Christof Seiler

Random Spatial Structure of Geometric Deformations and Bayesian Nonparametrics

Our work is motivated by the geometric study of lower back pain from patient CT images. In this paper, we take a first step towards that goal by introducing a data-driven way of identifying anatomical regions of interest. We propose a probabilistic model of the geometrical variability and describe individual patients as noisy deformations of a random spatial structure (modeled as regions) from a common template. The random regions are generated using the distance dependent Chinese Restaurant Process. We employ the Gibbs sampler to infer regions from a set of noisy deformation fields. Each step of the sampler involves model selection (Bayes factor) to decide about fusing regions. In the discussion, we highlight connections between image registration and Markov chain Monte Carlo methods.

CD38 is a key regulator of enhanced NK cell immune responses during pregnancy through its role in immune synapse formation

Natural killer (NK) cells use a diverse array of activating and inhibitory surface receptors to detect threats and provide an early line of defense against viral infections and cancer. Here, we demonstrate that the cell surface protein CD38 is a key human NK cell functional receptor through a role in immune synapse formation. CD38 expression marks a mature subset of human NK cells with a high functional capacity. NK cells expressing high levels of CD38 display enhanced killing and IFN-γ secretion in response to influenza virus-infected and tumor cells. Inhibition of CD38 enzymatic activity does not influence NK cell function, but blockade of CD38 and its ligand CD31 abrogates killing and IFN-γ expression in response to influenza-infected cells. Blockade of CD38 on NK cells similarly inhibits killing of tumor cells. CD38 localizes and accumulates at the immune synapse between NK cells and their targets, and blocking CD38 severely abrogates the ability of NK cells to form conjugates and immune synapses with target cells. Thus, CD38 plays a critical role in NK cell immune synapse formation. These findings open new avenues in immunotherapeutic development for cancer and infection by revealing a critical role for CD38 in NK cell function.Once sentence summary CD38 is responsible for the enhanced immune responses of NK cells to influenza virus infection during pregnancy through immune synapse formation. Abstract Pregnant women are particularly susceptible to complications of influenza A virus infection, which may result from pregnancy-induced changes in the function of immune cells, including natural killer (NK) cells. To decipher mechanisms driving enhanced NK cell activity during pregnancy, we profiled NK cells from pregnant and non-pregnant women, which showed significantly increased CD38 expression during pregnancy. CD38 expression defines a phenotypically distinct and mature subset of NK cells that display increased ability to secrete IFN-γ and to kill influenza-infected and tumor cells. This enhanced function is based on the ability of CD38 to promote the formation of the NK cell immune synapse. Thus, increased CD38 expression directly promotes enhanced NK cell responses during pregnancy through its role in immune synapse formation. These findings open new avenues in immunotherapeutic development for cancer and viruses by revealing a critical role for CD38 in the formation of the NK cell immune synapse.

Multi-Table Differential Correlation Analysis of Neuroanatomical and Cognitive Interactions in Turner Syndrome

Girls and women with Turner syndrome (TS) have a completely or partially missing X chromosome. Extensive studies on the impact of TS on neuroanatomy and cognition have been conducted. The integration of neuroanatomical and cognitive information into one consistent analysis through multi-table methods is difficult and most standard tests are underpowered. We propose a new two-sample testing procedure that compares associations between two tables in two groups. The procedure combines multi-table methods with permutation tests. In particular, we construct cluster size test statistics that incorporate spatial dependencies. We apply our new procedure to a newly collected dataset comprising of structural brain scans and cognitive test scores from girls with TS and healthy control participants (age and sex matched). We measure neuroanatomy with Tensor-Based Morphometry (TBM) and cognitive function with Wechsler IQ and NEuroPSYchological tests (NEPSY-II). We compare our multi-table testing procedure to a single-table analysis. Our new procedure reports differential correlations between two voxel clusters and a wide range of cognitive tests whereas the single-table analysis reports no differences. Our findings are consistent with the hypothesis that girls with TS have a different brain-cognition association structure than healthy controls.

Differential Induction of IFN-α and Modulation of CD112 and CD54 Expression Govern the Magnitude of NK Cell IFN-γ Response to Influenza A Viruses

In human and murine studies, IFN-γ is a critical mediator immunity to influenza. IFN-γ production is critical for viral clearance and the development of adaptive immune responses, yet excessive production of IFN-γ and other cytokines as part of a cytokine storm is associated with poor outcomes of influenza infection in humans. As NK cells are the main population of lung innate immune cells capable of producing IFN-γ early in infection, we set out to identify the drivers of the human NK cell IFN-γ response to influenza A viruses. We found that influenza triggers NK cells to secrete IFN-γ in the absence of T cells and in a manner dependent upon signaling from both cytokines and receptor–ligand interactions. Further, we discovered that the pandemic A/California/07/2009 (H1N1) strain elicits a seven-fold greater IFN-γ response than other strains tested, including a seasonal A/Victoria/361/2011 (H3N2) strain. These differential responses were independent of memory NK cells. Instead, we discovered that the A/Victoria/361/2011 influenza strain suppresses the NK cell IFN-γ response by downregulating NK-activating ligands CD112 and CD54 and by repressing the type I IFN response in a viral replication–dependent manner. In contrast, the A/California/07/2009 strain fails to repress the type I IFN response or to downregulate CD54 and CD112 to the same extent, which leads to the enhanced NK cell IFN-γ response. Our results indicate that influenza implements a strain-specific mechanism governing NK cell production of IFN-γ and identifies a previously unrecognized influenza innate immune evasion strategy.

Strain-Specific Human Natural Killer Cell Recognition of Influenza A Virus

Abstract Innate Natural killer (NK) cells employ an array of surface receptors to detect ‘altered self’ induced by infection or malignancy. Despite their decisive role in early antiviral immunity, the cellular mechanisms governing if or how they discriminate between viral infections remain unresolved. Here, we demonstrate that while human NK cells are capable of reducing infection levels of distinct influenza A strains, the A/California/07/2009 (pH1N1) strain induces a significantly more robust IFN-γ response than A/Victoria/361/2011 (H3N2) and all other strains tested. This surprising degree of strain specificity results in part from the inability of the pH1N1 strain to downregulate the activating ligands CD112 (Nectin-2) and CD54 (ICAM-1) as efficiently as the H3N2 strain, leading to enhanced NK cell detection and IFN-γ secretion. A network analysis of differentially expressed transcripts identifies the interferon α/β receptor (IFNAR) pathway as an additional, critical determinant of this strain-specific response. Strain-specific downregulation of NK cell activating ligands and modulation of type I IFN production represents a previously unrecognized influenza immunoevasion tactic and could present new opportunities to modulate the quality and quantity of the innate antiviral response for therapeutic benefit. One Sentence Summary Human natural killer cells distinguish between Influenza A strains using a combinatorial cytokine priming and receptor-ligand signaling mechanism.

Discussion of "Geodesic Monte Carlo on Embedded Manifolds"

Contributed discussion and rejoinder to Geodesic Monte Carlo on Embedded Manifolds (arXiv:1301.6064)

Bayesian Statistics in Computational Anatomy

Computational anatomy is the science of anatomical shape examined by deforming a template organ into a subject organ. It compares and contrasts organ shapes to inspire personalized treatments or find group differences in case-control studies. Independently of the transformation model used, the task of finding deformations between organs is a statistical task concerned with estimating parameters. Recently it has become important to go beyond “best” estimates and quantify the variability of estimates. The variability is caused by noise in the image, model misspecification, or sampling variability in an observational study. Bayesian statistics provides a rigorous framework to build models that can quantify uncertainty. In this book chapter, we will review some of the basics of Bayesian statistics and relate it to our own experience in applying Bayesian ideas in computational anatomy. We will divide the presentation into two parts. First, we formulate image registration using parametric Bayesian statistics and elaborate on some of the practical difficulties that we encountered in our own work. Second, we will give an example of nonparametric Bayesian statistics applied to clustering of deformation fields into parcels of contiguous voxels.

Multivariate Heteroscedasticity Models for Functional Brain Connectivity

Functional brain connectivity is the co-occurrence of brain activity in different areas during resting and while doing tasks. The data of interest are multivariate timeseries measured simultaneously across brain parcels using resting-state fMRI (rfMRI). We analyze functional connectivity using two heteroscedasticity models. Our first model is low-dimensional and scales linearly in the number of brain parcels. Our second model scales quadratically. We apply both models to data from the Human Connectome Project (HCP) comparing connectivity between short and conventional sleepers. We find stronger functional connectivity in short than conventional sleepers in brain areas consistent with previous findings. This might be due to subjects falling asleep in the scanner. Consequently, we recommend the inclusion of average sleep duration as a covariate to remove unwanted variation in rfMRI studies. A power analysis using the HCP data shows that a sample size of 40 detects 50% of the connectivity at a false discovery rate of 20%. We provide implementations using R and the probabilistic programming language Stan.