Coordination of Metabolism and Virulence Factors Expression of Extraintestinal Pathogenic Escherichia coli Purified from Blood Cultures of Patients with Sepsis

Deep Learning-Based Prediction of Decoy Spectra for False Discovery Rate Estimation in Spectral Library Searching

With the advantage of extensive coverage, predicted spectral libraries are becoming an attractive alternative in proteomic data analysis. As a popular false discovery rate estimation method, target decoy search has been adopted in library search workflows. While existing decoy methods for curated experimental libraries have been tested, their performance in predicted library scenarios remains unknown. Current methods rely on perturbing real spectra templates, limiting the diversity and number of decoy spectra that can be generated for a given library. In this study, we explore the shuffle-and-predict decoy library generation approach, which can generate decoy spectra without the need for template spectra. Our experiments shed light on decoy method performance for predicted library scenarios and demonstrate the quality of predicted decoys in FDR estimation.

The role of bacterial ATP-binding cassette (ABC) transporters in pathogenesis and virulence: Therapeutic and vaccine potential

The ATP-binding cassette (ABC) transporter superfamily is found in all domains of life, facilitating critical biological processes through the translocation of a wide variety of substrates from, ions to proteins, across cellular membranes in an ATP-coupled process. The role of ABC transporters in eukaryotes has been well established: the facilitation of genetic diseases and multi-drug resistance (MDR) in cancer patients. In contrast, the role of ABC transporters in prokaryotes has been ambiguous due to their diverse functions and the sheer number of organisms in which they reside. This review examines the role of bacterial ABC transporters in pathogenesis and virulence, and their potential for therapeutic and vaccine application. We demonstrate how ABC transporters play a vital role in the virulence and pathogenesis of several pathogenic bacteria through the import of essential molecules, such as metal ions, amino acids, peptides, vitamins and osmoprotectants, as well as, the export of virulent determinants involved in glycoconjugate biosynthesis and Type I secretion. Furthermore, ABC exporters facilitate the persistence of pathogenic bacteria through the export of toxic xenobiotic substances, thus, contributing to the development of antimicrobial resistance. We also show that ABC transporters display considerable potential for therapeutic application through immunisation and resistance reversal. In conclusion, bacterial ABC transporters play an immense role in virulence and pathogenesis and display desirable traits for clinical use, therefore, potentially aiding in the battle against MDR.

Cellular RNA Targets of Cold Shock Proteins CspC and CspE and Their Importance for Serum Resistance in Septicemic Escherichia coli

The RNA chaperones, cold shock proteins CspC and CspE, are important in stress response and adaptation. We studied their role in the pathogenesis of a virulent Escherichia coli, representative of extraintestinal pathogenic E. coli (ExPEC) which are serum resistant and septicemic. We performed a global analysis to identify transcripts that interact with these cold shock proteins (CSPs), focusing on virulence-related genes. We used CLIP-seq, which combines UV cross-linking, immunoprecipitation and RNA sequencing. A large number of transcripts bound to the CSPs were identified, and many bind both CspC and CspE. Many transcripts were of genes involved in protein synthesis, transcription and energy metabolism. In addition, there were virulence-related genes, (i.e., fur and ryhB), essential for iron homeostasis. The CLIP-seq results were validated on two transcripts, clpX and tdcA, reported as virulence-associated. Deletion of either CspC or CspE significantly decreased their transcript levels and in a double deletion mutant cspC/cspE, the transcript stability of tdcA and clpX was reduced by 32-fold and 10-fold, respectively. We showed that these two genes are important for virulence, as deleting either of them resulted in loss of serum resistance, a requirement for sepsis. As several virulence-related transcripts interact with CspC or CspE, we determined the importance of these proteins for growth in serum and showed that deletion of either gene significantly reduced serum survival. This phenotype could be partially complemented by cspE and fully complemented by cspC. These results indicate that the two RNA chaperones are essential for virulence, and that CspC particularly critical.

IMPORTANCE Virulent Escherichia coli strains that cause infections outside the intestinal tract—extraintestinal pathogenic E. coli (ExPEC)—constitute a major clinical problem worldwide. They are involved in several distinct conditions, including urinary tract infections, newborn meningitis, and sepsis. Due to increasing antibiotic resistance, these strains are a main factor in hospital and community-acquired infections. Because many strains, which do not cross-react immunologically are involved, developing a simple vaccine is not possible. Therefore, it is essential to understand the pathogenesis of these bacteria to identify potential targets for developing drugs or vaccines. One of the least investigated systems involves RNA binding proteins, important for stability of transcripts and global gene regulation. Two such proteins are CspC and CspE (“cold shock proteins”), RNA chaperones involved in stress adaptation. Here we performed a global analysis to identify the transcripts which are affected by these two chaperones, with focus on virulence-associated transcripts.

Proteomic Approaches to Unravel Mechanisms of Antibiotic Resistance and Immune Evasion of Bacterial Pathogens

The profound effects of and distress caused by the global COVID-19 pandemic highlighted what has been known in the health sciences a long time ago: that bacteria, fungi, viruses, and parasites continue to present a major threat to human health. Infectious diseases remain the leading cause of death worldwide, with antibiotic resistance increasing exponentially due to a lack of new treatments. In addition to this, many pathogens share the common trait of having the ability to modulate, and escape from, the host immune response. The challenge in medical microbiology is to develop and apply new experimental approaches that allow for the identification of both the microbe and its drug susceptibility profile in a time-sensitive manner, as well as to elucidate their molecular mechanisms of survival and immunomodulation. Over the last three decades, proteomics has contributed to a better understanding of the underlying molecular mechanisms responsible for microbial drug resistance and pathogenicity. Proteomics has gained new momentum as a result of recent advances in mass spectrometry. Indeed, mass spectrometry-based biomedical research has been made possible thanks to technological advances in instrumentation capability and the continuous improvement of sample processing and workflows. For example, high-throughput applications such as SWATH or Trapped ion mobility enable the identification of thousands of proteins in a matter of minutes. This type of rapid, in-depth analysis, combined with other advanced, supportive applications such as data processing and artificial intelligence, presents a unique opportunity to translate knowledge-based findings into measurable impacts like new antimicrobial biomarkers and drug targets. In relation to the Research Topic “Proteomic Approaches to Unravel Mechanisms of Resistance and Immune Evasion of Bacterial Pathogens,” this review specifically seeks to highlight the synergies between the powerful fields of modern proteomics and microbiology, as well as bridging translational opportunities from biomedical research to clinical practice.

Liquid chromatography mass spectrometry-based proteomics of Escherichia coli single colony

The Escherichia coli proteome is the most extensively characterized and studied of all prokaryotic proteomes. Despite this, large scale bacterial proteomics experiments performed on E. coli cells grown in liquid cultures have failed to identify key virulence factors thought to be important determinants in establishing bacterial infections. It seems likely that many important determinants associated with virulence and host cell adhesion are exclusively expressed during growth in biofilms, which can be crudely mimicked on solid media. This method describes a simple workflow to characterize the unique proteome signature of individual, isolated single colonies, using E. coli K12 strain grown on solid media as a model system. The workflow thus provides a means to explore the proteomes of minimally passaged clinical isolates of bacteria grown on primary culture plates and to identify both unique and differentially expressed proteins contained therein.

Value of the method:

- Simple mass spectrometry-based proteomics workflow to characterise the proteome of single colony forming units

- Enables exploration of the proteomes of minimally passaged clinical isolates from primary culture plates

- Identification of virulence factors expressed in true or mimicked biofilms that may be missed in liquid cultures

Method name: E. coli single colony proteome analysis

Distinct Metabolic Features of Pathogenic Escherichia coli and Shigella spp. Determined by Label-Free Quantitative Proteomics

Escherichia coli and Shigella spp. causing illnesses in humans represent a genotypically and phenotypically diverse group of pathogens. Although E. coli diversity has been studied by comparative genomics, the intra-species variation at the proteome level is currently unknown. The proteomes of 16 pathogenic E. coli, 2 non-pathogenic E. coli, and 5 Shigella strains originating from 18 phylogenetic lineages are investigated. By applying label-free quantitative proteomics on trypsin-digested cell extracts from bacteria grown on blood agar, 4018 proteins are detected, 3285 of which arequantified, and 261 represented virulence factors. Of 753 proteins quantified in all strains, the levels of 153 vary substantially between strains and are functionally associated mostly with stress response and peripheral metabolism. The levels of proteins associated with the central metabolism vary considerably less than the levels of proteins from other metabolic pathways. Hierarchical clustering analysis based on the protein levels results in strains grouping that differ from that obtained by gene-based phylogenetic analysis. Finally, strains of some E. coli pathotypes have more similar protein profiles even when the strains are not genetically closely related. The results suggest that the degree of genetic relatedness may not necessarily be a good predictor of E. coli phenotypic characteristics.

Comparison between the proteome of Escherichia coli single colony and during liquid culture

Most bacterial proteomic studies done to date utilise bacterial cells harvested from liquid culture media. However, it is widely accepted that many important determinants associated with virulence and host cell adhesion are exclusively expressed during growth on solid media, as a crude mimic of true biofilms. Here, we compare the observed proteome of Escherichia coli K12 from isolated single colonies on solid media with those observed at different growth phases in liquid culture; i.e. early-log, mid-log, early-, mid- and late-stationary growth phases. A total of 2044 protein groups covering approximately 47% of the total proteome were identified across all studied conditions, including 1650 proteins identified from single colonies and 1679 proteins from liquid cultured cells. Label-free quantitative analysis revealed that the E. coli proteome of single colonies on a solid agar differs from that observed in liquid culture. Notably, the presence of proteins in the Suf-operon that are involved in iron mobilisation and swarming motility was associated exclusively with single colony profiles, whereas proteins involved in motility such as motA, motB, fliH, flip, fliD and fliJ were associated exclusively with cells grown in liquid culture. The data presented here provide a valuable resource for understanding the role of key proteins within microenvironments surrounding E. coli single colonies. SIGNIFICANCE: To date, most proteomics studies have used E. coli cells harvested from liquid culture media even though many important determinants associated with virulence and host cell adhesion are exclusively expressed during growth on solid media. In this study, we compare the observed proteome of E. coli K12 from isolated single colonies on solid media with those observed at different growth phases in liquid culture; i.e. early-log, mid-log, early-, mid- and late-stationary growth phases. By using label-free quantitative analysis we demonstrate that the E. coli proteome of single colonies on a solid agar differs from that observed in liquid culture with an overlap of 68% of proteins between the two culture conditions. Our analysis further reveal the presence of proteins in the Suf-operon that are involved in iron mobilisation and swarming motility was associated exclusively with single colony profiles. While those proteins involved in motility such as motA, motB, fliH, flip, fliD and fliJ were associated exclusively with cells grown in liquid culture. By comparison to E. coli proteomic data available on liquid culture and solid media, this research represents a first effort to describe the differential expression of key E. coli proteins within microenvironments surrounding single colonies.

Genomic characterization of multidrug-resistant ESBL-producing Escherichia coli ST58 causing fatal colibacillosis in critically endangered Brazilian merganser (Mergus octosetaceus)

Even though antimicrobial‐resistant bacteria have begun to be detected in wildlife, raising important issues related to their transmission and persistence of clinically important pathogens in the environment, little is known about the role of these bacteria on wildlife health, especially on endangered species. The Brazilian merganser (Mergus octosetaceus) is one of the most threatened waterfowl in the world, classified as Critically Endangered by the International Union for Conservation of Nature. In 2019, a fatal case of sepsis was diagnosed in an 8‐day‐old Brazilian merganser inhabiting a zoological park. At necropsy, major gross lesions were pulmonary and hepatic congestion. Using microbiologic and genomic methods, we identified a multidrug‐resistant (MDR) extended‐spectrum β‐lactamase (ESBL) CTX‐M‐8‐producing Escherichia coli (designed as PMPU strain) belonging to the international clone ST58, in coelomic cavity, oesophagus, lungs, small intestine and cloaca samples. PMPU strain harboured a broad resistome against antibiotics (cephalosporins, tetracyclines, aminoglycosides, sulphonamides, trimethoprim and quinolones), domestic/hospital disinfectants and heavy metals (arsenic, mercury, lead, copper and silver). Additionally, the virulence of E. coli PMPU strain was confirmed using a wax moth (Galleria mellonella) infection model, and it was supported by the presence of virulence genes encoding toxins, adherence factors, invasins and iron acquisition systems. Broad resistome and virulome of PMPU contributed to therapeutic failure and death of the animal. In brief, we report for the first time a fatal colibacillosis by MDR ESBL‐producing E. coli in critically endangered Brazilian merganser, highlighting that besides colonization, critical priority pathogens are threatening wildlife. E. coli ST58 clone has been previously reported in humans, food‐producing animals, wildlife and environment, supporting broad adaptation and persistence at human–animal–environment interface.

Variant O89 O-Antigen of E. coli Is Associated With Group 1 Capsule Loci and Multidrug Resistance

Bacterial surface polysaccharides play significant roles in fitness and virulence. In Gram-negative bacteria such as Escherichia coli, major surface polysaccharides are lipopolysaccharide (LPS) and capsule, representing O- and K-antigens, respectively. There are multiple combinations of O:K types, many of which are well-characterized and can be related to ecotype or pathotype. In this investigation, we have identified a novel O:K permutation resulting through a process of major genome reorganization in a clade of E. coli. A multidrug-resistant, extended-spectrum β-lactamase (ESBL)-producing strain - E. coli 26561 - represented a prototype of strains combining a locus variant of O89 and group 1 capsular polysaccharide. Specifically, the variant O89 locus in this strain was truncated at gnd, flanked by insertion sequences and located between nfsB and ybdK and we apply the term O89m for this variant. The prototype lacked colanic acid and O-antigen loci between yegH and hisI with this tandem polysaccharide locus being replaced with a group 1 capsule (G1C) which, rather than being a recognized E. coli capsule type, this locus matched to Klebsiella K10 capsule type. A genomic survey identified more than 200 E. coli strains which possessed the O89m locus variant with one of a variety of G1C types. Isolates from our collection with the combination of O89m and G1C all displayed a mucoid phenotype and E. coli 26561 was unusual in exhibiting a mucoviscous phenotype more recognized as a characteristic among Klebsiella strains. Despite the locus truncation and novel location, all O89m:G1C strains examined showed a ladder pattern typifying smooth LPS and also showed high molecular weight, alcian blue-staining polysaccharide in cellular and/or extra-cellular fractions. Expression of both O-antigen and capsule biosynthesis loci were confirmed in prototype strain 26561 through quantitative proteome analysis. Further in silico exploration of more than 200 E. coli strains possessing the O89m:G1C combination identified a very high prevalence of multidrug resistance (MDR) - 85% possessed resistance to three or more antibiotic classes and a high proportion (58%) of these carried ESBL and/or carbapenemase. The increasing isolation of O89m:G1C isolates from extra-intestinal infection sites suggests that these represents an emergent clade of invasive, MDR E. coli.

Potential virulence factors of bacteria associated with tail fan necrosis in the spiny lobster, Jasus edwardsii

Tail fan necrosis (TFN) is a common condition found in commercially exploited spiny lobsters that greatly diminishes their commercial value. Bacteria possessing proteolytic, chitinolytic and lipolytic capabilities were associated with TFN in spiny lobsters, Jasus edwardsii. In this study, 69 bacterial isolates exhibiting all the three enzymatic capabilities from the haemolymph and tail fans of J. edwardsii with and without TFN were further characterized and compared, including morphology, biofilm formation, antimicrobial activity, antimicrobial resistance, and production of siderophores, melanin and ammonia. The genomic patterns of the most common Vibrio crassostreae isolates were also compared between TFN-affected and unaffected lobsters. Biofilm formation was stronger in bacterial isolates from both haemolymph and tail fans of TFN-affected lobsters compared to those from the unaffected lobsters, while melanin production and siderophore production were stronger in the isolates from tail fans of lobsters with TFN. By contrast, the other characteristics of isolates were similar in lobsters with and without TFN. The Vib. crassostreae isolates from the affected lobsters had similar genomic patterns. Overall, the results indicate that in addition to proteolytic, chitinolytic and lipolytic activities, the bacteria associated with TFN commonly have enhanced activity of important virulence factors, including biofilm formation, melanin production and siderophore production.

Selective substrate uptake: The role of ATP-binding cassette (ABC) importers in pathogenesis

The uptake of nutrients, including metals, amino acids and peptides are required for many biological processes. Pathogenic bacteria scavenge these essential nutrients from microenvironments to survive within the host. Pathogens must utilize a myriad of mechanisms to acquire these essential nutrients from the host while mediating the effects of toxicity. Bacteria utilize several transport proteins, including ATP-binding cassette (ABC) transporters to import and expel substrates. ABC transporters, conserved across all organisms, are powered by the energy from ATP to move substrates across cellular membranes. In this review, we will focus on nutrient uptake, the role of ABC importers at the host-pathogen interface, and explore emerging therapies to combat pathogenesis. This article is part of a Special Issue entitled: Beyond the Structure-Function Horizon of Membrane Proteins edited by Ute Hellmich, Rupak Doshi and Benjamin McIlwain.

Invasion and translocation of uropathogenic Escherichia coli isolated from urosepsis and patients with community-acquired urinary tract infection

Uropathogenic Escherichia coli (UPEC) strains are found in high numbers in the gut of patients with urinary tract infections (UTIs). We hypothesised that in hospitalised patients, UPEC strains might translocate from the gut to the blood stream and that this could be due to the presence of virulence genes (VGs) that are not commonly found in UPEC strains that cause UTI only. To test this, E. coli strains representing 75 dominant clonal groups of UPEC isolated from the blood of hospitalised patients with UTI (urosepsis) (n = 22), hospital-acquired (HA) UTI without blood infection (n = 24) and strains isolated from patients with community-acquired (CA)-UTIs (n = 29) were tested for their adhesion to, invasion and translocation through Caco-2 cells, in addition to the presence of 34 VGs associated with UPEC. Although there were no differences in the rate and degree of translocation among the groups, urosepsis and HA-UTI strains showed significantly higher abilities to adhere (P = 0.0095 and P < 0.0001 respectively) and invade Caco-2 cells than CA-UTI isolates (P = 0.0044, P = 0.0048 respectively). Urosepsis strains also carried significantly more VGs than strains isolated from patients with only UTI and/or CA-UTI isolates. In contrast, the antigen 43 allele RS218 was found more commonly among CA-UTI strains than in the other two groups. These data indicate that UPEC strains, irrespective of their source, are capable of translocating through gut epithelium. However, urosepsis and HA-UTI strains have a much better ability to interact with gut epithelia and have a greater virulence potential than CA-UPEC, which allows them to cause blood infection.

Route of infection alters virulence of neonatal septicemia Escherichia coli clinical isolates

Escherichia coli is the leading cause of Gram-negative neonatal septicemia in the United States. Invasion and passage across the neonatal gut after ingestion of maternal E. coli strains produce bacteremia. In this study, we compared the virulence properties of the neonatal E. coli bacteremia clinical isolate SCB34 with the archetypal neonatal E. coli meningitis strain RS218. Whole-genome sequencing data was used to compare the protein coding sequences among these clinical isolates and 33 other representative E. coli strains. Oral inoculation of newborn animals with either strain produced septicemia, whereas intraperitoneal injection caused septicemia only in pups infected with RS218 but not in those injected with SCB34. In addition to being virulent only through the oral route, SCB34 demonstrated significantly greater invasion and transcytosis of polarized intestinal epithelial cells in vitro as compared to RS218. Protein coding sequences comparisons highlighted the presence of known virulence factors that are shared among several of these isolates, and revealed the existence of proteins exclusively encoded in SCB34, many of which remain uncharacterized. Our study demonstrates that oral acquisition is crucial for the virulence properties of the neonatal bacteremia clinical isolate SCB34. This characteristic, along with its enhanced ability to invade and transcytose intestinal epithelium are likely determined by the specific virulence factors that predominate in this strain.

Relative Quantitative Proteomic Analysis of Brucella abortus Reveals Metabolic Adaptation to Multiple Environmental Stresses

Brucella spp. are facultative intracellular pathogens that cause chronic brucellosis in humans and animals. The virulence of Brucella primarily depends on its successful survival and replication in host cells. During invasion of the host tissue, Brucella is simultaneously subjected to a variety of harsh conditions, including nutrient limitation, low pH, antimicrobial defenses, and extreme levels of reactive oxygen species (ROS) via the host immune response. This suggests that Brucella may be able to regulate its metabolic adaptation in response to the distinct stresses encountered during its intracellular infection of the host. An investigation into the differential proteome expression patterns of Brucella grown under the relevant stress conditions may contribute toward a better understanding of its pathogenesis and adaptive response. Here, we utilized a mass spectrometry-based label-free relative quantitative proteomics approach to investigate and compare global proteomic changes in B. abortus in response to eight different stress treatments. The 3 h short-term in vitro single-stress and multi-stress conditions mimicked the in vivo conditions of B. abortus under intracellular infection, with survival rates ranging from 3.17 to 73.17%. The proteomic analysis identified and quantified a total of 2,272 proteins and 74% of the theoretical proteome, thereby providing wide coverage of the B. abortus proteome. By including eight distinct growth conditions and comparing these with a control condition, we identified a total of 1,221 differentially expressed proteins (DEPs) that were significantly changed under the stress treatments. Pathway analysis revealed that most of the proteins were involved in oxidative phosphorylation, ABC transporters, two-component systems, biosynthesis of secondary metabolites, the citrate cycle, thiamine metabolism, and nitrogen metabolism; constituting major response mechanisms toward the reconstruction of cellular homeostasis and metabolic balance under stress. In conclusion, our results provide a better understanding of the global metabolic adaptations of B. abortus associated with distinct environmental stresses. The identification of proteins necessary for stress resistance is crucial toward elucidating the infectious process in order to control brucellosis, and may facilitate the discovery of novel therapeutic targets and effective vaccines.

Identification of New Virulence Factors and Vaccine Candidates for Yersinia pestis

Earlier, we reported the identification of new virulence factors/mechanisms of Yersinia pestis using an in vivo signature-tagged mutagenesis (STM) screening approach. From this screen, the role of rbsA, which encodes an ATP-binding protein of ribose transport system, and vasK, an essential component of the type VI secretion system (T6SS), were evaluated in mouse models of plague and confirmed to be important during Y. pestis infection. However, many of the identified genes from the screen remained uncharacterized. In this study, in-frame deletion mutants of ypo0815, ypo2884, ypo3614-3168 (cyoABCDE), and ypo1119-1120, identified from the STM screen, were generated. While ypo0815 codes for a general secretion pathway protein E (GspE) of the T2SS, the ypo2884-encoded protein has homology to the βγ crystallin superfamily, cyoABCDE codes for the cytochrome o oxidase operon, and the ypo1119-1120 genes are within the Tol-Pal system which has multiple functions. Additionally, as our STM screen identified three T6SS-associated genes, and, based on in silico analysis, six T6SS clusters and multiple homologs of the T6SS effector hemolysin-coregulated protein (Hcp) exist in Y. pestis CO92, we also targeted these T6SS clusters and effectors for generating deletion mutants. These deletion mutant strains exhibited varying levels of attenuation (up to 100%), in bubonic or pneumonic murine infection models. The attenuation could be further augmented by generation of combinatorial deletion mutants, namely ΔlppΔypo0815, ΔlppΔypo2884, ΔlppΔcyoABCDE, ΔvasKΔhcp6, and Δypo2720-2733Δhcp3. We earlier showed that deletion of the lpp gene, which encodes Braun lipoprotein (Lpp) and activates Toll-like receptor-2, reduced virulence of Y. pestis CO92 in murine models of bubonic and pneumonic plague. The surviving mice infected with ΔlppΔcyoABCDE, ΔvasKΔhcp6, and Δypo2720-2733Δhcp3 mutant strains were 55-100% protected upon subsequent re-challenge with wild-type CO92 in a pneumonic model. Further, evaluation of the attenuated T6SS mutant strains in vitro revealed significant alterations in phagocytosis, intracellular survival in murine macrophages, and their ability to induce cytotoxic effects on macrophages. The results reported here provide further evidence of the utility of the STM screening approach for the identification of novel virulence factors and to possibly target such genes for the development of novel live-attenuated vaccine candidates for plague.

Proteomic comparison of three clinical diarrhoeagenic drug-resistant Escherichia coli isolates grown on CHROMagar™STEC media

Shiga-toxin-producing Escherichia coli (STEC) and enteropathogenic Escherichia coli (EPEC) are key diarrhoea-causing foodborne pathogens. We used proteomics to characterize the virulence and antimicrobial resistance protein profiles of three clinical pathogenic E. coli isolates (two EPEC [one resistant to ciprofloxacin] and one STEC) cultured on CHROMagar™STEC solid media after minimal laboratory passage. We identified 4767 unique peptides from 1630 protein group across all three clinical E. coli strains. Label-free proteomic analysis allowed the identification of virulence and drug resistance proteins that were unique to each of the clinical isolates compared in this study. The B subunit of Shiga toxin, ToxB, was uniquely detected in the STEC strain while several other virulence factors including SheA, OmpF, OmpC and OmpX were significantly more abundant in the STEC strain. The ciprofloxacin resistant EPEC isolate possessed reduced levels of key virulence proteins compared to the ciprofloxacin susceptible EPEC and STEC strains. Parallel reaction monitoring assays validated the presence of biologically relevant proteins across biologically-replicated cultures. Propagation of clinical isolates on a relevant solid medium followed by mass spectrometry analysis represents a convenient means to quantify virulence factors and drug resistance determinants that might otherwise be lost through extensive in vitro passage in enteropathogenic bacteria.

SIGNIFICANCE

Through the use of quantitative proteomics, we have characterized the virulence and antimicrobial resistance attributes of three clinically isolated, pathogenic E. coli strains cultured on solid media. Our results provide new, quantitative data on the expressed proteomes of these tellurite-resistant, diarrhoeagenic E. coli strains and reveal a subset of antimicrobial resistance and virulence proteins that are differentially abundant between these clinical strains. Our quantitative proteomics-based approach should thus have applicability in microbiological diagnostic labs for the identification of pathogenic/drug resistant E. coli in the future.

De novo peptide sequencing by deep learning

De novo peptide sequencing from tandem MS data is the key technology in proteomics for the characterization of proteins, especially for new sequences, such as mAbs. In this study, we propose a deep neural network model, DeepNovo, for de novo peptide sequencing. DeepNovo architecture combines recent advances in convolutional neural networks and recurrent neural networks to learn features of tandem mass spectra, fragment ions, and sequence patterns of peptides. The networks are further integrated with local dynamic programming to solve the complex optimization task of de novo sequencing. We evaluated the method on a wide variety of species and found that DeepNovo considerably outperformed state of the art methods, achieving 7.7-22.9% higher accuracy at the amino acid level and 38.1-64.0% higher accuracy at the peptide level. We further used DeepNovo to automatically reconstruct the complete sequences of antibody light and heavy chains of mouse, achieving 97.5-100% coverage and 97.2-99.5% accuracy, without assisting databases. Moreover, DeepNovo is retrainable to adapt to any sources of data and provides a complete end-to-end training and prediction solution to the de novo sequencing problem. Not only does our study extend the deep learning revolution to a new field, but it also shows an innovative approach in solving optimization problems by using deep learning and dynamic programming.