TrackSOM: Mapping immune response dynamics through clustering of time-course cytometry data

Mapping the dynamics of immune cell populations over time or disease-course is key to understanding immunopathogenesis and devising putative interventions. We present TrackSOM, a novel method for delineating cellular populations and tracking their development over a time- or disease-course cytometry datasets. We demonstrate TrackSOM-enabled elucidation of the immune response to West Nile Virus infection in mice, uncovering heterogeneous subpopulations of immune cells and relating their functional evolution to disease severity. TrackSOM is easy to use, encompasses few parameters, is quick to execute, and enables an integrative and dynamic overview of the immune system kinetics that underlie disease progression and/or resolution.

Peripheral B-cell dysregulation is associated with relapse after long-term quiescence in patients with multiple sclerosis

B cells play a major role in multiple sclerosis (MS), with many successful therapeutics capable of removing them from circulation. One such therapy, alemtuzumab, is thought to reset the immune system without the need for ongoing therapy in a proportion of patients. The exact cells contributing to disease pathogenesis and quiescence remain to be identified. We utilized mass cytometry to analyze B cells from the blood of patients with relapse-remitting MS (RRMS) before and after alemtuzumab treatment, and during relapse. A complementary RRMS cohort was analyzed by single-cell RNA sequencing. The R package "Spectre" was used to analyze these data, incorporating FlowSOM clustering, sparse partial least squares-discriminant analysis and permutational multivariate analysis of variance. Immunoglobulin (Ig)A+ and IgG1 + B-cell numbers were altered, including higher IgG1 + B cells during relapse. B-cell linker protein (BLNK), CD40 and CD210 expression by B cells was lower in patients with RRMS compared with non-MS controls, with similar results at the transcriptomic level. Finally, alemtuzumab restored BLNK, CD40 and CD210 expression by IgA+ and IgG1 + B cells, which was altered again during relapse. These data suggest that impairment of IgA+ and IgG1 + B cells may contribute to MS pathogenesis, which can be restored by alemtuzumab.

Detection and characterization of chemotaxis without cell tracking

Swarming has been observed in various biological systems from collective animal movements to immune cells. In the cellular context, swarming is driven by the secretion of chemotactic factors. Despite the critical role of chemotactic swarming, few methods to robustly identify and quantify this phenomenon exist. Here, we present a novel method for the analysis of time series of positional data generated from realizations of agent-based processes. We convert the positional data for each individual time point to a function measuring agent aggregation around a given area of interest, hence generating a functional time series. The functional time series, and a more easily visualized swarming metric of agent aggregation derived from these functions, provide useful information regarding the evolution of the underlying process over time. We extend our method to build upon the modelling of collective motility using drift-diffusion partial differential equations (PDEs). Using a functional linear model, we are able to use the functional time series to estimate the drift and diffusivity terms associated with the underlying PDE. By producing an accurate estimate for the drift coefficient, we can infer the strength and range of attraction or repulsion exerted on agents, as in chemotaxis. Our approach relies solely on using agent positional data. The spatial distribution of diffusing chemokines is not required, nor do individual agents need to be tracked over time. We demonstrate our approach using random walk simulations of chemotaxis and experiments investigating cytotoxic T cells interacting with tumouroids.

Integrated immune dynamics define correlates of COVID-19 severity and antibody responses

SARS-CoV-2 causes a spectrum of COVID-19 disease, the immunological basis of which remains ill defined. We analyzed 85 SARS-CoV-2-infected individuals at acute and/or convalescent time points, up to 102 days after symptom onset, quantifying 184 immunological parameters. Acute COVID-19 presented with high levels of IL-6, IL-18, and IL-10 and broad activation marked by the upregulation of CD38 on innate and adaptive lymphocytes and myeloid cells. Importantly, activated CXCR3⁺cT_FH1 cells in acute COVID-19 significantly correlate with and predict antibody levels and their avidity at convalescence as well as acute neutralization activity. Strikingly, intensive care unit (ICU) patients with severe COVID-19 display higher levels of soluble IL-6, IL-6R, and IL-18, and hyperactivation of innate, adaptive, and myeloid compartments than patients with moderate disease. Our analyses provide a comprehensive map of longitudinal immunological responses in COVID-19 patients and integrate key cellular pathways of complex immune networks underpinning severe COVID-19, providing important insights into potential biomarkers and immunotherapies.

Analyses of 184 immune features define kinetics of immune responses to SARS-CoV-2

Circulating T_FH1 cells in acute COVID-19 correlate with antibodies

sIL-6R levels are elevated in severe COVID-19 but do not correlate with IL-6

Elevated IL-6 and IL-18 correlate with immune cell hyperactivation

Using single-cell cytometry to illustrate integrated multi-perspective evaluation of clustering algorithms using Pareto fronts

MOTIVATION

Many 'automated gating' algorithms now exist to cluster cytometry and single cell sequencing data into discrete populations. Comparative algorithm evaluations on benchmark datasets rely either on a single performance metric, or a few metrics considered independently of one another. However, single metrics emphasise different aspects of clustering performance and do not rank clustering solutions in the same order. This underlies the lack of consensus between comparative studies regarding optimal clustering algorithms and undermines the translatability of results onto other non-benchmark datasets.

RESULTS

We propose the Pareto fronts framework as an integrative evaluation protocol, wherein individual metrics are instead leveraged as complementary perspectives. Judged superior are algorithms that provide the best trade-off between the multiple metrics considered simultaneously. This yields a more comprehensive and complete view of clustering performance. Moreover, by broadly and systematically sampling algorithm parameter values using the Latin Hypercube sampling method, our evaluation protocol minimises (un)fortunate parameter value selections as confounding factors. Furthermore, it reveals how meticulously each algorithm must be tuned in order to obtain good results, vital knowledge for users with novel data. We exemplify the protocol by conducting a comparative study between three clustering algorithms (ChronoClust, FlowSOM and Phenograph) using four common performance metrics applied across four cytometry benchmark datasets. To our knowledge, this is the first time Pareto fronts have been used to evaluate the performance of clustering algorithms in any application domain.

AVAILABILITY

Implementation of our Pareto front methodology and all scripts to reproduce this article are available at https://github.com/ghar1821/ParetoBench.

Enabling rational gut microbiome manipulations by understanding gut ecology through experimentally-evidenced in silico models

The gut microbiome has emerged as a contributing factor in non-communicable disease, rendering it a target of health-promoting interventions. Yet current understanding of the host-microbiome dynamic is insufficient to predict the variation in intervention outcomes across individuals. We explore the mechanisms that underpin the gut bacterial ecosystem and highlight how a more complete understanding of this ecology will enable improved intervention outcomes. This ecology varies within the gut over space and time. Interventions disrupt these processes, with cascading consequences throughout the ecosystem. In vivo studies cannot isolate and probe these processes at the required spatiotemporal resolutions, and in vitro studies lack the representative complexity required. However, we highlight that, together, both approaches can inform in silico models that integrate cellular-level dynamics, can extrapolate to explain bacterial community outcomes, permit experimentation and observation over ecological processes at high spatiotemporal resolution, and can serve as predictive platforms on which to prototype interventions. Thus, it is a concerted integration of these techniques that will enable rational targeted manipulations of the gut ecosystem.

Cytotoxic T cells swarm by homotypic chemokine signalling

Cytotoxic T lymphocytes (CTLs) are thought to arrive at target sites either via random search or following signals by other leukocytes. Here, we reveal independent emergent behaviour in CTL populations attacking tumour masses. Primary murine CTLs coordinate their migration in a process reminiscent of the swarming observed in neutrophils. CTLs engaging cognate targets accelerate the recruitment of distant T cells through long-range homotypic signalling, in part mediated via the diffusion of chemokines CCL3 and CCL4. Newly arriving CTLs augment the chemotactic signal, further accelerating mass recruitment in a positive feedback loop. Activated effector human T cells and chimeric antigen receptor (CAR) T cells similarly employ intra-population signalling to drive rapid convergence. Thus, CTLs recognising a cognate target can induce a localised mass response by amplifying the direct recruitment of additional T cells independently of other leukocytes.

Fecal microbiota transplantation from high caloric-fed donors alters glucose metabolism in recipient mice, independently of adiposity or exercise status

Studies suggest the gut microbiota contributes to the development of obesity and metabolic syndrome. Exercise alters microbiota composition and diversity and is protective of these maladies. We tested whether the protective metabolic effects of exercise are mediated through fecal components through assessment of body composition and metabolism in recipients of fecal microbiota transplantation (FMT) from exercise-trained (ET) mice fed normal or high-energy diets. Donor C57BL/6J mice were fed a chow or high-fat, high-sucrose diet (HFHS) for 4 wk to induce obesity and glucose intolerance. Mice were divided into sedentary (Sed) or ET groups (6 wk treadmill-based ET) while maintaining their diets, resulting in four donor groups: chow sedentary (NC-Sed) or ET (NC-ET) and HFHS sedentary (HFHS-Sed) or ET (HFHS-ET). Chow-fed recipient mice were gavaged with feces from the respective donor groups weekly, creating four groups (NC-Sed-R, NC-ET-R, HFHS-Sed-R, HFHS-ET-R), and body composition and metabolism were assessed. The HFHS diet led to glucose intolerance and obesity in the donors, whereas exercise training (ET) restrained adiposity and improved glucose tolerance. No donor group FMT altered recipient body composition. Despite unaltered adiposity, glucose levels were disrupted when challenged in mice receiving feces from HFHS-fed donors, irrespective of donor-ET status, with a decrease in insulin-stimulated glucose clearance into white adipose tissue and large intestine and specific changes in the recipient's microbiota composition observed. FMT can transmit HFHS-induced disrupted glucose metabolism to recipient mice independently of any change in adiposity. However, the protective metabolic effect of ET on glucose metabolism is not mediated through fecal factors.

Strategies for calibrating models of biology

Computational and mathematical modelling has become a valuable tool for investigating biological systems. Modelling enables prediction of how biological components interact to deliver system-level properties and extrapolation of biological system performance to contexts and experimental conditions where this is unknown. A model's value hinges on knowing that it faithfully represents the biology under the contexts of use, or clearly ascertaining otherwise and thus motivating further model refinement. These qualities are evaluated through calibration, typically formulated as identifying model parameter values that align model and biological behaviours as measured through a metric applied to both. Calibration is critical to modelling but is often underappreciated. A failure to appropriately calibrate risks unrepresentative models that generate erroneous insights. Here, we review a suite of strategies to more rigorously challenge a model's representation of a biological system. All are motivated by features of biological systems, and illustrative examples are drawn from the modelling literature. We examine the calibration of a model against distributions of biological behaviours or outcomes, not only average values. We argue for calibration even where model parameter values are experimentally ascertained. We explore how single metrics can be non-distinguishing for complex systems, with multiple-component dynamic and interaction configurations giving rise to the same metric output. Under these conditions, calibration is insufficiently constraining and the model non-identifiable: multiple solutions to the calibration problem exist. We draw an analogy to curve fitting and argue that calibrating a biological model against a single experiment or context is akin to curve fitting against a single data point. Though useful for communicating model results, we explore how metrics that quantify heavily emergent properties may not be suitable for use in calibration. Lastly, we consider the role of sensitivity and uncertainty analysis in calibration and the interpretation of model results. Our goal in this manuscript is to encourage a deeper consideration of calibration, and how to increase its capacity to either deliver faithful models or demonstrate them otherwise.

Memory B cells are reactivated in subcapsular proliferative foci of lymph nodes

Vaccine-induced immunity depends on the generation of memory B cells (MBC). However, where and how MBCs are reactivated to make neutralising antibodies remain unknown. Here we show that MBCs are prepositioned in a subcapsular niche in lymph nodes where, upon reactivation by antigen, they rapidly proliferate and differentiate into antibody-secreting plasma cells in the subcapsular proliferative foci (SPF). This novel structure is enriched for signals provided by T follicular helper cells and antigen-presenting subcapsular sinus macrophages. Compared with contemporaneous secondary germinal centres, SPF have distinct single-cell molecular signature, cell migration pattern and plasma cell output. Moreover, SPF are found both in human and mouse lymph nodes, suggesting that they are conserved throughout mammalian evolution. Our data thus reveal that SPF is a seat of immunological memory that may be exploited to rapidly mobilise secondary antibody responses and improve vaccine efficacy.

Automated multi-objective calibration of biological agent-based simulations

Computational agent-based simulation (ABS) is increasingly used to complement laboratory techniques in advancing our understanding of biological systems. Calibration, the identification of parameter values that align simulation with biological behaviours, becomes challenging as increasingly complex biological domains are simulated. Complex domains cannot be characterized by single metrics alone, rendering simulation calibration a fundamentally multi-metric optimization problem that typical calibration techniques cannot handle. Yet calibration is an essential activity in simulation-based science; the baseline calibration forms a control for subsequent experimentation and hence is fundamental in the interpretation of results. Here, we develop and showcase a method, built around multi-objective optimization, for calibrating ABSs against complex target behaviours requiring several metrics (termed objectives) to characterize. Multi-objective calibration (MOC) delivers those sets of parameter values representing optimal trade-offs in simulation performance against each metric, in the form of a Pareto front. We use MOC to calibrate a well-understood immunological simulation against both established a priori and previously unestablished target behaviours. Furthermore, we show that simulation-borne conclusions are broadly, but not entirely, robust to adopting baseline parameter values from different extremes of the Pareto front, highlighting the importance of MOC's identification of numerous calibration solutions. We devise a method for detecting overfitting in a multi-objective context, not previously possible, used to save computational effort by terminating MOC when no improved solutions will be found. MOC can significantly impact biological simulation, adding rigour to and speeding up an otherwise time-consuming calibration process and highlighting inappropriate biological capture by simulations that cannot be well calibrated. As such, it produces more accurate simulations that generate more informative biological predictions.

Towards an Integrative Understanding of Diet-Host-Gut Microbiome Interactions

Over the last 20 years, a sizeable body of research has linked the microbiome and host diet to a remarkable diversity of diseases. Yet, unifying principles of microbiome assembly or function, at levels required to rationally manipulate a specific individual’s microbiome to their benefit, have not emerged. A key driver of both community composition and activity is the host diet, but diet–microbiome interactions cannot be characterized without consideration of host–diet interactions such as appetite and digestion. This becomes even more complex if health outcomes are to be explored, as microbes engage in multiple interactions and feedback pathways with the host. Here, we review these interactions and set forth the need to build conceptual models of the diet–microbiome–host axes that draw out the key principles governing this system’s dynamics. We highlight how “units of response,” characterizations of similarly behaving microbes, do not correlate consistently with microbial sequence relatedness, raising a challenge for relating high-throughput data sets to conceptual models. Furthermore, they are question-specific; responses to resource environment may be captured at higher taxonomic levels, but capturing microbial products that depend on networks of different interacting populations, such as short-chain fatty acid production through anaerobic fermentation, can require consideration of the entire community. We posit that integrative approaches to teasing apart diet–microbe–host interactions will help bridge between experimental data sets and conceptual models and will be of value in formulating predictive models.

Diet-Microbiome Interactions in Health Are Controlled by Intestinal Nitrogen Source Constraints

Diet influences health and patterns of disease in populations. How different diets do this and why outcomes of diets vary between individuals are complex and involve interaction with the gut microbiome. A major challenge for predicting health outcomes of the host-microbiome dynamic is reconciling the effects of different aspects of diet (food composition or intake rate) on the system. Here we show that microbial community assembly is fundamentally shaped by a dichotomy in bacterial strategies to access nitrogen in the gut environment. Consequently, the pattern of dietary protein intake constrains the host-microbiome dynamic in ways that are common to a very broad range of diet manipulation strategies. These insights offer a mechanism for the impact of high protein intake on metabolic health and form the basis for a general theory of the impact of different diet strategies on host-microbiome outcomes.

Leukocyte Motility Models Assessed through Simulation and Multi-objective Optimization-Based Model Selection

The advent of two-photon microscopy now reveals unprecedented, detailed spatio-temporal data on cellular motility and interactions in vivo. Understanding cellular motility patterns is key to gaining insight into the development and possible manipulation of the immune response. Computational simulation has become an established technique for understanding immune processes and evaluating hypotheses in the context of experimental data, and there is clear scope to integrate microscopy-informed motility dynamics. However, determining which motility model best reflects in vivo motility is non-trivial: 3D motility is an intricate process requiring several metrics to characterize. This complicates model selection and parameterization, which must be performed against several metrics simultaneously. Here we evaluate Brownian motion, Lévy walk and several correlated random walks (CRWs) against the motility dynamics of neutrophils and lymph node T cells under inflammatory conditions by simultaneously considering cellular translational and turn speeds, and meandering indices. Heterogeneous cells exhibiting a continuum of inherent translational speeds and directionalities comprise both datasets, a feature significantly improving capture of in vivo motility when simulated as a CRW. Furthermore, translational and turn speeds are inversely correlated, and the corresponding CRW simulation again improves capture of our in vivo data, albeit to a lesser extent. In contrast, Brownian motion poorly reflects our data. Lévy walk is competitive in capturing some aspects of neutrophil motility, but T cell directional persistence only, therein highlighting the importance of evaluating models against several motility metrics simultaneously. This we achieve through novel application of multi-objective optimization, wherein each model is independently implemented and then parameterized to identify optimal trade-offs in performance against each metric. The resultant Pareto fronts of optimal solutions are directly contrasted to identify models best capturing in vivo dynamics, a technique that can aid model selection more generally. Our technique robustly determines our cell populations’ motility strategies, and paves the way for simulations that incorporate accurate immune cell motility dynamics.

Advances in imaging technology allow investigators to monitor the movements and interactions of immune cells in a live animal, processes essential to understanding and manipulating how an immune response is generated. T cells in the brains of Toxoplasma gondii-infected mice have previously been described as performing a Lévy walk, an optimal strategy for locating sparsely, randomly distributed targets. Determining which motility model best characterizes a population of cells is problematic; multiple metrics are required to specify as intricate and nuanced a process as 3D motility, and the tools to evaluate model-parameter combinations have been lacking. We have developed a novel framework to perform this model evaluation through simulation, a popular tool for exploring complex immune system phenomena. We find that Lévy walk offers an inferior capture of our data to another class of motility model, the correlated random walk, and this determination was possible because we are able to explicitly evaluate several motility metrics simultaneously. Further, we find evidence that leukocytes differ in their inherent translational and rotational speeds. These findings facilitate more accurate immune system simulations aimed at unravelling the processes underpinning this critical biological function.