Direct liquid extraction surface analysis mass spectrometry of cell wall lipids from mycobacteria: Salt additives for decreased spectral complexity

• Users may freely distribute the URL that is used to identify this publication. • Users may download and/or print one copy of the publication from the University of Birmingham research portal for the purpose of private study or non-commercial research. • User may use extracts from the document in line with the concept of ‘fair dealing’ under the Copyright, Designs and Patents Act 1988 (?) • Users may not further distribute the material nor use it for the purposes of commercial gain.


| INTRODUCTION
Tuberculosis (TB) continues to be the primary cause of global morbidity and mortality caused by a single infectious agent. The etiological agent of TB, Mycobacterium tuberculosis, is a unique bacterial pathogen with a highly complex cell envelope that is essential for its pathogenicity, virulence and ability to combat molecular attack from host defence mechanisms and antibiotic therapy. The mycobacterial cell envelope is loaded with a complex polysaccharide, arabinogalactan, which serves to covalently connect the inner peptidoglycan layer with an outer mycolate layer that is highly hydrophobic. This complex cell wall structure has been directly linked with natural antibiotic resistance. 1 Phosphatidylinositol mannosides (PIMs) are essential phospholipid components of the cell envelope. PIMs are predominantly located in the mycobacterial plasma membrane, which is highly asymmetric and also contains other phospholipids such as phosphatidylinositol (PI), phosphatidylethanolamine (PE) and cardiolipin (CL). The PIMs appear in different glycosylation and acylation states and have been shown to play an important role in host-pathogen interactions. 2 The specific acylation states of these lipids have been linked with the initiation of granuloma formation; 3  however, only a handful have been described for the direct analysis of bacterial colonies. Desorption electrospray ionisation (DESI), which desorbs analytes from a surface in a continuously sprayed jet of solvent ions, has recently been described for imaging of recombinant small molecule biocatalysts from E. coli. 4 Rapid evaporative ionisation mass spectrometry (REIMS), also known as the iKnife, which continuously ablates a sample and then detects gaseous ions removed from the sample, has been described for the analysis of phospholipids from bacterial colonies including P. aeruginosa, B. subtilis, and S. aureus grown on agar. 5 Flowprobe MS, which samples a surface via a continuous solvent flow over the sample, before aspirating away and ionising via conventional electrospray ionisation (ESI), has been described for small molecule metabolites produced by a large variety of bacterial and fungal species grown on agar. 6 Finally, liquid extraction surface analysis (LESA), which probes a surface via a liquid interface formed between a conductive pipette tip and the sample surface prior to conventional ESI, has been described for the direct analysis of intact proteins from E. coli, Pseudomonas aeruginosa and Staphylococcus aureus grown on agar 7,8 and for lipid analysis from tissues, 9 and other surfaces. 10 This approach is of particular interest as LESA provides higher sensitivity than the three continuous ablation/extraction methods described above.
Lipids are a diverse class of molecules ranging from polar to apolar species, neutral to basic and acidic; thus different solvent systems are preferentially used to extract different classes of lipids. 11 15 Furthermore, the addition of lithium acetate to ESI solvents has been described to be optimal for lipid dissociation (MS/MS) experiments. 16 Finally, the inclusion of ammonium-based salts in DESI extraction solvents has been shown to provide benefits in purified protein analysis from glass slides. 17 Here we describe suitable LESA extraction solvents for the direct analysis of acyl-phosphatidylinositol mannoside (Acyl-PIM) glycolipids, triacylglycerols and phosphoethanolamine species from bacterial colonies of Mycobacterium smeglatis (M. smegmatis) a bacterial model system for Mycobacterium tuberculosis. We also explore acetate salt addition to LESA extraction solvents for spectral simplification and structural characterisation, aiding accurate mass matching of lipid species directly to online mycobacterial databases. Our approach offers a rapid method for bacterial phenotyping via direct lipid analysis.

| Materials
All salts (lithium acetate, sodium acetate and potassium acetate) were purchased from Sigma-Aldrich (Gillingham, UK). HPLC grade chloroform and methanol were sourced from J. T. Baker (Deventer, The Netherlands). Middlebrook 7H9, Triptic Soy Broth (TSB) and Bacto Agar mycobacterial growth media were also purchased from Sigma-Aldrich.

| Bacterial sample preparation
Mycobacterium Smegmatis mc 2 155 was initially inoculated into 5 mL of TSB media and cultured to an OD 600 of 0.5. Cells were then harvested by centrifugation and resuspended in 5 mL of 7H9 media before 2-μL aliquots were inoculated onto 7H9 agar plates (6 cm, 100 mL agar), supplemented with 1.2% w/v Bacto Agar, using a dropping pipette to ensure that the colonies formed were sufficiently large enough for LESA sampling (~5 mm diameter). Agar plates were incubated at 37°C for 3-4 days and then transferred into a dark fridge environment where the temperature was kept at 5°C for storage.

| LESA sampling
LESA was carried out by use of the Triversa Nanomate (Advion Biosciences, Ithaca, NY, USA) using the advanced user interface feature. The extraction/ionisation solvent was 2:1 chloroform: methanol either with or without the inclusion of 10 mM acetate salt (lithium, sodium or potassium). During extraction, 10 μL of solvent was aspirated from the solvent well, before sampling the bacterial colony with 5 μL of this solvent for 3 s. Finally, 6 μL of the sampling solvent was re-aspirated and infused into the mass spectrometer at a gas pressure of 0.15 psi and a potential of 1.7 kV.

| Mass spectrometry
All data were acquired on an Orbitrap Elite instrument (Thermo Fisher  Lithium adducts do not feature in the database; therefore, assignments made upon lithium adduct addition were searched against a calculated neutral mass after subtracting the most abundant isotopic mass of lithium from the detected m/z value.

| LESA-MS sampling of lipids from bacterial colonies
Direct analysis of bacterial colonies using extraction solvents suitable for lipid analysis presents numerous challenges. Previously, intact protein species have been extracted from bacterial colonies of E. coli, Pseudomonas aeruginosa and Staphylococcus aureus grown on agar 7,8 using a mixture of acetonitrile and water containing formic acid.
However, this was achieved via contact-LESA, by disrupting the colony. This type of analysis has also been demonstrated to allow analysis of surfaces such as tissue using volatile solvent systems that are required for lipid analysis (e.g. mixtures of chloroform, methanol and propanol). 14 The LESA sampling regimes in these studies describe extraction times upwards of 30 s. This is not compatible with liquid-junction sampling of bacterial colonies with the volatile solvents required for lipid extraction as the solvent simply either evaporates or seeps into the colony, and the liquid-junction is lost. An extraction time of 3 s was found to be optimal when using these solvent systems and was used throughout this study. In addition, the inclusion of formic acid in the extraction solvent is not required to achieve high signal intensities (unlike typical intact protein protocols) because the lipid species detected are highly ionisable.
Sampling of the bacterial colonies with chloroform:methanol  Figure S1, supporting information). Figure 1A shows a typical mass Furthermore, peaks at m/z 1257.8 and 1273.8 can be tentatively assigned as the [M + Na] + and [M + K] + adducts, respectively, of the same species (AcPIM1(50:2)) based on accurate mass, see Table 1. In order to address these problems salt additives were introduced into the extraction solvent with the aim of directing dominant adduct formation to reduce spectral complexity and afford greater confidence in assignment by requiring searching against a single adduct type only.

| Salt additive addition to extraction solvents for spectral simplification
The addition of 10 mM lithium, sodium or potassium acetate to the sampling solvent led to a shift in dominant adduct formation to the respective cationic adduct and thus much reduced spectral complexity. Note that only acetate salts were investigated in this study as they are volatile salts that are compatible with the ESI process; other salt forms are less appropriate for electrospray based techniques as they reduce sensitivity. 19 For brevity, trends observed upon salt addition will be described for the most abundant ions only; a full list of detected lipids is provided in Table 1.   However, when lithium acetate was included in the sampling solvent a number of peaks were dominant in the region m/z 1200-1320 that were not assigned via the searches. For example, m/z 1273.8 corresponds to the potassium adduct of the most abundant species in this region (Ac1PIM1(50:2) [M + K] + ). Thus, the inclusion of lithium acetate in the LESA sampling solvent did not provide the same degree of spectral simplification as that described above when either sodium or potassium acetate was incorporated.
The inclusion of salt additives in MALDI matrices has become a commonplace approach for spectral simplification through directing lipid adduct formation. 20 Previous ESI studies have utilised acetate salts to drive adduct formation towards a desired lipid adduct for phosphocholine and triacylglycerol species; 16 however, this methodology has not previously been described for the LESA of lipids directly extracted from biological samples. The inclusion of ammonium salt additives in sampling solvents for DESI-MS analysis of intact proteins has been shown to increase sensitivity. 17 The salt additive approach in this study has enabled assignment (by reducing matches to a single result from a single adduct type and based on accurate mass only) of the detection of 20 different acyl-phosphatidylinositol mannoside species in agreement across three experiments, as shown in Table 2. There is also evidence of cardiolipin species which are important structural cell membrane components that enable cell curvature and represent an attractive drug target. 21 Here we show for the first time that similar approaches can be extended to LESA and for the analysis of biologically relevant bacterial cell wall lipids. It should be noted that certain cardiolipin species and acylphosphatidylinositol mannoside species have masses within a few ppm of one another and therefore cannot be distinguished from one another without structural characterisation.
Although these experiments targeted analysis of acylphosphatidylinositol mannoside lipids (and these were the most abundant species detected in this study) a number of other lipid species were also indicated from the LIPID MAPS database searches.
In the region m/z 700-900 a total of nine phosphatidylethanolamine assignments within 5 ppm accurate mass were suggested by the  Table 2. Furthermore, twenty triacylglycerol species were also suggested. These lipid classes also comprise part of the complex cell wall structure of mycobacterium and present biomolecularly informative ions for bacterial phenotyping and/or monitoring bacterial stress.

| Improved structural information
The addition of salt additives to promote a specific adduct type was also used to investigate the extent of structural information afforded from each adduct type via high-energy collision-induced dissociation (HCD). Further investigations are required to better understand the mechanisms of dissociation of these particular lipid species.

| CONCLUSIONS
Here we show suitable volatile extraction solvents and rapid sampling regimes for the extraction and analysis of lipid species directly from Future work will focus on monitoring bacterial stress via changes in lipid signals upon drug treatments and understanding the data obtained upon HCD fragmentation for the structural characterisation of these complex lipids.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of the article.