Optimized superficially porous particles for protein separations

Abstract

Continuing interest in larger therapeutic molecules by pharmaceutical and biotech companies provides the need for improved tools for examining these molecules both during the discovery phase and later during quality control. To meet this need, larger pore superficially porous particles with appropriate surface properties (Fused-Core^® particles) have been developed with a pore size of 400 Å, allowing large molecules (<500 kDa) unrestricted access to the bonded phase. In addition, a particle size (3.4 μm) is employed that allows high-efficiency, low-pressure separations suitable for potentially pressure-sensitive proteins. A study of the shell thickness of the new fused-core particles suggests a compromise between a short diffusion path and high efficiency versus adequate retention and mass load tolerance. In addition, superior performance for the reversed-phase separation of proteins requires that specific design properties for the bonded-phase should be incorporated. As a result, columns of the new particles with unique bonded phases show excellent stability and high compatibility with mass spectrometry-suitable mobile phases. This report includes fast separations of intact protein mixtures, as well as examples of very high-resolution separations of larger monoclonal antibody materials and associated variants. Investigations of protein recovery, sample loading and dynamic range for analysis are shown. The advantages of these new 400 Å fused-core particles, specifically designed for protein analysis, over traditional particles for protein separations are demonstrated.

1. Introduction

Superficially porous particles (SPPs) have had an important impact on the manner in which HPLC separations are performed. Large (~30 μm) silica SPPs called Zipax^® superficially porous particles were introduced by DuPont in the late 1960s as the first particles specifically designed for HPLC separations [1,2]. Shortly thereafter, these SPPs were covalently modified with a bonded silicone stationary phase to form particles called Permaphase^® by DuPont [3,4]. Columns of these particles were used to form the basis for the now widely used reversed-phase HPLC separations because of the stability in various mobile phases and the capability of gradient elution applications. The advent of 5–6 μm totally porous silica microspheres in the early 1970s [5,6] quickly ended the use of the larger SPPs, and it was more than 20 years before this technology re-emerged. 6-μm SPPs with wide pores for separating large solutes were reported in 1992 [7], and 5-μm SPPs for separating peptides and proteins were described in 2000 [8]. Even smaller (2.7 μm) SPPs were introduced in 2006 by Advanced Materials Technology, Inc. (AMT) as Halo^® fused-core particles [9]. The strong success of columns with these superficially porous particles, based on unusual high efficiency at modest operating pressure, quickly resulted in the commercial introduction of closely similar competing products often called porous shell or core–shell particles by other manufacturers. Many reports have been prepared on Halo and similar particles as a result of the superior performance available with these materials [e.g., 10–16].

The fused-core particles were designed to fit the strategy that different specifications are needed for the optimum HPLC separation of various compound classes, usually based on solute size. Table 1 summarizes the physical properties and utility ranges for fused-core particles prepared to fit this strategy. Some of these data were given in a previous report [17], but are presented here for convenience. The original fused-core particles were synthesized with a 2.7 μm overall diameter, having a 0.50 μm-thick porous shell with 90 Å pores [18]. This specification resulted in a particle with a surface area of 135 m²/g and a 75% overall particle porosity. Such particles have the capability of separating small molecules, extending to about 5 kDa, without solute diffusion restriction that would degrade column efficiency. The second fused-core particle introduced by AMT (called Halo^® Peptide) had the same overall size and porous shell thickness, but has a pore size of 160 Å allowing unhindered high-efficiency separation of larger solutes such as protein fragments and polypeptides up to about 15 kDa without restricted diffusion [19].

Table 1.

Physical characteristics of fused-core particles.

Fused-core particle	Particle size (μm)	Pore size (Å)	BET surface area (m2/g)	Shell thickness (μm)	% Porosity	Pore volume (cm3/g)	Separation utility
Halo	2.7	90	135	0.50	75	0.26	Small molecules, <5 kDa
Halo peptide	2.7	160	80	0.50	75	0.29	Peptides, <15 kDa
Wide pore	3.4	400	10	0.15	25	0.068	Proteins, <500 kDa
Wide pore	3.4	400	15	0.20	31	0.113	Proteins, <500 kDa
Wide pore	3.4	400	18	0.25	37	0.125	Proteins, <500 kDa

The subject of this presentation is fused-core particles with even wider pores (400 Å) that are designed to separate proteins and other large molecules up to about 500 kDa (Table 1). Particles with such wide pores are again required to eliminate restricted diffusion effects and allow high-efficiency fast separations for very large solutes. Data in a previous report had shown that larger SPPs with wide pores produce columns of higher efficiency, so the particle size for the new wide-pore particles was set at 3.4 μm [17] to retain high efficiency while permitting operation at lower pressure to protect pressure-sensitive proteins from degradation. An outstanding unknown regards the thickness of the porous shell needed for practical optimization of particles for separating proteins and other large molecules.

Theory and previous studies [17,20] indicated that a thin shell should be used to compensate for the small diffusion coefficients and resulting poor mass transfer of large molecules, but how thin? To answer this question, three different batches of 3.4 μm particles with 400 Å pores were synthesized with 0.15, 0.20 or 0.25 μm-thick shells so that these particles had surface areas of 10, 15 and 18 m²/g, respectively, as indicated in Table 1. This presentation describes studies performed on these particles and conclusions drawn from appropriate experimental data. In this report, particles with the same specifications are sometimes called Halo 400, which denotes a research identity, or Halo^® Protein, the same particles which have recently been commercially delivered.

2. Experimental

Proteins used as sample probes and mobile phase modifiers, including formic acid and ammonium formate, were obtained from Sigma–Aldrich (St. Louis, MO). Trifluoroacetic acid was from Pierce Chemicals (Rockford, IL) and acetonitrile from EMD (Gibbstown, NJ). Silanes for the bonding reactions were obtained from Gelest, Inc. (Morrisville, PA). IgG1 monoclonal antibodies (provided by Pfizer) were reduced by treatment with 100 mM dithiothreitol (DTT) in 8 M guanidine HCl at 50 °C for 35 min. Additional samples of IgGs were reduced and alkylated by sequential treatment with 10 mM DTT, 15 mM iodoacetamide, then quenched with an additional 10 mM DTT, all in 6 M guanidine HCl/20 mM Tris–HCl buffer at pH 7.8. The reduced and alkylated IgG solutions were buffer exchanged into 0.1% TFA using VivaSpin (Sartorius Stedim Biotech, Goettigen, Germany) centrifugal concentrators with 5 kDa cut-off HY polymeric membranes. The reduced and alkylated IgGs were adjusted to 2 mg/mL protein in 0.1% TFA and stored at −25 °C until use. Electrophoretic analysis of reduced IgG1 monoclonal antibody components were conducted on Bio-Rad (Hercules, CA) Criterion TGX 4–20% gradient SDS-PAGE gels that were visualized using Bio-Safe Coomassie G-250 stain, using conditions as essentially recommended by the manufacturer.

Particle sizes were determined with a Coulter Multisizer 3 instrument (Fullerton, CA). Surface areas and pore size distributions were measured by nitrogen adsorption and desorption with a Micromeritics Tristar II instrument (Norcross, GA). These measurements were made on 1.5–2.0 g particle samples using 10 s equilibration times for each point. Shell thicknesses were determined by the difference in Coulter counter measurements for the starting cores and the final particles. This parameter was also confirmed by cross-section focused ion beam scanning electron micrographs prepared at the Department of Materials Science and Engineering at the University of Delaware.

Columns of the Halo 400 fused-core particles were prepared at Advanced Materials Technology (Wilmington, DE). A column of 1.7 μm totally porous particles with 300 Å (BEH-300) was obtained from Waters Corporation (Milford, MA), and a column of 2.6 μm core–shell particles with 200 Å pores from Phenomenex (Aeris Widepore, Torrence, CA). HPLC data were collected with an Agilent 1100 liquid chromatograph (Palo Alto, CA), and with this instrument, data acquisition and instrument control used version B.01.03 ChemStation software. To minimize extra-column band broadening effects, a 5 μL flow cell with the heat exchanger bypassed was used. Some measurements were made with an Agilent 1200 SL instrument with a 2 μL flow cell. All of the capillary tubing connections and needle seats were 0.12 mm ID for both Agilent instruments. LC/MS studies on monoclonal antibody samples were performed with a Shimadzu Nexera LC connected to the MS-2020 single quadrupole MS instrument using ESI at +4.5 kV capillary voltage, 2 pps scan rate from 500 to 2000 amu m/z. Component masses for these measurements were determined by deconvolution using MagTran software, v. 1.02, based on the ZScore algorithm developed by Zhang and Marshall [21]. Corrections for instrumental extra-column band broadening effects were not applied to any of the data obtained in this study.

3. Results and discussion

3.1. Effect of particle characteristics

The effect of porous shell thickness is demonstrated by the data in Fig. 1. These plate height vs. linear mobile phase velocity plots for C4 stationary phase (described later) were constructed with experimental measurements on carbonic anhydrase (29 kDa) using an isocratic mobile phase of acetonitrile/aqueous trifluoroacetic acid at 60 °C. Slight differences in acetonitrile concentration were used to produce comparable solute retention factor k values. Data were fitted to the Knox equation [22], a relationship similar to that used to produce so-called van Deemter plots. The thinnest shell (0.15 μm) appears to have produced data with the highest efficiency and with slightly superior mass transfer (shallower plot slope and a smaller van Deemter C-term) than thicker porous shells. Plate heights and plot slopes increased with shell thickness increase. Although the data for the three shell thicknesses may not be statistically different, the trend is in keeping with the change in shell thickness. Table 2 shows the effect of porous shell thickness on the Knox equation coefficients. Values for the B-term are not reported since insufficient data were available at the required very low mobile phase velocities for this large solute. However, A- and C-term values are consistent and in keeping with the change in shell thickness.

Fig. 1.

Effect of porous shell thickness. Columns: 100 mm × 2.1 mm; temperature: 60 °C; mobile phase: 0.15 μm shell, 42% acetonitrile/56% aqueous 0.1% (v/v) trifluoroacetic acid; 0.20 μm shell, 42.3% acetonitrile/57.7% aqueous 0.1% (v/v) trifluoroacetic acid; 0.25 μm shell, 42.7% acetonitrile/57.3% aqueous 0.1% (v/v) trifluoroacetic acid; solute: carbonic anhydrase, 29 kDa.

Table 2.

Effect of porous shell thickness on Knox equation coefficients (h = Av^1/3 + B/v + Cv).

Knox coefficient	A	C
Halo 400, 0.15 μm	0.0251 ± 0.0008	0.0069 ± 0.0003
Halo 400, 0.20 μm	0.0256 ± 0.0007	0.0074 ± 0.0002
Halo 400, 0.25 μm	0.0258 ± 0.0014	0.0074 ± 0.0005

Note: values for B-term not reported as insufficient data were available for the protein at very low mobile phase velocities.

Sample loading studies for the C4-bonded particles, with different shell thicknesses are summarized in Fig. 2. Here, an acetonitrile/trifluoroacetic acid mobile phase gradient at 60 °C was used to elute the protein, myoglobin (17.7 kDa), which was injected in increasing mass amounts. Plots of myoglobin peak width vs. micrograms of protein injected show that the thicker shell and higher surface area tolerates higher amounts of protein before peak broadening occurs. Thinner shells show greater increase in peak widths with protein mass injected. Therefore, the effect of sample loading is opposite from that found for column efficiency (Fig. 1). Thinner shells give higher efficiency whereas thicker shells are more tolerant to increased sample loading (and greater solute retention).

Fig. 2.

Sample loading studies. Column: 100 mm × 4.6 mm; gradient: 39–49% B in 10 min; mobile phase A: aqueous 0.1% (v/v) trifluoroacetic acid; mobile phase B: acetonitrile with 0.1% (v/v) trifluoroacetic acid; temperature: 60 °C; flow rate: 0.5 mL/min; injection volume: 5 μL; solute: myoglobin, 17.7 kDa.

As a result of these two studies, a decision was made to settle on a fused-core particle with a 0.20 μm-thick porous shell as a viable practical compromise. This report now uses a 3.4 μm particle with a 0.20 μm-thick shell and 400 Å pores as the particle of study. The drawing on the left of Fig. 3 illustrates the physical specifications of this particle. The figure at the right of Fig. 3 shows a focused ion beam scanning electron micrograph (FIB/SEM) of the subject fused-core particles, indicating the consistent spherical shape for these wide-pore particles. More detailed information on these fused-core particles is shown in Fig. 4 with high-resolution FIB/SEM pictures of particles. The picture at the left of a single fused-core particle in Fig. 4 has sufficient resolution so that individual pores (400 Å) can be seen in the surface of the porous shell as a result of the very high resolution of the FIB/SEM method. Such resolution is not obtained by conventional SEM. On the right of Fig. 4 is a cross-section of a particle created by cleavage with the focused ion beam. Looking slightly downward on the particle, one can see the structure of the porous shell. The bottom of this particle micrograph clearly shows the shell thickness of 200 nm (0.20 μm) based on the scale provided in the bottom right-hand corner.

Fig. 3.

Wide-pore fused-core particles for protein separations. Left: cartoon of particle dimensions. Right: scanning electron micrograph of particles.

Fig. 4.

Focused ion beam scanning electron micrographs of wide-pore fused-core particles. Left: single particle showing porous outer shell. Right: particle cross-section showing shell thickness.

Fig. 5 shows pore size distribution plots for the subject particles measured by both nitrogen adsorption and desorption. The adsorption plots on three different batches of these particles show pore medians of slightly more than 400 Å. Compared to conventional wide-pore particles with 300 Å pores or smaller, a pore diameter of 400 Å was selected to allow the access of larger proteins to the pores without restricted diffusion that would degrade column efficiency. The reproducibility of data in Fig. 5 gives evidence of the developed fused-core synthesis process to produce particles with closely similar characteristics.

Fig. 5.

Pore size characterization of wide-pore fused-core particles by nitrogen adsorption and desorption.

3.2. Particle comparisons

Plate height vs. mobile phase velocity plots for columns of the Halo 400 fused-core particles and widely-used 1.7 μm totally porous particles are compared in Fig. 6. Data on carbonic anhydrase (29 kDa) were obtained using an isocratic acetonitrile/trifluoroacetic acid mobile phase at 60 °C. Slightly different acetonitrile concentrations were used to ensure that the retention factor k values were similar for this comparison. The Knox equation-fitted data clearly show that the plate heights and mass transfer for the fused-core particles are actually slightly superior even though the particle size is twice that of the 1.7 μm totally porous particles. If the particle size is taken into account by reduced plate height comparison (plate height divided by particle size), the comparison is even more dramatic, as shown in Fig. 7. The large difference in performance is further illustrated in Fig. 8 where the pressures required for comparable separations are taken into account. Here, peak widths multiplied by operating pressure are plotted versus the retention times for various proteins. Using the same separation conditions for acetonitrile/trifluoroacetic acid gradients, the fused-core Halo Protein columns show almost a three-fold superiority because of the much lower operating pressure required. This peak width times pressure plot is somewhat related to the separation impedance value developed by Bristow and Knox [23].

Fig. 6.

Plate height comparison. Columns 100 mm × 2.1 mm; mobile phase: 1.7 μm totally porous - 40.5% acetonitrile/59.5% aqueous 0.1% (v/v) trifluoroacetic acid, = 3.4; 3.4 μm Halo 400 – 42.5% acetonitrile/57.5% aqueous 0.1% (v/v) trifluoroacetic acid, k = 3.6; solute: carbonic anhydrase (29 kDa), 0.1 mg/mL in 6 M urea/1.0% acetic acid; temperature: 60 °C.

Fig. 7.

Reduced plate height comparison. Conditions as for Fig. 6.

Fig. 8.

Peak width times pressure comparison for proteins. Columns: 100 mm × 2.1 mm, 3.4 μm Halo Protein, 400 Å, C4 and 1.7 μm totally porous, 300 Å, C4; gradient: 20–69% acetonitrile/aqueous 0.1% (v/v) trifluoroacetic acid in 45.7 min; flow rate: 0.2 mL/min; temperature: 40 °C; solutes and column pressures as shown.

It is well known that the retention factor of proteins can vary with flow rate and pressure used for the separation [24–26]. This effect is further illustrated in Fig. 9 for the fused-core and totally porous particles of this discussion. Isocratic data with comparable acetonitrile/trifluoroacetic acid mobile phases (data from Fig. 6) for carbonic anhydrase (29 kDa) show significant retention variations with both particles for flow rate (and pressure) changes. One might speculate that these variations are caused by changes in the solvated state of the stationary phase, or by changes in the conformation of the large protein molecules with pressure (and flow rate) changes, or both.

Fig. 9.

Effect of flow rate on retention factor k. Same conditions as Fig. 6.

3.3. Stationary phases

Three different monofunctional stationary phases were studied for reversed-phase protein separations with the new wide-pore fused-core particles, as listed in Table 3. The C4 ligand was n-butyldimethyl that was densely bound to the particles with a concentration of 4.3 μmol/m² using special bonding techniques. This dense C4 stationary phase resulted in superior properties for protein separations, as indicated in later sections of this report.

Table 3.

Stationary phases for Halo Protein column packings.

Stationary phase	Si Ligand type	Concentration (μmol/m2)	Endcapped
C4	n-butyldimethyl	4.3	Yes
C8	n-octyldiisopropyl	2.2	Yes
C18	n-octadecyldiisobutyl	2.2	Yes

Sterically-protected phases of both ES-C8 and ES-C18 [27,28], n-octyldiisopropyl and n-octadecyldiisobutyl, were used to ensure stationary phase stability, and these ligands were bonded at the maximum concentration of 2.2 μmol/m² for these bulky functional groups. All bonded stationary phases for this study were endcapped.

Protein separations of columns with the three bonded phases are shown in Fig. 10. As might be expected, retention for components in this mixture of seven proteins ranging from 12.4 to 200 kDa generally is slightly larger for particles with longer ligands for this acetonitrile/trifluoroacetic acid mobile phase gradient. Slight variations in selectivity are seen, which also might be expected for different stationary phases.

Fig. 10.

Comparison of bonded phase ligands for fast protein separations. Columns: 100 mm × 2.1 mm; gradient: 35–66% in 5.0 min; mobile phase A: aqueous 0.1% trifluoroacetic acid; mobile phase B: 80/20 acetonitrile/aqueous 0.1% trifluoroacetic acid; flow rate: 0.5 mL/min; temperature: 60 °C; injection volume: 1 μL; detection: 280 nm; proteins as shown.

Column stability data for two of the bonded ligands are summarized in Fig. 11. Acetonitrile/trifluoroacetic acid 10 min gradient separations of five proteins at 90 °C were continuously sequentially used to develop these data. Shown at the left of Fig. 11, the C4 stationary phase shows almost no change in retention (or in peak efficiency or peak shape – not included) after almost 15,000 column volumes of mobile phase at 90 °C. No change in the characteristics of the sterically-protected ES-C8 phase (right, Fig. 11) was detected after almost 15,000 column volumes of mobile phase passed through the column at 90 °C. These results clearly indicate that these columns have excellent stability when used with low pH mobile phases at high temperatures. The stability of the C4 stationary phase might be considered remarkable in light of the short chain for this ligand. It is speculated that the very dense bonding of this group may inhibit the access of acidic water to the site-attachment of the bonded ligand to the silica support, so that hydrolysis is minimized.

Fig. 11.

Column stability studies. Columns: 100 mm × 2.1 mm as shown; mobile phase gradient: 25–40% acetonitrile/aqueous 0.1% trifluoroacetic acid in 10 min, continuously repeated; temperature: 90 °C; flow rate: 0.5 mL/min; detection: 215 nm; solutes, proteins as shown.

The effect of temperature variations on protein separations is shown in Fig. 12. The peaks for five proteins ranging from 14.3 to 66.4 kDa increase in size and sharpness and retention decreases as the temperature increases from 30 to 60 and 90 °C for this fast gradient. This difference in peak size (height and area) phenomenon is not completely understood. However, this and other evidence suggest that the solubility of proteins might be exceeded during the passage of a fast acetonitrile gradient at lower temperatures. Higher temperatures apparently maintain protein solubility so that total elution can occur. Slower gradients also increase peak size elution at lower temperatures. We find that at lower temperatures and fast gradients, reversed-phase columns sometimes tend to show separation changes after continued use, suggesting a buildup of non-eluted protein on the packing particles. When this occurs, a thorough purging with 90% acetonitrile/10% water/0.1% trifluoroacetic acid fully restores the column to original performance (data not shown).

Fig. 12.

Effect of temperature on protein separations. Column: 100 mm × 2.1 mm Halo Protein C4; gradient: 28–58% B in 10.0 min; mobile phase A: aqueous 0.1% trifluoroacetic acid; mobile phase B: acetonitrile with 0.1% trifluoroacetic acid; flow rate: 0.45 mL/min; instrument: Agilent 1200 SL; injection volume, 2 μL; detection: 215 nm; temperatures as indicated; solutes, proteins as shown.

3.4. Other studies

Tests were conducted to determine the effect of fused-core particle surface chemistry on protein peaks in mobile phases other than those with trifluoroacetic acid. Fig. 13 shows the elution of model proteins with an acetonitrile/formic acid mobile phase gradient that is typically used to conduct mass spectrometry (MS) identifications. Protein peaks for the column of 3.4 μm Halo 400 particles are sharp, well-defined and formed at a much lower operating pressure with increased separation selectivity compared to that for a 1.7 μm totally porous particle with 300 Å pores. Studies have shown that even further improvement in efficiency and peak shapes can be obtained by using 10–20 mM ammonium formate in the formic acid mobile phase [29]. This salt can be added without deleterious change in the mass spectrometric ionization efficiency for peptides and proteins, although protein and peptide specific changes in the ionization charge state distributions are often observed.

Fig. 13.

Effect of formic acid mobile phase. Columns: 100 mm × 2.1 mm, 1.7 μm totally porous, 300 Åand3.4 μm Halo 400 Å, both C4; gradient: 20–70% B in 10.0 min; mobile phase A: aqueous 0.1% formic acid; mobile phase B: acetonitrile with 0.1% formic acid; flow rate: 0.3 mL/min; temperature: 60 °C; pressures as shown; injection, 2 μL; detection: 280 nm; solutes, proteins as shown.

Particle pore size has a significant influence on column performance and peak shapes for very large molecules. For the data in Fig. 14, a mixture of small and very large proteins was separated with an acetonitrile/trifluoroacetic acid mobile phase gradient at 60 °C. The first two eluting peaks for smaller proteins show comparable peak widths for 3.4 μm Halo 400 Å C4 particles and a commercial C4 column of 3.6 μm core–shell particles with 200 Å pores. However, as the size of the proteins gets larger, the 200 Å particles show broader peaks, suggesting that restricted diffusion was involved. The fused-core particles with larger 400 Å pores continue to elute sharp, well-defined peaks, with no evidence of restricted diffusion for these large proteins.

Fig. 14.

Effect of pore size in protein separations. Columns: 100 mm × 2.1 mm, 3.4 μm Halo Protein C4 with 400 Å pores and a 3.6 μm pore–shell C4 particle with 200 Å pores; gradient: 40–47% in 10.0 min; mobile phase A: 0.1% aqueous trifluoroacetic acid; mobile phase B: 80/20 acetonitrile/water with 0.1% trifluoroacetic acid; flow rate: 0.3 mL/min; temperature: 60 °C; instrument: Shimadzu Nexera; injection volume: 1 μL; detection: 280 nm; solutes, proteins as shown; Peak widths in minutes.

Protein recovery studies were also undertaken with the Halo 400 C4 column. Protein peaks were collected from a 4.6 mm × 100 mm column using a 10 min acetonitrile/0.1% aqueous trifluoroacetic acid gradient at 60 °C. Mobile phase blanks were obtained by replacing the column with a union. Fractions were lyophilized and reconstituted, and then directly measured colorimetrically, using the method given in Section 2. Each sample was determined in duplicate and recoveries of 100 ± 5.8% and 92 ± 18% were found for cytochrome c and catalase, respectively. These results suggest that these proteins elute from the fused-core columns without significant change under the gradient separating conditions used.

3.5. Application studies

When the sulfide bonds for IgG1 subtype monoclonal antibodies are reduced, two light chains plus variants and two heavy chains plus variants are expected, as indicated by the drawings at the top of Fig. 15. It is well established that the variants can include specific amino acid side-chain modifications, as well as N-linked glycosylation variants. A fast gradient separation using acetonitrile/aqueous trifluoroacetic acid at 80 °C produced the chromatogram shown at the bottom of Fig. 15, where the light and heavy chains are well separated. Enlargement of the chromatogram near the heavy chain peak shows some variants more clearly. The components in this sample also were confirmed by electrophoretic analysis (not shown here).

Fig. 15.

Reduced monoclonal antibody IgG1 separation with TFA mobile phase. Top: reduction reaction cartoon. Bottom: separation with 100 mm × 2.1 mm column of Halo Protein C4; Gradient: 33–40% B in 10.0 min; mobile phase A: 0.1% (v/v) aqueous trifluoroacetic acid; mobile phase B: 80/20 acetonitrile/0.1% (v/v) aqueous trifluoroacetic acid; flow rate: 0.25 mL/min; temperature: 80 °C; instrument: Shimadzu Nexera; injection: 1 μL; detection: 280 nm; enlargement of 5–6.5 min shown.

Changes in the gradient can greatly increase the separation of the variants from the main chains, as shown in Fig. 16. Here, a shallower gradient using an acetonitrile-isopropanol/formic acid-ammonium formate gradient separation was performed at 80 °C on a reduced and alkylated IgG1 using serial UV and MS detection to monitor the much improved resolution of the variants from the main chains formed. The addition of a small amount of ammonium formate to formic acid is known to improve column efficiency and protein peak shapes. This mobile phase is highly satisfactory for LC/MS studies because of superior ionization of samples without instrumentation difficulties [29]. Shown below the main light chain peak are the raw MS data that was obtained during this separation. To the right of these raw data is a table of the mass values calculated for the peaks as a result of deconvolution of MS data for all peaks.

Fig. 16.

High resolution separation of mAb IgG1 light and heavy chains with LC/MS. Column: 100 mm × 2.1 mm Halo Protein C4; gradient: 29–32% B in 20 min; mobile phase A: 0.5% (v/v) formic acid with 20 mM ammonium formate; mobile phase B: 45% acetonitrile, 45% isopropanol/0.5% (v/v) formic acid with 20 mM ammonium formate; temperature: 80 °C; flow rate: 0.4 mL/min; instrument: Shimadzu LCMS 2020; detection: 280 nm and MS as described in Section 2. Top: high-resolution separation chromatogram. Bottom: left – MS raw data for main light chain peak; right – Mass values for deconvoluted peaks.

4. Conclusions

This study has shown that columns of 3.4 μm fused-core particles with a 0.20 μm-thick shell and 400 Å pores give superior chromatographic performance for separating proteins. Stationary phases of C4, ES-C8 and ES-C18 provide excellent separation of proteins with sharp peaks and high column efficiency. High recoveries were found for typical proteins when collected from fast acetonitrile/aqueous trifluoroacetic acid gradient separations at 60 °C. Excellent stability was demonstrated for particles with C4 and ES-C8 stationary phases at 90 °C using gradients of acetonitrile/aqueous trifluoroacetic acid mobile phase. Superior peak shapes and column efficiencies were shown when using formic acid mobile phases suitable for separations with MS detection. The wide pores of the fused-core particles are especially useful for separating very large proteins with excellent efficiency without the restricted diffusion of particles with smaller pores. Columns of these particles have been found uniquely useful for the high-resolution separation of monoclonal antibody components after reduction and alkylation. This study has shown improved technology for superior reversed-phase separations of intact proteins and other large biomolecules.