Michel
Prudent
a,
Manuel A.
Méndez
a,
Daniel F.
Jana
b,
Clémence
Corminboeuf
b and
Hubert H.
Girault
*a
aLaboratoire d’Electrochimie Physique et Analytique, Ecole Polytechnique Fédérale de Lausanne, Station 6, CH-1015 Lausanne, Switzerland. E-mail: hubert.girault@epfl.ch
bLaboratory for Computational Molecular Design, Ecole Polytechnique Fédérale de Lausanne, Batochime, CH-1015 Lausanne, Switzerland
First published on 20th May 2010
Single metal ion–phospholipid complexes are observed in biphasic electrospray ionization mass spectrometry (BESI-MS) using a dual-channel microsprayer. Such a microsprayer makes it possible to put into contact two immiscible liquids within the Taylor cone. Thus, L-α-dipalmitoyl phosphatidylcholine (DPPC) dissolved in 1,2-dichloroethane (DCE) reacts with aqueous metal cations (M = Na+, K+, Ca2+, Cu2+, La3+) yielding the formation of [M-DPPCn]z+ complexes. The number of phospholipid molecules ranges from 1 to 4 for monovalent ions, to 8 for divalent and to more than 10 for trivalent ions respectively. The large number of ligands observed involves the formation of solvent free single ion–phospholipid complexes.
The interaction of monovalent ions with lipids is generally accepted to be rather weak, although non-negligible. As a matter of fact, alkali ions can induce phase transitions as it was observed over multilamellar vesicles.2 Analogously, divalent metal cations, such as calcium, essential to a large number of life processes, effectively interact with cell membranes, modifying their conformation, structure and/or stability, and playing a key role in membrane fusion processes. The interactions between calcium and lipidic bilayers have been thoroughly investigated over the years.3–5 Experimental evidence obtained by multiple techniques (NMR,6 atomic force microscopy,7 X-ray diffraction,8 scanning electrochemical microscopy,9 fluorescence,10 light scattering measurements,8etc.) strongly suggests the formation of calcium–phospholipid complexes6 whose importance has lead to the introduction of calcium within the minimal fusion machinery.8 Indeed, calcium ion bridges generated by the local dehydration of phospholipid head groups and the calcium ions themselves, ultimately leads to the membrane fusion as it has been corroborated by classical molecular dynamics (MD).11 Additional classical MD simulations also predicted the sequential multistep binding and coordination of Ca2+ cations by three or four lipid carbonyl oxygens, supporting this idea.4 Copper is also known to participate in lipid metabolism and may react at the cell surface.12,13
Other recent theoretical studies (MD) have shown that interfacial charge differences stemming from electrostatic interactions in both cell membrane leaflets can also be responsible in a great extent for the creation of intense electric fields across the lipidic bilayer.3
In the field of electrochemistry at polarized liquid–liquid interfaces, the presence of lipidic monolayers adsorbed at the interface between two immiscible electrolyte solutions (ITIES) facilitates the transfer of ions from the aqueous to the organic phase. It indicates that phospholipids have ionophoric properties forming complexes in the organic phase.14 Analogously, Monzón and Yudi have shown that alkali metal ions like Li+ increase lipid organization inducing the interfacial packing of monolayers.15
In a more general sense, the analysis of the entire lipidic content of a biological system (i.e. the lipidome) has given rise to a new emerging branch of metabolomics denominated lipidomics.16–18 Among the techniques most commonly employed to analyze the vast number of compounds present in a biological sample, mass spectrometry (MS) has proven to be a very versatile and powerful technique. Owing to its soft ionization process, electrospray ionization (ESI) has even allowed the transfer of surfactant micelles into the gas phase conserving their structures.19 Nevertheless, there are only a few examples in the literature reporting the ion–lipid interactions in ESI-MS with an emphasis on the fragmentation process.20–24
The major difficulties in studying lipid–ion interactions stem from the low solubility of lipids in water and the high hydrophilic character of small inorganic ions. Both species are normally found in different and immiscible solvents unless prior formation of vesicles,25 monolayers26 or bilayers.27 Nonetheless, recent developments of ESI sources in MS provide adequate tools for such kind of studies.28 Efficient analysis of complexes occurring at the interface formed between water and 1,2-dichloroethane (DCE) has been recently carried out29 combining dual-channel microsprayers30 with the classic ESI technique. These microsprayers enable the injection of two immiscible liquids placing them in contact only at the tip where the Taylor cone is formed.29 The application of a high external voltage leads to electrocapillary emulsification31 along with intense swirls at the Taylor cone,32 making the analysis by MS of interfacial complexes feasible. In order to obtain valuable insight information of metal ion–phospholipid complexation reactions, we here apply this new approach, called biphasic electrospray ionization (BESI).28,29,33 Reactions between different metal ions dissolved in an aqueous phase and a phospholipid dissolved in DCE will thus be followed by MS.
DPPC was diluted to 200 μM in DCE. The salts (NaCl, KCl, CaSO4, CuSO4 and LaCl3) were diluted in water to 200 μM or 1 mM. La3+ solution was titrated with a complexometric method at 0.16 or 0.79 mM.
Mass spectrometric measurements were carried out in an LCT time of flight (TOF) mass spectrometer (Micromass, Manchester, UK) used in positive ionization mode. The commercial electrospray interface was removed and the BESI source was mounted in front of the MS. The high voltage was connected to the stainless steel needle of the syringe containing the aqueous solution. The pump was switched on (1 μL min−1 for each line, i.e. a total flow rate of 2 μL min−1) and the MS power supply was set at 5.0 kV. The source temperature was fixed at 130 °C. The ion optics parameters were tuned in order to maximize high molecular weights. The mass spectra were averaged during 1 min.
Conversion rates χ of [M-DPPCn]z+ were calculated as follows:
![]() | (1) |
Mz+ + nDPPC → [M-DPPCn]z+ | (2) |
Fig. 1a shows the mass spectrum obtained for K+. It clearly exhibits a mass shift of 38 Th of one, two and three DPPC, the most abundant being [K-DPPC]+ at m/z = 772.4 Th. A fragment of 2 DPPC was also observed at m/z = 1230.0 Th, which has been characterized with other metal ions.22,23 Na+ ions exhibit the same trend (MS not shown). Interaction with the divalent ion Ca2+ induces a different peak pattern with doubly charged clusters observed up to 8 DPPC (Fig. 1b). [Ca-DPPC4]2+ was the most abundant cluster. The tandem mass spectra (obtained with an ion trap MS) of the dimer and tetramer bound to Ca2+ ion showed the loss of DPPC ligands and fatty acid (palmitate) residues.23 Similar coordination was observed using a Cu2+ salt (Fig. 1c), i.e. copper(II) ions bound to 3–7 DPPC forming doubly charged clusters, and in addition copper(I) ions bound to one and two DPPC. [Cu-DPPC4]2+ was the most abundant cluster, in agreement with MS data reported for diacylglycerophosphocholines.20 The formation of copper(I) complexes from a copper(II) salt is well documented in ESI-MS and stems from the reduction of CuII to CuI during the ESI process.38–42 Moreover, different fragments observed and attributed to copper(II) ions, result from a higher yield of fragmentation induced by transition metal ions.24 Finally, La3+ was tested. The mass spectrum (Fig. 1d) shows high conversion rates despite a rather lower salt concentration used (0.8 mM instead of 1 mM) compared to the other ions tested and higher values of n, from 4 to 10. All the complexes were triply charged and no special fragments were induced by this cation.
![]() | ||
Fig. 1 Mass spectra of metal ion-DPPC complexes obtained with a BESI source coupled to a TOF-MS. (a) KCl, (b) CaSO4 and (c) CuSO4 at 1 mM each, and (d) LaCl3 at 0.8 mM, in water and DPPC 200 μM in DCE. Mass spectra were summed during 1 min. ![]() |
Fig. 2 summarizes the conversion rates of [M-DPPCn]z+ as a function of the number of DPPC molecules. Except for the monovalent ions K+, Na+ and Cu+ where χ decreases monotonically as n increases, the distributions exhibit Gaussian curves and La3+ gives the highest abundance taking into account the sum of all the [La-DPPCn]3+. In particular, the most abundant complexes were of stoichiometry 1:
1 for K+, Na+ and Cu+, 1
:
4 for Ca2+ and Cu2+ and 1
:
5 for La3+. The total conversion rate of the [M-DPPCn]z+ complexes decreases according to the following order La3+ > Ca2+ > Cu2+ > K+ > Na+ (Cu+ complexes were the least abundant because they stemmed from the reduction of Cu2+; therefore their concentration was much lower, affecting also the abundance of copper(II)-phospholipid complexes).
![]() | ||
Fig. 2 Conversion rates of the metal ion–DPPCn complexes as a function of n. The curve shapes are explained by the presence of a first and a second coordination spheres (marked with dashed lines). Colored dashed lines were added between markers for clarity. |
![]() | ||
Fig. 3 Enthalpy of binding for (a) K+ and (b) Ca2+ in the exchange process between free ion and water, and free ion and lipid molecules. (Computational data). Colored dashed lines were added between markers for clarity. |
![]() | ||
Fig. 4 Computational structure of Ca2+-DPPC4 (top), Ca2+-DPPC (left) and K+-DPPC (right). Green: calcium, purple: potassium and red: oxygen from phosphate (thickened). The dashed blue lines represent metal–O bonds. The aliphatic chains were cut for computational reasons. |
In the case of calcium–water interactions, Fig. 3b shows that the first six water molecules provide a bit more than half of the total hydration energy (−1180 kJ mol−1)43 also with a monotonic variation of the binding energy. Indeed, Bako et al. have shown that for clusters with 7 and 8 water molecules, those which are formed with six water molecules in the first hydration shell and with one or two in the second shell are more stable than those which are hepta- or octa-coordinated.43 They further suggested that these six water molecules of the first hydration shell in solution are arranged in well-defined octahedral geometry, each of these molecules being linked by hydrogen bonds to three molecules in the secondary shell. Here, the average binding energy for molecules of the first hydration shell is about 140 kJ mol−1 reflecting the higher charge of Ca2+ when compared to K+. With DPPC, computations show that four lipid molecules saturate the space around the cation by seven oxygen ligands coming from phosphate groups in close vicinity as shown in Fig. 4. Again, although the two sets of calculations cannot be compared directly, it is clear that complexation by a phospholipid is more favorable than that by four water molecules. While attempts to coordinate additional DPPC molecules to the calcium cation did not succeed, experimental results clearly show that complexes with 8 DPCCs are formed in the gas phase. From a steric viewpoint, the hexa-coordination of the calcium ion is ensured by 4 DPPC molecules (Fig. 4, top). Additional DPPC ligands could therefore result either from longer range ion–dipole interactions with the polar heads of the DPPC squeezed between the alkyl chains of the primary solvation shell to form a second solvation shell, or from van der Waals interactions thereby forming single ion–phospholipid complexes with the polar heads of the outer phospholipid molecules being located on the outside of the structures.
In the case of lanthanum, the number of water molecules in the inner and second hydration layer is equal to 9 and 16 respectively, as recently reviewed by Dognon et al.44 Here, we can observe up to 10 DPPC molecules. The maximum shown in Fig. 2 suggests that five must be involved in the inner coordination layer. As in the case of calcium, the additional DPPCs can either be aligned with the inner shell or forming a single ion complex in a head-to-tail conformation.
In the case of copper, Fig. 2 shows a mixed curve shape first decreasing before increasing. This stems from the two possible oxidation states of copper, Cu+ behaving as K+ and Cu2+ as Ca2+.
[M(H2O)x]z+(w) + nDPPC(o) → [M-(DPPC)n]z+(g) + xH2O(g) | (3) |
Regarding the gas phase, and taking into account the equality of the flow rates of both phases, two different situations can be envisaged: formation and ejection of biphasic droplets or ejection of monophasic aqueous and/or DCE droplets (Fig. 5). In the first case, biphasic droplets are likely to comprise an inner organic core surrounded by an outer aqueous shell (o@w, pathway I). One can show that the aqueous phase will surround spontaneously the organic core only if the following condition is fulfilled:
γDCE-a > γw-a + γw-DCE | (4) |
![]() | ||
Fig. 5 Schematic drawing of the two postulated pathways for the complexation/ionization process taking place after spraying an aqueous and an organic (containing DPPC) solution using the dual-channel microsprayer during BESI experiments. |
In the second case, the presence of monophasic aqueous and organic droplets within the fine mist formed during the electrospray process is considered (pathway II). In this case, inverted micelles formed within the Taylor cone are dissolved in the organic phase.
Finally, for both pathways, consecutive desolvation steps lead to stable cation–phospholipid complexes, as observed by MS. The ionization mechanisms would differ for the two pathways. Droplets fission proposed by the charge residue model (CRM)47,48 would be favored when o@w droplets are considered (Pathway I). On the contrary, the ion evaporation model (IEM)49 would predominate for the pathway II due to the prior presence of the complexes in the organic phase and absence of a blocking barrier at the DCE–air interface.
Independent of the ionization mechanism, important implications at the membrane scale can be drawn from the mass spectrometric evidence of these cation–DPPC clusters. First, structuring and packing effects caused by the cations adsorption/binding are known to have major influences on biologic membranes50,51 and can be rationalized in terms of the cation–lipid interactions. Hence, cations without the adequate charge and/or incapable of forming a second coordination sphere (e.g. K+ and Na+) will have a smaller impact on the membrane properties than cations with this ability (e.g. Ca2+, Cu2+ or La3+). In the second group of cations, La3+ will perturb in a greater extent the lipidic membrane. Indeed, it accommodates around it a higher number of lipid molecules (either at the first or at the second coordination sphere), as it has been reported by Lehrmann and Seelig.52 Therefore, intra-micellar interactions, specially taking place in the second coordination shell, should be responsible for the effects experimentally observed of metal cations over lipidic membranes, like packing effect. Analogously, it is shown that such interactions are strong enough to preserve the structure of the complexes in the gas phase.
As a last remark, clusters of [M-DPPCn]z+ can be formed at the liquid–liquid interface within either the Taylor cone or the expelled droplets from it, according to our previous results and discussion on the BESI source.29 The multistep binding of calcium to bilayers has been reported to be sub–microseconds,4 which is compatible with reactions driven under our conditions. These comparisons consolidate the validity of BESI sources for the analysis of in situ formed lipid complexes.
This journal is © The Royal Society of Chemistry 2010 |