Time series prediction of plasma hormone concentration. Evidence for differences in predictability of parathyroid hormone secretion between osteoporotic patients and normal controls.

Recent evidence links osteoporosis, a disease of bone remodeling, to changes in the dynamics of parathyroid hormone secretion. We use nonlinear and linear time series prediction to characterize the secretory dynamics of parathyroid hormone in both healthy human subjects and patients with osteoporosis. Osteoporotic patients appear to lack the periods of high predictability found in normal humans. Our results may provide an explanation for why an intermittent administration of parathyroid hormone is effective in restoring bone mass in osteoporotic patients.

Mechanisms of cellular information processing.

Living cells in multicellular organisms are in simultaneous contact with many regulatory factors such as hormones or neurotransmitters. Many of these factors vary with time in their local concentrations, owing to pulsatile release or production. Therefore, complex patterns of signaling factors act on each living cell in vivo, stimulating or inhibiting second-messenger pathways with potentially complex dynamics. These intracellular pathways do not operate independently but are extensively interconnected, creating complex networks and patterns of intracellular signals that combine to determine the cell's response. The potential significance of cross-signaling between second-messenger pathways and of dynamic stimulation of receptors for cellular information processing in physiology and pathophysiology are discussed.

Pulsatile patterns in hormone secretion.

Endocrine systems are regulated dynamically. With the development of sensitive methods for hormone measurements and high-frequency blood sampling, it has been shown in many endocrine systems that hormonal information is encoded in distinct pulses varying in frequency from minutes to hours. Focusing on pituitary hormones as an example, this review discusses the relevance of this pulsatile pattern of secretion on the regulation of endocrine systems and its implications on diagnosis and therapy o f endocrine diseases.

Pulsatile hormone secretion: Analysis and biological significance.

Pulsatile hormone release is a general phenomenon that can be observed in numerous endocrine systems. The analysis and biological significance of pulsatile hormone release were discussed at a Ferring satellite symposium of the Third European Congress of Endocrinology, held on July 23 and 24, 1994, in Hannover, Germany.

Ca2+/calmodulin inhibition and phospholipase C-linked Ca2+ Signaling in clonal beta-cells.

Neurotransmitters and hormones, such as arginine vasopressin (AVP) and bombesin, evoke frequency-modulated repetitive Ca2+ transients in insulin-secreting HIT-T15 cells by binding to receptors linked to phospholipase C (PLC). The role of calmodulin (CaM)-dependent mechanisms in the generation of PLC-linked Ca2+ transients was investigated by use of the naphthalenesulfonamide CaM antagonists W-7 and W-13 and their dechlorinated control analogs W-5 and W-12. W-7 (10-30 microM) and W-13 (30-100 microM), but not W-5 (100 microM) and W-12 (300 microM), reversibly inhibited the AVP- and bombesin-induced Ca2+ transients. As the generation of PLC-linked Ca2+ transients requires mobilization of internal Ca2+ and Ca2+ influx through voltage-sensitive (VSCC) and -insensitive (VICC) Ca2+ channels, the effects of the W compounds on these processes were further investigated. First, W-7 dose dependently diminished K+ (45 mM)-induced Ca2+ signals (IC50, approximately 25 microM), and W-13 (100 microM) reduced the K+ (45 mM)-induced [Ca2+]i rise by about 40-60%, whereas W-5 (100 microM) and W-12 (300 microM) had no effect. In addition, W-7 (100 microM) inhibited whole cell Ca2+ currents in mouse beta-cells by about 60%. Second, pretreatment of cells (5 min) with W-7 (30 microM), but not W-5 (30 microM), inhibited agonist-induced internal Ca2+ mobilization by about 75% in Ca2+-free medium. Neither W-7 (30 microM) nor W-5 (30 microM) affected AVP (100 nM)-stimulated formation of IP3. Third, capacitative Ca2+ influx through VICC activated by thapsigargin (2 microM) in the presence of verapamil (50 microM) was inhibited by W-7 (30 microM) but not by W-5 (30 microM). As all of the W compound effects corresponded well to their reported anticalmodulin activity, a specific anticalmodulin action can be assumed. Thus, Ca2+ via activation of CaM-dependent processes could provide positive feedback on the generation of PLC-linked Ca2+ transients in HIT-T15 cells. This appears to involve CaM-dependent regulation of both mobilization of internal Ca2+ and Ca2+ influx through VSCC and VICC.

Hypothalamic regulation of pulsatile thyrotopin secretion.

To determine the mechanism underlying pulsatile TSH secretion, 24-h serum TSH levels were measured in three groups of five healthy volunteers by sampling blood every 10 min. The influence of an 8-h infusion of dopamine (200 mg), somatostatin (500 micrograms), or nifedipine (5 mg) on the pulsatile release of TSH was tested using a cross-over design. The amount of TSH released per pulse was significantly lowered by these drugs, resulting in significantly decreased mean basal TSH serum levels. However, pulses of TSH were still detectable at all times. The TSH response to TRH (200 micrograms) tested in separate experiments was significantly lowered after 3 h of nifedipine infusion compared to the saline control value. Nifedipine treatment did not alter basal, pulsatile, or TRH-stimulated PRL secretion. The persistence of TSH pulses under dopamine and somatostatin treatment and the blunted TSH responses to nifedipine infusion support the hypothesis that pulsatile TSH secretion is under the control of hypothalamic TRH. The 24-h TSH secretion pattern achieved under stimulation with exogenous TRH in two patients with hypothalamic destruction through surgical removal of a craniopharyngioma provided further circumstantial evidence for this assumption. No TSH pulses and low basal TSH secretion were observed under basal conditions (1700-2400 h), whereas subsequent repetitive TRH challenge (25 micrograms/2 h to 50 micrograms/1 h) led to a pulsatile release of TSH with fusion of TSH pulses, resulting in a TSH secretion pattern strikingly similar to the circadian variation. These data suggest that pulsatile and circadian TSH secretions are predominantly controlled by TRH.

Pulsatile secretion of thyrotropin during fasting: a decrease of thyrotropin pulse amplitude.

The effect of fasting on circadian and pulsatile TSH secretion was investigated in eight healthy subjects (four men and four women in the follicular phase). Each subject was studied twice, once during 24 h with normal food intake and once during the last 24 h of a 60-h fast. Blood was sampled every 10 min during 24 h for measurement of TSH by a sensitive immunoradiometric assay. Fasting induced a decrease in plasma T3 [1.73 +/- 0.06 vs. 1.36 +/- 0.04 nmol/L; P less than 0.01 (mean +/- SE), control period vs. fasting] and thyroglobulin (52 +/- 8 vs. 35 +/- 7 pmol/L; P less than 0.001) and an increase in plasma rT3 (0.30 +/- 0.06 vs. 0.44 +/- 0.09 nmol/L; P less than 0.02). Plasma T4, thyroid hormone binding index, and free T4 were not statistically different in both periods. The mean plasma 24-h TSH concentration was lower during fasting than in the control period (2.0 +/- 0.3 vs. 1.0 +/- 0.2 mU/L; P less than 0.005). This was associated with a decrease in mean TSH pulse amplitude during fasting (Desade program: 0.6 +/- 0.1 vs. 0.3 +/- 0.1 mU/L; P less than 0.01; Cluster program: 0.5 +/- 0.1 vs. 0.2 +/- 0.1 mU/L; P less than 0.05), whereas TSH pulse frequency during fasting was unchanged (Desade program: 8.4 +/- 0.9 vs. 9.8 +/- 0.8 pulses/24 h; Cluster program: 9.5 +/- 0.5 vs. 7.9 +/- 0.9 pulses/24 h). There was a highly significant correlation between the mean 24-h TSH concentration and the mean TSH pulse amplitude during both the control period and fasting. Although the decrease in TSH concentration during fasting was evident over 24 h, fasting especially decreased the absolute (1.3 +/- 0.3 vs. 0.4 +/- 0.1 mU/L, P less than 0.02) and the relative (101 +/- 18% vs. 40 +/- 14%; P less than 0.02) nocturnal TSH surge (mean TSH 0000-0400 h vs. mean TSH 1500-1900 h). The decreased nocturnal TSH surge during fasting was associated with a significantly decreased TSH pulse amplitude, but with an unaltered number of TSH pulses between 2000-0400 h. In conclusion, fasting decreases 24-h TSH secretion and the nocturnal TSH surge in the absence of a change in plasma T4 concentration. This is associated with a decreased TSH pulse amplitude, whereas TSH pulse frequency remains unchanged.

Temporal pattern of pancreatic insulin and C-peptide secretion and of plasma glucose levels after nutritional stimulation.

The dependency of the secretory pattern of insulin and C-peptide on either oral ingestion of the energy substrates glucose and protein or gastric distension was determined in nine healthy male subjects. To analyze secretion dynamics, high frequency blood sampling, computed estimation of individual hormone half-lives, deconvolution of data, and pulse analysis of the deconvoluted data by the Cluster program were used. After stimulation with oral glucose and protein, baseline insulin, C-peptide, and glucose levels increased in parallel, forming two or three large increases (macropulses), with a mean duration of 63.8 min. The frequency of high frequency insulin and C-peptide pulses was unchanged, whereas a significantly increased amplitude formed the basis of insulin/C-peptide macropulses after both oral stimulations. No changes in baseline insulin/C-peptide concentrations or in amplitude or frequency were observed after a challenge with 400 mL H2O (n = 3). Gastric distension with an equal volume of H2O (400 mL) did not influence pancreatic hormone secretion. Insulin and C-peptide secretions were pulsatile, with a frequency of approximately one pulse per 12 min correlated to C-peptide pulses. When calculated by multiple regression analysis glucose, insulin and C-peptide plasma levels increased simultaneously after the challenge with either glucose or protein, suggesting a neuronal or humoral intestinal-pancreatic regulation of pancreatic hormone secretion. These findings suggest that high frequency insulin and C-peptide pulses form the basis of insulin and C-peptide plasma levels after meal stimulation.

Physiological regulation of circadian and pulsatile thyrotropin secretion in normal man and woman.

The circadian and pulsatile TSH secretion profiles were investigated in 5 females at the time of menstruation and 21 healthy males by sampling blood every 10 min for 24 h. Computer-assisted analysis, i.e. the Cluster and Desade programs, revealed means of 9.9 +/- 1.7 (Cluster) and 11.4 +/- 3.9 (Desade) pulses/24 h. More than 50% of the TSH pulses were detected between 2000-0400 h. Male and female subjects showed no significant difference in the basal mean and pulsatile secretion of TSH or in the TSH response to TRH (200 micrograms). Repetition of the TSH secretion analysis in 4 healthy subjects after 1, 2, and 6 months (2 subjects) revealed a significantly better cross-correlation within than between individuals (P less than 0.0001). We modulate the circadian TSH secretion pattern by acute sleep withdrawal or prolonged sleep after a night of sleep withdrawal in six healthy male volunteers. Sleep withdrawal augmented the nightly TSH secretion (mean serum TSH, 2.1 +/- 1.3 mU/L; mean TSH in sleep, 1.3 +/- 0.5 mU/L; P less than 0.05), whereas sleep after sleep withdrawal almost completely suppressed the circadian variation (mean TSH, 1.1 +/- 0.7 mU/L; P less than 0.01). This modulation is due to a significant decrease in pulse amplitude, but not to an alteration in the frequency or temporal distribution of TSH pulses.

Prediction and significance of the temporal pattern of hormone secretion in disease states.

Comparison of the temporal pattern of hormone secretion in health and disease reveals distinct differences in many systems. Analysis of these visually apparent differences conventionally rests on computer-assisted programs based on either model assumptions, or estimations of hormonal decay rates or threshold values, all of which may not accurately reflect physiological and/or pathophysiological states. Only recently have new methods evolved which are independent of preexisting knowledge of the system under study. Apart from the widely used approximate entropy statistic (ApEn), a measure for the regularity of a time-series, artificial neural networks are able to capture temporal structures in endocrine rhythms without any previous assumptions. In particular, non-linear dynamical systems may be delineated and separated from random behaviour. This is achieved by mapping complex input data to a given complex output by propagating data from the input layer to the output layer through a larger number of interconnections, so-called hidden layers. The networks are capable to extract relevant features from training samples and store this information in the distributed structure of interconnections. Using this approach on growth hormone (GH) rhythms of healthy controls, fasted healthy subjects, untreated acromegalic patients and acromegalics under octreotide suppressive therapy we were recently able to demonstrate the power of this approach to differentiate the temporal pattern of GH secretion following normalization of the data for absolute amplitudes. In a second approach we were able to significantly reduce the number of data points required to characterize the temporal structure of these rhythms. This latter quality of the networks may help to transfer analysis of changes in the temporal pattern of hormone secretion on a more routine basis.

Humoral coding and decoding.

Humoral communication systems are dynamically regulated. Most hormones are released in a pulsatile or burst-like manner into the bloodstream. It is well known that information coded in the frequency and amplitude of secretory pulses allows for the differential regulation of specific target cell function and structure. However, despite intensive study of transmembrane signalling relatively little is known about how the temporal dynamics of extracellular humoral stimuli specifically regulates the temporal pattern of intracellular signalling pathways, such as Ca2+-dependent signalling. Repetitive spikes of Ca2+ encode this information in their amplitude, duration and frequency, and are in turn decoded into the pattern of gene expression and phosphorylation of target proteins. Using a mathematical model for G protein-coupled Ca2+ signalling and information-theoretic approaches to stimulus reconstruction we have systematically quantified the amount of information coded in the Ca2+-signal about the dynamics of the stimulus, which allows us to explore the temporal bandwidth of transmembrane signalling. These in silico approaches permit us to differentiate the amount of information coded in the frequency, temporal precision, amplitude and the complete Ca2+-signal. This may open an avenue to the quantification of information flow and processing in the intra- and intercellular coding and decoding machinery.

Twenty-four-hour rhythms of plasma catecholamines and their relation to cardiovascular parameters in healthy young men.

Diurnal and ultradian rhythms of plasma norepinephrine and epinephrine and their role in the regulation of cardiovascular parameters were investigated over 24 h of recumbency in a group of five men. Catecholamines were measured at 10 min intervals, and blood pressure and heart rate were recorded continuously. Norepinephrine and epinephrine rapidly fluctuated in each subject, with no obvious diurnal rhythm. Spectral analysis suggested two ultradian rhythms with periods of around 12 h and 50-100 min for both catecholamines. The pulse detection programs PULSAR and CLUSTER revealed 20-30 pulses/24 h for norepinephrine and epinephrine, with a significant correlation between the two rhythms (r = 0.63-0.80, P < 0.001). Neither the frequency nor the amplitude of these rapid fluctuations differed between day and night. Arousal in the morning caused a small increase in plasma catecholamines and getting out of bed a large increase. Thus changes in posture and activity are the main influences on the diurnal variations of plasma catecholamines reported previously, while the ultradian rhythms of sympathoadrenomedullary activity appear to be of intrinsic origin. Blood pressure and heart rate exhibited a diurnal rhythm with a nightly decrease. Arousal and rising from bed increased blood pressure and heart rate significantly. Although the amplitude of the rapid fluctuations of plasma catecholamines at times exceeded those caused by postural changes in the morning, when both plasma norepinephrine and epinephrine levels correlated highly with all cardiovascular parameters, correlations were not significant during recumbency. Thus the intrinsic ultradian fluctuations of plasma catecholamines appear not to be involved in the control of cardiovascular parameters during recumbency, and the increase in blood pressure and heart rate in the morning appears to be controlled by direct sympathetic neural input to the heart and vasculature in response to changes in activity and posture rather than by an endogenous surge of plasma catecholamines.

Noise enhanced hormonal signal transduction through intracellular calcium oscillations.

In a wide range of non-linear dynamical systems, noise may enhance the detection of weak deterministic input signals. Here, we demonstrate this phenomenon for transmembrane signaling in a hormonal model system of intracellular Ca(2+) oscillations. Adding Gaussian noise to a subthreshold extracellular pulsatile stimulus increased the sensitivity in the dose-response relation of the Ca(2+) oscillations compared to the same noise signal added as a constant mean level. These findings may have important physiological consequences for the operation of hormonal and other physiological signal transduction systems close to the threshold level.

Coding efficiency and information rates in transmembrane signaling.

A variety of cell types responds to hormonal stimuli by repetitive spikes in the intracellular concentration of calcium ([Ca(2+)](i)) which have been demonstrated to encode information in their frequency, amplitude, and duration. These [Ca(2+)](i)-spike trains are able to specifically regulate distinct cellular functions. Using a mathematical model for receptor-controlled [Ca(2+)](i) oscillations in hepatocytes we investigate the encoding of fluctuating hormonal signals in [Ca(2+)](i)-spike trains. The transmembrane information transfer is quantified by using an information-theoretic reverse-engineering approach which allows to reconstruct the dynamic hormonal stimulus from the [Ca(2+)](i)-spike trains. This approach allows to estimate the accuracy of coding as well as the rate of transmembrane information transfer. We found that up to 87% of the dynamic stimulus information can be encoded in the [Ca(2+)](i)-spike train at a maximum information transfer rate of 1.1 bit per [Ca(2+)](i)-spike. These numerical results for humoral information transfer are in the same order as in a number of sensory neuronal systems despite several orders of magnitude different time scales of operation suggesting a universal principle of information processing in both biological systems.

Differential coding of humoral stimuli by timing and amplitude of intracellular calcium spike trains.

The ubiquitous Ca2(+)-phosphoinositide pathway transduces extracellular signals to cellular effectors. Using a mathematical model, we simulated intracellular Ca2+ fluctuations in hepatocytes upon humoral stimulation. We estimated the information encoded about random humoral stimuli in these Ca2+ spike trains using an information-theoretic approach based on stimulus estimation methods. We demonstrate accurate transfer of information about random humoral signals with low temporal cutoff frequencies. In contrast, our results suggest that high-frequency stimuli are poorly transduced by the transmembrane machinery. We found that humoral signals are encoded in both the timing and amplitude of intracellular Ca2+ spikes. The information transmitted per spike is similar to that of sensory neuronal systems, in spite of several orders of magnitude difference in firing rate.

Precision of intracellular calcium spike timing in primary rat hepatocytes.

Extracellular stimuli are often encoded in the frequency, amplitude and duration of spikes in the intracellular concentration of calcium ([Ca2+]i). However, the timing of individual [Ca2+]i-spikes in relation to the dynamics of an extracellular stimulus is still an open question. To address this question, we use a systems biology approach combining experimental and theoretical methods. Using computer simulations, we predict that more naturalistic pulsed stimuli generate precisely-timed [Ca2+]i-spikes in contrast to the application of constant stimuli of the same dose. These computational results are confirmed experimentally in single primary rat hepatocytes upon alpha1-adrenergic stimulation. Hormonal signalling in analogy to neuronal signalling thus has the potential to make use of temporal coding on the level of single cells. The [Ca2+]i-signalling cascade provides a first example for increasing the information capacity of an intracellular regulatory signal beyond the known coding mechanisms of amplitude (AM) and frequency modulation (FM).

Neural networks in the analysis of episodic growth hormone release.

Pulsatile secretion of growth hormone (GH) has been observed in healthy controls as well as acromegalic patients. In healthy adults, highly regulated secretory pulses of GH occur 4-8 times within 24 h. This episodic pattern of secretion seems to be related to the optimal induction of physiological effects at the peripheral level. In contrast to normal subjects, acromegalic patients demonstrate an irregular pattern of excessive GH release. This pattern of secretion is responsible for many systemic effects, such as the stimulation of connective tissue growth, cardiovascular and cerebrovascular disease, diabetes mellitus and arthritis. Standard methods for the analysis of pulsatile patterns of hormone secretion did not consistently separate the temporal dynamics of GH release in healthy controls and acromegalic patients under various study conditions. Using the cutting edge technology of artificial neural networks for time series prediction, we were able to achieve significant separation of both groups under various conditions by means of the predictability of their GH secretory dynamics. Improving the predictive results by using a more refined system of multiple neural networks acting in parallel (adaptive mixtures of local experts), we found that this system performed a self-organized segmentation of hormone pulsatility. It separated phases of secretory bursts and quiescence without any prior knowledge of the form of a GH pulse or a model of secretion. Comparing the predictive results for the GH dynamics with those for computer-stimulated stochastic processes, we were able to define the irregular pattern of GH secretion in acromegaly as a random autonomous process. The introduction of neural networks to the analysis of dynamic endocrine systems might help to expand the existing analytical approaches beyond counting frequency and amplitude of hormone pulses.

Algorithmic complexity of growth hormone release in humans.

Most hormones are secreted in an pulsatile rather than in a constant manner. This temporal pattern of pulsatile hormone release plays an important role in the regulation of cellular function and structure. In healthy humans growth hormone (GH) secretion is characterized by distinct pulses whereas patients bearing a GH producing tumor accompanied with excessive secretion (acromegaly) exhibit a highly irregular pattern of GH release. It has been hypothesized that this highly disorderly pattern of GH release in acromegaly arises from random events in the GH-producing tumor under decreased normal control of GH secretion. Using a context-free grammar complexity measure (algorithmic complexity) in conjunction with random surrogate data sets we demonstrate that the temporal pattern of GH release in acromegaly is not significantly different from a variety of stochastic processes. In contrast, normal subjects clearly exhibit deterministic structure in their temporal patterns of GH secretion. Our results support the hypothesis that GH release in acromegaly is due to random events in the GH-producing tumorous cells which might become independent from hypothalamic regulation.

[Pulsatile hormone secretion for control of target organs]. / Die Bedeutung pulsatiler Hormonsekretion für die Steuerung von Zielorganen.

Pulsatile secretion of hormones are observed in many endocrine systems. Here we discuss the significance of pulsatile patterns of hormone secretion for the regulation of endocrine target tissues in physiology and pathophysiology. New approaches to analyze endocrine rhythms are introduced that may enable to better define the temporal patterns of secretion relevant for the regulation of distinct processes in complex in vivo systems. This may lead to improved diagnostic and therapeutic strategies of endocrine diseases.