Chronic Kidney Disease–associated Lung Injury Is Mediated by Phosphate-induced MAPK/AKT Signaling

Abstract

Chronic kidney disease (CKD) is associated with systemic phosphate elevations, called hyperphosphatemia. Translational studies have shown that hyperphosphatemia contributes to CKD-associated inflammation and injury in various tissues, including the kidney, heart, liver, and parathyroid gland. Mechanisms underlying pathologic actions of elevated phosphate on cells are not well understood but seem to involve uptake of phosphate through sodium phosphate cotransporters and phosphate-induced signaling via FGFR1 (fibroblast growth factor receptor 1). Clinical studies indicate patients with CKD are more likely to develop inflammatory and restrictive lung diseases, such as fibrotic interstitial lung diseases, and here we aimed to determine whether hyperphosphatemia can cause lung injury. We found that a mouse model of CKD and hyperphosphatemia, induced by an adenine-rich diet, develops lung fibrosis and inflammation. Elevation of systemic phosphate concentration by administration of a high-phosphate diet in a mouse model of primary lung inflammation and fibrosis, induced by bleomycin, exacerbated lung injury in the absence of kidney damage. Our in vitro studies identified increases of proinflammatory cytokines in human lung fibroblasts exposed to phosphate elevations. Phosphate activated ERK 1/2 (extracellular signal-related kinase 1/2) and PKB/AKT (protein kinase B) signaling, and pharmacological inhibition of ERK, AKT, FGFR1, or sodium phosphate cotransporters prevented phosphate-induced proinflammatory cytokine upregulation. In addition, inhibition of FGFR1 or sodium phosphate cotransporters decreased the phosphate-induced activation of ERK and AKT. Our study suggests that phosphate can directly target lung fibroblasts and induce an inflammatory response and that hyperphosphatemia in CKD and non-CKD models contributes to lung injury. Phosphate-lowering strategies might protect from CKD-associated lung injury.

Chronic kidney disease (CKD) results from irreversible reductions in renal function that persist for at least 3 months due to several heterogeneous disease etiologies (1, 2). Accumulations of phosphate in the blood arise from a decrease of nephron function as CKD progresses (3, 4). The resulting hyperphosphatemia has several clinical implications, including systemic inflammation and vascular calcification (5–7), that can affect distant organs (8).

Our group has previously discovered that elevated phosphate can induce inflammation in the bronchial epithelium of mice and human bronchial epithelial cells (HBECs) in vitro. Furthermore, we found that cigarette smoke–induced injury to HBECs is exacerbated in a hyperphosphatemic environment (9).

The prevalence of restrictive lung disease (RLD), a general classification that includes fibrotic interstitial lung diseases such as idiopathic pulmonary fibrosis (IPF), increases dramatically as renal disease progresses and is an independent predictor of mortality in patients with CKD (10). RLD is present in 9–14% of patients with CKD stage 1–4 and increases to 36% in those with stage 5, or end-stage renal disease (10, 11). Only half of patients with end-stage renal disease present with normal pulmonary function, based on lung spirometry (11). RLD is characterized by a reduction in the distensibility of the lungs, which restricts lung expansion and reduces overall lung capacity (12). In addition to finding an association between RLD and low glomerular filtration rate, research has shown that several biomarkers of inflammation, such as IL6 (13), TNF (tumor necrosis factor) (14, 15), and leukocyte counts (16), increase in step with each progressive escalation of CKD severity stage (11).

Although RLD is not as common as obstructive lung disease, the prognosis for many of these ailments is very poor (17). Median survival rates for IPF, one of the most progressive types of fibrotic interstitial lung disease, are less than 5 years (18), and the medications available only mitigate symptoms, leaving no cure for these patients (19).

Many of the inflammatory markers upregulated in CKD are also precursors to the fibrosis that develops as a result of repeated injury to the lungs in fibrotic interstitial lung disease. Involvement of ERK 1/2 (extracellular signal-related kinase 1/2) and PKB/AKT (protein kinase B) signaling network has been described as a player in phosphate-mediated inflammation in the kidney (2), lung (9), vascular smooth muscle (20), and parathyroid gland (21). Elevated production of several proinflammatory cytokines, including IL1B, IL6, and IL8 (22), as well as TGF-β/TGFB (transforming growth factor-β) upregulation (23), is key in determining the progression of CKD. Research suggests that activation of the ERK pathway is required for the TGFB-induced epithelial-to-mesenchymal transition that is a hallmark of fibrosis in the lung (24). Furthermore, TGFB can activate AKT (25), leading to a vicious cycle of inflammation and fibrosis that provides a plausible crosstalk mechanism between the kidney and lung, particularly in the presence of aberrant phosphate homeostasis during CKD.

Previous studies in bone and kidney cells suggest that phosphate activates many of these inflammatory pathways through type III inorganic sodium phosphate cotransporters PiT-1/2, and FGFR (fibroblast growth factor receptor) activation (20, 26, 27). In addition, inhibition or deletion of FGFR1 or PiT-1/2 attenuates phosphate-induced activation of ERK and AKT downstream (28–30).

Although little is known about whether phosphate signals through similar pathways in other tissues, there are a few studies that have examined the effects of phosphate signaling in the lungs. In K-ras^LA1 mice, a model susceptible to the development of lung cancer, a high-phosphate diet stimulated heightened lung tumorigenesis compared with control animals. NPT2B (sodium-dependent phosphate cotransporter protein 2B) was found to be upregulated as a result of the high-phosphate diet, leading to an increased activation of AKT at Ser473 (31). In addition, deletion of this phosphate cotransporter led to pulmonary alveolar microlithiasis because of excess circulating phosphate. Microliths isolated from the lungs of those mice caused pulmonary inflammation when transferred into the lungs of C57BL/6J mice (32). To our knowledge, the effect of phosphate or hyperphosphatemia on interstitial fibrotic lung diseases has not been addressed experimentally, leaving a gap of knowledge to bridge CKD, phosphate, and fibrotic lung disease.

Therefore, the aims of this study were twofold: 1) establish whether CKD-associated hyperphosphatemia has an impact on the lung and pulmonary fibroblasts; and 2) determine if hyperphosphatemia increases the susceptibility to bleomycin-induced inflammation and/or fibrosis in murine lungs. Our previous findings (9, 33, 34), along with studies showing that hyperphosphatemia can induce inflammation in the heart (35), liver (36), kidney (2), parathyroid glands (21), and lungs of aged mice (37), suggest that these CKD-associated mineral and hormonal imbalances may be a culprit of RLD susceptibility. Given this evidence, we hypothesize that increased phosphate concentrations contribute to a proinflammatory response in lung fibroblasts and exacerbate bleomycin-induced murine lung inflammation and fibrosis.

Methods

Mice

Animal studies were performed as approved by the Animal Research Committee at the University of Alabama at Birmingham (UAB) School of Medicine and in conformity with applicable laws and guidelines for their use in research studies. Housing and use of the experimental vertebrate animals in this study and associated protocols was reviewed and approved by the UAB Institutional Animal Care and Use Committee. Animal Research: Reporting of In Vivo Experiments reporting guidelines were followed for reporting animal research (38). All mice used in this study were maintained on a C57BL/6 background. Studies were performed using mice maintained on an NIH-31 rodent diet (Harlan Teklad) and fed ad libitum until noted otherwise in each study.

For experiments involving adenine-induced kidney failure, an established dietary timeline was used to achieve desired renal effects (6, 9, 39). Briefly, 10- to 14-week-old male mice (40) were placed on a customized diet containing 0.2% adenine (TD.140290; Envigo) for 6 weeks, subsequently transferred to another customized diet containing 0.15% adenine (TD.170304) for 2 weeks, and finally switched back to the original 0.2% adenine diet for a final 6 weeks. Littermates were placed on a control diet (TD.170303; Envigo) to serve as controls for the study. After a 1-week diet acclimation stage, the mice were randomly assigned to the control or adenine group during the study. After the 14-week experimental dietary period, mice were killed under 2.5% isoflurane anesthesia, and samples were prepared as described below. This diet is commonly used to study CKD due to the kidney tubule–interstitial damage caused by insoluble 2,8-dihydroxyadenine crystals as a result of the excess dietary adenine (41).

For experiments investigating the pathologic effects of dietary phosphate load on bleomycin-induced lung injury, 8-week-old male mice (42) were randomly assigned to receive a 0.7% phosphate control diet (TD.00588; Envigo) or a customized 3% high-phosphate diet (TD.180286; Envigo) as treatment for 6–7 weeks, as established previously (6). At the end of Week 4 on the custom diet, mice from each group were again randomly assigned to receive a single oropharyngeal administration of bleomycin (1.25 U/kg) or saline (control) as previously described (33). Two weeks after administration during the inflammatory phase or 3 weeks after administration during the fibrotic phase of bleomycin-induced lung damage, mice were killed under 2.5% isoflurane anesthesia, and samples were prepared as described below.

Serum Chemistry

Cardiac puncture was used to collect murine blood, which was then transferred to Microvette serum tubes (365967; Becton, Dickinson and Co.) and allowed to sit undisturbed for 30 minutes. Blood samples were then centrifuged at 10,000 × g for 5 minutes at room temperature, and serum supernatants were separated and stored at −80°C. Clinical chemistry analyses to quantify blood urea nitrogen (BUN), creatinine, and serum phosphate concentration of the samples were performed by IDEXX BioAnalytics (6006-UC, comprehensive panel). Serum intact FGF23, the biologically active form before C- and N-terminal cleaved fragments are created, was assessed using a commercial ELISA kit (60-6800, Quidel) according to the manufacturer’s protocol.

BAL Collection and Cell Quantification

BAL fluid (BALF) was collected by methods previously established in our lab (34). Briefly, mice were killed, and a flexible cannula was inserted into a small hole cut near the top of the trachea. The cannula was tied off above the opening to create a leak-proof seal. Cold PBS (1 ml) was used to lavage the lungs while visualizing lung distension to verify effective fluid dispense and was then immediately withdrawn. Cells were pelleted via centrifugation and resuspended in nine parts sterile water for 30 seconds to lyse any red blood cells, followed by the addition of one part 10 × PBS to return the fluid to an isotonic concentration of 1 × PBS. Prepared BALF was loaded into a CytoSep funnel (M9641; Simport Scientific) along with charged microscope slides (12-550-15; Fisher Scientific), and each set was centrifuged at 300 × g for 5 minutes (CytoSpin 4; Thermo Scientific). Each slide was allowed to dry overnight and subsequently stained with Hema 3 (123-869; Fisher Scientific). Macrophages, lymphocytes, and neutrophils were quantified by visual inspection under an upright microscope using randomly selected fields until the total count reached 400 cells.

Hydroxyproline Assay

Mice were killed either 14 or 21 days after bleomycin instillation, and lung tissue was collected from each mouse. Lung tissues were dried at 70°C in an oven for 48 hours and then hydrolyzed with 6 N HCl at 95°C for an additional 20 hours. The hydroxyproline content in each lung was assessed using manufacturer’s instructions provided with a colorimetric quantitative kit (AZBhypro1; QuickZyme Biosciences) using hydroxyproline as a standard, as previously described (33).

Lung Histology and Immunohistochemical Staining

After euthanizing mice from the study, the lungs of each animal were perfused with 3 ml of PBS via the right ventricle. The lungs were inflated with 1 ml of 10% formalin solution using a catheter (26751; Exel International) inserted through a small slit in the trachea and submerged in 10% formalin solution at room temperature for 24 hours. The lungs were subsequently transferred to 70% ethanol and then subjected to paraffin embedding, sectioning, and Masson’s trichrome or Picrosirius red staining by the Comparative Pathology Laboratory at UAB. In addition, paraffin-embedded lung sections were also subjected to immunohistochemical (IHC) staining with IL1B (AF-401-NA; R&D Systems) or αSMA (α-smooth muscle actin; ACTA2) using a mouse and rabbit HRP/DAB detection kit (ab64264; abcam). Images of all stains were captured on a Nikon Ts2-FL inverted microscope with a Digital Sight 10 camera attached.

Furthermore, paraffin-embedded lung sections were subjected to an immunofluorescence staining protocol adapted from the aforementioned IHC protocol, as previously published (43). Briefly, paraffin sections were melted for 50 minutes in an oven set to 60°C. Slides were then deparaffinized and rehydrated using Clear-Rite 3 (Epredid; 6901) and 100%, 95%, and 70% ethanol, in order. Antigen retrieval was obtained using antigen unmasking solution (100×, citrate; Vector Labs, H3300) in a steamer for 20 minutes. Tissue sections were then rinsed in PBS and permeabilized for 5 minutes using 0.1% Triton in PBS before blocking in a 5% BSA solution. Tissues were then incubated overnight in primary antibodies IL1B (1:20, AF-401-NA; R&D Systems) and ASMA (1:500, PA5-85070; Invitrogen). After several 5-minute washes with 1 × PBS, the tissue sections were incubated for 45 minutes with the respective secondary antibodies (1:1,000, donkey anti-mouse 488 and donkey anti-rabbit 594; Invitrogen). The slides were then subjected to several 5-minute 1 × PBS washes again, and the tissue was quenched using a Vector TrueView Autofluorescence Quenching Kit (SP-8400-15; Vector Laboratories) for 5 minutes. Next, slides were washed with 1 × PBS for 3 minutes and incubated with diluted NucBlue (2 drops in 1 ml PBS) for 10 minutes to stain the nuclei before mounting with Prolong Gold Mounting Media and sealing with coverslips. After mounting, microscope slides were protected from light and stored at 4°C until imaging. Images of all stains were captured on a Nikon Ts2-FL inverted microscope with a Digital Sight 10 camera attached.

Cell Culture

Human lung fibroblast cells (IMR-90, CCL-186; ATCC) were cultured in Dulbecco’s Modified Eagle Medium, high glucose (10-017-CV; Corning) on tissue culture–treated plates as previously described (33). Cells were treated with the stated amounts of sodium phosphate (0.5 M; pH 7.4) or sodium sulfate (0.5 M; pH 7.4) buffers, which were prepared as previously described, to produce the final desired concentrations. Because physiological concentrations of phosphate in serum are around 1 mM in healthy adults (44), we used this as our control, as well as sodium sulfate, for an ion-balanced control at each respective concentration. In experiments in which inhibitors were used, cells were pretreated with an ERK inhibitor (U0126, iERK), AKT inhibitor (MK-2206, iAKT), or FGFR1-3 inhibitor (AZD4547, iR1), or phosphonoformic acid (PFA, P6801; Sigma) as a sodium phosphate cotransporter inhibitor for 1 hour before the addition of phosphate.

Primary human IPF and non-IPF fibroblasts were derived from age-matched deidentified explanted lung tissue under a protocol approved by the UAB Institutional Review Board and conformed to the Declaration of Helsinki as previously described (33). IPF diagnosis was made using a multidisciplinary approach in accordance with guidelines given by the American Thoracic Society and European Respiratory Society. Primary fibroblasts were cultured following the same protocol as described above with IMR-90 human lung fibroblasts.

ELISA

Concentrations of IL6 and IL8 cytokines from conditioned cell culture media were quantified using a human IL6 (88-7066-88; Invitrogen) and human IL8 (88-8086-86; Invitrogen) ELISA. Media collected from IMR-90 cell cultures at specified time points after treatment was clarified by centrifugation at 500 × g for 10 minutes at 4°C. Clarified medium was loaded onto an assay plate coated with either anti-IL6 or anti-IL8 capture antibody and incubated on a plate shaker at 220 revolutions per minute for 2 hours at room temperature. After following the manufacturer’s protocol, absorbance values were measured at 450 nm, and concentrations of each cytokine were determined using a serially diluted standard included with each ELISA kit.

Concentrations of TGFB from mouse serum were quantified using a mouse TGFB1 DuoSet ELISA (DY1679; R&D Systems) using the manufacturer’s provided instructions.

RNA Purification and Quantitative RT-PCR

Total RNA was extracted from murine lung tissue with the Mini Total RNA Kit (IB47302; IBI Scientific) and from IMR-90 cell cultures using the GeneJET RNA purification kit (K0732; Thermo Scientific). Using a two-step reaction method, 1 μg of total RNA was reverse-transcribed into cDNA using Maxima H Minus cDNA Synthesis Master Mix (M1669; Thermo Scientific) in a thermocycler (PCT-100; MJ Research), using the manufacturer’s suggested amplification protocol. For analysis of gene expression, quantitative RT-PCR (qRT-PCR) was performed with 25 ng of cDNA using the following Taqman probes (Life Technologies/Applied Biosystems): Mm00436450_m1 for mouse Cxcl (C-X-C Motif Chemokine Ligand), 2/Mip2 (macrophage inflammatory protein 2/MIP-2), Mm00434228_m1 for mouse Il1b, Mm00446190_m1 for mouse Il6, Mm00801666_g1 for mouse Col1a1, Mm00725412_s1 for mouse Acta2, Mm00438930_m1 for mouse Fgfr1, Mm01341852_m1 for mouse Fgfr4, Mm01178820_m1 for mouse Tgfb1, Hs01555410_m1 for human IL-1B, Hs00174131 for human IL6, and Hs00174103_m1 for human IL8. The internal control used to normalize gene expression values for mouse samples was Mm99999915_g1 for Gapdh and for human samples was Hs02758991 for GAPDH. All qRT-PCR was performed on a StepOnePlus Real-Time PCR System (Applied Biosystems), and results were then evaluated using the 2^-ΔΔCT method.

Western Immunoblotting

Lysates from cells were prepared in radioimmunoprecipitation assay buffer (9806; Cell Signaling) with Halt protease and phosphatase inhibitor cocktail (78430; Thermo Scientific). Bradford protein assay (1856210; Pierce) was used to measured protein yield from the samples. Precast gradient Ready Gels (4–20%, Bio-Rad) were used to separate 40 μg total protein and blotted onto nitrocellulose blotting membranes with 0.45 μm pores (10600003; Amersham). Membranes were subsequently blocked with 5% BSA for phosphorylated proteins or 5% nonfat milk for βactin in Tris-buffered saline (pH 7.4) with 0.05% Tween-20 (TBST) for 30 minutes. Primary antibodies against specific antigens were diluted 1:1,000 and incubated with the membranes overnight. AKT (9272; Cell Signaling), pAKT at Ser473 (4060; Cell Signaling), ERK1/2 (4695; Cell Signaling), and pERK1/2 (9101; Cell Signaling) were diluted in TBST.

The next day, membranes were washed and incubated with horseradish peroxidase–conjugated goat anti-mouse (31430; Invitrogen) or goat anti-rabbit (31466; Invitrogen) secondary antibodies diluted 1:6,000 in TBST at room temperature for 1 hour. Membranes were washed again, and horseradish peroxidase activity was detected using enhanced chemiluminescence detection solution (1705060; Bio-Rad) and imaged on a Bio-Rad Gel Doc XR Gel Documentation System for quantification. All immunoblots were repeated in at least three independent trials with comparable results. Densitometry was performed using Fiji/ImageJ software (45) as previously described (46).

Statistics

All presented results are expressed as mean ± SEM with all illustrated points representing individual samples. To analyze group differences between continuous variables, Student’s t tests were used. In instances where data sets contained two or more predictor variables, a one-, two-, or three-way ANOVA was used to analyze differences between the groups, followed by Tukey’s multiple comparison test for post hoc analysis. Although comparisons between the means of all groups were performed, graphical representation of significance was shown only for those referenced in the test to allow for visual clarity. Details on the specific tests used for each figure are included in the corresponding legend. A P value < 0.05 was deemed statistically significant for the purposes of data analysis. Organization, graphical representation, and analysis were all performed using Prism version 9 (GraphPad). Assumptions for each respective test were verified as part of this analysis. Sample size was determined based on sample availability, literature published on the study in question, and experimental studies performed by our laboratory previously.

Results

Mice with Adenine-induced CKD Exhibit Increased Markers of Pulmonary Fibrosis

To evaluate whether hyperphosphatemia resulting from kidney injury induces lung injury in vivo, we analyzed the lungs of an established mouse model of CKD induced by an adenine-enriched diet (47). Consistent with the previously characterized renal dysfunction, we found that mice on the 14-week adenine-enriched diet had significantly increased creatinine and BUN concentrations when compared with those on a control diet. Furthermore, serum phosphate and FGF23 concentrations were also significantly elevated (Figure 1A). Masson’s trichrome, Picrosirius red, and ACTA2-stained lung sections (Figure 1B and Figure E1B in the data supplement) revealed altered lung pathology compared with control mice, confirmed by decreases in mean linear intercepts (L_m) and an increase in red-stained area for each imaged field of the mice on the adenine diet (Figure 1C). Arrows indicate areas of increased interstitial fibrosis in the lungs of adenine-fed mice based on their respective staining (Figure 1B). Immunofluorescence-stained lung sections revealed higher expression of ACTA2 and IL1B in adenine-fed mice, with certain areas demonstrating colocalized positive signal for both analytes (Figure E1A). These findings reflected similar results from a previous study examining the lungs of adenine-induced CKD mice (48). Gene expression of markers commonly assessed in fibrosis, Col1a1 (collagen type I, α 1 chain) and Acta2, were also significantly upregulated in the total lung homogenate from the adenine diet mice (Figure 1D). Moreover, hydroxyproline concentration, a direct measure of collagen content (49), was significantly higher in the total lung homogenate from the adenine-fed mice as well (Figure 1E). In summary, an established diet-induced CKD mouse model demonstrates increased markers of pulmonary fibrosis.

Figure 1.

Adenine mice exhibit signs of pulmonary fibrosis. (A) Creatinine, BUN, phosphate, and intact FGF23 concentrations from the serum of mice on the adenine diet and littermates on the control diet. (B) Representative pathology of trichrome, Picrosirius red, and ASMA-stained lung sections from experimental mice. Scale bars, 100 μm. (C) Quantification of mean linear intercept (L_m) from trichrome-stained lung sections and percent Picrosirius red–stained area per field from adenine and control diet experimental mice. Values from three separate fields were averaged for each animal. (D) Relative mRNA transcript expression of Col1a1 and Acta2 and (E) hydroxyproline concentrations in whole-lung homogenates from adenine and control diet experimental mice. All values are mean ± SEM, with *P ≤ 0.05, **P ≤ 0.01, and ***P < 0.001; n = 5–15 mice/group. Statistical significance was analyzed via unpaired Student’s t test. BUN = blood urea nitrogen.

Phosphate Induces Expression of Proinflammatory Cytokines in Human Lung Fibroblasts

Given our in vivo results suggested a fibrotic phenotype in the lungs of CKD mice, we wanted to test whether phosphate was the culprit of this damage. Primary human lung fibroblasts (IMR-90) were treated with increasing concentrations (1–5 mM) of inorganic sodium phosphate or sodium sulfate. Starting at a 3-mM phosphate concentration in the cell culture media, IL1B, IL6, and CXCL8 mRNA expression was significantly increased when compared with the 1-mM phosphate physiological control and corresponding ion-balanced controls (Figure 2A). Similarly, IL6 and IL8 protein expression was also increased in the conditioned media at 3, 4, and 5 mM phosphate concentrations (Figure 2B). Expression of COL1A1 and ACTA2 were unchanged as a result of 24-hour phosphate exposure (data not shown). To confirm that these findings were also reflected in primary lung fibroblast cells derived from control donors and donors with fibrotic lung disease, and to assess whether differences in response to phosphate could be measured between lung fibroblasts derived from patients with and without IPF, we subjected cells to 1-mM control and 3-mM high phosphate for 24 hours as well. Although ACTA2 gene expression was unchanged in non-IPF lung fibroblasts after phosphate stimulation, baseline and post–3-mM Pi treatments both stimulated an upregulation of ACTA2. Treatment with 3-mM phosphate increased IL6 and IL8 secretion in both non-IPF and IPF primary lung fibroblasts, with an exacerbated response in IPF fibroblasts, when compared with the non-IPF cells (Figure 2C).

Figure 2.

Phosphate induces proinflammatory cytokine expression and secretion in human lung fibroblasts. (A) Relative mRNA transcripts of IL1B, IL6, and CXCL8 in human lung fibroblasts after 24 hours of Pi treatment. (B) IL6 and IL8 protein concentrations in conditioned media from human lung fibroblasts after 24 hours of Pi treatment. (C) Relative mRNA transcripts of ACTA2 (α-smooth muscle actin) and IL6 and IL8 protein concentrations in conditioned media from primary human lung fibroblasts isolated from patients with and without idiopathic pulmonary fibrosis (IPF). All experiments were reproduced three to five times in triplicates. Individual points for non-IPF and IPF graphs represent separate donors. All values are mean ± SEM with *P ≤ 0.05, **P ≤ 0.01, and ***P < 0.001. Statistical significance was analyzed via one-way (A and B) or two-way (C) ANOVA followed by Tukey’s multiple comparison post hoc test for groups depicted with significance bars.

Given that CKD leads to not just a state of hyperphosphatemia but also elevated FGF23 as a regulator of phosphate metabolism, and that FGF23 can induce a proinflammatory response in hepatocytes (6), IMR-90 cells were also treated with increasing concentrations of FGF23. Selected concentrations included median treatment concentrations consistent with an increase in inflammatory cytokine expression from our previous studies examining effects on HBECs from patients with chronic obstructive pulmonary disease (COPD) (34). Treatment with FGF23 had no effect on IL1B, IL6, or CXCL8 expression concentrations after 24 hours in IMR-90s (Figure E2A). Similarly, conditioned media collected after treatment with FGF23 also showed no significant difference in secreted IL6 or IL8 cytokine concentrations (Figure E2B). These data indicate that phosphate, but not FGF23, induces proinflammatory cytokine mRNA expression in human lung fibroblasts.

Inhibition of ERK or AKT Signaling Prevents Phosphate-induced Upregulation of Proinflammatory Cytokines in Human Lung Fibroblasts

To better understand the underlying molecular pathways leading to phosphate-mediated lung inflammation and fibrosis, we examined several signaling mediators involved in phosphate-related sensing and mineral dyshomeostasis (30, 50, 51). In addition to our previous study of HBECs (9), reports describing extrapulmonary manifestations suggest that phosphate can induce inflammation through ERK and AKT activation (9, 27). Based on these findings, we decided to analyze these pathways in human lung fibroblasts, singling out the individual effects of phosphate and FGF23. Treatment with 3-mM phosphate for 24 hours induced significant ERK phosphorylation in the fibroblast cultures (Figure 3A). To elucidate whether the phosphate-induced activation of ERK ultimately led to proinflammatory cytokine upregulation in human lung fibroblasts, we next focused on inhibiting ERK phosphorylation. Before stimulating with 3-mM phosphate, human lung fibroblasts were treated with iERK as previously described (9). ERK phosphorylation was effectively inhibited in both control and high phosphate–treated cultures (Figure 3B). Surprisingly, AKT phosphorylation increased with ERK inhibition, both at the 1-mM phosphate control and 3-mM phosphate concentrations (Figure 3C). In addition, the ERK inhibitor completely abolished the phosphate-induced increases in IL1B, IL6, and CXCL8 mRNA expression (Figure 3D). A similar decrease of IL6 and IL8 inflammatory cytokine release in the conditioned media from lung fibroblasts was also observed after ERK inhibition (Figure 3E).

Figure 3.

Inhibition of phosphate-driven ERK (extracellular signal-related kinase) activation attenuates phosphate-induced proinflammatory cytokine expression and secretion. (A) Representative immunoblots and quantification of ERK phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation. βactin served as loading control in all blots. (B) Representative immunoblots and quantification of ERK phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation ± 1 hour pretreatment with 5 μm ERK inhibitor (iERK). (C) Representative immunoblots and quantification of AKT phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation ± iERK pretreatment. (D) Relative mRNA transcripts of IL1B, IL6, and CXCL8 in human lung fibroblasts after 24 hour Pi stimulation ± iERK pretreatment. (E) IL6 and IL8 protein concentrations in conditioned media after Pi stimulation of human lung fibroblasts ± iERK pretreatment. All experiments were reproduced five to six times with triplicates. All values are mean ± SEM, with *P ≤ 0.05, **P < 0.01, and ***P < 0.001. Statistical significance was analyzed via one-way (A) or two-way (B–E) ANOVA followed by Tukey’s multiple comparison post hoc test for groups depicted with significance bars.

Interestingly, activation of AKT was observed starting at 2-mM phosphate (Figure 4A), suggesting this pathway is activated with a lower phosphate concentration in lung fibroblasts. To determine whether AKT inhibition also attenuates the effects of phosphate-induced proinflammatory cytokine expression, human lung fibroblasts were pretreated with iAKT before phosphate treatment. Although ERK phosphorylation was unaffected by iAKT (Figure 4B), AKT phosphorylation was significantly lower at both 1-mM phosphate control and 3-mM phosphate concentrations (Figure 4C). Inhibition of AKT phosphorylation significantly reduced phosphate-mediated IL1B, IL6, and CXCL8 mRNA expression (Figure 4D). This trend was also observed in the phosphate-induced secretion of IL6 and IL8, with AKT inhibition significantly reducing concentrations of both cytokines in the conditioned media after 24 hours (Figure 4E). Collectively, these data suggest that phosphate-induced inflammatory cytokine upregulation is, in part, dependent on both ERK and AKT activation.

Figure 4.

Inhibition of phosphate-driven AKT activation attenuates phosphate-induced proinflammatory cytokine expression and secretion. (A) Representative immunoblots and quantification of AKT phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation. βactin served as loading control for all blots. (B) Representative immunoblots and quantification of ERK phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation ± 1 hour pretreatment with 1 μm iAKT. (C) Representative immunoblots and quantification of AKT phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation ± iAKT pretreatment. (D) Relative mRNA transcripts of IL1B, IL6, and CXCL8 in human lung fibroblasts after 24 hours of Pi stimulation ± iAKT pretreatment. (E) IL6 and IL8 protein concentrations in conditioned media after Pi stimulation of human lung fibroblasts ± iAKT pretreatment. All experiments were reproduced five to six times with triplicates. All values are mean ± SEM with *P ≤ 0.05, **P < 0.01, and ***P < 0.001. Statistical significance was analyzed via one-way (A) or two-way (B–E) ANOVA followed by Tukey’s multiple comparison post hoc test for groups depicted with significance bars.

To examine whether concomitant stimulations with FGF23 would alter these effects, the same fibroblast cultures were costimulated with 1- or 3-mM phosphate and 40 ng/ml FGF23. As expected, ERK was activated after phosphate stimulation. However, FGF23 did not alter these effects in a significant manner (Figure E2C). AKT stimulation followed a similar trend, with FGF23 having no effect on the AKT phosphorylation at baseline or with 3-mM phosphate (Figure E2D). In summary, these data indicate that phosphate is sufficient to activate ERK and AKT, and FGF23 costimulation does not induce any further effect.

Inhibition of Phosphate Transport and FGFR1 Signaling Mitigates Phosphate-induced ERK and AKT Activation as Well as Inflammation

Previous reports have shown that bleomycin-induced lung inflammation and fibrosis can be modulated by inhibition of FGFRs (52). Interestingly, we found significant increases of Fgfr1 expression in the lungs of our adenine-induced CKD mice (Figure 5A). In addition, several groups have highlighted that phosphate-induced activation of ERK and AKT can be mitigated with inhibition of both sodium phosphate cotransporters and FGFR1 (50, 53–55), suggesting that this pathway may be important to understand if CKD-associated disruptions in phosphate homeostasis are tied to inflammation and fibrotic transition of fibroblasts in the lung. To determine whether phosphate must enter IMR-90s through sodium phosphate cotransporters and/or signal through FGFR1, we used known inhibitors of each component. PFA has been previously shown to suppress PiT-1/2 expression and function (50, 56), an important channel for phosphate to be imported into cells. Pretreatment of IMR-90 human lung fibroblasts with PFA, as previously described (50), significantly reduced the phosphate-induced increased expression of IL1B, IL6, and CXCL8 (Figure 5B). Similar reductions in IL6 and IL8 cytokines secreted into the media after 3-mM phosphate treatment were also seen with PFA addition (Figure 5C). Pretreatment of IMR-90 human lung fibroblasts with a well-established and clinically used FGFR1 inhibitor, iR1, also significantly decreased phosphate-induced increases of IL1B, IL6, and CXCL8 mRNA expression (Figure 5D) and secretion of IL6 and IL8 into the media (Figure 5E). Expression and secretion of these cytokines were unchanged by PFA or iR1.

Figure 5.

Inhibition of phosphate transporters or FGFR1 (fibroblast growth factor receptor 1) decreases proinflammatory cytokine upregulation and ERK/AKT activation. (A) Relative mRNA transcript expression of Fgfr1 and Fgfr4 in whole-lung homogenates from adenine and control diet experimental mice. (B) Relative mRNA transcript expression of IL1B, IL6, and CXCL8 in human lung fibroblasts after 24 hours of Pi stimulation ± 1 hour pretreatment with 5 mM phosphonoformic acid (PFA). (C) IL6 and IL8 protein concentrations in conditioned media after Pi stimulation of human lung fibroblasts ± PFA pretreatment. (D) Relative mRNA transcript expression of IL1B, IL6, and CXCL8 in human lung fibroblasts after 24 hours of Pi stimulation ± 1 hour pretreatment with 0.1 μm iR1. (E) IL6 and IL8 protein concentrations in conditioned media after Pi stimulation of human lung fibroblasts ± iR1 pretreatment. (F) Representative immunoblots and quantification of ERK phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation ± PFA or iR1 pretreatment. βactin served as loading control for all blots. (G) Representative immunoblots and quantification of AKT phosphorylation in total protein extracts from human lung fibroblasts after Pi stimulation ± PFA or iR1 pretreatment. All in vitro experiments were reproduced five to six times with triplicates. All values are mean ± SEM with *P ≤ 0.05, **P < 0.01, and ***P < 0.001. Statistical significance was analyzed via unpaired Student’s t test (A) or two-way (B–E) or three-way (F and G) ANOVA followed by Tukey’s multiple comparison post hoc test for groups depicted with significance bars. ns = not significant.

Finally, we examined ERK and AKT activation after PFA or iR1 pretreatment of IMR-90 cells to determine if these proteins were also part of the molecular pathways activated by phosphate. After iR1 addition, ERK phosphorylation was significantly reduced in the fibroblasts in both the 1-mM phosphate control and 3-mM phosphate-treated cells. PFA reduced 3-mM phosphate ERK activation without significant change at the 1-mM phosphate control concentration (Figure 5F). AKT phosphorylation was unchanged with PFA or iR1 pretreatment at a 1-mM phosphate concentration, but addition of both inhibitors was sufficient to prevent 3-mM phosphate–induced AKT activation (Figure 5G). All together, these data suggest that PiT-1/2 and FGFR1 are important potentiators of the phosphate-mediated proinflammatory response in pulmonary fibroblasts. Furthermore, it appears that phosphate transport into the fibroblasts and signaling through FGFR1 is upstream of ERK and AKT activation in a high-phosphate environment.

A High-Phosphate Diet Exacerbates 14-Day Bleomycin-induced Lung Inflammation In Vivo

To assess whether excess phosphate and/or FGF23 contributes to lung inflammation in the absence of CKD, C57BL/6 mice were placed on control (0.7%) or elevated (3%) dietary phosphate load for 4 weeks, following a previously established model (6, 57). Mice were subsequently given a single oropharyngeal instillation of saline or 1.25 U/kg bleomycin and continued on the diet for an additional 2 weeks to reach a proinflammatory phase (58) of lung injury (Figure 6A). Groups were classified as control diet + saline (CS), high-phosphate diet + saline (PS), control diet + bleomycin (CB), or high-phosphate diet + bleomycin (PB). At the conclusion of the 6-week experiment, serum creatinine and BUN were unchanged across all four mouse groups, indicating normal kidney function. As expected, both serum phosphate and FGF23 were significantly higher in the mice on a high-phosphate diet, regardless of saline or bleomycin instillation (Figure 6B). To determine the effects of phosphate on lung inflammation ± bleomycin, we collected BALF at the time of harvest. Total cell count was increased in both bleomycin groups, compared with the CS mice, with counts in the CB group significantly higher than the PB group. This trend held true in macrophage, lymphocyte, and neutrophil cell counts as well, with inflammatory cell counts in the BALF of PB mice consistently higher than those in the CB mice (Figure 6C). This proinflammatory response was further confirmed by elevated mRNA expression of Il1b and Il6 in the total lung homogenate from bleomycin-treated mice, which was further exacerbated in the PB group. Cxcl2 proinflammatory cytokine mRNA expression was significantly higher in the PB group but not in the CB group, compared with mice in the PS group (Figure 6D).

Figure 6.

Bleomycin-induced inflammatory lung injury is exacerbated by a high-phosphate diet. (A) Experimental timeline of 6-week mouse diet (control or high phosphate), followed by intratracheal instillation of saline or bleomycin. (B) Serum creatinine, BUN, phosphate, and intact FGF23 concentrations and (C) total cell count, macrophage, lymphocyte, and neutrophil counts from BAL fluid of experimental mouse groups. (D) Relative transcript expression of Il1b, Il6, and Cxcl2 in whole murine lung homogenates from each group. All values are mean ± SEM with *P ≤ 0.05, **P ≤ 0.01, and ***P < 0.001; n = 7–8 mice/group. Statistical significance was analyzed via two-way ANOVA followed by Tukey’s multiple comparison post hoc test for groups depicted with significance bars. CB = control diet + bleomycin; CS = control diet + saline; PB = high-phosphate diet + bleomycin; PS = high-phosphate diet + saline.

To confirm profibrotic and proinflammatory gene expression values and to examine protein expression location within the mouse lung, immunofluorescence was performed on sections from each treatment group. ACTA2 expression was higher in the lungs of mice treated with bleomycin, as was expected, but was slightly higher in the PB group (Figure 7A). IL1B was primarily present in the bleomycin groups as well, with more present in the PB group than the CB group (Figure 7A). IHC of IL1B showed higher protein expression in the lungs from bleomycin-treated mice, with further increase in bleomycin-treated mice on a high-phosphate diet, compared with the CS and PS groups. Increased IL1B expression was observed in areas of interstitial fibrosis in the lungs of PB mice (Figure 7B, arrows). In summary, a murine model of hyperphosphatemia without kidney injury exacerbates bleomycin-induced lung inflammation as a precursor to fibrosis.

Figure 7.

Bleomycin-induced inflammatory lung pathology is exacerbated by a high-phosphate diet and corresponds to fibrotic regions. (A) Representative pathology of ACTA2 (red) and IL1B (green) immunofluorescence-stained lung sections from each mouse group. Scale bars: A, 100 μm (20×); 200 μm (40×). (B) Representative pathology of IL1B immunohistochemical-stained lung sections from each mouse group. Scale bars, 100 μm. Arrows indicate areas of increased IL1B-positive staining.

High Phosphate Exacerbates the Fibrotic Response in Bleomycin-treated Mice

Given our results surrounding bleomycin-induced inflammation in mice on a high-phosphate diet, we wanted to determine if lung fibrosis was also exacerbated by hyperphosphatemia. Groups followed the previously described 4 weeks of 0.7% control or 3% high-phosphate diet load, were subsequently administered a single dose of 1.25 U/kg of bleomycin to induce lung injury, and were allowed to proceed for 3 weeks to reach the fibrotic phase (58, 59) (Figure 8A). As expected, no differences in serum creatinine or BUN were measured between groups after the full 7-week experiment. Serum phosphate and FGF23 concentrations were significantly higher in both phosphate diet groups (Figure 8B). Although this timeline was focused on lung fibrosis, cell counts in BALF were still amplified in CB and PB groups, with phosphate leading to a further increase. Macrophage, lymphocyte, and neutrophil counts were significantly higher in the BALF of groups exposed to bleomycin, but only macrophages were elevated in PB mice compared with the CB group (Figure 8C). Hydroxyproline concentration was significantly higher in the lungs of both bleomycin-exposed groups compared with CS and further elevated in the PB group over CB (Figure 8D). Serum TGFB was also higher in the bleomycin-treated mice, as is commonly observed as part of the fibrotic model (60), and phosphate exacerbated this response in the PB group (Figure 8E). Both Col1a1 and Tgfb mRNA expression from total lung homogenates were significantly higher in the CB groups than in the CS mice, whereas PB mice exhibited further significant increases in the expression of each fibrotic marker gene (Figure 8F).

Figure 8.

Bleomycin-induced fibrotic lung injury is worsened by a high-phosphate diet. (A) Experimental timeline of 7-week mouse diet and saline or bleomycin intratracheal installation with analysis after 3 weeks of bleomycin exposure. (B) Serum creatinine, BUN, phosphate, and intact FGF23 concentrations and (C) total cell count, macrophage, lymphocyte, and neutrophil counts from BAL fluid of experimental mice. (D) Hydroxyproline quantified from whole-lung homogenates and (E) serum transforming growth factor-β (TGFB) concentrations of different experimental mouse groups. (F) Relative transcript of Col1a1 and Tgfb in whole murine lung homogenates from each group. All values are mean ± SEM with *P ≤ 0.05, **P ≤ 0.01, and ***P < 0.001; n = 4–7 mice/group. Statistical significance was analyzed via two-way ANOVA followed by Tukey’s multiple comparison post hoc test for groups depicted with significance bars.

Similar to the mice harvested 14 days after bleomycin, the lungs of those harvested 21 days after bleomycin instillation exhibited an increase of ACTA2 expression in the bleomycin groups. IL1B expression was also primarily limited to regions of the lung also expressing ACTA2 (Figure 9A), but differences in the amount of IL1B between CB and PB groups were not as obvious at this later time point, mimicking the BALF inflammatory cell data (Figure 8C). Masson’s trichrome (Figure 9B) and Picrosirius red–stained (Figure 9C) murine lung sections revealed altered lung pathology in the CB group compared with the CS and PS groups, confirmed by decreases in mean linear intercept (Figure 9D) and an increase in the red-stained area for each imaged field (Figure 9E), consistent with signs of restrictive lung disease that accompany pulmonary fibrosis (61, 62). These observations were exacerbated when the mice were also on a high-phosphate diet, compared with mice in the CB group (Figures 9B–9E). Collectively, these data suggest that hyperphosphatemia worsens bleomycin-induced fibrotic lung injury in preclinical models of interstitial fibrotic lung disease.

Figure 9.

Bleomycin-induced fibrotic lung pathology is exacerbated by a high-phosphate diet. (A) Representative pathology of ACTA2 (red) and IL1B (green) immunofluorescence-stained lung sections from each mouse group. Scale bars: A, 100 μm (top; 20×); 200 μm (bottom; 40×). (B) Representative pathology of trichrome-stained and (C) Picrosirius red–stained lung sections from experimental mice. Scale bars, 100 μm. (D) Quantification of L_m from trichrome-stained lung sections and (E) percent Picrosirius red–stained area per field from adenine and control diet experimental mice. Values from three separate fields were averaged for each animal. All values are mean ± SEM with *P ≤ 0.05, **P ≤ 0.01, and ***P < 0.001; n = 4–7 mice/group. Statistical significance was analyzed via two-way ANOVA followed by Tukey’s multiple comparison post hoc test for groups depicted with significance bars.

Discussion

In this study, we describe the association between CKD and lung injury as well as identify a novel underlying signaling pathway that points to elevated phosphate as a potential mediator of pathological kidney–lung crosstalk. In addition, we characterize pulmonary fibroblasts as important effector cells in this pathway.

There are a few existing studies, primarily clinical, that demonstrate a link between CKD and RLD, with IPF as a primary focus (10, 11). CKD seems to not only make subjects more susceptible to IPF but also worsen the prognosis of patients with IPF, as their glomerular filtration rate decreases with the progression of kidney failure (11, 63). Current clinical research of the lung–kidney axis focuses on treating fibrosis in both organs independently (64), but investigating alterations in mineral metabolism for mechanism and treatment has been left relatively unexplored. Therefore, it is of utmost importance to further identify the underlying pathomechanisms of this multiorgan crosstalk to ultimately develop therapies that can address both CKD and lung disease, which increase in incidence with age (65, 66) and will continue to increase in incidence overall in the future (67, 68).

FGF23 has been widely described as a prognostic biomarker for CKD and some of its associated comorbidities (69), and we have previously shown that FGF23 exerts diverse effects in chronic inflammatory lung diseases, COPD, and cystic fibrosis (34, 70). In the case of COPD, FGF23 can directly affect the bronchial epithelium, inducing IL1B secretion via FGFR4 signaling (34). Interestingly, our recent report shows that FGF23 does not exert those same effects on the non-COPD bronchial epithelium and, on the contrary, improves the proinflammatory response elicited by phosphate stimulation (9). In relation to pulmonary fibrosis, we have shown previously that, although FGF23 plasma concentrations are elevated in individuals with IPF, FGF23 stimulation of pulmonary fibroblasts seems to have a compensatory effect that leads to a reduction in profibrotic genes when its coreceptor, α-Klotho, is sufficiently expressed (33). These previous results were the impetus for our current study, which specifically investigated circulating factors present in CKD that may affect the lung and mediate the proposed kidney–lung crosstalk. Confirming results from another group, who briefly describe fibrosis in the lungs of adenine-induced CKD mice (48), our slightly modified 14-week diet-based CKD model also revealed profibrotic gene upregulation and histopathological changes in lung tissue consistent with signs of restrictive lung disease that accompany fibrosis (Figures 1 and E1). We have previously described significantly increased inflammation in the lungs of these mice as well, compared with those without kidney disease (9). Interestingly, FGF23 itself did not have any effect on lung fibroblasts (Figure E2), but phosphate led to a robust proinflammatory response, which was exacerbated in fibroblasts from patients with IPF (Figure 2). Although phosphate itself did not induce changes in genes associated with fibrosis in IMR-90 lung fibroblasts, we did observe higher ACTA2 expression in fibroblasts derived from patients with IPF after phosphate exposure, compared with non–IPF-derived fibroblasts (Figure 2C). Furthermore, inflammation is an important precursor and accomplice to pulmonary fibrosis (71), meaning that the phosphate-induced inflammation observed in our study may condition these cells to be more susceptible to additional fibrotic stimuli and could explain why adenine-induced CKD mice have increases in pulmonary Col1a1 and Acta2 mRNA expression. In addition to IPF itself, restrictive lung diseases include various fibrotic interstitial lung diseases that have been shown to exhibit an inflammatory phase often during initial injury or environmental exposure and with continued exposure progress to fibrosis (72, 73).

To further characterize the pathways behind this response, we found that phosphate led to activation of both AKT and ERK in our fibroblast cell line (Figures 3 and 4), which seem to be parallel pathways and downstream of FGFR1 (Figures 3–5). Interestingly, there seems to be a complex crosstalk between AKT and ERK, because inhibition of ERK abolished the phosphate-induced increase in proinflammatory cytokines but increased AKT phosphorylation, whereas AKT inhibition itself also led to an attenuation of IL1B, IL6, and IL8. One explanation is that AKT could be interacting with ERK either directly or indirectly, and inhibition of ERK induces a positive feedback loop, which was not strong enough to induce inflammation of the specific markers we have tested. Other studies with similar findings have speculated that suppression of ERK signaling can cause AKT activation through phosphorylation of other mediators (such as HER2 Thr701) that we do not explore in our current study (74, 75). Further studies could investigate a larger set of pro- and antiinflammatory markers or targets in this signaling pathway, which might be regulated differently.

Future opportunities exist to investigate whether phosphate acts intra- and/or extracellularly to activate FGFR1 in lung fibroblasts. Previous research suggests that phosphate can interact both directly with FGFRs and with downstream proinflammatory pathways independent of FGFRs in renal cells (76), leaving opportunities to further explore these interactions in the lung. We have also shown here that ERK/AKT activation requires phosphate to enter the fibroblasts because of the inhibition of both by a phosphate transporter inhibitor (Figure 5).

This manuscript describes, for the first time, this kidney–lung crosstalk in vivo, characterizing phosphate as a key mediator in increasing susceptibility to lung injury in preclinical models of inflammatory and fibrotic interstitial lung disease. Interestingly, this model does not exhibit CKD, a limitation of the study; however, inducing secondary lung injury after establishment of CKD in mouse models would risk premature death, as the injury to both organs is severe. Therefore, a high-phosphate diet was used in combination with bleomycin to study both the proinflammatory and fibrotic phase of the bleomycin model under conditions mimicking CKD-associated mineral dyshomeostasis. Our results showed that phosphate, in combination with bleomycin, led to both increased proinflammatory (Figures 6 and 7) and profibrotic gene expression (Figures 8 and 9), depending on the length of bleomycin stimulation.

In summary, this study is one of the first studies to characterize a pathological kidney–lung crosstalk linking CKD with susceptibility to lung injury and lung fibrosis and characterizes phosphate as a key mediator, in addition to describing the associated underlying signaling pathway. Further confirmatory studies will be needed to clarify our findings, but this is one of the first studies to characterize this pathological kidney–lung crosstalk and potentially identify therapeutic targets for the future to treat or even prevent CKD-associated restrictive lung disease. Several therapeutic strategies exist (with more in the making) (77) to lower serum phosphate concentrations with the goal of protecting the heart and vasculature in patients with CKD (78, 79), and this study suggests that these phosphate-lowering therapies could be important in protecting the lung as well.