Single-cell Image Analysis Research Articles

A cross-reactive imaging matrix of membrane protein profiling for single-cell analysis

Membrane proteins play a critical role in cellular processes such as chemical transport, signal transduction, and enzymatic catalysis. A comprehensive understanding of membrane protein types and accurately measuring their expression levels are essential for advancing cell biology. However, current methods often fail to characterize the types and amounts of multiple membrane proteins in specific cell types due to inherent measurement limitations. Here, we have developed a single-cell image analysis technique based on a cross-reactive imaging matrix using membrane protein tags. This technique allows quantitative and qualitative analysis of multiple membrane protein types and their expression levels within single cells. As a proof of concept, we used four specific DNA aptamers to tag four different membrane proteins in seven different cell types. The four proteins are MUC1, EGFR, HeR2, and TLR4. The seven cell lines are 3t3, 4T1, A549, CT26, HeLa, Hepa1–6, Raw, respectively. We captured a comprehensive image matrix using wide-field fluorescence imaging, allowing precise quantification of protein expression at the single-cell level. Our approach introduces a multi-labeling analysis method for membrane proteins and reveals the influence of membrane protein diversity on cell heterogeneity analysis. We found that the accuracy of cell heterogeneity recognition can be affected by adjusting the number of membrane proteins. As the diversity of membrane protein species increases, the accuracy of cell type recognition increases. Our approach facilitates the evaluation of heterogeneity in diverse cells, providing new insights into cellular diversity and a powerful tool for studying single-cell biology.

Machine learning aided single cell image analysis improves understanding of morphometric heterogeneity of human mesenchymal stem cells

The multipotent stem cells of our body have been largely harnessed in biotherapeutics. However, as they are derived from multiple anatomical sources, from different tissues, human mesenchymal stem cells (hMSCs) are a heterogeneous population showing ambiguity in their in vitro behavior. Intra-clonal population heterogeneity has also been identified and pre-clinical mechanistic studies suggest that these cumulatively depreciate the therapeutic effects of hMSC transplantation. Although various biomarkers identify these specific stem cell populations, recent artificial intelligence-based methods have capitalized on the cellular morphologies of hMSCs, opening a new approach to understand their attributes. A robust and rapid platform is required to accommodate and eliminate the heterogeneity observed in the cell population, to standardize the quality of hMSC therapeutics globally.Here, we report our primary findings of morphological heterogeneity observed within and across two sources of hMSCs namely, stem cells from human exfoliated deciduous teeth (SHEDs) and human Wharton jelly mesenchymal stem cells (hWJ MSCs), using real-time single-cell images generated on immunophenotyping by imaging flow cytometry (IFC). We used the ImageJ software for identification and comparison between the two types of hMSCs using statistically significant morphometric descriptors that are biologically relevant. To expand on these insights, we have further applied deep learning methods and successfully report the development of a Convolutional Neural Network-based image classifier. In our research, we introduced a machine learning methodology to streamline the entire procedure, utilizing convolutional neural networks and transfer learning for binary classification, achieving an accuracy rate of 97.54%. We have also critically discussed the challenges, comparisons between solutions and future directions of machine learning in hMSC classification in biotherapeutics.

Abstract A115: Collective invasion facilitates liver metastasis in pancreatic ductal adenocarcinoma

Abstract Fifty percent of PDAC patients present with distant metastasis and face a dismal prognosis of 8 to 11 months. However, patients with lung metastases have better prognosis than patients with metastatic spread to other sites. In our PDAC metastasis cohort, patients with lung-only metastasis survived longer than patients with liver metastasis, despite sharing the same tumor subtype. To understand features of the primary tumor linked to metastatic site, we generated a novel gene signature for primary organotropism (pORG) that distinguishes primary tumors that developed liver metastasis from those that developed lung metastasis, independent of known tumor subtypes. Our pORG signature predicts survival independently of transcriptional subtype, genetic alterations and clinicopathological variables and in a validation cohort consisting of TCGA pancreatic cancer. Primary tumors with high pORG scores, i.e., liver-tropic primaries, have increased transcriptomic signatures of proliferation and replication stress whereas low pORG tumors have increased lymphocyte gene expression. To validate our findings, we generated cyclic immunofluorescence multiplex imaging data from 34 primary tumors. As anticipated, single cell image analysis revealed high B and T cell infiltration in low-pORG tumors, and increased proliferation and replication stress foci in high-pORG tumors. Intriguingly, malignant cells from high pORG tumors had an increased fraction KRT19-high tumor cells relative to low pORG tumors and were distinctly epithelial in appearance. To reconcile the metastatic behavior of high pORG tumors with their apparent enrichment for epithelial-state cells, we hypothesized that liver-tropism was associated with a partial epithelial-to-mesenchymal transition (EMT) state and metastasis via collective invasion. Bulk transcriptional analysis of 200 PDAC primary tumors revealed a distinct subset of high pORG tumors have increased expression of markers linked to partial EMT and collective invasion, including RAB11a, Rho-like GTPases, and proteases MMP7 and KLK7. Our results support a model of E-cadherin internalization facilitating partial EMT and collective invasion. Current efforts include in-vitro characterization of migration modes in patient-derived cell lines and organoids. Our work suggests unique and targetable tumor intrinsic features linked to organotropism in PDAC. Citation Format: Jennifer R. Eng, Carl Pelz, Koei Chin, Sam Sivagnanam, Link M. Jason, Dove Keith, Lisa Coussens, Rosalie C. Sears. Collective invasion facilitates liver metastasis in pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr A115.

MtFociCounter for automated single-cell mitochondrial nucleoid quantification and reproducible foci analysis.

Mitochondrial DNA (mtDNA) encodes the core subunits for OXPHOS, essential in near-all eukaryotes. Packed into distinct foci (nucleoids) inside mitochondria, the number of mtDNA copies differs between cell-types and is affected in several human diseases. Currently, common protocols estimate per-cell mtDNA-molecule numbers by sequencing or qPCR from bulk samples. However, this does not allow insight into cell-to-cell heterogeneity and can mask phenotypical sub-populations. Here, we present mtFociCounter, a single-cell image analysis tool for reproducible quantification of nucleoids and other foci. mtFociCounter is a light-weight, open-source freeware and overcomes current limitations to reproducible single-cell analysis of mitochondrial foci. We demonstrate its use by analysing 2165 single fibroblasts, and observe a large cell-to-cell heterogeneity in nucleoid numbers. In addition, mtFociCounter quantifies mitochondrial content and our results show good correlation (R=0.90) between nucleoid number and mitochondrial area, and we find nucleoid density is less variable than nucleoid numbers in wild-type cells. Finally, we demonstrate mtFociCounter readily detects differences in foci-numbers upon sample treatment, and applies to Mitochondrial RNA Granules and superresolution microscopy. mtFociCounter provides a versatile solution to reproducibly quantify cellular foci in single cells and our results highlight the importance of accounting for cell-to-cell variance and mitochondrial context in mitochondrial foci analysis.

S3-CIMA: Supervised spatial single-cell image analysis for identifying disease-associated cell-type compositions in tissue

The spatial organization of various cell types within the tissue microenvironment is a key element for the formation of physiological and pathological processes, including cancer and autoimmune diseases. Here, we present S3-CIMA, a weakly supervised convolutional neural network model that enables the detection of disease-specific microenvironment compositions from high-dimensional proteomic imaging data. We demonstrate the utility of this approach by determining cancer outcome- and cellular-signaling-specific spatial cell-state compositions in highly multiplexed fluorescence microscopy data of the tumor microenvironment in colorectal cancer. Moreover, we use S3-CIMA to identify disease-onset-specific changes of the pancreatic tissue microenvironment in type 1 diabetes using imaging mass-cytometry data. We evaluated S3-CIMA as a powerful tool to discover novel disease-associated spatial cellular interactions from currently available and future spatial biology datasets.

Development of an integrin αv-based universal marker, capable of both prediction and direction of stem cell fate.

Integrin-mediated focal adhesion (FA) and subsequent cytoskeletal reorganization influence cell morphology, migration, and ultimately cell fate. Previous studies have used various patterned surfaces with defined macroscopic cell shapes or nanoscopic FA distributions to explore how different substrates affect the fate of human bone marrow mesenchymal stem cells (BMSCs). However, there is currently no straightforward relationship between BMSC cell fates induced by patterned surfaces and FA distribution substrates. In this study, we conducted single-cell image analysis of integrin αv-mediated FA and cell morphological features of BMSCs during biochemically induced differentiation. This enabled the identification of distinct FA features that can discriminate between osteogenic and adipogenic differentiation, demonstrating that integrin αv-mediated focal adhesion (FA) can be used as a non-invasive biomarker for real time observation. Based on these results, we developed an organized microscale fibronectin (FN) patterned surface where the fate of BMSC could be precisely manipulated by these FA features. Notably, even in the absence of any biochemical inducers, such as those contained in the differentiation medium, BMSCs cultured on these FN patterned surfaces exhibited upregulation of differentiation markers comparable to BMSCs cultured using conventional differentiation methods. Therefore, the present study reveals the application of these FA features as universal markers not only for predicting differentiation status, but also for regulating cell fate by precisely controlling the FA features with a new cell culture platform. STATEMENT OF SIGNIFICANCE: Although the effects of material physiochemical properties on cell morphology and subsequent cell fate decisions have been extensively studied, a simple yet intuitive correlation between cellular features and differentiation remains unavailable. We present a single cell image-based strategy for predicting and directing stem cell fate. By using a specific integrin isoform, integrin αv, we identified distinct geometric features that can be used as a universal marker for discriminating between osteogenic and adipogenic differentiation in real-time. From these data, new cell culture platform capable of regulating cell fate by precisely controlling FA features and cell area can be developed.

Abstract 2091: Open-source single-cell spatial image analysis pipeline for ten-plex immunofluorescence of histological prostate cancer tissue sections

Abstract Introduction: Multiplexed fluorescence immunohistochemistry (mfIHC) allows the detection of multiple protein markers on the same histological tissue section, generating valuable information on specific cell subsets and cellular states in their spatial tissue context. Here, we demonstrate a proof-of-concept study for profiling different tumor microenvironment cell subsets in formalin-fixed paraffin-embedded prostate cancer tissue microarrays. Materials and methods: We applied automated mfIHC simultaneous staining of 31 slides with 3.5-µm sections using an Autostainer (Epredia, Kalamazoo, USA). The protocol included four rounds of cyclic staining and scanning of the slides with AxioScanZ.1 (Zeiss, Germany). Each cycle included DAPI for nuclear staining. Here, we applied this pipeline to detect the following targets: D2-40 (lymphatics), CD3 (T lymphocytes), CD56 (NK cells/neural cells/neuroendocrine cells), CD11b+ myeloid cells, PDGFRB (fibroblasts/pericytes), CD20 (B lymphocytes), CD31 (endothelium), CD45 (leukocytes), HLADR+ antigene-presenting cells, as well as a cocktail with PanCK and ECADH to detect epithelial cells. Three different prostate cancer TMA cohorts with n=907 patient samples were analyzed here using this approach.The scanned images were exported as Big TIFF individual channel whole-slide images. Individual TMA spot images were marked with Fiji ROI1 1-Click Tool macro (ImageJ 1.53t), and separate rounds were registered with Python (version 3.11.0, Python Software Foundation) using DAPI channel as reference. Nuclear segmentation was created with nucleAIzer trained modified Mask R-CNN model (https://github.com/lopaavol/nuclei_segmentation). All TMA spot images were run through image curation, where artefacts are excluded by machine learning using ilastik software (version 1.3.3post3). Then, CellProfiler (version 4.2.1) was applied to combine image curated tissue masks and segmented nuclei for cell-based channel intensity measurements. Finally, Jupyter Notebook was applied to classify cell subsets into specific phenotypes using intensity-based thresholding for each channel. Results and Conclusions: We present a detailed pipeline for 10-plex mfIHC with open-source single-cell image analysis tools. The approach can be applied to histological laboratories when combined with a standard five fluorescent channel scanner or imager and with basic computing skills. As the system applies secondary antibodies for the detection, it is flexible for optimizing new antibody panels, and the user is not limited to pre-labelled antibody panels often connected with commercial staining and imaging systems. The approach is especially suited for large-scale high-throughput tissue profiling studies when high n-values are needed for linking associated patient clinical data with spatial phenotype data. Citation Format: Jenni Säilä, Annabrita Schoonenberg, Katja Välimäki, Timo-Pekka Lehto, Carolin Stürenberg, Andrew Erickson, Antti Rannikko, Olli Kallioniemi, Lassi Paavolainen, Tuomas Mirtti, Teijo Pellinen. Open-source single-cell spatial image analysis pipeline for ten-plex immunofluorescence of histological prostate cancer tissue sections [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 2091.

Abstract PR010: Multiplex imaging reveals features of organotropism in pancreatic ductal adenocarcinoma

Abstract PDAC patients have a dismal prognosis with an 11% 5-year survival rate, driven in part by the fact that 90% of patients are diagnosed with metastatic disease. In a cohort of 300 PDAC patients, we observed that patients who had lung-only metastasis had longer overall survival than the majority of patients with metastasis to the liver. We used gene expression data to develop a signature based on gene expression in primary tumors of liver versus lung organotropism, which we called pORG. Our pORG signature predicts long term survival in PDAC patients and reveals unique features of the primary tumor, including increased interferon and immune activity in lung tropic tumors and increased proliferation and replication stress in liver-tropic tumors. To validate our findings, we generated cyclic immunofluorescence multiplex imaging data for a subset of 34 patients. Single cell image analysis revealed B and T cell infiltration and spatial clustering in low pORG/ lung tropic tumors, and markers of replication stress in high pORG patients. Our work suggests unique targetable tumor and microenvironment features linked to organotropism in PDAC. Citation Format: Jennifer Eng, Jason Link, Carl Pelz, Rosalie Sears. Multiplex imaging reveals features of organotropism in pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the AACR Special Conference: Cancer Metastasis; 2022 Nov 14-17; Portland, OR. Philadelphia (PA): AACR; Cancer Res 2022;83(2 Suppl_2):Abstract nr PR010.

Pyruvate Supports RET-Dependent Mitochondrial ROS Production to Control Mycobacterium avium Infection in Human Primary Macrophages.

Macrophages deploy a variety of antimicrobial programs to contain mycobacterial infection. Upon activation, they undergo extensive metabolic reprogramming to meet an increase in energy demand, but also to support immune effector functions such as secretion of cytokines and antimicrobial activities. Here, we report that mitochondrial import of pyruvate is linked to production of mitochondrial ROS and control of Mycobacterium avium (M. avium) infection in human primary macrophages. Using chemical inhibition, targeted mass spectrometry and single cell image analysis, we showed that macrophages infected with M. avium switch to aerobic glycolysis without any major imbalances in the tricarboxylic acid cycle volume or changes in the energy charge. Instead, we found that pyruvate import contributes to hyperpolarization of mitochondria in infected cells and increases production of mitochondrial reactive oxygen species by the complex I via reverse electron transport, which reduces the macrophage burden of M. avium. While mycobacterial infections are extremely difficult to treat and notoriously resistant to antibiotics, this work stresses out that compounds specifically inducing mitochondrial reactive oxygen species could present themself as valuable adjunct treatments.

A multi-encoder variational autoencoder controls multiple transformational features in single-cell image analysis

Image-based cell phenotyping relies on quantitative measurements as encoded representations of cells; however, defining suitable representations that capture complex imaging features is challenged by the lack of robust methods to segment cells, identify subcellular compartments, and extract relevant features. Variational autoencoder (VAE) approaches produce encouraging results by mapping an image to a representative descriptor, and outperform classical hand-crafted features for morphology, intensity, and texture at differentiating data. Although VAEs show promising results for capturing morphological and organizational features in tissue, single cell image analyses based on VAEs often fail to identify biologically informative features due to uninformative technical variation. Here we propose a multi-encoder VAE (ME-VAE) in single cell image analysis using transformed images as a self-supervised signal to extract transform-invariant biologically meaningful features, including emergent features not obvious from prior knowledge. We show that the proposed architecture improves analysis by making distinct cell populations more separable compared to traditional and recent extensions of VAE architectures and intensity measurements by enhancing phenotypic differences between cells and by improving correlations to other analytic modalities. Better feature extraction and image analysis methods enabled by the ME-VAE will advance our understanding of complex cell biology and enable discoveries previously hidden behind image complexity ultimately improving medical outcomes and drug discovery.

A high-throughput technique to map cell images to cell positions using a 3D imaging flow cytometer

We develop a high-throughput technique to relate positions of individual cells to their three-dimensional (3D) imaging features with single-cell resolution. The technique is particularly suitable for nonadherent cells where existing spatial biology methodologies relating cell properties to their positions in a solid tissue do not apply. Our design consists of two parts, as follows: recording 3D cell images at high throughput (500 to 1,000 cells/s) using a custom 3D imaging flow cytometer (3D-IFC) and dispensing cells in a first-in-first-out (FIFO) manner using a robotic cell placement platform (CPP). To prevent errors due to violations of the FIFO principle, we invented a method that uses marker beads and DNA sequencing software to detect errors. Experiments with human cancer cell lines demonstrate the feasibility of mapping 3D side scattering and fluorescent images, as well as two-dimensional (2D) transmission images of cells to their locations on the membrane filter for around 100,000 cells in less than 10 min. While the current work uses our specially designed 3D imaging flow cytometer to produce 3D cell images, our methodology can support other imaging modalities. The technology and method form a bridge between single-cell image analysis and single-cell molecular analysis.

Single-cell image analysis reveals over-expression of organic anion transporting polypeptides (OATPs) in human glioblastoma tissue.

Glioblastoma (GBM) is the most common and aggressive primary brain tumor in adults. Whilst the role of the efflux transporters are well established in GBM, the expression and function of uptake transporters, such as the organic anion transporting polypeptide (OATP) family, are not well understood. OATPs possess broad substrate specificity that includes anti-cancer agents; therefore, we sought to investigate the expression of four OATP isoforms in human GBM cell types using patient tumor tissue. We used fluorescent immunohistochemical labeling of paraffin-embedded surgically resected tissues and single-cell image analysis methods to explore the expression of the OATP isoforms in different tumor cell types through co-labeling with cell-type specific markers, such as IBA1 (pan-myeloid), GFAP (tumor cell), PDGFRβ (stromal cell), and UEA-1-lectin (endothelial). We found significant over-expression of all the OATP isoforms (OATP1A2, 2B1, 1C1 and 4A1) in GBM tumor sections when compared to non-neoplastic brain. A single-cell image analysis revealed that OATPs were significantly upregulated throughout the tumor parenchyma, with significantly higher expression found on lectin-positive blood vessels and IBA1-positive myeloid cells in GBM compared to non-tumor brain tissue. Qualitative analysis of the four OATP isoforms demonstrated greater expression of OATP4A1 in peri-necrotic regions of GBM tissue, which correlated with hypoxia-related markers within the Ivy GAP RNAseq dataset. Here, we demonstrate, for the first time, the protein expression of four OATPs in human GBM tissue, including upregulation within the tumor microenvironment by myeloid cells and tumor vasculature, and isoform-specific upregulation within hypoxic niches.

Revealing the spatio-phenotypic patterning of cells in healthy and tumor tissues with mLSR-3D and STAPL-3D.

Despite advances in three-dimensional (3D) imaging, it remains challenging to profile all the cells within a large 3D tissue, including the morphology and organization of the many cell types present. Here, we introduce eight-color, multispectral, large-scale single-cell resolution 3D (mLSR-3D) imaging and image analysis software for the parallelized, deep learning-based segmentation of large numbers of single cells in tissues, called segmentation analysis by parallelization of 3D datasets (STAPL-3D). Applying the method to pediatric Wilms tumor, we extract molecular, spatial and morphological features of millions of cells and reconstruct the tumor’s spatio-phenotypic patterning. In situ population profiling and pseudotime ordering reveals a highly disorganized spatial pattern in Wilms tumor compared to healthy fetal kidney, yet cellular profiles closely resembling human fetal kidney cells could be observed. In addition, we identify previously unreported tumor-specific populations, uniquely characterized by their spatial embedding or morphological attributes. Our results demonstrate the use of combining mLSR-3D and STAPL-3D to generate a comprehensive cellular map of human tumors.

Single-cell image analysis to explore cell-to-cell heterogeneity in isogenic populations.

Single-cell image analysis provides a powerful approach for studying cell-to-cell heterogeneity, which is an important attribute of isogenic cell populations, from microbial cultures to individual cells in multicellular organisms. This phenotypic variability must be explained at a mechanistic level if biologists are to fully understand cellular function and address the genotype-to-phenotype relationship. Variability in single-cell phenotypes is obscured by bulk readouts or averaging of phenotypes from individual cells in a sample; thus, single-cell image analysis enables a higher resolution view of cellular function. Here, we consider examples of both small- and large-scale studies carried out with isogenic cell populations assessed by fluorescence microscopy, and we illustrate the advantages, challenges, and the promise of quantitative single-cell image analysis.

Single cell morphological metrics and cytoskeletal alignment regulate VCAM-1 protein expression

In the initial stages of atherosclerosis, vascular adhesion molecule-1 (VCAM-1) is a surface protein that mediates leukocyte adhesion to the endothelium’s luminal surface. VCAM-1 expression is upregulated on endothelial cells (ECs) under pro-inflammatory conditions and is known to be modulated by fluid shear stress (FSS). High, pulsatile FSS induces endothelial elongation and cytoskeletal alignment and downregulates pro-inflammatory induced VCAM-1 expression, which is associated with an athero-protective EC phenotype. In contrast, athero-prone ECs under low, oscillatory FSS fail to elongate and maintain a cobblestone morphology with random cytoskeletal alignment, while VCAM-1 expression is upregulated. Whether EC shape and cytoskeletal alignment play a role in the regulation of VCAM-1 protein expression independent of FSS has not been previously determined. The goal of this study was to determine the effect of EC morphology, specifically cell elongation and alignment, and cytoskeletal alignment on VCAM-1 protein expression using topographical micropatterning of an endothelial monolayer and single cell image analysis techniques. Elongated ECs with an aligned cytoskeleton significantly downregulated VCAM-1 protein expression in the absence of FSS compared to planar controls. In addition, linear correlations between morphological metrics and protein expression showed that actin alignment had a significantly stronger effect on VCAM-1 expression than cell elongation. Functionally, monocytic U937 cells statically adhered less on micropatterns compared to planar substrates, in a VCAM-1 dependent manner. Therefore, endothelial cellular elongation and alignment as well as cytoskeletal alignment regulate VCAM-1 protein expression and immunogenic functions to produce a less inflammatory phenotype in the absence of hemodynamic effects.

Predicting the Estrogen Receptor Activity of Environmental Chemicals by Single-Cell Image Analysis and Data-driven Modeling

A comprehensive evaluation of toxic chemicals and understanding their potential harm to human physiology is vital in mitigating their adverse effects following exposure from environmental emergencies. In this work, we develop data-driven classification models to facilitate rapid decision making in such catastrophic events and predict the estrogenic activity of environmental toxicants as estrogen receptor-α (ERα) agonists or antagonists. By combining high-content analysis, big-data analytics, and machine learning algorithms, we demonstrate that highly accurate classifiers can be constructed for evaluating the estrogenic potential of many chemicals. We follow a rigorous, high throughput microscopy-based high-content analysis pipeline to measure the single cell-level response of benchmark compounds with known in vivo effects on the ERα pathway. The resulting high-dimensional dataset is then pre-processed by fitting a non-central gamma probability distribution function to each feature, compound, and concentration. The characteristic parameters of the distribution, which represent the mean and the shape of the distribution, are used as features for the classification analysis via Random Forest (RF) and Support Vector Machine (SVM) algorithms. The results show that the SVM classifier can predict the estrogenic potential of benchmark chemicals with higher accuracy than the RF algorithm, which misclassifies two antagonist compounds.

Single-cell image analysis reveals a protective role for microglia in glioblastoma.

BackgroundMicroglia and tumor-associated macrophages (TAMs) constitute up to half of the total tumor mass of glioblastomas. Despite these myeloid populations being ontogenetically distinct, they have been largely conflated. Recent single-cell transcriptomic studies have identified genes that distinguish microglia from TAMs. Here we investigated whether the translated proteins of genes enriched in microglial or TAM populations can be used to differentiate these myeloid cells in immunohistochemically stained human glioblastoma tissue.MethodsTissue sections from resected low-grade, meningioma, and glioblastoma (grade IV) tumors and epilepsy tissues were immunofluorescently triple-labeled for Iba1 (pan-myeloid marker), CD14 or CD163 (preferential TAM markers), and either P2RY12 or TMEM119 (microglial-specific markers). Using a single-cell-based image analysis pipeline, we quantified the abundance of each marker within single myeloid cells, allowing the identification and analysis of myeloid populations.ResultsP2RY12 and TMEM119 successfully discriminated microglia from TAMs in glioblastoma. In contrast, CD14 and CD163 expression were not restricted to invading TAMs and were upregulated by tumor microglia. Notably, a higher ratio of microglia to TAMs significantly correlated with increased patient survival.ConclusionsWe demonstrate the validity of previously defined microglial-specific genes P2RY12 and TMEM119 as robust discriminators of microglia and TAMs at the protein level in human tissue. Moreover, our data suggest that a higher proportion of microglia may be beneficial for patient survival in glioblastoma. Accordingly, this tissue-based method for myeloid population differentiation could serve as a useful prognostic tool.

Abstract PO-084: Protein marker heterogeneity in breast cancer subtypes measured using immunofluorescence protein multiplexing and quantitative, single cell image analysis

Abstract The extent of intra-tumoral heterogeneity - the variation in the composition of cells in a given tumor and those in the local tumor microenvironment - could have potential impact on diagnosis, treatment planning and subsequent response to treatment. To evaluate the extent of cancer cellular heterogeneity, we conducted quantitative protein marker multiplex imaging to study the variations in protein marker expression patterns on individual cells and spatial localizations. A multiplexed immunofluorescence imaging platform (MxIF, Cell DIVE) was used to measure the cellular expression of Estrogen Receptor (ER), Progesterone Receptor (PR), Epidermal Growth Factor Receptor 2 (HER2), Ki67, p53, p21WAF1 and p16INK4A in cancer epithelium. Analysis was conducted on a tissue microarray (TMA) representing subtypes classified as Luminal A-like, Luminal B-like (HER2-negative), Luminal B-like (HER2-positive), HER2-positive (non-luminal) or Triple-negative based on tumor grade and immune activity according to the St. Gallen surrogate classification. Of the 101 cores from 59 cases studied, high levels of heterogeneity were observed in ER and PR expression among the hormonal receptor-positive tumors. As expected Luminal A-like cancers exhibited higher proportions of individual cells co-expressing ER and PR, while cells in Luminal B-like, HER2-negative cancers showed ER expression only. Luminal B-like, HER2-positive cores were composed of cells with strong HER2 staining, and some cells co-expressing PR and HER2. Single cells with strong ER and HER2 labelling were rarely observed. Spatial visualizations illustrated that cells with similar expression signatures tend to be clustered together. Among cases which showed p53 overexpression with immunohistochemistry, the overall MxIF-measured p53 level was highest in TNBC compared to HER2+ and Luminal B-like cases. TNBC exhibited the highest proliferative fraction and most incidence of abnormal p53 and p16. We did not observe an association of p21 expression to P53 or P16 patterns, yet a slightly higher proportion of Luminal B-like cancers showed increased P21 levels compared to the other subtypes. Our study demonstrated the application of protein marker multiplexing and quantitative image analysis in measuring heterogeneity of protein co-expression signatures within breast cancer subtypes. Our next step is to apply the methods developed here to study a cohort where molecular profiling and radiomics were conducted (Bayani et al.) to reveal the extent of heterogeneity of breast cancer with a multi-omics approach. Citation Format: Alison M. Cheung, Dan Wang, Kela Liu, Fiona Ginty, Sharon Nofech-Mozes, Jane Bayani, John M.S. Bartlett, Anne Martel, Martin J. Yaffe. Protein marker heterogeneity in breast cancer subtypes measured using immunofluorescence protein multiplexing and quantitative, single cell image analysis [abstract]. In: Proceedings of the AACR Virtual Special Conference on Tumor Heterogeneity: From Single Cells to Clinical Impact; 2020 Sep 17-18. Philadelphia (PA): AACR; Cancer Res 2020;80(21 Suppl):Abstract nr PO-084.

Identification of a dysfunctional microglial population in human Alzheimer\u2019s disease cortex using novel single-cell histology image analysis

In Alzheimer’s disease (AD), microglia are affected by disease processes, but may also drive pathogenesis. AD pathology-associated microglial populations have been identified with single-cell RNA-Seq, but have not been validated in human brain tissue with anatomical context. Here, we quantified myeloid cell markers to identify changes in AD pathology-associated microglial populations. We performed fluorescent immunohistochemistry on normal (n = 8) and AD (n = 8) middle temporal gyri, co-labelling the pan-myeloid cell marker, Iba1, with one of 11 markers of interest (MOIs): CD45, HLA-DR, CD14, CD74, CD33, CD206, CD32, CD163, P2RY12, TMEM119, L-Ferritin. Novel image analyses quantified the single-cell abundance of Iba1 and each MOI. Each cell was gated into one Iba1-MOI population (Iba1low MOIhigh, Iba1high MOIhigh, or Iba1high MOIlow) and the abundance of each population was compared between AD and control. Triple-labelling of L-Ferritin and Iba1 with a subset of MOIs was performed to investigate L-Ferritin-MOI co-expression on Iba1low cells. Iba1low MOIhigh myeloid cell populations delineated by MOIs CD45, HLA-DR, CD14, CD74, CD33, CD32, and L-Ferritin were increased in AD. Further investigation of the Iba1low MOIhigh populations revealed that their abundances correlated with tau, but not amyloid beta, load in AD. The Iba1low microglial population highly expressed L-Ferritin, reflecting microglial dysfunction. The L-Ferritinhigh CD74high HLA-DRhigh phenotype of the Iba1low population mirrors that of a human AD pathology-associated microglial subpopulation previously identified using single-cell RNA-Seq. Our high-throughput immunohistochemical data with anatomical context support the microglial dysfunction hypothesis of AD.

Abstract PR02: Profiling intratumoral heterogeneity of bladder cancer subtypes at the single-cell level using machine-learning assisted histopathology

Abstract Purpose: Machine-learning assisted histopathology using markers of basal and luminal differentiation was employed to profile the intratumoral heterogeneity of bladder cancer from cystectomy patients and predict disease-free survival in this high-risk patient population. Methods: Urothelial carcinomas are biologically heterogeneous and vary greatly in clinical progression as well as treatment response. Delineation of molecular subtypes by gene expression analysis of luminal and basal markers has indicated differential outcomes associated with basal and luminal subtypes. However, histologic validation of this classification using protein markers (basal = KRT5/6, P63; luminal = KRT20/GATA3) has been challenging. While using multiplex-immunofluorescence to subtype a retrospective cystectomy cohort (a TMA of 380 patients), we determined that nearly 50% of tumors did not exhibit cytokeratin markers. Subtyping was further confounded by frequent loss of basal-to-luminal stratification and the emergence of intratumoral spatial heterogeneity with the basal and luminal subtypes being completely intermixed throughout the tumor. These observations caused us to hypothesize that previously undefined but clinically relevant subtypes might exist. To address this challenge we developed a single-cell image analysis pipeline that leveraged machine learning to classify molecular subtype and spatial heterogeneity within each tumor. Using the informatics software KNIME we achieved single-cell segmentation and extracted 285 features for 5 protein markers (P63, GATA3, collagen, nuclear stain, and pan-cytokeratin) from each ~20,000 cells contained in 2 cores of tumor and adjacent benign for each patient. Under guidance from a pathologist, definitive urothelial cells (luminal, intermediate, and basal cells) as well as stromal cells were selected from 25 cores normal urothelium to form the ground truth for XGboost-based machine-learning. Summary Findings: Single-cell profiling with machine learning on transcription factors could classify basal and luminal subtypes with greater than 97% accuracy according to validation in normal urothelium using keratin markers. While we were able to recapitulate differential survival associated with a pure basal subtype, it was the intratumoral heterogeneity of basal and luminal cells that was the predominant driver of disease-free survival. Conclusion: A newly identified bladder cancer subtype defined by intratumoral heterogeneity is a clinically relevant driver of disease-free survival. This abstract is also being presented as Poster B22. Citation Format: Adel Eskaros, Tatiana Novitskaya, Andries Zijlstra. Profiling intratumoral heterogeneity of bladder cancer subtypes at the single-cell level using machine-learning assisted histopathology [abstract]. In: Proceedings of the AACR Special Conference on Bladder Cancer: Transforming the Field; 2019 May 18-21; Denver, CO. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(15_Suppl):Abstract nr PR02.