Human blood and semen are issued from the following cohort “Fertiprotect”, including healthy men and their respective normal semen quality, according to the WHO (2010) guidelines. Exclusion criteria include seropositivity for HIV, HBV or HCV, smoking and male explained infertility (chemotherapy, varicocele or genital surgery). Patients were enrolled into the Assisted Reproductive Centers (Tours, FERTIPROTECT protocol) for couple infertility exploration during a medical consultation. Forty participants gave full-informed written consent to participate in the study, and ethical approval was obtained from the Ethics Committee of the Vinci Clinic and CHRU Bretonneau. Patients (n = 40, 29–53 years old) ( Supplementary Data Sheet 1 ) were separated in two groups depending on the body mass index (BMI), with 18.5-25 kg/m² considered as normal BMI (and noted BMI ≤ 25 kg/m²) and BMI ≥ 30 kg/m² considered as obese ( Table 1 ). However, blood and seminal fluid were recovered from the same individual in a group of 20 patients (BMI ≤ 25, n=10; BMI ≥ 30, n=10). Blood and seminal fluid analyses were obtained in fasted patients, and semen collection was obtained after a recommendation of 2-5 days of abstinence. For in vitro analysis, recombinant human FGF21 from Sigma-Aldrich was prepared in water (100 µg/mL, stock solution). The selective FGFR1 and FGFR3 inhibitor PD173074 was obtained from Tocris Bioscience (Minneapolis, MN, USA) and used at the concentration of 100 nM (20, 21). Analysis of FGF21 was performed on fresh washed human spermatozoa from others patients. Spermatozoa were exposed for 15 min (for Western blot analysis) or 30 min to 0–10 ng/mL recombinant human FGF21, as described in the legends.

ATP concentrations was measured by using the CellTiter-Glo™ ATP Assay Kit (Promega, France), and total-cholesterol concentration was measured by using the spectrophotometric assays (Biolabo, France), according to the manufacturer’s instructions. Plasma and seminal fluid levels of FGF21 were measured using the commercial Human FGF-21 Quantikine ELISA Kit (Bio-Techne, France).

Pellets of human spermatozoa were lysed [Tris 1 M (pH 7.4), NaCl 0.15 M, EDTA 1.3 mM, EGTA 1 mM, VO43−23 mM, NaF 0.1 M, NH2PO41%, Triton 0.5%] and the protein concentration of samples was measured using a kit bicinchoninic acid (BCA) protein assay (Interchim, Montluçon, France) and equal protein concentrations were electrophoresed (40µg). Saturation of membrane was done with Tris-Buffered Saline Tween buffer (0.05% of Tween 20 and 5% of milk) for 30 min at room temperature. Then, the membranes were incubated at 4°C overnight with the following antibodies (all diluted at 1/1000): phospho-Akt (ser473), Akt, phospho-ERK (Thr202/Tyr204), and ERK (Cell Signalling Technologies, USA). Experiments were performed on five different patients.

Immunoprecipitation of FGF21 in seminal fluid was performed on a pool of human seminal fluid devoid of spermatozoa (2 mL), which was incubated overnight with rabbit polyclonal anti‐FGF21 antibody (Thermo-Fisher Scientific, USA) and immunoprecipitated by 100 mg protein G agarose beads. After several washes, the immunoprecipitated proteins and depleted seminal fluid extract were analyzed by Western blot. Experiments were performed on five pools of patients. Detection of proteins was done by using chemiluminescence (Western Lightning Plus-ECL, Perkin Elmer, Villebon-sur-Yvette, France) with a G-box SynGene (Ozyme, St Quentin en Yvelines, France).

Paraffin-embedded testis, epididymis, prostate and seminal vesicle samples were retrieved from autopsy specimens from the Department of Histopathology and Urology/Andrology of the CHRU Bretonneau Tours Hospital and Hospices Civils de Lyon, France, following approval by the ethical committees of these institutions. Sections (7 µm) of the following human tissues (testis, epididymis, prostate and seminal vesicle) were deparaffinized and rehydrated in xylene and in various baths containing decreasing concentrations of alcohol (100, 90, 75%) for 10 minutes for each step. Immunohistochemical slides were washed in a PBS bath and microwaved for 2–3 min in antigen unmasking solution (Vector Laboratories, Inc., AbCys, Paris, France). An incubation with PBS 1X/0.1 M Glycine for 15 minutes at room temperature, has been performed to ensure saturation of aldehyde groups. To permeabilize cells on sections, an incubation for 15 min with a solution of 0.1% Triton X-100 (w/v) in PBS has been done. Finally, all nonspecific binding sites have been obstructed in 2% BSA solution for 15 min. For FGF21 immunostaining, sections were incubated overnight at 4°C with PBS/1% bovine serum albumin (BSA) containing primary antibody against FGF21 (Sigma-Aldrich, USA) at a 1:100 final dilution. Then, the sections were incubated with a “ready to use” labelled polymer-HRP anti-rabbit for 30 min (DAKO Cytomation Envision Plus HRP System, Dako, Ely, UK). Visualization was achieved by incubation in a DAB peroxidase substrate solution (Invitrogen, Cergy-Pontoise, USA).

Fresh human spermatozoa were fixed in 4% paraformaldehyde (PFA)/PBS for 15 min, then washed in a PBS bath. Spermatozoa were permeabilized with PBS-Triton 0.1%, and nonspecific binding sites were blocked in 2% BSA for 15 min, then incubated for 60 min at room temperature with FGFR3 (Thermo-Fisher, USA), FGFR1, FGFR4 and FGF21 (Sigma-Aldrich, USA) and β-klotho (Thermo-Fisher, USA) antibodies at a 1:100 final dilution. Rabbit or mouse IgG (Sigma-Aldrich, USA) antibodies (Sigma-Aldrich, USA) were used as negative controls. Analyses were performed in five different patients.

Before FGF21 incubation, fresh semen samples were washed and centrifuged at 1200 rpm), resuspended in DMEM (Ref D6546, Sigma-Aldrich, USA) (4500 mg/L glucose, sodium pyruvate, and sodium bicarbonate, without L-glutamine, without albumin and with 5% serum) and counted. From each patient, 5 million spermatozoa were incubated at 37°C in a water bath with increasing concentrations of recombinant FGF21 (0, 0.01, 0.1, 1, 10, and 100 ng/mL) for 30 min, with or without preincubation with the specific FGFR inhibitor PD173074 for 15 min. Sperm motility, as a percentage of motile spermatozoa, was evaluated by using a computer-assisted semen analyzer (CASA) (Hamilton-Thorne Sperm Analyser IVOS version 12.2l, Hamilton Thorne Biosciences, USA) with a Makler Counting Chamber (0.01 sq.mm/0.01 mm Deep). Three microscopic fields were analyzed, and a minimum of 200 spermatozoa per field were evaluated. The following parameters were measured: percentage of motile sperm, percentage of progressively motile spermatozoa, average path velocity (VAP, average velocity/smoothed average position of the spermatozoa), progressive velocity (VSL, straight line distance between the beginning and the end of the track), curvilinear line velocity (VCL, average velocity measured over the actual point-to-point track followed by the cell), straightness (STR, a measure of side-to-side movement of the VCL determined by the ratio VSL/VAP × 100), linearity (LIN, a measure of the departure of the cell track from a straight line), amplitude of lateral head (µm) (ALH), and beat cross frequency (BCF) (Hz). All results are presented in Table 2 .

Sperm membrane integrity was assessed with dual fluorescent probes, SYBR-14 and propidium iodide (PI) (Live/Dead Sperm Viability Kit, InvitrogenTM, Eugene, OR, USA) and semen were analyzed by using flow cytometry (MoFlo Astrios^EQ, USA). Mitochondrial activity was determined using a 200 nM mitotracker (Orange CM-H2TMRos, Invitrogen, Fisher Scientific, France) and samples were analyzed by using flow cytometry. Twenty thousand events were collected per sample. Only sperm emitting orange fluorescence (R1, R2) were classified with a high mitochondrial membrane potential (HMMP), which is associated with mitochondrial activity. Orange fluorescence is in y axis and SSC in x axis. We have separated a R1 population with a very high orange fluorescence and a mild fluorescence population R2, from the negative control (without mitotracker).

All results are presented in Supplementary Data Sheet 4 .

The contents of the ROS hydrogen peroxide (H2O2) in human sperm (n=4 patients per condition) was measured by Ros-Glo H2O2 assay kit (Promega, Charbonnières-les-Bains, France). Two million spermatozoa previously stimulated by FGF21 were incubated with H2O2 substrate solution during 4 hours. H2O2 present in samples degradate the H2O2 substrate into Luciferin Precursor and produce luminescence which is measured with a luminometer Luminoskan Ascent (Thermo-Fisher, USA). Luminescence is correlated with the concentration of H2O2 and the ROS activity as detailed in the Ros-Glo H2O2 assay kit.

A total of 2 million spermatozoa were loaded in a 96 microwell plate for fluorescent plate reader analysis, and the kinetics of intracellular calcium measurement were performed after a 2 μM Fluo-4 AM incubation. Fluorescence was measured every 30 s. The intracellular calcium intensity was plotted as the percentage change in fluorescence (ΔF/F0, %) compared with baseline (F0).

After a 30 minutes exposition with/without FGF21, spermatozoa were incubated with 10 μM calcium ionophore A23187 to induce acrosome reaction and stained with 25 μg/mL FITC-conjugated pisum sativum agglutinin (FITC-PSA; Sigma-Aldrich) for another 30 min at room temperature. The same experiment has been performed without calcium ionophore A23187. The percentage of acrosome reaction was estimated by counting 200 spermatozoa per patient. Only spermatozoa without FITC-PSA staining or FITC-PSA staining at the equatorial segment were identified as those with acrosome reactions.

Data were tested for homogeneity of variance by Bartlett’s test and for normal distribution by the Shapiro-Wilk test. One-way ANOVAs were performed with Tukey’s multiple comparisons tests or Dunnett’s multiple comparisons tests as appropriate. Data from the CASA system were compared by the Friedman test and Dunn’s multiple comparison test. All statistical analyses were performed using GraphPad Prism 6 (La Jolla, CA, USA). The results are expressed as mean ± SEM. Values were determined to be significant when * p < 0.05, ** p < 0.01, and *** p < 0.001, **** p < 0.0001, indicating a significant difference between the groups and control (p < 0.05).

Firstly, we have investigated the presence of FGF21 in human seminal fluid by an ELISA and immunoprecipitation assay. We have analyzed the FGF21 levels in fasting conditions in blood and seminal fluid from two groups of patients using the BMI, a normal BMI in the range 18.5–25 kg/m² and with obese patients with a BMI ≥ 30 kg/m² ( Figure 1 and Table 1 ). FGF21 plasma levels were 2.4-fold higher in obese patients in comparison to control patients ( Figure 1A ). However, a similar FGF21 level was measured in the seminal fluid of both groups ( Figure 1A ). We observed that the FGF21 concentration was nearly 12 to 28 times lower in the seminal fluid as compared to that in the plasma, and the FGF21 seminal fluid/FGF21 plasma ratio was lower in obese patients ( Figure 1B ). Due to the lower levels in seminal fluid, we had confirmed the presence of FGF21 in seminal fluid after immunoprecipitation and its absence in the depleted protein extract ( Figure 1C ). No significant relationship between plasma or seminal FGF21 levels and semen parameters (semen volume, sperm concentration, motility, or sperm abnormality) was observed ( Table 1 and Supplementary Data Sheets 1 , 2 ).

Because the majority of seminal fluid proteins are produced by the epididymis, seminal vesicles and prostate, immunochemistry against FGF21 has been performed on the male reproductive tract ( Figure 2 ). The expression of FGF21 was reported in Leydig cells in the testis, in the epithelium of the epididymis, and in the seminal vesicles, with weaker staining in the epithelium of the prostate gland ( Figure 2 ). However, no staining was observed in human spermatozoa ( Figure 3B .1).

As reported in the literature, FGF21 signaling is transduced by activating the β-klotho-FGFR1c/FGFR3 complex, which stimulates the components of the MAPK and PI3K pathways, and the calcium-dependent protein. In our condition, we have confirmed, by Western blot, the presence of FGFR1, FGFR3, FGFR4 and KLB proteins in human spermatozoa ( Figure 3A ). In order to localize the FGFR complex, a confocal microscopic analysis was performed and demonstrated the colocalization of the two FGF21 receptors (FGFR1 and FGFR3) with the cofactor KLB ( Figure 3B ; 5, 8) in the middle piece of the spermatozoa, behind the head and weakly in the tail. Weak staining of the FGFR4 receptor was reported in the head and in the neck of the spermatozoa ( Figure 3B ; 9).

After 15 min of stimulation of fresh human spermatozoa with increasing concentrations of recombinant human FGF21 (0.01 ng/mL-10 ng/mL), we observed a dose-dependent increase in the phosphorylation of both Akt and ERK, which was significant at the 10 ng/mL FGF21 concentration (in comparison to control, phospho-Akt had a 4.3-fold increase and phospho-ERK had a 2-fold increase) ( Figures 4A, B ). We observed that, at the low dose of 0.1 ng/mL, phospho-Akt increased 2.6-fold, but this increase was statistically insignificant. In addition, we observed that 15 min after exposure to recombinant human FGF21, the intracellular flux of Ca²⁺ was increased in a dose-dependent manner and significantly at the 10 ng/mL concentration of FGF21 ( Figure 4C ).

To determine the effect of FGF21 on spermatozoa, we incubated fresh human spermatozoa with recombinant FGF21 for 30 min. The investigation of time effect was performed on spermatozoa with progressive motility, as presented in Supplementary Data Sheet 3A , showing that 30 min is the optimal time. No consequence on the viability of sperm after FGF21 stimulation was measured by SYBR-14 or propidium iodide staining ( Supplementary Data Sheet 3B ).

We have analyzed the effect of FGF21 on the motility of sperm motility after exposing human spermatozoa to FGF21 for 30 min at 37°C. Compared to the control condition, FGF21 was able to significantly increase progressive motility at the 0.1 ng/mL and higher concentrations of FGF21 ( Figure 5A ). Preincubation with the FGFR inhibitor PD173074 was able to eliminate the stimulatory effect induced by 10 ng/mL FGF21 ( Figure 5A ). Moreover, the average velocity and curvilinear line velocity of the spermatozoa determined by VCL and VAP were also improved by about 20% to 27% (0.1-10 ng/mL FGF21) and were returned to control values if a preincubation with PD173074 was performed ( Table 2 ). Because motility is highly associated with mitochondria activity, the ATP level and mitochondrial membrane potential were analyzed. Despite spermatozoa still having a high mitochondrial membrane potential (control: 45.71% ± 1.89; FGF21: 0.1 ng/mL: 46.33% ± 2.12; FGF21: 1 ng/mL: 47.19% ± 2.19; FGF21: 10 ng/mL: 46.72% ± 2.15) ( Supplementary Data Sheet 4 ), a significant increase of both the ATP and cAMP levels in the spermatozoa was measured at the 10 ng/mL FGF21 concentration as compared to those in the control ( Figures 5B, C ).

Furthermore, we did not observe any consequence of FGF21 exposure to the already high percentage of acrosome-reacted sperm induced with by calcium ionophore [ Figure 5D -(2)] or without [ Figure 5D -(1)]. But, FGF21 improved oxidative stress by reducing the levels of reactive oxygen species (ROS) in a dose-dependent manner in human sperm ( Figure 5F ). Moreover, the stimulation of sperm by FGF21 induced a significant increase in cholesterol levels ( Figure 5E ).