Norepinephrine as the Intrinsic Contributor to Contact Lens–Induced Pseudomonas aeruginosa Keratitis

Abstract

Purpose: Contact lens wear (CLW) is one of the leading risk factors for Pseudomonas aeruginosa keratitis (PAK). However, the intrinsic factors that contribute to the high susceptibility to keratitis during CLW remain to be elucidated. CLW over an extended period can elevate corneal norepinephrine (NE) concentration. In this study, we investigated the role of NE in promoting PAK.

Methods: We constructed an injury-induced PAK model and a CLW-induced PAK model to confirm the impact of NE during corneal infection. Pharmacological blockage of NE and gene knockdown mouse were used to investigate the downstream effector of NE. RNA sequencing was performed to explore the cellular alterations during NE treatment. Non-parametric Mann-Whitney U test or Kruskal-Wallis test were used to ascertain the significance (P < 0.05).

Results: Supplementation of NE led to PAK even without artificial corneal injury during CLW. The effect was mediated by the β2-adrenergic receptor (β2-AR) in the corneal epithelium. The β2-AR blockage by the NE antagonist ICI118,551 (ICI) or by deleting of its encoding gene Adrb2 significantly alleviated infection during CLW. Conversely, β2-AR activation compromised the integrity of the epithelium and significantly increased the cortical plaque marker ezrin. Transcriptome analysis identified that the protective effect of ICI on the keratitis was mediated by dual-specificity phosphatases. Suramin, a Dusp5 antagonist, abrogated the protective effect of ICI.

Conclusions: These data reveal a new mechanism by which NE acts as an intrinsic factor that promotes CLW-induced PAK and provide novel therapeutic targets for treating keratitis by targeting NE-β2-AR.

Pseudomonas aeruginosa keratitis (PAK) is a sight-threatening, rapidly progressive eye disease.¹ Without timely treatment, the infected cornea will suffer from necrosis and perforation. The current treatment for PAK involves repeated administration of antibiotics. P. aeruginosa is highly virulent and has been identified as one of the pathogens that urgently requires new antibiotic strategies by the World Health Organization.² 

Contact lens (CL) usage is a major risk factor for bacterial keratitis,³ with an incidence of two to 20 corneal infections per 10,000 wearers annually.⁴ The average age of individuals with CL wear (CLW)–related PAK (39.2 years) is significantly lower than that of non-CL wearers (71.9 years).⁵ Mechanical injuries, hypoxia, alteration of tear fluid composition and function, and alteration of microbial composition and virulence have all been proposed as contributors to serious CL-related corneal complications.⁴ 

In our previous study, we demonstrated that extended CLW in mice resulted in increased corneal norepinephrine (NE) levels, which facilitated P. aeruginosa adhesion and biofilm formation on the CL.⁶ Moreover, NE was found to exacerbate PAK severity in scratched corneas by enhancing the host's inflammatory response and bacterial virulence.⁷ NE is naturally secreted from sympathetic nerves that are present in the corneal epithelium and stroma in mice^8,9; however, sympathetic nerves are either absent or sparse in the corneas of primates and healthy humans.^10,11 Nevertheless, during herpes simplex virus stromal keratitis, sympathetic nerves have been observed to invade the sensory-deprived cornea.¹² The cornea is heavily innervated with sensory nerves from the trigeminal ganglion, which are distributed throughout the corneal epithelium,¹³ that not only play a role in perception but also secrete neuropeptides that nourish the surrounding epithelial cells.¹⁴ The degeneration of sensory nerves and invasion of sympathetic nerves can cause changes in neuropeptides and neurotrophic factors, leading to disruptions in cornea's homeostasis.¹⁴ Acute psychosocial stress has been reported to increase plasma NE levels and delay skin barrier function recovery.¹⁵ Interestingly, our previous study demonstrated that the addition of NE aids in the transepithelial penetration of riboflavin during corneal cross-linking in mice.¹⁶ A healthy cornea with an intact epithelium possesses robust resistance to bacterial invasion.¹⁷ However, once the epithelial layer is compromised, the cornea becomes susceptible to keratitis.¹⁸ Current mouse models for PAK depend on manual disruption of the corneal epithelial barrier.⁴ It is not well understood whether sympathetic nerves and NE can compromise the barrier function of the cornea during ocular surface challenges such as extended CLW and infection. Therefore the present study aimed to investigate whether NE acts as an intrinsic factor in promoting PAK during extended CLW. 

A previous study has identified that a Gram-negative bacterium Neisseria meningococcus induces the opening of intercellular junctions in endothelial cells to cross the blood-brain barrier.¹⁹ Similarly, other Gram-negative bacteria such as P. aeruginosa and Neisseria gonorrhoeae have the ability to recruit host actin and junction proteins to form cortical plaques that facilitate bacterial attachment and cell invasion.^20,21 N. meningococcus has been shown to delocalize junctional proteins via the β2-adrenergic receptor (β2-AR)/ β-arrestin pathway.¹⁹ The β2-AR is a downstream receptor of NE, whereas β-arrestin is a downstream mediator that facilitates β2-AR desensitization and internalization.²² Whether this pathway also plays a role in P. aeruginosa invasion of the cornea warrants further investigation. 

In this study, we successfully induced PAK in CLW mice by subconjunctival injection of NE instead of by manual disruption of the cornea. This confirms that NE is one of the intrinsic factors involved in CLW-related keratitis. Additionally, our study identified β2-AR as the downstream mediator of NE in the cornea. These results enhance our comprehension of the risk factors that contribute to the development of PAK in individuals with extended CLW and suggest a new therapeutic target for this condition. 

Material and Methods

Ethics

Animal experiments were conducted in compliance with the guidelines provided by The Association for Research in Vision and Ophthalmology and approved by the Animal Investigation Committee of Eye Institute of Shandong First Medical University with approval number SDSYKYJS.No20210123. 

Bacterial Strain

The reference strain of P. aeruginosa (ATCC 19660) was used to establish the keratitis models. P. aeruginosa was cultured in Luria-Bertani broth (LB; L8291; Solarbio Life Science, Beijing, China) and diluted to the appropriate concentration before use. For in vivo experiments, 10⁵ colony-forming units (CFU) of bacteria in 5 µL were administered, whereas for in vitro experiments, 10⁷ CFU of bacteria in 5 µL were used. 

For the construction of the green fluorescent protein (GFP) strain, the pBBR1MCS5-Tac-EGFP plasmids (P9677; MiaoLing Plasmid Sharing Platform) were used. For electronical transfection, 5 mL P. aeruginosa were spun in a centrifuge, and the bacterial pellets were washed twice with a 1 mM MgSO₄ solution. The pellets were resuspended in 200 µL of the 1 mM MgSO₄ solution. Plasmids of 3 to 4 µg were mixed with the bacteria in a sterile electroporation cuvette, and an electrical field (2.5 KV, 25 µF, 600 Ω) was applied using an electroporation system (Gene Pulser Xcell Electroporation System; Bio-Rad Life Science, Hercules, CA, USA). After electroporation, the cells were immediately incubated in 1 mL of BHI medium for five minutes at room temperature and subsequently rotated in a 37°C incubator for three hours. The culture was then spread onto a BHI plate containing 25 µg/mL of gentamicin, and single colonies exhibiting GFP fluorescence were selected for future study. 

Preparation of Contact Lenses and Bacterial Inoculation

Hilafilcon B soft contact lens (Bausch & Lomb, Rochester, NY, USA) were trimmed with a 3.5 mm trephine and polished with a microkeratome to fit the mouse eye. A 96-well plate was used, with each well containing 195 µL of LB broth and 5 µL of bacterial culture at a concentration of 2 × 10⁷ CFU/mL (10⁵ CFU in total) being added. The trimmed lens was then rinsed in the well and incubated at 37°C with agitation at 100 rpm for 24 hours to facilitate bacteria attachment. 

Contact Lens Wear Induced PAK Model

The contact lens-induced PAK model was established in mice as previously described.^6,23 Before the procedure, the mouse was anesthetized with 0.6% pentobarbital sodium and its whiskers and eyelashes were trimmed. Proparacaine hydrochloride ophthalmic solution (Alcaine; Alcon Laboratories, Inc., Geneva, Switzerland) was applied to the cornea for topical anesthesia. The bacteria-inoculated lens was then placed onto the mouse cornea, and the eyelids were closed with 10-0 suture. After 30 hours, the suture and lens were removed, and the cornea was examined using slit lamp microscopy and collected for further analysis. The severity of keratitis was scored using a previously published scale²⁴: 0 for a clear or slightly opaque cornea partially or fully covering the pupil; +1 for a slightly opaque cornea partially or fully covering the anterior segment; +2 for a dense opacity partially or fully covering the pupil; +3 for a dense opacity covering the entire anterior segment; +4 for corneal perforation or phthisis. 

The Combination Model of Contact Lens Wear and Scratch-Induced PAK

The mouse underwent the same preparation steps as described in the previous section. In addition, three parallel incisions of 1 mm each were made on the corneal epithelium using a sterile 25-gauge needle. The bacteria-inoculated lens was then placed onto the mouse cornea, and the eyelids were closed with 10-0 suture. After 30 hours, the suture and lens were removed, and the cornea was imaged and collected for further experiments. 

Drug Application

The drugs were administered via subconjunctival injection using a 33-gauge syringe after mouse anesthesia, immediately before contact lens placement. The volume and concentration of drugs applied to each group were as follows: NE group: 5 µL 0.5 mg/mL NE ((R)-(−)-norepinephrine L-bitartrate monohydrate, Grand Pharma, Wuhan, China); NE + ICI group: 5 µL 0.5 mg/mL NE and 5 µL 2 mg/mL ICI (HY-13951, ICI 118,551 hydrochloride; MedChemExpress, Monmouth Junction, NJ, USA); NE + ICI + Suramin group: 5 µL 0.5 mg/mL NE, 5 µL 2 mg/mL ICI, and 5 µL 3.5 mg/mL suramin (HY-B0879, MedChemExpress); Control group: 5 µL 0.9% saline solution. 

Adrb2 Knockout Mouse Model

Adrb2 knockout mice (B6/JGpt-Adrb2em1Cd/Gpt) with a C57BL/6JGpt background were generated by GemPharmatech Co., Ltd. (Nanjing, China). The knockout area of Adrb2 was screened using PCR primers JS00580-Adrb2-5wt-tF2 (5ʹ-ACTGCTCCAAGAAGCAGACTCTG-3ʹ) and JS00580-Adrb2-5wt-tR2 (5ʹ-AAGACGTGAGAGCACAGAAGAGC-3ʹ). The primers for wild-type Adrb2 were JS10580-Adrb2-wt-tF2 (5ʹ-ATCACAGCCATTGCCAAGTTC-3ʹ) and JS10580-Adrb2-wt-tR2 (5ʹ-AAGTCTCCTCGGTGTAACAATCG-3ʹ). TreliefTM Mouse Direct PCR Kit (TSE14; Tsingke Biotechnology Co., Beijing, China) was used for genotyping of the Adrb2 knockout mice. The tail tips of the mice were collected in a 1.5 mL centrifuge tube and processed according to the kit protocol for direct PCR amplification. PCR was performed using 2 × Rapid Taq Master Mix (P222-01; Vazyme Biotech Co., Ltd., Nanjing, China) with an initial denaturation at 95°C for five minutes, followed by denaturation at 95°C for 30 seconds, annealing at 58°C for 30 seconds, extension at 72°C for 30 seconds, and final extension at 72°C for three minutes for 40 cycles. 

Quantitative Real-Time PCR (qPCR)

Complementary DNA (cDNA) was synthesized using the PrimeScript RT reagent Kit (Perfect Real Time) (RR037A; TaKaRa Bio Inc., Shiga, Japan). The expression levels of Adrb2 were measured by qPCR using FastStart Universal SYBR Green Master (Rox) (4913714001; Roche, Basel, Switzerland) with the qPCR cycler Rotor-Gene Q (Qiagen, Hilden, Germany). The qPCR primers for Adrb2 were T007691-Adrb2-RT-qtF1 (5ʹ-ACGGCTACTCTAGCAATAGCAACG-3ʹ) and T007691-Adrb2-RT-qtR1 (5ʹ-GCTAAGGCTAGGCACAGTACCTTG-3ʹ). The qPCR program was set to an initial denaturation at 95°C for 10 minutes, followed by 40 cycles of 95°C for 15 seconds and 60°C for one minute. 

Cell Invasion Assay

The human corneal epithelial cell line (HCEC), which was provided as a gift by Prof. Choun-Ki Joo,²⁵ was used for the cell invasion assay. HCECs were seeded onto sterile glass coverslips in a 24-well plate and starved with DF12 supplemented with 0.1% FBS one day before drug treatment. The cells were then treated with 7.5 µL of 23.6 µM ICI, or 6 µL of 23.6 µM isoproterenol (ISO), along with 2 µL of 23.6 µM NE. Actin inhibition was achieved by treating the cells with 4 µL of 10 µM L-B (Latrunculin B, ab144291; Abcam, Cambridge, MA, USA). After a 30-minute drug treatment, the HCECs were co-cultured with 10⁷ CFU P. aeruginosa for 30 minutes at 37°C and subsequently cultured in fresh medium for an additional 2.5 hours before being collected for immunostaining. The CFU measurement was conducted by treating the cells with 50 µg/mL gentamicin (L1312; Solarbio Life Science) for one hour to eliminate extracellular bacteria, followed by lysis. 

Bacterial Load Measurement

Corneas at 48 hours after infection or cell samples were homogenized in 100 µL sterile saline solution. Serial dilutions of 100 µL aliquots were plated onto LB agar plates in triplicate and cultured at 37°C for 24 hours. The bacterial colonies were counted to determine the CFU of the sample. 

Immunofluorescence Staining of Frozen Sections

Wild-type mouse corneas were collected and embedded in Tissue-Tek O.C.T. compound (Sakura Finetek, Tokyo, Japan). The corneas were sliced into 7 µm–thick sections and fixed with 4% PFA for 15 minutes. After washing with PBS, the sections were circled with an immunohistochemistry pen, permeabilized with 0.3% Triton for five minutes, and blocked with goat serum for one hour. The Adrb2 antibody (EPR707[N], ab182136; Abcam) was diluted in antibody dilution solution (WB500D; New Cell & Molecular Biotech Co., Ltd, Suzhou, China) at a ratio of 1:200 and incubated overnight at 4°C. The sections were then washed and incubated with the secondary antibody (1:500, ab150076, Donkey Anti-Rabbit IgG H&L Alexa Fluor 594; Abcam) for 1.5 hours at room temperature. Finally, and sections were stained with DAPI (c0065; Solarbio Life Science) for 30 seconds, mounted with anti-fade mounting medium (s2100; Solarbio Life Science), and imaged using confocal laser microscopy (Zeiss LSM880; Zeiss, Oberkochen, Germany). Negative control sections were only stained with the secondary antibody. 

Immunofluorescence Staining of HCEC

For ezrin staining, after drug treatment, cells were washed three times with PBS and fixed with 200 µL of 4% PFA for 20 minutes. After washing, cells were permeabilized with 200 µL of 0.1% Triton for one minute, washed three times with PBS for five minutes each time, and blocked with goat serum overnight at 4°C. The Ezrin antibody (no. 3145; Cell Signaling Technology, Danvers, MA, USA) was diluted with antibody dilution solution (WB500D; New Cell & Molecular Biotech Co., Ltd) at a ratio of 1:200. The cells were incubated with the antibody at 4°C overnight. After washing, the cells were incubated with the secondary antibody (1:500, ab150076, Donkey Anti-Rabbit IgG H&L Alexa Fluor 594; Abcam) for 1.5 hours at room temperature and then stained with DAPI (c0065; Solarbio Life Science) for 30 seconds. Cells were mounted with anti-fade mounting medium (s2100; Solarbio Life Science) on a coverslip and imaged with an Echo Laboratories RVL-100-G Microscope (Discover Echo Inc., San Diego, CA, USA). Five visual fields including upper, lower, left, right, and center were selected from microscopic images, and the stained positive cells were manually counted. 

For ZO-1 and F-actin staining, the cell permeabilization time was five minutes with 200 µL of 0.1% Triton. The primary antibodies were ZO-1 (1:500, 40-2200; Invitrogen, Carlsbad, CA, USA), Phalloidin-iFluor 594 (1:500, ab176757; Abcam). The secondary antibody was Donkey Anti-Rabbit IgG H&L Alexa Fluor 488 (1:500, ab150073; Abcam). 

Western Blot

HCEC lysates were prepared using RIPA buffer (r0010; Santa Cruz Biotechnology, Dallas, TX, USA) supplemented with 1% phenylmethylsulfonyl fluoride. A volume of 50 µL lysate was added into each well of a 24-well plate and incubated for 15 minutes on ice. The lysate was collected by scraping the cells with a 1 mL pipette tip and transferred to a 1.5 mL centrifuge tube. After centrifugation at 12,000 rpm for 20 minutes at 4°C, the supernatant was collected and mixed with 5 × loading buffer. The mixture was then heated at 95°C for 10 minutes. The sample was either directly used for Western blot or stored at −80°C for future use. 

For Western blot analysis, protein samples were subjected to 10% acrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes for antibody incubation. Dusp5 antibody (1:1000, ab200708; Abcam) was used, and GAPDH antibody (1:1000, LK9002; Tianjin Sungene Biotech Co., Ltd, Tianjin, China) was used as a loading control. The protein bands were quantified by ImageJ grayscale image peak analysis, which involved selecting the bands with the rectangle tool after the images were inverted to the 8-bit format. 

Scratch-Induced PAK Mouse Model

The mouse model of PAK induced by corneal scratch was constructed as previous described.⁷ Briefly, C57BL/6J mice aged six to eight weeks old were anesthetized with 0.6% pentobarbital sodium and had their whiskers and eyelashes trimmed. Proparacaine hydrochloride ophthalmic solution (Alcaine; Alcon Laboratories, Inc.) was applied to the cornea for topical anesthesia. Using a sterile 25-gauge needle, three parallel 1 mm incisions were made on the cornea of their left eye. Next, 5 µL of 2 × 10⁷ CFU/mL P. aeruginosa culture (10⁵ CFU in total) were topically applied to the scratched cornea. 

Cornea Epithelium Collection and RNA Sequencing

The scratch-induced PAK models were established, and either ICI or ISO was applied to the cornea every two hours after infection as the treatment, whereas saline solution was applied as the control solution. After six hours of treatment, the mice were euthanized, and their enucleated eyeballs were placed in a dish filled with PBS. The eyeball was punched through the sclera outside the corneal limbus with a 1 mL syringe needle for incision. The cornea was dissected by a circular cut using Vannas scissors (MR-S121A; Suzhou Mingren, Jiangsu, China), and the iris was gently removed under a stereoscope. After washing with PBS, the cornea was placed on a clean glass slide with the epithelial side facing up. The epithelial layer was scraped off with a sharp blade, and the collected cells were transferred into a 1.5 mL centrifuge tube. Five corneal epithelia were placed in each tube, and 50 µL of Trizol was added to each tube. The samples were stored at −80°C for subsequent analysis. 

Total RNA was extracted from the corneal epithelial cells, and the concentration was measured using a NanoDrop ND-1000 instrument. A cDNA library was constructed by Aksomics Inc. (Shanghai, China) using KAPA Stranded RNA-Seq Library Prep Kit (Illumina, San Diego, CA, USA). The resulting libraries were assessed for quality with an Agilent 2100 Bioanalyzer (Agilent, Santa Clara, CA, USA), quantified using the absolute quantification qPCR method, and sequenced using the Illumina HiSeq 4000 instrument. 

Gene Profiling in the Single-Cell Sequencing Dataset

Single-cell sequencing dataset of the mouse cornea was obtained from a previous study.²⁶ The libraries were mapped to the mouse genome (mm10 build) using Cell Ranger software (version 3.1.0, 10 × Genomics).²⁷ The gene expressions of the ARs were calculated, and their average expression was depicted. 

Statistics

The non-parametric Mann-Whitney U test was used to compare the daily mouse corneal clinical scores and CFU values. The nonparametric Kruskal-Wallis test was used to compare the numbers of ezrin-P. aeruginosa–positive staining among ICI, ISO, NE-treated HCECs. A P value < 0.05 was considered statistically significant. 

Results

NE as the Intrinsic Factor During CLW-Related Infection

Our previous studies have revealed that extended CLW,⁶ corneal infection, and corneal injury⁷ could lead to an increase in the endogenous level of NE in the cornea (Supplementary Fig. S1). To investigate whether NE acts as an intrinsic factor that promotes P. aeruginosa infection during CLW, we attached P. aeruginosa–inoculated CL to the cornea of mice and administered subconjunctival injections of either NE or saline solution (Fig. 1A). When saline solution was used, no observable infection was detected in mice wearing P. aeruginosa–inoculated contact lenses, when no corneal scratch was made (Fig. 1B). However, these mice developed keratitis on corneal injury by scratching (Fig. 1B), indicating that the integrity of the integrity of the corneal epithelium is crucial in preventing infection. Conversely, when NE was administered, mice developed keratitis even without corneal injury (Fig. 1C), suggesting that NE plays a role in inducing spontaneous CLW-related keratitis. The severity of NE-induced PAK was similar to that of injury-induced PAK in the CLW mice (Fig. 1D), providing further support for the idea that NE is involved in the development of PAK during CLW. 

Figure 1.

 

NE induced P. aeruginosa keratitis without corneal injury. (A) A schematic diagram depicting the construction of the extended CLW-induced PAK mouse model. (B) Slit-lamp photos of mice after wearing P. aeruginosa contaminated CL for 30 hours. These mice received subconjunctival injection of saline solution, and one group had corneal epithelial scratches whereas the other group did not. Mice with corneal epithelial scratches displayed corneal infection (n = 4). (C) Slit-lamp photos of mice after wearing P. aeruginosa contaminated contact lens for 30 hours. Neither group had corneal scratches, but one group received subconjunctival injection of saline solution whereas the other received NE. Mice with subconjunctival injection of NE displayed corneal infection (n = 4). (D) Clinical scores of corneas shown in panels B and C (n = 4).

Epithelial Adrb2 is the Downstream Receptor of NE During Corneal Infection

NE is a receptor ligand to both α- and β-ARs. To determine which receptor is vital during P. aeruginosa infection, the expression of ARs was analyzed in a single-cell RNA sequencing dataset of the mouse cornea.²⁶ Of the nine ARs tested (Adra1a, Adra1b, Adra1d, Adra2a, Adra2b, Adra2c, Adrb1, Adrb2, Adrb3), only Adrb2 showed significant expression in the mouse cornea, with the highest concentration in the corneal epithelium (Fig. 2A). The localization of Adrb2 protein in the corneal epithelial layer was confirmed by immunostaining (Fig. 2B). 

Figure 2.

 

Adrb2 is enriched in corneal epithelium. (A) Normalized gene expression of ARs in mouse corneal epithelium based on single-cell sequencing dataset. Adrb2 shows the highest expression level among all ARs. (B) Immunofluorescent staining of Adrb2 (red) in mouse cornea. Adrb2 is mainly expressed in the superficial layer of corneal epithelium. (C) Negative control staining without primary antibody for (B).

To confirm the specificity of the screened receptor, we used CRISPR/Cas9 to generate a knockout mouse model for Adrb2, the gene encoding β2-AR (Supplementary Fig. S2A). There was barely detectable Adrb2 mRNA expression in the cornea samples collected from homozygous Adrb2^−/− mice (Supplementary Fig. S2B). Heterozygous Adrb2^+/− mice showed partial resistance to CLW-induced PAK even with the presence of NE (Fig. 3A). CFU results demonstrated that Adrb2 knockdown reduced bacterial burden in the cornea (Fig. 3B). These results provide evidence that epithelial Adrb2 is the downstream receptor of NE during CLW-induced PAK. 

Figure 3.

 

Heterozygous Adrb2 knockout partially rescued PAK. (A) Slit-lamp photos of mice wearing P. aeruginosa contaminated CL for 30 hours. Heterozygous Adrb2 knockout partially rescued corneal infection (n = 4). (B) CFU quantification in mouse corneas treated with NE and Adrb2 knockdown. Heterozygous Adrb2 knockout reduced CFU compared to wild-type littermates (n = 4). P = 0.029. Mann-Whitney U test.

Adrb2 Activation Induced Paracellular Permeability Via Modulating Cytoskeleton

To explore the molecular mechanisms underlying Adrb2-mediated corneal infection, we topically applied the β2-AR agonist ISO, antagonist ICI 118,551,²⁸ and saline solution to the cornea after inducing PAK by scratch injury. We then performed transcriptomic analysis on epithelial cells to identify changes in gene expression profiles. 

We observed that the top upregulated gene ontology term in the ICI-treated epithelium was actin cytoskeleton organization (Fig. 4A). Therefore we stained the cortical plaque marker, ezrin (Fig. 4B), and found that the number of ezrin-positive cortical plaques was significantly higher in the NE and ISO groups compared to the ICI group (Fig. 4C). Furthermore, the colocalization of GFP-tagged P. aeruginosa with ezrin staining suggested that P. aeruginosa uses the cortical plaques as the initial attachment point for invasion (Fig. 4B). These findings provide critical insights into the role of Adrb2 activation in inducing paracellular permeability by modulating the cytoskeleton and facilitating bacterial invasion during corneal infection. The increase in cytoskeleton organization in the ICI group may act as a counterforce against bacterial recruitment of actin. 

Figure 4.

 

Adrb2 modulates corneal resistance by regulating epithelial integrity. (A) The top 10 upregulated gene ontology terms of the ICI-treated corneal epithelium in RNA sequencing (n = 15) for each group. (B) Colocalization of the cortical plaque marker ezrin (red) on the corneal epithelial cell surface and the GFP-tagged P. aeruginosa. (C) Quantification of ezrin in human epithelial cells treated with different chemicals. Eight images were used for ImageJ quantification in each group. Kruskal-Wallis test P < 0.001. Pairwise multiple comparisons showed P = 0.018 between ICI and NE groups and P < 0.001 between ICI and ISO groups. (D) Cytoskeleton (red) and tight junctions (green) staining on the epithelial cell line treated with actin toxin L-B with or without ICI treatment (n = 4). (E) CFU measurement of the epithelial cell line treated with different chemicals. NE was added to the cell culture to mimic the in vivo environment (n = 3). Kruskal-Wallis test P = 0.002. Pairwise multiple comparisons showed P = 0.002 between ICI and L-B groups, and P = 0.008 between ICI and L-B + ICI groups. (F) Cytoskeleton (green) and tight junctions (red) staining on the epithelial cell line with different chemical treatment. Con, HCECs infected with PA. White arrows indicate the paracellular gaps. (G) Schematic illustration of the weakening of adherence between epithelial cells as the concentration of Adrb2 agonists increases. The arrows indicate the breaking annealing points.

The generation of cellular adhesive forces by ZO-1 and counteradhesive forces by actinomyosin molecular motors mediate paracellular permeability.²⁹ To further elucidate the role of the cytoskeleton in bacterial defense, we treated cultured HCECs with the actin polymerization toxin latrunculin B (L-B), which significantly disrupted the cellular F-actin network and caused the aggregation of ZO-1 inside the cells, leading to a loss of cell adhesion (Fig. 4D). However, the addition of ICI to L-B–treated cells partially rescued the disrupted cytoskeleton and tight junction (Fig. 4D). Cell invasion assay demonstrated that the protective effect of ICI could be antagonized by L-B (Fig. 4E), suggesting that ICI strengthens host-pathogen defense by restoring the stability of cytoskeleton and tight junctions. Disruption of these junctions can lead to pathogen invasion, and immunostaining of ZO-1 for tight junctions in HCECs showed that ISO treatment induced a universal onset of intracellular gaps, whereas ICI treatment tightened the linkage (Figs. 4F, 4G). These findings suggest that strengthening the cytoskeleton and tight junctions through the inhibition of Adrb2 signaling can enhance host-pathogen defense and prevent bacterial infiltration. 

Inactivation of MAPK by Dusps Leads to the Protective Effects Against PA Infection Generated by Adrb2 Inhibition

Based on the RNA sequencing analysis, we observed a significant decrease in mitogen-activated protein kinase (MAPK) activity after ICI treatment (P = 0.00, Enrichment score = 0.63) (Fig. 5A). Dual-specificity phosphatases (Dusps), which are a group of phosphatases that can dephosphorylate substrates at tyrosine or serine/threonine residues,³⁰ were found to be among the most differentially expressed genes after ICI treatment, with Dusp5 being highly upregulated (Fig. 5B). Immunoblotting analysis confirmed a significant increase in Dusp5 protein levels (Figs. 5C, 5D) and a significant decrease in phosphorylated extracellular signal-regulated kinases (ERK) (Figs. 5C, 5E) in the ICI-treated HCECs. Because MAPK phosphorylation of cytoskeleton-associated proteins can regulate cytoskeleton rearrangement,³¹ we hypothesized that inhibiting Dusp5 would reduce the protective effect of ICI against infection. Our results demonstrated that the addition of suramin, a DUSP5 antagonist,³² significantly worsened CLW-induced PAK (Figs. 5F, 5G). Therefore our findings suggest that ICI treatment inhibits MAPK activity through the upregulation of DUSPs, especially Dusp5, and that this effect is responsible for the protective effect of ICI against PAK. Consequently, targeting Dusp5 may be a potential therapeutic approach for the treatment of corneal diseases. 

Figure 5.

 

Dusp-mediated inactivation of MAPK contributes to the protective effects of ICI. (A) Gene set enrichment analysis showing that the ICI group exhibits enrichment in the “inactivation of MAPK activity” compared to the Con group. (B) Heatmap of differentially expressed genes involved in the “inactivation of MAPK activity.” The Dusp family members are among the top upregulated genes in the ICI group. (C) Western blot analysis of Dusp5, pERK, ERK, and GAPDH at 40 minutes after drug treatment. Dusp5 expression is dramatically increased in the ICI group compared with the ISO group, whereas pERK is the lowest in the ICI group. (D) Dusp5 expression was significantly upregulated in the ICI group compared to ISO (P = 0.002) and Con (P = 0.026), as determined by Kruskal-Wallis test with all pairwise multiple comparisons. (E) ERK phosphorylation is significantly downregulated in the ICI group compared with ISO (P = 0.014) and Con (P = 0.002). Kruskal-Wallis test with all pairwise multiple comparisons. (F) Eye photos of mice wearing P. aeruginosa–inoculated CL for 30 hours and treated with subconjunctival injections of NE, NE + ICI, and NE + ICI + Suramin, respectively (n = 4). (G) Clinical scores of the mouse eye shown in (F). P = 0.029.

Discussion

Corneal infections caused by P. aeruginosa often result in poor visual outcomes, and there is a need to develop novel therapies. Prior investigations into CLW-induced keratitis have relied on the use of PAK mouse model constructed by mechanical injury on the cornea. In this study, we proposed that not only injury, but elevated NE levels also contribute to PAK. NE binds to Adrb2 and triggers the intracellular signaling, resulting in epithelial cytoskeleton reorganization. This in turn leads to a reduction in the integrity of the corneal epithelium, which facilitates adhesion and invasion by P. aeruginosa. 

Previous studies have demonstrated that NE can increase P. aeruginosa virulence, motility, proliferation, and epithelial invasion in a concentration-dependent manner.^7,33 In this study, we investigated the effect of NE on the host, particularly the disruption of the corneal epithelial barrier caused by exogenous NE. We observed an elevation of NE levels during extended CLW, where corneal NE levels were about 2.2-fold higher than those in control mice.⁶ To mimic this situation, we performed subconjunctival injections of NE. In a previous study, we found that subconjunctival injection of 2 µg NE could facilitate passive transepithelial penetration of riboflavin.¹⁶ In this study we applied 2.5 µg NE and observed P. aeruginosa invasion without additional injury. Although we did not test the lowest effective dosage of NE during corneal infection, the fact that P. aeruginosa is alive and is self-driving during invasion, suggesting that a lower dosage of NE, slightly above the physiological level, may have the potential to break down the epithelial barrier. 

The β2-AR is a G protein–coupled receptor that activates a complex signaling network upon ligand binding. Typically, NE binding to β2-AR preferentially activates the stimulatory G protein, leading to the activation of adenylyl cyclase and cyclic adenosine monophosphate or MAPK pathways.³⁴ In this study we focused on the most canonical pathway, the phosphorylation and activation of the MAPK ERK1/2 pathway. Our transcriptional analysis indicated that MAPK activity was inhibited after ICI treatment. Although ICI was traditionally regarded as a β2-AR antagonist, it can also act as a partial agonist or inverse agonist.³⁵ Continuous agonist exposure can lead to β2-AR desensitization, which is mediated by β-arrestins.³⁴ The recruitment of β-arrestins also drives the activation of MAPK.³⁶ The β-arrestins have the ability to regulate cytoskeleton organization in response to upstream signals.¹⁹ Several theories exist on how β-arrestins influence actin organization, including the upstream G protein–coupled receptor actions, the downstream molecules, and the extracellular environment, all of which may contribute to actin organization.³⁷ Further exploration is needed to determine whether β-arrestins are the major β2-AR downstream regulator during bacterial invasion. 

Dusps are a family of MAPK-specific phosphatases that are involved in the downregulation of ERK signaling.³⁸ Our results demonstrate that 12 members of the Dusp family were upregulated after ICI treatment, resulting in a corresponding downregulation of ERK phosphorylation. However, it is still unclear whether there are other intracellular mediators that link β2-AR/β-arrestins and Dusps. Our observation that inhibition of Dusp5 by suramin worsened corneal infection suggests Dusps play a role in the protective effect of ICI against PAK. Dusps have also been reported to regulate immune cells,³⁸ and further investigation is needed to determine whether there is a synergic action between immune cells and the cytoskeleton in regulating keratitis. 

Besides its effects on corneal cells, NE has been shown to have modulatory effects on immune cells. Studies have shown that the ablation of the sympathetic signaling axis limited the maturation of suppressive myeloid cells³⁹ and mitigated the immunosuppressive function of myeloid-derived suppressor cells.⁴⁰ In macrophages, β-AR signaling reduced the deformability of macrophages by regulating the actin cytoskeleton⁴¹ and stimulated M2 macrophages to produce anti-inflammatory properties.⁴² The recruitment of more neutrophils into the cornea and secretion inflammatory cytokines after corneal abrasion was also observed on sympathetic activation and elevated NE levels.^7,9 Moreover, there is direct regulation of sympathetic nerves to the surgical pain during cataract surgeries through upregulation of granulocyte colony-stimulating factor CSF3.⁴³ Modification of natural killer cells extrinsic or intrinsic β2-AR signaling has been shown to influence the cells’ reactions toward viral challenges.^44,45 The β2-AR agonist has been reported to reduce group 2 innate lymphoid cell responses by negatively regulating its proliferation and effector functions.⁴⁶ Additionally, the hyperactivity of sympathetic nerves has been found to limit the development of experimental autoimmune encephalomyelitis and the generation of pathogenic T cells through β2-AR.⁴⁷ It appears that NE and its ARs have a multifaceted role in regulating immune cell functions, which could impact both myeloid and lymphoid lineages. Therefore, the anti-infective effects of β2-AR antagonists may be attributed to several factors, including bacterial characteristics, epithelial barrier function, and immune cell activities. Given the rise of antibiotic-resistant strains, β2-AR antagonists represent a promising therapy for P. aeruginosa infections. 

There has been speculation that CLW may lead to microtrauma on the cornea.^48,49 However, scanning electron microscopy studies have indicated that there is little change to corneal surface ultrastructure after CLW.⁵⁰ Thus it is believed that CLW increases the risk of microbial keratitis via its impact on the ocular environment.⁴ In this study, we constructed a CLW mouse model with bacterial-inoculated contact lenses. Although no direct injury was manually made on the mouse cornea, we sutured the eyelid for 30 hours to keep the lens attached to the cornea. The suture of the eyelid may be a potential confounder, because this action can lead to a systematic increase in tissue NE as well. However, there was no other way to effectively keep the CL inside a mouse eye. After the mice woke up from anesthesia, they might scratch their eye with their paws to get rid of irritations on the cornea, which possibly made new corneal injuries. Nevertheless, we used inoculated CLW and eyelid suture in the control mice as well. The only difference between the control and treatment group was the solution subconjunctivally injected, with one being saline solution and the other being NE. Therefore the results we obtained robustly demonstrate that NE was the key to induce noninjury keratitis, rather than eyelid suture. 

In conclusion, the findings of this study suggest that NE is a potential intrinsic stimulator of PAK in the context of CLW, because it can disrupt the corneal epithelial barrier and facilitate P. aeruginosa invasion. Moreover, NE can modulate immune cell activity, making it a promising target for P. aeruginosa infection therapy, especially for antibiotic-resistant strains. Further research is warranted to explore the potential use of β-blockers in PAK treatment and to investigate the association between sympathetic nerve activation and ocular infection. 

Acknowledgments

Supported by the Shandong Provincial Natural Science Foundation (ZR2020QH140 to B.N.Z., ZR2022MH277 to X.M.), the National Natural Science Foundation of China (82101091 to B.N.Z., 82000852 to F.S.), and the Taishan Scholar Program (tstp20221163 to Q.Z.). 

Author Contributions: B.N.Z., W.K.C., and Q.Z. designed research; B.N.Z., B.Q., F.S., Q.D., Z.S. performed experiments; B.Z. analyzed sequencing data; B.N.Z. wrote the manuscript; S.L. and X.D. provided clinical data; V.J., W.K.C., Q.Z. provided suggestions to the manuscript. 

Data and Materials Availability: Mouse transcriptome data has been deposited in the Genome Sequence Archive (GSA, https://ngdc.cncb.ac.cn/gsa/) with the accession number CRA010787. 

Disclosure: B.N. Zhang, None; B. Qi, None; W.K. Chu, None; F. Song, None; S. Li, None; Q. Dong, None; Z. Shao, None; B. Zhang, None; X. Du, None; X. Ma, None; V. Jhanji, None; Q. Zhou, None