A transient receptor potential vanilloid 1-dependent corneal–trigeminal neuroinflammatory circuit promotes corneal neuropathy
Manuela Pizzano, Alexia Vereertbrugghen, Maia J. Martinez Gomez, Juliana Bernatowiez, Douglas Vera Aguilar, Florencia Sabbione, Irene A. Keitelman, Federico Fuentes, Mirta N. Giordano, Analía S. Trevani, Jeremías G. Galletti
TL;DR
The study reveals a neuroinflammatory pathway involving TRPV1 channels that contributes to corneal neuropathy in dry eye disease, offering new therapeutic targets.
Contribution
The paper identifies a TRPV1-dependent corneal–trigeminal neuroinflammatory circuit as a novel mechanism in corneal neuropathy.
Findings
TRPV1 activation in dry eye disease triggers neuroinflammation and macrophage reactivity in the trigeminal ganglion.
Blocking substance P reverses TRPV1-driven corneal nerve abnormalities.
TRPV1 overactivation alone can cause corneal nerve degeneration and neuroinflammation.
Abstract
Corneal neurosensory abnormalities cause pain and discomfort in ocular surface disease, yet their pathophysiology is poorly understood. Here we show that in a mouse dry eye model, the ocular (over)activation of transient receptor potential vanilloid 1 (TRPV1) channels in response to tear deficiency and tissue damage promotes neuroinflammatory gene expression and macrophage reactivity in the trigeminal ganglion, where the cornea-innervating sensory neurons are located. This is accompanied by ocular surface macrophage activation, impaired corneal sensitivity to mechanical and non-TRPV1-mediated chemical stimulation, reduced corneal nerve density and the sensitization of ocular TRPV1 channels, thus establishing a vicious neurosensory cycle. Isolated corneal TRPV1 activation without ocular desiccation recapitulates macrophage reactivity, corneal nerve degeneration and trigeminal…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8- —https://doi.org/10.13039/100004440Wellcome Trust (Wellcome)
- —https://doi.org/10.13039/501100003074Ministry of Science, Technology and Productive Innovation, Argentina | Agencia Nacional de Promoción Científica y Tecnológica (National Agency for Science and Tech
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsOcular Surface and Contact Lens · Ion Channels and Receptors · Corneal Surgery and Treatments
Introduction
The cornea must remain clear, smooth, wet and uninflamed to fulfill its essential visual role^1^. Its abundant nerve fibers act as guardians of the ocular surface by sensing changes in wetness, temperature and tear osmolarity, detecting potential threats and regulating the local immune response accordingly^2^. Conversely, corneal nerve abnormalities, that is, corneal neuropathy, are observed across diverse ocular surface disorders, including dry eye disease (DED)^2,3^. An increasingly prevalent disorder, DED is caused by decreased tear quantity and/or quality and constitutes a chief complaint in ophthalmology^4,5^. Remarkably, corneal neuropathy is linked to the most burdensome symptoms of DED, which range from ocular discomfort to disabling pain^6^, yet its underlying pathophysiology is not fully understood.
Recent work has begun to unravel the pathogenic factors that drive DED-associated corneal neuropathy. Transient elevations of tear osmolarity (due to excessive evaporation) are sufficient to elicit corneal nerve abnormalities and T cell activation in rodents^7,8^ and DED-like pain in humans^9^. In line with this, activation of CD4^+^ T cells infiltrating the conjunctiva in response to ocular desiccation is required for nerve injury to occur^10–12^. Corneal transient receptor potential vanilloid 1 (TRPV1) signaling is also necessary for corneal nerve damage to propagate in the context of desiccation^13^. These channels are sensitive to hyperosmolarity and cell membrane-derived lipids released during inflammation and tissue damage^14^. In the cornea, TRPV1 is selectively expressed in polymodal nociceptors, one of the three main types of corneal sensory nerve fiber along with mechanonociceptors and cold thermoreceptors^2^. Enhanced corneal TRPV1 activation is likely in DED, as both patients and animal models exhibit increased ocular sensitivity to capsaicin (a selective TRPV1 agonist), and TRPV1 activation mediates ocular pain perception in this condition^13,15,16^. However, whether additional local or extraocular factors contribute to corneal nerve injury in DED is yet to be established.
Damage to sensory nerve endings in the cornea or in other tissues triggers an inflammatory response in the sensory ganglia housing the affected neurons^17,18^. This process is known as neuroinflammation and, in the case of corneal injury, affects the trigeminal ganglia and the brainstem^19–23^. Corneal chemical burns^19,24^, herpetic keratitis^25^ and DED^11,13,20,21^ present with trigeminal neuroinflammation. Although the extensive corneal nerve damage in these ocular disorders is a likely cause of trigeminal neuroinflammation, other mechanisms such as the persistent activation of nociceptors by tissue-derived inflammatory signals might also be at play^26^. At any rate, emerging evidence shows that trigeminal neuroinflammation is associated with peripheral and central sensitization, microglial activation and other synaptic plastic changes in the trigeminal brainstem nuclei^20,21,27^, all of which sustain the chronic ocular pain state^22^. Nonetheless, the exact mechanisms that lead to trigeminal neuroinflammation in DED are unknown. In parallel, the activation of polymodal nociceptors in the cornea^28,29^ and other tissues^30,31^ leads to the local release of proinflammatory substance P and other neuropeptides from the same nerve endings after antidromic activation. This phenomenon is known as neurogenic inflammation^32^, and animal models demonstrate that it compounds ocular surface inflammation in corneal chemical burns^28,33–35^, ocular graft-versus-host disease^36^ and DED^36–40^. Although TRPV1 signaling triggers neurogenic inflammation after a corneal alkali burn^28^, the afferent signal that leads to ocular neuropeptide release in DED remains uncharacterized. Moreover, whether neurogenic and trigeminal neuroinflammatory processes contribute to corneal neuropathy progression in DED has not been explored.
Here, we hypothesized that ocular TRPV1 activation is the initiating event of both trigeminal neuroinflammation and neurogenic inflammation in DED, and in turn, that these processes worsen corneal neuropathy and, ultimately, ocular surface disease (working hypothesis; Fig. 1a). We explored this in a unilateral DED model in mice that allowed us to separate the effects induced by ocular desiccation from those elicited by trigeminal neuroinflammation and neurogenic inflammation.Fig. 1. Unilateral dry eye does not lead to corneal epitheliopathy in the contralateral eye in both WT and Trpv1KO mice.a Working hypothesis: ocular TRPV1 activation is the initiating event of both trigeminal neuroinflammation and neurogenic inflammation in DED, and in turn, these processes worsen corneal neuropathy and, ultimately, ocular surface disease. b The right extraorbital lacrimal gland was excised in WT or Trpv1KO mice of both sexes, leading to unilateral DED. Thus, the right and left eyes are referred to as ipsilateral (Ipsi) and contralateral (Contra), respectively. Sham-operated animals were included as controls (Ct). c The tear production on day 5 as measured by phenol red paper-wetting length. d The cumulative data (left) and representative micrographs (right) of corneal dextran-fluorescein uptake in Ct and unilateral DED mice from both strains. Data are shown as the MFI calculated with ImageJ software (Materials and methods). e The representative micrographs (left) and number (right) of proliferating (Ki67^+^, green) cells within the epithelial basal layer of corneal wholemounts obtained 10 days after unilateral DED induction; E-cadherin (epithelial-specific) stained in red. All experiments were performed twice or more with six mice per group per experiment. To compare means, two-way ANOVA was used for d, e (strain and treatment) with Dunnett’s post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant.
Materials and methods
Mice
C57BL/6 (C57BL/6NCrl) mice were originally obtained from Charles River Laboratories. Trpv1-knockout (Trpv1KO, B6.129×1-Trpv1^tm1Jul^/J, JAX stock no. 003770) and recombination-activating gene 1 (Rag1)-knockout (Rag1KO, B6.129S7-Rag1tm1Mom/J, JAX stock no. 002216) mice were purchased from The Jackson Laboratory. Mice were bred and maintained at the Institute of Experimental Medicine’s conventional animal facility. All mice were 6–8 weeks old at the beginning of the experiments, and both male and female mice were included. All protocols were approved by the Institute of Experimental Medicine animal ethics committee (approval no. 084/2020) and adhered to the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research.
Reagents and antibodies
Supplementary Table 1 lists all antibodies and the most relevant reagents. Unless otherwise specified, all chemical and biological reagents were from Sigma-Aldrich.
Lacrimal gland excision surgery
Mice were anesthetized by intraperitoneal injection of ketamine (80 mg/kg) and xylazine (8 mg/kg) and placed on a heated pad. The extraorbital lacrimal glands from both sides were excised sequentially in the bilateral DED model, whereas only the right gland was removed in the unilateral model. The excision surgery comprised four steps^13^: first, a 3-mm-long incision was made along the middle third of the line joining the lateral canthus of the ear and the pinna; second, the superior pole of the extraorbital lacrimal gland was exposed by incising the ensheathing fibrous capsule; third, the lacrimal gland was pulled out gently and excised, taking special care not to damage the blood vessels next to its inferior pole; and fourth, the skin was closed using 6-0 nylon thread. Sham surgery consisted of only steps 1 and 4. In all cases, a single dose of 10 mg/kg diclofenac sodium was injected subcutaneously in the scruff for postoperative analgesia, and ciprofloxacin ointment was applied over the wound once the surgery was completed. The eyes were protected from desiccation with 0.4% sodium hyaluronate (Dropstar LC, Laboratorio Poen) until the mice recovered from anesthesia.
Corneal topical treatment with capsaicin
The mice were anesthesized as described above. Topical anesthesia (5 μl ophthalmic 0.5% proparacaine, Poen-caína, Laboratorio Poen) was applied onto the right eye and 0.4% sodium hyaluronate on the left eye. While the mouse was on a heated pad, a 2-mm-diameter filter paper disk embedded in 0.5 mg/ml capsaicin solution (50 mg/ml stock solution in dimethylsulfoxide (DMSO) diluted in PBS before use) or vehicle was applied onto the central right cornea. After 5 min, the paper disk was removed, and the right eye was washed with 3 ml of PBS and then covered with 0.4% sodium hyaluronate until the mouse recovered from anesthesia.
Substance P antagonist treatment
In several experiments, 5 μl of fosaprepitant (10 mg/ml in PBS) was applied onto the contralateral (left) eye of mice immediately after completing the unilateral DED-inducing surgery on the right side. After the mice recovered from anesthesia, fosaprepitant treatment was continued as four times per day for 10 days. Control mice received the same volume of PBS.
Tear production measurement
Tear production was measured by inserting a 1-mm-wide phenol red-impregnated filter paper strip in the inferior conjunctival fornix adjacent to the lateral canthus, where it was held in place for 60 s while restraining the mouse gently and enabling normal blinking. The wetted length of the strip was used as a data point. The right eye of each mouse was measured at least 2 h after the left eye to rule out the irritative effect of the contralateral test.
Assessment of corneal epithelial barrier function
Corneal fluorescein uptake of dextran-fluorescein isothiocyanate was used as an indicator of epithelial integrity^10,13,41^. In brief, 0.5 μl of dextran-fluorescein isothiocyanate (average molecular weight, 3000 to 5000; 10 mg/ml in PBS) was applied to each eye, and then, the mouse was returned to its cage. After 3 min, a 10-s-long video of each eye under blue light was captured with the aid of a fluorescence-adapted dissection microscope (NightSea SFA-RB; Electron Microscopy Sciences). For analysis, a masked observer (M.P.) exported a representative video frame as an image and selected the corneal area suitable for analysis, excluding reflections and other artifacts, using ImageJ software version 2.15.0 (NIH; https://imagej.net/software/fiji). Then, the green channel was extracted and the mean fluorescence intensity (MFI) within the resulting region of interest was calculated after background subtraction (50-pixel rolling ball radius).
Assessment of corneal mechanical sensitivity
Mechanical thresholds were determined using a mouse-adapted version of Cochet-Bonnet esthesiometry^2,10,11,13^. Nylon 6-0 monofilament was cut into segments of varying lengths (1.0–5.5 cm in 0.5-cm steps). With the mouse held firmly in one hand, the cornea was touched six times with each filament, starting with the longest segment. A positive response was defined as blinking and retraction of the eye in reaction to at least three of the six tries. The longest segment yielding a positive response was used as the sensitivity threshold. Corneal sensitivity was measured in the morning (8:00 to 11:00) before any other experimental handling.
Eye-closing ratio
One mouse at a time was placed on an elevated platform and allowed to habituate for 2 min. Then, a >1-min-long video was recorded with a camera placed at the same height. For analysis, a masked observer (M.P.) selected snapshots in which each eye was clearly visible. The distance between canthi (x) and between the upper and lower lids (y) was measured using ImageJ software, and then the corresponding eye-closing ratio was calculated as y/x. At least two snapshots per eye were analyzed, and then, the results from both eyes were averaged to obtain one data point per mouse.
Ocular capsaicin and AITC sensitivity
Eye-wiping behavior was measured in response to 100 μmol/l capsaicin (10 mg/ml stock solution in ethanol, diluted 1:328 in PBS right before use) and 1 mmol/l allyl isothiocyanate (AITC; 1 mg/ml stock solution in DMSO, diluted 1:10 in PBS right before use). Immediately after applying 2.5 μl of solution onto either the left or right eye, the mouse was placed in a separate cage and recorded with a camera placed above for at least 45 s. For the analysis, a masked observer (M.P.) counted the number of eye wipes during the first 30 (capsaicin) or 40 s (AITC) using a slow playback speed. AITC sensitivity was measured at least 3 h after capsaicin stimulation, and for each stimulus, left eyes were challenged at least 2 h before the right eyes.
Preparation and flow cytometry analysis of conjunctival cell suspensions
Conjunctivas were collected in serum-free RPMI 1640 medium, minced with scissors, digested with collagenase and DNAse, filtered, stained and fixed as previously described^10,13^. The entire cell suspension resulting from one eye was stained and acquired as one independent sample on a Cytek Northern Lights cytometer (Cytek) and analyzed using FlowJo software (FlowJo v10.3, Treestar).
Collection of eye tissue and trigeminal ganglia for imaging
After killing, the enucleation was performed by gently proptosing the eye globe and cutting the optic nerve with curved scissors. The two eyes of each mouse were collected separately in ice-cold formaldehyde-containing buffer for 75 min, washed and stored in methanol at −20 °C until processed for staining. Mice were killed one at a time so that all ocular tissue was collected within 5 min of the time of death to ensure adequate corneal nerve preservation^42^. For trigeminal ganglion imaging, the mice were euthanized and then transcardially perfused with 4% formaldehyde in PBS (10 ml) before dissecting the skull, removing the brain and exposing the base of the skull. The trigeminal glanglia were carefully dissected, then fixed in 4% paraformaldehyde for 1 h, washed in PBS and finally stored in 30% sucrose at 4 °C until processed for cryosectioning.
Corneal immunostaining and imaging
The eyes were processed as previously described^10,13^. After blocking, the dissected corneas were stained overnight with anti-tubulin β3, anti-mouse/human Ki67 and anti-mouse/human CD324 (E-cadherin) antibodies. The image acquisition was performed with a FluoView FV1000 confocal microscope (Olympus) equipped with Plapon 60×/1.42 and UPlanSapo 20×/0.75 objectives. Z stacks spanning the entire corneal epithelium were acquired and analyzed at three different levels^10,13^. For epithelial cell turnover analysis, a blind observer (M.P., J.B.) selected a single section from the Z stack encompassing the basal epithelial cells and manually counted the number of Ki67^+^ cells. For the macrophage analysis, a Z projection encompassing ten slices (1-μm step size) beneath the corneal epithelium was created, background corrected (30-μm rolling ball) and thresholded with ImageJ software before using the analyze particles tool (minimum size set to 250 μm^2^) to calculate the number of cells per field and their individual MFI.
Trigeminal ganglion immunostaining and imaging
Trigeminal ganglia were embedded in optimal cutting temperature compound for cryosectioning. The staining was done at room temperature: trigeminal ganglion sections were first hydrated with PBS + 5% fetal calf serum for 30 min, then permeabilized with PBS + 1% fetal calf serum + 0.4% Triton X-100 for 1 h and blocked with PBS + 5% bovine serum albumin for 1 h. Tissue sections were incubated with primary antibodies diluted in blocking buffer for 90 min and protected from light, followed by three 15-min washes with PBS, and finally mounted as for corneal samples. Image acquisition was performed with an Olympus IX83 inverted motorized microscope (Olympus) equipped with a UPlanSapo 10×/0.4 objective and a disk scanning unit. Trigeminal ganglion macrophages were analyzed as described for corneal macrophages.
RNA isolation from trigeminal ganglia and RNA-seq analysis
The trigeminal ganglia were dissected after killing and cardiac perfusion with PBS to remove contaminating blood cells^43^, collected in ice-cold TRI Reagent and stored at −80 °C until processing. For RNA isolation, each sample (containing both trigeminal ganglia from one mouse in the bilateral model and either the right or left ganglion from one mouse in the unilateral model) was homogenized in 1 ml of TRI Reagent, and then, 0.2 ml of isopropanol was added. After centrifugation, RNA was purified from the aqueous phase using the Direct-zol RNA MiniPrep kit (R2052, Zymo Research) following the manufacturer’s instructions. The concentration and purity of RNA were assessed with a NanoDrop 1000 spectrophotometer (ThermoFisher Scientific). RNA sequencing (RNA-seq) was performed by NovoGene using the Illumina NovaSeq platform to generate 150-bp paired-end reads. The sequenced reads were mapped to the mouse reference genome (assembly GRCm38/mm10) using STAR v2.7.11a, and the quantification of reads per gene was estimated by RSEM v1.3.1. Subsequent analyses were performed on the iDEP platform^44^. Gene filtering was performed by selecting features with ≥0.5 counts per million reads in at least three samples. For quality controls, the filtered data were subjected to a principal component analysis and hierarchical clustering (Ward.D2 method, Euclidean distance) after trimmed mean of M-values normalization (edgeR v4.0.16). The gene information (ENSEMBL ID, external gene name and gene biotype) was extracted using biomaRt (v2.58.2). Differential gene expression analyses were performed by the DESeq2 methodology (v1.42.1). Genes with log_2_ fold change >0.263 and adjusted P value <0.05 were considered as upregulated, and genes with log_2_ fold change <−0.263 and adjusted P value <0.05 were categorized as downregulated. The list of genes was ranked by log_10_ adjusted P value × sign(log_2_ fold change) and subjected to a Gene Set Enrichment Analysis with the Gene Ontology database, using enrichGO and simplify (clusterProfiler v4.10.1). Enriched Gene Ontology terms with adjusted P value of 0.05 were considered significant. All raw data files are available at ArrayExpress (https://www.ebi.ac.uk/biostudies/arrayexpress), accession number E-MTAB-15193.
Statistical analysis
Student’s t-test and one- or two-way analysis of variance (ANOVA) with Dunnet’s or Sidak’s post hoc tests were used to compare the means of two or more samples, respectively. Significance was set at P < 0.05, and two-tailed tests were used in all experiments. All data are shown as mean ± standard error of measurement and each data point represents one animal. Calculations were performed using GraphPad Prism version 9 software (GraphPad Software).
Results
Contralateral changes in corneal epithelium and macrophages after unilateral DED in WT and Trpv1KO mice
We have previously reported that Trpv1KO mice are shielded from DED-induced corneal neuropathy^13^. Remarkably, the protective effect was not as evident in the superficial nerve endings interspersed among the apicalmost corneal epithelial cells, which are directly exposed to the abnormal tear film in DED, as it was in the intraepithelial nerves situated deeper within the corneal epithelium. This pattern suggested that TRPV1-mediated signaling within the nerve endings propagates corneal axonal loss in DED. As the overactivation of these channels promotes neuroinflammation in nonocular tissues^45^, we hypothesized that TRPV1-initiated trigeminal neuroinflammation and neurogenic inflammation could be pathogenic factors that worsen the ocular surface phenotype in DED (Fig. 1a). Supporting our hypothesis, it has been shown that corneal and trigeminal inflammation induced by a unilateral corneal alkali burn spread to the contralateral trigeminal ganglion and eye through a corneal–trigeminal axis^19,28^. Whether there is similar intereye crosstalk in DED is unknown, in part owing to the typically bilateral presentation of this disorder. To investigate this, we resorted to a unilateral (right) extraorbital lacrimal gland excision model to separate ocular surface changes induced directly by desiccation from those potentially instigated indirectly in the contralateral (left) side (Fig. 1b). First, we verified that the reduction in tear production was comparable in the operated (desiccated) side of wild-type (WT) and Trpv1KO DED mice (Fig. 1c). By contrast, tear production did not change significantly in the opposite (nondesiccated) eye of either strain (Fig. 1c), indicating that any contralateral effects could not be ascribed to desiccation. Then, we quantified corneal epitheliopathy, a typical DED finding characterized by barrier disruption and increased cell turnover in the corneal epithelial layer^46^. As previously described^13,29^, both WT and Trpv1KO mice exhibited increased corneal dye uptake in the desiccated eye on days 5 and 10, a sign of worsening epithelial barrier function (Fig. 1d). However, dye uptake in the contralateral eyes of both strains did not change from their baseline levels (Fig. 1d). Consistently, we observed a similar increase in cell proliferation (Ki67^+^) in the corneal epithelium of the desiccated eyes of WT and Trpv1KO mice after 10 days, whereas there was no change in the contralateral eyes of either strain (Fig. 1e).
Next, we examined corneal and conjunctival macrophages after 10 days of DED because they are critical to ocular surface homeostasis, particularly of corneal nerves^47,48^, and respond to desiccation-induced inflammation^49–52^. The confocal microscopy of corneal macrophages revealed their number did not change in the contralateral eyes of unilateral DED mice of either strain (Supplementary Fig. 1a). However, their expression of CD206, a marker associated with an antiinflammatory role^53,54^, decreased only in the contralateral WT eyes (Supplementary Figs. 1b,c). Consistently, the proportion of conjunctival macrophages (CD45^+^ CD11b^+^ F4/80^+^ cells) in the contralateral eyes of WT and Trpv1KO mice with unilateral DED remained the same (Supplementary Fig. 1d). However, their expression of MHC-II and CD64, two proinflammatory markers^53^, increased in the desiccated eyes of both strains but only in the contralateral eyes of WT mice (Supplementary Fig. 1e). Thus, our results show that unilateral DED progression is not accompanied by the typical corneal epithelial disease signs in the contralateral eye of either strain. However, contralateral macrophage activation is detectable in the cornea and conjunctiva of WT but not Trpv1KO mice, suggesting TRPV1-dependent spreading of subclinical inflammation to the contralateral ocular surface via trigeminal inflammatory signaling.
WT but not Trpv1KO mice with unilateral DED develop corneal nerve dysfunction in the contralateral eye
Corneal neurosensory abnormalities are another frequent finding in DED that contribute greatly to patients’ symptoms and may develop independently of corneal epitheliopathy^2,6,11,55^. Moreover, if contralateral reactivity in ocular surface macrophages were triggered by trigeminal inflammatory signaling, corneal neurosensory changes would also be expected. Therefore, we also assessed this aspect of disease in the contralateral eye of the unilateral DED model. We first measured corneal mechanosensitivity, a widely used clinical measurement of corneal nerve function (Fig. 2a,b). We reported that the decreased corneal mechanosensitivity characterizes the corneal neuropathy in this model and that Trpv1KO mice display slightly higher baseline corneal mechanosensitivity^11,13^. As previously described, WT mice experienced a progressive reduction in corneal mechanosensitivity in the operated side, whereas Trpv1KO mice did not^13^. Most of the mechanosensitivity impairment in the desiccated eye of WT mice took place by day 5 of DED induction and remained until day 10. Regarding the nondesiccated side, WT unilateral DED mice had normal mechanosensitivity in the contralateral eye on day 5 but showed a reduction by day 10, which amounted to a statistically significant dysfunction but of lesser magnitude and with lower tempo than in the desiccated eyes. The contralateral eyes of Trpv1KO unilateral DED mice did not show any change, indicating that the phenomenon was entirely dependent on TRPV1 signaling. We also measured corneal sensitivity to capsaicin (a TRPV1 agonist) as an indicator of polymodal nociceptor function, which increases (sensitization) in the bilateral DED model^11,13^. On day 5 of unilateral DED induction, capsaicin sensitivity increased in the ipsilateral eyes but not in the contralateral eyes of WT mice (Fig. 2c). However, the contralateral eyes of WT mice developed comparably increased capsaicin sensitivity by day 10 of DED induction, reflecting again the delayed pattern observed for mechanosensitivity. As expected, Trpv1KO mice were indifferent to capsaicin. We also assessed the eye-closing ratio as an indicator of spontaneous ocular pain in mice, which depends on TRPV1 activation^13^. The WT strain showed decreased ratios in the operated eye but not in the contralateral eye, suggesting that corneal neurosensory abnormalities in the nondesiccated side did not include spontaneous pain. By contrast, the Trpv1KO strain did not display this sign of ocular pain in either eye (Fig. 2d), highlighting the role of this channel in pain perception^13,29^. Finally, we measured corneal sensitivity to AITC, a selective agonist of the transient receptor potential ankyrin 1 (TRPA1) channels that are also expressed in polymodal nociceptor fibers. Baseline TRPA1 sensitivity was higher in WT than in Trpv1KO mice (Fig. 2e,f). In addition, both strains displayed signs of habituation, that is, decreased responses after repeated TRPA1 sensitivity testing, as previously reported^56^. Nonetheless, the operated eyes of WT mice developed a further decrease in TRPA1 sensitivity, whereas the contralateral eyes of WT mice and both eyes of Trpv1KO mice with unilateral DED remained unchanged compared the the same-strain control littermates (Fig. 2e,f). Altogether, these findings show that the contralateral eyes of WT mice with unilateral DED develop corneal nerve dysfunction, whereas Trpv1KO mice are completely protected. Remarkably, DED-associated corneal nerve dysfunction entails increased TRPV1 sensitivity, whereas Piezo2- and TRPA1-mediated sensory modalities are impaired, suggesting that TRPV1 overactivation probably takes place as the disease progresses.Fig. 2. Unilateral dry eye leads to corneal nerve dysfunction in the contralateral eye of WT but not Trpv1KO mice.a, b Corneal mechanosensitivity in WT (a) and Trpv1KO (b) mice on days 0, 5 and 10 of unilateral DED induction shown as individual data points (left) and overall progression curves (right, the dotted line corresponds to Ct mice of the same strain). c The ocular capsaicin sensitivity in WT mice on days 0, 5 and 10 of unilateral DED induction shown as individual data points (left) and overall progression curves (right, the dotted line corresponds to WT Ct mice). d The eye-closing ratio in WT and Trpv1KO mice on day 10 of unilateral DED induction, shown as cumulative data (left) and representative video frames (right). e, f The ocular AITC sensitivity in WT (e) and Trpv1KO (f) mice on days 0, 5 and 10 of unilateral DED induction shown as individual data points (left) and overall progression curves (right, the dotted line corresponds to Ct mice of the same strain). The right extraorbital lacrimal gland was excised in WT or Trpv1KO mice of both sexes, leading to unilateral DED. Thus, the right and left eyes are referred to as ipsilateral (Ipsi) and contralateral (Contra), respectively. Sham-operated animals were included as controls (Ct). All experiments were performed twice or more with six mice per group per experiment. To compare means, two-way ANOVA was applied with Dunnett’s post hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. ns, not significant.
As corneal neurosensory dysfunction often correlates with morphological changes in the nerve fibers, we also analyzed the intraepithelial corneal innervation pattern after 10 days of DED induction. As reported elsewhere^13,57^, Trpv1KO mice had higher innervation density at the subapical level and lower at the subbasal level than WT mice at baseline (Fig. 3a–d). Regardless of these strain-specific differences, unilateral DED induction reduced nerve density at the three levels analyzed (subapical, midepithelial and subbasal) in the ipsilateral eyes of WT mice. However, it only decreased the density of the most superficial (subapical) nerve endings in Trpv1KO mice. These findings in the desiccated eye of unilateral DED were consistent with previous reports on bilateral DED models^11,13,58,59^. By contrast, the contralateral eyes of WT mice displayed reduced nerve density at the midepithelial and subbasal but not at the subapical levels, whereas there were no detectable changes in the innervation of the contralateral eyes of Trpv1KO mice. Altogether, these findings confirm that DED-induced corneal neurosensory abnormalities spread to the contralateral eye in WT but not in Trpv1KO mice, underscoring the key role of TRPV1 signaling in this phenomenon.Fig. 3. Unilateral dry eye leads to corneal nerve abnormalities in the contralateral eye of WT but not Trpv1KO mice.a–d The quantification of intraepithelial corneal innervation at three different levels evidenced by anti-tubulin β3 immunostaining of corneal wholemounts after 10 days of DED induction: subapical (a) (the percent area occupied by nerve endings) and midepithelial (b) (count of nerve endings per field) nerve ending density and complexity of subbasal nerves (c) (sum of intersections at all Sholl radii) and representative micrographs (d). The right extraorbital lacrimal gland was excised in WT or Trpv1KO mice of both sexes, leading to unilateral DED. Thus, the right and left eyes are referred to as ipsilateral (Ipsi) and contralateral (Contra), respectively. Sham-operated animals were included as controls (Ct). All experiments were performed twice or more with six mice per group per experiment. To compare means, two-way ANOVA was used with Dunnett’s post hoc test. *P < 0.05, **P < 0.01. ns, not significant.
Trpv1KO mice with DED have a reduced trigeminal neuroinflammatory transcriptomic signature
The delayed onset of contralateral abnormalities in corneal nerve function and structure and ocular surface macrophages in WT but not Trpv1KO mice with unilateral DED suggested spreading via TRPV1-dependent trigeminal neuroinflammatory signaling (Fig. 1a). Peripheral nerve injury in the cornea elicits molecular and functional changes in primary sensory neurons and neuroinflammation in the trigeminal ganglia^3,11,13,60^, but whether this is mediated by TRPV1 signaling in DED is unknown. As TRPV1 is a tissue-damage and inflammation-sensing channel and TRPV1 activation is associated with neuroinflammation, we hypothesized that Trpv1KO mice would develop less trigeminal neuroinflammation than WT mice in response to ocular desiccation. To investigate this, we performed bulk RNA-seq of trigeminal ganglia obtained 10 days after DED induction by bilateral extraorbital lacrimal gland excision in WT and Trpv1KO mice (Fig. 4a). Sham-operated mice were included as controls. We have previously characterized this model, which develops extensive corneal neuropathy findings in WT but not in Trpv1KO mice^13^. Trigeminal gene expression was considerably influenced by strain (Fig. 4b), which might be explained by the unique molecular signature of TRPV1^+^ trigeminal nociceptive neurons^61^. Comparing the same-treatment mice of both strains (Fig. 4c, volcano plots), there were 3121 differentially expressed genes (DEGs) in control mice (1400 up- and 1721 downregulated genes in WT mice) and 5231 DEGs in DED mice (2441 up- and 2790 downregulated genes in WT mice). Considering all 16,622 analyzed genes (Fig. 4d), 6108 (36.8%) were differentially expressed in control and/or DED WT mice relative to same-treatment Trpv1KO mice. Of these, 2247 (36.8%) were differentially expressed in both control and DED mice, which suggested strain- and not treatment-specific differences in gene expression. All but one (Shisa3) of the common DEGs were regulated in the same direction in control and DED WT mice relative to the same-treatment Trpv1KO mice, confirming that their differential expression was mainly due to strain-specific factors. Among the 3861 treatment-specific remaining DEGs, 876 (22.7%) were exclusive to control WT versus Trpv1KO mice, and 2985 (77.3%) were exclusive to DED WT versus Trpv1KO mice.Fig. 4. Gene expression changes induced by DED in the trigeminal ganglion of WT versus Trpv1KO mice.a DED was induced surgically by bilateral extraorbital lacrimal gland excision in WT or Trpv1KO mice for 10 days, and then, the trigeminal ganglia were collected for bulk RNA-seq analysis (female mice, n = 3 per group). Sham-operated (Ct) mice were used as controls, and differential gene expression was calculated between same-treatment mice (either Ct or DED) of the two strains. b A principal component analysis plot of the four experimental groups. c Volcano plots of DEGs (fold change >1.2, adjusted P value <0.05) between the same-treatment mice of the two strains. Upregulated genes in WT mice are shown in red, and downregulated genes are shown in blue. d The number of non-DEGs and DEGs that were detected in both WT Ct versus Trpv1KO Ct and WT DED versus Trpv1KO DED (strain-specific), only in WT Ct versus Trpv1KO Ct (Ct-specific) and only in WT DED versus Trpv1KO DED (DED-specific) analyses. e The proportion of DEGs annotated in the Gene Ontology database as inflammatory process- (GO:0006954) and immune response-related (GO:0006955). f The gene set enrichment analysis (Gene Ontology Biological Process) showing the 30 most significantly up- (red) and downregulated (blue) pathways.
To better focus on DED-induced changes and not strain-specific differences in gene expression, we restricted the analysis to DEGs that were exclusive to either control or DED mice (Fig. 4e, bar graph of inflammatory and immune response genes). According to the Gene Ontology database, there was a comparable fraction of DEGs annotated as inflammatory process-related (GO:0006954) in control and DED WT mice relative to same-treatment Trpv1KO mice; however, there was a larger fraction of DEGs annotated as immune response-related (GO:0006955) in DED WT mice than in Ct WT mice relative to same-treatment Trpv1KO mice. These treatment-specific DEGs between control and DED WT and Trpv1KO mice are shown in Supplementary Fig. 2. Inflammation-associated genes that were upregulated exclusively in DED WT mice included: Ccl2 and Ccr2, a proinflammatory chemokine axis associated with macrophage infiltration, neuropathic pain and neuroinflammation in other peripheral trigeminal nerve injury models^62,63^; the trigeminal ganglion macrophage markers Cd68 and Cd74^64^; Il1b, the proinflammatory cytokine interleukin 1β; Cyba and Cybb, components of the reactive oxygen species-generating NADPH oxidase complex in neutrophils, monocytes and macrophages; Hmgb1 and Hmgb2, proinflammatory alarmins released by nervous tissue-resident cells in response to injury^65^; and P2ry14, a purinergic receptor linked to inflammatory pain in the trigeminal ganglion and increased IL-1β and CCL2 in satellite cells^66,67^. Several immune response-associated genes that were upregulated only in DED WT mice were the T and B cell markers Cd3d, Cd3e, Cd8a, Cd19, Cd79a, Cd79b, Ighg2c and Igkc and several major histocompatibility complex H2 genes.
An unbiased gene set enrichment analysis (Fig. 4f, Gene Ontology molecular function and biological process databases) supported our hypothesis. Compared with control Trpv1KO mice, gene expression in control WT mice was enriched in biological processes associated with natural killer cell cytokine production and the regulation of the immune response by immunoglobulin. By contrast, in DED WT mice, it was enriched in immune response-related signaling, MHC class II-mediated antigen presentation and B cell proliferation when compared with DED Trpv1KO mice. Genes involved in complement receptor and CXCR and CCR1 chemokine receptor signaling were also upregulated in DED WT mice (Supplementary Fig. 3). Collectively, our findings suggest that DED induction leads to less trigeminal neuroinflammation in Trpv1KO mice than in WT mice. However, whether trigeminal neuroinflammation spreads, thus possibly impacting the ocular surface phenotype in this disease, remained undetermined.
WT but not Trpv1KO mice develop contralateral trigeminal neuroinflammatory and macrophage responses after unilateral DED
In the setting of unilateral corneal lesions, the resulting ipsilateral trigeminal neuroinflammation spreads to the opposite side via a corneal–trigeminal axis^19,28^. As we found more trigeminal neuroinflammation in WT than in Trpv1KO mice with bilateral DED, we hypothesized that the contralateral corneal neurosensory abnormalities that we observed in WT but not Trpv1KO mice with unilateral DED could be instigated antidromically by more severe contralateral spread of trigeminal neuroinflammation facilitated by TRPV1 signaling on the desiccated side. To investigate this, we performed bulk RNA-seq of the contralateral trigeminal ganglia obtained 10 days after unilateral DED induction in WT and Trpv1KO mice and included sham-operated mice as reference (Fig. 5a). As for the bilateral DED model (Fig. 4), gene expression in the left trigeminal ganglia of WT and Trpv1KO mice was influenced by strain independently of desiccation acting on the contralateral eye^61^ (Fig. 5b). To account for this, we resorted to the strategy applied to the bilateral DED model: we compared gene expression in mice from both strains receiving the same treatment (Fig. 5c). There were 3305 DEGs between control mice (1509 up- and 1796 downregulated genes in WT mice) and 1818 DEGs between unilateral DED mice (717 up- and 1101 downregulated genes in WT mice) of the two strains. Considering all 17,092 analyzed genes (Fig. 5d), 4476 (26.2%) were differentially expressed in control and/or DED WT mice relative to Trpv1KO mice under the same treatment. Of these 4476 DEGs, 647 (3.8%) were differentially expressed in both control and DED mice, which suggested strain- and not treatment-specific differences in gene expression. All but 34 common DEGs were regulated in the same direction in control and DED WT mice relative to same-treatment Trpv1KO mice, confirming that their differential expression was due mainly to strain-specific factors. Among the remaining 3829 treatment-specific DEGs, 2658 (69.4%) were exclusive to control WT versus Trpv1KO mice, and 1171 (30.6%) were exclusive to WT versus Trpv1KO mice with unilateral DED in the opposite side.Fig. 5. Gene expression changes induced by unilateral DED in the contralateral trigeminal ganglion of WT versus Trpv1KO mice.a Unilateral DED was surgically induced in WT or Trpv1KO mice of both sexes by excising only the right extraorbital lacrimal gland. After 10 days, the contralateral (Contra) trigeminal ganglia were collected for bulk RNA-seq analysis (female mice, n = 3 per group). Sham-operated (Ct) mice were used as controls, and differential gene expression was calculated between same-treatment mice (either Ct or DED) of the two strains. b A principal component analysis plot of the four experimental groups. c Volcano plots of DEGs (fold change >1.2, adjusted P value <0.05) between the same-treatment mice of the two strains. Upregulated genes in WT mice are shown in red, and downregulated genes are shown in blue. d The number of non-DEGs and DEGs that were detected in both WT Ct versus Trpv1KO Ct and WT Contra versus Trpv1KO Contra (strain-specific), only in WT Ct versus Trpv1KO Ct (Ct-specific) and only in WT Contra versus Trpv1KO Contra (Contra-specific) analyses. e The proportion of DEGs annotated in the Gene Ontology database as inflammatory process- (GO:0006954) and immune response-related (GO:0006955). f The gene set enrichment analysis (Gene Ontology Biological Process) showing the 30 most significantly up- (red) and downregulated (blue) pathways.
As before, we restricted the analysis to DEGs that were exclusive to either control or contralateral trigeminal ganglia to focus on contralateral DED-induced changes and not strain-specific differences. Based on the Gene Ontology database, the fraction of inflammatory process-related DEGs was similar in control and unilateral DED WT mice relative to same-treatment Trpv1KO mice (Fig. 5e). However, there was a larger fraction of immune response-related DEGs in unilateral DED WT mice than in Ct WT mice relative to same-treatment Trpv1KO mice. Treatment-specific DEGs between control and unilateral DED WT and Trpv1KO mice are shown in Supplementary Fig. 4. Inflammation- and immune response-associated genes that were upregulated exclusively in the contralateral trigeminal ganglia of unilateral DED WT mice included: Ccl21a, a chemokine that favors T cell homing; Vip, an immunomodulatory neuropeptide; and complement-associated C1ra, C3, CD46 and Cfh. These findings are consistent with a neuroinflammatory transcriptional signature^68^. The unbiased gene set enrichment analysis (Fig. 5f, Gene Ontology molecular function and biological process databases) aligned with our hypothesis. Compared with the same-treatment Trpv1KO mice, the gene expression in control WT mice was enriched mainly in non-immune related biological processes, except for T cell activation via antigen presentation. By contrast, the contralateral trigeminal ganglia from unilateral DED WT mice were enriched in T cell polarization, myeloid cell chemotaxis and alternative complement activation (Supplementary Fig. 5).
Next, we analyzed the number of trigeminal ganglion macrophages and their expression of activation markers (Supplementary Fig. 6a), because these cells react to peripheral nerve injury in other models of trigeminal neuroinflammation and contribute to neuropathic pain^69,70^. After 10 days, there was no change in the number of macrophages in the contralateral trigeminal ganglion of unilateral DED mice of both strains (Supplementary Fig. 6b). However, MHC-II expression decreased, whereas CD206 levels increased in the contralateral trigeminal ganglion macrophages of WT mice with unilateral DED (Supplementary Fig. 6c,d), in line with the transcriptomic signature conducive to alternative macrophage activation (upregulation of Cypa, Casp6, Gas6, Il33, Romo1 and Vip)^71–76^. By contrast, both markers increased in the contralateral trigeminal ganglion macrophages of Trpv1KO mice with unilateral DED, indicating activation but no clear polarization^77^. Altogether, our findings show that unilateral DED induction leads to significantly more neuroinflammation-associated gene expression changes^68^ and alternative macrophage activation in the contralateral trigeminal ganglion of WT than in Trpv1KO mice, which may in turn account for the altered corneal neurosensory phenotype in the contralateral eye previously found only in WT mice.
Unilateral corneal topical treatment with capsaicin induces corneal neuropathy in the contralateral eye
Considering the trigeminal neuroinflammation and corneal neuropathy-associated findings in the opposite side of WT mice with unilateral DED and the lack thereof in the Trpv1KO strain, we reasoned that corneal TRPV1 signaling in the desiccated eye is a key initiating event. However, DED elicits TRPV1-independent ocular inflammation and immune activation, which may also promote corneal neuropathy and trigeminal neuroinflammation^10,11^. Therefore, as a proof of concept, we resorted to unilateral corneal topical treatment with capsaicin to induce isolated TRPV1 activation without the additional inflammation elicited by ocular desiccation (Fig. 6a). The right cornea of WT mice was topically treated with vehicle- or capsaicin-embedded filter paper disks on days 0 and 2. After 4 days of corneal capsaicin exposure (Fig. 6b), there were neither appreciable changes in the external eye appearance nor increased corneal dye uptake, indicating that the corneal epithelial barrier was not compromised by this treatment. By contrast, capsaicin treatment led to the marked degeneration of corneal nerve fibers in the exposed eye (Fig. 6c), as it has been reported in other extraocular tissues^78,79^. In line with this, capsaicin treatment decreased corneal mechanosensitivity in the exposed eye after 4 days, and more importantly, a smaller but consistent reduction was also detectable in the contralateral eye (Fig. 6d), thus supporting our hypothesis. Contrasting with ocular desiccation (Fig. 2c), unilateral capsaicin treatment decreased capsaicin sensitivity 24 h later in the exposed eye (Fig. 6e). However, capsaicin sensitivity consistently increased in the left eyes of mice exposed to capsaicin in their right corneas (Fig. 6e), the same pattern observed in the contralateral eyes of the unilateral desiccation model (Fig. 2c). Of note, the quantification of the eye-closing ratio revealed that capsaicin-treated mice did not suffer spontaneous pain in either eye (Fig. 6f). Next, we analyzed the corneal nerve morphology on both sides. As expected, capsaicin-exposed eyes evidenced a significant reduction in corneal nerve fiber density at the subapical, midepithelial and subbasal levels (Fig. 6g–i). The contralateral eyes, however, only showed a drop in the subapical nerve endings. To better characterize the contralateral sensorineural abnormalities, we followed another cohort of capsaicin-treated mice for 15 days (Supplementary Fig. 7a–d). We found these contralateral effects to be reversible but with different tempos. Capsaicin sensitivity returned to baseline levels faster than mechanosensitivity, whereas the drop in subapical nerve density peaked later and did not completely recover during this follow-up period. Finally, we examined conjunctival and corneal macrophages owing to their intimate relationship with corneal nerve fibers and their regulatory role on corneal nerve function. In line with what we previously observed in the unilateral DED model (Supplementary Fig. 1), the number of corneal macrophages did not change in the contralateral eyes of mice receiving unilateral ocular capsaicin treatment (Supplementary Fig. 8a) but their CD206 expression was reduced (Supplementary Fig. 8b,c). Consistently, conjunctival macrophages did not increase in number (Supplementary Fig. 8d,e) but upregulated their expression of MHC-II and CD64, two prototypical M1-like proinflammatory markers (Supplementary Fig. 8f).Fig. 6. Unilateral corneal topical treatment with capsaicin induces corneal neuropathy in the contralateral eye.a Unilateral corneal topical treatment with capsaicin (Cap) was performed in WT mice by applying a 2-mm filter paper disk soaked in 0.5 mg/ml Cap or vehicle (Ct) onto the right cornea for 5 min under anesthesia (days 0 and 2). b Representative micrographs (day 4) of external eye appearance (left) and corneal epithelial dye uptake (right). c Representative micrographs of Ct and Cap-treated right corneas with anti-tubulin β3 (nerve-specific) staining. d–f The corneal mechanosensitivity thresholds (d), ocular Cap sensitivity (e) and eye-closing ratio (f) as measured separately on the right and left sides of Ct and Cap-treated mice. g–i The density of intraepithelial corneal innervation in anti-tubulin β3 staining of corneal wholemounts from the right and left eyes of Ct and Cap-treated mice assessed at three different levels: subapical (g) (percent area occupied by nerve endings) and midepithelial (h) (count of nerve endings per field) nerve ending density and complexity of subbasal nerves (i) (sum of intersections at all Sholl radii). All experiments were performed twice or more with six mice per group per experiment. To compare means, two-way ANOVA was used with Dunnett’s post hoc test. *P < 0.05, ***P < 0.001, ****P < 0.0001. ns, not significant.
Finally, because we previously observed DED-induced upregulation of T and B cell-associated genes in the trigeminal ganglion of WT but not Trpv1KO mice (Fig. 4 and Supplementary Figs. 2 and 3), we repeated the experiment in Rag1KO mice (lacking T and B cells) to delineate the adaptive immune response’s role in the corneal–trigeminal axis induced by unilateral corneal capsaicin exposure (Supplementary Fig. 9a). Corneal mechanical and capsaicin sensitivity dropped in the capsaicin-exposed eyes of Rag1KO mice (Supplementary Fig. 9b,c) comparably with WT mice (Fig. 6d,e). By contrast, the effects on corneal nerve function observed in the contrateral eyes of WT were not replicated in Rag1KO mice: the corneal mechanical sensitivity did not decrease (Supplementary Fig. 9b) and capsaicin sensitivity did not increase (Supplementary Fig. 9c). Thus, the capsaicin-induced effects in the treated (right) eye are not dependent on adaptive immunity, whereas the TRPV1-initiated, neuroinflammation-driven abnormalities in corneal nerve function in the contralateral eye do rely on T or B cells. Collectively, these results highlight the critical role of corneal TRPV1 signaling in promoting trigeminal neuroinflammation, corneal neuropathy and ocular surface macrophage activation.
Local substance P blockade ameliorates contralateral corneal nerve dysfunction associated with unilateral DED
Substance P is a proinflammatory neuropeptide that modulates the ocular surface immune response in DED and other disorders^33,37,39,40,80–82^. As TRPV1 activation in nociceptive fibers is often accompanied by the release of substance P from the same nerve endings, and this neuropeptide enhances TRPV1 activity in primary sensory neurons^83^, we hypothesized that TRPV1-induced substance P release in our model promotes corneal nerve dysfunction independently from ocular desiccation. To test this, we applied a substance P receptor antagonist topically onto the contralateral eyes of mice with unilateral DED (Fig. 7a). First, we verified that substance P blockade did not impair corneal epithelial barrier function (Fig. 7b) nor did it increase cell turnover (Ki67^+^ cells, Fig. 7c), indicating that it did not affect corneal epithelial integrity. Next, we assessed its effect on corneal neurosensory function. Compared with the saline instillation, substance P blockade ameliorated the impairment in corneal mechanosensitivity that develops in the contralateral eyes of unilateral DED mice over 10 days (Fig. 7d). Moreover, it abrogated the sensitization to capsaicin (TRPV1-specific) stimulation in the contralateral eyes on days 5 and 10 (Fig. 7e). However, substance P blockade did not improve ocular pain in the contralateral eyes of DED mice, as the eye-closing ratio was not significantly different from their saline-treated cagemates (Fig. 7f). Finally, we analyzed corneal nerve morphology in the same contralateral eyes. Contrasting with the findings in Trpv1KO mice, substance P blockade over 10 days did not prevent the reduction in corneal nerve density at the subapical, midepithelial and subbasal levels (Fig. 7g–j). Altogether, these results indicate that substance P release is one of the effector mechanisms in the corneal–trigeminal axis initiated by ocular TRPV1 activation that leads to corneal neurosensory abnormalities in DED. However, several aspects seem to be substance P-independent, which contrasts with the critical initiating role of TRPV1 activation in this neuroinflammatory circuit.Fig. 7. Local substance P blockade ameliorates contralateral corneal nerve dysfunction associated with unilateral dry eye.a Unilateral DED was surgically induced in WT mice of both sexes by excising the right extraorbital lacrimal gland only and and either 10 mg/ml fosaprepitant (SPbl), a neurokinin-1 receptor antagonist or saline (PBS) was applied four times per day for 10 days on the contralateral (Contra) eye starting immediately after the surgery. b Cumulative data (left) and representative micrographs (right) of corneal dextran-fluorescein uptake in the Contra eyes of PBS and SP blocker-treated mice. Data are shown as the MFI calculated with ImageJ software (Materials and methods). c The number (left) of proliferating (Ki67^+^, green) cells within the epithelial basal layer and representative micrographs (right) of corneal wholemounts obtained 10 days after DED induction. E-cadherin (epithelial-specific) stained in red. d, e The corneal mechanosensitivity (d) and ocular capsaicin sensitivity (e) in Contra eyes of PBS and SP blocker-treated mice on days 0, 5 and 10 of unilateral DED induction shown as individual data points (left) and overall progression curves (right). f The eye-closing ratio in Contra eyes of PBS and SP blocker-treated mice on day 10 of unilateral DED induction. g–j The quantification of intraepithelial corneal innervation at three different levels by anti-tubulin β3 staining of corneal wholemounts of the PBS- and SP blocker-treated Contra eyes after 10 days of DED induction: subapical (g) (percent area occupied by nerve endings) and midepithelial (h) (count of nerve endings per field) nerve ending density and complexity of subbasal nerves (i) (sum of intersections at all Sholl radii) and representative micrographs (j). In b–h the black and red dotted lines correspond to sham-operated mice and ipsilateral eyes of unilateral DED eyes, respectively. To compare means, two-way ANOVA was used for b c and e (time and treatment) with Dunnett’s post hoc test, and Student’s t-test was applied in c and f–i. *P < 0.05, **P < 0.01. ns, not significant.
Discussion
Neurosensory abnormalities are a defining aspect of ocular surface disease, yet the processes that underlie corneal neuropathy in DED and other disorders remain incompletely understood^2,55^. Central and peripheral sensitization of corneal nociceptors contribute to the increased ocular pain and discomfort symptoms in affected patients, but whether neuroinflammation per se affects the disease course in the ocular surface is unknown. Here, we uncover a pathogenic TRPV1-dependent mechanism in DED by which ocular desiccation elicits trigeminal neuroinflammation, which in turn propagates corneal nerve dysfunction and contributes to disease progression (Fig. 8). These results have profound implications for the pathophysiology of ocular surface disorders, showing that trigeminal neuroinflammation is not an epiphenomenon but a core mechanism and that ocular surface TRPV1 activation serves as a gatekeeper in the afferent arm of this neuroimmune pathway.Fig. 8. Proposed model.Corneal desiccation in DED (or any inflammatory stimulus in other ocular surface disorders) affects corneal nerve endings and causes overactivation of TRPV1 channels in these fibers. Orthodromic TRPV1-initiated signaling serves as a danger signal to sensory neurons and leads to trigeminal ganglion neuroinflammation, which involves the activation of neuron-associated macrophages (Mφ) and adaptive immunity. The bilateral trigeminal neuroinflammation is fostered by intereye crosstalk through intercommissural neurons in the brainstem, which may facilitate the neuroinflammatory spread in the setting of unilateral corneal injury. In turn, the neuroinflammation and TRPV1-initiated activation of trigeminal sensory neurons induce antidromic signaling to the ocular surface, leading to proinflammatory neuropeptide release and the worsening of corneal dysfunction. Thus, a TRPV1-driven corneal–trigeminal neuroinflammatory circuit that promotes corneal neuropathy is established.
The development of trigeminal neuroinflammation upon ocular desiccation has been observed in different DED models^11,13,20,21,24,33^, and there is mounting evidence that it contributes to peripheral and central pain mechanisms in this disease^27,84^. The published gene and protein expression patterns indicate that neuronal stress responses, satellite glial cell activation, macrophage infiltration and an adaptive immune response take place in the trigeminal ganglion in DED^11,20,23^. In line with these reports, we found a DED-induced transcriptomic signature in the trigeminal ganglion of WT but not Trpv1KO mice that involves neuroinflammation with macrophage activation (upregulation of Cd68, Cd74, Ccl2 and Ccr2, among other genes) and an adaptive immune component (upregulation of T and B cell-associated Cd3d, Cd3e, Cd8a, Cd19, Cd79a, Cd79b and immunoglobulin genes). These trigeminal gene expression changes are consistent with those reported in dorsal root ganglia after peripheral nerve ligation^68,85^. However, the inciting events in the DED-affected ocular surface remained unsolved. Here, we identify TRPV1 activation as a critical step conducive to trigeminal neuroinflammation, which is in line with a previous report showing that the ocular instillation of capsazepine, a nonselective TRPV1 antagonist, decreases the upregulation of inflammation-associated genes in the trigeminal ganglion^23^. Capsazepine, however, also blocks transient receptor potential vanilloid 4, ankyrin 1 and melastatin 8 channels, all of which are expressed in corneal nerve fibers, as well as nicotinic acetylcholine receptors^86–89^. Thus, our results conclusively indicate that ocular TRPV1 signaling is required for trigeminal neuroinflammation to develop upon DED onset.
Our finding also complements the previously described role of ocular TRPV1 signaling in propagating the corneal sensory nerve damage from the subapical endings, which are the most superficial and, consequently, the most affected by desiccation, toward the subbasal nerves^13^. Although the most straightforward interpretation of our first report is that TRPV1 overactivation potentiates nerve retraction locally, these new results indicate that a TRPV1-triggered neuropathogenic corneal–trigeminal axis also contributes to the propagation and maintenance of corneal neuropathy. TRPV1 channels are sensitive to tear hyperosmolarity, a common finding in patients with DED that results from increased tear evaporation and/or inadequate production^90^. In mice, transient tear hyperosmolarity without desiccation is sufficient to cause corneal nerve dysfunction^8^, and both TRPV1 gene knockout and pharmacological blockade prevent corneal neuropathy in a DED model^13^. However, TRPV1 channels are also gated by multiple endogenous, tissue damage-derived mediators and become sensitized in the context of inflammation^91^. Intriguingly, corneal neuropathy development in our DED model^11,13^ (Fig. 2) entails a progressive decrease in corneal mechanical sensitivity (mediated by Piezo2 channels) and AITC sensitivity (a TRPA1 agonist) with a concomitant increase in capsaicin sensitivity (a TRPV1 agonist). We put forth that TRPV1 channel sensitization is a key step in establishing a self-reinforcing cycle of ocular TRPV1 overactivation inducing corneal nerve damage, which in turn leads to enhanced TRPV1 function in the injured corneal fibers. Supporting this model, TRPV1 expression increases in the disrupted and adjacent trigeminal sensory neurons following mandibular nerve injury^92,93^. Ocular desiccation upregulates TRPV1 expression in TRPM8^+^ cold-sensing trigeminal neurons^15,29^, which constitutes the basis of cold allodynia in DED and probably contributes to the capsaicin sensitization phenomenon observed in our model (Fig. 2). The cooling of the ocular surface by tear evaporation, which increases during ocular desiccation^94^, might constitute another stimulus for corneal TRPV1 overactivation in DED. Thus, although the role of TRPV1 channels in the development of trigeminal neuroinflammation in DED is firmly established, the stimuli causing the overactivation of these channels are probably diverse and remain undefined. In the trigeminal ganglion, the upregulation of the Ccl2–Ccr2 axis (linked to macrophage activity), as we observed in DED WT mice, is associated with the hyperexcitability of small-diameter sensory neurons, which is in part mediated by increased TRPV1 expression^62,63^. Thus, increased ocular capsaicin sensitivity might be a telltale sign of trigeminal ganglion neuroinflammation, as we consistently observe this phenomenon in murine models whenever there is corneal nerve damage and independently of the disease context (unpublished observations). In line with this, one study found that capsaicin-elicited ocular pain sensitivity varies across clinical DED subtypes in patients^16^, but the correlation with corneal nerve morphology was not explored. More research on this aspect of DED in humans is needed.
Although the gene expression analysis of trigeminal ganglia in the bilateral DED model confirmed a neuroinflammatory transcriptomic signature, the corresponding analysis of the contralateral ganglion in the unilateral model revealed that the DED-induced neuroinflammation spreads to the other side. Although the systemic influence of bloodborne inflammatory mediators cannot be completely ruled out, the existence of a neural pathway for contralateral neuroinflammatory extension has already been outlined as a corneal–trigeminal axis in other ocular disease models. Unilateral DED in mice elicits neuroinflammatory changes in the ipsilateral trigeminal brainstem complex, which include microglial activation, neuronal stress responses and astrogliosis^20,22^. By contrast, unilateral chemical corneal injury models, which exhibit more severe ocular inflammation than DED, develop similar reactive changes in the contralateral trigeminal brainstem complex and trigeminal ganglion^19,24^. Here, we show that this also applies to unilateral DED in WT mice, where the most upregulated gene pathways in the contralateral trigeminal ganglion relate to leukocyte recruitment, alternative complement activation, antigen presentation and T cell activation. In line with this, neuron-associated macrophages in the contralateral trigeminal ganglion of unilateral DED WT mice expressed higher levels of M2 and lower levels of M1 markers (Supplementary Fig. 6), a pattern that has been observed in the post-acute phase of nonocular neuropathic pain models^95^. More importantly, we found that this contralateral trigeminal neuroinflammatory response is dependent on ocular TRPV1 signaling (Fig. 5), underscoring the key role of this channel in DED pathophysiology. We also observed that the adaptive immune response serves an amplifying role in this process: T and B cell-deficient mice develop TRPV1-initiated corneal nerve dysfunction in the capsaicin-treated eye but do not exhibit the same extent of functional impairment in the contralateral eye as WT mice (Supplementary Fig. 9). This finding aligns with the previously discussed macrophage and T and B cell-associated transcriptomic signature that we found in the trigeminal ganglia of WT DED but not Trpv1KO DED mice (Fig. 4). Thus, our data indicate that in DED, macrophage-coordinated adaptive immunity promotes TRPV1-initiated neuroinflammation in the trigeminal ganglion, in agreement with findings from nonocular neuropathic pain models concerning the dorsal root ganglia^70,96^.
Finally, our results also imply that TRPV1-initiated trigeminal neuroinflammation is not an epiphenomenon in DED because it contributes to the ocular surface phenotype and, conversely, that corneal nerve dysfunction in this disease is partly neurogenic. Although, strictly speaking, this was demonstrated using a unilateral disease model that separates desiccation from contralateral neurogenic effects, it is highly likely that the desiccated eye in this model is also affected by ipsilateral trigeminal neuroinflammation. Neurogenic inflammation resulting from antidromic activity in sensory nerve fibers followed by release of proinflammatory neuropeptides such as substance P and calcitonin gene-related product is well documented in other organs^97^. Trigeminal nociceptive sensory neurons coexpress TRPV1 and substance P^29,98^, and substance P-mediated neurogenic inflammation plays a role in ocular surface disorders including DED^38^. In line with this, our findings indicate that substance P mediates some of the ocular surface effects of TRPV1 overactivation because instillation of a blocker for this neuropeptide reduces corneal nerve dysfunction in the opposite eye (Fig. 7). In our model, contralateral substance P release necessarily involves dromic activity in the trigeminal sensory neurons innervating the capsaicin-stimulated ipsilateral cornea, followed by antidromic activity in the contralateral sensory neurons supplying the contralateral ocular surface. However, because topical substance P blockade also improves ocular surface phenotype in bilateral DED models^37–40^, we hypothesize that the local release of substance P from the same nerve fibers on which TRPV1 channels are activated constitutes yet another potential source of this neuropeptide. It should be mentioned that although substance P blockade ameliorates TRPV1-induced corneal nerve dysfunction (Fig. 7c,d), it does not prevent morphological abnormalities in the distal (subapical) nerve endings (Fig. 7f). This agrees with recent findings on the striking redundancy of substance P in pain perception and neurogenic inflammation^99^, as these phenomena are conserved in substance P-deficient mice. By contrast, TRPV1 signaling constitutes a critical, nonredundant event in DED-associated corneal neuropathy and ocular pain, as Trpv1KO mice with DED do not show signs of corneal neurodegeneration and eye discomfort^13^.
Collectively, this study shows that DED-induced trigeminal neuroinflammation is dependent on corneal TRPV1 signaling, worsens corneal neuropathy and has proinflammatory consequences on the ocular surface, some of which are mediated by substance P. Furthermore, in the rare setting of unilateral disease, this circuit may spread ocular surface inflammation and corneal neurosensory dysfunction to the other eye. The delineation of this pathogenic neuroinflammatory axis represents an interesting target for therapeutic intervention and sheds insight into the peripheral and central sensitization mechanisms that drive pain perception and neurosensory dysfunction in DED. We believe that strategies that impede the afferent arm of this neural circuit warrant further investigation.
Supplementary information
Supplementary Information