An Ocular Renin-Angiotensin System

Immunohisrochemisrry of Angiotensinogen

Stephen J. Sramek,* Ingolf H. L. Wallow,* Duane A. Tewksbury,-f Curtis R. Brandt,*^: and Gretchen L. Poulsen"

The circulating renin-angiotensin system (RAS) is an important determinant in maintaining adequate systemic blood pressure, and it also may modify organ-specific blood flow. All recognized RAS components have been identified in the eye. In this study, angiotensinogen (ANG) was localized using an affinity-purified antibody and paraffin sections of seven human eyes. An antibody for human serum albumin was used for comparison. The ANG was present selectively in the cytoplasm of the nonpigmented ciliary epithelium (NPCE), more prominently in the pars plana than in the pars plicata. Both ANG and albumin were present in the blood vessel lumina of the uvea and retina. Both antibodies also stained perivascular tissue in the uvea, but not in the retina, reflecting the relative tightness of blood- tissue barriers. The detection of ANG in the NPCE may be significant in view of previous descriptions localizing prorenin and angiotensin-converting enzyme in the same cell layer. Invest Ophthalmol Vis Sci 33:1627-1632,1992

The renin-angiotensin system (RAS), through the tissues are limited. However, it has been measured in production of angiotensin II (All) in the circulation, surgical samples of human vitreous, subretinal fluid, plays an important role in the control of blood pres- and aqueous; concentrations in the vitreous were sure and electrolyte homeostasis.1 Recent studies correlated with plasma levels.5 The authors concluded have suggested that components of the RAS also are that ANG diffused into the vitreous from the plasma. present in peripheral tissues,2 including the eye. All We were interested (1) in the immunostaining pat- recognized RAS components have been identified in tern of ANG to continue our work on the anatomic the eye.3"13 The function of "local" RAS systems is relationship of RAS components in ocular tissue6 and unknown. However, because All is a potent vasocon- (2) in determining if the immunostaining pattern we strictor and angiogenic factor,14 it is possible that an found would support previous work suggesting that ocular RAS may be involved in the control of retinal plasma diffusion may be the source of intraocular blood flow or in retinovascular disease. ANG.5 To accomplish this, we determined the immu- Angiotensinogen (ANG), or renin substrate, is an nostaining pattern of ANG in ocular tissue and com- obligatory component for the eventual production of pared it with that of albumin, a serum glycoprotein of AIL Plasma ANG is derived primarily from the liver. approximately the same molecular weight. Although it is difficult to demonstrate ANG in hepatocytes by immunostaining, its messenger RNA can Materials and Methods be found by both in situ hybridization and northern Human ANG was purified from fresh-frozen blot analysis.1516 Currently, studies of ANG in ocular plasma according to previously published proce- dures.1718 Antiserum specific for human ANG was prepared by immunizing rabbits with purified human From the Departments of ""Ophthalmology and ^Medical Micro- biology and Immunology, University of Wisconsin Medical ANG. The immunoglobulin (Ig) G fraction of the an- School, Madison, and the fMarshfield Medical Research Founda- tiserum was isolated by affinity chromatography on tion, Marshfield, Madison, Wisconsin. Protein A-agarose (Pierce, Rockford, IL). A 0.9 X 10- Supported by a grant from the Diabetes Research and Education cm column was equilibrated with a mixture of 0.05 M Foundation, Bridgewater, New Jersey (SJS), by National Institutes Tris and 0.15 M NaCl, pH 7.0, and 2 ml of antiserum of Health (Bethesda, Maryland) grants EY01634 (IHLW) and EY07336 (CRB), and by a grant of the Miller Foundation (IHLW, (diluted 1:1 with the equilibration buffer) was applied Marshfield, Wisconsin). Also supported in part by Research to Pre- to the column. We eluted the column with equilibra- vent Blindness, Inc., New York, New York. tion buffer until the absorbance at 280 nm returned to Submitted for publication: August 19, 1991; accepted November baseline value. The IgG fraction was eluted with 0.1 8, 1991. M glycine HC1, pH 2.5. The pH of the eluted fractions Reprint requests: Stephen J. Sramek, MD, PhD, Department of Ophthalmology, Clinical Sciences Center, 600 Highland Avenue, was raised immediately by addition of 1 M Tris HC1, Madison, WI 53792. pH 8.0. The IgG fraction was concentrated, and the

1627

Downloaded from iovs.arvojournals.org on 10/02/2021 1628 INVESTIGATIVE OPHTHALMOLOGY G VISUAL SCIENCE / April 1992 Vol. 33

buffer was exchanged with the equilibrating buffer on analysis were resuspended in a volume of sodium do- an Amicon (Beverly, MA) membrane concentrator. decyl sulfate (SDS) sample buffer (0.0625 M Tris Purified ANG was coupled to a solid support HC1, pH 6.8, glycerol 0.1%, SDS 0.2%, 0.6 M 2-mer- (Aminolink gel; Pierce, Rockford, IL) according to captoethanol, and bromphenol blue 0.001%) equal to the manufacturer's instructions. The conjugated the tissue volume. The samples then were sonicated ANG was equilibrated with a mixture of 0.05 M Tris three times for 30 sec at 50% power using a Branson HC1 and 0.15 M NaCl, pH 7.0, and the IgG fraction model 200 sonifier cell disruptor (Branson Ultrason- was applied with gentle agitation for 1 hr at room ics, Corp., Danbury, CT) and boiled for 3 min. A 5-/*l temperature. The unbound protein was eluted with aliquot was removed and used to determine the pro- the equilibration buffer, and the ANG-specific anti- tein concentration with a commercially available pro- bodies were eluted with 0.1 M glycine HC1, pH 2.5. tein assay (Biorad, Richmond, CA). The samples then We collected 100 ^1 fractions of the eluate in tubes were stored at -20°C. For analysis, the samples were containing 100 /*1/M Tris, pH 8.0. The ANG-specific thawed, boiled for 3 min, and equal amounts of pro- antibodies were dialyzed against a mixture of 0.01 M tein underwent electrophoresis in discontinuous poly- sodium phosphate and 0.15 M NaCl, pH 7.0, and acrylamide 10% gels as described previously.19 The concentrated on an Amicon ultrafilter. Nonimmune proteins were transferred electrophoretically to nitro- rabbit IgG was purified by affinity chromatography cellulose using a Biorad Transblot cell. on a Protein A column as described. Goat anti-hu- Immunoblotting was done using an earlier proce- man serum albumin was purchased from Cappel dure.20 The primary rabbit anti-ANG and antialbu- (West Chester, PA). min antibodies were used at a final concentration of Seven eyes from seven subjects were selected from 0.5 /ig/ml total IgG. The secondary goat anti-rab- the University of Wisconsin Eye Bank based on the bit IgG alkaline phosphatase-conjugated antibody following criteria: no known eye disease, normoten- (Sigma, St. Louis, MO) was used at a final concentra- sive, nondiabetic, no history of use of an angiotensin- tion of 1.7 /ig/ml. Immunoblots were developed using converting enzyme inhibitor, and no more than 5 hr the nitro blue tetrazolium-bromochloroindoyl phos- after death to the beginning of fixation. The ages of phate substrate system.21 the donors were 38, 44, 44, 54, 61, 69, and 74 years. The eyes were bisected vertically and placed in neu- Results tral buffered formalin 10%, pH 7.4 for 24-48 hr. One Figure 1A shows an immunoblot demonstrating eye was used for both immunoblots and immuno- the specificity of the anti-ANG antibody, the presence staining of tissue sections. The cornea was removed at of ANG in various ocular tissue fractions, and an esti- the limbus, and the iris diaphragm was excised. The mate of relative ANG amounts in these fractions. The eye then was cut at the ora serrata, and vitreous was lane containing purified human ANG was overloaded collected at the vitreous base and from the posterior to reveal minor components. It shows a family of ma- eyecup. A 7-mm wedge of tissue was removed from jor bands in the molecular weight range appropriate the posterior cup, including ciliary body, retina, and for human ANG (reportedly, 61.4 and 65.4 kD,17 and sclera, and placed in formalin 10% for later immuno- approximately 58.3 and 61.8 kD, as calculated from staining. In the remaining cup, the neurosensory ret- our blot). It also contains high molecular weight ag- ina was detached from the retinal pigment epithelium gregates of antigen as described previously for stored 17 and dissected around the optic nerve. Finally, sclera samples. The lanes for vitreous, retina, iris, and cili- and much of the ciliary muscle were dissected from ary body obtained from fresh ocular tissue fractions, the ciliary body, and the rest of the pars plana and by contrast, showed only the two bands of nonaggre- pars plicata were pooled. In summary, our dissection gated ANG. The ANG immunostaining was more in- procedure resulted in one tissue wedge fixed for immu- tense for vitreous and ciliary body than for retina. nostaining of tissue sections, and in four pooled sam- Because the total protein amount loaded in the re- ples (iris, ciliary body, vitreous, and retina), which spective lanes was equivalent, vitreous and ciliary were frozen at —20°C for later immunoblotting. body seem relatively enriched compared with retina. Immunostaining was done on formalin-fixed, par- When we ran identical samples in a separate gel and affin-embedded sections 6-^m thick, using the avi- stained them with silver stain or Coomassie blue for din-biotin-peroxidase technique (Vector, Burlin- total proteins, we found that the iris lane appeared to game, CA).6 Negative control samples consisted of contain less protein (data not shown), suggesting the normal rabbit serum (anti-ANG) or normal goat reduced levels of ANG in the iris sample may be a serum (antialbumin) in place of the primary antibody result of our loading less protein. The total protein at the same immunoglobulin concentration. staining in the other lanes appeared equivalent, allow- Protein gel electrophoresis and immunoblotting ing an approximate comparison of the amount of were done as follows. Tissue samples for immunoblot ANG in retina, ciliary body, and vitreous.

Downloaded from iovs.arvojournals.org on 10/02/2021 No. 5 OCULAR RENIN-ANGIOTENSIN SYSTEM / Sramek er ol 1629

T3 o O CO CD 03 ->, 03 o E a> o? O 03

ili w a> E > DC b < E O DC ±1 > • i kd kd

106- 80-

49.5-

32.5-

27.5-

L R

Fig. 1. Immunoblot analysis of angiotensinogen (ANG) and albumin in ocular samples. (Left) ANG. Fifty micrograms of protein were loaded in each lane, except for the standard ANG control, which contained 2.5 ^g of purified ANG. (Right) Albumin. Twenty fivemicrogram s of protein were loaded in each lane except the normal human serum control, which contained 0.5 n% of total protein. Protein concentrations were determined using the Biorad (Richmond, CA) protein assay kit. Commercially available prestained molecular weight markers (Biorad) were used as standards. The size of the marker proteins (kd) are shown for each immunoblot.

Figure IB shows immunostaining for serum albu- sels in all (albumin, all seven) or nearly all eyes (ANG, min. Similar to ANG, serum albumin staining was five of seven). In all sections examined, positive and more intense for vitreous and ciliary body than for seemingly negative blood vessels existed side by side. retina. Again, the iris samples appeared to contain less This was thought to reflect the different perfusion sta- total protein than the ciliary body and retina (data not tus of different blood vessels at the time of harvesting shown). the tissue and/or different retention of coagulated in- Immunostaining of human eye tissue sections travascular protein during tissue processing and sec- showed similar, but not identical, patterns for ANG tioning. In general, intravascular staining was more and albumin. Both ANG and albumin were present in intense with albumin than with ANG. Perivascular intra- and extravascular spaces with their apparent and anterior stroma staining with ANG was less in- concentration corresponding to known ocular diffu- tense and less consistent than intravascular staining. sion barriers. However, in the nonpigmented ciliary The ciliary body (Figs. 3, 4) showed staining of epithelium (NPCE), only ANG was detected. some blood vessels of the pars plicata and consistent In the iris, intra vascular staining and staining of the dense staining of the connective tissue core with both perivascular and anterior stromal border were ob- albumin and ANG. The cytoplasm of the NPCE of served (Fig. 2). Intravascular staining was present the pars plicata was negative for albumin (with one with both antigens in the lumina of some blood ves- exception where the staining was weak), but positive

Downloaded from iovs.arvojournals.org on 10/02/2021 1630 INVESTIGATIVE OPHTHALMOLOGY b VISUAL SCIENCE / April 1992 Vol. 33

Figs. 2-6. (top to bottom). Each row of four horizontally arranged illustrations represents one figure. In each row, illustrations are referred to as (A) (far left) through (D)(far right). Fig. 2. Immunostaining of normal human iris with normal goat serum (NGS: A), anti-human albumin (B). antihuman angiotensinogen (ANG) (C). and normal rabbit serum (NRS: D). For albumin and ANG staining, residual vascular contents can be seen as a dense band on the inner surface of endothelial cells, while the cells themselves appear negative. In addition, staining of the anterior iris stroma is seen with antialbumin. Paraffin sections. X70. Fig. 3. Immunostaining of a pars plicata process for albumin (B; x240) and for ANG (C; X23O), with NGS (A, x 160) and NRS (D, X150) as controls. Concentrated stain is seen in the core of the process for both antigens{B, C). With albumin (B), stain also is seen along the inner surface of the nonpigmented ciliary epithelium (NPCE), but the underlying cells are negative. With ANG (C) the staining along the inner surface is not detected, but the cytoplasm of some cells of the NPCE is diffusely positive. Paraffin sections. Fig. 4. Area of the pars plana immunostained for albumin (B; X480) and ANG (C; X47O), with NGS (A, X27O) and NRS (D; x260) as controls. With albumin (B) the NPCE is negative, but segments of stain along the inner surface can be seen. With ANG (C). the cytoplasm of the NPCE is diffusely positive. Paraffin sections. Fig. 5. Immunostaining of the choriocapillaris with NGS (A; x27O). antialbumin (B; X460), anti-ANG (C; x460). and NRS (D; x27O). Staining of the connective tissue septae is seen for albumin and ANG. Paraffin sections. Fig. 6. Immunostaining of retina and vitreous using NGS (A), antialbumin (B). anti-ANG (C), and NRS (D). The lumina of retinal blood vessels contain antigens; the adjacent retinal tissue is negative. The vitreous is positive for both antigens. Paraffin sections; original magnification XI40.

Downloaded from iovs.arvojournals.org on 10/02/2021 No. 5 OCULAR RENIN-ANGIOTENSIN SYSTEM / Sromek er ol 1631

for ANG in almost one half of specimens (three of Intra- and perivascular tissue staining seemed to seven). In the pars plana, the cytoplasm of the NPCE confirm these assumptions. Immunostaining of the was consistently negative for albumin (all seven), but vitreous was positive for both proteins. In the retina, it usually was positive for ANG (five of seven). Along staining was observed only inside vascular spaces, re- the inner surface of the NPCE, there were segments of flecting its tight blood-retinal barrier. The iris, with its positive staining with albumin but not with ANG. vascular endothelial junctions intermediate between The adjacent vitreous was almost consistently posi- the retina and ciliary body, showed mild perivascular tive for both (all six). (In one specimen vitreous was staining of its stroma. The ciliary body and choroid, removed and not available for tissue sections, but it with its fenestrated capillaries, demonstrated dense was used for immunoblotting which was positive for stromal staining. We also observed one difference in ANG, Fig. 1A.) immunostaining between ANG and albumin. The In the choroid (Fig. 5), the connective tissue septa NPCE showed cellular staining with anti-ANG but of the choriocapillaris stained heavily and consis- not with antialbumin. Determining the origin of tently (all seven) for both proteins, and staining ex- ANG in the NPCE will require additional study, but tended into the connective tissue portion of Bruch's possible sources include uptake from the vitreous or membrane. The basal portion of the adjacent retinal vascular core or synthesis by the NPCE. pigment epithelium was negative. Weak staining of its Although evidence from other groups suggests that central and apical portion may be undetectable be- an ocular RAS may exist, the precise localization of cause of the presence of melanin and lipofuscin pig- the RAS components and the sites for their synthesis ment. and/or activation remain to be elucidated. The detec- The lumina of retinal blood vessels (Fig. 6) were tion of ANG in the cytoplasm of the NPCE may be consistently positive for both albumin (all seven) and significant in view of previous descriptions of prore- ANG (six of seven). The adjacent perivascular retinal nin and angiotensin-converting enzyme in the tissue and the inner limiting membrane were negative NPCE.6'7 If prorenin is converted to renin in NPCE, (all seven). all components necessary for All generation are present. However, prorenin has not been found to be con- Discussion verted to active renin in the eye, and renin has not been studied in the NPCE to our knowledge. There This is the first report to our knowledge of the im- has been speculation that membrane-bound prorenin munohistochemical staining of ocular tissue for the is an active form,24 which would then provide a com- glycoprotein ANG. It extends previous work by plete RAS in the NPCE. The RAS components also others in which ANG levels were measured in have been found in the vitreous,5 and therefore, the aqueous, vitreous, and subretinal fluid.5 Our experi- vitreous itself may be a site for all All generation, ments confirm the presence of ANG in vitreous and, either alone or in concert with the NPCE. In view of in addition, demonstrate the pattern of distribution in the potent vasoconstrictive effects of All,1 its ability to the uvea and retina. stimulate angiogenesis,14 the demonstrated sensitivity As a reference, we also stained for albumin, a serum of retinal blood vessels to intravitreal injections of glycoprotein of approximately the same molecular All,11 and localization of RAS components to the weight as ANG. Albumin is assumed to be "inert" NPCE and vitreous, further investigations are because it is neither an enzyme or a substrate. If both warranted concerning the mechanism by which an ANG and albumin were derived solely from plasma, ocular RAS might function in both healthy and dis- we might expect immunostaining for ANG and albu- eased eyes. min to be similar in various ocular tissues. With re- spect to the vitreous, plasma has been suggested as the Key words: renin-angiotensin system, angiotensinogen, im- probable source of soluble protein,522 although the munohistochemistry, localization, ocular, human pathways for movement of protein through the barrier of the NPCE have not been identified. Report- References edly, the ratio of concentrations in plasma versus vitre- 5 ous fluid is the same for ANG and albumin, ie, 20:1. 1. Oparil S and Haber E: The renin-angiotensin system. N Engl J Ratios for the concentrations of ANG and albumin in Med 291:389, 1974. plasma versus the stroma of the retina, iris, and ciliary 2. Re RN: Cellular biology of the renin-angiotensin systems. Arch body and choroid are not known, but they should Intern Med 144:2037, 1984. reflect the tightness of their respective blood-tissue 3. Igic R and Kojovic V: Angiotensin I converting enzyme (kinin- 23 ase II) in ocular tissues. Exp Eye Res 30:299, 1980. barriers. 4. Vita JB, Anderson JA, Hulem CD, and Leopold IH: Angioten-

Downloaded from iovs.arvojournals.org on 10/02/2021 1632 INVESTIGATIVE OPHTHALMOLOGY & VISUAL SCIENCE / April 1992 Vol. 33

sin-converting enzyme activity in ocular fluids. Invest Ophthal- 14. Fernandez LA, Twickler J, and Mead A: Neovascularization mol Vis Sci 30:255, 1981. produced by angiotensin II. J Lab Clin Med 105:141, 1985. 5. Danser AHJ, van den Dorpel MA, Deinum J, Derkx FHM, 15. Campbell DJ and Habener JF: Cellular localization of angio- Franken AAM, Peperkamp E, de Jong PTVM, and Schale- tensinogen gene expression in brown adipose tissue and mesen- kamp MADH: Renin, prorenin and immunoreactive renin in tery: Quantification of messenger ribonucleic acid abundance vitreous fluidfro m eyes with and without diabetic retinopathy. using hybridization in situ. Endocrinology 121:1616, 1987. J Clin Endocrinol Metab 68:160, 1989. 16. Gaillard-Sanchez I, Bruneval P, Clauser E, Belair MF, da Silva 6. Sramek SJ, Wallow IHL, Day RP, and Ehrlich EN: Ocular JL, Bariety J, Camilleri JP, and Corvol P: Successful detection renin-angiotensin: Immunohistochemical evidence for the by in situ cDNA hybridization of three members of the serpin presence of prorenin in eye tissue. Invest Ophthalmol Vis Sci family: Angiotensinogen, a, protease inhibitor, and antithrom- 29:1749, 1988. bin III in human hepatocytes. Mod Pathol 3:216, 1990. 7. Strittmatter SM, Braas KM, and Snyder SH: Localization of 17. Tewksbury DA: Angiotensinogen. Fed Proc 42:2724, 1983. angiotensin converting enzyme in the ciliary epithelium of the 18. Tewksbury DA, Dart RA, and Travis J: The amino terminal rat eye. Invest Ophthalmol Vis Sci 30:2209, 1989. amino acid sequence of human angiotensinogen. Biochem 8. Laliberte M, Laliberte F, Alhenc-Gelas F, and Chevillard C: Biophys Res Commun 99:1311, 1981. Immunohistochemistry of angiotensin I converting enzyme in 19. Brandt CR, Knupfer PB, Boush GA, Gausas RE, and Chandler rat eye structures involved in aqueous humor regulation. Lab JW: In vivo inductions of la expression in murine cornea after Invest 59:263, 1988. intravitreal injection of interferon-7. Invest Ophthalmol Vis 9. Rockwood EJ, Fantes F, Davis EB, and Anderson DR: The Sci 31:2248, 1990. response of the retinal vasculature to angiotensin. Invest Oph- 20. Towbin H, Staehelin T, and Gordon J: Electrophoretic transfer thalmol Vis Sci 28:676, 1987. of protein from polyacrylamide gels to nitrocellulose sheets: 10. Ferari-Dileo G, Davis EB, and Anderson DR: Angiotensin Procedure and some applications. Proc Natl Acad Sci USA binding sites in bovine and human retinal blood vessels. Invest 76:4350, 1979. Ophthalmol Vis Sci 28:1747, 1987. 21. Harlow E and Lane D: Antibodies: A Laboratory Manual 11. Ferari-Dileo G, Ryan JW, Rockwood EJ, Davis EB, and An- Cold Spring Harbor, NY, Cold Spring Harbor Laboratory, derson DR: Angiotensin-converting enzyme in bovine, feline, 1988, p. 505. and human ocular tissues. Invest Ophthalmol Vis Sci 29:876, 22. Chen CH and Chen SC: Studies on soluble proteins of vitreous 1988. in experimental animals. Exp Eye Res 32:381, 1981. 12. Ferari-Dileo G, Davis EB, and Anderson DR: Angiotensin II 23. Moses RA and Hart WM Jr, editors: Adler's Physiology of the binding receptors in retinal and optic nerve head blood vessels. Eye: Clinical Application, 8th ed. St. Louis, CV Mosby, 1987, Invest Ophthalmol Vis Sci 32:21, 1991. pp. 188-190. 13. Deinum J, Derkx FHM, Danser AHJ, and Schalekamp 24. Sealey JE, Glorioso N, Itskovitz J, and Laragh JH: Prorenin as MADH: Identification and quantification of renin and prore- a reproductive hormone: New form of the renin system. Am J nin in the bovine eye. Endocrinology 126:1673, 1990. Med 81:1041, 1986.

Downloaded from iovs.arvojournals.org on 10/02/2021