Diabetes Exacerbates Pseudomonas aeruginosa Keratitis in Streptozotocin-Induced and db/db Mice via Altering Programmed Cell Death Pathways

Abstract

Purpose: Patients with diabetes have a higher incidence of infections, which are often more severe. This study aimed to investigate the impact of hyperglycemia on bacterial keratitis caused by Pseudomonas aeruginosa (Pa) in two mouse models of diabetes, streptozotocin-induced type 1 diabetes mellitus (T1DM) and db/db type 2 diabetes mellitus.

Methods: The susceptibility of corneas to Pa was assessed by determining the inocula required to cause infectious keratitis. Dead or dying cells were identified using TUNEL staining or immunohistochemistry. Specific inhibitors were used to evaluate the role of cell death modulators in Pa keratitis. Cytokines and Treml4 expressions were analyzed using quantitative PCR, and the role of Treml4 in keratitis was determined using small interfering RNA technology.

Results: DM corneas required significantly fewer inocula to develop Pa keratitis, with T1DM corneas requiring 750 inocula and type 2 diabetes mellitus corneas requiring 2000 inocula, compared with 10,000 inocula required for normal (NL) mice. T1DM corneas had more TUNEL-positive and fewer F4/80-positive cells than NL corneas. Phospho-caspase 8 (apoptosis) and -RIPK3 (necroptosis) staining was more intense in the epithelial and stromal layers of NL and T1DM corneas, respectively. Pa keratitis was augmented by targeting caspase-8 and prevented by RIPK3 inhibition in both NL and T1DM mice. Hyperglycemia suppressed IL-17A/F and augmented IL-17C, IL-1β, IL-1Ra, and TREML4, the downregulation of which protected T1DM corneas from Pa infection by suppressing necroptosis. RIPK3 inhibition blocked Pa infection in db/+ mice and significantly decreased the severity of keratitis in db/db mice.

Conclusions: Hyperglycemia exacerbates bacterial keratitis in B6 mice by skewing apoptosis toward necroptosis. Preventing or reversing this transition may serve as an adjunct therapy for treating microbial keratitis in patients with diabetes.

Approximately 26 million people in the United States suffer from diabetes mellitus (DM).¹ These patients are three times more likely to develop sepsis and suffer far worse outcomes.² They have a higher incidence of foot, urinary tract, and surgical site infections.^3,4 Moreover, multidrug-resistant infections are also higher in patients with DM.⁵ The cornea is susceptible to DM complications, including compromised barrier function, delayed wound healing, and neuropathy, predisposing patients with DM to bacterial keratitis (BK).⁶ A study conducted in England found that people with type 2 DM (T2DM) had a 26% increase in eye infections compared with age-matched controls, and similarly, people with type 1 DM (T1DM) had a 36% increase in eye infections compared with age-matched controls.⁷ Another retrospective large-scale cohort study of 238,701 patients with DM and the same number of age- and sex-matched patients without DM revealed that the incidence rate for corneal ulcers, caused mainly by infectious keratitis, is 1.27 times higher in patients with DM than in the controls.⁸ These large-scale population studies indicate an increased susceptibility and severity of BK in patients with DM. 

Pseudomonas aeruginosa (Pa) is an opportunistic pathogen capable of infecting barrier function-compromised corneas.⁹ The invading pathogens are recognized by pathogen recognition receptors, such as Toll-like receptors (TLR), which recognize pathogen-associated molecular patterns.¹⁰ TLRs activate innate immune responses, including the expression of inflammatory cytokines such as IL-1β, IL-17, and antimicrobial peptides such as β-defensins and CRAMP.^11–13 The expressed cytokines recruit innate immune cells, such as neutrophils and macrophages, which participate in the clearance of the invading pathogen. However, if an infection is not controlled sufficiently under pathogenic conditions, such as diabetes and dry eye diseases, an overwhelming host inflammatory response and pathogen-virulent factors can cause tissue destruction and ulcerations.^14,15 For example, neutrophils are crucial for bacterial clearance, but persistent and/or elevated neutrophil recruitment was observed in injured diabetic corneas, resulting in delayed wound healing and severe sensory nerve degeneration.^16,17 To investigate the biological processes and molecules associated with delayed wound healing in diabetic conditions, we conducted mRNA sequencing of mouse corneas during Pa infection, comparing normoglycemic and hyperglycemic mice. The raw sequence data have been deposited in the National Center for Biotechnology Information's Gene Expression Omnibus as GSE154507. Our Gene Ontology enrichment analysis revealed that negative regulation and positive regulation of TLR signaling were uniquely associated with Pa-infected NL and DM corneas, respectively (unpublished data from Gao and Yu, accessed February 3, 2020). We found that, among the many genes involved in these biological processes, hyperglycemia augmented the expression of the TLR-positive regulator Triggering Receptor Expressed On Myeloid Cells Like 4 (TREML4),^18–20 while downregulating the negative regulator ACOD1 (IRG1)21 in diabetic corneas compared with normoglycemic corneas²¹ and was downregulated in DM compared with NL corneas. These results suggested that targeting specific modulators of TLR signaling pathways may alter the outcomes of BK. 

The activation of TLRs is also known to regulate programmed cell death (PCD), including apoptosis and necroptosis.²² Apoptosis is a caspase-dependent, nonlytic, and usually an immunologically silent form of cell death.²³ Apoptosis is a crucial host defense mechanism and an effective means to eliminate pathogens.²⁴ In contrast, necroptosis is a programmed lytic cell death carried out by the RIPK1–RIPK3–MLKL pathway.²⁵ Inhibition of Casp8, a key modulator of extrinsic apoptosis, results in the phosphorylation and release of RIPK1.²⁶ Phosphorylated RIPK1 then phosphorylates RIPK3, which in turn phosphorylates MLKL. Phosphorylated MLKL forms oligomer pores in the plasma membrane, releasing intracellular components such as high mobility group box 1, IL-1α, and IL33 that act as alarmins, further activating inflammatory receptors, thereby initiating a detrimental inflammatory response.^25,27 Hyperglycemia enhanced necroptosis, but inhibited extrinsic apoptosis in vivo.²⁸ Hence, we propose that hyperglycemia exacerbates microbial keratitis by promoting necroptosis, leading to hyperinflammation and corneal destruction in the corneas. 

In this study, we aimed to establish Pa keratitis models in T1DM and T2 DM mice. We assessed the minimal number of bacteria needed to cause corneal infection and progression of Pa keratitis in DM, compared with nondiabetic (NL) corneas. Our results showed more necroptotic cells in Pa-infected DM corneas. Our findings suggest that hyperglycemia shifts PCD from apoptosis to necroptosis in Pa-infected corneas and that necroptosis contributes to increased susceptibility to and severity of Pa infection in DM mice. 

Methods

Animals and Induction of Diabetes

All animal studies were approved by the Wayne State University Institutional Animal Care and Use Committee. All investigations conformed to the regulations of the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and the National Institutes of Health. Six-week-old male wild-type B6 and 10-weeks-old male and female db/db, with db/+ as the control mice, were purchased from The Jackson Laboratory. Wild-type B6 mice were induced to develop diabetes following the Diabetic Complications Consortium's Low-Dose Streptozotocin Induction Protocol (mouse, 50 mg streptozotocin [STZ]/kg, daily for 5 days) without fasting before STZ injections. The control animals were injected with 0.1 M citrate buffer (pH 4.5). Mice were housed in a non–pathogen-free WSU animal facility managed by DLAR technicians and checked for ocular surface abnormality daily starting on day –3. Mice, 10 weeks after initial STZ injections or 10-week-old bd/bd, were used for Pa infection. 

Infection Procedure

For STZ T1DM mice, each group consisted of five mice. Both eyes were infected, and corneas from the same mouse were pooled (Fig. 1). For the treated groups, five mice were used; left eyes were treated with vehicle or control small interfering RNA (siRNA), and right eyes were treated with test reagents (Fig. 3) or Treml4 siRNA (Fig. 6). T2DM db/db and the control db/+ mice, three male and three female mice were used, and both corneas were infected, scored, and pooled for sample collection. For inhibitor treatment, five male db/db and db/+ mice were used; left eyes were treated with vehicle, and right eyes were treated with a test reagent. For bacterial inoculation, mice were anesthetized with ketamine and xylazine and placed beneath a stereoscopic microscope at a magnification of 4×. The corneas were scratched gently with three 1-mm incisions using a 26G needle. A 5 µL of suspension containing 10,000 Pa ATCC19660 was applied to the scarified cornea. The anesthetized mice were kept in a fixed position with the instilled solution remaining on the ocular surface until they awakened (usually >20 minutes). The infected corneas were photographed. Bacterial load and myeloperoxidase (MPO) assay were determined. For experiments presented in Figures 5 and 8, the controls were uninfected corneas from NL and DM mice. 

Figure 1.

 

Susceptibility and severity of Pa keratitis in NL and T1DM B6 mice. The corneas of STZ-T1DM B6 (duration 10 weeks) and age-matched NL mice were scarified and inoculated with indicated numbers of Pa (CFU) at 0 h. The infected corneas were photographed (A) and assigned clinical scores (A’) at 1 and 3 dpi. At 3 dpi, the corneas were excised and subjected to bacterial counting (B) presented as CFU of Pa per cornea and to MPO determination (units/cornea) (C). The results are representative of three independent experiments (n = 5 each), and P values were generated using the nonparametric Mann–Whitney U test for clinical scores and two-way ANOVA for CFU and MPO. *P < 0.05; ** P < 0.01. The lowercase letters in (A’–C) referred to the corneas presented in (A).

Subconjunctival Injection of siRNAs and Small Molecular Inhibitors

Subconjunctival injection is a routine procedure used in the ophthalmology clinic to treat ocular diseases because it allows injected materials to diffuse into the cornea slowly. The subconjunctival injection volume for mice was 5 µL per injection. Anesthetized mice were injected with ON-TARGETplus mouse Treml4, and ON-TARGETplus nontargeting pool (the control) siRNAs were purchased from Dharmacon (Pittsburgh, PA, USA); 20 µmol/L solutions of Treml4 and the control siRNA were injected twice (24 and 6 hours) before Pa inoculation on left eyes with the controls into right eyes. For small molecular inhibitors, 5 µL at the indicated concentrations were injected –4 h before Pa inoculation; 0.1% DMSO in PBS was injected into the contralateral eyes as the vehicle control. 

Clinical Examination

Eyes were examined daily to monitor disease progression with a dissection microscope equipped with a digital camera. For clinical scores, mice were color coded and analyzed in a blinded fashion by two independent observers at 1 and 3 days postinfection (dpi) to visually grade the disease severity. The ocular disease was graded in clinical scores ranging from 0 to 12: A grade of 0 to 4 was assigned to each of the following three criteria—the area of opacity, density of opacity, and surface irregularity—resulting in a possible total score of 12. At 1 and 3 dpi, all infected corneas were photographed with a dissection microscope to illustrate the disease progression.²⁹ 

Bacteria Load and MPO Determination

We used our previously modified methods that allowed bacteria load and MPO to be determined with a single mouse cornea.^30,31 the corneas were excised from the enucleated eyes, minced, and homogenized in 100 µL of PBS with a TissueLyser (Retsch, Haan, Germany). The homogenates were divided into two parts. The first part was subjected to the counting of bacterial colonies. Aliquots (100 µL) of serial dilutions were plated onto Pseudomonas isolation agar (Sigma/Millipore, Burlington, MA, USA; 17208) plates in triplicate. The plates were incubated overnight at 37°C, and bacteria colonies were counted.³² 

The second part was used to MPO activity. The homogenates was mixed with 5 µL of 1% SDS and 10% Triton-X 100; 30 µL of homogenates were mixed with 270 µL of hexadecyltrimethylammonium bromide buffer. The samples were then subjected to three freeze-thaw cycles, followed by centrifugation at 16,000×g for 20 minutes. Twenty microliters of the supernatant was mixed with 180 µL of PBS (pH 6.0) containing 16.7 mg/ml O,O-dianisidine hydrochloride, and 0.0005% hydrogen peroxide at a 1:30 ratio in a 96-well plate. The change in absorbance at 460 nm was monitored continuously for 5 minutes in a Synergy2 Microplate reader (BioTek, Winooski, VT, USA). One unit of MPO activity corresponded with approximately 2.0 × 10⁵ polymorphonuclear leucocytes. 

RNA Extraction and Real-Time PCR

RNA was extracted from the collected corneas using rNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. cDNA was generated with an oligo(dT) primer (Invitrogen, Life Technologies, Grand Island, NY, USA) followed by analysis using real-time PCR with the Power SYBR Green PCR Master Mix (AB Applied Biosystems, University Park, IL, USA). The expression of β-actin was used as an internal control to normalize the amount of RNA used. The primer pairs are listed in Table 1.³¹ 

Table 1.

 

Primer Sequences Used for PCR

 

Immunohistochemistry and TUNEL Staining

Mouse corneas were embedded in the Tissue-Tek OCT compound and frozen in liquid nitrogen. Six-micrometer-thick sections were cut and mounted to polylysine-coated glass slides. After a 10-minute fixation in 4% paraformaldehyde, slides were blocked with 1PBS containing 2% BSA and Fc block for 1 hour at room temperature. Sections were then incubated with first and secondary antibodies. Slides were mounted with DAPI mounting media. Controls were treated similarly, but the primary antibody was replaced with rat IgG. To detect dead and dying cells, corneal cryostat sections were stained using a fluorescein in situ apoptosis-detection kit (TUNEL staining). The stained slides were examined with a fluorescence microscope, as described for the immunohistochemical studies.³³ TUNEL-stained corneal sections were also costained with different antibodies. 

Statistical Analyses

The statistical analyses were performed using GraphPad Prism 6 software (La Jolla, CA, USA). Data were presented as means ± SD. Experiments with two treatments and/or conditions were analyzed for statistical significance using a two-tailed Student t-test. Experiments with more than two conditions were analyzed using one-way ANOVA. A Bonferroni post-test was performed to determine statistically significant differences. In addition, a nonparametric Mann–Whitney U test was also performed for clinical scores. Mean differences were considered significant at a P value of less than 0.05. To ensure reproducibility, all experiments were repeated at least twice, and the results were consistent. To increase the scientific rigor of the study, in each experiment, NL and DM corneas with or without infection were used as the controls, giving multiple replications of the effects of DM on the susceptibility and severity of keratitis (photographs, clinical score, bacterial burden [CFU], and MPO activities). 

Results

Susceptibility and Severity of Pa Keratitis in NL and T1DM B6 Mice

Six-week-old female B6 mice were induced to develop diabetes using the low-dose STZ induction protocol. Average blood sugar levels reached 300 dg/mL by week 2 and were 350 to 600 dg/mL by weeks 8 to 10 after completion of the STZ regimen; control mice had approximately 100 dg/mL. Our previous studies showed that at this time or thereafter, diabetic corneal complications, including delayed wound healing, neuropathy, and dry eye, were observed.^34,35 In our Pa infection model, we used 10⁴ CFU Pa to infect needle-scratched B6 mouse corneas and observed that the infected corneas perforate at 5 dpi. 

Figure 1 shows the pathologies of Pa keratitis presented as micrographs and clinical scoring (CS with mild (CS 3.40 ± 0.45) and moderate (CS 8.0 ± 0.7) keratitis at 1 and 3 dpi of NL mice (Figs. 1Aa and 1Ag), respectively. The same amount of innocula caused severe keratitis: central corneal melting and a ring of opacification (CS 7.20 ± 0.83) and corneal perforation (CS 12) in T1DM mice at 1 and 3 dpi, respectively. Inocula at 2000 CFU Pa exhibited no infection of NL mice but caused moderate keratitis (CS 3.00 ± 0.71) at 1 dpi and close to perforation (CS 12) in DM corneas at 3 dpi (Figs. 1d, 1j). We then decreased the inoculate to 1000 CFU, which caused noticeable keratitis (1.60 ± 0.55) at 1 dpi and severe keratitis (CS 10.60 ± 0.54) at 3 dpi (Figs. 1e, 1k). Pa at 750 CFU exhibited almost signs of infection (CS 1.20 ± 0.45) at 1dpi (Fig. 1f), but the pathology in these corneas was similar to NL-infected corneas (CS 9.00 ± 0.94) at 3 dpi (Figs. 1l vs. 1g). The bacterial burdens (Fig. 1B) and MPO activities (Fig. 1C) per cornea were matched with the average clinical scores (Table 2). Our findings indicate that, like in patients with DM, mice with T1DM are highly susceptible to and have much worse outcomes from Pa infection. 

Table 2.

 

Characterization of Pa-infected NL and DM Corneas Correlated With the Micrograph Shown in Figure 1

 

Hyperglycemia Exacerbated Cell Death Detected by TUNEL Staining in Pa-infected Corneas

PCD pathways are an integrated part of immunity against microbial infection. DM has been known to adversely affect PCD, leading to the pathogenesis of diabetic complications, including diabetic nephropathy, cardiomyopathy, skin wound healing, and periodontitis.^36–38 Having shown high susceptibility of Pa keratitis in mice with T1DM, we next investigated cell death using TUNEL staining, which is commonly used to study PCDs, including apoptosis, necrosis, and autolytic cell death.³⁹ Figure 2 shows a lesion site where the epithelium was lost with massive infiltrated cells detected by DAPI in both NL and DM corneas. TUNEL staining was observed in the NL cornea, with most TUNEL-positive cells located apically (Figs. 2A and C). There were significantly more TUNEL-positive cells and fewer macrophages in DM compared with NL corneas. Many F4/80-positive cells were present in the entire stroma, some TUNEL positive (marked by arrows), whereas many more TUNEL-positive cells/debris were away from macrophages known to remove dead cell corpses by efferocytosis in DM corneas (arrowheads in Figs. 2B and D). 

Figure 2.

 

TUNEL and macrophage staining of NL and T1DM corneas at 1 dpi at a lesion site. STZ-T1DM (duration 10 weeks) and age-matched NL mice were infected with 2000 and 10,000 CFU Pa, respectively. At 1 dpi, the infected NL (A, C) and DM (B, D) corneas were excised, processed for cryostat section, and immunostained with TUNEL staining for dead cells (green) and F4/80 for macrophages (red). The TUNEL and F4/80 images were merged and superimposed with DAPI staining for nuclei. Arrows, macrophages costained with TUNEL; arrowheads, TUNEL-positive cells or cell debris that are not near macrophages. (B and D) High magnification of (A and B), respectively.

Hyperglycemia Skews Infection-induced PCD From Apoptosis to Necroptosis in B6 Mouse Corneas

Hyperglycemia has been shown to potentiate a shift from apoptosis to RIPK3-dependent necroptosis.⁴⁰ To determine which cell types are undergoing PCD in Pa-infected NL and DM corneas, we investigated cellular distributions of cleaved Casp8 (c-Casp8) (Fig. 3) and phosphorylated RIPK3 (p-RIPK3) (Fig. 4) using immunohistochemistry (IHC). Figure 3 shows a region next to a lesion site where individual immune cells can be visualized. In NL corneas, intense c-Casp8 staining was observed in most epithelial cells and some in the stroma, there were cells neither polymorphonuclear leucocytes nor macrophages. In contrast, the c-Casp8 staining intensity was much weaker in DM than in NL corneas with fewer c-Casp8–positive epithelial cells and not in the stroma. 

Figure 3.

 

Immunohistochemical detection of c-Casp8–positive innate immune cells in NL and DM corneas at 1 dpi. The cryostat sections of Pa-infected NL and DM corneas at 1 dpi used in Figure 2 were stained with c-Casp8 antibody and costained with F4/80 (macrophages) or NIMP-R14 (polymorphonuclear leucocytes). Arrows, representative of neutrophils or macrophages that were also c-Casp8 positive, indicative of apoptosis. E, epithelial layer; S, stroma. Scale bar, 100 µm.

Figure 4.

 

Immunohistochemical detection of p-RIPK3–positive innate immune cells in NL and DM corneas at 1 dpi. The cryostat sections of Pa-infected NL and DM corneas at 1 dpi used in Figure 2 were stained with antibodies against phospho-RIPK3 and co-stained with F4/80 (macrophages) or NIMP-R14 (polymorphonuclear leucocytes). Arrows, representative of neutrophils or macrophages that were also p-RIPK3 positive, indicative of necroptosis. E, epithelial layer; S, stroma. Scale bar, 100 µm.

Figure 4 shows an opposing pattern of RIPK3 staining to c-Casp8 stain shown in Figure 3. Fewer RIPK3 cells are in the NL corneas, whereas strong and markedly more RIPK3 cells are in the epithelium and stroma of DM corneas. In addition, many RIPK3 cells were polymorphonuclear leucocytes, but not macrophages. Our findings suggest that there are more necroptotic and fewer apoptotic cells in DM compared with NL Pa-infected B6 mouse corneas. 

cCasp8 and pRIPK3 were also co-localized staining to CD45 for myeloid cells. There were more CD45/c-Casp8–positive cells in NL and more CD45/pRIPK3–positive cells were in DM corneas (Fig. 2). 

Downregulation of Necroptosis and Apoptosis Had Opposing Effects on the Outcome of Pa Keratitis

Having shown that hyperglycemia may promote necroptosis and suppress apoptosis, we investigated the contributions of these PCD pathways to corneal response to infection using c-Casp8 and RIPK3 specific inhibitors (Fig. 5). At 1 dpi, inhibition of Casps8 led to more severe keratitis with significantly higher clinical score (2.5 vs. 5.25 in NL and 2.25 vs. 5.5 in DM corneas), bacterial burden (2.06 × 10⁵ vs. 5.06 × 10⁵ CFU in NL and 1.83 × 10⁵ vs. 5.87 × 10⁵ CFU in DM corneas), and MPO levels (11.80 vs. 26.30 units in NL and 11.28 vs. 28.04 units in DM corneas), compared with that of the control corneas. In contrast, inhibition of RIPK3 protected the corneal from infection with no sign of infection and inflammation (CS 0), no recoverable bacteria, and a significantly reduced polymorphonuclear leucocytes infiltration detected by MPO activity in both NL and DM corneas (Fig. 5). 

Figure 5.

 

Effects of Casp-8 or RIPK3 inhibition on Pa keratitis in NL and T1DM corneas. STZ-T1DM (duration 10 weeks) and age-matched NL mice were inoculated with 2000 and 10,000 CFU Pa, respectively. At 1 dpi, the corneas were photographed (A) and clinically scored (B). The corneas were then excised and subjected to CFU counting (CFU/corneas) (C) and MPO determination (units/cornea) (D). The results represent two independent experiments (n = 5 each), and P values were generated using the nonparametric Mann–Whitney U test for clinical scores and one-way ANOVA for CFU and MPO. **P < 0.01. The lowercase letters in (B–D) refer to the corneas presented in (A).

Genes Differentially Expressed in NL vs. DM Corneas During Pa Infection

Using real-time PCR, we investigated the expressions of innate and adaptive immunity related genes, including IL-6, −12, −22, −23, and −36 (α and γ), cytokines that were influenced by hyperglycemia (Supplementary Fig. S1), in Pa-infected NL and T1DM corneas with similar severity (clinical scores and bacterial burden) at 6 hours postinfection and 1 and 3 dpi. Figure 6 shows the expressions of the IL-1 and IL-17 families of cytokines that we previously reported to play a critical role in the pathogenesis of microbial keratitis and/or diabetic corneal complications.^{11,12,41–43} We observed that hyperglycemia downregulated IL-17C expression and upregulated IL-17F in uninfected corneas, whereas the levels of IL-1β and IL-Ra remained unchanged. Upon infection, all three IL-17 isoforms were upregulated, and hyperglycemia suppressed IL-17A and -17F but augmented IL-17C expression at 6 hours postinfection and 1 dpi. At 3 dpi, hyperglycemia increased IL-17A and IL-17C expression. IL-1β and its antagonist IL-1Ra were upregulated in both NL and DM corneas at the early stage of infection. The infection-induced upregulation IL-1Ra, but not IL-1β, was dampened in DM compared with NL corneas. The expression of TREML4, a key regulator of inflammation and immune cell death in sepsis,¹⁹ was slightly upregulated at 6 hours postinfection, and significantly upregulated at 1 and 3 dpi. Hyperglycemia further augmented the infection-induced upregulation of TREML4 (Fig. 6). 

Figure 6.

 

Effects of hyperglycemia on the expressions of the IL-1, IL-17 families of cytokines and TremL4 in B6 mouse corneas during Pa infection. STZ-T1DM (duration 10 weeks) mice were inoculated with 10,000 (DM 6pi), 2000 (DM 1dpi), or 750 (DM750 3dpi) CFU of Pa while age-matched NL mice inoculated with 10,000 CFU. Infected corneas were collected at the indicated time with noninfected cornea as the controls (0 h, NL, and DM) and were subjected to real-time PCR analysis. The results are presented as the increase (fold) over the value for noninfected NL and DM corneas (set at value 1) after normalization to the level of β-actin as the internal control. P values were generated by a two-tailed Student's t-test to compare NL with DM corneas at the indicated time; * P < 0.05 and ** P < 0.01. Data are representative of three independent experiments with three mice per group.

Modulation of PCD Through Downregulation of TREML4 Promotes Innate Immune Defense Against Pa Infection

Recently, it was reported that mice lacking TREML4 were protected from lethal sepsis and secondary pneumonia caused by Pa infection.¹⁸ To investigate the role of TREML4 in Pa keratitis, its expression in DM corneas was downregulated using siRNA technology. Figure 7 shows that Treml4 siRNA protected DM B6 mouse corneas from Pa infection with no sign of inflammation or opacification. Furthermore, the downregulation of Treml4 by siRNA markedly decreased the number of p-MLKL–positive cells in epithelial layers and the stroma, indicating a role for Treml4 in necroptosis. 

Figure 7.

 

TREML4 affecting the outcome of Pa keratitis by promoting necroptosis in DM corneas. STZ-T1DM (duration 10 weeks) were treated with the control (Aa) and Treml4 siRNA (Ab and Ac) via subconjunctival injection of 5 µL of siRNA –24 and –6 hours and inoculated with 2000 (Ab) or 1000 (Aa and Ac) CFU of Pa at 0 h. At 1 dpi, all corneas were photographed (A) and a clinical score was assigned (B). One group (n = 3) was subjected to CFU (C) and MPO (D) determination. Another group of corneas was subjected to IHC analysis of neutrophils (NIMP-R14, green) and p-MLKL (necroptotic cells, red) (E). Arrows, NIMP-R14 and p-MLKL positive; arrowheads, p-MLKL positive, NIMP-R14 negative. E, epithelium; S, stroma. Scale bar, 10 µm.

Susceptibility and Severity of Pa Keratitis in T2DM Mice

The Lepr^db/db mouse is a commonly used model owing to an autosomal recessive mutation in the leptin receptor. These mice exhibited hyperphagic, obese, hyperinsulinemic, and hyperglycemic characteristics, with hyperinsulinemia becoming apparent at approximately 2 weeks, obesity at 3 to 4 weeks, and hyperglycemia at week 4 of age. We evaluated the susceptibility of 10-week-old db/db mice as a diabetic model and lean db/+ mice as nondiabetic controls. The severity of Pa keratitis caused by 10,000 CFU inocula (Figs. 8Aa and Ae) in db/+ mice was similar to that of NL corneas (Fig. 1). In contrast, db/db mice had much more severe keratitis, resulting in a perforated eye at 3 dpi (Figs. 8Ab and 8Af). When 5000 CFU inoculum was used, no keratitis was observed in db/+ mice, whereas db/db mice had mild keratitis at 1 dpi (Fig. 8Ac) and corneal perforation at 3 dpi (Fig. 8Ag). Inoculation with 2500 CFU was sufficient to cause Pa keratitis at 3 dpi in db/db mice with bacterial loads similar to that of db/+ mice inoculated with 10,000 CFU Pa. The bacterial burden (CFU) (Fig. 8C) and MPO activities (Fig. 8D) were consistent with micrographs and clinical scores (Fig. 8B), illustrating the severity of Pa keratitis. Overall, these findings suggest that T2DM mice are also more susceptible to and develop more severe Pa keratitis than lean db/+ mice. 

Figure 8.

 

Susceptibility and severity of Pa keratitis in db/db mice. (A) Ten-week-old db/db and db/+ (n = 6, 3 male and 3 female) were scarified and inoculated with indicated numbers of Pa (CFU) at 0 h. The infected corneas were photographed (A) and assigned clinical scores at 1 (B) and 3 dpi (C). At 3 dpi, the corneas were excised and subjected to bacterial counting (white number in A) presented as CFU of Pa per cornea and to MPO determination (units/cornea) (D). The letters in (B–D) correspond with the letters in (A). The results represent three independent experiments (n = 5 each). P values were generated using the nonparametric Mann–Whitney U test for clinical scores and two-way ANOVA for CFU and MPO. * P < 0.05; ** P < 0.01.

Protective Effect of RIPK3 Inhibition on Pa Keratitis in B6 Mouse Corneas

Given the greater susceptibility and severity of Pa keratitis in T2DM corneas, we investigated whether inhibiting necroptosis reverses the adverse effects of T2DM on Pa keratitis. To achieve similar severity, we used different inoculum numbers (10,000 CFU for db/+ and 5000 CFU for db/db mice) and treated the infected mice with GSK-872, a potent, selective inhibitor of RIPK3, a modulator and effector of necroptosis,^44,45 or vehicle control at –4 h before Pa inoculation. The corneas were photographed (Fig. 9A) and clinically scored (Fig. 8B) at 1 dpi. GSK-872 prevented infection in db/+ mice and protected db/db mice from Pa infection with slight opacification, lower clinical scores (Fig. 9B), and bacterial burden (Fig. 9C). We also assessed the expression of IL-1β and IL-1Ra in the infected corneas. We found that infection upregulated their expressions in lean db/+ mice, which was further increased in db/db mice. However, GSK-872 blocked IL-1β and suppressed IL-1Ra expression in db/+ mouse corneas. Consistent with mild keratitis, IL-1β expression was detectable, whereas IL-1Ra expression was significantly augmented in Pa-infected db/db T2DM mice. These findings suggest that increased necroptosis may contribute to severe keratitis in the corneas of T2DM mice, as shown in kidney disease⁴⁶ and Alzheimer’s disease.⁴⁷ 

Figure 9.

 

Suppression of necroptosis improves the outcome of Pa keratitis in db/+ and db/db mice. RIPK3 inhibitor GSK-872 was administrated through subconjunctival injection 4 hours before Pa inoculation. The infected corneas were photographed (A) and assigned clinical scores at 1 dpi (numbers in corneal micrograph). At 1 dpi, the corneas were excised and subjected to bacterial counting presented as CFU of Pa per cornea (B), to MPO determination (units/cornea, C), and to qPCR analysis of IL-1β and IL-1Ra (fold increase) (D). The results are representative of two independent experiments (n = 6), and P values in (B–D) were generated using one-way ANOVA. ** P < 0.01.

Discussion

Studies have shown that impaired immunity in the hyperglycemic environment can lead to an increased incidence and severity of bacterial infections.^48,49 Chronic complications of diabetes, such as neuropathy, peripheral vascular disease, and increased expression of proinflammatory cytokines, have also been linked to microbial infection-related ulcerations in the skin and corneas.⁵⁰ In this study, we aimed to compare the impact of T1 and T2 DM on PCD pathways in response to Pa infection. Our findings suggest that apoptosis and necroptosis play critical and opposing roles in modulating innate immunity, and elevated necroptosis contributes to the increased incidence and severity of BK in T1 and T2DM mouse corneas. Targeting necroptosis may be a promising approach to improve the outcome of BK in patients with DM.⁵¹ 

In our previous studies, we observed that the minimal inocula of Pa that can cause corneal infection in B6 mice is 10,000 CFU. By varying the inocula, we found that diabetic corneas required drastically fewer bacteria to cause keratitis, with as few as 750 CFU in T1DM and 2000 CFU in T2DM. The differences between T1DM and T2DM in requiring inocula for infection may be due to the differences in the duration of hyperglycemia, 10 weeks for T1DM and 6 weeks for T2DM. Thus, mice with uncontrolled or undercontrolled hyperglycemia are highly susceptible to Pa infection. When 10,000 CFU inocula were used, BK developed rapidly, leading to corneal melting at 1 dpi and perforation at 3 dpi, which typically occurs at days 5 to 7 in NL corneas. Furthermore, when lower inocula were used (2000, 1000, and 750 CFU), the severity of keratitis was similar (2000 CFU) or lower (1000 and 750 CFU) at day 1, but by day 3, it was either higher than (1000) or similar to (750) that of NL corneas with 10,000 inocula. Hence, hyperglycemia promotes rapid progression and tissue destruction of BK. Similar responses to Pa infection were also observed in T2DM corneas of db/db mice, despite the vastly different phenotypes of the two mouse lines, STZ- and db/db, such as insulin deficiency, resistance, and altered lipid metabolism in T2DM patients.⁵² This finding suggests that the elevated glucose level is an underlying mechanism shared by the two DM models. 

Infections elicit diverse responses in the host that include activation of the innate immune system, inflammation, and PCD. By adjusting the numbers of inocula to achieve similar severity of keratitis, particularly the bacterial burdens, the mouse models of DM offer a unique opportunity to study the underlying causes for the increased susceptibility of the corneas to bacterial infection at the molecular and cellular levels. We focused on PCD pathways activated in response to Pa infection and altered by hyperglycemia in B6 mice. DAPI staining revealed massive infiltrated cells at lesion sites where the epithelial layer was lost. There were more TUNEL-positive cells or cell debris with higher intensity in T1DM compared with the control NL corneas, indicating that hyperglycemia exacerbates infection-induced overall cell death in B6 mouse corneas. 

Our IHC study revealed that apoptosis of epithelial and stromal cells was abundant in NL corneas, with no detectable p-RIPK3 staining in the stroma. However, in DM corneas, the intensity of RIPK3 staining increased and c-Casp8 staining was greatly suppressed in the epithelium and stroma. Based on these findings, we conclude that DM skews PCD from apoptosis to necroptosis in Pa-infected B6 mouse corneas. Apoptosis is the best-characterized form of cell death and is recognized as an effective way to eradicate invading pathogens⁵³; our observation that epithelial cells near the lesion site undergo apoptosis in NL corneas is consistent with the concept that apoptosis is necessary for deleting infected and damaged epithelial cells, reestablishing epithelial cell barrier function and generating signals important for mucosal immunity. Necroptosis is a caspase-independent mode of PCD that mimics the characteristics of apoptosis and necrosis.⁵⁴ When damaged epithelial cells undergo necroptosis, they release cellular contents such as high mobility group box , IL1α, and IL-33, exacerbating the inflammation generated by the pathogen–host interaction.^25,27,55 This circumstance can lead to hyperinflammation and tissue destruction, thereby decreasing the host's immune defense capacity. 

In NL corneas, in addition to epithelial cells, many neutrophils were found to be c-Casp8 positive. In a healthy host, neutrophils can phagocytose up to 10 to 20 bacteria per cell in a short time,⁵⁶ and the engulfed bacteria are killed by antimicrobial molecules.⁵⁷ The phagocytosis of bacteria prompts the neutrophils to undergo apoptosis, which is beneficial for the host because it allows the safe removal of neutrophil apoptotic bodies containing live or killed bacteria by macrophages via a biological process termed efferocytosis without causing inflammation.^58,59 However, in DM corneas, many neutrophils were p-RIPK3 positive, indicating neutrophil necroptosis. Unlike apoptosis, necroptosis of neutrophils would release live bacteria and alarmins, resulting in the exacerbation of inflammation, dissemination of invading pathogens to uninfected cells, and rapid progression of Pa keratitis.⁶⁰ The PCDs are highly interconnected, particularly apoptosis and necroptosis, in which Casp8 and RIPK1 are key modulators dedicating the occurrence of these pathways.⁶¹ Inhibition of RIPK3, the substrate of activated RIPK1 and effector of necroptosis, not only decreases inflammation, but also increases pathogen eradication.^25,62 Functional analyses using Casp8- and RIPK3-specific inhibitors in NL and DM corneas revealed that the application of Z-IETD-FMK, a potent Casp8 inhibitor, greatly exacerbated Pa keratitis in both NL and DM corneas at 1 dpi, whereas the blockade of RIPK3 activation with GSK872 prevented Pa keratitis from occurring, resulting in no opacification, no recoverable Pa (0 CFU), and significantly decreased MPO activity. Importantly, no significant differences were observed in Pa keratitis between GSK872-treated NL and DM corneas, suggesting that targeting necroptosis is effective in protecting the corneas from Pa infection. 

In addition to altering innate immune cell infiltration and PCD pathways, DM markedly affects the expression of cytokines in response to infection, including IL-17A, F, C, IL-1β, and its antagonist IL-1Ra, which we have shown to play critical role in mediating corneal innate immune response to Pa infection.⁶³ We, therefore, focused on a recently characterized inflammation and immunity regulator, TREML4.^19,64 Its role in the infection of solid tissue, such as the cornea, remains elusive. Our qPCR and IHC studies revealed that TREML4 expression was upregulated at 1 and 3 dpi in response to Pa infection in NL corneas and further augmented in DM corneas. Knockdown of TREML4 resulted in a decreased severity of Pa keratitis, conforming that hyperglycemia-upregulated Treml4 expression had a detrimental role in mediating innate defenses against bacterial infection and keratitis. More important, the downregulation of Treml4 altered PCD pathways, resulting in an increase in apoptosis and a decrease in necroptosis, the inhibition of which removed pathogens, inhibited the development of lesions, and promoted the remodeling of tissue.²⁵ Moreover, because TREML4 is a positive regulator of TLR signaling, whether it acts through alteration of TLR signaling or is directly involved in mediating PCD remains elusive. 

Similar to T1DM, lower inocula of Pa caused more severe keratitis in T2DM db/db mice compared with db/+ mice, indicating increased susceptibility and disease severity. At 1 dpi, infected corneas of db/db mice (5000 CFU) and control db/+ mice (10,000 CFU) with similar keratitis severity had similar levels of bacterial burden and MPO activity, suggesting similar levels of inflammatory and immune responses in obese T2DM and lean normoglycemic controls. However, the IL-1β and IL-1Ra were upregulated by infection and further augmented by DM. Although RIPK3 inhibitor treatment blocked the expression of IL-1β completely, the expression of IL-1Ra remained upregulated at lower levels than without RIPK3 inhibitor treatment in db/+ mice. In db/db mice, RIPK3 inhibitor treatment significantly decreased the levels of both IL-1β and IL-1Ra, suggesting that hyperglycemia-driven upregulation of IL-1β is related to necroptosis, and the inhibition of RIPK3 favors the expression of anti-inflammatory IL-1Ra over IL-1β. Hyperglycemia-driven upregulation of IL-1β is related to necroptosis.⁶⁵ Interestingly, IL-1Ra may alter the balance between apoptosis and necroptosis by directly interacting with Casp-8.⁶⁶ Hence, in addition to targeting IL-1R signaling, human recombinant IL-1Ra (anakinra) may also modulate necroptosis, resulting in an increase in apoptosis and a decrease in the release of alarmins that exacerbated the inflammatory response.⁶⁷ 

In summary, we demonstrated two major PCD pathways in Pa-infected corneas of both T1 and T2DM models. Hyperglycemia skews PCD from apoptosis and necroptosis, resulting in hyperinflammation and tissue destruction. Targeting these two pathways results in opposing outcomes of Pa keratitis. Hence, manipulation of PCD may be an effective adjunct therapy to treating infectious keratitis, particularly in patients with DM. 

Acknowledgments

Supported by NIH/NEI R01EY10869, EY17960 (to F.Y.), p30 EY004068 (NEI core to WSU), and Research to Prevent Blindness (to Kresge Eye Institute). 

Author Contributions: NG performed laboratory testing, sample collection/ analysis, and edited and checked the accuracy of the manuscript. RM performed laboratory testing and data analysis. FY was responsible for study design and recruitment, contributed to sample collection and data analysis, and reviewed and edited the manuscript. FY is the guarantor of this work and, as such, has full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. 

Disclosure: N. Gao, None; R. Me, None; F.X. Yu, None