Search

Gram-positive bacteria are some of the earliest known life forms, diverging from gram-negative bacteria 2 billion years ago. These organisms utilize sortase enzymes to attach proteins to their peptidoglycan cell wall, a structural feature that distinguishes the two types of bacteria. The transpeptidase activity of sortases make them an important tool in protein engineering applications, e.g., in sortase-mediated ligations or sortagging. However, due to relatively low catalytic efficiency, there are ongoing efforts to create better sortase variants for these uses. Here, we use bioinformatics tools, principal component analysis and ancestral sequence reconstruction, in combination with protein biochemistry, to analyze natural sequence variation in these enzymes. Principal component analysis on the sortase superfamily distinguishes previously described classes and identifies regions of relatively high sequence variation in structurally-conserved loops within each sortase family, including those near the active site. Using ancestral sequence reconstruction, we determined sequences of ancestral Staphylococcus and Streptococcus Class A sortase proteins. Enzyme assays revealed that the ancestral Streptococcus enzyme is relatively active and shares similar sequence variation with other Class A Streptococcus sortases. Taken together, we highlight how natural sequence variation can be utilized to investigate this important protein family, arguing that these and similar techniques may be used to discover or design sortases with increased catalytic efficiency and/or selectivity for sortase-mediated ligation experiments.

Protein-protein interactions that include recognition of short sequences of amino acids, or peptides, are critical in cellular processes. Protein-peptide interaction surface areas are relatively small and shallow, and there are often overlapping specificities in families of peptide-binding domains. Therefore, dissecting selectivity determinants can be challenging. PDZ domains are an example of a peptide-binding domain located in several intracellular signaling and trafficking pathways, which form interactions critical for the regulation of receptor endocytic trafficking, tight junction formation, organization of supramolecular complexes in neurons, and other biological systems. These domains are also directly targeted by pathogens, and a hallmark of many oncogenic viral proteins is a PDZ-binding motif. However, amidst sequences that target PDZ domains, there is a wide spectrum in relative promiscuity. For example, the viral HPV16 E6 oncoprotein recognizes over double the number of PDZ domain-containing proteins as the cystic fibrosis transmembrane conductance regulator (CFTR) in the cell, despite similar PDZ targeting-sequences and identical motif residues. Here, we determine binding affinities for PDZ domains known to bind either HPV16 E6 alone or both CFTR and HPV16 E6, using peptides matching WT and hybrid sequences. We also use energy minimization to model PDZ-peptide complexes and use sequence analyses to investigate this difference. We find that while the majority of single mutations had a marginal effect on overall affinity, the additive effect on the free energy of binding accurately describes the selectivity observed. Taken together, our results describe how complex and differing PDZ interactomes can be programmed in the cell.

Background Volumetric laser endomicroscopy (VLE) allows for near-microscopic imaging of the superficial esophageal wall and may improve detection of early neoplasia in Barrett’s esophagus (BE). Interpretation of a 6-cm long, circumferential VLE “full scan” may however be challenging for endoscopists. We aimed to evaluate the accuracy of VLE experts in correctly diagnosing VLE full scans of early neoplasia and non-dysplastic BE (NDBE). Methods 29 VLE full scan videos (15 neoplastic and 14 NDBE) were randomly evaluated by 12 VLE experts using a web-based module. Experts were blinded to the endoscopic BE images and histology. The 15 neoplastic cases contained a subtle endoscopically visible lesion, which on endoscopic resection showed high grade dysplasia or cancer. NDBE cases had no visible lesions and an absence of dysplasia in all biopsies. VLE videos were first scored as “neoplastic” or “NDBE.” If neoplastic, assessors located the area most suspicious for neoplasia. Primary outcome was the performance of VLE experts in differentiating between non-dysplastic and neoplastic full scan videos, calculated by accuracy, sensitivity, and specificity. Secondary outcomes included correct location of neoplasia, interobserver agreement, and level of confidence. Results VLE experts correctly labelled 73 % (95 % confidence interval [CI] 67 % – 79 %) of neoplastic VLE videos. In 54 % (range 27 % – 66 %) both neoplastic diagnosis and lesion location were correct. NDBE videos were consistent with endoscopic biopsies in 52 % (95 %CI 46 % – 57 %). Interobserver agreement was fair (kappa 0.28). High level of confidence was associated with a higher rate of correct neoplastic diagnosis (81 %) and lesion location (73 %). Conclusions Identification of subtle neoplastic lesions in VLE full scans by experts was disappointing. Future studies should focus on improving methodologies for reviewing full scans, development of refined VLE criteria for neoplasia, and computer-aided diagnosis of VLE scans.

There has been a vast increase in GI literature focused on the use of machine learning in endoscopy. The relative novelty of this field poses a challenge for reviewers and readers of GI journals. To appreciate scientific quality and novelty of machine learning studies, understanding of the technical basis and commonly used techniques is required. Clinicians often lack this technical background, while machine learning experts may be unfamiliar with clinical relevance and implications for daily practice. Therefore, there is an increasing need for a multidisciplinary, international evaluation on how to perform high-quality machine learning research in endoscopy. This review aims to provide guidance for readers and reviewers of peer-reviewed GI journals to allow critical appraisal of the most relevant quality requirements of machine learning studies. The paper provides an overview of common trends and their potential pitfalls and proposes comprehensive quality requirements in six overarching themes: terminology, data, algorithm description, experimental setup, interpretation of results and machine learning in clinical practice.

Identification of the molecular networks that facilitated the evolution of multicellular animals from their unicellular ancestors is a fundamental problem in evolutionary cellular biology. Choanoflagellates are recognized as the closest extant nonmetazoan ancestors to animals. These unicellular eukaryotes can adopt a multicellular‐like “rosette” state. Therefore, they are compelling models for the study of early multicellularity. Comparative studies revealed that a number of putative human orthologs are present in choanoflagellate genomes, suggesting that a subset of these genes were necessary for the emergence of multicellularity. However, previous work is largely based on sequence alignments alone, which does not confirm structural nor functional similarity. Here, we focus on the PDZ domain, a peptide‐binding domain which plays critical roles in myriad cellular signaling networks and which underwent a gene family expansion in metazoan lineages. Using a customized sequence similarity search algorithm, we identified 178 PDZ domains in the Monosiga brevicollis proteome. This includes 11 previously unidentified sequences, which we analyzed using Rosetta and homology modeling. To assess conservation of protein structure, we solved high‐resolution crystal structures of representative M. brevicollis PDZ domains that are homologous to human Dlg1 PDZ2, Dlg1 PDZ3, GIPC, and SHANK1 PDZ domains. To assess functional conservation, we calculated binding affinities for mbGIPC, mbSHANK1, mbSNX27, and mbDLG‐3 PDZ domains from M. brevicollis. Overall, we find that peptide selectivity is generally conserved between these two disparate organisms, with one possible exception, mbDLG‐3. Overall, our results provide novel insight into signaling pathways in a choanoflagellate model of primitive multicellularity.

Drugs which independently inhibit a shared target or pathway can have synthetic activities that result in multiplicative instead of merely additive potencies. This characteristic of drug combinations can be quantified by expressing the potency of the combination as if it were a single agent. We show that by optimizing this quantity we can prospectively design drug cocktails with apparent potencies that far exceed any of its individual components. We illustrate the power of this approach, which is based on statistical design of experiments to select optimal drug combinations, and response surface methodology to determine optimal drug ratios, by building a drug cocktail comprised of three antibodies for treating C. difficile infection that is almost 1000-fold more potent than the current, clinically approved antibody monotherapy. High synthetic activities do not require unusual drug interactions, and therefore may be achievable much more readily than generally appreciated.One-Sentence SummaryA development pathway is described for designing antibody cocktails with potencies that far exceed what is achievable with single antibodies

Gram-positive bacteria contain sortase enzymes on their cell surfaces that catalyze transpeptidation reactions critical for proper cellular function. In vitro, sortases are used in sortase-mediated ligation (SML) reactions for a variety of protein engineering applications. Historically, sortase A from Staphylococcus aureus (saSrtA) has been the enzyme of choice for SML reactions. However, the stringent specificity of saSrtA for the sequence motif LPXTG limits its uses. Here, we use principal component analysis to identify a structurally conserved loop with a high degree of variability in all classes of sortases. We investigate the contribution of this β7-β8 loop, located between the catalytic cysteine and arginine residues and immediately adjacent to the target binding cleft, by designing and testing chimeric sortase enzymes. Our chimeras utilize natural sequence variation of Class A sortases from 8 species engineered into the SrtA sequence from Streptococcus pneumoniae (spSrtA). While some of our chimeric enzymes mimic the activity and selectivity of the wild-type protein from which the loop sequence is derived (e.g., that of saSrtA), others result in chimeric spSrtA enzymes able to accommodate a range of residues in the final position of the substrate motif (LPXTX). Using mutagenesis, structural, and sequence analyses, we identify three interactions facilitated by β7-β8 loop residues that appear to be broadly characteristic of Class A sortase enzymes. These studies provide the foundation for a deeper understanding of sortase target selectivity and can expand the sortase toolbox for future SML applications.

Computer-aided detection (CAD) approaches have shown promising results for early esophageal cancer detection using Volumetric Laser Endoscopy (VLE) imagery. However, the relatively slow and computationally costly tissue segmentation employed in these approaches hamper their clinical applicability. In this paper, we propose to reframe the 2D tissue segmentation problem into a 1D tissue boundary detection problem. Instead of using an encoder-decoder architecture, we propose to follow the tissue boundary using a Recurrent Neural Network (RNN), exploiting the spatio-temporal relations within VLE frames. We demonstrate a near state-of-the-art performance using 18 times less floating point operations, enabling real-time execution in clinical practice.

Patients with Barrett’s esophagus, in which the normal lining of the esophageal wall has been replaced under influence of chronic acid reflux, have an increased risk of developing esophageal adenocarcinoma. Barrett’s patients therefore undergo regular endoscopic surveillance. When esophageal cancer is detected in an early stage it can be treated endoscopically with an excellent prognosis. However, early cancer has a subtle appearance and can therefore be missed. To increase neoplasia detection in Barrett’s esophagus (BE), many advanced endoscopic imaging techniques have been developed over the past decades. Several of these techniques have been discussed in this thesis, including optical chromoscopy, magnification endoscopy and volumetric laser endomicroscopy (VLE). In this thesis, we have investigated the application of these novel imaging techniques for detection and characterization of BE neoplasia. We described three major challenges of Barrett surveillance: 1) a subjective interpretation of endoscopic imagery by endoscopists, 2) the increase of available information within a Barrett segment, and 3) sampling error associated with random biopsies. To address these challenges, application of artificial intelligence (AI) was explored for enhanced endoscopic image interpretation. Additionally, VLE was studied, which offers the possibility to visualize the different esophageal wall layers and thereby guide targeted biopsies. Finally, we developed an AI-tool to aid in the interpretation of the complex information within VLE imagery in order to improve BE neoplasia detection. Future studies should evaluate our AI-tools during live clinical procedures.

Because the current Barrett's esophagus (BE) surveillance protocol suffers from sampling error of random biopsies and a high miss-rate of early neoplastic lesions, many new endoscopic imaging and sampling techniques have been developed. None of these techniques, however, have significantly increased the diagnostic yield of BE neoplasia. In fact, these techniques have led to an increase in the amount of visible information, yet endoscopists and pathologists inevitably suffer from variations in intra- and interobserver agreement. Artificial intelligence systems have the potential to overcome these endoscopist-dependent limitations.

Patients suffering from Barrett's Esophagus (BE) are at an increased risk of developing esophageal adenocarcinoma and early detection is crucial for a good prognosis. To aid the endoscopists with the early detection for this preliminary stage of esophageal cancer, this work concentrates on the development and extensive evaluation of a state-of-the-art computer-aided classification and localization algorithm for dysplastic lesions in BE. To this end, we have employed a large-scale endoscopic data set, consisting of 494,355 images, in combination with a novel semi-supervised learning algorithm to pretrain several instances of the proposed neural network architecture. Next, several Barrett-specific data sets that are increasingly closer to the target domain with significantly more data compared to other related work, were used in a multi-stage transfer learning strategy. Additionally, the algorithm was evaluated on two prospectively gathered external test sets and compared against 53 medical professionals. Finally, the model was also evaluated in a live setting without interfering with the current biopsy protocol. Results from the performed experiments show that the proposed model improves on the state-of-the-art on all measured metrics. More specifically, compared to the best performing state-of-the-art model, the specificity is improved by more than 20% points while simultaneously preserving high sensitivity and reducing the false positive rate substantially. Our algorithm yields similar scores on the localization metrics, where the intersection of all experts is correctly indicated in approximately 92% of the cases. Furthermore, the live pilot study shows great performance in a clinical setting with a patient level accuracy, sensitivity, and specificity of 90%. Finally, the proposed algorithm outperforms each individual medical expert by at least 5% and the average assessor by more than 10% over all assessor groups with respect to accuracy.

Volumetric laser endomicroscopy (VLE) has been shown to improve detection of early neoplasia in Barrett’s esophagus (BE). However, diagnostic performance using histopathology-correlated VLE regions of interest (ROIs) has not been adequately studied. We evaluated the diagnostic accuracy of VLE assessors for identification of early BE neoplasia in histopathology-correlated VLE ROIs. In total, 191 ROIs (120 nondysplastic and 71 neoplastic) from 50 BE patients were evaluated in a random order using a web-based module. All ROIs contained histopathology correlations enabled by VLE laser marking. Assessors were blinded to endoscopic BE images and histology. ROIs were first scored as nondysplastic or neoplastic. Level of confidence was assigned to the predicted diagnosis. Outcome measures were: (i) diagnostic performance of VLE assessors for identification of BE neoplasia in all VLE ROIs, defined as accuracy, sensitivity, and specificity; (ii) diagnostic performance of VLE assessors for only high level of confidence predictions; and (iii) interobserver agreement. Accuracy, sensitivity, and specificity for BE neoplasia identification were 79% (confidence interval [CI], 75–83), 75% (CI, 71–79), and 81% (CI, 76–86), respectively. When neoplasia was identified with a high level of confidence, accuracy, sensitivity, and specificity were 88%, 83%, and 90%, respectively. The overall strength of interobserver agreement was fair (k = 0.29). VLE assessors can identify BE neoplasia with reasonable diagnostic accuracy in histopathology-correlated VLE ROIs, and accuracy is enhanced when BE neoplasia is identified with high level of confidence. Future work should focus on renewed VLE image reviewing criteria and real-time automatic assessment of VLE scans.

The evolution of signaling pathways is complex and well-studied. In particular, the emergence of animal multicellularity had a major impact on protein-protein interactions and signaling systems in eukaryotic cells. However, choanoflagellates, our closest non-metazoan ancestor, contain a number of closely related signaling and trafficking proteins and domains. In addition, because choanoflagellates can adopt a rosette-/multicellular-like state, a lot can be gained by comparing proteins involved in choanoflagellate and human signaling pathways. Here, we look at how selectivity determinants evolved in the PDZ domain. There are over 250 PDZ domains in the human proteome, which are involved in critical protein-protein interactions that result in large multimolecular complexes, e.g., in the postsynaptic density of neuronal synapses. Binding of C-terminal sequences by PDZ domains is often transient and recognition typically involves 6 residues or less, with only 2 residues specifying the binding motif. We solved high resolution crystal structures ofMonosiga brevicollisPDZ domains homologous to human Dlg1 PDZ2, Dlg1 PDZ3, GIPC, and SHANK1 PDZ domains to investigate if the non-motif preferences are conserved, despite hundreds of millions of years of evolution. We also calculated binding affinities for GIPC, SHANK1, and SNX27 PDZ domains fromM. brevicollis. Overall, we found that peptide selectivity is conserved between these two disparate organisms, with one exception, mbDLG-3. In addition, we identify 178 PDZ domains in theM. brevicollisproteome, including 11 new sequences, which we verified using Rosetta and homology modeling. Overall, our results provide novel insight into signaling pathways in the choanoflagellate organism.

Background and aims Patients in the intensive care unit (ICU) with acute pancreatitis (AP) are at risk for extra-pancreatic complications given their severe illness and prolonged length of stay. We sought to determine the rate of extra-pancreatic complications and its effect on length of stay (LOS) and mortality in ICU patients with AP. Methods We performed a retrospective cohort study of ICU patients admitted to a tertiary-care center with a diagnosis of AP. A total of 287 ICU patients had a discharge diagnosis of AP, of which 163 met inclusion criteria. We calculated incidence rates of extra-pancreatic complications and performed a univariate and multi-variable analysis to determine predictors of LOS and mortality. Results There were a total of 158 extra-pancreatic complications (0.97 extra-pancreatic complications per patient). Ninety-five patients had at least one extra-pancreatic complication, whereas 68 patients had no extra-pancreatic complications. Patients with extra-pancreatic complications had a significantly longer LOS (14.7 vs 8.8 days, p