Establishing a Mouse Model of Chlorpromazine-Induced Corneal Trigeminal Denervation

Abstract

Purpose: This study aimed to establish a mouse model of chlorpromazine-induced corneal trigeminal denervation (CCTD).

Methods: Retrobulbar chlorpromazine injections were administered to 6- to 8-week-old C57BL/6j mice to induce corneal denervation. Additionally, apoptosis was assessed in isolated primary trigeminal ganglion cells after culturing in a conditioned medium containing chlorpromazine. Finally, the success rate of model generation, mortality and complication rates, and model-preparation learning curves were compared between the CCTD model and the electrocoagulation and axotomy models.

Results: Chlorpromazine retrobulbar injections resulted in trigeminal denervation, leading to a reduced blink reflex, corneal nerve density, and corneal epithelium thickness. Furthermore, 90% (9/10) of the mice developed epithelial defects, accompanied by increased apoptosis and inhibited proliferation of corneal epithelial cells. In vitro, trigeminal ganglion cell apoptosis increased after culturing in a conditioned medium containing chlorpromazine. Moreover, the CCTD model exhibited a higher success rate, longer survival rate, and lower complication rate compared to the electrocoagulation and axotomy models. Crucially, the learning curve demonstrated that the method used to generate the CCTD model was easy to learn.

Conclusions: The CCTD model is a user-friendly mouse model for studying corneal trigeminal denervation that offers a less invasive alternative to existing models.

Translational Relevance: The CCTD model serves as a valuable tool for investigating the functional mechanisms of corneal trigeminal nerves and their interactions with corneal cells.

Introduction

The cornea is the most densely innervated tissue in the human body, with approximately 10,000 nerve endings per square millimeter.¹ Sensory nerve fibers in the cornea mainly originate from the ophthalmic branch of the trigeminal ganglion (TG).² Myelinated nerve fibers lose their myelin sheath approximately 1 to 2 mm from the limbus and penetrate the corneal stromal layer to form a stromal nerve plexus.³ Then, part of the stromal nerve plexus develops into the subepithelial or subbasal nerve plexus.⁴ The close anatomical proximity between the corneal sensory nerve and corneal cells, including limbal stem cells, epithelia, and keratocytes, underscores their close physiological and pathological relationships.⁵ 

Corneal nerves play pivotal roles in regulating corneal sensation,³ maintaining epithelial integrity and proliferation, wound healing,⁶ influencing local corneal inflammation, and modulating immune responses.^7–9 Corneal innervation impairment can lead to neurotrophic keratopathy, which is characterized by persistent epithelial defects, progressive corneal melting, and, ultimately, perforation, potentially resulting in permanent vision loss or blindness.¹⁰ Consequently, various animal models of corneal trigeminal denervation have been developed in rabbits,¹¹ monkeys,¹² rats,¹³ and mice^14,15 using techniques such as hot metal probes, electrocoagulation needles, and microscopic tweezers to damage the ophthalmic branch of the trigeminal nerve through oral, cranial, and lateral conjunctival approaches. However, these models have limitations, as they are technically challenging, invasive, and associated with numerous complications related to the nervous system and eyeball. For example, surgical axotomy of the trigeminal nerve branch necessitates cutting the conjunctiva and lateral rectus to expose the nerves, and the small ciliary nerves are challenging to separate, increasing the risk of damage to the surrounding blood vessels.¹⁴ Stereotactic electrocoagulation requires electrode penetration into the cerebral cortex, inevitably causing physical damage to the brain tissue and raising the risk of intracranial infection.¹⁵ Furthermore, these models have exhibited low success rates and high mortality rates, limiting their widespread application.¹⁵ 

A previous study demonstrated the toxic effects of chlorpromazine on neuroblastoma cells in vitro.¹⁶ Additionally, retrobulbar chlorpromazine injection causes neurotrophic corneal ulcers.¹⁷ Consequently, we hypothesized that chlorpromazine might induce sensory denervation of the cornea. Notably, retrobulbar chlorpromazine injections induced minimal inflammation and damage to the surrounding tissues.¹⁸ Based on these observations, we developed a mouse model of chlorpromazine-induced corneal trigeminal denervation (CCTD) by retrobulbar chlorpromazine injections to address the limitations of the current models. 

Materials and Methods

Animals

C57BL/6 mice, 6 to 8 weeks old, were housed at the Experimental Animal Center of the Zhongshan Ophthalmic Center of Sun Yat-sen University. The study received approval from the Animal Experiment Ethics Committee of the Experimental Animal Center of the Zhongshan Ophthalmic Center of Sun Yat-sen University and the Third Affiliated Hospital of Guangzhou Medical University. All surgical procedures adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Chlorpromazine-Induced Corneal Trigeminal Denervation Model

Mice were anesthetized with 2% isoflurane and placed in a lateral position. Following disinfection of the upper eyelids with povidone, the eyeballs were protruded by gently pressing on the periorbital skin. A 34-gauge needle was vertically inserted into the retrobulbar space, with a penetration depth of 4 to 5 mm, situated in the outer third of the upper eyelid (Figs. 1A, 1B). Subsequently, a 10-µL solution of chlorpromazine (20 mg/mL) was injected steadily into the retrobulbar space.¹⁹ A sham control group was prepared by injecting equal volumes of saline. Finally, antibiotic ointment was applied to both eyes. 

Figure 1.

 

Retrobulbar chlorpromazine injections induce decreased blink reflexes and corneal epithelial defects. (A, B) Illustration of the chlorpromazine-induced corneal trigeminal denervation (CCTD) model. The needle was inserted vertically into the retrobulbar space at the outer one-third of the upper eyelid. (C) Fluorescein sodium staining revealed corneal defects following retrobulbar chlorpromazine injections. A dot-like corneal epithelial defect first appeared on day 1, progressed to a patch-like defect on days 2 and 3, and evolved into a corneal ulcer on days 5, 7, and 10.

It is important to note that we recommend a 34-gauge microsyringe, as it can be stabilized during the procedure to prevent damage to the surrounding tissues. Additionally, chlorpromazine possesses sedative properties and enhances the anesthesia while inhibiting the function of the hypothalamic thermoregulatory center in mice. Therefore, a lower degree of inhalational anesthesia is recommended during surgery. Following the procedure, the mice were placed on a heating pad and returned to their respective cages after resuscitation. 

Blink Reflex Assay

The blink reflex was assessed in all experimental mice using an 8-0 nylon thread under an anatomical microscope on postoperative days 1, 2, 3, 5, 7, and 10. 

Fluorescein Sodium Staining

A drop of fluorescein sodium was administrated to the conjunctival sacs, and slit-lamp microscope images were captured to document changes in corneal transparency, epithelial defects, edema, and neovascularization. 

Hematoxylin and Eosin, TUNEL, and Ki-67 Staining

The corneas were meticulously excised and fixed in 4% paraformaldehyde. Subsequently, the corneas were embedded in paraffin and sectioned into 8-µm-thick slices. These sections were stained with hematoxylin and eosin (H&E, #60524ES60; Yeasen Biotechnology, Shanghai, China) and a TUNEL staining kit (#12156792910; Roche, Basel, Switzerland) in accordance with the manufacturer's instructions. The sections were observed and imaged using a fluorescence microscope (Axio Imager Z2; Zeiss, Oberkochen, Germany). The thickness of the corneal epithelium was determined using ImageJ software (National Institutes of Health, Bethesda, MD). 

For Ki-67 immunofluorescence staining of the corneal cryosections to detect epithelial proliferation, 8-µm-thick cryosections were incubated with an anti-Ki-67 antibody (1:100, #ab279653; Abcam, Cambridge, UK) at 4°C overnight. The following day, the sections were incubated with a secondary antibody (1:500, #ab150118; Abcam) for 1 hour at room temperature and observed using a Zeiss confocal microscope. 

Whole-Mount Corneal Staining

The mouse eyeballs were fixed in Zamboni fixation solution (#RS2690; G-CLONE, Beijing, China) for 1 hour. Then, the corneas were fixed in Zamboni fixation solution for another hour and then sectioned into petals. Following this, permeabilization and blocking were performed using 0.3% Triton X-100 and 2% goat serum in 1× phosphate-buffered saline for 2 hours at room temperature. The corneas were incubated with III β-tubulin antibody (1:200, #ab18207; Abcam) at 4°C overnight. After three washes, the specimens were incubated with the secondary antibody (1:400, #11008; Thermo Fisher Scientific, Waltham, MA) for 1 hour at room temperature. Finally, whole-mount corneal samples were observed and imaged using a confocal microscope with a Zeiss Airyscan LSM 880 microscope. 

Cell Culture and Treatment

Primary TG cells were isolated following a previously reported protocol²⁰ and cultured in a neuronal basal medium supplemented with 10% fetal bovine serum (#10099-141; Thermo Fisher Scientific), 1% penicillin (# P1400; Solarbio Life Sciences, Beijing, China), 2% Gibco B-27 supplement (#17504044; Thermo Fisher Scientific), 1% non-essential amino acid solution (#N1250; Solarbio Life Sciences), and 1% HEPES (#H1095; Solarbio Life Sciences). Primary cells were fixed for immunofluorescence staining. For the chlorpromazine treatment, primary TG cells were treated with 10^–3, 10^–4, and 10^–5 M chlorpromazine for 40 minutes. Subsequently, TUNEL staining was performed according to the manufacturer's guidelines. 

Immunofluorescence Staining of TG Cells

TG cells were fixed in 4% paraformaldehyde, followed by permeabilization with 0.1% Triton X-100 for 30 minutes, then blocking with 5% goat serum for 30 minutes at room temperature. The cells were then incubated with III β-tubulin antibody (1:100, #ab18207; Abcam) at 4°C overnight. Next, the specimens were incubated with Alexa Fluor–conjugated secondary antibody (1:500, #A-11008; Thermo Fisher Scientific) for 1 hour at room temperature. Nuclei were counterstained using 4′,6-diamidino-2-phenylindole (DAPI, #FD9637; Fude Biological Technology, Hangzhou, China). Finally, the samples were observed and imaged using a Zeiss LSM 880 confocal microscope. 

Visual Electrophysiological Examination

Flash Electroretinogram

After mydriasis treatment and 30 minutes of dark adaptation, the mice were immobilized on the experimental platform under general anesthesia. The ring electrode was placed on the corneal surface, and the reference and grounding electrodes were placed on the same side of the cheek and tail. The amplitudes and peak times of the a- and b-waves were recorded following standard white flash stimulation (1.5–3.0 cd·s/m²). 

Flash Visual Evoked Potential

The recording electrode was placed at the midpoint of the line connecting the two ears. Reference and ground electrodes were located on the same side of the cheek and tail, respectively. The opposite eye was covered with black opaque material. Standard full-field white flash stimulation (1.5–3.0 cd·s/m², 1 Hz, 80 repetitions) was applied, followed by analysis of the amplitudes of the P2 and N2 waves. 

Model Procedure Learning Curves

The learning curve was described by the equation Y = A × X^–B, where Y represents the average time for completion per model when making the Xth model, A signifies the time required to complete the first model, X is the calculated number of completed models, and B represents the learning coefficient.²¹ In the experiment, Y, A, and X were recorded by three learners during the modeling procedure. Based on these results, the average learning times and number of animals required to complete three models were calculated and compared. 

Statistics

Data were analyzed using SPSS Statistics 25 (IBM, Chicago, IL). P values less than 0.05 were considered statistically significant. 

Results

Effects of Retrobulbar Chlorpromazine Injections

Following retrobulbar chlorpromazine injections for the establishment of the CCTD model, the success of the model was confirmed through the blink reflex and corneal fluorescein sodium staining experiments. The blink reflex was absent in 90% (9/10) of the CCTD model mice on all postoperative days, but the control mice maintained normal blink reflexes (Table 1). The corneal staining results revealed the presence of dot-like corneal epithelial defects and patch-like defects, as well as the progressive development of corneal ulcers, in the CCTD model on postoperative days 1 through 10. In contrast, the corneal epithelium remained intact in the control group throughout the study (Fig. 1C). Consequently, we successfully established a mouse corneal denervation model via retrobulbar chlorpromazine injections. 

Table 1.

 

Ratios of Absent Blink Reflexes in Total Mice

 

A denervated cornea, deprived of nerve nourishment, leads to neurotrophic keratitis.^14,15 Therefore, we conducted a TUNEL assay 3 days after the injections to identify apoptotic cells and investigate the effects of corneal denervation on corneal cells in the CCTD model. A higher number of TUNEL-positive cells were detected in the epithelium of the CCTD model mice compared to the control mice (Figs. 2A, 2B), indicating an increased incidence of apoptotic corneal cells following retrobulbar chlorpromazine injections. Furthermore, we observed a significant reduction (77%) in Ki-67–positive cells in the basal layer of the corneal epithelia of the CCTD model 3 days after the injections (Figs. 2C, 2D). Notably, the corneal epithelium was significantly thinner in the CCTD model mice compared to the control mice at the same time point (Figs. 2E, 2F). These findings further validate the successful corneal denervation in mice achieved through retrobulbar chlorpromazine injections. 

Figure 2.

 

Corneal epithelial apoptosis, proliferative inhibition, and decreased corneal epithelial thickness after retrobulbar chlorpromazine injections. (A) Immunohistochemistry and (B) quantification of TUNEL-positive apoptotic cells in the corneas of both control and CCTD mice. Scale bar: 30 µm. (C) Ki-67 staining of the cornea and (D) quantitative analysis of the Ki-67–positive cells comparing the control and CCTD model groups. Scale bar: 100 µm. (E) H&E staining of the corneas and (F) comparison of corneal epithelial thickness between the control and CCTD model groups. Scale bar: 100 µm. Data are presented as mean ± SD. *P < 0.05.

Corneal Nerve Density

Whole-mount corneal staining was performed to investigate variations in corneal nerve density in the CCTD model. In comparison to the control, both subbasal and stromal plexus nerve densities exhibited a significant reduction in the CCTD model 24 hours after the retrobulbar chlorpromazine injection. Moreover, no distinct nerve fibers were detected in the subbasal plexus or stromal plexus in the CCTD model over time (Fig. 3). This indicates that chlorpromazine has a toxic impact on the ophthalmic branch of the trigeminal nerve in vivo. 

Figure 3.

 

Reduction in corneal nerve density in the CCTD model. Whole-mount staining depicts corneal nerve density in the subbasal and the stromal plexus. Scale bar: 50 µm.

Chlorpromazine Treatment of TG Cells In Vitro

In order to investigate the effects of chlorpromazine in vitro, we isolated and cultured primary TG cells. The primary TG cells exhibited a fusiform morphology (Fig. 4A), consistent with the literature.²² Additionally, immunofluorescence staining of III β-tubulin was performed for cell identification (Fig. 4B). The III β-tubulin–positive cells were identified as TG cells.²³ Subsequently, primary TG cells were subjected to chlorpromazine treatment at different concentrations (10^–3, 10^–4, and 10^–5 M) for 40 minutes, followed by the TUNEL assay. We observed a notable increase in apoptotic cells at chlorpromazine concentrations of 10^–3 and 10^–4 M, but almost no apoptotic cells were observed at the 10^–5 M concentration (Figs. 4C, 4D). These results indicate that chlorpromazine is cytotoxic to TG cells in a dose-dependent manner in vitro. 

Figure 4.

 

Chlorpromazine increases apoptosis in primary TG cells in vitro. (A) Morphology of primary TG cells. Scale bar: 500 µm. (B) Immunofluorescence staining with DAPI (nuclear marker; blue) and III β-tubulin (neuronal marker; green) for primary TG cells. Scale bar: 300 µm. (C) TUNEL staining of primary TG cells treated with varying chlorpromazine concentrations. Scale bar: 300 µm. (D) Quantification of TUNEL-positive cells. Data are presented as mean ± SD. *P < 0.05.

CCTD, Electrocoagulation, and Axotomy Model Comparisons

We compared the novel CCTD model with the existing electrocoagulation and axotomy models. The CCTD model had the highest success rate (90%), followed by the axotomy (80%) and electrocoagulation (60%) models. The electrocoagulation model had the highest mortality rate (26.7%; 8/30 mice) due to brain lesions and infections, whereas the axotomy and CCTD models had lower mortality rates of 6.7% (2/30 mice) and 0% (0/30 mice), respectively (P < 0.05) (Table 2). 

Table 2.

 

Comparison of the Three Models

 

Moreover, we utilized flash visual evoked potential (f-VEP) and flash electroretinography (f-ERG) assays to evaluate the integrity of the optic pathway and retinal cell function and compare visual electrophysiology characteristics among the three models at 7 days after surgery. In the f-VEP assay, the P2–N2 wave reflects the visual function of the entire pathway from the retina to the visual cortex, and decreasing P2–N2 wave amplitudes indicate optic pathway damage.²⁴ The P2–N2 wave amplitudes were significantly lower in the axotomy and electrocoagulation model groups than in the control groups (Figs. 5A, 5C–5E). In contrast, almost no change was observed in the CCTD model group compared to the control group (Figs. 5A, 5B, 5E). Therefore, although the optic pathways in the axotomy and electrocoagulation models were severely damaged, the optic pathway was preserved in the CCTD mice. 

Figure 5.

 

Comparative analysis of the CCTD, electrocoagulation, and axotomy models. (A–D) Flash visual evoked potential (f-VEP) tests of the control (A), CCTD (B), axotomy (C), and electrocoagulation (D) models for assessing the integrity of the visual pathway. P2–N2 waves reflect the functionality of the visual pathway. (E) Quantification of the P2–N2 wave amplitudes. (F–I) Flash electroretinography (f-ERG) tests of the control (F), CCTD (G), axotomy (H), and electrocoagulation (I) model groups for evaluating retinal function. The a-wave reflects photoreceptor cell function, and the b-wave reflects bipolar cell function. (J) Quantification of the a-wave and b-wave times. (K) Quantification of the a-wave and b-wave amplitudes. Data are presented as mean ± SD. *P < 0.05.

In the f-ERG assay, the a-wave reflects photoreceptor cell function, and the b-wave reflects bipolar cell function. Decreasing a- and b-wave amplitudes indicate photoreceptor and bipolar cell dysfunction.²⁴ The amplitudes of both waves were significantly lower in the axotomy model group compared to the control group (Figs. 5F, 5H, 5K), indicating that the photoreceptor and bipolar cells were destroyed in the axotomy model. Minimal changes were observed in the CCTD model group (Figs. 5F, 5G, 5K). Additionally, the latencies of the a- and b-waves did not differ among the three model groups (Fig. 5J). 

Overall, photoreceptor and bipolar cells and the optic pathway integrity were severely damaged in the axotomy model. The optic pathway was also affected in the electrocoagulation model, but neither cell nor pathway damage occurred in the CCTD model. 

Learning Curve Results

We established learning curves following published guidelines to evaluate the difficulty of the three models.²¹ The CCTD model exhibited a shorter time to proficiency compared to the electrocoagulation and axotomy models (Fig. 6A). The average time required to complete the model procedure was 6.10 ± 1.34 minutes in the CCTD group, which was notably shorter than for the electrocoagulation (31.30 ± 5.62 minutes) and axotomy (11.40 ± 3.33 minutes) groups (Fig. 6B). Additionally, the number of animals required to successfully complete the model was 4.50 ± 1.10 in the CCTD model, fewer than for the electrocoagulation (10.50 ± 0.09 mice) and axotomy (7.30 ± 1.10 mice) models (Fig. 6C). Consequently, the CCTD model procedure is easier to learn compared to the other two models. 

Figure 6.

 

The learning curves of the CCTD, electrocoagulation, and axotomy models. (A) The learning curves of the three different approaches. (B) Average time required and (C) the number of animals needed to complete the models. Data are presented as mean ± SD. *P < 0.05.

Discussion

Chlorpromazine, widely known for its antagonistic effects on dopamine receptors as an antipsychotic, has shown a peculiar capacity to alleviate eye pain by retrobulbar injection,^18,25 as evidenced by clinical case records. For example, a 12-year-old patient with orbital gliosarcoma complained of unbearable pain that was relieved immediately after retrobulbar chlorpromazine injection.²⁶ Moreover, a retrospective study on patients that documented their blinded eyes with severe pain demonstrated a remarkable 80% relief rate at least 3 months after such injection.²⁷ A study by Ortiz et al.²⁸ further emphasized the efficacy of retrobulbar chlorpromazine injection in managing severe eye pain in blind patients; they reported 90% satisfaction with pain relief over an average 2.1-year follow-up period. Despite these promising observations, the precise mechanisms responsible for pain relief remain enigmatic. It is noteworthy that Hauck et al.¹⁷ reported instances of trigeminal denervation characterized by loss of corneal perception and corneal epithelial defects in an elderly patient experiencing retrobulbar chlorpromazine injection. Moreover, chlorpromazine injections into the sciatic nerve of rats resulted in severe damage, degeneration, and fewer myelinated nerve fibers,²⁹ and 10^–4 to 10^–3 M chlorpromazine solutions have been found to have direct and concentration-dependent toxic effects on neuroblastoma cells.¹⁶ These intriguing clinical insights and experimental observations highlight the possibility that retrobulbar chlorpromazine injection may affect the ophthalmic branch of the trigeminal nerves and inspired the development of our CCTD mouse model. 

In our CCTD model, we successfully induced damage to the ophthalmic branch of the trigeminal nerve, which resulted in corneal changes similar to those observed in prior corneal denervation models. These changes encompassed a decreased blink reflex and the emergence of corneal epithelial defects. Importantly, it is noteworthy that retrobulbar chlorpromazine injections predominantly affected the ophthalmic branch of the trigeminal nerve while sparing the optic nerve. Histological examination via H&E staining provided further support, indicating no discernible differences between the optic nerves of the CCTD model and those of the control group (Supplementary Fig. S1A). Electrophysiological examinations of the CCTD model confirmed the unimpaired functions of the retina and optic nerve (Fig. 5). The anatomical protection afforded by the meningeal layers encasing the intraorbital segment of the optic nerve likely explains this phenomenon, effectively safeguarding it from damage.³⁰ 

Although various models, such as axotomy and electrocoagulation, have been employed in corneal denervation studies, they have often presented complications involving surrounding tissues and intracranial infections. The CCTD model circumvents these issues by employing a relatively straightforward retrobulbar chlorpromazine injection, thus simplifying the experimental procedure, as evidenced by the efficient learning curve. This simple operation ensures minimal damage to the surrounding tissues, a fact corroborated by the electrophysiological results. Notably, our results indicated that the amplitudes of the a-, b-, and P2–N2 waves in the CCTD model remained largely unchanged when compared to the control group, signifying preservation of surrounding tissues, including the retina and optic nerve (Fig. 5). 

A considerable difference in mortality and long-term survival rates was observed in the electrocoagulation model, likely attributed to complications involving brain damage and infection. Consequently, this model is not recommended for long-term observational studies. In contrast, the CCTD model demonstrated an extended survival time for experimental mice, thereby offsetting the shortcomings of other models. As such, the CCTD model emerges as a valuable tool for examining the complex interactions between nerves and corneal cells. 

The precise mechanisms through which CCTD induces cell death in the trigeminal nerves remain an intriguing area for future investigation. Prior studies have illustrated the cytotoxic effects of chlorpromazine on tumor and glioma cells, including G2/M phase arrest and autophagic cell death via the inhibition of the phosphoinositide 3-kinase (PI3K)/protein kinase B (AKT)/mammalian target of rapamycin (mTOR) pathway.^31,32 Moreover, the induction of cytotoxicity in glial cells via the Ca²⁺ signaling pathway by chlorpromazine suggests the need for further mechanistic inquiries into the context of the trigeminal nerve.³³ Furthermore, extending the exploration of the chlorpromazine-induced corneal trigeminal denervation model to other species, such as rats, guinea pigs, and rabbits, appears both feasible and worthwhile. 

Acknowledgments

Supported by grants from the National Natural Science Foundation of China (81870631), the Natural Science Foundation of Guangdong Province (2023A1515012571), and the Guangzhou Basic and Applied Basic Research Municipal-School (College)-Enterprise Joint Funding Project (2023A03J0385). 

Disclosure: X. Lin, None; P. Xu, None; Y. Tian, None; H. Xiao, None; X. Dong, None; S. Wang, None