The Binding of SLE Autoantibodies to Mitochondria

Abstract

Systemic lupus erythematosus (SLE) is a prototypic autoimmune disease characterized by immune complexes. Because these complexes contain mitochondrial components, we assessed the presence of antibodies to whole mitochondria (wMITO) using an ELISA in which mitochondria from mouse liver are bound to microtiter plates pre-coated with poly-L-lysine. Studies with this ELISA demonstrated that SLE plasmas contain abundant anti-wMITO activity. While digestion with DNase 1 did not affect anti-wMITO activity, adsorption of plasma on DNA affinity columns could reduce binding activity. Assay for anti-mitochondrial antibodies (AMA) by immunofluorescence and an ELISA with the M2 antigen (2-oxo-acid dehydrogenase protein complex) showed a low frequency of positivity, indicating that AMA and anti-wMITO are distinct specificities. In the study of 204 patients with SLE, the levels of anti-wMITO were higher in active SLE and correlated with levels of anti-DNA. These findings suggest that anti-wMITO can form immune complexes with mitochondria which may drive pathogenesis.

1. Introduction

Systemic lupus erythematosus (SLE) is a prototypic autoimmune disease characterized by the expression of autoantibodies to a wide variety of cellular antigens (1, 2). Of these antibodies, antibodies to the cell nucleus (antinuclear antibodies or ANAs) are highly characteristic, with ANAs to specific molecules representing valuable markers for classification, diagnosis and disease activity. The technologies for assaying ANAs are continuously evolving; in addition to the classical immunofluorescence assay (IFA), many different assay platforms are now available for antibody determinations using purified or cloned proteins as antigens to allow more precise identification of autoreactivity targets (3).

While ANAs can bind purified antigens, serological studies on SLE have recently analyzed the antigenicity of microparticles (MPs) and utilized MPs as a source of nuclear and other cellular antigens (4–10). Microparticles are small, membrane-bound vesicles that emanate from dead and dying cells by a blebbing process; in blebbing, bubble-like structures on the surface of a cell fill with molecules that have translocated from the nucleus (11, 12). While blebbing is a prominent feature of apoptosis, production of MPs may occur with other death forms including necroptosis, a form of programmed cell death mediated by RIP kinase enzymes (13).

As demonstrated in studies of MPs obtained from blood as well as cultured cells, MPs contain an ensemble of nuclear, cytoplasmic and membrane molecules (14–16). In addition, MPs can contain organelles such as mitochondria whose presence can be demonstrated by microscopy as well as biochemical analysis. While mitochondria can be a component of particles, mitochondria can also exist as particles and cells may release mitochondria into the extracellular space by an active process (17–24). Thus, in preparations of MPs from in vivo or in vitro sources, mitochondria can occur as free particles as well as particle constituents.

Previous studies using flow cytometry have demonstrated that mitochondrial components are present in the MP fraction of blood from SLE patients and furthermore bear IgG (4–10). In this study, we have explored whether SLE autoantibodies can bind directly to mitochondria, extending findings of a study by Becker et al (25). For our study, we used mitochondria prepared from murine liver and have also assessed whether a pre-coating with poly-L-lysine (PLL) can increase assay sensitivity as shown with supernatants of apoptotic cells (26). As results presented herein indicate, patients with SLE express antibodies that bind to mitochondria, with levels increased in active disease.

2. Materials and methods

2.1. Study samples

For the initial development of the assay, plasma samples from 15 individuals diagnosed with SLE were obtained from Plasma Services Group (Huntingdon Valley, PA, USA). For these samples, EDTA was the anti-coagulant. Ten plasma samples from healthy controls with EDTA as the anti-coagulant were purchased from Innovative Research, Novi, Michigan, USA. After the conditions for the ELISA assay had been established, 204 serum samples from individuals diagnosed with SLE were analyzed. The samples were obtained from the Karolinska lupus cohort. All participants were diagnosed with SLE and all fulfilled a minimum of four of the 1982 revised SLE classification criteria (27). Medical charts were reviewed by a rheumatologist who performed a clinical examination at inclusion following a structured protocol. SLE disease activity was determined with Systemic Lupus Activity Measure (SLAM) and SLE Disease Activity Index 2000 (SLEDAI-2K) (28–30).

Approval to work with the samples was obtained in accordance with the Department of Veterans Affairs policy. The samples were stored at −80°C. During the performance of the experiments, an aliquot of the original material was stored at 4°C to avoid freeze-thaw damage to the sample.

Twenty normal serum controls were also purchased from Innovative Research. Samples from patients with primary biliary cholangitis (PBC) were purchased from Plasma Services Group.

Before use, all the samples underwent a brief centrifugation (3000 × g for 2 min at room temperature [19 to 23°C]) to remove larger sized particulates. The resulting supernatant was transferred to sterile, microcentrifuge tubes for storage at 4°C.

2.2. Isolation of mitochondria from mouse liver

Mitochondria were isolated as described previously (31). Briefly, flash-frozen liver from C57BL/6 mice (obtained from the Jackson Laboratory and housed at the animal facility of the Durham VAMC) were thawed rapidly (30–60 s) at 45°C in thaw media (0.01 M Tris [Sigma-Aldrich, St. Louis, MO, USA], pH 7.5, 0.25 M sucrose [Sigma-Aldrich] and 0.4% bovine serum albumin [BSA; Sigma-Aldrich]). Once thawed, the liver was weighed, washed with ice cold isolation buffer (0.32 M sucrose, 1 mM EDTA [K⁺ salt; Sigma-Aldrich] and 10 mM Tris-HCl) and left in 10% wt./vol. of isolation buffer.

The liver was then homogenized on ice using a chilled Dounce homogenizer. The homogenate was then mixed at a 1:1 ratio with 24% Percoll (Sigma-Aldrich) in 0.32 M sucrose, 1 mM EDTA (K⁺ salt) and 10 mM Tris-HCl solution. This mixture was added as the top layer of an ultracentrifuge tube containing two pre-formed layers of Percoll (26% and 40%). The tubes were then ultracentrifuged (LC-7 model, Beckman-Coulter, Brea, CA, USA) at 30,700 g for 20 min at 4°C using a swing out rotor (SW 41 Ti; Beckman-Coulter). The band at the 26% and 40% interface containing the mitochondria was carefully removed and stored on ice for use on the same day or at −80°C for long-term storage.

After isolation, the mitochondria were analyzed for number and staining with MitoTracker Green FM (MTG; 10 nM; Thermo-Fisher-Invitrogen, Waltham, MA, USA). Analyses were performed using a FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). Using sizing beads (Thermo-Fisher-Invitrogen) that ranged in size from 0.2 μm to 2 μm, parameters were determined to allow detection of events within this size range by side scatter (SSC). For determination of mitochondrial numbers, the isolate was diluted serially in 1X TBS (Tris Buffered Saline; 50 mM Tris, pH 7.5 plus 155 mM NaCl) to produce a sample with events/seconds of no more than 1000. Samples were counted for 30 seconds.

For staining with MTG, the mitochondrial isolate was diluted to an event number of 1000 events/s and incubated with either 10 nM MTG or an equivalent amount of DMSO (Santa Cruz Biotechnology, Dallas, Texas, USA). DMSO was kept to no more than 0.5 μl per 200 μl staining reaction. Stained samples were incubated for 15 min at room temperature in the dark before analysis.

Data were collected using FlowJo Collector’s Edition software (Tree Star, Inc., Ashland, OR, USA). Using the flow rate of the machine (μl/s), the number of mitochondrial events per second was calculated.

2.3. Measurement of DNA by PicoGreen

The concentration of dsDNA present in the mitochondrial isolate was measured using the PicoGreen reagent (Thermo-Fisher-Invitrogen) according to the manufacturer’s instructions using λ DNA (Thermo-Fisher-Invitrogen) as standard.

2.4. ELISA to detect antibodies to mitochondria

Immulon 2HB plates (ThermoFisher Scientific) were incubated for 15 min at RT with 100 μl of 0.5 μg/ml of poly-L-lysine (PLL; MW 70 – 150K; Sigma-Aldrich). The plates were then washed 1X with 100 μl/well of ultra-pure H₂O (Thermo-Fisher – Invitrogen). After drying for 1 h at RT, 200 μl of block buffer II (2% bovine serum albumin [BSA], 0.05% Tween-20 [both from Sigma Aldrich] in 1X phosphate buffered saline, pH 7.4 [PBS-free; Ca²⁺ free, Mg²⁺ free]) was distributed to each well and incubated for 1 h at RT. Wells were washed with PBS-free. At this point, isolated mitochondria were diluted to the equivalence of 10 ng/ml of DNA in ELISA coating buffer (0.1 M sodium phosphate buffer, pH 9.0). Control wells were incubated with 100 μl/well of ELISA coating buffer/well. The prepared plates were incubated overnight at 4°C.

The plates were then washed with 1X PBS-free and 100 μl/well of diluted plasma or serum was added in ELISA dilution buffer (EDB; 0.1% BSA, 0.05% Tween-20 in 1X PBS-free). All reactions were performed in duplicate. The plates were incubated for 1 h at RT and then washed with 1X PBS-free. The wells were then incubated for 1 h at RT with 100 μl/well of a 1:1000 dilution of anti-human IgG (γ chain specific)-HRP conjugated (Sigma-Aldrich) in EDB. At this point, the plates were washed with 1X PBS-free and then incubated for 30 min at RT in the dark with 100 μl/well of TMB substrate solution (0.01% H₂O₂, 0.015% 3,3’,5,5’-tetramethylbenzidine dihydrochloride [both from Sigma-Aldrich] in 0.1 M citrate buffer, pH 4.0). The reaction was terminated by the addition of 100 μl, 2 M H₂SO₄ per well. The plates were then analyzed for absorbance at 450 nm using a spectrophotometer (Molecular Devices, San Jose, CA, USA).

For initial experiments to enhance the binding of mitochondria to polystyrene microtiter plates, the PLL concentration used was 2 μg/ml (100 μl per well). This concentration was reduced to 0.5 μg/ml to reduce non-specific background.

2.5. Proteomic analysis of isolated mitochondria by mass spectroscopy

Before analysis by mass spectroscopy, the isolated mitochondria were washed to remove Percoll reagent. Briefly, thawed mitochondrial isolate was mixed with mitochondrial isolation buffer (see above) at a ratio of at least 1:5 of isolate to isolation buffer. The mix was transferred to Ultra-Clear tubes (Beckman-Coulter) and subjected to ultracentrifugation using a swing-out rotor (Sw 41 Ti; Beckman-Coulter) in a LC-7 ultracentrifuge. Centrifugation was performed at 17800 × g (RCF average) for 15 min at 4°C. On completion, a loose pellet was seen. The supernatant was carefully removed and the pellet resuspended in mitochondrial isolation buffer. The samples were ultracentrifuged again at 7900 × g for 12 min at 4°C. On completion, the supernatant was carefully removed and the pellet was resuspended in 500 μl isolation buffer, transferred to an autoclaved, 1.5 ml microcentrifuge tube and centrifuged at 6900 × g for 10 min at 4°C. On completion, the pellet was drained and stored at −80°C until further analysis.

Samples were analyzed by mass spectrometry by the Duke University Proteomics & Metabolomics Shared Resource. A mitochondrial pellet was lysed by probe sonication in 50 μl of 50 mM triethylammonium bicarbonate buffer, pH 8.5 (TEAB) containing 5% (w/v) SDS. 20 μg of protein was reduced and alkylated, followed by trypsin using an S-trap micro device (Protifi). The sample was reconstituted in 1% trifluoracetic acid and 2% acetonitrile, and 1 µg of peptides were analyzed by LC-MS/MS used a nanoACQUITY UPLC system (W aters) coupled to a Fusion Lumos high resolution accurate mass tandem mass spectrometer (Thermo) via a nanoelectrospray ionization source. The analytical separation used a 75 µm × 250 mm column (Waters) with a 90 min gradient of 5 to 30% MeCN with 0.1% formic acid at a flow rate of 400 nl/min and column temperature of 55 °C.

Data collection on the Fusion Lumos MS was performed in data-dependent acquisition (DDA) mode with a 120,000 resolution (@ m/z 200) full MS scan ion trap MS/MS using charge state filtering, monoisotopic precursor selection (MIPS) and a dynamic exclusion of 20 s. MS/MS was performed using HCD with a collision energy of 30 ± 5% with detection in the ion trap using rapid scanning, an AGC target of 5e3, max IT of 100 ms and max parallelization enabled. Data were converted to Mascot generic file (mgf) using Thermo Proteome Discoverer 2.2 followed by database searching against a Swissprot mouse database with reverse decoys using Mascot Server (v. 2.5) using a precursor tolerance of 5 ppm, a product ion tolerance of 0.8 Da, fixed carbamidomethyl (C), variable deamidation (NQ), oxidation (M) and acetylation (protein N-termini). Data were annotated at a 1% peptide and protein false discovery rate using Scaffold (Proteome Software). 1420 proteins were identified in the sample. Scaffold v4.8.4 was used for further analysis of the identified proteins.

2.6. Effects of DNase 1 digestion on antigenicity

To determine the contribution of DNA to the antigenicity of the mitochondria, the effects of DNase 1 (Ambien-ThermoFisher; 10 U/ml per reaction) were tested. The following preparations were assessed: mitochondrial isolate, CT DNA, and no antigen with and without DNase 1. The amount of isolate used was determined to allow for the addition of 1 ng of DNA per well of the ELISA assay. This amount was calculated using the post-digest DNA concentrations obtained from samples without DNase 1; an identical volume equivalent per well was added from the corresponding conditions with DNase 1. Mitochondrial isolate contains 1 mM EDTA. The online program, MAXCHELATOR webmaxC standard version (Stanford University, Palo Alto, CA, USA), was used to calculate the amounts of additional magnesium and calcium required to be added to produce a DNase 1 reaction with free calcium and magnesium concentrations of 0.5 mM and 2.5 mM, respectively.

The reactions were incubated for 1 h at 37°C. On completion, the amount of DNA remaining after DNase 1 reaction was measured by PicoGreen assay. Average reductions in DNA were 4.7 % and 92.0 % (n= 3) for mitochondria and CT DNA, respectively. Based on these post DNase 1 reaction values, the without DNase 1 reactions were diluted to a DNA equivalent of 10 ng/ml and 50 ng/ml per well of mitochondrial isolate DNA and CT DNA, respectively, with the appropriate ELISA coating buffer (see above). 100 μl per ELISA well were aliquoted to prepared ELISA plates (see above). For each antigen, equal volumes equivalent to the DNase 1 reactions were also diluted in the appropriate coating buffer. For no antigen DNase 1 reactions, the same volumes used for the mitochondrial isolate with and without DNase 1 reactions were used to prepare the ELISA plating solutions. The coated ELISA plates were incubated overnight at 4°C. The next day the ELISA experiment proceeded as described above.

2.7. Effects of adsorption with DNA on antibody binding

For DNA adsorption experiments, stock solutions of cellulose (Type 50, 50 μm) and dsDNA-cellulose, both from Sigma-Aldrich, were incubated overnight in 1X TE buffer. The next day, the cellulose was washed with TE and left in a ratio of 1:3 of resin to 1X TE. All centrifugation steps to pellet the resin were performed at 1000 rpm for 5 min at RT. Required amounts of resin were transferred to a 15 ml conical tube and washed with EDB. After the final wash, the resin was left in a 1:1 ratio of resin to EDB. For each plasma, two treatment reactions were performed: 1) cellulose, and 2) dsDNA cellulose.

Each absorbance reaction consisted of 2 ml of a 1:300 dilution of plasma plus 666 μl of packed resin, as appropriate. The reactions were tumbled for 4 h at RT, followed by tumbling overnight at 4°C. On completion, the samples were spun at 1000 rpm for 5 min at RT and the supernatant removed and stored on ice.

Antigens for assay were mitochondria and native double-stranded calf thymus DNA (CT DNA, Worthington Biochemical Corp., Lakewood, New Jersey, USA). Six SLE plasmas from the Plasma Services Group that showed strong binding to mitochondria and DNA during initial ELISA experiments were selected for the assay. Plasma from a healthy donor was also included. ELISA plates coated with mitochondria were set up as described above. Plates with CT DNA as antigen were prepared by dispensing 100 μl per well of 5 μg/ml of CT DNA in SSC (saline-sodium-citrate buffer; 150 mM NaCl, 15 mM Na citrate, pH 7.0). Plates were incubated overnight at 4°C. The following day, the plates were washed with PBS-free and incubated with block buffer II (see above) for 2 h at RT. Control plates with no antigen were also included to show background levels of binding.

The dilutions of treated plasma ranging from 1:500 to 1:12150 by 3-fold dilution were prepared in EDB. On completion of the blocking of CT DNA and control plates, all ELISA plates were washed with PBS-free. The diluted plasma samples were distributed to the ELISA plate and the ELISA procedure performed as described above.

2.8. The effects of adsorption with mitochondria on antibody binding

To have a preparation for adsorption, mitochondria were isolated from mouse liver and then washed immediately as described above to remove the Percoll reagent. The final pellet of mitochondria was resuspended in isolation buffer. The total isolate from one mouse liver yielded a washed mitochondrial pellet that was resuspended in 400 μl of isolation buffer. DNA present in the washed preparation was measured by PicoGreen, with the content of mitochondria assessed by flow cytometry using MitoTracker Green FM.

To assess the effects of adsorption with mitochondria, three plasmas obtained from Plasma Services Group were diluted with EDB to an expected ELISA activity level at OD 450nm of approximately 1.5 for the anti-wMITO or anti-DNA assays and were incubated with washed mitochondria or a preparation control. Plasma dilutions ranged from 1:1200 – 1:7500; for two of the plasmas, the dilution differed for anti-wMITO and anti-DNA assays because of the respective activities in the two assays.

For adsorption washed mitochondria, individual reactions containing an equivalence of DNA at 0, 3, 6 or 12 ng per ELISA assay well (100μl/well) were established. Reactions containing equivalent volumes of preparation control were assembled. All reactions were incubated for 3 h on ice with occasional mixing by gentle tapping. At the end of the incubation, the reactions were centrifuged in a Symphony 2417R centrifuge at 6,900 x g for 10 min at 4°C to pellet the mitochondria. The supernatants were carefully removed and stored at 4°C before assay for anti-wMITO and anti-DNA activity.

2.9. Detection of anti-mitochondrial antibodies by ELISA

Levels of anti-mitochondrial antibodies directed to the mitochondrial M2 antigen (2-oxo-acid dehydrogenase protein complex) were measured using the Anti-M2-3E ELISA from EuroImmun (Mountain Lakes, NJ, USA). The ELISA was used as per the manufacturer’s directions. Serum samples were diluted 1:100 with kit provided dilution buffer before use in the assay. Plasma samples from donors diagnosed with PBC were included in the assay to act as positive controls. PBC plasmas were also diluted 1:100 with dilution buffer provided in the kit. Plasmas and sera from normal healthy donors were also included as negative controls. On completion of the ELISA, sample absorbance at 450 nm was measured using a spectrophotometer.

2.10. Detection of anti-mitochondrial antibodies (AMA) by immunofluorescence assay

204 serum samples from individuals diagnosed with SLE and 20 serum samples from normal donors were examined for the presence of AMA. The samples were analyzed using the AMA indirect immunofluorescence test from EuroImmun. The kit was used according to the manufacturer’s directions except that SlowFade Diamond Antifade Mountant (ThermoFisher Scientific) was used instead of the mounting agent in the kit to help preserve sample fluorescence during analysis. Samples were used at a dilution of 1:100 for the assay. The slides were analyzed using an EVOS FL Cell Imaging System (ThermoFisher Scientific). An objective lens of 20X was used and the light source was an adjustable intensity LED.

2.11. Statistical methods

Anti-wMITO determined at a dilution of 1:800 within quartiles of disease activity were compared using ANOVA, with Student’s paired t-test as post-hoc analysis.

3. Results

3.1. Mitochondrial preparation

In these studies, we have explored the antigenic properties of mitochondria that have been isolated from mouse liver using a well-validated method to prepare mitochondria for studies of mitochondrial structure and function (31). Following purification, the mitochondrial preparations were stained with MitoTracker Green FM (MTG) and analyzed by flow cytometry using buffer as control. Figure 1 shows a representative example of the percentage of the mitochondrial isolate staining above background (buffer only isolate). The isolated mitochondria showed staining of 98.4% when compared to background MTG staining levels. Analysis of six mitochondrial preparations showed staining of 91.4% ± 8.2 (mean ± S.D.).

Figure 1. Analysis of mouse liver mitochondria with MitoTracker Green FM.

Mitochondria were isolated from mouse liver as described in the Materials and methods section, with the percentage of mitochondria in the preparation determined by staining with MitoTracker Green (MTG) FM. The data show the results of staining mitochondria with MTG FM (gray peak) as compared to buffer which had gone through the same purification steps (unfilled peak). Collected data were analyzed in terms of the percentage of events in the mitochondrial preparation with staining above background.

3.2. Development of the ELISA

For the ELISA, we first compared the binding of SLE plasma to plates either directly coated with mitochondria or plates pre-coated with poly-L-lysine (PLL) to increase mitochondrial adherence. The use of a PLL pre-coat was previously used in studies to assess antibody binding to MPs and supernatants of apoptotic cells that contain MPs (26). Studies indicate that mitochondria have a net negative charge and therefore would be likely to bind to PLL (32–34).

For the development of this assay, we used a set of SLE plasmas that showed binding activity in preliminary experiments. We have designated antibodies detected in this assay as antibodies to whole mitochondrial antibodies or anti-wMITO. Figure 2 presents the results of these experiments. As these data indicate, the presence of PLL leads to a significant increase in the binding of antibodies to mitochondria, although reactivity of mitochondria directly coated to plates could also be detected.

Figure 2. Effect of poly-L-lysine (PLL) on anti-wMITO ELISA.

Mitochondria prepared from mouse liver were incubated with ELISA plates coated with and without 0.5 μg/ml PLL. After washing, wells were then incubated with dilutions of plasma of SLE patients (A, B and C) or normal plasma (D) and antibody binding determined. Results with circles indicate values with PLL while results without PLL are indicated with squares. Filled markers represent results with mitochondria present whereas open markers show results with no mitochondria. Mean and standard deviations of duplicate reactions per condition are shown. Results shown are representative of 15 SLE and three normal plasmas tested.

Because PLL can bind to nuclear molecules, likely on the basis of charge, positivity in this assay could result from binding of circulating ICs that contain DNA and RNA. For the purpose of this assay, the coating concentration of PLL was limited and plasmas were tested with and without bound mitochondria. In general, the extent of binding to just PLL was low, indicating that antibodies detected were directed to antigens in the mitochondrial preparation.

3.3. Proteomic analysis of mitochondrial preparation

In these experiments, we used mitochondria that had been prepared during a standard procedure developed for isolating mitochondria (31). The proteome of mitochondria is complicated and dynamic and can include proteins from other organelles that can associate with mitochondria depending on the tissue, state of the cell and isolation techniques (35–37). We, therefore, used proteomics to define further the protein composition of our preparation. The relative amount of the most abundant proteins was ranked based on spectral counts (Table 1), suggesting that 28 of the top 50 proteins were mitochondrial (data not shown). In addition, eight of the proteins were peroxisomal, eight were from the endoplasmic reticulum and five were cytoplasmic.

Table 1:

Proteomic analysis of mitochondrial preparation used as antigens

Protein	Molecular Weight (kDa)	Exclusive Spectrum Count
A: Top 10 proteins identified/ranked by exclusive spectra count per protein
1 Carbamoyl-phosphate synthase [ammonia] (mitochondria)	165	3473
2 Uricase, (peroxisome)	35	1171
3 Catalase, (peroxisome)	60	1053
4 ATP synthase subunit beta (mitochondria)	56	803
5 Peroxisomal acyl-coenzyme A oxidase 1 (peroxisome)	75	746
6 ATP synthase subunit alpha (mitochondria)	60	705
7 Non-specific lipid-transfer protein (endoplasmic reticulum)	59	696
8 3-ketoacyl-CoA thiolase (mitochondria)	42	581
9 Glutamate dehydrogenase 1 (mitochondria)	61	468
10 60 kDa heat shock protein (mitochondria)	61	472
B: Identified histone proteins
1 Histone H2B type 1-B	14	10
2 Histone H4	11	7
3 Histone H2A type 1-F	14	6
4 Histone H3.1	15	6
C: Identified known SLE target antigens
1 60S acidic ribosomal protein P0	34	15
2 60S acidic ribosomal protein P2	12	6

The protein composition of mitochondria prepared from mouse liver was analyzed by mass spectroscopy. Proteins identified by exclusive spectra were determined as described in Materials and methods. Section A presents the top 10 proteins in terms of their abundance with intracellular location indicated in parentheses. Section B lists histones while Section C lists known target autoantigens in SLE other than histones.

As these studies indicate, a large number of mitochondrial proteins were well represented, although proteins from other cellular sources were present. Of note, we did not demonstrate the presence of known target antigens such as Sm, RNP, Ro, or La. We did find peptides from the ribosomal P proteins as well as sequences related to histones. The preparation we used can be operationally defined as mitochondria recognizing that components other than mitochondria may be present.

3.4. Effects of DNase 1 digestion on antigenicity

Studies by Becker et al (25) showed a relationship between antibodies to mitochondria and anti-DNA in SLE sera. To further assess this relationship, we first determined the effect of DNase 1 digestion on the antigenicity of mitochondria. As these results indicate (Figure 3), concentrations of DNase 1 adequate to eliminate the antigenicity of a DNA preparation did not reduce the binding to mitochondria.

Figure 3. Effects of DNase 1 digestion on the antigenicity of mitochondria and DNA.

Mitochondria prepared from mouse liver and CT DNA were digested with DNase 1 (10U/ml for 1 h at 37°C). Control reactions without DNase 1 were also conducted. DNA present in samples following treatment was analyzed by PicoGreen to assess the extent of the digestion. The treated preparations were used as antigens in the ELISA. Results from two SLE plasmas are shown with mitochondria and DNA represented as squares and circles, respectively. Open markers show results after DNase 1 digestion; filled markers indicate results without DNase 1 digestion. Mean and standard deviations of duplicate reactions per condition are shown. Results are representative of the five plasma tested.

3.5. Effects of DNA adsorption on antibody binding

As another approach, we adsorbed plasma with DNA cellulose to deplete anti-DNA reactivity; we then tested the adsorbed plasma in the ELISA. Figure 4 shows the change in reactivity of an SLE plasma in response to anti-DNA removal. Table 2 contains the data from all of the SLE plasmas tested. These experiments indicate that adsorption of DNA variably reduces the anti-wMITO activity depending on the plasma suggesting that at least some of the reactivity to mitochondria results from anti-DNA antibodies.

Figure 4. Effects of DNA adsorption on reactivity of plasma to DNA and mitochondria.

SLE (A) and normal plasma (B) were adsorbed with DNA cellulose to remove anti-DNA antibodies while controls reactions were adsorbed with cellulose. Dilutions (1:500 – 1:121500) of the adsorbed plasma were used to assess binding to DNA and mitochondria by ELISA. Squares show binding to mitochondria; circles show binding to CT DNA. Open markers show results of pretreatment with DNA cellulose and filled markers indicate plasma pretreated with cellulose. Mean and standard deviations of duplicate reactions per condition are shown. Result shown for SLE plasma is representative of six plasma tested.

Table 2:

Effect of DNA adsorption on antibody binding to mitochondria and DNA

Adsorbent Pretreatment ELISA Antigen	Cellulose Mitochondria	DNA Cellulose Mitochondria	Cellulose DNA	DNA Cellulose DNA
Plasma
Normal	0.21	0.12	0.08	0.07
SLE 5	2.32	0.83	2.20	0.08
SLE 9	1.92	1.00	1.89	0.10
SLE 10	2.39	1.34	2.32	0.12
SLE 14	1.53	0.55	0.53	0.11
SLE 16	2.48	2.00	0.38	0.09
SLE 19	0.56	0.19	1.71	0.07

Results are presented as terms of OD 450 nm values for reactivity of plasmas following adsorption of anti-DNA antibodies by DNA. Control adsorptions were performed with cellulose. Results reported are values for the ELISA assays using plasmas at a dilution of 1:500.

3.6. Effects of mitochondria adsorption on antibody binding

The next experiments to elucidate the relationship between anti-wMITO and anti-DNA involved adsorption of plasma with mitochondria, with the amount of mitochondria used for the adsorption determined by the amount of DNA as measured with PicoGreen. Three plasmas with significant levels of both anti-wMITO and anti-DNA were tested. As results in Figure 5 indicate, adsorption with mitochondria leads to a reduction of both anti-wMITO and anti-DNA, providing further evidence that antibodies to wMITO include anti-DNA antibodies.

Figure 5. The effects of adsorption with mitochondria on anti-wMITO and anti-DNA activity of SLE plasma.

Three different SLE plasma (#5, 9 and 10) were each adsorbed with different concentrations of washed, mitochondria isolated from mouse liver. The concentrations were 0 (no adsorption control) 3 ng, 6 ng or 12 ng. After removal of the mitochondria by centrifugation, the plasmas were analyzed by ELISA for reactivity to either wMITO (A) or CT DNA (B).

3.7. Relationship of anti-wMITO and anti-DNA

Having explored the antigenicity of mitochondria, we then performed ELISA assays for anti-DNA and anti-wMITO with a large number of sera (204) from patients with SLE. As these results indicate (Figure 6), antibody binding to mitochondria occurred commonly among sera. Levels of these antibodies were related to those of anti-DNA.

Figure 6. Relationship of anti-DNA and anti-wMITO in patients with SLE.

204 sera of patients with SLE and sera from 20 normal, healthy patient samples were assessed for binding by ELISA to wMITO and DNA. Samples were tested in duplicate and the mean and standard deviation for each condition were calculated. The mean OD 450 nm values of SLE serum in the anti-wMITO and anti-DNA assays were plotted. Linear regression analysis was performed on the data and an R² value of 0.461 was determined.

3.8. Anti-mitochondrial activity of SLE sera

In view of these data, we next assessed the presence of antibodies to mitochondrial antigens of the kind present in PBC. This autoimmune condition is characterized by the anti-mitochondrial antibodies (AMA) that can be detected by either immunofluorescence assays using organ tissues as a substrate or ELISA or immunoblotting with purified or cloned antigens (38). Among AMA specificities are antibodies directed against the M2 subunits of the 2-oxo-acid dehydrogenase protein complex. As shown in Table 3, the lupus sera showed infrequent activity to mitochondria in either of these formats. The reactivity in SLE is therefore distinct and suggests a novel set of specificities.

Table 3:

Detection of anti-mitochondrial antibodies (AMA) in SLE patient sera by ELISA and IIFT

	Anti-M2-3E ELISA	IIFT AMA
Positive	11	8
Negative	193	196

204 sera of patients were tested for anti-mitochondrial antibody (AMA) using an ELISA and Indirect immunofluorescence test (IIFT), both manufactured by EuroImmun. For both assays, the samples were diluted 1:100 in dilution buffer provided with the kits. Sera from normal healthy individuals were also included in both tests (n=20); all tested negative by ELISA and one sample was very weakly positive by indirect immunofluorescence testing. Of the SLE samples that tested positive, only two showed agreement between both test methods.

3.9. Relationship of anti-wMITO and disease activity

To assess a relationship of antibody levels to disease activity, the results of the ELISA were compared to measures of disease activity obtained using the SLEDAI and SLAM. Anti-wMITO were associated with SLAM (p=0.006) and SLEDAI (p<0.0001) overall, analyses by quartiles shown in Figure 7. Among SLAM-specific items, anti-wMITO levels were significantly associated with low hemoglobin levels (p<0.0001); higher sedimentation rates (p<0.0001); abnormal urine samples (p=0.02); the presence of lymph nodes (p=0.006); myalgia (p=0.007); and fatigue (p=0.04). Among the SLEDAI items, anti-wMITO levels were associated with arthritis (p<0.0001), hematuria (p=0.01), urinary casts (p=0.05), proteinuria (p=0.04), dsDNA antibodies (p<0.00001) and low complement (p=0.0004). Taken together, these results indicate the presence of anti-wMITO is associated with various manifestations of disease activity.

Figure 7. Relationship between anti-wMITO and disease activity in SLE.

The figure shows the relationship between anti-wMITO binding and SLAM and SLEDAI values for 204 patients plotted in terms of quartiles of disease activity. P values for differences between the specified quartiles are indicated on the figure.

4. Discussion

The studies provide new insights into the binding of SLE antibodies to mitochondria and demonstrate a relationship of anti-wMITO and anti-DNA. Thus, in a set of samples assessed in detail, reactivity was reduced by adsorption on a DNA affinity column; correspondingly, adsorption with mitochondria reduced anti-DNA activity. In addition, we showed that levels of antibodies to mitochondria are correlated with those for anti-DNA. These studies extend observations that ICs in patients with SLE contain mitochondrial antigens (10); they are also consistent with another study using isolated mitochondria as antigens for solid phase assays (25). Together, these results suggest that mitochondria could be the nidus for the formation of ICs with anti-DNA as well as other antibodies which operationally can be called antibodies to whole mitochondria (anti-wMITO).

To develop an ELISA for determination of anti-wMITO, we used solid phase matrix pre-coated with PLL. As our studies showed, this pre-coat dramatically increases the binding of mitochondria to the solid phase. In previous studies, we have used a PLL pre-coat to adhere supernatants from cells undergoing apoptosis to the solid phase; these supernatants contain microparticles along with soluble autoantigens (26). The pre-coat was also effective in allowing detection of microvesicles in studies on the role of these structures in malignancy (39). Thus, the pre-coat provides a much more robust assay to explore the biochemical basis of antibody binding to mitochondria and assess quantitatively anti-wMITO activity in clinical studies.

In these experiments, we used mitochondria from mouse liver prepared using a well-validated method for isolation of functionally intact forms of this organelle (31). The use of these preparations for assessing antigenicity can be complicated because of uncertainties in the nature of proteins in the mitochondrial preparations (35–37). Mitochondria are dynamic structures and can bear proteins from other organelles and cellular compartments; in addition, tissues may differ in the proteins found in mitochondria. Given these issues, the presence of other organelle components may not be surprising since the procedure involves Dounce homogenization to break open cells. Furthermore, the similarity in the structure, size and topological relationship of different organelles makes physical separation by techniques such as gradient centrifugation difficult and data suggest that many proteins can occur in more than one organelle (40).

The proteomic study provides further insight into the nature of the preparation we used to assess anti-wMITO. These studies indicated that, as expected, our preparation abundantly contained mitochondrial proteins, although we did identify a number of proteins with peroxisomal localization. As shown in many studies, mitochondria and peroxisomes are structurally and functionally related by the “peroxisome-mitochondria connection” (41–44). Mitochondria can also be associated with other organelles at membrane site contacts (MSCs). This situation limits identification of the organelle location of certain proteins and there is evidence, for example, that catalase, which is usually considered a peroxisomal protein, can occur in liver mitochondria as our proteomics study suggests (45).

We also used the proteomic analysis to assess the relative abundance of characteristic lupus autoantigens. As these studies showed, we did not find evidence of Sm, RNP, Ro or La although some histone-related proteins were present. The presence of histone-like proteins have been found in mitochondria and there is trafficking of proteins between mitochondria and the nucleus (46–48). Finally, we did find peptides related to ribosomal P proteins. Thus, while we found proteins usually considered as components of other organelles, we believe that the mitochondria preparations that we used provide a relevant and reliable representation of mitochondrial proteins and that mitochondrial components are the main target of reactivity for the SLE sera. Indeed, our analysis by flow cytometry using MTG indicates the high content of mitochondria in our preparations. Other studies have reached similar conclusions although these studies did not perform an in-depth proteomic study (25).

While the SLE sera bound well to the intact mitochondria, the ELISA against the M2 subunits of the 2-oxo-acid dehydrogenase protein complex showed little evidence of antibody binding. Similarly, using an IFA, only eight sera showed reactivity. These results indicate that the reactivity detected in the ELISA is distinct from the more classically defined AMA that are found in the sera of patients with PBC. Prior studies indicated that AMA, while sometimes observed in SLE serum, are not a common specificity, although patients with PBC can develop SLE and vice versa (49–54). In this regard, we did not detect significant binding of the PBC sera in the anti-wMITO ELISA suggesting that PBC target antigens are not readily accessible. As such, these results imply that the mitochondria in our preparations are relatively intact.

Our studies suggest that DNA or a structurally related cross-reactive antigen is a relevant target of anti-wMITO antibodies since adsorption with DNA can reduce anti-wMITO activity; for the three plasmas tested, adsorption with mitochondria also reduced binding to both wMITO and DNA. Furthermore, the levels of anti-DNA and anti-wMITO were related in a cross-sectional study. In contrast, in preliminary studies, we did not find a relationship to levels of anti-wMITO and antibodies to cardiolipin in assays assessing IgG, IgM and IgA responses (preliminary observations). Cardiolipin is highly characteristic of mitochondria and is an important component of the inner mitochondrial membrane. Depending on the state of the cell, cardiolipin can undergo movement in processes such as apoptosis (55, 56). Our preparation came from normal liver, however, where the amount of apoptosis would be expected to be limited.

While DNA may be one of the antigens present in the mitochondrial preparation, its origin is less clear. Mitochondria has its own DNA which codes for tRNA and ribosomal proteins (57). Like bacterial DNA, mitochondrial DNA is not organized into nucleosomes; rather than histones, a protein called TFAM is important in compacting and condensing DNA to fit into the mitochondria. These structures are termed nucleoids and contain a variety of other proteins involved in transcription and replication (57). For these experiments, we used a method that yields functionally active mitochondria, implying an intact structure with DNA on the inside of the mitochondria. It is possible, however, that, during the preparation which involves Dounce homogenization, some mitochondria are damaged or made permeable to exposed DNA.

While mitochondrial DNA may be on the interior of mitochondria, these organelles are dynamic and undergo structural changes including fission-fusion changes. These changes may allow exposure of DNA. Inversion of particle components can also occur as shown with proteins (58). Finally, it is possible that DNA released from the nucleus could bind to the mitochondria (59). In preliminary studies, we have shown by PCR that our preparation does contain nuclear as well as mitochondrial DNA. Studies are in progress to determine the mechanisms by which DNA arises on the surface of mitochondria in an antigenically accessible form.

As our studies showed, levels of anti-wMITO are related to measures of disease activity as assessed by both the SLAM and SLEDAI indices, comparing levels of anti-wMITO with levels of disease activity expressed quartiles. In this analysis, we specifically noted clinical associations with measures of arthritis and renal disease. In addition to showing a relationship of anti-wMITO and dsDNA antibodies and low complement, we also observed positive associations with laboratory measures such as low hemoglobin levels and high sedimentation rates. We also found a similar relationship between levels of disease activity and anti-DNA antibody levels (data now shown). Together, these results suggest that antibodies to wMITO and DNA can both contribute to disease activity. The most likely mechanism involves the formation of immune complexes, with circulating mitochondria as one source of DNA antigen. Our study on the presence of mitochondrial components in IgG-containing microparticles provides evidence for this possibility (10).

As we have shown previously, cells undergoing both apoptosis and necroptosis release mitochondria either as particles or as components of MPs (13). Other investigators have documented the release or extrusion of mitochondria during a variety of cell processes (17–24). Like microparticles which are also released during cell death, mitochondria may be able to bind other molecules, either specifically or non-specifically, to create an extracellular structure that displays relevant autoantigenic moieties; such structures can also contain cytokines which further increase the potential to drive inflammation and induce autoantibody production. Future studies will identify these autoantigens and determine the role in the pathogenesis of SLE.

Highlights:

• Blood of patients with SLE contain antibodies to mitochondria

• Levels of antibodies to mitochondria are related to those of anti-DNA

• Antibodies to mitochondria are associated with disease activity