Professor Robert Braidwood Sim-"Bob"-A Career in Complement Research Spanning 1973-2021.

Bob joined the MRC Immunochemistry Unit, within the Department of Biochemistry at Oxford University, as a D [...].

Heuristic hyperparameter optimization of deep learning models for genomic prediction.

There is a growing interest among quantitative geneticists and animal breeders in the use of deep learning (DL) for genomic prediction. However, the performance of DL is affected by hyperparameters that are typically manually set by users. These hyperparameters do not simply specify the architecture of the model; they are also critical for the efficacy of the optimization and model-fitting process. To date, most DL approaches used for genomic prediction have concentrated on identifying suitable hyperparameters by exploring discrete options from a subset of the hyperparameter space. Enlarging the hyperparameter optimization search space with continuous hyperparameters is a daunting combinatorial problem. To deal with this problem, we propose using differential evolution (DE) to perform an efficient search of arbitrarily complex hyperparameter spaces in DL models, and we apply this to the specific case of genomic prediction of livestock phenotypes. This approach was evaluated on two pig and cattle datasets with real genotypes and simulated phenotypes (N = 7,539 animals and M = 48,541 markers) and one real dataset (N = 910 individuals and M = 28,916 markers). Hyperparameters were evaluated using cross-validation. We compared the predictive performance of DL models using hyperparameters optimized by DE against DL models with "best practice" hyperparameters selected from published studies and baseline DL models with randomly specified hyperparameters. Optimized models using DE showed a clear improvement in predictive performance across all three datasets. DE optimized hyperparameters also resulted in DL models with less overfitting and less variation in predictive performance over repeated retraining compared to non-optimized DL models.

Serial use of existing clinical decisions aids can reduce computed tomography pulmonary angiography for pulmonary embolism.

Pulmonary embolism (PE) remains a diagnostic challenge in emergency medicine. Clinical decision aids (CDAs) like the Pulmonary Embolism Rule-Out Criteria (PERC) are sensitive but poorly specific; serial CDA use may improve specificity. The goal of this before-and-after study was to determine if serial use of existing CDAs in a novel diagnostic algorithm safely decreases the use of CT pulmonary angiograms (CTPA). This was a retrospective before-and-after study conducted at an urban ED with 105,000 annual visits. Our algorithm uses PERC, Wells' score, and D-dimer in series, before moving to CTPA. The algorithm was introduced in January, 2017. Use of CDAs and D-dimer in the 24 months pre- and 12 months post-intervention were obtained by chart review. The algorithm's effect on CTPA ordering was assessed by comparing volume 5 years pre- and 3 years post-intervention, adjusted for ED volume. Mean CTPAs per 1000 adult ED visits was 11.1 in the 5 pre-intervention years and 9.9 in the 3 post-intervention years (p < 0.0001). Use of PERC, Wells' score and D-dimer increased from 1.1%, 1.1%, and 28% to 8.8% (p = 0.0002) 8.1% (p = 0.0005), and 35% (p = 0.0066), respectively. Pre-intervention, there were six potentially missed PEs compared to three in the post-intervention period. Introduction of our serial CDA diagnostic algorithm was associated with increased use of CDAs and D-dimer and reduced CTPA rate without an apparent increase in the number of missed PEs. Prospective validation is needed to confirm these results.

Left atrium perforation with lung injury after catheter ablation.

Left atrial perforation is a known complication following pulmonary vein catheter ablation. Our case of a 62-year-old female underwent urgent surgery for repair of left atrium perforation with left pleural effusion as a late complication after multiple transcatheter radiofrequency pulmonary vein ablations for persistent atrial fibrillation.

Improving Timeliness of Oncology Assessment and Cancer Treatment Through Implementation of a Multidisciplinary Lung Cancer Clinic.

PURPOSE: Timely lung cancer care has been associated with improved clinical outcomes and patient satisfaction. We identified improvement opportunities in lung cancer management pathways at Kingston Health Sciences Centre. Quality improvement strategies led to the implementation of a multidisciplinary lung cancer clinic (MDC). METHODS: We set an outcome measure of decreasing the time from diagnosis to first cancer treatment by 10 days within 6 months of clinic implementation. We implemented a weekly MDC that involved respirologists, medical oncologists, and radiation oncologists at which patients with new lung cancer diagnoses were offered concurrent oncology consultation. We used Plan-Do-Study-Act cycles to guide our improvement initiatives. A total of five Plan-Do-Study-Act cycles spanned 14 months and consisted of an MDC pilot clinic, large-scale MDC launching, debriefing meetings, and clinic expansion. Pre-MDC data were analyzed retrospectively to establish baseline and prospectively for improvement. Statistical Process Control XmR(i) charts were used to report data. RESULTS: Since MDC initiation, 128 patients have been seen in 34 MDC clinics (3.8 patients per clinic). Mean days from diagnosis to first oncology assessment decreased from 12.4 days to 3.9 days, and mean days from diagnosis to first cancer treatment decreased from 39.5 to 15.0 days, both of which demonstrated special cause variation. Time to assessment and treatment improved for patients with every stage of lung cancer and for both small-cell and non-small-cell subtypes. CONCLUSION: MDC shortens the time from lung cancer diagnosis to oncology assessment and treatment. Time to treatment improved more than time to oncology assessment, which suggests the improvement is related to benefits beyond faster oncology assessment.

Complement Component C1q: Historical Perspective of a Functionally Versatile, and Structurally Unusual, Serum Protein.

Complement component C1q plays an important recognition role in adaptive, and innate, immunity through its ability to interact, via its six globular head regions, with both immunoglobulin and non-immunoglobulin activators of the complement system, and also in the clearance of cell debris, and by playing a role in regulation of cellular events by interacting with a wide range of cell surface molecules. The presence of collagen-like triple-helical structures within C1q appears crucial to the presentation, and multivalent binding, of the globular heads of C1q to targets, and also to its association with the proenzyme complex of C1r2-C1s2, to yield the C1 complex. The possible role that movement of these collagen-like structures may play in the activation of the C1 complex is a controversial area, with there still being no definitive answer as to how the first C1r proenzyme molecule becomes activated within the C1 complex, thus allowing it to activate proenzyme C1s, and initiate and the consequent cascade of events in the activation of the classical pathway of complement. The globular heads of C1q are similar to domains found within the tumor necrosis factor (TNF) superfamily of proteins, and have been shown to bind to a very wide range of ligands. In addition to its well-defined roles in infection and immunity, a variety of other functions associated with C1q include possible roles, in the development of problems in the central nervous system, which occur with aging, and perhaps in the regulation of tumor growth.

Crystal Structure of a Complex of Surfactant Protein D (SP-D) and Haemophilus influenzae Lipopolysaccharide Reveals Shielding of Core Structures in SP-D-Resistant Strains.

The carbohydrate recognition domains (CRDs) of lung collectin surfactant protein D (SP-D) recognize sugar patterns on the surface of lung pathogens and promote phagocytosis. Using Haemophilus influenzae Eagan strains expressing well-characterized lipopolysaccharide (LPS) surface structures of various levels of complexity, we show that bacterial recognition and binding by SP-D is inversely related to LPS chain extent and complexity. The crystal structure of a biologically active recombinant trimeric SP-D CRD complexed with a delipidated Eagan 4A LPS suggests that efficient LPS recognition by SP-D requires multiple binding interactions utilizing the three major ligand-binding determinants in the SP-D binding pocket, with Ca-dependent binding of inner-core heptose accompanied by interaction of anhydro-Kdo (4,7-anhydro-3-deoxy-d-manno-oct-2-ulosonic acid) with Arg343 and Asp325. Combined with enzyme-linked immunosorbent assays (ELISAs) and fluorescence-activated cell sorter (FACS) binding analyses, our results show that extended LPS structures previously thought to be targets for collectins are important in shielding the more vulnerable sites in the LPS core, revealing a mechanism by which pathogens with complex LPS extensions efficiently evade a first-line mucosal innate immune defense. The structure also reveals for the first time the dominant form of anhydro-Kdo.

Complement and non-complement activating functions of C1q: a prototypical innate immune molecule.

C1q is a versatile innate immune molecule that serves as the initiation subcomponent of the classical complement pathway. In addition, it is also a potent pattern recognition molecule, the versatility of which has fuelled its functional flexibility. C1q recognises an array of self, non-self and altered-self ligands. The broad-spectrum ligand-binding potential of C1q is facilitated by the modular organisation of the heterotrimeric globular head region, its ability to change its conformation in a very subtle way, and the manner in which this ancient molecule appears to have evolved to deal with the different types of ligands. Over recent years, molecules that resemble C1q have been put together to form the C1q family. In this review, we briefly summarise complement-dependent and complement-independent functions of C1q, its cognate receptors and key members of the rapidly growing C1q family.

Physiological concentration of calcium inhibits elastase-induced cleavage of a functional recombinant fragment of surfactant protein D.

Surfactant protein D (SP-D) plays an important role in lung host defence. SP-D levels have been shown to be depleted in cystic fibrosis (CF) patients. A recombinant fragment of the human SP-D (rfhSP-D) which consist of a hydrophobic neck and a CRD has been shown to be active in vivo and partially reverses the symptoms of the SP-D deficiency in the lungs when administered to SP-D knock-out mice. In this paper we studied the in vitro effect of different proteolytic enzymes commonly found in CF patients lungs, such as neutrophil elastase, cathepsin G and protease 3 as well as Pseudomonas elastase, on rfhSP-D. It was also shown that cleavage was inhibited by physiological concentration of calcium. When Western blot was compared with ELISA, we show that an anti-SP-D ELISA is a not a reliable assay of functional SP-D levels since non-functional fragments of SP-D are also detected. Thus, ELISA cannot be used as a reliable "diagnostic" tool for SP-D deficiency. Finally, we observe that SP-D is not cleaved in control patients but is degraded in about half the samples from cystic fibrosis patients, indicating that degradation of endogenous SP-D, by enzymes present in CF bronchioalveolar lavage fluid (BALF), may lead to deficiency of the protein as seen in CF and therefore rfhSP-D may be a useful future therapy.

New insights into the molecular mechanisms of classical complement activation.

C1q, the initiator of the classical complement cascade, is a versatile molecule with numerous ligands and variety of functions. Recent mutagenesis, epitope mapping and structural data brought novel understanding of the molecular mechanisms of C1q binding to target molecules, and subsequent C1 activation. Evidence has emerged suggesting that residues located within the C1q apical surface, and the exposed side surface of the B chain, facilitate the interaction of C1q with the majority of C1q ligands. The binding of C1q to IgG, IgM, CRP, and PTX3 is most likely a contiguous process, developing in different phases. During the initial phase, residues located within the gC1q apex, and shared between the three chains, are involved in the interaction with the ligands. After this initial recognition event, the Ca(2+) ion is attracted by the negatively charged C1q ligand. This loss of the Ca(2+) ion induces a rotation of the globular C1q head, facilitating further ligand binding, and transmitting an activation signal to C1r-C1s. This review summarizes these data, and offers a unifying model for C1 activation by negatively charged gC1q targets.

Susceptibility of mice genetically deficient in SP-A or SP-D gene to invasive pulmonary aspergillosis.

Pulmonary surfactant proteins, SP-A and SP-D, are carbohydrate pattern recognition molecules of innate immunity, which significantly enhance phagocytosis and killing of Aspergillus fumigatus, a pathogenic fungus, by neutrophils and macrophages. The present study examined the susceptibility of immunosuppressed SP-A gene deficient (SP-A(-/-)) or SP-D gene deficient (SP-D(-/-)) mice to A. fumigatus conidia challenge compared to wild-type (WT) mice. A. fumigatus-challenged SP-A(-/-) (SP-A(-/-) IPA) mice showed less mortality (40%) than the WT-IPA mice (100%) and increased mortality (60%) following administration of SP-A with decreased TNF-alpha and IFN-gamma to IL-4 ratio than SP-A(-/-) IPA mice. The SP-D(-/-) IPA mice (57.14%) showed similar mortality as WT-IPA mice (60%). However, the SP-D (-/-) IPA mice (42.86% mortality on day 2) died earlier than the WT-IPA mice (20% mortality on day 2), showed a higher hyphal density and tissue injury in lungs. Treatment with SP-D or a recombinant fragment of human SP-D rhSP-D reduced the mortality to 50% and 33%, respectively, concomitant with higher IFN-gamma to IL-4 ratios in treated SP-D(-/-) mice, compared to untreated control group. The results showed that SP-D gene deficient mice are more susceptible to IPA while SP-A gene deficient mice acquire resistance to IPA.

Yap is a novel regulator of C2C12 myogenesis.

The expression, regulation and function of mammalian Hippo pathway members in skeletal muscle is largely unknown. The aim of this study was thus to test the hypothesis that core members of the mammalian Hippo pathway are expressed in skeletal muscle and that the transcriptional co-factor Yap, a core member of the Hippo pathway, regulates C2C12 myogenesis. We found that the major components of the mammalian Hippo pathway including Yap are all expressed in skeletal muscles, C2C12 myoblasts and myotubes. In C2C12 myoblasts, Yap Ser127 phosphorylation is low and Yap localises to nuclei. Upon differentiation, Yap Ser127 phosphorylation increases approximately 20-fold and Yap translocates from the nucleus to the cytosol. To test whether the observed increase of Yap Ser127 phosphorylation is required for differentiation we overexpressed hYAP1 S127A, a mutant that can not be phosphorylated at Ser127, in C2C12 myoblasts. We found that overexpression of hYAP S127A prevented myotube formation, whereas the overexpression of wildtype hYAP1 or empty vector had no effect. In addition, more hYAP1 S127A overexpressing cells progressed through the S phase of the cell cycle and the expression of MRFs (myogenin, Myf5), Mef2c and cell cycle regulators (p21, cyclin D1) was significantly changed when compared to wildtype hYAP1 and empty vector overexpressing cells. This data suggests that the phosphorylation of Yap at Ser127 leads to a changed expression of MRFs and cell cycle regulators and is required for C2C12 myoblasts to differentiate into myotubes.

The class A macrophage scavenger receptor type I (SR-AI) recognizes complement iC3b and mediates NF-κB activation.

The macrophage scavenger receptor SR-AI binds to host tissue debris to perform clearance and it binds to bacteria for phagocytosis. In addition, SR-AI modulates macrophage activation through cell signaling. However, investigation of SR-AI signaling on macrophages is complicated due to its promiscuous ligand specificity that overlaps with other macrophage receptors. Therefore, we expressed SR-AI on HEK 293T cells to investigate its ligand binding and signaling. On 293Tcells, SR-AI could respond to E. coli DH5α, leading to NF-κB activation and IL-8 production. However, this requires E. coli DH5α to be sensitized by fresh serum that is treated with heat-inactivation or complement C3 depletion. Anti-C3 antibody inhibits the binding of SR-AI to serum-sensitized DH5α and blocks DH5α stimulation of SR-AI signaling. Further analysis showed that SR-AI can directly bind to purified iC3b but not C3 or C3b. By mutagenesis, The SRCR domain of SR-AI was found to be essential in SR-AI binding to serum-sensitized DH5α. These results revealed a novel property of SR-AI as a complement receptor for iC3b-opsonized bacteria that can elicit cell signaling.

Surfactant protein A (SP-A) binds to phosphatidylserine and competes with annexin V binding on late apoptotic cells.

The role of surfactant protein A (SP-A) in the recognition and clearance of apoptotic cells is well established, but to date, it is still not clear which surface molecules of apoptotic cells are involved in the process. Here we present evidence that phosphatidylserine (PS) is a relevant binding molecule for human SP-A. The binding is Ca(2+)-dependent and is not inhibited by mannose, suggesting that the sugar-binding site of the carbohydrate recognition domain (CRD) of SP-A is not involved. Flow cytometry studies on apoptotic Jurkat cells revealed apparent inhibition of annexin V binding by increasing concentrations of SP-A in late apoptotic but not early apoptotic cells, and this was consistent for Jurkat cells and neutrophils. Supporting these data, confocal microscopy results show a co-localisation of annexin V and SP-A in late apoptotic but not early apoptotic cells. However, we cannot conclude that this inhibition is exclusively due to the binding of SP-A to PS on the cell surface, as annexin V is not wholly specific for PS and SP-A also interacts with other phospholipids that might become exposed on the apoptotic cell surface.

Surface-bound myeloperoxidase is a ligand for recognition of late apoptotic neutrophils by human lung surfactant proteins A and D.

Surfactant proteins A (SP-A) and D (SP-D), both members of the collectin family, play a well established role in apoptotic cell recognition and clearance. Recent in vitro data show that SP-A and SP-D interact with apoptotic neutrophils in a distinct manner. SP-A and SP-D bind in a Ca(2+)-dependent manner to viable and early apoptotic neutrophils whereas the much greater interaction with late apoptotic neutrophils is Ca(2+)-independent. Cell surface molecules on the apoptotic target cells responsible for these interactions had not been identified and this study was done to find candidate target molecules. Myeloperoxidase (MPO), a specific intracellular defense molecule of neutrophils that becomes exposed on the outside of the cell upon apoptosis, was identified by affinity purification, mass-spectrometry and western blotting as a novel binding molecule for SP-A and SP-D. To confirm its role in recognition, it was shown that purified immobilised MPO binds SP-A and SP-D, and that MPO is surface-exposed on late apoptotic neutrophils. SP-A and SP-D inhibit binding of an anti-MPO monoclonal Ab to late apoptotic cells. Fluorescence microscopy confirmed that anti-MPO mAb and SP-A/SP-D colocalise on late apoptotic neutrophils. Desmoplakin was identified as a further potential ligand for SP-A, and neutrophil defensin as a target for both proteins.

The human lung surfactant proteins A (SP-A) and D (SP-D) interact with apoptotic target cells by different binding mechanisms.

The role of the lung surfactant proteins SP-A and SP-D in immune defence is well established. They bind to foreign organisms that invade the lungs and target them for phagocytic clearance by resident alveolar macrophages. SP-A and SP-D also bind to various apoptotic cells and facilitate their phagocytic uptake. To date, the molecular mechanisms by which the lung surfactant proteins interact with apoptotic cells and phagocytes are poorly understood. The aims of this study were to investigate further the interactions between SP-A and SP-D and apoptotic cells using human neutrophils and Jurkat cells as model systems. Specifically the binding behaviour of SP-A and SP-D with viable, early apoptotic and late apoptotic cells was investigated and compared. SP-A and SP-D show very distinct binding to the various cell types. SP-A bound to viable and early apoptotic cells in a predominantly Ca(2+)-dependent manner but the interaction with late apoptotic cells was Ca(2+)-independent, suggesting involvement of other than the lectin- or Ca(2+)-binding sites. This was consistent for neutrophils and Jurkat cells. SP-D in contrast, did not interact with viable and early apoptotic Jurkat cells but strongly and in a Ca(2+)-independent manner with late apoptotic Jurkat cells. SP-D-binding to viable and early apoptotic neutrophils was inhibited by maltose and ethylene-diamin-tetra-acetate (EDTA), suggesting lectin-binding site involvement whereas the binding to late apoptotic neutrophils was predominantly Ca(2+)-independent. These results represent a detailed study of the binding behaviour of SP-A and SP-D with different cell types and stages of viability. The mechanisms of these interactions appear to involve preferential recognition of different ligands on the apoptotic cell surface, which may include nucleic acid, phospholipid, protein and glycan structures.

Analogous interactions in initiating complexes of the classical and lectin pathways of complement.

The classical and lectin pathways of complement activation neutralize pathogens and stimulate key immunological processes. Both pathways are initiated by collagen-containing, soluble pattern recognition molecules associated with specific serine proteases. In the classical pathway, C1q binds to Ab-Ag complexes or bacterial surfaces to activate C1r and C1s. In the lectin pathway, mannan-binding lectin and ficolins bind to carbohydrates on pathogens to activate mannan-binding lectin-associated serine protease 2. To characterize the interactions leading to classical pathway activation, we have analyzed binding between human C1q, C1r, and C1s, which associate to form C1, using full-length and truncated protease components. We show that C1r and C1s bind to C1q independently. The CUB1-epidermal growth factor fragments contribute most toward binding, but CUB2 of C1r, but not of C1s, is also important. Each C1rs tetramer presents a total of six binding sites, one for each of the collagenous domains of C1q. We also demonstrate that subcomponents of the lectin and classical pathways cross-interact. Thus, although the stoichiometries of complexes differ, interactions are analogous, with equivalent contacts between recognition and protease subcomponents. Importantly, these new data are contrary to existing models of C1 and enable us to propose a new model using mannan-binding lectin-mannan-binding lectin-associated serine protease interactions as a template.

A recombinant fragment of human surfactant protein D lacking the short collagen-like stalk fails to correct morphological alterations in lungs of SP-D deficient mice.

Emphysema-like pathology is a characteristic feature of surfactant protein D (SP-D) knock-out mice. Treatment with a recombinant fragment of human SP-D consisting of a short collagen-like stalk (but not the entire collagen-like domain of native SP-D), neck, and carbohydrate recognizing domain (CRD) inhibits development of emphysema-like pathology in SP-D deficient mice. On the other hand, it has been shown that the entire collagen-like domain is necessary for preventing SP-D knock-out mice from pulmonary emphysema development. Thus, in the present study, we aimed to elucidate the role of the short collagen-like stalk for the function of the recombinant fragment of human SP-D. We treated SP-D knock-out mice with a fragment of human SP-D lacking the short collagen-like stalk and compared the effects on lung morphology with results from untreated wild-type and SP-D knock-out mice and from SP-D knock-out mice treated with a recombinant fragment of human SP-D including the short collagen-like stalk. The fragment of SP-D lacking the short collagen-like stalk failed to correct pulmonary emphysematous alterations demonstrating the importance of the short collagen-like stalk for the biological activity of the recombinant fragment of human SP-D.