Role of VGF-Derived Carboxy-Terminal Peptides in Energy Balance and Reproduction: Analysis of “Humanized” Knockin Mice Expressing Full-Length or Truncated VGF

Abstract

Targeted deletion of VGF, a secreted neuronal and endocrine peptide precursor, produces lean, hypermetabolic, and infertile mice that are resistant to diet-, lesion-, and genetically-induced obesity and diabetes. Previous studies suggest that VGF controls energy expenditure (EE), fat storage, and lipolysis, whereas VGF C-terminal peptides also regulate reproductive behavior and glucose homeostasis. To assess the functional equivalence of human VGF_1–615 (hVGF) and mouse VGF_1–617 (mVGF), and to elucidate the function of the VGF C-terminal region in the regulation of energy balance and susceptibility to obesity, we generated humanized VGF knockin mouse models expressing full-length hVGF or a C-terminally deleted human VGF_1–524 (hSNP), encoded by a single nucleotide polymorphism (rs35400704). We show that homozygous male and female hVGF and hSNP mice are fertile. hVGF female mice had significantly increased body weight compared with wild-type mice, whereas hSNP mice have reduced adiposity, increased activity- and nonactivity-related EE, and improved glucose tolerance, indicating that VGF C-terminal peptides are not required for reproductive function, but 1 or more specific VGF C-terminal peptides are likely to be critical regulators of EE. Taken together, our results suggest that human and mouse VGF proteins are largely functionally conserved but that species-specific differences in VGF peptide function, perhaps a result of known differences in receptor binding affinity, likely alter the metabolic phenotype of hVGF compared with mVGF mice, and in hSNP mice in which several C-terminal VGF peptides are ablated, result in significantly increased activity- and nonactivity-related EE.

VGF is a secreted granin protein and peptide precursor that is expressed in neurons throughout the brain and in several neuroendocrine and endocrine tissues (1–3). Homozygous germline VGF knockout mice are lean and hypermetabolic, and resist developing obesity and diabetes when fed a high-fat diet (HFD) (4), suggesting that VGF regulates energy balance by modulating sympathetic outflow. Targeted deletion of Vgf also suppresses obesity, hyperinsulinemia, and hyperglycemia in A^y/a agouti and melanocortin 4 receptor knockout mice (5, 6), supporting a role for VGF in the melanocortin pathway.

VGF proteins are conserved throughout vertebrate evolution (3), although the biological significance of small differences that exist in the sequences of several mammalian VGF proteins has not been studied. Central administration of the VGF-derived peptide TLQP-21 (named by the 4 N-terminal amino acids and length; mouse VGF_556–576) increases energy expenditure (EE) (7), whereas peripheral administration decreases adipocyte diameter (8) and enhances glucose-stimulated insulin secretion (9). VGF signals at least in part, by binding to 2 recently identified TLQP-21 receptors, the G protein-coupled complement 3a (C3a) complement receptor (10), and/or the gC1q complement receptor (11). Here, we use newly generated “humanized” mice (12) to determine whether human and mouse VGF coding sequences are functionally equivalent and whether combined germline ablation of several bioactive C-terminal VGF peptides, including TLQP-62, TLQP-21, and AQEE-30, impacts energy balance and reproductive function in knockin mice.

Materials and Methods

Mouse strains and diets

Humanized VGF knockin mouse lines were generated using mouse Vgf genomic, and human BAC and VGF genomic clones, as previously detailed (12). A 2.2-kb SfiI-SphI fragment that contained human VGF coding, 5′-UTR (untranslated region), and 3′-UTR sequences replaced the 2.3-kb KpnI-XbaI fragment that included mouse Vgf coding, 5′-UTR, and 3′-UTR sequences (Figure 1A). Inserted human sequences encoded full-length human VGF (amino acids 1–615), or after introduction of a single nucleotide polymorphism (SNP) (rs35400704) by site-directed mutagenesis (Mutagenex, Inc), creating a stop codon, encoded a truncated human VGF protein (amino acids 1–524). Targeting constructs were electroporated into 129Sv/J-derived R1 embryonic stem (ES) cells by the Mouse Genetics and Gene Targeting Core Facility, Icahn School of Medicine at Mount Sinai (4). Male chimeras were mated with C57BL/6J females to produce F1 breeders and experiments were performed on N2F1 mice.

Figure 1.

Generation of hVGF and hSNP knockin mice. A, Targeting construct and strategy used to knock the full-length human VGF_1–615 or a SNP-encoded, C-terminally truncated VGF_1–524 coding sequence, flanked by human 5′- and 3′-UTR sequences, into the mouse Vgf locus, is shown. B, Southern blot analysis of the 14.6-kb BglII restriction fragment from the mouse Vgf locus, and the 8-kb BglII fragment from the targeted alleles of hSNP lines 57 and 96, and hVGF line 124. C, Western blot analysis of hippocampal tissue lysates with anti-VGF antisera identified full-length hVGF_1–615 and mVGF_1–617 proteins, and the truncated hVGF_1–524 protein in hippocampal lysates from homozygous male mice. D and E, Quantification of Western blot analyses comparing levels of hVGF_1–615 to mVGF_1–617 (D) and hVGF_1–524 to mVGF_1–617 (E) (*, P < .05, Student's t test), in hippocampal lysates from homozygous male mice. F, Quantification of Western blot analyses comparing levels of hVGF_1–524 to hVGF_1–615 in total brain lysates from homozygous hSNP and hVGF male mice, respectively (*, P < .05, Student's t test).

The VGF-deficient line used here was generated by Regeneron Pharmaceuticals, Inc (13) using 129B6-derived F1H4 ES cells and a BAC-based targeting vector with deletion of the entire Vgf coding sequence and insertion of an in frame lacZ reporter gene and neomycin-selection cassette. Experiments were performed on N2F1 mice (>83% C57Bl6 background). The phenotype of the Regeneron VGF-deficient line (14) is extremely similar to a line generated in our lab using R1 ES cells (4).

The floxed VGF-overexpressing mouse line was generated by inserting a 5′-flanking loxp site into the Vgf 5′-UTR (KpnI site), and a 3′-flanking loxp site and flippase recombinase target (FRT)-flanked neomycin selection cassette, derived from p-loxP-2FRT-PGKneo (Dr David Gordon, University of Colorado Health Science Center), into the Vgf 3′-UTR (XbaI site), using previously described mouse Vgf genomic sequences (4). The construct was electroporated into hybrid 129B6 ES cells (inGenious Targeting Laboratory), and male chimeras were mated with C57BL/6J females to produce F1 breeders having germ line transmission of the targeted, floxed Vgf allele. Experiments were performed on mixed background N3F1 Vgf^flox/Vgf^flox and wild-type mice.

Mice were housed at room temperature in a 12-hour light, 12-hour dark cycle, with chow and water available ad libitum unless otherwise specified. Mice fed control chow (standard chow diet [STD]) received a 10% fat, 70% carbohydrate, 20% protein, 3.85-kcal/g diet, and mice fed HFD received a 60% fat, 20% carbohydrate, 20% protein, 5.24-kcal/g diet (Research Diets), for 12 weeks starting at weaning (21 da of age; n = 8–24 of each sex, genotype per group). In a subgroup, animals were fed HFD or STD for 10 weeks starting at 8–9 weeks of age (n = 5 of each sex, genotype); daily food consumption was measured over 5 consecutive days during the 10th week. Mice were group housed in a standard, nonenriched environment, unless individually housed for metabolic monitoring. All animal studies were conducted in accordance with the Guide for Care and Use of Experimental Animals, using protocols approved by Institutional Animal Care and Use Committees at the Icahn School of Medicine at Mount Sinai, the University of Minnesota, and the University of Cincinnati.

Metabolic cage and body composition analyses

Metabolic monitoring systems were used to evaluate activity, food consumption, and EE, including comprehensive animal metabolic monitoring systems (Columbus Instruments; PhenoMaster/LabMaster, TSE Systems). Mice were individually housed for metabolic testing and were habituated to the metabolic chamber for 1 day before collection of data over 2 days (cohort sizes: n = 8–24 of each sex, genotype per group). For a subset of the feeding-related experiments, mice were group housed (n = 5 of each sex, genotype). VO₂ (volume O₂), VCO₂ (volume CO₂), and respiratory exchange ratio (RER) were calculated from the gas-exchange data. Activity was measured on the x- and z-axes with the use of infrared beams. Live and carcass cohorts were analyzed for total body fat, lean tissue, and body water content using an EchoMRI quantitative magnetic resonance system (Echo Medical Systems).

Calculation of EE components was carried out using previously described methods (15). Briefly, 90 minutes time bins were used to analyze recordings of activity (x-axis beam break counts) and EE (kcal/h) over 24 hours. The bin average for EE was regressed against the bin average for activity for each individual mouse. The intercept for EE was calculated and subtracted from the total EE; this value was classified as activity EE (AEE), whereas the values below the intercept were classified as non-AEE (NAEE).

Glucose tolerance test

After 10 weeks on either a control or HFD, mice were fasted overnight for 16 hours, injected ip with 1.5 g/kg body weight D(+)-glucose, and blood glucose was measured at 0 (immediately before glucose administration), 15, 30, 60, 90, and 120 minutes, via tail sampling using a One Touch UltraMini blood glucose meter (LifeScan).

Western blot analysis

Tissues (white adipose tissue [WAT] and brain) were homogenized and lysed, protein samples (25 μg) separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Millipore), and membranes blocked, as previously described (16). Protein lysates were obtained from iBAT (interscapular brown adipose tissue) by homogenization in 25mM HEPES, 150mM NaCl, 5mM EDTA, 5mM EGTA, 5mM glycerophosphate, 0.9% Triton X-100, 0.1% Nonidet P-40, 5mM sodium pyrophosphate, 1% glycerol, 1mM phenylmethylsulfonyl fluoride, 20-μg/mL aprotinin, 1-μg/mL leupeptin, 0.5mM sodium vanadate, and protease inhibitor cocktail tablet (Roche), and 20-μg samples were resolved on 12% polyacrylamide gels and transferred to polyvinylidene difluoride membranes. Membranes were incubated overnight at 4°C with polyclonal rabbit antisera, diluted in blocking buffer, as described in Table 1. Specificity and use of antiperilipin (17) and antiuncoupling protein-1 (UCP-1) (18) antibodies has been previously described. Membranes were then washed, incubated with secondary antibody, washed, and bound antibodies detected using enhanced chemiluminescence (Thermo Scientific) and quantified with NIH ImageJ, or scanned and quantified using the LI-COR Odyssey (LI-COR), as previously described (16). Band densities were compared by ANOVA and Tukey's post hoc test, or by 2-tailed Student's t test, using Prism 5.0 (GraphPad Software, Inc).

Table 1.

Antibody Table

Peptide/Protein Target	Antigen Sequence (if known)	Name of Antibody	Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody	Species Raised in; Monoclonal or Polyclonal	Dilution Used
Phospho-ACC	Phospho-ACC	Anti-pACC	Cell Signaling Technology, 3661	Rabbit polyclonal	0.001
ACC		Anti-ACC	Cell Signaling Technology, 3662	Rabbit polyclonal	0.001
Phospho-Akt Ser473	Phospho-Akt Ser473	Anti-pAkt Ser473	Cell Signaling Technology, 4058S	Rabbit polyclonal	0.001
ATGL		Anti-ATGL	Cell Signaling Technology, 2138	Rabbit polyclonal	0.001
Phospho-ATPCL	Phospho-ATPCL	Anti-pATPCL	Cell Signaling Technology, 4331	Rabbit polyclonal	0.001
ATPCL		Anti-ATPCL	Cell Signaling Technology, 13390	Rabbit polyclonal	0.001
β-Actin		Anti-β-actin	Abcam, ab3661	Rabbit polyclonal	0.0001
β-Tubulin		Anti-β-tubulin	Cell Signaling Technology, 2146	Rabbit polyclonal	0.001
FAS		Anti-FAS	BD Bioscience, 610 962	Rabbit polyclonal	0.001
FSHβ		Anti-FSHβ	National Hormone and Peptide Program	Rabbit polyclonal	0.0001
GAPDH		Anti-GAPDH	Abcam, ab9485	Rabbit polyclonal	0.0002
Phospho-GSK	Phospho-GSK	Anti-pGSK	Cell Signaling Technology, 9331	Rabbit polyclonal	0.001
GSK total		Anti-GSK total	Cell Signaling Technology, 9338	Rabbit polyclonal	0.001
Phospho-HSL Ser563	Phospho-HSL Ser563	Anti-pHSL Ser563	Cell Signaling Technology, 4139S	Rabbit polyclonal	0.001
Phospho-HSL Ser565	Phospho-HSL Ser565	Anti-pHSL Ser565	Cell Signaling Technology, 4137S	Rabbit polyclonal	0.001
Phospho-HSL Ser660	Phospho-HSL Ser660	Anti-pHSL Ser660	Cell Signaling Technology, 4126L	Rabbit polyclonal	0.001
HSL total		Anti-HSL total	Cell Signaling Technology, 4107S	Rabbit polyclonal	0.0005
Insulin receptor β		Anti-IRβ	Santa Cruz Biotechnology, Inc, sc-711	Rabbit polyclonal	0.002
LHβ		Anti-LHβ	National Hormone and Peptide Progra	Rabbit polyclonal	0.001
Perilipin		Antiperilipin	Dr Andrew Greenberg, Tufts University	Rabbit polyclonal	0.0005
Phospho-mTOR	Phospho-mTOR	Anti-pmTOR	Cell Signaling Technology, 2971L	Rabbit polyclonal	0.001
Phospho-PKA substrate	Phospho-PKA substrate	Anti-pPKA substrate	Cell Signaling Technology, 9621	Rabbit polyclonal	0.001
Tyrosine hydroxylase		Anti-TH	Millipore AB152	Rabbit polyclonal	0.001
UCP-1		Anti-UCP-1	Dr Thomas Gettys, Pennington	Rabbit polyclonal	0.001
UCP-1		Anti-UCP-1	Abcam, ab10983	Rabbit polyclonal	0.001
VGF	AQEE-30 (rat VGF aa 588–617)	Anti-AQEE-30	Stephen Salton	Rabbit polyclonal	0.001
VGF	Human VGF (aa 159–223)	Anti-VGF B-8	Santa Cruz Biotechnology, Inc, sc-365397	Mouse monoclonal	0.01
Rabbit IgG		Antirabbit IgG	LI-COR, IRDye800LT	Donkey polyclonal	0.0001
Rabbit IgG		Antirabbit IgG	LI-COR, IRDye680LT	Donkey polyclonal	0.0001

Abbreviations: ACC, acetyl-CoA carboxylase; ATGL, adipose triglyceride lipase; ATPCL, ATP citrate lyase; FAS, fatty acid synthase; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GSK, glycogen synthase kinase; HSL, hormone sensitive lipase; IRβ, insulin receptor β; mTOR, mammalian target of rapamycin; PKA, protein kinase A; TH, tyrosine hydroxylase.

Quantitative real-time PCR

TRIzol reagent (Life Technologies) was used to isolate total RNA from frozen tissue. iScript cDNA synthesis kit (Bio-Rad) was used to synthesize 4000 ng of RNA. Gene transcription was measured with iQ SYBR Green supermix (Bio-Rad) in duplicate for target genes (Supplemental Table 1 lists primer information) on a CFX96 Real-Time PCR system (Bio-Rad). Gene expression data is presented as normalized linear-transformed values (2-Δcycle threshold) and normalized over expression levels of controls.

Statistics

Values are presented as the mean ± SEM. Comparisons between groups were made with unpaired, 2-tailed Student's t test, or among groups, with ANOVA and Tukey's or Bonferroni post hoc testing. Differences were considered statistically significant at P < .05.

Results

Functional replacement of mouse VGF_1–617 by germline knockin of human VGF_1–615 or a truncated human VGF_1–524 coding sequence into the mouse Vgf locus

To determine whether human VGF was functional in mice and could rescue the lean, hypermetabolic phenotype resulting from germline VGF ablation (4), we knocked full-length human VGF_1–615 into the mouse Vgf locus. Human VGF coding, 5′-UTR, and 3′-UTR sequences replaced equivalent mouse sequences downstream of the mouse Vgf promoter (hVGF line) (Figure 1). We additionally introduced a known SNP (rs35400704) into the human VGF coding sequence, creating a stop codon, which resulted in the translation of a truncated human VGF_1–524 protein that lacked several bioactive C-terminal peptides (hSNP [human VGF_1–524] line) (Figure 1A). Southern blotting (Figure 1B) and PCR analysis confirmed correct gene targeting. Western blot analysis of hippocampal lysates using mouse monoclonal anti-hVGF_159–223 demonstrated that hVGF_1–615 and hVGF_1–524 proteins of the anticipated size were expressed from the targeted alleles (Figure 1C). Expression in hippocampal lysates of hVGF_1–524 protein from homozygous hSNP mice was reduced by approximately 75% compared with levels of mVGF (mouse VGF_1–617) in wild-type mice (Figure 1, C and E), whereas levels of hVGF_1–615 and mVGF were similar (Figure 1, C and D). In total brain lysates, levels of hVGF_1–524 in homozygous hSNP mice were approximately 50% of the levels of hVGF_1–615 in homozygous hVGF mice (Figure 1F).

Homozygous knockin mice expressing human VGF_1–615 or a truncated human VGF_1–524 are fertile

VGF knockout mice are infertile (4), potentially a result of impaired hypothalamus-pituitary-gonadal axis function and/or abnormal fat metabolism. In addition, VGF-derived peptides have also been shown to regulate male sexual behavior and female reproductive function (19–22). We therefore investigated whether hVGF or hSNP mice were fertile by quantifying litter size and frequency for several breeding pairs (Figure 2). Similar litter sizes and number of litters were noted, independent of male or female expression of hVGF_1–615, hVGF_1–524, or mVGF_1–617. In addition, content of pituitary LHβ and FSHβ subunits was similar in homozygous male or female mice expressing hVGF_1–615, hVGF_1–524, or mVGF_1–617. These results indicate that human VGF_1–615 knocked into the mouse Vgf locus functionally replaced mouse mVGF_1–617 and that N-terminal human VGF_1–524, lacking several bioactive C-terminal peptides, was sufficient for reproductive function in mice.

Figure 2.

Homozygous hVGF and hSNP knockin mice have normal reproductive function. Number and size of litters over 6 months (*, previously reported in Ref. 4) for the breeding pairs with the genotypes noted in A. No significant differences in ovary and uterus weights were noted among female 12-week-old homozygous hVGF_1–615, hVGF_1–524, and mVGF_1–617 mice (n = 3 mice per group) (B). No significant differences in pituitary FSHβ and LHβ protein levels, normalized to β-actin as a loading control, were noted in sex-matched, 12-week-old hVGF_1–615, hVGF_1–524, and mVGF_1–617 mice (n = 3 mice per group) (C).

Metabolic analysis of homozygous knockin mice expressing human VGF_1–615 or truncated human VGF_1–524

Mice were weaned at 21 days and were fed either STD or HFD for 12 weeks. Female but not male homozygous hVGF mice showed higher body weights at weaning compared with wild-type mVGF mice, whereas both male and female homozygous hSNP mice showed lower body weights at weaning compared with the other genotypes (Supplemental Figure 1) (mice referred to as hVGF, hSNP, or mVGF are homozygous unless otherwise noted). An interesting gender × diet × genotype interaction emerged from the analysis of metabolic phenotype (Figure 3). Female hVGF showed a significantly higher body weight gain and adiposity compared with mVGF when fed either STD or HFD (Figure 3, right panels). At week 12 after weaning, hVGF females showed an overall increase in fat mass compared with mVGF that was more evident on STD (Figure 3, G and H and K and L). Body weights of female hSNP mice, expressing truncated hVGF_1–524, were lower than hVGF and indistinguishable from mVGF mice (Figure 3, C and D). Despite having the same body weight compared with mVGF, hSNP females had reduced adiposity, in particular on a HFD (Figure 3L), suggesting that removal of C-terminal VGF peptides results in a lean phenotype similar to Vgf−/Vgf− knockout mice (4, 5).

Figure 3.

Metabolic analyses of homozygous male and female mice expressing human VGF_1–615 (hVGF), truncated human VGF_1–524 (hSNP), or mouse VGF_1–617 (mVGF). Body weights (grams) of male and female hVGF, hSNP, and mVGF mice, fed either STD or HFD, were measured weekly for 12 weeks after weaning at 3 weeks of age. A–D, F (24,2040) = 4.15 (P < .00001; **, P at least < .01 vs mVGF). E–L, Body composition of male and female hVGF, hSNP, and mVGF, measured 12 weeks after weaning (n = 8–24 per group). E–H, Expressed as absolute grams of fat or lean mass. I–L, Expressed as % fat mass.

The metabolic phenotypes shown by male hVGF and hSNP fed a HFD (Figure 3, B, F, and J) substantially overlap those of females (Figure 3, D, H, and L). Male hVGF fed a STD showed a transient increase in body weight in the 3rd and 6th weeks after weaning (Figure 3A), and when body composition was assessed at week 12 after weaning, they showed a leaner phenotype than mVGF (Figure 3, E and I). This opposes the phenotype shown on a HFD (Figure 3, F and J) and also that of females fed a STD (Figure 3, G and K). Male hSNP mice also showed a lean phenotype despite their body weight being undistinguishable from the other groups.

Food intake of singly housed mice during metabolic testing on a HFD was significantly higher in 3-month-old hSNP males and females compared with age- and sex-matched wild-type mice, whereas on a STD only, hSNP females ate significantly more than wild-type mVGF mice (Figure 4, A–D). We also measured food intake in group-housed cohorts (n = 5) of male hSNP or wild-type mice, and here noted increased food intake in both HFD and STD in male hSNP mice (Supplemental Figure 2). Analysis of EE revealed that VO₂ and VCO₂ were increased in hSNP compared with wild-type mice, in both the dark, active and light, inactive phases (Figure 4, E–H, and Supplemental Figure 3). EE in hVGF was only increased in male hVGF fed a STD (Figure 4, E–H, and Supplemental Figure 3). Conversely RER was not affected, suggesting similar substrate use for all experimental groups (Supplemental Figure 3). Importantly locomotor activity was increased in male and female hSNP mice compared with wild-type and hVGF mice (Figure 4, I–L).

Figure 4.

Food intake, VO₂, locomotor activity, and components of total EE of homozygous male and female mice expressing human VGF_1–615 (hVGF), truncated human VGF_1–524 (hSNP), or mouse VGF_1–617 (mVGF). A–D, Food intake (genotype: F_2,64 = 15.2, P < .00001; genotype × gender × diet: F_2,64 = 2.4, P = .09). E–H, VO₂ (genotype: F_2,69 = 42.8, P < .00001; genotype × gender × diet × LD: NS). I–L, Locomotor activity (genotype: F_2,69 = 22.2, P < .00001; genotype × gender × diet × LD: F_2,64 = 2.5, P = .08) on STD or HFD. M–P, Average total EE was calculated from calorimetry data. AEE was estimated using the method described in Ref. 15, and NAEE was calculated as EE minus AEE (total EE: genotype, F_2,69 = 60.5, P < .00001; genotype × diet, F_2,69 = 7.3, P < .01; genotype × diet × gender: NS; AEE: genotype, F_2,69 = 60.5, P < .00001; genotype × diet, F_2,69 = 7.3, P < .01; NAEE: genotype, F_2,69 = 25.2, P < .00001; genotype × diet, F_2,69 = 4.8, P < .01; genotype × diet × gender, NS). All parameters were measured in 10- to 12-week-old homozygous male and female hVGF, hSNP, and mVGF mice. Tukey‘s HSD post hoc; *, P < .05; **, P < .01 vs mVGF (n = 8–24 per group).

To get mechanistic insight into the contribution of locomotor activity to EE (Energy Expenditure), we calculated the total EE and estimated the AEE (Activity Energy Expenditure) and the NAEE (Non Activity Energy Expenditure), as described by Virtue et al (15). All hSNP groups showed higher total EE compared with mVGF (Figure 4), a finding that is consistent with increased VO₂ and VCO₂. Globally, the contribution of AEE to total EE was 17 ± 1.5% in mVGF and 18.6 ± 0.6% in hVGF mice, which is in line with previous data (15). The contribution of AEE to total EE drastically increased in hSNP mice to 23.4 ± 0.9%, suggesting a relevant role for activity in the hypermetabolic phenotype of these mice. Importantly AEE was significantly increased in all hSNP groups, with the exception of STD females, compared with controls. Conversely, NAEE was increased only in hSNP fed a HFD but not in hSNP fed a STD. Overall, this analysis suggests that activity-related EE significantly contributes to the increased total EE shown by hSNP mice. The increased NAEE observed only when mice were fed a HFD suggests that other components of EE (either adaptive brown adipose tissue related [see below] or basal metabolism) might play a substantial role as well.

Analysis of glucose homeostasis in homozygous knockin mice expressing human VGF_1–615 or truncated human VGF_1–524

Germline Vgf−/Vgf− knockout mice show increased glucose clearance in oral or ip glucose tolerance tests (4) and increased insulin sensitivity based on hyperinsulinemic euglycemic glucose clamp analysis (6). Glucose tolerance testing of male and female hVGF and hSNP mice fed STD showed similar clearance of an ip glucose bolus by hVGF and mVGF mice, whereas hSNP mice cleared glucose significantly more rapidly than mVGF mice (Figure 5, A and B). No significant differences in plasma insulin levels between homozygous hSNP and mVGF mice fed STD ad libitum were noted in males or females (male mVGF, 0.99 ± 0.23 ng/mL; male hSNP, 0.87 ± 0.19 ng/mL; female mVGF, 0.44 ± 0.06 ng/mL; female hSNP, 0.56 ± 0.06 ng/mL; n = 7 per group; ANOVA with Bonferroni post hoc test). After being fed a HFD for 10 weeks, hSNP mice cleared glucose significantly more rapidly than mVGF mice fed either regular or high fat chow, whereas hVGF and mVGF mice fed a HFD had similar glucose tolerance (Figure 5C).

Figure 5.

Glucose tolerance testing (GTT) of homozygous male mice expressing human VGF_1–615, truncated human VGF_1–524, or mouse VGF_1–617. Homozygous hVGF_1–524 mice (hSNP) or hVGF_1–615 mice (hVGF) and control mVGF_1–617 male mice (mVGF) were fed either STD or HFD for 10 weeks, and GTT was carried out. A and C, GTT comparing regular chow-fed homozygous male and female hSNP to mVGF mice (n = 4 per treatment group). B and D, GTT comparing regular chow-fed homozygous male and female hVGF to mVGF mice (n = 4 per treatment group). Data were analyzed by 2-tailed Student's t test (*, P < .05; **, P < .01; ***, P < .001). E and F, ANOVA with Tukey's post hoc analysis of plasma glucose levels between mVGF and hSNP males fed HFD and between mVGF and hVGF males fed HFD, respectively (*, P < .05; **, P < .01; ***, P < .001; ****, P < .0001).

Analysis of WAT and iBAT protein expression in homozygous knockin mice expressing human VGF_1–615 or truncated human VGF_1–524 is similar to wild-type controls

Increased glucose clearance in germline Vgf−/Vgf− knockout mice is associated with increased circulating free fatty acids, decreased adipose fat mass, and alterations in a number of key lipolytic proteins in WAT, all consistent with increased lipolysis (6, 14, 16). We therefore used Western blot analysis to measure protein expression in WAT from homozygous hSNP and mVGF mice. No consistent alterations in levels and/or phosphorylation of proteins that regulate lipolysis or lipogenesis in WAT were detected in hSNP males or females (Figure 6, A and B).

Figure 6.

Expression of WAT proteins involved in lipolysis and lipogenesis, iBAT UCP-1 protein levels, and skeletal muscle gene expression, in homozygous male and female mice expressing hVGF or hSNP compared with mVGF. A and B, WAT was collected from hSNP (n = 4) and mVGF (n = 4) male and female mice, and protein expression was determined using Western blot analysis as described in Materials and Methods. Values are expressed as fold change in hSNP WAT compared with mVGF WAT, normalized to β-actin as a loading control. C, UCP-1 protein levels in iBAT from male and female homozygous hSNP, hVGF, and mVGF, and heterozygous hSNP/mVGF mice (n = 3 per group) were measured by Western blot analysis as described in Materials and Methods. UCP-1 levels, normalized to β-tubulin as a loading control, are expressed as fold change relative to mVGF levels. Statistical significance (P < .05) was determined using the 2-tailed Student's t test. D, Gene expression in the quadriceps muscle of mice fed a HFD was quantified by QPCR (n = 4–6 per group). UCP-2 mRNA levels were increased in hSNP male mice compared with mVGF; *, P < .05. See Supplemental Table 1 for QPCR primer sequences. ADRP, adipose differentiation-related protein; COXII, cytochrome c oxidase subunit II; Foxo1, forkhead box O1; PDK4, pyruvate dehydrogenase kinase 4; PPARδ, peroxisome proliferator-activated receptor delta; UCP2, uncoupling protein 2.

To determine whether increased UCP-1-mediated energy dissipation in iBAT, in addition to increased locomotor activity, might drive the hypermetabolic phenotype of hSNP mice, iBAT UCP-1 protein levels were quantified in homozygous mVGF, hVGF, and hSNP mice, and heterozygous hSNP/mVGF mice (Figure 6C). No significant changes in UCP-1 levels in any of the groups compared with mVGF were noted. These results suggest that the hypermetabolic phenotype of hSNP mice is not primarily driven by increased energy UCP-1-driven thermogenesis in iBAT but rather by hyperactivity and/or hyperphagia.

Analysis of skeletal muscle gene expression that regulates energy metabolism

In addition to BAT, skeletal muscle is an important site of nonshivering thermogenesis, regulating temperature and energy homeostasis (23). Our data suggest that activity-associated EE plays a significant role in total EE (Figure 4). In the quadriceps muscle of hSNP male mice fed HFD (the group which most consistently demonstrated increased EE compared with controls), we detected a significant increase in UCP-2 mRNA levels and a slight but not significant increase in sarcolipin and cytochrome C oxidase II mRNA levels (Figure 6D). Conversely, no change was observed in Foxo1 (Forkhead box protein O1), PDK4 (pyruvate dehydrogenase lipoamide kinase isozyme 4), PPARδ (peroxisome proliferator-activated receptor delta), or ADRP (adipose differentiation-related protein) mRNA levels in male hSNP mice compared with mVGF mice, and no change in gene expression was observed in female hSNP mice, nor in male or female hVGF mice. Overall, these data suggest that increased EE in male mice lacking VGF C-terminal peptides might at least in part be explained by increased mitochondrial metabolism in skeletal muscle.

Gene dosage effects of VGF on body weight and metabolism

Despite robust effects of complete Vgf gene ablation on body weight, adiposity, and EE (4–6, 14, 16), the metabolic phenotype of mice with a single Vgf allele has not been investigated and is particularly relevant to the interpretation of data obtained from the hSNP line, which in addition to making no VGF C-terminal peptides, synthesize reduced levels of VGF protein compared with wild-type and hVGF mice (Figure 1). We therefore metabolically characterized heterozygous Vgf+/Vgf− knockout, hSNP/mVGF, and wild-type mice. Presence of a single mouse Vgf allele normalized EE, food intake, and locomotor activity in hSNP/mVGF mice, compared with homozygous hSNP mice (Supplemental Figure 4). Interestingly, heterozygous Vgf+/Vgf− knockout mice had reduced body weight and adiposity, increased lean mass/body weight, increased VO₂ (dark phase, d 2 and 4), and normal locomotor activity, compared with wild type (Figure 7, A–C). Lastly, we analyzed homozygous floxed VGF mice that overexpress VGF mRNA and protein by virtue of the placement of the pgk-neo cassette in the 3′-UTR region of the Vgf gene. This leads to premature mRNA termination and polyadenylation using a cryptic poly-A addition site in the inverted pgk-neo cassette, eliminating 3 3′ microRNA binding sites, and resulting in increased central nervous system expression of VGF, approximately 150% wild-type VGF protein levels in hypothalamus (Figure 7D). When fed STD, homozygous Vgf^flox/Vgf^flox male mice weigh more than age-matched wild-type littermates (Figure 7D) and show a small increase in food intake and somewhat reduced locomotor activity, with no change in RER or VO₂ (Figure 7, E–H).

Figure 7.

Vgf gene dosage regulates metabolic phenotype. Male heterozygous VGF knockout mice (Vgf+/Vgf−) that express lower levels of VGF, Vgf^flox/Vgf^flox mice that overexpress VGF, and wild-type (Vgf+/Vgf+) mice, were analyzed for metabolic phenotype. A–C, Analyses of VO₂, locomotor activity, body weight, fat mass/body weight, and lean mass/body weight for Vgf+/Vgf− and Vgf+/Vgf+, measured in 10- to 12-week-old male mice. D, Western blot analysis and quantification of VGF protein levels in hypothalamus at 10 weeks of age and body weights, measured weekly for 7 weeks starting at weaning at 3 weeks of age, for Vgf^flox/Vgf^flox and Vgf+/Vgf+ male mice. E–H, Analyses of food intake, VO₂, RER, and locomotor activity, respectively, measured in male Vgf^flox/Vgf^flox and Vgf+/Vgf+ mice at 10–12 weeks of age. Statistical significance was determined using the 2-tailed Student's t test (*, P < .05; **, P < .01).

Discussion

Functional and evolutionary conservation of human and mouse VGF peptides

Our studies support the functional conservation of human and mouse VGF with respect to the control of reproduction, although small differences in body weight between homozygous hVGF_1–615 and mVGF_1–617 mice may be driven by sequence differences between the species homologues. The next are the 8 major known VGF-derived peptides and their respective amino acid sequence changes between mVGF and hVGF (in square brackets) (3): NERP (neuroendocrine regulatory peptide) 1 [4], NERP2 [1], NERP3 [0], NERP4 [0], TLQP-62 [10], TLQP-21 [5], AQEE-30 [2], and LQEQ-19 [1] (see Figure 8). NERP3 and NERP4 are fully conserved and are therefore unlikely to be responsible for the different metabolic phenotypes of hVGF and mVGF mice. NERP1, which regulates penile erection (21) and suppresses angiotensin II-induced vasopressin release from the pituitary (24), has 4-amino acid differences, and NERP2, which regulates feeding, locomotor activity, and EE via an orexin-dependent mechanism (25), has a single amino acid difference between the respective mouse and human peptide orthologs. Of note, both the human and mouse VGF-derived peptides NERP2 and TLQP-21, which have been shown to regulate feeding, adiposity, and/or EE, do differ in sequence (Figure 8). Thus, the single amino acid substitution in NERP2 could drive the metabolic phenotype of hVGF mice and remains to be investigated.

Figure 8.

Evolutionary conservation of VGF and VGF C-terminal domains. Regions of conservation between mouse and human VGF proteins, in the C-terminal domain that is ablated in hSNP (VGF_1–524) mice (amino acids numbers indicated are for human VGF), are shown. Note that human and mouse TLQP-21 and TLQP-62 peptides differ in several amino acids. Positions of the major bioactive NERP peptides, found in the truncated hVGF_1–524 protein, are also shown. NERP-3 and NERP-4 peptides are 100% conserved in mouse and human, NERP-2 has a single G/D substitution in the human peptide at position 36, whereas NERP-1 has 4-amino acid substitutions localized at its N terminus. An additional region that is highly conserved in mammals, reptiles, and fish and has been detected by peptidomic strategies but has no known bioactivity (44) has been aligned for the next species: Mus musculus (NP_001034474; 617 aa), Homo sapiens (NP_003369; 615 aa), Alligator mississippiensis (XP_006266396; 614 aa), and Danio rerio (XP_003198998; 642 aa).

The C-terminal peptides appear to be less evolutionarily conserved than the NERP peptides (encoded by the more N-terminal portion of VGF), suggesting that the metabolic phenotype of hVGF mice is potentially driven by sequence differences in the C-terminal peptides. TLQP-21, ablated in hSNP mice, is the only VGF-derived peptide for which site-specific mutagenesis coupled with structure/function analysis has been conducted (26). The results demonstrate that the C-terminal domain of TLQP-21 is critical for its biological activity and binding to the mouse C3a receptor 1 (C3aR1) (26), a recently identified TLQP-21 receptor (10). Notably, human TLQP-21 binds with lower affinity than mouse TLQP-21 to C3aR1, suggesting that the S20A substitution in the human ortholog of TLQP-21 is responsible for the lower activity of this peptide toward human and mouse C3aR1 (26). Thus, the obese phenotype of hVGF mice might, at least in part, result from a defect in the prolipolytic mechanism that is regulated by TLQP-21/C3aR1 in adipocytes (26).

EE in hVGF_1–524 mice lacking VGF C-terminal peptides is increased

Ablation of human C-terminal peptides in hVGF_1–524 (hSNP) mice was associated with a robust increase in VO₂, food intake, and locomotor activity, measured during active and inactive periods, suggesting that their metabolic phenotype may be driven predominantly by hyperactivity. Indeed, we demonstrated that the component of EE directly associated with activity (15) is significantly increased in hSNP groups. Administration of the VGF-derived peptide TLQP-21 (mVGF_556–576) to mice robustly potentiates β-adrenergic receptor-induced lipolysis, increases sympathetic tone, increases EE, and prevents diet-induced obesity (7, 8). Somewhat paradoxically, germline ablation of VGF results in a lean, hypermetabolic phenotype (6, 14), like TLQP-21 administration, indicative of increased sympathetic tone and lipolysis in these mice (16). Both male Vgf+/Vgf− and hSNP mice have reduced fat mass, and both have increased VO₂ during the active period, whereas male hSNP mice are hypermetabolic during both active and inactive periods, and unlike Vgf+/Vgf−, have body weights that are indistinguishable from wild-type mice. Increased EE in hSNP mice was also noted under fasted conditions (data not shown), when plasma glucose levels are significantly decreased, suggesting potential dysfunction in the neuronal circuitry that senses glucose and controls BAT thermogenesis, perhaps in part via regulation of WAT lipolysis and circulating free fatty acids levels.

Thus, despite similarities in VGF levels between hSNP and Vgf+/Vgf− mice, differences in their EE and body weights suggest that total ablation of C-terminal peptides in hSNP mice may predominantly drive the hypermetabolic, hyperactive phenotype. Moreover, the significant increase in NAEE in hSNP groups fed HFD suggests that additional regulatory mechanisms remain to be identified. The only C-terminal peptide for which a metabolic function has been systematically investigated is TLQP-21. Our analysis of hSNP and hSNP/mVGF mice suggests that TLQP-62, AQEE-30, and/or LQEQ-19 (all ablated in hSNP homozygotes) might have a positive effect on energy balance or alternatively could act as functional antagonists of TLQP-21 under physiological conditions. In addition, these data also suggest distinct developmental and adult functional roles for the VGF precursor and peptides in the regulation of EE and lipolysis. Further studies using currently available floxed VGF mouse models are required to address this issue.

Retrograde viral tract tracing from BAT has implicated neurons in the hypothalamus, including lateral hypothalamus, dorsomedial hypothalamus, and paraventricular nucleus, as well as in the locus coeruleus, raphe pallidum (RPa), and rostroventrolateral medulla in the regulation of sympathetic output to BAT (27–30). Blockade of GABA_A (gamma-Aminobutyric acid) receptors in RPa increases BAT SNA and thermogenesis that is resistant to inhibition by glucoprivation, consistent with inhibition that is mediated by activation of GABAergic input to BAT sympathetic premotor neurons in RPa (31). Thus, it remains possible that these latter circuits in the brainstem are compromised in homozygous hVGF_1–524 and Vgf−/Vgf− mice, because both have increased EE in the face of reduced fasting plasma glucose levels (4–6), and VGF is expressed widely, and notably is detected in inhibitory interneurons (32). In addition, development of the lean, hypermetabolic Vgf−/Vgf− phenotype was blocked by neonatal guanethidine-mediated destruction of sympathetic pathways (14), suggesting that increased sympathetic nerve activity, perhaps driven by the circuits above, increases basal metabolism and lipolysis (16).

Finally, because overall hVGF_1–524 protein levels were reduced approximately 50%–75% in hSNP mice (Figure 1, C–E), body composition and metabolic phenotype were also investigated in heterozygous Vgf+/Vgf− knockout mice that express approximately 50% the level of wild-type VGF protein (4). Decreased hVGF_1–524 protein levels in hSNP mice are most likely a result of reduced protein stability and/or missorting and degradation, as part of a previously defined VGF sorting signal, located in the C terminus (33), is deleted in hVGF_1–524. Although unlikely, we cannot rule out differential detection during Western blotting of sodium dodecyl sulfate-denatured hVGF_1–615 or mVGF_1–617 and hVGF_1–524 by monoclonal anti-hVGF_159–223.

Mechanisms underlying altered EE in hVGF_1–524 mice

BAT and skeletal muscle are important sites of nonshivering thermogenesis in mammals, regulating temperature and energy homeostasis (23, 34). Total ablation of C-terminal peptides in hSNP mice may predominantly drive the hypermetabolic, hyperactive phenotype, but investigation of iBAT protein expression did not reveal increased UCP-1 levels, previously noted in lean, hypermetabolic, homozygous germline VGF knockout mice (14). Our analysis reveals that a significant component of total EE can be explained by activity, highlighting the potential utility of the hSNP mouse model in disentangling the molecular mechanisms underlying activity-related thermogenesis. We therefore examined gene expression that impacts mitochondrial function in skeletal muscle, the predominant tissue making up lean mass, which makes the major contribution to EE during rest and exercise (35). A number of potential mechanisms underlie the contribution of skeletal muscle to thermogenesis and whole-body energy metabolism. Increased mitochondrial biogenesis in skeletal muscle leads to increased EE and resistance to diet-induced obesity in mice (36–39). In addition, Akt1 (also known as protein kinase B) ablation in mice, which increased EE and resistance to diet-induced obesity as does VGF ablation (4, 5), was associated with increased UCP-2, UCP-3, and PDK4 mRNA levels and increased palmitate oxidation, in skeletal muscle (40). Lastly, sarcolipin, a regulator of the sarco/endoplasmic reticulum Ca(2+)-ATPase pump, is required for muscle-based thermogenesis, and its ablation predisposed mice to diet-induced obesity (41). Investigation of skeletal muscle gene expression in our mice revealed an increase in UCP-2 mRNA levels, and a slight but not significant increase in Sarcolipin and cytochrome c oxidase subunit II mRNA levels, in male hSNP mice, suggesting that the EE phenotype could be driven in part by increased mitochondrial function and muscle thermogenesis. On the other hand, we did not detect changes in gene expression in female mice, which showed a similar metabolic phenotype to males, thus suggesting that other mechanisms are likely involved or that activity-related thermogenesis is mediated by many independent pathways in skeletal muscle as well as in other tissues.

Additionally, male and female hSNP mice fed a HFD showed increased NAEE as well, thus suggesting that adaptive and/or basal thermogenesis could also be increased. However, we did not detect increased iBAT UCP-1 protein levels (the main mechanism for nonshivering thermogenesis in iBAT), suggesting that increased thermogenesis might also be driven by changes in global metabolism that perhaps reflect altered mitochondrial function or morphology, as noted in VGF knockout mice (14), which remain to be investigated in humanized mice.

Increased glucose tolerance in hVGF_1–524 mice is associated with decreased adiposity

Previous studies demonstrated increased glucose clearance in oral and ip glucose tolerance tests in homozygous germline VGF knockout mice (4), and based on hyperinsulinemic euglycemic glucose clamp analysis, increased insulin sensitivity (6). Increased glucose tolerance in hSNP mice may result from increased glucose use and/or increased insulin sensitivity, the former likely a result of increased basal metabolic rate and the latter a result of reduced adiposity.

Recent studies have demonstrated that the VGF C-terminal peptide TLQP-21 increases glucose-stimulated insulin secretion in vivo and in vitro, improves glucose tolerance, and enhances pancreatic β-cell survival (9), suggesting there are significant differences between the function of endogenous and exogenous VGF C-terminal peptides, including TLQP-21, in the adult, where they regulate glucose homeostasis (9) and lipolysis (7, 8), and the developmental roles for VGF in the pancreatic β-cell and sympathetic nervous system that would be impacted in germline knockin hSNP mice, reported here, and also in germline VGF knockout mice. Moreover, in hSNP mice, there may be compensatory actions of peptides that are found in the hVGF_1–524 protein, including NERP-2, which also regulates β-cell function (9, 42). Further study of VGF actions in pancreatic β-cells is thus required, particularly given recent data demonstrating that the adipokine adipsin increases glucose-stimulated insulin secretion from β-cells by catalyzing the generation of the complement factor C3a (43); C3a and TLQP-21 are both ligands for the complement receptor C3aR1 (10). Use of conditional VGF and C3aR1 β-cell knockout approaches should improve our understanding of endogenous VGF function in the pancreas and the receptor system it activates.

Conclusions

We have demonstrated that human and mouse VGF proteins are largely functionally conserved and that C-terminal VGF peptides are not required for reproductive function. Importantly, mice in which the human VGF gene replaces the mouse Vgf ortholog are significantly heavier and are more obese than control animals, suggesting species-specific differences in the metabolic functions of VGF peptide(s). Finally, we showed that mice carrying a C-terminal truncated form of human VGF are hypermetabolic and hyperactive, do not show a consistent increase in iBAT UCP-1 expression but at least in males, and have higher levels of skeletal muscle UCP-2 mRNA, suggesting that these mice could represent a novel model to determine the role of activity on thermogenesis, and that VGF C-terminal peptides may be critical regulators of this process.