Pseudomonas aeruginosa MucD Protease Mediates Keratitis by Inhibiting Neutrophil Recruitment and Promoting Bacterial Survival

Abstract

Purpose.: Pseudomonas aeruginosa is a leading pathogen of blinding keratitis worldwide. In this study, the role of the serine protease in the pathogenesis of P. aeruginosa keratitis in the mouse cornea was investigated by comparing the parent and rescue strains.

Methods.: Cornea of C57BL/6 mice were infected with P. aeruginosa strain PAO1, serine protease (MucD protease or PA3535) mutants (ΔmucD or ΔPA3535), or a complemented strain. Corneal virulence was evaluated by determining clinical scores and bacterial enumeration. A myeloperoxidase assay was performed to determine the number of polymorphonuclear (PMN) cells infiltrating the cornea. An ELISA was used to quantify inflammatory cytokines and chemokines in the cornea.

Results.: The clinical score and bacterial numbers in eyes infected with ΔmucD were significantly lower than in those infected with PAO1, ΔPA3535, or the MucD rescue strain after 48 hours (P < 0.001). A larger number of infiltrating PMN cells was observed in eyes infected with ΔmucD at 12 and 24 hours, compared with eyes infected with PAO1. IL-1β, KC, and MIP2 levels were higher in eyes infected with ΔmucD than in those infected with PAO1 after 12 hours.

Conclusions.: The MucD protease suppressed IL-1β, KC, and MIP2 during the early stages of the infection and inhibited neutrophil recruitment in the cornea. Therefore, the MucD protease contributes significantly to the pathogenesis of keratitis. MucD protease plays a critical role in the establishment of Pseudomonas aeruginosa keratitis by facilitating evasion of the immune response.

Introduction

P seudomonas aeruginosa is a common opportunistic bacterial pathogen that causes a variety of human infections, and is a leading cause of blinding keratitis worldwide. Keratitis caused by P. aeruginosa occurs following injury, ocular surgery, or in association with contact lens wear, and can progress rapidly with suppurative infiltration; it can lead to corneal perforation and melt, and result in the loss of vision. ¹ To understand the mechanism of pathogenesis in keratitis caused by P. aeruginosa , investigators have attempted to identify the virulence factors and their associated cellular structures, such as the flagella, ^2,3 pili, ⁴ and lipopolysaccharide, ⁵ as well as extracellular products, including proteases, ^6–10 exotoxin A, ¹¹ and biofilm. ¹² Of these, proteases, including metalloproteases such as alkaline protease, elastase A, and elastase B, ^13–15 have been investigated extensively. However, infection of animal models with metalloprotease-deficient mutants demonstrated that none of these enzymes are essential for corneal virulence. ¹⁶ Protease IV has been shown to be an important virulence factor in rabbit cornea. ^17,18 Moreover, P. aeruginosa small protease (PASP), a more recently discovered protease, plays a critical role in the pathogenesis of keratitis. ^7,19,20 Along with these virulence factors, the Type III secretion system (TTSS) is involved in the pathogenesis of keratitis. ^21–25 The TTSS probes the host cell and transports toxins using a needle-like prong apparatus. 

The corneal response to P. aeruginosa infection is critical for a better understanding of the natural defense mechanisms of the cornea, which will in turn facilitate the development of novel new treatments and preventive measures. ²⁶ IL-1 and chemotactic cytokines (e.g., IL-8) play a critical role in neutrophil recruitment and the innate immune response. ^26,27 Sun et al. ²⁵ showed that ExoS and ExoT ADP ribosyltransferase (ADPR) activities mediated P. aeruginosa keratitis in mice by promoting neutrophil apoptosis and bacterial survival. Moreover, ExoS ADPR activities inhibited IL-1β and IL-18 secretion by repressing the activation of caspase-1 in the host cell. ²⁸ Thus, P. aeruginosa virulence factors, such as TTSS, may influence the cytokine profile and reduce neutrophil recruitment or activity, promoting bacterial survival in the cornea. Recently, Okuda et al. ²⁹ demonstrated IL-8 degradation following infection of Caco-2 cells with the wild-type, but not the ΔExoS, strain; purified ExoS protein did not degrade IL-8. ²⁹ However, ExoS may degrade IL-8 indirectly; IL-8 degradation by P. aeruginosa was blocked by the addition of serine protease inhibitors. Thus, serine proteases may influence the cytokine profile and the immune response in the cornea. Protease IV, a serine protease, has been shown to be a critical virulence factor in rabbit and mouse corneas, ^17,18 and can degrade various proteins, including complement, fibrinogen, plasminogen, immunoglobulin, and surfactant proteins. ^30,31 However, little is known of the role of P. aeruginosa serine proteases, other than protease IV, in corneal infection. In this study, the role of serine proteases in P. aeruginosa keratitis in mice was investigated. 

Materials and Methods

Animal Care and Use

Female C57BL/6 mice were obtained from CLEA, Japan, Inc. (Tokyo, Japan). Mice at 7 to 9 weeks of age were used in all experiments. All animals were humanely treated according to the guidelines of the ARVO Resolution on the Use of Animals in Research. 

Bacterial Strains and Culture Conditions

The wild-type PAO1 strain and mutant strains were used (Table). Searching of the Pseudomonas Genome Database (http://www.pseudomonas.com) revealed two serine protease genes, PA3535 and mucD (PA0766). ³² Thus, ΔmucD and ΔPA3535 strains, constructed using the suicide vector pEX18Tc as described previously, were used, ²⁹ as well as the mucD (ΔmucD/mucD) complemented strain, transformed by a pBBR1MCS5-tac plasmid harboring mucD. Bacteria were grown to mid-log phase in brain–heart infusion broth, washed, and diluted in sterile PBS to 1 × 10⁵ cells per 2.5 μL. 

Table

 

Bacterial Strains Used in This Study

Strain	Phenotype	Genotype and/or
PAO1	Prototroph	Wild type
ΔmucD	MucD protease-deficient	PAO1 ΔmucD
ΔmucD/mucD	MucD protease-complement strain	PAO1 ΔmucD/mucD, Gmr
ΔPA3535	PA3535 deficient	PAO1 ΔPA3535

 Gm^r, gentamicin resistant.

In Vivo Model of Corneal Infection

The mouse keratitis model used in this study has been described previously. ²⁷ Mice were anesthetized by intraperitoneal (IP) injection of 0.4 mL 1.2% 2,2,2-tribromoethanol (Sigma-Aldrich Japan, Tokyo, Japan) in saline. Central corneas were scarified with three parallel 1-mm-long abrasions using a 27-gauge needle. A 2.5-μL aliquot containing 1 × 10⁵ colony-forming units (CFUs) of P. aeruginosa was applied to the scarified cornea. Sterile PBS was applied to the abraded cornea as a trauma control. A sterile trepan (Biopsy Punch; Kai Medical, Seki, Japan) was used to generate a 2-mm-diameter punch of a silicone hydrogel contact lens (1-DAY ACUVUE True Eye; Johnson & Johnson, New Brunswick, NJ), which was placed over the central cornea to maintain placement of the bacterial suspension. 

Scoring of Corneal Opacity

Corneal disease was graded using an established scale: 0: clear or slight opacity, partially or fully covering the pupil; 1: slight opacity, fully covering the anterior segment; 2: dense opacity, partially or fully covering the pupil; 3: dense opacity, covering the entire anterior segment; and 4: corneal perforation or phthisis. ³³ A clinical score was recorded for each mouse after infection for statistical comparison of disease severity. 

Quantification of Bacterial Growth in the Cornea

Whole eyes or corneas were homogenized twice under sterile conditions using the Micro Smash MS-100 system (Tomy Seiko, Tokyo, Japan) at 825g for 1 minute. Serial log dilutions were performed and bacteria were plated onto brain–heart infusion agar (BD Biosciences, Franklin Lakes, NJ). Plates were incubated at 37°C for 18 hours, and the number of CFUs was determined by direct counting. 

Histology and Immunohistochemistry

Eyes were enucleated at predetermined time points and fixed in 4% paraformaldehyde or methanol overnight at 4°C. They were then embedded in paraffin, and 5-μm sections were cut through the central cornea and stained with hematoxylin and eosin. Rabbit anti-mouse polymorphonuclear (PMN) antibody (1:1000 dilution; Cedarlane Laboratories, Ontario, Canada) was used. Fluorescein isothiocyanate–labeled anti-rabbit IgG (1:500 dilution; Vector Laboratories, Burlingame, CA) was used as the second antibody. The cornea sections were examined and photographed with a charge-coupled device (CCD) camera (model DP-50; Olympus, Tokyo, Japan) attached to a model BX-50 microscope (Olympus) or an inverted fluorescence microscope (Observer Z1; Carl Zeiss Micro Imaging, Thornwood, NY). 

Myeloperoxidase Assay

A myeloperoxidase (MPO) assay was modified and used to enumerate active PMN cells that infiltrated the corneal stroma after infection. ³⁴ A 150-μL aliquot, containing homogenized corneas in PBS, was added to 150 μL 100 mM phosphate buffer (pH: 6.0) containing 0.5% hexadecyltrimethylammonium bromide. Samples were freeze-thawed three times and centrifuged at 10,000g for 15 minutes at 4°C. Then, a 20-μL aliquot of the supernatant was added to 80 μL of a 50-mM phosphate buffer containing ortho-dianisidine dihydrochloride (16.7 mg per 100 mL) and 0.0005% hydrogen peroxide. The change in absorbance at 450 nm was measured continuously for 5 minutes, and the rate of change for each sample was determined. MPO/cornea units were calculated from a standard curve generated with purified MPO (product number M6908; Sigma-Aldrich Japan). One unit of MPO activity was equivalent to approximately 2 × 10⁵ PMN cells mL⁻¹. 

ELISA

Cytokine protein levels were determined using ELISA kits (R&D Systems, Minneapolis, MN). Corneas from mice were collected individually (n = 5–6 per group per time point). The corneas were homogenized in 0.5 mL PBS. All samples were centrifuged at 15,500g for 5 minutes, and an aliquot of each supernatant was assayed in duplicate for IL-1β, TNF-α, keratinocyte-derived cytokine (KC), and macrophage inflammatory protein-2 (MIP-2) protein, according to the manufacturer's instructions. 

Neutrophil Depletion Mice

Mice were rendered neutropenic by IP injection of 100 μg anti-mouse Gr-1 (RB6-8C5 MAb; R&D Systems) in 0.2-mL PBS at 24 hours, and were then inoculated with P. aeruginosa . ³⁵ The control mice received the same dose of rat IgG. 

Statistical Analysis

An unpaired, two-tailed Student's t-test was used to determine statistical significance. A value of P < 0.05 was taken to indicate statistical significance. 

Results

MucD Protease Is Essential for the Development of P. aeruginosa Keratitis

To determine the role of serine proteases in the development of P. aeruginosa keratitis, mice were infected with the PAO1 parent strain of P. aeruginosa , or each of two serine-protease–deficient mutants. Infection with PAO1 and the ΔPA3535 mutant increased the clinical score for corneal opacity, with severe disease apparent at 48 hours (Figs. 1A, 1B); in contrast, infection with the ΔmucD mutant significantly reduced the clinical scores (Figs. 1A, 1B) Complementation with a plasmid expressing MucD (ΔmucD/mucD) completely restored the capacity of the ΔmucD mutant to cause infection (Figs. 1A, 1B). To determine the role of serine protease in bacterial survival in the cornea, eyes were homogenized and CFUs quantified. Numbers of bacteria in corneas infected by PAO1 and ΔPA3535 were significantly higher than those in corneas infected with ΔmucD (Fig. 1C). The number of CFUs in the cornea infected by the ΔmucD mutant was lower than the inoculum, indicating that these bacteria were being killed. Infection with the complemented ΔmucD/mucD strain restored the wild-type phenotype in terms of survival in the cornea (Fig. 1C). 

Figure 1

 

Role of the MucD serine protease in P. aeruginosa keratitis. Corneas of C57BL/6 mice were abraded and infected with PAO1, ΔmucD, ΔPA3535, or ΔmucD/mucD. (A) Representative images of the cornea at 48 hours postinfection. (B) Clinical scores of infected eyes at 48 hours postinfection. The scores of the ΔmucD group were significantly lower than those of the other strains. Data are means ± SEM (n = 5). ***P < 0.001. (C) Number of viable organisms recovered from mouse corneas at 48 hours postinfection. The number of ΔmucD recovered was significantly lower than that of the other strains. Data are means ± SEM (n = 5). ***P < 0.001.

To evaluate the course of infection, clinical scores and bacterial numbers were compared in eyes infected with PAO1 and ΔmucD at 0, 6, 12, 24, and 48 hours. The corneal opacity in the eye infected with PAO1 was elevated over the course of observation (Figs. 2A, 2B). In contrast, the corneal opacity scores increased slightly for ΔmucD; however, the clinical scores in eyes infected with ΔmucD were significantly lower than those in eyes infected with PAO1 at 12, 24, and 48 hours (Figs. 2A, 2B). The number of PAO1 recovered from infected eyes increased over the course of the infection, whereas the number of ΔmucD decreased from 24 to 48 hours (Fig. 2C). 

Figure 2

 

Time course of corneas infected with PAO1 or ΔmucD. (A) Representative images of the cornea at 6, 12, 24, and 48 hours postinfection. (B) Clinical scores of the infected eyes. The scores of the ΔmucD group were significantly lower than those of the PAO1 group at 12, 24, and 48 hours. Data are means ± SEM (n = 8). *P < 0.05, **P < 0.01, ***P < 0.001. (C) Numbers of viable organisms recovered from mouse corneas. The number of recovered ΔmucD was significantly lower than PAO1 at 24 and 48 hours postinfection. Data are means ± SEM (n = 8). *P < 0.05, ***P < 0.001.

Effect of the MucD Protease on Neutrophil Recruitment

Because the growth of ΔmucD was inhibited 12 hours after bacterial inoculation, we expected the corneal immune response, such as neutrophil infiltration, to be effective in clearing ΔmucD. Histological sections showed a prominent cellular infiltrate and destruction of the collagen layer in PAO1-infected corneas after 24 hours, and corneal edema along with destruction of the collagen layer after 48 hours (Fig. 3A). However, corneas infected with the ΔmucD showed cellular infiltrate in all layers from 24 hours, but not destruction of the collagen layer (Fig. 3A). To assess neutrophil recruitment in the corneal stroma, sections were stained with anti-PMN antibodies. Infiltrating cells, stained with anti-PMN in the PAO1-infected cornea, were evident in all layers of the cornea at 24 hours (Fig. 3B). In contrast, stained cells in ΔmucD-infected cornea were evident in all layers at 12 hours (Fig. 3B). PMN cells were enumerated in the cornea of eyes infected with PAO1 and ΔmucD. The number of PMN cells in ΔmucD-infected corneas was significantly higher than in PAO1-infected corneas at 12 and 24 hours (Fig. 3C). 

Figure 3

 

Polymorphonuclear recruitment in infected corneas. (A) Representative histological appearance of corneas infected with PAO1 or ΔmucD at 12, 24, and 48 hours postinfection. Scale bar: 100 μm. (B) Anti-PMN antibody immunohistochemistry staining of corneas infected with PAO1 or ΔmucD at 12 and 24 hours postinfection. Scale bar: 100 μm. (C) Myeloperoxidase activity in infected corneas. Significantly higher MPO activity was evident in corneas infected with ΔmucD than in those with PAO1 at 12 and 24 hours postinfection. Data are means ± SEM (n = 5–6). Time 0 represents uninjured corneas. *P < 0.05.

The MucD Protease Inhibits Cytokines and Chemokines in the Early Stages of Infection

To determine the mechanism by which the MucD protease inhibits neutrophil recruitment, we quantified proinflammatory cytokines (IL-1β and TNF-α), and chemokines (KC and MIP-2), in eyes infected with PAO1 and ΔmucD. IL-1β, KC, and MIP-2 levels in PAO1-infected corneas at 12 hours were lower than in those infected with ΔmucD (Fig. 4). 

Figure 4

 

Levels of the chemokines MIP-2 and KC and the proinflammatory cytokines IL-1β and TNF-α, as determined by ELISA in corneas infected with PAO1 or ΔmucD at 0, 12, and 24 hours postinfection. Time 0 represents unwounded corneas. Data are means ± SEM (n = 5–6). *P < 0.05, ***P < 0.001.

The ΔmucD Strain Causes Infection Similar to PAO1 in Neutrophil-Depleted Mice

To confirm the association between MucD protease and neutrophil recruitment, the eyes of neutrophil-depleted mice were infected with either PAO1 or ΔmucD. As a control, ΔmucD was used to infect the eyes of mice receiving rat IgG. The eyes that had abscess inside the eye were determined as endophthalmitis. Both PAO1 and ΔmucD induced corneal perforation and endophthalmitis in neutrophil-depleted mice. In contrast, ΔmucD did not induce corneal perforation or endophthalmitis in control mice. The number of bacteria recovered from ΔmucD- and PAO1-infected whole eyes of neutrophil-depleted mice was significantly greater than that from ΔmucD-infected whole eyes of control mice (Fig. 5). 

Figure 5

 

Recovery of viable organisms from corneas of mice administered Gr-1 or control at 48 hours postinfection. The number of recovered ΔmucD was similar to PAO-1 in mice administered with Gr-1. Data are means ± SEM (n = 5). N.S., not significant; ***P < 0.001.

Discussion

Previous studies indicated that proteases contributed directly to corneal tissue damage, as well as indirectly via activation of the matrix metalloproteinase. ^36,37 Moreover, protease IV and PASP may degrade complement or antimicrobial peptides and thus escape from the host defenses. ^19,30,31 PMNs play a major role in bacterial eradication. Bacteria are killed by PMN phagocytosis, lysosomal degranulation, and oxidative stress. PMN recruitment in the infected cornea is induced by cytokines and chemokines. ²⁶ In the mouse, two members of the C-X-C family of chemokines, MIP-2 (a functional homologue of human IL-8) and KC, are potent chemoattractants and activators of PMN. ³⁸ Along with chemokines, IL-1 regulates MIP-2 levels and is involved in PMN recruitment. ²⁶ Thus, it is important to understand the dynamics of PMNs, cytokines, and chemokines in the P. aeruginosa–infected cornea. However, the relationship between P. aeruginosa proteases and the corneal immune response, including PMNs and chemokines, has not been investigated extensively to date. The P. aeruginosa MucD serine protease was involved in degradation of IL-8 in CaCo-2 cells infected with PAO1 ²⁹ ; thus, we expected that similar phenomena would be evident in corneal infection. To identify the proteases that play a critical role in keratitis, we focused on serine proteases other than protease IV. 

In this study, the MucD protease-deficient mutant demonstrated significantly reduced virulence in the mouse keratitis model. Complementation with a plasmid-expressing MucD restored the capacity of the ΔmucD mutant to cause infection. This implies that MucD protease plays an important role in the pathogenesis of keratitis. The molecular weight of MucD is predicted to be 50.3 kDa. ³² The mucD gene is located in the algT–mucA–mucB–mucC–mucD operon; MucD is a negative regulator of alginate production and a positive regulator of heat-shock stress. ³⁹ The mucD gene itself inhibits the production of extracellular polysaccharide. ^40,41 Moreover, MucD has been reported to be an endoserine protease with a GNSGGAL motif, which localizes to the periplasmic space. ^39,41 Thus, MucD is a multifunctional protein. Unlike other serine proteases that were investigated in keratitis, MucD protease could have roles for regulator of virulence factors as well as protease. Yorgey et al. ⁴² reported that the mucD-deficient mutant exhibited reduced virulence in several infection models, such as in plants, nematodes, and mice. There are several possible explanations for the effect of mucD in this infection model. First, the mucD-deficient mutant exhibits increased sensitivity to temperature and oxidative stress. ³⁹ Thus, mucD mutants may be more sensitive to PMN oxidative attack in tissue. Second, MucD may be required for the production of extracellular virulence factors. ⁴² Third, MucD could degrade secreted mucin that protected tissues in previous report. ⁴³ However, the exact role of MucD in infection models has not been determined. Our data suggest that MucD reduced IL-1β, KC, and MIP-2 levels in the early stages of infection and inhibited PMN recruitment in the cornea. Furthermore, the PMN-depletion model facilitated establishment of infection by the ΔmucD strain. Thus, in this case, MucD inhibits PMN recruitment via the suppression of cytokines and chemokines. MucD may leak from the periplasm to the extracellular space and affect cytokines and chemokines. However, the exact mechanism underlying the suppression of cytokines and chemokines by MucD is not known. One possibility is that MucD directly degrades cytokines and chemokines. Another is that MucD induces production of extracellular virulence factors, and so indirectly degrades cytokines and chemokines. The regulation of MucD secretion has not been investigated; however, several regulators, including the type III effector ExoS, are candidates. ²⁹ Further investigation is needed to understand the roles and mechanisms underlying the role of MucD in keratitis. 

In conclusion, our results showed that infection with the MucD-deficient mutant resulted in increased cytokine and chemokine levels, as well as neutrophil recruitment during the early stages of infection. This observation is consistent with previous in vitro experiments using Caco-2 cells. ²⁹ Thus, MucD plays a critical role in the establishment of infection by facilitating evasion of the immune response. Collectively, these observations suggest the MucD protease as a useful target for therapies that aim to treat or prevent keratitis. 

Acknowledgments

Supported by the Department of Bioscience, Integrated Center for Science, Ehime University, and in part by research grants from the Japanese Ministry of Education, Culture, Sports, Science and Technology, Tokyo, Japan. 

Disclosure: Y. Mochizuki, ROHTO Pharmaceutical Co., Ltd. (E); T. Suzuki, None; N. Oka, None; Y. Zhang, None; Y. Hayashi, None; N. Hayashi, None; N. Gotoh, None; Y. Ohashi, None