Abstract

   Emerging evidence highlights cellular senescence’s pivotal role in
   chronic kidney disease (CKD). Proximal tubule epithelial cells (PTECs)
   and fibroblasts are major players in CKD and serve as cellular sources
   of senescence. The generation of a conditionally immortalized human
   kidney cell model would allow to better understand the specific
   mechanisms and factors associated with cellular senescence in a
   controlled setting, devoid of potential confounding factors such as age
   and comorbidities. In addition, the availability of human kidney cell
   lines for preclinical research is sparse and most cell lines do not
   reflect their in vivo counterparts due to their altered behavior as
   immortalized cancer-like cells. In this study, PTECs and fibroblasts
   from human kidneys were isolated and transduced with
   doxycycline-inducible simian virus 40 large T antigen (SV40LT) vector.
   By comparing their gene expression with single-cell RNA sequencing data
   from human kidneys, the newly produced human kidney cell lines
   demonstrated significant resemblances to their in vivo counterparts. As
   predicted, PTECs showed functional activity and fibroblasts responded
   to injury with fibrosis. Withdrawal of the immortalizing factor
   doxycycline led to p21^+ cell-cycle arrest and the key hallmarks of
   senescence. The obtained senescence gene set largely overlapped between
   both cell lines and with the previously published SenMayo set of
   senescence-associated genes. Furthermore, crosstalk experiments showed
   that senescent PTECs can cause a profibrotic response in fibroblasts by
   paracrine actions. In 76 human kidney sections, the number of p21^+
   cells correlated with the degree of fibrosis, age and reduced
   glomerular filtration, validating the role of senescence in CKD. In
   conclusion, we provide a novel cellular ex vivo model to study kidney
   senescence which can serve as a platform for large scale compounds
   testing.

   Subject terms: End-stage renal disease, Senescence

Introduction

   Chronic kidney disease (CKD) is a growing public health problem
   affecting more than 10% of people worldwide [[56]1]. Accumulating
   evidence indicates that cellular senescence, an irreversible situation
   involving cell-cycle arrest, contributes to CKD [[57]2]. In particular,
   proximal tubule epithelial cells (PTECs) are the major location of
   senescent cells after kidney injury or during aging [[58]3]. Senescent
   cells obtain a senescence-associated secretory phenotype (SASP) and
   secrete numerous proinflammatory and profibrotic cytokines that can
   adversely affect neighboring healthy cells [[59]4]. It is likely that
   the SASP in injured PTECs is a key mechanism through which fibroblast
   activation drives kidney fibrosis in CKD [[60]5–[61]8]. As interstitial
   fibrosis correlates with kidney function and preclinical data suggest
   that targeting fibrosis can stabilize kidney function, fibrosis is
   widely accepted as a therapeutic target for CKD [[62]9]. Therefore,
   investigating the role of PTECs and fibroblasts in cellular senescence
   and the crosstalk between these two cell types will be highly important
   for understanding SASP-mediated kidney fibrosis. Moreover, senescent
   cells in the kidney may represent novel targets for preventing the
   progression of CKD [[63]10]. However, robust human kidney in vitro cell
   models for senolytic drug screening are lacking, as most cell lines do
   not recapitulate their in vivo counterparts due to their altered
   behavior as immortalized cancer-like cells.

   In the present study, we aimed to generate a conditionally immortalized
   cell model by transducing a doxycycline-inducible simian virus 40 large
   T antigen (SV40LT) vector into human PTECs and kidney fibroblasts.
   SV40LT is commonly introduced into primary cells to establish an
   immortalized cell line [[64]11]. The T antigen binds and inhibits
   several proteins, including the tumor suppressor p53 and the
   retinoblastoma suppressor gene (pRB), to promote cell division
   [[65]12]. Creating a conditional model enables the proliferation of
   cells under the influence of doxycycline-induced T antigen expression,
   maintaining a state of conditional immortality. Additionally, this
   approach provides the opportunity to investigate cells that will enter
   the senescence process upon the removal of doxycycline, as these cells
   exhibit characteristics similar to those of primary cells. This cell
   model would also enable the investigation of cellular senescence within
   a controlled environment, eliminating potential confounding factors
   such as age and comorbidities. Here, we demonstrate that the newly
   generated human kidney cell lines indeed exhibit significant
   similarities to their in vivo counterparts, as evidenced by comparisons
   with publicly available human kidney single-cell data from the Kidney
   Precision Medicine Project (KPMP) [[66]13] and by the use of functional
   assays. Furthermore, the withdrawal of doxycycline led to the
   expression of a senescence signature similar to the SenMayo consensus
   signature [[67]14] in both PTECs and fibroblasts. We were able to model
   the cellular crosstalk between senescent PTECs and fibroblasts,
   indicating that senescent PTECs can cause a profibrotic response in
   fibroblasts through paracrine actions. Thus, we present a newly
   developed scalable platform for investigating kidney senescence. This
   platform holds promise for extensive compound screening initiatives
   aiming to discover novel senolytics for the treatment of CKD on a large
   scale.

Results

Induction of cellular senescence in conditionally immortalized human kidney
CD10^+ PTECs and PDGFRβ^+ fibroblasts

   Conditionally immortalized human kidney CD10^+ PTECs and PDGFRβ^+
   fibroblast lines were generated using the pTet-SV40LT vector (Fig.
   [68]1A, Supplementary Fig. [69]1A). SV40LT mRNA expression was tightly
   controlled by doxycycline in both cell lines (Fig. [70]1B, C). As
   expected, the growth of CD10^+ PTECs and PDGFRβ^+ cells was attenuated
   by the removal of doxycycline, while the resupplementation of
   doxycycline increased the expression of SVLT40 and proliferation
   (Supplementary Fig. [71]1B, C). This observation indicated that the
   block of proliferation was reversible in our cell lines. These findings
   were supported by the results of the cell proliferation WST-1 assay
   (Fig. [72]1D, E).

Fig. 1. Generation of doxycycline (Dox)-inducible SV40LT in human kidney
CD10^+ proximal tubule epithelial cells (PTECs) and PDGFRβ^+ fibroblasts.

   [73]Fig. 1
   [74]Open in a new tab

   A Schematic representation of the generation of dox-inducible SV40LT in
   cell lines from human kidneys using the pTet-SV40 vector. Figure was
   created with BioRender. B, C Relative gene expression of SV40LT in
   CD10^+ PTECs and PDGFRβ^+ cell lines treated with or without
   doxycycline for 7 days. ***P < 0.001, ****P < 0.0001, n = 3. Gene
   expression was measured using RT-qPCR and visualized as the mean fold
   change. P values were determined by a two-tailed unpaired t test. D, E
   WST-1 assay in both cell lines treated with or without doxycycline for
   7 days; n = 5–9. ****P < 0.0001, two-tailed unpaired t test. Mean
   values are shown ± SD.

   This cell model allows the expansion of cells far beyond their regular
   life span through conditional immortalization. As we chose aged
   individuals to generate these cell lines (85 years old for PTECs and 74
   years old for PDGFRb^+ fibroblasts), we hypothesized that withdrawal of
   the immortalizing factor would return the aged cells to a primary state
   of cellular senescence characterized by shortened telomeres,
   SA-ß-galactosidase activity, and other relevant features. Indeed, when
   we measured telomere length in both cell lines (Supplementary Fig.
   [75]2), we found critically shortened telomeres, nearing the Hayflick
   limit, which is the threshold at which a cell can no longer undergo
   further cell division. Furthermore, by removing doxycycline from the
   cell culture medium, we were also able to induce cellular senescence in
   both cell lines, as evidenced by increased SA-β-gal activity in
   enlarged cells (Fig. [76]2A, B).

Fig. 2. The cell lines matched the respective cell types according to the
human single-cell RNA sequencing data and exhibited senescent signatures upon
doxycycline withdrawal.

   [77]Fig. 2
   [78]Open in a new tab

   Representative phase-contrast microscopy images (A) and SA-β-gal
   staining (B) of both cell lines treated with or without doxycycline for
   7 days. Scale bars, 90 μm in (A) and 200 μm in (B). n = 4. C PCA
   dimensions of both cell lines treated with and without doxycycline for
   7 days. The data were obtained from bulk RNA-seq. D–G Running mean
   expression (window size = 10) of the top 3 ranked overlapping marker
   genes for each cell type is shown for CD10^+ PTECs and for PDGFRβ+
   cells, with and without doxycycline. Irrespective of the doxycycline in
   the cell culture medium, the KPMP data of CD10^+ PTECs and PDGFRβ+
   fibroblasts showed the highest overlap with human proximal tubules and
   fibroblasts, respectively. H, I Hallmark pathways associated with the
   normalized enrichment score (NES) determined via gene set enrichment
   analysis (GSEA) of CD10^+ PTECs and PDGFRβ^+ cells treated with or
   without doxycycline for 7 days. A normalized score greater than or less
   than 0 represented positively or negatively associated pathways,
   respectively. n = 3 each. *P ≤ 0.05. J, K Heatmap showing the
   differentially regulated genes (adjusted p values ≤ 0.05 and log2-fold
   change >1 or < 1). Color scales represent after-scaling expression
   data. Venn diagram (L) and KEGG pathway analysis (M, N) of common up-
   and downregulated DEGs and pathways across both cell lines.

Bulk transcriptome analysis of conditionally immortalized human kidney CD10^+
PTECs and PDGFRβ^+ fibroblasts

   We next performed bulk RNA sequencing (RNA-seq) to dissect mRNA
   expression alterations in both cell lines in response to doxycycline
   withdrawal, placing them in a sudden senescent cell state. Principal
   component analysis (PCA) indicated that the replicates clustered
   together with strong differences based on cell identity (epithelial vs.
   mesenchymal) and a strong effect of senescence (doxycycline vs.
   doxycycline withdrawal; Fig. [79]2C). On the basis of gene expression,
   our cell lines displayed notable resemblances to their in vivo
   counterparts based on comparisons with publicly accessible human kidney
   single-cell data from the Kidney Precision Medicine Project (KPMP)
   (Fig. [80]2D–G). Thus, our CD10^+ PTECs were best matched with proximal
   tubule cells according to the KPMP data (Fig. [81]2D, E). Similarly,
   for PDGFRβ+ cells, the best matching cell type according to the KPMP
   data was interstitial fibroblasts (Fig. [82]2F, G).

   Next, we assessed the pathways associated with doxycycline withdrawal
   in our cell lines. Gene set enrichment analysis (Hallmarks) revealed
   enrichment of NF-κB, hypoxia, complement, inflammatory response, P53,
   IL-2 and IL-6-STAT signaling in senescent cells (-dox 7D) compared to
   nonsenescent cells (Fig. [83]2H, I). In contrast, pathways related to
   cell proliferation and DNA replication were downregulated in senescent
   cells (Fig. [84]2H, I). The expression of the senescence marker gene
   CDKN1A (p21) was significantly upregulated in both cell lines upon
   doxycycline withdrawal, while Lamin B1 expression was downregulated
   (Fig. [85]2J, K, Supplementary Table [86]S2). In addition, upon
   doxycycline withdrawal, the expression of genes related to inflammation
   (IL1A, IL6 and TNFα), fibrosis (VEGF and CTGF), chemotaxis (CXCL1,
   CXCL2, CXCL3 and CXCL8), and the complement system (C3) was markedly
   increased in both cell lines, while the expression of G2/M and mitosis
   genes, such as CDC25C and CDK1, was decreased in senescent cells (Fig.
   [87]2J, K). These observations were confirmed by RT‒qPCR (Supplementary
   Figure [88]3A‒D). There was a notable convergence of differentially
   regulated genes (DEGs) in the same direction across both cell lines
   (Fig. [89]2L), suggesting a common response in cellular senescence.
   KEGG pathway analysis of the common DEGs revealed increased TNF-α and
   p53 signaling in senescent cells, while pathways associated with the
   cell cycle and DNA replication were downregulated (Fig. [90]2M, N).

   We used DoRothEA to predict transcription factor activities and found
   that JUN, TP53, RELA, and NFKB1 were increased during senescence in
   both cell lines (Fig. [91]3A, B). In addition, increased cell
   type-specific transcription factor activities were found in PTECs (SP3,
   STAT2, and KLF5) and PDGFRβ^+ cells (TCF12, RUNX2, and IRF4) (Fig.
   [92]3A, B). We also inferred the importance of cell signaling via
   PROGENy and observed that P53, TGFβ, NF-kB, WNT, and TNF pathway
   activity increased during the cellular senescence process in both cell
   lines (Fig. [93]3C, D). In contrast, the PI3K, estrogen and hypoxia
   pathways were downregulated in senescent cells. In addition, we found a
   large overlap with the publicly available SenMayo senescence panel of
   125 genes, which has been described to predict aging and senescence
   across tissues and species (Fig. [94]3E, Supplementary Table [95]S3)
   [[96]14]. In summary, these findings suggest that our two novel cell
   lines can acquire a conserved senescence phenotype.

Fig. 3. Pathway/transcription factor activities in human kidney CD10^+ PTECs
and PDGFRβ^+ cell lines with doxycycline-inducible SV40LT expression and
staining for p21 in human kidneys.

   [97]Fig. 3
   [98]Open in a new tab

   A, B Transcription factor (TF) activity as estimated by the DoRothEA
   algorithm across senescent cells (7 days without doxycycline) and
   nonsenescent cells (7 days with doxycycline) from the human CD10^+ PTEC
   and PDGFRβ+ fibroblast bulk RNA-seq dataset. Color scales represent
   after-scaling expression data. C, D Pathway activity as estimated by
   the PROGENy algorithm in senescent cells versus nonsenescent cells
   using bulk RNA-seq data. Color scales represent after-scaling
   expression data. E Venn diagram of the differentially expressed genes
   (adjusted p ≤ 0.05 and log2-fold change >1 or <1) in senescent CD10^+
   PTECs/PDGFRβ^+ cells and the “SenMayo” senescence gene set. F
   Comparison of significantly differentially expressed genes (DEGs)
   (adjusted p value < 0.01) between CD10^+ PTECs treated with doxycycline
   (nonsenescent) and those not treated with doxycycline (senescent) and
   between PTECs with temperature-sensitive SV40LT on day 9 at 37 °C
   (senescent; [[99]15] referred to as day9noIS) and cells at 33 °C on day
   0 (immortalized cells, referred to as day0noIS) revealed a Spearman
   correlation coefficient of 0.54. G Immunohistochemical staining of p21
   in healthy and fibrotic human kidneys (n = 76). Scale bars, 100 μm. H,
   I Spearman correlations between the p21^+ cell population/area and
   estimated glomerular filtration (n = 74) or age (n = 73). J, K
   Immunofluorescence costaining of p21 and LTA and quantification of the
   proteins in human kidneys (n = 5 for healthy kidney, n = 3 for fibrotic
   kidney). Mean values are shown ± SD.

   To rule out the possibility that alterations in gene expression after
   doxycycline withdrawal result from direct doxycycline effects rather
   than downstream effects of SV40LT, we compared our PTECs with a
   previously reported conditionally immortalized PTEC culture model in
   which SV40LT is regulated by temperature [[100]15] (Fig. [101]3F).
   Spearman’s correlation between all common DE genes and statistically
   significant DE genes (adjusted p value < 0.01) was 0.38 and 0.54,
   respectively. This finding shows a similar signature between our
   senescent PTECs and previously published PTECs, indicating that the
   changes observed in our model are unlikely to be attributed to direct
   effects of doxycycline but rather to the downstream effects of SV40LT
   degradation. Additionally, we established another CD10^+ human PTEC
   line that was permanently immortalized with human telomerase reverse
   transcriptase (hTERT) and SV40LT. These non-conditionally immortalized
   cells were then cultured with or without doxycycline to examine the
   doxycycline-induced changes in the transcriptome (Supplementary Fig.
   [102]4A‒C). Compared to the cells treated with doxycycline, the cells
   without doxycycline exhibited increased gene expression related to
   coagulation, peroxisome function, cholesterol homeostasis, and
   proliferation pathways, while pathways associated with p53 and
   inflammation were downregulated (Supplementary Fig. [103]4A, B).
   Therefore, these changes did not reflect those observed in our
   conditionally immortalized PTECs upon doxycycline withdrawal. This is
   further emphasized by the poor overlap of DEGs between the two PTEC
   lines (Supplementary Fig. [104]4C). Overall, this indicates that the
   alterations observed in our conditionally immortalized model are not
   caused by doxycycline itself, but instead result from SV40LT
   degradation and the onset of senescence.

   To assess the significance of these findings within the context of CKD
   in humans, we analyzed p21 expression in tissue microarrays with kidney
   cores from 76 patients. Compared to that in healthy kidney tissue, p21
   expression was increased in fibrotic tissue and correlated with lower
   kidney function and higher age (Fig. [105]3G–I). Coimmunostaining of
   p21 with the PTEC marker LTA revealed increased p21 expression in LTA^+
   PTECs, LTA^− tubular epithelial cells (degenerative PTECs and other
   tubule segments) and interstitial cells (fibroblasts and immune cells)
   in fibrosis (Fig. [106]3J, K). These results suggest that p21
   expression may represent a clinically relevant senescence marker in
   CKD.

Functional analysis of conditionally immortalized CD10^+ PTECs and PDGFRβ^+
fibroblasts

   To evaluate whether transporter proteins known to be expressed in
   proximal tubules [[107]16, [108]17] are functionally expressed in our
   human kidney CD10^+ PTEC line, the efflux protein inhibitors PSC833 and
   MK571 were used to block the cell transporter activity of P-gp and
   MRP4, respectively. Using these inhibitors, we detected increased
   fluorescence intensity of corresponding metabolites of calcein-AM and
   CMFDA, indicating the presence and activity of these transporters in PT
   cells (Fig. [109]4A, B). Based on these findings, we postulate that our
   PTEC line exhibit functional activity comparable to that of in vivo
   proximal tubule cells.

Fig. 4. Functional analysis of human kidney CD10^+ PTECs and PDGFRβ^+ cell
lines with doxycycline-inducible SV40LT expression and crosstalk between
PTECs and fibroblasts.

   [110]Fig. 4
   [111]Open in a new tab

   A, B Fluorescence intensity of CD10^+ PTECs with and without efflux
   transport blockers. The cells were first incubated with the substrates
   calcein-AM (1 μM) and CMFDA (1.25 μM), after which the efflux transport
   blockers PSC833 (5 μM) and MK571 (5 μM) were added. n = 9–18,
   **P < 0.01, ****P < 0.0001, two-tailed unpaired t test. C, D Gene
   expression of COL1A1 and FN1 in human kidney PDGFRβ^+ cells exposed to
   IL-1β (100 ng/ml) or TGFβ (10 ng/ml). *P ≤ 0.05, **P < 0.01,
   ****P < 0.0001; two-tailed unpaired t test. N = 6. E Cytokine arrays
   incubated with pooled supernatant from senescent (−dox) and
   nonsenescent (+dox) PTECs (n = 3 per condition was pooled and added to
   one membrane). F Quantification of (E). G Mass spectrometry signal of
   MMP-7 in cell culture supernatant from nonsenescent (dox+) and
   senescent (dox−) PTECs. N = 4. H Schematic representation of the
   coculture experiments. F Supernatants from day 5 senescent CD10^+ PTECs
   were transferred to PDGFRβ^+ cells for 3 days. I Gene expression of
   ACTA2, COL1A1, and FN1 in PDGFRβ+ cells. *P ≤ 0.05, two-tailed unpaired
   t-test. N = 5–6. Mean values are shown ± SD. J Immunofluorescence
   staining of αSMA and Col1a1 in PDGFRβ^+ cells. Scale bars, 50 μm.
   N = 3.

   Next, we tested whether PDGFRβ^+ cells respond to increased expression
   of fibrosis-related genes upon stimulation with the proinflammatory,
   profibrotic stimuli IL1β and TGFβ. As predicted, RT-qPCR revealed
   increased expression of typical myofibroblast-related genes (COL1A1 and
   FN1) after IL-1β and TGFβ stimulation compared to that in
   vehicle-treated cells (Fig. [112]4C, D). One hypothesis of kidney
   fibrosis is that cytokines related to the SASP in senescent PTECs
   induce fibrosis through the paracrine stimulation of PDGFRβ+. As our
   transcriptome data revealed the increased expression of several
   proinflammatory and profibrotic cytokines in senescent cells, we
   assessed its relative expression on the protein level using a cytokine
   array kit (Fig. [113]4E, F). As predicted from our transcriptome data,
   the supernatant of senescent PTECs showed a relative increase of CCL2,
   CXCL1, GM-CSF, IL-6 and IL-8 levels, compared to supernatant of
   nonsenescent PTECs (Fig. [114]4E, F). Most of these chemokines have
   been shown to play a role in kidney senescence and fibrosis [[115]3,
   [116]8, [117]18]. In contrast, CCL5 and MIF levels were lower in
   supernatant of senescent PTECs compared to supernatant of nonsenescent
   PTECs. Both CCL5 and MIF play a role promoting inflammation, but also
   proliferation [[118]19, [119]20]. While one might expect higher levels
   of CCL5 and MIF due to its inclusion in the proinflammatory SASP, the
   reduced expression could also be aligned with the low proliferative
   state of senescent cells [[120]19, [121]20].

   As our transcriptome data showed increased gene expression of MMP-7 and
   we recently identified MMP-7, released primarily by PTECs, as biomarker
   for kidney disease progression [[122]21], we also used mass
   spectrometry to measure MMP-7 levels in the cell culture supernatant of
   PTECs. This revealed increased expression of MMP-7 levels upon the
   induction of senescence, which is consistent with a previous study
   reporting increased expression of MMP-7 in aging kidneys [[123]22]
   (Fig. [124]4G).

   Next, we transferred the supernatant from our senescent CD10^+ PTECs to
   PDGFRβ^+ fibroblasts cultured with doxycycline (Fig. [125]4H). The
   supernatant was mixed with fresh medium at a 1:1 ratio. Control
   fibroblasts received supernatant from CD10^+ PTECs cultured in
   doxycycline-containing medium. RT‒qPCR revealed increased expression of
   myofibroblast differentiation/fibroblast activation marker genes
   (ACTA2, COL1A1, and FN1) in PDGFRβ^+ cells cultured in medium from
   senescent PTECs compared to fibroblasts cultured in medium from
   nonsenescent PTECs (Fig. [126]4I). Immunofluorescence staining of
   PDGFRβ+ cells for alpha smooth muscle actin (αSMA, encoded by ACTA2)
   and collagen 1 alpha 1 (Col1a1) further supported these mRNA expression
   results at the protein level (Fig. [127]4J). Thus, here, we provide
   proof that senescent human PTECs can drive fibroblast activation and
   myofibroblast differentiation in a paracrine manner through the SASP.

Discussion

   In this study, we isolated and conditionally immortalized human PTECs
   and kidney fibroblasts with doxycycline-inducible SV40LT expression. By
   comparing the gene expression data with human kidney single-cell RNA
   sequencing data, we showed that these cell lines retained substantial
   similarity to their in vivo counterparts. We demonstrated that this
   model enables reversible immortalization of cells for expansion, while
   withdrawal of the immortalization factor immediately initiates a
   senescent cell state characterized by cell-cycle arrest and key
   hallmarks of senescence. The obtained gene set exhibited significant
   overlap with the publicly accessible SenMayo panel comprising 125
   senescence-related genes [[128]14] recognized for their expression
   during both aging and senescence across various tissues and species.
   Furthermore, we provided proof that senescent PTECs drive fibroblast
   activation through paracrine actions. Overall, this conditionally
   immortalized cell culture model can be employed for the examination of
   cellular senescence and may serve as a potential platform for screening
   senolytic drugs to restore kidney function.

   Primary cells have a limited capacity for division in culture,
   primarily due to the progressive shortening of telomeres with each cell
   division. By expanding and subsequently withdrawing the immortalizing
   expansion stimulus, we observed that the cells were propelled into a
   future state marked by shortened telomeres, triggering the onset of
   cellular senescence. SV40LT [[129]23] or hTERT [[130]24, [131]25] are
   commonly used to overexpress SV40LT in immortalized cells. In our
   pursuit of conditional immortalization, we reasoned that the
   immortalizing factor must remain entirely inactive when uninduced and
   should be capable of rapid and uniform induction to attain sufficient
   expression levels in all cells. Previous studies used
   temperature-sensitive SV40LT to generate conditionally immortalized
   glomerular endothelial cells [[132]26], podocytes [[133]27] or proximal
   tubule cells [[134]28]. In these models, cells proliferate at a
   permissive temperature of 33 °C but cease growth at 37–39 °C. However,
   this temperature shift can influence various cellular functions and is
   not particularly practical. In addition, a large T often does not
   degrade entirely at increased temperatures, necessitating even higher
   temperatures, which affect various cellular processes. Here, we used an
   alternative approach and controlled cell proliferation or arrest by
   tightly controlling the doxycycline-inducible expression of SV40LT at a
   consistent temperature of 37 °C. We showed that our system efficiently
   regulates the expression of SV40LT: in the absence of doxycycline, no
   SV40LT expression was observed, while doxycycline administration caused
   robust SV40LT expression with dramatic changes in cell proliferation,
   morphology and gene expression. Importantly, the observed alterations
   were a result of the downstream effects of SV40LT and not attributable
   to doxycycline itself. This was evident in the comparison with
   temperature-sensitive SV40LT-expressing PTECs, which exhibited similar
   transcriptomic changes upon induction of senescence.

   Both PTECs and fibroblasts are two key players in kidney senescence and
   fibrosis [[135]29]. By using bulk RNA sequencing, we were able to
   reveal the senescence gene sets in PTECs and fibroblasts. In the
   absence of doxycycline, both cell lines expressed common senescence
   markers and exhibited senescence-related morphological alterations,
   consistent with the observations in CKD [[136]30]. In particular, we
   found robust upregulation of the expression of the CDKN1A (p21) gene,
   which is a main driver involved in cell-cycle arrest during senescence
   [[137]31]. In contrast, induction of cellular senescence causes a
   reduction in LMNB1 expression, a widely recognized senescence feature
   that is linked to impaired nuclear integrity, resulting in the
   enlargement and irregular shape of cells [[138]32]. We noted an
   enrichment of genes known to play a role in vivo in failed proximal
   tubule cell repair, senescence, inflammation and fibrosis, including
   NFKB1, TNF, STAT3, VCAM1 and RELA [[139]14, [140]33]. Another hallmark
   of cellular senescence is the secretion of cytokines, chemokines,
   proteinases, and other SASP components, which are involved in several
   biological processes and contribute to tissue dysfunction in aging and
   disease [[141]32, [142]34]. As such, upon senescence induction, we
   found increased gene expression of the proinflammatory genes IL1, IL6,
   and CCL2 in both cell lines. To a large extent, the SASP is a
   transcriptional program mediated by nuclear factor-kB (NF-kB), which
   drives the gene expression of IL1 and TNFA [[143]35, [144]36]. IL6,
   IL10, and TGFβ are also upregulated through other pathways, such as the
   JAK2/STAT3 pathway [[145]37, [146]38]. We also found increased
   expression of connective tissue growth factor (CTGF) in senescent
   cells. CTGF is a key downstream mediator of TGFβ signaling and has been
   demonstrated to regulate profibrotic effects in the kidney [[147]39,
   [148]40].

   Cellular senescence is a complex process that involves multiple
   signaling pathways. We found decreased PI3K pathway activity and
   estrogen and hypoxia pathway activity in senescent cells. PI3K
   signaling is essential for cell growth and proliferation. It has been
   shown earlier that PI3K/AMPK/mTOR activity is reduced during the
   senescence of irritated endothelial cells [[149]41], leading to
   decreased levels of the downstream targets EIF4E, RPS6, and EIF3 and
   many ribosomal proteins relevant for cell growth and protein
   translation. Furthermore, PTEC senescence can be induced by the
   inhibition of AMPK–mTOR signaling [[150]42]. Similarly, the estrogen
   pathway promotes cell growth, and recent research has suggested that
   cellular senescence may underlie sex differences [[151]43]. Estrogen
   can suppress the release of reactive oxygen species (ROS), thereby
   preventing ROS-induced DNA damage. In addition, estrogen suppresses p21
   expression, SASP factor expression and the JAK/STAT pathway [[152]44]
   and is important for telomere maintenance and autophagy. Consistent
   with the findings of previous reports, we also found increased
   expression of HIF1A in senescent cells, which accumulates under hypoxic
   conditions and regulates cellular senescence by affecting p21, p53, and
   lamin B1 levels [[153]45]. In contrast to these three pathways, we
   found increased activity of several pathways (JAK-STAT, p53, MAPK,
   NF-kB, TNFalpha, TGFbeta, and Wnt), which are involved in senescence
   [[154]31]. Overall, our results indicate that our conditional cell
   culture model can recapitulate the hallmarks of cellular senescence.

   Interactions between epithelial cells and mesenchymal components,
   specifically fibroblasts and pericytes, have been recognized as
   hallmarks of CKD [[155]7]. In nearly every instance of kidney injury,
   PTECs sustain damage, whether secondary to glomerular injury-induced
   proteinuria or direct toxic or hypoxic injury. Consequently, we propose
   that the induction of a senescent phenotype in PTECs could lead to the
   subsequent activation of fibroblasts through the SASP. There is some
   evidence from murine models that crosstalk between PTECs and
   fibroblasts via paracrine actions contributes to the pathophysiological
   processes associated with cellular senescence and kidney fibrosis
   [[156]46]. Indeed, we identified several proinflammatory and
   profibrotic proteins, known to be components of the SASP, in the
   supernatant of senescent PTECs, and demonstrated that human senescent
   PTECs drive fibroblast activation in a paracrine manner. Furthermore,
   the number of senescent PTECs correlated positively with the severity
   of kidney fibrosis and negatively with kidney function in a human
   cohort of >70 patients.

   In summary, our study offers a novel conditional senescence cell
   culture model for studying cellular senescence in the human kidney.
   These cell lines may serve as valuable tools for potential therapeutic
   target identification and drug testing and provide evidence that
   paracrine signaling in senescent PTECs might be involved in kidney
   fibrosis and CKD.

Materials and methods

Ethics and Human Kidney Tissue Collection

   The present study was approved by the local ethics committee of the
   University Hospital RWTH Aachen (EK-016/17). Unaffected healthy kidney
   tissues were obtained from patients who underwent nephrectomy for
   kidney cancer (distant from any cancer as confirmed by a blinded
   pathologist). Written informed consent was obtained from all patients,
   and the study was performed in accordance with the Declaration of
   Helsinki. The work at biosafety level S2 was approved by the German
   authorities (reference number 53.05-01-K-21-093).

Isolation of primary CD10^+ PTECs and PDGFRβ^+ fibroblasts

   PTECs can be marked as clusters of 10 differentiated (CD10^+ cells),
   and our recent single-cell atlas of human kidney fibrosis suggested
   that PDGFRβ^+ fibroblasts are the main cellular source of scar-forming
   myofibroblasts [[157]5]. Thus, in this study, we sorted CD10^+ PTECs
   and PDGFRβ^+ fibroblasts as described below. We aimed to include kidney
   tissue from individuals over 65 years old with a eGFR of less than
   60 ml/min/1.73 m². These inclusion criteria were chosen to isolate
   cells with already shortened telomeres due to aging and CKD.

   Kidney tissue (85 years old, female, eGFR: 30 ml/min/1.73 m^2) was
   mechanically dissociated into 0.5–1 mm^3 pieces, and a single-cell
   suspension was obtained by using gentle-MACS (Miltenyi Biotec, Bergisch
   Gladbach, Germany) as described previously [[158]5]. The cells were
   first incubated with Fc-Block (Trustain anti-human, 1:50; Biolegend,
   Cat#422302; RRID: AB_2818986) for 30 min and then stained with
   anti-CD10 (clone HI10a, 1:40; Biolegend Cat#312214; RRID: AB_2146548),
   anti-CD45 (clone HI30, 1:40; Biolegend Cat#304006; RRID: AB_314394),
   and anti-CD31 (clone WM59, 1:40; Biolegend, Cat#303106; RRID:
   AB_314332) antibodies for another 30 min at room temperature. DAPI
   (#D9542, Sigma) was used for dead cell staining.
   CD10^+/CD31^−/CD45^−/dapi^- PTECs were sorted using a SONY SH800 sorter
   (Sony Biotechnology, San Jose, CA, USA) (Supplementary Fig. [159]S1).

   To isolate PDGFRβ^+ fibroblasts from the kidney (74 years old, male,
   eGFR: 59 ml/min/1.73 m^2), single-cell suspensions were subjected to
   magnetic-activated cell sorting (MACS) according to the manufacturer’s
   instructions. In brief, cells were stained with a human PDGFRβ^+
   antibody (Clone PR7212, 1:100; R&D Systems, Cat#MAB1263; RRID:
   AB_2162792), followed by magnetic separation using anti-mouse IgG1
   microbeads (Miltenyi Biotec, Cat#130-047-102; RRID: AB_244355) to
   obtain PDGFRβ^+ fibroblasts.

   Primary CD10^+ PTECs and PDGFRβ^+ fibroblasts were cultured in DMEM/F12
   and GlutaMAX^TM Medium (Gibco Cat# 31331026, Grand Island, New York,
   USA) supplemented with 10% fetal calf serum (PAA, Cat# A01125-499,
   Cölbe, Germany) and 100 U/ml penicillin‒streptomycin (Gibco) at 37 °C
   and 5% CO[2] with saturating humidity. All cell models were regularly
   tested for mycoplasma contamination.

Conditional immortalization of CD10^+ epithelial cells and PDGFRβ^+
fibroblasts

   Primary CD10^+ PTECs and PDGFRβ^+ fibroblasts were immortalized via
   lentiviral infection with SV40LT. Here, we generated a
   doxycycline-inducible lentiviral vector (pTet-SV40LT-Puro) expressing
   SV40LT using a Gibson assembly kit (Cat#E5510S, NEB, Frankfurt,
   Germany; Gibson et al., 2009). In brief, pCW-Cas9 (Cat# 50661, Addgene)
   [[160]47] was opened using NheI (Cat# R3131V, NEB) and KpnI (Cat
   #R3142V, NEB) as the backbone. SV40LT was amplified from pBABE-puro
   SV40LT (Cat #13970, Addgene) [[161]48], and the EF1α-Puro-rTetR-WPRE
   element was amplified from TLCV2 (Cat #87360, Addgene) [[162]49] by PCR
   using Q5 High-Fidelity DNA Polymerase (Cat #M0491V, NEB). For each cell
   line, 0.4 million primary cells were infected with lentiviral particles
   expressing SV40LT from five 6 cm dishes. HEK293T cells (0.8 million,
   70–80% cell confluence) were transected with 5 μg of pTet-SV40LT-Puro,
   2.5 μg of psPAX2 (Cat #12260, Addgene) or 2.5 μg of pMD2. G (Cat
   #12259, Addgene) per 6 cm dish by calcium phosphate precipitation
   [[163]50]. The virus particles were concentrated by chondroitin sulfate
   sodium salt (CSS) and polybrene precipitation [[164]51] as described
   previously [[165]52]. Primary CD10^+ PTECs and PDGFRβ^+ fibroblasts
   were infected with SV40LT lentiviral particles for 4 h, after which the
   medium was replaced with complete growth medium supplemented with
   10 μg/ml doxycycline (Cat #D9891, Sigma). The CSS precipitates were
   removed by daily washing with PBS, and the medium was changed. Two days
   after infection, 2 μg/ml puromycin (Cat #P8833, Sigma‒Aldrich) was used
   to select SV40LT-expressing cells for 7 days, and doxycycline was added
   to maintain SV40LT expression and maintain cell immortalization.

Measurement of telomere length using flow FISH

   Fluorescence in situ hybridization (flow-FISH) was carried out as
   described previously [[166]53]. Briefly, vital sterile frozen PTECs and
   fibroblasts treated with or without doxycycline were subjected to
   flow-FISH analysis of telomere length. Samples were prepared for cell
   denaturation and mixed with a fluorescein isothiocyanate (FITC)-labeled
   telomere-specific (CCCTAA) 3-peptide nucleic acid FISH probe
   (Eurogentec, Liège, Belgium) for DNA hybridization, followed by DNA
   counterstaining with LDS 751 (Sigma). Bovine thymocytes were used as
   internal controls. The predeterminate telomere length of bovine
   thymocytes was used to calculate the length in kilobases within PTECs
   and fibroblasts (Supplementary Fig. [167]2).

Cell proliferation assay

   Cell proliferation was monitored using a WST-1 assay (Cat #5015944001,
   Merck) according to the manufacturer’s instructions. In brief, 2 × 10^3
   CD10^+ or PDGFRβ^+ cells were seeded into a 96-well plate with or
   without doxycycline. The cells were incubated with WST-1 reagent for
   2 h at 37 °C and 5% CO[2] on day 7. The OD value was then determined
   using a spectrometer (CLARIOstar, Ortenberg, Germany) at 450–600 nm
   absorbance.

Senescence assay

   Senescence β-galactosidase (SA-β-Gal; Cat#9860S; Cell Signaling
   Technology, Boston, MA, USA) was used to detect SA-β-gal in senescent
   cells. In brief, 2 × 10^4 CD10^+ PTECs or PDGFRβ^+ cells were cultured
   in 12-well plates with or without doxycycline for 7 days and then
   stained with SA-β-Gal according to the manufacturer’s instructions.
   Images were acquired with an optical microscope (ZEISS1846, Oberkochen,
   Germany).

PDGFRβ^+ fibroblast stimulation with IL-1β and TGFβ

   Immortalized PDGFRβ^+ fibroblasts at 70% confluence were starved for
   24 h in DMEM containing 0.5% FCS. Then, IL1β (100 ng/ml; Cat# SRP3083,
   Sigma‒Aldrich) and TGFβ (10 ng/ml; Cat #100-21, PeproTech, Hamburg,
   Germany) were added to the medium. The medium was changed after 24 h,
   and the cells were harvested after 48 h of treatment.

Doxycycline treatment of non-conditionally immortalized CD10+ human proximal
tubule cells

   CD10^+ PTECs were isolated from the healthy kidney cortex of a
   55-year-old male nephrectomy specimen. To isolate PT cells, a
   single-cell suspension was prepared as described above, incubated with
   CD10-Microbeads (Miltenyi Biotec, #130-093-452), and separated using
   MACS technology (Miltenyi Biotec, autoMACS Pro Separator, #130-092-545,
   autoMACS Columns #130-021-101). The isolated primary PTECs were
   cultured in DMEM/F12 (1:1) with Glutamax (Gibco, #31331), supplemented
   with 10% FCS and 1% Pen/Strep for three weeks. To immortalize the PT
   cells, retroviral particles SV40LT and HTERT were employed. Retroviral
   particles were produced by transient transfection of HEK293T cells
   using TransIT-LT (Mirus). Two types of amphotropic particles were
   generated by co-transfecting plasmids pBABE-puro-SV40-LT (Addgene
   13970) or xlox-dNGFR-TERT (Addgene 69805) along with a packaging
   plasmid pUMVC (Addgene #8449) and a pseudotyping plasmid pMD2.G
   (Addgene 12259). Retroviral particles were concentrated 100-fold using
   Retro-X concentrator (Clontech) 48 h post transfection. Cell
   transduction was performed by incubating the target cells with serial
   dilutions of the retroviral supernatants (a 1:1 mix of concentrated
   particles containing SV40-LT or hTERT) for 48 h. Subsequently, the
   infected PT cells were selected using 2 μg/mL puromycin 72 h after
   transduction for 7 days. These immortalized CD10^+ PT cells were then
   cultured in 10 μg/ml doxycycline (Cat #D9891, Sigma) for 7 days to
   investigate the doxycycline-induced changes in the transcriptome.

RT-qPCR assay

   Total RNA was isolated using a RNeasy Mini Kit (Cat# 74106; Qiagen,
   Hilden, Germany) according to the manufacturer’s instructions. The RNA
   concentration was measured with a Nanodrop 2000 (Thermo Fisher,
   Waltham, MA, USA). Total RNA (1 μg) was reverse transcribed with a
   High-Capacity cDNA Reverse Transcription Kit (Cat# 43-688-13; Applied
   Biosystems, Foster, CA, USA). A real-time PCR system (Applied
   Biosystems) was used to perform RT-qPCR using SYBR Green fluorescence
   (Cat# 1725125, Bio-Rad, Feldkirchen, Germany). The primers used for
   RT-qPCR are listed in Supplementary Table [168]1. Human GAPDH was used
   for normalization of gene expression.

Bulk RNA sequencing and data analysis

   RNA was isolated as described above. rRNA was depleted within 200 ng of
   total RNA, and sequencing libraries were generated using the NEBNext
   Ultra II Directional RNA library kit (Cat#E7760L, NEB) according to the
   manufacturer’s protocol. In the initial preprocessing step, nextflow
   workflows (version 21.04.1) [[169]54] were applied, specifically the
   nf-core/rnaseq (version 3.1) pipeline [[170]55]. This involved the use
   of STAR (version 2.7.9a) [[171]56] for read alignment, Salmon (version
   1.5.0) [[172]57] for read quantification, TrimGalore (version 0.6.6)
   for read trimming, and GENCODE (version 38) for gene annotation
   [[173]57]. Subsequently, the generated count matrix file from Salmon
   was filtered, excluding genes labeled ‘Mt_tRNA,’ ‘rRNA,’ ‘Mt_rRNA,’ and
   ‘rRNA_pseudogene’ in the GENCODE annotation file. Additionally, genes
   associated with low expression were removed using HTSFilter (version
   1.32.0) [[174]58]. In the final step, DESeq2 (version 1.32.0) [[175]59]
   was used to identify DEGs from the filtered count matrix file for the
   specified comparisons. PCA was performed by using the glmpca package
   [[176]60]. For hierarchical clustering, the distance between each
   sample was calculated by the “dist” function with the “average” method
   in R. Afterwards, all the samples were clustered by “hclust” and
   visualized by the “gplots” package. The DEGs were ranked by the Wald
   statistic for pathway analysis. P values were adjusted for multiple
   testing using the Benjamini and Hochberg method. Genes and pathways
   with an FDR < 0.05 were considered significant. We used GSEA-preranked
   to test for enrichment of senescence genes in the phenotypes using the
   fgsea R package (v.1.14.0) [[177]61]. In the following heatmaps, a
   subset of significant genes was displayed with the “Pheatmap” R
   package. The PROGENy tool was used to estimate pathway activity in
   senescent and nonsenescent cells [[178]62]. The DoRothEa database was
   used to predict transcription factor-binding sites and infer regulatory
   networks [[179]63, [180]64]. Venn diagrams were generated with the
   “VennDiagram” package in R (version 4.2.1). The upregulated (log fold
   change ≥1 and adjusted p value ≤ 0.05) and downregulated (log fold
   change ≤ 1 and adjusted p value ≤ 0.05) genes in both CD10^+ PTECs and
   PDGFRβ^+ cells were chosen as the intersections in the Venn diagrams.
   The “ClusterProfiler” package in R (version 4.2.1) was used to perform
   the KEGG pathway enrichment analysis based on the 990 common
   upregulated and 1038 common downregulated genes. Pathways with adjusted
   p values < 0.05 were considered significant. Furthermore, the overlap
   between the previously published panel of 125 senescence genes
   [[181]13] and our two cell lines was compared.

Cell type comparison with publicly available human kidney single-cell
transcriptomics data

   Single-cell transcriptomics can be used to understand the cellular
   heterogeneity of complex biological systems such as the kidney. To
   estimate the similarity of our cell lines to their relevant cell types
   in vivo, we used publicly available single-cell transcriptomics data
   from the KPMP consortium (atlas.kpmp.org/repository). Briefly,
   overlapping differentially expressed genes were identified for each
   cell line and cell type in the single-cell data and ranked based on
   their average gene expression in that cell type. The cell types with
   the highest average gene expression were considered the best matches to
   our cells. To this end, KPMP cell type annotation data consisting of
   200338 single cells were obtained from KPMP.org. The raw count data
   were log-normalized using the normalize_total and log1p functions in
   Scanpy (version 1.9.2), and the marker genes for each cell type were
   calculated using the Wilcoxon method implemented in the
   rank_genes_groups function. Marker genes with adjusted p values < 0.01
   were used for downstream analyses. The overlap of these marker genes
   with the top 100 DE genes (logFC > 0 and adjusted p value < 0.05) for
   each cell line (CD10^+ PTECs vs PDGFRβ+ fibroblasts) obtained using
   DESeq2 was then used to identify the common genes. The running average
   (window size = 10) of the top overlapping genes with respect to the
   log2-fold change was then calculated for each cell type in the
   single-cell data.

Correlation with the publicly available conditionally immortalized PTEC model

   To rule out the possibility that alterations in gene expression after
   doxycycline withdrawal result from direct doxycycline effects rather
   than downstream effects of SV40LT degradation, we compared our PTECs
   with a previously reported conditionally immortalized PTEC culture
   model (ciPTEC-OAT1) in which SV40LT expression is regulated by
   temperature [[182]15]. In this model, p53 and p21 are activated at a
   permissive temperature (33 °C) through the expression of the
   temperature-sensitive mutant U19tsA58 of the SV40 large T antigen
   (SV40T). Moreover, the cells exhibit senescence-like arrest upon
   shifting to a nonpermissive temperature (37 °C). In their study, cells
   were cultured until 90% confluence at 33 °C and then transferred to
   37 °C with 5% (v/v) CO[2] to allow them to mature for 0, 1, 3 or 9
   days. For comparison with our RNA-seq data, we used day 0 at 33°C and
   day 9 at 37 °C, both without indoxyl sulfate treatment. To compare the
   dysregulated genes in the two cell lines, we calculated the
   differentially expressed genes in our CD10+ PTECs (doxycycline negative
   vs doxycycline treated) and proximal tubule cells (senescent vs
   immortalized) obtained from [[183]15].

Fluorescence-based functional assays of CD10^+ PTECs

   The functional activities of the ATP-dependent transporter
   P-glycoprotein (P-gp) and the organic anion transporter human multidrug
   resistance protein 4 (MRP4) in CD10^+ PTECs were evaluated using
   fluorescent substrates that accumulate intracellularly upon the
   inhibition of efflux transporters. Calcein-AM (Cat# 425201; BioLegend,
   San Diego, CA, USA) and CMFDA (5-Chlormethylfluorescein diacetate)
   (Cat# C7025; Thermo Fisher Scientific, Eugene, OR, USA) were used to
   evaluate transporter activity. Despite the significant substrate
   overlap among the transporters, their distinct binding properties
   enable different substrate specificities [[184]65]. Calcein-AM and
   CMFDA acquire fluorescent characteristics following hydrolysis
   (cleavage of methyl groups) in the cytoplasm. In the case of CMFDA,
   this process is followed by conjugation to
   carboxyfluorescein-glutathione, involving the replacement of the
   chlorine group. These resultant fluorescent metabolites were utilized
   to assess the activities of P-gp (calcein) and MRPs (CMFDA
   metabolites), respectively. The inhibitors MK571 (Cat# HY-19989A,
   MedChem Express, Monmouth Junction, NJ, USA) and PSC833 (Cat# HY-17384,
   MedChem Express) were used as known inhibitors of MRP4 and P-gp,
   respectively. The dilutions of the different inhibitors were varied
   starting at a concentration of 5 μM. To test for CMFDA metabolites
   retention, the cells were incubated with 1.25 μM CMFDA in combination
   with 5 μM MK571 for 30 min with doxycycline. Subsequently, the cells
   were washed with Krebs–Henseleit buffer (Cat# K3753, Merck), and
   fluorescence was immediately measured (wavelengths: excitation at
   492 nm; emission at 517 nm). Calcein retention was assessed by
   incubating the cells with 1 μM calcein-AM for 1 h in combination with
   the other inhibitor. Afterward, the cells were washed with
   Krebs–Henseleit buffer and lysed with 1% Triton X-100 (Cat#T8787,
   Sigma‒Aldrich). Fluorescence was measured after 45 min (wavelengths:
   excitation at 485 nm; emission at 530 nm). The pH of the
   Krebs–Henseleit buffer was maintained at 7.4, and fluorescence readings
   were performed using a CLARIOstar microplate reader (BMG LABTECH). All
   tests were performed in 96-well plates, three times in at least three
   independent experiments.

PTEC-Fibroblast crosstalk experiments

   Supernatants from day 5 CD10^+ PTECs not treated with doxycycline were
   mixed with fresh medium at a ratio of 1 to 1 and were referred to as
   senescent medium. The PDGFRβ^+ cells were cultured for 3 days with
   doxycycline and senescent medium and then harvested. Control PDGFRβ^+
   cells received supernatant from PTECs that were cultured with
   doxycycline in the medium.

Cytokine array

   The relative levels of various secreted cytokines were measured in the
   cell culture supernatants of conditionally immortalized PTECs using a
   cytokine array (Proteome Profiler Human Cytokine Array Kit, R&D
   Systems, Cat# ARY005B). PTECs were plated on 6-well plates at a density
   of 60,000 cells/well in 2 ml/well medium, respectively. Seven days
   after plating, with and without doxycycline treatment, cell
   supernatants were collected, pooled for each condition (n = 3), and
   centrifuged twice (1600 rpm, 5 min). The relative cytokine levels were
   visualized according to the manufacturer’s instructions, using 0.5 ml
   of the supernatant. The membranes were simultaneously imaged using the
   Chemi-Doc MP system (Bio-Rad) with a consistent exposure time. The mean
   pixel intensities, normalized to the internal control spots, were
   quantified using ImageJ2 software (version 2.14).

Quantification of MMP-7 using matrix-assisted laser desorption/ionization
(MALDI)–time-of-flight mass spectrometry

   Mass spectrometry was utilized to measure MMP-7 levels since it was not
   included in the cytokine array kit. Twenty micrograms of protein from
   the supernatant of PTECs with or without doxycycline was separated by
   SDS-polyacrylamide gel electrophoresis. Proteins were stained using
   Coomassie Brilliant Blue G-250 (BioRad, Munich, Germany). The gel plugs
   were manually excised and then washed and equilibrated using ammonium
   bicarbonate in acetonitrile. The isolated protein was digested with
   trypsin and analyzed by matrix-assisted laser
   desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF-TOF)
   as previously described [[185]66, [186]67]. A database search
   (Swiss-Prot) was performed using the Mascot 2.2 search engine (Matrix
   Science, Boston, MA) and Bruker Bio-Tool 3.2 software. The calibrated
   and annotated spectra were used to calculate the peptide mass signals
   for each entry in the sequence database, compare the experimental
   MALDI-MS and MALDI-MS/MS datasets, and assign a statistical weight to
   each individual peptide match using empirically determined factors.

Immunofluorescence

   For cultured cells, the following antibodies were used: anti-COL1α1
   (1:100; Abcam Cat# ab34710; RRID: AB_731684), anti-actin alpha SMA-FITC
   (1:100; Sigma Cat# F3777; RRID: AB_476977), and AF555-conjugated
   anti-rabbit (1:100; Vector Cat# BA-1000; RRID: AB_2313606). PDGFRβ^+
   cells were fixed in 4% formalin for 10 min at room temperature.
   Triton-X (0.1%) was used to permeabilize the cells for 20 min. The
   cells were blocked in 2% bovine serum, incubated overnight with a
   primary antibody, washed with PBS and subsequently incubated with the
   secondary antibody for 2 h.

   For human tissue sections, deparaffinization and antigen retrieval
   (Cat# H-3300, Vector) were performed as described below. The sections
   were incubated with blocking buffer (1% donkey serum in PBS) followed
   by incubation with a p21 antibody (1:100; Cell Signaling Cat#2947;
   RRID: AB_823586). The sections were washed with PBS and then incubated
   with Cy3-conjugated secondary antibodies (1:100; Jackson ImmunoResearch
   Cat# 711-165-152; RRID: AB_2307443) and Lotus Tetragonolobus Lectin LTA
   (1:200; Vector Cat#FL1321; RRID: AB_2336559) for 60 min. Cells and
   tissue sections were stained with 4′,6-diamidino-2-phenylindole (DAPI)
   (1:10,000, Sigma‒Aldrich). Images were acquired with a Nikon A1R
   confocal microscope. The raw imaging data were processed using Nikon
   Software and ImageJ.

Immunohistochemical staining and scanning of tissue microarrays

   Microarrays containing cores of human kidney tissue were used for p21
   staining. Deparaffinization was achieved by immersing the slides in
   100% xylene for three rounds of 5 min each. This was followed by 2 min
   of immersion in decreasing ethanol concentrations (3 × 100%, 2 × 96%,
   and 1 × 70%) to rehydrate the tissue, followed by a citrate-based
   antigen retrieval step. A 10-min step involving the addition of 3%
   hydrogen peroxide (Cat# 8070.4; Carl Roth, Karlsruhe, Germany) was then
   carried out to block peroxidase activity. Subsequently, the tissue was
   washed twice with PBS, after which avidin and biotin were added to the
   tissue for 10 min each. The tissue was stained with a p21 antibody
   (1:100; Cell Signaling Cat#2947) at room temperature for 90 min and
   then washed with PBS, followed by incubation with a biotinylated
   anti-rabbit antibody (1:100; Vector Cat#BA-1000; RRID: AB_2213606) for
   30 min. This was followed by washing with PBS and incubating with the
   avidin-biotin complex (Vector Cat# SP-2001, RRID: AB_2336231) for
   30 min. After two washes, the HRP substrate DAB (Vector Cat# SK-4100,
   RRID: AB_2336382) was added to the tissue for 10 min. The entire slide
   was scanned. The image files were imported into QuPath [[187]68], and
   p21^+ cells were manually counted in 76 cores. Cores were excluded from
   the analysis when age (N = 3) or estimated glomerular filtration rate
   (eGFR) (N = 2) were missing.

Statistics

   Variance homogeneity was tested using F tests. Additionally, the data
   were assessed for a normal distribution using the D’Agostino and
   Pearson test. To compare two groups, an unpaired t test or the
   Mann‒Whitney test was applied when the data were normally distributed.
   When comparing more than two groups, one-way analysis of variance
   (ANOVA) with Brown–Forsythe test was used. The correlation between two
   variables was calculated using the Spearman correlation coefficient.
   The ROUT test was employed to identify outliers. A p value less than
   0.05 was considered to indicate statistical significance. All the
   experiments were repeated at least three times, and the experimental
   data were statistically analyzed using GraphPad Prism 9.0. The data are
   presented as the means ± standard deviations (SDs).

Supplementary information

   [188]Supplemental Figures^ (2.5MB, pdf)
   [189]Supplemental Data as Excel File^ (2.2MB, xlsx)

Acknowledgements