Co-differential genes between DKD and aging: implications for a diagnostic model of DKD

Introduction

Diabetic kidney disease (DKD) is a severe complication of diabetes mellitus (DM). Over the past two decades, the incidence and mortality rates of this disease have increased rapidly (Xiong & Zhou, 2019). In addition, the prognosis is poor. DKD is a principal factor in the development of end-stage renal disease (Atkins & Zimmet, 2010). DKD is responsible for 34%–36% of fatalities among various types of chronic kidney diseases (Thomas, 2019). However, the understanding of the pathogenesis of DKD remains incomplete, and a definitive cure is yet to be discovered. Therefore, a deeper understanding of its pathogenesis and treatment methods is necessary to enhance its diagnosis and therapy.

Studies have indicated that aging is a significant contributor to the development of DKD, and can increase the prevalence of nephrosclerosis (Alicic, Rooney & Tuttle, 2017). Cellular senescence (cellular aging) is a critical factor in aging (Xiong & Zhou, 2019). Hyperglycemia stimulation in both in vitro or in vivo (DKD rats and mice) experiments has been found to accelerate cellular senescence in various types of renal cells, including proximal tubular epithelial cells (Kitada et al., 2014; Tsai et al., 2018; Verzola et al., 2008), endothelial cells (D’Onofrio et al., 2016), mesangial cells (Cao et al., 2019), and podocytes (Verzola et al., 2008), as well as shorten telomere length (Verzola et al., 2008). These findings suggest a strong correlation between aging and DKD.

In addition, hyperglycemia can promote the formation of the senescence-associated secretion phenotype (SASP) (Prattichizzo et al., 2018). This refers to the secretome produced by senescent cells. The SASP encompasses interleukins, chemokines, inflammatory factors, and growth factors (Vicente et al., 2016). SASP not only promotes cell senescence but also damages the immune system function, resulting in persistent inflammation and fibrosis (Xiong & Zhou, 2019). In addition, cellular senescence affects somatic cells, such as kidney cells, and causes cumulative alterations in the immune system. These changes result in immunosenescence (immune aging), which is characterized by a decline in adaptive immunity and an increase in low-level chronic inflammation (Barbé-Tuana et al., 2020). Immunosenescence also alters the quantity and functionality of immune cells, such as T and B lymphocytes, macrophages, and mast cells (Yang & Mou, 2017).

There is accumulating evidence suggests that the immune system plays a crucial role in DKD, challenging the traditional view that DKD is solely caused by metabolic and hemodynamic changes. The infiltration of immune cells is usually detected in renal samples at various phases of DKD (Klessens et al., 2017). Immune cells establish a pro-fibrotic microenvironment for renal cells by releasing cytokines and pro-fibrotic factors (Pérez-Morales et al., 2019). Owing to the activation of numerous signaling pathways, such as nuclear factor-kappa B (NF-κB), tumor necrosis factor-β (TGF-β), and mitogen-activated protein kinase (MAPK), chronic and continuing inflammation eventually leads to progressive renal fibrosis (Tang & Yiu, 2020). Moreover, some studies have explored the potential protective effects of anti-inflammatory therapy in DKD (Pérez-Morales et al., 2019). These findings provide evidence that inflammation promotes DKD.

Therefore, cellular senescence may contribute to DKD by exacerbating immune dysregulation. Immune cell alterations during immunosenescence play a significant role in the pathological progression of DKD (Barbé-Tuana et al., 2020).

Therefore, aging and DKD are inextricably linked and both present a similar low-level chronic inflammation state. This suggests a common genetic background between the two. However, few studies have integrated both the genetic background factors. This study aimed to conduct integrative research on type 2 DKD and aging by analyzing their co-differential genes. A genetic diagnostic model of DKD was developed by differentially expressed gene (DEG) analysis (Ritchie et al., 2015) and LASSO regression analysis (Tibshirani, 1997). Additionally, this study used consensus clustering analysis (Wilkerson & Hayes, 2010), WGCNA (Langfelder & Horvath, 2008), functional enrichment analysis (Consortium, 2019; Kanehisa & Goto, 2000), PPI network analysis (Szklarczyk et al., 2019), and immune cell infiltration analysis (Rusk, 2019) to investigate the biological mechanisms of DKD and aging co-differential genes, with a particular focus on understanding the mechanism of immune cells. Finally, the expression of these co-differential genes was verified using single-cell RNA sequencing (scRNA-seq) data from patients with early DKD and qRT-PCR tests on DKD mouse kidneys.

Materials & Methods

Results

Gene expression analysis and enrichment analysis of DKD vs control

GSEA of all genes in the GSE96804 dataset revealed that aging-related pathways were significantly enriched in DKD samples compared to control samples (Figs. 1A–1B). A total of 12,245 genes with FDR < 0.05 were identified as DKD-DEGs when compared to the control group. DKD-DEGs contained 6,369 upregulated and 5,876 downregulated genes. Genes with FDR < 0.05 and —LogFC— > 1 were identified as significant DKD-DEGs, with 306 significantly upregulated and 338 significantly downregulated DKD-DEGs. Genes were visualized using a volcano plot (Fig. 1C). Each gene was represented by a dot, and significant DKD-DEGs were colored in red or blue. A total of 159 DKD and aging co-differential genes were identified by the intersection of 12245 DKD-DEGs and 307 aging-related genes (Fig. 1D). GO enrichment analysis revealed that biological processes associated with aging (cellular senescence, replicative senescence, regulation of cellular senescence, positive regulation of cellular senescence, and negative regulation of cellular senescence), stress (response to oxidative stress and cellular response to chemical stress), apoptosis (regulation of apoptotic signaling pathway), and nutritional metabolism (regulation of carbohydrate metabolic process and response to nutrient levels) were enriched by the co-differential genes associated with DKD and aging (Fig. 1E). The minimum enrichment score for these biological processes was 7.33. The maximum enrichment score was 56.92 (replicative senescence). KEGG enrichment analysis also revealed pathways related to aging (cellular senescence, longevity regulating pathway-multiple species, longevity regulating pathway), apoptosis, metabolism (insulin signaling pathway, insulin signaling pathway, insulin resistance, endocrine resistance, growth hormone synthesis, lipid and atherosclerosis, fluid shear stress and atherosclerosis, neurotrophin signaling pathway, adipocytokine signaling pathway), and several signaling pathways related to inflammation, apoptosis or metabolism (FoxO, MAPK, PI3K-Akt, AGE-RAGE, HIF-1, Ras, ErbB, and P53) (Fig. 1F).

Identification of co-expressed gene modules

A WGCNA co-expression network was established by analyzing all 10,000 gene expressions from 41 DKD samples in GSE96804. The dataset was verified and two missing values were identified and removed (cut height = 95) (Fig. 3A). All missing values belonged to group 3 of the aging and DKD co-differential DKD subtypes. After removing them, no members of group 3 remained. The sample dendrogram was clustered using 10,000 gene expression profiles. However, as shown in Fig. 3B, the dendrogram result was identical to the consensus clustering result, which was based only on 159 DKD and aging co-differential genes. As shown in Fig. 3C, we set 3 as the power of soft-thresholding. Co-expression modules were created using dynamic tree cut. Subsequently, at a height cut of 0.25, 11 modules were merged into 10 modules and unique colors were assigned (Fig. 3D). As shown in Fig. 3E, the turquoise module was most associated with aging and DKD co-differential DKD subtypes, implying that genes in the turquoise module were highly involved in the grouping of co-differential DKD subtypes. The features of the turquoise module genes are shown in Fig. S1. A total of 49.58% of the turquoise module genes were DKD-DEGs. 73 turquoise module genes were aging-related, accounting for 23.78% of all 307 aging-related genes. Five genes (FOS, EGR1, PTGS2, IGF1, and PLCG2) were identified from the intersection of the turquoise module genes and significant DKD-DEGs.

Analysis of significant co-differential genes associated with DKD and aging

The intersection of 644 significant DKD-DEGs (FDR < 0.05 and —LogFC— > 1) and 307 aging-related genes resulted in the identification of 15 genes (Fig. 5A). These genes (FOS, HSPA1A, HSPA1B, MXD1, JUN, PTGS2, EGR1, PCK1, EGF, CETP, KL, SST, IGF1, SLC13A1, and PLCG2) were identified as significant co-differential genes associated with DKD and aging (Table 2). The expression levels of all genes were significantly downregulated (LogFC < −1) (Figs. 5B–5C). PPI network analysis indicated a close interaction between the proteins encoded by DKD and aging significant co-differential genes (Fig. 5D). Furthermore, Pearson’s correlation test revealed strong positive correlations between DKD and aging significant co-differential genes (Fig. 5E). In addition, KEGG (Fig. 5F) and GO (Fig. 5G) enrichment analysis revealed that DKD and aging significantly co-differential genes were involved in the biological process of regulating stress (temperature, chemical stress, and oxidative stress), aging, glucolipid metabolism, and glycosylation (AGE-RAGE pathway in DM complications, lipid and atherosclerosis, and response to fructose and lipopolysaccharide), as well as immune-related signaling pathways (apoptosis, TNF, MAPK, IL-17, Ras, NF-κB, Toll-like receptor, PI3K-Akt, and T cell and B cell receptor). The minimum enrichment scores for KEGG and GO enrichment analyses of DKD and aging significant co-differential genes were 9.9 and 5.31, respectively.

Development of a genetic diagnostic model for DKD

The LASSO method was used to screen for significant co-differential genes associated with DKD and aging. Based on this, a genetic diagnostic model with five co-differential genes (FOS, HSPA1A, IGF1, CETP, and PCK1) was developed (Figs. 6A–6C). The AUC value of the validation set was 0.876 (Fig. 6D), indicating that the model had an excellent diagnostic performance. The box plots showed the predicted values of the LASSO model for DKD and normal controls separately in the training and validation sets (Figs. 6E–6F), demonstrating the strong discrimination ability of this model.

Table 2:

Diabetic kidney disease (DKD) and aging co-differential genes.

Abbreviation	Full name
CETP	Cholesteryl ester transfer protein
EGF	Epidermal growth factor
EGR1	Early growth response 1
FOS	Fos proto-oncogene
HSPA1A	Heat shock protein family A (Hsp70) member 1A
HSPA1B	Heat shock protein family A (Hsp70) member 1B
IGF1	Insulin like growth factor 1
JUN	Jun proto-oncogene
KL	Klotho
MXD1	MAX dimerization protein 1
PCK1	Phosphoenolpyruvate carboxykinase 1
PLCG2	Phospholipase C gamma 2
PTGS2	Prostaglandin-endoperoxide synthase 2
SLC13A1	Solute carrier family 12 member [sodium/chloride] 3
SST	Somatostatin

DOI: 10.7717/peerj.17046/table-2

Annotation and analysis of cluster subtypes of single-cell data

After screening the scRNA-seq data of GSE131882, there were 16,504 cells with an average of 1,301 genes and 13,726 counts per cell in three kidney samples from early DKD patients and three from normal controls. UMAP analysis revealed significant changes in single-cell data between patients with early DKD and controls (Fig. 7A). Eleven renal cell types and leukocytes were annotated based on renal cell type-specific DEGs (Figs. 7B–7C, Table 3). Subsequently, the four leukocyte subtypes were classified using marker genes (Table 3). The classification results suggested that only one T lymphocyte and six monocytes were present in control kidney samples. Meanwhile, 47 T lymphocytes, 24 B lymphocytes, 31 monocytes, and 14 plasma cells were detected in the early DKD kidney samples (Fig. 7D). These findings demonstrated that immune cells were significantly increased in patients with early DKD.

Table 3:

Cell-type-specific differentially expressed genes for cell types annotation of single cell data.

Cell types	Abbreviations	Marker gene
Proximal convoluted tubule	PCT	CUBN, SLC5A, ALDOB
Complement factor H	CFH	CFH
Loop of Henle	LOH	SLC12A1
Distal convoluted tubule	DCT	SLC12A3, SLC12A2
Distal convoluted tubule/connecting tubule	DCT/CT	SLC8A1
Collecting duct-principal cell	CD-PC	AQP2
Collecting duct- intercalated cell A	CD-ICA	AQP6, KIT, ATP6V0D2
Collecting duct- intercalated cell B	CD-ICB	SLC26A4
Podocyte	PODO	NPHS1, NPHS2
Endothelium	ENDO	PECAM1, FLT1
Mesangial cell	MES	ITGA8, PDGFRB
Leukocyte	LEUK	PTPRC
T lymphocyte	T_cell	CD247, CD96, CD28
Monocytes	Monocytes	FCGR2A, CSF1R, CSF2RA, CD86
B lymphocyte	B_cell	MS4A1, PAX5
Plasma cells	Plasma	CD38, SDC1

DOI: 10.7717/peerj.17046/table-3

Histopathology of the DKD and control mice

The DKD mouse model (Fig. 8A) was established by injecting STZ and feeding the mice a high-fat diet. In the kidney histology section of DKD mice, some pathological alterations were observed in the kidney histological section of DKD mice (Fig. 8B). Hematoxylin and eosin staining revealed diffuse mesangial expansion, including mesangial cell proliferation, mesangial matrix hyperplasia, nodular changes, and tubular cell hypertrophy. Masson’s Trichrome staining revealed a significant quantity of blue-stained collagen accumulated in the glomeruli, which suggested fibrosis. PAS staining revealed glycogen deposition, mesangial cell proliferation, and thickened basement membrane in the glomeruli.

Discussion

In this study, DKD and aging co-differential genes and DKD and aging significant co-differential genes were comprehensively analyzed. Based on these genes, a diagnostic model for DKD was developed.

Aging is a significant risk factor for DKD (Alicic, Rooney & Tuttle, 2017). High glucose stimulation has been shown to accelerate cellular senescence in several types of kidney cells (Cao et al., 2019; D’Onofrio et al., 2016; Kitada et al., 2014; Tsai et al., 2018; Verzola et al., 2008). Through whole-gene GSEA, GO, and KEGG enrichment analyses, we found that aging-related pathways were significantly enriched, demonstrating that aging plays an important role in the pathogenesis and progression of DKD at the genetic level. We established aging and DKD co-differential DKD subtypes based on heterogeneity in the expression of DKD and aging co-differential genes. Subsequently, we observed that different aging and DKD co-differential DKD subtypes had significant differences in whole-genome expression and biological processes closely associated with both aging and DKD development, such as glucolipid metabolism and epithelial-mesenchymal transition. Furthermore, subtype categorization based on co-differential genes yielded results similar to those of the whole-gene-based WGCNA dendrograms. These findings suggest that integrative research on aging and DKD is important for the development of novel diagnostic and therapeutic strategies.

Through an integrative study of aging and DKD, co-differential genes were obtained by gene intersection. These DKD and aging co-differential genes, as well as DKD and aging significant co-differential genes, are involved in not only aging, but also various biological processes, including stress, apoptosis, glucolipid metabolism, glycosylation, and several immune-related signaling pathways. These biological processes are similar to those involved in the pathophysiology of DKD. This suggests that these genes may play a decisive role in DKD progression and could serve as novel markers for DKD diagnosis.

Therefore, a genetic diagnostic model for DKD has been developed based on these genes. ROC curves demonstrated the excellent diagnostic efficiency of this model. The expression of these model genes was validated using single-cell sequencing data and mouse models. In our DKD mouse model, kidney histopathology revealed increased glycogen deposition, mesangial proliferation, thickened basement membrane, and increased fibrous collagen, all of which are pathological hallmarks of DKD (Alicic, Rooney & Tuttle, 2017). The expression levels of FOS, HSPA1A, PCK1, and IGF1 differed between the two assays.

In line with our findings, previously published studies have shown that FOS, HSPA1A, PCK1, and IGF1 are directly or indirectly associated with DKD. FOS and HSPA1A are closely associated with immunity.

IGF1, which encodes insulin-like proteins and activates the type I IGF receptor, is a component of the growth hormone-IGF-somatostatin system that plays a critical role in DKD (Segev et al., 2004). Research has indicated that IGF1 can inhibit apoptosis and DNA damage in high glucose-stimulated mesangial cells (Kamenický et al., 2014). Consistent with our study, significantly decreased serum IGF1 and decreased renal Igf1 RNA levels were found in db/db mice with DKD in a previous study (Segev et al., 2007). A lifestyle intervention trial also found that low IGF1 levels predicted DM onset in patients with prediabetes (Meyer et al., 2022). However, unlike our study, which demonstrated low IGF1 expression in DKD patients, some studies have reported that IGF expression increases with DKD progression (Dong et al., 2019). Elevated IGF-1 levels may be associated with adverse renal outcomes in patients with DKD (Mohebi et al., 2023). However, studies have also demonstrated that IGF1 expression varies with stage of DM (Mohebi et al., 2023). Moreover, both low and high normal levels of IGF-I may be associated with insulin resistance (Friedrich et al., 2012). These results may explain the inconsistent data obtained from different experiments. At the same time, these findings also suggest that the treatment strategy for DKD patients with high IGF1 gene expression should differ from those with low IGF1 gene expression.

PCK1, a critical regulator of gluconeogenesis, is a key regulator of renal tubular cell metabolism that can alter mitochondrial function (Verissimo et al., 2023). Furthermore, PCK1 shows significantly higher expression in the kidneys than in other organs (Fagerberg et al., 2014). Consistent with our DEG analysis and mouse qPCR results, previous studies have shown that Pck1 expression is downregulated in STZ-induced DKD mice. Moreover, overexpression of Pck1 has been found to improve proteinuria and collagen deposition (Hasegawa et al., 2023). These results suggest that PCK1 plays a protective role in DKD and could be a potential diagnostic and therapeutic target for future studies. In contrast, some researchers have reported that Pck1 mRNA is significantly increased in the kidneys of DKD db/db mice (Watanabe et al., 2018). This finding corroborated the conclusion of our single-cell analysis of patients with early DKD, revealing that Pck1 expression levels may vary at different stages of DKD, which requires further investigation.

FOS encodes c-Fos, a transcription factor subunit of activator protein 1 (AP-1). AP-1 is one of the targets of the MAPK signaling cascade (Atsaves et al., 2019) and can be activated by various DKD-related stimuli (Huang et al., 2014). Furthermore, AP-1 can affect inflammation through different mechanisms, such as regulating naive T cell differentiation and regulating the activity of the innate immune system (Wagner & Eferl, 2005). Furthermore, some studies investigations have shown that AP-1 can also affect renal fibrosis (Huang et al., 2014).

HSPA1A encodes heat shock protein 70 (HSP70), a type of molecular chaperone involved in protein folding and remodeling (Rosenzweig et al., 2019). It is also effective as an immunological adjuvant. Hsp70 can directly inhibit heat shock factor 1, which is a vital component of the heat shock response. Moreover, intracellular Hsp70 downregulates NF-κB activation and exhibits immunosuppressive activity. When Hsp70 is extracellular, the reverse effect occurs (Tukaj, 2020). Previous research has indicated that serum HSP70 levels are higher in patients with DM with albuminuria than in those without albuminuria (Nargesi et al., 2016), which is contrary to the trend of decreased HSPA1A gene expression in the kidneys of DKD patients that we investigated. Intracellular and extracellular HSP70 play distinct roles in inflammation (Zhou et al., 2023), which may explain the different trends in the kidney compared to the serum. However, in our single-cell analysis of patients with early DKD, the expression of HSPA1A in endothelial, mesangial, and proximal renal tubular epithelial cells increased, whereas podocyte HSPA1A expression decreased. Other studies have shown contradictory results, suggesting that both intracellular and extracellular Hsp70 play a dual role in inflammation (Tukaj, 2020). Its complex and confusing relationship with DKD warrants further investigation.

In contrast, studies have revealed that DKD shows a low-level of chronic inflammation (Pérez-Morales et al., 2019). Immune disorders in patients with DKD were observed in our study. Whole-gene GSEA also revealed significant gene enrichment in immune-related pathways in patients with DKD compared to controls. Furthermore, immune cell infiltration analysis revealed significant heterogeneity between DKD and normal tissues. Similarly, scRNA-seq data analysis revealed that early DKD patients had significantly higher levels of T lymphocytes, B lymphocytes, monocytes, and plasma cells than controls. Consistent with our findings, many studies have demonstrated a promoting effect of immunity on DKD development. Diabetic stress recruits leukocytes, such as macrophages and mast cells, via danger-associated molecular patterns, resulting in kidney damage (Tesch, 2017). T cells, which are recruited to the kidneys of patients with DKD, also contribute to the development of DKD (Chen et al., 2022).

Aging also causes low-level chronic inflammation (Barbé-Tuana et al., 2020), and senescent cells may trigger an immunological response to DKD. Our findings revealed that different aging and DKD co-differential DKD subtypes exhibited significant immunological heterogeneity. ssGSEA revealed differences in genes encoding inflammatory responses between these subtypes. Furthermore, a noticeable discrepancy in the immune cell infiltration was observed. In addition, enrichment analysis of key WGCNA genes revealed that these genes had a significant biological role in DKD pathogenesis and aging, as well as a close relationship with immunity and inflammation.

These findings suggest that DKD progression may be linked to immunosenescence and that aging plays a significant role in DKD immunological disorders. Moreover, aging-targeted therapy may represent a novel treatment for immunological dysfunction in patients with DKD. Our diagnostic model genes (IGF1, CETP, PCK1, FOS, and HSPA1A), particularly IGF1, PCK1, FOS, and HSPA1A, are potential targets.

Our study has several limitations. First, our findings were based on analysis of the mRNA expression levels of genes rather than protein levels. Second, owing to publication bias of available data, a small sample size, and limitations of animal models, we plan to conduct experiments in subsequent studies to validate the conclusions and address these limitations.

Conclusions

This study identified the co-differential genes between DKD and aging, and developed a DKD diagnostic model based on these genes. Our diagnostic model genes (IGF1, CETP, PCK1, FOS, and HSPA1A), particularly IGF1, PCK1, FOS, and HSPA1A, are potential diagnostic and therapeutic targets for DKD treatment. Additionally, DKD progression may be linked to immunosenescence.

Supplemental Information