Primary In Vivo Oscillations of Metabolism in the Pancreas

Reagents of analytical grade and water purified by a Milli-Q filter (Millipore, Bedford, MA) were used. Tetramethylbenzidine and insulin peroxidase were purchased from Sigma (St. Louis, MO); aprotinin (Trasylol), from Bayer (Leverkusen, Germany); and heparin, from Leo/Løvens (Ballerup, Denmark). The rat insulin standard was obtained from Novo Nordisk (Bagsvaerd, Denmark). IgG-certified microtiter plates were purchased from Nunc (Roskilde, Denmark). The antibodies to insulin were raised in guinea pigs in our laboratory.

Male SD rats (n = 7) weighing 300–350 g were obtained from a local breeding colony (Biomedical Center, Uppsala, Sweden). The animals had free access to tap water and pelleted food at all times. The experiments were approved by the local animal ethics committee at Uppsala University.

The animals were anesthetized with an intraperitoneal injection of thiobutabarbital sodium (Inactin; Research Biochemicals International, Natick, MA; 120 mg/kg body wt), placed on a heated operating table maintained at body temperature (38°C), heparinized, and tracheostomized. Polyethylene catheters were inserted into the right femoral artery and the right femoral vein, respectively, to monitor the mean arterial blood pressure with a Statham P 23dB pressure transducer (Statham Laboratories, Los Angeles, CA) and to infuse Ringer solution (5 ml · h⁻¹ · kg⁻¹) to substitute for body fluid loss. Next the pancreas of each animal was exposed through a midline abdominal incision and a catheter was placed in the portal vein by inserting a sharpened cannula, connected to the catheter, through the vascular wall and in the same direction as the blood flow. The cannula was loosely placed and no ligatures were used that could affect portal blood flow. Only 40 μl blood/min were diverted into the sampling catheter, with the rest of the portal blood being drained, as usual, into the liver. The catheter was connected to a peristaltic pump (Ismatec Reglo 4/12; Glattburg-Zürich, Switzerland), which allowed continuous sampling of portal blood. The pancreas was immobilized and superficial islets were visualized in situ by intravenous injection of 0.8-ml sterile-filtered 2% (wt/vol) neutral red (Kebo Grave AB, Stockholm, Sweden) (22). The staining procedure made it possible to insert one oxygen tension electrode into a superficially located islet and one into the adjacent exocrine parenchyma by the use of micromanipulators under a stereo microscope (23). The electrodes were placed ∼10 mm apart. The injection with neutral red does not affect blood concentrations of glucose or insulin, blood flow of the whole pancreas or islet, or pancreatic oxygen tension measurements (22). The animals were allowed to rest for at least 30 min after the surgical preparation before recordings of oxygen tension and portal blood sampling were begun. Blood glucose concentration was determined in the beginning and end of each experiment with test reagent strips (Medisense; Baxter Travenol, Deerfield, IL). Oxygen tension recordings and blood samples from the portal vein were collected continuously (40 μl/min) for ∼30 min. At the end of the experiment, blood was also collected from the femoral vein for determination of the hematocrit.

The oxygen tension electrodes were modified Clark microelectrodes (24,25), with an outer tip diameter of 2–6 μm and an inner tip diameter of 1–2 μm. The electrodes were polarized at −0.8 V, which gave a linear response between the oxygen tension and the electrode current. The electrical current was measured by picoamperemeters (University of Aarhus, Aarhus, Denmark). The electrodes were calibrated in water saturated with Na₂S₂O₅ or air at 37°C before and after the experiments. The drift of the microelectrode recordings was less than 0.5% per hour. The data were acquired at 4 Hz and processed by a MacLab Instrument (AD Instruments, Hastings, UK). In figures displaying pO₂ recordings and blood insulin measurements, averages of pO₂ data corresponding to 30 s were calculated and used for further data presentations.

Simultaneous with the oxygen tension recordings, portal blood samples (20 μl) were collected every 30 s in tubes containing heparin (25 IU/ml) and aprotinin (Trasylol, 500 kIU/ml). Plasma was immediately separated and frozen until assayed for insulin content. Determinations of insulin concentrations were made by a competitive enzyme-linked immunosorbent assay with the insulin-capturing antibody immobilized directly in the solid phase (11). Amounts of insulin were calculated from linear standard curves in semilogarithmic plots. The method had inter- and intra-assay variations of <10% in the specified range (11).

In separate experiments (n = 3), the passage time for insulin in the pancreas was estimated by measuring the time for substances to pass the splanchnic bed and reach the sampling point in the portal vein. This was performed by injection of 1 ml 30% (wt/vol) d-glucose into the ascending aorta. Simultaneously, blood was withdrawn from the portal vein at the same rate as above (40 μl/min). Glucose concentrations were determined in the withdrawn portal blood samples at 0, 10, and 20 s after the injection.

The occurrence and frequency of regular oscillations in the oxygen tension and plasma insulin recordings were assessed from power spectra obtained by Fourier transformation of the original data. Analyses of the frequencies were performed with Igor software (Wave Metrics, Lake Oswego, OR). The passage times of insulin through the pancreas as well as from the portal vein to the sampling vial were determined to be <10 and <75 s, respectively. These values were taken into account when plotting the data. Results are presented as means ± SE. Statistical evaluation was performed with Student’s paired t test.

In vivo pO₂ of the rat endocrine and exocrine pancreas was monitored simultaneously with determinations of portal blood insulin concentrations in seven animals. The average islet pO₂ was 39 ± 5 mmHg at the beginning of the experiment and remained stable throughout the experiment. Simultaneous measurements of pO₂ in adjacent exocrine pancreatic parenchyma were significantly lower (16 ± 6 mmHg; P < 0.05) throughout the experiment. Both the endocrine and exocrine pO₂ recordings showed a complex pattern of slow (0.21 ± 0.03 and 0.14 ± 0.02 min⁻¹, respectively) and superimposed rapid (3.1 ± 0.3 and 2.7 ± 0.2 min⁻¹, respectively) oscillatory activities (Fig. 1).

The portal plasma insulin concentration was oscillatory with a frequency of 0.22 ± 0.02 min⁻¹ and had an average value of 5.0 ± 0.6 nmol/l (Fig. 2). When analyzing portal insulin determinations and corresponding endocrine measurements of pO₂ for oscillatory activity, averages of data points corresponding to 30 s of pO₂ recordings were used. The frequency observed in portal insulin measurements was almost identical to that identified in the islet pO₂ measurements.

The blood glucose concentration of the portal blood at the start of these experiments was 6.2 ± 0.1 mmol/l and did not differ from that at the end of the experiments (6.6 ± 0.2 mmol/l). Similarly, no change was observed when the mean arterial pressures at the start (116 ± 5 mmHg) and end (110 ± 6 mmHg) of the experiments were compared. The hematocrit was normal (46 ± 1%) at the end of the experiments.

Portal insulin measurements represent the activity of the whole endocrine pancreas, which consists of islets situated at different distances from the portal sampling point. Synchronous changes in insulin release from islets situated in different parts of the pancreas to the portal sampling point could therefore perturb the pulsatile pattern if there were substantial differences in passage time among islets. To determine the passage time for insulin in the pancreas, d-glucose was injected into the ascending aorta in separate experiments. The glucose concentration was then determined in portal blood samples at 0, 10, and 20 s after the injection (n = 3). We observed that the blood glucose concentration had already risen significantly after 10 s, indicating an intrapancreatic passage time of 10 s or less. In control saline-injected animals (n = 3), portal blood glucose concentrations were not affected (data not shown).

The present in vivo measurements recorded oscillations in islet pO₂ and plasma insulin with similar frequencies, but also recorded rapid oscillatory activity in pO₂, present in the endocrine and exocrine pancreas. The values of pO₂ of the endocrine and exocrine pancreas were similar to previously obtained values (23). The significantly lower pO₂ in the exocrine tissue is likely to have been related to the lower blood flow in this part of the pancreas (22,26). Recordings from the endocrine and exocrine pancreas showed a slow and a superimposed rapid oscillatory activity, with approximate frequencies of 0.2 and 3 min⁻¹, respectively. The slower activity had a frequency similar to that of the oscillations in the portal vein insulin, which matched results obtained when portal plasma insulin was analyzed in a canine model (3), but was about double the frequency previously reported for plasma insulin oscillations in the rat (27) and humans (2). The lower frequency observed in the latter studies was most likely attributable to the peripheral sampling and lower sampling rate, which make pulse detection more difficult, as has been shown in the canine model (3). Indeed, when using a sensitive insulin assay and deconvolution, plasma insulin pulses with a period of 4.7 ± 0.1 min have been detected in peripheral blood samples from humans (4). A correlation between oscillations in pO₂ and insulin has already been described in the isolated islet (21) and in groups of islets (17). From this point of view, the present in vivo finding of a slow oscillatory activity in pO₂ was not unexpected. However, from the perspective that slow oscillations in neither [Ca²⁺]_i nor pO₂ have so far been demonstrated in vivo, the finding is important and represents the first in vivo recording of an oscillatory islet parameter with a frequency corresponding to that of plasma insulin oscillations.

Slow in vitro oscillations in [Ca²⁺]_i (12,18,19,28,29) as well as in metabolism (17,21,30–35), with a frequency similar to that of the plasma insulin oscillations, have been recorded in the isolated islet. The key role of [Ca²⁺]_i (36) and metabolism (37) for exocytosis of insulin and the similar frequency of the slow [Ca²⁺]_i and metabolic oscillations and those in plasma insulin have promoted the notion that regular variations in plasma insulin may reflect either [Ca²⁺]_i or metabolic oscillations of the β-cell, or a combination of the two. More direct support for this concept was obtained by parallel measurements of [Ca²⁺]_i and insulin release (12,18,19,29) and pO₂ and insulin release (17,21) from individual islets. In these studies, slow oscillations in [Ca²⁺]_i and pO₂ were always synchronous with pulses of insulin release. However, when islet [Ca²⁺]_i was nonoscillatory after exposure to low glucose concentrations or depolarizing conditions, the corresponding insulin release was still pulsatile (38,39). In contrast, when pO₂ was measured under the same conditions, oscillations were recorded that were synchronized with pulsatile release of insulin (21). Oscillations in metabolism have been proposed to be the result of spontaneous glycolytic oscillations through allosteric regulation of phosphofructokinase-1-M (30). In support of such a role of the enzyme, impaired plasma insulin oscillations have been observed in individuals with mutations in the gene for the enzyme (40). The present report of in vivo pO₂ oscillations that are synchronous with plasma insulin oscillations further supports a primary role of metabolic oscillations in the induction of plasma insulin oscillations.

Until now, [Ca²⁺]_i and membrane potential were the only islet parameters besides insulin that had been monitored in vivo (41–43). In vivo measurements of [Ca²⁺]_i and membrane potential have demonstrated oscillatory activity that was faster (2–5 min⁻¹) than the frequency of the plasma insulin oscillations. The rapid oscillations in membrane potential consist of “slow waves,” which are periods of depolarization, when accumulations of action potentials (or “bursts”) promote influx of Ca²⁺, interspersed with periods of hyperpolarization when no Ca²⁺ influx is present. These “slow waves” give rise to the rapid oscillations in [Ca²⁺]_i, which have been extensively characterized in the isolated islet (28,44–47). The observed superimposed, rapid oscillatory in vivo activity of pO₂ has shown a similar frequency. Such rapid changes in pO₂ have been previously observed in vitro and have been related to rapid oscillations in [Ca²⁺]_i (34,35). The relation between the rapid oscillatory activity in metabolism and ionic movements is complex; two explanations that have been suggested are that ionic changes drive metabolism by activating mitochondrial dehydrogenases (33,48–50) and that oscillatory metabolic changes precede the ionic movements (31,51). Irrespective of causality, it is reasonable to assume that these rapid ionic and metabolic events have their secretory counterpart in individual islets in vivo, as has been demonstrated in the isolated islet in vitro (19,47,52–56). With the presently found estimated passage time of insulin in the pancreas of <10 s, these rapid secretory events would theoretically be detectable in the portal vein. However, given that no synchronization of the rapid oscillations in electrical activity of different islets from the same pancreas in vivo has been observed (57), it is less plausible that the rapid secretory events are present in plasma insulin. Another consequence of the rapid passage time of the blood through the pancreas is that blood-borne factors could play a role in coordinating the secretory activities of the islets in the pancreas (10,58). It has been suggested that the intrinsic nervous system of the pancreas is responsible for interislet coordination (13–16,59), or that the relaying of membrane potential changes among islets via the exocrine tissue is a possible mechanism for synchronizing the islets in the pancreas (10).

In conclusion, the recorded oscillations in pO₂, with a frequency similar to that of plasma insulin oscillations, support a primary role of metabolic oscillations in the induction of plasma insulin oscillations.

The study was supported by grants from the Swedish Medical Research Council (72X-109, 72X-14019); the Swedish Diabetes Association; the Swedish-American Diabetes Research Program, funded by the Juvenile Diabetes Foundation and the Wallenberg Foundation; the Novo Nordisk Foundation; the Family Ernfors Foundation; the Marcus and Amalia Wallenberg Foundation; the Magnus Bergvalls Foundation; the Swedish Foundation for Strategic Research; the Swedish Society of Medicine; and the Göran Gustafsson Foundation.