Protease-Activated Receptor 2 (PAR2) expressed in sensory neurons contributes to signs of pain and neuropathy in paclitaxel treated mice

Abstract

Chemotherapy-Induced Peripheral Neuropathy (CIPN) is a common, dose-limiting side effect of cancer therapy. Protease-activated receptor 2 (PAR2) is implicated in a variety of pathologies, including CIPN. In this study, we demonstrate the role of PAR2 expressed in sensory neurons in a paclitaxel (PTX)-induced model of CIPN in mice. PAR2 knockout/WT mice and mice with PAR2 ablated in sensory neurons were treated with paclitaxel administered via intraperitoneal injection. In vivo behavioral studies were done in mice using von Frey filaments and the Mouse Grimace Scale. We then examined immunohistochemical staining of dorsal root ganglion (DRG) and hind paw skin samples from CIPN mice to measure satellite cell gliosis and intra-epidermal nerve fiber (IENF) density. Pharmacological reversal of CIPN pain was tested with the PAR2 antagonist C781. Mechanical allodynia caused by paclitaxel treatment was alleviated in PAR2 knockout mice of both sexes. In the PAR2 sensory neuronal conditional knockout (cKO) mice, both mechanical allodynia and facial grimacing were attenuated in mice of both sexes. In the dorsal root ganglion of the paclitaxel-treated PAR2 cKO mice, satellite glial cell activation was reduced compared to control mice. IENF density analysis of the skin showed that the paclitaxel-treated control mice have a reduction in nerve fiber density while the PAR2 cKO mice had a comparable skin innervation as the vehicle-treated animals. Similar results were seen with satellite cell gliosis in the DRG where gliosis induced by PTX was absent in PAR cKO mice. Finally, C781 was able to transiently reverse established PTX-evoked mechanical allodynia.

INTRODUCTION

More than 16.9 million cancer survivors were alive in 2019, and due to better early detection of cancer and advancements in chemotherapeutic agents, this number is projected to increase to more than 22.1 million by 2030^{18, 44, 71}. With increased cancer survivorship, a common consequence of treatment that cancer patients and survivors deal with are the long-lasting neuropathic symptoms from peripheral nerve damage caused by neurotoxic chemotherapeutics, a condition known as chemotherapy-induced peripheral neuropathy (CIPN)^{30, 33, 72, 86, 97}. CIPN affects over 60% of patients in the first month after chemotherapy and, although the prevalence of CIPN decreases over time, 30% of patients continue to suffer from CIPN at 6 months or more⁹¹. Common symptoms of CIPN include numbness, tingling, hyperalgesia, and spontaneous pain which can be disabling and impair quality of life^{25, 61, 78, 91, 97, 100}. Furthermore, CIPN symptoms can lead to dose reduction or even treatment cessation, potentially affecting patient survival^{9, 25, 78, 94}. Despite this, few effective therapeutics are currently available to remedy or prevent the effects of CIPN due to a poor understanding of the underlying mechanism^{26, 37, 64}.

The effects of chemotherapy drugs on the nervous system can vary depending on the dosage and drug class, the most neurotoxic classes being platinum-based drugs, taxanes, and thalidomide⁷. Paclitaxel (PTX) is a taxane chemotherapeutic drug used in the treatment of breast, lung, and ovarian cancer, among others^{105, 108}. Taxanes impair cancer cell division and growth by stabilizing microtubules, thereby preventing the chromosomes from separating and leading to apoptosis in dividing cells^{1, 6, 41, 111}. Though neurons are not dividing cells, they are still susceptible to complications from PTX exposure. PTX has poor penetration into the CNS but has been shown to accumulate in the DRG at high concentrations, thus resulting in mainly sensory neuropathic symptoms^{14, 23}. Effects on the PNS by PTX are multifaceted and include dysfunction of microtubules, mitochondria, and calcium homeostasis, axon degeneration, sensitization of ion channels, hyperexcitability of sensory neurons, and activation of glial and immune cells^{77, 86, 111}.

Protease-Activated Receptor 2 (PAR2) has been proposed as a possible mediator of CIPN pain via the neuroimmune interactions that are catalyzed by chemotherapy treatment^{16, 85, 98}. PAR2 is a G-protein coupled receptor (GPCR) implicated in a number of pathologies including inflammatory, cancer, migraine, and CIPN pain^{16, 35, 50, 53, 54, 68, 83, 85, 98, 102, 113}. The extracellular portion of the receptor is cleaved by proteases, such as those released by immune cells. Cleavage by trypsin-like proteases exposes a tethered ligand that can then bind to the receptor, leading to the recruitment of G_q and β-arrestins, while cleavage by elastase promotes the recruitment of G_12/13 and Rho-dependent activation of MAPK^{2, 38, 46, 51, 82}. The β-arrestin/MAPK pathway in particular causes nascent protein synthesis in a small subpopulation of sensory neurons and causes mechanical hypersensitivity and facial grimacing after PAR2 activation^{36, 99}.

Here, we test the hypothesis that sensory neuronal PAR2 plays a key role in CIPN pain using behavioral genetics, pharmacology and immunohistochemistry. We measured both mechanical and spontaneous nociceptive behaviors using von Frey filaments and the Mouse Grimace Scale (MGS), respectively, in paclitaxel-treated PAR2 global knockout mice and Pirt^CreF2rl1^flox mice. We also measured nerve fiber density in the hind paw skin and satellite cell gliosis in the dorsal root ganglion (DRG) in the Pirt^CreF2rl1^flox mice after paclitaxel treatment. Finally, we tested a PAR2 antagonist, C781, in a paclitaxel CIPN model. Our findings demonstrate that PAR2 expressed on sensory neurons contributes to both mechanical and spontaneous nociceptive behaviors as well as changes in nerve fiber density in the epidermis and satellite cell gliosis.

MATERIALS AND METHODS

Animals

All experiments and procedures were approved by the Institutional Animal Care and Use Committee at University of Texas at Dallas. To generate Pirt^CreF2rl1^flox mice, loxP sites were inserted on exon 2 of the F2rl1 gene and crossed with the Pirt^Cre mouse, as described previously³⁶. Our experimental mice were homozygous for the F2rl1^flox gene and heterozygous for the Pirt^Cre gene due to the insertion of Cre recombinase into the Pirt promoter, thus rendering Pirt expression null in the allele. For control groups, we used vehicle-treated mice as well as Pirt^Cre mice that have wildtype F2rl1 expression. The mice were bred in house at the University of Texas at Dallas and housed in climate-controlled rooms set to 22°C with a 12-hour light/dark cycle and given food and water ad libitum. Male and female mice were used in all experiments. Because we have not previously found any sex differences in PAR2 signaling related to pain, sex differences were not assessed in experiments described here and the study was not powered to evaluate sex differences^{36, 65}. Because of this, we also do not reach any conclusions on sex differences in paclitaxel effects on mice.

Experimental reagents

Paclitaxel was purchased from Sigma Aldrich (Y0000698) and dissolved in a 50/50 Kolliphor^® EL (Sigma Aldrich C5135)/ethanol solution and then further diluted in sterile Dulbecco’s phosphate buffered saline (DPBS; Thermo Scientific) for injections. Paclitaxel or vehicle control (50/50 Kolliphor^® EL/ethanol solution diluted in DPBS) was injected into the peritoneal space using a 27G ½” needle. C781 was synthesized using solid-phase synthesis as described previously⁸⁸. C781 was injected into the peritoneal space using a 27G ½” needle at a dosage of 10 mg/kg.

Behavioral Methods

Experimenters were blinded to treatment and genotype groups in all experiments. Mice were treated with PTX (4 mg/kg IP) every other day for a total of four injections (cumulative dose of 16 mg/kg)^{43, 69, 101}. Intermittent systemic administration of PTX was done to mimic cyclic chemotherapy dosing regimens in human patients^{28, 40}. Animals were habituated to suspended acrylic chambers with each individual chamber measuring 11.4 × 7.6 × 7.6 cm and a wire mesh bottom (1 cm²). Mechanical hypersensitivity was assessed using the manual von Frey hair test after spontaneous pain assessment using the Mouse Grimace Scale (MGS). This was done to assure that grimacing responses were not a result of mechanical stimulation.

Mouse grimace scores were determined using the Mouse Grimace Scale (MGS) test as described by Langford and colleagues⁵⁶. After habituation, the experimenter observed the mice and assigned intensity ratings (0 = not present, 1 = moderately present, 2 = severe) to the facial action units relative to baseline. The five action units identified are 1) orbital tightening, 2) nose bulge, 3) cheek bulge, 4) ear position, and 5) whisker change. The scores for all five action units are averaged to obtain the MGS score for each time point. The MGS has been shown to be an accurate way of measuring spontaneous acute nociception in response to multiple types of stimuli, including PTX-induced CIPN^{3, 5, 58, 67, 103}.

Hind paw withdrawal thresholds were calculated via the Dixon up-down method after applying calibrated von Frey filaments from Stoelting, Co. (cat# 58011) to the volar surface of the left hind paw¹⁵. Increasing pressure was applied until the filament bent against the skin at a right angle and a positive response was recorded when the mouse showed nocifensive behaviors, such as rapid paw withdrawal, flicking, and/or licking of the stimulated paw. The maximum filament weight used was 2 g.

Immunohistochemistry

After the conclusion of behavioral experiments, animals were anesthetized using 4% isoflurane and euthanized by cervical dislocation and decapitation. Hind paw skin was carefully dissected and fixed overnight in Zamboni’s fixative solution at 4°C. The skin was washed three times (30 min each) in PBS to remove excess fixative and incubated in 30% sucrose at 4°C until it sunk. Samples were blotted dry on paper towels, embedded in Optimum cutting temperature (OCT), and flash-frozen in dry ice. Tissues were sectioned at 50 μm and immersed into a netted well with PBS. Sections were washed three times in PBS to remove excess OCT and then incubated in blocking buffer (10% normal goat serum, 0.3% Triton X-100 in 0.1 M PB) for 1 hr at room temperature. Sections were then incubated overnight in primary antibody (rabbit anti-PGP9.5; CL7756AP-50; Cedarlane) at 1:1000. The next day, sections were washed in PBS three times and incubated in secondary antibody, cross-adsorbed goat anti-rabbit IgG (H+L) Alexa Fluor 555 (1:2000; A21428; Invitrogen, Thermo Fisher Scientific) for 1 hr at room temperature. Finally, sections were incubated for 5 min in 1:5000 DAPI (ACD), washed in PBS two times, moved and flattened onto slides, and cover-slipped using ProLong Gold mounting medium (Thermo Fisher Scientific).

Dorsal root ganglia (DRGs) were dissected and embedded in OCT and immediately flash-frozen in dry ice. Tissues were sectioned at 20 μm onto charged slides where they were briefly thawed to adhere to the slide but immediately returned to the 20°C cryostat chamber until completion of sectioning. The sections were fixed in ice-cold 10% formalin for 15 min followed by incubation in 50%, 70%, and then 100% ethanol for 5 min each. Slides were placed in a light-protected, humidity-controlled tray and incubated in blocking buffer for 1 hr at room temperature. Slides were then incubated overnight in mouse anti-GFAP (N206A/8; NeuroMab) and chicken anti-peripherin (CPCA-Peri; Encor Biotechnology) at 1:1000 in blocking buffer overnight at 4°C. The next day, slides were washed in PBS three times and incubated in goat anti-chicken (H&L) Alexa Fluor 488 (A11039; Invitrogen, Thermo Fisher Scientific) goat anti-mouse IgG1 Alexa Fluor 555 (A21127; Invitrogen, Thermo Fisher Scientific) at 1:2000 for 1 hr at room temperature. Finally, sections were incubated for 5 min in 1:5000 DAPI, washed in PBS two times, and then cover-slipped using ProLong Gold mounting medium.

Image analysis for skin sections

Three sections per mouse were imaged on an Olympus FV3000 confocal microscope at magnification x20 and then analyzed using Cellsens (Olympus) software. Acquisition parameters were maintained constant. To determine the number of intraepidermal nerve fibers (IENF), the number of free nerve endings that cross the dermo-epidermal junction was counted in a blinded manner according to counting rules recommended by the European Federation of Neurological Societies (EFNS)⁵⁷. The IENF density was calculated as the number of fibers per millimeter of epidermal length.

Image analysis for DRG sections

Three sections per mouse were imaged on an Olympus FV3000 confocal microscope at magnification x20 and then analyzed using Cellsens (Olympus) software. Acquisition parameters were maintained constant. Satellite glial cells were distinguished via GFAP staining, as well as shape and position in relation to neurons, which were stained with peripherin. Neurons that were surrounded by satellite glial cells by half or more of their circumference were categorized as GFAP⁺ neurons. The total number of GFAP⁺ neurons were quantified and divided by the total number of neurons in the field and presented as a percentage.

Data and Statistical Analysis

Sample sizes for in vivo experiments were estimated based on effect sizes observed in PAR2 sensory neuron-specific conditional knockout (cKO) mice as reported previously³⁶. All statistical tests were done using GraphPad Prism version 9.4.1 (GraphPad Software, Inc.). Differences between groups were calculated using repeated measures 2-way ANOVAs with Dunnett’s or Sidak’s multiple-comparison test. Effect size was determined by calculating the cumulative differences between the baseline value and the experimental value at each time point. One-way ANOVA with Dunnett post-test was used to compare effect sizes. Statistical analysis for IENF and SGC analysis is described in figure legends. All data are represented as mean ± SEM.

RESULTS

Mechanical hypersensitivity and grimace caused by PTX treatment are profoundly reduced in mice lacking PAR2 expressed on sensory neurons

We tested whether global PAR2 knockout mice exhibit reduced mechanical hypersensitivity in a paclitaxel-induced CIPN model. We injected 4 mg/kg of PTX into the intraperitoneal space every other day for a total of 4 injections. Using von Frey testing, we found that PTX caused long-lasting mechanical hypersensitivity in the wildtype control mice that lasted out to 23 days. In contrast, this mechanical hypersensitivity was significantly attenuated in the global PAR2 knockout mice (Figure 1A). The vehicle-treated control animals showed paw withdrawal thresholds similar to baseline levels. The WT PTX mice showed significantly higher effect size when compared to both vehicle groups but not to the PTX-treated PAR2 knockout group (Figure 1B).

Figure 1: PAR2^−/− mice show attenuated mechanical hypersensitivity in response to paclitaxel treatment.

Baseline (BL) paw withdrawal thresholds for male and female ICR mice were obtained before paclitaxel treatment (4 mg/kg IP every other day for 4 days). (A) The WT control mice that received PTX treatment exhibited long-lasting mechanical allodynia that was significantly different from the WT Veh group on days 7, 9, 11, 14, 19, and 23 after the initial PTX injection. On the other hand, the PAR2^−/− mice treated with PTX were significantly different from the WT Veh group only on day 7. Purple asterisks denote significant differences between the WT PTX group and the WT Veh. Black asterisks denote significant differences between the PAR2^−/− PTX group and the WT Veh group. (B) Effect size was determined by calculating the cumulative differences between the baseline value and the experimental value at each time point. The WT PTX mice show a significantly higher effect size when compared to the WT Veh group. However, both the effect sizes of the PAR2^−/− groups were not significantly different from the WT Veh group. For the WT Veh group, n = 3 females and n = 3 males. For the WT PTX group, n = 4 females and n = 3 males. For the PAR2^−/− Veh group, n = 3 females and n = 3 males. For the PAR2^−/− PTX group, n = 3 females and n = 3 males. *p <0.05, **p<0.01, ***p<0.001, ****p<0.0001. For paw withdrawal threshold, repeated measures two-way ANOVA and Dunnett’s multiple comparisons test was used with group mean comparisons made to the WT Veh group. For effect size, one-way ANOVA with Dunnett’s multiple comparisons test was used with all group mean comparisons made to the WT Veh group.

Our previous research on PAR2-evoked pain using the Pirt^CreF2rl1^flox mouse, a sensory neuron-specific cKO PAR2 mouse, revealed that mechanical hypersensitivity and spontaneous pain caused by PAR2 activation is dependent on PAR2 expressed on a small subpopulation of sensory neurons^{36, 47, 48}. Therefore, we treated these mice with PTX to determine whether we would we see similar results after PTX treatment. As previously seen, PTX caused long-lasting mechanical hypersensitivity in the Pirt^CreF2rl1^+/+ control animals that lasted out to 23 days (Figure 2A). PTX also caused facial grimacing in these animals that lasted out to 23 days (Figure 2C). Compared to the Pirt^CreF2rl1^+/+ control animals, the Pirt^CreF2rl1^flox mice showed similar withdrawal thresholds and facial grimace scores to the vehicle-treated animals. (Figure 2A and 2B). For facial grimace scoring, the Pirt^CreF2rl1^+/+ PTX group showed significantly higher grimace scores when compared to the Pirt^CreF2rl1^+/+ Veh group. However, both Pirt^CreF2rl1^flox groups showed similar grimace scores to the Pirt^CreF2rl1^+/+ Veh group (Figure 2C and 2D). These results suggest that the mechanical hypersensitivity and facial grimacing caused by PTX can be mediated by PAR2 expressed on sensory neurons.

Figure 2: Pirt^CreF2rl1^flox mice show attenuated mechanical hypersensitivity and facial grimacing in response to paclitaxel treatment.

Baseline (BL) paw withdrawal thresholds and facial grimacing scores for male and female Pirt^CreF2rl1^flox and control mice were obtained before paclitaxel treatment (4 mg/kg IP every other day for 4 days). (A) The control mice that received PTX treatment exhibited long-lasting mechanical allodynia that was significantly different from the Pirt^CreF2rl1^+/+ Veh group on days 3, 7. 9, 13, 17, and 23 after the initial PTX injection. On the other hand, the Pirt^CreF2rl1^flox mice treated with PTX show similar paw withdrawal thresholds to the Pirt^CreF2rl1^+/+ Veh group. Pink asterisks denote significant differences between the Pirt^CreF2rl1^+/+ PTX group and the Pirt^CreF2rl1^+/+ Veh group. Gray plus signs denote significant differences between the Pirt^CreF2r1l^flox Veh group and the Pirt^CreF2rl1^+/+ Veh group. (B) Effect size was determined by calculating the cumulative differences between the baseline value and the experimental value at each time point. The Pirt^CreF2rl1^+/+ PTX mice show a significantly higher effect size when compared to the Pirt^CreF2rl1^+/+ Veh group. The effect sizes of the Pirt^CreF2rl1^flox Veh and Pirt^CreF2rl1^flox PTX groups were not significantly different from the Pirt^CreF2rl1^+/+ Veh group. (C) The control mice that received PTX treatment also exhibited long-lasting facial grimacing that was significantly different from the Pirt^CreF2rl1^+/+ Veh group on days 3, 7, 9, 13, and 17 after the initial PTX injection. However, both of the Pirt^CreF2rl1^flox groups showed similar grimace scores to the Pirt^CreF2rl1^+/+ Veh group. Pink asterisks denote significant differences between the Pirt^CreF2rl1^+/+ PTX group and the Pirt^CreF2rl1^+/+ Veh group. Black asterisks denote significant differences between the Pirt^CreF2rl1^flox PTX group and the Pirt^CreF2rl1^+/+ Veh group. (D) The Pirt^CreF2rl1^+/+ PTX mice show a significantly higher effect size when compared to the Pirt^CreF2rl1^+/+ Veh group. The effect sizes of the Pirt^CreF2rl1^flox Veh and Pirt^CreF2rl1^flox PTX groups were not significantly different from the Pirt^CreF2rl1^+/+ Veh group. For the Pirt^CreF2rl1^+/+ Veh group, n = 3 females and n = 3 males. For the Pirt^CreF2rl1^+/+ PTX group, n = 4 females and n = 3 males. For the Pirt^CreF2rl1^flox Veh group, n = 2 females and n = 4 males. For the Pirt^CreF2rl1^flox PTX group, n = 3 females and n = 6 males. *p <0.05, **p<0.01, ***p<0.001, ****p<0.0001. For paw withdrawal threshold, repeated measures two-way ANOVA and Dunnett’s multiple comparisons test was used. For facial grimace scores, repeated measures two-way ANOVA with Dunnett’s multiple comparisons test was used. For effect size, one-way ANOVA with Dunnett’s multiple comparisons test was used. All group mean comparisons were made to the Pirt^CreF2rl1^+/+ Veh group.

Epidermal innervation loss caused by PTX treatment is reduced in mice lacking PAR2 expressed on sensory neurons

PTX treatment has been demonstrated to result in a loss of IENF density in both humans and rodents^{10, 17, 49, 62, 92, 109}. Our previous analysis of single-cell RNA-sequencing and RNAscope validation experiments showed that the PAR2 gene, F2rl1, is expressed specifically in a small subpopulation of neurons that also express IL31ra, Nppb, Hrh1, and Mrgprx1, all genes believed to be important for mediating itch and expressed in sensory neurons that innervate the epidermis^{36, 55, 60, 70, 73}. To determine whether PAR2 expression on this subpopulation of skin-innervating sensory neurons is critical for PTX-induced IENF loss, we dissected the hind paw skin from the mice used in the in vivo behavioral assays after completion of experiments. The skin samples were stained for PGP9.5 to visualize the intraepidermal nerve fibers and the free nerve endings that cross the dermo-epidermal junction were counted.

The vehicle-treated animals showed dense nerve fibers that innervate the epidermis. After PTX treatment, the Pirt^CreF2rl1^+/+ control mice show a clear loss of nerve fibers in the skin that is not seen in the Pirt^CreF2rl1^flox PTX group (Figure 3A & 3B). The IENF density was calculated as the number of fibers that crossed the dermo-epidermal junction per millimeter of epidermal length. The Pirt^CreF2rl1^+/+ PTX group showed significantly lower IENF/mm than the Pirt^CreF2rl1^+/+ Veh group. Both Pirt^CreF2rl1^flox groups were not significantly different from the Pirt^CreF2rl1^+/+ Veh group. These results indicate that the Pirt^CreF2rl1^flox mice are protected from nerve fiber loss as a result of PTX treatment.

Figure 3: Pirt^CreF2rl1^flox mice are protected from nerve fiber loss as a result of paclitaxel treatment.

(A) Hind paw skin was dissected, cryosectioned at 50 μm, and stained for PGP9.5 (green) to visualize skin nerve fibers and DAPI (red) to visualize the dermo-epidermal junction. The Pirt^CreF2rl1^+/+ PTX mice show a clear loss in nerve fibers in the epidermis that is not seen in the Pirt^CreF2rl1^flox PTX group or vehicle groups. (B) To determine the number of intraepidermal nerve fibers (IENF), the number of free nerve endings that cross the dermo-epidermal junction was counted and the IENF density was calculated as the number of fibers per millimeter of epidermal length. The Pirt^CreF2rl1^+/+ PTX group showed significantly lower IENF/mm than the Pirt^CreF2rl1^+/+ Veh group. Both Pirt^CreF2rl1^flox groups were not significantly different from the Pirt^CreF2rl1^+/+ Veh group. ***p<0.001. For IENF/mm comparisons, one-way ANOVA with Dunnett’s multiple comparisons test was used. All group mean comparisons were made to the Pirt^CreF2rl1^+/+ Veh group. Three sections per mouse were analyzed and averaged together. For the Pirt^CreF2rl1^+/+ Veh group, n = 6 animals. For the Pirt^CreF2rl1^+/+ PTX group, n = 6 animals. For the Pirt^CreF2rl1^flox Veh group, n = 6 animals. For the Pirt^CreF2rl1^flox PTX group, n = 8 animals.

Satellite cell gliosis caused by PTX treatment is reduced in mice lacking PAR2 expressed on sensory neurons

Another hallmark of CIPN is satellite glial cell activation in the DRG¹⁰⁶. Satellite glial cells surround the cell bodies of DRG neurons and play an important role in intercellular communication with the neurons they surround³⁴. In pathological conditions such as nerve injury or inflammation, satellite cell gliosis is observed as increased expression of glial fibrillary acidic protein (GFAP)^{76, 96, 107}. It is possible that similar to nerve axotomy, the loss in epidermal innervation could lead to satellite cell gliosis¹⁰⁶. To determine whether PAR2 expression on sensory neurons is important for satellite cell gliosis as a result of PTX treatment, we dissected the DRG from the mice used in the in vivo behavioral assays after completion of experiments. The DRG samples were stained for GFAP to visualize the satellite glial cells and the percentage of neurons surrounded by GFAP-positive satellite glial cells was determined.

In the vehicle-treated animals, few DRG neurons were surrounded by GFAP-positive satellite glial cells. PTX treatment caused an increase in GFAP-positive satellite glial cells in the Pirt^CreF2rl1^+/+ mice. However, this increase was not seen in the PTX-treated Pirt^CreF2rl1^flox mice, indicating that PAR2 plays a role in satellite cell gliosis after PTX treatment (Figure 4A & 4B).

Figure 4: Pirt^CreF2rl1^flox mice are protected from satellite gliosis as a result of paclitaxel treatment.

(A) Dorsal root ganglion (DRG) was dissected, cryosectioned at 20 μm, and stained for Peripherin (green) to visualize peripheral neurons, GFAP (red) to visualize satellite glial cells, and DAPI (blue) to visualize nuclei. The Pirt^CreF2rl1^+/+ PTX mice show clear satellite cell gliosis as represented by an increase in GFAP expression in the DRG that is not seen in the Pirt^CreF2rl1^flox PTX group or vehicle groups. (B) To determine the percentage of neurons that are surrounded by satellite glial cells, the number of neurons that were surrounded by 50% or more by GFAP-positive satellite glial cells was divided by the total number of neurons in the field of view. The Pirt^CreF2rl1^+/+ PTX group showed significantly higher number of neurons surrounded by GFAP-positive satellite glial cells than the Pirt^CreF2rl1^+/+ Veh group. Both Pirt^CreF2rl1^flox groups were not significantly different from the Pirt^CreF2rl1^+/+ Veh group. *p <0.05. For GFAP⁺/total neurons comparisons, one-way ANOVA with Dunnett’s multiple comparisons test was used. All group mean comparisons were made to the Pirt^CreF2rl1^+/+ Veh group. Three sections per mouse were analyzed and averaged together. For the Pirt^CreF2rl1^+/+ Veh group, n = 6 animals. For the Pirt^CreF2rl1^+/+ PTX group, n = 3 animals. For the Pirt^CreF2rl1^flox Veh group, n = 4 animals. For the Pirt^CreF2rl1^flox PTX group, n = 6 animals.

A PAR2 antagonist transiently reverses established mechanical hypersensitivity caused by PTX treatment

To determine whether established mechanical hypersensitivity caused by PTX treatment could be reversed by PAR2 signaling inhibition we used a β-arrestin signaling biased PAR2 antagonist that we developed, C781⁸⁸. C781 is a specific inhibitor of PAR2 signaling among other PAR receptors, but the antagonist has not been assessed in broad off-target screens for other GPCRs or other targets⁸⁸. C781 was given by intraperitoneal injection and mechanical threshold and facial grimacing was assessed following treatment. Time points were chosen based on previously published pharmacokinetic studies done in mice administered with C781⁵². C781 treatment created a transient reversal of mechanical hypersensitivity but did not have an influence on facial grimacing (Figure 5A & 5B).

Figure 5: C781 transiently reverses mechanical hypersensitivity in response to paclitaxel treatment.

Baseline (BL) paw withdrawal thresholds and facial grimace scores for male and female ICR mice were obtained before paclitaxel treatment (4 mg/kg IP every other day for 4 days). Before C781 treatment (red arrow), establishment of CIPN was checked to ensure that animals were mechanically hypersensitive and grimacing. C781 was injected at 10 mg/kg into the intraperitoneal space. (A) Mice treated with C781 transiently showed higher mechanical thresholds. (B) C781 treatment had no effect on facial grimacing. For the Veh group, n = 3 females and n = 4 males. For the C781 group, n = 4 females and n = 3 males. **p<0.01. For both paw withdrawal threshold and grimace scores, repeated measures two-way ANOVA with Sidak’s multiple comparisons test was used.

DISCUSSION

In this study we demonstrate that PAR2 expressed on sensory neurons plays an important role in mediating paclitaxel-induced CIPN nociceptive behaviors as well as pathophysiological changes in the skin and DRG that are consistent with neuropathy that is seen in patients. Our findings show that both mechanical hypersensitivity and facial grimacing as a result of paclitaxel treatment are reduced in both global PAR2 knockout mice and mice with PAR2 ablated selectively in sensory neurons. We also demonstrate that paclitaxel-induced skin innervation loss and satellite cell gliosis in the DRG are prevented in Pirt^CreF2rl1^flox mice. Finally, treatment with a PAR2 antagonist, C781, transiently reversed paclitaxel-evoked mechanical hypersensitivity. We conclude that PAR2 expressed in sensory neurons serves as an important mediator of paclitaxel-evoked pain and neuropathy and may be a possible therapeutic target for the treatment of CIPN more broadly.

Previous studies have made a case for PAR2 as an important contributor to CIPN using PAR2 antagonists and global knockout mice, but the cells that express PAR2 to mediate these effects have been unclear¹⁶. A key contribution of our work is clarifying that PAR2 expressed by sensory neurons is the site of action for the role of PAR2 in CIPN, at least for paclitaxel treatment. PAR2 has been shown to sensitize the transient receptor potential vanilloid 4 (TRPV4) ion channel to cause mechanical hyperalgesia, and TRPV4 has also been implicated to promote paclitaxel-induced mechanical allodynia^{31, 32, 66, 81}. PAR2 has also been shown to sensitize transient receptor potential ankyrin 1 (TRPA1) and vanilloid type 1 (TRPV1) in rodent models of CIPN¹⁶. In other studies, the use of a PAR2 antagonist attenuated paclitaxel-induced cold allodynia¹⁶. Although we did not study the role of sensory neuronal PAR2 in paclitaxel-induced heat hyperalgesia or cold allodynia, future studies using the Pirt^CreF2rl1^flox mice could be done to investigate this. Previous studies using sensory neuron ablation of PAR2 have shown that sensory neuronal PAR2 is critical for pain caused by injection of exogenous proteases, pain caused by inflammatory mediators, pain caused by proteases released from immune and cancer cells and certain types of visceral pain^{36, 46}. In the DRG of mice, PAR2 is expressed by a specific subset of neurons that express markers that are consistent with these neurons innervating the outermost layer of the skin as free nerve endings in the epidermis^{36, 55, 70, 73}. Our finding that intraepidermal nerve fiber loss is abrogated with sensory neuron-specific ablation of PAR2 is consistent with this previous literature. Although PAR2 is only expressed by a small subset of sensory neurons, we noted a nearly complete loss of satellite cell gliosis caused by paclitaxel treatment in the DRGs of mice with sensory neuron cKO of PAR2. The reasons for this effect are currently unclear, but suggest that PAR2 is a key mediator of cascading signaling events that lead to a broad satellite cell gliosis that eventually envelops neurons that likely do not express PAR2. This could be a mechanism through which a neuropathy that usually first affects long axons that innervate the epidermis can spread to other neuronal subtypes causing spontaneous pain in CIPN patients.

Paclitaxel is considered one of the most neurotoxic chemotherapeutics, causing one of the highest CIPN rates in patients⁷⁴. This is in line with preclinical CIPN models, in which paclitaxel or cisplatin treatment in mice showed the highest efficacy in causing CIPN²⁸. Although many clinical trials have investigated potential therapies for the prevention or treatment of CIPN, as of 2020, only duloxetine has emerged as a moderately recommended treatment by the American Society of Clinical Oncology (ASCO) clinical practice guidelines^{22, 39, 64, 110}. However, duloxetine has been shown to be more effective in alleviating CIPN as a result of platinum-based therapies like cisplatin rather than taxane-treated patients, suggesting a difference in CIPN mechanism⁹³. Most clinical trial studies have been based on evidence from other types of neuropathies (e.g., diabetic neuropathy) but the mechanism(s) through which CIPN develops differentially from other chronic neuropathic pain conditions, and also how these mechanisms may differ between various chemotherapeutic agents, remain unclear^{11, 42}. CIPN can develop through multiple targets, and while the neurotoxic mechanisms that initiate and maintain CIPN potentially overlap with its cytotoxic mechanisms, it is imperative that CIPN-blocking drugs do not function at the expense of the antitumor efficacy of chemotherapeutics^{11, 42}. PAR2 has been associated with cancer where it is thought to play a role in promoting oncogenesis and cancer cell invasiveness^{95, 104}. While PAR2 antagonists as a treatment for CIPN would require testing on specific cancer types, we would not expect that blocking PAR2 would cause interference with the efficacy of chemotherapeutics given the existing literature on PAR2 and cancer⁴⁵.

Paclitaxel has been shown to cause immune cell infiltration and activation in the skin and in the DRG, including increased infiltration and activation of macrophages, mast cells, and neutrophils^{16, 29, 63, 79, 80, 112}. PAR2 thus emerged as a possible mediator of CIPN as these activated immune cells release proteases which act on PAR2, causing sensitization of sensory neurons and hyperalgesia^{8, 16, 75, 85, 98}. Macrophages are known to release PAR2-activating proteases including elastase and cathepsin S, and the use of minocycline, a macrophage inhibitor, prevents macrophage recruitment into the DRG as well as mechanical hypersensitivity caused by paclitaxel treatment^{10, 12, 13, 19, 62}. Similar results were also seen with quercetin, a mast cell stabilizer²⁹. Furthermore, paclitaxel triggers the release of high mobility group box 1 (HMGB1) from macrophages via ATP signaling and increased reactive oxygen species (ROS) production. While there is evidence for sex differences in HMGB1 signaling, PAR2 depletion protects both male and female mice from developing disulfide HMGB1-induced mechanical hypersensitivity^{4, 20, 84, 89, 90}. This possibly suggests that HMGB1 induces different signaling pathways in males and females that then converge on PAR2 activation which then mediates at least part of the pain produced by HMGB1⁴. Collectively, these studies point to an important role of PAR2 in integrating immune-mediated signaling in a small subset of neurons that innervates the outermost layer of the skin.

Our work here fits with previous studies on signaling within sensory neurons that causes many features of CIPN, including mitochondrial and lysosomal dysfunction and altered translation regulation that causes nociceptor hyperexcitability^{21, 24}. The model emerging from the present experiments, and previous work on mechanisms of paclitaxel-evoked pain, is that PAR2 activation on nociceptors induces β-arrestinmediated mitogen activated protein kinase (MAPK) activation including phosphorylation of extracellular regulated protein kinase (ERK) and mitogen activated protein kinase interacting kinase (MNK)⁹⁹. This then induces changes in gene expression driven by phosphorylation of proteins that control translation of mRNAs that are involved in nociceptor hyperexcitability, and lysosomal and mitochondrial dysfunction via increased translation of RagA and altered mechanistic target of rapamycin (mTOR) signaling at the surface of lysosomes⁶⁹. This pathology likely starts in neurons that innervate the skin, in particular the skin of the hands and feet, but it may spread to a broader subset of neurons as paclitaxel causes hyperexcitability in many nociceptor populations both in mice and in humans (Figure 6). It will be important to understand how this broader effect of paclitaxel spreads and what role PAR2 plays in initiating this cascade of signaling events. As mentioned above, one testable hypothesis is satellite cell gliosis which is well known to occur in CIPN mouse models. Interestingly we recently showed that meteorin, which is thought to target satellite glial cells, can both prevent and reverse paclitaxel-evoked pain and signs of neuropathy⁸⁷.

Figure 6: Summary diagram of potential role of PAR2 and downstream signaling in paclitaxel-induced neuropathy.

The figure summarizes our hypothesis of how PAR2 signaling on a specific subset of nociceptors might lead to loss of epidermal nerve fibers, satellite cell gliosis and pain. A key feature of the model is that paclitaxel-induced changes may begin with a small subset of sensory fibers, but gene expression changes in DRG cell bodies driven by PAR2 activation could cause spread of satellite cell gliosis causing sensitization of additional subsets of DRG nociceptors. Some conclusions summarized in the figure are from previously published studies including PAR2 signaling inducing changes in translation regulation⁹⁹, the important role of MNKeIF4E signaling in paclitaxel CIPN model⁶⁹, and the key role of satellite cell gliosis in CIPN^{27, 59, 87}.

C781 only transiently reversed the mechanical hypersensitivity caused by paclitaxel in mice and had no effect on grimacing. This transient effect is likely attributable to the pharmacokinetics of C781. We previously published that C781 achieves a rapid spike in plasma concentration with systemic administration and then is likely excreted via the kidneys, although we cannot rule out excretion via feces⁵². It is unlikely that the compound is metabolized in the liver because C781 has a long half-life in liver microsomes. Our pharmacological study should be viewed as proof of concept for the efficacy of a PAR2 antagonist that selectively blocks β-arrestin signaling downstream of the receptor for the treatment of CIPN⁸⁸. Future development of similar PAR2 antagonists with a more favorable pharmacokinetic profile will allow for longer target coverage in vivo. Based on our findings with PAR2 cKO mice, we hypothesize that such a molecule would have efficacy against pain endpoints, as well as reversing signs of small fiber neuropathy such as loss of intraepidermal nerve fibers.

In conclusion, our work provides clarity on how PAR2 is involved in CIPN. Sensory neuron expressed PAR2 is responsible for mechanical hypersensitivity, facial grimacing, intraepidermal nerve fiber loss and satellite cell gliosis caused by paclitaxel in mice. Our data with C781 suggests, in agreement with previous studies, that PAR2 can be targeted pharmacologically for the alleviation of CIPN pain even after chemotherapeutic treatment.

Perspective:

Our work demonstrates that PAR2 expressed in sensory neurons plays a key role in paclitaxel-induced mechanical allodynia, spontaneous pain and signs of neuropathy, suggesting PAR2 as a possible therapeutic target in multiple aspects of paclitaxel CIPN.

Highlights.

• Protease-activated receptor-2 (PAR2) knockout mice show reduced pain responses to paclitaxel treatment

• Paclitaxel-evoked pain and neuropathy is decreased in sensory neuron-specific PAR2 knockout mice

• Pharmacological targeting or PAR2 transiently reverses mechanical hypersensitivity in paclitaxel mice

• PAR2 is an attractive target for alleviation of paclitaxel neuropathy