To target cellular senescence in diabetic kidney disease: the known and the unknown

Cellular senescence is a physiological and natural process in organic systems. The word ‘senescence’ originates from the Latin word ‘senex’ which means ‘old man’ or ‘old age.’ In 1891, Minot used the word to describe the passage of youth to old age in the biological context [1] and in the 1960s, Hayflick and Moorhead described the degeneration and passage limitation of human fibroblasts in culture, which was first explained by ‘senescence’ [2]. However, at that time, the concept of senescence was unclear both at the cellular level and at the level of the organism.

The understanding of senescence has significantly advanced over the last few decades. Initially, the phenomenon of cellular senescence referred to events in cultured cells where cell replication was attenuated with age. The current definition refers to a permanent cell growth arrest state in response to distinct stresses, such as oncogene activation, DNA damage and oxidative stress [3]. Since senescence is a phenotype, resulting from multiple effector mechanisms, the identification of senescence is based on the presence of a combination of multiple biomarkers, such as senescence-associated β-galactosidase (SA-β-Gal), cell cycle arrest, senescence-associated secretory phenotype (SASP), DNA damage response (DDR), abnormal expansion and flattening in cell morphology [4,5].

Diabetic kidney disease (DKD) is the most common case of chronic kidney disease globally and remains a major cause of end-stage renal disease (ESRD) [6]. Approximately 40% of diabetic patients develop some features of DKD, leading to a major healthcare burden [7]. The molecular mechanisms underlying the development of DKD have been widely explored but the therapeutic strategies for DKD are still limited [8]. The pathogenic role of senescence in DKD and treatments that target this process are gaining increasing attention [9]. Here, we review the current understanding of senescence in DKD and summarize the possible therapeutic interventions relevant to senescence in the field.

SA-β-Gal activity was reported to be a biomarker for senescent cells in culture and in vivo approximately 30 years ago [10]. The enzymatic activity of SA-β-Gal at pH 6.0 can be detected in senescent cells, but not in most non-senescent, quiescent or terminally differentiated cells, whereas acidic β-Gal activity can be detected in all cells at pH4.0 [10] SA-β-Gal activity originates from lysosomes and the enzyme is the classical lysosomal β-D-galactosidase, encoded by the gene GLB1 [11]. Senescent cells have low activity of β-Gal as a result of knockdown of GLB1 [11]. Interestingly, lysosomal β-Gal activity can be detected in all non-senescent cells at pH 4.0, the optimal condition for the enzyme. In a suboptimal environment at pH6.0, the activity of β-Gal is usually not detectable in most non-senescent cells. In senescent cells, lysosomal biogenesis is increased leading to an increased number of lysosomes and an increased abundance of lysosomal enzymes, including the lysosomal enzyme β-Gal. The enzymatic activity of the abundant lysosomal β-Gal in these senescent cells becomes detectable under the suboptimal condition at pH 6.0 [12–14]. The β-D-galactose residues in β-D-galactosides are the substrates of β-Gal [8]. To detect the activity of β-Gal in an assay, the X-gal chromogen (5-bromo-4-chloro-3-indoyl β-d-galactopyranoside) is the most popular chemical used which serves as an exogenous substrate, and can be cleaved into galactose and 5-bromo-4-chloro-3-hydroxyindole. 5-bromo-4-chloro-3-hydroxyindole is then oxidized to a blue product, 5,5′-dibromo-4,4′-dichloro-indigo, which reflects the activity of β-Gal activity in that assay [15,16].

However, the detected activity of β-Gal at pH 6.0 is not specific to senescent cells. The increased lysosome biogenesis due to autophagy activation can also lead to increased β-Gal activity [17,18]. The staining is also increased in immortal cells when they reach high cell density [19]. The β-Gal activity at pH 6.0 is also detectable in tissue-resident macrophages [20]. Furthermore, the osteoclasts in bone marrow show a high level of β-Gal activity, regardless of the staining pH [21]. Therefore, this marker, despite being useful, cannot be used alone as a definitive marker to reliably define cellular senescence. Thus, it is recommended that several other hallmarks of senescence should be used in combination with SA-β-Gal to reliably identify senescent cells.

Irreversible cell cycle arrest is an important feature of senescent cells, which has attracted researchers’ attention for >30 years [22,23]. The p53-p21- retinoblastoma protein (RB) pathway plays an important role in regulating the cell division cycle [24]. p53, encoded by the TP53 gene, is a tumor suppressor, whose loss of function mutations are associated with malignant proliferation [25]. The protein p53 is a powerful transcription factor regulating transcriptional expression of multiple genes and biological processes in response to cellular stresses such as oncogene activation and DNA damage [26]. p21 is encoded by the cyclin-dependent kinase inhibitor 1A (CDKN1A), also known as wildtype activating factor-1 (WAF1) or cyclin-dependent kinase inhibitory protein-1 (CIP1) [27]. The p21 protein interacts with cyclin-dependent kinase (CDK), inhibiting CDK activity and acting as a negative cell cycle regulator [28,29]. Retinoblastoma protein (RB), encoded by RB1, belongs to a family comprising three members including RB, retinoblastoma-like 1 (RBL1 or p107), and retinoblastoma-like 2 (RBL2 or p130) [30]. RB is a critical regulator inhibiting the G1/S transition, whose deficiency is associated with many malignant tumors including retinoblastoma, prostate cancer, lung cancer, and breast cancer [30]. RB modulates the cell cycle via interactions with the early region 2 binding factors (E2Fs), a transcription factor family that serves as a positive regulator for cells to progress to the S phase [31,32]. In detail, hypo-or monophosphorylated RB forms a complex with E2Fs, attenuating the function of E2Fs as a cell division promoter [31,33]. However, the highly activated CDK would hyperphosphorylate RB, dissociating the RB/E2Fs complex thus leading to E2Fs activation [31]. Collectively, p21 is activated by p53 at the transcriptional level, followed by the inhibition of CDKs, such as CDK1 and CDK2, by the p21 protein [34]. The inhibition of CDKs results in hypo- or monophosphorylation of RB and the inactivation of E2Fs, ultimately leading to cell cycle arrest [24]. Apart from the p53-p21-RB pathway, another critical CDK inhibitor named p16, encoded by cyclin-dependent kinase inhibitor 2A (CDKN2A), is also involved in inhibiting cell cycle progression [35], and its up-regulation is associated with cellular senescence. Indeed, tissue accumulation of p16 positive cells accelerates organ aging and elimination of these cells delays ageing-associated disorders in vivo [36,37]. p16 belongs to the inhibitor of cyclin-dependent kinase 4 (INK4) family, which consists of four members including p16/INK4A, p15/INK4B, p18/INK4C and p19/ INK4D [38]. p16 binds to CDK4/6 and inhibits their activity, resulting in the hypo- or monophosphorylation of RB and subsequent abrogation of E2F function [39]. Transcriptional regulation of p16 is mediated by two complexes comprised of multiple proteins, such as B-cell-specific Moloney murine leukemia virus integration region 1 (BMI-1) and Enhancer of zeste homolog 2 (EZH2) [39]. The two protein complexes negatively regulate the expression of p16 by methylating its gene promoter [39].

In the 1990s, the up-regulation of p53, p21, and p16 in senescent cells was reported to be useful biomarkers for cellular senescence [40–43]. Generally, the p53-p21 pathway is activated upon DNA damage response (DDR) and p16 can be induced by oncogenic activation and by various other stressors [44]. Interestingly, it was reported that in early senescent cells, p21 was increased but p16 remained at a low level. However, in late senescent cells, the level of p21 could return to a nearly normal level but p16 accumulated to a high level [45]. Thus, the p53-p21 axis plays a role in regulating senescence in the early stage and might be important for senescence induction [46], whereas p16 is necessary for the maintenance of cellular senescence during the relatively late stage [45,46]. Collectively, p53-p21 and p16 are responsible for cell cycle arrest jointly or separately according to the stressor and cell type [47].

SASP was reported in senescent cells >20 years ago when overexpression of certain cytokines such as interleukin 1 α (IL-1α), and intercellular cell adhesion molecule (ICAM) were observed in senescent cells [22]. However, the concept of SASP wasn’t clearly defined until 2008 [48], when Campisi and colleagues studied the molecules secreted by different senescent cells. They initially found that all the different cell populations showed a similar feature upon senescence induction, e.g. having high levels of secreted inflammatory cytokines and chemokines, immunomodulatory cytokines, and growth factors, with increased expression of inflammatory cytokines such as IL-6 and IL-8 considered the core feature of the SASP [48]. However, more extensive studies subsequently showed that SASP composition could be heterogeneous, and affected by the cell type and the nature of the stresses [49]. In fibroblasts, senescent cells induced by ionizing radiation were found to secrete 343 SASP proteins [50]. The most highly up-regulated proteins in the senescent fibroblasts included inflammatory cytokines, hemostasis-related factors and extracellular matrix (ECM) proteins [50]. In another study, quantitative unbiased proteomic analysis of secreted soluble proteins and exosomes in the medium of cells treated with three different senescence inducers revealed that hundreds of proteins were significantly changed in senescent fibroblasts and renal epithelial cells [51]. That study demonstrated that chemokine C-X-C motif ligands (CXCLs), matrix metalloproteinases (MMPs), and tissue inhibitors of metallopeptidase (TIMPs), were frequently detectable SASP molecules [51]. It was also observed that there were fewer elevated proteins detected in renal epithelial cells than in fibroblasts, indicating the distinct composition of the SASP in different cell populations [51].

Stimuli such as chemotherapeutic drugs, chemicals, γ-irradiation, ultraviolet radiation, and endogenous oxidative damage, can cause single-strand or double-strand breaks in chromosomal DNA molecules, which trigger the DDR in affected cells [52,53]. The protein kinases Ataxia Telengectasia Mutated (ATM) and Ataxia Telangiectasia and RAD3-related (ATR), belonging to the phosphatidylinositol 3-kinase-like protein kinases (PIKKs) family, are key sensors of DNA damage or genotoxic stress [54,55]. ATR/ATM phosphorylates checkpoint kinase 1/checkpoint kinase 2 (CHK1/CHK2)which can phosphorylate p53. The phosphorylation of p53 induces accumulation of p53 in the cells, which transcriptionally up-regulates p21, leading to cycle arrest and cellular senescence [56]. DNA damage is known to induce not only cellular senescence but also apoptosis, with speculation that prolonged moderate DNA damage leads to cellular senescence, while severe DNA damage results in apoptosis [57].

The DDR is both an activator and a feature of senescence. Most of the stimuli activating senescence could affect DNA directly or indirectly. For example, telomere shortening during replicative senescence leads to DNA breaks [58]. In oncogene-induced senescence, cellular hyperproliferation triggered by oncogene activation is accompanied by accumulated genomic damage, which will activate the DDR [56]. Therefore, the activation of DDR signaling can serve as a marker to identify cells in the senescent state. There are many DNA damage response related proteins, which have been extensively studied, such as γH2AX, ATM, p53-binding protein 1 (53BP1) and mediator of DNA damage checkpoint protein 1 (MDC1) [55,59]. Phosphorylation of Serine 139 of histone H2AX (γH2AX) has been recognized as the activation signal of the DDR [60]. When DNA double-strand breaks (DSB) occur, the PI3K-like kinases such as ATM are activated and phosphorylate H2AX [61]. γH2AX foci were found in the region near the DSB site shortly after the cells were exposed to DNA damage stimuli [61]. Increased deposition of γH2AX has been widely used as a marker of DDR or a genotoxic stress marker in research [62,63]. Similarly, increased levels of 53BP1, MDC1 and the ATM/ATR/CHK1/CHK2/p53 cascade can also be detected to characterize the cellular senescence state [64,65].

Heterochromatin is a modified and condensed chromatin structure where some proliferation-promoting genes are included and transcriptionally inactivated. Facultative heterochromatin formation is considered to be the key mechanism for X inactivation in female cells where one of the two X chromosomes is transcriptionally inactivated. In 2003, Narita and colleagues reported that the formation of facultative heterochromatins can be stimulated by senescence inducers in cultured human diploid fibroblast cells (IMR-90 cell line). In these induced senescent cells, the heterochromatins were shown to contain E2F target genes of the proliferation promoting pRB/E2F pathway as well as binding of a histone H2A variant, macroH2A and common heterochromatin marker proteins such as heterochromatin proteins 1 (HP1) and Histone 3 with its Lysine [9] residue methylated (Me-K⁹-H3) [66]. This finding is consistent with the previous observation that senescent cells have extensive chromatin remodeling [67]. The unique domains of the transcriptionally silent foci observed in senescent cells are known as SAHF. SAHF can be detected by dense 4′,6′-diamidino-2-phenylindole (DAPI) staining and visualized by dense puncta staining patterns enriched and colocalized for repressive chromatin marks such as trimethylated histone H3 Lys9 (H3K9me3), macroH2A and HP1 [66,68,69]. However, SAHF is not always obviously detected in senescent cells induced by different senescence stimuli. For example, oncogene-induced senescence by exogenous expression of oncogenic Ras is associated with stronger SAHF staining, when compared with other inducers, such as ionizing radiation and H₂O₂ treatment [70].

Oxidative stress is a senescence driver and also a feature of cellular senescence. Reactive oxygen species (ROS) can promote senescence by inducing mitochondrial DNA (mtDNA) damage and activating the p53, p21, and p16 pathways [71]. In fact, ROS accumulation also results in the oxidization of mitochondrial proteins and enzymes, which further impairs mitochondrial function [72]. Mitochondrial dysfunction can lead to a decrease in the efficiency of oxidative phosphorylation and more ROS production, which, together, can further accelerate cellular senescence [73].

H₂O₂ can induce a senescent phenotype in cultured cells and has been widely used as a cellular senescence inducer [74–76]. In contrast, antioxidants can inhibit cellular senescence by reducing ROS levels [77]. The activation of antioxidant signaling pathways such as the Nrf2 pathway is associated with the amelioration of senescent phenotypes [78]. Therefore, the detection of ROS is a useful parameter to reflect the cellular status of senescence.

The accelerated cellular senescence in a diabetic environment has been demonstrated in both animal models and in human kidney biopsy samples. p21 is persistently induced in kidneys from STZ-induced diabetic mice and db/db mice, at both the mRNA and protein levels [79]. Other senescence markers including SA-β-Gal staining, γH2AX, and p16 are also observed, supporting the presence of premature senescence in the kidneys of diabetic mice [79,80]. Diabetic Akita mice show higher levels of senescence-associated gene expression, higher oxidative stress, and more mitochondrial DNA damage when compared with wild-type counterparts [81]. The bradykinin B2 receptor (B2R), a G-protein coupled receptor (GPCR), involved in neurodevelopment, neuroprotection, blood pressure regulation, and inflammation control [82], is reported to inhibit oxidative stress- and high-glucose-induced senescence, via the p53 and RB pathways [83,84]. Indeed, Akita mice with a mutation at the B2R locus (Bdkrb2–/–Ins2Akita/+) have a more severe senescence phenotype in the kidney [81]. By contrast, genetic deletion of p66^ShcA, an adaptor protein mediating the intracellular signal transduction of GPCRs, can alleviate the aging phenotype in the kidneys of Akita diabetic mice [85]. Indeed, p66^ShcA has been shown to increase the levels of ROS in cells treated with high glucose and promote oxidative stress in DKD mice [86,87].

Indeed, oxidative stress as a result of increased intracellular accumulation of ROS has been shown to be a critical pathological stimulus in DKD. NADPH Oxidase 4 (NOX4) is a major NOX isoform expressed in the mouse kidney, and it is the key enzyme producing ROS in the kidney in diabetic mice. Previous studies have shown that either genetic deletion or pharmacological inhibition of NOX4 reduces renal ROS levels leading to attenuation of experimental DKD, including key profibrotic and proinflammatory parameters [88]. On the other hand, transgenic expression of the human Nox5 gene in mouse kidneys, which is otherwise absent in rodents, further increases renal ROS levels and consequently exacerbates DKD in mice [89]. In the latter study, renal expression of p21, one of the key cellular senescence markers, and renal levels of ROS were concurrently increased in diabetic mice, with these changes further increased by transgenic expression of human NOX5 [89]. Both NOX4 and NOX5 are expressed in the human kidney and their expression levels are elevated in DKD. These experimental data support the view that approaches to reduce oxidative stress are effective to confer renoprotection in DKD, likely as a result of reduced oxidative stress-induced cellular senescence.

In human samples, SA-β-Gal staining and p16 expression are increased in Type 2 diabetic nephropathy biopsies compared with normal age-matched control tissues, supporting the presence of increased cellular senescence in DKD [90]. Furthermore, the markers of senescence are more pronounced in tubular cells than in glomeruli [90]. Interestingly, glomerular p16 expression was found to be associated with proteinuria, while tubular p16 was associated with key risk factors for diabetes, such as body mass index, LDL cholesterol, and HbA1c [90]. Increased γH2AX staining is also reported in the kidney from DKD patients [79]. Mechanistically, high glucose has been shown to induce senescence in tubular epithelial cells, glomerular mesangial cells, endothelial cells, and immortalized podocytes [91–94].

Taken together, these biological characteristics are observed in senescent cells derived from various cell types and/or in various contexts (Figure 1). Since some of these changes also occur and can be detected in other biological processes unrelated to cellular senescence, detection of multiple biomarkers for these characteristics is essential in order to identify and characterize senescent cells in DKD. It should be appreciated that the potential pathological roles of these cellular senescence related biological characteristics are not yet fully elucidated. Possible pathogenic roles of some of these biological characteristics and related biomarkers in DKD are summarized in Table 1.

Urinary p21 is detectable in samples from DKD patients, but not in healthy controls or in patients with other diseases with normal kidney function [79]. The association of levels of urinary p21 with an increasing grade of CKD using KDIGO criteria has also been confirmed in a large cross-sectional cohort [79]. Plasma Activin A, a marker of SASP, is quantitatively associated with senescence and kidney fibrosis in DKD. Compared with individuals without diabetes or CKD, patients with DKD and reduced kidney function have significantly increased levels of plasma Activin A [95]. Some urinary inflammatory cytokines such as IL-6, IL-10, the molecules present in the SASP, are also proposed to be indicators of kidney injury in DKD [96]. Although the relationships among these senescence markers and kidney damage are reported as described above, all these studies show no direct evidence of the association between the levels of these urinary senescent molecules and the degree of senescence in the kidney.

Oxidative stress and the associated signaling pathways play important roles in regulating cellular senescence in DKD. Glomerular endothelial senescence could be driven by M1 macrophages and is dependent on intracellular ROS [94]. As mentioned above, NADPH oxidase (NOX) is a significant contributor to oxidative stress in DKD. NOX1 is reported to promote premature senescence in DKD by modulating the p38/p27 signaling pathway and activating PKCα/β [80]. The kelch-like ECH-associated protein 1 (KEAP1)/nuclear factor erythroid 2-related factor 2 (NRF2) pathway is a powerful modulator of redox balance, which also plays an important role in aging [97]. Pyrroloquinoline quinone (PQQ), a redox coenzyme, can inhibit oxidative stress and senescence in HK-2 cells in a high glucose environment by affecting the KEAP1/NRF2 pathway [98]. The lower senescence is associated with enhanced NRF2 translocation to the nucleus and increased levels of certain downstream antioxidants [98]. β-Hydroxybutyrate is one of the ketone bodies derived from fatty acid metabolism and used as an energy source when glucose cannot be efficiently used. β-Hydroxybutyrate has been shown to inhibit ROS generation, lipid peroxidation and protein oxidation [99]. Moreover, β-hydroxybutyrate can mitigate senescence in the kidney in STZ-induced diabetic mice as a result of NRF2 restoration, which has been confirmed in vitro in immortalized mouse podocytes cultured in high glucose media containing TGFβ1 [93]. Indeed, NRF2 is repressed in podocytes in diabetic kidney by glycogen synthase kinase 3β (GSKβ), a redox-sensitive transducer [100].

Autophagy is an important cellular process that degrades and reuses dysfunctional cellular components to regenerate new and functional cellular components. The effect of autophagy on cellular senescence appears to be complex [101]. Autophagy inhibition can either delay or promote cellular senescence, depending on cell type, senescence stimulus used, and the timing and duration of autophagy inhibition [101,102]. Inhibition of autophagy with 3-methyladenine (3-MA) contributes to advanced glycation end product (AGE)-induced senescence in mesangial cells [103]. Autophagy is associated with the alleviation of senescence in experimental DKD partially because of the degradation of SASP molecules as shown in vivo, and in HK-2 cells [104]. Mitophagy, an autophagy process to remove damaged mitochondria, has also been shown to affect cellular senescence in the context of DKD. Inhibition of mitophagy with Mdivi-1 (mitochondrial division inhibitor 1) results in enhanced senescence in renal tubular epithelial cells (RTECs) treated with high glucose, while a mitophagy agonist Torin1 reduces cellular senescence markers [91]. Enhancing mitophagy and mitochondrial function by overexpressing yeast mitochondrial escape 1-like 1 (YME1L), a metalloprotease localized to the inner mitochondrial membrane with activity to maintain mitochondrial integrity protects RTECs against senescence in DKD mice [105]. However, the molecules and pathways mediating the effect of autophagy in influencing cellular senescence in DKD remain to be fully elucidated. Interestingly, the autophagic degradation of GATA4 mediated by the autophagic receptor protein p62 is considered to be a possible mechanism [106]. The accumulation of GATA4 in human fibroblast IMR-90 cells activates SASP and promotes cell cycle arrest, which could be eliminated by autophagy [106]. Indeed, GATA4 has been found to promote senescence in DKD mice [107].

The ER is an important cellular enclosed compartment where membrane proteins and secretory proteins are synthesized by the ribosomes attached to the ER (rough ER), and then modified and properly folded within the ER to become functional proteins which are exported from the ER. The ER is also responsible for the synthesis of lipids, phospholipids and cholesterol as well as secretion of steroid hormones. Therefore, the ER is responsible for the modification, folding and trafficking of proteins. Misfolded or unfolded proteins cannot be exported and are accumulated within the ER, causing ER stress [108]. ER stress has been shown to affect multiple cell types and plays a vital role in the progression of DKD [109]. Importantly, ER stress might be involved in the progression of premature senescence in DKD [110,111]. Colocalization of the ER stress marker glucose-regulated protein 78 kDa (GRP78) and the cellular senescence marker p16 or p21 was confirmed in proximal tubular epithelial cells (PTECs) [110,111]. Furthermore, inhibition of ER stress by 4-phenylbutyrate (4-PBA) alleviated premature senescence induced by AGEs, and ER stress inducers were shown to increase the proportion of senescent PTECs in vitro [110,111].

Like mitochondria which are considered to be the major source of ROS, the ER is also a significant contributor and is estimated to produce ∼25% of total ROS [112]. Indeed, ROS are natural by-products of the key biological process of protein folding within ER. ER stress interacts with oxidative stress, which might be the critical mechanism for cell fate regulation [113]. The specific role of ER stress networks in cellular senescence modulation remains to be fully defined.

The evidence for the regulation of senescence by the TGFβ signaling pathway was reported approximately three decades ago [114–116]. Mechanistically, TGFβ promotes senescence by inducing cell cycle arrest-related genes and SASP molecules via Smad and non-Smad pathways, albeit this has only been shown in cancer cell lines and non-kidney cell lines [117]. Furthermore, TGFβ1 is a member of SASP. TGFβ1 levels are associated with levels of senescence markers such as p21, SASP molecules and oxidative stress in DKD [118,119]. TGFβ has been shown to be a critical pathogenic growth factor in DKD and anti-TGFβ approaches have been shown to attenuate diabetes induced renal injury in animal models of DKD [120–123]. Furthermore, cell division autoantigen 1 (CDA1), a molecule linked to TGFβ levels and action [124], was initially shown to up-regulate p21 expression via the p53 pathway leading to cell cycle arrest in HeLa cells [125–127], and in DNA damage response in various cancer cells [128], was subsequently found to synergistically enhance TGFβ signaling in HK-2 cells and to promote diabetes associated renal injury in DKD [129–131]. In those studies, a molecular approach to knockdown CDA1 in cells as well as genetic and pharmacological approaches to knockout the CDA1 gene or inhibit CDA1’s activity in vivo are effective attenuating TGFβ signaling and ameliorating kidney injury in animal models of diabetes. These findings support the view that the CDA1/TGFβ axis plays a key pathological role in DKD, probably as a result of promoting cellular senescence thus compromising tissue repair in the kidney upon injury by diabetes, as well as by enhancing the profibrotic process leading to tissue scar formation, known as fibrosis, at a later stage of disease (Figure 2).

There are other proteins reported to influence senescence in DKD. For instance, protease-activated protein C is reported to reverse senescence in DKD by reducing tubular p21 expression [79]. In addition, valproic acid can mitigate senescence in STZ-induced DKD mice [132], and deletion of the C5a receptor attenuates cellular senescence and the senescence-associated secretory phenotype in mice with DKD [132].

The SASP contains pro-inflammatory cytokines, such as IL-1β, IL-6, TNFα, and MCP-1 [48], and pro-fibrotic molecules such as TGFβ1, resulting in chronic inflammation and fibrosis. The pro-inflammatory cytokines can cause sustained inflammation which contributes to the pathogenesis and progression of DKD [133]. The pro-fibrotic molecules, such as TGFβ1, are powerful drivers of kidney fibrosis [134] Indeed, concurrent senescent cell accumulation and kidney fibrosis have been widely observed in mice upon renal injury [135–137].

Proximal tubules with senescent phenotypes show maladaptive repair after injury [138]. Proximal tubules appear to be the source of specialized epithelial cells with co-expression of the stem cell markers CD133 and CD24, which are able to differentiate into various renal cell populations [139,140]. The CD24⁺/CD133⁺ scattered tubular-like renal cells (CD24⁺/CD133⁺STC) are single cells scattered throughout the tubules and have self-renewal potential, playing an important role in renal recovery after acute injury [141,142]. Indeed, it was demonstrated that infusion of renal CD24⁺/CD133⁺ STCs attenuated ischemia-induced renal injury with improved renal parameters such as less hypoxia, reduced fibrosis, decreased inflammation and diminished capillary loss in mouse, while senescent renal CD24⁺/CD133⁺ STCs had impaired reparative capacity [143]. Therefore, cellular senescence plays a key role in impairing repairing capacity of renal cells, leading to disease progression (Figure 3).

Senescent cells appear to equip themselves with anti-apoptotic capability, via regulation of the Bcl-2 family of proteins [144]. Apoptosis resistance contributes to the accumulation of senescent cells in the kidney due to the ineffective elimination of these senescent cells [145,146]. In DKD, Decoy receptor 2 (DCR2), a p53 target gene influencing chemosensitivity by inhibiting the pro-apoptotic activity of the tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), was recently shown to interact with peroxiredoxin 1 (PRDX1) to induce senescence of RTECs, which play an important role in promoting renal fibrosis in DKD [137]. DcR2 was shown to induce senescent RTECs with an apoptosis-resistant phenotype by interacting with GRP78, a major ER chaperone protein playing a key role in regulating the Unfolded Protein Response in ER stress [147]. Taken together, these studies show that increased accumulation of senescent cells is associated with inflammation and fibrosis as well as the consequent loss of renal function in the diabetic kidney [37,79,148].

Some therapies that aim to improve glycemic control also show renoprotection in patients with Type 2 diabetes (T2D) [7]. Interestingly, the glucose-lowering drugs with renal protective effects are able to mitigate cellular senescence.

Metformin is one of the most popular oral glucose-lowering agents and is often the first drug used in T2D. In DKD, metformin has been shown to decrease oxidative stress and augment autophagy [149]. It can alleviate inflammation and retard fibrosis via regulation of the NFκB and TGFβ pathways [150]. It reduces senescence in HK-2 cells treated with high glucose and in db/db mice via the MBNL1/miR-130a-3p/STAT3 pathway or by regulating the expression of E2F1 [151,152]. Indeed, metformin has been proposed as a possible anti-aging drug with multiple target pathways [153]. In several human fibroblast cell lines (WI-38 cells, IMR‐90 cells, and IDH 4 cells), this drug can decrease markers of senescence, including p16, p21, IL-6, and IL-8 [154]. In murine olfactory ensheathing cells, metformin was shown to inhibit the NFκB pathway and decrease various cellular senescence markers such as SA-β-Gal activity, oxidative stress as well as inflammatory cytokines [155]. In addition, metformin down-regulates senescence and SASP via the Nrf2 and AMPK pathways [156,157].

Dapagliflozin, a sodium-glucose cotransporter-2 inhibitor (SGLT2i), can mitigate the senescence phenotype induced by high glucose in primary human RTECs [119]. SGLT2i exposure for 8 weeks is able to decrease senescence markers in the kidneys of db/db mice, including the protein levels of 53BP1, p16, p21, γH2AX, as well as SASP [158]. However, in another study, treatment with SGLT2i alone for 6 weeks failed to reduce the levels of p21 and γH2AX in STZ-induced diabetic mice, despite a significant reduction in UACR [79]. The reason for the inconsistent results might be different animal models and different treatment protocols used, such as dosage and treatment durations. The latter study also showed a trend toward decreased senescence markers, although the differences were not statistically significant [79].

Glucagon-like peptide-1 (GLP-1) receptor agonists are also renoprotective in diabetic patients [159]. Dulaglutide, a GLP-1 receptor agonist, has been reported to alleviate high-glucose-induced senescence-related gene expression and protect against diabetic retinopathy [160]. Since diabetic retinopathy and DKD are both microvascular complications of diabetes [161], it is postulated that these drugs can also alleviate senescence in DKD.

Dipeptidyl peptidase 4 (DPP-4) inhibitor is a new class of glucose-lowering medicines. Although no renal protection was found in RCTs according to the result of the composite renal outcome, DPP-4 inhibitors could partially delay the progression of albuminuria [162–164]. Recently, researchers found that DPP-4 on the surface of senescence-associated extracellular vesicles appears to hinder the internalization of the vesicles by HeLa cells, which implicated a possible role of DPP-4 in mediating the effect of senescent cells on other cells in proximity [165]. However, the role of DPP-4 inhibitor in influencing cellular senescence and/or cellular senescence-associated pathological effects in DKD needs to be elucidated by future studies.

Senotherapy is a treatment strategy to specifically target senescent cells (SNCs) in order to restore tissue integrity by mitigating the adverse effects associated with cellular senescence [166]. This strategy focuses on the blockade of SASP and the clearance of SNCs, the latter approach known as senolysis or senolytics [166]. Senolytics eliminate SNCs by targeting the pathways that are differentially activated in SNCs but not in non-SNCs, such as the anti-apoptotic pathway [166,167].

One of the classical senolytics is the combination of Dasatinib and Quercetin (D + Q). Dasatinib, a tyrosine kinase inhibitor, was approved for the treatment of leukemia by the FDA in 2006 [168]. Quercetin is a natural flavonoid targeting anti-apoptosis network members including BCL-xL and HIF1α [169]. Such an approach has been shown to have multiple beneficial effects, such as reduced inflammation and being an anti-oxidant. Kirkland's team has for the first time used D+Q as a therapy for senescence [170]. They creatively explored the Senescent Cell Anti‐Apoptotic Pathway (SCAP) networks to identify a potential drug target to eliminate SNCs. First, they confirmed that the inhibition of the SCAP network by siRNA could effectively promote apoptosis in SNCs without toxicity to the non-SNCs. They used bioinformatic approaches to identify D+Q as a promising candidate from 46 potentially senolytic drugs and then demonstrated D+Q as an efficacious agent for SNC elimination in several cell populations. In a Phase II clinical study (NCT02848131), subjects with chronic kidney disease and diabetes were treated with D+Q leading to decreased senescent cells in adipose tissue and skin [171]. The study is still ongoing with the renal outcome yet to be reported.

Furthermore, in a preclinical study, the heat shock protein 90 (HSP90) inhibitor (alvespimycin), a senolytic, reduces kidney senescence in diabetic mice with bilateral renal ischemia–reperfusion injury [172]. As described above, oxidative stress is an important characteristic and inducer of cellular senescence with redox pathways involved in senescence modulation in DKD. Although there is no direct evidence showing that antioxidants retard senescence in DKD, targeting the key ROS producers in the diabetic mouse kidney has been shown to reduce renal injury in DKD as a result of reduced oxidative stress [88,89]. Epigallocatechin gallate (EGCG) is an antioxidant compound in green tea, and has been previously shown to prevent oxidative stress-induced cellular senescence in human mesenchymal stem cells, with this action mediated by Nrf2 [173,174]. The EGCG-Nrf2 axis has been demonstrated to play a key role in reducing renal oxidative stress, inflammation and fibrosis as well as albuminuria in experimental DKD [175].

The incidence of DKD is strongly associated with the duration of diabetes, indicating that a prolonged and complex process of molecular and cellular changes plays a causal role in initiating DKD, a condition of progressive decline in kidney function as a result of progressive tissue injury within the kidney. A number of factors as a result of hyperglycemia and/or insufficient intracellular insulin signaling due to insulin resistance or insulin deficiency, which are the major features of diabetes, have an adverse impact on renal cell health, and can cause cell death in the kidney. Damaged kidney tissue needs to be repaired in order to maintain tissue homeostasis and renal function. Decreased reparative capability is likely to be a key mechanism driving the development and progression of DKD with increasing renal injury. Indeed, increased senescent renal cells are observed in DKD, including CD24⁺/CD133⁺STCs which are responsible for repairing tissue injury in the kidney. These senescent cells not only lose their ability to proliferate in order to participate in the tissue repairing process, but also secrete proinflammatory and profibrotic cytokines which can also significantly impact on other cells leading to inflammation and fibrosis. Therefore, the onset of DKD occurs as a result of concurrent continuous injuring of the kidney by diabetes and cellular senescence induced impairment of tissue repair. This leads to increasing loss of functional renal tissue. Without targeting the senescent cells, the key culprit of the imbalance of tissue homeostasis, continuous loss of kidney tissue and renal function leads to progression of DKD to kidney failure. Interestingly, most of the drugs with renoprotective effects have been shown to counteract the adverse effects of senescent cells in DKD. Furthermore, Senotherapeutics, the therapeutic strategy to specifically target senescent cells is emerging with some agents being evaluated in clinical trials exploring renal outcomes in diabetes.

The detailed regulation of cellular senescence at a mechanistic level in DKD is yet to be fully elucidated. Further research in this field to understand the interaction or crosstalk among the various pathways related to key pathophysiological responses in diabetes, such as oxidative stress, autophagy, ER stress, cell death, and the kidney repair process, is warranted. In addition, the impact of senescence on various specific renal cell population, such as tubular cells, podocytes, and immune cells, are worthy further study.

Although the underlying mechanisms need further investigation, current therapeutic approaches for senescent cell elimination appear to show kidney benefits. Therefore, an optimized therapeutic strategy to target senescent cells in the kidney represents a novel, promising and potentially effective treatment for DKD patients.

Data sharing is not applicable to this review paper.

The authors declare that there are no competing interests associated with the manuscript.

Open access for this article was enabled by the participation of Monash University in an all-inclusive Read & Publish agreement with Portland Press and the Biochemical Society under a transformative agreement with CAUL.

Yuehan Wei: Conceptualization, Data curation, Writing—original draft, Writing—review & editing. Shan Mou: Writing—review & editing. Qing Yang: Writing—review & editing. Fang Liu: Writing—review & editing. Mark E. Cooper: Supervision, Writing—original draft, Writing—review & editing. Zhonglin Chai: Conceptualization, Resources, Data curation, Supervision, Validation, Writing—original draft, Project administration, Writing—review & editing.

3-MA

3-methyladenine

β-Gal

β-galactosidase

AGE

advanced glycation end product

CDA1

cell division autoantigen 1

DCR2

Decoy receptor 2

DDR

DNA damage response

DPP-4

dipeptidyl peptidase 4

DKD

diabetic kidney disease

EGCG

epigallocatechin gallate

ESRD

end-stage renal disease

ER

endoplasmic reticulum

GLP-1

glucagon-like peptide-1

HSP90

heat shock protein 90

PRDX1

peroxiredoxin 1

PTEC

proximal tubular epithelial cell

RTEC

renal tubular epithelial cell

SASP

senescence-associated secretory phenotype

SA-β-Gal

senescence-associated β-galactosidase

SCAP

senescent cell anti‐apoptotic pathway

SGLT2i

sodium-glucose cotransporter-2 inhibitor

SNC

senescent cell

T2D

Type 2 diabetes

TGFβ

transforming growth factor-β

TRAIL

tumor necrosis factor-related apoptosis-inducing ligand