A Bayesian gene network reveals insight into the JAK-STAT pathway in systemic lupus erythematosus

Abstract

Systemic lupus erythematosus (SLE) is a chronic, remitting, and relapsing, inflammatory disease involving multiple organs, which exhibits abnormalities of both the innate and adaptive immune responses. A limited number of transcriptomic studies have characterized the gene pathways involved in SLE in an attempt to identify the key pathogenic drivers of the disease. In order to further advance our understanding of the pathogenesis of SLE, we used a novel Bayesian network algorithm to hybridize knowledge- and data-driven methods, and then applied the algorithm to build an SLE gene network using transcriptomic data from 1,760 SLE patients' RNA from the two tabalumab Phase III trials (ILLUMINATE-I & -II), the largest SLE RNA dataset to date. Further, based on the gene network, we carried out hub- and key driver-gene analyses for gene prioritization. Our analyses identified that the JAK-STAT pathway genes, including JAK2, STAT1, and STAT2, played essential roles in SLE pathogenesis, and reaffirmed the recent discovery of pathogenic relevance of JAK-STAT signaling in SLE. Additionally, we showed that other genes, such as IRF1, IRF7, PDIA4, FAM72C, TNFSF10, DHX58, SIGLEC1, and PML, may be also important in SLE and serve as potential therapeutic targets for SLE. In summary, using a hybridized network construction approach, we systematically investigated gene-gene interactions based on their transcriptomic profiles, prioritized genes based on their importance in the network structure, and revealed new insights into SLE activity.

Trial registration: ClinicalTrials.gov NCT01205438 NCT01196091.

Conflict of interest statement

Thomas Dörner has received grant support from Chugai, Janssen, Novartis, and Sanofi. He has received consultancy support from AbbVie, Celgene, Eli Lilly and Company, Janssen, Novartis, Roche, Samsung, and UCB, and speaker bureau fees from Eli Lilly and Company and Roche. Michelle Petri has received consultancy support from Eli Lilly and Company. Daniel Wallace has received consulting support from Amgen, Eli Lilly and Company, EMD Merck Serono, and Pfizer. Yupeng Li, Richard Higgs, Robert Hoffman, Ernst Dow, Xiong Liu, Brian Eastwood, Bradley Miller and Yushi Liu are employees and stockholders of Eli Lilly and Company. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

Figures

Fig 1. Overall analysis workflow.

Fig 2. Algorithm workflow.

Fig 3. Stability and accuracy assessment using the simulation.

The stability (A) is measured by the mean of Hamming distances between pairs of 100 individual networks in 100 runs for each final network. The accuracy (B) is measured by the Hamming distance between the synthetic true network and a predicted network. The simulation was repeated 20 times for each precision setting. Precision represents the accuracy of the prior information, except precision 0, which means no prior information. The error bar is +/-1 standard deviation.

Fig 4. The prior network.

The node represents the gene, the edge is the regulatory relationship between two genes, and the color darkness of the edge corresponds to the edge weight, i.e. the number of documents showing the regulatory relationship. The top 12 hub genes are highlighted.

Fig 5. Reliability cutoff selection and the degree distribution of the final network.

(A) The reliability cutoff selection was based on a scale-free criterion and the cutoff was set to 90, where the scale-free criterion first achieved above 0.8 when increasing the cutoff from 50 to 100. (B) For the degree distribution of the final network, the distribution was log-transformed to show it generally fitted the power-law distribution.

Fig 6. The final network.

The node represents gene and the edge is the regulatory relationship between two genes. The color of a node represents one of the WGCNA modules, Yellow, Green or Tan, and the color darkness of the edge corresponds to the edge weight, i.e. the edge frequency in the 100 runs. The top 12 hub genes are highlighted.

Fig 7. The relationship of node degrees for two networks.

(A) with prior information and (B) without prior information.

Fig 8. The subnetwork of IFR1, IFR7, STAT1, STAT2, JAK2 and their neighbors.

The color of a node represents one of the WGCNA modules, Yellow, Green or Tan.

Fig 9

The first-degree neighbors of (A) FAM72C and (B) PDIA4 genes. The color of a node represents one of the WGCNA modules, Yellow, Green or Tan, and the width of an edge corresponds to the edge weight, i.e. the edge frequency in the 100 runs.