Dorsal root ganglion neurons and tyrosine hydroxylase – An intriguing association with implications for sensation and pain

Abstract

Tyrosine hydroxylase (TH) is a rate-limiting enzyme broadly expressed in noradrenergic and dopaminergic neurons in the central nervous system [57,70]. TH is also expressed by peripheral sympathetic neurons [98] as well as by enteric neurons within the gut [81,84]. Over 30 years ago, TH was unexpectedly discovered in developing and adult rodent cranial and dorsal root ganglion (DRG) neurons. Today, TH-expressing DRG neurons are being re-discovered as a relevant subpopulation. This review addresses the emerging importance of TH-expressing DRG neurons in sensation and pain mechanisms, focusing specifically on: 1) their nature as C-low threshold mechanoreceptors (C-LTMRs); 2) their involvement in nociception/pain; and 3) their catecholaminergic phenotype.

TH-expressing DRG neurons – a unique subpopulation

The presence of TH in peripheral sensory neurons was initially described in the field of neuronal development in the 1980s and 90s in rat [33,35,36,40,41,82,83] and avian [55,102] embryonic cranial and spinal ganglia. This was later confirmed in mouse [3,20,32] and also by means of TH gene promoter-driven expression of the Escherichia coli reporter gene lacZ [42,92], supporting the concept of dynamic developmental changes in neurotransmitter phenotype. However, it was soon discovered that many sensory neurons maintain TH protein and transcript synthesis during adulthood (Figure 1A). Moreover, today we know that they comprise one of seven key neuronal lineages in mouse DRGs [54,95]. Interestingly, whereas neuronal precursors are identified for most of these lineages, ancestors of TH-expressing DRG neurons remain unknown [54].

Figure 1. Morphological features of TH-expressing DRG neurons (quantitative and qualitative data in B, C is based on mouse).

(A) Adult TH-expressing DRG neurons have been detected in a number of species, including rat, guinea pig and mouse (—, not determined; * in humans, only embryonic expression has so far been studied [43]). (B) TH-expressing mouse DRG neurons are primarily small and medium-sized. Interestingly, differences can be observed when comparing visceral and non-visceral neurons, the former being represented by both small and medium-sized neurons, while the latter most often are small in size [data summarized from [7,10] and pooled to show differences between subpopulations. Percentage refers to the distribution among neuronal soma sizes of the total number of neurons measured for each target tissue/organ (921 non-visceral; 35 colonic and 74 urinary bladder)]. (C) On average, 15% of mouse DRG neurons express TH, although variations between DRGs innervating visceral versus non-visceral tissues or glabrous versus hairy skin are observed. Micrographs show examples of TH- and CGRP-expressing DRG neurons innervating non-visceral or visceral tissues, and the patterns of coexpression between the two markers. (Upper micrograph; L4-5 DRG neurons) TH-expressing DRG neurons (in green) innervating non-visceral tissues very seldom coexpress CGRP (in blue), and only rarely the two markers can be detected in the same neuron (double arrowhead). (Lower micrograph; S1 DRG neurons; in red, neurons innervating the urinary bladder) TH-expressing DRG neurons innervating the urinary bladder often coexpress CGRP (arrows). Additional bladder-projecting DRG neurons expressing CGRP but lacking TH are observed (arrowheads). Finally, several TH- (in green) or CGRP-only (in red) neurons are detected in the S1 DRG, and again a rare TH/CGRP coexpressing neurons (double arrowhead) Scale bar: 50 μm, applies for both micrographs.

Various immunohistochemical [7,10,41,45,48,59,82,83,96], in situ hybridization [7] and more recent large-scale single-cell RNA sequencing [95] studies documented that most adult rodent TH-expressing DRG neurons are small (Figure 1B). Only in mice is TH also detected in medium-sized adult visceral DRG neurons [10], or transiently in large DRG neurons during prenatal life [64]. The number of TH-expressing DRG neurons varies between studies, depending on species and DRG level (Figure 1C). On average, ~15% of all adult DRG neurons express TH, although TH mRNA is reportedly expressed in ~37% of L4–L6 mouse DRG neurons [95]. Whether TH transcripts are translated into protein in all neurons remains to be established.

The classical characterization of small and medium-sized DRG neurons as “peptidergic” (calcitonin gene-related peptide (CGRP)-expressing) or “non-peptidergic” (binding isolectin B4, IB4) [13,56,66] applies only partially to TH-expressing neurons (Figure 2A) [54,95], and appears to depend on the targeted tissue/organ. In fact, non-visceral TH-expressing DRG neurons are non-peptidergic and do not bind IB4 [7,10,32,54,59,82,95]. Even galanin, upregulated in small DRG neurons after peripheral nerve injury [15,90,97], remains absent in TH-expressing neurons under such conditions in mouse [7]. In contrast, a large percentage of TH-expressing DRG neurons targeting the colorectum or the urinary bladder of mouse are peptidergic [10].

Figure 2. Neurochemical features of TH-expressing DRG neurons.

(A) Neurochemical expression in DRG somata of unidentified L4-5 DRG neurons (not identified to tissue of innervation), and those innervating hairy skin or viscera ( , present; , absent; —, not determined). TH-expressing DRG neurons exhibit distinctive neurochemical features, some of which appear to be dependent on target (e.g., compare CGRP expression between neurons innervating hairy skin and viscera). Although VGLUT2 expression remains to be confirmed in somata of TH-expressing DRG neurons projecting to hairy skin and viscera (δ), the presence of the transporter in nerve terminals in close apposition to hair follicles [8], and the fact that virtually all visceral DRG neurons express VGLUT2 [11,12], suggest that TH-expressing DRG neurons synthesize both VGLUT2 and VGLUT3. (B) The presence of enzymes associated with catecholamine synthesis and metabolism, including TH and associated molecules (NET-1), supports the existence of dopaminergic (1) or noradrenergic (2) phenotypes for TH-expressing DRG neurons (shading denotes enzymes and catecholamines present or secreted from DRG somata). Abbreviations: Cav3.2, T-type calcium channel Cav.3.2; Gfrα2, GDNF family receptor alpha-2; MRGPRs, mas related G-protein coupled receptors; NaV1.8/9, voltage gated sodium channel 1.8/9; NFH, neurofilament high; NtsR1, neurotensin receptor type 1; P2X3, P2X purinoceptor 3; Piezo2, Piezo-type mechanosensitive ion channel component 2; RET, rearranged in transformation proto-oncogene; SP, substance P; TAFA4, chemokine-like protein TAFA4; TrkA, Tropomyosin receptor kinase A; TRPA1, Transient receptor potential cation channel, member A1; VGLUT, vesicular glutamate transporter type; Y2R, neuropeptide tyrosine receptor, type 2.

Expression of a number of other neurochemical markers in TH-expressing DRG neurons has been analyzed (Figure 2A). The expression of transient receptor potential channel, subfamily V, member 1 (TRPV1) remains, however, controversial, since absence [54,56,95], minor [22] or abundant presence [27,53] in TH-expressing DRG neurons in mouse [27] has been reported.

Altogether, mammalian sensory ganglia contain a select, lineage-specific, subpopulation of permanently or transiently expressing TH neurons with unique neurochemical coding, possibly influenced by the target tissue/organ (see [54,56,59]).

Effects of nerve injury

TH expression in primary afferent neurons is affected by peripheral nerve injury. This was first observed in nodose and petrosal ganglia after axotomy of the vagus nerve or its branches as a reduction in the percentage of TH-expressing neurons [29,41,104], accompanied by a decrease in TH catalytic activity [39]. In mouse DRGs, axotomy of the sciatic [7] or pelvic nerves [65], or spinal nerve ligation [101] results in downregulated expression of TH mRNA [7] and/or protein [7,65], or of TH promoter-driven expression of enhanced green fluorescent protein [101]. In rat, sciatic nerve chronic constriction injury (CCI) leads to reduced TH mRNA content in L4-5 DRGs [30].

Changes in TH expression after injury could derive from alterations in peripheral trophic support, as suggested by TH downregulation in nodose ganglion neurons after colchicine-induced inhibition of vagal trophic factor axoplasmic transport [103]. In contrast, dorsal rhizotomy, a “central nerve injury” model, fails to alter TH expression in DRG neurons [39]. However, exactly which neurotrophic factors influence TH-expressing adult DRG neurons appears controversial. In fact, while their lack of TrkA and expression of RET and Gfrα2 (Figure 2A) [49,59,95] suggest a role for neurturin and glial cell line-derived neurotrophic factor (GDNF) [75], Gfrα2 or RET deletion in mice does not affect the development of non-peptidergic DRG neurons [49,62] or their expression of TH [49]. In contrast, TH-expressing neurons in developing mouse nodose and petrosal ganglia are reported to rely on peripherally available brain-derived neurotrophic factor, GDNF or both for their survival [18] while remaining nerve growth factor-independent [20].

In conclusion, the expression of TH is typically downregulated by peripheral nerve injury. If other insults such as tissue inflammation also result in changes in the expression of TH in sensory neurons, and what trophic factor/s may be involved, it remains to be established.

Roles in normal sensation

Li and colleagues [59] reported that all C-LTMRs innervating hairy skin as longitudinal lanceolate nerve endings expressed TH. Moreover, they noted that together with Aβ- and Aδ-LTMRs, C-LTMRs comprise functionally distinct mechanosensory end organs, with their central projections integrated within discrete dorsal horn LTMR columns [59]. C-LTMRs normally convey innocuous mechanical (hair deflection and light touch) and cooling sensations [56,59,105]. This has been confirmed in studies analyzing the role of TAFA4, a chemokine-like protein [16], or the T-type calcium channel Cav3.2 [22], both highly co-expressed with TH in C-LTMRs [16,22]. In addition, C-LTMRs appear to be involved in processing gentle and affective touch (e.g., nurturing), in a manner common to humans and rodents [61,72]. Thus, TH-expressing C-LTMRs are likely participants of “emotional touch” (see [105]).

Roles in nociception/pain

Recent studies using transgenic mice revealed a role of TH-expressing DRG neurons in non-visceral pain mechanisms. The vesicular glutamate transporter 2 (VGLUT2) is essential for the processing of acute pain and heat hyperalgesia [89]. When VGLUT2 is selectively deleted in TH-expressing mouse DRG neurons [53], responses to noxious mechanical stimulation are unaffected whereas responses to noxious heat (potentially suggesting coexpression with TRPV1) and inflammation are decreased. Deletion of TAFA4, selectively coexisting with TH in C-LTMRs, results in enhanced and prolonged mechanical hypersensitivity after either inflammation or CCI, suggesting that TAFA4 modulates the excitability of these neurons [16]. Finally, deletion of Cav3.2, a key regulator of sensory neuron excitability expressed both in Aδ-LTMRs and TH-expressing C-LTMRS, reduces spared nerve injury-induced mechanical and cold allodynia and attenuates tissue inflammation-induced hypersensitivity [22].

Less is known of the role of TH-expressing DRG neurons in visceral pain. However, pharmacological blockade of Cav3.2 in rats prevents the occurrence of colonic hypersensitivity induced by intracolonic sodium butyrate [21]. This effect appears to take place in peptidergic, IB4-negative, Cav3.2-expressing colorectal DRG neurons [63], although it remains to be established if they also express TH, as mentioned above for mice [22]. Also, patients with classic or non-ulcerative interstitial cystitis show increases in the number of TH-IR nerve fibers innervating the urinary bladder, likely due to outgrowth of sympathetic neuron projections [77]. However, TH-expressing DRG neurons may also participate and thus, along with sympathetic input, modulate urinary bladder sensitivity.

Potential catecholaminergic phenotype and pain modulation

There are currently three hypotheses regarding the catecholaminergic phenotype of TH-expressing DRG neurons. The first, for which evidence is strongest, is that TH-expressing DRG neurons synthesize and use dopamine or L-DOPA as neurotransmitters. Thus, a number of dopamine-associated enzymes and enzymatic products were identified in rat [5,51,52,99] and mouse [27] DRG (Figure 2B). Synthesis of dopamine was also reported in chick [80] and rat [5,99] DRGs, and additional support comes from studies in cranial sensory neurons, where the dopaminergic nature of TH-expressing neurons has been confirmed [19,31,34,67]. In humans, dopamine and associated metabolites were identified in peripheral nerves [50].

Dopamine release by DRG neurons would modulate pain at two specific targets: 1) Central DRG nerve terminals, and 2) Local spinal interneurons and projection neurons. Dopamine receptors 1–5 (D1–5Rs) have been reported in rat DRG neurons using RT-PCR [100], western blot and immunohistochemistry [23]. These receptors are functional, as established by the inhibitory effects of D1- and D2Rs [58,68], or D1- and D5Rs [23], on the electrical activity of predominantly small DRG neurons [23,58,68]. Such effects could modulate both the excitability of DRG nociceptors as well as neurotransmitter release from their central terminals. In support, activation of D2Rs inhibits presynaptic N-type calcium channel currents in rat chemosensory cranial visceral afferents, resulting in inhibition of neurotransmitter release [44]. Functional D2Rs have not only been documented at pre-, but also at postsynaptic sites in the substantia gelatinosa (lamina II) of rat [94]. Importantly, stimulation of the A11 cell group, possibly the major supraspinal source of dopamine (see [46]), results in antinociception, most likely by D2R-dependent inhibition of substantia gelatinosa neurons, both in rats exposed to noxious stimuli [94] or after peripheral nerve injury [2,71]. Activation of spinal D2Rs also attenuates carrageenan-induced hyperalgesia [24,25,60,73]. Finally, an antinociceptive role has been ascribed to dopamine in human patients suffering fibromyalgia [69] or restless leg syndrome [93].

Potential synthesis and release of L-DOPA by DRG neurons could also serve in antinociception. L-DOPA, after conversion to dopamine, reduces nociceptive behavior induced by intrathecal administration of substance P in rats, via a D2R-dependent mechanism [91]. In rats with CCI, intrathecally injected L-DOPA reduces mechanical and thermal hyperalgesia, again through D2R actions [14]. Interestingly, the D2R-dependent antinociceptive effect seems to be enhanced after CCI [14] or spinal nerve ligation [2,76], possibly due to injury-induced D2R upregulation in pre- and/or post-synaptic sites in the dorsal horn [2].

The second hypothesis, based on a single study in rats with bilateral lumbar sympathectomy and adrenalectomy, suggests that TRPV1-expressing DRG neurons release noradrenaline (but not dopamine) from their peripheral terminals upon knee joint capsaicin administration (Figure 2B). The authors also detected in these rats presence of norepinephrine transporter-1 (NET-1, for reuptake of extracellular norepinephrine and dopamine [74]), dopamine β-hydroxylase (DBH) and monoamino oxidase type A (MAO-A) in DRGs. They also found that neonatal capsaicin treatment alters the expression of DBH and MAO-A [17]. Therefore, a sensory neuron source of peripheral noradrenaline release should not be discounted when discussing the effects of adrenoceptor blockers on pain states [79].

The third hypothesis is that TH is not active in primary afferent neurons, based on the reported weak expression or absence of any other cathecolaminergic trait in guinea pig [48], mouse [7] and rat [82,83,96] non-visceral DRG neurons, and in mouse visceral DRG neurons [10]. However, mouse diencephalic A11 dopaminergic neurons have been shown to lack, for example, the dopamine transporter [46]. Therefore, absence of related molecules should not necessarily rule out a specific catecholaminergic phenotype.

Summary

In conclusion, more than 30 years after being discovered, TH-expressing DRG neurons are recognized as a key subpopulation during DRG development, with emerging roles in adulthood in several species. However, future studies will be necessary to (1) provide definitive proof of the synthesis of catecholamines by TH-expressing DRG neurons. 2) If synthesis of dopamine, L-DOPA or both is established, their potential somatic release and action on dopamine receptor-expressing DRG neurons should be explored (as observed for substantia nigra neurons, which engage in somatodendritic release [1,4,6,26,86]). 3) Determine whether TH-expressing (catecholaminergic?) DRG neurons are present in humans. 4) Analyze the impact of selective deletion or pharmacological manipulation of TH in DRG neurons on visceral and non-visceral pain. 5) Analyze the implications for functionality and/or trophic factor support of the observed differences in abundance of TH-expressing DRG neurons, relative to spinal segmental level and target organ/tissue.

Finally, while data on cell lineage or molecular signature of whole DRGs is relevant for understanding the role of sensory neurons, their morphological and functional diversity is importantly reliant on the tissue innervated. A more comprehensive characterization of DRG subpopulations, including target-identified TH-expressing ones, bears the potential for an improved understanding of the involvement of each DRG neuron subpopulation in discrete types of sensory and pain mechanisms.