Electrospray ionization (ESI)-MS was initially developed by Fenn and colleagues for analysis of biomolecules (Fenn et al., 1989). It depends on the formation of gaseous ions from polar, thermally labile and mostly non-volatile molecules. The "soft" electrospray ionization causes little or no fragmentation and virtually all phospholipid species are detected as molecular ion species. The identification and structural information about a certain peak is acquired by fragmentation (tandem mass spectrometry, MS/MS) (Han and Gross, 1994; Kim et al., 1994). Advances in mass spectrometry (MS) have created powerful new avenues for the measurement of global changes in the cellular lipid inventory.

Electrospray ionization (ESI) mass spectrometry has been successfully used to analyze lipids in various cell types. However, for less abundant and more complicated chemical compositions, HPLC is needed for upfront separation before ionization by ESI and detection by MS (Watson, 2006). In the field of lipidomics HPLC or TLC have been coupled with electrospray ionization (ESI)/atmospheric pressure chemical ionization (APCI)-MS and matrix-assisted laser desorption/ionization (MALDI)-MS, respectively (Wenk, 2005). These methods provide the identification of lipid composition and quantification of cellular lipids, such as fatty acids, phospholipids, cholesterols, and sphingolipids.

The identification of lipid composition and quantification of cellular lipids provide molecular signature for a lipid related pathway. One of the most obvious applications of lipidomics will include profiling lipid extracts in order to identify metabolic pathways and enzymes that are affected (Wenk, 2005). A major advantage of such lipidomics-based discovery is that, together with our relatively good understanding of many biosynthetic and metabolic pathways, it will lead to the identification of pathways and enzymes governing human diseases (Cascante et al., 2002). A recent review (Wenk, 2005) well summarized diverse aspects and fundamental principles of lipidomics.

Phosphoinositides are less abundant lipids in eukaryotic cell membranes. Therefore, analysis of phosphoinositides requires special cautions during sample treatment. According to several reports (Hsu and Turk, 2000; Wenk et al., 2003; Milne et al 2005; Shui et al., 2007) to minimize the interference of other phospholipids during analysis, a selective two-step extraction was used (Scheme 1). First, the cell material was extracted with neutral solvents (mixture of chloroform (CHCl₃) and methanol (MeOH)) and the resulting pellet was extracted with acidified solvents (containing HCl or citric acid) for quantitative recovery of phosphoinositides.

After extraction of phosphoinositides, HPLC and TLC separation have been employed in order to classify molecular species such as phosphatidylinositol (PI), phosphatidylinositol 3-phosphate (PI3P), phosphatidylinositol 4-phosphate (PI4P), phosphatidylinositol 5-phosphate (PI5P), phosphatidylinositol (3,4)-bisphosphate (PI(3,4)P₂), phosphatidylinositol (3,5)-bisphosphate (PI(3,5)P₂), phosphatidylinositol (4,5)-bisphosphate (PI(4,5)P₂), and phosphatidylinositol (3,4,5)-trisphosphate (PI(3,4,5)P₃, PIP₃). Traditionally, the molecular species have been detected, identified and quantified by autoradiograph (Wenk et al., 2003). However, the identification based on radio-labeling could not provide information of fatty acyl composition of phosphoinositides that can be solved by the application of modernized mass spectrometry methods. Current mass spectrometry is able to analyze lipid extracts of phosphoinositides without prior separation through HPLC or TLC due to its high resolving power.

Several methods such as fast atom bombardment mass spectrometry (FAB-MS), matrix-assisted laser desorption and ionization/time of flight mass spectrometry (MALDI-TOF-MS) and ESI-MS have been used for the analysis of phosphoinositides. MALDI-TOF-MS has been recently applied for phosphatidylinositols in murine brain extracts (Berry et al., 2004) and ESI-MS has been used for PIPs and PIP₂s in lipid extracts from cultures cells and rat brain (Wenk et al., 2003). ESI-MS has also been utilized for PIPs, PIP₂s and PIP₃s in murine macrophage extracts (Milne et al., 2005). These biological applications of ESI-MS can provide us of understanding of the acyl chain contents of phosphoinositides due to its technical advance and advantage. The fatty acyl composition of phosphoinositides is assigned by precursor ion scan or fragmentation analysis in ESI-MS in negative ion mode. Among these applications for biological samples, a report (Milne et al., 2005) for the analysis of phosphoinositides in murine macrophage cells presented the identification of 28 PIP and PIP₂ compounds as well as 8 PIP3 compounds. Additionally, fatty acid composition of lyso-phosphoinositides in plasma from patients with ovarian cancer was identified through the application of ESI-MS (Sutphen et al., 2004). According to a report (Milne et al., 2005), the limit of detection using ESI-MS is < 9 pmol/ml for 38:4 PIP₃.

In general, single-stage ESI-MS is enough to investigate the fatty acid composition of relatively abundant phosphoinositides from biological samples. The inspection of samples using tandem mass spectrometer (MS/MS) combined with collision-induced dissociation (CID), can identify and quantify less abundant phosphoinositides with their fatty acid composition and degree of unsaturation (Wenk et al., 2003). MS/MS can monitor transition from a parent ion to a product ion of small quantity of phosphoinositides and provide increasing sensitivity.

In the analysis of phosphoinositides using MS, we need to consider suppression effect of ion signal of phosphoinositides by other components in lipid extracts. In order to solve this effect, they also presented the effect of the addition of piperidine onto cellular extracts on the change of signal to noise (S/N) ratio (Wenk et al., 2003; Milne et al., 2005). This addition of a organic buffer, piperidine results in enhancement of ion signal of phosphoinositides, similarly as in the analysis of phosphopeptides and oligonucleotides.

Using these methods, there are several investigations on detailed analysis of cellular phosphoinositides. They identified approximately 75 species of phosphoinositides with detailed fatty acyl chain composition and profiled them from mouse brain tissue, yeast mutant cells (Wenk et al., 2003), RAW 264.7 cell, primary murine peritoneal macrophage (Milne et al., 2005) and Mycobaterium bovis (Shui et al., 2007). Interestingly, they detected elevated concentrations of PIP2 in human fibroblasts from patients with Lowe syndrome (Wenk et al., 2003). Lowe syndrome is a genetic disorder characterized by bilateral congenital cataracts, renal Fanconi syndrome, and mental retardation. It is coming from mutations of a phosphoinositide 5-phosphatase. Thus, it results in the disruption of phosphoinositide metabolism and increment of PIP2 concentrations in human cells.

In order to suggest the significance of phosphoinositides profiling, we shortly summarize cellular signaling mechanism of phosphoinositides. The various phosphoinositides are maintained at steady state levels in the inner leaflet of the plasma membrane by a continuous and sequential series of phosphorylation and dephosphorylation reactions by specific kinases and phosphatases, respectively, which are regulated and/or relocated through cell surface receptors for extracellular ligands (Di Paolo and De Camilli, 2006). This has been termed a 'futile cycle', and can consume a significant proportion of cellular ATP production. Controlled synthesis of these different phosphoinositides can occur in different intracellular compartments for distinct and independently regulated functions with differing target enzymes (Roth, 2004). Among the enzymes relating to phosphoinositides phosphorylation and dephosphorylation, phosphoinositide 3-kinase (PI3K) related pathway is well characterized (Vivanco and Sawyers, 2002). The lipid kinase, PI3K is involved in the regulation of a number of cellular processes, like transcription, migration, angiogenesis, cell growth, proliferation, apoptosis and glucose metabolism. PI3K phosphorylates phosphatidylinositols of the cell membrane, thus generating, e.g., phosphatidylinositol-3,4,5-trisphosphate (PIP₃) from phosphatidylinositol 4,5-bisphosphate (PIP₂). PIP₃ at the cell membrane recruits protein kinases such as protein kinase B or Akt (PKB/Akt) and phosphoinositide-dependent kinase-1 (PDK1) which bind with their PH-(pleckstrin homology)-domain to PIP₃. The downstream signaling of PI3K phosphorylate a wide variety of cellular signaling molecules relevant for the regulation of cell growth, cell cycle and cell proliferation, for apoptosis, or for transport (Figure 2). PKB/Akt further activates mammalian target of rapamycin (mTOR), which is a kinase stimulating the uptake of nutrients such as glucose, amino acids, cholesterol and iron. mTOR further activates eIF4E-binding protein-1 and thus is involved in the regulation of translation (Pitt and Chen, 2008). Besides, phosphoinositides are key components of the nucleus of the cell, where they have many essential functions, such as DNA repair, transcription regulation and RNA dynamics (Hammond et al., 2004). They are considered as activity switches responsible for such processes, with the phosphorylation state of the inositol ring. Different forms of phosphoinositides appear to have specific functions at each level of gene expression, so extracellular events must coordinate the production of these compounds in a highly organized manner (Han and Gross, 2005).

The choice of matrix, one of important issues for MALDI-MS, is not simple for in vitro and in situ lipid analysis due to the special properties of lipids, including lower molecular weights (<1,500 Da), rather wide molecular diversity, very close presented m/z ratio, relative insolubility in aqueous systems and the obvious concentration difference within a crude sample. Various matrices had been screened and suggested to be useful for phospholipid analysis (Figure 3, Table 1). Besides the low chemical noise and interferences from matrix relevant ions, the important prerequisite of a matrix for lipid analysis are namely, good absorptivity at the given laser wavelength, good solubility in analyte solvents, suitable acidity and basicity and high ionization efficiency of analyte molecules.

Among the matrices shown in Figure 3, 2,5-dihydroxybenzoic acid (DHB), so far, is regarded as the most suitable for lipid analysis in rat brain and neutrophilic granulocytes because it can produce cleaner spectra and fewer fragments of analyte ions (Schiller et al., 1999). Furthermore relevant matrix peaks do not interfere with analyte identification (Petkovic et al., 2001). Additionally, suitable vacuum stability of DHB is a key factor for its popular application in imaging mass spectrometry (IMS) of lipids (Wang et al., 2007). When used for phospholipids analysis in mouse brain, rat brain, mouse cerebellum, mouse retina, single zooplankter individuals, and lipid mixtures extracted from tissues, matrix solution can be prepared with a number of solvent system, such as ethanol/water (1:1, v/v with or without 0.1% trifluoroacetic acid (TFA)) (Schwartz et al., 2003; Jackson et al., 2005b; Woods et al., 2006; Puolitaival et al., 2008) ethanol/water (9:1, v/v, with or without 0.1% TFA) (Fujiwaki et al., 2002) acetonitrile/water (1:1, v/v, with 0.1% TFA) (Chen et al., 2008) methanol/water (7:3, v/v) (Hayasaka et al., 2008; Shimma et al., 2008) and chloroform/methanol (2:1 v/v ) (Ishida et al., 2003). Addition of small amounts of TFA in the matrix solution was found to enhance considerably signal to noise (S/N) ratio in the case of phosphatidylcholines and phosphoinositides, most likely due to enhanced solubility of lipids (Schiller et al., 1999; Petkovic et al., 2001).

One obvious drawback of DHB as MALDI matrix is to its tendency to form large crystals, where the analyte is not evenly distributed, leading to poor spot-to-spot reproducibility and lowering mass resolution (Bouschen and Spengler, 2007). DHB has some disadvantages for phospholipid analysis as shown in Figure 4. Due to its relatively higher acidity, DHB produced weak or no signal intensity and exhibited poor sensitivity for phospholipids, especially for phosphatidic acids (PAs), phosphatiylserines, phosphatidylinositols, phosphatidylglycerols, phosphatidylethanolamines in negative-ion mode (Petkovic et al., 2001; Estrada and Yappert, 2004a).

Several neutral, basic or low acidic compounds were also tested their suitability for MALDI-MS analysis of phospholipids (Estrada and Yappert, 2004b; Jackson et al., 2005a; Woods et al., 2006; Wang et al., 2007) to alleviate difficulties in the detection and identification of phospholipid classes with ionization efficiencies lower than those of sphingomyelins and phosphatidylcholines and to improve the sensitivity of negative-ion mass spectra. Among the matrices shown in Figure 3, 2,6-dihydroxyacetophenone is one of most employed matrices. 2,6-Dihydroxyacetophenone in 50-70% ethanol or methanol with or without 0.1%TFA is commonly used for lipid analysis (Jackson et al., 2005a; Shimma et al., 2007). Cesium iodide (Wang et al., 2007) and lithium chloride (Jackson et al., 2005a) can be added. 2,6-Dihydroxyacetophenone can be used for both positive and negative ionization mode analysis of phospholipids (Woods et al., 2006; Jackson et al., 2007; Wang et al., 2007), and works more efficient than DHB in terms of detection sensitivity, resolution and signal-to-noise of mass spectra. 2,6-Dihydroxyacetophenone and another neutral matrix of 6-aza-2-thiothymine (ATT) also can be used to detect noncovalent complexes between chlorisondamine and phosphatidylcholine molecular species phosphatidylcholine32:0 and phosphatidylcholine34: 1 on a rat brain tissue added with 1 nmol of chlorisondamine. Noncovalent complex formation was not observed when highly acidic matrices such as sinapinic acid (SA) and cyano-4-hydroxycinnamic acid (CHCA) were used (Jackson et al., 2005a). p-Nitroaniline proved to be another good option for the use of DHB for MALDI-MS analysis of phospholipid in a lens tissue (Figure 4). Due to its higher absorptivity and its lower acidity than that of DHB, p-nitroaniline provided superior sensitivity for analysis of phospholipids either positive-ion or negative-ion modes and increased the relative amount of deprotonated species in the negative-ion mode than DHB (Estrada and Yappert, 2004a, 2004b). Therefore, it was possible to confirm peak assignments for phospholipid classes (phosphatidylglycerols and phosphatiylseriness) that normally give weak signals when DHB is used. p-Nitroaniline allowed the identification of phosphatidylethanolamines and phosphatidylethanolamines plasmalogens (PEps) even in mixtures containing sphingomyelins and phosphatidylcholines (Estrada and Yappert, 2004b). However, p-nitroaniline had the same drawback as 2,6-dihydroxyacetophenone, a high vapor pressure. Although addition of CsCl in p-nitroaniline solution can minimize matrix (Rujoi et al., 2004), p-nitroaniline is precluded in MALDI-MS of phospholipids due to poor vacuum stability.

Most recently, 2-mercaptobenzothiazole as an advantageous alternative to the use of DHB for MALDI-MS of phospholipids in a brain and liver tissue was developed (Astigarraga et al, 2008). Several special features of 2-mercaptobenzothiazole allows its superior matrix efficiency to that of the commonly used matrices of DHB, 2,6-dihydroxyacetophenone and p-nitroaniline for profiling and imaging phospholipids in liver and brain samples both in vitro and in situ, directly on tissue slices. Its low vapor pressure allows acquisition times of hours, which makes it possible to create well-defined mass images by scanning the whole tissue slices. Its low acidity allows identification of several additional species in negative-ion mode and enables the detection of the largest number of lipids from liver and brain samples. The formation of small and uniform crystals of 2-mercaptobenzothiazole in the IMS sample preparation is a paramount factor for the achievement of good spatial resolution and uniform spectra. Its crystallization in very small and homogeneous crystals also allows high detection reproducibility.

Some ionic liquid or solid matrices have exhibited some advantages over traditional single solid matrixes such as DHB for in vitro or in situ phospholipid analysis by MALDI-MS in a brain tissue and biological fluids. Due to their notable features of strong absorptivity, low vacuum volatility (negligible vapor pressure), promote ionization and homogeneous solution, CHCA-based ionic matrices, such as CHCA/n-butylamine (nBA-CHCA) (Jones et al., 2005; Li, et al., 2005), CHCA/1-methylimidazole, CHCA/pyridine, CHCA/tripropylamine, CHCA/tributylamine and CHCA/aniline (Li et al., 2005), have been reported to give better quality data in terms of ionization efficiency, signal intensities, sensitivity, resolution, signal reproducibility, number of compounds detected, and contaminant tolerance. The utilization of p-nitroaniline-based solid ionic matrix, p-nitroaniline/butyric acid, can reduce mass spectrum complexity by reliably appearing of only protonated molecules [M + H]⁺ of lipids containing phosphatidylcholine head groups, such as lysophosphatidylcholine, phosphatidylcholine, and platelet-activating factor (PAF) and of monosodium adducts [M + Na]⁺ as the major molecular ions for anionic phospholipids, such as phosphatidylglycerol, PA, and phosphatiylserine in the mass spectra (Ham et al., 2005). Another feature of p-nitroaniline/butyric acid is its ability to simultaneously detect phosphorylated lipids of phosphatidylcholines, sphingomyelins, PAFs and phosphatidylethanolamines in the positive-ion mode.