Estrogen protects female mice with chronic kidney disease from fibroblast growth factor 23-induced left ventricular hypertrophy

Abstract

Introduction:

Fibroblast growth factor 23 (FGF23) contributes to left ventricular hypertrophy (LVH) and mortality in chronic kidney disease (CKD). Males have a higher risk of cardiovascular events than females. The protective effects of estrogen on the heart are well established, but it remains unclear if cardiac FGF23 signaling is modified in females with CKD.

Methods:

We studied mineral metabolism, kidney and heart phenotypes of age-matched wild-type (WT) and Col4a3 knockout (Col4a3KO) mice during CKD progression and used transcriptomics to identify proximal targets of FGF23 involved in CKD-associated LVH. Additionally, we tested the effects of FGF23 and estradiol (E2) in vivo and on cultured neonatal mouse cardiomyocytes (NMCMs). All results were separated by sex.

Results:

Compared to WT, CKD males showed progressive increases in blood urea nitrogen (BUN) and FGF23 levels, overt LVH at 20 weeks, and premature death at 22 weeks. In contrast, CKD females showed earlier increases in BUN and FGF23 levels but did not develop LVH and lived longer than males. RNA sequencing analyses revealed that the proximal targets of FGF23 identified in males with CKD are established downstream targets of estrogen signaling. In vitro, FGF23 induced the hypertrophic growth of NMCMs isolated from male and female mice and E2 co-treatment prevented this effect. Finally, E2 prevented FGF23-induced calcineurin activity in the heart of male mice, whereas ovariectomy triggered the development of LVH in female mice with CKD.

Conclusions:

We identified common molecular targets of FGF23 and E2 signaling in the heart and show that estrogen antagonizes the hypertrophic effects of FGF23, supporting female-specific cardioprotective mechanisms in CKD.

Graphical Abstract

Introduction

Cardiovascular disease is a leading cause of death in patients with CKD.^1–3 Despite a higher prevalence of CKD, the risk of cardiovascular events is lower among women, which suggests a protective effect of female sex on the heart.^4–9 Although the cellular and molecular mechanisms underlying this sexual dimorphism are still unclear, sex hormones and their receptors may play an important role.^10,11 Indeed, estrogen is known to exert effects on numerous cells, including endothelial and vascular smooth muscle cells, fibroblasts and cardiomyocytes.¹² In premenopausal women, estrogens promote cardioprotective effects. As a result, women under 40 years of age with premature menopause have a higher risk of cardiovascular disease,¹³ and estrogen replacement therapy decreases the left ventricular mass in both normotensive and hypertensive postmenopausal women.^14,15 Estrogens might reduce left ventricular hypertrophy by regulating calcium signaling^16–18 and suppressing the calcineurin / nuclear factor of activated T cell (NFAT) pathway.^19–21 Whether estrogens provide the same level of cardiac protection in men is unknown.¹¹ Despite these observations, the role of sex hormones in cardiovascular disease has not been extensively studied in CKD. This is of outmost importance given that substantial reductions in estrogen levels, early menopause, menstrual disorders, and infertility are often observed in women with CKD, which may negatively impact cardiovascular health.^22,23

In CKD, increased levels of circulating fibroblast growth factor 23 (FGF23) are independently and positively associated with cardiovascular disease and all-cause mortality.^24–27 FGF23 is a bone-produced hormone that primarily targets the kidney to inhibit phosphate reabsorption and vitamin D production. In CKD, FGF23 levels rise progressively to reach up to a 1,000-fold the normal levels in end-stage kidney disease.^24,25,28,29 Although elevated FGF23 maintains normal serum phosphate levels, FGF23 also exerts extra-renal effects, and the heart is one of the major FGF23 targets. Indeed, at sustained high concentrations, FGF23 induces left ventricular hypertrophy (LVH), which is an important precursor of heart failure in patients with CKD.^30,31 In the heart, activation of FGF receptor 4 (FGFR4) by FGF23 results in downstream activation of phospholipase Cγ / calcineurin / NFAT signaling and hypertrophic growth of the cardiomyocytes.^32,33 Although the impact of elevated FGF23 on cardiac hypertrophy is well described, it is unknown if the effects of FGF23 vary by sex and remains to be investigated independently in each sex.

In the present study, we investigated the role of FGF23 in the development of LVH in male and female mice. We used the established Col4a3 knockout mouse model of progressive CKD ³⁴ to assess the longitudinal modifications of kidney function, mineral metabolism, and heart morphology and function in each sex during CKD progression. We report that unlike males, female mice with advanced CKD do not develop LVH despite earlier onset of CKD and FGF23 excess. We used transcriptomics analyses in hearts and cardiomyocytes to identify FGF23-regulated genes involved in CKD-associated cardiac hypertrophy and putative regulators of these targets in males and females. Finally, based on these analyses, we demonstrate the direct role of estrogen in antagonizing the hypertrophic targets of FGF23 in males and females. Together, this study shows that FGF23 and estrogen share downstream molecular targets in the heart and proposes a new molecular mechanism leading to female-specific cardioprotective effects in mice with CKD.

Methods

Animal studies

We studied age-matched male and female wild-type (Col4a3^+/+, WT) and homozygous mutant (Col4a3^−/−, Col4a3^KO) littermates, C57BL/6J, and transgenic mCherry^Myh6-TG expressing a cardiomyocyte-specific mCherry tag driven by a modified alpha myosin heavy chain promoter as detailed in Supplemental Data. All animal studies were performed in accordance with the Northwestern University Institutional Animal Care and Use Committee approved protocols.

Calcineurin activity assay

We assessed calcineurin activity in heart protein extracts of C57BL/6J male and female mice using the Calcineurin Cellular Activity Kit (BML-AK816, Enzo Life Sciences Inc., Farmingdale, NY, USA) as detailed in Supplemental Data.

Serum and urine biochemistry

We performed biochemical analyzes of serum and urine as previously described ^34–38 and detailed in Supplementary data.

Echocardiography

One week prior to sacrifice, we assessed heart morphology and function in all mice using a Vevo 770^® and Vevo 3100 high-resolution micro-imaging echocardiography system (FUJIFILM VisualSonics, Toronto, Canada). While imaging, we maintained all mice under 1.5% isoflurane anesthesia, and body temperature at normal level. We analyzed the acquired images using the Vevo^® 770 Workstation Software and VevoLab (FUJIFILM VisualSonics, Toronto, Canada), as previously described.^34,37

Histology

Histological analyzes of hearts were performed as previously described and detailed in Supplementary data.

Flow Cytometry

Freshly isolated cells harvested from the hearts of mCherry^Myh6-TG mice were sorted as detailed in Supplementary data.

RNA Sequencing

We performed bulk RNAseq on heart samples from 20-week-old WT and Col4a3^KO mice, on mCherry-sorted cardiomyocytes 2 hours post-FGF23 and vehicle injection, and on NMCM cultures 6h post-treatment with recombinant mouse FGF23 and/or beta-Estradiol (E2) as detailed in Supplementary data.

Neonatal murine cardiomyocyte cultures

We separated mCherry^Myh6-TG mouse neonates by sex as previously described,³⁹ and isolated primary neonatal murine cardiomyocytes (NMCMs) from neonates of each sex using the Pierce^™ Primary Cardiomyocyte Isolation kit (88281, Thermo Fisher Scientific Inc., Waltham, MA, USA). We cultured cells in presence of recombinant mouse FGF23 (mFGF23, 26-FG-025, R&D Systems, Minneapolis, MN, USA) and E2 (E2758–250MG, Sigma-Aldrich, St. Louis, MO, USA) as detailed in Supplementary data.

Immunofluorescence

We quantified NMCMs mean area based on the measurements of 100 individual cells per treatment condition, using LAS X software (Leica Microsystems Inc., Buffalo Grove, USA) as detailed in Supplementary data.

Ischemia-reperfusion injury

Female C57BL/6J mice underwent sham (sham) or unilateral kidney ischemia reperfusion injury (uIRI, 23min) at 10 weeks of age. At 12 weeks of age, mice underwent either a second sham surgery or a contralateral nephrectomy and ovariectomy (Nx-OVX).

Blood pressure measurements

Blood pressure was assessed in 20-week-old WT and Col4a3^KO male and female mice using a computerized mouse tail-cuff system (CODA, Kent Scientific, Torrington, CT) as previously described.^34,37

RT-qPCR

We performed RT-qPCR as previously described ^34–38 and detailed in Supplementary data.

Statistics

All values are represented as mean ± SEM. We used Gehan’s Wilcoxon test for comparative survival analysis. We used parametric analysis of variance (ANOVA) followed by post hoc Bonferroni tests for multiple groups comparisons. (Statistica software, Statsoft, Tulsa, OK, USA). Differences were considered statistically significant at P values < 0.05.

Results

Females show earlier onset of CKD, but longer lifespan, than males with CKD

We investigated the lifespan and kidney morphology and function of WT and CKD male and female littermate mice using the Col4a3 knockout mouse model of progressive CKD. Consistent with previous reports,^34,40,41 Col4a3^KO male mice showed a significant reduction in lifespan compared to WT males. Col4a3^KO females also showed a reduced lifespan but lived on average 2 weeks longer than Col4a3^KO males (24.0 ± 1.0 vs. 22.0 ± 0.6 weeks, p<0.05, Fig 1a). We next analyzed CKD progression in a separate set of mice from 4 to 20 weeks of age. All mice gained weight throughout the study period. At 20 weeks of age, male mice with advanced CKD showed a lower body weight than WT controls (Fig 1b). Consistent with impaired kidney function, male and female mice with CKD developed polyuria (Fig 1c), increased blood urea nitrogen (BUN) levels (Fig 1d), and urinary albumin to creatinine ratio (ACR) (Fig 1e) compared to WT. Interestingly, these increases either occurred earlier during CKD progression (Fig 1c–d) or were more pronounced (Fig 1e) in Col4a3^KO females than in Col4a3^KO males. By 20 weeks, all males and females with CKD showed similar levels of glomerulosclerosis, tubular atrophy and severe interstitial fibrosis (Fig 1f–g).

Figure 1: CKD females show longer lifespan than CKD males despite earlier onset of CKD.

a) Kaplan Meier cumulative proportion of wild-type (WT) and Col4a3^KO male and female mice surviving. Both Col4a3^KO males and Col4a3^KO females show reduced lifespan (p<0.05 vs. WT; n = 11 per group). Col4a3^KO females survive longer than Col4a3^KO males (p<0.05). b) Body weight, c) 24h urine volume, d) blood urea nitrogen levels (BUN), e) urinary albumin to creatinine ratio (ACR) and f) kidney fibrosis in 4 to 20-week-old WT male (n = 7–21) and female (n = 5–18) and Col4a3^KO male (n = 5–19) and female (n = 5–16) mice. Values are represented as mean ± SEM. P < 0.05 in age-matched (^a) Col4a3^KO vs. WT males, (^b) Col4a3^KO vs. WT females, (*) WT females vs. males, and (^#) Col4a3^KO females vs. males. g) Representative periodic acid-Schiff (PAS, top row) and picro-sirius red (PSR) staining by bright-field (BF, middle row) and polarized light microscopy (POL, bottom row) of kidneys from 20-week-old WT and Col4a3^KO male and female mice (scale bar = 50 μm).

Females with CKD show earlier increases in FGF23 levels than males with CKD

All mice with CKD showed alterations of mineral metabolism compared to WT littermates. Consistent with previous studies in this model,^34–38,42 Col4a3^KO males showed increased circulating cFGF23, iFGF23, Pi, and PTH levels by 20 weeks of age (Fig 2a–d), while serum 1,25-dihydroxyvitamin D (1,25(OH)₂D) and calcium levels remained stable throughout the study period (Fig 2e–f). In contrast, Col4a3^KO females showed earlier elevations in cFGF23, iFGF23 and PTH levels (16 weeks). cFGF23 and iFGF23 levels were higher in females than males with CKD, albeit not significantly (Fig 2a–b). Despite earlier increases in FGF23 and PTH levels, females showed trends to lower urinary Pi excretion than males throughout the study period that became significant at 20 weeks when overt hyperphosphatemia occurs (Fig 2c, S1). At 20 weeks, kidney Fgfr1 mRNA remained unchanged and kidney Klotho mRNA was decreased to similar levels in Col4a3^KO males and females (Fig S1). Compared to sex-matched WT, 1,25(OH)₂D levels also remained unchanged in Col4a3^KO females, albeit significantly lower than in Col4a3^KO males at 20 weeks (Fig 2e). From 8 weeks, Col4a3^KO females show a steady increase in urinary calcium excretion that plateau after 16 weeks possibly leading to hypercalcemia observed at 20 weeks (Fig 2f, S1).

Figure 2: Mineral metabolism is altered earlier in CKD females than in CKD males.

Serum levels of a) total FGF23 (cFGF23), b) intact FGF23 (iFGF23), c) phosphate (Pi), d) parathyroid hormone (PTH), e) 1,25-dihydroxyvitamin D (1,25(OH)₂D) and f) calcium (Ca²⁺) in 4 (a-c) and 8 (d-f) to 20-week-old WT male (n = 6–16) and female (n = 5–15) and Col4a3^KO male (n = 6–15) and female (n = 5–16) mice. Values are represented as mean ± SEM. P < 0.05 in age-matched (^a) Col4a3^KO vs. WT males, (^b) Col4a3^KO vs. WT females, (*) WT females vs. males, and (^#) Col4a3^KO females vs. males.

Males, but not females, with CKD develop LVH

Elevated levels of FGF23 directly contribute to the development of LVH and mortality in patients and mice with CKD.^25,32 Col4a3^KO male mice with CKD exhibit similar characteristics as patients with CKD, including increased FGF23 levels, LVH and a short lifespan.³⁴ However, in the current study, we report that Col4a3^KO females have a longer lifespan than males despite earlier increases in FGF23 levels and onset of CKD. Thus, we next compared the heart morphology and function of WT and Col4a3^KO males and females from 8 to 20 weeks of age. Consistent with previous studies,^34,35,37,38 heart histology analyses of 20-week-old males with advanced CKD showed an increase in heart size (Fig 3a), and a significant increase in cardiomyocyte area and perimeter (Fig 3b–d) compared to WT males. Echocardiography analyses also revealed significant increases in LV mass and LV posterior wall thickness, and unchanged LV internal diameter at 20 weeks (Fig 3e–h, Table S1–4). Before 20 weeks of age, the heart morphology of Col4a3^KO males was similar to WT. Together, this supports the development of LVH in males with advanced CKD. In contrast, the heart morphology of Col4a3^KO females remained unchanged at all timepoints suggesting that they did not develop LVH (Fig 3a–h). Ejection fraction and stroke volume remained unchanged in all mice (Fig 3i–j). At 20 weeks, blood pressure and cardiac index were increased to similar levels in both Col4a3^KO males and females (Fig 3k, S2).

Figure 3: Male, but not female, Col4a3^KO mice with CKD develop LVH.

a) Representative hematoxylin and eosin (H&E) staining (scale bar = 2 mm) and b) wheat germ agglutinin (WGA) staining (scale bar = 50 μm) of hearts from 20-week-old WT and Col4a3^KO male and female mice. c) Cardiomyocyte perimeter and d) cardiomyocyte area quantified from WGA staining of hearts from 8 to 20-week-old WT and Col4a3^KO male and female mice. e) Representative echocardiography images in M-mode, f) Left ventricular (LV) mass, g) LV posterior wall thickness at diastole (LVPW;d), h) LV internal diameter at diastole (LVID;d), i) ejection fraction (EF), j) stroke volume, and k) cardiac index in 8 to 20-week-old WT male (n = 5–14) and female (n = 5–10) and Col4a3^KO male (n = 5–10) and female (n = 5–12) mice. Values are represented as mean ± SEM. P < 0.05 in age-matched (^a) Col4a3^KO vs. WT males, (^b) Col4a3^KO vs. WT females, (*) WT females vs. males, and (^#) Col4a3^KO females vs. males.

Estrogen signaling is a predicted regulator of CKD-associated cardiac hypertrophy genes

To identify the molecular mechanisms involved in this sexual dimorphism, we performed RNA-sequencing in whole hearts isolated from 20-week-old WT and Col4a3^KO male and female mice. Using comparative pathway analyzes (IPA), we identified a cluster of signaling pathways predicted to be upregulated only in Col4a3^KO males (vs. WT males), but not in Col4a3^KO females (vs. WT females) (Fig 4a). A total of 116 genes (Table S5) significantly modified in Col4a3^KO males, but not in Col4a3^KO females, were present in these pathways. The modified expression of calcium/calcineurin/NFAT signaling molecules, known to be induced by FGF23 in the heart, predicted cardiac hypertrophy in Col4a3^KO males but not in females (Fig S3). We performed a secondary pathway analysis on these 116 genes and show that the top 6 pathways are all predicted to be activated in Col4a3^KO vs. WT males, including cardiac hypertrophy signaling and estrogen receptor signaling (Fig 4b). Of these 116 genes, we identified 40 genes involved specifically in cardiac hypertrophy signaling (Fig 4c, Table S5), 24 of which shared estrogen as their top upstream regulator (Fig 4d, Table S5). This suggests that genes that are differentially regulated in CKD-associated LVH are established targets of estrogen signaling. Of note, the expressions of estrogen receptor 1 (Esr1) and receptor 2 (Esr2) were similar between males and females (Fig S4).

Figure 4: Estrogen signaling is a predicted upstream regulator of cardiac hypertrophy genes in mice with CKD.

Ingenuity pathway analysis (IPA) of RNA sequencing performed on WT and Col4a3^KO male and female hearts at 20 weeks of age (n = 3 per group). a) Pathways are clustered according to patterns of predicted activation and inhibition. The frame delineates the main cluster of interest containing predicted upregulated canonical pathways in Col4a3^KO males (vs. WT males - red), but not in Col4a3^KO females (vs. WT females - black). In this cluster, all pathways cumulate 116 genes with significantly altered expression (p < 0.05, FDR < 0.1 cut-off). b) Canonical pathway analysis of the 116 genes identified in a) was performed in the male datasets. The represented top 6 canonical pathways are all predicted to be activated in Col4a3^KO vs. WT males. c) Forty of the 116 genes are related to cardiac hypertrophy signaling and show significantly altered expression (p < 0.05, FDR < 0.1 cut-off) represented as a heatmap. d) Upstream regulators analysis performed on these cardiac hypertrophy genes identified that 24 of the 40 genes share estrogen as their top upstream regulator (p < 0.05, FDR < 0.1 cut-off). The heatmap represents the normalized expression of the 24 estrogen-regulated genes in each group.

Estrogen signaling is a predicted regulator of FGF23-regulated cardiac hypertrophy genes in cardiomyocytes

To determine the cardiomyocyte-specific proximal targets of FGF23, we performed RNAseq in primary cardiomyocytes isolated from 20-week-old WT male and female mice 2h after a single injection of mouse recombinant FGF23 (50 ng/g) or vehicle. Compared to sex-matched Ctr mice, we identified 1014 differentially regulated genes in cardiomyocytes from FGF23-treated male mice, and only 396 genes in cardiomyocytes from FGF23-treated female mice (Fig 5a). Subsequent canonical pathway analysis revealed predicted inhibitions and activations of multiple signaling pathways in male but not in female cardiomyocytes after FGF23 treatment, including the activation of cardiac hypertrophy signaling represented by a set of 30 differentially regulated genes (Fig 5b). Among these 30 FGF23-regulated genes (Table S6), 11 genes shared estrogen as their top upstream regulator (Fig 5c). This indicates that genes that are differentially regulated in cardiomyocytes in response to FGF23 are established targets of estrogen signaling.

Figure 5: Estrogen signaling is a predicted upstream regulator of cardiomyocyte hypertrophy genes in FGF23-treated mice.

Ingenuity pathway analysis (IPA) of RNA sequencing performed on mCherry-sorted cardiomyocytes isolated from 20-week-old mCherry^Myh6-TG+ male and female mice 2h after a single injection of 50 ng/g recombinant FGF23 or saline vehicle (n = 5 per group). a) Venn diagram representing male-specific and female-specific differentially regulated genes. b) Differentially regulated canonical pathways include a set of 30 genes involved in cardiac hypertrophy signaling, predicted to be upregulated in male (red box), but not in female (black box) cardiomyocytes. c) The heatmap represents the expression of the 30 FGF23-regulated genes involved in cardiac hypertrophy signaling. Upstream regulators analysis performed on these cardiac hypertrophy genes identified that 11 of the 30 genes share estrogen as their top upstream regulator (black boxes). The cutoff values for all analyses were p<0.05, FC=5 and FDR < 0.05 vs. vehicle.

Estradiol prevents FGF23-induced cardiomyocyte hypertrophy

To test the direct effects of estrogen-mediated signaling on FGF23-induced hypertrophy, we cultured primary cardiomyocytes isolated from male and female mouse neonates and treated these NMCMs with incremental doses of FGF23, estradiol (E2), or both for 48h (Fig 6a–b). Consistent with previous reports,^32,33 FGF23 induced the hypertrophic growth of male NMCMs. In absence of E2, FGF23 also induced the hypertrophic growth of female NMCMs. Contrary to FGF23, E2 alone had no effect on male cells and slightly reduced the size of female NMCMs. Irrespective of sex, the size of NMCMs did not change when co-treated with FGF23 and E2, indicating that E2 fully prevented the FGF23-induced hypertrophic growth. Next, we performed RNAseq of male and female NMCMs 6h post-FGF23 and E2 treatment (Fig 6c–d, Table S7). We identified a cluster of 45 cardiac hypertrophy genes that were significantly regulated by FGF23 and E2 in male and female NMCMs. Among these genes, 33 changed in response to FGF23 in male NMCMs while only 8 genes were modified 6h post-treatment in female NMCMs. FGF23 effect was partially prevented by E2 co-treatment in male NMCMs (Fig 6c, Table S7). Consistent with these results, pathway analysis showed a significant increase in cardiac hypertrophy and calcium signaling in FGF23-treated male NMCMs that was partially prevented by E2 co-treatment. Female NMCMs showed a mild increase in cardiac hypertrophy signaling that was fully prevented by E2 co-treatment (Fig 6d, S5). To further demonstrate that estrogen directly inhibits FGF23-induced cardiac hypertrophy signaling, we injected male and female mice with 50 ng/g recombinant FGF23 or vehicle and pre-treated the male mice with 45 ng/g E2 or vehicle thirty minutes before FGF23 injection. We measured the activity of the calcineurin in heart protein extracts collected fifteen minutes post-FGF23 injection (Fig 6e). Consistent with previous results, FGF23 increased calcineurin activity in males, but not in females. This effect was cancelled by pre-treatment with E2 in males. Together, these results show that estrogen antagonizes the hypertrophic effects of FGF23 in cardiomyocytes.

Figure 6: Estradiol fully prevents FGF23-induced hypertrophy in female, but not in male, cardiomyocytes in vitro.

Primary mouse cardiomyocytes (NMCMs) were isolated from male and female neonates, differentiated for 7d in culture, and treated for 48h with 0, 25, 50 or 100 ng/mL recombinant FGF23 (a-b) or 6h with 0 or 50 ng/mL recombinant FGF23 (c-d) and 0 or 100 nM beta-estradiol (E2). a) Representative immunofluorescence staining of cultured (NMCMs). Green: anti-sarcomeric alpha actinin, Blue: NucBlue (nuclei) (scale bar = 50 μm). b) Quantification of cardiomyocyte area in male (squares) and female (circles) cells. Values are represented as mean ± SEM. N = 4 per group. P < 0.05 vs. sex-matched (^a) 0 ng/mL FGF23 / 0 nM E2, (^b) 0 ng/mL FGF23 / 100 nM E2, (^c) FGF23 concentration-matched / 0 nM E2, and vs. treatment-matched (*) males. c) Heatmap of mRNA expression of differentially regulated cardiac hypertrophy genes, and d) R plot of differentially regulated pathways in male and female NMCMs vs. untreated controls. N = 3–4 per group. P < 0.1 vs. untreated controls. e) CaN activity in hearts of male (squares) and female (circles) mice treated with 0 or 50 ng/g i.p. recombinant FGF23 and 0 or 45 ng/g s.c. E2. Values are represented as mean ± SEM and fold change of male vehicle control. N = 6–12 per group. (^a) P < 0.05 vs. sex-matched vehicle control.

Ovariectomy induces LVH in females with CKD, but not in healthy females

To demonstrate that estrogen deficiency triggers the development of LVH in female mice with CKD, we performed sham and ovariectomy (OVX) surgeries in the Col4a3^KO model of CKD (Fig 7) and in the surgical injury model of renal unilateral ischemia reperfusion with contralateral nephrectomy (uIRI-Nx – Fig S6). Compared to sham control mice, OVX did not significantly affect parameters of kidney function, mineral metabolism and heart morphology and function in WT females (Fig 7). Consistent with previous results (Fig 1–3), sham-Col4a3^KO females showed impaired kidney function, proteinuria, elevated serum FGF23 levels and hyperphosphatemia. OVX did not further affect the parameters of kidney function and mineral metabolism in Col4a3^KO mice (Fig 7b–f). In contrast, OVX-Col4a3^KO females showed increased parameters of cardiac hypertrophy compared to sham-Col4a3^KO mice (Fig 7g–j, Table S8), supporting a protective effect of estrogen in CKD. We observed similar results in the model of uIRI-Nx, in which OVX also triggered the development of LVH despite milder kidney injury than in Col4a3^KO females (Fig S6, Table S9).

Figure 7: Estrogen deficiency triggers the development of LVH in Col4a3^KO female mice.

a) Body weight, b) blood urea nitrogen (BUN), c) urinary albumin to creatinine ratio (ACR), serum levels of d) total FGF23 (cFGF23), e) intact FGF23 (iFGF23), f) phosphate (Pi), g) representative echocardiography images in M-mode, h) left ventricular (LV) mass, i) LV anterior wall thickness at diastole (LVAW;d), and j) ejection fraction (EF) in 20-week-old WT and Col4a3^KO female mice 8 weeks after sham surgery (sham) or ovariectomy (OVX). Values are represented as mean ± SEM. N = 6–12 per group. P < 0.05 vs. (^a) sham WT, (^b) OVX WT, and (^c) sham Col4a3^KO.

Discussion

Increased FGF23 levels are independently associated with the development of LVH and with cardiovascular mortality in CKD.^24,25,30,32 Despite a higher prevalence of CKD in women, male sex is associated with a higher risk of cardiovascular events and mortality.^4–9 It is unclear if FGF23 contributes to this sexual dimorphism. The Col4a3^KO mouse is an established model of progressive CKD that recapitulates many characteristics observed in patients with CKD, including a shortened lifespan.^34,35,37,38 In the present study using this model, we report that female mice with CKD live longer than male mice with CKD. To understand this difference, we investigated the mineral metabolism factors that contribute to mortality in CKD, including circulating levels of FGF23, PTH and Pi, as well as onset and progression of CKD, and their combined impact on cardiac morphology and function. Interestingly, compared to males, female mice with CKD show an earlier onset of kidney disease, and earlier elevations of FGF23 and PTH levels. As males with advanced CKD develop LVH, we show that females surprisingly do not, possibly contributing to their extended lifespan. Because this occurred despite higher FGF23 levels, we investigated if FGF23 signaling was differentially regulated in male and female mice. We show that molecular targets of FGF23 are counter-regulated by estrogen in the heart. As a result, FGF23 induces the development of LVH in male with CKD, but not in females. This study is the first to investigate how cardiac FGF23 signaling is modified by sex-specific mechanisms in CKD.

The genetic mutation of Col4a3 results in Alport syndrome in humans and mice.^40,43 The Col4a3 knockout mouse is used as a well-established model of CKD that is characterized by a spontaneous and progressive decline in kidney function, altered bone and mineral metabolism, cardiovascular disorders, and shortened lifespan. We previously reported, and show again in the present study, that Col4a3^KO male mice on a C57BL6/J genetic background develop LVH in advanced CKD.^34,37 In addition, we show for the first time that C57BL6/J Col4a3^KO females display an earlier onset of CKD than males, indicating a sexual dimorphism in this model. The early onset of CKD did not negatively impact the lifespan of Col4a3^KO females as they surprisingly lived longer than Col4a3^KO male littermates. Consistent with a dimorphic phenotype, and contrary to Col4a3^KO males, Col4a3^KO female littermates with advanced CKD did not develop LVH. This finding provides a new model to study cardiovascular disorders and sexual dimorphism associated with CKD. Interestingly, these differences in heart morphology were not due to differences in blood pressure, which was increased to the same levels in males and females with CKD. To understand why females with CKD did not develop LVH like males, we investigated if FGF23 signaling was modified in the heart of female mice.

As shown in previous studies, Col4a3^KO male mice display increased levels of serum cFGF23, iFGF23, PTH and Pi, consistent with altered mineral metabolism.^34–38,42 We provide new data showing that the same is true for Col4a3^KO female mice, who show earlier elevations of FGF23 levels than males. The factors contributing to this early increase are still unclear. Higher serum Pi levels have been previously reported in women compared to men, including in patients with CKD.^44–46 In the Heart and soul study, among participants with stable cardiovascular disease and reduced kidney function, women had significantly higher serum FGF23 levels than men, and the use of estrogen in women decreased FGF23 levels.^47,48 Since serum phosphate levels in female mice with CKD mirrored those of males despite higher circulating FGF23, might indicate increased end-organ resistance to FGF23. This is consistent with observations of non-significantly reduced urinary phosphate excretion and kidney Klotho expression in females. Additional female-specific mechanisms might also contribute to earlier FGF23 production and secretion. For instance, the administration of estrogen to ovariectomized rats with CKD has been shown to increase FGF23 levels leading to a reduction in PTH synthesis and secretion.^49,50 In addition, calcium was previously shown to stimulate FGF23 production.⁵¹ In the current study, contrary to males, Col4a3^KO females developed hypercalcemia. Because Col4a3^KO females showed similar PTH levels and lower 1,25(OH)₂D levels than males, it suggests that intestinal calcium absorption was not increased. Urinary calcium excretion was higher in Col4a3^KO females throughout CKD progression which together suggests that an earlier calcium release from the bone might explain hypercalcemia in females. Whether this contributes to early increases in FGF23 in females, consistent with the stimulatory effect of calcium on FGF23,⁵¹ is unclear.

Given that females with CKD do not develop LVH despite FGF23 excess suggests the presence of additional mechanisms protecting females from developing FGF23-induced cardiac hypertrophy. In RNAseq analyses, we show that most genes involved in the development of CKD-associated LVH, and early, acute targets of FGF23, are also downstream targets of estrogen signaling. Cardioprotective mechanisms have been described in females in previous studies, and sex hormones and their receptors play an important role.^10,11 Estrogen has been shown to reduce infarct size and to improve post-ischemic myocardial function in models of cardiac ischemia.¹⁰ It was also shown to have antihypertrophic effects in pressure overload models, inhibiting the development of LVH.¹⁰ Estrogen might reduce left ventricular and cardiomyocyte hypertrophy in part directly by regulating calcium signaling ^17,18 and suppressing the NFAT pathway.^19–21 Notably, FGF23-induced hypertrophic growth of cardiomyocytes is similarly mediated by the NFAT signaling cascade, through activation of FGFR4. This suggests that estrogen might counteract the action of FGF23 on the heart and prevent females with CKD from developing LVH. This was supported by additional data showing that ovariectomy led to the development of LVH in female with CKD, but not in healthy controls. Therefore, we hypothesized that there is a possible overlap between estrogen- and FGF23-mediated signaling and tested the effects of estrogen in the context of FGF23-induced hypertrophy.

In vitro, RNAseq analyses performed 6 hours post-FGF23 treatment show that the hypertrophic signal was less pronounced in female than in male NMCMs, but 48 hours of FGF23 treatment induced a similar hypertrophic growth of both male and female NMCMs. This suggests a delayed hypertrophic response of female cardiomyocytes to FGF23. Whether this was due to potential increased estrogen signaling, decreased FGFR signaling or other modified female-specific factors interfering with FGF23 activity at baseline remains unclear. In both sexes, FGF23 hypertrophic signal was blunted by co-treatment with E2. This correction was more potent in female cells, in particular in cells that received the highest dose of FGF23, which suggests that some beneficial effects of estrogen are lost in males under saturating FGF23 concentrations. In previous study, estrogen receptor expression was reported to be lower in males,⁵² but in contrast we found similar levels of estrogen receptor expression in the heart of adult male and female mice in the current study. In addition, pre-treatment of male mice with E2 fully prevented the acute activation of calcineurin by FGF23, demonstrating that estrogen antagonize FGF23-induced hypertrophic signaling in the heart irrespective of sex.

To conclude, our results show that FGF23 and estrogen share gene targets involved in the development of cardiac hypertrophy, and that estrogen counteracts the direct hypertrophic effects of FGF23 on cardiomyocytes, preventing the development of FGF23-induced LVH in females. Additional translational and clinical studies will be needed to determine whether FGF23 is associated with LVH to the same extent in men and women with CKD, and with varying estrogen levels in women with CKD. Importantly, the peri- and post-menopausal decrease in estrogen levels alone or combined with potentially exacerbated FGF23 excess in women with CKD might trigger a higher risk of LVH due to the decline in antihypertrophic estrogen. As a result, post-menopausal women with CKD with the highest FGF23 levels might be especially vulnerable.

Translational statement.

Increased fibroblast growth factor 23 (FGF23) levels directly contribute to the development of cardiac hypertrophy and mortality in patients with chronic kidney disease (CKD). Studies that investigate the sex-specific effects of FGF23 on the heart are needed. We show that in mice, estrogen and FGF23 regulate similar molecular targets in the heart. We also show that, as a result, estrogen antagonizes the cardiac hypertrophic effects of FGF23. Our study demonstrates female-specific cardioprotective mechanisms and suggests that the risk of FGF23-induced cardiac hypertrophy might increase with reduced estrogen levels in women with CKD.