Etv4 regulates nociception by controlling peptidergic sensory neuron development and peripheral tissue innervation

Understanding the development and establishment of neural circuits that regulate pain transduction is an important issue in neural development. However, little is known about the molecular mechanisms underlying this process. Primary afferent neurons, located in the dorsal root ganglion (DRG), are sensory neurons with the capacity to detect and transduce external physical stimuli into electrical signals that are then conveyed to the central nervous system (CNS). Primary afferent neurons have a single axon, which bifurcates into a peripheral branch that innervates diverse targets such as the skin or muscle, and a central branch, which establishes synaptic contact with CNS neurons. According to their perception modalities, sensory neurons can be classified into proprioceptors, which sense body positions, mechanoreceptors, which mediate touch, and nociceptors, which are involved in the perception of pain and temperature (Basbaum et al., 2009). These major subpopulations exhibit different physiological, morphological and molecular properties (Usoskin et al., 2015).

It has been established that the expression of neurotrophic factor receptors defines the functionality of DRG neurons (Marmigère and Ernfors, 2007). Neurons that express TrkA respond to nerve growth factor (NGF), whereas TrkB⁺ or TrkC⁺ neurons are activated by brain-derived neurotrophic factor (BDNF) and neurotrophin 4/5 (NT4/5) or neurotrophin 3 (NT3), respectively. These neurothrophic signals are essential for the peripheral innervation of appropriate targets, for cell survival, and for the expression of ion channels, receptors and molecules that define their functional properties (Lallemend and Ernfors, 2012; Meltzer et al., 2021).

Neurons that express the TrkA receptor during embryonic development project fibers into the epidermis in response to NGF and originate nociceptive neurons. Animals deficient in TrkA and NGF lack nociceptive neurons and their projections, and are defective in response to noxious stimuli (Carroll et al., 1992; Crowley et al., 1994; Smeyne et al., 1994). In humans, mutations in TrkA or NGF have been associated with deficits in nociceptor development and pain sensation (Einarsdottir et al., 2004; Indo, 2010; Indo et al., 1996). Later throughout development, these neurons segregate into two subpopulations: peptidergic and non-peptidergic neurons. Peptidergic nociceptors are defined by the expression of calcitonin gene-related peptide (CGRP) and other specific neuropeptides, such as substance P (Ju et al., 1987; Xu et al., 1990), and maintain the expression of TrkA. On the other hand, non-peptidergic neurons extinguish the expression of TrkA and respond to glial-derived neurotrophic factor (GDNF) family ligands (GFLs) through the receptor tyrosine kinase Ret and its co-receptor molecules: GFRα proteins (Lallemend and Ernfors, 2012; Meltzer et al., 2021).

As mentioned, in adult DRGs, TrkA expression exclusively persists in peptidergic neurons, and signaling through NGF plays an important role in setting pain sensitivity. Many reports indicate that endogenous NGF levels are elevated in chronic pain (Denk et al., 2017; Pezet and McMahon, 2006; Woolf et al., 1994). Moreover, the administration of NGF in the periphery leads to hyperalgesia (Dyck et al., 1997; Lewin et al., 1993; Petty et al., 1994; Rukwied et al., 2010).

Different ion channels expressed in peptidergic neurons are involved in the transduction of painful stimuli, including transient receptor potential vanilloid 1 (TRPV1), which is a heat- and capsaicin-evoked channel expressed predominantly by peptidergic afferents (Caterina et al., 1997; Yang and Zheng, 2017). The genetic deletion of the Trpv1 gene results in an impaired response to heat stimuli and capsaicin (Caterina et al., 2000). Interestingly, the expression of TRPV1 and its phosphorylation levels are controlled by the activity of NGF and the TrkA receptor (Amaya et al., 2004; Ji et al., 2002; Xue et al., 2007; Zhang et al., 2012; Zhu and Oxford, 2007).

NGF signaling in sensory neurons involves its binding to the TrkA receptor at the axon terminal, subsequent activation of the MEK/ERK and PI3K/Akt pathways, and retrograde signal transduction to the nucleus, where it triggers a specific transcriptional program. This signaling is essential for neuronal viability, neurite outgrowth and pain sensitivity transduction (Chowdary et al., 2012; Harrington and Ginty, 2013). In mammals, members of the Pea3 (polyoma enhancer activator 3) subfamily of ETS (E26 transformation specific) transcription factors have been implicated as mediators of neurotrophic factor signaling downstream of the MEK/ERK pathway and in regulating the establishment of synaptic connections (Arber et al., 2000; Fontanet et al., 2013). This family includes three members: Etv1, Etv4 and Etv5, also named Er81, Pea3 and Erm, respectively. These factors are associated with the development and progression of different types of tumors and also with tissue remodeling processes (de Launoit et al., 2006; Oh et al., 2012; Oikawa and Yamada, 2003; Yordy and Muise-Helmericks, 2000). In particular, Etv4 is highly expressed in embryonic DRG TrkA⁺ sensory neurons. Furthermore, in vitro studies have shown that Etv4 is induced by distal NGF and is required for proper axonal growth of TrkA⁺ neurons (Fontanet et al., 2013). However, the physiological role of Etv4 in the development of the peripheral nervous system (PNS) remains unclear.

We show here that Etv4 is required for a correct cutaneous peptidergic innervation and pain transduction. Our findings indicate that Etv4 is necessary for the expression of enzymes involved in extracellular matrix (ECM) remodeling, and for the proper formation of growth cones that mediate NGF-dependent axonal growth. We also demonstrate that Etv4 mutant mice are less susceptible to heat nociception and chemical stimuli. Furthermore, we provide evidence indicating that Etv4 is essential for the development of nociceptors, by mediating the expression of the pain transducing channel TRPV1.

Based on the fact that Etv4 mediates NGF signaling (Fontanet et al., 2013), and that NGF is required for the epidermal innervation of DRG neurons that express TrkA (peptidergic nociceptive cells) during embryonic development (Patel et al., 2000), we examined whether Etv4 deletion plays a role in such event by analyzing Etv4-deficient mice. For this purpose, we examined presence of free nerve endings in transverse sections of the glabrous skin of the hindpaw of Etv4^−/− and wild-type mice at two different developmental stages (newborn P0 and 2-month-old mice). The peptidergic fibers were visualized using antibodies against calcitonin gene-related peptide (CGRP), which label neurons that respond to NGF. In these assays, we found fewer CGRP⁺ axon terminals penetrating through the basal membrane into the epidermis in Etv4-null mice compared with wild-type littermates at P0 (Fig. 1A,C). A similar deficiency in CGRP⁺ epidermal innervation was also observed in adult Etv4^−/− compared with wild type (Fig. 1B,D). Likewise, epidermal CGRP⁺ fibers were shorter in adult mutant than in control mice (Fig. 1E). This shortening did not result from external constraints, such as a reduced thickness of the skin, as no difference in skin thickness was observed between control and Etv4-deficient mice (data not shown). These findings indicate that developing peptidergic nociceptive sensory neurons do not adequately innervate the epidermis in the absence of Etv4. In order to analyze epidermal innervation of non-peptidergic nociceptive neurons, we performed staining using antibodies against the GFRα1 and GFRα2 receptors, which labeled non-peptidergic DRG sensory fibers (De Vincenti et al., 2021; Sakai et al., 2017). Quantification of this innervation indicated that the density of cutaneous non-peptidergic fibers was unchanged between wild-type and Etv4 knockout adult mice (Fig. 1F,G). Taken together, these observations indicate that Etv4 is essential for cutaneous peripheral peptidergic projections, but is not required for non-peptidergic innervation.

We next examined the expression of Etv4 mRNA in DRG at different embryonic and postnatal stages, using RT-PCR. Interestingly, while Etv4 mRNA is expressed during embryonic development, its expression decreases in the postnatal stages (Fig. 1H). Immunostaining of DRG sections of E14.5 and P20 mice confirmed this result at the protein level, indicating that at embryonic stages Etv4 colocalizes with a large number of TrkA-positive neurons and that after birth its levels are substantially reduced (Fig. 1I). The specificity of the Etv4 antibodies was evaluated on tissue from Etv4-deficient mice (Fig. S1) (Fontanet et al., 2018). In agreement with previous work, we observed that the vast majority of TrkA-positive neurons express Etv4 at early developmental stages (Fontanet et al., 2013). However, the expression of Etv4 exceeds the TrkA-positive cell population, raising the possibility that Etv4 might have physiological contributions in other populations of DRG neurons. At postnatal stages, we observed the presence of Etv4 in the nuclei of both TrkA-positive and TrkA-negative neurons, indicating that Etv4 expression is not exclusive to peptidergic cells (Fig. 1I).

To assess whether Etv4 is required for the proper development of central spinal projections of CGRP⁺ afferents, we evaluated the distribution of CGRP projections in tissue from Etv4-deficient and wild-type animals at P15. In wild-type mice, CGRP-labeled fibers were found extending into the dorsal horn and projecting predominantly towards superficial laminas I and II. No differences were observed either in the area covered by CGRP⁺ fibers or in the number of CGRP⁺ terminals that innervate the dorsal horn of the spinal cord between Etv4 mutant and wild-type mice (Fig. 2). Moreover, aberrant CGRP⁺ projections were not detected in the spinal cord of Etv4-deficient mutant mice compared with wild-type animals (Fig. 2D). Intriguingly, a reduction in the intensity of CGRP⁺ staining was detected in Etv4-null mice (Fig. 2E). Despite this, no differences in CGRP mRNA levels were observed between wild-type and Etv4 knockout mice (Fig. S2). To further analyze the role of Etv4 in peptidergic central innervation, we labeled central spinal projections of small-diameter neurons using TrkA staining. No differences were observed in the area covered by TrkA-positive fibers or in the intensity of the TrkA-positive terminals innervating the spinal cord dorsal horn between Etv4-mutant and wild-type mice (Fig. 2C,F). Thus, together these findings suggest that Etv4 affects the morphology and density of peripheral but not central axons of peptidergic neurons.

Based on the defects in peptidergic epidermal innervation observed in Etv4-null mice, we decided to further analyze whether the Etv4 deletion affects neuronal viability in the development of DRG sensory neurons. To this end, sections of P15 Etv4-mutant and wild-type mice were stained with the pan-neuronal marker NeuN or with the sensory neuron marker PGP9.5. No significant differences were observed in the total number of sensory neurons, suggesting that Etv4 is not associated with the survival of DRG neurons (Fig. 3A,B), and therefore the defect observed in peripheral peptidergic innervation is not a consequence of a change in neuronal viability. According to these data, the size of lumbar sensory ganglia of the Etv4 mutants was comparable with that of wild-type mice (Fig. S3).

We then decided to examine whether the altered epidermal innervation in Etv4-knockout mice could be explained by changes in the number of nociceptive neurons. As mentioned above, these neurons are classified into non-peptidergic, i.e. expressing the GFL receptor Ret, GFRα1 and GFRα2, or peptidergic, i.e. characterized by the expression of TrkA. No changes were observed in the percentage of TrkA⁺ cells relative to the total number of DRG neurons (NeuN⁺) in Etv4 mutant mice compared with wild-type mice in young P15 animals (Fig. 3A,C). The expression of TrkA (Ntrk1), Ret and Gfra1 mRNAs was analyzed by RT-PCR. No differences were observed in the expression levels of the mRNA of these receptors in Etv4-deficient mice compared with wild-type animals (Fig. 3D,E).

Taken together, these results allow us to conclude that, whereas Etv4 ablation does not impact either the number of TrkA⁺ nociceptive neurons or the expression of neurotrophic factor receptors, it leads to a deficit in peptidergic peripheral innervation. The change in peripheral innervation observed in Etv4 mutants is therefore attributable neither to cell death nor to a cell fate conversion.

To further characterize the role of Etv4 in axonal extension, we performed in vitro assays on dissociated DRG primary neurons. Embryonic DRG neurons from Etv4^−/− or wild-type mice were cultured in the presence of NGF and the extent of axonal processes was measured after 24-36 h. Etv4 mutant neurons showed shorter neurites compared with neurons derived from wild-type animals (Fig. 4A-C). These differences were not due to variations in cell viability (Fig. 4D). In addition, the growth of cranial sensory neurons obtained from the trigeminal ganglia (TG), which also express Etv4 and respond to NGF, was analyzed (Fig. S4). Consistent with the results obtained using DRG neurons, the absence of Etv4 in TG sensory neurons also resulted in poor axonal growth in the presence of NGF (Fig. 4E-G). Therefore, our data indicate that Etv4 is required for proper axonal growth of the NGF-responsive sensory neurons.

Developing axons are guided to their target tissues by extracellular physical and molecular cues detected by the growth cones at their tips. Extracellular signals are locally integrated in the growth cone and converge on dynamic changes in the actin cytoskeleton that play a key role in generating the force required for neurite extension. As Etv4 is involved in the growth of neurites, we decided to examine whether it does so by regulating the morphology of the growth cone. We measured growth cone area of Etv4-deficient and wild-type DRG and TG neurons, cultured in the presence of NGF. Cells were labeled with phalloidin, which stains polymerized actin. DRG sensory neurons derived from Etv4^−/− mice exhibited a significant reduction in growth cone area compared with their wild-type counterparts, in the presence of NGF (Fig. 4H,I). A similar reduction in the growth cone size was observed in Etv4-deficient trigeminal sensory neurons (Fig. 4J,K).

Several studies indicate that Pea3 transcription factors regulate the expression of different metalloproteinases (MMPs), which are involved in axonal growth of DRG sensory neurons, likely via their known activity on ECM remodeling (Sanz et al., 2017; Yan and Boyd, 2007). Consistent with a role for metalloproteinases in NGF-induced DRG sensory neurite growth, a significant reduction in the neurite length and growth cone area was observed when wild-type neurons were treated with the pan-metalloproteinase inhibitor GM6001 (Fig. 5A,B), mimicking the phenotype of Etv4^−/− neurons. These results indicate that metalloproteinases are required for the correct growth of NGF-dependent embryonic sensory neurons.

Previous evidence suggests that expression of metalloproteinases, such as MMP3, MMP13 and MMP10, as well as urokinase-type plasminogen activator receptor (UPAR), a GPI-anchor molecule that regulates the MMPs activation, are induced by NGF (Chen et al., 2008; Farias-Eisner et al., 2000; Fontanet et al., 2013; Yan and Boyd, 2007). Moreover, the analysis of the promoter sequences of the corresponding genes reveals the existence of specific sites for the binding of PEA3 transcription factors (Bhattacharya et al., 2001; Yan and Boyd, 2007). To analyze the possibility that Etv4 could mediate their expression in response to NGF, we used the neuronal PC12 cell line, which expresses the NGF receptor TrkA and responds to NGF stimulation by increasing Etv4 levels. The cells were transfected with a plasmid containing a specific Etv4-shRNA and stimulated with NGF for 8 h. Although an increase in the expression of Etv4, Mmp3, Mmp13, Mmp10 and Upar (Plaur) mRNAs was observed in response to NGF, transfection of PC12 cells with the shRNA for Etv4 inhibited these changes (Fig. 5C). Subsequently, we analyzed the changes in the expression of these molecules in dissociated DRG neuronal cultures obtained from wild-type and Etv4-deficient mice, in response to 8 h of stimulation with NGF. As shown in Fig. 5D,E, NGF promoted a significant increase in the mRNA expression of Mmp3, Mmp13 and Mmp10 in DRG neuronal cultures obtained from wild type but not in cultures from Etv4-deficient mice, confirming that Etv4 is required for the induction by NGF of these ECM-remodeling molecules. Although a substantial, but not significant, increase in Upar mRNA expression was observed upon NGF stimulation, a significant reduction in the levels of this molecule was observed in cultures obtained from Etv4 knockout mice, indicating that Etv4 is required for promoting Upar mRNA expression in NGF-responsive sensory neurons (Fig. 5D,E). The absence of the expression of Etv4 mRNA in DRG neurons obtained from Etv4-deficient mice was confirmed by RT-PCR (Fig. S5).

In order to analyze whether the deficit in neurite outgrowth observed in Etv4^−/− neurons is attributable to reduced levels of MMPs, we next asked whether exogenously supplied MMPs could revert the phenotype of Etv4-deficient DRG cells. To this end, HEK-293 cells were transfected with control or Mmp3- and Mmp13-expressing vectors. After confirming the expression of these molecules (Fig. S6), the conditioned medium was added to wild-type and Etv4-deficient DRG neurons cultured in the presence of NGF. Our results indicate that these MMPs were able to compensate for the outgrowth deficiency exhibited by Etv4 knockout neurons, supporting the idea that their induction by Etv4 is necessary for the axonal growth of NGF-dependent sensory neurons (Fig. 5F).

Together, these results support a model in which Etv4 would be required for NGF-induced expression of ECM remodeling proteins and for NGF-induced axonal growth. Alteration of this pathway may partially account for the deficiencies in peptidergic nociceptive innervation observed in the epidermis of Etv4-mutant mice

Given that Etv4 is highly co-expressed with TrkA during the development of DRG neurons, and is required to mediate the effect of NGF/TrkA signaling on sensory axonal outgrowth and epidermal innervation, we next asked whether ablation of Etv4 could also affect other NGF-induced sensory properties, such as the sensitivity to painful stimuli. As NGF/TrkA signaling mediates thermal and inflammatory sensory modalities (Khodorova et al., 2017; Ueda et al., 2010), we analyzed the behavioral response of Etv4-deficient mice to noxious thermal and chemical stimuli. Hot-plate and tail-flick tests were applied to analyze the responses to acute thermal pain, and the formalin test was used as a model of chemically induced pain.

As heat nociceptors are known to be activated by any temperature above 43°C, mice were exposed to the hot-plate test, subjecting them to a heat stimulus of 48°C and 50°C, and the withdrawal latency at different temperatures was measured in wild-type and Etv4-deficient animals. As shown in Fig. 6A, Etv4 mutant mice exhibited significantly delayed escape latency at a plate temperature of 50°C (but not 48°C) compared with wild-type animals, indicating that Etv4 knockout mice are less sensitive to noxious heat stimuli than wild-type littermates.

We used the tail immersion test, in which the tail of each mouse is immersed in water at different temperatures and the time from onset of stimulation to rapid withdrawal of the tail from the water is recorded. Although none of the experimental groups showed differences in response to a water bath temperature of 4°C or 48°C, exposure of Etv4-deficient mice to 55°C caused a significant delay in the withdrawal of the tail from the water compared with wild-type animals (Fig. 6B). Taken together, these results suggest that Etv4-mutant animals are less sensitive to noxious heat, normally transduced by peptidergic neurons that express TrkA, but not to noxious cold, which is mediated by Ret-expressing non-peptidergic sensory neurons (Cavanaugh et al., 2009; Lindfors et al., 2006; Lippoldt et al., 2013; Wang et al., 2013).

The formalin test was then applied to examine responses to painful chemical stimuli. Both Etv4-mutant and wild-type animals showed two-phase nociceptive responses after intraplantar injection of formalin into the skin of the left hindpaw. No significant differences were detected during the early phase of the test; however, in the second phase, Etv4-deficient mice exhibited reduced pain-like responses compared with wild-type mice (Fig. 6C). Finally, the sensitivity to mechanical stimuli was studied by applying the Von Frey test. No significant differences were observed between the two experimental groups (Fig. 6D).

Overall, we conclude that Etv4-mutant animals are less sensitive to noxious heat stimuli and to the second phase of the formalin test, but show normal responses to cold and mechanical stimuli, and in the early phase of the formalin test. These findings suggest that Etv4 is required for the nociceptive responses mediated by peptidergic sensory neurons.

To gain additional insight on the mechanism through which Etv4 controls nociception, we examined the mRNA expression levels of different receptors and ion channels, including the transient receptor potential (TRP) family of cation channels Trpv1, Trpv2, Trpm8 and Trpa1, and the voltage-gated sodium channels (VGSCs) Nav1.7 (Scn9a) and Nav1.8 (Scn10a), all of which are expressed in DRG neurons and are known to contribute to the transmission of somatosensory signals (Dib-Hajj et al., 2010). Expression of these genes was analyzed in DRGs obtained from wild-type and Etv4-deficient mice at P20. Trpv1 and Trpv2 mRNA levels were significantly reduced in Etv4-deficient mice, whereas the other molecules analyzed were unaffected. Interestingly, TRPV1 and TRPV2 are known to be associated with pain induced by noxious high temperatures, whereas TRPM8 and TRPA1 detect pain induced by low temperatures (Mickle et al., 2015). No differences were observed in the Nav1.7 and Nav1.8 mRNA levels, which have been previously linked to the maintenance of inflammatory pain states (Levinson et al., 2012) between wild-type and Etv4 knockout mice (Fig. 7A).

Considering that TRPV1 is critically involved in the transduction of noxious heat and taking into account that its expression is regulated by NGF (Ji et al., 2002; Zhang et al., 2005), we analyzed possible differences in the number of TRPV1-expressing DRG neurons between Etv4-deficient and wild-type mice. A significant reduction in the proportion of TRPV1-positive DRG neurons was detected in Etv4-mutant mice (Fig. 7B).

This result raises the possibility that reduction of TRPV1-expressing neurons may result in a loss of TRPV1-dependent peripheral signaling in Etv4-mutant mice. To test this hypothesis, we performed the capsaicin test, a well-characterized model for TRPV1-dependent sensitization of peripheral nerve terminals, using capsaicin, a natural TRPV1 agonist. To this end, capsaicin was injected into the plantar surface of wild-type and Etv4-deficient mice, followed by monitoring and quantifying the time spent licking, flinching and guarding (nociceptive behaviors) per minute, for the first 5 min after injection. As shown in Fig. 7C, both wild-type and Etv4-mutant animals spent a similar amount of time developing nociception-associated behaviors during the first 2 min after capsaicin injection. However, these nociception-associated behaviors did not persist beyond 3 min after injection in Etv4-deficient animals, in contrast to controls, in which the behavior pattern was maintained for the rest of the time evaluated (Fig. 7C). Overall, Etv4-mutant mice were less sensitive to capsaicin, and this result is consistent with the decreased levels of TRPV1 expression observed in these mice.

Our findings point to an essential role of specific transcriptional programs in the pain transduction processes. In particular, they suggest that the transcription factor Etv4 plays a crucial role in the development of primary sensory afferent neurons specialized in the transduction of noxious heat and chemogenic inflammatory pain. This molecule participates in the regulation of genes relevant for the transduction of noxious stimuli in the adult. In addition, we found that, although Etv4 is dispensable for the viability of DRG peptidergic neurons, it is involved in the outgrowth and arborization of their projections into the epidermis. Consistently, in vitro assays indicate that Etv4 is required for growth cone expansion and axonal growth of NGF-dependent sensory neurons. From a mechanistic point of view, we show that this process involves the Etv4-mediated regulation of the expression of a group of ECM remodeling enzymes. Moreover, our data indicate that Etv4 is also necessary for the correct maturation of sensory nociceptive neurons expressing TRPV1, and that Etv4 ablation result in the loss of the sensory modality associated with TRPV1 expression.

Based on these data, and our previous work, we propose a model in which, during development, NGF-derived from the target tissue triggers signals that are transported retrogradely along nociceptive axons. These signals promote, during embryonic development, the expression of the Etv4 transcription factor, which in turn, regulates the expression of effector genes, including those involved in ECM remodeling, which allow an efficient target innervation, maturation of nociceptive neurons and peripheral pain transduction (Fig. 8).

Pea3 transcription factors are expressed during normal development and in oncogenic processes, characterized by extensive remodeling of the ECM (Lu et al., 2009; Oh et al., 2012; Wang and Zhang, 2009). Many studies have emphasized the key role played by Pea3 subfamily proteins in central and peripheral neurons (Chotteau-Lelièvre et al., 1997; Hagedorn et al., 2000; Paratore et al., 2002). Etv1 has been described in fast-spiking cortical interneurons (Dehorter et al., 2015), Etv4 and Etv5 in hippocampal pyramidal neurons (Fontanet et al., 2018), and Etv1 and Etv4 in spinal cord motoneurons (Arber et al., 2000; Haase et al., 2002; Helmbacher et al., 2003; Lin et al., 1998; Livet et al., 2002; Vrieseling and Arber, 2006). The three Pea3 transcription factors are expressed in different subpopulations of DRG neurons. Whereas Etv1 is found in TrkC⁺ neurons, Etv4 and Etv5 are reported to be expressed in TrkA⁺ and Ret⁺ sensory neurons, and their expression in TrkB⁺ cells is still unclear (Arber et al., 2000; Fontanet et al., 2013; Patel et al., 2000). These transcription factors are induced by the neurotrophic factors corresponding to each Trk receptor: Etv1 is upregulated in response to NT3 in proprioceptive neurons, whereas Etv4 and Etv5 are induced in response to NGF in nociceptive cells (Fontanet et al., 2013; Patel et al., 2003). Previously, we have shown that the reduction of Etv4 expression through the use of specific shRNA sequences inhibits axonal growth of sensory neurons in response to NGF, whereas its overexpression is sufficient to induce axonal growth in response to this neurotrophin (Fontanet et al., 2013). In the present work, we extend these findings, showing that peptidergic DRG and trigeminal sensory neurons from Etv4-mutant mice do not respond adequately to NGF.

Despite several studies analyzing the contribution of Etv4 in the CNS, its physiological role in the development of the PNS still remains unknown. Previous evidence has demonstrated that targeting a transcription factor resulting from the fusion of Ewing sarcoma (Ewsr1) with the Etv4 DNA-binding domain to the Etv1 locus results in its ectopic and precocious expression in postmitotic DRG proprioceptive sensory neurons (which normally do not express Etv4, but express Etv1) and leads to perturbations of axonal projections, the acquisition of differentiation markers and a switch in their dependence on neurotrophic support. This indicates that the precise timing and location of ETS transcription factor signaling is essential for normal sensory neuron development (Hippenmeyer et al., 2005). Later, we described that Etv4, when expressed in embryonic DRG peptidergic neurons, plays a role in sensory axonal growth in response to NGF (Fontanet et al., 2013). However, the physiological contribution of this NGF/Etv4 pathway in peptidergic neuron development was not analyzed. Here, we evaluated the physiological consequences of Etv4 depletion in the development of nociceptive sensory neurons. We show that it is necessary for the expression of molecules induced by NGF, which behave as effectors of sensory axonal growth during embryonic development, and is required for correct peptidergic epidermal innervation and pain transduction.

In contrast, Etv4 deficiency did not affect the epidermal innervation of GFRα1- and GFRα2-positive fibers or the nociceptive cold behavior, both of which are mediated by non-peptidergic Ret-positive sensory fibers. Many transcriptional factors have been described as mediators of NGF function in DRG neurons. Although most of them are associated with the survival and maturation of TrkA-expressing sensory neurons (Huang et al., 2015; Lonze et al., 2002), few have been described as being involved in neurite outgrowth and target innervation independently of survival (Harrington and Ginty, 2013). Among them, the serum response factor, SRF, acting downstream of NGF signaling, has been shown to promote in vivo axonal growth and target innervation of embryonic DRG sensory neurons, by regulating key cytoskeletal genes (Wickramasinghe et al., 2008).

Previously, we have shown that Etv4 is expressed at high levels in ∼80% of the TrkA⁺ DRG neurons during the critical developmental period of peripheral innervation. Moreover, we have described that the expression of Etv4 in sensory neurons is induced in response to NGF, functioning as a critical mediator of NGF signaling involved in sensory neurite outgrowth (Fontanet et al., 2013). Here, we extend these observations showing that DRG peptidergic neurons in Etv4-null mice have deficient axonal growth and target innervation. Our results indicate that Etv4 is required for the expression of adequate levels of metalloproteinases – such as MMP3, MMP10 and MMP13 – and UPAR, a receptor involved in ECM remodeling and DRG neurite outgrowth, in response to NGF (Farias-Eisner et al., 2000; Fontanet et al., 2013; Sanz et al., 2017). The promoters of Mmp3, Mmp10 and Mmp13 are characterized by the presence of a binding site for Pea3 factors. The presence of these sites in many Mmp promoters makes these genes responsive to different trophic factors (Yan and Boyd, 2007). On the other hand, UPAR is also an important regulator of ECM remodeling and cell signaling. By binding the serine protease urokinase-type plasminogen activator (uPA), UPAR activates plasminogen, generating plasmin, which cleaves different ECM components and activates MMPs. In addition, UPAR is also a signaling receptor that promotes cell motility and invasion (Smith and Marshall, 2010).

Different subpopulations of nociceptors are responsible for the encoding of specific nociceptive stimuli. Peptidergic nociceptors have been involved in the transduction of pain induced by thermal and inflammatory stimuli. Our results indicate that Etv4 deletion, although not affecting the number of TrkA⁺ sensory neurons, critically perturbs pain transduction by altering the development of an adequate functional modality of peptidergic epidermal innervation.

At adult stages, peptidergic and non-peptidergic sensory neurons can also be distinguished by the expression of a combination of ion channels involved in the transduction of different external stimuli into electrical activity. Although peptidergic neurons have been associated primarily with heat sensitivity, non-peptidergic neurons play a role in cold detection. The transient receptor potential (TRP) family of ion channels are the main sensors of temperature, cold or heat, and are expressed in different DRG subpopulations (Wang and Siemens, 2015). Among them, TRPV1 has been identified as the molecular target of capsaicin (the main active component in hot peppers) and is predominantly expressed in peptidergic neurons (Caterina et al., 1997). Genetic ablation of TRPV1 in mice eliminates capsaicin responsiveness and results in an altered reaction to noxious heat stimuli (Caterina et al., 2000). In humans, pharmacological inhibition of TRPV1 reduces their ability to detect painful heat stimuli (Gavva et al., 2008; Rowbotham et al., 2011; Szallasi and Sheta, 2012). When TRPV1 is sensitized in the context of inflammation, it also plays a dominant role in heat detection (Amaya et al., 2004; Caterina et al., 2000, 1997; Davis et al., 2000; Huang et al., 2006). Moreover, activation of TRPV1 by intradermally injected capsaicin was shown to induce neurogenic inflammation by promoting the antidromic release of CGRP and substance P (Lin et al., 2007). Furthermore, different studies indicate that the expression, trafficking and phosphorylation of TRPV1 are controlled by the activity of NGF/TrkA (Amaya et al., 2004; Ji et al., 2002; Xue et al., 2007; Zhang et al., 2005). In our study, we found reduced expression levels of Trpv1 mRNA in DRGs of Etv4^−/− animals. Nevertheless, it is not clear how Etv4 would regulate the expression of TRPV1 as Pea3-binding sites have not been described in the promotor of Trpv1 gene. Interestingly, the transcription factor Sp1 was found to directly regulate TRPV1 in response to NGF, and Etv4 is known to be an interactor of Sp1 (Chu et al., 2011; Jiang et al., 2007). Thus, Etv4 could induce TRPV1 through the formation of a complex with Sp1. Alternatively, Etv4 may induce TRPV1 expression in an indirect manner, by activating expression of other transcription factors acting as relays.

In summary, our data demonstrate that genetic ablation of Etv4 results in reduced responses to noxious heat, formalin and capsaicin, and suggest that these behavioral phenotypes are the consequence of the lack of a correct peptidergic innervation and/or reduced expression of TRPV1 channels. Although Etv4 is not exclusively expressed in TrkA-positive peptidergic neurons, the absence of behavioral differences between wild-type and Etv4-mutant mice exposed to cold or mechanical stimuli, suggests that Etv4 is dispensable for the function of both cold- and mechanoreceptor-sensing nociceptive neurons, characterized by the expression of Ret and TrkB receptors. Indeed, the fact that both the expression of the Trpv8 cold-sensing channel and the non-peptidergic cutaneous innervation are preserved in Etv4-null mice additionally supports the idea that Etv4 is not critically involved in the non-peptidergic nociceptive function analyzed here. Taken together, our findings are in accordance with a model in which Etv4 is required for target-secreted NGF-dependent mechanisms that regulate peptidergic axonal growth and the correct maturation of TrkA⁺ nociceptive neurons. Therefore, these results provide a key insight into the transcriptional program by which growth factors regulate axonal growth, innervation and transduction of sensory pain. Understanding the molecular basis of the perception of noxious stimuli by sensory neurons is essential for the design of new therapeutic treatments for pain.

PC12 cells were grown in DMEM in the presence of 5% horse serum and 10% FBS (Invitrogen) (Shirazi Fard et al., 2010). Etv4-shRNA, HA-tagged-MMP3 and Flag-tagged-MMP13 vectors were purchased from Cellogenetics. The retroviral vector pRetro-U6G shRNA was used for expression of Etv4-shRNA (Fontanet et al., 2013, 2018). NGF was purchased from Promega; and the matrix metalloproteinase inhibitor GM6001 was purchased from Calbiochem.

Etv4^−/− mice were kindly provided by Dr Olivia Bermingham-McDonogh (University of Washington, Seattle, USA). The Etv4-mutant mice have been described in detail by Livet et al. (2002). The transgenic strain was genotyped using PCR-based strategy. The use of animals was approved by the Animal Care and Use Committee (CICUAL) of the School of Medicine, University of Buenos Aires, Instituto Leloir and Austral University, according to the Principles for Biomedical Research involving animals of the Council for International Organizations for Medical Sciences and according to previsions stated in the Guide for the Care and Use of Laboratory Animals.

DRG neurons obtained from Etv4^+/+ and Etv4^−/− mice at embryonic (E) day 14.5 were prepared as previously described (Baudet et al., 2000). Briefly, the ganglia were dissociated with collagenase (2% w/v, Sigma-Aldrich), trypsin (0.1% w/v, Invitrogen) and DNAse I (10 µg/ml, Invitrogen). For cranial sensory neuron culture, trigeminal ganglia (TG) from Etv4^+/+ and Etv4^−/− mice at embryonic day 15.5 were collected and dissociated using trypsin (0.05% w/v, Invitrogen) for 30 min. Cells were seeded onto poly-ornithine-coated (0.5 mg/ml, Sigma-Aldrich) and laminin-coated (10 μg/ml, Sigma-Aldrich) 24-well plates at a density of 15,000 cells/well and were incubated for 24-36 h. Neurons were maintained in DMEM:F12 (1:1) medium supplemented with 60 mg/ml penicillin, 100 mg/ml streptomycin, 2 mM glutamine (Invitrogen) and 1 mg/ml BSA (Sigma-Aldrich) (Sensory neuron media). DRG and TG neurons were maintained with NGF (Promega) (50 ng/ml and 10 ng/ml, respectively).

Primary cultures of DRG or TG neurons were performed as previously described (see above). Cells were fixed with 4% PFA and stained using anti-βIII-tubulin antibodies to identify neurites (mouse anti-βIII-tubulin; G7121, clone 5G8, 1/5000, Promega) or with 4% PFA in the presence of 4% sucrose and stained using anti-βIII-tubulin antibodies and phalloidin (ThermoFisher) to analyze growth cone area. Neuronal survival was evaluated using the nuclear stain 4′,6-diamidino-2-phenylindole dihydrochloride (DAPI, Sigma-Aldrich). Neurons containing fragmented or condensed nuclear staining were scored as apoptotic cells and not computed in the differentiation assays. Neuron viability was quantified as the number of live neurons in relation to the total amount of neurons (live and apoptotic cells). For DRG dissociated neurons, images were acquired using a Zeiss Observer microscope. For TG dissociated neurons, images were acquired using an Olympus IX-81 inverted microscope. For growth cone analysis, images were acquired using a Zeiss LSM710 NLO confocal microscope (Plan-APOCHROMAT 63×, a.4 N/A). Images were analyzed using the ImageJ software.

Outgrowth assays in the presence of the pan-metalloproteinase inhibitor GM6001 (Calbiochem) were carried out using 5 µm of the inhibitor. The cells were cultured for 30 h and then maintained for 6 h in the presence of GM6001. Cells were fixed with 4% PFA supplemented with 4% and sucrose stained using anti-βIII-tubulin antibodies and phalloidin. All the experiments were carried out independently at least with three mice/condition or genotype in duplicate.

For outgrowth recovery assays with exogenous MMP3 and MMP13, the HEK-293T cell line was transfected using polyethylenimine (PEI from Polyscience) with control or HA-tagged-MMP3 and Flag-tagged-MMP13 plasmids, and incubated in sensory neuron media. After 2 days, the conditionate medium containing MMP3 and MMP13 was collected, followed by centrifugation at 20,000×g for 20 min. DRG neurons obtained from Etv4^+/+ and Etv4^−/− mice at embryonic (E) day 14.5 were cultured for 36 h in presence of control or MMP3/MMP13-containing media.

PC12 cells were grown in DMEM supplemented with 5% horse serum and 10% FBS (Invitrogen). Cells were transfected using X-treme GENE (Roche) following the manufacturer's instructions. For downregulation of Etv4, PC12 cells were transfected with control or Etv4-shRNA-containing vector and selected with puromycin to enrich in the transfected cells. PC12 Etv4-shRNA clones were generated by isolation of colonies transfected with the construct and selected with puromycin as described previously (Fontanet et al., 2013). PC12 control and PC12-Etv4-shRNA were stimulated for 8 h with NGF (50 ng/ml).

DRG cultures from Etv4^+/+ and Etv4^−/− mice at embryonic day (E) 14.5 were prepared as previously described (De Vincenti et al., 2021). Neurons were maintained with NGF 20 ng/ml for 44 h, serum-starved for 12 h and treated with or without NGF (50 ng/ml for 8 h).

For immunofluorescence assays of DRG and spinal cord, mice at the indicated developmental stages were euthanized and perfused transcardially with 4% paraformaldehyde (PFA) in PBS under deep anesthesia. The lumbar section (L1-L4) containing the DRG and spinal cord was dissected and post-fixed overnight in 4% PFA. DRGs and spinal cord immunostaining were carried out on 20 µm cryostat transverse sections on glass at the different developmental stages. Tissue sections were permeabilized for 30 min in PBS containing 0.3% Triton X-100, washed with PBS and then blocked for 1 h with 5% donkey normal serum (DNS) (Jackson ImmunoResearch). Antibodies were applied in PBS containing 0.3% Triton X-100 and 3% DNS. For double labeling with the neuronal marker NeuN and TrkA or TrpV1, a serial protocol was performed. Sections were first permeabilized for 1 h in PBS containing 1.5% Triton X-100 and incubated with NeuN antibody for 48 h followed by TrkA or TrpV1 antibody incubation overnight. Secondary antibodies were from Jackson ImmunoResearch: Cy2-anti-goat (705-225-147), Cy3-anti-goat (705-165-147), Cy2-anti-mouse (715-225-151), Cy3-anti-mouse (715-165-151), Cy2-anti-rabbit (711-225-152), Cy3-anti-rabbit (711-165-152). All secondary antibodies were diluted 1/300.

For neuron quantification or central peptidergic projections quantification assay, images were obtained using an Olympus IX83 DSU 20X objective at a resolution of 2048×2048 pixels (16 bits) using identical settings between control and experimental images with no saturation. Each image corresponds to a merge of 15 optical sections of 0.9 µm each. The number of CGRP⁺ fibers in the dorsal horn was measured as the total number of fibers that extend through the lamina I/II into the medial zone of the dorsal horn/section (Hancock et al., 2011). For CGRP intensity levels, corrected CGRP fiber fluorescence was calculated for each dorsal horn of the spinal cord section using a hand-drawn ROI. The intensity level was calculated as the integrated density (IntDen) of the selected dorsal horn area minus the area of the selected dorsal horn times the background (Bck) reading for each image [CGRP intensity level=IntDen – (Area×Bck)]. All images were analyzed using ImageJ software.

For peripheral innervation analysis, 2-month-old mice were euthanized and perfused transcardially with 4% PFA in PBS under deep anesthesia. Glabrous skin from the hindpaw of newborn or 2-month-old mice was sectioned on glass at 20 µm. For peptidergic innervation, images were obtained using an Olympus IX83 DSU 20× objective at a resolution of 2048×2048 pixels (16 bits) using identical settings between control and experimental images with no saturation. Each image corresponds to a merge of 15 optical sections of 0.9 µm each. For non-peptidergic innervation, images were obtained using a Zeiss LSM 880 Airyscan 20× objective (0.8 NA). Each image corresponds to a merge of 12 optical sections of 0.9 µm each. Only fibers crossing the dermis-epidermis line were analyzed on segments of 500 µm of skin using ImageJ software.

Immunofluorescence was performed using the following antibodies: mouse anti-Etv4 (sc-166629, 1/200; Santa Cruz Biotechnology), rabbit anti-NeuN (D4G40, 1/600; Cell Signaling), rat anti-TrkA (AF1056, 1/800; R&D Systems), rabbit anti-TRPV1 (ACC-030, 1/200; Alomone labs), rabbit anti-CGRP (calcitonin gene-related peptide; PC205L, 1/1000, Sigma-Aldrich), rabbit anti-PGP9.5 (NE1013, 1/1000, Calbiochem), goat polyclonal anti-GFRα1 (AF560, 1/200, R&D Systems) and goat polyclonal anti-GFRα2 (AF429, 1/200, R&D Systems).

For mRNA analysis, total RNA was isolated from lumbar DRG (L1-L6) or DRG cultures, using RNA columns (NucleoSpin RNA) according to manufacturer's instructions and cDNA was synthesized using M-MLV reverse transcriptase (Invitrogen) and random hexamers (Applied Biosystems). The cDNA was amplified using the following primer sets: mouse Etv4 forward, 5′-CAG CCC TTT TCC AGG GCA GAA C-3′; mouse Etv4 reverse, 5′-GCT GGA AGA CGG AGC TGT GCT-3′; mouse TrkA forward, 5′-CTG CCA GGA CAA ACA GAA CAC-3′; mouse TrkA reverse, 5′-GTG GTT GGC TTC GTC TGA GTA-3′; mouse Ret forward, 5′-TGA AGA AAA GCA AGG GCC GG-3′; mouse Ret reverse, 5′-ACA ATC TCC CAG AGC AGC AC-3′; mouse GFRα1 forward, 5′-GAC CGT CTG GAC TGT GTG AAA G-3′; mouse GFRα1 reverse, 5′-TTA GTG TGC GGT ACT TGG TGC T-3′; rat Etv4 forward, 5′-ACT CAG ATG TCC CTG GAT GTG-3′; rat Etv4 reverse, 5′-TCA CTC CAT CAC CTG AAG AGG-3′; rat Mmp3 forward, 5′-GCT ATT CTG GCC ACC TTC TTC-3′; rat Mmp3 reverse, 5′-TCA TCA TCA CCA TAG GGA AGG-3′; rat Mmp13 forward, 5′-GAC AGA TTC TTC TGG CGT CTG-3′; rat Mmp13 reverse, 5′-GAA GTT CTG GCC AAA AGG ACT-3′; mouse TrpV1 forward, 5′-GAC GGC AAG GAT GAC TTC CG-3′; mouse TrpV1 reverse, 5′-AGT TGC CTG GGT CCT CGT T-3′; mouse Mmp3 forward, 5′-TAA AGA CAG GCA CTT TTG GCG-3′; mouse Mmp3 reverse, 5′-TTA GTA ACC TCA TAT GCA GCA TCC A-3′; mouse Mmp13 forward, 5′-TGA TGC CAT TAC CAG TCT CC-3′; mouse Mmp13 reverse, 5′-ACA TGG TTG GGA AGT TCT GG-3′; rat and mouse Mmp10 forward, 5′-AGG AAG TCA GTT CTG GGC AG-3′; rat and mouse Mmp10 reverse, 5′-GCA TCA ATC TTC TTC ACG GTG G-3′; rat and mouse Upar forward, 5′-AAT GGT GGC CCA GTT CTG GA-3′; reverse rat and mouse Upar, 5′-CCC AGC ACA TCT AAG CCT GT-3′; mouse Trpv2 forward, 5′-TTA CTG GGC CAG CTG TGG TA-3′; mouse Trpv2 reverse, 5′-CCT GGG ACA GCA CTG TGA G-3′; Trpm8 forward, 5′-AAG AAG TGT TTC AAA TGC TG-3′; Trpm8 reverse, 5′-AAT TCT CCT TCA TGA CAC CC-3′; Trpa1 forward, 5′-ATG CAA GAA ACA CGA CAA GA-3′; Trpa1 reverse, 5′-TGA GCT CAT GCT GCT TTT AA-3′; NaV1.7 forward, 5′-AAG CTT GGA GGT CAA GAT AT-3′; NaV1.7 reverse, 5′-TCA AAA GCT TGG TTT GTC ACT-3′; NaV1.8 forward, 5′-CAT GAA GAA GCT GGG CTC CA-3′; NaV1.8 reverse, 5′-TGA TGT CAA ATG CTT GCC TGG-3′; rat Tata binding protein (Tbp) forward, 5′-GGG GAG CTG TGA TGT GAA GT-3′; rat Tata binding protein (Tbp) reverse, 5′-CCA GGA AAT AAT TCT GGC TCA-3′; mouse Gapdh forward, 5′-TGT TCC TAC CCC CAA TGT GT-3′; mouse Gapdh reverse, 5′-AGG AGA CAA CCT GGT CCT CA-3′.

HEK-193 cells were lysed at 4°C in lysis buffer containing 0.5% Triton X-100, 0.1% SDS plus protease and phosphatase inhibitors. Protein lysates were clarified by centrifugation and analyzed by immunoblotting as previously described (Fontanet et al., 2013). The blots were scanned in a Storm 845 PhosphorImager (GE Healthcare Life Sciences). The following antibodies were used: rat monoclonal anti-HA (3F10, 1/1000, Roche), rabbit polyclonal anti-MMP13 (sc-30073, 1/500, Santa Cruz Biotechnology) and mouse anti-α-tubulin (T9026, Sigma, 1/6000).

Mice were acclimatized to the different behavioral testing apparatus at least 2 h for 2-3 days before the testing day. To measure heat pain, mice were placed within a Plexiglass observation chamber surrounding a hot plate, and the latency to hindpaw flicking, licking or jumping was measured (using a stopwatch). The hot plate was set at two different temperatures, 48°C and 50°C, and all animals were tested sequentially at each temperature with 90 min between test. For the tail immersion test, the water bath was set at three different temperatures, 4°C, 48°C and 55°C, controlled by a digital thermometer. Animals were gently restrained using a restrainer tube, presented previously in the habituation session in order to reduce stress. The tail withdrawal latency was measured (using a stopwatch) by immersing the terminal 3 cm of tail into the water bath. Mice were tested three times at each bath temperature every 2 min with a 90 min interval between different temperatures.

To measure responses to chemical nociceptive stimuli, mice were placed in Plexiglass cubicles, and after 30 min habituation they received an intraplantar injection of 2.5% formalin (30 µl). The amount of time spent licking, flinching and guarding was measured in phases I and II of the formalin test. Phase I included observation for the first 5 min after formalin injection; phase II ran from 10 min to 45 min after formalin injection (Bannon and Malmberg, 2007).

To measure capsaicin-evoked pain, mice were given a 3 µg/10 µl injection of capsaicin (Sigma-Aldrich) in the intraplantar hind leg. The time spent licking, flinching and guarding was measured every 1 min for a period of 5 min after capsaicin injection (Seal et al., 2009).

To measure responses to mechanical stimuli, animals were placed into Plexiglass domes on an elevated wire grid. Von Frey nylon monofilaments of different bending forces (0.16, 0.4, 0.6 and 1 g; Stoelting) were used to stimulate the center of the plantar surface of hind legs, following the up-and-down method of Dixon, as described previously (Chaplan et al., 1994). Quantitative assessment of tactile allodynia in the mouse paw was performed to establish the 50% withdrawal (g) threshold. Results were obtained using the formula: 50% g threshold=(10 Xf+kδ)/10,000 [where Xf=value (in log units) of the final von Frey hair used; k=tabular value for the pattern of positive/negative responses; and δ=mean difference (in log units) between stimuli]. For all behavioral experiments, only 2-month-old male mice were used to avoid sex difference in nociception.

Data were analyzed using GraphPad Prism 8.0 and are expressed as mean±s.e.m. Significance was accepted at P<0.05. No statistical method was performed in order to define sample size, but our sample sizes were similar to those used in the field and the n of each experiment is indicated in the figure legends. For animal studies, the handling of the data was performed in a blinded manner. The selection of the mice was unbiased in terms of size and weight. Statistical analysis was performed in GraphPad Prism 8.0. The normal distribution of the variable was evaluated with the Shapiro-Wilk test. Only for small datasets was normality was assumed. For the statistical analysis of the data, two-tailed Student's t-test, one-way ANOVA or two-way ANOVA analysis followed by a respective post-hoc test was preformed; the type is indicated in figure legends.

We thank Dr O. Bermingham McDonogh (University of Washington, Seattle, USA) for sharing transgenic mice; N. Villalba, A. Rossi and C. Pascuale for technical assistance with microscopy; M. Ponce and Maria Jimena Afonso for animal care; and Innova-t, UBATEC and the Fundación Instituto Leloir for research grant administration.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200583