PathoGraph: An Attention-Based Graph Neural Network Capable of Prognostication Based on CD276 Labelling of Malignant Glioma Cells.

Computerized methods have been developed that allow quantitative morphological analyses of whole slide images (WSIs), e.g., of immunohistochemical stains. The latter are attractive because they can provide high-resolution data on the distribution of proteins in tissue. However, many immunohistochemical results are complex because the protein of interest occurs in multiple locations (in different cells and also extracellularly). We have recently established an artificial intelligence framework, PathoFusion which utilises a bifocal convolutional neural network (BCNN) model for detecting and counting arbitrarily definable morphological structures. We have now complemented this model by adding an attention-based graph neural network (abGCN) for the advanced analysis and automated interpretation of such data. Classical convolutional neural network (CNN) models suffer from limitations when handling global information. In contrast, our abGCN is capable of creating a graph representation of cellular detail from entire WSIs. This abGCN method combines attention learning with visualisation techniques that pinpoint the location of informative cells and highlight cell-cell interactions. We have analysed cellular labelling for CD276, a protein of great interest in cancer immunology and a potential marker of malignant glioma cells/putative glioma stem cells (GSCs). We are especially interested in the relationship between CD276 expression and prognosis. The graphs permit predicting individual patient survival on the basis of GSC community features. Our experiments lay a foundation for the use of the BCNN-abGCN tool chain in automated diagnostic prognostication using immunohistochemically labelled histological slides, but the method is essentially generic and potentially a widely usable tool in medical research and AI based healthcare applications.

An Open-Source AI Framework for the Analysis of Single Cells in Whole-Slide Images with a Note on CD276 in Glioblastoma.

Routine examination of entire histological slides at cellular resolution poses a significant if not insurmountable challenge to human observers. However, high-resolution data such as the cellular distribution of proteins in tissues, e.g., those obtained following immunochemical staining, are highly desirable. Our present study extends the applicability of the PathoFusion framework to the cellular level. We illustrate our approach using the detection of CD276 immunoreactive cells in glioblastoma as an example. Following automatic identification by means of PathoFusion's bifocal convolutional neural network (BCNN) model, individual cells are automatically profiled and counted. Only discriminable cells selected through data filtering and thresholding were segmented for cell-level analysis. Subsequently, we converted the detection signals into the corresponding heatmaps visualizing the distribution of the detected cells in entire whole-slide images of adjacent H&E-stained sections using the Discrete Wavelet Transform (DWT). Our results demonstrate that PathoFusion is capable of autonomously detecting and counting individual immunochemically labelled cells with a high prediction performance of 0.992 AUC and 97.7% accuracy. The data can be used for whole-slide cross-modality analyses, e.g., relationships between immunochemical signals and anaplastic histological features. PathoFusion has the potential to be applied to additional problems that seek to correlate heterogeneous data streams and to serve as a clinically applicable, weakly supervised system for histological image analyses in (neuro)pathology.

Customizing Maxpar Direct Immune Profiling Assay with Additional Surface Marker and Intracellular Cytokine Staining Workflows for Expanded Mass Cytometry Panels.

Mass cytometry, or cytometry by time-of-flight (the basis for Fluidigm® CyTOF® technology), is a system for single-cell detection using antibodies tagged with metal probes. Without the need for compensation, the highly parametric Helios™ mass cytometer has a detection range of 135 distinct mass channels (75-209 Da). Optimized for mass cytometry, the Maxpar® Direct™ Immune Profiling Assay™ is a dry, metal-tagged antibody cocktail for immunophenotyping 37 immune cell populations found in human peripheral blood in a single tube. The Maxpar Direct Assay utilizes 31 mass channels for marker detection and live/dead viability staining, with at least 14 additional marker channels available from the Fluidigm catalog for flexible custom panel design. Here, we describe a workflow combining the assay with additional surface and intracellular cytokine antibodies for peripheral blood mononuclear cell (PBMC) staining using lanthanide-, bismuth-, and cadmium-tagged antibodies.

PathoFusion: An Open-Source AI Framework for Recognition of Pathomorphological Features and Mapping of Immunohistochemical Data.

We have developed a platform, termed PathoFusion, which is an integrated system for marking, training, and recognition of pathological features in whole-slide tissue sections. The platform uses a bifocal convolutional neural network (BCNN) which is designed to simultaneously capture both index and contextual feature information from shorter and longer image tiles, respectively. This is analogous to how a microscopist in pathology works, identifying a cancerous morphological feature in the tissue context using first a narrow and then a wider focus, hence bifocal. Adjacent tissue sections obtained from glioblastoma cases were processed for hematoxylin and eosin (H&E) and immunohistochemical (CD276) staining. Image tiles cropped from the digitized images based on markings made by a consultant neuropathologist were used to train the BCNN. PathoFusion demonstrated its ability to recognize malignant neuropathological features autonomously and map immunohistochemical data simultaneously. Our experiments show that PathoFusion achieved areas under the curve (AUCs) of 0.985 ± 0.011 and 0.988 ± 0.001 in patch-level recognition of six typical pathomorphological features and detection of associated immunoreactivity, respectively. On this basis, the system further correlated CD276 immunoreactivity to abnormal tumor vasculature. Corresponding feature distributions and overlaps were visualized by heatmaps, permitting high-resolution qualitative as well as quantitative morphological analyses for entire histological slides. Recognition of more user-defined pathomorphological features can be added to the system and included in future tissue analyses. Integration of PathoFusion with the day-to-day service workflow of a (neuro)pathology department is a goal. The software code for PathoFusion is made publicly available.

Multiple tolerance defects contribute to the breach of B cell tolerance in New Zealand Black chromosome 1 congenic mice.

Lupus is characterized by a loss of B cell tolerance leading to autoantibody production. In this study, we explored the mechanisms underlying this loss of tolerance using B6 congenic mice with an interval from New Zealand Black chromosome 1 (denoted c1(96-100)) sufficient for anti-nuclear antibody production. Transgenes for soluble hen egg white lysozyme (sHEL) and anti-HEL immunoglobulin were crossed onto this background and various tolerance mechanisms examined. We found that c1(96-100) mice produced increased levels of IgM and IgG anti-HEL antibodies compared to B6 mice and had higher proportions of germinal center B cells and long-lived plasma cells, suggesting a germinal center-dependent breach of B cell anergy. Consistent with impaired anergy induction, c1(96-100) double transgenic B cells showed enhanced survival and CD86 upregulation. Hematopoietic chimeric sHEL mice with a mixture of B6 and c1(96-100) HEL transgenic B cells recapitulated these results, suggesting the presence of a B cell autonomous defect. Surprisingly, however, there was equivalent recruitment of B6 and c1(96-100) B cells into germinal centers and differentiation to splenic plasmablasts in these mice. In contrast, there were increased proportions of c1(96-100) T follicular helper cells and long-lived plasma cells as compared to their B6 counterparts, suggesting that both B and T cell defects are required to breach germinal center tolerance in this model. This possibility was further supported by experiments showing an enhanced breach of anergy in double transgenic mice with a longer chromosome 1 interval with additional T cell defects.

IL-10 Production Is Critical for Sustaining the Expansion of CD5+ B and NKT Cells and Restraining Autoantibody Production in Congenic Lupus-Prone Mice.

The development and progression of systemic lupus erythematosus is mediated by the complex interaction of genetic and environmental factors. To decipher the genetics that contribute to pathogenesis and the production of pathogenic autoantibodies, our lab has focused on the generation of congenic lupus-prone mice derived from the New Zealand Black (NZB) strain. Previous work has shown that an NZB-derived chromosome 4 interval spanning 32 to 151 Mb led to expansion of CD5+ B and Natural Killer T (NKT) cells, and could suppress autoimmunity when crossed with a lupus-prone mouse strain. Subsequently, it was shown that CD5+ B cells but not NKT cells derived from these mice could suppress the development of pro-inflammatory T cells. In this paper, we aimed to further resolve the genetics that leads to expansion of these two innate-like populations through the creation of additional sub-congenic mice and to characterize the role of IL-10 in the suppression of autoimmunity through the generation of IL-10 knockout mice. We show that expansion of CD5+ B cells and NKT cells localizes to a chromosome 4 interval spanning 91 to 123 Mb, which is distinct from the region that mediates the majority of the suppressive phenotype. We also demonstrate that IL-10 is critical to restraining autoantibody production and surprisingly plays a vital role in supporting the expansion of innate-like populations.

mir-181a-1/b-1 Modulates Tolerance through Opposing Activities in Selection and Peripheral T Cell Function.

Understanding the consequences of tuning TCR signaling on selection, peripheral T cell function, and tolerance in the context of native TCR repertoires may provide insight into the physiological control of tolerance. In this study, we show that genetic ablation of a natural tuner of TCR signaling, mir-181a-1/b-1, in double-positive thymocytes dampened TCR and Erk signaling and increased the threshold of positive selection. Whereas mir-181a-1/b-1 deletion in mice resulted in an increase in the intrinsic reactivity of naive T cells to self-antigens, it did not cause spontaneous autoimmunity. Loss of mir-181a-1/b-1 dampened the induction of experimental autoimmune encephalomyelitis and reduced basal TCR signaling in peripheral T cells and their migration from lymph nodes to pathogenic sites. Taken together, these results demonstrate that tolerance can be modulated by microRNA gene products through the control of opposing activities in T cell selection and peripheral T cell function.

microRNAs regulate cell-to-cell variability of endogenous target gene expression in developing mouse thymocytes.

The development and homeostasis of multicellular organisms relies on gene regulation within individual constituent cells. Gene regulatory circuits that increase the robustness of gene expression frequently incorporate microRNAs as post-transcriptional regulators. Computational approaches, synthetic gene circuits and observations in model organisms predict that the co-regulation of microRNAs and their target mRNAs can reduce cell-to-cell variability in the expression of target genes. However, whether microRNAs directly regulate variability of endogenous gene expression remains to be tested in mammalian cells. Here we use quantitative flow cytometry to show that microRNAs impact on cell-to-cell variability of protein expression in developing mouse thymocytes. We find two distinct mechanisms that control variation in the activation-induced expression of the microRNA target CD69. First, the expression of miR-17 and miR-20a, two members of the miR-17-92 cluster, is co-regulated with the target mRNA Cd69 to form an activation-induced incoherent feed-forward loop. Another microRNA, miR-181a, acts at least in part upstream of the target mRNA Cd69 to modulate cellular responses to activation. The ability of microRNAs to render gene expression more uniform across mammalian cell populations may be important for normal development and for disease.

Suppression of inflammation with conditional deletion of the prostaglandin E2 EP2 receptor in macrophages and brain microglia.

Prostaglandin E2 (PGE2), a potent lipid signaling molecule, modulates inflammatory responses through activation of downstream G-protein coupled EP(1-4) receptors. Here, we investigated the cell-specific in vivo function of PGE2 signaling through its E-prostanoid 2 (EP2) receptor in murine innate immune responses systemically and in the CNS. In vivo, systemic administration of lipopolysaccharide (LPS) resulted in a broad induction of cytokines and chemokines in plasma that was significantly attenuated in EP2-deficient mice. Ex vivo stimulation of peritoneal macrophages with LPS elicited proinflammatory responses that were dependent on EP2 signaling and that overlapped with in vivo plasma findings, suggesting that myeloid-lineage EP2 signaling is a major effector of innate immune responses. Conditional deletion of the EP2 receptor in myeloid lineage cells in Cd11bCre;EP2(lox/lox) mice attenuated plasma inflammatory responses and transmission of systemic inflammation to the brain was inhibited, with decreased hippocampal inflammatory gene expression and cerebral cortical levels of IL-6. Conditional deletion of EP2 significantly blunted microglial and astrocytic inflammatory responses to the neurotoxin MPTP and reduced striatal dopamine turnover. Suppression of microglial EP2 signaling also increased numbers of dopaminergic (DA) neurons in the substantia nigra independent of MPTP treatment, suggesting that microglial EP2 may influence development or survival of DA neurons. Unbiased microarray analysis of microglia isolated from adult Cd11bCre;EP2(lox/lox) and control mice demonstrated a broad downregulation of inflammatory pathways with ablation of microglial EP2 receptor. Together, these data identify a cell-specific proinflammatory role for macrophage/microglial EP2 signaling in innate immune responses systemically and in brain.

Regulation of immune responses and tolerance: the microRNA perspective.

Much has been learned about the molecular and cellular components critical for the control of immune responses and tolerance. It remains a challenge, however, to control the immune response and tolerance at the system level without causing significant toxicity to normal tissues. Recent studies suggest that microRNA (miRNA) genes, an abundant class of non-coding RNA genes that produce characteristic approximately 22 nucleotides small RNAs, play important roles in immune cells. In this article, we discuss emerging knowledge regarding the functions of miRNA genes in the immune system. We delve into the roles of miRNAs in regulating signaling strength and threshold, homeostasis, and the dynamics of the immune response and tolerance during normal and pathogenic immunological conditions. We also present observations based on analyzes of miR-181 family genes that indicate the potential functions of primary and/or precursor miRNAs in target recognition and explore the impact of these findings on target identification. Finally, we illustrate that despite the subtle effects of miRNAs on gene expression, miRNAs have the potential to influence the outcomes of normal and pathogenic immune responses by controlling the quantitative and dynamic aspects of immune responses. Tuning miRNA functions in immune cells, through gain- and loss-of-function approaches in mice, may reveal novel approach to restore immune equilibrium from pathogenic conditions, such as autoimmune disease and leukemia, without significant toxicity.

TLR tolerance reduces IFN-alpha production despite plasmacytoid dendritic cell expansion and anti-nuclear antibodies in NZB bicongenic mice.

Genetic loci on New Zealand Black (NZB) chromosomes 1 and 13 play a significant role in the development of lupus-like autoimmune disease. We have previously shown that C57BL/6 (B6) congenic mice with homozygous NZB chromosome 1 (B6.NZBc1) or 13 (B6.NZBc13) intervals develop anti-nuclear antibodies and mild glomerulonephritis (GN), together with increased T and B cell activation. Here, we produced B6.NZBc1c13 bicongenic mice with both intervals, and demonstrate several novel phenotypes including: marked plasmacytoid and myeloid dendritic cell expansion, and elevated IgA production. Despite these changes, only minor increases in anti-nuclear antibody production were seen, and the severity of GN was reduced as compared to B6.NZBc1 mice. Although bicongenic mice had increased levels of baff and tnf-α mRNA in their spleens, the levels of IFN-α-induced gene expression were reduced. Splenocytes from bicongenic mice also demonstrated reduced secretion of IFN-α following TLR stimulation in vitro. This reduction was not due to inhibition by TNF-α and IL-10, or regulation by other cellular populations. Because pDC in bicongenic mice are chronically exposed to nuclear antigen-containing immune complexes in vivo, we examined whether repeated stimulation of mouse pDC with TLR ligands leads to impaired IFN-α production, a phenomenon termed TLR tolerance. Bone marrow pDC from both B6 and bicongenic mice demonstrated markedly inhibited secretion of IFN-α following repeated stimulation with a TLR9 ligand. Our findings suggest that the expansion of pDC and production of anti-nuclear antibodies need not be associated with increased IFN-α production and severe kidney disease, revealing additional complexity in the regulation of autoimmunity in systemic lupus erythematosus.

Epistatic suppression of fatal autoimmunity in New Zealand black bicongenic mice.

Numerous mapping studies have implicated genetic intervals from lupus-prone New Zealand Black (NZB) chromosomes 1 and 4 as contributing to lupus pathogenesis. By introgressing NZB chromosomal intervals onto a non-lupus-prone B6 background, we determined that: NZB chromosome 1 congenic mice (denoted B6.NZBc1) developed fatal autoimmune-mediated kidney disease, and NZB chromosome 4 congenic mice (denoted B6.NZBc4) exhibited a marked expansion of B1a and NKT cells in the surprising absence of autoimmunity. In this study, we sought to examine whether epistatic interactions between these two loci would affect lupus autoimmunity by generating bicongenic mice that carry both NZB chromosomal intervals. Compared with B6.NZBc1 mice, bicongenic mice demonstrated significantly decreased mortality, kidney disease, Th1-biased IgG autoantibody isotypes, and differentiation of IFN-γ-producing T cells. Furthermore, a subset of bicongenic mice exhibited a paucity of CD21(+)CD1d(+) B cells and an altered NKT cell activation profile that correlated with greater disease inhibition. Thus, NZBc4 contains suppressive epistatic modifiers that appear to inhibit the development of fatal NZBc1 autoimmunity by promoting a shift away from a proinflammatory cytokine profile, which in some mice may involve NKT cells.

Abrogation of pathogenic IgG autoantibody production in CD40L gene-deleted lupus-prone New Zealand Black mice.

New Zealand Black (NZB) mice spontaneously develop a lupus-like autoimmune disease. Since CD40-CD40L interactions are important for B cell class-switch recombination and germinal center formation, we sought to understand the impact of these interactions on the immune abnormalities in NZB CD40L gene-deleted (CD40L(-/-)) mice in vivo. NZB.CD40L(-/-) mice demonstrated abrogation of all IgG autoantibodies tested and attenuated kidney disease. However, polyclonal B cell activation in vivo and B cell proliferation and class-switching in response to TLR ligands in vitro were preserved in the absence of CD40L in NZB mice. Although, plasmacytoid dendritic cell expansion and elevated BAFF production were unaffected by the absence of CD40L, there was some evidence that IFN-α-induced gene expression was reduced in the bone marrow of NZB.CD40L(-/-) mice. Our results suggest that CD40-CD40L interactions play an important role in promoting pathogenic IgG autoantibody production and kidney disease in NZB mice.

An intrinsic B-cell defect supports autoimmunity in New Zealand black chromosome 13 congenic mice.

Introgression of a New Zealand Black (NZB) chromosome 13 interval onto a C57BL/6 (B6) background (B6.NZBc13) is sufficient to produce many hallmarks of lupus, including high-titre anti-chromatin antibody production, abnormal B- and T-cell activation, and renal disease. In this study we sought to characterize the immune defects leading to these abnormalities. By generating hematopoietic chimeras and BCR transgenic mice, we show that the congenic autoimmune phenotype can be transferred by BM cells and requires the presence of autoreactive B cells. Using the hen egg white lysozyme immunoglobulin transgenic mouse model, we demonstrate that B-cell anergy, deletion, and receptor editing are intact. Nevertheless, congenic B cells exhibit altered peripheral B-cell selection, as demonstrated by enhanced survival and activation of endogenous B cells with autoreactivity to chromatin and Sm/ribonucleoprotein. Given the autoantibody specificities to nuclear antigens, TLR signalling was assessed. B6.NZBc13 B cells were hyper-responsive to poly(I:C), a TLR3 ligand, demonstrating enhanced proliferation and survival as compared to B6 B cells. Our findings indicate the presence of an intrinsic B-cell defect on NZB chromosome 13 that results in hyper-responsiveness to a dsRNA analogue and implicates its potential supporting role in the generation of autoimmunity in B6.NZBc13 mice.

B cell activating factor (BAFF) and T cells cooperate to breach B cell tolerance in lupus-prone New Zealand Black (NZB) mice.

The presence of autoantibodies in New Zealand Black (NZB) mice suggests a B cell tolerance defect however the nature of this defect is unknown. To determine whether defects in B cell anergy contribute to the autoimmune phenotype in NZB mice, soluble hen egg lysozyme (sHEL) and anti-HEL Ig transgenes were bred onto the NZB background to generate double transgenic (dTg) mice. NZB dTg mice had elevated levels of anti-HEL antibodies, despite apparently normal B cell functional anergy in-vitro. NZB dTg B cells also demonstrated increased survival and abnormal entry into the follicular compartment following transfer into sHEL mice. Since this process is dependent on BAFF, BAFF serum and mRNA levels were assessed and were found to be significantly elevated in NZB dTg mice. Treatment of NZB sHEL recipient mice with TACI-Ig reduced NZB dTg B cell survival following adoptive transfer, confirming the role of BAFF in this process. Although NZB mice had modestly elevated BAFF, the enhanced NZB B cell survival response appeared to result from an altered response to BAFF. In contrast, T cell blockade had a minimal effect on B cell survival, but inhibited anti-HEL antibody production. The findings suggest that the modest BAFF elevations in NZB mice are sufficient to perturb B cell tolerance, particularly when acting in concert with B cell functional abnormalities and T cell help.

Insights into the genetic basis and immunopathogenesis of systemic lupus erythematosus from the study of mouse models.

Systemic lupus erythematosus (SLE) is a multisystem autoimmune disease characterized by production of autoantibodies directed against nuclear antigens resulting in formation of immune complexes that deposit in multiple organs causing tissue damage. SLE is a complex genetic disease in which variations in multiple genes, each with a modest effect size, contribute to disease genesis. Given this genetic complexity, identification of the role of individual polymorphisms is challenging. In this context, studies of mouse models of lupus have been particularly informative. Here we review the findings arising from the study of gene deleted, transgenic and congenic lupus-prone mouse models.

Dissociation of the genetic loci leading to b1a and NKT cell expansions from autoantibody production and renal disease in B6 mice with an introgressed New Zealand Black chromosome 4 interval.

Previous mapping studies have linked New Zealand Black (NZB) chromosome 4 to several lupus traits, including autoantibody production, splenomegaly, and glomerulonephritis. To confirm the presence of these traits, our laboratory introgressed homozygous NZB chromosome 4 intervals extending from either 114 to 149 Mb or 32 to 149 Mb onto the lupus-resistant C57BL/6 background (denoted B6.NZBc4S and B6.NZBc4L, respectively). Characterization of aged cohorts revealed that B6.NZBc4L mice exhibited a striking increase in splenic B1a and NKT cells in the absence of high titer autoantibody production and significant renal disease. Tissue-specific expansion of these subsets was also seen in the peritoneum and liver for B1a cells and in the bone marrow for NKT cells. Staining with CD1d tetramers loaded with an alpha-galactosylceramide analog (PBS57) demonstrated that the expanded NKT cell population was mainly CD1d-dependent NKT cells. The lack of both cellular phenotypes in B6.NZBc4S mice demonstrates that the genetic polymorphism(s) that result in these phenotypes are on the proximal region of NZB chromosome 4. This study confirms the presence of a locus that promotes the expansion of B1a cells and newly identifies a region that promotes CD1d-restricted NKT cell expansion on NZB chromosome 4. Taken together, the data indicate that neither an expansion of B1a cells and/nor NKT cells is sufficient to promote autoantibody production and ultimately, renal disease.

Colocalization of expansion of the splenic marginal zone population with abnormal B cell activation and autoantibody production in B6 mice with an introgressed New Zealand Black chromosome 13 interval.

Polyclonal B cell activation is a prominent feature of the lupus-prone New Zealand Black (NZB) mouse strain. We have previously demonstrated linkage between a region on NZB chromosome 13 and increased costimulatory molecule expression on B cells. In this study we have produced C57BL/6 congenic mice with an introgressed homozygous NZB interval extending from approximately 24 to 73 cM on chromosome 13 (denoted B6.NZBc13). We show that B6.NZBc13 female mice not only have enhanced B cell activation but also share many other B cell phenotypic characteristics with NZB mice, including expansion of marginal zone and CD5+ B cell populations, increased numbers of IgM ELISPOTs, and increased serum levels of total IgM and IgM autoantibodies. In addition these mice have increased T cell activation, increased numbers of germinal centers, mild glomerulonephritis, and produce high-titer IgM and IgG anti-chromatin Abs. Male B6.NZBc13 mice have a less pronounced cellular phenotype, lacking expansion of the marginal zone B cell population and IgG anti-chromatin Ab production, indicating the presence of gender dimorphism for this locus. Thus, we have identified a genetic locus that recapitulates with fidelity the B cell phenotypic abnormalities in NZB mice, and we demonstrate that this locus is sufficient to induce an autoimmune phenotype. The data provide further support to the contention that immune abnormalities leading to altered B cell activation and selection contribute to the development of autoimmunity in NZB mice.