Fibroblast growth factor 21 in breast milk controls neonatal intestine function

FGF21 is a hormonal factor with important functions in the control of metabolism. FGF21 is found in rodent and human milk. Radiolabeled FGF21 administered to lactating dams accumulates in milk and is transferred to neonatal gut. The small intestine of neonatal (but not adult) mice highly expresses β-Klotho in the luminal area. FGF21-KO pups fed by FGF21-KO dams showed decreased expression and circulating levels of incretins (GIP and GLP-1), reduced gene expression of intestinal lactase and maltase-glucoamylase, and low levels of galactose in plasma, all associated with a mild decrease in body weight. When FGF21-KO pups were nursed by wild-type dams (expressing FGF21 in milk), intestinal peptides and digestive enzymes were up-regulated, lactase enzymatic activity was induced, and galactose levels and body weight were normalized. Neonatal intestine explants were sensitive to FGF21, as evidenced by enhanced ERK1/2 phosphorylation. Oral infusion of FGF21 into neonatal pups induced expression of intestinal hormone factors and digestive enzymes, lactase activity and lactose absorption. These findings reveal a novel role of FGF21 as a hormonal factor contributing to neonatal intestinal function via its presence in maternal milk. Appropriate signaling of FGF21 to neonate is necessary to ensure optimal digestive and endocrine function in developing intestine.


Control females
3 Supplemental Table 2 independent assays from 5-month-old male FGF21-KO mice and WT controls, that had been fed by FGF21-KO dams or WT dams, respectively. P values for statistical comparison are shown.

Acquisition and analysis of rodent milk
To study the concentration of FGF21 in milk, adult female Swiss mice and Wistar rats were mated with appropriate males. After delivery, all litters were adjusted to 9-10 pups. Milk samples were collected from dams on day 15 of lactation. Nursing dams were separated from their pups for 12 h (to ensure that the mammary glands were full of milk), anesthetized with isoflurane and manually milked. Milk samples were stored frozen at -20ºC. For analysis, each sample was centrifuged (13000 rpm for 10 min at 4ºC) for removal of the fat layer. Blood samples were collected from the end of the tail. The FGF21 protein levels in milk and plasma from mice and rats were determined by ELISA (RD291108200R, BioVendor R&D, Brno -Modřice, Czech Republic).
The milk composition was determined as follows. Triglycerides were measured using a commercial kit (Sigma-Aldrich). The protein content was measured by

Measurement of FGF21 in human breast milk
Mature breast milk (sampled at least 1 month after parturition) and blood samples were obtained from donors who voluntarily consented to participate in the study.
The study subjects comprised 22 healthy women between 27 and 37 years of age.
According to the time of lactation, milk samples were divided in colostrum (day 1 to 5 after birth) (n=4), transitional milk (day 6 to 21) (n=4), fully mature milk under 6 months (n=10) and fully mature milk over 6 months (n=4). For analysis of FGF21 analysis, skim milk was prepared by centrifugation of whole milk at 13000 rpm for 10 min at 4ºC. Human FGF21 protein levels in milk and serum were determined by ELISA (RD191108200R, BioVendor R&D). This portion of the study was approved by Bioethics Committee at the University of Barcelona, Spain. and RNA was purified as described in the below section "RNA isolation, cDNA synthesis, and real-time PCR". Three liver samples obtained from biopsies were used as positive controls of FGF21 expression. In both cases, informed consent from all subjects were obtained and the study was approved by the Bioethics Committee at the University of Barcelona, Spain.

Studies in lactating mice
Lactating dams (days 4, 8, 15 and 21 after delivery) and virgin female mice of the same age (controls) were sacrificed by decapitation. Blood was collected, plasma was obtained, and mammary gland, liver, periovaric WAT and interscapular BAT were dissected and frozen in liquid nitrogen for further analysis.
Pups were killed by decapitation; blood was collected for preparation of serum; and tissues were immediately frozen in liquid nitrogen, and stored at -80ºC until 13 processing. For comparison, tissues were also obtained from adult mice (male, 5 month-old).
To study the specific localization of β-Klotho expression in the intestinal mucosa, small intestines were removed from 15 day-old lactating mice. Samples of the duodenum, jejunum and ileum were cut longitudinally to expose the luminal areas, and a blade was used to scrape the mucosa and separate it from the rest of the tissue, following the previously reported procedure (15). Unprocessed tissue (total), mucosa and tissue remaining after scraping were frozen for RNA isolation and gene expression analysis. and frozen for further processing.

Intestinal explant incubation
Jejunum samples were collected from 8 day-old pups, divided into 0.5 cm sections, cut longitudinally to expose the mucosa, and incubated with DMEM with or without 10 nM mouse recombinant FGF21 (RD272108100, BioVendor R&D) at 37 °C in a humidified 95% air/5% CO 2 incubator. At the indicated times (10 min, 20 min and 3h), tissue and media samples were collected together and centrifuged, and pellets were frozen for subsequent RNA isolation (for gene expression studies) or protein homogenization (for Western blot analysis).

RNA isolation, cDNA synthesis, and real-time PCR
Dissected tissues were homogenized using an IKA ® T25 digital ULTRA-TURRAX (Staufen, Germany), and total RNA was isolated using a column-affinity based methodology (NucleoSpin RNA II; Macherey-Nagel, Düren, Germany). Total RNA was isolated from rodent samples (tissues, scraped intestinal mucosa, and intestine explants), as well as from cells that had been isolated from human breastmilk using a previously described method (28

Immunohistochemical detection of β-Klotho
During tissue collection, intestinal segments were fixed with 4% formalin for 24 h and then stored in ethanol 70% at 4 ºC until paraffin infiltration was performed.
Paraffin-embedded tissues were sectioned to obtain transverse and longitudinal orientations, and the sections were mounted on glass slides. The samples were incubated with an anti-β-Klotho antibody (LS-B3568, LSBio, LifeSpan BioSciences Inc., Seattle, WA) in a humidified chamber overnight at 4 ºC and then incubated with an ABC-complex-conjugated secondary antibody, and the results were visualized with DAB. Finally, samples were stained with hematoxylin and photographed under a microscope (50x magnification).

Quantification of intestinal peptides in plasma
The concentrations of glucagon-like peptide 1 (GLP-1), gastric inhibitory peptide (GIP) and peptide YY (PYY) in plasma samples from the various pups were determined by MILLIPLEX ® MAP Mouse Metabolic Disease Multiplex Assay (Cat. No. MMHMAG-44K, Millipore).

Lactase activity
Lactase activity was assessed with a slight modification of the method reported by Dahlqvist (29). Approximately 25 mg of duodenum, jejunum or ileum were homogenized in 6 volumes of NaCl 0,9%. The homogenates were incubated with 4 volumes of 150 mM lactose in 0,1 M maleate buffer (pH 6,5) supplemented with 0,2 mM of the protease inhibitor, 4-chloromercuribenzoic acid, for 20 minutes at 37 ºC. The reaction was stopped with 6% perchloric acid, and samples were centrifuged at maximal speed. The pellet was removed, the supernatant was neutralized with 30% KOH, and the glucose concentration was determined with the Glucose Assay Reagent (Sigma-Aldrich). Enzymatic activity was normalized with respect to the quantity of input protein.

Lactase absorption rate measurement
Lactase absorption rate was assessed by quantifying the exhalation of 14 CO 2 produced from the oxidation of 14 C-lactose infused to gut, following a previously described method (30) adapted to mouse neonates. Briefly, after 4 hours of oral infusion of 100 ng/ml FGF21 (described above in Studies in neonatal mice and animal experimental designs), pups were infused with 2.5 µC i 14 C-lactose (MC-1466; Hartmann Analytic GmbH, Braunschweig, Germany) dissolved in 100 µl of water and maintained in a chamber which contained a 6 cm 2 Whatman ® paper (GE Healthcare, Little Chalfont, UK) impregnated with phenylethylamine (Sigma), in order to capture the radiolabeled CO 2 breathed by the pups after lactose oxidation.
Papers were changed every 15 minutes during the first half hour after 14 C-lactose administration and every 30 minutes during the following 2.5 hours, in order to obtain serial quantifications. CO 2 paper traps were placed in scintillation vials containing 5 ml of scintillation fluid (Ecoscint TM H, National Diagnostics, Atlanta, GA) and the samples were counted using a Packard 2100TR TriCarb Liquid Scintillation Counter (Packard Instrument Company Inc., Meriden, CT). The linear rate of 14 CO 2 counts appearance was calculated.

Statistics
All results are expressed as means ± SEM. Differences were tested for statistical significance using unpaired t-test, one-way analysis of variance (ANOVA) with Tukey's Multiple Comparison Test, or two-way ANOVA with Bonferroni post-test, as appropriate. 20