Introduction
The process of aging is a multifaceted and widespread occurrence that is shaped by a combination of environmental factors, stochastic events, genetic variations, and epigenetic changes across various cell types, tissues, and their intricate interplays throughout an individual's lifespan [
1‐
3]. At the same time, aging always goes with persistent and low-grade inflammation as well as metabolic disorder, both of which are considered to take critical roles in age-related pathologies [
4,
5]. Nevertheless, the precise etiology, particular pathological effect, and potential role in contributing to adverse conditions remain unknown.
Different from acute inflammatory process in response to harmful insults such as traumatic tissue injury or an invading pathogen, chronic inflammation during aging tends to be of low grade and persistence, leading to tissue degeneration finally. Possible mechanisms behind the special process includes (1) abnormal production of cytokines from activated immune cells and damaged nonimmune cells [
6,
7]; (2) the modulation of “anabolic signaling”; for example, metabolic synthesis induce production of tumor necrosis factor-α and IL-6 to lower metabolic synthesis back [
8,
9]. The contributions of above studies require further study and there is still less acknowledgement about the complete mechanisms and pathways.
Disturbed metabolic process (i.e., obesity, dyslipidemia, glucose intolerance, insulin resistance) are common during unhealthy aging, which are featured by ectopic fat accumulation and abnormal lipid metabolism. Although there is agreement that familial aggregation and cluster are notable characteristic of these conditions, much debate remains unsettled over its pathogenesis. Metabolic perturbances may closely involve in aging development as metabolic changes like increased glycolysis and disturbed lipid metabolism are present in senescence [
10,
11]. In addition, lipid metabolism dysfunction caused by epigenetic changes may accelerates aging [
12] and the impairment of genes involved in lipids synthesis extends lifespan.
Fibroblast growth factor 21 (FGF21) initially attract interests as a great metabolic regulator of glucose and lipids metabolism. As evidence shown, FGF21 is reported to prevent fatty liver formation and insulin resistance in obese mice model and numerous studies showed FGF21 analogue reduced fat mass and alleviated hyperglycaemia, insulin resistance, dyslipidaemia in NASH [
13,
14]. Furthermore, when subjected to a ketogenic diet, mice lacking FGF21 (FGF21 KO mice) exhibited significant fatty liver accumulation and compromised fatty acid oxidation [
15]. With the deepening of the research, FGF21 was reported to possess the excellent abilities of anti-inflammation [
16‐
18]. Although the role of FGF21 in the regulation of inflammatory process and metabolic process has been well elucidated in previous studies, the effect of FGF21 deficiency on aging remain unsettled. Regarding that disrupted metabolism and sterile inflammation are significant phenotype of aging, hypothesis was put forward that FGF21 may take a critical role in aging as most if not all aging-involved diseases share an inflammatory and disturbed-metabolic signature. Previous study shows that circulating levels of FGF21 increase with aging in both rodents and humans [
19]. Though report have been made in clarifying the altered level of FGF21 in aging, the targeted organs, histological changes and pathological mechanism still need exploration.
After genomics, transcriptomics and proteomics, metabolomics emerged as a new approach to characterize metabolic profiling of biological processes [
20‐
22]. When an individual life is in response to physiological stimuli as well as pathological insults, or influenced by genetic or environmental factors, the metabolic profile changed. Although the FGF21 is a key metabolic-regulatory factors, the reports about specific metabolic events of disease pathogenesis and progression remain limited. To fill in the gap, a targeted metabolomics profiling and enriched pathway analysis on liver samples was performed to identify hepatic metabolic alterations and investigate the potential effect induced by FGF21 deletion. Exploring metabolic changes enables the screening of potential biomarkers or therapeutic targets in FGF21-lacking conditions [
23]. Therefore, our study not only deepens our comprehension of the role of FGF21 during aging but provides inflammatory and metabolic mechanism underlying it.
Discussion
Fibroblast growth factor 21 (FGF21) is an endocrine member of the FGFs family. Huge numbers of studies have revealed that FGF21 treatment can alleviate many age-related metabolic disorders [
26] and exhibit anti-inflammatory property in multiple diseases like obesity [
27], bacterial infection [
28] and neurodegeneration [
29]. Those aging-related phenotypes is always accompanied with metabolic dysfunction and long-term inflammatory condition, which may be a potential factor for cell senescence. Consequently, we wondered how FGF21 takes a role in aging process under normal balanced lifestyle.
FGF21 knockout mice and littermate control mice were kept under the same feeding and environmental conditions. This study unprecedentedly reported that compared with littermate control mice, FGF21 KO mice aged at 36–40 weeks had elevated inflammatory cytokines in perivascular serum alongside with substantial pathological changes of organs such as spontaneous inflammatory response in lung and abnormal accumulation of fat in liver. But the phenotypes mentioned above are not observed in young mice of 4–6 weeks between two groups, which indicated that it takes long time for FGF21 to take a role in aging. Above results may account for the inflammatory conditions and metabolic disorder during aging from genetic perspectives as well.
Other results also support that FGF21 take an anti-inflammatory role in degradative diseases [
30]. One study has showed that FGF21 attenuates neurodegeneration by reducing neuroinflammation and oxidant stress through regulating the NF-κB pathway and AMPKα/AKT pathway, which enhances the protective effect on mitochondria in neurons [
29]. Another report identified the conclusion that FGF21 is essential to counteract the renal inflammation during aging on a low-protein diet [
31]. Those results make similar conclusion that FGF21 inhibited inflammation in aging-related diseases, which is consistent with our study. We further figure out whether exactly lack of FGF21 results in chronic and long-term inflammatory response in murine model. Regarding the limitation that the half-life of FGF21 is short and aging model is chronic, we further investigated the ability of FGF21 to respond to external inflammatory shock caused by LPS using FGF21-deletion mice and exogenous administration of FGF21. Result showed that FGF21 exert significantly anti-inflammatory properties, and the loss of FGF21 further aggravated the LPS-induced inflammatory phenotype. Our study deepens our understanding of FGF21 against inflammation in senescence.
Regarding the substantial inflammatory response of lung, transcriptome was further performed in this study to explore the mechanism of FGF21 on pulmonary inflammation. Result indicated that the enriched pathways were cell-cell adherens junction, cadherin binding involved in cell-cell adhesion, and focal adhesion, all of which was part of barrier integrity. Experiments on murine lung were made to measure the pulmonary permeability and results showed that FGF21 administration alleviated leakage caused by LPS-induced inflammation. Western blot of occludin also supports above conclusion. Recent findings also demonstrated an active role of FGF21 in barrier repairing, FGF21 was reported to decrease retinal permeability via regulating cellular tight junction, Chen
et al. reported rhFGF21 preserved blood-brain barrier integrity through activation of PPARγ [
17,
32‐
34]. Parallel investigation concluded that FGF21 improved LPS-induced HPMEC dysfunction and inflammatory response through SIRT1-mediated NF-κB deacetylation
in vitro, which was consistent to this study.
As KEGG enriched analysis annotated, MAPK signaling pathway may take part in the effect of FGF21 on lung inflammation and hyperpermeability. The MAPK family mainly consists of Erk1/2 MAPK, JNK MAPK and P38 MAPK [
35]. Our data suggested that FGF21 knockout upregulated the phosphorylation of JNK in murine lung while the phosphorylation of Erk1/2 and P38 showed no significant alterations. JNK MAPK signaling pathway is reported to exert biological functions such as endothelial inflammation [
25,
36] and cellular tight junction (Q. [
20‐
22]). The findings from the immunoblot analysis indicated that FGF21 suppressed the activation of JNK triggered by LPS. The study by Kang
et al. demonstrated that the transplantation of FGF21-secreting adipose-derived stem cells suppressed the phosphorylation of JNK in the liver, aligning with our findings [
37]. Nevertheless, the precise mechanisms underlying the relationship between FGF21 and JNK remain inadequately understood, necessitating further comprehensive investigation.
Previous studies reported that occludin degradation activated by the E3 ubiquitin ligase Itch is a critically regulatory mechanism in lung [
38]. Itch was considered as a substrate of JNK [
39]. JNK possibly facilitates occludin degradation via the E3 ubiquitin ligase Itch, which will need comprehensive exploration in the future. Our data showed that FGF21 inhibited the phosphorylation of JNK induced by LPS and rescued the downregulated occludin caused by LPS at the same time, which indicated that FGF21 possibly increased the occludin expression via inhibiting E3 ubiquitin ligase Itch by suppressing the activation of JNK.
For another, murine liver developed accumulated storage of fat when FGF21 is lacking in this investigation. Liver is a major metabolic organ for glucose and lipid metabolism and toxic metabolites degradation. In this study, we speculated that FGF21 deletion induced over accumulation of lipids in liver by metabolic perturbance of glucose and lipids. It is well-known that FGF21 is able to regulate metabolism via maintaining glucose tolerance and insulin sensitivity [
40,
41]. A previous investigation determined that lipid accumulation exacerbates liver aging, while FGF21 enhances lipophagy in the liver to decrease lipid accumulation [
42]. FGF21 also has been shown to inhibit hepatocyte senescence through the activation of the AMPK-dependent autophagic pathway [
42] and by regulating macrophage polarization [
43].
In skeletal muscle FGF21 is sufficient to activate muscle atrophy by activating the removal of damaged mitochondria through Bnip3-dependent mitophagy [
44], which possesses anti-inflammatory effects [
44]. Ablation of the mitophagy mediator PINK1 has been shown to induce lipid accumulation and trigger liver steatosis in preclinical models [
45], which could be a strategy for balancing metabolic homeostasis, protecting or halting NAFLD. It is suggested that FGF21 may work coordinately with autophagy to regulate energy homeostasis in the context of aging [
46]. Therefore, the effect of FGF21 on inflammation and metabolism during aging may be closely related to autophagy.
As for glucose metabolism, there is much dynamic balance between glucose and lipid metabolism to support energy balance in body. Surprisingly, the level of hepatic glucose showed no significant difference between FGF21 KO mice and widetype mice in this test, seemingly paradox to FGF21’s glucose-downregulating effect. Previous study indicated that 24-week-old male global FGF21-KO mice are insulin-resistant [
47]. In this study, fructose 6-phosphate and glucose 6-phosphate, which are the key metabolites in glycolysis, are downregulated in FGF21-KO liver. This may be caused by the decrease of HK kinase activity in insulin resistance [
48]. FGF21 knockout should have shown insulin resistance and elevated blood sugar in mice, but the active inflammatory processes also are in demand of high energy consumption, which result in no significant increase of glucose in FGF21 liver. Further regulatory mechanism on glucose and lipid metabolism resulted from genetic deletion of FGF21 is needed.
Although the perturbance of glucose and lipid metabolism is the predominant histological character of all types of liver disease, other less abundant metabolites are also key cellular factors in the development and progression of fatty liver. Firstly, higher level of aminoadipic acid [
49] and 2-methylbutyroylcarnitine [
50,
51] were highly predictive of patients who later progress to metabolic disorder. Especially, aminoadipic acid reduces fasting plasma glucose levels and increases insulin secretion from human islets. This study indicated that aminoadipic acid levels are elevated in liver of FGF21 KO mice, also supporting the hypothesis that increased insulin secretion and insulin resistance exist in FGF21 KO mice. In addition, a clinical study has been reported that compared with healthy individuals, patients with NASH and NAFLD have significantly higher levels of ceramides and other sphingolipid in liver, as well as the result that the ceramide levels are positively correlated with the extent of inflammation and oxidative stress [
52,
53]. Analogously, Theodore
et al. conducted a single-cell metabolomic analysis on the metabolic condition of human hepatocytes lines stimulated by fatty acids, and found that ceramides were highly concentrated in inflammatory steatosis hepatocytes. Those results were validated
in vivo and consistent with previous reports [
54]. Consistent with above studies, ceramide level is increased in FGF21 KO mice when compared with wide type mice in our result, as well as accumulation of fatty acids and TAG and increased cytokines in serum.
Enriched pathway analysis based on those significantly different metabolites were carried out further to explore the specific pathway influenced by FGF21. The glycerophospholipids metabolism, regarded as one of the metabolic pathways to characterize liver injury [
55], is the top enriched pathway in current investigation. The disturbance of glycerophospholipids metabolism could impair hepatic metabolic function, cause steatohepatitis and metabolic syndrome, and finally develop liver fibrosis [
56]. The excessive accumulation of LPC in hepatocytes has the potential to induce lipid toxicity, ultimately impairing hepatocyte metabolism [
57,
58]. Previous study suggested that LPCAT3, the predominant LPCAT in the liver [
59], plays a crucial role in LPC conversion, insulin-sensitivity regulation and systemic metabolism [
59,
60]. The finding that loss of LPCAT3 has been shown to induce FGF21 secretion indirectly highlights a potential connection between FGF21 and LPCAT3 [
60]. In this study, murine liver lacking FGF21 showed lower LPCAT3, seemly accounting for more hepatic accumulation of LPC. Considering that Glycerophospholipids metabolism is regulated by diversity of enzymes and complicated metabolic pathway, feedback regulation between FGF21 and LPCAT3 are possibly present.
Material and Method
Animal Model
4–6 weeks C57BL/6 J male mice weighted at 20-25 g were all purchased from Vital River Laboratory Animal Technology. FGF21 KO mice backgrounded at C57BL/6 J were presented as a gift from Dr. Steve Kliewer (University of Texas Southwestern Medical Center, Dallas, TX, USA). FGF21 KO and WT littermates used in this study were bred from heterozygous mice. All mice were housed at 20–24 °C and fed with normal balanced diet.
LPS exposure was intratracheally injected with dose of 10 mg/kg LPS, while those in the control group were injected with an equal volume of PBS. Human recombinant FGF21 freeze-dried powder (97.9% purity and 124% activity) (D. [
20‐
22,
61,
62]) comes from School of Pharmacy, Wenzhou Medical University. All experimental protocols were approved by the Institutional Animal Care and Use Committee of the Laboratory Animal Center at Wenzhou medical University.
Measurement of Inflammatory Cytokines by ELISA
The perivascular blood sample was collected by removing eyeballs. After centrifugation, the supernatant was collected. The serum levels of IL-6, IL-1β, TNF-α, ICAM-1, VCAM-1 and TGF-β were measured using mouse ELISA kit (Shanghai Boyun Biotech, Co., Ltd., Shanghai, China). All serum samples were stored at − 80 °C until experiment and the procedures were performed following the manufacturer’s instructions. Each sample was analyzed in triplicate wells.
Bronchoalveolar Lavage Fluid (BALF) Analysis
By lavaging the left lung four times repeatedly with 0.2 mL PBS each time, bronchoalveolar lavage fluid (BALF) was collected. Every lavaging was collected and recovery rate was kept above 80%. After centrifuged, the protein concentration was measured by Coomassie brilliant blue Assay and cell counts of BALF were analyzed by automated cell counter.
Wet/Dry Ratio Analysis
After removing the lungs, the wet weight of the right lung from all groups was measured at once. Then all lungs were put in oven at 65 °C for 7 days to get dry wight measurement. Pulmonary edema was calculated as the ratio of Wet-to-Dry weight.
Haematoxylin and Eosin (HE) Staining
Tissue slices from lungs were fixed in 4% paraformaldehyde overnight and embedded in paraffin. Subsequently they were cut into 5 μm using microtome, then deparaffinized, rehydrated and finally stained with haematoxylin and eosin. Images were acquired with a light microscope. The score was based on lung inflammatory condition. Scoring in each picture was based on the percentage of inflammatory area according to the following standard: 0 = no inflammation, 1 = up to 25%, 2 = 25%–50%, 3 = 50%–75%, 4 = 75%–100%.
Immunohistochemistry
Firstly, the lung sections were were cut into 5 μm using microtome, deparaffinized and rehydrated. With incubation at 3% H2O2 at 37 °C for 10 min, the endogenous peroxidase activity was eliminated. The lung sections were boiled in antigen retrieval buffer containing citrate-hydrochloric acid for 15 min and blocked with 5% normal goat serum (OriGENE Technologies, Inc.) for 30 min. Then the sections were incubated with anti-MPO antibody (22225-1-AP, proteintech) or anti-CD68 antibody (28058-1-AP, proteintech) overnight at 4 °C. The next day, the sections were incubated with biotin-labeled secondary antibody working solution (1:200, A0277, Beyotime, Shanghai, China; Goat anti-rabbit IgG-HRP) at 37 °C for 30 min. The morphology of lung slides was scanned with Nikon inverted microscope (Nikon, Tokyo, Japan).
Evans Blue Permeable Assessment and FITC-dextran Permeable Assessment
One hour before sacrifice, FITC-dextran (70 kDa; 10 mg/kg, Sigma-Aldrich) or Evans blue dye (25 mg/kg, BBI solutions) was injected intravenously via tail vein. Lungs were perfused through cardiac lavage with PBS and were put in oven at 60 °C for 24 h. Lung tissues were weighed and soaked in methanamide (0.03 mL/m) for 24 h at 60 °C and then centrifuged. Absorbance at 620 nm and 740 nm of the supernatant was measured. Adjustment was calculated on the following formula: corrected absorbance at 620 nm = actual absorbance at 620 nm − [1.426 × (absorbance at 740 nm) + 0.03].
After FITC-dextran injection for 1 h, lung tissues were perfusing and frozen by liquid nitrogen. After cut into 8 μm, the leaking FITC-dextran was recorded by Confocal laser scanning microscopy (Nikon, Tokyo, Japan).
Immunofluorescence Detection
The lung immunofluorescence frozen sections were used for immunofluorescence after fixed by 4% formaldehyde. Then slides were blocked with 5% donkey serum one hour at room temperature and incubated with anti-occludin antibody (1:100, Huabio, R1510-33) overnight at 4 °C. After incubated with secondary antibodies anti-rabbit IgG-594 (1:100, abcam, ab150076) And nuclei were stained with 4,6-diamidino-2-phenylindole (DAPI). Fluorescence images were recorded by confocal laser scanning microscopy (Nikon, Tokyo, Japan).
PCR Array
Total RNA extraction was performed with Trizol reagent (Takara). OD260/OD280 ratio greater than 1.8 were set for standard sample selection. Then cDNA was prepared according to iScript cDNA Synthesis kit (Bio-Rad, USA) following the manufacturer’s instructions. Then cDNA was added into a mouse PCR arrary plate (Wcgene Biotech, Shanghai, China) to detect gene expression profiling. The qPCR was performed using the Real-Time PCR System (CFX96 Real-Time System, Bio-Rad, USA), and the total experiment was performed according to the manufacturer's protocol. Data were normalized to the gene expression based on the cycle threshold (CT) values and were used to calculate relative gene expression with the 2−ΔΔCT method. The significantly different genes were selected with fold regulation greater than 2 and P-value < 0.05.
RNA-Sequencing
Total RNA from the lung was obtained for RNA-sequencing library construction and following sequencing. Then a comprehensive transcriptome from all samples were obtained from paired-end sequencing, which was performed on an Illumina Hiseq X-Ten (LC Bio, China) and analyzed by perl scripts. After the final transcriptome was generated, FPKM values (fragments per kilobase of exon per million fragments mapped) by the Cufflinks software was applied to evaluate gene expression levels. The P-value was identified by the FDR (false discovery rate) control method and the differentially-expressed messenger RNAs (mRNAs) were selected with P-value < 0.05 and |log2 (fold change)|> 1, all above was performed by R software.
Cell Culture
Human umbilical vascular endothelial cell (HUVEC) was purchased from ATCC and cultured in an atmosphere of 5% CO2 at 37 °C. Overexpression of FGF21 in HUVEC was carried out by lentiviral infection following the manufacturer’s protocol. Lentiviral vectors with FGF21-green fluorescent protein (GFP) genes (Lv-FGF21) or without FGF21 genes (Lv-CON) were constructed by the Ji Kai Gene Company (Shanghai, China). 3 days after the lentivirus infection, fresh medium containing 1 μg/mL puromycin was added to select stable puromycin-resistant cells. Overexpression efficiency was testified by western blot and PCR. Cells in the corresponding group were treated with LPS at a concentration of 0.5 mg/mL. Activation of JNK MAPK signaling pathway was applied by Anisomycin with dose of 1 μg/mL (Aladdin, A102397).
Real-Time qPCR
Trizol reagent (Takara) was used to extract total RNA from cells. And only samples with an OD260/OD280 ratio greater than 1.8 were used for experiments. The cDNA was prepared according to iScript cDNA Synthesis kit (Bio-Rad, USA) according to the manufacturer’s instructions. And qPCR was performed using the Real-Time PCR System (CFX96 Real-Time System, Bio-Rad, USA). β-actin served as endogenous control. The relative fold change was calculated using the 2–△△CT method. Forward primer ATGGATCGCTCCACTTTGACC and reverse primer GGGCTTCGGACTGGTAAACAT for FGF21. Forward primer CTGGAACGGTGAAGGTGACA and reverse primer AAGGGACTTCCTGTAACAATGCA for β-actin.
Western Blot
Cell and tissue lysates were homogenized and mixed with loading buffer. Samples were then boiled for 10 min. Separate samples were loaded onto a 10% SDS-PAGE gel. Separated proteins were transferred to PVDF (Millipore, Billerica, MA, USA). The membranes were blocked with 5% nonfat milk at room temperature for 1.5 h and incubated with specific primary antibodies overnight at 4°C. All primary antibodies concentration were diluted 1000 times. After incubation with secondary antibodies, the stripes were treated with ECL reagent (Pierce, WI, USA) for detection. The protein signals were quantified with Image J software after visualized by Bio-Rad Electrochemiluminescence detector. The primary antibodies used were as follows: rabbit anti-occludin (1:1000, R1510-33, HUABIO), rabbit anti-ZO-1 (1:1000, ab276131, Abcam), rabbit anti-claudin5 (1:1000, ET1703-58, HUABIO), rabbit anti-E-cadherin (1:1000, #3195, CST), rabbit anti-phospho-JNK (1:1000, ET1609-42, HUABIO), rabbit anti-JNK (1:1000, ET1601-28, HUABIO), rabbit anti-FGF21 (1:1000, ab171941, Abcam), rabbit anti-phospho-Erk (1:1000, #4370, CST), rabbit anti-Erk (1:1000, #4695, CST), rabbit anti-phospho-P38 (1:1000, ab195049, Abcam) and rabbit anti-P38 (1:1000, ab170099, Abcam), mouse anti-LPCAT3 (1:1000, 678821-Ig, proteintech), rabbit anti-β-tubulin (1:1000, #2128, CST) and rabbit anti-β-actin (1:1000, 81115-1-RR, proteintech). β-tubulin or β-actin was used as an internal control.
Measurement of Monolayer Permeability
For endothelial monolayer permeability evaluation, HUVEC (5 × 103 cells/well) transfected with Lv-FGF21 or Lv-CON were seeded on polyethylene terephthalate transwell filters (Corning Costar 3470; Corning Inc.). After reaching confluence, LPS was administrated for 24 h (0.5 mg/mL, Sigma-Aldrich) and then FITC-dextran (10 mg/mL, Sigma-Aldrich) was added to the top well for 1 h. The fluorescence intensity of leaking FITC-dextran in the lower room was measured by multifunctional microplate reader (TECAN, Maennedorf, Switzerland) at 488 nm excitation and 520 nm emission.
The liver samples were harvested and underwent targeted metabolomics profiling analysis. Firstly, Q300 Kit (Metabo-Profile, Shanghai, China) was used to generate all raw metabolites data by UPLC-MS/MS. Then data was processed at iMAP platform. For another, lipid profiling of liver was also generated on a Waters ACQUITY Ultra-Performance LC (UPLC) system. A Waters XEVO TQ-S mass spectrometry and MassLynx 4.1 software (Waters, Milford, MA) were also used to generate and process data. At first, distribution characteristics of metabolite profile between groups was identified by Principal component analysis (PCA) and orthogonal partial least squares discriminant analysis. According to the OPLS-DA model, VIP (variable importance in projection) was obtained and was used for following process. Metabolites with p-value < 0.05 and VIP ≥ 1 were screened as statistically significant metabolites. Following analysis of enriched pathway was performed with selected metabolites.
Statistics
ImageJ (NIH) software was used to analyze protein bands gray scale scanning. GraphPad Prism 7.0 (GraphPad Software, San Diego, CA, USA) was applied to perform statistical analyses. Comparisons between two groups were analyzed by unpaired two-tailed Student's t-test and multiple comparisons were analyzed by ANOVA followed by Bonferroni post hoc test. P < 0.05 was indicated as significance. All data were from at least three independent experiments and presented as the mean ± SD.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.