Role of fibroblast growth factor 23 and klotho cross talk in idiopathic pulmonary fibrosis

Abstract

Idiopathic pulmonary fibrosis (IPF) is a chronic, progressive fibrosing interstitial pneumonia that mainly affects the elderly. Several reports have demonstrated that aging is involved in the underlying pathogenic mechanisms of IPF. α-Klotho (KL) has been well characterized as an “age-suppressing” hormone and can provide protection against cellular senescence and oxidative stress. In this study, KL levels were assessed in human plasma and primary lung fibroblasts from patients with idiopathic pulmonary fibrosis (IPF-FB) and in lung tissue from mice exposed to bleomycin, which showed significant downregulation when compared with controls. Conversely, transgenic mice overexpressing KL were protected against bleomycin-induced lung fibrosis. Treatment of human lung fibroblasts with recombinant KL alone was not sufficient to inhibit transforming growth factor-β (TGF-β)-induced collagen deposition and inflammatory marker expression. Interestingly, fibroblast growth factor 23 (FGF23), a proinflammatory circulating protein for which KL is a coreceptor, was upregulated in IPF and bleomycin lungs. To our surprise, FGF23 and KL coadministration led to a significant reduction in fibrosis and inflammation in IPF-FB; FGF23 administration alone or in combination with KL stimulated KL upregulation. We conclude that in IPF downregulation of KL may contribute to fibrosis and inflammation and FGF23 may act as a compensatory antifibrotic and anti-inflammatory mediator via inhibition of TGF-β signaling. Upon restoration of KL levels, the combination of FGF23 and KL leads to resolution of inflammation and fibrosis. Altogether, these data provide novel insight into the FGF23/KL axis and its antifibrotic/anti-inflammatory properties, which opens new avenues for potential therapies in aging-related diseases like IPF.

INTRODUCTION

Idiopathic pulmonary fibrosis (IPF) is a chronic, progressive fibrosing interstitial pneumonia that is associated with a poor prognosis and affects more than 100,000 older adults in the United States each year (53). Presently, approved medications for IPF are only modestly efficient. Several reports have demonstrated that cell senescence contributes to the underlying pathogenic mechanisms of IPF (69, 70). Interestingly, α-klotho (KL), a single-transmembrane protein that exists in a membrane bound and soluble form, has been well characterized as an “age-suppressing” hormone (32, 34). Mice lacking klotho develop an aging phenotype including a shortened lifespan, atherosclerosis, skin atrophy, osteoporosis, and widened alveolar spaces, which is consistent with pulmonary emphysema (67). Furthermore, it has been shown that KL expression is reduced in airways from COPD patients (18) and that recombinant KL protects the alveolar epithelium against oxidative damage (27, 55). Little is known about KL signaling; however, it has been shown to antagonize transforming growth factor-β (TGF-β) receptor signaling (30). Furthermore, recent data revealed that KL is a physiological regulator of lipid raft structure in the myocardium (8). The function of KL as a coreceptor for fibroblast growth factor (FGF) 23, a proinflammatory circulating hormone, has also been documented (5).

FGF23 has been well characterized as a key regulator in maintaining serum phosphate homeostasis (3, 44, 58). Clinical studies also demonstrate a strong association between increased FGF23 plasma levels and risk of chronic kidney disease (CKD) progression, cardiovascular events, and mortality (21, 61). Signaling occurs via FGF receptors (FGFRs), which are a subfamily of receptor tyrosine kinases (13). FGF23 signaling requires KL as a coreceptor in organs such as the kidney and parathyroid gland, where it regulates phosphate homeostasis (20, 29). In the absence of KL, FGF23 signaling can induce hypertrophic growth in cardiac myocytes and production of inflammatory cytokines in hepatocytes (19, 63). Our previous studies have shown that FGF23 plasma levels are elevated in cystic fibrosis patients and synergistically induce TGF-β mediated airway inflammation, which can be attenuated by KL supplementation (30).

Here, we hypothesize that KL deficiency can also contribute to the development of pulmonary fibrosis. The role of the KL and FGF23 cross talk and their effect on TGF-β dysregulation in the pathophysiology of IPF are investigated and discussed.

METHODS

Study approval.

Plasma samples from IPF patients and controls were obtained from the Pulmonary Clinical Database and Biospecimen Repository at the University of Alabama at Birmingham (UAB) after approval from the UAB Institutional Review Board. Written informed consent was received from each participant before inclusion in the study, and samples were deidentified. Human primary control and IPF fibroblasts were obtained from organ donors whose lungs were rejected for transplant. Institutional Review Board-approved consent for research was obtained by UAB and conformed to the Declaration of Helsinki. All animal protocols were approved by the Institutional Animal Care and Use Committees at The University of Alabama at Birmingham. Both KL^+/− and C57BL/6/kl overexpressed (OE) mouse models were provided by G. King.

ELISA.

Plasma was used to measure intact FGF23 levels and KL levels in pg/ml using commercially available ELISA kits in our human cohort (FGF23, Immutopics, Clemente, CA; and for KL, IBL International, Hamburg, Germany) as previously described (30). For the detection of intact FGF23 in mouse serum, we also used a commercially available ELISA kit from the same vendor according to the vendor’s provided protocol (63).

Cell culture and recombinant proteins.

Primary lung fibroblasts were isolated from lung explants of patients undergoing lung transplantation with IPF or failed donors as previously described (23). Since we experienced heterogeneity of the different cultures, we were not able to show clear differences in collagen content between IPF or control lungs. Furthermore, IPF fibroblasts had the tendency to become senescent and did not grow well. To increase our experiment numbers, we also included IMR-90 cells (Coriell, Camden, NJ), which were used mainly for the cell signaling experiments as described previously (10, 23). Primary murine lung fibroblasts were isolated from lungs of wild-type (WT) and klotho-overexpressing mice as described previously (23, 72).

Human recombinant soluble klotho was purchased from PeproTech (Rocky Hill, NJ) and diluted in sterile filtered phosphate-buffered saline (PBS) containing 0.15% bovine serum albumin (BSA) at a stock concentration of 100 μg/ml, which was stored at −80°C for no longer than 3 months. Human recombinant FGF23 was purchased from R&D Systems (Minneapolis, MN) and PeproTech as well and reconstituted at 20 μg/40 μl in sterile water and then brought to 20 μg/ml in sterile PBS containing 0.1% BSA as recommended by the manufacturer. Control cells were incubated with PBS/0.1% BSA only.

Bleomycin lung injury model.

Six-month-old S129J/Kl^+/+, S129J/Kl^+/−, and C57BL/6/Kl WT mice and C57BL/6/Kl OE mice were anesthetized with inhaled isoflurane (2% vol/vol). Bleomycin (1.25 U/kg) or saline (control) was administered oropharyngeally as described previously (23). Mice were euthanized by CO₂ inhalation at different time points as described in this article, and lung tissues were collected. All procedures involving animals were approved by the Institutional Animal Care and Use Committees at The University of Alabama at Birmingham.

Analysis of lung function in mice.

Three weeks after bleomycin exposure, mice were anesthetized with ketamine/zylazine and pulmonary function was evaluated on a flexiVent as previously described (1, 47), using tracheal insertion of a 18-gauge Angiocath, which was fixed with a ligature of 3–0 silk. Measurements made included total resistance (R; which contains Rn, G, and chest wall resistance, which in the mouse is essentially 0) and compliance (C).

Bronchoalveolar lavage in mice.

Lung lavage was obtained following established protocols of the laboratory (17). Briefly, a tracheal cannula was inserted and the bronchoalveolar lavage (BAL) procedure was performed under direct visualization of lung distension (maximum of 2 ml), as previously described (73, 74). Cells were pelleted by centrifugation (500–1,100 g for 5 min at 4°C) and resuspended in 100–150 μl phosphate-buffered saline for total cell count determination. Cytospin preparations were stained with modified Wright-Giemsa staining for differential cell counts.

IL-8 and IL-6 ELISA and mRNA assessment.

Gene expression was performed by qPCR using Taqman probes (Life Technologies/Applied Biosystems, Carlsbad, CA) with the following probes used: Hs00934627_m1 and Mm00502002_m1 for KL, Hs00174103_m1 for IL-8, Mm00446190_m1 for IL-6, and Hs02758991_g1 for GAPDH.

Western immunoblotting and immunoprecipitation studies.

Cell lysates were prepared in RIPA buffer with Halt protease and phosphatase inhibitor cocktail (Thermo Scientific, Waltham, MA), and Western immunoblotting was performed as previously described (24). Briefly, SDS-PAGE separation of proteins in cell or tissue homogenates (20–50 μg) was performed, and proteins were transferred to polyvinylidene difluoride (PVDF) membranes.

Blots were incubated overnight at 4°C with the respective primary antibodies (1:1,000), washed three times with Tris-buffered saline, and incubated with secondary antibody (1:1,000) as previously described (30, 31). The same blots were reprobed for β-actin (primary antibody 1:4,000; secondary antibody; 1:1,000) to compare protein loading. Blots were developed using species-specific horseradish peroxidase (HRP)-conjugated secondary antibodies and ECL reagents (GE Healthcare Life Sciences, Piscataway, NJ) on an Amersham Biosciences 600 Imager (GE Healthcare). Exposures were 8–15 s for β-actin and 30 s to 10 min for all other primary antibodies. Densitometric analyses were performed using ImageJ (59). For the detection of murine klotho in plasma and total lung lysates, a recently published antibody was used for immunoprecipitation as described previously (41). Antibodies were validated by using TGF-β as a control stimulus for induction of α-smooth muscle actin (α-SMA), COL1A1 and pSMAD3 showing upregulation of the protein band at the expected molecular size. FGF23 (MBS854462; MyBioSource; 1:1,000) has been commonly used to detect endogenous levels of total FGF23 protein and was validated by expected size on immunoblots, reproducibility of the results with a second FGF23 antibody (Immutopics), and control stimulus (low iron) in osteocytes, which are known FGF23-producing cells (9, 30). The klotho antibody was validated using tissue from klotho-deficient mice (41).

Hydroxyproline assay.

Mice were euthanized 21 days after bleomycin injection, and lung tissue was retrieved from each mouse. Lung tissues were dried at 70°C in an oven for 48 h and then hydrolyzed with 6 N HCl at 95°C for 20 h. The hydroxyproline content per lung was assessed as previously described (35).

Lung histology and immunohistochemical staining.

Mice were euthanized and perfused via the right ventricle with 3 ml of normal saline, and inflated with 1 ml of 10% neutral buffered formalin and fixed overnight in formalin, and then dehydrated in 70% ethanol. With the use of standard procedures, lungs were processed and embedded in paraffin. Sections were cut (3–5 mm), mounted on slides, and stained with hematoxylin and eosin or Masson’s trichrome blue for collagen as previously described, which was done by the UAB Comparative Pathology Laboratory (22). Slides were also immunostained with anti-FGF23 according to the following protocol: Antigen retrieval was performed using Tris-EDTA buffer and a pressure cooker retrieval at 3% H₂O₂ at room temperature was done for 20 min to block the endogenous hydrogen peroxidase; slides were incubated in 3% normal goat serum at room temperature for 30 min to block the nonspecific binding from the secondary antibody. Then, primary antibody (FGF23; MBS854462; MyBiosource; rabbit polyclonal) was used to incubate overnight in 4°C (for human use 1:300, for mouse use 1:150). Slides were incubated in secondary antibody (goat anti-rabbit conjugated with HRP; ab6721; Abcam) at room temperature for 45 min (for human use 1:800, for mouse use 1:200) and after washes developed with DAB chromogen and counterstained with hematoxylin.

Statistics.

Data were analyzed with Prism5 (GraphPad Software, La Jolla, CA) and are shown as means ± SD using Student's t-test and analysis of variance or Kruskal-Wallis test with appropriate post tests for at least three independent experiments. Significance was accepted at P < 0.05. For correlation analysis, the Spearman correlation coefficient was used.

RESULTS

Soluble plasma KL levels are decreased in individuals with IPF.

We assessed soluble KL levels in plasma samples from 46 individuals, including 23 individuals with IPF and 23 controls, without history of lung disease. IPF was diagnosed by expert clinicians from the Interstitial Lung Disease Center at the University of Alabama according to current consensus criteria (53). Clinical characteristics, including age, sex, and race are shown in Table 1. Plasma KL levels were significantly decreased in the IPF patient cohort when compared with the control cohort (Fig. 1A and Table 1). Parameters to assess baseline renal function [blood urea nitrogen (BUN) and creatinine] and evidence for uncontrolled diabetes (random blood glucose) were also included, which showed no difference between both groups except creatinine levels were higher in the IPF group (Table 1). Since KL is inversely related to age as observed in our study (Fig. 1B, Pearson correlation coefficient r = −0.5068 with P = 0.0006) and by others (76), samples were age-matched from both cohorts, which still showed a significant reduction of KL in the IPF cohort compared with the controls (P = 0.014; Fig. 1C). Furthermore, we show that there was a significant inverse correlation between plasma KL levels and creatinine levels in the IPF cohort (Fig. 1D, Pearson correlation coefficient r = −0.4533 with P = 0.0341). We did not see any correlation among KL plasma levels and forced expiratory volume in 1 s (FEV1) or forced vital capacity (FVC; Fig. 1, E and F). Additionally, assessment of KL mRNA expression in primary human interstitial lung fibroblasts from IPF lungs (6, 38) demonstrated a marked decrease in KL levels, when compared with control cells (Fig. 1G). In summary, we demonstrate that KL is downregulated in lung fibroblasts and serum from individuals with IPF and that there is a correlation with age and renal function; however, KL downregulation does not correlate with lung function in this cohort.

Table 1.

Patient characteristics

	Control (23)	IPF (23)	P Value
Age	60.6 ± 8.02	71.6 ± 5.59	P < 0.001
Sex
Female	20 (87.0%)	3 (13.0%)	P < 0.001
Male	3 (13.0%)	20 (87.0%)	P < 0.001
Race
Caucasian	6 (26.1%)	21 (91.3%)
African American	17 (73.9%)	2 (8.70%)
Lung function
FVC%	93 ± 21	61.5 ± 15.9	P < 0.001
FEV1%	91.3 ± 18.9	68.7 ± 20	P < 0.005
Renal function
Creatinine, mg/dl	0.8 ± 0.2	1.1 ± 0.4	P = 0.04
BUN, mg/dl	14.8 ± 8.2	17.9 ± 9.2	P = 0.4
Glucose, mg/dl	110.5 ± 36.5	111.0 ± 24.1	P = 0.96

Values are means ± SD. Patient characteristics of our study population, divided into patients with and without idiopathic pulmonary fibrosis (IPF), including age, race, sex, spirometry data, renal function, and random glucose levels. FVC%, percent forced vital capacity; FEV1%, percent forced expiratory volume in 1 s; BUN, blood urea nitrogen.

Fig. 1.

Klotho (KL) levels are decreased in pulmonary fibrosis. A: dot blot showing soluble KL levels in plasma from controls and idiopathic pulmonary fibrosis (IPF) patients. B: correlation analysis between KL levels and age of the whole cohort (control and IPF patients) using the Pearson correlation coefficient. C: subgroup analysis of KL plasma levels from control individuals and age-matched IPF patients and patients excluded with elevated creatinine levels. D: correlation analysis between KL levels and serum creatinine of the IPF cohort. E and F: correlation analysis between KL levels and percent forced vital capacity (FVC%) predicted (E) and percent forced expiratory volume in 1 s (FEV1%) predicted (F) of the IPF cohort. G: KL mRNA fold change in IPF fibroblasts, when compared with those from control donors (3 different donors per group). Statistical analysis was done using ANOVA or Student’s t-test showing means ± SD with *P < 0.05 and ***P < 0.001.

The effect of KL overexpression on lung function and airway inflammation.

To determine whether KL can prevent or slow fibrosis, we used 6- to 8-mo-old C57BL/6/Kl wild-type (WT) mice and C57BL/6/Kl mice that ectopically overexpressed KL (OE) (39) with bleomycin to induce lung fibrosis. We used a previously published klotho antibody (41) and could show a significant upregulation of the 110-kDa band in the serum of klotho-overexpressing mice (Fig. 2A), which is consistent with previous reports (34). We could also detect klotho transcript expression in primary lung fibroblasts isolated from these mice (Fig. 2B). Furthermore, bronchoalveolar lavage fluid obtained from both mouse groups did not show any difference in total cell count and monocyte/macrophage count (Fig. 2C). The body weights of all groups including the mice before and after exposure to bleomycin were similar (Fig. 2D, top). FGF23 serum levels were also indifferent among all mouse groups except for a significant increase in the bleomycin-treated Kl OE mice (Fig. 2D, bottom). Serum creatinine and phosphorous levels showed no difference among groups (Fig. 2E, top and bottom). Lung function was assessed by using flexiVent and indicated that there was no change in baseline static resistance (Fig. 2F, top), but a significant decrease in static compliance solely in the wild-type mice, after treatment with bleomycin (Fig. 2F, bottom). Altogether, these data indicate that klotho-overexpressing mice are not phenotypically different in lung and renal function, when compared with their WT littermates.

Fig. 2.

Characterization of klotho-overexpressing mice (KL OE) mice. A: immunoprecipitation of serum and total lung lysate from wild-type (WT) and KL OE mice using a klotho antibody. B: analysis of klotho transcript levels from lung fibroblasts from WT and klotho-overexpressing (OE) mice. C: dot blots indicating total cell count and macrophage/monocyte cell count from both mouse groups. BALF, bronchoalveolar lavage fluid. D: bar graphs showing pre- and postsaline/bleomycin (3 wk exposure) body weights and fibroblast growth factor 23 (FGF23) serum levels of these mouse groups. E: assessment of renal function by serum creatinine and phosphate levels in mice treated with saline or bleomycin. F: dot blot showing total lung resistance (Rrs) and compliance (Crs) of these mice. All n = 3–6 mice per group showing means ± SD; *P < 0.05 and ***P < 0.005.

KL expression was downregulated in a murine lung fibrosis model.

Lung tissue was obtained at different time points from 3-mo-old mice after oropharyngeal instillation of bleomycin as described previously (22). KL mRNA levels were significantly downregulated starting at day 3, and this reduction was sustained until day 28. On day 56, KL levels were partially restored to baseline levels, which coincided with resolution of fibrosis in this fibrosis mouse model (Fig. 3A) (22).

Fig. 3.

Klotho (KL) overexpression (OE) ameliorates bleomycin-induced lung injury. A: bar graphs indicating KL mRNA levels from mice (3–5 per group) that were injected with bleomycin via oropharyngeal route and harvested at the indicated time points. B: bar graphs indicating hydroxyproline fold-change/lung, compared with the PBS treated wild-type (WT) mice, with black bars representing WT mice and gray bars representing the KL OE mice. C: trichrome/hematoxylin staining of lung tissue slides from vehicle (PBS)-treated WT and KL-overexpressing (OE) mice, harvested after 21 days (×4 and ×20 magnification). D: trichrome/hematoxylin staining of same groups after bleomycin injection (2 separate experiments from 4 to 6 lungs per group). All bar graphs are means ± SD. *P < 0.05, **P < 0.01, and ***P < 0.005.

Overexpression of KL in vivo leads to attenuation of bleomycin-induced lung fibrosis.

We assessed the Kl overexpressing mice for collagen deposition using the method of quantitation of total lung hydroxyproline (22). Total hydroxyproline content per mouse lung in the WT bleomycin-treated lungs showed a significant increase when compared with saline treated controls. This increase was attenuated in the lungs of Kl OE mice (Fig. 3B). Immunohistochemistry using a combination of hematoxylin and trichrome stain showed patchy fibrotic areas in the bleomycin treated WT mouse lungs, which were reduced in the Kl OE mouse lungs (representative images shown in Fig. 3, C and D, ×4 and ×20). Altogether, these in vivo data indicate that overexpression of KL protects against bleomycin-induced lung fibrosis.

Recombinant KL does not inhibit the TGF-β-induced increase of smooth muscle actin or COL1A1 deposition and inflammation in human lung fibroblasts.

Since KL has been shown to inhibit TGF-β-induced renal fibrosis (12), the effect of KL on TGF-β-induced COL1A1 or α-smooth muscle actin (α-SMA) deposition and inflammation was assessed. Lung fibroblasts were stimulated with TGF-β in the presence and absence of KL preincubation. Our results were consistent with previous reports showing that TGF-β increased COL1A1 and α-smooth muscle actin expression (11) but the effects of TGF-β were not inhibited by KL in human lung fibroblasts (Fig. 4, A and B).

Fig. 4.

Effect of klotho (KL) on fibroblast inflammatory markers and matrix deposition. A: representative immunoblot (IB) of COL1A1, α-smooth muscle actin, and β-actin from primary human fibroblasts stimulated with transforming growth factor-β (TGF-β; 5 ng/ml), KL (100 ng/ml), or both for 48 h. B: densitometric analysis of the COL1A1/ACTIN ratio summarizing 4 different experiments. C: bar graphs indicating mRNA fold change of IL-1β, IL-6, and IL-8 levels in human fibroblasts with and without stimulation with human recombinant KL for 24 h. D and E: IL-6 (D) and IL-8 (E) mRNA fold-change in these cultures, when stimulated with TGF-β ± KL (3 different experiments with 6 different donors). All bar graphs are means ± SD. *P < 0.05 and ***P < 0.001.

Stimulation with recombinant KL alone led to a decrease in IL-6 and IL-8 mRNA levels (Fig. 4C), but TGF-β treatment led to a significant increase of IL-6 and IL-8 expression, which was not inhibited by KL; IL-6 expression was further increased when cells were cotreated with TGF-β (Fig. 4, D and E). These findings indicate that administration of recombinant KL alone is not sufficient to inhibit TGF-β-induced COL1A1 and α-SMA deposition and local inflammation in primary lung fibroblasts cultures.

FGF23 is elevated in IPF patients.

Among its physiological functions, KL has been well characterized as a coreceptor for FGF23 (20, 52). Interestingly, it has been shown that KL knockout mice have increased FGF23 expression (30, 63). Using the same patient cohort as analyzed above (Table 1), we showed that FGF23 plasma levels were elevated in IPF patients when compared with control patients (Fig. 5A). There was no significant correlation between FGF23 and lung function (FVC%, and FEV1% of these patients (Fig. 5, B and C). However, lung sections from control patients (Fig. 5D) and patients with IPF (Fig. 5E), stained for FGF23 expression, displayed an increase in the bronchial epithelium distribution as well as parenchymal staining of FGF23 in fibrotic areas of IPF patient lungs compared with control lungs.

Fig. 5.

Fibroblast growth factor 23 (FGF23) levels are increased in pulmonary fibrosis. A: dot plot showing plasma FGF23 levels from control and idiopathic pulmonary fibrosis (IPF) patients (same cohort as described in Fig. 1). *P < 0.05. B and C: correlation analysis between FGF23 levels and percent forced vital capacity (FVC%) predicted (B) and percent forced expiratory volume in 1 s (FEV1%) predicted (C) of the IPF cohort using the Pearson correlation coefficient. D and E: negative control and anti-FGF23 staining (×4 and ×20 magnifications) from human lung tissue sections obtained from control (D) and IPF (E) patients.

FGF23 is increased in total lung tissue and lung fibroblasts from bleomycin-treated mice.

FGF23 expression was also assessed in total lung tissue from mice treated with bleomycin. To determine the localization of FGF23 expression in the lung, lung tissue sections from these mice, euthanized on day 28, were stained with hematoxylin and anti-FGF23 (Fig. 6A, top). Both the bronchial epithelium and the fibrotic parenchyma stained positive for FGF23 expression. Furthermore, there was a significant increase of FGF23 mRNA levels in these animals starting at day 3 until day 28 (Fig. 6A). On day 56, when pulmonary fibrosis resolves in this animal model (36, 44a), FGF23 levels were reduced, comparable to control levels.

Fig. 6.

The effect of fibroblast growth factor 23 (FGF23) and -klotho (KL) on markers of fibrosis in murine fibroblasts. A, top: hematoxylin and anti-FGF23 staining of murine tissue sections, harvested after 28 days. A, bottom: bar graphs indicating FGF23 mRNA levels in total mouse lung tissue at different time points following bleomycin lung injury. B: bar graphs indicating KL transcript levels in total lung tissue and primary isolated lung fibroblasts from mouse lungs and representative immunoblots (IB) of FGF23 and β-actin protein levels in primary mouse lung fibroblasts, exposed to FGF23 (40 ng/ml), transforming growth factor-β (TGF-β; 5 ng/ml), or both for 24 h. (All n = 3 independent experiments from 3 different lungs showing means ± SD with *P < 0.05 and **P < 0.01.)

To further decipher a source of FGF23, we assessed FGF23 transcript levels in primary mouse fibroblasts isolated from these lungs, which showed a significant increase when compared with total lung levels (Fig. 6B, top). When these fibroblasts were treated with TGF-β or TGF-β and FGF23, there was a significant increase in FGF23 protein levels (Fig. 6B, bottom). These data combined suggest that lung fibroblasts express FGF23 that is augmented following the addition of TGF-β or TGF-β and FGF23.

Mice deficient in KL are more susceptible to lung fibrosis.

We have shown that overexpression of KL is protective against bleomycin-induced lung fibrosis. Furthermore, we wanted to assess whether deficiency in KL leads to disease susceptibility. One caveat to the Kl knockout mouse model is that, due to the aging phenotype, these mice die within 8–12 wk (32) and due to their frailty, they do not tolerate bleomycin injections. Therefore, we used Kl haplodeficient (Kl^+/−) mice, which are known to be more susceptible to lung injury (27), and exposed them to bleomycin for a total of 3 wk. As described previously, these mice develop normally and have insignificant changes in serum phosphate levels and kidney function as previously described (14, 68). Kl^+/+ developed the bleomycin-induced patchy fibrosis (Fig. 7A), which was exaggerated in the Kl^+/− mice (Fig. 7B). Interestingly, there was an increase in FGF23 levels after bleomycin injection in Kl^+/+ mice that was further augmented in the Kl^+/− mice (Fig. 7C). Similarly, the Kl^+/− mice also had an increased hydroxyproline content/lung when compared with their Kl^+/+ littermates (Fig. 7D). These data suggest that partial loss of KL results in increased FGF23 levels and enhanced collagen deposition indicative of increased fibrotic lung disease susceptibility.

Fig. 7.

Bleomycin-induced fibrosis is exaggerated in the klotho-haplodeficient (Kl^+/−) mouse model. A: trichrome staining of lung tissue slides from vehicle (PBS) treated wild type (Kl^+/+) and Kl^+/− mice, harvested after 21 days (×4 and ×20 magnification). B: trichrome staining of same groups after bleomycin injection. C and D: dot blots showing serum fibroblast growth factor 23 (FGF23) levels (C) and hydroxyproline content/lung (D) in Kl^+/+and Kl^+/− mice, injected with either vehicle (saline) or bleomycin (n = 3–6 mice for each group). All bar graphs are means ± SD. *P < 0.05 and ***P < 0.005.

Coadministration of KL and FGF23 regulates collagen and Kl expression in lung fibroblasts.

Human lung fibroblasts were stimulated with FGF23 ± KL to determine the effect on collagen synthesis. FGF23 led to a slight decrease in COL1A1 levels, but coadministration of Kl and FGF23 caused a more dramatic decrease (Fig. 8A). Furthermore, FGF23 treatment led to increased KL levels in fibroblast cultures; however, this was not significant. Interestingly, the combined stimulation of FGF23 and KL led to a significant upregulation of KL mRNA levels in these lung fibroblasts (Fig. 8A, bottom). Collectively, these data suggest that the combination of KL/FGF23 reduces collagen expression in primary lung fibroblast cultures possibly via partial restoration of KL levels.

Fig. 8.

Fibroblast growth factor 23 (FGF23) and klotho (KL) supplementation attenuates transforming growth factor-β (TGF-β)-induced COL1A1 deposition and IL-6 and IL-8 expression in human lung fibroblasts. A, top: a representative immunoblot of COL1A1 and β-actin from primary human lung fibroblasts after stimulation with FGF23 (25 ng/ml) ± KL (100 ng/ml) for 48 h and densitometry analysis of the COL1A1/ACTIN ratio. A, bottom: bar graphs showing KL mRNA levels in human fibroblasts after stimulation with KL, FGF23 or both for 48 h. B: representative immunoblots of human lung fibroblasts, prestimulated with either different concentrations of KL alone (10, 100, and 1,000 ng/ml) or KL combined with FGF23 and subsequent stimulation with TGF-β for 48 h. C and D: summarized densitometry analysis of the COL1A1/ACTIN fold change (C; for samples where highest KL concentration was used) and quantification of COL1A1 mRNA changes (D) assessed by quantitative real-time PCR. E and F: IL-6 (E) and IL-8 (F) mRNA fold change in human fibroblasts after stimulation with TGF-β + FGF ± KL. (All n = 3 independent experiments from 3 different lungs showing means ± SD with *P < 0.05, **P < 0.01, ***P 0.005.)

KL and FGF23 combined ameliorate TGF-β-induced fibrotic and inflammatory markers in human lung fibroblasts.

To determine the impact of KL /FGF23 combined on TGF-β-induced fibrosis and inflammation, primary human lung fibroblasts were stimulated with TGF-β for 24 h ± preincubation with KL and/or FGF23. As determined by Western blot analysis, administration of KL did not significantly block TGFβ-induced collagen expression at any concentration tested, which was similar to, when FGF23 alone was coadministered with TGF-β (Fig. 8B). Surprisingly, the combination of KL/FGF23 led to a dose-dependent downregulation of COL1A1 (Fig. 8, B and C, densitometric analysis, and Fig. 8D, qRT-PCR) with concomitant downregulation of phospho-SMAD3 and total SMAD3 level expression. Furthermore, TGF-β and FGF23 coadministration resulted in a significant increase in IL-6 and IL-8 levels that was similar to TGF-β alone indicating TGF-β was responsible for the upregulation. Following preincubation with KL and FGF23, the upregulation of IL-6 and IL-8 mediated by TGF-β was partially blocked (Fig. 8, E and F). These data collectively indicate that KL/FGF23 may work synergistically to protect against fibrosis and inflammation in fibrotic lung diseases such as IPF.

To further support these findings, we isolated primary murine lung fibroblasts from WT and Kl OE mice and treated these with TGF-β, FGF23, or both. Our results showed that Kl OE lung fibroblasts have significantly increased klotho levels, when compared with lung fibroblasts from WT mice. Furthermore, TGF-β in combination with FGF23 led to a significant decrease of klotho mRNA levels (Fig. 9A). After stimulation with TGF-β ± FGF23, there was a significant increase in COL1A1, which was associated with an increase in SMAD3 phosphorylation (Fig. 9B). In comparison to lung fibroblast cultures from WT mice, Kl OE cultures showed a significant decrease in TGF-β ± FGF23-induced SMAD3 phosphorylation and COL1A1 levels (Fig. 9C). Together, these data suggest that murine lung fibroblasts that overexpress klotho also show an attenuated COL1A1 response and decreased SMAD3 phosphorylation, when stimulated with TGF-β ± FGF23 and when compared with fibroblasts from WT mice.

Fig. 9.

Fibroblast growth factor 23 (FGF23) and klotho (KL) supplementation attenuates transforming growth factor-β (TGF-β) signaling in lung fibroblasts from KL-overexpressing (OE) mice. A: bar graphs indicating klotho transcript levels in fibroblasts isolated from wild-type (WT) and KL OE mice, at baseline or treated with FGF23 (40 ng/ml) and TGF-β (5 ng/ml) for 24 h. B: immunoblots (IB) of COL1A1, phospho and total SMAD3, and β-actin from protein lysates of fibroblasts, treated for 24 h with FGF23, TGF-β or both and C: densitometric analysis of the change in TGF-β + FGF23-induced change in p-SMAD3/SMAD3/ACTIN and COL1A1/ACTIN ratios of both WT and KL OE fibroblasts. (All n = 3 independent experiments from 3 different lungs showing means ± SD with *P < 0.05 and ***P < 0.005.)

DISCUSSION

Our findings characterize FGF23 and KL signaling as a novel mechanism involved in the pathogenesis of idiopathic pulmonary fibrosis. Whereas soluble serum KL levels are significantly downregulated in individuals with IPF, FGF23 levels are markedly elevated in this patient cohort. Furthermore, KL seems to be downregulated in primary human lung fibroblasts, which were derived from IPF lungs. In vivo analysis using the bleomycin mouse model validated these findings indicating that bleomycin treatment decreased pulmonary mRNA expression of KL, while it increased local FGF23 expression. Our in vitro experiments using human lung fibroblasts revealed that KL itself had no effect on TGF-β-induced collagen deposition and inflammation, but when it was combined with FGF23, KL was able to ameliorate the profibrotic and proinflammatory TGF-β effect in IPF (Fig. 8), which could be attributed to a decrease in SMAD3 phosphorylation.

Lastly, overexpression of KL in vivo attenuated the bleomycin-induced fibrosis in mice. Therefore, KL downregulation seems to be a crucial step in pathogenesis of IPF with FGF23 elevation being more complex.

Analysis of klotho protein levels has been very challenging; different assays seem to measure different forms and these results do not correlate with each other. Furthermore, there are no established reference values (51) and klotho antibodies are either not sufficiently sensitive and/or the klotho protein itself, which exists in a full-length form and soluble form, easily degrades (25, 43, 62). Therefore, most of our results regarding klotho expression are shown as transcripts by quantitative real-time PCR (most antibodies that are still available detect klotho the kidney, where it is most abundantly expressed in its full-length form; Refs. 7, 49), since this was the most reliable and sensitive method. The ELISA, which is predominantly used to analyze soluble KL in human plasma, shows negative correlation of KL levels with age, consistent with previously discussed in vivo data characterizing KL as an antiaging protein (27, 32, 34). There are multiple reports showing klotho serum downregulation in chronic kidney disease, diabetes, and Alzheimer’s disease (37, 60); one report shows upregulation in healthy athletes attempting to associate positive regulation with exercise (45), but Pako et al. (50) assessed KL levels in COPD patients undergoing rehabilitation and did not see any correlation with lung disease or exercise. Therefore, our study is one of the few studies and so far the first one demonstrating significantly reduced serum KL levels in a chronic pulmonary disease. One limitation is that our control and IPF cohort differ significantly in sex distribution due to the fact that both incidence and prevalence are generally higher among men than women (54). Furthermore, creatinine levels were slightly higher in the IPF group, but there was no difference between BUN and creatinine and glucose levels (Table 1). We also do not see any correlation with lung function. Therefore, sufficiently powered clinical trials are needed for further evaluation.

Our results characterizing FGF23 as antifibrotic and anti-inflammatory protein in pulmonary fibrosis and with klotho supplementation was interesting and also surprising to us. We have previously shown that FGF23 is elevated in individuals with cystic fibrosis and in COPD chronic inflammatory diseases (21, 46, 63). In COPD, FGF23 acts proinflammatory via PLCγ/NFAT signaling (31). In the kidney, Doi et al. (12) demonstrated that klotho inhibits TGF-β signaling thereby suppressing renal fibrosis and cancer metastasis. This opposes our human cell culture results showing that KL alone is not sufficient to prevent TGF-β-induced collagen synthesis in lung fibroblasts. A possible underlying mechanism involves FGF23 and KL-mediated suppression of SMAD3 phosphorylation. One explanation for this discrepancy could be that authors from previous studies did not consider the possible involvement of FGF23-induced signaling at that time. Recently, it has been shown that FGF23 can be produced locally by renal tubules and can activate injury-primed renal fibroblasts via FGFR4-dependent signaling and enhancement of TGF-β autoinduction (33, 65, 66). In these reports, the authors did not supplement with KL and evaluate for attenuation. Furthermore, a recent studies have shown that KL can affect FGF23 binding affinities (33). Doi et al. (12) worked with rat renal epithelial cell cultures and human embryonic kidney (HEK293) cells, which are neither primary cells nor fibroblasts. Lastly, the models used to inflict kidney fibrosis (unilateral ureteral obstruction) represent “acute diseases” rather than chronic diseases such as lung fibrosis, which is the focus of our research. There are also other fibroblast growth factors that might be involved in the pathogenesis of lung fibrosis including FGF9 and FGF18 (26), which were not focus of our studies.

In summary, all these effects are klotho independent or more precisely in a state of “klotho downregulation or deficiency.” Therefore, we suggest that the initial upregulation of FGF23 in various “aging” diseases including pulmonary fibrosis can be also seen as a compensatory mechanism. If the system is able to restore KL levels, FGF23 will act together with KL trying to resolve local inflammation and/or fibrosis by attenuation of SMAD3 phosphorylation (Fig. 10). This hypothesis would be consistent with our findings that both FGF23 and KL together can increase transcript levels of KL in human lung fibroblasts. If the system cannot restore the decrease in KL, then inflammation and fibrosis persist, leading to disease such as lung fibrosis, renal disease, or cardiovascular disease. Future studies are needed to decipher the exact signaling mechanisms and source of klotho involved in these pathways, e.g., WNT signaling, which has been shown to induce an inflammatory response in pulmonary fibroblasts (42, 71) as well as its link to KL in the kidney (77). We are also aware that the preclinical model of bleomycin-induced pulmonary fibrosis is less than optimal and lung function often does not correlate with the extent of fibrosis as shown previously (15); we therefore focused on the 3-wk time point, when there was a significant decrease in lung compliance and klotho and increase in FGF23 levels.

Fig. 10.

Model. A cartoon illustrates our working hypothesis of the interplay between fibroblast growth factor 23 (FGF23) and klotho (KL) in healthy aging and disease [i.e., idiopathic pulmonary fibrosis (IPF)]. In IPF, insults from increased aging/disease-related transforming growth factor-β (TGF-β) cause inflammation and fibrosis, which is compounded through the downregulation of KL. A compensatory upregulation of FGF23 during disease progression is an attempt to resolve fibrosis through the upregulation/restoration of KL levels. Therapies designed with a combination of FGF/KL may have benefit in aging and IPF disease.

Currently, there are only few medications approved for the treatment of IPF. Pirfenidone, an oral antifibrotic, partially due to inhibition of TGF-β signaling (16, 40) has been studied and showed a reduction in disease progression compared with placebo (28, 48). Nintendanib is a second drug, which is currently approved for IPF treatment but also shows its main effect in disease progression (56, 57). This inhibitor includes as targets platelet-derived growth factor (PDGF), TGF-β, but also FGF receptor signaling (4, 75). Both antifibrotic agents have immensely improved our care for IPF patients, but we are far from optimal treatment and a possible cure. Reinstitution of KL levels might be an important aspect and could therefore be part of future treatment strategies for IPF patients, which are still desperately needed.

GRANTS

This work was supported by the Flight Attendant Medical Research Institute Grant YFAC152003 (to S. Krick), Cystic Fibrosis Foundation Grant CFF-KRICK1610 (to S. Krick), and National Institutes of Health Grants K99HL131866 and R00HL131866 (to J. W. Barnes) and R03AG059994 (to S. Krick).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.B. and S.K. conceived and designed research; D.D., S. Helton, S. Hutcheson, D. Kurundkar, M.L., R.D., C. Farver, H.T.V., G.K., D. Kentrup, Z.Y., and S.K. performed experiments; J.B., D.D., S. Helton, and S.K. analyzed data; J.B., D.D., C. Faul, V.T., and S.K. interpreted results of experiments; J.B. and S.K. prepared figures; J.W.B. and S.K. drafted manuscript; J.B., N.J.L., J.G., G.K., C. Faul, T.K., J.A.D.A., S.M., V.T., and S.K. edited and revised manuscript; J.B., D.D., S. Helton, S. Hutcheson, D. Kurundkar, N.J.L., M.L., J.G., C. Farver, G.K., C. Faul, T.K., J.A.D.A., S.M., V.J.T., and S.K. approved final version of manuscript.