Septic serum induces glucocorticoid resistance and modifies the expression of glucocorticoid isoforms receptors: a prospective cohort study and in vitro experimental assay

Sepsis is a leading cause of death in intensive care units (ICUs) around the world [1]. In spite of new developments in critical care and sepsis therapy, the mortality rates associated with severe sepsis and septic shock are still over 30% in most reports [1‐5].

A protective role for glucocorticoid (GC) therapy in animal models of sepsis was shown many decades ago [6, 7], leading to a proposal as early as 1940 for the use of GCs to treat patients with severe sepsis [8]. However this therapeutic strategy has had several dramatic shifts with time. The early approach of high dose GCs for sepsis therapy was abandoned when the potential benefits initially reported [9‐11] could not be replicated, and higher mortality associated with secondary infections was suggested [12, 13]. A renewed interest in GC therapy in sepsis, at what has been termed a physiological dose, was seen after reports of improved response to vasopressor drugs and decreased mortality in selected groups of patients who had an inadequate response to adrenocorticotropin hormone (ACTH) (defined as an increase in total plasma cortisol <9 μg/dL) [14, 15]. However, in 2008, the CORTICUS study [16] did not confirm these results. This contradictory evidence still leaves much uncertainty about the real benefits of GC therapy in sepsis [17‐19]. The most accepted conclusion from earlier studies was that positive GC effects in sepsis were seen only in a restricted group of patients. Until recently, these patients had been identified through a reduced ACTH response, which has been termed adrenal insufficiency or inadequate response to stress [14]. This selection method and its role as a prognostic factor has also been challenged [20‐25]. Therefore it is possible that to demonstrate any benefits for GC therapy in sepsis, it will be necessary to define further to whom GC therapy should be prescribed.

It is possible that target cells in some patients with appropriate cortisol plasma levels do not fully respond to GCs, being less sensitive to a given GC concentration - a phenomenon called GC resistance [26‐28]. These patients, if they exist, would not be identified by the ACTH response test but they could benefit from steroid therapy.

The actions of GCs are mediated by their receptor (GR), which acts as a ligand-dependent transcription factor [29‐31]. Although the GR is the product of a single gene, several isoforms have been described [32, 33]. The most studied are GR alpha (GRα), the classical receptor that mediates GC actions, and GR beta (GRβ), which can act as a dominant negative [34‐37]. A change in the expression of the GR isoforms with greater GRβ expression has been proposed as a mechanism of GC resistance in chronic inflammatory conditions such as rheumatoid arthritis, ulcerative colitis and others [38‐47]. Previous studies of the effect of sepsis on GR expression in humans and endotoxin-treated rats have shown apparently contradictory results in different cell types and tissues, leading the authors to suggest a cell-specific GR modulation of expression in sepsis [48‐52]. However, some of the above studies have evaluated only the mRNA levels of the GR, which do not necessarily correlate with the receptor protein expression [47], while others have evaluated the GR protein expression using antibodies that do not discriminate between GRα and GRβ or by GC binding assays, which evaluate only GRα [51].

It is known that normal mononuclear cells (lymphocytes and monocytes) are sensitive to the actions of GCs. These hormones can influence the T helper 1 and 2 cells balance and modulate cytokine production, which subsequently influences the inflammatory response in sepsis. It is therefore important to know if there is a change in GC receptor expression and/or sensitivity in mononuclear cells in sepsis.

It has been demonstrated that inflammatory cytokines in vitro can preferentially increase the expression of GRβ in lymphocytic cell lines and induce GC resistance [51]. It is possible that proinflammatory cytokines or other factors present in the serum of septic patients may induce a greater expression of GRβ leading to a GC-resistant state. This response could be an appropriate feedback loop in infectious diseases but it also might lead to an unbalanced inflammatory response, which contributes to the morbidity of sepsis.

This hypothesis must be evaluated because, at least in theory, these patients could benefit from GC supplementation to overcome their lower cell GC sensitivity, even in cases where an adequate basal cortisol level or adequate ACTH response can be documented.

The aims of this study were:

1. To evaluate whether there is an overexpression of the dominant-negative GCβ in peripheral mononuclear cells (PBMCs) of septic patients.

The role of GC therapy in septic patients is still controversial [19, 53‐55]. Dose, duration of treatment and selection of patients influences the benefit of the therapy [19, 54]. Insufficient adrenal GC secretion and peripheral GC resistance are factors likely to influence the response of patients to corticosteroids [26‐28]. GC actions are mediated through their cellular receptors. Because a dominant negative effect of GRβ was reported, a potential role for this receptor in GC-resistant states has been proposed [39‐41, 45]. Therefore our study evaluated the expression of GRα and β in septic patients. The expression of GR can have high inter-individual variability [52], so we chose to evaluate the effect of sepsis on the expression of the GR within the same individual. As it is impossible to know in advance who will develop sepsis, we compared the mononuclear GR expression in a septic state (admission to ICU) with the same patient's GR expression at discharge from hospital (clinical recovery). This strategy also allowed us to evaluate if the potential change in expression induced by sepsis is transient or not. However, only the survivors from sepsis could be studied (n = 5).

Our data showed no difference in the expression of GRα in PBMC in sepsis (Figure 1). However it is important to stress that we studied a limited number of septic patients (nine) and only the survivors (five) were included for this analysis. It is possible in a small sample like ours that a difference between groups can not be demonstrated, especially when there is high variability in inter-individual responses. However, even with this limited number of patients, our results showed that, in sepsis, there was a significant transient increased expression of GRβ in PBMC (Figure 1). Ledderose et al. [52] also found an increase of GRβ mRNA expression in T cells from septic patients compared to stimulated or non-stimulated T-cells of healthy donors. That study also had very high inter-individual variability with many overlapping results between groups [19]. Our results showed that all patients had a higher expression of GRβ on the ICU admission day.

In sepsis, there is first a severe inflammatory reaction, followed by an anti-inflammatory phase as a compensatory response, corresponding in general to the third to fifth days of sepsis evolution. Both stages are characterized by different cytokines (TNFα, IL6 and IL1 for the inflammatory stage, and IL10 and TGFβ for the anti-inflammatory stage), with potentially different effects on GR expression. We therefore evaluated if there was a change in GR expression by the third day in ICU. We showed that the levels of expression of both GRα and GRβ observed on ICU admission day were maintained on the third day in the ICU (Figure 2). To study further whether a change in expression of GR isoforms was associated with 28-day mortality rate, we evaluated the changes of GR expression between the first and third days in the individuals who survived versus those who did not survive the septic event. These results were not significantly different (Figure 3). However, it may be of interest that a tendency for an increased expression of GRα on the third day was observed only in the non-survivor group (Figure 3).

It has previously been shown that pro-inflammatory cytokines such as IL1 and TNFα increase the expression of GRα and GRβ in lymphocytic cell lines in vitro, with a stronger effect on the expression of the β isoform [52]. The serum of septic patients has high levels of proinflammatory cytokines, as well as other factors such as GCs, which could at least in part influence the expression of GR. To assess whether soluble factors present in the serum of septic patients induce changes in the cellular expression of GR, we tested the effect of serum taken both from the patients on admission to the ICU and from the control group of healthy individuals, matched by age and sex to the patients, on the expression of GRα and β in cells in culture. Because the PBMC are primarily T lymphocytes but also B-lymphocytes and monocytes, we evaluated the effect of septic serum on T (CEM) and B (Raji) lymphoid cell lines and in a myeloid cell line (K562). Our results showed that, compared to control serum, the septic serum induced an increased expression of both GR isoforms in T and B lymphoid cells, with a stronger effect in the expression of GRβ (Figure 4). This result is similar to the in vitro effect of proinflammatory cytokines on GR [51] and is also similar to the results obtained from PBMC in the same septic patients (Figure 1), supporting a role for factors present in serum during sepsis in the induction of GRβ expression. However, this same serum induced the contrary effect on K562 cells (Figure 4).

We have insufficient information to know whether these results reflect a specific regulatory effect of GCs in different cell types with particular roles in the immune response or if they are secondary to an abnormal behavior of different tumor cell lines. However, we have shown that septic serum can regulate the expression of GR cells in culture. Because the dominant negative effect of GRβ has been proved in vitro only when it has been overexpressed five times or more than GRα, the increased expression of GRβ in itself does not necessarily imply a cell GC-resistant state [34]. We went on to investigate whether septic serum is also able to impair the cell GC sensitivity in cells in culture. Figure 5A shows that dexamethasone significantly inhibits the LPS-induced TNFα secretion from PBMC cultured in the presence of pooled serum from control participants. However, when these cells were cultured in the presence of pooled septic serum, dexamethasone lost its inhibitory effect (Figure 5B). The concentration of dexamethasone used in this assay is greater than the equivalent reported free plasma cortisol levels in septic shock patients treated with low-dose hydrocortisone therapy [56]. Therefore, our results show that septic serum not only influences the expression of cell GR in culture, but can also induce a GC-resistant state.

Finally, to explore further if the serum from patients who survived the septic event had a different effect on the expression of GR and/or GC sensitivity of cells in culture compared to the serum of non-survivors, we tested the effect of the pooled serum obtained on admission to the ICU from both groups of patients on GR expression and on cell GC sensitivity, as described previously. Our results showed that the serum from both survivors and non-survivors induced the same effects in GR expression and in the GC sensitivity test as the pooled serum from all septic patients studied together (data not shown). Further studies with higher number of patients are required to corroborate this data due to the limitations of the small sample evaluated.

We found a transient increased expression of the dominant negative GRβ in PBMC from septic patients. We also demonstrated that serum from septic patients influences the expression of GR transcriptional isoforms α and β in cell lines in culture, increasing the dominant negative GRβ in T and B cell lines. Septic serum also induced a GC-resistant state in PBMC from normal participants in vitro. Therefore our results support the presence of a peripheral GC-resistant state in septic patients associated with a higher expression of GRβ which could be caused, at least in part, by agents present in the serum of septic patients. The small sample of patients and controls (n = 9) included in this work means that further study is required to corroborate our data and to evaluate whether this peripheral GC resistance could be a factor that influences the response to corticosteroid treatment in septic patients.