Hormones are characteristically secreted in distinct pulses. Corticosterone (in rodents) and cortisol (in human) show profound circadian and ultradian rhythmicity which results in rapid changes in ligand concentration in the circulating blood. The pattern of ultradian glucocorticoid levels shows genetic modulation, can change according to the physiological and pathological status of the animal and can be programmed by neonatal events. Since the tissue response to changes in circulating glucocorticoid levels can be very rapid, changes in ultradian rhythm could allow tissue-specific regulation of glucocorticoid signalling without major changes in mean plasma hormone levels.
The utilization of frequency modulation as a mechanism for communication both between and within cells is well reported . The endocrine system is no stranger to the use of frequency encoding, which has been particularly well described for gonadotropin-releasing hormone  and growth hormone . It is also becoming clear that the binding of steroid hormones to nuclear receptors is a far from static process, and ligands, DNA and other proteins interact in a highly dynamic manner [4,5]. We now have evidence that in addition to the cyclic activity of the transcriptional machinery, there is also a cyclic rhythmicity of access of glucocorticoids to cells and their receptors. We have investigated this using an automatic blood sampling system in free-running animals.
In female rats, there is a complex ultradian rhythm of corticosterone secretion of approx. 1 h duration, and it is modulation of the amplitude of these pulses that generates the well-known circadian profile of the hormone . An important feature of this glucocorticoid ultradian rhythm both in human and in the rat is the rapid decline of the hormone levels after each pulse. In the rat, this decline in corticosterone parallels the known half-life of approx. 10 min [6,7]. This implies a totally non-secretory phase after each endogenous pulse of corticosterone secretion. We now have evidence that glucocorticoid secretion is actually inhibited during this post-secretory (or refractory) phase of the cycle. Thus mild stress applied during a secretory phase, when basal hormone levels are rising, results in a robust glucocorticoid response, whereas the same stress applied during the putative inhibitory phase, when basal levels are falling, results in a markedly suppressed glucocorticoid response .
There are clear genetic influences on ultradian rhythmicity and we have been able to demonstrate very marked differences in ultradian activity in different strains of rat (Fischer and Lewis) that differ in their stress-responsiveness and their susceptibility to autoimmune disease . There is also a marked sexual diergism , and furthermore a susceptibility to programming from neonatal exposure to gonadal steroids [10,11]. Interestingly, neonatal stressors are also able to programme ultradian rhythmicity into adulthood as well as modify the HPA (hypothalamic–pituitary–adrenal) responsiveness to stress .
The pattern of HPA activity can also be modified by the physiological state. Thus lactation, a time of stress hypo-responsiveness [13,14], results in a loss of the diurnal variability in ultradian secretion of corticosterone, while chronic inflammation in an animal model of inflammatory arthritis  results in an approximate doubling of pulse-frequency. In humans, there are also characteristic changes in ultradian rhythm during disease [16,17].
These results put us in a unique position to investigate whether it is the pattern – rather than the absolute levels – of corticosterone that mediate feedback and the adaptive and mal-adaptive response to increased HPA activity. Glucocorticoid feedback inhibition of the HPA axis activity can involve rapid and probably non-genomic responses . Indeed, in pilot studies, we have shown that administration of exogenouos corticosteroid inhibits HPA activity at a hypothalamic level within 20 min (M.A. Andrews, R.J. Windle, S.A. Wood, H.C. Atkinson, C.D. Ingram and S.L. Lightman, unpublished work). These effects are rapid enough to function within the time-span of ultradian rhythms and terminate the active phase of the glucocorticoid pulse; whether such a mechanism could directly affect the pulse generated is unknown.
At the tissue level, Kitchener et al.  have already shown that different brain areas show differential nuclear translocation and DNA binding of glucocorticoid receptors during stress and the circadian cycle. We are now developing an in vivo model in which we can alter HPA pulse characteristics at will and are using this to investigate tissue-specific responses to different patterns of glucocorticoid infusion (C. Wiles, B. Conway-Campbell, M. McKenna, H.C. Atkinson, S.A. Wood, F. Spriga and S.L. Lightman, unpublished work). Preliminary studies suggest remarkably rapid responses to individual pulses and confirm and extend the tissue specificity suggested by Kitchener et al. .
Nuclear Receptors: Structure, Mechanisms and Therapeutic Targets: A Focus Topic at BioScience 2006, held at SECC Glasgow, U.K., 23–27 July 2006. Edited by C. Bevan (Imperial College London, U.K.), D. Black (Organon, U.K.) and I. McEwan (Aberdeen, U.K.).