posterior and anterior permeability defects? morphologic

Posterior ond Anterior Permeability Defects?Morphologic Observations onStreptozotocin-treated Rats

Ingolf H. L. Wallow

Structural abnormalities of the blood-ocular barrier were examined in streptozotocin (STZ)-treatedhyperglycemic rats, after 9 days, 6 months, and 10 months' duration of "diabetes," and in normo-glycemic control animals using the horseradish peroxidase tracer technique combined with light andelectron microscopy. The most frequent abnormalities consisted of small areas of diffuse dense stainingby the tracer of (1) the retinal pigment epithelium and (2) the nonpigmented ciliary epithelium.Pigment epithelium abnormalities occurred occasionally in both groups of animals with approximatelyequal frequency and extent. Ciliary body abnormalities occurred also in both groups, but were frequent;statistically, the probability of these changes was not significantly different between the two groups.At the ora serrata, tracer escape was present through the retinal pigment epithelium into subretinalspace and retina. Retinal vascular leakage occurred rarely and may be related to tracer toxicity ratherthan hyperglycemia. Thus, using the HRP method, we cannot confirm the claim that sustained STZ-induced hyperglycemia causes breakdown of the blood-retinal barrier in the rat. Invest OphthalmolVis Sci 24:1259-1268, 1983

Streptozotocin (STZ)-induced hyperglycemia inrats has been used as an experimental model to studyalterations of the blood-retinal barrier (BRB).1"6 Vit-reous fluorophotometry has been employed to mea-sure such alterations.1"4 Increased levels of vitreousfluorescence were reported in three studies1"3 and in-terpreted as evidence for a breakdown of the barrier.Recovery of barrier function was ascribed to insulinadministration2 and to pancreatic islet transplanta-tion.3

Some morphologic studies have suspected the ret-inal pigment epithelium as a site of barrier break-down.5*6 Fluorescein produced large areas of relativelyuniform pigment epithelial staining5 while horserad-ish peroxidase (HRP) showed no difference betweenhyperglycemic and control rats in one study5 but inanother, showed several types of leaky RPE lesionsalmost exclusively in hyperglycemic animals.6 Bothstudies agreed that the blood-retinal barrier to HRPremained unaltered at the level of the retinal bloodvessels, yet others described increased permeability

From the Department of Ophthalmology, University of Wis-consin, Madison, Wisconsin.

Supported in part by NIH research grant 5 R01 EY01634 andby NIH Research Career Development Award 5 K04 EYOO135.

Submitted for publication: October 26, 1981.Reprint requests addressed to Ingolf H. L. Wallow, MD, De-

partment of Ophthalmology, Clinical Science Center, 600 High-land Avenue, Madison, WI 53792.

to HRP of retinal capillaries in STZ-hyperglycemicrats7 and in alloxan-diabetic dogs.8

We report here observations in the acute, inter-mediate and chronic stages on the distribution ofblood-borne HRP in retina and ciliary body of ratsmade hyperglycemic with STZ, and in appropriatecontrol animals.

Materials and Methods

Animals

A total of 35 male hooded rats were studied at oneof three intervals following the induction of hyper-glycemia by STZ injection: (A) 9 days (acute stage),(B) 6 months (intermediate stage), and (C) 10 months(chronic stage). Eighteen rats received the STZ injec-tion and became hyperglycemic while 17 of the ratsserved as controls. Fourteen of the 17 control ratsnever received STZ, three received one dose of STZ,but remained normoglycemic.

STZ was administered as a freshly prepared 10%solution in citrate buffer at pH 4.5 in a dose of 65mg/kg BW by injection into the femoral vein aftera 24-hr fast.

All 18 treated rats remained consistently hypergly-cemic and glucosuric. Their blood glucose was mea-sured periodically by the Eyetone-dextrostix systemwith the result occasionally monitored by the hexo-kinase enzymatic endpoint method. Urine sugar and

0146-0404/83/0900/1259/$ 1.30 © Association for Research in Vision and Ophthalmology

1259

Downloaded From: http://iovs.arvojournals.org/pdfaccess.ashx?url=/data/journals/iovs/933116/ on 04/11/2018

1260 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / September 1983 Vol. 24

Table 1. Nonfasting blood glucose levels and weight change in STZ-treated hyperglycemic and innormoglycemic control rats after 9 days (group A), 6 months (group B), and 10 months (group C). Meanvalues are followed by standard deviations of the mean

Group

AHyperglycemicControls

BHyperglycemicControls

CHyperglycemicControls

No. ofanimals

6

8

4

4

8

5

Nonfasting blood sugar values mg/dl by

Colorimetry

Before STZ

140 ±21

201 ± 46

106 ± 15

100 ± 14

169 ± 22

180 ± 14

After STZ

577 ± 144

685 ± 97

736 ± 50

Hexokinase

618 ± 179

170 ± 15

—

:

Weight loss* (—) or gain (+),g, at end of experiment

(-) 58 ± 14

(+) 26 ± 32

(-) 17 ±60

(+) 106 ± 31

(+) 62 ± 51

(+)212± 53

* Measured against initial body weight at time of STZ-injection/assignment to control group.

acetone were measured daily. Test and control ani-mals in each group were housed and fed together(Table 1). Ether anesthesia was used throughout.

Group A consisted of 14 rats approximately 4months old weighing between 390 and 450 g. SixSTZ-treated animals became consistently hypergly-cemic and glucosuric. They received no insulin. Eightanimals served as normoglycemic controls and in-

cluded one that had been ineffectually injected withSTZ. This rat is treated separately in the morphologicresults section (Table 2).

Group B consisted of eight rats also about 4 monthsold and weighing between 350 and 500 g. Four STZ-treated animals became consistently hyperglycemicand glucosuric. They received 4-5 units of NPH in-sulin weekly, which was enough to insure sustained

Table 2. Frequency of tissue blocks with ocular barrier abnormalities per number of tissue blocks studied bylight microscopy in different locations of the eyes of STZ-treated and control rats at 9 days (A), 6 months(B), and 10 months' duration (C) of experimental condition

Group

STZ

AControl*

STZ

Control

STZ

CControl§

RPE*

Posterior+ midperiph.

#

0/50(1/50)

4/87

2/47(2/47)

0/33

1/63(6/63)

5/35

%

0(2.0)

4.5

4.3(4.3)

0

1.6(9.5)

14.3

Anterior

#

1/26

0/54

0/17

0/11

0/29

2/16

%

3.8

0

0

0

0

12.5

Location of tissue blocks

Retina

Posterior+ midperiph.

# %

0/45 0

2/85 2.3

1/43 2.3

0/35 0

0/60 0

0/35 0

Anterior

#

0/21

0/52

1/17

0/10

0/28

0/18

%

0

0

5.8

0

0

0

Ora\

#

16/27

25/45

11/13

2/5

14/31

18/26

%

59.2

55.5

84.6

40.0

45.1

69.2

Ciliary body]

#

27/33

31/51

7/11

5/8

33/47

29/35

%

81.8

60.7

63.6

62.5

70.2

82.8

* RPE abnormalities listed here are first those that consisted of diffusecytoplasmic staining by the tracer HRP. Listed in brackets are changes offocal RPE proliferation ("group 3 lesions").

t Ora and ciliary body were processed within the same tissue blocks sothat favorably oriented sections from these showed both areas.

% This group of 8 "control" animals contains 1 STZ-treated but normo-

glycemic rat, which showed no RPE or retinal vascular barrrier abnormal-ities, but frequent ora and ciliary body changes, as the other controls.

§ This group of 5 "control" animals contains 2 STZ-treated but normo-glycemic rats, which showed no RPE or retinal vascular barrier abnormal-ities, but frequent ora and ciliary body changes, as the other controls.


No. POSTERIOR AND ANTERIOR PERMEABILITY DEFECT5 /WQI IOW 1261

Fig. 1. Retina of rat, 9 days following streptozotocin (STZ) treatment. Horseradish peroxidase reaction product diffusely and denselystains cytoplasm of individual or groups of RPE cells (arrows). Tracer does not extend into subretinal space around photoreceptor outersegments (OS); ONL = outer nuclear layer (original magnification X1100).

hyperglycemia while preventing hyperglycemic orketotic death. Four animals, not injected with STZ,served as normoglycemic controls.

Group C consisted of 13 rats approximately 2months old weighing between 200 to 340 g. EightSTZ-treated animals became consistently hypogly-cemic and glucosuric. They received NPH insulinapproximately every 12 days. Five animals served asnormoglycemic controls and included two that hadbeen ineffectually injected with STZ. These two aretreated separately in the morphologic results section(Table 2). In groups B and C the last insulin injectiongiven was 4 and 5 units, respectively, 2 days prior tokilling.

Tracer Technique and Tissue Processing

The permeability of the BRB to the protein tracer,HRP, was studied9 in one eye of each of the animalsof group A, B, and C, at 9 days, 6 months and 10months, respectively, following STZ injection. Inboth normal and "diabetic" animals occasional vas-cular leakage was present in groups A and B (seeTable 2) and was attributed to an HRP-induced in-crease of blood vessel permeability as previously ob-served in rodents by others.10 Thus, in group C, 5min prior to HRP injection, diphenhydramine hy-drochloride and methysergide maleate, each in a doseof 1 mg/kg were given intravenously in order to pre-vent an HRP-induced increase of blood vessel per-meability. For all groups, HRP type VI was freshlydissolved in normal saline in a concentration of 180mg/ml. The femoral vein was surgically prepared and180 mg/kg of tracer were injected. The preparationsite was then sutured. From 10 to 20 min after in-jection one eye of each animal was enucleated,

promptly fixed in 4% glutaraldehyde in cacodylatebuffer at pH 7.4 and opened coronally at the limbus.Fixation was continued for 2 hrs at 4 C.

Comparable tissue samples were taken from alleyes, consisting of posterior retina (adjacent to theoptic nerve), intermediate retina (midway betweendisc and ora), ora serrata, and ciliary body. Thesewere embedded in agar, sectioned into 50-100 /xmthick slices, and processed for peroxidase activity.9

Subsequently, the tissue was postfixed for 1 hr in1% OSO4 solution in S-collidine buffer, stained in theblock with 2% uranyl acetate for 1 hr, dehydrated ingraded alcohols and embedded in an epoxy resin.Sections were cut 1.5 /im thick; half were stained withtoluidine blue, the other half left unstained. Theywere examined by light microscopy at 400-1000X magnification for evidence of tracer leakage. Fromfour to seven tissue blocks were made from each tis-sue sample and four to five sections were cut andevaluated from each block. More than 3,000 sectionswere reviewed. Representative areas showing leaks(specified under Results) were also thin sectioned,stained with lead citrate and examined in an electronmicroscope.

Results

The most interesting abnormality found was dif-fuse dense cytoplasmic staining by the tracer HRPof a single or group of adjacent cells in the retinalpigment epithelium (RPE) (Figs. 1, 2) or nonpig-mented ciliary epithelium (NPCE) (Figs. 3, 4).

Table 2 shows that no consistent differences in fre-quency, location or extent of these abnormalities werefound between normal and hyperglycemic animalsregardless of the initial age (4 months in group A and


Fig. 2. Retina of control rat, 9 days following entry into experiment. Tracer reaction product is diffusely present within cytoplasm ofRPE cell at center. APV = apical vill; OS = outer segments. Adjacent normal RPE cells show tracer only within a few vesicles (arrow) nearthe prominent basal infoldings of the plasma membrane (original magnification X6400).

Fig. 3. Pars plicata of ciliary body of rat, 6 months following STZ treatment. One cell (arrow) of the nonpigmented ciliary epitheliumis diffusely stained by dark tracer reaction product (XI100).


No. 9 POSTERIOR AND ANTERIOR PERMEABILITY DEFECTS/Wollow 1260

Fig. 4. Electron micrograph of another tissue block of the same animal shows comparable area. Tracer has leaked out of the central vascularcore (black) and stained basal, lateral and apical spaces around the pigmented ciliary epithelium (PE). At the level of the nonpigmented ciliaryepithelium (NPE) tracer is seen within (a) numerous vesicles (lower arrow) of normal appearing cells, and (b) as diffuse dense cytoplasmicstaining of several other cells (upper arrows) (X7000).


1264 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / September 1980 Vol. 24

Fig. 5. STZ-treated rat, 10 months in experiment. Two cell layers are seen, hypopigmented RPE (A) in contact with Bruch's membraneand an irregularly hyperpigmented group of cells (B) extending away from Bruch's membrane into the subretinal space. Tracer did notpenetrate the basal hypopigmented layer. This was the most extensive lesion of its kind (original magnification XI100).

B vs. 2 months in group C) or weight (350-500 g ingroups A and B vs. 200-340 g in group C). Also nodifference was recognized between insulin-treated(groups B and C) and noninsulin-treated hypergly-cemic animals (group A), or between streptozotocin(STZ)-treated hyperglycemic animals and a few STZ-treated animals that remained normoglycemic. Stain-ing of the RPE occurred occasionally, staining of theNPCE was very frequent.

At times another RPE change was encounteredconsisting of one or several hyperpigmented cells lo-cated on top or in between hypopigmented RPE cellsthat were connected laterally to unaltered RPE (Fig.5). Tracer did not stain the RPE cells or the subretinalspace. Such changes were similar to the "group 3 le-sions" described by others.6 Table 2 shows that theywere present in nine blocks of six STZ-treated ani-mals. They were not observed in control animals.

At the ora serrata tracer extended into the subret-inal space and into the intercellular spaces of the neu-rosensory retina (Figs. 6, 7). Frequently, a few adja-cent cells of the nonpigmented ciliary epitheliumwere diffusely stained. Tracer leaks were seen in everyanimal, hyperglycemic or control, with satisfactorytissue sections (Table 2). In groups A and B, but notin group C, several retinal blood vessels showed dif-fuse impregnation of their walls by tracer reactionproduct with extension into the paravascular tissuespaces. One STZ-treated and two control animalswere involved (Tables 2 and 3).

The probability of barrier abnormalities at ora andciliary body was not significantly different betweenSTZ-treated and control animals when tested in the

following manner: For each of the three groups ofrats (A, B, and C) a chi-square statistic with Yatescontinuity correction was calculated, and the respec-tive P values were determined. For the ora the P val-ues for group A, B, and C were 0.95, 0.19, and 0.12,for the ciliary body they were 0.07, 0.67, and 0.29.

Electron microscopy confirmed our light micro-scopic observations in five blocks with diffuse, denseRPE staining, in two blocks with "group 3 lesions,"in four blocks each with ora and ciliary body changesand three blocks with tracer escape around retinalblood vessels. The accuracy of light microscopic ex-amination was tested in the following way: four areasof diffuse, dense RPE staining in four different blocksfrom three different animals appeared very mild, eg,on light microscopic observation, there was only aquestionable mild cytoplasmic densification as com-pared to adjacent normal cells. Electron microscopicobservation of these cases failed to confirm diffusecytoplasmic staining with HRP. Therefore, such areaswere not included into Tables 2 and 3.

We found 13 tissue blocks with dense diffuse RPEstaining in nine animals (four STZ-treated, five con-trols). In seven blocks the retina was attached; in twominimally detached by artifact, but present and eval-uated; in four blocks the retina was artifactitiouslydetached and not available for evaluation. Tracer es-cape around retinal blood vessels occurred in fourtissue blocks of three animals.

Discussion

The horseradish peroxidase (HRP) method of ex-amining the blood-retinal barrier in STZ-hypergly-



Fig. 6. Ora serrata of rat, 6 months following STZ treatment. Dark tracer extends into subretinal space (SR) and into outer nuclear layer(ONL). Diffusely tracer-stained cells of the nonpigmented ciliary epithelium are also seen (arrow) (original magnification XI100).

cemic and in normal control rats demonstrated ab-normal staining of the retinal pigment epithelium(RPE) and of the nonpigmented ciliary epithelium(NPCE). Furthermore, tracer leakage was found reg-ularly at the ora and, rarely, around retinal bloodvessels.

HRP was given circulation times of 10 to 20 minproviding ample opportunity for the tracer to escapethrough barrier defects by orthograde flow. Times of10 to 20 min were considered together as "long" incontrast to "short" times (less than 1 min). The latterallow orthograde escape of tracer through an openbarrier yet limit retrograde tracer diffusion into thetissues that can complicate interpretation. We did notemploy short times since the long ones failed to yieldconsistent results.

Diffuse Cytoplasmic Staining of Cells by HRP

In the RPE this has been noted in both STZ-treatedhyperglycemic and in untreated normoglycemic con-trol rats.56 In the two studies cited experimental con-ditions and results varied. One group of authors ex-

amined both eyes of young rats, six diabetic and sixcontrol animals, initially weighing 145-185 g, 3 to5 weeks after the beginning of the experiment allow-ing ten minutes tracer circulation time. The authorsfound occasional RPE staining in both diabetic andcontrol eyes and attributed this not to the experi-mental condition.5 Instead, they directed attention tothe fact that staining of the cytoplasm by HRP hasbeen considered by some an artifact of the histo-chemical tracer method.10

The second group of authors6 examined old rats,four STZ diabetic and four control animals, initiallyweighing 400-500 g, 8 days after the beginning of theexperiment, allowing 30 min tracer circulation time.They found variable numbers of tracer stained RPElesions. In tabulating their data they showed RPElesions in six of seven diabetic eyes quantitativelystudied, but in only one of four normal control eyes.RPE lesions in their STZ-hyperglycemic animals dif-fered also in quality from RPE lesions in normogly-cemic rats. In hyperglycemic animals three types oflesions were distinguished: (1) Diffuse dense tracer


1266 INVESTIGATIVE OPHTHALMOLOGY 6 VISUAL SCIENCE / Seprember 1983 Vol. 24

Fig. 7. Electron micrograph of area adjacent to Figure 6. At posterior pars plana (left), tracer (arrows) is present within lacunae betweenthe pigmented (PE) and nonpigmented (NPE) ciliary epithelium. Tracer is seen also within subretinal space (SR) and within intercellularspaces of retina. OS = whirls of outer segment lamellae; ONL = nucleus of photoreceptor cell in outer nuclear layer; V = vitreous. Note alsoportions of diffusely stained cells at transition between pars plana and retina (original magnification X56O0).

staining, (2) RPE necrosis with tracer extensionwithin the subretinal space, and (3) RPE regenerationover degenerating RPE cells. In their control rats onlytype 1 lesions were seen.

Our results are consistent with some of their find-ings, but inconsistent with others. In control rats wealso saw type 1 lesions only. Equivalents to their type3 lesions were found only in STZ-treated animals andconsisted of areas of hypopigmented RPE arrangedin a single line associated with other pigmented cells

underneath this line or on top of this line. On theother hand, we cannot confirm a significantly in-creased frequency of type 1 lesions in STZ animals.Also, type 2 lesions were not found in our material.

Type 1 and type 2 lesions may be seen as stagesof the same process.6 Both indicate cell degeneration.Type 1 lesions may occur initially. They are describednow in three reports in both groups of experimentalanimals, hyperglycemic and normoglycemic. Thus,the lesion seems nonspecific. Only frequency is con-



troversial making further study an affair of samplesize and quantitative morphology.

Type 2 lesions were seen by only one group.6 RPEcells altered enough to permit cytoplasmic floodingby the foreign protein tracer HRP may at some timebreak down further. Tracer could then escape intothe subretinal space, but be demonstrable only iftracer injection, death of the animal, and tissue sam-pling would have occurred at precisely the right timeand location. Further clarification again would de-pend on more quantitative data.

Type 3 lesions were not found in one study5; in theother,6 they were present only in STZ-treated hyper-glycemic rats. We found type 3 lesions also in STZ-treated animals only but see difficulties in acceptingthis as a significant finding. Pathologically, type 3 le-sions represent areas of RPE regeneration. Thesemust have originated from acute changes comparableto those seen in type 1 and type 2 lesions. Why shouldacute changes be found in both control and STZ rats,but evidence of their repair only in the latter?

Diffuse cytoplasmic staining was also seen in theciliary body which is particularly sensitive to changesof its environment."12 In the normal stage HRP en-ters the lateral and apical intercellular spaces of thepigmented epithelium, but passage between the cellsof the nonpigmented layer is prevented by zonulaeoccludentes.13'14 Micro vesicular transport across thecytoplasm bypassing the zonulae occludentes is con-sidered insignificant, and does not amount to tracerdeposition in posterior or anterior chambers. Simi-larly, in our animals, no intercellular passage of HRPwas seen. In all animals some cells of the nonpig-mented ciliary epithelium showed diffuse dense cy-toplasmic staining. Statistically, no significant differ-ence was recognized between hyperglycemic and con-trol animals. A particular situation is present at theora serrata where a physiologic leak of the blood-ret-inal barrier has been described by others in monkey.15

We confirmed, in the rat, the presence of an ora leakpresent in all STZ-treated and control animals withsatisfactory tissue sections.

HRP Leakage from Retinal Blood Vessels

In experimentally diabetic dogs following five yearsof intentionally poor metabolic control, HRP seem-ingly permeated junctions between degenerate en-dothelial cells of retinal blood vessels. The junctionsin normoglycemic control dogs remained tight.8 InSTZ "diabetic" rats, maintained 3 to 6 months, asimilar junctional insufficiency was demonstrated. Itwas also reported that after 2 months, the numberof tracer filled pinocytotic vesicles in the cytoplasmof endothelial cells of retinal capillaries was signifi-

Table 3. Frequency of barrier abnormalities peranimal in different ocular locations of STZ-treatedand control rats at 9 days (A), 6 months (B), and10 months' duration (C) of experimental condition

A

B

C

Group

STZ

Control*

STZ

Control

STZ

Controlf

No. ofanimals

6

8

4

4

8

5

RPEt

1 ( 1 )

3

2 ( 1 )

0

1 ( 2 )

2

Retina

0

2

1

0

0

0

Ora

6

8

4

1§

7§

4§

Ciliarybody

6

8

4

4

8

5

* t See footnotes $, and §, respectively, of Table 2.t First figures give rats with diffuse cytoplasmic staining of the RPE. Listed

in brackets are the animals which had "group 3 lesions" only. In addition,group 3 lesions were present in some STZ-rats that also had diffuse cyto-plasmic staining, ie, in one rat each at 6 and 10 months.

§ In these groups only one, seven, or four animals, respectively, had sec-tions passing through the ora.

cantly higher than those in the control group and inthe diabetic group maintained for only 1 month.7

Both studies included extensive quantitative evalua-tion or morphometric analysis by electron micros-copy.

In two other studies discussed before, quantitativedata based on electron microscopy were not given.5'6

No leakage from the retinal vasculature was shown.Occasional dense diffuse staining of endothelial cellswas noted in both diabetic and control eyes despitepremedication of the animals with antihistamines.5

Our findings in this present study were based onlight microscopic screening. If a blood vessel was notsuspected on light microscopic examination to leaktracer, a specific electron microscopic evaluation ofthe neurosensory retina was not performed. Bloodvessel leakage occurred in two vessels of two controlanimals and in two vessels of one STZ-hyperglycemicanimal. The appearance of leakage was comparableto the description given by others.7 Tracer impreg-nated the vascular basement membranes and filledintercellular tissue spaces around the vessels. We didnot determine the morphologic pathway of leakage.The three animals showing vascular leakage belongedto groups not pretreated with antihistamines (groupsA, B). In the third group (C) where such treatmentwas applied, no retinal vascular leakage was observed.We suspect, therefore, that leakage in our animalswas influenced by HRP-toxicity known to be signif-icant in rats.10 Additional information is needed inorder to more conclusively assess the possible role ofSTZ-hyperglycemia on the permeability of retinalvessels.


1268 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / September 1983 Vol. 24

Key words: blood-ocular barrier, retinal pigment epithelium,retinal blood vessels, horseradish peroxidase, streptozotocin-induced hyperglycemia, rats.

Acknowledgments

Mr. Richard Hibma, BS, and Eunice Scharpf providedvaluable technical assistance.

References

1. Waltman S, Krupin T, Hanish S, Oestrich C, and Becker B:Alteration of the blood-retinal barrier in experimental diabetesmellitus. Arch Ophthalmol 96:878, 1978.

2. Jones CW, Cunha-Vaz J, Zweig KO, and Stein M: Kinetic vit-reous fluorophotometry in experimental diabetes. ArchOphthalmol 97:1941, 1979.

3. Krupin T, Waltman SR, Scharp DW, Oestrich C, Feldman SD,Becker B, Ballinger WF, and Lacy PE: Ocular fluorophotometryin streptozotocin diabetes mellitus in the rat: effect of pancreaticislet isografts. Invest Ophthalmol Vis Sci 18:1185, 1979.

4. Klein R, Wallow IHL, and Ernest JT: Fluorophotometry. III.Streptozocin-treated rats and rats with pancreatectomy. ArchOphthalmol 98:2235, 1980.

5. Kirber WM, Nichols CW, Grimes PA, Winegrad AI, and LatiesAM: A permeability defect of the retinal pigment epithelium.Occurrence in early streptozocin diabetes. Arch Ophthalmol98:725, 1980.

6. Tso MOM, Cunha-Vaz JG, Shih C-Y, and Jones CW: Clini-

copathologic study of blood-retinal barrier in experimental di-abetes mellitus. Arch Ophthalmol 98:2032, 1980.

7. Ishibashi T, Tanaka K, and Taniguchi Y: Disruption of blood-retinal barrier in experimental diabetic rats: An electron micro-scopic study. Exp Eye Res 30:401, 1980.

8. Wallow IHL and Engerman RL: Permeability and patency ofretinal blood vessels in experimental diabetes. Invest OphthalmolVis Sci 16:447, 1977.

9. Graham RC and Karnovsky MJ: The early stages of absorptionof injected horseradish peroxidase in the proximal tubules ofmouse kidney: Ultrastructural cytochemistry by a new technique.J Histochem Cytochem 14:291, 1966.

10. Cotran RS and Karnovsky MJ: Vascular leakage induced byhorseradish peroxidase in the rat. Proc Soc Exp Biol Med 126:557,1967.

11. Tormey JMcD: Differences in membrane configuration betweenosmium tetroxide-fixed and glutaraldehyde-fixed ciliary epithe-lium. J Cell Biol 23:658, 1964.

12. Smith RS: Ultrastructural studies of the blood-aqueous barrier.I: Transport of an electron-dense tracer in the iris and ciliarybody of the mouse. Am J Ophthalmol 71:1066, 1971.

13. Shiose Y: Electron microscopic studies on blood-retinal andblood-aqueous barriers. Jap J Ophthalmol 14:73, 1970.

14. Shabo AL and Maxwell DS: The blood-aqueous barrier to tracerprotein: a light and electron microscopic study of the primateciliary process. Microvasc Res 4:142, 1972.

15. Shabo AL and Maxwell DS: Structural organization of the parsplana-ora serrata transition in the human and monkey eye withemphasis on protein barriers. Lab Invest 29:511, 1973.


posterior and anterior permeability defects? morphologic

Documents