Session L11: Immune Sensing and Response I

Live

Paradoxically, the pro-inflammatory cytokine Tumor Necrosis Factor α (TNFα) simultaneously activates opposing pro-apoptotic and pro-survival signaling. We show that the activation of antagonistic pathways changes the properties of cell fate decision transitioning cells from a “slow and accurate” to “fast and error-prone” decision mode. Mathematical modeling predicts, and experiments in vitro and in vivo validate, that the regulation of the decision mode of non-immune cells by innate immune cell production of TNFα is key to prevent viral spread. Overall our results demonstrate how a collective phenotype emerges from the regulation of fundamental tradeoffs within cellular cognitive processes.   

Live

B cells are a crucial player in the adaptive immune system as antibodies of high affinity are essential to swiftly eradicate pathogens. A specific response is made possible by processes of recombination and subsequent mutation of the immunoglobulin-coding genes that ensure an extraordinary diversity of the B cell repertoire. Thanks to hypermutations, affinity maturation converges promptly to deliver antibodies of desired specificity. We use high-throughput repertoire sequencing data to reconstruct B cell lineages from non-productive sequences and model the inherent biases of the mutation process. We describe preferential targeting of motifs and regions along the sequence and observe that mutations occurring concomitantly along B-cell lineages tend to co-localise, suggesting a possible mechanism for accelerating affinity maturation

Reference:
Natanael Spisak, Aleksandra M Walczak, Thierry Mora, Learning the heterogeneous hypermutation landscape of immunoglobulins from high-throughput repertoire data, Nucleic Acids Research, gkaa825, doi.org/10.1093/nar/gkaa825   

Live

Adaptive immunity's success relies on the extraordinary diversity of protein receptors on B cell membranes. Recent progress in deep sequencing methods has been followed by the development of probabilistic models characterizing repertoire sequence distributions. Here we present a statistical approach defined in terms of a probabilistic V(D)J recombination model enhanced by a selection factor that describes repertoire diversity and that predicts with high accuracy its level of publicness, i.e. the number of sequences that will be shared between any number of individuals. While the model shows perfect agreement with healthy repertoires it clearly underestimates sharing between repertoires affected by a common antigen. This deviation is a sign of a stronger antigen driven selection that opens the possibility of finding antigen-specific antibodies.   

Live

T cells exhibit remarkable sensitivity and selectivity in detecting and responding to agonist peptides (p) bound to MHC molecules in a sea of self pMHC molecules. Despite much work, understanding of the underlying mechanisms of distinguishing such ligands remains incomplete. Here, we quantify T cell discriminatory capacity using channel capacity, a direct measure of the signaling network’s ability to discriminate between antigen-presenting cells (APCs) displaying either self ligands or a mixture of self and agonist ligands. This metric shows how differences in information content between these two types of peptidomes are decoded by network topology, feedback loops, and rates of kinetic proofreading signaling steps inside T cells. Using channel capacity, we constructed numerically substantiated hypotheses to explain the discriminatory role of a recently identified slow LAT Y132 phosphorylation step. Biochemical and imaging experiments support these findings.   

Live

Adaptive immune systems neutralize pathogens using diverse receptors on B-cells (BCR). Their neutralization power is increased during a Darwinian maturation process in each individual, where random sequence modifications are selected. We show that these modification include large scale deletions and insertions (indels) of non-templated genomic material. We describe the statistics of these hyperindels in healthy and HIV-infected people, where they seem to play an important role. We introduce a statistical model for the occurrence of these indels, inferring the parameters of such a model from real data through maximum-likelihood approaches.   

Live

CRISPR-Cas has become a ubiquitous genetic editing tool for developing cell lines, disease models, and gene therapies, yet there is still more to be understood about the specificity and potential off-target activity of this tool. Here, a stochastic model is developed to study the targeting activity of the endogenous CRISPR-Cas immune system in bacteria against viruses. Based on the energy landscape of the recognition reaction, the significance of two biological modules is explored: (1) a short DNA motif called the PAM and (2) an RNA guide that matches a roughly 30-bp DNA sequence in the target genome. The model results indicate the further division of the RNA guide into two modules based on differences in the sensitivity to DNA mismatches, which is in line with experimental observations of a "seed" region. The modularity of the CRISPR-Cas immune system appears to manage the tradeoff among having enough specificity to avoid self-targeting, having enough cross-reactivity to recognize mutated virus DNA, and launching a fast enough immune response to protect against an unfolding infection.   

Live

Some bacteria and archaea possess an adaptive immune system that maintains a memory of past infections in viral DNA elements called spacers stored in the CRISPR loci of their genomes. This memory is used to mount targeted responses against threats. However, the cross-reactivity of CRISPR interference and primed spacer acquisition suggests that incorporation of foreign DNA can also lead to auto-immunity. We show that balancing viral defense against auto-immunity predicts a scaling law relating spacer length and CRISPR repertoire size. By analyzing a database of microbial genomes, we find that the predicted scaling law is realized empirically across prokaryotes, partly through proportionate use of different CRISPR types in strains carrying multiple loci. We show that strains with nonfunctional CRISPR loci do not show this scaling, and demonstrate that simple population-level selection mechanisms can generate the trend, along with observed variations between strains of a given species.   

Live

We consider a stochastic bistable two-species generalized Lotka-Volterra model of the microbiome and use it as a testbed to analytically and numerically explore the effects of direct (e.g. fecal microbiota transplantation) and indirect (e.g. changes in diet) bacteriotherapies. Two types of noise are included in this model, representing the immigration of bacteria into and within the gut (additive noise) and variations in growth rate associated with the spatially inhomogeneous distribution of resources (multiplicative noise). The efficacy of a bacteriotherapy is determined by comparing the mean first-passage times (the average time required for the system to transition from one basin of attraction to the other) with and without the intervention. We use concepts from transition path theory to investigate how the role of noise affects these bacteriotherapies, and probe the relationship between the deterministic and stochastic systems by comparing isocommittor surfaces of the stochastic system to the separatrix of the deterministic system.   

Live

The adaptive immune system is a very diverse ecosystem built to respond to pathogens by selecting clones of cells with specific receptors. While clonal expansion in response to a particular antigen has been studied, we still do not know the neutral dynamics that drive the immune system to develop in absence of strong pathogenic threats. We investigate the forces shaping T-cells clonal dynamics in healthy individuals.
Using a model that has shown good consistency with stationary data and longitudinal data from different Repertoire Sequencing (RepSeq) techniques, we use Bayesian Inference to learn parameters underlying the healthy T-cells population dynamics for individuals of various ages and both sexes. Quantifying the experimental noise accurately for each RepSeq technique enables us to robustly infer the dynamics parameters and extract biological information from it. We find that the majority of clones follow a first order non-Markovian dynamic and the inferred turn-over time-scales of the repertoire seem to be strongly correlated with the age of the patients. As an application of our method we developed a classifier that discriminates different individuals based on their immune samples and show the discrimination task is stable over time.   

Live

Storing memory for molecular recognition is an efficient strategy for sensing and responding to external stimuli. Biological processes use different strategies to store memory. For example, the olfactory system uses general receptors that can bind to a large number of natural molecular mixtures, while the immune memory is encoded by a multitude of specialized receptors to target diverse antigens. To probe the emergence of distinct memory strategies, we propose a dynamical Hebbian network model that can efficiently learn and store memory against evolving patterns. Specifically, we demonstrate that specialized memory emerges as an optimal strategy against evolving patterns, whereas a general distributed memory can be used to recognize static patterns. Our results shed light on the distinct encoding of memory in the olfactory and the immune system, where the former uses general receptors against static molecular mixtures and the latter uses specialized receptors against evolving pathogens.   

Live

Subclasses of immune B and T-cells have different functional roles to work together to produce an immune response and lasting immunity.
Additionally to these functional roles, T and B-cell lymphocytes rely on the diversity of their receptor chains to recognize different pathogens.
The lymphocyte subclasses emerge from common ancestors generated with the same diversity of receptors during selection processes.
I will show how to leverage biophysical models of receptor generation with machine learning models of selection to identify specific sequence features characteristic of functional lymphocyte repertoires and subrepertoires.
Specifically using only repertoire level sequence information, we classify CD4 and CD8 T-cells, find correlations between receptor chains arising during selection and identify T-cells subsets that are targets of pathogenic epitopes.
I also show examples of when linear classifiers do as well as deep methods.