- Email: [email protected]
A role for the lipoxygenase pathway of arachidonic acid metabolism in glucose- and glucagon-induced insulin secretion
Life Sciences, Vol. 32, pp. 903-910 Printed in the U.S.A.
Pergamon Press
A ROLE FOR THE LIPOXYGENASE PATHWAY OF ARACHIDONIC ACID METABOLISM IN GLUCOSE- AND GLUCAGON-INDUCED INSULIN SECRETION Stewart A. Metz, M.D., Wilfred Y. Fujimoto, M.D., R. Paul Robertson, M.D. Divisions of Clinical Pharmacology, and Endocrinology and Metabolism, and the Departments of Medicine University of Colorado Health Sciences Center Denver, Colorado 80262 and The University of Washington Seattle Washington, 98195 (Received in final form November 15, 1982)
Summary Although the cyclo-oxygenase pathway of arachidonic acid (AA) metabolism inhibits glucose-stimulated insulin release through synthesis of prostaglandins, very l i t t l e attention has been given to the effects of lipoxygenase pathway products on beta cell function. We have examined the effects of two structurally-dissimilar lipoxygenase inhibitors on insulin release from monolayer-cultured rat islet cells. Both nordihydroguaiaretic acid (NDGA, 20-50 ~M) and BW755c (100-2501JM) caused a dose-responsive inhibition of glucoseinduced insulin release. This inhibitory effect occurred despite concomitant inhibition of prostaglandin E synthesis. Lipoxygenase inhibitors also impeded cyclic AMP accumulation. Insulin and cyclic AMP release induced by glucagon were also blunted. These studies suggest the hypothesis that AA released in or near the beta cell is metabolized to lipoxygenase product(s) which have feedforward properties important to glucose- and glucagon-stimulated cyclic nucleotide accumulation and insulin release. Glucose-induced insulin (IRI) secretion is known to be associated with turnover of membrane phospholipids (1). More recently, it has been suggested that glucose can activate arachidonic acid uptake and release in the endocrine islet (2,3). This is of interest to studies of IR[ release since others have recently demonstrated that exogenous phospholipase mimics the effects of glucose on phospholipid turnover and insulin release (4). The link between such membrane effects and insulin release has remained elusive. A role for long chain, polyunsaturated fatty acids such as arachidonic acid, which might be released during such membrane perturbations, has only recently been studied. We have recently shown both in vivo (5) and in vitro (6) that a product of the cyclo-oxygenase pathway, prostaglandin E, inhibits glucose-induced insulin secretion; correspondingly, inhibitors of PGE synthesis such as sodium salicylate and ibuprofen potentiate glucoseinduced insulin release. On the other hand, arachidonic acid released from cell membrane phospholipids can also be metabolized via an alternative pathway, the lipoxygenase pathway, whose function in the endocrine islet is largely unexplored. Since many cell membrane-active compounds that are known to activate phospholipases (e.g., bradykinin, calcium ionophore A 23187, phorbol esters, and furosemide) stimulate, rather than inhibit, insulin release (7-9), we considered the possibility that the lipoxygenase cascade might represent a stimulatory pathway within the beta cell. Therefore, we studied the effect of combined lipoxygenase and cyclo-oxygenase inhibitors (since no selective lipoxygenase
0024-3205/83/080903-08503.00/0 Copyright (c) 1983 Pergamon Press Ltd.
904
Lipoxy~enase
and Insulin
Vol.
32, No. 8, ]983
inhibitor is available) upon glucose-induced insulin secretion using monolayer cultures of neonatal rat pancreatic islet ceils. We reasoned that if products of the Iipoxygenase pathway have either no e f f e c t on insulin release or an inhibitory e f f e c t similar to that seen with PCs, the e f f e c t of these combined inhibitors would be stimulatory. In contrast, if lipoxygenase products are stimulatory, the e f f e c t on insulin release might be null or even inhibitory. M a t e r i a l s and M e t h o d s
BW755c ( 3 - a m i n o - l - _ t r i f l u o r o m e t h y l - p h e n y ~ -2-pyrazoline) was generously provided by Dr. P.3. McHale of the Welicome Research l a b o r a t o r i e s (Kent, England). The antioxidant nordihydroguaiaretic acid (NDGA) was purchased from Sigma Chemical Co. (St. Louis, Missouri). Radioimmunoassay kits for the d e t e r m i n a t i o n of cyclic AMP released into the medium were purchased f r o m New England Nuclear (Boston, Mass.). The methods used for establishing monolayer cultures of neonatal rat pancreatic islet cells, enriched for endocrine ceils, have been previously described in detail (S). In brief, this method determines the cumulative release of insulin, prostaglandin E and cyclic nucleotides into the medium over a static 1-hr incubation period; thus, values reported represent t o t a l release and do not r e f l e c t specifically first or second phase insulin secretion. Control plates contained the same diluent (water for BW755c and glucagon, and ethanol for NDQA) as the experimental plates. Prostaglandin E (10), insulin (11), and cyclic AMP release (12) were measured by previously published radioimmunoassay procedures. The assay for cyclic AMP has a sensitivity ( m i n i m u m - d e t e c t a b l e - a m o u n t ) of less than 25 (routinely 10) f e m t o m o l e s / m l (2 standard deviations from the mean of replicate determinations of Bo=2.9%, B/B o at 25 femtomoles/ml=94%; n=10 determinations each). The radioimmunoassay is able to detect clearly changes of 15-20 f e m t o m o l e s / m l at cyclic AMP levels in culture medium of 150200 f e m t o m o l e s / m l . Neither of the drugs studied had significant effects upon recoveries of the relevant compounds in the radioimmunoassays or HPLC procedure. I d e n t i f i c a t i o n and measurement of 12-hydroxyeicosatetraenoic acid (12-HETE), as an index of lipoxygenase a c t i v i t y , were performed a f t e r incubating dispersed endocrinec ell-enriched pancreatic cells or intact rat islets w i t h 3H-arachidonic acid in the absence and presence of lipoxygenase inhibitors, and analyzing the medium, a f t e r e x t r a c t i o n into m e t h y l f o r m a t e , using reverse phase high performance liquid chromatography and s c i n t i l l a tion counting. For the ceil labelling procedure, pancreases from 30 rats were e n z y m a t i caily digested and mechanically dispersed, as for the monolayer cultures (6). However~ a f t e r the first overnight culture period, rather than being re-plated, the cells were harvested by collection of the medium containing non-attached cells and equally aliquoted into 3-4 test tubes. Ceils were labelled with t r i t i a t e d arachidonic acid (2 p C i / m l ) in a low glucose concentration (30 mg/dl) p u t a t i v e l y to m i n i m i z e phospholipase a c t i v i t y (3) and medium was collected a f t e r a 2-hour incubation period. Esterified arachidonate was then released f r o m membrane stores over the ensuing 30 minutes using either the ionophore A23187 or bradykinin in a high glucose concentration (300 mg/dl) to m a x i m i z e phospholipase a c t i v a t i o ~ (2,3). 12-HETE values were calculated as the sum of the CPM Muting with authentic ~H-12-HETE in the direct labelling and deacylation steps, and expressed as percent of t o t a l CPMs eluting from the H P / C column. These procedures have been described in detail (13). The primary column used for HPLC was a free f a t t y acid column (Waters Associates, M i l f o r d , MA). The solvent system ( a c e t o n i t r i l e / w a t e r / a c e t i c acid) was applied isocratieaUy for 50 minutes, and then increased fro, m 23% to 80% a c e t o n i t r i l e over a 30 min gradient. 12-HETE elutes at 67 to 69 minutes. -'H-12-HETE was identified by precise co-elution with authentic t r i t i a t e d 12-HETE when samples were analyzed as the free acid (in this and three additional HPLC systems) or w i t h 12-HETE methyl ester (in a fourth system) a f t e r m e t h y l a t i o n of samples using diazomethane in ether. Statistical analyses were by Newman-Keuls test for multiple comparisons and Student's t test for isolated comparisons.
Vol. 32, No. 8, 1983
Lipoxygenase and Insulin
905
Results With increasing concentrations of BW755c, there was a progressive decline in prostagtandin E synthesis that was maximal by 10 #M, and which was accompanied at this concentration by a modest inhibition of 12-HETE f o r m a t i o n (-b,1%). These effects were unaccompanied by any change in [RI secretion (Fig. 1). However, at higher concentrations (50-250 ]~M), 12-HETE was further inhibited (by 54% at 250 #M) whereas PGE synthesis did not fall further, thus providing a selective increase in inhibition of lipoxygenase a c t i v i t y at progressively higher concentrations of BW755c. A t these higher drug concentrations, there was a dose-dependent inhibition of glucose-stimulated (glucose = 300 mg/dt) insulin release (Fig. 1) to values indistinguishable (at 250 MM) from the low glucose (30 mg/dl) control (50 +- 2 vs. 52 + 2 p U / m l ; p=ns). Additional studies were then performed to study further the glucose-dependency of this inhibitory effect. BW755c did not reduce insulin release at a sub-stimulatory glucose concentration (30 mg/dI), had an intermediate inhibitory e f f e c t near the glucose threshold (100 mg/dl), and was m a x i m a l l y inhibitory e f f e c t at 300 mg/dl (Table 1). Thus BW755c t o t a l l y abrogated the glucose responsivity of these cells. Effects of this drug were unaccompanied by evidence of cell t o x i c i t y at any concentration studied, as assessed by phase contrast microscopy and trypan blue exclusion.
600
240 ~+SEM
500
4
n -- 2 - 4
PLATES
EACH
200
m
\E 400Q.
160 (/) I-
--//-5 ",
Z
I~~ 300-
t20
\ \\
£9
\
Q.
\
ULIN \
200.
-80
40
100-
"--%
.....
....... PGE
o
(6)#o,.5
~5
1'0
2"5
50 I00
250
o
BW755c, pM FIG. 1 Effects of various concentrations of BW755c on insulin release at 300 mg/dl of glucose (solid line) and PGE synthesis (dashed line). Inhibition of PQE was significant (p<.001) at 2 IJM or higher and inhibition of insulin release was significant from 100 IJM on (p<.005 or greater). Synthesis of 12-HETE was inhibited by 41% at 1O NM but by 54% at 250 1JM.
906
Lipoxygenase and Insulin
TABLE
Vol. 32, No. 8, 1983
I
EFFECT OF BW755c (250 pM) ON INSULIN RELEASE ( p U / m l ) AT VARIOUS GLUCOSE (Q~ mg/dl) C O N C E N T R A T I O N S Condition
Control
BW755c
G 30
, 33 +_ 2 " ~
35 _+ 0.4 :.- p<. 001
i G I00 p<.O01< G 300
' 4 8 _+ I - ' i I -
p<.O01 . . . . . .
~-59 _+ 4
33 + 2 L 36 + 2
p<.Ol Values are ~ _+ SEM; n=4 p l a t e s each
40 I I
14OO-
d
(.9 (3_
~
..... ~
140 -H
o
,oo
0 . 5 2 IO 20 SODIUMmg/dlSALICYLATE,
v I
y -Iz SEM
n=4 PLATES EACH 170
12OO
~50
I OOO
130
800
IlO
soo
9O
c
400
70
3 --
200
50
O
,
(0)
//
o) C r-
30 o. I
0.5
2
5
20
5'O
NORDIHYDROGUAIARETIC ACID, uM FIG. 2
Effects of various concentrations of nordihydroguaiaretic acid on insulin release (solid line) and POE synthesis (dashed line) at 300 mg/dl of glucose. Decrements in PGE synthesis were significant (I)<.005 or greater) aL 0.1 pM or higher. Changes in insulin release were significant (13<.02 or greater) at 2-50 pM. Insert shows effects of sodium salicylate for comparison (data from ref. 6). Synthesis of 12-HETE was inhibited by only 17% at 5 pM but by 89% at 50 pM.
Vol. 32, No. 8, 1983
Lipooxgenase
and Insulin
907
NDGA also was a potent inhibitor of cyclo-oxygenase at low concentrations (maximal effect seen at 5 pM), at which concentration 1 2 - H E T E formation was only minimally inhibited (-17%). This selective cyclo-oxygenase inhibition was accompanied by a progressive rise in insulin release (Fig. 2) which closely mimicked the effects of the prostaglandin synthesis inhibitor (6) sodium salicylate (Fig. 2, insert). At higher concentrations of N D G A , P G E synthesis did not decline further but inhibition of 1 2 - H E T E formation reached 8 9 % at 50 ]aM. Concomitantly there was a decline in insulin release to values significiantly less than the high glucose (300 mg/dl) control (control = 70 _+ 4; N D G A , 20 p M = 38 +_ 2 pU/ml; p<.01) and indistinguishable from the low glucose value (glucose, 30 mg/dl = 40 +_ 3 ]aU/ml). At 0-20 ]aM, N D G A was unaccompanied by evidence of cell toxicity, although at concentrations of ~. 50 ]aM, there was rounding and detachment of some cells assessed by phase contrast microscopy, an effect which appeared to preferentially affect fibroblasts. Simultaneous addition of maximally effective concentrations of BW755c and NDGA led to an inhibitory effect on insulin release not greater than that of either drug alone (data not shown). To examine the possibility that the inhibitory effects of lipoxygenase inhibitors were associated with changes of cyclic nucleoLide accumulation, measurements were made of cyclic AMP released into the medium, a sensitive index of the effects of drugs on synthesis and release of this nucleotide (12). Both lipoxygenase inhibitors caused a dosedependent inhibition of cyclic AMP accumulation (Fig. 3). The dose-response relationships showed that this inhibition of cyclic AMP could be readily dissociated from the coincident
180"
I } 160" D.
T. . . .
....
L
~ 140. <
\
I ..~-4 ~.axes eACN
\~
010
BW755C, pM 250
iz~u .-~-4 PLATES~ACH
2OO ¢L '150 O
\
\
T00
\ X
\-L
U 50
(0) "YO~I 05 I 5 NORDIHYOROGUAIARETIC ACID,
10
50
pM
FIG. 3 Effects of BW755c and NDGA on cyclic AMP accumulation in the medium. Decrements were significant as compared to glucose 300 mg/dl alone (p<.05 or greater) at 50 pM and 20 [aM, respectively. inhibition of prostaglandin synthesis (for example, compare Fig. 3 to Fig. i and 2) but seemed to parallel closely the inhibition of lipoxygenase and of insulin release.
908
Lipoxygenase and Insulin
Vol. 32, No. 8, 1983
Furthermore, BW755c also reduced cyclic AMP accumulat!on and insulin release induced by glueagon, 250 ng/ml (Fig. 4).
200
7
IOOO 900
+ SEM
n = 4 plates(INSULIN) 150
i
600
3 plates (cAMP)
I00
4OO 2
D r.,O z
INS I
50
~\\\,
200
i
o--
CONTROL (H20)
GLUCAGON, 250 ng/ml
[
BW 755c, 250 pM
BW 755c (~) GLUCAGON
o
FIG. 4 Effect of BW755c (250 pM) and glucagon (250 ng/ml) alone and in combination on insulin (wide, open or shaded bars) and cyclic AMP (narrow, cross-hatched bars) release. Differences between any two groups were statistically significant (p< .05 or greater). Discussion The e f f e c t s on insulin r e l e a s e of BW755c and N D G A (two s t r u c t u r a l l y - d i s s i m i l a r i n h i b i t o r s of a r a c h i d o n i c acid m e t a b o l i s m ) m u s t t a k e into a c c o u n t t h e d i f f e r e n t i a l sensitivities of cyclo-oxygenase and lipoyxgenase to these drugs, as we and others (14-16) have observed. BW755c and NDQA inhibit glucose-induced insulin secretion at concentrations of greater than 50 pM and 5 BM, respectively. These same concentrations inhibited lipoxygenase, as assessed by 12-HETE measurements. At lower concentrations, prostaglandin E synthesis was maximally and relatively selectively inhibited by NDGA, with 12-HETE formation being inhibited by less than 20%. A comparison to the effects of sodium salicylate and ibuprofen in the same experimental system (6,12) suggests that the stimulatory effect of NDGA at these lower concentrations was likely due to a selective inhibition of prostaglandin synthesis. Although some authors find that certain prostaglandins increase insulin release, especially at lower glucose concentrations (17), in the current preparation POE inhibits glucose-stimulated insulin release (6,12), and, conversely, inhibition of prostaglandin synthesis induces a progressive, curvilinear augmentation of glucose-induced insulin release (6). The partial inhibition of lipoxygenase (about 40%) by BW755c at low concentrations (10-100 pM) was submaxima] but was greater than that due to 5 wM NDGA and consequently was sufficient to block the stimulation of insulin release otherwise associated with cyclo-oxygenase inhibition. The
Vol. 32, No. 8, 1983
Lipoxygenase and Insulin
909
progressive reduction in insulin secretion below control values seen at higher concentra tions of either agent and associated with further inhibition of 12-HETE synthesis is compatible with the conclusion that the lipoxygenase pathway is stimulatory to insulin release. This formulation is supported by our observations that exogenously provided or endogenously released arachidonie acid potently stimulates insulin release (13,18). This effect can be further enhanced by cyelo-oxygenase inhibition and totally blocked by lipoxygenase inhibition. Thus our data suggest that arachidonic acid released in (or near) the beta cell can be metabolized to products which modulate insulin secretion by dual pathways--cyclo-oxygenase (inhibitory) and lipoxygenase (stimulatory). They do not, however, address the issue of whether insulin secretagogues augment release of lipoxygenase metabolites, or whether a tonic level of lipoxygenase activity plays a requisite permissive role for insulin release. In favor of the former possibility are the data of Evans and Clements (2) and of Laychock (3) that glucose activates arachidonic acid uptake and release in the endocrine islet, and other data (for review~ see 19) that glucose modifies lipoxygenase-mediated arachidonate metabolism in several cell systems. It is also possible that both drugs decreased insulin release by some unrecognized toxic effect on the cells (or via a primary effect to inhibit cyclic AMP accumulation; see below). While we cannot fully exclude these possibilities, it seems unlikely that two structurally-dissimilar drugs which both inhibit lipoxygenase should also share an undefined toxic effect, especially since morphologic and functional evidence of cell injury was sought and not found. We emphasize that 12-HETE measurements are used in the current study only as a specific marker for the lipoxygenase pathway; the arachidonate metabolite(s) responsible for the stimulatory effect remain to be elucidated. In fact, exogenous 12-HETE fails to alter glucose-stimulated insulin release at concentrations from 0.5-2000 ng/ml (S. Metz, R. Murphy, W. Fujimoto, R.P. Robertson, manuscript in preparation). In the absence of identification and measurement of the active metabolite of arachidonic acid over a full range of drug concentrations and under identical conditions as those used to monitor insulin release, perfect correlations between lipoxygenase activation (as assessed by 12HETE synthesis) and insulin release are not to be expected. However, our data (and those of others, 20) using 12-HETE as a marker, show that a lipoxygenase exists in the endocrine islet; that it is potently inhibited by concentrations of BW755c and NDGA which also inhibit insulin release; and that at lower drug concentrations, the effect of lipoxygenase inhibition becomes less evident and is counterbalanced (e.g, BW755c at 10-50 ]JM) or superseded (e.g., NDQA at 5 pM) by the effects of relatively selective cyclo-oxygenase inhibition. Our results suggesting that lipoxygenase inhibition can prevent glucose-induced insulin release may be relevant to the suggestion (2,3) that glucose activates a phospholipase in the pancreatic islet, an effect which could increase the availability of arachidonate. Taken in concert with the observation that exogenously provided phospholipase can mimic the effects of glucose on phospholipid turnover and insulin release (4), these observations permit the hypothesis that lipoxygenase-mediated metabolism of arachidonic acid might be an intermediate step in stimulus-secretion coupling during glucose-induced (or glucagon-induced) insulin release. This conclusion is compatible with reports that compounds which inhibit phospholipase(s), such as trifluoperazine (21) and mepacrine (21), also inhibit glucose-stimulated insulin release (9,22). Trifluoperazine has been shown to act as a combined cyclo-oxygenase/lipoxygenase inhibitor by virtue of inhibiting a calmodulin-dependent phospholipase (21,23); conversely, the activation of phospholipase (23) and of adenylate cyclase (24) by calcium and calmodulin raises the possibility that glucose-induced alterations in calcium flux may, in turn, a c t i v a t e oxygenated arachidonate metabolites as "second messengers." Indeed, the current data also demonstrate that an adequate degree of lipoxygenase inhibition impedes cyclic AMP production by cultured islet cells; therefore, cyclic nucleotides could mediate part of the stimulatory effect of lipoxygenase products upon insulin secretion. However, such a causal relationship is not established by the current studies and we cannot totally exclude
910
Lipoxygenase and Insulin
Vol. 32, No. 8, 1983
a non-specific impairment of cyclic AMP accumulation by NDQA and BW755c. The possibility of a specific relationship between lipoxygenase products and cyclic nucleotide formation is supported by recent studies of Levasseur and co-workers (25) who have provided evidence that a lipoxygenase pathway in the thyroid may mediate a part of stimulatory actions of TSH on cyclic AMP accumulation. Others have also found evidence that lipoxygenase products augment cyclic AMP accumulation (26). Products of the lipoxygenase pathway are also known to potently augment cyclic QMP formation in certain systems (27). Whatever the mechanism, the current studies suggest that oxygenation products of arachidonic acid may provide both stimulatory and inhibitory influences on glucose-induced insulin release, depending upon whether the lipoxygenase or the cyclo-oxygenase pathway is dominant in a given circumstance. Acknowledgments The authors gratefully acknowledge the excellent technical assistance of Ms. 3eanette Teague, Ms. Lytian Fuller, and Mr. David Federighi. This work was supported by the Veterans Administration, a grant from the Upjohn Company, and grants AM-15312 and AM-17847 from the National Institutes of Health. References 1. 2. 3. 4.
R.S. CLEMENTS Jr. and W.B. RHOTEN, J. Clin. Invest. 57 684-691 (1976). M.H. EVANS and R.S. CLEMENTS Jr., Clin. Res. 29 404a---~1981). S.G. LAYCHOCK, Fed. Proc. 40 458a (1981). K. TANIGAWA, H. KUZUYA, H. IMURA, H. TANIGUCHI, S. BABA, Y. T A K A I and Y. MISHIZUKI, FEBS Letters. 38 183-186 (1982). 5. R.P. ROBERTSON and M. CHEN, J. Clin. Invest. 60 747-753 (1977). 6. S.A. METZ, R.P. ROBERTSON and W. FU3IMOTO, Diabetes. 30 551-557 (1981). 7. 3.E. VANCE and S.C. BRAGG, Diabetes (Suppl). 18 326-327 (1969). 8. W..]. MALAISSE, A. SENER, A. HERCHUELZ, A.R. CARPINELLI, P. POLOCZEK, 3. WINAND and M. CASTAGNA, Cancer Res. 40 3827-3831 (1980). 9. R. LANDGRAF and M.M.C. LANDQRAF-LEURS, Prostaglandins. 17 599-613 (1979). 10. R.P. ROBERTSON. Am. 3. Physiol. 228 68-70 (1975). 11. D.S. ZAHARKO and L.V. BECK, Diabetes. 17 444-457 (1968). 12. S. METZ, W. FUSIMOTO and R.P. ROBERTSON. Metabolism, in press (1982). 13. S. METZ, S. LITTLE, W. FUJIMOTO and R.P. ROBERTSON, Adv. Prost. Thromb. Res., in press (1982). 14. M.I. SIEGEL, R.T. MCCONNELL, R.W. BONSER and P. CUATRECASAS, Prostaglandins. 21 123-132 (1981). 15. C.L. ARMOUR, 3.M. HUES, 3.P. SEALE and D.M. TEMPLE, Eur. 3. Pharmacol. 72 93-96 (1981). 16. M.A. BRAY, A.W. FORD-HUTCHINSON, M.E. SHIPLEY and M.J.H. SMITH, Br. 3. Pharmacol. 71 507-512 (1980). 17. T.P. HEANEY and R.G. LARKINS, Diabetes 30 824-828 (1981). 18. S. METZ, S. LITTLE, W. FU3IMOTO and R.P. ROBERTSON, Clin. Res. 30 63a (1982). 19. S.A. METZ, Prostagland. Med. 7 581-589 (1981). 20. K.I. K E L L Y and S.G. L A Y C H C ~ C K , Prostaglandins. 2__l_759-769 t (1981). 21. R . W . W A L E N G A , E.E. Q P A S and M.B. FEINSTEIN, 3. Biol. Chem. 256 12523-12528 (1981). 22. Y. KRAUSZ, C.B. WOLLHEIM, E. SIEGEL and G.W.G. SHARP, J. Clin. Invest. 66 603-607 (1980). 23. P.A. C R A V E N , R.K. S T U D E R and F.R. DeRUBERTIS, .]. Clin. Invest. 68 722-732 (i98i). 2/4. G.W.G. S H A R P , D.E. W E I D E N K E L L E R , D. KAELIN, E.G. SIEGEL and C.B. W O L L H E I M , Diabetes. 29 7/4-77 (1980). 25. S. L E V A S S E U R , F.F. SUN, Y. F R I E D M A N , and Q. B U R K E , Prostaglandins 22 663673 (1981). 26. H.-E. C L A E S S O N , FEBS Letters 139 305-308 (1982) 27. H. H I D A K A and T. A S A N O , Proc. Nat'l. Acad. Sol. USA. 7/4 3657-3661 (1977).
Pergamon Press
A ROLE FOR THE LIPOXYGENASE PATHWAY OF ARACHIDONIC ACID METABOLISM IN GLUCOSE- AND GLUCAGON-INDUCED INSULIN SECRETION Stewart A. Metz, M.D., Wilfred Y. Fujimoto, M.D., R. Paul Robertson, M.D. Divisions of Clinical Pharmacology, and Endocrinology and Metabolism, and the Departments of Medicine University of Colorado Health Sciences Center Denver, Colorado 80262 and The University of Washington Seattle Washington, 98195 (Received in final form November 15, 1982)
Summary Although the cyclo-oxygenase pathway of arachidonic acid (AA) metabolism inhibits glucose-stimulated insulin release through synthesis of prostaglandins, very l i t t l e attention has been given to the effects of lipoxygenase pathway products on beta cell function. We have examined the effects of two structurally-dissimilar lipoxygenase inhibitors on insulin release from monolayer-cultured rat islet cells. Both nordihydroguaiaretic acid (NDGA, 20-50 ~M) and BW755c (100-2501JM) caused a dose-responsive inhibition of glucoseinduced insulin release. This inhibitory effect occurred despite concomitant inhibition of prostaglandin E synthesis. Lipoxygenase inhibitors also impeded cyclic AMP accumulation. Insulin and cyclic AMP release induced by glucagon were also blunted. These studies suggest the hypothesis that AA released in or near the beta cell is metabolized to lipoxygenase product(s) which have feedforward properties important to glucose- and glucagon-stimulated cyclic nucleotide accumulation and insulin release. Glucose-induced insulin (IRI) secretion is known to be associated with turnover of membrane phospholipids (1). More recently, it has been suggested that glucose can activate arachidonic acid uptake and release in the endocrine islet (2,3). This is of interest to studies of IR[ release since others have recently demonstrated that exogenous phospholipase mimics the effects of glucose on phospholipid turnover and insulin release (4). The link between such membrane effects and insulin release has remained elusive. A role for long chain, polyunsaturated fatty acids such as arachidonic acid, which might be released during such membrane perturbations, has only recently been studied. We have recently shown both in vivo (5) and in vitro (6) that a product of the cyclo-oxygenase pathway, prostaglandin E, inhibits glucose-induced insulin secretion; correspondingly, inhibitors of PGE synthesis such as sodium salicylate and ibuprofen potentiate glucoseinduced insulin release. On the other hand, arachidonic acid released from cell membrane phospholipids can also be metabolized via an alternative pathway, the lipoxygenase pathway, whose function in the endocrine islet is largely unexplored. Since many cell membrane-active compounds that are known to activate phospholipases (e.g., bradykinin, calcium ionophore A 23187, phorbol esters, and furosemide) stimulate, rather than inhibit, insulin release (7-9), we considered the possibility that the lipoxygenase cascade might represent a stimulatory pathway within the beta cell. Therefore, we studied the effect of combined lipoxygenase and cyclo-oxygenase inhibitors (since no selective lipoxygenase
0024-3205/83/080903-08503.00/0 Copyright (c) 1983 Pergamon Press Ltd.
904
Lipoxy~enase
and Insulin
Vol.
32, No. 8, ]983
inhibitor is available) upon glucose-induced insulin secretion using monolayer cultures of neonatal rat pancreatic islet ceils. We reasoned that if products of the Iipoxygenase pathway have either no e f f e c t on insulin release or an inhibitory e f f e c t similar to that seen with PCs, the e f f e c t of these combined inhibitors would be stimulatory. In contrast, if lipoxygenase products are stimulatory, the e f f e c t on insulin release might be null or even inhibitory. M a t e r i a l s and M e t h o d s
BW755c ( 3 - a m i n o - l - _ t r i f l u o r o m e t h y l - p h e n y ~ -2-pyrazoline) was generously provided by Dr. P.3. McHale of the Welicome Research l a b o r a t o r i e s (Kent, England). The antioxidant nordihydroguaiaretic acid (NDGA) was purchased from Sigma Chemical Co. (St. Louis, Missouri). Radioimmunoassay kits for the d e t e r m i n a t i o n of cyclic AMP released into the medium were purchased f r o m New England Nuclear (Boston, Mass.). The methods used for establishing monolayer cultures of neonatal rat pancreatic islet cells, enriched for endocrine ceils, have been previously described in detail (S). In brief, this method determines the cumulative release of insulin, prostaglandin E and cyclic nucleotides into the medium over a static 1-hr incubation period; thus, values reported represent t o t a l release and do not r e f l e c t specifically first or second phase insulin secretion. Control plates contained the same diluent (water for BW755c and glucagon, and ethanol for NDQA) as the experimental plates. Prostaglandin E (10), insulin (11), and cyclic AMP release (12) were measured by previously published radioimmunoassay procedures. The assay for cyclic AMP has a sensitivity ( m i n i m u m - d e t e c t a b l e - a m o u n t ) of less than 25 (routinely 10) f e m t o m o l e s / m l (2 standard deviations from the mean of replicate determinations of Bo=2.9%, B/B o at 25 femtomoles/ml=94%; n=10 determinations each). The radioimmunoassay is able to detect clearly changes of 15-20 f e m t o m o l e s / m l at cyclic AMP levels in culture medium of 150200 f e m t o m o l e s / m l . Neither of the drugs studied had significant effects upon recoveries of the relevant compounds in the radioimmunoassays or HPLC procedure. I d e n t i f i c a t i o n and measurement of 12-hydroxyeicosatetraenoic acid (12-HETE), as an index of lipoxygenase a c t i v i t y , were performed a f t e r incubating dispersed endocrinec ell-enriched pancreatic cells or intact rat islets w i t h 3H-arachidonic acid in the absence and presence of lipoxygenase inhibitors, and analyzing the medium, a f t e r e x t r a c t i o n into m e t h y l f o r m a t e , using reverse phase high performance liquid chromatography and s c i n t i l l a tion counting. For the ceil labelling procedure, pancreases from 30 rats were e n z y m a t i caily digested and mechanically dispersed, as for the monolayer cultures (6). However~ a f t e r the first overnight culture period, rather than being re-plated, the cells were harvested by collection of the medium containing non-attached cells and equally aliquoted into 3-4 test tubes. Ceils were labelled with t r i t i a t e d arachidonic acid (2 p C i / m l ) in a low glucose concentration (30 mg/dl) p u t a t i v e l y to m i n i m i z e phospholipase a c t i v i t y (3) and medium was collected a f t e r a 2-hour incubation period. Esterified arachidonate was then released f r o m membrane stores over the ensuing 30 minutes using either the ionophore A23187 or bradykinin in a high glucose concentration (300 mg/dl) to m a x i m i z e phospholipase a c t i v a t i o ~ (2,3). 12-HETE values were calculated as the sum of the CPM Muting with authentic ~H-12-HETE in the direct labelling and deacylation steps, and expressed as percent of t o t a l CPMs eluting from the H P / C column. These procedures have been described in detail (13). The primary column used for HPLC was a free f a t t y acid column (Waters Associates, M i l f o r d , MA). The solvent system ( a c e t o n i t r i l e / w a t e r / a c e t i c acid) was applied isocratieaUy for 50 minutes, and then increased fro, m 23% to 80% a c e t o n i t r i l e over a 30 min gradient. 12-HETE elutes at 67 to 69 minutes. -'H-12-HETE was identified by precise co-elution with authentic t r i t i a t e d 12-HETE when samples were analyzed as the free acid (in this and three additional HPLC systems) or w i t h 12-HETE methyl ester (in a fourth system) a f t e r m e t h y l a t i o n of samples using diazomethane in ether. Statistical analyses were by Newman-Keuls test for multiple comparisons and Student's t test for isolated comparisons.
Vol. 32, No. 8, 1983
Lipoxygenase and Insulin
905
Results With increasing concentrations of BW755c, there was a progressive decline in prostagtandin E synthesis that was maximal by 10 #M, and which was accompanied at this concentration by a modest inhibition of 12-HETE f o r m a t i o n (-b,1%). These effects were unaccompanied by any change in [RI secretion (Fig. 1). However, at higher concentrations (50-250 ]~M), 12-HETE was further inhibited (by 54% at 250 #M) whereas PGE synthesis did not fall further, thus providing a selective increase in inhibition of lipoxygenase a c t i v i t y at progressively higher concentrations of BW755c. A t these higher drug concentrations, there was a dose-dependent inhibition of glucose-stimulated (glucose = 300 mg/dt) insulin release (Fig. 1) to values indistinguishable (at 250 MM) from the low glucose (30 mg/dl) control (50 +- 2 vs. 52 + 2 p U / m l ; p=ns). Additional studies were then performed to study further the glucose-dependency of this inhibitory effect. BW755c did not reduce insulin release at a sub-stimulatory glucose concentration (30 mg/dI), had an intermediate inhibitory e f f e c t near the glucose threshold (100 mg/dl), and was m a x i m a l l y inhibitory e f f e c t at 300 mg/dl (Table 1). Thus BW755c t o t a l l y abrogated the glucose responsivity of these cells. Effects of this drug were unaccompanied by evidence of cell t o x i c i t y at any concentration studied, as assessed by phase contrast microscopy and trypan blue exclusion.
600
240 ~+SEM
500
4
n -- 2 - 4
PLATES
EACH
200
m
\E 400Q.
160 (/) I-
--//-5 ",
Z
I~~ 300-
t20
\ \\
£9
\
Q.
\
ULIN \
200.
-80
40
100-
"--%
.....
....... PGE
o
(6)#o,.5
~5
1'0
2"5
50 I00
250
o
BW755c, pM FIG. 1 Effects of various concentrations of BW755c on insulin release at 300 mg/dl of glucose (solid line) and PGE synthesis (dashed line). Inhibition of PQE was significant (p<.001) at 2 IJM or higher and inhibition of insulin release was significant from 100 IJM on (p<.005 or greater). Synthesis of 12-HETE was inhibited by 41% at 1O NM but by 54% at 250 1JM.
906
Lipoxygenase and Insulin
TABLE
Vol. 32, No. 8, 1983
I
EFFECT OF BW755c (250 pM) ON INSULIN RELEASE ( p U / m l ) AT VARIOUS GLUCOSE (Q~ mg/dl) C O N C E N T R A T I O N S Condition
Control
BW755c
G 30
, 33 +_ 2 " ~
35 _+ 0.4 :.- p<. 001
i G I00 p<.O01< G 300
' 4 8 _+ I - ' i I -
p<.O01 . . . . . .
~-59 _+ 4
33 + 2 L 36 + 2
p<.Ol Values are ~ _+ SEM; n=4 p l a t e s each
40 I I
14OO-
d
(.9 (3_
~
..... ~
140 -H
o
,oo
0 . 5 2 IO 20 SODIUMmg/dlSALICYLATE,
v I
y -Iz SEM
n=4 PLATES EACH 170
12OO
~50
I OOO
130
800
IlO
soo
9O
c
400
70
3 --
200
50
O
,
(0)
//
o) C r-
30 o. I
0.5
2
5
20
5'O
NORDIHYDROGUAIARETIC ACID, uM FIG. 2
Effects of various concentrations of nordihydroguaiaretic acid on insulin release (solid line) and POE synthesis (dashed line) at 300 mg/dl of glucose. Decrements in PGE synthesis were significant (I)<.005 or greater) aL 0.1 pM or higher. Changes in insulin release were significant (13<.02 or greater) at 2-50 pM. Insert shows effects of sodium salicylate for comparison (data from ref. 6). Synthesis of 12-HETE was inhibited by only 17% at 5 pM but by 89% at 50 pM.
Vol. 32, No. 8, 1983
Lipooxgenase
and Insulin
907
NDGA also was a potent inhibitor of cyclo-oxygenase at low concentrations (maximal effect seen at 5 pM), at which concentration 1 2 - H E T E formation was only minimally inhibited (-17%). This selective cyclo-oxygenase inhibition was accompanied by a progressive rise in insulin release (Fig. 2) which closely mimicked the effects of the prostaglandin synthesis inhibitor (6) sodium salicylate (Fig. 2, insert). At higher concentrations of N D G A , P G E synthesis did not decline further but inhibition of 1 2 - H E T E formation reached 8 9 % at 50 ]aM. Concomitantly there was a decline in insulin release to values significiantly less than the high glucose (300 mg/dl) control (control = 70 _+ 4; N D G A , 20 p M = 38 +_ 2 pU/ml; p<.01) and indistinguishable from the low glucose value (glucose, 30 mg/dl = 40 +_ 3 ]aU/ml). At 0-20 ]aM, N D G A was unaccompanied by evidence of cell toxicity, although at concentrations of ~. 50 ]aM, there was rounding and detachment of some cells assessed by phase contrast microscopy, an effect which appeared to preferentially affect fibroblasts. Simultaneous addition of maximally effective concentrations of BW755c and NDGA led to an inhibitory effect on insulin release not greater than that of either drug alone (data not shown). To examine the possibility that the inhibitory effects of lipoxygenase inhibitors were associated with changes of cyclic nucleoLide accumulation, measurements were made of cyclic AMP released into the medium, a sensitive index of the effects of drugs on synthesis and release of this nucleotide (12). Both lipoxygenase inhibitors caused a dosedependent inhibition of cyclic AMP accumulation (Fig. 3). The dose-response relationships showed that this inhibition of cyclic AMP could be readily dissociated from the coincident
180"
I } 160" D.
T. . . .
....
L
~ 140. <
\
I ..~-4 ~.axes eACN
\~
010
BW755C, pM 250
iz~u .-~-4 PLATES~ACH
2OO ¢L '150 O
\
\
T00
\ X
\-L
U 50
(0) "YO~I 05 I 5 NORDIHYOROGUAIARETIC ACID,
10
50
pM
FIG. 3 Effects of BW755c and NDGA on cyclic AMP accumulation in the medium. Decrements were significant as compared to glucose 300 mg/dl alone (p<.05 or greater) at 50 pM and 20 [aM, respectively. inhibition of prostaglandin synthesis (for example, compare Fig. 3 to Fig. i and 2) but seemed to parallel closely the inhibition of lipoxygenase and of insulin release.
908
Lipoxygenase and Insulin
Vol. 32, No. 8, 1983
Furthermore, BW755c also reduced cyclic AMP accumulat!on and insulin release induced by glueagon, 250 ng/ml (Fig. 4).
200
7
IOOO 900
+ SEM
n = 4 plates(INSULIN) 150
i
600
3 plates (cAMP)
I00
4OO 2
D r.,O z
INS I
50
~\\\,
200
i
o--
CONTROL (H20)
GLUCAGON, 250 ng/ml
[
BW 755c, 250 pM
BW 755c (~) GLUCAGON
o
FIG. 4 Effect of BW755c (250 pM) and glucagon (250 ng/ml) alone and in combination on insulin (wide, open or shaded bars) and cyclic AMP (narrow, cross-hatched bars) release. Differences between any two groups were statistically significant (p< .05 or greater). Discussion The e f f e c t s on insulin r e l e a s e of BW755c and N D G A (two s t r u c t u r a l l y - d i s s i m i l a r i n h i b i t o r s of a r a c h i d o n i c acid m e t a b o l i s m ) m u s t t a k e into a c c o u n t t h e d i f f e r e n t i a l sensitivities of cyclo-oxygenase and lipoyxgenase to these drugs, as we and others (14-16) have observed. BW755c and NDQA inhibit glucose-induced insulin secretion at concentrations of greater than 50 pM and 5 BM, respectively. These same concentrations inhibited lipoxygenase, as assessed by 12-HETE measurements. At lower concentrations, prostaglandin E synthesis was maximally and relatively selectively inhibited by NDGA, with 12-HETE formation being inhibited by less than 20%. A comparison to the effects of sodium salicylate and ibuprofen in the same experimental system (6,12) suggests that the stimulatory effect of NDGA at these lower concentrations was likely due to a selective inhibition of prostaglandin synthesis. Although some authors find that certain prostaglandins increase insulin release, especially at lower glucose concentrations (17), in the current preparation POE inhibits glucose-stimulated insulin release (6,12), and, conversely, inhibition of prostaglandin synthesis induces a progressive, curvilinear augmentation of glucose-induced insulin release (6). The partial inhibition of lipoxygenase (about 40%) by BW755c at low concentrations (10-100 pM) was submaxima] but was greater than that due to 5 wM NDGA and consequently was sufficient to block the stimulation of insulin release otherwise associated with cyclo-oxygenase inhibition. The
Vol. 32, No. 8, 1983
Lipoxygenase and Insulin
909
progressive reduction in insulin secretion below control values seen at higher concentra tions of either agent and associated with further inhibition of 12-HETE synthesis is compatible with the conclusion that the lipoxygenase pathway is stimulatory to insulin release. This formulation is supported by our observations that exogenously provided or endogenously released arachidonie acid potently stimulates insulin release (13,18). This effect can be further enhanced by cyelo-oxygenase inhibition and totally blocked by lipoxygenase inhibition. Thus our data suggest that arachidonic acid released in (or near) the beta cell can be metabolized to products which modulate insulin secretion by dual pathways--cyclo-oxygenase (inhibitory) and lipoxygenase (stimulatory). They do not, however, address the issue of whether insulin secretagogues augment release of lipoxygenase metabolites, or whether a tonic level of lipoxygenase activity plays a requisite permissive role for insulin release. In favor of the former possibility are the data of Evans and Clements (2) and of Laychock (3) that glucose activates arachidonic acid uptake and release in the endocrine islet, and other data (for review~ see 19) that glucose modifies lipoxygenase-mediated arachidonate metabolism in several cell systems. It is also possible that both drugs decreased insulin release by some unrecognized toxic effect on the cells (or via a primary effect to inhibit cyclic AMP accumulation; see below). While we cannot fully exclude these possibilities, it seems unlikely that two structurally-dissimilar drugs which both inhibit lipoxygenase should also share an undefined toxic effect, especially since morphologic and functional evidence of cell injury was sought and not found. We emphasize that 12-HETE measurements are used in the current study only as a specific marker for the lipoxygenase pathway; the arachidonate metabolite(s) responsible for the stimulatory effect remain to be elucidated. In fact, exogenous 12-HETE fails to alter glucose-stimulated insulin release at concentrations from 0.5-2000 ng/ml (S. Metz, R. Murphy, W. Fujimoto, R.P. Robertson, manuscript in preparation). In the absence of identification and measurement of the active metabolite of arachidonic acid over a full range of drug concentrations and under identical conditions as those used to monitor insulin release, perfect correlations between lipoxygenase activation (as assessed by 12HETE synthesis) and insulin release are not to be expected. However, our data (and those of others, 20) using 12-HETE as a marker, show that a lipoxygenase exists in the endocrine islet; that it is potently inhibited by concentrations of BW755c and NDGA which also inhibit insulin release; and that at lower drug concentrations, the effect of lipoxygenase inhibition becomes less evident and is counterbalanced (e.g, BW755c at 10-50 ]JM) or superseded (e.g., NDQA at 5 pM) by the effects of relatively selective cyclo-oxygenase inhibition. Our results suggesting that lipoxygenase inhibition can prevent glucose-induced insulin release may be relevant to the suggestion (2,3) that glucose activates a phospholipase in the pancreatic islet, an effect which could increase the availability of arachidonate. Taken in concert with the observation that exogenously provided phospholipase can mimic the effects of glucose on phospholipid turnover and insulin release (4), these observations permit the hypothesis that lipoxygenase-mediated metabolism of arachidonic acid might be an intermediate step in stimulus-secretion coupling during glucose-induced (or glucagon-induced) insulin release. This conclusion is compatible with reports that compounds which inhibit phospholipase(s), such as trifluoperazine (21) and mepacrine (21), also inhibit glucose-stimulated insulin release (9,22). Trifluoperazine has been shown to act as a combined cyclo-oxygenase/lipoxygenase inhibitor by virtue of inhibiting a calmodulin-dependent phospholipase (21,23); conversely, the activation of phospholipase (23) and of adenylate cyclase (24) by calcium and calmodulin raises the possibility that glucose-induced alterations in calcium flux may, in turn, a c t i v a t e oxygenated arachidonate metabolites as "second messengers." Indeed, the current data also demonstrate that an adequate degree of lipoxygenase inhibition impedes cyclic AMP production by cultured islet cells; therefore, cyclic nucleotides could mediate part of the stimulatory effect of lipoxygenase products upon insulin secretion. However, such a causal relationship is not established by the current studies and we cannot totally exclude
910
Lipoxygenase and Insulin
Vol. 32, No. 8, 1983
a non-specific impairment of cyclic AMP accumulation by NDQA and BW755c. The possibility of a specific relationship between lipoxygenase products and cyclic nucleotide formation is supported by recent studies of Levasseur and co-workers (25) who have provided evidence that a lipoxygenase pathway in the thyroid may mediate a part of stimulatory actions of TSH on cyclic AMP accumulation. Others have also found evidence that lipoxygenase products augment cyclic AMP accumulation (26). Products of the lipoxygenase pathway are also known to potently augment cyclic QMP formation in certain systems (27). Whatever the mechanism, the current studies suggest that oxygenation products of arachidonic acid may provide both stimulatory and inhibitory influences on glucose-induced insulin release, depending upon whether the lipoxygenase or the cyclo-oxygenase pathway is dominant in a given circumstance. Acknowledgments The authors gratefully acknowledge the excellent technical assistance of Ms. 3eanette Teague, Ms. Lytian Fuller, and Mr. David Federighi. This work was supported by the Veterans Administration, a grant from the Upjohn Company, and grants AM-15312 and AM-17847 from the National Institutes of Health. References 1. 2. 3. 4.
R.S. CLEMENTS Jr. and W.B. RHOTEN, J. Clin. Invest. 57 684-691 (1976). M.H. EVANS and R.S. CLEMENTS Jr., Clin. Res. 29 404a---~1981). S.G. LAYCHOCK, Fed. Proc. 40 458a (1981). K. TANIGAWA, H. KUZUYA, H. IMURA, H. TANIGUCHI, S. BABA, Y. T A K A I and Y. MISHIZUKI, FEBS Letters. 38 183-186 (1982). 5. R.P. ROBERTSON and M. CHEN, J. Clin. Invest. 60 747-753 (1977). 6. S.A. METZ, R.P. ROBERTSON and W. FU3IMOTO, Diabetes. 30 551-557 (1981). 7. 3.E. VANCE and S.C. BRAGG, Diabetes (Suppl). 18 326-327 (1969). 8. W..]. MALAISSE, A. SENER, A. HERCHUELZ, A.R. CARPINELLI, P. POLOCZEK, 3. WINAND and M. CASTAGNA, Cancer Res. 40 3827-3831 (1980). 9. R. LANDGRAF and M.M.C. LANDQRAF-LEURS, Prostaglandins. 17 599-613 (1979). 10. R.P. ROBERTSON. Am. 3. Physiol. 228 68-70 (1975). 11. D.S. ZAHARKO and L.V. BECK, Diabetes. 17 444-457 (1968). 12. S. METZ, W. FUSIMOTO and R.P. ROBERTSON. Metabolism, in press (1982). 13. S. METZ, S. LITTLE, W. FUJIMOTO and R.P. ROBERTSON, Adv. Prost. Thromb. Res., in press (1982). 14. M.I. SIEGEL, R.T. MCCONNELL, R.W. BONSER and P. CUATRECASAS, Prostaglandins. 21 123-132 (1981). 15. C.L. ARMOUR, 3.M. HUES, 3.P. SEALE and D.M. TEMPLE, Eur. 3. Pharmacol. 72 93-96 (1981). 16. M.A. BRAY, A.W. FORD-HUTCHINSON, M.E. SHIPLEY and M.J.H. SMITH, Br. 3. Pharmacol. 71 507-512 (1980). 17. T.P. HEANEY and R.G. LARKINS, Diabetes 30 824-828 (1981). 18. S. METZ, S. LITTLE, W. FU3IMOTO and R.P. ROBERTSON, Clin. Res. 30 63a (1982). 19. S.A. METZ, Prostagland. Med. 7 581-589 (1981). 20. K.I. K E L L Y and S.G. L A Y C H C ~ C K , Prostaglandins. 2__l_759-769 t (1981). 21. R . W . W A L E N G A , E.E. Q P A S and M.B. FEINSTEIN, 3. Biol. Chem. 256 12523-12528 (1981). 22. Y. KRAUSZ, C.B. WOLLHEIM, E. SIEGEL and G.W.G. SHARP, J. Clin. Invest. 66 603-607 (1980). 23. P.A. C R A V E N , R.K. S T U D E R and F.R. DeRUBERTIS, .]. Clin. Invest. 68 722-732 (i98i). 2/4. G.W.G. S H A R P , D.E. W E I D E N K E L L E R , D. KAELIN, E.G. SIEGEL and C.B. W O L L H E I M , Diabetes. 29 7/4-77 (1980). 25. S. L E V A S S E U R , F.F. SUN, Y. F R I E D M A N , and Q. B U R K E , Prostaglandins 22 663673 (1981). 26. H.-E. C L A E S S O N , FEBS Letters 139 305-308 (1982) 27. H. H I D A K A and T. A S A N O , Proc. Nat'l. Acad. Sol. USA. 7/4 3657-3661 (1977).