Chronic kidney disease (CKD) is a highly heterogeneous disease with multiple etiologies, risk factors, and outcomes, that is commonly defined as the persistence of structural or functional abnormalities of the kidney for greater than 3 months that severely affect health. CKD has been recognized as one of the world’s growing public health problems. According to the Global Burden of Kidney Disease (2017), there are 697.5 million CKD cases worldwide, almost a third of which are in China and India, and the total number of individuals receiving kidney replacement therapy is over 2.5 million and is expected to double by 2030, making it a major health and economic challenge across regions (Bikbov et al., 2020). The prevalence of CKD is still increasing, and clinical outcomes are unsatisfactory. Very often, the burden of CKD is not restricted to the impact on the need for renal replacement therapy after progression to end-stage renal disease (ESRD). In fact, patients with CKD tend to have a significantly increased risk of cardiovascular disease and are at higher risk of hospitalization and death (Webster et al., 2017). And since there are no symptoms in the early stages, as many as nine out of ten adults with CKD are unaware of their kidney damage, and CKD acts as a “silent killer”. Despite the high burden and impact caused by the disease, available therapy options to stop or slow the progression of CKD continue to be limited. This may be attributed to the fact that the underlying mechanisms of CKD pathogenesis and progression are not fully understood. Therefore, in-depth study of CKD-related pathologies and understanding of its underlying molecular mechanisms are important for early diagnosis of the disease and providing new therapeutic targets.

Influencing the rate of progression in CKD patients is usually associated with genes polymorphisms of inflammatory response, fibrosis, metabolism, CKD progression, and renin-angiotensin-aldosterone system (RAAS) (Yan et al., 2021). Notably, there is mounting evidence that renal immune dysregulation plays a critical role in the physiological dysfunction and development associated with CKD (Espi et al., 2020; Tang et al., 2021). Regardless of the underlying etiology, the inflammatory response is continuously activated in CKD, characterized by the recruitment of immune cell throughout the kidney, along with local overproduction of growth factors (e.g., tumor necrosis factor (TNF)-α, interleukins, interferon (IFN)-γ, chemokines) and pro-fibrotic cytokines (Suárez-Fueyo et al., 2017; Cantero-Navarro et al., 2021). Under this regulation, different cellular and molecular processes are activated, leading to persistent kidney injury and disease progression in CKD. Therefore, exploring the correlation between immune-related genes and CKD progression may help to elucidate the pathogenesis of CKD patients at the molecular level. However, little is known about the precise regulatory genetic and molecular mechanisms of immune involvement in the pathogenesis and progression of CKD.

Thanks to the rise and advancement of bioinformatics analysis and high-throughput sequencing, new approaches will contribute to a better understand the biology of kidney disease through the use of genetic, epigenetic modification and transcriptomic studies, providing new clues and opening up new unknown avenues for the study of disease mechanisms (Pareek et al., 2011; O’seaghdha and Fox, 2011). Several bioinformatics studies have made significant advances in the identification of diagnostic biomarkers for CKD (Ahmed et al., 2022; Wang et al., 2022). However, purely assessing the differences between CKD cases and normal controls is far from meeting the growing need for its complex pathological features, risk stratification for progressive end-stage renal disease, and high-risk clinical outcomes. Early characterization of this heterogeneity is an important step in developing an individualized follow-up strategy for patients with CKD (Zheng et al., 2021). Several studies have explored the importance of dysregulated signaling events in various renal and environmental cells, suggesting that immune cell infiltration patterns have predictive power for clinical guidance (Şenbabaoğlu et al., 2016). A comprehensive and integrated exploration of immune subtypes based on multiple transcriptome expression profiles may be a nice addition to the mechanisms of CKD progression (Li et al., 2022).

Therefore, in this study, we systematically examined the differentially expressed IRGs and immune profiles of renal tissues among normal and CKD individuals for the first time. The biological features, enrichment pathways and immune profiles among the two subtypes were further investigated by dividing 440 CKD patients into two immune-related subgroups through the expression profiles of 15 characterized IRGs. Predictive models disclosing patients with diverse molecular clusters were developed by combining several machine learning algorithms. Finally, we also investigated the correlation between the diagnostic value of prospective markers and clinical characteristics. We hope that a feature-based classification model for IRGs will reveal important molecular mechanisms underlying the role of immunity in CKD and provide for improved diagnosis, management and treatment of CKD.

Chronic kidney disease affects 11.7%–15.1% of the global population and is one of the world’s most serious health problems (Lv and Zhang, 2019). Regardless of the potential cause, CKD progresses gently and contributes to irreversible nephron loss, end-stage renal disease, and increased mortality. The differential phenotypes of CKD are not always well known in clinical practice due to the presence of marked heterogeneity in pathogenic processes, risk factors, pathological changes, and outcomes, and thus their overall treatment outcomes are extremely limited. Previous studies have emphasized the development of tools for characterizing this potential heterogeneity with different phenotypic data (Zheng et al., 2021), but no in-depth analysis of CKD subgroup-specific biomarker data has been performed. The addition of distinguishing CKD clusters at the molecular level will undoubtedly increase our comprehension of CKD heterogeneity and is important for guiding individualized treatment of CKD. Furthermore, despite significant efforts to explore new targets for CKD, the current knowledge appears to be inadequate and there is still an urgent need for potential biomarkers with high specificity and sensitivity.

Immunomodulation has been extensively studied due to its close relationship with CKD. As previously mentioned, CKD is a chronic and confusing process that affects a variety of normal metabolic functions and kinase-mediated pathways. Chronic inflammation underlies one of the major pathogenic mechanisms of chronic kidney disease, a process that can be characterized as a complicated interaction among renal cells and resident immune cells, such as dendritic cells and macrophages, and the recruitment of circulating monocytes, neutrophils and lymphocytes (Andrade-Oliveira et al., 2019). With further exploration of the mechanisms of monocytes (Heine et al., 2012), macrophages (Guiteras et al., 2016), mast cells (Holdsworth and Summers, 2008), T-cell and dendritic cells (Weisheit et al., 2015) in CKD revealed that they are closely related with the progression and development of CKD. Characterizing the molecular role of immune cell infiltration into CKD is crucial, and immunogene features may be a prognostic or predictive factor for CKD.

In this study, we first comprehensively analysed the expression profiles of IRGs in normal and CKD patient‘s renal interstitial tissue samples and identified 15 important core genes that may be involved in the immune-related pathological processes of CKD. Most of these IRGs were higher in CKD patients than in normal controls, suggesting that immune regulation may play a key role in the pathogenesis of CKD. In addition, we explored the differences in immune cell levels between them and found that CKD patients showed higher levels of T-cell, macrophages and mast cells, which is consistent with earlier studies (Zeisberg and Neilson, 2010; Lee et al., 2020).

Subsequently, we investigated for the first time the identification of two distinct immune-related clusters based on central IRGs. Cluster 2 showed higher immune scores and levels of immune infiltration than Cluster 1. GO enrichment analysis of the two clusters revealed that leukocyte cell adhesion and regulation, as well as immune-related processes (e.g., antigen processing and presentation, immune response activation, T-cell activation regulation) were significantly different immune processes. The buildup of interstitial leukocytes is a feature of numerous CKD subtypes. Leukocyte adhesion is a key link in the pro-inflammatory, pro-apoptotic, and pro-fibrotic mediators produced by interstitial leukocyte infiltration, thereby contributing to the progression of CKD (Anders et al., 2006). Locally secreted chemokines play a key role in the progression of CKD by mediating the recruitment of inflammatory leukocytes and activation of immune cells (Anders et al., 2006). Additionally, enrichment analysis suggests that clustering distinguishes CKD patients with different levels of immune response and immune activation. KEGG enrichment analysis reveals distinction between immunometabolic disorder pathways and abnormal immune diseases in CKD subtypes.

Given that immune molecular features may be key regulators of individual heterogeneity in CKD. Based on three machine learning algorithms, we next evaluated immune-related DEGs between the two clusters and selected five important immune-related genes (CD48, CTSS, LYZ, ITGB2, and ISG20) to predict subtypes of CKD patients. The CD48 molecule is a glycosylphosphatidylinositol (GPI)-anchored cell surface protein of the CD2 family of molecules, and through its interaction with the ligands CD2 and CD244, it contributes to a variety of immune processes, including immune prominence of tissues, cell adhesion and co-stimulation, regulation of target cell lysis, and promotion of T-cell activation (Mcardel et al., 2016). ISG20 (interferon stimulated exonuclease gene 20) is an RNA nucleic acid exonuclease that stimulates the progression of a variety of tumors (Gao et al., 2019), and enhanced ISG20 expression is correlated with greater infiltration of monocyte-derived macrophages and neutrophils and suppresses adaptive immune responses (Gao et al., 2019). The expression of ISG20 was significantly elevated in renal fibrosis compared to normal samples, and in vitro knockdown of ISG20 significantly inhibited fibrotic protein expression in HK-2 (Sun et al., 2022).

As a member of the cysteine protease family, CTSS (cathepsin S) levels are elevated in a variety of diseases and are strongly associated with diseases such as IgA nephropathy, diabetes mellitus and atherosclerosis (Liu et al., 2006; Zhao et al., 2021). According to a recent study, macrophage-derived CTSS may hasten endothelial damage and nephrosclerosis in diabetic nephropathy (Kumar Vr et al., 2016). In addition, CTSS mediates the TGF-β/smad signalling pathway to promote extracellular matrix deposition and epithelial-mesenchymal transition (EMT) in the kidney (Yao et al., 2019). ITGB2 is capable of encoding integrin β-chains that binds to other α-chains to form distinct integrin heterodimers. It is involved in cell surface-mediated signalling and immune responses, and functions to promote leukocyte adhesion, extravasation and extracellular matrix remodelling (Tan, 2012; Boguslawska et al., 2016). LYZ (lysozyme) is an innate immune protein released from neutrophil and macrophage granules (Faust et al., 2000), which has antimicrobial properties and is involved in immune regulation, inflammatory signalling, vasodilation and myocardial inhibition (Mink et al., 2008). A recent study indicated a significant association of LYZ with all CKD stages, eGFR and survival (Makridakis et al., 2020). LYZ may be a promising therapeutic target for chronic kidney disease. Subsequently, we constructed a nomogram model for the diagnosis of CKD subtype based on the 5 gene and found that the model had significant predictive power, suggesting the value of this prediction model in clinical application. In other words, the model can provide clinicians with a trial strategy to pre-emptively identify patients at high risk of CKD immunisation and to develop early interventions to slow the trajectory of renal decline in patients.Notably, all of these biomarkers were overexpressed in the CKD samples compared to controls. The potential diagnostic value of these markers was also further analysed and CTSS, LYZ, and ISG20 were found to be highly accurate in differentiating CKD from normal tissue and to correlate closely with renal function.

For the immune infiltration results between clusters, we discovered that Cluster 1 significantly infiltrated cells were naive B-cell, plasma cells, naive CD4 T-cell, memory resting CD4 T-cell, and activated mast cells, whereas Cluster 2 had a higher abundance of CD8 T-cell, T-cell gamma delta, and M1 and M2 macrophages. Studies between immune metabolism and naive B-cell and plasma cells have been extensively explored. When naive B-cell are activated, oxidative phosphorylation (OXPHOS), TCA cycling, and nucleotide biosynthesis increas (Waters et al., 2018). Plasma cells are able to rely on glutamine and long-chain fatty acids as substrates for oxidative metabolism to provide underlying oxidative phosphorylation (Raza and Clarke, 2021). Whether these immune cells are associated with the enhanced metabolic pathway in cluster 1 is unclear. Mast cells and CD4 T-cell exert a role in regulating inflammatory cell infiltration, but their role in CKD should be interpreted with caution (Kim et al., 2009; Tapmeier et al., 2010). Depletion of naive T-cell and CD4⁺ central memory cells is associated with progressive renal function decline (Litjens et al., 2006). γδ T-cell promote renal interstitial fibrosis (Law et al., 2019). Macrophages, as important heterogeneous cells of the innate immune system, can exhibit different phenotypes in response to the local microenvironment. Most forms of renal inflammation are characterized by M1 macrophage infiltration in the early stages, but M2 macrophage infiltration predominates in the chronic phase, and polarization of macrophages is noted during CKD (Lee et al., 2020). It is reasonable to expect that elevated M1 and M2 macrophages are observed in CKD. Overall, these results suggest that subtype 2 has higher levels of immune infiltrating cells that promote CKD progression than subtype 1, representing a worse prognosis for CKD. Finally, we evaluated the correlation between the biomarkers CTSS, LYZ and ISG20 and infiltrating immune cells. CTSS, ISG20, and LYZ were all positively correlated with T-cell gamma delta, M1 and M2 macrophages, and memory B-cell and negatively correlated with naive B-cell, naive CD4 T-cell, memory resting CD4 T-cell, and naive CD4 T-cell, confirming that these genes play an important role in differentiating subtypes. Furthermore, these genes are positively correlated with T-cell costimulation, T-cell coinhibition, parainflammation, MHC class I, inflammation-promoting genes, cytolytic activity, human leukocyte antigen (HLA), checkpoint activation, chemokine C-C-motif receptor (CCR), and antigen-presenting cells (APC) costimulation. Overall, a deeper understanding of immune cell-biomarker correlations may provide new insights into immune mechanisms, and targeting CTSS, LYZ, and ISG20 to improve abnormal immune status may be a promising method for treating CKD.

We collected sufficient samples and used comprehensive bioinformatics analysis methods, such as WGCNA and machine learning, which made this study more comprehensive and reliable. However, some limitations of our study have to be acknowledged. First, under long-term chronic kidney disease pathophysiological conditions, the gene expression patterns of certain immune cell types may be influenced and adjusted due to a number of immunologic or non-immunologic complex pathogenetic factors. Further work exploring this differential variation is necessary. Secondly, our study mainly focused on interstitial renal data and a parallel comparison of glomerular and tubulointerstitial data seems to be better convincing and reliable. Furthermore, it is necessary to collect additional important clinical information such as age, gender, stage, prognosis, etc. To further verify the diagnostic properties of the model in predicting CKD clusters as well as to adjust the clinical prediction curves. Finally, this study is based on published data, and although external datasets were applied for validation, more experiments are needed to demonstrate the biological function of the studied biomarkers.

In summary, this study identified two immune-related clusters in CKD patients by unsupervised clustering and used to elucidate the biological functions between different subtypes using a bioinformatics analysis approach.A diagnostic model based on five characteristic IRGs (CD48, CTSS, LYZ, ITGB2, and ISG20) identified by three machine algorithms was shown to have good ability to discriminate between subtypes. CTSS, LYZ, and ITGB2 were demonstrated to be promising diagnostic biomarkers for CKD by using external datasets, and they all showed a negative correlation with GFR. Immune cell infiltration is heterogeneous among subtypes, and CTSS, LYZ, and ITGB2 are closely associated with multiple immune cells. Our study preliminarily explored the underlying immunomolecular mechanisms leading to CKD heterogeneity, and these results may provide new insights to explore the pathophysiology and prognostic heterogeneity of CKD patients.