Eye diseases and proteins controlling visual transduction

Biochimie, 69 (1987) 371- 377

371

© Soci~t~ de Chimie biologique/Elsevier, Paris

Eye diseases and proteins controlling visual transduction Eric VADOT Service d'Ophtalmologie, Centre Hospitalier, B.F. 1125, 73011 Chambdrv Cedex, France (Received 3-3-1987, accepted after revision 4-5-1987)

Summary - Retinal S antigen is a soluble protein found in abundance in photoreceptor cells. Immunization of laboratory animals with this antigen in adjuvant induces experimental autoimmune uveoretinitis. Cellular immunity plays a major role in this condition. Autoimmune responses toward retinal S antigen are often observed in patients with retinal inflammatory diseases, however, these responses are usually secondary to local tissue damage. The S antigen is identical to the 48 K protein characterized in rod outer segments by its light-dependent binding to the disk membrane in the presence of ATP. This protein binds specifically to photoexcited and phosphorylated rhodopsin, and quenches the activity of the light-dependent cGMP-phosphodiesterase, most probably because it competes with transducin. There is no evidence for any direct inactivation of phosphodiesterase by 48 K protein. In view of the numerous similarities between the photoreceptor enzyme cascade and hormone-activated cyclase systems, a related protein could be involved in the desensitization of hormonal systems. experimental autoimmune uveoretinitis / retinal S antigen / 48 K protein I desensitization

R6sum6 - Pathologie ophtalmologique et prot6ines de contr61e de la transduction visueHe. L"antig~ne S rdtinien est une protdine soluble abondante, localisde aux photordcepteurs, dont l'injection en presence d'adjuvant induit chez l'animal une uvdordtinite autoimmune expdrimentale. Une rdponse autoimmune contre cet antigone peut ~tre observde au cours de nombreuses affections rdtiniennes humaines, habituellement de nature inflammatoire. II a dtd possible rdcemment de l'identifier ~ la protdine 48 K, caractdrisde par sa capacitd de fixation aux disques des segments externes de batonnets aprbs illumination en prdsence d'A TP. La fixation de cette protdine se fait dlectivement sur la rhodopsine photoactivde et phosphorylde par la rhodopsine kinase, ce qui bloque complbtement l'activation de la transducine et de la cGMP-phosphodiestdrase. Ce processus pourrait ~tre un modble de la ddsensibilisation des rdcepteurs hormonaux. uvdo-r~tinite autoimmune expdrimentale I antigone S rdtinien I protdine 48 K I d~sensibUisation

Introduction

The study of autoimmune ocular diseases has led immunologists to the purification of a soluble protein, called 'retinal S antigen', found in abundance in the photoreceptor cells of the retina.

The so-called '48 K protein' was independently characterized in rod outer segments by its lightdependent binding to disk membranes. The properties of retinal S antigen, its identity with the 48 K protein, and its biochemical and physiological properties are reviewed here.

372

E. Vadot

Retinal S antigen and its immunological properties The notion of autoimmunity as a pathogenic mechanism in ocular disease was first introduced by Elschnig (1910) to explain sympathetic ophthalmia. This very rare disease is a severe bilateral intraocular inflammation (uveitis) arising a variable period of time following a penetrating injury to one eye, or ocular surgery. Blindness was formerly its usual outcome: this explains why ophthalmologists are so interested in this disease. That the uveal pigment (melanin) liberated at the time of ocular injury could be the immunizing autoantigen was suggested by Elschnig as a working hypothesis. This notion was unfortunately accepted as an established fact during more than 50 years, until the demonstration in 1965 by Wacker [1] of a much stronger autoimmune activity in the retina than in the uvea: an experimental autoimmune uveoretinitis is much easier to induce by immunization of laboratory animals with retinal extract in complete Freund's adjuvant than with uveal tissue.

Purification and distribution o f retinal S antigen Retinal S antigen, identified as the main active fraction in soluble retinal extracts for the induction of experimental autoimmune uveoretinitis, was subsequently purified by Wacker et al. [2] and by Dorey and Faure [3]. The isolation procedures include extraction of the retina into aqueous solution, ammonium sulfate precipitation at 50°70 saturation, gel Filtration on Sephadex G - 1 5 0 or Ultrogel AcA34, and ion-exchange chromatography or isoelectrofocusing. A rapid and simple method recently published by Wilden et al. [4] uses affinity purification. The purified S antigen is a soluble protein of about 50 kDa. Immunohistochemical staining methods show that S antigen localizes selectively in the photoreceptor cells. Its local concentration is very high, and close to that of rhodopsin [5]. Retinal S antigen has also been found in the pineal gland [6, 7]. This is not surprising, since homologies between photoreceptors and pinealocytes are well known, and the gland has a photoreceptive function in lower vertebrates. Moreover, the pineal gland is often affected by an inflammatory process in the course of experimental autoimmune uveoretinitis [7]. However, S antigen has never been found in any other animal

tissue. This is a good example of what immunologists call organ specificity. On the other hand, antibodies raised against retinal S antigen are almost devoid of species specificity: proteins from different mammalian species are antigenically very similar, and some epitopes are present in all vertebrates and even in some invertebrates [8]. This implies a high degree of conservation for the primary structure, and an important physiological function.

Experimental autoimmune uveoretinitis This disease is characterized by an inflammatory process involving the outer retina, but originating in posterior uvea (choroid) and retinal vessels, whence the name uveoretinitis (Fig. 1). The histopathological changes differ markedly from species to species, owing to different vascular patterns in the retina. Moreover, significant dose-related differences are found in the nature of the lesions produced: an inoculum of 1 #g of homologous retinal S antigen with complete Freund's adjuvant suffices to induce a mild lymphocytic infiltration in the choroid of most injected guinea pigs, whereas a panophthalmitis with complete destruction of the retina follows the injection of a 50 #g dose [9]. Purified retinal S antigen preparations obtained from different mammalian species are similarly effective in inducing the uveore!initis: the disease is easily observed after inoculation of a guinea pig with an extract of its own retina [10], but the bovine S antigen is similarly active, and is therefore the most often used preparation. A detailed account of the immunopathological characteristics of experimental autoimmune uveoretinitis is not attempted here. The interested reader will find such information in reviews by Faure [9], and Gery, Mochizuki and Nussenblatt [11]. Numerous studies show that cell-mediated immunity plays the major role in the pathogenesis of uveoretinitis: for example, the injection of activated lymphocytes from an immunized animal into a naive recipient induces the disease [12, 13]. Only the helper/inducer T lymphocytes are capable of transferring the uveoretinitis. However, they do not necessarily induce the ability to synthesize autoantibodies [14]. Another interesting finding has been reported by de Kozak and her colleagues: the injection of S antigen alone, without any adjuvant, fails to induce uveoretinitis. On the contrary, it appears to prevent the disease if its induction is further attempted according to the usual protocol. Similarly, active experimental autoimmune uveoretinitis can be reversed by injections of

Proteins controlling visual transduction

373

Fig. 1. Experimental autoimmune uveoretinitis in the rat, immunized by a single footpad injection of 30/zg of bovine S antigen mixed with 0.1 ml of complete Freund's adjuvant. Left: first manifestations, 2 weeks after injection. A few inflammatory cells can be seen, notably in the subretinal space (bottom). The retinal structure is otherwise normal. Right: typical histopathological picture, 3 weeks after injection. Abundant mononuclear cell infiltrates are seen in choroid and subretinal space (bottom), and around retinal vessels (top right). The photoreceptor cell layer is almost entirely destroyed. These micrographs were made by Dr. Y. de Kozak (INSERM U86, Paris).

S antigen alone, and a complete cure may even be obtained [!5].

Retinal autoimmunity in human diseases Experimental allergic uveoretinitis is often considered as a model for some ocular inflammations suspected of having autoimmune etiology: (1) Sympathetic ophthalmia is a severe bilateral uveitis that occurs following a penetrating injury to one eye and after a variable period of time (from 10 days to many years). A small degree of meningeal irritation is commonly observed. (2) Vogt-KoyanagiHarada disease, a bilateral uveitis, is clinically similar to sympathetic ophthalmia, with symptoms including meningeal irritation, dysacusis and eventually whitening of hair and eyelashes. (3) Birdshot retinochoroidopathy is a rare bilateral posterior uveitis with cream colored lesions throughout the eye fundus, and a prominent retinal vasculitis. The HLA-A 29 antigen is found in most patients with this condition. However, autoimmune phenomena are not always present in the course of these d!seases, and differences exist between the above conditions and

experimental autoimmune uveoretinitis, so that the ,~L~ cannot u~ ~:umidered a good model for any of them. Autoimmune responses against retinal S antigen have been described in a number of retinal diseases, whether degenerative (retinitis pigmentosa, retinal detachment) or inflammatory (onchocerciasis, toxoplasmic retinochoroiditis). Our own work on the latter disease is worth detailing, since it is a good example of the difficulties found in such studies. Toxoplasmic retinochoroiditis is a relatively frequent disease characterized by white inflammatory foci in the retina, usually in the vicinity of healed, pigmented scars. It heals within some months, but relapses occur in more than 50% of all cases. It usually follows a congenital infection by a sporozoan parasite, Toxoplasma gondii. It is commonly assumed that the recurrences follow cyst rupture in the retina. A specific cell-mediated autoimmunity against retinal S antigen is frequently found in these patients. It is more often observed during the acute stage than after healing, and in recurrences than during the first outbreak. Therefore, hypersensitivity to retinal S antigen might initiate or support recurrences of toxoplasmic retinochoroiditis.

374

E. Vadot

However, by correlating detailed clinical data with immunological results, a very different picture is obtained: the a u t o i m m u n e response is present less often during the first week t h a n afterwards. This result excludes the possibility o f an a u t o i m m u n e component in the triggering of relapses. Even a supporting role in ocular inflammation is not very l:kely, since the correlation between activity of the lesions and abnormal autoimmune responses is not very strong [16]. Thus, in spite o f their frequency, autoimmune responses are only epiphenomena following tissue damage. It should be emphasized that only a negative result allows a firm conclusion to be obtained in such studies. A careful analysis of clinical data is necessary for that purpose. This is an inherent limitation of such clinical studies. Consequently, a pathogenic role for autoimmunity to the retina has not been unequivocally demonstrated in any h u m a n ocular disease to date.

The 48 K protein regulating light-dependent phosphodiesterase Quite independently, H. Kiihn (1978) characterized the soluble 48 K protein in rod outer segments by its fight-dependent binding to the disk membrane [17], a property suggestive of a regulatory role in visual transduction. The collaboration of vision biochemists with immunologists and ophthalmolo~sts has allowed this protein to be identified as the retinal S antigen [18, 19].

Retinal S antigen characterized as the 48 K protein Biochemical, functional and immunological tests show the identity of these proteins: (1) both proteins migrate in sodium dodecyl sulfate-polyacrylamide gel electrophoresis as polypeptides of about 50 k D a (see Fig. 2, lanes 1, 2, 4, 6). In their native form, they are eluted from gel filtration columns as proteins of about 50 kDa, and are therefore monomeric. (2) Purified retinal S antigen binds to illuminated disk membranes (Fig. 2, lanes 6 and 7). (3) Monoclonal and polyclonal antibodies specific for S antigen specifically recognize the 48 K protein in extracts of rod outer segments. The radioimmunoassay titration curve obtained with purified 48 K protein is practically identical with the standard curve for S antigen (Fig. 3). (4) Purified 48 K protein induces the development o f ::~perimental autoimmune uveoretinitis with the same efficiency as that observed with authentic retinal S antigen,

123

46

92.5 K -66.2

67

89

m -.. P D E

K --qlll~ ~

45 K--

-,,,,,48 K" , "

=i: ;

~

.m*

b~

i

,.,,.,

•',~ T~ --,Tp

U

31 K - - .

_pur,,,.0t]

o

5-antigen-J purified ~ 4 8 K ....

RO S P-~

;ks

S-anti

en ligh t

4'

D

4~

L

4'

4'

4'

4P

.~

.Ib

~

~

D

L

D L

Fig. 2. Biochemical and functional identity of retinal S antigen and 48 K protein, analyzed by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE). Lanes 1 and 2: overloaded gels of purified bovine 48 K protein and retinal S antigen, respectively. Lane 3: extract at low ionic strength (5 mM Hepes, 1 mM dithiothreitol, pH 7.0) from dark-kept bovine rod outer segments. The major bands are, in addition to 48 K protein, the cyclic GMP-phosphodiesterase (PDE) and the transducin subunits (To, Tfl, Ty). At a more physiological ionic strength, transducin and PDE are membrane associated, and are therefore not visible in lanes 4-7. Lanes 4-9: lightinduced binding of 48 K protein and S antigen to disk membranes, shown by comparing supernatant,~ obtained from darkkept (D) or illuminated (L) _mem_bra_nesuspensions° The protein disappears from the supernatants obtained after illumination. Lanes 4-7 are supernatants obtained at physiological ionic strength (120 mM KCI, 20 mM Hepes, 1 mM MgCI2, 1 mM dithiothreitol) from rod outer segments (6 mg/ml of rhodopsin) supplemented with 2 mM ATP. Purified S antigen was added to lanes 6 and 7. Lanes 8 and 9: previously phosphorylated, regenerated and washed disk membranes (P-disks)(cf. text), with purified retinal S antigen. This figure was prepared by C. Pfister (Laboratoire de Biophysique Mol~culaire et Cellulaire. C.E.N. Grenoble).

and the antibody levels in rats immunized with either protein are similar.

Interactions with photoexcited and phosphorylated rhodopsin The retinal S antigen/48 K protein is characterized by its ability to associate with illuminated rod outer segment disk membranes. This light-dependent binding is strongly stimulated by rhodopsin phosphorylation. It is inhibited by a transducin excess [20], and it suppresses the light activation of cyclic

!

Proteins controlling visual transduction 123

123

100

S-antigen

~so m

aSnf.EL ED

gel

immu----no detection

1 10 100 IO~X) antigen concentration (ng/rnl)

[]

Fig. 3. Immunological identity of 48 K protein and retinal S antigen, a. Recognition of the 48 K protein transferred from an SDS gel to a nitrocellulose sheet (Western blotting technique). The stained gel (left) corresponds to the immunodetection blot (right). Lane 1 : purified retinal S antigen. Lane 2: light extract (EL), at physiological ionic strength, from a rod outer segment suspension in the presence of 2 mM ATP. This extract is devoid of 48 K protein; see lane 5 in Fig. 2. Lane 3: dark extract tED), as in lane 2, but without illumination. This extract contains 48 K protein; see lane 4 in Fig. 2. Four different monoclonal antibodies to bovine S antigen were tested and gave identical results. No protein was recognized in E L . The major protein recognized in E D corresponds to the 48 K protein, with an intensity similar to that for S antigen, b. Radioimmunoassay of 48 K protein and S antigen. Different concentrations of either S antigen or 48 K protein competed with a constant amount of m25Ilabeled S antigen for a constant amount of rabbit antibody to bovine S antigen. The vertical scale shows the percentage B / B o of mzsIbound to the antibody. The two titration curves are practically identical. This figure was prepared by C. Pfister (Laboratoire de Biophysique Mol~culaire et Cellulaire, C.E.N. Grenoble).

GMP-phosphodiesterase (see the paper by P. Deterre et al., in this issue, on visual transduction mechanism). The use of reconstituted systems has allowed a better understanding of this property [211: (1) Disk membranes with highly phosphorylated and then regenerated rhodopsin can be prepared by incubating rod outer segment suspensions under continuous illumination in the presence of ATP. The resulting phosphorylated opsin (5-7 phosphates bound per opsin chain) is then regenerated to phosphorylated rhodopsin by incubation with an excess of 11-cis retinal in the dark [20]. These membranes are then thoroughly washed to eliminate all soluble and peripheral proteins, and reconstituted with pure transducin and phosphodiesterase. Here too, the 48 K protein is able to suppress the lightinduced activation of cyclic GMP-phosphodiesterase in a dose dependent manner, and proportionally

375

to the quantity of bleached rhodopsin. Moreover, the initial velocity of phosphodiesterase is strongly affected by the 48 K protein in this system, but not in a standard rod outer segment preparation: this is due to the fact that rhodopsin is unphosphorylated initially and becomes phosphorylated only slowly during the assay. Rhodopsin phosphorylation exerts by itself an inhibitory effect on phosphodiesterase activation, since phosphorylated disk membranes have a capacity to dactivate membrane-associated transducin and phosphodiesterase reduced to about 30% of control. Nevertheless, the presence of the 48 K protein is required for a complete inhibition. (2) Urea-treated disk membranes are completely free of rhodopsin kinase, and 48 K protein has no significant effect on phosphodiesterase activation in a system reconstituted from such disks, even in the presence of ATP: rhodopsin phosphorylation is therefore necessary for this purpose. (3) Phosphodiesterase can be directly activated by long-term activated transducin Ta~-GTP [yS], thus bypassing the normal activation pathway through photoactivated rhodopsin. Under these conditions, the 48 K protein has no influence on phosphodiesterase activity, even in the presence of ATP and/or photoactivated phosphorylated rhodopsin. This clearly shows that 48 K protein has no direct influence on activated phosphodiesterase, in disagreement with the proposal of Zuckerman et al. [22, 23]. For this reason, we prefer to avoid the name 'arrestin' coined by this author and referring to such a mechanism. (4) Finally, the inhibitory effect of 48 K protein can be observed with a fully synthetic system obtained with purified phosphorylated rhodopsin (6 phosphates/rhodopsin) recombined with phospholipids, transducin and phosphodiesterase. Therefore, it may be safely concluded that S antigen/48 K protein binds to phosphorylated photoactivated rhodopsin, thus competing with transducin and preventing phosphodiesterase activation. Rhodopsin phosphorylation and the binding of 48 K protein are both necessary in order to fully inhibit visual transduction. This corresponds to a homologous desensitization of rhodopsin, since rhodopsin kinase phosphorylates only photoactivated rhodopsin. Such a mechanism should be very efficient in vivo, since the amount of S antigen/48 K protein in retinal rods is similar to that of rhodopsin [5], and exceeds that of transducin. Each photoactivated rhodopsin molecule may then bind the former protein. Moreover, immunofluorescence studies show S antigen/48 K protein in the whole cytoplasm of dark-adapted rods, but only in the outer segments after illumination [5].

376

E. Vadot

Structural studies on S antigen~48 K p r o t e i n

References

The primary structure of bovine retinal S antigen has now been completely elucidated through sequencing of its eDNA [25, 26]. The complete protein consists of 403 amino acid residues, with a predicted molecular weight of 45 296. The hydrophobicity profile does not show any hydrophobic stretch characteristic of a signal peptide or of a transmembrane sequence. Of course this is not surprising for a soluble cytoplasmic protein. Some sequence similarities with the transducin-~ subunit have been described, but the two proteins are obviously only very distantly related. The sites of interaction with rhodopsin and with antibodies, and the uveitogenic sites have not yet been defined. However, information on these topics should become available in the near future. It is not known at this moment if the same protein exists both in rods and cones, or if the latter contain a similar but distinct molecule. This latter possibility seems more likely, since some antibodies specific for S antigen label only rods [27] and some others label both photoreceptor types [8]. Moreover, distinct transducin types are found in rods and in cones [28]. Further structural studies should resolve this issue.

1 Wacker W.B. & Lipton M.M. (1965) Nature 206, 253-254 2 Wacker W.B., Donoso L.A., Kalsow C.M., Yankeelov J.A. & Organisciak D.T. (1977) J. Immunol. 119, 1949-1958 3 Dorey C. & Faure J.P. (1977) Ann. Immunol. (Inst. Pasteur) 128 C, 229-232 4 Wilden U., Wiist E., Weyand I. & Kiihn H. (1986) FEBS Lett. 207, 292-295 5 Broekhuyse R.M., Tolhuizen E.F.J., Janssen A.P.M. & Winkens H.J. (1985) Curr. Eye Res. 4, 613-618 6 Mirshahi M., Faure J.P., Brisson P., Falcon J., Guerlotte J. & Collin J.P. (1984) Biol. Cell. 52, 195-198 7 Kalsow C.M. & Wacker W.B. (1978) Invest. Ophthalmol. Vis. Sci. 17, 774-783 8 Mirshahi M., Boucheix C., Collenot G., Thillaye B. & Faure J.P. (1985) Invest. OphthalmoL Vis. Sci. 26, 1016-1021 9 Faure J.P. (1980) Curr. Top. Eye Res. 2, 215-302 10 Faure J.P., de Kozak Y., Dorey C. & Tuyen V.V. (1977) Arch. OphthalmoL (Paris) 37, 47-60 11 Gery I., Mochizuki M. & Nussenblatt R.B. (1986) Prog. Retinal Res. 5, 75-109 12 Faure J.P. & de Kozak Y. (1981) in: Immunology of the Eye, Workshop H (Helmsen R.J., Suran A., Gery I. & Nussenblatt R.B., eds.), Information Retrieval Inc., Washington, pp. 33-48 13 Mochizuki M., Kuwabara T., McAllister C., Nussenblatt R.B, & Gery I. (1985) Invest. Ophtha!mo!o V'_tSoScio 26, ! - 9 14 Caspi R.R., Roberge F.C., McAllister C.G., El Sa'fed M., Kuwabara T., Gery T., Hanna E. & Nussenblatt R.B. (1986)J. lmmunol. 136, 928-933 15 de Kozak Y., Faure J.P., Ardy H., Usui M. & Thillaye B. (1978)Ann. lmmunol. (Inst. Pasteur) 129 C, 73-88 16 Vadot E. & R6my C. (1980) J. Fr. Ophtalmol. 3, 301-302 17 Kiihn H. (1978) Biochemistry 17, 4389-4395 18 Pfister C., Dorey C., Vadot E., Mirshahi M., Deterre P., Chabre M. & Faure J.P. (1984) C.R. Acad. Sci. Paris Ser. III 299, 261-265 19 Pfister C., Chabre M., Plouet J., Tuyen V.V., de Kozak Y., Faure J.P. & Kiihn H. (1985) Science 228, 891-893 20 Kiihn H., Hall S.W. & Wilden U. (1984)FEBS Lett. 176, 473-478 21 Wilden U., Hall S.W., Kiihn H. (1986) Proc. Natl. Acad. Sci. USA 83, 1174-1178 22 Zuckerman R., Buzdygon B. & Liebman P. (1985) Invest. Ophthalmol. Vis. Sci. 26, Suppl., 45 23 Zuckerman R. & Cheasty J.E. (1986) FEBS Lett. 207, 35-41 24 Chabre M. (1985) Annu. Rev. Biophys. Chem. 14, 331-360

Concluding remarks Homologous (agonist-promoted) desensitization of the ~-adrenergic receptor is associated with its phosphorylation by a specific kinase which specifically acts upon the agonist-occupied, or active, form of the receptor [29]. This/3-adrenergic receptor kinase can phosphorylate rhodopsin in a totally light-dependent fashion. Moreover, rhodopsin kinase is also able to specifically phosphorylate the agonist-occupied form of/3-adrenergic receptor [30]. Since both receptors display a strikingly similar structure (7 membrane-spanning helices; glycosylated N-terminal sequence; serine-rich Cterminal sequence) and interact with a coupling protein, it is tempting to envisage a similar mechanism for their homologous desensitization, involving phosphorylation of the active form, and subsequent binding of a specific protein, thus preventing coupling protein activation. The interaction of S antigen/48 K protein with photoactivated phosphorylated rhodopsin could thus be a model for hormonal receptor desensitization.

Proteins controlling visual transduction 25 Wistow G.J., Katial A., Craft C. & Shinohara T. (1986) FEBS Lett. 196, 23-28 26 Yamaki K., Takahashi Y., Sakuragi S. & Matsubara K. (1987)Biochem. Biophys. Res. Commun. 142, 904-910 27 McKechnie N.M., AI-Mahdawi S., Dutton G. & Forrester J.V. (1986) Exp. Eye Res. 42, 479-487

377

28 Lerea C.L., Somers D.E., Hurley J.B., Klock I.B. & Bunt-Milam A.H. Science 234, 77-80 29 Benovic J.L., Strasser R.H., Caron M.G. & Lefkowitz R.J. (1986) Proc. Natl. Acad. Sci. USA 83, 2797-2801 30 Benovic J.L., Mayor F., Somers R.L., Caron M.G. & Lefkowitz R.J. (1986) Nature 321,869-872