A Glucocorticoid-Mediated Immunoregulatory Circuit Integrated at Brain Levels: Our Early Studies and a Present View

Probably the first link between the adrenal gland and immune organs was identified when Henry Jaffe showed at the beginning of the twentieth century that adrenalectomy results in increased thymus weight, generalized hyperplasia of the lymph nodes and bone marrow, and blood lymphocytosis [1]. He also wrote that “severe chronic infections may sometimes bring about a pronounced pathological involution of the thymus,” which could be explained retrospectively based on studies that followed.

Subsequently, Hans Selye [2] reported that “toxic substances,” including bacterial toxins, elicit hyperplasia of the adrenal cortex and involution of the thymus. This work opened a field that is up to now extensively studied: stress and immunity. More or less at the same time, Edward Kendall isolated several compounds from the adrenal gland, which included cortisone and cortisol (at that time named compound E and compound F, respectively), and Tadeus Reichstein isolated seven products, including these compounds. The potential clinical relevance of this material was unraveled when Philip Hench reported in the 1940s that the adrenal gland produces a hormone that ameliorated the symptoms of inflammatory diseases such as arthritis and rheumatic fever, and the first patient with rheumatoid arthritis received cortisone in 1948. Edward Kendall, Tadeus Reichstein, and Philip Hench were awarded the Nobel Prize in Physiology or Medicine in 1950, “for their discoveries relating to the hormones of the adrenal cortex, their structure, and biological effects” [3]. These seminal discoveries led to a milestone in the treatment of autoimmune and inflammatory diseases.

In 1951, Peter Medawar, who, together with Macfarlane Burnet, also received the Nobel Prize in 1960 for the discovery of acquired immunological tolerance, obtained early evidence of immunologic actions of glucocorticoids. Together with his co-workers, he showed that local application of cortisol delays rejection of skin allografts [4]. Six synthetic steroids were introduced between 1954 and 1958 for systemic anti-inflammatory therapy [5]. Since then, the use of glucocorticoids has largely expanded, but also many secondary effects of high doses of synthetic glucocorticoids have been described [5]. Later, evidence about effects of endogenous glucocorticoids was reported. For example, it was shown in 1973 that immune responses are increased in adrenal deficient animals [6]. Around this time, it was reported that the sensitivity to glucocorticoids differs among subpopulations of human lymphocytes [7].

The discovery of the glucocorticoid receptor (GR) was probably the next step in understanding the effects of this hormone on immune processes. The first indication of the nature and functioning of the GR in lymphoid cells was reported by Allan Munck and Truls Brinck-Johnsen [8].

The early evidence mentioned led to the generalized concept that glucocorticoids are anti-inflammatory and immune-suppressive hormones. However, later studies showed that this view can be misleading when applied to basal and moderately increased levels of glucocorticoids. Although stimulatory effects of glucocorticoids on mitogen-stimulated B cells were already reported in 1977 [9], we have probably been the first to provide evidence that low doses of glucocorticoids are needed to elicit a specific adaptive immune response in vitro. We reported that murine lymphocytes fail to mount an immune response to sheep red blood cells (SRBC) unless a dose of glucocorticoids comparable to a physiologic concentration was added to the cultures, indicating that this hormone is necessary to prime immune cells during specific immune responses [10].

Today, there is clear evidence that low blood levels of glucocorticoids can also enhance innate immunity by supporting inflammation, a protective physiologic response that contributes to eliminate the initial cause of cell injury, clear out damaged cells and tissues, and initiate tissue repair. We shall not expand here on proinflammatory effects of low levels of endogenous glucocorticoids but just quote two examples to illustrate this aspect. For example, at physiologic levels, glucocorticoids can induce upregulation of some pattern-recognition receptors and favor pathogen recognition via evolutionarily conserved damage-associated molecular pattern molecules [11]. Also, Busillo and colleagues [12] showed that glucocorticoids positively regulate the expression of NLRP3, a receptor that recognizes pathogen-associated molecular patterns, which, in turn, leads to enhanced production of the pro-inflammatory cytokines IL-1β and IL-18. In summary, it can be concluded that low physiologic levels of glucocorticoids enhance a protective response in reaction to dangerous signals [13, 14]. Opposite effects on inflammatory and adaptive immune responses are observed when high levels of endogenous glucocorticoids are reached during acute or chronic conditions, or when exogenous analogs are therapeutically administered. In these cases, the anti-inflammatory and immune suppressive effect of natural or synthetic glucocorticoids predominate mainly by mechanisms that result, for example, in inhibition of AP-1 and NF-κB and upregulation of their inhibitors [15].

Another important aspect is the evidence that endogenous glucocorticoids exert a permanent control of immune cell activity under basal conditions. This modulatory effect was proven in non-overtly immunized animals by evaluating the number of B cells producing antibodies independently of their specificity. The number of activated B cells is an indication that, under natural conditions, immune responses are permanently triggered by exposure to environmental and internal antigens. A clear increase in the number of antibody-producing cells in the spleen was detected when the effect of endogenous glucocorticoids was reduced by adrenalectomy. The opposite situation was observed in animals exposed to the stress of transportation from supplier to the housing place, which resulted in increased levels of corticosterone. The change in the number of antibody-producing cells between both conditions, adrenalectomy and transportation stress, was around 10-fold [16]. In addition, an inverse correlation between immune cell mass and B-cell activity in the spleen and corticosterone levels was observed when the endogenous concentration of the hormone was increased during stressful situations and reduced by adrenalectomy. These simple experiments illustrated the impact of endogenous glucocorticoid levels on the activity of the immune system.

We shall concentrate here on different inputs that lead to levels of endogenous glucocorticoids that can affect immunity. Psychologic stress results in the stimulation of glucocorticoid release together with other hormones via activation of the hypothalamus-pituitary axes. The most accepted concept is that, while increased glucocorticoid levels resulting from acute stress support natural immunity, they exert variable actions on adaptive immunity [17]. In contrast, anti-inflammatory and immune suppressive effects are manifested when glucocorticoid levels increase during chronic stress. The initial protective function of glucocorticoids is lost and the effect of these hormones is either blunted or becomes erratic and deleterious (see below).

The release of glucocorticoids is also elicited by homeostatic inputs such as changes in glucose homeostasis and osmolarity. As known since long time, this endocrine response, together with other neuroendocrine signals, contributes to regulate these processes. For example, glucocorticoids stimulate glucose production and are released as a regulatory response to hypoglycemia. As we shall discuss in the following section, immune regulatory effects of endogenous glucocorticoids took longer to be unraveled, and, already in 1969, Beisel and Rapoport [18] discussed that, although much was known about the effect of exogenous steroids at that time, there were still many questions about the role of the host’s own adrenal response to infectious illness.

In the following, we refer to how early evidence that a release of glucocorticoids with immunoregulatory effects is triggered during inflammatory and adaptive immune responses was obtained. Although we mainly concentrate on evidence obtained from our early work, it is clear that this would have not been possible without the background provided by previous and contemporary researchers.

As pointed above, glucocorticoids were already widely used in the clinics as anti-inflammatory and immunosuppressive agents at the beginning of 1970s. It was also known that changes in endogenous glucocorticoid levels either by adrenalectomy or stress can affect the immune response. However, this evidence did not automatically indicate that changes in endogenous glucocorticoids are part of a physiological immunoregulatory process. A general principle in physiology is that a regulatory process is triggered when there is a change in the conditions and/or activity of its target. For example, the regulation of blood pressure by neuroendocrine mediators is triggered when brain centers receive information from pressure receptors that blood pressure is altered. These receptors convey information to the brain, which in turn reacts with a regulatory response. Such response should either reestablish basal normal blood pressure or result in an adaptive adjustment to body activity, for example, during exercise. In case of immunoregulation by glucocorticoids, immune cells should send information to brain centers, for example, those located in the hypothalamus, so that they orchestrate a response that results in a change in glucocorticoid release. This change would either contribute to maintain the basal activity of immune cells or modulate their operation when they are stimulated.

Our first approach to explore whether the activity of the immune system is under neuroendocrine regulation was to immunize rodents and use blood levels of corticosterone as readout. The hypothesis was that if there are interactions between the immune and the nervous systems, immune activation should elicit a regulatory neuroendocrine response that in turn affects the course of the immune response. This interaction should occur under situations in which one system does not harm the other and also under non-stressful, toxic, or pathological conditions. Thus, it was also necessary to deal with the confounding factors that, under natural conditions, immune responses are often associated with direct or indirect tissue damage and altered organ functions and that illness, as stressor, can per se elicit neuroendocrine responses. To circumvent these problems, we immunized animals with innocuous antigens that can elicit a strong adaptive immune response without causing any disease. One of the antigens classically used at that time to immunize mice and rats was SRBC, a model of immunization that permits to have animals that receive the same number of syngeneic red blood cells as control. The first approach was to measure blood levels of corticosterone in animals that received this antigen since it was already known that this hormone can affect immune processes. A clear increase in corticosterone levels was detected few days after immune challenge with SRBC [19]. This delay already indicated that the corticosterone response was not the consequence of a possible stress linked to the intraperitoneal administration of the antigen. This was further supported by the lack of endocrine response in animals that had received syngeneic erythrocytes. The only precedent to this result that we have retrospectively found was a report showing that a pyrogenic endotoxin preparation from Pseudomonas quickly (within hours) induces histological alterations and changes in ascorbic acid and cholesterol content in the adrenals [20]. At that time, it was discussed whether this effect was caused by direct stimulation of the hypophysis by the endotoxin or indirectly due to an unspecific stress response to the pyrogen, as quoted by Wexler and colleagues [20]. The possibility that the release of glucocorticoids triggered by the preparation of endotoxin used was immunologically induced was not considered.

To study whether the increase in corticosterone levels observed at the time of the peak of the immune response to SRBC was a more general phenomenon, we also used two unrelated antigens, keyhole limpet hemocyanin and horse erythrocytes, which are highly immunogenic in rats and mice [19]. An increase in corticosterone blood levels was also detected in parallel to the immune response to these antigens.

The evidence that the immune response resulted in increased glucocorticoid levels was replicated by different laboratories [21‒27]. It was further shown that the increase in glucocorticoid levels following SRBC challenge is dependent on the intensity of the immune response [26]. This study used the Biozzi high and low responder mice and it was found that corticosterone blood levels were increased on the day of the peak plaque-forming cell response to SRBC or to the trinitrophenyl hapten, only in high responder animals but not in antigen-injected low responder animals, suggesting a direct correlation between immune responsiveness and hormonal changes. The increase in corticosterone levels following immunization also affected the circadian rhythms of this hormone in high responder mice.

It is worth noting that not all types of immune responses result in stimulation of the pituitary-adrenal axis. For example, corticosterone blood levels are lower in animals rejecting a skin allograft than in animals bearing autografts [28]. A later report showed that major alterations in the processing of proopiomelanocortin in the pituitary gland are observed during skin graft rejection [29].

Using the same antigen (SRBC), we also showed that the immune response evokes a clear increase in the rate of firing of hypothalamic neurons [30] and in noradrenaline turnover rate in the hypothalamus [31] and a decrease in the activity of the sympathetic nervous system [32] (for more references, see [33]). These results implied that, as with other physiologic systems, the brain is informed about the activity of the immune system. This and other evidence led us to propose that the immune system is a receptor sensorial system that can detect the presence of nonself or modified-self molecules [34].

After establishing that the immune response can elicit a glucocorticoid response, our next step was to search for possible immune cell-derived products that could mediate this effect. For this purpose, immune cells were stimulated in vitro and the cell-free supernatant was injected into naïve animals. The first approach was to stimulate human peripheral blood mononuclear cells (PBMCs) or rat spleen cells with concanavalin A, a mitogenic lectin that mainly affects T cells, and the supernatant of these cultures was injected into rats. Corticosterone blood levels started to increase 30 min after intraperitoneal injection of this material and reached levels that were more than 3-fold higher than that of control animals injected with medium from non-stimulated cells after 2 h [35]. The increase was similar to that of rats and mice after injection of SRBC or trinitrophenyl-hemocyanin, suggesting that factors produced in vitro by stimulated immune cells can elicit the glucocorticoid increase that occurs during the course of the immune response in vivo. Another approach was to study if these factors are also produced during an in vitro-induced immune response (a mixed human lymphocyte culture) practically in the absence of macrophages. Less than 10⁵ human lymphocytes, so the amount contained in less than 0.1 mL blood, can produce enough material to stimulate the HPA axis, showing the high potency of these supernatants. We termed these products glucocorticoid-increasing factors (GIF) [36]. The increase in corticosterone levels mediated by GIF did not occur in hypophysectomized animals, showing that this effect was not mediated by a direct effect on the adrenals. These results also indicated that not only mitogens but also antigens can stimulate cells to produce GIF.

A further approach was to stimulate human peripheral blood leukocytes or mouse spleen cells with Newcastle disease virus (NDV), a natural infective agent. In contrast to chicken, this virus is innocuous in mice unless they are adrenalectomized, already indicating the protective role of endogenous glucocorticoids [37]. A clear increase in ACTH and corticosterone blood levels was also detected in this model [36, 38].

Due to our limited resources at that time, the only studies that we performed aiming at characterizing these factors indicated that they were proteins in the molecular weight range of 15–20 kDa (unpublished). Thus, we had to wait until the mid-eighties that the first cytokines became available in pure or recombinant form to test their effects on the HPA axis. Our first study was possible when we had available a highly purified preparation and also recombinant human IL-1β to explore whether this cytokine can stimulate this axis. The choice of this cytokine was based on our previous studies showing that the increase in ACTH and corticosterone levels induced by supernatants from in vitro stimulated immune cells with NDV was inhibited by an antibody to IL-1 [39]. Thus, the next step was to administer low, subpyrogenic doses of IL-1β into LPS-resistant C3H/HeJ mice. An increase in ACTH and corticosterone levels was observed 30 min after intraperitoneal administration of the cytokine, and the maximal response was observed after 2 h. A dose-response study indicated that this effect could be elicited by 0.25 μg IL-1β/mouse. One year after the publication of these results, three back-to-back published papers provided more details about the capacity of IL-1 to stimulate ACTH and corticosterone release. While the report by Bernton et al. [40] indicated that IL-1 can affect ACTH release in in vitro cultures of rat pituitary cells [40], the studies in vivoby Sapolsky and colleagues [41] and by our group [42] showed that the effect of the cytokine on the HPA axis is induced by stimulation of the release of hypothalamic CRH.

We have also compared the capacity of IL-β, TNFα, and IL-6 to stimulate the HPA axis. Although TNFα and IL-6 could also induce an increase in ACTH and corticosterone blood levels upon administration to rats, IL-1 was much more potent than these other cytokines when compared on a weight basis [43]. Today, it is well established that multiple cytokines can stimulate glucocorticoid release. More than 20 years ago, we quoted that also IL-2, IL-3, IFNγ, and some colony-stimulating factors are among the cytokines that can exert this endocrine effect (for review [44]). Since then, this list has been considerably enlarged by adding IL-11, IL-12, IL-17, IL-18, IL-37, and IFNα and β, but it is beyond the scope of this paper to discuss more details on this aspect. In our view, the important concept to emphasize here is that this cytokine redundancy probably indicates the biological relevance of the immune-mediated capacity to stimulate the HPA axis.

To explore whether the increase in corticosterone levels during a specific adaptive immune response is of immunoregulatory relevance, we took advantage from the well-established phenomenon of sequential antigenic competition, in which the immune response to one antigen, for example, horse red blood cells, results in inhibition of the response to an unrelated antigen, for example, SRBC, injected later. We have shown that the increase in corticosterone levels induced by the first antigen results in inhibition of the response to the second one, as shown by the fact that adrenalectomy inhibits antigenic competition [10]. This study permits to interpret, for example, why the immune response to a viral stimulation, for example, influenza virus, compromises the response to a subsequent bacterial infection with Listeria monocytogenes [45].

There is large evidence showing that increased levels of glucocorticoids are detected during infectious diseases. A recently published meta-analysis, which included 42 studies involving 32 host and 32 pathogen species, shows how universal this endocrine response is [46]. The evidence that activation of immune cells results in peripheral and central cytokine production that stimulates the release of glucocorticoids to levels that can affect the immune response was the basis for the postulation of the glucocorticoid-mediated immunoregulatory circuit [47] (Fig. 1).

Most reports support the view that increased glucocorticoids levels play a protective role during infections by impeding a harmful overstimulation of the immune response [17]. However, there is also evidence indicating a more complex role of this hormone during immune processes. Unfortunately, there is not much information on how dynamic changes in endogenous levels of glucocorticoids affect the course of immune responses and how the levels attained during such responses regulate the activity of each type of immune cells at each given step. Taking together the evidence available at present, in the following we attempt to briefly analyze how endogenous glucocorticoids could exert regulatory actions during the course of an immune response. A possible interpretation is that basal and initially slowly increasing glucocorticoid levels would cause an increase in the sensitivity of inflammatory cells that favors the detection of invading microorganisms and the initiation of the process tending to their neutralization [48] by facilitating antigen presentation and specific clonal expansion [49]. Afterward, glucocorticoid levels continuously increase, reaching a plateau that is sustained for several days, as shown in the models of immune response described above. The increase in glucocorticoid levels when the immune response is fully expressed would serve to shape the specificity of the response by reducing the probability of bystander effects, such as, for example, the recruitment of unrelated lymphocyte clones or those with low affinity for the antigen. Endogenous glucocorticoids are likely to exert such control without interfering with the ongoing immune response since activated T cells, for example, become resistant to the apoptotic effect of this hormone [50‒52]. Furthermore, increased levels of glucocorticoids during adaptive immune responses increase the expression of IL-7 receptor α in activated T cells. IL-7 is a cytokine that not only prevents apoptosis of activated T cells [53] but that also plays a positive role on the immune responses by promoting the production and function of effector and memory T cells [54]. Thus, the protection of antigen-specific T cells from endogenously increased glucocorticoid levels during the effector phase would contribute to refine the specificity of the immune response. Endogenous glucocorticoids also promote a switch from Th1 toward Th2 functions since Th1 cells are sensitive to glucocorticoid-induced apoptosis and cytokine suppression, while Th2 are sensitive to cytokine suppression by glucocorticoids but not to apoptosis, and Th17 are resistant to both effects [55]. Thus, it is expected that endogenous glucocorticoids would promote antibody production by favoring Th2 and Th17 activity since not only Th2 but also Th17 can strongly promote B-cell differentiation and class switch recombination [56].

Another way by which endogenous glucocorticoids could also contribute to regulate an immune response is by affecting regulatory T cells (Tregs) since in vivo and in vitro treatment with dexamethasone enhances the relative proportion and selective expansion of this cell type [57‒60]. It is also worth noting that, as part of a permanent control of immune cell activity, endogenous glucocorticoids also modulate immune cell distribution [61]. For example, circadian changes in glucocorticoid levels contribute to control daily changes of T cell distribution between lymphoid organs and blood, and an accumulation of T cells in lymphoid organs at night promotes CD8 T-cell responses to infection. In our view, aspects that determine glucocorticoid effects during an immune response as those mentioned above should be considered when the immunoregulatory relevance of these hormones is evaluated during pathologic conditions.

As mentioned, we have shown that the increased levels of ACTH and corticosterone following infection with NDV is mediated by endogenously produced IL-1. Considering that this virus is lethal in adrenalectomized animals [37], this was probably the first indication that a cytokine-HPA axis circuit allows survival from the infection. Probably the first evidence that cytokines mediate a protective stimulation of the HPA axis during the course of a viral infection that can be lethal was provided by Ruzek and colleagues [62, 63]. They showed that IL-6 mediates a marked increase in glucocorticoid levels that protects mice against life-threatening infection with murine cytomegalovirus infection, with IL-1 contributing to IL-6 production. Since then, other examples became available showing that the immune response rather than the infectious agent mediates the increase in glucocorticoid levels during different viral infections (for review [64]).

There is also evidence that an increase in endogenous corticosterone levels influences the susceptibility and clinical course of parasitic infections. While C57Bl/6 mice develop a severe disease following infection with Trypanosoma cruzi, an intracellular parasite that causes Chagas’ disease, Balb/c mice are more resistant. An intense (more than 10-fold) stimulation of the HPA axis was observed in both strains 3 weeks after infection with the parasite, but glucocorticoid levels were already increased two- to threefold in the less susceptible Balb/c strain during the first week following T. cruzi injection, when no evidence of disease is detected. Blockade of GRs with the glucocorticoid antagonist RU486 accelerated death in C57Bl/6J mice and was lethal in 100% Balb/c mice. These results indicate that endogenous glucocorticoids do not only restrict the course of the disease caused by the parasite but also control the susceptibility to express it [65]. A more recent study by Kugler and colleagues [66] is in line with these observations. They show that CD4⁺ T cells trigger a glucocorticoid response in mice infected with the unicellular parasite Toxoplasma gondii, an endocrine response that, in turn, decreases CD4⁺ T-cell effector function and is crucial for preventing tissue damage and promoting host survival.

The glucocorticoid immunoregulatory circuit does not only operate during infections but also during autoimmune diseases, such as experimental autoimmune encephalomyelitis (EAE). Endogenous glucocorticoid levels are increased during EAE in Lewis rats and this increase is essential for survival [67]. Increased blood glucocorticoid levels are also observed in EAE-resistant PVG rats following immunization with the encephalitogenic antigen myelin basic protein (MBP) [68]. However, MBP-injected adrenalectomized PVG rats develop a severe disease from which they do not recover [69], illustrating the relevance of the increase in endogenous glucocorticoid levels in this disease. We have also shown that immune cells from animals with EAE can produce mediators that increase glucocorticoid levels when administered to naive recipients. Furthermore, acute in vivo blockade of IL-1 receptors inhibits to large extent the increase in corticosterone levels during EAE. These results indicate that a glucocorticoid response, which is decisive for the prevention or moderation of EAE, is mainly the result of the activation of a protective glucocorticoid-associated immunoregulatory circuit [68].

Increased levels of glucocorticoids are also detected at early stages of tumor growth, either when the tumor is de novo induced or following transplantation of syngeneic tumor cells. This effect is noticed even before the tumor becomes palpable and without signs of disease [70]. At least in one model, we have shown that this early effect is mediated by a host-derived factor/s present in the ascitic fluid of animals bearing a fully syngeneic EL-4 lymphoma [71]. In this model, the presence of T cells in the host is required to elicit changes in glucocorticoid levels, indicating that the hormonal response is immunologically mediated [72]. We have also shown that the levels of glucocorticoids attained as response to EL-4 cell inoculation are anti-inflammatory [73]. Altogether, this evidence indicates that the immune-glucocorticoid circuit is also operative in this tumor model.

On the other hand, there are also models showing that the glucocorticoid-associated immunoregulatory circuit is not functioning and thus cannot contribute to control or moderate the disease. For example, the increase in corticosterone blood levels was not sustained in Dark Agouti rats with rheumatoid arthritis, and it was even followed by a progressive decrease in the levels of the hormone at a time when high levels of cytokines that can stimulate the HPA axis are produced during this disease [74]. It was also shown in a model of induced arthritis in Lewis rats that there is a central nervous system defect in the biosynthesis of CRH, which contributes to the susceptibility to develop the disease [75]. Disturbances in the circuit have been also demonstrated in obese chicken that spontaneously develop autoimmune thyroiditis and are used as model of human Hashimoto’s disease [25].

In general, it has been shown in animal models that increased levels of endogenous glucocorticoids are beneficial for the host during acute viral, bacterial, and parasitic infections [44, 63‒65, 76] and during certain autoimmune diseases [25, 67, 68, 75]. Unfortunately, this protective effect tends to dissipate during more prolonged conditions, as shown in animal models and also in humans. The impairment in the functioning of the circuit can be the consequence of several causes, among which are glucocorticoid deficit or glucocorticoid resistance. Decreased glucocorticoid levels are commonly due to adrenal insufficiency, as observed during sepsis [77, 78], but also to significantly lower ACTH levels as observed in critically ill septic patients [79] or a blunted pituitary response to CRH [80]. Glucocorticoid resistance might be due to defects at the level of the GR, for example, down regulation [81], or less expression of receptor [82], inhibition of GR transcriptional activity [83], reduced glucocorticoid binding capacity [84], or lower GR affinity [85], or overexpression of the GR beta isoform [86], which has been implicated in the impairment of the classical isoform of the receptor responsible for glucocorticoid actions [87]. In a rat model of arthritis, we found that the defect in this circuit can be also due to a disruption between peripheral and central signals or a desynchronization in the responses of the systems [74].

These alterations can result in an uncontrolled release of pro-inflammatory cytokines, as shown, for example, during advanced stages of sepsis, accelerating its lethal course [88‒90]. Besides sepsis, there are clinical studies indicating that the glucocorticoid-associated immunoregulatory circuit is or could be also defective in infections that do not necessarily end in sepsis, like rhinovirus infection [91], SARS-CoV-2 infection [92], or severe malaria [93]. Alterations in the circuit are also observed in several other diseases in humans, for example, arthritis [86, 94, 95], relapsing remitting multiple sclerosis [96], Sjögren’s syndrome [80], psoriasis [97], chronic inflammatory bowel disease [81, 83], fibromyalgia [82], asthma [94], and atopic dermatitis [84] (Fig. 2). The references quoted do not pretend to provide a complete list of the studies supporting this point and are included only as examples.

Two common endocrine human diseases that involve alteration in glucocorticoid levels are Cushing’s syndrome and Addison’s disease. Chronic hypercortisolism (Cushing’s syndrome) induces a variety of alterations in the immune system, often leading to severe clinical complications such as sepsis and opportunistic infections [98]. On the other hand, infections have been reported as a major cause of premature death in patients with primary adrenal insufficiency (Addison’s disease), a disease of autoimmune origin [99, 100], and also in patients with secondary adrenal insufficiency due to hypopituitarism [101]. Although it would be possible to propose a tentative explanation based on alterations in the glucocorticoid-associated immunoregulatory circuit, it is beyond the scope of this article to further discuss these diseases.

We have briefly discussed above the impact of adrenal insufficiency and glucocorticoid resistance among the factors that can lead to failures in the operation of the glucocorticoid-mediated immunoregulatory circuit. However, the final regulatory efficacy of glucocorticoids is conditioned also by other variables that, in our view, have not been sufficiently explored during diseases involving the immune system.

The availability of endogenous glucocorticoids to a cell depends on extracellular binding proteins, and a large proportion of circulating glucocorticoids is bound to corticosteroid-binding globulin (transcortin) or albumin. There is evidence that, for example, corticosteroid-binding globulin is reduced during sepsis leading to increased free cortisol availability to immune cells [102]. This effect, which can initially restrict pro-inflammatory cytokine production, is blunted at later steps when glucocorticoid resistance develops [103].

As in other cells, physiological effects of glucocorticoids on immune cells are mediated by the GR. The variable proportion of GR and its isoforms determines the diversity of glucocorticoid effect and their potential actions on different types of immune cell targets [104]. It would be necessary to explore changes in the expression of these isoforms that might occur at different stages of a disease on a particular type of immune cell and would also help to decide at which stage of the immune response to a pathogen the administration of glucocorticoids could be beneficial.

A better identification of the cell target of glucocorticoids would be also necessary since a simplified interpretation or extrapolation of the effect of glucocorticoids might turn out to be wrong. Just as an example, glucocorticoids can increase or decrease toll-like receptor expression depending on which toll-like receptor is addressed and on which cell type [105].

The availability of active glucocorticoid to a target cell is also determined by 11β-hydroxysteroid dehydrogenase (11β-HSD) enzymes. Type 1 11β-HSD converts cortisone and 11-dehydrocorticosterone into the more powerful glucocorticoids cortisol and corticosterone, respectively, while type 2 11β-HSD functions in the opposite direction. The response of a tissue to glucocorticoids depends on the activity of these enzymes and on the intracellular glucocorticoid concentration. The expression and activity of these enzymes depend on the degree of activation of particular types of immune cells. For example, 11β-HSD1 is highly expressed in activated inflammatory cells and is downregulated as inflammation resolves [106‒108]. Besides glucocorticoids, other hormones and mediators such as cytokines can affect the intracellular levels of 11β-HSDs and modify glucocorticoid sensitivity during stress and inflammatory/immune responses (for references, see [106]). In a more recent and comprehensive article, Van den Berghe and colleagues [109] reviewed these and several other factors involving the HPA axis in critically ill patients with sepsis and hyperinflammation. Among several other factors, they discuss the relevance of considering the time-dependent pattern of changes in ACTH and cortisol levels, several variables that determine the tissue-specific regulation of cortisol action and availability, and the suppression of cortisol breakdown in liver and kidney. Another interesting aspect included in this review is the discussion of the gaps that are still needed to fill for future translational research and therapy, which might allow to increase the efficiency of endogenous glucocorticoids and eventually decrease their exogenous administration.

Another variable to be considered when evaluating effects of endogenous glucocorticoids is that ACTH, besides corticosterone/cortisol secretion, also stimulates the release of other hormones from the adrenal cortex, such as aldosterone, which can also affect immune function [110] and whose release is also stimulated by IL-1 [111], and dehydroepiandrosterone (DHEA), whose effect on some immune cells opposes that of cortisol or corticosterone [112, 113]. Thus, the final effects of these hormones would depend on the relative amount of DHEA and aldosterone released by the adrenal gland [114, 115].

A further important aspect that needs consideration is that the immunoregulatory glucocorticoid circuit does not operate in isolation since cytokines and other immune products can affect many neuroendocrine mechanisms [44, 116]. Thus, when possible, it would be desirable to include a profile of cytokines, hormones, and neurotransmitters simultaneously evaluated at different stages of a disease and following its treatment in clinical studies that address the relevance of endogenous glucocorticoids.

In collaboration with several colleagues, we started with this type of approach studying endocrine and cytokine profiles during different stages of lung tuberculosis caused by Mycobacterium tuberculosis in humans and their potential changes after treatment. We have studied endocrine responses involving pituitary, adrenal, gonadal, and thyroid hormones in parallel to circulating levels of several cytokines in newly diagnosed, untreated male patients with mild, moderate, and advanced lung tuberculosis and matched healthy controls [117]. IFN-γ, IL-10, and IL-6 levels were elevated in patients. DHEA and testosterone levels were profoundly decreased and growth hormone levels were markedly elevated in patients in the absence of changes in IGF-1, which indicates a high degree of growth hormone resistance. A modest increase in cortisol, estradiol, thyroid hormone, and prolactin blood levels was detected in parallel. A predominant feature was a marked increase in the glucocorticoid/DHEA ratio, which indicates that, in patients, the effect of glucocorticoid is relieved from the opposing effects of the adrenal androgen DHEA on the immune system [115]. Most of the changes observed in cytokine and hormonal levels depended on the degree of pulmonary involvement. Supernatants of PBMCs from the patients and stimulated in vitro with antigens from M. tuberculosis significantly inhibited DHEA secretion by a human adrenal cell line. This effect was reverted when TGF-β was neutralized in these supernatants [118]. Furthermore, it was demonstrated that cortisol inhibits mycobacterial antigen-driven lymphoproliferation and IFN-γ production and suppresses TGF-β production in DHEA-treated human PBMCs [117]. It was also observed that a lower body mass index in patients coexists with reduced levels of leptin and increased concentrations of IL-6, cortisol, IL-1β, and adiponectin [119]. Finally, it was shown that the clinical recovery of patients undergoing specific treatment is associated with changes in immune and neuroendocrine responses [120]. In our view, another interesting observation was that low blood levels of DHEA were also detected in non-overtly sick cohabitants [119]. Taken together, these results support the view that at least some of the endocrine changes observed in the patients may be mediated by endogenous cytokines. The endocrine profile of tuberculosis patients would favor a reduction of protective cell-mediated immunity and an exacerbation of inflammation, leading to perpetuation of lung injury and to the hypercatabolic condition (consumption) that characterizes this disease. These studies indicate that an integrative multiparametric clinical research in human pathologies would provide the rationale for interventions on both, immune defenses and basic adaptive homeostatic regulatory mechanisms, aiming at improving recovery. This would be particularly important during severe infections considering the increasing incidence of antibiotic resistance.

The functioning of the immune system in a healthy individual goes beyond prevention of disease, as shown by the multiple homeostatic mechanisms that are influenced by immune cell products [44, 116]. Cytokines contribute to modulate basic brain functions, such as learning and memory, and systemic mechanisms, such as intermediate metabolism, cardio-respiratory function, and general electrolytic balance. These mediators are released when the physiologic mechanisms they can affect need to be adjusted as consequence of internal or external challenges that require plasticity for adaptation. This would imply that molecular and cellular mechanisms integrate such broader network of interactions at brain level.

We have postulated that astrocytes, a third component of the now called tripartite synapse [121], play a central role in the integration of immunoregulatory neuroendocrine circuits that operate under healthy conditions [122]. The reason is that astrocytes, the predominant cell type in the brain, are considered as neuro-immune cells because, besides neural effects, they contribute to mediate brain immune defenses. Astrocytes produce cytokines such as IL-1, IL-6, and IL-18 [122‒127] that, acting as transmitters between neural cells, are also essential for the regulation of synaptic transmission during long-term potentiation (LTP). Increased expression of IL-1β, IL-6, and other cytokines are observed during LTP, and while IL-1 supports this process, IL-6 exerts opposite effects [128, 129]. The same effects are also exerted on the learning process [130‒132]. IL-1, either directly, via self-induction or via induction of a cascade of cytokines that are produced in the brain during peripheral immune stimulation and under sensorial and physical stress conditions [133‒138] mediates the stimulation of the HPA axis. Such stimulation contributes to the glucocorticoid-immune circuit that regulates immunity both in the periphery and in the brain. Both neurons and astrocytes can produce IL-1, and its concentration is increased following AMPA glutamate receptor stimulation. Acting in an autocrine/paracrine way, IL-1 contributes to increase glucose incorporation by these cells to assist in their functioning [122].

Based on this evidence, it can be accepted that brain cytokines contribute to (a) immunoregulation mediated by neuroendocrine mechanisms; (b) maintain physiologic brain functions that range from basic homeostatic mechanisms to complex behaviors. Considering the broad effects of brain cytokines, we have proposed that the tripartite synapse could play a central role in processing immune signals in the brain and in their integration with neuro-sensorial signals (references in [47]). When cytokine production in the brain is immunologically triggered, their effects would predominate in brain areas such as of the hypothalamus and exert immunoregulatory actions by activating the HPA axis. Furthermore, there is evidence that peripheral IL-1 has the capacity to induce its own production in other brain areas that finally may contribute to the stimulation of the HPA (for references, see [122]).

The panorama changes during infections and other diseases with immune components. In most cases, systemic infectious diseases lead to neuroinflammation, which is considered an initially adaptive process in the central nervous system [139]. A degree of neuroinflammation can occur even during mild diseases, such as a flu, and may play a neuroprotective role [140]. Often, this neuroinflammation is induced without a direct effect of pathogens in the brain but via peripheral signals released during the immune response. In addition, peripheral inflammatory processes might induce a degree of neuroinflammation, as suggested by over-expression of pro-inflammatory cytokines in the brain during disease [74]. The involvement of peripheral and central cytokines in symptoms of sickness such as fever, hypophagia, lethargy, listlessness, anhedonia, aversion, and reduced social communications [136, 141‒144] should also be integrated in this already complicated scenario, but it is beyond the scope of this review to discuss these aspects. Under these disease conditions, changes in the activity of a glucocorticoid-associated circuit mediated by cytokines produced during the immune response and by stress signals [133, 135, 137, 138], including the stress of being sick [136], seem to be a universal host response.

During neuroinflammation associated with peripheral advanced diseases, there is an overstimulation of glial cells [140]. While activation of microglia becomes central in the neuroinflammatory process, the physiologic function of astrocytes is not only lost but becomes detrimental, for example, by overproduction of IL-1. Instead of supporting LTP, IL-1 becomes inhibitory during immune activation and stress [133‒138]. It is known that the effect of this cytokine follows an inverted U, supportive or inhibitory depending on its concentration [132]. A similar situation occurs with effects of endogenous glucocorticoids on synaptic plasticity. While increased glucocorticoid levels, for example, during acute stress, support synaptic plasticity, this process is inhibited by the levels of these hormones attained during chronic stress [145].

In summary, the evidence discussed above showed that the glucocorticoid-mediated immunoregulatory circuit operates during health and disease. The initial indications were obtained using innocuous antigens to prove that this circuit operates independently of disease. Later, it was explored how this circuit operates during experimental models of disease and there are studies indicating that this circuit is protective during human infections and other pathologies involving the immune system. However, there is also evidence that this circuit is altered during the course of some infectious diseases and chronic pathologies. As an attempt to integrate regulatory neuroendocrine mechanisms that control immune and brain functions, we have discussed evidence supporting the view that the tripartite synapse integrates immunoregulatory neuroendocrine mechanisms, such as the glucocorticoid-associated circuit. However, it should be emphasized that, although this circuit plays a central role in maintaining health, it is known since a long time that cytokines and other immune products have many other neuroendocrine and metabolic effects [44]. Thus, it is expected that different neuroendocrine-immune circuits could become predominant at different stages of a pathology expressed at distinct phases of life, such as during development, growth, reproduction, and aging [44].

Finally, we like to add that maintenance of health is an active process involving well-balanced neuroendocrine regulatory and immune mechanisms that affect general homeostasis and are integrated at brain levels. The World Health Organization defines health as “a state of complete physical, mental, and social well-being and not merely the absence of disease or infirmity” [146]. In fact, health maintenance is at least as important as protection from disease.

The authors have no conflicts of interest to declare.

The work of the authors quoted in this contribution has been supported by different grants from the Swiss National Science Foundation and the Deutsche Forschungsgemeinschaft.

H. Besedovsky wrote the first draft of the manuscript and collected the work published by other groups. A. del Rey contributed to write the draft, edited the final version, and designed the figures.