Functional mass cytometry for reclassification and precise diagnosis of systemic autoimmune diseases

Resumen

Systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), systemic sclerosis (SSC), Sjögren’s syndrome (SJS), mixed connective tissue disease (MCTD) and primary antiphospholipid syndrome (PAPS) are classified as systemic autoimmune diseases (SADs). These diseases are characterized by signs of autoimmunity that include the production of autoantibodies and the damage of different organs. Although having separated clinical definitions and clinical diagnostic criteria, these diseases are difficult to diagnose differentially, as patients have highly overlapping symptoms and varied clinical signs, particularly at early disease stages. This overlapping clinical landscape impedes the correct diagnosis and early drug administration. While molecular resemblance between SADs was suspected for a long time, the lack of well described, shared biomarkers make treatment and diagnosis difficult. Therefore, molecular and cellular-based studies need to be undertaken to classify the patients based on the physiopathological mechanism underlying the diseases in a strategy of personalized medicine. In order to study the complexity of the immune system at the single cell level, proper technologies need to be used. Mass cytometry (Cytometry by Time-Of-Flight, CyTOF, MC) is a high-dimensional technique that allows to measure more than 50 markers in one single cell. Thus, it is a good tool to perform deep-phenotyping studies tracking several cell types or levels of cellular activation markers. However, in order to observe patient-specific cellular patterns, significant amounts of individuals need to be recruited, involving often different research centers. Hence a proper experimental design needs to be established. Whole blood preservation seems to be an attractive way to gather samples from centers located far away from MC-core facility, yet not many blood-preservation protocols were validated for MC so far. Additionally, because samples acquired through the CyTOF instrument suffer from cell clogging, signal drop associated to long acquisition and batch effects, special care needs to be taken when analyzing the data when multiple groups of samples are studied. Thus, a data analysis pipeline that considers data normalization, quality control and the high-dimensional nature of MC data needs to be used. Additionally, in order to analyze hundreds of samples the analysis pipeline needs to be adapted for large-scale studies and ideally be automatized as much as possible. However up to now, no such workflow was developed. In this PhD thesis we studied 7 different SADs in order to find new biomarkers that allow for patient reclassification according to immune cell signatures. We aimed at performing a deep phenotyping study including functional markers relevant for SADs. As we wanted to have the most complete picture of the immune system, we decided to collect whole blood samples and use MC cytometry to analyze them. In order to do this, we collected blood samples in different centers located in Granada and Córdoba. Thus, we had to establish a cryopreservation protocol suitable for multicenter studies. As in total more than one hundred samples were collected, we established also an experimental protocol minimizing experimental variation, and a quality control and analysis pipeline was also optimized together with automatized data preprocessing. Using these settings, we have demonstrated that high-content immunophenotyping studies can be successfully performed with small amounts of fixed/frozen blood. Immediate whole blood fixation benefits from shorter manipulation times, hence preventing cell death specially in the neutrophil compartment. We designed an experimental workflow that limits experimental variation and reported an R-based data curation workflow that cleans collected data and corrects the batch effects introduced during the sample preparation and staining. This pipeline is semi-automated and optimized for large studies involving human blood phenotyping, together with functional markers. Finally, we showed that MC can be successfully used to detect groups (clusters) of patients having similar immune landscapes, supporting the personalized medicine development in SADs. So far, we constructed a patient reclassification framework using cell frequencies and expression levels of functional markers. The four detected clusters differed in the frequency and activation state of both myeloid and lymphoid cells. Additionally, they were also characterized by different levels of pro and anti-inflammatory cytokines. Each cluster contained a mixture of different diseases, confirming the high heterogeneity of each diagnosis label.