Lectin-Binding Specificity of the Fertilization-Relevant Protein PDC-109 by Means of Surface Plasmon Resonance and Carbohydrate REcognition Domain EXcision-Mass Spectrometry

Seminal plasma proteins are relevant for sperm functionality and some appear responsible for establishing sperm interactions with the various environments along the female genital tract towards the oocyte. In recent years, research has focused on characterizing the role of these proteins in the context of reproductive biology, fertility diagnostics and treatment of related problems. Herein, we focus on the main protein of bovine seminal plasma, PDC-109 (BSP-A1/-A2), which by virtue of its lectin properties is involved in fertilization. By means of surface plasmon resonance, the interaction of PDC-109 with a panel of the most relevant glycosidic epitopes of mammals has been qualitatively and quantitatively characterized, and a higher affinity for carbohydrates containing fucose has been observed, in line with previous studies. Additionally, using the orthogonal technique of Carbohydrate REcognition Domain EXcision-Mass Spectrometry (CREDEX-MS), the recognition domain of the interaction complexes between PDC-109 and all fucosylated disaccharides [(Fuc-α1,(3,4,6)-GlcNAc)] has been defined, revealing the specific glycotope and the peptide domain likely to act as the PDC-109 carbohydrate binding site.


Introduction
In the male genital tract, secretions from the testes, epididymis, seminal vesicles, and other accessory glands contribute to the fluid portion of semen (i.e., seminal plasma), in which sperm cells (i.e., spermatozoa) are suspended. Mammal seminal plasma is a complex fluid that serves as carrier for the spermatozoa in their journey to their target, the uterus. However, since the beginning of the 20th century, researchers realized that seminal plasma played a more important role in sperm biology than that of a mere vehicle for spermatozoa. Growing evidence has pointed to this "acellular" part of the semen as a fundamental player in mammalian fertilization, and in recent years intensive research has led to the identification and functional characterization of several seminal plasma proteins. Since sperm cells have a limited biosynthetic ability [1], the interaction of different seminal plasma proteins with spermatozoa, as well as with proteins from the female genital tract encountered during the journey to the site of fertilization, will have a profound impact on sperm functionality. Proteins from the seminal fluid will be adsorbed on the sperm surface, remodeling

Synthesis of Neo-Glycoprobes Exposing Relevant Mammalian Carbohydrate Epitopes
We aimed to study the lectin-binding profile of PDC-109 using purified protein material from biological origin, hence initial work was focused on the synthesis, purification and validation of new glycoprobes displaying some of the most relevant glycotopes in the mammalian system (Table 1).
All the glycopeptides were synthesized as previously reported by our group [16] using oxime ligation between the oligosaccharides in Table 1 and the peptide module N[Me]-O-Aoa-GFKKG at 25 and 20 mM, respectively. Ligation reactions were performed at pH 3.5 (NAc-hexoses) or pH 4.6 (hexoses) and 37 • C for 72 h, the conjugates purified by high performance liquid chromatography (HPLC) and characterized by mass spectrometry (MS) as detailed in Materials and Methods. In addition to choosing an optimal pH for every hexose type, the use of aniline as catalyst to reduce incubation time was evaluated, as some authors have reported significant yield improvements (nearly double in 3 h) in conjugations involving non-methylated aminoxyacetic acid (Aoa)-containing peptides and monosaccharides [17]. However, with our N[Me]-O-Aoa-peptide the yield increased only marginally after 72 h incubation, except for reactions with Fuc-containing i.e., [Fuc-α1, (3,4,6)-GlcNAc] disaccharides (e.g., from 8% to 16% for Fuc-α1,4-GlcNAc). Table 1. Carbohydrate epitopes commonly encountered in mammalian specimens and corresponding lectins with well-defined binding properties.

Epitope Carbohydrate Lectin
Disaccharides with terminal β-galactose Gal-β1,4-Glc Erythrina cristagalli (ECA) Gal-β1,4-GlcNAc Gal-β1, 3 A substantial problem during the production of the glycopeptides was the purification by reversed phase (RP) HPLC, as the small change in hydrophobicity between the glycopeptide product and unreacted N[Me]-O-Aoa-peptide precursor rendered separations difficult, particularly in conjugations with <20% conversion. Attempts to increase the retention time of unreacted peptide by addition of carbonyl scavengers such as formaldehyde and acetone, or the alkylating agent N-ethylmaleimide [18], did not improve resolution. An additional difficulty, also related to purification, was that residual (trifluoroacetic) acid from HPLC eluents caused degradation of the N[Me]-O-Aoa-peptides during lyophilization. This problem was conveniently solved by adjusting to pH 5 the glycoprobe solution immediately after collection. A further test of the suitability of the conjugation conditions involved the highly acid-labile trisaccharides containing terminal sialic (N-acetyl neuraminic) acid (Sia), which could be successfully conjugated to the N[Me]-O-Aoa-peptide module at mildly acidic pH, purified and characterized by MS like other glycoconjugates. Finally, the long-term stability of all compounds was shown to be satisfactory (>95% unaltered after 2 years) provided they are stored in lyophilized form. This renders our glycopeptide probe approach useful for longitudinal studies in carbohydrate research.

Neo-Glycoprobe Evaluation by SPR Studies with Known Lectins
With neo-glycopeptides displaying the most relevant glycotopes of the mammalian system in hand, their functionality was validated in SPR interaction studies using plant lectins (Table 1) with known selectivity for the indicated carbohydrate epitopes. Lectin solutions were flown across a surface with the immobilized cognate sugar unit. Ac-GFKKG-amide, a non-glycosylated probe, was used as reference surface, given that the previously described N[Me]-O-Aoa-GFKKG-amide [16] displayed undesirable, non-specific binding with PDC-109.
Subsequently, glycoprobes displaying mannobioses only differing in their glycosidic linkage (Man-α1,2-Man, Man-α1,3-Man and Man-α1,6-Man) were immobilized on a sensor chip and their interaction with concanavalin A (Con A) was studied. Con A is a relatively complex lectin, organized as a β-barrel-like tetramer at pH > 7, each dimer subunit (D2 symmetry) with two CRDs situated on opposing faces of the protein. Sensorgrams of different Con A concentrations were recorded and kinetic constants determined by simultaneously fitting the experimental curves to a bivalent kinetic model ( Figure 1). This model allows the first and second binding events to be separately described by two sets of rate constants (Table 2).  Table 2. Kinetic constants of the interaction of Con A with mannobiose-containing glycoprobes determined by SPR. The unconventional response units (RU −1 ) used to describe the second association event made comparison with constants derived from other methods (e.g., Isothermal Titration Calorimetry, ITC) difficult, and only the affinity constants for the first event, K A1 (K A1 = k a1 /k d1 ) could be accurately compared. For this event, a higher affinity of Man-α1,3-Man over Man-α1,6-Man was found, in agreement with previous ITC data [19], although the K A1 values obtained with the ITC method were one order of magnitude lower (~10 4 ) than with our SPR approach (~10 5 ). One possible explanation for this difference, keeping in mind the two-CRDs-per-dimer model of Con A, is that although the bivalent binding model used in SPR only allows to derive standard-unit values for K A1 , the apparent k d1 , used for K A1 determination is also influenced by the second event. If this contribution is not factored in, a lower k d1 , and consequently a higher K A1 , results.
The last group of disaccharide glycoprobes to be evaluated included those with terminal Fuc units (Fuc-α1,3-GlcNAc, Fuc-α1,4-GlcNAc and Fuc-α1,6-GlcNAc). The latter epitope corresponds to the core-fucosylation of N-glycans, and the former two are partial Lewis x and Lewis a epitopes, respectively. The glycoprobe surfaces were tested against two Fuc-specific lectins from Lotus tetragonolobus (LTA) and Ulex europeaus (UEA). For both lectins, sensorgrams clearly demonstrated affinity for Fuc epitopes (Figure 2; representative example for Fuc-α1,4-GlcNAc), even though the carbohydrate affinity for both lectins has been described to be strongly specific for the blood group H determinant (Fuc-α1,2-Gal) [20,21], not considered in our study. -O-Aoa-GFKKG-amide glycoprobes, representative of the two existing linkage types in adult mammalian glycoproteins, was prepared and the two Sia-specific lectins were flown over the surfaces. As the MAA lectin reportedly requires three intact sugar units for binding [22], our results support a native ring-closed structure for the first monosaccharide. The binding responses observed for the two lectins were in perfect agreement with their reported carbohydrate specificity. Thus, whereas SNA showed a marked preference for the Sia-α2,6-lacNAc isomer, MAA recognized only the Sia-α2,3-lacNAc containing glycoprobe ( Figure 3). For kinetic analysis, several concentrations in the 250 nM to 1.9 µM range for MAA and in the 74 to 563 nM range for SNA were analyzed. A 1:1 Langmuir binding model was used for sensorgram fitting ( Figure 4, Table 3). As expected for interactions involving trisaccharides, the K A equilibrium constants obtained were in the 10 6 -10 7 M −1 range. For MAA, the affinity constant determined by our approach was 9.12 × 10 5 M −1 ; of the same magnitude than reported previously [23], but one order lower than other constants determined using complex neoglycoproteins and immobilized lectins [24], in both cases by SPR. The higher affinity for neoglycoproteins can easily be explained by their multivalent nature. For SNA, a K A of 6.27 × 10 6 M −1 was determined, consistent with the previously reported value [25] of 6.7 × 10 6 M −1 .  Table 3. Kinetic constants of MAA and SNA to their specific sialic acid-containing glycoprobes determined by SPR.

PDC-109 Binding Profile and Kinetic Studies
While the lipid selectivity and kinetics of PDC-109 has been characterized in considerable detail [11,26], its interaction with carbohydrate epitopes remains largely uncharted, despite its potential role in sperm reservoir formation in the female genital tract. We have therefore performed SPR experiments with PDC-109 and the most relevant mammalian glycotopes, particularly those containing Fuc epitopes. Once SPR sensor chips were functionalized with the different synthetic glycoprobes and evaluated with specific lectins (Section 2.1.2) they were used to characterize both qualitatively and quantitatively the interaction of PDC-109. For kinetic analysis, several concentrations in the 6.25 to 100 µM range were analyzed and a 1:1 Langmuir binding model was chosen for sensorgram fitting.
In general, considerable variability was observed in the kinetic experiments, with CV (from 3 to 6 repeat experiments) exceeding 50% in particular cases. Data dispersion was attributable to the unusual sensorgrams observed, reflecting a complex binding event, particularly in the association phase of PDC-109. As shown in Figure 5a, the usual signal rise observed right after sample injection and due to bulk effects-refractive index differences between analyte solution and flow buffer-was followed by a rather intriguing association phase. In a conventional, straightforward binding event, the response-i.e., the signal rise over time, reflecting protein binding to the sensor surface-reaches a constant value once equilibrium is established, and remains so until the protein solution passed over the sensor chip is replaced with buffer, giving way to the spontaneous dissociation process and subsequent surface regeneration. However, in the present case visual inspection of all PDC-109 sensorgrams (those in Figure 5a being representative) clearly shows a biphasic phenomenon during the association phase (displayed enlarged in Figure 5a). Thus, response increased until a high value (maximum binding) was reached, then slowly decreased even though PDC-109 was still being flown across the surface. Such biphasic behavior during the association phase was observed not only with all immobilized glycoprobes but also with the reference surface ( Figure 5a; green and blue trace sensorgrams, respectively), hence the differential curve resulting from reference subtraction ( Figure 5b) displayed an uneven shape, with large spikes at both the beginning and the end of the injection that complicated data fitting. As a result, values for the kinetic constants may include some variability, although a reliable epitope preference could nonetheless be established.
This unusual biphasic profile, reflecting loss of mass from the sensor surface during the association phase, was not observed with any of the standard lectins studied earlier (Figure 5c), and suggested an uncharacteristic binding behavior of PDC-109 that could be confirmed after meticulous evaluation of all possible variables (surface, ligand, analyte, buffer conditions, competitive reagents, solute stability, etc.) and with some precedents observed in other systems [26].
Plausible explanations for the unusual behavior of PDC-109, accounting for the biphasic shape of the SPR sensorgrams, would assume the existence of two different affinities, namely two competing analyte reactions, in the protein preparation. This can be due either to a single analyte with two different affinities, or to two different analytes, each with its own affinity.
Regarding the former situation, the possibility of non-specific binding between PDC-109 and the SPR surface was investigated, since the carboxymethyled dextran matrix on the sensor surface is structurally not too distant from the glycoprobe ligands, hence a weak interaction with the lectin could not be excluded. To this end, a "saturated" solution of free carboxymethylated dextran (CM) (identical to that present as CM5 surface coating) was added to the analyte solution; if the analyte had any affinity for either dextran or carboxymethylated variant, it would bind preferentially the free polymer over that on the sensor chip, thus reducing non-specific binding and avoiding competing reactions. This modification, however, did not have any effect on the binding curves, which remained biphasic in all cases. In a complementary attempt, a C1 chip lacking the dextran matrix (only carboxymethyl groups directly bound to the gold surface) was tested to exclude any influence of the matrix; again, the same biphasic profile could be observed during the association phase. From these results, we concluded that no competing reactions with the SPR surface existed, hence the biphasic behavior must proceed from analyte heterogeneity, i.e., two analytes competing for the same ligand. If such was the case, the decreasing region in the association phase curve could result from a smaller molecule with higher ligand affinity but slower kinetics competing with a faster but lower-affinity interaction.
Different possibilities of analyte heterogeneity can be envisaged: (i) PDC-109 may undergo conformational changes upon binding to the epitope; (ii) more than one isoform/glycoform is present in the PDC-109 sample; or (iii) PDC-109 has different aggregation states; all these situations might produce sensorgrams with biphasic association curves, such as those in Figure 6. Previous reports suggest PDC-109 undergoes a conformational change upon binding to PC membranes [27]. However, as SPR can only detect refractive index variations related to mass changes over time, subtle conformational changes not resulting in mass changes will go unnoticed. However, while beyond the scope of the SPR technique, heterogeneity due to conformational change would not explain the SPR signal reduction. To investigate the possibility of a heterogeneous analyte, several different approaches were followed. Firstly, considering that PDC-109 exists in two major forms (i.e., glycosylated and non-glycosylated) [10], enzymatic deglycosylation was performed in order to obtain a single analyte isoform. Given the limited availability of purified PDC-109, the amount of protein used for deglycosylation did not allow a quantitative purification step. The extent of the reaction was assessed by matrix assisted laser desorption ionization-time of flight (MALDI-TOF) MS, which showed that glycan removal could not be brought to completion. Even so, the partially deglycosylated PDC-109 was tested on the same sensor surfaces previously used with native protein. The resulting sensorgrams clearly showed a rising association phase without biphasic behavior, yet the incomplete deglycosylation prevented formulating clear conclusions, as the standard binding response could not be unequivocally attributed to a single isoform. Secondly, to investigate the possibility of aggregation, PDC-109 was flown across a chip where itself was immobilized. The resulting sensorgrams displayed the previously observed biphasic shape, suggesting that interaction of PDC-109 with itself is the cause of the anomalous biphasic sensorgrams, and that the glycosylated version plays a significant role in this self-recognition, possibly by virtue of the single sialic acid residue present.
Evidence from the literature points out that PDC-109 is naturally produced as a mixture of several protein forms [28] and aggregation states [29] that play an important role in modulating its interaction with other biomolecules. This would allow concluding that, whichever of the situations proposed in Figure 6 applies -including a combination thereof-, PDC-109 acts as a lectin-like molecule albeit with quite unique features that, though not abolishing its ability to recognize carbohydrates, give rise to unusual SPR sensorgrams.
Despite the above-discussed limitations, we were able to determine the SPR binding parameters of PDC-109. Affinity constants ( Table 4) indicated weak to moderate binding for carbohydrates, at least if compared to the considerably higher affinity (K A~1 0 7 M −1 ) found for PC membranes [26]. This would suggest that PDC-109 binding to carbohydrates in oviductal epithelial cells must likely involve some sort of multivalency in order to efficiently retain the sperm in the oviductal reservoir.  From the above data it can also be seen that PDC-109 appears to have the highest binding affinity for Fuc-α1,4-GlcNAc among all glycotopes in this study. This is in agreement with earlier findings which propose that bull sperm binds to an oligosaccharide ligand on the oviductal epithelium that resembles Lewis a trisaccharide (Galβ1-3[Fucα1-4]GlcNAc) [30]. For PDC-109 vs. Fuc-α1,4-GlcNAc, the higher affinity was a consequence of a lower dissociation rate constant, consistent with a privileged interaction with this glycosidic configuration. The observation that PDC-109 displays favored interaction with the Neu5Ac-α2,6-Gal-β1,4-GlcNAc epitope may account for the observed self-recognition. Part of PDC-109 bears an O-linked trisaccharide consistent with this mass [10] and, even though this aspect has never been fully elucidated, it could well be that the Sia residue, when exposed in an α-2,6-configuration, facilitates such binding. Finally, the absence of significant binding with the mannobiose epitopes is in contrast to what was reported by Amari and coworkers. However, they employed oligomannose type pentassacharides and the complex architecture of this oligosaccharide may well account for the difference in binding characteristics.

Carbohydrate Recognition Domain of PDC-109 by CREDEX-MS
Carbohydrate Recognition Domain Excision (CREDEX) mass spectrometry is a technology to map the interacting domains of sugar-lectin complexes, providing structural information by limited proteolysis (epitope excision/extraction) coupled with MS [31].
In order to test the complementarity of CREDEX-MS with the SPR-based glycoprobe approach for PDC-109, and to define the carbohydrate-binding site of this seminal plasma protein, glycotopes that in the SPR study had displayed higher affinities, i.e., all fucosylated epitopes (Fucα-1,3-GlcNAc, Fucα-1,4-GlcNAc and Fucα-1,6-GlcNAc), were immobilized onto DVS-Sepharose. To guarantee data reliability, prior to the excision experiments with PDC-109, columns were validated for their Fuc-recognition ability by binding experiments with the Fuc-specific UEA-I lectin followed by SDS-PAGE detection. Excision experiments were also carried out and the CRD of UEA-I was successfully delimited (data not shown). The affinity-bound peptides eluted and identified by MS were in agreement with the previously reported structural basis of UEA-I carbohydrate specificity [21].
Once the functionality of the columns was confirmed, excision experiments with PDC-109 were run. Briefly, PDC-109 was incubated with the Fuc-α1, (  In the Fuc-α1,4-GlcNAc excision experiment (Figure 7), five out of seven peptides (colored) identified in the elution fraction correlated directly with the predicted ones. In addition, two other peptides (highlighted in gray) observed in the elution fraction corresponded to miscleaved fragments. The only unique peptide in the excision experiments with all fucosylated glycoprobes was the disulfide-bridged fragment [69-78, 86-102], as well as its reduced components. Hence it is likely that these two peptides of the 2Fn2 domain, either disulfide-bound or individually, constitute the carbohydrate binding domain of PDC-109.

Discussion
Our SPR-based approach, using an extensive collection of well-defined glycopeptide probes exposing sugar epitopes in native-like form, has not only served to determine kinetic and affinity constants of these epitopes with specific plant lectins consistent with previously published results, but also shown its applicability to PDC-109, a fertilization-relevant protein in bovines, and possibly also humans. While the binding affinity of PDC-109 for heparin [32] or PC [26,33], as well as its chaperone-like activity [34] have received considerable attention, our analysis of the lectin-like activity of PDC-109 is, to the best of our knowledge, the first quantitative study of the carbohydrate binding behavior of PDC-109.
Although several glycoconjugates have been synthesized by other laboratories using also N[Me]-Aoa-peptide versions as mimics of natural glycoproteins, few of them have been used as probes for interaction studies [35,36], and not always with conclusive results. Thus, this work represents a further step to establish our SPR-approach as a reliable alternative in carbohydrate-lectin interaction studies. The complex nature of the natural PDC-109 preparation has exposed the difficulties in biomolecular interaction analysis when not dealing with well-defined and homogenous samples. Yet these challenges have also helped our understanding of biological phenomena where heterogeneous mixtures are common. The admitted limitations of the SPR approach in the particular case of PDC-109 have been satisfactorily overcome by CREDEX-MS, again showing both techniques as a suitable duet to decipher protein-carbohydrate interactions not only at the demonstration level [16] (i.e., with "academic" lectins) but with "real", biologically relevant samples.
The PDC-109 lectin binding profile has been completed, and its specificity assessed against various validated glycosidic epitopes. Specifically, the interaction with common mammal epitopes has been qualitatively and quantitatively characterized by SPR, with a larger affinity found for the Fuc-α1,4-GlcNAc disaccharide, a result in agreement with previous studies which proposed that, during sperm reservoir formation, bovine sperm binds to an oligosaccharide ligand of the oviductal epithelium similar to the Lewis a epitope (Galβ1-3[Fucα1-4]GlcNAc) [15,30]. On a different note, it is worth mentioning that unusual PDC-109 SPR sensorgrams posing difficulties for kinetic adjustment most likely reflected intrinsic PDC-109 properties, confirming that this protein is naturally a mixture of various isoforms and aggregation states that play an important role in modulating the interaction with other biomolecules [28,29,[37][38][39]. In this context, self-aggregation of PDC-109 could be driven by the Neu5Ac-Hex-HexNAc trisaccharide found in roughly 50% of its isoforms in circulation. The lack of quantitative glycosylation might serve as a limiting factor in PDC-109-induced sperm coating, i.e., regulating the layer density around the spermatozoon.
The role of PDC-109 in the formation of the sperm reservoir, through its lectin specificity for carbohydrates containing Fuc, was corroborated by SPR and fine-tuned to Fuc-α1,(3,4,6)-GlcNAc by the orthogonal CREDEX-MS technique. In all experiments, a common peptide fragment has been identified that is very likely to be part of the PDC-109 CRD.
In conclusion, our study sheds new light on the interaction profile of the PDC-109 seminal plasma protein, and increases understanding of the molecular events involved in the capacitation and fecundation processes. Also importantly, such an understanding can make inroads into the design and development of novel fertility-related drugs.
N-deglycosylation of PDC-109 was done using PNGase F (1 µL; 250 units/0.25 mL) in phosphate buffer, 50 mM pH 7.3, 16 h, 37 • C. After filtration (10 kDa filter) to eliminate buffer salts, the efficiency of the procedure and the resulting compound/s was checked by SDS-PAGE and MALDI-TOF MS.

Peptide Module Synthesis
The GFKKG-amide substrate was made by Fmoc solid-phase synthesis on a Rink amide MBHA resin (0.70 mmol/g) at a 0.1 mmol scale in an Applied Biosystems 430A automated synthesizer.

Boc-N[Me]-Aoa-OH coupling
Prior to coupling to the GFKKG-amide resin, Boc-methylaminooxyacetic acid·DCHA salt (500 mg Boc-N[Me]-Aoa-OH/DCHA) was shaken with 0.1 M HCl and EtOAc (50 mL each) to release the free carboxyl form into the organic phase (3 extractions). After drying with MgSO 4 . EtOAc was evaporated and the residue used in manual coupling (3 equivalents each of Boc-amino acid and HBTU, 6 equivalents of DIEA) to the peptide-resin for 1 h, room temperature (r.t.), in DMF.

Chemoselective Ligation of Carbohydrates
Conjugation between N[Me]-O-Aoa-GFKKG-amide peptide and oligosaccharides was done at 20 and 25 mM, respectively in 0.1 M NaOAc for 72 h at 37 • C and pH 3.5 for NAc-hexosamines or pH 4.6 for hexoses. Conjugations involving Fuc-containing disaccharides (Fuc-α1,(3,4,6)-GlcNAc) were performed both in the presence and the absence of 100 mM aniline as a nucleophile catalyst, to explore improvements in conjugation yields.

MALDI-TOF Mass Spectrometry (MS)
Peptides and glycopeptides were dissolved in water and mixed with the corresponding matrix solution (1:1 v/v) and 1 µL of the mixture was applied to the MALDI target and allowed to dry at room temperature.

Amino Acid Analysis (AAA)
As an alternative to UV quantification, AAA was used for both N[Me]-O-Aoa-GFKKG-amide and Ac-GFKKG-amide and glycopeptides (Table 5). Prior to AAA, ca. 10 nmol (10 µg) peptide or glycopeptide sample were dissolved in 200 µL of 12 M HCl and 200 µL of 0.1 M norleucine in water (20 nmol, internal standard) in a Pyrex glass tube. The tube was stoppered and hydrolysis was performed for 24 h at 110 • C. HCl was removed in a rotary evaporator prior to analysis of the hydrolysate by Gas Chromatography-Mass Spectrometry (Agilent Technologies, Barcelona, Spain). Amino acids (in both samples and calibration standards) were converted to tert-butyl dimethylsilyl (TBDMS) derivatives by treatment with 50 µL of N-tert-butyldimethylsilyl-N-methyltrifluoroacetamide (MTBSTFA) and 50 µL ACN for 4 h at 100 • C. Derivatized samples were next analyzed by GC-MS in an Agilent 6890N Network GC system fitted with an Agilent 5973N mass selective detector. A Phenomenex Zebron ZB-5 cross-linked 5% phenyl polymethyl siloxane capillary column (15 m × 0.25 mm i.d., 0.25 µm film thickness) was used. Oven temperature was initially held at 100 • C then raised to 310 • C at 25 • C/min. The interface and ion source temperatures were 150 and 230 • C, respectively. Column gas flow was 20 mL/min. The MS was operated in full scan mode (50-650 amu; scan rate 2.46 scans/s) with at least two characteristic mass fragments being recorded for each amino acid. Four replicates per sample were run. Quantification was performed from interpolation of calibration standard curves.

Surface Plasmon Resonance (SPR)
All SPR experiments were performed in a BIAcore 3000 instrument (GE Healthcare, Uppsala, Sweden) using carboxymethyl-functionalized CM5 or C1 sensor chips and HBS-P (0.01 M HEPES pH 7.4; 0.15 M NaCl; 0.005% v/v surfactant P20) as running buffer, supplemented when necessary with CaCl 2 (5 mM) and MnCl 2 (1 mM) or other competitive reagents/adjuvants such as bovine serum albumin (BSA), carboxymethyldextran (CMD) or polivinylpirrolidone (PVP). SPR sensorgram analysis and curve fitting were done with the BIAevaluation 4.1.1 sofware package (GE Healthcare, Uppsala, Sweden). In all experiments, specific binding responses were obtained by subtracting from each channel the reference channel response. Moreover, in most cases a double referencing was applied (i.e., reference Fc1 channel plus an internal reference standard, namely buffer injection subtraction).

Immobilization of Glycoprobes
Glycoprobes immobilized on each sensor chip and their immobilization levels are given on Table 6. On each chip, a reference surface (preferentially, first flow cell) was created with a non-specific probe immobilized (N[Me]-O-Aoa-GFKKG-amide or Ac-GFKKG-amide peptides) to subtract the non-specific lectin binding to the dextran surface. For all immobilizations, the carboxyl functionalities of the CM5 (or C1) sensor chip surfaces were activated by injecting a mixture of 0.05 M NHS and 0.2 M EDC (1:1 v/v; 50 µL) at 5 µL/min. Then, glycopeptide solutions were passed at 1-2 mg/mL in 10 mM NaOAc for approximately 12 min over the activated surface. For immobilization, pH 4 or pH 6 conditions were used, as required. In the specific case of PDC-109, protein solution was passed at 0.05 µg/µL in 10 mM NaOAc pH 4 over the activated surface. Afterwards, unreacted groups on all chip surfaces were blocked with ethanolamine-HCl (1 M; pH 8.5; 60 µL) or H 2 O (60 µL). The difference, in resonance units (RU), between readings after surface activation and final response was converted into amount of immobilized glycopeptide (0.981 RU~1 pg/mm 2 for glycopeptides, 1 RU = 1 pg/mm 2 for proteins; see legend Table 6).

Binding and Kinetic Experiments
All kinetic measurements were carried out by running a manually programmed sequence of binding experiments using several protein range concentrations and two flow rates. Regeneration of the sensor surface was accomplished in a specific-manner by injecting the complementary carbohydrate (10 mM lac, 100-500 mM αMe-Man, 0.5 M αMe-Fuc or L-Fuc, 0.5 M GlcNAc, 0.5 M Glc, and 10 mM Sia). Alternatively, other procedures such as glycoprotein solutions (e.g., fetuin at 2-5 g/L), low pH-buffers (e.g., 10 mM glycine pH 1.5-2.5), high salt (e.g., 1 M NaCl), high pH or specific chemicals were also used to break the interaction.
For the evaluation of the sensor chips, binding experiments and kinetic studies were carried out with the various glycoprobes and specific well-characterized lectins (ECA, Con A, LTA, UEA-I, MAA, and SNA). Several concentrations of each lectin were prepared by serial 2/3-fold dilutions of the most concentrated sample in running buffer supplemented with CaCl 2 (5 mM) and MnCl 2 (1 mM). Thus, concentrations varied from 66 nM to 2.5 µM for ECA; from 2.5 to 80 nM for Con A; from 31.25 nM to 1 µM for LTA and UEA-I; from 250 nM to 1.9 µM for MAA and from 74 to 563 nM for SNA. Analyses were performed at 20 µL/min and kinetic studies at 10 and 40 µL/min. After lectin injection, the sample solution was replaced by running buffer and the dissociation phase was monitored for 3-6 min. Sensor surfaces were regenerated with a 10-50 µL-injection of 10 mM lac for ECA; 500 mM αMe-Man for Con A; 0.5 M L-Fuc for LTA and UEA-I and 5 g/L fetuin for MAA and SNA.
For determination of PDC-109 kinetic parameters, several concentrations in the 6.25-200 µM range were prepared by serial 2/3-fold dilutions of the most concentrated sample in HBS-P running buffer as a standard condition. Other buffer conditions (i.e., HBS-P buffer supplemented with CaCl 2 (5 mM) and MnCl 2 (1 mM), or adding other reagents such as BSA 0.5% carrier protein, CMD 0.5% and PVP 0.5% competitors) were tested to improve SPR sensorgrams and subsequent kinetic fittings. Binding experiments were performed at 25 • C and 30 µL/min flow rate. After PDC-109 injection (3-min pulse), the sample solution was replaced by running buffer and the dissociation phase was monitored for 6 min. Sensor surface was regenerated with a 60 µL-injection of specific carbohydrate (10 mM lac, 500 mM αMe-Man, 0.5 M αMe-Fuc, and 10 mM Sia) followed by a 60 µL-injection of NaCl 1 M.
For PDC-109 binding behavior studies, CM5 surface vs. C1 surface without dextran matrix were tested to assess the influence of this structural element. Binding experiments were performed by injecting both PDC-109 untreated and after deglycosylation at 50 µM, 25 • C, and 20 µL/min flow. After protein injection, the sample solution was replaced by running buffer and the dissociation phase was monitored for 4 min. Sensor surfaces were regenerated with a 30 µL-injection of 1 M NaCl. Alternatively, PDC-109 aggregation tests were monitored by SPR. Briefly, protein solution of 50 µM PDC-109 was passed over immobilized PDC-109 and reference flow cells at 20 µL/min. After dissociation phase, surface regeneration was accomplished by injection of 30 µL of 1 M NaCl. Two replicates were performed for each injection in all binding behavior studies. For disaccharide immobilization, 5 mg of sugar (Fuc-α1,(3,4,6)-GlcNAc for UEA-I/PDC-109) was dissolved in 50 µL of 0.5 M K 2 CO 3 (pH 11), and the solution was incubated in a microcolumn with 50 µg of dry DVS-activated Sepharose, overnight at r.t. under continuous shaking (800 rpm). Then, the microcolumn was washed sequentially with 50 mM NH 4 OAc (pH 4) and 50 mM NH 4 HCO 3 (pH 8). Finally, it was equilibrated with binding buffer (10 mM HEPES, 25 mM NaCl, 5 mM CaCl 2 and 1 mM MnCl 2 ) and stored at 4 • C.
Before performing excision experiments, columns were tested for correct binding. Briefly, 20 µg of UEA-I or PDC-109 were added onto the Fuc-α1,(3,4,6)-GlcNAc-Sepharose columns, and were incubated in binding buffer for 24 h at 37 • C. Unbound materials were removed by extensive washing with running buffer. Bound lectins were eluted with 60% ACN (0.1% TFA), and the protein content of each fractions (flow through; wash and elution) was analyzed by 1D-SDS-PAGE electrophoresis.
For excision experiments, 20 µg of UEA-I or PDC-109 in running buffer (75 µL) were loaded onto the Fuc-α1,4-GlcNAc-Sepharose microcolumn and incubated for 24 h at 37 • C. Unbound lectin was removed by washing with running buffer until no signal could be detected by MS. Then, the sugar-lectin complex was digested overnight with trypsin (enzyme:substrate ratios as above) in 25 mM NH 4 HCO 3 , pH 8.5 at 37 • C. After 15 h, the flow through containing digestion products was removed and the column was washed. After washes with running buffer, specific-bound peptides were eluted with 60 µL of 60% ACN (0.1% TFA), concentrated and lyophilized. Prior to MALDI-TOF analysis, each sample was desalted by means of micro-column purification (Poros R2) or by ZipTip ® Pipette Tips.

Conflicts of Interest:
The authors declare no conflict of interest.