Ultradian Oscillations of Insulin Secretion in Humans Chantal Simon and Gabrielle Brandenberger

Ultradian rhythmicity appears to be characteristic of several endocrine systems. As described for other hor- mones, insulin release is a multioscillatory process with rapid pulses of about 10 min and slower ultradian oscillations (50–120 min). The mechanisms underlying the ultradian circhoral oscillations of insulin secretion rate (ISR), which arise in part from a rhythmic amplifi- cation of the rapid pulses, are not fully understood. In humans, included in the same period range is the alter- nation of rapid eye movement (REM) and non-REM (NREM) sleep cycles and the associated opposite oscil- lations in sympathovagal balance. During sleep, the glucose and ISR oscillations were amplified by about 150%, but the REM-NREM sleep cycles did not entrain the glucose and ISR ultradian oscillations. Also, the latter were not related to either the ultradian oscilla- tions in sympathoagal balance, as inferred from spectral analysis of cardiac R-R intervals, or the plasma fluctu- FIG. 1. Plasma glucose concentration and ISR profiles in one represen- ations of glucagon-like peptide-1 (GLP-1), an incretin tative subject studied during continuous enteral nutrition with a hormone known to potentiate glucose-stimulated insu- 10-min blood-sampling procedure. ISR was estimated by deconvolution analysis from plasma C-peptide concentrations. lin. Other rhythmic physiological processes are cur- rently being examined in relation to ultradian insulin as compared with constant administration at the same release. Diabetes 51 (Suppl. 1):S258–S261, 2002 average rate (5,6), suggests that, like the rapid pulses, these ultradian oscillations may have a functional signifi- cance. On the other hand, alterations in the glucose and ISR patterns have been described in patients with type 2 n addition to the rapid insulin pulses that recur diabetes (7,8), and the tight coupling between glucose and every 5–10 min, slow and large ultradian oscillations ISR pulses is altered even in conditions where glucose of insulin secretion with a period range of 50–120 tolerance is only minimally impaired (9). Whether these Imin have been described in both humans and ani- alterations, which occur early in the course of ␤-cell mals. These oscillations are best seen in situations of dysfunction (10), play a pathogenetic role in the onset of insulin stimulation, and have been observed after meal hyperglycemia is yet to be determined. Periodically mod- ingestion (1), during continuous enteral nutrition (2), and ulated signals are thought to induce a larger response by during intravenous glucose infusion (3). They are closely the target cell than constant signals (11). It has been associated with similar oscillations of plasma glucose suggested that oscillatory insulin infusion has a more concentration. Figure 1 shows the insulin secretion rate potent inhibitory effect on hepatic glucose production. (ISR) profile estimated from plasma C-peptide levels as Alternatively, it might enhance glucose uptake by revers- proposed by Eaton (4) and the concomitant glucose profile ing downregulation of insulin receptors or due to its obtained in one subject studied during continuous enteral synchrony with periodic oscillations of intracellular pro- nutrition in an experiment lasting 24 h with blood sam- cesses. Such circhoral oscillations are characteristic of pling at 10-min intervals. In these conditions, the mean numerous intracellular processes, including enzyme activ- amplitude of the oscillations reaches 50% of the mean ities (specifically ATP) or protein synthesis (12). levels for ISR and 20% for glucose. In the postprandial The precise mechanisms that generate the ultradian state, their amplitude is maximal immediately after food insulin oscillations are not fully understood. In particular, ingestion and then decreases progressively (1). The higher it is not established whether the glucose oscillations have hypoglycemic effect of intermittent insulin administration, an active role in the origin of the ISR oscillations or whether the latter are independently generated by an From the Laboratoire des Re´gulations Physiologiques et des Rythmes Bi- intrapancreatic pacemaker. ISR oscillations persist after ologiques chez l’Homme, 67085 Strasbourg Cedex, France. pancreas transplantation (13) or in denervated islet cell Address correspondence and reprint requests to chantal.simon@medecine. autografts (14), which suggests that their origin does not u-strasbg.fr. Accepted for publication 22 June 2001. depend on the central nervous system; also, they do not EEG, electroencephalographic; GLP-1, glucagon-like peptide-1; HF, high seem to involve the counterregulatory hormones (3). frequency; ISR, insulin secretion rate; LF, low frequency; NREM, non-rapid eye movement; REM, rapid eye movement; SWS, slow-wave sleep. Based on a mathematical model, it has been hypothesized The symposium and the publication of this article have been made possible that they could result from instability in the insulin- by an unrestricted educational grant from Servier, Paris. glucose feedback loop (15,16). This nonlinear model, S258 DIABETES, VOL. 51, SUPPLEMENT 1, FEBRUARY 2002 ULTRADIAN OSCILLATIONS OF INSULIN

FIG. 2. Plasma insulin concentration and corresponding ISR profiles in one representative subject studied during continuous enteral nutrition with a 2-min blood-sampling procedure. ISR was estimated by decon- volution analysis from plasma insulin concentrations. which is based on different time-delayed feedback loops (glucose stimulates insulin secretion, insulin stimulates glucose uptake and inhibits hepatic glucose production, and glucose stimulates its own uptake), exhibits self- sustained oscillations when a constant glucose infusion is simulated and is consistent with several experimental data. Yet, it does not account for the existence, in some type 2 diabetic patients (7), of glucose pulses in the absence of any concomitant oscillations. The existence of an intrapancreatic pacemaker cannot be excluded. Chou FIG. 3. Mean (؎SE) time courses of delta wave activity (an index of sleep depth), LF/HF ratio (an index of heart rate variability obtained et al. (17) have identified low-frequency insulin oscilla- from spectral analysis of the R-R intervals), plasma glucose concentra- tions in perifused rat pancreatic islets with a period tions, and ISR during SWS. Individual data were normalized in order to ranging from 50 to 100 min. Ultradian rhythms have also compensate for the individual differences in the duration of the SWS episodes. Data are expressed as percentages of the mean overnight been described for numerous physiological processes— values in each subject. The axis gives the real time of sleep episodes, hormonal release, such as norepinephrine (18), as well as averaged for nine subjects. ISR was estimated by deconvolution anal- behavioral, gastrointestinal, and sleep processes (12)— ysis from plasma C-peptide concentrations. St 2, stage 2. indicating that other control mechanisms, although not exclusive, may be involved. Using spectral analysis of the sleep electroencephalo- Relationship with the rapid oscillations. To analyze graphic (EEG) activity, which provides a dynamic descrip- the relationship between the two rhythmic components of tion of sleep processes, it has been demonstrated that the ISR, plasma insulin was measured with a 2-min blood- delta wave activity (0.5–3.5 Hz), which characterizes the sampling procedure in four subjects studied during con- depth of sleep, oscillates with a similar period (24). tinuous enteral nutrition for 8 h. A deconvolution method We previously established (25) that the oscillations of based on a biexponential disappearance rate of insulin, plasma glucose and ISR are amplified by about 150% assuming half-lives for insulin of 2.8 and 5 min with a during sleep without any modification of their frequency fractional slow component of 28% (19), was used to (Fig. 1). This amplification is not an endogenous circadian quantify the dynamics of insulin secretion. The plasma clock effect since it occurs whatever the time of sleep. To insulin and estimated ISR profiles from one subject are better evaluate the temporal link of the glucose and ISR represented in Fig. 2, which shows that the slow insulin oscillations with the internal structure of sleep and its oscillations are partly due to a slow rhythmic amplification ultradian component, we used spectral analysis of sleep of rapid ISR pulses. Spectral analysis of the deconvoluted EEG. Concomitantly, spectral analysis of cardiac R-R ISR profiles confirmed the presence of two significant intervals over 5-min periods was used to calculate various rhythmic components: one with a mean periodicity of indexes of heart rate variability, thought to reflect the 48–96 min, the other with a mean periodicity of 6–12 min. sympathovagal balance. In particular, we calculated the Altogether, these results indicate that the temporal orga- ratio between the power densities in the low-frequency nization of insulin secretion is analogous to that of lutein- (LF: 0.04–0.15 Hz) to the high-frequency (HF: 0.15–0.50 izing hormone or growth hormone, for which the presence Hz) bands. of both rapid and slow secretory pulses has been demon- To establish the time courses of plasma glucose levels, strated (20,21). ISR estimated from plasma C-peptide, and heart rate Relationship with sleep and sympathovagal activity. variability indexes during the different sleep states, we Beside a circadian rhythm, sleep presents an ultradian selected for each of nine subjects studied during continu- component, which is responsible for the alternation of the ous enteral nutrition, uninterrupted NREM sleep episodes two basic sleep states, non–rapid eye movement (NREM) containing slow-wave sleep (SWS). Figure 3 illustrates the sleep and rapid eye movement (REM) sleep. The REM- mean normalized profiles of sleep pattern obtained in the NREM cycles, which are associated with inverse fluctua- nine subjects, together with the concomitant LF/HF ratio, tions of sympathovagal balance (22,23), have a mean ISR, and plasma glucose profiles. As expected, the oscil- periodicity of 80–20 min close to that reported for ISR. lation of delta wave activity, which reflects sleep deepen-

DIABETES, VOL. 51, SUPPLEMENT 1, FEBRUARY 2002 S259 C. SIMON AND G. BRANDENBERGER

between the GLP-1 fluctuations and the insulin or the glucose oscillations.

CONCLUSION Insulin secretion is characterized by a complex temporal organization with two distinct rhythmic components, rapid pulses being associated with slower ultradian oscil- lations that are best seen when insulin secretion is stimu- lated. Ultradian rhythms with similar periods have been identified for numerous physiological processes with no evident coupling with ISR oscillations, which does not support the view of one common pulse generator. What- ever the mechanisms involved, different data suggest that the ISR circhoral rhythm has functional significance.

ACKNOWLEDGMENTS The authors are indebted to Daniel Joly and Jean Ehrhart for data acquisition. They thank Miche`le Simeoni for experimental and editorial assistance.

FIG. 4. Plasma glucose and ISR oscillations and concomitant plasma GLP-1 concentrations in one representative subject during continuous REFERENCES enteral nutrition. Blood was sampled at 10-min intervals. 1. Simon C, Follenius M, Brandenberger G: Postprandial oscillations of plasma glucose insulin and C-peptide in man. Diabetologia 30:769–773, 1987 ing and lightening, and the oscillation of LF/HF ratio were 2. Simon C, Brandenberger G, Follenius M: Ultradian oscillations of plasma glucose, insulin and C-peptide in man during continuous enteral nutrition. coupled and varied in opposite ways, with a predominant J Clin Endocrinol Metab 64:669–674, 1987 parasympathetic tone during SWS. By contrast, the mean 3. Shapiro ET, Tillil H, Polonsky KS, Fang VS, Rubenstein AH, Van Cauter E: glucose or ISR profiles did not show any significant Oscillations in insulin secretion during constant glucose infusion in normal variation during deep sleep, indicating that the oscillations man: relationship to changes in plasma glucose. J Clin Endocrinol Metab 67:307–314, 1988 in delta wave activity trigger neither ISR nor glucose 4. Eaton RP, Allen RC, Schade DS, Erickson KM, Standefer J: Prehepatic oscillations. Inspection of the individual profiles con- insulin production in man: kinetic analysis using peripheral connecting firmed that the fluctuations of insulin secretion were not peptide behavior. J Clin Endocrinol Metab 51:520–528, 1980 systematically associated with concomitant oscillations in 5. Sturis J, Scheen AJ, Leproult R, Polonsky KS, van Cauter E: 24-hour glucose profiles during continuous or oscillatory insulin infusion: demon- delta wave activity. Conversely, sleep deepening was stration of the functional significance of ultradian insulin oscillations. associated with a valley, a peak, an increase, or a decrease J Clin Invest 95:1464–1471, 1995 in ISR, which explains the smooth aspect of the averaged 6. Lefebvre PJ, Paolisso G, Scheen AJ, Henquin JC: Pulsatility of insulin and curves. glucagon release: physiological significance and pharmacological implica- tions. Diabetologia 30:443–452, 1987 ISR oscillations and glucagon-like peptide-1. Gluca- 7. Simon C, Brandenberger G, Follenius M, Schlienger JL: Alteration in the gon-like peptide-1 (GLP-1) is a peptide hormone secreted temporal organisation of insulin secretion in type 2 (non-insulin-depen- by the intestinal mucosa in response to meal or glucose dent) diabetic patients under continuous enteral nutrition. Diabetologia ingestion (26). It is known to act as an incretin hormone, 34:435–440, 1991 8. Polonsky KS, Given BD, Hirsch LJ, Tillil H, Shapiro ET, Beebe C, Frank BH, potentiating glucose-stimulated insulin release and has Galloway JA, Van Cauter E: Abnormal patterns of insulin secretion in been shown to increase ␤-cell sensitivity to minor ultra- non-insulin-dependent diabetes mellitus. N Engl J Med 318:1231–1239, dian glucose excursions in patients with impaired glucose 1988 tolerance (27). GLP-1 itself is secreted in a pulsatile 9. O’Meara NM, Sturis J, Van Cauter E, Polonsky KS: Lack of control by manner with a high-frequency periodicity (28). In order to glucose of ultradian insulin secretory oscillations in impaired glucose tolerance and in non-insulin-dependent diabetes mellitus. J Clin Invest establish whether GLP-1 secretion also presents ultradian 92:262–271, 1993 fluctuations that could modulate the ISR oscillations, 10. Byrne M, Sturis J, Sobel R, Polonsky K: Elevated plasma glucose 2-hour GLP-1 plasma levels and ISR were determined in six postchallenge predicts defects in beta-cell function. Am J Physiol 270: healthy male subjects (20–26 years) studied during con- E572–E579, 1996 11. Goldbeter A: Cell to Cell Signalling: From Experiments to Theoretical tinuous enteral nutrition with a 10-min blood-sampling Models. London, UK, Academic Press, 1989 procedure. GLP-1 (7–36 amide) was measured by radioim- 12. Lloyd D, Rossi EL: Ultradian Rhythm in Life Processes. An Inquiry into munoassay with an intra-assay variation coefficient of 6% Fundamental Principles of Chronobiology and Psychobiology. London, (Peninsula, San Carlos, CA). Plasma GLP-1, ISR, and UK, Springer-Verlag, 1992 glucose profiles obtained in one subject are illustrated in 13. Kern W, Born J, Fehm HL, Kerner W: Sustained oscillatory insulin secretion after pancreas transplantation. Horm Metab Res 22:644–646, Fig. 4. Contrasting with the regular oscillations observed 1990 for glucose and ISR, the GLP-1 profile exhibited irregular 14. Shapiro ET, Strasser S, Polonsky KS, Alejandro R, Mintz DH: Persistence fluctuations. Pulse analysis revealed the presence of 5.0 Ϯ of oscillatory insulin secretion in denervated islet cell autografts. Trans- 0.8 significant pulses during the 8-h experiment. These plantation 52:574–576, 1991 15. Sturis J, Polonsky KS, Mosekilde E, Van Cauter E: Computer model for pulses had a mean period of 45–70 min depending on the mechanisms underlying ultradian oscillations of insulin and glucose. Am J subject. Their relative amplitude was 22.8 Ϯ 2.1%. Coinci- Physiol 260:E801–E809, 1991 dence analysis did not reveal any systematic association 16. Tolic IM, Mosekilde E, Sturis J: Modeling the insulin-glucose feedback

S260 DIABETES, VOL. 51, SUPPLEMENT 1, FEBRUARY 2002 ULTRADIAN OSCILLATIONS OF INSULIN

system: the significance of pulsatile insulin secretion. J Theor Biol 23. Baharav A, Kotagal S, Gibbons V, Rubin BK, Pratt G, Karin J, Akselrod S: 207:361–375, 2000 Fluctuations in autonomic nervous activity during sleep displayed by 17. Chou HF, Berman N, Ipp E: Evidence for pancreatic pacemaker for insulin power spectrum analysis of heart rate variability. Neurology 45:1183–1187, oscillations in low-frequency range. Am J Physiol 266:R1786–R1791, 1994 1995 18. Levin BE, Rappaport M, Natelson BM: Ultradian variations of plasma 24. Aeschbach D, Borbely AA: All night dynamics of the human sleep EEG: J noradrenaline in humans. Life Sci 25:621–628, 1979 Sleep Res 2:70–81, 1993 19. Song SH, McIntyre SS, Shah H, Veldhuis JD, Hayes PC, Butler PC: Direct 25. Simon C, Brandenberger G, Saini J, Ehrhart J, Follenius M: Slow oscilla- measurement of pulsatile insulin secretion from the portal vein in human tions of plasma glucose and insulin secretion rate are amplified during subjects. J Clin Endocrinol Metab 85:4481–4499, 2000 sleep in humans under continuous enteral nutrition. Sleep 17:333–338, 1994 20. Veldhuis JD, Evans WS, Rogol AD, Drake CR, Thorner MO, Merriam GR, 26. Orskov C: Glucagon-like peptide-1, a new hormone of the entero-insular Johnson ML: Intensified rates of venous sampling unmask the presence of axis. Diabetologia 25:701–711, 1992 spontaneous, high frequency pulsations of luteinizing hormone in man. J Clin Endocrinol Metab 59:96–102, 1984 27. Byrne MM, Gliem K, Wank U, Arnold R, Katschinski M, Polonsky KS, Goke 21. Evans WS, Faria AC, Christiansen E, Ho KY, Weiss J, Rogol AD, Johnson B: Glucagon-like peptide 1 improves the ability of the cell to sense and ML, Blizzard RM, Veldhuis JD, Thorner MO: Impact of intensive venous respond to glucose in subjects with impaired tolerance. Diabetes 47:1259– sampling on characterization of pulsatile GH release. Am J Physiol 1265, 1998 252:E459–E556, 1987 28. Balks HJ, Holst JJ, von zur Muhlen A, Brabant G: Rapid oscillations in 22. Brandenberger G, Ehrhart J, Piquard F, Simon C: Inverse coupling between plasma glucagon-like peptide (GLP-1) in humans: cholinergic control of ultradian oscillations in delta wave activity and heart rate variability during GLP-1 secretion via muscarinic receptors. J Clin Endocrinol Metab sleep. Clin Neurophysiol 112:992–996, 2001 82:786–790, 1997

DIABETES, VOL. 51, SUPPLEMENT 1, FEBRUARY 2002 S261