Long-term effects of a fat-directed FGF21 gene therapy in aged female mice

Abstract

Fibroblast growth factor 21 (FGF21) has been developed as a potential therapeutic agent for metabolic syndromes. Moreover, FGF21 is considered a pro-longevity hormone because transgenic mice overexpressing FGF21 display extended lifespan, raising the possibility of using FGF21 to promote healthy aging. We recently showed that visceral fat directed FGF21 gene therapy improves metabolic and immune health in insulin resistant BTBR mice. Here, we used a fat directed rAAV-FGF21 vector in 17-month-old female mice to investigate whether long-term FGF21 gene transfer could mitigate aging-related functional decline. Animals with FGF21 treatment displayed a steady, significant lower body weight over 7-month of the study compared to age-matched control mice. FGF21 treatment reduced adiposity and increased relative lean mass and energy expenditure associated with almost 100 folds higher serum level of FGF21. However, those changes were not translated into benefits on muscle function and did not affect metabolic function of liver. Overall, we have demonstrated that a single dose of fat-directed AAV-FGF21 treatment can provide a sustainable, high serum level of FGF21 over long period of time, and mostly influences adipose tissue homeostasis and energy expenditure. High levels of FGF21 alone in aged mice is not sufficient to improve liver or muscle functions.

INTRODUCTION

The size and proportion of the aging population is increasing worldwide, with life expectancy expected to continue to increase over the next few decades (1). It is estimated that the aged population will double from 750 million in 2019 to 1.5 billion in 2050 and the proportion of aged people (65 years and older) will increase from 9 per cent in 2019 to 16 per cent in 2050 (2). Despite the increase in life expectancy over the past decades, healthspan has not similarly increased (3, 4). Age is the main risk factor for serious diseases including cancer, cardiovascular disease, and neurodegeneration (5). The majority of aged people suffer from multimorbidity, the consequences of which include disability and functional decline, poor quality of life, and high health care costs (6). One common disease among older people is metabolic syndrome, which is characterized by a co-occurrence of several cardiovascular risk factors including obesity, insulin resistance, glucose resistance, dyslipidemia, and hypertension. (7, 8). Recently, the hormone fibroblast growth factor 21 (FGF21) has garnered attention mainly as a possible therapeutic agent for the treatment of metabolic syndrome (9). Studies have shown that FGF21 decreases hepatic triglycerides, enhances insulin sensitivity, increases energy expenditure, decreases body weight, and regulates macronutrient preference in various physiological and pathological conditions (10, 11, 12, 13, 14).

Due to the important role of FGF21 in energy homeostasis, researchers have recently become interested in using FGF21 gene therapy to treat metabolic diseases. Exogenous administration of recombinant FGF21 protein in different murine models of obesity has been found to reduce adiposity, improve insulin sensitivity, and lower blood glucose and triglycerides (15, 16, 17). However, due to the short half-life and biophysical deficiencies of the native FGF21 peptide (18), several researchers have focused on the use of adeno-associated viral (AAV) vectors to deliver FGF21 in order to assure long-term and sustained protein production in vivo. Jimenez et al found that administration of FGF21 via an AAV vector in ob/ob or high fat diet-fed mice resulted in reductions in body weight, adipocyte size and inflammation, and hepatic steatosis and improvements in glucose homeostasis and energy expenditure for over a year (19).

Previously our lab conducted a study investigating the potential of visceral adipose tissue (VAT)-targeted FGF21 gene therapy for improving metabolic dysfunction in the insulin-resistant BTBR T+Itpr3tf/J (BTBR) mouse under both normal chow diet and diet-induced obesity (DIO) conditions. Liver is the most insulted organ when metabolic dysfunction starts to emerge. To reduce liver burden in a metabolic dysfunctional state, we developed a dual cassette AAV vector which, when packed to engineered hybrid serotype Rec2, selectively transduces adipose tissues while restricting off-target transgene expression in liver (20). A single intraperitoneal (IP) injection of adipose-targeting Rec2-FGF21 dual-cassette vector was able to sustain transduction of FGF21 in the VAT over the course of a 21-week experiment and improved systemic metabolism in DIO BTBR mice, reduced adipose tissue macrophage (ATM) inflammation, altered adipose and hypothalamic gene expression, and altered serum adipokine and inflammation markers (21).

FGF21 is thought to be a longevity factor as life-long overexpression of FGF21 extends lifespan in transgenic mice (22). Davidsohn et al. reported that AAV vector-based antiaging gene therapies to express three longevity-related genes (FGF21, αKlotho, and soluble form of mouse transforming growth factor-β receptor 2 [sTGFβR2]) can treat several age-related diseases (23) raising the possibility that FGF21 may be used to exert metabolic beneficial effect on aging. Based on the promising results from these studies, we hypothesized that implementing AAV-FGF21 gene therapy at old age could mitigate aging-related metabolic decline and exert positive effects on normal aging. In this study, we investigated the long-term effects of a single IP injection of Rec2-FGF21 to old female mice at the age of 17 months with a dose (2×10¹⁰ vg/mouse) substantially lower than the doses commonly used for systemic delivery of AAV.

MATERIALS AND METHODS

Mice and Diet

Aging female C57BL/6 (National Institute on Aging) (NIA) mice were used to investigate the effects of AAV mediated FGF21 expression. The experiment was not replicated. Mice were received from NIA at 10- month-old of age and were fed a normal chow diet (NCD) (11% fat, caloric density 3.4 kcal/g, Teklad) over the duration of the experiment. At 17 months old, mice were randomized to receive IP injections of either Rec2-Empty (carrying the same dual-cassette but no transgene, n=9) or Rec2-FGF21 (n=10) at a dosage of 2×10¹⁰ viral genome (vg) per mouse. Our past studies have demonstrated that using groups of 9–10 mice is adequate to detect a pre-specified effect size. Investigators were not blinded for analysis of outcomes. Two mice from the Rec2-Empty group and one mouse from the Rec2-FGF21 group died naturally over the course of the experiment and were excluded from subsequent analysis. One mouse from the Rec2-FGF21 group had a large mass on the liver at euthanasia and therefore was excluded from analyses of liver mass, liver function panel, liver triglyceride, and muscle function. Ten 12-month-old female mice who received no treatment were used as middle age comparisons for endpoint body weight, tissue weight, and serum FGF21 measurements. Weekly food consumption and body weights were recorded. All mice had ad libitum access to food and water. All mice were group housed (3–5 mice) in standard laboratory environment cages and housed in temperature (22–23°C) and humidity (30–70%) controlled rooms under a 12:12 light:dark cycle. All animal experiments were approved by The Ohio State University Institutional Animal Care and Use Committee.

Recombinant AAV (rAAV) Vector Construction and Packaging.

rAAV vector contains two expression cassettes flanked by AAV2 inverted terminal repeats. The first cassette consists of the CMV enhancer and chicken β-actin (CBA) promoter, FGF21 transgene (NM_020013.4), woodchuck post-transcriptional regulatory element (WPRE), and bovine growth hormone poly-A. The second cassette encodes a synthetic microRNA targeting the WPRE driven by the albumin basic promoter to prevent transgene expression in the liver. The empty control vector lacks a transgene insertion in CBA backbone plasmid. All vectors were packaged into Rec2 capsid and purified by iodixanol gradient centrifugation as previously described (20, 24, 25, 26). Rec2-Empty and Rec2-FGF21 rAAV vectors (2 × 10¹⁰ vg) were administered to mice via IP injections (in 150 μL AAV buffer).

Body Composition

echoMRI was utilized to measure body composition of fat, lean, free water, and total water masses in live mice without anesthesia at 1-, 5-, and 9-weeks post-injection (WPI). Body composition analysis was performed with an echoMRI 3-in-1 Analyzer at the Small Animal Imaging Core of The Dorothy M. Davis Heart & Lung Research Institute, The Ohio State University.

Energy Expenditure

At 27 WPI, mice underwent indirect calorimetry using the Comprehensive Laboratory Animal Monitoring System (CLAMS) (Columbus Instruments, Columbus, OH). Mice were singly housed with ad libitum access to food and water. Mice were acclimatized in the metabolic chambers for 18 hours, then behavior and physiological parameters (heat, O₂ consumption, CO₂ production, respiratory exchange ratio, and physical activity) were recorded for 24 hours at room temperature.

Glucose Tolerance Test

A glucose tolerance test (GTT) was conducted at 22 WPI. Mice were fasted for 16 hours overnight, then injected with glucose solution intraperitoneally (1.0 mg glucose/kg body weight). Blood was collected from the tail at baseline, 15, 30-, 60-, 90-, and 120-mins post glucose injection. Blood glucose concentrations were measured with a portable glucose meter (Bayer Contour Next).

Open Field Test

Mice were subjected to an open field test (OFT) at 14 WPI and 29 WPI. Mice were individually placed in the center of an open square arena (60 cm × 60 cm, enclosed by opaque walls 48 cm in height) and allowed to explore for 10 minutes. During this time, time spent, and distance traveled in both the center and periphery of the arena were recorded and scored by TopScan software (Clever Sys, Inc.). The arena was cleaned with OptiCide between each trial to remove odor cues.

Grip Strength Test and Rotarod Test

To assess motor function, grip strength and rotarod tests were performed. Bilateral forelimb and hind limb grip strength was assessed using a force transducer with a T-bar grip attachment (BIO-GS3, Bioseb, Vitrolles, France). During grip assessments, mice were scruffed and held perpendicular to the direction of the force transducer while both forelimb or hindlimb paws were allowed to grip the bar attachment. Mice were then pulled away from the transducer in plane with the vector of force recording until grip was lost. Three trials each of forelimb and hind limb grip strength were measured, and peak force was quantified in grams. Both absolute values and values normalized to body weight were compared between groups. A rotarod device with automatic timer and fall sensors was used to assess latency to fall as a measure of motor function (Harvard Apparatus). Briefly, mice were placed in lanes of the rotarod device while the rod was slowly rotating at 4 rotations per minute (RPM). Up to 5 mice were tested at a time, each placed in a single lane. Once mice were placed in the lanes, the test was initiated, and the rod accelerated at 20 RPM/min. The duration of time between initiation of the rod acceleration and when each mouse fell off the rod was recorded as latency to fall. Mice performed three trials of the testing separated by a minimum of 1 minute rest, and the average latency to fall was used for comparison between groups.

Tissue Harvest

Mice were euthanized at 33 WPI. Mice were anesthetized by isoflurane and decapitated. Brown adipose tissue (BAT), gonadal white adipose tissue (gWAT), inguinal WAT (iWAT), retroperitoneal WAT (rWAT), gastrocnemius, soleus, and liver were collected and weighed. Tissues were flash-frozen on dry ice and stored at −80° C until further analysis.

Quantitative Real-Time PCR

Total RNA was isolated from BAT, iWAT, gWAT, gastrocnemius, and liver using the RNeasy Mini Kit plus RNase-free DNase treatment (Qiagen #74804). First-strand cDNA was generated using TaqMan Reverse Transcription Reagent (Applied Biosystems #N8080234). Quantitative real-time PCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems #A25742) on a StepOnePlus Real-Time PCR System (Applied Biosystems). Primer sequences can be viewed in Supplementary Table 1. Data were calibrated to endogenous control Actb and the relative gene expression was quantified using the 2 ^−ΔΔCT method (27).

Serum Analysis

R&D Systems DuoSet ELISA kits were used to assay serum FGF21 (#DY3057), adiponectin (#DY1119), and IGF-1 (#DY791). Serum samples were provided to the Comparative Pathology & Mouse Phenotyping Shared Resource at the Ohio State University’s College of Veterinary Medicine for a liver function panel consisting of serum triglycerides, cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), alkaline phosphatase, aspartate aminotransferase, alanine aminotransferase, creatine kinase, lactate dehydrogenase, albumin, globulin, direct bilirubin, total bilirubin, and total protein.

Hepatic triglyceride quantification

Lipids were extracted from liver tissue by chloroform/methanol (2:1 v/v), followed by rinses in 50 mM NaCl and CaCl₂ (0.36 M)/methanol (1:1 v/v) (28). Hepatic triglyceride quantification was performed with a Caymen Chemical triglyceride assay kit (#10010303).

Immunoblotting

BAT was homogenized in ice-cold Pierce RIPA buffer containing 1x Roche Phosstop and Calbiochem protease inhibitor cocktail III, then was subjected to a brief sonication on ice. Tissues lysates were spun at 13,000 rpm for 15 min at 4 °C. The supernatant was collected, and the protein concentration was determined with a BCA protein assay kit (Pierce). Protein from each sample was loaded (7 μg) and separated by gradient gel (4–20%, Mini-PROTEAN TGX, Bio-Rad), then transferred to a nitrocellulose membrane (BIO-RAD). Blots were incubated overnight at 4 °C with the following primary antibodies: Beta-actin (Cell Signaling #4970, 1:1000), UCP1 (Abcam ab10983, 1:3000). Chemiluminescence signal was detected and visualized by LI-COR Odyssey Fc imaging system (LI-COR Biotechnology, Lincoln, NE). Quantification analysis was carried out with Image Studio software version 5.2 (LI-COR Biotechnology).

Statistical Analysis

Data are expressed as mean ± SEM. GraphPad Prism 7 software (GraphPad, La Jolla, CA) was used to analyze our data, using two-tail Student’s t-tests. P <0.05 was considered statistically significant. Data were tested for normality using the Shapiro-Wilk test. If data violated assumptions of normality, either log2 transforms or Mann-Whitney tests were performed, and analysis was repeated. Welch’s correction was performed for data that violated assumptions of homogeneity of variance. An ANOVA mixed effects model was used to analyze all longitudinal data including the energy expenditure, and Bonferroni’s test was used for corrections post-hoc.

RESULTS

VAT-directed FGF21 gene transfer moderately improves metabolic outcomes in aged female mice

Female mice at the age of 17-months were randomized to receive IP injection of Rec2-FGF21 or a control vector carrying no transgene Rec2-Empty (Fig. 1A). Over the course of the 33 weeks experiment, body weight and food intake were measured weekly. Mice injected with Rec2-FGF21 gained less total weight (Fig. 1B) and displayed reduced percent body weight gained as compared to Rec2-Empty injected mice (Fig 1C). Both groups did not consume significantly different amounts of relative food, as measured by grams per body weight per day (Fig. 1D). Mice were subjected to body composition measurements via an in vivo echoMRI at 1 WPI, 5 WPI, and 9 WPI. At 1 WPI there was no difference in percent fat mass (Fig. 2A) or lean mass (Fig. 2B) between the two groups. At 5 WPI there was no difference in percent fat mass (Fig. 1E) between groups, but Rec2-FGF21 mice displayed significantly higher percent lean mass as compared to Rec2-Empty mice at 5 and 9 WPI (Fig. 2B), and significantly lower percent fat mass at 9 WPI (Fig. 2A). A GTT was conducted at 22 WPI. There was no difference in glycemic control between groups, as measured by an ANOVA mixed effects model and area under the curve (AUC) calculation (Fig. 1G).

Figure 1. VAT-directed FGF21 gene transfer decreases body weight gain in aged female mice.

(A) Experimental timeline. (B) Body weight (BW). Empty: n=9, FGF21: n=10. (C) Percent body weight gain. Empty: n=9, FGF21: n=10. (D) Relative food intake. Empty: n=9, FGF21: n=10. (E) Glucose tolerance test (GTT). Empty: n=8, FGF21: n=10. (F) GTT area under the curve (AUC). Empty: n=8, FGF21: n=10. Data are means ±SEM. * P<0.05.

Figure 2. VAT-directed FGF21 gene transfer decreases adiposity and increases lean mass in aged female mice.

(A) Percent fat mass at 1 week post injection (WPI), 5 WPI, and 9 WPI as measured by echoMRI. Empty: n=9, FGF21: n=10 for 1 and 5 WPI. Empty: n=8, FGF21: n=10 for 9 WPI. (B) Percent lean mass at 1 WPI, 5 WPI, and 9 WPI as measured by echoMRI. Empty: n=9, FGF21: n=10 for 1 and 5 WPI. Empty: n=8, FGF21: n=10 for 9 WPI. (C) Tissue weight (brown adipose tissue, inguinal white adipose tissue, retroperitoneal white adipose tissue, gonadal white adipose tissue, gastrocnemius, soleus). Empty: n=7, FGF21: n=9. (D) Relative tissue weight, normalized to body weight. Empty: n=7, FGF21: n=9. Data are means ±SEM. * P<0.05, ** P<0.01, *** P<0.001.

At 27 WPI, mice underwent indirect calorimetry measurements using CLAMs. Body composition is known to have effects on energy expenditure (EE) (29). Heavier animals typically have higher absolute rates of EE, partially due to increases in total metabolically active mass, while lighter animals typically have higher per-kilogram rates of EE (30). At the time of indirect calorimetry, Rec2-FGF21 animals displayed a significantly lower body weight (Fig. 3A). We compared the relationship between EE and body weight for the two groups. Rec2-FGF21 animals showed increased EE per gram of body weight, as measured by an ANOVA mixed effects model, denoting increased energy efficiency over the Rec2-Empty group. (Fig. 3B) Rec2-FGF21 animals also showed a significantly higher average RER overall and during the dark phase (Fig. 3C, D). There were no differences in ambulation between the groups (Fig. 3E, F). Measurements of heat, VO₂, and VCO₂, normalized to body weight and not normalized to body weight can be seen in Supplementary Figure 1.

Figure 3. VAT-directed FGF21 gene transfer increases energy expenditure per gram body weight in aged female mice.

(A) Body weight at time of indirect calorimetry. (B) Energy expenditure. (C) Respiratory exchange ratio (RER). (D) RER AUC. (E) Ambulation. (F) Ambulation AUC. Data are means ±SEM. Empty: n=7, FGF21: n=9. * P<0.05, ** P<0.01.

At sacrifice when the mice were 24-months old, body weight of Rec2-FGF21 animals was significantly lower than Rec2-Empty animals of the same age and a cohort of naïve mice of 12-months of age. (Supplementary Fig. 2). The absolute and relative weights of iWAT, rWAT, and gWAT were significantly lower in AAV-FGF21 mice as compared to Rec2-Empty mice while the relative weight of BAT was significantly higher (Fig. 2C, D). Age effect on tissue weights was observed as 24-month-old mice showed reduced tissue weights of BAT, rWAT, gWAT, and gastrocnemius as well as increased liver weight compared to 12-month-old mice. FGF21 treatment did not alter the age-related changes in liver and gastrocnemius (Supplementary. Fig. 2B–K).

VAT-directed FGF21 gene transfer does not alter measures of anxiety or locomotion in aged female mice

To explore whether long-term overexpression of FGF21 affects behavior, open field tests, classically used to assess exploratory behavior, general locomotion, and anxiety (31, 32), were performed at 14 WPI and 29 WPI, respectively. Open field test is based on the natural conflict between the tendency to explore a new environment and to avoid an exposed open area (33). Alteration in time spent in the center of the open field is considered to reflect a change in anxiety level. Total distance traveled, distance traveled in the periphery/distance traveled total ratio, and distance traveled in the center/distance traveled total ratio were measured. There were no significant differences between groups for any measures at either 14 WPI or 29 WPI (Supplementary. Fig. 3).

VAT-directed FGF21 gene transfer does not enhance muscle performance in aging female mice

Assessment of motor strength using grip testing demonstrated no improvement of forelimb or hind limb strength when compared as absolute force values or force normalized to body weight (Fig. 4A–D). In fact, absolute hind limb strength (Fig. 4C) showed a significant decrease in the 24-month-old Rec2-FGF21 group. Additionally, latency to fall on rotarod testing showed no significant differences between groups (Fig. 4E).

Figure 4. VAT-directed FGF21 gene transfer does not improve muscle function in aged female mice.

(A) Average bilateral forelimb strength. (B) Bilateral forelimb strength, normalized to body weight. (C) Average bilateral hindlimb strength. (D) Bilateral hindlimb strength, normalized to body weight. (E) Average rotarod. Data are means ±SEM. Empty: n=7, FGF21: n=8. * P<0.05.

VAT-directed FGF21 gene transfer leads to high level sustained elevation of circulating FGF21

Serum FGF21 levels were measured in the 12-month-old naïve group, 24-month-old Rec2-Empty group, and 24-month-old Rec2-FGF21 group. A single injection of Rec2-FGF21 resulted in sustained elevation of circulating FGF21 at 33 WPI, approximately 94 folds higher than the Rec2-Empty group (Table 1). Serum FGF21 level was progressively decreased with aging from 0.21 ± 0.05 ng/mL in the 12-month-old naïve group to 0.03 ± 0.01 ng/mL in the 24-month-old Rec2-Empty group (Supplementary Fig. 4). Adiponectin is an adipokine involved in regulating glucose levels and insulin sensitivity (34) and is thought to mediate FGF21’s effect on glycemic control. Serum adiponectin levels were not significantly different between the Rec2-Empty and Rec2-FGF21 groups (Table 1). Leptin is a major adipokine that is positively corelated with adiposity. Unfortunately, serum samples were not enough to assay leptin after liver panel measurement. Insulin-like growth factor 1 (IGF1) is a downstream mediator of growth hormone (GH), which promotes lipolysis and fat oxidation. Serum levels of IGF1 are inversely correlated with age (35). Serum levels of IGF1 between groups were not different (Table 1).

Table 1.

Serum biomarkers.

Serum Measurement	Empty	FGF21	p value
FGF21 (ng/mL)	0.03 ± 0.007	2.83 ± 0.213	<0.0001
Adiponectin (μg/mL)	11.45 ± 0.78	11.01 ± 0.76	0.695
IGF-1 (ng/mL)	21.53 ± 3.912	16.08 ± 4.775	0.139

Empty: n=7, FGF21: n=9. Data are means ±SEM.

VAT-directed FGF21 gene transfer alters gene expression in adipose tissue, but not in muscle

Overexpression of Fgf21 was confirmed in the main targeting tissue gWAT (Fig. 5B). A highly significant increase of Fgf21 mRNA was observed in the non-targeted iWAT (3.4-folds) but at 2–3 orders smaller magnitude compared to that in the gWAT (>1800-folds) (Fig. 5D). Our previous study identified a gene expression signature in DIO mice following fat-directed FGF21 gene therapy (21) including FGF21 receptor genes (Fgfr1, Klb), inflammatory genes (Ccl2, Pai1), ATM markers (Egr2, F4/80), major adipokines (Lep, Adipoq), and thermogenic genes (Ppargc1a, Cidea, Dio2, Ucp1). In both gWAT and iWAT, mRNA expression of Fgfr1 (encoding fibroblast growth factor receptor 1), Lep (encoding leptin), and Adipoq (encoding adiponectin) was significantly lower in Rec2-FGF21 mice (Fig. 5A, C) while other genes profiled were similar. Aging-related muscle functional decline is associated with mitochondrial dysfunction. The gene expression of Fgf21 and its receptors as well as genes involved in mitochondrial functions (Cs, Got2, Mdh1, Pdha1, Sdhb) was profiled in the gastrocnemius. There were no differences in gene expression in gastrocnemius between the two groups (Fig. 5E, F).

Figure 5. VAT-directed FGF21 gene transfer alters white adipose tissue gene expression in aged female mice.

(A) Relative mRNA expression in gonadal white adipose tissue (gWAT). (B) Relative mRNA expression of Fgf21 in gWAT. (C) Relative mRNA expression in inguinal white adipose tissue (iWAT). (D) Relative mRNA expression of Fgf21 in iWAT. (E) Relative mRNA expression in gastrocnemius. (F) Relative mRNA expression of Fgf21 in gastrocnemius. Data are means ±SEM. Empty: n=5, FGF21: n=5. * P<0.05, *** P<0.001, **** P<0.0001.

Overexpression of Fgf21 was also found in BAT associated with significant downregulation of Lep and Adipoq in Rec2-FGF21 mice (Fig. 6A). Aging is associated with a decline of BAT activity (36). BAT plays an important role in thermogenesis, a process primarily mediated by uncoupling protein-1 (UCP1) (37). Although Ucp1 mRNA was not changed (Fig. 6A), the UCP1 protein levels in Rec2-FGF21 mice were significantly higher than that in Rec2-Empty mice (Fig. 6B, C).

Figure 6. VAT-directed FGF21 gene transfer alters brown adipose tissue gene expression in aged female mice.

(A) Relative mRNA expression in BAT. (B) Western blotting of UCP1 in BAT. (C) Quantification of Western blotting. Data are means ±SEM. Empty: n=5, FGF21: n=5. * P<0.05, *** P<0.001, **** P<0.0001

VAT-directed FGF21 gene transfer exerts limited effects on liver

Liver function panel was profiled at euthanasia. Serum HDL was significantly lower in the Rec2-FGF21 group compared to the Rec2-Empty group (Table 2). There were no differences between groups for all other serum markers. To assess whether FGF21 gene transfer alters liver steatosis, hepatic triglycerides were measured and there was no difference between groups (Fig. 7A). Moreover, the absolute and relative liver weight were similar between the two groups (Fig. 7B, C). Liver gene expression profiling found limited changes, only downregulation of Srebp1c (encoding sterol regulatory element-binding protein 1c), an enzyme regulating de novo lipogenesis, (Fig. 7D) and upregulation of Fgf21 (Fig. 7E).

Table 2.

Serum liver function panel

Serum Liver Panel	Empty	FGF21	p value
Triglycerides (mg/dL)	58.6 ± 4.66	52.4 ± 7.51	0.500
Cholesterol (mg/dL)	82.6 ± 6.17	72.1 ± 4.9	0.210
HDL (mg/dL)	41.4 ± 1.84	32 ± 3.32	0.029
LDL (mg/dL)	30.4 ± 5.69	29.7 ± 2.77	0.912
Alkaline phosphatase (U/L)	136.7 ± 24.36	100.7 ± 12.42	0.213
Aspartate aminotransferase (U/L)	340.4 ± 17.87	344.0 ± 17.15	0.888
Alanine aminotransferase (U/L)	75.9 ± 4.96	69.1 ± 4.85	0.352
Creatine kinase (U/L)	10780.0 ± 2017.34	10385.7 ± 10385.7	0.871
Lactate dehydrogenase (U/L)	1528.6 ± 148.89	1463.4 ± 173.31	0.780
Albumin (g/dL)	2.6 ± 0.14	2.7 ± 0.12	0.704
Globulin (mg/dL)	2.5 ± 0.19	2.4 ± 0.12	0.494
Direct bilirubin (mg/dL)	0.4 ± 0.06	0.4 ± 0.08	0.884
Total bilirubin (mg/dL)	0.3 ± 0.05	0.4 ± 0.06	0.500
Total protein (g/dL)	5.1 ± 0.13	5.1 ± 0.21	0.733

Empty: n=7, FGF21: n=7. Data are means ±SEM.

Figure 7. VAT-directed FGF21 gene transfer does not alter liver function in aged female mice.

(A) Liver triglycerides. Empty: n=7, FGF21: n=8. (B) Liver weight. Empty: n=7, FGF21: n=8. (C) Relative liver weight, normalized to body weight. Empty: n=7, FGF21: n=8. (D) Relative mRNA expression in liver. Empty: n=5, FGF21: n=5. (E) Relative mRNA expression of Fgf21 in liver. Empty: n=5, FGF21: n=5. Data are means ±SEM. * P<0.05.

DISCUSSION

In this study, we utilized a VAT-targeting AAV Rec2 vector to investigate the effects of FGF21 overexpression on aging in mice of normal weight. Previously, we used this vector in a non-aged insulin-resistant BTBR model to validate the effectiveness and explore its effects on metabolism. To our knowledge, Davidsohn et al. (23) has been the only group to utilize an AAV-FGF21 vector to investigate the potential of FGF21 on treating symptoms of aging. Our current study differs in the methodology of studying the effect of FGF21 on aging. Davidsohn et al. used several models of age-related diseases using non-aged mice. One experiment in their study examined the effects of AAV-FGF21 on age-related obesity using 18- month-old male mice, but only reported weight change. In addition, the Davidsohn study chose AAV8 serotype to deliver FGF21 gene primarily to the liver because liver is the major source of endogenous FGF21 expression. However, gene therapy targeting the liver exhibits significant toxicity in some conditions (38). As such, we investigated adipose-directed Rec2 vector as an alternative for systemic FGF21 gene delivery. Moreover, this study presented new evidence on the long-term effects of FGF21 overexpression in normal aged C57BL/6 mice to mirror the natural aging process of humans.

Consistent with previous reports on the use of AAV-FGF21 for treatment of obesity (19, 21), we found that animals treated with Rec2-FGF21 displayed a decrease in body weight gain and a decrease in adiposity. This is also consistent with the aging study by Davidsohn et al., which reported that AAV8-FGF21 administration in aged male mice resulted in a weight loss (23). Several studies in mice have found that generally weight increases with age, but then weight declines near the end of the life at advanced age (39, 40, 41). We found that the 24-month-old AAV-Empty animals displayed lower body weight and lower relative weights of rWAT and gWAT compared to middle aged 12-month-old animals. Rec2-FGF21 treatment lowered body weight and relative WAT weight even further. Whether this AAV-FGF21 mediated weight loss and decrease in adiposity can confer beneficial effects on lifespan or health may well depend on age of mice or study subjects. In this study, we did not observe the beneficial effect of long term FGF21 treatment on glucose tolerance, muscle function, and liver function in very old mice. Due to the complex nature of aging and metabolism, we must examine additional measures of aging to assess whether the Rec2-FGF21 treatment is beneficial.

This study is the first to investigate the AAV-based gene transfer to the adipose tissue in aged mice and monitor the long-term transgene expression. We found that the VAT-directed Rec2 vectors led to high level transgene expression sustained at least 7.5 months in aged mice, suggesting low toxicity and low immunogenicity of the Rec2 vector platform. FGF21 overexpression remained primarily in the targeted tissue VAT although some leakiness was observed in the iWAT, BAT, and liver. Our previous studies in young mice and in a shorter duration demonstrated minimal off-target transgene expression in the liver (20, 21, 42, 43, 44). As the dual-cassette design to restrict transgene expression in the liver is dependent on the high potency of the albumin promoter, it is possible that the suppression of transgene expression in the liver becomes less efficient in the aged liver when albumin promoter is weakened.

A single injection of Rec2-FGF21 at a dose of 2×10¹⁰ vg/mouse, resulted in an approximately 94 folds increase in the serum level of FGF21 at 33 WPI. Contrary to several studies in humans up to 80 years old showing that FGF21 serum levels increase with age (45, 46), we found that FGF21 serum levels were significantly lower in our 24- month-old Rec2-Empty group as compared to the 12-month-old group, a 7 folds drop. In this study, we used a dose of Rec2-FGF21 that had been shown to be effective and safe in mice younger than middle age (21). The exceptional elevation of FGF21 in the current study suggests the dose might not be optimal in the context of advanced age. We also observed a downregulation of Fgfr1 in WAT. One study investigating FGF21 in obesity found elevated serum levels of FGF21 accompanied by downregulation of expression of receptor machinery adipose tissue, suggesting a state of FGF21 resistance (39, 47). A study in humans and mice found that serum levels of FGF21 in humans are significantly higher in elderly individuals (≥70 years old) compared to young individuals (≤40 years old), but this is not accompanied by changes in receptor machinery in subcutaneous fat. These results were mirrored in old mice (16 months old) and young mice (5 months old) while data in very old mice were not reported (46). It is possible that the huge increase in FGF21 induced by our vector led to a state of FGF21 resistance.

In addition to reduced body weight gain and adiposity, FGF21 treated animals showed increased energy expenditure as measured by indirect calorimetry, which was further supported by an increase in BAT relative weight. Relative tissue weight of BAT was decreased in 24-month-old Rec2-Empty animals as compared to 12-month-old animals, and FGF21 treatment increased the relative BAT weight to 12-month-old levels, suggesting prevention of age-related decline of BAT. BAT is an important contributor to energy dissipation via adaptive thermogenesis and Ucp1 mediates this process. UCP1 protein level was significantly increased in the BAT of Rec2-FGF21 mice consistent with elevated energy expenditure. Moreover, Ucp1 gene expression in iWAT was trending upward toward significance in Rec2-FGF21 animals. Ucp1 can be increased with the process of induction of thermogenic beige cells in WAT (48). Several studies have found that adipose-derived FGF21 can act in a paracrine/autocrine manner to activate Ucp1 and other thermogenic genes to induce beiging of WAT (49, 50). Other studies have suggested that FGF21 increases energy expenditure by acting via the central nervous system from the sympathetic nervous system to brown adipose tissue (51, 52). It is possible that in our study Rec2-mediated FGF21 overexpression in VAT acts both directly on adipocytes and through the central nervous system to increase energy expenditure. Although precise mechanisms remain to be elucidated, these results suggest that Rec2-FGF21 treatment could mitigate the decreased energy expenditure associated with aging (53).

Despite the changes in adiposity and energy expenditure, Rec2-FGF21 treatment did not improve other markers of aging that we measured. In addition to increased adiposity and decreased energy expenditure, impaired glucose tolerance and insulin sensitivity (54) is associated with the aging process. We performed a GTT and saw no difference between groups for baseline glucose levels or glucose clearance. There could be several explanations for this lack of improvement. “A Toolbox for the Longitudinal Assessment of Healthspan in Aging Mice” (55) recommends using groups of at least 20 mice per group to observe physiologically relevant changes in the GTT. Our groups contained 9 and 10 mice per group due to the availability of aged mice for this study. It is also possible that impairments in glucose tolerance were not severe enough to benefit from FGF21 treatment. An insulin tolerance test (ITT) could have provided more information regarding changes in insulin sensitivity, but we did not want to risk mortality from the stress of conducting both a GTT and an ITT in close temporal proximity.

A decrease in skeletal muscle function and mass is also associated with aging (56). Skeletal muscles are known to express and secrete FGF21, and recent studies have begun investigating the direct and indirect actions of FGF21 on skeletal muscle including regulation of skeletal muscle fiber type and size as well as muscle mass. FGF21 may play a causative role in muscle atrophy in certain pathologies via regulation of mitophagy and autophagy (57, 58). FGF21 is also suggested to regulate glucose uptake and insulin sensitivity in skeletal muscles, both by acting directly on skeletal muscle and indirectly via adiponectin released from adipocytes (59). We performed grip strength and rotarod to test whether overexpression of FGF21 would improve muscle function and found no improvements. In fact, FGF21 treatment decreased average bilateral hindlimb strength. Moreover, FGF21 treatment did not prevent age-related decrease in gastrocnemius weight. Studies have found a positive correlation of serum FGF21 levels and aging-sarcopenia (45, 60). It is possible that Rec2-FGF21 treatment resulted in extreme elevation of serum FGF21 level in aged mice contributing to detrimental effect on skeletal muscle. Further investigations on dosage and timing of the treatment, muscle phenotyping, and molecular signatures are warranted.

Aging is associated with an increased risk of a number of liver diseases, including liver steatosis, which can lead to cirrhosis, fibrosis, and cancer (61, 62). It has been suggested that FGF21 may be a beneficial treatment for liver diseases, as it stimulates fatty acid oxidation in the liver, leading to decreased hepatic fat accumulation. Additionally, FGF21 improves insulin resistance in obese models (63). Our previous study showed that VAT-directed Rec2-FGF21 treatment reversed hepatic steatosis in obese BTBR mice (21). In this study, Rec2-FGF21 treatment showed unremarkable effects on liver. Srebp1c is an important regulator of lipid metabolism in the liver and is associated with the development of liver steatosis (64). Although Srebp1c mRNA was significantly downregulated in FGF21 treated animals, this did not seem to have effects on hepatic triglyceride levels or liver weight. For the serum liver function panel, only HDL was different between groups, with HDL being lower in Rec2-FGF21 animals. Previous studies have shown an inverse correlation between HDL levels and risk of cardiovascular disease (CVD), but more recent studies have shifted to dysfunction of HDL being correlated with risk of CVD (65). Our current data measuring the level of HDL may be insufficient to determine if the HDL is acting beneficially or detrimentally.

The notion that FGF21 is a prolongevity hormone is largely based on the lifespan study in transgenic mice selectively overexpressing FGF21 in the hepatocytes. Lifelong overexpression of FGF21 increases median survival time in mice by ~30% in males and ~40% in females (22). Zhang et al. found that 26–27 months old female FGF21-Tg mice displayed lower body weight, but measures of energy expenditure, body composition, and serum levels of leptin and adiponectin did not differ from the wild-type counterparts. Hepatic triglyceride levels were lower in 26–27 months old FGF21-Tg females compared to wild-type females. Additionally, both male and female 26–27-month-old FGF21-Tg mice displayed dramatically enhanced glucose tolerance and whole-body insulin sensitivity compared to wild-type mice. In contrast to the transgenic mouse model, we found that overexpression of FGF21 in aged female wild type mice increased energy expenditure and decreased adiposity and increased lean mass while there were no differences in glucose tolerance or in hepatic triglyceride levels. We did not examine lifespan in our study, but we measured many markers associated with aging and healthspan. The mixed results are insufficient to either support or rule out an impact on normal aging by overexpressing FGF21 in advanced age. Admittedly, our study was conducted in female mice only and future investigation should enroll both sexes. The differences observed between our study and the Zhang study could be due to several reasons. Zhang et al. overexpressed FGF21 in the liver while our study targeted FGF21 overexpression in the WAT. FGF21 plays different roles in different tissues, so it is possible that FGF21 is eliciting tissue-specific effects in both studies and acting through different mechanisms. Additionally, the animals in the Zhang study were exposed to elevated FGF21 for their entire lives while our study introduced FGF21 overexpression only in old age. Whether lifelong exposure to FGF21 is necessary to increase lifespan remains to be determined (22).

In this study we have found that a single injection of Rec2-FGF21 vector led to sustained overexpressing of transgene in targeted adipose tissues and robust elevation of circulating FGF21 in aged mice, establishing the feasibility of studies on healthspan and lifespan. The effects on markers of aging are mixed when FGF21 gene therapy was implemented at advanced age. Future studies should consider the age of treatment, the dosage, targeting tissues of gene transfer, enrolling both sexes, and expanding phenotypic assessments such as immune-aging in which FGF21 is thought to play a role (66).

DATA AVAILABILITY STATEMENT

Data supporting the results can be found within the published article and its supplementary files.