α1-Acid Glycoprotein Decreases Neutrophil Migration and Increases Susceptibility to Sepsis in Diabetic Mice

Care and treatment of animals was based on the Guide for the Care and Use of Laboratory Animals (18), and research was approved by the Animal Research Ethics Committee of the School of Medicine Ribeirão Preto. Balb/C and C57BL/6 (8–10 weeks old) mice were housed in an adequate animal room at 23–25°C, with free access to water and food.

Diabetes was induced by a single intravenous injection of 50 mg/kg body wt alloxan (Sigma) in saline (100 μL). The control group received an intravenous injection of saline (100 μL). Diabetic mice with a blood glucose level ≥300 mg/dL 3 days after alloxan treatment were used. In the fifth day, alloxan-treated animals were treated subcutaneously daily with saline or insulin for 5 days. One dose of bovine insulin (Sigma) was administered in the morning, and one dose of insulin glargine (long-acting insulin; sanofi-aventis) was administered at 5 p.m. Both were administered at a dose of 1 UI for each 200 mg/dL blood glucose. Blood samples were collected from the tail vein to measure blood glucose levels by a glucometer (Precision Xtra; Abbott). Ten days after alloxan treatment, mice were submitted to sham operation (SH), mild sepsis (MS), or severe sepsis (SS) by the cecum ligation and puncture (CLP) model and several parameters were analyzed. In another set of experiments, at day 10 after alloxan treatment, the mice were used for neutrophil migration assay in vivo or for blood neutrophil chemotaxis assay in vitro.

Sepsis was induced by CLP (19). Two punctures were made in the cecum using a 21-gauge needle to induce MS or using an 18-gauge needle to induce SS. SH animals were submitted to laparotomy without cecal puncture.

Neutrophil migration to peritoneal cavity or to lung cavity was evaluated 6 h after sepsis induction or 6 h after intraperitoneal injection of macrophage inflammatory protein (MIP)-2 (chemokine [C-X-C motif] ligand (CXCL)-2, 3–30 ng/mice; R&D Systems).

Extent of leukocyte accumulation in the lung was measured 6 h after sepsis by myeloperoxidase activity as previously described (20).

Bacterial count was determined 6 h after sepsis in peritoneal cavity and blood as previously described (21).

Using colorimetric assay kits, serum alanine aminotransferase and aspartate aminotransferase (Labtest Diagnóstica, Lagoa Santa, Brazil) levels were estimated 6 h after sepsis.

Using an ELISA assay kit (Millipore), serum insulin levels were estimated 6 h after sepsis.

Briefly, mice were anesthetized with tribromoethanol (250 mg/kg i.v.) and mesenteric tissue was withdrawn for intravital microscopic examination 3 h after surgery as previously described (20).

Blood neutrophils were isolated by Percoll gradient as previously described (14). Next, neutrophil chemotaxis assay (10⁶ cells/mL in RPMI-1640, 0.01% BSA, 97% of viable cells, assessed by trypan blue) was performed using a 48-well modified Boyden chamber (Neuro Probe) in response to RPMI-1640 or MIP-2 (10 or 30 ng/mL).

AGP was measured in serum by an immunoturbidimetric method using a kit manufactured by Quibasa Química Básica (Belo Horizonte, Minas Gerais, Brazil).

MIP-2, keratinocyte-derived chemokine (KC), and interleukin (IL)-6 levels were detected in serum and peritoneal cavity by ELISA 6 h after surgery.

Mouse blood samples were stained with rat anti-mouse R-phycoethrin-CD62L (MEL-14), fluorescein isothiocyanate–CD11b (M1/70), Peridinin-chlorophyll proteins (PerCP)-Cy5.5-GR1 (Ly-6G) (BD Bioscience), and R-phycoethrin-CXCR2 (R&D Systems) monoclonal antibodies or appropriate isotype controls. After staining, cells were fixed, washed, and analyzed using a flow cytometer (FACSort; Becton Dickinson).

Neutrophils were affixed on glass slides and then incubated with rabbit anti-mouse GRK2 antibody or isotype control (Santa Cruz Biotechnology). Next, Alexa-Fluor 594–conjugated goat anti-rabbit IgG antibody (Invitrogen) was added. Cells were incubated with DAPI (Invitrogen) to stain the cell nucleus. Microscopic analysis of fluorescent images was performed using an Olympus BX-50 epifluorescence microscope.

Four hours after surgery, mice were killed and the liver was collected. Total RNA was extracted with TRIzol reagent (Invitrogen) using instructions from the manufacturer. Reverse transcription of total RNA to cDNA was carried out with reverse transcription reaction (Superscript II; Gibco Life Technologies). Real-time PCR quantitative mRNA analysis was performed in an ABI Prism 7500 Sequence Detection System using the SYBR–green fluorescence system (Applied Biosystems) as indicated by the manufacturer. Data were analyzed with the comparative cycle threshold method. Primer pairs for mouse β-actin and AGP were as follows: AGP forward, 5′-CACAGCAAATAGTTCAACCA-3; AGP reverse, 5′-CAAAGGTTTCTACTCCTCCT-3′; β-actin forward, 5′-AGCT-GCGTTTTACACCCTTT-3′; and β-actin reverse, 5′-AAGCCATGCCAATGTTGTCT-3′. All samples were submitted to dissociation characteristics of double-stranded DNA during heating (melting curve analysis). To validate and standardize the method, we used positive or negative known control samples for both AGP and β-actin genes.

Data are reported as means ± SEM (except for survival curves) and are representative of two or three independent experiments with five mice in each group. Differences between two unpaired groups were compared by Student t test. The mean differences between three or more groups within an experiment were compared by ANOVA. When significant differences were identified, individual comparisons were tested subsequently with Bonferroni t test. Statistical significance was set at P < 0.05. Survival rates (n = 10–12) were expressed as percentages, and a log-rank test (χ² test) was used to examine differences between survival curves.

Alloxan-induced diabetes has commonly been used as an animal model of insulin-dependent diabetes (22–24). Compared with those in control nondiabetic mice, blood glucose levels were found to be elevated in alloxan-treated mice (Fig. 1A). Moreover, body weight gain was observed in nondiabetic mice during the observation period (10 days, 2 ± 0.2 g), whereas diabetic mice presented significant weight loss (−3.2 ± 1.4 g). Insulin-replacement therapy reversed the state of metabolic catabolism, as observed by decreased blood glucose levels (Fig. 1A) and recovery of weight gain: increase of 0.25 ± 0.31 g. Taken together, these data confirmed the presence of severe diabetes in this mouse model.

We next investigated the influence of alloxan treatment on sepsis outcome. Diabetic mice submitted to MS (10 days after alloxan treatment) showed 100% mortality after 4 days, whereas ∼85% of control mice survived until 7 days after surgery (Fig. 1B). Induction of MS 5 days after alloxan treatment also decreases survival rate compared with nondiabetic mice (Supplementary Fig. 1A). All SH control and diabetic mice survived during the observation period (data not show). As previously demonstrated (21), all nondiabetic mice submitted to SS died during the 3 day period after sepsis induction. As expected, diabetic mice exposed to SS died faster than nondiabetic mice (Supplementary Fig. 1B). Similar effects on blood glucose, weight loss, and mortality were also observed in a different mouse strain (Supplementary Fig. 2A–C). In accordance with survival data, diabetic mice subjected to MS or control mice subjected to SS failed to control the infection, as demonstrated by the higher number of colony-forming units (CFUs) in peritoneal cavity and blood compared with MS control mice (Fig. 1C and D).

Deficiencies in neutrophil migration have been described in experimental diabetic models (7). Here, we observed that diabetic animals with MS displayed a significant impairment of neutrophil migration into the peritoneal cavity, which was similar to that observed in nondiabetic mice subjected to SS (Fig. 1E).

The reduction of neutrophil migration to the infection site in diabetic mice was not a consequence of low concentration of neutrophil chemotactic cytokine in the focus of infection. The levels of MIP-2 (CXCL2) in the peritoneal cavities were 77.2 ± 18.0 pg/mL (SH), 423.4 ± 132.6 pg/mL (MS), and 1,893 ± 85.9 pg/mL (SS) for control mice (n = 5) and 99.6 ± 32.0 pg/mL (SH) and 2,017 ± 51.6 pg/mL (MS) for diabetic mice (n = 5). The peritoneal levels of tumor necrosis factor-α were 28.0 ± 3.0 pg/mL (SH), 119.3 ± 28.2 pg/mL (MS), and 505.8 ± 87.5 pg/mL (SS) for control mice (n = 5) and 25 ± 9.5 pg/mL (SH) and 310.8 ± 45.4 pg/mL (MS) for diabetic mice.

The elevated serum glucose levels and/or low levels of insulin associated with diabetes may be responsible for disturbance in the host immune response (25). Therefore, we subjected diabetic mice to insulin therapy followed by CLP. Expectedly, insulin treatment significantly prevented high mortality rate of diabetic MS mice (Fig. 1B), decreased the bacterial load in the focus of infection and blood (Fig. 1C and D), and improved neutrophil migration toward the infection site (Fig. 1E).

To investigate whether the observed effects of insulin treatment in diabetic septic mice were specific to a diabetes-associated phenomenon, we also evaluated the outcome of control nondiabetic mice exposed to insulin during sepsis. Nondiabetic animals subjected to SH, MS, or SS did not present significant changes in serum levels of insulin 6 h after surgery (Supplementary Fig. 3A). As expected, alloxan-treated mice displayed reduced serum insulin amounts, which increased during treatment.

After 2 h postsepsis, all animal groups displayed similar blood glucose concentration, indicating that the anesthesia (ketamine/xilasine) rather than sepsis induced hyperglycemia, as previously published by Saha et al. (26). After the observed hyperglycemia, nondiabetic septic, but not sham or naïve animals, displayed low levels of blood glucose (Supplementary Fig. 3B and data not shown). In contrast with the observed effect on diabetic septic mice (Fig. 1B), insulin administration (2 or 1 UI s.c. 30 min after surgery) in nondiabetic septic mice promoted significant mortality rates (Supplementary Fig. 3C–F), which could be explained by exacerbation of hypoglycemic state, i.e., treatment of already hypoglycemic mice with insulin (27). Together, these data suggest that specific defects associated with diabetes enhance sepsis sensitivity, which can be rescued by insulin treatment.

High levels of systemic inflammatory cytokines, as well as leukocyte sequestration into remote tissues of infection site, were correlated with the severity of sepsis (13). Therefore, we determined leukocyte sequestration into the lung and serum levels of cytokines 6 h after CLP. MS diabetic mice had significantly higher serum concentrations of MIP-2 and KC than found in resistant MS nondiabetic mice (Fig. 2A and B). Serum levels of MIP-2, KC, and IL-6 in SS nondiabetic mice were similar to the levels found in MS diabetic mice. Notably, insulin treatment in MS diabetic mice significantly inhibited the elevation of MIP-2, KC, and IL-6 serum levels.

Leukocyte sequestration (Fig. 2D) in the lungs and neutrophil migration to broncoalveolar cavity (Supplementary Fig. 4) of MS diabetic mice were significantly higher than in MS nondiabetic mice. Insulin-treated diabetic mice presented significantly reduced leukocyte sequestration and neutrophil migration in lungs. In addition to systemic inflammatory response, diabetic mice subjected to MS presented significantly higher serum levels of aspartate aminotransferase and alanine aminotransferase (biochemical markers of liver injury) compared with nondiabetic MS mice. Insulin treatment was able to prevent the increased alanine aminotransferase and aspartate aminotransferase serum levels (Supplementary Fig. 5).

Since neutrophil migration depends on leukocyte-endothelium interactions and chemotaxis (28), we examined the effect of diabetes on these parameters. Diabetic MS mice and nondiabetic SS mice displayed a significant decrease in leukocyte rolling and adhesion compared with nondiabetic MS animals (Fig. 3A and B). Importantly, insulin treatment of diabetic mice significantly prevented the reduction of leukocyte rolling and adhesion after CLP surgery (Fig. 3A and B).

As shown in Fig. 3C, MS induction in diabetic mice or SS induction in nondiabetic mice resulted in significant inhibition of MIP-2–induced chemotaxis of circulating neutrophils compared with neutrophils from MS nondiabetic mice. Insulin treatment also restored the neutrophil chemotaxis response to MIP-2 in MS diabetic mice. Interestingly, blood neutrophils from diabetic mice without sepsis presented a significant decline in neutrophil chemotaxis response (Supplementary Fig. 6A). In vivo, intraperitoneal administration of MIP-2 also induced a lower neutrophil migration in diabetic mice compared with cell migration in nondiabetic mice (Supplementary Fig. 6B).

High levels of AGP have been reported in human diabetes (16,17), and this acute-phase protein is involved in neutrophil migration failure during sepsis (15). Figure 4A shows that serum levels of AGP were significantly increased 6 h after surgery in SH and MS diabetic mice compared with those in SH and MS control mice. Moreover, serum levels of AGP were comparable in MS diabetic mice and SS nondiabetic mice. Similarly, expression of AGP mRNA measured by real-time quantitative PCR in liver tissues was significantly higher in SH or MS diabetic mice than in SH or MS nondiabetic mice, respectively. Treatment of diabetic mice with insulin resulted in significant inhibition of AGP mRNA expression in the liver (Fig. 4B). AGP serum levels were negatively correlated with neutrophil migration (Fig. 4C) and positively correlated with serum levels of alanine aminotransferase and aspartate aminotransferase (Supplementary Fig. 5C and D, respectively). Confirming the inhibitory role of AGP in neutrophil migration, we observed intravenous administration of AGP (50 μg/mice) 30 min before induction of MS inhibits neutrophil migration to the focus of infection (Fig. 4D).

Adhesion molecules such as CD11b and CD62L are important for rolling and adhesion as well as useful markers of neutrophil activation (29). Thus, expression of l-selectin (CD62L) and CD11b in neutrophils was evaluated. Neutrophils obtained from SS nondiabetic mice or from diabetic mice subjected to MS presented a significant reduction in CD62L expression (Fig. 4F) and upregulation of CD11b (Fig. 4G) compared with MS nondiabetic mice. Insulin-treated diabetic mice impaired exacerbation of neutrophil l-selectin shedding and upregulation of CD11b after sepsis.

Remarkably, when AGP was administrated to insulin-treated MS diabetic mice, the effect of insulin in preventing the reduction of neutrophil migration as well as l-selectin shedding and rise of CD11b in blood neutrophils was abolished (Fig. 4D, F, and G, respectively). Figure 4E shows representative histograms of CD62L and CD11b expression in blood neutrophils (cells with high expression of GR-1).

Since CXCR2 expression in leukocytes is necessary for neutrophil migration to infection site (30) and chemotaxis induced by MIP-2 (14), expression of CXCR2 in neutrophils was evaluated. Neutrophil CXCR2 expression in MS diabetic mice or SS nondiabetic mice was significantly decreased compared with neutrophils from MS nondiabetic mice. Treatment of diabetic mice with insulin prevented CXCR2 internalization in neutrophils after MS induction (Fig. 5A). Administration of AGP to MS nondiabetic or to insulin-treated MS diabetic mice reduced CXCR2 expression in blood neutrophils. Figure 5A shows the representative histograms of CXCR2 expression in blood neutrophils (cells with high expression of GR-1).

High expression of GRK2 has been described as a key modulator of CXCR2 receptor desensitization (14,31). Therefore, we examined by immunofluorescence microscopy whether diabetes development modulates the expression of GRK2 in neutrophils after sepsis. Significant upregulation of intracellular GRK2 protein expression was found in SS nondiabetic mice and MS diabetic mice compared with neutrophils from MS nondiabetic or SH diabetic mice. Insulin treatment prevented the GRK2 increase observed in the MS diabetic mice (Fig. 5B and C). Administration of AGP to MS diabetic mice treated with insulin increased GRK2 expression in blood neutrophils.

Using the alloxan model of diabetes followed by polymicrobial infection, we demonstrated that diabetic mice with poor metabolic control are highly susceptible to sepsis. After MS induction, diabetic mice exhibited a reduction in rolling, adhesion, chemotaxis, and migration of leukocytes. As a consequence, these mice had bacterial spreading, systemic inflammatory response, and 100% mortality. Blood neutrophils from diabetic mice subjected to MS expressed high levels of GRK2 and downregulation of CXCR2 expression. Interestingly, the immunosuppression found in diabetic mice appears be mediated by AGP, an acute-phase protein.

Neutrophils are essential for bacterial infection eradication (13,28,32). For diabetes, several studies have demonstrated a defect in neutrophil migration/chemotaxis in human and experimental models (33,34). Extending these findings, we observed that diabetic mice subjected to MS exhibited reduced neutrophil migration to the focus of infection, resulting in bacteremia, systemic inflammatory response, and 100% mortality. The levels of tumor necrosis factor-α and MIP-2 in the peritoneal lavage of diabetic mice were higher than in nondiabetic mice subjected to the same septic stimulus, suggesting that reduction of neutrophil recruitment is not due to the absence of inflammatory mediators.

Next, we investigated the mechanism by which diabetic mice display reduction of neutrophil recruitment toward the infection site. We demonstrated that diabetic mice subjected to MS had a significant decrease in leukocyte rolling and adhesion in postcapillary venules of mesentery. Furthermore, we observed that diabetic MS mice presented impaired intracellular adhesion molecule-1 mesenteric endothelium expression compared with nondiabetic MS mice (data not shown). Changes in adhesion molecule expression also could contribute to the impairment of rolling and adhesion of leukocytes to endothelial cells in diabetes (28). Nevertheless, the role of adhesion molecules such as intracellular adhesion molecule-1 and vascular cell adhesion molecule-1 in the sensitivity to MS remains to be elucidated. Our results are in accordance with those of Fortes et al. (35), which showed reduced leukocyte rolling on venular endothelium of the internal spermatic fascia in diabetic rats.

The inappropriate activation of leuckocytes in circulation induced the shedding of l-selectin and upregulation of CD11b, which could decrease the interactions of these cells with the endothelium (36). Our results demonstrated that neutrophils from diabetic mice had a significant CD62L shedding and upregulation of CD11b compared with control mice. In accordance, a significant increase in neutrophil surface CD11b expression, as well as a decrease in surface CD62L expression, was observed in patients with diabetic microangiopathy (37).

Chemotactic factors and their receptors also play a pivotal role in regulating the activation and movements of leukocytes through the extracellular matrix (28). In a polymicrobial infection, the CXCR2 blockade inhibits the neutrophil migration and increases the severity of sepsis (30). Moreover, in human (38) and experimental sepsis (14), the reduction in neutrophil chemotaxis is associated with decreased CXCR2 expression in neutrophils. In the current study, we found that blood neutrophils from diabetic mice submitted to MS had a significant decrease in CXCR2 expression in these cells compared with neutrophils from control mice. The levels of CXCR2 expression in diabetic mice were similar to those found in nondiabetic mice subjected to SS. Accordingly, neutrophils from diabetic mice subjected or not subjected to sepsis displayed decreased MIP-2–induced chemotaxis response. In vivo, MIP-2 administration in diabetic mice without sepsis also induced a low neutrophil migration.

Subsequently, we investigated whether the downregulation of CXCR2 expression in neutrophils during sepsis development in diabetic mice was mediated by GRK2 activation. GRK2 has been implicated in the downregulation of chemokine receptors (e.g., chemokine [C-C motif] receptors [CCRs] and CXCRs) and innate chemoattractant receptors (e.g., N-formylpeptide chemoattractant receptor [fMLP] and C5a receptor) (39,40). Moreover, increased levels of GRK2 in neutrophils from SS patients or mice are associated with decreased chemotactic responses (14,41). Notably, blood neutrophils from diabetic mice subjected to MS have significant increase in GRK2 expression compared with control mice.

However, despite the data presented here, a fundamental question remains unanswered. Which is the main factor that triggers the susceptibility of diabetic mice to sepsis? AGP is one of the major acute-phase proteins in humans and mice and has been described as an inhibitor of leukocyte functions, such as inhibition of rolling/adhesion and migration in vivo and the chemotaxis response in vitro (15). AGP serum concentration increases in response to systemic inflammation, infection, or diseases like diabetes (16,17,42). Interestingly, we found that AGP expression in serum of alloxan-treated diabetic mice started to increase at 2 h postsepsis and persisted at least until 12 h postsepsis (Fig. 4A and data not shown). Likewise, after MS induction in diabetic mice, serum levels of AGP were similar to those of nondiabetic mice subjected to SS. In addition, the hepatic AGP mRNA expression was significantly higher in diabetic mice after CLP-induced sepsis. Moreover, when AGP was administrated to nondiabetic or diabetic mice treated with insulin, this protein inhibited the neutrophil migration to the focus of infection. These results suggest that AGP might play an important role as an immunosuppressing molecule associated with neutrophil migration failure and therefore in inducing the predisposition of diabetic mice to sepsis as well.

Other studies with diabetic rats and mice have also shown a decreased neutrophil migration (6,34,35), phagocytosis capacity (8), and hydrogen peroxide production (9). It should be emphasized that abnormalities in granulocyte chemotaxis, phagocytosis, and microbicidal activity have been described in poorly controlled, but not controlled, diabetic patients (31,43). Furthermore, the reduction of blood glucose levels by insulin treatment of diabetic patients (44) or rats (9) has been reported to be significantly correlated with improvement of neutrophil phagocytosis capacity. Here, we demonstrated that insulin treatment prevented the failure of neutrophil migration to the focus of infection, improved the infection control, and increased the survival rate of diabetic mice subject to sepsis by ∼70% at 7 days after CLP. Moreover, insulin treatment inhibited AGP expression found in diabetic mice subjected to sepsis. As a consequence, the downregulation of CXCR2 and CD62L and the high expression of CD11b and GRK2 in neutrophils of diabetic mice subjected to sepsis were prevented. However, when AGP was administrated to diabetic mice treated with insulin, the protective effect of insulin was abolished.

The beneficial effect of insulin underpins that at least a moderate control of glucose levels by insulin administration enables diabetic mice to have a better profile of immune response. In contrast, treatment of control nondiabetic septic mice with insulin decreased survival rate. Insulin infusion has been used to normalize hyperglycemia; however, the beneficial effects of insulin and tight glucose control may be overcome by hypoglycemia and an increased death rate in critically ill patients (45). Therefore, further works are necessary to evaluate the adverse/beneficial effect of strict glucose control by insulin during sepsis.

In summary, our study provides evidence for a novel diabetes-insulin-sepsis-AGP axis in the regulation of neutrophil migration as well as control of susceptibility to infection. Moreover, these observations suggest that the regulation of AGP pathway, in conjunction with insulin therapy, merits further investigation as a potential immunopharmacologic intervention (monoclonal antibody) for enhancing the control of sensitivity to sepsis in diabetes.

This study was supported by grants from Fundação de Amparo à Pesquisa do Estado de São Paulo and Conselho Nacional de Desenvolvimento Cientifico e Tecnológico, Brazil.

No potential conflicts of interest relevant to this article were reported.

F.S. designed research, performed research, analyzed data, and wrote the manuscript. D.C., F.O.S., A.d.F., F.S.S., and S.M.V. performed research. F.J.A.P. designed research. J.C.A.-F. designed research, performed research, and analyzed data. F.Q.C. designed research, analyzed data, and wrote the manuscript. F.S. and F.Q.C. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank G.B. Francisco and F.L. Mestriner of the School of Medicine Ribeirão Preto, University of São Paulo, for technical assistance.