted by others (Price and Inyang 2015; Walters and Moroz 2009). Not only are learning-like cellular and physiological changes observed during injury-induced sensitization but processes such as nociceptive priming (which may be analogous to the learning and memory concept of savings) and reconsolidation may also be present (Costa et al. 2015; Huberdeau et al. 2015; Murre and Dros 2015; Price and Inyang 2015). Most of these discussions have involved comparisons of hyperalgesia or allodynia to sensitization-type learning. However, understanding how habituation, another form of nonassociative learning, might influence nociceptive memories also needs to be examined. This is because 1) the cellular mechanisms of habituation my provide insights into how to reverse chronic pain and 2) some forms of chronic pain are accompanied by deficits in habituation (Coppola et al. 2013; Peters et al. 1989). That the mechanisms of learning and memory should be so intimately associated with nociception is not surprising given, as stated at the beginning of this review, the adaptive importance for an animal to alter its behavior appropriately following nociceptive stimuli (Crook et al. 2014).

Endocannabinoids and TRPs

There is substantial interest in the antinociceptive potential of cannabinoid-acting drugs, both compounds based on phytocannabinoids in cannabis (e.g., 9-Δ-THC and cannabidiol) and those that boost endogenous cannabinoid (endocannabinoid) levels (Di Marzo 2009). Endocannabinoids are lipid neurotransmitters that are not stored in vesicles but are instead synthesized and released in an activity-dependent manner (Katona and Freund 2012). They are often synthesized and released in the postsynaptic neuron and function as retrograde transmitters. The first identified receptors were named cannabinoid receptors 1 and 2 (CB1 and CB2) and are both G protein-coupled receptors (Matsuda et al. 1990; Munro et al. 1993). Later, the transient receptor potential vanilloid 1 (TRPV1) channel was found to directly respond to endocannabinoids as well (Zygmunt et al. 1999; Zygmunt et al 2013). Both CB1- and TRP-mediated endocannabinoid signaling can elicit long-term depression (LTD) of synaptic transmission (Chávez et al. 2010; Edwards 2014; Gibson et al. 2008; Grueter et al. 2010; Katona and Freund 2012; Kim et al. 2012b).

Elements of the endocannabinoid signaling system are conserved throughout the animal kingdom (Elphick 2012). The two most common endocannabinoids in the mammalian brain, 2-arachydonylglycerol (2-AG) and anandamide, have also been found in the invertebrates Hirudo, C. elegans, Drosophila, and in Hydra (De Petrocellis et al. 1999; Matias et al. 2001; Lehtonen et al. 2008; Khaliullina et al. 2015). Furthermore, genes encoding the enzymes involved in endocannabinoid synthesis or metabolism have been found in Drosophila (Khaliullina et al. 2015; Leung et al. 2008), C. elegans (Pastuhov et al. 2012), and Hirudo (accession no. KU500007). Protostomal invertebrates lack orthologs of CB1 or CB2 receptors (Elphick 2012) but do possess a wide range of TRP channels (Vriens et al. 2004). There is evidence that, similar to vertebrates, invertebrate TRP channels respond to a variety of lipid-signaling molecules, including endocannabinoids (Kahn-Kirby et al. 2004; Leung et al. 2008; Yuan and Burrell 2013a). This would imply that central TRP channels fulfill a role as endocannabinoid receptors before the existence of CB1 or CB2 receptors, although this does not exclude the presence of other endocannabinoid receptors in invertebrates.

There is considerable untapped potential in using invertebrates to study how endocannabinoids may be modulating nociception. Endocannabinoid/TRPV-mediated LTD of synapses in mammals and Hirudo share a requirement for postsynaptic increases in intracellular Ca²⁺ and presynaptic activation of calcineurin (Edwards 2014; Yuan and Burrell 2012; Yuan and Burrell 2013a), suggesting an evolutionarily conserved mechanism of neuromodulation. Another advantage of studying endocannabinoid-mediated modulation of nociception in invertebrates is that, since they lack orthologs of CB1 and CB2 receptors, it is possible to isolate the effects of endocannabinoid action via TRP receptors without having to resort to pharmacological agents or genetic knockout of CB1/CB2 receptors. One question that invertebrate studies could address is whether there are other TRPs that respond to endocannabinoid transmitters and whether such signaling modulates nociception. In the Drosophila retina there is evidence of an interaction between the TRPL channel and DAG lipase, the enzyme responsible for 2-AG synthesis (Leung et al. 2008), but there has been no direct investigation of 2-AG’s interaction with this TRP channel or whether such signaling is found in other regions of the fly CNS.

Given the accessibility to key neurons and synapses in identified nociceptive circuits, plus the presence of relevant genes and molecular pathways, studies using invertebrates may provide useful information for the basic mechanisms by which endocannabinoids modulate nociceptive signaling. One example would be understanding the processes mediating the pro- vs. antinociceptive effects of these transmitters. Endocannabinoids are traditionally thought to be analgesic due, at least in part, to depression of glutamatergic transmission at primary afferent synapses (Kato et al. 2012; Kinsey et al. 2009; Liang et al. 2004; Morisset and Urban 2001; Suplita et al. 2006). However, endocannabinoid signaling has also been observed to enhance nociception, most likely because of depression of GABAergic/glycinergic inhibitory transmission resulting in disinhibition of spinal nociceptive circuits (Carey et al. 2016; Pernía-Andrade et al. 2009). Interestingly, injury-induced allodynia resulting from TRPV1-mediated disinhibition in the spinal cord has been observed, but it is not known what is activating TRPV1 (Kim et al. 2012b). This capacity of endocannabinoids to have both pro- and antinociceptive effects may compromise clinical studies attempting to ascertain the effectiveness of cannabinoid-based therapies to treat chronic pain (Christie and Mallet 2009; Huggins et al. 2012; Naef et al. 2003). One explanation for these opposing effects of endocannabinoids is that the analgesic effects are due to depression of nociceptive synapses, whereas the pronociceptive effects are due to disinhibition of nonnociceptive inputs, allowing these afferents to have access to the nociceptive circuits that are normally suppressed. In pharmacological studies in Hirudo, endocannabinoid/TRPV signaling depressed nociceptive synapses but potentiated nonnociceptive synapses via a disinhibitory mechanism (Higgins et al. 2013; Wang and Burrell 2016; Yuan and Burrell 2010; Yuan and Burrell 2012; Yuan and Burrell 2013a). Whether such effects are observed in other species, their functional relevance in terms of regulating nociceptive signaling, and how these opposing effects may be selectively activated can be addressed using Hirudo and other invertebrates. Such comparative studies provide a powerful approach for answering a fundamentally important question about cannabinoid modulation of nociception.

Inhibition of Nociceptive Circuits and the Role of Chloride Gradients

Inhibitory synapses, mediated by the neurotransmitters GABA and glycine, play a critical role in nociception by controlling the “gate” that allows nociceptive signals to reach the brain (Melzack and Wall 1965). These inhibitory synapses also control access by nonnociceptive afferents to spinal cord nociceptive circuits, and decreases in inhibitory signaling (disinhibition) can contribute to allodynia or hyperalgesia (Kim et al. 2012b; Lu et al. 2013; Torsney and MacDermott 2006;). GABA is also found as an inhibitory transmitter in arthropods, nematodes, mollusks, and annelids (Buckingham et al. 2005; Cline 1986; Darlison et al. 1993; Florey 1991; Schuske et al. 2004;). Furthermore, there is evidence that disinhibition can alter synaptic signaling by nociceptive and nonnociceptive afferents to produce sensitization in Aplysia and Hirudo (Trudeau and Castellucci 1993; Wang et al. 2015).

Inhibition mediated by ionotropic GABA or glycine receptors requires the influx of Cl⁻, and therefore the regulation of intracellular Cl⁻ concentrations plays a critical role in the effects of these receptors. In most neurons the concentration of intracellular Cl⁻ is lower compared with extracellular levels so that activation of GABA or glycine ionotropic receptors permits Cl⁻ influx, resulting in hyperpolarization of the membrane potential and inhibition (inhibition can also be accomplished through membrane current shunts) (Kaila et al. 2014). However, Cl⁻ gradients in spinal cord afferents appear to be reversed (high inside relative to outside) so that GABA/glycine receptor activation leads to Cl⁻ efflux and membrane depolarization. Complicating matters further, this Cl⁻-mediated depolarization can be excitatory or inhibitory. The excitatory effect is due to activation of voltage-gated cation channels, whereas the inhibitory effect is due to either Na⁺ channel inactivation or shunting inhibition (see Doyon et al. 2016 for an excellent explanation of shunting inhibition). Intracellular Cl⁻ levels are regulated by a Cl⁻ exporter, K⁺-Cl⁻ cotransporter 2 (KCC2), and by a Cl⁻ importer, Na⁺-K⁺-Cl⁻ cotransporter (NKCC1) (Kaila et al. 2014).

Nociceptive sensitization appears to alter Cl⁻ gradients in both primary afferents and projection neurons in the vertebrate spinal cord. Sensitization elicited by peripheral inflammation is blocked by drugs that inhibit activity of the Cl⁻ importer NKCC1 (Granados-Soto et al. 2005; Pitcher and Cervero 2010). In neuropathic pain models, nociceptive projection neurons undergo a shift in their Cl⁻ gradient caused by an increase in expression of NKCC1 and concomitant increase in intracellular Cl⁻ levels (Coull et al. 2003; Coull et al. 2005). Consequently, these cells transition from being hyperpolarized and inhibited by GABA or glycine to being depolarized (and presumably excited) by these transmitters.

There are questions in the mammalian literature regarding whether both nociceptive and nonnociceptive afferents have elevated intracellular Cl⁻ levels, what cotransporters are involved in regulating Cl⁻ gradients, and the functional effect of depolarizing Cl⁻ conductances (Duchen 1986; Gilbert et al. 2007; Mao et al. 2012; Price et al. 2006; Sung et al. 2000; Wang et al. 2015; Wei et al. 2013; Willis 1999). These are questions that studies using invertebrates could potentially address, taking advantage of the improved ability to selectively record from and manipulate nociceptive and nonnociceptive afferents. Examples of GABA-induced depolarization in invertebrates have been observed in both afferents and the targets of afferent neurons (Burrows 1996; Cattaert et al. 1999; Cheung et al. 2006; Norekian and Malyshev 2005; Pfeiffer et al. 2009; Sargent et al. 1977). In Hirudo, nociceptive neurons have a Cl⁻ equilibrium potential that is above the resting potential and are depolarized by GABA, whereas nonnociceptive pressure-sensitive (P) cells have a Cl⁻ equilibrium potential that is below the resting potential and are hyperpolarized by GABA (Wang et al. 2015). A similar difference in Cl⁻ gradient between nociceptive and nonnociceptive afferents has been reported in some mammalian studies (Gilbert et al. 2007; Mao et al. 2012).

A consequence of these differences in Cl⁻ gradients between nociceptive and nonnociceptive afferents is that decreases in GABAergic signaling produce opposing effects. Hirudo P cell synapses are disinhibited by depression of GABA signaling, but N cell synapses are actually depressed by reductions in tonic GABAergic input (Wang et al. 2015). These distinctions in Cl⁻ gradients may explain, at least in part, the pro- and antinociceptive effects of cannabinoid-based treatments mentioned earlier. Studies in Hirudo have found that endocannabinoids enhance nonnociceptive (P cell) synaptic transmission (even though P cells lack TRPV channels) due to decreases in GABAergic inhibition of P cell synapses (disinhibition) (Higgins et al. 2013; Summers et al. 2014; Wang and Burrell 2016). On the other hand, polymodal nociceptors do possess TRPV channels, and endocannabinoids directly depress these synapses in a TRPV-dependent manner (Yuan and Burrell 2010; Yuan and Burrell 2013a). Although these nociceptive synapses receive GABAergic input, it is depolarizing, and therefore the nociceptive synapses do not experience endocannabinoid-mediated disinhibition, i.e., they are “protected” from disinhibition as a result of increased intracellular Cl⁻ concentrations (Wang et al. 2015; Wang and Burrell 2016). An additional consequence of depolarizing Cl⁻ conductances is that it opens up a potential novel mechanism of GABAergic modulation. Whereas polymodal nociceptors in Hirudo possess TRPV channels that can be directly modulated by endocannabinoids, mechanical nociceptors lack TRPV channels but are nevertheless depressed by these endocannabinoids (Summers et al. 2014; Wang and Burrell 2016). This endocannabinoid-mediated depression of mechanical nociceptor synapses is due to decreases in excitatory GABAergic signaling (Wang and Burrell 2016). Similar to how such decreases in GABAergic input disinhibit nonnociceptive (P cell) synapses, the mechanical nociceptor synapses are “disexcited” by depression of this excitatory, and presumably tonic, GABAergic input (Wang and Burrell 2016). At least one mammalian study has reported similar effects in which inhibition of GABA receptors increased transmission by nonnociceptive synapses (presumably a disinhibitory effect) but depressed nociceptive synapses (possibly a disexcitation effect) (Melin et al. 2013).

A second related issue that could be addressed using invertebrates is what levels of Cl⁻-mediated depolarization produce excitation (due to activating voltage-gated cation channels) vs. inhibition (due to current shunt or Na⁺ channel inactivation) (Doyon et al. 2016). In primary afferent circuits, the assumption has been that, as the magnitude of the Cl⁻-mediated depolarization increases, the effect shifts from being inhibitory to excitatory (Duchen 1986; Willis 1999). However, this does not appear to have been directly measured in the spinal cord, and studies in the brain indicate that low levels of GABA-mediated depolarization are excitatory, with higher levels becoming inhibitory (Song et al. 2011). It should be noted that this study monitored tonic GABA conductances, and the issue of examining the contribution of tonic vs. phasic GABAergic signaling, whether excitatory or inhibitory, is yet another issue that could be addressed in invertebrate studies of nociception.

The Ethological Relevance of Nociception

As stated in the Introduction, there is considerable concern about the lack of progress in developing novel antinociceptive therapies that can treat a wide range of pain conditions. One facet of this problem is that animal studies have relied heavily on reflexive behaviors as a measure of changes in the nociceptive circuitry, and these may not capture the complexity of chronic pain conditions nor the efficacy of potential therapies (Mogil 2009). Furthermore, the stimuli used to elicit these nocifensive responses, while experimentally advantageous and reproducible, may not be ethologically relevant (Williams 2016). A greater understanding of how and why the nervous system is modulated to be hypersensitive during recovery from injury (e.g., hyperalgesia or allodynia) or hyposensitive at times when it is critical to be able to escape from the source of injury (e.g. stress-induced analgesia) would improve our understanding of the basic mechanisms of nociception (Williams 2016).

Recent studies using the squid by Crook, Walters, and colleagues are excellent examples of studying nociceptive sensitization in an ethologically relevant context. Squid with injuries to one arm are sensitized to subsequent mechanical and visual stimuli (Crook et al. 2011). This injury-induced sensitization did not alter food-seeking behavior but enhanced defensive responses to natural predators, e.g., exhibiting alert behavior when predators were farther away and then more readily initiated subsequent escape responses (Crook et al. 2011; Crook et al. 2014). Transient anesthetic treatment during the injury prevented subsequent behavioral sensitization and sensitization of nociceptive neurons (Crook et al. 2013; Crook et al. 2014). However, this treatment had a cost in that injured animals that had been anesthetized during the injury later had a lower probability of surviving an encounter with a predator than did an injured unanesthetized squid (Crook et al. 2014). A similar consideration of ethological relevance also informed studies addressing why Drosophila larvae undergo a characteristic corkscrew locomotory response when exposed to stimuli that activated nociceptive afferents, a behavior that would not seem to be effective in moving away from noxious stimuli (Hwang et al. 2007). Hwang et al. demonstrated that this behavior is effective in protecting Drosophila larvae from parasitic wasps that are a routine threat in the larvae’s natural environment. Studies in Manduca sexta caterpillars have also examined nociceptive sensitization via stimuli that mimics a predator’s attack (in this case, birds) (McMackin et al. 2016; Walters et al. 2001). This nociceptive sensitization has been found to require activation of NMDARs and hyperpolarization cyclic nucleotide-gated (HCN) channels in the CNS (Tabuena et al. 2016). In the studies of Aplysia behavioral sensitization and its neural correlates discussed earlier, the observed modulatory effects were produced by artificial stimuli such as electric shock or exogenous application of serotonin. Recent experiments have demonstrated that these sensitization-induced behavioral and physiological changes in Aplysia can also be produced following an attack by a lobster, a natural predator of this mollusk (Mason et al. 2014; Watkins et al. 2010). Finally, Elwood and colleagues have examined whether crustaceans can learn to avoid noxious electric shock stimuli and whether such noxious stimuli can produce behavioral changes related to how these animals make decisions regarding resource quality (Appel and Elwood 2009a, 2009b; Magee and Elwood 2013).

In addition to illustrating the potential evolutionary significance of injury-induced sensitization, this study demonstrates how using ethologically relevant behavioral measures can provide a more nuanced and informative “readout” of nociceptive modulation. Invertebrates lend themselves to this approach of using ethologically relevant behaviors in a controlled experimental setting as demonstrated by experiments using the squid described in the previous paragraph. Additional examples that could be used in the study of nociception include responses to visual threat in Drosophila (Gibson et al. 2015), tests of “anxiety-like” behavior in crayfish (Fossat et al. 2014), aversive conditioning in crabs (Magee and Elwood 2013), and prey localization behavior in Hirudo (Harley et al. 2011).

A significant issue that these studies are beginning to raise is whether invertebrates (and nonmammalian vertebrates) experience an emotional response comparable to how humans experience pain (Crook and Walters 2011; Sneddon 2015; Walters 2016). Establishing what constitutes emotion-like behavior, identifying the potential physiological substrates involved, and even developing the terminology for how one refers to such behaviors is a challenge in both invertebrates and vertebrates (Anderson and Adolphs 2014; LeDoux 2012). Nevertheless, these are important issues from both a scientific and ethical perspective (Crook and Walters 2011; Sneddon 2015). One approach to this issue is provided by Anderson and colleagues who use the term “emotion primitives” to describe behaviors that are evolutionarily conserved across phyla (Anderson and Adolphs 2014; Gibson et al. 2015). As with emotional behaviors ascribed to humans, emotion primitives are observable behaviors or changes in behavior that represent information about the internal state of the animal. Findings from Crook et al. (2014) suggest that injured squid demonstrate a persistent increase in avoidance behaviors to predators (Crook et al. 2014), and this may indicate an emotion primitive corresponding to anxiety since the response was not specific to the context or the actual source of the injury.

Addressing the ethological significance/evolutionary relevance of nociceptive modulation should not be limited to sensitizing processes but also to processes that attenuate nociception. Stress-induced analgesia and diffuse nociceptive inhibitory control (DNIC) are two modulatory processes in mammals that reduce responses to nociceptive stimuli immediately following exposure to a stressful stimulus or to an actual nociceptive stimulus, respectively (Butler and Finn 2009). It has been hypothesized that both of these processes preserve an animal’s response in the face of dangerous conditions, such as an attack by a predator, so that an escape or defensive behavior can be carried out. Little or no research has been conducted on stress-induced analgesia or DNIC in invertebrates despite the fact that 1) these are likely evolutionarily conserved processes and 2) it may be possible to use more ethologically relevant stressors in invertebrate studies compared with the electric shocks or constraint stressors used in most mammalian studies.

Another form of antinociceptive modulation is gate control, which refers to a decrease in nociceptive signaling due to repetitive activation of nonnociceptive Aβ-fiber afferents in mammals (Melzack and Wall 1965). The principle of gate control has been used in both transcutaneous nerve stimulation (TENS) and spinal cord stimulation (SCS) therapies to control pain. However, it has become clear that gate control of nociception involves more than sensory gating and that repetitive nonnociceptive stimulation elicits additional modulatory processes that persist even when Aβ activity has ceased (Meyerson and Linderoth 2000; Sandkühler 2000; Schechtmann et al. 2008; Song et al. 2009; Vance et al. 2014). In Hirudo, repetitive activation of touch afferents (T cells) decreases both synaptic transmission by nociceptive afferents (N cells) and the magnitude of a withdrawal reflex elicited by N cell stimulation (Yuan and Burrell 2010; Yuan and Burrell 2013b). Both the synaptic and behavioral effects are due to endocannabinoid modulation, which may explain, at least in part, how the attenuating effects of repetitive nonnociceptive afferent activity can persist once stimulation has ended. Similar findings have also been observed in the rat in which repetitive stimulation of Aβ-fibers elicits a long-term depression of C-fiber synapses that is mediated by endocannabinoid signaling (Yang et al. 2016). This pattern of Aβ stimulation also reduced behavioral sensitization in nerve-injured animals, and this effect is endocannabinoid dependent as well. Although this is a nice example of similar processes of nociceptive modulation being observed at multiple phylogenetic levels, it would be fair to ask what is the functional or evolutionary relevance of this form of modulation. That is, under what ethologically relevant conditions do animals encounter such repetitive nonnociceptive stimuli, and why should such stimulus impact nociceptive stimulus-response pathways? One possibility is that the stimuli that produce “gate control” are also eliciting a form of generalization or transfer of habituation (Poon and Young 2006) from nonnociceptive pathways to nociceptive pathways. Applying the concepts of habituation to TENS and SCS may provide an avenue to increase the efficacy of these therapies.

Perspectives and Significance

Invertebrates provide an opportunity to uncover basic mechanisms of nociceptive signaling and its modulation due to the ability to selectively manipulate nociceptive vs. nonnociceptive afferents and their synapses through either molecular genetic or physiological approaches. Furthermore, the capacity to examine the adaptive significance of injury-induced sensitization in invertebrates may uncover new approaches for how to treat chronic pain (Williams 2016). With efforts to develop full genome sequencing and annotation in more invertebrate species and the arrival of new molecular genetic tools such as CRISPR/Cas9 gene editing, the utility of invertebrate nociception studies will only increase. These advantages would complement the pain/nociceptive research already being conducted in mammals and contribute to the shared goal of improving the treatment of pain in humans. However, it is not clear that the pain research community is paying sufficient attention to invertebrate studies. In a Medline search of invertebrate articles on nociception (“invertebrate OR Drosophila OR C. elegans OR mollusk OR Hirudo” and “pain OR nocicept*”) over the last 10 years, only 2 out of the 276 articles appeared in journals with the word pain in their title, with the rest appearing in more general neuroscience journals. This is unfortunate given that such comparative approaches have been repeatedly validated in neuroscience. Topics such as learning and memory, neurodevelopment, and central pattern generator function have seen considerable advancement in large part because of research spanning invertebrate and vertebrate species (Burrell and Sahley 2001; De Schutter et al. 2005; Dickinson 2006; Hawkins et al. 2006; Kandel et al. 2014; Munno and Syed 2003; Yamaguchi and Miura 2015). Re-engaging this comparative approach is critical for understanding the fundamental processes of pain and using that knowledge to improve treatments.

GRANTS

This work was supported by Centers of Biomedical Research Excellence (COBRE) Grant P20-RR-015567, National Science Foundation Grant IOS-1051734, National Institute of Neurological Disorders and Stroke R01NS092716, and by the University of South Dakota Division of Basic Biomedical Sciences.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.D.B. drafted manuscript; B.D.B. edited and revised manuscript; B.D.B. approved final version of manuscript.