MALDI-TOF/MS analysis of archaebacterial lipids in lyophilized membranes dry-mixed with 9-aminoacridine.

A method of direct lipid analysis by MALDI mass spectrometry in intact membranes, without prior extraction/separation steps, is described. The purple membrane isolated from the extremely halophilic archaeon Halobacterium salinarum was selected as model membrane. Lyophilized purple membrane were grinded with 9-aminoacridine (9-AA) as dry matrix, and the powder mixture was crushed in a mechanical die press to form a thin pellet. Small pieces of the pellet were then attached to the MALDI target and directly analyzed. In parallel, individual archaebacterial phospholipids and glycolipids, together with the total lipid extract of the purple membrane, were analyzed by MALDI-TOF/MS using 9-AA as the matrix in solution. Results show that 9-AA represents a suitable matrix for the conventional MALDI-TOF/MS analysis of lipid extracts from archaeal microorganisms, as well as for fast and reliable direct dry lipid analysis of lyophilized archaebacterial membranes. This method might be of general application, offering the advantage of quickly gaining information about lipid components without disrupting or altering the membrane matrix.


Microorganism growth conditions
The engineered high-producing BR strain of Hbt. salinarum used in this study was kindly provided by Richard Needleman ( 10 ). The Hbt. salinarum cells were grown in light at 37°C in liquid growth medium containing neutralized peptone (L34, Oxoid), prepared as previously described ( 11 ). The MdS1 strain, a representative of Halorubrum sp. , was isolated from the salterns Margherita di Savoia and grown in our laboratory as previously described ( 12 ).

Isolation and lyophilization of archaebacterial membranes
Purple membranes were isolated from the high-producing BR strains of Hbt. salinarum as previously described ( 11 ) and when indicated were used in lyophilized form. The red membranes of the extremely halophilic microorganism MdS1 were isolated as previously described ( 13 ) and lyophilized before lipid analysis.

Lipid extraction of the purple membrane
Total lipids of the PM were extracted using the Bligh and Dyer method ( 14 ), as modifi ed for extreme halophiles ( 15 ). The extracts were carefully dried under N 2 before weighing, and then dissolved in chloroform (10 mg/ml).

Isolation and purifi cation of individual lipids from the total extract
The lipid components of PM were separated by preparative TLC (Merck 20 × 20 cm × 0.5 mm thick layer) in Solvent A. Lipids were visualized by staining with iodine vapor and were eluted and recovered from the scraped silica as previously described ( 15 ). Isolated and purifi ed phospholipids were dissolved in chloroform at the concentration of 1 mg/ml; glycolipids and cardiolipin analogs in the mass range 1000-2000 amu were resuspended at the fi nal concentration of 2 mg/ml.

Sample preparation of lyophilized archaebacterial membranes for MALDI-TOF/MS
The procedure for sample preparation is analogous to the preparation of KBr samples for infrared analysis ( 16,17 ). Lyophilized archaebacterial membranes (PM and red membrane of MdS1) and dry 9-AA matrix were mixed in 2:1 ratio (w/w). Lyophilized membranes were grinded with 9-AA in an agate mortar, and the powder mixture was crushed in a mechanical die press to are considered adaptive traits of microorganisms able to thrive in harsh or extreme environments, such as saturated salt, anoxic, and high-temperature waters ( 5 ).
As many of the lipid components of archaeal cell membranes are not commercially available, we have isolated and purifi ed individual archaeal lipids for MALDI-TOF/ MS analysis. The purple membrane (PM) of the extreme halophile Hbt. salinarum was selected as a model membrane for our studies.
The PM domain has an important role in the bioenergetics of the microorganism, as it contains the photoactivated proton pump bacteriorhodopsin. Bacteriorhodopsin is able to convert green sunlight in a proton gradient across the cell membrane (more acid outside); the protons fl owing back through ATP synthase allow ATP synthesis inside the cell. The purple membrane is constituted of a 2D crystalline lattice, formed by bacteriorhodopsin as only protein and a small number of lipid molecules (about 10 per each protein molecule). In the past we determined the PM lipid-protein stoichiometries by means of nuclear magnetic resonance (NMR) analysis ( 9 ).
In the present study we illustrate a new method of direct MALDI-TOF/MS analysis of lyophilized PM fi nely grinded with 9-AA as dry matrix. The simplifi ed method of sample preparation for MALDI lipid analysis is also suitable for other archaebacterial membranes in lyophilized form. In parallel, to validate results obtained with the new method, individual lipid components and the total lipid extract of archaebacterial membranes were analyzed for the fi rst time by MALDI-TOF/MS.
Glycolipids (diphytanylglycerol ether analogs). In the genus Halobacterium , the major glycolipid is the sulfated triglyco- has also been found ( 18 ). S-TGD-1 is the main glycolipid of the purple membrane, while S-TeGD is present in minor amount very likely as a contaminant.
Furthermore a glycosylated cardiolipin is present in the purple membrane of Hbt. salinarum ( 15 ). Glycosyl derivatives of cardiolipin have also been described in bacteria ( 8,19 ). Fig. 2 shows the MALDI-TOF mass spectra of isolated and purifi ed S-TGD-1 and S-TGD-1-PA and, at the bottom, their chemical structures.
The glycosylated cardiolipin of PM has the structure of a complex phosphosulfoglycolipid, having two diphytanylglycerol moieties, can be considered a dimeric phospholipid. The structure of S-TGD-1-PA was fi rst determined in our laboratory by combining TLC, NMR, ESI-MS, and MS/MS analysis of the isolated and purifi ed glycolipid and its hydrolysis products ( 15 ). It has also been shown that S-TGD-1 is the precursor of S-TGD-1-PA ( 20 ). form a thin pellet disc (less than 1 mm thickness). Small pieces of the disc were then fi xed to the MALDI target with double-sided adhesive tape and directly analyzed.

MALDI-TOF/MS
MALDI-TOF mass spectra were acquired on a Bruker Microfl ex mass spectrometer (Bruker Daltonics, Bremen, Germany). The system utilizes a pulsed nitrogen laser, emitting at 337 nm; the extraction voltage was 20 kV. 600 single laser shots were averaged for each mass spectrum. In the analysis of lipids in solution, the spectrum was obtained by moving the laser within the spot. The laser fl uence was kept at about 80% of maximum value to have a good signal-to-noise ratio. At variance in the analysis of lipids in intact membranes, it was not necessary to move the laser during the acquisition of the spectrum, thanks to the thickness of the pellet (see below). It was necessary to keep the laser fl uence at 100% for good desorption and ionization of the sample in the pellet; at lower laser intensity, it was not possible to have good signals. All spectra were acquired in negative ion mode using the delayed pulsed extraction. Spectral mass resolutions and signalto-noise ratios were determined by the software for the instrument, "Flex Analysis 3.0" (Bruker Daltonics). When indicated, spectra were acquired with a Bruker Autofl ex mass spectrometer (Bruker Daltonics) in refl ector mode; the laser fl uence was kept about 10% above threshold to optimize the signal-to-noise ratio.

Application of MALDI-TOF/MS to the analysis of archaebacterial lipids in solution
Phospholipids (diphytanylglycerolphosphate ether analogs). The MALDI-TOF mass spectra of some archaebacterial phospholipids present in the membranes of the extremely halophilic archaeal microorganisms are reported in Fig. 1 . In particular the following lipids were analyzed: PG, PGP-Me, PGS, and BPG. The chemical structures of analyzed phospholipids are illustrated at the bottom of the Fig. 1 . All these archaebacterial phospholipid standards have been isolated and purifi ed from the lipid extract of Hbt. salinarum . PGP-Me is the major phospholipid in all extreme halophiles and extreme haloalkalophiles as well ( 4 ). BPG is the archaeal analog of eukaryotic or bacterial cardiolipin, an interesting kind of dimeric phospholipid having four phytanyl chains in the hydrophobic tail.
The basic unit of all archaebacterial phospholipids is represented by the archaetidyl group or phosphatidic acid, which is generally detected only in traces in the cellular extracts as biosynthetic intermediates or is produced by molecular fragmentation in mass spectra. Table 1 reports formulas, exact masses, and MALDI-TOF/MS signals of archaeal phospholipids.
In some spectra, fragmentation ions are present together with the molecular ion; fragments produced in the MALDI-TOF/MS lipid analysis of equimolar mixtures of archaebacterial phospholipids, including BPG, showed that the BPG signal is always very small compared with those of other phospholipids (data not shown).

Lipid analysis of intact lyophilized archaebacterial membranes dry mixed with 9-aminoacridine
Recently MALDI-TOF/MS analysis of insoluble organic compounds of high molecular weight has also been performed by dry mixing the analyte and matrix, and then crushing it in a mechanical die press to form a thin disc or pellet that can be directly scanned by the MALDI laser beam ( 16,17 ). We tested the possibility of applying this approach to the direct lipid analysis of isolated membranes in lyophilized form, thus avoiding the steps of membrane disruption, lipid solubilization, and phase partition.
It is well known that the purple membrane of Hbt. salinarum has an extraordinary thermal and chemical stability; we have verifi ed that is possible to keep PM in lyophilized form at room temperature for months without changing their lipid composition (unpublished observations).
Lyophilized purple membranes were fi nely mixed with the powder of 9-AA. The dry mixture was kept under high fi nally the small peak at m/z 731.7 corresponds to PA, a lipid fragment. At variance from ESI-MS lipid profi le, in MALDI-TOF mass spectrum the peak of BPG can be barely detected by enlarging the y axis, although it gives a good signal when analyzed individually (see mass spectrum in Fig. 1 ).
It has been shown that BPG is only a minor lipid component of PM, likely a contaminant, while S-TGD-1-PA is present in stoichiometric ratio 1:1 with bacteriorhodopsin ( 9 ).
membranes in lyophilized form was also verifi ed. Fig. 6 illustrates the MALDI-TOF/MS analysis of lyophilized red membranes isolated from the strain MdS1, another extremely halophilic microorganism of the Archaea world. MdS1 red membranes are highly enriched in BPG ( 12,13 ).
The lyophilized MdS1 red membranes and dry 9-AA were mixed in 2:1 ratio, obtaining a thin red pellet. A fragment of the pellet was used for MALDI-TOF/MS analysis. The resulting MALDI-TOF/MS lipid profi le was perfectly superimposable onto the lipid profi le obtained by previous TLC and ESI-MS analysis ( 12,13 ). Fig. 6A reports the TLC lipid profi le of red membranes and lipid assignments based on data in previous literature and mass spectrometric analysis. Experimental lipid masses obtained by MALDI-TOF/MS analysis correspond well to calculated lipid masses and to those previously obtained by ESI-MS analysis ( 12,13 ). It can be seen that in the MALDI spectra of membrane/matrix pellets most of archaebacterial lipids are represented by their sodium adducts. The excellent quality of the MALDI-TOF/MS spectrum of lyophilized MdS1 membranes in Fig. 6 is likely due to the optimal homogeneity of the red membranes/9-AA pellet. pressure in a die press to form a thin disc, similar to those obtained by mixing substrates and KBr for IR analysis. Fig.  4 shows the fragment of PM/9-AA pellet attached to the MALDI target.
A representative MALDI-TOF mass spectrum obtained by analyzing the PM/9-AA pellet is reported in Fig. 5 . The MALDI-TOF/MS lipid profi le obtained in these experimental conditions corresponds quite well to the one reported in Fig. 3 that was obtained by analyzing the PM lipid extract. The glycosylated cardiolipin S-TGD-1-PA is represented by the signal at m/z 1955.5, 1933.5, and 1853.3, corresponding to the sodium adduct of the molecular ion, the molecular ion, and the desulfated glycolipid, respectively. The signals at m/z 1380.4 and 1218.3 correspond well to the glycolipid S-TeGD and S-TGD-1, respectively. PGS and PGP-Me, both phospholipids having two acidic groups in the polar head, are represented by a mixture of molecular ions and sodium adducts, whereas PG is represented only by its molecular ion. The mass deviation in the high range of lipid masses might be due to differences in the sample height in the dry mixture of the pellet.
The possibility of extending the novel method of MALDI-TOF/MS lipid analysis to other archaebacterial density to the halophile membranes. No nitrogenous base-containing phospholipids, such as phosphatidylserine or phosphatidylethanolamine, are present in extreme halophiles. This may be characteristic of the halophilic Archaea, in contrast to the methanogenic and thermophilic Archaea. Information on lipid biosynthesis in Archaea is available thanks to labeling studies with whole cells ( 5,22 ). In a previous study, we analyzed the lipids of the purple membrane by combining TLC and NMR analysis. The It is noteworthy that the BPG signal at m/z 1542.9 is comparable to that of other membrane lipid components; therefore, the novel method of lipid analysis offers the advantage of overcoming the problem of diffi cult ionization of BPG when it is mixed with other phospholipids in solution.

DISCUSSION
Recent applications of MALDI-TOF/MS to the analysis of cellular lipids have shown the potential of a technique long considered of practical utility only for protein analysis ( 2,21 ).
It has been shown that MALDI mass spectrometry allows the analysis of glycerolipids, glycerophospholipids, triglycerides, cholesterol, and cholesterol derivatives, while reports describing the application of MALDI to the analysis of archaebacterial lipids are not yet available.
In this study, we illustrate for the fi rst time the application of MALDI-TOF/MS to the analysis of archaebacterial lipids in solution, and we show that it is possible to directly analyze lipids of membranes isolated from archaeal microorganisms without prior extraction/separation steps.
We have analyzed archaeal membrane lipid standards and the total lipid extract of the archaeal membranes isolated from halophilic microorganisms, obtaining results comparable with those obtained by ESI-MS analysis.
Most of the phospholipids and glycolipids of extreme halophilic archaeons are anionic, so that their negatively charged groups would impart a high negative charge  ( 15,20 ). B: MALDI-TOF mass spectrum of the total lipid extract of purple membranes of Hbt. salinarum acquired in the negative ion mode using 9-AA as the matrix.   results obtained allowed the estimation of stoichiometric ratio of individual lipid component and the sole protein of the purple membrane bacteriorhodopsin ( 9 ).
In this study, we considered the purple membrane as a model membrane to investigate the potential of MALDI-TOF/MS in the analysis of lipids of archaeal microorganisms. Results indicate that 9-AA in solution is a suitable matrix for the desorption/ionization of archaeal phospholipids and glycolipids. The method, originally developed by Sun G. et al. ( 2 ), allows good ionization and easy detection of the dimeric archaeal phospholipids or cardiolipins. The sensitivity of the method is particularly high for the archaeal sulfated glycolipids, which in analogy with sulfatides of eukaryotic cells ( 23 ), exhibit an extremely high tendency to ionize in the presence of 9-AA.
The MALDI-TOF/MS profi le of the lipid extract of the purple membranes contains the same main signals present in the ESI mass spectrum ( 15 ). All the lipids in the extract can be easily revealed by this technique of analysis, with the exception of the minor PM lipid component BPG, whose signal is very low.
Interestingly we also show that it is possible to perform a direct lipid analysis of the intact lyophilized archaebacte-rial membranes by MALDI-TOF/MS dry mixed with the matrix. We recommend care in the preparation of the membrane/matrix pellets, because the incomplete homogeneity of the samples might cause some inaccuracy in the measurements.
The study of lipid composition of cell membranes normally requires two separate steps: 1 ) extraction of lipids by harsh organic solvents and 2 ) combined chromatographic and mass spectrometry analysis. The novel approach to analyze lipids of intact membranes is particularly useful if doubts about the presence of artifacts introduced by the lipid extraction procedure are raised (e.g., highly polar lipids as sulfated polyglycolipids may preferentially partition into the methanol-water phase and therefore be absent in the chloroform phase). In a previous study of 31 P NMR analysis of solubilized purple membrane, the phosphorus signal of the glycosylated cardiolipin S-TGD-1-PA could be not detected, raising doubts about the presence of the archaeal cardiolipins in the PM ( 24 ). In the present study, our data confi rm that archaeal cardiolipins are true components of the membrane matrix of the purple membrane.
The direct lipid analysis of lyophilized membranes appears to be of general applicability for archaebacterial lipids, whose structures are chemically stable even in extreme environmental conditions. During the dehydration process, the polar heads of phospholipids lose the associated water, whereas sodium ions likely remain bound to membrane. This might explain why most of highly polar lipids of archaebacterial membranes appear as sodium adducts in the MALDI spectra of lyophilized membranes.
The method of MALDI-TOF/MS lipid analysis of lyophilized membranes is particularly suited to the analysis of the complex archaebacterial phospholipid BPG, allowing an easier detection of it. The sodium adduct of BPG represents one of the main signals in the MALDI-TOF/MS lipid profi le in Fig. 6 , although it appears diffi cult to obtain the ionization of BPG in solution in the presence of other phospholipids and glycolipids.
The novel method we describe might be of general applicability. It offers the advantage of quickly gaining information about membrane lipid components without disrupting or altering the membrane matrix. Experiments are in progress to evaluate the potential of this novel method in the study of lipid composition of bacterial and eukaryotic membranes.