Cross-scale multi-instance learning for pathological image diagnosis

Analyzing high resolution whole slide images (WSIs) with regard to information across multiple scales poses a significant challenge in digital pathology. Multi-instance learning (MIL) is a common solution for working with high resolution images by classifying bags of objects (i.e. sets of smaller image patches). However, such processing is typically performed at a single scale (e.g., 20× magnification) of WSIs, disregarding the vital inter-scale information that is key to diagnoses by human pathologists. In this study, we propose a novel cross-scale MIL algorithm to explicitly aggregate inter-scale relationships into a single MIL network for pathological image diagnosis. The contribution of this paper is three-fold: (1) A novel cross-scale MIL (CS-MIL) algorithm that integrates the multi-scale information and the inter-scale relationships is proposed; (2) A toy dataset with scale-specific morphological features is created and released to examine and visualize differential cross-scale attention; (3) Superior performance on both in-house and public datasets is demonstrated by our simple cross-scale MIL strategy. The official implementation is publicly available at https://github.com/hrlblab/CS-MIL.

Inhibition of autocrine HGF maturation overcomes cetuximab resistance in colorectal cancer

Although amplifications and mutations in receptor tyrosine kinases (RTKs) act as bona fide oncogenes, in most cancers, RTKs maintain moderate expression and remain wild-type. Consequently, cognate ligands control many facets of tumorigenesis, including resistance to anti-RTK therapies. Herein, we show that the ligands for the RTKs MET and RON, HGF and HGFL, respectively, are synthesized as inactive precursors that are activated by cellular proteases. Our newly generated HGF/HGFL protease inhibitors could overcome both de novo and acquired cetuximab resistance in colorectal cancer (CRC). Conversely, HGF overexpression was necessary and sufficient to induce cetuximab resistance and loss of polarity. Moreover, HGF-induced cetuximab resistance could be overcome by the downstream MET inhibitor, crizotinib, and upstream protease inhibitors. Additionally, HAI-1, an endogenous inhibitor of HGF proteases, (i) was downregulated in CRC, (ii) exhibited increased genomic methylation that correlated with poor prognosis, (iii) HAI-1 expression correlated with cetuximab response in a panel of cancer cell lines, and (iv) exogenous addition of recombinant HAI-1 overcame cetuximab resistance in CC-HGF cells. Thus, we describe a targetable, autocrine HAI-1/Protease/HGF/MET axis in cetuximab resistance in CRC.

Coronavirus M Protein Trafficking in Epithelial Cells Utilizes a Myosin Vb Splice Variant and Rab10

The membrane (M) glycoprotein of coronaviruses (CoVs) serves as the nidus for virion assembly. Using a yeast two-hybrid screen, we identified the interaction of the cytosolic tail of Murine Hepatitis Virus (MHV-CoV) M protein with Myosin Vb (MYO5B), specifically with the alternative splice variant of cellular MYO5B including exon D (MYO5B+D), which mediates interaction with Rab10. When co-expressed in human lung epithelial A549 and canine kidney epithelial MDCK cells, MYO5B+D co-localized with the MHV-CoV M protein, as well as with the M proteins from Porcine Epidemic Diarrhea Virus (PEDV-CoV), Middle East Respiratory Syndrome (MERS-CoV) and Severe Acute Respiratory Syndrome 2 (SARS-CoV-2). Co-expressed M proteins and MYO5B+D co-localized with endogenous Rab10 and Rab11a. We identified point mutations in MHV-CoV M that blocked the interaction with MYO5B+D in yeast 2-hybrid assays. One of these point mutations (E121K) was previously shown to block MHV-CoV virion assembly and its interaction with MYO5B+D. The E to K mutation at homologous positions in PEDV-CoV, MERS-CoV and SARS-CoV-2 M proteins also blocked colocalization with MYO5B+D. The knockdown of Rab10 blocked the co-localization of M proteins with MYO5B+D and was rescued by re-expression of CFP-Rab10. Our results suggest that CoV M proteins traffic through Rab10-containing systems, in association with MYO5B+D.

Molecular cartography uncovers evolutionary and microenvironmental dynamics in sporadic colorectal tumors

Colorectal cancer exhibits dynamic cellular and genetic heterogeneity during progression from precursor lesions toward malignancy. Analysis of spatial multi-omic data from 31 human colorectal specimens enabled phylogeographic mapping of tumor evolution that revealed individualized progression trajectories and accompanying microenvironmental and clonal alterations. Phylogeographic mapping ordered genetic events, classified tumors by their evolutionary dynamics, and placed clonal regions along global pseudotemporal progression trajectories encompassing the chromosomal instability (CIN+) and hypermutated (HM) pathways. Integrated single-cell and spatial transcriptomic data revealed recurring epithelial programs and infiltrating immune states along progression pseudotime. We discovered an immune exclusion signature (IEX), consisting of extracellular matrix regulators DDR1, TGFBI, PAK4, and DPEP1, that charts with CIN+ tumor progression, is associated with reduced cytotoxic cell infiltration, and shows prognostic value in independent cohorts. This spatial multi-omic atlas provides insights into colorectal tumor-microenvironment co-evolution, serving as a resource for stratification and targeted treatments.

Patient-derived enteroids provide a platform for the development of therapeutic approaches in microvillus inclusion disease

Microvillus inclusion disease (MVID), caused by loss-of-function mutations in the motor protein myosin Vb (MYO5B), is a severe infantile disease characterized by diarrhea, malabsorption, and acid/base instability, requiring intensive parenteral support for nutritional and fluid management. Human patient-derived enteroids represent a model for investigation of monogenic epithelial disorders but are a rare resource from MVID patients. We developed human enteroids with different loss-of function MYO5B variants and showed that they recapitulated the structural changes found in native MVID enterocytes. Multiplex immunofluorescence imaging of patient duodenal tissues revealed patient-specific changes in localization of brush border transporters. Functional analysis of electrolyte transport revealed profound loss of Na+/H+ exchange (NHE) activity in MVID patient enteroids with near-normal chloride secretion. The chloride channel-blocking antidiarrheal drug crofelemer dose-dependently inhibited agonist-mediated fluid secretion. MVID enteroids exhibited altered differentiation and maturation versus healthy enteroids. γ-Secretase inhibition with DAPT recovered apical brush border structure and functional Na+/H+ exchange activity in MVID enteroids. Transcriptomic analysis revealed potential pathways involved in the rescue of MVID cells including serum/glucocorticoid-regulated kinase 2 (SGK2) and NHE regulatory factor 3 (NHERF3). These results demonstrate the utility of patient-derived enteroids for developing therapeutic approaches to MVID.

A Specialized Epithelial Cell Type Regulating Mucosal Immunity and Driving Human Crohn's Disease

Crohn's disease (CD) is a complex chronic inflammatory disorder that may affect any part of gastrointestinal tract with extra-intestinal manifestations and associated immune dysregulation. To characterize heterogeneity in CD, we profiled single-cell transcriptomics of 170 samples from 65 CD patients and 18 non-inflammatory bowel disease (IBD) controls in both the terminal ileum (TI) and ascending colon (AC). Analysis of 202,359 cells identified a novel epithelial cell type in both TI and AC, featuring high expression of LCN2, NOS2, and DUOX2, and thus is named LND. LND cells, confirmed by high-resolution in-situ RNA imaging, were rarely found in non-IBD controls, but expanded significantly in active CD. Compared to other epithelial cells, genes defining LND cells were enriched in antimicrobial response and immunoregulation. Moreover, multiplexed protein imaging demonstrated that LND cell abundance was associated with immune infiltration. Cross-talk between LND and immune cells was explored by ligand-receptor interactions and further evidenced by their spatial colocalization. LND cells showed significant enrichment of expression specificity of IBD/CD susceptibility genes, revealing its role in immunopathogenesis of CD. Investigating lineage relationships of epithelial cells detected two LND cell subpopulations with different origins and developmental potential, early and late LND. The ratio of the late to early LND cells was related to anti-TNF response. These findings emphasize the pathogenic role of the specialized LND cell type in both Crohn's ileitis and Crohn's colitis and identify novel biomarkers associated with disease activity and treatment response.

Feasibility of Universal Anomaly Detection without Knowing the Abnormality in Medical Images

Many anomaly detection approaches, especially deep learning methods, have been recently developed to identify abnormal image morphology by only employing normal images during training. Unfortunately, many prior anomaly detection methods were optimized for a specific "known" abnormality (e.g., brain tumor, bone fraction, cell types). Moreover, even though only the normal images were used in the training process, the abnormal images were often employed during the validation process (e.g., epoch selection, hyper-parameter tuning), which might leak the supposed "unknown" abnormality unintentionally. In this study, we investigated these two essential aspects regarding universal anomaly detection in medical images by (1) comparing various anomaly detection methods across four medical datasets, (2) investigating the inevitable but often neglected issues on how to unbiasedly select the optimal anomaly detection model during the validation phase using only normal images, and (3) proposing a simple decision-level ensemble method to leverage the advantage of different kinds of anomaly detection without knowing the abnormality. The results of our experiments indicate that none of the evaluated methods consistently achieved the best performance across all datasets. Our proposed method enhanced the robustness of performance in general (average AUC 0.956).

GammaGateR: semi-automated marker gating for single-cell multiplexed imaging

Motivation

Multiplexed immunofluorescence (mIF) is an emerging assay for multichannel protein imaging that can decipher cell-level spatial features in tissues. However, existing automated cell phenotyping methods, such as clustering, face challenges in achieving consistency across experiments and often require subjective evaluation. As a result, mIF analyses often revert to marker gating based on manual thresholding of raw imaging data.

Results

To address the need for an evaluable semi-automated algorithm, we developed GammaGateR, an R package for interactive marker gating designed specifically for segmented cell-level data from mIF images. Based on a novel closed-form gamma mixture model, GammaGateR provides estimates of marker-positive cell proportions and soft clustering of marker-positive cells. The model incorporates user-specified constraints that provide a consistent but slide-specific model fit. We compared GammaGateR against the newest unsupervised approach for annotating mIF data, employing two colon datasets and one ovarian cancer dataset for the evaluation. We showed that GammaGateR produces highly similar results to a silver standard established through manual annotation. Furthermore, we demonstrated its effectiveness in identifying biological signals, achieved by mapping known spatial interactions between CD68 and MUC5AC cells in the colon and by accurately predicting survival in ovarian cancer patients using the phenotype probabilities as input for machine learning methods. GammaGateR is a highly efficient tool that can improve the replicability of marker gating results, while reducing the time of manual segmentation.

Availability and Implementation

The R package is available at https://github.com/JiangmeiRubyXiong/GammaGateR.

Apposition of fibroblasts with metaplastic gastric cells promotes dysplastic transition

BACKGROUND & AIMS

Elements of field cancerization including atrophic gastritis, metaplasia and dysplasia promote gastric cancer development in association with chronic inflammation. However, it remains unclear how stroma changes during carcinogenesis and how the stroma contributes to progression of gastric preneoplasia. Here we investigated heterogeneity of fibroblasts, one of the most important elements in the stroma, and their roles in neoplastic transformation of metaplasia.

METHODS

We utilized single cell transcriptomics to evaluate the cellular heterogeneity of mucosal cells from human gastric cancer patients. Tissue sections from the same cohort and tissue microarrays were used to identify the geographical distribution of distinct fibroblast subsets. We further evaluated the role of fibroblasts from pathologic mucosa in dysplastic progression of metaplastic cells using patient-derived metaplastic gastroids and fibroblasts.

RESULTS

We identified four subsets of fibroblasts within stromal cells defined by the differential expression of PDGFRA, FBLN2, ACTA2 or PDGFRB. Each subset was distributed distinctively throughout stomach tissues with different proportions at each pathologic stage. The PDGFRα⁺ subset expanded in metaplasia and cancer compared with normal, maintaining a close proximity with the epithelial compartment. Co-culture of metaplasia- or cancer-derived fibroblasts with gastroids showing the characteristics of spasmolytic polypeptide-expressing metaplasia (SPEM) induced disordered growth, loss of metaplastic markers and increases in markers of dysplasia. Culture of metaplastic gastroids with conditioned media from metaplasia- or cancer-derived fibroblasts also promoted dysplastic transition.

CONCLUSIONS

These findings indicate that fibroblast associations with metaplastic epithelial cells can facilitate direct transition of metaplastic SPEM cell lineages into dysplastic lineages.

Intestinal Epithelial Digestive, Transport, and Barrier Protein Expression Is Increased in Environmental Enteric Dysfunction

Environmental enteric dysfunction (EED) is characterized by malabsorption and diarrhea that result in irreversible deficits in physical and intellectual growth. We sought to define the expression of transport and tight junction proteins by quantitative analysis of duodenal biopsies from patients with EED. Biopsies from Pakistani children with confirmed EED diagnoses were compared to those from age-matched North American healthy controls, patients with celiac disease, and patients with nonceliac disease with villous atrophy or intraepithelial lymphocytosis. Expression of brush border digestive and transport proteins and paracellular (tight junction) proteins was assessed by quantitative multiplex immunofluorescence microscopy. EED was characterized by partial villous atrophy and marked intraepithelial lymphocytosis. Epithelial proliferation and enteroendocrine, tuft, and Paneth cell numbers were unchanged, but there was significant goblet cell expansion in EED biopsies. Expression of proteins involved in nutrient and water absorption and that of the basolateral Cl- transport protein NKCC1 were also increased in EED. Finally, the barrier-forming tight junction protein claudin-4 (CLDN4) was significantly upregulated in EED, particularly within villous enterocytes. In contrast, expression of CFTR, CLDN2, CLDN15, JAM-A, occludin, ZO-1, and E-cadherin was unchanged. Upregulation of a barrier-forming tight junction protein and brush border and basolateral membrane proteins that support nutrient and water transport in EED is paradoxical, as their increased expression would be expected to be correlated with increased intestinal barrier function and enhanced absorption, respectively. These data suggest that EED activates adaptive intestinal epithelial responses to enhance nutrient absorption but that these changes are insufficient to restore health.

Monitoring Cochlear Health With Intracochlear Electrocochleography During Cochlear Implantation: Findings From an International Clinical Investigation

OBJECTIVES

Electrocochleography (ECochG) is emerging as a tool for monitoring cochlear function during cochlear implant (CI) surgery. ECochG may be recorded directly from electrodes on the implant array intraoperatively. For low-frequency stimulation, its amplitude tends to rise or may plateau as the electrode is inserted. The aim of this study was to explore whether compromise of the ECochG signal, defined as a fall in its amplitude of 30% or more during insertion, whether transient or permanent, is associated with poorer postoperative acoustic hearing, and to examine how preoperative hearing levels may influence the ability to record ECochG. The specific hypotheses tested were threefold: (a) deterioration in the pure-tone average of low-frequency hearing at the first postoperative follow-up interval (follow-up visit 1 [FUV1], 4 to 6 weeks) will be associated with compromise of the cochlear microphonic (CM) amplitude during electrode insertion (primary hypothesis); (b) an association is observed at the second postoperative follow-up interval (FUV2, 3 months) (secondary hypothesis 1); and (c) the CM response will be recorded earlier during electrode array insertion when the preoperative high-frequency hearing is better (secondary hypothesis 2).

DESIGN

International, multi-site prospective, observational, between groups design, targeting 41 adult participants in each of two groups, (compromised CM versus preserved CM). Adult CI candidates who were scheduled to receive a Cochlear Nucleus CI with a Slim Straight or a Slim Modiolar electrode array and had a preoperative audiometric low-frequency average thresholds of ≤80 dB HL at 500, 750, and 1000 Hz in the ear to be implanted, were recruited from eight international implant sites. Pure tone audiometry was measured preoperatively and at postoperative visits (FUV1 and follow-up visit 2 [FUV2]). ECochG was measured during and immediately after the implantation of the array.

RESULTS

From a total of 78 enrolled individuals (80 ears), 77 participants (79 ears) underwent surgery. Due to protocol deviations, 18 ears (23%) were excluded. Of the 61 ears with ECochG responses, amplitudes were < 1 µV throughout implantation for 18 ears (23%) and deemed "unclear" for classification. EcochG responses >1 µV in 43 ears (55%) were stable throughout implantation for 8 ears and compromised in 35 ears. For the primary endpoint at FUV1, 7/41 ears (17%) with preserved CM had a median hearing loss of 12.6 dB versus 34/41 ears (83%) with compromised CM and a median hearing loss of 26.9 dB ( p < 0.014). In assessing the practicalities of measuring intraoperative ECochG, the presence of a measurable CM (>1 µV) during implantation was dependent on preoperative, low-frequency thresholds, particularly at the stimulus frequency (0.5 kHz). High-frequency, preoperative thresholds were also associated with a measurable CM > 1 µV during surgery.

CONCLUSIONS

Our data shows that CM drops occurring during electrode insertion were correlated with significantly poorer hearing preservation postoperatively compared to CMs that remained stable throughout the electrode insertion. The practicality of measuring ECochG in a large cohort is discussed, regarding the suggested optimal preoperative low-frequency hearing levels ( < 80 dB HL) considered necessary to obtain a CM signal >1 µV.

Influence of Cell-type Ratio on Spatially Resolved Single-cell Transcriptomes using the Tangram Algorithm: Based on Implementation on Single-Cell and MxIF Data

The Tangram algorithm is a benchmarking method of aligning single-cell (sc/snRNA-seq) data to various forms of spatial data collected from the same region. With this data alignment, the annotation of the single-cell data can be projected to spatial data. However, the cell composition (cell-type ratio) of the single-cell data and spatial data might be different because of heterogeneous cell distribution. Whether the Tangram algorithm can be adapted when the two data have different cell-type ratios has not been discussed in previous works. In our practical application that maps the cell-type classification results of single-cell data to the Multiplex immunofluorescence (MxIF) spatial data, cell-type ratios were different, though they were sampled from adjacent areas. In this work, both simulation and empirical validation were conducted to quantitatively explore the impact of the mismatched cell-type ratio on the Tangram mapping in different situations. Results show that the cell-type difference has a negative influence on classification accuracy.

Predicting Crohn's disease severity in the colon using mixed cell nucleus density from pseudo labels

Crohn's disease (CD) is a debilitating inflammatory bowel disease with no known cure. Computational analysis of hematoxylin and eosin (H&E) stained colon biopsy whole slide images (WSIs) from CD patients provides the opportunity to discover unknown and complex relationships between tissue cellular features and disease severity. While there have been works using cell nuclei-derived features for predicting slide-level traits, this has not been performed on CD H&E WSIs for classifying normal tissue from CD patients vs active CD and assessing slide label-predictive performance while using both separate and combined information from pseudo-segmentation labels of nuclei from neutrophils, eosinophils, epithelial cells, lymphocytes, plasma cells, and connective cells. We used 413 WSIs of CD patient biopsies and calculated normalized histograms of nucleus density for the six cell classes for each WSI. We used a support vector machine to classify the truncated singular value decomposition representations of the normalized histograms as normal or active CD with four-fold cross-validation in rounds where nucleus types were first compared individually, the best was selected, and further types were added each round. We found that neutrophils were the most predictive individual nucleus type, with an AUC of 0.92 ± 0.0003 on the withheld test set. Adding information improved cross-validation performance for the first two rounds and on the withheld test set for the first three rounds, though performance metrics did not increase substantially beyond when neutrophils were used alone.

Topological-Preserving Membrane Skeleton Segmentation in Multiplex Immunofluorescence Imaging

Multiplex immunofluorescence (MxIF) is an emerging imaging technology whose downstream molecular analytics highly rely upon the effectiveness of cell segmentation. In practice, multiple membrane markers (e.g., NaKATPase, PanCK and β-catenin) are employed to stain membranes for different cell types, so as to achieve a more comprehensive cell segmentation since no single marker fits all cell types. However, prevalent watershed-based image processing might yield inferior capability for modeling complicated relationships between markers. For example, some markers can be misleading due to questionable stain quality. In this paper, we propose a deep learning based membrane segmentation method to aggregate complementary information that is uniquely provided by large scale MxIF markers. We aim to segment tubular membrane structure in MxIF data using global (membrane markers z-stack projection image) and local (separate individual markers) information to maximize topology preservation with deep learning. Specifically, we investigate the feasibility of four SOTA 2D deep networks and four volumetric-based loss functions. We conducted a comprehensive ablation study to assess the sensitivity of the proposed method with various combinations of input channels. Beyond using adjusted rand index (ARI) as the evaluation metric, which was inspired by the clDice, we propose a novel volumetric metric that is specific for skeletal structure, denoted as c l D i c e S K E L . In total, 80 membrane MxIF images were manually traced for 5-fold cross-validation. Our model outperforms the baseline with a 20.2% and 41.3% increase in c l D i c e S K E L and ARI performance, which is significant (p<0.05) using the Wilcoxon signed rank test. Our work explores a promising direction for advancing MxIF imaging cell segmentation with deep learning membrane segmentation. Tools are available at https://github.com/MASILab/MxIF_Membrane_Segmentation.

Therapy Development for Microvillus Inclusion Disease using Patient-derived Enteroids

Microvillus Inclusion Disease (MVID), caused by loss-of-function mutations in the motor protein Myosin Vb (MYO5B), is a severe infantile disease characterized by diarrhea, malabsorption, and acid-base instability, requiring intensive parenteral support for nutritional and fluid management. Human patient-derived enteroids represent a model for investigation of monogenic epithelial disorders but are a rare resource from MVID patients. We developed human enteroids with different loss-of function MYO5B variants and showed that they recapitulated the structural changes found in native MVID enterocytes. Multiplex Immunofluorescence imaging of patient duodenal tissues revealed patient-specific changes in localization of brush border transporters. Functional analysis of electrolyte transport revealed profound loss of Na + /H + exchange (NHE) activity in MVID patient enteroids with near-normal chloride secretion. The chloride channel-blocking anti-diarrheal drug, Crofelemer, dose-dependently inhibited agonist-mediated fluid secretion. MVID enteroids exhibited altered differentiation and maturation versus healthy enteroids. Inhibition of Notch signaling with the γ-secretase inhibitor, DAPT, recovered apical brush border structure and functional Na + /H + exchange activity in MVID enteroids. Transcriptomic analysis revealed potential pathways involved in the rescue of MVID cells including serum- and glucocorticoid-induced protein kinase 2 (SGK2), and NHE regulatory factor 3 (NHERF3). These results demonstrate the utility of patient-derived enteroids for developing therapeutic approaches to MVID.

Conflict-of-interest statement

The authors have declared that no conflict of interest exists.

Rab11FIP1-deficient mice develop spontaneous inflammation and show increased susceptibility to colon damage

The small GTPase, Rab11a, regulates vesicle trafficking and cell polarity in epithelial cells through interaction with Rab11 family-interacting proteins (Rab11-FIPs). We hypothesized that deficiency of Rab11-FIP1 would affect mucosal integrity in the intestine. Global Rab11FIP1 knockout (KO) mice were generated by deletion of the second exon. Pathology of intestinal tissues was analyzed by immunostaining of colonic sections and RNA-sequencing of isolated colonic epithelial cells. A low concentration of dextran sodium sulfate (DSS, 2%) was added to drinking water for 5 days, and injury score was compared between Rab11FIP1 KO, Rab11FIP2 KO, and heterozygous littermates. Rab11FIP1 KO mice showed normal fertility and body weight gain. More frequent lymphoid patches and infiltration of macrophages and neutrophils were identified in Rab11FIP1 KO mice before the development of rectal prolapse compared with control mice. The population of trefoil factor 3 (TFF3)-positive goblet cells was significantly lower, and the ratio of proliferative to nonproliferative cells was higher in Rab11FIP1 KO colons. Transcription signatures indicated that Rab11FIP1 deletion downregulated genes that mediate stress tolerance response, whereas genes mediating the response to infection were significantly upregulated, consistent with the inflammatory responses in the steady state. Lack of Rab11FIP1 also resulted in abnormal accumulation of subapical vesicles in colonocytes and the internalization of transmembrane mucin, MUC13, with Rab14. After DSS treatment, Rab11FIP1 KO mice showed greater body weight loss and more severe mucosal damage than those in heterozygous littermates. These findings suggest that Rab11FIP1 is important for cytoprotection mechanisms and for the maintenance of colonic mucosal integrity.NEW & NOTEWORTHY Although Rab11FIP1 is important in membrane trafficking in epithelial cells, the gastrointestinal phenotype of Rab11FIP1 knockout (KO) mice had never been reported. This study demonstrated that Rab11FIP1 loss induces mistrafficking of Rab14 and MUC13 and decreases in colonic goblet cells, resulting in impaired mucosal integrity.

Cross-scale Attention Guided Multi-instance Learning for Crohn's Disease Diagnosis with Pathological Images

Multi-instance learning (MIL) is widely used in the computer-aided interpretation of pathological Whole Slide Images (WSIs) to solve the lack of pixel-wise or patch-wise annotations. Often, this approach directly applies "natural image driven" MIL algorithms which overlook the multi-scale (i.e. pyramidal) nature of WSIs. Off-the-shelf MIL algorithms are typically deployed on a single-scale of WSIs (e.g., 20× magnification), while human pathologists usually aggregate the global and local patterns in a multi-scale manner (e.g., by zooming in and out between different magnifications). In this study, we propose a novel cross-scale attention mechanism to explicitly aggregate inter-scale interactions into a single MIL network for Crohn's Disease (CD), which is a form of inflammatory bowel disease. The contribution of this paper is two-fold: (1) a cross-scale attention mechanism is proposed to aggregate features from different resolutions with multi-scale interaction; and (2) differential multi-scale attention visualizations are generated to localize explainable lesion patterns. By training ~250,000 H&E-stained Ascending Colon (AC) patches from 20 CD patient and 30 healthy control samples at different scales, our approach achieved a superior Area under the Curve (AUC) score of 0.8924 compared with baseline models. The official implementation is publicly available at https://github.com/hrlblab/CS-MIL.

Cancer-Associated Fibroblasts and Squamous Epithelial Cells Constitute a Unique Microenvironment in a Mouse Model of Inflammation-Induced Colon Cancer

The tumor microenvironment plays a key role in the pathogenesis of colorectal tumors and contains various cell types including epithelial, immune, and mesenchymal cells. Characterization of the interactions between these cell types is necessary for revealing the complex nature of tumors. In this study, we used single-cell RNA-seq (scRNA-seq) to compare the tumor microenvironments between a mouse model of sporadic colorectal adenoma (Lrig1^CreERT2/+;Apc^2lox14/+) and a mouse model of inflammation-driven colorectal cancer induced by azoxymethane and dextran sodium sulfate (AOM/DSS). While both models develop tumors in the distal colon, we found that the two tumor types have distinct microenvironments. AOM/DSS tumors have an increased abundance of two populations of cancer-associated fibroblasts (CAFs) compared with APC tumors, and we revealed their divergent spatial association with tumor cells using multiplex immunofluorescence (MxIF) imaging. We also identified a unique squamous cell population in AOM/DSS tumors, whose origins were distinct from anal squamous epithelial cells. These cells were in higher proportions upon administration of a chemotherapy regimen of 5-Fluorouracil/Irinotecan. We used computational inference algorithms to predict cell-cell communication mediated by ligand-receptor interactions and downstream pathway activation, and identified potential mechanistic connections between CAFs and tumor cells, as well as CAFs and squamous epithelial cells. This study provides important preclinical insight into the microenvironment of two distinct models of colorectal tumors and reveals unique roles for CAFs and squamous epithelial cells in the AOM/DSS model of inflammation-driven cancer.

A Myosin Vb Splice Variant Regulates Coronavirus M Protein Trafficking in Polarized Epithelial Cells

The membrane (M) glycoprotein of coronaviruses (CoVs) serves as the nidus for virion assembly. Using a yeast two‐hybrid screen, we identified the interaction of the cytosolic tail of Murine Hepatitis Virus (MHV‐CoV) M protein with Myosin Vb (MYO5B) (Figure 1A). MHV‐CoV M interacts specifically with the alternative splice variant of cellular MYO5B including Exon D (MYO5B+D), that also mediates interaction with cellular Rab10. When co‐expressed in human lung epithelial A549 and canine kidney epithelial MDCK cells, MYO5B+D co‐localized with MHV‐CoV M protein, as well as with M proteins from porcine epidemic diarrhea Virus (PEDV‐CoV), Middle East Respiratory Syndrome (MERS‐CoV) and Severe Acute Respiratory Syndrome 2 (SARS‐CoV‐2) (Figure 2). M‐GFP chimeric proteins co‐expressed with mCherry‐MYO5B+D also co‐localized with endogenous Rab10 and Rab11a (Figure 1B,C). We identified point mutations in MHV‐CoV M that blocked the interaction with MYO5B+D in yeast 2‐hybrid assays (Figure 1B). One of these point mutations (E121K) was previously shown to block MHV‐CoV virion assembly, and it blocked interaction with MYO5B+D. The E to K mutation at homologous positions in PEDV‐CoV, MERS‐CoV and SARS‐CoV‐2 M proteins also blocked colocalization with MYO5B+D. Knockdown of Rab10 blocked the co‐localization of the M proteins with MYO5B+D (Figure 2). Re‐expression of Cerulein‐Rab10 in Rab10 KD cells re‐established colocalization between M proteins and MYO5B+D (Figure 2). Our results suggest that the interaction of CoV M proteins with MYO5B+D may play a role in regulating their trafficking through Rab10‐containing membrane systems in epithelial cells.

Quantifying and correcting slide-to-slide variation in multiplexed immunofluorescence images

MOTIVATION

Multiplexed imaging is a nascent single-cell assay with a complex data structure susceptible to technical variability that disrupts inference. These in situ methods are valuable in understanding cell-cell interactions, but few standardized processing steps or normalization techniques of multiplexed imaging data are available.

RESULTS

We implement and compare data transformations and normalization algorithms in multiplexed imaging data. Our methods adapt the ComBat and functional data registration methods to remove slide effects in this domain, and we present an evaluation framework to compare the proposed approaches. We present clear slide-to-slide variation in the raw, unadjusted data and show that many of the proposed normalization methods reduce this variation while preserving and improving the biological signal. Furthermore, we find that dividing multiplexed imaging data by its slide mean, and the functional data registration methods, perform the best under our proposed evaluation framework. In summary, this approach provides a foundation for better data quality and evaluation criteria in multiplexed imaging.

AVAILABILITY AND IMPLEMENTATION

Source code is provided at: https://github.com/statimagcoll/MultiplexedNormalization and an R package to implement these methods is available here: https://github.com/ColemanRHarris/mxnorm.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Circle Representation for Medical Object Detection

Box representation has been extensively used for object detection in computer vision. Such representation is efficacious but not necessarily optimized for biomedical objects (e.g., glomeruli), which play an essential role in renal pathology. In this paper, we propose a simple circle representation for medical object detection and introduce CircleNet, an anchor-free detection framework. Compared with the conventional bounding box representation, the proposed bounding circle representation innovates in three-fold: (1) it is optimized for ball-shaped biomedical objects; (2) The circle representation reduced the degree of freedom compared with box representation; (3) It is naturally more rotation invariant. When detecting glomeruli and nuclei on pathological images, the proposed circle representation achieved superior detection performance and be more rotation-invariant, compared with the bounding box. The code has been made publicly available: https://github.com/hrlblab/CircleNet.

MITI minimum information guidelines for highly multiplexed tissue images

The imminent release of tissue atlases combining multi-channel microscopy with single cell sequencing and other omics data from normal and diseased specimens creates an urgent need for data and metadata standards that guide data deposition, curation and release. We describe a Minimum Information about highly multiplexed Tissue Imaging (MITI) standard that applies best practices developed for genomics and other microscopy data to highly multiplexed tissue images and traditional histology.

Multi-modal Learning with Missing Data for Cancer Diagnosis Using Histopathological and Genomic Data

Multi-modal learning (e.g., integrating pathological images with genomic features) tends to improve the accuracy of cancer diagnosis and prognosis as compared to learning with a single modality. However, missing data is a common problem in clinical practice, i.e., not every patient has all modalities available. Most of the previous works directly discarded samples with missing modalities, which might lose information in these data and increase the likelihood of overfitting. In this work, we generalize the multi-modal learning in cancer diagnosis with the capacity of dealing with missing data using histological images and genomic data. Our integrated model can utilize all available data from patients with both complete and partial modalities. The experiments on the public TCGA-GBM and TCGA-LGG datasets show that the data with missing modalities can contribute to multi-modal learning, which improves the model performance in grade classification of glioma cancer.

Inpainting Missing Tissue in Multiplexed Immunofluorescence Imaging

Multiplex immunofluorescence (MxIF) is an emerging technique that allows for staining multiple cellular and histological markers to stain simultaneously on a single tissue section. However, with multiple rounds of staining and bleaching, it is inevitable that the scarce tissue may be physically depleted. Thus, a digital way of synthesizing such missing tissue would be appealing since it would increase the useable areas for the downstream single-cell analysis. In this work, we investigate the feasibility of employing generative adversarial network (GAN) approaches to synthesize missing tissues using 11 MxIF structural molecular markers (i.e., epithelial and stromal). Briefly, we integrate a multi-channel high-resolution image synthesis approach to synthesize the missing tissue from the remaining markers. The performance of different methods is quantitatively evaluated via the downstream cell membrane segmentation task. Our contribution is that we, for the first time, assess the feasibility of synthesizing missing tissues in MxIF via quantitative segmentation. The proposed synthesis method has comparable reproducibility with the baseline method on performance for the missing tissue region reconstruction only, but it improves 40% on whole tissue synthesis that is crucial for practical application. We conclude that GANs are a promising direction of advancing MxIF imaging with deep image synthesis.

MIRIAM: A machine and deep learning single-cell segmentation and quantification pipeline for multi-dimensional tissue images

Increasingly, highly multiplexed tissue imaging methods are used to profile protein expression at the single‐cell level. However, a critical limitation is the lack of robust cell segmentation tools for tissue sections. We present Multiplexed Image Resegmentation of Internal Aberrant Membranes (MIRIAM) that combines (a) a pipeline for cell segmentation and quantification that incorporates machine learning‐based pixel classification to define cellular compartments, (b) a novel method for extending incomplete cell membranes, and (c) a deep learning‐based cell shape descriptor. Using human colonic adenomas as an example, we show that MIRIAM is superior to widely utilized segmentation methods and provides a pipeline that is broadly applicable to different imaging platforms and tissue types.

A cross-platform informatics system for the Gut Cell Atlas: integrating clinical, anatomical and histological data

The Gut Cell Atlas (GCA), an initiative funded by the Helmsley Charitable Trust, seeks to create a reference platform to understand the human gut, with a specific focus on Crohn's disease. Although a primary focus of the GCA is on focusing on single-cell profiling, we seek to provide a framework to integrate other analyses on multi-modality data such as electronic health record data, radiological images, and histology tissues/images. Herein, we use the research electronic data capture (REDCap) system as the central tool for a secure web application that supports protected health information (PHI) restricted access. Our innovations focus on addressing the challenges with tracking all specimens and biopsies, validating manual data entry at scale, and sharing organizational data across the group. We present a scalable, cross-platform barcode printing/record system that integrates with REDCap. The central informatics infrastructure to support our design is a tuple table to track longitudinal data entry and sample tracking. The current data collection (by December 2020) is illustrated with types and formats of the data that the system collects. We estimate that one terabyte is needed for data storage per patient study. Our proposed data sharing informatics system addresses the challenges with integrating physical sample tracking, large files, and manual data entry with REDCap.

Random Multi-Channel Image Synthesis for Multiplexed Immunofluorescence Imaging

Multiplex immunofluorescence (MxIF) is an emerging imaging technique that produces the high sensitivity and specificity of single-cell mapping. With a tenet of "seeing is believing", MxIF enables iterative staining and imaging extensive antibodies, which provides comprehensive biomarkers to segment and group different cells on a single tissue section. However, considerable depletion of the scarce tissue is inevitable from extensive rounds of staining and bleaching ("missing tissue"). Moreover, the immunofluorescence (IF) imaging can globally fail for particular rounds ("missing stain"). In this work, we focus on the "missing stain" issue. It would be appealing to develop digital image synthesis approaches to restore missing stain images without losing more tissue physically. Herein, we aim to develop image synthesis approaches for eleven MxIF structural molecular markers (i.e., epithelial and stromal) on real samples. We propose a novel multi-channel high-resolution image synthesis approach, called pixN2N-HD, to tackle possible missing stain scenarios via a high-resolution generative adversarial network (GAN). Our contribution is three-fold: (1) a single deep network framework is proposed to tackle missing stain in MxIF; (2) the proposed "N-to-N" strategy reduces theoretical four years of computational time to 20 hours when covering all possible missing stains scenarios, with up to five missing stains (e.g., "(N-1)-to-1", "(N-2)-to-2"); and (3) this work is the first comprehensive experimental study of investigating cross-stain synthesis in MxIF. Our results elucidate a promising direction of advancing MxIF imaging with deep image synthesis.

Cell differentiation is disrupted by MYO5B loss through Wnt/Notch imbalance

Functional loss of myosin Vb (MYO5B) induces a variety of deficits in intestinal epithelial cell function and causes a congenital diarrheal disorder, microvillus inclusion disease (MVID). The impact of MYO5B loss on differentiated cell lineage choice has not been investigated. We quantified the populations of differentiated epithelial cells in tamoxifen-induced, epithelial cell–specific MYO5B-knockout (VilCre^ERT2 Myo5b^fl/fl) mice utilizing digital image analysis. Consistent with our RNA-sequencing data, MYO5B loss induced a reduction in tuft cells in vivo and in organoid cultures. Paneth cells were significantly increased by MYO5B deficiency along with expansion of the progenitor cell zone. We further investigated the effect of lysophosphatidic acid (LPA) signaling on epithelial cell differentiation. Intraperitoneal LPA significantly increased tuft cell populations in both control and MYO5B-knockout mice. Transcripts for Wnt ligands were significantly downregulated by MYO5B loss in intestinal epithelial cells, whereas Notch signaling molecules were unchanged. Additionally, treatment with the Notch inhibitor dibenzazepine (DBZ) restored the populations of secretory cells, suggesting that the Notch pathway is maintained in MYO5B-deficient intestine. MYO5B loss likely impairs progenitor cell differentiation in the small intestine in vivo and in vitro, partially mediated by Wnt/Notch imbalance. Notch inhibition and/or LPA treatment may represent an effective therapeutic approach for treatment of MVID.

Abstract 1578: Heating it up: Targeting RAS/RAF/PI3K pathway to make melanoma tumors ‘immunologically hot' and suitable for checkpoint blockade immunotherapies

Immune checkpoint blockade (ICB) is the current first-line treatment for metastatic melanoma. However, ICB fails in many patients and has dangerous side effects. Rigosertib (RGS), a non-ATP-competitive small molecule RAS mimetic, has the potential to block oncogenic RAS-RAF-MEK-ERK and/or PI3K-AKT-mTOR signaling pathways. Using immunocompetent mouse models of B16F10 (BRAFwt) and YUMM3.3 (BRAFmut) melanoma, we identified that RGS treatment (300mg/kg) of melanoma tumor-bearing mice is well tolerated and results in ~50% inhibition of tumor growth as monotherapy and ~70% inhibition in synergy with αPD1+αCTLA4. RGS-induced tumor inhibition depends on the induction of CD40 expression on melanoma cells, followed by immunogenic cell death, leading to an inflamed tumor microenvironment with enrichment of dendritic cells and activated CD8+ Tc cells. The RGS-initiated suppression of tumor growth was partially reversed by either suppression of CD40 expression in the melanoma cells by shRNA, or depletion of CD8+ Tc cells. Analysis of multiple published patient melanoma studies confirms that a high CD40 expression level correlates with CD80, ICOS-L, beneficial type-I T-cell responses, and better survival in melanoma patients. The CD40 levels in melanoma cells are prognostic for therapeutic response to RGS, RAF inhibitor and ICB. Notably, multiplex IHC analysis showed that BRAF inhibitor treatment significantly induces CD40+SOX10+ melanoma cells in the tumors of melanoma patients and patient-derived xenografts. Our preclinical data support the therapeutic use of RGS plus αPD1+αCTLA4 in RAS/RAF/MEK and/or PI3K pathway-activated melanoma tumors and point to the need for clinical trials to determine the clinical benefit of RGS plus ICB for metastatic melanoma patients who do not respond to ICB alone.

The authors sincerely thank Onconova Therapeutics, Newtown, PA 18940 for kindly supplying Rigosertib for this work.

Citation Format: Chi Yan, Nabil Saleh, Caroline Nebhan, Anna E. Vilgelm, E. Premkumar Reddy, Joseph T. Roland, Douglas B. Johnson, Rebecca L. Shattuck-Brandt, Sheau-Chiann Chen, Gregory D. Ayers, Ann Richmond. Heating it up: Targeting RAS/RAF/PI3K pathway to make melanoma tumors ‘immunologically hot' and suitable for checkpoint blockade immunotherapies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1578.

Targeting MYCN-expressing triple-negative breast cancer with BET and MEK inhibitors

Triple-negative breast cancer (TNBC) is an aggressive form of breast cancer that does not respond to endocrine therapy or human epidermal growth factor receptor 2 (HER2)-targeted therapies. Individuals with TNBC experience higher rates of relapse and shorter overall survival compared to patients with receptor-positive breast cancer subtypes. Preclinical discoveries are needed to identify, develop, and advance new drug targets to improve outcomes for patients with TNBC. Here, we report that MYCN, an oncogene typically overexpressed in tumors of the nervous system or with neuroendocrine features, is heterogeneously expressed within a substantial fraction of primary and recurrent TNBC and is expressed in an even higher fraction of TNBCs that do not display a pathological complete response after neoadjuvant chemotherapy. We performed high-throughput chemical screens on TNBC cell lines with varying amounts of MYCN expression and determined that cells with higher expression of MYCN were more sensitive to bromodomain and extraterminal motif (BET) inhibitors. Combined BET and MEK inhibition resulted in a synergistic decrease in viability, both in vitro and in vivo, using cell lines and patient-derived xenograft (PDX) models. Our preclinical data provide a rationale to advance a combination of BET and MEK inhibitors to clinical investigation for patients with advanced MYCN-expressing TNBC.

Novel induction of CD40 expression by tumor cells with RAS/RAF/PI3K pathway inhibition augments response to checkpoint blockade

BACKGROUND

While immune checkpoint blockade (ICB) is the current first-line treatment for metastatic melanoma, it is effective for ~ 52% of patients and has dangerous side effects. The objective here was to identify the feasibility and mechanism of RAS/RAF/PI3K pathway inhibition in melanoma to sensitize tumors to ICB therapy.

METHODS

Rigosertib (RGS) is a non-ATP-competitive small molecule RAS mimetic. RGS monotherapy or in combination therapy with ICB were investigated using immunocompetent mouse models of BRAF^wt and BRAF^mut melanoma and analyzed in reference to patient data.

RESULTS

RGS treatment (300 mg/kg) was well tolerated in mice and resulted in ~ 50% inhibition of tumor growth as monotherapy and ~ 70% inhibition in combination with αPD1 + αCTLA4. RGS-induced tumor growth inhibition depends on CD40 upregulation in melanoma cells followed by immunogenic cell death, leading to enriched dendritic cells and activated T cells in the tumor microenvironment. The RGS-initiated tumor suppression was partially reversed by either knockdown of CD40 expression in melanoma cells or depletion of CD8⁺ cytotoxic T cells. Treatment with either dabrafenib and trametinib or with RGS, increased CD40⁺SOX10⁺ melanoma cells in the tumors of melanoma patients and patient-derived xenografts. High CD40 expression level correlates with beneficial T-cell responses and better survival in a TCGA dataset from melanoma patients. Expression of CD40 by melanoma cells is associated with therapeutic response to RAF/MEK inhibition and ICB.

CONCLUSIONS

Our data support the therapeutic use of RGS + αPD1 + αCTLA4 in RAS/RAF/PI3K pathway-activated melanomas and point to the need for clinical trials of RGS + ICB for melanoma patients who do not respond to ICB alone.

TRIAL REGISTRATION

NCT01205815 (Sept 17, 2010).

Key histopathologic features in idiopathic noncirrhotic portal hypertension: an interobserver agreement study and proposal for diagnostic criteria

Histologic features of idiopathic noncirrhotic portal hypertension (INCPH), loosely termed as obliterative portal venopathy (OPV), are heterogenous, often subtle, and overlap with other entities. To this date, no consensus histopathologic diagnostic criteria have been established for INCPH. For these reasons, rendering a reproducible consensus histologic diagnosis of OPV on a liver biopsy may often be challenging even for experienced hepatopathologists. We report herein a two-phase interobserver agreement study on the diagnosis of OPV and assessed the relative value of histologic features in 104 liver biopsies in distinguishing between INCPH and non-INCPH with the goal to obtain a consensus on specific practical diagnostic criteria. Six hepatopathologists blinded to clinical information and original pathologic diagnosis reviewed internet-based case study sets with high-resolution whole-slide images. The initial interobserver agreement on OPV was expectedly low, but significantly improved (moderate agreement in most categories) upon adopting a consensus view recognizing portal vein sclerosis as the only strong independent histologic predictor for INCPH, and that contrary to the conventional view, aberrant portal/periportal vessels does not significantly contribute to the positive assignment of OPV status. We propose a three-tiered classification with diagnostic criteria to facilitate the histologic assignment of OPV status for the evaluation of INCPH. Furthermore, we have validated the performance of the proposed criteria either based on histology alone or coupled with clinicopathologic correlation. This classification may aid in practical histologic assessment of liver biopsies with or without portal hypertension and help to improve diagnostic consistency and accuracy.

The BDNF rs6265 Polymorphism is a Modifier of Cardiomyocyte Contractility and Dilated Cardiomyopathy

Brain-derived neurotrophic factor (BDNF) is a neuronal growth and survival factor that harbors cardioprotective qualities that may attenuate dilated cardiomyopathy. In ~30% of the population, BDNF has a common, nonsynonymous single nucleotide polymorphism rs6265 (Val66Met), which might be correlated with increased risk of cardiovascular events. We previously showed that BDNF correlates with better cardiac function in Duchenne muscular dystrophy (DMD) patients. However, the effect of the Val66Met polymorphism on cardiac function has not been determined. The goal of the current study was to determine the effects of rs6265 on BDNF biomarker suitability and DMD cardiac functions more generally. We assessed cardiovascular and skeletal muscle function in human DMD patients segregated by polymorphic allele. We also compared echocardiographic, electrophysiologic, and cardiomyocyte contractility in C57/BL-6 wild-type mice with rs6265 polymorphism and in mdx/mTR (mDMD) mouse model of DMD. In human DMD patients, plasma BDNF levels had a positive correlation with left ventricular function, opposite to that seen in rs6265 carriers. There was also a substantial decrease in skeletal muscle function in carriers compared to the Val homozygotes. Surprisingly, the opposite was true when cardiac function of DMD carriers and non-carriers were compared. On the other hand, Val66Met wild-type mice had only subtle functional differences at baseline but significantly decreased cardiomyocyte contractility. Our results indicate that the Val66Met polymorphism alters myocyte contractility, conferring worse skeletal muscle function but better cardiac function in DMD patients. Moreover, these results suggest a mechanism for the relative preservation of cardiac tissues compared to skeletal muscle in DMD patients and underscores the complexity of BDNF signaling in response to mechanical workload.

Lysophosphatidic Acid Increases Maturation of Brush Borders and SGLT1 Activity in MYO5B-deficient Mice, a Model of Microvillus Inclusion Disease

BACKGROUND & AIM

Myosin VB (MYO5B) is an essential trafficking protein for membrane recycling in gastrointestinal epithelial cells. The inactivating mutations of MYO5B cause the congenital diarrheal disease, microvillus inclusion disease (MVID). MYO5B deficiency in mice causes mislocalization of SGLT1 and NHE3, but retained apical function of CFTR, resulting in malabsorption and secretory diarrhea. Activation of lysophosphatidic acid (LPA) receptors can improve diarrhea, but the effect of LPA on MVID symptoms is unclear. We investigated whether LPA administration can reduce the epithelial deficits in MYO5B-knockout mice.

METHODS

Studies were conducted with tamoxifen-induced, intestine-specific knockout of MYO5B (VilCre^ERT2;Myo5b^flox/flox) and littermate controls. Mice were given LPA, an LPAR2 agonist (GRI977143), or vehicle for 4 days after a single injection of tamoxifen. Apical SGLT1 and CFTR activities were measured in Üssing chambers. Intestinal tissues were collected, and localization of membrane transporters was evaluated by immunofluorescence analysis in tissue sections and enteroids. RNA sequencing and enrichment analysis were performed with isolated jejunal epithelial cells.

RESULTS

Daily administration of LPA reduced villus blunting, frequency of multivesicular bodies, and levels of cathepsins in intestinal tissues of MYO5B-knockout mice compared with vehicle administration. LPA partially restored the brush border height and the localization of SGLT1 and NHE3 in small intestine of MYO5B-knockout mice and enteroids. The SGLT1-dependent short-circuit current was increased and abnormal CFTR activities were decreased in jejunum from MYO5B-knockout mice given LPA compared with vehicle.

CONCLUSIONS

LPA may regulate a MYO5B-independent trafficking mechanism and brush border maturation, and therefore be developed for treatment of MVID.

CircleNet: Anchor-free Glomerulus Detection with Circle Representation

Object detection networks are powerful in computer vision, but not necessarily optimized for biomedical object detection. In this work, we propose CircleNet, a simple anchor-free detection method with circle representation for detection of the ball-shaped glomerulus. Different from the traditional bounding box based detection method, the bounding circle (1) reduces the degrees of freedom of detection representation, (2) is naturally rotation invariant, (3) and optimized for ball-shaped objects. The key innovation to enable this representation is the anchor-free framework with the circle detection head. We evaluate CircleNet in the context of detection of glomerulus. CircleNet increases average precision of the glomerulus detection from 0.598 to 0.647. Another key advantage is that CircleNet achieves better rotation consistency compared with bounding box representations.

Trop2 is upregulated in the transition to dysplasia in the metaplastic gastric mucosa

Intestinal-type gastric adenocarcinoma arises in a field of pre-existing metaplasia. While biomarkers of cancer and metaplasia have been identified, the definition of dysplastic transition as a critical point in the evolution of cancer has remained obscure. We have evaluated Trop2 as a putative marker of the transition from metaplasia to dysplasia in the stomach in multiple mouse models of metaplasia induction and progression. In addition, TROP2 expression was evaluated in human samples by immunostaining tissue microarrays for metaplasia, dysplasia, and gastric cancer. Dysplastic mouse organoids were evaluated in vitro following shRNA knockdown of Trop2 expression. In mouse models, no Trop2 was observed in the normal corpus and Trop2 was not induced in acute models of metaplasia induction with either L635 or DMP-777. In Mist1-Kras mice, Trop2 expression was not observed in metaplasia at 1 month after Kras induction, but was observed in dysplastic glands at 3-4 months after Kras induction. In human tissues, no Trop2 was observed in normal corpus mucosa or SPEM, but Trop2 expression was observed in incomplete intestinal metaplasia, with significantly less expression in complete intestinal metaplasia. Trop2 expression was observed in all dysplastic and 84% of gastric cancer lesions, although expression levels were variable. Dysplastic mouse organoids from Mist1-Kras mice expressed Trop2 strongly. Knockdown of Trop2 with shRNA markedly reduced organoid growth and budding behavior, and induced the upregulation of apical villin expression. We conclude that Trop2 is upregulated in the transition to dysplasia in the stomach and promotes dysplastic cell behaviors. © 2020 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Bcl2-Expressing Quiescent Type B Neural Stem Cells in the Ventricular-Subventricular Zone Are Resistant to Concurrent Temozolomide/X-Irradiation

The ventricular-subventricular zone (V-SVZ) of the mammalian brain is a site of adult neurogenesis. Within the V-SVZ reside type B neural stem cells (NSCs) and type A neuroblasts. The V-SVZ is also a primary site for very aggressive glioblastoma (GBM). Standard-of-care therapy for GBM consists of safe maximum resection, concurrent temozolomide (TMZ), and X-irradiation (XRT), followed by adjuvant TMZ therapy. The question of how this therapy impacts neurogenesis is not well understood and is of fundamental importance as normal tissue tolerance is a limiting factor. Here, we studied the effects of concurrent TMZ/XRT followed by adjuvant TMZ on type B stem cells and type A neuroblasts of the V-SVZ in C57BL/6 mice. We found that chemoradiation induced an apoptotic response in type A neuroblasts, as marked by cleavage of caspase 3, but not in NSCs, and that A cells within the V-SVZ were repopulated given sufficient recovery time. 53BP1 foci formation and resolution was used to assess the repair of DNA double-strand breaks. Remarkably, the repair was the same in type B and type A cells. While Bax expression was the same for type A or B cells, antiapoptotic Bcl2 and Mcl1 expression was significantly greater in NSCs. Thus, the resistance of type B NSCs to TMZ/XRT appears to be due, in part, to high basal expression of antiapoptotic proteins compared with type A cells. This preclinical research, demonstrating that murine NSCs residing in the V-SVZ are tolerant of standard chemoradiation therapy, supports a dose escalation strategy for treatment of GBM. Stem Cells 2019;37:1629-1639.

Loss of Rab25 promotes the development of skin squamous cell carcinoma through the dysregulation of integrin trafficking

Rab25 can function as both a tumor suppressor and a tumor promoter across different tissues. This study sought to clarify the role of Rab25 as a tumor suppressor in skin squamous cell carcinoma (SCC). Rab25 loss was closely associated with neoplastic transition in both humans and mice. Rab25 loss was well correlated with increased cell proliferation and poor differentiation in human SCC. While Rab25 knockout (KO) in mice did not induce spontaneous tumor formation, it did significantly accelerate tumor generation and promote malignant transformation in a mouse two-stage skin carcinogenesis model. Xenografting of a Rab25-deficient human keratinocyte cell line, HaCaT, also elicited neoplastic transformation. Notably, Rab25 deficiency led to dysregulation of integrins β1, β4, and α6, which matched well with increased epidermal proliferation and impaired desmosome-tight junction formation. Rab25 deficiency induced impairment of integrin recycling, leading to the improper expression of integrins. In line with this, significant attenuation of integrin β1, β4, and α6 expression was identified in human SCCs where Rab25 was deficient. Collectively, these results suggest that loss of Rab25 promotes the development and neoplastic transition of SCC through dysregulation of integrin trafficking. © 2019 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Abstract 3934: Bcl2-expressing quiescent type B neural stem cells in the SVZ are resistant to concurrent temozolomide/X-irradiation

The subventricular zone (SVZ) is the largest source of neural stem cells (NSCs) in the adult brain. Emerging research indicates that NSCs within the SVZ may be cells of origin for WHO grade IV astrocytoma (glioblastoma, GBM)1. GBM consists of multiple fractions of proliferative and/or quiescent stem-like cells that are thought to be lineally related. GBM located adjacent to the SVZ are very resistant to standard of care concurrent temozolomide (TMZ)/X-irradiation (XRT) therapy, a consequence, it is hypothesized, of their NSC origin2. An important, unanswered question is the origin of this resistance. While a significant effort has been undertaken to study proliferating cells, the origins of quiescent cell resistance are not well understood. Normal NSCs adjacent to the SVZ are mainly quiescent. We rationalized that a fundamental understanding of the response of quiescent NSCs to TMZ/XRT would be informative and aid in our understanding of GBM resistance. For 5 consecutive days cohorts of C57BL/6 mice were administered TMZ (0 or 50 mg.kg i.p.). One hr later 0 or 2 Gy was administered to the brain. Transcardial perfusion was performed on day 6 for half the mice. The remaining mice received adjuvant TMZ (100 mg/kg) or vehicle on days 19-22 and transcardial perfusion was performed on day 82. 10 µm coronal brain sections were obtained and immunostained for well characterized markers of type B NSCs (GFAP and Sox2) and type A neuroblasts (Dcx). Immunofluorescence was imaged using a Leica Aperio Versa 200 slide scanning microscope. Cell Profiler software was used to quantify type B and A cells in the SVZ in all cohorts. Proliferating type A cells were exquisitely sensitive to 5 days of concurrent TMZ/XRT treatment whereas quiescent NSCs located within 30 µm of a dorsal or dorsolateral ventricle were very resistant. NSCs in mice exposed to concurrent and adjuvant therapy were also resistant and importantly, able to repopulate type A cells to sham/control levels. 53BP1 foci formation, a surrogate for DNA DSBs, was quantified in Sox2- and Dcx-expressing cells using confocal microscopy following a single TMZ/XRT exposure. Foci formation, measured 6 min to 24 hrs after TMZ/XRT, was not statistically different between cell types (P&gt;0.05). Because TMZ/XRT induced an apoptotic response in A but not in B cells, as marked by cleaved Caspase-3 staining, we investigated expression of Bax and Bcl2 on a per cell basis. Bax expression was not significantly different for type A or B cells (P&gt;0.05). In contrast, type B NSCs expressed 5-fold more Bcl2 than type A neuroblasts (P&lt; 0.001). In conclusion, we demonstrate that type A neuroblasts are sensitive to TMZ/XRT but can be repopulated by inherently resistant type B NSCs given sufficient time. The resistance of quiescent NSCs to TMZ/XRT is associated with high basal expression of anti-apoptotic proteins. 1Lee et al Nature 2018, 560:243-47; 2Smith et al J Neurooncol 2016, 128:207-16

Citation Format: Brent D. Cameron, Geri Traver, Joseph T. Roland, Daniel Dean, Levi Johnson, Kelli L. Boyd, Rebecca A. Ihrie, Jialiang Wang, Michael L. Freeman. Bcl2-expressing quiescent type B neural stem cells in the SVZ are resistant to concurrent temozolomide/X-irradiation [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 3934.

Advances in Evaluation of Chronic Diarrhea in Infants

Diarrhea is common in infants (children less than 2 years of age), usually acute, and, if chronic, commonly caused by allergies and occasionally by infectious agents. Congenital diarrheas and enteropathies (CODEs) are rare causes of devastating chronic diarrhea in infants. Evaluation of CODEs is a lengthy process and infrequently leads to a clear diagnosis. However, genomic analyses and the development of model systems have increased our understanding of CODE pathogenesis. With these advances, a new diagnostic approach is needed. We propose a revised approach to determine causes of diarrhea in infants, including CODEs, based on stool analysis, histologic features, responses to dietary modifications, and genetic tests. After exclusion of common causes of diarrhea in infants, the evaluation proceeds through analyses of stool characteristics (watery, fatty, or bloody) and histologic features, such as the villus to crypt ratio in intestinal biopsies. Infants with CODEs resulting from defects in digestion, absorption, transport of nutrients and electrolytes, or enteroendocrine cell development or function have normal villi to crypt ratios; defects in enterocyte structure or immune-mediated conditions result in an abnormal villus to crypt ratios and morphology. Whole-exome and genome sequencing in the early stages of evaluation can reduce the time required for a definitive diagnosis of CODEs, or lead to identification of new variants associated with these enteropathies. The functional effects of gene mutations can be analyzed in model systems such as enteroids or induced pluripotent stem cells and are facilitated by recent advances in gene editing procedures. Characterization and investigation of new CODE disorders will improve management of patients and advance our understanding of epithelial cells and other cells in the intestinal mucosa.

Head and Neck MRI Findings in CHARGE Syndrome

Coloboma of the eye, Heart defects, Atresia of the choanae, Retardation of growth and/or development, Genital and/or urinary abnormalities, and Ear abnormalities and deafness (CHARGE) syndrome is a disorder with multiple congenital anomalies seen on imaging. A retrospective review of 10 patients with CHARGE syndrome who underwent MR imaging of the brain as part of a preoperative evaluation for cochlear implantation was conducted. Structural abnormalities of the entire MR imaging of the head were evaluated, including the auditory system, olfactory system, face, skull base, and central nervous system. The most frequent MR imaging findings included dysplasias of the semicircular canals and hypoplasia of the frontal lobe olfactory sulci. Less frequent findings included cleft lip/palate and coloboma. Our study uncovered new findings of a J-shaped sella, dorsal angulation of the clivus, and absent/atrophic parotid glands, not previously described in patients with CHARGE. Our results emphasize the utility of MR imaging in the diagnosis and management of patients with CHARGE syndrome.

Unsupervised Trajectory Analysis of Single-Cell RNA-Seq and Imaging Data Reveals Alternative Tuft Cell Origins in the Gut

Modern single-cell technologies allow multiplexed sampling of cellular states within a tissue. However, computational tools that can infer developmental cell-state transitions reproducibly from such single-cell data are lacking. Here, we introduce p-Creode, an unsupervised algorithm that produces multi-branching graphs from single-cell data, compares graphs with differing topologies, and infers a statistically robust hierarchy of cell-state transitions that define developmental trajectories. We have applied p-Creode to mass cytometry, multiplex immunofluorescence, and single-cell RNA-seq data. As a test case, we validate cell-state-transition trajectories predicted by p-Creode for intestinal tuft cells, a rare, chemosensory cell type. We clarify that tuft cells are specified outside of the Atoh1-dependent secretory lineage in the small intestine. However, p-Creode also predicts, and we confirm, that tuft cells arise from an alternative, Atoh1-driven developmental program in the colon. These studies introduce p-Creode as a reliable method for analyzing large datasets that depict branching transition trajectories.

The C-terminal region of A-kinase anchor protein 350 (AKAP350A) enables formation of microtubule-nucleation centers and interacts with pericentriolar proteins

Microtubules in animal cells assemble (nucleate) from both the centrosome and the cis-Golgi cisternae. A-kinase anchor protein 350 kDa (AKAP350A, also called AKAP450/CG-NAP/AKAP9) is a large scaffolding protein located at both the centrosome and Golgi apparatus. Previous findings have suggested that AKAP350 is important for microtubule dynamics at both locations, but how this scaffolding protein assembles microtubule nucleation machinery is unclear. Here, we found that overexpression of the C-terminal third of AKAP350A, enhanced GFP-AKAP350A(2691-3907), induces the formation of multiple microtubule-nucleation centers (MTNCs). Nevertheless, these induced MTNCs lacked "true" centriole proteins, such as Cep135. Mapping analysis with AKAP350A truncations demonstrated that AKAP350A contains discrete regions responsible for promoting or inhibiting the formation of multiple MTNCs. Moreover, GFP-AKAP350A(2691-3907) recruited several pericentriolar proteins to MTNCs, including γ-tubulin, pericentrin, Cep68, Cep170, and Cdk5RAP2. Proteomic analysis indicated that Cdk5RAP2 and Cep170 both interact with the microtubule nucleation-promoting region of AKAP350A, whereas Cep68 interacts with the distal C-terminal AKAP350A region. Yeast two-hybrid assays established a direct interaction of Cep170 with AKAP350A. Super-resolution and deconvolution microscopy analyses were performed to define the association of AKAP350A with centrosomes, and these studies disclosed that AKAP350A spans the bridge between centrioles, co-localizing with rootletin and Cep68 in the linker region. siRNA-mediated depletion of AKAP350A caused displacement of both Cep68 and Cep170 from the centrosome. These results suggest that AKAP350A acts as a scaffold for factors involved in microtubule nucleation at the centrosome and coordinates the assembly of protein complexes associating with the intercentriolar bridge.

Optimized multiplex immunofluorescence single-cell analysis reveals tuft cell heterogeneity

Intestinal tuft cells are a rare, poorly understood cell type recently shown to be a critical mediator of type 2 immune response to helminth infection. Here, we present advances in segmentation algorithms and analytical tools for multiplex immunofluorescence (MxIF), a platform that enables iterative staining of over 60 antibodies on a single tissue section. These refinements have enabled a comprehensive analysis of tuft cell number, distribution, and protein expression profiles as a function of anatomical location and physiological perturbations. Based solely on DCLK1 immunoreactivity, tuft cell numbers were similar throughout the mouse small intestine and colon. However, multiple subsets of tuft cells were uncovered when protein coexpression signatures were examined, including two new intestinal tuft cell markers, Hopx and EGFR phosphotyrosine 1068. Furthermore, we identified dynamic changes in tuft cell number, composition, and protein expression associated with fasting and refeeding and after introduction of microbiota to germ-free mice. These studies provide a foundational framework for future studies of intestinal tuft cell regulation and demonstrate the utility of our improved MxIF computational methods and workflow for understanding cellular heterogeneity in complex tissues in normal and disease states.

Abnormal Rab11-Rab8-vesicles cluster in enterocytes of patients with microvillus inclusion disease

Microvillus inclusion disease (MVID) is a congenital enteropathy characterized by accumulation of vesiculo-tubular endomembranes in the subapical cytoplasm of enterocytes, historically termed "secretory granules." However, neither their identity nor pathophysiological significance is well defined. Using immunoelectron microscopy and tomography, we studied biopsies from MVID patients (3× Myosin 5b mutations and 1× Syntaxin3 mutation) and compared them to controls and genome-edited CaCo2 cell models, harboring relevant mutations. Duodenal biopsies from 2 patients with novel Myosin 5b mutations and typical clinical symptoms showed unusual ultrastructural phenotypes: aberrant subapical vesicles and tubules were prominent in the enterocytes, though other histological hallmarks of MVID were almost absent (ectopic intra-/intercellular microvilli, brush border atrophy). We identified these enigmatic vesiculo-tubular organelles as Rab11-Rab8-positive recycling compartments of altered size, shape and location harboring the apical SNARE Syntaxin3, apical transporters sodium-hydrogen exchanger 3 (NHE3) and cystic fibrosis transmembrane conductance regulator. Our data strongly indicate that in MVID disrupted trafficking between cargo vesicles and the apical plasma membrane is the primary cause of a defect of epithelial polarity and subsequent facultative loss of brush border integrity, leading to malabsorption. Furthermore, they support the notion that mislocalization of transporters, such as NHE3 substantially contributes to the reported sodium loss diarrhea.

MAGI-2 Is a Sensitive and Specific Marker of Prostatic Adenocarcinoma: A Comparison With AMACR

OBJECTIVES

We compared the utility of membrane-associated guanylate kinase, WW and PDZ domain-containing protein 2 (MAGI-2) and α-methylacyl CoA (AMACR) by immunohistochemistry in diagnosing prostatic adenocarcinoma.

METHODS

Seventy-eight radical prostatectomies were used to construct three tissue microarrays with 512 cores, including benign prostatic tissue, benign prostatic hyperplasia, high-grade prostatic intraepithelial neoplasia (HGPIN), and adenocarcinoma. AMACR and MAGI-2 immunohistochemistry were evaluated by visual and image analysis.

RESULTS

MAGI-2 and AMACR were significantly higher in adenocarcinoma and HGPIN compared with benign tissue. At H-score cutoffs of 300 and 200, MAGI-2 was more accurate in distinguishing benign from malignant glands than AMACR. Areas under the curve by image and visual analysis were 0.846 and 0.818 for MAGI-2 and 0.937 and 0.924 for AMACR, respectively. The accuracy of MAGI-2 in distinguishing benign from malignant glands on the same core was higher (95% vs 88%).

CONCLUSIONS

MAGI-2 could represent a useful adjunct for diagnosis of prostatic adenocarcinoma, especially when AMACR is not discriminatory.

Distinct metaplastic and inflammatory phenotypes in autoimmune and adenocarcinoma-associated chronic atrophic gastritis

BACKGROUND

Autoimmune gastritis (AIG) and adenocarcinoma-associated chronic atrophic gastritis (CAG) are both associated with oxyntic atrophy, but AIG patients demonstrate an increased risk of carcinoid tumors rather than the elevated risk of adenocarcinoma observed with CAG. We therefore sought to compare the characteristics of the metaplastic mucosa in AIG and CAG patients.

METHODS

We examined markers for metaplasia (spasmolytic polypeptide expressing metaplasia (SPEM) and intestinal metaplasia) as well as proliferation (Ki67) and immune cell populations (neutrophils, macrophages, and eosinophils) in gastric sections from 16 female patients with autoimmune thyroiditis and AIG and 17 patients with CAG associated with gastric adenocarcinoma.

RESULTS

Both AIG and CAG patients demonstrated prominent SPEM and intestinal metaplasia. However, AIG patients displayed significantly lower numbers of infiltrating macrophages and significantly reduced mucosal cell proliferation as compared to CAG patients.

CONCLUSIONS

These findings indicate that, while both AIG and CAG patients display prominent oxyntic atrophy and metaplasia, the AIG patients do not show proliferative metaplastic lineages that would predispose to adenocarcinoma.