Time courses of corneal nerves and myeloid cells after injury
To assess the loss of corneal nerves and the influx of myeloid cells following injury, we first defined the time courses of corneal nerve density, cornea-infiltrating and blood-circulating myeloid cells, and clinical ocular manifestations after chemical and mechanical injury to the cornea. The injury was made by applying absolute ethanol to the cornea and scraping off its epithelium in BALB/c mice.28, 31 At 1 min, 30 min, 2 h, 6 h, 1 d, 7 d, 14 d, 21 d and 28 d after injury, the corneas were observed clinically by slit-lamp biomicroscopy and harvested for immunostaining in order to detect corneal nerves, Ly6G+ cells and Ly6C+ cells (Fig. 1A). Also, peripheral blood cells were collected and subjected to flow cytometry for detection of CD11b, Ly6G and Ly6C expression.
The injury caused visually-significant corneal stromal opacity and neovascularization as examined by slit-lamp biomicroscopy (Fig. 1A, B). Time-course analysis indicated that corneal epithelial defect was completely healed by 14 d after injury (day 14) in 80% of mice. Corneal opacity was most noticeable at day 7 and gradually decreased thereafter. Corneal neovascularization ensued thereafter, and reached a peak at days 21─28.
Corneal nerve density progressively declined with time after injury (Fig. 1C, D). The immunostaining of corneal whole-mounts with Ab against β-tubulin III (pan-neuronal marker) showed that both thick stromal nerves and thin sub-basal nerves were significantly lost, beginning within 1 min after injury and resulting in a maximal loss at days 1─7. Then, partial regeneration of deep stromal nerves followed but without recovery of the sub-basal hairpin-like nerve plexus until day 28 (Fig. 1C).
At the same time, the injury triggered a robust inflammatory response with an influx of myeloid cells into the cornea, of which Ly6G+ cells were the predominant population, covering up to 20% of the whole corneal area at day 7 as evaluated by Ly6G immunostaining of corneal flat-mounts (Fig. 1E, F). By comparison, there was less infiltration of Ly6C+ cells into the cornea (Fig. 1G). Interestingly, the kinetics of Ly6G+ cells infiltration into the cornea were in parallel with those of corneal nerve loss. Ly6G+ cells, after injury, infiltrated the cornea progressively from the periphery to the center, reaching a peak at day 7 and disappearing over days 7 to 28 (Fig. 1E, F). A similar time course was observed for the percentage of CD11bhiLy6Ghi cells in the peripheral blood, as analyzed by flow cytometry, whereas there was no change in the percentage of circulating CD11bhiLy6ChiLy6Glo cells at day 1 or 7 (Fig. 1H, I).
The finding that the corneal infiltration of Ly6G+ cells corresponded to the reduction of corneal nerves led us to posit that Ly6G+ myeloid cells might play a role in corneal nerve loss after sterile injury.
CD11bhiLy6Ghi cell depletion exacerbates corneal nerve loss while increasing CD11bhiLy6ChiLy6Glo cells and IL-6
To investigate the role of Ly6G+ myeloid cells, we depleted CD11bhiLy6Ghi cells in BALB/c or C57BL/6 mice by intraperitoneal (IP) injection of 1A8 Ab or isotype-matched control IgG after injury (Fig. 2A-H).32, 33 A significant 75% reduction in corneal Ly6G+ cells was noted by the immunohistochemistry of the corneal whole-mounts (Fig. 2C, D), and a near-complete depletion of systemic CD11bhiLy6Ghi cells in the blood was confirmed by flow cytometry at both days 1 and 7 after injury (Fig. 2E, F).
Depletion of CD11bhiLy6Ghi cells significantly aggravated corneal stromal opacity, delayed corneal epithelial healing, and worsened corneal nerve loss after injury (Fig. 2A-D). The transcript levels of the pro-inflammatory cytokines/chemokines such as IL-1β, CCL2 and CXCL2 in the cornea were significantly lower in mice treated with 1A8 Ab than in those treated with control IgG, which suggests that CD11bhiLy6Ghi cells were the major source of these cytokines/chemokines (Fig. 2G). By contrast, the levels of IL-6, the most highly-upregulated cytokine upon corneal injury (348-fold increase relative to normal cornea at day 1), were markedly upregulated in the cornea and the blood of 1A8-treated mice, as compared to control IgG-treated mice (Fig. 2H). Also, the percentage of CD11bhiLy6ChiLy6Glo cells in the blood was significantly increased in 1A8-treated mice at day 7 (4.0 ± 0.6% in 1A8-treated vs. 1.0 ± 0.4% in control IgG-treated mice, p < 0.0001) (Fig. 2E, F).
These results suggest that CD11bhiLy6Ghi cells play a protective role in corneal nerve loss under sterile injury, notwithstanding the fact that they induce inflammation through the production of pro-inflammatory cytokines/chemokines. Remarkably, CD11bhiLy6Ghi cell depletion led to reciprocal increases in CD11bhiLy6ChiLy6Glo cells and IL-6.
CD11bhiLy6ChiLy6Glo cell depletion aggravates corneal nerve loss while increasing IL-6
In the above experiments, after CD11bhiLy6Ghi cell depletion, CD11bhiLy6ChiLy6Glo cells were increased and corneal nerve loss got more severe (Fig. 2). We thus went on to evaluate whether CD11bhiLy6ChiLy6Glo cells might contribute to corneal nerve damage after injury. To this end, we used IP injection of Monts-1 Ab in BALB/c mice for selective depletion of CD11bhiLy6ChiLy6Glo cells (Fig. 3A-H).34 Near-complete depletion of CD11bhiLy6ChiLy6Glo cells was confirmed by flow cytometry at both days 1 and 7 after injury (Fig. 3E, F). However, CD11bhiLy6ChiLy6Glo cell depletion by Monts-1 Ab did not affect the percentage of CD11bhiLy6Ghi cells in the blood (Fig. 3E, F).
Similarly to CD11bhiLy6Ghi cell depletion (Fig. 2), depletion of CD11bhiLy6ChiLy6Glo cells further aggravated corneal nerve loss as well as corneal opacity after injury (Fig. 3A-D). The mRNA levels of IL-1β, CCL2 and CXCL2 in the cornea as well as corneal Ly6G+ cell infiltration were significantly higher in mice treated with Monts-1 Ab, as compared to those treated with control IgG (Fig. 3C, D, G), indicating that CD11bhiLy6ChiLy6Glo cells had an anti-inflammatory effect on the cornea by suppressing pro-inflammatory cytokines/chemokines. Similarly, the levels of IL-6 in the cornea and blood were enhanced in Monts-1-treated mice (Fig. 3H).
Together, the data demonstrate that CD11bhiLy6ChiLy6Glo cells protect against corneal nerve loss by sterile injury and inhibit the production of pro-inflammatory cytokines/chemokines. Moreover, CD11bhiLy6ChiLy6Glo cell depletion resulted in IL-6 increase, a common finding in cases of CD11bhiLy6Ghi cell depletion.
IL-6 blockade attenuates, while IL-6 supplementation aggravates, corneal nerve loss
Based on the observation that both CD11bhiLy6Ghi cell depletion (Fig. 2) and CD11bhiLy6ChiLy6Glo cell depletion (Fig. 3) commonly led to upregulation of IL-6 alongside aggravation of corneal nerve loss, we next tested the possibility that IL-6 might be responsible for corneal nerve damage after sterile corneal injury. For this purpose, we locally blocked or replenished IL-6 in BALB/c mice by subconjunctival injection of neutralizing IL-6 Ab or recombinant IL-6 protein immediately after injury.
The anti-IL-6 treatment significantly ameliorated clinical ocular manifestations after injury, as reflected by reduced corneal opacity and facilitated epithelial healing, whereas IL-6 addition deteriorated corneal opacity and epithelial defect (Fig. 4A, B). Corneal nerve density was significantly preserved in the anti-IL-6-treated mice, as compared to the control IgG-treated mice, while corneal nerves were further lost in mice treated with recombinant IL-6 (Fig. 4C, D). Moreover, the numbers of both cornea-infiltrating Ly6G+ cells and systemic CD11bhiLy6Ghi cells were significantly repressed by anti-IL-6 treatment and, contrastingly, increased by IL-6 replenishment (Fig. 4D-F). Similar findings were observed with regard to the percentages of CD11bhiLy6ChiLy6Glo cells in the blood (Fig. 4G). It should be noted that neither anti-IL-6 nor recombinant IL-6 treatment had a significant impact on the cornea or systemic myeloid cells in the steady-state (i.e., in mice without injury).
Collectively, the results indicate that IL-6, which is released at high levels in the cornea upon sterile injury, contributes to corneal nerve loss and induces recruitment of mostly Ly6G+ myeloid cells (and, to a lesser extent, Ly6C+ myeloid cells) to the cornea. The myeloid cells, in turn, protect corneal nerves by suppressing IL-6 through a negative-feedback loop (Fig. 4H).
Absence of TLR2, not TLR4, reduces IL-6 and attenuates corneal nerve loss after injury
We further searched for the upstream signaling that stimulates IL-6 and induces corneal nerve loss upon sterile injury to the cornea. Given that activation of Toll-like receptor 2 (TLR2) elicits IL-6 trans-signalling,35 and also in light of our previous data that TLR2 signaling is the principal stimulus for innate immune response in the cornea after sterile injury,31 we determined to explore the roles of TLR2 and TLR4 by utilizing mice lacking TLR2 or TLR4.
Corneal nerve loss and clinical ocular manifestations (corneal stromal opacity and epithelial defect) were markedly attenuated in TLR2 knockout (KO) mice after injury, as compared to wild-type C57BL/6 and TLR4 KO mice (Fig. 5A-D). Corneal Ly6G+ cell infiltration as well as circulating CD11bhiLy6Ghi cells in the blood were reduced in TLR2KO mice, as compared to wild-type or TLR4 KO mice (Fig. 5E-H). The IL-6 mRNA levels in the cornea and blood as well as the IL-1β level in the cornea were significantly lower in TLR2KO mice than in wild-type or TLR4 KO mice (Fig. 5I). On the other hand, no differences were observed in any examined parameters between TLR4 KO and wild-type mice. Therefore, the results demonstrate that TLR2, not TLR4, is a signaling receptor upstream of the IL-6-myeloid cell axis under sterile corneal injury (Fig. 5J).
TLR2 stimulation induces IL-6 production in corneal stromal fibroblasts
We next sought to identify the cell population responsible for IL-6 production in response to TLR2 activation upon corneal injury. Since IL-6 was highly elevated in the cornea denuded of its epithelium at day 1 after injury (Fig. 2H, 3H, 5I), corneal epithelial cells are unlikely to be the source of IL-6. And because CD11bhiLy6Ghi or CD11bhiLy6Chi cell depletion rather resulted in IL-6 increase (Fig. 2H, 3H), it is also unlikely that cornea-infiltrating Ly6G+ or Ly6C+ cells are the main IL-6 producers. We thus examined the possibility that IL-6 might be secreted by keratocytes, which are fibroblasts and a major cell population constituting the corneal stroma.
Primary cultures of keratocytes isolated from human corneal stroma were treated with various concentrations of Pam2csk4, a TLR2 agonist (0 ─ 100 ng/mL) for 24 h. Keratocytes expressed TLR2, TLR3 and TLR4 as assessed by flow cytometry (Fig. 6A, B). Pam2csk4 increased the expression levels of TLR2 in keratocytes in a dose-dependent manner, while TLR3 or TLR4 levels were not affected by Pam2csk4 (Fig. 6A, B). Also, Pam2csk4 upregulated the mRNA level of IL-6 and elicited the production of IL-6 protein in keratocytes in a dose-dependent manner (Fig. 6C, D). Hence, the data suggest that keratocytes are activated to produce IL-6 by TLR2 stimulation, and support the notion that keratocytes are one of the major contributors to IL-6 production after sterile corneal injury.