Glycoprotein analysis using lectin microcolumns and capillary electrophoresis: characterization of alpha₁-acid glycoprotein by combined separation methods

Abstract

Separations based on combinations of 2.1 mm I.D. high-performance affinity microcolumns and capillary electrophoresis were developed and used to characterize the glycoforms of an intact glycoprotein. Human alpha₁-acid glycoprotein (AGP) was used as a model analyte due to its heterogeneous glycosylation resulting from variations in its degree of branching, fucosylation, and number of sialic acids. Three separation formats were examined based on microcolumns that contained the lectins concanavalin A (Con A) or Aleuria aurantia lectin (AAL). These microcolumns were used with one another or in combination with capillary electrophoresis. N-Glycan analysis of the non-retained and retained AGP fractions was carried out by using PNGase F digestion and nanoflow electrospray ionization mass spectrometry. Con A microcolumns were found to selectively enrich AGP that contained bi-antennary N-glycans, while AAL microcolumns retained AGP with fucose-containing N-glycans. Results from these separation methods indicated that fucosylation of the N-linked glycans was more abundant when a high degree of branching was present in AGP. Sialic acid residues were more abundant when higher degrees of branching and more fucose residues were present in AGP. The separation and analysis methods that were developed could be used with relatively small amounts of AGP and can be adapted for use with other intact glycoproteins.

1. Introduction

Alpha₁-acid glycoprotein (AGP) is an acute phase glycoprotein that functions as a serum transport protein for neutral and basic drugs [1]. This protein consists of a single peptide chain that contains 183 amino acid residues and has a molecular mass of 4.1–4.3 × 10⁴ g/mol (i.e., 41–43 kDa), with two generic variants that differ by 22 amino acid residues [2]. AGP and its glycoforms have been of interest as biomarkers for various disease states [3–7]. The normal human serum concentration of AGP spans from 0.5 to 1.0 mg/mL (i.e., 12 to 24 μM); however, this level can increase by up to 10-fold in some clinical conditions [3]. A change in the degree of branching for the glycans of AGP has been found in patients with infection and systemic lupus erythematosus [4]; an increase in α1–3 fucosylation has been observed in pancreatic cancer; and hypersialylation has been reported in patients with ovarian cancer or lymphoma [5–7].

AGP has five glycosylation sites to which various N-linked complex-type glycans are attached (see examples in Figure 1) [2]. These glycans account for 45% (mass/mass) of the molecular mass of AGP [2]. Heterogeneous glycoform compositions of AGP can result due to glycan variations in the degrees of branching, amount of fucose residues, or extent of sialylation [2,3]. A large number of variants based on changes in the glycans of AGP have been reported [8]. These variants typically appear as 9–11 major glycoform bands when analyzed by methods such as capillary electrophoresis (CE) [6,9–11].

Figure 1.

Basic structure of AGP, including examples of N-linked glycan chains that may occur at the five glycosylation sites of AGP [2].

Previous separation methods for the analysis of intact AGP have mainly relied on single techniques. For instance, CE has been used for the analysis of AGP based on its electrophoretic mobility [6,9–16] or isoelectric point [17]. Affinity separations based on lectins have also been used to examine AGP [4,18–27]. The lectin concanavalin A (Con A) from Canavalia ensiformis has been used to fractionate AGP based on the presence of bi-antennary N-linked glycans vs. tri- and tetra-antennary N-linked glycans [4,18–27]. Aleuria aurantia lectin (AAL) has been used to separate AGP with α1–6 and α1–3 linked fucose residues from AGP without these residues [20,21,23,25,28,29]. One previous report employed a low-performance Con A affinity column that was coupled off-line with CE; however, this column was used for mainly preparative purposes and provided only limited quantitative information on AGP glycoforms [30].

In this report, high-performance Con A and AAL affinity microcolumns will be combined with each other or with CE to provide separations based on multiple methods for the characterization of AGP glycoforms. The term “affinity microcolumn”, as used here, refers to a column that has a volume in the low- to mid-microliter range and that contains an immobilized binding agent for the selective capture or isolation of a given target [31,32]. This work will make use of 2.1 mm I.D. microcolumns that have recently been prepared with Con A or AAL and characterized for their overall binding to AGP [32]. In this current study, these microcolumns will be evaluated, both alone and combined, with regards to their binding specificity for AGP glycoforms through use of nanoflow electrospray ionization mass spectrometry (nESI-MS). The combined separation methods will be used to examine the degree of branching, fucosylation, and amount of sialic acid (i.e., N-acetylneuraminic acid) for the N-glycans on intact AGP. The advantages and potential limitations of these separation methods will also be considered, as pertaining to use of these techniques in future work with other glycoproteins.

2. Experimental section

2.1. Materials

The following chemicals and materials were from Sigma-Aldrich (St Louis, MO, USA): AGP (from pooled human plasma, product G9885, ≥ 99% pure), L-fucose (> 99%), Con A (product C7275, type V, highly purified lyophilized powder), poly(ethylene oxide) (PEO; viscosity-average molecular mass, 8,000 kDa), Brij 35 (product 858366, number-average molecular mass, 1.198 kDa), Amicon Ultra centrifugal filter units (Ultra-4, 30 kDa cutoff), PNGase F (from Elizabethkingia meningoseptica, product P7367, ≥ 95%), iodoacetamide (IAA, ≥ 99%) and dithiothreitol (DTT, ≥ 98%). AAL (product L-1390, homogeneous by SDS-PAGE) was purchased from Vector Laboratories (Burlingame, CA, USA). Reagents for a micro bicinchoninic acid (BCA) protein assay were acquired from Thermo Fisher Scientific (Waltham, MA, USA). The NuTip HyperCarb PGC tips (1–10 μL) were from Glygen (Columbia, MD, USA), and the Nucleosil silica (300 Å pore size, 5 μm particle diameter) was from Macherey-Nagel (Duren, Germany). All aqueous solutions and samples were prepared using water purified by a Milli-Q Advantage A10 system (EMD Millipore, Billerica, MA, USA) and were filtered through 0.20 μm GNWP nylon membranes from Millipore.

2.2. Apparatus

The chromatographic studies were performed with a Jasco HPLC system (Tokyo, Japan) that contained two PU-2080 isocratic pumps, an AS-2057 autosampler, a CO-2067 column oven, and a UV-2075 UV detector. A two position/six port valve (MX Series II, IDEX Health & Science LLC, Rohnert Park, CA, USA) was used to switch between the isocratic pumps for sample loading and elution. The HPLC system was controlled with ChromNav software from Jasco. The chromatograms were analyzed with Peakfit 4.12 (Jandel Scientific Software, San Rafael, CA, USA).

CE was performed with a P/ACE MDQ instrument that was equipped with a UV detector and controlled with 32 Karat 7.0 software (Beckman Instruments, Fullerton, CA, USA). This system used 50 μm I.D. fused silica capillaries from Polymicro Technologies (Phoenix, AZ, USA). Electropherograms were analyzed with Peakfit 4.12 software.

The analysis of glycans was carried out with a Synapt G2-S HDMS quadrupole time-of-flight hybrid mass spectrometer (Q-TOF-MS; Waters, Manchester, UK). The sample solutions were placed in a homemade borosilicate emitter for introduction by nESI in the static mode. The emitters were created from 1.5–1.8 mm I.D. × 100 mm Corning Pyrex melting point capillaries (Corning, NY, USA) by using a vertical micropipette puller (David Kopf Instruments, Tujunga, CA, USA). The emitter was fitted to a home-built static nESI stage at the inlet of the MS instrument [33]. A platinum wire from the stage was placed inside the emitter so that it was in contact with the sample solution. This wire was used to deliver a capillary potential for nESI. The Q-TOF-MS system was controlled by MassLynx v4.1 software from Waters.

2.3. Preparation and use of lectin microcolumns

Nucleosil silica (pore size 300 Å, particle size 5 μm) was converted into a diol form and used to immobilize Con A or AAL through reductive amination (i.e., the Schiff base method), as described previously [32]. The final immobilized lectin content for these supports was determined in triplicate by using a micro BCA protein assay [32]. For this assay, a small portion (i.e., around 4 mg) of each lectin support was washed thoroughly with water to remove the Tris buffer and then dried at 50 °C prior to measurement. A control support was used as the blank during this measurement. The activities of these supports were previously verified by using fluorescent sugars that had known binding properties for these lectins (i.e., p-nitrophenyl α-D-mannopyranoside for Con A and p-nitrophenyl α-L-fucopyranoside for AAL, respectively) [32]. Each lectin support was downward slurry packed into a 2.1 mm I.D. × 50 mm stainless steel column (product 25119, Restek, Bellefonte, PA, USA) at a pressure of 4000 psi (27.6 MPa). The packing solution for the Con A support was 10 mM Tris-HCl (pH 7.4, 10 mM Tris base titrated by HCl) buffer that contained 0.15 M NaCl, 0.5 mM CaCl₂, 0.5 mM MgCl₂ and 0.5 mM MnCl₂ [32]. The packing solution for the AAL support was 10 mM Tris-HCl (pH 7.4, 10 mM Tris base titrated by HCl) buffer that contained 0.15 M NaCl [32]. These microcolumns, and remaining supports, were stored in their corresponding packing buffers, with no additives included, at 4 °C until use.

The separation of AGP glycoforms on Con A microcolumns was performed under isocratic conditions, as described in a recent study [32]. The injected sample volume for AGP was 20 μL and the AGP concentration was 10 mg/mL. The mobile phase was 10 mM Tris-HCl (pH 7.4, 10 mM Tris base titrated by HCl) containing 0.15 M NaCl, 0.5 mM CaCl₂, 0.5 mM MgCl₂ and 0.5 mM MnCl₂. The AGP samples were prepared in this mobile phase. The flow rate was 50 μL/min and the column temperature was 50 °C. The elution of AGP glycoforms was monitored at 280 nm on both the Con A microcolumns and AAL microcolumns (see next paragraph). All samples in this study were injected in triplicate.

The AAL microcolumns were used to separate AGP glycoforms by using step elution that employed L-fucose as a competing agent [32]. This separation was carried out at 0.75 mL/min and 50 °C. A 20 μL injection volume was used along with 1 mg/mL samples of AGP that were prepared in the application buffer. The application buffer consisted of 10 mM Tris-HCl (pH 7.4, 10 mM Tris base titrated by HCl) that contained 0.15 M NaCl. L-Fucose was added at a concentration of 2 mM to this buffer to elute the retained AGP glycoforms (Note: no further improvement in the elution profile or amount of released AGP has been seen when the fucose concentration, flow rate, or temperature is varied slightly from the given optimized conditions) [32]. This elution step was carried out at 2.7 min after sample injection onto the AAL microcolumn, with the application buffer then being passed through the microcolumn at 5.3 min for column regeneration.

When a Con A microcolumn was combined with an AAL microcolumn, both the non-retained and retained fraction of AGP were collected from the Con A column (e.g., 0.75 mL for the non-retained fraction and 1.75 mL for the retained fraction, with each fraction after two runs containing ~200 μg AGP under the maximum loading conditions used in this study). These fractions were then exchanged into the application buffer for the AAL microcolumn by using a 30 kDa ultrafiltration filter that was spun at 3000 × g (i.e., 4400 rpm with the given equipment) for 7 min at room temperature. The AGP samples were concentrated during this step to around 200 μL (i.e., around 1.0 mg/mL AGP for a sample that initially contained ~200 μg). These fractions were then applied to an AAL microcolumn using an injection volume of 20 μL for further separation and analysis. A 1.0 mg/mL standard containing non-fractionated AGP was applied to the same AAL microcolumn as a reference.

2.4. Glycoform analysis of AGP by CE

Fractions of AGP glycoforms were collected in 0.25 to 1.5 mL portions upon their elution from the lectin microcolumns. For a Con A microcolumn, fractions were collected between 2.5–7.5 min and 7.5–30 min. For an AAL microcolumn, fractions were collected between 0–2 min and 3–5 min. The amount of AGP in each fraction was estimated by using the product of the total AGP loaded onto the system and relative areas for the observed non-retained and retained peaks. The fractions were desalted by combining them with water three times in a 30 kDa centrifugal filter, followed by centrifugation at 3000 × g (i.e., 4400 rpm) for 7 min at room temperature after each addition of water. The remaining AGP solution was diluted with water to a final concentration of 1 μg/mL and stored at 4 °C until use. A previously-developed CE technique with electrophoretic injection was used to allow the analysis of AGP at this concentration level [11].

The intact AGP glycoforms in the collected fractions were examined by using CE with a neutral coated capillary (i.e., based on PEO) and electrophoretic injection, according to a previous method [11]. Electroosmotic flow due to the coated capillary was negligible under the conditions used for CE in this study, as demonstrated previously [10,11]. The samples were injected at −5 kV for 5 min and analyzed with the coated capillary at 25 °C and in the presence of a running buffer that consisted of 20 mM acetate buffer (pH 4.2, 20 mM sodium acetate plus acetic acid, titrated with HCl or NaOH, using 0.036 g sodium acetate and 89 μL or 0.094 g glacial acetic acid per 100 mL buffer) containing 0.05% (mass/volume) PEO and 0.1% (mass/volume) Brij 35. The separation was performed in the reversed-polarity mode of the separation voltage (i.e., the negatively-charged cathode was located at the injection end of the capillary). This separation was carried out on the coated capillary (50 μm I.D., 360 μm O.D., 60.2 cm total length, 50 cm length to detector) and at an applied voltage of −30 kV. The detection of AGP glycoforms was carried out at 200 nm.

2.5. Release and purification of glycans from AGP

A portion of each collected fraction of AGP (i.e., about 0.3 mg) was digested by PNGase F for analysis by MS. These fractions were first desalted and placed into water, as described in the previous section, with the remaining AGP solution being dried by using a SpeedVac (Savant Instruments, Farmingdale, NY, USA) without heating. The dried product was stored at −80 °C until its use in digestion and MS analysis.

Digestion of the AGP samples with PNGase F was performed according to a previous procedure with minor modifications [34]. In this method, a 0.3 mg portion of an AGP fraction was dissolved in 30 μL water and combined with 25 μL of 25 mM sodium phosphate buffer (Ph 7.5–8.0). The AGP in this sample was reduced by adding 2.5 μL of 200 mM DTT and incubating the mixture at 60 °C for 45 min. The AGP sample was then alkylated by adding 10 μL of 200 mM IAA, with this mixture being reacted in the dark for 1 h at room temperature. A 100 μL portion of 25 mM sodium phosphate buffer (pH 7.5–8.0) was added, followed by boiling for 10 min to denature the AGP. After allowing the sample to cool to room temperature, 15 μL of PNGase F (7.5 units) was added, and the mixture was incubated at 37 °C for 18–22 h. The digestion was stopped by boiling the mixture for 3 min. A 800 μL portion of cold ethanol was added to the mixture to induce protein precipitation. The mixture was then chilled for 30 min at −20 °C, followed by centrifugation using a Micro V Microcentrifuge (Thermo Fisher) for 10 min at maximum speed (10,000 rpm, or ~9400 × g). The supernatant was dried to a total volume of approximately 20 μL by placing it into a SpeedVac. The remaining glycans were resuspended in water to give a total solution volume of 200 μL.

The glycan solution was purified prior to MS analysis by using graphitized carbon tips [34]. Each tip was first cleaned with 100 μL of 80% (v/v) acetonitrile containing 0.1% (v/v) trifluoroacetic acid (TFA). The tip was then washed with 100 μL water. A 20 μL portion of a glycan solution made from digested AGP was loaded onto the tip, followed by washing with 100 μL water. The neutral and sialylated glycans were eluted from the graphitized carbon tips by applying 20 μL of 25% (v/v) acetonitrile containing 0.05% TFA.

2.6. Glycan analysis by MS and MS/MS

Each glycan solution that was prepared from digested AGP was loaded into a borosilicate emitter and placed onto the platinum wire of a home-built nESI stage [33]. A capillary potential (1.30 – 1.40 kV) was applied to the prepared solution and optimized for each sample and emitter. The source temperature was held at 80 °C. The gas used for nebulization was nitrogen. The sampling cone and source offset voltages were each set between 0–10 V. Both the MS and tandem mass spectrometry (MS/MS) experiments were carried out in a positive ion mode. The MS/MS studies were conducted by performing collision-induced dissociation (CID) on a quadrupole-selected m/z value in the “trap” region of the instrument (i.e., a stacked ring ion guide). The selected ions were accelerated by using a set potential difference, ranging from ΔU = 7.5–15 V, and collided with argon. For CID analysis, the trap gas flow was set at 1.2 mL/min to keep the trap cell pressure at approximately 5.0 × 10⁻³ mbar. All MS and MS/MS spectra were acquired using MassLynx 4.1 and further visualized by using IGOR Pro 7 (WaveMetrics, Lake Oswego, OR, USA).

3. Results and discussion

3.1. Retention and elution properties of Con A and AAL microcolumns

The lectin supports used in this report contained 88 (± 6) mg Con A per gram of silica or 29.4 (± 1.7) mg AAL per gram of silica [32]. This corresponded to a lectin content for a 2.1 mm I.D. × 50 mm microcolumn of 6.9 mg Con A (65 nmol) or 2.3 mg AAL (32 nmol). As described previously, linear elution conditions were obtained with this type of Con A microcolumn up to a total sample load of > 200 μg AGP, and the AAL microcolumn provided linear elution conditions up to a load of 100 μg AGP. The Con A and AAL microcolumns were stable for at least seven months (> 15 months for Con A) and more than 80–90 sample injections or application cycles [32].

It is known Con A has weak-to-moderate binding towards bi-antennary N-glycans, while AAL binds to α1–6, α1–2, α1–3 and α1–4 fucose residues (Note: α1–6 and α1–3 linked fucose residues are typically present in AGP) [2,35,36]. Some typical chromatograms obtained with the Con A and AAL microcolumns during the separation of AGP glycoforms are shown in Figure 2; these chromatograms are based on application and elution conditions that have been optimized for use with these supports and microcolumns, such as related to the resolution of retained vs non-retained AGP fractions [32]. Con A microcolumns could separate the non-retained and retained fractions of AGP at a resolution of 0.93 within 20 min when using isocratic elution conditions at 50 μL/min and 50 °C (see Ref. [32] for more details on the effects of temperature or flow rate on this separation). The non-retained and strongly-retained fractions of AGP were separated at a baseline resolution within 6 min on an AAL microcolumn at 0.75 mL/min and 50 °C when using 2 mM L-fucose as a competing agent for step elution [32].

Figure 2.

Chromatograms obtained for AGP using (a) a Con A microcolumn followed off-line by (b) an AAL microcolumn. The results in (b) are for the AGP retained fraction from the Con A microcolumn; a similar chromatogram was obtained for the non-retained fraction from this microcolumn (see Supplementary Material). The separation in (a) was obtained under isocratic conditions at 50 °C and 50 μL/min. The separation in (b) was carried out at 50 °C and 0.75 mL/min using a step elution scheme, in which the elution buffer (containing 2 mM L-fucose) was applied to the microcolumn at 2.7 min. Other conditions are provided in Section 2.

3.2. Analysis of selectivity of Con A microcolumn for AGP glycoforms

The binding specificity of the Con A support and microcolumns for AGP was examined by using nESI-MS to look at the glycans that were released from the retained and non-retained AGP fractions after treatment with PNGase F. The mass spectra that were acquired from these samples are provided in Figure 3(a).

Figure 3.

Mass spectra for glycans released from AGP fractions that were retained (top) or non-retained (bottom) by (a) a Con A microcolumn or (b) an AAL microcolumn. Abbreviations: N, N-acetylhexosamine; H, hexose; F, fucose; and S, sialic acid (i.e., N-acetylneuraminic acid), where the subscripts appearing after these letters indicate the number of each monosaccharide unit in the glycan. Protonated species are the predominant types of ions that are seen in this type of analysis; however, some sodiated, potassiated, and hybrid species (e.g., protonated and sodiated ions) have been observed for other types of glycan samples even after these have been treated and purified as described in Section 2.5. The identity of the base peak for each major type of glycan is indicated in these mass spectra; most of the remaining peaks represent isotopic envelopes of the main, identified peaks or the same glycan species with different charge carriers, although some trace amounts of other non-identified peaks are also present.

For the Con A-retained fraction of AGP, the observed glycan structures ranged in size from bi-antennary to tri- and tetra-antennary chains. The base peak was a bi-antennary N-glycan with two sialic acids and no fucose residues; this peak was designated as [N₄H₅S₂]²⁺, in which N is N-acetylhexosamine, H is hexose and S is sialic acid. The composition of this base peak was confirmed by MS/MS (see Supplementary Material). All other peaks that were identified represented tri- or tetra-antennary N-glycans from the remaining glycosylation sites and that were in addition to the bi-antennary N-glycans. The total relative abundance for the identified peaks due to tri-antennary glycans in this AGP sample was 123% (i.e., when using their combined signals compared to that of the base peak), and the total relative abundance for the assigned peaks due to tetra-antennary glycans was 62%.

Glycans that were released from the non-retained fraction of AGP on a Con A microcolumn had a bi-antennary peak for [N₄H₅S₂]²⁺ with a relative abundance of only 13% compared to the base peak (Note: no change in relative abundance for this ion was seen in the non-retained vs retained fractions from the AAL microcolumn, as shown in Figure 3(b) and which acted as a reference and surrogate control for non-fractionated AGP with respect to bi- vs tri-/tetra-antennary N-glycans). The base peak for the non-retained fraction of AGP on a Con A microcolumn was now [N₅H₆S₃]²⁺, which corresponded to a tri-antennary N-glycan with three sialic acid residues and no fucose residues. The composition of this peak was again confirmed by using MS/MS, as shown in the Supplementary Material. In addition, this peak showed no appreciable difference in intensity when examined in the retained vs non-retained fractions of AGP from an AAL microcolumn. All other peaks identified in the MS spectrum of the non-retained AGP fraction from a Con A microcolumn were determined to be tri- and tetra-antennary glycans with variable charge states and degrees of sialylation and fucosylation. The difference in peak intensity of bi-antennary glycans between the retained and non-retained AGP fractions (i.e., going from a relative intensity of 1.54 to 0.13 vs. the peak for [N₅H₆S₃]²⁺) confirmed that the Con A microcolumn could selectively enrich AGP that contained bi-antennary N-glycans. This result was consistent with previous work that has examined the selectivity of Con A towards glycopeptides containing bi-antennary vs. tri- and tetra-antennary glycans [37].

3.3. Analysis of selectivity of AAL for AGP glycoforms

The N-glycans of AGP that were present in the retained and non-retained fractions from an AAL microcolumn were also released by treatment with PNGase F and analyzed by nESI-MS. The mass spectra that were obtained for these fractions are shown in Figure 3(b). These mass spectra were also compared to those obtained in Figure 3(a) for fractions collected from the Con A microcolumn (i.e., which was used as a reference, as it gave a separation of AGP based on bi-vs tri-/tetra-antennary glycans rather than fucosylation).

In the AAL-retained fraction of AGP, the observed structures in the mass spectra ranged from bi-antennary to tri- and tetra-antennary glycans that contained no fucosylation or one-to-two fucose residues. The base peak was a non-fucosylated bi-antennary glycan, as represented here by [N₄H₅S₂]²⁺. The two peaks with the next highest relative abundances (i.e., 84% and 90%) corresponded to doubly- and triply-charged forms of a tri-antennary glycan with one fucose and three sialic acid residues; these peaks were labelled [N₅H₆F₁S₃]²⁺ and [N₅H₆F₁S₃]³⁺, where F is fucose and the other symbols are the same as defined earlier. A handful of peaks that corresponded to other fucosylated N-glycans, including singly- and doubly-fucosylated tri- and tetra-antennary structures, were also detected. The composition for representative peaks in this group (i.e., those for [N₄H₅S₂]²⁺ and [N₅H₆S₃]²⁺) was confirmed by MS/MS, as shown in the Supplementary Material.

Glycans that were released from the non-retained AGP fraction on an AAL microcolumn ranged in size from bi-antennary to tri- and tetra-antennary groups, with either no fucosylation or a single fucose residue being present. In this case, the signals corresponding to fucosylated glycans appeared to be minor, with relative intensities < 15% versus the base peak. The differences in glycan content of the two AGP fractions from the AAL microcolumn could be seen in the disappearance or decrease in intensity of fucosylated glycans when going from the retained to non-retained fractions, as illustrated in Figure 3(b). This change demonstrated the selectivity of an AAL microcolumn in retaining AGP that contained fucosylated N-glycans.

The amounts of bi-antennary N-glycans in the retained and non-retained fractions of AGP from an AAL microcolumn were not significantly different. This meant the relative intensities of other ions could be compared by using the non-fucosylated bi-antennary glycan ion [N₄H₅S₂]²⁺ as a reference in each fraction. The results of this comparison are shown in Table 1. It was found that the relative intensity of fucosylated ions in the retained fraction of AGP from an AAL microcolumn was 2.5- to 13.3-fold higher than the relative intensity due to fucosylated ions in the non-retained fraction. This confirmed that AAL was selectively interacting with AGP glycoforms that contained fucose residues, as has been noted in previous studies with complex-type oligosaccharides [38,39]. In contrast to this, the fucosylated glycan ions that showed an increase in the retained AGP fraction of an AAL microcolumn gave either similar amounts in the retained and non-retained fractions from the Con A microcolumn (e.g., [N₅H₆F₁S₃]³⁺ and [N₆H₇F₁S₄] ³⁺) or a larger amount in the non-retained fraction (e.g., [N₅H₆F₁S₂]²⁺ and [N₅H₆F₁S₃]²⁺).

Table 1.

Relative intensities of fucosylated ions in retained vs. non-retained AGP on AAL^a

Fucosylated ions	Retained fraction (%)	Non-retained fraction (%)	Ratio, % Retained vs. % Non-retained
Bi-antennary ions
[N4H5F1S2]2+	32.3	4.8	6.7
Tri-antennary ions
[N5H6F1S2]3+	4.7	1.9	2.5
[N5H6F1S3]3+	89.7	13.7	6.5
[N5H6F2S3]3+	9.9	1.7	5.8
[N5H6F1S2]2+	31.7	2.9	10.9
[N5H6F1S3]2+	84.0	6.3	13.3
[N5H6F2S3]2+	7.6	1.5	5.1
Tetra-antennary ions
[N6H7F1S3]3+	25.1	8.4	3.0
[N6H7F1S4]3+	23.8	3.2	7.4
[N6H7F2S4]3+	16.5	2.5	6.6

^aThe listed relative intensities use the measured intensity for [N₄H₅S₂]²⁺ as the base peak ion.

It was found that the there was one type of ion detected for a fucosylated bi-antennary N-glycan, while there were three to six ions detected for tri- and tetra-antennary N-glycans (see Table 1). The large distribution of ions for the tri- and tetra-antennary glycans in the mass spectra was due to the loss of sialic acid (i.e., during the ionization and analysis of glycans by MS) and the presence of various charge states. Fucosylation was most abundant in the tri-antennary N-glycans, giving a combined relative intensity of 230% vs. the base peak, followed by tetra-antennary N-glycans, with an overall relative intensity of 65%. Fucosylation was much less common for the bi-antennary N-glycans, with a relative intensity of 32% in the AAL-retained fraction of AGP. N-Glycans with two fucose residues were also observed in the AAL-enriched AGP fraction. Even though more types of tri-antennary N-glycan ions that contained one fucose vs two fucose residues were detected in the AAL-retained fraction, the tri-antennary N-glycan ions in this group that contained two fucose residues were more abundant (combined relative intensity, 18%). Tetra-antennary N-glycan ions in the same fraction and that contained two fucose residues, as represented by a single type of detected ion ([N₆H₇F₂S₄]³⁺, had a slightly lower relative intensity (17%). No fucosylated ions with two fucose residues were found in the bi-antennary glycans from the AAL-retained fraction of AGP.

3.4. Combined use of Con A and AAL microcolumns for analysis of AGP glycoforms

A Con A microcolumn was coupled off-line with an AAL microcolumn for a multi-column fractionation of AGP glycoforms. The AAL microcolumn was used in the second part of this scheme because its strong binding to fucosylated AGP glycoforms made it possible to concentrate these glycoforms as they eluted under isocratic conditions from the Con A microcolumn. As shown in Figure 2(a), a Con A microcolumn gave a non-retained peak that accounted for 55% of the total area for AGP and a retained peak that made up the remaining 45%. As demonstrated previously in this report (Section 3.2), the non-retained peak for AGP on the Con A microcolumn represented glycoforms that contained tri- and tetra-antennary N-glycans and minor amounts of bi-antennary N-glycans. The retained AGP peak represented glycoforms that contained bi-antennary N-glycans plus co-existing tri- or tetra-antennary N-glycans.

The non-retained and retained AGP fractions were collected from the Con A microcolumn. These fractions were exchanged into the application buffer for the AAL microcolumn and concentrated to a volume of about 200 μL (i.e., ~1 mg/mL AGP). A typical separation for these samples, as illustrated for AGP that was retained by a Con A microcolumn, is provided in Figure 2(b). Similar results were obtained for the non-retained fraction of AGP from a Con A microcolumn (see Supplementary Material). For the example in Figure 2(b), the retained peak for AGP from the AAL microcolumn made up ~10% of the total area for this protein (Note: this result agreed with prior observations made in Ref. [32] for intact AGP using similar AAL supports). As noted previously (see Section 3.3), this retained fraction contained the majority of AGP glycoforms with some level of fucosylation. This meant intact fucosylated AGP, and which contained fucose groups that were accessible to AAL for binding (i.e., without any significant steric hindrance effects), accounted for only about one tenth of the original set of AGP glycoforms.

A closer examination of the results for this two-column fractionation (i.e., as given in Figure 2(b) and the Supplementary Material) indicated the level of fucosylation was different in the retained and non-retained fractions of AGP from a Con A microcolumn. For instance, 10.4 (± 0.3)% of the AGP that was non-retained by a Con A microcolumn was retained by an AAL microcolumn (i.e., around 10% of this AGP was fucosylated), while only 6.9 (± 0.4)% of the AGP retained by a Con A microcolumn was also retained by an AAL microcolumn. In other words, the non-retained fraction from a Con A microcolumn had 1.5-fold more AGP with AAL-binding activity than the AGP fraction that was retained by Con A (i.e., a difference significant at the 95% confidence level). This meant fucosylation of the N-linked glycans, as measured on an AAL microcolumn, was slightly more abundant when a higher degree of branching was present in AGP, as detected by using the Con A microcolumn. These results were consistent with the structural information that was obtained by nESI-MS for the AGP fractions from an AAL microcolumn. Similar results have been observed for control human serum, where a high degree of fucosylation was associated with a low content (or total absence) of bi-antennary glycans and a high content of tri- or tetra-antennary glycans [2].

3.5. Combined use of Con A microcolumn and CE for analysis of AGP glycoforms

A Con A microcolumn was next combined with CE for the separation of AGP glycoforms. CE was used to separate intact AGP glycoforms based on their overall charge, as is related to the amount of sialic residues that were present [10,11]. Electrophoretic injection was used in the CE method to concentrate the applied AGP and compensate for sample dilution in the first part of the fractionation step [10,11] (i.e., based on a Con A microcolumn). A typical separation that was obtained for AGP by the CE method is shown in Figure 4(a). In this example, AGP was separated into nine major glycoform bands within ~20 min. It has been shown in prior work that the neighboring glycoform bands in this method differ by one charge [10], with the AGP band at the shortest migration time having the largest negative charge and largest number of sialic acids [10,11].

Figure 4.

(a) Electropherograms and (b) corresponding glycoform patterns obtained for the non-retained and retained fractions of AGP on a Con A microcolumn. The numbers in (a) and (b) refer to the relative order of migration for the glycoform bands. The error bars shown for the migration times and % peak areas in (b) represent ± 1 S.D. (n = 3) and are often comparable in size to the symbols in this plot. Conditions for CE: separation voltage, −30 kV; temperature, 25 °C; running buffer, 20 mM sodium acetate buffer (pH 4.2) containing 0.05% PEO and 0.1% Brij 35; injection conditions, −5 kV for 5 min: capillary, 50 μm I.D., 360 μm O.D., 60.2 cm total length, 50 cm length to detector. The capillary coating procedure is described in Section 2.4.

The CE results were compared for the non-retained and retained AGP fractions from a Con A microcolumn. The glycoform patterns seen here for the retained and non-retained fractions followed the general pattern of nine glycoform bands that has been observed previously with this same CE method for non-fractionated AGP [11]. Glycoform band 5 had the largest relative peak area for the non-retained Con A fraction of AGP, while glycoform bands 6–7 gave the largest relative peak areas for the retained fraction. However, non-fractionated AGP has been found to have its most abundant glycoforms in band 6 followed by band 5, which is a result intermediate between those shown in Figure 4 for the Con A non-retained and retained fractions of AGP) [11].

The Con A-retained fraction of AGP was found to contain a larger level of AGP in the glycoform bands that had smallest amounts of sialic acid. This is illustrated in Figure 4(b), where glycoform bands 6–9 in the Con A-retained AGP fraction had relative contributions to the total peak area that were up to 15% higher than the contributions by the same bands for the non-retained fraction. The percent areas of these bands had precisions of ± 0.01–0.13% (n = 3), with the difference in percent areas for the retained vs. non-retained AGP fractions being significant at 95% confidence level. This change was accompanied by a corresponding decrease in the percent area contributions by bands 1–5 in the Con A-retained fraction of AGP vs. the non-retained fraction; the percent areas of these bands had precisions of ± 0.03–0.27% (n = 3) and were also significantly different at the 95% confidence level.

The change in distribution for the glycoform peaks in Figure 4 was further reflected by a shift in the composition-weighted average migration time for the combined bands of AGP. This shift was represented by an increase in the average migration time of 1.4 min (i.e., a change significant at the 95% confidence level) in going from the non-retained to retained AGP fractions. This shift between the non-retained and retained fractions of AGP indicated there was a decrease in the number of sialic acids in going from AGP glycoforms that had a high degree of branching to those with a relatively low degree of branching. This observation is consistent with the fact that the number of sialic acids is usually paired with the degree of branching in complex type N-glycans [2]. This general observation was confirmed by the MS results in Figure 3(a), where 2–4 sialic acids were associated with the observed tri- and tetra-antennary glycans while only 2 sialic acids were noted on the bi-antennary glycan.

3.6. Combined use of AAL microcolumn and CE for analysis of AGP glycoforms

An AAL microcolumn was also combined with CE for the separation of AGP glycoforms. The glycoform patterns that were obtained are shown in Figure 5 (see Supplementary Material for corresponding electropherograms). These patterns were similar for the retained and non-retained fractions from the AAL microcolumn, and were also consistent with results found in prior work with this CE method for non-fractionated AGP [11]. When compared by CE with the non-retained AGP fraction, the fraction that was retained by an AAL microcolumn contained a slightly higher percentage of AGP glycoform bands at smaller migration times and with more sialic acid residues. For instance, bands 1–5 in the AAL-retained fraction of AGP had up to 2.5% higher contributions to the total peak area than the same bands in the non-retained AGP fraction, while peaks 6–9 (i.e., with lower sialic acid content) had up to 2.8% lower contributions to the total peak area. These differences were significant at the 95% confidence level but were much smaller than those seen by CE for the non-retained vs. retained fractions of AGP from a Con A microcolumn. However, these results did indicate that sialic acid residues were slightly more abundant in AGP when more fucose residues were present in the N-glycans.

Figure 5.

Glycoform patterns obtained by CE for the non-retained and retained fractions for AGP when using an AAL microcolumn. The numbers in the plot refer to the relative order of migration for the glycoform bands. The error bars shown for the migration times and % peak areas represent ± 1 S.D. (n = 3) and are often comparable in size to the symbols in this plot. The electropherograms for these fractions are provided in the Supplementary Material.

The average migration times of the AGP fractions from the AAL microcolumn were also compared by CE. It was found that glycoform bands 5 to 9 for AGP in the AAL-retained fraction were consistently 5–14 s longer in their average migration times than the bands for the non-retained AGP fraction that had the same charges and number of sialic acids. This difference in migration time was significant at the 95% confidence level. A similar shift was seen in bands 1–4. This small change in migration time is believed to be due to the change in average mass for AGP glycoforms when fucose residues were present on the glycans of AGP [10].

4. Conclusion

In this study, three formats for combined separation methods, based on lectin microcolumns and CE, were developed and used for the analysis of AGP glycoforms. The release of glycans by digestion with PNGase F and the analysis of this digest by nESI-MS was also used to examine the glycan composition of AGP that had been fractionated by lectin microcolumns. It was found that Con A microcolumns tended to retain AGP glycoforms that contained bi-antennary N-glycans, while AAL microcolumns retained fucose-containing AGP glycoforms. A two-column fractionation based on both Con A and AAL microcolumns indicated that fucosylation of N-linked glycans was slightly more abundant when a higher degree of branching was present in AGP. The combined use of a Con A microcolumn with CE showed that sialic acids were more abundant in AGP when higher degrees of glycan branching were present. A combined separation approach using an AAL microcolumn and CE demonstrated that sialic acid residues were slightly more abundant in AGP when there was a higher level of fucosylation present. Results from these combined methods verified that N-glycans with a higher degree of branching are associated with more sialic acid and fucose residues in AGP [2].

This work with AGP demonstrated the feasibility of coupling multiple lectin microcolumns with each other or with CE to provide combined separation methods for the analysis of an intact glycoprotein and its glycoforms. The lectin microcolumns were found to be useful in fractionating intact AGP according to classes of glycan structures and provided complementary information to both each other and CE. Future studies will explore how these methods can be coupled on-line with each other and with MS for the separation and analysis of AGP and other glycoproteins or glycoconjugates with complex compositions [40].

Highlights.

• Lectin microcolumns were used to fractionate alpha₁-acid glycoprotein glycoforms.

• Microcolumns containing concanavalin A and Aleuria Aurantia lectin were examined for their binding selectivity to these glycoforms.

• Mass spectrometry was also used to examine the selectivity of the lectin microcolumns.

• The lectin microcolumns were used combined with each other and with capillary electrophoresis.

• These methods can be adapted for the fractionation and analysis of other glycoproteins.

Abbreviations:

AAL

Aleuria aurantia lectin

AGP

alpha₁-acid glycoprotein

BCA

bicinchoninic acid

CID

collision-induced dissociation

Con A

concanavalin A

DTT

dithiothreitol

F

fucose

H

hexose

IAA

iodoacetamide

MS/MS

tandem mass spectrometry

N

N-acetylhexosamine

nESI-MS

nanoflow electrospray ionization mass spectrometry

PEO

poly(ethylene oxide)

Q-TOF-MS

quadrupole time-of-flight hybrid mass spectrometer

S

sialic acid

TFA

trifluoroacetic acid