Introduction
Patients suffering from critical illnesses, which can be evoked by sepsis, major trauma, extensive burn injuries or surgery, typically present with high plasma concentrations of total and even more so of free cortisol [
1,
2]. The increased systemic cortisol availability during critical illness is crucial for survival as it plays a key role in providing essential energy substrates and in regulating the immune and hemodynamic responses necessary for restoring homeostasis [
3‐
5]. Both very high and very low levels of systemic cortisol have been associated with poor outcome, underlining the importance of a thorough understanding of this response [
2].
Hypercortisolism of critical illness has long been assumed to be exclusively brought about by a centrally activated hypothalamic–pituitary–adrenocortical (HPA) axis. This centrally activated HPA axis implies activated hypothalamic corticotropin-releasing hormone (CRH) and vasopressin (AVP) expression, in turn driving pituitary production of the 31-kDa large precursor poly-peptide hormone, pro-opiomelanocortin (POMC). Newly synthesized POMC proteins are predominantly sorted in dense secretory core granules within the corticotropic cells and are subsequently cleaved into adrenocorticotropic hormone (ACTH) by proprotein convertase 1 (PC1/3) [
6‐
8]. Subsequently, ACTH is released in the systemic circulation and rapidly activates the adrenal cortex to synthesize and release cortisol [
7]. However, the hypercortisolism of critical illness is not accompanied by elevated plasma ACTH. This has been referred to as ‘ACTH–cortisol dissociation’ [
1,
9].
Over the last decade, it has been shown that the rise in systemic cortisol availability during critical illness is to a large extent explained by suppression of the cortisol-binding proteins, increasing the free fraction of cortisol in the circulation, and by suppression of cortisol breakdown in liver and kidney [
1,
10]. Whether the low plasma ACTH is the consequence of glucocorticoid-receptor (GR)-mediated feed-back inhibition exerted by high circulating cortisol driven through these peripheral mechanisms remains debated [
2,
9]. In addition, the site of such feed-back inhibition (hypothalamus, pituitary or both), if present, and the affected pathways remain incompletely understood.
In the subacute and chronic phases of illness, incremental ACTH responses to a bolus injection of CRH were shown to be suppressed [
11]. Upon recovery, one week after intensive care discharge, rebound rises in plasma ACTH and cortisol to supra-normal levels have been reported [
10]. These data suggest the possibility of a centrally suppressed adrenocortical function when critical illness lingers. However, the finding that ACTH is not fully suppressed while circulating free cortisol is substantially elevated, suggests an ongoing central stimulation [
2,
9]. Indeed, while ACTH was low, cortisol production rates documented via tracer technology were found to be doubled as compared with healthy subjects [
1]. Hence, adrenocortical stimulation not exerted by ACTH may contribute to steroidogenesis in these patients, while elevated free cortisol could exert central feed-back inhibition.
We hypothesized that sepsis-induced critical illness, further referred to as ‘sepsis,’ indeed immediately and continuously activates the hypothalamus to generate, via CRH and AVP, ACTH-induced hypercortisolism, but as soon as free cortisol is elevated, feedback inhibition at the pituitary level interferes with normal processing of POMC into ACTH, explaining the typical ACTH–cortisol dissociation. In this constellation, unprocessed POMC could leach from the pituitary into the systemic circulation, which could in theory stimulate the adrenal cortex [
12,
13]. To test this hypothesis, we first documented plasma concentrations of POMC in relation to ACTH and cortisol in acute and prolonged human critical illness evoked by sepsis. Subsequently, we performed a study in septic mice to document the hypothesized alterations within the hypothalamus, the pituitary and the adrenal cortex in relation to duration of illness.
Part of these results has been previously reported in the form of an abstract [
14].
Discussion
The two studies of human patients suffering from sepsis revealed that, in the face of the known ACTH–cortisol dissociation, plasma concentrations of the ACTH precursor POMC were substantially elevated from the acute into the prolonged phase of sepsis-induced critical illness. In the mouse model of sepsis, this hormonal phenotype was confirmed. In the mice, sepsis was found to acutely, though transiently, increase hypothalamic expression of CRH and AVP, followed by an upregulation of both CRH and AVP receptor expression at the pituitary level. Also, from acute throughout prolonged sepsis, pituitary POMC gene expression was elevated. Together, these findings are suggestive of a centrally activated HPA axis irrespective of illness duration. In contrast, markers of processing POMC into ACTH and of ACTH secretion, known to be negatively regulated by glucocorticoid receptor ligand binding, were suppressed at all time points, offering explanation for the low ACTH and the high POMC plasma levels. Although adrenocortical structure was distorted, markers of adrenocortical steroidogenic activity were increased. The possibility that the latter is driven by high circulating POMC requires further investigation.
The first important and novel finding was the high levels of circulating POMC in both the acute and prolonged human sepsis studies as well as in the mouse model of sepsis. This observation corroborates a hypothalamic activation in response to sepsis. The mouse study indeed revealed that hypothalamic activators of POMC expression, CRH and AVP, were ubiquitously expressed, their pituitary receptors were upregulated, and pituitary levels of POMC gene expression were high. The preserved CRH and AVP expression is in line with a previous study of experimental septic shock in rats and of human septic shock non-survivors [
19], in which molecules drive the preserved CRH and AVP expression remains speculative, but could involve inducible nitric oxide synthase (iNOS) [
19,
20], cytokines [
21] and catecholamines [
22]. These data could all point to a centrally activated cortisol production, were it not that pituitary ACTH levels were low, and circulating levels never increased. Preserved pituitary POMC expression coinciding with reduced pituitary ACTH and normal to low circulating ACTH suggests impaired pituitary processing of POMC into ACTH and secretion hereof in the systemic circulation. Indeed, gene and protein expression of prohormone convertase 1/3, the dominant GR-regulated enzyme responsible for proteolytic cleavage of POMC [
8], was suppressed during all phases of sepsis-induced critical illness. In addition, Annexin A1, a GR-regulated potent inhibitor of pituitary exocytosis of stored ACTH [
7], was upregulated at the mRNA level.
The differentially altered pituitary gene expression profile of two POMC-activating transcription factors, Nur77 and Tpit, of which the first is positively regulated by CRH-R ligand binding and negatively by GR ligand binding [
23] and the second is positively regulated by CRH-R ligand binding but not affected by GR ligand binding [
24], confirms the presence of both stimulating CRH and AVP signaling and suppressing GR-ligand-binding-induced signaling, during sepsis. Such centrally stimulated ongoing POMC gene expression together with suppressed downstream processing into ACTH could result in an increased availability of POMC protein. This POMC can leach out from the corticotrope cells via the constitutive secretory pathway, bringing about the high circulating levels [
7,
25]. Such POMC leaching into the circulation could explain why POMC protein does not appear to accumulate in the pituitary gland despite increased POMC gene expression and impaired POMC processing into ACTH.
Whether increased circulating POMC levels are solely explained by increased CORT-induced feedback inhibition at the level of the pituitary GR, in the presence of ongoing central activation by CRH/AVP, remains unknown. The surprising finding that in human sepsis patients, both ACTH and cortisol rose upon recovery, suggests that the driving GR-binding ligands may be eliminated upon recovery. Hence, these GR-binding ligands could be distinct from glucocorticoids, as cortisol was found to be even higher during recovery than during the prolonged ICU phase [
10], in which other molecules that activate the GR in the context of sepsis-induced critical illness remain currently unknown. However, bile acids, of which the circulating levels are known to rise in this condition [
26] and to normalize upon recovery, could be one of other possible candidates [
27,
28].
In the adrenal cortex, gene expression of regulators and markers of steroidogenesis, which are assumed to be predominantly ACTH stimulated, was found to be upregulated in the absence of increased plasma ACTH. One possible explanation could be an increased ACTH sensitivity [
29], a possibility that is supported by the here observed increased expression of MRAP, a facilitator of MC2-R expression and signaling [
30], coinciding with normal or increased expression of the ACTH receptor MC2-R itself. However, in patients with sepsis, ACTH responses to cosyntropin are never elevated as shown previously [
10], not supporting this hypothesis. Alternatively, the MC2-R could be activated by ligands other than ACTH. In theory, small ACTH fragments, not detected by highly specific immunoenzymometric assays, could exert such steroidogenic effects. However, small ACTH fragments have previously shown to be absent in septic and non-septic ICU patients [
9]. In addition, pituitary expression of PC2, which could have increased further processing to smaller fragments, was increased only in the acute phase of sepsis. Alternatively, increased circulating POMC could play such a role and contribute to steroidogenesis. Such a steroidogenic role for POMC has already been suggested by cases of clinically overt Cushing’s syndrome revealing high plasma concentrations of POMC and (very) low plasma ACTH [
12,
13]. If POMC can drive steroidogenesis, this could be either a direct effect through binding of POMC to MC2-R or an indirect effect if POMC is locally cleaved into ACTH at the adrenocortical level [
13]. Whether such a POMC effect contributes to the increased mRNA of the steroidogenic enzymes that we found in the septic mice, and thus to the elevated plasma CORT while ACTH is not increased and while the adrenocortical structure and cholesterol ester storage was compromised, remains to be investigated. Indeed, the observed contrast between the increased mRNA levels of the steroidogenic enzymes, particularly those involved in cholesterol supply (HDL-R, LDL-R and HMG-CoA reductase) and the pronounced lipid depletion in the adrenal cortex, is striking. Whether the upregulation of these enzymes at the mRNA level without effectively resulting in augmented adrenocortical cholesterol content is the consequence of low circulating cholesterol, is unknown. Alternatively, a local inflammation-driven steroidogenesis or increased sympathetic activation in response to sepsis has been suggested to explain high systemic CORT in the face of low plasma ACTH [
31,
32]. However, we here report only a transient rise in adrenal TNF-α mRNA with suppressed levels present in the prolonged phase of illness, not supporting a local inflammatory process driving adrenocortical steroidogenesis.
This study has some limitations. First, in the human studies, plasma hormone concentrations were compared between septic patients and matched healthy subjects; a descriptive analysis only and the impact of frequently used drugs, baseline BMI and nutritional status or other potential confounders were not investigated. Second, the mouse model of sepsis-induced, antibiotics-treated and fluid-resuscitated critical illness is a model of fecal peritonitis not treated surgically, unlike what would be done for human patients. Third, with the chosen experimental design, we could not provide insights in alterations occurring prior to the 1-day time point. Fourth, the interpretation of the data as suggestive for suppressed POMC processing into ACTH during critical illness was based only on gene and protein expression and not on quantification of enzyme activity. Fifth, because of the small size of a mouse pituitary, we used whole pituitary homogenates for gene and protein expression and because of a limited amount of proteins yielded from pituitary homogenates, it was not possible to perform western blot analysis for all transcriptional regulators. Finally, the scarce knowledge on any potential steroidogenic capacity of POMC requires further research.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.