Genetic syndromes screening by facial recognition technology: VGG-16 screening model construction and evaluation.

BACKGROUND: Many genetic syndromes (GSs) have distinct facial dysmorphism, and facial gestalts can be used as a diagnostic tool for recognizing a syndrome. Facial recognition technology has advanced in recent years, and the screening of GSs by facial recognition technology has become feasible. This study constructed an automatic facial recognition model for the identification of children with GSs. RESULTS: A total of 456 frontal facial photos were collected from 228 children with GSs and 228 healthy children in Guangdong Provincial People's Hospital from Jun 2016 to Jan 2021. Only one frontal facial image was selected for each participant. The VGG-16 network (named after its proposal lab, Visual Geometry Group from Oxford University) was pretrained by transfer learning methods, and a facial recognition model based on the VGG-16 architecture was constructed. The performance of the VGG-16 model was evaluated by five-fold cross-validation. Comparison of VGG-16 model to five physicians were also performed. The VGG-16 model achieved the highest accuracy of 0.8860 ± 0.0211, specificity of 0.9124 ± 0.0308, recall of 0.8597 ± 0.0190, F1-score of 0.8829 ± 0.0215 and an area under the receiver operating characteristic curve of 0.9443 ± 0.0276 (95% confidence interval: 0.9210-0.9620) for GS screening, which was significantly higher than that achieved by human experts. CONCLUSIONS: This study highlighted the feasibility of facial recognition technology for GSs identification. The VGG-16 recognition model can play a prominent role in GSs screening in clinical practice.

Automated Facial Recognition for Noonan Syndrome Using Novel Deep Convolutional Neural Network With Additive Angular Margin Loss.

BACKGROUND: Noonan syndrome (NS), a genetically heterogeneous disorder, presents with hypertelorism, ptosis, dysplastic pulmonary valve stenosis, hypertrophic cardiomyopathy, and small stature. Early detection and assessment of NS are crucial to formulating an individualized treatment protocol. However, the diagnostic rate of pediatricians and pediatric cardiologists is limited. To overcome this challenge, we propose an automated facial recognition model to identify NS using a novel deep convolutional neural network (DCNN) with a loss function called additive angular margin loss (ArcFace). METHODS: The proposed automated facial recognition models were trained on dataset that included 127 NS patients, 163 healthy children, and 130 children with several other dysmorphic syndromes. The photo dataset contained only one frontal face image from each participant. A novel DCNN framework with ArcFace loss function (DCNN-Arcface model) was constructed. Two traditional machine learning models and a DCNN model with cross-entropy loss function (DCNN-CE model) were also constructed. Transfer learning and data augmentation were applied in the training process. The identification performance of facial recognition models was assessed by five-fold cross-validation. Comparison of the DCNN-Arcface model to two traditional machine learning models, the DCNN-CE model, and six physicians were performed. RESULTS: At distinguishing NS patients from healthy children, the DCNN-Arcface model achieved an accuracy of 0.9201 ± 0.0138 and an area under the receiver operator characteristic curve (AUC) of 0.9797 ± 0.0055. At distinguishing NS patients from children with several other genetic syndromes, it achieved an accuracy of 0.8171 ± 0.0074 and an AUC of 0.9274 ± 0.0062. In both cases, the DCNN-Arcface model outperformed the two traditional machine learning models, the DCNN-CE model, and six physicians. CONCLUSION: This study shows that the proposed DCNN-Arcface model is a promising way to screen NS patients and can improve the NS diagnosis rate.

Automatic Facial Recognition of Williams-Beuren Syndrome Based on Deep Convolutional Neural Networks.

Background: Williams-Beuren syndrome (WBS) is a rare genetic syndrome with a characteristic "elfin" facial gestalt. The "elfin" facial characteristics include a broad forehead, periorbital puffiness, flat nasal bridge, short upturned nose, wide mouth, thick lips, and pointed chin. Recently, deep convolutional neural networks (CNNs) have been successfully applied to facial recognition for diagnosing genetic syndromes. However, there is little research on WBS facial recognition using deep CNNs. Objective: The purpose of this study was to construct an automatic facial recognition model for WBS diagnosis based on deep CNNs. Methods: The study enrolled 104 WBS children, 91 cases with other genetic syndromes, and 145 healthy children. The photo dataset used only one frontal facial photo from each participant. Five face recognition frameworks for WBS were constructed by adopting the VGG-16, VGG-19, ResNet-18, ResNet-34, and MobileNet-V2 architectures, respectively. ImageNet transfer learning was used to avoid over-fitting. The classification performance of the facial recognition models was assessed by five-fold cross validation, and comparison with human experts was performed. Results: The five face recognition frameworks for WBS were constructed. The VGG-19 model achieved the best performance. The accuracy, precision, recall, F1 score, and area under curve (AUC) of the VGG-19 model were 92.7 ± 1.3%, 94.0 ± 5.6%, 81.7 ± 3.6%, 87.2 ± 2.0%, and 89.6 ± 1.3%, respectively. The highest accuracy, precision, recall, F1 score, and AUC of human experts were 82.1, 65.9, 85.6, 74.5, and 83.0%, respectively. The AUCs of each human expert were inferior to the AUCs of the VGG-16 (88.6 ± 3.5%), VGG-19 (89.6 ± 1.3%), ResNet-18 (83.6 ± 8.2%), and ResNet-34 (86.3 ± 4.9%) models. Conclusions: This study highlighted the possibility of using deep CNNs for diagnosing WBS in clinical practice. The facial recognition framework based on VGG-19 could play a prominent role in WBS diagnosis. Transfer learning technology can help to construct facial recognition models of genetic syndromes with small-scale datasets.

B Cell Tolerance to Deiminated Histones in BALB/c, C57BL/6, and Autoimmune-Prone Mouse Strains.

Deimination, a posttranslational modification of arginine to citrulline carried out by peptidylarginine deiminases, may compromise tolerance of self-antigens. Patients with connective tissue autoimmunity, particularly rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), or Felty's syndrome, present with autoantibodies to deiminated histones (dH), which thus form a category of antibodies to citrullinated protein antigens (ACPA). In general, ACPA are a sensitive diagnostic for RA and may form in response to the release of nuclear chromatin (DNA plus dH) from granulocytes, usually referred to as neutrophil extracellular traps. The aim of this study was to examine spontaneously autoimmune mice for autoantibodies and T cell responses to dH. We compared IgG binding to deiminated and non-deiminated histones (nH) by ELISA and Western blotting in spontaneously autoimmune strains of (NZB × NZW) F1 and NZM2410 together with their derivative congenic strains, C57BL/6.Sle1 and C57BL/6.Sle1.Sle3, which display profound autoreactivity against nuclear self-antigens. The splenocyte proliferation against the two antigens was determined in the spontaneously autoimmune (NZB × NZW) F1 strain from which other autoimmune strains used in the study were derived. Immunizations with dH and nH were attempted in BALB/c mice to assess their splenocyte response. Splenocytes from BALB/c mice and from autoimmune mice at the time of conversion to autoimmunity proliferated strongly in response to dH, yet serum IgG from autoimmune (NZB × NZW) F1, NZM2410, and C57BL/6.Sle1.Sle3 mice displayed a remarkable bias against binding to dH. At the time of seroconversion, the antibodies already exhibited preference for nH, and only nH were recovered from circulating immune complexes. Analysis of histone deimination showed constitutive deimination in thymic extracts from C57BL/6 and C57BL/6.Sle1.Sle2.Sle3 triply congenic mice and in spleens of autoimmune triply congenic mice. Our study demonstrates that tolerance mechanisms against dH are intact in BALB/c and C57BL/6 mice and continue to be effective in mice with overt autoimmunity to nH. We conclude that, in contrast to human RA and SLE patients, where we frequently observe autoantibodies against dH, autoimmune mice maintain strong tolerance mechanisms to prevent the development of autoantibodies to dH.

Genetic and cellular dissection of the activation of AM14 rheumatoid factor B cells in a mouse model of lupus.

The RF-specific AM14 tg BCR has been used as a model to dissect the mechanisms of B cell tolerance to ICs containing nucleic acids. We have shown previously that AM14 RF B cells break tolerance in the TC mouse model of lupus through the dual engagement of the AM14 BCR and TLR9. In this study, we showed that neither the expression of Sle1 or Sle2 susceptibility loci alone was sufficient to activate AM14 RF B cells, suggesting that the production of antichromatin IgG2a(a) autoAg mediated by Sle1 and an intrinsically higher B cell activation mediated by Sle2 were required. We also showed that the B6 genetic background enhanced the selection of AM14 RF B cells to the MZB cell compartment regardless of the expression of the Sle loci and therefore, of their activation into AFCs. Furthermore, some AM14 RF B cells were selected into the B-1a compartment, where they did not differentiate into AFCs. Therefore, it is unlikely that the selection of AM14 RF B cells to the MZB or B-1a cell compartments in TC.AM14(a) mice is responsible for their breach of tolerance. Finally, we showed that the presence of expression of Sle1 in non-tg cells, most likely T cells, is necessary for the activation of AM14 RF B cells into AFCs. Overall, these results suggest a threshold model of activation of AM14 RF B cells on the B6 background with additive genetic and cellular contribution of multiple sources.

Dysregulated cytokine production by dendritic cells modulates B cell responses in the NZM2410 mouse model of lupus.

The breakdown in tolerance of autoreactive B cells in the lupus-prone NZM2410-derived B6.Sle1.Sle2.Sle3 (TC) mice results in the secretion of autoantibodies. TC dendritic cells (DCs) enhance B cell proliferation and antibody secretion in a cytokine-dependent manner. However, the specific cytokine milieu by which TC DCs activate B cells was not known. In this study, we compared TC and C57BL/6 (B6) control for the distribution of DC subsets and for their production of cytokines affecting B cell responses. We show that TC DCs enhanced B cell proliferation through the production of IL-6 and IFN-γ, while antibody secretion was only dependent on IL-6. Pre-disease TC mice showed an expanded PDCA1(+) cells prior to disease onset that was localized to the marginal zone and further expanded with age. The presence of PDCA1(+) cells in the marginal zone correlated with a Type I Interferon (IFN) signature in marginal zone B cells, and this response was higher in TC than B6 mice. In vivo administration of anti-chromatin immune complexes upregulated IL-6 and IFN-γ production by splenic DCs from TC but not B6 mice. The production of BAFF and APRIL was decreased upon TC DC stimulation both in vitro and in vivo, indicating that these B cell survival factors do not play a role in B cell modulation by TC DCs. Finally, TC B cells were defective at downregulating IL-6 expression in response to anti-inflammatory apoptotic cell exposure. Overall, these results show that the TC autoimmune genetic background induces the production of B cell-modulating inflammatory cytokines by DCs, which are regulated by the microenvironment as well as the interplay between DC.

A novel pathway of rapid TLR-triggered activation of integrin-dependent leukocyte adhesion that requires Rap1 GTPase.

Rapid ß2-integrin activation is indispensable for leukocyte adhesion and recruitment to sites of infection and is mediated by chemokine- or P-selectin glycoprotein ligand-1-induced inside-out signaling. Here we uncovered a novel pathway for rapid activation of integrin-dependent leukocyte adhesion, triggered by toll-like receptor (TLR)-mediated signaling. TLR2 or TLR5 ligation rapidly activated integrin-dependent leukocyte adhesion to immobilized ICAM-1 and fibronectin. Consistently, in vivo administration of the TLR2-ligand Pam3CSK4 increased integrin-dependent slow rolling and adhesion to endothelium within minutes, as identified by intravital microscopy in the cremaster model. TLR2 and TLR5 ligation increased ß2-integrin affinity, as assessed by the detection of activation-dependent neoepitopes. TLR2- and TLR5-triggered integrin activation in leukocytes required enhanced Rap1 GTPase activity, which was mediated by Rac1 activation and NADPH oxidase-2-dependent reactive oxygen species production. This novel direct pathway linking initial pathogen recognition by TLRs to rapid ß2-integrin activation may critically regulate acute leukocyte infiltration to sites of pathogen invasion.

Activation of rheumatoid factor-specific B cells is antigen dependent and occurs preferentially outside of germinal centers in the lupus-prone NZM2410 mouse model.

AM14 rheumatoid factor (RF) B cells in the MRL/lpr mice are activated by dual BCR and TLR7/9 ligation and differentiate into plasmablasts via an extrafollicular (EF) route. It was not known whether this mechanism of activation of RF B cells applied to other lupus-prone mouse models. We investigated the mechanisms by which RF B cells break tolerance in the NZM2410-derived B6.Sle1.Sle2.Sle3 (TC) strain in comparison with C57BL/6 (B6) controls, each expressing the AM14 H chain transgene in the presence or absence of the IgG2a(a) autoantigen. The TC, but not B6, genetic background promotes the differentiation of RF B cells into Ab-forming cells (AFCs) in the presence of the autoantigen. Activated RF B cells preferentially differentiated into plasmablasts in EF zones. Contrary to the MRL/lpr strain, TC RF B cells were also located within germinal centers, but only the formation of EF foci was positively correlated with the production of RF AFCs. Immunization of young TC.AM14 H chain transgenic mice with IgG2a(a) anti-chromatin immune complexes (ICs) activated RF B cells in a BCR- and TLR9-dependent manner. However, these IC immunizations did not result in the production of RF AFCs. These results show that RF B cells break tolerance with the same general mechanisms in the TC and the MRL/lpr lupus-prone genetic backgrounds, namely the dual activation of the BCR and TLR9 pathways. There are also distinct differences, such as the presence of RF B cells in GCs and the requirement of chronic IgG2a(a) anti-chromatin ICs for full differentiation of RF AFCs.

Contributions of B cells to lupus pathogenesis.

Systemic lupus erythematosus (SLE) is an autoimmune disease characterized by the production of autoantibodies. This review summarizes first the results obtained in the mouse that have revealed how B cell tolerance is breached in SLE. We then review the B cell subsets, in addition to the autoAb producing cells, which contribute to SLE pathogenesis, focusing on marginal zone B cells, B-1 cells and regulatory B cells. Finally, we review the interactions between B cells and other immune cells that have been implicated in SLE, such as dendritic cells, macrophages, neutrophils and T cells.

The granulocyte colony stimulating factor pathway regulates autoantibody production in a murine induced model of systemic lupus erythematosus.

INTRODUCTION: An NZB-derived genetic locus (Sle2c2) that suppresses autoantibody production in a mouse model of induced systemic lupus erythematosus contains a polymorphism in the gene encoding the G-CSF receptor. This study was designed to test the hypothesis that the Sle2c2 suppression is associated with an impaired G-CSF receptor function that can be overcome by exogenous G-CSF. METHODS: Leukocytes from B6.Sle2c2 and B6 congenic mice, which carry a different allele of the G-CSF receptor, were compared for their responses to G-CSF. Autoantibody production was induced with the chronic graft-versus-host-disease (cGVHD) model by adoptive transfer of B6.bm12 splenocytes. Different treatment regimens varying the amount and frequency of G-CSF (Neulasta®) or carrier control were tested on cGVHD outcomes. Autoantibody production, immune cell activation, and reactive oxygen species (ROS) production were compared between the two strains with the various treatments. In addition, the effect of G-CSF treatment was examined on the production autoantibodies in the B6.Sle1.Sle2.Sle3 (B6.TC) spontaneous model of lupus. RESULTS: B6.Sle2c2 and B6 leukocytes responded differently to G-CSF. G-CSF binding by B6.Sle2c2 leukocytes was reduced as compared to B6, which was associated with a reduced expansion in response to in vivo G-CSF treatment. G-CSF in vivo treatment also failed to mobilize bone-marrow B6.Sle2c2 neutrophils as it did for B6 neutrophils. In contrast, the expression of G-CSF responsive genes indicated a higher G-CSF receptor signaling in B6.Sle2c2 cells. G-CSF treatment restored the ability of B6.Sle2c2 mice to produce autoantibodies in a dose-dependent manner upon cGVHD induction, which correlated with restored CD4+ T cells activation, as well as dendritic cell and granulocyte expansion. Steady-state ROS production was higher in B6.Sle2c2 than in B6 mice. cGVHD induction resulted in a larger increase in ROS production in B6 than in B6.Sle2c2 mice, and this difference was eliminated with G-CSF treatment. Finally, a low dose G-CSF treatment accelerated the production of anti-dsDNA IgG in young B6.TC mice. CONCLUSION: The different in vivo and in vitro responses of B6.Sle2c2 leukocytes are consistent with the mutation in the G-CSFR having functional consequences. The elimination of Sle2c2 suppression of autoantibody production by exogenous G-CSF indicates that Sle2c2 corresponds to a loss of function of G-CSF receptor. This result was corroborated by the increased anti-dsDNA IgG production in G-CSF-treated B6.TC mice, which also carry the Sle2c2 locus. Overall, these results suggest that the G-CSF pathway regulates the production of autoantibodies in murine models of lupus.

Animal models of molecular pathology systemic lupus erythematosus.

Systemic lupus erythematosus (SLE) is an autoimmune disease that affects multiple organ systems. A hallmark of SLE is the production of antinuclear antibodies against nuclear antigens such as chromatin and DNA. High levels of autoAbs promote the formation of immune complexes which can lead to the development of glomerulonephritis and progress to end-stage renal failure. Although the exact etiology of SLE is unknown, it is thought to be multifactorial in nature. A combination of environmental, hormonal, and a predisposed genetic background lead to the development of this disorder. Here, we review the various mouse models that have been used to study SLE and discuss how their study has led to a better understanding of the genetic and cellular factors involved in the development of systemic autoimmunity and lupus-like clinical symptoms. We also review the mouse studies that have explored the molecular pathways that are altered in this disease and the investigation of their therapeutic potentials.

Murine models of systemic lupus erythematosus.

Systemic lupus erythematosus (SLE) is a multifactorial autoimmune disorder. The study of diverse mouse models of lupus has provided clues to the etiology of SLE. Spontaneous mouse models of lupus have led to identification of numerous susceptibility loci from which several candidate genes have emerged. Meanwhile, induced models of lupus have provided insight into the role of environmental factors in lupus pathogenesis as well as provided a better understanding of cellular mechanisms involved in the onset and progression of disease. The SLE-like phenotypes present in these models have also served to screen numerous potential SLE therapies. Due to the complex nature of SLE, it is necessary to understand the effect specific targeted therapies have on immune homeostasis. Furthermore, knowledge gained from mouse models will provide novel therapy targets for the treatment of SLE.

Autoreactive marginal zone B cells enter the follicles and interact with CD4+ T cells in lupus-prone mice.

BACKGROUND: Marginal zone B cells have been implicated in the production of autoantibodies in murine models of lupus. It has been suggested that they contribute to lupus immunopathogenesis through their enhanced effector functions and their repertoire that is biased toward autoreactive specificities. In the B6.NZM2410.Sle.Sle2.Sle3 (B6.TC) model of lupus, the majority of marginal zone B cells are located outside the marginal zone and inside the follicles. Genetic alterations of this strain have shown a correlation between autoimmune pathogenesis and the presence of intrafollicular marginal zone B cells. This study was designed first to strengthen our original observations and to determine how the marginal zone B cells from the lupus-prone mice respond to stimulations and interact with T cells. RESULTS: The intrafollicular location of B6.TC MZB cells starts before disease manifestations and puts MZB cells in direct contact with CD4+ T cells. Two different autoreactive B cell receptor (BCR) transgenic models showed that the expression of the Sle susceptibility loci enhances the presence of MZB cells inside the follicles. In vitro, B6.TC MZB cells were better effectors than B6 MZB cells with enhanced proliferation and antibody (Ab) production, including anti-DNA Ab, in response to stimulation with TLR ligands, immune complexes or anti-CD40. Furthermore, B6.TC MZB and CD4+ T cells showed a reciprocally enhanced activation, which indicated that their contacts inside B6.TC follicles have functional consequences that suggest an amplification loop between these two cell types. CONCLUSIONS: These results showed that the NZM2410 susceptibility loci induce MZB cells to locate into the follicles, and that this breach of follicular exclusion occurs early in the development of the autoimmune pathogenesis. The enhanced responses to stimulation and increased effector functions of MZB cells from lupus-prone mice as compare to non-autoimmune MZB cells provide a mechanism by which the failure of MZB cell follicular exclusion contributes to the autoimmune process.

Hepatitis delta antigen requires a minimum length of the hepatitis delta virus unbranched rod RNA structure for binding.

Hepatitis delta virus (HDV) is a subviral pathogen that increases the severity of liver disease caused by hepatitis B virus. Both the small circular RNA genome and its complement, the antigenome, form a characteristic unbranched rod structure in which approximately 70% of the nucleotides are base paired. These RNAs are associated with the sole virally encoded protein, hepatitis delta antigen (HDAg), in infected cells; however, the nature of the ribonucleoprotein complexes (RNPs) is not well understood. Previous analyses of binding in vitro using native, bacterially expressed HDAg have been hampered by a lack of specificity for HDV RNA. Here, we show that removal of the C-terminal 35 amino acids of HDAg yields a native, bacterially expressed protein, HDAg-160, that specifically binds HDV unbranched rod RNA with high affinity. In an electrophoretic mobility shift assay, this protein produced a discrete, micrococcal nuclease-resistant complex with an approximately 400-nucleotide (nt) segment of HDV unbranched rod RNA. Binding occurred with several segments of HDV RNA, although with various affinities and efficiencies. Analysis of the effects of deleting segments of the unbranched rod indicated that binding did not require one or two specific binding sites within these RNA segments. Rather, a minimum-length HDV RNA unbranched rod approximately 311 nt was essential for RNP formation. The results are consistent with a model in which HDAg binds HDV unbranched rod RNA as multimers of fixed size rather than as individual subunits.

Del-1, an endogenous leukocyte-endothelial adhesion inhibitor, limits inflammatory cell recruitment.

Leukocyte recruitment to sites of infection or inflammation requires multiple adhesive events. Although numerous players promoting leukocyte-endothelial interactions have been characterized, functionally important endogenous inhibitors of leukocyte adhesion have not been identified. Here we describe the endothelially derived secreted molecule Del-1 (developmental endothelial locus-1) as an anti-adhesive factor that interferes with the integrin LFA-1-dependent leukocyte-endothelial adhesion. Endothelial Del-1 deficiency increased LFA-1-dependent leukocyte adhesion in vitro and in vivo. Del-1-/- mice displayed significantly higher neutrophil accumulation in lipopolysaccharide-induced lung inflammation in vivo, which was reversed in Del-1/LFA-1 double-deficient mice. Thus, Del-1 is an endogenous inhibitor of inflammatory cell recruitment and could provide a basis for targeting leukocyte-endothelial interactions in disease.