Volume 180, Issue 22 p. 2898-2915
THEMED ISSUE ARTICLE
Open Access

Single cell G-protein coupled receptor profiling of activated kidney fibroblasts expressing transcription factor 21

Harmandeep Kaur

Harmandeep Kaur

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Contribution: Conceptualization (lead), Data curation (lead), Formal analysis (lead), ​Investigation (lead), Methodology (lead), Validation (lead), Visualization (lead), Writing - original draft (supporting), Writing - review & editing (supporting)

Search for more papers by this author
Veera Ganesh Yerra

Veera Ganesh Yerra

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Contribution: Data curation (supporting), ​Investigation (supporting), Methodology (supporting), Writing - review & editing (supporting)

Search for more papers by this author
Sri Nagarjun Batchu

Sri Nagarjun Batchu

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Contribution: Data curation (supporting), ​Investigation (supporting), Methodology (supporting), Writing - review & editing (supporting)

Search for more papers by this author
Duc Tin Tran

Duc Tin Tran

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Contribution: ​Investigation (supporting), Validation (supporting)

Search for more papers by this author
M. D. Golam Kabir

M. D. Golam Kabir

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Contribution: ​Investigation (supporting), Methodology (supporting)

Search for more papers by this author
Youan Liu

Youan Liu

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Contribution: ​Investigation (supporting), Methodology (supporting), Project administration (supporting)

Search for more papers by this author
Suzanne L. Advani

Suzanne L. Advani

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Contribution: ​Investigation (supporting), Methodology (supporting), Project administration (supporting)

Search for more papers by this author
Phelopater Sedrak

Phelopater Sedrak

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Temerty Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada

Contribution: Data curation (supporting), ​Investigation (supporting), Methodology (supporting)

Search for more papers by this author
Laurette Geldenhuys

Laurette Geldenhuys

Department of Pathology, Dalhousie University, Halifax, Nova Scotia, Canada

Contribution: Resources (supporting), Writing - review & editing (supporting)

Search for more papers by this author
Karthik K. Tennankore

Karthik K. Tennankore

Department of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada

Contribution: Resources (supporting), Writing - review & editing (supporting)

Search for more papers by this author
Penelope Poyah

Penelope Poyah

Department of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada

Contribution: Resources (supporting)

Search for more papers by this author
Ferhan S. Siddiqi

Ferhan S. Siddiqi

Department of Medicine, Dalhousie University, Halifax, Nova Scotia, Canada

Contribution: Resources (supporting), Writing - review & editing (supporting)

Search for more papers by this author
Andrew Advani

Corresponding Author

Andrew Advani

Keenan Research Centre for Biomedical Science and Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, Ontario, Canada

Correspondence

Andrew Advani, Keenan Chair in Medicine, Associate Professor/Clinician Scientist, St. Michael's Hospital, 6-151 61 Queen Street East, Toronto, Ontario, Canada, M5C 2T2.

Email: [email protected]

Contribution: Conceptualization (lead), Data curation (lead), Formal analysis (lead), Funding acquisition (lead), ​Investigation (lead), Project administration (lead), Resources (lead), Supervision (lead), Visualization (lead), Writing - original draft (lead), Writing - review & editing (lead)

Search for more papers by this author
First published: 28 April 2023
Citations: 3

Abstract

Background and Purpose

Activated fibroblasts deposit fibrotic matrix in chronic kidney disease (CKD) and G-protein coupled receptors (GPCRs) are the most druggable therapeutic targets. Here, we set out to establish a transcriptional profile that identifies activated kidney fibroblasts and the GPCRs that they express.

Experimental Approach

RNA sequencing and single cell qRT-PCR were performed on mouse kidneys after unilateral ureteral obstruction (UUO). Candidate expression was evaluated in mice with UUO or diabetes or injected with adriamycin or folic acid. Intervention studies were conducted in mice with diabetes or UUO. Correlative histology was performed in human kidney tissue.

Key Results

Transcription factor 21 (Tcf21)+ cells that expressed 2 or 3 of Postn, Acta2 and Pdgfra were highly enriched for fibrogenic genes and were defined as activated kidney fibroblasts. Tcf21+ α-smooth muscle actin (α-SMA)+ interstitial cells accumulated in kidneys of mice with UUO or diabetes or injected with adriamycin or folic acid, whereas renin-angiotensin system blockade attenuated increases in Tcf21 in diabetic mice. Fifty-six GPCRs were up-regulated in single Tcf21+ kidney fibroblasts, the most up-regulated being Adgra2 and S1pr3. Adenosine receptors, Adora2a/2b, were up-regulated in Tcf21+ fibroblasts and the adenosine receptor antagonist, caffeine decreased Tcf21 upregulation and kidney fibrosis in UUO mice. TCF21, ADGRA2, S1PR3 and ADORA2A/2B were each detectable in α-SMA+ interstitial cells in human kidney samples.

Conclusion and Implications

Tcf21 is a marker of kidney fibroblasts that are enriched for fibrogenic genes in CKD. Further analysis of the GPCRs expressed by these cells may identify new targets for treating CKD.

LINKED ARTICLES

This article is part of a themed issue on Translational Advances in Fibrosis as a Therapeutic Target. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v180.22/issuetoc

Graphical Abstract

Abbreviations

  • bHLH
  • basic helix-loop-helix
  • CAD
  • coronary artery disease
  • CKD
  • chronic kidney disease
  • DKD
  • diabetic kidney disease
  • DM-HFD
  • diabetic high fat diet-fed
  • FPKM
  • fragments per kilobase per million mapped reads
  • FSGS
  • focal segmental glomerulosclerosis
  • GO
  • Gene Ontology
  • GPCR
  • G-protein coupled receptor
  • IFC
  • integrated fluidic circuit
  • KEGG
  • Kyoto Encyclopedia of Genes and Genomes
  • KIM-1
  • kidney injury molecule-1
  • MI
  • myocardial infarction
  • NGAL
  • neutrophil gelatinase-associated lipocalin
  • PCA
  • principal component analysis
  • qRT-PCR
  • quantitative reverse transcription polymerase chain reaction
  • RAS
  • renin angiotensin system
  • S1P
  • sphingosine-1-phosphate
  • Tcf21
  • transcription factor 21
  • t-SNE
  • t-distributed stochastic neighbour embedding
  • UUO
  • unilateral ureteral obstruction
  • α-SMA
  • α-smooth muscle actin
  • What is already known?

    • Chronic kidney disease (CKD) is characterized by progressive fibrosis mediated by activated kidney fibroblasts.
    • As they are expressed on cell surfaces, GPCRs are common targets for therapeutic agents.

    What does this study add?

    • Tcf21 expression helps to identify kidney fibroblasts that are enriched for fibrogenic genes in CKD.
    • Fifty-six GPCRs were up-regulated in Tcf21+ kidney fibroblasts including Adgra2, S1pr3, Adora2a and Adora2b.

    What is the clinical significance?

    • Study of the GPCRs expressed by Tcf21+ fibroblasts could lead to new treatments for CKD

    1 INTRODUCTION

    Regardless of their underlying aetiologies, all kidney injury processes that ultimately cause chronic kidney disease (CKD) follow the same common pathway of progressive kidney fibrosis (Zeisberg & Neilson, 2010). Consequently, there are many academic and industrial teams, worldwide, actively exploring strategies to stop or potentially even reverse fibrosis, as a means of improving kidney outcomes. As they are expressed on the cell surface and known to be susceptible to pharmacological modulation, GPCRs are often viewed as appealing therapeutic targets. Indeed, about a third of all approved drugs target GPCRs, making GPCR modulators the single largest class of any drug (Hauser et al., 2018). GPCRs, though, are also often low in abundance and heterogeneous in their patterns of expression (Kaur et al., 2017). These characteristics can hamper efforts to discover GPCRs that may play a role in tubulointerstitial fibrogenesis.

    The cells principally responsible for the deposition of extracellular matrix within the tubulointerstitium are activated kidney fibroblasts (Kuppe et al., 2021). Activated fibroblasts (also called myofibroblasts) are commonly identified by the de novo expression of α-smooth muscle actin (α-SMA) (Hewitson, 2009). However, α-SMA expression alone is an inconsistent marker of fibroblast activation and collagen production (Sun et al., 2016). In cardiac tissue, transcription factor 21, and its gene, Tcf21, is garnering attention as a marker of resting fibroblasts that give rise to myofibroblasts responsible for scar formation after myocardial infarction (MI) (Kanisicak et al., 2016). Tcf21 is sometimes referred to as Pod1 (also called capsulin or epicardin) and, in the kidney, expression of Tcf21/Pod1 was necessary for the differentiation and maintenance of glomerular podocytes (Maezawa et al., 2014; Quaggin et al., 1999). Tcf21, though, is also expressed by a subset of kidney interstitial cells into adulthood (Quaggin et al., 1999) and, in a recent expression profiling study, Tcf21 was the most highly up-regulated transcription factor in kidney fibroblasts of mice injected with folic acid or following unilateral ureteral obstruction (UUO) (Higashi et al., 2019). The utility of Tcf21 as a marker for identifying fibroblasts in kidney disease, however, has not previously been explored.

    In the present study, we sought to take advantage of recent developments in single cell isolation technologies (Kaur & Advani, 2021) to define the GPCRs that are expressed by activated kidney fibroblasts. In doing so, we established that Tcf21+ fibroblasts are highly enriched for fibrogenic genes in mouse models of kidney disease.

    2 METHODS

    2.1 Mouse models of kidney disease

    All animal care and experimental procedures adhered to the guidelines of the Canadian Council of Animal Care and were approved by the St. Michael's Hospital Animal Care Committee. Animal studies are reported in compliance with the ARRIVE guidelines (Percie du Sert et al., 2020) and with the recommendations made by the British Journal of Pharmacology (Lilley et al., 2020).

    Animals were housed in a specific pathogen free facility in IVC cages, temperature 24 ± 2°C, 12-h light/dark cycle and 45%–65% relative humidity. Each ventilated plastic cage was supplemented with Bed-o'Cobs bedding material, cotton fiber Nestlets and a PVC tube. A maximum of five mice were housed in each cage. Mice were acclimatized to the environment for a minimum of 5 days before study.

    Unilateral ureteral obstruction (UUO) or sham surgeries were performed in male C57BL/6N mice (C57BL/6N/Crl; Charles River Laboratories, Senneville, Quebec, Canada) aged ~8 weeks, as previously described (Batchu et al., 2021). Briefly, mice were anaesthetized with 2% isoflurane and an incision was made in the left flank before occlusion of the left ureter distal to its origin using two 5-0 silk sutures. Sham mice underwent the same procedure without ligation of the left ureter. Analgesia was achieved with slow-release buprenorphine 0.5mg/kg s.c. administered preoperatively. Precautions to ensure surgical asepsis included shaving of the surgical site and application of alcohol and then iodine solution, repeated twice, before draping with sterile gauze. Post-operatively, mice were recovered separately in a clean cage without bedding and on a warm water blanket (37oC). Mice were followed for 3, 7 or 14 days post-surgery and were killed by cervical dislocation after anaesthesia with 5% isoflurane

    For adriamycin (doxorubicin) nephrotoxicity studies, 8 week old male BALB/c mice (Charles River Laboratories) received a single tail-vein injection of adriamycin (10 mg·kg−1) in PBS or PBS alone and were followed for 10 days, as previously described (Majumder et al., 2018). For folic acid nephrotoxicity studies, male C57BL/6N mice (Charles River Laboratories) aged ~8 weeks received a single intraperitoneal injection of folic acid (250 mg·kg−1 in 300 mM NaHCO3) or NaHCO3 alone and were followed for 14 days.

    For diabetes studies, diabetic high fat diet-fed (DM-HFD) mice were generated and treated with either the ACE inhibitor ramipril or the AT1 receptor antagonist telmisartan, as previously described (Batchu et al., 2021). Briefly, male C57BL/6N mice (Charles River Laboratories) aged 7–8 weeks were fed a high fat diet (HFD; 45% kcal fat, 35% kcal carbohydrate, 0.05% w/w cholesterol; Research Diets Inc., New Brunswick, NJ) for 15 weeks before receiving an intraperitoneal injection of streptozotocin (90 mg·kg−1 in 0.1 mol·L−1 sodium citrate, pH 4.5, after a 4 h fast) and were maintained on a HFD for a further 21 weeks (Batchu et al., 2021). After 19 weeks, mice were treated with either ramipril (10 mg·kg−1·day−1) or telmisartan (3 mg·kg−1·day−1) in drinking water, or drinking water alone for 2 weeks. Male age-matched controls were fed standard chow and given normal drinking water (Batchu et al., 2021).

    For studies with caffeine, male C57BL/6N mice (Charles River Laboratories) aged ~8 weeks underwent UUO or sham surgery as already described and were treated with either drinking water or caffeine at either 0.4 mg·ml−1 (Lu et al., 2007) or 1 mg·ml−1 (Costa et al., 2008) in drinking water beginning on the day of surgery and continued for 14 days. Systolic blood pressure was measured using a CODA noninvasive blood pressure system (Kent Scientific, Torrington, CT) (Yuen et al., 2012). Plasma levels of cystatin C was measured by ELISA (ab201280, Abcam, Cambridge, MA).

    Mice were selected for study because the goal was to identify GPCRs that are expressed by single kidney fibroblasts isolated from fibrotic kidneys and to generate proof-of-principle evidence that pharmacological antagonism of the identified GPCR(s) attenuates tissue fibrosis. The mouse models studied (UUO, folic acid nephrotoxicity, adriamycin nephrotoxicity and diabetic high fat diet) were considered appropriate to reflect a range of fibrotic kidney diseases of different aetiologies. The study was designed to generate groups of equal size using randomization, with blinded analysis where feasible. Sample size was based on practicability and initial phenotyping experiments performed in sham and UUO mice (Figure S1). Humane endpoints were assessed and determined according to the vivarium's Welfare Assessment Scoring System

    2.2 Picrosirius red staining

    Kidney fibrosis was assessed by staining kidney cross sections with picrosirius red. Stained sections were digitized with an Axio Scan.z1 (Carl Zeiss Microscopy, Jena, Germany) and the proportional area positively staining red was analysed using HALO® image analysis platform (Indica Labs, Albuquerque, NM) or ImageJ version 1.39 (National Institutes of Health, Bethesda, MD) in a masked manner.

    2.3 Isolation of primary fibroblast-enriched kidney cells

    Primary fibroblast-enriched cells were isolated from the left (obstructed) and right (unobstructed) kidneys of mice 3, 7 or 14 days after UUO or from control mice as previously described (Batchu et al., 2021) and as adapted from a method previously used in mouse hearts (Kaur et al., 2016). Briefly, tissue was mechanically and enzymatically digested using digestion mix containing 2 mg·ml−1 collagenase II (Worthington Biochemical Corp. Lakewood, NJ) and 5 units·ml−1 DNase 1 (MilliporeSigma) at 37°C with gentle shaking for 60–90 min. Cells were filtered through 70 μm and 40 μm sieves and washed with PBS before plating for 1 h at 37°C and 5% CO2 in DMEM containing 10% FBS, L-glutamine and 1% penicillin/streptomycin. Adherent cells were washed thoroughly with PBS before RNA isolation or single cell isolation.

    2.4 RNA sequencing

    RNA sequencing was performed using the 6G RNA Sequencing Service (150 bp paired-end, 40 million reads) for mouse kidneys (n = 5 per group) and the 12G RNA Sequencing Service (150 bp paired-end, 80 million reads) for fibroblast-enriched kidney cells (n = 3 control, n = 4 day 3 UUO, n = 4 day 7 UUO) by Arraystar Inc. (Rockville, MD). Briefly, following RNA quantification using a Nanodrop ND-1000 instrument, total RNA was enriched using oligo (dT) magnetic beads and RNA sequencing libraries were prepared using a KAPA Stranded RNA-Seq Library Prep Kit (Illumina, San Diego, CA). Sequencing (150 cycles for both ends) was performed on an Illumina Novaseq 6000. Image analysis and base calling were performed using Solexa pipeline v1.8 and sequence quality was examined using FastQC. Trimmed reads (trimmed 5′, 3′-adaptor bases using cutadapt) were aligned to the reference genome (GRCm38) using Hisat2 software. Transcript abundances were estimated using StringTie (Pertea et al., 2015) and the fragments per kilobase per million mapped reads (FPKM) and differential gene expression were determined using the R package Ballgown (Frazee et al., 2015). Volcano plots, Principal Component Analysis (PCA), Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed with the differentially expressed genes in R, Python or shell environment (Batchu et al., 2020). RNA sequencing data are available at Gene Expression Omnibus (accession numbers GSE198962 and GSE198829).

    2.5 Quantitative reverse transcription polymerase chain reaction in fibroblast-enriched cells and NRK-49F cells

    RNA was isolated from fibroblast-enriched cells using TRIzol reagent (Thermo Fisher Scientific, Waltham, MA) and cDNA was reverse transcribed from 1 μg RNA using Superscript III Reverse Transcriptase (Thermo Fisher Scientific). SYBR green-based quantitative RT-PCR (qRT-PCR) was performed on a QuantStudio 7 Flex Real-Time PCR System (Thermo Fisher Scientific) using the following primer sequences from Integrated DNA Technologies (Coralville, IA): Postn, forward 5′-CCATTGGAGGCAAACAACTCC-3′, reverse 5′-TTGCTTCCTCTCACCATGCA-3′; Pdgfra, forward 5′-CAAAGGGAGGACGTTCAAGAC-3′, reverse 5′-TGCGTCCATCTCCAGATTCA-3′; Ccn2, forward 5′-AAGCTGACCTGGAGGAAAACA-3′, reverse 5′-TGCAGCCAGAAAGCTCAAAC-3′; Tcf21, forward 5′-CATTCACCCAGTCAACCTGA-3′, reverse 5′-CCACTTCCTTCAGGTCATTCTC-3′; Gapdh, forward 5′-AGACGGCCGCATCTTCTT-3′, reverse 5′-TTCACACCGACCTTCACCAT-3′. For adenosine A2A and A2B receptor agonism/antagonism studies, NRK-49F renal fibroblasts (CRL-1570, ATCC, Manassas, VA) were incubated with 4-mmol·L−1 caffeine (H. Wang et al., 2014), 1-μmol·L−1 CGS 21680 (#1063, Bio-Techne, Toronto, Ontario, Canada) (H. Wang et al., 2014) and/or 5-μmol·L−1 BAY60-6583 (Gao et al., 2014) (HY-103171, MedChemExpress, Monmouth Junction, NJ) for 48 h and qRT-PCR was performed using the following primer sequences: Tcf21, forward 5′-GATGCTGGACTGTGACTCCC-3′, reverse 5′-CCTTCTGTGGAGACCCGTTC-3′; Ccn2, forward 5′-CCCGATGGCGAGATCATGAA-3′, reverse 5′-TGTCCCTTACTCCCTGGCTT-3′; Gapdh, forward 5′-GAACGGGAAGCTCACTGG-3′, reverse 5′-GCCTGCTTCACCACCTTCT-3′. Data analysis was performed using the comparative ΔΔCT method.

    2.6 Single cell qRT-PCR

    Fibroblast-enriched cells were isolated from the kidneys of mice 3 days after UUO surgery as already described. Approximately 2000 cells at a time were loaded in a volume of 5 μl C1 suspension reagent (Fluidigm, San Francisco, CA) onto the Fluidigm C1 Single-Cell Auto Prep System (17-25 μm integrated fluidic circuits [IFCs], Fluidigm), followed by RNA isolation and cDNA synthesis according to the manufacturer's protocol. Individual chambers were inspected by optical microscopy to eliminate empty chambers or chambers containing more than one cell. qRT-PCR was performed on individual cells for the housekeeper gene Hprt forward 5′- CAGTACAGCCCCAAAATGGTTA-3′, reverse 5′-AGTCTGGCCTGTATCCAACA-3′ and only cells in which Hprt CT was <25 were studied. High throughput quantitative PCR was performed on single cell cDNAs using a BioMark™ 96.96 Dynamic Array (Fluidigm) with Sso-Fast EvaGreen Supermix low ROX (BioRad, Hercules, CA) and Delta Gene primer assays (Fluidigm) as listed in Table S1. Statistical analyses were performed by the Canadian Centre for Computational Genomics (C3G, Montreal, Quebec, Canada). To determine relative gene expression in individual cells, qRT-PCR based expression values in individual cells were first converted to Log2Ex values as previously described (Livak et al., 2013), where Log2Ex = LoD CT − CT [Gene], if CT ≤ LoD, then Log2Ex = LoD − CT; if CT ˃ LoD, then Log2Ex = 0; LoD default value 24. Differential expression was performed using the R package limma (Ritchie et al., 2015), comparing cells from designated cell-types to all remaining cells. Nominal P values were corrected for multiple testing using the Benjamini-Hochberg method. Clustering and visualization of cells were performed using t-distributed stochastic neighbour embedding (t-SNE; R package Rtsne). Identification of outlier samples was achieved by performing unsupervised hierarchical clustering of cells based on 5 housekeeping genes (Gapdh, Rpl13a, Gusb, Hprt, Hsp90ab1). A total of 37 cells, which did not express at least 2 housekeeping genes were removed. After removing outliers, cell-types were defined using a combination of markers (Acta2, Cdh5, Epcam, Pdgfra, Postn, Ptprc, Tcf21, Nphs2). Violin plots were generated using the function geom_violin from the R package ggplot2 (https://ggplot2.tidyverse.org/reference/geom_violin.html). Heatmaps were generated using the Complex Heatmap R package (https://jokergoo.github.io/ComplexHeatmap-reference/book/).

    2.7 RNAscope in situ hybridization

    RNAscope in situ hybridization (Advanced Cell Diagnostics [ACD], Hayward, CA) was performed according to the manufacturer's instructions and using custom software as previously reported (F. Wang et al., 2012) with the following probesets: TCF21 (mouse #508661, human #470371), ADGRA2 (mouse #318031, human #53651), S1PR3 (mouse #435951, human #537881), ADORA2A (mouse #409431, human #500081) and ADORA2B (mouse #115281, human #457661). Hybridization signals were detected using RNAscope 2.5HD Detection Reagent-RED (#322360) for dual stains and RNAscope 2.5HD detection reagent-BROWN (#322310) for single stains (ACD). After hybridization, immunohistochemistry was performed with the following antibodies: α-smooth muscle actin (αSMA) (Abcam rabbit polyclonal, catalogue number ab5694, batch/lot number GR283004–13, epitope N terminal, isotype IgG, RRID:AB_2223021, diluted 1:80 in PBS, solution not re-used; secondary antibody, Agilent Dako Envision+ System-HRP Labelled Polymer Anti-Rabbit, catalogue number K400311-2, lot/batch number 11240197, reacts with heavy and light chains of IgG, isotype IgG (Agilent Technologies, Inc., Santa Clara, CA); Agilent Dako Protein Free Serum Block Ready to Use X0909, lot number 11401570) and nephrin (R & D Systems (Minneapolis, MN) goat polyclonal, catalogue number AF3159, batch/lot number CBIK0215071, epitope extracellular domain of mouse nephrin Gln37–Thr1049, isotype IgG, RRID:AV_2155023, diluted 1:400 in PBS, solution not re-used; Cedarlane HRP Affinity Purified Rabbit anti-Goat, catalogue number CLDB2000, lot number A1618, isotype IgG, diluted 1:200, solution not re-used; Agilent Dako Protein Free Serum Block Ready to Use X0909, lot number 11401570). Experimental details conform with British Journal of Pharmacology guidelines (Alexander et al., 2018). Cells positive for Tcf21 positive puncta were counted in the glomeruli and intersitium in six randomly selected cortical fields (x 100 magnification) in kidney sections of control and DM-HFD mice by an investigator masked to the study groups.

    2.8 Human kidney tissue

    Single RNAscope in situ hybridization was performed on post mortem tissue from patients without history of kidney disease (control), obtained from the National Disease Research Interchange (NDRI: Philadelphia, PA). Dual RNAscope in situ hybridization and immunostaining for α-SMA was performed on archival formalin-fixed, paraffin-embedded kidney tissue obtained at the time of nephrectomy for conventional renal carcinoma from patients without diabetes and with normal kidney function (control) or from patients with diabetic nephropathy (Majumder et al., 2018). A minimum of four kidney sections were examined for each stain and from each study group. The study was approved by the Research Ethics Board of St. Michael's Hospital and was conducted in accordance with Declaration of Helsinki principles. Research ethics board approval was also provided by the Nova Scotia Health Authority Research Ethics Board for the archival kidney tissue studies, together with a waiver of consent based on impracticability criteria.

    2.9 Data and statistical analysis

    Data analysis in this manuscript complies with the British Journal of Pharmacology's recommendations and requirements on experimental design and analysis (Curtis et al., 2022). Data are expressed as mean ± S.D.. Statistical significance was determined by one-way ANOVA with a Fisher least significant difference post-test (three groups), Tukey's post-test (four groups) or unpaired two-tailed Student's t test (two groups), unless otherwise stated. For multiple comparisons, post hoc testing was only conducted if F in ANOVA achieved P <0.05 and there was no significant variance inhomogeneity. Where variance inhomogeneity was detected by Brown–Forsythe test, statistical comparison was performed after log-transformation or with Welch's ANOVA followed by Dunnett T3 post hoc test as stated. Statistical analysis was undertaken only for studies where each group's size was at least n = 5. RNA sequencing of fibroblast-enriched kidney cells was performed on n = 3–4 samples per condition to curate a list of candidate GPCRs which did not require statistical analysis. Outlying values were included in the statistical analysis except as stated for the single cell qRT-PCR screen. Representative in situ hybridization photomicrographs were taken on n ≥ 4 kidney sections per probe and were not subjected to statistical analysis unless stated. Group size is the number of independent values, and statistical analysis was done using these values. Statistical analyses were performed using GraphPad Prism 8 for macOS (GraphPad Software Inc., San Diego, CA). A P value <0.05 was considered statistically significant.

    2.10 Materials

    Ramipril (#15558) and telmisartan (#11615) were supplied by Cayman Chemicals (Ann Arbor, MI, USA); caffeine (C0750) and streptozotocin by MilliporeSigma (Oakville, Ontario, Canada); folic acid by BioShop Canada Inc., (Burlington, Ontario, Canada). CGS 21680 (#1063) was supplied by Bio-Techne (Toronto, Ontario, Canada); BAY60-6583 (HY-103171) and adriamycin by MedChemExpress (Monmouth Junction, NJ, USA). The HFD was supplied by Research Diets Inc. (New Brunswick, NJ, USA).

    2.11 Nomenclature of targets and ligands

    Key protein targets and ligands in this article are hyperlinked to corresponding entries in the IUPHAR/BPS Guide to PHARMACOLOGY (http://www.guidetopharmacology.org) and are permanently archived in the Concise Guide to PHARMACOLOGY 2021/22 (Alexander, Christopoulos et al., 2021; Alexander, Fabbro et al., 2021a,b; Alexander, Kelly et al., 2021)

    3 RESULTS

    3.1 Inflammation and fibrosis in the kidneys of mice with unilateral ureteral obstruction

    We first focused our experiments on the well-established mouse UUO model of kidney inflammation and fibrosis (Chevalier et al., 2009). In comparison to sham-operated mice, 14 days after UUO, mice exhibited a slightly lower body weight (Figure S1a), an increase in weight of the obstructed kidney (Figure S1b,c), and an increase in the proportion of interstitium that stained positively with picrosirius red (Figure S1d), indicative of interstitial fibrosis. By RNA sequencing of bulk kidney tissue (40 million paired-end reads), differential gene expression analysis identified 5037 genes that were increased in their expression and 1244 genes that were decreased in their expression (fold change ≥1.5, P value ≤0.05) (Figure S1e). Among the top 10 genes most increased in their mRNA levels were Havcr1 the gene encoding kidney injury molecule-1 (KIM-1) and Lcn2 the gene encoding neutrophil gelatinase-associated lipocalin (NGAL), as well as the inflammatory genes C3 and Lyz2 and the fibrosis-associated genes Timp1, Col1a1, Mmp7 and Col3a1 (Figure S1f). Down-regulated genes included the organic ion transporters Slc22a7, Slco1a1 and Slc7a13 and the growth factor Egf (Figure S1g). As expected, mRNA levels of the genes that were later used to identify activated kidney fibroblasts were increased in UUO kidneys (Postn fold change 5.38241426, P value q.39045E-07; Acta2 fold change 8.534418206, P value 1.3141E−7; Pdgfra fold change 4.39694863, P value 0.000113857; Tcf21 fold change 5.914765582, P value 5.10223E−5). The top 40 genes encoding GPCRs in bulk kidney tissue 14 days after UUO are shown in Figure S1h, the top ranked GPCR being Adgre1 the gene encoding the mouse monocyte–macrophage marker F4/80 (fold change 11.87353138, P value 2.93024E−6). Tables S2 and S3 show GO (Table S2) and KEGG pathway (Table S3) analyses for the comparison of sham and UUO kidneys 14 days after surgery.

    3.2 Curation of a cell marker, fibrosis gene and GPCR library in fibroblast-enriched kidney cells

    Next, we set out to curate a library of cell markers, fibrosis genes and GPCRs that we could subsequently probe for in single cells. We first isolated plastic adherent fibroblast-enriched cells from the left (obstructed) and right (unobstructed) kidneys of mice 3, 7 and 14 days after UUO. Quantitative RT-PCR revealed that the transcriptional markers of activated fibroblasts Postn (Kaur et al., 2016), Pdgfra (Higashi et al., 2019) and Ccn2 (Heusinger-Ribeiro et al., 2001) were all increased in these cells 3 days after UUO, with increases in Postn and Pdgfra persisting to day 14, albeit at lower levels of expression (Figure S2a–c). Because GPCR expression can be both heterogeneous and at low abundance (Kaur et al., 2017), we decided to perform targeted qRT-PCR for fibroblast GPCRs in single cells isolated from UUO kidneys. To curate a GPCR panel for targeted qRT-PCR, we performed sequencing of RNA isolated from plastic adherent cells 3 or 7 days after UUO or from control mouse kidneys (80 million paired-end reads). PCA plots showed a distinguishable gene expression profile between the three experimental conditions, with cells isolated from kidneys 3 or 7 days after UUO clustering apart from control fibroblasts (Figure S2d). We designed intron-spanning assays for 178 genes, including 157 GPCRs and 2 GPCR-associated proteins (Gprasp1, Gprasp2) that were detected in fibroblast-enriched kidney cells with a FPKM ≥ 0.5 by RNA sequencing, together with marker genes, fibrosis-associated genes and housekeepers (Table 1).

    TABLE 1. Genes included in the custom-designed GPCR quantitative reverse transcription polymerase chain reaction panel for single cell analysis.
    Genes Number
    3 and 7 days after UUO Ackr3, Adgra2, Adgra3, Adgrd1, Adgre1, Adgre4, Adgre5, Adgrf1, Adgrf5, Adgrg1, Adgrg3, Adgrg6, Adgrl1, Adgrl2, Adgrl4, Adora1, Adora2a, Adora2b, Adra1d, Adra2a, Adra2b, Adrb1, Adrb2, Agtr1a, Aplnr, Avpr1a, Avpr2, Bdkrb1, Bdkrb2, C3ar1, C5ar1, C5ar2, Calcrl, Casr, Ccr1, Ccr2, Ccr5, Ccr7, Ccrl2, Celsr1, Celsr2, Chrm4, Cmklr1, Cnr2, Cxcr2, Cxcr3, Cxcr4, Cxcr5, Cxcr6, Cysltr1, Ednra, Ednrb, F2r, Fpr1, Fpr2, Fzd1, Fzd2, Fzd4, Fzd5, Fzd6, Fzd7, Fzd8, Fzr1, Gabbr1, Gper1, Gpr107, Gpr108, Gpr132, Gpr135, Gpr137, Gpr137b, Gpr141, Gpr146, Gpr153, Gpr155, Gpr157, Gpr160, Gpr161, Gpr162, Gpr171, Gpr176, Gpr18, Gpr180, Gpr182, Gpr183, Gpr19, Gpr27, Gpr3, Gpr31b, Gpr34, Gpr35, Gpr39, Gpr4, Gpr65, Gpr68, Gpr84, Gpr85, Gpr89, Gprasp1, Gprc5a, Gprc5b, Gprc5c, Hcar2, Hrh2, Htr1b, Htr3a, Lgr4, Lpar1, Lpar2, Lpar3, Lpar5, Lpar6, Ltb4r1, Mrgprf, Npy6r, Opn3, Oxgr1, P2ry1, P2ry10, P2ry12, P2ry13, P2ry14, P2ry2, P2ry6, Ptger1, Ptger2, Ptger3, Ptger4, Ptgfr, Ptgir, Pth1r, S1pr1, S1pr2, S1pr3, S1pr4, Smo, Sucnr1, Tbxa2r, Tpra1 139
    3 days after UUO only Adra1b, Ffar2, Vipr1 3
    7 days after UUO only Ackr1, Adcyap1r, Adgrg2, Adgrg5, Agtr2, Fzd3, Gpr173, Gpr82, Gpr88, Gprasp2, Htr2a, Htr2b, Htr7, Lpar4, Mrgpre 15
    Control only Adra1a, Mas1 2
    Cell identity Acta2 (smooth muscle), Cdh5 (endothelial), Epcam (epithelial), Pdgfra (fibroblast), Postn (fibroblast), Ptprc (leukocyte), Tcf21 (fibroblast), Nphs2 (podocyte) 8
    Fibrosis-associated

    Timp1, Col1a1, Col1a2, Ccn2, Fn1, Tgfb1

    6
    Housekeeping Rpl13a, Gusb, Hprt, Hsp90ab1, Gapdh 5
    • Abbreviation: UUO = unilateral ureteral obstruction.

    3.3 Single cell qRT-PCR reveals that Tcf21+ fibroblasts isolated from UUO kidneys are enriched for fibrotic genes

    Next, single fibroblast-enriched cells were isolated from mouse kidneys 3 days after UUO using the Fluidigm C1 Auto Prep System. Several quality control steps were applied to define bona fide single fibroblasts. First, individual chambers in the Fluidigm C1 Auto Prep System were inspected by light microscopy to exclude empty wells or chambers containing more than one cell. Second, qRT-PCR was performed for the housekeeping gene Hprt, with only those cells with a CT for Hprt < 25 included in high throughput BioMark quantitative PCR of single cell cDNA. In total, 357 cells were included in the high throughput GPCR screen. Of these, 37 cells that did not express at least two housekeeping genes were excluded (Figure S3). We next applied stringent gene expression marker criteria to define specific cell populations using the marker genes Acta2, Cdh5, Epcam, Ptprc, Tcf21, Postn, Pdgfra and Nphs2 and we excluded a total of 180 cells with mixed marker expression patterns. Taking this approach, cells expressing only Acta2 were defined as smooth muscle cells (n = 4), cells expressing only Cdh5 were defined as endothelial cells (n = 9), cells expressing only Epcam were defined as epithelial cells (n = 18), cells expressing only Ptprc were defined as leukocytes (n = 63) and cells expressing only Nphs2 were defined as podocytes (n = 0). We compared 2 populations of possible fibroblasts based on the presence or absence of expression of Tcf21. Tcf21+ fibroblasts (n = 13) expressed high levels of 2 or 3 of Postn, Acta2 and Pdgfra and were negative for all other cell markers (Figure 1a–c). Tcf21− possible fibroblasts (n = 8) expressed at least 1 of Postn and Pdgfra and were negative for Tcf21, Ptprc, Epcam and Cdh5 (Figure 1a–c). All six fibrosis markers (Col1a2, Timp1, Col1a1, Fn1, Ccn2 and Tgfb1) were differentially up-regulated in Tcf21+ cells in comparison to all other cell-types, whereas Col1a1 and Timp1 were up-regulated in Tcf21− cells (Table 2). Furthermore, fibrotic gene expression was markedly increased in Tcf21+ fibroblasts in comparison to Tcf21− cells (Table 2). Thus, kidney cells expressing Tcf21 and at least two of Acta2, Postn and Pdgfra are bona fide activated kidney fibroblasts.

    Details are in the caption following the image
    Classification of single Tcf21+ activated kidney fibroblasts and Tcf21− possible activated kidney fibroblasts. (a) Heat map for 140 single cells isolated from the kidneys of mice 3 days after unilateral ureteral obstruction (UUO) surgery. Cells labelled Tcf21+ (n = 13) express Tcf21 and 2 or 3 of Postn, Acta2 and Pdgfra and are designated Tcf21+ activated kidney fibroblasts. Cells labelled Tcf21− (n = 8) are negative for Tcf21− and express at least one of Postn or Pdgfra and are designated as Tcf21− possible activated kidney fibroblasts. (b) Violin plots showing expression patterns of marker genes in the Tcf21+ and Tcf21- cells shown in the heatmap (a). (c) Clustering and visualization of cells using t-distributed stochastic neighbour embedding (t-SNE). (d) Dual RNAscope in situ hybridization for Tcf21, Adgra2 and S1pr3 (red puncta) with immunohistochemistry for α-smooth muscle actin (α-SMA; brown) in the kidneys of sham-operated mice or mice 14 days after UUO. Original magnification X 400; images are representative of n = 4 per group. Scale bars = 50 μm. The insets are enlargements of the dashed areas.
    TABLE 2. Expression of fibrosis markers in Tcf21+ activated kidney fibroblasts and Tcf21− possible activated kidney fibroblasts.
    Tcf21+
    Gene Log2 fold change Adjusted P value
    Col1a2 13.20896096 1.49E−31
    Timp1 11.08488519 5.32E−26
    Col1a1 10.76418485 1.13E−20
    Fn1 9.095080367 1.72E−11
    Ccn2 6.667181401 2.22E−8
    Tgfb1 3.513880597 0.001980266
    Tcf21-
    Gene Log2 fold change Adjusted P value
    Col1a1 6.186223022 0.004325526
    Timp1 4.791708633 0.046296749

    3.4 GPCR upregulation in Tcf21+ activated kidney fibroblasts

    By single cell qRT-PCR, we identified 56 differentially up-regulated GPCRs, 54 of which were up-regulated in only Tcf21+ cells, two of which were up-regulated in both Tcf21+ and Tcf21− cells (Gpr153 and Adgrd1) and none of which were up-regulated only in Tcf21− cells (Table 3). The two most up-regulated GPCRs in Tcf21+ activated fibroblasts were Adgra2 (also called Gpr124) and S1pr3 (Table 3). We performed RNAscope in situ hybridization for Tcf21, Adgra2 and S1pr3 (co-stained by immunohistochemistry for α-SMA) in the kidneys of mice 14 days after sham or UUO surgery and observed Tcf21+, Adgra2+ and S1pr3+ RNAscope puncta in (but not necessarily limited to) α-SMA+ interstitial cells in UUO kidneys (Figure 1d).

    TABLE 3. Differentially expressed GPCRs in Tcf21+ activated kidney fibroblasts and Tcf21− possible activated kidney fibroblasts.
    Tcf21+
    Gene Log2 fold change Adjusted P value Gene Log2 fold change Adjusted P value Gene Log2 fold change Adjusted P value
    Adgra2 9.516859931 1.50E−25 Fzd2 4.799925373 3.85E−12 Ptgfr 2.506584386 0.001261481
    S1pr3 9.364695752 1.92E−28 Gpr107 4.740223881 4.85E−6 Tbxa2r 2.35543054 5.69E−9
    Gpr153 9.052095293 1.56E−28 Lgr4 4.481624569 2.65E−5 Gpr135 2.267979334 0.001544928
    Adgrl1 7.762858783 2.91E−17 Gpr89 4.440683123 6.37E−6 Opn3 2.075315729 0.009911016
    Adgrd1 7.352990815 5.72E−22 Gpr108 4.434971297 4.85E−6 Adra1b 2.014758898 1.95E−10
    Lpar1 7.191010333 4.39E−28 Fzr1 4.340700344 2.29E−5 Gpr27 1.93847302 0.012266618
    Fzr 7.075838117 2.23E−11 Fzd3 4.319276693 1.28E−12 Gpr39 1.717990815 0.023333272
    Mrgprf 6.490183697 1.13E−32 Adora2a 4.291974742 4.24E−12 Bdkrb2 1.695373134 0.009911016
    Ednra 6.284500574 1.03E−27 Adgrg2 4.179115959 4.74E−17 Gprasp2 1.551389208 0.045071981
    Avpr1a 6.269070034 1.62E−21 Adgre5 4.006383467 0.000687185 Adgra3 1.541130884 0.030047162
    Fzd1 6.26771527 3.26E−13 Gpr137 3.923461538 9.00E−6 Ptger2 1.477342135 0.000387464
    Smo 6.0536969 1.44E−8 Gabbr1 3.792766935 0.000108757 Mrgpre 1.404517796 0.001636989
    Gpr180 6.023564868 5.38E−10 Pth1r 3.752996556 0.000567125 Ackr3 1.304977038 0.033667111
    Gprc5b 5.831159587 1.58E−12 Htr2a 3.54424225 1.01E−10 Ackr1 1.29847876 3.59E−5
    Ptgir 5.638725603 4.88E−35 Calcrl 3.377135476 0.00392894 Gpr88 1.288461538 9.58E−8
    Gpr161 5.465551091 3.28E−18 Adgrl2 3.203903559 3.96E−5 Adra1d 1.174207807 0.009911016
    Adora2b 5.133639495 1.93E−9 Lpar4 3.142313433 5.89E−10 Tcf21-
    Agtr1a 4.934712974 7.90E−13 Ptger4 2.691991963 0.03298021 Gene Log2 fold change Adjusted P value
    Gprasp1 4.885315729 1.69E−6 Fzd7 2.564081515 0.000233568 Gpr153 4.542329137 0.004828828
    Gpr176 4.864156142 2.45E−25 Ptger1 2.546331803 1.21E−7 Adgrd1 4.383597122 0.001974658
    • Note: Tcf21+ cells express Tcf21 and 2 or 3 of Postn, Acta2 and Pdgfra and were designated activated kidney fibroblasts. Tcf21- cells do not express Tcf21, but express at least 1 of Postn or Pdgfra.

    3.5 Tcf21+ activated fibroblasts are increased in the kidneys of diabetic mice and mice injected with folic acid or adriamycin, and blockade of the renin-angiotensin system (RAS) diminishes kidney Tcf21 expression in diabetic mouse kidneys

    Having defined Tcf21 as a marker useful in the identification of activated kidney fibroblasts and the GPCRs that these cells express in UUO kidneys, we set out to determine whether Tcf21 expression is altered in other mouse models of kidney disease. To do this, we studied aged, high fat diet-fed C57BL/6N mice with streptozotocin-induced diabetes (DM-HFD) (Batchu et al., 2021), BALB/c mice injected with adriamycin (Majumder et al., 2018) and C57BL6/N mice injected with folic acid. In each case, dual RNAscope in situ hybridization and immunohistochemical staining confirmed increased abundance of Tcf21+ α-SMA+ interstitial cells in mice with kidney disease in comparison to their respective controls (Figure 2a). To determine whether Tcf21 expression levels are affected by therapies known to improve outcomes in kidney disease we performed qRT-PCR of bulk kidney tissue from DM-HFD mice and we compared this to age-matched controls and DM-HFD mice treated with either the ACE inhibitor ramipril or the AT1 receptor antagonist telmisartan (Batchu et al., 2021). Tcf21 mRNA levels were increased by ~50% in the kidneys of vehicle-treated DM-HFD mice compared to controls, whereas Tcf21 levels were significantly lower in the kidneys of DM-HFD mice treated with either ramipril or telmisartan (Figure 2b). Co-staining of kidney sections with RNAscope in situ hybridization for Tcf21 and immunohistochemistry for the podocyte marker nephrin also demonstrated Tcf21 transcripts in glomerular podocytes, under both normal and diseased conditions; with an observable increase in Tcf21 RNAscope puncta in the interstitium in UUO kidneys and a decrease in nephrin immunostaining in BALB/c kidneys (Figure S4). To compare the relative contributions of podocyte and interstitial cells to whole kidney Tcf21 content, and to explore how these levels may change with disease and with therapy, we counted Tcf21+ cells in the glomeruli and in the interstitium of kidney sections from control and DM-HFD mice (Figure 2c). In these mice, the number of Tcf21+ glomerular cells was unaffected either by comorbid diabetes or with RAS blockade (Figure 2c). In contrast, Tcf21+ cells were increased in number in the tubulointerstitium of DM-HFD mice, with interstitial Tcf21+ cell number being normalized with telmisartan but not with ramipril (Figure 2c).

    Details are in the caption following the image
    Kidney Tcf21 expression is increased in experimental kidney disease and attenuated by block of the renin-angiotensin system, in experimental diabetic kidney disease. (a) RNAscope in situ hybridization for Tcf21 (red puncta) and immunohistochemistry for α-smooth muscle actin (α-SMA; brown) in kidney tissue from control or aged, diabetic high fat diet-fed (DM-HFD) mice; BALB/c mice injected via tail vein with PBS or adriamycin (10 mg·kg−1 in PBS) and followed for 10 days; or from C57BL/6N mice injected intraperitoneally with 300 mM sodium bicarbonate (control) or 250-mg·kg−1 folic acid in 300-mM sodium bicarbonate and followed for 14 days. Original magnification X 400; images are representative of n = 5 per group. Scale bars = 50 μm. The insets are enlargements of the dashed areas. (b) Quantitative reverse transcription polymerase chain reaction (qRT-PCR) for Tcf21 in the kidneys of control and DM-HFD mice and DM-HFD mice treated with ramipril (10 mg·kg−1·day−1) or telmisartan (3 mg·kg−1·day−1) for 2 weeks (Batchu et al., 2021). n = 7 per group except DM-HFD+ ramipril (n = 6). (c) Quantitation of Tcf21 positive cells in the glomeruli and interstitium of control and DM-HFD mice treated with ramipril or telmisartan. n = 6 per group except DM-HFD+ telmisartan (n = 5). Values shown are means ± SD, from the number (n) of animals shown. *P < 0.05, significantly different as indicated; one-way ANOVA followed by Tukey's post hoc test.

    3.6 The adenosine receptors Adora2a and Adora2b are up-regulated in Tcf21+ activated kidney fibroblasts and adenosine receptor antagonism reduces Tcf21 expression and kidney fibrosis in UUO mice

    Next, we set out to perform proof-of-principle experiments that GPCRs found to be up-regulated in Tcf21+ activated kidney fibroblasts could aid in the identification of interventions that affect kidney fibrosis. We observed that the expression of the genes for the adenosine receptors, Adora2a and Adora2b were each up-regulated in Tcf21+ activated fibroblasts (Table 3) and we confirmed their expression in α-SMA+ interstitial cells in UUO kidneys by RNAscope in situ hybridization (Figure 3a). We noted that caffeine is an adenosine receptor antagonist with activity against both A2A and A2B receptors (Alnouri et al., 2015; Fredholm et al., 2001) and that recent epidemiological data have pointed to an inverse association between caffeine consumption and CKD outcomes (Bigotte Vieira et al., 2019; Kanbay et al., 2021; Srithongkul & Ungprasert, 2020). Sham-operated and UUO mice were then treated with 1 of 2 doses of caffeine (0.4 and 1 mg·ml−1 in drinking water) for 2 weeks (Figure 3b and Table 4). As expected, given that UUO mice have an intact functioning contralateral kidney, plasma cystatin C was largely unchanged across treatment groups (Table 4). In contrast, expression of the fibrogenic genes Postn, Pdgfra, Cola1a and Ccn2 was increased in UUO mice and reduced with caffeine (Table 4). Tcf21 mRNA levels were increased ~20-fold in UUO mouse kidneys, whereas caffeine treatment caused a dose-dependent reduction in Tcf21 expression (Figure 3c), and this was associated with a reduction in interstitial fibrosis, as determined by picrosirius red staining (Figure 3d).

    Details are in the caption following the image
    The adenosine receptor antagonist caffeine attenuates Tcf21 expression and interstitial fibrosis in mice after unilateral ureteral obstruction (UUO) and A2A receptor agonist up-regulates Tcf21 and Ccn2 in NRK-49F fibroblasts. (a) RNAscope in situ hybridization for the adenosine receptors Adora2a and Adora2b (red puncta) and immunohistochemistry for α-smooth muscle actin (α-SMA; brown) in kidney tissue from mice 14 days after sham or UUO surgery. Original magnification X 400; images are representative of n = 6 per group. Scale bars = 50 μm. The insets are enlargements of the dashed areas. (b) Experimental design for the treatment of sham or UUO mice with caffeine in drinking water for 14 days. (c) Quantitative reverse transcription polymerase chain reaction (qRT-PCR) for Tcf21 in the kidneys of sham and UUO mice treated with vehicle (drinking water) or caffeine dosed at either 0.4 or 1 mg·ml−1 in drinking water for 14 days. Sham+ vehicle (n = 13), Sham+ caffeine (0.4 mg·ml−1) (n = 10), Sham+ caffeine (1 mg·ml−1) (n = 12), UUO+ vehicle (n = 13), UUO+ caffeine (0.4 mg·ml−1) (n = 11), UUO+ caffeine (1 mg·ml−1) (n = 12). (d) Representative photomicrographs of picrosirius red stained kidney sections and quantification of picrosirius red staining for sham and UUO mice treated with vehicle (drinking water) or caffeine dosed at either 0.4 or 1 mg·ml−1 in drinking water for 14 days. Sham+ vehicle (n = 13), sham+ caffeine (0.4 mg·ml−1) (n = 11), sham+ caffeine (1 mg·ml−1) (n = 12), UUO+ vehicle (n = 12), UUO+ caffeine (0.4 mg·ml−1) (n = 10), UUO+ caffeine (1 mg·ml−1) (n = 12). Original magnification x 400. Scale bars = 50 μm. (E and F) qRT-PCR for Tcf21 (E) and Ccn2 (F) in NRK-49F fibroblasts incubated for 48 h with the A2A receptor agonist CGS 21680 (1 μmol·L−1) or the A2B receptor agonist BAY60–6583 (5 μmol·L−1) in the presence or absence of 4 mmol·L caffeine (n = 6 per condition). Values are mean ± S.D.. *P <0.05, significantly different as indicated; in (c), data from sham and UUO groups were analysed, following log-transformation, with one-way ANOVA and Tukey's post test; in (c and d), data from caffeine and vehicle groups or, in (e and f), data from adenosine receptor agonists and vehicle were analysed with one-way ANOVA and Fisher least significant differences post hoc test.
    TABLE 4. Functional parameters and expression of fibrogenic genes in the kidneys of sham-operated mice and mice with unilateral ureteral obstruction (UUO) and treated with vehicle (drinking water) or caffeine in drinking water at doses of either 0.4 or 1 mg·ml−1 for 14 days.
    Sham+ vehicle Sham+ caffeine (0.4 mg·ml−1) Sham+ caffeine (1 mg·ml−1) UUO+ vehicle UUO+ caffeine (0.4 mg·ml−1) UUO + caffeine (1 mg·ml−1)
    n 13 11 12 13 11 12
    Body weight (g) 20.6 ± 1.2 20.7 ± 1.1 21.1 ± 1.5 20.9 ± 1.6 20.4 ± 1.1 20.6 ± 0.7
    Left kidney weight (g) 0.148 ± 0.014 0.132 ± 0.013 0.141 ± 0.007 0.221 ± 0.041* 0.219 ± 0.039 0.191 ± 0.040
    Left kidney weight: Body weight (%) 0.72 ± 0.07 0.64 ± 0.05 0.67 ± 0.04 1.07 ± 0.24* 1.08 ± 0.19 0.93 ± 0.21
    Systolic blood pressure (mmHg) 102 ± 10 96 ± 7 105 ± 10 119±10* 115 ± 11 110 ± 7
    Plasma cystatin C (ng·ml−1)

    252 ± 34

    (n = 11)

    242 ± 25

    (n = 11)

    222 ± 25

    (n = 12)

    272 ± 59

    (n = 12)

    240 ± 43

    (n = 11)

    265 ± 33

    (n = 12)

    Postn:Gapdh mRNA (fold) 1.5 ± 1.3

    1.0 ± 1.5

    (n = 9)

    0.2 ± 0.2 73.7 ± 66.3* 11.3 ± 7.7# 6.9 ± 4.4#
    Pdgfra:Gapdh mRNA (fold) 1.1 ± 0.6 1.1 ± 1.3 0.5 ± 0.2 10.3 ± 9.5* 4.9 ± 1.3# 4.9 ± 1.1#
    Col1a1:Gapdh mRNA (fold) 1.1 ± 0.6 1.0 ± 0.2 0.7 ± 0.2 111.0 ± 80.6* 80.3 ± 24.5 57.4 ± 20.8#
    Ccn2:Gapdh mRNA (fold) 1.2 ± 0.8 0.8 ± 0.2 0.9 ± 0.3 23.5 ± 20.8* 9.6 ± 5.5# 13.5 ± 3.7
    • Note: Values shown are the means ± SD from the number (n) of animals shown.
    • * P < 0.05, significantly different from Sham+ vehicle;
    • # P < 0.05, significantly different from UUO+ vehicle;
    • Welch's ANOVA followed by Dunnett's T3 post hoc test.

    To explore whether either A2A or A2B receptors regulate either fibroblast Tcf21 expression or fibrogenic gene expression, we next incubated NRK-49F kidney fibroblast cells with either the A2A receptor agonist CGS 21680 (Jarvis et al., 1989) or the A2B receptor agonist BAY60–6583 (Eckle et al., 2007) in the presence of absence of caffeine. In these experiments, the A2A receptor agonist increased mRNA levels of both Tcf21 and Ccn2 in NRK-49F fibroblasts and this effect was negated by caffeine (Figure 3e,f). The A2B receptor agonist had a more variable effect, increasing Tcf21 mRNA but not Ccn2 mRNA, with Ccn2 mRNA levels being reduced by the combination of caffeine and BAY60-6583 (Figure 3e,f).

    3.7 TCF21, ADGRA2, S1PR3, ADORA2A and ADORA2B are expressed by α-SMA+ interstitial cells in human kidneys

    In our final experiments, we sought to explore the relevance to human disease of the candidates identified through our single cell GPCR screen of fibroblasts from UUO mouse kidneys. By RNAscope in situ hybridization, we observed transcripts for TCF21, ADGRA2, S1PR3, ADORA2A or ADORA2B in interstitial cells of human kidneys (Figure 4). We also combined RNAscope in situ hybridization with immunostaining for α-SMA in kidney tissue from controls or from patients with diabetic kidney disease (DKD, the most common cause of end-stage kidney disease). In each case, TCF21, ADGRA2, S1PR3, ADORA2A or ADORA2B mRNA was detectable in (although not necessarily limited to) α-SMA+ interstitial cells (Figure 4).

    Details are in the caption following the image
    TCF21, ADGRA2, S1PR3, ADORA2A and ADORA2B are expressed by α-smooth muscle actin (α-SMA) expressing interstitial cells in human control kidneys and in kidney tissue from patients with diabetic kidney disease (DKD). Single RNAscope in situ hybridization (left column) and dual RNAscope in situ hybridization and immunohistochemistry for α-SMA (brown) (middle and right columns) in archival kidney tissue from normal human kidney tissue (controls) or patients DKD. Original magnification x 400; images are representative of n ≥ 4 per group. Scale bars = 50 μm. The insets are enlargements of the dashed areas. The arrows point to RNAscope puncta in interstitial cells (brown puncta) or in α-SMA+ interstitial cells (red puncta).

    4 DISCUSSION

    Activation of interstitial fibroblasts and the excessive production of extracellular matrix by these cells are, in large part, responsible for the progressive decline of kidney function in CKD. With the goal of advancing the development of new interventions that alter pathological fibrogenesis in the kidneys, we performed a single cell GPCR screen of activated kidney fibroblasts. In doing so, we established that activated kidney fibroblasts express the transcription factor, Tcf21 and we identified 56 GPCRs that are differentially up-regulated in Tcf21+ activated kidney fibroblasts, including both recognized and novel targets. The results of the single cell GPCR screen reported here may be useful to others in identifying new treatment opportunities to improve outcomes in CKD.

    Tcf21 is a basic helix–loop–helix (bHLH) transcription factor that, in the kidneys, is expressed by developing and mature podocytes and, during metanephric development, also by Six2+ nephron progenitors and the Foxd1+ stromal mesenchyme (Maezawa et al., 2014; Quaggin et al., 1999). bHLH transcription factors regulate cell fate determination and differentiation. Accordingly, knockout of Tcf21 impairs podocyte differentiation and causes an FSGS-like phenotype (Maezawa et al., 2014); and Tcf21 null mice fail to develop differentiated peritubular interstitial cells (Cui et al., 2003). Given the persistence of Tcf21+ cells in the interstitium into adulthood (Quaggin et al., 1999) and the limited contribution that circulating cells make to the activated fibroblast population in diseased kidneys (Kramann et al., 2018; Kuppe et al., 2021), we expect that Tcf21+ activated fibroblasts arise from resident kidney cells.

    Whereas Tcf21 has been best studied in the kidneys for its role in podocyte biology (Maezawa et al., 2014; Quaggin et al., 1999), to date Tcf21+ fibroblasts have been most extensively researched in a cardiac context. In the developing heart, Tcf21+ epicardial progenitors can form either coronary vascular smooth muscle cells or cardiac fibroblasts (Acharya et al., 2012). However, with time, Tcf21 expression becomes limited to cardiac fibroblasts where it is necessary for cell fate determination (Acharya et al., 2012). Following MI, Tcf21+ cardiac fibroblasts become periostin-expressing myofibroblasts responsible for collagen production and scar formation (Kanisicak et al., 2016). However, even in cardiac tissues, the role of TCF21 is not straightforward. For instance, TCF21 polymorphisms have been linked to coronary artery disease (CAD) risk (Schunkert et al., 2011), knockdown of TCF21 results in upregulation of differentiation markers in human coronary artery smooth muscle cells (Nurnberg et al., 2015), and higher tissue TCF21 expression has been associated with lower CAD risk (Wirka et al., 2019). In the kidney, Tcf21 was the most up-regulated transcription factor identified through microarray analysis of fibroblasts isolated from mouse kidneys after folic acid or UUO (Higashi et al., 2019); and, in a mouse kidney progenitor cell line, overexpression of Tcf21 increased smooth muscle and myofibroblast gene expression, whereas knockdown of Tcf21 had the opposite effect (Plotkin & Mudunuri, 2008). On the contrary, TCF21 has not been identified as a marker of activated fibroblasts in recent single cell RNA sequencing studies of human kidney disease (Kuppe et al., 2021; Muto et al., 2022). We observed that TCF21 is expressed by α-SMA+ interstitial cells in adult mouse and human kidneys. Furthermore, as suggested by the response to RAS blockade in diabetic mice or A2A receptor activation in cultured cells, increases in its mRNA levels likely reflect both increases in the number of Tcf21+ fibroblasts and upregulation in its expression in individual cells. Whether Tcf21 plays a role in kidney fibroblast activation has not, however, been defined. In summary, in the present study we have defined Tcf21 as a marker of activated kidney fibroblasts, but not necessarily a marker of kidney fibroblast activation.

    Among the GPCRs in Tcf21+ activated kidney fibroblasts, the two most up-regulated were Adgra2 (also called GPR124) and S1pr3. ADGRA2 is an orphan GPCR that has twice been reported in abstract form as a potential mediator of kidney fibrosis (Blank et al., 2022; Ikeda et al., 2017); and a patent (WO2017040916A3) has been filed for modulators of GPR124 for reduction of pathological fibrosis. The gene S1pr3 codes for S1P3 one of five receptors (named S1P1 - S1P5) for the sphingolipid metabolite, sphingosine-1-phosphate (S1P). S1P has been studied for its role in kidney (patho)biology (see Drexler et al., 2021; Jo et al., 2008), with extracellular S1P having been reported to have profibrotic actions in the kidneys (Drexler et al., 2021; X. Zhang et al., 2018). Thus, the utility of the single cell strategy we employed to identify GPCRs putatively involved in the fibrogenic function of activated kidney fibroblasts is supported by the detection of candidates that have been validated for their role in kidney fibrosis (e.g. S1pr3), are being actively targeted by investigational antifibrotic therapies (e.g. Adgra2), or alternatively that are already the targets of approved therapies (e.g. Atgr1a). Our screen, however, also identified a number of GPCRs not previously linked, or less well-established, as mediators of fibrosis and these GPCRs warrant further scrutiny (e.g. Gpr153, Mrgprf, Gpr176, Adgrd1, Fzd3, Gpr88, Gpr107, Gpr108, Gpr89, Gpr137, Gpr27).

    As proof-of-concept that antagonism of GPCRs identified in our single cell screen can improve kidney outcomes, we focused further on the adenosine A2A and A2B receptors. Like the S1P receptors, adenosine receptors have been studied for their role in kidney biology previously (see Roberts et al., 2014), in which short term activation of A2A and A2B receptors may decrease inflammation, but persistent activation of A2B receptors promotes kidney fibrosis (Roberts et al., 2014). Adenosine is present at low concentrations in normal kidneys and its levels increase during kidney injury, primarily because of hydrolysis of ATP released from injured or dying cells (Roberts et al., 2014). In vitro studies have identified expression of A2B receptors (but not of A2A receptors) in kidney fibroblasts and they have described a fibrogenic role for activation of A2B receptors in these cells (Tang et al., 2015; Wilkinson et al., 2016). In the present study, the non-selective adenosine receptor antagonist, caffeine reduced Tcf21 mRNA levels, fibrogenic gene expression and interstitial fibrosis in UUO mice. Furthermore, in cultured cells caffeine prevented A2A receptor-dependent Tcf21 and Ccn2 upregulation. These findings complement an existing body of literature describing the roles of A2A receptors in fibrogenesis (Herman-de-Sousa et al., 2022; Kim et al., 2022; Perez-Aso et al., 2014; Shaikh et al., 2016; J. Zhang et al., 2017) and the effects of caffeine on tissue fibrosis. For instance, caffeine has been reported to inhibit the activation of BHK-21 kidney fibroblasts induced by hypoxia (Nilnumkhum et al., 2019) and it has also been reported to attenuate fibrosis of the liver (Modi et al., 2010; Shan et al., 2022) and lungs (Tatler et al., 2016). Conversely, however, caffeine worsened fibrosis in a mouse model of cystic kidney disease (Meca et al., 2019), activation of A2A receptors attenuated fibrosis in experimental glomerulonephritis (Garcia et al., 2011), and a dual-acting A2A receptor agonist and A3A receptor antagonist (LJ-4459) reduced tubulointerstitial fibrosis in UUO mice (Pak et al., 2021). It is also noteworthy that adenosine and caffeine exert other effects that may alter kidney function, independent of those on interstitial fibroblasts (e.g., regulation of afferent arteriolar tone and sodium reabsorption; Vallon et al., 2006). Thus, additional research is required to define the specific actions of individual adenosine receptors in activated kidney fibroblasts.

    In addition to the above limitations, the relatively small number of cells studied in our single cell qRT-PCR screen merits consideration. Because of the stringent quality control criteria we employed, our final analysis included 140 single cells, among them 13 Tcf21+ activated kidney fibroblasts. Recent single cell RNA sequencing experiments in experimental kidney fibrosis have studied up to hundreds of thousands of cells (Balzer et al., 2022; Li et al., 2022). Single cell RNA sequencing, however, favours more abundant transcripts and is capable of detecting only approximately 10% of cellular mRNA (Wilson & Humphreys, 2020). In contrast, qRT-PCR can detect more GPCRs (Kaur et al., 2017) and, accordingly, we employed this approach to identify which GPCRs are differentially expressed by activated kidney fibroblasts. Had a larger number of cells been studied, or had less stringent criteria been employed, it is possible that we would have identified additional GPCRs. Nevertheless, the top candidate GPCRs have established roles, or are being actively investigated, as fibrogenic GPCRs, attesting to the specificity of the approach that we have taken.

    In summary, the present study established Tcf21 as a transcriptional marker that is helpful in identifying activated fibroblasts in kidney disease. The single cell qRT-PCR screen that was performed in these cells may serve as a resource for investigators interested in studying therapeutically modulated GPCRs that are up-regulated in activated kidney fibroblasts and that may be involved in pathological kidney fibrogenesis.

    DECLARATION OF TRANSPARENCY AND SCIENTIFIC RIGOUR

    This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Design & Analysis, Immunoblotting and Immunochemistry, and Animal Experimentation, and as recommended by funding agencies, publishers and other organisations engaged with supporting research.

    AUTHOR CONTRIBUTIONS

    Harmandeep Kaur and Andrew Advani conceived the study; Harmandeep Kaur, Veera Ganesh Yerra, Sri Nagarjun Batchu, Duc Tin Tran, M. D. Golam Kabir, Youan Liu, Suzanne L. Advani and Phelopater Sedrak performed the experiments; Harmandeep Kaur and Andrew Advani analysed the data; Laurette Geldenhuys, Karthik K. Tennankore, Penelope Poyah and Ferhan S. Siddiqi provided access to archived human kidney tissue; Harmandeep Kaur and Andrew Advani drafted and revised the paper.

    ACKNOWLEDGEMENTS

    The authors gratefully acknowledge the services and facilities of the Princess Margaret Genomics Centre (Toronto, Ontario, Canada) and the Canadian Centre for Computational Genomics (C3G, Montreal, Quebec, Canada). We acknowledge the use of tissues procured by the National Disease Research Interchange (NDRI) with support from NIH grant 2 U42 OD011158. The studies were supported by a Banting and Best Diabetes Centre - Sun Life Financial Pilot and Feasibility Grant to A. A. and, in part, by project grants from the Canadian Institutes of Health Research to A.A. (PJT153284 and PJT166083) and a Grant-in-Aid from the Heart and Stroke Foundation of Canada to A. A. (G22-0031981). H. K. was supported by a KRESCENT Post-Doctoral Fellowship from the Kidney Foundation of Canada. V. G. Y. was supported by a Diabetes Canada Post-Doctoral Fellowship and a D. H. Gales Family Charitable Foundation Post-Doctoral Fellowship from the Banting and Best Diabetes Centre. P.S. was supported by a Charles Hollenberg Summer Studentship from the Banting and Best Diabetes Centre. A. A. was a recipient of a Diabetes Investigator Award from Diabetes Canada and holds the Keenan Chair in Medicine from St. Michael's Hospital and University of Toronto. Work in the Advani Lab is supported, in part, by the RDV Foundation. Graphics were created with BioRender.com.

      CONFLICT OF INTEREST STATEMENT

      K. K. T. has received advisory board or consultancy fees from Otsuka, Vifor, GSK, Bayer and AstraZeneca and has received unrestricted grant funding from Otsuka for an investigator initiated study. All other authors declare no conflict of interest.

      DATA AVAILABILITY STATEMENT

      The data that support the findings of this study are available from the corresponding author upon reasonable request. Some data may not be made available because of privacy or ethical restrictions.