Background Cellular immune responses are phenotypically and functionally perturbed in patients with lupus nephritis (LN), leading to severe renal tissue inflammation. Although multiple gene expression landscapes have been identified for LN development, longitudinal cell type-specific immune responses and prognostic signatures during treatment of LN remain largely unknown.

Methods To uncover transcriptome changes during treatment and identify immune markers to predict treatment response, we performed sequential single-cell RNA sequencing using peripheral blood mononuclear cells (PBMCs) obtained from patients with biopsy-proven LN who received mycophenolate mofetil in combination with glucocorticoids. Single-cell libraries were generated using a commercially available droplet method, the Chromium System from 10× Genomics, Inc. (Pleasanton, CA, USA).

Results We profiled ~239,000 PBMCs from 10 female patients with LN. After receiving standard therapy for 1 year, number of individuals with complete response (CR) and non-response (NR) according to ACR response criteria was 5 each. Peripheral blood B cells in patients with NR showed a significant expansion of double-negative switched memory cells (DN2), a distinct subset expressing T-box transcription factor T-bet, at renal flare and the increased proportion was maintained after immunomodulatory treatment. Next, we directly compared myeloid cells between CR and NR groups. Analysis of differentially expressed genes revealed that NR was characterized by up-regulation of various interferon-stimulated genes (ISGs), including IFITM1, IFI44, and ISG15. Both IFN-α and IFN-γ responsive genes were enriched. Patients with CR were accompanied by repression of inflammatory pathways across all types of myeloid cells, with noticeable down-regulation of unphosphorylated IFN-stimulated gene factor 3 (U-ISGF3)-inducible signatures.

Conclusions We provide the first evidence of comparative transcriptional signatures depending on the treatment response in patients with LN. Our results highlight that detailed analysis on immune cells and a dissection of type I IFN-driven inflammatory features enhance the understanding of treatment response dynamics, which might guide the selection of optimal therapeutics for LN.