Dysregulation of the OGF–OGFr pathway correlates with elevated serum OGF and ocular surface complications in the diabetic rat

Type 1 diabetes rat model

T1D was induced in six-week old male Sprague-Dawley rats (Charles River) by injection of 50 mg/kg STZ (Sigma-Aldrich) dissolved in buffer at pH 4.5 for two consecutive days.^13,17 Rats were fasted for 18 h prior to each injection; this method produced hyperglycemia (blood glucose levels > 250 mg/dL) within 72 h and resulted in minimal lethality. A cohort of hyperglycemic rats was implanted with insulin implants (LinShin, Canada) within 24 h of detecting hyperglycemia, and identified as the T1D-INS group. LinShin insulin implants designed for 60 days of insulin release (7 mm in length) were subcutaneously implanted using the manufacturer’s instructions. Insulin was released within 1 h of minipump implantation at a rate of 2 U/h. Hyperglycemic animals not treated with insulin were identified as the T1D group. Finally, rats not rendered hyperglycemic secondary to STZ receiving intraperitoneal injections of sodium citrate buffer only were removed from the study. All animal protocols adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and were approved by the Penn State College of Medicine Institutional Animal Care and Use Committee.

Measurement of physiological markers of diabetes

At weekly intervals for eight weeks, rats were weighed and glucose levels were monitored from tail blood. At each time point, at least 15 rats (randomized) for each group (T1D, T1D+insulin, and Normal) were evaluated. Some rats in each group were maintained until week 11 post STZ injection, but the experiment was halted when T1D animals required humane euthanasia according to the IACUC regulations.

Assessment of ocular surface complications

Prior to induction of hyperglycemia and at 2, 3, 4, 5, 6, and 8 weeks following STZ injection, 12 to 20 male rats from each treatment group (T1D, T1D-INS, Normal) and representative of different experiment/shipments were examined to determine the extent of ocular surface corneal complications. All measurements were conducted in unanesthetized animals, and followed previous protocols.^12,15 Dry eye was determined by the Schirmer 1 test, which consisted of inserting Schirmer paper strips (1 mm × 17 mm long) (Alcon) into the lower lid cul-de-sac for 1 min and measuring the wetting length to the nearest half millimeter using a scale provided by the manufacturer. Corneal surface sensitivity was assessed by measuring the blink rate using the Cochet-Bonnet aesthesiometer (Boca Raton, FL). The aesthesiometer pressure (g/mm²) was determined directly from the protocol supplied by the manufacturer. Increased pressure required to induce blinking when the non-anesthetized cornea was touched with the aesthesiometer filament indicated decreased corneal sensitivity.

Rate of corneal reepithelialization

Corneal epithelial wound closure was measured as the percentage of residual defect following 5 mm diameter abrasions created in the right eye; procedures and protocols followed those published previously.^13,17 All surgeries were conducted between 0800 and 0900 h to prevent disparities from diurnal rhythm. Wound closure rates were determined by photographing the residual defect stained with moistened fluorescein strips at 0 (baseline), 16, 24, 32, 40, 48, 56, 72, and 90 h. No eyes were photographed at intervals ≤16 h in order to prevent disruption of the epithelial healing process. The residual area of the epithelial defect was calculated as the percentage of the original epithelial defect.

Tissue levels of OGF and OGFr

All rats were humanely euthanized with an intraperitoneal sodium pentobarbital (Euthansol) injection (>100 mg/kg; >0.2 mL/100 g body weight) at 90 h following wounding and eyes proptosed for immunohistochemistry. To assess abnormalities in OGF or OGFr expression in the cornea, the unwounded eye from three rats in each group was removed, quickly frozen, and stored at −80°C until used. Frozen corneal tissue was sectioned (8–10 µm) and stained with validated antibodies to OGF (1:200) or OGFr (1:150; Bethyl Laboratories, Inc., Montgomery, TX) according to the published procedures.^22,23 Control sections of each treatment group were stained with secondary antibody only. Stained sections were photographed using confocal microscopy (Keyence Microscopy). Images were calibrated using optical density step tablet software provided by ImageJ such that densitometric analyses removed the mean gray background values and converted data from pixels to optical density units. At least 10 tissue sections from different animals were stained for immunohistochemical semi-quantitative assessment. Expression levels of peptide or receptor were analyzed by performing densitometric analyses on photographs captured at similar settings and magnification using computer software; brightness was converted to numerical values.

Serum levels of OGF, OGFr, and CD10 enkephalinase

Beginning on week 2 following STZ injections, three to six rats from each group (T1D, T1D-INS, Normal) were humanely euthanized with an intraperitoneal sodium pentobarbital injection (>100 mg/kg; >0.2 mL/100 g body weight). Whole trunk blood was collected on ice, centrifuged to obtain serum, and aliquots were stored at −80°C prior to assay. Enzyme-linked immunosorbent assay (ELISA) kits were purchased from MyBioSource for OGF (MBS756126) or OGFr (#MBS109224). CD10 (Cluster of differentiation 10, also termed neprilysin, is a neutral endopeptidase that degrades enkephalin. CD10 is not selective for methionine enkephalin (i.e. OGF), but was measured by an ELISA kit from MyBioSource (#MBS764927) with rat sensitivity of 0.094 ng/mL to determine if elevated OGF levels were related to changes in enzymatic degradation.

All serum analyses involved multiple assays with duplicate samples from at least three different rats at each time point collected across several independent experiments analyzed in a single ELISA. Each assay also included samples from rats tested in all assays as positive controls.

Data analyses

Rigor and reproducibility of the study were ensured by conducting multiple consecutive experiments with smaller cohorts of rats and using replicate samples from multiple animals in different groups and experiments. All parametric data (e.g. body weights, glucose levels, Schirmer measurements, serum values) were analyzed with GraphPad Prism 8.0 software using two-way (condition, time point) analysis of variance (ANOVA) to determine significant interaction. Subsequent comparisons were made using Newman–Keuls tests; 95% confidence levels were required for further analyses.

Clinical characteristics

Across all four independent experiments, more than 95% of the rats injected with STZ had blood glucose levels of >250 mg/dL within 72 h. All T1D rats used in the studies were hyperglycemic within five days. Three rats did not convert to being hyperglycemic and were removed from the study. For the T1D-INS group, blood glucose levels began to level off within hours of insulin pump implantation. Insulin implants were replaced in a few rats that were studied until week 11. No deaths occurred in the first two weeks post STZ injection; T1D-related deaths were only reported at weeks 10 and 11.

Body weights and blood glucose levels are presented in Figure 1(a) and (b), respectively. Two-factor ANOVA identified a significant interaction (P < 0.0001) between condition (T1D, Normal, T1D-INS) and time (weeks), and subsequent analyses identified significant differences in body weight and glucose levels between T1D and Normal animals, as well as T1D and T1D-INS. Analyses indicated that hyperglycemia was evident within two weeks and that rats receiving systemic insulin (T1D-INS) had normal body weights and glucose levels throughout the course of the study. Normal body weights increased from approximately 180 g to over 470 g in eight weeks, whereas the body weights of T1D rats were approximately 300 g after two months of hyperglycemia. In comparison to blood glucose levels for normal and T1D-INS rats that ranged from 128 to 152 mg/dL or 126 to 200 mg/dL, respectively, glucose levels for T1D rats were ≥ 500 mg/dL.

OPEN IN VIEWER

Figure 1.

(a) Body weights (grams) and (b) blood glucose levels (mg/dL) for male Sprague-Dawley rats randomly selected from T1D, T1D-INS, or Normal groups (n ≥ 14 per group). Values represent means ± S.E. M. Two-factor ANOVA revealed a significant interaction between condition (T1D, T1D-INS, Normal) and time (weeks) at P < 0.0001. Comparisons of condition at each time point were significantly different from Normals at P < 0.001 (***).

Temporal expression of abnormalities in non-invasive measures on the corneal surface

For both parameters, two factor ANOVA revealed a strong interaction (condition × time) of P < 0.0001, and significant main effects of condition and time (each P < 0.0001). Corneal surface sensitivity (Figure 2(a)) and tear production (Figure 2(b)) were monitored on a weekly basis. T1D rats demonstrated changes in aesthesiometry after four weeks, with both T1D and T1D-INS animals having abnormalities on weeks 5 through 8. For T1D rats, the pressure required to elicit a response rose from 0.4 to 0.5 g/mm² (week 4) to greater than 0.6 g/mm² in weeks 5 through 8. In comparison, the pressure required was less than 0.5 g/mm² during weeks 1–4 in the T1D-INS group, but increased to 0.6–0.8 g/mm² during weeks 5–8; normal animals required pressure levels of 0.4 g/mm² or less to elicit a blink response.

OPEN IN VIEWER

Figure 2.

(a) Corneal surface sensitivity measured with a Cochet-Bonnet aesthesiometer. Three readings of pressure (g/mm²) were averaged for each rat at every time point. (b)Tear production (mm) measured by the Schirmer 1 test. Values represent means ± S.E.M. for male rats randomly selected at each time point from T1D, T1D-INS, and Normal conditions (n ≥ 14 per group). Two-factor ANOVA of corneal sensitivity and tear production revealed significant interactions (condition × time) at P < 0.0001. Post hoc one-way ANOVA revealed differences among groups at each time point (P < 0.001) that were significantly different from Normals at P < 0.01 (**), P < 0.001 (***) using Newman–Keuls tests.

Tear production, which was measured as the mm distance of wetting of the Schirmer paper strip, was consistent for all groups for the first four weeks (Figure 2(b)) and ranged between 5 and 6 mm. Two factor ANOVA revealed a strong interaction (condition × time) of P < 0.0001, with time being P < 0.0002. At five weeks post STZ injection, tear production was significantly reduced (P < 0.001) in controlled and uncontrolled diabetic rats, being 40% less than Normals (3.5 vs. 5.8 mm). Tear fluid measurements remained significantly lower than Normals for both T1D and T1D-INS rats, with mean values at weeks 5–8 ranging between 3.4 and 4.1 mm for T1D and T1D-INS rats in comparison to 5.7–6.1 mm for Normals.

Some T1D rats survived to week 11, along with T1D-INS and Normal animals, and body weight, blood glucose, tear production, and corneal surface sensitivity were monitored. Although n values were too low for reliable statistical analyses of the T1D-INS group, male T1D rats continued to have blood glucose levels exceeding 550 mg/dL through week 11, and significantly reduced tear production (2.8 mm) in comparison to Normal values of 5.8 mm. At approximately 80 days post initial hyperglycemia, all rats were humanely euthanized to adhere to the institutional IACUC regulations.

Magnitude of delayed corneal re-epithelialization

A measure of corneal epithelial dysfunction in T1D rats was the assessment of re-epithelialization following the creation of a 5-mm diameter central corneal epithelial wound (Figure 3(a) and (b)). Although there was no significant interaction of condition and time, by 16 h following wounding, the T1D rats showed evidence of slower wound healing, exhibiting significantly more residual wound area relative to their baseline wound areas and in comparison to the residual wounds in both T1D-INS and Normal rats. Residual wounds were 4.5% at 56 h for Normals in comparison to 13% and 23% for T1D-INS and T1D animals, respectively. By 64 h, wounds in the Normal rats were completely healed, whereas the male T1D rats had residual wounds of approximately 9%. T1D rats had residual wounds greater than 2% of their original area until 86 h. Corneal wounds in T1D-INS rats appeared to heal at approximately the same rate as those in Normal rats, except for measurements at 40 h, which indicated slower healing for the T1D-INS animals, as well as for the T1D rats. When wounds were less than 1% of their original size, they were considered “healed”.

OPEN IN VIEWER

Figure 3.

Photographs (a) and percent residual corneal defects (b) for T1D, T1D-INS, and Normal rats (n= 6–9 animals per group). Eyes were stained with fluorescein strips and photographed prior to wounding (0 h) and at 16, 24, 32, 40, 48, 56, and 72 h after abrasion. Two-factor ANOVA did not reveal a significant interaction with condition, but time was significant with one-factor ANOVA, P < 0.001. Residual areas (means ± S.E.M.) are presented as the percentage of remaining wound relative to the area of the original wound for the same rat. Significantly different from Normal animals at P < 0 05 (*), P < 0.01 (**), P < 0.001 (***); significantly different between T1D and T1D-INS at P < 0.05 (+). (A color version of this figure is available in the online journal.)

Dysregulation of OGF and OGFr expression in the corneal epithelium

For both parameters, two factor ANOVA revealed a significant interaction between condition and time (P < 0.01), as well as significant main effects (each P < 0.01). OGF expression in corneal epithelial tissue was semi-quantitatively measured at two, three, four, and six weeks (Figure 4(a) and (b). OGF protein expression was comparable for all animals on weeks 2 and 3, and was recorded as approximately 35 optical density units. OGF expression was substantially higher in both diabetic groups on weeks 4 and 6. By week 4, OGF values in the T1D group were semi-quantitatively assessed at 75 optical density units in comparison to 35 units for Normal tissue, and the T1D-INS group had mean measurements of 74 and 69 optical units for weeks 4 and 6, respectively, representing more than twice the OGF expression as Normal rats (Figure 4(b)).

OPEN IN VIEWER

Figure 4.

OGF expression in corneal epithelium. (a) Photomicrographs of intact corneal epithelium at two, three, four, and six weeks following hyperglycemia in T1D, T1D-INS, and Normal rats stained with validated anti-OGF antibodies. Tissue stained with only secondary antibody is shown in the insert at week 6. Images were taken at 10× magnification. (b) Tissue images were scanned with a confocal microscope and semi-quantitatively evaluated for OGF expression using Image J software. Values represent means ± S.E.M. for 15–20 readings collected from at least three tissue specimens. Two-factor ANOVA revealed a significant interaction between condition and time, P < 0.01, with specific differences between conditions noted at four and six weeks. Significantly different from Normal levels at P < 0.001 (***). (A color version of this figure is available in the online journal.)

OGF receptor expression in corneal epithelium was also measured in sequential sections from the same animals (Figure 5(a) and (b)). Receptor protein expression levels in Normal rats ranged between 15 and 20 optical density units, whereas both controlled (T1D-INS) and uncontrolled (T1D) diabetic rats had elevated levels of OGFr within two weeks of becoming hyperglycemic. At weeks 2 and 3, optical density levels were 22 to 30 optical density units, but at weeks 4 and 5, OGFr expression increased to greater than 60 optical density units in both the T1D and T1D-INS groups, representing a greater than 3-fold increase over Normals within 2 weeks. Thus, insulin treatment did not reduce corneal epithelial levels of OGF or its receptor in diabetic animals compared to uncontrolled diabetic rats.

OPEN IN VIEWER

Figure 5.

OGFr expression in corneal epithelium. (a) Photomicrographs of intact corneal epithelium at two, three, four, and six weeks following hyperglycemia in T1D, T1D-INS, and Normal male rats. Tissue stained with only secondary antibody is shown in the insert at week 6. Images were taken at 10× magnification. (b) Tissue images were scanned with a confocal microscope and semi-quantitatively evaluated for OGFr expression using Image J software. Two-factor ANOVA revealed a significant interaction between condition and time, P < 0.01, with specific differences between conditions noted at four and six weeks as determined by one-way ANOVA (P < 0.001). Values represent means ± S.E.M. for 15–20 readings collected from at least three tissue specimens. Significantly different from Normal levels at P < 0.001 (***). (A color version of this figure is available in the online journal.)

Serum levels of OGF, OGFr, and CD10

Serum OGF levels for both controlled and uncontrolled hyperglycemic rats were comparable to Normal values on weeks 2 and 3 after injection of STZ (Figure 6(a)). Two-factor ANOVA revealed an interaction between condition and time (P < 0.01), and subsequent ANOVA and Newman–Keuls tests demonstrated that there were substantial changes in serum OGF levels during the eight-week examination period. However, between three and four weeks, OGF levels doubled and were significantly elevated at weeks four, five, six, and eight for T1D rats and at four, five, and six weeks for the T1D-INS rats ranging from 49 to 57 pg/mL relative to normal values. Baseline or normal values ranged between 21 and 26 pg/mL for the entire eight-week experimental period.

OPEN IN VIEWER

Figure 6.

Serum levels of OGF (a) and OGFr (b) collected weekly following hyperglycemia in T1D, T1D-INS, and Normal male rats. Histograms represent means ± S.E.M. for values from multiple ELISA tests. No significant interaction (condition × time) was noted for serum levels of OGFr; however, one-way ANOVA and Newman–Keuls post hoc analyses at three, four, and six weeks revealed significant differences from Normal levels at P < 0.05 (*), P < 0.01 (**), and P < 0.001 (***).

OGFr values were 1000-fold greater than OGF levels for male rats in all conditions reaching ng quantities (Figure 6(b)). The mean OGFr levels for Normal males ranged from1.8 ng/mL to 3.2 ng/mL over the eight-week period. Two-way ANOVA analyses did not indicate a significant interaction between condition and time, but a main effect was noted. At all-time points, serum OGFr values were less than respective Normal values and reached significance on weeks 3 and 8 for the T1D group and weeks 3, 4, and 8 for T1D-INS male rats. Serum OGFr in the T1D-INS groups was less than 1 ng/mL across all eight weeks.

Analyses of CD10 in serum at three, four, and eight weeks post STZ revealed that mean CD10 values did not differ and ranged between 0.1 pg/µL and 1.2 pg/µL (data not shown).