Affinity Binding of Intact Fat Cells and Their Ghosts to Immobilized Insulin

Transcription

1 Proc. Nat. Acad. Sci. USA Vol. 70, No. 3, pp , March 1973 Affinity Binding of Intact Fat Cells and Their Ghosts to Immobilized Insulin (insulin receptor/insulin-sepharose/affinity chromatography/buoyant density/plasma membranes) DENIS D. SODERMAN, JOHN GERMERSHAUSEN, AND HOWARD M. KATZEN Department of Biochemistry, Merck Institute for Therapeutic Research, Rahway, New Jersey Communicated by Max Tishler, January 4, 1973 ABSTRACT Immobilized insulin, prepared by coupling insulin directly to agarose or through hydrocarbon "connecting arms," was demonstrated to be capable of firmly binding intact adipocytes and their ghosts. Various lines of evidence indicate that the insulin receptor on the plasma membrane, in addition to the insulin coupled to the agarose, was responsible for the observed binding. This evidence includes: (a) the finding that increasing the "arm" length increased the binding capacities of insulin- Sepharose affinity chromatographic columns, (b) specific inhibition and reversal by insulin and antiserum to insulin of the binding, as compared to lesser effects by other peptide hormones, (c) the indication that only the plasmamembrane sacs, not the other cellular contaminants in the crude ghosts, are capable of binding, and (d) the impairmerit and restoration of trypsin-sensitive membrane binding sites that are also required for insulin biosensitivity. These findings support the idea that the insulin receptor is the trypsin-sensitive site. By use of the differential btloyant densities of the various cell-bead complexes that resulted from the interaction of adipocytes with insulin- Sepharose, a new procedure was developed to demonstrate and study the binding. These complexes could also be demonstrated by interference contrast microscopy. Binding readily occurred under conditions favorable for insulin stimulation of the cells. By coupling tracer amounts of [1251]insulin to Sepharose oj insulin-sepharose, the effects of anti-insulin antisera, free insulin, and other peptide hormones and supplemental factors on the buoyant-density distribution of the complexes could be measured, as well as the effects of other ligands coupled to Sepharose. Despite the rapidly accumulating literature on the various applications of proteins coupled ("immobilized") to insoluble matrices, most of the published research on the insulin (I) receptor used free radioiodinated insulin, rather than immobilized native insulin, as the interacting hormonal agent. Except for our earlier preliminary report (1), there has been no published evidence demonstrating the ability of immobilized insulin to bind insulin receptor-containing intact cells or membranes. Other recent studies iih our laboratory (1, 2) and by Cuatrecasas (3) have indicated the potential of immobilized insulin for the affinity chromatographic isolation and purification of solubilized membrane fractions capable of binding [1251]insulin. Abbreviations: ILysS or IPheS, insulin (I) coupled at ph 9 and 5, respectively, to cyanogen bromide-activated Sepharose (S) as described by Cuatrecasas and Anfinsen (13). According to Cuatrecasas (4), coupling at these ph values results in I coupled through its lysine (Lys) (ph 9) or N-terminal phenylalanine (Phe) (ph 3). ILys5S or IPhe5S, ILys, or IPhe coupled at these ph values to the w-bromoacetamidoalkylamino-s derivatives (15), thereby resulting in insulin coupled to S via diamine-connecting arms-s(1.) of from 2 to 7 carbons in length; ILys5S and IPhe5S represent the 5-carbon (pentyl) derivatives. Insulin-S, general term not specifying any particular linkage. To whom to address correspondence. 792 It had been indicated previously that insulin coupled to Sepharose (S) retains the biological properties characteristic of the native hormone (4-6). Because of the relatively large size of the Sepharose beads, it was concluded that the hormone acts on the surface membrane of the cell (4). However, since Oka and Topper (6) could observe no actual binding of insulin-phe-sepharose beads to mammary epithelial cells that were sensitive to the immobilized-but not the free-hormone, they questioned the involvement of binding, from a mechanistic point of view, in the biological response to the hormone. Specific binding of [ln2i]insulin(monoiodinated) to liverand fat- cell membranes and intact fat cells has been demonstrated (7-9) and shown to correlate well with the differing bioactivities of various insulins and insulin derivatives (7). However, the question of the effect of radioiodination on the biological activity of the hormone appears to remain a controversial subject (10, 11). The present study was undertaken to determine if immobilized insulin can be utilized to study the interaction of the hormone with its receptor on intact fat cells and their plasma membranes. MATERIALS AND METHODS Male CDF albino rats (Charles River) weighing between g were used, and allowed free access to Purina laboratory chow. Sources of reagents were: Sepharose 4B (agarose), Pharmacia Fine Chemicals; beef insulin (recrystallized, 25.9 U/mg) and i4c-labeled algal protein hydrolysate (100,LCi /ml), Schwarz/Manin Labs; tryl)sin (twice-crystallized), Worthington Biochemicals; guinea pig antiserum to bovine insulin, Miles Laboratories; glucagon (crystalline, anhydrous), Mann Laboratories; and bovine-serum albumin (Fract. V), Sylvana Co. Ovine growth hormone (5-5 powder, 50 mg: 50 USP units), ovine prolactin (crystalline, P-S-9), leutinizing hormone (137, lyophilized), and ovine follicle-stimulating hormone (S-3) were gifts of the NIH Endocrinology Study Section, Bethesda, Md. Beef (standard activity) and human growth hormones (lyophilized) were gifts of I)rs. E. Ham and U. Lewis of this Institute. [l25iiinsulin(monoiodiinated) was prepared according to Freychet et al. (11). Other materials were purchased from commercial sources. Isolated fat cells and their ghosts were prepared by the procedures of Rodbell (12, 13) and derived from the distal half of the epididymal adipose tissue. The cells were highly sensitive to insulin. 20-Fold stimulations of the oxidation of [1- i4c]glucose to i4co2 by 35,AU of insulin added per ml were routinely obtained, and with a sensitivity to 2,U/ml, when measured by the procedure of Rodbell (12) as modified slighty by Glieman (14). "Lysing" and "wash" solutions for preparation of ghosts contained 1 mmi KHC03-pH 7.2, 2.5 mai MgCl2, and 0.1 mm\ CaC12.

2 The various insulin-sepharoses were prepared according to the general procedures for coupling protein amino groups described by Cuatrecasas and Anfinsen (15), and extensively washed immediately before use, for periods of up to 7 days with (each of) NaHCO3 buffer (ph 7.2), 6 M guanidine HCI, 0.01 N NaOH, 0.1 N HCl, and the suspending buffer at ph 7.4, until protein and ['25I]insulin (coupled in tracer amounts to the insulin-s) were no longer detectable in the washings. The insulin linked to Sepharose was determined by amino-acid analysis and specific-activity measurements of the [1251linsulin. All insulin-s preparations used in this study exhibited biological activities qualitatively characteristic of the native hormone and attributable, on the basis of [1251 linsulin measurements, to the immobilized hormone (2). Membrane-bound hexokinase was determined by the nitroblue-tetrazollum: glucose-6-phosphate dehydrogenasecoupled staining procedure (16). The intensities of the resultant membrane-associated blue color were estimated visually and found to require glucose, ATP, and Mg++. According to the studies of Rodbell (13), 15% of the total cellular hexokinase is distributed in fractionated ghost preparations in proportions similar to those of plasma membrane adenylate cyclase. RESULTS In Fig. 1 is depicted an example of the ability of affinity columns of insulin coupled directly to cyanogen bromideactivated Sepharose (17) to bind firmly significant portions of the insoluble fat-cell ghost preparations. While all of the turbidity attributable to the ghosts applied to Sepharose columns passed through unretarded immediately after the void volumes, considerable amounts could not be washed through in the fractions collected from the insulin-s columns, B 00 0 Proc. Nat. A cad. Sci. USA 70 (1973) FIG. 1. I FRACTION NUMBER Insulin-S affinity chromatographic "wash" profiles of ghost preparations. ILysS (A- - -A). ILysS (AX--- A), and IPheS (@-.) contained insulin coupled at 1.52, 2.64, and 2.15 mg/ml of packed Sepharose, respectively; Sepharose 4B (O--O) was the control column. A 0.5-ml aliquot of ghosts suspended in Krebs-Ringer bicarbonate buffer (ph 7.4) with a total turbidity at 280 nm of 6.30 per aliquot was applied to each siliconized glass column (6 X 105 mm) containing 1 ml of packed bed. Buffer was then continuously applied to the columns and 0.5-ml fractions were collected. Mixing the contents of each column (by repeatedly inverting) after collecting fraction 10 led to no further turbidity in subsequent fractions. I Insulin-Sepharose Binding Fat Cells and Membranes 793 TABLE 1. Binding of ghosts to columns of insulin-s derivatives with connecting arms of different lengths Washed Pooled washings beads Column (hexokinase) (hexokinase) (turbidity) Sepharoset ILysS ILys2S ILys3S ILys4S ILys5S llys7s Sepharoset IPheS IPhe2S IPhe3S IPhe5S Insulin-S was prepared with insulin added at a concentration of 8 mg/ml of packed Sepharose. t An equivalent of turbidity units was applied to each affinity column in the lysyl series, with this column as control Turbidity units were applied to each affinity column in the phenylalanyl series. Hexokinase activities in the washings were not determined. even after extensive washings and mixing of the beads. The maximum capacity of ILysS to bind membranes was reached when the Sepharose contained less than 1.7 mg of insulin coupled per ml of packed beads. Under these conditions, about a third of the membrane preparation was retained. At lower concentrations of immobilized hormone less binding occurred. IPheS exhibited a greater capacity than ILysS to bind ghosts, although the IPheS wash fraction profile was not as sharp as that from ILysS. While determinations of the membrane-marker hexokinase (see Methods) are only semiquantitative, results from such estimations of both the beads and washings (not shown) were completely consistent with the results in Fig. 1. In an examination of the specificity and reversibility of the binding seen in Fig. 1, it was found that free insulin at 6 mg/ ml was capable of slowly, but progressively, eluting about 80% of the ghosts, whereas albumin at this concentration was ineffective. However, some membrane could also be eluted with 4% bovine-serum albumin. Inasmuch as some of the insulin used for elution may be expected to become "trapped" on (i.e., bound to) free receptor sites on the bound membranes, the concentration of insulin required to elute the ghosts may not be unreasonably high. In addition, the local concentration of immobilized insulin near the membrane binding sites may be relatively high. Guanidine HCl at 6 M was very effective in eluting the ghosts, whereas.1 M NaCl, 0.05 N acetic acid, and 0.05 N NaOH had little effect. Significantly, the insulin- S preparations treated with antiserum to bovine insulin exhibited no binding capacity. Earlier studies on protein purification by affinity chromatography suggested the importance of interposing a hydrocarbon chain ("arm") between the Jigand and the insoluble polysaccharide backbone (18). Table 1 shows that insulin-s preparations with progressively increasing arm lengths exhibited increasing binding capacities, although the insertion of only a 2-carbon length arm resulted in decreased binding

3 794 Biochemistry: Soderman et al. Proc. Nat. Acad. Sci. USA 70 (1973) TABLE 2. Binding to affinity columns determined tkith '4C-labeled ghosts Pooled Washed washings beads % Column Preparationt (cpm) (cpm) Bound Sepharose [14C] ghosts ILys5S [4C] ghosts ILys5S ['4C]aminoacids 8 X <1 IPhe5S I14C] ghosts Insulin-S contained 1.5 mg of insulin per ml of packed Sepharose. t Ghosts were prepared after incubation of fat cells for 90 min at 370 in Krebs-Ringer phosphate buffer (ph 7.4) containing 4% bovine-serum albumin, 3 mm glucose, MACi of "4C-labeled algal protein hydrolysate per ml, and 10 MAM carrier [12C] amino-acid mixture. Resultant cells and ghosts were washed with buffer until cell washings were essentially free of 14C, and ghost washings had low and constant amounts of 14C. Incubation of a separate batch of cells with 9 jig of actinomycin-d per ml for 15 min before addition of [14C] aminoacids inhibited incorporation by 65%. An equal amount of 14C (10,500 cpm) as ghosts was applied to each column. Free [14C]aminoacids applied to columns instead of [14C]- ghosts. compared to the derivative with no arm. This finding suggests that the arm and components necessary for interposing the arm, in contradistinction to the insulin, are not responsible for the binding. Measurements were also made of the binding of 14Clabeled ghosts to insulin-s (Table 2). Results comparable to those determined by the enzyme staining and turbidimetric methods were obtained. Kono (19) has shown that trypsin (1 mg/ml) completely abolishes the insulin sensitivity of fat cells without significantly affecting the baseline rate of glucose utilization; sensitivity to insulin was restored after a 90-min incubation of the insulin-insensitive cells with soybean trypsin inhibitor (8, 19). Kono attributed these effects on the insulin sensitivity to the insulin receptor (19). We have confirmed these findings. TABLE 3. FIG. 2. Spontaneous buoyant-density distribution at 300 of fat cells, Sepharose, and ILys5S 10 min after mixing different proportions of cells and beads in Krebs-Ringer buffer (ph 7.4) containing 0.2%7o bovine-serum albumin. To each tube was added 0.50 ml of packed Sepharose (tube A) or ILys5S (700 jg of insulin coupled per ml of packed Sepharose) (tubes B, C, and D), , 0.3, and 0.2 ml of packed cells (to A, B, C, and D, respectively), and buffer to a final volume of 1.5 ml. Table 3 shows that cells treated under conditions identical to those resulting in the loss and regain of insulin sensitivity demonstrated a concomitant loss and restoration of binding of the ghosts derived from these cells. Trypsin at a concentration of 4 mg/ml apparently caused irreversible damage to the insulin binding sites, since the additional 90-min incubation (in the presence of trypsin inhibitor) failed to regenerate any binding capacity. However, incubation after treatment with lower concentrations of trypsin resulted in from 70 to 95% restoration of binding capacity. During the course of these studies we discovered that fat cells readily associate with insulin-s beads, resulting either in floatation or sedimentation of the cells and beads, depending upon the proportion of cells to beads. The illustration in Fig. 2 shows that after mixture of Sepharose beads with viable fat cells (tube A), the beads and cells completely and rapidly Effect of trypsin treatment of cells on binding of ghosts to ILys5S columns Ghosts bound After trypsin treatmentt After reincubationt Trypsin STI Added Pooled wash % Bound STI Added Pooled wash % Bound (mg/ml) (turbidity) (turbidity ) N.P. N.P N.P. N.P Cells were incubated with trypsin for 15 min at 37 in Krebs-Ringer bicarbonate buffer (ph 7.4) in the presence (+) and absence (-) of soybean trypsin inhibitor (STI) (19). t Ghosts were prepared from an aliquot of trypsin-treated cells after addition of inhibitor and immediately tested for binding capacities. t Remaining portion of treated cells was incubated for 90 min at 370 in the presence or absence of inhibitor, after which inhibitor was added (when previously absent) and ghosts were prepared for testing binding capacities. Determined at 450 nm on ghosts "added" to, and eluted from, columns. N.P. refers to step not performed.

4 Insulin-Sepharose Binding Fat Cells and Membranes Proc. Nat. A cad. Sci. USA 70 (1978) 795 TABLE 4. Binding specificity of insulin-s according to buoyant density distribution Beads sedi- S Trypsin- mentedt Buoyant: Preparation treatment (cpm) (normalized) ILys5S ILys5S ILysS Glycine5S ILys5S + antiserum to insulin ILys5S + normal serum Insulin-S (750 lug of insulin per ml of packed Sepharose) was labeled with tracer [l25i]insulin; glycine5s was prepared identically, except that glycine was substituted for native insulin. t The ratio of cells to beads was adjusted so that about 15% (138 cpm) of the JLys5S sedimented after mixing cells and beads; this ratio was also used in the other experiments. I The cpm sedimenting 15 min after mixing beads and trypsin (1 mg/ml)-treated (+) or untreated (-) cells were recorded as percent of the total cpm in each tube (about 890 cpm) compared (normalized) to the percent of ILys5S that sedimented from the cells untreated with trypsin. Antiserum to insulin or normal serum was added (to 1 mg/ml) with gentle mixing 5 min after cells were complexed with beads. separated from each other, leaving distinct layers of buoyant cells and sedimented beads. However, under identical conditions, except substituting ILys5S for Sepharose, all of the beads floated with the cells (tube B). Conversely, an excess of ILys5S completely sedimented the cells (tube D), while an intermediate proportion of cells to beads (tube C) resulted in cells associated with beads distributed throughout the tube. It became clear that the number of cells bound per bead substantially determined the buoyant density of the resultant complex. In Fig. 3 are illustrated examples of cell-ilys5s complexes as seen by Nomarski interference contrast microscopy. When a mixture (taken from tube B, Fig. 2) of insulin-s beads and cells were examined, beads (in virtually the same plane of focus as cells) could clearly be seen to be surrounded by cells bound to each bead (Fig. 3, panel A). However, since control Sepharose beads settle beneath the floating cells on the microscope slide, almost no beads could be seen in a mixture of Sepharose and cells when the cells were in focus (not shown). More dilute mixtures containing from one to three cells bound per ILys5S bead are seen in panels B, C, and D of Fig. 3. Because of the rapid separation of cells from Sepharose beads noted above, cells and ILys5S beads could not be seen simultaneously in focus unless they were bound to each other, as in these photographs. It is important to note that these cells were seen to stay bound to, and move with, the insulin-s beads when convection currents created movement of the fluid medium on the slides. In Tables 4 and 5 are examples of results of a study of the specificity of the binding of insulin-s to intact fat cells, as determined by buoyant density distributions of insulin-s tracer-labeled with [1251I]insulin. Treatment of the cells with 1 mg of trypsin per ml reduced the binding capacity of the cells to an extent that, when measured under these conditions, TABLE 5. Relative effects of proteins on binding of cells to ILys5S Beads A % Beads Addition sedimentedt sedimented (mg/ml) (cpm) (above baseline) None Insulin (0.5) BSA, fract. V (40) Gelatin (40) Beef GH (4.8) Ovine GH (4.8) Human GH (4.8) Prolactin (3.0) LH (0.5) FSH (3.0) Glucagon (3.0) Anti-insulin antiserum (2.0) Each compound added to cell suspension to its final designated concentration immediately before addition of tracer [1 I] insulinlabeled ILys5S containing 1.9 mg of insulin per ml of packed Sepharose. t Ratio of cells to beads was adjusted, resulting in baseline of 437 cpm of beads sedimented; if desired, 0% of beads could be sedimented at higher ratios. I The sedimented cpm greater than baseline were compared to the total cpm (about 1700 sedimented and floating) in each tube. Measurements of the volumes of cells and beads (instead of cpm) were consistent with these results. BSA, bovine-serum albumin; GH, growth hormone; LH, leutinizing hormone; FSH, folliclestimulating hormone. about 50% of the beads lost their buoyancy. Higher concentrations of trypsin more extensively abolished the binding capacity of the cells. Insulin-S was appreciably more effective with, than without, the connecting arm, and glycine linked to Sepharose via an arm exhibited no binding capacity. Binding could be extensively blocked (Table 5) and reversed (Table 4) by antiserum to insulin. Significantly, prior exposure of insulin-s to antiserum to insulin followed by extensive washings of the beads with buffer completely inhibited the beads' binding capacity, while similar prior treatment with excess FIG. 3. Nomarksi interference contrast microscopy (X 500) of concentrated (A) and diluted (B, C, and D) mixtures of cells and ILys5S beads. Larger spheres are beads. See Fig. 2.

5 796 Biochemistry: Soderman et al. insulin had no effect. The binding capacity of these antiserumblocked beads could be completely restored by washings with 6 M guanidine HCl. Insulin at 10-5 M specifically inhibited the binding of 4 X 10-5 M immobilized hormone to the cells, while other proteins and hormones (Table 5) had lesser effects at higher concentrations. It is likely that at the point of attachment of membrane to bead, multiple receptor sites of binding are involved. Also, the local concentration of immobilized insulin near these sites may be high, so that the concentration of insulin required to inhibit or reverse binding may also be high. Whether the lesser effects of the other proteins indicate some nonspecific binding capacity inherent in the insulin-s, or reflect binding of these proteins to membrane sites adjacent to the insulin receptor thereby blocking it (directly or indirectly), is not known. The possibility that zinc associated with the immobilized insulin may be involved in the binding was eliminated by our findings that (a) zinc-free ILys5S exhibited binding properties identical to those seen in Tables 4 and 5, and (b) 10-2 M ZnC12 had no effect. DISCUSSION The specific binding of isolated fat cells and their plasma membranes to immobilized insulin may represent a promising new tool to study the insulin receptor. In a reference to membrane binding to insulin-agarose cited as "unpublished data" in a recent review (20), the possibility of utilizing insoluble hormone derivatives to separate cell populations according to specific functions was suggested. Previous investigations in our laboratory (1, 2) and by Cuatrecasas (3) have indicated the potential of affinity chromatography for the isolation and purification of detergent-solubilized membrane fractions capable of binding [1251 linsulin. In addition, the present findings may provide a method, possibly in conjunction with the use of free [125I]insulin (7-9), to test the hypothesis of Oka and Topper that rather than a tight binding, the collision with, or detachment of the hormone from, the cell may be required for hormonal activity (6). This hypothesis was proposed on the basis that no binding could be observed between IPheS beads and mammary epithelial cells sensitive to the immobilized, but not the free, hormone (6). While further studies will be required to relate specific binding to biological activity, the present studies demonstrate the ability of several insulin-s preparations to bind firmly viable, intact fat cells under conditions compatible with the expression of bioactivity of these (see Methods) and other (4-6) insulin-agarose preparations. Although several questions have recently arisen (21, 22) concerning the validity of some of the earlier studies (4) on whether "free" or "immobilized" insulin was responsible for the observed bioactivities, a demonstrated binding to insoluble beads, by definition, does not involve solubilized hormone. Several lines of evidence presented demonstrate the specificity of the binding by insulin-s. They are also consistent with the concept that such firm binding is involved in, or at least does not mitigate against (6), insulin stimulation Proc. Nat. Acad. Sci. USA 70 (1978) of the fat cell. Included in this evidence is the specific inhibition and reversal of this binding by antiserum to insulin and by insulin, as well as the presence of trypsin-sensitive sites on the fat-cell membrane required for both binding and biosensitivity. It is interesting to note that no more than 75% of the total amount of ghost preparation applied to each affinity column could be found bound (Fig. 1 and Tables 1, 2, and 3). This may be comparable to the proportion of sacs of plasma membranes in the ghost preparation. It is known that this preparation contains, in addition to ghosts, free nuclei and other unenclosed intracellular particles (13). Thus, these affinity columns may be capable of purifying receptor-containing plasma membranes. While none of these lines of evidence exclude the possible involvement of biologically "nonproductive" insulin-binding sites, they do demonstrate the ability of immobilized insulin to bind specific receptor sites on the membrane in a manner comparable to the binding and biological properties indicative of native insulin. We thank Mliss Brenda Halstead for valuable technical assistance with the fat-cell bioassays, and Dr. Harry Carter for the microscopic examinations of the cell-bead mixtures. 1. Soderman, D. D., Germershausen, J. & Katzen, H. M. (1972) Fed. Proc. 31, Katzen, H. -AI. & Soderman, D. D. (1972) in Membranes in Metabolic Regulation, eds. Mehlman, M. A. & Hanson, R. W. (Academic Press, New York), pp Cuatrecasas, P. (1972) Proc. Nat. Acad. Sci. USA 69, Cuatrecasas, P. (1969) Proc. Nat. Acad. Sci. USA 63, Blatt, L. X. & Kim, K.-H. (1971) J. Biol. Chem. 246, Oka, T. & Topper, Y. J. (1971) Proc. Nat. Acad. Sci. USA 68, Freychet, P., Roth, J. & Neville, D. M. (1971) Proc. Nat. Acad. Sci. USA 68, Kono, T. & Barham, F. W. (1971) J. Biol. Chem. 246, Cuatrecasas, P. (1971) Proc. Nat. Acad. Sci. USA 68, Lambert, B., Sutter, B. & Jacquemin, C. (1972) Horm. Metab. Res. 4, Freychet, P., Roth, J. & Neville, D. WI. (1971) Biochem. Biophys. Res. Commun. 43, Rodbell, M. (1964) J. Biol. Chem. 239, Rodbell, MI. (1967) J. Biol. Chem. 242, Gliemann, J. (1967) Diabetologia 3, Cuatrecasas, P. & Anfinsen, C. (1971) Methods Enzymol. 26, Katzen, H. W., Soderman, D. D. & Wiley, C. (1970) J. Biol. Chem. 245, Axen, R., Porath, J. & Ernback, S. (1967) Nature 214, Cuatrecasas, P., Wilchek, M. & Anfinsen, C. B. (1968) Proc. Nat. Acad. Sci. USA 61, Kono, T. (1969) J. Biol. Chem. 244, Cuatrecasas, P. (1971) in Biochemical Aspects of Reactions on Solid Supports, ed, Stark, G. R. (Academic Press, New York), pp Davidson, M. B., Gerschenson, L. E. & Van Herle, A. J. (1972) Diabetes 21, Suppl. No. 1, Katzen, H. M. & Vlahakes, G. (1973) Science, in press.