The cereus Matter of Bacillus Endophthalmitis

Abstract

Bacillus cereus (B. cereus) endophthalmitis is a devastating intraocular infection primarily associated with post-traumatic injuries. The majority of these infections result in substantial vision loss, if not loss of the eye itself, within 12-48 hours. Multifactorial mechanisms that lead to the innate intraocular inflammatory response during this disease include the combination of robust bacterial replication, migration of the organism throughout the eye, and toxin production by the organism. Therefore, the window of therapeutic intervention in B. cereus endophthalmitis is quite narrow compared to that of other pathogens which cause this disease. Understanding the interaction of bacterial and host factors is critical in understanding the disease and formulating more rational therapeutics for salvaging vision. In this review, we will discuss clinical and research findings related to B. cereus endophthalmitis in terms of the organism’s virulence and inflammogenic potential, and strategies for improving of current therapeutic regimens for this blinding disease.

1. Introduction

Endophthalmitis, of any etiology, is a rare but destructive ocular infection which is a potentially blinding complication following surgical procedures (post-operative), trauma (post-traumatic), or septicemia (endogenous) (Callegan et al., 2007; Durand 2017; Kernt and Kampik, 2010; Mahabadi et al., 2019; Nes, 2018; Safneck, 2012). In most cases of endophthalmitis, patients infected with avirulent organisms maintain good visual acuity following proper treatment. However, patients infected with virulent organisms may retain only navigational vision, despite proper and aggressive treatment. These patients may also require enucleation or evisceration of the eye in severe cases that are otherwise refractory to treatment (Callegan et al., 2002b; Kernt and Kampik, 2010; Safneck, 2012). Various organisms, including Gram-positive and Gram-negative bacteria and fungi are associated with endophthalmitis. Among all organisms associated with endophthalmitis, 85% of bacterial isolates were reported to be Gram-positive. (Parkunan and Callegan, 2016a; Schwartz et al., 2016). Post-operative endophthalmitis is the most prevalent and usually results from contamination with microbiota colonizing the skin surrounding the eye, the conjunctiva, or the eyelid. Organisms involved in these infections are typically avirulent coagulase-negative staphylococci. Staphylococcus aureus (S. aureus) is also a significant cause of these infections. (Durand, 2013; Nes, 2018; Coburn and Callegan, 2012). Twenty-five percent of all endophthalmitis cases have been reported to be attributed to non-surgical trauma, whereas 2-7% of penetrating ocular injuries have been reported to result in confirmed endophthalmitis (Agrawal, 2012; Ahmed et al., 2012; Bhagat et al., 2011; Dehghani et al., 2014; El Chehab et al., 2016; Essex et al., 2004; Gokce et al., 2015; Huber-Spitzy et al., 1986; Parke et al., 2012; Rommel and Kaplan, 1986; Schmidseder et al., 1998). Post-traumatic endophthalmitis cases are commonly caused by pathogens found in the environment, the most common being S. aureus and Bacillus (Agrawal, 2012; Ahmed et al., 2012; Bhagat et al., 2011; Carifi, 2012; Dehghani et al., 2014; El Chehab et al., 2016; Gokce et al., 2015; Jindal et al., 2014). Endogenous endophthalmitis is an unfortunate sequela of septicemia in which pathogens gain access into the eye via the retinal vasculature. This form of endophthalmitis is fairly uncommon, accounting for merely 2-8% of all endophthalmitis cases. Table 1 references the frequency and contribution of organisms in different types of endophthalmitis. However, the threat of bilateral infections with this type of endophthalmitis is highly concerning. Endogenous endophthalmitis is caused by a variety of pathogens, including S. aureus, streptococci, Bacillus cereus (B. cereus), Gram-negative Klebsiella pneumoniae (K. pneumoniae) and E. coli, and fungi such as Candida albicans (Callegan et al., 2002b; Callegan et al., 2007; Cornut and Chiquet, 2011; Cunningham et al., 2018; Jackson et al., 2003, 2014; Ness, 2007; Coburn and Callegan, 2012).

Table 1:

Frequency and contribution of organisms in different types of endophthalmitis.

Organisms	Post-Traumatic Endophthalmitis		Post-Operative Endophthalmitis		Endogenous Endophthalmitis
	Role	Frequency	Role	Frequency	Role	Frequency
Bacillus cereus	Y(1)	11-29%(2-5)	Y(6)	NA	Y(7)	NA
Staphylococcus aureus	Y(8)	NA	Y(9)	7-11.5%(10)	Y(11)	25%(12)
Staphylococcus epidermidis	Y(13)	22-42%(2-5)	Y(14)	45-50%(10)	NA	NA
Klebsiella pneumoniae	Y(15)	10-22%(2-5)	Y(16)	3-15%(10,17)	Y(10,17,18)	NA
Enterococcus faecalis	Y(19)	NA	Y(20)	NA	Y(21)	NA
Streptococcus pneumoniae	Y(22)	11-14%(2-5)	Y(22)	24-37.7%(10)	Y(22)	32% (12)
Candida albicans	Y(23)	5-15%(23)	Y(24)	NA	Y(25)	NA
Pseudomonas aeruginosa	Y(26)	NA	NA	NA	NA	NA

Y: Yes, NA: Not Available or Not Applicable

¹Schemmer et al., 1987;

²Affeldt et al., 1987;

³Essex et al., 2004;

⁴Kunimoto et al., 1999;

⁵Peyman et al., 1980;

⁶Rishi et al., 2013;

⁷David et al., 1994;

⁸Major et al., 2010;

⁹Seedor et al., 1990;

¹⁰Chiquet et al., 2010;

¹¹Ranjan et al., 2018;

¹²Avni et al., 2016;

¹³Sanchez et al., 2019;

¹⁴Pichi et al., 2014;

¹⁵Sridhar et al., 2014;

¹⁶Sanghi et al., 2014;

¹⁷Fisch et al., 1991;

¹⁸Kashani et al., 2013;

¹⁹Bhagat et al., 2016

²⁰Todokoro et al., 2017;

²¹Barge et al., 2013;

²²Miller et al., 2004;

²³Gupta et al., 2008;

²⁴Vinekar et al., 2014;

²⁵Sridhar et al., 2013;

²⁶Durand et al., 2016

Compared to other pathogens associated with any type of endophthalmitis, the bacterium B. cereus has emerged as an exceptionally virulent danger to the eye (Alfaro et al., 1996; Callegan et al., 2002c; David et al., 1994; Drobniewski, 1993; Kotiranta et al., 2000; Miller et al., 2008; Parkunan and Callegan, 2016). Although B. cereus has been mostly associated with self-limiting food poisoning cases, this organism is frequently isolated in blinding cases of post-traumatic and endogenous endophthalmitis (Abu el-Asrar et al., 1999; Bazinet, 2017; Eski et al., 2018; Galie et al., 2018; Glasset et al., 2018; Granum, 2017; Huber-Spitzy et al., 1986; Lima et al., 2017; Rommel and Kaplan, 1986). Compared to Bacillus infections in other body sites, B. cereus infections in the eye include an extremely robust inflammatory response (Ansell et al., 1980; Bottone, 2010; Bouza et al., 1979; Callegan et al., 1999a; David et al., 1994; Drobniewski, 1993; Lam, 2015; Miller et al., 2008). Prior to the mid 1940s, classification of Bacillus isolates from endophthalmitis cases did not extend beyond the genus level. Most of these isolates were incorrectly identified as Bacillus subtilis, a relatively avirulent laboratory contaminant (David et al., 1994; Shamsuddin et al., 1982). After 1948, criteria were established to discriminate B. cereus from other members of the Bacillus genus. However, in the late 19^th century, and even today, B. thuringiensis can be misidentified as B. cereus due to the genotypic and phenotypic similarities of these organisms (Baumann et al., 1984; Brumlik et al., 2004; Callegan et al., 2002a, 2003, 2006b; Carlson et al., 1994; Ivanova et al., 2003; Read et al., 2003).

B. cereus is a hardy and ubiquitous pathogen compared to other pathogens associated with endophthalmitis (Parkunan and Callegan, 2016a). B. cereus can thrive in extreme environments, namely because of its capacity for spore formation (McDowell and Friedman, 2018). Because of this, B. cereus is a common post-traumatic endophthalmitis isolate. B. cereus grows rapidly in the eye and possesses a complex cohort of virulence factors, and therefore, the window for successful therapeutic intervention is relatively short (Callegan et al., 2002c, 2005; Moyer et al., 2008, 2009). Despite recent progress in discerning the underlying mechanisms of disease development, the prognosis of B. cereus endophthalmitis remains poor (Callegan et al., 2002b; Parkunan and Callegan, 2016a; Wiskur et al., 2008b). In this chapter, we will review the epidemiology, clinical characteristics, and treatment of B. cereus endophthalmitis. Current knowledge regarding host/pathogen interactions in this disease will also be explored. Collectively, this information will assist clinicians and scientists in better understanding and treating this blinding disease.

2. Bacillus Endophthalmitis

2.1. Incidence and Epidemiology

B. cereus is more frequently associated with cases of post-traumatic endophthalmitis than it is with post-operative cases, although post-operative outbreaks of B. cereus endophthalmitis following contamination of surgical equipment have been reported (Altiparmak et al., 2007). Reports of B. cereus endophthalmitis associated with cataract surgery did not identify the source of contamination (Basak et al., 2012; Roy et al., 1997). It has been reported that 3-17% of total open globe injuries resulted in some form of endophthalmitis (Ahmed et al., 2012; Bhagat et al., 2016; Callegan et al., 2002b; Dehghani et al., 2014; Essex et al., 2004; Parke et al., 2012).

Bacillus spp. has primarily been reported as the causative organism in endophthalmitis associated with penetrating ocular trauma cases in adults, with incidences ranging from 9% to 45% (Affeldt et al., 1987; Asencio et al., 2014; Bhagat et al., 2016; Boldt et al., 1989; Chhabra et al., 2006; Essex et al., 2004; Jindal et al., 2014; Kunimoto et al.,1999; Long et al., 2014; Peyman et al.,1980; Schemmer and Driebe, 1987). Epidemiologic studies suggested that young people and males have a greater incidence of post-traumatic endophthalmitis, as these groups experience more ocular injuries (Cao et al., 2012; Jafari et al., 2010; May et al., 2000; Omolase et al., 2011; Pandita and Merriman, 2012; Soylu et al., 2010). One study suggested that compared to post-operative and endogenous cases of endophthalmitis, patients with post-traumatic endophthalmitis tended to be younger (Bhagat et al., 2016). In another report, open globe injuries occurred in children at a significantly greater incidence than that reported for the same type of injuries in adults (Agrawal et al., 2013; Ahmed et al., 2012; Altintas et al., 2011; Andreoli et al., 2009; Dehghani et al., 2014; Duch-Samper et al., 1997, 1998; Essex et al., 2004; Hooi and Hooi, 2003; Hosseini et al., 2011; Jonas et al., 2000; Khan et al., 2014; Lee et al., 2009; Long et al., 2014; Narang et al., 2003; Sabaci et al., 2002; Saxena et al., 2002; Schrader, 2004; Thevi and Abas, 2017; Zhang et al., 2010). The primary underlying cause for pediatric endophthalmitis has been reported to be ocular trauma (Narang et al., 2004). Associations between the incidence of B. cereus endophthalmitis and age or gender have not been reported. However, patients with traumatic ocular injuries have a high probability of being infected with B. cereus if the injury is associated with a retained intraocular foreign body (IOFB) (Boldt et al., 1989; Duan et al., 2016; Forster et al., 1980; Mieler et al., 1990; Shamsuddin et al., 1982; Shroff et al., 2016). It has been reported that in B. cereus post-traumatic endophthalmitis, greater than 70% of patients lost some degree of vision in the infected eye, and as few as 9% of those retained 20/70 or better visual acuity (David et al., 1994).

B. cereus is also a common cause of endogenous endophthalmitis. The organism gains access into the eye from the bloodstream, which is typically seeded from a distant infection site. The potential exists for bilateral endogenous endophthalmitis. In this disease, several risk factors, such as diabetes, immunocompromise, and drug abuse have been reported. Of these, the incidence of B. cereus endogenous endophthalmitis has been reported most often in association with the abuse of illicit intravenous drugs (Cowan et al., 1987; Masi, 1978; Miller et al., 2008; Okada et al., 1994; Vahey and Flynn, 1991). Associations of B. cereus infections with use of a specific type of drug have not been reported. However, reports suggest that the injection of B. cereus spores from contaminated drug paraphernalia or the drugs themselves are the likely source (Grossniklaus et al., 1985; Luong et al., 2017, 2019; Masi, 1978; Shamsuddin et al., 1982).

Most patients with Bacillus endophthalmitis develop acute signs of endophthalmitis within a few hours of injury. In endogenous cases, the time course of B. cereus endophthalmitis from the initial bloodstream infection to clinical signs is less concrete. However, because the majority of cases of B. cereus endophthalmitis occur in specific risk groups, being aware of clinical histories in patients who present with rapidly developing ocular inflammation and vision loss is critical to early and effective therapeutic intervention. Without appropriate treatment, patients with B. cereus endophthalmitis have a high chance of permanent vision loss.

2.2. Clinical Signs and Symptoms

Patients with bacterial or fungal endophthalmitis may exhibit loss of vision over a range of a few hours to several days. Most patients with endophthalmitis report severe ocular pain and discomfort (Johnson et al., 1997). Clinical signs and symptoms in B. cereus endophthalmitis are similar to that of endophthalmitis caused by other organisms, but occur at an accelerated rate (Parkunan and Callegan, 2016a). This characteristic is likely the result of faster replication in the eye and subsequently, a more robust inflammatory response (Callegan et al., 1999a; Parkunan and Callegan, 2016a). The time between injury and deterioration of vision in B. cereus endophthalmitis has been reported to be typically less than 48 hours (Bhagat et al., 2016; Carifi, 2012; Das et al., 2001; Hemady et al., 1990; Huber-Spitzy et al., 1986; Long et al., 2014; Rishi et al., 2016; Tran et al., 2003). Evolving clinical signs of B. cereus endophthalmitis may include severe ocular pain, chemosis of the conjunctiva, periorbital swelling, cells and flare in the aqueous humor, a hypopyon in the anterior chamber, and proptosis of the globe (Figure 1A) (Pan et al., 2017). Development of a corneal ring abscess is a classic clinical sign of Bacillus endophthalmitis (Figure 1B) (Bhagat et al., 2011; Chan et al., 2003; Miller et al., 2008). Systemic symptoms such as fever, polymorphonuclear leukocytosis, and malaise are also associated with Bacillus endophthalmitis (Boldt et al., 1989; Das et al., 2001; Davey and Tauber, 1987; David et al., 1994; Huber-Spitzy et al., 1986; Rommel and Kaplan, 1986; Roy et al., 1997; Schemmer and Driebe, 1987).

Figure 1:

(A) Photograph of a B. cereus-infected eye after penetrating trauma. This image demonstrates chemosis, corneal opacification, and proptosis. Copyright © 2017 Zhisheng Ke, MD, The Eye Hospital of Wenzhou Medical University, Wenzhou, China. (B) Photograph of a patient’s left eye infected with B. cereus following cataract surgery. The patient had diffuse eyelid swelling, conjunctival chemosis, an intact corneal wound, a corneal ring abscess, an anterior chamber hypopyon and fibrin infiltration, and an obscure fundus view. Reproduced with permission from Oxford University Press.

2.3. Diagnosis and Laboratory Identification

In comparison to all organisms which cause endophthalmitis, bacteria are isolated with the greatest frequency (Teweldemedhin et al., 2017). To confirm an accurate diagnosis of endophthalmitis and begin proper treatment, rapid microbiological identification is required. Proper diagnosis and laboratory identification are important for verification of the etiological agent, to initiate appropriate antibiotic treatment, to make decisions for or against ocular surgery, and to clearly document the epidemiology of the disease. In bacterial endophthalmitis, clinical diagnosis goes hand-in-hand with laboratory identification using cultures of the vitreous and/or aqueous, and the addition of blood cultures in endogenous cases. (Ledbetter et al., 2018; Mishra et al., 2018; Schwartz et al., 2016; Teweldemedhin et al., 2017; Yossa et al., 2017; Zghal et al., 2017).

Post-operative endophthalmitis is initially suspected based on clinical presentation and subsequently confirmed by laboratory testing of vitreous or aqueous specimens (Barza et al., 1997). It has been reported that vitreous specimens provide more accurate and reliable culture results than do cultures from aqueous humor (Donahue et al., 1993; Forster et al., 1980). Harvesting vitreous specimens require a similar anesthetic and operative setup as do intravitreal antibiotic injections, which are commonly performed immediately afterward (Vaziri et al., 2015). Diagnosis of post-traumatic endophthalmitis may be challenging if injury-induced symptoms are also present, but the presence of a hypopyon, vitritis, and/or ocular pain should be considered probable signs of infection (Ahmed et al., 2012; Gokce et al., 2015). For endogenous cases, systemic signs and symptoms of infection could aid in the diagnosis. However, patients with no overt signs of systemic infection could also suffer from endogenous endophthalmitis (Shankar et al., 2009). Endogenous cases might be confirmed only when clinical findings of endophthalmitis are linked with blood culture results from the systemic infection (Cowan et al., 1987; Dehghani et al., 2014; Donahue et al., 1993; Rahmani and Eliott, 2018; Vaziri et al., 2015).

Initial diagnosis of endophthalmitis is typically based on patient history and initial examination (Durand, 2016). Laboratory identification is then initiated by submitting a sample of purulent discharge and vitreous and/or aqueous fluid to the laboratory for microbiological stains and cultures (Bhagat et al., 2011, 2016; Iyer et al., 2001; Rommel and Kaplan, 1986; Schemmer and Driebe, 1987). Local antisepsis is required to collect intraocular samples. Samples should be obtained before antibiotic treatment is initiated, but antibiotics are rarely withheld pending laboratory results. Samples are submitted for Gram and/or calcofluor staining, as well as for routine microbiologic cultures and antimicrobial susceptibility testing. Routine microbiologic cultures include the direct plating of undiluted specimens on select liquid and solid media for simultaneous recovery and identification of organisms from the intraocular environment. Diluted specimens must be concentrated by using filters or cytospin (Ma et al., 2011; Miller, 2016; Ramakrishnan et al., 2009). The efficiency of microbiological identification depends on the quality of specimen collection and transport. For post-operative endophthalmitis, reported culture positive rates range from 10 to 70%, which is greater than that from post-traumatic cases (Safneck, 2012; Durand, 1996; Kunimoto et al., 1999; Melo et al., 2011; Pathengay et al., 2004). This traditional method of bacterial culture techniques has drawbacks, as sample concentration may be time consuming, and inoculated media must be incubated immediately (Miller, 2016). The Bactec technique offers a unique means of recovering ocular pathogens from vitreous samples. This method has advantages over the traditional method, including inoculation of a single medium, no need for immediate incubation, and no need for a supply of fresh agar media (Eser et al., 2007; Kratz et al., 2006; Rachitskaya et al., 2013; Tan et al., 2011; Tanaka et al., 2017). With both methods, laboratory results are typically reported 24-48 hours later.

A negative bacterial culture report does not exclude a diagnosis of endophthalmitis, and samples can be tested by molecular diagnostic methods (Durand, 2016). This is particularly important as volumes from these samples may be quite low. In recent years, molecular techniques such as PCR have gained clinical acceptance for identification of microbes from ocular infections, particularly in culture negative cases. (Chiquet et al., 2008; Ogawa et al., 2012a; Ogawa et al., 2012b; Sugita et al., 2012). In addition to the routine culture techniques noted above, ocular microbiologists have reported the use of new and more efficient technologies for rapid organism recovery and identification of pathogens from intraocular fluids, including real time and multiiplex PCR, DNA microarrays, matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF), peptide nucleic acid fluorescent in situ hybridization (PNA-FISH), and next-generation sequencing of ocular samples (Chiquet et al., 2008; Cornut et al., 2014; Hong et al., 2015; Kenchappa et al., 2006; Murray, 2012; Ogawa et al., 2012a; Ogawa et al., 2012b; Safneck, 2012; Sakai et al., 2012; Sugita et al., 2012; Van Gelder, 2001; Vaziri et al., 2015). In pan-bacterial PCR, 16S rRNA is amplified using universal primers complementary to DNA regions that are conserved among all bacterial species (Brillat-Zaratzian et al., 2014; Fisch et al., 1991; Lohmann et al., 1998; Sugita et al., 2011), covering a wide bacterial spectrum and providing a large panel of bacterial species (Aarthi et al., 2012; Sowmya and Madhavan, 2009; Sponsel et al., 2002; Sugita et al., 2011; Van Gelder, 2001; Yeung et al., 2009). For Bacillus and other organisms, 16S rDNA can be detected by PCR, sequenced, and compared to preselected genera or species, or to a larger database for speciation. Molecular identification methods facilitate more rapid and accurate microbiological diagnosis, can semi-quantify bacterial loads, and are able to screen for specific antigens from infectious agents (Sugita et al., 2011, 2013; Taravati et al., 2013). Since routine microbiological and molecular techniques are complementary, it is important to acquire intraocular samples as early as possible to enhance the rapid identification of microorganisms causing endophthalmitis, facilitating more rapid therapeutic intervention. This is especially critical for properly treating Bacillus endophthalmitis.

2.4. Risk Factors

Acquisition of a detailed patient history and thorough clinical examination is vital for each suspected case of endophthalmitis. For post-traumatic endophthalmitis, important information should include the patient’s location or activities at the time of the injury, as well as the mechanism of injury. B. cereus infection of a penetrating eye injury usually occurs from trauma with a soil-contaminated IOFB. This is more frequent in agricultural settings. (Ahmed et al., 2012; Boldt et al., 1989; Essex et al., 2004; Tran et al., 2003). A major risk factor associated with vision loss during post-traumatic endophthalmitis is the time between an injury and treatment. If the time to repair of an open globe injury extends past 24 hours, the risk of endophthalmitis is increased, regardless of the presence of an IOFB (Ahmed et al., 2012; Bialasiewicz et al., 2008; Nicoara et al., 2014; Schmidseder et al., 1998). Studies also indicate that the risk of endophthalmitis is greater if surgical intervention to remove an IOFB from the eye occurs greater than 24 hours after the injury (13.4%) compared to surgical removal of the IOFB within 24 hours after the injury (3.5%) (Thompson et al., 1993). This is highly critical in isolated or rural areas where an emergent ophthalmology consult may not be available, or in combat settings where high-velocity projectiles from explosions or gunfire enter the eye (Blanch et al., 2011; Gokce et al., 2015; Sabaci et al., 2002; Yeh et al., 2008). Any delay in the surgical procedure to close the globe and/or remove the IOFB may result in endophthalmitis. Wound characteristics are also considered as a risk factor in post-traumatic endophthalmitis cases (Kong et al., 2015). Studies on open globe injuries suggest that the presence of a laceration, such as a penetrating or perforating injury, and globe ruptures were individual risk factors for the development of endophthalmitis (Zhang et al., 2010). Injury with a soil-contaminated IOFB was also linked to a greater risk of endophthalmitis (Jindal et al., 2014; Kong et al., 2015; Nicoara et al., 2014; Zhang et al., 2010).

Reports suggest that the association of intraocular tissue prolapse, the site and extent of the ocular wound, and posterior segment involvement may increase the risk of Bacillus endophthalmitis. However, other reports disagree (Faghihi et al., 2012; Nicoara et al., 2014; Soheilian et al., 2007; Zhang et al., 2010), so it is unclear whether these factors contribute to an increased risk of endophthalmitis. These variables may be interrelated. For example, a larger wound affects a greater surface area that is subject to tissue contamination and an increased risk of tissue prolapse (Essex et al., 2004; Gupta et al., 2007; Zhang et al., 2010). Further studies are necessary to evaluate the significance of these variables in the development of post-traumatic endophthalmitis. Associations between these risks and the development of Bacillus endophthalmitis have not been made, but again, the rapidly blinding nature of a potential Bacillus infection in these scenarios should alert clinicians that appropriate treatment is critical.

2.5. Treatment, Prognostic Factors, and Prevention

In general, endophthalmitis caused by virulent pathogens is difficult to treat. Clinical severity is dependent upon on a number of factors, including the mechanism of entry, virulence of the infecting organism, and the immunocompetence of the patient (Callegan et al., 2007). The identity of the infecting pathogen is usually unknown when endophthalmitis is initially diagnosed, so clinicians must empirically treat the infected eye (Bhagat et al., 2016). Therapeutic options which target the infection site (i.e. intravitreal injections) pose a risk due to the delicate and nonregenerative nature of retinal tissues lining the interior of the eye (Schwartz et al., 2016). Moreover, not all empirically used antibiotics are effective against all endophthalmitis organisms. The threat of antibiotic resistance in ocular pathogens further complicates therapy (Bispo et al., 2016).

B. cereus endophthalmitis is an aggressive infection, likely due to its rapid replication and synthesis of toxins in the eye. Therefore, proper treatment should be initiated as early as possible. For ocular trauma, primary globe repairs are often performed with the removal of the IOFB (if found by B-scan ultrasonography, ultrasound biomicroscopy, plain film x-rays, or magnetic resonance imaging) (Agrawal, 2012; Bhagat et al., 2016; Dehghani et al., 2014; Rommel and Kaplan, 1986). If soil contamination of the IOFB is suspected, intravitreal injection with the appropriate antibiotics could be introduced while collecting the vitreal sample or removing the IOFB (Ren et al., 2014). If signs of endophthalmitis appear, a pars plana vitrectomy with antibiotic injection could be performed before vision deteriorates further. Although there is no unanimous standard of care which requires intravitreal injection of antibiotics if an IOFB is suspected, it is a reasonable practice to employ antibiotics as a preventative measure supported by many (Seal and Kirkness, 1992; Thevi and Abas, 2017; Yang et al., 2010). The infected eye should also be covered with a rigid shield to protect it from environmental exposure (Bhagat et al., 2011, 2016; Pan et al., 2017).

Upon the initial diagnosis of acute bacterial endophthalmitis, patients typically receive intravitreal injections of vancomycin and ceftazidime as empiric treatment options until culture results become available (Bhagat et al., 2016; Essex et al., 2008; Gupta et al., 2007; Jiang and Zhang, 2003; Nguyen and Hartnett, 2017; Talu et al., 2010; Vahey and Flynn, 1991). Aminoglycosides are effective against ocular pathogens, but are usually avoided because of reports of retinal toxicity. Vancomycin works effectively against all Bacillus isolates, but intravitreal clindamycin may also be used in cases where infection with Bacillus is suspected (Astley et al., 2016; Callegan et al., 2002b, 2007; Wiskur et al., 2008b). The use of systemic fluoroquinolones as adjunctive therapy in Bacillus endophthalmitis has also been suggested (Bhagat et al., 2016). In experimental rabbit B. cereus endophthalmitis, early intravitreal treatment at 2 hours or 4 hours with antibiotics such as vancomycin or gatifloxacin killed bacteria at a time when toxin levels in the eye were low. This early treatment limited retinal damage and preserved vision. A delay in treatment beyond 6 hours postinfection led to a substantial loss in retinal function and severe endophthalmitis (Wiskur et al., 2008b), highlighting the critical need for rapid identification and proper treatment for this blinding infection.

In most cases of endophthalmitis, patients receiving initial antibiotic treatment show clinical improvement by the time the culture findings are available. Endophthalmitis patients may then be treated with antibiotics which target the specific pathogen identified by culture. If cultures are negative, treatment is usually continued if the patient is improving. If the patient is not improving after 24-48 hours, further examination and imaging are conducted to confirm the absence of IOFBs and to identify other potential infectious etiologies, such as fungi or a polymicrobial infection, or whether multi-antibiotic resistance is an issue (Bhagat et al., 2016; Jindal et al., 2013; Mukherjee et al., 2014; Parkunan and Callegan, 2016). In the case polymicrobial or antibiotic resistant infections, a combination of different antibiotics may be required.

The final visual outcome in endophthalmitis depends on timely recognition and intervention. Administration of antibiotics is a mainstay in the management of endophthalmitis. However, significant intraocular inflammation may persist (Ching Wen Ho et al., 2018; Kim et al., 2017). To prevent potential negative effects from inflammation, anti-inflammatory drugs such as corticosteroids are often used in the treatment of endophthalmitis. However, the effectiveness of anti-inflammatory adjuncts for endophthalmitis is controversial (Bui and Carvounis, 2014; Callegan et al., 2001, 2002b, 2007; Novosad and Callegan, 2010; Wiskur et al., 2008b; Miller et al., 2019). A more direct therapeutic option, especially for Bacillus and other fulminant cases of endophthalmitis, is vitrectomy (Mieler et al., 1990). During routine vitrectomy, some or all of the vitreous humor is surgically removed and replaced with saline or an air or gas bubble. This process removes bacteria, products which cause inflammation, and inflammatory cells, clearing the visual tract. Fluid-gas or fluid-air exchange after vitrectomy is another possible treatment option (Lambrou et al., 1988). Vitrectomy with antibiotic administration has been reported to be useful in endophthalmitis caused by virulent organisms only when treatment is initiated early during infection (Bhagat et al., 2016). In experimental B. cereus endophthalmitis in rabbits, treatment with vitrectomy and vancomycin at the early time of 4 hours postinfection resulted in significantly greater retinal function than use of intravitreal vancomycin only (Callegan et al., 2011). Vitrectomy is designed to further reduce the infection and inflammation and help to control the infection, which is necessary to clear the visual tract and increase the potential for a positive clinical outcome (Bhagat et al., 2016; Callegan et al., 2007).

Bacillus have remained sensitive to ophthalmic antibiotics. Although antibiotics can kill Bacillus, these drugs do not affect the many toxins produced during infection which can damage tissue or activate inflammatory pathways. We recently demonstrated that biomimetic nanosponges (nanoparticles surrounded by erythrocyte membranes) can neutralize pore-forming toxins from Bacillus cereus and other Gram-positive ocular pathogens (Coburn et al., 2019). Although the use of nanosponges with or without gatifloxacin did not affect retinal function loss during experimental endophthalmitis, retinas in eyes treated with nanosponges were structurally intact and inflammation was greatly reduced, suggesting an improvement in clinical outcome. Additional studies testing the efficacy of anti-toxin therapy are needed as a follow up to these promising results.

Because of the high rate of vision and globe loss in B. cereus endophthalmitis, an adequate patient history and prompt and proper antibiotic, anti-inflammatory, and surgical intervention may be necessary in these cases (Davey and Tauber, 1987; David et al., 1994; Miller et al., 2008; Rommel and Kaplan, 1986; Schemmer and Driebe, 1987). Realistically, using protective eyewear while working in agricultural or mechanical settings could potentially avert traumatic ocular injuries, which would ultimately prevent this blinding disease. Regardless of the sensitivity of B. cereus to routinely used ophthalmic antibiotics, continued poor visual outcomes suggest that successful treatment options for B. cereus endophthalmitis have yet to be established.

3. Pathogenesis of Bacillus Endophthalmitis

3.1. Bacteriology

Among all the Bacillus species isolated from blinding cases of endophthalmitis, B. cereus is the most common. B. cereus is a Gram-positive, rod-shaped bacterium (Figure 2) that can grow aerobically and anaerobically. B. cereus forms spores, can resist desiccation, and is found in both temperate and extreme environments. This organism is β-hemolytic, producing a myriad of pore-forming and membrane-damaging toxins and enzymes. B. cereus is also a motile organism and is decorated with flagella and adhesive pili. B. cereus can assume a hyperflagellated state to migrate through viscous liquids. (McDowell and Friedman, 2018; Sankararaman and Velayuthan, 2013).

Figure 2:

(A) Gram staining of B. cereus. Magnification, 400X (B) B. cereus rods stained with Leifson Flagella Stain. Magnification, 1000X. (C) 1μL of B. cereus was pipetted onto trypticase soy agar (TSA) + 5% sheep blood agar and incubated for 18h at 37°C. Colonies with characteristics of double zone of hemolysis were observed at each inoculation point. (D) Transmission electron microscopy (TEM) of B. cereus. Flagella are denoted by black arrows, while white arrows denote pili.

Similar to other Gram-positive ocular pathogens, B. cereus have an inner membrane, a thick layer of peptidoglycan, teichoic acid, and membrane-bound lipoprotein on its surface. Unlike other Gram-positive ocular pathogens, B. cereus are decorated with flagella, pili, and an outer membrane protein called the S-layer which covers the entire surface of the organism (Budzik et al., 2007; Fouet and Mesnage, 2002; Mesnage et al., 2001; Mignot et al., 2001; Missiakas and Schneewind, 2017; Murakami et al., 1991; Tagawa, 2014; Zheng et al., 2013). The specific B. cereus factors that have been examined as contributing to endophthalmitis include its quorum sensing-dependent transcriptional regulator PlcR which controls extracellular virulence factor expression, individual membrane damaging toxins, its motility phenotype, and components of its cell wall.

B. cereus is a well-characterized organism that is typically thought of as a contaminant or a low-threat bacterium associated with self-resolving outbreaks of food poisoning (Becker et al., 1994; Granum and Lund, 1997; Kawashima and Obana, 1999; McDowell and Friedman, 2018; Schoeni and Wong, 2005; Tewari and Abdullah, 2015). Non-gastrointestinal sites that can be infected with B. cereus include skin (local infection/burns), blood (bacteremia/septicemia), respiratory tract (pneumoniae), cardiac tissue (endocarditis), central nervous system (brain abscesses and meningitis), and the eye (Barraud et al., 2012; Bekemeyer and Zimmerman, 1985; Brouland et al., 2018; Dolan et al., 2012; Frankard et al., 2004; Gaur et al., 2001; Gurler et al., 2012; Hansford et al., 2014; Henrickson, 1990; Hori et al., 2017; Hosein et al., 2013; Ikeda et al., 2015; Kato et al., 2016; Kelley et al., 2013; Martinez et al., 2007; Miyata et al., 2013; Ngow and Wan Khairina, 2013; Ozkocaman et al., 2006; Schaefer et al., 2016; Shimoyama et al., 2017; Soudet et al., 2017). These spore-formers are durable organisms, resistant to harsh environments, and can be found in all manner of particulates, contaminating objects responsible for penetrating injuries (Davey and Tauber, 1987; Gupta et al., 2007; Le Saux and Harding, 1987; Martinez et al., 2007; Rommel and Kaplan, 1986; Schemmer and Driebe, 1987).

3.2. Bacteria/Host Dynamics in B. cereus Endophthalmitis

As discussed earlier, the contribution of B. cereus to ocular infections is of clinical importance due to its rapid evolution and blinding complications. Endophthalmitis is a multifactorial infection, as both bacterial (envelope components, secreted products, and others) and host factors contribute to the pathogenesis of this disease. Table 2 references bacterial and host factors and their roles in endophthalmitis caused by different bacterial pathogens. Below we will discuss these factors in the context of the pathogenesis of B. cereus endophthalmitis.

Table 2:

Contributors to the Pathogenesis of Bacterial Endophthalmitis

Role of Bacterial Factors in Endophthalmitis
	B. cereus	S. aureus	E. faecalis	S. epidermidis	S. pneumoniae	K. pneumoniae
Secreted products	Yes(1-5)	Yes(1,6)	Yes(7)	Yes(8)	Yes(9)	Yes(10)
Quorum sensing	Yes(1,11)	Yes(12)	Yes(13,14)
Cell wall	Yes(1)	Yes(1)	Yes(1)	Yes(15)		Yes(10)
Capsule					Yes(16)	Yes(10,17)
Flagella	Yes(18)
Pili	Yes(19)
S-layer	Yes(20)
Motility	Yes(2,21)
Role of Host Factors in Bacterial Endophthalmitis
	B. cereus	S. aureus	E. faecalis	S. epidermidis	S. pneumoniae	K. pneumoniae
TLR2	Yes(22)	Yes(23,24)
TLR5	No(18)
TLR4	Yes(25)					Yes(26)
NOD2	No(27)
NLRP3	No(27)
Inflammatory mediators	Yes(28,29)

¹Callegan et al., 1999a;

²Callegan et al., 2005;

³Beecher et al. 1995;

⁴Beecher et al., 2000;

⁵Moyer et al., 2008, 2009;

⁶Kumar et al., 2015;

⁷Todokoro et al., 2017;

⁸Mahajan et al., 1982;

⁹Ng et al., 1997;

¹⁰Wiskur et al. 2008a;

¹¹Callegan et al., 2003;

¹²Booth et al., 1995, 1997;

¹³Engelbert et al. 2004;

¹⁴Mylonakis et al. 2003;

¹⁵Pleyer et al. 1992;

¹⁶Sanders et al., 2011;

¹⁷Hunt et al., 2011;

¹⁸Parkunan et al., 2014a;

¹⁹Callegan et al., 2017;

²⁰Mursalin et al., 2019;

²¹Callegan et al., 2006a;

²²Novosad et al., 2011;

²³Shamsuddin and Kumar 2011a;

²⁴Kumar et al., 2010;

²⁵Parkunan et al., 2015;

²⁶Hunt et al., 2014;

²⁷Astley et al., 2016;

²⁸Ramadan et al., 2006, 2008;

²⁹Parkunan et al., 2016b.

3.2.1. Role of Bacterial Factors in B. cereus Endophthalmitis

During replication, B. cereus synthesizes numerous potential virulence factors, including secreted toxins whose expression is coordinated by the quorum sensing regulator PlcR (Agaisse et al., 1999; Brillard and Lereclus, 2007; Callegan et al., 2003; Declerck et al., 2007; Gilois et al., 2007; Gohar et al., 2008; Jia et al., 2015; Sastalla et al., 2010). Toxins regulated by this system include hemolysin BL (HBL), hemolysin I (cereolysin O), hemolysin III, hemolysin IV, non-hemolytic enterotoxin (NHE), cereulide, collagenases, cytotoxin K, phosphatidylinositol-specific phospholipase C (PI-PLC), sphingomyelinase, and phosphatidylcholine-specific phospholipase C (PC-PLC) (Agaisse et al., 1999; Arslan et al., 2014; Chon et al., 2012; De Santis et al., 2008; Fagerlund et al., 2010; Fluer, 2007; Forghani et al., 2014; Gohar et al., 2008; Granum, 1994; Jessberger et al., 2014; Oh et al., 2015; Ramarao and Sanchis, 2013; Schoeni and Wong, 2005). HBL was the first toxin of B. cereus investigated for its involvement in intraocular virulence (Beecher et al., 1995). Although crude B. cereus supernatants containing HBL and purified HBL caused intraocular inflammation when injected into rabbit eyes, deletion of HBL in B. cereus did not attenuate its virulence in an experimental rabbit endophthalmitis model (Callegan et al., 1999b). Commercial PI-PLC and PC-PLC, and purified HBL, cereolysin O, and sphingomyelinase were individually examined for toxicity on rabbit retinal tissue in vitro. In that study, HBL and PC-PLC were toxic to retinal buttons (Beecher et al., 2000). However, the virulence of isogenic B. thuringiensis mutants deficient in PC-PLC or PI-PLC in the rabbit eye was not attenuated (Callegan et al., 2002a). Together, these findings suggested that in isolation, some toxins may not be necessary for B. cereus endophthalmitis pathogenesis, and the combination of these bacterial products are what confers intraocular virulence to this organism.

Quorum sensing-regulation of toxins and enzymes as a group is significant in the context of B. cereus ocular virulence (Gohar et al., 2008; Salamitou et al., 2000; Sastalla et al., 2010). The contribution of the quorum sensing regulator PlcR was examined in the rabbit experimental endophthalmitis model (Callegan et al., 2003). Bacillus deficient in PlcR regulation do not produce several of the toxins listed above. Inflammation and retinal function loss were significantly delayed in eyes infected with the isogenic PlcR mutant compared to eyes infected with wild type Bacillus (Callegan et al., 2003). This result suggested the importance of PlcR-regulated virulence factors as a group in rapidly evolving Bacillus endophthalmitis. To investigate the role of PlcR-regulated toxins on blood ocular barrier (BOB) function, we injected sterile cell-free supernatants generated from wild-type or plcR-deficient B. cereus into mouse eyes. Non-plcR regulated secreted factors caused BOB permeability. These results demonstrated that specific tight junction loss in the BOB was not dependent on the presence of a functional PlcR. Together, these findings suggest that different bacterial products play multifactorial roles in the pathogenesis of B. cereus endophthalmitis (Moyer et al., 2008, 2009).

The rapid replication of B. cereus may contribute to intraocular virulence because this phenotype rapidly introduces cell wall associated factors which trigger intraocular inflammation (Callegan et al., 1999a). The B. cereus envelope includes cell wall components, the cell membrane, and associated proteins, and varies structurally from other Gram-positive ocular pathogens (Dufresne and Paradis-Bleau, 2015; Rajagopal and Walker, 2017; Siegel et al., 2016). When injected into the rabbit eye, cell wall preparations of B. cereus caused greater intraocular inflammation than that of preparations of cell walls from S. aureus and E. faecalis (Callegan et al., 1999a). Flagella aid B. cereus in migration through the eye, but are weak activators of Toll-like receptor (TLR)-5, the canonical innate immune receptor that recognizes flagella (Parkunan et al., 2014a). Deficiencies in motility and swarming in B. cereus hamper bacterial migration within the eye, resulting in less severe inflammation (Callegan et al., 2005; Callegan et al., 2006). Infections with B. cereus deficient in pili (an adhesion appendage) led to a reduction in intraocular bacterial burden compared to that of wild-type infected mouse eyes, suggesting a possible function for pili in defending B. cereus from intraocular clearance (Callegan et al., 2017). The inflammatory capacity of common Gram-positive envelope components is well documented for other types of infections. However, the roles of common and unique envelope components in B. cereus have not been addressed in the context of endophthalmitis (Ginsburg, 2002; Iyer et al., 2010; Langer et al., 2008; Nguyen and Gotz, 2016). We recently demonstrated that infection with a S-layer deficient Bacillus thuringiensis resulted in significantly retained retinal function, reduced disease severity and inflammation compared to wild-type infected mouse eyes (Mursalin et al., 2019). This suggests that Bacillus S-layer protein contributes to the pathogenesis of endophthalmitis, potentially by triggering innate inflammatory pathways in the retina.

3.2.2. Role of Host Factors in B. cereus Endophthalmitis

The eye enjoys protection from the immune system, a phenomenon known as “immune privilege” (Streilein, 1995, 1999; Streilein et al., 2002). During endophthalmitis in mice, rabbits, and perhaps humans, immune privilege is altered and inflammation results. In a novel zebrafish endophthalmitis model, eyes maintained immune privilege, rapidly cleared staphylococci, and did not develop endophthalmitis (Mei et al., 2019), so the acute response appears to be model-specific. The primary immune response that provides the initial defense against invading microbes is host innate immunity (Akira, 2009; Jang et al., 2015; Takeuchi and Akira, 2007). This system senses the existence of foreign invaders and activates pathways which function to eradicate these hazards. Recognition of microbes and their associated products is accomplished through pattern-recognition receptors (PRRs). Surface receptors scan nearby extracellular spaces, while intracellular receptors inspect engulfed microbial determinants known as pathogen associated molecular patterns (PAMPs) that may indicate the presence of a foreign invader (Brubaker et al., 2015). Toll-like receptors (TLRs) are a group of PRRs that are present in the retina and have been identified on Muller cells, retinal pigment epithelial cells (RPE), glial cells, and photoreceptor cells (Chang et al., 2006; Singh and Kumar, 2015). Peptidoglycan, teichoic acids, lipoproteins, flagella, and pili are PAMPs associated with the B. cereus envelope. Theoretically, immune responses triggered by the interaction of retinal cell PRRs with B. cereus PAMPs are designed to combat intraocular infection by recruiting inflammatory cells into the eye. The predominant immune cells that infiltrate into the posterior segment during bacterial endophthalmitis are neutrophils. Infiltration of these cells into the clear vitreous environment causes loss of vitreal clarity and may cause bystander damage to the retina, but should also serve to limit infection. Therefore, the absence of or diminished activation of the immune response should result in a more severe infection.

The role of TLRs in the inflammatory response during endophthalmitis has been extensively studied in experimental mouse models using S. aureus, B. cereus, and K. pneumoniae as the offending pathogens (Coburn et al., 2018; Hunt et al., 2014; Kumar and Shamsuddin, 2012; Kumar et al., 2010; Kumar and Yu, 2006; Novosad et al., 2011; Parkunan et al., 2014). The B. cereus envelope contains putative TLR2 agonists such as Peptidoglycan, lipoproteins, and teichoic acids. Intraocular infections in mice with defective TLR2 receptors were muted. B. cereus-infected eyes of TLR2^−/− mice had significantly lower concentrations of proinflammatory mediators and therefore, reduced inflammatory cell influx. These mouse eyes also exhibited greater retained retinal function and less retinal damage compared to infected eyes of wild-type mice (Novosad et al., 2011). This suggested a significant role for TLR2-mediated responses in B. cereus endophthalmitis. TLR2 was also shown to be a significant contributor to inflammation in S. aureus endophthalmitis (Kumar et al., 2010). TLR4 was demonstrated to be important for the pathogenesis of Gram-negative endophthalmitis caused by K. pneumoniae (Hunt et al., 2014). This PRR recognizes the lipopolysaccharide (LPS) of Gram-negative bacteria (Lu et al., 2008). However, B. cereus is a Gram-positive pathogen and would not be expected to incite a TLR4-mediated response. Unexpectedly, TLR4^−/− mouse eyes infected with B. cereus had a significantly reduced inflammatory response, as well as less retinal damage and retinal function loss compared to infected wild-type mouse eyes (Parkunan et al., 2015). The expression and levels of proinflammatory mediators were also significantly less in infected TLR4^−/− mouse eyes, suggesting that inflammation in B. cereus endophthalmitis was also mediated through TLR4 (Parkunan et al., 2015). Involvement of TLRs were also confirmed as the roles of their adaptors myeloid differentiation primary response gene-88 (MyD88) and Toll/interleukin-1 receptor (TIR) domain containing adaptor-inducing interferon- β (TRIF) were confirmed in B. cereus endophthalmitis (Parkunan et al., 2015).

The terminal downstream event of innate receptor activation is the synthesis of inflammatory mediators. The concentrations of cytokines tumor necrosis factor alpha (TNF-α), interleukin-1 beta (IL-1β), IL-6, and macrophage inflammatory protein 1 alpha (MIP-1α), and chemokine KC (C-X-C motif ligand 1 ([CXCL1]) increased in parallel with BRB permeability and inflammatory cell influx during experimental B. cereus endophthalmitis (Moyer et al., 2009; Novosad et al., 2011; Parkunan et al., 2014; Parkunan et al., 2015, 2016b; Ramadan et al., 2008). Moreover, we recently identified an additional cohort of inflammatory mediators (CXCL2 [MIP2-α], CXCL10 [IP-10], chemokine (C-C) motif ligand 2 [CCL2]/monocyte chemoattractant protein 1 [MCP1], and CCL3 [MIP1-α]) whose expression is dependent upon a functional TLR4 during B. cereus intraocular infection (Coburn et al., 2018). Taken together, these studies suggest potential targets for bacterial endophthalmitis by interfering with the activation of TLR-regulated pathways.

Production of these inflammatory mediators often occurs in parallel with a worsening clinical picture in endophthalmitis models. The significance of TNF-α in the acute innate response in experimental B. cereus endophthalmitis was demonstrated in infections in the TNF-α^−/− mouse (Ramadan et al., 2008). When TNF-α was absent, few neutrophils were recruited to the infection site and B. cereus replicated more rapidly, causing a faster decline in retinal function compared to that of infected wild type mouse eyes (Ramadan et al., 2008). In experimental B. cereus endophthalmitis, secretion of IL-6 and KC/CXCL1 occurred in parallel with increased influx of inflammatory cells. The expression of these inflammatory mediators was dependent on functional TLR2, TLR4, and their adaptors (Novosad et al., 2011; Parkunan et al., 2014, 2015). KC/CXCL1 recruits immune cells into infected tissue. Inflammatory responses and disease severity were significantly reduced in the eyes of KC^−/− mice infected with B. cereus (Parkunan et al., 2016b). In contrast, the absence of IL-6 did not radically alter the overall immune response to B. cereus (Parkunan et al., 2016b). It has been reported that the caspase-1/nucleotide binding domain, leucine-rich repeat protein 3 (NLRP3) inflammasome complex contributes to intraocular IL-1β expression during endotoxin-induced uveitis (Rosenzweig et al., 2012). The upregulation of IL-1β in experimental B. cereus endophthalmitis suggested a potential involvement of NLRP3. However, eyes of NLRP3^−/− mice had a similar retinal function decline and inflammatory response during B. cereus endophthalmitis as did infected eyes of wild type mice (Astley et al., 2016). We have reported that some of these pathways (NOD2, NLRP3, TLR5) and mediators (IL-6) did not play significant roles in B. cereus intraocular infections. However, their participation in slowly-evolving forms of endophthalmitis caused by other pathogens should not be excluded.

Both bacterial and host factors contribute to the devastating progression of B. cereus endophthalmitis. This is also true for other forms of bacterial endophthalmitis. Therefore, to be successful, therapeutic design might include inhibition of pathways involved in global bacterial toxin production, direct inhibition of toxin activities, or blocking innate interactions that would ultimately prevent the activation of inflammatory pathways or downstream mediators which would blunt further recruitment of inflammatory cells. The next logical step to designing more rational therapies is the testing of these novel strategies as adjuncts to antibiotics with the goal of improving visual outcomes in microbial endophthalmitis.

3.3. Significance of B. cereus as an Ocular Pathogen

B. cereus, B. thuringiensis, and B. anthracis are members of the B. cereus sensu lato group. B. cereus has been reported to be the most commonly isolated member of this group and of the genera Bacillus isolated from blinding cases of endophthalmitis (Awan, 1992; Callegan et al., 2006b; David et al., 1994; Le Saux and Harding, 1987). B. cereus is one of the most prevalent virulent pathogens associated with post-traumatic endophthalmitis, ranked second after S. aureus (Parkunan and Callegan, 2016a). Visual outcomes in any form of endophthalmitis depend on fast and appropriate therapeutic intervention, which is typically targeted toward the type of infecting pathogen. Fortunately, intraocular infections by an avirulent organisms (such as coagulase-negative staphylococci) are usually cleared with the use of efficacious topical antibiotics and an appropriate immune response in the patient, and vision loss is minor or absent. However, infection with a virulent pathogen may result in a fulminant but insufficient inflammatory response, and an infection that may be intractable to even intravitreal antibiotics and corticosteroids (Astley et al., 2016; Callegan et al., 2002b, 2005, 2007; Durand, 2002, 2017; Durand and Dohlman, 2009; Parkunan and Callegan, 2016).

Experimental animal models of endophthalmitis have been utilized to understand bacterial virulence strategies as well as the host response to infection (Astley et al., 2016). In general, an invading pathogen is recognized by innate immune receptors, initiating a molecular cascade which results in an inflammatory response. For endophthalmitis, this response includes the production of inflammatory mediators which recruit inflammatory cells in an effort to clear the infection from the posterior segment. At the same time, replicating bacteria produce numerous toxins and enzymes that damage the retina and impair its function (Astley et al., 2016; Callegan et al., 1999a; Durand, 2002, 2016; Durand and Dohlman, 2009; Gregory et al., 2007; Parkunan and Callegan, 2016; Ramadan et al., 2006).

We reported that secreted products of B. cereus were extremely toxic on human retinal cells and that the intraocular presence of B. cereus toxins caused rapid retinal destruction and significant inflammation (Moyer et al., 2008, 2009; Callegan et al., 2006b). In vitro analysis of the cytotoxicity of bacterial secreted products on human retinal Muller cells is demonstrated in Figure 3. Filter-sterilized supernatant from overnight cultures of B. cereus, S. aureus, Enterococcus faecalis (E. faecalis) (LaGrow et al., 2017), Staphylococcus epidermidis (S. epidermidis), and Streptococcus pneumoniae (S. pneumoniae) were incubated with human retinal Muller cells. Cell death was semi-quantified by release of lactate dehydrogenase (LDH). B. cereus secreted products killed 60% of retinal Muller cells in 45 minutes. In comparison, little or no cytotoxic effects were observed in Muller cells exposed to the secreted products of other Gram-positive pathogens over the same amount of time. While these studies were limited to a controlled in vitro setting, these results suggest that B. cereus secreted products are more harmful for retinal Muller cells compared to that of other endophthalmitis pathogens.

Figure 3:

Comparative analysis of the cytotoxicity of bacterial secreted products on human retinal Muller cells (MIO-M1). Muller cells were treated with overnight culture supernatants from B. cereus, S. aureus, E. faecalis, S. epidermidis, or S. pneumoniae for 45 minutes. For E. faecalis, the cytotoxicity of supernatants from cytolysin component LL- and LS-producing strains were tested as previously described in LaGrow et al., 2017. The levels of LDH release from affected retinal Muller cells were measured using the Pierce LDH Cytotoxicity Assay Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. Data is represented as the mean ± SEM of % cytotoxicity of retinal Muller cells based on the positive LDH control. Statistical significance was determined by unpaired Student t-test (**P ≤ 0.01, N=3/group).

To further illustrate the distinctive virulence of B. cereus compared to other pathogens in endophthalmitis, we conducted a snapshot comparative analysis of experimental endophthalmitis (Figure 4). We intravitreally injected B. cereus, E. faecalis, or S. pneumoniae at 100 CFU/eye, or injected S. aureus or S. epidermidis at 5000 CFU/eye to induce experimental endophthalmitis in C57BL/6J mice (Astley et al., 2016). Figure 4A illustrates the intraocular bacterial growth of these organisms at 12 hours postinfection. The intraocular burden of B. cereus and S. epidermidis reached 10⁸ CFU/eye at 12 hours postinfection. We and others have reported that K. pneumoniae, S. pneumoniae, and E. faecalis reach a similar concentration at 24 hours postinfection, and S. aureus reach a similar concentration in the eye at 48 hours postinfection (Callegan et al., 1999a; Hunt et al., 2014; Meredith et al., 1990; Parkunan and Callegan, 2016a; Ramadan et al., 2006). By 12 hours postinfection, B. cereus, E. faecalis, and S. pneumoniae grew from 100 CFU to 7.2X10⁷, 2.2X10⁷, and 3X10⁴ CFU/eye, respectively. S. aureus and S. epidermidis grew from 5000 CFU/eye to 1.4X10⁵ and 1.4X10⁷ CFU/eye over the same time period. Figure 4B depicts the intraocular growth rates of B. cereus and other endophthalmitis pathogens in this experiment. B. cereus grew at a rate of 1.9 hour ⁻¹, whereas S. aureus, E. faecalis, S. epidermidis, and S. pneumoniae grew at rates of 0.2 hour ⁻¹, 1.7 hour ⁻¹, 0.7 hour ⁻¹ and 0.5 hour ⁻¹, respectively. In comparison, B. cereus grew 8.4x, 1.1x, 2.5x, and 3.9x times faster in the mouse eye than S. aureus, E. faecalis, S. epidermidis, and S. pneumoniae, respectively. One limitation to these studies was the use of 5000 CFU to initiate staphylococcal endophthalmitis in mice. Staphylococcal infections in mouse eyes are often cleared when inocula less than 1000 CFU/eye are used (Engelbert and Gilmore, 2005), so to ensure reproducible infections, 5000 CFU were used. Overall, these results demonstrate that B. cereus growth inside the eye is faster compared to most other Gram-positive bacteria which cause endophthalmitis, further illustrating the potential for more severe intraocular infections with B. cereus in humans.

Figure 4:

Comparative analysis of intraocular bacterial growth (A, B), quantification of myeloperoxidase (C), retinal function (D) and pathology (E) in Gram-positive experimental endophthalmitis in mice. 100 CFU/eye of B. cereus, E, faecalis, or S. pneumoniae, or 5000 CFU/eye of S. aureus or S. epidermidis were intravitreally injected into the posterior segment of the eyes of C57BL/6J mice. Infection parameters were measured at 12 hours postinfection. Figure 4A demonstrates the intraocular bacterial burden of these pathogens. The thick black bar indicates the initial bacterial inoculum at the time of infection. Figure 4B demonstrates intraocular bacterial growth rate in terms of hour⁻¹ over 12 hours of infection. Figure 4C demonstrates the level of myeloperoxidase in those infected eyes. Figure 4D demonstrates the retinal A-wave and B-wave retention of these pathogens. Figure 4E illustrates hematoxylin and eosin staining of whole eyes. Quantitative data represents the mean ± SEM of N≥5 eyes per time point. Statistical significance was determined by unpaired Student t-test (ns P ≥ 0.05, * P ≤ 0.0159, ** P ≤ 0.0079). Sections are representative of three eyes per group.

In Bacillus endophthalmitis, the primary infiltrating cells are polymorphonuclear leucocytes (PMN). These cells are present inside the infected eyes as early as 4 hours postinfection (Ramadan et al., 2006). Activation of TLRs during Bacillus endophthalmitis generates proinflammatory chemokines that recruit PMN into the posterior segment (Novosad et al., 2011; Parkunan et al., 2015, 2016). Infiltration of PMN in the eyes can be measured by quantifying the myeloperoxidase (MPO) enzyme. MPO is present in the granules of PMNs and contributes to microbicidal activity (Nauseef et al., 1983). Hence, PMN might cause damage to delicate retinal tissue by producing harmful biproducts in an effort to clear the infection. Figure 4C depicts that the levels of MPO in B. cereus infected eyes were significantly higher compared to that of S. aureus, E. faecalis, S. epidermidis, or S. pneumoniae infected eyes at 12 hours postinfection. These results suggest that inflammatory cell infiltration into infected eyes were faster in B. cereus- infected eyes than eyes infected with other ocular pathogens.

Increasing bacterial mass and secreted products inside the eye directly or indirectly affect retinal function and architecture (Astley et al., 2016). A comparison of retinal function outcomes in experimental endophthalmitis caused by different pathogens is depicted in Figure 4D. The level of retained A-wave, which is a function of retinal photoreceptors cells, was approximately 10% at 12 hours postinfection in eyes infected with B. cereus, suggesting that majority of photoreceptor cell function was lost in these eyes within 12 hours postinfection. The retained A-wave functions of eyes infected with S. aureus, E. faecalis, S. epidermidis, or S. pneumoniae were significantly greater compared to B. cereus at 12 hours postinfection, and were collectively at approximately 70-100%. At 12 hours postinfection, the retained function of B-waves, which arises from secondary retinal cells such as Muller cells and bipolar cells, was also significantly reduced in eyes infected with B. cereus compared to that of eyes infected with other pathogens. Loss of B-wave function in eyes infected with B. cereus was approximately 90%, compared to loss of approximately 0-40% for the other pathogens. These results demonstrate that compared to other Gram-positive endophthalmitis pathogens, retinal function declines rapidly in eyes infected with B. cereus, suggesting the potential for rapidly developing blindness in B. cereus endophthalmitis. It has been reported that in other forms of experimental endophthalmitis, retinal function is lost at a much slower rate compared to that of B. cereus (Callegan et al., 1999a; Engelbert and Gilmore, 2005; Sanders et al., 2011). In a comparative meta-analysis using previously published experimental animal data, we reported that almost complete loss of retinal function occurs after 24 hours for K. pneumoniae, 72 hours for S. aureus, 48 hours for E. faecalis, and 48 hours for S. pneumoniae. B. cereus reaches this critical level of retinal function loss in only 12 hours (Parkunan and Callegan, 2016; Ramadan et al., 2006). Figure 4E illustrates the histological changes in eyes infected with various pathogens at 12 hours postinfection. In eyes infected with B. cereus, retinas were completely detached, retinal layers were not distinguishable, and significant accumulation of inflammatory cells throughout the eye were noted. In stark contrast, eyes infected with S. aureus, E. faecalis, S. epidermidis, or S. pneumoniae at this same time point were significantly less inflamed. In these eyes, retinas were intact, retinal layers were clearly distinct, and inflammation was minimal. In our comparative meta-analyses of experimental endophthalmitis in different animal models, B. cereus endophthalmitis reached the peak of disease severity, bacterial growth, inflammatory influx, and retinal function loss within 12-24 hours postinfection (Astley et al., 2016; Parkunan and Callegan, 2016). Compared to B. cereus endophthalmitis, infections with S. epidermidis, S. aureus, S. pneumonia, E. faecalis, and K. pneumoniae were relatively slow to develop. Together, these studies highlight how critical immediate identification and therapeutic intervention is for this particular infection.

B. cereus endophthalmitis reaches the top of the list of rapidly blinding ocular diseases, but the level of understanding of the complexities in the host/pathogen relationship in this disease is still limited. Treating endophthalmitis is difficult due to the delicate anatomy and physiology of the interior of the eye. Currently, there remains no collectively agreed upon course of therapy for endophthalmitis which adequately prevents the loss of vision. Furthermore, the beneficial use of anti-inflammatory drugs are not convincing due to the wide variability in clinical and experimental outcomes (Aguilar et al., 1996; Bui and Carvounis, 2014; Ermis et al., 2005; Liu et al., 2000; Meredith et al., 1996; Pollack et al., 2004; Shah et al., 2000; Yildirim et al., 2002). Despite therapeutic intervention which would otherwise be beneficial in less severe cases of endophthalmitis, B. cereus-infected eyes can lose significant vision very rapidly, emphasizing the importance of improved therapeutics required to fight this disease.

4. Conclusions

With the increased use of intravitreal injections to combat intraocular inflammation and retinal degenerations, and the increasing number of invasive ocular surgeries for cataracts and glaucoma, the prevalence of endophthalmitis may increase. Compared to other etiologic forms of endophthalmitis, B. cereus endophthalmitis is a somewhat infrequent event, but has significant sight-threatening consequences. A uniquely rapid infection course, high therapeutic failure rate, and poor functional and anatomic outcomes are typical of B. cereus endophthalmitis. In this review, we discussed the current clinical information and research into the mechanisms underlying this blinding infection. Figure 5 depicts a model of B. cereus endophthalmitis, which illustrates the progression of this disease from infection initiation to retinal damage. So far, researchers have identified several virulence determinants in Bacillus endophthalmitis. However, the mechanisms by which these factors drive and/or impact inflammation and retinal tissue damage remain under investigation. Recent progress has provided a detailed depiction of the interactions between innate immunity and B. cereus in the eye and has identified targetable inflammatory mediators in the context of endophthalmitis. A more thorough understanding of the intraocular inflammatory signals driven by this pathogen and its products is critical for deriving better therapeutic strategies against this blinding infection.

Figure 5:

Model of B. cereus endophthalmitis. (A) Initiation of B. cereus infection in the eye after an ocular trauma or surgery. (B) Rapid bacterial growth in the vitreous and migration toward the retina. (C) Production bacterial toxins and enzymes. These bacterial products can cause dysfunction and permeability of the blood-retinal barrier. (D) Activation of TLR pathways by bacterial envelope constituents. (E) TLR pathway activation leads to the generation of proinflammatory mediators which recruit inflammatory cells (PMN) into the eye, beginning at the optic nerve head. These mediators may also disrupt the blood-retinal barrier. (F) Infiltrating PMNs may block light pathways to the retina and cause bystander damage to retinal cells, leading to vision loss. Vit: Vitreous, ON: Optic nerve.