Matrix Gla Protein Expression in Pericytes and Myofibroblasts Contributes to Renal Fibrosis

Abstract

Renal fibrosis is the main pathologic change observed with the progression of chronic kidney disease (CKD), which predicts kidney outcomes. The ability to detect fibrosis early in the disease course may be crucial to identify those at the highest risk of CKD progression. Clinical studies have observed increased serum matrix Gla protein (MGP), a potent inhibitor of soft tissue calcification, in patients with CKD. In a cross-sectional study of patients with CKD, serum MGP levels were found to be associated with albuminuria and waist circumference after controlling for kidney function. To examine the impact of MGP on the progression of CKD, various mouse models were used in the current study. Using Cre-reporter, Rosa^Tomato;Mgp-Cre mice and a new knock-in model expressing hemagglutinin epitope-tagged MGP, it was identified that pericytes and myofibroblasts in healthy and folic acid (FA)–injured kidneys produce MGP. FA injection in Mgp^–/– mice induced significantly less renal fibrosis in comparison to the control mice because of a reduced number of pericytes and attenuated Notch signaling. In a complementary experiment, restoration of Mgp expression in myofibroblasts in Mgp^–/– mice leads to renal fibrosis as severe as in control mice. This work suggests that MGP expression in myofibroblasts exacerbates renal fibrosis in FA-injured kidneys.

Graphical abstract

Chronic kidney disease (CKD) is a progressive and irreversible loss of kidney function affecting 10% of the world population. In developed countries, the leading causes of CKD are diabetes, glomerulonephritis, and hypertension.1, 2, 3 CKD is divided into five stages according to the estimated glomerular filtration rate (eGFR) and urine albumin/creatinine ratio (uACR) test results. The last stage of CKD is also known as kidney failure or end-stage kidney disease.¹ As CKD advances, an imbalance of blood phosphate and calcium occurs, which contributes to the development and progression of cardiovascular diseases and bone disorders.⁴ It is estimated that 37 million people in the United States meet the criteria for CKD; however, in its early stages, most people are asymptomatic and unaware that they have the condition. Treatments that target the control of blood pressure and reduction of albuminuria attenuate the progression of CKD. Hence, preventing or detecting the disease before it progresses to the severe stage is critically important.

When kidney cells are injured, a wound-healing process is initiated to repair the tissue.⁵ Recruited immune cells secrete growth factors, including transforming growth factor-β (TGF-β), platelet-derived growth factor, and vascular endothelial growth factor, that act to stimulate the differentiation and proliferation of myofibroblasts.⁶ Tubular interstitial cells, such as resident fibroblasts and pericytes, are well-known progenitor cells of myofibroblasts. These cells are localized between the interstitial area or perivascular region of the kidney, where they support tissue structure and maintain kidney homeostasis under physiological conditions. When the kidney is injured, detachment of pericytes from the peritubular vessel structure occurs, and the profibrotic cytokines released by immune cells stimulate these cells to differentiate into myofibroblasts.^5,7, 8, 9 Myofibroblasts produce extracellular matrix (ECM) proteins for wound closure and tissue regeneration. If the injury is repeated, the wound-healing process enters a maladaptive phase called fibrosis.⁶ Fibrosis is the main pathologic change observed with the progression of CKD, where, in addition to interstitial fibrosis, it ultimately manifests as arteriosclerosis and glomerulosclerosis. Although the degree of fibrosis predicts kidney outcomes, there are currently no established tools to monitor kidney fibrosis in the clinic. The ability to identify fibrosis early in the disease course may be crucial to identify those at the highest risk of progression.

In addition to eGFR and uACR, several other biomarkers have been assessed for their ability to detect kidney disease progression. Kidney injury molecule-1 (KIM-1), neutrophil gelatinase-associated lipocalin, liver-type fatty acid–binding protein (L-FABP), and cystatin C are some of the known biomarkers produced by injured tubular epithelial cells.¹⁰ Previous studies demonstrated that the kidney expression of matrix Gla protein (MGP) increased progressively with CKD and predicted kidney outcomes.11, 12, 13, 14 MGP is a small secreted protein that acts as a potent inhibitor of vascular and cartilaginous soft tissue ectopic calcification.¹⁵ Two post-translational modifications—carboxylation of four glutamic acid residues and phosphorylation of three serine residues—have been linked to MGP's anti-mineralization function.^16,17

A previous study reported the association between MGP mRNA expression in the glomeruli and tubulointerstitial compartment and disease activity/histologic features in kidney biopsies of patients enrolled in the Nephrotic Syndrome Study Network (NEPTUNE) study.¹¹ Tubulo-interstitial expression of MGP correlated strongly with the degree of interstitial fibrosis and tubular atrophy independent of the level of kidney function. Furthermore, a higher level of tubulo-interstitial MGP expression was associated with an increased risk for the composite end point of decline in eGFR and progression to end-stage kidney disease. Finally, there was a significant negative correlation between baseline measure of proteinuria and both glomerular and interstitial expression of MGP. Although this was the first study to suggest a role for MGP in fibrosis, the observational study design cannot resolve whether the increase in MGP expression is a consequence of CKD or a cause of CKD progression. Finally, although their study confirms that MGP expression occurs in the kidney, it is not yet known from which cells it is expressed.¹¹

Here, patients with stage 3 to 5 CKD who do not require dialysis showed high circulating levels of MGP, and complementary data in mice demonstrated Mgp induction in two kidney injury models: ischemia-reperfusion (I/R) surgery and folic acid (FA) injection. Using Mgp promoter-driven reporter gene expression in compound mutant mice [B6.Cg-Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze}/J;Mgp^Cre/+] and a novel knock-in model, Mgp^HA/HA mice expressing hemagglutinin (HA)–tagged MGP, pericytes, and myofibroblasts were revealed as the primary cell types that express Mgp in both healthy and injured kidneys. Interestingly, Mgp ablation significantly reduced kidney fibrosis following FA-induced injury compared with controls, most likely because of the decreased numbers of pericytes and down-regulation of Notch signaling. In contrast, restoration of Mgp expression in Mgp^–/– mice, driven by an Sm22α promoter fragment active in myofibroblasts, resulted in kidney fibrosis of similar severity to that observed in the control mice. Overall, this study identifies MGP-expressing cells in healthy and diseased kidneys and highlights the profibrotic role of MGP in the development of kidney fibrosis.

Materials and Methods

Human Study

Participants with CKD were enrolled in a study designed to examine the cross-sectional association between single-slice epicardial fat area and kidney function as well as traditional and kidney-related cardiovascular disease risk factors.¹⁸ Participants were eligible to participate if they were aged >18 years and had CKD stage 3 to 5 (not requiring dialysis and excluding acute kidney injury). All patients gave informed consent to participate, and the study protocol was approved by the Queen's University (Kingston, ON, Canada) Health Sciences and Affiliated Teaching Hospitals Research Ethics Board. Those participants with a measurement of MGP were included in this sub-study.

Laboratory measures: Blood samples were stored at –80°C and, after a single freeze-thaw cycle, the following were measured from fasting plasma in duplicate at the Ontario Cancer Biomarker Network (Toronto, ON, Canada): total MGP by enzyme immunoassay (BioMedica Diagnostics, Windsor, NS, Canada), 1,25-dihyroxyviatmin D, enzyme immunoassay (Immunodiagnostic Systems Inc., Fountain Hills, AZ), 25-hydroxyvitamin D, enzyme immunoassay (Immunodiagnostic Systems Inc.), fetuin A, enzyme-linked immunosorbent assay (ALPCO Diagnostics, Salem, NH), and fibroblast growth factor-23 enzyme-linked immunosorbent assay (ALPCO Diagnostics). The remaining measurements were measured in fasting serum at Kingston General Hospital's (Kingston, ON, Canada) Core Laboratory and included creatinine [Jaffe rate method; Beckman Coulter (Fullerton, CA) UniCel DxC 800 SYNCHRON Clinical System assay; traceable to isotope dilution mass spectroscopy], glucose, phosphate, total calcium, intact parathyroid hormone (chemiluminescent immunoassay; Beckman Couter UniCel DxI 800 Access Immunoassay System), albumin, high-sensitivity C-reactive protein (Beckman Coulter UniCel DxC 600/800; SYNCHRON Clinical System), total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein, and triglycerides.

Albuminuria was assessed by the uACR (mg/mmol).

The four-variable Modification of Diet in Renal Disease Study equation, re-expressed for standardized creatinine, was used to estimate eGFR. Weight and height were collected on each individual to assess body mass index in kg/m². Abdominal obesity was assessed by waist circumference.

Summary statistics are expressed as means and SDs, medians and interquartile ranges, or numbers/percentages, as appropriate. Spearman R was used to assess the relationship between variables and MGP with partial adjustment for eGFR. MGP was modeled using an ordinary least squares linear regression model, including the following independent variables: waist circumference, uACR, eGFR, and the product of uACR and eGFR. Extreme outliers and overly influential observations based on Cooks-D >0.1 were considered for removal (n = 2), and six participants had missing data for uACR.

Mouse Models

Generation and characterization of B6.Cg-Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze}/J (Rosa26^Tomato/+), Mgp-Cre, Mgp^–/–, and Sm22α-Mgp mice have been reported previously.^15,19, 20, 21 To generate the knock-in mice carrying Mgp with the human influenza HA sequence at the 3ʹ end, guide RNAs targeting the gene sequence adjacent to the stop codons were designed using online software CRISPOR version 4.98 (https://crispor.gi.ucsc.edu). Three different guide RNAs were ordered from Synthego (Redwood City, CA), and the targeting efficiency (ability to introduce insertions/deletions at the Mgp loci) of each of them was determined by in vitro culture of guide RNA–clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 (Cas9)–injected blastocysts and subsequent DNA sequencing. Next, the validated single-guide RNA mixed with Cas9 protein and the single-stranded oligodeoxynucleotide repair template carrying the HA sequence were microinjected into the fertilized eggs from C57BL/6J mice at the Transgenic Core Facility of Goodman Cancer Research Centre at McGill University (Montreal, QC, Canada). The insertion of the HA sequence in the 3ʹ end of Mgp was confirmed by PCR amplification of the targeted sequence using flanking primers, followed by Sanger sequencing. Subsequent genotyping of the knock-in mice was performed using PCR. All the primers used for the genotyping are listed in Table 1. All mouse lines were maintained at the vivarium of Shriners Hospital for Children (Montreal, QC, Canada) following an animal use protocol approved by the animal care committee of McGill University.

Table 1.

List of Primers Used in This Study

Variable	Primer	Sequence	PCR product size, bp
Genotyping	MGP-5ʹ UTR/F	5ʹ-GGAGAACACTACAGCTCTGTC-3ʹ	524
	MGP-5ʹ E1/R	5ʹ-AGGTTTCTCACCGTAGCACAG-3ʹ
	Pthrp-3/F	5ʹ-ATCAGGATGATCTGGACGAAGA-3ʹ	600
	Mgp-E4/R	5ʹ-TGCCTGAAGTAGCGGTTGTA-3ʹ
	Sv40/qF2	5ʹ-CAAGTTAACAACAACAATTGCAT-3ʹ	400
	Sv40/qR2	5ʹ-TTCAGAGCAGAATTGTGGAGTG-3ʹ
	Rosa locus/F	5ʹ- AAAGTCGCTCTGAGTTGTTAT-3ʹ	347
	CMV enhancer/R	5ʹ-TCGTTGGGCGGTCAGCCAG-3ʹ
	MgpCre/F	5ʹ-CTGGAGTTTCAATACCGGAG-3ʹ	690
	MgpCre/R	5ʹ-ATGTGGTTACACCTCCACAC-3ʹ
	HA/F3	5ʹ-ATGGTCTACGGCTACAACGC-3ʹ	297 (WT) 324 (MgpHA)
	HA/R3	5ʹ-AAAGAAACCTGACTTCACAAAGTG-3ʹ
Quantitative real-time-PCR	mMgp/qF	5ʹ-AGGACTCCATGCTTTCGTGA-3ʹ	97
	mMgp/qR	5ʹ-ACCCGAGACACCATGAAGAG-3ʹ
	mCol1a1/qF	5ʹ-ACATGTTCAGCTTTGTGGACC-3ʹ	139
	mCol1a1/qR	5ʹ-GGTTTCCACGTCTCACCATT-3ʹ
	mCol3a1/qF	5ʹ-GCTTTGTGCAAAGTGGAACC-3ʹ	143
	mCol3a1/qR	5ʹ-TGGTTCTGGCTTCCAGACAT-3ʹ
	mEln/qF	5ʹ-TCTTGCTCAACCTCCTCCAT-3ʹ	111
	mEln/qR	5ʹ-CAATACCAGCCCCTGGATAA-3ʹ
	mActa2/qF	5ʹ-GGGATCCTGACGCTGAAGTA-3ʹ	147
	mActa2/qR	5ʹ-GTTCAGTGGTGCCTCTGTCA-3ʹ
	mHey1/qF	5ʹ-TGAAGAGAGCTCACCCAGACT-3ʹ	145
	mHey1/qR	5ʹ-GCCAAAACCTGGGACGATGT-3ʹ
	mTagln/qF	5ʹ-ATCCTGAGGGATCGAAGCCA-3ʹ	167
	mTagln/qR	5ʹ-TGCTGCCATATCCTTACCTTCA-3ʹ
	mHprt/qF	5ʹ-GTTGAGAGATCATCTCCACC-3ʹ	341
	mHprt/qR	5ʹ-AGCGATGATGAACCAGGTTA-3ʹ

CMV, cytomegalovirus; F, forward; HA, hemagglutinin; m, mouse; MGP, matrix Gla protein; Pthrp, parathyroid hormone-related protein; qF, forward primer for quantitative PCR; qR, reverse primer for quantitative PCR; R, reverse; SV40, simian virus 40; UTR, untranslated region; WT, wild type.

Animal Studies

Renal injury was induced by either FA (F8758; Sigma, St. Louis, MO) administration or I/R surgery. The sample size was calculated on the basis of pilot experiments examining Mgp expression after the FA administration or I/R surgery. The power of the experiment was set to 80%. A minimum of three mice per group was considered necessary. The 8-week–old Mgp^HA/HA mice and 2-week–old wild-type (WT), Mgp^–/–, and Mgp^–/–;Sm22α-Mgp mice randomly received vehicle (0.3 mol/L sodium bicarbonate) or 250 mg/kg FA by i.p. injection to induce renal injury. Mice were euthanized 2, 14, and/or 21 days after the administration. Both sexes were used for this study, and three to six mice were analyzed per group. The animal experimental protocols were reviewed and approved by the animal care committee of McGill University.

I/R injury by unilateral renal artery clamping and contralateral nephrectomy were performed on 8-week–old C57BL/6 mice, as previously described.²² Mice were euthanized 2 and 21 days after the surgery, and kidneys were collected for analysis. Only female mice were used for I/R injury surgery because of the high mortality of male mice after the injury, and three mice were analyzed per group. The animal experimental protocols were reviewed and approved by the Centre Hospitalier de l’Université de Montréal (Montreal, QC, Canada) Comité Institutionnel de Protection des Animaux.

Gene Expression Analysis

Total RNA extraction and first-strand cDNA synthesis from the kidneys were performed as previously described.²³ PCRs were performed with an Advanced real-time quantitative PCR master mix (800-437-UL; Wisent Inc., Saint-Jean-Baptiste, QC, Canada). Mouse Hprt was amplified to be used as an endogenous control. All primer sequences are listed in Table 1.

Immunofluorescence

For immunofluorescence staining, mouse kidney sections were deparaffinized, processed for antigen retrieval with 10 mmol/L sodium citrate (pH 6), and blocked with 5% normal goat serum in phosphate-buffered saline (PBS) with 0.3% Triton X-100 for 1 hour at room temperature. The sections were then incubated overnight at 4°C in a humidified chamber with the primary antibodies. Slides were washed three times with PBS and incubated with the secondary antibodies for an hour at room temperature. The nucleus was counterstained using Hoechst (63493; Sigma), and slides were mounted by coverslips using Fluoromount-G mounting medium (00-4958-02; Invitrogen, Waltham, MA). Images were captured by a LSM780 laser-scanning confocal microscope (Zeiss, Oberkochen, Germany) at the Molecular Imaging Platform, Research Institute of the McGill University Health Centre (RI-MUHC; Montreal, QC, Canada) or a LSM900 Airyscan2 (Zeiss) at the Cellular Imaging Core Facility, Research Centre of the Hospital Centre of the University of Montreal (CRCHUM; Montreal, QC, Canada). The following primary antibodies were used: anti-HA tag antibody (1:50; 2367; Cell Signaling Technology, Danvers, MA), anti-CD31 (1:200; 77699; Cell Signaling Technology), anti–platelet-derived growth factor receptor β (1:50; 3169; Cell Signaling Technology), and anti–α smooth muscle actin (SMA) (1:200; 19245; Cell Signaling Technology). The secondary antibodies used were anti-rabbit IgG (H+L), F (ab')₂ fragment (Alexa Fluor 555 conjugate; 1:500; 4413; Cell Signaling Technology) and anti-mouse IgG (H+L), F (ab')₂ fragment (Alexa Fluor 647 conjugate; 1:500; 4410; Cell Signaling Technology).

Immunoprecipitation

Mgp^HA/HA serum was diluted in PBS with 0.02% Tween-20 containing 1 mmol/L phenylmethanesulfonyl fluoride (P7626; Sigma), 2 mmol/L sodium fluoride (S6776; Sigma), 0.5 mmol/L sodium orthovanadate (S6058; Sigma), and 2 μg/mL leupeptin (L851; Sigma). Dynabeads Protein A (10001D; Invitrogen) was conjugated to anti-HA antibody (3724; Cell Signaling Technologies), and the antibody-Dynabeads complex was incubated with the diluted serum overnight at 4°C. The beads were precipitated using a magnetic separator and washed three times with cold PBS with 0.02% Tween-20. The immunoprecipitants were denatured by boiling with the 1× Laemmli buffer for 5 minutes at 95°C.

Western Blot Analysis

Serum immunoprecipitants or kidney lysates were resolved on Tris-glycine SDS-PAGE gel, transferred onto a 0.2-μm polyvinylidene difluoride membrane (1620177; Bio-Rad, Hercules, CA) and were detected using an anti-HA antibody (1:1000; 3724; Cell Signaling Technologies), anti–T cell immunoglobulin cumin 1/kidney injury molecule-1/hepatitis A virus cellular receptor (TIM-1/KIM-1/HAVCR) antibody (1:1000; AF1817; Novus Biologicals, Littleton, CO), or anti–β-tubulin antibody (1:1000; 2146; Cell Signaling Technologies) as primary antibodies and rabbit anti–goat–horseradish peroxidase IgG (1:1000; HAF017; Novus Biologicals) or anti-rabbit IgG horseradish peroxidase–linked antibody (1:2000; 7074; Cell Signaling Technologies) as the secondary antibody.

Serum Biochemistry

Blood urea nitrogen and serum creatinine were measured using the QuantiChrom Urea Assay Kit (DIUR-100; BioAssay Systems, Hayward, CA) and Creatinine Assay Kit (265-30; Sekisui Diagnostics P.E.I. Inc., Charlottetown, PE, Canada), according to the manufacturer's recommendations.

uACR Measurement

Mouse urine was collected before and 14 days after the FA injection and analyzed for uACR. The urine albumin and creatinine were measured using a mouse albumin enzyme-linked immunosorbent assay kit (ab108792; Abcam, Cambridge, UK) and a Creatinine Assay Kit (265-30; Sekisui Diagnostics P.E.I. Inc.), according to the manufacturer's recommendation. Urine albumin levels were normalized by urine creatinine levels.

Histologic Analysis

Mouse kidneys were fixed overnight in 10% formalin (SF100-20; Thermo Fisher Scientific, Fair Lawn, NJ), embedded in paraffin, and sectioned (5 μm thick). Sections were deparaffinized and stained by picrosirius red (43665; Sigma) to visualize the collagenous fibers.²⁴ Images were taken using a light microscope (DM200; Leica Microsystems, Wetzlar, Germany). All histologic images were captured using a digital camera (DP72; Olympus Canada Inc., Richmond Hill, ON, Canada), and acquired with DP2-BSW software (XV3.0; Olympus Canada Inc.).

Statistical Analysis

Numeric variables were described as means ± SEM, means ± SD, or median (quartile 1 to quartile 3). Comparisons for the mice data were performed via t-test, one-way analysis of variance, followed by the Tukey multiple-comparison test, or two-way analysis of variance, followed by the Sidak multiple-comparison test using Prism software version 7 (GraphPad, Boston, MA). For patient data, SAS software version 9.4 (SAS Institute Inc., Cary, NC) was used. Bivariate association was assessed by Pearson or Spearman correlation coefficient and with adjustment using partial correlations. Multivariable ordinary linear regression was used to model MGP as a function of waist circumference, eGFR, uACR, and the product of eGFR and uACR.

Results

Elevated MGP Levels Are Highly Associated with Low eGFR and High uACR Levels

The mean age of the 99 patients with CKD was 64 years, and the mean eGFR was 25 mL/minute per m². In bivariate analysis, MGP was associated with body mass index and waist circumference but not with eGFR or uACR. There was no association between MGP and any laboratory parameters related to CKD–mineral and bone disorder, including parathyroid hormone, fibroblast growth factor-23, or fetuin (Supplemental Table 1). In a multivariable linear regression model, MGP was independently associated with waist circumference and uACR after controlling for eGFR, and there was a significant interaction between eGFR and uACR (Table 2). The raw data are overlaid on the model fit (Figure 1). For participants with higher levels of uACR, MGP level increased as eGFR declined; however, this association attenuated as uACR decreased. The highest MGP values tended to be among patients with low eGFR and high uACR.

Table 2.

Linear Regression Model Predicting MGP in Patients with CKD

Parameter estimates (n = 89)
Variable	Parameter estimate	SEM	P value	95% CI
Intercept	8.66	1.66	<0.0001	5.35	11.98
eGFR	0.03	0.03	0.24	–0.02	0.08
uACR	0.02	0.01	0.002	0.01	0.04
WC	0.06	0.02	0.0002	0.03	0.09
Interaction between uACR and eGFR	–0.001	0.0003	0.002	–0.002	–0.0004
Model summary R2 = 0.24 P = 0.0001

Ten observations were excluded because missing variable (n = 6), extreme MGP (n = 1), extreme uACR (n = 1), or being overly influential (n = 2).

CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; MGP, matrix Gla protein; uACR, urine albumin/creatinine ratio; WC, waist circumference.

Figure 1.

Elevated matrix Gla protein (MGP) levels are highly associated with low estimated glomerular filtration rate (eGFR; mL/min/m²) and high urine albumin/creatinine ratio (uACR; mg/mmol) levels. The color of the background represents the predicted MGP (nmol/L) estimated by the following linear model: MGP = 8.66 + 0.058 ∗ waist circumference (WC) + 0.030 ∗ eGFR + 0.022 ∗ uACR – 0.0011 ∗ eGFR ∗ uACR, with WC fixed at 100 cm. The bubbles represent the observed data, with the x and y location of the bubble depicting the eGFR and uACR values, respectively, and the area of the bubble proportional to the WC. The top right corner is missing because this region extrapolates the predicted MGP values beyond the range of the data. The horizontal dashed line depicts the uACR value for which the slope of eGFR on MGP is 0, and the vertical dashed line depicts the value of eGFR for which the slope of uACR on MGP is 0.

Increased MGP Expression in Murine Injured Kidneys

MGP expression was examined in mouse renal injury models using I/R surgery and FA injection. For the surgical method, 30 minute unilateral renal pedicle clamping followed by contralateral nephrectomy was done on 8- to 12-week–old female WT mice. For the nonsurgical procedure, 250 mg/kg FA was intraperitoneally administrated to the 8-week–old male WT mice (Figure 2A). Mgp expression was significantly increased in both models at the late stages—21 days after the I/R surgery (Figure 2B) and 14 days after the FA injection (Figure 2C). The FA administration model was used for the subsequent in vivo experiments to induce the injuries because of its procedural simplicity and high reproducibility.

Figure 2.

Matrix Gla protein (MGP) is highly expressed in injured kidney. A: Scheme showing the steps using ischemia-reperfusion (I/R) surgery and folic acid (FA) injection procedure to induce kidney injury in wild-type mice. B: Quantitative real-time PCR analysis of Mgp expression levels in the kidneys before and after I/R surgery in reference to Hprt. C: Quantitative real-time PCR analysis of Mgp expression levels in the kidneys before and after folic acid administration in reference to Hprt. Statistical analysis: one-way analysis of variance, followed by Dunnett correction, was used to calculate P. The results are presented as means ± SEM (B and C). n = 3 for each group (B and C). ∗P < 0.05, ∗∗∗P < 0.001.

In Vivo Labeling of Endogenous MGP

Mgp expression in the kidney has been studied in previous publications11, 12, 13, 25; however, it is still unknown which renal cells express Mgp. To identify them, Rosa26^Tomato/+ reporter mice were bred with Mgp-Cre mice to generate Rosa26^Tomato/+;Mgp-Cre mice, which expressed robust red fluorescence, tdTomato, following Mgp promoter-driven Cre recombinase expression in MGP-producing cells (Figure 3A). The intense tdTomato signal was observed in some tubular epithelial cells and interstitial cells in the kidney at 5 weeks of age (Figure 3B).

Figure 3.

Pericytes express Mgp in the kidney. A: Scheme represents Rosa^Tomato/+;Mgp-Cre mice generation. Cre recombinase expression following the Mgp gene expression deleted the stop cassette and expressed red fluorescent protein (RFP) fluorescence. B: Fluorescence microscopy showing the endogenous RFP expression in tubular interstitial cells of control (Rosa^Tomato/+) and Rosa^Tomato/+;Mgp-Cre kidneys. Nuclei were stained with Hoechst (blue). Bottom panels: Magnified images of yellow dotted boxed areas are presented. C:Mgp target sequence, the sequence of the guide RNA (blue), and single-stranded oligodeoxynucleotide (ssODN) to introduce the desired insertion. The ssODN carries homology arms flanking each side of the target nucleotide with hemagglutinin (HA) sequences (yellow boxed area). As shown, the guide RNA anneals on the stop codon (TAG) of Mgp (red boxed area), complementary to the protospacer adjacent motif (green circle), where co-injected clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 (Cas9) is expected to generate a double-stranded break. D: Left: A representative agarose gel image of genotyping PCR using HA/F3 and HA/R3 primers. Wild-type (WT) Mgp allele produces a 297-bp fragment, and HA-Mgp allele produces a 324-bp fragment. Right: Sanger sequencing analyses confirm the insertion of the HA sequence (yellow) before the stop codon (red). E: Confocal microscopy showing matrix Gla protein (MGP)–expressing cells in a healthy kidney. Tissues were stained using rabbit anti-CD31 or rabbit anti–platelet-derived growth factor receptor (PDGFR)-β with mouse anti-HA antibody, followed by anti-rabbit Alexa 555 (green or white) and anti-mouse Alexa 647 (red) antibodies. Nuclei were stained with Hoechst (blue). Right: Magnified images of yellow dotted boxed areas are presented. Scale bars: 10 μm (B, bottom panels, and E, left); 20 μm (B, top panels); 2 μm (E, right). IRES, internal ribosome entry site.

Although the tdTomato expression was regulated by the Mgp promoter, it may not always reflect the promoter activity at the time of sample collection. Furthermore, multiple cell types present in the interstitial region may also synthesize MGP. To identify these cells, co-immunostaining using antibodies against specific cell markers and MGP was required. However, multiple in-house and commercially available antibodies against mouse MGP failed to yield specific MGP detection in the experiments. To circumvent this issue, a new knock-in mouse expressing MGP carrying a human influenza HA tag at the C-terminal end was generated using the CRISPR/Cas9-based mutagenesis approach. The guide RNA–Cas9 complex together with the single-stranded oligodeoxynucleotide repair template (Figure 3C) were microinjected into the mouse fertilized eggs. The insertion of the desired sequence was examined by PCR and Sanger sequencing (Figure 3D). Homozygote mice (Mgp^HA/HA) were generated and used for subsequent experiments.

To identify the MGP-producing cells, kidneys were collected from 5-week–old Mgp^HA/HA mice, paraffin embedded, sectioned, and stained with the HA-tag antibody together with specific antibodies against platelet-derived growth factor receptor-β, a pericyte marker, or against CD31, an endothelial cell marker. HA staining was only observed in pericytes, not in endothelial cells (Figure 3E).

Myofibroblasts Are the Major Source of Up-Regulated MGP in Injured Kidneys

A significant increase in Mgp expression was observed in the FA-injured kidney (Figure 2C), but it was unclear which type of cells up-regulated MGP expression. To identify MGP-producing cells in injured kidneys, 250 mg/kg of FA was intraperitoneally administrated to the 8-week–old Mgp^HA/HA male mice, and the kidneys were collected after 2 weeks. Co-immunofluorescence staining revealed that most MGP-expressing cells were α-SMA–positive cells (myofibroblasts), but not CD31-positive cells (Figure 4A). Moreover, elevated MGP levels were also detectable from sera after the FA injections (Figure 4, B and C).

Figure 4.

Myofibroblasts in the injured kidney express matrix Gla protein (MGP). A: Confocal microscopy showing MGP-expressing cells in injured kidney. Tissues were stained using rabbit anti–α-smooth muscle actin (⍺-SMA) or rabbit anti-CD31 with mouse anti-hemagglutinin (HA) antibodies, followed by anti-rabbit Alexa 555 (green or white) and anti-mouse Alexa 647 (red) antibodies. Nuclei were stained with Hoechst (blue). Right: Magnified images of yellow dotted boxed areas are presented. B: Serum samples from Mgp^HA/HA mice after the folic acid administration were subjected to immunoprecipitation and blotted using anti-HA antibody. Bottom: The immunoblot of rabbit IgG heavy chain (IgG H) used for immunoprecipitation (loading control). C: Densitometric analysis demonstrates comparable protein expression. Statistical analysis: one-way analysis of variance, followed by Dunnett correction, was used to calculate P. n = 3 to 4 for each group (C). ∗P < 0.05. Scale bars: 5 μm (A, right); 10 μm (A, left).

Lack of MGP Up-Regulates Kidney Injury Marker without Affecting Kidney Function

Next, 250 mg/kg of FA was intraperitoneally administrated to 2-week–old WT and Mgp^–/– mice. Injections were performed at an early age, because of the short life span of Mgp^–/– mice. The changes in blood urea nitrogen, serum creatinine, and uACR levels were similar between these two groups after the FA injection (Figure 5, A–C). However, Mgp^–/– kidneys showed significantly higher levels of KIM-1 both before and 2 days after FA injections compared with WT kidneys, whereas this difference was no longer observed by 14 days after FA injections (Figure 5D). Interestingly, despite the significantly higher KIM-1 expression in Mgp^–/– kidneys, they appeared less affected than WT kidneys 14 days after the FA injection (Figure 5E).

Figure 5.

Loss of matrix Gla protein (MGP) aggravates the early tubular injury without affecting kidney function. A: Bar graphs showing the blood urea nitrogen (BUN) levels in wild-type (WT) and Mgp^–/– mice before and 2 and 14 days after the folic acid (FA) administration. B: Bar graphs showing the serum creatinine levels in WT and Mgp^–/– mice before and 2 and 14 days after the folic acid administration. C: Bar graphs showing the urine albumin/creatinine ratio levels in WT and Mgp^–/– mice before and 14 days after the folic acid administration. D: Kidney lysates from WT and Mgp^–/– mice before and 2 and 14 days after the FA administration were subjected to immunoblot analyses using anti–kidney injury molecule (KIM)-1 antibody. Bottom: The immunoblot of β-tubulin (loading control). Left: Densitometric analysis demonstrates comparable protein expression. E: WT and Mgp^–/– kidneys before or 14 days after the FA administration. A–D: Statistical analysis: two-way analysis of variance, followed by Sidak correction, was used to calculate P. The results are presented as means ± SD (A–D). n = 3 for each group (A–D). ∗P < 0.05, ∗∗∗P < 0.001. Scale bars = 2 mm (E). NS, not significant.

Lack of MGP Suppresses FA-Induced Fibrosis

Because MGP expression was detected in collagen-producing myofibroblasts, its impact on renal fibrosis was examined using Mgp^–/– mice. The degree of renal fibrosis in Mgp^–/– kidneys was compared with WT kidneys after FA-induced injury by sirius red staining of paraffin sections and quantitative real-time PCR. Sirius red staining showed significantly less fibrosis in Mgp^–/– kidneys compared with WT kidneys 14 days after the FA injection (Figure 6A). This was supported by the fibrosis markers, Col1a1, Col3a1, Eln, and Acta2 gene expression analysis by quantitative real-time PCR. These markers were not significantly increased in the Mgp^–/– kidneys even after FA injection (Figure 6B). To identify the reason for attenuated renal fibrosis in FA-injured Mgp^–/– kidneys, the number of pericytes was examined. Pericytes are the MGP-expressing cells in kidney that may differentiate to myofibroblasts and up-regulate the synthesis of collagen-rich ECMs. The imaging data demonstrated that there were significantly fewer numbers of pericytes in Mgp^–/– kidney (Figure 6C). In addition, a significant down-regulation of Hey1, a downstream target of Notch signaling, was confirmed by quantitative real-time PCR (Figure 6D).

Figure 6.

Loss of matrix Gla protein (MGP) suppresses the extracellular matrix expression and kidney fibrosis. A: Sirius red–stained histologic sections and quantification of stained area confirmed the collagen fibers in wild-type (WT) or Mgp^–/– kidneys at 2 weeks after the folic acid (FA) administration. For quantification, 10 fields were examined from each sample. B: Quantitative real-time PCR analysis of fibrotic marker gene expression levels in WT and Mgp^–/– kidneys before and after FA (or vehicle) administration in reference to Hprt. C: Confocal microscopy showing pericytes (arrowheads) in the WT and Mgp^–/– kidneys. Tissues were stained using rabbit anti–platelet-derived growth factor receptor (PDGFR)-β, followed by anti-rabbit Alexa 555 (red). Nuclei were stained with Hoechst (blue). Right: The bar graph represents the number of pericytes over total interstitial cells per field. For each sample, 10 fields were examined. D: Quantitative real-time PCR analysis of Hey1 expression levels in WT and Mgp^–/– kidneys before or 2 weeks after the FA administration in reference to Hprt. A–D: Statistical analysis: t-test (A and C) or two-way analysis of variance, followed by Bonferroni correction (B and D), was used to calculate P. The results are presented as means ± SEM (A–D). n = 4 for each group (A and C); n = 3 for each group (B); n = 3 to 5 for each group (D). ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗∗P < 0.0001. Scale bars: 100 μm (A, left); 20 μm (C). HPF, high-power field; NS, not significant.

Restoration of MGP Induces Renal Fibrosis

Next, Mgp expression was restored by introducing the Sm22⍺-Mgp transgene to Mgp^–/– mice (Mgp^–/–;Sm22⍺-Mgp mice), and renal fibrosis was examined. Although Sm22⍺, also known as Tagln, is a well-known marker for vascular smooth muscle cells, it is also expressed in renal pericytes (Supplemental Figure S1).^26,27 Additionally, Tagln expression was up-regulated after the FA injury in the kidneys (Figure 7A). Considering these, the Sm22⍺-Mgp transgene expression in Mgp^–/–;Sm22⍺-Mgp mice was expected to restore the MGP expression in the renal pericytes and myofibroblasts. Two-week–old WT, Mgp^–/–, and Mgp^–/–;Sm22⍺-Mgp mice were injected with 250 mg/kg of FA intraperitoneally, and kidneys were collected after 14 days. The Sm22⍺-Mgp transgene expression significantly increased in Mgp^–/–;Sm22⍺-Mgp kidneys after the FA injection (Figure 7B). The degree of renal fibrosis in Mgp^–/–;Sm22⍺-Mgp kidney was examined by sirius red staining of paraffin sections and quantitative real-time PCR and compared with WT and Mgp^–/– kidneys. Severe fibrosis comparable to that of FA-injured WT kidneys was detected in FA-injured Mgp^–/–;Sm22⍺-Mgp kidneys (Figure 7C). Furthermore, the expression of fibrotic ECM markers was significantly up-regulated in Mgp^–/–;Sm22⍺-Mgp kidneys after the FA injection (Figure 7D).

Figure 7.

Restoration of Mgp expression in myofibroblasts re-establishes renal fibrosis in Mgp^–/– mice. A: Quantitative real-time PCR analysis of Tagln expression levels in wild-type (WT) kidneys that received folic acid (FA) administration relative to noninjected kidneys in reference to Hprt. B: Quantitative real-time PCR analysis of Mgp expression levels in Mgp^–/–; Sm22a-Mgp kidneys before and after FA administration in reference to Hprt. C: Sirius red–stained histologic sections confirmed the collagen fibers in WT, Mgp^–/–, and Mgp^–/–; Sm22a-Mgp kidneys at 2 weeks after the FA injection. D: Quantitative real-time PCR analysis of fibrotic marker expression levels in Mgp^–/–; Sm22a-Mgp kidneys relative to WT or Mgp^–/– kidneys before and after FA administration in reference to Hprt. A, B, and D: Statistical analysis: t-test (A and B) or two-way analysis of variance, followed by Sidak correction (D), was used to calculate P. The results are presented as means ± SEM (A, B, and D). n = 3 to 6 for each group (A); n = 4 to 5 for each group (B); n = 3 to 4 for each group (D). ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001. Scale bars = 100 μm (C). NS, not significant.

Discussion

MGP is primarily known for its anti-mineralization functions in the vascular and cartilaginous tissues, where its expression levels are high.^15,28 Although low-level MGP expression has been detected in other tissues,²⁸ its expression is up-regulated in multiple pathologic conditions, including CKD.11, 12, 13, 14^,29 In a small cohort of patients, the present analysis revealed that after adjusting for waist circumference, MGP levels increased with albuminuria in patients with low eGFR and reached the highest levels among patients with CKD with both low eGFR and high albuminuria. These findings are consistent with that of Puzantian et al¹³ and Wei et al,²⁹ showing an inverse relationship between the levels of dephosphorylated uncarboxylated MGP and eGFR. In the same line of observation, Miyata et al¹¹ reported high interstitial MGP expression in patients with proteinuria and low eGFR that predicted poor kidney outcomes.

The critical importance of albuminuria lies in its association with the progression of kidney disease and the occurrence of cardiovascular events. The relationship between albuminuria and loss of kidney function is still being elucidated; however, an emerging narrative focuses on the response of tubular epithelial cells to filtered protein. Albuminuria stresses renal tubular cells and causes a shift toward production of profibrotic mediators, such as TGF-β1, which, in turn, leads directly to the production of ECM proteins and fibrosis. A reduction in albuminuria explained 36% of the beneficial effect of canagliflozin on cardiovascular events.³⁰ Understanding various signaling pathways altered in response to proteinuria, and the development of fibrosis is growing.

Although systemic MGP levels may influence kidney pathology, the local induction of MGP expression in affected kidney cells can also play a significant role in driving disease progression. This notion is supported by multiple animal studies. For example, Miyata et al¹¹ and Kida and Yamaguchi²⁵ used rat 5/6 nephrectomy and mouse unilateral ureteral obstruction (UUO) to investigate the correlation between increased Mgp expression and disease progression in rodent models. The increased Mgp expression was detected in the tubular-interstitial area of nephrectomized kidneys, but the specific cell types were not identified.¹¹ Moreover, the role of MGP in the kidney was not thoroughly examined in other studies. One key obstacle to detecting MGP in mouse tissues is the unavailability of specific antibodies. Several anti-MGP antibodies failed to show specific MGP detection in mouse tissue extracts in our hands. To overcome this technical limitation, two complementary genetic approaches were used to detect Mgp promoter activity, as well as the localization of MGP in healthy and fibrotic kidneys.

In the current study, an Mgp promoter-driven Cre expression model²¹ was used to induce red fluorescent protein, tdTomato, expression in cells producing MGP. Although this method is effective for confirming gene expression in tissues without extra steps of immunostaining, it does not provide real-time insights into promoter activity or confirm the presence of the protein in the tissue at the time of sample collection. Because gene expression levels can vary depending on the developmental and cell differentiation stages, additional characterization at the translational level is necessary. To achieve this, a second approach by tagging MGP in vivo with an HA tag was used, enabling direct detection of the protein using an anti-HA tag antibody.

The fluorescence tags are frequently used to easily visualize the protein of interest without immunostaining. However, this may not always be suitable for small proteins like MGP (approximately 14 kDa) because of the heavy molecular weight of these reporter proteins (approximately 25 kDa). Hence, epitope tags are good alternatives, which are smaller than the full-length protein tags and are less likely to affect the target protein's function. After a thorough literature review, an HA tag was selected for the present model because of its reported low non-specific background detection in various in vivo experiments.³¹ The HA tag is an epitope tag derived from the glycoprotein of the influenza virus, which is not known to interfere with protein function, while enabling highly specific detection of tagged proteins in immunoassays.³¹ Considering the critical functional role for MGP's N-terminal sequence,^17,23 this tag was added at the C-terminal end of MGP.

The in vivo reporter assays identified pericytes and myofibroblasts as the primary sources for local MGP production in the kidney under both physiological and pathologic conditions. Pericytes play critical roles in the perivascular system, where they stabilize the capillary structure and mediate angiogenesis and immune cell migration.^8,32 Reduction or detachment of pericytes is known to cause arteriovenous malformation and other vascular anomalies.³³ Interestingly, previous studies reported arteriovenous malformation in the brain, lungs, and kidneys of Mgp^–/– mice.^34,35 Our group also reported this pathology in the skeletal tissues in Mgp^–/– mice.³⁶ The current observation that pericyte numbers are reduced in Mgp^–/– kidneys might be a possible cause for arteriovenous malformation in these mice.

Pericytes are one of the major ancestral cell types giving rise to myofibroblasts, which play a major role in ECM synthesis in fibrotic kidneys.^9,37,38 Abnormal ECM synthesis by myofibroblasts worsens kidney function by reducing the number of functional nephrons, inducing capillary rarefaction, tubular atrophy, and vascular stiffening, eventually leading to irreversible progression to end-stage kidney disease.^5,6 It is, therefore, critical to understand the mechanisms of myofibroblast differentiation and identify the key molecules associated with this process. To date, well-known pathways involved in myofibroblast differentiation and ECM formation are TGF-β, Wnt/β-catenin, Hedgehog, and Notch signaling.^6,38 It was shown that inhibiting Notch signaling via transgenic or pharmacologic approaches significantly reduced kidney fibrosis in a UUO-injured model.^7,39

A previous study showed more severe renal fibrosis in Mgp heterozygous (Mgp^+/–) mice after UUO injury and suggested that MGP has a vascular protective effect in the injured kidney.²⁵ This contradicts the current findings demonstrating that the loss of MGP attenuates kidney fibrosis after FA-induced injury. This discrepancy could be caused by the use of mice with different Mgp gene dosage and methods to induce the renal injury in each study. Although Mgp^+/– mice carry one allele of the Mgp gene, the total Mgp expression levels in these mice can still be up-regulated to levels comparable to WT mice. Alternatively, Mgp^+/– mice may produce a reduced amount of MGP that nevertheless results in comparable adverse effects in the tissue. Interestingly, the initial study only showed increased ⍺-SMA–positive staining in UUO–injured Mgp^+/– kidneys, without examining Mgp expression and renal fibrosis after the injury.

KIM-1 is a phosphatidylserine receptor expressed in injured kidneys that is required for phagocytosis of apoptotic cells by the neighboring epithelial cells.⁴⁰ Although KIM-1 expression in Mgp^–/– kidneys was significantly higher than in WT kidneys, there were no significant alterations of kidney functions in these two genotypes, as evident by blood urea nitrogen, serum creatinine, and uACR levels. Interestingly, on FA-induced injury, Mgp^–/– kidneys recovered faster, as evident by reduced fibrotic lesions and lower expression of fibrotic markers by 14 days after injections. These observations suggest a role for MGP as a mediator of fibrosis in injured kidneys.

The reduced renal fibrosis in Mgp^–/– kidneys after FA treatment might be caused by a reduced number of pericytes, which, in turn, reduced the number of differentiated myofibroblasts, ultimately resulting in reduced ECM synthesis by these cells. In addition, significantly reduced expression of Hey1, a Notch downstream gene, in injured Mgp^–/– kidneys indicates that MGP deficiency may impair Notch signaling, thereby preventing kidney fibrosis directly. Several previous studies showed the relevance between MGP expression and cell survival, death, and proliferation in other tissues.41, 42, 43, 44, 45 Marulanda et al⁴⁵ showed that the loss of MGP causes chondrocyte cell death in the nasal septum. Newman et al⁴³ also showed down-regulation of MGP in chondrogenic ATDC5 cells, inducing apoptosis. On the other hand, increased MGP expression resulted in augmented cell proliferation^41,42 and increased stemness in cancer cells.⁴⁴ Although MGP might have similar roles in kidney pericytes, further studies will be needed to examine the mechanism(s) by which MGP may affect pericyte number and function in fibrotic kidneys.

The current study shows that there is an association between albuminuria and circulating levels of MGP in patients with CKD after adjustment for eGFR and obesity. Using in vivo reporter assays and genetic tagging, this study is the first to identify the specific cell types that express MGP in healthy and fibrotic kidneys in murine models. The negative role of MGP in the progression of renal fibrosis is also demonstrated. Overall, this research lays the foundation for future therapeutic strategies aimed at mitigating kidney fibrosis through the targeted down-regulation of MGP in the identified cell types. The newly developed mouse model expressing HA-tagged MGP offers a powerful tool to explore the spatiotemporal functions of MGP across various tissues in both health and disease.

Disclosure Statement

None declared.

Supplemental Data

Supplemental Figure 1

Transgelin (Tagln) expression in pericytes in multiple tissues. Dot blot showing Tagln expression in different pericytes. The color of the dot approximates average gene expression. Its size represents the percentage of cells within each cell type that expresses the gene. CZ CELLxGENE Discover: a single-cell data platform for scalable exploration, analysis, and modeling of aggregated data.²⁷