Insulin-like Growth Factor I Receptors in Retinal Rod Outer Segments*

We have previously reported that the GDP-bound a- subunit of the GTP-binding protein transducin, present in outer segments of retinal rod cells (ROS), serves as a high affinity in vitro substrate (K, = 1 MM) for the insulin receptor kinase. The present study demon- strates that transducin also serves as in vitro substrate for an endogenous IGF-I receptor kinase isolated from ROS membranes. The presence of insulin-like growth factor I (IGF-I) receptors in ROS is evident from the high affinity and specific binding of 12'I-IGF-I to ROS membranes (Kd = 3 nM) which contain 110 fmol of IGF-I binding sites/mg of membrane protein. Further-more, cross-linking of 12'I-IGF-I labels the 135-kDa a- subunit of this receptor. '2SI-Insulin binding capacity to ROS membranes is less than 5% that of IGF-I. The IGF-I-stimulated tyrosine kinase activity in solubilized and partially purified receptors from ROS autophos- phorylates its own 95-kDa &subunits as well as other substrates like transducin. Insulin, which is 200-fold less potent than IGF-I in competing for '261-IGF-I binding, is only 5-fold less potent than IGF-I in stimulating the receptor kinase activity. This suggests that insulin is much more potent than IGF-I in coupling ligand binding with kinase activation. The

Transducin, the GTP-binding protein present in retinal rod outer segments (ROS),' plays a key role in phototransduction by coupling the bleached rhodopsin to the cGMP-phosphodiesterase (1)(2)(3). We have recently shown that the GDPbound a-subunit of transducin (TD-a-GDP) serves as a high affinity in vitro substrate for both the insulin receptor kinase and protein kinase C (Ca'+/phospholipid-dependent enzyme) ( 4 ) . In order to assess whether this phosphorylation is of *This work was supported in part by the Israeli Academy of Sciences and Humanities (R. S. E.), the Israel Cancer Research Fund, and the Israel Cancer Association (Y. Z. and R. S. E.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The abbreviations used are: ROS, retinal rod outer segments; IGF, insulin-like growth factor; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; SDS, sodium dodecyl sulfate; TD, transducin.
physiological significance, we attempted to identify insulin receptors in ROS. Here we demonstrate that ROS membranes contain a much higher concentration of insulin-like growth factor I (IGF-I) receptors than of insulin receptors. Receptors for insulin and for IGF-I share a great deal of both structural and functional homology (5-7). Both act as tyrosine kinases capable of phosphorylating themselves as well as exogenously added substrates (8)(9)(10)(11)(12)(13)(14)(15)(16)(17). Furthermore, although not identical, the substrate specificity of these kinases toward exogenous substrates is very similar (16)(17)(18). Both insulin and IGF-Idependent tyrosine kinases were suggested to mediate at least part of the metabolic and growth-promoting activities of these hormones (8)(9)(10)(11)(12)(13)(14)(15)(16). IGF-I itself is known to be present in the vitreous. Moreover, elevated concentrations of this hormone were found in the vitreous of most diabetic subjects with severe proliferative retinopathy (19). The presence of IGF-I receptors in ROS raises the possibility that through phosphorylation of endogenous substrates like transducin, IGF-I and insulin could mediate at least part of the normal or pathological metabolic activities manifested by retinal rod cells.

Methods
Purification of Membranes from Rat Liuer and Outer Segments of Rod Cells-Transducin and its subunits were purified as in Ref. 20. ROS membranes were purified from freshly dissected bovine retinas by a modification (21) of the method of Papermaster and Dreyer (22). Purity of ROS membranes was assessed by electron microscopy. They were free from any contaminating neuronal retinal membranes as assessed by the lack of GO, a GTP-binding protein which is particularly abundant in brain (23). Rat liver plasma membranes were purified by the procedure of Neville as previously described (24). Membranes were stored at -70 "C prior to use.
Partial Purification of IGF-I and Insulin Receptor Kinase-IGF-I receptor kinase and insulin receptor kinase were partially purified from ROS membranes and rat liver membranes (respectively) essentially as described (24). The membranes were solubilized in 1% Triton X-100, and the receptors were further purified by affinity chromatography over a column of wheat germ agglutinin coupled to agarose. The IGF-I and insulin receptors were eluted from the column in buffer containing 50 mM Hepes, 150 mM NaCI, 0.1% Triton X-100, 0.5 M N-acetyl-D-glucosamine, and 10% (v/v) glycerol (pH 7.6) and were kept at -70 "C prior to use.
Affinity Cross-linking of '251-Insulin and lZ5I-IGF-I to ROS and Rat Liuer Plasma Membrunes-ROS or liver membranes (1.5 and 1.7 mg, respectively) were suspended in 50-p1 aliquots of buffer A containing Na2HP04 (pH 7.5). Fifty-pl aliquots of insulin, IGF-I (each at 1 X M), or buffer A alone were added, together with 50-pl tracer amounts of either '251-insulin or lZ5I-IGF-I (each at 1 pCi/ml). Incubation was carried out for 4 h at 15 "C with occasional shaking. The membranes were spun down ( 5 min, 12,000 X g) and the pellets resuspended in 0.5 ml of buffer A. Ten-p1 aliquots of 0.1 M dimethyl suberimidate were added, and incubation continued for 15 min at 4 "C. The samples were spun for 5 min at 12,000 X g. The pellets 118 mM NaCI, 1.2 mM MgSO4, 1.2 mM KHzPO4, 5 mM KCI, 10 mM Segments were resuspended in 1 ml of buffer containing 10 mM Tris-HCI and 1 mM EDTA, pH 7.4, spun again, and the pellets resuspended in 100 pl of Laemmli's sample buffer. After boiling for 5 min, samples were separated by SDS-10% polyacrylamide gel electrophoresis, dried, and subjected to autoradiography.
"'I-Insulin and 12sI-IGF-Z Binding to ROS Membranes-ROS membranes (1.5 mg) were suspended in 50 pl of buffer A. Fifty-pl aliquots of 12sII-insulin or %IGF-I (-30,000 cpm/tube) were added together with 50-pl aliquots of the appropriate unlabeled ligand. Following 4 h of incubation at 15 "C, the membranes were spun down (5 min, 12,000 X g), and the pellets were counted in a gamma spectrometer.

Materials
[y-"PJATP (3000 Ci/mmol; 1 Ci = 37 gigabecquerels), [12sI]monoiodo-labeled (at tyrosine 14 of the A-chain) human insulin (2000 Ci/ mmol), and lZsII-IGF-I (2000 Ci/mmol) were from Amersham, Buckinghamshire, Great Britain. Sera from a patient with autoantibodies to the insulin receptor (B-10) were a generous gift of Dr. P. Gorden, Diabetes Branch, National Institutes of Health, Bethesda, MD. ATP, CTP, the synthetic tyrosine-containing polymer (GluRO-Tyr2"),, pork insulin, insulin A-chain, insulin B-chain, and N-acetyl-D-glucosamine were from Sigma. Tyrss-labeled insulin-like growth factor-I was from Amgen Biologicals, Thousand Oaks, CA. Protein A coupled to Sepharose was from Pharmacia P-L Biochemicals. + during the binding period was less than 2% and could not account for the reduced binding of '2'I-insulin.

I2'I-IGF-I and '2'I-Insulin Binding to ROS
Affinity Cross-linking of '"I-IGF-I and '251-Insulin to ROS and Liver Membranes-'2'I-IGF-I bound exclusively to the cy (135-kDa)-subunit of the IGF-I receptor present in ROS membranes. This is evident from the specific labeling of a 135-kDa protein following the incubation of ROS membranes with '2'I-IGF-I and subsequent cross-linking of the ligand with dimethyl suberimidate (Fig. 2, lune B ) . A protein of that size was previously shown to be the cy-subunit of the IGF-I receptors in non-neuronal tissues (6,25). Binding of I2'I-IGF-I to the 135-kDa protein was specific and could be completely abolished when binding and cross-linking were conducted in the presence of M IGF-I (Fig. 2, lune A). In contrast, no detectable amount of '2sI-insulin could be cross-linked to ROS membranes under these experimental conditions (Fig. 2, lanes C and D). When a similar experiment was carried out using rat liver plasma membranes, a 135-kDa protein was specifically labeled when cross-linked to '2sI-insulin but not to IGF-I (Fig. 2, lanes G and H ) . This band most likely represents the a-subunit of the insulin receptor present in rat liver membranes. These findings confirm previous results which demonstrated the presence of receptors for insulin but not for IGF-I in rat liver membranes (26). '"I-IGF-I Binding to Solubilized and Lectin-purified ROS-Purified ROS membranes were solubilized with Triton X-100 (1%) and further subjected to affinity chromatography on columns of wheat germ agglutinin coupled to agarose. Elution of the columns with 0.5 M N-acetyl-D-glucosamine yielded a preparation that was enriched in IGF-I receptors as evidenced by its capability to specifically bind '2sI-IGF-I (Fig. 3). Similar to the results obtained with intact ROS membranes, both IGF-I and insulin inhibited '2sI-IGF-I binding in a dosedependent manner. Half-maximal inhibition occurred at lo-' and 2 X M IGF-I and insulin, respectively. 12sI-Insulin was significantly less potent than '2sI-IGF-I in binding to these receptor preparations. Quantitation of 'ZSI-insulin binding was difficult to assess since its binding was only slightly inhibited by excess of either insulin or IGF-I.
IGF-I-stimulated Tyrosine Kinase Activity Associated with the IGF-I Receptors in ROS-Since both insulin and IGF-I receptors are known tyrosine kinases, attempts were made to characterize the tyrosine kinase activity associated with the IGF-I receptor present in ROS membranes. Lectin-purified IGF-I receptors from ROS were incubated with IGF-I or insulin for 30 min at 22 "C to allow ligand binding. [-y-"'P] ATP and MnZ+ were subsequently added, and autophosphorylation was carried out for 15 min at 22 "C. Autoradiography of the phosphorylated proteins resolved on SDS gels revealed that both IGF-I and insulin stimulated the phosphorylation of a 95-kDa protein (Fig. 4). This phosphorylated protein was of the same molecular weight as are the P-subunits of the receptors for both IGF-I and insulin (8,13). The lectin- purified IGF-I receptor could also phosphorylate exogenously added substrates such as the tyrosine-containing polymer (GIuRo-Tyr20),. IGF-I markedly stimulated this reaction in a dose-dependent manner (Fig. 5), half-maximal stimulation occurring at a concentration of 5 X 10"' M. This is comparable to the potency of IGF-I to inhibit binding '2sI-IGF-I. Interestingly, insulin was only 5-fold less potent than IGF-I in stimulating the receptor kinase, which is in sharp contrast to insulin being only 1% as effective as IGF-I in inhibiting 12sI-IGF-I binding. Furthermore, the effects of insulin and of IGF-I on the tyrosine kinase activity were not additive. Incubation of receptor preparations with saturating concentrations of both IGF-I and insulin did not result in an increased kinase activity beyond that obtained with each of the ligands alone (Fig. 5). The tyrosine kinase was specifically stimulated only by IGF-I or insulin; insulin A-or B-chains (2 X M) or other ligands such as ACTH (1 X M) or epidermal growth factor (1 X M) failed to stimulate the tyrosine kinase activity.
Failure of Antibodies Directed against the Insulin Receptors to Immunoprecipitate the IGF-I Receptors-Antibodies directed against the insulin receptor (B-10) which do not recognize type I IGF receptors (15) were used to selectively deplete insulin receptors from lectin-purified preparations of ROS membranes. Following incubation with the B-10 antiserum and removal of immune complexes with protein A, there was no reduction in either the insulin or the IGF-Istimulated kinase activities remaining in the supernatant (Fig. 6). In contrast, when a similar experiment was carried out with solubilized and lectin-purified receptor preparations from rat liver, insulin-stimulated receptor kinase activity was depleted by more than 50% even though it was present in a 20-fold higher concentration than the receptor preparations from ROS. These results are in agreement with the notion that rat liver membranes contain insulin but lack IGF-I receptors (see also Fig. 1).
IGF-I and Insulin-stimulated Phosphorylation of TD-a-GDP by IGF-I Receptor Kinase from ROS-We have previously shown that the GDP-bound a-subunit of transducin undergoes phosphorylation on tyrosine residues catalyzed by rat liver insulin receptor kinase (4). In order to determine whether TD-a-GDP also serves as a substrate for the endogenous IGF-I receptor kinase present in ROS, purified prepa- Immunodepletion of insulin but not of IGF-I-dependent kinase activity by antibodies against the insulin receptor. Two-hundred p1 of solubilized and partially purified receptor preparations from rat liver and ROS membranes were incubated for 16 h a t 4 "C either with normal serum or with polyclonal antibody specific for the insulin receptor (B-10) (each a t a final dilution of 1:lOO). Three mg of protein A coupled to Sepharose were then added, and incubation was continued for an additional 60 min a t 4 "C. Samples were then spun down (12,000 X g, 5 min, 4 "C), and the supernatants were incubated for 30 min a t 22 "C with buffer (empty bars), 10" M insulin (filled bars), or IGF-I (hatched bars). Phosphorylation of (GIuRO-TyrZ"), was carried out as described under "Experimental Procedures." rations of TD-a-GDP were incubated with IGF-I receptors in the presence or absence of insulin or IGF-I. As seen in Fig. 7, both IGF-I and insulin markedly enhanced phosphorylation of TD-a-GDP by the IGF-I receptor kinase. These results are compatible with our previous studies where we could demonstrate phosphorylation of TD-a-GDP by the insulin receptor kinase derived from rat liver plasma membranes.
Substrate Specificity of the IGF-I Receptor Kinase-The capability of the IGF-I receptor kinase to phosphorylate different exogenous substrates was studied. Out of three tyrosine-containing polymers studied, (GluR"-Tyr2"), was the most effective one (highest V,,,,J. (Gl~~"-Ala~"-Tyr'~),, and (Alafin-G1u2"-Lyssn-Tyr'"), were about 20% as effective. Casein and histone H2b were less than 10% effective as substrates (Table   I). These results are in agreement with the previously described substrate specificity of the IGF-I receptor kinase (16).

DISCUSSION
This study documents the presence of receptors for IGF-I in retinal rod outer segments. About 110 fmol of IGF-I binding sites are present in 1 mg of purified ROS membranes. In general, the IGF-I receptors in ROS appear to be both structurally and functionally indistinguishable from IGF-I receptors present in non-neuronal tissues (25,27). They are glycoproteins as judged by their binding to columns of wheat germ agglutinin coupled to agarose. They contain a t least two subunits: a 135-kDa a-subunit which specifically binds IGF-I and a 95-kDa @-subunit which possess an IGF-I-stimulated tyrosine kinase activity. This protein tyrosine kinase is capable of phosphorylating itself as well as exogenously added substrates with a similar specificity to that reported for IGF-I receptors from non-neuronal tissues (16). Both binding and stimulation of the receptor kinase occur a t physiological concentrations of IGF-I with half-maximal effects on kinase activity a t 0.5 nM. Such low concentrations are comparable to those known to be present in the vitreous of normal subjects (2.7 f 0.96 ng/ml) (19). The size of the a-subunit (determined by affinity cross-linking of '*'I-IGF-I) distinguishes these IGF-I receptors from insulin and IGF-I receptors present in whole brain. The latter are smaller in size with an a-subunit of 115 kDa (28-31). It has been suggested that the lower molecular weight of brain a-subunit may be due to differences in carbohydrate residues when comparing the brain a-subunit to non-neuronal tissues (29-31). Whether this means that Incorporation of 32P into endogenous proteins was subtracted in each case. This was less than 2% of that incorporated into the exogenously added substrates. 'The ratio between activities determined in the presence and in the absence of stimulant. e Activities assaved in the uresence of either IGF-I or insulin using (Gluso-TyrZo), as substrate. ND, not determined.
IGF-I receptors from ROS have a different function from that played by brain IGF-I receptors still remains to be determined. Although IGF-I receptors share a great deal of structural and functional homology with insulin receptors, we conclude that ROS membranes contain few (if any) insulin receptors. This conclusion is based on the following observations. (a) binding of tracer amounts of lZ5I-IGF-I and of '251-insulin either to intact membranes or to solubilized and partially purified receptor preparations reveals the presence of exceedingly large numbers of 'T-IGF-1 binding sites compared to those for '251-insulin. ( 6 ) IGF-I is at least 100-fold more potent than insulin in displacing tracer lZ5I-IGF-I bound. (c) lZ5I-IGF-I, but not insulin, can be specifically cross-linked to the 135-kDa a-subunit of the receptor. Insulin could stimulate the receptor kinase activity, yet it was &fold less potent than IGF-I. Furthermore, at saturating concentrations M) the effects of insulin and of IGF-I were not additive. These observations suggest that both ligands stimulate a single pool of receptor kinases and that insulin acts through binding to the IGF-I receptor. This conclusion is further supported by the fact that antibodies that react specifically with insulin receptors, but not with IGF-I receptors, do not precipitate the insulin or the IGF-I-stimulated tyrosine kinase activities present in ROS. In contrast, these antibodies precipitate insulin receptor kinase activity present in rat liver.
Several possibilities could account for the greater potency of insulin (relative to IGF-I) for stimulating phosphorylation of type I IGF-receptor and exogenous substrates than for binding to the IGF-I receptor. For example, insulin might act t.hrough a subset of IGF-I receptors that has a higher affinity for insulin. Alternatively, insulin could be much more potent in coupling ligand binding to receptor kinase activation. These and other possibilities are still speculative and not supported by experimental results. The only clue we have is that this is probably a general phenomenon, since similar results were reported for IGF-I receptors isolated from BRL cells (16).
The physiological significance of the presence of IGF-I (19) and its receptors in ROS is as yet unclear. It should be noted that in addition to their known role in phototransduction, retinal rod cells exert an active metabolism including both anaerobic and aerobic glycolysis as well as active lipid and protein synthesis (32). Based on our results we propose that binding of IGF-I or insulin to the IGF-I receptor in ROS generates intracellular signals that regulate part of the rod cellular metabolism. The latter could be carried out either in the outer segments themselves or in the inner segment as a result of signals transmitted from the outer segments following binding of these hormones.
Phosphorylation of substrate proteins by the IGF-I receptor kinase is likely to be the next step in the chain of events that follow IGF-I binding. Since we have demonstrated that transducin is a high affinity in vitro substrate for the endogenous IGF-I receptor kinase in ROS, we would like to speculate that signal transduction initiated through IGF-I binding could be mediated through phosphorylation of transducin or of other transducin-like GTP-binding proteins present in these membranes.
Finally, it should be noted that IGF-I concentration in the vitreous of diabetic subjects with severe retinopathy is increased 2-fold compared to normal controls (19). These elevated IGF-I concentrations (6.3 2 0.93 ng/ml) are sufficient to account for almost maximal in vitro activation of the IGF-I receptor kinase. It still remains to be determined whether persistent elevation of IGF-I concentrations impairs the activity of the IGF-I receptor kinase in vivo and whether such impairment contributes to the pathogenesis of diabetic retinopathy, a disease which is often associated with impaired or even complete loss of vision.