Predicting susceptibility to tuberculosis based on gene expression profiling in dendritic cells

www.nature.com/scientificreports OPEN Received: 13 February 2017 Accepted: 5 June 2017 Published: xx xx xxxx Predicting susceptibility to tuberculosis based on gene expression profiling in dendritic cells John D. Blischak 1,2, Ludovic Tailleux3, Marsha Myrthil1, Cécile Charlois4, Emmanuel Bergot5, Aurélien Dinh6, Gloria Morizot7, Olivia Chény8, Cassandre Von Platen8, Jean-Louis Herrmann 9,10, Roland Brosch 3, Luis B. Barreiro11,12 & Yoav Gilad1,13 Tuberculosis (TB) is a deadly infectious disease, which kills millions of people every year. The causative pathogen, Mycobacterium tuberculosis (MTB), is estimated to have infected up to a third of the world’s population; however, only approximately 10% of infected healthy individuals progress to active TB. Despite evidence for heritability, it is not currently possible to predict who may develop TB. To explore approaches to classify susceptibility to TB, we infected with MTB dendritic cells (DCs) from putatively resistant individuals diagnosed with latent TB, and from susceptible individuals that had recovered from active TB. We measured gene expression levels in infected and non-infected cells and found hundreds of differentially expressed genes between susceptible and resistant individuals in the non-infected cells. We further found that genetic polymorphisms nearby the differentially expressed genes between susceptible and resistant individuals are more likely to be associated with TB susceptibility in published GWAS data. Lastly, we trained a classifier based on the gene expression levels in the non-infected cells, and demonstrated reasonable performance on our data and an independent data set. Overall, our promising results from this small study suggest that training a classifier on a larger cohort may enable us to accurately predict TB susceptibility. Tuberculosis (TB) is a major public health issue. Worldwide, over a million people die of TB annually, and millions more currently live with the disease1–3. Successful treatment requires months of antibiotic therapy4, and drug-resistant strains of Mycobacterium tuberculosis (MTB) continuously emerge5. Approximately a third of the world’s population is estimated to be infected with MTB, but most are asymptomatic. While these naturally resistant individuals are able to avoid active disease, MTB might persist in a dormant state, known as latent TB6. In contrast, approximately 10% of individuals will develop active TB after infection with MTB7, 8. Unfortunately, we are currently unable to predict if an individual is susceptible. While twin and family studies have indicated a heritable component of TB susceptibility9–12, genome wide association studies (GWAS) have only identified a few loci with low effect size13–19. Due to the highly polygenic architecture, it may be informative to examine differences between susceptible and resistant individuals at a higher level of organization, e.g. gene regulatory networks. 1 Department of Human Genetics, University of Chicago, Chicago, Illinois, USA. 2Committee on Genetics, Genomics, and Systems Biology, University of Chicago, Chicago, Illinois, USA. 3Integrated Mycobacterial Pathogenomics, Institut Pasteur, Paris, France. 4Centre de Lutte Antituberculeuse de Paris, DASES Mairie de Paris, 75013, Paris, France. 5Service de pneumologie et oncologie thoracique, CHU Côte de Nacre, 14033, Caen, France. 6Maladies Infectieuses, AP-HP, Hôpital Universitaire Raymond-Poincaré, Garches, 92380, France. 7Clinical Investigation & Access Biological Resources (ICAReB), Institut Pasteur, Paris, France. 8Clinical Core, Centre for Translational Science, Institut Pasteur, Paris, France. 9INSERM, U1173, UFR Simone Veil, Université de Versailles Saint Quentin, Saint Quentin en Yvelines, France. 10APHP, Groupe Hospitalo-Universitaire Paris Île-de-France Ouest, Garches et Boulogne-Billancourt, France. 11Department of Genetics, CHU Sainte-Justine Research Center, Montreal, Québec, Canada. 12Department of Pediatrics, University of Montreal, Montreal, Québec, Canada. 13Department of Medicine, University of Chicago, Chicago, Illinois, USA. John D. Blischak and Ludovic Tailleux contributed equally to this work. Correspondence and requests for materials should be addressed to L.T. (email: ) or L.B.B. (email: ) or Y.G. (email: ) Scientific Reports | 7: 5702 | DOI:10.1038/s41598-017-05878-w 1 www.nature.com/scientificreports/ Using this approach, previous studies have characterized gene expression profiles in innate immune cells isolated from individuals known to be susceptible or resistant to infectious diseases, including those with latent or active TB20 and acute rheumatic fever21. We hypothesized that gene expression profiles in innate immune cells may be used to classify individuals with respect to their susceptibility to develop active TB. To test this hypothesis, we differentiated dendritic cells (DCs) from monocytes isolated from individuals that had recovered from a past episode of active TB, which we refer to as susceptible, and from individuals with confirmed latent TB, which we refer to as putatively resistant (this group is enriched in resistant individuals but we cannot exclude that some still have the potential to develop active TB22). We infected the DCs with MTB because these innate immune cells help shape the adaptive immune response, which is critical for fighting MTB23, 24. We discovered that the gene expression differences between innate immune cells from resistant and susceptible individuals were present primarily in the non-infected state, that these differentially expressed genes were enriched for nearby SNPs with low p-values in TB susceptibility GWAS, and furthermore, that these gene expression levels could be used to classify individuals based on their susceptibility status. Results Susceptible individuals have an altered transcriptome in the non-infected state. We obtained whole blood samples from 25 healthy male Caucasian individuals (Supplementary Data S1). Six of the donors had recovered from active TB, and are thus putatively susceptible. The remaining 19 tested positive for latent TB without ever experiencing symptoms of active TB, and are thus putatively resistant. We isolated dendritic cells (DCs) and treated them with Mycobacterium tuberculosis (MTB) or a mock control for 18 hours. To measure genome-wide gene expression levels in infected and non-infected samples, we isolated and sequenced RNA using a processing pipeline designed to minimize the introduction of unwanted technical variation (Supplementary Fig. S1). We obtained a mean (±SEM) of 48 ± 6 million raw reads per sample. We performed quality control analyses to remove non-expressed genes (Supplementary Fig. S2; Supplementary Data S2), identify and remove outliers (Supplementary Figs S3, S4 and S5), and check for confounding batch effects (Supplementary Figs S6 and S7). Ultimately, data from 6 of the 50 samples (25 individuals × 2 treatments) failed the quality chec (...truncated)