Indeed, although both cortisol and aldosterone levels increased during the morning hours, the ratio between aldosterone and cortisol was much higher during the early night, when the effects of spironolactone on T cell counts were apparent. This tempts to speculate that rather than MR activation per se, the balance between MR and GR activation is more crucial for the regulation of T cell migration. On the other hand, the effect of spironolactone fading in the morning hours can be taken to

exclude that MR signaling is involved in the prominent circadian decline in T cell numbers at that time. This decline in T cells was paralleled not only by an increase in cortisol but also in CXCR4 expression, i.e.,

a pattern in line with the view derived from previous studies small molecule library screening that cortisol via activation of GR induces CXCR4 expression which in turn accelerates the migration of T cells, presumably into the bone marrow (Dimitrov et al., 2009, Fauci, 1975 and Okutsu et al., 2005). GR and MR can form heterodimers thereby increasing the functional diversity of these receptors (Liu et al., 1995, Nishi et al., 2004 and Trapp et al., 1994). The fact that spironolactone did neither affect CXCR4 expression nor the decrease in blood ATR inhibition T cell counts in the morning shows that this pattern is GR driven, and does not require concomitant activation of MR. Of note, in the absence of the enzyme 11β-hydroxysteroid dehydrogenase 2 cortisol binds MR with even higher affinity than GR (Krozowski et al., 1999, Rupprecht et al., 1993 and Zhang et al., 2005). Estimates from animal studies indicate that during the circadian nadir of glucocorticoid

levels about 50 per cent of MR are occupied by endogenous AMP deaminase corticosteroids (Kalman and Spencer, 2002). Therefore, the increasing effect of spironolactone on naïve T cell counts might basically stem also from a blockade of low cortisol levels acting on the MR. However, in humans, there is evidence for a threefold higher affinity of lymphocytic MR for aldosterone than cortisol (Armanini et al., 1985), making it unlikely that cortisol substantially contributes to MR mediated T cell trafficking during early nocturnal sleep. Also, an unspecific mediation of the effects via non-lymphocytic MR seems unlikely, as the effect was cell-subset specific, with no impact of spironolactone on CD62L− T cells, and we did not observe any effects on blood pressure or sleep architecture, nor did the subjects report any side effects. Though unlikely, it cannot be fully ruled out that non-MR mediated effects of spironolactone, like a down-regulation of IL-2 production (Sonder et al., 2006), added to the observed increase in circulating T helper cells. Testing with more specific MR antagonists or agonists might help to resolve this issue in future studies.