Human autoantibodies to diacyl-phosphatidylethanolamine recognize a specific set of discrete cytoplasmic domains

Abstract

The aim of this study was to characterize a novel human autoantibody–autoantigen system represented as cytoplasmic discrete speckles (CDS) in indirect immunofluorescence (IIF). A distinct CDS IIF pattern represented by 3–20 discrete speckles dispersed throughout the cytoplasm was identified among other cytoplasmic speckled IIF patterns. The cytoplasmic domains labelled by human anti-CDS-1 antibodies did not co-localize with endosome/lysosome markers EEA1 and LAMP-2, but showed partial co-localization with glycine–tryptophan bodies (GWB). CDS-1 sera did not react with several cellular extracts in immunoblotting and did not immunoprecipitate recombinant GW182 or EEA1 proteins. The typical CDS-1 IIF labelling pattern was abolished after delipidation of HEp-2 cells. Moreover, CDS-1 sera reacted strongly with a lipid component co-migrating with phosphatidylethanolamine (PE) in high performance thin-layer chromatography (HPTLC)-immunostaining of HEp-2 cell total lipid extracts. The CDS-1 major molecular targets were established by electrospray ionization–mass spectrometry (ESI-MS), HPTLC-immunostaining and chemiluminescent enzyme-linked immunosorbent assay as diacyl-PE species, containing preferentially a cis-C18 : 1 fatty acid chain at C-2 of the glycerol moiety, namely 1,2-cis-C18 : 1-PE and 1-C16 : 0-2-cis-C18 : 1-PE. The clinical association of CDS-1 sera included a variety of systemic and organ-specific autoimmune diseases but they were also observed in patients with no evidence of autoimmune disease.

Introduction

Several autoantibody systems have been identified and characterized based initially on a distinctive indirect immunofluorescence (IIF) pattern followed by more definitive identification by double immunodiffusion, immunoblotting, immunoprecipitation and/or expression cloning and immunoscreening of cDNA libraries [1,2]. Such approaches have allowed the characterization of a large array of target nuclear antigens; however, relatively few cytoplasmic autoantigens have been characterized or identified at the molecular level. The aim of the present study was to characterize a peculiar autoantibody–autoantigen system associated with the cytoplasmic discrete speckled (CDS) staining pattern, referred to sometimes in the literature as the ‘lysosomal pattern’[3]. This latter terminology may be an oversimplification, as different morphologically similar domains exist in the cytoplasm, including a number of related organelles comprising the endocytosis, exocytosis and lysosome pathways [4], and the recently described distinct glycine–tryptophan body (GWB) system that is involved in mRNA processing [5].

Endocytosis is a process whereby cells internalize extracellular molecules into cytoplasmic vesicles, regulate cell-surface receptor expression, maintain cell polarity and modulate antigen presentation. In mammalian cells, cargo can be endocytosed via clathrin-coated pits or via the clathrin-independent pathway or caveolae. All internalization pathways deliver cargo molecules to early endosomes. From early endosomes, cargo can be recycled back to the plasma membrane, sorted to the late recycling compartments or delivered to late endosomes. Late endosomes either mature into lysosomes or transfer cargo to lysosomes through vesicular intermediates. Several vesicles of the endosome–lysosome system can be identified by their surface markers. Early endosome markers include GTP-binding proteins (Rab4, Rab5 and Rab11), transferrin receptors [4] and early endosome antigen 1 (EEA1) [6]. Late endosome and lysosome markers include Rab7, Rab9 and LAMPs/lgp (lysosomal membrane glycoproteins) [4].

Recently, a novel discrete cytoplasmic domain not co-localizing with markers of the Golgi complex, endosome, lysosomes or peroxisomes was reported [5]. The probe used to identify this cytoplasmic complex was the serum from a patient with motor and sensory neuropathy containing autoantibodies against a 182 kDa protein (GW182) that binds mRNA and contains numerous glycine–tryptophan (GW) repeats. The cytoplasmic structures recognized by anti-GW182 antibodies were designated GWBs and were shown to contain the mRNA decapping enzymes hDcp1 and hLSm4 [7].

In this report we characterized a novel autoantibody system associated with the CDS IIF pattern. CDS-1 sera recognized specific phosphatidylethanolamine (PE) species located at vesicles not related to known markers of the endosome–lysosome pathway, but related partially to GWBs.

Materials and methods

Sera

Among sera depicting CDS IIF pattern obtained from routine screening for anti-nuclear antibodies (ANA), we selected a set that shared a peculiar distribution and size of speckles and labelled them as CDS-1 sera. The CDS-1 IIF pattern was confirmed and titrated by IIF on HEp-2 cells, HeLa cells and mouse peritoneal macrophages. Clinical information was derived from retrospective chart review and interviews with the assistant physicians. A diagnosis of autoimmune rheumatic disease was ascribed when classification criteria were met for systemic lupus erythematosus (SLE) [8], rheumatoid arthritis (RA) [9], polymyositis/dermatomyositis (PM/DM) [10,11], Sjögren's syndrome (SS) [12] and systemic sclerosis (SSc) [13]. Additionally, human anti-cardiolipin antibody-positive sera from 66 patients with primary and secondary anti-phospholipid antibody syndromes were obtained from the out-patient clinic at UNIFESP Medical School Hospital.

Cells

HEp-2 (ATCC, CCl-23) and HeLa (ATCC CCL-2·2) cells were cultured in RPMI-1640 medium (Sigma, St Louis, MO, USA), supplemented with 10% fetal calf serum (FCS) and 2 mM l-glutamine. Mouse resident peritoneal macrophages (kindly provided by Dr Michel Rabinovitch, UNIFESP, São Paulo, Brazil) were obtained from peritoneal washouts of BALB/c female mice, 7–9 weeks old. Cells (2 × 10⁴) suspended in Ca²⁺ and Mg²⁺-free phosphate buffered saline were allowed to attach onto 13 mm-diameter glass coverslips placed in 2 cm² wells and kept in 0·5 ml antibiotic-free Dulbecco's modified Eagle's medium (DMEM) (Sigma) containing 15 mM 4-(2-hydroxyethyl)-1-piperazine-ethanesulphonic acid (HEPES, Sigma), 2 g/l sodium bicarbonate, 1 mM l-glutamine and 5% FCS [14]. Cultures and macrophage isolates were kept at 37°C in 5% CO₂ atmosphere.

IIF

HEp-2 cells, HeLa cells and peritoneal mouse macrophages on coverslips were fixed in 3% paraformaldehyde in phosphate-buffered saline (PBS) and permeabilized with 0·1% Triton X-100 in PBS. Commercially available HEp-2 cells (Virgo/Hemagen, Columbia, MD, USA) were also used for standard IIF. Cells were blocked with 0·25% gelatin in PBS (blocking buffer) and incubated with either human sera diluted 1 : 100 or undiluted affinity-purified human antibodies. Secondary antibody was fluorescein isothiocyanate (FITC)-labelled goat antibody to human IgG (Virgo/Hemagen, Columbia, MD, USA). Cells were imaged by fluorescence confocal scanning laser microscopy on a Bio-Rad 1024-UV system (Bio-Rad, Hercules, CA, USA) attached to a Zeiss Axiovert 100 microscope, using either 40 × (water immersion) or 63 × (oil immersion) PlanApochomatic objectives.

Double immunofluorescence confocal microscopy

After preincubation for 30 min at room temperature (RT) in blocking buffer, HEp-2 cells were incubated with anti-LAMP-2 (clone H4B4, Development Studies Hybridoma Bank, IA, USA), anti-EEA1 (clone 14, Transduction Laboratories, Lexington, KY, USA) or anti-GW182 (clone 4B6, kindly provided by Dr T. Eystathioy, University of Calgary, Canada) [15] mouse monoclonal antibodies (mAb) and in a second step with human anti-CDS-1 serum. Human sera and mAb to EEA1 and GW182 were used at 1 : 100 dilution in blocking buffer and anti-LAMP-2 mAb was used undiluted. Secondary antibodies were FITC-labelled anti-human IgG (Sigma) diluted 1 : 60 and Cy3-labelled anti-mouse IgG (Sigma) or rhodamine-labelled anti-mouse IgG (Sigma) diluted 1 : 80 in blocking buffer. In some experiments nuclei were labelled with 4′,6-diamidino-2-phenylindole (DAPI) (Molecular Probes, OR, USA). Cells were imaged as described above.

Cellular extracts

HEp-2 or HeLa cells were harvested and resuspended in PBS with a mixture of protease inhibitors: 1 µg/ml leupeptin, 1 µg/ml pepstatin, 1 mMphenylmethylsulphonyl fluoride (PMSF) (Calbiochem-Behring Corp., La Jolla, CA, USA) and 2 µg/ml aprotinin (Sigma). For total extract, whole cells were solubilized in Laemmli sample buffer [16], heated at 100°C for 5 min, centrifuged for 10 min at 3000 g and the supernatant kept at −20°C until use. For cytoplasmic extracts, HEp-2 cells were washed three times with 130 mM NaCl, 5 mM KCl, 1·5 mM MgCl₂, and resuspended in hypotonic HEPES buffer (10 mm HEPES pH 7·9, 10 mM KCl, 1·5 mM MgCl₂) with protease inhibitors, as described above. The cells were kept on ice for 30 min and broken by seven strokes of Dounce homogenization with an S-pestle. Subsequently, 0·1 volume of isotonic buffer (300 mM HEPES pH 7·9, 1·4 mM KCl, 30 mM MgCl₂) was added and the cellular lysate was centrifuged at 3000 g for 5 min to pellet nuclei. The supernatant was kept as the cytoplasmic fraction. After determining the protein content, the cell fractions were solubilized in Laemmli sample buffer and brought to 100°C before electrophoresis [17].

Immunoblotting

Cellular extracts were subjected to 10% sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membranes as described [17]. Membranes were blocked in 0·05% Tween-20, 5% non-fat milk in PBS (PBS/T/M) and cut into 5 mm-wide strips, which were exposed individually to human sera and murine mAb diluted 1 : 100 in PBS/T/M. Strips were incubated sequentially with biotinylated goat anti-human IgG (Amersham Pharmacia Biotech, Amersham, UK) diluted 1 : 2000 in 0·5% bovine serum albumin (BSA) in PBS and then with streptavidin–horseradish peroxidase (Amersham Pharmacia Biotech) 1 : 2000 in 0·5% BSA in PBS. Antibody–antigen complexes were detected by autoradiography using the ECL reagent (Amersham Pharmacia Biotech) followed by exposure to radiographic film (Hyperfilm, Amersham Pharmacia Biotech) [17].

Outchterlony double immunodiffusion (DID)

All CDS-1 sera were tested against calf thymus extractable nuclear antigens by DID in 0·4% agarose gels [18]. A prototype anti-SS-A/Ro serum was used to compare the precipitin lines obtained.

In vitro transcription and translation

In vitro transcription and translation was performed as described previously [19]. Purified and linearized plasmid (1 µg each) − EEA1 cDNA [19] or GW182 cDNA [5] − was used as a template for in vitro transcription with T₃ RNA polymerase. RNA transcripts were analysed in 0·8% agarose gel containing 2·2 M formaldehyde. Transcribed RNA (1 µg) was added into a 50 µl-translation reaction containing rabbit reticulocyte lysate, [³⁵S]-methionine (Trans-[³⁵S] label, 70% methionine, 15% cysteine; ICN Biochemicals, Irvine, CA, USA) and RNase block II (Stratagene Inc., La Jolla, CA, USA) as suggested by the manufacturer (Promega Biotec, Madison, WI, USA). Translation was carried out at 30°C for 1 h, followed by SDS-PAGE to monitor translation products. Samples were stored at −80°C.

Immunoprecipitation

Immunoprecipitation of [³⁵S]-labelled in vitro translation products was performed as described [20]. Briefly, 10 µl of human serum and 2–5 µl of in vitro translation products were incubated with protein A-Sepharose beads for 4 h at RT. After washing five times with NET-2 buffer [50 mM Tris-HCl, pH 7·4, 150 mM NaCl, 5 mM ethylenediamine tetraacetic acid (EDTA), 0·5% Nonidet P-40, 0·5% deoxycholic acid, 0·1% SDS, 0·02% sodium azide] the beads were resuspended in Laemmli sample buffer [16]. Samples were analysed by SDS-PAGE and autoradiography.

Indirect immunofluorescence after extraction with organic solvents

HEp-2 cells grown onto glass coverslips were fixed with 3% paraformaldehyde in PBS as above, permeabilized with 0·1% Triton X-100 in PBS for 4 min and extracted with an organic solvent mixture consisting of isopropyl alcohol : hexane : water (55 : 20 : 25, v/v/v) for 15 min at RT. The extraction was repeated twice and cells were then processed for IIF with undiluted anti-LAMP-2 mAb, 1 : 100 anti-EEA1 mAb and a set of autoimmune sera (CDS-1, EEA1, Sm/U1-RNP, native DNA, Scl-70, Golgi, Jo-1 and fibrillarin). As a control the same set of anti-sera and monoclonal antibodies were tested in cells fixed and permeabilized as above but not subjected to organic solvent extraction.

HEp-2 cell total lipid extract

HEp-2 cells grown to subconfluency were washed three times with PBS and detached from plastic bottles with a cell scraper. Lipids were extracted as described by Kates [21], with some modifications. Briefly, the cell pellet was extracted three times with isopropyl alcohol : hexane : water (55 : 20 : 25, v/v/v) for 1 h at RT under vigorous shaking, followed by centrifugation at 3000 g for 10 min at RT. The organic phase was saved and cells were re-extracted with chloroform : methanol (2 : 1, v/v) under the same conditions. All organic phases were combined and concentrated to near dryness in a rotary evaporator (Büchi, Flawil, Switzerland). The lipid extract was resuspended in 5 ml deionized water, sonicated for 5 min and loaded onto a 1 ml-Supelclean™ LC-18 SPE column (Supelco Inc, Bellefonte, PA, USA). After exhaustive washing with deionized water the column was eluted sequentially with 20 ml methanol and 15 ml chloroform : methanol (2 : 1, v/v). The combined eluates were dried under nitrogen, solubilized in 1 ml chloroform : methanol (2 : 1, v/v) and kept at −20°C as the HEp-2 cell total lipid extract.

Lipid standards

Commercial phospholipid standards for 1-[cis-9-octadecenoyl]-2-hexadecanoyl-sn-glycero-3- phosphatidylcholine (C16 : 0/C18 : 1-PC); diphosphatidylglycerol (cardiolipin, C18 : 2-CL); 1-oleoyl-2-sn-glycero-3-phosphate (lyso-phosphatidic acid, C18 : 1-LPA); and 1-hexadecanoyl-2-[cis-9-octadecenoyl]-sn-glycero-3-phosphoethanolamine (C16 : 0/C18 : 1-PE) were obtained from Sigma. From Avanti Polar Lipids (Alabaster, AL, USA) we obtained the following phospholipid standards: 1-myristoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine (lyso-C14 : 0-PE); 1-palmitoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine (lyso-C16 : 0-PE); 1-stearoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine (lyso-C18 : 0-PE); 1-oleoyl-2-hydroxy-sn-glycero-3-phosphoethanolamine (lyso-C18 : 1-PE); 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (C18 : 1/C18 : 1-PE); 1,2-dielaidoyl-sn-glycero-3-phosphoethanolamine (trans-C18 : 1/C18 : 1-PE); 1-stearoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (C18 : 0/18 : 1-PE); 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine (C14 : 0/C14 : 0-PE); 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine (C16 : 0/C16 : 0-PE); 1,2-distearoyl-sn-glycero-3-phosphoethanolamine (C18 : 0/C18 : 0-PE), l-α-phosphatidylserine (C18 : 0/C18 : 1-PS); oleoyl semilysobisphosphatidic acid (S,R isomer) sn-(3-oleoyl-2-hydroxy)glicerol-1phospho-sn-3′-(1′,2′-dioleoyl)-glycerol, ammonium salt (C18 : 1-LBPA); and 1-O-hexadecyl-2-acetoyl-sn-glycero-2-phosphocholine (C16 : 0-PAF), l-α-phosphatidylinositol (PI).

High-performance thin-layer chromatography (HPTLC)

HEp-2 cell total lipid extract and lipid standards were separated on silica-gel HPTLC plates (J.T. Baker, Phillipsburg, NJ, USA) using chloroform : methanol : 0·02% CaCl₂ (90 : 40 : 13·5, v/v/v) as solvent. In order to obtain additional resolution of individual components, HEp-2 cell total lipids were separated by two-dimensional chromatography on silica-gel HPTLC plates. The first dimension was carried out with the solvent chloroform : methanol : 0·02% CaCl₂ (90 : 40 : 13·5, v/v/v) and the second dimension with chloroform : methanol : water (65 : 25 : 4, v/v/v). The lipids were visualized by exposure to iodine vapour for 5 min or by spraying the plate with 30% sulphuric acid, followed by heating at 120°C for 5 min [21].

HPTLC-immunostaining

Immunostaining was performed according to Karasawa et al. [22], with slight modifications. HPTLC plates were dried and dipped for 30 s into 0·1% polyisobutylmethacrylate in hexane. After drying at RT, plates were soaked in 1% BSA in PBS for 1 h and incubated with the appropriate serum diluted 1 : 1500 in 1% BSA in PBS for 1 h at RT in a moist chamber. After washing with PBS, plates were incubated under the same conditions with biotinylated goat anti-human IgG 1 : 2000 in 0·5% BSA in PBS. Plates were washed and incubated with streptavidin–horseradish peroxidase 1 : 2000 in 0·5% BSA in PBS for 1 h. After washing in PBS and in 50 mM carbonate/bicarbonate buffer (pH 9·6), antibody–antigen complexes were detected by autoradiography using the chemiluminescence SignalPlus reagent (Pierce Biotechnology, Rockford, IL, USA) followed by exposure to radiographic Hyperfilm.

Electrospray ionization–mass spectrometry analysis

Lipid standards and samples obtained by scraping off immunostained spots from two-dimensional HPTLC were analysed in a Finnigan LCQ-Duo electrospray ionization-ion trap-mass spectrometer (ESI-MS) (ThermoFinnigan, San Jose, CA, USA). The lipids were dissolved in chloroform : methanol (1 : 1, v/v), containing 0·1% formic acid and 5 mM ammonium acetate (CM : FA : AA), and introduced into the ESI-MS source through a fused-silica capillary (50 µm internal diameter) at a flow rate of 5 µl/min, with the ion source temperature set at 200°C. Mass analysis (ESI-MS) was performed in both negative- and positive-ion modes using a capillary voltage of ∼47 eV. Fragmentation analysis (ESI-MS/MS) was carried out using 30% (1·5 eV) relative collision energy. All equipment parameters were optimized using phospholipid (PE, PC, PS, PA and PI) standards (from Sigma and Avanti Polar Lipids) at 7–70 pmol/µl in CM : FA : AA [23].

Detection of anti-phospholipid antibodies by chemiluminescent enzyme-linked immonosorbent assay (CL-ELISA)

CL-ELISA tests for detection of anti-phospholipid antibodies were carried out as described previously [24], with modifications for sensitivity improvement using chemiluminescent detection [25]. Briefly, microtitre wells (FluoroNunc MaxiSorp™ Plate, NUNC, Roskilde, Denmark) were coated with 50 µl phospholipid standard in 100% methanol at various concentrations (1·75, 0·88, 0·44, 0·22 and 0·11 nmol/well). The solution was evaporated for 2 h at RT and then overnight at 4°C. The wells were blocked with 100 µl 0·1% BSA in PBS for 1 h at 37°C and incubated for 1 h at 37°C with 50 µl of the appropriate serum diluted 1 : 2000 in 0·1% BSA in PBS. Plates were then incubated with biotinylated goat anti-human IgG 1 : 2000 in PBS (50 µl/well) for 30 min at 37°C followed by streptavidin–horseradish peroxidase 1 : 2000 in PBS (50 µl/well) for 30 min at 37°C. Reactions were developed using the ECL™ reagent 1 : 2000 in 50 mM pH 9·6 carbonate/bicarbonate buffer (50 µl/well). Plates were read immediately in a microplate luminometer Fluoroskan FL (Labsystems, Helsinki, Finland), and the results were expressed in relative luminescent units (RLU).

Absorption of anti-PE antibodies

CDS-1 prototype serum (no. 11) diluted 1 : 1000 in 0·5% BSA in PBS was incubated for 1 h with a HPTLC plate (5 × 6 cm) spotted with PE, as described above. Unbound serum was saved and bound antibodies were eluted by incubating the plate with 10 mM citric acid pH 2·5 for 3 min at RT. Unbound serum was incubated again with the HPTLC-PE plate followed by elution of bound antibodies as before. Unprocessed serum no. 11 and unbound serum no. 11 recovered after five absorption rounds were tested by HPTLC-immunostaining and IIF on HEp-2 cells. As a control, an aliquot of serum no. 11 underwent the same absorption protocol with a non-spotted HPTLC plate.

Results

Definition of CDS-1 pattern by indirect immunofluorescence

CDS-1 was characterized as 3–20 bright sharply defined discrete speckles spread throughout the cytoplasm of interphase cells without clustering or any preferential localization (Fig. 1a,d,g). According to this morphological classification, 27 sera were identified as having the typical CDS-1 pattern. The CDS-1 pattern was observed consistently in commercial and in-house HEp-2 cell preparations as well as in HeLa cells and mouse peritoneal macrophages. The CDS-1 sera titre varied from 1 : 320 to 1 : 2560, the majority ranging from 1 : 640 to 1 : 1280. Some sera showed an additional nuclear fine speckled staining (Table 1). The CDS-1 pattern was not observed in a series of anti-cardiolipin antibody-positive sera from 66 patients with the anti-phospholipid antibody syndrome.

Fig. 1.

Dual immunofluorescence confocal microscopy in HEp-2 cells with human CDS-1 sera and mouse monoclonal antibodies to LAMP-2, EEA1 or GW182. Cells were incubated sequentially with human serum diluted 1 : 100, traced with fluorescein isothiocyanate (FITC)-labelled goat antibody to human IgG, and monoclonal antibodies (mAb) traced with Cy3-labelled goat antibody to mouse IgG. (a) CDS-1 serum no. 35. (b) Anti-LAMP-2 mAb. (c) Merged image. (d) CDS-1 prototype serum no. 11. (e) Anti-EEA1 mAb. (f) Merged image. (g) CDS-1 prototype serum no. 11. (h) Anti-GW182 mAb. (i) Merged image, arrows point to immunoco-localization spots. Bars represent 20 µm.

Table 1.

Clinical and serological synopsis of patients with antibodies to CDS-1.

No.	Ageaa	Gender	Clinical presentationbb	CDS-1 titre	Associated IIF pattern	Immunoblotting	Immunodiffusion to anti-ENA
1	47	F	n.a.	1 : 320	None	No reactivity	None
7	52	F	n.a.	1 : 640	NfSp	52 kDa	None
8	n.a.	F	Nephrolithiasis, pyielonephritis, bladder carcinoma, chronic cystitis and uveitis	1 : 320	None	No reactivity	None
11	61	F	SLE	1 : 1280	None	No reactivity	None
21	60	F	DLE and hepatitis	1 : 320	NfSp	52 kDa	SS-A/Ro
26	46	F	Rheumatoid arthritis	1 : 1280	None	No reactivity	None
35	57	F	n.a.	1 : 1280	None	No reactivity	None
36	72	F	n.a.	1 : 640	None	No reactivity	None
38	n.a.	n.a.	n.a.	1 : 640	None	No reactivity	None
43	53	F	Diffuse myalgia and osteoporosis	1 : 320	None	No reactivity	None
45	52	F	Hashimoto's thyrioiditis	1 : 1280	None	No reactivity	None
51	61	F	Alopecia	1 : 1280	None	No reactivity	None
52	n.a.	F	SLE and Hashimoto's tyrioiditis	1 : 1280	None	No reactivity	None
53	n.a.	F	n.a.	1 : 1280	None	No reactivity	None
54	n.a.	F	n.a.	1 : 1280	NfSp	52 kDa	SS-A/Ro
55	n.a.	F	SLE	1 : 2560	NfSp	52 kDa	SS-A/Ro
56	42	F	Leukopenia	1 : 1280	None	No reactivity	None
58	31	F	MCTD	1 : 1280	None	No reactivity	None
60	53	F	Non-specific muscle-skeletal complaints	1 : 640	None	No reactivity	None
61	31	F	Polyarthralgia	1 : 640	None	No reactivity	None
62	68	F	n.a.	1 : 640	None	No reactivity	None
63	34	F	Non-specific muscle-skeletal complaints	1 : 320	NfSp	No reactivity	None
66	43	F	Unilateral chronic uveitis	1 : 2560	None	No reactivity	None
69	59	F	Arthrosis and fibromialgia	1 : 640	None	No reactivity	None

^aNA, not available.

^bMCTD, mixed connective tissue disease.

^cSLE, systemic lupus erythematosus.

^dDLE, discoid lupus erythematosus.

^eNfSp, nuclear fine speckles.

Cytotopographic characterization of domains associated with CDS-1

Using immunofluorescence confocal microscopy, CDS-1 sera depicted no co-localization with LAMP-2 mAb (Fig. 1a–c) or EEA1 mAb (Fig. 1d–f), but showed partial co-localization with GW182 mAb (Fig. 1g–i). Virtually all GW182-rich bodies contained the antigen recognized by anti-CDS-1 sera but only one-third of CDS-1 speckles were recognized by anti-GW182 mAb. In a systematic analysis of 50 sequential cells stained with anti-CDS-1 serum and anti-GW182 mAb, 305 cytoplasmic speckles were detected: 107 (35%) were positive for GW182 and CDS-1 labelling; 191 (62·7%) were positive exclusively for CDS-1 labelling; and 7 (2·3%) were positive solely for GW182 labelling. CDS-1 staining was observed in 107 (94%) of the 114 GW182-positive speckles. Conversely, GW182 staining was observed in 107 (36%) of the 298 CDS-1-positive speckles.

Immunological characterization of CDS-1 autoantigen

No consistent immunoblotting reactivity was achieved with the 27 CDS-1 sera against total cell extracts. Immunoblotting experiments with either HEp-2 cell total extract under native conditions or with cytoplasmic fractions were also negative (data not shown). Additionally representative CDS-1 sera did not immunoprecipitate labelled HeLa cell total extract (three sera), in vitro synthesized GW182 (five sera) and in vitro synthesized EEA1 (one serum) (data not shown). Lipid extraction from HEp-2 cells with organic solvents prior to IIF procedure severely impaired CDS-1 reactivity (Fig. 2a,b) but did not affect the reactivity of human antibodies to EEA1 (Fig. 2c,d), Sm/U1-RNP, native DNA, Scl-70, Golgi, Jo-1 and fibrillarin or of LAMP-2 and EEA1 mAbs (data not shown). Lipid extraction also did not interfere with the associated nuclear fine speckled pattern yielded by some CDS-1 sera. In HPTLC-immunostaining, 24 of the 27 CDS-1 sera reacted with HEp-2 total lipid extract represented as a common band with a relative mobility equivalent to phosphatidylethanolamine (PE) standard and four of them depicted an additional reactivity with a band with relative mobility equivalent to phosphatidylcholine (PC) standard (Fig. 3a). Using a pool of six CDS-1 sera, it was shown that CDS-1 reactivity in HPTLC against total HEp-2 lipid extract had a similar migration pattern as the PE standard (Fig. 3b,c). The three CDS-1 sera with no reactivity in HPTLC showed positive reactivity against GWB recombinant proteins in an addressable laser bead immunoassay. In addition, two of four representative PE-reactive CDS-1 sera also showed some reactivity to GWB recombinant proteins (data not shown).

Fig. 2.

Indirect immunofluorescence of CDS-1 and anti-EEA1 sera on delipidated HEp-2 cells. Sera were diluted 1 : 100. (a, b) CDS-1 serum no. 11. (c, d) Human anti-EEA1 serum. (a, c) Non-delipidated HEp-2 cells (control); (b, d) HEp-2 cells extracted with isopropyl alcohol/hexane/water (delipidated). Antibodies were traced with fluorescein isothiocyanate (FITC)-labelled goat antibody to human IgG. Bars represent 20 µm.

Fig. 3.

Characterization of CDS-1 sera reactivity against total lipid extract of HEp-2 cells by high performance thin-layer chromatography (HPTLC). (a) HPTLC-immunostaining. 1, human anti-EEA1 serum; 2, human anti-cardiolipin serum; 3, human anti-Scl-70 serum; 4 and 5, normal human sera; 6–15, CDS-1 sera (nn. 38, 51, 52, 53, 35, 11, 43, 36, 54 and 61). Phospholipid standards: phosphatidylethanolamine (PE); phosphatidylcholine (PC); phosphatidylserine (PS). (b) HPTLC of total lipid extract of HEp-2 cells and phospholipid standards. HL, HEp-2 cell total lipids; PE, PS and PC, phospholipid standards. Lipids were visualized after exposure to iodine vapour. C, HPTLC-immunostaining. A replica of the plate used in (b) was assayed with a pool of six CDS-1 sera (nos 11, 35, 38, 51, 52 and 53). O, origin.

ESI-MS analysis

Total lipids extracted from HEp-2 cells were separated by two-dimensional HPTLC (Fig. 4, inset). The spot corresponding to CDS-1 reactivity (Fig. 4, inset, delimited by a dashed circle) was scraped off the silica and analysed by ESI-MS and ESI-MS/MS. In negative-ion mode analysis, several singly charged [(M-H)^–] ion species were observed between m/z 478 and 773 (Fig. 4a). ESI-MS/MS analysis of the dominant ion species at m/z 478 gave origin to daughter ions which could be assigned to a lyso-PE species containing C18 : 1 fatty acid (m/z 281) in its composition (Fig. 4b). The second most abundant ion observed at m/z 743 was also subjected to fragmentation and could be assigned as diacyl-PE species, containing two C18 : 1 fatty acids (Fig. 4c). Other minor ion species observed at m/z 506, 536 and 546 were assigned as lyso-PE species containing C20 : 1, C22 : 0 and C24 : 2, respectively. The remaining ion species observed at m/z 745, 771 and 773 were assigned tentatively as diacyl-PE species containing C18 : 0/C18 : 1, C18 : 1/C20 : 1 and C18 : 1/C20 : 0, respectively. Analysis of eluates from the other spots (Fig. 4, inset) was compatible with neutral lipids (NL), PC, PS and lyso-phosphatidylcholine (LPC), respectively (data not shown).

Fig. 4.

ESI-MS (negative-ion mode) analysis of HEp-2 cell phosphatidylethanolamine (PE) fraction reactive with CDS-1 sera. (a) electrospray ionization–mass spectrometry (ESI-MS) spectrum. HEp-2 cell total lipids were separated by two-dimensional chromatography on silica-gel high performance thin-layer chromatography (HPTLC) plate (inset). The first dimension run was performed using chloroform : methanol : 0·02% CaCl₂ (90 : 40 : 13·5, v/v/v), and the second dimension was carried out in chloroform : methanol : water (65 : 25 : 4, v/v/v). The lipids were visualized after exposure to iodine vapour. The encircled area represents the spot scrapped off for ESI-MS analysis. (b and c) Tandem ESI-MS spectra of ions m/z 478·4 and 742·6, respectively. Fragmentation assignments and putative structures for the two PE species are depicted. O, origin; NL, neutral lipids; LPE, lyso-phosphatidylethanolamine; PE, phosphatidylethanolamine; PS, phosphatidylserine; PC, phosphatidylcholine; LPC, lyso-phosphatidylcholine; m/z, mass to charge ratio.

HPTLC-immunostaining and CL-ELISA with an expanded panel of phospholipid standards

In order to confirm and define further the fine specificity of CDS-1 antibodies, a pool of CDS-1 sera (nos 11, 35, 38, 51, 52 and 53) was tested against additional phospholipid standards on HPTLC-immunostaining and CL-ELISA (Fig. 5). A strong reactivity was observed in HPTLC-immunostaining with C16 : 0/C18 : 1-PE and C18 : 1/C18 : 1-PE (Fig. 5b). A weaker reactivity was observed with trans-C18 : 1/C18 : 1-PE and C18 : 0/C18 : 1-PE. There was no or little reactivity with the other phospholipid standards, such as C18 : 0/C18 : 1-PS, C18 : 1-LPA, C16 : 0/C18 : 1-PC, C18 : 2-CL, lyso-C14 : 0-PE, lyso-C16 : 0-PE, lyso-C18 : 0-PE, lyso-C18 : 1-PE, C18 : 1-LBPA and C16 : 0-PAF (Fig. 5b). No HPTLC-immunostaining reactivity against HEp-2 cell lipid extract or phospholipid standards was observed with autoimmune sera to Scl-70, centromere, Jo-1, SS-A/Ro and native DNA (data not shown).

Fig. 5.

Reactivity of CDS-1 sera on high performance thin-layer chromatography (HPTLC)-immunostaining and chemiluminescent enzyme-linked immonosorbent assay (CL-ELISA) using different phospholipid standards. HPTLC was performed using chloroform : methanol : CaCl₂ 0·02% (90 : 60 : 13·5, v/v/v) as solvent. (a) HPTLC. Phospholipid standards (7 nmol each) were visualized with iodine vapour. (b) HPTLC-immunostaining. A replica of the plate used in (a) was assayed with a pool of six CDS-1sera (nn. 11, 35, 38, 51, 52 and 53). (c) CL-ELISA. All phospholipid standards were assayed by CL-ELISA, at 0·88 nmol/well, using the pool of six CDS-1 sera (black bars) and a pool of 10 normal human sera (white bars). O, origin.

All previous and additional phospholipid standards were assayed with a pool of six CDS-1 sera by CL-ELISA. Overall, the results were similar to those obtained by HPTLC-immunostaining; however, a weak but noticeable reactivity was now observed with C18 : 2-CL and C16 : 0/C18 : 1-PC standards (Fig. 5c). Nevertheless, CDS-1 serum pool reactivity was again much stronger with C18 : 1/18 : 1-PE and C16 : 0/C18 : 1-PE standards. A weaker reactivity was observed with trans-C18 : 1/C18 : 1-PE, C18 : 0/C18 : 1-PE and no reactivity was noticed with an extended panel of phospholipids, including PE species with saturated lipid tails (C14 : 0/C14 : 0-PE, C16 : 0/C16 : 0-PE, C18 : 0/C18 : 0-PE), several lyso-PE species, PS, PAF and PI (Fig. 5c).

Removal of IIF CDS-1 reactivity by absorption of CDS-1 sera with PE

In order to confirm that the CDS-1 IIF pattern is caused by anti-PE antibodies we absorbed CDS-1 prototype serum no. 11 with PE as described in Materials and methods and could abolish successfully the CDS-1 fluorescence pattern in IIF (Fig. 6a,b) and anti-PE HPTLC-immunostaining (Fig. 6c,d). Remarkably, sera that also depicted an associated nuclear fine speckled pattern maintained this additional IIF pattern. CDS-1 serum no. 11 absorbed with non-spotted HPTLC plate maintained its original HPTLC-immunostaining profile and characteristic IIF pattern (data not shown).

Fig. 6.

Abolishment of immunofluorescence reactivity and high performance thin-layer chromatography (HPTLC)-immunostaining of CDS-1 serum by absorption of anti-phosphatidylethanolamine (PE) antibodies. CDS-1 serum no. 11 prediluted 1 : 1000 in 0·5% bovine serum albumin (BSA) in phosphate-buffered saline (PBS) was absorbed extensively with PE as described in Material and methods. (a) Indirect immunofluorescence of HEp-2 cells stained with anti-CDS-1 serum before absorption. (b) Indirect immunofluorescence of HEp-2 cells stained with anti-CDS-1 serum after absorption. Bars represent 10 µm. (c) HPTLC-immunostaining of PE standard with anti-CDS-1 serum no. 11 before (lane 1) and after (lane 2) absorption. (d) The same HPTLC plate as in (c) visualized with iodine vapour, depicting the PE band in both lanes. O, origin.

Clinical associations of anti-CDS-1 antibodies

Reliable information was available for 16 of the 24 patients with CDS-1/anti-PE antibodies. Table 1 depicts the demographic, clinical and immunological features. There was a predominance of an array of autoimmune diseases; however, some conditions unrelated to autoimmunity were also represented. Remarkably, none of the patients presented evidence of the anti-phospholipid antibody syndrome. The mean age of patients with CDS-1/anti-PE antibodies was 54 years old, ranging from 31 to 72. All patients were female, except for one whose gender was not known.

Discussion

Cytoplasmic discrete speckles (CDS) observed by IIF have long been referred to in the literature as lysosomes [3]. Recently some reports have been able to associate some of these IIF patterns with Golgi complex [20], early endosomes [26] and discrete cytoplasmic structures designated as GWBs [5]. In the present work we report on the immunological characterization of a novel autoantibody system, designated herein as CDS-1, which recognized discrete structures in the cell cytoplasm. Immunobiochemical experiments showed that CDS-1 sera recognized phospholipid-rich motifs. Based on data obtained with the use of mAb to EEA1, LAMP-2 and GW182 we were able to show that CDS-1 speckles co-localized partially with the recently described GWB but showed no co-localization with components of the endosome–lysosome pathway. The lack of association with the endosome–lysosome pathway is consistent with the unique characteristics of GWBs [5]. Further studies are being held in order to confirm the co-localization of CDS-1 and GWB domains at the ultrastructural level.

Despite the intense and high titre IIF staining, CDS-1 sera presented no consistent reactivity in immunoblotting and immunoprecipitation with HEp-2 and HeLa cell extracts, even when an enriched cytoplasmic fraction was used. The lack of reactivity in immunoblotting could be due to the requirement for conformational epitopes by CDS-1 antibodies that might not have been preserved in the regular immunoblotting protocol. Alternatively, the target antigen could be a lipid molecule rather than a protein. This latter hypothesis was supported by the finding that CDS-1 IIF staining was reduced drastically when lipids were extracted from cells by organic solvents. Moreover, 24 of the 27 CDS-1 sera recognized a common band in lipid extract from HEp-2 cells separated by HPTLC. The ESI-MS analysis of the lipid component co-migrating with PE and recognized strongly by CDS-1 sera showed the presence of major lyso-PE species (mainly C18 : 1-lyso-PE) and minor diacyl-PE species (mainly C18 : 1/C18 : 1). It is worthwhile to point out that the major lyso-PE species might have been originated by artefactual release of fatty acids from diacyl-PEs during lipid extraction procedures and/or ESI-MS analysis. In order to explore further the initial ESI-MS results, different phospholipid standards were separated by HPTLC and the reactivity of CDS-1 sera towards diacyl-PE was confirmed by immunostaining. Interestingly, CDS-1 sera reacted strongly with diacyl-PE species containing one or more unsaturated (C18 : 1) fatty acid chain in its composition. In contrast, CDS-1 sera showed no reactivity towards lyso-PE species, suggesting that at least one unsaturated fatty acid chain is crucial for antigen recognition, possibly by interfering with the final conformation of the relevant epitope [27]. It should be noted that the phospholipid head was also fundamental for the antigen recognition process, as PS and PC species containing one C18 : 1 fatty acid chain were weakly or not recognized by the CDS-1 serum pool. These findings were confirmed in the CL-ELISA system. Interestingly, four of the 24 anti-PE CDS-1 sera also recognized PC in the HPTLC system. As for the three anti-PE-negative CDS-1 sera, reactivity to GWB recombinant proteins GW182, GWN2, GWS2 and GW3 was observed in an addressable laser bead immunoassay [28]. This reactivity was also observed in two of four anti-PE-positive CDS-1 sera (data not shown). In view of the extensive co-localization of CDS-1 and GWB domains it is not unexpected that some CDS-1 sera with autoantibodies to diacyl-PE also recognize other molecular species in GWB. It has been documented extensively in the literature that the autoimmune response is frequently comprehensive to different epitopes and different molecules in the supramolecular complex or cellular domain targeted by autoantibodies [1]. Epitope spreading within GWB and CDS-1 domains is under current investigation in the laboratory and will be explored in a future publication.

In dual IIF confocal microscopy, CDS-1 antibodies did not co-localize with anti-LAMP-2 or with anti-EEA1 mAb, suggesting that CDS-1 antibodies target cytoplasmic structures not belonging to the endocytic–exocytic pathway. On the other hand, CDS-1 autoantibodies showed partial co-localization with mAb to GW182, a marker of a distinct set of cytoplasmic structures, the GW bodies. In a recent study, only four of 18 (22%) human sera staining GWB structures bound recombinant GW182 [28], suggesting an autoantigenic multiplicity for this cytoplasmic organelle. Due to the possibility of antigenic spreading, one could envision that in addition to anti-PE reactivity, some CDS-1 sera might target other GWB antigens. In fact, two anti-PE-positive CDS-1 sera presented moderate reactivity to GW proteins GW-182, GW-2 and GW-3 in an addressable laser bead immunoassay [28] (data not shown). It will be important to confirm these preliminary data and to test the reactivity with other GWB-related proteins, such as hLSm4 or hDcp1 [7]. It is relevant that virtually all GWB speckles were also reactive with anti-CDS-1/PE antibodies but only one-third of the CDS-1 speckles were GW-182-positive. Taken together these findings suggest that CDS-1 sera recognize PE moieties associated with cytoplasmic bodies, some of which represent GW bodies. The nature of the remaining PE-positive bodies is still to be determined. It is conceivable that PE-positive domains and GWB may represent different metabolic stages of a similar set of cytoplasmic domains in which there is a spectral variation in the content of PE.

Anti-phospholipid autoantibodies have been extensively documented in the so-called anti-phospholipid antibody syndrome (APS). Although cardiolipin is the best-known target, anti-PC, anti-PS and anti-PE antibodies are also detected in some patients with APS [29]. Phospholipids constitute the backbone of all cellular membranes, including the cytoplasmic membrane. It is evident that the various anti-phospholipid antibodies recognize restricted sets of epitopes not available widely because the vast majority of APS sera with anti-phospholipid reactivity in ELISA depict no specific IIF staining pattern. The observation that antibodies to such an antigen as ubiquitous as PE depicted a rather discrete IIF staining pattern (CDS-1) finds resonance with the previous report on autoantibodies to lysobisphosphatidic acid (LBPA) that also display a cytoplasmic discrete speckled IIF pattern [30]. It is possible that the peculiar epitopes recognized by CDS-1 sera are available in restricted sets of cytoplasmic domains. Another possibility is that the PE content is high enough to yield visible immunofluorescence in only a restricted set of cytoplasmic domains. Evidence is accumulating that some lipids are not distributed randomly in endosomal membranes along these recycling and degradation routes, contributing to the notion that endosomes contain a mosaic of structural and functional membrane domains [30,31]. A similar phenomenon occurs to cardiolipin, which is restricted to the inner membrane of mitochondria [29]. In fact, it has been shown previously that the distribution of PE among cellular membranes is heterogeneous [32]. It is also relevant that binding of CDS-1 antibodies to PE in HPTLC immunostaining and in CL-ELISA was shown to be highly dependent on the oleic chain of the phospholipid. Finally, it should be emphasized that the conformation of phospholipids may account for their antigenicity. For example, in the mid-1980s, Rauch and Janoff [33] were the first to show that lupus anti-coagulants were neutralized by hexagonal PE, but not by lamellar phase PE. Thus, one could hypothesize that the discrete IIF pattern observed with CDS-1 antibodies might be due to a highly selective representation of certain PE oleic chains in specific conformational arrangements across several cytoplasmic domains.

It is interesting that none of CDS-1/diacyl-PE sera depicted significant reactivity to cardiolipin in HPTLC, CL-ELISA and in the standard ELISA assay. In addition, careful examination of a series of anti-cardiolipin-positive sera showed no CDS staining on indirect immunofluorescence on HEp-2 cells. This might indicate that the CDS-1 relevant epitopes are expressed in a context devoid of this phospholipid so that epitope spreading would not occur. On the other hand, four of the 24 anti-PE-positive CDS-1 sera also recognized PC in the HPTLC system, suggesting that the epitope spreading mechanism might be relevant for these two antigenic systems.

The clinical data obtained in the present report should be considered preliminary, due to the design of the clinical information collection. Although the study group was small and non-random in design, it appears that circulating autoantibodies to CDS-1/diacyl-PE occur more frequently in female than in male patients. Anti-CDS-1/diacyl-PE antibodies were associated with a variety of clinical conditions, including SLE and other autoimmune diseases, as well as some illnesses with no apparent clinical autoimmune features. Interestingly, no patient with anti-CDS-1/diacyl-PE antibodies presented with manifestations characteristic of the anti-phospholipid syndrome.

In conclusion, the present study demonstrated a novel autoantibody system associated with a particular cytoplasmic discrete speckled IIF pattern. The molecular and topographic nature of the cognate antigens was characterized. These autoantibodies might be instrumental for further elucidation of the nature of the recently reported GWB and related cytoplasmic organelles. Preliminary clinical associations were pointed out but further studies are required to establish the clinical significance of this novel human autoantibody specificity.