Hypoglycosylated hFSH Has Greater Bioactivity Than Fully Glycosylated Recombinant hFSH in Human Granulosa Cells

Abstract

Context:

Previous studies suggest that aging in women is associated with a reduction in hypoglycosylated forms of FSH.

Objective:

Experiments were performed to determine whether glycosylation of the FSHβ subunit modulates the biological activity of FSH in human granulosa cells.

Design and Setting:

Recombinant human FSH (hFSH) derived from GH₃ pituitary cells was purified into fractions containing hypoglycosylated hFSH^21/18 and fully glycosylated hFSH²⁴. The response to FSH glycoforms was evaluated using the well-characterized, FSH-responsive human granulosa cell line, KGN at an academic medical center.

Interventions:

Granulosa cells were treated with increasing concentrations of fully- or hypoglycosylated FSH glycoforms for periods up to 48 hours.

Main Outcome Measure(s):

The main outcomes were indices of cAMP-dependent cell signaling and estrogen and progesterone synthesis.

Results:

We observed that hypoglycosylated FSH^21/18 was significantly more effective than fully glycosylated FSH²⁴ at stimulating cAMP accumulation, protein kinase A (PKA) activity, and cAMP response element binding protein (CREB) (S133) phosphorylation. FSH^21/18 was also much more effective than hFSH²⁴ on the stimulation CREB-response element–mediated transcription, expression of aromatase and STAR proteins, and synthesis of estrogen and progesterone. Adenoviral-mediated expression of the endogenous inhibitor of PKA, inhibited FSH^21/18- and FSH²⁴-stimulated CREB phosphorylation, and steroidogenesis.

Conclusions:

Hypoglycosylated FSH^21/18 has greater bioactivity than fully glycosylated hFSH²⁴, suggesting that age-dependent decreases in hypoglycosylated hFSH contribute to reduced ovarian responsiveness. Hypoglycosylated FSH may be useful in follicle stimulation protocols for older patients using assisted reproduction technologies.

FSH stimulates the growth and maturation of ovarian follicles by acting directly on FSH receptors (FSHR) located on granulosa cells (1–3). Glycosylation of FSH is critical for FSHR activation (4, 5). Recent evidence suggests that human pituitary FSH consists of multiple glycoforms (6–9) and that FSH glycoform abundance is under physiological regulation (10, 11). Analysis of human FSH (hFSH) glycosylation revealed macroheterogeneity in FSHβ subunit N-glycosylation (6, 7, 11, 12). Given that the FSHα subunit always possesses both Asn⁵² and Asn⁷⁸ N-glycans, FSH glycoforms are identified by their FSHβ subunit variants, which can be accomplished by Western blot analysis using anti-hFSHβ antibodies, such as RFSH20 (6) and 15–1.C3.C5 (13). Fully glycosylated FSHβ²⁴ possesses both Asn⁷ and Asn²⁴ N-glycans; partially glycosylated FSHβ²¹ possesses only the Asn⁷ glycan; partially glycosylated FSHβ¹⁸ possesses only the Asn²⁴ glycan; whereas completely deglycosylated FSHβ¹⁵ lacks both FSHβ subunit N-glycans (12). Recent studies (9) suggest that hypoglycosylated pituitary hFSH preparations exhibited 9–20-fold higher FSH receptor binding activity compared with fully glycosylated FSH²⁴. It seems, therefore, that the extent of glycosylation of the FSHβ subunit may contribute to its bioactivity.

The Stages of Reproductive Aging Workshop (STRAW) reported that the course of reproductive aging through the menopause transition is characterized by an early monotonic increase in FSH followed by a characteristic steep trajectory during the late menopausal transition reaching levels greater than 25 mIU/mL (14, 15). Recent evidence shows that fully glycosylated FSH²⁴ represents approximately 80% of hFSH in pooled pituitary and urinary hFSH samples from postmenopausal women, whereas partially glycosylated FSH²¹ represents 52–70% of the hFSH in samples isolated from pituitaries derived from autopsies of women in their twenties (7, 9, 11). Furthermore, the abundance of the low molecular weight glycoform, FSH²¹, is correlated with the age of the woman (11). The FSH²¹ glycoform is more abundant in pituitaries of younger women and decreases over the reproductive life span. The ratio of FSH²¹ to FSH²⁴ decreases with increasing age such that in postmenopausal women hFSH²⁴ is the dominant glycoform. Although the reasons for the switch from hypoglycosylated hFSH to fully glycosylated hFSH are not understood at present, a study by Selman et al (16) reported that FSH preparations with different glycosylation patterns differentially affect clinical outcomes in patients being treated for infertility. Moreover, the profound increase in circulating levels of hFSH at menopause (15) highlights the importance of understanding how FSH glycosylation variants alter ovarian function.

The FSHβ subunit is essential for female fertility and sex steroid hormone production (17, 18). However, little is known regarding the changes in cellular responsiveness that occur in granulosa cells as a result of age-dependent alterations in FSHβ subunit glycosylation. The present study makes use of purified recombinant hFSH^21/18 and hFSH²⁴ glycofoms, which represent the changes in FSH glycoform expression that occur during aging in women. Our recent report (13) describes the purification, detailed characterization, and ligand-binding characteristics of these glycoforms expressed in GH₃ cells. Here we report that compared with the fully glycosylated hFSH²⁴, hypoglycosylated hFSH^21/18 displays enhanced ability to stimulate cAMP accumulation, protein kinase A (PKA) activity, and cAMP response element binding protein (CREB) phosphorylation in human granulosa cells. The enhanced cellular signaling events in response to treatment with hFSH^21/18 correlated with greater estrogen and progesterone secretion than observed in response to hFSH²⁴. Our findings suggest that glycosylation of the FSHβ subunit plays an important role in determining the bioactivity of the hormone.

Materials and Methods

Reagents

Recombinant hFSH was purified from transformed rat pituitary GH₃ cells that express both fully glycosylated hFSH²⁴ and hypoglycosylated hFSH¹⁸ and hFSH²¹. The purification and characterization are described in detail by Butnev et al (13). Western blot analysis of FSH preparations was performed with the FSHβ subunit-specific primary antibody RFSH20 (1:5000) and Precision Plus Protein WesternC standards (No. 161-0376). Pituitary hFSH reference preparation AFP7298 was provided by Dr A.F. Parlow, National Hormone and Pituitary Program. Follistim was from Merck & Co, Inc and Gonal-f was from EMD Serono, Inc. Antibodies used were: rabbit polyclonal anti-phospho-CREB (Ser133) and phospho-PKA substrate (RRXS*/T*) (Cell Signaling), anti-CREB (New England Biolabs), anticytochrome P450 aromatase (AbD Serotec), anti-STAR (Abcam). Horseradish peroxidase–conjugated antirabbit and antimouse secondary antibodies were from Sigma. The adenovirus expressing a canonical CREB-response element (CRE)-luciferase reporter (Ad.CRE-Luc) and the corresponding adenoviral control constructs Ad.MCS-Luc (firefly luciferase) and internal control construct Ad.pRL-Luc (Renilla luciferase) were from Vector Biolabs. The adenoviruses expressing green fluorescent protein (Ad.GFP) and the endogenous inhibitor of PKA, Ad.PKI were previously described (19). All reagents without a source listed were purchased from either Sigma or Fisher Scientific.

Cell culture

Human KGN granulosa cells (20) were kindly provided by Dr Ayae Fukuzawa (RIKEN Cell Bank, Koyadai, Japan) and cultured in DMEM/F12 medium supplemented with 10% fetal bovine serum, and 1% antibiotics as described (21, 22). For experiments determining FSH-responsive signaling pathways, granulosa cells were treated with 0–100 ng/mL of FSH^21/18 or FSH²⁴ for up to 120 minutes. For experiments measuring steroid synthesis, granulosa cells were treated with FSH glycoforms for 48 hours. To determine estradiol synthesis the medium was supplemented with 100nM 4-androstene-3,17-dione.

Western blot analysis

Western blot analysis and quantification were performed as described (21, 22). The membranes were incubated with primary antibodies at 4°C overnight: 1:1000 antiphospho-CREB, 1:1000 antiphospho-PKA substrates, 1:500 anti-aromatase, 1:5000 anti-STAR, 1: 5000 antiactin. Antibody binding was detected using HRP-conjugated antirabbit secondary antibody, followed by chemiluminescence detection (22).

cAMP assay

The cAMP levels in the medium were measured using a high-sensitivity direct cAMP chemiluminescent immunoassay kit (Arbor Assays).

Immunofluorescence assay

Granulosa cells cultured on glass coverslips were treated for 30 minutes with 0, 10, or 100ng/mL of FSH²¹ or FSH²⁴. After fixing in ice-cold 4% paraformaldehyde for 10 minutes, staining was performed as previously described (21, 22). Fixed cells were incubated with CREB or phosphospecific CREB antibodies at 4°C overnight. Antigens were visualized by applying Alexa Fluor-594 conjugated donkey antirabbit second antibodies. Coverslips were examined under epifluorescence (Olympus fluorescence microscope equipped with a DP71 camera, and the images were captured using MicroSuite image analysis software (Olympus). Cells containing nuclei positive for phospho-CREB were quantified and expressed as a percentage of the total cells counted on each coverslip; two coverslipper treatment and two to three images per coverslip.

CRE-luciferase reporter assay

To determine the effect of FSH glycoforms on transcription driven by the CRE, we employed an adenoviral CRE-driven luciferase (Luc) reporter construct (Ad.CRE-Luc; firefly luciferase), and the corresponding adenoviral control constructs Ad.MCS-Luc (firefly luciferase) and internal control construct Ad.pRL-Luc (Renilla luciferase). Briefly, human KGN granulosa cells were infected with 1 × 10⁶ pfu/mL of Ad.CRE-Luc plus Ad.pRL-Luc or Ad.MCS-Luc plus Ad.pRL-Luc as previously described (23). After 24 hours cultures were rinsed and treated with FSH^21/18 and FSH²⁴ for 6 hours. The Dual-luciferase reporter assay system kit (Promega) was used to detect the luciferase signal. Luminescence was detected using a FLUOstar OPTIMA (BMG LABTECH) plate reader. The firefly luciferase reporter activity was normalized to the Renilla luciferase signal and expressed as relative luciferase activity units (rlu).

Progesterone RIA and 17β-estradiol ELA

Progesterone in the culture medium was determined by RIA using a Coat-A-Count kit from Siemens. The 17β-estradiol assay was conducted using a commercial 17β-estradiol enzyme immunoassay kit from Arbor Assays.

Statistical analysis

All experiments were performed at least three times with granulosa cell cultures prepared on different days. Data analysis was performed using GraphPad Prism 5.03 software. The data are presented as means ± SEM. Best fit lines for concentration-response studies were drawn with a three-parameter nonlinear fit analysis (log agonist vs response). Comparison of a single response vs control was performed using the Student t test. Repeated measures were analyzed using ANOVA followed by the Student-Newman-Keuls test Multiple Comparison Test. Two-way ANOVA followed by Bonferroni post tests was used to compare concentration responses for FSH^21/18 and FSH²⁴. P < .05 was considered significant.

Results

FSH glycoforms

The FSH glycoforms used in this study are represented diagrammatically in Figure 1, A–C. Fully glycosylated hFSH²⁴ possesses glycans attached to all four potential N-glycosylation sites. The hypoglycosylated hFSH preparation possessed both FSH²¹ and FSH¹⁸ glycoforms and is hereafter designated FSH^21/18.

Figure 1.

Human FSH glycoforms used in these studies. A, Fully glycosylated, recombinant GH₃-hFSH²⁴, which possesses glycans at all four N-glycosylation sites. B, Hypoglycosylated GH₃-hFSH²¹, which lacks Asn²⁴ N-glycan. C, Hypoglycosylated GH₃-hFSH¹⁸, which lacks Asn⁷ glycan. The glycan diagrams depict abundant glycans found in the nano-electrospray ionization mass spectrometry analysis of GH₃-hFSH glycans using the Oxford Glycobiology Institute system (38, 39). D, Western blot comparing reduced, 1-μg samples of recombinant GH₃-hFSH glycoforms with commercially available recombinant hFSH and pituitary hFSH. The primary antibody was RFSH20, diluted 1:5000. Lane 1, MW markers, as indicated; lane 2, Follistim; lane 3, Gonal f; lane 4, hypoglycosylated GH₃-hFSH^21/18 mixture; lane 5, GH₃-hFSH²⁴; lane 6, pituitary hFSH (AFP7298A); lane 7, MW markers.

The recombinant hFSH glycoforms employed in the present study were compared with pituitary hFSH reference preparation AFP7298 and commercially available recombinant hFSH preparations expressed in Chinese hamster ovarian cells (Figure 1D). The hFSH glycoforms were expressed in a rat pituitary somatotrope GH₃ cell line (24) and isolated by procedures aimed at capturing all forms of the hormone (13). Both Follistim and Gonal-f possess largely 24 kDa hFSHβ, which migrates in the same region of the gel as pituitary hFSH²⁴ (compare lanes 2 and 3 with lane 6). Hypoglycosylated FSH represents less than 30% of the immuno-activity (lanes 2, 3, and 6). The unusually rapid migration of recombinant GH₃ FSHβ²⁴ and slow migration of GH₃ FSHβ²¹ were recently reported (13) along with evidence for both Asn⁷ and Asn²⁴ glycans in the former and the absence of the Asn²⁴ glycan in the latter.

FSH^21/18 is more effective than FSH²⁴ at stimulating cAMP accumulation and activating PKA.

Treatment of KGN human granulosa cell cultures for 30 minutes with increasing concentrations of FSH^21/18 and FSH²⁴ resulted in concentration-dependent increases in cAMP accumulation (Figure 2A). Two-way ANOVA revealed that cAMP levels in cells treated with 10, 30, or 100 ng/mL FSH^21/18 were significantly greater than cAMP levels in cells treated with comparable levels of FSH²⁴. At the highest concentration, FSH^21/18 stimulated 20-fold increases in cAMP vs 12-fold increases for FSH²⁴. We observed that 10 ng/mL FSH^21/18 was the lowest effective concentration of FSH^21/18 to significantly increase cAMP accumulation (5.7-fold increase; P < .05). Although 30 ng/mL FSH²⁴ increased cAMP by 3.9-fold, the response was not statistically significant. Treatment with 100 ng/mL FSH²⁴ was required to significantly increase cAMP (12-fold increase; P < .05).

Figure 2.

FSH^21/18 is more effective than FSH²⁴ at stimulating cAMP accumulation and PKA activation in human KGN granulosa cells. Granulosa cells were treated for 30 minutes with increasing concentrations of FSH^21/18 or FSH²⁴. A, cAMP levels in the medium were measured using a cAMP chemiluminescent immunoassay kit as described in the Methods. Results are means ± SEM; n = 3 separate experiments. *, P < .05 vs control. **, P < .05, FSH^21/18 vs FSH²⁴. B, Representative Western blot analysis of FSH-induced phosphorylation of PKA-substrates using an antibody directed against the PKA-consensus phosphorylation site. Actin served as a loading control.

To determine whether increases in cAMP couple to activation of PKA we performed Western blot analysis on whole cell lysates using an antibody directed against PKA consensus phosphorylation sites (RRXS*/T*). Consistent with cAMP accumulation results, FSH^21/18 was more effective than FSH²⁴ at increasing phosphorylation of PKA substrates (Figure 2B). Compared with control, 3 ng/mL of FSH^21/18 stimulated an increase phosphorylation of PKA substrates. In contrast, treatment with 10–30 ng/mL of FSH²⁴ was required to observe an increase in the phosphorylation of PKA substrates.

FSH^21/18 is more effective than FSH²⁴ at stimulating the phosphorylation of CREB

The transcription factor CREB is important for regulation of FSH-dependent transcription in granulosa cells. Furthermore, PKA-dependent phosphorylation of the Ser133 residue enhances the activity of CREB (25). Treatment of granulosa cells for 30 minutes with 1–100 ng/mL of FSH^21/18 and FSH²⁴ resulted in concentration-dependent increases in the CREB phosphorylation (Figure 3A). The graph and Western blot show that FSH^21/18 was much more effective than FSH²⁴ at stimulating CREB phosphorylation. At the highest concentration, FSH^21/18 stimulated an 18-fold increase in CREB-phosphorylation vs an 11-fold increase for FSH²⁴. We observed a slight increase (3.6-fold; P < .05) in phospho-CREB with 3 ng/mL FSH^21/18 and larger increases with 10 ng/mL FSH^21/18 (10-fold increase; P < .01). On average, 30 ng/mL FSH²⁴ stimulated an 8-fold increase in phospho-CREB, but the response was variable and not significant. Stimulation with 100 ng/mL FSH²⁴ was required to significantly increase (P < .05) levels of CREB phosphorylation. Two-way ANOVA revealed that phospho-CREB levels in cells treated with 10, 30, or 100 ng/mL FSH^21/18 were significantly greater than phospho-CREB levels in cells treated with comparable levels of FSH²⁴.

Figure 3.

FSH^21/18 is more effective than FSH²⁴ at stimulating CREB phosphorylation in human KGN granulosa cells. A, Granulosa cells were treated for 30 minutes with increasing concentrations of FSH^21/18 or FSH²⁴. Cell lysates were prepared and Western blot analysis performed using phospho-Ser 133-CREB (P-CREB), CREB, and β-actin antibodies. Bars represent means ± SEM; n = 3 separate experiments. *, P < .05 vs control. **, P < .05, FSH^21/18 vs FSH²⁴. B, Granulosa cells were treated for up to 120 minutes with 30 ng/mL FSH^21/18 or FSH²⁴. Shown is a representative Western blot analysis of the time course of FSH-induced phosphorylation of P-CREB and PKA-substrates. Levels of β-actin protein served as a loading control

Treatment of granulosa cells for 5–120 minutes with 30 ng/mL FSH^21/18 or FSH²⁴ revealed temporal increases in the phosphorylation of CREB and other PKA substrates (Figure 3B). We observed increases in CREB phosphorylation within 5 minutes following FSH^21/18 treatment. Maximal increases in CREB phosphorylation in response to FSH^21/18 were observed from 15–60 minutes before returning to near basal levels after 120 minutes. For FSH²⁴, the response was delayed, with increases in CREB phosphorylation observed after 15 minutes. Maximal responses were observed after 30 minutes and CREB phosphorylation returned to control levels within 120 minutes. When examining the onset of PKA-dependent phosphorylation we observed early temporal patterns similar to those observed for CREB phosphorylation for both FSH^21/18 and FSH²⁴. Increases in phosphorylation of PKA substrates were more rapid and more robust in response to FSH^21/18 compared with FSH²⁴. Our findings that the level of PKA-dependent phosphorylation remained elevated throughout the 120-minute incubation period contrasts with the reduction in CREB phosphorylation observed at 120 minutes.

FSH^21/18 is more effective than FSH²⁴ at stimulating CRE-mediated transcription activity

We used two approaches to evaluate whether the observed alterations in CREB phosphorylation coupled with changes in transcriptional activity. First, we used immunofluorescence to analyze the subcellular localization of phosphorylated CREB. As expected, we found that both phospho-CREB and nonphosphorylated CREB localized to granulosa cell nuclei (not shown). Consistent with the Western blot results (Figure 3A), we observed that treatment with FSH^21/18 was more effective than treatment with FSH²⁴ at increasing the percentage of granulosa cells with nuclear localized phospho-CREB (Figure 4A). We observed significant increases in cells positive for nuclear phospho-CREB following treatment with 10 ng/mL FSH^21/18, whereas it required 100 ng/mL FSH²⁴ to stimulate significant increases in nuclear phospho-CREB. Second, we examined the ability of FSH glycoforms to stimulate CRE-mediated gene expression using a cAMP-responsive-element-driven luciferase reporter gene (CRE-luc). Figure 4B shows that 10 ng/mL FSH^21/18 stimulated a significant increase (13.7-fold; P < .05) in transcription. In contrast, the 6-fold increase in CRE-mediated transcription in response to 10 ng/mL of FSH²⁴ was not statistically significant. Treatment with 100 ng/mL FSH^21/18 and FSH²⁴ significantly stimulated transcription of the CRE-luc reporter (22- and 19-fold increases, respectively). Two-way ANOVA revealed that the number of phospho-CREB-positive cells and extent of CRE-mediated transcription in cells treated with 10 ng/mL FSH^21/18 were significantly greater than in cells treated with 10 ng/mL FSH²⁴.

Figure 4.

FSH^21/18 is more effective than FSH²⁴ at stimulating CRE-mediated transcription. A, Granulosa cells were treated for 30 minutes with 0, 10, and 100 ng/mL of either FSH²¹ or FSH²⁴. The number of phospho-CREB-positive nuclei were quantified and expressed as a percent of the total cells counted in each experiment. Bars represent means ± SEM; n = 3 separate experiments. *, P < .05 vs control. B, Granulosa cells were infected with an adenovirus expressing a CRE-luciferase reporter (CRE-Luc) or green fluorescent protein as a control. After 24 hours, granulosa cells were treated for 6 hours with 0, 10, or 100 ng/mL FSH^21/18 or FSH²⁴. Relative CRE-luciferase activity (rlu) was determined as described in the Methods. Bars represent means ± SEM; n = 3 separate experiments. *, P < .05 vs control. **, P < .05, FSH^21/18 vs FSH²⁴. ns, not significant.

FSH^21/18 is more effective than FSH²⁴ at stimulating estradiol and progesterone synthesis

Treatment for 48 hours with FSH^21/18 was much more effective than FSH²⁴ at stimulating estradiol synthesis (Figure 5A). At the highest concentration (100 ng/mL), FSH^21/18 stimulated an 11-fold increase in estradiol vs a 6-fold increase for FSH²⁴. We observed 6-fold increases in estradiol with 10 ng/mL FSH^21/18, whereas 100 ng/mL FSH²⁴ was required to increase estradiol by 6.1-fold. Two-way ANOVA revealed that estradiol levels in granulosa cells treated with 30 or 100 ng/mL FSH^21/18 were significantly greater than estradiol levels in cells treated with comparable levels of FSH²⁴ (P < .05).

Figure 5.

FSH^21/18 is more effective than FSH²⁴ at stimulating estradiol and progesterone synthesis in human KGN granulosa cells. Granulosa cells were treated for 48 hours with control medium (C) or increasing concentrations (1–100 ng/mL) of FSH^21/18 or FSH²⁴. A and B, Conditioned medium was collected and estradiol and progesterone levels were measured as described in the Methods. Symbols represent means ± SEM; n = 3 experiments. *, P < .05 vs control. **, P < .05, FSH^21/18 vs FSH²⁴. C, Cell lysates were prepared and Western blot analysis of aromatase (CYP19), STAR, and β-actin proteins was performed.

FSH^21/18 was much more effective than FSH²⁴ at stimulating progesterone synthesis (Figure 5B). At 100 ng/mL, FSH^21/18 stimulated a 12.1-fold increase in progesterone vs a 6.8-fold increase for FSH²⁴. We observed a significant increase in progesterone with 10 ng/mL FSH^21/18 (9 ± 3.0-fold increase; P < .01), whereas 100 ng/mL FSH²⁴ was required to significantly increase progesterone (6.8 ± 0.9-fold increase; P < .05). Two-way ANOVA revealed that progesterone levels in granulosa cells treated with 10, 30, or 100 ng/mL FSH^21/18 were significantly greater than progesterone levels in cells treated with comparable levels of FSH²⁴ (P < .05).

Western blot analysis evaluated aromatase (CYP19) and STAR protein expression in granulosa cells incubated for 48 hours with increasing concentrations of FSH glycoforms (Figure 5C). Similar to the results obtained when analyzing steroid synthesis, we observed that FSH^21/18 was more effective than FSH²⁴ at increasing CYP19 and STAR proteins.

To determine whether the responses to the FSH glycoforms are mediated, at least in part by PKA, we overexpressed the endogenous inhibitor of PKA (PKI) (19). Compared with the adenovirus control (GFP) expression of PKI effectively blocked the ability of FSH^21/18 and FSH²⁴ to stimulate the phosphorylation of CREB and PKA substrates (Supplemental Figure 1A). Expression of PKI effectively blocked the ability of FSH^21/18 and FSH²⁴ to stimulate estrogen synthesis and CYP19 protein expression (Supplemental Figure 1B). Similar findings were observed for progesterone synthesis and STAR protein expression (not shown).

Discussion

We employed GH₃ cell–derived hFSH glycoform preparations, defined as FSH variants differing in the number of N-glycans attached to the FSHβ subunit, to demonstrate the effect of FSHβ subunit glycosylation on FSHR-mediated signaling events and steroid secretion in human granulosa cells. The present study used purified recombinant hFSH glycoforms, FSH^21/18, and FSH²⁴, and the KGN human granulosa cell line as our model system. Analysis of the GH₃ cell–derived FSH glycoforms revealed oligosaccharides similar to those present in pituitary FSH glycoforms (13). Hypoglycosylated hFSH^21/18 glycoforms exhibited significantly greater ability to stimulate cAMP- and PKA-dependent signaling compared with fully glycosylated FSH²⁴. The enhanced PKA signaling in response to hypoglycosylated FSH was associated with enhanced biological activity as evidenced by increased estrogen and progesterone synthesis. Our studies demonstrate for the first time the important contribution of FSHβ subunit N-glycosylation to the biologic activity of FSH.

Fully glycosylated hFSH²⁴ was much less effective than hypoglycosylated hFSH^21/18 at initiating cAMP/PKA signaling in the human granulosa cell line. This conclusion is supported by our observation that treatment with hFSH²⁴ concentrations greater than 30 ng/mL was required to provoke significant increases in cAMP accumulation, whereas, treatment with 10 ng/mL hFSH^21/18 caused a significant increase in cAMP accumulation. Consistent with these findings were observations that phosphorylation of PKA substrates and the transcription factor CREB were elevated at lower concentrations of hFSH^21/18 than hFSH²⁴. Furthermore, the CREB and PKA-dependent phosphorylation responses to hFSH^21/18 were more rapid than these responses to hFSH²⁴. The greater ability of FSH^21/18 to activate cellular signaling is presumptively due, at least in part, to a greater affinity of FSH^21/18 vs FSH²⁴ for the FSHR (9, 13). Studies using rat or bovine testis homogenates or CHO cells stably expressing the human FSHR clearly demonstrated that FSH^21/18 occupied more FSH binding sites than FSH²⁴ (9, 13). The relative ability of FSH glycoforms to bind the FSHR was not affected by the numbers of FSHR present in the ligand-binding assay, given that testis homogenates and CHO cells, which overexpress the FSHR, represent conditions of low vs high receptor density, respectively. This information is useful for the present analysis given that others have shown that cellular signaling responses to FSH, such as cAMP accumulation (26), are proportional to the number of occupied receptors. It seems, therefore, that the extent of FSHβ subunit N-glycosylation may contribute to the ability of FSH to activate the cAMP/PKA signaling pathway in granulosa cells.

The GH₃ cell–derived hFSH glycoforms, FSH^21/18 and FSH²⁴, displayed different bioactivities. In addition to effects on cAMP and PKA activity, compared with FSH²⁴, hypoglycosylated FSH^21/18 more effectively stimulated CRE-mediated transcription, aromatase and STAR protein expression, as well as estrogen and progesterone secretion. With regard to estrogen and progesterone synthesis, approximately 10-fold greater concentrations of FSH²⁴ were required to stimulate increases in steroid comparable to those observed in response to FSH^21/18. For example, when used at 100 ng/mL FSH²⁴ stimulated a 6.1-fold increase in estrogen production, whereas 10 ng/mL FSH^21/18 stimulated a 6.0-fold increase. A similar relationship exists for the effects of FSH²⁴ and FSH^21/18 on progesterone synthesis. It is instructive to note that adenoviral-mediated expression of the PKA inhibitor effectively suppressed the stimulatory effects of both FSH²⁴ and FSH^21/18 on PKA activity and steroid synthesis. These findings suggest that cAMP/PKA signaling is a common mediator of the actions of both FSH glycoforms.

Although the present findings suggest that FSHβ subunit macroheterogeneity differentially influences granulosa cell function, the effects of FSH microheterogeneity have also been studied on follicle development and estradiol secretion. Collectively, these earlier studies showed that FSH oligosaccharide structures affect biological responses (27–29). A recent report describes the effects of recombinant hFSH glycosylation variants that were isolated according to either degree of sialylation or complexity of oligosaccharides using preparative isoelectric focusing and lectin affinity chromatography, respectively (30). This study revealed that less acidic/sialylated FSH isoforms and FSH isoforms enriched for high mannose/hybrid oligosaccharides have greater effects on steroidogenic enzyme expression and steroid secretion in KGN cells. The idea that less acidic/sialylated isoforms contain some glycoforms similar to the hypoglycosylated FSH^21/18 preparation used in the present report has been reported (10). Microarray analysis revealed that the genes affected by glycoforms bearing incomplete oligosaccharides were associated with essential aspects of granulosa cell function, such as ovarian follicle development and steroid biosynthesis (30). Further analysis of the actions of FSH^21/18 and FSH²⁴ is required to fully understand their ability to influence gene expression and ovarian function.

The observation that FSH^21/18 has greater bioactivity than FSH²⁴ is highly relevant because circulating levels of FSH in premenopausal women rarely exceed 30 mIU/mL (31), which is approximately 3–3.5 ng/mL FSH (based on 8.56–10 mIU/ng as potencies for highly purified hFSH preparations). Importantly, the FSH glycoform produced in young women is predominately hypoglycosylated FSH^21/18 (7, 11). Whereas, the in vitro KGN human granulosa cell model may not accurately predict the sensitivity of granulosa cells to FSH in vivo, it is worth noting that 3–10 ng/mL FSH^21/18 provoked increases in steroid synthesis and 10-fold higher concentrations were required for similar responses to FSH²⁴. Thus, the more biologically active hypoglycosylated FSH may play a prominent role in regulating ovarian function during the reproductive years. In the postmenopausal period, FSH levels are considerably higher and can approach 120 mIU/mL (approximately 12–14 ng/mL). Given the loss of the FSH^21/18 glycoform during menopausal transition (11), we speculate that the high levels of the less bioactive FSH²⁴ glycoform may be insufficient to stimulate normal ovarian function in vivo. Underscoring the significance of the present results is the observation that recombinant hFSH preparations used for assisted reproduction protocols are largely composed of the FSH²⁴ glycoform. Given that clinical studies suggest a direct dose-response relationship between amount of FSH administered and follicle development (32), it will be important to determine whether hypoglycosylated FSH, which has greater bioactivity in vitro, has improved performance in a clinical setting.

The results demonstrate that the extent of hFSHβ subunit N-glycosylation affect signal transduction pathways and expression of proteins essential for granulosa cell steroid secretion. The results clearly suggest that hypoglycosylated FSH^21/18 is much more biologically active than fully glycosylated FSH²⁴. The present findings demonstrate the potential affect of naturally occurring age-related FSH glycoforms on FSHR-mediated signaling events and biological responses in granulosa cells. Given that clinical studies have found improved ovarian stimulation with mixtures of acidic and less acidic hFSH isoforms (16) it will be important to determine whether mixtures of hypoglycosylated FSH and fully glycosylated FSH contribute to enhanced or diminished biological responses. Understanding how FSH glycoforms direct activities in multiple FSH target tissues [ovary, bone (33, 34), endothelium (35), uterus, and placenta (36, 37)] offers a unique opportunity to develop novel approaches for diagnosis and treatments to improve fertility and reduce age-associated morbidity.