Hormone Secretion and Glucose Metabolism in Islets of Langerhans of the Isolated Perfused Pancreas from Normal and Streptozotocin Diabetic Rats*

The glucose responsiveness of (Y- and P-cells of normal as well as untreated and insulin-treated streptozotocin diabetic rats was tested in the extracorporeal perfusion system. Also assessed was the possible in vitro effect of added insulin on the glucose sensitivity of islets from untreated diabetic animals. Insulin and glucagon secretion served as functional indicators of glucose responsiveness of the two cell types. The rate of glucose entry into islet tissue was estimated, and the effect of glucose on the tissue supply of ATP and lactate and the cyclic 3’:5’-AMP level of islets was measured under the above in vitro conditions. It was demonstrated that P-cells are more accessible to glucose than a-cells, that glucose entry into islet cells is not significantly modified by insulin and that glucose had no effect on ATP, lactate and cyclic 3’:5’-AMP levels of islet tissue under any of the conditions investigated. High insulin in oitro elevated ATP levels of a-cell islets independent of extracellular glucose.

The glucose responsiveness of (Y-and P-cells of normal as well as untreated and insulin-treated streptozotocin diabetic rats was tested in the extracorporeal perfusion system. Also assessed was the possible in vitro effect of added insulin on the glucose sensitivity of islets from untreated diabetic animals.
Insulin and glucagon secretion served as functional indicators of glucose responsiveness of the two cell types. The rate of glucose entry into islet tissue was estimated, and the effect of glucose on the tissue supply of ATP and lactate and the cyclic 3':5'-AMP level of islets was measured under the above in vitro conditions.
It was demonstrated that P-cells are more accessible to glucose than a-cells, that glucose entry into islet cells is not significantly modified by insulin and that glucose had no effect on A new concept of bihormonal etiology of diabetes mellitus has evolved during the last few years implying that both a-and p-cells of the pancreatic islets might be primarily impaired (1,2), whereas previously the primary lesion was thought to be confined to the P-cells (3,4). It is conjectured that both cell types have the same basic defects, i.e. an impairment or even inability of the endocrine cells to recognize and respond to altered glucose concentration in the blood. The reduced sensitivity to glucose manifests itself in insulin deficit and glucagon excess in plasma relative to the blood sugar level or, expressed differently, in increased glucagon to insulin ratios in the blood. The hypothesis has gained momentum as a result of the recent observation that the hypothalamic oligopeptide somatostatin reduces blood sugar in the diabetic organism (5,6) 6053 isolated perfused rat pancreas (7,8) for assessing the secretory function of 01-and P-cells in vitro (9,10) and applying quantitative histochemical methods (11,12)  into islet cells is not significantly modified by insulin, and that glucose had no effect on ATP, lactate, and cyclic AMP levels of islet tissue under any of the conditions. Glucose caused insulin release from normal but not from diabetic islets, and rapidly and efficiently suppressed glucagon secretion of the normal and of the insulin treated diabetic pancreas. However, glucose was less effective in inhibiting glucagon secretion from untreated diabetic pancreas whether insulin was added to the perfusate or not. These functional and biochemical results are compatible with the classic view that a primary lesion of the P-cells can explain the diabetic syndrome and that it seems unnecessary to postulate an additional primary lesion in a-cells.

Animals
and Perfusion System-Male Sprague Dawley rats weighing 300 to 400 g, fed ad libitum with Purina rat chow and water were used in all experiments.
Diabetes was induced with streptozotocin (65 t.o 70 mg/kg intravenously) (9). Control animals received no treatment. After diabetes had been demonstrated by blood glucose analysis (usually within '2 to 3 days following streptozotocin) all diabetic animals were treated with insulin. During the first week of treatment they received 5 units subcutaneously of NPH insulin daily. During the ensuing treatment period the daily dose was reduced to 2 units subcutaneously.
This treatment allowed normal weight gain in the diabetic animals but did not normalize the blood sugar. In animals designated "untreated diabetics" insulin was withdrawn for 2 to 3 days prior to the perfusion experiment.
In contrast to the previous studies (9,10) in this series all animals entered the experiment in the fed state in order to avoid possible complications resulting from hypoglycemia in the insulin-treated group.
Blood samples were obtained just before removal of the pancreas from the heparin-treated animals. The samples were drawn by puncturing the aorta closely above the most cranial ligature. The blood was injected into tubes containing 500 Kallikrein inactivating units/ml of Trasylol (FBA Pharmaceuticals, N. Y.) and 1.25 mg of EDTA/ml and was kept on ice no longer than 30 min. The plasma was then separated by centrifugation and stored at -20" until assayed. The plasma samples were used to determine glucose, lactate, ketone bodies, glucagon, and insulin and all values were corrected for dilution with the Trasylol.
The pancreas was isolated and perfused using the procedure of Grodsky et al. (7) with minor modifications (8). The anesthetic and surgical preparation of the animals and composition of the perfusion media have been described in detail (8)(9)(10). Inulin was present at 0.33 mM throughout to serve as marker of the extracellular water space of islets.
From 10 to 24 separate perfusions were performed for different experimental conditions. The secretory stimulus was a 10 rnM concentration of a mixture of 19 amino acids in proportions found in rat serum ( Fig. 1, protocols 1 through 3). As compared to previous studies (9, 10) cystine was left out of the amino acid mixture.
The amino acid mixture was present throughout the entire perfusion which lasted 23 min in protocols 1 and 2 and 35 min in protocol 3. amino acid mixture for 20 min, 10 mM glucose was superimposed in half the experiments in each group, either for 3 min, as in protocols 1 and 2, or for 15 min, as in protocol 3. In one set of experiments, I ag/ml of insulin was included in the perfusion medium throughout (protocol 2). Protocol 1 was used with the pancreas from controls, treated diabetics, and untreated diabetics. Protocols 2 and 3 were used with untreated diabetics only. In all perfusion experiments 0.5-ml samples of perfusate were obtained at suitable intervals from the cannula in the portal vein, cooled on ice, and, after completing the experiment, frozen at -20" until assayed.
In protocols 1 and 2, usually nine samples were collected, and in protocol 3 the number of samples was increased to 14. The rates of insulin and glucagon release were calculated by multiplying the concentration of the respective sample by the flow rate, (5 to 6 ml/min) which was measured at frequent intervals. Sampling of Pancreas and Tissue Preparation for Quantitative Histochemistry-The pancreas was sampled at the termination of the perfusion at 23 min ( Fig. 1, protocols 1 and 2) using quick freezing as described previously (12,13). Microtome sections of 30 wrn thickness were cut in a cryostat at -20" and then freeze-dried as described in previous publications (12,13). Islets were dissected freehand without prior staining and weighed using a quartz fiber fishpole balance (14). To obtain samples for lactate measurements, gloves were worn during dissection and the materials that come in contact with the tissue during the process of dissecting and weighing were thoroughly cleaned with alcohol and acetone. Inulin, Glucose, Lactate, ATP, and Cyclic AMP in Islet Tissue-Oil well methods in combination with enzymatic -cycling and fluorometry were used for measuring inulin, glucose, lactate, and ATP in islet tissue. Methods for inulin (15), glucose, and ATP (11,12) have been described previously.
The inulin method was modified in that the tissue sample was heated in 0.1 N NaOH. This pretreatment destroyed free glucose and enzymes that would interfere in the enzymatic assay for fructose. Subsequently, fructose is liberated from the inulin by hydrolysis in weak acid. The actual procedure for inulin was as follows. The samples, placed in the oil well rack, were suspended in 0.07 pl of 0.1 N NaOH, covered with oil, and heated for 25 min in an oven set at 105". After cooling to 25" 0.05 pl of 0.2 N HCl was added and the heating step was repeated (25 min in an oven set at 105") in order to hydrolyze inulin. After cooling, 1.5 ~1 of the reagent for measuring fructose was added. The composition of the reagent was described in the original procedures for measuring fructose (12,15). The racks were incubated for 45 min and the reaction was terminated by adding 0.32 ~1 of 1 N NaOH. A heating step followed (20 min at 75"). TPNH resulting from fructose was then measured by enzymatic cycling as described previously.
The enzymes were adjusted to allow amplification of 3ooO times. Inulin standards ranged from 0.3 to 1.6 pmol (Fig. 2). The lactate assay was a micro modification of a procedure published previously (16,17). Islet samples were suspended in 0.07 ~1 of 0.05 N HCl under oil and were heated for 20 min at 75". After cooling and neutralizing with 0.017 ~1 of 2-amino-2-methyl-propyl alcohol base 0.125 ~1 of the following reagent was added: 0. propyl alcohol buffer, pH 9.9; DPN, 650 pM; sodium glutamate, 5 mM; lactic acid dehydrogenase (beef heart), 75 wg/ml; and glutamate-pyruvate transaminase, 75 rg/ml. The oil well rack was then incubated for 45 min at 22" and the reaction was terminated with alkali (1.08 ~1 of 0.125 N NaOH) and heating (20 min at 75"). Tbe DPNH formed was measured by enzymatic cycling according to the procedure of Kato et al. (18). This reaction was performed in 50 ~1 in glass tubes (10 x 75 mm). The amplification needed for this application was 4006fold. Lactate standards ranging from 5 x lo-l3 to 3 x lo-'* mol were carried through the entire process. Great care was taken to avoid contamination of tissue and materials with lactate and therefore gloves had to be worn while carrying out this enzymatic assay. For measurements of cyclic AMP, 5 to 7 islet or exocrine tissue samples were dissected, weighed, and pooled in glass tubes (6 x 50 mm). The pooled samples weighed between 0.3 and 1.0 Kg. Forty microliters of 10% trichloroacetic acid were added to the tubes which were allowed to sit at room temperature for 15 to 30 min. The tubes were gently mixed and 35 ~1 of the extract were removed and placed into clean tubes. (A dissecting microscope was used during this transfer to prevent inadvertent transfer of tissue samples.) The trichloroacetic acid extracts were then evaporated to dryness in a vacuum centrifuge.
This procedure completely removed all trichloroacetic acid. The dried residue was dissolved in 50 ~1 of H,O and then acetylated and assayed for cyclic AMP as previously described (19). Measurements were made in triplicate. Treatment of tissue extracts with beef heart phosphodiesterase destroyed all immunoreactive material indicating that the measured substance was indeed cyclic AMP.
Other Analytical Methods-Immunoreactive glucagon and insulin were measured by using double antibody systems as previously described (10,20,21 (9,17). Data given in Table  I  high glucagon release (Fig. 3). Since no samples were taken during the early period, only the second phase of release is seen (compare Refs. 9 and 10). The rate of release is almost twice that observed in previous studies (9,10). This may be due to the fed state of the animals or due to the change of the perfusion protocol. Consistent with published data is the complete absence of a P-cell response to the amino acid mixture alone (10) (Fig. 4) greatly reduced a-cell response to the amino acid stimulus (approximately one-third of the control rate) consistent with the results of a previous study (9) (Fig. 3). Glucose suppression of a-cell function was delayed and was clearly seen only when of the a-cells in the in vitro setting (Fig. 3).  The results for each pancreas are shown individually. These individual values represent the means * S.E. from usually six islet samples. Also given are the means * S.E. for each group. Perfusions were performed in the presence and absence of 10 mM glucose in the perfusion medium. The exposure to glucose following a preperfusion of 20 min lasted for 3 min, after which period the pancreas was sampled by quick freezing.  a Inulin levels are expressed in terms of hexose equivalents (30 eq/mol of inulin). oitro, the increase being similarly 22.4 mmolikg of dry tissue during the 3 min. Tissue glucose had nearly equilibrated with the perfusate glucose within the short period chosen here, as indicated by results of a few measurements performed on islets sampled after 15 min of exposure to 10 mM glucose (protocol 3). In islets from two normal animals the absolute glucose levels were 31.2 * 3.2 and 34.8 & 3.3 mm/kg of dry tissue and from two untreated diabetics they were 26.9 * 1.9 and 21.5 + 2.1 mmol/kg of dry tissue, all within the range found after brief (3 min) exposure to glucose (compare with Tables II and III).
Islet tissue dissected from the pancreas following perfusion in the absence of extracellular glucose contained substantial amounts of glucose, confirming the results of a previous study from this laboratory (17). The levels found here ranged from 4.4 mM in islets of controls to 9.4 mM in islets from untreated diabetics receiving insulin treatment in uitro. That these levels are about twice as high as were previously reported (17) may be due to the nutritional state and the shorter duration of perfusion (fed uerws fasted and 23 uersus 75 min). Because of the specificity of the assay it is reasonable to assume that glucose is being measured. It is less likely that fructose release from inulin or glucose released from dextran can explain the data since this tissue glucose blank lacks proportionality to the inulin levels of the islets (compare the results in controls with the results of islets following in vitro treatment with high insulin) (Table II). Further, it was found that neither inulin nor dextran carried through the assay procedures resulted in TPNH formation as measured here. It is also unlikely that this glucose pool is derived from glycogen since dilute acid fails to hydrolyze the cu-glucosidic bond (11,12).
In order to allow some insight into the tissue distribution of glucose, the inulin spaces of islets were determined. When the pancreas was perfused with 0.33 mM inulin and in the absence of glucose, the extracellular space was remarkably constant no matter what the condition, i.e. 0.87, 0.91, 0.91, and 0.90 liters of water/kg of dry tissue in controls, untreated diabetics, treated diabetics, and untreated diabetics plus insulin in uitro, respectively. The extracellular space measured here is about 1.5 times larger than found in uiuo. The most plausible explanation is that in the perfusion situation this compartment is expanded as a result of the artificial circulatory conditions. Surprisingly, it was found that in the presence of 10 mM glucose in the perfusate the inulin spaces of diabetic islets was reduced following pharmacological doses of insulin in vitro (from 1.11 to 0.67 liters/kg of dry tissue, i.e. -38%, Table III). There is currently no obvious biochemical correlate that might explain the phenomenon.
Using these data on inulin distribution it can be calculated that the intracellular glucose space of controls is about two-thirds of an assumed total water space of 3 liters/kg dry tissue (17), whereas the intracellular glucose space of the islets from untreated diabetics contributes about one-third of the total water space. Brief insulin treatment in vitro with 1 rg/ml of insulin increased the intracellular glucose space from 1.0 to 1.5 liter/kg (Table III). This change does not attain statistical significance, however. The level of free intracellular glucose is, within experimental error, the same in the islets from all three diabetic groups, with the lowest levels at 5 and the highest level of 6.6 mM. Similarly, in a previous study (17) insulin, when given in uiuo or in vitro had no effect on the intracellular glucose level.
Lactate Levels of Islet Tissue (  The values of this table are derived from the results represented in solids).
Here also means f S.E. are recorded. The inulin spaces repre- Table II. Free glucose represents the difference between the glucose sent the quotient of inulin tissue values in terms of rn~ and of 0.35 values found in the presence and absence of 10 mM glucose in the per-mM, the concentration of inulin in the perfusate water. The intracellufusate, calculated for each individual pancreas. The means + S.E. of lar (I.C.) glucose space represents the difference between total gluthe differences are given. Inulin represents the means * S.E. of the case space and inulin (or extracellular) space. The total glucose space values obtained in the presence of 10 mM glucose (fructose equivalents of normal islets is of comparable magnitude as the assumed total water divided by 30). The total glucose space was obtained by dividing the space (about 3 liters/kg of dry tissue). Also given is a statistical individual values representing the free glucose levels by 10.5 (i.e  groups, whether glucose was present in the perfusate or not (Table IV). The values ranged from about 6.5 mM in the islets of control pancreas perfused without glucose to 9.7 mM in islets of the pancreas of untreated diabetics perfused without glucose but containing high insulin (1 fig/ml). There was substantial scatter from animal to animal. These high values are consistent with the extremely high serum lactates observed at the time of killing. Further, expressed on a wet weight basis, the values are within the range of tissue lactate concentrations measured in many other tissues i.e. between 1.5 to 2.5 mM. It would seem from this that under certain experimental circumstances lactate is a poor indicator of glycolysis (26).

ATP
Content of Normal and Diabetic Islets (Table   IL')-Consistent with previous results of ATP measurements in islets of normal (13, 17) and diabetic pancreas (17) it was found that for short term perfusions as used here (23 min) the energy supply was in the physiological range. These results in the perfused pancreas differ from the results obtained with isolated islets obtained either by freehand dissection or by collagenase digestion (27). In the latter cases it was found that ATP levels had dropped substantially in the course of islet isolation and that high glucose was needed to normalize ATP concentration. This high degree of ATP preservation is certainly a distinguishing feature of the perfused pancreas system. Exposure to 10 mM glucose from 3 min in addition to the amino acid mixture has no significant effect on ATP levels under any circumstances (Table IV). It cannot be overlooked that with insulin treatment in vivo and in vitro the ATP values of islets from diabetic animals are significantly higher, whether glucose is added to the perfusate or not. The ATP increment is most pronounced after perfusion with high insulin for 23 min (+22%, p 5 0.01). A similar increase of ATP in a-cell islets due to insulin has been observed previously in the isolated perfused pancreas (17), the change being +24% but lacking statistical significance. These data indicate that insulin is capable of influencing the phosphate potential of o-cells. Levels of Cyclic AMP in Islet Tissue of Perfused Pancreas (Table   V&The recent development of a highly sensitive radioimmunoassay (19,28) allowed us to measure cyclic AMP in islet tissue and the corresponding exocrine samples. These measurements revealed a striking histochemical heterogeneity of the cyclic AMP system of pancreatic tissue: the cyclic AMP levels of islets were 5 to 7 times higher than the levels of exocrine tissue. The difference is so pronounced that for the pancreas the cyclic nucleotide can be considered an islet tissue marker, again attesting to the reliability of the dissection procedure used here. The levels found in the perfusion setting are similar to the levels reported previously with isolated perifused pancreatic islets (29). In the quoted study, cyclic AMP levels between 1.5 to 4 pmol/lOO islets were observed. Because the average dry weight of islets is 1.3 rg (30), this is equivalent to 11.5 to 30 pmol/mg of dry weight. Perfusion with glucose for 3 min did not alter the cyclic AMP level under any circumstances. Since only one time point was analyzed here no statement can be made regarding whether or not the cyclic AMP system is involved in glucose potentiation or glucose suppression of amino acid-induced insulin and glucagon release, respectively. However, the data presented clearly demonstrate the feasibility of studying dynamics of the cyclic AMP system of pancreatic islets in the isolated perfused pancreas, thus avoiding possible complications that might arise from using isolated islets obtained by the collagenase procedure.

Reduced
Glucose Suppressibility of a-Cells in Diabetes-In order to help elucidate how glucagon secretion from a-cells is regulated physiologically and how this regulation might be impaired in diabetes, studies with the isolated perfused pancreas system have been performed by several investigators using the pancreas from diabetic animals (9,24,25). Even though the forms of diabetes differed markedly, it was shown in    In each experiment 5 to 7 islets were pooled for preparing the acid extract (see "Methods"). The numbers of experiments entering the analyses are given in parentheses.

Condition
Islets Acinar tissue pmollmg dry tissue No glucose Controls Untreated Diabetics Treated Diabetics Glucose, 10 mM 23.5 * 8.8 (7) 3.3 i 0.4 (8)" 31.9 + 6.4 (6) 4.9 * 1.2 (7)" 23.2 zt 3.6 (10) 4.0 * 0.4 (9)" Controls Untreated Diabetics Treated Diabetics 23. 5 + 4.0 (8) 3.6 + 0.5 (7)" 30.8 zt 6.9 (7) 5.2 + 0.8 (8)" 28.6 * 9.1 (8) 4.4 + 0.5 (8) In the preceding study of this series (9) preperfusion of the pancreas even with very low levels of glucose (5 mM In islet samples dissected here from the normal pancreas, P-cells contribute most of the tissue, e.g. 75%. They are considered here as @cell islets. In islets from streptozotocin-treated rats, b-cells have been virtually eliminated (9,17,35) but the relative contribution of al-and a,-cells, and non-endocrine cells is not established. However, it seems to be a fair approximation to consider these islets as a-cell islets since this cell type contributes probably as much as 75% of the remaining tissue mass. It has been postulated that the glucose metabolism of a-cells is impaired in diabetes, that this leads to a fuel shortage of the and metabolism as well as the energy potential of the o-cell might be entirely insulin-independent (9,17). Because of the present data this latter view needs to be modified. The capacity of high insulin