long-standing glucose intolerance, delayed hyperinsulinism following oral glucose, and retinal microaneurysms typical of diabetes. Plasma that was essentially ...
"BIG INSULIN": A NEW COMPONENT OF PLASMA INSULIN DETECTED BY IMMUNOASSAY BY JESSE ROTH, PHILLIP GORDEN, AND IRA PASTAN NATIONAL INSTITUTE OF ARTHRITIS AND METABOLIC DISEASES, NATIONAL INSTITUTES OF HEALTH, BETHESDA, MARYLAND
Communicated by James A. Shannon, April 17, 1968
Since the advent of radioimmunoassay, I the concentration of insulin in plasma of normal and diabetic subjects in response to many stimuli has been studied extensively.2 However, the chemical nature of the circulating hormone is not entirely clear. We find that endogenous plasma insulin has two components. One is very similar to pancreatic insulin. The other component is larger in size and immunologically less reactive than pancreatic insulin. With our antiserum, it accounts for up to 50 per cent of plasma insulin measured by immunoassay. Materials and Methods.-Purified crystalline porcine and bovine insulin were gifts of Eli Lilly. Highly purified A-chain and B-chain sulfonates of bovine insulin were gifts of Dr. Arnold Marglin.' Cytochrome c, pancreatic ribonuclease, and porcine adrenocorticotropin (ACTH) were purchased from Mann, crystalline human serum albumin from Armour, rabbit fraction II and normal guinea pig serum from Pentex, guinea pig fraction II from Hyland, and NaI'25 and NaI"11 from Union Carbide. Guinea pig antiporcine insulin serum was a gift of Drs. A. Kagan and S. M. Glick. Antibody to guinea pig gamma globulin was generated in sheep by repeated injection of guinea pig fraction II in complete Freund's adjuvant (Difco). Albumin and ACTH were labeled with 112 and insulin with I'll by the chloramine T method at specific activities of 5, 5, and 450 uc/igg, respectively.4 The albumin-I125 was freed of radioactive contaminants by batch adsorption with Dowex 1X10. The ACTH-I'25 was purified by adsorption to silica and elution with acetic acid and acetone.' The insulin-I131, prepared fresh for each assay, was purified on a cellulose column.1 The normal subjects for the experiments were healthy male volunteers in their early twenties without a family history of diabetes. For the standard oral glucose tolerance tests 100 gm of glucose was administered in the morning after an overnight fast; for 3 days prior to the studies the subjects received a normal diet that included 300 gm of carbohydrate daily. For the starvation tests, food was withheld until the urine reacted strongly positive for acetone; the fast, about 36 hr in duration, was terminated by the administration of glucose, 100 gm orally. Patient MIT is an obese adult female with long-standing glucose intolerance, delayed hyperinsulinism following oral glucose, and retinal microaneurysms typical of diabetes. Plasma that was essentially free of insulin was obtained after a 72-hr fast from a patient who had had an islet-cell adenoma removed several years earlier. None of the subjects had ever been treated with insulin. Blood was drawn into heparin, refrigerated, and centrifuged, and the plasma stored at -20°. Plasma, 1-3 ml, was enriched with albumin-I125 and iodidel25 and applied to a 1 X 50-cm column of Sephadex G-50 fine (Pharmacia). Shorter columns or coarser grades of gel were unsatisfactory. Since the results of radioimmunoassay can be affected by variations in the salt content of the samples,' the column was equilibrated and developed in the same diluent that was used in the immunoassay (see below); also, none of the column fractions corresponding to the salt peak (marked by the iodidel25) were assayed. Fractions of 1-1.5 ml were collected, the radioactivity counted to localize the plasma protein and salt peaks, and the fractions stored at 40 until assayed. Diluent for the insulin assays and for the gel filtration was veronal (0.05 11, pH 8.6) to which had been added human serum albumin (2.5 mg/ml), rabbit fraction II (0.1 mg/ml), and toluene as a preservative. For the assay7 a series of tubes were prepared each of which contained, in a total volume of 1 ml, insulin-I"ll (approx. 0.05 mug), insulin 138
VOL. 61, 1968
BIOCHEMISTRY: ROTH, GORDEN, AND PASTAN
139
antibody (1:80,000), ethylenediaminetetraacetate (EDTA) (0.01 4l), normal guinea pig serum (0.005 ml), and either crystalline porcine insulin (0-10 mug), or plasma (0.1 ml), or an aliquot of the fractions from gel filtration (0.4-0.8 ml). After 3 days at 40, sheep anti-guinea pig gamma globulin (0.1 ml) was added to precipitate the antibody-bound insulin-I3Il8' 9 After another 18 hr at 40 the mixtures were centrifuged and the supernates were decanted. Both the precipitates and supernates were counted in a well-type scintillation counter that had been set to exclude 1125 radioactivity. The B/F of insulin-I13l was expressed as a function of the concentration of unlabeled porcine insulin. The precipitates were not washed. However, in our experiments, in the absence of antiinsulin serum, radioactivity in the precipitates was uniformly 6% or less. Likewise, damage to the insulin-I'3l during incubation in our assay was slight and quite similar in all samples of a given assay (see below). In the presence of an excess of an anti-insulin serum, 85% or more of the radioactivity was in the precipitate. To calculate recoveries of plasma insulin following gel filtration, an aliquot of the plasma that had been applied to the column was assayed at the same time as the column fractions. Cytochrome c was measured by its absorbance at 410 mIA, ribonuclease by its hydrolysis of denatured RNA,10 and unlabeled ACTH by its fluorescence at 350 mM.
Results.-When plasma was filtered on Sephadex G-50, the endogenous insulin was recovered in two peaks. One peak, quite discrete, appeared at 0.45 of the distance between the protein and salt peaks ("little insulin") (Fig. 1A). The other peak ("big insulin") was detected halfway between "little insulin" and the plasma protein peak and comprised from 0 to 50 per -ent of the total plasma insulin (Fig. 1). When highly purified crystalline porcine insulin (pancreatic insulin) was added to insulin-free plasma, only a single peak was observed which corresponded to "little insulin" (Fig. 2). Likewise, when pancreatic insulin was added to plasma from patient MIT, in which "big insulin" constituted about one third of the endogenous insulin, the added pancreatic insulin appeared only with the "little insulin" peak (Fig. 1B). To prepare large amounts of "big" and "little" insulin, 40 ml of plasma from patient /1\IT were filtered on Sephadex G-50 (4 X 54 cm) (Fig. 3A). The fractions corresponding to "big insulin" and to "little insulin" were pooled, dialyzed, lyophilized, and redissolved in 1 ml of diluent. Each fraction was filtered again on Sephadex G-50 (1 X 50 cm). Each of the two components retained its characteristic migration pattern and remained free of the other component (Fig. 3). Purified human and porcine insulins differ by one amino acid. With our guinea pig antiporcine insulin serum, human insulin in plasma and purified porcine insulin reacted identically;" purified bovine insulin reacted less strongly (Fig. 4). The isolated A and B chains were essentially unreactive12 (Fig. 4). "Little insulin," tested over a 20-fold range of dilutions, reacted identically with porcine insulin (Fig. 5). However, "big insulin," tested over a 40-fold range of dilution, reacted nonidentically (Fig. 6). The nonidentical cross-reactivity of "big insulin" accounts in part for the truncated appearance of its peak on gel filtration. When "big insulin" was measured against a "big insulin" standard, instead of the porcine insulin standard, the "big insulin" peak had a more usual appearance. Because "big insulin" and porcine insulin did not react identically, and since in all other experi-
to
BIOCHEMISTRY: ROTH, GORDEN, A AND PASTAAT
140
PROC. N. A. S.
ments we measured "big insulin" by reference to porcine insulin, we usually underestimated the fraction of total insulin that was "big insulin." One other aspect of the cross-reactivity deserves comment. Although one of the two components of circulating insulin reacts with our antiserum nonidentically with porcine insulin, insulin in unfractionated human plasma was not distinguished, under our experimental conditions, from purified porcine insulin. An analogous result was obtained by mixing bovine with porcine insulin; they differ in their reactivities about as much a.s "big" and "little" insulin. Although the individual components were distinguished, mixtures of bovine and porcine, even up to 1: 1, were barely distinguishable from pure porcine insulin (Fig. 4). To obtain an approximate estimate of the size of the two components, markers 0, 1~~ ~ ~ ~~~~~~~~~~~~~.28
0.8
1A
Z
16
~~~~~~~~~3A
-21 0 6061-4
_
PERIPHERAL ARTERY AT 25 MINUTES
A,.,
U)
-
0 5
ZA
cn
-
z-
z
u
0
E 0.4
.51
20 7 5
z 0
I
.4
ZL -
_---AM_ .
53
Hr
.11
10
4 M LI
1.05
.12 .13 -.16 ,19
PANCREATIC VEIN AT 10 MINUTES
u
V
10 t
.08
)
ZAZ J) D
PANCREATIC VEIN ]07 AT 25 MINUTES
1,0,
,
Z L. i.ui
3L
E
40 9.-
OD
.18
z(1) I0.4
35
1.
U.
a40
.28
0 0
s
E E
1.08
4
0.5 ---
_.23
,~ -.26o5
0
--46-9O-.956
FIG. 7.-Arteriovetotus difference of insulin across the pancreas. An anaesthetized dog infused with glucose, 0.5 gm/kg body weight. Ten and 25 min later blood was collected from the pancreatic vein and femoral artery. Because of variation in the volume of the effluent fractions with the 25-min samples, the insulin was expressed as m/g/ml of effluent/2 ml of plasma applied. The other notations are the same as in the legend to Fig. 1.
was
Vol. I1, 1968
BIO('HE.IIIS7'1TY: ROTH, GORDE.X, .1AD 1PASTA'X.Y
14:3
GTT 15 MINUTES
STARVATION GTT 15 MINUTES
0.8 -
.20 21 0.6
6-_
