Self-Supervised Machine Learning to Characterise Step Counts from Wrist-Worn Accelerometers in the UK Biobank.

PURPOSE: Step count is an intuitive measure of physical activity frequently quantified in health-related studies; however, accurate step counting is difficult in the free-living environment, with error routinely above 20% in wrist-worn devices against camera-annotated ground truth. This study aims to describe the development and validation of step count derived from a wrist-worn accelerometer and assess its association with cardiovascular and all-cause mortality in a large prospective cohort. METHODS: We developed and externally validated a self-supervised machine learning step detection model, trained on an open-source and step-annotated free-living dataset. 39 individuals will free-living ground-truth annotated step counts were used for model development. An open-source dataset with 30 individuals was used for external validation. Epidemiological analysis was performed using 75,263 UK Biobank participants without prevalent cardiovascular disease (CVD) or cancer. Cox regression was used to test the association of daily step count with fatal CVD and all-cause mortality after adjustment for potential confounders. RESULTS: The algorithm substantially outperformed reference models (free-living mean absolute percent error of 12.5%, versus 65-231%). Our data indicate an inverse dose-response association, where taking 6,430-8,277 daily steps was associated with 37% [25-48%] and 28% [20-35%] lower risk of fatal CVD and all-cause mortality up to seven years later, compared to those taking fewer steps each day. CONCLUSIONS: We have developed an open and transparent method that markedly improves the measurement of steps in large-scale wrist-worn accelerometer datasets. The application of this method demonstrated expected associations with CVD and all-cause mortality, indicating excellent face validity. This reinforces public health messaging for increasing physical activity and can help lay the groundwork for the inclusion of target step counts in future public health guidelines.

ORF3c is expressed in SARS-CoV-2-infected cells and inhibits innate sensing by targeting MAVS.

Most SARS-CoV-2 proteins are translated from subgenomic RNAs (sgRNAs). While the majority of these sgRNAs are monocistronic, some viral mRNAs encode more than one protein. One example is the ORF3a sgRNA that also encodes ORF3c, an enigmatic 41-amino-acid peptide. Here, we show that ORF3c is expressed in SARS-CoV-2-infected cells and suppresses RIG-I- and MDA5-mediated IFN-ß induction. ORF3c interacts with the signaling adaptor MAVS, induces its C-terminal cleavage, and inhibits the interaction of RIG-I with MAVS. The immunosuppressive activity of ORF3c is conserved among members of the subgenus sarbecovirus, including SARS-CoV and coronaviruses isolated from bats. Notably, however, the SARS-CoV-2 delta and kappa variants harbor premature stop codons in ORF3c, demonstrating that this reading frame is not essential for efficient viral replication in vivo and is likely compensated by other viral proteins. In agreement with this, disruption of ORF3c does not significantly affect SARS-CoV-2 replication in CaCo-2, CaLu-3, or Rhinolophus alcyone cells. In summary, we here identify ORF3c as an immune evasion factor of SARS-CoV-2 that suppresses innate sensing in infected cells.

Quantitative nanohistology of aging dermal collagen.

The skin is the largest organ in the body and is essential for protecting us from environmental stressors such as UV radiation, pollution, and pathogens. As we age, our skin undergoes complex changes that can affect its function, appearance, and health. These changes result from intrinsic (chronological) and extrinsic (environmental) factors that can cause damage to the skin's cells and extracellular matrix. As higher-resolution microscopical techniques, such as Atomic Force Microscopy (AFM), are being deployed to support histology, it is possible to explore the biophysical properties of the dermal scaffold's constituents, such as the collagen network. In this study, we demonstrate the use of our AFM-based quantitative nanohistology, performed directly on unfixed cryosections of 30 donors (female, Caucasian), to differentiate between dermal collagen from different age groups and anatomical sites. The initial 420 (10 × 10 µm2) Atomic Force Microscopy images were segmented into 42,000 (1 × 1 µm2) images before being classified according to four pre-defined empirical collagen structural biomarkers to quantify the structural heterogeneity of the dermal collagen. These markers include interfibrillar gap formation, undefined collagen structure, and registered or unregistered dense collagen fibrillar network with evident D-banding. The structural analysis was also complemented by extensive nanoindentation (∼1,000 curves) performed on individual fibrils from each section, yielding 30,000 indentation curves for this study. Principal Component Analysis was used to reduce the complexity of high-dimensional datasets. The % prevalence of the empirical collagen structural biomarkers between the papillary and reticular dermis for each section proves determinant in differentiating between the donors as a function of their age or the anatomical site (cheek or breast). A case of abnormal biological aging validated our markers and nanohistology approach. This case also highlighted the difference between chronological and biological aging regarding dermal collagen phenotyping. However, quantifying the impact of chronic and pathological conditions on the structure and function of collagen at the sub-micron level remains challenging and lengthy. By employing tools such as the Atomic Force Microscope as presented here, it is possible to start evaluating the complexity of the dermal matrix at the nanoscale and start identifying relevant collagen morphology which could be used toward histopathology standards.

Development and Validation of a Machine Learning Wrist-worn Step Detection Algorithm with Deployment in the UK Biobank.

Background: Step count is an intuitive measure of physical activity frequently quantified in a range of health-related studies; however, accurate quantification of step count can be difficult in the free-living environment, with step counting error routinely above 20% in both consumer and research-grade wrist-worn devices. This study aims to describe the development and validation of step count derived from a wrist-worn accelerometer and to assess its association with cardiovascular and all-cause mortality in a large prospective cohort study. Methods: We developed and externally validated a hybrid step detection model that involves self-supervised machine learning, trained on a new ground truth annotated, free-living step count dataset (OxWalk, n=39, aged 19-81) and tested against other open-source step counting algorithms. This model was applied to ascertain daily step counts from raw wrist-worn accelerometer data of 75,493 UK Biobank participants without a prior history of cardiovascular disease (CVD) or cancer. Cox regression was used to obtain hazard ratios and 95% confidence intervals for the association of daily step count with fatal CVD and all-cause mortality after adjustment for potential confounders. Findings: The novel step algorithm demonstrated a mean absolute percent error of 12.5% in free-living validation, detecting 98.7% of true steps and substantially outperforming other recent wrist-worn, open-source algorithms. Our data are indicative of an inverse dose-response association, where, for example, taking 6,596 to 8,474 steps per day was associated with a 39% [24-52%] and 27% [16-36%] lower risk of fatal CVD and all-cause mortality, respectively, compared to those taking fewer steps each day. Interpretation: An accurate measure of step count was ascertained using a machine learning pipeline that demonstrates state-of-the-art accuracy in internal and external validation. The expected associations with CVD and all-cause mortality indicate excellent face validity. This algorithm can be used widely for other studies that have utilised wrist-worn accelerometers and an open-source pipeline is provided to facilitate implementation.

Violin Varnishes: Microstructure and Nanomechanical Analysis.

The aim of the current work is twofold: to demonstrate the application of in situ non-invasive imaging by portable atomic force microscopy (AFM) on the surfaces of a violin and to integrate compositional and mechanical analysis at the nano scale level on model samples of varnished wood. These samples were prepared according to traditional recipes by an Italian lute-maker family well practised in the art. Samples of oil and spirit-based varnishes on maple wood, naturally and accelerated light aged, were studied. AFM was used to measure the nanomechanical properties of the model samples and established that the spirit-based varnish was stiffer than the oil-based. Synchrotron radiation micro- Fourier Transform Infra-red analysis of the layer structure revealed that stiffer spirit-based varnish showed less penetration into the wood than the oil-based. Further PeakForce Quantitative Nanomechanical Mapping (QNM) demonstrated a difference in adhesion values between the oil- and spirit-based samples.

Phenotypic Properties of Collagen in Dentinogenesis Imperfecta Associated with Osteogenesis Imperfecta.

INTRODUCTION: Dentinogenesis imperfecta type 1 (OIDI) is considered a relatively rare genetic disorder (1:5000 to 1:45,000) associated with osteogenesis imperfecta. OIDI impacts the formation of collagen fibrils in dentin, leading to morphological and structural changes that affect the strength and appearance of teeth. However, there is still a lack of understanding regarding the nanoscale characterization of the disease, in terms of collagen ultrastructure and mechanical properties. Therefore, this research presents a qualitative and quantitative report into the phenotype and characterization of OIDI in dentin, by using a combination of imaging, nanomechanical approaches. METHODS: For this study, 8 primary molars from OIDI patients and 8 primary control molars were collected, embedded in acrylic resin and cut into longitudinal sections. Sections were then demineralized in 37% phosphoric acid using a protocol developed in-house. Initial experiments demonstrated the effectiveness of the demineralization protocol, as the ATR-FTIR spectral fingerprints showed an increase in the amide bands together with a decrease in phosphate content. Structural and mechanical analyses were performed directly on both the mineralized and demineralized samples using a combination of scanning electron microscopy, atomic force microscopy, and Wallace indentation. RESULTS: Mesoscale imaging showed alterations in dentinal tubule morphology in OIDI patients, with a reduced number of tubules and a decreased tubule diameter compared to healthy controls. Nanoscale collagen ultrastructure presented a similar D-banding periodicity between OIDI and controls. Reduced collagen fibrils diameter was also recorded for the OIDI group. The hardness of the (mineralized) control dentin was found to be significantly higher (p<0.05) than that of the OIDI (mineralized) dentine. Both the exposed peri- and intratubular dentinal collagen presented bimodal elastic behaviors (Young's moduli). The control samples presented a stiffening of the intratubular collagen when compared to the peritubular collagen. In case of the OIDI, this stiffening in the collagen between peri- and intratubular dentinal collagen was not observed and the exposed collagen presented overall a lower elasticity than the control samples. CONCLUSION: This study presents a systematic approach to the characterization of collagen structure and properties in OIDI as diagnosed in dentin. Structural markers for OIDI at the mesoscale and nanoscale were found and correlated with an observed lack of increased elastic moduli of the collagen fibrils in the intratubular OIDI dentin. These findings offer an explanation of how structural changes in the dentin could be responsible for the failure of some adhesive restorative materials as observed in patients affected by OIDI.

An engineered, quantifiable in vitro model for analysing the effect of proteostasis-targeting drugs on tissue physical properties.

Cellular function depends on the maintenance of protein homeostasis (proteostasis) by regulated protein degradation. Chronic dysregulation of proteostasis is associated with neurodegenerative and age-related diseases, and drugs targeting components of the protein degradation apparatus are increasingly used in cancer therapies. However, as chronic imbalances rather than loss of function mediate their pathogenesis, research models that allow for the study of the complex effects of drugs on tissue properties in proteostasis-associated diseases are almost completely lacking. Here, to determine the functional effects of impaired proteostatic fine-tuning, we applied a combination of materials science characterisation techniques to a cell-derived, in vitro model of bone-like tissue formation in which we pharmacologically perturbed protein degradation. We show that low-level inhibition of VCP/p97 and the proteasome, two major components of the degradation machinery, have remarkably different effects on the bone-like material that human bone-marrow derived mesenchymal stromal cells (hMSC) form in vitro. Specifically, whilst proteasome inhibition mildly enhances tissue formation, Raman spectroscopic, atomic force microscopy-based indentation, and electron microscopy imaging reveal that VCP/p97 inhibition induces the formation of bone-like tissue that is softer, contains less protein, appears to have more crystalline mineral, and may involve aberrant micro- and ultra-structural tissue organisation. These observations contrast with findings from conventional osteogenic assays that failed to identify any effect on mineralisation. Taken together, these data suggest that mild proteostatic impairment in hMSC alters the bone-like material they form in ways that could explain some pathologies associated with VCP/p97-related diseases. They also demonstrate the utility of quantitative materials science approaches for tackling long-standing questions in biology and medicine, and could form the basis for preclinical drug testing platforms to develop therapies for diseases stemming from perturbed proteostasis or for cancer therapies targeting protein degradation. Our findings may also have important implications for the field of tissue engineering, as the manufacture of cell-derived biomaterial scaffolds may need to consider proteostasis to effectively replicate native tissues.

Quantitative nanohistological investigation of scleroderma: an atomic force microscopy-based approach to disease characterization.

Scleroderma (or systemic sclerosis, SSc) is a disease caused by excess crosslinking of collagen. The skin stiffens and becomes painful, while internally, organ function can be compromised by the less elastic collagen. Diagnosis of SSc is often only possible in advanced cases by which treatment time is limited. A more detailed analysis of SSc may provide better future treatment options and information of disease progression. Recently, the histological stain picrosirius red showing collagen register has been combined with atomic force microscopy (AFM) to study SSc. Skin from healthy individuals and SSc patients was biopsied, stained and studied using AFM. By investigating the crosslinking of collagen at a smaller hierarchical stage, the effects of SSc were more pronounced. Changes in morphology and Young's elastic modulus were observed and quantified; giving rise to a novel technique, we have termed "quantitative nanohistology". An increase in nanoscale stiffness in the collagen for SSc compared with healthy individuals was seen by a significant increase in the Young's modulus profile for the collagen. These markers of stiffer collagen in SSc are similar to the symptoms experienced by patients, giving additional hope that in the future, nanohistology using AFM can be readily applied as a clinical tool, providing detailed information of the state of collagen.

Demineralization-remineralization dynamics in teeth and bone.

Biomineralization is a dynamic, complex, lifelong process by which living organisms control precipitations of inorganic nanocrystals within organic matrices to form unique hybrid biological tissues, for example, enamel, dentin, cementum, and bone. Understanding the process of mineral deposition is important for the development of treatments for mineralization-related diseases and also for the innovation and development of scaffolds. This review provides a thorough overview of the up-to-date information on the theories describing the possible mechanisms and the factors implicated as agonists and antagonists of mineralization. Then, the role of calcium and phosphate ions in the maintenance of teeth and bone health is described. Throughout the life, teeth and bone are at risk of demineralization, with particular emphasis on teeth, due to their anatomical arrangement and location. Teeth are exposed to food, drink, and the microbiota of the mouth; therefore, they have developed a high resistance to localized demineralization that is unmatched by bone. The mechanisms by which demineralization-remineralization process occurs in both teeth and bone and the new therapies/technologies that reverse demineralization or boost remineralization are also scrupulously discussed. Technologies discussed include composites with nano- and micron-sized inorganic minerals that can mimic mechanical properties of the tooth and bone in addition to promoting more natural repair of surrounding tissues. Turning these new technologies to products and practices would improve health care worldwide.

Regulation of post-Golgi LH3 trafficking is essential for collagen homeostasis.

Post-translational modifications are necessary for collagen precursor molecules (procollagens) to acquire final shape and function. However, the mechanism and contribution of collagen modifications that occur outside the endoplasmic reticulum and Golgi are not understood. We discovered that VIPAR, with its partner proteins, regulate sorting of lysyl hydroxylase 3 (LH3, also known as PLOD3) into newly identified post-Golgi collagen IV carriers and that VIPAR-dependent sorting is essential for modification of lysines in multiple collagen types. Identification of structural and functional collagen abnormalities in cells and tissues from patients and murine models of the autosomal recessive multisystem disorder Arthrogryposis, Renal dysfunction and Cholestasis syndrome caused by VIPAR and VPS33B deficiencies confirmed our findings. Thus, regulation of post-Golgi LH3 trafficking is essential for collagen homeostasis and for the development and function of multiple organs and tissues.

Optical coherence tomography use in the diagnosis of enamel defects.

Molar incisor hypomineralization (MIH) affects the permanent incisors and molars, whose undermineralized matrix is evidenced by lesions ranging from white to yellow/brown opacities to crumbling enamel lesions incapable of withstanding normal occlusal forces and function. Diagnosing the condition involves clinical and radiographic examination of these teeth, with known limitations in determining the depth extent of the enamel defects in particular. Optical coherence tomography (OCT) is an emerging hard and soft tissue imaging technique, which was investigated as a new potential diagnostic method in dentistry. A comparison between the diagnostic potential of the conventional methods and OCT was conducted. Compared to conventional imaging methods, OCT gave more information on the structure of the enamel defects as well as the depth extent of the defects into the enamel structure. Different types of enamel defects were compared, each type presenting a unique identifiable pattern when imaged using OCT. Additionally, advanced methods of OCT image analysis including backscattered light intensity profile analysis and enface reconstruction were performed. Both methods confirmed the potential of OCT in enamel defects diagnosis. In conclusion, OCT imaging enabled the identification of the type of enamel defect and the determination of the extent of the enamel defects in MIH with the advantage of being a radiation free diagnostic technique.