Allogeneic kidney transplantation is the preferred treatment option for patients with end-stage renal disease, as it can significantly improve overall survival and quality of life [32]. However, graft rejection is a significant risk factor for long-term graft loss [5]. Clinically, allogeneic transplantation often leads to TCMR, significantly elevating the risk of later chronic rejection, worse outcomes, and graft loss [8]. Currently, histopathological changes in graft tissue are commonly used to diagnose TCMR. This method is subject to subjective interpretation among pathologists, has nonspecific lesions, and relies on arbitrary rules, which can make the diagnosis less reliable than expected [33]. By evaluating molecular changes in tissue that precede morphological changes, a new diagnostic method could overcome the limitations of current histopathological diagnosis. A growing body of literature suggests that abnormal changes in glycans are associated with the T cell immune response during graft rejection [20, 21, 34]. Based on this, we obtained renal graft rejection-related microarray expression data from GEO and explored the role of GTGs in renal graft rejection at a molecular level. DE-GTGs were identified to establish a TCMR diagnostic model for distinguishing TCMR from other types of graft injury. Furthermore, we used the DE-GTGs to construct a prognostic model for predicting long-term renal graft outcomes.
Glycosylation, a common post-translational modification of proteins, plays a crucial role in regulating fundamental processes such as cell division, differentiation, immune response, and cell-to-cell interaction [35]. In this study, we identified 15 DE-GTGs by integrating DEGs and GTGs. Functional enrichment analysis showed that the glycosphingolipid biosynthesis biological process played a critical role in the progression of graft rejection. Furthermore, the expression levels of most of the DE-GTGs showed significant differences between the TCMR group and nonTCMR group, suggesting that DE-GTGs may play an important part in rejection, especially in TCMR. We then attempted to construct a TCMR diagnostic model to distinguish TCMR from other type of graft injury. By integrating LASSO and XGboost machine learning methods, we obtained four DE-GTGs related to TCMR: ST3GAL5, GCNT1, CSGALNACT2, and B4GALT1. It is worth noting that the four hub DE-GTGs are closely linked to T cell-mediated immune responses. ST3GAL5 was positively correlated with the infiltration of CD8+ T cells, which was involved in tumor immunoregulation in ccRCC [19]. Notch signaling plays a critical role in regulating T cell development, activation, and function [22]. GCNT1 modify Notch receptors and regulate Notch signaling [36, 37], and they are required for optimal in vitro stimulation of T cells [38]. CSGALNACT2 co-localized with macrophages inside the plaque during atherosclerosis development, enabling oxidized low-density lipoprotein binding on the macrophage surface and increasing macrophage foam cell formation and mice atherosclerosis [39]. Nilius et al reported that a high expression level of B4GALT1 in T-lymphocytes, but not in monocytes, was associated with a lower risk of relapse with a hazard ratio (HR) of 0.66 (95% confidence interval (CI) of HR: 0.45–0.97; p = 0.02) upon multivariate Cox regression analysis [40]. In our study, we found that ST3GAL5, GCNT1, and CSGALNACT2 had positive correlation with TCMR, while B4GALT1 had the negative correlation with TCMR. The TCMR diagnostic model was constructed by performing Logistics regression based on the four hub DE-GTGs. To our delight, the TCMR diagnostic model not only showed excellent diagnostic values in the train cohort (the AUC was 0.833 in the GSE36059 dataset), but also presented good diagnostic values in the two validation cohorts (the AUC was 0.792 and 0.831 in the GSE48581 and GSE72925 datasets, respectively). The immune infiltration analysis showed that the high-risk group presented more immune cells infiltration than low-risk group. It is reported that effector T cells, dendritic cells and activated macrophages are the main acting cells in the pathogenesis of TCMR [41], which are also significantly associated with the four TCMR-related hub DE-GTGs in the present study.
The GSVA-enrichment plot showed that the inflammatory response, graft rejection, IL6-JAK-STAT3-signaling, TNFA-signaling-via-NFKB, and IL2-STAT5 signaling were significantly activated in the high-risk group. As we all know, when T cells are stimulated during graft rejection, cytokines such as IL6, TNFA and IL2 are secreted [42]. IL6, when bound to its receptor, can initiate a cascade through JAK activation, and activated JAK kinase phosphorylates STAT3, inducing dimerization, leading to inflammation [43]. TNFA can activate NFKB, further causing a magnified inflammatory reaction [44]. During an immune response, activated CD4+ and CD8+ T lymphocytes secrete large quantities of IL2. The duration of the IL2-STAT5 signal controls the expansion of antigen-specific CD8+ T cell populations [45]. It is well known that changes in glycan structure are well documented in association with inflammatory response [46]. For example, the loss of the β1,6-GlcNAc branch on N-glycans through mutation of the Mgat5 locus leads to hyperresponsive T cells and autoimmunity [47], while loss of complex N-glycans through ablation of Mgat2 leads to a loss of commensal-derived polysaccharide antigen presentation to regulatory T cells [48]. Based on the evidence above, we propose that when graft rejection occurs, a large number of cytokines are secreted, which activates various inflammation related signaling pathways, resulting enhanced glycosylation. This leads to an increasing number of immune cells, mainly including T cells, migrating and infiltrating to the graft.
The occurrence of graft rejection after transplantation is closely related to the occurrence of long-term graft failure. In order to effectively predict the survival of long-term graft, we further explored the potential of DE-GTGs for predicting graft survival time after a biopsy. Through univariable Cox and LASSO COX regression, five DE-GTGs (ST3GAL5, GCNT1, POGLUT1, GALNT1, and POMT1) were selected to construct a long-term graft survival predictive model. It has been reported that GALNT1 plays an important role in the process of leukocyte attachment and rolling. GALNT1 initiates core-type protein O-glycosylation, and the loss of GALNT1 leads to reduced leukocyte rolling and recruitment, as well as increased rolling velocity, indicating a predominant role a significant role for GALNT1 in attaching functionally relevant O-linked glycans to selectin ligands [49]. Moreover, some investigations have demonstrated that POMT1 is crucial for E-cadherin-mediated cell adhesion [50] and plays an important role in muscular dystrophy-dystroglycanopathies [51]. Previous research has shown that POGLUT1 regulates myeloid and lymphoid cell development through Notch signaling [52]. Thus, these five DE-GTGs play important roles in the immune response. In terms of transplanted kidneys, persistent and intense immune response is the main culprit for graft loss. The time-dependent ROC curves showed the AUCs (0.76, 0.81, and 0.70) of predictive model for 1–3 years of survival in the train cohort, indicating the predictive model showed excellent predicted value. Furthermore, K-M analysis revealed significant differences in survival time between high- and low-risk groups, indicating that outcomes worsened in the high-risk group as time increased. Therefore, the GTGs had shown their potential in predicting the risk of graft loss, which is a prerequisite for treatment to avoid further deterioration in those symptomatic patients.
In summary, our findings demonstrated that GTGs could be used be used for diagnosing TCMR. In addition, GTGs can also be applied to predict long-term graft outcomes. These results offer novel schemes for assessing kidney transplant diseases and provide a valuable direction for future research.