Kidney transplantation is currently a primary treatment for patients with end-stage renal disease, with higher life-quality and less medical resource consumption compared with other renal replacement therapies (1). Despite the dramatic improvement of short-term outcomes over decades, long-term outcomes of kidney transplantation are still remaining an obstacle (2). Acute renal allograft rejection was noted as the leading primary or secondary cause of short-term and long-term graft failure (3). Prompt recognition and aggressive treatment for acute rejection (AR) are essential for preventing the progression of the disease. Except for renal biopsy, other monitoring measures of AR, such as renal function, donor-specific antibodies, and serum immunosuppressive drug level, are much less specific (4). Accurate noninvasive novel biomarkers have high negative predictive value and may substitute biopsy surveillance. Therefore, researchers are now paying attention to exploring ideal noninvasive novel biomarkers related to renal allograft rejection.

Previous studies have detected gene expression profiles through microarray measurements in peripheral blood, urine, and biopsy samples to differentiate rejection patients from non-rejection (5). In 2012, a five-gene set (DUSP1, PBEF1, PSEN1, MAPK9, and NKTR) was locked by qPCR and penalized logistic regression verification and validation in peripheral blood samples (6). Chen et al. identified and validated three cross-organ AR biomarkers (PECAM1, CXCL9, and CD44) by integrating three datasets of pediatric renal transplantation, adult renal transplantation and heart transplantation (7). Bioinformatic analyses of the updated datasets could aid in discovering novel candidate genes and provide deeper understandings of the molecular mechanisms of acute renal transplant rejection.

The development of new computational techniques remains a rapidly expanding area. Weighted gene co-expression network analysis (WGCNA) is a systems biology method to transform the gene expression data into highly relevant modules and investigate genes in the selected modules, which is beyond traditional differential expression detection (8). CIBERSORT, a deconvolution algorithm, could calculate the levels and examine the roles of immune cells in the immune microenvironment of the target disease (9). Currently, both methods have been applied effectively in various diseases to identify hub genes or therapeutic agents (10, 11), while studies using WGCNA or immune infiltration to search for the potential biomarkers in renal transplant AR are still scarce.

In this study, differentially expressed genes (DEG) between AR and non-rejection (NR) samples were identified from five gene expression omnibus (GEO) microarray datasets (GSE1563, GSE25902, GSE36059, GSE98320, and GSE174020). We conducted the WGCNA analysis, and two up-regulated modules, medium purple 3 and dark orange, and one down-regulated module, midnight blue, were chosen as rejection-related modules, as well as immune cell infiltration and protein-protein interaction (PPI) network analyses to screen PDCD1, a hub gene involved in kidney transplant rejection. Flow cytometry analyses with an exhaustion-related phenotype, PD1+CD57-, validated the identified hub gene.

In recent years, acute allograft rejection rates in kidney transplant recipients have dramatically improved with the evolution of immunosuppressive therapy (17). However, prompt differential diagnosis is still essential as acute rejection could rapidly and progressively affect graft function whenever it occurs (18). Identification of biomarkers might provide new insights for early diagnosis or targeted treatment of AR. The current study investigated the KEGG and GO enrichment and the immune cell infiltration of five AR-related GEO datasets from renal biopsy samples, to deeper understand the molecular involvement in AR. We identified 7194 AR-related genes based on WGCNA analysis. By intersection with a specific KEGG pathway and constructing a PPI network, PDCD1, which was then validated with flow-cytometry analysis, was selected as a potential diagnostic marker.

The results of GO biological process terms indicated that the DEGs were involved in T cell activation, mononuclear cell differentiation, regulation of cell-cell adhesion, and lymphocyte differentiation, consistent with the previous enrichment results of TCMR by Zhou et.al (19). The KEGG results mainly enriched the DEGs with multiple autoimmune diseases and various T cell-related pathways, involving differentiation of T helper cells and the T cell receptor signaling pathway. T helper cells, including Th1, Th2, Th17 and follicular helper T cells, act as important immune factors and regulate effective cell pathways in the acute rejection progress (20). The immune infiltration analysis also proved the association of acute rejection with multiple subtypes of T cells, including follicular helper T cells, CD8⁺ T cells, memory-activated CD4⁺ T cells, and gamma delta T cells. Other immune cells, such as B cells, dendritic cells, Neutrophils, Macrophages, etc., were also involved in acute rejection by comparison with non-rejection samples. The participation of these lymphocyte cells was not demonstrated in the enrichment results of the rejection-related DEGs.

In the acute rejection episodes, recognition of the foreign antigen or allograft by T cell receptors generates and mediates alloimmune response by recruiting inflammatory cells or antibodies and destroying the transplant renal function (21–23). Effective T cell activation by T cell receptors would lead to clonal expansion of specific phenotypes of T cells, which can be detected by clinical testing measures (24). Indirect activated CD4⁺ T cells are also capable and unique helpers to alloreactive B cells in the alloantibody-mediated response (25). KEGG is a manually annotated metabolic pathway database that contains highly-ordered functional information on the complex interrelationships between the genes and their products (13). Compared with GO, KEGG is more interactive and more closely linked to the biological progress in reality. For the above reasons, we selected a KEGG enriched pathway, the T cell receptor signaling pathway (hsa04660), from the DEGs-enriched pathways or gene sets for further hub gene selection.

The co-stimulatory and co-inhibitory receptors are have a pivotal role in the T cell receptor signaling, as they modulate the throughout T cell biology, including activation, subset differentiation, effector function, memory formation, and survival (26). Immune checkpoint inhibitors targeting co-stimulatory signals in anti-cancer settings have long been heavily adopted forms of immunosuppressants (27). Knowledge of costimulatory blockade agents in transplantation tolerance covers not only initial T cell activation but is also gradually extending to multiple steps in alloantibody production (28). Leibler C et al. suggested that Belatacept, a recombined CTLA4-Ig, could inhibit the Tfh (PD1⁺ICOS⁺)–B-cell crosstalk in humans and decrease differentiation of B cells into plasmablasts (29). Its direct effects on Ig production by activated B cells are independent of T-cell intervention (28). Therefore, identification of a pivotal biomarker in the T cell receptor pathway, especially a co-stimulatory or co-inhibitory receptor by bioinformatics methods, might provide clues for developing new immunotherapy effective in both T cell- and antibody-related rejections.

PDCD1, which encodes programmed cell death protein 1 (CD279, PD1), is a transmembrane receptor protein in CD28/CTLA-4 subfamily and is broadly expressed in various hematopoietic cells and other tissues (30). There have been great concerns about PD1-related pathways in the induction and maintenance of immune tolerance in cancer, autoimmune diseases, and chronic infection (30). Antibodies blocking the PD1/PDL1 axis have been developed to protect T cells from exhausted status in anti-cancer immunity (31). However, previous studies of PD1 expression in allograft rejection showed conflicting results. Studies by Wang and Afaneh C et al. reported higher levels of PD1 mRNA before renal transplantation and at rejection in both blood and urine samples, which were also associated with disease prognosis (32, 33). While another study by Carmona-Escamilla et al. showed that PD1 was downregulated in CD4⁺ T cells in acute rejection renal recipients (34). PD1 expressions on CD4 and CD8 T cells were also significantly decreased in acute rejection recipients in a previous liver transplantation research (35). Anti-PD1/PDL1 treatments for malignant or metastatic tumors were reported associating with immune checkpoint inhibitors (ICI)-associated transplant rejection. Most of them were acute cellular rejections, and a small proportion was also presented as antibody-mediated rejections (36). The association between PD1 and acute rejection in kidney transplantation might be because of its negative regulations of multiple T cell subsets and humoral immunity (30). Its wide expression in diverse immune cells and tissues might also bring higher diagnostic values and new therapeutic insights, although no agonists of PD1 have yet been available for graft acceptance.

In our flow-cytometry results, acute renal transplant rejection recipients present a significant decrease of the PD1⁺CD57^- phenotype in both CD4⁺ and CD8⁺ T cells. PD1⁺CD57^- has been recognized as an exhausted T cell phenotype, which refers to reduced capacity to secrete cytokines and expression of inhibitory receptors (37). Clinical and experimental evidence have supported the protective role of exhausted lymphocyte cells in organ transplantation (38). A previous study by Fribourg M et al. showed that increased percentages of PD1⁺CD57^- in CD4⁺ and CD8⁺ T cells significantly improved posttransplant renal function and were consistent with the acute rejection incidence (39). We may speculate that monitoring the phenotypic markers, such as PD1⁺CD57^-, and tapering T-cell depleting induction therapy could effectively reduce acute rejection incidence and immune-related adverse outcomes.

There are several limitations in the present study. Firstly, the sample size for validation was relatively small and restricted to a single center. Secondly, we know that there are two typical classifications of acute rejection, T cell-mediated rejection, and antibody-mediated rejection. We did not respectively investigate each classification. Thirdly, although the PD1⁺CD57^- phenotype typically indicates T cell exhaustion, the correlation between CD57 and acute rejection, and interaction between PD1 and CD57 still need to be confirmed.

In conclusion, this study performed comprehensive bioinformatic methods, including WGCNA and immune infiltration analysis. PDCD1 was identified as a diagnostic biomarker in acute renal rejection. The flow-cytometry validation results of the PD1⁺CD57^- phenotype are consistent with our hypothesis and suggest the potential role of the PDCD1 gene and the PD1⁺CD57^- phenotype in allograft protection. Future research may focus on investigating specific functions of PD1 and the PD1⁺CD57^- phenotype in diverse T cell subtypes, such as effector T cell, regulatory T cell, or follicular T cell, as well as B cells, to figure out the underlying mechanisms and to develop new strategies for preserving protective immune responses.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

The studies involving human participants were reviewed and approved by Clinical research committee of the First Affiliated Hospital, Zhejiang University School of Medicine. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

YW, XLin: review of public databases, flow cytometry, writing the manuscript. CW: flow cytometry. XLiu, XW, YQ, YC, QZ, HZ: collect clinical samples, overall review of results, data and statistical analysis. JC, HH: supervise the study, funding acquisition, review the manuscript. All authors agree to be accountable for the content of the work.