Ablation of sensory nerves favours melanoma progression

Abstract The tumour mass is composed not only of heterogeneous neoplastic cells, but also a variety of other components that may affect cancer cells behaviour. The lack of detailed knowledge about all the constituents of the tumour microenvironment restricts the design of effective treatments. Nerves have been reported to contribute to the growth and maintenance of numerous tissues. The effects of sensory innervations on tumour growth remain unclear. Here, by using state‐of‐the‐art techniques, including Cre/loxP technologies, confocal microscopy, in vivo‐tracing and chemical denervation, we revealed the presence of sensory nerves infiltrating within the melanoma microenvironment, and affecting cancer progression. Strikingly, melanoma growth in vivo was accelerated following genetic ablation or chemical denervation of sensory nerves. In humans, a retrospective analysis of melanoma patients revealed that increased expression of genes related to sensory nerves in tumours was associated with better clinical outcomes. These findings suggest that sensory innervations counteract melanoma progression. The emerging knowledge from this research provides a novel target in the tumour microenvironment for therapeutic benefit in cancer patients.


| INTRODUC TI ON
Melanoma exhibits one of the most aggressive behaviours among cancers. 1 It affects millions of people in the world. 2 Promising therapeutic strategies have been refined in recent years, nevertheless the 5-year overall survival of patients with metastatic cutaneous melanoma remains between 5% and 19%. 3,4 The disease is commonly originated as a consequence of the stepwise agglomeration of genetic and epigenetic alterations in melanocytes; nonetheless the tumour microenvironment plays a dynamic role in regulating the subsequent tumour growth. 5 Physiologically, the skin microenvironment in healthy people is a physical and chemical barrier that protects from tumorigenesis; nevertheless, cancer cells evoke numerous changes to transform the adjacent normal cells into pathological entities. 6 The orchestration of such events implicates expansion, migration and contribution of various cell types. 7 The cellular composition of the tumour microenvironment is heterogeneous and yet not fully uncovered. While some components of this microenvironment promote, [8][9][10] others limit tumour progression. [11][12][13] Given the complexity and plasticity of the cancer-associated cells, further studies are necessary in order to develop more effective therapies.
The skin is densely innervated, and nerves have been reported to contribute to its growth, functionality and maintenance. 14 It has been recently reported that the two branches of the autonomic nervous system regulate cancer progression in different organs by controlling cancer initiation, progression and metastasis. [15][16][17] Encouragingly, clinical studies describe positive effects of drugs that interfere with the autonomic nervous system in melanoma patients. 18 In addition to autonomic nerves, the skin is also innervated by sparsely distributed sensory fibers, most of which express the voltage-gated sodium (Nav) channel Nav1.8. 19 The Nav1.8 channel can be specifically used as a molecular marker for sensory nerves, and has been targeted in research as a way to manipulate these nerves and study them within the skin as well as other tissues. [20][21][22] Interestingly, in vitro co-culture systems using sensory neurons from the dorsal root ganglion have suggested that nerve interactions may affect cancer cell proliferation. 23 Sensory nerves can eventually contribute to tumour-associated pain 24,25 ; yet, whether sensory nerve fibers are involved in tumour progression in vivo is unclear. Here, we have detected the presence of sensory innervations in the melanoma microenvironment and have tested the hypothesis that sensory nerves affect melanoma behaviour by evaluating the effect of genetic and pharmacologic ablations of these nerves.

| In vivo analyses of tumour growth
For tumour growth, 8-to 10-week-old mice were injected with 5 × 10 4 B16F10 cells subcutaneously in the right flank. Tumours were removed 16 days after injection and weighted. Length (L) and width (W) were measured for calculating tumour volume (V) using the formula V = 0.5 × (L × W 2 ). 28 Tumour area was determined using calibrated photographs of each tumour using Fiji software ® , version 1.53 (National Institute of Health, Bethesda, MD).

| RTX treatment
For chemical ablation of sensory nerves, WT mice were treated with resiniferatoxin (RTX, Sigma) as previously described. 27 To perform ablation of sensory nerves before tumour implantation, 4-week-old mice were injected subcutaneously on consecutive days with increasing doses of RTX (30, 70 and 100 μg/kg) dissolved in 2% DMSO with 0.15% Tween 80 in PBS. Mice rested for 20 days before injection of B16F10 cells. For depletion of sensory nerves after tumour implantation, mice were injected subcutaneously with B16F10 cells and, after 48 hours, underwent treatment with increasing doses of RTX (30, 70 and 100 μg/kg). For both conditions, control mice were injected with vehicle alone. To access sensory depletion efficiency, mice were subjected to a behavioural test to measure the sensitivity to capsaicin, confirming the ablation of sensory nerves, as previously described. 29 After intra-plantar injection of capsaicin (3 μg in 20 μL), total licking time was recorded in a 5-minute interval.

| TUNEL assay
For analysing DNA fragmentation, tumour fragments were embedded in paraffin and 5 μm sections were used for TUNEL staining according to the manufacturer's protocol. 30 Sections were deparaffinized, rehydrated and maintained in PBS. Permeabilization was performed for 20 minutes using Proteinase K (0.2 mg/mL in TBS). Endogenous peroxidases were inactivated using H 2 O 2 (3% diluted in methanol).
Sections were then labelled with terminal deoxynucleotidyl transferase (TdT) for one and a half hour and the reaction was terminated.
Detection was performed using the conjugate and DAB provided by the manufacturer. Methyl green was used as counterstain and sections were analysed using a light microscope. Multiple fields of each section were analysed using Fiji software ® , version 1.53 (National Institute of Health) to quantify the number of positive cells per area.

| Kaplan-Meier curves
Survival curves were plotted using the website R2: Genomics Analysis and Visualization Platform (http://r2.amc.nl). Genes commonly expressed by sensory nerves (Nav1.8, TRVP1 and VIP) were selected and the parameters chosen following the protocol given by the website.
Shapiro-Wilk normality test was performed, and unpaired t test was used to determine statistical significance.

| Nav1.8+ sensory nerves infiltrate the tumour microenvrionment
To examine whether sensory innervations are present within the tumour microenvironment, we have analysed an orthotopic melanoma tumour mouse model in which only sensory nerves are labelled with the red fluorophore TdTomato. We have crossed Nav1.8-Cre mice with a mouse line conditionally expressing TdTomato. 21 In Nav1.8-Cre/TdTomato mice, upon removal of loxP-

| Genetic depletion of Nav1.8+ sensory nerves enhances melanoma growth and tumoural angiogenesis
To explore the role of Nav1.8+ sensory nerves within the tumour microenvironment, we have induced targeted diphtheria toxinbased cell ablation. 31 We crossed Nav1.8-Cre mice with inducible diphtheria toxin A (iDTA) transgenic mice to specifically deplete all sensory neurons. 26 Nav1.8-Cre/iDTA mice were previously shown to be devoid of all Nav1.8-expressing nociceptors, 32 and have no response to mechanical stimuli, noxious heat or capsaicin. 22 Genetic depletion of sensory nerves was confirmed by immunohistochemistry to Nav1.8 in the dorsal root ganglions of these animals ( Figure A2A,B). We have analysed the growth of B16F10 melanoma cells subcutaneously injected into Nav1.8-Cre/ iDTA mice (genetically depleted of Nav1.8+ sensory nerves; Figure 2A). These experiments revealed that after 16 days, melanoma size increased in the absence of Nav1.8+ sensory nerves (tumour weight increased from 1.05 ± 0.17 to 2.21 ± 0.11 g; tumour weight per body weight increased from 0.06 ± 0.004 to 0.11 ± 0.005 g; tumour area increased from 2.77 ± 0.09 to 4.61 ± 0.10 cm 2 , tumour volume increased from 1980 ± 264 to 3559 ± 76 mm 3 ; Figure 2B-F). Also there was an enhancement in the intra-tumoural blood vessels' area (from 2.8 ± 0.3 to 4.6 ± 0.8 µm 2 ), diameter (from 12.3 ± 0.7 to 21.4 ± 1.7 µm) and length (from 0.12 ± 0.01 to 0.21 ± 0.02 mm/µm 2 ; Figure 2G,H). Additionally, tumour cells death decreased in the mice without sensory nerves (from 0.67 ± 0.7 to 1.01 ± 0.11 cells per mm 2 ; Figure 2I,J). Genetic depletion of sensory nerves also led to an increase in proliferating cells within the tumour (from 26.36 ± 3.098 to 43.83 ± 3.394 per cent of cells within the tumour; Figure A3A-C). Animal weights and blood counts were not affected by genetic ablation of sensory nerves in melanoma-bearing mice (data not shown). Our results indicate that sensory nerves within the tumour microenvironment are trying to inhibit cancer progression.

| Chemical depletion of sensory nerves before cancer cells implantation enhances melanoma growth and intra-tumoural blood vessel formation
We sought an alternative method, for comparison, to confirm the role of sensory nerves in the melanoma microenvironment. Thus, we also achieved sensory denervation by treating wild-type mice with resiniferatoxin (RTX), a capsaicin analogue. 33 Pharmacologic depletion of sensory nerves was confirmed by immunohistochemistry to Nav1.8 in the dorsal root ganglions of these animals ( Figure A2C, D). After pre-treatment with 3 consecutive doses of RTX (30, 70 and 100 μg/kg), followed by melanoma cells transplantation, we analysed the tumour growth ( Figure 3A,B).
Additionally, there was an increase in the intra-tumoural blood vessels' area (from 8.2 ± 1.9 to 14.5 ± 1.1 µm 2 ) and length (from 1.60 ± 0.25 to 2.75 ± 0.43 mm/µm 2 ; Figure 3G,H). Moreover, tumour cells death decreased in RTX-treated mice (from 0.54 ± 0.04 to 0.21 ± 0.03 cells per mm 2 ; Figure 3I,J). Chemical depletion of sensory nerves before cancer cells transplantation also led to an increase in proliferating cells within the tumour (from 28.12 ± 3.84 to 44.73 ± 3.808 per cent of cells within the tumour; Figure A3D-F). Animal weights and blood counts were not affected by sensory nerves depletion (data not shown). These results were similar to the ones achieved by genetic ablation of Nav1.8+ sensory nerves.
Together, our results support the idea that sensory nerves try to block melanoma progression. F I G U R E 1 Sensory Nav1.8+ nerve fibers are present within the melanoma tumour microenvironment. A, Schematic representation for subcutaneous allograft melanoma growth. 5 × 10 4 B16F10 melanoma cells were subcutaneously injected into Nav1.8-Cre/TdTomato mice, and tumour tissues were surgically removed 16 days later. B, Whole melanoma tumour viewed from a fluorescent dissecting microscope showing Nav1.8 + sensory nerves labelled with TdTomato fluorescence (red) and brightfield images. C, Representative image of a Nav1

| D ISCUSS I ON
Here, we show the presence and importance of sensory nerves within the melanoma microenvironment. The sensory innervations are not inert within the tumours, but rather participate actively during tumour progression, as sensory nerve loss in a model of genetic depletion of Nav1.8+ sensory fibers or chemical depletion using RTX can induce changes in melanoma growth in vivo. Our results show that the sensory denervation leads to worse outcomes in melanoma-bearing mice ( Figure 6). Remarkably, we show that low expression of genes related to sensory nerves correlates with worse outcomes also in human melanoma patients, suggesting that sensory nerves try to block cancer growth.
While several cellular and molecular mechanisms have been proposed to affect cancer progression, the identity and role of all tumour microenvironment components remain to be defined. The present results highlight a key constituent in the complex tumour microenvironment: sensory nerves (Figure 7). Previous studies have used chemical denervation of sensory nerves, by using capsaicin, nevertheless, given the possible broad unspecific effects of this drug, the question whether sensory nerves play a role in tumour progression remained open. Here, we proved by specific genetic ablation of sensory nerves that sensory innervation affects melanoma progression. Interestingly, our results are different from some recent studies that have shown capsaicin reducing cancer progression of tumour-bearing mice, in models of pancreatic 41 and prostate cancer. 42 We show here that this is probably due to the moment of drug injection. As when we administered RTX after cancer cells implantation, our results also suggested that this drug inhibits tumour growth. Various previous studies have shown the direct effect of capsaicin on melanoma cells in vitro, inhibiting proliferation and migration, and increasing the death of these cells. 35,[43][44][45][46] Thus, when capsaicin, or its analog RTX, is administered after cancer cells implantation, it is already expected that tumour cells will be affected, and the role of sensory denervation cannot be analysed in this model without the interference of capsaicin' direct effects on the tumour cells. On the other hand, sensory nerves' genetic ablation or the treatment of mice with RTX before the cancer cells transplantation allowed us to analyse more specifically the role of sensory nerves in the tumour microenvironment in vivo.
Sensory nerves can contribute to tumour-associated pain as demonstrated in pancreatic 24 and prostate cancers. 25 In vitro co-culture systems using sensory neurons from the dorsal root ganglion have suggested that these nerves may interact with cancer cells. 23  Developmental studies have provided evidence that ingrowth of sensory nerves precedes arterial blood vessel formation, which follows axons branching pattern in the embryonic skin. 53,54 However, it was not studied whether sensory nerves affect tumoural new blood vessel formation (angiogenesis). Our results indicate that sensory nerves inhibit tumoural angiogenesis within the melanoma.
It remains unexplored how this happens, is this a direct or indirect effect of sensory nerves? Are sensory nerves affecting endothelial cells, pericytes or both? Sensory nerves endings can release neuropeptides, including substance P (SP), VIP, tachykinins, calcitonin gene-related peptide (CGRP) and others. 55 In this context, it will be interesting to discover which key signalling molecules are responsible for this phenotype in the tumour microenvironment. Distinct peripheral nerves have been shown to be present in the tumour microenvironment of various organs, being implicated as regulators of cancer progression. Most studies suggest a pro-tumorigenic neural role. The two branches of the autonomic nervous system have been shown to regulate prostate cancer progression: sympathetic adrenergic nerves are required for cancer initiation, while parasympathetic cholinergic fibers promote cancer metastasis. 15,16 Additionally, adrenergic and cholinergic signalling have been F I G U R E 5 Low expression of sensory nerve-related genes correlates with worse prognosis in melanoma. High expression of genes expressed in sensory nerves correlates with best outcomes in patients with melanoma. The prognostic impact of sensory nerve-related genes in melanoma patients was evaluated using the R2: Genomics Analysis and Visualization Platform (http://r2.amc. nl). A, B, and C, We evaluated the survival probability of patients with melanoma based on their tumour transcriptome 38 (n = 214). Kaplan-Meier plots depicting the survival probability of patients with melanoma based on the expression of Nav1.8, TRPV1 and VIP in the tumours. A, High Nav1.8 expression is associated with improved outcome in patients with melanoma. B, High TRPV1 expression is associated with improved outcome in patients with melanoma. C, High VIP expression is associated with improved outcome in patients with melanoma. We queried a Kaplan-Meier plotter data set. 38 Log-rank test was used implicated in pancreatic 56 and gastric 17,57 tumour progression, respectively. Whether our findings on sensory nerves apply to other organs remains unknown. Also, given the potential links within the peripheral nervous system, future research should evaluate whether different peripheral nerves within the tumour microenvironment communicate between them, and whether one innervation compensates for the absence of the other.
Histological proximity between malignant cells and peripheral innervations was described for many years exclusively as perineural invasion. 58 In recent years, this concept has evolved, and new investigations are revealing that nerves may also play pro-active roles within the tumour microenvironment, by regulating cancer progression. 17,41,56,57,59,60 Nevertheless, several questions still need to be answered. It remains to be elucidated whether functions of innervations are the same in distinct cancer types. Along the same lines, which cells in the tumour microenvironment are attracting the nerves? Are the attractants coming directly from cancer cells? For the progress of our knowledge on the roles of various innervations within the tumour microenvironment, more functional studies are needed. For instance, in vivo genetic conditional knockout experiments, by deleting key genes from specific intra-tumoural nerves, will advance our understanding. These studies will clarify molecular mechanisms involved in the intra-tumoural axonogenesis as well as in the interactions between nerves and other components present within the tumours.
A big challenge for the future will be to seek the translational value of these new findings to patients. Nerves may have promising clinical use not only as biomarkers, but also as therapeutic targets. Yet, more experimental investigations are needed to clarify the potential and discover specific targets within the tumour nerves for cancer F I G U R E 6 Schematic illustration summarizing the results of sensory nerves depletion from the tumour microenvironment. Sensory innervations, identified by their expression of Nav1.8, can be found in melanomas generated by B16F10 cell inoculation. (Top). Genetic ablation of sensory Nav1.8+ nerve fibers increases tumour size. Similar results are obtained with chemical depletion of sensory nerves when resiniferatoxin (RTX) is given before tumour implantation (Bottom left). In contrast, when sensory nerves are chemically ablated after tumour implantation, tumour size decreases, probably due to the cytotoxic effect of RTX acting directly on tumour cells (Bottom right) management. The development of methods to manoeuvre intra-tumoural innervations is yet at the earliest phases, and will demand the assembly of multidisciplinary groups to create nerves-based treatments that can be used in the clinic. Among novel approaches, bioelectronic medicine may bring the possibility to control, by inhibition or stimulation, individual nerve fibers within the tumours, avoiding the off-target side effects caused by most pharmacological drugs. 61 Although enormous advancement has been achieved in targeting the tumour microenvironment to treat cancer, the best is yet to come.
In conclusion, our results suggest that protection of endogenous sensory nerves in cancer patients may thus provide an stimulating new avenue for anti-cancer therapy. Taking into account our evidence regarding sensory nerves anti-tumoural role, expanding the research into the mechanisms by which sensory innervations act blocking tumour growth may lead to the identification of potential new therapeutic pathways. This work raises the exciting and novel concept that targeting the peripheral sensory nervous system in the tumour might provide a novel approach to treat melanoma. Additionally, this study lays the groundwork for a paradigm that may have a broad impact on our understanding and the management of other cancers as well.