Quantitative Effects of Glucose, Sulfonylureas, Salicylate, and Indole-3-Acetic Acid on the Secretion of Insulin Activity Into Pancreatic Venous Blood

Home , Metahexamide

QUANTITATIVE EFFECTS OF GLUCOSE, SULFONYLUREAS, SALICYLATE, AND INDOLE-3-ACETIC ACID ON THE SECRETION OF INSULIN ACTIVITY INTO PANCREATIC VENOUS BLOOD

Holbrooke S. Seltzer

J Clin Invest. 1962;41(2):289-300. https://doi.org/10.1172/JCI104482.

Research Article

Find the latest version: https://jci.me/104482/pdf Journal of Clinical Investigation Vol. 41, No. 2, 1962 QUANTITATIVE EFFECTS OF GLUCOSE, SULFONYLUREAS, SALICYLATE, AND INDOLE-3-ACETIC ACID ON THE SECRETION OF INSULIN ACTIVITY INTO PANCREATIC VENOUS BLOOD * By HOLBROOKE S. SELTZER (From the Metabolic Section, Veterans Administration Hospital, and the Department of In- ternal Medicine, University of Texas Southwestern Medical School, Dallas, Texas) (Submitted for publication July 5, 1961; accepted October 12, 1961)

It is now clear that the predominant metabolic ever, it shortly became evident that the normal hyper- effect of sulfonylurea compounds is to release in- glycemic response to anesthetic and surgical trauma caused premature discharge of insulin, which masked sulin from pancreatic beta cells. In addition, as- insulin-releasing activity of all agents except glucose says by Yalow, Black, Villazon and Berson (1) itself (Table I). Accordingly, animals were starved and of insulin concentrations in peripheral blood indi- phlorhizinized for 3 days preoperatively, to deplete hepatic cate that the secretory potency of sulfonylureas is glycogen stores maximally and lower peripheral blood substantially weaker than that of glucose. More glucose to hypoglycemic levels. In dogs weighing 13 quantitative comparison of insulin-releasing ac- kg or less initially, 5 ml of a 40 per cent solution of phlorhizin in propylene glycol was injected subcutane- tivities of individual substances, however, re- ously approximately 64, 48, 40, 24, and 16 hours before quires knowledge of both the concentration of in- operation; the total phlorhizin dosage was 10 g. Dogs sulin in pancreatic venous blood, and the latter's weighing above 13 kg were given 6 ml per dose, and rate of flow. A suitable closed circuit enabling total dosage was 12 g. All animals also received two these measurements was devised by Metz (2), injections of 1: 500 epinephrine-in-oil, 1 ml intramuscu- larly at 40 and 16 hours before surgery. Animals were who determined total insulin outflow from the anesthetized with pentobarbital sodium (Nembutal), 25 pancreaticoduodenal vein draining the right pan- mg per kg intravenously, and laparotomized by midline creatic limb in dogs during glucose infusion into incision. The posterior surface of the right pancreatic the femoral vein. The latter technic was employed limb was exposed by delivering the free loop of the du- in the present study to determine total secretion of odenum into the wound and reflecting it to the left. The cranial pancreaticoduodenal vein, draining all of the insulin activity from a segment of dog pancreas for right limb of the pancreas except its distal tip, was usu- 30 minutes after glucose or a hypoglycemia-induc- ally visible and accessible with variable ease beneath the ing agent was injected into a peripheral vein. surface of the gland. The vessel was cannulated in Although glucose and three sulfonylurea com- retrograde direction (upstream against blood flow) pounds all promptly initiated release of insulin, about 2 cm distal to its confluence with the portal vein, using a plastic catheter of large bore (PE 240, OD 0.095 glucose was by far the most potent insulogenic inch) to allow free flow of effluent. The vein was ligated substance. In contrast, neither salicylate nor proximal to the point of cannulation to prevent back- indole-3-acetic acid altered the rate of insulin se- flow of blood from the portal vein. Immediately before cretion. cannulation, the dog was heparinized with 1,500 U per kg of body weight. After cannulation, the entire venous METHODS outflow was collected uninterruptedly in a baseline (con- 10 and trol) aliquot during the first 10 minutes; and then in Mongrel dogs of either sex, weighing between three postinjection collections, each of 10 minutes' dura- 15 kg, were used in all experiments. During a quaran- tion. Total volume and hematocrit were determined on tine of 14 animals were stabi- period days, metabolically all aliquots. In some animals an additional femoral ve- lized on a ration of meat and Purina daily laboratory nous aliquot was obtained before injection, and 10 min- which 60 g 20 g chow, provided approximately protein, utes thereafter, for estimation of insulin activity in fat and 90 g carbohydrate. In early experiments, intact peripheral plasma. At the end of the baseline collection, dogs were used after a 24-hour fasting period. How- the test substance was inj ected into a femoral vein. * This work was supported in part by Grant A-4708 Eight dogs received 15 g of glucose (30 ml of 50 per from the United States Public Health Service, Bethesda, cent solution) in 5 minutes. The sodium salts of tolbuta- Md. This paper was presented in part at the Southern mide (50 mg per kg to 6 dogs), chlorpropamide (50 mg Society for Clinical Research, New Orleans, La., Janu- per kg to 5 dogs), metahexamide (10 mg per kg to 5 ary 23, 1960. dogs), salicylic acid (50 mg per kg to 4 dogs), and in- 289 290 HOLBROOKE S. SELTZER

TABLE I Responsiveness of intact versus phlorhizinized dogs to insulin-releasing stimuli

Pancreatic venous plasma insulin activity No. of Preinjection Preinjection Postinjection Substance dogs blood glucose -10 to 0 min 0 to 20 min p Value mg/lOO ml ,uU/ml pU/ml Intact dogs Glucose 4 105 i 9* 100 i 9 678 i 161 <0.005 Tolbutamide 4 127 4 9 94 i 13 118 ± 10 <0.3 Phlorhizinized dogs Glucose 8 48 ± 4 13 4 8 615 i 103 <0.001 Tolbutamide 6 53 4 5 12 4t 5 73 i 6 <0.001 * Mean i SEM. dole-3-acetic acid 1 (50 mg per kg to 4 dogs), were dis- Warburg apparatus for 90 minutes at 37.2° C under an solved in 10 ml of distilled water and injected within 2 atmosphere of 95 per cent 02 and 5 per cent CO2. minutes. As control substances, 4 dogs were given 50 Residual glucose was determined by the iodometric mg of sodium sulfadiazine per kg, and 3 others received method of King (5). Diaphragms were dried for 2 10 ml of 0.85 per cent sodium chloride solution. Oxa- hours at 1100 C, cooled, and weighed. Glucose uptake lated samples for blood glucose were obtained from the was expressed as micrograms per 10 mg of dried dia- femoral vein during the baseline period and every 15 phragm. "Net glucose uptake" signified total glucose minutes for 1 hour after administration of test material. uptake from a sample minus baseline glucose uptake by The animal was then killed with pentobarbital, and the diaphragm incubated in GBSS alone. Technical limi- segment of pancreas which had been drained verified by tations prevented simultaneous performance of both a retrograde injection of 20 ml of dilute methylene blue dose-response curve and all samples from a single ex- solution into the cannula. periment. Accordingly, at intervals of 2 to 4 weeks a Blood glucose was determined in duplicate by the dose-response curve was done, by determining net glu- Somogyi-Nelson method (3). Aliquots of pancreatico- cose uptake from triplicate 2-ml aliquots of GBSS con- duodenal and femoral venous blood were centrifuged im- taining 10, 100, and 1,000 AU of insulin per ml, re- mediately, and the separated plasma was refrigerated at spectively. The mean of 24 individual dose-response 70 C. All samples were assayed within a week of pro- curves yielded the standard dose-response curve (Figure curement, and all specimens from an animal were assayed 1) from which experimental values for net glucose up- simultaneously. Plasma insulin activity was assayed by take were converted into microunits of "effective insulin determining the glucose uptake from 2 ml of undiluted concentration" per milliliter of plasma. Conversion val- plasma by isolated rat hemidiaphragm, adhering strictly ues between 1 and 10 AU per ml of plasma were esti- to the technic described by Vallance-Owen and Hurlock mated by extrapolation. The few glucose uptakes ex- (4). Male Wistar rats, weighing between 90 and 130 g, ceeding the equivalent for an insulin concentration of were used after food was withheld for 18 to 24 hours. 1,000 /AU per ml of GBSS (Table II) were included Glucose concentration of plasma or buffered salt solu- in calculations as the latter value. tion (GBSS) was 300 mg per 100 ml. When glucose Pancreaticoduodenal venous plasma flow (PP flow) was the test substance, and its concentration in post- was derived by multiplying the rate of pancreaticodu- injection plasmas exceeded 300 mg per 100 ml, the glu- odenal venous blood flow (PB flow) by the "plasma- cose level of GBSS and all plasma samples was brought crit": up to that in the most hyperglycemic plasma. Samples of plasma, GBSS alone, or GBSS containing 10, 100, or PB flow (ml/min) X (100 - hematocrit) 1,000 uU of insulin2 per ml GBSS, were incubated in a 100 = PP flow (ml/min) [1] 1Indole-3-acetic acid was obtained from Nutritional Biochemicals Corp., Cleveland, Ohio. The sodium salt The concentration of insulin activity in pancreaticodu- was prepared by dissolving the acid in 5 ml of warm odenal venous plasma (PIA), multiplied by pancreatico- 0.5 N NaOH, and neutralizing with dilute HCl to pH 7 duodenal venous plasma flow, yielded outflow rate of on pHydrion paper. plasma insulin activity (PIA flow): 2 Five times recrystallized zinc insulin was kindly sup- plied by Dr. W. R. Kirtley, Eli Lilly and Company, PIA (MU/ml) X PP flow (ml/min) Indianapolis, Ind. = PIA flow (MU/min) [2] INSULINOGENIC EFFECTS OF GLUCOSE AND HYPOGLYCEMIA-INDUCING SUBSTANCES 291

0CO3 0 06~ .. O- 0

_ ! NO 0 m NO v. 00 v t)in0 0 00 "000 V CM CV tC o Cddd C 0 v

14 u u in U) 00" V u; _ t)o) U) 0 0 v 00 C- 000 N e u 0% 0 V -C4 00 M Us > ._ a~~~~~~~N _

. ermx -o e oo -4 r 0 en aq 0 in *'0 4en _l 00 m 0 0 o o N uo QS _ v 0eu)+e5 C4 VI) o 0 V 0s0 sU oo% o) o *00 0 CU 00 %Uco c in oo r 00 Nro 00 aV 00 No r\ C4 _ csi _4t0'C4 6 N6 cl cl C4c

O _ S 0t cU CM - C

U. 0d C' O :N C LnX \mm Y) m' in m-O~ t- \0 00- u4 00N 'o\ 0e - o NO

Cd _ v .E _ +' _Co C d a" V V 0 0%0%t-.U00ccMg 00 w - m00 -0 C6 V4 e60 i el; C; N6 Il \0 -sO'Oi C' -tt00V)) 04t CMO 00 r-00 00 0 0c cM

O VN 0 uz NO~~ ~ O 0U0O-I 0 '100' U0 0 ".k"r00 C_0 u O m 0- O.O.-0.0C4 C4 atoy>

) _ tMm r ) - °O O 0 s: U) o o _ r + e v mwNV 0% V _ e c t 7o -. 4' -O_U)N+ W u4 O - W)\0 UpO O 0 00 inU .0 .a Vo 00 V° o o o rU o * '-( * O o~ .E n q V V

U) ._~~ ~ ~ ~ ~ ~ *00; go CU rE< -4 -4 -4 4r-4 C -¢ °° Q-0oO 0e 0 0--nCMV ooo o r V 0 U)0e O t r Oo -~ ~ ~ o -X e X X N CS X o N N o so-UC tn o . 0 dd o eU 6 0 C"S Coo .E o O = in N 0 -~ °b0 CoMt"-onMto0 _ CU 0 . t M O N Ooo-in . 00o o 4 N m0)_4. LnCo 'O to 0 o 0 V ? O N ) ON v E d ...... Cd ,C0) ) VZ-I_ -0 MO

0 U) _) U) bs 'osoin 'D t- _ C4 0 0C 0 c0 0 U 0 CM0% _ uo V -4~~~~~~0MN- nt

d - o'm0Mto MU) O ) o 'o o 4 to -4'o_0o o0 C4 0 t%Comto_ Vr4 o -4 U4% 0.. ~ ~ ~ ~ ) .o in C r C to c;~~o-t - 0 0_ Ne-4 V N-4) "WbV)°

______dd _ _ _lN_ CMV O0 CMs0 O - 0o)n- V 0 0 t 00 U)~~~~~~~%WCM00)). 8 8CC

r )rU)~ C" Ceq O00 o O 0 )0 W V 0 0o , \

bO 0 ~ 292 HOLBROOKE S. SELTZER

oo oo U) d 6 6 00C;

co o m U) N 00 0 cooNU)Umm0 eS 00-4 NN -U)M0- N V 0 Cd 00 ._ I- to 0N0V U> N U) .2C .0

CU) C \, LeN vCNV N N V _r N V

0 t~~~~ OCO No Cd 00 0 kn o Noo 0) 4 V o) * m7 m4m o - V 0 . v

N Xa NV o 0 NCCN 0 - 0- v >~~~~ Q~0000 m 0, - eo* " co V: cli~~c CUd .) CNOW 4) )4 d CO4 0 V E r b N 00 00 3) o s o o _ o )0 - N 0 o- v oSo :CU)U)0 4- 0 CLX U) U) °0 0 U)C- - U 4 C6 C4 v i C;u Coo 0V N CO0 ta) oo r o V) . 0) v m C cd .t W ocN o't u u ) ) O0 m0+ E a) V~~~~o N o NVWOe ''0o X U)No -X-~ ~ ~ ~o ~ C4-- X o~~~~~~N cdO as oo X o CU -

0~~~~~~'

O o 0 0 Ct- '- '0o- V 0 v 0 .) - - 0% '0 --v CO _ coCOo _4 _4 t- N O; 0 < V 0 Q £

W)) -4 W) U) 0 '0 .E U)- - eN

o0 .14 0%

- I - -0 - I - V O%~ CU 0

C) N

.e) Cli a) . Q4 0 C'- o N 00 0 M C4 00 -4 W) Co N 00 0 08 U)NC U U) U) U) N v 0 O V m_rVoC4

0% 4 4-s t U * - O '0 r V C- C14 1-4 to-4 V

'O \0 0 o" 00 -.0 N rC N N t .-Coo0 o'-. 10 NOOdW)

a b0 0 d ~ 0 inrUNo coaCCa .4 Ncd m i o - 0 c N ) A a I A Q X :~ INSULINOGENIC EFFECTS OF GLUCOSE AND HYPOGLYCEMIA-INDUCING SUBSTANCES 293 (MEAN ± S. E. M.) and essentially the same as recent ones obtained by Gundersen (8).

B. Secretory rates of insulin activity in pancreaticoduodenal venous plasma after administration of test substances (Table II, Figures 2 and 3) 1. After glucose. At the end of the 5-minute injection of 15 g of glucose, peripheral blood glucose level (mean ± SEM) was 417 ± 21 mg per 100 ml, and 25 minutes later it had fallen to 236 + 12 mg per 100 ml. Glucose evoked a prompt and prodigious rise in pancreatic outflow of insulin CUBE ROOT OF INSULIN CONCENTRATION IMICRO-UNITO PER ML GLUCOSE-BUFFER$ activity, with only a slight decrease during the 30 FIG. 1. STANDARD DOSE-RESPONSE CURVE. Experimental minutes of observation. The increased secretion values for net glucose uptake from plasma were converted was due to greatly enhanced titers of insulin into "microunits of effective insulin concentration per activity in pancreatic venous blood, since neither milliliter of plasma" from this curve. Points on the pancreatic blood flow nor hematocrit changed sig- abscissa are cube roots of 10, 100, and 1,000 ,uU insulin nificantly. Respective concentrations of effective per ml of glucose-buffered salt solution (GBSS), re- insulin activity were 13 ± 8 MU per ml of plasma, spectively. Conversion values below 10 AU per ml were estimated from the extrapolated segment of the curve. baseline; 634 ± 134 /U per ml between 0 and 10 minutes; 596 + 92 uU per ml between 10 and RESULTS 20 minutes; and 748 ± 124 /%U per ml between 20 and 30 minutes. Derived secretory rates of in- A. Dose-response curve (Figure 1) sulin activity (Figure 2) were 48+ 29 pU per + All 24 individual dose-response curves, and minute, baseline; 3,561 813 uU per minute be- hence the mean curve derived from them, were tween 0 and 10 minutes; 2,983 ± 686 tzU per curvilinear when plotted on the usual semiloga- minute between 10 and 20 minutes; and 2,295 ± rithmic coordinates. However, as Vallance-Owen, 346 MU per minute between 20 and 30 minutes. Hurlock and Please (6) originally demonstrated and validated statistically, and as Wright (7) and MHEANISEM Gundersen (8) subsequently confirmed, the dose- >1_ GLUCOSE TOLBUTAMIDE CHLORPROPAMIDE METAHEXAMIOE 15 GM 50 MG/KG SO KG/KG 10 MG/KG response curve formed virtually a straight line *-5000 when net glucose uptake was plotted linearly on the ordinate, and insulin concentration as cube root on the abscissa. Respective mean net glu- U- cose uptakes for insulin concentrations of 10, 100, 4 and 1,000 uU per ml of GBSS were 63 + 6 I500 (SEM) Mg per 10 mg diaphragm; 151 ±8 ug per 10 mg diaphragm; and 313 + 10 ug per 10 v0 mg diaphragm. On the extrapolated portion of the curve, net glucose uptake of 20 ug per 10 mg _J M O 0 10 20 so 0 10 20 30 0 10 20 30 0 10 20 30 diaphragm was equivalent to 1 MU per ml GBSS. U. <.001 <.001 <.WH <.01 <.01 <.05 <.0S <.1 <.0S5 .5 <.1 <.2 The slope of this dose-response curve (31.9 fug of CONTINUOUS 10-MINUTE COLLECTION PERIODS glucose uptake per cube root of insulin level in FIG. 2. INSULIN ACTIVITY IN PANCREATIC VENOUS BLOOD microunits per milliliter) was considerably steeper AFTER GLUCOSE AND SULFONYLUREAs. Glucose and sul- of insulin than those of Vallance-Owen and co-workers and fonylurea compounds caused prompt release activity into pancreatic venous blood. Beta-cell stimula- Wright (23.1 and 23.7 Mg of glucose uptake per tion by glucose, however, was greater and more sus- cube root of insulin concentration, respectively), tained than by any sulfonylurea. 294 HOLBROOKE S. SELTZER (MEAN±SEM1 208 uU per minute (p < 0.05) within 10 minutes, >_ SALICYLATE IND.-3-ACET. AC. SULFADIAZINE SALINE _.- 50 MG/KG 50 MG/KG 50 MG/KG 10 ML then subsided to 275 ± 94 F/U per minute (p < 0.1) in the 10 to 20 minute interval, and remained at 333 ± 110 fxU per minute (p < 0.2) during the last period. All increases of insulin flow rates above baseline levels were due to enhanced con- *- 5000 centrations of insulin activity in pancreaticoduodenal venous plasma, and not to increased I. 500 .4 plasma flow rates. 3. After salicylate and indole-3-acetic acid. Neither sodium salicylate nor sodium indole-3- acetate stimulated release of insulin activity (Fig- J 0 10 220 3 10 20 01 0 20 30 0 10 20 50 U. na <.4 <.8 <.3 <,7 <.9 <.a <.4 <.6 <.o <.6 <.S <.S ure 3). Interval flow rates in animals given 10-MINUTE COLLECTION PERIODS CONTINUOUS salicylate were 19 + 10 p&U per minute, baseline; FIG. 3. INSULIN ACTIVITY IN PANCREATIC VENOUS 121 ± 86 MAU per minute between 0 and 10 min- BLOOD AFTER HYPOGLYCEMIA-INDUCING AND CONTROL SUB- -+ STANCES. The two nonsulfonylurea, hypoglycemia-induc- utes; 25 16 uU per minute between 10 and 20 ing substances, salicylate and indole-3-acetic acid, were as minutes; and 106 ± 28 MU per minute between devoid of insulin-releasing activity as the control sub- 20 and 30 minutes. Mean plasma salicylate con- stances, sulfadiazine and saline. centration 15 minutes after injection was 28 ± 3 mg per 100 ml. For indole-3-acetic acid, baseline flow rate was 76 ± 40 MuU per minute; and All of these increases above baseline exceeded ± ± p < 0.001 level of significance. the three postinjection values were 54 21, 89 2. After sulfonylurea compounds. The effects 80, and 96 ± 71 j1XU per minute, respectively. of tolbutamide, chlorpropamide, and metahexa- 4. After sulfadiazine and saline. Similarly, ad- mide on release of insulin activity resembled that ministration of sodium sulfadiazine or sodium chlo- of glucose qualitatively, but not quantitatively. ride as control substances elicited no change in As a group, they sharply augmented insulin ac- preinjection outflow rates of insulin activity (Fig- tivity in pancreatic venous blood during the first ure 3). 10 minutes, although maximal secretory rates in individual experiments occurred in any of the C. Comparison of insulagenic potencies three collection intervals. The magnitude of response to these agents, however, was only a frac- The last column in Table II shows the mean in- tion of that following glucose (Figure 2). Se- crease of insulin outflow rate above baseline value cretion rates of insulin activity in animals given during the entire 30-minute interval. After glu- tolbutamide were 40 ± 17 /LU per minute, base- cose administration, the average increase in pan- line; rising to 379 + 116 utU per minute (p < creatic venous insulin activity was 2,968 ± 567 0.01) in the first 10 minutes; reaching a maximum ,uU per minute (100.0 +19.1 per cent). The of 433 ± 124 ,AU per minute (p < 0.01) between average increase after tolbutamide was 294 + 70 10 and 20 minutes; and dropping to 188 ± 54 pU MAU per minute (9.9 + 2.4 per cent).. After chlor- per minute (p < 0.05) between 20 and 30 minutes. propamide the mean increase was 553 ± 188 uU Dogs injected with chlorpropamide had- corre- per minute (18.6 + 6.4 per cent), and after meta- sponding values of 40 ±+9 ,uU per minute, baseline; hexamide the corresponding mean enhancement 645 ± 254 1.U per minute (p < 0.05) between 0 was 347 ± 87 puU per minute (11.7 ± 2.9 per and 10 minutes; dropping slightly to 428 ± 175 MuU cent). All of these increments were significant at per minute (p < 0.1) between 10 and 20 minutes; p < 0.02 or better. In contrast, after salicylate and rising again to 706 ± 270 uU per minute the increase above baseline value was only 57 ± (p < 0.05) during the last period. Insulin out- 42 uU per minute (1.9 ± 1.4 per cent), and after put after metahexamide rose from baseline of 36 + indole-3-acetic acid it was 52 ± 38 uU per min- 12 ,uU per minute to a maximal value of 559 i ute (1.8 ± 1.3 per cent). INSULINOGENIC EFFECTS OF GLUCOSE AND HYPOGLYCEMIA-INDUCING SUBSTANCES 295

TABLE III Insulin activity in peripheral versus pancreatic venous plasma of phlorhizinized dogs (Mg net glucose uptake/JO mg diaphragm)

Femoral vein Pancreaticoduodenal vein Substance Baseline 10 min p Value Baseline 0-20 min p Value Glucose 5 A= 16* 22 :1 13 <0.3 55 i 12 293 :1 40 <0.001 Tolbutamide 8 4t 28 22 ± 15 <0.7 59 i 13 147 :1 8 <0.001 Salicylate -12 ±- 6 8 4- 18 <0.4 38 41 16 57 i 18 <0.6 Indole-3-acetic acid 24 4t 6 -8 ± 10 <0.2 64 ± 15 52 i 21 <0.6 Saline 15 i 3 27 i 21 <0.5 50 4± 7 36 i 4 <0.8 * Mean 4t SEM.

D. Changes in pancreaticodiuodenal veCnous plasma F. In vitro effects on glucose uptake of test sub- flow (Table II) stances added to plasma (Table IV) Except in the glucose group, in which the Sulfonylureas, added to baseline plasma in larger volume of injected solution caused signifi- clinically effective concentrations and incubated cant rise in pancreatic venous blood flow during in vitro, had no effect on net glucose uptake by the first collection period, no important increases rat diaphragm. In contrast, when salicylate was in blood flow were observed. Toward the end of added to baseline plasma, net glucose uptake did the 40 minutes of continuous hemorrhage, blood increase from the control value of 69 11 to 119 flow declined progressively and was accompanied 13 /ug per 10 mg diaphragm (p < 0.005). by slight hemodilution. Only in the tolbutamide- treated animals was the change significant (base- G. Changes in peripheral blood glucose (Table V) line of 3.14 0.18 ml per minute fell to 2.48 0.17 ml per minute, p < 0.005). None of the test substances caused depression of femoral venous blood glucose during 60 min- E. Insulin activity in femoral venous plasma after utes after injection. In particular, mean baseline administration of test substances (Table III) blood glucose of 55 + 3 mg per 100 ml for all three sulfonylurea compounds actually rose Baseline concentrations of insulin activity in slightly to 57 3 mg per 100 ml at 30 minutes, were lower than peripheral plasma consistently and 69 5 mg per 100 ml at 60 minutes. respective baseline titers in pancreaticoduodenal venous plasma. Ten minutes after administration of glucose, tolbutamide, salicylate, indole- DISCUSSION 3-acetic acid or saline, no enhancement of insulin Vallance-Owen correctly emphasizes that his activity was detectable in femoral venous blood bioassay technic does not measure total insulin from these phlorhizinized dogs. content of plasma, but estimates instead, "plasma-

TABLE IV Effect of hypoglycemia-inducing substances, added to plasma in vitro, on glucose uptake by rat hemidiaphragmn

Glucose uptake Baseline Plasma Substances Conc. plasma +substance p Value mg/100 ml pg/1O mg diaphragm Tolbutamide 25-50 55 4± 22t 60 i 11 <0.9 Chlorpropamide 50 15 i 15 8 i 11 <0.8 Metahexamide 50 15 4t 20 -2 i 13 <0.4 Salicylate 30-50 69 + 11 119 ± 13 <0.005 I[iidole-3-acetic acid 50 59 i 14 57 4t 22 <0.9 Stilfadiaziiie 50 7(0 i 23 83 i 23 <0.7 * As sodium salt. t Mean ± SEM. 296 HOLBROOKE S. SELTZER

TABLE V Femoral venous blood glucose concentrations after administration of hypoglycemic agents

Minutes after injection No. of Substance dogs Preinjection 15 30 45 60

mg per 100 ml Tolbutamide 6 53 5* 50 i 7 54 4 6 63 ± 6 73 ± 6 Chlorpropamide 5 54 5 51 i 4 51 ± 5 52 ±5 56 ± 8 Metahexamide 5 61 ±t 3 65 ±i 5 67 ± 3 72 3 76 i 3 Salicylate 4 43 ± 4 42 ± 4 50 ± 6 52 i 8 57 7 Indole-3-acetic acid 4 45 +t 3 46 i 3 42 i 3 44 ± 4 49 i 5 Sulfadiazine 4 56 ±t 6 62 ± 5 65 ± 8 66 ± 7 67 9 Saline 3 43 ± 6 40 ±t 6 40 ± 7 38 ±t 6 41 ± 6 * Mean ±1 SEM. insulin activity, or the effective plasma-insulin con- specific increase of glucose uptake by rat dia- centration, i.e., the sum of insulin and its syner- phragm. gists, if any, on the one hand and its antagonists At the same time, experimental values for on the other" (9). Nevertheless, in the present concentrations and flow rates of insulin activity study, the conclusion seems reasonable that the denoted only relative quantitation of insulin se- flow rates of insulin activity in pancreaticodu- cretion. Since the cranial pancreaticoduodenal odenal venous effluent truly reflected the rate of vein in the dog drains a segment of gland con- insulin secretion by beta cells. The isolated rat taining about 30 per cent of total islet cell volume diaphragm is sufficiently specific for assaying in- (14), only a fraction of total insulin secretion was sulin activity that it does not detect "insulin ac- measured. A more critical dissuader from mis- tivity" in blood from totally depancreatized animals interpreting relative values as absolute, however, (10, 11), in contrast to the more sensitive but ap- is the fact that reported "normal" values for in- parently less specific rat epididymal fat pad (12, sulin content of blood differ widely (15). This 13). Similarly, by Vallance-Owen's method, two discrepancy even exists between workers using totally pancreatectomized patients exhibited no cir- undiluted plasma and rat diaphragm as assay tis- culating insulin activity either before or after in- sue, but varying otherwise in method. Metz (2) gesting 100 g of glucose (Table VI). Further in- found baseline insulin activity in pancreaticodu- dications that augmented outflow of insulin activity odenal venous blood of dogs to vary from 10 to 31 from a pancreatic segment actually signified in- mU per 100 ml of plasma (100 to 310 /tU per ml) creased secretion of insulin were the absence of at blood glucose levels between 37 and 86 mg per concomitant enhancement of insulin activity in 100 ml, with enhancement to values of 330 and peripheral blood, and the demonstration that in 570 mU per 100 ml plasma (3,300 and 5,700 vitro addition of sulfonylureas did not cause non- FLU per ml) when blood glucose was 300 mg per 100 ml. These absolute values, about ten times TABLE VI greater than those recorded under similar ex- Lack of insulin activity in plasma of totally depancreatized patients perimental conditions in the present study, are largely due to his ten times less sensitive dose- Total glucose uptake Plasma insu- response curve, since actual net glucose uptakes lin activity Days Plasma from plasma were essentially the same in both without GBSS Fasting after After Patient insulin alone plasma glucose* Fasting glucose laboratories. In contrast, the specific immunoassay of Yalow and Berson pAg/1O mg diaphragm pAU/ml plasma (16) yielded values L.F. 5 287t 292 290 <1+ <1 for actual insulin concentrations in human plasma +6 416 +22 similar to for J. B. 2 289 284 309 0 l very those biologic "insulin activity" +20 +14 +t20 found by Vallance-Owen and co-workers (4, 6), 100 g by mouth. Wright (7), .nd Seltzer and Sluith (17). These 1 Mean A-SEM. + 20 pg glucose uptake/10 mg diaphragm =1 puU insulin activity/mnl. comparable results by two entirely different pro- INSULINOGENIC EFFECTS OF GLUCOSE AND HYPOGLYCEMIA-INDUCING SUBSTANCES 297 cedures suggest that absolute concentrations of circulating insulin are, in fact, low rather than GLUCOSE TOLBUTAMIDE CHLORPROPAMIDE METAHEXAMIDE high. T Even in relative terms, however, the parameter -j, 100- of insulin-releasing activity enabled sharp quali- tative and quantitative classification of test sub- u- 75- stances. Glucose and all three sulfonylureas stim- z ulated prompt discharge of insulin, whereas salicylate and indole-3-acetic acid had no effect. The insulogenic potency of glucose far exceeded that of sulfonylureas. Rapidly induced hyper- U) Wil glycemia evoked a tremendous outpouring of insulin, whose secretory rate promptly reached FIG. 4. COMPARATIVE INSULOGENIC POTENCIES OF GLU- COSE AND SULFONYLUREAS. Respective vertical bars indi- a maximal level which was essentially sustained cate the mean increase above baseline value in secretory for 30 minutes. The slight drop during the last rate of insulin activity during the entire 30 minutes after period may have reflected the waning glycemic injection of test substance. The insulogenic omnipotence stimulus as blood glucose level fell rapidly, since of hyperglycemia is obvious. Metz (18) found that glucose infused at mount- ing rates for 45 minutes continued to elicit pro- betics (21). Similarly, both in normal subjects portional increase of insulin secretion. In con- and in untreated mild diabetics, Yalow and co- trast to the massive response to glucose, the in- workers (1) found greater enhancement of plasma sulin-releasing potency of all three sulfonylurea insulin levels after glucose than after tolbutamide. compounds was much smaller, both in magnitude In nondiabetics the average increase after glu- and duration. Mean outflow rates for the nine cose was 119 uU per ml, compared to only 28 IU periods of observation never exceeded 20 per per ml after tolbutamide; in diabetic patients cent of the maximal rate for glucose (Table II corresponding mean increases were 132 ,tU per and Figure 2). Moreover, in Figure 4 are com- ml after glucose and 52 uU per ml after tolbuta- pared respective mean increases of insulin secre- mide. In these clinical studies, the full insulogenic tion during the entire 30 minutes after injection superiority of glucose over sulfonylurea compounds of glucose and sulfonylureas. Considering en- was probably masked for several reasons. In ad- hancement by glucose as 100 per cent, correspond- dition to significant hepatic binding of intra-portal- ing insulogenic potencies were 9.9 per cent for venous insulin during its first traverse of the liver tolbutamide, 18.6 per cent for chlorpropamide, and (22), and dilution of the remainder upon enter- 11.7 per cent for metahexamide. Although blood ing the post-hepatic circulation, isolated values of levels of sulfonylurea were not measured, main- insulin concentration tend to obscure the sustained tenance of effective concentrations was assured nature of the glycemic stimulus compared to the by the known half-life disappearance rates for transient one invoked by sulfonylureas. tolbutamide, chlorpropamide, and metahexamide An opposite interpretation might be that the of 4.6, 35, and 19 hours, respectively (19, 20). observed insulogenic superiority of glucose was The insulogenic omnipotence of glucose has an artifactual sequela of the chronic hypoglycemia previously been intimated in published clinical induced in experimental animals. The hypogly- observations. One hour after giving 50 g of cemic state is combated by the counterregulatory glucose orally, Vallance-Owen and associates release of epinephrine (23), adrenocorticosteroids found an increment in plasma insulin activity of (24), and glucagon (25). Epinephrine and cor- 333 utU per ml in normal subjects (4), and 219 tisol strongly inhibit the enhancing effect of insu- tU per ml in maturity-onset diabetics (6). Two lin on glucose uptake by rat diaphragm (26, 27), and one-half hours after similar individuals took wherefore their presence in excess in test plasmas 2 g of tolbutamide by mouth, the average increase could have blunted the full betacytotropic effects of plasma insulin activity was 114 /U per ml in of sulfonylurea agents. However, even in ani- normal subjects, and 130 /U per ml in adult dia- mals whose hypoglycemia was obliterated by glu- 298 HOLBROOKE S. SELTZER cose administration, it is unlikely that preinjection blood glucose was optimally responsive to tolbu- plasma levels of epinephrine or cortisol changed tamide, fasting normoglycemia was indefinitely significantly during the next 30 minutes, since maintained at well tolerated blood levels between epinephrine-in-oil had been used in preparing all 25 and 35 mg per 100 ml (34). In the present animals, and circulating cortisol titers are known study, failure of the latter concentrations to mani- to be markedly elevated during surgical trauma fest betacytotropic activity confirms the exclusively per se (28). extrapancreatic action of salicylate. Manchester, The implication that the physiological stimulus Randle and Smith (35) showed significant aug- represented by a rising blood sugar level can mentation of glucose uptake by rat diaphragm stimulate normal beta cells maximally and indefi- when salicylate was added in vitro, a finding veri- nitely, suggests that glucose and sulfonylureas fied in Table IV; and Falcone (36) subsequently impinge differently upon the beta cell. Glucose localized a salicylate-induced block in oxidative appears to be truly "insulogenic" in that it not phosphorylation, which presumably results in ac- only triggers release of stored insulin granules, celerated disposal of glucose via anaerobic glycoly- but also engenders formation and secretion of new sis. The metabolic action of salicylate thus closely insulin in proportion to the concentration of glu- stimulates the peripheral mechanism whereby cose in pancreatic arterial blood. In contrast, phenethylbiguanide substances also cause disap- sulfonylurea compounds seem merely capable of pearance of circulating glucose (37). "solubilizing" preformed, stored insulin and en- The hypoglycemia-inducing property of indole- abling it to be rapidly discharged. Thereafter, 3-acetic acid was discovered by Mirsky, Diengott so long as effective sulfonylurea blood levels per- and Perisutti (38) during studies elucidating its sist, spontaneously synthesized insulin is promptly ability to inhibit insulinase activity. Clinical ti- secreted regardless of arterial glucose concentra- tration in diabetic patients revealed this physio- tion. Volk and Lazarus (29) demonstrated totally logical end-product of tryptophan metabolism to degranulated beta cells during chronic tolbuta- be fully as effective as tolbutamide in reducing mide-induced hypoglycemia in rabbits and dogs. fasting hyperglycemia (39). Because indole-3- Moreover, after 9 months of daily tolbutamide acetic acid is also a powerful plant growth hor- administration to rats, Baender (30) found that mone, it might have possessed an "insulotropic" several days after stopping the drug, normal beta action similar to that by which animal growth cell granulation had been restored. Of similar hormone stimulates anatomical and functional hy- significance may be the repeated clinical observa- pertrophy of beta cells (40). However, not only tion that, although sulfonylureas normalize glu- was the agent devoid of betacytotropic activity, cose levels in optimally responsive ketoacidosis- but its addition to the in vitro assay system did resistant diabetics, the oral glucose tolerance test not increase glucose uptake by rat diaphragm. itself frequently shows little improvement (31, In addition, even profound hypoglycemic re- 32). This apparent paradox is actually a pre- sponses of elderly diabetics to the substance were dictable sequela, if pancreatic islets continuously not accompanied by increased plasma insulin ac- bathed in a sulfonylurea solution cannot conserve tivity (39). In a negative sense, therefore, these a depot of insulin to release upon hyperglycemic findings support the conclusion of Mirsky and co- demand. workers that indole-3-acetic acid preponderantly Recent clinical trials with salicylate in diabetic blocks destruction of insulin by insulinase. patients have shown that its ability to lower blood Finally, the apparent failure of proved hypo- sugar is equal to that of sulfonylurea and phen- glycemia-inducing agents to cause further lower- ethylbiguanide compounds. Unfortunately, nor- ing of blood sugar was an expected sequela of the malization of fasting blood glucose in maturity- experimental procedure. Preliminary efforts in onset diabetics was achieved by Reid and Mac- intact dogs had consistently shown that the neces- Dougall (33) at blood levels of salicylate (above sarily extensive surgical manipulations caused 40 mg per 100 ml) too toxic to encourage its peripheral hyperglycemia, which not only evoked therapeutic use. However, when this agent was premature insulin secretion but also canceled the selectively used only in mild adult diabetics whose glucose-depressing activity of the agent under INSULINOGENIC EFFECTS OF GLUCOSE AND HYPOGLYCEMIA-INDUCING SUBSTANCES 299 scrutiny. When the animal was depleted of gly- 6. Vallance-Owen, J., Hurlock, B., and Please, N. W. cogen by starvation, phlorhizin, and epinephrine Plasma-insulin activity in diabetes mellitus, meas- to prevent ured by the rat diaphragm technique. Lancet 1955, otherwise uncontrollable hyperglycemic 2, 583. interference, the normal hypoglycemic response 7. Wright, P. H. Plasma-insulin estimation by the rat- to the test substance was also blunted. diaphragm method. Lancet 1957, 2, 621. 8. Gundersen, K. Personal communication. 9. Vallance-Owen, J. Measurement of insulin activity SUMMARY in blood. Diabetes 1956, 5, 248. Respective betacytotropic activities of glucose, 10. Groen, J., Kamminga, C. E., Willebrands, A. F., and sulfonylurea compounds, salicylate, and indole-3- Blickman, J. R. Evidence for the presence of insulin in blood serum. A method for an approxi- acetic acid were semiquantitatively compared by mate determination of the insulin content of blood. measuring the effect of each agent on the outflow J. clin. Invest. 1952, 31, 97. rate of insulin activity in pancreaticoduodenal ve- 11. Vallance-Owen, J., and Lukens, F. D. W. Studies nous blood of phlorhizinized dogs. on insulin antagonism in plasma. Endocrinology Glucose and three sulfonylureas 1957, 60, 625. (tolbutamide, 12. Leonards, J. R. Insulin-like activity of blood. What chlorpropamide, and metahexamide) all promptly is it? Fed. Proc. 1959, 18, 272. enhanced concentration of insulin activity in the 13. Goldberg, H. L., and Egdahl, R. H. Studies sug- pancreatic venous effluent. However, whereas gesting the extra-pancreatic production of sub- the response to glucose was both intense in mag- stances with insulin-like activity. Fed. Proc. 1961, nitude and sustained, insulin release by sulfonyl- 20, 190. 14. Haist, R. E. Factors affecting insulin content of ureas was only moderate in degree and transient pancreas. Physiol. Rev. 1944, 24, 409. in duration. Mean enhancement of insulin secre- 15. Yalow, R. S., and Berson, S. A. Plasma insulin in tion by glucose was five to ten times the output man. Amer. J. Med. 1960, 29, 1. after sulfonylurea compounds. Neither salicylate 16. Yalow, R. S., and Berson, S. A. Plasma insulin nor indole-3-acetic acid stimulated pancreatic re- concentrations in nondiabetic and early diabetic subjects. Determinations by a new sensitive im- lease of insulin activity. muno-assay technic. Diabetes 1960, 9, 254. The data suggest that: 1) both glucose and sul- 17. Seltzer, H. S., and Smith, W. L. Plasma insulin fonylurea compounds cause prompt release of in- activity after glucose: An index of insulogenic re- sulin from pancreatic beta cells; 2) the betacyto- serve in normal and diabetic man. Diabetes 1959, tropic activity of glucose is far more potent than 8, 417. that of sulfonylureas; 3) glucose is truly insulo- 18. Metz, R. Personal communication. 19. Knauff, R. E., Fajans, S. S., Ramirez, E., and Conn, genic; i.e., capable of engendering formation of J. WV. Metabolic studies of chlorpropamide in new insulin, whereas sulfonylureas seem only able normal men and in diabetic subjects. Ann. N. Y. to release preformed stores of insulin; and 4) Acad. Sci. 1959, 74, 603. salicylate and indole-3-acetic acid cause peripheral 20. Hamwi, G. J., Skillman, T. G., Kruger, F. A., hypoglycemia via extrapancreatic mechanisms. Roush, W. H., and Freedy, L. R. Comparative pharmacology and clinical responses to metahexamide. Ann. N. Y. Acad. Sci. 1959, 82, 547. REFERENCES 21. Vallance-Owen, J., Joplin, G. F., and Fraser, R. 1. Yalow, R. S., Black, H., Villazon, M., and Berson, Tolbutamide control of diabetes mellitus. Clini- S. A. Comparison of plasma insulin levels fol- cal responsiveness and insulin reserve. Lancet lowing administration of tolbutamide and glucose. 1959, 2, 584. Diabetes 1960, 9, 356. 22. Madison, L. L., and Kaplan, N. The hepatic binding 2. Metz, R. The effect of blood glucose concentration of P13`-labeled insulin in human subj ects during a on insulin output. Diabetes 1960, 9, 89. single transhepatic circulation (abstract). J. Lab. 3. Nelson, N. A photometric adaptation of the Somogyi clin. Med. 1958, 52, 927. method for the determination of glucose. J. biol. 23. Goldfien, A., Moore, R., Zileli, S., Havens, L. L., Chem. 1944, 153, 375. Boling, L., and Thorn, G. W. Plasma epinephrine 4. Vallance-Owen, J., and Hurlock, B. Estimation of and norepinephrine levels during insulin-induced plasma-insulin by the rat diaphragm method. Lan- hypoglycemia in man. J. clin. Endocr. 1961, 21, cet 1954, 1, 68. 296. 5. King, E. J. Microanalysis in Medical Biochemistry, 24. Christy, N. P., Longson, D., Horwitz, XV. A., and 1st ed. New York, Grune & Stratton, 1947, p. 17. Knight, M. M. Inhibitory effect of chlorproma- 300 HOLBROOKE S. SELTZER

zine upon the adrenal cortical response to insulin 32. Fajans, S. S., and Conn, J. W. Tolbutamide-induced hypoglycemia in man. J. clin. Invest. 1957, 36, improvement in carbohydrate tolerance of young 543. people with mild diabetes mellitus. Diabetes 1960, 25. Unger, R. H., Eisentraut, A. M., McCall, M. S., and 9, 83. Madison, L. L. The hormonal role of endogenous 33. Reid, J., MacDougall, A. I., and Andrews, M. M. glucagon in blood: Glucose homeostasis as dem- Aspirin and diabetes mellitus. Brit. med. J. 1957, onstrated by a specific immunoassay (abstract). 2, 1071. J. clin. Invest. 1960, 39, 1036. 34. Seltzer, H. S. Unpublished observations. 26. Groen, J., v.d. Geld, H., Bolinger, R. E., and Wille- 35. Manchester, K. L., Randle, P. J., and Smith, G. H. brands, A. F. The anti-insulin effect of epineph- Some effects of sodium salicylate on muscle metab- rine. Its significance for the determination of olism. Brit. med. J. 1958, 1, 1028. serum insulin by the rat diaphragm method. Dia- 36. Falcone, A. B. Studies on the mechanism of action betes 1958, 7, 272. of salicylates; effects on oxidative phosphoryla- 27. Stadie, W. C., Haugaard, N., and Marsh, J. B. The tion (abstract). J. clin. Invest. 1959, 38, 1002. effect of growth hormone and cortisone on the 37. Kruger, F. A., Skillman, T. G., Hamwi, G. J., action of bound insulin. J. biol. Chem. 1952, 198, Grubbs, R. C., and Danforth, N. The mechanism 785. of action of hypoglycemic guanidine derivatives. 28. Hume, D. M. The secretion of epinephrine, nor- Diabetes 1960, 9, 170. epinephrine and corticosteroids in the adrenal ve- 38. Mirsky, I. A., Diengott, D., and Perisutti, G. The nous blood of the dog following single and re- hypoglycemic and insulinase-inhibitory action of peated trauma. Surg. Forum 1957, 8, 111. some plant growth hormone regulators. Endo- 29. Volk, B. W., and Lazarus, S. S. Effect of insulo- crinology 1956, 59, 715. genic agents on the pancreatic islets. Diabetes 39. Seltzer, H. S., and Smith, W. L. "Plasma insulin 1960, 9, 264. activity" in human diabetes during hypoglycemic response to tolbutamide and indole-3-acetic acid. 30. Baender, A. Toxicological and histological studies Proc. Soc. exp. Biol. (N. Y.) 1959, 100, 171. with tolbutamide. Ann. N. Y. Acad. Sci. 1957, 71, 40. Batts, A. A., Bennett, L. L., Ellis, S., and George, 152. R. A. Growth hormone-induced reduction of 31. de Salcedo, I., and Borges, F. The clinical results glycosuria and partial repair of the islets of Langer- of the treatment of diabetes mellitus with chlor- hans in partially pancreatectomized diabetic rats. propamide. Ann. N. Y. Acad. Sci. 1959, 74, 891. Endocrinology 1956, 59, 620.

CORRECTION On page 1798 of the article "The Influence of Barbiturates on Coumarin Plasma Levels and Prothrombin Response," by Dayton et al. (J. clin. Invest. 1961, 40) the legend to Figure 1 should have the following addendum: "which was preceded by the same dose of barbital daily for 3 prior days."