蛋白质组学肽段鉴定可信度评价方法

蛋白质组学基于质谱数据鉴定肽段和蛋白质，从而给出基因表达的直接证据，帮助解析蛋白质的结构和功能，研究蛋白质与疾病的关系，提供靶向治疗方案，而这些都取决于鉴定的肽段和蛋白质的准确性。蛋白质组学常采用目标-诱饵库方法（target-decoy approach，TDA）对鉴定的肽段和蛋白质进行质量控制，并对其进行改进演化后应用到子类肽段（比如突变肽段和修饰肽段等）和交联肽段等特殊鉴定结果的可信度评价中。然而，TDA存在两个局限，即错误率估计值不够准确以及不能评价单个鉴定结果的可信度，经过TDA质量控制后的结果还需要进一步检验，因此领域内也提出了一系列其他方法（本文统称为Beyond-TDA方法），协同加强肽段的可信度评价。本文对数据依赖模式下采集的质谱数据肽段层面的TDA常规方法和特殊方法进行了综述，对Beyond-TDA方法进行了分类阐述，并总结了各种方法的优势与不足。     

Integrated Identification and Quantification Error Probabilities for Shotgun Proteomics

1. Download : Download high-res image (96KB)
2. Download : Download full-size image

Highlights

• •Unified identification and quantification error rates for protein quantification.
• •Error propagation using graphical models and Bayesian statistics.
• •Account for uncertainty of missing values instead of overconfident point estimates.
• •Control of differential expression false discovery rate at increased sensitivity.

     

Identification and Validation of Novel Spinophilin-associated Proteins in Rodent Striatum Using an Enhanced ex Vivo Shotgun Proteomics Approach

Spinophilin regulates excitatory postsynaptic function and morphology during development by virtue of its interactions with filamentous actin, protein phosphatase 1, and a plethora of additional signaling proteins. To provide insight into the roles of spinophilin in mature brain, we characterized the spinophilin interactome in subcellular fractions solubilized from adult rodent striatum by using a shotgun proteomics approach to identify proteins in spinophilin immune complexes. Initial analyses of samples generated using a mouse spinophilin antibody detected 23 proteins that were not present in an IgG control sample; however, 12 of these proteins were detected in complexes isolated from spinophilin knock-out tissue. A second screen using two different spinophilin antibodies and either knock-out or IgG controls identified a total of 125 proteins. The probability of each protein being specifically associated with spinophilin in each sample was calculated, and proteins were ranked according to a χ² analysis of the probabilities from analyses of multiple samples. Spinophilin and the known associated proteins neurabin and multiple isoforms of protein phosphatase 1 were specifically detected. Multiple, novel, spinophilin-associated proteins (myosin Va, calcium/calmodulin-dependent protein kinase II, neurofilament light polypeptide, postsynaptic density 95, α-actinin, and densin) were then shown to interact with GST fusion proteins containing fragments of spinophilin. Additional biochemical and transfected cell imaging studies showed that α-actinin and densin directly interact with residues 151–300 and 446–817, respectively, of spinophilin. Taken together, we have developed a multi-antibody, shotgun proteomics approach to characterize protein interactomes in native tissues, delineating the importance of knock-out tissue controls and providing novel insights into the nature and function of the spinophilin interactome in mature striatum.Genomic sequencing has revealed the full repertoire of ∼20,000 proteins that can be expressed in most mammals. Innate biochemical or enzymatic activities of many proteins are critical to their function, but these activities are often modified by interactions with other proteins. Moreover, many proteins have no known catalytic activity and are thought to serve structural roles in assembling protein complexes, greatly increasing the efficiency and fidelity of intracellular processes. Thus, systematic definition of protein interactomes promises tremendous insight into biochemical mechanisms underlying the functions of many proteins.A prime example of the importance of protein-protein interactions for modifying biological function is the postsynaptic density (PSD),¹ an actin-rich organelle localized to neuronal dendritic spines that contains receptors, kinases, phosphatases, and scaffolding proteins (1, 2). Dynamic changes in enzymatic activities and protein-protein interactions underlie changes in the size and shape of both PSDs and dendritic spines as well as the modulation of PSD-targeted neurotransmitter receptors that are critical for synaptic plasticity, learning, and memory. Furthermore, dendritic spine morphology and number are altered in many neurological disorders, including Parkinson disease (PD), Angelman syndrome, and fragile X syndrome (3–7).Spinophilin (neurabin II) is an F-actin- and protein phosphatase 1 (PP1)-binding protein with no known catalytic function (8–10). It is highly expressed in brain and is localized to dendritic spines and PSDs where it plays a key role targeting PP1 to regulate synaptic plasticity, learning, and memory (11–14). Spinophilin associates with its homolog neurabin, which is also a PP1- and F-actin-binding protein that regulates synaptic plasticity and dendrite morphology (14–16). The interaction between spinophilin and the γ1 isoform of PP1 is enhanced in an animal model of PD (17), perhaps contributing to the altered phosphorylation of synaptic proteins, such as CaMKII and glutamate receptor subunits observed following dopamine (DA) depletion (18–20). DA depletion also decreases the number of dendritic spines on striatal medium spiny neurons (4, 5). Spine density is regulated by dynamic changes in the F-actin cytoskeleton, and spinophilin regulates dendritic spine density during development (21). Indeed, candidate protein or generic protein-protein interaction screens have identified many additional spinophilin-associated proteins (SpAPs) that modulate F-actin dynamics and/or cell morphology (22–27; for a review, see Ref. 28), consistent with the idea that spinophilin is an archetypical scaffolding protein. However, these interactions have mostly been characterized in vitro and/or following protein overexpression in cultured cells, and the inter-relationship of these interactions in vivo is largely unknown. Although the spinophilin interactome appears to dictate the biological roles of spinophilin, the composition of these complexes in the mature brain is poorly understood.Co-immunoprecipitation is commonly used to confirm the biological relevance of specific bivalent protein-protein interactions in native tissues that were initially identified using generic molecular approaches, such as yeast two-hybrid screening. Prior studies combined this approach with mass spectrometry-based proteomics methods to more broadly characterize the composition of mammalian signaling complexes and the PSD interactome, such as the signalosome associated with synaptic N-methyl-d-aspartate receptors (29) and complexes associated with other PSD-enriched proteins (30). In addition, proteomics methodologies were used to identify over 1100 protein components of the PSD (30). Indeed, the potential for shotgun proteomics studies to provide novel insights into protein function in the brain is increasingly recognized (31). Moreover, computational approaches are being developed to identify potential protein-protein interactions (32). However, validation of specific interactions among the very large data sets of candidates typically identified using these approaches can be daunting. In addition, most proteomics analyses have relied on a single antibody to the target protein of interest with, at best, an unrelated non-immune IgG as a negative control, necessitating the use of very high quality antibodies.We developed a systematic shotgun proteomics approach to define protein interactomes in a native tissue context. We used this approach to characterize the composition of spinophilin complexes isolated from rodent striatum and confirmed the association of multiple, novel SpAPs. Furthermore, we extensively characterized the interaction of two additional SpAPs, α-actinin and densin, using biochemical and imaging techniques. Our studies directly illustrate the importance of appropriate subcellular fractionation conditions, using multiple antibodies to the protein of interest, and the underappreciated, critical role of analyzing parallel samples prepared from knock-out (KO) animals. Thus, our findings demonstrate a methodological framework with key controls that can be broadly applied to characterizing protein interactomes, in addition to providing novel insights into the role of spinophilin in controlling synaptic signaling.     

Scavager: A Versatile Postsearch Validation Algorithm for Shotgun Proteomics Based on Gradient Boosting

Shotgun proteomics workflows for database protein identification typically include a combination of search engines and postsearch validation software based mostly on machine learning algorithms. Here, a new postsearch validation tool called Scavager employing CatBoost, an open‐source gradient boosting library, which shows improved efficiency compared with the other popular algorithms, such as Percolator, PeptideProphet, and Q‐ranker, is presented. The comparison is done using multiple data sets and search engines, including MSGF+, MSFragger, X!Tandem, Comet, and recently introduced IdentiPy. Implemented in Python programming language, Scavager is open‐source and freely available at https://bitbucket.org/markmipt/scavager .     

MaxQuant Software for Ion Mobility Enhanced Shotgun Proteomics

1. Download : Download high-res image (140KB)
2. Download : Download full-size image

Highlights

• •Highly parallelizable 4D feature detection in ion mobility enhanced shotgun proteomics.
• •Multidimensional non-linear mass, retention time and ion mobility recalibration.
• •Collision cross section aware matching between runs.
• •Label-free quantification of ion mobility MS data.

     

Elastase Digests: New Ammunition for Shotgun Membrane Proteomics

Despite many advances in membrane proteomics during the last decade the fundamental problem of accessing the transmembrane regions itself has only been addressed to some extent. The present study establishes a method for the nano-LC-based analysis of complex membrane proteomes on the basis of a methanolic porcine pancreatic elastase digest to increase transmembrane coverage. Halobacterium salinarium purple and Corynebacterium glutamicum membranes were successfully analyzed by using the new protocol. We demonstrated that elastase digests yield a large proportion of transmembrane peptides, facilitating membrane protein identification. The potential for characterization of a membrane protein through full sequence coverage using elastase is there but is restricted to the higher abundance protein components. Compatibility of the work flow with the two most common mass spectrometric ionization techniques, ESI and MALDI, was shown. Currently better results are obtained using ESI mainly because of the low response of MALDI for strictly neutral peptides. New findings concerning elastase specificity in complex protein mixtures reveal a new prospect beyond the application in shotgun experiments. Furthermore peptide mass fingerprinting with less specific enzymes might be done in the near future but requires an adaptation of current search algorithms to the new proteases.Upon the introduction of modern mass spectrometric ionization techniques, such as MALDI (1) and ESI (2), extremely powerful and valuable tools were given to researchers for the identification and characterization of proteins. Nevertheless the intricate analysis of membrane subproteomes still represents one of the major challenges despite the amount of “success” reported in the literature. Until now, there was no general protocol available to address membrane proteomes as a whole, demonstrating their degree of difficulty and complexity.During the past years, two different strategies in proteome analysis have evolved: the more widespread bottom-up and top-down proteomics. The latter approach has been shown to provide access to the transmembranal regions of membrane proteins and has the power to characterize complete protein primary structures including labile covalent modifications (3, 4). Nevertheless there are limitations due to sample complexity, emphasizing the need for a liquid chromatographic separation on the protein level, which is comparable to LC peptide separation. Therefore, the bottom-up variant is the most commonly used method for the proteomics analysis of complex membrane samples. Improvements in the bottom-up work flow were achieved by adapting sample cleanup and prefractionation processes (5–10) and subsequently by the development of modified and optimized separation techniques. The two main work flows, the gel-based approach mainly carried out with 2D¹ SDS-PAGE (11, 12) and the shotgun identification of proteolytically digested protein mixtures and their multidimensional separation via liquid chromatography (13), had to be adapted. The separation of proteins via classical 2D SDS-PAGE is only possible up to a GRAVY score of ∼0.4 (14, 15). Despite lacking the separation power of the IEF-SDS-PAGE system, derived techniques like the doubled SDS- (16) and 16-benzyldimethyl-n-hexadecylammonium chloride/SDS-PAGE (17) represent an improvement for hydrophobic proteins. Advances in the nLC separation of hydrophobic peptides, e.g. the separation at elevated temperatures (18) and the use of LC-compatible detergents (19), yielded significant success. When combining both prominent separation techniques, the one-dimensional SDS-LC analysis proved to be more effective than 2D PAGE as well (20, 21).One of the remaining steps still offering room for improvements is the proteolytic procedure. The original multidimensional protein identification technique using trypsin has been successfully improved by the application of proteinase K under high pH conditions, yielding an increased number of membrane protein identifications (22). Further modification of this method by recleaving the isolated membranous parts with cyanogen bromide increased the accessibility to membrane-spanning peptides (18).Despite the suitability of proteinase K for the shotgun analysis of membrane proteomes, trypsin is the prevalent enzyme choice in most current proteomics approaches because of its very specific cleavage behavior. Calculations have implied that alternative proteases are preferentially suited for the analysis of membrane proteomes (23). Proteases other than trypsin have been utilized only to a small degree and mostly for the targeted analysis of protein complexes. Pepsin, for example, was used for the characterization of an aquaporin (24), and elastase and subtilisin were used as proteases in a triple digest approach of protein complexes and lens tissue (25). Further improvements in the accessibility of the membrane proteome have been shown for tryptic (26, 27) and tryptic/chymotryptic (28) digests by performing the proteolytic treatment in the presence of methanol.During membrane proteomics method development, purple membranes from Halobacterium sp. NRC-1, consisting to a large extent of the H⁺-ATPase bacteriorhodopsin, have widely been used as reference in a variety of cases. Full BR sequence coverage has been reported by several publications using different techniques (27, 29). Such exorbitant sequence coverage amounts are only achievable for the most abundant proteins within a complex mixture. The identification of less expressed proteins, which certainly are more in the researcher''s interest, occurs via lower peptide numbers (27).Generally the basic tryptic cleavage sites are predominantly located in the loops facing the cytoplasm and to a lesser extent in the extracellular or periplasmic loops but are very scarce within the transmembrane helix stretches. This mostly limits the potential tryptic fragments to loop components and larger peptides containing at least one TM helix. The use of another protease, preferentially cleaving after neutral aliphatic residues, should not succumb to this lack of cleavage sites.One of the previously mentioned proteases, porcine pancreatic elastase, was reported to possess potential cleavage specificity at the carboxyl-terminal side of small neutral amino acids (30). It has been used previously for the examination of single protein phosphorylations (31) but was described not to be suited for the cleavage of complex membrane-containing samples because of a supposedly limited activity when applied to them (22). Despite this finding, elastase has been successfully utilized in a mass spectrometric analysis as early as 1974 when it was regarded as an ideal protease for future mass spectrometric studies (32). At the same time, most of the experiments to characterize elastase by analyzing its S₁ pocket (for nomenclature, see Ref. 33) binding capabilities (34) as well as cleavage preferences of ester and protein substrates (35–39) were performed.Based on previous research concerning elastase, we set up an nLC-based membrane proteomics analysis. This large data set was used to characterize the protease behavior of elastase and the physicochemical properties of the detected peptides. The PM model was used to establish a method to analyze more complex Corynebacterium glutamicum membranes, which have been analyzed previously using different 2D techniques (7, 28). We demonstrated that a combination of elastase and methanol is particularly suitable for an nLC-based membrane proteome analysis and results in a significantly increased number of TM peptides. Additionally the promising PMF application of elastase digests is discussed.     

Bayesian Confidence Intervals for Multiplexed Proteomics Integrate Ion-statistics with Peptide Quantification Concordance,

1. Download : Download high-res image (74KB)
2. Download : Download full-size image

Highlights

• •Bayesian Beta-Binomial model integrates ion statistics with peptide ratio agreement.
• •Model appropriately interprets information from low signal peptides.
• •Confidence can be assigned even without replicates.
• •Model adds sensitivity to detection of small changes.

     

Computing Exact p-values for a Cross-correlation Shotgun Proteomics Score Function

The core of every protein mass spectrometry analysis pipeline is a function that assesses the quality of a match between an observed spectrum and a candidate peptide. We describe a procedure for computing exact p-values for the oldest and still widely used score function, SEQUEST XCorr. The procedure uses dynamic programming to enumerate efficiently the full distribution of scores for all possible peptides whose masses are close to that of the spectrum precursor mass. Ranking identified spectra by p-value rather than XCorr significantly reduces variance because of spectrum-specific effects on the score. In combination with the Percolator postprocessor, the XCorr p-value yields more spectrum and peptide identifications at a fixed false discovery rate than Mascot, X!Tandem, Comet, and MS-GF+ across a variety of data sets.A high-throughput proteomics experiment generates many thousands of candidate hypotheses, only a fraction of which are true and an even smaller fraction of which are of significant biological interest. Consequently, the accurate and efficient assignment of statistical confidence estimates to identified fragmentation spectra is critical to making efficient use of shotgun proteomics data sets.Historically, the field has shifted from focusing on discrimination—the ability of a search engine to distinguish between correct and incorrect spectrum identifications—to calibration. If a score function is well-calibrated, then the score of x assigned to one spectrum is directly comparable to a score of x assigned to a different spectrum. A well-known example of poor calibration is the distribution of SEQUEST XCorr scores produced on spectra of varying charges (Fig. 1A), such that a score of 1.8 for a doubly charged (2+) spectrum indicates a good quality identification, whereas the same score assigned to a 3+ spectrum corresponds to a much poorer match. Improving the calibration of a given score function across spectra can lead to large improvements in the number of identified spectra at a fixed statistical confidence threshold. Score calibration can be carried out using empirical curve fitting procedures to estimate p-values (1, 2, 3) or, more recently, using dynamic programming to calculate an exact p-value for each observed score (4). Machine learning postprocessors such as PeptideProphet (5) and Percolator (6) simultaneously calibrate scores and incorporate additional information, leading to even larger improvements in identification rates.Open in a separate windowFig. 1.Distribution of scores for charge 2+ and charge 3+ spectra from yeast data set. A, Standard XCorr scores. B, XCorr Šidák-corrected p-values.In this work, we describe a dynamic programming method for computing exact p-values for the oldest and one of the most widely used score functions, SEQUEST XCorr (7, 8). We demonstrate analytically and empirically that the resulting p-values are valid relative to a widely accepted null model. Furthermore, we show that, across a variety of data sets, our XCorr p-value yields significantly improved statistical power relative to competing, state-of-the-art methods, including SEQUEST, Mascot (9), X!Tandem (10), and Comet (3), and is competitive with other dynamic programming-based calibration methods like MS-GF+ (11). Strikingly, the improved calibration from our scoring scheme is complementary to that provided by Percolator, so that the combination of the two methods yields even better results, evaluated both at the spectrum and peptide levels.     

Temporal Analyses of Barley Malting Stages Using Shotgun Proteomics

Malt derived from barley malting is an essential raw material for beer brewing. In this study, we performed the first dynamic proteome survey during barley malting using a gel‐free proteomics approach. This entailed in‐solution tryptic digestion of precipitated proteins and analysis of peptides by nanoliquid chromatography coupled with tandem mass spectrometry. A total of 1418 proteins were identified from the five malting stages: Steep, 1, 3, 5 days after germination, and end of Kiln. About 900 proteins identified in this analysis were uncharacterized or predicted proteins. Integrating information from Uniprot90, Uniprot50, Pfam, Interpro databases and gene ontologies from EnsemblPlants, 796 of the predicted and uncharacterized proteins were provided functional annotations. Nearly 63% of the identified proteins were present during all the five time points suggesting a coordinated activation of major metabolic pathways during malting. GO enrichment analysis showed over‐representation of proteins associated with translation, carbohydrate metabolism, and stress response. Analysis of variance of the spectral counts of proteins present in all the five malting stages identified 205 proteins with significant differences in their abundance. Proteins associated with carbohydrate metabolism especially enzyme activity regulation provide novel targets for malting barley breeding and for predicting malting quality.     

Shotgun Redox Proteomics: Identification and Quantitation of Carbonylated Proteins in the UVB-Resistant Marine Bacterium,Photobacterium angustum S14

UVB oxidizes proteins through the generation of reactive oxygen species. One consequence of UVB irradiation is carbonylation, the irreversible formation of a carbonyl group on proline, lysine, arginine or threonine residues. In this study, redox proteomics was performed to identify carbonylated proteins in the UVB resistant marine bacterium Photobacterium angustum. Mass-spectrometry was performed with either biotin-labeled or dinitrophenylhydrazide (DNPH) derivatized proteins. The DNPH redox proteomics method enabled the identification of 62 carbonylated proteins (5% of 1221 identified proteins) in cells exposed to UVB or darkness. Eleven carbonylated proteins were quantified and the UVB/dark abundance ratio was determined at both the protein and peptide levels. As a result we determined which functional classes of proteins were carbonylated, which residues were preferentially modified, and what the implications of the carbonylation were for protein function. As the first large scale, shotgun redox proteomics analysis examining carbonylation to be performed on bacteria, our study provides a new level of understanding about the effects of UVB on cellular proteins, and provides a methodology for advancing studies in other biological systems.     

Bayesian Nonparametric Modeling for Comparison of Single-Neuron Firing Intensities

Summary .  We propose a fully inferential model-based approach to the problem of comparing the firing patterns of a neuron recorded under two distinct experimental conditions. The methodology is based on nonhomogeneous Poisson process models for the firing times of each condition with flexible nonparametric mixture prior models for the corresponding intensity functions. We demonstrate posterior inferences from a global analysis, which may be used to compare the two conditions over the entire experimental time window, as well as from a pointwise analysis at selected time points to detect local deviations of firing patterns from one condition to another. We apply our method on two neurons recorded from the primary motor cortex area of a monkey's brain while performing a sequence of reaching tasks.     

Bayesian Inference for Gene Expression and Proteomics.

A text that has a systematic account of Bayesian analysis incomputational biology has been needed for a long time. Thisbook is a timely publication entirely devoted to cutting-edgeBayesian methods in genomics and proteomics research and manyof its contributors are leading authorities in the field. Itis thus an indispensable reference for researchers who are interestedin applying Bayesian techniques in their own biological research.Moreover, the book calls for more methodological and theoreticalresearch to     

A Bayesian Nonparametric Approach for Mapping Dynamic Quantitative Traits

In biology, many quantitative traits are dynamic in nature. They can often be described by some smooth functions or curves. A joint analysis of all the repeated measurements of the dynamic traits by functional quantitative trait loci (QTL) mapping methods has the benefits to (1) understand the genetic control of the whole dynamic process of the quantitative traits and (2) improve the statistical power to detect QTL. One crucial issue in functional QTL mapping is how to correctly describe the smoothness of trajectories of functional valued traits. We develop an efficient Bayesian nonparametric multiple-loci procedure for mapping dynamic traits. The method uses the Bayesian P-splines with (nonparametric) B-spline bases to specify the functional form of a QTL trajectory and a random walk prior to automatically determine its degree of smoothness. An efficient deterministic variational Bayes algorithm is used to implement both (1) the search of an optimal subset of QTL among large marker panels and (2) estimation of the genetic effects of the selected QTL changing over time. Our method can be fast even on some large-scale data sets. The advantages of our method are illustrated on both simulated and real data sets.     

目标?鄄诱饵库搜索策略在蛋白质组质谱鉴定和质控中的应用及研究进展

基于串联质谱的蛋白质组研究会产生海量的质谱数据,这些数据通常使用数据库搜索引擎进行鉴定分析,并根据肽段谱图匹配(PSM)反推真实的样品蛋白质．对于高通量蛋白质组数据的处理,其鉴定结果的可信是后续分析应用的前提,因此对鉴定结果的质量控制尤为重要,而基于目标-诱饵库(target-decoy)搜索策略的质量控制是目前应用最为广泛的方法．本文首先介绍了基于目标-诱饵库搜索策略搜库和质量控制的实施流程,然后综述了基于目标-诱饵库搜索策略的质量控制工具,并提出了目标-诱饵库搜索策略的不足及改善方法,最后对目标-诱饵库搜索策略进行了总结与展望．     

Identification and Validation of Specific Markers of Bacillus anthracis Spores by Proteomics and Genomics Approaches

Bacillus anthracis is the causative bacteria of anthrax, an acute and often fatal disease in humans. The infectious agent, the spore, represents a real bioterrorism threat and its specific identification is crucial. However, because of the high genomic relatedness within the Bacillus cereus group, it is still a real challenge to identify B. anthracis spores confidently. Mass spectrometry-based tools represent a powerful approach to the efficient discovery and identification of such protein markers. Here we undertook comparative proteomics analyses of Bacillus anthracis, cereus and thuringiensis spores to identify proteoforms unique to B. anthracis. The marker discovery pipeline developed combined peptide- and protein-centric approaches using liquid chromatography coupled to tandem mass spectrometry experiments using a high resolution/high mass accuracy LTQ-Orbitrap instrument. By combining these data with those from complementary bioinformatics approaches, we were able to highlight a dozen novel proteins consistently observed across all the investigated B. anthracis spores while being absent in B. cereus/thuringiensis spores. To further demonstrate the relevance of these markers and their strict specificity to B. anthracis, the number of strains studied was extended to 55, by including closely related strains such as B. thuringiensis 9727, and above all the B. cereus biovar anthracis CI, CA strains that possess pXO1- and pXO2-like plasmids. Under these conditions, the combination of proteomics and genomics approaches confirms the pertinence of 11 markers. Genes encoding these 11 markers are located on the chromosome, which provides additional targets complementary to the commonly used plasmid-encoded markers. Last but not least, we also report the development of a targeted liquid chromatography coupled to tandem mass spectrometry method involving the selection reaction monitoring mode for the monitoring of the 4 most suitable protein markers. Within a proof-of-concept study, we demonstrate the value of this approach for the further high throughput and specific detection of B. anthracis spores within complex samples.Bacillus anthracis is a highly virulent bacterium, which is the etiologic agent of anthrax, an acute and often lethal disease of animals and humans (1). According to the Centers for Disease Control and Prevention, B. anthracis is classified as a category A agent, the highest rank of potential bioterrorism agents (http://www.bt.cdc.gov/agent/agentlist-category.asp). The infectious agent of anthrax, the spore, was used as a bioterrorism weapon in 2001 in the United States when mailed letters containing B. anthracis spores caused 22 cases of inhalational and/or cutaneous anthrax, five of which were lethal (2). These events have emphasized the need for rapid and accurate detection of B. anthracis spores.Bacillus anthracis is a member of the genus Bacillus, Gram-positive, rod-shaped bacteria characterized by the ability to form endospores under aerobic or facultative anaerobic conditions (3). The genus Bacillus is a widely heterogeneous group encompassing 268 validly described species to date (http://www.bacterio.net/b/bacillus.html, last accessed on August 9^th 2013). B. anthracis is part of the B. cereus group which consists of six distinct species: B. anthracis, B. cereus, B. thuringiensis, B. mycoides, B. pseudomycoides, and B. weihenstephanensis (4, 5). The latter three species are generally regarded as nonpathogenic whereas B. cereus and B. thuringiensis could be opportunistic or pathogenic to mammals or insects (5, 6). B. cereus is a ubiquitous species that lives in soil but is also found in foods of plant and animal origin, such as dairy products (7). Its occurrence has also been linked to food poisoning and it can cause diarrhea and vomiting (6, 8). B. thuringiensis is primarily an insect pathogen, also present in soil, and often used as a biopesticide (9).B. anthracis is highly monomorphic, that is, shows little genetic variation (10), and primarily exists in the environment as a highly stable, dormant spore in the soil (1). Specific identification of B. anthracis is challenging because of its high genetic similarity (sequence similarity >99%) with B. cereus and B. thuringiensis (5, 11). The fact that these closely related species are rather omnipresent in the environment further complicates identification of B. anthracis. The main difference among these three species is the presence in B. anthracis of the two virulence plasmids pXO1 and pXO2 (1), which are responsible for its pathogenicity. pXO1 encodes a tripartite toxin (protective antigen (PA), lethal factor (LF), and edema factor (EF)) which causes edema and death (1), whereas pXO2 encodes a poly-γ-d-glutamate capsule which protects the organism from phagocytosis (1). B. anthracis identification often relies on the detection of the genes encoded by these two plasmids via nucleic acid-based assays (12–14). Nevertheless, the occasionally observed loss of the pXO2 plasmid within environmental species may impair the robustness of detection (1). In addition, in recent years a series of findings has shown that the presence of pXO1 and pXO2 is not a unique feature of B. anthracis. Indeed, Hu et al. have demonstrated that ∼7% of B. cereus/B. thuringiensis species can have a pXO1-like plasmid and ∼1.5% a pXO2-like plasmid (15). This was particularly underlined for some virulent B. cereus strains (i.e. B. cereus strains G9241, B. cereus biovar anthracis strains CA and CI) (16–20).Because of these potential drawbacks, the use of chromosome-encoded genes would be preferable for the specific detection of B. anthracis. Such genes (rpoB, gyrA, gyrB, plcR, BA5345, and BA813) have been reported as potential markers (21–25), but concerns have also been raised about their ability to discriminate B. anthracis efficiently from closely related B. cereus strains (26). Ahmod et al. have recently pointed out, by in silico database analysis, that a specific sequence deletion (indel) occurs in the yeaC gene and exploited it for the specific identification of B. anthracis (27). Nevertheless, a few B. anthracis strains (e.g. B. anthracis A1055) do not have this specific deletion and so may lead to false-negative results (27).In the last few years, protein profiling by MS, essentially based on matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF MS), has emerged as an alternative (or a complement) to genotypic or phenotypic methods for the fast and efficient identification of microorganisms (28, 29). Such an approach is based on the reproducible acquisition of global bacterial protein fingerprints/patterns. The combination of MS-based protein patterns and chemometric/bioinformatic tools has been demonstrated to efficiently differentiate members of the B. cereus group from other Bacillus species (30). However, the specific discrimination of B. anthracis from the closely related B. cereus and B. thuringiensis remains difficult (30). This study of Lasch and coworkers, performed on vegetative cells, identified a few ribosomal and spore proteins as being responsible for this clustering (30). Closer inspection of the data revealed that B. anthracis identification was essentially based on one particular isoform of the small acid-soluble spore protein B (SASP-B)¹ (30–34), which is exclusively expressed in spores, as the samples were shown to contain residual spores. However, the specificity of SASP-B has recently been questioned as the published genomes of B. cereus biovar anthracis CI and B. thuringiensis BGSC 4CC1 strains have been shown to share the same SASP-B isoform as B. anthracis (35). Altogether these results underline that the quest for specific markers of B. anthracis needs to be pursued.Mass spectrometry also represents a powerful tool for the discovery and identification of protein markers (36, 37). In the case of B. anthracis, this approach has more commonly been used for the comprehensive characterization of given bacterial proteomes. For example, the proteome of vegetative cells with variable plasmid contents has been extensively studied (38–40), as the proteomes of mature spores (41, 42) and of germinating spores (43, 44). Only one recent study, based on a proteo-genomic approach, was initiated to identify protein markers of B. anthracis (45). In this work, potential markers were characterized but using a very limited number of B. cereus group strains (three B. cereus and two B. thuringiensis). Moreover, this study was done on vegetative cells, whereas the spore proteome is drastically different. To our knowledge, no study has characterized and validated relevant protein markers specific to B. anthracis spores, which constitute the dissemination form of B. anthracis and are often targeted by first-line immunodetection methods (46).Here we report comparative proteomics analyses of Bacillus anthracis/cereus/thuringiensis spores, undertaken to identify proteoforms unique to B. anthracis. Preliminary identification was performed on a limited set of Bacillus species both at the peptide (after enzymatic digestion) and protein levels by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) using a high resolution/high mass accuracy LTQ-Orbitrap instrument. The pertinence of 11 markers was further demonstrated using proteomics and genomics approaches on a representative larger set of up to 55 different strains, including the closely related B. cereus biovar anthracis CI, CA, and B. thuringiensis 9727. Lastly, as a proof-of-concept study, we also report for four B. anthracis markers the implementation of a targeted LC-MS/MS method using selected reaction monitoring (SRM), based on the extension of a previous one focused on SASP-B (35). Preliminary results regarding method usefulness for the high throughput and accurate detection of B. anthracis spores in complex samples were also obtained and will be reported herein.