Ca2+ binding capacity of cytoplasmic proteins from rod photoreceptors is mainly due to arrestin.

Arrestin (also called S-antigen or 48-kDa protein) binds to photoexcited and phosphorylated rhodopsin and, thereby, blocks competitively the activation of transducin. Using Ca2+ titration in the presence of the indicator arsenazo III and 45Ca2+ autoradiography, we show that arrestin is a Ca2(+)-binding protein. The Ca2+ binding capacity of arresting-containing protein extracts from bovine rod outer segments is about twice as high as that of arrestin-depleted extracts. The difference in the Ca2+ binding of arrestin-containing and arrestin-depleted protein extracts was attributed to arrestin. Both, these difference-measurements of protein extracts and the measurements of purified arrestin yield dissociation constants for the Ca2+ binding of arrestin between 2 and 4 microM. The titration curves are consistent with a molar ratio of one Ca2+ binding site per arrestin. No Ca2+ binding in the micromolar range was found in extracts containing mainly transducin and cGMP-phosphodiesterase. Since arrestin is one of the most abundant proteins in rod photoreceptors occurring presumably up to millimolar concentrations in rod outer segments, we suggest that aside from its function to prevent the activation of transducin, arrestin acts probably as an intracellular Ca2+ buffer.

No Ca2+ binding in the micromolar range was found in extracts containing mainly transducin and cGMP-phosphodiesterase.
Since arrestin is one of the most abundant proteins in rod photoreceptors occuring presumably up to millimolar concentrations in rod outer segments, we suggest that aside from its function to prevent the activation of transducin, arrestin acts probably as an intracellular Ca2+ buffer.
After absorption of a photon by the visual pigment, the photoexcited rhodopsin activates transducin, a GTP-binding protein, which in turn activates a cGMP-phosphodiesterase (PDE)' (1,2). The light-induced activation of the PDE uia transducin leads to a decrease in the intracellular cGMP level that entails the closure of cGMP-gated cation channels in the plasma membrane (3,4). This closure shuts down the Na' and Ca'+ ion influx which persists during darkness (5). As a consequence, the intracellular free Ca2+ concentration decreases due to the light-independent Na+/Ca'+ exchange activity in the plasma membrane (4). It was recently proposed that intracellular free Ca')+ is involved in the sensitivity control of the photoreceptor (6)(7)(8) more, the presence of a significant Ca" buffering capacity in ROS has been infered from electrophysiological data (5). We report here that at least one-half of the observed Ca'+ binding capacity with micromolar affinity of cytoplasmic ROS proteins is due to arrestin, an abundant protein in rod cells, which is involved in the deactivation of the cGMP-PDE enzyme cascade (11,12 The reliability of this titration method was verified using the Ca'+-binding protein calmodulin which binds Ca2+ in the micromolar range ( Fig. 1, inset). The values for the stoichiometry, 1:4.07 + 0.27 (S.D., n. = 4), and for the dissociation constant, K = 2.42 + 0.19 pM (SD., n = 4), obtained with calmodulin concentrations between 1 and 2.5 FM, are in good agreement with previously reported values obtained by other methods (stoichiometry, 1:4 and K = 2.4 PM) (21). Hypotonic and Isotonic Dark-extracts-All extracts obtained from ROS, which were purified in the presence of 1 mM EDTA, were washed extensively by ultrafiltration which reduced their initial concentration of EDTA (about 10 PM) by a factor of >103. Extracts obtained from ROS prepared in the absence of EDTA were used without the washing and concentration procedure. In these latter extracts the concentration of Ca2+ binding sites was below 1 FM. To enlarge the measuring signals only concentrated protein extracts were used if not stated otherwise.
The Ca'+ binding capacity of unconcentrated hypotonic dark-extracts, 2.22 + 0.45 nmol of Ca*+/mg protein (S.D., n = 4), was identical with the Ca*+ binding capacity of concentrated hypotonic dark-extracts, 2.29 -C 0.40 nmol of Ca*+/mg protein (S.D., n = 7). This implies that no substantial loss of Ca2+ binding sites results from the washing and concentration procedure.
Protein extracts prepared under isotonic conditions (isotonic dark-extracts) contained only low amounts of transducin and PDE since most of these proteins were bound to the membranes. The Ca2+ binding capacity of isotonic dark-extracts was obtained to be 5.63 f 0.38 nmol of Ca*'/mg protein (S.D., n = 4). This apparent increase of the Ca2+ binding capacity of isotonic dark-extracts as compared to hypotonic dark-extracts can be fully attributed to the lower protein content of isotonic extracts which is due to the low content of transducin and PDE. Therefore, the amounts of transducin and PDE were subtracted from the total protein amount of the hypotonic extracts in order to compare the Ca*+ binding capacities of hypotonic and isotonic dark-extracts. Since transducin and PDE make up about 60% of the proteins in hypotonic extracts, the Ca*+ binding capacity of concentrated hypotonic dark-extracts was re-calculated to be 5.52 + 1.57 nmol of Ca'+/mg of protein (minus transducin and PDE) (S.D., n = 7), i.e. undistinguishable from the value obtained for isotonic dark-extracts.
This consideration suggests that both transducin and PDE have no Ca2+ binding sites in the micromolar range, at least not in their undissociated states. This presumption was tested with hypotonic re-extracts obtained from ROS membranes after an isotonic dark-extraction.
In these re-extracts which contained mainly transducin and PDE we could not detect a noticeable Ca2+ binding in the micromolar range (Fig. 2). This finding confirms that the presence of transducin and PDE in hypotonic dark-extracts does not increase the concentration of Ca*+ binding sites.
Arrestin-containing and Arrestin-depleted Extracts-In contrast to transducin and PDE we observed that the absence of arrestin reduces markedly the concentration of Ca*+ binding sites. Illumination of ROS in the presence of ATP and Mg2+ leads to the binding of arrestin to photoexcited and phosphorylated rhodopsin (11); therefore, isotonic protein extracts from ROS illuminated in the presence of ATP contain only trace amounts of arrestin (arrestin-depleted lightextracts; Fig. 3 cles) has a greater Ca2' binding capacity than the arrestin-depleted extract (R, solid circles). The difference in the concentration of arrestin between these two extracts is 1.5 gM; the difference in the Ca'+ binding of these extracts is 1.6 KM. The binding curves are plotted like those in Fig. 2. Inset, densitograms of the isotonic extracts, whose Ca'+ binding curves are shown. The ordinate of these densitograms (not shown) represents the relative absorbance at 580 nm of Coomassie Blue-stained SDS gels. The arrestin-depleted extract (B, 0.79 mg of protein) shows a markedly lower amount of arrestin (arrolus) than the arrestin-containing extract (A, 0.87 mg of protein). The lower Ca2+ binding capacity of arrestin-depleted extracts is explainable with the lower amount of arrestin in these extracts. MW, molecular weight. thus arrestin does not bind (arrestin-containing light-extracts; Fig. 3, inset A).
The arrestin contents of arrestin-containing light-extracts and isotonic dark-extracts are similar. The same Ca2+ binding capacity was observed for arrestin-containing light-extracts and isotonic dark-extracts.
However, the Ca*+ binding capacity of arrestin-depleted light-extracts is definitively lower, namely 2.81 ? 0.57 nmol of Ca'+/mg protein (SD., n = 3), although the amount of proteins in arrestin-depleted lightextracts is slightly reduced due to the lack of arrestin in these extracts (arrestin makes up about 10% of the protein mass in arrestin-containing light-extracts).
Since the Ca*+ binding capacity of arrestin-depleted light-extracts is only one-half as high as that of arrestin-containing light-extracts, this finding suggests that about one-half of the Ca'+ binding capacity of arrestin-containing light-extracts is due to arrestin. To enlarge the concentration of arrestin in ROS, we prepared ROS from illuminated eyes since arrestin is shifted in a light-dependent manner from the inner to the outer segment of rods (22). The isotonic extracts from these ROS contained about three times as much arrestin as isotonic extracts from ROS of dark-kept eyes. Arrestin-depleted extracts were prepared by removal of arrestin from arrestin-containing extracts via its light-dependent binding to photoexcited and phosphorylated isorhodopsin (see "Experimental Procedures"). The difference in the Ca2+ binding between these arrestincontaining and arrestin-depleted extracts (Fig. 3) was about three times higher than that of extracts from dark-kept eyes, i.e. similar to the increase in arrestin content in these preparations. Fig. 3 shows Ca'+ binding curves of arrestin-containing and arrestin-depleted extracts. The difference curve which is due to the Ca'+ binding of arrestin (Fig. 4, open  Ca*' bound to arrestin). The abscissa shows the free Ca"+ concentration. Each point is the average of data from six experiments (difference measurements) resp. from 12 experiments (purified arrestin). The data from crude and FPLC-purified arrestin have been combined (solid circles) since no differences were observed in their Ca" binding capacities. The error bars indicate the standard deviations.
yields a dissociation constant of 3.45 f 0.82 pM (S.D., n = 6) and a stoichiometry for Ca*+ binding to arrestin of 1:1.07 + 0.11 (S.D., n = 6). These results strongly suggest that arrestin contains one Ca" binding site with micromolar affinity. The Ca2+ binding to arrestin was further detected by autoradiography (Fig. 5) of ROS proteins revealing binding of 4hCa2+ to arrestin and presumably to the separated a-subunit of transducin and to rhodopsin. This presumption was supported by autoradiography of protein extracts containing primarily transducin and PDE and of hypotonically washed ROS membranes (data not shown). The more intensive labeling of arrestin in ROS obtained from illuminated eyes (Fig. 5A, lane 5) as compared to ROS from dark-kept eyes (lane 4) corresponds to the higher amount of arrestin in these preparations, suggesting that the Ca2+ binding is not due to a comigrating protein. This inference is corroborated by the Ca'+ binding of crude (lane 2) and FPLC-purified (lane 3) arrestin. Purified Arrestin-Ca" binding of arrestin was established for both crude (purity 80-95%) and FPLC-purified (purity >98%) arrestin. This finding was obtained for arrestin isolated from many retinae and from single retinae of individual animals. Fig. 4 (solid circles) shows that one Ca2+ is bound per arrestin, independent of whether crude (1:1.03 f 0.16, S.D., n = 10) or FPLC-purified arrestin was measured (1:1.03 resp. l:l.Ol, n = 2). The dissociation constant calculated from the Ca*+ binding curves is 3.21 & 0.81 pM (S.D., n = 10; crude arrestin) and 1.78 resp. 2.23 pM (n = 2; FPLC-purified arrestin).
Since arrestin can be fractionated in different isoelectric subspecies2 we investigated the Ca*+ binding capacity of the main subspecies of arrestin (p1 -5.8) from pooled retinae. We obtained one Ca2+ binding site per arrestin, implying that the various arrestin subspecies probably do not differ in their molar ratio of Ca2+ binding site per molecule. concentration range. Since arsenazo III binds Ca'+ in the micromolar range, it is more suited to measure Ca"+ affinities in this range than other indicators like Quin 2 (24) or tetramethylmurexide (25) which do not have micromolar dissociation constants. Beeler et al. (26) reported that binding of arsenazo III to muscle proteins may occur which affects the Ca" affinity of the indicator. In our experiments, binding of arsenazo III to one or more protein components is unlikely for the following reasons: (a) we found the correct Ca'+ affinity and binding stoichiometry for calmodulin; (b) no systematic deviations were observed between the titration curves of protein-free (control) and protein-containing solutions of PDE (up to 1.9 FM) and transducin (up to 13.4 FM); (c) binding of arsenazo III to proteins could not be detected with equilibrium dialysis of arsenazo III in the presence of the investigated proteins (ROS extracts, calmodulin).

DISCUSSION
Therefore, this titration technique measures correctly the Ca'?+ binding capacity of ROS extracts and purified arrestin.
Although Ca')+ plays an important role in photoreceptors (6-8), presumably through sensitivity control, only little is known about Ca'+-binding proteins in ROS. Calmodulin has been identified in very low amounts in bovine and frog ROS but its role in visual excitation processes is questionable (27,28). Recently, the guanylate cyclase activity of bovine ROS has been reported to be controlled by nanomolar Ca'+ concen-trations presumably by a Ca'+-dependent regulatory protein (10). We report here that the concentration of Ca" binding sites with micromolar affinity is in the range from 2 to 6 nmol of Ca'+/mg protein in protein extracts from ROS of darkkept eyes and that at least half of these Ca'+ binding sites can be attributed to arrestin. Transducin and PDE, the major proteins in hypotonic extracts from ROS displayed no micromolar Ca"' affinity. This does not exclude the possibility, however, that these proteins may adopt conformational states which allow Ca" binding as it is indicated for the separated (k-subunit of transducin by autoradiography. Remarkably, analysis of the amino acid sequence of arrestin (29) displays no typical CaL+ binding domain of the EF-hand t.ype which comprises normally 12 amino acids (30). This does not exclude Ca'+ binding sites of other types, though. In fact, comparison of the amino acid sequences of Ca'+ binding sites of calmodulin (Asn"''-Glut "'), parvalbumin (Asp"'-Glu""), and troponin C (Asp"-G~u'~, Asp"'-Glu") with the arrestin amino acid sequence yields a conspicuous conservation or conserved substitution of four from five negatively charged amino acids in the amino acid stretch 362-373 of arrestin (Asp-X-Asp-X-X-X-Glu-X-X-X-Asp-Glu).
Although the glytine residue which is conserved in many Ca'+ binding sites does not occur in this sequence we suggest tentatively that this amino acid stretch is involved in Ca'+ binding since negatively charged amino acids are essential to form Ca"' binding sites (30).
One may suspect that Cal+ binding to arrestin influences its binding to photoexcited and phosphorylated rhodopsin; however, we did not observe an influence of micromolar Ca2+ concentrations on the binding of arrestin to photoexcited and highly phosphorylated rhodopsin:' Several lines of evidence indicate that Ca'+ buffering of arrestin might be important in photoreceptors: (a) arrestin is one of the major proteins in ROS, and its molar ratio to rhodopsin in whole photoreceptor cells was reported to be close to 1:l (22); (6) the molar ratio of arrestin to rhodopsin in ROS is about 1:lO (31); (c) arrestin binds one Ca"+ with an affinity of about 3 FM, and at least one-half of the Ca" binding capacity with micromolar affinity of isotonic darkextracts is due to arrestin (this work). Given the above molar ratio of arrestin to rhodopsin in ROS one estimates that the arrestin concentration may well be in the millimolar range, in particular since the cytoplasmic volume of ROS is largely restricted by the disk stack. This means that CaL+ buffering of arrestin may become substantial. High local concentrations of arrestin may thus readily explain the electrophysiological observation (5) that about 95% of the Ca'+ entering a rod cell is bound and is not accessible to injected aequorin. Of course, other Ca'+ binding sites, as those reported here, may add to the Ca'+ buffering capacity as well. Intracellular Mg" may also influence the effective Ca'+ affinity of arrestin; however, autoradiography indicates that arrestin binds Ca" even in the presence of a 7-lo-fold excess of Mg'+ over Ca'+. The Ca"+ buffering capacity of ROS is further increased upon illumination since arrestin is shifted from the inner to the outer segment of rods after light-activation of the photoreceptor (e.g. Ref. 22). Aside from its function to prevent transducin activation by competitive binding to light-activated rhodopsin we suggest that a second physiological function of arrestin is probably to act as an internal Ca"+ buffer in photoreceptors.