Sepsis remains a leading cause of morbidity and mortality in the intensive care unit (ICU), and one of the top 10 causes of death worldwide 12. The underlying pathophysiological cascade of sepsis is highly complex and far from fully understood 345. From an immunological standpoint, the activation of the complement cascade, a potent arm of the innate immune response, has been associated with fatal outcomes in septic patients 6789. Particularly, the complement anaphylatoxin C5a, a small inflammatory peptide derived from complement activation, has been characterised as a 'key' mediator of sepsis and septic organ dysfunction 1011121314, and was recently labelled as 'too much of a good thing' or to reveal 'a dark side in sepsis' 1516.

Although intentionally beneficial, disproportionate activation of complement during sepsis has been found to contribute to thymocyte and adrenomedullary apoptosis 1718, paralysis of innate immunity 1920, deterioration of the coagulation/fibrinolytic system 21 and multiple organ failure 22. Accordingly, blockade of C5a or its receptors has been shown to prevent multiple organ failure and to greatly attenuate mortality after caecal ligation and puncture (CLP)-induced sepsis 1011141922.

Encephalopathy syndrome is a well described complication of sepsis in the ICU. This phenomenon is thought to represent a consequence of inflammation-mediated dysfunction of the blood-brain barrier (BBB), thus allowing neurotoxic mediators to extravasate from the peripheral circulation into the subarachnoid space or into the brain parenchyma. Noteworthy, the focus of research studies have only addressed in more depth the neuro-inflammatory and metabolic intracerebral changes in sepsis 23242526272829. The complement anaphylatoxin C5a has been characterised as a mediator of BBB dysfunction in a variety of central nervous system (CNS) disorders, including traumatic brain injury and bacterial meningitis 303132. In addition, the detection of the C5a receptor (C5aR) on neurons and the observed upregulation of neuronal C5aR expression under inflammatory conditions 31333435 renders the brain more vulnerable to C5a-mediated neuropathophysiological sequelae secondary to a disruption of the BBB 30313637. The complement cascade has only recently been implicated in the pathophysiology of septic encephalopathy 38.

Based on the established functions of C5a in the pathophysiology of sepsis and on experimental data which imply C5a is a potent mediator of BBB damage and neuroinflammation, we designed the present study to investigate the effect of antibody-mediated C5a-blockade on preventing the development of encephalopathy in experimental sepsis. We hypothesised that blockade of C5a would reverse the dysfunction of the BBB and restore the immunological and endocrinological homeostasis in the septic brain.

Under physiological conditions, the BBB maintains the cerebral micro-environment by tightly regulating the passage of molecules into and out of the brain to protect the brain from microorganisms and neurotoxic substances. During sepsis, however, blood-borne proinflammatory mediators are released, coincidental with diffuse endothelial activation and subsequent upregulation of vascular adhesion molecule-1 (VCAM-1), intracellular adhesion molecule-1 (ICAM-1), E-selectin on cerebral endothelia 42434445, facilitating adhesion and transmigration of neutrophils and monocytes into the brain tissue. Especially the anaphylatoxin C5a is known to be a strong inducer of ICAM-1, VCAM-1 and various selectins 4647484950. In addition, cerebral endothelia produce IL-1β, TNF and IL-6 515253, all of which have been shown to directly induce a disruption of the BBB in vitro 54. Thus, these mediators interact with surrounding brain cells, relaying into the brain inflammatory response and jeopardising the functional integrity of the BBB 555657. In the present study, we found that, during experimental sepsis, the antibody-mediated blockade of anaphylatoxin C5a prevented breakdown of the blood-brain barrier, reducing cerebral and pituitary edema formation, as assessed by extravasation of albumin and EB (Figure 1).

Although the CNS itself was traditionally thought to be immunologically privileged, recent studies have demonstrated that the CNS is a rich source of inflammatory mediators, such as cytokines, chemokines and complement components 58596061626364, and has therefore been termed both 'culprit' and 'victim' during sepsis 57. Thus, during sepsis, the BBB is exposed to harmful proinflammatory mediators deriving from two different compartments, the brain as well as the systemic circulation. This results in an extrinsic, as well as intrinsic, attack on the BBB, accelerating the deterioration of its barrier function. As described above, we demonstrate significant upregulation of pituitary expression of TNF and IL-6 mRNA following CLP (Figure 3), indicating that the pituitary might be an additional source of cerebral proinflammatory mediators. Blood-derived proinflammatory mediators reach the hypophyseal circulation via the anterior hypophyseal arteries and cytokines can diffuse into the pituitary because these areas are free from CNS BBB 65. Therefore, we decided that mRNA analysis for TNF and IL-6 might shed a more accurate light on the origin of these proinflammatory mediators.

Resident cells of the brain, such as neurons, astrocytes and microglia, are capable of synthesising essentially all complement proteins, complement regulatory molecules and complement receptors 316667686970. It has been reported that the pituitary expresses the complement receptors C3aR, C5aR and C5a-like receptor 2 (C5L2), and that these molecules may contribute to regulation of the immune response 7172. We found upregulation of C5aR in the pituitary based on mRNA and protein levels following CLP and reversal of these changes by administration of anti-C5a (Figure 2). Upregulation of C5aR during sepsis has been described in various organs, such as lung, liver, kidney and heart 12. Such upregulation infers that these tissues may develop highly undesirable outcomes after encounters with C5a.

Sepsis is known to induce an abnormal pituitary response 73. Hormonal changes in cortisol, mineralocorticoids, thyroid hormones, GH and vasopressin have all been described during sepsis. Although the acute phase of sepsis is characterised by high levels of GH, GH insufficiency is reported in the late stage of sepsis 7374. In line with these findings, we have found elevated levels of GH protein 24 hours after CLP, while pituitary mRNA expression is significantly reduced, most likely because of a negative feedback mechanism.

Hypercortisolism during the early stages of sepsis is usually followed by cortisol insufficiency in 60% of septic patients 73. Significantly increased levels of corticosterone occurred in rats following CLP (Figure 4). In the current study, pituitary POMC mRNA expression was completely abolished during experimental sepsis (Figure 4). POMC is a polypeptide precursor of multiple molecules, including ACTH and melanocyte-stimulating hormones (MSH) α, β and γ 75. Interestingly, MSH-α has recently emerged as a molecule with potent antiinflammatory effects, which orchestrates descending neurogenic anti-inflammatory pathways and ameliorates the inflammatory response of immune cells 76. At a molecular level, MSH-α decreases the intracellular production of proinflammatory cytokines and chemokines by inhibiting nuclear factor-κB activation and reduces cellular expression of VCAM-1, ICAM-1 and E-selectin 76. During human sepsis, plasma concentrations of MSH-α have been found to be significantly decreased during the early course of the disease and gradually recovered as a function of time 77. More importantly, its concentrations negatively correlated with plasma concentrations of TNF and IL-1β 77. Thus, it is tempting to speculate, whether the observed pituitary upregulation of TNF and IL-6 mRNA (Figure 3) is related to the complete absence of POMC expression (Figure 4), which will result in a lack of pituitary MSH-α production with uninhibited proinflammatory activation of pituitary cells.

It remains to be determined if our findings can be extrapolated into humans. Several reports have stressed the disconnection between rodent and human sepsis 787980, making it difficult to draw definitive conclusions from an experimental study for the clinical setting. Moreover, in the current study, the C5a-neutralising antibody was administered immediately after the CLP procedure. Follow-up studies need to address the effects of delayed anti-C5a treatment, because diagnosis of the sepsis syndrome in patients might be delayed due to several co-morbidities. Thus, it is imperative to address if anti-C5a treatment also reverses BBB and pituitary dysfunction after onset of CLP, which would greatly enhance the therapeutic impact of a potential C5a-blockade in humans.

We describe amelioration of BBB breakdown and partial reversal of pituitary dysfunction by neutralisation of C5a during experimental sepsis. Similarly, blockade of C5aR has recently been described to reduce the LPS-induced activation in the paraventricular nucleus and the central amygdala 81. Thus, as we are beginning to gain novel insights into the crucial role of C5a in the development of sepsis-induced BBB dysfunction, we might be able to immunomodulate its detrimental effects and improve the outcome of septic encephalopathy.

• Anti-C5a prevents break-down of the BBB during experimental sepsis.

• The C5aR is robustly upregulated in the pituitary gland during CLP-induced sepsis.

• Pituitary mRNA expression of proinflammatory TNF and IL-6 is upregulated during experimental sepsis.

• Experimental sepsis induces pituitary dysfunction which is ameliorated by a neutralising anti-C5a antibody.

ACTH: adrenocorticotropic hormone; BBB: blood-brain barrier; C5aR: C5a receptor; C5L2: C5a like receptor 2; CLP: caecal ligation and puncture; CNS: central nervous system; C_T: cycle threshold; EB: Evans Blue; ELISA: enzyme immunosorbent assay; GAPDH: glyceraldehyde 3-phosphate dehydrogenase; GH: growth hormone; ICAM: intracellular adhesion molecule; ICU: intensive care unit; Ig: immunoglobulin; IL: interleukin; MSH: melanocyte-stimulating hormone; PBS: phosphate buffered saline; PCR: polymerase chain reaction; POMC: proopiomelanocortin; RIPA: radio immunoprecipitation assay; TBST: Tris-buffered saline Tween-20; TNF: tumour necrosis factor; VCAM: vascular adhesion molecule

The authors declare that they have no competing interests.

MAF, PFS, MHL, PAW and KI designed the study and supervised the experiments. MAF, DR, AND, LMH and BMT performed the experiments. MAF, PFS, SJM, WRS and KI analysed the data and drafted the manuscript. All authors revised the manuscript for important scientific content, read and approved the final manuscript.