Quantitative Assessment of the Integrity of the Blood–Retinal Barrier in Mice

Abstract

purpose. The purpose of this study was to develop and characterize a quantitative assay of blood–retinal barrier (BRB) function in mice and to determine the effect of several purported vasopermeability factors on the BRB.

methods. Adult C57BL/6J mice were treated with three regimens of increasingly extensive retinal cryopexy and subsequently were given an intraperitoneal injection of 1 μCi/g body weight of [³H]mannitol. At several time points, the amount of radioactivity per milligram tissue was compared in retina, lung, and kidney. Time points that maximize signal-to-background differential in the retina were identified, and the ratio of counts per minute (CPM) per milligram retina to CPM per milligram lung (retina-to-lung leakage ratio, RLLR) or kidney (retina-to-renal leakage ratio, RRLR) were calculated. This technique was then used to compare the amount of BRB breakdown that occurs after intravitreous injection of vascular endothelial growth factor (VEGF), insulin-like growth factor (IGF)-1, prostaglandin (PG) E₁, PGE₂, interleukin (IL)-1β, or tumor necrosis factor (TNF)-α.

results. Twenty-four hours after retinal cryopexy, there was a higher level of radioactivity in treated than in control retinas, and the signal-to-background difference was optimal when measurements were obtained 1 hour after injection of [³H]mannitol. In untreated mice, the RLLR was 0.30 ± 0.02 and the RRLR was 0.22 ± 0.01. Twenty-four hours after one 5-second application of retinal cryopexy, the RLLR was 0.73 ± 0.20 and the RRLR was 0.71 ± 0.23. With increasing amounts of cryopexy, there was an increase in the RLLR and RRLR, so that after two 10-second applications, the RLLR was 1.66 ± 0.31 and the RRLR was 1.47 ± 0.20. Intravitreous injection of VEGF, IGF-1, PGE₁, PGE₂, IL-1β, or TNF-α each caused significant increases in the RLLR and RRLR, but there were some differences in potency and time course. VEGF caused prominent BRB breakdown at 6 hours that returned to near normal by 24 hours. IL-1β also caused relatively rapid breakdown of the BRB, but its effect was more prolonged than that caused by VEGF. There was delayed, but substantial breakdown of the BRB after injection of TNF-α. IGF-1, PGE₂, and PGE₁ caused less severe, relatively delayed, and more prolonged BRB breakdown.

conclusions. Measurement of the RLLR or RRLR after intraperitoneal injection of [³H]mannitol in mice provides a quantitative assessment of BRB function that is normalized and can therefore be compared from assay to assay. Comparison of the extent and duration of BRB breakdown after intravitreous injection of vasoactive substances shows that agents can be grouped by resultant extent and time course of leakage. Additional studies are needed to determine whether this grouping has its basis in shared mechanisms of BRB disruption.

Extravasation of plasma from the vasculature is well tolerated in most tissues of the body but results in dysfunction in the brain and retina. Adaptations of the retinal vessels and retinal pigmented epithelium (RPE), the inner and outer blood–retinal barriers (BRBs) prevent extracellular accumulation of fluid. Perturbation of the BRBs occurs in several disease processes, resulting in macular edema and loss of vision. Macular edema is the most common cause of visual loss in patients with diabetes and constitutes a major public health problem. ^{1 2} It is also a major source of visual loss in other ischemic retinopathies, such as branch and central retinal vein occlusion. ^{3 4} Focal laser treatment is the only treatment that has been demonstrated to have any benefit, and the benefit is very modest. ⁵ There is a great need for new treatments. 

To develop new treatments it is necessary to have good techniques to assess BRB function, to investigate mechanisms of BRB breakdown, and to test efficacy of potential therapeutic agents. Immunohistochemical staining for serum proteins provides a means to evaluate the BRB at both the light and electron microscopic levels in animal models or human pathologic specimens. ^{6 7} This technique has been extremely useful for investigating the site and mechanism of BRB breakdown in several disease processes, ^{6 7 8 9 10 11 12 13} but quantitative assessments are difficult and labor-intensive. Radiolabeled sucrose or mannitol have been used to obtain quantitative assessment of BRB breakdown in rats, ^{14 15} and this model is potentially a valuable system that can be used to compare effects of vasopermeability factors on the BRB and to investigate effects of drug treatments. A similar technique in mice would have the same benefits, with the added advantage of allowing investigations in transgenic and knockout mouse models. In this study, we report a comparison of purported vasopermeability agents on the BRB, with a newly developed and validated technique that allows quantitative assessment of the BRB in mice. 

Materials and Methods

Retinal Cryopexy

Six- to 7-week-old C57BL/6J mice were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. They were weighed and anesthetized with 25 mg/kg ketamine (Fort Dodge Animal Health, Fort Dodge, IA) and 4 mg/kg xylazine (Vedco, St. Joseph, MO). Pieces of dry ice were carved and molded to provide a uniform point on one end and were grasped with forceps. The point of the dry ice was applied to the sclera of one eye and held in contact for exactly 5 or 10 seconds, after which irrigation with balanced salt solution resulted in rapid withdrawal of the dry ice and thawing of the tissue. In preliminary experiments, the optimal intervals were determined for the time between cryopexy and intraperitoneal injection of tracer, [³H]mannitol (New England Nuclear, Boston, MA), and the time between tracer injection and death, followed by rapid isolation of the retina and determination of radioactivity in the retina. In subsequent experiments, 48 mice were divided into three groups (n = 16 in each group). In group 1, cryopexy was applied for 5 seconds to one quadrant of one eye; in group 2, two cryotreatments of 5 seconds were administered, one in each of two quadrants of one eye; and in group 3, two cryotreatments of 10 seconds were administered, one in each of two quadrants of one eye. Twenty-four hours after cryopexy, mice were sedated with methoxyflurane (Schering-Plough, Union, NJ) and given an intraperitoneal injection of 1 μCi/g body weight of [³H]mannitol. Sixty minutes after injection, the mice were sedated with methoxyflurane and killed by cervical dislocation. 

Processing of Retinas

Retinas from both the treated and untreated (control) eyes were rapidly removed by a technique described by Winkler. ¹⁶ Briefly, the posterior portion of the globe was firmly grasped with forceps, and a razor blade was used to cut across the cornea and extrude the lens, vitreous, and retina. Retinas were dissected from lens and vitreous and any RPE that extruded and were placed in preweighed scintillation vials within 30 seconds of death. The thoracic cavity was opened, and the left superior lobe of the lung was removed and placed in another preweighed scintillation vial. A left dorsal incision was made, and the retroperitoneal space was entered without entering the peritoneal cavity. The renal vessels were clamped with forceps, and the left kidney was removed, cleaned of all fat, and placed into a preweighed scintillation vial. All liquid was removed from the vials, and remaining droplets were allowed to evaporate over 20 minutes. The vials were weighed and the tissue weights were recorded. One milliliter of NCSII solubilizing solution (Amersham, Chicago, IL) was added to each vial, and the vials were incubated overnight in a 50°C water bath. The solubilized tissue was brought to room temperature and decolorized with 20% benzoyl peroxide in toluene in a 50°C water bath. The vials were brought to room temperature and 5 mL scintillation fluid (Cytoscint ES; ICN, Aurora, OH) and 30 μL glacial acetic acid were added. The vials were stored for several hours in darkness at 4°C to eliminate chemoluminescence. Radioactivity was counted with a liquid scintillation counter (model 1409; Wallac, Gaithersburg, MD). 

Intraocular Injections

Intraocular injections were performed as previously described with a pump microinjection apparatus (Harvard Apparatus, South Natick, MA) and pulled glass micropipets. ¹⁷ Each micropipet was calibrated to deliver 1 μL of vehicle in response to depression of a foot switch. Mice were anesthetized with 25 mg/kg ketamine and 4 mg/kg xylazine, and pupils were dilated with 1% tropicamide. Under a dissecting microscope, the sharpened tip of a micropipet was passed through the sclera just behind the limbus into the vitreous cavity, and the foot switch was depressed. The following agents were tested: human vascular endothelial growth factor (VEGF; R&D Systems, Minneapolis, MN), human insulin-like growth factor (IGF)-1 (R&D Systems), prostaglandin (PG) E₁ and PGE₂ (Sigma, St. Louis, MO), human tumor necrosis factor (TNF)-α (Sigma), and human interleukin (IL)-1β (Chemicon, Temecula, CA). These agents were selected because previous studies using other techniques in other animals have suggested that they cause breakdown of the BRB. ^{8 12 13 18 19 20 21 22 23 24 25} The left eye was injected with agent and the right eye was injected with vehicle or in some cases was left uninjected. Tracer injections were performed at 6 or 24 hours after injection of 100 U IL-1β; 10⁻⁵ M TNF-α, IGF-1, PGE₁, or PGE₂; or 10⁻⁶ M VEGF. Depending on the results, additional doses and time points were examined. Mice were killed 1 hour after tracer injection, and tissue radioactivity was measured as described earlier. If there was significant BRB breakdown with the initial dose at 6 and 24 hours, then longer and shorter time points were examined and the effects of lower doses were investigated. Intermediate time points were also examined if a more detailed assessment of time course was considered potentially worthwhile for a particular agent. Doses of 0.05, 0.5, 10, and 100 U IL-1β were investigated. 

Analysis of Data

Counts per minute (CPM) per milligram tissue was measured in lung, kidney, treated retina, and untreated retina, retina-lung, retina-kidney, and lung-kidney ratios were calculated. The ratios obtained for retinas treated with a particular agent at a specific concentration and time point were compared with those for untreated or saline-treated retinas by Student’s unpaired t-test for populations with unequal variances. 

Results

Entry of [³H]mannitol into the Retina

In initial experiments, the amount of radioactivity in the retina and the superior lobe of the left lung was measured at several time points after intraperitoneal injection of 1 μCi/g body weight of [³H]mannitol. Radioactivity in the retina was low at all time points, whereas the level was high in the lung at 1 hour and then gradually decreased to retinal levels by 8 hours after injection (Fig. 1) . Based on these experiments, an interval of 1 hour between tracer injection and death was selected for subsequent experiments. 

Validation of the Technique by Measurement of BRB after Different Amounts of Retinal Cryopexy

Retinal cryopexy results in BRB breakdown, and with increasing amounts of treatment there is increasing disruption of the BRB. ²⁶ We used retinal cryopexy to assess the validity of our technique designed to assess BRB function in mice. Comparison of eyes treated with different amounts of cryopexy showed that there was substantial breakdown of the BRB that increased in proportion to the area and duration of treatment (Fig. 2) . The ratios of CPM per milligram retina to CPM per milligram lung (retina-to-lung leakage ratio; RLLR) and CPM per milligram retina to CPM per milligram kidney (retina-to-renal leakage ratio; RRLR) were almost identical. There was significantly more leakage in treated than in control retinas with all treatments and significantly more leakage in eyes treated with two 10-second freezes compared with eyes treated with one 5-second freeze. 

Comparison of Vasopermeability Factors

Figures 3A and 3B show the effect of intravitreous injection of 1 μL 10⁻⁶ M VEGF or TNF-α. At 6 hours after intravitreous injection of VEGF, there was a significant increase in the RLLR and RRLR, which decreased thereafter, but was still significantly greater than control at both 24 and 48 hours after injection. There was no significant increase in the RLLR or RRLR 6 hours after intravitreous injection of 10⁻⁵ M TNF-α, but 24 hours after injection of 10⁻⁶ M TNF-α, a 10-fold lower dose, there was BRB breakdown similar to that seen 6 hours after injection of VEGF. The RLLR and RRLR decreased between 24 and 48 hours, but were still elevated above control levels. Administration of 10⁻⁷ M VEGF or TNF-α did not increase the RLLR or RRLR above control levels at time points ranging from 3 to 48 hours (not shown). 

At 6, 24, and 48 hours after intravitreous injection of 1 μL 10⁻⁶ M IGF-1, PGE₁, or PGE₂, there was no significant increase in the RLLR or RRLR (not shown). Injection of a 10⁻⁵ M dose resulted in significant increases at both 6 and 24 hours with IGF-1 and PGE₂, and also at 72 hours with PGE₂ (Figs. 3C 3D) . With PGE₁, the elevations of RLLR and RRLR were modest, with only the RLLR showing significance at 6 hours and only the RRLR showing significance at 24 hours. 

Injection of 0.5, 10, or 100 U IL-1β resulted in significant elevation of the RLLR and RRLR at both 6 and 24 hours after injection, and injection of 0.05 U resulted in elevation of the RLLR at 24 but not 6 hours after injection (Fig. 3E 3F) . At their optimal doses, VEGF and IL-1β induced the greatest BRB breakdown at 6 hours, and TNF-α caused the greatest disruption of the BRB at 24 hours. 

Discussion

In this study, we developed a new technique for quantitative assessment of BRB function in mice. We used [³H]mannitol as a tracer, because it does not enter cells, and the rate of its accumulation in a tissue is dependent on how rapidly small nontransported molecules enter the extracellular fluid from the vasculature. Hence, it is a measure of the leakiness of the vasculature in that tissue. In rats, intravenous injection of [³H]mannitol has provided reproducible measurements of BRB function. ¹⁴ However, intravenous injections are more difficult in mice than rats, and therefore we investigated the potential use of intraperitoneal injections. The rate at which [³H]mannitol entered the circulation from the peritoneal cavity appeared to be quite constant, because at a given time point after injection of 1 μCi [³H]mannitol, the amount of radioactivity per milligram lung or kidney, two tissues that do not have a blood–tissue barrier, was reproducible. Furthermore, by calculating the ratio of CPM per milligram retina to CPM per milligram lung or kidney, any small variation from one mouse to another in the amount of [³H]mannitol administered or the rate of absorption into the vasculature was negated. Also, comparisons from assay to assay were enhanced, because any small protocol variations, such as the time from tracer injection to death, are controlled for by normalization with the reference tissues. We selected lung and kidney as our reference tissues, because they are separated from the peritoneal cavity and are unlikely to be contaminated by any unabsorbed [³H]mannitol. There was a higher rate of entry of [³H]mannitol into the extracellular fluid of kidney than of lung, with a remarkably consistent difference between the two. The lung-kidney ratio calculated from all our experiments was 0.70 ± 0.02, with a range of 0.68 to 0.78. The same information was provided by the RLLR and the RRLR, with the former consistently slightly higher than the latter. The RLLR and RRLR in untreated retinas were 0.30 ± 0.02 and 0.22 ± 0.01, respectively. This strong concordance bolsters confidence in the assay. Either the RLLR or RRLR would be sufficient for profiling the effect of a vasopermeability factor, but having both increases confidence in the validity of the measurements. 

Previous studies using other techniques have demonstrated that retinal cryopexy causes BRB breakdown that increases with duration and area of treatment. ²⁶ Using our technique, we found that an increase in the amount of retinal cryopexy in mice, resulted in a reproducible, highly significant increase in the RLLR and RRLR, demonstrating that this new technique in mice provides information similar to that provided by well-established techniques for measuring BRB function in rabbits. Although this new technique allows rapid, reproducible measurement of extent of BRB breakdown, it does not provide any localization. Concurrently performing immunohistochemical staining for albumin would allow some assessment of whether breakdown of the inner, outer, or both components of the BRB were responsible for the leakage. 

We used our new technique to assess the effect of intraocular injections of purported vasopermeability factors. Compared with the other agents tested, VEGF caused a relatively rapid and transient disruption of the BRB. IL-1β also caused relatively rapid disruption of the barrier, but it was more sustained. The effects of TNF-α were delayed compared with those of VEGF and IL-1β. IGF-1 and PGE₂ were relatively less potent than VEGF, but at higher doses caused more delayed and sustained disruption of the BRB. Although direct dose–response comparisons between IL-1β and the other agents cannot be made, it appears that IL-1β is an extremely potent vasopermeability agent, as are VEGF and TNF-α, whereas PGE₁, PGE₂, and IGF-1 are less potent. 

These results correlate well with previously reported results using different techniques to assess BRB breakdown. Several studies have suggested that VEGF causes breakdown of the BRB. ^{12 13 18 19 20 21 22} Six hours after intravitreous injection of VEGF in rabbits, there is the opening of a significant number of tight junctions between retinal endothelial cells that is partially reversed by 24 hours after injection. ¹³ That result predicts exactly what was found in the present study: severe BRB breakdown 6 hours after intravitreous injection of VEGF that is reduced 24 hours after injection. Compared with VEGF, TNF-α caused delayed and sustained leakage. This is consistent with findings in rats in which, using a similar radiolabeled tracer technique, Bamforth et al. ²⁴ found no significant leakage 4 hours after intravitreous injection of TNF-α, but sustained leakage starting 24 hours after injection. We found that intravitreous injection of IL-1β also caused sustained disruption of the BRB, but the onset of leakage was earlier than with TNF-α. In rats, IL-1β caused leakage that peaked at 4 to 8 hours and 24 to 48 hours after injection, ²³ which is consistent with our findings in mice. Morphologic evaluations in rabbits demonstrating that IL-1β causes rapid and prolonged opening of tight junctions between retinal vascular endothelial cells are also consistent. Therefore, our findings in mice with purported vasopermeability factors correlate well with both morphologic and functional results in other animals. 

A possible interpretation of these data is that the agents that have a delayed effect on the BRB may act indirectly through other agents. Each has been shown to increase production and release of VEGF. ^{27 28 29 30 31 32 33 34 35 36 37 38 39 40 41} Proinflammatory factors recruit monocytes and may indirectly promote VEGF synthesis through monocyte-smooth muscle cell interactions. ⁴² Additional experiments are needed to determine whether VEGF receptor blockers can decrease the effects of TNF-α, IGF-1, and PGE₂ on the BRB. The technique described in this report will facilitate those studies. 

 Supported by Grants EY10017, EY05951, EY12609, and P30EY1765 from the National Eye Institute; the Foundation Fighting Blindness; the Steinbach Foundation; Lew R. Wasserman Merit Awards (SAV, PAC), and unrestricted funds from Research to Prevent Blindness. PAC is the George S. and Dolores Dore Eccles Professor of Ophthalmology and Neuroscience.

 Submitted for publication September 21, 2001; revised January 16, 2002; accepted February 28, 2002.

 Commercial relationships policy: N.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Time course of radioactivity measurements in retina and lung at various times after intraperitoneal injection of [³H]mannitol. At several times after intraperitoneal injection of 1 μCi/g body weight of [³H]mannitol, mice were killed, and radioactivity (CPM) was measured in the superior lobe of the left lung and the retinas. Each point represents the mean of two measurements for lung and four measurements for retina. The level of radioactivity in the retina was low throughout the duration of the study, whereas that in lung peaked early and decreased to a low baseline level similar to that seen in retina. The greatest difference between retina and lung was 1 hour after injection, the earliest time point tested.

Figure 2.

 

Breakdown of the BRB increases with increasing amounts of cryopexy. Adult C57BL/6 mice were treated with the following amounts of cryopexy: a single 5-second application in one quadrant of one eye (Treatment 1, n = 16), two 5-second applications in each of two different quadrants (Treatment 2, n = 16), or two 10-second applications in each of two different quadrants (Treatment 3, n = 16). After 24 hours, the mice were given an intraperitoneal injection of 1 μCi/g body weight of [³H]mannitol. Sixty minutes after injection, the mice were killed and retinas from both the treated and untreated (control) eyes were rapidly removed, as were a kidney and the superior lobe of the left lung. The tissues were solubilized and assayed for radioactivity. Each bar represents the mean (± SEM) ratio of CPM per milligram retina to CPM per milligram lung (RLLR) (A) or kidney (RRLR) (B). *P < 0.05; **P < 0.01 for difference from contralateral control retina by unpaired t-test.

Figure 3.

 

Time course of breakdown of the BRB after intravitreous injection of purported vasopermeability agents. Adult C57BL/6 mice were anesthetized and given in an intravitreous injection of PBS (control), VEGF, TNFα, IGF-1, PGE₁, PGE₂, or IL-1β in one eye. At various time points, the RLLR (A, C, E) or the RRLR (B, D, F) were measured, as described in the legend to Figure 2 . (A, B) Mean (±SEM) RLLR (A) or RRLR (B) calculated at various times after intravitreous injection of 1 μL of a 10⁻⁶ M solution of the listed agents (except the 6-hour time point for TNF-α, which shows the ratios that resulted after injection of 10⁻⁵ M (n = 5), the highest dosage used). For TNF-α, 5 mice were used to calculate the mean ratios at 24 and 48 hours; for VEGF, 5 mice at 3 hours, 15 at 6 hours, 8 at 24 hours, and 6 at 48 hours; for the control the number of eyes used was 19 at 3 hours, 84 at 6 hours, 95 at 24 hours, and 64 at 48 hours (*P < 0.05; **P < 0.01 for difference from control by unpaired t-test for populations with unequal variances). (C, D) Mean (±SEM) RLLR (C) or RRLR (D) calculated at various time points after intravitreous injection of 1 μL of a 10⁻⁵-M solution of the listed agents. For IGF-1, the number of mice used to calculate the mean ratios was 15 at 6 and 24 hours; for PGE₁, 13 mice at 6 hours and 19 at 24 hours; and for PGE₂, 9 mice at 6 hours, 10 at 24 hours, and 10 at 72 hours; for the control, the number of eyes used was 84 at 6 hours, 95 at 24 hours, and 20 at 72 hours (*P < 0.05 and **P < 0.01 for difference from control by unpaired t-test for populations with unequal variances). (E, F) Mean ± SEM RLLR (E) or RRLR (F) calculated at various time points after intravitreous injection of 1 μL of one of the listed concentrations of IL-1β. For 100 U, 5 mice were used to calculate the mean ratios at each time point of 6, 24, and 48 hours; for 10 U, 5 mice at 6 hours and 7 each at 24 and 48 hours; for 0.5 U, 8 mice at 6 hours and 10 at 24 hours; for 0.05 U, 6 mice at 6 hours and 10 at 24 hours; for the control, the number of eyes used was 84 at 6 hours, 95 at 24 hours, and 64 at 48 hours (*P < 0.05 and **P < 0.01 for difference from control by unpaired t-test for populations with unequal variances).