• arthritis
• rheumatoid
• inflammation
• autoimmune diseases

We appreciate the interest and comments shown by Wang et al about our study in their letter.1 2 We would like to clarify and discuss some issues Wang et al indicated.

First, matching design is generally used in case–control studies to eliminate the interference of potential confounding factors. Ideally, the subjects are perfectly matched in case–control (1:R) groups. In our research design stage, we also planned the ratio (1:1) of case–control that is commonly used in genetic and/or epigenetic studies. However, several steps of quality control were adopted in DNA/RNA preparation and microarray experiment, and some subjects were excluded, which led to the mismatched numbers, even though the sample ratio was originally designed as 1:1. The mismatched sample probably introduced some confounding effects, but such effects would be excluded to the greatest extent through our multiomics integration, replication in independent samples and in-depth functional validation. Actually, it is frequently observed that the numbers of cases and controls were mismatched in the genome-wide profiling study.3

Second, it is a common strategy that the identified DNA methylation sites and messenger RNAs (mRNAs) from microarray assays are validated using different technical methods (herein, bisulfate sequencing for DNA methylation and reverse transcription PCR for mRNA). Our research strategy is to integrate multiomics expression profiling (methylome and transcriptome) by using high-throughput microarray analysis and then to technically and biologically validate the significant findings in additional study samples with larger sample size. Although the three assays had different levels of sensitivity, we believe that the consistent results from different methodology and biological validation at DNA methylation level and mRNA expression level would be reasonable to warrant their significance for rheumatoid arthritis (RA), especially for the significance of poly (ADP-ribose) polymerase family member 9 (PARP9).

Third, the PARP9 gene was the most interesting gene after exploratory analysis and a series of confirmatory analyses.1 PARP9 was affirmed to be causative among the regulatory chains of DNA methylation–mRNA–RA and highlighted in interaction networks constructed by the differentially methylated genes/differentially expressed genes. Among five validated methylation sites, three (cg00959259, cg08122652 and cg22930808) were located in the PARP9 gene. The significant correlations between methylation levels in PARP9 and gene expression were verified in peripheral blood mononuclear cells and Jurkat T cells, as well as in primary T cells. The above evidence taken together could justify the priority of PARP9.

Fourth, the study also used Jurkat cells as cell models to investigate the functional effect of the PARP9 gene. Based on current evidence, we cannot conclude that the effect is specific. As shown in the Discussion section, we have discussed that some of our findings were also reported in other autoimmune diseases,4–6 suggesting that the identified sites in our study may serve as common sites shared by other autoimmune diseases. Further research would be needed to explore whether the RA-related methylation sites identified in the present study are unique to RA or common to other autoimmune diseases including systemic lupus erythematous.

Last, as shown in the Results section, a significant correlation between the methylation level (cg00959259) and PARP9 gene expression (r=0.752, p=0.019) was detected in the active RA cases. This experiment is to investigate whether the detected methylation sites have regulation effects on mRNA expression of PARP9 in the patients with RA with active disease status. Such patient selection strategy is consistent with those in the discovery and replication stages. The original purpose of this study is to find the abnormal methylation sites between active RA cases and healthy controls. Therefore, all the patients with RA in our study were recruited according to their active disease status but not limited to newly detected and newly diagnosed cases.