FGF21 prevents low-protein diet-induced renal inflammation in aged mice

Abstract

Low-protein (LP) diets extend lifespan through a comprehensive improvement in metabolic health across multiple tissues and organs. Many of these metabolic responses to protein restriction are secondary to transcriptional activation and release of FGF21 from the liver. However, the effects of an LP diet on the kidney in the context of aging has not been examined. Therefore, the goal of the current study was to investigate the impact of chronic consumption of an LP diet on the kidney in aging mice lacking FGF21. Wild-type (WT; C57BL/6J) and FGF21 knockout (KO) mice were fed a normal protein diet (20% casein) or an LP (5% casein) diet ad libitum from 3 to 22 mo of age. The LP diet led to a decrease in kidney weight and urinary albumin-to-creatinine ratio in both WT and FGF21 KO mice. Although the LP diet produced only mild fibrosis and infiltration of leukocytes in WT kidneys, the effects were significantly exacerbated by the absence of FGF21. Accordingly, transcriptomic analysis showed that inflammation-related pathways were significantly enriched and upregulated in response to LP diet in FGF21 KO mice but not WT mice. Collectively, these data demonstrate that the LP diet negatively affected the kidney during aging, but in the absence of FGF21, the LP diet-induced renal damage and inflammation were significantly worse, indicating a protective role of FGF21 in the kidney.

NEW & NOTEWORTHY Long-term protein restriction is not advantageous for an otherwise healthy, aging kidney, as it facilitates the development of renal tubular injury and inflammatory cell infiltration. We provide evidence using FGF21 knockout animals that FGF21 is essential to counteract the renal injury and inflammation during aging on a low-protein diet.

INTRODUCTION

Convincing evidence from a number of previous studies has established the long-term beneficial effects of low-protein (LP) diets in fruit flies, insects, and mammals (1, 2). LP diets increase food intake and energy expenditure concomitantly (3, 4) and, through improvements in whole body metabolism, extend life span (5). Studies on mice have established that protein restriction conveys these effects largely through increasing circulating fibroblast growth factor 21 (FGF21), a metabolic hormone that is essential in protein sensing (4, 6, 7). Expression and release of hepatic FGF21 increase rapidly within days after the initiation of LP diets in rodents (8). Many of the initial studies of protein restriction were focused on the liver, adipose tissue (both white and brown), and brain (4, 9, 10). However, much less is known about the potential effects of an LP diet on the kidney over time and the role FGF21 may play in the aging kidney. FGF21 mediates its effects through fibroblast growth factor receptors (FGFRs), which belong to the tyrosine kinase receptor family. Many of the FGFRs, specifically FGFR1c, FGFR2c, FGFR3, and other FGF21-related molecules, such as klotho, are expressed in the kidney (11).

Much of the literature on dietary protein intake and kidney function is largely focused on high dietary protein levels and kidney function/filtration (12–15). It appears, however, that protein restriction does not prevent the decline of kidney function during healthy aging (16) and in fact may even increase overall mortality and cancer-related mortality in adults who are 66 yr and older (17). Interestingly, rodent models and human data show some discrepancies regarding renal disease. Patients with chronic kidney disease (CKD) typically show elevated FGF21 levels and higher serum FGF21 correlates with a poorer metabolic profile, higher inflammatory markers, and increased mortality (18–22). In rodents, FGF21 seems to be protective in models of diabetic kidney disease or lipotoxicity (11, 23, 24), and deletion of FGF21 further exacerbates the associated pathology (24).

In the current study, we set out to evaluate the long-term effects of an LP diet on the kidney in normal, healthy aging mice with or without FGF21. We demonstrate that chronic consumption of an LP diet adversely affects renal tissue and, specifically, tubular integrity in mice. Importantly, we provide evidence that FGF21 is essential in protecting the kidney from the deleterious effects of LP diets, including tubular injury and infiltration of inflammatory cells. Finally, using a high-content transcriptomic approach and rigorous bioinformatic analysis, we show that ablation of FGF21 leads to overactivation of several inflammatory pathways that exacerbate the deleterious effects of the LP diet in the aging kidney.

METHODS

Animal and Diets

All experiments were approved by the Pennington Biomedical Research Center Institutional Animal Care and Use Committee in accordance with the guidelines and regulations of the National Research Council, Animal Welfare Act, and Public Health Service Policy on humane care and use of animals in scientific research. Male C57BL/6J [wild-type (WT)] mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Fgf21 whole body knockout (KO) mice on the B6 background, as previously described (25), were generously provided by Dr. Steven Kliewer (University of Texas Southwestern). Normal protein (NP) and LP diets were formulated and produced by Research Diets, Inc. as previously described (4, 8). The two diets were isocaloric with same fat and varying protein and carbohydrate content. The salt content of the diet (2.59 g/kg) was consistent and was same in both NP and LP diets. Casein was the protein source of both diets, which made up 20% and 5% weight in NP and LP diets, respectively. The following four groups of animals were used: WT_NP, WT_LP, KO_NP, and KO_LP. All animals were group housed (3–4 animals in 1 cage) on a 12:12-h light-dark cycle with ad libitum access to food and water. After acclimation on the NP diet for ∼5 days, half of the WT and FGF21 KO mice (∼3 mo of age) were switched to LP diets. Mice remained on their respective diets throughout the study. At 22 mo of age, mice were euthanized for tissue collection. Trunk blood and kidneys were collected. Kidneys were also weighted and sliced into two halves. One-half of the kidney was fixed in 10% formalin for histological analysis, whereas the other half and the other kidney was flash frozen in liquid nitrogen and then stored at −80°C. At 21 mo of age, spot urine was collected in an empty cage daily for 5 days and then pooled together and stored at −80°C.

Histology/Immunohistochemistry

Kidneys were processed for paraffin embedding after 48 h of fixation. Embedded kidneys were cut into 5 µm sections. Sections were mounted on SuperFrost slides (Fisher Scientific) and deparaffinized before being stained. Sections were stained with 1) Periodic Acid-Schiff staining to evaluate glomerular size, sclerosis, protein casts, and inflammatory cell infiltration; 2) Masson’s trichrome staining for collagen deposits and other trichrome-positive material; 3) picrosirius red staining (with 1% fast green to enhance tissue visibility on background) for fibrosis; and 4) CD3-positive staining using anti-CD3 antibody (Abcam) and biotinylated secondary anti-rabbit antibody (Vector BioLabs). CD3-positive cells were visualized using a peroxidase/DAB kit (Vector Biolabs). Histopathology parameters were evaluated blindly by two independent investigators. The scores and trends were the same; therefore, one investigator’s scores are shown. The detailed criteria are described below. Picrosirius red and CD3-positive staining images were analyzed using ImageJ software, by RGB stacking the images, applying threshold to the staining area, and then analyzing the area percentage (3–4 areas per sample where fibrosis or immune cells were present in a given viewing area).

Cortical Tubular Injury

Tubular injury was assessed semiquantitatively with a scoring scale adapted from various publications (26–28). Briefly, injury was defined as the varying degrees of paracellular spaces, brush-border loss, tubular dilation, loss of tubular and epithelial structure and brush border (sloughing of the epithelium and discernibility of cellular/tubular pattern), morphology of nuclei (round versus oblong), and cast formation. The scale ranged from 0 to 3 with half scores being applicable. A score of 0 represented an instance where <15% of the tubules had a combination of the abovementioned markers. The remainder of the scores followed suit with a percent basis as follows: 1 = 15−30% injured tubules, 2 = 31−50% injured tubules, and 3 = >50% of tubules being injured. In addition to this tubular injury tally, leukocyte infiltration received a separate score with the following system: 0 = >15% leukocyte infiltration, 1 = 15−25% leukocyte infiltration, 2 = 26− 50% leukocyte infiltration, and 3 = >50% leukocyte infiltration. Fifteen to twenty viewing areas per animal were scored at ×20 (Hamamatsu NDP.view2).

Glomerular Injury

For renal damage evaluation, we used previously described methods (29). Briefly, glomerular injury analysis was performed on Masson trichrome-stained scanned slides using NanoZoomer software (Hamamatsu). Glomeruli were scored on a 0–4 scale, where 0 is a healthy glomerulus with no sclerosis. The score of 1 represented 1−25% of mesangial expansion and sclerosis, 2 represented 26−50% of mesangial expansion and sclerosis, glomeruli with 51−75% of mesangial expansion and sclerosis had a score of 3, and a score of 4 was given to 76−100% of mesangial expansion and sclerosis in glomeruli; ∼300 glomeruli per sample were scored.

Urine Albumin and Creatinine Analysis

Urine albumin was determined using a Mouse Albumin ELISA, Albuwell M (Ethos Biosciences, Philadelphia, PA). Urine creatinine was measured by the Creatinine (Urinary) Colorimetric Assay Kit (Cayman Chemical, Ann Arbor, MI).

Serum/Plasma Parameters

Serum FGF21 levels were measured using an ELISA kit according to the manufacturer’s procedures (Mouse and Rat FGF21 ELISA, RD291108200R, BioVendor). Blood glucose levels were determined using a glucometer (Accu Check, Roche, Indianapolis, IN). Blood urea nitrogen (BUN) was measured using a BUN kit (“EIABUN,” ThermoFisher).

Western Blot Analysis

For Western blot experiments, the kidneys were homogenized in lysis buffer containing a mixture of protease and phosphatase inhibitors, 2% Triton X-100, 300 mM NaCl, 20 mM Tris, 2 mM EDTA, and 1% Nonidet P-40. Samples were centrifuged at 10,000 rpm for 15 min at 4°C. Protein concentrations were determined using a DC Protein Assay Kit (Bio-Rad, Hercules, CA). Equal amounts of protein (30 µg) were mixed with SDS sample buffer including 2% β-mercaptoethanol as a reducing agent, boiled for 10 min, loaded and separated on reducing gels, and then transferred to a PVDF membrane. The membrane was blocked in 5% BSA for 1 h. Phosphorylated NF-κB p65 (1:1,000) and total NF-κB p65 (1:2,000) were detected using primary antibodies from Cell Signaling Technology (NF-κB p65 no. 4764 and phospho-NF-κB p65 no. 3033, Danvers, MA) incubated overnight at 4°C and horseradish peroxidase-tagged secondary antibody (GE Healthcare UK Limited; 1:5,000) for 1 h at room temperature. Bands were visualized using an ECL chemiluminescent substrate (ThermoFisher Scientific, Waltham, MA) and X-ray film.

Real-Time PCR

Total RNA was extracted from the kidney cortex using an RNeasy Mini Kit (Qiagen, Valencia, CA). cDNA was synthesized by reverse transcription using iScript (Bio-Rad Laboratories). Gene expression was measured by quantitative PCR (Applied Biosystems, Foster City, CA) using SYBR green (Bio-Rad Laboratories). Cyclophilin was used to normalize the expression of target genes. The primers used were as follows: cyclophilin, forward 5′- CTTCGAGCTGTTTGCAGACAAAGT-3′ and reverse 5′- AGATGCCAGGACCTGTATGCT-3′; kidney injury molecule-1 (Kim-1), forward 5′- TGGCACTGTGACATCCTCAGA-3′ and reverse 5′- GCAACGGACATGCCAACATA-3′; F4/80, forward 5′- TGGGATGTACAGATGGGGGA-3′ and reverse 5′- TCCTGGGCCTTGAAAGTTGG-3′; monocyte chemoattractant protein-1 (Mcp-1), forward 5′- TTCCTCCACCACCATGCAG-3′ and reverse 5′- CCAGCCGGCAACTGTGA-3′; and tumor necrosis factor (Tnf), forward 5′- GGCTTTCCGAATTCACTGGAG-3′ and reverse 5′- CCCCGGCCTTCCAAATAAA-3′. All primers were from Integrated DNA Technologies.

RNA Sequencing and Data Analysis

RNA samples (6 from each experimental group) were first analyzed by an Agilent Bioanalyzer RNA 1000 chip to verify their quality and integrity. The library was then constructed using a Lexogen Quant-Seq 3′ mRNA-Seq Library Prep Kit. Completed libraries were analyzed on the Agilent Bioanalyzer High Sensitivity DNA chip to verify correct library size. All libraries were pooled in equimolar amounts and sequenced on the Illumina NextSeq 500 at 75-bp forward read and 6-bp forward index read. Primary data analysis was performed using Lexogen Quantseq pipeline 2.3.6 FWD on the Bluebee platform for quality control, mapping, and read count tables.

Sequencing depth-normalized count data of samples from the four experimental groups (WT_NP, WT_LP, KO_NP, and KO_LP) were analyzed by principal component analysis (R package “prcomp”) to identify potential outliers. Samples were clustered based on gene expression similarities. One outlier each from the KO_LP and WT_NP groups was removed from the subsequent analysis.

Differential gene expression analysis was performed using the Bioconductor “limma” package after precision weighting of genes and empirical Bayes methods for posterior variance estimation. Pathway enrichment analysis was conducted using a gene set enrichment analysis-based approach to determine the enrichment of a priori defined biological pathways from the Kyoto Encyclopedia of Genes and Genomes repository, obtained from the Molecular Signature Database. Statistical significance of pathway enrichment was ascertained by permutation testing over size-matched random gene sets. Adjustments for multiple testing were performed via control of the false discovery rate.

Ingenuity Pathway Analysis

Differentially expressed genes with nominal P < 0.05 were used for canonical pathway and upstream regulator analyses in ingenuity pathway analysis (IPA; Qiagen, https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis). Comparisons between KO versus WT animals with NP and LP diets and between LP versus NP diets in WT and KO animals were investigated. Heatmaps were constructed based on the top 20 of canonical pathway and upstream regulators (sorted by absolute Z score) in each.

Statistical Analysis

All data are presented as means ± SE and were analyzed using two-way ANOVA with diet and genotype as main effects (Prism 6, GraphPad, San Diego, CA). Significant difference was set at P < 0.05. Tukey’s post hoc analysis was performed when analyzing all data.

Data Availability

Limma-analyzed RNA sequencing data with normalized counts are made available from GEO under Accession No. GSE162967.

RESULTS

LP Diet Adversely Affects Renal Structure in an FGF21-Dependent Manner

Kidneys from all experimental groups were harvested at 22 mo of age, and blood samples were drawn (Fig. 1A). WT mice on the LP diet had significantly higher circulating FGF21 and lower blood glucose levels compared with the NP diet-fed group (Table 1). FGF21 levels were essentially not detectable in FGF21 KO groups (Table 1). Aged mice on the LP diet had significantly smaller kidneys regardless of their genotype (WT or FGF21 KO; Fig. 1B). However, kidney-to-body weight ratios were only markedly lower in the FGF21 KO group on the LP diet (Fig. 1C). None of the experimental groups developed significant albuminuria (as shown by normal or close to normal urinary albumin-to-creatinine ratios in Fig. 1G); but, interestingly, mice on the LP diet showed significantly lower ratios compared with NP diet-fed groups of the corresponding genotype. This could mean improved creatinine clearance on LP diets; however, as we did not measure urinary flow, this can only be speculated. Importantly, both WT and FGF21 KO mice on the LP diet had increased Kim-1 mRNA transcript abundance in kidney tissue compared with kidneys of mice on the NP diet (Fig. 1D), indicating tubular injury. Consistently with this observation, histopathological analyses of kidneys from mice on the LP diet showed tubular damage (evaluated with periodic acid-Schiff and trichrome staining) and tubular scores in both WT and FGF21 KO mice (Fig. 2, A–C). Lack of FGF21, however, markedly exacerbated tubular injury, together with significant leukocyte infiltration (Fig. 2C) and increased fibrosis (as shown in Fig. 2, E and F, by sirius red staining). Interestingly, the effects on glomerular integrity were much milder. Lack of FGF21 seems to increase the number of sclerotic glomeruli or glomeruli with mesangial space expansion regardless of dietary protein content, suggesting a protective role of FGF21 against the aging-related decline of glomeruli function (Fig. 2D). A significant portion of immune cell infiltration was also CD3 positive in FGF21 KO mice, indicating a T cell-driven inflammatory response (Fig. 2, E and G). BUN levels were not significantly different between groups (Table 1) and were all in the low range. This is not surprising with mice being on an LP diet as their urea levels are low (since urea is a protein metabolism byproduct). Thus, BUN is not a good marker to estimate kidney function in such dietary models. Taken together, these data indicate that long-term LP diet feeding in aging mice is not beneficial and, in fact, can accelerate renal injury in an FGF21-dependent manner.

Figure 1.

Experimental design and changes in basic kidney parameters on the low-protein (LP) diet. A: experimental design showing the diets and end points for both wild-type (WT) and FGF21 knockout (KO) mice. B and C: kidney weights (B) and kidney-to-body weight ratios (C) at 22 mo of age. Kidney injury molecule-1 (Kim-1) levels (D), urinary albumin (E), creatinine (F), and albumin-to-creatinine ratio (G) in kidney tissues at 22 mo of age. Values are means ± SE; n = 8–10/group. *P < 0.05 vs. normal protein (NP) diet (two-way ANOVA with a Tukey post hoc test). FGF21, fibroblast growth factor 21.

Table 1.

Body weights and serum measurements

	WT NP	WT LP	FGF21 KO NP	FGF21 KO LP
Body weight, g	45.94 ± 2.77	33.98 ± 1.88*	43.34 ± 1.60	39.11 ± 2.50
Blood glucose, mg/dL	166.42 ± 10.4	128.6 ± 10.45*	183.7 ± 12.01	179.5 ± 10.41
Serum FGF21, ng/mL	0.47 ± 0.11	2.44 ± 0.105*	ND	ND
BUN, mg/dL	13.11 ± 1.56	8.97 ± 0.43	10.64 ± 0.62	13.23 ± 2.92

Data are presented as means ± SE. n = 8–10. *P < 0.05. BUN, blood urea nitrogen; FGF21, fibroblast growth factor 21; KO, knock out; LP, low protein; ND, no data; NP, normal protein; WT, wild-type.

Figure 2.

Lack of FGF21 exacerbates renal pathology and inflammatory cell infiltration on a low-protein (LP) diet. A: summary of histology findings and representative microphotographs of kidney sections stained with periodic acid-Schiff (PAS) stain (top row: ×2, middle top row: ×10, and middle bottom row: ×40) and trichrome stain (bottom row, ×10) showing exacerbated pathology in FGF21 knockout (KO) animals on an LP diet. B−D: quantification of renal injury and pathology scores. Tubular injury scores (B), leukocyte infiltration scores (C), and glomerular injury scores (D) were evaluated from the PAS staining. n = 4/group, ∼30 viewing areas per section or 300 glomeruli per group were evaluated. E: picrosirius red (PSR)/fast green staining (top row) showing fibrosis in red and CD3-positive staining (counterstained with hematoxylin, bottom row) showing the population of T cells within inflammatory infiltrations. Quantification of fibrosis (F) and CD3-positive areas (G). Values are means ± SE; n = 4/group, ∼10–16 viewing areas total selected from within a fibrotic or inflammatory cell infiltration area were evaluated. *P < 0.05 vs. normal protein (NP); #P < 0.05 vs. wild-type (WT) animals on the corresponding diet (two-way ANOVA with a Tukey’s post hoc test). FGF21, fibroblast growth factor 21.

Lack of FGF21 Increases Expression of Inflammatory Markers in the Kidney on an LP Diet

As we observed significant immune cell infiltration in the kidneys of LP diet-fed groups, we next investigated the expression of some of the common inflammatory markers in these kidneys. LP diet in itself did not increase the transcript abundance of the markers TNF-α, MCP-1, or F4/80; however, all of these markers showed significant increases in mRNA expression levels in the FGF21 KO group on the LP diet (Fig. 3, A–C). These data are in agreement with previous observations and collectively indicate that FGF21 plays an important anti-inflammatory role during protein restriction in peripheral tissues including the kidney. To strengthen our analysis further, we compared NF-κB p65 and phosphorylated NF-κB p65 protein levels using Western blot analysis. Indeed, we found that phosphorylated NF-κB p65 levels were significantly increased by the LP diet compared with the NP diet in FGF21 KO mice, whereas no difference was observed in WT animals (Fig. 3D). These data are consistent with the inflammatory phenotype in LP diet-fed FGF21 KO mice.

Figure 3.

Lack of FGF21 potentiates the expression and levels of inflammatory markers on a low-protein (LP) diet. Transcript abundance of mRNA was evaluated using quantitative PCR for tumor necrosis factor-α (TNF-α; A), monocyte chemoattractant protein-1 (MCP-1; B), and F4/80 (C) markers from 22-mo-old kidneys. n = 8–10/group. D: total NF-κB and phosphorylated NF-κB p65 protein levels were measured using Western blot and quantified by densitometry analysis of the bands. Values are means ± SE; n = 5/group. *P < 0.05 vs. the normal protein (NP) diet (two-way ANOVA with a Tukey’s post hoc test). AU, arbitrary units; FGF21, fibroblast growth factor 21; KO, knockout; WT, wild-type.

Transcriptomics Analysis Reveals the Upregulation of Several Inflammatory, Apoptotic, and Fibrosis Pathways in Mice Lacking FGF21 and Fed an LP Diet

To gain further insight into the potential pathways involved and to further verify the role of FGF21 in preventing renal inflammation on LP diets in aged mice, we conducted RNA sequencing followed by pathway enrichment analysis using a gene set enrichment analysis and IPA. In contrast to clustered NP and LP samples in WT mice, they were clearly separated in FGF21 KO mice in principal component analysis, indicating a differential transcriptional profile (Fig. 4A). The LP diet together with the lack of FGF21 showed the highest number of genes dysregulated compared with the NP diet-fed group (Fig. 4B). Next, comparison analysis of canonical pathway and upstream regulators was performed between LP and NP diets with FGF21 KO versus WT animals (Fig. 4C). Ablation of FGF21 inhibited more canonical pathways in LP diet-fed kidneys than NP diet-fed kidneys, including a common inhibition of the nuclear factor erythroid 2-related factor 2 (NRF2)-mediated oxidative stress response. Consistently, inhibition of nuclear factor erythroid-derived 2-like 2 (NFE2L2; NRF2) by deletion of FGF21 was predicted in upstream regulators in both diets (Fig. 4C). When we used comparison analysis in IPA and compared canonical pathways in FGF21 KO mice on LP versus NP diets and WT mice on LP versus NP diets, we found several inflammation-related pathways activated (based on activation Z scores) in the FGF21 KO LP group, including PKC-θ signaling in T lymphocytes, production of nitric oxide and ROS in macrophages, and IL-6 signaling, among others (Fig. 4D). Oxidative phosphorylation, however, was downregulated. Similarly, when we examined upstream regulators in the same comparison, we found activation of mediators in lipopolysaccharide-related signaling, including interferon-γ, TNF, and IL-1B as well as interferon-α, transforming growth factor-β1, and NF-κB in the FGF21 KO group on the LP diet (Fig. 4D). Furthermore, when we specifically focused on inflammatory pathways, we found that ablation of FGF21 had the most prominent effects on such pathways as shown by Kyoto Encyclopedia of Genes and Genomes analysis in Fig. 5, A and B. Several inflammation-related pathways were significantly enriched in FGF21 KO mice on the LP diet compared with those on the NP diet. Finally, we analyzed and compared the expression of inflammation, oxidative stress, apoptosis, and fibrosis-related genes and found many of those, including il18, chemokine (C-C motif) ligand 2 (ccl2), nfkb1, nfkb2, chemokine (C-X-C motif) ligands 10 and 11 (cxc10 and cxcl1), neutrophil cytosolic factors 1 and 2 [ncf1 (p47phox) and ncf2 (p67phox)], caspase-1 and -2 (casp1 and casp2), and jun to be differentially expressed in the FGF21 KO group on the LP diet (Fig. 5, C and D). Collectively, these data demonstrate that, indeed, FGF21 is essential in preventing the undesired long-term effects of LP diet in aged mice, with the emphasis on preventing activation of inflammation and fibrosis in the kidney.

Figure 4.

Low-protein (LP) diet differently affects transcriptional profile of wild-type (WT) and FGF21 knockout (KO) mice. A: principal component analysis showing the separation and distribution of each group based on transcriptional analysis. B: number of differentially regulated genes in the kidneys of each experimental group of mice at various statistical cutoff levels. C: top 20 canonical pathways and upstream regulators (sorted by absolute Z scores) in comparison of normal protein (NP) and LP diets in either WT or FGF21 KO mice. D: top 20 canonical pathways and upstream regulators in FGF21 KO and WT mice on either the NP or LP diet. For both C and D, activated pathways or regulators are shown in blue and inhibited pathways/regulators are shown in red. Nonsignificant changes are shown in gray. n = 6/group. FGF21, fibroblast growth factor 21; nuclear factor erythroid 2-related factor 2 (NRF2).

Figure 5.

Lack of FGF21 exacerbates the involvement of inflammatory pathways in the kidneys of mice on a low-protein (LP) diet. A: gene set enrichment analysis showing the top scoring pathways [false discovery rate (FDR) ≤ 0.1]. Rows refer to biological pathways derived from the Kyoto Encyclopedia of Genes and Genomes or custom annotations, and columns refer to pathway regulation (up- or downregulated) under diet [normal protein (NP), LP]- or genotype [wild-type (WT), knockout (KO)]-based comparisons. Upregulated pathways are shown in blue and downregulated pathways are shown in red. B: gene set enrichment analysis showing a comparison of regulation of inflammatory pathways as a function of diet (NP or LP) in WT and FGF21 KO mice. The negative logarithm of the FDR is plotted on the x-axis and inflammatory pathways are listed on the y-axis. Pathway enrichment −log(FDR) values in LP versus NP in WT mice are shown in blue, and those in KO mice are shown in red. The dashed line indicates the FDR = 0.1 level. The majority of the pathways in LP versus NP diets in KO mice were highly significant [−log(FDR) > 1] in KO LP versus NP but not in WT LP versus NP. Sample-level expression box plots of some of the most prominent inflammatory and fibrosis genes (C) and selected apoptosis and oxidative stress genes (D) in each experimental group. n = 8–10/group. Graphs are boxes with medians. *P < 0.05 vs. the NP diet; #P < 0.05 vs. WT animals (two-way ANOVA with a Tukey’s post hoc test). ccl2, chemokine (C-C motif) ligand 2 (ccl2); cxc10 and cxcl1, chemokine (C-X-C motif) ligands 10 and 11; FGF21, fibroblast growth factor 21; ncf1 (p47phox) and ncf2 (p67phox), neutrophil cytosolic factors 1 and 2; casp1 and casp2, caspase-1 and -2, Tnf, tumor necrosis factor.

DISCUSSION

Research focused on the metabolic and physiological effects of LP diets has undergone a renaissance over the past decade due to the beneficial metabolic effects of an LP diet. A number of groups have shown that LP diets increase energy expenditure, reduce fat deposition, enhance insulin sensitivity, and reduce lipids in rodents and humans (3, 4, 8). Soon after the introduction of an LP diet, the liver responds by increasing transcription and release of FGF21, and several recent studies have shown that FGF21 plays a central and indispensable role in the detection and subsequent responses to protein restriction (4, 6, 8). A particularly important finding was that the increase in FGF21 was essential to the ability of an LP diet to increase energy expenditure and limit fat deposition, and FGF21 produces these effects through a direct action in the central nervous system (6). However, it seems likely that FGF21 is also acting as in other tissues to produce the overall improvements in healthspan resulting from LP diets. Potential actions in the kidney have received essentially no attention and are the focus of the present work, with special emphasis on the potential role of FGF21 in the aging kidney.

Here, we demonstrate that the lack of FGF21 worsens renal injury and inflammation in aging mice on a long-term LP diet. Our findings further highlight the important role of circulating FGF21 levels in metabolic homeostasis during protein restriction. High-protein intake, especially protein from red meat, has been associated with the acceleration of age-related diseases (30–32), where protein restriction and LP diets may offer significant benefits (17). Although protein restriction may slow the progression of preexisting kidney disease (33) (this is still a matter of debate; see Refs. 34 and 35), this may not be the case for healthy, aging kidneys. In fact, some studies have shown no measurable benefit to restricting protein intake in healthy subjects without kidney disease (16). Although in our experiments WT aging mice on LP diet did not develop kidney disease per se, they did show some degree of inflammatory cell infiltration and, importantly, tubular damage by the time they reached old age compared with those on an NP diet. As Kim-1 measurements are a sensitive biomarker for early proximal tubular injury (36, 37), our data show that even normal, WT mice develop some degree of damage by old age during protein restriction. Renal injury and especially infiltration of immune cells and tubular pathology were, however, significantly exacerbated by the lack of FGF21. It is, therefore, likely that normal circulating FGF21 levels are essential to slow the age-related kidney damage that occurs during protein restriction. Our data also strongly suggest that restricting protein intake, at least in C57BL/6 mice, may not offer long-term benefits for the kidney during healthy aging. Similar findings have previously been shown in other organs and tissues, where lack of FGF21 aggravated pathology, such as fibrosis and inflammation. For example, in streptozotocin-induced diabetic mice, lack of FGF21 was shown to exacerbate cardiomyopathy by aggravating cardiac lipid accumulation (38). In the high-fat-fed mouse model, FGF21 improved lipid metabolism and ameliorated hepatic steatosis (39, 40). Similarly, FGF21 KO mice had more severe hepatic steatosis, inflammation, and fibrosis in a diet-induced model (methionine and choline deficiency) of steatotic hepatitis (39). The most likely explanation for the anti-inflammatory effects of FGF21 is the activation of Nrf2 signaling-driven antioxidant mechanisms and, especially, inhibition of the NF-κB pathway (41). The latter was confirmed by both IPA upstream analysis, which predicted that NF-κB is activated by LP in FGF21 KO but not in WT mice, and by measuring phosphorylated NF-κB p65 protein levels. The Nrf2-mediated oxidative stress response was inhibited by deletion of FGF21 regardless of dietary protein content, as shown by the canonical pathway and upstream analysis compared in NP and LP diets in KO versus WT animals in IPA, respectively (Fig. 4C). Our gene expression data corroborate this notion as the LP diet in FGF21 KO animals increased the expression of genes related to organizing subunits of NADPH oxidases (which then can be sources of increased oxidative stress), NF-kB, and genes encoding certain chemokine ligands. Our transcriptomic data further supports these observations and are consistent with previous data from kidney injury models. For example, FGF21 deletion produced the most robust and prominent effect on the transcriptomic pathways examined, with the majority of inflammation-related pathways significantly enriched when FGF21 KO mice but not WT mice were put on an LP diet.

The discrepancies between FGF21 levels during CKD in humans and rodents are puzzling. In patients with CKD or diabetic nephropathy, FGF21 (and FGF23) levels are elevated and increase progressively with the loss of renal function (42, 43). Higher FGF21 levels correlate with higher inflammation, a poorer prognosis, and higher mortality (21). In diabetic nephropathy, higher circulating FGF21 levels correspond to decline of glomerular filtration rate and increased albuminuria (44). The high FGF21 levels are only partially explained by lower filtration and the decrease in renal FGF21 clearance. For instance, under disease conditions, not only the liver but tissues, such as cardiac and skeletal muscle, can also contribute to elevation of FGF21 (29, 45). In contrast to human data, FGF21 is protective in mouse models of diabetic nephropathy (44). Consistent with our data, FGF21 deletion further accelerates pathology and aggravates injury in rodent models of renal lipotoxicity or acute kidney injury induced by cisplatin (24, 46). Another recent study has suggested that FGF21 is required to survive CKD, but at the expense of blood pressure regulation and homeostasis (47). One explanation is that FGF21 may function as an adaptive response to stress and to multiple stimuli. For example, FGF21 may act to counteract lipotoxicity and oxidative stress in obese and diabetic patients, but this adaptive response is in fact not able to counteract and alleviate renal injury.

Our study paves the way for future research into the role of FGF21 in renal protein handling, leaving questions and directions open for future investigation. For example, it is essential to understand the mechanistic kidney-centric cascades orchestrated by FGF21, and especially, what the direct versus indirect effects of FGF21 on the kidney are and if FGF21 has any impact on the renal protein reuptake in the proximal tubules (48). Furthermore, it remains to be determined if supplementation of FGF21 or switching to an NP diet could reverse and improve the phenotype in the aging kidney. Another limitation of the current study is the exclusive use of male mice. As such, it is unclear if the effects of either protein restriction or FGF21 deletion on kidney function would be replicated in females. An interesting previous study focusing on concentrating mechanisms of the kidney in a collecting duct (CD)-specific nitric oxide synthase 1 (NOS1) ablation model showed that female CDNOS1KO mice had significantly higher FGF21 levels, driving their water intake down (49). Importantly, recent work has indicated that female mice are less sensitive to the effects of LP or amino acid diet (50, 51), and therefore it remains possible that these effects we observe in male mice may not directly translate to females.

Perspectives and Significance

Taken together, we provide evidence herein that FGF21 is essential to counteract the renal injury and inflammation that occurs during normal aging in the kidney under protein restriction conditions. Our data suggest that long-term protein restriction may not be advantageous for an otherwise healthy kidney and support the notion that an LP diet does not provide additional benefits for renal outcomes in patients.

DATA AVAILABILITY

Limma-analyzed RNA sequencing data with normalized counts are made available from GEO under Accession No. GSE162967.

GRANTS

This work was supported in part by National Institutes of Health (NIH) Grants R01DK115749 (to K.S.), HL148114 (to D.V.I.), DK105032 and DK121370 (to C.D.M.), DK096311 (to T.W.G.), and F32DK115137 (C.M.H.). S.G. was partially supported by NIH Grant U54GM104940, which funds the Louisiana Clinical and Translational Science Center, and by funding from the National Medical Research Council, Ministry of Health Singapore (WBS R913200076263). This work used the following core facilities at Pennington Biomedical: Animal Behavior and Metabolism, Genomics and Cell Biology and Bioimaging Core, which are all supported in part by COBRE (P20GM103528) and NORC (P30DK072476) center grants (NIH).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

H.F., K.P.S., C.M.H., C.D.M., and T.W.G. conceived and designed research; H.F., L.C.S., K.P.S., and C.M.H. performed experiments; H.F., S.G., L.C.S., K.P.S., C.M.H., D.S., D.V.I., and T.W.G. analyzed data; H.F., K.P.S., C.M.H., D.S., D.V.I., C.D.M., T.W.G., and K.S. interpreted results of experiments; H.F., S.G., L.C.S., and D.V.I. prepared figures; K.S. drafted manuscript; H.F., S.G., C.M.H., D.S., D.V.I., C.D.M., T.W.G., and K.S. edited and revised manuscript; T.W.G. and K.S. approved final version of manuscript.