Hyperosmotic stress enhances cytokine production and decreases phagocytosis in vitro

Hyperglycemia is common in acutely ill patients and may be attributed to the release of stress hormones, peripheral insulin resistance, and certain drugs. Hyperglycemia is an independent risk factor for increased morbidity and mortality in critically ill patients with a variety of conditions. After myocardial infarction, hyperglycemia is associated with an increased risk of in-hospital mortality in patients with or without diabetes [1, 2]. After cardiac surgery, hyperglycemia has been associated with increased postoperative complications [3]. A randomized and prospective trial among 1,548 surgical intensive care unit (ICU) patients receiving mechanical ventilation demonstrated that intensive insulin therapy (IIT) reduced mortality from 20.2% to 10.6% in patients who remained in the ICU for more than 5 days [4]. These results were subsequently partially confirmed in patients admitted to a medical ICU [5].

The mechanisms underlying the detrimental effects of hyperglycemia and the beneficial effects of IIT are not fully understood but may be related to alterations in immune functions and inflammation [6]. Hyperglycemia enhances plasma levels of interleukin (IL)-6 and tumor necrosis factor-alpha (TNF-α) in normal healthy volunteers [7, 8]. Hyperglycemia after myocardial infarction is associated with higher levels of C-reactive protein (CRP) and IL-18 [9], and in the abovementioned study [4] IIT decreased CRP levels significantly [10]. In a rabbit model, strict glycemic control with insulin improved innate immune function such as phagocytosis and prevented excessive inflammation [11]. After endotoxin injection in pigs, insulin treatment reduced the cytokine content of various organs [12].

These in vivo studies indicate an important relationship between hyperglycemia and immune cell function. However, the precise cellular and molecular mechanisms of hyperglycemia and insulin therapy have yet to be fully characterized. In living animals, multiple interacting mechanisms are often difficult to separate. In particular, the protective effect of insulin therapy may be due to insulin alone, the lowering of glucose, its endocrine effects, or a combination thereof. To address this question, we examined the effects of various hyperosmotic substances (glucose and mannitol), as well as insulin, on cytokine production in human peripheral blood mononuclear cells (PBMCs) in vitro. In addition, we investigated the effects of these hyperosmotic substances (glucose and mannitol) on phagocytic and oxidative burst activity in human granulocytes.

This study was approved by the Ethical Committee on Human Research of Charité University Hospital (Berlin, Germany) and was performed in accordance with the Declaration of Helsinki (1964). Informed consent was obtained from all volunteers.

Our results demonstrate that high concentrations of glucose enhance LPS-induced cytokine production in fresh human PBMCs. In addition, glucose decreases phagocytosis and oxidative burst in whole blood. The effects of glucose appear to be mediated by hyperosmotic stress since similar, though weaker, effects could be observed using mannitol. Our findings are in line with observations in animal models of critical illness [11, 12]. In a porcine model, stress hyperglycemia led to a significantly enhanced inflammation as shown by elevated TNF-α and IL-6 levels [12]. This effect could be partly reversed by the addition of insulin. Similarly, burn injury in rabbits [11] leads to stress hyperglycemia and elevation of inflammatory markers. The addition of insulin to maintain normoglycemia resulted in reduced inflammation 3 days after the injury compared with rabbits under hyperglycemia. Experimental hyperglycemia was also shown to elevate circulating cytokine levels in healthy volunteers as well as in diabetic subjects [7]. Interestingly, diabetic patients had significantly higher cytokine levels even before the glucose challenge leading to hyperglycemia. After endotoxin stimulation, the induction of hyperglycemia led to a slight but significant increase in IL-6 in healthy volunteers, whereas TNF-α levels were unaffected [8]. The addition of insulin to reach normoglycemia, however, did not alter the cytokine response in healthy volunteers. In critically ill surgical patients, however, IIT led to a reduced level of inflammation [4]. In this prospective randomized controlled trial involving 1,548 adult patients, van den Berghe and colleagues [4] demonstrated that IIT reduced mortality during intensive care from 8.0% to 4.6%.

The greatest reduction in mortality involved deaths due to multiple-organ failure with a proven septic focus. IIT also reduced overall in-hospital mortality, bloodstream infections, and acute renal failure requiring renal replacement therapy. Although cytokine levels were not measured, the level of inflammation as expressed by the level of CRP was significantly lower in patients under IIT compared with conventional insulin therapy [4]. Furthermore, the rate of secondary infections and the rate of severe sepsis decreased. The reason why critically ill patients under IIT have fewer severe infections is not fully understood. These results, however, indicate that functions of the immune system may be altered.

Several experimental studies investigating the influence of glucose, insulin, and osmotic stress on the immune system on a cellular level have been undertaken [14‐19]. Glucose was shown to induce cytokine production in cell lines [14, 20]. However, cell lines may react differently from freshly isolated cells due to long and artificial culture conditions. Besides the present study, only one group investigated fresh PBMCs and, in agreement with our results, reported enhancement of IL-6 and TNF production by glucose [17]. Osmotic stress by hypertonic saline in PBMCs led to conflicting results. Whereas Shapiro and Dinarello [18] found an elevation of cytokine levels, others report of reduced cytokine production under osmotic stress [15]. The present investigation confirms the results that glucose is able to induce an enhanced cytokine response in fresh human PBMCs. Although high concentrations of glucose have been used in our study, it is important to consider that only viable cells are able to produce and release cytokines. Cytokines are not stored in intracellular compartments but rather are newly synthesized and released in response to inflammatory stimuli [21, 22]. It is thus very unlikely that the observed effects are due to nonspecific toxicity.

There are several potential explanations about the possible underlying molecular mechanism of cytokine induction under increasing levels of glucose and changes in osmolarity. The protein kinase C seems to be involved in the elevated cytokine production by monocytes [14]. Furthermore, it has been suggested that other mitogen-activated protein (MAP) kinases and nuclear factor-kappa-B (NF-κB) are involved as well [23‐25]. Sherry and colleagues [25] could block LPS-induced TNF-α production by inhibition of p38 MAP kinase in diabetic mice. In another study, Németh and colleagues [26] observed an increase in IL-8 production when exposing intestinal epithelial cells to hyperosmotic conditions. This effect seemed to be mediated by activation of p38 and p42/44 MAP kinase as well as an increase in NF-κB [26]. In a sideline to the present study, preliminary data of our group suggest the involvement of p38 MAP kinase.

In a clinical setting, the application of insulin seems to be important as it lowers glucose levels and therefore improves metabolic and osmotic homeostasis. A direct effect of insulin on immune function, however, has not been demonstrated before. In agreement, in the present study, we did not observe that high doses of insulin alone alter the LPS-stimulated cytokine response. The effects of insulin, even in combination with glucose, were rather small and significant in only one experiment. We therefore do not believe that insulin exerts major direct effects on cytokine response.

During hyperglycemic conditions, such as inadequately controlled diabetes, phagocytosis has been reported to be impaired. Clinical trials showed a direct correlation between metabolic control of diabetes and the phagocytic capacity of polymorphonuclear (PMN) cells [27]. These findings are supported by several animal studies: in one study, rabbits with experimental diabetes were randomly assigned to receive insulin treatment followed by a 30% burn injury. In the group receiving insulin, phagocytic activity of monocytes improved by 150%; similarly, a twofold augmentation of oxidative killing compared with controls was observed [11].

In accordance with our findings, several previous studies support the notion that not only hyperglycemia but also changes in osmolarity induce alterations in PMN cell function [19, 27‐29]. Impairment of phagocytic/oxidative burst activity under increasing levels of glucose and changes in osmolarity may have different explanations. Intracellular killing and phagocytosis are partly transmitted via the formation of oxygen species (ROS). Hyperglycemia was shown to inhibit ROS formation in several studies [30]. Alexiewicz and colleagues [31] linked the impaired phagocytic activity to elevated intracellular calcium [Ca²⁺]_I levels as they observed a significant direct relation between [Ca²⁺]_I and the degree of hyperglycemia. They also suggested a glucose-induced calcium influx as well as an increased activity of protein kinase C to be responsible for this effect. They hypothesized that the acute increase in intracellular calcium would deplete the cell's adenosine triphosphate content necessary for the conformational changes during phagocytosis [18, 32]. However, whether the observed in vitro effects of hyperglycemia and hyperosmolarity can explain the reported clinical benefits of IIT remains unknown. A study in healthy human subjects undergoing a 4-hour hyperglycemic or hyperinsulinemic euglycemic clamp test demonstrated no significant changes in PMN cell function [33], indicating that a short acute hyperglycemia may not be sufficient to provoke changes in PMN cell function. Only a more complete understanding of innate immunity during alterations of the host's physiologic milieu will enable the correlation of the observed immunologic and metabolic alterations to the clinical outcome of the patients. Many other aspects of innate immunity, such as neutrophil apoptosis, chemotactic migration of monocytes and neutrophils, complement-mediated cell lysis, the kininogen-bradykinin system, or the role of the mast cell, still remain poorly understood regarding their functional alterations during acute hyperglycemia.

Clinical trials demonstrate considerable evidence for the benefits of IIT in critically ill surgical patients. Our study focused on the in vitro aspects of acute hyperglycemia, hyperinsulinema, and hyperosmotic stress on the human innate immune system in vitro. We could demonstrate that cytokine production by fresh human PBMCs is enhanced under hyperglycemic and hyperosmotic conditions. Furthermore, an impairment of neutrophil phagocytic and oxidative burst activity by glucose was shown. Our in vitro findings may help to explain the beneficial effects of IIT not only in diabetic but also in critically ill patients.