| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Other Original Articles |
Department of Biological Sciences, Wichita State University, Wichita, Kansas 67260-0026; Institute of Self-Organizing Systems and Biophysics, North-Eastern Hill University (V.S.), Shillong 793022, India; and Protein Chemistry Laboratory, University of Pennsylvania School of Medicine (W.T.M.), Philadelphia, Pennsylvania 19104
Address all correspondence and requests for reprints to: Dr. George R. Bousfield, Department of Biological Sciences, Box 26, Wichita State University, 1845 Fairmount, Wichita, Kansas 67260-0026. E-mail: george.bousfield{at}wichita.edu
Abstract
Human FSH consists of a mixture of isoforms that can be separated on the basis of differences in negative charge conferred by variations in the numbers of sialic acid residues that terminate oligosaccharide branches. Western analysis of human FSH isoforms separated by chromatofocusing revealed the presence of two human FSHß isoforms that differed in size. A low mol wt human FSHß isoform was associated with all FSH isoform fractions. A high mol wt human FSHß isoform was associated with the more acidic fractions and increased in relative abundance as the pI decreased. Characterization of representative human FSHß isoforms by mass spectrometry and automated Edman degradation revealed a low mol wt isoform that was not glycosylated. A high mol wt isoform was N-glycosylated at Asn residues 7 and 24. These results indicate that pituitary human FSH consists of two classes of molecules: those that possess a nonglycosylated ß-subunit and those that possess a glycosylated ß-subunit. Glycoprotein hormones are known to be elliptical molecules, and the ß-subunit oligosaccharides project outward from the short diameter, thereby increasing it. It is interesting to speculate that this change in shape might affect ultrafiltration rates, leading to differences in delivery rates to target tissues and elimination by filtration in the kidney.
FSH IS KNOWN to exist as a dynamic
population of isoforms that exhibit changes in relative abundance
reflecting the endocrine status of the individual or the stage of the
female reproductive cycle (1, 2, 3, 4). An early report of FSH
heterogeneity demonstrated the presence of two overlapping FSH isoforms
in rhesus monkey pituitaries by Sephadex G-100 gel filtration
(5). In intact females, the high mol wt form represented
37% of the total FSH immunoactivity, with the low mol wt form
constituting the remainder. Ovariectomy produced an increase in the
abundance of the high mol wt form to 58%. Elimination of
N-glycosylation sites by site-directed mutagenesis has been
shown to alter the Sephadex G-75 elution volumes for recombinant human
FSHß (hFSHß) glycosylation mutants (6). Studies of FSH
obtained from intact and gonadectomized rats and monkeys indicated that
steroid hormones regulated size variation in this hormone
(7, 8, 9). Application of high resolution electrophoretic and
chromatographic techniques revealed the existence of multiple FSH
isoforms in a variety of species, including our own (2).
The biological activities of the isoforms varied, with the less acidic
forms more active in various in vitro bioassays than the
more acidic isoforms. Changes in isoform abundance were observed during
the menstrual cycle (10). Removal of sialic acid from FSH
oligosaccharides by neuraminidase digestion demonstrated that charge
variation was largely due to differences in the sialic acid content of
the different isoforms (11). Estradiol has been reported
to reduce levels of sialyl transferase mRNA in rat pituitaries, which
may contribute to the increase in less acidic FSH isoform abundance
associated with elevated circulating levels of this hormone
(12). Analysis of highly purified hFSH isoform
preparations revealed that the sialic acid content ranged from 1.5
mol/mol hFSH in a less acidic isoform to 13.7 mol/mol in a more acidic
isoform (13). Oligosaccharides isolated from hFSH differed
primarily in the number of branches, which for the most part were
terminated with sialic acid (14, 15). These were
distributed across four N-glycosylation sites, two in the
-subunit and two in the ß-subunit (16).
We became interested in characterizing the changing patterns of hFSH
glycosylation at individual glycosylation sites when our studies with
equine gonadotropin hybrids indicated equine FSH (eFSH) binding to its
cognate receptor was less affected by Asn56
oligosaccharide size than eLH (17). Taking the sialic acid
content as an indicator of oligosaccharide branching, it seemed that
hFSH isoforms could be isolated that possessed
Asn52 oligosaccharides with three or more
branches for our studies. Our plan was to isolate a representative
sample of hFSH isoforms from a human pituitary glycoprotein extract,
separate the isoforms by means of chromatofocusing, and characterize
the glycosylation of each isoform using techniques that are routine in
our laboratory (18). We expected that oligosaccharide
mapping would reveal a greater abundance of branched oligosaccharides
in the more acidic hFSH isoforms. Instead, we encountered a
nonglycosylated hFSHß isoform. Because of the potential physiological
significance of this isoform (19), our studies focused on
characterizing it instead of counting oligosaccharide branches.
Materials and Methods
Materials
Dr. Anne Hartree provided the human pituitary glycoprotein fraction, GTN, to Dr. Darrell N. Ward when she retired. Dr. Ward provided us with this fraction upon his retirement. The monoclonal antibody 46.3H6.B7 (20) was provided by Dr. James A. Dias (Wadsworth Center, New York State Department of Health, Albany, NY). Monoclonal antibodies RFSH20 (21), HT13 (22), and ECG01 (23) were provided by Dr. Jean-Michel Bidart (Insitut Gustave-Roussy, Villejuif, France). The monoclonal antibody 518B7 was provided by Dr. Janet Roser (University of California, Davis, CA) (24). The rabbit polyclonal antibody to eLHß was prepared in this laboratory (25). The hLH and hFSH reference preparations were provided by the NIDDK National Hormone and Pituitary Program and Dr. A. F. Parlow. Dr. Gordon D. Niswender (Colorado State University, Bolder, CO) provided the antiserum against progesterone. Charcoal was purchased from EM Science (Gibbstown, NJ). Antimouse IgG or antirabbit IgG conjugated with horseradish peroxidase, amino acid standards, plus general and mammalian protease inhibitor cocktails were purchased from Sigma (St. Louis, MO). Tritiated progesterone was purchased from Amersham Pharmacia Biotech (Arlington Heights, IL). Dextran T500, ECL Western blotting detection reagent, carboxymethyl (CM)-Sephadex C-50, phenyl-Sepharose 6 fast flow, Sephacryl S-100, Sephadex G-75, and PolyBufferExchanger-94 (PBE-94) resins as well as Polybuffer 74 were obtained from Amersham Pharmacia Biotech (Piscataway, NJ). The semimicro pH electrode was purchased from Corning, Inc. (Big Flats, NY). Polyvinylidene difluoride Immobilon-P membranes and Amicon Centricon ultrafiltration chambers were obtained from Millipore Corp. (Bedford, MA). Prestained Broad Range Precision Protein Standard markers were purchased from Bio-Rad Laboratories, Inc. (Richmond, CA). Gel-code Blue stain was purchased from Pierce Chemical Co. (Rockford, IL). Scintisafe and Kodak Biomax ML Scientific Imaging Film were purchased from Fisher Scientific (St. Louis, MO). Medium 199, chicken serum, gentamicin, calf serum, and Benchmark Prestained Protein Ladder were obtained from Life Technologies, Inc. (Gaithersburg, MD). The plastic beads used in the immunoradiometric assay (IRMA) were purchased from Precision Plastic Ball Co. (Chicago, IL). The IRMAs were performed in 60-well plates obtained from Abbott Laboratories (Abbott Park, IL). All other reagents were sequencing grade, HPLC grade, or reagent grade as required.
The monoclonal antibody 46.3H6.B7 was raised against hFSH
(20). It binds free FSHß or FSH dimer, but does not
cross-react with either -subunit or hLHß (26).
Monoclonal antibody, RFSH20, was raised against recombinant hFSH
(21). It binds hFSH and hFSHß, but does not bind hLH,
hCG, or -subunit. It does exhibit a low level of binding to hTSH.
Monoclonal antibody HT13 was raised against hCG (22) and
recognizes an epitope spanning loops 1 and 3 that is specific to human
-subunits (22, 23). It can bind all human glycoprotein
hormones via the common -subunit. The monoclonal antibody, ECG01,
was raised against eCG (27). It recognizes a
conformational epitope in equine and human -subunit loops 1 and 3
either as the isolated subunit or as part of a heterodimer
(23). Bovine LH was used to prepare the monoclonal
antibody 518B7, which binds free CG/LHß and CG/LH dimer at a
conformation-dependent epitope located on the ß-subunit (24, 28). The polyclonal antibody, anti-eLHß, was raised against
eLHß (29). This antibody recognizes denatured human and
equine LH ß-subunits when employed in Western blots and also
recognizes denatured equine -subunit (due to the 1.3% intact
eLH content of the immunogen), but not human -subunit.
Purification of hFSH
A 1.2-g batch of GTN was dissolved in 200 ml 4 mM ammonium acetate buffer, pH 5.7, at 4 C. Cation exchange chromatography using a 2.5 x 60-cm CM-Sephadex column, equilibrated in the same buffer, proceeded as described by Hartree (30), except that CM-Sephadex (C50) was substituted for CM-cellulose. The CM-Sephadex-bound fractions were subjected to phenyl-Sepharose chromatography. Samples ranging from 44–268 mg protein were dissolved in 1 M ammonium sulfate in 0.05 M sodium phosphate, pH 7.0, and applied to 2.5 x 13.2-cm columns of Amersham Pharmacia Biotech fast flow phenyl-Sepharose. The chromatograms were developed by stepwise elution as described by Hiyama et al. (31) at a flow rate of 270 ml/h. The FSH fractions obtained from three phenyl-Sepharose chromatograms were pooled and applied to a 2.5 x 190-cm column of Sephacryl S-100 that was equilibrated and developed with 0.126 M ammonium bicarbonate at a flow rate of 75 ml/h.
Separation of hFSH isoforms by chromatofocusing
Polybuffer 74 (PB-74) was diluted 1:8 with distilled water, and the pH was adjusted to 4.0 with HCl. A 0.9 x 40-cm PBE-94 column was equilibrated with 25 mM imidazole-HCl, pH 7.4, and developed with PB-74-HCl, pH 4.0, at a flow rate of 15 ml/h (32). The 46-mg hFSH sample was dissolved in PB-74 and equilibrated by gel filtration chromatography using a 2.5 x 33-cm Sephadex G-25 column equilibrated and developed with PB-74. The PBE-94 chromatogram was developed with PB-74, pH 4.0, at a flow rate of 16 ml/h. After the chromatogram returned to baseline during PB-74 elution, 1 M sodium chloride was applied to the column to elute the remaining bound protein. Protein peaks were detected by UV absorbance at 280 nm. The pH of each tube was measured using a semimicro pH electrode. A 0.3-ml aliquot of 1.1 M Tris-HCl, pH 7, was added to all fractions with a pH of 6 or less. Tubes were initially pooled into 16 different fractions, concentrated by lyophilization or by ultrafiltration using 10,000 mol wt cut-off Amicon Centriprep cartridges. Each fraction was then applied to a 2.5 x 94-cm Sephadex G-75 column that was equilibrated and developed with 0.126 M ammonium bicarbonate at a flow rate of 36 ml/h. Proteins detected by absorbance at 280 nm were recovered by lyophilization.
Immunoaffinity chromatography
Samples of the two most basic isoforms, 4.7 mg Cf-A and 1.0 mg Cf-B, were subjected to immunoaffinity chromatography at 4 C. One of the hFSH subunit fractions, Cf-C4, was also purified by immunoaffinity chromatography at 25 C. The hFSHß-specific monoclonal antibody, 46.3H6.B7, was purified by protein G affinity chromatography, and the IgG fraction was coupled to a 1-ml Amersham Pharmacia Biotech HiTrap-NHS column following the instructions provided by the manufacturer. FSH or FSH subunit samples were diluted in 20 mM sodium phosphate buffer, pH 7.0. During initial experiments this buffer did not include protease inhibitors, whereas in subsequent experiments 1 mg/ml general protease inhibitors were included. The samples were vortexed, incubated for 15 min at 25 C, applied to the affinity column, and slowly recycled through the column for 30 min. The unbound material was eluted with 20 mM sodium phosphate, pH 7.0, and detected by absorbance at 280 nm using an in-line Uvicord II detector (Amersham Pharmacia Biotech). After the recorder pen returned to baseline, the bound fraction was eluted with 100 mM glycine-HCl, pH 2.7, and immediately neutralized by the addition of 180 µl/ml 1 M Tris-HCl, pH 9.0. Fractions were concentrated and desalted in Centricon P10 or Centriplus ultrafiltration devices and analyzed by Western analysis using RFSH20, HT13, and anti-eLHß as described below. Amino acid analysis was performed on all four purified isoforms.
Subunit isolation
Two hFSH isoform samples, 2.9 mg Cf-C and 1.8 mg Cf-D, were incubated in 6 M guanidine hydrochloride at 37 C overnight with shaking. Each sample was then applied to a Phenomenex Jupiter C18 reverse phase HPLC column equilibrated with 80% 0.1 M sodium phosphate-triethylamine/20% 20 mM sodium phosphate-triethylamine, pH 6.5, containing 60% acetonitrile at a flow rate of 1 ml/min using a model 600 high performance liquid chromatograph (Waters Corp., Milford, MA). The column was washed for 10 min, then a 60-min linear gradient from 20–40% phosphate-triethylamine/acetonitrile was applied to elute the hFSH subunit fractions (31). Protein was detected by simultaneous measurement of absorbance at 280 and 210 nm using a model 490E detector (Waters Corp.). Proteins were desalted and concentrated using Amicon Centricon P10 devices. Fractions were analyzed by SDS-PAGE and Western analysis under reducing and nonreducing conditions with HT13 and RFSH20 antibodies. One fraction was subjected to immunoaffinity chromatography as described above. Samples of interest were subjected to amino acid analysis, mass spectrometry analysis, protein sequence analysis, and anti-eLHß Western analysis.
SDS-PAGE
SDS-PAGE was routinely conducted using a Mini-PROTEAN II electrophoresis apparatus (Bio-Rad Laboratories, Inc., Richmond, CA). Samples were subjected to electrophoresis on 15% polyacrylamide slab gels, using the discontinuous buffer system of Laemmli (33) at a constant voltage of 200 V. Gels were fixed for at least 30 min in 9% acetic acid/45% methanol/46% water, stained with Gel-code Blue for at least 1 h, then destained in distilled water. Gel images were captured using a Gel Doc 1000 (Bio-Rad Laboratories, Inc.), and Mr estimates were made using the software package Quantity One (Bio-Rad Laboratories, Inc.).
Western blotting
Western blotting on Millipore Corp. Immobilon-P polyvinylidene difluoride membranes was performed as recommended by the manufacturer. Following SDS-PAGE, proteins were transferred to Immobilon-P membranes at 4 C for 2 h using a constant voltage of 100 V in a Mini-PROTEAN II transfer apparatus. The transfer buffer was composed of 25 mM Tris-HCl (pH 7.4), 190 mM glycine, and 20% methanol. After electroblotting, the membranes were blocked with 50 ml 5% (wt/vol) nonfat dry milk suspended in Western blotting buffer (WBB) composed of 150 mM sodium chloride, 1 mM EDTA, 50 mM Tris-HCl, and 0.05% Tween 20. After 1 h, the membranes were washed with distilled water followed by three 100-ml washes in WBB for 5 min each. Membranes were next incubated with one of the following protein G-purified monoclonal antibodies (MAb) RFSH20, HT13 (diluted 1:1000), or protein G-purified polyclonal anti-eLHß antibody (5 µg) in 5 ml WBB containing 5% nonfat dry milk for 3–4 h. Membranes were washed three times in WBB. Secondary antibodies (5 µl) conjugated to horseradish peroxidase that were specific for mouse (RFSH20 and HT13) or rabbit (anti-eLHß) IgG were added to 10 ml 5% milk in WBB, incubated with a membrane for 1 h, and then washed as described above. Finally, membranes were treated with Amersham Pharmacia Biotech enhanced chemiluminescence Western blotting detection reagents following the manufacturer’s instructions. Blots were then exposed to Biomax ML Scientific Imaging Film for various times and processed with a film processor. Either Life Technologies, Inc., Benchmark Prestained protein ladder or Bio-Rad Laboratories, Inc., Prestained Broad Range Precision Protein Standards were used to estimate the sizes of the immunoreactive protein bands. X-Ray film images were captured using a Gel Doc 1000 (Bio-Rad Laboratories, Inc.). Mr estimation and volume analysis were carried out using the software package Quantity One. Volume analysis is densitometric analysis of a rectangular region of the SDS gel or x-ray film used in a Western blot.
Mass and sequence analysis
Matrix-assisted laser desorption ionization, time of flight mass
spectrometry (34) was performed on a Micromass TofSpec 2E
time of flight mass spectrometer (1.0-m flight tube) outfitted with a
N2 UV laser (337 nm). Samples of purified hFSH and hFSHß isoform
preparations were dissolved in 50% acetonitrile/0.1% trifluoroacetic
acid and diluted before analysis using -cyano-4-hydroxycinnamic acid
as the matrix (35). Samples of the hFSHß preparations
were subjected to 24 cycles of automated Edman degradation using a
PE Applied Biosystems (Foster City, CA) model 473A protein
sequencer. Potential signal peptidase cleavage sites were investigated
following the method of von Heijne (36) as implemented by
the Macintosh software package AnalyzeSignalase version 2.03, written
by Ned Mantei (Department of Biochemistry, Swiss Federal Institute of
Technology, Zurich, Switzerland).
IRMAs
IRMAs (37) involved combinations of the common
-specific MAb HT13 or ECG01 with either the FSHß-specific MAb
RFSH20, or the LHß-specific MAb 518B7. Tracer antibodies were diluted
to 100,000 cpm/200 µl. Cold hormone and tracer dilutions were
prepared in a solution of 50% 0.1 M sodium phosphate, pH
7.4, and 50% decomplemented calf serum. For each cold hormone
dilution, two beads were incubated with capture antibodies at 25 C for
13 h with slow shaking. Beads were incubated with capture antibody
HT13 to assess hLH contamination or with RFSH20 to measure hFSH. After
incubation, the beads were washed with Milli-Q water three times and
transferred to 60-well plates containing 100 µl cold hormone
dilutions and 100 µl 50% 0.1 M sodium phosphate, pH 7.4,
and 50% decomplemented calf serum. The plates were incubated at 25 C
for 2 h with slow shaking. Blank control wells with buffer and
tracer but no cold hormone served to measure background counts per min.
After incubation, free hormone was aspirated from wells, and the beads
were subsequently washed three times with Milli-Q water. After washing
the beads, 200 µl tracer were added to each well. The tracer
antibodies used were [125I]518B7 for measuring
hLH and [125I]HT13 or
[125I]ECG01 for measuring hFSH. Tracer
antibodies were incubated with the bead-hormone complex for 1 h at
25 C. After incubation, the supernatant was aspirated, and the beads
were washed three times with Milli-Q water, transferred to
polypropylene tubes, and counted in a Packard Cobra II AutoGamma
counter. The counting efficiency was more than 74%. Statistical
analysis of the families of dose-response curves was accomplished using
the software package Allfit (38).
Radioligand assays
The LH RLA employed [125I]hCG as radioligand and rat testis homogenate as LH receptor preparation (29). The specific activities ranged from 12–19 µCi/µg, and specific binding was 12–19% of the total counts added. The hLH standards employed were AFP-4179C (2162 IU/mg) and NIDDK/National Hormone and Pituitary Program AFP-4395A (6100 IU/mg). The FSH radioligand assay employed Chinese hamster ovary cells stably transfected with a human FSH receptor cDNA as the receptor preparation and [125I]eFSH as radioligand (17). The specific binding of this tracer ranged from 15–36% of the total counts added. The hFSH reference preparation was NIDDK/National Hormone and Pituitary Program AFP-7298A SIAFP-B-3, which has a FSH potency of 8560 IU/mg. The program Allfit (38) was used to compare the dose-response curves and calculate relative potencies.
Granulosa cell assay
To compare the in vitro biological activities of the purified hFSH isoforms, progesterone secretion by diethylstilbestrol-primed granulosa cells obtained from immature female rats was measured in response to increasing concentrations of hormone preparation (17, 39). The progesterone content of 100-µl samples of conditioned medium was determined by RIA (40). The ED50 for each dose-response curve was determined using Allfit (38), and relative potency calculations were based on comparison of these values.
Results
Purification of hFSH
The isolation procedure consisted of chromatographic steps
designed to minimize separation of the various hFSH isoforms. Indeed,
the bulk of the FSH activity was associated with a single fraction
recovered from the CM-Sephadex column and each phenyl-Sepharose column.
After gel filtration of the combined hFSH fractions, we recovered 49 mg
purified hFSH in Sephacryl S-100 fraction B that exhibited an FSH
receptor binding potency of 2925 IU/mg. This represented 25% of the
FSH activity in the GTN fraction. Comparison of the SDS-PAGE profiles
for the hFSH preparations revealed more contaminant protein bands in
S100-B than in AFP7298A (Fig. 1A
). The
electrophoretic mobilities of the two major bands associated with
S100-B were greater than that of the major band associated with
AFP7298A. The former resembled those of purified hLH, suggesting
possible contamination. However, a receptor binding assay indicated
only 4% LH receptor-binding activity associated with S100-B, and the
results of an IRMA indicated only 2% LH immunoactivity. Western
analysis of GTN, S100-B, and human gonadotropin reference preparations
with an hFSHß-specific monoclonal antibody revealed the presence of
two bands in all FSH-containing fractions (Fig. 1B). The 24,000 mol wt
hFSHß isoform band was more abundant than the 21,000 mol wt isoform.
Densitometric analysis of the x-ray film with Quantity One software
indicated that the 21,000 mol wt band represented 14% of the
immunoreactive material in GTN. Its relative abundance increased to
24% in S100-B. An 11,000 mol wt hFSHß fragment band was observed in
the GTN fraction, but not AFP7298A. This apparent fragment band
appeared to be associated with an hFSHß fraction that was partially
eliminated from S100-B by gel filtration chromatography.
|
The chromatofocusing chromatogram consisted of very broad,
overlapping peaks (Fig. 2A). Western
analysis of the individual fractions revealed that the 21,000 mol wt
hFSHß isoform was present in all chromatofocusing fractions that
possessed detectable hFSHß (Fig. 2B). The 24,000 mol wt isoform was
present only in the more acidic fractions, those eluted below pH 5.5,
and its relative abundance increased as the pI decreased. The
-subunit band intensity appeared to be related to the amount of
hLHß present (Fig. 2C). Although hLH was expected in basic fractions,
the level of hLHß contamination in many of these fractions was
surprisingly high (Fig. 2D), as we had only detected a 2–4% LH in the
S100-B fraction.
|
-specific antibody, as
the capture antibody and detected hLH with the LHß-specific
antibody [125I]518B7. FSH also binds HT13, and
this could have limited hLH binding, producing a reduced estimate of LH
present in S100-B. Reduction of the FSH concentration after
chromatofocusing may have been responsible for the nearly 3-fold
increase in LH immunoactivity that accompanied chromatofocusing.
Western analysis indicated that Cf-B (fractions 4–9) was also
predominantly hLH, although there was not enough material to include in
the IRMA. We used immunoaffinity chromatography to purify hFSH from
both Cf-A and Cf-B. The antibody-bound fractions were designated Cf-A1
and Cf-B1. These preparations along with Cf-C (fractions 10–13) and
Cf-D (fraction 15) provided us with four highly purified hFSH isoforms.
Cf-A1 and Cf-B1 possessed primarily the low mol wt hFSHß isoform,
whereas Cf-C and Cf-D possessed both hFSHß isoforms (Fig. 3A). Western analysis indicated that we
had succeeded in largely eliminating hLHß immunoactivity (Fig. 3C),
which could only be detected after prolonged exposure of the x-ray
film. However, densitometric analysis indicated that the intensity of
the -subunit bands varied considerably, with 30% that of AFP7298A
hFSH present in Cf-A1, 7% in Cf-B1, 100% in Cf-C, and 66% in Cf-D
(Fig. 3B). In contrast, the hFSHß immunoactivity possessed by these
fractions was relatively uniform, representing 56–64% that of
AFP7298A hFSH. As the same variations in -subunit immunoactivity
were obtained after probing three different Immobilon-P membranes with
HT13, we suspected that Cf-A1 and Cf-B1 possessed more hFSHß than
hFSH. Indeed, a subsequent dimer-specific IRMA that used RFSH20 as
capture antibody and ECG01 as tracer indicated that that Cf-B1
possessed only 2% hFSH immunoactivity.
|
FSH radioligand assays of Cf-A1, Cf-C, and Cf-D (Fig. 4A) indicated the same level of FSH
receptor-binding activity in these three fractions (mean potency
± SD, 3,548 ± 1,515, 1,791 ± 272, and
1,421 ± 352 IU/mg, respectively). This suggested that despite the
reduced -subunit immunological activity observed in Cf-A1, it was
primarily an hFSH dimer preparation. When the FSH activities were
compared in the more sensitive granulosa cell assay that permitted
inclusion of Cf-B1 (Fig. 4B), this preparation exhibited only 5% FSH
biological activity (457 IU/mg). It possessed 8 times more hFSHß
immunoactivity than -subunit immunoactivity as indicated by Western
analysis and 2% hFSH by IRMA, consistent with the presence of more
biologically inactive hFSHß than functional hFSH heterodimer.
Although the Cf-A1 FSH receptor-binding activity was not significantly
greater than that of Cf-C and Cf-D (P > 0.05), its FSH
potency in the granulosa cell assay increased 4-fold, giving Cf-A1 a
greater FSH activity (14,545 IU/mg) than the hFSH reference preparation
AFP7298A. Cf-C and Cf-D had FSH potencies of 1,238 and 1,615 IU/mg,
respectively. Unfortunately, there was not enough material to further
characterize Cf-A1.
|
The more abundant hFSH isoform preparations, Cf-C and Cf-D, were
used for subunit purification. The former possessed less than 50% of
the 24,000 mol wt hFSHß isoform, whereas the latter possessed 77% of
this isoform. The relative abundance of the 24,000 mol wt hFSHß
isoform was reflected in the HPLC chromatograms (Fig. 5). The Cf-D subunit chromatogram
included a peak at 27 min that produced an electrophoretic pattern
indicating that it was probably the 24,000 mol wt hFSHß isoform (Fig. 5B, inset, lane 4). This was confirmed by Western analysis
(Fig. 6C). The absence of hFSH
immunoactivity from Cf-D3 indicated that it was suitable for
characterization of its glycosylation. Fractions Cf-D4 and Cf-D5
appeared to be hFSH. Cf-D6 was probably hFSH, as both - and
ß-subunits were present, whereas Cf-D7 was composed primarily of the
24,000 mol wt hFSHß isoform. The results were different when we tried
to isolate subunits from Cf-C. There was only a shoulder at 27 min that
produced a much broader Coomassie blue-stained band than that
associated with Cf-D3. Western analysis indicated that Cf-C2 included
the 24,000 mol wt hFSHß isoform along with hFSH. In fact, all of
the fractions recovered from this chromatogram included both subunits.
Fraction Cf-C4 appeared to be the best candidate for isolating the
21,000 mol wt hFSHß isoform. As Western analysis of nonreduced
samples of this fraction indicated that it was predominantly
dissociated subunit (data not shown), we purified the hFSHß component
by immunoaffinity chromatography using the same procedure employed to
obtain Cf-A1 and Cf-B1.
|
|
The mass of the major component in the low mol wt hFSHß
isoform preparation, Cf-C4 (Fig. 7A), was
12,361.3. This was 124.9 mass units less than the average isotopic mass
of 12,486.2 predicted from the amino acid sequence of hFSHß, whereas
the minor component was 80.4 mass units larger. The mass of the major
component in the high mol wt hFSHß isoform preparation, Cf-D3 (Fig. 7B), was 13,450.0, which was 963.8 mass units larger than the average
isotopic mass of the peptide moiety of hFSHß, whereas the minor
component was 127.3 mass units smaller. Only the high mol wt hFSHß
isoform represented by Cf-D3 appeared to be glycosylated, as at least
five overlapping peaks could be discerned in the expanded mass spectrum
(Fig. 7B, inset). The Edman sequencing results corroborated
the absence of glycosylation in the low mol wt hFSHß isoform Cf-C4
and revealed that the smaller than predicted mass for the main
component was due to a two-residue deletion at the amino-terminus. A
major sequence beginning with Cys3 and a minor
sequence beginning at Asn1 were apparent in the
>PhNCS-amino acid chromatograms. The 205.6-mass unit difference
between the two components mass labeled 12,566.6 and 12,361.0 shown in
Fig. 7A was very close to the 201.2-mass unit difference calculated
from the hFSHß 1–111 and 3–111 peptide sequences using the
PeptIdent tool available at the SWISS-PROT/TrEMBL web site
(41). Amino-terminal heterogeneity has been reported for
FSH ß-subunits derived from pituitary FSH (42, 43, 44).
Until recombinant hFSHß was analyzed, the heterogeneity was commonly
attributed to postmortem degradation (45). However, the
report that insect cell-expressed hFSHß possessed an amino-terminal
Cys3 suggested this resulted from alternative
signal peptidase cleavage (46). Analysis of the predicted
sequence for pre-hFSHß using the method of von Heijne
(36) produced a lower score for cleavage at
Cys-1–Asn1
(3.36) than for cleavage at
Ser2–Cys3 (5.37). In
contrast, similar scores (3.93 and 3.35, respectively) were obtained
for pre-eFSHß, which corresponded to the roughly equivalent amounts
of the 1–111 and 3–111 forms found in this species (44).
It was interesting that evidence for only the 3–111 hFSHß form was
obtained during automated Edman analysis of the glycosylated hFSHß
isoform (see below), because both the 1–111 and 3–111 forms were
associated with the diglycosylated and monoglycosylated eFSHß
isoforms (Bousfield, G. R., unpublished data), suggesting there is
probably no relationship between signal peptidase cleavage and
N-glycosylation. The fact that one signal peptidase cleavage
site appeared to be favored over the other was crucial in permitting us
to examine the glycosylation status of both hFSHß N-linked
sites. >PhNCS-Asn was observed during Edman degradation cycles
corresponding to Asn7 and
Asn24, which indicated that some of the hFSHß
isoform contained in Cf-C4 had never been glycosylated. Amino acid
sequence analysis of Cf-D3 indicated that it was also a truncated
hFSHß isoform consisting of residues 3–111. No >PhNCS-Asn was
detected at either glycosylation site, indicating that the hFSHß
isoform CD-D3 was glycosylated at both Asn7 and
Asn24. Subtracting the mass predicted for hFSHß
3–111 from the mass assigned to Cf-D3 yielded a combined mass of only
1,165 for the two oligosaccharides. This seemed small for the hFSHß
isolated from the most acidic hFSH isoform, especially since the most
abundant hFSH oligosaccharides were reported to be triantennary
structures (15) with formula weights of 3,066. The
presence of triantennary oligosaccharides predicted a mass of 18,700;
however, no convincing signal was obtained in that region of the
spectrum. In contrast, nonglycosylated hFSHß 3–111 was observed
even though its presence was undetectable by Western analysis (Fig. 6C
).
|
-subunits obtained from hFSH isoform
preparations Cf-C3 and Cf-D5 yielded masses of 13,796.3 and 14,377.1,
respectively. These peaks covered very broad ranges compared with the
glycosylated hFSHß isoform. The corresponding heterodimer masses
ranged from 26,363 for the nonglycosylated hFSHß isoform to 27,827
for the diglycosylated hFSHß isoform. These were unexpectedly
low because the electrophoretic mobilities of hFSH subunits were less
than those for eFSH, indicating a greater mass for hFSH
(47), yet the masses for the eFSH isoforms ranged from
28,199–31,059 (44). Discussion
Partial glycosylation of FSH subunits was first reported for recombinant bFSHß (48). Equine FSHß was the first naturally occurring FSH subunit reported to exhibit partial glycosylation. This occurred at a single glycosylation site, Asn7 (44). When we observed the two-band pattern after Western analysis of immunopurified hFSHß, we anticipated that the 24,000 mol wt band represented the diglycosylated band, whereas the 21,000 mol wt band represented the monoglycosylated band. The existence of partially glycosylated FSHß should have been observed during amino acid sequence determination; however, the low abundance of this isoform combined with subunit purification techniques that favored isolation of more heavily glycosylated FSHß isoforms probably ensured that only the diglycosylated FSHß isoform was obtained. Typically FSH subunits were separated by anion exchange chromatography (49, 50, 51, 52, 53). Nonglycosylated FSHß probably was eluted in the unabsorbed fraction, whereas the absorbed fraction was enriched for the diglycosylated form. Even reverse phase HPLC, once touted as the universal solution to glycoprotein hormone subunit isolation (31, 54), was inadequate to separate the subunits derived from Cf-C in which the nonglycosylated ß-subunit predominated. The chromatogram obtained during HPLC isolation of pFSH subunits indicated that porcine FSHß was obtained from a peak that resembled the hFSHß 27 min peak from which diglycosylated FSHß was obtained in the present study (43).
The high biological potency of Cf-A1 in the granulosa cell assay corresponded to results reported by Bishop et al. (55) for recombinant hFSH possessing a mutated, nonglycosylated hFSHß expressed by CHO cells. Although elimination of both hFSHß glycosylation sites had no effect on FSH receptor-binding activity, there was a 6-fold increase in FSH biological activity in the rat Sertoli cell bioassay. Recombinant hFSH expressed in COS-7 cells displayed increased FSH receptor-binding activity after hFSHß glycosylation site deletion that was accompanied by decreased FSH biological activity in the rat granulosa cell bioassay (6). These contrasting results could be accounted for by differences in glycosylation resulting from expression in different cell types. However, no carbohydrate data were available, and other explanations, such as the use of RIA (55) vs. radioligand assay (6) to quantify the recombinant hFSH included in the assay, Sertoli cells vs. granulosa cells, proteolytic damage, or serum factor interference with the assays, made interpretation difficult. Our results contrast with those involving purified hFSH preparations, which suggested the less acidic hFSH isoforms were less active in RIAs and radioligand assays, whereas more acidic isoforms were more active (56). In the present study the least acidic preparation, Cf-A1, exhibited greater biological activity than the more acidic preparations, Cf-C and Cf-D. This result was consistent with studies involving crude hFSH isoform preparations; less acidic isoforms exhibited greater in vitro biological activity (2).
Based on ß-subunit glycosylation, FSH could be divided into two categories: diglycosylated ß-subunit, which possessed both N-linked oligosaccharides, and underglycosylated ß-subunit. The latter appeared to take two forms. In horses, partially glycosylated eFSHß lacked carbohydrate at Asn7, but appeared to be glycosylated at Asn24. In hFSHß, it appeared that neither glycosylation site was occupied by carbohydrate. This was contrary to our expectation that a monoglycosylated hFSHß isoform would be encountered. As recombinant hFSHß includes an isoform lacking carbohydrate at Asn24 (26), the existence of a similar, naturally occurring hFSHß isoform remains an open question. For the present discussion we will refer to the low mol wt hFSHß isoform as a nonglycosylated isoform. All hFSH isoforms were found to possess hFSH with nonglycosylated hFSHß as determined by characterization of representative high and low mol wt hFSHß isoforms. More acidic hFSH isoforms possessed both the nonglycosylated and the diglycosylated hFSHß isoforms. The relative abundance of the diglycosylated hFSHß isoform increased with decreasing pI of the hFSH isoform. Nevertheless, even the most acidic hFSH isoforms possessed some of the nonglycosylated FSHß isoform. Shorter circulatory half-lives have been reported for the less acidic FSH isoforms (57). Site-directed mutagenesis of recombinant hFSH indicated that the ß-subunit oligosaccharides determined the MCRs (19). The present study demonstrates that the relative abundance of nonglycosylated hFSHß is an additional factor contributing to the MCRs for various hFSH isoforms. Particularly relevant for our studies, a 5-fold increase in the MCR was observed after elimination of the Asn7 glycosylation site and a 10-fold increase in clearance rate was noted after elimination of both glycosylation sites (19). Because less acidic hFSH isoforms possessed only the nonglycosylated ß-subunit isoform, the more rapid clearance of these forms reflected this structural feature. The fact that a 10-fold difference in clearance rates has not been noted for naturally occurring hFSH isoforms probably reflected the heterogeneous nature of the more acidic isoform population, as both nonglycosylated and diglycosylated hFSHß comprised the FSH molecules found in these isoform preparations.
Both hepatic and renal clearance mechanisms must be considered for glycoprotein hormones (4). Renal clearance may be more important for largely sialylated glycoprotein hormones, such as hFSH (58), as exclusively sialylated hCG was not cleared by the liver asialoglycoprotein receptor (59). For proteins with masses less than 66,200, the mass of serum albumin, factors that determine glomerular ultrafiltration rate are size, shape, and charge (60). Carbohydrate can change all three factors, although the greatest effects are likely to be on molecular shape and charge. The crystal structure of hCG revealed an elongated molecule with molecular dimensions of 75 x 30 x 35 Å (61). The recent report of the crystal structure for hFSH confirmed the expectation that it has a similar elongated shape (26). Modeling of hCG oligosaccharides showed the two ß-subunit oligosaccharides extending out from the narrow axis of the protein moiety (62). As this would increase the narrow diameter by as much as 30 Å (63), a substantial reduction in glomerular sieving could be expected. Indeed, reduced molecular size during gel filtration experiments has been demonstrated for recombinant hFSH mutants lacking one or more ß-subunit glycosylation sites (6).
Less acidic FSH isoforms are elevated in the preovulatory phase of the reproductive cycle. Studies with deglycosylated ovine FSH indicated a more rapid uptake by the follicular granulosa cells (64). Perhaps partial or total elimination of FSHß oligosaccharides is necessary for more efficient delivery to this avascular tissue, and the faster clearance rate is a by-product of a more rapid ultrafiltration rate necessary to facilitate ovarian uptake. Studies involving hFSH isoforms indicated that less acidic isoforms were more active in a bioassay that measured germinal vesicle breakdown in cumulus-enclosed mouse oocytes (65). Although recombinant hFSH mutants lacking both ß-subunit glycosylation sites did not support increased ovarian weight gain in rats, there could be a specialized function for them. The cumulus-surrounded oocyte in a preovulatory follicle is located at the greatest distance from the vascularized theca cell layer. Less acidic FSH isoforms may be adapted to activating oocytes in preovulatory follicles because the absence of ß-subunit oligosaccharides facilitates delivery to the interior of the follicle, increasing exit from follicle capillaries via ultrafiltration and more rapid diffusion through the interstitial fluid.
FSH glycosylation appears to be more complex than that of the other glycoprotein hormones. Partial glycosylation of the ß-subunit must be considered along with the distribution of variably branched oligosaccharides (15, 58). Glycosylation mutants of the FSH ß-subunit (6, 55) are now relevant subjects for studies on the physiological effects of FSH isoforms, but analysis of the impact of various oligosaccharide structures is rendered more difficult because FSH isoforms consist of mixtures of ß-subunit isoforms.
Acknowledgments
We thank Drs. A. S. Hartree and D. N. Ward for providing the human glycoprotein hormone fraction that made these studies possible. We thank Drs. J.-M. Bidart, J. A. Dias, J. Roser, and G. D. Niswender for providing antibodies, and the NIDDK/National Hormone and Pituitary Program and Dr. A. F. Parlow for the hFSH and hLH reference preparations. We are grateful to Dr. Viktor Y. Butnev for his advice on chromatographic procedures.
Footnotes
This work was supported by NIH Grant AG15428. Protein sequencing and mass spectrometry were provided by the Protein Chemistry Laboratory of the Medical School of the University of Pennsylvania supported by core grants of the Diabetes and Cancer Centers (DK-19525 and CA-16520).
1 Present address: Program in Molecular Biophysics, Florida State
University, Tallahassee, Florida 32306. 
Abbreviations: CM, Carboxymethyl; eFSH, equine FSH; hFSH, human FSH; IRMA, immunoradiometric assay; MAb, monoclonal antibody; PBE-94, PolyBufferExchanger-94; WBB, Western blotting buffer.
Received November 22, 2000.
Accepted March 30, 2001.
References
2,3-sialyltransferase messenger ribonucleic acid
levels in the female rat. J Mol Endocrinol 23:153–165[Abstract]
-subunit Asn56 oligosaccharide on in
vitro subunit association and FSH receptor binding of equine
gonadotropins. Biol Reprod 58:458–469 using monoclonal
antibody 3A identifies a potential receptor binding sequence. Endocrinology 128:1485–1495[Abstract]
-subunit. I. Characterization of topographic sites recognized by
monoclonal antibodies. J Biol Chem 263:10364–10369-subunit. Endocrinology 124:923–929[Abstract]
- and ß-subunits. Biochim Biophys Acta 1159:74–80[CrossRef][Medline]
and eCG-ß. Endocrinology 132:205–211[Abstract]
-Cyano-4
hydroxycinnamic acid as a matrix for matrix-assisted laser desorption
mass spectrometry. Org Mass Spectrom 27:156–158[CrossRef]
hCG, and ßhCG as measured by specific monoclonal
immunoradiometric assays. Endocrinology 120:549–558[Abstract]
- and ß-subunits of human
follicle-stimulating hormone. Endocrinology 94:655–664[Medline]
Subunit. J Biol Chem 250:6735–6746 and
ß subunits. Eur J Biochem 86:105–113[Medline]
This article has been cited by other articles:
 |
|
 |
|
  A. Ulloa-Aguirre, C. Timossi, J. Barrios-de-Tomasi, A. Maldonado, and P. Nayudu Impact of Carbohydrate Heterogeneity in Function of Follicle-Stimulating Hormone: Studies Derived from in Vitro and in Vivo Models Biol Reprod, August 1, 2003; 69(2): 379 - 389. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
