Diabetes Mellitus Leads to Accumulation of Dendritic Cells and Nerve Fiber Damage of the Subbasal Nerve Plexus in the Cornea

Abstract

Purpose.: To evaluate whether nerve fibers of the subbasal nerve plexus (SNP) and dendritic cells (DCs) are in association with each other leading to neuropathy in the diabetic cornea.

Methods.: BALB/c mice were injected with streptozotocin (STZ) for 5 days for induction of diabetes mellitus (DM) or with vehicle solution (control). B6.VLep^ob/ob (ob/ob) mice served as an obese and glucose-intolerant DM type 2 (DM II) model and lean B6.VLep^ob/+ (ob/+) mice as respective controls. Using in vivo corneal confocal microscopy (CCM), nerve fibers and DCs were quantified over a period of 9 weeks and additionally analyzed by in vitro immunofluorescence whole-mount staining.

Results.: In STZ-diabetic mice, CCM revealed an increase of DC density (DCD) in contrast to controls, whereas nerve fiber density (NFD) was decreased with duration of DM. In ob/ob mice, DCD was 3-fold higher than in both ob/+ mice and STZ-diabetic mice. Whole-mount staining displayed CD11c(+) and major histocompatibility complex (MHC) class II(+) mature DCs in colocalization with class III β-tubulin(+) nerve fibers in the cornea.

Conclusions.: Hyperglycemia leads to corneal DC infiltration, and obesity aggravates this immune response. The direct contact between DCs and the SNP can be assumed to be a trigger of nerve fiber damage and thus a contributing factor to polyneuropathy in diabetic corneas.

Introduction

Diabetes mellitus (DM) is a common metabolic disease characterized by an increased blood glucose (BG) level due to a relative or absolute deficit of insulin and is classified into DM I and DM II. ¹ The insulin-dependent type, DM I, is caused by an autoimmune destruction of the β-cells in the pancreas, whereas the insulin-independent type, DM II, is characterized by insulin resistance. International analyses show a doubling of the diabetic population over the last three decades. ¹ The increased BG level leads to various symptoms based on pathologic changes in organs and tissues of the whole body, including eyes, kidneys, heart, blood vessels, and nerves. ² However, the underlying mechanisms are not yet fully understood. 

The most common type of neuropathy is diabetic neuropathy, ³ defined as change in morphology and function as well as the loss of nerve fibers in different tissues. To determine diabetic neuropathy, only invasive methods like skin biopsy exist. The cornea displays the highest nerve fiber density (NFD) in the human body, comprising innervating nerves with Aδ and C fibers ⁴ that originate from the ophthalmic division of the trigeminal nerve. These fibers are damaged in the earliest stage of DM, leading to peripheral neuropathy. ⁵ Corneal confocal microscopy (CCM) represents an alternative and highly attractive method to assess corneal pathology due to its ability to display detailed information on the different corneal cell layers, including the visualization of nerve fibers of the corneal subbasal nerve plexus (SNP). Degeneration or regeneration of nerve fibers results in changes in corneal NFD. ⁶ Diabetes mellitus leads to the loss of nerve fibers, thus resulting in a lower level of NFD ^7–9 that can be displayed by CCM. ¹⁰  

Besides being able to examine the corneal SNP, CCM has also the ability to detect immunologic cells like the highly reflective dendritic cells (DCs). These bone marrow–derived cells can be found in the noninflamed cornea with a high density in the periphery in contrast to the center, where only a few sessile DCs appear. ¹¹ Due to their function to recognize, process, and present antigens, they are also called professional antigen-presenting cells (APCs), reflecting a kind of guard against foreign influences on the cornea. Dendritic cells exhibit a distinct morphology, showing specific cell processes growing especially to the apical surface of the corneal epithelium. ¹² As DCs are found to be increased in patients with no or mild diabetic neuropathy but decreased with moderate and severe diabetic neuropathy, it has been postulated that they might be involved in the initial stage of diabetic nerve destruction. ¹³ In addition, a strong correlation between an increase of DCs and a decrease of nerve fibers in inflamed corneas has been shown. ¹⁴ Within this context, the term neuroimmunologic synapse was coined to refer to interconnections of the immune and nervous systems. ¹⁵ In previous studies, an interplay between the immune and nervous systems has been demonstrated in skin ¹⁶ and gut. ¹⁷ To add to our current understanding about the role of DCs and nerve fibers in DM, we hypothesize that corneal polyneuropathy is closely linked to an infiltration of DCs. For this purpose, we provide a comprehensive longitudinal analysis of corneal DCs and nerve fibers in experimental mouse models of both streptozotocin (STZ)-diabetic mice and ob/ob mice. 

Materials and Methods

Animals

Female 8- to 12-week-old B6.VLep^ob/ob and BALB/c mice were purchased from Charles River Laboratories (Sulzfeld, Germany). Animals were kept on water and standard laboratory chow ad libitum. All experiments were approved by the local animal welfare committee and were performed in accordance with the German legislation and the principles of laboratory animal care, as well as with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Induction of Diabetes; Blood Glucose and Weight Measurements

For induction of DM, BALB/c mice (n = 7) received an intraperitoneal (i.p.) injection of 50 mg/kg per body weight STZ (Sigma-Aldrich, Steinheim, Germany) dissolved in 0.05 M sodium citrate buffer on 5 days. BALB/c mice, treated with the vehicle solution, served as controls (n = 7). Ob/ob mice (n = 5) exhibit obesity and hyperglycemia. ¹⁸ Lean and normoglycemic ob/+ mice (n = 5) were used as controls. The onset and progression of diabetes were monitored by BG (mmol/L) assessment using the BG meter Contour (Bayer Vital, Leverkusen, Germany). All animals underwent regular weight analysis. 

Corneal Confocal Microscopy

By means of in vivo CCM, nerve fibers of the SNP and DCs were quantified biweekly over a period of 9 weeks using the combination of Heidelberg Retina Tomograph II (Heidelberg Engineering, Heidelberg, Germany) and the in-house–developed Rostock Cornea Module ¹⁹ (University of Rostock, Germany) adapted to mouse imaging. ²⁰ Mice were anesthetized with 75 mg/kg per body weight ketamine (bela-pharm, Vechta, Germany) and 5 mg/kg per body weight xylazine i.p. (Bayer Health Care, Leverkusen, Germany) and fixed in a MouseFix animal holder (Steven GmbH, Ochtrup, Germany). The body temperature was maintained at 37°C with a heating pad. The corneas were wetted with gel (Vidisic; Bausch & Lomb/Dr. Mann Pharma, Berlin, Germany) to prevent desiccation. The curved cornea was then lightly touched with a TomoCap (Heidelberg Engineering GmbH) (Fig. 1A) that was also filled with gel, and prescanned for differentiation between the center and the periphery. In contrast to a previous work, ²¹ in this study planar images revealing clear and sharp morphology throughout the whole area were interpreted as located “inside the center” and served for subsequent analysis (Fig. 1B). Upon movement of the microscope as images revealed a loss of focus due to the convexity of the cornea toward the periphery, the images were interpreted as located “outside the center” and were not considered for further analysis. Additionally, the nerve fiber vortex of the SNP located in the center of the cornea served as a landmark, being characteristic of the corneal center. Within one of the nine images taken of each animal's eye, captured in a meandering pattern (Fig. 1C), the vortex was routinely observed, which further underscored the exclusive analysis of the corneal center. However, the vortex was found located at the border of the central regions and not necessarily located in the most central image. Since nine different pictures were captured, the entire analyzed region contained tissue in very close proximity to the vortex and represents the central cornea. Digital images (0.3 × 0.3 mm, 384 × 384 pixels) were taken using a 0.3-mm² field-of-view lens. Using the sequence mode, three images per second were captured. Finally, offline analysis of these images was performed in a masked manner. Dendritic cell density (DCD) and NFD were assessed by the use of ImageJ (available in the public domain at http://rsb.info.nih.gov/ij/) NeuronJ as plug-in for ImageJ, Analysis 3.1 (Soft Imaging System, Münster, Germany). ¹⁰ Dendritic cell density is defined as the total number of counted DCs per square millimeter (number/mm²), and NFD includes the total accumulative length of all nerve fibers and branches of the SNP (mm/mm²). 

Figure 1

 

(A) Schematic drawing of in vivo CCM using the combination of the Heidelberg Retina Tomograph II and the in-house–developed Rostock Cornea Module. (B) In vivo CCM image of a DC (white arrow) and nerve fibers of the SNP as well as stromal nerves (yellow arrows) of a normoglycemic BALB/c mouse. Scale bar: 50 μm. (C) Schematic drawing of the nine images, captured in a meandering pattern, for in vivo CCM analysis of the corneal center. (D) Corneal flattening of whole mounts by use of four to six radial incisions toward the center.

Corneal Whole-Mount Staining

Before enucleation, mice were killed and subsequently the corneas were processed as described elsewhere. ²² In brief, the separated globes were fixed for 20 minutes at room temperature (RT) in 4% paraformaldehyde (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), 0.2% picric acid in 0.1 M phosphate-buffered saline (PBS; Life Technologies GmbH, Darmstadt, Germany). Afterward the anterior segment was excised and placed back in the fixative solution for 20 minutes, followed by the extraction of the lens and a last incubation in the fixative solution for 20 minutes. After the corneas were freed from the iris, four to six radial incisions toward the center of the cornea were performed with an ophthalmic scalpel (pfm medical ag, Köln, Germany) to flatten the tissue (Fig. 1D). The preparations were stored in 0.1 M PBS with 30% sucrose (Sigma-Aldrich Chemie GmbH Taufkirchen, München, Germany) for 24 hours. Corneas were incubated at 37°C by gentle agitation in 0.1% EDTA (Sigma-Aldrich, Inc., St. Louis, MO, USA) and 0.01% hyaluronidase (type IV-S, product no. H4272; Sigma-Aldrich) in 0.1 M PBS, pH 5.3. Tissues were rinsed for 90 minutes in PBS with 0.3% Triton X-100 (PBS-TX; Roth, Karlsruhe, Germany) and were blocked at RT for 2 hours in PBS-TX containing 1% bovine serum albumin (BSA; Santa Cruz Biotechnology, Inc.). To identify mature DCs, we used CD11c (purified anti-mouse CD11c, clone N418, 1:200; BioLegend, San Diego, CA, USA) combined with major histocompatibility complex (MHC) class II (I-A/I-E) (purified anti-mouse MHC class II, clone M5/114.15.2, 1:200; eBioscience, Inc., San Diego, CA, USA). Neuronal class III β-tubulin antibody (TUJ1) (clone TUJ1 1-15-79, 1:1000; Covance, Princeton, NJ, USA) was used as a pan-neuronal surface marker for corneal nerve fibers. Corneas were rinsed for 90 minutes in PBS-TX followed by the application of DyLight 488-conjugated anti-hamster (Armenian) IgG (1:200; BioLegend), Alexa Fluor 555-conjugated anti-rat (1:400; Life Technologies GmbH), and Alexa Fluor 633-conjugated anti-rabbit (1:400; Life Technologies GmbH) for 2 hours. After rinsing of the tissue for 90 minutes in PBS-TX, corneas were stained with 4′,6-diamidino-2-phenylindole (1:1000, AppliChem GmbH, Darmstadt, Germany) as nuclear staining and mounted in fluorescent mounting medium (Dako, Jena, Germany). 

Confocal Imaging of Corneal Whole Mounts

Images from processed corneas were captured with a 60-fold oil immersion objective lens of an Olympus Fluoview 10i confocal microscope (Olympus, Hamburg, Germany). To assess the three-dimensionality with precise information on the SNP and DCs including their processes, z-stacks were performed. Slices were captured in steps of 0.5 μm and reconstructed to one image. Images were taken from the center of each cornea to allow comparability between CCM and confocal microscopy of corneal whole mounts. Offline analysis of the histologic specimen was performed in a masked manner. 

Statistical Analysis

The SPSS version 20.0. (IBM Corp., Armonk, NY, USA) General Linear Model (GLM) repeated measures ANOVA was used for statistical analysis of the data to test the effect of both the between-subject factor (treatment) and the within-subject factor (time). By means of SigmaStat 3.5 software (SigmaStat; Jandel Corporation, San Rafael, CA, USA), differences between groups were calculated with one-way repeated measures ANOVA with an appropriate post hoc test, and linear regression was performed for analyzing correlations between BG, DCD, and NFD. A P value of <0.05 was considered statistically significant. 

Results

Blood Glucose Response

Two weeks after the first STZ application, mice showed an increase in BG level of 10.8 mmol/L and an average blood BG level of 13.5 mmol/L during the remainder of the experiment (Fig. 2A). 

Figure 2

 

Average BG levels of STZ- and sodium citrate–treated BALB/c mice (n = 7) (A) as well as ob/ob and ob/+ mice (n = 5) (B). Quantitative analysis of DCD (number/mm²) (C) and NFD (mm/mm²) (E) in the corneal epithelium of healthy mice and mice with STZ-induced DM over an observation period of 9 weeks. Sodium citrate–treated BALB/c mice (n = 7) served as normoglycemic controls (sham), and STZ-treated BALB/c mice (n = 7) represent hyperglycemic DM I mice (STZ). Quantitative analysis of DCD (number/mm²) (D) and NFD (mm/mm²) (F) in the corneal epithelium of healthy mice and mice with DM II over an observation period of 9 weeks. B6.VLep^ob/+ mice (n = 5) served as normoglycemic controls (ob/+), and B6.VLep^ob/ob mice (n = 5) represent hyperglycemic DM II mice (ob/ob). Means ± SEM; *P < 0.05 vs. sham at the respective time points; #P < 0.05 vs. 1 week.

Ob/ob mice revealed an average BG level of 15.9 mmol/L from the first detection of BG until week 4 and showed decreased BG levels (10.6 and 11.8 mmol/L) at the last two time points (Fig. 2B). 

CCM-Based Analysis of the Cornea in STZ-Diabetic Mice

The parallel arrangement of single nerve fibers of the SNP and between residing DCs is exemplified in Figure 1B. After treatment with STZ, BALB/c mice over the period of 1 to 9 weeks showed a gradual increase of corneal DCD in contrast to the constant level of DCD in control mice. At weeks 7 and 9, DCD displayed a significant rise (Fig. 2C) when compared to both the corresponding values for control animals (Fig. 2C) and within-group values at week 1, with an approximately 3-fold increase at week 9 (17 ± 2 / mm²). In addition, the increase of DCD in STZ-diabetic mice showed a significant positive correlation with increasing BG levels (Fig. 3C) in contrast to sham mice (Fig. 3A). The NFD in the cornea of STZ-diabetic mice significantly decreased at weeks 5, 7, and 9, in contrast to constant levels in sham mice (Fig. 2E). Regression analysis of STZ-diabetic mice showed a significant negative correlation between DCD and NFD (Fig. 4C) but not in sham mice (Fig. 4A). The GLM repeated measures ANOVA revealed that for BG and NFD, the time effect and the interaction between time and treatment were significant (P < 0.05). For DCD, the time effect was also significant (P < 0.05). 

Figure 3

 

Regression analysis between BG (mmol/L) and DCD (number/mm²) of sodium citrate–treated sham mice (n = 7) (A), STZ-treated mice (n = 7) (C), ob/+ mice (n = 5) (B), and ob/ob mice (n = 5) (D). R, regression coefficient.

Figure 4

 

Regression analysis between DCD (number/mm²) and NFD (mm/mm²) of sodium citrate–treated sham mice (n = 7) (A), STZ-treated mice (n = 7) (C), ob/+ mice (n = 5) (B), and ob/ob mice (n = 5) (D).

CCM-Based Analysis of the Cornea in ob/ob Mice

Both ob/+ and ob/ob mice exhibited different levels of all measured parameters during the entire period when compared to BALB/c mice. Ob/ob mice showed a nearly constant high level of DCD (65 ± 6 / mm²) (Fig. 2D), which was statistically different over the entire observation period in contrast to ob/+ mice (23 ± 5 / mm²) (Fig. 2D). The values for DCD in ob/ob mice showed a significant negative correlation with BG levels (Fig. 3D) in contrast to ob/+ mice (Fig. 3B). Nerve fiber density in ob/ob mice was lower than in control mice (Fig. 2F). Regression analysis on ob/ob and ob/+ mice showed no correlation between DCD and NFD (Figs. 4D, 4B). The GLM repeated measures ANOVA revealed that for NFD the time effect was significant (P < 0.05). 

Corneal Whole-Mount Staining in Mice

Whole-mount staining procedure enables an accurate presentation of different structures in the cornea. CD11c(+) and MHC class II(+) DCs reside in the basal layer of the corneal epithelium and represent mature cells. Their processes originating from cell bodies follow different directions leading to the surface of the cornea (Figs. 5A, 5B). The SNP is located between the epithelium and the anterior stroma, and class III β-tubulin(+) single nerve fibers sprout within the superficial epithelial layers to the surface of the cornea (Fig. 5C). 

Figure 5

 

Localization of DCs and nerve fibers within the corneal epithelium and anterior stroma assessed by performing a z-stack through the corneal whole mount of a normoglycemic BALB/c mouse. DCs express CD11c (A) and MHC class II (B). In general, they are localized between the epithelium and anterior stroma. The cell bodies reside beneath the basal cell layer, and the processes of the DCs are directed toward the corneal surface. Class III β-tubulin(+) (TUJ1) nerve fiber bundles of the stroma (C) sprout into single nerve fibers forming the SNP between the anterior stroma and the epithelium. Associated small fibers resulting from the SNP are arranged among epithelial cells. Scale bars: 30 μm.

Corneal Whole-Mount Staining in Both DM Models

Control BALB/c and ob/+ mice at week 9 showed a gradient of corneal DCD with a decrease from the periphery toward the center (Figs. 6A, 6C). Streptozotocin-diabetic and ob/ob mice at week 9 also presented a gradient of DCD from the periphery to the center, however with much higher numbers of mature DCs (Figs. 6B, 6D). 

Figure 6

 

Distribution of CD11c(+) and MHC class II(+) DCs in corneal whole mounts. Images were taken with a 10× objective lens. All groups revealed a decreasing gradient of DCs from the periphery to the center. In contrast to sodium citrate–treated BALB/c and B6.VLep^ob/+ mice (A, C), STZ-treated BALB/c and B6.VLep^ob/ob mice display a higher density of DCs (B, D). Scale bars: 500 μm.

All corneal whole mounts of each treatment group displayed DCs, which are positive for both CD11c and MHC class II. Control BALB/c mice at week 9 showed many class III β-tubulin(+) nerve fibers forming the typical parallel structure of the SNP (Figs. 7B, 7E) and sparsely dispersed CD11c(+) and MHC class II(+) DCs (Figs. 7B–D). In agreement with the quantitative analysis of CCM images, corneal whole mounts of STZ-diabetic mice at week 9 revealed a higher density of DCs and a lower density of nerve fibers (Figs. 8B–E). In contrast to ob/+ mice (Figs. 9B–D), ob/ob mice showed a much higher level of DCD (Figs. 10B–D). Higher magnification of corneal whole mounts, as exemplified in Figure 11, reveals that DCs often colocalize with nerve fibers and seem to establish close contacts. 

Figure 7

 

In vivo CCM and confocal imaging of corneal whole mounts in normoglycemic BALB/c mice at week 9. (A) CCM of nerve fibers. Scale bar: 50 μm. (B–E) Confocal imaging of corneal whole mounts with a single CD11c(+) (B, C, green) and MHC class II(+) (B, D, red) DC and class III β-tubulin(+) (TUJ1) nerve fibers (B, E, yellow). DAPI (blue) as nuclear staining. Scale bars: 30 μm.

Figure 8

 

In vivo CCM and confocal imaging of corneal whole mounts in hyperglycemic BALB/c mice at week 9. (A) CCM of DCs (arrows) and nerve fibers. Scale bar: 50 μm. (B–E) Confocal imaging of corneal whole mounts with three CD11c⁺ ([B, C], green) and MHC class II(+) ([B, D], red) DCs and class III β-tubulin(+) (TUJ1) nerve fibers ([B, E], yellow). DAPI (blue) as nuclear staining. Scale bar: 30 μm.

Figure 9

 

In vivo CCM and confocal imaging of corneal whole mounts in normoglycemic B6.VLep^ob/+ mice at week 9. (A) CCM of DCs (arrows) and nerve fibers. Scale bar: 50 μm. (B–E) Confocal imaging of corneal whole mounts with three CD11c(+) ([B, C], green) and MHC class II(+) ([B, D], red) DCs and class III β-tubulin(+) (TUJ1) nerve fibers ([B, E], yellow). DAPI (blue) as nuclear staining. Scale bar: 30 μm.

Figure 10

 

In vivo CCM and confocal imaging of corneal whole mounts in hyperglycemic B6.VLep^ob/ob mice at week 9. (A) CCM of DCs (arrows) and nerve fibers. Scale bar: 50 μm. (B–E) Confocal imaging of corneal whole mounts with several CD11c(+) ([B, C], green) and MHC class II(+) ([B, D], red) DCs and class III β-tubulin(+) (TUJ1) nerve fibers ([B, E], yellow). DAPI (blue) as nuclear staining. Scale bars: 30 μm.

Figure 11

 

Higher magnification of a confocal image of a corneal whole mount, displaying the close position between a mature DC and nerve fibers of the SNP. CD11c(+) (green), MHC class II(+) (red) DC, and class III β-tubulin(+) (TUJ1) nerve fibers (yellow). DAPI (blue) as nuclear staining. Scale bar: 20 μm.

Discussion

Methodological Considerations

In contrast to invasive skin biopsy, CCM represents a noninvasive method for repetitive in vivo investigations, allowing immediate identification of morphologic characteristics such as changes in corneal nerve fibers and DCs. For evaluation of corneal neuropathy, the study focused on the SNP and not on stromal nerve fibers. However, one cannot exclude that possibility that in individual images, stromal nerve fibers were detected as well (Fig. 1B, yellow arrows). Nevertheless, for offline quantification of the in vivo CCM images, only single nerve fibers of the SNP were quantified. In general, the visualization of whole nerve fiber bundles, such as those present in the stromal level, does not allow identification of the loss of single nerve fibers and thus does not allow the exact quantification of a reduction of NFD. Analysis of single nerve fibers of the SNP might have an additional advantage in that the loss of an individual nerve fiber might precede whole-bundle degeneration. Therefore, in contrast to stromal nerve fiber bundles, alteration of the SNP might serve as an earlier indicator for diabetic polyneuropathy. To ensure optimal reproducibility of the distribution of subbasal nerve fibers and DCs, nine images per cornea were quantified in this study. The question whether diabetic neuropathy occurs from the center or from the periphery cannot be answered by the present study because the peripheral area was not investigated. Nevertheless, we suppose that corneal neuropathy would start from the periphery and not within the center of the cornea due to the fact that corneal neuropathy, among others, might be initiated by paracrine signals reaching the cornea via the limbal vascular plexus located in the very periphery, forming the border between the conjunctiva and the avascular cornea. 

The in vitro analysis of corneal whole mounts allowed discrimination of CD11c(+) DCs and their immature or mature status through the use of MHC class II, supporting further interpretation of CCM images. With the concomitant use of the neuronal class III β-tubulin(+) antibody as pan-neuronal marker, a comprehensive analysis of both DCs and nerves under different BG levels was possible. 

Induction of DM can be achieved through a single high-dose or multiple low-dose injections of STZ. The application of high-dose STZ leads to a severe necrosis of pancreatic β-cells, in contrast to the multiple low-dose application of STZ that leads to specific β-cell death. ²³ BALB/c mice underwent an application of STZ in multiple low doses over 5 days and as lean hyperglycemic animals represent a model of DM I. ^24–26  

Due to the leptin deficiency, B6.VLep^ob/ob mice display a spontaneous DM II model exhibiting impaired glucose tolerance, hyperglycemia, hyperinsulinemia, and obesity. ²⁷ This mouse strain is commonly used as a model for a diabetes-like syndrome. ^25,27,28 It has to be noted that ob/ob mice display only a transient hyperglycemia between weeks 8 and 12, compensated later by increasing insulin levels. ²⁹  

Nerve Fibers of the SNP in the Diabetic Cornea

Several studies using CCM have reported changes in morphology of corneal nerve fibers during the progress of DM and described a decrease in NFD. ^9,10,30,31 The present study could confirm these observations in DM I by the significant decrease of NFD at weeks 5, 7, and 9 in STZ-treated mice. The NFD of ob/ob mice remained approximately constant over time with a lower level compared to that in ob/+ mice. The absence of significant differences between the two groups relates to the tendency toward higher NFD in ob/ob mice from week 5 on. Notably, the higher NFD values inversely correlate with lower BG values, whereas both parameters remain constant in ob/+ mice. Another study reported the development of peripheral neuropathy in ob/ob mice. ¹⁸ However, that study used 4-week-old animals that showed high BG (>13.8 mmol/L). ¹⁸ In contrast, our study used 8- to 12-week-old animals, which had lower BG at later time points, most likely due to hyperinsulinemia occurring at this age. 

DCs in the Diabetic Cornea

In line with Knickelbein et al., ¹² DCs were found beneath cells of the basal epithelium and displayed several long processes with direction toward the corneal surface. In addition, the current data confirm those of Knickelbein et al. ¹² and Lee et al., ³² in that the DCD revealed a decreasing gradient from the periphery toward the center of the cornea under normal conditions, but extend the current knowledge by demonstrating that diabetic mice have much higher DCD than found in nondiabetic controls. To our knowledge this is the first report that STZ-induced DM is associated with a constant increase of DCs over 9 weeks, while ob/ob mice show a constantly high level of DCD, most probably due to enhanced inflammatory and altered immunologic status of these mice. ³³  

The increase of DCs upon diabetes induction might be interpreted as a cellular response to inflammation, as DM is known to be associated with systemic inflammation. ^34,35 This view is supported by the fact that inflammatory stimuli like electric cautery to the ocular surface or application of lipopolysaccharide and tumor necrosis factor-α induce corneal DC infiltration and maturation ^32,36 and are closely linked to a decrease of subbasal nerve fibers in patients with infectious keratitis. ¹⁴ Accordingly, Popper and coworkers (Popper M, et al. IOVS 2005;46:ARVO E-Abstract 879) described diabetic corneal neuropathy in association with an increase of very highly reflective cells in patients with DM II. Tavakoli et al. ¹³ also showed these features in the cornea of DM I and DM II patients with an additional correlation to the loss of corneal nerve fibers. 

In addition to DCs, Langerhans cells (LCs), which similarly function as professional APCs, and macrophages reside in the cornea. ^36,37 Mayer et al. ³⁸ demonstrated that corneal Langerin⁺ LCs coexpress MHC class II and CD11c characterizing the DC lineage. Furthermore, these cells are exclusively located in the corneal epithelium and display a decreasing gradient from the periphery to the center. ³⁸ During inflammation, the cornea shows an influx of LCs expressing costimulatory markers similar to DCs, ³⁶ suggesting a likewise behavior of these cells under diabetic conditions. Moreover, CD11c(−) and CD11b(+) macrophages could be identified in the cornea, mainly located within the posterior stroma ^37,38 where they show an association to nerve fiber bundles. ³⁹ This morphologic proximity of the two cell types might also play a role in corneal diabetic polyneuropathy and needs further investigation. 

Hamrah et al. ³⁶ demonstrated that DCs in the uninflamed epithelium of the corneal center are MHC class II negative. Under inflammatory conditions, however, they described a subset of DCs expressing MHC class II. This observation is partly in contrast to our findings in corneal whole mounts, demonstrating that all DCs, independently of diabetic or nondiabetic conditions, were CD11c(+) and MHC class II(+). This might indicate a permanently activated status of corneal DCs, confirming them as highly sensitive cells against foreign antigens. 

Due to the significant negative correlation between DCD and nerve fibers in STZ-induced DM, it might be concluded that the increase of DCs could play an initial role in the manifestation of diabetic corneal neuropathy. The close position of DCs and single nerve fibers of the SNP suggests communication between the two cell structures. In other organ systems like the skin, a colocalization of LCs with epidermal nerves was observed. ⁴⁰ Furthermore, in lung ⁴¹ and in the gut, ⁴² an interaction between DCs and nerve fibers was described. A morphologic association between neurites and mast cells and the additional presence of vesicles at the points of contact were also shown in a nerve–mast cell coculture. ⁴³ These findings support our assumption that a delicate interaction between nerve fibers and DCs could also exist in the diabetic cornea. 

However, corneal nerve fiber loss and DC infiltration might also be different phenomena that occur accidentally upon hyperglycemia at the same time but independently of each other. There could be confounding factors, for example, vascularization that develops after denervation resulting in an influx of DCs. While Ferrari et al. ⁴⁴ reported on corneal neovascularization after denervation upon trigeminal stereotactic electrolysis with a concomitant increase of CD45(+) inflammatory cells in the cornea, ⁴⁴ we could not observe corneal neovascularization as a potential attractor of DC infiltration. However, due to the ability of MHC class II(+) DCs to form membrane nanotubes, mounting a possible transfer system of different molecules in healthy and inflamed murine corneas, ^45,46 these cell processes might reach the nerve fibers of the SNP and contribute to corneal polyneuropathy. 

In summary, the current data underline the possible association between an increase of DCs and a decrease of nerve fibers in the cornea. As DM is an inflammatory disease, interaction of the two cell types could lead to neurogenic inflammation, resulting in the release of neuropeptides and thus contributing to a bidirectional interaction ⁴⁷ of corneal DCs and nerve fibers. 

Acknowledgments

The authors thank Carl F. Marfurt, PhD, for his support in whole-mount staining (Department of Anatomy and Cell Biology, Indiana University School of Medicine-Northwest, Gary, Indiana), Heike Weiss, PhD, for her helpful advice on confocal microscopy of corneal whole mounts (Institute of Medical Biochemistry and Molecular Biology, University of Rostock), and Günther Kundt, PhD (Institute for Biostatistics and Informatics in Medicine and Aging Research, University of Rostock), for his excellent help and advice in the statistical revision of the data. 

Supported by a grant from the Federal Ministry of Education and Research (894193; REMEDIS [Höhere Lebensqualität durch neuartige Mikroimplantate]). The authors alone are responsible for the content and writing of the paper. 

Disclosure: K. Leppin, None; A.-K. Behrendt, None; M. Reichard, None; O. Stachs, None; R.F. Guthoff, None; S. Baltrusch, None; J.C. Eule, None; B. Vollmar, None