Background
Endophthalmitis is a serious infection of the posterior segment of the eye which occurs from introduction of microbes following a surgical procedure (post-operative endophthalmitis [POE]), a traumatic penetrating injury (post-traumatic endophthalmitis [PTE]), or bloodstream spread from an infection of a distant site in the body (endogenous endophthalmitis [EE]) [
1‐
3]. Bacterial endophthalmitis is considered a medical emergency and often results in poor visual outcomes [
4]. Much of the intraocular damage in endophthalmitis is due, in part, to the host inflammatory response [
5‐
9]. Immediate and aggressive intervention to stop the progression of the disease is critical to salvaging vision. There is currently no universal therapeutic regimen which prevents the significant inflammation and vision loss associated with severe forms of endophthalmitis.
The Gram-positive pathogen
Bacillus cereus is a leading cause of PTE and EE. PTE infections due to
B. cereus progress rapidly and result in a fulminant endophthalmitis characterized by severe intraocular inflammation, eye pain, and loss of visual acuity within hours [
1‐
4]. Complete blindness can result in 1 or 2 days, and in nearly half of these infections, evisceration or enucleation is required to salvage healthy tissue in the orbit [
10]. The severity and rapid progression of this infection has been recapitulated in a mouse model [
1,
2,
6‐
9]. Infection of mouse eyes with as few as 100 colony-forming units (CFU) of
B. cereus results in significant inflammation and loss of visual function within hours, similar to that observed in human infections. Because inflammation in the eye causes damage to non-regenerative neural structures, it is important to identify host factors that lead to the events that contribute to this bystander damage.
Robust inflammation in response to intraocular bacterial infection is triggered by the early recognition of cellular components via a class of pattern recognition receptors called Toll-like receptors (TLRs) that are expressed on host cells [
11,
12]. Parkunan et al. recently published findings implicating the TLR4/TRIF/MYD88 axis in intraocular
B. cereus infections [
8].
B. cereus infected eyes of TLR4
−/− mice had significantly less polymorphonuclear leukocytes (PMN) influx and reduced concentrations of four inflammatory mediators relative to infected eyes of C57BL/6 J wild type mice. These parameters correlated with a significant retention of retinal function. These results suggested that the inflammatory cascade in
B. cereus endophthalmitis is initiated, in part, by TLR4 signaling through a potentially novel TLR4 ligand either expressed or induced by
B. cereus [
8].
The attenuated course of infection observed in TLR4
−/− mice implicated downstream mediators of the TLR4 pathway as important in the robust, early response in eyes infected with
B. cereus [
8]. In the current study, we sought to identify host TLR4-dependent factors upregulated in response to
B. cereus intraocular infection. Based on previous observations of a less severe inflammatory response in TLR4
−/− mice [
8], we hypothesized that the retinal gene expression profile would be significantly different between TLR4-deficient mice and C57BL/6 J mice following infection. Microarray analysis identified 15 genes involved in the acute inflammatory response, neutrophil recruitment, photoreceptor cell survival, and pathogen recognition and clearance that were upregulated 5-fold or greater in infected C57BL/6 J wild type mice compared to their levels in uninfected control mice (Table
1). The expression of 14 out of 15 of these genes was found to be unaltered in TLR4
−/− mice relative to uninfected controls, indicating their dependency on TLR4 (Table
1). These genes included key mediators in neutrophil recruitment, and activation of photoreceptor survival and pathogen clearance mechanisms in response to
B. cereus infection. These results further suggest that the TLR4 pathway might serve as a target for new anti-inflammatory treatments critically needed to not only control the explosive inflammation seen in
B. cereus ocular infection, but also ocular infections due to
Klebsiella pneumoniae and other Gram negative pathogens.
Table 1
Microarray analysis of retinal genes upregulated 5-fold and higher 4 h postinfection with B. cereus ATCC14579
CXCL1 | chemokine (C-X-C motif) ligand 1 | NM_008176 | 34 | 0.0106 | NC | NS |
CXCL2 | chemokine (C-X-C motif) ligand 2 | NM_009140 | 29 | 0.0225 | NC | NS |
IL-6 | interleukin 6 | NM_031168 | 25 | 0.0114 | NC | NS |
CXCL10 | chemokine (C-X-C motif) ligand 10 | NM_021274 | 21 | 0.0328 | NC | NS |
CCL2 | chemokine (C-C motif) ligand 2 | NM_011333 | 20 | 0.0355 | NC | NS |
CCL3 | chemokine (C-C motif) ligand 3 | NM_011337 | 16 | 0.0006 | 5 | 0.0446 |
PTX3 | pentraxin related gene | NM_008987 | 15 | 0.0376 | NC | NS |
ICAM1 | intercellular adhesion molecule 1 | NM_010493 | 11 | 0.0026 | NC | NS |
SOCS3 | suppressor of cytokine signaling 3 | NM_007707 | 10 | 0.0034 | NC | NS |
CYR61 | cysteine rich protein 61 | NM_010516 | 10 | 0.0116 | NC | NS |
MOBP | myelin-associated oligodendrocytic basic protein | NM_008614 | NC | NS | 10 | 0.0270 |
MBP | myelin basic protein | NM_001025245 | NC | NS | 10 | 0.0223 |
PTGS2 | prostaglandin-endoperoxide synthase 2 | NM_011198 | 9 | 0.0027 | NC | NS |
STEAP4 | STEAP family member 4 | NM_054098 | 6 | 0.0116 | NC | NS |
LIF | leukemia inhibitory factor | NM_008501 | 6 | 0.0055 | NC | NS |
CH25H | cholesterol 25-hydroxylase | NM_009890 | 6 | 0.0232 | NC | NS |
PLP1 | proteolipid protein (myelin) 1 | NM_011123 | NC | NS | 6 | 0.0236 |
EGR2 | early growth response 2 | NM_010118 | 5 | 0.0062 | NC | NS |
Discussion
B. cereus infection of the eye leads to a rapid and destructive inflammatory response that has devastating consequences for vision. TLR2 and TLR4 are key mediators of the innate immune response to bacterial pathogens during the early stages of endophthalmitis [
21,
22]. TLR2 plays an important role in both
B. cereus [
21] and
S. aureus [
22] endophthalmitis. TLR4 also plays a significant role in mediating inflammation in
B. cereus endophthalmitis [
8] and, as expected, in
Klebsiella pneumoniae endophthalmitis [
23]. Parkunan et al. [
8] demonstrated an increase in TLR4 mediated chemokines and inflammatory markers following
B. cereus intraocular infection. Levels of the chemokines CXCL1 (KC), TNF-α (tumor necrosis factor alpha), IL-6, IL-1β were significantly reduced in TLR4
−/− mice when compared to wild type mice infected with
B. cereus [
8]. These findings corroborate our microarray and qPCR results. Our previous studies assessed the expression of a limited set of proinflammatory chemokines and cytokines and used whole globes for analysis, which did not permit identification of the source of proinflammatory mediators. Using a transcriptomics approach, we identified 15 proinflammatory and immunomodulatory TLR4-dependent genes whose expression was increased 5-fold or greater specifically in the retinas of eyes 4 h after infection with
B. cereus. Since these genes were highly upregulated early during the course of infection, they potentially represent targetable host factors that mediate the initial response to
B. cereus ocular infection.
In the current study, CXCL1 (KC) was the most highly upregulated gene by microarray analysis in the retinas of eyes infected with
B. cereus. Quantitative PCR confirmed upregulation of CXCL1. Elevated CXCL1 expression was not surprising, considering its chemoattractant properties and the neutrophil burden observed in the posterior segment following infection [
9]. We also observed upregulation of CXCL2 (MIP-2α), CXCL10 (IP-10), and CCL2 (MCP1) chemokines in these retinas. CXCL2 is highly homologous and shares many of the same roles in acute inflammation as CXCL1, including interaction with the CXCR2 receptor, secretion by monocytes and macrophages
, and attraction of neutrophils to sites of infection and inflammation [
24‐
26]. CXCL10 is secreted by monocytes, endothelium, and fibroblasts after IFN-γ (interferon gamma) stimulation in response to viral infection, and after LPS stimulation in response to Gram-negative infection. CXCL10 serves as a chemoattractant that recruits monocytes/macrophages, T cells, NK cells, and dendritic cells [
27,
28]. Since CXCL10 production can occur as a result of stimulation by LPS through TLR4 [
27], we hypothesize that the observed upregulation of CXCL10 after
B. cereus infection may have resulted from the activation of TLR4 by novel ligand. Given that
B. cereus is a Gram-positive bacterium and does not produce LPS, CXCL10 upregulation might occur due to the activation of TLR4 by another ligand produced or elicited by
B. cereus. Rajamani et al. did not observe CXCL10 upregulation following intraocular infection with
S. aureus, however inflammation in this infection is primarily driven by TLR2, and not by TLR4 [
29].
IL-6 expression was upregulated 25-fold in C57BL6/J mice on the microarray analysis relative to uninfected control mice and was not upregulated in TLR4
−/− mice relative to uninfected controls. IL-6 is both a proinflammatory chemoattractant and a modulator of inflammation via signaling that increases the expression of TNF-α and IL-1β antagonists. Rajamani et al., demonstrated that IL-6 and IL-1β were both significantly upregulated during
S. aureus endophthalmitis and suggested that these genes were important for the response to
S. aureus infection [
29]. Parkunan reported that IL-6 levels were markedly increased at 8 and 12 h postinfection in a TLR4-dependent manner following intraocular infection with
B. cereus [
8]. In contrast to the findings of Parkunan et al., we did not detect increased levels of TNFα transcript in retinas at 4 h following infection. This suggests that the source of TNFα seen by Parkunan may not be from cells in the retina at this stage of infection, but rather from infiltrating neutrophils, given that cytokine assays were performed on homogenized whole globes [
8]. Neutrophils enter the eye as early as 4 h postinfection with
B. cereus, but do not infiltrate the retinal layers until approximately 8 h postinfection [
6]. IL-6 is a pro-inflammatory mediator expressed when bacterial recognition induces inflammation via TLR4 activation. While IL-6 has also been shown to be anti-inflammatory due to its ability to induce soluble TNFα and IL-1β receptor antagonist expression [
30], this is unlikely the case in our model, given that neither TNFα or IL-1β expression was altered in the retina at 4 h. The increase in IL-6 seen in our microarray analysis done at 4 h implicates IL-6’s inflammatory role as a chemoattractant, while not precluding its role as a signal to downregulate TNF-α and IL-1β expression at a later time point as a means to limit inflammation thus preserving susceptible cells in the retina.
IL-6 is expressed by several cell types in the eye, including RPE cells, ganglion cells, and resident microglia [
31,
32]. Others have reported that levels of IL-6 are markedly increased in the retina upon damage to the optic nerve [
33], while others have reported IL-6 protects mature retinal ganglion cells from pressure-induced death [
34]. In contrast, IL-6 was dispensible for an inflammatory response following
B. cereus infection of the eye. A similar course of infection, proinflammatory mediator profile, neutrophil infiltration, and architectural changes to the retinal layers were observed in both IL-6
−/− and C57BL/6 J eyes [
9]. Redundancy due to additional gp130-dependent cytokines, such as LIF [
19], which is expressed at significantly higher levels in our analysis, might explain why the inflammatory response in IL-6
−/− mouse eyes was not significantly dampened following
B. cereus infection. Additionally, the increased expression of IL-6 seen in this experiment might serve the dual role of helping to initiate the early inflammatory response seen in
B. cereus endophthalmitis as well as serving to protect sensitive neuroretinal cells by limiting further inflammation and preventing apoptosis, as has been shown in glaucoma models [
34].
CCL2 recruits monocytes, basophils, memory T-cells, and dendritic cells, but not neutrophils or eosinophils [
35,
36], and CCL3 activates neutrophils [
24]. CCL2 and CCL3 are critical to the recruitment of neutrophils in the context of keratitis [
37]. In a mouse model of
Pseudomonas aeruginosa-induced corneal infection, antibodies directed against CCL2 or CCL3 significantly reduced neutrophil infiltration into the cornea and decreased corneal damage. Both CCL2 and CCL3 were upregulated in the context of TLR4-mediated inflammation after
B. cereus ocular infection, and their blockade might reduce neutrophil infiltration into the vitreous and decrease damage to the retina.
While the mechanisms of neutrophil infiltration into the eye during endophthalmitis are not completely understood,
S. aureus is capable of inducing expression of E-selectin and ICAM1 on macrovascular endothelial cells in a rat model of endophthalmitis [
38]. In the current study, we demonstrated that
B. cereus infection also induces ICAM1 expression in the retina. Expression of ICAM1 is upregulated by a variety of stimuli including retinoic acid, oxidant stress, and the proinflammatory cytokines IL-1β, TNFα, and IFNγ [
39]. Lipopolysaccharide (LPS) was shown to induce expression of ICAM1 in human pulmonary alveolar epithelial cells through the TLR4/c-Src/NADPH oxidase/ROS-dependent NF-κB pathway [
40]. ICAM1 serves as the ligand for LFA-1 (Lymphocyte function-associated antigen 1integrin) on leukocytes [
14,
15]. Leukocytes that bind to endothelium via ICAM1 / LFA-1 complex are able to initiate transmigration across the endothelial membrane [
41]. The finding that ICAM1 is upregulated in wild type, but not in TLR4
−/− eyes, correlates with our previous findings that neutrophil recruitment is decreased in TLR4
−/− eyes following infection [
8]. Interestingly, ICAM1 possesses signal-transducing functions that are associated primarily with proinflammatory pathways. Ligation of ICAM1 on the surface of endothelial cells elicits a signaling cascade resulting in production of IL-8 and RANTES (regulated on activation, normal T cell expressed and secreted) [
42], as well as additional ICAM1 in a positive feedback loop [
43]. Upregulation of ICAM1 early during infection and prior to PMN infiltration might indicate an additional role in signaling at this stage in addition to leukocyte trafficking.
Another gene product implicated in leukocyte migration, cysteine-rich protein 61 (CYR61), was also significantly upregulated following
B. cereus infection. CYR61 is a modular protein that functions as a bridge between cells and the extracellular matrix, binding to integrins and to extracellular matrix proteins [
16]. CYR61 is primarily involved in regulating adhesion and chemotaxis, as well as angiogenesis [
16]. In contrast to the rapid and significant upregulation of CYR61 following
B. cereus infection, Rajamani and colleagues did not observe upregulation of CYR61 until 12 h after intraocular infection with
S. aureus [
29]. This finding correlates with the delay in neutrophil influx observed in
S. aureus endophthalmitis as compared to
B. cereus endophthalmitis. It is likely that CYR61 is upregulated in order to mediate neutrophil invasion following
B. cereus infection of the eye. PTX3 was also upregulated following
B. cereus infection, and is known to be produced in response to TLR engagement [
17]. PTX3 activates the classical complement pathway via C1q [
18]. Therefore it could be hypothesized that PTX3 might function to activate an anti-bacterial response to
B. cereus intraocular infection. However, it is currently unknown whether complement plays a role in
B. cereus endophthalmitis. While complement is present in the eye [
44], its absence did not alter the outcome in a mouse model of
S. aureus endophthalmitis [
45].
SOCS3 is a negative regulator of cytokine signaling induced by IL-6, IL-10, and INFγ (mediators of both the MYD88-dependent and MYD88-independent TLR pathways). SOCS3 functions to inhibit STAT3 phosphorylation and this negative regulatory function prevents excessive activation of proinflammatory genes [
19]. The rapid and destructive inflammatory response observed following
B. cereus infection suggests that SOCS3 may not adequately inhibit STAT3 phosphorylation in the cells in the retina which function as the initial responders. Wang et al. demonstrated that prolonged STAT3 activation occurs as a result of the IL-6 receptor associating with the epidermal growth factor receptor [
19]. This complex is capable of STAT3 activation but is not inhibited by SOCS3.
The pathogen Mycobacterium tuberculosis directly activates SOCS3 and therefore inhibits NF-κB/rel-mediated proinflammatory cytokine production [
46], which functions to suppress the inflammatory response. It is currently unknown as to whether
B. cereus is capable of directly activating SOCS3 in the retina, however the robust inflammatory response incited by
B. cereus might override the inhibitory effects of SOCS3 activation.
LIF upregulation in response to
B. cereus infection might serve to enhance the protection of the delicate, nonregenerative photoreceptors. LIF is an IL-6 regulated neurocytokine that is upregulated in Muller cells in response to retinal stress [
47,
48]. Chucair-Elliott et al. demonstrated that LIF downregulates the expression of RPE65, which ultimately leads to a decrease in 11-cis-retinal, a chromophore that might be toxic in excessive amounts [
49].
B. cereus infection may be a stressor that results in upregulation of LIF in order to protect this layer of cells in the retina. While it is unknown which cells are expressing LIF at increased levels following infection, given that Muller cells upregulate LIF in response to stress [
47,
48], Muller cells might serve as at least one of the cell types that produce LIF in response to
B. cereus infection.
Induction of PTGS2/COX-2 during infection is mediated by TLR4 and NFκB (nuclear factor kappa-light-chain-enhancer of activated B cells) [
50]. This enzyme converts arachidonic acid to prostaglandin endoperoxide H2 (PGE
2) and is expressed during inflammation. PGE
2 has an immunomodulatory effect and serves to prevent the activation of neutrophils, which is an immune evasion strategy utilized by some bacterial pathogens.
Streptococcus pneumoniae induces PGE
2 production by human neutrophils and prevents activation [
51]. In a model of
Pseudomonas pneumonia, lack of PTGS2/COX-2 was proven to be beneficial and resulted in increased clearance of bacteria from the lungs [
52]. The mechanism for this was linked to PGE
2 inhibiting superoxide production by immune effectors and therefore hindering bacterial killing. PTGS2/COX-2 is usually an inflammation inducible enzyme not normally expressed in most tissues. However, PTGS2/COX-2 is constitutively expressed throughout both murine and human eyes [
53], in the cornea, iris, ciliary body, and retina. Wang et al. suggested that PTGS2/COX-2 might play a protective role in eye tumorigenesis. Significant upregulation of PTGS2/COX-2 and PGE
2 synthesis in the retina following
B. cereus infection might serve to modulate the function of invading neutrophils and prevent activation.
The fact that 15 genes associated with TLR4 activation were upregulated 5-fold or greater in C57BL6/J mice compared to uninfected control eyes, but were not induced in TLR4
−/− suggests that
B. cereus is capable of activation of TLR4. This implicates a component of
B. cereus as a novel ligand for activating the TLR4 receptor. TLR4 typically mediates the inflammatory response to LPS in conjunction with MD2 (lymphocyte antigen 96), CD14, and MYD88 [
54]. However, TLR4 has been shown to recognize additional exogenous and endogenous ligands, including respiratory syncytial virus, heat-shock proteins, fibronectin, fibrinogen, and hyaluronic acid [
55‐
61]. Alternatively,
B. cereus might instigate a response that results in the formation of an endogenous ligand for the TLR4 pathway. In the current study, we did not observe a transcriptional upregulation of any of the reported endogenous ligands. However, we cannot rule out the possibility that
B. cereus infection might result in the posttranscriptional production or modification of an endogenous ligand.
The ability of
B. cereus to activate both the TLR4 [
8] and TLR2 [
21] pathways might explain why
B. cereus endophthalmitis results in an explosive inflammatory response that results in poor visual outcomes in affected patients. The early onset of TLR4-associated inflammation could play a key role in therapies designed to prevent further inflammation and damage to the sensitive and nonregenerative structures of the eye. This study identified retinal genes that were significantly upregulated early during
B. cereus infection that might prove tractable as targets for intervention. However, inherent redundancies in these pathways and the potential for exacerbating inflammation might complicate the design of new therapeutics. Our study also presented key differences between the inflammatory mediators that are elicited by
B. cereus and those by
S. aureus [
29] following intraocular infection. Endophthalmitis severity and outcome as result of infection with these two pathogens is starkly different, and the results of this study shed light on the differences in types and timing of inflammatory mediator production that contribute to the distinctive courses and outcomes. Given the limitations inherent to microarray analysis for assessing global gene changes, future studies confirming these results by proteomics and the analysis of the severity of
B. cereus infections in mice specifically deficient in these pathways will be required. Future studies will evaluate the retinal and global ocular inflammatory responses over the course of
B. cereus endophthalmitis to identify pathway-based anti-inflammatory targets, specifically, those that are regulated by TLR4, and to identify the
B. cereus produced or induced ligand for TLR4.