Download Elevated FGF21 Leads to Attenuated Postnatal Linear Growth in

J C E M
O N L I N E
Hot Topics in Translational Endocrinology—Endocrine Research
Elevated FGF21 Leads to Attenuated Postnatal Linear
Growth in Preterm Infants Through GH Resistance in
Chondrocytes
Leonardo Guasti,* Sanna Silvennoinen,* Neil W. Bulstrode, Patrizia Ferretti,
Ulla Sankilampi,* and Leo Dunkel*
Centre for Endocrinology (L.G., L.D.), William Harvey Research Institute, Barts and the London, Queen Mary
University of London, London EC1M 6BQ, United Kingdom; Department of Pediatrics (S.S., U.S.), Kuopio
University Hospital, 70210 Kuopio, Finland; Department of Plastic Surgery (N.W.B.), Great Ormond Street
Hospital for Children National Health Service Trust, London WC1N 3JH, United Kingdom; and Developmental
Biology Unit (P.F.), University College London Institute of Child Health, London WC1N 1EH, United Kingdom
Context: The hormone fibroblast growth factor 21 (FGF21) is a key metabolic regulator in the
adaptation to fasting. In food-restricted mice, inhibition of skeletal growth is mediated by the
antagonistic effect of FGF21 on GH action in the liver and growth plate.
Objective: The objective of the study was to assess the role of FGF21 in growth regulation in humans
using postnatal growth failure of very preterm infants as a model.
Design: FGF21 levels were measured serially in very preterm infants, and their linear growth evaluated from birth to term-equivalent age. Primary chondrocytes obtained from pediatric donors
were used to test whether FGF21 can directly interfere with GH signaling.
Results: A negative association (␤ ⫺.415, P ⬍ .005, linear regression model) of FGF21 levels with the
change in SD score for length was found. In primary chondrocytes, FGF21 upregulated basal and
GH-induced SOCS2 expression and inhibited GH-induced signal transducer and activator of transcription 5 (STAT5) phosphorylation as well as GH-induced COLII and ALP expression. Finally, FGF21
inhibited GH-induced IGF-1 expression and cell proliferation, indicating GH resistance. However,
FGF21 did not affect IGF-1–induced cell proliferation.
Conclusions: Elevated FGF21 serum levels during the first weeks of life are independently associated with postnatal growth failure in preterm infants. Furthermore, our data provide mechanistic
insights into GH resistance secondary to prematurity and may offer an explanation for the growth failure
commonly seen in chronic conditions of childhood. (J Clin Endocrinol Metab 99: E2198–E2206, 2014)
A
pproximately 5% to 10% of children are born prematurely, and 1% of children are born ⬍32 weeks
gestation, as very preterm (VPT) infants. Despite improvements in neonatal intensive care including standardized
nutrition, growth retardation is common in these infants,
and final height is often compromised (1, 2). Postnatally
reduced GH/IGF-1 action impairs linear growth both in
term (3) and preterm (4) infants, but at the same time, even
healthy infants display biochemical features of GH resis-
tance (high GH and low IGF-1 levels) through yet unknown mechanisms (5, 6).
Recently, a new mechanism of GH resistance and
growth failure was proposed in mice. In a seminal study by
Kubicky et al (7), induction of fibroblast growth factor 21
(FGF21) by undernutrition was shown to cause GH resistance. It was also shown that FGF21 antagonized GH
actions on chondrogenesis directly at the growth plate and
that high concentrations of FGF21 directly suppress
ISSN Print 0021-972X ISSN Online 1945-7197
Printed in U.S.A.
Copyright © 2014 by the Endocrine Society
Received February 26, 2014. Accepted July 18, 2014.
First Published Online August 19, 2014
* L.G., S.S., U.S., and L.D. contributed equally to this work.
Abbreviations: CV, coefficient of variation; FBS, fetal bovine serum; FGF21, fibroblast
growth factor 21; FGFR, FGF receptor; GHR, GH receptor; RT-qPCR, quantitative RT-PCR;
SDS, SD score; STAT5, signal transducer and activator of transcription 5; SOCS2, suppressor
of cytokine signaling 2; VPT, very preterm.
E2198
jcem.endojournals.org
J Clin Endocrinol Metab, November 2014, 99(11):E2198 –E2206
doi: 10.1210/jc.2014-1566
doi: 10.1210/jc.2014-1566
growth plate chondrocyte proliferation and differentiation (8). Importantly, growth failure due to undernutrition
was attenuated in FGF21-knockout compared with wildtype mice (7).
FGF21, along with FGF15/19, lacks the FGF heparinbinding domain; therefore, it can be released from the site
of synthesis and function as an endocrine factor (9). During fasting, increased hepatic secretion of FGF21 induces
gluconeogenesis, fatty acid oxidation, and ketogenesis.
Thus, FGF21 is considered a key regulator of the metabolic adaptation to fasting (10). In target tissues, FGF21
binds to FGF receptors (FGFRs). A preferential binding to
the complex FGFR1/␤-klotho has been described (11, 12).
In this study, the role of FGF21 in mediating GH resistance was examined in a cohort of VPT infants. The
molecular mechanisms underlying FGF21 actions were
investigated using an in vitro model of primary human
chondrocytes. We demonstrate for the first time that elevated FGF21 levels impair linear growth in VPT infants
via a mechanism likely involving direct inhibition of GH
action on chondrocytes. This suggests an important role
for FGF21 in the regulation of postnatal linear growth in
humans.
jcem.endojournals.org
E2199
Table 1. Clinical Characteristics of the 32 VPT Infants
(19 Males)
Median/n Range
At birth
Gestational age, wk
Weight, kg
Weight, SDa
Length, cm
Length, SDa
At term-equivalent age
Postmenstrual age, wk
Weight, kg
Weight, SDb
Length, cm
Length, SDb
Nutrition at sampling
Week 1 (n ⫽ 29)
Enteral
Combined parenteral and enteral
Total parenteral
Week 3 (n ⫽ 32)
Enteral
Combined parenteral and enteral
Total parenteral
Week 5 (n ⫽ 28)
Enteral
Combined parenteral and enteral
Total parenteral
28.2
0.99
⫺0.2
36.8
⫺0.1
23.4 –31.9
0.48 –1.92
⫺2.2 to 1.7
28.0 – 44.0
⫺2.6 to 1.6
40.0
2.99
⫺1.2
47.5
⫺1.1
37.0 – 41.6
1.55–3.79
⫺4.3 to 0.4
40.0 –51.5
⫺4.5 to 0.7
5
17
7
19
11
2
22
5
1
a
Birth weight and length were converted to SDS using the populationbased birth size reference for singletons or twins (16).
Patients and Methods
b
Weight and length at term-equivalent age were converted to SDS
using the population-based birth size reference for singletons (16).
Patients
As part of the Finnish PreBaby study on metabolism and
growth in VPT infants, 32 infants (19 boys, 57%) were recruited
during the first week of life at the Kuopio University Hospital
neonatal intensive care unit (Table 1).
The median gestational age at birth was 28.2 (range 23.4 –
31.9, SD 2.9) weeks, and the median length of hospitalization
was 67 (range 28 –182, SD 45.4) days. At discharge, the median
postmenstrual age was 38.6 (range 35.4 –51.0, SD 4.1) weeks.
Two sets of unidentical twins and one set of unidentical triplets
were included in the study cohort. All the VPT infants survived,
but presented with typical morbidity associated with prematurity: 5 infants (16%) had early or late-onset septic infection, three
(9%) had necrotizing enterocolitis, 8 (25%) developed bronchopulmonary dysplasia (diagnosed at 36 postmenstrual weeks by
the oxygen reduction test) (13), 2 (6%) had intraventricular hemorrhage grade III or IV, and 2 infants (6%) developed retinopathy of prematurity. Their nutrition adhered to current recommendations (14, 15). The VPT infants were fed immediately with
a high-dose iv protein and carbohydrate infusion, an early lipid
infusion, and minimal enteral feeding to avoid undernutrition
and starving and to achieve full enteral feeding as early as possible. Data on nutrition (enteral nutrition, combined enteral and
parenteral nutrition, or total parenteral nutrition) were registered. The median age when the full enteral nutrition was
achieved was 10 (range 3–101, SD 22.9) days. Weight and recumbent length were measured at birth, at weeks 1, 3, and 5, and
at term-equivalent age using the Giraffe OmniBed inbed scale
(Ohmeda Medical) in intensive care or a baby scale (Seca model
376) and neonatometer (Pedihealth). These measures were trans-
formed into SD scores (SDSs) using the contemporary population-based reference (16). Small for gestational age was defined
as birth weight or length at least 2 SD below the sex- and gestational age-specific reference mean (17). Extrauterine growth
retardation was defined as weight or length at least 2 SD below
the sex-specific reference mean at the term-equivalent age. Only
2 of the 32 VPT infants had birth weight or length below ⫺2.0
SD (ie, were small for gestational age at birth).
Mixed umbilical blood samples were obtained immediately
after birth and peripheral venous or arterial samples in weeks 1,
3, and 5. The plasma and serum samples were prepared by centrifugation after blood collection, separated into aliquots, and
stored at ⫺70°C until analyzed. Because of the extremely small
size of our patients, blood samples were not obtained after overnight fasting.
Plasma FGF21 concentrations were measured by an FGF21
human ELISA kit (BioVendor) according to the manufacturer’s
instructions. Plasma was diluted 1:1.5 or 1:2. The measuring
range with 1:1.5 dilution was 11.3 to 2880 pg/mL and with 1:2
dilution 15 to 3840 pg/mL. Within-run coefficient of variation
(CV) was 4.6% (mean FGF21, 223 pg/mL), and between-run CV
was 13.9% (mean 611 pg/mL).
Serum IGF-1 concentrations were measured by an ELISA kit
(Mediagnost GmbH). The within-run CV was 4.7% (mean
IGF-1 4.8 nmol/L), and between-run CV for control serum was
6% (mean IGF-1, 13.6 nmol/L) in the range of 1.84 to 65 nmol/L.
E2200
Guasti et al
FGF21 in Preterm Infants
Cells
Chondrocyte cultures were established from rib cartilage of
consented pediatric patients undergoing facial reconstruction
(because of hemifacial microsomia with microtia) at Great Ormond Street Hospital, London, UK. In particular, patients (n ⫽
3 males, aged 9, 10, and 11 years old) underwent autologous
costal cartilage to ear graft, and they had normal linear growth. The
protocol to establish primary cultures, their markers’ expression,
and chondrogenic potential has been described previously (18, 19).
Cells were grown in high-glucose (4500 mg/L) DMEM (Life Technologies) supplemented with 10% embryonic stem cells qualified
fetal bovine serum (FBS) (Life Technologies) and 1% penicillin/
streptomycin (Life Technologies) at 37°C in a humidified incubator
with 5% CO2. Cells were used at passage 4 or lower.
RT-PCR and quantitative RT-PCR
Human brain RNA was obtained from Agilent Technologies.
Human liver RNA was from Clontech. Human rib cartilage
RNA from pediatric donors was extracted using TRIzol (Life
Technologies) followed by RNeasy Mini kit (QIAGEN). RNA
from primary chondrocytes was extracted using RNeasy Mini
kit. cDNAs were synthesized using Moloney Murine Leukemia
Virus (M-MLV) reverse transcriptase (Promega) according to
the manufacturer’s instructions.
PCR to detect FGFR1 (isoforms IIIb and IIIc), ␤KLOTHO,
GHR, and GAPDH was performed in a GS1 thermocycler (GStorm) using 1.25 U Taq polymerase (New England Biolabs),
200␮M of each dNTP, 1.5mM MgCl2, and 0.5␮M specific
primers.
For quantitative RT-PCR (RT-qPCR), amplification was in a
10-␮L reaction containing 2 ␮L cDNA template, 5 ␮L 2⫻ SYBR
Green I Master Mix (KAPA Biosystems), 0.2 ␮L low ROX
(KAPA Biosystems), 0.5␮M of each primer and 2.3 ␮L nucleasefree H2O and performed using an Mx3000 thermocycler (Stratagene). Data were analyzed with MxPro software (Stratagene).
For RT-PCR and RT-qPCR, GAPDH was used as housekeeping gene. Primers and amplification conditions are reported
in Supplemental Table 1. Each reaction was carried out in triplicate, and each experiment was repeated 3 times.
Normalization and quantification of RT-qPCR data were
performed using the relative cycle threshold method. Data are
expressed as fold change relative to GAPDH.
Cell treatments
GH and IGF-1 responsiveness
Cells (1 ⫻ 105 cells per well in 24-well plates) were serumstarved and when 80% confluent (24 hours later) treated with
recombinant GH (500 ng/mL; Life Technologies) or IGF-1 (100
ng/mL; Life Technologies). Cells were lysed in 2⫻ Laemmli buffer (Sigma) after 0, 5, 10, 20, 30, and 60 minutes and then processed for Western blotting. In another set of experiments, cells
were serum-starved for 24 hours in the absence or presence of
recombinant FGF21 (5 ␮g/mL; Abnova) before GH or IGF-1
challenge for 20 minutes.
J Clin Endocrinol Metab, November 2014, 99(11):E2198 –E2206
free medium) for 24 hours, cells were challenged with recombinant human GH (500 ng/mL) for 30 minutes or 8, 16, or 24 hours
before RNA extraction. FGF21 was kept in the medium during
GH treatment. FGF21 concentrations lower than 0.5 ␮g/mL did
not elicit any SOCS2 or IGF-1 transcriptional response (not
shown).
To assess the effect of FGF21 in GH-induced chondrogenic
differentiation, cells that had been confluent for 72 hours were
treated with recombinant GH (500 ng/mL) in DMEM/1% FBS
in the absence or presence of recombinant FGF21 (5 ␮g/mL) for
7 days before RNA extraction. Medium was changed every 2
days.
Proliferation
Cells (1 ⫻ 104 cells per well in 96-well plates) were serumstarved for 24 hours and then incubated with or without 5 ␮g/mL
recombinant FGF21 for 24 hours. Cells were then incubated
with or without 500 ng/mL GH or 100 ng/mL recombinant
IGF-1, all diluted in DMEM/1% FBS, for 96 hours. FGF21 was
kept in the medium during GH and IGF-1 treatment. Cells were
fixed in 4% paraformaldehyde in ice-cold PBS for 30 minutes
and then stained with methylene blue (Sigma) in 0.01M boric
acid (pH 8.5) (1% wt/vol). After washes with boric acid, the dye
was extracted by incubating plates for 16 hours with a solution
consisting of 50% ethanol and 50% 0.1M HCl. Absorbance was
measured at 650 nm with a Microplate Reader (Bio-Rad).
Protein analysis
Anti-FGF21 (Abnova) and control rabbit IgG (Insight Biotechnology Limited) were cross-linked to the Thermo Scientific
Ammonolink Plus coupling resin (part of the Microlink protein
coupling kit; Pierce Thermo Scientific) following the instructions
provided. Pooled serum (100 ␮L) from VPT infants was diluted
10 times with PBS and incubated with the columns overnight at
4°C before washes and elution as reported in the kit protocol.
Aliquots of eluates were mixed with reducing or nonreducing
Laemmli buffer.
Samples were size-separated through a NuPAGE Novex BisTris 4% to 12% gradient gel (Life Technologies), along with the
Novex Sharp prestained protein ladder (Life Technologies) and
blotted onto nitrocellulose membranes (Fisher). Membranes
were incubated with blocking buffer, consisting in 5% nonfat
dry milk (Asda) in PBS containing 0.1% Tween 20 for 2 hours at
room temperature. They were then incubated with primary antibodies (information in Supplemental Table 2) diluted in blocking buffer.
After washes, membranes were incubated with goat antimouse IRDye800 and goat antirabbit IRDye680 (Life Technologies; 1:10000 dilution). Immunoblots were scanned using the
Odyssey Infrared Imaging System (LI-COR).
In other sets of experiments, 1 ␮g recombinant FGF21 was
size-separated as described above and gels silver stained (silver
stain for mass spectrometry kit; Pierce), following the instructions provided.
Study approval
Quantitative RT-PCR
Cells (1 ⫻ 105 cells per well in 24-well plates) were serumstarved for 24 hours before treatment with recombinant proteins
or VPT patient serum with a low or high endogenous FGF21.
After recombinant FGF21 treatment (0.5 or 5 ␮g/mL in serum-
This study was carried out with ethical approval (Ethics Committees of the Pohjois-Savo Health Care District, Finland, and
Camden and Islington Community Local Research Ethics Committee). Informed consent was obtained from both parents of all
study participants.
doi: 10.1210/jc.2014-1566
jcem.endojournals.org
E2201
SPSS software version 19.0 (SPSS Inc)
was used for statistical analysis. P values
⬍.05 were considered significant.
For in vitro experiments, statistical
significance was evaluated by ANOVA
and Student’s t test (Vassar Stats Online
Calculator) and P values ⬍.05 were considered significant. Data are presented as
means ⫾ SEM. Each experiment was performed in triplicate on chondrocytes
from at least 3 donors.
Results
FGF21 levels and growth failure
in VPT infants
FGF21 levels were below the sensitivity of the assay in 85% of infants
in the umbilical blood, and in 31%,
25%, and 18% of the infants at
weeks 1, 3, and 5, respectively. A significant increase in FGF21 level was
Figure 1. FGF21 levels increased by age and were significantly associated with the change in
observed from week 1 (median 33.2,
⌬SDS for length but not for weight. A, A significant increase in FGF21 level from week 1 (median
33.2, range ⬍11.3–1921.0 pg/mL) to week 3 (median 44.0, range ⬍11.3–928.2) and to week 5
range ⬍ 11.3–1921.0 pg/mL) to
(median 92.6, range ⬍11.3–2384.0 pg/mL) was observed. B and C, Association of the postnatal
week 3 (median 44.0, range ⬍ 11.3–
FGF21 level (mean of log-transformed FGF21 concentrations at weeks 1, 3, and 5) and ⌬SDS for
928.2 pg/mL; P ⫽ .001), and from
length from birth to term-equivalent age (B) and ⌬SDS for weight from birth to term-equivalent
age (C) in 32 VPT infants. Standardized regression coefficients (␤) and P values were obtained
weeks 3 to 5 (median 92.6, range
from the linear regression model adjusted for gestational age at birth, birth size, sex, nutrition,
⬍11.3–2384.0 pg/mL; P ⫽ .008)
and average postnatal IGF-1 level at weeks 1, 3, and 5.
(Figure 1A).
The clinical characteristics of our
cohort are reported in Table 1. At
Statistics
FGF21 and IGF-1 concentrations that were below the limit of
birth, the median birth weight SDS was ⫺0.2 (range ⫺2.2
detection of the assays, were substituted with a constant value
to 1.7; 2 infants had a birth weight SDS ⬍ ⫺2). At termthat was half of the limit of detection (for FGF21, 5.6 pg/mL, and
equivalent age (range 37.0 – 41.6 weeks), the median
for IGF-1, 0.92 nmol/L, respectively) to avoid their dropping out
weight SDS was ⫺1.2 (range ⫺4.3 to 0.4; 11 infants had
of the statistical analyses. To identify factors associated with
a weight SDS ⬍ ⫺2). Mean ⌬SDS (individual changes) for
FGF21 levels in VPT infants, between- and within-group comweight was ⫺1.3 (range ⫺3.8 to 0.0). Median birth length
parisons were done using a mixed-models analysis, suitable because of the potential correlation structure of the data caused by
SD was ⫺0.1 (range ⫺2.6 to 1.6; only 1 VPT infant had the
twins/triplets and repeated measurements. Gestational age, sex,
birth length SDS ⬍ ⫺2). At term-equivalent age, median
time point at sampling, birth weight SDS, birth length SDS, parlength SD was ⫺1.1 (range ⫺4.5 to 0.7; 11 infants had
enteral nutrition (yes/no), enteral nutrition (yes/no), and IGF-1
length SDS below ⫺2). Mean ⌬SDS for length (individual
concentration were included in the model as fixed effects, and the
changes) was ⫺1.5 (range ⫺4.2 to 0.2). A negative ⌬SDS
subject and pair (twins or triplets) as random effects. The FGF21
data were right-skewed and therefore transformed logarithmifor length and weight was observed in every VPT infant
cally to achieve normality of residuals in the mixed-models
after birth.
analysis.
Factors potentially contributing to growth failure after
To assess factors associated with postnatal growth failure
birth were analyzed and are shown in Table 2. Higher
from birth to term-equivalent age, linear regression analysis was
gestational age at birth was associated with a less severe
carried out. The outcome measures were ⌬SDS for weight and
length from birth to term. As explanatory factors, birth weight
growth failure postnatally (standardized regression coefSDS, birth length SDS, parenteral nutrition (yes/no), enteral nuficient for ⌬SDS for length was 0.475, P ⬍ .005, and for
trition (yes/no), gestational age, sex, IGF-1, and FGF21 were
⌬SDS for weight was 0.441, P ⬍ .005), whereas a higher
included. The average FGF21 and IGF-1 before term-equivalent
birth weight SDS was negatively associated with ⌬SDS for
age (mean of log-transformed FGF21 at weeks 1, 3, and 5 and
weight (⫺.518, P ⬍ .001) but not with ⌬SDS for length.
mean of IGF-1 at weeks 1, 3, and 5) were used in the analysis as
Need for total parenteral nutrition at week 1, 3, or 5 was
a summary measure.
E2202
Guasti et al
FGF21 in Preterm Infants
J Clin Endocrinol Metab, November 2014, 99(11):E2198 –E2206
Table 2. Associations of Gestational Age at Birth, Birth Weight and Length SDSs, Sex, Postnatal FGF21 and IGF-1
Levels, and Nutrition With the Postnatal Growth From Birth Until Term-Equivalent Age in 32 VPT Infants
⌬SDS for Length From
Birth to Term-Equivalent Age
⌬SDS for Weight From Birth to
Term-Equivalent Age
Linear
Regression Model
Standardized Regression
Coefficients
P Value
Standardized Regression
Coefficients
P Value
Gestational age at birth
Birth length, SD
Birth weight, SD
Female sex
Any enteral nutrition
Total parenteral nutrition
FGF21a
IGF-1a
.475
⫺.249
.192
⫺.035
.026
⫺.180
⫺.415
.057
.000
.055
.179
.555
.683
.006
.000
.440
.441
.273
⫺.518
⫺.208
.028
⫺.182
⫺.045
.144
.000
.059
.001
.002
.686
.012
.571
.081
a
Mean of weeks 1, 3, and 5 was used to reflect the average postnatal FGF21 (log-transformed values to normalize the distribution for the linear
regression model) and IGF-1 levels.
negatively associated with ⌬SDS for both weight and
length (⫺.182, P ⫽ .012, and ⫺.180, P ⫽ .006,
respectively).
A significant negative association between the average
FGF21 level during the first 5 weeks of life and ⌬SDS for
length (Figure 1B) was found (standardized regression coefficient -.415 and P ⬍ .005 after adjustment for gestational age at birth, birth length SDS, birth weight SDS, sex,
nutrition, and IGF-1). The association of FGF21 levels
with ⌬SDS for weight was not statistically significant
(standardized regression coefficient ⫺.045, P ⫽ .571, Figure 1C). Average IGF-1 level during the first 5 weeks of life
was not associated with ⌬SDS for length or weight (standardized regression coefficient .057, P ⫽ .440, and .144,
P ⫽ .081, respectively).
Effect of FGF21 on GH signaling in human primary
chondrocytes
We hypothesized that FGF21 acts on chondrocytes, as
shown in mouse studies (7, 8). We tested the effect of recombinant FGF21 and patients’ serum on primary chondrocyte
cultures established from pediatric patients with normal linear growth. These cells express lower levels of cartilage markers compared with cartilage, have high chondrogenic potential, and are also highly proliferative (18, 19); therefore, they
appeared to be suitable in vitro correlates of the growth plate
because in the latter, cells with high GH receptor (GHR) and
signal transducer and activator of transcription 5 (STAT5)b
levels are proliferative/prehypertrophic (20, 21). Moreover,
chondrocyte primary cultures responded canonically to recombinant GH and IGF-1 (Supplemental Figure 1).
These cells also expressed the FGF21 receptor complex
FGFR1/␤-KLOTHO (11, 12) (Figure 2A), indicating that
they can respond to FGF21. However, their FGF21 expression levels were significantly lower compared with
that of liver (Supplemental Figure 2). Suppressor of cyto-
kine signaling 2 (SOCS2) is an important negative regulator of the GH signaling cascade, and its expression is
stimulated by GH and regulated by GH-dependent STAT5
activity (22). Interestingly, a significant increase of Socs2
mRNA and Socs2 protein is found in the liver of Fgf21transgenic mice (10). In primary chondrocytes, FGF21 upregulated basal and GH-induced SOCS2 expression (Figure 2B, middle panel). SOCS1 and SOCS3 expression
(Figure 2B, left and right panels, respectively) showed a
similar trend, albeit not significant. Remarkably, patient
serum of VPT infants, containing high endogenous levels
of FGF21, increased SOCS2 expression in chondrocytes
(Figure 2C). FGF21 also inhibited GH-induced STAT5
phosphorylation but had no apparent affect on GHinduced AKT phosphorylation, whereas GH-induced
ERK1/2 phosphorylation increased upon FGF21 treatment (Figure 2D). FGF21 had no effect on IGF-1–induced
AKT and ERK1/2 phosphorylation (not shown). The increased levels of ALP and COLII expression during longterm GH treatment (23) were also attenuated by FGF21
(Figure 2E), whereas neither GH nor FGF21 had any effect
on COLX expression (Figure 2E, right panel). FGF21 also
reduced both basal and GH-induced IGF-1 expression
(Figure 2F) and inhibited GH- but not IGF-1–induced
chondrocyte proliferation (Figure 3G).
To further elucidate the fact that the concentrations of
FGF21 needed to elicit a biological response in vitro were
⬎3000 times higher than those normally seen in the circulation, we immunoisolated FGF21 from a pooled serum
sample. When eluates were subjected to Western blot with
anti-FGF21 antibody, a band with an apparent molecular
mass of ⬃50 kDa could be detected both in reducing and
nonreducing conditions (Figure 3, left panel). No bands
were present in eluates from columns conjugated with a
control antibody. An aliquot of recombinant FGF21 run
doi: 10.1210/jc.2014-1566
jcem.endojournals.org
E2203
Figure 2. FGF21 affects GH signaling in primary chondrocytes. A, RT-PCR analysis of FGFR1 IIIb and IIIc isoforms, ␤-KLOTHO, and GAPDH
expression in human chondrocytes and liver; negative controls (⫺con) were PCR samples where cDNA was omitted. Human brain cDNA was used
as positive control for both IIIb and IIIc isoforms. B, Cells were incubated with FGF21 (0.5 and 5 ␮g/mL) for 24 hours and then challenged with GH
(500 ng/mL) for 1 hour before analysis of SOCS1, SOCS2, and SOCS3 expression by RT-qPCR. C, Cells were incubated for 1 hour with VPT infants’
pooled sera with low (66.12 pg/mL) and high (746.67 pg/mL) FGF21 levels before SOCS2 RT-qPCR. The levels of GH were similar in the 2 samples
(37.5 and 38.6 ng/mL). D, Cells were incubated in the absence or presence of FGF21 (5 ␮g/mL) for 24 hours and then challenged with GH (500
ng/mL) for 20 minutes before analysis of STAT5, AKT, and ERK1/2 phosphorylation by Western blot. Data indicate the ratio of phosphorylated vs
total protein and are normalized to GH treatment alone. E, Cells were incubated with GH (500 ng/mL) with or without FGF21 (5 ␮g/mL) for 7 days
before analysis of COLII, ALP, and COLX expression by RT-qPCR. F, Cells were incubated with FGF21 (5 ␮g/mL) for 24 hours and then challenged
with GH (500 ng/mL) for 30 minutes or 8, 16, or 24 hours before analysis of IGF-1 expression by RT-qPCR. G, Cells were incubated with FGF21 (5
␮g/mL) for 24 hours and then challenged with GH (500 ng/mL) or IGF-1 (100 ng/mL) for 96 hours before methylene blue assay. The experiments
were performed on cells obtained from 3 patients, in triplicate. Error bars represent the mean ⫾ SEM. *, P ⬍ .05; **, P ⬍ .01; ***, P ⬍ .001.
E2204
Guasti et al
FGF21 in Preterm Infants
J Clin Endocrinol Metab, November 2014, 99(11):E2198 –E2206
to GH resistance have yet to be elucidated. Extrauterine growth failure
in VPT infants is a well-characterized model of GH resistance of unknown origin. Lack of association
between IGF-1 levels and postnatal
growth failure in VPT infants can be
best explained by the fact that circulating IGF-1 levels reflect only hepatic production, whereas the GH
Figure 3. Left panel, FGF21 was immunoisolated from pooled sera of VPT infants with antiresistance is generalized and is also
FGF21 antibodies. Rabbit IgG were used as control. Eluates were size-separated under reducing
(R) or nonreducing (NR) conditions and Western blotted with anti-FGF21 antibodies; 0.1 ␮g
localized in the growth plates.
recombinant FGF21 was run alongside these samples. Middle panel, Recombinant FGF21 (1 ␮g)
We have shown that FGF21 inrun under R and NR conditions was Western blotted with a different anti-FGF21 antibody. Right
hibits
GH-induced proliferation in
panel, Silver staining of 1 ␮g recombinant FGF21 run under NR conditions.
primary chondrocytes by acting downstream of GHR signaling, specifically
alongside resulted in 2 bands, a major one of ⬃25 kDa and at the level of SOCS2 expression. Because both SOCS2
a very faint one at ⬃50 kDa. This last finding was cor- activity (22) and extracellular GH levels (34) have been
roborated with a different antibody to FGF21 (Figure 3, shown to induce loss of GHR in the plasma membrane, it
center panel), with the 50-kDa band being sensitive to is intriguing that FGF21 can potentiate GH-dependent
reducing conditions. These results indicate that FGF21
SOCS2 expression in primary chondrocytes; one likely
forms dimers in vivo and that ⬎ 95% of recombinant
scenario is that the concomitant high levels of GH and
FGF21 is monomeric. The dimer could not be detected in
FGF21 might induce a state of GH resistance through
a silver-stained gel after size separation of recombinant
short-term inhibition of GH signaling and/or increased
FGF21 (Figure 3, right panel). We speculate that monoGHR turnover, eventually resulting in decreased chonmeric FGF21 has a lower bioactivity as shown for FGF2
drocyte proliferation. FGF21 also reduced GH-induced
(24 –26).
chondrocyte maturation, as seen by the attenuation of
COLII and ALP expression, suggesting that chronic high
levels of FGF21 could affect various stages of chondrocyte
Discussion
maturation/differentiation.
In this prospective clinical cohort study, we measured
Even though we detected extremely low FGF21 expresplasma FGF21 levels in VPT infants below 32 gestation sion in primary chondrocytes compared with the liver
weeks at birth and at 3 serial time points during the first (Supplemental Figure 2), the ability to express FGF21
weeks of life. Elevated FGF21 during the first 5 weeks of from the human growth plate in vivo must not be excluded
life was specifically associated with failure in linear because it has been reported in mice (8). Importantly, our
growth but not with weight gain.
data indicate that patients’ serum, containing high endogPremature infants provide an applicable model in clin- enous levels of FGF21, increased the levels of SOCS2 in
ical studies on mechanisms of growth failure in children.
chondrocytes in a dose-dependent manner, further conDespite vast improvements in neonatal intensive care,
firming mechanistic data obtained with recombinant
postnatal growth failure in VPT infants is still a frequent
FGF21.
phenomenon that is usually followed by a period of
We also report here for the first time that FGF21 likely
catch-up growth (27, 28). However, despite this, adult
forms dimers in vivo. It is well established that FGF2
height is often compromised.
A limitation of this study was that data on daily caloric dimerization is crucial to enhance its receptor binding and
and protein intake was unavailable, as well as other fac- biological activity such as the activation of downstream
tors that might contribute to the growth failure in VPT signaling (24 –26). We speculate that the almost exclusive
presence of monomeric FGF21 in the recombinant prepinfants (29 –31).
The GH–IGF-1 axis has been shown to be important for aration explains the inability of FGF21 to elicit significant
linear growth in infancy (3, 32, 33), and there is evidence cellular responses when used at physiological concentrathat impaired linear growth is due to GH resistance (nor- tions in this study and in others (8). Biochemical and crysmal or elevated GH and low IGF-1). However, the precise tallography studies, along with computerized molecular
molecular mechanisms by which chronic conditions lead docking prediction of FGF21 with FGFR1/␤-klotho will
doi: 10.1210/jc.2014-1566
help understand the relative bioactivity of monomeric vs
dimeric FGF21.
However, the need to use higher concentrations of
FGF21 in our in vitro system to trigger a biological response could have other explanations such as 1) higher
levels of FGF21 cleavage and/or 2) the inability of recombinant FGF21 to be concentrated near/around its receptor
complex in a 2-dimensional environment with low expression of extracellular matrix components (18, 19), which
are known to potentiate signaling of FGF cytokines.
GH and IGF-1 are the 2 key hormones regulating linear
growth and its associated metabolic processes. Both hormones have independent, overlapping, and complementary effects on bone, cartilage, muscle, and fat to promote
linear growth, increase lean body mass, and decrease fat
mass. Additionally, GH and IGF-1 have distinct cellular
targets at the level of the growth plate chondrocyte. For
example, in mice lacking both GHR and IGF-1, the
growth phenotype is more severe than in mice lacking
either gene (35). Because IGF-1–induced cell proliferation
is unaffected by FGF21 (unlike GH-induced cell proliferation), we propose that IGF-1 treatment could provide a
novel therapeutic strategy to increase growth in patients
with growth failure due to elevated FGF21 levels.
Although further studies both in experimental animals and humans are needed to fully elucidate the effects of FGF21 on growth at the systemic level and in the
growth plate, our findings have identified novel mechanisms of GH-resistant growth failure in humans that
could lead to development of new growth-promoting
treatment modalities.
Acknowledgments
Address all correspondence and requests for reprints to: Leo
Dunkel, Centre for Endocrinology, William Harvey Research
Institute, Barts and the London, Queen Mary University of London, Charterhouse Square, London EC1M 6BQ, UK. E-mail:
[email protected].
Disclosure Summary: The authors have declared that no conflict of interests exists.
References
1. Hack M, Schluchter M, Cartar L, Rahman M, Cuttler L, Borawski
E. Growth of very low birth weight infants to age 20 years. Pediatrics. 2003;112(1 Pt 1):e30 – e38.
2. Wehkalampi K, Hovi P, Dunkel L, et al. Advanced pubertal growth
spurt in subjects born preterm: the Helsinki study of very low birth
weight adults. J Clin Endocrinol Metab. 2011;96(2):525–533.
3. Mehta A, Hindmarsh PC, Stanhope RG, et al. The role of growth
hormone in determining birth size and early postnatal growth, using
congenital growth hormone deficiency (GHD) as a model. Clin Endocrinol (Oxf). 2005;63(2):223–231.
jcem.endojournals.org
E2205
4. Kajantie E, Dunkel L, Rutanen EM, et al. IGF-I, IGF binding protein
(IGFBP)-3, phosphoisoforms of IGFBP-1, and postnatal growth in
very low birth weight infants. J Clin Endocrinol Metab. 2002;87(5):
2171–2179.
5. Ogilvy-Stuart AL, Hands SJ, Adcock CJ, et al. Insulin, insulin-like
growth factor I (IGF-I), IGF-binding protein-1, growth hormone,
and feeding in the newborn. J Clin Endocrinol Metab. 1998;83(10):
3550 –3557.
6. Deiber M, Chatelain P, Naville D, Putet G, Salle B. Functional hypersomatotropism in small for gestational age (SGA) newborn infants. J Clin Endocrinol Metab. 1989;68(1):232–234.
7. Kubicky RA, Wu S, Kharitonenkov A, De Luca F. Role of fibroblast
growth factor 21 (FGF21) in undernutrition-related attenuation of
growth in mice. Endocrinology. 2012;153(5):2287–2295.
8. Wu S, Levenson A, Kharitonenkov A, De Luca F. Fibroblast growth
factor 21 (FGF21) inhibits chondrocyte function and growth hormone action directly at the growth plate. J Biol Chem. 2012;
287(31):26060 –26067.
9. Angelin B, Larsson TE, Rudling M. Circulating fibroblast growth
factors as metabolic regulators–a critical appraisal. Cell Metab.
2012;16(6):693–705.
10. Inagaki T, Lin VY, Goetz R, Mohammadi M, Mangelsdorf DJ,
Kliewer SA. Inhibition of growth hormone signaling by the fastinginduced hormone FGF21. Cell Metab. 2008;8(1):77– 83.
11. Yang C, Jin C, Li X, Wang F, McKeehan WL, Luo Y. Differential
specificity of endocrine FGF19 and FGF21 to FGFR1 and FGFR4 in
complex with KLB. PLoS One. 2012;7(3):e33870.
12. Ding X, Boney-Montoya J, Owen BM, et al. betaKlotho is required
for fibroblast growth factor 21 effects on growth and metabolism.
Cell Metab. 2012;16(3):387–393.
13. Walsh MC, Yao Q, Gettner P, et al. Impact of a physiologic definition on bronchopulmonary dysplasia rates. Pediatrics. 2004;
114(5):1305–1311.
14. Koletzko B, Goulet O, Hunt J, Krohn K, Shamir R. Report on the
guidelines on parenteral nutrition in infants, children and adolescents. Clin Nutr. 2005;24(6):1105–1109.
15. Agostoni C, Buonocore G, Carnielli VP, et al. Enteral nutrient supply for preterm infants: commentary from the European Society of
Paediatric Gastroenterology, Hepatology and Nutrition Committee
on Nutrition. J Pediatr Gastroenterol Nutr. 2010;50(1):85–91.
16. Sankilampi U, Hannila ML, Saari A, Gissler M, Dunkel L. New
population-based references for birth weight, length, and head circumference in singletons and twins from 23 to 43 gestation weeks.
Ann Med. 2013;45(5– 6):446 – 454.
17. Lee PA, Chernausek SD, Hokken-Koelega AC, Czernichow P; International Small for Gestational Age Advisory Board. International
Small for Gestational Age Advisory Board consensus development
conference statement: management of short children born small for
gestational age, April 24-October 1, 2001. Pediatrics. 2003;111(6
Pt 1):1253–1261.
18. Guasti L, Prasongchean W, Kleftouris G, et al. High plasticity of
pediatric adipose tissue-derived stem cells: too much for selective
skeletogenic differentiation? Stem Cells Transl Med. 2012;1(5):
384 –395.
19. Guasti L, Vagaska B, Bulstrode NW, Seifalian AM, Ferretti P. Chondrogenic differentiation of adipose tissue-derived stem cells within
nanocaged POSS-PCU scaffolds: a new tool for nanomedicine.
Nanomedicine. 2014;10(2):279 –289.
20. Gevers EF, Hannah MJ, Waters MJ, Robinson IC. Regulation of
rapid signal transducer and activator of transcription-5 phosphorylation in the resting cells of the growth plate and in the liver by
growth hormone and feeding. Endocrinology. 2009;150(8):3627–
3636.
21. Gevers EF, van der Eerden BC, Karperien M, Raap AK, Robinson
IC, Wit JM. Localization and regulation of the growth hormone
receptor and growth hormone-binding protein in the rat growth
plate. J Bone Miner Res. 2002;17(8):1408 –1419.
22. Vesterlund M, Zadjali F, Persson T, et al. The SOCS2 ubiquitin
E2206
23.
24.
25.
26.
27.
28.
29.
Guasti et al
FGF21 in Preterm Infants
ligase complex regulates growth hormone receptor levels. PLoS
One. 2011;6(9):e25358.
Maor G, Hochberg Z, von der Mark K, Heinegard D, Silbermann M.
Human growth hormone enhances chondrogenesis and osteogenesis in a tissue culture system of chondroprogenitor cells. Endocrinology. 1989;125(3):1239 –1245.
Safran M, Eisenstein M, Aviezer D, Yayon A. Oligomerization reduces heparin affinity but enhances receptor binding of fibroblast
growth factor 2. Biochem J. 2000;345(Pt 1):107–113.
Schlessinger J, Ullrich A. Growth factor signaling by receptor tyrosine kinases. Neuron. 1992;9(3):383–391.
Spivak-Kroizman T, Lemmon MA, Dikic I, et al. Heparin-induced
oligomerization of FGF molecules is responsible for FGF receptor
dimerization, activation, and cell proliferation. Cell. 1994;79(6):
1015–1024.
Ehrenkranz RA, Younes N, Lemons JA, et al. Longitudinal growth
of hospitalized very low birth weight infants. Pediatrics. 1999;104(2
Pt 1):280 –289.
Niklasson A, Engstrom E, Hard AL, Wikland KA, Hellstrom A.
Growth in very preterm children: a longitudinal study. Pediatr Res.
2003;54(6):899 –905.
Clark RH, Thomas P, Peabody J. Extrauterine growth restriction
J Clin Endocrinol Metab, November 2014, 99(11):E2198 –E2206
30.
31.
32.
33.
34.
35.
remains a serious problem in prematurely born neonates. Pediatrics.
2003;111(5 Pt 1):986 –990.
Berry MA, Conrod H, Usher RH. Growth of very premature infants
fed intravenous hyperalimentation and calcium-supplemented formula. Pediatrics. 1997;100(4):647– 653.
Marks KA, Reichman B, Lusky A, Zmora E. Fetal growth and postnatal growth failure in very-low-birthweight infants. Acta Paediatr.
2006;95(2):236 –242.
Mayer M, Schmitt K, Kapelari K, Frisch H, Köstl G, Voigt M. Spontaneous growth in growth hormone deficiency from birth until 7
years of age: development of disease-specific growth curves. Horm
Res Paediatr. 2010;74(2):136 –144.
Ranke MB, Lindberg A, Albertsson-Wikland K, Wilton P, Price DA,
Reiter EO. Increased response, but lower responsiveness, to growth
hormone (GH) in very young children (aged 0 –3 years) with idiopathic GH Deficiency: analysis of data from KIGS. J Clin Endocrinol
Metab. 2005;90(4):1966 –1971.
Deng L, He K, Wang X, et al. Determinants of growth hormone
receptor down-regulation. Mol Endocrinol. 2007;21(7):1537–
1551.
Lupu F, Terwilliger JD, Lee K, Segre GV, Efstratiadis A. Roles of
growth hormone and insulin-like growth factor 1 in mouse postnatal
growth. Dev Biol. 2001;229(1):141–162.