Ultrastructural Features of Retinal Capillary Basement Membrane Thickening in Diabetic Swine

Abstract

Purpose

To evaluate the ultrastructural features of retinal capillary basement membranes (BMs) in a swine model of type 1 diabetes.

Materials and Methods

Yorkshire pigs were rendered diabetic with streptozotocin and dyslipidemic with a high fat and cholesterol diet. At 18, 26 and 32 weeks of diabetes, the retina sections within 3 disc diameters from the optic disc were examined under transmission electron microscope to evaluate the ultrastructural features of retinal capillary BM. For measurement of the thickness of these capillary BMs, digital morphometric analysis was adapted.

Results

Diabetic swine had significantly thicker retinal capillary BMs compared to controls. Pigs that sustained diabetes for longer periods or experienced severe diabetes tended to have thicker BMs. Those pigs which did not maintain glucose levels above 200 mg/dL did not demonstrate thicker retinal capillary BMs. Characteristic ultrastructural features of diabetic vasculopathy were observed including rarefaction as an early stage of Swiss cheese cavitation, lamellation with multiplication of electron dense layers, and fibrillar materials within capillary BM.

Conclusions

Diabetic Yorkshire pigs develop characteristic features of an early retinal microangiopathy fairly rapidly and may serve as a higher order animal model for basement membrane studies of type 1 diabetes.

INTRODUCTION

Diabetic retinopathy (DR) is the leading cause of new cases of legal blindness in the working population in Western countries.¹ Loss of retinal function occurs due to vascular and neural dysfunction in DR. Diabetic microvascular pathology is a traditional development in DR. Clinically, microaneurysms, retinal hemorrhage, cotton wool spots and neovascularization represent classic markers of retinal vascular dysfunction. Prior to these clinical manifestations, the histological changes seen in early DR include thickened capillary basement membrane (BM), pericyte loss or ghost remnants, acellular capillary structures, endothelial proliferation and microaneurysm formation. While capillary BM thickening is considered as an important manifestation of DR, Ashton observed that it is a pathological event not confined to DR but may also observed in advancing age, hypertensive retinopathy, and Coats’ disease.² Since it is the most common and consistent feature of DR, however, BM thickening has been adopted as an early marker in experimental DR and has been measured to assess the effect of various treatments.^3–5

As some animals develop similar features of human DR, a variety of experimental animals have been studied including the rat, mouse, dog and monkey. While rodent models may be useful to investigate pathologic mechanisms, larger diabetic animal models are highly valuable because of their use in bridging the gap from the results of small animal studies to clinical applications in human beings. The dog develops morphologically indistinguishable lesions from those of early DR in humans; however, it takes up to 5 years to develop noticeable lesions in bovine somatotropin or alloxan injected dogs.⁶ The retinal vascular structure of monkeys is the most similar to that of humans, but no clinical sign of DR was noticed for 4 to 13 years unless the monkeys were hypertensive and retinal microaneurysms were infrequently found in alloxan diabetic monkeys after 10–15 years of diabetes.^7–8

The pig has been increasingly used for biomedical research because of the morphological and physiological similarities between porcine and human organs, especially the skin, cardiovascular system, gastrointestinal tract, and the urinary system.⁹ The pig also develops atheroma in a relatively short period in the hyperglycemic setting and is a useful model in cardiovascular research including diabetic macrovasculopathy.^10–11 Here, we investigated the potential role of the pig as a large animal model of DR, using electron microscopy (EM) to examine ultrastructural features and measuring the thickness of retinal capillary BM with modern digital morphometric analysis.

MATERIALS AND METHODS

Animals and induction of diabetes

Eight Yorkshire pigs (19.8kg ± 1.9) were purchased from Animal Biotech Industries, Inc (Danboro, Philadelphia, USA) and induced diabetes in five animals by intravenous injections of filter-sterilized streptozotocin (STZ; Sigma, St. Louis, Missouri, USA; 50mg/kg in citrate buffer, pH 4.5) on three consecutive days. Diabetes was confirmed by measuring fasting blood glucose was monitored on day seven and monitored monthly thereafter. Two out of the five diabetic pigs developed diabetic ketoacidosis, necessitating the chronic administration of long-acting insulin, insulin glargine (2U daily). All the diabetic pigs were rendered dyslipidemic with a high fat diet containing 1.2% cholesterol and 15% lard (Harlan Teklad, Madison, Wisconsin, USA), and the three control pigs were given a normal diet.¹⁰ The pigs were sacrificed at 18, 26 and 32 weeks post-STZ treatment with an injection of Euthasol (pentobarbital; 100–120 mg/kg/intravenous, Henry Schein, Melville, New York, USA). All experiments were performed in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee at Columbia University.

Transmission electron microscopy

The enucleated eye was fixed in the half-strength Karnovsky’s fixative (Electron Microscopy Sciences, Hatfield, Pennsylvania, USA). A 2 mm diameter trephine (Katena products, Inc., Denville, New Jersey, USA) was used to cut out the posterior retina within 3 diameters from the optic disc and processed by standard techniques of EM. To minimize variability, capillaries in nerve fiber layer and ganglion cell layer which have relatively thicker BM were excluded, and only the capillaries in the inner nuclear layer and outer plexiform layer were included for the morphometric analysis. The EM photos were digitally taken of all the capillaries encountered in those regions at the original magnification between 15,000and 30,000 (JEOL 100S equipped with digital camera, Peabody, Massachusetts, USA).

Morphometric measurement of BM width

For quantitative measurement of BM thickness, Photoshop® (version 7.0, Adobe®, San Jose, California, USA) and ImageJ (National Institute of Health, Bethesda, Maryland, USA) software were employed. The inner and outer boundary lines of the BM were drawn without any feathering edges and the area of outer BM outlined were colored using the paint bucket tool in the Photoshop software (Fig 1). The pixels of the colored BM area were counted using the histogram tool and then the number of pixels was converted into the metric unit (nm²) using the conversion constant calculated from the scale bar. The length of BM was measured along the center of BM in the ImageJ software. Finally, the average of outer BM thickness in each capillary was calculated by dividing the area by the length of BM. A minimum of 6 capillaries in the retinal layers described above were determined in each eye.

Figure 1. Digital morphometric analysis of basement membrane thickness.

For quantitative measurement of BM thickness, the inner and outer boundary of the BM were drawn (A) and the area of outer BM between the glial cells and pericyte or endothelium (gray color) was measured digitally (B). The average outer BM thickness was calculated by dividing the area by the length of the outer BM.

Statistical analyses

Analysis of variance (ANOVA) and Tukey post hoc multiple comparison tests were run to compare the BM thickness of pigs (SPSS, SPSS Inc., Chicago, IL, USA).

RESULTS

Serum glucose levels

The average fasting glucose levels of diabetic pigs were between 248 and 436 mg/dL, which were 4 to 7 fold higher than basal levels. While the fasting glucose levels of pig A, B, and D were well maintained above 200 mg/dL after STZ injection, the glucose levels of pig C and E fell below 200 mg/dL at least one time during the follow-up (Fig 2A). Both pig B and pig D experienced diabetic ketoacidosis within 1 month of the STZ injections, requiring chronic insulin therapy. However, the dose of insulin administered (2U daily) prevented ketoacidosis without attaining euglycemia.

Figure 2. Fasting glucose level and body weight change of diabetic pigs.

(A) While the fasting glucose of pig A, B, and D were well maintained at high level after STZ injection, the glucose levels of pig C and E were fluctuating below 200 mg/dL at least one time during the follow-up.

(B) The diabetic pigs B and D which experienced the ketoacidosis gained the smallest body weight during the follow-up, while the moderate diabetic pigs C and E gained body weight more than twice that of the severe diabetic pigs B and D.

Body weight change

The STZ-injected Yorkshire pigs showed a diminished growth trajectory, gaining an average of 0.59 kg/week (range: 0.12 – 1.08 kg/week, Fig 2), compared to the 5.49 kg/week seen in normal control pigs. In particular, the pigs (B and D) who experienced diabetic ketoacidosis demonstrated the smallest weight gain (0.12 kg/week and 0.37 kg/week, respectively) (Fig 2B).

Lens changes

All diabetic pigs showed a certain degree of cataract formation. The pig A at 32 week diabetes had mature cataracts with no view of fundus because of their severity. The eyes of pigs B and D that experienced ketoacidosis had denser nuclear cataracts compared to the age-matched diabetic pigs C and E which did not experience ketoacidosis.

Basement membrane width

The mean retinal capillary BM widths of pigs A, B and D were significantly increased compared to those of three control pigs (p<0.001, <0.001, <0.01 respectively). Pigs C and E that had fluctuating hyperglycemia did not demonstrate a significant change in thickness of capillary BM (Fig 3). These results are summarized in table 1.

Figure 3. Thickening of retinal capillary basement membrane in diabetic pigs.

The mean retinal capillary BM widths of pigs A, B and D were significantly increased compared to those of controls after 32, 26, and 18 weeks of hyperglycemia, respectively. Pigs C and E that had variable hyperglycemia did not demonstrate a statistically significant change in thickness of capillary BM.

Table 1.

The effects of hyperglycemia in streptozotocin injected pigs and controls

	Pig A	Pig B	Pig C	Pig D	Pig E	Controls
Duration of diabetes	32 weeks	26 weeks	26 weeks	18 weeks	18 weeks
Average glucose level (mg/dL)	319	385	248	436	305	68
Ketoacidosis		yes		yes
Average body weight gain (kg/week)	0.52	0.12	1.12	0.37	1.08	5.49
Hyperglycemic cataract	Mature	Moderate nuclear	Minimum	Moderate nuclear	Mild nuclear
Thickness of BM	121.5 ± 21.8 (131.2%)	124.8 ± 19.1 (134.8%)	88.1 ± 6.6	117.1 ± 32.3 (126.5%)	116.1 ± 17.4	92.6 ± 15.0
Number of capillaries	26	16	6	18	7	25

Ultrastructural Features within Thickened BM

Characteristic ultrastructural features of diabetes were observed in capillary BM of diabetic pigs. The EM revealed intervening accumulation of acellular irregular electron dense materials within the rarefaction created between the BM replications (white arrows in Fig 4A,B). The capillary BM which had fibrillar materials inside in a parallel alignment extended outwards toward adjacent cells (black arrows in Fig 4A,C). BM lamellation with abnormal multiplication of electron dense layers was also observed, leading to thickening of BM in affected regions (Fig 4D,E).

Figure 4. The ultrastructural features of retinal capillary basement membrane in diabetic pigs.

The EM of capillary from diabetic pig A reveals rarefaction with irregular electron dense materials splitting the BM, which appears to be an early stage of Swiss cheese vacuolization (white arrows in A, magnified in B). BM thickening with fibrillar materials extending outwards to adjacent cells was also observed (black arrows in A and magnified in C). The capillary from diabetic pig D demonstrated lamellation, a multiplication of electron dense layers leading thickening of BM (arrows in D and magnified in E).

DISCUSSION

As streptozotocin is particularly toxic to the beta cells in pancreas, it is been widely used, along with alloxan, to chemically induce diabetes in variety of animal models. STZ-injected dogs develop thickened BM after 4 to 5 years of diabetes. EM studies on the STZ-injected rat have revealed thicker BM after 12 months.^12–13 The response to STZ-induced hyperglycemia seems more variable in the mouse depending on the strain. While diabetic BALB/c mice had thicker BM after 20 months, STZ injected C57Bl/6J mice did not have any change in the BM, even after 18 months.^14–15 Yucatan miniature pigs, considerably smaller than Yorkshire swine, had thicker BMs at 20 week after alloxan treatment.¹⁶

The Yorkshire swine model of STZ induced diabetes was established by Gerrity as a diabetic macroangiopathy model and demonstrates accelerated atherosclerosis in diabetes.¹⁰ In this model, pancreatic β-cells are reduced to 10% of the normal value at 2 weeks and subsequently some of them regenerate, recovering to 19% of normal levels at 20 weeks. As described, the diabetic swine is a type 1 model but also shares several features with type 2 diabetes, such as usually not requiring insulin treatment and demonstrating a high triglyceridemia.

In our study, Yorkshire swine showed potential use as a larger order diabetic microangiopathy model. BM thickening was evident with several important ultrastructural features seen in human beings as early as 18 weeks after STZ treatment. The results also indicate that a pig with longer period of diabetes, more severe hyperglycemia, less weight gain and advanced cataract developed a highly thickened retinal capillary BM. Though Pig B and Pig D had shorter period of diabetes (26 weeks and 18 weeks, respectively) than Pig A (32 weeks), these animals had the most significantly thickened BM. These pigs had severe diabetes, gained less weight, had moderate cataract and were treated with insulin because of diabetic ketoacidosis. On the other hand, Pig C and Pig E, in which BM was not thickened after 26 weeks and 18 weeks of hyperglycemia respectively, did not maintain consistent levels of hyperglycemia above 200 mg/dL and gained body weight more than twice that observed in the Pig B or Pig D. Consistent with the less severe nature of their diabetic state, their eyes demonstrated minimal cataract formation. As the swine erythrocyte is highly impermeable to glucose, the glycated hemoglobin such as HbA1C does not reflect the overall hyperglycemia very well in diabetic pigs.^17–18 Instead the average weight gain per week and cataract formation may well be alternative indicators of the severity of diabetes.

Kurtz et al’s observed the importance of the neighboring cell’s contribution to BM maintenance and showed that the BM is being continually removed by depolymerization and replaced, always on the epithelial side not endothelial side in the kidney.¹⁹ Ashton also emphasized that the basal lamina of the pericyte does not usually share in the thickening of the whole basal lamina in DR.² The basal lamina thickening in retinal capillaries has been attributed to glial cells, implying a disturbance of Muller cell metabolism in diabetes.^{2, 13, 20–21} In this study, we measured only the outer basal lamina which exists between the glial cells and pericyte or endothelium, excluding the inner basal lamina, as the outer basal lamina is a more reliable indicator of BM thickness in DR and shows more dramatic change associated with the metabolic activity of Muller cells.

There are two general patterns of BM thickening. First, the BM can become thicker homogeneously which makes it indistinguishable from original BM. Second, a variety of electron dense materials can accumulate, usually consisting of large or small granules or by microfibrillar material or cellular debris. Interestingly, this diabetic swine model developed not only homogenous thickening in BM but also unique ultrastructural features within the BM, such as rarefaction (Fig 4A,B), lamellation (Fig 4D,E) and accumulation of fibrillar materials (Fig 4A,C). The Swiss-cheese vacuolization and lamellation in DR has been well described in the human eye and possibly results from a disturbed ratio of the extracellular matrix components.³ Lamellation is probably the result of repeated injury and regeneration, creating newer thin layers of BM materials that will lend a lamellated pattern to the thickened BM. In animal models, such features had not been typically observed as in humans, though some diabetic animals such as galactose-fed rats at the age of 15 or 21 months demonstrated these unique characters.^22–24 An encouraging sign of this diabetic Yorkshire pig model is the development of these features between 18 and 32 weeks of hyperglycemia, quite an early time point for a large animal model.

The structure of the retinal vascular system in pig is very similar to that of the human.²⁵ The pig usually has four major retinal arteries, two to four layers of capillaries, and a periarterial capillary-free zone. On the other hand, a few differences from human retinal vasculature should be emphasized. There is no fovea or central capillary free area, and the retinal capillaries of pig have much less pericytes than endothelial cells like dogs, while human retinal capillaries have about equal number of pericytes and endothelial cells.

In this study, we showed that the STZ-injected Yorkshire pig displayed thickened retinal capillary BM with several ultrastructural features, such as lamellation and rarefaction within BM. As it develops ultrastructural features in a relatively short-term period, the STZ-injected swine holds promise as a large animal model of diabetes, and further studies of classic histologic features of diabetic retinopathy should be pursued. This swine model may aid in improving our understanding of diabetic vasculopathy and permit the investigation of new treatments identified as promising in small animal studies prior to clinical trials in humans.