Automated Lipid A Structure Assignment from Hierar

Authors: Ying S Ting Scott A Shaffer Jace W Jones Wailap V Ng Robert K Ernst David R Goodlett

Publish Date: 2011/03/05

Volume: 22, Issue: 5, Pages: 856-866

PDF Link

Abstract

Infusionbased electrospray ionization ESI coupled to multiplestage tandem mass spectrometry MS n is a standard methodology for investigating lipid A structural diversity Shaffer et al J Am Soc Mass Spectrom 186 1080–1092 2007 Annotation of these MS n spectra however has remained a manual expertdriven process In order to keep up with the data acquisition rates of modern instruments we devised a computational method to annotate lipid A MS n spectra rapidly and automatically which we refer to as hierarchical tandem mass spectrometry HiTMS algorithm As a firstpass tool HiTMS aids expert interpretation of lipid A MS n data by providing the analyst with a set of candidate structures that may then be confirmed or rejected HiTMS deciphers the signature ions eg A Y and Ztype ions and neutral losses of MS n spectra using a speciesspecific library based on general prior structural knowledge of the given lipid A species under investigation Candidates are selected by calculating the correlation between theoretical and acquired MS n spectra At a false discovery rate of less than 001 HiTMS correctly assigned 85 of the structures in a library of 133 manually annotated Francisella tularensis subspecies novicida lipid A structures Additionally HiTMS correctly assigned 85 of the structures in a smaller library of lipid A species from Yersinia pestis demonstrating that it may be used across speciesLipid A the endotoxic portion of lipopolysaccharides LPS responsible for gramnegative bacterial virulence is imbedded in the outer leaflet of the gramnegative bacterial outer membrane As an essential component of gramnegative bacterial membranes lipid A exhibits speciesspecific structural diversity 1 The general structure consists of a backbone of two glucosamine residues present as a β16linked dimer 2 This backbone can be diversified in response to specific environmental signals or between bacterial species Specifically changes in the fatty acid content varying both in the length and number of fatty acid side chains eg tetraacylated or hexaacylated and phosphorylation patterns Supplemental Figure S1 3 Additional glycan modifications of the phosphate residues by monosaccharides such as aminoarabinose or galactosamine are also observed 2 4 Environmentally induced changes to lipid A diversity is not only observed among species but also within species Supplemental Figure S1 Variability in lipid A structures is an adaptive mechanism that increases bacterial survival often increasing resistance to host killing mechanism or in the avoidance of the host innate immune system 5 6 The diversity of such species and environmentaldriven structural modification makes complete lipid A structural analysis challengingOverview of HiTMS interpretation of lipid A hierarchical ESIMS n data Bacterial lipid A is isolated from LPS extraction and analyzed by ESI tandem mass spectrometry with hierarchical MS n strategy that acquires tandem mass spectra on each precursor ion and all of the derived fragment ions The collection of MS n spectra is searched against the theoretical signature ion TSI database for observed signature ions The neutral losses of signature ions in each spectrum are then searched against the theoretical neutral losses TNL database to identify dissociation formulae Lipid A preliminary structures for each MS n spectral set are then proposed Every assignment of preliminary structures is given a Xscore based on the correlation between theoretical and acquired spectra All candidate structures that pass the Xscore cutoff are considered as accurate assignmentsA thorough review of the literature revealed a number of software tools available for lipid structure assignment however none of these tools are compatible with lipid MS n data analysis Some of these tools are limited to the type of instrument used to obtain the data such as Fatty Acid Analysis Tool FAAT Lipid Profiler and Lipid Inspector 13 14 15 16 Some are limited to the lipid classes covered by their fragment ion databases such as LipidQA 17 Another software tool lipID supported more comprehensive lipid classes but unfortunately lacked the ability to analyze MS n data 18 In 2009 AMDMSSL was developed to automate the analyses of multidimensional mass spectrometry MDMSbased shotgun lipidomics 19 20 AMDMSSL was able to identify individual lipid molecular species within a predetermined individual lipid class 21 However the general structure of lipid A is very different from the majority of lipid classes eg glycerolipids glycerophospholipids and sphingolipids which were well characterized by AMDMSSL builtin database In addition the initial analysis of structures by tandem mass spectrometry alone is often not sufficient to decipher the structures of lipid A which we have found often require exhaustive MS n analysis to successfully characterize the isobaric components 7 10 11 22 23Thus while there are a number of computational tools available for the analysis of ESI generated data of lipids in general none deal with the structural diversity of lipid A specifically and none have the ability to analyze HTP hierarchical MS n mass spectra As mentioned previously recent work from our laboratory demonstrated the use an infusionbased HTP ESIMS n strategy to characterize lipid A structural diversity in Fn 7 In this study 30 lipid A structures were determined by manual spectral interpretation for two growth conditions 25 °C and 37 °C The extensive tandem mass spectral library generated in this prior study was used as the initial training dataset for the development of an automated structure assignment tool that we term hierarchical tandem mass spectrometry HiTMS algorithm Data analysis by HiTMS is based on predicted ion dissociations and complementary neutral losses including fatty acids phosphate and monosaccharide substituents that are used to construct a speciesspecific library Here we describe HiTMS as well as use of a crosscorrelation scoring routine to assign individual lipid A structures objectively from complex mixtures of Fn and Yersinia pestis YpFrancisella tularensis subspecies novicida Fn strain U112 was grown with aeration in tryptic soy broth Gibco BRL Grand Island NY USA supplemented with 01 cysteine at 25 °C and harvested in the stationary phase Yp strain KIM6+ was grown in Luria broth pH 74 at 37 °C with aeration and harvested in the late exponential phase referred to as Yp wild type Yp WT 24 Lipid A C1 and C4 phosphatase LpxE and LpxF respectively have been expression cloned in Fn 25 The individual plasmids with the structural genes of LpxE or LpxF and an ampicillin resistance gene were incorporated into KIM6+ cell via electroporation 11 The phosphatase expressing strains were then grown in Luria broth containing 100 μg/mL ampicillin pH 74 at 37 °C with aeration and harvested in the late exponential phase and designated as Yp LpxE and Yp LpxF Fn and Yp LPS were extracted using the hot phenol/water extraction method as previously described 26 Lipid A was isolated after LPS was treated with RNase A DNase I and proteinase K by the method of Caroff et al 27The isolated Fn lipid A was analyzed by electrospray ionization ESI in the negative ion mode on a hybrid linear ion trap Fourier transform ion cyclotron resonance FTICR mass spectrometer LTQFT Thermo Scientific San Jose CA USA Lipid A was prepared at ~05 mg/mL in methanol/chloroform 21 and infused at 10 μL/min into a heated capillary inlet maintained at 400–450 °C MS n spectra were acquired according to a “target” MS scheme predetermined from previous studies 7 Briefly in this scheme 15 deprotonated molecular ions were selected individually for MS2 for the initial loss of 120 number of carbonsnumber of double bonds 140 160 180 and 200 fatty acids each of which were determined previously to be esterified through the 2 position fatty acid at the 3hydroxy position of the lipid A deprotonated anions Each of the subsequent ions was selected for MS3 for the combined loss of galactosamine and the 3position 3hydroxy fatty acid 120 140 160 180 and 200 Ions representing these combined losses were in turn selected for MS4 and monitored for the observation of Y1/Z1 ion pairs using 1 min scan averaging Each MS2 and MS3 “channel” was selected regardless of product ion spectra observed 28 and only the MS4 were used for the determination of the individual lipid A structures Ion population in the LTQ was set at 10000 and collision energies employed for MS n ranged from 25–35 For Yp Lipid A data were acquired on an LTQFT Ultra Thermo Scientific as described elsewhere 10 11

Keywords: