J Biol Chem, Vol. 274, Issue 36, 25193-25196, September 3, 1999

COMMUNICATION
Phospholipid Membranes Form Specific Nonbilayer Molecular Arrangements That Are Antigenic^*

Leopoldo Aguilar, Guadalupe Ortega-Pierres^§, Begoa Campos^¶, Rocio Fonseca^§, Miguel Ibáñez, Carlos Wong, Norberto Farfán, Jorge M. Naciff^¶, Marcia A. Kaetzel^¶, John R. Dedman^¶, and Isabel Baeza^**

From the  Department of Biochemistry, National School of Biological Science, Instituto Politécnico Nacional, Apartado Postal 4-897, Admon. 4, México City 06401, México, the ^§ Department of Genetic and Molecular Biology and the  Department of Chemistry, Center for Research and Advanced Studies, México City 07000, México, and the ^¶ Department of Molecular and Cellular Physiology, University of Cincinnati, Ohio 45267

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Hexagonal phase (H_II)-preferring lipids such as phosphatidate, cardiolipin, and phosphatidylserine form nonbilayer molecular arrangements in lipid bilayers. While their presence in biological membranes has not been established, in vitro studies suggest that alterations in membrane properties modify their function. In this study, antiphospholipid monoclonal antibodies were developed against nonbilayer structures. One of the monoclonal antibodies identifies nonplanar surfaces in liposomes and in membranes of cultured cells. These results are the first evidence that natural membranes maintain a fragile balance between bilayer and nonbilayer lipid arrangements. Therefore, these antibodies can be used to evaluate the role of H_II-preferring lipids in the modulation of membrane activities. Our studies demonstrated that nonplanar surfaces are highly immunogenic. Although these structures are normally transient, their formation can be stabilized by temperature variations, drugs, antibiotics, apolar peptides, and divalent cations. Our studies demonstrated that abnormal exposure of nonbilayer arrangements may induce autoimmune responses as found in the antiphospholipid syndrome.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

The lamellar or bilayer phase is the most common molecular association adopted by phospholipids and glycolipids in the matrix of cell membranes. However, membrane lipids are in dynamic transition, 30-40% can adopt nonbilayer arrangements (1-5). In isolated form, some phospholipids form hexagonally packed cylinders of the hexagonal II (H_II)¹ tubular phase. The H_II-preferring lipids can also form intermediate structures of the H_II phase, or nonbilayer arrangements, in lipid bilayers of experimental model systems, such as liposomes (2-4). Electron microscopy studies demonstrate that these nonbilayer structures appear as protuberances on the surface of liposomes (2, 3, 6). These "rough" surfaces are transient and may represent a physical intermediate of natural membranes. Their formation can be triggered by temperature variations, drugs, antibiotics, apolar peptides, and divalent cations (2, 3, 6, 7). Nonbilayer lipids may be involved in specific cellular activities, such as membrane fission-fusion processes (2-4, 6, 7), organization of tight junctions between mammalian cells (8), and the specific activation of several membrane enzymes (9). Nonbilayer structures have been shown in model systems by physical methods (2-4, 6, 7, 10-12); their presence in biological membranes has not been confirmed (10, 13), however, they appear to be essential for the viability of Escherichia coli cells (14, 15).

In this study, antiphospholipid monoclonal antibodies were developed against nonbilayer phospholipid arrangements produced on liposomes. These antibodies also identify nonbilayer structures in membranes of cultured cells. These results are the first evidence that nonbilayer lipid structures are present in cells and may be involved in the production of antiphospholipid autoantibodies which can participate in the development of the human antiphospholipid syndrome (16).

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Preparation and Characterization of Liposomes-- Liposomes prepared from specific lipids in TBS buffer (10 mM Tris-HCl, 1 mM NaCl, pH 7) (10) were characterized by freeze-fracture electron microscopy (6) and by their ³¹P nuclear magnetic resonance (NMR) spectra (6, 12).

Immunization Scheme to Obtain Polyclonal Antiphospholipid Antibodies against Nonbilayer Phospholipid Arrangements-- Polyclonal antibodies were obtained as described previously (10) using egg yolk L--phosphatidylcholine/egg yolk L--phosphatidic acid (PC/PA) 2:1 (mole ratio) liposomes in 5 mM Mn²⁺ as antigen.

Preparation of Antiphospholipid Monoclonal Antibodies against Nonbilayer Phospholipid Arrangements (APmAb)-- Hybridomas were obtained by fusing myeloma P3X63Ag8U.1 with spleen cells isolated from mice immunized with PC/PA liposomes in 5 mM Mn²⁺. Supernatants were screened for APmAb by a liposomal ELISA method and flow cytometry (10, 17). Individual clones were obtained by limited dilution (17). Clones producing APmAb were expanded by in vitro culture and by inoculation in BALB/C mice to obtain ascites fluid. APmAb was detected by a liposomal ELISA method and flow cytometry.

Liposomal ELISA-- ELISA wells were coated with liposomes (0.1 µmol/100 µl of TBS) alone or treated with 3 mM Ca²⁺ or 5 mM Mn²⁺. Wells blocked with 0.4% gelatin in TBS alone or containing the appropriate cation were incubated with inactivated immune mice sera or APmAb diluted in the correspondent TBS buffer. Peroxidase-conjugated goat anti-mouse polyvalent or anti-mouse IgM antibodies were used. Bound antibodies were detected by reaction with o-phenylenediamine solution.

Liposomal Flow Cytometric Analysis-- Liposomes (0.1 µmol/100 µl of TBS) alone or treated with 3 mM Ca²⁺ or 5 mM Mn²⁺ were incubated at 37 °C for 1 h in the dark with FITC-H308-APmAb at 1:10 dilution in TBS alone or containing 3 mM Ca²⁺ or 5 mM Mn²⁺. Liposomes were washed by sedimentation at 200,000 × g for 50 min at 18 °C with the appropriate TBS buffer. Flow cytometry was accomplished using a Becton Dickinson FACSCalibur Flow Cytometer. Ten-thousand liposomes were analyzed using CELLQuest software at FL1 compensation of 0.8% and a detector compensation threshold FSC-H of 52 V, with the detectors: FSC of E00, SSC of 401 V, and FL1 of 748 V. Results are reported as relative fluorescence (FL1) and relative forward (FSC) and side scatter (SSC) light histograms in logarithmic mode (18). In a similar way as was described for cells (18), the diffraction of the laser beam (FSC) is proportional to the liposome surface area and/or liposomal aggregation, while refraction plus reflection of the beam (SSC) are proportional to the complexity of liposomal bilayers (10). As a negative control an irrelevant monoclonal antibody of IgM isotype directed to a surface protein of Trichinella spiralis (Ts-mAb) was used (19).

Affinity Purification of APmAb from Ascites Fluid-- PC/PA 2:1 liposomes in 5 mM Mn²⁺ were incubated with ascites fluid in presence of 5 mM Mn²⁺ for 30 min at 37 °C. Mixture was centrifuged at 200,000 × g and the pellet washed with TBS plus 5 mM Mn²⁺. APmAb was released from liposomes by addition of TBS containing 5 mM EDTA. Purity of APmAb was analyzed by 12% SDS-polyacrylamide gel electrophoresis followed by Western blot analysis.

Fluorescence Resonance Energy Transfer Assay (FRET)-- Relative fluorescence of PC/1-acyl-2-(12[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]dodecanoyl) (NBD)-PA 2:1 liposomes was measured by excitation at 465 nm and emission at 530 nm using a Perkin-Elmer LS-5B spectrofluorometer. Quenching percentage was calculated by the equation, (F_o  F)/F_o × 100 (20), where F_o, is fluorescence intensity of liposomes in 3 mM Ca²⁺ or 5 mM Mn⁺² prior to addition of APmAb, and F is fluorescence following addition of APmAb. As a negative control Ts-mAb was used (19).

Immunolocalization of Nonbilayer Structures in Mammalian Cell Membranes-- HeLa, C5337, C33, and MCF-7 cells were analyzed as described in Naciff et al. (21).

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Detection of Nonbilayer Phospholipid Structures on Liposomes-- Liposomes containing the H_II-preferring lipid PA were unilamellar in TBS buffer, with "smooth" surfaces, and varied from 100 to 400 nm in size (Fig. 1A). Ca²⁺ ions (3 mM) induced nonbilayer structures (33 Å thick), which form a complex network of interconnected strings. These structures show fusogenic properties by forming considerably larger liposomes up to 1 µm of radius. A small portion of a fused liposome is shown in Fig. 1A. Mn²⁺ (5 mM) induce nonbilayer structures (37 Å thick) which are organized in quadrangular and pentagonal strings that did not modify the size of liposomes (Fig. 1A). Both cations also induced nonlamellar structures in liposomes containing the H_II-preferring lipids bovine heart cardiolipin (CL) or bovine brain L--phosphatidylserine (PS), which were similar to those described for PC/PA liposomes. In contrast, PC and egg yolk L--phosphatidyl-DL-glycerol (PG)/cholesterol (Chol) 1:1 liposomes, without H_II-preferring lipids, displayed smooth surfaces either alone or when they were incubated with the cations (data not shown). Liposomes treated with Ca²⁺ or Mn²⁺ remained nonaggregated, since it was not necessary to add EDTA to visualize them by freeze-fracture as individual entities (6).

View larger version (39K): [in this window] [in a new window]   Fig. 1.   Detection of nonbilayer lipid structures in liposomes. A, PC/PA 2:1 liposomes (9 µmol) incubated at 37 °C for 30 min with TBS buffer alone or containing 3 mM CaCl2 or 5 mM MnCl2 were used for freeze-fracture electron microscopy. The shadow direction is indicated by the large arrow in TBS panel. The solid line represents 200 nm in all micrographs. Small arrows show nonbilayer structures. B, 31P NMR spectra were obtained at 37 °C using PC/PA 2:1, PC/CL 2:1, or PC liposomes (70 µmol) incubated alone or with the cations as described for freeze-fracture. The same liposome preparation were used for these determinations and for immunological assays.

After the addition of Ca²⁺ (3 mM) to liposomes containing PA or CL, the ³¹P NMR spectrum became sharper than those without cations and had a slight shift to a lower field (Fig. 1B), which indicate isotropic motion of PA or CL in the nonbilayer structures (2-4, 6, 11). In the presence of Mn²⁺ (5 mM) peaks were not evident (Fig. 1B). Mn²⁺ induces the formation of nonbilayer structures. Mn²⁺ penetrates the vesicle exerting paramagnetic effects on phospholipids from both the outer and inner liposomal bilayers, which completely broadens the NMR spectra (6, 12). Similar changes in NMR spectra of all liposomes were obtained with 0.5 mM Ca²⁺ or Mn²⁺. PC liposomes did not show spectral changes when they were incubated with Ca²⁺ (Fig. 1B). However, in the presence of Mn²⁺ (3 mM) the spectrum signal intensity decreased. This change is due to the Mn²⁺ paramagnetic effect and was exerted only on lipids from outer liposomal bilayers (Fig. 1B); similar results were obtained with PG/Chol 1:1 liposomes.

Freeze-fracture and NMR studies clearly showed the nonbilayer structures on liposomes. Both methods demonstrated that only liposomes containing H_II-preferring lipids produce nonbilayer structures.

Antiphospholipid Antibody Specificity-- Antibodies were evaluated by an ELISA method using PC/PA 2:1 liposomes as the antigen. Both polyclonal and hybridoma supernatant containing APmAb bind to nonbilayer structures. The highest binding was attained with APmAb clone H308 and Mn²⁺-induced nonbilayer structures (Fig. 2A). Polyclonal antibodies, but not H308-APmAb, also recognized the smooth liposomes incubated with TBS buffer. The irrelevant monoclonal antibody Ts-mAb did not show reaction with smooth liposomes or bearing nonbilayer structures (Fig. 2A). Using this method, 42 hybridomas producing APmAb were selected. The isotype of all APmAb obtained was IgM as determined by ELISA. Since the routine method to clinically measure antiphospholipid autoantibodies is an ELISA assay that uses a coated plate with lipids dissolved in ethanol (22), the binding of H308-APmAb to purified PC, PA, or CL was evaluated. Our data showed that H308-APmAb did not interact with the nonliposomal lipid-coated plates. In contrast, polyclonal antibodies bind to these lipid-coated surfaces (Fig. 2B). These results strongly suggest that H308-APmAb interacts with nonplanar lipid surfaces.

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Detection of antiphospholipid antibodies by ELISA. A, PC/PA 2:1 liposomes incubated with TBS buffer alone or containing 5 mM MnCl2 or 3 mM CaCl2 were used as antigens. Liposomes were tested with Ts-mAb, polyclonal antibodies from immune BALB/c mouse sera at 1:50 dilution, and H308 hybridoma supernatant containing APmAb at 1:10 dilution. B, purified PC, PA, or CL was used as antigen. Lipids were tested with polyclonal antibodies from immune BALB/c mouse sera or H308 hybridoma supernatant containing APmAb. Absorbance values were determined in triplicate and represent the means ± S.D.

The fact that H308-APmAb specifically recognizes nonbilayer structures on liposomes, an antigen similar to biological membranes, and not lipid-coated surfaces suggests that antibodies similar to H308-APmAb could be produced in human autoimmune disorders such as the antiphospholipid syndrome.

Binding of H308-APmAb to Nonbilayer Structures-- H308-APmAb was affinity-purified directly from ascites by taking advantage of the property of H308-APmAb to bind to liposomes bearing Mn²⁺-induced nonbilayer structures. All H308-APmAb bound to liposomes was specifically eluted by chelation of Mn²⁺ (Fig. 3C). A considerable portion of purified H308-APmAb was not glycosylated (Fig. 3C).

View larger version (38K): [in this window] [in a new window]   Fig. 3.   Flow cytometric and FRET analysis of the reactivity of H308-APmAb with nonbilayer structures on liposomes. A, liposomes containing the HII-preferring lipids PA or CL and those made from PC, DPPC, and DPPA were used as antigens. PC/PA 2:1, PC/CL 2:1, and PC liposomes alone or treated with 3 mM CaCl2 or DPPC/PC/DPPA 1.2:0.8:1 liposomes treated with 5 mM BaCl2 or 0.05 mM LaCl3 were tested with affinity-purified H308-APmAb. Changes in liposomal fluorescence, bilayer complexity (SSC), and liposomal aggregation (FSC) due to the incubation of liposomes with the cations and/or FITC H308-APmAb or FITC Ts-mAb were evaluated. Controls represent both liposomes alone or incubated with Ca2+ or with Ca2+ plus FITC-Ts-mAb. One experiment representative of three is shown. B, fluorescent quenching was induced by binding of H308-APmAb to PC/NBD-PA or PC/NBD-PS 2:1 liposomes in 3 mM CaCl2 or 5 mM MnCl2. Measurements were obtained using a 95% confidence level (p = 0.05). C, affinity purification of APmAb. Ascites fluid was mixed with PC/PA 2:1 liposomes in 5 mM MnCl2. SDS-polyacrylamide gel electrophoresis and immunoblot show purified APmAb. The IgM heavy chain (HC) is glycosylated (HC-glycos) and nonglycosylated (HC), and both forms were identified with peroxidase-labeled goat-anti-mouse IgM heavy chain. LC, light chain.

H308-APmAb did not show any interaction with smooth PC/PA liposomes, since liposomal fluorescence (Fig. 3A, panel a) was similar to that of liposomes alone or Ca²⁺-treated (Fig. 3A, panel m). In contrast, the 60-fold increase in fluorescence, compared with that of smooth liposomes (with a difference among populations in a logarithmic scale (D) = 0.9, at p < 0.001), clearly showed the interaction of H308-APmAb with Ca²⁺-induced nonbilayer structures (Fig. 3A, panel a). In addition, SSC values, which show the complexity of liposomal bilayers, i.e. the presence of nonbilayer arrangements, indicated that the pattern of Ca²⁺-induced structures was different after the immunoreaction, compared with the pattern of these structures on Ca²⁺-treated liposomes (Figs. 3A, panels b and n). These profiles reflect the dynamic feature of nonbilayer arrangements. Liposomal aggregation (FSC), which can produce nonspecific increases in fluorescence, was not evident, since the FSC values, after the antibody binding, were similar to those of liposomes bearing nonbilayers structures without H308-APmAb (Figs. 3A, panels c and o). Liposomal fluorescence, SSC, and FSC values in the presence of affinity-purified Ts-mAb were similar to that of Ca²⁺-treated liposomes, demonstrating that Ts-mAb does not bind vesicles. Analysis of liposomes containing PA incubated with Ca²⁺ and Ts-mAb is also shown in Fig. 3A, panels m to o. H308-APmAb also binds to liposomes bearing Ca²⁺-induced nonlamellar structures formed by PS or CL. Analysis of liposomes containing CL is shown in Fig. 3A, panels d to f. This reaction, although positive, was clearly distinct from the reaction with liposomes containing PA (Fig. 3A, panels b and e). H308-APmAb did not show interaction with PC liposomes alone or incubated with Ca²⁺ (Fig. 3A, panel g). Furthermore, values of FSC and SSC did not show liposomal aggregation or the presence of nonbilayer structures (Fig. 3A, panels h and i). Similar results were obtained with PG/Chol 1:1 liposomes. In addition, similar binding of H308-APmAb was obtained when all liposomes were incubated with 0.5 mM Ca²⁺. These results support the fact that only liposomes containing nonbilayer structures have an immunoreaction with H308-APmAb.

An alternative conclusion is that the antibody recognizes the lipid-divalent cation complex and/or the resultant reduction in the liposomal charge in the presence of divalent cations. These possibilities were eliminated by the use of "rigid" liposomes. Liposomes made from dipalmitoyl-L--phosphatidylcholine (DPPC)/PC/dipalmitoyl-L--phosphatidate (DPPA) 1.2:0.8:1 or DPPC/PC/E. coli CL 1.2:0.8:1 have a higher content of saturated fatty acids in their DPPC, DPPA, or E. coli CL (23). Consequently the bilayers of these liposomes are more rigid than the bilayers of the liposomes containing similar lipids isolated from egg yolk or bovine heart such as those from Fig. 3A (panels a-f and m-o). Divalent (Ba²⁺ or Ca²⁺) or trivalent (La³⁺) ions can interact with rigid liposomes and modify their surface charge, but these interactions are not associated with the formation of nonbilayer structures as is shown for DPPC/PC/DPPA liposomes (Fig. 3A, panels k and l). These results were corroborated by NMR spectroscopy (data not shown). H308-APmAb did not show interaction with rigid liposomes incubated with Ba²⁺ or La³⁺ (Fig. 3A, panels j-l). These results confirm that the H308-APmAb recognition of the anionic lipid-metal ion complex is only associated with the transition of lipids to nonbilayer structures.

The binding of H308-APmAb was also analyzed by the quenching of NBD-chromophore induced by clustering of fluorescent NBD phospholipids upon H308-APmAb-membrane binding. Affinity-purified H308-APmAb bound to liposomes containing NBD-PA or NBD-PS, while Ts-mAb did not present any binding (Fig. 3B). These results corroborate that H308-APmAb specifically binds to Ca²⁺ or Mn²⁺-induced nonbilayer structures.

Liposomal ELISA, flow cytometry, and FRET demonstrated that the specificity of H308-APmAb is against a nonbilayer structure formed by the H_II-preferring anionic phospholipids PA, CL, or PS. These data suggest that the specificity of H308-APmAb is not against a chemical structure but against a specific lipid molecular arrangement.

Immunolocalization of Nonbilayer Structures in Mammalian Cell Membranes-- In mammalian cells, negatively charged surfaces are formed by the exposure of anionic phospholipids on the outer leaflet of plasma membranes. In the present study, it has been demonstrated that the anionic phospholipids PA, CL, and PS form antigenic nonbilayer structures recognized by our APmAb. Thus, we analyzed whether nonbilayer structures are formed on the membranes of mammalian cells. Cell lines HeLa, C5337, C33, and MCF-7 were directly immunostained with a Cy3-conjugated H308-APmAb. Immunostaining of cells showed that H308-APmAb recognizes cellular membranes possible in apoptosis and necrosis (Fig. 4A). This finding indicates that nonbilayer phospholipid arrangements occur in these biological membranes. These findings support the studies that demonstrate a strong correlation between nonbilayer arrangements and the viability of E. coli (14, 15).

View larger version (143K): [in this window] [in a new window]   Fig. 4.   Direct immunostaining of mammalian cells. HeLa cells were subjected to a direct immunofluorescence using Cy3-H308-APmAb (A) and Cy3-Ts-mAb (C). B and D show phase contrast of the same fields. Arrows show cells immunostaining.

These results indicate that cells maintain a fragil balance between bilayer and nonbilayer lipid arrangements during the dynamic events of membrane function. The presence of these unique phospholipid arrangements suggests their involvement in membrane functions. Therefore, this monoclonal antibody can be used to evaluate the role of H_II-preferring lipids in the modulation of physiological and pathological membrane activities. These unique antinonbilayer antibodies could be different antibodies to those currently detected in clinical assays using lipid-coated ELISA plates, including anti-cardiolipin and anti-phosphatidylserine antibodies. It has been reported that lupus anticoagulant antibodies specifically recognize the H_II tubular phase of phosphatidylethanolamine but not the bilayer arrangements of this lipid (13, 24). However, it is difficult to reconcile membrane features with extensive areas of the lipid H_II phase. Instead, we propose that intermediate structures of the H_II phase in lipid bilayers, such as those analyzed here, are more compatible with the structure and the modulation of physiological and pathological membrane activities. Since nonbilayer structures are immunogenic, their abnormal exposure may induce autoimmune responses as found in patients with primary antiphospholipid syndrome, systemic lupus erythematosus, stroke, and cancer.

	ACKNOWLEDGEMENTS

We thank Dr. R. Mondragón and Dr. S. González from the Electron Microscopy Unit, Center for Research and Advanced Studies, México for electron microscopy studies.

	FOOTNOTES

^* This work was supported in part by Instituto Politécnico Nacional Grant DEPI16.19 and Consejo Nacional de Ciencia y Tecnologia-México Grant 1157P-N9507, American Heart Association Grant 9930171 (to B. C.), American Heart Association Ohio Valley Affiliate Grant 9806236 (to B.C.), and the National Institutes of Health Grant DK46433 (to J. R. D).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^** To whom correspondence should be addressed. Tel.: 52-5-729-6000 (ext. 62326); Fax: 52-5-396-3503; E-mail: ibaeza@vmredipn.ipn.mx.

	ABBREVIATIONS

The abbreviations used are: H_II, hexagonal II; PC, L--phosphatidylcholine; PA, L--phosphatidic acid; APmAb, antiphospholipid monoclonal antibodies against nonbilayer phospholipid arrangements; ELISA, enzyme-linked immunosorbent assay; FITC, fluorescein isothiocyanate; FSC, forward scatter light; SSC, side scatter light; FL1, relative fluorescence; FRET, fluorescence resonance energy transfer assay; NBD, 1-acyl-2-(12[(7-nitro-2,1,3-benzoxadiazol-4-yl)amino]dodecanoyl); CL, cardiolipin; PS, L--phosphatidylserine; PG, L--phosphatidyl-DL-glycerol; Chol, cholesterol; DPPC, dipalmitoyl-L--phosphatidylcholine; DPPA, dipalmitoyl-L--phosphatidate.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES