Abstract

   Coagulation factor XII (FXII) conveys various functions as an active
   protease that promotes thrombosis and inflammation, and as a zymogen
   via surface receptors like urokinase-type plasminogen activator
   receptor (uPAR). While plasma levels of FXII are increased in diabetes
   mellitus and diabetic kidney disease (DKD), a pathogenic role of FXII
   in DKD remains unknown. Here we show that FXII is locally expressed in
   kidney tubular cells and that urinary FXII correlates with kidney
   dysfunction in DKD patients. F12-deficient mice (F12^-/-) are protected
   from hyperglycemia-induced kidney injury. Mechanistically, FXII
   interacts with uPAR on tubular cells promoting integrin β1-dependent
   signaling. This signaling axis induces oxidative stress, persistent DNA
   damage and senescence. Blocking uPAR or integrin β1 ameliorates
   FXII-induced tubular cell injury. Our findings demonstrate that
   FXII-uPAR-integrin β1 signaling on tubular cells drives senescence.
   These findings imply previously undescribed diagnostic and therapeutic
   approaches to detect or treat DKD and possibly other
   senescence-associated diseases.

   Subject terms: Chronic kidney disease, Molecular biology, Risk factors,
   Experimental models of disease
     __________________________________________________________________

   The role of coagulation factor FXII in diabetic kidney disease (DKD)
   remains unknown. Here, the authors show that zymogen FXII induces
   kidney injury via receptor (uPAR)-mediated signaling by inducing
   oxidative DNA damage and senescence, identifying new potential
   therapeutic targets for DKD.

Introduction

   Coagulation factor XII (FXII, gene F12) is activated upon interaction
   with negatively charged surfaces (contact activation). The activated
   protease (FXIIa) initiates the intrinsic coagulation pathway and
   inflammatory reactions via the kallikrein-kinin-system (KKS)^[76]1.
   Furthermore, FXII zymogen signals through plasma membrane receptors
   such as the urokinase‐type plasminogen activator receptor (uPAR) in
   different cells promoting cell and context- specific responses,
   including angiogenic effects in endothelial cells, activation of immune
   cells such as neutrophils and macrophages, and profibrotic effects in
   lung fibroblasts^[77]2–[78]6.

   Proinflammatory and profibrotic signaling is a hallmark of diabetic
   kidney disease (DKD), a serious microvascular complication in patients
   with diabetes mellitus^[79]7,[80]8. The pathomechanisms underlying
   proinflammatory and profibrotic signaling in DKD involve hemodynamic
   and metabolic changes as well as DNA damage and
   senescence^[81]9,[82]10. Cellular stressors, such as increased reactive
   oxygen species (ROS) generation in diabetic kidneys, trigger DNA damage
   and senescence, which is characterized by permanent cell cycle arrest,
   macromolecular damage, morphological changes, and a specific secretome
   (SASP, senescence-associated secretory phenotype) that induces
   inflammatory and fibrotic changes and compromises kidney
   function^[83]10,[84]11.

   In addition, diabetes mellitus, in general, and DKD, in particular, are
   linked with alterations of the coagulation system that predispose to
   inflammatory and fibrotic changes^[85]12,[86]13. Despite its known role
   in thrombosis and bradykinin-driven inflammation, the role of FXII in
   the pathophysiology of DKD is not yet defined. Earlier reports showed
   upregulation of hepatic FXII production in patients with insulin
   resistance and increased levels of circulating FXII and FXIIa in
   patients with diabetes mellitus or chronic kidney disease
   (CKD)^[87]14–[88]17. Neutrophil-derived FXII activates neutrophils in
   an autocrine fashion, demonstrating that non-hepatic FXII promotes
   inflammation^[89]4, a key feature of DKD.

   FXII signaling via uPAR involves uPAR coreceptors, such as integrins,
   promoting cell-specific responses. Interference with FXII-uPAR binding
   can inhibit FXII-associated signaling effects^[90]3,[91]4. uPAR is
   associated with senescence, and targeting uPAR with CAR-T cells
   eliminates senescent cells and associated pathologies^[92]18.
   Furthermore, the role of uPAR in renal diseases, including DKD, is
   established^[93]19,[94]20.

   Whether FXII induction in DKD contributes to the inflammatory state
   through the activation of the intrinsic coagulation pathway, the KKS,
   or is mechanistically linked with DKD through a signaling mechanism
   independent of its protease function remains unknown. It is possible,
   but it remains to be shown that the interaction of FXII with uPAR
   promotes senescence in diabetic kidneys. Deciphering the relevance of
   FXII binding to uPAR for induction of senescence may be therapeutically
   relevant, as strategies inhibiting zymogen FXII or its active form do
   not increase the risk of bleeding and are hence considered safe^[95]21.

   In the current study, we combined unbiased approaches, analyses of
   diabetes patient samples, and murine diabetes models to uncover a
   function of zymogen FXII signaling for DKD pathology, which promotes
   oxidative DNA damage and tubular senescence via uPAR-integrin β1
   signaling.

Results

Kidney FXII induction correlates with impaired function in human DKD

   Transcriptomic analysis of the Nephroseq® and the Karolinska kidney
   research center (Karokidney^[96]22) databases revealed increased kidney
   tubular F12 expression in DKD patients, while the glomerular expression
   remained unchanged (Fig. [97]1a and Supplementary Fig. [98]S1a, b). F12
   expression was inversely correlated with the estimated glomerular
   filtration rate (eGFR; Fig. [99]1b). In contrast, tubular expression of
   F11, which codes for coagulation factor XI, the FXIIa substrate in the
   intrinsic pathway, was suppressed in DKD as compared to nondiabetic
   kidneys (Fig. [100]1c and Supplementary Fig. [101]S1a,b).

Fig. 1. Upregulation of FXII correlates with impaired kidney function in
human DKD.

   [102]Fig. 1
   [103]Open in a new tab

   a Dot-plot summarizing F12 expression in the tubulointerstitial and the
   glomerular compartments (Karokidney public RNA-sequencing database).
   Dot-plots reflecting mean ± SEM of 20 controls (C) and 19 DKD samples;
   two-tailed unpaired student’s t test. b Line-graph representing the
   negative correlation of F12 expression in the tubulointerstitium with
   the estimated glomerular filtration rate (eGFR) in CKD patients and in
   healthy living donors (n = 147) from the Ju CKD Tublnt Dataset of the
   Nephroseq® database. The confidence interval of r (Pearson coefficient)
   and P value (two-tailed) were calculated by linear regression. c
   Dot-plot summarizing F11 expression in the tubulointerstitial and the
   glomerular compartments (Karokidney public RNA-sequencing database).
   Dot-plots reflecting mean ± SEM of 20 controls (C) and 19 DKD samples;
   two-tailed unpaired student’s t test. d Exemplary histological images
   of human kidney sections stained for FXII (top) and magnified areas
   (bottom) obtained from nondiabetic controls (C; n = 6) or diabetic
   patients with DKD (DKD; n = 5). Scale bars represent 20 μm. e, f
   Dot-plots showing the distribution of the urinary levels of FXII
   (ng/ml; ELISA) in urine samples obtained from the LIFE-ADULT (e) and
   HEIST-DiC (f) cohorts (number of samples are provided in Supplementary
   Tables [104]S2 and [105]S3 respectively). Urinary FXII was measured in
   normoglycemic controls (C) and in diabetic individuals (CKD grade
   according to KDIGO criteria). Dot-plots reflecting mean ± SEM;
   Kruskal-Wallis test with Dunn’s multiple comparison test. g Receiver
   operating characteristic (ROC) analyses of urinary FXII (ng/ml; ELISA)
   in diabetic individuals with low risk of CKD compared to nondiabetic
   controls (blue) or in diabetic individuals with moderate risk of CKD
   (green), high risk (yellow), and very high risk (red) compared to
   low-risk patients in the LIFE-ADULT cohort. AUC: area under the curve.
   h, i Line graphs representing the positive correlation of urinary FXII
   (ng/ml) with urinary albumin creatinine ration (UACR; mg albumin/g
   creatinine; (h) n = 140) and the negative correlation with the
   estimated glomerular filtration rate (eGFR, ml/min/1.73 m²; (i)
   n = 138) in diabetic individuals from the LIFE-ADULT cohort. The
   confidence interval of r (Pearson coefficient) and P values
   (two-tailed) were calculated by linear regression. Source data are
   provided as a “Source Data” file.

   To investigate whether increased F12 expression translates into
   increased protein expression, we next analyzed FXII protein expression
   in human kidney biopsies of diabetic patients with established DKD. DKD
   in these patients was characterized by tubulointerstitial fibrosis,
   thickening of the tubular basement membrane, and glomerular mesangial
   matrix expansion (Supplementary Fig. [106]S1c and Supplementary
   Table [107]S1). FXII was markedly increased in the tubular compartment
   in DKD biopsies compared to nondiabetic controls (Fig. [108]1d).

   Considering that FXII is a secreted protein, we investigated whether
   urinary FXII levels were increased in patients with DKD in two
   independent large cross-sectional cohorts of type-2 diabetic patients
   with different stages of CKD (LIFE-ADULT; Supplementary Table [109]S2
   and HEIST-DiC, Supplementary Table [110]S3). Urinary FXII levels were
   increased in DKD patients compared to healthy controls and were
   associated with DKD severity in both cohorts (ELISA; Fig. [111]1e, f).
   Receiver operating characteristic (ROC) curve analyses revealed an area
   under the curve (AUC) of 0.7957 for urinary FXII levels in diabetic
   patients with a low CKD risk compared to healthy controls, and the AUC
   was even higher when patients with a higher CKD risk were compared to
   those with low risk in the LIFE-ADULT cohort (Fig. [112]1g). Similarly,
   urinary FXII levels were associated with CKD severity in the HEIST-DiC
   cohort (SupplementaryFig. [113]S1d). Concomitantly, urinary FXII levels
   positively correlated with albuminuria and cystatin C levels
   (Fig. [114]1h and Supplementary Fig. [115]S1e, f) and negatively with
   eGFR (Fig. [116]1i and Supplementary Fig. [117]S1g) in both cohorts.
   Taken together, kidney FXII expression is induced in DKD patients, and
   increased urinary FXII levels reflect impaired kidney function.

F12^-/- mice have reduced kidney dysfunction and histopathological changes in
experimental DKD

   To investigate the role of FXII in DKD, we induced persistent
   hyperglycemia in wild-type (WT) and FXII deficient mice (F12^-/-) mice
   using streptozotocin (STZ) for 24 weeks (Fig. [118]2a). Similar to the
   findings in humans, FXII expression (mRNA and protein) was upregulated
   predominantly in the tubular compartment in the kidneys of
   hyperglycemic WT mice compared to normoglycemic controls (Fig. [119]2b
   and Supplementary [120]S2a). In db/db mice (representing type-2
   diabetes^[121]23), FXII expression (protein and mRNA) was likewise
   increased in the tubular compartment compared to nondiabetic control
   db/m mice (Supplementary Fig. [122]S2b, c).

Fig. 2. F12^-/- mice are protected from DKD.

   [123]Fig. 2
   [124]Open in a new tab

   a Experimental scheme of the DKD model. Wild type (WT) and F12^-/- mice
   were age-matched, and persistent hyperglycemia was induced using
   streptozotocin (STZ) for 24 weeks. b Exemplary histological images of
   kidney sections stained for FXII comparing normoglycemic controls (C)
   and hyperglycemic (DM) WT and F12^-/- mice. FXII is detected by HRP-DAB
   reaction (brown); hematoxylin nuclear counter stain (blue). Scale bars
   represent 20 μm. c Line graphs showing urinary albumin-creatinine ratio
   (UACR, μg albumin/mg creatinine) in experimental groups (as described
   in b) after 8, 16, or 24 weeks of persistent hyperglycemia. Line graphs
   reflecting mean ± SEM of 6 mice per group; two-way ANOVA with Tukeys’s
   multiple comparison test comparing hyperglycemic WT and F12^-/- mice at
   the 3 time points. d, e Dot-plots summarizing blood urea nitrogen (BUN,
   mmol/l; (d) and adjusted kidney weight (KW/BW, mg kidney weight /g body
   weight; (e) in the experimental groups (as described in b). Dot-plots
   reflecting mean ± SEM of 6 mice per group; two-way ANOVA with Tukeys’s
   multiple comparison test. f Exemplary histological images of periodic
   acid Schiff staining (PAS) showing glomeruli (top panel), podocyte
   number reflected by Wilms tumor 1 immunostaining (WT-1, brown,
   hematoxylin nuclear counterstain, blue; middle panel), and transmission
   electron microscopy of podocytes (TEM; bottom panel) in experimental
   groups (as described in b); scale bars of top and middle panels
   represent 20 μm, while scale bars of bottom panel represent 1 μm. g
   Exemplary histological images of periodic acid Schiff staining (PAS)
   showing tubuli (top panel), interstitial fibrosis (middle panel,
   Masson’s trichrome stain, MTS), and kidney injury molecule-1
   immunostaining (bottom panel, KIM-1, red; DAPI nuclear counterstain,
   blue) in experimental groups (as described in b); all scale bars
   represent 20 μm. Source data are provided as a “Source Data” file.

   Kidney functional and histopathological markers were indistinguishable
   between normoglycemic WT and F12^-/- mice (Fig. [125]2c–g and
   Supplementary Fig. [126]S2d–k). While blood glucose level and body
   weight were comparable in hyperglycemic WT and F12^-/- mice after 24
   weeks of persistent hyperglycemia (Supplementary Fig. [127]S2d, e), the
   latter group displayed improved kidney function, as indicated by
   reductions in the urinary albumin/creatinine ratio (UACR), blood urea
   nitrogen (BUN), and normalized kidney weight (Fig. [128]2c–e).
   Histological analyses of hyperglycemic F12^-/- kidneys revealed reduced
   DKD-associated glomerular lesions, including mesangial matrix
   expansion, loss of podocytes, and thickening of the glomerular basement
   membrane (GMB) (Fig. [129]2f and Supplementary Fig. [130]S2f–h).
   Furthermore, tubular pathology (tubular dilation, atrophy, and loss of
   brush borders), tubulointerstitial fibrosis, and the expression of
   kidney injury molecule-1 (KIM-1) appeared less pronounced in
   hyperglycemic F12^-/- mice (Fig. [131]2g and Supplementary
   Fig. [132]S1i, k). Thus, FXII deficiency ameliorates glomerular and
   tubular damage in experimental murine DKD.

FXII deficiency induces differential gene expression in DKD

   To gain mechanistic insights on FXII roles in DKD, we performed bulk
   RNA sequencing (RNA-seq) on the kidneys of WT and F12^-/- mice.
   Expression patterns of normoglycemic WT and F12^-/- mice were similar,
   while the expression signatures of hyperglycemic WT and F12^-/- mice
   differed, as revealed by principal component analysis (PCA)
   (Fig. [133]3a). The clustering of mRNA indicated differential gene
   expression with 614 genes downregulated and 428 genes upregulated in
   hyperglycemic F12^-/- compared to WT mice (Fig. [134]3b). Gene set
   enrichment analysis (GSEA) using the hallmark gene sets revealed a
   negative enrichment of fibrosis-related pathways
   (epithelial-mesenchymal transition; EMT), cell cycle arrest (p53
   pathway and G2/M checkpoints), stress pathways (mTORC1 and the unfolded
   protein response), and inflammation (IL2-STAT5 signaling) in
   hyperglycemic F12^-/- mice (Fig. [135]3c and Supplementary
   Fig. [136]S3a). Further pathway analysis of the differentially
   expressed genes (DEGs) indicated that FXII deficiency was associated
   with the downregulation of pathways involved in cell cycle regulation,
   cell adhesion, integrin signaling, inflammation, and hemostasis
   (Fig. [137]3d and Supplementary Fig. [138]S3b- d). On the other hand,
   compared to WT mice, F12 deficiency (i) upregulated DNA damage repair
   pathways and (ii) upregulated metabolic pathways such as lipid
   metabolism, amino acid metabolism, and organic acid
   metabolism^[139]24,[140]25 (Supplementary Fig. [141]S4). Collectively,
   FXII regulates gene sets related to pathways linked to DKD in
   hyperglycemic kidneys^[142]12,[143]26–[144]28.

Fig. 3. FXII deficiency induces differential gene expression in DKD.

   [145]Fig. 3
   [146]Open in a new tab

   a Principal component analysis (PCA) on gene sets of normoglycemic (C)
   and hyperglycemic (DM) WT and F12^-/- mice kidneys. b Heatmap of the
   RNA-seq data showing the differentially expressed genes (DEGs) in WT-DM
   and F12^-/--DM mice. Each column represents data from an individual
   mouse. Color intensity represents row Z-score. c Gene set enrichment
   analysis (GSEA) plots of the hallmark gene sets representing key
   negatively enriched pathways when comparing F12^-/--DM to WT-DM
   kidneys. Significance is represented by the false discovery rate (FDR).
   d Bar graph representing the top enriched pathways based on the
   downregulated differentially expressed genes (DEGs) in F12^-/--DM
   compared to WT-DM kidneys using KEGG (Kyoto Encyclopedia of Genes and
   Genomes), WikiPathways, Reactome, PID (Pathway Interaction Database),
   and GO (Gene Ontology: Biological processes) databases. The pathways
   were ranked by the false discovery rate (FDR). Source data are provided
   as a “Source Data” file.

Hyperglycemic F12^-/- mice are less susceptible to kidney DNA
damage-associated senescence

   DNA damage activates cell cycle arrest and cell cycle checkpoints to
   prevent genomic instability^[147]29,[148]30. The negative enrichment of
   pathways related to cell cycle arrest (p53 pathway) and cell cycle
   checkpoints (G1/S and G2/M checkpoints) in hyperglycemic F12^-/- mice
   (Fig. [149]3d and Supplementary Fig. [150]S4a–c) prompted us to analyze
   the extent of reactive oxygen species (ROS)-induced DNA damage. Kidney
   sections were stained for 8-hydroxy-2’-deoxyguanosine (8-O-dG), a
   marker of oxidative DNA damage^[151]31, and phosphorylated H2A histone
   X (γ-H2AX), a marker of unrepaired DNA damage^[152]32. ROS-induced DNA
   damage was reduced in hyperglycemic F12^-/- mice (Fig. [153]4a–c),
   suggesting that FXII promotes ROS generation while impairing the DNA
   damage response in DKD.

Fig. 4. F12^-/- mice kidneys are less susceptible to DNA damage and
associated senescence.

   [154]Fig. 4
   [155]Open in a new tab

   a–c Exemplary images (a) and dot blots summarizing results (b, c) of
   8-hydroxy-2’-deoxyguanosine (8-O-dG) and phosphorylated H2A histone X
   (γ-H2AX) staining comparing normoglycemic (control, C) and
   hyperglycemic (DM) wild type (WT) and F12^-/- mice. 8-O-dG and γ-H2AX
   are immunofluorescently detected, green and red, respectively; DAPI
   nuclear counterstain, blue. Insets show higher magnification of the
   marked areas. Scale bars represent 20 μm. Dot-plots reflecting
   mean ± SEM of 6 mice per group; two-way ANOVA with Tukeys’s multiple
   comparison test. CTCF: corrected total cell fluorescence. Arb. un.:
   arbitrary units. d Heatmap of the RNA-seq data showing gene expression
   changes of senescence-associated genes in WT-DM and F12^-/--DM mice.
   Each column represents data from an individual mouse. Color intensity
   represents row Z-score. e Bar graphs summarizing expression (qRT-PCR)
   of selected senescence-associated genes in experimental groups (as
   described in a). Bar graphs reflecting mean ± SEM of 4 mice per group;
   two-way ANOVA with Tukeys’s multiple comparison test. f Exemplary
   images of mouse kidney sections stained for senescence-associated
   β-galactosidase (top panel, SA-β-gal, blue; eosin counterstain), p21
   (middle panel, detected by HRP-DAB reaction, brown; hematoxylin nuclear
   counter stain, blue), and senescence-associated heterochromatin foci of
   tri-methyl-histone H3 (Lys9) (bottom panel, H3K9-3me SAHF,
   immunofluorescently detected, red; DAPI nuclear counterstain, blue,
   insets show higher magnification of the marked areas) in experimental
   groups (as described in a). Scale bars represent 20 μm. g Line graph
   representing the positive correlation of urinary FXII (ng/ml) with
   urinary p21 (pg/ml) in diabetic individuals from the LIFE-ADULT cohort
   (n = 146). The confidence interval of r (Pearson coefficient) and P
   values (two-tailed) were calculated by linear regression. Source data
   are provided as a “Source Data” file.

   Persistent DNA damage and defective repair during DKD trigger premature
   senescence that further promotes DKD progression^[156]33. The cell
   cycle inhibitor p21 (Cdkn1a), which we have identified as a driver of
   tubular cell senescence in DKD^[157]34, was downregulated in
   hyperglycemic F12^-/- compared to WT mice (Supplementary Fig. [158]S3b
   and Supplementary Fig. [159]S5a–c). Furthermore, we analyzed a panel of
   genes known to be associated with senescence in mice^[160]35,[161]36.
   These genes were downregulated in the kidneys of hyperglycemic F12^-/-
   compared to WT mice (Fig. [162]4d). The downregulation of genes related
   to cell cycle arrest, the senescence-associated secretory phenotype
   (SASP), cell adhesion, and fibrosis in the kidneys of hyperglycemic
   F12^-/- mice was confirmed by qRT-PCR (Fig. [163]4e). In addition,
   analysis of selected SASP-related cytokines and chemokines using Olink
   technology revealed increased SASP-related cytokines and chemokines in
   some hyperglycemic WT, but not in hyperglycemic F12^-/- mice kidneys
   (Supplementary Fig. [164]S5d). Reduction of tubular senescence in the
   kidneys of hyperglycemic F12^-/- mice was verified by immunostaining
   for the senescence-associated β-galactosidase (SA-β-gal), p21, the
   proliferation marker Ki-67, the nuclear envelope protein lamin-B1, and
   senescence-associated heterochromatin foci (SAHF) (Fig. [165]4f and
   Supplementary Fig. [166]S6a–f). In humans, urinary FXII was positively
   correlated with urinary p21 in DKD patients of the LIFE-ADULT cohort
   (Fig. [167]4g).

   SASP chemokines promote the infiltration of inflammatory cells^[168]37,
   and hence we determined macrophage proportions using F4/80
   immunostaining. We observed reduced macrophage infiltration in the
   tubulointerstitium of hyperglycemic F12^-/- mice compared to
   hyperglycemic WT mice (Supplementary Fig. [169]S6g, h). Senescent cells
   induce anti-apoptotic regulators and are not hallmarked by increased
   apoptosis^[170]38,[171]39. The anti-apoptotic regulators BCL-2 and
   BCL-XL were induced in hyperglycemic WT compared to F12^-/ - mice
   kidneys, and the number of cleaved caspase-3 positive cells was not
   different in the kidneys of both genotypes (Supplementary
   Fig. [172]S7). These findings are consistent with increased kidney
   senescence in WT, but not F12^-/ - mice. Thus, loss of FXII expression
   protects mice from senescence and inflammation in DKD.

FXII is predominantly expressed by murine and human renal tubular cells

   Analysis of the Nephroseq® database confirmed higher F12 expression in
   the tubulointerstitial compartment compared to the glomerular
   compartment (Supplementary Fig. [173]S8a), consistent with our
   immunostaining data in human and mouse kidneys. Furthermore, analysis
   of F12 expression in a single cell transcriptomic database (Kidney
   Interactive Transcriptomics; KIT) indicated the highest F12 expression
   in clusters of proximal tubular cells in healthy adult kidneys^[174]40
   (Supplementary Fig. [175]S8b). In addition, F12 expression was induced
   in proximal tubular clusters of DKD patients^[176]41,[177]42
   (Supplementary Fig. [178]S8c, d). FXII expression was readily
   detectable in a human proximal tubular cell line (HKC-8) and in murine
   primary proximal tubular cells (PTCs) and was increased following
   stimulation with high glucose (25 mM, 24 h) (Supplementary
   Fig. [179]S8e–h). Thus, FXII is predominantly upregulated in the kidney
   tubular compartment under hyperglycemic conditions, supporting a model
   in which the induction of FXII in tubular cells promotes senescence in
   DKD.

FXII induces DNA damage and associated senescence in kidney tubular cells in
vitro

   Considering the zymogen FXII’s direct cellular effects beyond
   coagulation and KKS activation^[180]43, we next examined whether FXII
   can directly promote DNA damage and premature senescence in kidney
   tubular cells. Markers of DNA damage and senescence were induced in
   HKC-8 cells exposed to increasing concentrations of purified human FXII
   for 24 h in the presence of Zn^2+, a cofactor for FXII cell surface
   binding^[181]2 (Supplementary Fig. [182]S9a, b). FXII (62 nM) induced
   time-dependently the expression of KIM-1, p21, and γ-H2AX in HKC-8
   cells starting from 6 h (Fig. [183]5a, b). Mouse PTCs exposed to a
   similar dose of recombinant murine FXII showed comparable induction of
   these markers (Supplementary Fig. [184]S9c, d). The induction of
   tubular cell injury and senescence by FXII was paralleled by increases
   in intracellular ROS levels (H2DCFDA) and oxidative DNA damage (8-O-dG)
   in HKC-8 cells and in PTCs (Fig. [185]5c–e and Supplementary
   Fig. [186]S9e, f). Cell cycle analysis of PTCs exposed to FXII revealed
   an increase of cells accumulating in the G2/M phase compared to
   untreated cells (Supplementary Fig. [187]S9g, h). Furthermore, FXII
   induced SA-β-gal signal in PTCs and the expression of genes related to
   cell cycle arrest, SASP, and inflammation in HKC-8 cells
   (Fig. [188]5f–h). Thus, FXII induces ROS accumulation, DNA damage and
   senescence in human and murine kidney tubular cells in vitro.

Fig. 5. FXII induces DNA damage and associated senescence in kidney tubular
cells in vitro.

   [189]Fig. 5
   [190]Open in a new tab

   a, b Representative immunoblots (a loading control: α-Tubulin) and
   dot-plots summarizing results (b) for γ-H2AX, p21, and KIM-1 expression
   in HKC-8 cells exposed to purified human FXII (62 nM) in the presence
   of Zn^2+ (10 µM) for 6, 24, and 48 h. Dot-plots reflecting mean ± SEM
   of 3 independent experiments; one-way ANOVA with Tukeys’s multiple
   comparison test. c–e Exemplary images (c) and dot-plots summarizing
   results (d, e) of staining with the intracellular ROS detector
   2’,7’-dichlorodihydrofluorescein diacetate (top panel, H2DCFDA, green)
   and 8-hydroxy-2’-deoxyguanosine (bottom panel, 8-O-dG, red; DAPI
   nuclear counterstain, blue) in experimental groups (as described in a).
   Scale bars represent 20 μm. Dot-plots reflecting mean ± SEM of 3
   independent experiments; one-way ANOVA with Tukeys’s multiple
   comparison test. CTCF corrected total cell fluorescence. Arb. un.:
   arbitrary units. f, g Exemplary images of senescence-associated
   β-galactosidase (f, SA-β-gal, blue) and dot-plot summarizing
   quantification of SA-β-gal staining (g) in PTCs treated with
   recombinant mouse FXII (62 nM) in the presence of Zn^2+ (10 µM) for 6,
   24, and 48 h. Dot-plots reflecting mean ± SEM of 3 independent
   experiments; one-way ANOVA with Tukeys’s multiple comparison test. h
   Bar graphs summarizing expression (qRT-PCR) of selected
   senescence-associated genes in experimental groups (as described in a).
   Bar graphs reflecting mean ± SEM of 3 independent experiments; one-way
   ANOVA with Tukeys’s multiple comparison test. Source data are provided
   as a “Source Data” file.

FXII interacts with uPAR on tubular cells under hyperglycemic conditions

   To address a possible role of uPAR in FXII-mediated tubular senescence,
   we analyzed its expression in hyperglycemic WT and F12^-/- kidneys.
   uPAR induction was strong in WT kidneys, however, the receptor was
   hardly detectable in F12^-/- mouse kidneys (Fig. [191]6a, b and
   Supplementary Fig. [192]S10a). Exposure of HKC-8 cells to human FXII
   increased the surface expression of uPAR compared to control cells
   (Fig. [193]6c,d and Supplementary Fig. [194]S10b). To investigate
   whether FXII binds to uPAR on the tubular cell surface with glucose
   stimulation, we exposed PTCs and HKC-8 cells to high glucose (25 mM,
   24 h) and determined FXII-uPAR interaction by proximity ligation assay
   (PLA) and co-immunoprecipitation. Both assays revealed an increase in
   the FXII-uPAR interaction in high glucose conditions (Fig. [195]6e, f
   and Supplementary Fig. [196]S10c). Furthermore, increased interaction
   of FXII and uPAR was detected in human DKD biopsies compared to control
   biopsies using PLA (Fig. [197]6g, h). To investigate whether other
   plasma proteins that compete with FXII to uPAR binding such as high
   molecular weight kininogen (HK) and vitronectin^[198]2 may interfere
   with FXII-uPAR binding, we analyzed the Nephroseq® and the Karokidney
   transcriptomic databases. While tubular F12 expression was increased in
   DKD, the expression of KNG1 (encoding HK) was downregulated, and VTN
   (encoding vitronectin) was not changed compared to controls
   (Supplementary Fig. [199]S11a, b). Analysis of FXII, HK, and uPAR
   expression in human DKD biopsies revealed upregulation and
   colocalization of FXII and uPAR, while the HK signal was barely
   detectable and showed little if any colocalization with the upregulated
   uPAR (Supplementary Fig. [200]S11c g). To experimentally address the
   effects of competing proteins on FXII-uPAR binding on renal tubular
   cells, we exposed HKC-8 cells to FXII (62 nM; ± 10 µM zinc) in the
   presence of equimolar (62 nM) or excess molar (120 nM) concentrations
   of HK and determined FXII binding to uPAR by coimmunoprecipitation. The
   presence of an equimolar concentration of HK did not affect FXII
   binding to uAPR in the presence of zinc, while the excess molar
   concentration of HK reduced the binding even in the presence of zinc
   (Supplementary Fig. [201]S11h). Taken together, these data support a
   model in which the local expression of FXII in renal tubular cells in
   early DKD initiates uPAR signaling and thus promotes DNA damage.

Fig. 6. FXII interacts with uPAR to signal on tubular cell surface.

   [202]Fig. 6
   [203]Open in a new tab

   a, b Representative immunoblots (a loading control: β-Actin) and
   dot-plot summarizing results (b) for uPAR expression in kidney lysates
   of normoglycemic controls (C) and hyperglycemic (DM) wild type (WT) and
   F12^-/- mice. Dot-plot reflecting mean ± SEM of 6 mice per group;
   two-way ANOVA with Tukeys’s multiple comparison test. c, d
   Representative histogram (c) and dot-plot summarizing the results (d)
   of uPAR surface staining determined by flow cytometry (mean
   fluorescence intensity, MFI) in HKC-8 cells exposed to purified human
   FXII (62 nM) in the presence of Zn^2+ (10 µM) for 6, 24, and 48 h.
   Dot-plot reflecting mean ± SEM of 3 independent experiments; one-way
   ANOVA with Tukeys’s multiple comparison test. e, f Representative
   images of proximity ligation assay (PLA, e) and dot-plot summarizing
   results (f) in PTCs exposed to normal (5 mM) or high (25 mM) glucose
   for 24 h. PLA signals representing FXII and uPAR interaction are
   immunofluorescently detected, red; DAPI nuclear counterstain, blue;
   phalloidin for cytoskeleton, green. Scale bars represent 20 μm.
   Dot-plot reflecting mean ± SEM of 3 independent experiments quantifying
   30 cells from each condition with each dot representing the number of
   PLA signals/cell; two-tailed unpaired student’s t test. g, h
   Representative histological images of proximity ligation assay (g, PLA,
   red dots representing FXII and uPAR interaction) and dot-plot
   summarizing results (h) in human kidney sections of nondiabetic
   controls (C) or diabetic patients with DKD (DKD); DAPI nuclear
   counterstain (blue). Scale bars represent 20 μm. Dot-plot reflecting
   mean ± SEM of 5 samples per group with each dot representing the mean
   of PLA signals/field for one sample; two-tailed unpaired student’s t
   test. Source data are provided as a “Source Data” file.

FXII interacts with multiple sites on uPAR through its heavy chain domains

   To identify the interacting residues of FXII and uPAR, we performed
   computational modeling of the heavy chain of FXII and uPAR using
   AlphaFold2_multimer_v3 (see [204]Supplementary Methods), which
   identified candidate binding residues in the fibronectin type II (FN2)
   and Kringle domains of FXII and corresponding residues in domains 1 and
   2 of uPAR (Fig. [205]7a and Supplementary Table [206]S4). To confirm
   the importance of the identified FXII residues, we designed sequential
   peptides covering these residues on the FN2 (HR13: ^54HRQLYHKCTHKGR^66)
   and Kringle (TY10: ^246TYRNVTAEQA^255, PW15: ^275PWCFVLNRDRLSWEY^289)
   domains (Supplementary Table [207]S5). The area covered by the peptide
   HR13 derived from the FN2 domain of FXII has been previously shown to
   mediate binding to negatively charged surfaces^[208]44. Furthermore,
   HR13 shares sequence similarity in 8 amino acids out of 13 with the
   previously published peptide YHK9, which blocks FXII binding to HUVEC
   cells^[209]2. Pretreatment of HKC-8 cells with the peptides HR13 and
   PW15 (300 µM) reduced FXII binding to uPAR as determined by
   coimmunoprecipitation, while TY10 had no effect (Fig. [210]7b).
   Pretreatment with the peptides HR13 and PW15 reduced FXII-induced DNA
   damage and senescence suggesting a functional relevance of the amino
   acid stretches H54-R66 in FN2 domain and P275-Y289 in Kringle domain
   for FXII-induced uPAR signaling (Fig. [211]7c and Supplementary
   Fig. [212]S12a). We next tested the stability of the newly generated
   peptide PW15 that prevented FXII binding to uPAR in our experimental
   conditions. The peptide showed high stability for up to 24 h in HKC-8
   cell culture medium (Supplementary Fig. [213]S12b, c). Treatment of
   HKC-8 cells simultaneously with the peptides HR13 and PW15 induced
   markers of DNA damage and senescence, while single peptides (HR13 or
   PW15) failed to induce this response (Supplementary Fig. [214]S12d, e).
   These results suggest that the simultaneous interaction of FXII with
   different uPAR binding sites is required for signal transduction and
   induction of DNA damage and senescence and that a combination of
   FXII-derived peptides can mimic the effect. Furthermore, we have
   synthesized 3 uPAR-based peptides based on the identified amino acids
   required for the uPAR-FXII interaction, namely RL20
   (^52RLWEEGEELELVEKSCTHSE^71) and DL19 (^96DLCNQGNSGRAVTYSRSRY^114) from
   domain 1 and DV20 (^146DVVTHWIQEGEEGRPKDDRH^165) from domain 2
   (Supplementary Table [215]S5). The area covered by the peptide DL19 has
   been reported previously to interfere with the binding of HK to
   uPAR^[216]45. In addition, the peptide DV20 shares some sequence
   similarities with the previously reported peptide IQE13, a peptide from
   uPAR domain 2 that binds integrins, in addition to sharing a few amino
   acid sequences with the previously reported peptides QCR20 and EEG20
   from uPAR domain 2^[217]3. We used AlphaFold2_multimer_v3 to predict
   the binding of these newly synthesized peptides to the heavy chain of
   FXII. Global docking of the peptides using AlphaFold2_multimer_v3
   predicted binding to the FXII interface for the 3 peptides
   (Supplementary Fig. [218]S12f). Pretreatment of HKC-8 cells with the
   peptides DL19 and DV20 (300 µM) reduced FXII binding to uPAR as
   determined by coimmunoprecipitation, while RL20 had no effect in our
   cellular model (Fig. [219]7d). Pretreatment with the peptides DL19 and
   DV20 reduced the induction of DNA damage and senescence markers by FXII
   or by the combination of FXII-based peptides HR13 and PW15, suggesting
   functional relevance of the amino acid stretches D96-Y114 and D146-H165
   in uPAR domains 1 and 2, respectively, for FXII-induced uPAR signaling
   (Fig. [220]7e and Supplementary Fig. [221]S12g–i).

Fig. 7. FXII interacts with uPAR through its FN2 and kringle domains.

   [222]Fig. 7
   [223]Open in a new tab

   a Computational model of the FXII-uPAR complex with magnified areas
   depicting the interaction of FXII’s fibronectin type II (FN2) and
   Kringle domains with uPAR domain 2 at multiple sites. b Representative
   immunoblots for FXII and uPAR from uPAR coimmunoprecipitation (IP, top)
   and immunoblots for uPAR, and α-Tubulin from the input (input, bottom)
   of HKC-8 cells exposed to purified human FXII (62 nM) in the presence
   of Zn^2+ (10 µM) for 24 h with and without pretreatment with FXII
   sequential peptides (HR13, TY10 and PW15; 300 µM) for 1 h compared to
   control non-treated cells (C). Input serves as a loading control.
   Immunoblots represent 3 independent experiments. c Representative
   immunoblots (loading control: α-Tubulin) for γ-H2AX, p21, and KIM-1
   expression in experimental groups (as described in b). Immunoblots
   represent 3 independent experiments. d Representative immunoblots for
   FXII and uPAR from uPAR coimmunoprecipitation (IP, top) and immunoblots
   for uPAR, and α-Tubulin from the input (input, bottom) of HKC-8 cells
   exposed to purified human FXII (62 nM) in the presence of Zn^2+ (10 µM)
   for 24 h with and without pretreatment with uPAR based peptides (RL20,
   DL19, and DV20; 300 µM) for 1 h compared to control non-treated cells.
   Input serves as a loading control. Immunoblots represent 3 independent
   experiments. e Representative immunoblots (loading control: α-Tubulin)
   for γ-H2AX, p21, and KIM-1 expression in experimental groups (as
   described in d). Immunoblots represent 3 independent experiments.

   To further determine the relevance of the FN2 domain of FXII and domain
   2 of uPAR for the observed effects, we first blocked domain 2 of the
   human uPAR using the previously reported peptide PGS20 (Supplementary
   Table-[224]S5)^[225]3, which reduced FXII-induced DNA damage and
   senescence in HKC-8 cells (Supplementary Fig. [226]S13a, b). To confirm
   the role of the FXII’s FN2 domain for binding to uPAR, we ablated FXII
   expression in HKC-8 cells (F12-null HKC-8) using CRISPR/Cas9 technology
   (Supplementary Fig. [227]S13c) and transfected F12-null cells either
   with wild type FXII (WT-FXII) or a FXII deletion mutant lacking FN2
   domain (∆Fib-II)^[228]46 (Supplementary Fig. [229]S13d). DNA damage and
   senescence markers were reduced in tubular cells transfected with
   ∆Fib-II-FXII compared to WT-FXII transfected cells (Supplementary
   Fig. [230]S13e, f). To determine whether the observed effects require
   FXII’s proteolytic activity, we used two different approaches: i) by
   using the FXII cyclic peptide inhibitors FXII-618 and
   FXII-900^[231]47,[232]48 (Supplementary Table [233]S5) or ii) by
   rescuing F12-null HKC-8 cells with the FXII Locarno mutant (lacking the
   enzymatic activity^[234]49). Markers of DNA damage and senescence were
   not affected by the cyclic inhibitors or the expression of the Locarno
   mutant (Supplementary Fig. [235]S14). Furthermore, transcriptomic
   analysis of the Nephroseq® and the Karokidney public databases revealed
   that in addition to KNG1 (Supplementary Fig. [236]S11a, b), the
   expression of the contact pathway genes KLKB1 (encoding kallikrein) and
   F11 (encoding FXI) were downregulated in DKD patients (Supplementary
   Fig. [237]S15a, b). We confirmed the downregulation of Klkb1 and Kng1
   in hyperglycemic mice kidneys using qPCR and did not find a difference
   between WT and F12^-/- mice (Supplementary Fig. [238]S15c).
   Furthermore, in hyperglycemic WT and F12^-/- mice, mRNA levels of F11
   were comparable, as were D-dimer plasma levels, reflecting that
   coagulation activation was not different between genotypes
   (Supplementary Fig. [239]S15c, d). Collectively, these results suggest
   a direct interaction between zymogen FXII and uPAR on kidney tubular
   cells, which regulates tubular uPAR expression and tubular cell
   senescence independent of FXII’s proteolytic activity.

Integrin β1 is required for FXII signaling via uPAR in renal tubular cells

   uPAR itself is not signaling competent and requires coreceptors such as
   integrins^[240]50. Functional annotations of the DEGs indicated
   FXII-dependent regulation of integrin signaling (Fig. [241]3d), and
   integrin β1 had the highest FDR among the downregulated integrins in
   hyperglycemic F12^-/- mice kidneys (Fig. [242]8a). The downregulation
   of several integrins in the kidneys of hyperglycemic F12^-/- mice was
   confirmed by qRT-PCR and immunoblotting (Fig. [243]8b and Supplementary
   Fig. [244]S16a–c). To study a possible FXII-driven interaction of uPAR
   with integrins β1 and β3, two integrins known to interact with uPAR and
   are expressed by kidney cells^[245]3,[246]20,[247]51, we exposed HKC-8
   cells to FXII and performed coimmunoprecipitation assays. The
   uPAR-integrin β1 interaction increased upon exposure of HKC-8 cells to
   FXII, but binding to integrin β3 was not affected (Fig. [248]8c). The
   increased uPAR-integrin β1 interaction in FXII-treated HKC-8 cells was
   confirmed by PLA (Fig. [249]8d, e). To confirm whether FXII’s FN2
   domain is required to induce uPAR interaction with integrin β1, we
   transfected F12-null HKC-8 cells with WT-FXII or the ∆Fib-II-FXII
   mutant. The latter reduced the interaction of uPAR with integrin β1
   (Fig. [250]8f, g). Blocking integrin β1 with a monoclonal antibody
   abolished FXII-induced induction of γ-H2AX, p21, and KIM-1 in HKC-8
   cells while blocking integrin β3 had no effect (Supplementary
   Fig. [251]S16d–g). To determine whether uPAR-integrin β1 interaction is
   sustained in the tubular compartment under hyperglycemic conditions in
   vivo, we performed PLA on normoglycemic and hyperglycemic WT and
   F12^-/- mice. The uPAR-integrin β1 interaction was readily detectable
   in hyperglycemic WT but almost absent in hyperglycemic F12^-/- mice
   (Supplementary Fig. [252]S17a, b). To validate these findings in human
   DKD, we conducted uPAR-integrin β1 PLA on kidney biopsies of
   nondiabetic controls and DKD patients. The strong interaction of uPAR
   and integrin β1 was readily detectable in the kidneys of DKD patients,
   which was accompanied by FXII upregulation (Fig. [253]8h and
   SupplementaryFig. [254]S17c). Furthermore, immunostaining for FXII,
   uPAR, and integrin β1 showed colocalization of the three proteins in
   DKD patient biopsies compared to controls (Supplementary
   Fig. [255]S18).

Fig. 8. Integrin β1 is required for FXII-uPAR signaling on tubular cells.

   [256]Fig. 8
   [257]Open in a new tab

   a Volcano plot comparing hyperglycemic F12^-/- mice to hyperglycemic WT
   mice based on Log fold change (FC) values and the false discovery rate
   (FDR); integrins are shown in green. Integrin β1 (Itgb1) is the most
   downregulated integrin in hyperglycemic F12^-/- mice kidneys by FDR. b
   Bar graphs summarizing the expression (qRT-PCR) of selected integrin
   genes comparing normoglycemic controls and hyperglycemic WT and F12^-/-
   mice. Bar graphs reflecting mean ± SEM of 4 mice per group; two-way
   ANOVA with Tukeys’s multiple comparison test. c Representative
   immunoblots for active integrin β1, active integrin β3, and uPAR from
   uPAR coimmunoprecipitation (IP, top) and immunoblots for FXII, uPAR,
   and α-Tubulin from the input (input, bottom,) of HKC-8 cells exposed to
   purified human FXII (62 nM) in the presence of Zn^2+ (10 µM) for 24 h
   (FXII) compared to control non-treated cells (C). Input serves as a
   loading control. Immunoblots represent 3 independent experiments. d, e
   Representative images of proximity ligation assay (PLA, d; red, uPAR
   and active integrin β1 interaction) and dot-plot summarizing results
   (e) in experimental groups (as described in c); DAPI nuclear
   counterstain, blue; phalloidin for cytoskeleton, green. Scale bars
   represent 20 μm. Dot-plot reflecting mean ± SEM of 3 independent
   experiments quantifying 30 cells from each condition with each dot
   representing the number of PLA signals/cell; two-tailed unpaired
   student’s t test. f, g Representative images of proximity ligation
   assay (PLA, f; red dots representing uPAR and active integrin β1
   interaction) and dot-plot summarizing results (g) in F12-null HKC-8
   cells transfected with wild type FXII (WT-FXII) or a FXII deletion
   mutant lacking fibronectin type II domain (∆Fib-II-FXII) compared to
   empty-vector transfected cells (C, controls); DAPI nuclear
   counterstain, blue; and phalloidin for cytoskeleton, green. Scale bars
   represent 20 μm. Dot-plot reflecting mean ± SEM of 3 independent
   experiments quantifying 30 cells from each condition with each dot
   representing the number of PLA signals/cell; one-way ANOVA with
   Tukeys’s multiple comparison test. h Representative images of proximity
   ligation assay (PLA, f; red dots representing uPAR and active integrin
   β1 interaction) in human kidney sections of nondiabetic controls (C) or
   diabetic patients with DKD (DKD); DAPI nuclear counterstain, blue; and
   FXII, green. Scale bars represent 20 μm. Source data are provided as a
   “Source Data” file.

   To identify the relevant integrin α-subunit interacting with integrin
   β1 upon FXII-uPAR stimulation in DKD, we focused on integrins α6 and
   α5, which were the most downregulated alpha subunits (highest FDR
   values next to integrin β1) in hyperglycemic F12^-/- mice kidneys
   compared to WT mice (Fig. [258]8a). To determine a possible role of
   integrins α6 and α5, we exposed HKC-8 cells to FXII and performed
   coimmunoprecipitation. While exposure of HKC-8 cells to FXII increased
   the integrin β1/α6 interaction, the integrin β1/α5 interaction was not
   affected (Supplementary Fig. [259]S19a). To scrutinize whether the
   integrin α6β1 heterodimer mediates FXII-uPAR-dependent tubular injury
   and senescence, we blocked integrin α6 with a functional blocking
   monoclonal antibody (1 µg/ml). Blocking integrin α6 abolished
   FXII-induced induction of injury markers in HKC-8 cells (Supplementary
   Fig. [260]S19b, c).

   Aberrant integrin β1 signaling is associated with abnormal focal
   adhesions contributing to senescence^[261]52. To investigate whether
   signaling of FXII-uPAR-integrin β1 axis modulates focal adhesions, we
   determined phosphorylation of focal adhesion kinase (FAK) and Src
   kinase in HKC-8 cells exposed to FXII. FXII time-dependently induced
   phosphorylation and activation of FAK and Src, which was associated
   with upregulation of the Rho family GTPase Rac1 and of the ROS
   regulator NADPH oxidase 1 (NOX1) (Supplementary Fig. [262]S20a, b). To
   determine whether this pathway is activated upon FXII-uPAR interaction
   we pretreated cells with the inhibitory uPAR-based peptides DL19 and
   DV20 (300 µM) or a functional blocking monoclonal antibody targeting
   integrin β1 (10 µg/ml). Both interventions abolished FXII’s effects on
   focal adhesion kinase activation or the upregulation of Rac1 and NOX1
   (Supplementary Fig. [263]S20c–f). Collectively, these data support a
   model in which the integrin α6β1 heterodimer mediates FXII-uPAR
   intracellular signaling via FAK-Src, thereby promoting DNA damage and
   senescence in kidney tubular cells.

Targeting FXII as a therapeutic strategy for DKD

   To determine whether reducing FXII levels could ameliorate kidney
   tubular senescence and experimental DKD, we treated mice after 16 weeks
   of persistent hyperglycemia with a vivo morpholino targeting FXII for
   an additional 6 weeks (Fig. [264]9a). The FXII vivo morpholino markedly
   reduced FXII expression in hyperglycemic kidneys compared to untreated
   mice or to mice treated with a scrambled mismatch morpholino
   (Supplementary Fig. [265]S21a, b). The FXII vivo morpholino reduced
   kidney dysfunction, structural changes, DNA damage, and senescence in
   experimental DKD (Fig. [266]9b–e and Supplementary Fig. [267]S21c–h).
   Importantly, targeting FXII expression reduced the uPAR-integrin β1
   interaction in hyperglycemic mice (Supplementary Fig. [268]S22). Thus,
   targeting FXII or the FXII-uPAR interaction may represent a therapeutic
   approach to reduce senescence and the associated progression of DKD.

Fig. 9. Targeting FXII ameliorates established DKD.

   [269]Fig. 9
   [270]Open in a new tab

   a Experimental scheme of the DKD model with interventions. WT mice were
   age-matched, and persistent hyperglycemia was induced using
   streptozotocin (STZ) and maintained for 16 weeks. Subgroups of mice
   were treated with PBS, mismatch morpholino, or FXII translational
   blocking morpholino (FXII morph.) starting from 16 weeks of
   hyperglycemia for a further 6 weeks. Wks: weeks. b Dot-plot showing
   average urinary albumin-creatinine ratio (UACR, μg albumin/mg
   creatinine) in experimental groups (as described in a) after 16 or 24
   weeks of persistent hyperglycemia. Dot-plot reflecting mean ± SEM of 5
   mice per group; two-way ANOVA with Tukeys’s multiple comparison test. c
   Dot-plot summarizing blood urea nitrogen (BUN, mmol/l) in experimental
   groups (as described in a). Dot-plot reflecting mean ± SEM of 5 mice
   per group; one-way ANOVA with Tukeys’s multiple comparison test. d
   Exemplary histological images of periodic acid Schiff staining (top
   panel, PAS), interstitial fibrosis (middle panel, Masson’s trichrome
   stain, MTS), and kidney injury molecule-1 immunostaining (bottom panel,
   KIM-1, red; DAPI nuclear counterstain, blue) in experimental groups (as
   described in a); scale bars represent 20 μm. e Exemplary histological
   images of senescence-associated β-galactosidase (top panel, SA-β-gal,
   blue; eosin counterstain), p21 immunostaining (middle panel, detected
   by HRP-DAB reaction, brown; hematoxylin nuclear counter stain, blue),
   and phosphorylated H2A histone X (bottom panel, γ-H2AX,
   immunofluorescently detected, red; DAPI nuclear counterstain, blue,
   insets show higher magnification of the marked areas) in experimental
   groups (as described in a); scale bars represent 20 μm. Source data are
   provided as a “Source Data” file.

Discussion

   This study identifies a previously undescribed role of zymogen FXII
   signaling in the progression of DKD. Based on our results, we propose
   that the zymogen FXII binds to uPAR and signals via integrin β1 on
   tubular cells, promoting DNA damage and senescence. The current results
   suggest that therapies targeting FXII or the FXII-uPAR interaction
   constitute modern strategies to ameliorate DKD. Further preclinical and
   translational studies are required to validate the therapeutic
   potential of this identified role of FXII-uPAR signaling in DKD.

   The current study identifies tubular uPAR as a receptor for the
   detrimental effects of FXII. uPAR induction and signaling have been
   previously reported in human and murine DKD in glomerular
   cells^[271]19,[272]53,[273]54. We observed a strong induction of uPAR
   in tubular cells upon exposure to a hyperglycemic milieu, which was not
   apparent in hyperglycemic F12^-/- kidneys. Moderately increased uPAR
   expression upon exposure to FXII has been previously shown in
   neutrophils^[274]55. The mechanism through which FXII regulates uPAR
   surface expression remains to be fully understood.

   uPAR is expressed by senescent cells and has been considered as a
   therapeutic target for senolytic strategies^[275]18. Here we propose a
   mechanism through which uPAR induces senescence, thus identifying
   previously undescribed approaches to therapeutically target
   uPAR-mediated senescence. FXII-induced uPAR expression on tubular cells
   was associated with ROS accumulation, DNA damage, and senescence in
   human and murine kidney tubular cells. Kidney tubular cells that are
   normally resting in the G1 phase of the cell cycle respond to
   DNA-damaging stressors under hyperglycemic conditions, aiming to
   maintain genomic stability, by inducing cell cycle inhibitors and cell
   cycle checkpoint regulators^[276]56,[277]57. However, persistent
   hyperglycemic stress beyond the cellular repair capacity drives
   premature senescence^[278]58. Expression profiling revealed that the
   protection against kidney injury in hyperglycemic F12^-/- mice was
   linked to negative enrichment of key pathological pathways related to
   DKD, including cell cycle arrest. Based on the mouse model used and the
   clinical data, we speculate that FXII-uPAR signaling is an early event
   in the course of DKD leading to tubular ROS and senescence. We cannot
   exclude that at advanced stages, excess molar concentrations of
   competing proteins, such as HK, may cross the dysfunctional glomerular
   filtration barrier and modulate FXII-uPAR signaling. Whether this would
   have an impact on already established tubular senescence remains to be
   studied.

   Senescent cells are characterized by their adhesive phenotype, which is
   associated with increased focal adhesions and reduced
   motility^[279]59,[280]60. Furthermore, secreted metalloproteinases in
   the SASP modulate the extracellular matrix (ECM) of senescent cells and
   cause tissue remodeling^[281]61. Consistently, pathways related to ECM
   remodeling and integrin signaling, which are typically induced in
   DKD^[282]62, were negatively enriched in hyperglycemic F12^-/-mice.
   Reducing FXII levels in vivo attenuated premature senescence and the
   associated SASP. Taken together, our data suggest that locally produced
   FXII interacts with uPAR and induces tubular senescence and associated
   changes, leading to kidney damage.

   Signaling effects of the zymogen FXII have been linked to its
   interaction with uPAR via its heavy chain, which mediates binding and
   uPAR-dependent signaling^[283]3,[284]4,[285]63. The heavy chain domains
   protect the circulating zymogen FXII from autoactivation through steric
   hindrance provided by the interaction between the FN2 and the Kringle
   domains, keeping FXII in a closed conformation^[286]46,[287]64–[288]66.
   In response to surface binding, this closed conformation is relaxed,
   exposing the cleavage site for activation^[289]64,[290]65. Considering
   the importance of FXII heavy chain domains for surface interactions, we
   focused on the heavy chain for computational structural modeling and
   identified candidate binding residues in FXII’s FN2 and Kringle domains
   and corresponding residues on domains 1 and 2 of uPAR mediating the
   FXII-uPAR interaction. This modeling and our experimental data identify
   new molecular interactions of FXII and uPAR, extending previous
   findings^[291]3. Furthermore, the blocking effects of the peptides DL19
   and PW15 derived from uPAR domain 1 and FXII’s Kringle domains,
   respectively, suggest previously undescribed binding sites of the
   FXII-uPAR interaction. Previous studies suggested the involvement of
   FXII’s Kringle domain in binding to artificial surfaces^[292]67.
   Binding of the Kringle domain of FXII to domains 1 and 2 of uPAR seems
   possible, taken into account the following evidence: (i) the important
   role of the Kringle domain of urokinase-type plasminogen activator
   (uPA) to uPAR binding^[293]68, (ii) the high sequence similarity
   between the Kringle domains of FXII and uPA, and (iii) the inhibition
   of this binding by peptides targeting uPAR domains 1 and 2^[294]69. Our
   computational docking predicted the binding of the synthesized peptides
   to their corresponding residues, yet not all of them showed inhibitory
   effects on the FXII-uPAR interaction in our experiments. These
   differences may reflect discrepancies between the in silico and in
   vitro approach or indicate that the interacting residues between FXII
   and uPAR depend on the specific cell type and experimental conditions.
   Interestingly, the experimental results indicate that the simultaneous
   interaction of FXII with different uPAR binding sites is required for
   signal transduction and induction of DNA damage and senescence.
   Accordingly, blocking one binding site is sufficient to inhibit FXII’s
   effect. Further detailed analyses of the FXII-uPAR interaction may
   hence identify new molecular targets, allowing to therapeutically
   modulate this interaction.

   In addition to reduced tubular injury, glomerular injury was
   ameliorated in hyperglycemic F12^-/- mice in the current study.
   Glomerular injury is likewise reduced in F12^-/- mice with sickle cell
   disease^[295]55, but the authors did not investigate tubular injury.
   The direct effect of zymogen FXII on podocytes or other glomerular
   cells seems possible, given the established role of uPAR in glomerular
   pathologies^[296]19,[297]20. Yet, the relation of tubular and
   glomerular injury is bidirectional and some studies suggest that
   tubular damage is in part independent of and may even precede
   glomerular injury in DKD^[298]70–[299]72. We observed FXII expression
   in kidney tubular cells, which increased in response to high glucose in
   vitro and in vivo. Local FXII production by cells other than
   hepatocytes^[300]4,[301]6,[302]73 (and human protein atlas) and
   FXII-mediated effects independent of hepatic FXII have been
   reported^[303]4, suggesting that tubular-released FXII may induce
   tubular damage in an auto- or paracrine fashion. Based on the current
   results we assume that the protective phenotype observed with FXII
   morpholino treatment as reflected by reduced tubular senescence and
   ameliorated albuminuria reflects the contribution of tubular cell
   injury to albuminuria in DKD^[304]74. The precise role of tubular FXII
   in kidney injury, e.g., a differential effect in acute versus chronic
   renal injury, requires further investigation, including, for example,
   mouse models with cell-specific FXII-inactivation.

   Increased tubular FXII expression is expected to increase urinary FXII
   levels. We show that urinary FXII levels correlated with increasing
   severity of DKD in two independent cohorts of type-2 diabetic patients,
   indicating the utility of urinary FXII as a tubular injury biomarker.
   We currently do not know whether the increased urinary FXII selectively
   results from increased tubular FXII expression or increased filtration
   of plasma-derived FXII. Given its size (80 kDa), FXII may cross the
   glomerular filtration barrier once barrier dysfunction is established.
   In addition, while our data propose that urinary FXII reflects tubular
   injury, the levels measured in urine most likely do not reflect levels
   reached locally in the tubular compartment or at the surface of tubular
   cells. Further studies are required to elucidate the exact role and the
   regulation of tubular FXII in the pathogenesis of DKD. Another
   interesting question is whether the recently identified
   nephroprotective therapies (SGLT2-inhibitors and GLP-1 agonists)
   regulate tubular FXII expression. These questions need to be addressed
   in the future.

   Consistent with previous reports showing that zinc is a cofactor
   required for FXII surface binding^[305]2,[306]4,[307]55,[308]75, the
   FXII-uPAR dependent effects on tubular cells were found to be
   zinc-dependent. In blood, activated platelets release zinc from
   internal stores^[309]76. Furthermore, activated platelets are known to
   contribute to DKD^[310]77, but it remains currently unclear how zinc
   released from the platelets may reach the intratubular lumen to
   facilitate FXII-uPAR binding. Alternatively, renal tubular cells
   express zinc transporters and reabsorb zinc^[311]78. In addition,
   deficiency of metallothioneins, a family of heavy metal binding
   proteins expressed by tubular cells and maintaining zinc homeostasis,
   exacerbates DKD in murine models by inducing oxidative
   stress^[312]79,[313]80. Therefore, it appears possible that tubular
   injury in hyperglycemic conditions releases zinc from its intracellular
   stores, enhancing FXII binding to uPAR and, hence FXII signaling.
   Indeed, urinary excretion of zinc is increased in CKD patients^[314]81,
   which may provide sufficient zinc in the tubular lumen. The origin of
   zinc needs to be characterized in the future and may provide new
   therapeutic options to restrict the FXII-uPAR interaction on tubular
   cells.

   Zinc mediates FXII surface binding but may also facilitate FXII
   activation. Our data suggest that the effects of FXII on tubular cells
   are independent of its activation, in line with previous reports on
   other cell types^[315]2–[316]4. Furthermore, the differential
   regulation of the contact pathway proteins in the kidneys of DKD
   patients and in hyperglycemic mice and the absence of coagulation
   changes in F12^-/- mice compared to WT mice support the notion that the
   effect of FXII is independent of coagulation activation. Data
   describing a function of the KKS, which is activated by the protease
   FXII, in kidney injury are controversial as earlier studies suggested
   that bradykinin promotes tubular and glomerular injury^[317]82,[318]83,
   while other more recent studies imply protective effects of exogenous
   kallikrein and bradykinin receptors in DKD^[319]84,[320]85. Our data
   suggest a detrimental effect of zymogen FXII which is hence independent
   of the KKS. Further analyses are required to delineate a possible
   additional effect of the activated FXII in DKD.

   The lack of a transmembrane domain in uPAR requires its association
   with membrane spanning receptors, such as integrins^[321]50,[322]86.
   Integrins form complexes with FXII and uPAR, promoting FXII-dependent
   signaling in a cell and context specific manner^[323]87,[324]88.
   Upregulation of integrins such as β1, β3, and β6 is linked to EMT,
   fibrosis, and senescence in injured tubular
   cells^[325]51,[326]89,[327]90. Persistent integrin β1 signaling is
   associated with sustained focal adhesions and ROS production that
   contribute to senescence^[328]52. In epithelial cells, the most
   abundant integrins are β1-containing heterodimers, and integrin β1
   signaling can induce diverse cellular responses depending on the cell
   type, the α subunit forming the integrin heterodimer, the binding
   ligand, and the cellular microenvironment^[329]91–[330]93. Our results
   showed (i) that the uPAR/FXII complex required active integrin β1 but
   not β3, that (ii) a blocking antibody against integrin β1 almost
   completely prevented FXII-induced tubular cell injury, that (iii) the
   integrin β1 forms a heterodimer with integrin α6, which promotes
   FXII-associated DNA damage and tubular cell injury. The integrin α6β1
   heterodimer induces ROS, DNA damage and senescence in human
   fibroblasts^[331]94, similar to the phenotype we observed in kidney
   tubular cells in our study. Integrin signaling activates the Rho family
   GTPase Rac1 which is promoted by FAK^[332]94–[333]96. Rac1 activates
   NADPH oxidases (NOXs), increasing ROS generation, DNA damage and
   senescence^[334]97,[335]98. Constitutive integrin signaling associated
   with abnormal focal adhesions induces senescence through increased ROS
   production^[336]52. Integrin β1 signaling mediates cellular adhesion
   through phosphorylation of FAK, and the latter acts as a scaffolding
   platform for other kinases including Src kinase^[337]95,[338]99. Src
   activation maintains a senescence phenotype in fibroblasts in response
   to DNA damage^[339]100. In the current study, FXII deficiency was
   associated with the downregulation of pathways related to integrin
   signaling, focal adhesions, and Rho GTPase signaling in hyperglycemic
   kidneys. Furthermore, interference with FXII-uPAR binding or blocking
   integrin β1 reduced the activation of focal adhesion kinases and the
   upregulation of Rac1 and NOX1 in tubular cells, suggesting the
   involvement of abnormal focal adhesions in FXII-mediated oxidative DNA
   damage. Collectively, these results are consistent with a model in
   which FXII-uPAR interaction induces tubular senescence via integrin β1
   signaling. Considering the wide range of functions and effects of
   integrins, therapies targeting upstream receptor mechanisms, such as
   the FXII-uPAR interaction, could be superior approaches to target
   integrin-mediated detrimental effects.

   In conclusion, our findings show a previously undescribed function of
   FXII in the progression of DKD in which FXII-dependent stabilization of
   and interaction with uPAR on tubular cells induces integrin β1
   signaling that in turn promotes DNA damage and senescence. Targeting
   FXII or its interaction with uPAR may represent promising therapeutic
   avenues for DKD.

Methods

   The research conducted in this study complies with the ethical
   regulations of the University of Leipzig (Ethic vote no:
   263-2009-14122009 and 201/17-ek), the local animal care and use
   committee (TVV 70/21, Landesverwaltungsamt, Leipzig, Germany), and the
   University of Heidelberg (Ethic vote no: S-383/2016).

Study design

   The objectives of this study were to identify the possible role of FXII
   in the pathogenesis of DKD and to elucidate the involved mechanism.
   FXII expression was analyzed in human kidney biopsies (immunostaining),
   in urine samples (ELISA) and in various murine models of DKD
   (streptozotocin induced DKD, db/db mice). Kidneys of WT and F12^-/-
   mice were analyzed using RNA-sequencing, immunostaining, and
   immunoblotting. Furthermore, time kinetics and mechanistic studies were
   conducted in human and murine kidney tubular cell lines in vitro. In
   silico computational structural modeling of the interacting proteins
   was conducted to elucidate the signaling effect of FXII. Details of all
   experimental procedures are provided in the supplementary methods.

Human kidney biopsies and urine samples

   Human kidney biopsies of type-2 diabetic patients with established DKD
   and non-diabetic controls were obtained from the tissue bank of the
   National Center for Tumor Diseases, Heidelberg, Germany (Supplementary
   Table [340]S1). Human urine samples of control and type-2 diabetic
   individuals with different CKD stages were obtained from the LIFE-ADULT
   cohort (Ethic vote no: 263–2009–14122009 and 201/17-ek, University of
   Leipzig, Supplementary Table [341]S2)^[342]101 and from the
   Heidelberger Study on Diabetes and Complications (HEIST-DiC, Ethic vote
   no: S-383/2016, University of Heidelberg, Supplementary
   Table [343]S3)^[344]102. In both cohorts, CKD severity was classified
   according to the KDIGO criteria^[345]103.

Murine model of DKD and in vivo interventions

   Male wild-type (WT) C57BL/6 mice, nondiabetic C57BLKsJ-db/+ (db/m), and
   diabetic C57BL/KsJ-db/db (db/db) mice were obtained from Janvier
   (S.A.S., St. Berthevin Cedex, France). Male F12 deficient (F12^-/-)
   mice on C57BL/6 background were provided by Thomas Renné and have been
   previously described^[346]104. Mice were maintained at a temperature of
   21 ± 2 °C and a humidity of 55 ± 15% with free access to standard chow
   diet (V1534-0, ssniff, Germany) and water. Persistent hyperglycemia was
   induced in male WT and F12^-/- mice at the age of 8 weeks using
   intraperitoneal injections of streptozotocin (STZ) freshly dissolved in
   0.05 mM sterile sodium citrate (pH 4.5, 60 mg/kg body weight for 5
   consecutive days)^[347]105. Age-matched control mice of both genotypes
   received intraperitoneal sodium citrate injections. Blood glucose
   levels were determined routinely using ACCU-CHEK glucometer with blood
   samples collected from the tail vein. Mice that had a blood glucose
   level above 17 mM after 2 weeks of STZ injections were considered
   diabetic. Mice that had a blood glucose level of more than 28 mM
   received 1–2 units of insulin glargine (Lantus®; SANOFI, France)
   subcutaneously to avoid potentially lethal hyperglycemia. Mice were
   maintained under persistent hyperglycemia with continuous monitoring
   for 24 weeks, then they were sacrificed to collect blood and organs for
   further analyses.

   A subset of WT mice were treated after 16 weeks of persistent
   hyperglycemia with a vivo morpholino oligomer
   (5’-ACCCCAGGAACAACAGAGCCGTCAT-3’) targeting the translation of mouse
   F12 gene (6 mg/kg body weight in PBS, every other day, intraperitoneal)
   or a mismatch vivo morpholino (5’-CCCCGCTGCCTGCCCAGGA-3’) for further 6
   weeks according to an established protocol^[348]106. Control mice
   received PBS every other day (intraperitoneal) for the same duration.
   All animal experiments were conducted following standards and
   procedures approved by the local animal care and use committee (animal
   proposal number: TVV 70/21, Landesverwaltungsamt, Leipzig, Germany).

Cell culture and in vitro interventions

   Immortalized human-derived renal proximal tubular cells (HKC-8) were
   cultured in a mixture of DMEM glucose-free medium and Ham’s F12
   nutrient mixture in a ratio of 1:1 to achieve a final glucose
   concentration of 5 mM in the medium. The medium mixture was
   supplemented with 10% FBS, and the cells were maintained at 37 °C.
   Mouse primary proximal tubular cells (PTCs) were isolated and cultured
   according to an established protocol^[349]107. Freshly isolated mouse
   kidneys were decapsulated and the cortical area was separated and
   minced into small pieces and collected in HBSS containing Collagenase-I
   at a concentration of 200 units/ml. The small cortical fragments were
   digested at 37 °C for 30 min and the proximal tubule fragments were
   enriched using horse serum. The Proximal tubule containing fragments
   were pelleted down and washed twice with HBSS, then the pellet was
   resuspended in DMEM:F12 medium in a ratio of 1:1, supplemented with
   insulin/transferrin/selenium (Invitrogen, 5 μg/ml, 2.75 μg/ml and
   3.35 ng/ml respectively), APO transferrin (Sigma-Aldrich, 2.0 μg/ml),
   hydrocortisone (Sigma-Aldrich, 40 ng/ml), recombinant human epidermal
   growth factor (rhEGF, R&D Systems, 0.01 μg/ml), and 1%
   antibiotic/antimycotic solution (Sigma-Aldrich, 10,000 units/ml
   penicillin, 0.1 mg/ml streptomycin and 0.25 μg/ml amphotericin B). The
   medium was changed after 72 h, and the cells were maintained in the
   same culture medium without rhEGF.

   For dose selection of FXII, HKC-8 cells were exposed to increasing
   concentration of purified human FXII (30, 62, or 150 nM) in the
   presence of 10 µM of Zn^2+ for 24 h. For the time kinetics study, HKC-8
   cells or PTCs were exposed to purified human FXII (62 nM) or
   recombinant mouse FXII (62 nM), respectively, in the presence of 10 µM
   of Zn^2+ for 6, 24, and 48 h. In a subset of experiments, HKC-8 cells
   were pretreated with the human uPAR domain-1 based peptides RL20 and
   DL19, domain-2 based peptides DV20 and PGS20^[350]3 or the human
   FXII-based peptides (HR13, TY10, and PW15), at a concentration of
   300 µM for 1 h followed by treatment with purified human FXII (62 nM)
   in the presence of 10 µM of Zn^2+ for 24 h. In another set of
   experiments, HKC-8 cells were pretreated with the peptides HR13 and
   PW15 (300 µM) alone or in combination for 24 h with and without
   pretreatment with the peptides DL19 and DV20 for 1 h (300 µM). For FXII
   inhibitor experiments, HKC-8 cells were pretreated with the activated
   FXII cyclic peptide inhibitors, FXII-618^[351]48 and FXII-900^[352]47,
   at a concentration of 10 µM for 30 min, followed by treatment with
   purified human FXII (62 nM) in the presence of 10 µM of Zn^2+ for 24 h.
   For blocking integrins β1, β3, and α6 experiments, HKC-8 cells were
   preincubated with the function-blocking monoclonal antibodies against
   integrin β1 (clone P5D2)^[353]108, integrin β3 (clone B3A)^[354]109 at
   10 µg/ml, or integrin α6 (clone GOH3)^[355]110 at 1 µg/ml for 30 min
   followed by treatment with purified human FXII (62 nM) in the presence
   of 10 µM of Zn^2+ for 24 h.

Generation of F12-null HKC-8 cells and FXII mutant transfection

   HKC-8 cells were seeded in a 6-welll plate at a density of 4 × 10^5
   cells/ well. Confluent cells were transfected with GFP-expressing human
   FXII CRISPR/Cas9 KO plasmid (sc-409611; Santa Cruz Biotechnology) using
   Turbofect transfection reagent following manufacturer’s protocol. After
   24 h, transfected cells were sorted, and GFP-positive single cells were
   collected in 96-well plates. Single-cell clones were expanded and
   screened for FXII expression by immunoblotting. In different sets of
   experiments, F12-null HKC-8 cells were transfected with wild-type human
   FXII (pcDNA3-FXII WT), a FXII mutant lacking the FN2 domain
   (pcDNA3-FXII-∆FibII), a FXII mutant lacking enzymatic activity
   (pcDNA3-FXII-Locarno), or empty vector (pcDNA3.1 +) using Turbofect
   transfection reagent following manufacturer’s protocol. After 24 h,
   proteins were isolated for immunoblotting or cells grown on cover slips
   were fixed for staining.

Statistical analysis

   The data are summarized as mean ± SEM (standard error of the mean).
   Statistical analyses were performed with parametric Student’s t test,
   one-way ANOVA, two-way ANOVA, or non-parametric Mann–Whitney and
   Kruskal–Wallis test, as appropriate, and post-hoc comparison with the
   method of Tukey or Dunnett’s multiple comparisons. The
   Kolmogorov–Smirnov (KS) test or D’Agostino–Pearson-Normality-test was
   used to determine whether the data were consistent with a Gaussian
   distribution. Prism 9 ([356]www.graphpad.com) software was used for
   statistical analyses. Statistical significance was accepted at values
   of P < 0.05.

Reporting summary

   Further information on research design is available in the [357]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [358]Supplementary Information^ (4.6MB, pdf)
   [359]Peer Review File^ (2.3MB, pdf)
   [360]Reporting Summary^ (1.4MB, pdf)

Source data

   [361]Source Data^ (7.9MB, xlsx)

Acknowledgements