FlowSOM: Using self-organizing maps for visualization and interpretation of cytometry data

Quantitative Comparison of Conventional and t-SNE-guided Gating Analyses

Dimensionality reduction using the t-Distributed Stochastic Neighbor Embedding (t-SNE) algorithm has emerged as a popular tool for visualizing high-parameter single-cell data. While this approach has obvious potential for data visualization it remains unclear how t-SNE analysis compares to conventional manual hand-gating in stratifying and quantitating the frequency of diverse immune cell populations. We applied a comprehensive 38-parameter mass cytometry panel to human blood and compared the frequencies of 28 immune cell subsets using both conventional bivariate and t-SNE-guided manual gating. t-SNE analysis was capable of stratifying every general cellular lineage and most sub-lineages with high correlation between conventional and t-SNE-guided cell frequency calculations. However, specific immune cell subsets delineated by the manual gating of continuous variables were not fully separated in t-SNE space thus causing discrepancies in subset identification and quantification between these analytical approaches. Overall, these studies highlight the consistency between t-SNE and conventional hand-gating in stratifying general immune cell lineages while demonstrating that particular cell subsets defined by conventional manual gating may be intermingled in t-SNE space.

High Salt Inhibits Tumor Growth by Enhancing Anti-tumor Immunity

Excess salt intake could affect the immune system by shifting the immune cell balance toward a pro-inflammatory state. Since this shift of the immune balance is thought to be beneficial in anti-cancer immunity, we tested the impact of high salt diets on tumor growth in mice. Here we show that high salt significantly inhibited tumor growth in two independent murine tumor transplantation models. Although high salt fed tumor-bearing mice showed alterations in T cell populations, the effect seemed to be largely independent of adaptive immune cells. In contrast, depletion of myeloid-derived suppressor cells (MDSCs) significantly reverted the inhibitory effect on tumor growth. In line with this, high salt conditions almost completely blocked murine MDSC function in vitro. Importantly, similar effects were observed in human MDSCs isolated from cancer patients. Thus, high salt conditions seem to inhibit tumor growth by enabling more pronounced anti-tumor immunity through the functional modulation of MDSCs. Our findings might have critical relevance for cancer immunotherapy.

Development, application and computational analysis of high-dimensional fluorescent antibody panels for single-cell flow cytometry

The interrogation of single cells is revolutionizing biology, especially our understanding of the immune system. Flow cytometry is still one of the most versatile and high-throughput approaches for single-cell analysis, and its capability has been recently extended to detect up to 28 colors, thus approaching the utility of cytometry by time of flight (CyTOF). However, flow cytometry suffers from autofluorescence and spreading error (SE) generated by errors in the measurement of photons mainly at red and far-red wavelengths, which limit barcoding and the detection of dim markers. Consequently, development of 28-color fluorescent antibody panels for flow cytometry is laborious and time consuming. Here, we describe the steps that are required to successfully achieve 28-color measurement capability. To do this, we provide a reference map of the fluorescence spreading errors in the 28-color space to simplify panel design and predict the success of fluorescent antibody combinations. Finally, we provide detailed instructions for the computational analysis of such complex data by existing, popular algorithms (PhenoGraph and FlowSOM). We exemplify our approach by designing a high-dimensional panel to characterize the immune system, but we anticipate that our approach can be used to design any high-dimensional flow cytometry panel of choice. The full protocol takes a few days to complete, depending on the time spent on panel design and data analysis.

Learning Single-Cell Distances from Cytometry Data

Recent years have seen an increased interest in employing data analysis techniques for the automated identification of cell populations in the field of cytometry. These techniques highly depend on the use of a distance metric, a function that quantifies the distances between single-cell measurements. In most cases, researchers simply use the Euclidean distance metric. In this article, we exploit the availability of single-cell labels to find an optimal Mahalanobis distance metric derived from the data. We show that such a Mahalanobis distance metric results in an improved identification of cell populations compared with the Euclidean distance metric. Once determined, it can be used for the analysis of multiple samples that were measured under the same experimental setup. We illustrate this approach for cytometry data from two different origins, that is, flow cytometry applied to microbial cells and mass cytometry for the analysis of human blood cells. We also illustrate that such a distance metric results in an improved identification of cell populations when clustering methods are employed. Generally, these results imply that the performance of data analysis techniques can be improved by using a more advanced distance metric. © 2019 International Society for Advancement of Cytometry.

diffcyt: Differential discovery in high-dimensional cytometry via high-resolution clustering

High-dimensional flow and mass cytometry allow cell types and states to be characterized in great detail by measuring expression levels of more than 40 targeted protein markers per cell at the single-cell level. However, data analysis can be difficult, due to the large size and dimensionality of datasets as well as limitations of existing computational methods. Here, we present diffcyt, a new computational framework for differential discovery analyses in high-dimensional cytometry data, based on a combination of high-resolution clustering and empirical Bayes moderated tests adapted from transcriptomics. Our approach provides improved statistical performance, including for rare cell populations, along with flexible experimental designs and fast runtimes in an open-source framework.

Analysis of High-Dimensional Phenotype Data Generated by Mass Cytometry or High-Dimensional Flow Cytometry

Recent advances in single cell multi-omics methodologies significantly expand our understanding of cellular heterogeneity, particularly in the field of immunology. Today's state-of-the-art flow and mass cytometers can detect up to 50 parameters to comprehensively characterize the identity and function of individual cells within a heterogeneous population. As a consequence, the increasing number of parameters that can be detected simultaneously also introduces substantial complexity for the experimental setup and downstream data processing. However, this challenge in data analysis fostered the development of novel bioinformatic tools to fully exploit the high-dimensional data. These tools will eventually replace cumbersome serial, manual gating in the two-dimensional space driven by a priori knowledge, which still represents the gold standard in flow cytometric analysis, to meet the needs of the today's immunologist. To this end, we provide guidelines for a high-dimensional cytometry workflow including experimental setup, panel design, fluorescent spillover compensation, and data analysis.

A novel data fusion method for the effective analysis of multiple panels of flow cytometry data

Multicolour flow cytometry (MFC) is used to measure multiple cellular markers at the single-cell level. Cellular markers may be coloured with different panels of fluorescently-labelled antibodies to enable cell identification or the detection of activated cells in pre-defined, ‘gated’ specific cell subsets. The number of markers that can be used per measurement is technologically limited however, requiring every panel to be analysed in a separate aliquot measurement. The combined analyses of these dedicated panels may enhance the predictive ability of these measurements and could enrich the interpretation of the immunological information. Here we introduce a fusion method for MFC data, based on DAMACY (Discriminant Analysis of Multi-Aspect Cytometry data), which can combine information from complementary panels. This approach leads to both enhanced predictions and clearer interpretations in comparison with the analysis of separate measurements. We illustrate this method using two datasets: the response of neutrophils evoked by a systemic endotoxin challenge and the activated immune status of the innate cells, T cells and B cells in obese versus lean individuals. The data fusion approach was able to detect cells that do not individually show a difference between clinical phenotypes but do play a role in combination with other cells.

Ultrafast clustering of single-cell flow cytometry data using FlowGrid

Background

Flow cytometry is a popular technology for quantitative single-cell profiling of cell surface markers. It enables expression measurement of tens of cell surface protein markers in millions of single cells. It is a powerful tool for discovering cell sub-populations and quantifying cell population heterogeneity. Traditionally, scientists use manual gating to identify cell types, but the process is subjective and is not effective for large multidimensional data. Many clustering algorithms have been developed to analyse these data but most of them are not scalable to very large data sets with more than ten million cells.

Results

Here, we present a new clustering algorithm that combines the advantages of density-based clustering algorithm DBSCAN with the scalability of grid-based clustering. This new clustering algorithm is implemented in python as an open source package, FlowGrid. FlowGrid is memory efficient and scales linearly with respect to the number of cells. We have evaluated the performance of FlowGrid against other state-of-the-art clustering programs and found that FlowGrid produces similar clustering results but with substantially less time. For example, FlowGrid is able to complete a clustering task on a data set of 23.6 million cells in less than 12 seconds, while other algorithms take more than 500 seconds or get into error.

Conclusions

FlowGrid is an ultrafast clustering algorithm for large single-cell flow cytometry data. The source code is available at https://github.com/VCCRI/FlowGrid.

High-Throughput Analysis of Clinical Flow Cytometry Data by Automated Gating

Advancements in flow cytometers with capability to measure 15 or more parameters have enabled us to characterize cell populations at unprecedented levels of detail. Beyond discovery research, there is now a growing demand to dive deeper into evaluating the immune response in clinical trials for immune modulating compounds. However, for high-volume, complex flow cytometry data generated in clinical trials, conventional manual gating remains the standard of practice. Traditional manual gating is resource intense and becomes a bottleneck and an impractical method to complete high volumes of flow cytometry data analysis. Current efforts to automate “manual gating” have shown that computational algorithms can facilitate the analysis of daunting multi-parameter data; however, a greater degree of precision in comparison with traditional manual gating is needed for wide-scale adoption of automated gating methods. In an effort to more closely follow the manual gating process, our automated gating pipeline was created to include negative controls (Fluorescence Minus One [FMO]) to enhance the reliability of gate placement. We demonstrate that use of an automated pipeline, heavily relying on FMO controls for population discrimination, can analyze multi-parameter, large-scale clinical datasets with comparable precision and accuracy to traditional manual gating.

Expansion and activation of distinct central memory T lymphocyte subsets in complex regional pain syndrome

BACKGROUND

Complex regional pain syndrome (CRPS) is a debilitating condition where trauma to a limb results in devastating persistent pain that is disproportionate to the initial injury. The pathophysiology of CRPS remains unknown; however, accumulating evidence suggests it is an immunoneurological disorder, especially in light of evidence of auto-antibodies in ~ 30% of patients. Despite this, a systematic assessment of all circulating leukocyte populations in CRPS has never been performed.

METHODS

We characterised 14 participants as meeting the Budapest clinical criteria for CRPS and assessed their pain ratings and psychological state using a series of questionnaires. Next, we performed immunophenotyping on blood samples from the 14 CRPS participants as well as 14 healthy pain-free controls using mass cytometry. Using a panel of 38 phenotypic and activation markers, we characterised the numbers and intracellular activation status of all major leukocyte populations using manual gating strategies and unsupervised cluster analysis.

RESULTS

We have shown expansion and activation of several distinct populations of central memory T lymphocytes in CRPS. The number of central memory CD8+ T cells was increased 2.15-fold; furthermore, this cell group had increased phosphorylation of NFkB and STAT1 compared to controls. Regarding central memory CD4+ T lymphocytes, the number of Th1 and Treg cells was increased 4.98-fold and 2.18-fold respectively, with increased phosphorylation of NFkB in both populations. We also found decreased numbers of CD1c+ myeloid dendritic cells, although with increased p38 phosphorylation. These changes could indicate dendritic cell tissue trafficking, as well as their involvement in lymphocyte activation.

CONCLUSIONS

These findings represent the first mass cytometry immunophenotyping study in any chronic pain state and provide preliminary evidence of an antigen-mediated T lymphocyte response in CRPS. In particular, the presence of increased numbers of long-lived central memory CD4+ and CD8+ T lymphocytes with increased activation of pro-inflammatory signalling pathways may indicate ongoing inflammation and cellular damage in CRPS.

Predicting Cell Populations in Single Cell Mass Cytometry Data

Mass cytometry by time‐of‐flight (CyTOF) is a valuable technology for high‐dimensional analysis at the single cell level. Identification of different cell populations is an important task during the data analysis. Many clustering tools can perform this task, which is essential to identify “new” cell populations in explorative experiments. However, relying on clustering is laborious since it often involves manual annotation, which significantly limits the reproducibility of identifying cell‐populations across different samples. The latter is particularly important in studies comparing different conditions, for example in cohort studies. Learning cell populations from an annotated set of cells solves these problems. However, currently available methods for automatic cell population identification are either complex, dependent on prior biological knowledge about the populations during the learning process, or can only identify canonical cell populations. We propose to use a linear discriminant analysis (LDA) classifier to automatically identify cell populations in CyTOF data. LDA outperforms two state‐of‐the‐art algorithms on four benchmark datasets. Compared to more complex classifiers, LDA has substantial advantages with respect to the interpretable performance, reproducibility, and scalability to larger datasets with deeper annotations. We apply LDA to a dataset of ~3.5 million cells representing 57 cell populations in the Human Mucosal Immune System. LDA has high performance on abundant cell populations as well as the majority of rare cell populations, and provides accurate estimates of cell population frequencies. Further incorporating a rejection option, based on the estimated posterior probabilities, allows LDA to identify previously unknown (new) cell populations that were not encountered during training. Altogether, reproducible prediction of cell population compositions using LDA opens up possibilities to analyze large cohort studies based on CyTOF data. © 2019 The Authors. Cytometry Part A published by Wiley Periodicals, Inc. on behalf of International Society for Advancement of Cytometry.

BAFF-driven B cell hyperplasia underlies lung disease in common variable immunodeficiency

BACKGROUND

Common variable immunodeficiency (CVID) is the most common symptomatic primary immunodeficiency and is frequently complicated by interstitial lung disease (ILD) for which etiology is unknown and therapy inadequate.

METHODS

Medical record review implicated B cell dysregulation in CVID ILD progression. This was further studied in blood and lung samples using culture, cytometry, ELISA, and histology. Eleven CVID ILD patients were treated with rituximab and followed for 18 months.

RESULTS

Serum IgM increased in conjunction with ILD progression, a finding that reflected the extent of IgM production within B cell follicles in lung parenchyma. Targeting these pulmonary B cell follicles with rituximab ameliorated CVID ILD, but disease recurred in association with IgM elevation. Searching for a stimulus of this pulmonary B cell hyperplasia, we found B cell-activating factor (BAFF) increased in blood and lungs of progressive and post-rituximab CVID ILD patients and detected elevation of BAFF-producing monocytes in progressive ILD. This elevated BAFF interacts with naive B cells, as they are the predominant subset in progressive CVID ILD, expressing BAFF receptor (BAFF-R) within pulmonary B cell follicles and blood to promote Bcl-2 expression. Antiapoptotic Bcl-2 was linked with exclusion of apoptosis from B cell follicles in CVID ILD and increased survival of naive CVID B cells cultured with BAFF.

CONCLUSION

CVID ILD is driven by pulmonary B cell hyperplasia that is reflected by serum IgM elevation, ameliorated by rituximab, and bolstered by elevated BAFF-mediated apoptosis resistance via BAFF-R.

FUNDING

NIH, Primary Immune Deficiency Treatment Consortium, and Rare Disease Foundation.

RIPK3-Induced Inflammation by I-MDSCs Promotes Intestinal Tumors

Myeloid-derived suppressor cells (MDSC) promote colorectal cancer by several mechanisms, including suppression of antitumor T cells and production of tumorigenic factors. We previously showed that an intermediate MDSC subset (I-MDSC) is expanded in an intestinal tumor model (Apc^Min/+ mice), but the importance of this subset in promoting tumors is unclear. Here, we show that I-MDSCs are a distinct heterogeneous subset due to differential and reduced expression of the monocytic marker, Ly6C, and granulocytic marker, Ly6G. Besides causing necroptotic cell death, receptor-interacting protein kinase 3 (RIPK3) has an alternate function as a signaling component inducing cytokine synthesis. We evaluated whether RIPK3 regulates inflammatory cytokines in I-MDSCs to assess the nonimmunosuppressive function of I-MDSCs in promoting tumors. Inhibition of RIPK3 with the commercially available small-molecule inhibitor GSK 872 showed that RIPK3-mediated inflammation promoted intestinal tumors in two intestinal tumor models, Apc^Min/+ mice and an MC38 transplantable tumor model. Mechanistically, RIPK3 signaling in I-MDSC increased tumor size by expanding IL17-producing T cells in MC38 tumors. Collectively, these data suggest RIPK3 signaling as a potential therapeutic target in colorectal cancer. SIGNIFICANCE: The specific role of RIPK3 in intestinal tumors and MDSC function sheds light on a key inflammatory mechanism driving tumorigenesis and allows for possible therapeutic intervention.

Evaluation of Clustering Methods in Compression of Topological Models and Visual Place Recognition Using Global Appearance Descriptors

This paper presents an extended study about the compression of topological models of indoor environments. The performance of two clustering methods is tested in order to know their utility both to build a model of the environment and to solve the localization task. Omnidirectional images are used to create the compact model, as well as to estimate the robot position within the environment. These images are characterized through global appearance descriptors, since they constitute a straightforward mechanism to build a compact model and estimate the robot position. To evaluate the goodness of the proposed clustering algorithms, several datasets are considered. They are composed of either panoramic or omnidirectional images captured in several environments, under real operating conditions. The results confirm that compression of visual information contributes to a more efficient localization process through saving computation time and keeping a relatively good accuracy.

RORγt inhibition selectively targets IL-17 producing iNKT and γδ-T cells enriched in Spondyloarthritis patients

Dysregulated IL-23/IL-17 responses have been linked to psoriatic arthritis and other forms of spondyloarthritides (SpA). RORγt, the key Thelper17 (Th17) cell transcriptional regulator, is also expressed by subsets of innate-like T cells, including invariant natural killer T (iNKT) and γδ-T cells, but their contribution to SpA is still unclear. Here we describe the presence of particular RORγt⁺T-bet^loPLZF^- iNKT and γδ-hi T cell subsets in healthy peripheral blood. RORγt⁺ iNKT and γδ-hi T cells show IL-23 mediated Th17-like immune responses and were clearly enriched within inflamed joints of SpA patients where they act as major IL-17 secretors. SpA derived iNKT and γδ-T cells showed unique and Th17-skewed phenotype and gene expression profiles. Strikingly, RORγt inhibition blocked γδ17 and iNKT17 cell function while selectively sparing IL-22⁺ subsets. Overall, our findings highlight a unique diversity of human RORγt⁺ T cells and underscore the potential of RORγt antagonism to modulate aberrant type 17 responses.

Analysis of the Murine Bone Marrow Hematopoietic System Using Mass and Flow Cytometry

The hematopoietic system produces erythrocytes (red blood cells), leukocytes (white blood cells), and thrombocytes (platelets) throughout the life of an organism. Long-lived hematopoietic stem cells give rise to early progenitors with multi-lineage potential that progressively differentiate into lineage-specific progenitors. Following lineage commitment, these progenitors proliferate and expand, before eventually differentiating into their mature forms. This process drives the up- and downregulation of a wide variety of surface and intracellular markers throughout differentiation, making cytometric analysis of this interconnected system challenging. Moreover, during inflammation, the hematopoietic system can be mobilized to re-prioritize the production of various lineages, in order to match increased demand, often at the expense of other lineages. As such, the response of the hematopoietic system in the bone marrow (BM) is a critical component of both immunity and disease. Because of the complexity of the hematopoietic system in steady state and disease, high-dimensional cytometry technologies are well suited to the exploration of these complex systems. Here we describe a protocol for the extraction of murine bone marrow, and preparation for examination using high-dimensional flow or mass cytometry. Additionally, we describe methods for performing cell cycle assays using bromodeoxyuridine (BrdU) or iododeoxyuridine (IdU). Finally, we describe an analytical method that allows for a system-level analysis of the hematopoietic system in steady state or inflammatory scenarios.

Analysis of Mass Cytometry Data

The CyTOF system produces single cell protein expression data similar to that from flow cytometry, but with an increased number of features measured. Traditionally, analysis of these data is carried out using manual gating, but with the increased dimensionality, manual gating becomes a suboptimal analysis strategy in some cases. To address this, a number of data analysis tools for tasks such as clustering, differential abundance analysis, and visualization have been developed and made freely available. We here introduce some of the more popular tools for CyTOF analysis and exemplify their utility in a common analysis workflow.

Flow-Cytometric Monitoring of Minimal Residual Disease in Pediatric Patients With Acute Myeloid Leukemia: Recent Advances and Future Strategies

Minimal residual disease (MRD) by multiparametric flow cytometry (MFC) has been recently shown as a strong and independent prognostic marker of relapse in pediatric AML (pedAML) when measured at specific time points during Induction and/or Consolidation therapy. Hence, MFC-MRD has the potential to refine the current strategies of pedAML risk stratification, traditionally based on the cytogenetic and molecular genetic aberrations at diagnosis. Consequently, it may guide the modulation of therapy intensity and clinical decision making. However, the use of non-standardized protocols, including different staining panels, analysis, and gating strategies, may hamper a broad implementation of MFC-MRD monitoring in clinical routine. Besides, the thresholds of MRD positivity still need to be validated in large, prospective and multi-center clinical studies, as well as optimal time points of MRD assessment during therapy, to better discriminate patients with different prognosis. In the present review, we summarize the most relevant findings on MFC-MRD testing in pedAML. We examine the clinical significance of MFC-MRD and the recent advances in its standardization, including innovative approaches with an automated analysis of MFC-MRD data. We also touch upon other technologies for MRD assessment in AML, such as quantitative genomic breakpoint PCR, current challenges and future strategies to enable full incorporation of MFC-MRD into clinical practice.

Stabilization of cytokine mRNAs in iNKT cells requires the serine-threonine kinase IRE1alpha

Activated invariant natural killer T (iNKT) cells rapidly produce large amounts of cytokines, but how cytokine mRNAs are induced, stabilized and mobilized following iNKT activation is still unclear. Here we show that an endoplasmic reticulum stress sensor, inositol-requiring enzyme 1α (IRE1α), links key cellular processes required for iNKT cell effector functions in specific iNKT subsets, in which TCR-dependent activation of IRE1α is associated with downstream activation of p38 MAPK and the stabilization of preformed cytokine mRNAs. Importantly, genetic deletion of IRE1α in iNKT cells reduces cytokine production and protects mice from oxazolone colitis. We therefore propose that an IRE1α-dependent signaling cascade couples constitutive cytokine mRNA expression to the rapid induction of cytokine secretion and effector functions in activated iNKT cells.

Invariant natural killer T (iNKT) cells rapidly enhance cytokine secretion and effector function following activation, but the underlying mechanism is still unclear. Here the authors show that an endoplasmic reticulum stress sensor, inositol-requiring enzyme 1α, activates the p38 kinase to stabilize cytokine mRNA for enhanced iNKT functions.

Expression of PD-1 by T Cells in Malignant Glioma Patients Reflects Exhaustion and Activation

PURPOSE

Glioblastoma (GBM) is the most common primary malignant tumor in the central nervous system. Our recent preclinical work has suggested that PD-1/PD-L1 plays an important immunoregulatory role to limit effective antitumor T-cell responses induced by active immunotherapy. However, little is known about the functional role that PD-1 plays on human T lymphocytes in patients with malignant glioma.Experimental Design: In this study, we examined the immune landscape and function of PD-1 expression by T cells from tumor and peripheral blood in patients with malignant glioma.

RESULTS

We found several differences between PD-1⁺ tumor-infiltrating lymphocytes (TIL) and patient-matched PD-1⁺ peripheral blood T lymphocytes. Phenotypically, PD-1⁺ TILs exhibited higher expression of markers of activation and exhaustion than peripheral blood PD-1⁺ T cells, which instead had increased markers of memory. A comparison of the T-cell receptor variable chain populations revealed decreased diversity in T cells that expressed PD-1, regardless of the location obtained. Functionally, peripheral blood PD-1⁺ T cells had a significantly increased proliferative capacity upon activation compared with PD-1^- T cells.

CONCLUSIONS

Our evidence suggests that PD-1 expression in patients with glioma reflects chronically activated effector T cells that display hallmarks of memory and exhaustion depending on its anatomic location. The decreased diversity in PD-1⁺ T cells suggests that the PD-1-expressing population has a narrower range of cognate antigen targets compared with the PD-1 nonexpression population. This information can be used to inform how we interpret immune responses to PD-1-blocking therapies or other immunotherapies.

Immunophenotypic measurable residual disease (MRD) in acute myeloid leukemia: Is multicentric MRD assessment feasible?

Flow-cytometric detection of now termed measurable residual disease (MRD) in acute myeloid leukemia (AML) has proven to have an independent prognostic impact. In a previous multicenter study we developed protocols to accurately define leukemia-associated immunophenotypes (LAIPs) at diagnosis. It has, however, not been demonstrated whether the use of the defined LAIPs in the same multicenter setting results in a high concordance between centers in MRD assessment. In the present paper we evaluated whether interpretation of list-mode data (LMD) files, obtained from MRD assessment of previously determined LAIPs during and after treatment, could reliably be performed in a multicenter setting. The percentage of MRD positive cells was simultaneously determined in totally 173 LMD files from 77 AML patients by six participating centers. The quantitative concordance between the six participating centers was meanly 84%, with slight variation of 75%-89%. In addition our data showed that the type and number of LAIPs were of influence on the performance outcome. The highest concordance was observed for LAIPs with cross-lineage expression, followed by LAIPs with an asynchronous antigen expression. Our results imply that immunophenotypic MRD assessment in AML will only be feasible when fully standardized methods are used for reliable multicenter assessment.

Computational Immune Monitoring Reveals Abnormal Double-Negative T Cells Present across Human Tumor Types

Advances in single-cell biology have enabled measurements of >40 protein features on millions of immune cells within clinical samples. However, the data analysis steps following cell population identification are susceptible to bias, time-consuming, and challenging to compare across studies. Here, an ensemble of unsupervised tools was developed to evaluate four essential types of immune cell information, incorporate changes over time, and address diverse immune monitoring challenges. The four complementary properties characterized were (i) systemic plasticity, (ii) change in population abundance, (iii) change in signature population features, and (iv) novelty of cellular phenotype. Three systems immune monitoring studies were selected to challenge this ensemble approach. In serial biopsies of melanoma tumors undergoing targeted therapy, the ensemble approach revealed enrichment of double-negative (DN) T cells. Melanoma tumor-resident DN T cells were abnormal and phenotypically distinct from those found in nonmalignant lymphoid tissues, but similar to those found in glioblastoma and renal cell carcinoma. Overall, ensemble systems immune monitoring provided a robust, quantitative view of changes in both the system and cell subsets, allowed for transparent review by human experts, and revealed abnormal immune cells present across multiple human tumor types.

Global immune fingerprinting in glioblastoma patient peripheral blood reveals immune-suppression signatures associated with prognosis

Glioblastoma (GBM) remains uniformly lethal, and despite a large accumulation of immune cells in the microenvironment, there is limited antitumor immune response. To overcome these challenges, a comprehensive understanding of GBM systemic immune response during disease progression is required. Here, we integrated multiparameter flow cytometry and mass cytometry TOF (CyTOF) analysis of patient blood to determine changes in the immune system among tumor types and over disease progression. Utilizing flow cytometry analysis in a cohort of 259 patients ranging from benign to malignant primary and metastatic brain tumors, we found that GBM patients had a significant elevation in myeloid-derived suppressor cells (MDSCs) in peripheral blood but not immunosuppressive Tregs. In GBM patient tissue, we found that increased MDSC levels in recurrent GBM portended poor prognosis. CyTOF analysis of peripheral blood from newly diagnosed GBM patients revealed that reduced MDSCs over time were accompanied by a concomitant increase in DCs. GBM patients with extended survival also had reduced MDSCs, similar to the levels of low-grade glioma (LGG) patients. Our findings provide a rationale for developing strategies to target MDSCs, which are elevated in GBM patients and predict poor prognosis.

CytoBinning: Immunological insights from multi-dimensional data

New cytometric techniques continue to push the boundaries of multi-parameter quantitative data acquisition at the single-cell level particularly in immunology and medicine. Sophisticated analysis methods for such ever higher dimensional datasets are rapidly emerging, with advanced data representations and dimensional reduction approaches. However, these are not yet standardized and clinical scientists and cell biologists are not yet experienced in their interpretation. More fundamentally their range of statistical validity is not yet fully established. We therefore propose a new method for the automated and unbiased analysis of high-dimensional single cell datasets that is simple and robust, with the goal of reducing this complex information into a familiar 2D scatter plot representation that is of immediate utility to a range of biomedical and clinical settings. Using publicly available flow cytometry and mass cytometry datasets we demonstrate that this method (termed CytoBinning), recapitulates the results of traditional manual cytometric analyses and leads to new and testable hypotheses.

BRAF and MEK inhibitor therapy eliminates Nestin-expressing melanoma cells in human tumors

Little is known about the in vivo impacts of targeted therapy on melanoma cell abundance and protein expression. Here, 21 antibodies were added to an established melanoma mass cytometry panel to measure 32 cellular features, distinguish malignant cells, and characterize dabrafenib and trametinib responses in BRAFV600mut melanoma. Tumor cells were biopsied before neoadjuvant therapy and compared to cells surgically resected from the same site after 4 weeks of therapy. Approximately 50,000 cells per tumor were characterized by mass cytometry and computational tools t-SNE/viSNE, FlowSOM, and MEM. The resulting single-cell view of melanoma treatment response revealed initially heterogeneous melanoma tumors were consistently cleared of Nestin-expressing melanoma cells. Melanoma cell subsets that persisted to week 4 were heterogeneous but expressed SOX2 or SOX10 proteins and specifically lacked surface expression of MHC I proteins by MEM analysis. Traditional histology imaging of tissue microarrays from the same tumors confirmed mass cytometry results, including persistence of NES- SOX10+ S100β+ melanoma cells. This quantitative single-cell view of melanoma treatment response revealed protein features of malignant cells that are not eliminated by targeted therapy.

High-Dimensional Profiling Reveals Heterogeneity of the Th17 Subset and Its Association With Systemic Immunomodulatory Treatment in Non-infectious Uveitis

Background: Non-infectious uveitis (NIU) is a severe intra ocular inflammation, which frequently requires prompt systemic immunosuppressive therapy (IMT) to halt the development of vision-threatening complications. IMT is considered when NIU cannot be treated with corticosteroids alone, which is unpredictable in advance. Previous studies have linked blood cell subsets to glucocorticoid sensitivity, which suggests that the composition of blood leukocytes may early identify patients that will require IMT. Objective: To map the blood leukocyte composition of NIU and identify cell subsets that stratify patients that required IMT during follow-up. Methods: We performed controlled flow cytometry experiments measuring a total of 37 protein markers in the blood of 30 IMT free patients with active non-infectious anterior, intermediate, and posterior uveitis, and compared these to 15 age and sex matched healthy controls. Results from manual gating were validated by automatic unsupervised gating using FlowSOM. Results: Patients with uveitis displayed lower relative frequencies of Natural Killer cells and higher relative frequencies of memory T cells, in particular the CCR6+ lineages. These results were confirmed by automatic gating by unsupervised clustering using FlowSOM. We observed considerable heterogeneity in memory T cell subsets and abundance of CXCR3-CCR6+ (Th17) cells between the uveitis subtypes. Importantly, regardless of the uveitis subtype, patients that eventually required IMT in the course of the study follow-up exhibited increased CCR6+ T cell abundance before commencing therapy. Conclusion: High-dimensional immunoprofiling in NIU patients shows that clinically distinct forms of human NIU exhibit shared as well as unique immune cell perturbations in the peripheral blood and link CCR6+ T cell abundance to systemic immunomodulatory treatment.

Cytofast: A workflow for visual and quantitative analysis of flow and mass cytometry data to discover immune signatures and correlations

Multi-parametric flow and mass cytometry allows exceptional high-resolution exploration of the cellular composition of the immune system. A large panel of computational tools have been developed to analyze the high-dimensional landscape of the data generated. Analysis frameworks such as FlowSOM or Cytosplore incorporate clustering and dimensionality reduction techniques and include algorithms allowing visualization of multi-parametric cytometric analysis. To additionally provide means to quantify specific cell clusters and correlations between samples, we developed an R-package, called cytofast, for further downstream analysis. Specifically, cytofast enables the visualization and quantification of cell clusters for an efficient discovery of cell populations associated with diseases or physiology. We used cytofast on mass and flow cytometry datasets based on the modulation of the immune system upon immunotherapy. With cytofast, we rapidly generated visual representations of group-related immune cell clusters and showed correlations with the immune system composition. We discovered macrophage subsets that significantly decrease upon cancer immunotherapy and distinct prime-boost effects of prophylactic vaccines on the myeloid compartment. Cytofast is a time-efficient tool for comprehensive cytometric analysis to reveal immune signatures and correlations. Cytofast is available at Bioconductor.

Murine myeloproliferative disorder as a consequence of impaired collaboration between dendritic cells and CD4 T cells

Dendritic cells (DCs) are a key cell type in the initiation of the adaptive immune response. Recently, an additional role for DCs in suppressing myeloproliferation was discovered. Myeloproliferative disorder (MPD) was observed in murine studies with constitutive depletion of DCs, as well as in patients with congenital deficiency in DCs caused by mutations in GATA2 or IRF8 The mechanistic link between DC deficiency and MPD was not predicted through the known biology and has remained an enigma. Prevailing models suggest numerical DC deficiency leads to MPD through compensatory myeloid differentiation. Here, we formally tested whether MPD can also arise through a loss of DC function without numerical deficiency. Using mice whose DCs are deficient in antigen presentation, we find spontaneous MPD that is characterized by splenomegaly, neutrophilia, and extramedullary hematopoiesis, despite normal numbers of DCs. Disease development was dependent on loss of the MHC class II (MHCII) antigen-presenting complex on DCs and was eliminated in mice deficient in total lymphocytes. Mice lacking MHCII and CD4 T cells did not develop disease. Thus, MPD was paradoxically contingent on the presence of CD4 T cells and on a failure of DCs to activate CD4 T cells, trapping the cells in a naive Flt3 ligand-expressing state. These results identify a novel requirement for intercellular collaboration between DCs and CD4 T cells to regulate myeloid differentiation. Our findings support a new conceptual framework of DC biology in preventing MPD in mice and humans.

Machine learning reveals chronic graft-versus-host disease phenotypes and stratifies survival after stem cell transplant for hematologic malignancies

The application of machine learning in medicine has been productive in multiple fields, but has not previously been applied to analyze the complexity of organ involvement by chronic graft-versus-host disease. Chronic graft-versus-host disease is classified by an overall composite score as mild, moderate or severe, which may overlook clinically relevant patterns in organ involvement. Here we applied a novel computational approach to chronic graft-versus-host disease with the goal of identifying phenotypic groups based on the subcomponents of the National Institutes of Health Consensus Criteria. Computational analysis revealed seven distinct groups of patients with contrasting clinical risks. The high-risk group had an inferior overall survival compared to the low-risk group (hazard ratio 2.24; 95% confidence interval: 1.36-3.68), an effect that was independent of graft-versus-host disease severity as measured by the National Institutes of Health criteria. To test clinical applicability, knowledge was translated into a simplified clinical prognostic decision tree. Groups identified by the decision tree also stratified outcomes and closely matched those from the original analysis. Patients in the high- and intermediate-risk decision-tree groups had significantly shorter overall survival than those in the low-risk group (hazard ratio 2.79; 95% confidence interval: 1.58-4.91 and hazard ratio 1.78; 95% confidence interval: 1.06-3.01, respectively). Machine learning and other computational analyses may better reveal biomarkers and stratify risk than the current approach based on cumulative severity. This approach could now be explored in other disease models with complex clinical phenotypes. External validation must be completed prior to clinical application. Ultimately, this approach has the potential to reveal distinct pathophysiological mechanisms that may underlie clusters. Clinicaltrials.gov identifier: NCT00637689.

Multiplexed protein maps link subcellular organization to cellular states

Obtaining highly multiplexed protein measurements across multiple length scales has enormous potential for biomedicine. Here, we measured, by iterative indirect immunofluorescence imaging (4i), 40-plex protein readouts from biological samples at high-throughput from the millimeter to the nanometer scale. This approach simultaneously captures properties apparent at the population, cellular, and subcellular levels, including microenvironment, cell shape, and cell cycle state. It also captures the detailed morphology of organelles, cytoskeletal structures, nuclear subcompartments, and the fate of signaling receptors in thousands of single cells in situ. We used computer vision and systems biology approaches to achieve unsupervised comprehensive quantification of protein subcompartmentalization within various multicellular, cellular, and pharmacological contexts. Thus, highly multiplexed subcellular protein maps can be used to identify functionally relevant single-cell states.

A Structured Tumor-Immune Microenvironment in Triple Negative Breast Cancer Revealed by Multiplexed Ion Beam Imaging

The immune system is critical in modulating cancer progression, but knowledge of immune composition, phenotype, and interactions with tumor is limited. We used multiplexed ion beam imaging by time-of-flight (MIBI-TOF) to simultaneously quantify in situ expression of 36 proteins covering identity, function, and immune regulation at sub-cellular resolution in 41 triple-negative breast cancer patients. Multi-step processing, including deep-learning-based segmentation, revealed variability in the composition of tumor-immune populations across individuals, reconciled by overall immune infiltration and enriched co-occurrence of immune subpopulations and checkpoint expression. Spatial enrichment analysis showed immune mixed and compartmentalized tumors, coinciding with expression of PD1, PD-L1, and IDO in a cell-type- and location-specific manner. Ordered immune structures along the tumor-immune border were associated with compartmentalization and linked to survival. These data demonstrate organization in the tumor-immune microenvironment that is structured in cellular composition, spatial arrangement, and regulatory-protein expression and provide a framework to apply multiplexed imaging to immune oncology.

Automatically generate two-dimensional gating hierarchy from clustered cytometry data

Cytometry is an important technique widely used in medicine and biological research. Biologists traditionally analyze single-cell cytometry data by manual gating, which can be subjective and labor intensive. To address this issue, many automated and semiautomated methods have been developed. These advanced methods are designed to speed up and standardize the analysis of cytometry data, but their popularity is limited by their visualizations which are not intuitive to biologists who are accustomed to the conventional biaxial gating plots. In this article, we present a new method called Cluster-to-Gate (C2G) that can take clustering results as input, and automatically generate a nested two-dimensional gating hierarchy, which is a visualization representation that biologists are familiar with. This method can generate gating sequences for multiple target populations simultaneously and summarize them in one hierarchical tree that represents the gating hierarchy. We have tested this method on target populations defined by manual gating, automated clustering algorithms (k-means for example), and visualization-assisted methods (SPADE and tSNE). We have demonstrated that C2G is able to generate gating sequences that capture cell populations defined by the various clustering strategies, and robust to over-clustered and overlapping target populations. © 2018 International Society for Advancement of Cytometry.

Formation and phenotypic characterization of CD49a, CD49b and CD103 expressing CD8 T cell populations in human metastatic melanoma

Integrins α1β1 (CD49a), α2β1 (CD49b) and αEβ7 (CD103) mediate retention of lymphocytes in peripheral tissues, and their expression is upregulated on tumor infiltrating lymphocytes (TIL) compared to circulating lymphocytes. Little is known about what induces expression of these retention integrins (RI) nor whether RI define subsets in the tumor microenvironment (TME) with a specific phenotype. Human metastatic melanoma-derived CD8 TIL could be grouped into five subpopulations based on RI expression patterns: RI^neg, CD49a⁺ only, CD49a⁺CD49b⁺, CD49a⁺CD103⁺, or positive for all three RI. A significantly larger fraction of the CD49a⁺ only subpopulation expressed multiple effector cytokines, whereas CD49a⁺CD103⁺ and CD49a⁺CD49b⁺ cells expressed IFNγ only. RI^neg and CD49a⁺CD49b⁺CD103⁺ CD8 TIL subsets expressed significantly less effector cytokines overall. Interestingly, however, CD49a⁺CD49b⁺CD103⁺ CD8 expressed lowest CD127, and highest levels of perforin and exhaustion markers PD-1 and Tim3, suggesting selective exhaustion rather than conversion to memory. To gain insight into RI expression induction, normal donor PBMC were cultured with T cell receptor (TCR) stimulation and/or cytokines. TCR stimulation alone induced two RI⁺ cell populations: CD49a single positive and CD49a⁺CD49b⁺ cells. TNFα and IL-2 each were capable of inducing these populations. Addition of TGFβ to TCR stimulation generated two additional populations; CD49a⁺CD49b^negCD103⁺ and CD49a⁺CD49b⁺CD103^+. Taken together, our findings identify opportunities to modulate RI expression in the TME by cytokine therapies and to generate subsets with a specific RI repertoire in the interest of augmenting immune therapies for cancer or for modulating other immune-related diseases such as autoimmune diseases.

Niche signals and transcription factors involved in tissue-resident macrophage development

Tissue-resident macrophages form an essential part of the first line of defense in all tissues of the body. Next to their immunological role, they play an important role in maintaining tissue homeostasis. Recently, it was shown that they are primarily of embryonic origin. During embryogenesis, precursors originating in the yolk sac and fetal liver colonize the embryonal tissues where they develop into mature tissue-resident macrophages. Their development is governed by two distinct sets of transcription factors. First, in the pre-macrophage stage, a core macrophage program is established by lineage-determining transcription factors. Under the influence of tissue-specific signals, this core program is refined by signal-dependent transcription factors. This nurturing by the niche allows the macrophages to perform tissue-specific functions. In the last 15 years, some of these niche signals and transcription factors have been identified. However, detailed insight in the exact mechanism of development is still lacking.

Advancing systems immunology through data-driven statistical analysis

Systems biology provides an effective approach to decipher, predict, and ultimately manipulate the complex and inter-connected networks that regulate the immune system. Advances in high-throughput, multiplexed experimental techniques have increased the availability of proteomic and transcriptomic immunological datasets, and as a result, have also accelerated the development of new data-driven computational algorithms to extract biological insight from these data. This review highlights how data-driven statistical models have been used to characterize immune cell subsets and their functions, to map the signaling and intercellular networks that regulate immune responses, and to connect immune cell states to disease outcomes to generate hypotheses for novel therapeutic strategies. We focus on recent advances in evaluating immune cell responses following viral infection and in the tumor microenvironment, which hold promise for improving vaccines, antiviral and cancer immunotherapy.

MetaCyto: A Tool for Automated Meta-analysis of Mass and Flow Cytometry Data

While meta-analysis has demonstrated increased statistical power and more robust estimations in studies, the application of this commonly accepted methodology to cytometry data has been challenging. Different cytometry studies often involve diverse sets of markers. Moreover, the detected values of the same marker are inconsistent between studies due to different experimental designs and cytometer configurations. As a result, the cell subsets identified by existing auto-gating methods cannot be directly compared across studies. We developed MetaCyto for automated meta-analysis of both flow and mass cytometry (CyTOF) data. By combining clustering methods with a silhouette scanning method, MetaCyto is able to identify commonly labeled cell subsets across studies, thus enabling meta-analysis. Applying MetaCyto across a set of ten heterogeneous cytometry studies totaling 2,926 samples enabled us to identify multiple cell populations exhibiting differences in abundance between demographic groups. Software is released to the public through Bioconductor (http://bioconductor.org/packages/release/bioc/html/MetaCyto.html).

A systematic performance evaluation of clustering methods for single-cell RNA-seq data

Subpopulation identification, usually via some form of unsupervised clustering, is a fundamental step in the analysis of many single-cell RNA-seq data sets. This has motivated the development and application of a broad range of clustering methods, based on various underlying algorithms. Here, we provide a systematic and extensible performance evaluation of 14 clustering algorithms implemented in R, including both methods developed explicitly for scRNA-seq data and more general-purpose methods. The methods were evaluated using nine publicly available scRNA-seq data sets as well as three simulations with varying degree of cluster separability. The same feature selection approaches were used for all methods, allowing us to focus on the investigation of the performance of the clustering algorithms themselves. We evaluated the ability of recovering known subpopulations, the stability and the run time and scalability of the methods. Additionally, we investigated whether the performance could be improved by generating consensus partitions from multiple individual clustering methods. We found substantial differences in the performance, run time and stability between the methods, with SC3 and Seurat showing the most favorable results. Additionally, we found that consensus clustering typically did not improve the performance compared to the best of the combined methods, but that several of the top-performing methods already perform some type of consensus clustering. All the code used for the evaluation is available on GitHub ( https://github.com/markrobinsonuzh/scRNAseq_clustering_comparison). In addition, an R package providing access to data and clustering results, thereby facilitating inclusion of new methods and data sets, is available from Bioconductor ( https://bioconductor.org/packages/DuoClustering2018).

A systematic performance evaluation of clustering methods for single-cell RNA-seq data

Subpopulation identification, usually via some form of unsupervised clustering, is a fundamental step in the analysis of many single-cell RNA-seq data sets. This has motivated the development and application of a broad range of clustering methods, based on various underlying algorithms. Here, we provide a systematic and extensible performance evaluation of 14 clustering algorithms implemented in R, including both methods developed explicitly for scRNA-seq data and more general-purpose methods. The methods were evaluated using nine publicly available scRNA-seq data sets as well as three simulations with varying degree of cluster separability. The same feature selection approaches were used for all methods, allowing us to focus on the investigation of the performance of the clustering algorithms themselves.

We evaluated the ability of recovering known subpopulations, the stability and the run time and scalability of the methods. Additionally, we investigated whether the performance could be improved by generating consensus partitions from multiple individual clustering methods. We found substantial differences in the performance, run time and stability between the methods, with SC3 and Seurat showing the most favorable results. Additionally, we found that consensus clustering typically did not improve the performance compared to the best of the combined methods, but that several of the top-performing methods already perform some type of consensus clustering.

All the code used for the evaluation is available on GitHub ( https://github.com/markrobinsonuzh/scRNAseq_clustering_comparison). In addition, an R package providing access to data and clustering results, thereby facilitating inclusion of new methods and data sets, is available from Bioconductor ( https://bioconductor.org/packages/DuoClustering2018).