Human glycodelin (Gd) is an abundant glycoprotein from the lipocalin family and is involved in crucial biological processes such as reproduction and immune reaction. In females and males, Gd is found in four distinct glycoforms–A, C, F and S–that arise from different N-linked oligosaccharide side chains at amino acid residues Asn28 and Asn63. We have expressed Gd (carrying two amino acid substitutions to improve solubility) as a non-glycosylated protein in Escherichia coli via periplasmic secretion and determined its X-ray structure at 2.45 Å resolution. Gd reveals a classical lipocalin fold including two disulfide bridges, which is however unusually compact and lacks a pronounced central pocket inside the β-barrel, in line with its low affinity for hydrophobic ligands. Instead, this lipocalin exhibits a unique homodimeric quaternary structure that appears ideally suited as a scaffold for the presentation of specific glycans. In fact, the four oligosaccharides are presented in close proximity on the same side of the dimer surface, which increases avidity for cellular receptors, e.g. during sperm–egg recognition. A bioinformatic analysis indicated that Gd orthologues exclusively occur in certain suborders of primates that have a menstrual cycle, suggesting that this lipocalin with its role in fertility only recently emerged during evolution.

INTRODUCTION

Glycodelin (Gd) is a secretory glycoprotein that regulates critical steps during fertilization and also has immunomodulatory effects. It was discovered independently in several laboratories, which initially led to the introduction of different names for this protein [1]. However, deeper biochemical analysis revealed the same amino acid sequence conjugated to different N-linked glycans, which are associated with distinct biological functions. Consequently, these protein isoforms were collectively denominated as Gd [2]. Four glycoforms of Gd have been identified in reproductive tissues and fluids: Gd-A (amniotic fluid, endometrium/decidua and maternal serum) [3,4], Gd-C (cumulus cells) [5], Gd-F (follicular fluid, luteinized granulosa cells and the oviduct) [6], and Gd-S (seminal plasma and seminal vesicles) [7].

Uterine Gd-A is secreted by the glandular epithelium of the endometrium upon progesterone stimulation. It shows contraceptive activity by binding to the sperm head, which inhibits its interaction with the zona pellucida, i.e. the translucent glycoprotein matrix that surrounds the plasma membrane of the oocyte [8]. This correlates with the low abundance of Gd-A during the fertile window [9]. On the other hand, sperm-bound Gd-A may be involved in the immunological protection of spermatozoa against maternal lymphocytes [2].

Before binding to the zona pellucida, the spermatozoa have to migrate a long distance across the uterus, through the follicular fluid which contains the glycoform Gd-F and, finally, to the cumulus oophorus. Gd-F has a similar glycan structure to Gd-A and, thus, shows comparable inhibitory effects on spermatozoa binding to the zona pellucida. In fact, both Gd-A and Gd-F glycoforms bind to the sperm head via fucosyltransferase-5, an integral membrane protein located in the acrosomal region of the spermatozoon [10]. This binding site of Gd-A and Gd-F likely coincides with the site of interaction with the zona pellucida. In addition, Gd-F was shown to inhibit the progesterone-induced acrosome reaction [11,12].

Cumulus cells of the cumulus oophorus, which surrounds the oocyte, do not synthesize Gd de novo but take up and convert Gd-A and Gd-F via glycan remodelling to the Gd-C isoform. Gd-C can displace sperm-bound Gd-A and Gd-F and, thus, stimulate binding of spermatozoa to the zona pellucida [5].

Gd-S, the fourth glycoform, is one of the most abundant glycoproteins in human seminal plasma. Unlike Gd-A, Gd-F and Gd-C, Gd-S does not interfere with zona pellucida binding of spermatozoa and, hence, is not contraceptive [7]. Instead, it inhibits the capacitation of spermatozoa, a short-lived state required for the acrosome reaction. After entering the uterine cavity, Gd-S is released from the spermatozoa by the cervical mucus [13]. Contrasting with Gd-A in females, hormonal regulation of Gd-S in men is unknown.

Apart from its role during fertilization, Gd has been found to have significant immunomodulatory effects. It inhibits T-cell proliferation and reduces their stimulation in an oligosaccharide-dependent manner [14,15]. Furthermore, Gd-A, but not Gd-S, is immunosuppressive by inducing apoptosis of monocytes and activated T-cells [16].

The human Gd preprotein consists of 180 residues, 18 of which constitute the N-terminal signal peptide [17]. The Gd gene, which comprises seven exons, is localized on chromosome 9q34 [18] under the control of a promoter with four putative glucocorticoid/progesterone-response elements [19]. An allotypic sequence variation occurs at codon 77 of the mature protein, which can encode either Glu or Gly [17,19].

Gd is a member of the lipocalin protein family and shows pronounced amino acid sequence similarity to the mammalian β-lactoglobulins [17,19,20]. Although this homology initially suggested a similar ligand-binding function for Gd, its dissociation constant for retinol as well as retinoic acid in the μM range is much higher compared with bovine β-lactoglobulin [21,22]. The mature primary structure of Gd comprises 162 amino acids with three potential N-linked glycosylation sites at Asn residues 28, 63 and 85 [17], of which only Asn28 and Asn63 were shown to be glycosylated in vivo [3,7]. Notably, so far all known Gd glycoforms have been isolated as homodimers [23,24].

In the present study, we report the three-dimensional structure of human Gd in its dimeric state and propose a model for its glycoforms.

EXPERIMENTAL

Cloning, expression and purification of Gd

The coding sequence of human Gd (UniProt ID P09466) was derived from a previously described cDNA [21,25]. This sequence encodes Gly instead of Glu at position 77, which is a known polymorphism [17]. The structural gene for residues 2–162 of mature Gd was cloned on pASK75 [26], thus fusing the protein to the N-terminal OmpA signal peptide and the C-terminal Strep-tag II (Gd-strepII), under control of a tetracycline-inducible promoter as described before [21]. According to this procedure, Gd-strepII was secreted into the periplasm of Escherichia coli and successfully purified via Strep-Tactin affinity chromatography [27].

However, the bacterially produced, non-glycosylated protein showed low solubility and did not yield any crystals in sparse matrix screens. To improve this behaviour, three side chain substitutions, N28E, N63E and F162G, were introduced using the QuikChange site-directed mutagenesis kit (Agilent) with the primer pairs 5′-GGC CAT GGC GAC CAA CGAAAT CTC CCT CAT GGC G-3′/5′-CGC CAT GAG GGA GAT TTC GTT GGT CGC CAT GGC C-3′, 5′-CGT TCT GCA CAG ATG GGA GGAAAA CAG CTG TGT TGA GAA G-3′/5′-CTT CTC AAC ACA GCT GTT TTC CTC CCA TCT GTG CAG AAC G-3′, and 5′-GGA AGA GCC GTG CCG TGG TAG CGC TTG GTC TCA CC-3′/5′-GGT GAG ACC AAG CGC TAC CAC GGC ACG GCT CTT CC-3′ (mutated codons underlined), followed by individual analysis of protein expression and solubility. Correct sequences of all plasmids were verified by automated double-stranded DNA sequencing (ABIPrism™ 310 Genetic Analyzer; Applied Biosystems). Thereafter, combinations of these mutations were tested, resulting in the version Gd(28E/162G)-strepII, which showed decreased aggregation tendency but still did not yield protein crystals.

Finally, the C-terminal Strep-tag II was replaced by a His6-tag via subcloning the coding region for Gd(28E/162G) on pASK75-his using the restriction endonucleases XbaI and Eco47III. In addition, the Gly codon at position 77 was substituted by a codon for Glu via site-directed mutagenesis with the primer pair 5′-CCT TGG AGA GAA GAC TGA GAA TCC AAA GAA GTT CAA GAT-3′/5′-CTT GAA CTT CTT TGG ATT CTC AGT CTT CTC TCC AAG GAC-3′ as above. Gd(28E/77E/162G)-His6 was produced in E. coli JM83 [28], co-transformed with the chaperone helper plasmid pTUM4 [29], as previously described [21]. Cells were grown in 2 litres of LB medium in the shake flask at 22°C under agitation. The highest yield of recombinant protein was obtained after induction at a cell density of OD550 ≈ 1.6 with 0.2 mg/l anhydrotetracycline for 4 h.

Sedimented cells were resuspended in 20 ml of ice-cold periplasmic fractionation buffer (0.5 M sucrose, 1 mM EDTA, 0.1 M Tris/HCl pH 8.0) supplemented with 100 μg/ml hen egg-white lysozyme. After incubation for 30 min on ice, spheroplasts were thoroughly centrifuged and the resulting periplasmic extract was dialysed against immobilized metal ion affinity chromatography (IMAC) buffer (0.5 M NaCl, 40 mM NaPi pH 7.5) and sterile-filtered.

Recombinant Gd was purified by IMAC on an IDA-Sepharose column (GE Healthcare) charged with Zn2+ ions using elution with a linear imidazole concentration gradient. Appropriate fractions were pooled, supplemented with 1 mM EDTA, and dialysed against 20 mM NaCl, 20 mM Tris/HCl pH 8.0. Gd(28E/77E/162G)-His6 was finally purified to homogeneity by anion-exchange chromatography (AEX) using a Resource Q column (GE Healthcare) by applying a linear concentration gradient up to 400 mM NaCl in the same buffer.

The purified protein was dialysed against 50 mM NaCl, 10 mM Tris/HCl pH 8.0 and concentrated to 7.2 mg/ml using an Amicon Ultra-4 filter (Merck Millipore). The final concentration was determined with a NanoDrop 2000 UV-Vis spectrophotometer (Thermo Fisher Scientific) using a calculated molar absorption coefficient [30] of 22710 M−1·cm−1 and a molecular weight of 19.6 kDa. The yield of purified Gd(28E/77E/162G)-His6 was approximately 0.1 mg/l E. coli culture. This protein preparation allowed concentration to at least 12 mg/ml.

Analytical size-exclusion chromatography

Size-exclusion chromatography (SEC) was performed on a 24 ml Superdex 200 10/300 GL column (GE Healthcare) in 150 mM NaCl, 20 mM Tris/HCl pH 8.0 at a flow rate of 0.5 ml/min. The column was calibrated with the following molecular weight standard proteins (Sigma–Aldrich): apoferritin (443 kDa), alcohol dehydrogenase (150 kDa), albumin (66 kDa), carbonic anhydrase (29 kDa) and cytochrome c (12.4 kDa); the void volume was determined with Blue Dextran. Based on the elution volumes, the partition coefficients Kav were calculated for each standard protein and plotted in a semi-logarithmic manner. The resulting slope was used to determine the apparent molecular weight of the purified recombinant Gd.

Protein crystallization and X-ray structure determination

Purified Gd(28E/77E/162G)-His6 was subjected to sparse matrix crystallization screening [31] using an in-house set of 480 unique conditions derived from the Biomolecular Crystallization Database [32]. Crystallization drops composed of 150 nl of protein solution at a concentration of 7.2 mg/ml and 150 nl of reservoir solution were prepared with a Freedom Evo robotic system (Tecan) in CrystalQuick 96 well sitting drop plates (Greiner, Solingen, Germany) and incubated at 20°C. Progress of crystallization was inspected with an SZX12 stereomicroscope (Olympus). Crystals appeared within 2 days and reached their final sizes after 2 weeks.

Diffraction quality crystals were obtained in the presence of 25% (w/v) polyethylene glycol (PEG) 4000, 170 mM ammonium sulfate and 15% (v/v) glycerol. The initial pH value within the crystallization drop was estimated to be 7.2, by mixing equal amounts of protein buffer and reservoir solution (in a separate experiment), whereas the latter had a pH of 4.8. Protein crystals grew as bundles of hexagonal rods, which had to be manually separated during harvest with LithoLoops (Molecular Dimensions) prior to data collection. As the reservoir solution contained PEG4000 and glycerol at concentrations sufficient for cryo-protection, the crystals were directly flash frozen in liquid nitrogen.

X-ray diffraction data were collected at beamline BL14.2 of the BESSY electron storage ring (Berlin-Adlershof) operated by the Helmholtz-Zentrum Berlin [33]. Datasets were processed with the XDS package [34]. Gd crystals diffracted up to a resolution of 2.45 Å (1 Å=0.1 mn) and belonged to the space group P6122 (cf. Table 1). The asymmetric unit contained one polypeptide chain, corresponding to a solvent content of 48% and a Matthews coefficient of 2.37 Å3/Da. The crystal structure was solved by molecular replacement with PHASER [35] using the coordinate set of reindeer β-lactoglobulin (PDB entry 1YUP, chain A) as search model. Model building and refinement were carried out in repeated cycles with COOT [36] and REFMAC5 [37]. To account for the increased flexibility of the C-terminal residues 154–160 (followed by the disordered residues 161–162 and the His6-tag), Translation, Libration and Screw (TLS) groups [38] were defined manually and used during refinement.

Table 1
X-ray data collection and refinement statistics
Parameter Value 
Data collection  
 Space group P6122 
 Unit cell parameters a=b=57.00, c=198.64 Å 
 α=β=90°, γ=120° 
 Wavelength (Å) 0.9184 
 Resolution (Å) 30–2.45 (2.55–2.45)* 
 Completeness (%) 99.9 (100.0) 
 Unique reflections 7678 (835) 
 Multiplicity 13.5 (14.1) 
 Mean I/σ(I26.5 (3.9) 
Rmeas (%) 7.4 (83.1) 
 Wilson B-factor (Å257.2 
Refinement  
 Resolution (Å) 28.5–2.45 (2.51–2.45) 
 Reflections (working) 7321 (529) 
 Reflections (test) 355 (17) 
Rcryst (%) 20.0 (25.9) 
Rfree (%) 25.7 (35.3) 
 Number of polypeptide chains/ligands 1/0 
 Number of protein atoms/waters 1242/25 
B values of protein atoms/waters (Å263.9/52.6 
 Ramachandran plot: favoured/outliers (%) 93.4/0.7 
 RMSD bonds (Å)/angles (°) 0.01/1.18 
Parameter Value 
Data collection  
 Space group P6122 
 Unit cell parameters a=b=57.00, c=198.64 Å 
 α=β=90°, γ=120° 
 Wavelength (Å) 0.9184 
 Resolution (Å) 30–2.45 (2.55–2.45)* 
 Completeness (%) 99.9 (100.0) 
 Unique reflections 7678 (835) 
 Multiplicity 13.5 (14.1) 
 Mean I/σ(I26.5 (3.9) 
Rmeas (%) 7.4 (83.1) 
 Wilson B-factor (Å257.2 
Refinement  
 Resolution (Å) 28.5–2.45 (2.51–2.45) 
 Reflections (working) 7321 (529) 
 Reflections (test) 355 (17) 
Rcryst (%) 20.0 (25.9) 
Rfree (%) 25.7 (35.3) 
 Number of polypeptide chains/ligands 1/0 
 Number of protein atoms/waters 1242/25 
B values of protein atoms/waters (Å263.9/52.6 
 Ramachandran plot: favoured/outliers (%) 93.4/0.7 
 RMSD bonds (Å)/angles (°) 0.01/1.18 

*Values in parentheses refer to the highest resolution shell.

†Ramachandran statistics were calculated with MolProbity [66].

Structural superpositions and calculations of RMSD were performed with Chimera [39]. Electrostatics was calculated with APBS [40], and ligand pockets were searched with CASTp [41]. Secondary structures were analysed with DSSP [42], whereas sequence alignments were generated with ClustalW2 [43] and visualized with Aline [44]. Protein sequences for the phylogenetic analysis were retrieved from NCBI (www.ncbi.nlm.hih.gov/genome), UniProt (www.uniprot.org) and Ensembl (www.ensembl.org). A model of the fully glycosylated Gd-A was assembled using glycan structures generated by SWEET II [45], which were attached manually with default glycan orientations generated by GLYPROT [46]. Molecular graphics were prepared with PyMOL (Schrödinger). Coordinates and structure factors for the refined Gd model have been deposited in the Protein Data Bank (PDB; www.rcsb.org) under accession code 4R0B.

RESULTS

Protein engineering, purification and structure determination

Non-glycosylated Gd with a His6 affinity tag was expressed in the presence of chaperones and disulfide catalysts [29] in the periplasm of E. coli and purified to homogeneity by IMAC and AEX. Although sufficient amounts of correctly folded and functional protein could be prepared in this manner, the solubility of Gd was too low for protein crystallization, most likely as a result of the missing glycosylation after bacterial biosynthesis. Consequently, we sought to improve expression levels, overall solubility and crystallizability by replacing some amino acid side chains in exposed positions (cf. Experimental), in line with a strategy that previously proved to be successful for the structural analysis of another human lipocalin, ApoD [47].

To this end, Asn28, representing one of the two natural glycosylation sites of Gd, was exchanged for the more polar charged residue Glu, and the hydrophobic exposed Phe162 at the C-terminus (followed by the His6-tag in the recombinant protein) was replaced by Gly. In addition, Gly at position 77 was substituted by the polar and charged residue Glu according to a natural polymorphism in humans. Indeed, the version Gd(28E/77E/162G)-His6 was produced in E. coli as a native protein dimer at higher levels (Figure 1) and with better solubility, and this protein yielded crystals that diffracted synchrotron X-rays to 2.45 Å resolution (Table 1).

Purification and analytical SEC of recombinant human Gd

Figure 1
Purification and analytical SEC of recombinant human Gd

(A) Documentation of the purification process by SDS/PAGE (Coomassie-stained). Lane M: molecular size marker (kDa); lanes 1 and 2: E. coli whole cell extract, respectively, before and after 4 h induction of gene expression; lane 3: periplasmic protein fraction; lane 4: IMAC elution peak; lane 5: pool after AEX; lane 6: sample as in lane 5, but under non-reducing conditions. (B) Analytical SEC on a Superdex 200 10/300 GL column in the presence of 150 mM NaCl, 20 mM Tris/HCl pH 8.0. Kav values of five standard proteins (circles) were plotted against the logarithm of their molecular weight. Linear regression of these Kav values was used to determine the apparent molecular weight of the purified Gd (filled circle).

Figure 1
Purification and analytical SEC of recombinant human Gd

(A) Documentation of the purification process by SDS/PAGE (Coomassie-stained). Lane M: molecular size marker (kDa); lanes 1 and 2: E. coli whole cell extract, respectively, before and after 4 h induction of gene expression; lane 3: periplasmic protein fraction; lane 4: IMAC elution peak; lane 5: pool after AEX; lane 6: sample as in lane 5, but under non-reducing conditions. (B) Analytical SEC on a Superdex 200 10/300 GL column in the presence of 150 mM NaCl, 20 mM Tris/HCl pH 8.0. Kav values of five standard proteins (circles) were plotted against the logarithm of their molecular weight. Linear regression of these Kav values was used to determine the apparent molecular weight of the purified Gd (filled circle).

The crystal structure of Gd was solved in the space group P6122 by molecular replacement using reindeer β-lactoglobulin [48] as starting model, followed by refinement to 2.45 Å resolution. The final structure comprises residues 8–160 of the mature polypeptide chain. Notably, the C-terminal residues 161–162, including the Phe162 to Gly mutation, as well as the His6-tag were not resolved in the electron density map. For interpretation of the crystal structure and illustration purposes, the second non-natural mutation, N28E, which is solvent exposed without distinct contacts to other residues, was modelled as Asn in the present report to properly reflect local electrostatics and the structural effect of this glycosylation site.

Three-dimensional structure of human Gd

Gd adopts the classical lipocalin fold, comprising a central eight-stranded antiparallel β-barrel (i.e. β-strands A–H) with an N-terminal 310-helix, a C-terminal α-helix that packs against strands FGHA as well as a ninth β-strand (I). This single strand runs partially antiparallel to β-strand A and is finally fixed via a disulfide bond between Cys160 at the C-terminus of the polypeptide chain and Cys66 in β-strand D (Figure 2) [20,49]. Although this disulfide bridge is conserved in most lipocalins, another one, which is unique to Gd (but also is found in β-lactoglobulin, see below), occurs between Cys106 and Cys119, thus connecting the neighbouring β-strands G and H.

Crystal structure of human Gd

Figure 2
Crystal structure of human Gd

The Gd monomer from the asymmetric unit is shown in cartoon representation with β-strands, helical segments and loops coloured green, light blue and grey, respectively. The disulfide bridges (yellow) and Asn side chains (with atom-type colouring) responsible for N-linked glycosylation are depicted as ball-and-sticks.

Figure 2
Crystal structure of human Gd

The Gd monomer from the asymmetric unit is shown in cartoon representation with β-strands, helical segments and loops coloured green, light blue and grey, respectively. The disulfide bridges (yellow) and Asn side chains (with atom-type colouring) responsible for N-linked glycosylation are depicted as ball-and-sticks.

The β-barrel shows a rather ellipsoidal cross section and may also be described as a sandwich of the concave β-sheets A1BCD and EFGHA2, wherein segments A1 and A2 designate the lower and upper parts of β-strand A. According to the strict geometric criteria of DSSP [42], Trp19 within this segment does not obey β-strand conformation and, thus, appears to conformationally separate residues Thr18 in A1 and His20 in A2. At the wider end of the β-barrel, the eight strands A–H are connected in a pairwise manner by four loops designated #1 to #4.

These four loops vary considerably in length, comprising 15, 4, 2 and 6 residues (i.e. residues between the segments in β-strand conformation as identified by DSSP). A short segment of the long loop #1 locally adopts a 310-helical conformation (Figure 2). The glycosylation sites at residues Asn28 and Asn63 are part of loops #1 and #2, respectively. A third potential glycosylation site at Asn85 within β-strand E, although exposed to solvent, does not show post-translational modification in vivo [3,7]. Except for a slight steric hindrance caused by the upstream glycosylation site Asn28, there is no obvious structural explanation why Asn85 is not glycosylated in vivo. However, genome-wide analyses of glycoproteins have demonstrated that about one third of all N-glycosylation sequons are not modified by oligosaccharides, due to regulatory mechanisms that are not yet fully understood [50].

Interestingly, the three-dimensional structure of Gd lacks the typical deep hydrophobic ligand pocket that is characteristic for most lipocalins [20,49]. Loops #1 and #3 completely cover the wider end of the β-barrel (Figure 2) such that the interior is inaccessible to solvent. Analysis of possible internal cavities with CASTp [41] revealed a very small hydrophobic pocket with a volume of merely 54 Å3 underneath loop #3, which is lined by seven apolar residues: Ile43, Ile56, Leu58, Val71, Ile84, Tyr86 and Leu107 (see below). Especially the bulky side chain of Tyr86 at the end of β-strand E appears to act as a lid for the pocket entrance. Opening of the cavity would require a significant conformational change of both loop #3 and its adjoining β-strands E and F, which together form a type I β-turn [51].

Dimerization of Gd

Analytical SEC of the purified recombinant protein revealed a single peak corresponding to an apparent molecular weight of 40.3 kDa, which indicates stable dimer formation of non-glycosylated Gd in solution (Figure 1), in line with previous biochemical studies of the protein from natural sources [23,24]. Also, analysis of the crystal packing with Proteins Interfaces Surfaces and Assemblies (PISA) [52] clearly suggested association to a functional dimer. The dimer contact is formed along a crystallographic 2-fold axis (note that there is only one polypeptide chain in the asymmetric unit; cf. Table 1). The interface involves 17% of the 162 residues of mature Gd, resulting in a total buried surface area (BSA) of 1180 Å2 on each monomer, which involves 20 hydrogen bonds and 4 salt bridges (counting one salt bridge per charged side chain pair).

Residues that mediate the dimer contact are predominantly found in the sequence stretch 128–158, which comprises the C-terminal α-helix and β-strand I as well as their flanking loops. In addition, loop #1 contributes 96 Å2 of BSA and two hydrogen bonds to the interface (Figures 3 and 4). Strand I plays a key role in dimerization as it forms a symmetrical antiparallel β-sheet with strand I′ of the second, symmetry-related monomer. In this way, a larger contiguous antiparallel β-sheet is formed, running across the entire dimer and altogether involving the twelve strands EFGHA2II′A2′H′G′F′E′. Notably, residues Asn28, Glu77 and Phe162, which were mutated to improve the solubility and crystallization behaviour of Gd, are not part of the dimer contact. Strikingly, both monomers of the Gd dimer are oriented such that all four glycosylation sites are presented within a narrow surface area (Figure 3), which will be discussed further below.

Dimeric quaternary structure of Gd

Figure 3
Dimeric quaternary structure of Gd

(A) Cartoon representation of the Gd dimer along and perpendicular to its dyad, which coincides with a crystallographic symmetry axis. Colouring of the monomers and display of highlighted residues is the same as in Figure 2. (B) Illustration of the surface electrostatic potential from −5 kBT/e (red) to +5 kBT/e (blue) for the protein dimer. (C) Illustration of surface polarity. Residues are coloured according to increasing hydrophobicity from green (hydrophilic) through white (neutral, including the polypeptide backbone) to brown (hydrophobic) using a group-wise parameter set [67,68].

Figure 3
Dimeric quaternary structure of Gd

(A) Cartoon representation of the Gd dimer along and perpendicular to its dyad, which coincides with a crystallographic symmetry axis. Colouring of the monomers and display of highlighted residues is the same as in Figure 2. (B) Illustration of the surface electrostatic potential from −5 kBT/e (red) to +5 kBT/e (blue) for the protein dimer. (C) Illustration of surface polarity. Residues are coloured according to increasing hydrophobicity from green (hydrophilic) through white (neutral, including the polypeptide backbone) to brown (hydrophobic) using a group-wise parameter set [67,68].

Calculation of the surface electrostatics showed a slightly negative overall charge of the dimer, which is consistent with the calculated acidic pI of 5.4 [30] for the mature non-glycosylated protein. Nevertheless, this analysis also revealed two oppositely charged faces for the Gd dimer. The surface area that harbours the four glycosylation sites is slightly positively charged, whereas the antipodal area shows a negative charge (Figure 3). Similar to the asymmetric distribution of charged residues on the dimer surface, an uneven distribution of hydrophobic patches is evident. While the glycosylation sites are found in a rather apolar surface environment, the other side of Gd is much more polar. This pattern is consistent with the observed low solubility of the protein in the absence of the highly polar sugar side chains, which effectively shield hydrophobic surface patches (Figure 5).

Comparison between human Gd and bovine β-lactoglobulin

As expected from its amino acid sequence and also from the molecular replacement solution of the crystal structure (cf. Experimental), human Gd is structurally most closely related to ruminant β-lactoglobulins. Since bovine β-lactoglobulin constitutes one of the most intensely studied lipocalins so far [53], and the structure of reindeer β-lactoglobulin is very similar, we compared the fold of human Gd with that of bovine β-lactoglobulin (PDB code 1BEB) [54]. Interestingly, the latter crystal structure contains a dimer in the asymmetric unit, with both monomers exhibiting a ‘closed’ conformation [55]. Alternatively, bovine β-lactoglobulin can also adopt an ‘open’ conformation in a pH-dependent manner (the so-called Tanford transition), which allows complexation of physiological ligands such as retinol (vitamin A) [55,56]. The amino acid sequences of mature human Gd and bovine β-lactoglobulin have the same number of residues with overall 43% identity. A structural alignment prepared with DaliLite [57] revealed 149 equivalent Cα positions, including the two conserved intramolecular disulfide bridges (Figure 4).

Comparison of Gd with β-lactoglobulin

Figure 4
Comparison of Gd with β-lactoglobulin

(A) Ribbon representation of human Gd (green) in the same orientation as bovine β-lactoglobulin (orange; PDB code 1BEB, chain A) [55]. The cavities are shown with their internal surfaces, whereas adjacent amino acid side chains are depicted as sticks. Loops #1–4 and residues lining the internal pocket of Gd are labelled. (B) Superposition of Gd and β-lactoglobulin dimers using one monomeric subunit (right) to illustrate the differing orientation of a second monomer (left). Cα atoms of residues that are involved in the dimer interface are highlighted as spheres. (C) Structure-based amino acid sequence alignment of Gd and β-lactoglobulin including secondary structure elements. Disulfide bridges and glycosylation sites are coloured yellow and purple, respectively. The sequence shifts between the β-strands I (see text) are highlighted in red. Residues that participate in dimer formation or that line the cavities are marked by spheres and stars, respectively. The allelic polymorphism of Gd residue 77 is indicated in light blue. Loops #1–4 are labelled with boxes.

Figure 4
Comparison of Gd with β-lactoglobulin

(A) Ribbon representation of human Gd (green) in the same orientation as bovine β-lactoglobulin (orange; PDB code 1BEB, chain A) [55]. The cavities are shown with their internal surfaces, whereas adjacent amino acid side chains are depicted as sticks. Loops #1–4 and residues lining the internal pocket of Gd are labelled. (B) Superposition of Gd and β-lactoglobulin dimers using one monomeric subunit (right) to illustrate the differing orientation of a second monomer (left). Cα atoms of residues that are involved in the dimer interface are highlighted as spheres. (C) Structure-based amino acid sequence alignment of Gd and β-lactoglobulin including secondary structure elements. Disulfide bridges and glycosylation sites are coloured yellow and purple, respectively. The sequence shifts between the β-strands I (see text) are highlighted in red. Residues that participate in dimer formation or that line the cavities are marked by spheres and stars, respectively. The allelic polymorphism of Gd residue 77 is indicated in light blue. Loops #1–4 are labelled with boxes.

Among these equivalent positions, the 63 residues of β-strands A–H (including Trp19 connecting strand segments A1 and A2) that form the central β-barrel show a very similar backbone conformation in both structures. Superposition of the corresponding Cα atoms yielded an RMSD of just 0.63 Å. Nevertheless, as expected [49], significant structural differences were found in loops #1–4 at the wider end of the β-barrel, with mutual RMSD values of 1.41, 1.70, 1.63 and 3.56 Å, respectively (calculated after superposition of the 63 β-barrel positions). However, the most prominent structural differences occur in the C-terminal region downsteam of β-strand H, where residues 124–160 show an overall RMSD as high as 4.97 Å. This huge deviation is caused by a register shift of the extra β-strand I towards the C-terminus of Gd. Although strand I is in an overall similar position in both protein structures, it comprises residues 149–152 in Gd but residues 147–150 in β-lactoglobulin. As a consequence, the loops that flank strand I on both sides differ in length by two residues each (Figure 4).

The shift of the C-terminal secondary structural element of Gd compared with β-lactoglobulin is in agreement with their distinct modes of dimerization. Although both dimer contacts are mediated by loop #1 and, in particular, strand I, as explained above, the dimer interface of Gd also involves the long α-helix and several residues in the upstream loop, the downstream loop that leads into β-strand I as well as the coiled stretch at the C-terminal end (Figure 4C). As result, the Gd dimer shows a much tighter contact, with more than twice the BSA (compared with a BSA of just 528 A2 per monomer of β-lactoglobulin). Moreover, this requires a mutual rotation of both monomers by about 60° (compared with the orientation in β-lactoglobulin) to bring the α-helices into contact distance (Figure 4B).

As indicated above, Gd harbours a very small cavity inside the β-barrel whereas the crystallized apo-form of bovine β-lactoglobulin shows a significantly larger obscured hydrophobic pocket in its so-called ‘closed’ conformation. With 377 Å3, this cavity has about 7-fold the size as the one of Gd (Figure 4). Closer analysis revealed that in Gd the interior of the β-barrel is more densely packed. However, side chains that point into the β-barrel are comparable in size and biochemical properties in both lipocalin structures. Dissection of the Gd and β-lactoglobulin β-barrels into their concave β-sheets A1BCD (27 Cα positions) and EFGHA2 (35 Cα positions), followed by mutual superposition of the β-sheets A1BCD from Gd and β-lactoglobulin and subsequent independent RMSD calculation, yielded a value of 0.37 Å for this pair of β-sheets but a significantly larger RMSD of 1.11 Å for the opposite pair of β-sheets EFGHA2. This rather high RMSD results mainly from the different relative positions of strands F and G, which are moved considerably closer to the A1BCD sheet in Gd than in β-lactoglobulin. This leads to the more elliptical cross-section of the β-barrel in Gd and reduces the size of the buried pocket.

DISCUSSION

Elucidation of the three-dimensional structure of human Gd has revealed a tightly packed, physiologically relevant dimer, which is consistent with previous reports that demonstrated dimeric stoichiometry for all isolated glycoforms [23,24]. The distribution pattern of solvent-exposed charged and hydrophobic residues, resulting in two oppositely charged surface areas and one side with more hydrophobic character (Figure 3), suggests that in the absence of glycosylation the dimeric protein may become sticky, in particular at low ionic strength.

This is in line with our observation that, depending on the affinity tag, the bacterially produced non-glycosylated Gd could only be concentrated to approximately 2–5 mg/ml (at low ionic strength) without forming visible aggregates. Our rational mutagenesis focused at the glycosylation sites Asn28 and Asn63 as well as the C-terminally exposed hydrophobic residue Phe162 to improve solubility and crystallizability. The combination of the amino acid exchanges N28E and F162G, together with the allotypic mutation G77E, turned out to be a good choice. Glu28 appears to compensate for both the hydrophobic patch and the slightly positive charge around this N-glycosylation site. On the other hand, Gly162 reduces the hydrophobicity of the structurally flexible C-terminus. Finally, Glu77 leads to an exposed negative charge at the loop that connects strands D and E at the closed end of the β-barrel. In the presence of these mutations, it was possible to concentrate the protein solution to at least 12 mg/ml.

Gd shows a rather compact fold with just a small, fully entrapped hydrophobic pocket inside the wider part of the β-barrel. In order to possibly bind a ligand, loop #3 would need to undergo a conformational change that allows access to this cavity. Such an alternative conformation of loop #3 has been described for bovine β-lactoglobulin, where it occurs in a pH-dependent manner, known as the Tanford transition [55]. However, the pH dependence of such a conformational change may not apply to Gd due to the lack of the responsible Glu residue at position 89, which is replaced by Ala. In addition, the very small pocket size with its sterical restriction in Gd suggests very low affinity for physiologically relevant ligands that are typical for other human members of the lipocalin family, e.g. retinol, retinoic acid or fatty acids (Figure 4A). This is in line with the 100-fold higher KD value measured for retinol compared with β-lactoglobulin [21,22]. Thus, it seems that in Gd binding activity for a small molecule ligand may no longer be relevant for its biological function.

Instead, regulation of critical steps during fertilization in humans has been linked to distinct glycoforms of Gd [58]. Therefore, both the type and spatial distribution of attached oligosaccharides appear to be crucial. Our structural study reveals that the Gd dimer provides a robust scaffold for the presentation of two pairs of complex glycans. Both monomers are oriented in such a way that all four sugar chains are displayed on the same side of the dimer and point into one direction (Figure 5). Consequently, depending on the type of glycosylation–for example, Gd-A can carry four terminal sialic acid groups–Gd can engage up to four cell surface glycan receptors (lectins) on the sperm head simultaneously. This should cause a considerable avidity effect and may explain why Gd-A can outcompete the dense sugar matrix in the glycocalyx of the zona pellucida and, thus, prevent sperm binding to the oocyte. Notably, considering the conformational flexibility of large oligosaccharides, the terminal sugar moieties should be able to bind glycan receptors that are as far as 100 Å apart.

Model of the glycosylated Gd dimer

Figure 5
Model of the glycosylated Gd dimer

(A) Both monomers of the physiological dimer are shown in surface representation (blue and green), whereas the N-linked sugars displayed by residues Asn28 and Asn63 in glycoform Gd-A are depicted as Corey-Pauling-Koltun space-filling models (or CPK models). The orientation of the protein corresponds to the upper row in Figure 3. (B) Schematic sugar representation of typical glycan structures found for Gd-A (female), corresponding to the model shown in (A), compared with Gd-S (male).

Figure 5
Model of the glycosylated Gd dimer

(A) Both monomers of the physiological dimer are shown in surface representation (blue and green), whereas the N-linked sugars displayed by residues Asn28 and Asn63 in glycoform Gd-A are depicted as Corey-Pauling-Koltun space-filling models (or CPK models). The orientation of the protein corresponds to the upper row in Figure 3. (B) Schematic sugar representation of typical glycan structures found for Gd-A (female), corresponding to the model shown in (A), compared with Gd-S (male).

Interestingly, Gd is exclusively found in particular suborders of the primates. A comprehensive bioinformatic analysis and amino acid sequence alignment (Figure 6), utilizing its two functionally relevant N-glycosylation sites as marker, revealed that Gd is conserved among the hominoidea (humans, great apes and gibbons), cercopithecoidea (old world monkeys) and at least in part among the platyrrhini (new world monkeys). The sequence with the lowest identity to human Gd was detected in the new world monkey Callithrix jacchus with 68% identical amino acid residues (for the preprotein). Interestingly, this hit is directly followed by another sequence from C. jacchus with lower homology (63% identical residues), but which lacks the characteristic N-glycosylation sites and has to be classified as a β-lactoglobulin (see below and the alignments depicted in Figures 6B and 6C).

Amino acid sequence alignments of primate Gds as well as β-lactoglobulins

Figure 6
Amino acid sequence alignments of primate Gds as well as β-lactoglobulins

(A) Different branches of the primate evolution are illustrated for the hominoidae, cercopithecoidea, platyrrhini, tarsioidea and strepsirhini, respectively, which are coloured purple, red, orange, blue and green in this schematic phylogenetic tree. Reproduced from [69]: Bosinger, S.E., Johnson, Z.P. and Silvestri, G. (2011) Primate genomes for biomedicine. Nat. Biotechnol. 29, 983–984. (B) Sequence alignment of primate Gds from H. sapiens (UNIPROT P09466), P. troglodytes (NCBI XP_003312445), P. paniscus (NCBI XP_003816990), G. gorilla gorilla (NCBI XP_004048899), P. abelii (NCBI XP_003777533), N. leucogenys (ENSEMBL ENSNLET00000010767), M. mulatta (NCBI NP_001028099), M. fascicularis (NCBI XP_005580521), P. anubis (NCBI XP_003911151), C. sabaeus (NCBI XP_008003892) and C. jacchus (ENSEMBL ENSCJAT00000019359) reveals the presence of Gd only in hominoidae, cercopithecoidea and platyrrhini. Secondary structure elements, glycosylation sites, dimer contact residues, disulfide bridges and pocket-lining residues are highlighted as in Figure 4. (C) Sequence alignment of primate β-lactoglobulins from P. abelii (NCBI XP_002820419), N. leucogenys (NCBI XP_003279687), M. mulatta (UNIPROT G7NF83), M. fascicularis (NCBI XP_005580520), C. sabaeus (NCBI XP_008003890), P. anubis (XP_003911161), P. cynocephalus (UNIPROT O77511), C. jacchus (NCBI XP_002743484), S. boliviensis boliviensis (NCBI XP_003941475), T. syrichta (NCBI XP_008068855) and O. garnettii (UNIPROT H0XUX3) in the same format as in (B). Note that there is no functional β-lactoglobulin in humans, chimpanzee, bonobo and gorilla. Triangles (plum) indicate missing N-glycosylation sites in comparison with the Gds. Potential new N-glycosylation sites that occur upon insertion of residues in loop #2 for P. abelii and N. leucogenys are highlighted in light blue. N- and C-terminal residues that were not properly aligned or ambiguous are denoted by #. Secondary structure elements were assigned according to bovine β-lactoglobulin (PDB code 1BEB, chain A) [54], serving as a prototypic orthologue [62], and coloured as in Figure 4.

Figure 6
Amino acid sequence alignments of primate Gds as well as β-lactoglobulins

(A) Different branches of the primate evolution are illustrated for the hominoidae, cercopithecoidea, platyrrhini, tarsioidea and strepsirhini, respectively, which are coloured purple, red, orange, blue and green in this schematic phylogenetic tree. Reproduced from [69]: Bosinger, S.E., Johnson, Z.P. and Silvestri, G. (2011) Primate genomes for biomedicine. Nat. Biotechnol. 29, 983–984. (B) Sequence alignment of primate Gds from H. sapiens (UNIPROT P09466), P. troglodytes (NCBI XP_003312445), P. paniscus (NCBI XP_003816990), G. gorilla gorilla (NCBI XP_004048899), P. abelii (NCBI XP_003777533), N. leucogenys (ENSEMBL ENSNLET00000010767), M. mulatta (NCBI NP_001028099), M. fascicularis (NCBI XP_005580521), P. anubis (NCBI XP_003911151), C. sabaeus (NCBI XP_008003892) and C. jacchus (ENSEMBL ENSCJAT00000019359) reveals the presence of Gd only in hominoidae, cercopithecoidea and platyrrhini. Secondary structure elements, glycosylation sites, dimer contact residues, disulfide bridges and pocket-lining residues are highlighted as in Figure 4. (C) Sequence alignment of primate β-lactoglobulins from P. abelii (NCBI XP_002820419), N. leucogenys (NCBI XP_003279687), M. mulatta (UNIPROT G7NF83), M. fascicularis (NCBI XP_005580520), C. sabaeus (NCBI XP_008003890), P. anubis (XP_003911161), P. cynocephalus (UNIPROT O77511), C. jacchus (NCBI XP_002743484), S. boliviensis boliviensis (NCBI XP_003941475), T. syrichta (NCBI XP_008068855) and O. garnettii (UNIPROT H0XUX3) in the same format as in (B). Note that there is no functional β-lactoglobulin in humans, chimpanzee, bonobo and gorilla. Triangles (plum) indicate missing N-glycosylation sites in comparison with the Gds. Potential new N-glycosylation sites that occur upon insertion of residues in loop #2 for P. abelii and N. leucogenys are highlighted in light blue. N- and C-terminal residues that were not properly aligned or ambiguous are denoted by #. Secondary structure elements were assigned according to bovine β-lactoglobulin (PDB code 1BEB, chain A) [54], serving as a prototypic orthologue [62], and coloured as in Figure 4.

Indeed, the presence of Gd in the placenta of primates other than humans has been experimentally demonstrated for the cynomolgus monkey (Macaca fascicularis) and the olive baboon (Papio anubis) [59,60]. Intriguingly, the occurrence of Gd coincides with the evolutionary appearance of menstruation in the higher primates, which is slight to overt in platyrrhini, cercopithecoidea and hominoidea but absent or covert in strepsirhini and tarsioidea [61].

Apart from that, a distinct set of Gd-related sequences, without the two N-glycosylation motifs, emerges across the primate phylogenetic tree with the exception of humans, chimpanzee and gorilla. These sequences correspond to the primate orthologue of β-lactoglobulin, another well-known member of the lipocalin protein family in mammals [53,62]. Although in humans, the β-lactoglobulin gene is inactivated and present only as a pseudogene (NCBI AF403023-25) [53], evidence for the expression and secretion of β-lactoglobulin by the mammary gland of cercopithecoidea has been obtained for the hamadryas baboon (Papio hamadryas), the rhesus monkey (Macaca mulatta) and the cynomolgus monkey (Macaca fascicularis) [6365]. These studies also confirmed that the primate β-lactoglobulins are not glycosylated, like their bovine counterpart, as indicated by the missing sequons in their primary structures. Notably, the β-lactoglobulins of hominoidae, cercopithecoidae and at least some platyrrhini show an insertion of several amino acids in loop #2 (Figure 6C), which results in a slightly larger molecular weight than for ruminant β-lactoglobulins, a feature that may well be mistaken as a sign of glycosylation in these species.

It is tempting to speculate that the emergence of the N-glycosylated Gd upward from new world monkeys results from a gene duplication event of β-lactoglobulin, accompanied by a gain of function for Gd within the haplorrhini suborder, prior to the evolutionary separation of the catarrhini and platyrrhini parvorders. Moreover, the fact that all Gd sequences exhibit the two characteristic N-glycosylation sequons underlines their functional importance and suggests that they have evolved in a rather short period of time. This is in agreement with the protein sequences found in the strepsirhini suborder and tarsioidea infraorder (within the haplorrhini suborder), which reveal the presence of β-lactoglobulin only (without the insertion in loop #2).

The notion that the human lipocalin pseudogene mentioned above indeed constitutes a silenced β-lactoglobulin–and does not represent another Gd variant–is supported by two facts: (i) the translated amino acid sequence does not contain an N-glycosylation site and (ii) it is less similar to human Gd than to β-lactoglobulin from orangutan, i.e. the closest human relative whose genome carries an intact β-lactoglobulin gene. In contrast, also no β-lactoglobulin sequence was found in chimpanzee, bonobo and gorilla. Whether these primate species have the same pseudogene as humans remains to be seen. Nevertheless, our analysis strongly indicates that the inactivation of the β-lactoglobulin gene occurred in primate evolution only recently. It further raises the question about the physiological role of β-lactoglobulin in primate milk, which seems to be restricted to amino acid supply rather than to the storage of physiologically relevant hydrophobic ligands, such as retinoids and fatty acids, as it is the case for bovine β-lactoglobulin [62].

Taken together, our structural and bioinformatic study suggests that Gd evolved quite recently and exclusively in the higher primates. Although this lipocalin may have retained a rudimentary ligand-binding capability that is probably not important for its biological function any more, the stable homodimerization of Gd clearly is a new feature. With this quaternary structure Gd provides a compact protein scaffold for the presentation of two pairs of complex glycans in an arrangement optimal for cellular oligosaccharide recognition. This novel biochemical function of a lipocalin may have implications for the biology of reproduction in humans and other higher primates.

AUTHOR CONTRIBUTION

All authors designed and analysed experiments. André Schiefner, Fabian Rodewald and Irmgard Neumaier performed the experiments. André Schiefner, Fabian Rodewald and Arne Skerra wrote the paper.

FUNDING

We thank the Helmholtz Zentrum Berlin for allocation of synchrotron radiation beamtime and financial support.

Abbreviations

     
  • AEX

    anion-exchange chromatography

  •  
  • BSA

    buried surface area

  •  
  • Gd

    glycodelin

  •  
  • IMAC

    immobilized metal ion affinity chromatography

  •  
  • SEC

    size-exclusion chromatography

References

References
1
Seppälä
M.
Koistinen
H.
Koistinen
R.
Glycodelins
Trends Endocrinol. Metab.
2001
, vol. 
12
 (pg. 
111
-
117
)
[PubMed]
2
Seppälä
M.
Koistinen
H.
Koistinen
R.
Chiu
P.C.
Yeung
W.S.
Glycosylation related actions of glycodelin: gamete, cumulus cell, immune cell and clinical associations
Hum. Reprod. Update
2007
, vol. 
13
 (pg. 
275
-
287
)
[PubMed]
3
Dell
A.
Morris
H.R.
Easton
R.L.
Panico
M.
Patankar
M.
Oehniger
S.
Koistinen
R.
Koistinen
H.
Seppälä
M.
Clark
G.F.
Structural analysis of the oligosaccharides derived from glycodelin, a human glycoprotein with potent immunosuppressive and contraceptive activities
J. Biol. Chem.
1995
, vol. 
270
 (pg. 
24116
-
24126
)
[PubMed]
4
Koistinen
H.
Easton
R.L.
Chiu
P.C.
Chalabi
S.
Halttunen
M.
Dell
A.
Morris
H.R.
Yeung
W.S.
Seppälä
M.
Koistinen
R.
Differences in glycosylation and sperm-egg binding inhibition of pregnancy-related glycodelin
Biol. Reprod.
2003
, vol. 
69
 (pg. 
1545
-
1551
)
[PubMed]
5
Chiu
P.C.
Chung
M.K.
Koistinen
R.
Koistinen
H.
Seppälä
M.
Ho
P.C.
Ng
E.H.
Lee
K.F.
Yeung
W.S.
Cumulus oophorus-associated glycodelin-C displaces sperm-bound glycodelin-A and -F and stimulates spermatozoa-zona pellucida binding
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
5378
-
5388
)
[PubMed]
6
Tse
J.Y.
Chiu
P.C.
Lee
K.F.
Seppälä
M.
Koistinen
H.
Koistinen
R.
Yao
Y.Q.
Yeung
W. S.
The synthesis and fate of glycodelin in human ovary during folliculogenesis
Mol. Hum. Reprod.
2002
, vol. 
8
 (pg. 
142
-
148
)
[PubMed]
7
Morris
H.R.
Dell
A.
Easton
R.L.
Panico
M.
Koistinen
H.
Koistinen
R.
Oehninger
S.
Patankar
M.S.
Seppälä
M.
Clark
G.F.
Gender-specific glycosylation of human glycodelin affects its contraceptive activity
J. Biol. Chem.
1996
, vol. 
271
 (pg. 
32159
-
32167
)
[PubMed]
8
Oehninger
S.
Coddington
C.C.
Hodgen
G.D.
Seppälä
M.
Factors affecting fertilization: endometrial placental protein 14 reduces the capacity of human spermatozoa to bind to the human zona pellucida
Fertil. Steril.
1995
, vol. 
63
 (pg. 
377
-
383
)
[PubMed]
9
Seppälä
M.
Koistinen
H.
Koistinen
R.
Dell
A.
Morris
H.R.
Oehninger
S.
Clark
G.F.
Glycodelins as regulators of early events of reproduction
Clin. Endocrinol.
1997
, vol. 
46
 (pg. 
381
-
386
)
10
Chiu
P.C.
Chung
M.K.
Koistinen
R.
Koistinen
H.
Seppälä
M.
Ho
P.C.
Ng
E.H.
Lee
K.F.
Yeung
W.S.
Glycodelin-A interacts with fucosyltransferase on human sperm plasma membrane to inhibit spermatozoa-zona pellucida binding
J. Cell Sci.
2007
, vol. 
120
 (pg. 
33
-
44
)
[PubMed]
11
Chiu
P.C.
Koistinen
R.
Koistinen
H.
Seppälä
M.
Lee
K.F.
Yeung
W.S.
Zona-binding inhibitory factor-1 from human follicular fluid is an isoform of glycodelin
Biol. Reprod.
2003
, vol. 
69
 (pg. 
365
-
372
)
[PubMed]
12
Chiu
P.C.
Tsang
H.Y.
Koistinen
R.
Koistinen
H.
Seppälä
M.
Lee
K.F.
Yeung
W.S.
The contribution of D-mannose, L-fucose, N-acetylglucosamine, and selectin residues on the binding of glycodelin isoforms to human spermatozoa
Biol. Reprod.
2004
, vol. 
70
 (pg. 
1710
-
1719
)
[PubMed]
13
Chiu
P.C.
Chung
M.K.
Tsang
H.Y.
Koistinen
R.
Koistinen
H.
Seppälä
M.
Lee
K. F.
Yeung
W.S.
Glycodelin-S in human seminal plasma reduces cholesterol efflux and inhibits capacitation of spermatozoa
J. Biol. Chem.
2005
, vol. 
280
 (pg. 
25580
-
25589
)
[PubMed]
14
Rachmilewitz
J.
Borovsky
Z.
Riely
G.J.
Miller
R.
Tykocinski
M.L.
Negative regulation of T cell activation by placental protein 14 is mediated by the tyrosine phosphatase receptor CD45
J. Biol. Chem.
2003
, vol. 
278
 (pg. 
14059
-
14065
)
[PubMed]
15
Rachmilewitz
J.
Riely
G.J.
Tykocinski
M.L.
Placental protein 14 functions as a direct T-cell inhibitor
Cell Immunol.
1999
, vol. 
191
 (pg. 
26
-
33
)
[PubMed]
16
Mukhopadhyay
D.
SundarRaj
S.
Alok
A.
Karande
A.A.
Glycodelin A, not glycodelin S, is apoptotically active. Relevance of sialic acid modification
J. Biol. Chem.
2004
, vol. 
279
 (pg. 
8577
-
8584
)
[PubMed]
17
Julkunen
M.
Seppälä
M.
Janne
O.A.
Complete amino acid sequence of human placental protein 14: a progesterone-regulated uterine protein homologous to β-lactoglobulins
Proc. Natl. Acad. Sci. U.S.A.
1988
, vol. 
85
 (pg. 
8845
-
8849
)
[PubMed]
18
Van Cong
N.
Vaisse
C.
Gross
M.S.
Slim
R.
Milgrom
E.
Bernheim
A.
The human placental protein 14 (PP14) gene is localized on chromosome 9q34
Hum. Genet.
1991
, vol. 
86
 (pg. 
515
-
518
)
[PubMed]
19
Vaisse
C.
Atger
M.
Potier
B.
Milgrom
E.
Human placental protein 14 gene: sequence and characterization of a short duplication
DNA Cell Biol.
1990
, vol. 
9
 (pg. 
401
-
413
)
[PubMed]
20
Flower
D.R.
The lipocalin protein family: structure and function
Biochem. J.
1996
, vol. 
318
 (pg. 
1
-
14
)
[PubMed]
21
Breustedt
D.A.
Schönfeld
D.L.
Skerra
A.
Comparative ligand-binding analysis of ten human lipocalins
Biochim. Biophys. Acta
2006
, vol. 
1764
 (pg. 
161
-
173
)
[PubMed]
22
Fugate
R.D.
Song
P.S.
Spectroscopic characterization of β-lactoglobulin-retinol complex
Biochim. Biophys. Acta
1980
, vol. 
625
 (pg. 
28
-
42
)
[PubMed]
23
Koistinen
H.
Koistinen
R.
Dell
A.
Morris
H.R.
Easton
R.L.
Patankar
M.S.
Oehninger
S.
Clark
G.F.
Seppälä
M.
Glycodelin from seminal plasma is a differentially glycosylated form of contraceptive glycodelin-A
Mol. Hum. Reprod.
1996
, vol. 
2
 (pg. 
759
-
765
)
[PubMed]
24
Vigne
J.L.
Hornung
D.
Mueller
M.D.
Taylor
R.N.
Purification and characterization of an immunomodulatory endometrial protein, glycodelin
J. Biol. Chem.
2001
, vol. 
276
 (pg. 
17101
-
17105
)
[PubMed]
25
Dutta
B.
Mukhopadhyay
D.
Roy
N.
Das
G.
Karande
A.A.
Cloning, expression, purification, and immunocharacterization of placental protein-14
Protein Expr. Purif.
1998
, vol. 
14
 (pg. 
327
-
334
)
[PubMed]
26
Skerra
A.
Use of the tetracycline promoter for the tightly regulated production of a murine antibody fragment in Escherichia coli
Gene
1994
, vol. 
151
 (pg. 
131
-
135
)
[PubMed]
27
Schmidt
T.G.
Skerra
A.
The Strep-tag system for one-step purification and high-affinity detection or capturing of proteins
Nat. Protoc.
2007
, vol. 
2
 (pg. 
1528
-
1535
)
[PubMed]
28
Yanisch-Perron
C.
Vieira
J.
Messing
J.
Improved M13 phage cloning vectors and host strains: nucleotide sequences of the M13mp18 and pUC19 vectors
Gene.
1985
, vol. 
33
 (pg. 
103
-
119
)
[PubMed]
29
Schlapschy
M.
Grimm
S.
Skerra
A.
A system for concomitant overexpression of four periplasmic folding catalysts to improve secretory protein production in Escherichia coli
Protein Eng. Des. Sel.
2006
, vol. 
19
 (pg. 
385
-
390
)
[PubMed]
30
Gasteiger
E.
Gattiker
A.
Hoogland
C.
Ivanyi
I.
Appel
R.D.
Bairoch
A.
ExPASy: The proteomics server for in-depth protein knowledge and analysis
Nucleic Acids Res.
2003
, vol. 
31
 (pg. 
3784
-
3788
)
[PubMed]
31
Jancarik
J.
Kim
S.-H.
Sparse matrix sampling: a screening method for crystallization of proteins
J. Appl. Crystallogr.
1991
, vol. 
24
 (pg. 
409
-
411
)
32
Tung
M.
Gallagher
D.T.
The Biomolecular Crystallization Database Version 4: expanded content and new features
Acta Crystallogr. D Biol. Crystallogr.
2009
, vol. 
65
 (pg. 
18
-
23
)
[PubMed]
33
Mueller
U.
Darowski
N.
Fuchs
M.R.
Förster
R.
Hellmig
M.
Paithankar
K.S.
Pühringer
S.
Steffien
M.
Zocher
G.
Weiss
M.S.
Facilities for macromolecular crystallography at the Helmholtz-Zentrum Berlin
J. Synchrotron Radiat.
2012
, vol. 
19
 (pg. 
442
-
449
)
[PubMed]
34
Kabsch
W.
XDS
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
125
-
132
)
[PubMed]
35
McCoy
A.J.
Grosse-Kunstleve
R.W.
Adams
P.D.
Winn
M.D.
Storoni
L.C.
Read
R.J.
Phaser crystallographic software
J. Appl. Crystallogr.
2007
, vol. 
40
 (pg. 
658
-
674
)
[PubMed]
36
Emsley
P.
Lohkamp
B.
Scott
W.G.
Cowtan
K.
Features and development of Coot
Acta Crystallogr. D Biol. Crystallogr.
2010
, vol. 
66
 (pg. 
486
-
501
)
[PubMed]
37
Murshudov
G.N.
Skubak
P.
Lebedev
A.A.
Pannu
N.S.
Steiner
R.A.
Nicholls
R.A.
Winn
M.D.
Long
F.
Vagin
A.A.
REFMAC5 for the refinement of macromolecular crystal structures
Acta Crystallogr. D Biol. Crystallogr.
2011
, vol. 
67
 (pg. 
355
-
367
)
[PubMed]
38
Painter
J.
Merritt
E.A.
Optimal description of a protein structure in terms of multiple groups undergoing TLS motion
Acta Crystallogr. D Biol. Crystallogr.
2006
, vol. 
62
 (pg. 
439
-
450
)
[PubMed]
39
Pettersen
E.F.
Goddard
T.D.
Huang
C.C.
Couch
G.S.
Greenblatt
D.M.
Meng
E.C.
Ferrin
T.E.
UCSF Chimera—a visualization system for exploratory research and analysis
J. Comput. Chem.
2004
, vol. 
25
 (pg. 
1605
-
1612
)
[PubMed]
40
Baker
N.A.
Sept
D.
Joseph
S.
Holst
M.J.
McCammon
J.A.
Electrostatics of nanosystems: application to microtubules and the ribosome
Proc. Natl. Acad. Sci. U.S.A.
2001
, vol. 
98
 (pg. 
10037
-
10041
)
[PubMed]
41
Dundas
J.
Ouyang
Z.
Tseng
J.
Binkowski
A.
Turpaz
Y.
Liang
J.
CASTp: computed atlas of surface topography of proteins with structural and topographical mapping of functionally annotated residues
Nucleic Acids Res.
2006
, vol. 
34
 (pg. 
W116
-
W118
)
[PubMed]
42
Kabsch
W.
Sander
C.
Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features
Biopolymers
1983
, vol. 
22
 (pg. 
2577
-
2637
)
[PubMed]
43
Larkin
M.A.
Blackshields
G.
Brown
N.P.
Chenna
R.
McGettigan
P.A.
McWilliam
H.
Valentin
F.
Wallace
I.M.
Wilm
A.
Lopez
R.
, et al. 
Clustal W and Clustal X version 2.0
Bioinformatics
2007
, vol. 
23
 (pg. 
2947
-
2948
)
[PubMed]
44
Bond
C.S.
Schüttelkopf
A.W.
ALINE: a WYSIWYG protein-sequence alignment editor for publication-quality alignments
Acta Crystallogr. D Biol. Crystallogr.
2009
, vol. 
65
 (pg. 
510
-
512
)
[PubMed]
45
Bohne
A.
Lang
E.
von der Lieth
C.W.
SWEET-WWW-based rapid 3D construction of oligo- and polysaccharides
Bioinformatics
1999
, vol. 
15
 (pg. 
767
-
768
)
[PubMed]
46
Bohne-Lang
A.
von der Lieth
C.W.
GlyProt: in silico glycosylation of proteins
Nucleic Acids Res.
2005
, vol. 
33
 (pg. 
W214
-
W219
)
[PubMed]
47
Nasreen
A.
Vogt
M.
Kim
H.J.
Eichinger
A.
Skerra
A.
Solubility engineering and crystallization of human apolipoprotein D
Protein Sci.
2006
, vol. 
15
 (pg. 
190
-
199
)
[PubMed]
48
Oksanen
E.
Jaakola
V.P.
Tolonen
T.
Valkonen
K.
Åkerström
B.
Kalkkinen
N.
Virtanen
V.
Goldman
A.
Reindeer β-lactoglobulin crystal structure with pseudo-body-centred noncrystallographic symmetry
Acta Crystallogr. D Biol. Crystallogr.
2006
, vol. 
62
 (pg. 
1369
-
1374
)
[PubMed]
49
Skerra
A.
Lipocalins as a scaffold
Biochim. Biophys. Acta
2000
, vol. 
1482
 (pg. 
337
-
350
)
[PubMed]
50
Aebi
M.
N-linked protein glycosylation in the ER
Biochim. Biophys. Acta
2013
, vol. 
1833
 (pg. 
2430
-
2437
)
[PubMed]
51
Hutchinson
E.G.
Thornton
J.M.
A revised set of potentials for β-turn formation in proteins
Protein Sci.
1994
, vol. 
3
 (pg. 
2207
-
2216
)
[PubMed]
52
Krissinel
E.
Henrick
K.
Inference of macromolecular assemblies from crystalline state
J. Mol. Biol.
2007
, vol. 
372
 (pg. 
774
-
797
)
[PubMed]
53
Kontopidis
G.
Holt
C.
Sawyer
L.
β-lactoglobulin: binding properties, structure, and function
J. Dairy Sci.
2004
, vol. 
87
 (pg. 
785
-
796
)
[PubMed]
54
Brownlow
S.
Morais Cabral
J.H.
Cooper
R.
Flower
D.R.
Yewdall
S.J.
Polikarpov
I.
North
A.C.
Sawyer
L.
Bovine β-lactoglobulin at 1.8 Å resolution—still an enigmatic lipocalin
Structure
1997
, vol. 
5
 (pg. 
481
-
495
)
[PubMed]
55
Qin
B.Y.
Bewley
M.C.
Creamer
L.K.
Baker
H.M.
Baker
E.N.
Jameson
G.B.
Structural basis of the Tanford transition of bovine β-lactoglobulin
Biochemistry
1998
, vol. 
37
 (pg. 
14014
-
14023
)
[PubMed]
56
Kontopidis
G.
Holt
C.
Sawyer
L.
The ligand-binding site of bovine β-lactoglobulin: evidence for a function?
J. Mol. Biol.
2002
, vol. 
318
 (pg. 
1043
-
1055
)
57
Holm
L.
Park
J.
DaliLite workbench for protein structure comparison
Bioinformatics
2000
, vol. 
16
 (pg. 
566
-
567
)
[PubMed]
58
Seppälä
M.
Koistinen
H.
Koistinen
R.
Hautala
L.
Chiu
P.C.
Yeung
W.S.
Glycodelin in reproductive endocrinology and hormone-related cancer
Eur. J. Endocrinol.
2009
, vol. 
160
 (pg. 
121
-
133
)
[PubMed]
59
Fazleabas
A.T.
Donnelly
K.M.
Hild-Petito
S.
Hausermann
H.M.
Verhage
H.G.
Secretory proteins of the baboon (Papio anubis) endometrium: regulation during the menstrual cycle and early pregnancy
Hum. Reprod. Update
1997
, vol. 
3
 (pg. 
553
-
559
)
[PubMed]
60
Inaba
N.
Sato
N.
Fukazawa
I.
Ota
Y.
Shirotake
S.
Takamizawa
H.
Nozawa
S.
Bohn
H.
The immunocytochemical localization of new soluble placental tissue proteins (PP14, 16, 17, 19, 20 and PP21) in human and cynomolgus monkey placentae
Arch. Gynecol.
1987
, vol. 
240
 (pg. 
13
-
19
)
[PubMed]
61
Strassmann
B.I.
The evolution of endometrial cycles and menstruation
Q. Rev. Biol.
1996
, vol. 
71
 (pg. 
181
-
220
)
[PubMed]
62
Sawyer
L.
Kontopidis
G.
The core lipocalin, bovine β-lactoglobulin
Biochim. Biophys. Acta
2000
, vol. 
1482
 (pg. 
136
-
148
)
[PubMed]
63
Azuma
N.
Yamauchi
K.
Identification of α-lactalbumin and β-lactoglobulin in cynomolgus monkey (Macaca fascicularis) milk
Comp. Biochem. Physiol. B.
1991
, vol. 
99
 (pg. 
917
-
921
)
[PubMed]
64
Hall
A.J.
Masel
A.
Bell
K.
Halliday
J.A.
Shaw
D.C.
VandeBerg
J.L.
Characterization of baboon (Papio hamadryas) milk proteins
Biochem. Genet.
2001
, vol. 
39
 (pg. 
59
-
71
)
[PubMed]
65
Kunz
C.
Lönnerdal
B.
Isolation and characterization of a 21 kDa whey protein in rhesus monkey (Macaca mulatta) milk
Comp. Biochem. Physiol. Biochem. Mol. Biol.
1994
, vol. 
108
 (pg. 
463
-
469
)
[PubMed]
66
Davis
I.W.
Leaver-Fay
A.
Chen
V.B.
Block
J.N.
Kapral
G.J.
Wang
X.
Murray
L.W.
Arendall
3rd W.B.
Snoeyink
J.
Richardson
J.S.
Richardson
D.C.
MolProbity: all-atom contacts and structure validation for proteins and nucleic acids
Nucleic Acids Res.
2007
, vol. 
35
 (pg. 
W375
-
W383
)
[PubMed]
67
Eichinger
A.
Nasreen
A.
Kim
H.J.
Skerra
A.
Structural insight into the dual ligand specificity and mode of high density lipoprotein association of apolipoprotein D
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
31068
-
31075
)
[PubMed]
68
Tao
P.
Wang
R.
Lai
L.
Calculating partition coefficients of peptides by the addition method
J. Mol. Model.
1999
, vol. 
5
 (pg. 
189
-
195
)
69
Bosinger
S.E.
Johnson
Z.P.
Silvestri
G.
Primate genomes for biomedicine
Nat. Biotechnol.
2011
, vol. 
29
 (pg. 
983
-
984
)
[PubMed]