Interleukin-1β Increases Prostaglandin E2-Stimulated Adenosine 3′,5′-cyclic Monophosphate Production in Rabbit Pigmented Ciliary Epithelium

Exp. Eye Res. (1996) 63, 91–104

Interleukin-1β Increases Prostaglandin E2-Stimulated Adenosine 3«,5«-cyclic Monophosphate Production in Rabbit Pigmented Ciliary Epithelium L L O Y D N. F L E I S H ER*, M. C H R I S T I N E M  G A H A N, J E N N Y B. F E R R E L L    I N E S P A G A N North Carolina State University, College of Veterinary Medicine, 4700 Hillsborough Street, Raleigh, NC 27606, U.S.A. (Received Columbia 6 October 1995 and accepted in revised form 16 February 1996) This study was designed to determine the effects of interleukin-1 on basal and prostaglandin E -stimulated # adenosine 3«,5«-cyclic monophosphate production by primary and first passage cultures of non-transformed rabbit pigmented and non-pigmented ciliary epithelial cells. Confluent cultures of rabbit pigmented and non-pigmented ciliary epithelial cells were incubated for varying periods of time in serum-free medium with or without interleukin-1β, tumor necrosis factor-α, bacterial lipopolysaccharide, transforming growth factor-β2, cycloheximide, indomethacin and combinations of these agents. Cells were then preincubated for 10 min with serum-free medium plus the phosphodiesterase inhibitor, 3isobutyl-1-methylxanthine and then stimulated for 10 min with serum-free medium plus 3-isobutyl-1methylxanthine (for basal adenosine 3«,5«-cyclic monophosphate production) or serum-free medium containing several concentrations of prostaglandin E and 3-isobutyl-1-methylxanthine. In certain # experiments isoproterenol, vasoactive intestinal peptide, or forskolin was substituted for prostaglandin E . # Adenosine 3«,5«-cyclic monophosphate was then extracted into ice-cold absolute ethanol and measured by radioimmunoassay. Prostaglandin E stimulated adenosine 3«,5«-cyclic monophosphate production in # pigmented and non-pigmented ciliary epithelial cells in a dose-dependent manner. Incubation with interleukin-1β (150 U ml−") increased prostaglandin E -stimulated, but not basal adenosine 3«,5«-cyclic # monophosphate production in pigmented ciliary epithelial cells. This interleukin-1β-induced enhancement of prostaglandin E -stimulated adenosine 3«,5«-cyclic monophosphate production, called the # interleukin-1 effect, was not seen with non-pigmented ciliary epithelial cells. The interleukin-1 effect was dependent upon interleukin-1β concentration, time and de novo protein synthesis. The interleukin 1 effect could not be reproduced by replacing interleukin-1β with tumor necrosis factor-α or bacterial lipopolysaccharide and was specific for prostaglandin E , since interleukin-1β did not enhance # isoproterenol-, vasoactive intestinal peptide-, or forskolin-induced adenosine 3«,5«-cyclic monophosphate production. Chronic exposure to prostaglandin E (during the 3 hr incubation period), with or without # interleukin-1β in the incubation medium, reduced subsequent prostaglandin E -stimulated adenosine # 3«,5«-cyclic monophosphate production. Inhibition of de novo prostaglandin synthesis with indomethacin increased the interleukin-1 effect. The interleukin-1 effect was inhibited by the immunosuppressive cytokine, transforming growth factor-β2, in a dose-dependent manner. This is the first report of prostaglandin E -induced stimulation of adenosine 3«,5«-cyclic monophosphate production by pigmented # ciliary epithelial cells and of the unique ability of interleukin-1 to increase this effect. The results are consistent with interleukin-1-induced upregulation of prostaglandin E receptors. Since transforming growth factor-β2 inhibited this interleukin-1 effect, this immunosuppressive cytokine may exert negative feedback and thus regulate the physiological consequences of the interleukin-1 effect. " 1996 Academic Press Limited Key words : ciliary epithelium ; cyclic AMP, interleukin-1 ; prostaglandins ; transforming growth factorbeta ; pigmented ciliary epithelium.

1. Introduction Interleukin-1 (IL-1), a cytokine produced by many cells, but in greatest quantities by activated monocytes} macrophages, exerts a wide range of effects in the host. Indeed, IL-1 probably affects nearly every tissue and organ in the body. Originally identified as an ‘ endogenous pyrogen ’ and later as a ‘ lymphocyte activating factor ’, IL-1 has been implicated in nu* For correspondence.

0014-4835}96}070091­14 $18.00}0

merous inflammatory disease states including septic shock, rheumatoid arthritis, inflammatory bowel disease, and localized inflammation of diverse etiology (Dinarello, 1991). Although the α and β isoforms of IL-1 are the products of separate genes (Cannon et al., 1989), they exert remarkably similar biological effects and both have been shown to play important roles in inflammation of extraocular origin (Dinarello, 1991). IL-1 has also been implicated in the intraocular inflammatory response since IL-1β levels increase in aqueous humor during experimentally-induced uveitis # 1996 Academic Press Limited

92

(Kulkarni and Mancino, 1993) and both the α and β isoforms induce uveitis when injected intraocularly (Bhattacherjee and Henderson, 1987 ; Fleisher, Ferrell and McGahan, 1992 ; Rosenbaum et al., 1987). However, despite the presumed importance of IL-1 in intraocular inflammation, attempts to inhibit its activity have not achieved the anticipated antiinflammatory consequences. For example, treatment with IL-1 receptor antagonist affected neither endotoxin-induced uveitis in rats (Rosenbaum et al., 1993) and rabbits (Rosenbaum and Boney, 1992), nor immunogenically-induced uveitis in rabbits (Rosenbaum and Boney, 1992). Although these results are inconsistent with an essential role for IL-1 in uveitis, effective concentrations of IL-1 receptor antagonist may not have been achieved at critical intraocular anatomical sites. The importance of achieving effective concentrations of IL-1 receptor antagonist at critical intraocular sites is underscored by the fact that IL-1 requires less than 5 % receptor occupancy to induce its effects (Dinarello, 1991). One potential intraocular site for the actions of IL-1 is the ciliary epithelium of the iris–ciliary body. This bilayer of cells consists of a non-pigmented layer (NPE) opposing the aqueous humor and a pigmented layer (PE) directly contacting the vascular stroma of the ciliary processes. By virtue of its proximity to the vasculature of the ciliary processes, the PE layer would be especially accessible to the actions of cytokines such as IL-1. However, little information exists concerning the effects of IL-1 on cells of the anterior uvea. In an earlier study we reported that IL-1β did not affect adenosine 3«,5«-cyclic monophosphate (cAMP) production by isolated ciliary epithelial bilayers, in vitro, although these bilayers did exhibit increased cAMP production, ex vivo, following their removal from eyes in which inflammation had been induced by intravitreal injection of IL-1β, tumor necrosis factor-α (TNF-α), or bacterial lipopolysaccharide (endotoxin) (Fleisher, Ferrell and McGahan, 1995). However, since the bilayer consists of two distinct cell populations, the inability of IL-1 to alter cAMP production in vitro is complicated by the possibility that IL-1 could increase cAMP production in one cell type and decrease it on the other ; the net effect could be little, if any change in total cAMP production. Therefore, in the present investigation we utilized cultures of nontransformed rabbit PE and NPE cells to examine the effects of IL-1β on basal and prostaglandin (PG) E # stimulated cAMP production. PGE was chosen as a # cAMP-inducing agent since it has been shown to stimulate cAMP production by ciliary epithelial cells (Bhattacherjee, Rhodes and Paterson, 1993 ; Jumblatt et al., 1994), is produced in large quantities during intraocular inflammation (Eakins, 1977 ; Fleisher and McGahan, 1986), and has both inflammatory and anti-inflammatory activities in the eye (Cole and Unger, 1973 ; Hoyng et al., 1986 ; Protzman and Woodward, 1990 ; Wong and Howes, 1983).

L. N. F L E I S H E R E T A L.

2. Material and Methods Preparation of Rabbit Pigmented and Non-pigmented Ciliary Epithelial Cells Rabbit PE and NPE cell cultures were prepared based upon methods developed by Fain and Farahbakhsh (1989) and Cilluffo, Cohen, and Fain (1991). Although most experiments were performed with primary cultures, some were performed with primary and first passage cells. Descriptions of the state of passage of cells for each experiment are found in the legends for each figure. With respect to rabbit breed, although male New Zealand white rabbits (2–3 kg) were used in the initial studies with PE cells, most of the experiments on PE cells utilized tissues obtained from male Dutch Belted rabbits. The breeds used for each experiments are also indicated in the figure legends. All NPE cells were prepared from 2–3week-old Dutch Belted rabbits. Treatment of animals conformed to the NIH Guide for the Care and Use of Laboratory Animals, the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research, and was approved by the University Institutional Animal Care and Use Committee. In order to prepare PE cells, rabbits were killed by exposure to an atmosphere of 100 % CO and eyes # enucleated and rinsed in Hanks’ balanced salt solution (HBSS). The posterior segment was removed 4 mm posterior to the limbus and the anterior segment pinned to a dissecting dish (with 5 mm hardened Sylgard ; Dow Corning, Midland, MI, U.S.A.) containing HBSS. The lens was gently removed leaving the capsule in place and a 1 mm strip (around the ciliary body ring) dissected from the remaining anterior segment. The strip was transferred to fresh HBSS, cut into four segments, and transferred to growth medium (GM) consisting of minimum essential medium (MEM) with Earle’s salts and -glutamine (Gibco BRL, Grand Island, NY, U.S.A.) plus gentamicin (50 µg ml−"), kanamycin (100 µg ml−" ; all antibiotics from Gibco BRL, Grand Island, NY, U.S.A.), additional -glutamine (3 m), and fetal bovine serum (FBS, 15 % ; Hyclone, Logan, UT, U.S.A.). The strips from both eyes were rinsed with Hanks’ Ca- and Mg-free balanced salt solution (HCMF-BSS) and incubated with HCMF-BSS containing 2±4 % dispase (Boehringer Mannheim, Indianapolis, IN, U.S.A.) and 100 m sorbitol (Sigma Chemical Co., St. Louis, MO, U.S.A.) for 37 min. Strips were then rinsed with HCMF-BSS, transferred to GM, and the epithelial bilayer peeled off and transferred to fresh GM. Strips of epithelial cells were rinsed in HCMF-BSS and incubated (1 hr) in HCMF-BSS containing 0±5 m EDTA and 0±05 % trypsin (Difco, Detroit, ME, U.S.A.). The cells plus the trypsin-EDTA were mixed with GM, centrifuged (5 min at 300 g), the cell pellet rinsed by centrifugation (twice with GM), and the cells suspended by trituration of the pellet with 3 ml GM. Cells were then plated

I L-1β I N C R E A S E S P G E2-S T I M U L A T E D c A M P

(A)

(B)

F. 1. Primary cultures of (a) pigmented and (b) nonpigmented ciliary epithelial cells from Dutch Belted rabbits. Both cultures were stained for tight-junction associated protein using rat anti-ZO-1 antibody. Notice the absence of

93

(50 000 cells per well) on six-well tissue culture plates (Costar u3506, Cambridge, MA, U.S.A.) and incubated at 37°C in 95 % air}5 % CO and 95 % humidity. # These conditions support the growth of PE, but not NPE cells (Fain and Farahbakhsh, 1989). Cells reached confluence within 7–9 days. Purity of the PE cultures was verified by microscopic appearance, absence of mouse anti-vimentin antibody staining (Accurate Chemical & Scientific Corp., Westbury, NY, U.S.A.), a marker for fibroblasts (the mouse anti-vimentin antibody did stain rabbit dermal fibroblasts), and absence of rat anti-ZO-1 antibody staining (clone MAB1520 ; Chemicon, Temecula, CA, U.S.A.), a marker for the tight junctions of NPE cells (Stevenson et al., 1986). The absence of ZO-1 staining on PE cells and its presence on NPE cells (as a dark border around the cells) can be seen in Fig. 1(a) and (b). Monolayers were stained for ZO-1 using the streptavidin-biotin-peroxidase method (Histostain SP ; Zymed Lab., South San Francisco, CA, U.S.A.) with 3-amino-9ethylcarbazole as the chromogen. At confluency monolayers were rinsed with phosphate-buffered saline (PBS), fixed in fresh 4 % paraformaldehyde (1 hr at 4°C), rinsed with PBS (three times), and then preincubated in 10 % normal goat serum with 0±3 % Triton X-100 in PBS (20 min at 25°C) to reduce background staining. Monolayers were then incubated with a 1 : 25 dilution of rat anti-ZO-1 antibody (20 min at 37°C ; Chemicon, Temekula, CA, U.S.A.), washed with PBS (three times), incubated with a biotinylated second antibody (10 min at 25°C), washed with PBS, incubated with horseradish peroxidase–streptavidin conjugate (10 min at 25°C), washed with PBS, and incubated with hydrogen peroxide-chromogen mixture (8 min at 37°C). Cells were then rinsed in distilled water, counterstained with hematoxylin for 3 min, washed in tap water and then in PBS for 30 sec, and then rinsed in distilled water. Cells were permanently mounted and examined with a Zeiss microscope. Controls included monolayers incubated without the primary antibody or with normal rat serum. The positive reaction product appears as a dark border surrounding the NPE. The same method was used to stain vimentin, except that a 1 : 200 dilution of the mouse anti-vimentin antibody replaced the anti-ZO-1 antibody. In experiments involving the effects of cycloheximide on PGE -stimulated cAMP production, the # viability of PE cells was determined by Trypan blue exclusion as follows. After a 2 hr incubation with or without cycloheximide, monolayers were rinsed with MEM and a 1 : 1 mixture of 0±4 % Trypan blue : MEM was added. After 10 min, cells were examined under brightfield using a Nikon TMS microscope. Non-viable cells take up the stain and appear blue under these conditions. staining of PE and the presence of dark borders surrounding the NPE (magnification ¬33).

94

In order to prepare NPE cells, ciliary body strips were dissected from 2–3-week-old Dutch Belted rabbits, cut into two segments, transferred to a 1 : 1 mixture of MEM and Nutrient Mixture F-10 (Gibco BRL), rinsed with this mixture and incubated in this mixture plus 3±46 m EGTA (Sigma) for 2–4 hr at 37°C. NPE were then microdissected from PE and the NPE monolayers from several eyes were pooled and transferred to a chloride-free balanced salt solution consisting of 126 m Na glutamate, 3±8 m -aspartic acid (mono K+ salt), 1±08 m Na HPO , 14 m # % NaHCO , 6±1 m -glucose, 0±31 m -gluconic acid $ (hemi Ca#+ salt), 0±63 m MgSO and 0±0005 % phenol % red (all from Sigma). After rinsing with HCMF-BSS, NPE monolayers were incubated in HCMF-BSS containing 2 mg ml−" collagenase–dispase (Boehringer Mannheim, Indianapolis, IN, U.S.A.) for 45 min at 37°C. At the end of this incubation NPE culture growth medium (NPGM) consisting of NCTC-135 medium plus gentamicin (50 µg ml−"), kanamycin (100 µg ml−"), -glutamine (3 m), adenine HCl (0±49 µ), cholesterol (520 n), O-phosphoethanolamine (5 µ), FeSO (1 µ), ribose (3±3 µ), % sodium pyruvate 150 µ), ATP (1±9 µ), ethanolamine (5 µ), bovine serum albumin (0±79 mg ml−") and 15 % FBS was added (all from Sigma except FBS ; Hyclone, Logan, UT, U.S.A.). The mixture was then centrifuged (5 min at 300 g), the cell pellet rinsed by centrifugation (twice wtih NPGM), and the pellet suspended by trituration with 3 ml NPGM. Cells were then plated (50 000 cells per well) on six-well tissue culture plates coated with collagen 1 (Collaborative Research, Bedford, MA, U.S.A.). NPE cultures were verified by positive staining of their tight junctions with rat anti-ZO-1 antibody (see Fig. 1). Experimental Procedure and Extraction and Assay of cAMP Once cells reached confluence they were washed three times with serum-free GM (SFGM with no additives) and incubated with SFGM containing the appropriate test agent (e.g. IL-1β, TNF-α) at 37°C in 95 % air}5 % CO and 95 % humidity. The duration of # the incubation period varied depending upon the particular experiment. At the conclusion of the incubation period, PE monolayers were preincubated for 10 min in 3 ml SFGM containing the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX, 0±5 m, Calbiochem, La Jolla, CA, U.S.A.), aspirated, and then stimulated with the test agent for 10 min (usually PGE in SFGM plus IBMX or just # SFGM plus IBMX ; for basal cAMP production). Stimulation of cAMP production was terminated by aspirating the SFGM and adding 1 ml ice-cold absolute ethanol. Plates were then maintained at room temperature for 10 min, swirled, the extract transferred to microcentrifuge tubes and dried for 1 hr in a vacuum concentrator (Labconco, Kansas City, MO, U.S.A.).

L. N. F L E I S H E R E T A L.

The dried residue was dissolved in Tris–EDTA buffer (from the RIA kit ; 125 µl) and 50 µl aliquots used for the cAMP assay (Amersham, Arlington Heights, IL, U.S.A.). Cellular protein was determined by dissolving the contents of each dry tissue culture well in NaOH (700 µl, 1 ). After 3–4 hr of gentle rotatory swirling, 100 µl aliquots were assayed by the Lowry method (Lowry et al., 1951). Reagents Human recombinant IL-1β (specific activity : 1±5¬10) U mg−" protein) and human recombinant TNF-α (specific activity : 2±86¬10( U mg−" protein) were purchased from R & D Systems (Minneapolis, MN, U.S.A.). Simian recombinant transforming growth factor-β2 (TGF-β2) was provided by Celtrix Laboratories (Santa Clara, CA, U.S.A.). PGE , en# dotoxin (bacterial lipopolysaccharide from E. coli, O55 ; B5), cycloheximide, and VIP were purchased from Sigma ; isoproterenol and forskolin were purchased from Calbiochem. Statistical Analysis Differences between experimental groups were analysed using one-way ANOVA. When analysis involved more than two means, the significance of differences between them was determined using Tukey’s HSD test. The null hypothesis was rejected at P ! 0±05. Data represent mean³... Treatments were routinely run in duplicate or triplicate. 3. Results Response of PE and NPE Cells to PGE : Effects of Cell # Passage, Breed and IL-1β on Basal and PGE # Stimulated cAMP Production PGE induced dose-dependent cAMP production in # rabbit PE and NPE cells. Figure 2 illustrates that one passage of New Zealand white (NZW) PE cells did not significantly affect the ability of PGE to stimulate # cAMP production during a 10 min stimulation period. Primary NPE cells [which were always from Dutch Belted (DB) rabbits] were somewhat less sensitive to PGE compared to either primary or first passage PE # cells from NZW rabbits (Fig. 2). With respect to rabbit breed, Fig. 3 illustrates that after a 4 hr incubation in serum-free medium, primary PE cells from DB rabbits were less sensitive to high concentrations of PGE # (10–100 µ) than primary PE cells from NZW rabbits. However, both breeds exhibited similar responsiveness to 0±1–1 µ PGE , the concentrations primarily used # throughout this investigation. Figure 4 illustrates that incubation of NZW PE cells with IL-1β for 4 hr followed by stimulation with medium (for basal cAMP production) or PGE (0±1 µ) # for 10 min, resulted in a significant increase in PGE # stimulated (P ! 0±007), but not basal cAMP pro-

I L-1β I N C R E A S E S P G E2-S T I M U L A T E D c A M P

cAMP pmol (mg protein)–1

200

150

100

50

0

0

0.01

0.1 1 PGE2 (µM)

10

100

F. 2. PGE dose–response for pigmented and non# pigmented ciliary epithelial cells. Monolayers of primary (D) and first passage (E) NZW rabbit PE cells and primary (_) DB rabbit NPE cells were stimulated with PGE (0–100 µ) # for 10 min. Cyclic AMP was then extracted and measured by radioimmunoassay. Data expressed as mean³... ; n ¯ 4 (primary and first passage PE) and n ¯ 2 (primary NPE). Error bars smaller than the symbols do not appear on the figure.

duction. The IL-1β-induced increase in PGE # stimulated cAMP production, which will be referred to as the IL-1 effect, was observed at 15 U ml−" IL-1β and peaked at 150 U ml−". It should also be noted that incubation with IL-1β (150 U ml−") enhanced the ability of PGE to stimulate cAMP production to an # equal extent in both NZW and DB rabbits (Fig. 3 ; compare NZW}IL-1 with DB}IL-1). As shown in Fig. 5, similar exposure of NPE cells to IL-1β did not alter basal cAMP production, nor did it induce an IL-1 effect. Kinetics of the IL-1 Effect In order to determine the kinetics of the IL-1 effect, separate groups of PE cells were exposed to IL-1β (150 U ml−") for time periods up to 24 hr. At the end of each incubation period PE cells were stimulated

95

with medium or PGE (1 µ) and cAMP measured. As # illustrated in Fig. 6, basal cAMP production was constant over the 24 hr incubation period, and PGE # stimulated cAMP production was greater than the corresponding basal production at all time points. The effects of exposure to IL-1β, i.e., the IL-1 effect, became apparent at 1±5 hr when PGE -stimulated cAMP # production increased to 1±9-fold that at time zero. By 3 hr, PGE -stimulated cAMP production increased to # 3±5-fold that at zero time and remained elevated at 24 hr (2±5-fold increase compared to time zero). Specificity of the IL-1 effect for IL-1 and for PGE

# In order to assess whether enhancement of PGE # stimulated cAMP production was specific for IL-1, PE cells were exposed to TNF-α, a cytokine that shares many biological activities with IL-1 (Le and Vilcek, 1987), and to bacterial lipopolysaccharide (LPS), which works largely through release of cytokines such as IL-1 and TNF-α (Beutler and Cerami, 1987, Dinarello, 1991). As shown in Fig. 7, exposure of PE cells to TNF-α for 4 hr altered neither basal (medium stimulated) nor PGE -stimulated (0±1 and 1 µ) cAMP # production. Since these experiments were conducted at different times using different concentrations of TNF-α and PGE , the results are presented in Fig. 7 # as two separate experiments. Similar exposure of PE cells to LPS (0±01–10 µg ml−") also did not influence basal or PGE -stimulated cAMP production (Fig. 8). # Because of differences in baseline responses between different experiments, the LPS data were expressed as the ratio of stimulated}basal cAMP production. In this instance, ‘ Basal ’ was defined as the mean cAMP production by cells incubated without LPS and stimulated with medium. In order to assess whether IL-1β enhanced cAMP production by agents acting through receptors other than PGE receptors, the effects of IL-1β on

cAMP pmol (mg protein)–1

120 100 80 60 40 20 0

NZW/Medium

DB/Medium NZW/IL-1 Rabbit breed/4 hr incubation condition

DB/IL-1

F. 3. Effect of rabbit breed on PGE -stimulated cAMP production by PE cells. Primary cultures of PE cells from NZW # (n ¯ 2) and DB rabbits (n ¯ 4) were incubated in serum-free medium, with or without IL-1β (150 U ml−"), for 4 hr and then stimulated with PGE (*, 0 ; 8, 0±01 µ ; 7, 0±1 µ ; +, 1 µ ; 5, 10 µ ; 9, 100 µ) for 10 min. Cyclic AMP was # then extracted and measured by radioimmunoassay. Data expressed as mean³...

96

L. N. F L E I S H E R E T A L.

cAMP pmol (mg protein)–1

50 *†

40

30

*†

* *

20

*

10

0

0

1.5

15 IL-1β (U ml–1)

150

750

F. 4. Effect of IL-1β on basal and PGE -stimulated cAMP production by PE cells. PE cells from NZW rabbits were incubated # with medium or medium supplemented with IL-1β for 4 hr and then stimulated with medium (basal cAMP production) (*) or medium supplemented with PGE (0±1 µ) (+) for 10 min (all experiments, but one were conducted with primary cultures). # Cyclic AMP was then extracted and measured by radioimmunoassay. Data expressed as mean³... for n ¯ 4–6 (n ¯ 2 for − 750 U ml " IL-1β). * ¯ Significantly different from corresponding basal cAMP production (P ! 0±02) ; † ¯ Significantly different from PGE -stimulated cAMP production by cells incubated without IL-1β (P ! 0±007). #

cAMP pmol (mg protein)–1

40

30

20

10

0

0

15

150 750 3 hr incubation; IL-β (U ml–1)

1500

F. 5. Effect of IL-1β on basal and PGE -stimulated cAMP production by NPE cells. NPE cells from DB rabbits were incubated # with medium or medium supplemented with IL-1β (15–1500 U ml−") for 3 hr and then stimulated with PGE for 10 min # (0–10 µ). Cyclic AMP was then extracted and measured by radioimmunoassay. Two experiments were conducted with primary cultures ; the other two were conducted with first passage cultures derived from these primary cultures. Data expressed as mean³... for n ¯ 4. *, medium ; 8, PGE (1 µ) ; +, PGE (10 µ). # #

isoproterenol-, vasoactive intestinal peptide (VIP)- and forskolin-stimulated cAMP production were measured in PE cells. Isoproterenol stimulates cAMP production by a β-adrenergic receptor-Gs protein transduced mechanism (Asano et al., 1984), while VIP acts upon its own receptor, also utilizing a Gs protein transduced mechanism (Finch, Sreedharan and Goetzl, 1989). Forskolin bypasses cell membrane receptors by directly stimulating the catalytic subunit of adenylyl cyclase (Seamon and Daly, 1981). PE cells were incubated in medium or medium supplemented with IL-1β (150 U ml−") for 4 hr and then stimulated with isoproterenol (10 µ), VIP (1 and 10 n) or forskolin (1 and 10 µ) for 10 min. As illustrated in Fig. 9, although isoproterenol, VIP, and forskolin increased cAMP production well above basal levels, incubation with IL-1β did not enhance the cAMP-stimulating

efficacy of each agent compared to PE cells incubated in medium alone. It should be pointed out that VIPstimulated cAMP production was quite impressive (10 n VIP increased cAMP production 100-fold above basal levels). VIP has been reported to be a potent agonist for cAMP production in membrane preparations from rabbit ciliary processes (Mittag, Tormay and Podos, 1987) (9–20-fold increases at nanomolar concentrations ; EC ¯ 65 n) and in &! intact, excised rabbit ciliary processes (15-fold increase at 300 n) (McNellis and Bausher, 1991). Importance of De Novo Protein Synthesis to the IL-1 Effect The latency between exposure to IL-1β and expression of the IL-1 effect suggested a requirement for

I L-1β I N C R E A S E S P G E2-S T I M U L A T E D c A M P

97

cAMP pmol (mg protein)–1

150 125 100 75 50 25 0

0

0.17

0.5

1 1.5 2 3 Hours with IL-1β (150 U ml–1)

4

5

24

F. 6. Time course for basal and PGE -stimulated cAMP production by PE cells. Primary cultures of DB rabbit PE cells were # incubated in medium supplemented with IL-1β (150 U ml−") for periods ranging from 0–24 hr. At the conclusion of each incubation period cells were stimulated with medium (basal cAMP production ; ^) or medium supplemented with PGE (1 µ ; # _) for 10 min. Cyclic AMP was then extracted and measured by radioimmunoassay. Data expressed as mean³... for n ¯ 3–4. Error bars smaller than the symbols do not appear on the figure.

cAMP pmol (mg protein)–1

40

Experiment 1

30

Experiment 2

20

10

0

0

20

200

0

200

2000

–1

TNFa (U ml )

F. 7. Effect of TNF-α on basal and PGE -stimulated cAMP production by PE cells. In two separate groups of experiments # primary and first passage PE cells from NZW rabbits were incubated with medium or medium supplemented with TNF-α [20–200 U ml−" (experiment 1) ; 200–2000 U ml−" (experiment 2)] for 4 hr and then stimulated with medium (basal cAMP production ; *) or medium supplemented with PGE (0±1 µ for experiment 1, + ; 1 µ for experiment 2, 5) for 10 min. Cyclic # AMP was then extracted and measured by radioimmunoassay. Data expressed as mean³... for n ¯ 4–6 (experiment 1) and n ¯ 2–3 (experiment 2).

de novo protein synthesis. In order to determine whether de novo protein synthesis was required for the IL-1 effect, PE cells were incubated with medium or IL-1 (150 U ml−") with or without cycloheximide (0–100 µ) for 2 hr and then stimulated with medium or PGE (1 µ) for 10 min. The incubation period was # shortened to 2 hr to minimize the possibility of cycloheximide-induced cytotoxicity. As illustrated in Fig. 10, inhibition of de novo protein synthesis completely abolished the IL-1 effect. At 10 and 100 µ cycloheximide, PGE -stimulated cAMP production by # PE cells incubated with IL-1β was of similar magnitude to that observed in cells incubated without IL-1β. This inhibition of the IL-1 effect was not due to

cycloheximide-induced diminution of responsiveness to PGE since cycloheximide affected neither basal # cAMP production (in the presence or absence of IL1β), nor the ability of PE cells to respond to PGE . # PGE -stimulated cAMP production by PE cells incu# bated in medium alone was not significantly different from PGE -stimulated cAMP production by cells # incubated with medium plus cycloheximide. It should be noted that although PGE -stimulated cAMP pro# duction by cells incubated with IL-1β plus 10 µ cycloheximide was greater than the corresponding basal cAMP production, the difference between the two groups was not statistically significant. However, since there was a significant difference between these

98

L. N. F L E I S H E R E T A L. 4

Stimulated/basal

3

2

1

0

0

0.01

0.1 LPS (µg ml–1)

1

10

F. 8. Effect of LPS on basal and PGE -stimulated cAMP production by PE cells. Primary and first passage PE cells from DB # rabbits were incubated with LPS (0–10 µg ml−") for 3 hr and then stimulated with medium (basal cAMP production ; *) or medium supplemented with PGE (1 µ ; +) for 10 min. Cyclic AMP was then extracted and measured by radioimmunoassay. # Data expressed as mean³... for n ¯ 6 (n ¯ 3 for 10 µg ml−" LPS).

1200

cAMP pmol (mg protein)–1

1100 1000 700 600 500 400 300 200 10 0

Basal

Isoprot 10 µM

VIP 1 nM

VIP 10 nM

FSK 1 µM

FSK 10 µM

F. 9. Effect of isoproterenol, VIP, and forskolin on cAMP production by PE cells in the presence and absence of IL-1β. Primary PE cells from DB rabbits were incubated in medium (*) or medium supplemented with IL-1β (150 U ml−" ; +) for 4 hr. At the conclusion of the incubation period cells were stimulated for 10 min with medium (basal cAMP production) or medium supplemented with isoproterenol (10 µ), VIP (1 and 10 n) or forskolin (1 and 10 µ). Cyclic AMP was then extracted and measured by radioimmunoassay. Data expressed as mean³... for n ¯ 9.

groups at 100 µ cycloheximide (P ! 0±03), it does not appear that cycloheximide significantly inhibited the ability of PE cells to respond to PGE . Furthermore, # analysis of separate groups of PE cells with trypan blue indicated that cell viability was not affected by cycloheximide treatment (data not shown). Effect of Transforming Growth Factor (TGF)-β2 on cAMP Production in PE Cells TGF-β is a multifunctional cytokine with potent immunosuppressive and anti-inflammatory activities (Wahl, 1992). Since TGF-β can antagonize many of the actions of IL-1 (Gamble and Vadas, 1991 ; Junquero et al., 1992 ; Turner et al., 1991), PE cells were incubated with medium or IL-1 (150 U ml−")

with or without TGF-β2 (0–2000 pg ml−") for 3 hr and then stimulated with medium or PGE (1 µ) for # 10 min. Because of differences in baseline responses between different experiments, data were expressed as the ratio of stimulated}basal cAMP production. ‘ Basal ’ is defined as the mean cAMP production by cells incubated without IL-1β and TGF-β2 and stimulated with medium. Without IL-β in the incubation medium, basal and PGE -stimulated cAMP production were # unaffected by TGF-β2 (Fig. 11). However, if IL-1β was included in the 3 hr incubation medium, then TGF-β2 induced a dose-dependent decrease in the IL-1 effect (P ! 0±002 for 20–2000 pg ml−" TGF-β2 ; corresponding to TGF-β2 concentrations of 0±8– 80¬10−"# ). Inhibition was first seen at 20 pg ml−" of TGF-β2 (40 % decrease in the IL-1 effect compared

I L-1β I N C R E A S E S P G E2-S T I M U L A T E D c A M P

99

90

cAMP pmol (mg protein)–1

80

*†

70 60 50 40

*

30

* *

20

*

10 0

10

0

100

Cycloheximide in incubation medium (µM)

F. 10. Effect of cycloheximide on basal and PGE -stimulated cAMP production by PE cells in the presence and absence of # IL-1β. Primary PE cells from DB rabbits were incubated in medium with or without IL-1β (150 U ml−") plus cycloheximide (0–100 µ) for 2 hr. At the conclusion of the incubation period cells were stimulated with medium (basal cAMP production) or medium supplemented with PGE (1 µ) for 10 min. Cyclic AMP was then extracted and measured by radioimmunoassay. # *, incubated in medium with or without cycloheximide, stimulated with medium ; 8, incubated in medium with or without cycloheximide, stimulated with PGE ; 5, incubated in IL-1β with or without cycloheximide, stimulated with medium ; +, # incubated in IL-1β with or without cycloheximide, stimulated with PGE . Data expressed as mean³... for n ¯ 4. * ¯ # Significantly different from basal cAMP production by cells subjected to the same incubation conditions (P ! 0±04) ; † ¯ Significantly different from PGE -stimulated cAMP production by cells incubated without IL-1β and cycloheximide (P ! 0±02). # 8 7

Stimulated/basal

6

* *

5

*

4 3 2 1 0

0

0.2

2 20 200 TGF−β2 in incubation medium (pg ml–1)

2000

F. 11. Effect of TGF-β2 on basal and PGE -stimulated cAMP production by PE cells in the presence and absence of IL-1β. # Primary PE cells from DB rabbits were incubated in medium with or without IL-1β (150 U ml−") plus TGF-β2 (0–2000 pg ml−") for 3 hr. At the conclusion of the incubation period cells were stimulated with medium (basal cAMP production) or medium supplemented with PGE (1 µ) for 10 min. Cyclic AMP was then extracted and measured by radioimmunoassay. *, incubated # in medium with or without TGF-β2, stimulated with medium ; 8, incubated in medium with or without TGF-β2, stimulated with PGE ; 5, incubated in IL-1β with or without TGF-β2, stimulated with medium ; +, incubated in IL-1β with or without # TGF-β2, stimulated with PGE . Data expressed as mean³... for n ¯ 6–13 (n ¯ 3 for cells incubated with 2000 pg ml−" # TGF-β2, but no IL-1β). * ¯ Significantly different from PGE -stimulated cAMP production by cells incubated with IL-1β, but # no TGF-β2 (P ! 0±002).

to PGE -stimulated cAMP production in cells incubated # with IL-1β, but without TGF-β2) and was maximal at 2000 pg ml−" (67 % decrease). Higher concentrations of TGF-β did not further decrease the IL-1 effect (data # not shown).

Effect of Chronic Exposure to PGE on cAMP # Production in PE Cells Since PGE receptors are subject to homologous # down-regulation (Limas and Limas, 1987), PE cells

100

L. N. F L E I S H E R E T A L. 100

cAMP pmol (mg protein)–1

90 80 70 60 50 *

40 30 20 *

10 0

PGE2 (1 µM)

Medium

10 min stimulation conditions

F. 12. Effect of chronic exposure to PGE on cAMP production by PE cells in the presence and absence of IL-1β. First passage # PE cells from DB rabbits were incubated for 3 hr in medium (*) or medium supplemented with IL-1β (150 U ml−" ; 5), PGE # (1 µ ; 8), or IL-1β plus PGE (+). At the end of the incubation period cells were stimulated for 10 min with medium (basal # cAMP production) or PGE (1 µ). Cyclic AMP was then extracted and measured by radioimmunoassay. Date expressed as # mean³... for n ¯ 3. * ¯ significantly different from cells receiving the same stimulation and incubation conditions except that PGE was not present in the incubation medium (P ! 0±001). #

were incubated with medium or medium supplemented with IL-1β (150 U ml−"), PGE (1 µ), or IL-1β plus PGE for 3 hr and then stimulated with medium # or PGE (1 µ) for 10 min. As shown in Fig. 12, # chronic exposure to PGE significantly reduced PGE # # stimulated cAMP production (P ! 0±001 ; compared to cells incubated in medium without PGE and # stimulated with PGE ). Furthermore, although an # IL-1 effect was observed with PE cells incubated with medium­IL-1β­PGE and stimulated with PGE , the # # amount of cAMP produced under these conditions was significantly less than that produced by cells incubated in medium plus IL-1β and stimulated with PGE (P ! # 0±001). Effect of Indomethacin on cAMP Production in PE Cells Since chronic exposure of PE cells to PGE decreased # subsequent PGE -stimulated cAMP production, this # raised the possibility that PGE played a role in the # regulation of PGE receptor expression. In order to test this hypothesis, PE cells were incubated with medium or medium supplemented with the cyclo-oxygenase inhibitor, indomethacin (10 µ), IL-1β, (150 U ml−"), or indomethacin­IL-1β for 4 hr and then stimulated with PGE (0–1 µ) for 10 min. As shown in Fig. 13, # incubation with indomethacin did not alter basal cAMP production under any of the four incubation conditions, nor did it alter PGE -stimulated cAMP # production in PE cells incubated without IL-1β. However, when cells were incubated with indomethacin­IL-1β, a dose-dependent increase in PGE -stimulated cAMP production was observed com# pared to cells incubated with just IL-1β (P ! 0±03).

4. Discussion The ciliary epithelial bilayer serves as the anatomical site of the blood–aqueous barrier in the ciliary body and is the site of formation of aqueous humor. Despite these essential functions, virtually nothing is known about signal transduction mechanisms in the inner or pigmented layer of this epithelial bilayer. Attention has focused on the non-pigmented layer, primarily because of its tight junctions and secretory nature. However, the pigmented layer is also thought to play a role in formation of aqueous humor (Cole, 1984). This is underscored by the coupling of NPE and PE through gap junctions at their apical membranes (Raviola and Raviola, 1978). Furthermore, since it has the same neuroectodermal origin as the retinal pigment epithelium, a tissue with extensive biosynthetic and immunological properties, the PE could also be involved in barrier function and migration of leukocytes during the intraocular inflammatory response. In the current investigation we have characterized a unique action of the cytokine IL-1β on rabbit PE cells, i.e. an IL-1-induced enhancement of PGE # stimulated cAMP production (Fig. 4). This IL-1 effect occurred selectively with PE cells since it was not observed if rabbit NPE cells were exposed to IL-1β (Fig. 5). The IL-1 effect was dose- and time-dependent (Figs. 4 and 6), required de novo protein synthesis (Fig. 10), and was partially inhibited by the immunosuppressive and anti-inflammatory cytokine TGF-β2 (Fig. 11). Although PGE has been reported to decrease adenylyl # cyclase activity in porcine PE membrane fractions (Sano and Shichi, 1993), to the best of our knowledge, this is the first description of a stimulatory effect of

I L-1β I N C R E A S E S P G E2-S T I M U L A T E D c A M P

101

cAMP pmol (mg protein)–1

150

*

120

90 * 60

30

0

–IL-1/–Indo

–IL-1/+Indo +IL-1/–Indo 4 hr incubation conditions

+IL-1/+Indo

F. 13. Effect of indomethacin on basal and PGE2-stimulated cAMP production by PE cells in the presence and absence of IL-1β. Primary PE cells from DB rabbits were incubated for 4 hr in medium, medium­indomethacin (10 µ), medium­IL1β (150 U ml−"), or medium­indomethacin­IL-1β and then stimulated for 10 min with PGE (*, 0 ; 8, 0±1 µ ; +, 1 µ). # Cyclic AMP was then extracted and measured by radioimmunoassay. Data expressed as mean³... for n ¯ 4. * ¯ significantly different from cells receiving the same stimulation and incubation conditions except that indomethacin was not present in the incubation medium (P ! 0±001).

PGE on cAMP production in PE cells. It should be # pointed out that PGE can exert concentration# dependent effects on cAMP production (Ashby, 1986 ; Cohen-Luria and Rimon, 1992). In order to demonstrate inhibition of cAMP production, the effects of PGE on activator-driven (e.g. forskolin or isopro# terenol) cAMP production must be determined. Since the methods used in the present investigation would not reveal an inhibitory effect of PGE on PE cAMP # production, the possibility that PGE exerts a bimodal # effect on cAMP production in rabbit PE cells cannot be ruled out. The present results also differ with those of Mittag, Tormay and Podos (1987) who were unable to detect significant adenylyl cyclase activity in membrane preparations of rabbit PE cells. However, as they pointed out, the use of enzymes such as trypsin and collagenase can introduce artifacts, especially when membrane preparations are used. Indeed, even in their NPE membrane preparations, the enzyme treatments used caused a 50–60 % loss of total cyclase activity. In the present investigation, cells were never disrupted and were allowed 7–9 days to grow to confluence. Therefore, enzyme-induced damage to adenylyl cyclase activity was probably very limited. How could IL-1 induce an enhancement of PGE # stimulated cAMP production in rabbit PE cells ? Since PGE is thought to stimulate cAMP through a # receptor–G-protein coupled mechanism (Smith, 1989), IL-1 could act at one of several levels ; the PGE receptor, the transducing Gs protein, or the adenylyl cyclase enzyme. It is unlikely that IL-1 directly increased adenylyl cyclase activity since it did not augment basal cAMP production (Pieroni et al., 1993) nor did it increase isoproterenol-, VIP- or forskolinstimulated cAMP production (Fig. 6). One mechanism consistent with the results of the present investigation

is that IL-1 upregulated expression of PGE receptors, either by increasing receptor synthesis or by altering receptor turnover. Although the results do not distinguish between IL-1-induced effects on receptor synthesis and turnover, the following observations are consistent with IL-1-induced increases in functional PGE receptors and with a role for PGE in the # regulation of PGE receptor expression. (1) The IL-1 effect was dependent upon time (Fig. 6) and de novo protein synthesis (Fig. 10). (2) Inclusion of PGE in the # incubation medium reduced subsequent PGE -stimu# lated cAMP production whether or not IL-1 was included in the incubation medium (Fig. 12), suggesting that chronic exposure to PGE reduced expression # of functional PGE receptors. (3) Inhibition of endogenous PGE synthesis (by including indomethacin # in the incubation medium with IL-1β) increased the subsequent IL-1 effect when compared to cells incubated with only IL-1β (Fig. 13). Presumably, de novo synthesis of PGE during the incubation period, which # was probably induced by IL-1, exerted a negative effect on expression of active cell-surface PGE receptors. By inhibiting the cyclo-oxygenase enzyme and reducing de novo PGE synthesis, this negative influence on PGE # receptor expression was removed, allowing for an ‘ enhanced ’ IL-1 effect to occur. It is not likely that IL-1-induced synthesis of PGE contributed to the # IL-1 effect itself since basal cAMP was never seen to increase in PE cells incubated with IL-1β and stimulated with medium (Figs 4, 6 and 9–13). Additional studies, particularly Scatchard analysis of PGE binding to PGE receptors on PE cells, will be # required to uncover the specific molecular events responsible for the IL-1 effect. Although the present results do not address the physiological significance of the IL-1 effect, it is

102

interesting to note that the IL-1 effect was inhibited by TGF-β2 (Fig. 11). TGF-β has potent immunosuppressive and anti-inflammatory properties (Wahl, 1992), is found in high concentrations in normal intraocular fluids (Cousins et al., 1991 ; Jampel et al., 1990 ; Lutty et al., 1993), and is synthesized by several intraocular tissues (Knisely et al., 1991 ; Lutty et al., 1993 ; Pasquale et al., 1993). TGB-β immunoreactivity ( β2 and β3) has been detected in human and rabbit PE and NPE (Peress and Perillo, 1994) and bovine PE cells synthesize TGF-β (primarily β2) in culture (Helbig et al., 1990). In murine T cells, secretion of latent TGF-β and its processing into the active form were increased by IL-1 (Bristol et al., 1990). Furthermore, expression of TGF-β genes ( β2 and β3) is stimulated by cAMP through activation of protein kinase A and subsequent phosphorylation and activation of the cAMP response element binding protein (Kelly, O’Reilly and Rizzino, 1992 ; Lafyatis et al., 1990). This raises the possibility that one physiological consequence of the IL-1 effect is stimulation of TGF-β genes and increased expression of TGFβ. Since TGF-β inhibits the IL-1 effect, it is tempting to speculate that IL-1-induced enhancement of TGF-β2 expression is part of a negative feedback mechanism regulating the physiological consequences of the IL-1 effect. This would be consistent with the observations listed above and with the fact that TGF-β can counteract many of the actions of IL-1 (Gamble and Vadas, 1988, Girard, Matsubara and Fini, 1991, Turner et al., 1991). Furthermore, the TGF-β2 concentrations that inhibited the IL-1 effect (20–2000 pg ml−" ¯ 0±8–80¬10−"# ), were well within the physiological range and comparable to concentrations of TGF-β measured in aqueous humor by Allen et al. (1996) (active TGF-β ¯ 24¬10−"#  ; total TGF-β activity ¯ 204¬10−"# ) and Cousins et al. (1991) (active TGF-β ¯ 20¬10−"#  ; total TGF-β activity ¯ 90¬10−"# ). Since IL-1 is generally considered to be an inflammatory cytokine (Dinarello, 1988), the physiological consequences of the IL-1 effect could contribute to the overall intraocular inflammatory response. Protzman and Woodward (1990) reported that prostaglandins activating the EP receptor subtype, at which PGE is # # the most active endogenous ligand, induce breakdown of the blood-aqueous barrier in rabbits. However, since inhibition of prostaglandin biosynthesis with nonsteroidal antiinflammatory drugs produces only mild attenuation of the uveitic response (Bhattacherjee, Williams and Eakins, 1983 ; Fleisher, 1988), the IL-1 effect can only be one of several mechanisms contributing to intraocular inflammation. This should not be surprising since redundancy of effect with respect to the actions of inflammatory mediators is a salient characteristic of the inflammatory process. Furthermore, in light of the central role of the ciliary epithelium in secretion of aqueous humor and the importance of cAMP in the regulation of this secretory

L. N. F L E I S H E R E T A L.

process (Sears, 1985), the possibility that the IL-1 effect may play a role in changes in intraocular pressure that occur during uveitis should also be considered. Acknowledgements The authors would like to thank Dr Yasushi Ogawa, Celtrix Pharmaceuticals, Santa Clara, CA, U.S.A., for the generous gift of TGF-β2. Supported by NIH grant EY-08688 and funds from the state of North Carolina.

References Allen, J. B., McGahan, M. C., Ogawa, Y., Sellon, D. C., Clark, B. D. and Fleisher, L. N. (1996). Intravitreal transforming growth factor-β2 decreases cellular infiltration in endotoxin-induced ocular inflammation in rabbits. Curr. Eye Res. 15, 95–103. Asano, T., Pedersen, S., Scott, C. W. and Ross, E. W. (1984). Reconstitution of catecholamine-stimulated binding of guanosine 5«-O-(3-thiotriphosphate) to the stimulatory GTP-binding protein of adenylate cyclase. Biochemistry 23, 5460–5466. Ashby, B. (1986). Kinetic evidence indicating separate stimulatory and inhibitory prostaglandin receptors on platelet membranes. J. Cyclic Nucleotide Protein Phosphorylation Res. 11, 291–300. Beutler, B. and Cerami, A. (1987). The endogenous mediator of endotoxic shock. Clin Res. 35, 192–197. Bhattacherjee, P. and Henderson, B. (1987). Inflammatory responses to intraocularly injected interleukin 1. Curr. Eye Res. 6, 929–934. Bhattacherjee, P., Rhodes, L. and Paterson, C. A. (1993). Prostaglandin receptors coupled to adenylyl cyclase in the iris-ciliary body of rabbits, cats and cows. Exp. Eye Res. 56, 327–333. Bhattacherjee, P., Williams, R. N. and Eakins, K. E. (1983). A comparison of the ocular anti-inflammatory activity of steroidal and nonsteroidal compounds in the rat. Invest. Ophthalmol. Vis. Sci. 24, 1143–1146. Bristol, L. A., Ruscetti, F. W., Brody, D. T. and Durum, S. K. (1990). IL-1 alpha induces expression of active transforming growth factor-beta in nonproliferating T cells via a post-transcriptional mechanism. J. Immunol. 145, 4108–14. Cannon, J. G., Burton, B. D., Wingfield, P., Schmeissner, U., Losberger, C., Dinarello, C. A. and Shaw, A. R. (1989). Rabbit IL-1. Cloning, expression, biologic properties, and transcription during endotoxemia. J Immunol. 142, 2299–306. Cilluffo, M. C., Cohen, B. N. and Fain, G. L. (1991). Nonpigmented cells of the rabbit ciliary body epithelium. Tissue culture and voltage-gated currents. Invest. Ophthalmol. Vis. Sci. 32, 1619–29. Cohen-Luria R. and Rimon, G. (1992). Prostaglandin E can # bimodally inhibit and stimulate the epididymal adipocyte adenylyl cyclase activity. Cell Signal. 4, 331–5. Cole, D. A. (1984). Ocular fluids. In The Eye. (Ed. Davson, H.). Vol. 1A. Pp. 269–390. Academic Press : Orlando, U.S.A. Cole, D. F. and Unger, W. G. (1973). Prostaglandins as mediators for the responses of the eye to trauma. Exp. Eye Res. 17, 357–68. Cousins, S. W., McCabe, M. M., Danielpour, D. and Streilein, J. W. (1991). Identification of transforming growth factor-beta as an immunosuppressive factor in aqueous humor. Invest. Opthalmol. Vis. Sci. 32, 2201–11.

I L-1β I N C R E A S E S P G E2-S T I M U L A T E D c A M P

Dinarello, C. A. (1988). Biology of interleukin 1. Faseb J. 2, 108–15. Dinarello, C. A. (1991). Interleukin-1 and interleukin-1 antagonism. Blood 77, 1627–52. Eakins, K. E. (1977). Prostaglandin and nonprostaglandin mediated breakdown of the blood–aqueous barrier. Exp. Eye Res. 25 (Suppl.), 483–98. Fain, G. L. and Farahbakhsh, N. A. (1989). Voltageactivated currents recorded from rabbit pigmented ciliary body epithelial cells in culture. J. Physiol. (Lond.). 417, 83–103. Finch, R. J., Sreedharan, S. P. and Goetzl, E. J. (1989). Highaffinity receptors for vasoactive intestinal peptide on human myeloma cells. J. Immunol. 142, 1977–82. Fleisher, L. N. (1988). Effects of inhibitors of arachidonic acid metabolism on endotoxin-induced ocular inflammation. Curr. Eye Res. 7, 321–7. Fleisher, L. N., Ferrell, J. B. and McGahan, M. C. (1992). Synergistic uveitic effects of tumor necrosis factor-α and interleukin-1β. Invest. Ophthalmol. Vis. Sci. 33, 2120–7. Fleisher, L. N., Ferrell, J. B. and McGahan, M. C. (1995). Inflammation induced changes in adenosine 3«,5«-cyclic monophosphate production by ciliary epithelial cell bilayers. Exp. Eye Res. 60, 165–71. Fleisher, L. N. and McGahan, M. C. (1986). Time course for prostaglandin synthesis by rabbit lens during endotoxin-induced ocular inflammation. Curr. Eye Res. 5, 629–34. Gamble, J. R. and Vadas, M. A. (1988). Endothelial adhesiveness for blood neutrophils is inhibited by transforming growth factor-β. Science 242, 97–9. Gamble, J. R. and Vadas, M. A. (1991). Endothelial cell adhesiveness for human T lymphocytes is inhibited by transforming growth factor-beta 1. J. Immunol. 146, 1149–54. Girard, M. T., Matsubara, M. and Fini, M. E. (1991). Transforming growth factor-β and interleukin-1 modulate metalloproteinase expression by corneal stromal cells. Invest. Ophthalmol. Vis. Sci. 32, 2441–54. Helbig, H., Kittredge, K. L., Gurley, R. C., Thurau, S. R., Palestine, A. G. and Nussenblatt, R. B. (1990). Endotoxin-induced production of inflammatory mediators by cultured ciliary epithelial cells. Curr. Eye Res. 9, 501–5. Hoyng, P. F. J., Verbey, N., Thorig, L. and van Haeringen, N. J. (1986). Topical prostaglandins inhibit traumainduced inflammation in the rabbit eye. Invest. Ophthalmol. Vis. Sci. 27, 1217–25. Jampel, H. D., Roche, N., Stark, W. J. and Roberts, A. B. (1990). Transforming growth factor-β in human aqueous humor. Curr. Eye Res. 9, 963–9. Jumblatt, M. M., Neltner, A. A., Coca-Prados, M. and Paterson, C. A. (1994). EP -receptor stimulated cyclic # AMP synthesis in cultured human non-pigmented ciliary epithelium. Exp. Eye Res. 58, 563–566. Junquero, D. C., Scott-Burden, T., Schini, V. B. and Vanhoutte, P. M. (1992). Inhibition of cytokineinduced nitric oxide production by transforming growth factor-beta 1 in human smooth muscle cells. Journal of Physiology (Lond.) 454, 451–65. Kelly, D., O’Reilly, M. A. and Rizzino, A. (1992). Differential regulation of the transforming growth factor type-beta 2 gene promoter in embryonal carcinoma cells and their differentiated cells. Dev. Biol. 153, 172–5. Knisely, T. L., Bleicher, P. A., Vibbard, C. A. and Granstein, R. D. (1991). Production of latent transforming growth factor-beta and other inhibitory factors by cultured m‘urine iris and ciliary body cells. Curr. Eye Res. 10, 761–71. Kulkarni, P. S. and Mancino, M. (1993). Studies on intraocular inflammation produced by intravitreal

103

human interleukins in rabbits. Exp. Eye Res. 56, 275–9. Lafyatis, R., Lechleider, R., Kim, S. J., Jakowlew, S., Roberts, A. B. and Sporn, M. B. (1990). Structural and functional characterization of the transforming growth factor beta 3 promoter : a cAMP-responsive element regulates basal and induced transcription. J. Biol. Chem. 265, 19128–36. Le, J. and Vilcek, J. (1987). Tumor necrosis factor and interleukin 1 : cytokines with multiple overlapping biological activities. Lab. Invest. 56, 234–248. Limas, C. and Limas, C. J. (1987). Homologous regulation of PGE receptors in rat renal medulla. Arch. Biochem. # Biophys. 259, 89–97. Lowry, O. H., Rosebrough, N. J., Farr, A. L. and Randall, R. J. (1951). Protein measurement with the Folin phenol reagent. J. Biol. Chem. 193, 265–75. Lutty, J. A., Threlkeld, A. B., Crone, S. and McLeod, D. S. (1993). Heterogeneity in localization of isoforms of TGF-β in human retina, vitreous, and choroid. Invest. Ophthalmol. Vis. Sci. 34, 477–87. McNellis, E. L. and Bausher, L. P. (1991). Stimulatory and inhibitory cyclic AMP responses in rabbit ciliary processes after cervical ganglionectomy. Curr. Eye Res. 10, 399–407. Mittag, T. W., Tormay, A. and Podos, S. M. (1987). Vasoactive intestinal peptide and intraocular pressure : Adenylate cyclase activation and binding sites for vasoactive intestinal peptide in membranes of ocular ciliary processes. J. Pharmacol. Exp. Ther. 241, 230–5. Pasquale, L. R., Dorman-Pease, M. E., Lutty, G. A., Quigley, H. A. and Jampel, H. D. (1993). Immunolocalization of TGF-β1, TGF-β2, and TGF-β3 in the anterior segment of the human eye. Invest. Ophthalmol. Vis. Sci. 34, 23–30. Peress, N. S. and Perillo, E. (1994). TGF and TGF-β3 immunoreactivity within the ciliary epithelium. Invest. Ophthalmol. Vis. Sci. 35, 453–7. Pieroni, J., Jacobowitz, O., Chen, J. and Iyengar, R. (1993). Signal recognition and integration by Gs-stimulated adenylyl cyclases. Curr. Opinion Neurobiol. 3, 345–51. Protzman, C. E. and Woodward, D. F. (1990). Prostanoidinduced blood-aqueous barrier breakdown in rabbits involves the EP2 receptor subtype. Invest. Ophthalmol. Vis. Sci. 31, 2463–6. Raviola, G. and Raviola, R. (1978). Intercellular junctions in the ciliary epithelium. Invest. Ophthalmol. Vis. Sci. 17, 958–81. Rosenbaum, J. T. and Boney, R. S. (1992). Activity of an interleukin 1 receptor antagonist in rabbit models of uveitis. Arch. Ophthalmol. 110, 547–9. Rosenbaum, J. T., Rosenbaum, S. O., Waldman, M., Toth, C. and Mohlen, K. (1993). Do interleukin-1, tumor necrosis factor, or complement contribute to endotoxininduced uveitis ? Invest. Ophthalmol. Vis. Sci. 34 (Suppl.), 1143. Rosenbaum, J. T., Samples, J. R., Hefeneider, S. H. and Howes, E. L., Jr. (1987). Ocular inflammatory effects of intravitreal interleukin-1. Arch. Ophthalmol. 105, 1117–20. Sano, N. and Shichi, H. (1993). Bimodal regulation of adenylate cyclase by prostaglandin E receptors in # porcine ciliary epithelium. Prostaglandins Leukotrienes Essen. Fatty Acids. 49, 765–9. Seamon, K. B. and Daly, J. W. (1981). Activation of adenylate cyclase by the diterpene forskolin does not require the guanine nucleotide regulatory protein. J. Biol. Chem. 256, 9799–801. Sears, M. L. (1985). Regulation of aqueous humor flow by the adenylate cyclase receptor complex in the ciliary epithelium. Am. J. Ophthalmol. 100, 194–8.

104

Smith, W. L. (1989). The eicosanoids and their biochemical mechanisms of action. Biochem. J. 259, 315–24. Stevenson, B. R., Siliciano, J. D., Mooseker, M. S. and Goodenough, D. A. (1986). Identification of ZO-1 : a high molecular weight polypeptide associated with the tight junction (zonula occludens) in a variety of epithelia. J. Cell. Biol. 103, 755–66. Turner, M., Chantry, D., Katsikis, P., Berger, A., Brennan, F. M. and Feldmann, M. (1991). Induction of the

L. N. F L E I S H E R E T A L.

interleukin 1 receptor antagonist protein by transforming growth factor-beta. Eur. J. Immunol. 21, 1635–9. Wahl, S. M. (1992). Transforming growth factor beta (TGFβ ) in inflammation : a cause and a cure. J. Clin. Immunol. 12, 61–74. Wong, K. L. and Howes, E. L. J. (1983). Intraocular injection of prostaglandins : Modification of the response to circulating endotoxin. Arch. Ophthalmol. 101, 275–9.