Nociceptive primary afferents: they have a mind of their own

Abstract

Nociceptive primary afferents have three surprising properties: they are highly complex in their expression of neurotransmitters and receptors and most probably participate in autocrine and paracrine interactions; they are capable of exerting tonic and activity-dependent inhibitory control over incoming nociceptive input; they can generate signals in the form of dorsal root reflexes that are transmitted antidromically out to the periphery and these signals can result in neurogenic inflammation in the innervated tissue. Thus, nociceptive primary afferents are highly complicated structures, capable of modifying input before it is ever transmitted to the central nervous system and capable of altering the tissue they innervate.

In the early years of studying sensory processing, the peripheral afferent was thought to be no more than a pipeline or a conduit that faithfully transported sensory information from the stimulated region to conscious levels. This belief is most clearly demonstrated in the drawing by Descartes published in the 1680s, where he illustrated a boy experiencing burning pain as a result of his toes coming in contact with fire (see Fig.2 of Roper, 2014, Introduction, this issue). An uninterrupted line is drawn from the toes to the brain, suggesting there is no modification of the fiery stimulus at any point along the stimulus trajectory. Three hundred plus years of research later we know that this is not the case. On the contrary, the study of nociceptive primary afferents has demonstrated that these fibres have many surprising properties, three of which will be discussed in this review. First, although nociceptive terminals appear simple and uncomplicated (Fig.1), immunohistochemical studies have demonstrated that nociceptors are very complex in their expression of ligands, neurotransmitters and receptors. This allows for nociceptors to have autocrine and paracrine interactions. Second, as a result of this complexity, they are able to modify input before it reaches the central nervous system (CNS), including inhibition of input so that signals are dampened before ever leaving the peripheral terminal. Finally, these fibres can generate outgoing signals, termed dorsal root reflexes (DRRs), which alter the peripheral tissues they innervate. This antidromic activity contributes to disease states.

Figure 2. Schematic drawing of two nociceptors with their terminals expanded at the level of the skin.

The dorsal root ganglia (DRG) and central processes of these fibres (terminating in the spinal cord dorsal horn) are also sketched. Nociceptors express a large variety of receptors and channels and they contain numerous ligands, many of which they release. Receptor activation in the periphery occurs through volume transmission. This allows for autocrine (1) and paracrine (2) regulation of their activity. Ligands with non-neural sources are also listed. 5HT, serotonin; ACh, acetylcholine, ATP, adenosine triphosphate, Angio II, angiotensin II; Bomb, bombesin; BK, bradykinin; CCK, cholecystokinin; CGRP, calcitonin gene-related peptide; HA, histamine; iGluR, ionotropic glutamate receptor; mGluR, metabotropic glutamate receptors; mrgprd, MAS-related G protein-coupled receptor; NK1, neurokinin 1; NPY, neuropeptide Y; PGE, prostaglandin E; SP, substance P; TRPs, transient receptor potential receptors; TRKs, tyrosine kinase receptors. (Reproduced from Carlton & Coggeshall, 1998, with permission.)

Figure 1. Immunostained nociceptive primary afferents visualized as they penetrate the epidermis.

Modified from Zylka et al. 2005, with permission.

Complexity of nociceptive primary afferents

In the late 1990s, Richard Coggeshall and I wrote two review articles that focused on neurotransmitter and receptor localizations in primary afferents. One article surveyed primary afferents in general (Coggeshall & Carlton, 1997) and the second focused on nociceptors in particular (Carlton & Coggeshall, 1998). As a result of these endeavours, it became very clear that integration of information at the level of the peripheral terminal was a real possibility and the oversimplified view of the nociceptor as just a conduit for noxious information was a gross misrepresentation. To date numerous ligands/neurotransmitters have been localized in nociceptive primary afferents (Fig.2). Add to this list growth factors (nerve growth factor, brain-derived neurotrophic factor, glial-derived neurotrophic factor) and hormones (prolactin, somatostatin), and the capability of peripheral nociceptors to influence their microenvironment becomes quite impressive. Add to this an array of receptors that are expressed on these terminals and one can begin to imagine that each terminal resembles a tiny brain! The complexity of nociceptors has led us to think of them as having a ‘mind of their own’.

There are no classic synapses in relation to cutaneous nociceptors (pre- and postsynaptic elements that communicate via a synaptic specialization within a synaptic cleft). Therefore, the best way to describe the type of transmission that occurs in relation to nociceptors in the periphery is volume transmission (Zoli et al. 1998; Fig.2). This is when a ligand is released and through 3-dimensional diffusion in the extracellular space reaches its receptor. Travelling in this manner, a ligand is able not only to reach a larger number of targets but also can reach targets at an increased distance. Consistent with volume transmission, several lines of evidence strongly suggest that autocrine and paracine interactions can occur at the level of the nociceptor (Fig.2). This is exemplified by the fact that nociceptors can release glutamate (De Groot et al. 2000; Jin et al. 2009) and several receptors in the glutamate family (ionotropic and metabotropic) are expressed by both nociceptive and non-nociceptive peripheral terminals (for review see Carlton, 2001; Neugebauer & Carlton, 2002). Thus, there is compelling evidence that this compartment of the nervous system, once thought to be simple and straightforward in its processing capabilities, is capable of integrating input before it ever reaches the first synapse in the CNS.

Nociceptors modify incoming input prior to transmission to the CNS

There are now several lines of evidence that nociceptive primary afferents can modify signals before they are transmitted centrally. This ability can change the way we perceive the outside world. There are comprehensive reviews written on nociceptor sensitization, processes through which nociceptors can amplify signals (Gold & Gebhart, 2010). In this review, I will discuss the concept of nociceptor-initiated inhibition: processes through which nociceptors can inhibit signals. Two receptor families known to have inhibitory control over nociceptor activity will be discussed, each using a different mechanism of action. These phenomena take the form of tonic inhibitory control and activity-dependent inhibitory control.

Somatostatin (SST)

Subsets of dorsal root ganglion (DRG) cells express somatostatin peptide (Hökfelt et al. 1975) and SST receptors (SST1, 2A, 2B, 3 and 4; Carlton et al. 2001a, 2004; Bär et al. 2004). Nociceptors in particular express SST receptor 2A (Carlton et al. 2001a, 2004). Several lines of evidence demonstrate that SST agonists applied in the periphery will attenuate formalin or capsaicin (CAP)-induced nociceptive behaviours and block bradykinin-induced nociceptor sensitization (Carlton et al. 2001a, 2004; Ji et al. 2006). The surprising finding is that SST receptors maintain a tonic inhibitory control over nociceptors (Carlton et al. 2001b). In naive rats, blockade of these receptors on nociceptors by intraplantar injection of either the SST receptor antagonist cyclo-somatostatin (c-SOM) or a SST receptor antibody, generates nociceptive behaviours such as flinching and lifting/licking. In rats in acute pain (formalin), injection of these same drugs produces enhancement of the formalin-induced nociceptive behaviours. These actions are blocked when c-SOM is co-applied with three different SST agonists (SRIF-14, octreotide, vapreotide). Tonic inhibitory control is confirmed at the single fibre level using a glabrous skin–nerve preparation. Application of c-SOM to the receptive field of identified nociceptors innervating the skin results in the generation of activity that is blocked by the addition of the SST agonists. Additional studies demonstrate that SST receptors maintain tonic inhibitory control over transient receptor potential vanilloid 1 (TRPV1) receptors expressed on cutaneous nociceptors (Carlton et al. 2004). Blockade of peripheral SST receptors with c-SOM dramatically enhances TRPV1 agonist CAP-induced pain behaviours and nociceptor activity. SST receptors are the first receptors shown to have a tonic inhibitory control over nociceptor processing in the periphery.

Metabotropic glutamate receptors (mGluRs)

Our lab has a long history of studying glutamate receptors on nociceptors in the periphery (Carlton et al. 1995, 2009b; Zhou et al. 1996; Davidson et al. 1997; Carlton, 2001; Du et al. 2001, 2003, 2006, 2008; Neugebauer & Carlton, 2002; Govea et al. 2012). We have been most interested in the group II and III mGluRs because when activated, they produce neuronal depression/inhibition (Schoepp & Conn, 1993; Conn & Pin, 1997), actions that would be important in controlling pain transmission from the periphery to the CNS. When antibodies and pharmacological tools became available to investigate the presence and action of mGluRs, another exciting possibility for modulation of painful input by peripheral nociceptors was elucidated. Anatomical studies demonstrate that a significant percentage of DRG cells express either group II, III or both receptors. Importantly, 93% and 71% of DRG cells expressing TRPV1 express group II (Carlton et al. 2001b) or group III (Govea et al. 2012), respectively. Thus, the anatomical substrate is present supporting an action of the two receptors on the same peripheral nociceptor.

Do these mGluRs maintain a tonic inhibitory control, similar to what is seen with peripheral SST receptors? In contrast to what is observed following injection of an SST receptor antagonist, a group II/III antagonist LY341495 (LY, 100 μm) injected alone into the hindpaw does not generate nociceptive behaviours (Du et al. 2008). Thus, there is no tonic inhibitory control exhibited by group II/III mGluRs. This lack of effect of LY alone and lack of tonic control is confirmed in DRG cells using calcium imaging and in nociceptors using single fibre recordings (Carlton et al. 2011). However, several lines of evidence demonstrate that groups II and III maintain an activity-dependent inhibitory control over nociceptors (Carlton et al. 2011). Intraplantar injection of LY significantly enhances CAP-induced nociceptive behaviours (Fig.3). This is confirmed at the single fibre level where CAP-induced activity is enhanced not only by LY but also by (RS)-1-amino-5-phosphonoindan-1-carboxylic acid (APICA), a selective group II antagonist and by 2R,4R-4-aminopyrrolidine-2,4-dicarboxyate (UBP), a selective group III antagonist. In a dose-dependent fashion, LY can induce calcium mobilization in DRG cells when applied with CAP. The actions of LY on CAP-induced responses are attenuated by either group II or III selective agonists (Carlton et al. 2011).

Figure 3. Behavioural data demonstrating group II/III involvement in activity-dependent inhibition of TRPV1 receptors.

Intraplantar capsaicin (CAP) alone results in flinching (A and B) and Lift/Lick (C and D) behaviour. This behaviour is enhanced when CAP is injected with LY341495 (LY), a group II/III antagonist. CAP in one hindpaw and LY in the other results in behaviour that is no different from CAP alone, confirming that LY does not become systemic but is having a local effect. (Reproduced from Carlton et al. 2011, with permission.)

A series of experiments demonstrates that excess glutamate plays a role in inducing activity-dependent inhibition by group II/III mGluRs. Sensitivity to heat does not develop following injection of glutamate alone (300 μm), but injection of this concentration plus LY (to prevent group II/III mGluR activation) produces a robust and prolonged sensitivity to heat evidenced by a significant lowering of the paw withdrawal latency to a heat stimulus (Carlton et al. 2011). A similar result is observed at the single fibre level where 1 mm glutamate alone does not change the discharge rate or the unit response to heat, but in the presence of LY there is a 4-fold increase in the glutamate-induced discharge rate and the threshold to activation is lower. The experiments described above use exogenous glutamate. To determine if endogenous glutamate release has functional relevance in relation to mGluR activation, a protocol is used that causes endogenous release of glutamate, namely formalin injection (Omote et al. 1998). Formalin (2%) is injected alone or with LY. There is a 50% increase in formalin-induced nociceptive behaviours when it is accompanied by LY. This increase is prevented when a group II agonist is added. The data infer that release of endogenous glutamate plays a pivotal role in engaging group II/III inhibition, which dampens formalin-induced pain behaviours. Sources of endogenous glutamate include first and foremost the primary afferents themselves (Westlund et al. 1989; Jeftinija et al. 1991; Omote et al. 1998; De Groot et al. 2000; Keast & Stephensen, 2000), keratinocytes (Genever et al. 1999) and blood serum (Erdo, 1991). Based on this series of experiments we conclude that group II/III mGluRs do not influence nociceptive afferents when at rest and they do not modulate responses following brief activation (i.e. 10 s heat pulse). The mGluR inhibitory influence becomes apparent after high frequency and/or prolonged stimulation (as occurs in response to algogenic substances like CAP or formalin; Carlton et al. 2011).

The data are compelling that group II/III mGluRs function as built-in negative modulators of peripheral nociceptor activity. They have little or no role under basal, quiescent conditions, or when nociceptors respond to brief stimuli. However, the mGluRs clearly regulate nociceptors undergoing vigorous excitation. Endogenous inhibitory modulation of TRPV1 function is undoubtedly important given its critical role in pain transmission (Tominaga et al. 1998; Caterina & Julius, 2001). Our studies show that if mGluR activation is prevented, then prolonged enhancement of TRPV1 function occurs. Without the activity-dependent modulation of group II/III mGluRs, the TRPV1-induced peripheral sensitization is prolonged and the nociceptive signal transmitted to the cord or brainstem is amplified. The augmented signals would impact dorsal horn neurons and most likely promote central sensitization. This use-dependent modification by mGluRs is important, serving to prevent ‘runaway’ excitation of nociceptors. If the system fails, this could enhance both peripheral and/or central sensitization that is a hallmark of many chronic pain conditions. Thus, there is great potential for primary afferent nociceptors to protect us from exaggerated pain transmission that can produce, or transform into, chronic pain.

Generation of outgoing signals: dorsal root reflexes

Primary afferents can function as a two-way street. In addition to modifying input before it is transmitted to the CNS, they can also generate action potentials at their central terminals which are then conducted to their peripheral terminals, causing release of neurotransmitters. Thus, they can transmit information into the CNS and carry input out of the CNS (Gotch & Horsley, 1891; Barron & Matthews, 1935). Because this activity was originally recorded in dorsal roots that were disconnected from the periphery, the phenomenon was termed dorsal root reflexes (DRRs; Barron & Matthews, 1935). While recording from dorsal roots is the ‘cleanest’ way to isolate this antidromic activity and be confident of its origin, it can also be recorded distal to DRGs in nerve branches (Toennies, 1938; Sluka et al. 1995a; Rees et al. 1996) with the caveat that proper controls must be done to confirm the origin and the fibre type carrying the output. Originally all sizes of myelinated axons were reported to carry DRRs, but not C fibres (Toennies, 1938; Lisney, 1979). However, more recently, C fibres have been shown to carry DRRs as well (Lin et al. 2000; Weng & Dougherty, 2005).

The DRR circuitry that has been proposed involves the same circuitry underlying primary afferent depolarization (PAD) (Willis, 1999) and involves the release of γ-aminobutyric acid (GABA) from dorsal horn interneurons. This activates GABA_A receptors on presynaptic terminals of primary afferents, resulting in chloride ions (Cl^–) leaving the terminal. This depolarizes the terminal, bringing the membrane potential closer to the threshold for generation of an action potential. If enough Cl^– leaves the terminal and the membrane reaches threshold, an action potential is generated and propagated antidromically out to the periphery. Why does activation of GABA_A receptors lead to excitation here but to inhibition elsewhere in the CNS? The answer lies in the expression by primary afferents of a particular transporter: the Na⁺,K⁺,2Cl^– or NKCC cotransporter (Alvarez-Leefmans et al. 1988; Rocha-Gonzalez et al. 2008). This cotransporter allows afferent terminals to sequester Cl^– ions such that the concentration of Cl^– becomes greater inside the terminal compared to outside (Alvarez-Leefmans et al. 1998; Sung et al. 2000). When GABA_A receptors located on primary afferent terminals are activated, Cl^– flows out of the terminal and the membrane equilibrium potential moves toward the threshold potential. This proposed circuitry is depicted in Fig.4 (Alvarez-Leefmans et al. 1998). Non-NMDA receptors have also been shown to play a role in DRRs; it has been hypothesized that the first synapse in the spinal cord, where incoming input gains access to DRR circuitry, involves non-NMDA receptors since 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), a non-NMDA antagonist, but not dl-2-amino-7-phosphonoheptanoic acid (AP7), an NMDA antagonist, can block DRR activity (Sluka & Westlund, 1993; Sluka et al. 1995b). Additions have been made to this hypothesized circuitry which explain the phenomena of touch-evoked allodynia following peripheral injury (Cervero & Laird, 1996).

Figure 4. Producing primary afferent depolarization (PAD) in central terminals of sensory fibres.

The proposed circuitry underlying dorsal root reflex activity at the level of the primary afferent in the dorsal horn is shown. An axoaxonic contact is made by a GABAergic interneuron onto a primary afferent terminal. Release of GABA from the interneuron and activation of GABA_A receptors produces an efflux of Cl^– from the primary afferent terminal, resulting in its depolarization. The Cl^– is concentrated in the primary afferent terminal by the Na⁺,K⁺,Cl^– cotransporter symbolized by the dark oval in the membrane of the afferent. (Modified from Alvarez-Leefmans et al. 1998, with permission.)

Animal studies show us that DRRs, particularly those in C fibres, have an important efferent function that affects the functional state of peripheral tissues (Maggi et al. 1987). This antidromic activity in C fibres results in release of calcitonin gene-related peptide (CGRP) and substance P (SP) from nerve endings (Maggi, 1995). CGRP induces neurogenic inflammation as evidenced by vasodilatation (flare) and SP induces plasma extravasation (oedema; for review see Willis, 1999). Activation of TRPV1-expressing C fibres produces changes in blood flow, a flare response and neurogenic inflammation (Lin et al. 1999). Evidence that these events are due in part to DRR activity comes from the observation that acute dorsal rhizotomy nearly abolishes the blood flow changes and significantly reduces the flare and neurogenic inflammation at the injection site. Furthermore, intrathecal bicuculline (GABA_A antagonist) or CNQX (non-NMDA antagonist) dramatically reduces all of these signs. Using the same CAP injection protocol, DRRs can be recorded from the central stump of cut dorsal rootlets and this activity is significantly reduced with intrathecal bicuculline or CNQX (Lin et al. 1999). From these experiments it is concluded that DRRs play a major role in the development of neurogenic inflammation. In addition to CAP, electrical stimulation or noxious mechanical stimulation of cutaneous C fibres can also evoke DRRs and the receptive field to drive the DRRs can be as small as one digit or as large as the whole body (Weng & Dougherty, 2005). Importantly, spread of DRR activity can occur over many cord segments, rostral and caudal to the point of origin (Bagust et al. 1993). In animal models of acute knee arthritis, DRRs are detected in the central stump of the cut dorsal root filaments or from the proximal stump of the cut medial articular nerve which innervates the knee joint (Rees et al. 1994; Sluka et al. 1995a). Sympathectomy does not change the DRR activity but cutting the dorsal roots eliminates it confirming a CNS origin. Delivery of non-NMDA antagonists via a microdialysis fibre threaded into the spinal cord dorsal horn also changes the outcome of knee joint inflammation: CNQX reduces the joint swelling, prevents the major temperature increase in the inflamed knee and attenuates the heat hyperalgesia that normally develops in the ipsilateral hindpaw (Sluka & Westlund, 1993). One explanation for this outcome is a CNQX-induced reduction in DRR activity and a consequent decrease in release of SP and CGRP in the periphery (Sluka & Westlund, 1993). DRRs not only have consequences in the periphery but also affect dorsal horn neuron activity. Specifically, NKCC1 cotransporter activity is functionally linked to sensitization of dorsal horn nociceptive neurons (Pitcher & Cervero, 2010). Neurogenic inflammation occurs not only in skin and joints but also in the eye, middle ear, dura mater and the respiratory, genito-urinary reproductive and digestive systems (see Geppetti & Holzer, 1996).

In an animal model of spinal cord injury (Carlton et al. 2009a), the presence of DRR activity has been reported (Du & Carlton, 2007) and may contribute to the pathophysiological pain observed following this CNS injury. Thirty-five days after a thoracic 10 (T10) contusion, spontaneous and evoked DRRs are recorded in Aδ and C fibres in filaments teased from the median nerve in the forelimb (Fig.5). A significantly higher percentage of units show DRR activity compared to that found in shams (Fig.6). There is no change in the DRR activity following dorsal or ventral rhizotomy and no change following sympathectomy confirming that sensory fibres carry the output (S.M. Carlton, unpublished observations). The evoked DRR activity is attenuated by intrathecal bicuculline and the rats develop forepaw oedema evidenced by an increase in paw volume at 35 days post-injury. Thus, DRR activity can develop not only after a peripheral injury/inflammation but also after a CNS injury.

Figure 5. Evidence of DRR activity in fibres in the median nerve following a T10 spinal cord injury.

A, dorsal root reflex (DRR) activity is recorded from the proximal stump of the cut median nerve in vivo. B, trace of spontaneous DRRs and evoked DRRs (produced by mechanical stimulation of the forepaw) recorded from the median nerve proximal stump.

Figure 6. Evidence that DRR activity increases in spinal cord-injured (SCI) animals compared to naive and sham.

Histograms show the increased percentage of C (A) and Aδ (B) fibres exhibiting DRR activity in spinal cord-injured rats compared to naive and sham. *P < 0.05 compared to naive; †P < 0.05 compared to sham, one-way ANOVA.

Conclusions

Our understanding of primary afferent function has grown exponentially since the 1600s. In contrast to the belief that primary afferents are waiting to conduct input unmodified directly to the brain, we now know that they take an active role in modifying the signal before it is ever transmitted to the CNS. In terms of pain management, the variety of receptors expressed by nociceptive afferents presents numerous targets for drug development. The ability of some receptors to decrease/dampen nociceptive input before it is transmitted to the CNS should move them to the top of the drug development list! Finally, the phenomena of DRRs, their significant contributions to pain and swelling following inflammation, and their ability to propagate to many segments rostral and caudal in the cord, suggests DRRs could be contributing to many pain states but heretofore this influence has been undetected and/or unrecognized. Clearly, the more we understand about how nociceptive primary afferents ‘work’, the better chance we have of developing evidence-based therapies for peripheral pain management.

Glossary

CAP

capsaicin

CGRP

calcitonin gene-related peptide

CNQX

6-cyano-7-nitroquinoxaline-2,3-dione

c-SOM

cyclo-somatostatin

DRG

dorsal root ganglion

DRR

dorsal root reflex

GABA

γ-aminobutyric acid

LY

LY341495

mGluR

metabotropic glutamate receptor

NKCC

Na⁺,K⁺,Cl^– cotransporter

NMDA

N-methyl-d-aspartate

SP

substance P

SST

somatostatin

TRPV1

transient receptor potential vanilloid 1

VT

volume transmission

Biography

Susan M. Carlton is a native of upstate New York. She did her undergraduate studies at Mary Washington College (Fredericksburg, VA, USA) attaining a BS in Biology. She then moved to the Medical College of Virginia where she earned her PhD in anatomy, with a concentration in neuroanatomy. She did two postdoctoral fellowships, one at the University of Texus Medical Branch (UTMB) with Dr William D. Willis and one at Yale University (New Haven, CT, USA) with Dr Carole LaMotte. She became an assistant professor at UTMB, Department of Neuroscience and Cell Biology and rose through the ranks. She is now a tenured professor, teaching and researching in this department for over 30 years. Her area of expertise is the peripheral nociceptor, studying pain transmission in both acute and chronic pain states.

Additional information

Competing interests

None declared.

Funding

Funding for studies done in the Carlton lab: NIH grants R01 NS027910 and R01 NS054765.