Abstract

Growth differentiation factor 11 (GDF11) is member of the transforming growth factor β (TGF-β) superfamily of proteins. Circulating GDF11 concentrations appear to decline with age, and its depletion is associated with cardiac hypertrophy and other morbidities. Knowledge of GDF11 regulation is limited, and the effects of natural genetic variation on GDF11 levels are currently undefined. We tested whether genetic background determines serum GDF11 concentrations using two classical inbred mouse strains: C57BL/6J (B6) and BALB/cByJ (BALB). B6 mice exhibited significantly higher GDF11 levels than BALB mice, and these strain differences were consistent throughout the life span. Overall, interactions between age and genetic background determined GDF11 concentrations, which were unaffected by sex. We then surveyed a panel of 22 genetically diverse inbred mouse strains and discovered a sixfold range in GDF11 levels at middle age. We estimated that 74.52% of phenotypic variation in GDF11 levels was attributable to genetic background. We used the Mouse Phenome Database to screen for phenotypes that correlate with GDF11. Interestingly, GDF11 levels predicted median strain life spans. This study revealed high heritability of GDF11 levels. Furthermore, our correlative data suggest that GDF11 may serve as a novel predictor of mammalian life span.

Advanced age is a major risk factor for cardiovascular disease (CVD). Mechanisms that couple aging and CVD must be clarified, and members of transforming growth factor β (TGF-β) superfamily have emerged as potent regulators of this relationship. For example, growth differentiation factor 11 (GDF11) has generated considerable interest in recent years. Reports suggest that circulating GDF11 levels decrease during aging (1,2), and its depletion drives deterioration of neurogenic niche vasculature (3), skeletal muscle dysfunction (2), and cardiac hypertrophy (1). However, the mechanisms that control GDF11 levels throughout the life span have not yet been elucidated. A forward genetics approach could identify important genes and pathways that determine circulating GDF11 levels, but investigators must first establish whether its serum concentrations are heritable and therefore affected by natural genetic variation.

The current study defines interactions between genetic background, sex, and age on circulating GDF11 levels. Our initial experiments were performed using two classical inbred strains that are widely used in both cardiovascular and aging studies, C57BL/6J (B6) and BALB/cByJ (BALB), and these efforts guided the development of a strain survey. Using 22 genetically diverse inbred mouse strains, we estimated GDF11 heritability, and efficient mixed-model analysis (EMMA) identified novel candidate genes associated with serum GDF11 levels. Concurrently, we studied the relationship between GDF11 levels and age-related cardiac biomarkers: heart weight and circulating concentrations of growth differentiation factor 15 (GDF15), a related TGF-β superfamily protein that increases with age and predicts CVD risk (4). We also tested for correlations between GDF11 levels and phenotypes related to longevity. Our study is the first to demonstrate that GDF11 levels are genetically regulated, and it will inform future efforts to elucidate mechanisms that determine GDF11 levels and their effects on health.

Methods

Mice

All mice were purchased from The Jackson Laboratory or obtained from the National Institute on Aging (NIA) Aged Rodent Colonies. Blood collections were performed when young-adult cohorts were 3–4 months old, middle-aged were 12–13 months old (except WSB/EiJ and POHN/DehJ, which were 14–15 months old), and the old were 22–23 months old. All mice were fed standard chow (LabDiet, St. Louis, MO, product 5053) and housed on a 12-hour light-dark cycle at University of Georgia.

Blood Collection and ELISA Kits

Blood was collected by submandibular bleeding (ACUC-approved protocol A2013-08-011). GDF11 levels were quantified by ELISA kit manufactured by Biomatik (EKU04586; Wilmington, DE). GDF15 analyses were performed using the Quantkine Mouse/Rat GDF15 Immunoassay manufactured by R&D Systems, Inc. (product MGD150; Minneapolis, MN). Concentrations were quantified by optical density on a SpectraMax microplate reader (Molecular Devices, Sunnyvale, CA).

Statistical Analysis

All statistical analyses were conducted using GraphPad Prism 6 (GraphPad Software, La Jolla, CA), R, and SAS (SAS Institute Inc., Cary, NC). Data are shown as mean ± SEM; p values less than .050 are considered as significant.

Efficient Mixed-Model Analysis

Association mapping was performed by EMMA. Single nucleotide polymorphisms (SNPs) were categorized as significant when p values were less than 10–5. The normality of p values was evaluated by Q-Q plots. Output from the EMMA server was converted to Build 38 by Ensembl Assembly Converter. Genes and loci were identified in the Mouse Genome Informatics website (informatics.jax.org).

Results

Age, Sex, and Strain Interactions Regulate GDF11 Levels in B6 and BALB Mice

Mice were categorized according to strain (B6 or BALB), sex (male or female), and age (young, middle-aged, or old). GDF11 levels decreased over time in both strains. Young mice had higher GDF11 concentrations than middle-aged (B6, p = .0054; BALB, p = .0013) and old mice (B6, p = .0329; BALB, p = .0009; Figure 1A;

). No significant differences were discovered between males and females, indicating that sex was not a significant factor (p = .884). B6 mice had higher GDF11 levels than BALB at all ages. The most significant differences between strains were at 12 months (p = 3.44×10–7), the age at which GDF11 depletion is mostly complete.

Figure 1.

Genetic background determines growth differentiation factor 11 (GDF11) levels. (A) Serum GDF11 levels in B6 and BALB (*p < .05; **p < .001). (B) Serum GDF11 concentrations in 22 inbred strains. Samples were obtained from middle-aged female mice. (C) Manhattan plot of GDF11 levels from efficient mixed-model analysis. Cross indicates significant single nucleotide polymorphisms. Data are represented as mean ± SEM.

Figure 1.

Genetic background determines growth differentiation factor 11 (GDF11) levels. (A) Serum GDF11 levels in B6 and BALB (*p < .05; **p < .001). (B) Serum GDF11 concentrations in 22 inbred strains. Samples were obtained from middle-aged female mice. (C) Manhattan plot of GDF11 levels from efficient mixed-model analysis. Cross indicates significant single nucleotide polymorphisms. Data are represented as mean ± SEM.

Heritability of GDF11 Levels and Identification of Candidate Regulatory Genes

To capture the effect of genetic background on GDF11 levels, we performed a survey of 22 genetically diverse mouse strains. A sixfold range was observed between the strains with the lowest (C3H, 24.04±2.98 pg/mL) and highest (WSB, 147.15±16.33 pg/mL) GDF11 levels (Figure 1B). We estimated that 74.52% of phenotypic variation was attributable to strain (p < 2.2×10–16). EMMA identified candidate SNPs that regulate GDF11 variation (Figure 1C). We narrowed the SNPs to those embedded within protein-coding genes and identified seven candidate genes that may contribute to GDF11 regulation (Table1;

).

Table 1.

Significant Loci and Candidate GDF11 Regulatory Genes

Chromosome Location Protein-Coding Gene 
93403982 Ano7 
93404204 Ano7 
69115224 Kpna4 
11 50605075 Adamts2 
13 38220021 Dsp; Snrnp48 
13 38237800 Dsp 
13 38293619 Dsp 
13 47129920 Rnf144b 
17 90411299 Nrxn1 
Chromosome Location Protein-Coding Gene 
93403982 Ano7 
93404204 Ano7 
69115224 Kpna4 
11 50605075 Adamts2 
13 38220021 Dsp; Snrnp48 
13 38237800 Dsp 
13 38293619 Dsp 
13 47129920 Rnf144b 
17 90411299 Nrxn1 

Notes: Single nucleotide polymorphisms were identified by efficient mixed-model analysis and then narrowed to those embedded within protein-coding genes.

Relationship Between GDF11 Levels and Phenotypes Related to Health Span and Life Span

We tested for relationships between GDF11 levels and predictors of CVD, including GDF15 levels and heart weight (Figure 2). The relationship between GDF11 and GDF15 was weak (p = .0236) and disappeared when the sexes were separated, implying that a distinct pathway regulates these factors. GDF11 levels were negatively correlated with heart weight in males (r = −.7053, p < .001) but not in females (r = .0643, p = .7267). Using the Mouse Phenome Database, we screened for relationships between GDF11 levels and other age-related phenotypes (Table 2). A significant quadratic relationship was discovered between GDF11 levels and median strain life spans (r2 = .3400, p value = .0443; Figure 3A), which was largely driven by two strains: C3H and NZO. Without them, there was an obvious positive linear relationship (r2 = .5234, p = .0015; Figure 3B), indicating that higher GDF11 levels at middle age are associated with longer life spans. We also found that excluding short-lived strains (median life spans < 600 days) did not affect the relationship between GDF11 levels and life span (

). A potential linear pattern was observed between levels of GDF11 and IGF-1 among 18 inbred mouse strains (r = −.3334, p = .1764; Figure 3C). NZO, again an outlying strain with relatively high IGF-1 and GDF11, appeared to distort the fit. When it was excluded, the negative relationship became significant (r = −.5244, p = .0307; Figure 3D).

Figure 2.

Relationship between growth differentiation factor 11 (GDF11) levels and cardiac biomarkers. GDF11 concentrations were tested for relationships with (A) heart weight [r (male) = −.7053, p < .001; r (female) = .0643, p = .7267] and (B) heart weight (normalized to body weight; r (male) = −.5765, p = .0006; r (female) = .2085, p = .2521). (C) GDF15 versus heart weight [r (male) = .6461, p = 6.484 x 10-5; r (female) = .2058, p = .2584)]. (D) A weak relationship exists between serum levels of GDF11 and GDF15 in mice (r = -.2827, p = .0236).

Figure 2.

Relationship between growth differentiation factor 11 (GDF11) levels and cardiac biomarkers. GDF11 concentrations were tested for relationships with (A) heart weight [r (male) = −.7053, p < .001; r (female) = .0643, p = .7267] and (B) heart weight (normalized to body weight; r (male) = −.5765, p = .0006; r (female) = .2085, p = .2521). (C) GDF15 versus heart weight [r (male) = .6461, p = 6.484 x 10-5; r (female) = .2058, p = .2584)]. (D) A weak relationship exists between serum levels of GDF11 and GDF15 in mice (r = -.2827, p = .0236).

Table 2.

Correlations Between GDF11 Levels and Other Disease-relevant Phenotypes

Phenotype Correlation, if linear p Value Age 
Life span Quadratic .0443 N/A 
Life span (without C3H, NZO) .7234 .0015 N/A 
IGF-1 (6 month) −.3958 .1040 6 months 
IGF-1 (6 month; without NZO) −.6612 .0038 6 months 
IGF-1 (12 month) −.3334 .1764 12 months 
IGF-1 (12 month; without NZO) −.5244 .0307 12 months 
IGF-1 (18 month) −.3130 .2060 18 months 
IGF-1 (18 month; without NZO) −.6252 .0073 18 months 
Thyroxine (12 month) −.2519 .3132 12 months 
Thyroxine (18 month) −.0056 .9830 18 months 
Heart rate −.3804 .1461 12 months 
Glucose .1140 .6631 12 weeks 
Non-HDL cholesterol −.2542 .2936 10 weeks 
HDL cholesterol .1749 .4875 12 months 
Phenotype Correlation, if linear p Value Age 
Life span Quadratic .0443 N/A 
Life span (without C3H, NZO) .7234 .0015 N/A 
IGF-1 (6 month) −.3958 .1040 6 months 
IGF-1 (6 month; without NZO) −.6612 .0038 6 months 
IGF-1 (12 month) −.3334 .1764 12 months 
IGF-1 (12 month; without NZO) −.5244 .0307 12 months 
IGF-1 (18 month) −.3130 .2060 18 months 
IGF-1 (18 month; without NZO) −.6252 .0073 18 months 
Thyroxine (12 month) −.2519 .3132 12 months 
Thyroxine (18 month) −.0056 .9830 18 months 
Heart rate −.3804 .1461 12 months 
Glucose .1140 .6631 12 weeks 
Non-HDL cholesterol −.2542 .2936 10 weeks 
HDL cholesterol .1749 .4875 12 months 

Note: Phenotypic data from 22 genetically diverse inbred mouse strains were obtained from the Mouse Phenome Database (phenome.jax.org). Phenotypes were screened for correlations with GDF11 levels at middle age.

Figure 3.

Growth differentiation factor 11 (GDF11) levels predict longevity phenotypes. (A) Association between GDF11 levels and median life span (r2 = .3400, p = .0443). (B) Association between GDF11 and median life span, without C3H and NZO (r2 =.5234, p = .0015). (C) Correlation between levels of GDF11 and IGF-1 in middle-aged mice (r = −.3334, p = .1764) and (D) without NZO (r = −.5244, p = .0307).

Figure 3.

Growth differentiation factor 11 (GDF11) levels predict longevity phenotypes. (A) Association between GDF11 levels and median life span (r2 = .3400, p = .0443). (B) Association between GDF11 and median life span, without C3H and NZO (r2 =.5234, p = .0015). (C) Correlation between levels of GDF11 and IGF-1 in middle-aged mice (r = −.3334, p = .1764) and (D) without NZO (r = −.5244, p = .0307).

Discussion

GDF11 is a key regulator of developmental processes such as muscle progenitor cell proliferation (5). Previous studies identified GDF11 as an anti-aging factor that is lost as an organism grows older and suggested that administration of exogenous GDF11 reverses age-related deterioration (1). Recent studies challenged those findings and concluded that GDF11 levels instead increase with age (6,7). The divergent results appear to be rooted in methodology (8). Anti-GDF11 antibodies often exhibit cross-reactivity with growth differentiation factor 8 (GDF 8), or myostatin. The current study quantified serum GDF11 levels using a commercially available ELISA, which was designed to selectively detect GDF11. We demonstrated that GDF11 levels decrease with age, which aligns with original reports by Loffredo and colleagues (1) and a recent study by Poggioli and colleagues (8). However, the GDF11 protein may exist in multiple forms in the circulation, and it remains unclear whether distinct forms are differentially detected by anti-GDF11 antibodies.

Our study demonstrated for the first time that GDF11 levels are highly heritable, and we identified candidate genes associated with GDF11 levels at middle age. Using EMMA, we discovered that several SNPs were located within the gene that encodes for desmoplakin (Dsp). Dsp is the major component of desmosomes, which are prominent in cardiac muscle and epidermal cells (9). Interestingly, Dsp mutations cause cardiomyopathy (10). However, it is unclear how Dsp or the other candidate genes would directly regulate GDF11 levels. To address this issue, as well as the inherent limitations associated with EMMA analyses (11), future efforts should narrow this candidate gene list using linkage mapping. Molecular studies must also verify the mechanisms by which candidates regulate GDF11 levels.

A recent study indicated that GDF11 levels may predict human mortality by mitigating risk of cardiovascular events (12). To better understand the effects of GDF11 concentrations on health, we screened for relationships between GDF11 levels and disease-relevant phenotypes in the Mouse Phenome Database. We discovered a significant correlation between GDF11 levels and median strain life spans (13,14). Higher GDF11 levels predicted longer life spans, which interestingly contrasts with the effects of myostatin on maximal murine life span (15). The relationship between GDF11 levels and IGF-1 levels did not achieve statistical significance, but a relationship emerged when the outlying strain NZO was removed. IGF-1 is a major life-span regulator (16), and IGF-1 levels predict murine life span (13). Moreover, IGF-1Ea, one of the IGF-1 isoforms, also improves cardiac function after myocardial infarction (17,18). We predict that GDF11 may intersect with canonical aging pathways to regulate mammalian life span and CVD risk, a concept that must be tested in future studies with greater statistical power. Further characterization of GDF11 and its regulatory mechanisms will potentially illuminate novel therapeutic strategies to extend health span and life span.

Supplementary Material

Supplementary material can be found at:

Funding

This work was supported by University of Georgia Office of Vice-President for Research and College of Family and Consumer Sciences.

Conflict of Interest

There are no disclosures to report.

Acknowledgments

The authors thank Katie Marie Norris, Erica Jane Coe, and Victoria Piurowski for their assistance with this article.

References

1.
Loffredo
FS
Steinhauser
ML
Jay
SM
et al
.
Growth differentiation factor 11 is a circulating factor that reverses age-related cardiac hypertrophy
.
Cell
 .
2013
;
153
:
828
839
. doi:10.1016/j.cell.2013.04.015
2.
Sinha
M
Jang
YC
Oh
J
et al
.
Restoring systemic GDF11 levels reverses age-related dysfunction in mouse skeletal muscle
.
Science
 .
2014
;
344
:
649
652
. doi:10.1126/science.1251152
3.
Katsimpardi
L
Litterman
NK
Schein
PA
et al
.
Vascular and neurogenic rejuvenation of the aging mouse brain by young systemic factors
.
Science
 .
2014
;
344
:
630
634
. doi:10.1126/science.1251141
4.
Wallentin
L
Zethelius
B
Berglund
L
et al
.
GDF-15 for prognostication of cardiovascular and cancer morbidity and mortality in men
.
PLoS One
 .
2013
;
8
:
e78797
. doi:10.1371/journal.pone.0078797
5.
Nomura
T
Ueyama
T
Ashihara
E
et al
.
Skeletal muscle-derived progenitors capable of differentiating into cardiomyocytes proliferate through myostatin-independent TGF-beta family signaling
.
Biochem Biophys Res Commun
 .
2008
;
365
(
4
):
863
869
.
6.
Egerman
MA
Cadena
SM
Gilbert
JA
et al
.
GDF11 increases with age and inhibits skeletal muscle regeneration
.
Cell Metab
 .
2015
;
22
:
164
174
. doi:10.1016/j.cmet.2015.05.010
7.
Rodgers
BD
Eldridge
JA
.
Reduced circulating GDF11 is unlikely responsible for age-dependent changes in mouse heart, muscle, and brain
.
Endocrinology
 .
2015
;
156
(
11
):
3885
3888
.
doi:10.1210/en.2015-1628
8.
Poggioli
T
Vujic
A
Yang
P
et al
.
Circulating growth differentiation factor 11/8 levels decline with age
.
Circ Res
 .
2015
.
doi:10.1161/CIRCRESAHA.115.307521
9.
Green
KJ
Parry
DA
Steinert
PM
et al
.
Structure of the human desmoplakins
.
J Biol Chem
 .
1990
;
265
:
2603
2612
.
10.
Pilichou
K
Nava
A
Basso
C
et al
.
Mutations in desmoglein-2 gene are associated with arrhythmogenic right ventricular cardiomyopathy
.
Circulation
 .
2006
;
113
:
1171
1179
.
11.
Kang
HM
Zaitlen
NA
Wade
CM
et al
.
Efficient control of population structure in model organism association mapping
.
Genetics
 .
2008
;
178
:
1709
1723
. doi:10.1534/genetics.107.080101
12.
Olson
KA
Beatty
AL
Heidecker
B
et al
.
Association of growth differentiation factor 11/8, putative anti-ageing factor, with cardiovascular outcomes and overall mortality in humans: analysis of the Heart and Soul and HUNT3 cohorts
.
Eur Heart J
 .
2015
.
doi:10.1093/eurheartj/ehv385
13.
Yuan
R
Tsaih
SW
Petkova
SB
et al
.
Aging in inbred strains of mice: study design and interim report on median lifespans and circulating IGF1 levels
.
Aging Cell
 .
2009
;
8
:
277
287
. doi:10.1111/j.1474-9726.2009. 00478.x
14.
Yuan
R
Menga
Q
Nautiyalb
J
et al
.
Genetic coregulation of age of female sexual maturation and lifespan through circulating IGF1 among inbred mouse strains
.
Proc Natl Acad Sci USA
 .
2011
;
109
:
8224
8229
.
15.
Mendias
CL
Bakhurin
KI
Gumucio
JP
et al
.
Haploinsufficiency of myostatin protects against aging-related declines in muscle function and enhances the longevity of mice
.
Aging Cell
 .
2015
;
14
:
704
706
. doi:10.1111/acel.12339
16.
Gong
Z
Kennedy
O
Sun
H
et al
.
Reductions in serum IGF-1 during aging impair health span
.
Aging Cell
 .
2014
;
13
:
408
418
. doi:10.111.acel.12188
17.
Gallego-Colon
E
Sampson
RD
Sattler
S
et al
.
Cardiac-restricted IGF-1Ea overexpression reduces the early accumulation of inflammatory myeloid cells and mediates expression of extracellular matrix remodelling genes after myocardial infarction
.
Mediat Inflamm
 .
2015
;
2015
:
484357
. doi:10.1155/2015/484357
18.
Ungvari
Z
Csiszar
A
.
The emerging role of IGF-1 deficiency in cardiovascular aging: recent advances
.
J Gerontol A Biol Sci Med Sci
 .
2012
;
67
(
6
):
599
610
. doi:10.1093/gerona/gls072

Author notes

Address correspondence to Robert Pazdro, PhD, Department of Foods and Nutrition, University of Georgia, 305 Sanford Drive, Athens, GA 30602. E-mail: rpazdro@uga.edu
Decision Editor: Rafael de Cabo, PhD

Supplementary data