Evaluation of in vivo bioactivities of recombinant hypo- (FSH^21/18) and fully- (FSH²⁴) glycosylated human FSH glycoforms in Fshb null mice

Abstract

The hormone - specific FSHβ subunit of the human FSH heterodimer consists of N-linked glycans at Asn⁷ and Asn²⁴ residues that are co-translationally attached early during subunit biosynthesis. Differences in the number of N-glycans (none, one or two) on the human FSHβ subunit contribute to macroheterogeneity in the FSH heterodimer. The resulting FSH glycoforms are termed hypo-glycosylated (FSH^21/18, missing either an Asn²⁴ or Asn⁷ N-glycan chain on the β - subunit, respectively) or fully glycosylated (FSH²⁴, possessing of both Asn⁷ and Asn²⁴ N-linked glycans on the β - subunit) FSH. The recombinant versions of human FSH glycoforms (FSH^21/18 and FSH²⁴) have been purified and biochemically characterized. In vitro functional studies have indicated that FSH^21/18 exhibits faster FSH- receptor binding kinetics and is much more active than FSH²⁴ in every assay tested to date. However, the in vivo bioactivity of the hypoglycosylated FSH glycoform has never been tested. Here, we evaluated the in vivo bioactivities of FSH glycoforms in Fshb null mice using a pharmacological rescue approach. In Fshb null female mice, both hypo- and fully-glycosylated FSH elicited an ovarian weight gain response by 48h and induced ovarian genes in a dose- and time-dependent manner. Quantification by real time qPCR assays indicated that hypo-glycosylated FSH^21/18 was bioactive in vivo and induced FSH-responsive ovarian genes similar to fully-glycosylated FSH²⁴. Western blot analyses followed by densitometry of key signaling components downstream of the FSH-receptor confirmed that the hypo-glycosylated FSH^21/18 elicited a response similar to that by fully-glycosylated FSH²⁴ in ovaries of Fshb null mice. When injected into Fshb null males, hypo-glycosylated FSH^21/18 was more active than the fully-glycosylated FSH²⁴ in inducing FSH-responsive genes and Sertoli cell proliferation. Thus, our data establish that recombinant hypo-glycosylated human FSH^21/18 glycoform elicits bioactivity in vivo similar to the fully-glycosylated FSH. Our studies may have clinical implications particularly in formulating FSH-based ovarian follicle induction protocols using a combination of different human FSH glycoforms.

1. Introduction

Follicle-stimulating hormone (FSH) is synthesized in and secreted from pituitary gonadotropes [1, 2]. FSH and other pituitary (LH and TSH) and placental (hCG) glycoprotein hormones contain two non-covalently associated heterodimeric α– and β–subunits [1, 2]. Both the common α–subunit and hormone-specific FSHβ, possesses up to 2 N-linked glycan chains [1-3]. The co-translational addition of GlcNAC₂ Man₉Glc₂ precursors to conserved Asn residues and subsequent post-translational conversion of each precursor to complex-type glycan structures constitute a multi-enzyme driven process termed N-glycosylation that occurs in the ER-Golgi compartments [1-3]. Recent structural studies revealed differences in N-glycan chain number, sugar composition and branching pattern of the mature human FSH purified from pooled pituitary glands, pooled postmenopausal urine, or produced in vitro using a CHO cell expression system [4-11].

Previous studies have identified differences in the abundance of different FSH glycoforms as a function of age [12-16]. It was predicted that such age-related FSH glycoforms act differently in different target cells [4, 14, 17]. Direct purification of FSH glycoforms from human pituitaries and/or urine collected at different ages and their characterization indeed identified hypo-glycosylated FSH is a mixture of FSH¹⁸ and FSH²¹ glycoforms present as the dominant form in young age and fully-glycosylated FSH, designated as FSH²⁴ is the dominant form in old age [4, 6, 14, 17-20]. Recombinant expression in a heterologous GH₃ cell culture system followed by purification and biochemical characterization revealed that these FSH glycoforms differ in the number or location of N-linked glycan chains on FSHβ subunit [4-7, 14]. While both N-linked glycans are present at Asn⁷ and Asn²⁴ in the FSHβ subunit of fully-glycosylated FSH²⁴, a single N-linked glycan is present only at Asn⁷ in the FSHβ subunit of FSH²¹ or at Asn²⁴ in the FSHβ subunit of FSH¹⁸. The latter two glycoforms are collectively known as hypo-glycosylated FSH and referred to as FSH^21/18 [4-7, 14, 19].

Both the pituitary-derived preparations and GH₃ cell-derived recombinant FSH glycoforms have been biochemically characterized in vitro and compared to fully-glycosylated FSH²⁴ with regard to receptor binding and dissociation kinetics, glycan composition by mass spectrometry and in vitro bioactivity using radio-receptor assays [4-7, 14, 18, 19]. More recently, these FSH glycoforms have also been tested in an FSH-responsive KGN granulosa cell line [17]. However, the in vivo bioactivities of GH₃-derived recombinant human FSH glycoforms have not yet been directly tested. Moreover, whether hypoglycosylated FSH (FSH^21/18) elicits bioactivity in vivo is not yet known. Here, we have used a pharmacological rescue approach and injected the GH₃ cell-derived recombinant human FSH^21/18 and FSH²⁴ glycoforms into our well characterized Fshb null mice [21-24] and directly assayed ovarian and testis gene expression changes as a measure of glycoform in vivo bioactivity. We demonstrate that recombinant hypo-glycosylated FSH^21/18 glycoform elicits FSH bioactivity in vivo in the absence of endogenous FSH.

2. Materials and Methods

2.1. Production of GH₃-derived recombinant human FSH glycoforms

Recombinant human FSH glycoforms were purified from rat pituitary GH₃ clones that co- express fully-glycosylated hFSH²⁴ and both hypo-glycosylated glycoforms hFSH¹⁸ and hFSH²¹ as a mixture (Fig. 1A). The purification and characterization of the hormones from GH₃-conditioned media were described in detail elsewhere [6, 17, 18]. Western blot analysis of FSH preparations was performed with the FSHβ subunit-specific primary antibody RFSH20 (1:5000) that detected two bands corresponding to 18 KDa and 21 KDa in the FSH^21/18 preparation (Fig.1B, Lane 1), and a single band at 24KDa in the case of FSH²⁴ and hFSH AFP 7298 reference preparation (Fig. 1B, Lanes 2 and 3) as previously reported [6, 17, 18]. An independent blot similarly loaded as above with FSH preparations, was probed with an α-subunit antiserum, which detected a band in all the lanes (Fig. 1C). One μg aliquots of these purified and in vitro validated recombinant hormones were frozen and stored at −80C, thawed rapidly when required and immediately injected into mice. For the sake of simplicity, FSH^21/18 was referred to as FSH²¹, the predominant glycoform in the FSH^21/18 mixture in all Figures.

Fig. 1.

A summary of N-linked sugar chains attached at the Asn residues in human FSH subunits and the FSH glycoform nomenclature (A). Western blot analysis (B) using a mouse anti-human FSH monoclonal antibody (RFSH-20) indicates the molecular size of purified recombinant GH₃-derived FSH^21/18 and FSH²⁴ glycoforms (Lanes 1 and 2). Purified human pituitary FSH reference preparation (AFP 7298) was used as positive control (Lane 3). An independent blot with FSH samples loaded in the same order as in panel B, was probed with an anti-alpha 15-2 monoclonal antibody (C).

2.2. Fshb null mice

Fshb null mice were generated and genotyped by genomic PCR assays performed on tail DNA samples as described before [23, 24]. For the in vivo bioactivity testing experiments, immature Fshb null female mice at 21d-22d of age or Fshb null male pups at postnatal day 3 or 5 were used. All mice were maintained under 12-h dark, 12-h light cycles with food and water supplied ad libitum. All studies with mice were performed in accordance with the Guide for the Care and Use of Laboratory Animals per the NIH instructions and approved institutional protocols.

2.3. Dose response study in the female

For the dose response study, Fshb null females were injected (i.p.) with different doses (0.5 μg, 1 μg or 2 μg) of purified recombinant hormones prepared in 100 μL of PBS or with 100 μL of PBS alone. These doses correspond to ~ 5, 10 and 20 IU respectively of the standard reference human FSH preparation. Ovaries in each case were collected at different time points for either RNA, Western blot or immunofluorescence analysis as described below.

2.4. Time course study in the female

Three sets of experiments were designed. In the first set, mice were injected (i.p.) with different doses of purified recombinant hormones as described above, and ovaries were collected after 2h, 4h, 8h, 16h and 24h. In the second set, ovaries were collected after 0.5h, 1h and 2h. In the final set, ovaries were collected after 5 min, 10 min., and 20 min. For all experimental sets, ovaries, upon removal were immediately flash frozen on dry ice until further use and processed for either RNA or protein analysis.

2.5. In vivo bioassays in the male

For checking in vivo bioactivities of FSH glycoforms, the protocols indicated in Supplementary Fig. S1 were used. In one set of experiments (Supplementary Fig. S1 A, B), beginning postnatal day 5, pups were injected (i.p.) daily with 25 μL of either PBS or 1 μg of recombinant hormone (FSH^21/18 or FSH²⁴) for 5 days until postnatal day 9. On postnatal day 10, testes were collected and the wet weights recorded. One testis was fixed in formalin reagent for histological and immunochemical analysis and the other was snap frozen and processed for RNA isolation as described [24, 25]. In a second set of experiments (Supplementary Fig. S1 C), beginning postnatal day 3, mice were injected with PBS or glycoforms for 3 days - once daily on days 3, 4 and 5. On the morning of day 5, mice were injected with BrdU and after 2 h, blood and testes were collected and processed as described above.

2.6. RNA isolation, cDNA synthesis and Taqman Real-time qPCR assays

Total RNA was isolated from mouse ovaries or testis by RNEasy micro columns (Qiagen) and RNA quantified using a NanoDrop spectrophotometer (ThermoScientific) set at 260 nm. Approximately 500 ng-1 μg total RNA was reverse transcribed by the oligo dT method using the SuperScript III kit (ThermoScientific) as described [24-26]. Taqman real time qPCR assays were performed on the cDNA samples in triplicate using custom-made or pre-inventoried primer/combo mixes (ThermoScientific or IDT). Expression of Ppil1 was used as an internal control and the relative amounts of mRNA expression were calculated as described [24-26].

2.7. Histological analysis

Testes and ovaries were processed, later paraffin-embedded and 6 μm sections were cut and stained with either PAS reagent/ hematoxylin or hematoxylin-eosin for histological analysis as described before [24-26]. Quantification of testis tubule size was done on PAS reagent / hemtoxylin – stained testis section images that were digitally captured as described [21]. Per group, three mice were used and ~ 100 tubules were counted in each case.

2.8 Immunofluorescence

Formalin-fixed and paraffin-embedded tissue sections of approximately 6μm thickness were cut and processed for immunofluorescence as described before [24-26]. The following primary antibodies were used: rabbit polyclonal antibody against Sox9 (gift from Dr. K. Morohashi, 1:1,000), rat monoclonal antibody to GCNA (gift from Dr. George Enders, used undiluted) mouse anti BrdU monoclonal antibody (Roche, 1:100), rabbit anti phospho-CREB (Cell Signaling Systems) or rabbit anti phospho-PKA substrate (Cell Signaling Systems, 1:500) both at a dilution of 1:500. The sections were later incubated with the appropriate secondary antibodies conjugated with Alexa fluors (Invitrogen) and/or a nuclear dye and visualized using an epifluorescence microscope as described [24-26]. GCNA⁺ germ cells were counted in ~ 100 tubules per group and three mice were used per group.

2.9. Western blot analysis

Gonadal proteins were extracted in immuno-precipitation buffer (Santa Cruz Biotechnology) and collected by centrifugation at 10,000 X g at 4C. Protein content was estimated by Bio-Rad protein assay using bovine serum albumin as standard. Aliquots of 10-15 μg of protein were denatured in SDS-PAGE sample buffer (final concentration = 32 mM Tris-HCl, pH 6.8, 12.5% glycerol (v/v), 1% SDS, and 31 μM β-mercaptoethanol) at 100C for 5 min., separated on 12 % polyacrylamide gels, and transferred onto PVDF membranes as described [23-25]. The membranes were blocked in 5 % non-fat dry milk and incubated with primary antibodies (Cell Signaling Systems, 1:2, 000) at room temperature for 4-5 h, washed in 0.1 % Tween-20 containing buffer, incubated in goat anti-rabbit HRP conjugate (1:4,000) as described [23-25]. The antigen-antibody complexes were visualized by an enhanced chemiluminescence, ECL III detection kit (GE Health Care). Densitometry was performed using the NIH’s ImageJ software.

2.10. Serum estradiol assay

Serum estradiol was measured using a mouse ELISA kit (Calbiotech, ES180S-100) according to the manufacturer’s instructions. The concentration of the lowest standard measured was 3 pg/ml.

2.11. Statistics

For each experiment, 4-5 female Fshb null mice per group or 5-6 males were used. Data were reported as Mean ± SEM. Statistical analysis was performed by Graph Pad Prism software and included Student’s T-test or one-way ANOVA followed by Turkey’s post-hoc test. A P value < 0.05 was considered significant.

3. Results

3.1. Hypo (FSH^21/18)- and fully (FSH²⁴)-glycosylated FSH glycoforms elicit dose-and time-dependent changes in gene expression in ovaries of Fshb null female mice

FSH is essential for ovarian folliculogenesis and Fshb null female mice develop a preantral stage block in ovarian folliculogenesis [21-24]. Many FSH-responsive genes remain suppressed in the ovaries of Fshb null mice [24, 27, 28]. In addition to the expression of well-known FSH-responsive genes, Cyp19a1, Inhba and Inhbb, we evaluated Apaf1, S100g, Dok1, Kcnj8 marker genes. These marker genes were chosen because, the corresponding mRNAs were markedly upregulated in ovaries of Fshb null mice that were either pharmacologically [27, 29] or genetically [24] rescued. To test the in vivo bioactivities of FSH^21/18 and FSH²⁴, we first ip injected 1 μg of the recombinant hormones separately (~ equivalent to 10 IU based on highly purified human FSH Reference Standard, with an estimated potency of 8.6 -10 mIU/ng) and tested the ovarian gene responses after 2h, 4h, 8h, 16h and 24h by Taqmman real-time qPCR assays. Expression of 5 (Cyp19a1, Inhba, Inhbb, Ccnd1 and Ki67) out of the 9 (56 %) genes tested was significantly up-regulated by 2h compared to that at all other time points analyzed (Fig. 2). The induction of this set of genes was similar irrespective of whether FSH^21/18 or FSH²⁴ glycoform was injected into Fshb null female mice (Fig. 2). In the absence of FSH, Fshr expression is normally up-regulated and injection of either FSH glycoform suppressed the expression after 2h (Fig. 2). In order to determine whether 2h represented the optimal time-point, we next tested the response to injection of a fixed 1μg dose of the recombinant hormones by assaying ovarian gene expression after 0.5h, 1h and 2h (Fig. 3). We first compared the responses to FSH²¹ and FSH²⁴ with respect to time (after 2h) in a single qPCR assay and measured the expression of a set of 9 known FSH-responsive genes (Fig. 3).

Fig. 2.

Ovarian gene expression analyzed by Taqman real time qPCR assays in Fshb null female mice injected with recombinant FSH glycoforms at 1μg dose. Gene expression was assayed between 2h – 24h at different time points. For each time point, triplicate cDNA samples from ovaries from 4-5 mice were used. * P < 0.05 vs. PBS-injected control group. The genes analyzed are listed in the Supplementary Material Table - 1.

Fig. 3.

Ovarian gene expression analyzed by Taqman real time qPCR assays in Fshb null female mice injected with recombinant FSH glycoforms at 1μg dose. Gene expression was assayed at different time points between 0.5h – 2h as indicated. For each time point, triplicate cDNA samples from ovaries from 4-5 mice were used. * P < 0.05 vs. PBS-injected control group and ** P < 0.05 vs. the other glycoform at the same time point. The genes analyzed are listed in the Supplementary Material Table - 1.

Based on the initial gene expression profiling described above, we measured the ovarian gene expression more comprehensively by testing expression of 11 additional genes thus, evaluating the response of 20 genes to a 1 μg dose of FSH^21/18 or FSH²⁴ and plotted the data separately with respect to different time points 0.5h, 1h and 2h, after injection (Fig. 4). When the expression changes of 60 genes (3 time points and 20 genes per time point) were collectively compared, it was evident that the FSH^21/18 glycoform induced 14 genes significantly higher than the FSH²⁴ glycoform (23 %). The FSH²⁴ glycoform induced 16 genes significantly higher than FSH^21/18 (27 %) and 30 genes (50 %) were nearly equally induced by both FSH glycoforms (Fig. 4). At a fixed time point of 2h, changing the dose of the hormones towards either the higher (2 μg) or lower (0.5 μg) end did not result in significantly higher expression responses in the majority of the ovarian genes than achieved with 1 μg of the hormones (Supplementary Fig. S2). Furthermore, two independent batches of the GH₃-derived recombinant FSH glycoform preparations were tested for their in vivo bioactivity. We found that both elicited identical ovarian gene responses in Fshb null female mice indicating no batch-to batch variation in bioactivities of the FSH glycoform preparations (data not shown). Together, these data indicate that hypo - glycosylated FSH^21/18, like fully-glycosylated FSH²⁴ is bioactive in vivo and each FSH glycoform elicits temporally distinct gene responses in ovaries of Fshb null mice.

Fig. 4.

Ovarian gene expression analyzed by Taqman real time qPCR assays in Fshb null female mice injected with recombinant FSH glycoforms at 1μg dose. Expression of 11 known FSH-responsive marker genes was analyzed 0.5h, 1h and 2h after the FSH glycoform injection. Ovarian gene expression was separately compared for each time point. For each marker expression assay, triplicate cDNA samples from ovaries from 4-5 mice were used. * P < 0.05 vs. PBS-injected control group and ** P < 0.05 vs. the other glycoform. The genes analyzed are listed in the Supplementary Material Table - 1.

3.2. Differences in intra-ovarian signaling pathways regulated by FSH^21/18 and FSH²⁴ glycoforms

FSH binds to and signals via GPCR-coupled FSH-receptors (FSHR) expressed on ovarian granulosa cells [30-32]. FSHR-mediated signaling leads to regulation of multiple downstream signaling pathways in granulosa cells including the CREB-dependent protein kinase-A (PKA) pathway [30, 33-35]. To determine if FSH^21/18 and FSH²⁴ glycoforms elicit these responses in Fshb null females, first we tested ovarian expression of the activated forms of phospho-CREB and phospho-PKA substrate by qualitative immunofluorescence using specific antibodies. At a 1μg dose, both FSH^21/18 and FSH²⁴ glycoforms induced expression of phospho-CREB and phospho-PKA substrate in ovarian granulosa cells compared to those in PBS-injected controls (Fig. 5). This induction occurred robustly at 30 minutes (Fig. 5 A, B) remained elevated at 1h (Fig. 5 C, D), and by 2h, minimal expression was observed in granulosa cells (Fig. 5 E, F).

Fig. 5.

Immunolabeling of ovarian sections with anti-phospho-CREB (A, C and E) and anti-phospho-PKA substrate (B, D and F) antibodies indicate activation of FSH-receptor mediated signaling in ovaries of FSH glycoform injected Fshb null mice. Each FSH glycoform was injected at 1 μg dose and ovaries were harvested after 0.5h (A and B), 1h (C and D) and 2h (E and F), formalin-fixed and processed. Merged images are shown: specific antibody staining was visualized in green and the nuclei were stained red. Multiple sections from ovaries of 3 mice per group and per each time point were analyzed. Bottom panels represent enlarged images of follicles present in white squares in upper panels. White bar represents 100 μm.

We next tested expression of several FSH-R downstream signaling pathway components in ovarian lysates by Western blot analysis followed by densitometry. At a1μg dose, both FSH^21/18 and FSH²⁴ glycoforms induced phospho-CREB and phospho-PKA substrate 30 min., 1h and 2h after injection, compared to that in PBS-injected controls (Fig. 6). Both glycoforms also significantly induced the expression of phospho-Akt at the three time points tested when compared to PBS-injected controls (Fig. 6). The maximal induction of phospho-Akt was achieved after 2h by treatment with hypo-glycosylated FSH^21/18 (Fig. 6 E, F). In contrast, striking differences were observed when induction of phosoho-p38 and phospho-p44/42 was compared. First, the hypo-glycosylated FSH^21/18 glycoform significantly induced phosoho-p38 after 30′ compared to fully-glycosylated FSH²⁴ glycoform (Fig. 6 B). Hypo-glycosylated FSH^21/18 glycoform significantly induced phospho-p44/p42 only after 1h (Fig. 6D). Thus, the hypoglycosylated FSH^21/18 glycoform after 2h of treatment, significantly up-regulated expression of 4 of the 5 activated forms of FSH-R downstream signaling components in ovaries of Fshb null mice (Fig. 6F).

Fig. 6.

Western blot analysis of ovarian extracts probed with various antibodies against activated forms of signaling components downstream of FSH-receptor (A, C and E). Extracts were prepared from Fshb null mice injected with FSH glycoforms after different times (0.5h, 1h and 2h). Western blot analysis was performed on ovarian extracts from 2-3 mice per group and per time point. Expression of β-tubulin was used as an internal control and relative expression of each phospho-protein was plotted on the Y-axis. Densitometry data (B, D and F) represents an average of 3 independent blots. * P < 0.05 vs. PBS injected group and ** P < 0.05 vs. the other glycoform.

To test whether induction of FSH-R downstream signaling components by FSH glycoforms occurred before 30 minutes, we tested their expression at 5, 10 and 20 minutes after FSH glycoform injection. We found no significant induction of responses in the ovaries of Fshb null mice during these early time points after injection of the FSH glycoforms (Supplementary Fig. S3), although a trend towards upregulation was observed. Despite a better signaling response by the FSH^21/18 glycoform, both glycoforms induced a similar ovarian weight gain response (Fig. 7A). Moreover, ovarian histology by PAS/hematoxylin staining indicated that 1 μg of either FSH^21/18 or FSH²⁴ glycoform induced antrum formation by 48h (Fig. 7 B-E). Interestingly, serum estradiol was significantly higher in FSH^21/18 glycoform - injected immature Fshb null females compared to that in either FSH²⁴-injected immature Fshb null females or PBS-injected immature control mice (Fig. 7F). Collectively, all the above data indicate that hypo-glycosylated FSH^21/18 is biologically active in inducing FSH-R downstream signaling pathway components and estradiol production from ovaries of Fshb null mice, similar to that by fully-glycosylated FSH²⁴.

Fig. 7.

Loss of FSH results in hypoplastic ovaries and both the FSH^21/18 and FSH²⁴ glycoforms elicit the ovarian weight gain response in Fshb null mice (A). Ovarian histology shows antrum in sections obtained from control (Ctrl.) (B) and FSH glycoform-injected (D and E) but not PBS-injected (C) Fshb^−/− females. Serum estradiol was significantly high in FSH^21/18 injected immature Fshb null mice compared to either FSH²⁴ - injected Fshb null females or PBS- injected control mice (F). * P < 0.05 vs. PBS-injected Ctrl, ** P < 0.05 vs. PBS-injected Fshb^−/−, n=6 mice per group. Black bar represents 200 μm.

3.3. In vivo bioactivities of FSH^21/18 and FSH²⁴ glycoforms in Fshb null male mice

FSH binds to G-protein coupled FSH-receptors on Sertoli cells in the testis. FSH regulates Sertoli cell proliferation during pre-pubertal testis development [36-38]. Adult Fshb null male mice display reduced testis weight as a result of fewer Sertoli cells [39, 40]. To evaluate in vivo bioactivity of FSH glycoforms in the male, beginning at postnatal day 5, mice were injected once daily with either PBS, FSH^21/18 or FSH²⁴ for 5 days. By postnatal day 10, testis weights in PBS-injected Fshb null mice significantly increased compared with those at 5d (Fig. 8A). Compared to the PBS injected group, FSH glycoform-injected groups demonstrated a significant increase in testicular weights on postnatal day10. When injected with FSH^21/18, testis weights in Fshb null mice increased by 1.8 times (3.8 times compared to that in PBS injected Fshb null mice on postnatal day 5) by d10 (Fig. 8A). When injected with FSH²⁴, testis weights in Fshb null mice increased by 1.3 times (2.6 times compared to PBS injected controls on postnatal day 5) (Fig. 8A). When testis weight gain responses were compared between the two FSH glycoforms, although the hypo-glycosylated FSH^21/18 glycoform showed a greater response (1.4-1.5 times more) than that by the full-glycosylated FSH²⁴ glycoform, it was not statistically significant (Fig. 8A).

Fig. 8.

In vivo bioactivity of FSH glycoforms was evaluated following the protocols depicted in Supplementary Fig. S1. Both FSH^21/18 and FSH²⁴ elicited a testis weight gain response in Fshb^−/− male mice in a 5-day injection protocol (A). * P < 0.05 vs. 5d PBS group, ** P < 0.05 vs. 10d PBS group, n=7 mice. Hematoxylin-eosin stained testis histology (B) shows increased number of germ cells in sections from FSH^21/18 - injected mice compared to those in FSH²⁴ - injected mice resulting in increased tubule size (bar graph in panel B). Quantification of GCNA1, a pan-germ cell marker - positive germ cells by immunofluorescence indicates the same (C). In panel B and C, * P < 0.05 vs. PBS group, ** P < 0.05 vs. FSH²⁴ group, n=3 mice, approximately 100 tubules were counted. White bar represents 100 μm in panel B and 200 μm in panel C. Taqman real time qPCR analysis (D) shows that 4 (Cdo1, Clu Tbx22 and Zic3) out of the 9 FSH-responsive genes were upregulated by FSH^21/18 compared to FSH²⁴. * P < 0.05 vs. PBS group, ** P < 0.05 vs. the other glycoform. The genes analyzed are listed in the Supplementary Material Table -1. In panel E, testes sections from postnatal day 5 pups from control (Ctrl.) and experimental groups were immunolabeled with antibodies against SOX9, the Sertoli cell lineage - specific marker (green) and BrdU (red). The merged images are shown (E). White bar represents 200 μm. Quantification (F) shows that loss of FSH results in 45 % reduction in number of proliferating Sertoli cells (calculated as % of Sox9⁺BrdU⁺ /Sox9⁺ cells) in Fshb^−/− males. FSH^21/18 significantly increases the percentage of proliferating Sertoli cells in Fshb^−/− males compared to those in FSH²⁴ - injected Fshb^−/− males (B). Approximately 250 tubules from multiple testis sections from 3 mice per group were counted. In panel F, * P < 0. 05, vs. 5d Ctrl. group, ** P < 0.05 vs. 10d Fshb^−/− group and ^a P < 0.05 vs. FSH²⁴-injected group.

Consistent with the above testis weight data, histological analysis of hematoxylin - eosin-stained testis sections indicated that testis tubule size significantly increased in FSH glycoform-treated mice compared to those in PBS-injected Fshb null males (Fig. 8B). Further, immunolabeling with an antibody against GCNA1, a pan-germ cell marker showed that a greater number of germ cells were present in the testis sections from FSH glycoform - injected Fshb null mice compared to those in PBS - injected group (Fig. 8C) likely as a result of more number of Sertoli cells. Both the testis tubule size (Fig. 8B, right most panel, bar graph) and germ cell number (Fig. 8C, right most panel, bar graph) were significantly higher in testes of mice treated with FH^21/18 compared to that by FSH²⁴. Finally, expression analysis of FSH-responsive genes in testis by qPCR assays (Fig. 8D) indicated that the FSH^21/18 glycoform significantly upregulated 4 (Cdo1, Clu, Tbx22 and Zic3) out of the 9 (44 %) and down-regulated 1 (Amh) out of the 9 (11 %) and did not affect 4 (Car13, Fshr, Gsta2, Slc12a5) out of the 9 (44 %) genes. In contrast, only 1 out of the 9 (11%) genes was significantly up- (Car13) and downregulated (Fshr) by FSH²⁴. 3 (Cdo1, Tbx22, and Amh) out of the 9 genes (33 %) were not affected by only FSH²⁴. Two (Gsta2 and Slc12a5) out of the 9 (22 %) genes were unaffected by either of the FSH glycoform and their expression was nearly identical to that in testes of the PBS-injected control group (Fig. 8D).

Because, Sertoli cells in the mouse testis begin to cease to proliferate by postnatal day 10, we assayed Sertoli cell proliferation by in vivo BrdU labeling in testes of postnatal day 5 pups that received daily injections of FSH glycoforms for 3 days beginning postnatal day 3 (Supplementary Fig. 1 C and Fig. 8E). In the absence of FSH, Sertoli cell proliferation was significantly reduced by 45 % in Fshb null mice (Fig. 8F). Scoring the percentage of proliferating Sox9⁺/BrdU⁺ Sertoli cells confirmed that FSH glycoforms significantly enhanced the Sertoli cell proliferation in Fshb null mice. The hypo-glycosylated FSH^21/18 glycoform significantly enhanced Sertoli cell proliferation when compared to that by fully-glycosylated FSH²⁴ (Fig. 8F). Collectively, all the above data confirm that both FSH glycoforms are bioactive in vivo in Fshb null males in the absence of endogenous FSH. Furthermore, the hypo-glycosylated FSH^21/18 glycoform elicited a better response than the fully-glycosylated FSH²⁴ glycoform when testis tubule size and Sertoli cell proliferation were quantified.

4. Discussion

The existence of macro-heterogeneity in FSH preparations contributed by FSHβ subunits differing in N-glycosylation was recently confirmed [4-6, 14, 17-19]. Four FSH glycoforms were identified by direct purification from pituitary extracts [4-7, 14, 17-19] and following recombinant expression of genes encoding specific N-glycosylation mutant FSHβ subunits [4-7, 14, 17-19]. Both hypo-glycosylated FSH^21/18 and fully-glycosylated FSH²⁴ glycoform preparations were tested in vitro in radio-ligand and radio-receptor assays [4, 6, 18] and immortalized granulosa cell cultures [17]. In many of these in vitro assays, the hypo-glycosylated FSH^21/18 glycoform preparation was found more active than the fully-glycosylated FSH²⁴ glycoform preparation [4, 6, 17, 18]. It has also been proposed that the abundance of FSH glycoforms in tissue, serum and urine changes with age and could have implications for ovarian aging and for explaining at least some of the extra-gonadal actions of FSH [4, 14, 17-19]. However, whether these different FSH glycoforms elicit bioactivity in vivo and if there are differences in their in vivo bioactivity had not been tested to date.

Here, we have tested the recombinant GH₃ cell-derived FSH glycoform preparations (FSH^21/18 and FSH²⁴) by a pharmacological rescue strategy using immature Fshb null female mice. This strategy permitted us to monitor the in vivo bioactivity of the injected FSH glycoforms on an identical Fshb null genetic background and in the absence of endogenous FSH. We took advantage of our previously made observation that Fshb null mice retain full FSH-responsiveness to exogenous FSH [22] and these mice could be rescued both genetically [21, 23, 24] and pharmacologically [22, 41, 42]. Our data in this manuscript are consistent with these previously published reports and confirm that the GH₃-cell-derived recombinant FSH glycosylation variants are biologically active in vivo. As our candidate gene expression data set represents only a “snapshot” of the known FSH-responsive genes in the mouse ovary [27, 28], additional studies employing large-scale gene expression profiling such as microarray/RNA Seq techniques are required to completely characterize large-scale responses elicited by each FSH glycoform.

FSH^21/18 and FSH²⁴ glycoforms differently regulated 20 genes in the ovary whose expression was analyzed after 2h of treatment by real-time PCR. We could place these genes into distinct biological pathways, thereby implicating that FSH^21/18 and FSH²⁴ glycoforms act via different pathways in vivo. Based on these gene responses to different FSH glycoforms, we predict that the FSH^21/18 glycoforms may regulate cell cycle, apoptosis, cell adhesion events and growth factor-mediated signaling. The FSH²⁴ glycoform may regulate cell differentiation pathways and transcription factor-mediated events. In support of this, the FSH^21/18 and FSH²⁴ glycoforms also induced expression of different activated forms of key signaling pathways downstream of FSH-receptors. Whether the observed differences manifest and are indicative of in vivo clearance rates of each hormone is not known. Recombinant hFSH glycosylation mutants equivalent to FSH²¹ and FSH¹⁸ were reported to be cleared from rat serum significantly faster than wild-type FSH, which was probably an 80%/20% mixture of FSH²⁴ and FSH^21/18 [43]. More detailed kinetic studies involving tracer-labeled glycoform preparations injected into Fshb null mice and monitoring of their clearance in mouse blood need to be performed to correlate these gene expression differences to glycoform serum half-lives. Nevertheless, our in vivo bioactivity data including estradiol production are in close agreement with recent in vitro data that also demonstrate that hypo-glycosylated FSH is more active than fully-glycosylated FSH in a granulosa tumor-derived KGN cell line [17]. Together, the previous in vitro [6, 17, 18] and in vivo bioactivity data in this manuscript functionally validate the biochemical data that indicated enhanced receptor occupancy and high affinity binding by hypo-glycosylated FSH glycoforms compared to those by the fully-glycosylated form [6, 17, 18]. These results stand in sharp contrast to studies with classic FSH isoforms which have differences in charge due to terminal sialic acids contributing to micro-heterogeneity in human FSH. Less acidic FSH isoforms exhibited significantly higher in vivo biological activity than more acidic FSH isoforms [44, 45]. Studies in the sheep model have indicated that acidic mix of FSH isoforms were more potent in inducing follicular maturation and estradiol production than the less acidic isoforms [46, 47]. Thus, it appears that the ratio of different FSH glycoforms may be important for in vivo biological actions of FSH [16].

We previously showed that daily injections of recombinant FSH to neonatal Fshb null male pups resulted in an increase in testis weights compared to those in Fshb null pups injected with PBS [48]. We modified this protocol and injected different FSH glycoforms once daily for 5 days, beginning at postnatal day 5. In this modified FSH in vivo bioassay in the male, each FSH glycoform was found active. The Sertoli cell proliferation was significantly higher in mice injected with FSH^21/18 as compared to those in FSH²⁴ - injected mice. Since the number of Sertoli cells also dictates the germ cell- carrying capacity, more GCNA1⁺ germ cells were present in FSH glycoform-injected mice.

We also found differences in testis gene expression in Fshb null males when injected with FSH^21/18 or FSH²⁴ (Fig. 8). We evaluated expression of a set of candidate genes shown in Fig. 8D, because these were found to be FSH-responsive in mouse Sertoli cells during the proliferative phase (Wang H., and Kumar T.R., unpublished results) prior to puberty. Currently, we do not know the functional significance of these FSH-responsive genes in the mouse testis. Similar to what we observed in the ovary, it is likely that each glycoform may signal through different signaling pathways in the testis, resulting in changes in different networks of gene expression. For example, while the classical signaling pathway downstream of FSH receptor is the cAMP-CREB and PKA pathway, we previously reported that FSH acts via a non-classical MAPK pathway involving NF-κB transcription factor and regulates AMH production [48]. It would be interesting to determine in the future the different gene networks regulated by different FSH glycoforms in the testis. The ability to quantify the expression of different testicular genes in Fshb null males in response to different FSH glycoforms provides another convenient in vivo bioassay for testing different FSH analogs.

The existence of a naturally occurring hypo-glycosylated human FSH variant and its changing abundance with age have been reported [18]. The hypo-glycosylated variant was particularly abundant in younger women during ovulatory cycles [18] when rapid action and clearance were required [49]. Consistent with this, in a radioligand assay, hypo-glycosylated FSH exhibited faster kinetics and bound a higher number of FSH receptors on rat testicular homogenate, calf testis membranes and ovarian cells expressing hFSHRs, more rapidly than the fully-glycosylated FSH [6]. Recent in vitro studies also support these receptor binding studies. These studies showed that relatively low doses of hypo-glycosylated FSH more robustly induced cAMP, estradiol and progesterone production in cultured granulosa cells when compared to that by fully-glycosylated FSH glycoform [17]. Thus, it was proposed that loss of hypo-glycosylated FSH associated with aging may manifest in defective follicle maturation and oocyte development and ultimately result in loss of ovarian function [17, 18]. This loss of ovarian function leads to loss of estrogen production and results in production of elevated fully-glycosylated FSH glycoform, whose abundance predominates over the hypo-glycosylated FSH during peri- or post-menopausal age [17, 18]. It is possible that the hypo-and fully-glycosylated FSH glycoforms may signal differently and exhibit a biased agonism at the FSHR in granulosa cells. Such a likely biased agonism by gonadotropin glycoforms at the level of cellular responses has been previously reported [50-53].

Non-classical FSH action other than cAMP-PKA pathway has been reported in Sertoli cells [48], bone osteoclasts [54] and more recently in myometrial [55] and endometrial cells [56]. Because FSH-receptors have been identified on non-gonadal cells including bone osteoclasts [54], tumor blood vessels [57], and uterine and placental cells [55, 56, 58, 59], it would be interesting to determine whether different FSH glycoforms bind and signal via FSH receptors in these non-gonadal cells. Most likely, FSH^21/18 and FSH²⁴ glycoforms may signal through different pathways in these different cell types. Because the ratio of FSH^21/18 to FSH²⁴ and the tissue/cell specificity of FSH action changes with age, biased agonism could be one mechanism through which these different FSH ligands signal via FSHRs and utilize different downstream components. These possibilities need to be tested in the future using both in vitro and in vivo approaches.

In conclusion, we tested the in vivo bioactivities of the hypo-glycosylated FSH^21/18 and fully-glycosylated FSH²⁴ glycoforms using a pharmacological rescue approach involving Fshb null mice. Our studies confirm that the FSH glycoforms elicit FSH bioactivity in vivo as measured by ovarian and testicular gene expression responses. Our in vivo studies in the male reveal that FSH^21/18 is more efficacious than FSH²⁴. Our studies in the female demonstrate that each glycoform participates in a different downstream signaling pathway in the ovary. Changes in abundance of different FSH glycoforms could explain their age-specific actions in target tissues for example, ovary versus bone [14, 17]. The long-term in vivo effects of distinct FSH glycoforms and their age-specific abundance will be of considerable interest in the future for testing their efficacy in follicle maturation protocols and further evaluating the associated clinical benefits.

Highlights.

• In vivo bioactivity of recombinant hypo (FSH^21/18) - and fully (FSH²⁴) - glycosylated FSH was tested in Fshb^−/− mice.

• Both the FSH glycoforms elicited an ovarian weight gain response and induced antrum formation in Fshb^−/− female mice.

• In ovarian gene induction assays, FSH glycoforms were found active in the absence of endogenous FSH.

• Both the FSH glycoforms elicited a testis weight gain response in Fshb^−/− male mice.

• In Sertoli cell proliferation and testis gene expression assays, FSH^21/18 glycoform was found more active than FSH²⁴.

Abbreviations

AMH

anti-Müllerian hormone

cAMP

Cyclic adenosine 5′-monophosphate

CHO

Chinese hamster ovary

CREB

Cyclic AMP-responsive element binding protein

FSH

Follicle-stimulating hormone

LH

Luteinizing hormone

PKA

Protein kinase-A

PVDF

Polyvinylidine difluoride

RIA

Radioimmuno assay

RRA

Radioreceptor assay

r-h

Recombinant human