N-glycan data were processed using UNIFI 1.7 with Glycobase 3+ (Waters Corporation, Milford, MA) for N-glycan
structure. The peak area of chromatography was calculated for relative
glycan quantification. The mean relative content and SD were determined
for three replicates per sample. The MS2 data analysis was performed
with MassLynx 4.1 data system. The data of intact hFSH subunits were
deconvoluted and analyzed by UNIFI 1.7. The observed N-glycans structures were set as amino acid modification for deconvolution (mass error <50 ppm).
The LC-MSE data were processed using Biopharma Lynx 1.3.3 (Waters Corporation, Milford, MA) for N-glycosylation
sites analysis. The mass tolerance was set at 5 ppm for precursor and 5
ppm for fragment ions, respectively. The identified peptides were
confirmed by MSE spectra with at least five b/y fragment ions. The N-glycopeptide data were analyzed by ProteinLynx Global SERVER 3.0 (Waters Corporation, Milford, MA). In order to identify N-glycopeptides in rhFSH and uhFSH digest, N-linked glycosylation was selected as variable modification.
For O-glycopeptide identification, the data of
HCD-pd-ETD were searched separately using Byonic 2.5.6 (Protein Metrics,
San Carlos, CA) with the following search parameters: peptide tolerance
= 10 ppm; fragment tolerance = 0.02 Da for both HCD and ETD; missed
cleavages = 1; modifications: carbamidomethyl cysteine (fixed),
methionine oxidation (common2). Mucin type O-glycan including Hex
(mass of 162.0528), HexNAc (mass of 203.0794), HexNAc2 (mass of
406.1587), HexNAc2Hex1 (mass of 568.2116),NeuAc1HexNAc1Hex1 (mass of
656.2276) and NeuAc2HexNAc1Hex1 (mass of 947.3230) were set as database
in the Byonic data searching. Spectrum-level FDR was set as auto cut.
The PSM with score ≥100 and mass error <10 ppm were accepted.
In vivo bioassay for rhFSH and uhFSH biological activity
Protein contents were assessed by SEC-HPLC using BSA as standard. The in vivo bioactivity
of hFSH was assessed according to current US and European
Pharmacopoeias by the traditional Steelman-Pohley human chorionic
gonadotrophin (hCG) augmentation assay, which measures ovarian
hypertrophy following administration of exogenous FSH (in combination
with hCG) to immature female rats.23
Results and Discussion
SEC-HPLC results showed that rhFSH, which was
efficiently isolated using PuregonR,maintained its overall structural
integrity and purity. For more accurate structural comparison between
rhFSH and uhFSH, we employed a glycoproteomic strategy in characterizing
glycosylation, including determination of intact masses, sialic acid
contents, N-glycosylation sites, N-glycan structures, site-specific N- and O-glycans, and relative glycan quantities and specific activities (Figure 1a).
Molecular weight distribution of hFSHs
The intact rhFSH and uhFSH were analyzed using
SDS-PAGE. The bands showed these intact proteins comprising glycosylated
α- and β-subunits exhibited similar molecular weights of approximately
43 kDa (Figure 1b). The mass of each subunit was determined using
high-resolution MALDI-TOF/TOF mass spectrometer (Figure 1b). The α- and
β-subunits have the oretical molecular weights of 10 and 12 kDa,
respectively. The mass range of the two subunits of rhFSH was 14–15 kDa,
whereas that of the two subunits of uhFSH was 14–18 kDa. The mass of
each subunit was further analyzed by UPLC-ESI-QTOF MS (Supporting
Information Figure 1). The theoretical molecular masses of subunits and
glycan chains were searched in the deconvoluted data. The results showed
that the molecular masses of rhFSH subunits were in the range
13,710-17,264 Da, whereas those of uhFSH subunits were at 13,453-17,336
Da (Supporting Information Table 1). These results indicated that both
proteins contain heterogeneous glycosylated isoforms. Moreover, the
broader mass range of uhFSH suggested that glycosylation in uhFSH is
more complex than in rhFSH.
Sialylation level and sialic acid contents
Isoelectric focusing (IEF) was used in isolating the
hFSH charge analogues mainly according to their sialylation level, which
increases the acidity of the isoform. The pI range of the isoform
profile of rhFSH was 4–5, whereas that of uhFSH was 4–4.6 (Figure 1c).
Therefore, the difference between the isoform profile of rhFSH and uhFSH
indicated higher sialylation level in uhFSH.
Sialic acid content in each hFSH was further assessed.
The sialic acid content of rhFSH was 11.5 ± 0.04 mol/mol protein,
whereas that of uhFSH was 13.7 ± 0.04 mol/mol protein (Figure 1d).
Sialylation level was higher in uhFSH than in rhFSH, and the uhFSH/rhFSH
ratio was 1.19 ± 0.004 (p=0.0004). The sialic acid in hFSHs is highly
important in their receptor binding ability, biological activity, and
clearance from maternal circulation.24 Thus, the differences in the
sialylation level of the hFSHs results in the differences in their
biological properties.
Identification of N-Glycosylation sites
The N-glycosylation site generally contains an N-X-S/T sequence motif (where X ≠ Pro). The primary sequence of hFSH encodes two N-glycosylation sequons for each subunit;these N-glycosylation
sites are located at N52 and N78 of the α-subunit and at N7 and N24 of
the β-subunit. In high-resolution MS2 analysis, the conversion of N into
D with an increase of 0.98 Da after removing N-glycans using PNGase F can be used in identifying N-glycosylation sites. All of four N-glycosylation sites were identified in hFSHs. Most of the sites were fully occupied by N-glycans,
except βN24 in uhFSH, as the partial conversion was observed in mass
spectra after deglycosylation (Figure 1e). Partial N-glycosylation
of the FSH β-subunit was first reported in recombination bovine FSH.25
Both of the β-subunit carbohydrate residues were demonstrated to
determine the metabolic clearance rate and in vivo potency of
hFSH.26 The lack of oligosaccharides in the β-subunit possibly results
in differences in delivery rates to target tissues and in elimination
rates via filtration in the kidney.25
N-Glycan profiling and relative quantification
The N-linked glycans, released from each hFSH,
were labeled with 2-AB at their reducing ends. The chromatography peak
area was calculated for relative glycan quantification. The system was
calibrated using an external standard of dextran ladder from which the
retention time for individual glycan was converted into GU.27 The
glycans were analyzed on the basis of their GUs, which were then
compared with reference values in the “Glycobase” database for
preliminary structural assignment. The preliminary structures of the
glycans were further confirmed by their masses detected in MS.
Supporting Information Table 2 and Figure 2 show the profiling and relative contents of N-glycans
through HILIC separation and fluorescence detection. The results showed
that the two hFSHs possessed complexity in their carbohydrates, such as
(a) degree of complexity in branching; (b) mainly complex type
oligosaccharides; and (c) variations in core-/antenna-fucose and
terminal sialic acid residues. However, the mono- and di-sialylated
biantennary glycans were predominant in rhFSH at a relative amount of
46.23%, whereas the di- and tri-sialylated species, with bi- or
tri-antennary glycans, were the most common forms in uhFSH. Moreover,
the bisecting GlcNAc moieties linked to the core mannose residue were
present in uhFSH. rhFSH also exhibited a more fucosylated distribution
compared with uhFSH. The disialylated biantennary glycan
NeuAc2HexNAc4Hex5 showed the highest relative content in both hFSHs
despite the difference in sialic acid linkage of the two hFSHs.
Approximately 10.20% of the oligosaccharides of rhFSH were not
sialylated, whereas all of the glycan chains in uhFSH showed
sialylation.
Some interesting N-glycans were further
analyzed using tandem MS. LC-MS/MS analysis revealed the presence of
NeuGc-containing, antenna-fucosylated, and sulfated glycans in rhFSH.
The MS2 fragment ions obtained from the 2-AB labeled N-glycan at m/z 1180.43
described a complex type biantennary structure containing a NeuGc
residue (Figure 3a). The protonated B1α and B3α ions at m/z 308.1 and 673.2, respectively, suggested the presence of NeuGc residue. Sulfated glycans at m/z 1214.44 were also
identified (Figure 3b). Moreover, the NeuAc residue was found to be sulfated at m/z 376.1. The dehydrated sodiated ions at m/z 538.1 and 741.2 further suggested the occurrence of sulfation.
The N-glycans of uhFSH were also investigated
and data showed the presence of bisecting core-/antenna-fucosylated
isomers. Figure 3c and 3d show the MS2 spectrum of
Fuc2NeuAc1HexNAc5Hex5, [M+2H]2+. The two isomers were successfully
revealed based on the differences in retention times shown by HILIC. The
diagnostic ions indicated that the first isomer contains bisecting
biantennary structure with antenna fucosylation, whereas the second
isomer is indicative of a triantennary oligosaccharide with core and
antenna fucosylation.
rhFSH contains the NeuGc residue, which is an unusual
and immunogenic oligosaccharide.28 Moreover, core and antenna
fucosylations were both found in the two hFSHs. Fucose addition is as
important as addition of sulfate group or sialic acid because it ensures
proper binding to appropriate receptors.29 Furthermore, terminal sialic
acids and sulfate groups regulate the biological half-life of hFSH.30
Bisecting GlcNAc also increase liver and spleen uptake of glycoproteins
that possess the structural feature described above.31 The glycan
isomers of uhFSH have been detected in various peaks in chromatograms,
indicating the complexity of N-glycosylation.