| 초록 |
Objectives: Increasing evidence indicates that the contribution of innate immunity to renal inflammation plays a substantial role in the development of diabetic kidney disease (DKD). In this work, we attempted to apply a multi-omics approach with machine learning-based techniques to identify immune-derived new therapeutic targets in DKD. Methods: In total, we randomly partitioned the independent cohorts into a training set (control: n = 41, DKD: n = 49) and an external validation set (control: n = 20, DKD: n = 41). We attempted Lasso, support vector machine, random forest, supervised machine learning methods. Single-nucleus sequencing data were processed with Seurat v4.0.3. MPC5 cells were treated with the medium containing either normal glucose (5.5 mM), high glucose (30 mM), or high mannitol (30 mM mannitol) for the indicated time intervals. The primer sequences are listed in Table 1. Results: The optimal prediction model with the highest average C-index (0.8737), and the TYRO3 gene was identified. The high expression levels of TYRO3 subgroup had a significantly lower immune score (P < 0.001) (Fig. E). The proportion of M1/M2 macrophages was significantly higher in patients with DKD than in control groups, as confirmed by scRNA-seq results (Fig. G). There was a significantly positive association between TYRO3 expression and eGFR (R = 0.498, P = 0.001) (Fig. E). As shown in Fig. I, compared with the normal glucose (5.5 mM) group, the increased expression of Tyro3 mRNA in MPC5 cultured with high glucose (30 mM) was obviously time-dependent. Conclusions: We propose an exploratory approach for inflammatory risk assessment and stratification in DKD. Our results reveal that TYRO3 could be a potential target for regulating immune and inflammatory responses in DKD. Further studies will be required to gain a better understanding of the mechanisms of TYRO3 in the development of renal injury in DKD. |