Antineutrophil cytoplasmic antibodies (ANCA) binding to neutrophil elastase (NE) and proteinase 3 (PR3) are detectable in most patients with cocaine-induced midline destructive lesions (CIMDL), but the pathogenic role and antigen specificity of these antibodies are unknown. This study was undertaken to assess the effects of NE ANCA on the enzymatic activity of NE, to determine whether these antibodies interfere with the physiologic effect of secretory leukoprotease inhibitor (SLPI), and to investigate the antigen specificity of both NE and PR3 ANCA in patients with CIMDL. We also compared the binding of PR3 ANCA in patients with CIMDL with that in patients with Wegener's granulomatosis (WG).
PR3 ANCA and NE ANCA were detected by capture enzyme-linked immunosorbent assays (ELISAs) and by indirect immunofluorescence. IgG was purified from the patients' sera, and the influence of NE ANCA on the enzymatic activity of NE and on the inhibitory activity of SLPI was investigated by determining the hydrolysis of N-methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide by NE.
IgG from NE ANCA–positive sera of patients with CIMDL inhibited the enzymatic activity of NE and did not interfere with the activity of SLPI. In contrast to the findings in WG sera, measurement of PR3 ANCA in CIMDL sera showed only fair to moderate concordance between the 2 different capture ELISAs. Cross-inhibition experiments demonstrated that NE ANCA and PR3 ANCA represent distinct autoantibodies in patients with CIMDL.
The functional effects of NE ANCA on the enzymatic activity of NE or on the activity of SLPI cannot be implicated in the pathogenesis of CIMDL. The autoimmune reaction that targets neutrophil serine proteases in patients with CIMDL is frequently directed against more than one antigen. The ANCA response, including the reactivity of PR3 ANCA, in patients with CIMDL differs from what has been described in patients with WG.
Cocaine-induced midline destructive lesions (CIMDL) are an uncommon complication of habitual intranasal cocaine insufflation (1). Patients with CIMDL develop extensive destruction of the midfacial osteocartilagenous structures, which resembles involvement of the upper respiratory tract in patients with Wegener's granulomatosis (WG) (1).
Antineutrophil cytoplasmic antibodies (ANCA) are detectable in most patients with WG as well as in most patients with CIMDL. Whereas a cytoplasmic (cANCA) staining pattern predominates in WG, a perinuclear (pANCA) staining pattern is typical in CIMDL. In addition, most sera from patients with CIMDL react with human neutrophil elastase (NE) (1). ANCA specific for proteinase 3 (PR3), the typical finding in the sera of patients with WG, are present in ∼50% of patients with CIMDL (1). ANCA specific for NE represent a valuable diagnostic marker for CIMDL and are almost never detectable in patients with WG or in those with microscopic polyangiitis (MPA) (1).
It remains unknown whether ANCA contribute to the pathogenesis of CIMDL. Moreover, it has yet to be determined if the simultaneous recognition of both types, NE ANCA and PR3 ANCA, is attributable to cross-reaction of the antibodies or to the coexistence of distinct antigen-specific antibodies. An enzyme activity–enhancing effect of NE ANCA isolated from patients with MPA and from patients with osteomyelitis has been reported (2). Therefore, we investigated whether the binding of these ANCA to NE enhances the proteolytic activity of NE and/or interferes with its major physiologic antagonist, secretory leukoprotease inhibitor (SLPI) (3), on the mucosal surfaces of the upper respiratory tract. We also explored whether, in the sera of patients with CIMDL, the reacting NE ANCA and PR3 ANCA are cross-reacting antibodies, or whether each represents a distinct antibody population.
PATIENTS AND METHODS
Serum samples from 37 patients with CIMDL evaluated at Brescia University in Italy and at the Mayo Clinic in Rochester, Minnesota were analyzed. Samples from 5 healthy volunteers and from 5 cANCA-positive and PR3 ANCA–positive patients with WG served as controls. The Mayo Clinic Institutional Review Board approved the study. Due to the limited quantities of serum available, not all samples were included in all experiments.
All serum samples from the patients with CIMDL were evaluated with multimodality ANCA testing. ANCA staining patterns were assessed by indirect immunofluorescence (IIF) on ethanol-fixed neutrophils at serum dilutions of 1:4 and 1:16 (1). PR3 ANCA were measured by commercial direct enzyme-linked immunosorbent assay (ELISA), as well as by a capture ELISA with the MCPR3-2 capturing antibody and a PR3 anti–c-myc capture ELISA (4, 5). NE ANCA were determined using a sheep anti-NE capture ELISA (using polyclonal sheep anti-NE capturing antibody) and an NE anti–c-myc capture ELISA (1, 5). In the anti–c-myc assays, recombinant c-myc–tagged antigens were captured by an anti–c-myc antibody (P2241; Sigma-Aldrich, St. Louis, MO) (5).
IgG was purified by applying 0.5 ml of serum to a protein G column (Pierce, Rockford, IL), according to the manufacturer's instructions. The fractions with the highest spectrophotometrically measured absorbance (280 nm; A280) were pooled. The optical density (OD) of the pooled sample was measured at 280 nm, and the IgG concentration was calculated as follows: IgG (in mg/ml) = (A280/14) × 10.
Effect of NE ANCA–containing IgG on the hydrolytic activity of NE.
To determine the effect of NE ANCA on the hydrolysis of the peptide substrate N-methoxysuccinyl-Ala-Ala-Pro-Val p-nitroanilide (MeAAPV; Sigma-Aldrich), 12.5 ng of NE per well (Athens Research Technology, Athens, GA) was diluted in 25 μl of buffer (1% Triton X-100, 0.1M Tris, pH 8.0), followed by the addition of 25 μg of IgG in each well. Samples (50 μl/well) were incubated for 30 minutes at 37°C, and 50 μl of 2 mM MeAAPV in DMSO was added. The change in the OD per minute (mOD/minute) was measured every 30 seconds for 10 minutes, at 405–490 nm. Results are expressed as a percent of the values for the positive control (12.5 ng NE/well, MeAAPV in the absence of IgG).
Effect of NE ANCA–containing IgG on the inhibition of the hydrolytic activity of NE by SLPI.
To determine whether NE ANCA interfere with the inhibition of NE-mediated hydrolysis of MeAAPV by SLPI, 12.5 ng of NE/well was diluted in 20 μl of buffer (1% Triton X-100, 0.1M Tris, pH 8.0). Twenty-five micrograms of IgG was diluted in 20 μl of the same buffer and added to the NE in each well. Samples were incubated for 30 minutes at 37°C; 10 μl of SLPI (R&D Systems, Minneapolis, MN), diluted 1:100 in buffer (1% Triton X-100, 0.1M Tris, pH 8.0), and 50 μl of 2 mM MeAAPV in DMSO were added; this concentration of SLPI was previously determined to be sufficient to completely inhibit the enzymatic activity of 12.5 ng of NE. The mOD/minute was then measured every 30 seconds for 10 minutes, at 405–490 nm. Results were normalized by subtracting the values for the positive control (12.5 ng of NE and 10 μl of SLPI). This was followed by calculation of the residual enzymatic activity of NE, expressed as a percent of the values for the negative control (12.5 ng of NE without SLPI).
Serial dilutions of all serum samples that were found positive for the antibodies by the PR3 anti–c-myc capture ELISA and the NE anti–c-myc capture ELISA were analyzed. For each sample, the dilution yielding 50% of the maximum OD (OD50) was determined for each ANCA type. All subsequent experiments were conducted using this dilution. All assays were performed in duplicate.
Anti–c-myc–coated plates were loaded with a saturating dilution of serum-free medium supernatant of 293 cell clones expressing enzymatically inactive recombinant PR3 c-myc or NE c-myc for 1 hour at room temperature. Antigen-free control wells (anti–c-myc–coated plates incubated with serum-free medium supernatant from sham-transfected 293 cells, diluted 1:2 in IRMA buffer) served as negative controls. After 3 washes with 20 mM Tris, 500 mM NaCl, pH 7.5, and 0.05% Tween, 100 μl of patient sera per well, diluted in Tris buffered saline Immulite and 0.5% bovine serum albumin to the predetermined 50% saturation point, was preadsorbed for 24 hours at 4°C in antigen-coated wells and control wells. Following this preadsorption, the supernatants were analyzed by PR3 anti–c-myc capture ELISA and NE anti-c-myc capture ELISA (5).
GraphPad Prism 4.0 for Macintosh (GraphPad Software, San Diego, CA) was used for the statistical analyses. Differences in the ODs and changes in the enzymatic activity were compared with Wilcoxon's signed rank test and the Mann-Whitney U test, respectively. P values less than 0.05 were considered significant. Concordance was assessed using kappa statistics.
Effects of NE ANCA–containing IgG from patients with CIMDL on the enzymatic activity of NE.
To determine whether NE ANCA from patients with CIMDL enhance the enzymatic activity of NE, we measured the effect of NE ANCA–containing IgG on the hydrolysis of MeAAPV by NE. Of the 37 patients with CIMDL, 4 were excluded because their sera tested negative for the ANCA, 2 were excluded because of the detection of exclusively PR3 ANCA, and 5 were excluded because the quantity of serum needed to extract IgG was insufficient. Consequently, 26 serum samples were included in this experiment. Multimodality ANCA testing revealed that 16 of these serum samples were positive for both NE ANCA and PR3 ANCA, and 10 were positive for NE ANCA only. None of these IgG preparations enhanced the hydrolytic activity of NE.
Of the 26 CIMDL serum samples, IgG preparations from 16 samples inhibited the NE enzymatic activity by more than 20%. There was no difference in the extent of inhibition between the samples positive for both NE ANCA and PR3 ANCA and those positive for NE ANCA only. NE activity was decreased by more than 20% in only 1 of the 5 normal control IgG preparations and in none of the 5 PR3 ANCA–positive IgG preparations of samples from patients with WG (Figure 1A). These data indicate that NE ANCA–containing IgG from patients with CIMDL does not enhance, but rather inhibits the hydrolytic activity of NE.
Effect of NE ANCA–containing IgG on the inhibitory activity of SLPI in patients with CIMDL.
NE ANCA could interfere with the physiologic inhibition of NE activity on mucosal surfaces in patients with CIMDL. We therefore investigated the influence of NE ANCA–containing IgG on the inhibition of the enzymatic activity of NE by SLPI. Only 23 of the 26 CIMDL serum samples used in the previous experiment were available in sufficient quantity to be included in this experiment. On multimodality ANCA testing, 15 samples were positive for both NE ANCA and PR3 ANCA, and 8 were positive for NE ANCA only.
Our results showed that there was no interference with the inhibitory activity of SLPI on the hydrolysis of MeAAPV by NE in any of the 23 CIMDL serum samples, similar to the results obtained in the 5 WG serum samples and in the 5 normal control samples. Furthermore, no difference was noted between the CIMDL serum samples positive for both NE ANCA and PR3 ANCA and the samples positive for NE ANCA only. Representative results are shown in Figure 1B.
Distinguishing PR3 ANCA in patients with CIMDL from PR3 ANCA in patients with WG.
We also explored whether the PR3 ANCA in the sera of patients with CIMDL could be distinguished from the PR3 ANCA in the sera of patients with WG. Most capture ELISAs for detection of PR3 ANCA use monoclonal antibodies to capture the antigen, such as the monoclonal antibody against PR3 used in our MCPR3-2 capture ELISA (4); alternatively, c-myc–tagged recombinant antigens can be captured by antibodies recognizing the c-myc tag (5). We previously demonstrated very good concordance (99%; κ = 0.98) between the results obtained with each type of assay in samples from patients with WG (6). In contrast, in the 37 serum samples from patients with CIMDL, only a fair to moderate concordance (70% [26 of 37], 95% confidence interval 54–83%; κ = 0.42) between the 2 assays for PR3 ANCA was found. These observations suggest that the PR3 ANCA in patients with CIMDL target different epitopes than those targeted by the PR3 ANCA in patients with WG.
IIF was used to determine the ANCA staining pattern in the 26 CIMDL serum samples included in the function studies. In contrast to the typical cANCA pattern seen in WG sera, a pattern of pANCA staining was observed in the majority of the CIMDL sera (92.3% [24 of 26]), including all 10 samples that were positive for NE ANCA and 14 samples that reacted with both NE and PR3 by antigen-specific capture ELISAs. Only 2 patients double-positive for NE ANCA/PR3 ANCA had a cANCA IIF staining pattern. When tested by direct ELISA for PR3 ANCA, 11 (42%) of the 26 CIMDL serum samples tested positive for PR3 ANCA.
Cross-reactivity of NE ANCA and PR3 ANCA in patients with CIMDL.
Of the 37 CIMDL serum samples, 21 were NE ANCA positive by the NE anti–c-myc capture ELISA. Of these, 10 were also PR3 ANCA positive by the PR3 anti–c-myc capture ELISA. All 10 of these NE ANCA/PR3 ANCA double-positive samples were included in the cross-inhibition experiments.
To identify cross-reactivity between NE ANCA and PR3 ANCA, the CIMDL serum samples were incubated for 24 hours with PR3 or NE, or left without antigen. Following this preadsorption, the samples were reanalyzed by NE anti–c-myc capture ELISA and PR3 anti–c-myc capture ELISA. As expected, a significant reduction in ANCA reactivity was detected for NE ANCA following preadsorption with NE and for PR3 ANCA following exposure to PR3 (Figure 2). In contrast, the reactivity of NE ANCA was not significantly affected by preadsorption with PR3, or vice versa. Similar results were obtained for all 10 double-positive CIMDL samples included in this analysis (Figure 3). These data demonstrate that double-positive sera (for both NE ANCA and PR3 ANCA) from patients with CIMDL contain at least 2 distinct antibody specificities, rather than a single antibody cross-reacting with both NE and PR3.
NE ANCA from patients with MPA and from those with chronic staphylococcal osteomyelitis have been reported to enhance the enzymatic activity of NE (2). We speculated that this mechanism might also contribute to the pathogenesis of CIMDL. However, our data demonstrated an inhibitory effect of NE ANCA on NE enzymatic activity and revealed that there was no interference by NE ANCA with the activity of SLPI. These differences could be explained by the binding of different epitopes by NE ANCA in patients with CIMDL as compared with that observed in other diseases. Furthermore, our investigations examined the effect of the entire NE ANCA–containing IgG fraction, whereas the previous study examined the influence of affinity-purified NE ANCA (2). Given that, in vivo, patient sera contain a mixture of various antibodies rather than antibodies that target only a specific antigen, we believe that our approach more realistically reflects the interactions among the various antibodies. The cleavage of large molecular substrates of NE, such as elastin, is more sensitive to the effects of inhibitors than is the hydrolysis of MeAPPV by NE. Consequently, experiments with large molecular substrates were not pursued once we identified an inhibitory, rather than enhancing, effect for most of these NE ANCA.
Although our study shows that NE ANCA in patients with CIMDL do not enhance the enzymatic activity of NE, we cannot exclude the possibility that they are still involved in the pathogenesis of CIMDL through different mechanisms. Similar to the observed effects of PR3 ANCA and ANCA against myeloperoxidase (MPO) in patients with ANCA-associated vasculitides, NE ANCA may activate primed neutrophils, modify the clearance of apoptotic cells, or enhance and/or perturb neutrophil apoptosis (7, 8). Interestingly, a high frequency of apoptotic epithelial cells has been described in CIMDL (6).
Autoantibodies binding to both NE and PR3 are frequently detected in patients with CIMDL (1). Our cross-inhibition experiments clearly showed that the sera from patients with CIMDL that were double-positive for NE ANCA and PR3 ANCA contained at least 2 antigen-specific antibodies recognizing unique epitopes, rather than shared epitopes of NE and PR3.
The pANCA pattern is the predominant IIF staining pattern in patients with CIMDL, even in most of the CIMDL serum samples that also test positive for PR3 ANCA by direct or capture ELISAs (1). Concurrently, MPO ANCA are usually absent in CIMDL sera, suggesting that NE ANCA are responsible for most of the pANCA immunofluorescence staining.
Moreover, we demonstrated that detection of PR3 ANCA by antigen-specific capture ELISAs varies between sera from patients with CIMDL and sera from patients with WG (5). This is likely caused by competition of PR3 ANCA with the capturing antibody for specific target epitopes in selected patients with CIMDL, but not in patients with WG. Furthermore, whereas IgG preparations from many PR3 ANCA–positive patients with WG either inhibit or enhance the enzymatic activity of PR3 in the hydrolysis of MeAAPV, no change in the hydrolytic activity of PR3 was detected in experiments with IgG preparations from NE ANCA– and PR3 ANCA–positive CIMDL sera (Specks U, et al: unpublished observations). These results suggest that the nature of the antigen, and possibly the specific epitopes recognized by the specific autoantibodies, may determine the disease phenotype.
In conclusion, our data indicate that the functional effects of NE ANCA on the activities of NE or SLPI are unlikely to contribute to the pathogenesis of CIMDL. Moreover, despite some clinical similarities with the nasal lesions of WG, the autoantibody characteristics encountered in CIMDL are distinct from those of WG.
Dr. Specks had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Peikert, Specks.
Acquisition of data. Peikert, Hummel, McKenney, Gregorini, Trimarchi, Specks.
Analysis and interpretation of data. Peikert, Finkielman, Hummel, McKenney, Gregorini, Trimarchi, Specks.