Primary mouse MC cultures were previously established. MC were outgrown from glomeruli of male C57BL/6 wild-type or Smad3 knockout mice isolated using Dynabeads (Invitrogen). Cells were cultured in Dulbecco’s modified eagle’s medium supplemented with 16% fetal bovine serum, streptomycin (100 μg/mL) and penicillin (100 μg/mL) at 37 °C in 95% O₂, 5% CO_2. Cells from passages 10 to 13 were used for transfection experiments. They were serum deprived at 80%–90% confluence in 1% bovine serum albumin (BSA) for 24 h following transfection and prior to treatment with any of 0.5 ng/mL TGFβ1 (R&D Systems), 94 mM Stat5 inhibitor (Cayman Chemical), 1 µg actinomycin D (Thermo Fisher) or 5uM Smad3 inhibitor SIS3 (Cayman Chemical).

DNAzol (Invitrogen) was used to isolate genomic DNA from mouse MC. The promoter region 2048bp upstream of the start codon for inhba (NCBI Reference Sequence: XM_011244285.3) was amplified using the following primers with MluI and XhoI restriction enzyme sites attached to the 5′ and 3′ end respectively: F5′-ACTACGCGTAAAACTGAAGTTTAACCTAGTGTC-3′, R5′- CCTGGCAGCAAAAGTCGTG -3’. The following forward primers were used to create deletion constructs in conjunction with the above reverse primer: 686bp-actA F5′-ACTACGCGTGAGCAAGGAGCAGCAAGAA-3′, 647bp-actA F5′-ACTACGCGTCCCTAGTCACAGCTCATACT-3′, 577bp-actA F5′-ACTACGCGTAATCCCTCTCTCTCTCTCTC-3′, 437bp-actA F5′-ACTACGCGTCTCTCTTCCTTTCTCCCTCC-3′, 350bp-actA F5′-ACTACGCGTCTGTCTCTCCCTCCCATC-3′, 248bp-actA F5′-ACTACGCGTCTCCCTCTCTCCCTCCCTCC-3′, 175-actA F5′-ACTACGCGTTGCATTCAGAGAGGGAACC -3’. PCR products were purified after gel electrophoresis and ligated into the pGL3 firefly luciferase plasmid (Promega) using appropriate restriction enzymes. All plasmids were sequenced.

Using the −350bp plasmid as a template, one predicted FoxP1 (AAACA) and one Stat5 (TTCACAGA) binding site were identified using MatInspector (Genomatrix, Germany). Each site was mutated separately to generate the 350bp ActA-FoxP1-mut and 350bp ActA-Stat5-mut, using two sets of primers, P1 and P2, respectively. Briefly, each construct was created using P1 and P2 to mutate the binding site. Two parts of each construct were created to mutate the region where the binding site for FoxP1 or Stat5 is located. An additional PCR was completed to create one final construct containing the full −350bp sequence containing the mutated site. These mutated −350bp constructs were then cloned into the high efficiency TOPO TA cloning vector. Established clones were selected, purified DNA was digested and ligated into a pGL3 luciferase vector. The following primers were used for site-directed mutagenesis: P1 350bp-FoxP1-mut; sense, F5′-ACTACGCGTCTGTCTCTCCCTCCCATC-3’; antisense, R5′-CTCTCCCTTGGACGAAACAT-3’. P2 350bp-FoxP1-mut; sense, F5′- GAGAGGGAACCTGCAAACAA-3’; antisense, R5′- CCTGGCAGCAAAAGTCGTG -3’. P1 350bp-Stat5-mut; sense, F5′-ACTACGCGTCTGTCTCTCCCTCCCATC-3’; antisense, R5′- TGTTTTGAAGTGTCTTTTG -3’. P2 350bp-Stat5-mut; sense, F5′- ACAAAACTAGTGTCTTAAC -3’; antisense, R5′- CCTGGCAGCAAAAGTCGTG -3’.

The pcDNA3.1 FoxP1A plasmid was a gift from Dr. A. Rao (Addgene #16362).

For luciferase experiments, 3 × 10⁵ MC were plated in 12-well plates in triplicate at 60%–70% confluence and transfected with 0.5 μg of promoter luciferase plasmid along with 0.05 μg pCMV β-galactosidase (Clontech) using Effectene (Qiagen). After the cells were harvested, luciferase and β-galactosidase activities were measured using kits (Promega) and a SpectraMax L Microplate Reader (Molecular Devices) set to measure luminescence or 420 nm absorbance respectively.

For siRNA transfection, 4 × 10⁵ MC were seeded in a 6-well plate and allowed to reach a confluence of 30%–40%. Knockdown of Sox9 and FoxP1 was achieved using 50 nM specific siRNA (Thermo Fisher) with RNAiMAX (Thermo Fisher).

TRIzol Reagent (Invitrogen) was used for RNA extraction. For all samples, 1 µg RNA was reverse transcribed to cDNA using qScript cDNA SuperMix Reagent (Quanta Biosciences) for quantitative real-time PCR using Power SYBR Green PCR Master Mix (Thermo Fisher Scientific). Specific primers used were: Inhba F5′-ACAGCCAGGAAGACACTGCA-3′ and R5′-CAGGTCACTGCCTTCCTTGG-3’; 18S F5′-GCCGCTAGAGGTGAAATTCTTG-3′, R5′-CATTCTTGGCAAATGCTTTCG-3′ as internal control. Gene expression was calculated using the ∆∆C_T method.

Protein was extracted using lysis buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM EGTA, 2.5 mM sodium pyrophosphate, 1 mM β-glycerophosphate, 2 mM DTT, 1 mM sodium vanadate, 1 mM phenylmethylsulphonyl fluoride, 1 μg/mL leupeptin, 2 μg/mL aprotinin). Cell lysates were then centrifuged and the resulting supernatant used to measure activin A (R&D Systems), Smad3 (Abcam), FoxP1 (Cell Signaling), Stat5 (R&D Systems), Sox9 (R&D Systems), tubulin (Sigma) and GAPDH (Millipore) levels by immunoblotting or with the activin A Quantikine ELISA Kit (R&D Systems).

As described previously (Uttarwar et al., 2012), cells were lysed using hypotonic buffer for nuclear protein purification. Once centrifuged, pelleted nuclei were sonicated with hypotonic buffer composed of 0.4M NaCl and 10% glycerol. Equal amounts of nuclear protein were immunoprecipitated using 1 µg Smad3 antibody overnight at 4 °C, followed by incubation with protein G-agarose slurry for 2 h (4 °C). After washing, total immunoprecipitated products were separated on SDS-PAGE for immunoblotting.

This was conducted as previously described (Mehta et al., 2019). Briefly, after MC were treated with TGFβ1 for 6 h, protein-DNA complexes were crosslinked using formaldehyde, nuclear extracts were sheared by sonication followed by immunoprecipitation with FoxP1, Stat5 or nonspecific IgG antibodies (1 µg). After reversing crosslinking, DNA was isolated and used for qRT-PCR using primers for the −350bp promoter region. 10% of the original nuclear isolate before immunoprecipitation was used as input. Ct values were evaluated across 3 separate experiments, with target enrichment of Stat5 or FoxP1 calculated by normalizing to % input.

Animal studies were previously carried out in accordance with the principles of laboratory animal care and McMaster University and Canadian Council on Animal Care guidelines (protocol #22–05–17). Male C57BL/6 mice underwent resection of the upper and lower poles of the left kidney at 6–7 weeks of age, followed by right nephrectomy after a 1-week recovery period. Kidneys were harvested after 16 weeks and stored in OCT compound. For immunofluorescence, 8 µm sections were cut, fixed in 3.7% paraformaldehyde, permeabilized with 0.2% Triton X-100 and blocked prior to overnight incubation at 4 °C with a primary antibody against activin A (1:200, R&D Systems). The following day, primary antibodies against FoxP1 (1:50) or Stat5 (1:1000) were applied overnight at 4 °C, followed the next day by incubation with secondary antibodies (Alexa Fluor 488-conjugated anti-goat and Alexa Fluor 568-conjugated anti-rabbit from Invitrogen) for 30 min at room temperature. In a second 5/6 nephrectomy model, male CD-1 mice were used as they display greater injury than the C57BL/6 strain (protocol #23–40). Kidneys in this model were harvested 9 weeks after 5/6 nephrectomy and sections used for costaining with the MC marker integrin α8 (1:250, R&D Systems). Images were acquired using a Nikon Eclipse Ti2 microscope at ×20 magnification.

Values are presented as mean ± SEM. Statistical difference among multiple groups was determined using ANOVA with a Tukey’s post hoc test. Unpaired, two-tailed Student t tests were used for single comparisons. P value <0.05 was considered statistically significant, using GraphPad Prism 7 for calculations.

We and others have shown increased expression of activin A in both rodent and human chronic kidney disease (Zhang et al., 2019; Soomro et al., 2023; Tsai et al., 2023). To identify whether increased inhba transcript is also seen, we accessed the Nephroseq database (www.nephroseq.org) and extracted gene expression information for inhba from the Nakagawa study (GSE66494) which included pathologist-confirmed samples for chronic kidney disease (n = 5) and normal (n = 3) human kidney samples. These data show that inhba gene expression is significantly increased in chronic kidney disease (Figure 1A).

We have previously shown that TGFβ1 increases synthesis and secretion of activin A in MC (Soomro et al., 2023). To determine whether this was dependent on induction of inhba gene expression, MC were treated for 24 h with TGFβ1 and inhba assessed by qRT-PCR. Figure 1B shows a significant increase in transcript levels. We next assessed response to TGFβ1 in the presence of actinomycin D, an inhibitor of transcription, to determine whether RNA synthesis was required. MCs were treated with TGFβ1 for 3 h before the addition of actinomycin D and transcript levels were then assessed at 0.5, 1 and 3 h following treatment, as show in Figure 1C. As shown in the graph, TGFβ1 induced ongoing synthesis of inhba which was prevented by inhibition of transcription. These data show a direct transcriptional effect of TGFβ1 on inhba.

To understand how TGFβ1 regulates inhba transcription, we cloned 2048 base pairs of the inhba promoter upstream of its transcription start site (TSS) (NCBI gene ID 16323) and inserted it upstream of the luciferase gene. This is depicted in Figure 2A which also shows the inhba gene structure. We first confirmed activation of the promoter luciferase reporter by TGFβ1 (Figure 2B). To identify regions necessary for the TGFβ1 response, we then created a series of deletion constructs as shown in Figure 2C. These were tested with TGFβ1 0.5 ng/mL. As seen in Figure 2D, the full promoter as well as subsequent deletion constructs were all responsive to TGFβ1, with the exception of the shortest sequence (−175bp). These findings indicate that important TGFβ1-regulatory elements reside in the −350 to −175 region of the inhba promoter.

Interestingly, the −350bp construct showed a more robust response to TGFβ1 compared to the −437bp construct. Examination of this region identified a microsatellite region containing a high degree of repetition of the dinucleotides cytosine and thymidine, shown in purple in the boxed region in Figure 2A. Microsatellite regions control the degree of transcriptional activity depending on their location relative to the TSS, causing an expansion or contraction of the length of that region to form an open or closed secondary configuration, respectively (Bagshaw, 2017). The effect of the microsatellite region on overall promoter activity is highlighted in Figure 2E, in which the promoter activity data shown in Figure 2D are presented relative to the −2048bp basal promoter activity rather than to their own internal control. Here, removal of the CT island in the −350bp construct significantly increased basal activity of the promoter. These data imply that the microsatellite region has an overall silencing effect on inhba promoter activity.

To identify the regulatory elements within the −350 to −175bp region specific to TGFβ1 induction of inhba, a smaller deletion construct was created (−284bp). Figure 3A shows the absence of induction by TGFβ1 for this construct and indicates that important TGFβ1-responsive elements reside in the −350 to −284bp region. Analysis of this region using MatInspector identified several potential regulators as shown in Figure 3B. We chose three transcription factors for further analysis based on their previously identified role as mediators of TGFβ1 profibrotic effects. These include Sex-determining region Y-box gene 9 (Sox9), forkhead box transcription factor 1 (FoxP1) and signal transducer and activator of transcription 5 (Stat5) (Bot et al., 2011; Li et al., 2018; Massague and Sheppard, 2023). Further studies used the −350bp promoter construct.

Using siRNA downregulation, we first assessed whether Sox9 was required for inhba upregulation by TGFβ1. As shown in Figure 3C, this did not reduce TGFβ1-induced inhba promoter activation, with confirmation of Sox9 downregulation shown in Figure 3D. However, siRNA downregulation of FoxP1 abrogated induction of promoter activity by TGFβ1 (Figure 3E), with its downregulation also confirmed in Figure 3F. Mutation of the FoxP1 regulatory element (Figure 3G) similarly prevented promoter induction by TGFβ1 (Figure 3H). The importance of FoxP1 for regulation of the full promoter construct, as well as synthesis of activin A, was confirmed in Figures 3I,J respectively. Fold changes were calculated by normalizing the measured values of the treatment to the corresponding controls within each siRNA group.

We next assessed whether Stat5 was required for promoter activation. Figure 4A shows that a Stat5 inhibitor prevented TGFβ1-induced inhba promoter activation, as did mutation of the Stat5 binding element (Figures 4B,C). Stat5 inhibition also attenuated TGFβ1-induced activation of the full promoter, although not as effectively as for the shorter promoter (Figure 4D). However, a significant reduction in TGFβ1-induced activin A protein production as assessed by ELISA was seen with Stat5 inhibition (Figure 4E).

To confirm binding of FoxP1 and Stat5 to the −350bp promoter region, we performed ChIP assays. Figures 4F,G show binding of both of these transcription factors to this promoter region, with no binding seen when using control IgG for immunoprecipitation from nuclear lysates. These data overall identify an important role for FoxP1 and Stat5, but not Sox9, in regulation of inhba by TGFβ1.

Smad3 is the canonical mediator of TGFβ1 signaling. While we did not identify any Smad binding elements in the −350bp promoter region, several of these were identified upstream of this in the full promoter. We thus assessed the importance of Smad3 to promoter activation by TGFβ1 in MC derived from Smad3 knockout compared with wild-type mice. Western blotting for Smad3 confirmed absence of Smad3 in knockout MC compared with wild-type cells (Figure 5A). Figure 5B shows that activation of the full promoter by TGFβ1 was abolished in the absence of Smad3, identifying a critical role for this transcription factor in regulating TGFβ1-induced inhba expression.

Interestingly, Smad3 deletion also prevented TGFβ1-induced activation of the −350bp promoter (Figure 5C), although this region lacks a Smad binding element. This may be through Smad3 regulation of other transcription factors. We next confirmed lack of inhba transcript induction by TGFβ1 in Smad3 knockout cells, shown in Figure 5D. Data are presented normalized to their own genotype. Inhibition of Smad3 using SIS3 similarly prevented activation of both the full length and −350bp promoter regions by TGFβ1 (Figures 5E,F). These data highlight an important role for Smad3 in regulation of inhba synthesis in response to TGFβ1.

To further investigate the potential interaction of FoxP1 and Stat5 with Smad3, we immunoprecipitated Smad3 from nuclear lysate after treatment with TGFβ1 and probed for FoxP1 or Stat5. Figures 6A,B and show that association of both transcription factors increased significantly in response to TGFβ1. Furthermore, overexpression of FoxP1 in Smad3 wild-type MC dose-dependently increased activity of the −350bp construct (Figure 6C) as well as activin A protein production (Figure 6D), but promoter luciferase activation was completely abrogated in Smad3 knockout cells (Figure 6E). Together, these data suggest that Smad3 interaction with FoxP1 and Stat5 supports their regulation of the inhba promoter.

We next sought in vivo support for FoxP1 and Stat5 involvement in the upregulation of activin A. We used the well-established 5/6 nephrectomy model of chronic kidney disease to assess whether increased FoxP1 or Stat5 could be colocalized with elevated expression of activin A. We performed immunofluorescence staining of kidneys. In Figures 6F,G, increased colocalization of activin A with FoxP1 or Stat5 is seen after 5/6 nephrectomy. Boxes identify glomeruli which are shown magnified, and suggest presence within areas of MC in addition to the tubular cell staining that is seen. Colocalization using the MC marker integrin α8 is shown in Supplementary Figure S1. These data support the regulation of activin A by these transcription factors in vivo.