FGF23 is synthesised locally by renal tubules and activates injury-primed fibroblasts

In kidney disease, higher circulating levels of the mineral-regulating hormone fibroblast growth factor (FGF)-23 are predictive of disease progression but direct pathogenic effects on the kidney are unknown. We sought evidence of local renal synthesis in response to unilateral ureteric obstruction in the mouse, and pro-fibrotic actions of FGF23 on the fibroblast in vitro. Acute tubulointerstitial injury due to unilateral ureteric obstruction stimulated renal FGF23 synthesis by tubules, and downregulated inactivating proprotein convertases, without effects on systemic mineral metabolism. In vitro, FGF23 had divergent effects on fibroblast activation in cells derived from normal and obstructed kidneys. While FGF23 failed to stimulate fibrogenesis in normal fibroblasts, in those primed by injury, FGF23 induced pro-fibrotic signalling cascades via activation of TGF-β pathways. Effects were independent of α-klotho. Tubule-derived FGF23 may amplify myofibroblast activation in acute renal injury, and might provide a novel therapeutic target in renal fibrosis.


Growth factors, Inhibitors and antibodies
Inhibitors used, and their final concentrations are listed in SupplementaryTable S1.
None of the compounds were found to significantly reduce cell viability (MTT assay) over 24h, when used at the stated concentrations. Primary and secondary antibodies used are listed in Supplementary Table S2 and Supplementary Table S3 respectively.

Immunoperoxidase staining
Adjacent portions of renal tissue were fixed in methyl carnoys or 4% paraformaldehyde in phosphate buffered saline (PBS), processed and embedded in paraffin wax. For DAB immunohistochemistry, tissue sections were dewaxed in xylene, rehydrated in graded alcohol, endogenous peroxidase activity was quenched using 3% H 2 O 2 in water for 30 min, before nonspecific binding sites were blocked with 10% normal horse serum for a further 30 min at RT (Vector Laboratories, Burlinghame, CA, USA). Sections were then incubated with rabbit polyclonal anti-collagen I (methyl caronys) or rabbit monoclonal anti-vimentin (4% paraformaldehyde). In each case, antibody binding was visualised using an avidin-biotincomplex (ABC Elite; Vector Laboratories) and 3,3'-diaminobenzidine (DAB; Sigma). For detection of α -smooth muscle actin, parformaldehyde fixed sections were prepared as above and then incubated with mouse monoclonal αSMA, conjugated to biotin (ARK kit; Dako), and incubated with streptavidin peroxidase (ARK) and DAB. Finally sections were counterstained with haematoxylin (Dako), dehydrated, and mounted in DePex (Merck) 6 . Images were taken using an Olympus BX50 microscope using 10x magnification.

Immunofluorescence staining
For immunofluorescent co-staining of FGF23 and Lotus tetragonolobus lectin (LTL), parformaldehyde fixed sections were boiled under pressure in citrate buffer (pH 6.0), equilibrated in PBS for 30 min and free aldehyde groups reduced by immersion in NH 4 Cl (50 mmol/L) in PBS for 20 min at room temperature. Non-specific binding sites were blocked with 10% goat serum (Vector laboratories) in 3% BSA/PBS (pH 7.6) containing 0.1 M glycine for 1h at RT. Slides were then incubated with rat monoclonal anti-mouse FGF23 in 1% BSA in PBS overnight at 4°C. Sections were then incubated with biotin-conjugated goat anti-rat IgG (Vector Laboratories) and fluorescein labelled-LTL lectin (1:200; FL-1321; Vector Laboratories) in 1% BSA in PBS for 2h at RT. DAPI (2 µg/mL in PBS, 15 min, RT) was used a nuclear stain. For immunofluorescent co-staining of FGF23 and αSMA, parformaldehyde fixed sections were subjected to antigen retrieval and blocking as above before incubation with mouse monoclonal anti-αSMA in1% BSA/PBS for 2h at RT. Sections were then incubated with rat monoclonal anti-mouse FGF23 in in1% BSA/PBS overnight at 4°C, before the addition of biotinylated-anti-rat IgG secondary and Alexa Fluor 594 anti-mouse IgG (Life Technologies, Carlsbad, CA, USA) in1% BSA/PBS for 2h at RT. Since FGF23 is a low-abundance target, signal was amplified using Alexa Fluor 488 or 595 tyramide reagent (Alexa Fluor™ Tyramide SuperBoost™ Kit, streptavidin; Life Technologies) according to the manufacturer's instructions. Sections were washed in PBS and mounted in hard set Vectashield (#H-1400; Vector laboratories). Low power images were taken using a Zeiss AXIOSKOP2 microscope using 0.75x or 20x magnification or visualised on a Leica SP5 confocal microscope with a 63x oil objective for high power imaging. Post-acquisition processing was performed using Fiji-ImageJ (NIH).
The mouse FGF-23 antibody (#MAB2629; R&D systems) used here has been employed in the IHC detection of FGF23 in multiple rodent studies 7,8 , including ectopic detection in the kidney 9 . Specificity was confirmed here by blocking studies using antiserum pre-absorbed with a 10 molar excess of the immunising peptide (#2629-FG; R&D systems).
After centrifugation, the supernatant was used at the same working dilution, and in parallel with anti-FGF23 staining, as above ( Supplementary Fig.S6). Comparison of immunohistochemical detection without amplification using SuperBoost is also shown alongside.

Laser capture microdissection
Laser Capture Microdissection (LCMD) was used to isolate specific nephron segments from fixed paraffin embedded tissue sections using the Veritas™ instrument and Arcturus reagents (Mountain View, CA, USA) 10 .
Paraffin wax embedded methyl carnoys fixed tissue blocks prepared above were resectioned at 5-10µm. Sections were collected on RNase free plastic slides (Arcturus). The slides were de-waxed using RNase-free techniques, and stained using the HistoGene kit (Arcturus) to provide differential contrast for subsequent micro-dissection. HistoGene stain (Arcturus) provides differential staining of nuclei (purple) and cytoplasm (light pink) while preserving RNA integrity. The contrast provided by the HistoGene stain was used to identify cortical glomerular and tubular profiles for excision. A Veritas Microdissection Instrument (Arcturus), which utilizes the IR capture laser with a UV cutting laser, was used to laser capture a distinct population of glomeruli and tubules. After dissecting individual glomeruli and tubules (>50 per section) from surrounding tissue, each was captured into an Adhesive Cap (Arcturus) using the IR capture laser. After the cells were collected, the cap was immediately transferred to a microfuge tube containing extraction buffer (Arcturus) and processed as detailed below.

Biochemical studies
Serum obtained from animals at D0 and D3 post UUO were analysed for creatinine intact FGF23 (#60-6800; Immutopics, Inc) and C-terminal FGF23 (#60-6300; Immutopics, Inc), using commercially available colorimetric assays or ELISA and run according to manufacturer's instructions. Intact FGF23 and C-terminal FGF23 assays were calibrated using the same set of standards supplied by the manufacturer in the intact kit, allowing direct comparison of measurements in pg/mL. All samples were run in duplicate in assays with inhouse within-run analytical CVs <7%.

Proprotein convertase activity assay
Proprotein convertase activity was measured in kidney homogenates using previously

Fibroblast culture and treatments
Primary cultures of fibroblasts propagated from normal kidneys (NRKF) and fibrotic kidneys (3 days after UUO; UUOF) of Sprague-Dawley rats were utilised for these studies, as described before 12  were captured with a Leica SP5 confocal microscope using a 63x oil immersion objective.

Cell transfection and luciferase reporter assays
Transient gene silencing in UUOF was performed using pre-designed siRNA (13 nM

RNA was extracted from homogenised tissue (TissueLyser LT) or cultured cells in 6
well plates using the Qiagen miRNeasy Mini kit according to manufactures' instructions. 1 µg of RNA was reverse-transcribed using the iScript RT supermix kit (Bio-Rad). Quantitative  Table S4). PCR conditions were set according to the manufacturer's instructions. Melt curve analysis was performed to verify the purity and specificity of the amplicons. Direct sequencing was used to confirm the specific amplification of FGF23 transcripts in whole kidney lysates and LCMD tissue. Threshold cycles were calculated using CFX Manager Software (Bio-Rad), and the mRNA level of target genes was normalised to the reference gene, β-actin, and expressed relative to an appropriate control (D0 or vehicle-treated) using the 2^(−ΔΔCt) method. For detection analyses, PCR products were resolved on 10% TBE pre-cast gels and visualised with SYBR Safe DNA gel stain (Invitrogen, Carlsbad, CA, USA) on a ChemiDoc MP imaging platform (Bio-Rad).
For LCMD tissue, RNA extraction and purification was performed using the Paradise PLUS Reagent System (Arcturus) according to the manufacturer's instructions with modifications as described previously 10 . The quality of the extracted RNA was assessed using the Paradise Quality Assessment kit with the ratio of 3'-and 5'-directed b-actin transcripts <10. First and second strand cDNA was synthesised using Superscript III RT (Invitrogen) and purified, before Whole Transcriptome Amplification (Arcturus) and purification of the amplified RNA. A second round of cDNA synthesis was performed to generate the template for gene expression.

RT 2 Profiler PCR array
Total RNA was isolated from 25cm 2

Western blot analysis
Western blotting studies were performed as previously described 15  To substantiate the specificity of the anti-mouse FGF23 antibody (#MAB2629, R&D Systems) used for our Western blotting studies we analysed purified mouse intact FGF23 and isolated C-terminal FGF23 fragments (Supplementary Fig. S6). Here, bands for intact and Cterminal FGF23 identified in kidney tissue, co-electrophorese with their corresponding purified recombinant intact (#2629-FG) and C-terminal peptides (Genscript). Neither intact nor Cterminal fragments are present in liver (a negative control), or in the kidney when the primary antibody is omitted (a technical control). Incubation of intact FGF23 peptides with furin resulted in the generation of C-terminal FGF23 fragments, but not with ADHR FGF23

Statistics
Results were analysed by one-way ANOVA with Holm-Sidak multiple comparisons test when comparing 3 or more groups or using a two-tailed unpaired Student's t-test when comparing 2 groups (GraphPad Prism 7.0, La Jolla, CA, USA). All data in this paper are presented as the mean ± SD with P < 0.05 defined as statistically significant.  Representative immunofluorescent staining of cytoskeletal proteins vimentin (green) and αSMA (red) in fibroblasts cultured from D0 control (NRKF) and D3 OB (UUOF) rat kidney explants treated for 48h with either TGFβ1 or FGF23 (both at 1 ng/mL). Nuclei were stained with DAPI (blue). Scale bar =50µm.