Contribution of Leukocidin ED to the Pathogenesis of Staphylococcus aureus Endophthalmitis

Abstract

Purpose: To test the hypothesis that leukocidin ED (LukED) contributes to the pathogenesis of experimental Staphylococcus aureus endophthalmitis

Methods: Growth curves were generated for S. aureus strain JE2 and strain JE2 lukE::Tn, the transposon mutant of LukED, in brain heart infusion (BHI) and explanted rabbit vitreous. The expression of leukotoxins (lukSF-PV, lukED, hlgABC, and lukGH) was assessed in 18-hour overnight cultures in BHI, tryptic soy broth, and vitreous. S. aureus endophthalmitis was induced by intravitreal injection of 5000 colony-forming units of JE2 or JE2 lukE::Tn into C57BL/6J mice. At 6, 12, and 24 hours after infection, eyes were assessed for retinal function, intraocular colony-forming units and inflammation, and neutrophil infiltration by flow cytometry. RNA was isolated from infected eyes to assess leukotoxin expression.

Results: Strains JE2 and JE2 lukE::Tn grew similarly in BHI and vitreous. Transcript levels of leukotoxin subunits were lower in vitreous compared with laboratory media. In vivo, no differences in retinal function, intraocular growth, intraocular inflation, or neutrophil infiltration were observed in eyes infected with JE2 or JE2 lukE::Tn. During infection, other leukotoxins were expressed in vivo in the absence of LukED.

Conclusions: LukED does not seem to be essential for the pathogenesis of experimental S. aureus endophthalmitis. However, other leukotoxins are expressed in vivo, which may compensate for the effects of LukED during infection.

Bacterial endophthalmitis is a serious medical emergency that necessitates immediate treatment because the infection results in damage to sensitive intraocular tissue, which may lead to vision loss.^1–4 Bacteria can gain access to the posterior segment of the eye during ocular surgeries, from penetrating injuries to the eye or, in rare cases, from a systemic bloodstream infection.^1–7 The majority of endophthalmitis cases arise from the introduction of bacteria after a procedure, including during cataract surgery or intravitreal injections to treat neovascular eye diseases and intraocular inflammation.^8,9 Symptoms of endophthalmitis include acute inflammation, eye pain, redness, and rapidly decreasing vision.^1–8 Treatment for endophthalmitis includes intravitreal administration of antibiotics, such as vancomycin, ceftazidime, or fourth-generation fluoroquinolones, as well as corticosteroids to blunt inflammation.^3,8 In severe cases, vitrectomies are performed to clear the interior of the eye of bacteria, inflammatory cells, and debris that might obscure the visual axis.^1–8 

Endophthalmitis can arise from infection with bacteria or fungi, with coagulase-negative staphylococci (CoNS) being one of the most commonly cultured bacteria from postoperative and post-traumatic endophthalmitis.^3,10 Staphylococci account for the majority of endophthalmitis cases because staphylococci are part of the ocular microbiota.^3,11 Infections with CoNS typically resolve well, with minimal impact on vision. However, infection with Staphylococcus aureus often leads to poor clinical outcomes, with visual acuities of these cases reported at 20/400 or worse.^12–14 S. aureus endophthalmitis severity arises from the combination of its many virulence factors and the potential for multidrug resistance (MDR).¹⁵ Recent epidemiology studies report high incidences of methicillin-resistance in staphylococcal ocular isolates. ARMOR (Antibiotic Resistance Monitoring in Ocular Microorganisms) is an open-ended program that provides antibiotic susceptibility and resistance profiles of ocular pathogens. For staphylococci isolated between 2009 and 2018, 34.9% were methicillin-resistant S. aureus (MRSA), and these isolates were more likely to be resistant to another antibiotic.¹⁶ For antibiotic profiles in presumed endophthalmitis ocular isolates between 2009 and 2020, 47.9% of CoNS and 45.9% of S. aureus isolates were resistant to methicillin.¹⁷ There was a prevalence of MDR among isolates resistant to methicillin where 72.2% of methicillin-resistant CoNS and 85.7% of MRSA were resistant to at least three antibiotics.¹⁷ In addition, MDR complicates treatment for intraocular infections. S. aureus clinical isolates are increasingly resistant to multiple antibiotics, including methicillin, cephalosporins, aminoglycosides, and β-lactam antibiotics.^13,14,20,21 The rise of MDR among S. aureus isolates underscores the need to understand host–pathogen interactions during S. aureus endophthalmitis. By understanding these host–pathogen interactions, novel therapeutic strategies could be developed for this blinding infection beyond the use of conventional antibiotics. 

In experimental S. aureus endophthalmitis, neutrophils infiltrate the eye. Despite this infiltration, there is an inadequate clearance of intraocular bacteria, which can replicate into the millions.^22,23 Previous findings have shown that S. aureus toxins and transcriptional regulators contribute to intraocular inflammation and the loss of retinal function.^18,19 Transcriptional regulators control the expression of a variety of factors that help the bacterium to evade an immune response.²⁴ Among these factors, S. aureus produces a class of toxins called leukotoxins that can actively lyse innate and adaptive immune cells.^25–28 Leukotoxins are bicomponent pore-forming toxins, composed of slow (S) and fast (F) subunits. Subunit designation is based on their elution speed on a chromatography column.²⁹ Human S. aureus isolate genomes can encode up to five leukotoxins, including Panton–Valentine leukocidin (PVL), gamma-toxin AB (HlgAB), gamma-toxin CB, leukocidin ED (LukED), and LukGH (also referred to as LukAB).^25–27 Leukotoxins cause cell lysis in a receptor-mediated manner, and each leukotoxin binds to specific receptors. First, the S subunit binds to a specific chemokine, complement, or integrin receptor (Cd11b). Second, upon S subunit binding, the F subunit interacts with the S subunit to form a dimer. Third, additional S and F subunits are recruited until a prepore octamer structure is formed, and the octamer undergoes conformational changes in the stem domain of the S and F subunit. Finally, these conformational changes lead to insertion into the cell membrane, forming a β-barrel pore that facilitates the influx of cations, resulting in cell lysis.^25–28,30 

The role of individual leukotoxins in the pathogenesis of S. aureus endophthalmitis has not been studied extensively. Siqueira et al.³¹ reported that intravitreal injections of the purified PVL and gamma toxin subunits resulted in inflammation of the posterior and anterior chamber of rabbit eyes. However, in this study, a combination of different toxin subunits was tested, and it is not clear if all the combinations resulted in pore formation. Furthermore, purified toxins were injected, and the physiological relevance of the concentrations tested to ocular infection is unclear. Additional research investigated the role of gamma toxin using gamma toxin-deficient mutants in rabbit models of endophthalmitis. Supersac et al.³² reported that the absence of the gamma toxin resulted in no difference in inflammation, whereas Callegan et al.⁹ reported that the absence of the gamma toxin resulted in no difference in retinal function. More recently, intravitreal injections of PVL into the rabbit eye resulted in PVL localization in the retina. Liu et al.³³ reported that PVL binds to retinal ganglion cells via the C5aR1 receptor and transiently localized to displaced amacrine cells. PVL also led to the activation of Muller cells and microglia, leading to the production of interleukin-6, and some microglia cells underwent apoptosis.³³ These results were later replicated in an rabbit retinal explant model, where PVL localized to horizontal cells.³⁴ In the same retinal explant model, incubation with PVL for 4 hours induced oxidative stress and altered the concentration of retinal neurotransmitters.³⁵ These studies did not investigate whether PVL or the gamma toxin affected infiltrating neutrophils. Additionally, the contributions of LukGH and LukED in S. aureus endophthalmitis have not been investigated. 

Ocular clinical isolates have been screened for virulence genes in several epidemiological studies. Afzal et al.³⁶ published two separate reports on the virulence genes associated with infectious and noninfectious S. aureus ocular strains. One study reported that hlgABC was present in more than 96% of all S. aureus ocular isolates, and the second reported that hlgABC was associated with 100% of infection isolates.^36,37 In a separate report, Afzal et al.^36,37 reported incidences for pvl of 16% and 11% in infection isolates and 16% and 14% in noninfection strains. lukE was reported to be in 4% of all ocular isolates, but the frequency of the lukD subunit, the pairing subunit, was 79% in infectious strains and 50% in noninfectious strains. Sueke et al.³⁸ reported that 9.5% of S. aureus keratitis isolates were PCR positive for pvl, and these isolates were associated with worse clinical outcomes. An additional study from two hospitals in India reported 67% PCR positivity for pvl in S. aureus ocular isolates.³⁹ Rarey et al.,⁴⁰ using latex particles that were sensitized to antibodies against PVL, reported that 10% of S. aureus endophthalmitis isolates were positive for PVL. We reported that, among S. aureus ocular isolates, 76.7% were lukED positive, 19.4% were hlgABC positive, and 23.3% were pvl positive.¹⁰ Our screening showed that LukED was the most prevalent leukotoxin detected in a collection of 129 S. aureus isolates.¹⁰ 

LukED is capable of lysing neutrophils, macrophages, monocytes, dendritic cells, T cells, and natural killer cells by targeting CXCR1, CXCR2, or CCR5.^41,42 Additionally, LukED is the only leukotoxin reported to lyse erythrocytes by targeting the Duffy antigen receptor for chemokines.⁴³ Furthermore, LukED has been implicated to facilitate S. aureus replication in a murine bacteremia model by killing phagocytes, including neutrophils.²⁷ The high incidence of LukED in S. aureus ocular isolates and its potential activity against immune cells that might lead to unfettered replication led us to test the hypothesis that LukED contributes to the pathogenesis of S. aureus endophthalmitis. Here, we demonstrated that the absence of LukED did not alter intraocular growth of S. aureus, inflammation, or retinal function changes occurring during infection. Although we did not detect a difference in the outcome of S. aureus endophthalmitis in the absence of LukED, we detected the expression of other leukotoxins during infection in this model. The presence of other leukotoxins may have masked or compensated for the absence of LukED, also contributing to deficiency in S. aureus clearance. 

Materials and Methods

Bacterial Strains

The strains used here include S. aureus JE2, a USA300 strain, and the lukE::Tn mutant of JE2 (JE2 lukE::Tn), from the Nebraska Transposon Mutant Library (BEI Resources, Manassas, VA, USA). JE2 is a derivative of USA300 LAC, where two plasmids, one encoding for macrolide resistance and another cryptic, were cured for ease of genetic manipulation and avoid interference with transposition events. USA300 strains have the same sequence type (ST8), agr group (I), SCCmec (subtype IV), and spa motif (MBQBLO). USA300 strains also commonly encode PVL toxin genes. The JE2 lukE::Tn mutant was generated with a mariner transposon, bursa aurealis. The mariner transposon is inserted after nucleotide 311 from the 5′ end of the lukE gene.⁴⁴ 

Quantifying S. aureus In Vitro

All strains were cultured in brain heart infusion (BHI; VWR, Radnor, PA, USA) media with aeration at 37°C. After 18 hours, cultures were centrifuged for 5 minutes at 4150×g and washed with PBS (pH 7.4). Cultures were then diluted in either fresh BHI or filter-sterilized explanted rabbit vitreous (Pelfreez Biologics, Rogers, AR, USA) to approximately 10³ colony-forming units (CFU)/mL, then incubated at 37°C for 18 hours. Every 2 hours, 20-µL aliquots were diluted 10-fold, track diluted in PBS, then pipetted onto BHI agar plates and counted to generate a growth curve. Growth rates were calculated during exponential phase using the equation, N_t = N₀ × (1 + r)^t, where N_t is the bacterial concentration at the end of exponential phase, N₀ is the bacterial concentration at the start of exponential phase, r is the growth rate (hour⁻¹), and t is the time passed.⁴⁵ 

Mice

In vivo experiments used male 8- to 10-week-old C57BL/6 mice (stock number 000664; Jackson Laboratory, Bar Harbor, ME, USA), following the guidelines and recommendations of the University of Oklahoma Health Sciences Center Institutional Animal Care and Use Committee, the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and the Guide for the Care and Use of Laboratory Animals. All mice were housed in biosafety level 2 conditions on a 12-hour dark/light cycle for 2 weeks to equilibrate their microbiota. 

Murine S. aureus Endophthalmitis Model

S. aureus strain JE2 or the isogenic transposon mutant JE2 lukE::Tn were cultured in BHI for 18 hours at 37°C, then diluted in BHI to the appropriate inoculum. Mice were anesthetized with isoflurane before the right eye was intravitreally injected with approximately 5000 CFU/0.5 µL of either strain JE2 or strain JE2 lukE::Tn using a sterile capillary needle (Kimble Glass Company, Vineland, NJ, USA). The contralateral eye served as an uninfected control.^46–48 

ERG

A- and B-wave function of mouse retinas were quantified by ERG, as described previously.^45,47–49 Pre-ERG dark-adaptation periods lasted for at least 6 hours. ERGs were performed at 6, 12, or 24 hours after infection (Espion E2, Diagnosys LLC, Lowell, MA, USA). Mice were anesthetized with a cocktail of ketamine and xylazine (ketamine HCl, 85 mg/kg body, Covetrus, Portland, ME, USA; and AnaSed, 14 mg/kg, Akorn, Decatur, IL, USA). We administered 0.5% proparacaine HCl (Alcon Laboratories, Fort Worth, TX, USA) for topical anesthesia, and 10% topical phenylephrine hydrochloride (Paragon BioTeck, Portland, OR, USA) was administered for dilation. Gold-wire electrodes were placed on each cornea, and reference and ground electrodes were affixed to the forehead and tail, respectively. Eyes were stimulated with five consecutive flashes (1200 cd·s/m²), and the A- and B-waves were recorded for both eyes simultaneously. The percent retention of the A- and B-waves was calculated using the following formula: 100 − {(1 − [Experimental A-wave or Experimental B-wave amplitude/Control A-wave or Control B-wave amplitude])} × 100, respectively.^45,47–49 

Quantifying Intraocular Bacterial Growth

At 6, 12, or 24 hours after infection, mice were euthanized by CO₂ inhalation and eyes were harvested. Harvested eyes were homogenized in 400 µL of PBS containing 1-mm glass beads (BioSpec Products, Inc., Bartlesville, OK, USA). Eyes were homogenized twice for 60 seconds at 5000 RPM in a mini-bead beater (Biospec Products, Inc.). Eye homogenates were track-diluted 10-fold and pipetted onto BHI agar plates, then incubated overnight. After overnight incubation, CFU/eye was quantified as described previously.^45,47–49 

Myeloperoxidase (MPO) Assay

MPO was quantified using a sandwich ELISA (Hycult Biotech, Plymouth Meeting, PA, USA), as previously described.^45,47–49 At 6, 12, or 24 hours after infection, mice were euthanized, harvested eyes were homogenized in 400 µL of PBS supplemented with a proteinase inhibitor cocktail (Roche Diagnostics, Indianapolis, IN, USA), then homogenized as described above. Homogenates of uninfected eyes were used as controls. The lower limit of detection of this assay was 2 ng/mL. 

Flow Cytometry

At 6, 12, or 24 hours after infection, mice were anesthetized as described elsewhere in this article, then perfused with PBS. After euthanasia, eyes were harvested and temporarily stored in 2 mL of Roswell Park Memorial Media 1640 without L-glutamine (Gibco, Grand Island, NY, USA) and supplemented with 10% fetal bovine serum (Gibco), until further processing. Whole eye globes were dissociated gently by pressing through a cell strainer (70 µm; Thermo Fisher Scientific, Waltham, MA, USA) using the wide end of a sterile 1 mL syringe plunger. Eyes were then incubated in 1 mL of 0.2 mg/mL collagenase type I (Worthington, Lakewood, NJ, USA) solution dissolved in Roswell Park Memorial Media at 37°C on a rocker for 15 minutes. After 15 minutes, eyes were gently pressed through the strainer to create a cell suspension. The cell suspension was transferred to a 96-well plate, then cells were pelleted by centrifugation at 300×g for 5 minutes at 4°C, and the supernatant was discarded. For staining, cells were incubated with Live/Dead Aqua (Thermo Fisher Scientific) in PBS for 20 minutes at 4°C. After washing with staining buffer (Biolegend, San Diego, CA, USA), cells were labeled with an anti-mouse CXCR2 APC/Cyanine7 antibody (clone SA044G4) antibody (Biolegend) for 20 minutes at room temperature. After another wash, cells were stained for 20 minutes with the following antibodies (all from Biolegend unless otherwise noted): anti-CD45 PerCP/Cy5.5 (clone I3/2.3), Ly6G APC (clone 1A8), Ly6C Alexa Fluor 488 (clone HK1.4), Cd11b BV421 (clone M1/70), CD101 PE (clone Moushi/101; Thermo Fisher Scientific). Samples were analyzed using a four-laser spectral flow cytometer (Aurora Cytek, Cytek Biosciences, Fermont, CA, USA), containing 16 violet, 14 blue, 10 yellow–green, and 8 red channels (4L-16V-14B-10YG-8R). Spectral unmixing was performed using single stained reference controls and the SpectroFlo unmixing wizard (Cytek Biosciences, Fermont, CA, USA). Data were analyzed using FCS Express version 7 (DeNovo software, Pasadena, CA, USA). 

Histology

Eyes were harvested from euthanized mice at 6, 12, or 24 hours after infection, incubated in Perfix for 24 hours, then transferred to 70% ethanol for at least 24 hours. Eyes were paraffin embedded, sectioned, and stained with hematoxylin and eosin. 

Bacterial RNA Isolation and Quantitative RT-PCR

For in vitro experiments, expression of staphylococcal leukotoxin genes (lukSF-PV, hlgABC, lukED, and lukGH) was measured by quantitative real-time PCR (qRT-PCR). Total RNA was isolated from 18-hour overnight JE2 cultures grown in either BHI, tryptic soy broth (TSB), or filter-sterilized explanted rabbit vitreous. Overnight cultures were centrifuged at 4150×g for 5 minutes, and the resulting pellet was resuspended using RLT buffer (RNeasy minikit; Qiagen, Germantown, MD, USA). The resuspended pellet was transferred into tubes containing sterile 0.1-mm glass beads (Biospec Products Inc.) and homogenized for 60 s at 5000 rpm in a Mini-BeadBeater (Biospec Products Inc.). The tubes were then briefly pulsed and transferred to 1.5-mL snap cap tubes and centrifuged at 16,000×g for 2 minutes. RNA was isolated using the RNeasy mini kit, DNA was removed using DNase I (Zymo Research, Irvine, CA, USA), and RNA was purified (RNA clean and concentrator-5 kit; Zymo Research), all following the kit manufacturers’ instructions. RNA concentration and purity were determined using a Nanodrop (Thermo Fisher Scientific). qRT-PCR was performed (Applied BioSystems 7500; Thermo Fisher Scientific) using the iTaq Universal SYBR green one-step kit (Bio-rad, Hercules, CA, USA). We used 16s rRNA as the reference housekeeping gene and relative gene expression was calculated using the ΔC_T method. Primers used are listed in Table. 

Table.

 

Sequences of Primers Used for This Study

 

For total bacterial RNA isolation from eyes infected with strain JE2 or strain JE2 lukE::Tn, five infected eyes were pooled together in 400 µL PBS containing 1.0-mm sterile glass beads, then homogenized in a mini-bead beater as described elsewhere in this article. Ocular debris was pelleted by briefly pulsing using a mini benchtop centrifuge. Supernatant was then transferred to a 1.5-mL snap cap tube and bacteria was pelleted by spinning at 16,000×g for 2 minutes. Bacterial pellets were then resuspended using RLT buffer (RNeasy, Qiagen). The resuspended bacterial pellets were transferred into tubes containing sterile 0.1-mm glass beads (Biospec Products Inc.) and homogenized as described elsewhere in this article. The RNA was isolated, and DNA removed and purified following the manufacturers’ instructions as described elsewhere in this article. Quantitation of in vivo toxin gene expression was performed using RT-qPCR as described elsewhere in this article. 

Statistics

The Mann-Whitney U test was used for all statistical comparisons for all in vitro and in vivo analysis (GraphPad Prism 9 Software, Inc., La Jolla, CA, USA).^45,47–49 P values of less than 0.05 were considered significant. 

Results

JE2 and the LukED Transposon Mutant Exhibit Similar Growth Kinetics

The growth rates of JE2 and the isogenic JE2 lukE::Tn mutant were compared by quantifying CFU of these strains during 18 hours of incubation in BHI or explanted rabbit vitreous. Figure 1A shows that JE2 and JE2 lukE::Tn grew to similar concentrations (P > 0.05 at all time points), reaching stationary phase at 8 hours. Figure 1C shows an ex vivo growth curve in explanted vitreous. Similarly, JE2 and JE lukE::Tn grew to similar concentrations (P > 0.05 at all time points), reaching the stationary phase at 10 hours. The growth rate during exponential phase was also calculated in the two different environments. In BHI, JE2 grew at a growth rate of 3.79/hour ± 0.22/hour, whereas JE2 lukE::Tn grew at a rate of 3.90/hour ± 0.10/hour (Fig. 1B) (P > 0.9999). Similarly, there was no difference in the growth rate between JE2 (4.35/hour ± 0.16/hour) and JE2 lukE::Tn (4.50/hour ± 0.28/hour) in explanted vitreous (Fig. 1D) (P > 0.9999). These results show that JE2 and JE2 lukE::Tn exhibited similar growth kinetics in vitro and ex vivo, suggesting that any in vivo differences were unlikely to be influenced by the bacterial growth rate. 

Figure 1.

 

S. aureus JE2 and JE2 lukE::Tn exhibit similar growth kinetics in vitro and ex vivo. Overnight (18-hour) cultures of either S. aureus strain JE2 or its isogenic mutant JE2 lukE::Tn were diluted to approximately 10³ CFU/mL in either BHI or explanted rabbit vitreous. Growth curves calculated during growth in BHI (A) or vitreous (C) are shown. Growth rates calculated during exponential phase of growth in BHI (B) or vitreous (D) are shown. JE2 and JE2 lukE::Tn exhibited similar growth kinetics and growth rates in BHI and in explanted rabbit vitreous. Data represents the mean ± SEM of three independent cultures. Statistical significance is indicated as P < 0.05; ns, not significant.

Leukotoxin Subunits Are Expressed in Explanted Rabbit Vitreous

We reported that lukED was highly prevalent in the genomes of ocular S. aureus isolates.¹⁰ To determine if leukotoxins were expressed in an ocular environment, we performed qRT-PCR to detect expression of individual leukotoxin subunits (lukSF-PV, hlgABC, lukED, and lukGH) in explanted rabbit vitreous and normalized to 16s rRNA. Given that leukotoxins are under the control of the accessory gene regulator (Agr), we cultured JE2 in vitreous for 18 hours before isolating RNA from the samples. Agr-driven quorum sensing is activated at greater bacterial densities, increasing the expression of many virulence genes, including toxins, during the late logarithmic and stationary phases.^50,51 Results indicated that all leukotoxin subunits were expressed in explanted vitreous (Fig. 2A) and subunits within the same operon had similar CT values (lukS-PV vs. lukF-PV [P > 0.9999], lukE vs lukD [P = 0.4000], lukG vs lukH [P > 0.9999], hlgB and hlgC [P = 0.2000]). Given that the expression of leukotoxins depends on the type of growth media, we compared the expression of vitreous to that of laboratory media (BHI and TSB).^27,52–55 Transcript levels of all leukotoxin subunits were lower in vitreous compared with that in BHI, except for lukE and lukD. (Fig. 2B). Expression of lukE was similar in vitreous and BHI (fold change 1.13), although lukD expression was greater in vitreous (fold change 2.32). All leukotoxin subunits had lower expression in vitreous compared with expression in TSB (Fig. 2C). These results suggest that, although leukotoxin expression was generally lower in vitreous than in laboratory media, these toxins were still expressed, suggesting that these toxins may also be expressed in vivo. 

Figure 2.

 

Leukotoxin expression in BHI, TSB, and explanted rabbit vitreous. S. aureus strain JE2 was cultured in either BHI, TSB, or explanted rabbit vitreous for 18 hours. After 18 hours, RNA was isolated, and RT-qPCR was used to detect individual leukotoxin subunits in the three environments. The 16s rRNA was used as a control. (A) C_T values of leukotoxin subunits grown in vitreous. Fold change expression in vitreous relative to (B) BHI or (C) TSB. Leukotoxin subunits were expressed ex vivo. Transcript levels of lukE in explanted vitreous compared with BHI and a two-fold increase of lukD transcript levels were observed in vitreous compared with BHI. The remaining leukotoxin transcript levels were lower in vitreous compared with BHI. All leukotoxin transcripts were lower in explanted vitreous compared with TSB. Data are representative of mean C_T value ± SEM of n = 3 and mean fold change across three independent experiments.

Absence of LukED Does not Impact Retinal Function

To determine whether LukED is important in the pathogenesis of S. aureus endophthalmitis, we compared infection parameters in eyes infected with S. aureus JE2 or JE2 lukE::Tn, first assessing retinal function using scotopic ERG. After intravitreal injection of 5000 CFU of JE2 or the transposon mutant JE2 lukE:Tn into C57BL/6J mouse eyes, we dark-adapted mice for a minimum of 6 hours. At 6 hours after infection, no significant differences were observed in the percent A-wave retention between mice infected with JE2 or JE2 lukE::Tn (44.79% ± 5.39% vs 53.13% ± 5.68%; P = 0.3402) (Fig. 3A). Similarly, the percent retention for the B-wave was not statistically significant (50.51% ± 6.41% vs 58.81% ± 6.92; P = 0.5214) (Fig. 3B). Similarly, at 12 hours after infection, no differences were found between eyes infected with the JE2 or JE2 lukE::Tn in either the mean A-wave retention (40.78% ± 5.81% vs. 35.05% ± 3.80%; P = 0.6455) (Fig. 3A) or the mean B-wave retention (41.21% ± 7.90% vs. 35.62% ± 5.90%; P = 0.5054) (Fig. 3B). At 24 hours after infection, the mean A-wave retention (30.51% ± 1.69% vs 39.73% ± 2.81%; P = 0.3144) (Fig. 3A) and the mean B-wave retention (30.51% ± 5.11% vs 39.73% ± 5.13%; P = 0.3144) (Fig. 3B) did not differ between eyes infected with JE2 or JE2 lukE::Tn. These results show that the absence of LukED in S. aureus did not alter retinal function loss in this model. 

Figure 3.

 

The absence of LukE does not improve retinal function. Mouse eyes were infected with 5000 CFU of either S. aureus strain JE2 or S. aureus strain JE2 lukE::Tn. At 6, 12, or 24 hours after infection, retinal function was analyzed by ERG. Retinal function decline was observed in eyes infected with JE2 and JE2 lukE::Tn and these differences were not statistically significant. Values represent means ± SEM of eight eyes per group per time point. Statistical significance is indicated as P < 0.05; ns, not significant.

Absence of LukED Does not Affect Intraocular Growth

To determine whether LukED is important for the intraocular growth of S. aureus, we quantified intraocular CFUs in eyes infected with either JE2 or the JE2 lukE::Tn mutant. At 6 hours after infection, intraocular CFUs were significantly greater in eyes infected with JE2 compared with eyes infected with JE2 lukE::Tn (P = 0.0344) (Fig. 4). However, at 12 and 24 hours after infection, there was no difference in CFU/eye in these groups (P = 0.9723 and P = 0.2561, respectively) (Fig. 4). These results show that the intraocular bacterial growth of JE2 and JE2 lukE::Tn in infected eyes was similar, suggesting that LukED is not essential for S. aureus to establish an infection in the mouse eye. 

Absence of LukED Does not Affect Intraocular Inflammation

Inflammation was semiquantified by detecting the presence of MPO in eyes infected with JE2 or the JE2 lukE::Tn mutant. At 6 hours after infection, the mean MPO/eye between eyes infected with strain JE2 (365.37 ± 37.35 ng/eye) or JE2 lukE::Tn (431.46 ± 28.83 ng/eye) was not statistically significant (P = 0.1949) (Fig. 5). At 12 hours after infection, no significant difference between the mean MPO of eyes infected with JE2 (1243.85 ± 126.72 ng/eye) and the mean MPO of eyes infected with JE2 lukE::Tn (1024.82 ± 107.34 ng/eye) was observed (P = 0.1605) (Fig. 5). By 24 hours after infection, the mean MPO was greater in the JE2-infected eyes (2166.22 ± 166.53 ng/eye) compared with eyes infected with JE2 lukE::Tn (1790.28 ± 269.39 ng/eye), but the difference was not statistically significant (P = 0.2345) (Fig. 5). These results suggest that the LukED deficiency did not impact intraocular inflammation in this model. 

Absence of LukED Did not Affect Neutrophil Infiltration

During endophthalmitis, neutrophils are the primary and most abundant immune cells to infiltrate the eye. To investigate whether the absence of LukED affected neutrophil infiltration, we used flow cytometry to quantify neutrophil infiltration. Figure 6A illustrates the gating strategy used to identify neutrophils. 

We first gated on total CD45⁺ CD11b⁺ myeloid-derived cells (Fig. 6B). We observed that there was no statistical difference in infiltrating CD45⁺ Cd11b⁺ cells in eyes infected with JE2 or JE2 lukE::Tn at any time point. At 6 hours after infection, the mean number of infiltrating CD45⁺ Cd11b⁺ cells was 943 ± 269 in eyes infected with JE2 compared with 794 ± 266 in eyes infected with JE2 lukE::Tn (P = 0.6454). Similarly, no differences in infiltrating CD45⁺ Cd11b⁺ cells were found at 12 hours after infection (5449 ± 990 cells in eyes infected with JE2 and 4988 ± 619 cells in eyes infected with JE2 lukE::Tn; P > 0.9999) or at 24 hours after infection (53,582 ± 9027 cells in eyes infected with JE2 and 52,654 ± 15,998 cells in eyes infected with JE2 lukE::Tn; P = 0.4557). 

To identify neutrophils, CD45⁺ Cd11b⁺ myeloid-derived cells were then gated on Ly6G⁺ Ly6C⁺ cells. Similarly, there was no significant difference in neutrophil infiltration at any time point (Fig. 6C). At 6 hours, on average the total number of neutrophils was 282 ± 103 in eyes infected with JE2 compared with 188 ± 50 in eyes infected with JE2 lukE::Tn (P = 0.9812). At 12 hours after infection, the number of infiltrating neutrophils in eyes infected with JE2 was 4949 ± 916 compared with 4546 ± 501 in eyes infected with JE2 lukE::Tn (P = 0.9521). At 24 hours after infection, eyes infected with JE2 had 49,658 ± 8792 neutrophils compared with 46,484 ± 15,669 in eyes infected with JE2 lukE::Tn (P = 0.4557). At later time points, neutrophils comprised the majority of infiltrating myeloid cells (approximately 90% at 12 hours and 24 hours after infection). 

To determine whether infiltrating neutrophils were mature and active, we co-labeled cells with an antibody against maturation marker CD101. Neutrophils were then gated on CD101⁺ to determine the number and percentage of activated neutrophils. At 6 hours after infection (Fig. 6D), 141 ± 35 activated neutrophils were identified in eyes infected with JE2 compared with 163 ± 55 activated neutrophils in eyes infected with JE2 lukE (P = 0.7290). The percentage of activated neutrophils (Fig. 6E) in eyes infected with JE2 was 86.2% compared with 90% in eyes infected with JE2 lukE::Tn (P = 0.8629). At 12 hours after infection, the number of CD101⁺ neutrophils was 4691 ± 848 or 94.9% of infiltrating neutrophils in eyes infected with JE2 compared with 4497 ± 485 CD101⁺ neutrophils (P = 0.8785), or 99% of infiltrating neutrophils (P = 0.9015). At 24 hours, JE2-infected eyes had 46,665 ± 8631 CD101⁺ neutrophils, which accounted for 93.5% of all infiltrating neutrophils, compared with 43,447 ± 14,428 CD101⁺ neutrophils (P = 0.3829), or 94% of infiltrating neutrophils (P = 0.9015) in eyes infected with JE2 lukE::Tn. 

We also quantified the number and percentage of neutrophils expressing CXCR2, the receptor for LukED. Although the total number of CXCR⁺ neutrophils increased over time, the peak percentage of CXCR2⁺ neutrophils was observed at 12 hours after infection, followed by a decrease. At 6 hours after infection, the number of CXCR2⁺ neutrophils was 161 ± 44 in JE2-infected eyes compared with 105 ± 25 CXCR2⁺ neutrophils in JE2 lukE::Tn-infected eyes (P > 0.9999). The percentage of CXCR2⁺ neutrophils was 65.6% in JE2-infected eyes compared with 64% in JE2 lukE::Tn-infected eyes, with no significant difference between groups (P = 0.2190). At 12 hours after infection, 3518 ± 714 neutrophils or 68.8% of infiltrating neutrophils in eyes infected with JE2 were CXCR2⁺, whereas 3718 ± 416 neutrophils (P = 0.8785) or 85% of infiltrating neutrophils in eyes infected with JE2 lukE::Tn were CXCR2⁺ (P = 0.2190). At 24 hours after infection, the CXCR2⁺ neutrophil numbers in eyes infected with JE2 and JE2 lukE::Tn were 17,457 ± 2000 and 17,476 ± 4404, respectively (P = 0.6200). By this time point, the percentage of CXCR2⁺ neutrophils had decreased to 40.1% in JE2-infected eyes compared with 44.8% in JE2 lukE::Tn-infected eyes (P = 0.4557). These results demonstrate that even in the absence of LukED, the number of infiltrating neutrophils was similar, most infiltrating neutrophils were activated, and the percentage of CXCR2⁺ neutrophils declined throughout the course of infection. These data support our MPO data and suggest that the absence of LukED does not impact neutrophil recruitment in S. aureus endophthalmitis. 

Absence of LukED Did not Affect Changes in Ocular Architecture

Ocular pathological changes in eyes infected with either JE2 or JE2 lukE::Tn were compared by histology. In uninfected eyes, the retinal architecture was intact, there was an absence of inflammatory cells and fibrin, and the corneas were clear (Fig. 7). At 6 hours after infection, eyes infected with JE2 or JE2 lukE::Tn exhibited signs of inflammation and deposition of fibrin in the anterior and posterior chambers. At 12 hours after infection, eyes infected with JE2 or JE2 lukE::Tn exhibited an increase in fibrin and inflammatory cells in the posterior segment. Additionally, retinal detachments and corneal edema were observed. At 24 hours after infection, infected eyes continued to exhibit evolving inflammation and fibrin deposition, corneal edema, and damage to the retinal architecture. No differences in retinal architecture or inflammation were observed in histology sections of eyes infected with JE2 or the isogenic mutant at any time point, which is consistent with similarities in retinal function loss (Fig. 2) and inflammation (Figs. 5 and 6). Together, these results suggest that, although LukED is expressed ex vivo, this leukotoxin on its own may not play an important role in the pathogenesis of S. aureus endophthalmitis. 

All Leukotoxins Are Detected In Vivo During S. aureus Endophthalmitis

Our ex vivo data (Fig. 2A) showed that leukotoxins were expressed in an ocular environment (explanted vitreous). To confirm and compare in vivo expression of LukED and other leukotoxins, RT-qPCR was used to detect expression at 6, 12, and 24 hours after infection. We first compared the expression of leukotoxins at 12 hours with that at 6 hours (Fig. 8A). lukSF-PV and lukE had similar expressions at 6 and 12 hours (fold change lukS-PV = 1.26, lukF-PV = 0.97, and lukE = 0.97). We observed a downregulation of lukD and lukGH at 12 hours (fold change lukD = 0.38, lukG = 0.47, and lukH = 0.71). The only leukotoxin that was upregulated at 12 hours was the gamma toxin (fold change hlgA = 2.86, hlgB = 3.00, and hlgC = 3.75). We then compared the fold change in expression between 12 hours and 24 hours (Fig. 8B). At 24 hours, lukS-PV and the gamma toxin had similar expression compared with that at 12 hours (fold change lukS-PV = 0.95, hlgA = 0.85, hlgB = 1.32, and hlgC = 0.83). lukF-PV was the only subunit that had an increase in fold change at 24 hours compared with 12 hours (fold change lukF-PV = 1.76). lukED and lukGH were downregulated at 24 hours compared with 12 hours (fold change lukE = 0.62, lukD = 0.46, lukG = 0.38, and lukH = 0.22). 

Next, we wanted to determine if the expression of other leukotoxins would change in the absence of LukED. In eyes infected with JE2 lukE::Tn (Fig. 8C), there was no upregulation of lukS-PV and lukF-PV at 6 hours (fold change 1.03 and 1.54, respectively). lukG and lukH saw an approximate two-fold increase in expression (fold change 1.92 and 1.73, respectively). The gamma toxin subunits saw a greater than two-fold expression at 6 hours (fold change hlgA = 2.54, hlgB = 3.83 hlgC = 4.44). At 12 hours, we detected no differences in fold change expression in the absence of lukED (Fig. 8D). The fold changes for each leukotoxin subunit were lukS-PV (1.03), lukF-PV (1.54), lukG (0.89), lukH (0.71), hlgA (0.90), and hlgB (1.57) and hlgC (1.18). By 24 hours, we saw a downregulation of most leukotoxin subunits, except for lukGH (Fig. 8E). The fold change for each subunit were lukS-PV (0.35), lukF-PV (0.17), lukG (3.52), lukH (4.27), hlgA (0.33), hlgB (0.20), and hlgC (0.41). These results suggest that leukotoxin expression varies throughout the course of infection and that, in the absence of LukED, other leukotoxins may be expressed in vivo to compensate for the absence of this leukotoxin. 

Discussion

S. aureus is the most virulent organism among the genus Staphylococcus, and this virulence can be attributed to the many secreted virulence factors.⁵⁶ Among these virulence factors are leukotoxins, which are bicomponent pore-forming toxins that help the organism to evade the immune response. Leukotoxins are capable of lysing innate and adaptive immune cells based on cell surface receptors. However, all leukotoxins are capable of lysing neutrophils. Human S. aureus isolate genomes can encode up to five leukotoxins: PVL, HlgAB, gamma-toxin CB, LukED, and LukGH, the prevalence of which differs among S. aureus lineages.^23–25 HlgABC is encoded in the core genome and is highly conserved in all lineages; the gene is detected in 99% of sequenced strains.^57,58 LukGH is also encoded in the core genome, but whether this toxin is conserved among isolates is not known. LukGH is found in all publicly available S. aureus genomes.²⁸ LukED is encoded in a stable, yet highly variable pathogenicity island, vSaβ. This pathogenicity island is highly variable because lineages that do not encode LukED often encode genes for enterotoxins.^59,60 The presence of LukED is largely lineage specific, and its predicted frequency is approximately 70%.^57,58 PVL is located on temperate phase ϕSa2 and the prophage carrying the pvl gene is present in approximately 2% to 3% of all S. aureus isolates.⁶¹ However, PVL is found predominately in MRSA, especially in the USA300 lineage, the predominant strain of community-associated MRSA.^62,63 

We reported in a 10-year survey of ocular bacterial isolates that staphylococci comprised the vast majority of ocular infection isolates.¹⁰ For S. aureus isolates, we screened for virulence genes by PCR, including the leukotoxins. Among all S. aureus isolates, 76.7%, were PCR positive for lukED, 19.4% were positive for hlgABC, and 23.3% were positive for pvl.¹⁰ Previously, ocular isolates have been screened for pvl and hlgABC, but not many studies have reported the presence of lukED among ocular isolates and the presence of lukGH has not been reported.^37–40 Given that LukED has been implicated in the pathology of experimental S. aureus bacteremia and that, among ocular isolates lukED was the most prevalent, we decided to test whether LukED was essential in the pathogenesis of S. aureus endophthalmitis.²⁷ 

To address the hypothesis that LukED is essential for the pathogenesis of experimental S. aureus endophthalmitis, we first determined whether these toxins were expressed in an ex vivo ocular environment.⁶³ All leukotoxins were expressed ex vivo to varying degrees (Fig. 2A). Additionally, because it was reported that leukotoxin expression is media dependent, we compared expression in vitreous with expression in different laboratory media (BHI and TSB). Results indicated that, compared with laboratory media, most leukotoxins had lower transcript levels in explanted rabbit vitreous. Culturing S. aureus in BHI has been reported to increase expression of pvl. In addition, hlgA expression was greater in BHI compared with TSB.⁵² Transcriptional fusion reporters have been used in S. aureus USA300 to determine that the lukGH and pvl promoters are more transcriptionally active than are lukED and hlgABC when strains are grown in TSB.⁵³ LukED has been reported to be expressed at low levels in laboratory media, including BHI and TSB.⁵² LukED expression has been reported to increase in media containing yeast extract, casamino acids, and sodium pyruvate. LukED was initially identified and purified from culture supernatants of S. aureus strain Newman grown in media containing yeast extract, casamino acids, and sodium pyruvate.^27,54,55 Interestingly, our qPCR results showed that, when S. aureus is cultured in vitreous, LukED is the only leukotoxin with expression similar to that of BHI (Fig. 2). Although most leukotoxins were observed to be expressed to a lower degree in vitreous, these toxins were still detected ex vivo, suggesting that these toxins may be produced during an intraocular infection. Leukotoxins have been reported to have lytic effects in vitro at low concentrations (approximately 1 nM).^64,65 Additionally, the media-dependent differences in leukotoxin expression also demonstrate the complex regulation of these toxins. Although these toxins may be expressed at low levels ex vivo, other environmental factors may induce their expression in vivo. 

LukED binds to receptors CXCR1, CXCR2, CCR5, or Duffy antigen receptor for chemokines and causes lysis of different cells, depending on the availability of these receptors.^41–43 Deletion of LukED in S. aureus strain Newman and two clinical USA500 MRSA strains resulted in attenuation of virulence compared with that of the parental wild type and complemented strains in a murine bacteremia model.²⁷ That study also observed that LukED mutants had a 15-fold decrease in bacterial burden in the kidneys, suggesting that LukED is important for bacterial replication within hematogenous seeded tissue.²⁷ In our studies, we intravitreally injected 5000 CFU of strain JE2 or the transposon mutant JE2 lukE::Tn. We observed that, in the absence of LukED, there was no attenuation in the pathogenesis of S. aureus endophthalmitis. Inactivation of lukED in JE2 resulted in no difference in retinal function decline (Fig. 3), a result consistent with similarities in intraocular growth (Fig. 4), intraocular MPO concentrations (Fig. 5), inflammatory cell influx, fibrin deposition, and damage to the retinal architecture (Fig. 7) that would negatively affect A-wave and B-wave responses. During endophthalmitis, the retina and other ocular tissue are damaged by bacterial virulence factors and the host inflammatory response.^66,67 S. aureus cell wall components (peptidoglycan and wall teichoic acids) have been shown to contribute to ocular inflammation.^67,68 Purified toxins have also been shown to contribute to the inflammatory response, including intravitreal injections of purified PVL and gamma toxin subunits.³¹ The lack of differences in pathological changes in the S. aureus murine endophthalmitis model could be attributed to the fact that JE2 and JE2 lukE::Tn differ only in LukED, so the activities of the other instigators of pathology and inflammation result in infections that evolve similarly in this model. 

Figure 4.

 

The absence of LukE does not affect intraocular growth. Mouse eyes were infected with 5000 CFU of either S. aureus strain JE2 or S. aureus strain JE2 lukE::Tn. Infected eyes were harvested, homogenized, and serially diluted 10-fold at 6, 12, or 24 hours after infection to quantify intraocular CFU. Mean CFU/eye was statistically significant at 6 hours after infection with eyes infected with JE2 exhibiting greater growth. At 12 and 24 hours after infection, there was no difference in intraocular CFU in eyes infected with JE2 or JE2 lukE::Tn. Values represent mean CFU/eye ± SEM of n = 17 to 20 mice per group per time point. Statistical significance is indicated as P < 0.05; ns, not significant.

Figure 5.

 

The absence of LukE does not affect intraocular inflammation. Mouse eyes were infected with 5000 CFU of either S. aureus strain JE2 or S. aureus strain JE2 lukE::Tn-infected eyes were harvested and homogenized at 6, 12, and 24 hours after infection. Eye homogenates were used to assess intraocular inflammation by semiquantifying MPO using a sandwich ELISA. Mean MPO (ng/eye) was similar between eyes infected with JE2 or JE2 lukE::Tn at 6, 12, or 24 hours after infection. Values represent the mean ± SEM of MPO (ng/eye) of n = 8 eyes per group per time point. Statistical significance is indicated as P < 0.05; ns, not significant.

To quantify neutrophil infiltration, we first gated on all myeloid-derived cells (CD45⁺, Cd11b⁺), then further gated on double-positive Ly6G⁺ Ly6C⁺ cells to identify infiltrating neutrophils (Fig. 6A). Our flow cytometry results are consistent with intraocular MPO results, where we observed no difference in infiltrating immune cells whether eyes were infected with JE2 or JE2 lukE::Tn (Figs. 6B–C). We then gated on CD101⁺ to identify mature neutrophils. An increase in the CD101 receptor has been observed in mature murine neutrophils in the bone marrow and peripheral blood.⁶⁹ We quantified the number and percentage of CD101⁺ neutrophils and observed that the majority of the infiltrating neutrophils at all time points were CD101⁺. Quantities of CD101⁺ neutrophils were similar regardless of the infecting strain. CD101⁻ neutrophils have been shown to have incomplete nuclear development, aberrant MPO activity, and reduced phagocytic capability against Mycobacterium tuberculosis and Escherichia coli.^70,71 Other studies have suggested that CD101⁺ neutrophils are important to control Salmonella replication in vivo and in vitro. CD101 controls bacterial replication through metabolic modulation by the expression of Irg1 and Nox2, which produce itaconate and reactive oxygen species, respectively.⁷² Singh et al.⁷³ reported that, in the murine model of S. aureus endophthalmitis, Irg1 deficiency worsens ocular pathology and intraocular injection of itaconate reduces inflammation in Irg1-deficient and wild-type mice. Given that CD101⁺ neutrophils are important for bacterial clearance, yet there is no decrease in bacterial burden over the course of infection, this finding suggests that, although CD101⁺ neutrophils are infiltrating into the eye, their ability to clear bacteria is impaired. There is a need to characterize the heterogeneity of infiltrating neutrophils to explain why bacterial clearance fails. 

Figure 6.

 

The absence of LukE does not affect neutrophil infiltration. Mouse eyes were infected with 5000 CFU of either S. aureus strain JE2 or S. aureus strain JE2 lukE::Tn. Infected eyes were harvest and analyzed by flow cytometry at 6, 12, or 24 hours after infection. (A) Representative gating strategy used to identify infiltrating immune cells. Singlets were identified using forward scatter (FSC-A and FSC-H), followed by Live-Dead Aqua staining to identify live cells. Myeloid-derived cells were identified as CD45⁺ Cd11b⁺ and were further gated on Ly6G⁺ Ly6C⁺ to identify infiltrating neutrophils. Neutrophils were then gated on CD101⁺ and CXCR2⁺. (B) Quantitation of total number of CD45⁺ Cd11b⁺ cells at 6, 12, or 24 hours after infection. (C) Quantitation of total number of neutrophils at 6, 12, or 24 hours. Quantitation of total number (D) or percentage (E) of CD101⁺ neutrophils at 6, 12, or 24 hours after infection. Quantitation of total number (F) or percentage (G) of CXCR2⁺ neutrophils at 6, 12, or 24 hours after infection. Mean number of infiltrating CD45⁺ Cd11b⁺ and neutrophils were similar between eyes infected with JE2 or JE2 lukE::Tn at all time points. Similarly, the number and percentage of CXCR2⁺ or CD101⁺ neutrophils were similar between eyes infected with JE2 or JE2 lukE::Tn at all time points. Values represent mean ± SEM of seven to eight eyes per group per time point. Statistical significance is indicated as P < 0.05; ns, not significant.

Figure 7.

 

The absence of LukED does not protect retinal architecture. Mouse eyes were infected with 5000 CFU of either S. aureus strain JE2 or S. aureus strain JE2 lukE::Tn. Infected eyes were harvested at 6, 12, or 24 hours after infection and incubated in Prefer for 24 hours then transferred to 70% ethanol. Eyes were stained with hematoxylin and eosin. Eyes infected with JE2 or JE2 lukE::Tn showed evolving inflammation, deposition of fibrin, and retinal damage as the infection progressed. An uninfected eye is shown is shown as a reference. Sections are representative of n = 5 eyes per timepoint.

We also characterized the number and percentage of infiltrating neutrophils that expressed CXCR2, one of the receptors required for LukED to lyse their target cells. Our results indicated no difference in the number or percentage of live CXCR2⁺ neutrophils between eyes infected with JE2 or JE2 lukE::Tn. At 6 hours after infection, 65.6% of neutrophils in JE2-infected eyes were CXCR2⁺, whereas 64% of neutrophils were CXCR2⁺ in eyes infected in JE2 lukE::Tn. By 12 hours, the percentage of CXCR2⁺ neutrophils was 68.8% in JE2 infected eyes, whereas 85% of neutrophils were CXCR2⁺ in eyes infected with JE2 lukE::Tn. None of these differences were statistically significant. By 24 hours, the percent of CXCR2⁺ neutrophils decreased in both groups. In JE2-infected eyes, 40.1% of neutrophils were CXCR2⁺, whereas 44.8% of neutrophils were CXCR2⁺ in JE2 lukE::Tn-infected eyes (Fig. 8). LukED is also capable of lysing cells through interactions with CCR5 and CXCR1. Neutrophils lack expression of CCR5 and this receptor mediates the cell death of macrophages, T cells, and dendritic cells.⁴¹ Neutrophils can be also be lysed through interactions by CXCR1. However, given that we saw no difference in the quantity of neutrophils between the infection groups, it is unlikely that there would be differences in the quantities of CXCR1 neutrophils.⁴² In humans, lysis of neutrophils was prevented in the presence of CXCL8, which can bind to both CXCR1 and CXCR2. Administration of CXCL1, which binds to CXCR2, leads to lysis of human neutrophils, suggesting that both receptors need to be blocked to prevent cell lysis.⁴² The murine equivalent of these chemokines are keratinocyte chemoattractant and macrophage inflammatory protein 2. These chemokines have been expressed in retinal tissue as early as 3 and 6 hours after infection, respectively.^22,74,75 The presence of keratinocyte chemoattractant/CXCL1/CXCL8 and macrophage inflammatory protein-2 may block CXCR1 and CXCR2 on infiltrating neutrophils. This finding may explain why there was no difference in the percentage of live neutrophils when LukED is absent. 

Figure 8.

 

LukED and other leukotoxin are detected in vivo during endophthalmitis. Mouse eyes were infected with 5000 CFU of either S. aureus strain JE2 or S. aureus strain JE2 lukE::Tn. Infected eyes were harvested and bacterial RNA was isolated at 6, 12, or 24 hours. The expression of leukotoxin subunits were determined by qPCR. Fold change was compared in eyes infected with JE2 at (A) 12 hours relative to 6 hours and (B) 24 hours relative to 12 hours. Expression of leukotoxins in JE2 lukE::Tn-infected were determined and the fold change was compared with JE2 infected eyes. (C–E) Fold change of leukotoxin subunits with JE2 lukE::Tn-infected eyes relative to JE2 infected eyes at 6, 12, and 24 hours after infection. In eyes infected with JE2, hlgA, hlgB, and hlgC were observed to be upregulated at 12 hours after infection relative to 6 hours after infection. By 24 hours in eyes infected with JE2, there was an approximately two-fold change in lukS-PV relative to 12 hours. When comparing relative expression between JE2 or JE2 lukE::Tn-infected eyes, hlgA, hlgB, and hlgC were upregulated in JE2 lukE::Tn-infected eyes at 6 hours. At 12 hours after infection, there was similar expression between JE2 and JE2 lukE::Tn-infected eyes. At 24 hours after infection, lukG and lukH was upregulated in JE2 lukE::Tn-infected eyes relative to JE2 infected eyes. Data represent the fold change of three experiments with five eyes pooled per experiment. The 16s rRNA was used as a control for internal normalization.

qPCR results showed that all leukotoxin subunits were expressed in infected murine eyes starting at 6 hours after infection. In eyes infected with JE2, the fold change of lukS-PV, lukF-PV, and the lukE subunits at 12 hours was similar to that at 6 hours (Fig. 8A). There was a decrease in the fold change for lukD, lukG, and lukH at 12 hours relative to that at 6 hours. Also, for eyes infected with JE2, the fold change for lukS-PV, lukF-PV, hlgA, hlgB, and hlgC was similar at 24 hours after infection relative to that at 12 hours after infection. There was a decrease in fold change for lukE, lukD, lukG, and lukH at 24 hours after infection relative to that at 12 hours after infection (Fig. 8B). Previously, in a rabbit model of S. aureus endophthalmitis, PVL subunits hlgB and hlgC, but not hlgA, were detected after 24 hours after infection.⁷⁶ The expression of these genes was decreased when eyes were treated with 5 mg/kg of moxifloxacin at 24 hours after infection. There was also a downregulation of transcriptional regulators Agr (agr), staphylococcal accessory gene (sar), and an alternative sigma factor, sigB, after antibiotic treatment.⁷⁵ Previous findings have shown the importance of these transcriptional regulators in S. aureus virulence in the rabbit model of S. aureus endophthalmitis. The absence of Agr or the absence of both Agr and Sar led to improved retinal function and a decrease in the influx of inflammatory cells.^16,17 Previous findings have shown that leukotoxin transcript levels are more abundant during the late exponential and early stationary phases, suggesting that leukotoxins are regulated by Agr quorum sensing.^50,51 Sar is a transcriptional regulator of Agr quorum sensing that leads to the transcription of the P3 promoter to encode RNAIII. RNAIII negatively regulates the repressor of toxins, causing repressor of toxins to bind leukotoxin promoters to prevent the expression of leukotoxins.^51,77 LukED has been previously shown to be regulated by Sar and repressor of toxins.⁷⁸ Additionally, leukotoxins may be influenced by external stimuli. The two-component system, SaeRS, recognizes external stimuli and allows SaeR to bind leukotoxin promoters and activate leukotoxin expression.⁷⁹ RpiRC, a transcriptional regulator of enzymes involved in sugar catabolism, has been shown previously to regulate LukED and PVL expression negatively.⁸⁰ Although we would normally expect leukotoxin expression to increase as bacterial CFU increases, the complex regulation of leukotoxins may explain why leukotoxins did not seem to follow the expected gene expression pattern for genes under Agr control. 

Given that we observed no differences in infection parameters between eyes infected with JE2 and JE2 lukE::Tn, we investigated whether the upregulation of other leukotoxins in the absence of LukED had occurred (Figs. 8C–E). At 6 hours, we observed the upregulation of hlgABC and lukGH in eyes infected with JE2 lukE::Tn. By 12 hours, similar expression of all leukotoxin subunits was observed in the absence of LukED. At 24 hours, we observed a decrease in the expression of lukS-PV, lukF-PV hlgA, hlgB, and hlgC. However, there was an increased expression of lukG and lukH in eyes infected with JE2 lukE::Tn compared with expression of these genes in JE2. Using a lukGH reporter strain, LukGH has been reported to be expressed in renal abscesses, and, in the absence of LukGH, there was a 100-fold decrease in bacterial burden.⁵⁵ It has also been reported that LukGH is preferentially expressed in the presence of human neutrophils, allowing S. aureus to escape the phagolysosome.⁵³ Although those studies used human neutrophils, it may be possible that, in the absence of lukED, lukGH is upregulated in the presence of murine neutrophils to evade the immune response. The upregulation of LukGH at 24 hours after infection may indicate that, in the absence of LukED, LukGH may be expressed preferentially. If S. aureus is phagocytosed by neutrophils and LukGH expression is induced, this could mediate any internalized S. aureus to escape intracellular killing. 

All leukotoxins were expressed in vivo and the presence of other leukotoxins may explain why there was no difference in the pathogenesis of S. aureus endophthalmitis between eyes infected with JE2 and JE2 lukE::Tn. Although our results suggest that LukED may not individually contribute to the pathogenesis of S. aureus endophthalmitis, the same holds true for the gamma toxin in S. aureus endophthalmitis pathogenesis. Two independent studies showed that the absence of gamma toxin in S. aureus did not impact inflammation or retinal function loss in rabbit eyes.^9,32 However, in these studies, the expression of other leukotoxins was not tested. Our study and these two studies show that the absence of single leukotoxins resulted in no difference in the pathogenesis of S. aureus endophthalmitis. In a rabbit model of S. aureus keratitis, the absence of the gamma toxin resulted in a reduced slit-lamp examination score, but similar bacterial burden compared with the parental wild type.⁸¹ Leukotoxins are functionally redundant in their ability to lyse neutrophils, and in the absence of LukED, other leukotoxins are expressed in vivo. Collectively, leukotoxins may contribute to the pathogenesis of S. aureus endophthalmitis as a group. Although some leukotoxins, such as PVL, have a lower affinity to murine neutrophils compared with human neutrophils, it has been reported that, despite the lower affinity for murine neutrophils, deletion of leukotoxins in low-dose systemic infection and corneal infection models result in lower bacterial burdens.^55,82 Additionally, antibodies have been produced to target all leukotoxins, and neutralizing all leukotoxins has been shown to protect rabbit red blood cells from lysis in vitro and improve infection outcomes murine models of pneumonia and bacteremia.^83–85 The impact of leukotoxins should be assessed as a group to determine if these toxins may be functionally redundant in endophthalmitis. 

In conclusion, this study suggests that LukED may have a dispensable role in the pathogenesis of experimental S. aureus endophthalmitis. This may be attributed to the presence of keratinocyte chemoattractant and macrophage inflammatory protein-2 in the intraocular environment regardless of whether LukED is present, leading to competition of LukED binding to CXCR1 and CXCR2. Additionally, the expression of other leukotoxins may explain why we observed no differences in intraocular growth, intraocular inflammation, neutrophil infiltration, or retinal function. Expression of leukotoxins in vivo suggests that these toxins may have functional redundancies in vivo and the impact of all leukotoxins in vivo should be investigated. 

Acknowledgments

The authors thank Mark Dittmar and the personnel of the Dean McGee Eye Institute Animal Facility for their invaluable technical assistance. We also thank former lab member Md Huzzatul Mursalin for insightful discussions and technical assistance. We also acknowledge the OUHSC Live Animal Imaging and Analysis Core and the Ocular Immunobiology Core for technical assistance, and the Cellular Imaging Core for histology expertise (P30 Vision Core grant P30EY021725 to M.C.C). 

Funded by National Institutes of Health grant R01EY032073 (to M.C.C). Our research is also supported in part by National Institutes of Health grant R21EY035725 (to M.C.C) and an unrestricted grant to the Dean A. McGee Eye Institute from Research to Prevent Blindness, Inc. 

These data were presented at the 2024 Association for Research in Vision and Ophthalmology (ARVO) Annual Meeting, May 5–9, 2024, Seattle, Washington. The following reagents were provided by the Network on Antimicrobial Resistance in Staphylococcus aureus (NARSA) for distribution through BEI Resources, NIAID, NIH: Staphylococcus aureus subsp. aureus, Strain JE2, NR-46543, and Staphylococcus aureus subsp. aureus, Strain JE2, Transposon Mutant NE558 (SAUSA300_1769), NR-47101. 

Disclosure: L. Longoria-Gonzalez, None; P.S. Coburn, None; R. Astley, None; Y. Chen, None; M.C. Callegan, None