Recordings from muscle afferent fibers have provided a wealth of information about the response properties of these cells in physiological and pathophysiological situations, and how these responses are modified by sensory mediators . However, ion channel activation and the signal transduction cascades modulating primary afferent excitability are most directly studied by making electrophysiological or optical recordings from sensory neuron cell bodies, and the present study provides some of the first descriptions of the electrophysiological properties of isolated sensory neurons innervating muscle. The results highlight the differences in the molecular signatures of proprioceptive muscle afferents and other muscle afferents, as well as the differences between muscle afferents and those which innervate other structures in the head such as teeth [4, 8].
Sensory neuron modality can only be determined in in vivo or intact ex vivo preparations, thus we cannot assign a definitive physiological function to the cells in the present study. However, sensory neurons that detect potentially noxious stimuli (nociceptors) are thought to preferentially express a number of ion channels not normally found in other primary afferents. For example, expression of the TTX-resistant sodium channels NaV1.8 and NaV1.9 has been strongly correlated with a nociceptive sensory modality in in vivo recordings made from sensory neuron cell bodies [10, 11]. Channels such as the vanilloid receptor TRPV1 are thought to be expressed exclusively by nociceptors because they are normally activated by demonstrably noxious stimuli and there is a strong correlation between selective pharmacological activation of the channels and human sensations [7, 12]. Other channels are thought to be associated with nociceptors because their biophysical properties are sufficient to explain a response to a noxious stimulus by a subset of sensory neurons. For example, ASIC3 channels are activated by the modest changes in extracellular calcium and pH that accompany cardiac ischaemia and are highly expressed in a subset of cardiac sensory afferents that are presumed to transmit the pain of angina . The assignment of ion channels to nociceptive neurons has also been made based on correlating channel expression with other putative markers of nociceptors including small soma diameter, expression of substance P, calcitonin gene related peptide or TRPV1 and expression in sensory neurons projecting to tissues from which the only conscious sensation is pain .
Almost all the masseter afferents isolated from the TG expressed significant amounts of at least one of the putative "nociceptive" ion channels we examined in this study; capsaicin activated TRPV1 channels, acid activated ASIC-3-like channels or ATP activated P2X3-like channels. With the exception of cells expressing TRPV1, a nociceptive phenotype cannot be reasonably inferred from the expression of any one channel, however significant numbers of masseter afferents expressed two or more of the channels we examined. Almost all cells expressing P2X3-like channels also expressed ASIC3-like currents and about 25% also expressed TRPV1; most cells expressing ASIC3-like currents also expressed either P2X3-like currents or TRPV1, and more than 50% of the TRPV1 expressing cells also expressed either P2X3-like currents or ASIC3-like currents. These data indicate that most TG masseter afferents can detect a noxious stimulus, but whether this represents their primary or only function remains unknown. By contrast, masseter afferents isolated from the MeV nucleus did not express TRPV1, P2X3-like or ASIC3-like channels, consistent with their function as purely proprioceptive muscle spindle afferents .
There is considerable evidence that a proportion of muscle afferents can reliably signal stimuli in both the innocuous and noxious range, and the properties of some of these afferents are consistent with the expression patterns of the channels in muscle afferents found in the present study . In particular, afferents that are activated by the changing metabolic state of muscle (metaboreceptors)  appear to express channels classically associated with nociceptors, such as TRPV1, but clearly signal non-noxious information as well. Thus, lactic acid stimulation of muscle afferents in rat produces a classic cardiovascular pressor response that is sensitive to the ASIC channel antagonist amiloride but not to the TRPV1 antagonist capsazepine . However, the lactic acid-induced response is attenuated after pretretament with the potent TRPV1 agonist resiniferatoxin, which desensitizes or destroys TRPV1-expressing nerves . Further, while capsaicin produces a pressor response, blocking TRPV1 does not inhibit a contraction-induced pressor response . Recordings from muscle afferents also show that capsaicin activates a population of Group III and Group IV afferents, some of which also proton-sensitive . Thus it seems that a significant proportion of muscle afferents involved in producing activity-induced cardiovascular reflexes, perhaps mediated by activation of ASIC channels, also express TRPV1. Injection of capsaicin into human masseter muscle is painful , so there is no question that there are TRPV1 expressing afferents in muscle that transduce noxious stimuli. Our findings that about 50% of TRPV1 containing masseter afferents also expressed ASIC channels, and 30% of ASIC expressing afferents expressed TRPV1 are consistent with these results. There is no other information about the co-expression of TRPV1, ASIC and P2X receptors in afferents from the masseter muscle, although a relatively limited co-expression of TRPV1 and P2X3 receptors has been reported in gastrocnemius-soleus muscle afferents .
The ASIC channels in trigeminal masseter afferents seemed to be largely comprised of homomeric ASIC3 channels or ASIC channel heteromers containing ASIC3. ASIC3 channels are highly sensitive to changes in extracellular pH and lactate and if activated by a substantial change in pH they desensitize significantly more rapidly than other ASIC channels [5, 18, 19]. The pH 6-induced currents in most trigeminal ganglion masseter afferents desensitized more than 90% during the 1.5 s proton application, while in the remaining neurons the significant residual current (> 10% of the peak) suggested the presence of other ASIC subunits in these cells, probably ASIC1 . Thus, the majority of ASIC currents observed in TG ganglion afferents had similar properties to those found in rat cardiac afferents (18), which are thought to be ASIC3-mediated. However, in the absence of selective blockers of ASIC subunits, we cannot definitively assign the currents we observed to specific ASIC subunits or combinations of subunits. MeV masseter afferents exhibited robust acid-induced currents but these were less sensitive to changes in extracellular pH and desensitized much more slowly than ganglion neuron ASIC currents.
In the only previous study of ASIC channel function in muscle afferents, 50% of sensory neurons labeled from the gastrocnemius muscle responded to pH 5.0 solution with robust inward currents . The currents elicited in 30% of the cells were tentatively assigned to ASIC3/ASIC2b heteromers. The crucial role of ASIC3 channels in muscle-associated sensory function is underlined by the main finding of that study, which is that ASIC3 channel expression is required for the long lasting hyperalgesia produced by repeated acid injection in muscle . The currents we observed in the majority of rat masseter afferents differ from those reported in mouse dorsal root ganglion neurons , primarily due to the lack of a significant sustained current component at pH 6.0-in our experiments this component was only 3% of the peak current. However, our conclusion that ASIC3 forms an essential part of masseter afferent ASIC channel complexes, is similar to that reached by others based on experiments in mouse DRG neurons from ASIC-null mice .
Action Potentials and Sodium Channels in Masseter Afferents
The action potentials of the MeV masseter afferents were narrow and lacked an inflection on the downward component of the current, consistent with previous recordings from acutely isolated MeV neurons  and MeV neurons in brain slices [23, 24]. The I
Na recorded from MeV masseter afferents were completely blocked by TTX. These data are consistent with reports that muscle spindle afferents do not express detectable NaV1.8 or NaV1.9 immunoreactivity [10, 11]. The narrow action potentials of proprioceptive afferents are consistent with the very rapid firing rates that these neurons achieve-exceeding 200 Hz (e.g. ). By contrast, the masseter afferents isolated from the TG had a wide range of action potential widths and shapes, and most cells had a significant component of TTX-resistant I
Na. TTX-resistant I
Na are subject to acute regulation by a variety sensory mediators acting via G protein-coupled or tyrosine kinase-linked receptors, particularly prostaglandins, bradykinin and nerve growth factor . The changes in I
Na availability produced by these mediators mean that afferents expressing TTX-resistant I
Na are likely to be subject to rapid changes in excitability reflecting the state of the tissue they innervate. Wider action potentials and greater amounts of TTX-resistant I
Na are strongly correlated with a nociceptive modality, but these properties vary between afferents of different conduction velocity classes as well as between afferents of different modality within a class , and one cannot define a neuron as nociceptive simply on the basis of a action potential duration or its sensitivity to TTX. Nevertheless, within the TG masseter afferents, which presumably contain cells with a nociceptive function, smaller neurons tended to have wider action potentials. There was no such relationship apparent within the proprioceptive MeV masseter afferents.
We found a wide variety of ATP-induced currents in TG masseter afferents, similar to results from other studies in sensory neurons [28–30]. Messenger RNA and receptor-like immunoreactivity for 6 of the 7 cloned subtypes of P2X receptor are found in the trigeminal ganglion [31, 32] and the currents we recorded are likely to be comprised of a mixture of homo- and heteromeric P2X receptor channels. Although attempting to define the P2X subunits responsible for the variety of ATP currents was beyond the scope of this study (but see ) we attributed the rapidly desensitizing ATP current observed in some masseter afferents to P2X3 receptor activation. Rapidly desensitizing ATP currents in sensory neurons have been reported to depend on the presence of the P2X3 gene or have been identified pharmacologically as P2X3 receptors [33–35] and although the kinetically similar P2X1 receptor has been shown to be present in sensory neurons by immunohistochemical methods, there is little electrophysiological evidence for currents mediated by P2X1 receptors in rat or mouse sensory neurons [4, 29].
P2X3 subunits make a major contribution to the ATP currents in trigeminal neurons projecting to the tooth pulp, both as P2X3 receptor homomers and putative P2X2/P2X3 heteromers . Relatively fewer masseter afferents express P2X3-like immunoreactivity . In the present study we observed rapidly desensitizing ATP currents either alone or in combination with other P2X currents in about 55% of TG masseter afferents, which is similar to the proportion of tooth pulp nociceptors which displayed fast ATP currents (44%, ). P2X3-like immunoreactivity has been reported in about 25% of masseter afferents , which is a considerably smaller proportion than suggested by the present report. These differing results may reflect differing sensitivities of immunohistochemistry and electrophysiology, they may be due to the presence of some P2X1-containing currents in masseter afferents or perhaps arise from the short time the isolated neurons spend in culture. 30 – 40% of tooth pulp afferents can be labeled with P2X3 antiserum [4, 37] but it is interesting to note that more than 50% of tooth pulp afferents challenged with ATP also displayed persistent currents. This indicates that while P2X3-containing ATP receptors are found in many putative nociceptors, they are not the only P2X subunits that may detect noxious stimuli signalled by ATP.
Opioid Modulation of Calcium Channels
The relatively high μ-opioid receptor sensitivity of jaw muscle afferents is similar to that reported in afferents projecting to hindlimb muscles, where about 75% of cells were sensitive to DAMGO . The apparently high sensitivity of muscle afferent I
Ca to μ-opioid agonists contrasts with the reported low frequency of μ-opioid agonist modulation of I
Ca in skin and colonic afferents (approximately 10%, ). The relative insensitivity of MeV I
Ca to DAMGO (1 of 11 cells responding) is consistent with the extremely low mRNA abundance in these cells (2 of 72, ). The modulation of I
Ca in MeV neurons by the GABAB receptor agonist baclofen is in contrast to the lack of affect of baclofen on the membrane properties of MeV neurons in slices .
Interestingly, the μ-opioid receptor sensitivity of masseter muscle afferents differs markedly from that reported for the "purely nociceptive" afferents from tooth pulp [8, 40]. DAMGO inhibited I
Ca in more masseter afferents than tooth pulp afferents, regardless of cell size (80% versus 42% respectively ), and most strikingly, DAMGO was equally effective in small and large masseter afferents (85% and 82% of cells inhibited respectively). By contrast, DAMGO inhibited I
Ca in only 30% of large tooth pulp afferents . As 30% of large masseter afferents had substantial capsaicin currents (> 500 pA), indicating that these cells are likely to be nociceptors, our data suggest that μ-opioid receptors may be differentially expressed in distinct populations of nociceptors projecting to different tissues in the head, i.e. preferentially expressed in muscle nociceptors versus tooth pulp nociceptors. These data suggest that opioid analgesics may be better at relieving some types of head pain than others, and further that the endogenous opioid analgesic systems of the periphery may display differential effectiveness against nociceptive stimuli arising from distinct structures. The apparently high expression of opioid receptors in TG masseter afferents also suggests that these receptors may have functions in muscle physiology other than simple inhibition of nociception.