Neutrophil-Derived Hyperresistinemia in Severe Acute Streptococcal Infections¹

Despite adequate antibiotics and modern intensive care, severe sepsis remains a leading cause of morbidity and mortality in intensive care units. Severe sepsis is associated with a mortality rate of ∼30%, which rises to 40–60% if exacerbated by septic shock (1). In these severe systemic infections, the normally tightly controlled balance between the inflammatory, coagulatory, and neuroendocrine systems is lost. Systemic release of numerous inflammatory mediators, such as cytokines, anaphylatoxins, factors of coagulation, and fibrinolysis, has long been recognized as a major event in the pathophysiology of sepsis (2). However, it was recently shown that systemic cytokine responses are more prolonged and heterogeneous in nature than previously thought (3). Furthermore, the up-regulation of proinflammatory mediators that persist for a prolonged time has been suggested to influence the long-term outcomes of sepsis (4).

We recently reported on resistin as a powerful marker of severity of sepsis (5). Resistin belongs to a family of cysteine-rich peptides called resistin-like molecules, discovered in 2001 as an adipocyte-derived hormone that contributes to obesity-associated insulin resistance in mice (6, 7). In humans, resistin has emerged as a potent proinflammatory molecule associated with acute and chronic inflammatory conditions, whereas its role in glucose metabolism remains controversial (8, 9, 10, 11, 12). Most relevant to this study is our recent finding that systemic resistin levels strongly correlate with the severity of sepsis, and the highest concentrations are noted in patients with septic shock (5). This study aims to further our understanding of resistin responses in severe bacterial infections with respect to cellular sources and bacterial triggers. The study focuses on one of the most severe forms of Gram-positive septic shock, i.e., group A streptococcal toxic shock syndrome (STSS).⁴ The results demonstrate hyperresistinemia both locally and systemically in STSS patients and, importantly, identifies neutrophils as a novel dominant source of the mediator in patients with severe acute bacterial infections. In line with this, the recently described neutrophil activator, the streptococcal M1 protein, was found to be a potent trigger of resistin release from neutrophils.

Patients with defined STSS (n = 18) or culture-confirmed Gram-negative septic shock (n = 17) were selected from three studies (5, 13). Serum/plasma samples were collected at several time points during the acute septic episode. Because the STSS patients were part of a placebo-controlled trial of i.v. IgG (13), only samples at baseline before study drug administration were used from the patients that had received i.v. IgG. Consequently, only patients that had received placebo (n = 10) were included in the study of resistin kinetics. Among the 95 patients with septic shock and/or severe sepsis (5), 92 had baseline values for resistin. Baseline neutrophil counts were recorded for 39 of these. For analyses of intracellular resistin in whole blood, five patients with septic shock were included from an ongoing sepsis study at the Karolinska University Hospital (Stockholm, Sweden). Samples from nine healthy volunteers were included as controls.

Ten snap-frozen tissue biopsies from four patients with necrotizing fasciitis and STSS (n = 9 biopsies) as well as one biopsy from a patient with severe cellulitis, all caused by group A streptococci, were used. The biopsy material has previously been described in detail (14, 15). Snap-frozen tissue biopsies from five healthy individuals undergoing reconstructive surgery at Karolinska University Hospital served as controls.

All studies were ethically approved at Karolinska University Hospital, the University of Toronto, Toronto, Canada, or at each participating hospital in the multicenter trial (13). Written informed consent was obtained from all patients or their legal guardians.

Resistin in patient samples or cell culture supernatants was analyzed by either ELISA (BioVendor) or Luminex (BioSource). A selection of representative samples (n = 5) were analyzed by both assays, and essentially identical resistin levels were detected (p = 0.999). IL-8 was measured in a multiplex Luminex analysis (BioSource) together with resistin. Because our patient cohorts included both plasma and serum samples, ELISA resistin levels in serum and plasma samples collected from the same individual at the same time point were analyzed from three septic shock patients and one healthy control (data not shown). No significant difference between serum or plasma samples was seen (p = 1.0). MPO levels were measured in cell culture supernatants by ELISA (Immunology Consultant Laboratory).

Clinical isolates from blood cultures of patients with STSS or Gram-negative septic shock were used in the infection/stimulation assays, including group A streptococcal strains 5448 (M1T1 STSS isolate; Ref. 16) and 08/04 (M1T1 STSS isolate Ref. 13), as well as an Escherichia coli blood isolate from patient h1 in the Gram-negative patient cohort. Supernatants were prepared from overnight cultures of the above-mentioned group A streptococcal strains, as previously described (13). Such supernatants contain a mixture of secreted superantigens and other exotoxins. Supernatants were also prepared from cultures of the AP1 strain and its isogenic mutant MC25, which contains a truncated form of the M1 protein that lacks the transmembrane-spanning region, leading to M1 protein accumulation in the supernatant (17). The M1 protein was purified from the AP1-derived strain MC25 as described (18).

Biopsies were sectioned, fixed in 2% formaldehyde, stained, and analyzed essentially as previously detailed (14, 15). Human neutrophils, purified as described below, were added to adhesion slides and fixed with 2% paraformaldehyde before staining. The following Abs were used: anti-CD68 (DakoCytomation); anti-neutrophil elastase (DakoCytomation); anti-resistin mAb (R&D Systems); anti-MPO (Santa Cruz Biotechnology); anti-lactoferrin (Sigma-Aldrich); and a polyclonal rabbit antiserum specific for the Lancefield group A carbohydrate (Difco) (14, 15). Irrelevant isotype-specific murine Abs (DakoCytomation) were used to control for nonspecific staining reactions. The specificity of the anti-resistin Ab was verified by preincubation of the antiserum with a 20-fold excess of recombinant resistin (BioVendor), which blocked the reactivity of the Ab. The immunohistochemical staining procedure was modified to include an initial blocking step for 30 min at room temperature with 10% FCS in Earl’s balanced salt solution and 0.1% saponin. Despite the sizes of the biopsies, the whole section was analyzed by acquired computerized image analysis (ACIA). The results are presented as ACIA values, which equals the percentage of positively stained area × the mean intensity of positive staining. Single and dual immunofluorescence staining of tissue and cells was evaluated by a Leica confocal scanner TCS SP II coupled to a Leica DMR microscope.

Human neutrophils and PBMC were isolated from peripheral blood collected from healthy individuals by Polymorphprep or Ficoll-Hypaque gradient centrifugation, respectively. Monocytes were enriched through 2 h of plastic adherence followed by infection with bacteria at a high multiplicity of infection (MOI) of 10–20. In the case of neutrophils, the bacteria were preincubated with 10% human serum for 10 min and then mixed with autologous neutrophils at a MOI of 16–27. Neutrophils were also stimulated with different concentrations of the streptococcal M1 protein, LPS (Sigma-Aldrich), PMA (50 ng/ml; Sigma-Aldrich), superantigen-containing supernatants of overnight cultures of group A streptococcal strains, and fixed bacterial strains. Unstimulated cells served as negative controls. At defined times postinfection, the cells were centrifuged and supernatants collected. Triton-X (0.02%) lysates of neutrophils served as controls for the total content of resistin in unstimulated cells.

Intracellular resistin in monocytes and neutrophils was analyzed in whole blood treated with ammonium chloride lysis buffer. The cells were stained with Abs against CD14 PE and CD15 FITC (BD Biosciences) followed by washes, fixation with 2% formaldehyde, and permeabilization with 0.1% saponin in PBS supplemented with 1% HEPES. Cells were then incubated with the monoclonal anti-human resistin Ab labeled with a Zenon Alexa Fluor 647 mouse IgG2b labeling kit from Invitrogen or with an isotype control (BD Phosflow Alexa Fluor 647 Mouse IgG2b κ). Resistin expression in monocytes and granulocytes/neutrophils was analyzed using a FACSCalibur flow cytometer (BD Biosciences) and CellQuest software (BD Biosciences).

Subcellular fractionation of neutrophils was performed by density centrifugation on Percoll gradient as described by Kjedsel et al. (19), with some minor modifications (20). Briefly, polymorphonuclear neutrophils were purified and adjusted to 5 × 10⁷ cells/ml, followed by their disruption by nitrogen cavitation. After nuclei and unbroken cells were sedimented by centrifugation, the postnuclear supernatant was separated on Percoll gradients and the localization of subcellular organelles in the gradients was determined by marker analysis of the fractions. Relative amounts of MPO (marker for the azurophil granules), lactoferrin (specific granules), and CD35 (secretory vesicles) were tested by Western blotting for MPO and lactoferrin and by dot blot analysis for CD35 using Abs against respective marker, as described elsewhere (21).

Data were analyzed by GraphPad Prism version 4.0 for Windows (GraphPad software). Mann-Whitney U test was used for comparison between groups. Correlations between variables were determined by use of Pearson correlation test or, in the case of non-Gaussian distribution of the data, Spearman rank correlation coefficient. Differences were considered significant when p < 0.05.

To assess resistin levels in the circulation of STSS patients during the acute phase of infection, a cohort of patients with STSS was compared with a Gram-negative septic shock cohort, the latter of which was dominated by E. coli and Klebsiella. The two cohorts were well matched with respect to age, gender, and severity of infection based on a sepsis-associated organ failure assessment (SOFA) score at the day of enrollment in the study (STSS: median 11.50, range 6–19; G-negative cohort: median 13.00, range 5–19). Analyses of systemic resistin levels confirmed our previous finding (5) that septic shock patients have markedly elevated levels (28-fold higher) as compared with healthy controls (p < 0.0001) (Fig. 1,a). STSS patients had equally high levels of resistin in circulation as the Gram-negative septic shock cohort. Plasma samples in the STSS cohort were collected daily during the acute septic episode and revealed a prolonged hyperresistinemia (Fig. 1,b). Strikingly, even after 28 days of infection, high systemic resistin levels (median, 25 ng/ml; range, 6.9–55) could be detected in STSS patients, which is 7-fold higher than the levels in healthy controls (p < 0.0002). Comparison to resistin levels in the Gram-negative cohort revealed that this cohort had slightly, but not significantly, lower levels throughout the first week as compared with the STSS cohort (Fig. 1 b). The final time point for serum collection in the Gram-negative septic shock cohort was at day 14. In these samples, resistin was found to be lower (median, 13 ng/ml; range, 3–29) than the values detected at day 28 in STSS patients.

To assess resistin responses at the local site of streptococcal infection, snap-frozen tissue biopsy material collected from the epicenter of infection of patients with severe group A streptococcal tissue infections, i.e., necrotizing fasciitis or severe cellulitis, was used. The tissue biopsies were analyzed for the expression of resistin by intracellular immunohistochemical staining and ACIA (Fig. 2). Resistin was detected in all infected biopsies analyzed and showed distinct cytoplasmic staining within a large percentage of positive cells (Fig. 2, b and f). The tissue was further characterized with respect to bacterial load by staining for group A streptococci (Fig. 2, a and f) and inflammatory cell infiltration, including neutrophils (Fig. 2, c and f) and macrophages (Fig. 2, d and f). Image analyses of the stainings revealed a positive correlation between bacterial load and resistin levels in the tissue (r = 0.79, p = 0.0069; Pearson’s correlation). Staining of resistin and inflammatory cells in skin biopsies of healthy controls revealed macrophages scattered throughout the epidermis and dermis in all biopsies, whereas only a few cells stained positive for resistin and neutrophils. Comparison of image analysis data revealed significantly higher resistin levels and neutrophil and macrophage infiltration in the streptococcal infected tissue as compared with the healthy controls (p < 0.038) (Fig. 2 g). Hence, the results demonstrate that there is a pronounced hyperresistinemia in streptococcal tissue infections that correlates with bacterial load.

The source of resistin has been a matter of debate, as the initial studies were performed in mice in which resistin is produced by adipocytes (6). Subsequent studies in humans implicated monocytes as the main sources of resistin (22, 23, 24). In light of this, it seemed relevant to determine whether monocytes/macrophages were responsible for the pronounced resistin release seen during acute bacterial infections. To identify resistin-producing cells, tissue biopsies were immunostained for resistin in combination with cell markers and analyzed by confocal microscopy. Initial analyses focused on macrophages, and the results showed that macrophages are indeed a source of resistin in the tissue during these infections, as resistin-positive granules could be detected in CD68-positive cells (Fig. 2,h). However, a significant population of resistin-positive cells, which were not CD68 positive, was evident in the tissue (Fig. 2,h; indicated by arrow), inferring another source of resistin. Subsequent analyses of potential cell types identified these cells as neutrophils (Fig. 2,i). Quantification of the percentage of neutrophils vs macrophages in the resistin-positive cell population in the tissue revealed that on average 87% of the resistin-positive cells were neutrophils as compared with 34% of the macrophages. Consistent with this finding, in situ image values of resistin showed a positive correlation with neutrophil infiltration in the tissue (p < 0.012) (Fig. 2,j). As shown in Fig. 2 f, the two infected biopsies with low resistin levels also had the lowest neutrophil counts. Similar results were obtained in healthy skin biopsies, where the majority of the few resistin-positive cells found were identified as neutrophils. Hence, these studies revealed a novel significant source contributing to the hyperresistinemia present at the inflamed tissue site of infection.

To assess whether these cell types are also the ones responsible for the systemic resistin previously found to be associated with severity of sepsis (5), acute phase whole blood from septic shock patients were analyzed by flow cytometry. Such samples were not available from the above-described patients but were obtained through an ongoing septic shock study at our hospital. To this end, we received whole blood samples from five consecutively enrolled septic shock patients that were immediately prepared and analyzed for resistin-producing cells in defined cell populations. As in the case of locally produced resistin, the vast majority of resistin-positive cells in whole blood were found in the neutrophil population (ranging from 70 to 98%), whereas the corresponding monocyte frequencies ranged from 0.5 to 3% (Fig. 3 a). This was further supported by analyses of an extended severe sepsis/septic shock cohort (n = 39), which revealed a significant correlation between serum resistin levels and neutrophil counts at the day of diagnosis (r = 0.53, p = 0.0005; Spearman rank order test).

Whole blood of healthy controls was analyzed at the same time as that of the septic shock patients and showed a similar distribution, with 60–96% and 0.1–3% of resistin-positive neutrophils and monocytes, respectively (Fig. 3,a). Hence, in some individuals there was a subpopulation of resistin-positive cells that were neither CD15- nor CD14-positive. The identity of these cells is at present unknown. However, simultaneous analyses of systemic resistin measured in the serum of the five patients and the healthy controls revealed significantly higher levels in the sepsis patients compared with the healthy controls (p < 0.0079) (Fig. 3 a). It is worth noting that the two highest values of resistin (122 and 153 ng/ml) were found in patients 4 and 5, who died within 24–48 h of sampling. Taken together, the data suggested that resistin exists preformed in neutrophils and macrophages and can be rapidly released upon activation and degranulation. The total amount of resistin within neutrophils was assessed by Triton X-100 lysis of neutrophils isolated from healthy individuals (n = 8). The results demonstrated that there was an interindividual variation in the overall cellular content of resistin, with values ranging from 4.4 to 9.6 ng per 10⁶ neutrophils.

To determine the subcellular location of resistin, the cell membrane of neutrophils obtained from healthy donors was disrupted and the resulting cellular content was subjected to density centrifugation on a Percoll gradient. Fractions were collected and investigated by Western blotting or dot blot analysis using Abs against MPO (a marker protein for azurophilic granules), lactoferrin (specific granules), and CD35 (secretory vesicles). Assessment of resistin in each fraction revealed a complete overlap with resistin and MPO-containing fractions (Fig. 3,b). However, resistin could also be detected, although at lower levels, in fractions containing lactoferrin (Fig. 3,b), thus indicating that resistin may be stored within azurophilic and specific granules. To address this further, primary neutrophils were stained for resistin and granule marker proteins and evaluated by confocal microscopy. Both lactoferrin- and MPO-positive granules were more abundant than the resistin-positive ones (Fig. 3 c). However, 100% of the resistin-positive granules were double positive for MPO, whereas no colocalization with lactoferrin was seen. The fact that there was a MPO-positive subpopulation that lacks resistin may reflect the previously described heterogeneity in azurophilic granules (25).

To further investigate the resistin response during bacterial infections and to identify bacterial factors that induce resistin release, an in vitro stimulation assay was established. To mimic the clinical setting, primary neutrophils from healthy donors were exposed to high numbers (MOI, 16–27) of either group A streptococci or the Gram-negative bacterium E. coli for various spans. Both of the isolates used were obtained from septic shock patients. Levels of resistin were subsequently measured in culture supernatants and the data revealed that both clinical isolates triggered a significant release of resistin from primary neutrophils (Fig. 4,a). To identify the bacterial components responsible for this induction, neutrophils were stimulated with an array of bacterial factors, including the classical sepsis-associated factors LPS and the superantigen-containing supernatant of an overnight culture of group A streptococci, fixed Gram-negative and Gram-positive bacteria (proinflammatory cell wall components), and the streptococcal M1 protein (Fig. 4,b). Stimulation of neutrophils from different donors revealed that no resistin was induced by a superantigen-containing streptococcal supernatant, and only low responses were triggered by LPS in doses of up to 100 ng/ml. When formaldehyde-fixed bacteria were used as stimuli, group A streptococcus was found to be the most potent inducer of resistin release, as compared with fixed E. coli or a Staphylococcus aureus strain (p < 0.001; Tukey’s multiple comparison test). Another prominent inducer of resistin release from neutrophils was the M1 protein of group A streptococci. In fact, both the M1 protein and fixed group A streptococci triggered a resistin response equal to that of the positive control, the phorbol ester PMA (Fig. 4,b). Assessment of resistin levels in Triton X-100-lysed neutrophils demonstrated that ∼60% of the resistin stored in neutrophils is mobilized in response to the M1 protein (Fig. 4,b). Simultaneous analyses of IL-8 in the culture supernatants revealed that all stimuli, including the streptococcal culture supernatant that failed to trigger resistin release, induced IL-8 production and release (Fig. 4,b). Interestingly, stimulation with fixed bacteria showed that fixed E. coli induced an equal IL-8 response as the fixed group A streptococci, despite the noted difference in the induction of resistin release. Results from Triton X-100-lysed neutrophils demonstrated high levels of resistin but only low amounts of IL-8. This is consistent with the hypothesis that resistin exists preformed and stored in the azurophilic granules, whereas IL-8 is up-regulated in response to bacterial components or other stimuli. To address this further, MPO was measured in supernatants from activated neutrophils and found to be coreleased with resistin (Fig. 4 c).

A dose-response experiment with the M1 protein showed that maximum release of resistin was achieved at 1 μg/ml, and in some donors the response even declined when higher doses of the M1 protein was used (Fig. 4,d). This dose dependency of M1 protein-triggered activation of neutrophils has previously been described and found to be associated with the ability of soluble M1 protein to form complexes with fibrinogen (26). To assess this further, we used the isogenic AP1 mutant strain MC25, which has a truncated M1 protein lacking the COOH-terminal cell wall-anchoring motif (17). Therefore MC25 has no M1 protein on its surface. The anchorless protein is released and accumulates in the growth medium. Culture supernatants from MC25 were found to be potent triggers of resistin release from neutrophils, whereas supernatants from its wild-type AP1 strain did not (Fig. 4,e). Resistin release was also tested in the absence of fibrinogen by the removal of plasma from the cultures. This resulted in a complete loss of M1 protein-triggered responses, whereas PMA-induced responses were equal in the presence or absence of plasma (Fig. 4 e). Taken together, the data strongly suggest that M1 protein-triggered resistin release from neutrophils involves a complex formation with fibrinogen.

In contrast to neutrophils, monocyte-enriched PBMC isolated from healthy donors did not release detectable levels of resistin following infection with either group A streptococci, or E. coli, nor following stimulation with the M1 protein (data not shown).

Originally described as an adipokine associated with obesity-induced insulin resistance (6), resistin has recently emerged as a potent proinflammatory protein of potential clinical relevance for acute (5) and chronic inflammatory conditions (10, 11). This study, which focuses on resistin in severe invasive bacterial infections, was initially triggered by our previous finding that resistin levels in septic shock may differ depending on the causative microorganism (5). We now explore this further by the use of defined cohorts of patients with either STSS or patients with septic shock caused by Gram-negative bacteria. The former was selected as a Gram-positive bacterial cohort because it represents one of the most severe forms of Gram-positive sepsis in immunocompetent individuals, as evident by mortality rates commonly exceeding 50%. A comparison of systemic resistin responses in these patient cohorts demonstrated that STSS patients had equally high serum resistin levels at enrollment as compared with those with Gram-negative septic shock. Importantly, the analyses revealed that the elevated resistin levels persist over a prolonged time in both cohorts, and markedly elevated levels were noted even at later time points. Interestingly, resistin levels measured in day 28 samples from the STSS patients were higher (almost 2-fold) than the day 14 samples, i.e., the latest time point available from the Gram-negative cohort. These data are in agreement with the results of the bacterial stimulation assays in which the highest resistin levels were consistently seen when group A streptococcal stimuli (fixed bacteria or M1 protein) were used, whereas LPS or fixed E. coli proved to be fairly poor inducers of resistin. However, the question of whether Gram-positive toxic shock is associated with a more pronounced resistin response than Gram-negative septic shock remains to be answered, preferably by analyses of larger patient cohorts enrolled in the same study.

Although not yet demonstrated, it is natural to speculate that such a prolonged elevation of systemic resistin, with known proinflammatory activity, could play a central role in maintaining an inflammatory response believed to contribute to long-term mortality in sepsis. Our current patient material was too limited to allow for such conclusions, but it is noteworthy that in the STSS cohort six of seven patients had elevated resistin levels (using a conservative threshold of >15 ng/ml) 28 days after onset of the septic episode and, among those, four also demonstrated elevated IL-8 responses at this late time point (median, 1203 pg/ml; range, 183–11,133). Furthermore, preliminary in vitro experiments showed that although IL-8 (1–100 ng/ml) did not induce resistin, pretreatment with IL-8 increased resistin release by streptococcal M1 protein by ∼20% (data not shown).

Another intriguing finding of the current study arose from the studies of resistin responses at the local tissue site of infection in patients with severe group A streptococcal infections. These studies revealed that there is pronounced hyperresistinemia in streptococcal tissue infections and, most importantly, a novel significant source of resistin in these infections was identified. The source of resistin has been a matter of debate, as the initial studies were performed in mice in which resistin is produced by adipocytes (6). Subsequent studies in humans implicated monocytes as the main source of resistin (22, 23, 24). In contrast, our analyses of resistin-positive cells in group A streptococcus-infected patient tissue showed that although resistin-positive tissue macrophages could be identified, neutrophils represented the dominant source at the inflamed tissue site. Similarly, analyses of intracellular resistin in the whole blood of sepsis patients revealed that the majority of resistin-positive cells in circulation were neutrophils, indicating that this is likely the source for the pronounced systemic hyperresistinemia in severe sepsis/septic shock patients. Triton X-100 lysis of primary neutrophils of healthy donors and the subsequent measurement of resistin showed a total cellular content of >4 ng/10⁶ neutrophils. Hence the concentrations seen in the circulation (up to 300 ng/ml) of septic shock patients can readily be sustained by the mobilization and degranulation of neutrophils, especially in light of the pronounced neutrophilia (>10 × 10⁹ neutrophils/L), which is common in these patients. It is noteworthy that there was a marked interindividual variation in the total resistin content ranging from 4 to 10 ng/10⁶ neutrophils. Similarly, the flow cytometry data, i.e., mean fluorescent intensity values in individuals analyzed simultaneously, also indicated such an interindividual variation. This finding raises the question of whether such differences in resistin content could represent a predisposing factor for developing septic shock.

Further studies linked resistin to the azurophilic granules. Azurophilic granules are, of the different granule types, the latest to be mobilized, which could explain the accumulation and persistence of resistin at the infectious site. In contrast, the heparin-binding protein, highly expressed in severe streptococcal infections and linked to acute vascular leakage, is stored within the secretory vesicles, which are the most readily mobilized granules (20). Taken together, these findings are in line with recent studies underscoring the central role of streptococcus-triggered neutrophil activation and degranulation in the induction of shock and acute organ injury (21, 26). In vitro studies of neutrophil and resistin responses confirmed that neutrophils represent a major source of the resistin mobilized in response to bacterial infection. Interestingly, neither purified LPS nor superantigen-containing supernatant induced a significant resistin release from human neutrophils, whereas infection with live clinical streptococcal or E. coli isolates resulted in high levels of soluble resistin. Thus, bacterial triggers other than the classical sepsis-associated bacterial factors LPS and superantigen must be involved. One likely candidate was the streptococcal M1 protein that has recently received a lot of attention due to its immunostimulatory activity (18, 26, 27). Most pertinent to this study is its role as a mediator of potent activation and degranulation of neutrophils that result in vascular leakage and septic shock as well as acute lung injury (21, 26). This occurs when soluble M1 protein forms complexes with fibrinogen, which are potent activators of neutrophils through interaction with β₂-integrins. We now demonstrate that the streptococcal M1 protein is a prominent trigger of resistin release from neutrophils, and we show that the response is dependent on the presence of fibrinogen-containing plasma, implicating the above described mechanism in these responses.

The physiological function and regulation of resistin in humans is still not fully defined. Patel et al. (23) studied the regulation of expression of resistin in a range of human tissues and showed that the highest expression was in bone marrow. Resistin has been shown to be up-regulated during monocyte differentiation to macrophages (22, 23), as well as by proinflammatory cytokines (11, 28). At the transcriptional level, resistin expression is regulated by peroxisome proliferator-activated receptor γ (PPARγ) activators and can be efficiently blocked by the PPARγ agonist rosiglitazone (23). As previously mentioned, resistin has been ascribed numerous proinflammatory properties in humans, whereas its role in glucose metabolism and homeostasis remains controversial. Furthermore, two recent reports described immunosuppressive effects of resistin on immune cells, including interference with chemotactic and oxidative responses of neutrophils (29), as well as suppression of Ag uptake and presentation by dendritic cells (30). Yet another level of complexity is added by the fact that resistin has a tendency to form oligomers both in vitro and in vivo (31, 32, 33). This oligomerization has been suggested to affect resistin’s biological functions and hence will be an important factor to consider in future studies. In light of sepsis, several of the properties ascribed to resistin are important attributes in the pathophysiology of severe sepsis and septic shock, including the induction of proinflammatory responses (9, 11) as well as endothelial cell activation (34). The latter involves resistin-induced release of endothelin-1, as well as up-regulation of adhesion molecules such as VCAM-1 and ICAM-1 (34, 35). Endothelin-1 has been strongly linked to morbidity and mortality in sepsis and may contribute to organ dysfunction in septic shock (36). Hence, there are many plausible and not solely exclusive mechanisms by which resistin may contribute to sepsis. The findings presented in this report emphasize the importance of neutrophils as dominant sources of resistin during severe bacterial infections and place resistin among the potentially important neutrophil granule proteins that may contribute to the pathogenesis of inflammatory diseases. Further studies will be necessary to define the role of resistin in neutrophil function in normal immune responses and in sepsis.

We acknowledge Prof. Donald E. Low for providing the tissue biopsy material and research nurse Gunilla Herman for diligent collection of samples and data in the sepsis studies as well as the excellent technical assistance of Anette Hofmann, Monica Heidenholm, Erik Wennerberg, and Hernan Concha.

The authors have no financial conflict of interest.