Neutrophil extracellular traps as a potential source of autoantigen in cocaine-associated autoimmunity

Abstract

Objective. Exposure to illicit cocaine and its frequent adulterant, levamisole, is associated with ANCAs targeting neutrophil elastase (NE), neutropenia and vasculitic/thrombotic skin purpura. The mechanisms of cocaine/levamisole-associated autoimmunity (CLAA) are unknown. The aim of this study was to assess the ability of cocaine and levamisole to induce the release of neutrophil extracellular traps (NETs), a potential source of autoantigen and tissue injury.

Methods. We performed quantitative and qualitative assessment of NET formation in neutrophils from healthy donors exposed to either drug in vitro. In addition, IgG from sera of individuals with CLAA (CLAA-IgG) was assessed for its ability to enhance formation of, and to bind to, drug-induced NETs.

Results. Both cocaine and levamisole could induce formation of NETs enriched in NE and, potentially, inflammatory mitochondrial DNA. Both drugs could also augment simultaneous release of B cell-activating factor belonging to the TNF family (BAFF). CLAA-IgG, but not IgG from healthy individuals, could potentiate drug-induced NETosis. Furthermore, CLAA-IgG, but not ANCA⁺ control IgG, bound to drug-induced NETs in a pattern consistent with NE targeting.

Conclusion. Both cocaine and levamisole may contribute to the development of ANCAs by inducing release of potentially inflammatory NETs in association with NE autoantigen and BAFF. Enhancement of drug-induced NET release by CLAA-IgG provides a potential mechanism linking vasculitis/pupuric skin disease to acute drug exposure in patients with CLAA. Further study of this under-recognized form of autoimmunity will be likely to provide mechanistic insight into ANCA-associated vasculitis and other diseases associated with NETosis.

Rheumatology key message

• Drug-induced neutrophil extracellular traps may be an important source of autoantigen in cocaine/levamisole-associated autoimmunity.

Introduction

North America remains the world’s largest market for illicit cocaine, with an estimated 1.5 million users in the USA alone [1]. A link between cocaine and autoimmunity was noted as early as 1996 with two reports of destructive upper airway lesions and serum ANCAs in cocaine users [2, 3]. In 2008, reports began to emerge of high-titre serum ANCAs, thrombotic/vasculitic skin purpura and severe neutropenia in cocaine users, a syndrome that coincided closely with the adulteration of the North American illicit cocaine supply with the anti-helminth drug, levamisole [4, 5]. By 2009, the US Centres for Disease Control were reporting that 69% of illicit cocaine seized in the USA contained levamisole [6].

The mechanisms of cocaine/levamisole-associated autoimmunity (CLAA) and related clinical manifestations are unknown. Cocaine and levamisole have been independently linked to the development of ANCAs [5]. The antigen targets of ANCAs in cocaine users appear to be distinct from those in patients with idiopathic ANCA-associated vasculitis (AAV). In AAV, either MPO or PR3 is the dominant antigen target of ANCAs. In cocaine users, a dominant target of ANCAs is neutrophil elastase (NE) [5, 7]. NE is a chief enzymatic constituent of neutrophil extracellular traps (NETs), a recently recognized antimicrobial product released by activated neutrophils [8, 9]. NETs also play important pathogenic roles in certain inflammatory, autoimmune and thrombotic diseases [10, 11]. Formation of NETs (NETosis) is a type of neutrophil programmed cell death that results in extrusion of nuclear and/or mitochondrial DNA (mtDNA) mixed with granule contents. The antimicrobial, thrombotic and inflammatory properties of NETs appear to depend on their protein and nucleic acid composition, which itself is determined by the nature of stimulation [11–13]. The idea that certain drugs might favour the development of highly immunogenic NETs was supported by a recent animal study of propylthiouracil-induced ANCA vasculitis [14].

In AAV, IgG ANCA is believed to potentiate NETosis in activated neutrophils via binding externalized MPO and PR3 and simultaneous engagement of Fc receptors [15], but a pathogenic role for ANCAs in CLAA has not been reported. Purpura and neutropenia have been attributed to toxic effects of levamisole, because therapeutic use of this drug can cause both manifestations [4]. Accordingly, purpura and neutropenia appear to be provoked by active or recent illicit cocaine use; and these problems abate with abstinence, despite the persistence of serum ANCAs (C. Lood and G. C. Hughes, unpublished observations). Thus, ANCA alone may not be sufficient to cause disease in cocaine users.

In this brief study, we identify two mechanisms that could contribute to the development of ANCAs and related clinical manifestations in cocaine users. First, cocaine and levamisole can induce the release of autoantigen in the form of NETs enriched in mtDNA. Second, ANCA from cocaine users enhances drug-induced NETosis, a potential mechanism linking acute drug exposure to certain clinical manifestations.

Methods

Patients and study approval

Blood specimens from healthy control (HC) subjects were obtained with written donor consent according to the Declaration of Helsinki and with ethical approval of the University of Washington Institutional Review Board. For ANCA control (AC) and CLAA subjects, we obtained de-identified frozen serum samples that had been collected previously for the purpose of medical care. CLAA subjects had ANCA IF titres ⩾1:256 and evidence of cocaine exposure by urine toxicology testing. AC subjects had ANCA IF titres ⩾1:256 and no evidence of cocaine exposure by urine toxicology testing.

NET induction

Neutrophils were isolated from heparinized blood using Polymorphprep (Axis-Shield, Dundee, Scotland, UK). Neutrophils were resuspended in medium (RPMI 1640; Life Technologies, Waltham, MA, USA) supplemented with 0.01 M HEPES (Gibco, Waltham, MA, USA) and non-essential amino acids (Thermo Scientific, Waltham, MA, USA). Neutrophils (1 × 10⁶/ml) were incubated in poly-l-lysine-coated plates with or without the following inhibitors: Cl-amidine (200 μM; Calbiochem, San Diego, CA, USA), apocynin (100 μM; Sigma-Aldrich, St. Louis, MO, USA), (DPI, 25 μM; Sigma) or thenoyltrifluoroacetone (1 μM; Sigma-Aldrich) for 1 h before addition of levamisole (10 nM), cocaine (10 μM) or PMA (20 nM) for an additional 4 h. In some experiments, serum IgG (10 μg/ml), purified using the Melon Gel IgG kit (Thermo Scientific), was added in combination with other stimuli.

NET quantification

Plate-bound NETs were detached for 30 min at 37 °C with micrococcal nuclease (0.3 U/ml; Thermo Scientific), diluted in nuclease buffer containing 10 mM Tris–HCl pH 7.5, 10 mM MgCl₂, 2 mM CaCl₂ and 50 mM NaCl, and analysed by Sytox Green incorporation (Life Technology) as determined by fluorimetry (Synergy 2; BioTek, Winooski, VT, USA) using a DNA standard for quantification.

NET visualization

Activated neutrophils were fixed in 2% paraformaldehyde, permeabilized with saponin and stained with antibodies to NE (ab21595, diluted 1/100; Abcam, Cambridge, MA, USA) or purified human IgG (10 μg/ml) followed by addition of Alexa Fluor 555-conjugated donkey-anti-rabbit IgG (diluted 1/200; Invitrogen) or Alexa Fluor 647-conjugated donkey-anti-human IgG (diluted 1/1000; Jackson ImmunoResearch, West Grove, PA, USA), respectively, as well as DAPI or Sytox Green to identify DNA. NETs were visualized by IF microscopy (EVOS cell imaging system; Life Technology).

NE activity and protein levels

Detached NETs were incubated with NE substrate (N-succinyl-Ala-Ala-Ala-p-nitoranilide, 20 μM; Sigma-Aldrich), and absorbance was analysed at 405 nm (Synergy 2; BioTek). As a standard curve, purified porcine elastase was used (Sigma). Relative NE and DNA levels in NETs stained with DAPI and anti-NE (above) were quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

Quantification of mtDNA

DNA was purified from extruded NETs using phenol chloroform, and 8 ng of isolated DNA was mixed with SYBR green master mix (Thermo Scientific) and primers (50 nM) for 16S (forward: 5′-CGC ATA AGC CTG CGT CAG ATC AA-3′; and reverse: 5′-TGT GTT GGG TTG ACA GTG AGG G-3′) or 18S (forward: 5′-GTA ACC CGT TGA ACC CCA TT-3′; and reverse: 5′-CCA TCC AAT CGG TAG TAG CG-3′) at a final volume of 20 μl. Activation of enzyme at 95 °C for 15 min was followed by 40 cycles with 95 °C for 15 s and 60 °C for 60 s.

Anti-NE IgG ELISA

A 96-well plate was coated with 1 μg/ml purified NE (Innovative Research) for 16 h, followed by blocking in 1% BSA for 2 h. Serum samples, diluted 1/100 in 1% blocking buffer, were added for 2 h, followed by ALP-conjugated goat-anti-human IgG antibodies (diluted 1/100; Jackson ImmunoResearch) and analysis of absorption at 405 nm.

Analysis of BAFF

Neutrophils (2.5 × 10⁶/ml) were activated for 16 h, and cell-free supernatant was analysed for BAFF by ELISA (BAFF DuoSet; R&D Systems, Minneapolis, MN, USA).

Results

Cocaine and levamisole induce NET formation

Upon exposure of neutrophils to cocaine or levamisole, we observed by IF microscopy release of DNA complexed with NE plus chromatin decondensation characteristic of NETs (Fig. 1A). Drug concentrations used were those that yielded maximal NETosis (data not shown) and that fell within levels reported in human blood [16, 17]. Either drug could induce DNA release (Fig. 1B), although not as potently as PMA. In these conditions, drug exposure resulted in only ∼20% cell death, whereas PMA, as expected, induced 100% cell death (data not shown). Despite lower DNA levels, drug-induced NETs showed similar NE activity compared with PMA-induced NETs (Fig. 1C), suggesting relative enrichment for NE. Indeed, material extruded after cocaine or levamisole exposure showed higher NE/DNA ratios compared with PMA-induced NETs (Fig. 1D). Thus, drug-induced DNA release was qualitatively different from PMA-induced NETosis. To confirm that drug-induced DNA release was attributable to NETosis, we treated neutrophils with inhibitors of reactive oxygen species (ROS) production and peptidyl arginine deiminase, type 4 (PAD4)-mediated citrullination (Fig. 1E). As expected, phorbol myristate acetate (PMA)-induced DNA release was reduced by inhibitors of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (apocynin) or PAD4 (Cl-amidine), but not by an inhibitor of mitochondrial ROS (mtROS), thenoyltrifluoroacetone. In contrast, drug-induced DNA release was somehwat reduced by all three of these inhibitors and almost completely abrogated by diphenyleneiodonium (DPI), an inhibitor of both NADPH oxidase and mtROS. Thus, cocaine and levamisole could induce DNA release through engagement of both mitochondrial and NADPH oxidase pathways. We recently demonstrated that production of mtROS primes neutrophils to release NETs enriched in oxidized, inflammatory mtDNA [13]. Consistent with a requirement for mtROS (Fig. 1E), drug-induced NETs contained higher ratios of 16S mtDNA/18S nuclear DNA compared with PMA-induced NETs (Fig. 1F).

Fig. 1.

Cocaine and levamisole induce neutrophil extracellular trap formation

(A) IF microscopic images showing DNA (blue) and neutrophil elastase (NE, red) staining of neutrophils exposed to cocaine, levamisole or PMA before fixation and permeabilization. Scale = 10 μm. Insets are digitally enlarged. (B and C) DNA (B) and NE release (C) were quantified after enzymatic detachment of NETs (n = 16 experiments for B, and n = 6 for C). (D) Computer-aided analysis of NE/DNA intensity in images represented in A (median values indicated). (E) DNA release in cultures identical to those in A–D pre-incubated for 60 min with inhibitors of PAD4 (Cl-amidine), NADPH oxidase (apocynin), mitochondrial ROS (TTFA) or both NADPH oxidase and mitochondrial ROS (DPI) (n = 3–9 experiments). (F) Ratios of mitochondrial (16S) DNA/nuclear (18S) DNA in released NETs as determined by quantitative PCR (n = 10 experiments). (G) BAFF release by neutrophils stimulated for 16 h as determined by ELISA (n = 5–7 experiments). *P < 0.05, **P < 0.01 and ***P < 0.001, Student’s paired t-test vs control (unless otherwise indicated). For D, one-way analysis of variance with correction for multiple comparisons was used. Shown are mean values and s.d.s, except for D, which shows individual and median values. DPI: diphenyleneiodonium; NADPH: nicotinamide adenine dinucleotide phosphate; NET: neutrophil extracellular trap; PAD4: peptidyl arginine deiminase, type 4; PMA: phorbol myristate acetate; TTFA: thenoyltrifluoroacetone.

Neutrophils are an important source of BAFF, a potent B-cell survival and differentiation factor implicated in the development of human autoimmunity [18]. We observed that both cocaine and levamisole enhanced the release of BAFF (Fig. 1G), although not to the extent of PMA. In these experiments, we could not exclude the possibility that cell death pathways other than NETosis (e.g. apoptosis) contributed to BAFF release. Taken together, these results indicate that both cocaine and levamisole can induce in neutrophils the coincidental release of the following: (i) NETs enriched in potentially inflammatory mtDNA; (ii) NE, a major autoantigen in CLAA patients; and (iii) the B-cell survival factor, BAFF.

CLAA-IgG amplifies drug-induced NETosis and recognizes NET-associated antigens

ANCA IgG from AAV patients can enhance TNFα-potentiated NETosis [15]. Therefore, we wanted to know whether ANCA IgG from CLAA subjects (CLAA-IgG) could enhance drug-induced NETosis. Although the addition of IgG purified from HC donors did not enhance NET formation induced by either drug, addition of ANCA control IgG (AC-IgG), and in particular CLAA-IgG, enhanced drug-induced DNA release by about 3-fold (Fig 2A and B). None of the three IgG isolates was able to increase DNA release induced by PMA, a potent inducer of NETosis (Fig. 2C). Thus, IgG purified from CLAA patients markedly enhanced cocaine- and levamisole-induced NET formation.

Fig. 2.

Cocaine/levamisole-associated autoimmunity-IgG amplifies drug-induced neutrophil extracellular trap formation and binds to neutrophil extracellular trap-associated antigens

DNA release from neutrophils that were pre-incubated with IgG isolated from HC, AC or CLAA subjects, then stimulated with cocaine (A), levamisole (B) or PMA (C). The results are derived from three or four independent experiments with IgG from three individuals for each condition. (D) Serum anti-NE IgG levels in HC, AC and CLAA subjects. (E) IF microscopic images showing DNA (green) and IgG (red) staining of unstimulated and levamisole-stimulated neutrophils incubated with HC, AC or CLAA sera. Scale = 10 μm. Insets are digitally enlarged. Images are representative of at least three independent experiments for each condition. *P < 0.05, **P < 0.01 and ***P < 0.001, Student’s paired t-test vs control (unless otherwise indicated). Shown are mean values and s.d.s, or mean and individual values (D). AC: ANCA control; CLAA: cocaine/levamisole-associated autoimmunity; HC: healthy control; NE: neutrophil elastase; NET: neutrophil extracellular trap; PMA: phorbol myristate acetate.

Given that both cocaine and levamisole could induce release of NET-associated NE (Fig. 1A and C), and NE is a major antigen target of ANCAs in cocaine users, we wanted to know whether CLAA-IgG recognized NET-associated antigens. As expected, both AC-IgG and CLAA-IgG recognized cytoplasmic antigens in unstimulated neutrophils, whereas IgG from HC donors (HC-IgG) did not (Fig. 2E, upper row and insets). In levamisole-stimulated neutrophils, however, CLAA-IgG, but not AC-IgG or HC-IgG, showed prominent reactivity with NET-associated antigens (Fig. 2E, bottom row). The pattern of IgG staining was similar to that of NET-associated NE (Fig. 1A, cocaine or levamisole). Moreover, sera from CLAA but not AC or HC subjects showed marked IgG reactivity to NE (Fig. 2D). Thus, it is likely that CLAA-IgG uniquely recognizes NE associated with drug-induced NETs.

Discussion

Our results identify potential mechanisms by which cocaine and its frequent adulterant, levamisole, break immunological tolerance in humans. Both drugs can induce formation of NETs enriched in NE, a known target of ANCAs in cocaine users. Moreover, drug-induced NETs could be immunogenic owing to their relatively high content of inflammatory mtDNA, an idea that could be tested in animal models. Release of inflammatory NETs and NE in conjunction with BAFF might further promote the survival and differentiation of NE-reactive B cells. Indeed, IgG ANCAs in cocaine users can persist for years after apparent drug abstinence (C. Lood and G. C. Hughes, unpublished observations), suggesting the involvement of long-lived B cells having undergone immunoglobulin gene class switch recombination. Thus, chronic exposure to either drug may lead to loss of tolerance through repeated release of neutrophil antigens in the context of inflammatory DNA and BAFF. Whether or not cocaine and levamisole synergize in this regard remains to be determined.

However, ANCA alone appears insufficient to cause clinical disease in cocaine users; other factors, such as acute drug exposure, are required. Our observation that CLAA-IgG enhanced drug-induced NET formation suggests one way in which ANCAs could become pathogenic during acute drug exposure. For example, vasculitic purpura might involve ANCA enhancement of drug-induced NETosis among perivascular or tissue-infiltrating neutrophils, analogous to what is described in AAV [15]. Thrombotic purpura might involve the ability of NETs to cause intravascular thrombosis [10]. Additionally, binding of CLAA-IgG to drug-induced NETs might enhance their immunogenicity/pathogenicity by protecting them against degradation or by triggering complement- or Fc receptor-mediated inflammation, similar to mechanisms described for anti-NET antibodies in SLE [19].

Continued study of CLAA is important for several reasons. First, millions of people may be at risk. Second, CLAA is likely to be more prevalent among cocaine users than suggested by available data, because only those presenting with select clinical manifestations are tested for autoimmunity; thus, CLAA may be a relatively common form of autoimmunity. Further dissecting its mechanisms should provide much needed insight into drug-induced autoimmunity, idiopathic AAV and other forms of immunological disease involving NETs.