Increased Neutrophil Elastase and Proteinase 3 and Augmented NETosis Are Closely Associated With β-Cell Autoimmunity in Patients With Type 1 Diabetes

The global incidence of type 1 diabetes (T1D), an autoimmune disease caused by an interactive combination of genetic and environmental factors, has more than doubled in the past two decades (1,2). Although the triggering factors that are involved in the initiation of T1D remain unclear, it is widely accepted that organ-specific autoimmune destruction of the insulin-producing β-cells in the pancreatic islets of Langerhans is mediated primarily by autoreactive T cells, which is accompanied by the production of different autoantibodies to β-cell antigens, including glutamic acid decarboxylase autoantibody (GADA), insulinoma-associated protein 2 autoantibody (IA2A), and zinc transporter-8 autoantibody (ZnT8A) (3–5). These autoantibodies have been proven to be instrumental for the prediction and diagnosis of T1D but are deemed not to be pathogenic (6,7). A number of other immune cells, including dendritic cells, macrophages, B cells, and neutrophils, are also implicated in the development of insulitis in T1D (8,9).

Neutrophils, which are the most abundant (40–75%) type of white blood cells, have recently been implicated in the onset and progression of T1D (10,11). The primary functions of neutrophils are to eliminate extracellular pathogens by multiple strategies, including phagocytosis, degranulation to release lytic enzymes, and neutrophil extracellular traps (NETs) that are formed through a unique cell death process clearly differentiated from apoptosis and necrosis, termed “NETosis” (12–14). However, improper activation of neutrophils may lead to tissue damage during autoimmune or exaggerated inflammatory responses (15). Notably, circulating neutrophil counts are reduced in patients with T1D and in their nondiabetic first-degree relatives but not in patients with type 2 diabetes (11). In nonobese diabetic (NOD) mice (a spontaneous model of T1D), neutrophil infiltration and NET formation in the islets were observed as early as 2 weeks after birth, and blockage of neutrophil activities with an anti-Ly6G antibody reduced the subsequent development of insulitis and diabetes (9).

Neutrophil serine proteases, including neutrophil elastase (NE), proteinase 3 (PR3), and cathepsin G (CG), are the major components of neutrophil azurophilic granules that participate in the elimination of engulfed microorganisms (16). Neutrophil activation and degranulation can result in the release of neutrophil serine proteases into the extracellular medium and circulation, where they not only help to eliminate the invaded pathogens but also serve as the humoral regulators of the immune responses during acute and chronic inflammation, modulating cellular signaling network by processing chemokines, and activating specific cell-surface receptors (17–19). Abnormal activities of neutrophil serine proteases have been implicated in the pathogenesis of several inflammatory and autoimmune diseases, including chronic obstructive pulmonary disease, cystic fibrosis, Wegener granulomatosis, Papillon-Lefèvre syndrome, and small-vessel vasculitis (20). However, their association with T1D has not been explored so far.

In this study, we measured circulating levels of two main types of neutrophil serine proteases (NE and PR3) and their enzymatic activities in T1D patients with different disease duration together with age- and sex-matched healthy control subjects. Furthermore, we explored whether altered NET formation and α1-antitrypsin (A1AT), a major endogenous inhibitor of neutrophil serine proteases, are associated with reduced neutrophil counts and markedly increased activities of neutrophil serine proteases in patients with T1D. We also measured the dynamic changes of circulating NE/PR3 activities during the development of autoimmune diabetes in NOD mice.

A total of 149 patients with T1D were randomly selected from children diagnosed at the Diabetes Center, Second Xiangya Hospital of Central South University, from October 2000 to October 2013. Patients with T1D were diagnosed according to the criteria of the American Diabetes Association (21). All patients were treated with insulin. The median disease duration of T1D was 4.2 (interquartile range 1.7–7.1) years.

A total of 77 age- and sex-matched healthy control subjects were recruited from children in the community participating in a health screening at the Children Health Center of the Second Xiangya Hospital, Central South University, using the following inclusion criteria: fasting plasma glucose of less than 5.6 mmol/L and 2-h plasma glucose of less than 7.8 mmol/L, and no family history of diabetes or other autoimmune or chronic diseases.

A total of 25 adults with type 2 diabetes diagnosed within 1 year and 25 age- and sex-matched healthy control subjects were recruited at the Diabetes Center, Second Xiangya Hospital of Central South University, and the inclusion criteria were described in our previous study (22).

The Second Xiangya Hospital of Central South University Institutional Review Board approved the study, and written informed consents were obtained from the patients and healthy control subjects.

After overnight fasting, a venous blood specimen was collected in the morning (∼0800) for analysis of various biochemical parameters. Plasma glucose was measured enzymatically on a Hitachi 7170 analyzer (Boehringer Mannheim GmbH, Mannheim, Germany). HbA_1c was measured by automated liquid chromatography (VARIANT II Hemoglobin Testing System; Bio-Rad, Hercules, CA). Serum levels of C-peptide and C-reactive protein were quantified using a chemiluminescence immunoassay on a Bayer 180SE Automated Chemiluminescence System (Bayer AG, Leverkusen, Germany) and an immunoturbidimetric assay (Orion Diagnostica, Espoo, Finland), respectively. The titers of GADA, IA2A, and ZnT8A were determined by in-house radioligand assays, as previously described (22,23).

Circulating protein levels of NE, PR3, and A1AT were measured using ELISA kits established in our laboratory (Antibody and Immunoassay Services, The University of Hong Kong). The limits of detection for the NE, PR3, and A1AT ELISA kits were 0.156 ng/mL. No cross-reactivity among these proteins or with other proteins was detected. The intra- and interassay variations were, respectively, 4.5% and 5.1% for the NE ELISA kit, 3.9% and 4.3% for the PR3 ELISA kit, and 4.9% and 5.3% for the A1AT ELISA kit.

The combined enzymatic activities of PR3 and NE in serum were determined with a chromogen-based assay using N-methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide (Sigma-Aldrich, St. Louis, MO) as the substrate, which has a catalytic constant K_cat/K_m of 33,915 M⁻¹ s⁻¹ for NE and 499 M⁻¹ s⁻¹ for PR3 (24). Briefly, serum (20 µL) was incubated with 0.1 mol/L Tris-HCl buffer (180 µL; pH 8.0) containing 0.5 mol/L NaCl and 1 mmol/L substrate at 37°C for 24 h. The amount of p-nitroaniline released was measured spectrophotometrically at 405 nm. The enzymatic activities of PR3 and NE were calculated according to the Δ optical density (OD) values before and after 24-h incubation with substrate and expressed as mU/mL serum, where 1 unit was defined as the amount of PR3 and NE that hydrolyze the substrate to yield p-nitroaniline at 1 µmol/min at 37°C (25).

The levels of neutrophil NETosis were measured by quantifying the amount of circulating myeloperoxidase (MPO)-DNA complexes, a well-established marker of NET formation, as previously described (26). Briefly, 5 μg/mL mouse anti-MPO monoclonal antibody (ABD Serotec, Puchheim, Germany) was coated to 96-well microtiter plates overnight at 4°C. After blocking with 1% BSA, serum samples were added per well with the peroxidase-labeled anti-DNA monoclonal antibody (component No. 2 of the Cell Death Detection ELISA PLUS kit; Roche Diagnostics, Indianapolis, IN) according to the manufacturer’s instructions. After incubation for 2 h at room temperature on a shaking device (300 rpm), the wells were washed three times and then incubated with the peroxidase substrate at 37°C for 60 min. The OD at wavelength of 405 nm was measured using a μQuant microplate reader (BioTek Instruments, Winooski, VT).

NOD/ShiLtJ breeder mice were purchased from The Jackson Laboratory (Bar Harbor, ME). BALB/c and C57BL/6N mice were obtained from the University of Hong Kong Animal Unit. All mice were housed in a room under specific pathogen–free conditions and 12-h light/dark cycles at 22° to 24°C and with ad libitum access to water and standard chow (PicoLab Rodent Diet 20; LabDiet, St. Louis, MO). Blood was collected weekly from female mice from 2 to 30 weeks of age. Blood glucose was monitored using an Accu-Chek Advantage glucose meter (Roche Diagnostics, Indianapolis, IN), and diabetes was defined as two consecutive readings above 11.1 mmol/L (9). Circulating NE/PR3 enzymatic activities were measured as described above. All experimental procedures were approved by the University of Hong Kong Committee on the Use of Live Animals for Teaching and Research and were performed in accordance with the Guide for the Care and Use of Laboratory Animals.

All analyses were performed with SPSS 16.0 software (SPSS, Chicago, IL). Normality was tested using the Kolmogorov-Smirnov test. Data that were not normally distributed were logarithmically transformed before analysis. Differences between groups were assessed by χ² or unpaired Student t test. Comparisons among groups were performed using one-way ANOVA and independent Student t tests. Correlations were analyzed using Pearson correlation or partial correlation, as appropriate. Data are expressed as mean ± SD or median with the interquartile range (IQR), as appropriate. In all statistical comparisons, a P value <0.05 was used to indicate a statistically significant difference.

The clinical characteristics of T1D patients and their matched healthy control subjects are described in Table 1. T1D patients were further divided into three groups by disease duration, including patients within 1 year from diagnosis (n = 28), with a disease duration >1 and <5 years (n = 59), and with duration >5 years (n = 62). Compared with healthy subjects, T1D patients had higher fasting glucose and HbA_1c but lower fasting C-peptide levels. No significant differences in C-reactive protein were found among these groups (Table 1). Consistent with previous reports (10,11), circulating neutrophils were moderately reduced in T1D patients diagnosed within 1 year compared with the healthy control subjects (median 2.27 [IQR 1.80–3.52] vs. 3.63 [3.02–4.15] × 10⁶/mL; P < 0.05), but not in T1D patients with a disease duration of >1 and <5 years or with duration >5 years (Table 1).

In contrast to the mild reduction of peripheral neutrophils, we found that the circulating protein levels of NE and PR3 were dramatically increased in T1D patients compared with the healthy control subjects (median NE: 1594.7 [IQR 988.4–2284.6] vs. 397.0 [262.2–468.8] ng/mL, P < 0.001; PR3: 295.3 [206.0–430.4] vs. 107.4 [92.5–165.0] ng/mL, P < 0.001). Notably, the magnitude of increases in protein levels of NE and PR3 was significantly higher in T1D patients diagnosed within 1 year compared with the other two groups with longer disease duration (all P < 0.01, Fig. 1A and B). Circulating protein levels of NE and PR3 did not differ between boys and girls in patients or control subjects.

To further confirm these findings, we measured the combined enzymatic activities of NE and PR3 using N-methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide, a common substrate for NE and PR3 (24). The results showed that circulating NE/PR3 enzymatic activities in T1D patients were also substantially higher than those in healthy individuals (median 0.69 [IQR 0.41–1.03] vs. 0.14 [0.10–0.21] mU/mL, P < 0.001). Likewise, the most significant increase in NE/PR3 enzymatic activities was observed in T1D patients within 1 year from diagnosis (Fig. 1C). The correlation coefficient between NE/PR3 enzymatic activities and circulating protein levels was 0.915 for NE (P < 0.001) and 0.874 for PR3 (P < 0.001).

The activities of plasma NE and PR3 are tightly controlled by their associated endogenous inhibitors, especially A1AT, an archetype member of the serine protease inhibitor (SERPIN) superfamily. Because our data showed that the amplitude of increases in NE and PR3 enzymatic activities was higher than that of the circulating protein levels, we next investigated whether dysregulated A1AT contributed to the increased enzymatic activities of NE and PR3 in T1D. In contrast to elevated NE and PR3 levels, the circulating concentrations of A1AT in T1D patients diagnosed within 1 year were significantly decreased compared with healthy subjects (median 1.37 [IQR 1.07–1.83] vs. 1.80 [1.57–2.07] mg/mL, P < 0.01), whereas the decline in patients with disease duration for 1–5 years or with disease duration >5 years did not reach statistical significance (Fig. 1D).

To explore the underlying mechanism responsible for the markedly elevated circulating NE and PR3 levels, we examined the levels of neutrophil NETosis by quantifying the amount of circulating MPO-DNA complexes, a well-established marker of NET formation (26). In line with the increased NE and PR3 levels, a significant elevation of circulating MPO-DNA complexes was observed in T1D patients, especially in T1D patients with the disease duration of <1 year compared with the healthy individuals (median 0.197 [IQR, 0.049–0.412] vs. 0.026 [0.011–0.058] mean OD 405, P < 0.001; Fig. 2A). Furthermore, the amount of MPO-DNA complexes in serum was significantly correlated with the circulating protein levels of NE (r = 0.554, P < 0.001) and PR3 (r = 0.575, P < 0.001) as well as with NE and PR3 enzymatic activities (r = 0.527, P < 0.001; Fig. 2B–D), suggesting that the increased circulating NE and PR3 protein levels in T1D patients are at least partly attributed to enhanced neutrophil NETosis.

We next explored the relationship between circulating neutrophil serine proteases and GADA, IA2A, and ZnT8A, the three autoantibodies associated with β-cell autoimmunity in T1D patients. Among 149 T1D patients, 54 (36%) were negative for autoantibodies, and 61 (41%), 24 (16%), and 10 patients (7%) were positive for one, two, and three autoantibodies, respectively. Notably, circulating levels of NE and PR3 proteins and their enzymatic activities were increased progressively with increased numbers of the autoantibodies detected in these patients (Fig. 3A–C). Even for the autoantibody-negative T1D patients, the circulating protein levels and enzymatic activities of NE and PR3 were much higher than those in healthy control subjects (median protein levels for NE: 1154.90 [IQR 770.8–1749.5] vs. 397.0 [262.2–468.8] ng/mL, P < 0.0001; PR3: 237.4 [154.3–307.1] vs. 107.4 [92.5–165.0] ng/mL, P < 0.0001; enzymatic activities: 0.53 [0.37–0.79] vs. 0.14 (0.10–0.21) mU/mL, P < 0.001; Fig. 3A–C). Furthermore, a strong correlation between the titers of GADA and the circulating protein levels of NE (r = 0.296, P = 0.011) and PR3 (r = 0.270, P = 0.021) as well as NE and PR3 enzymatic activities (r = 0.275, P = 0.019) were detected in the GADA-positive T1D patients (n = 73; Fig. 3D–F). Likewise, the titers of IA2A in IA2A-positive T1D patients (n = 44) were also positively associated with the protein levels of NE, PR3, and their enzymatic activities (Supplementary Fig. 1A–C). After adjustment for disease duration, the circulating protein levels and enzymatic activities of NE and PR3 were still significantly correlated with the numbers and titers of these autoantibodies (all P < 0.05, Supplementary Table 1). However, no significant correlation between fasting blood glucose and circulating protein levels of NE (r = −0.103, P = 0.211) or PR3 (r = −0.097, P = 0.237) or NE/PR3 enzymatic activities (r = −0.078, P = 0.342) was observed in the current study cohort. We further measured and compared the circulating protein levels and enzymatic activities of NE and PR3 in 25 type 2 diabetes patients within 1 year from diagnosis and in 25 age- and sex-matched healthy control subjects (Supplementary Table 2). There was no significant difference in protein levels or enzymatic activities of NE and PR3 or NETosis between the two groups (Supplementary Table 2). Taken together, these data suggested that elevated NE and PR3 might be closely associated with β-cell autoimmunity but not glycemic status in T1D patients.

To further explore the relationship between neutrophil serine proteases and the development of T1D, we determined the dynamic changes of PR3 and NE in NOD mice (n = 30), a well-established animal model for autoimmune diabetes, from 2 to 30 weeks of age. These mice were retrospectively assigned to two groups: those that eventually developed diabetes (n = 22, called “diabetic”) and those that did not (n = 8, called “nondiabetic”). The circulating NE/PR3 enzymatic activities in diabetic mice were markedly elevated by fourfold as early as 4 weeks after birth compared with those in 2-week-old mice, and the elevation was sustained for more than 10 weeks before the onset of diabetes. Afterward, the NE/PR3 activities in diabetic mice were gradually decreased to the baseline levels, presumably due to the termination of autoimmune responses as a result of complete β-cell destruction (Fig. 4A). Although there was a transient and modest increase of circulating NE/PR3 activities in nondiabetic mice between 4 and 5 weeks after birth (Fig. 4B), the magnitude and duration of NE/PR3 elevation was substantially lower than in age-matched diabetic mice (Fig. 4C). In BALB/c and C57BL/6N mice, which do not develop insulitis and autoimmune diabetes, circulating NE/PR3 activities remained little changed throughout the 30-week observation period (Supplementary Fig. 2).

In this study, we demonstrated that a modest reduction of neutrophil counts in patients with T1D at onset is accompanied by a marked elevation of protein levels and enzymatic activities of NE and PR3, the two major neutrophil serine proteases. Furthermore, these changes in T1D patients are closely associated with increased neutrophil NETosis, as determined by quantification of MPO-DNA complexes in the circulation. These findings suggest that the reduction of neutrophil counts in T1D patients is partly attributed to augmented NETosis, which in turn leads to increased NET formation and the release of NE and PR3 into the blood stream.

We showed that the amplitude of elevation in circulating NE/PR3 enzymatic activities and NET formation in patients with disease duration of <1 year is substantially higher than those with disease duration of >1 year. A significant reduction in neutrophil counts is observed only in T1D patients with a disease duration of <1 year. Consistent with our findings, a previous study in Italy also found that neutrophil reduction is greatest in individuals with the highest risk of developing T1D (11). After the disease onset, mild neutropenia persists for a few years and then resolves at 5 years after diagnosis (as determined by a longitudinal analysis). In NOD mice with spontaneous development of autoimmune diabetes, neutrophil infiltration and NET formation in the islets are detected as early as 2 weeks after birth, well before the onset of overt diabetes (9). Furthermore, neutrophil depletion at the early stage reduces subsequent development of diabetes in NOD mice (9). Taken together, these data support an early role of neutrophil NETosis, NET formation, and augmented release of neutrophil serine proteases in the onset of β-cell autoimmunity in T1D. Indeed, increased neutrophil NETosis and NET formation have been implicated in a number of autoimmune diseases, including small vessel vasculitis, systemic lupus erythematosus (SLE), and multiple sclerosis (26–28).

In SLE, NETs have been demonstrated to stimulate plasmacytoid dendritic cells for releasing interferon-α, which in turn augments the autoreactivity of antigen-presenting and antibody-producing cells (29,30). NETosis leads to the release of intracellular proteins, including histones and high-mobility group protein B1, the latter of which is implicated in the initiation and/or perpetuation of autoimmunity in several types of autoimmune disorders, including T1D (30,31). Furthermore, NETs is associated with altered patterns of epigenetic and posttranslational modifications, such as methylation, acetylation, and citrullination, which may represent an important source of autoantigens promoting the generation of autoantibodies (32). In particular, a growing body of evidence supports a pathogenic role for citrullinated autoantigens in triggering autoimmune responses in SLE, rheumatoid arthritis, and multiple sclerosis (33). However, the pathophysiological roles of NETosis and its associated changes in T1D remain elusive.

The current etiopathological diagnosis of autoimmune T1D heavily relies on the detection of the autoantibodies against several β-cell antigens. However, these autoantibodies are rarely detectable in children before 6 months of age (34). Moreover, the diagnostic sensitivity of the single autoantibody measurement in T1D patients is as low as 59–67% (35). To capture the therapeutic window for this disease, identifying new biomarkers for detection of early immunological events that affect human islets is critically important. Our current study demonstrated an approximately fourfold increase of circulating protein levels and more than a fivefold elevation of enzymatic activities of NE and PR3 in T1D patients. Furthermore, elevated NE and PR3 were significantly associated with the positive numbers and titers of the autoantibodies detected in T1D patients. Even in those autoantibody-negative patients, the circulating enzymatic activities of NE and PR3 are still substantially higher than in healthy controls. Using the animal model of T1D, we found that elevated circulating NE/PR3 activities occur well before the onset of hyperglycemia and diabetes and that their activities gradually decline after the development of overt diabetes.

Taken together, our data suggest that circulating NE and PR3 may serve as sensitive biomarkers for early detection of those individuals with high risk of developing T1D. However, we found no significant association between increased levels of NE and PR3 and the severity of hyperglycemia in T1D patients. In fact, although hyperglycemia becomes more severe with the progression of T1D, circulating levels of NE and PR3 exhibit opposite changes, suggesting that increased neutrophil NETosis and augmented release of NE and PR3 are not the consequence of impaired glycemic controls but are related to β-cell autoimmunity. Indeed, our observation that NETosis and NE/PR3 levels in T1D patients with longer disease duration are much lower than new-onset patients (<1 year) may be attributed to the gradual attenuation of β-cell autoimmunity with the progression of diabetes to an advanced stage. This is also in line with the fact that the number of autoantibodies in new-onset T1D patients was much higher than those with longer disease duration (36).

In addition to its classical roles for host defense against infection, neutrophil serine proteases are an important regulator of inflammation and innate immunity (17,19,37,38). NE and PR3 are both involved in maturation and release of proinflammatory cytokines, such as tumor necrosis factor-α, interleukin-1β, and interleukin-18, and also induces expression and activation of Toll-like receptors (39–42), all of which are important mediators of insulitis and β-cell destruction (43,44). Furthermore, NE and PR3 play an indispensable role in recruiting neutrophils to the site of inflammation. Notably, neutrophil serine proteases have recently been implicated in high-fat diet–induced obesity, inflammation, and macrophage infiltration in adipose tissues in mice (45). Injection of recombinant PR3 alone is sufficient to induce hyperglycemia in mice (46). By contrast, treatment with A1AT, a major endogenous inhibitor of NE and PR3, decreases lymphocyte infiltration in the islets and prevents β-cell loss and diabetes in rodent models of T1D (47,48). These animal studies, in conjunction with our clinical and animal findings, suggest that elevated NE and PR3 may be the direct contributors to the pathogenesis of autoimmune diabetes by early involvement of autoimmune inflammatory responses in pancreatic islets.

A1AT, the most abundant circulating serpin secreted from hepatocytes, inhibits neutrophil serine proteases by covalent binding to the enzymes (49). Deficiency of A1AT has been implicated in a number of inflammatory disorders, including chronic obstructive pulmonary disease (50). Our present study observed a modest but significant reduction of circulating A1AT in patients with T1D, suggesting that augmented circulating NE and PR3 activities may result from a combination of increased release of these two enzymes from neutrophil NETosis and decreased production of their endogenous inhibitor A1AT.

Our study has several limitations, including the relatively small sample size and the cross-sectional design. In addition, because our samples were collected from Chinese only, whether the findings are replicable in other ethnic group remains to be determined. Further large scale, longitudinal studies on different ethnic groups are mandatory to clarify the roles of NE and PR3 in the initiation and progression of β-cell autoimmunity and to evaluate their clinical value for prediction and early diagnosis of T1D.

Funding. This work was supported by The University of Hong Kong (HKU) matching funds for the State Key Laboratory of Pharmaceutical Biotechnology, the Hong Kong Research Grants Council Collaborative Research Fund (HKU4/CRF/10R and HKU2/CRF/12R) and Germany/Hong Kong Joint Research Scheme (G-HK708/13) to A.X., the National Key Technology Research & Development Program (2012BAI02B04) to Z.Z., and the National Natural Science Foundation of China (81200600) to Y.X.

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

Author Contributions. Y.W., Y.X., and L.Z. conducted the experiments, analyzed data, and wrote the manuscript. D.Y., J.Z., and Y.T. conducted the experiments. S.R.B. and K.S.L.L. contributed to data analysis and edited the manuscript. Z.Z. and A.X. contributed to the experimental design, analyzed data, and wrote the manuscript. Z.Z. and A.X. 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.