Extra-Large Gα Protein (XLαs) Deficiency Causes Severe Adenine-Induced Renal Injury with Massive FGF23 Elevation

Abstract

Fibroblast growth factor-23 (FGF23) is critical for phosphate and vitamin D homeostasis. Cellular and molecular mechanisms underlying FGF23 production remain poorly defined. The extra-large Gα subunit (XLαs) is a variant of the stimulatory G protein alpha-subunit (Gsα), which mediates the stimulatory action of parathyroid hormone in skeletal FGF23 production. XLαs ablation causes diminished FGF23 levels in early postnatal mice. Herein we found that plasma FGF23 levels were comparable in adult XLαs knockout (XLKO) and wild-type littermates. Upon adenine-rich diet-induced renal injury, a model of chronic kidney disease, both mice showed increased levels of plasma FGF23. Unexpectedly, XLKO mice had markedly higher FGF23 levels than WT mice, with higher blood urea nitrogen and more severe tubulopathy. FGF23 mRNA levels increased substantially in bone and bone marrow in both genotypes; however, the levels in bone were markedly higher than in bone marrow. In XLKO mice, a positive linear correlation was observed between plasma FGF23 and bone, but not bone marrow, FGF23 mRNA levels, suggesting that bone, rather than bone marrow, is an important contributor to severely elevated FGF23 levels in this model. Upon folic acid injection, a model of acute kidney injury, XLKO and WT mice exhibited similar degrees of tubulopathy; however, plasma phosphate and FGF23 elevations were modestly blunted in XLKO males, but not in females, compared to WT counterparts. Our findings suggest that XLαs ablation does not substantially alter FGF23 production in adult mice but increases susceptibility to adenine-induced kidney injury, causing severe FGF23 elevations in plasma and bone.

An important hormone regulating phosphate and vitamin D homeostasis is fibroblast growth factor 23 (FGF23) (1, 2). Pro-FGF23 is synthesized as a 251 amino acid protein in the bone tissue by osteoblasts and osteocytes and is cleaved intracellularly to form the mature FGF23 protein, which is either secreted intact (iFGF23) or inactivated by another intracellular cleavage step, to form the inactive N- and C-terminal fragments (3). iFGF23 then inhibits the reabsorption of phosphate from the glomerular filtrate and diminishes the synthesis of the bioactive vitamin D metabolite 1,25-dihydroxyvitamin D3 in the renal tubule (4). Dysregulated action of FGF23 is associated with several human genetic and acquired disorders. For example, patients with X-linked hypophosphatemic rickets produce excess levels of FGF23 in bone and thus display renal phosphate wasting (5). Conversely, patients with familial hyperphosphatemic tumoral calcinosis display inactivating mutations in GALNT3 or FGF23 and are therefore deficient for iFGF23 levels (6). In addition, FGF23 levels in the circulation rise dramatically as a result of renal failure (7). Although this elevation is beneficial in early stages of the kidney injury by preventing soft tissue mineralization, the highly increased FGF23 levels in end-stage renal disease have been found to correlate with the occurrence of left ventricular hypertrophy and the increased risk of cardiovascular events (8, 9), thus contributing to the morbidity and mortality in patients with chronic kidney disease (CKD). The mechanisms underlying FGF23 production remain incompletely defined.

Osteocytes and osteoblasts in bone are already known to be important producers of FGF23, providing a substantial portion of serum FGF23 at baseline and in response to a high phosphate challenge (10). Several other tissues have also been shown to contribute to FGF23 production under different conditions, such as in inflammation and in response to kidney injury (1, 11–14). Mouse studies have shown that bone marrow, which contains hematopoietic and mesenchymal stem cells, is capable of producing FGF23 upon blood loss-induced erythropoietin (EPO) expression or in response to EPO injection, with a more significant or comparable contribution than bone, respectively (14–16). A recent study demonstrated that multiple tissues, including bone, contribute to the elevation of FGF23 in a folic acid-induced acute kidney injury (AKI) model but observed no elevation of FGF23 mRNA in bone marrow (17), suggesting that bone marrow is unlikely to be a source of FGF23 in renal failure. Relative contributions of bone and bone marrow to the excess FGF23 levels in the setting of CKD have remained uncertain.

Production of FGF23 is stimulated by a number of exogenous molecules, including 1,25-dihydroxyvitamin D, calcium, and leptin (18–20). Recently, inflammatory mediators, which are systemically increased in CKD, have also come to the focus of attention as stimulators of FGF23 synthesis (11, 21, 22). Another stimulator of osseous FGF23 production is parathyroid hormone (PTH), which acts via the stimulatory G protein Gsα, a ubiquitous signaling protein that mediates the actions of many hormones, autocrine and paracrine factors, and neurotransmitters via generation of 3′,5′-cyclic adenosine 5′-mono-phosphate (cAMP) (23–25). The PTH receptor can also couple to other heterotrimeric G protein alpha-subunits, including the extra-large variant of Gsα (XLαs) (26–28). XLαs can mimic Gsα with respect to cAMP generation and is expressed in bone marrow stromal cells, calvarial preosteoblasts, and osteocytes (28–31). It has been shown that early postnatal mice in which XLαs is ablated show hyperphosphatemia with diminished FGF23 levels in bone and in the circulation (32–34), indicating that XLαs is required for FGF23 production. However, unlike the action of Gsα in this regard, the action of XLαs in adulthood has remained unknown.

In this study, we investigated the role of XLαs in adult mice with respect to FGF23 production by examining XLαs knockout mice (XLKO) and wild-type (WT) littermates. Furthermore, to determine whether XLαs-mediated mechanisms are involved in kidney injury-induced FGF23 production, we employed a model of CKD involving the use of adenine-rich diet, as described (35). This model entails tubulointerstitial nephropathy due to the precipitation of highly insoluble adenine derivative 2,8-dihydroxyadenine in the renal parenchyma (36). While adult XLKO and WT littermates did not show significant differences in baseline plasma FGF23 levels, our findings have indicated that XLαs deficiency increases susceptibility to adenine-induced kidney injury. Moreover, we compared the levels of FGF23 mRNA in bone and bone marrow, revealing that bone, rather than bone marrow, is a substantial source of plasma FGF23 in the adenine-induced kidney injury model. We also employed the folic acid-induced AKI model, in which folic acid causes nephrotoxicity that leads to acute tubular necrosis (37). Folic acid injection resulted in a comparable degree of renal failure in XLKO and WT mice within 24 hours. In both CKD and AKI models, we have not detected differences between the 2 genotypes in the relationship between plasma FGF23 and phosphate levels.

Methods

Mice and experimental design

All animal experiments were conducted in accordance with the accepted standards of the Institutional Animal Care and Use Committee, and the studies were approved by the Massachusetts General Hospital Subcommittee on Research Animal Care.

XLKO mice were generated by disrupting the first exon of XLαs on the paternal allele, as previously described (38), and maintained in the CD1 background. WT and XLKO mice had free access to water and a standard chow diet (0.9% phosphate, 0.6% calcium). At 8 weeks of age, mice of each genotype were divided into 2 groups, of which 1 group was changed to receive an adenine-rich chow diet (0.2% adenine, 0.9% phosphate, 0.6% calcium) for a total period of 6 weeks. The mice and the food were weighed, and venous plasma samples were obtained from the tail vain on a weekly basis. Most male mice under the adenine group lost more than 25% of their weights before the completion of the study, and, therefore, the analyses were performed only on female WT and XLKO mice. An aggravated disease progression of adenine-induced kidney disease in male compared with female rats has been observed previously (39). This finding is in accordance with observations in humans, as men show a faster progression towards end-stage renal disease than women before menopause (40), and might be caused by higher estrogen and ER-α activation in females (39). The mice were weighed weekly and supplemented subcutaneously with lactated ringer solution after 5% body weight loss compared to baseline weight at 0 weeks adenine/control diet to reduce excessive water loss due to polyuria. The injected volume was equivalent to 4% of the current mouse weight and was given in two portions with four hours in between. AKI was induced in 12-week-old male and female XLKO and WT littermates by injecting 240 mg/kg folic acid in sodium bicarbonate subcutaneously, as described (41). Vehicle injection was used as control. Blood samples were taken at baseline and at 2, 8, and 24 hours after injection. Mice were euthanized after 24 hours, and kidneys were morphologically examined for signs of kidney injury.

Histopathology

Kidneys were fixed in 10% paraformaldehyde, embedded in paraffin, and serially sectioned at 5 μm. Hematoxylin and eosin staining was performed using standard protocols. The tissue sectioning and staining were conducted at the Endocrine Unit, Massachusetts General Hospital Center for Skeletal Research.

Measurement of plasma biochemistries

Venous plasma samples were collected from tail vain puncture at different timepoints. Plasma phosphorous and blood urea nitrogen (BUN) levels were determined with colorimetric assays, according to the manufacturer’s protocols (phosphate, Abcam Phosphate Assay Kit Colorimetric; BUN, Stanbio™ BUN Liquid Reagent for Diagnostic Set). Plasma FGF23 levels were quantitated by using the mouse/rat FGF-23 (C-Term) enzyme-linked immunosorbent assay (ELISA) kit or the mouse/rat FGF-23 (Intact) ELISA kit (both from Immutopics/Quidel).

cDNA synthesis and gene expression analysis

Cortical bone and bone marrow were separated by removing the epiphysis and metaphysis of the femur bone and flushing the diaphysis with phosphate-buffered saline. Bone marrow and cortical bone samples were homogenized in Trizol. The total RNA was separated using the ThermoFisher TRIzol™ Reagent Kit and precipitated using the Qiagen RNeasy® Plus Mini Kit. Two micrograms of RNA from each sample was run on a 1.5% Agarose gel to confirm RNA quality. cDNA was synthesized using the New England Biolabs ProtoScript II First strand cDNA synthesis kit according to the manufacturer’s protocol. qRT-PCR analysis on FGF23 was performed using the Applied biosystems TaqMan Fast Advanced Master Mix and predesigned TaqMan gene expression assays Fgf23-FAM, Actb-VIC (β-actin as reference control) and Gapdh-VIC (glyceraldehyde-3-phosphate dehydrogenase as reference control). qRT-PCR analysis on all other genes was performed with specific primers and the PowerUp SYBR Green Master Mix (Applied biosystems) with β-Actin as a reference gene. β-Actin (F: 5′-GATCTGGCACCACACCTTCT-3′; R: 5′-GGGGTGTTGAAGGTCTCAAA-3′), interleukin (IL)-6 (F: 5′-TACCACTTCACAAGTCGGAGGC-3′; R: 5′-CTGCAAGTGCATCATCGTTGTTC-3′), IL-1β (F: 5′-GGAGAACCAAGCAACGACAAAATA-3′; R: 5′-TGGGGAACTCTGCAGACTCAAAC-3′), Egr-1 (F: 5′-AGCGAACAACCCTATGAGCACC-3′; R: 5′-ATGGGAGGCAACCGAGTCGTTT-3′). Ratios of bone to bone marrow FGF23 mRNA levels (relative to β-actin) were calculated for each study mouse by dividing the bone value to bone marrow value at baseline or at 6 weeks of adenine-rich diet.

Statistical analysis

Leven’s F test was first performed to assess equality of variances between groups. If variances were equal, paired or unpaired Student’s t test (2-tailed) was used for determining the statistical significance of the difference between 2 means. Otherwise, Welch’s t test (2-tailed) was used. For multiple comparisons, Bonferroni correction was performed: the P value was multiplied by the number of comparisons. Outliers in each data group were identified by using the Grubbs’s test and excluded. Differences in plasma FGF23 values between different time points and between adenine-fed or folic acid-injected WT and XLKO mice were analyzed by using 2-way repeated measures ANOVA. Pearson correlation coefficient was calculated to determine the degree with which two variables are linearly related. A P < .05 was considered statistically significant and represented as follows: *P < .05, **P < .01, ***P < .001, ****P < .0001. Analyses were performed by using Prism 6 (GraphPad).

Results

XLKO mice develop a more advanced kidney injury upon adenine-rich diet

To understand the role of XLαs in kidney injury-induced FGF23 production we studied the XLαs knockout mice (XLKO). XLαs expression is monoallelic and occurs paternally in nearly all tissues, and XLKO mice carries the targeted gene only on the transcriptionally active paternal allele (38). After confirming the strong reduction of XLαs mRNA levels in bone (Fig. 1A) and bone marrow (Fig. 1B) of 8-week-old XLKO mice, we measured the levels of total FGF23 and iFGF23 in the plasma. At baseline, total FGF23 levels (measured by the C-terminal FGF23 assay) in XLKO mice tended to be lower than the levels in WT littermates (Fig. 1C). While this result was in line with findings from early postnatal XLKO and WT pups (32), no difference was detectable in plasma iFGF23 levels between the 2 genotypes (Fig. 1D). In order to investigate whether XLαs is required for the FGF23 synthesis in CKD, we fed 8-week-old XLKO and WT littermates a diet containing 0.2% adenine, as described previously (35, 42). Many of the male XLKO mice had to be euthanized due to excessive weight loss 2 weeks after the start of the adenine diet (Fig. 1E), and, therefore, we had to limit our data collection to female XLKO and WT littermates. Among female mice, only 1 WT and 1 XLKO had to be excluded from the study due to excessive weight loss (Fig. 1F). As soon as 1 week after introducing the adenine-rich diet, pale spots could be found grossly on the kidney surface (Fig. 2A). After 6 weeks of adenine-rich diet, the whole kidneys appeared pale and showed morphological signs of atrophy with an irregular surface and a reduced size (Fig. 2B). The hematoxylin and eosin staining of renal sections revealed extensive proximal and distal tubular luminal expansion, with more severe appearance of tubulopathy in XLKO than in WT kidneys (Fig. 2C). A significant increase in plasma phosphate levels was observed in both XLKO and WT littermates after 6 weeks of the adenine diet (Fig. 2D). In addition, BUN levels were elevated compared with baseline in both genotypes, and the elevation was significantly more pronounced in XLKO than in WT littermates (Fig. 2E). These findings confirmed the anticipated effect of adenine-rich diet on the kidney and suggested that XLKO mice had more advanced kidney injury than WT littermates. We found a slightly decreased adenine-rich food intake in XLKO mice compared with WT littermates (WT control diet: 0.73 ± 0.1 g; WT adenine diet: 0.78 ± 0.05 g; XLKO control diet: 0.62 ± 0.17 g; XLKO adenine diet: 0.42 ± 0.13 g; values reflect the consumed amount of food per day per gram mouse weight), thus making it unlikely that the more advanced kidney injury in XLKO mice reflected increased consumption of the adenine-rich diet.

Figure 1.

Extra-large Gα subunit (XLαs) and fibroblast growth factor-23 (FGF23) levels in wild-type (WT) and XLαs knockout (XLKO) mice, as well as their weights and survival during adenine-rich diet. Eight-week-old WT and XLKO mice were euthanized and XLαs-mRNA levels were measured in bone (A) and bone marrow tissue (B). Plasma samples from 8 week-old WT and XLKO females were collected and levels of total (C) and intact (D) FGF23 were determined. Unpaired Student’s t-test (two-tailed) was used for statistical comparison. Welch’s test was used for unequal variances. (E) Females and males were weighed on a weekly basis and the weight progression was compared between the genotypes and the diets. Figure (F) shows the survival data of males and females of both genotypes during the adenine-rich diet. (A) WT n = 3, XLKO n = 3; (B) WT n = 3, XLKO n = 3; (C) WT n = 18, XLKO n = 23; (D) WT n = 9, XLKO n = 11; (E) female WT adenine n = 12, control n = 10; females XLKO adenine n = 13, control n = 12; male WT adenine n = 4, control n = 6; male XLKO adenine n = 8, control n = 7; (F) females n = 25, males n = 12. *, P < .05.

Figure 2.

The effect of the adenine-rich diet on kidneys. (A) Kidneys from 3 female wild-type (WT) and 3 female XLαs knockout (XLKO) mice were harvested at 0, 1, or 2 weeks of adenine diet. Pale spots on the kidney surface display an early onset of the kidney damage. Morphological comparison reveals a faster progressing kidney injury in XLKO mice starting at 1 week of adenine diet. At 6 weeks (B), kidneys of adenine-fed mice were shrunk and pale with an irregular surface. (C) Histological analysis revealed an extensive tubular dilation upon six weeks of adenine diet with more pronounced kidney damage in the XLKO mice than WT mice. (For B, C) Six WT mice receiving control diet, 6 WT mice receiving adenine diet, 7 XLKO mice receiving control diet, and 9 XLKO mice receiving adenine diet were compared for histological and morphological analysis. Representative histology sections are shown. (D,E) Plasma phosphate and blood urea nitrogen (BUN) levels in WT and XLKO mice following adenine-rich diet. Plasma samples were collected from WT and XLKO mice before and after 6 weeks of adenine or control diet. Plasma phosphate (D) and BUN (E) levels were obtained. Unpaired Student’s t-test (two-tailed) was used, with or without Welch’s test (for unequal variances), followed by Bonferroni correction of the p values for multiple comparisons. (D) 0 week WT n = 11, XLKO = 11; 6 week WT n = 11, XLKO = 11; (E) 0 week WT n = 14, XLKO = 12; 6 week WT n = 11, XLKO = 11. **, P < .01; ****, P < .0001.

More advanced renal injury in XLKO mice is associated with higher plasma FGF23 levels

To characterize the FGF23 synthesis, we measured plasma total FGF23 levels, finding out that an elevation was detectable as soon as 1 week after introducing the adenine-rich diet in both XLKO (2251 ± 766.7 pg/mL) and WT (994.2 ± 232.4 pg/mL) mice (Fig. 3A). At 6 weeks, total FGF23 levels were markedly elevated, with higher levels in XLKO than WT (104 206 ± 21 741 vs 18 430 ± 3379 pg/mL, P < .0001; Fig. 3A). A mild increase of total FGF23 levels were also observed in control diet-fed animals (Fig. 3B). Plasma iFGF23 also measured higher in both WT and XLKO mice fed the adenine-rich diet compared to control diet for 6 weeks, with significantly higher levels in XLKO than WT mice (Fig. 3C). We then examined the relationship between plasma total FGF23 and phosphate levels. At baseline, no linear correlation was observed in either genotype (Fig. 3D). At 6 weeks, WT mice continued to show no linear correlation between plasma total FGF23 and phosphate, whereas XLKO mice displayed a statistically significant positive linear correlation (Fig. 3E). When all data—baseline and 6 weeks of adenine diet—are analyzed together, total FGF23 appeared to be linearly correlated with plasma phosphate only in XLKO mice (Fig. 3F). Due to the wide variation of FGF23 values between baseline and 6 weeks of adenine diet, we then reperformed the linear regression and correlation analysis after semi-log transformation of the data: logarithm of FGF23 vs. phosphate. This analysis indicated a positive correlation in both WT and XLKO mice, and furthermore, the linear relationship between log(FGF23) and phosphate in WT mice appeared to be comparable to that in XLKO mice according to the statistical analysis of the difference between the slopes (Fig. 3G).

Figure 3.

Time course of plasma FGF23 elevation, and the relationship between FGF23 and plasma phosphate. Eight-week-old wild-type (WT) and XLαs knockout (XLKO) mice were fed with 0.2% adenine-rich or control diet for 6 weeks. Plasma samples were collected at different timepoints. (A) The time course of total FGF23 is shown in WT (blue) and XLKO (red) plasma samples at 0, 1, 2, 4, and 6 weeks of adenine (A) and control diet (B). Two-way ANOVA multiple comparison analysis was used for statistical analysis. (C) iFGF23 levels are shown after 6 weeks of control or adenine diet. Unpaired Student’s t-test (2-tailed) was used for statistical comparison between the genotypes. Linear regression analysis was performed between total plasma FGF23 and phosphate in WT and XLKO mice before (0 wk, D) and after 6 weeks of adenine diet (E). Both timepoints were combined and linear regression analysis was performed between the genotypes before (F) and after semi-log transformation of total plasma FGF23 (G). The Pearson correlation coefficient was calculated. (A) WT adenine n = 8, XLKO adenine n = 10; (B) WT control n = 9, XLKO control n = 12; (C) Control WT n = 10, XLKO = 12; Adenine WT n = 9, XLKO = 11; (D) WT n = 9, XLKO n = 10; (E) WT n = 9, XLKO n = 10; (F, G) WT n = 18, XLKO n = 20. *, P < .05; **, P < .01; ****, P < .0001.

Bone is an important contributor to plasma FGF23 levels in CKD

Both osteoblasts/osteocytes and bone marrow have been described as significant producers of FGF23 (1, 14, 43, 44). In order to characterize the synthesis of FGF23 in more detail, we extracted femur bone and bone marrow samples from XLKO and WT mice after 6 weeks of adenine-rich or control diet. Consistent with the plasma FGF23 levels, qRT-PCR analysis revealed increased FGF23 mRNA levels in femur bone (Fig. 4A) and bone marrow samples (Fig. 4B) of mice fed the adenine-rich diet. The values obtained from the femurs of adenine-fed XLKO mice showed wide variation, and therefore, the difference between these mice and the control diet-fed XLKO mice did not reach statistical significance (Fig. 4A). In contrast, the increases observed in bone marrow were significant in both WT and XLKO littermates (Fig. 4B). These findings were confirmed by using another housekeeping gene as reference: Gapdh (Fig. 4C, D). Notably, these experiments strongly suggested that the levels of FGF23 mRNA are markedly higher in bone than in bone marrow. Calculating the bone-to-bone marrow ratio in individual mice of either genotype, we found that, at baseline, the mean FGF23 mRNA level relative to β-actin was 28.6 ± 8.3 (95% confidence interval: 7.3–49.8) times higher in bone than in bone marrow. Six weeks after the adenine-rich diet, the FGF23 mRNA level continued to be dramatically higher in bone (Fig. 4E), with the mean bone-to-bone marrow ratio being 20.5 ± 7.7 (95% confidence interval: 1.5–39.4). When FGF23 mRNA was quantified relative to Gapdh, the level in bone following the 6-week adenine-rich diet was 31.4 ± 10.5-fold (95% confidence interval: 5.8–56.9) higher than in bone marrow (Fig. 4E). To further investigate the contributions of bone and bone marrow to the plasma FGF23 levels, we analyzed the relationship between plasma FGF23 and tissue FGF23 mRNA levels. A positive correlation was observed in WT animals between plasma FGF23 and both bone (Fig. 5A) and bone marrow FGF23 mRNA levels (Fig. 5D). In contrast, a statistically significant correlation existed in XLKO animals only between plasma FGF23 and bone FGF23 mRNA levels (Fig. 5B and 5E). When the analysis was performed after pooling the values from both genotypes, plasma FGF23 correlated significantly with FGF23 mRNA in bone, but not in bone marrow (Fig. 5C and 5F). However, analyzing the logarithm of plasma FGF23 levels against bone marrow FGF23 mRNA levels, we detected a positive correlation in XLKO mice, as well as in pooled data from both of the genotypes (Fig. 5G and 5H).

Figure 4.

Bone and bone marrow FGF23 mRNA levels in response to adenine-rich diet. 8-week-old wild-type (WT) and, XLαs knockout (XLKO) mice were fed with 0.2% adenine or control diet for 6 weeks. Cortical bone (A,C) and bone marrow (B,D) tissue samples were collected and FGF23 mRNA levels were determined at 6 weeks by using β-actin (A,B) or Gapdh (C,D) as reference control. Unpaired Student’s t-test (2-tailed) was used for statistical analysis. Welch’s test was used for unequal variances. (E) Ratio of bone-to-bone marrow FGF23 mRNA level after 6-weeks of adenine diet, measured relative to either β-actin or GAPDH in individual mice from both WT and XLKO groups. (A) Control WT n = 5, XLKO = 5; Adenine WT n = 4, XLKO = 7; (B) Control WT n = 4, XLKO = 6; Adenine WT n = 5, XLKO = 9; (C) Control WT n = 5, XLKO = 5; Adenine WT n = 4, XLKO = 7; (D) Control WT n = 5, XLKO = 5; Adenine WT n = 5, XLKO = 7; (E) n = 7 for both groups. *, P < .05; **, P < .01.

Figure 5.

The relationship between plasma fibroblast growth factor-23 (FGF23) and bone or bone marrow FGF23 mRNA levels in wild-type (WT) and XLαs knockout (XLKO) mice. Linear regression analysis was performed between plasma FGF23 and tissue FGF23 mRNA expression in bone (A–C) and bone marrow (D–F), determining the Pearson correlation coefficient. Bone marrow FGF23 was compared to log (plasma FGF23) in XLKO (G) and both genotypes (H) performing linear regression analysis and determining the Pearson correlation coefficient. (A) WT n = 9 (Control n = 5, Adenine n = 4), (B) XLKO n = 11 (Control n = 5, Adenine n = 6); (C) WT + XLKO n = 20, (D) WT n = 9 (Control n = 4, Adenine n = 5), (E) XLKO n = 14 (Control n = 6, Adenine n = 8), (F) WT + XLKO n = 23, (G) XLKO n = 14 (Control n = 6, Adenine n = 8), (H) WT + XLKO n = 23.

Local inflammation is unlikely to drive the excess FGF23 production in bone and bone marrow in CKD

Inflammatory mediators have been shown to stimulate FGF23 expression (21, 45, 46). As CKD patients exhibit systemic inflammation, we subsequently addressed whether the observed increase of FGF23 in bone and bone marrow samples reflect increased inflammation in these tissues. Analysis of inflammation markers isolated after 6 weeks of adenine-rich diet revealed a tendency of IL-1β mRNA levels to rise in XLKO (bone: 1.7 ± 0.29-fold; P = .10; bone marrow: 1.84 ± 0.3-fold; P = .06) but not in WT mice compared to control diet (Fig. 6A and 6B). The mRNA levels of IL-6, which has been reported as an essential mediator of adenine-rich diet-induced FGF23 elevation (21), were modestly diminished in both bone and bone marrow of adenine-fed WT mice compared to control diet-fed WT mice (Fig. 6A and 6B). However, no significant differences were detected in IL-6 mRNA levels in XLKO bone and bone marrow samples between control and adenine-rich diet (Fig. 6A and 6B). In contrast, both IL-1β and IL-6 and mRNA levels were markedly elevated in whole kidneys from WT and XLKO mice (Fig. 6C), verifying that the adenine-induced kidney injury was associated with increased renal inflammation.

Figure 6.

Markers of inflammation in bone, bone marrow and kidney tissue following 6 weeks of adenine-rich diet. Eight-week-old wild-type (WT) and XLαs knockout (XLKO) mice were fed with 0.2% adenine (gray bars) or control (black bars) diet for 6 weeks. Cortical bone (A), bone marrow (B) and kidney samples (C) were collected and interleukin (IL)-1β and IL-6 mRNA levels were determined using β-actin as reference control. Unpaired Student’s t-test (two-tailed) was used for statistical comparisons. Welch’s test was used for unequal variances. (A) Control WT n = 5, XLKO = 5; Adenine WT n = 5, XLKO = 7; (B) Control WT n = 5, XLKO = 6; Adenine WT n = 6, XLKO = 9; (C) Control WT n = 10, XLKO = 12; Adenine WT n = 9, XLKO = 11. *, P < .05; ****, P < .0001.

Renal induction of FGF23 expression as a result of adenine-rich diet

It has recently been shown that FGF23 expression is induced in kidney tissue upon kidney injury (12, 13). We thus sought to determine the FGF23 expression in WT and XLKO kidneys. While no FGF23 mRNA could be detected in whole kidneys of mice receiving the control diet, FGF23 mRNA was readily detectable in this tissue 1 week after introducing the adenine-diet in both genotypes and throughout the 6-week experimental period (Fig. 7A, B). The actions of FGF23 in the kidneys are mediated by the activation of different pathways. One of these pathways initiates the suppression of renal Na/Pi cotransporters by increasing the renal expression of Egr1 through activation of the ERK1/2 signaling pathway (47). qRT-PCR experiments revealed an enhancement of renal Egr1 expression as soon as one week after introducing the adenine-rich diet in both WT and XLKO mice (Fig. 7C, D). Egr1 mRNA levels in kidney correlated positively with plasma FGF23 levels (Fig. 7E), whereas no correlation was detected between renal Egr1 and FGF23 mRNA levels (Fig. 7F).

Figure 7.

Adenine-induced alterations in kidney, including the induction of fibroblast growth factor-23 (FGF23) mRNA expression. Eight-week-old wild-type (WT) and XLαs knockout (XLKO) mice were fed with 0.2% adenine (gray bars) or control (black bars) diet and kidney samples were collected after 1, 2, or 6 weeks. FGF23 (A, B) and Egr1 (C, D) mRNA expression levels were determined using β-actin as reference control. ND = none detected. Linear relationship between kidney Egr1 mRNA and total plasma FGF23 levels (E) or kidney FGF23 mRNA expression (F) was examined by correlation analysis using the Pearson correlation coefficient. Unpaired Student’s t-test (2-tailed) was used for statistical comparisons. Welch’s test was used for unequal variances. (A) WT: 1 week adenine n = 3; 2 week adenine n = 3; 6 week adenine n = 10; (B) XLKO: 1 week adenine n = 3; 2 week adenine n = 3; 6 week adenine n = 10; (C) WT: 1 week adenine/control n = 3; 2 week adenine/control n = 3; 6 week adenine/control n = 9; (D) XLKO: 1 week adenine/control n = 3; 2 week adenine/control n = 3; 6 week adenine n = 9; 6 week control n = 11; (E) n = 17 (WT n = 8, XLKO n = 9); (F) n = 17 (WT n = 8, XLKO n = 9). *, P < .05; **, P < .01; ***, P < .001; ****, P < .0001.

Folic acid induces comparable degrees of AKI in XLKO and WT mice with no apparent induction of renal FGF23 expression

To understand whether the increased susceptibility of XLKO mice to renal injury is specific to adenine-rich diet, we examined the effects of folic acid, an agent that leads to acute kidney injury. Within 24 hours after folic acid injection, plasma FGF23 levels rise markedly (41), and FGF23 mRNA levels increase in multiple tissues, including bone but not bone marrow (17). As expected, we observed a gradual increase of BUN in folic acid-injected female and male mice, and the levels were comparable at each time point between WT and XLKO (Fig. 8B and 8E), suggesting a similar degree of renal injury in both genotypes. Plasma phosphate also increased similarly in WT and XLKO mice, with modestly lower levels in XLKO males than WT males 24 hours after the injection (Fig. 8A and 8D). Plasma total FGF23 also rose gradually without significant differences between genotypes; however, male XLKO mice showed significantly lower FGF23 levels than WT males at 24 hours (Fig. 8C and 8F). Linear regression analysis of total FGF23 and phosphate levels did not reveal any significant differences between WT and XLKO mice, showing positive linear correlations (Fig. 9A and 9C). However, on semi-log transformation of data, the slope of the regression line was steeper in XLKO males, but not in females, than their WT counterparts (Fig. 9B and 9D), suggesting that phosphate might have a greater magnitude of effect on total FGF23 levels in XLKO males in the setting of acute kidney injury.

Figure 8.

Plasma phosphate, blood urea nitrogen (BUN), and plasma fibroblast growth factor-23 (FGF23) dynamics after folic acid-induced acute kidney injury. Twelve-week old WT and XLKO males and females were injected subcutaneously with folic acid and euthanized after 24 hours. Blood samples were collected 0, 2, 8, and 24 hours after folic acid injection and levels of plasma phosphate (A, D), BUN (B, E) and FGF23 (C, F) were determined. Two-way ANOVA multiple comparison analysis was used for statistical analysis. (A) WT n = 10, XLKO n = 10; (B) WT n = 10, XLKO n = 10; (C) WT n = 9, XLKO n = 10; (D) WT n = 11, XLKO n = 8; (E) WT n = 11, XLKO n = 8; (F) WT n = 11, XLKO n = 7. *, P < .05; ****, P < .0001.

Figure 9.

The relationship between plasma fibroblast growth factor-23 (FGF23) and phosphate levels after folic acid-induced acute kidney injury. Twelve-week-old wild-type (WT) and XLαs knockout (XLKO) males and females were injected subcutaneously with folic acid and euthanized after 24 hours. Blood samples were collected 0, 2, 8, and 24 hours after folic acid injection. Levels of plasma FGF23 and phosphate were obtained. Data points of all four timepoints were pooled and linear regression analysis was performed determining the Pearson correlation coefficient (A,C). (B,D) The semilogarithmic graphs of data in A and C. (A,B) WT n = 36 (9 mice), XLKO n = 40 (10 mice); (C,D) WT n = 44 (11 mice), XLKO n = 28 (7 mice).

We also analyzed the kidneys of folic acid-injected mice with respect to the expression levels of FGF23 and inflammation markers. Unlike our findings in adenine-fed mice, we could not detect renal FGF23 expression in control or folic acid-injected mice, regardless of the genotype. IL-6 mRNA levels in kidneys of folic acid-injected mice, on the other hand, were elevated (Fig. 10A and 10B); however, the elevation in WT females was not statistically significant (Fig. 10A). Renal IL-1β mRNA levels rose mildly in WT and XLKO females in response to folic acid injection, but the differences were not statistically significant (Fig. 10C). WT males, but not XLKO males, displayed a slight, statistically significant elevation of renal IL-1β mRNA (Fig. 10D). Renal Egr1 mRNA levels were elevated in folic acid-injected WT and XLKO mice compared with controls (Fig. 10E and 10F); however, the difference in XLKO males did not reach statistical significance.

Figure 10.

Folic acid-induced alterations of inflammation markers and Egr1 expression in kidney. Twelve-week-old wild-type (WT) and XLαs knockout (XLKO) males and females were injected subcutaneously with vehicle (control) or folic acid (FA) and euthanized after 24 hours. Kidney samples were collected and mRNA expression levels of IL-6 (A,D), IL-1β (B,E) and EGR1 (C,F) were determined using β-actin as reference control. Unpaired Student’s t-test (2-tailed) was used for statistical comparisons. Welch’s test was used for unequal variances, followed by Bonferroni correction of p values. (A) WT control n = 8, FA n = 9; XLKO control n = 10, FA n = 9; (B) WT control n = 9, FA n = 10; XLKO control n = 10, FA n = 8; (C) WT control n = 8, FA n = 11; XLKO control n = 9, FA n = 8; (D) WT control n = 9, FA n = 11; XLKO control n = 11, FA n = 9; (E) WT control n = 8, FA n = 10; XLKO control n = 9, FA n = 10; (F) WT control n = 11, FA n = 15; XLKO control n = 12, FA n = 10. *, P < .05; **, P < .01; ****, P < .0001.

Discussion

Based on our recent observations of reduced FGF23 plasma levels in early postnatal XLKO mice, and based on the finding that XLαs is expressed in adult bone, we investigated the role of this protein in renal injury-induced FGF23 production. We used an established model of CKD involving an 0.2% adenine-containing diet. We chose to begin the diet in 8-week-old mice, because the initial study that established this protocol in mice, as well as many pursuant studies, had employed mice at this age and been able to reliably detect increases in FGF23 levels (35). Although mice at this age have typically not reached skeletal maturity, this is unlikely to confound our results, given that the control diet-fed WT and XLKO mice showed only a minor increase in their FGF23 levels over the 6-week treatment period. In addition, we also measured the difference in plasma phosphate levels in control diet-fed mice between 8 weeks and 14 weeks (ie, the end-point of our study), and revealed that, in keeping with findings in pediatric age humans (48), both WT and XLKO mice showed a modest decline in plasma phosphate levels (–0.572 ± 0.14 mM and –0.323 ± 0.1mM in 14-week-old control diet fed females compared with 8-week-old females, respectively). This finding indicates that the adenine diet-induced elevation of plasma phosphate at 6 weeks was in fact more pronounced than that assessed when comparing to the baseline (0 week).

Consistent with the previous literature, male mice fed the adenine-rich diet developed renal failure more severely than female mice and, therefore, had to be excluded from the study due to excessive weight loss before the end of the 6-week study period. Morphological and histological analyses of data from female mice, as well as BUN levels, indicated a more progressive kidney disease in XLKO than in WT littermates. It appears likely that adenine, rather than an intrinsic kidney pathology, is responsible for the enhanced renal injury in XLKO mice, considering that these mice did not display a similarly increased susceptibility to renal injury upon folic acid injection.

When adenine phosphoribosyltransferase, which converts adenine to adenosine monophosphate and inorganic pyrophosphate, is defective or when adenine concentration exceeds enzymatic capacity, adenine becomes increasingly metabolized by xanthine dehydrogenase, which converts adenine into the highly insoluble derivate 2,8-dihydroxyadenine. This molecule precipitates in the renal parenchyma, causing a crystalline nephropathy (36). XLKO phenotype is characterized by poor adaptation to feeding, early postnatal lethality, and defective glucose and energy metabolism. Furthermore, XLKO mice are lean, hypermetabolic, and show increased sympathetic nervous system activity (38, 49). This hypermetabolic state might cause an increase in turnover of adenine by xanthine dehydrogenase, producing higher concentrations of its insoluble derivates and, in turn, leading to a faster progression of kidney disease. Xanthine-dehydrogenase is expressed in liver and pancreas. As XLαs is expressed in WT pancreatic tissue (50), and depletion of XLαs leads to a pancreatic endocrine dysfunction (38), XLαs deficiency in pancreas might be another reason for the dissimilar progression of CKD in WT and XLKO mice. In addition, the ability to accumulate adenosine and hypoxanthine in adipose tissue (51) might act as an additional buffering mechanism for the WT animals, which have larger adipose tissue than their XLKO littermates.

Contrary to previous observations of significantly reduced plasma FGF23 levels in young XLKO pups, we detected only a slight reduction in total FGF23 plasma levels in 8-week-old XLKO mice. It may be that the actions of XLαs in FGF23 synthesis are compensated by other signaling proteins in adults. It is also possible that the systemic alterations in adult XLKO mice may mask or counteract the FGF23-related phenotype resulting from the skeletal XLαs deficiency.

We did not find a significant correlation at baseline between plasma total FGF23 and phosphate levels in WT or XLKO mice. Note that a correlation between plasma FGF23 and phosphate levels at baseline has not been consistently observed, and the data remain conflicting (52–54). The positive correlation between plasma FGF23 and phosphate levels in adenine-fed XLKO mice, as opposed to WT mice, may reflect the more severe renal injury and higher FGF23 levels in XLKO mice. However, on semi-log transformation of the entire data set, we could detect a significant positive correlation in both genotypes, and the relationship between plasma total FGF23 and phosphate did not appear to differ between XLKO and WT. We could not obtain a higher degree of renal injury, which could have perhaps revealed the effect of XLαs deficiency, if any, on the plasma FGF23 response to kidney injury. Nonetheless, XLKO mice proved valuable in our experiments to assess the relative contributions of bone and bone marrow to FGF23 levels by providing a more advanced adenine-induced renal injury model than WT mice and a wide range of plasma FGF23 values with severely elevated levels.

In folic acid injection experiments, we were able to study both males and females and observed a comparable degree of renal failure in XLKO and WT mice. Although the relationship between plasma phosphate and total FGF23 levels were not different between the two genotypes of mice, there was a modest blunting of the FGF23 elevation in XLKO males, but not females, 24 hours after folic acid injection. This finding may reflect the mild blunting of the plasma phosphate elevation in those mice at the same time point. We currently have no explanation for the mildly blunted increase in plasma phosphate and FGF23 levels in folic acid-injected XLKO males. However, although BUN levels were comparable in XLKO and WT males, the renal inflammation markers in XLKO males did not appear to be as robustly increased as in WT males, suggesting a lower degree of renal injury in XLKO than in WT males.

Upon hypoxia-stimulated EPO production, bone marrow has been found to be another important producer of FGF23 in AKI (14, 15, 44). In our adenine-induced CKD model, FGF23 mRNA levels increased in both bone and bone marrow, confirming that FGF23 production is stimulated in the latter, as well as in bone. However, we found FGF23 expression to be especially high in bone compared with bone marrow, both at baseline and 6 weeks after the adenine-rich diet. Although our quantitative RT-PCR analysis was not designed to measure absolute RNA amount, the dramatically higher FGF23 levels in bone than in bone marrow was detected by using two different house-keeping genes as reference control. Moreover, the positive linear correlation in WT mice between plasma FGF23 and bone marrow FGF23 mRNA levels was not detected in XLKO mice, which showed markedly high levels of plasma FGF23 levels upon the adenine-rich diet. The correlation was detected only after semi-log transformation of data. These results may suggest that the exceedingly high levels of plasma FGF23 reflect production in bone rather than in bone marrow. Nevertheless, our conclusion is based on FGF23 mRNA measurements, and we realize that the vast difference between the levels of bone and bone marrow FGF23 mRNA may not necessarily reflect the differences in protein levels. While this possibility is important to consider, the FGF23 mRNA expression changes observed in previous studies were concordant with protein level alterations (14, 15, 44). Moreover, in agreement with our findings, a recent study showed that the serum and bone FGF23 elevation in a similar adenine-induced renal injury model is reduced by ~90% in mice with conditional ablation of FGF23 in mature osteoblast/osteocytes (55).

Inflammation is an important regulator of FGF23 production, and IL-6 has been shown as an important stimulator of FGF23 production in response to adenine-induced nephropathy (11, 21, 22). We found unaltered IL-1β mRNA levels and modestly diminished IL-6 mRNA levels in bone and bone marrow samples from WT mice following 6 weeks of adenine-rich diet. Given that the levels of FGF23 mRNA in these tissues significantly rose during this time, our findings suggest that local inflammation has a minor role in FGF23 production in this model of CKD. In XLKO bone and bone marrow samples, however, the alterations observed in the levels of these inflammatory markers were different, with mildly increased levels of IL-1β mRNA and unaltered levels of IL-6 mRNA. This tendency of increased local inflammation in XLKO bone following the adenine diet may reflect the greater degree of renal injury in those mice than in WT mice. Our findings cannot rule out the possibility that full-blown inflammation in bone and/or bone marrow, which could develop as a result of highly advanced renal injury, contributes to the exceedingly high FGF23 production.

Recently it has been shown that the renal tubular epithelial cells also express FGF23 mRNA and protein in response to injury (12, 13). Corroborating those previous observations, our data also indicate that the ectopic renal FGF23 expression is detectable concurrently with the excess FGF23 levels in plasma, that is, soon after the impairment of renal function by the adenine-rich diet. However, the FGF23 mRNA found in kidney did not correlate with plasma FGF23 levels. This finding suggests that mechanisms governing the induction of FGF23 expression in kidney differ from those underlying the overproduction of this hormone in osseous tissues. Interestingly, renal expression of Egr1, which translates the action of FGF23 (47), correlated with plasma FGF23 levels, but not with renal FGF23 mRNA levels, strongly suggesting that the renal Egr1-response reflects global FGF23 elevation.

Despite evidence of renal injury with elevated BUN and phosphate levels, the folic acid injection did not lead to an induction of FGF23 expression in kidney. It is also possible that a longer duration of injury is required for the induction of renal FGF23 expression or that the latter requires a specific type of kidney injury, such as the one generated by adenine. Our finding, however, contrasts the data in a recent report, in which folic acid injection was stated, but not shown, to induce ectopic renal FGF23 expression (17). The latter study used the same dose and administration route of folic acid but employed a different genetic background (C57Bl/6J), which could perhaps explain the discrepancy. With respect to renal Egr1 mRNA levels in our AKI model, we did not detect an elevation in folic acid-injected XLKO males compared with vehicle-injected XLKO males. This may reflect the blunted increase of plasma FGF23 levels in those mice.

In summary, we examined the role of XLαs in adult mice regarding kidney injury-induced FGF23 production. Our investigations revealed that XLαs has a protective role in adenine-induced nephropathy, and that bone, rather than bone marrow, is a major source of excess FGF23 resulting from adenine-induced renal injury. Unlike the findings in early postnatal mice, our study did not reveal any evidence that XLαs ablation significantly impairs FGF23 production in adult mice.

Glossary

Abbreviations

AKI

acute kidney injury

BUN

blood urea nitrogen

cAMP

3′,5′-cyclic adenosine 5′-mono-phosphate

CKD

chronic kidney disease

EPO

erythropoietin

FGF23

fibroblast growth factor-23

IL

interleukin

PTH

parathyroid hormone

WT

wild type

XLαs

extra-large Gα subunit

XLKO

XLαs knockout

Additional Information

Disclosure Summary: All authors have nothing to disclose.