Article pubs.acs.org/ac
High Resolution CZE-MS Quantitative Characterization of Intact Biopharmaceutical Proteins: Proteoforms of Interferon-β1 David R. Bush,†,§ Li Zang,‡ Arseniy M. Belov,† Alexander R. Ivanov,† and Barry L. Karger*,† †
Barnett Institute, Northeastern University, 360 Huntington Ave, Boston, Massachusetts 02115, United States Analytical Development Department, Biogen, Cambridge, Massachusetts 02142, United States
‡
S Supporting Information *
ABSTRACT: New and improved methods are required for the enhanced characterization of complex biopharmaceuticals, especially those with charge and glycan heterogeneity. High resolution separation and mass spectrometry (MS) analysis of intact proteoforms can contribute significantly to the characterization of such proteins, many of which are glycoproteins. Here, we report on capillary zone electrophoresis (CZE) coupled via a commercial CESI sheathless interface to an Orbitrap ELITE MS for the intact analysis of recombinant human interferon-β1 (Avonex, rhIFN-β1), a biopharmaceutical with complex glycosylation at a single Nlinked site. Using a cross-linked polyethylenimine coating, column efficiencies between 350,000 and 450,000 plates were produced, allowing separation based on charge and subtle hydrodynamic volume differences. A total of 138 proteoforms were found, and 55 were quantitated. Charge species due to deamidation and sialylation were separated by CZE. Given the high column efficiency, isobaric positional isomers of a single sialic acid on biantennary glycan antennae were resolved. Further, triantennary isomers (antenna on α(1−3) or α(1− 6) arms) were separated and confirmed by exoglycosidase digestion. Proteoforms of the N-terminal cleavage of methionine were detected by precursor molecular weight and top-down ETD and HCD analysis of the reduced protein. Quantitative analysis suggested potential correlations between the methionine loss with the relative amount of the deamidation, as well as the level of deamidation with glycan structure. We demonstrate that high resolution CZE separation of intact glycoprotein species coupled to MS has significant potential for the in-depth characterization and quantitative analysis of biopharmaceutical proteoforms.
P
protein molecules, are not identified or quantitated. Intact MS (MS1 of the undigested glycoprotein) and top-down MS19−25 (fragmentation of the intact protein in the mass spectrometer) can be conducted with minimal sample preparation and can potentially reveal individual proteoform structures. This latter information is important because the specific proteoforms determine protein function, not the proteolytic peptides. Additionally, the correlated presence of multiple modifications on individual proteoforms can potentially lead to additional functional information. The analysis of intact proteoforms is a growing area of interest as mass spectrometers become ever higher resolving and sensitive. The need for high resolution separation prior to MS is essential where there are multiple proteoforms of biopharmaceuticals, some of which are isobaric or differ by only a few Daltons or less and/or are in low abundance. Efforts are being made to develop LC columns suitable for high resolution intact protein separation.26 At the same time, capillary zone electrophoresis (CZE) represents an alternative and comple-
rotein biopharmaceuticals belong to an important and continually growing sector of the pharmaceutical industry. In 2012, almost half of the new drugs approved by the FDA and 5 out of the 10 top-selling drugs were biopharmaceuticals.1 In contrast to synthetic small molecule drugs, biopharmaceuticals are produced from living organisms, leading to complex mixtures of proteoforms.2 Sources of heterogeneity can include, among others, a variety of glycoforms, disulfide bond scrambling, oxidation, deamidation, sequence mutation, and N- and C-terminal alteration.3 Detailed characterization and quantitation of proteoforms is important, as specific modifications can affect the efficacy and safety of the drug product. For example, oxidation has been linked to protein aggregation, which has potential for immunogenicity,3−5 sialic acid content of glycans can improve3,6 or lower efficacy,3,7 IgG fucosylation can limit Fc receptor binding,3,8 and deamidation can reduce activity.3,9−11 Common methods currently employed for analyzing biopharmaceutical products include peptide mapping and glycan profiling. Although the information provided by these bottom-up approaches is valuable, there are limitations: (1) Multiple sample preparation steps can introduce modifications in the protein product.12−18 (2) Specific proteoforms, i.e., the exact combinations of multiple modifications of the intact © XXXX American Chemical Society
Received: July 10, 2015 Accepted: December 7, 2015
A
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
For intact MS1-only scans, the resolution was 120,000 at 400 m/z with 1 microscan and an AGC target of 2 × 106. For the repeatability experiments, the MS resolution was 30,000 at 400 m/z. For top-down MS2, the resolution was 60,000 at 400 m/z. For top-down ETD fragmentation, 5 microscans were acquired with an AGC target of 1 × 106, a reaction time of 25 ms, and a target reagent current of 1 ×106. For all experiments, the ELITE was operated in “protein mode” with a reduced pressure difference between the HCD cell and the C-trap of 10 ×10−10 mTorr. Data Processing. Top-down data analysis was performed using both the high resolution MS1-only scans and MS2 scans as follows: (1) Monoisotopic masses and charge states of a peak of interest were determined from high resolution intact MS1 by averaging the spectra over a peak and exporting a deconvoluted .raw file using the XTRACT algorithm in Xcalibur (v. 3.0.63, Thermo Fisher Scientific). (2) Extracted ion electropherograms (XIEs) of targeted m/z in the low resolution MS1 segment of the top-down scan were used to locate the area over which to average targeted MS2. The resulting averaged spectra were then exported into a .raw file that was then converted to an .mgf file using the MSconvert utility of ProteoWizard (v. 3.0.4472). (3) The monoisotopic m/z and charge state, determined from the intact MS1 scan, were applied to the header of the individual .mgf file. (4) Each .mgf file was analyzed by Byonic (v. 2.0-25, Protein Metrics, San Carlos, CA) against the sequence of recombinant human interferon β1. ETD and HCD spectra were searched with 5 Da precursor mass tolerance, 15 ppm fragment mass tolerance, 1 deamidation (N), and 1 glycan (N) from the comprehensive human N-glycan database (‘Human 360.txt’) supplied with the Byonic.
mentary separation approach based on charge and hydrodynamic volume with the potential for high efficiency.27−29 Although the coupling of CE to MS has developed more slowly than that of LC to MS, recent strides have been made using sheath flow30,31 and sheathless interfacing.32 Because no sample dilution occurs in sheath-flow systems, the sheathless interface is more sensitive.33−35 We36 and others27,37 have demonstrated the potential of high resolution CE coupled to high resolution MS to analyze intact complex glycoproteins. Recently, groups have shown that CE-MS of intact proteins (MS1) can be used to analyze complex protein mixtures31,38−40 but much development remains in this area as well as top-down analysis of glycoproteins. In this work, we apply intact and top-down CZE-MS to Avonex, a commercial recombinant human interferon-β-1a (rhIFN-β1) expressed from CHO cells used to treat multiple sclerosis.41 rhIFN-β1 has known modifications, such as complex glycosylation, deamidation, succinimide, and oxidation.6,27,42,43 Here, CZE-MS is demonstrated to successfully characterize and quantitate individual molecular proteoforms with highresolution accurate mass analysis. We use a positively coated capillary, generating between 350,000 and 450,000 total plates with high migration reproducibility based on the free solution mobility. Although intact protein analysis of 80 proteoforms of rhIFN-β1 by CZE-MS has been reported,27 in the present study, we have significantly improved the separation and analysis of the proteoforms, leading to the determination of 138 individual proteoforms, some of which are identified via selective digestion by several exoglycosidases. Besides elucidating the main glycoforms of Avonex, we have observed intact separation of glycan positional isomers of the biantennary monosialic acid glycan as well as triantennary glycans with the third antenna attached to either the α(1−3) or α(1−6) arm. The spectra of 63 forms were deconvoluted, and 55 have been quantitated. We show that the extent of N-terminal methionine loss (des-1) and deamidation appear to be correlated, as well as the extent of deamidation with the glycan structure. The results of this work demonstrate the significant potential of high resolution CZE-MS for top-down analysis of glycosylated biotherapeutics.
■
RESULTS AND DISCUSSION CZE-MS Characterization. We separated the proteoforms of Avonex using acidic buffer conditions (3% acetic acid, pH 2.5) to protonate the various proteoforms for favorable CZE separation and electrospray ionization. We covalently coated the bare fused silica capillary with cPEI in contrast to dynamic polymeric coatings, such as polybrene or hydroxypropylmethylcellulose, which require frequent recoating and can potentially contaminate the MS source.44−48 Although the coating is positively charged and rhIFN-β1 has up to four sialic acids, the net charge of the various proteoforms is positive at pH 2.5, reducing the possibility of protein interaction with the coating. In addition, given the positively charged surface, the EOF migrated from the negatively charged electrode at the inlet to the porous ESI tip in contact with the positively charged electrode. The net positively charged proteins at pH 2.5 will electrophorese toward the negative electrode at the inlet but at the same time be pulled toward the outlet by the electroosmotic flow. The capillary is thus effectively lengthened with the slowest electrophoretically migrating species (i.e., least positive and largest) eluting first from the capillary and fastest eluting last. The run-to-run repeatability of rhIFN-β1 separation on a cPEI capillary is shown in Figure S-1 for four repeated injections. The RSD of the migration time is 0.5%, and in general, the RSD remained below 2% for run-to-run technical replicates as shown in Table S-1. However, the EOF varies over the number of injections or column to column as seen with the migration time of the neutral marker. To compensate for this, we have determined the free solution mobility, μe, of the various proteoforms (See Supporting Information for calculation of μe from the apparent mobility,
■
MATERIALS AND METHODS All materials, sample preparations, and the cross-linked PEI (cPEI) coating of the etched, porous tip capillaries are described in the Supporting Information. CZE-MS. The cPEI-coated Opti-MS capillary (Sciex, Brea, CA) was 30 μm ID, 150 μm OD, and 100 cm in length. CZE was conducted on a CESI 8000 instrument (Sciex) coupled to an LTQ Orbitrap ELITE Velos Pro mass spectrometer (Thermo Fisher Scientific, San Jose, CA) through an adapter for the Opti-MS sprayer tip that is attached to the commercial nESI source supplied with the ELITE. Except where stated, the BGE was 3% acetic acid at pH 2.5, and the separation voltage was 20 kV in reverse polarity. Between runs, the cPEI capillary was rinsed with 1 M NaCl for 2 min at 50 psi followed by rinsing the capillary and conductive line for 7 and 2 min, respectively, with BGE at 50 psi. Injection conditions were either 1 psi for 10 s or 0.5 psi for 5 s; the specific condition is indicated in the figure captions. After injection, a small plug of BGE was injected into the separation capillary at 0.5 psi for 5 s. The nESI conditions consisted of the sprayer tip positioned approximately 5 mm from the transfer capillary of the ELITE with an electrospray potential of 1.6 kV. B
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 1. Ion density map of CE-MS separation of rhIFN-β1. CE-MS was performed for 1.2 mg/mL of rhIFN-β1 in 50 mM ammonium acetate at pH 4.5 injected with 0.5 psi of pressure for 10s (7.6 nL). The BGE was 3% acetic acid at pH 2.5. (A) The map is focused on the 11+ species and shown as log2(intensity). The insert is a magnified view of the region between 13.8 and 14.6 min and 2181−2224 m/z to show the low abundance N981x0 and N87140 peaks with intensity plotted linearly. (B) Magnified view of the region between 14.8 and 16.75 min and 2021−2073 m/z showing the low abundance N54xx0 forms presented as log2(intensity). Glycans are highlighted with colors corresponding to the number of C
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry Figure 1. continued
LacNAc: 2, red; 3, green; 4, blue; 5, orange; and 6, pink. Glycan nomenclature: N(Hex) (HexNAc) (Fuc) (NeuAc) (NeuGc); modifications are denoted as follows: D, deamidation; S, succinimide; O, oxidation; M, des-1 methionine loss; and SG, sulfated glycan. Peaks deconvoluted within 5 ppm of the theoretical mass are shown in bold and are listed in Table 1.
Figure 2. Extracted ion electropherograms of the N54xxx (2 LacNAc) glycoforms of rhIFN-β1 with and without exoglycosidase treatment. Injection conditions for the different treatments are stated in their respective figure captions of the ion density maps (Figure 1 and Figures S-2−S-4). Each channel corresponds to the top six charge states (z = +11 to +16) of the indicated forms. Possible glycan structures corresponding to the peaks of the extracted signals are outlined with the same color as the peaks.
μapp). Table S-1 lists a number of proteoforms for which the various mobilities have been determined for run to run, day to day, and column to column. As can be seen, the reproducibility of the free solution mobilities of individual proteoforms is excellent, less than 0.6% RSD for different columns over multiple days. Figure 1 is an ion density map showing the various proteoforms to be discussed below. High column efficiency is essential for the separation of closely related proteoforms. Table S-2 provides plate counts for a variety of proteoforms, showing that plate counts are between 350,000 and 450,000. An example of the separation efficiency and the peak symmetry can be seen in Figures 2 and Figure S-1. Such large plate counts lead to the separation of subtly different isobaric proteoforms, as shown below. CZE-MS of rhIFN-β1. N-Glycans produced in CHO cells consist of a biantennary G2F core polysaccharide49 (Scheme S1A), to which are attached, among others, repeating Nacetyllactosamine (LacNAc) extensions and up to four LacNAc antennae, many of which are terminated with sialic acid (NeuAc) with very small amounts of NeuGc.39,45,50 The MS1 analysis of native glycans cannot distinguish the various isobaric hexoses (Hex) from each other or the various isobaric Nacetylhexosamines (HexNAc). We therefore have selected a previously developed notation to describe glycan composi-
tions,51 where the first character denotes an N- or O-glycan, and each number represents the number of Hex, HexNAc, Fuc, NeuAc, and NeuGc units. Examples of annotated structures are shown in Scheme S-1. The ion density map of intact rhIFN-β1 on the cPEI-coated capillary is shown in Figure 1A and an expanded section in Figure 1B with major proteoforms highlighted according to their sialic acid content. In total, 138 proteoforms were found. Sixty three proteoform identifications based on deconvoluted MS1 accurate mass data (5 ppm) are detailed in Table 1, and the remainder were identified based on their relative mass and free solution mobility. Extracted ion electropherograms (EIE) of the major glycoforms with 2−5 LacNAc units are shown in Figures 2 and 3 and Figures S-5 and S-6. As can be seen in Figure 1 and Table 1, the migration time for a given glycan form decreased with the number of sialic acids present. At the same time, for a given number of sialic acids, there were significant differences in migration due to the structure of the glycan on the proteoform from hydrodynamic volume differences. The crystal structure of rhIFN-β152 reveals that the glycosylation site is on the surface of the protein. Given the overall flexibility of glycans, variations in carbohydrate structure can potentially have a significant impact on hydrodynamic volume and thus migration time. For example, D
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
E
Des-1 Succ Ox SG
N→D N→D mixed
Des-1 Succ Ox SG N→D N→D, N→D, N→D, N→D,
N→D N→D (1) Des-1 (1) (1) N→D (2) Des-1 (2) (2)
N→D
MT (min)
17.50, 17.22 17.38 17.11 17.63 17.41 16.32 16.34 16.54 16.70 16.72 16.94 15.93 15.98 15.95 15.93 15.16 15.72 15.74 15.72 15.72 15.39 15.88 15.66 15.33, 15.53, 15.63, 15.83
b
theo. 21309.76 21472.81 21473.79 21781.92 21782.91 22074.01 21942.97 22073.02 22074.01 21942.97 22073.02 22364.12 22233.08 22347.09 22380.11 22444.08 22365.10 22234.06 22348.08 22381.10 22445.06 22510.18 22511.16 22656.19
21309.67 21472.74 21473.70 21781.91 21782.91 22074.10 21942.96 22072.99 22073.92 21942.96 22072.99 22364.14 22233.11 22347.18 22380.14 22444.10 22365.13 22234.08 22348.10 22381.18 22445.14 22510.09 22511.10 22656.23
obs. 4.3 3.2 4.3 0.6 0.3 4.2 0.2 1.4 3.7 0.2 1.5 1.0 1.2 4.1 1.0 1.2 1.4 0.6 1.2 3.5 3.5 3.8 2.8 1.7
error (ppm)
monoisotopic MH+ glycan
N→D mixed, N→D N→D
N98140
N→D
N→D mixed, N→D
N→D, Des-1 N→D
N→D, Des-1 N→D mixed mixed
N→D (1) (1) N→D (2) (2)
PTM
N98130
N87140
N87130
N87120
N76140
N76130
N76110 N76120
N65130
N65120
N65110
MT (min)
14.94, 15.16, 15.34
14.97 14.84 14.77
15.43, 15.52, 15.63
15.16 14.99 14.96
14.73 14.54 14.29 14.11 14.26, 14.42, 14.64 14.1 13.93
14.92, 14.78, 14.72, 14.44 14.27 14.73,
15.25 15.44 15.57 15.77 15.07, 14.93, 14.88, 15.92 15.33,
15.06, 15.28
proteoform theo.
23750.61 23751.59 24041.70 24042.69 24116.72 24406.84 24407.82
23385.48 23255.42 23386.46 23676.57 23677.56 23460.50
22730.23 22729.25 22730.23 22729.25 23020.34 22890.29 23021.33 22656.22 23095.37
22438.16
23750.63 23751.66 24041.74 24042.76 24116.74 24406.79 24407.83
23385.51 23255.36 23386.53 23676.55 23677.58 23460.46
0.8 3.0 1.5 3.2 0.5 2.1 0.3
1.2 2.5 3.0 0.9 0.9 1.8
1.7 0.7 0.5 1.8 3.0 0.9 1.8 5.5 0.5
NA
22730.19 22729.23 22730.24 22729.21 23020.41 22890.31 23021.37 22656.10 23095.35
error (ppm)
obs. NAc
monoisotopic MH+
Modifications: N → D, deamidation; Des-1, N-methionine loss; Ox, oxidation; Succ, succinimide; SG, sulfated glycan. bMT: migration time. Glycoforms with more than one pair of deamidated/ nondeamidated species are numbered in order of migration. “Mixed” indicates that the proteoforms for a given glycoform could not be accurately isolated for quantitation and the theoretical MH+ used in calculating the error was that of the indicated PTM. Multiple migrations times are noted with (1) or (2) where separated at baseline or as multiple MT values where inflection points were detectable but not separated at baseline. cN65110 overlaps slightly in migration and m/z space with the N54120 SG peaks, shown in the ion density map in Figure 1B.
a
N54130
N54220
N54120
N54110
N54100
PTM
N→D, mixed
glycan
proteoform
N34000 N44000
a
Table 1. Summary of Identified Glycoforms and Protein Modifications of Intact rhIFN-β1
Analytical Chemistry Article
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
Figure 3. Extracted ion electropherograms of the N65xxx (3 LacNAc) glycoforms of rhIFN-β1 with and without exoglycosidase treatment. Injection conditions for the different treatments are stated in their respective figure captions of the ion density maps (Figure 1 and Figures S-2−S-4). Each channel corresponds to the top six charge states (z = +11 to +16) of the indicated forms. Glycan structures corresponding to the extracted signals peaks are outlined with the same color as the peaks.
the order: deamidated, nondeamidated, deamidated, and nondeamidated. Because the glycan is biantennary with a single NeuAc, the two pairs of deamidated/nondeamidated species are most likely due to the separation of positional isomers with the sialic acid residing on either the α(1−3) or α(1−6) linked mannose antenna (N54110a and N54110b, Scheme S-1B). Because CHO cells are known to produce predominantly α(2−3) sialic acid linkage,54 the two pairs are not due to isomers of NeuAc attached to a given antenna. Compared to the N54110 forms, the N54120 with 2 sialic acids cannot form positional isomers and thus only two peaks are found. The N54130 glycoforms with three sialic acids, shown in Figure 1B, also produce a positional isomer pair, which along with deamidation, form a total of 4 peaks. Another peak splitting occurs with N65130 (Figure 3). Because each antenna of the trianntenary structure is terminated with a sialic acid (Scheme S-1F), the only possible difference between N65130a and N65130b is attachment of the third antenna to either the α(1−3) or α(1−6) arm. Positional isomer separation by CZE, although observed for glycans,55 has to our knowledge not been shown for intact proteins of 20 kDa or more. Taking the free solution mobility ratios for the above positional isomers, we find that the hydrodynamic differences for N54110 differ by 2.5% and for N65130 by only 0.9% (see Table S-1). The ability to separate such small hydrodynamic volume differences is due to the very high efficiency of the separations (>350,000 plates). Besides glycan proteoforms, 4 PTMs are observed in Table 1. In addition to deamidation, an associated succinimide was
with additional LacNAc units, migration time is seen to decrease due to an increase in hydrodynamic volume (e.g., N54120 > N65120 > N76120, Figures 1−3 and Figure S-3). Glycoforms are generally found in pairs of peaks that are attributed to nondeamidated and singly deamidated species based on accurate mass (the single deamidation occurs at N25, found from peptide mapping). Note that, without separation, it would be very difficult to identify the deamidated peak, which differs by roughly one Da from the nonmodified form by MS. Three less-common glycans were found in Figure 1 and are listed in Table 1. First, a glycan with a second Fuc, N54220, has been observed in CHO cells as a sialyl Lewis-X (SLex) structure.50 The associated deamidated and nondeamidated peaks migrate slightly slower than the N54210 form, as would be expected given the increase in hydrodynamic volume imparted by the additional neutral Fuc. Second, a +80 Da modification from the N54120 form was observed. This addition could be due to sulfation or phosphorylation, but evidence from exoglycosidase digests described later points to a sulfated glycan. Either of the modifications would result in the pronounced decrease in migration time due to the addition of a negative charge. Third, although not commonly observed, the N54130 glycoforms (biantennary with three sialic acids) in Figure 1 are known to exist.53 Some proteoforms are separated into more than two pairs of peaks (see Figures 1B, 2, and 3 and Figures S-4 and S-5). The most prominent example is the N54110 species, where two isobaric pairs of deamidated/nondeamidated peaks are found. On the basis of accurate mass (Table 1), the four peaks are in F
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
quantitation process and the quantitation in the exoglycosidase digests are provided in the Supporting Information. Table S-2 presents the quantitative results; all features with total relative abundance higher than 1.2% of total peak intensity, a total of 55 forms, were quantitated with less than 20% RSD (n = 6). The major species are N54120 and the 2-sialic acid-biantennary structure consisting of the unmodified form plus deamidation and other PTMs, representing 58% of the total of all forms. The next most abundant glycoforms are N65130 (9%) and N54110 (7.5%). These results are in general agreement with published data27 as well as peptide mapping (data not shown). However, Table S-2 provides significantly more structural detail than has previously been documented. Importantly, as many proteoforms consist of multiple modifications (glycans and PTMs), this quantitative information would not be available by bottom-up methods. Correlation of Protein Modifications. From the quantitative data for the intact forms shown in Tables S-2 and S-5, we can examine whether the appearance of a specific modification potentially correlates with one or more other modifications. Consider the N54120 glycoform, the most abundant form present. Examining the deamidation and des-1 modifications, we present in Table 2 and Table S-6 the relative
found based on accurate mass. Oxidation was also observed, the presence of which is known from peptide mapping analysis (data not shown). The fourth detected PTM contained a −131 Da mass difference, which is assigned as an N-terminal methionine loss (Figure 1). To confirm the des-1 structure, we reduced rhIFN-β1 and performed top-down CZE-MS2. As shown in the Supporting Information, ETD and HCD fragmentation gave c- and b-ions, respectively, that validated the des-1 assignment (Figure S-8). It is important to note that a previous CZE-MS study of rhIFN-β127 did not report the Nmethionine loss. As a consequence, several of the glycan structures reported in that paper were misassigned as glycans not commonly found in CHO cells (e.g., N64110 and N64100). When multiple PTMs and numerous glycan structures exist on a complex glycoprotein, structures corresponding to similar molecular masses can exist, potentially leading to misassignment. High resolution separation/mass spectrometry and the ability to conduct MS2 analysis (topdown) is thus important. CZE-MS Separations of Exoglycosidase-Treated rhIFN-β1. To explore more deeply the separation of the glycan isomers of the intact protein, we incubated rhIFN-β1 with sialidase, galactosidase, or both enzymes to selectively remove sialic acid and/or galactose. We believe this to be the first reported application of multiple exoglycosidase treatments to elucidate glycan structure intact on a protein of 25 kDa by CE-MS. The ion density maps for the treated samples are shown in Figures S-2−S-4 and the extracted ion electropherograms of the high abundance glycoforms in Figures 2 and 3 and Figures S-5 and S-6. Consider first the biantennary glycan structures shown in Figure 2. Sialidase and sialidase + galactosidase treatments collapse all forms (N54120, N54110, N54100) into two peaks, i.e., nondeamidated and deamidated. For the galactosidase treatment, the N54120 form remains unchanged as the galactoses are protected by NeuAc; however, as expected, N54110 and N54100 are digested to N44110 and N34100, respectively, because the unprotected galactoses are removed. The possibilities for structures with 3-LacNAc are more numerous, as the potential attachment of a third antenna to either the α(1−3) or α(1−6) mannose or a poly-LacNAc at the end of a biantennary structure becomes possible (see Figure 3). With respect to the triantennary 3-sialic acid (N65130) forms, galactosidase digestion, as expected, does not alter the EIE pattern. However, sialidase + galactosidase digestion leads to two major isobaric peaks (N35100), which is consistent with the positional isomer separation of the α(1−3) and α(1−6) structures. Examining next the 2-sialic acid N65120 structures in the untreated sample, more than 4 isobaric peaks are observed. As shown in Figure 3, treatment with galactosidase yields primarily the triantennary N55120 structure with an unprotected galactose cleaved, and additionally, the biantennary N65120 structure with both galactoses being protected by sialic acid. Further, the galactosidase digest results in 2 pairs of deamidated/nondeamidated peaks from the attachment of the third LacNAc to either the α(1−3) or α(1−6) arm. Thus, the N65120 structure is a complex mixture of positional isomers of tri- and biantennary glycans. The exoglycosidase treatment aids in the elucidation of these complex structures. Relative Quantitation of rhIFN-β1 Proteoforms. We next conducted quantitative analysis of all proteoforms whose MS spectra were deconvoluted. A description of the
Table 2. Relative Abundances of Deamidation and des-1 Methionine Loss for the N54120 Glycoforms of rhIFN-β1 (N→D) Des-1
− +
+
−
31 6.2
60 2.7
percentages of the combination of the 4 proteoforms (deamidated, nondeamidated, des-1 nondeamidated, des-1 deamidated) for the untreated and exoglycosidase-treated samples. It can be seen that the des-1 deamidated proteoform is more abundant than the des-1 nondeamidated form, whereas the deamidated form is less abundant than the nondeamidated (unmodified) form. If we assume that des-1 and deamidation are formed from the unmodified proteoform, then the species with the combined modifications cannot be more abundant than either of the single modified forms until such time that one of the single modified forms is higher in abundance than the unmodified form. This holds true even when the modifications occur sequentially. However, we find that the form with the two modifications is higher in abundance than that of the des-1 form, even though the deamidated form is lower in abundance than the unmodified form (Table 2), suggesting a potential correlation. The reason for this result could either be due to preferential enrichment during downstream batch processing or a synergistic effect between the two modifications. More studies are necessary to further examine this potential correlation. As a potential second correlation, it appears that the size of the glycan influences the extent of deamidation, as the fraction of the deamidated relative to the nondeamidated forms decreases with glycan size (see Table 3). Specifically, for the untreated and galactosidase-treated samples (Table 3 and Table S-7), all glycans of size N54120 and larger have a deamidation fraction of 40% or lower. The N54110 pairs and smaller glycans, on the other hand, are 50% or greater. In fact N54100 is roughly 70% deamidated, whereas N54110 is 53% and N54120 is only 34%. Note also that for a given glycan structure G
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
Analytical Chemistry
■
Table 3. Percent Deamidation for Glycoforms of rhIFN-β1
N44000 N54100 N34000 N54110
modificationsa
(1) (2) des-1 (1) des-1 (2)
N54120 des-1 succ. N54220 N65130c N76130 N76140 N87130 N87140 N98130 N98140
percent deam.
RSDb
50.3 69.7 53.9 53.9 51.6 83.3 79.3 34.3 69.9 22.9 32.1 35.4 35.8 28.3 41.4 33.5 40.4 40.0
11.5 9.2 4.9 8.7 5.2 5.0 3.8 4.0 4.6 5.3 5.3 6.6 6.7 5.5 3.3 6.5 10.3 8.7
ASSOCIATED CONTENT
S Supporting Information *
proteoform glycan
Article
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b03218. Supplemental materials and methods, ion density maps, extracted ion electropherograms, analysis of top-down HCD spectra, rhIFN-β1 isomer glycan structures, and tables with analyses of CZE-MS separations, rhIFN-β1 proteoform resolution, sialidase and galactosidease treatment responses, and deamidation responses (PDF) Table regarding repeatability and reproducibility of μe of rhIFN-β1 proteoforms (XLSX)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Fax: 616-373-2855. Present Address §
Genedata, Inc. 750 Marrett Road, One Cranberry Hill, Lexington, MA 02421
Notes
a
Modifications: des-1, des-1 methionine loss; succ., succinimide; SG, sulfated glycan. Glycoforms with more than one pair of deamidated/ nondeamidated species are numbered in order of migration. bRSD of n = 6. cIncludes both α(1−3) and α(1−6) peaks of N65130. Averages are stated as the percent of the deamidated form of the total of the deamidated and nondeamidated forms.
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We acknowledge Biogen Biopharmaceutical Development Department for their support through a Sponsored Research Agreement and Thermo Fisher Scientific through a technology alliance. We also thank Dr. Marcia Santos, Dr. Jeff Chapman, Dr. Andras Guttman, and Dr. Chitra Ratnayake at Sciex, Dr. Neloni Ranmali Wijeratne at USC, Dr. Rosa Viner at Thermo Fisher Scientific, and Dr. Marshall Bern at ProteinMetrics.
the increase in the extent of deamidation is up to 80% or more when des-1 is present. It should be pointed out that more work is necessary to verify such correlations and to understand their underlying causes, but this type of information from intact quantitative protein analysis presents intriguing possibilities.
■
■
REFERENCES
(1) Lindsley, C. W. ACS Chem. Neurosci. 2013, 4, 905. (2) Smith, L. M.; Kelleher, N. L. Nat. Methods 2013, 10, 186. (3) Eon-Duval, A.; Broly, H.; Gleixner, R. Biotechnol. Prog. 2012, 28, 608. (4) van Beers, M. C.; Sauerborn, M.; Gilli, F.; Brinks, V.; Schellekens, H.; Jiskoot, W. Pharm. Res. 2011, 28, 2393. (5) Vázquez-Rey, M.; Lang, D. A. Biotechnol. Bioeng. 2011, 108, 1494. (6) Dissing-Olesen, L.; Thaysen-Andersen, M.; Meldgaard, M.; Højrup, P.; Finsen, B. J. Pharmacol. Exp. Ther. 2008, 326, 338. (7) Takeuchi, M.; Inoue, N.; Strickland, T. W.; Kubota, M.; Wada, M.; Shimizu, R.; Hoshi, S.; Kozutsumi, H.; Takasaki, S.; Kobata, A. Proc. Natl. Acad. Sci. U. S. A. 1989, 86, 7819. (8) Shinkawa, T.; Nakamura, K.; Yamane, N.; Shoji-Hosaka, E.; Kanda, Y.; Sakurada, M.; Uchida, K.; Anazawa, H.; Satoh, M.; Yamasaki, M.; Hanai, N.; Shitara, K. J. Biol. Chem. 2003, 278, 3466. (9) Cacia, J.; Keck, R.; Presta, L. G.; Frenz, J. Biochemistry 1996, 35, 1897. (10) Vlasak, J.; Bussat, M. C.; Wang, S.; Wagner-Rousset, E.; Schaefer, M.; Klinguer-Hamour, C.; Kirchmeier, M.; Corvaïa, N.; Ionescu, R.; Beck, A. Anal. Biochem. 2009, 392, 145. (11) Huang, L.; Lu, J.; Wroblewski, V. J.; Beals, J. M.; Riggin, R. M. Anal. Chem. 2005, 77, 1432. (12) Boja, E. S.; Fales, H. M. Anal. Chem. 2001, 73, 3576. (13) Li, X.; Cournoyer, J.; Lin, C.; O’Connor, P. J. Am. Soc. Mass Spectrom. 2008, 19, 855. (14) Perdivara, I.; Deterding, L.; Przybylski, M.; Tomer, K. J. Am. Soc. Mass Spectrom. 2010, 21, 1114. (15) Hao, P.; Ren, Y.; Alpert, A. J.; Sze, S. K. Mol. Cell. Proteomics 2011, 10, 1−11. (16) Du, Y.; Wang, F.; May, K.; Xu, W.; Liu, H. Anal. Chem. 2012, 84, 6355.
CONCLUSIONS This paper has demonstrated the power of high resolution CZE separation coupled to high resolution MS for intact protein analysis. High efficiency separation is needed to resolve isobaric forms, e.g., positional glycan isomers, and those with small mass differences, e.g., deamidation. Given the potential for large proteoform combinations due to multiple glycan structures and PTMs, care must be exercised in identification-based intact mass. We used two forms of fragmentation (top-down MS/MS and exoglycosidase digestion) to aid in the elucidation of structures. Of course, proteins with multiple glycan sites would be more complicated. Nevertheless, the combination of high efficiency protein separation coupled to high resolution MS1 and MS2, along with a variety of enzymatic treatments should prove useful in advancing the characterization of intact biopharmaceutical proteins. Ultimately, a combination of intact/top-down and bottom-up (and even middle-down) analyses will provide the most information on the structure and heterogeneity of the protein. The study of the intact and top down MS analysis leads to the potential of finding correlations between individual glycoforms. A recent study showed that the number of C-terminal lysine and pyroglutamic acid moieties on mAbs influenced the size and processing rate of glycans.56 This result, along with the potential correlations shown in this paper, suggest that a detailed study of biopharmaceuticals by intact and top-down MS may lead to new ways of characterizing these complex therapeutics. H
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
Article
Analytical Chemistry
(48) Horka, M.; Dolnik, V. Electrophoresis 2001, 22, 644. (49) Jefferis, R. Biotechnol. Prog. 2005, 21, 11. (50) North, S. J.; Huang, H.-H.; Sundaram, S.; Jang-Lee, J.; Etienne, A. T.; Trollope, A.; Chalabi, S.; Dell, A.; Stanley, P.; Haslam, S. M. J. Biol. Chem. 2010, 285, 5759. (51) Song, T.; Ozcan, S.; Becker, A.; Lebrilla, C. B. Anal. Chem. 2014, 85, 5661. (52) Karpusas, M.; Nolte, M.; Benton, C. B.; Meier, W.; Lipscomb, W. N.; Goelz, S. Proc. Natl. Acad. Sci. U. S. A. 1997, 94, 11813. (53) Rudd, P. M.; Morgan, B. P.; Wormald, M. R.; Harvey, D. J.; van den Berg, C. W.; Davis, S. J.; Ferguson, M. A. J.; Dwek, R. A. J. Biol. Chem. 1997, 272, 7229. (54) Xu, X.; Nagarajan, H.; Lewis, N. E.; Pan, S.; Cai, Z.; Liu, X.; Chen, W.; Xie, M.; Wang, W.; Hammond, S.; Andersen, M. R.; Neff, N.; Passarelli, B.; Koh, W.; Fan, H. C.; Wang, J.; Gui, Y.; Lee, K. H.; Betenbaugh, M. J.; Quake, S. R.; Famili, I.; Palsson, B. O.; Wang, J. Nat. Biotechnol. 2011, 29, 735. (55) Mittermayr, S.; Bones, J.; Doherty, M.; Guttman, A.; Rudd, P. M. J. Proteome Res. 2011, 10, 3820. (56) Yang, J.-M.; Ai, J.; Bao, Y.; Yuan, Z.; Qin, Y.; Xie, Y.-W.; Tao, D.; Fu, D.; Peng, Y. Anal. Biochem. 2014, 448, 82.
(17) Nielsen, M. L.; Vermeulen, M.; Bonaldi, T.; Cox, J.; Moroder, L.; Mann, M. Nat. Methods 2008, 5, 459. (18) Bischoff, R.; Bronsema, K. J.; van de Merbel, N. C. TrAC, Trends Anal. Chem. 2013, 48, 41. (19) Zabrouskov, V.; Han, X.; Welker, E.; Zhai, H.; Lin, C.; van Wijk, K. J.; Scheraga, H. A.; McLafferty, F. W. Biochemistry 2006, 45, 987. (20) Pesavento, J. J.; Kim, Y.-B.; Taylor, G. K.; Kelleher, N. L. J. Am. Chem. Soc. 2004, 126, 3386. (21) Boyne, M. T.; Pesavento, J. J.; Mizzen, C. A.; Kelleher, N. L. J. Proteome Res. 2006, 5, 248. (22) Thomas, C. E.; Kelleher, N. L.; Mizzen, C. A. J. Proteome Res. 2006, 5, 240. (23) Garcia, B. A.; Pesavento, J. J.; Mizzen, C. A.; Kelleher, N. L. Nat. Methods 2007, 4, 487. (24) Pesavento, J. J.; Bullock, C. R.; LeDuc, R. D.; Mizzen, C. A.; Kelleher, N. L. J. Biol. Chem. 2008, 283, 14927. (25) Pesavento, J. J.; Yang, H.; Kelleher, N. L.; Mizzen, C. A. Mol. Cell. Biol. 2008, 28, 468. (26) Sandra, K.; Vandenheede, I.; Sandra, P. Journal of Chromatography A 2014, 1335, 81. (27) Haselberg, R.; de Jong, G. J.; Somsen, G. W. Anal. Chem. 2013, 85, 2289. (28) Lyubarskaya, Y.; Houde, D.; Woodard, J.; Murphy, D.; Mhatre, R. Anal. Biochem. 2006, 348, 24. (29) Wheeler, T. D.; Sun, J. L.; Pleiner, S.; Geier, H.; Dobberthien, P.; Studts, J.; Singh, R.; Fathollahi, B. Anal. Chem. 2014, 86, 5416. (30) Wojcik, R.; Dada, O. O.; Sadilek, M.; Dovichi, N. J. Rapid Commun. Mass Spectrom. 2010, 24, 2554. (31) Sun, L.; Knierman, M. D.; Zhu, G.; Dovichi, N. J. Anal. Chem. 2013, 85, 5989. (32) Moini, M. Anal. Chem. 2007, 79, 4241. (33) Busnel, J.-M.; Schoenmaker, B.; Ramautar, R.; CarrascoPancorbo, A.; Ratnayake, C.; Feitelson, J. S.; Chapman, J. D.; Deelder, A. M.; Mayboroda, O. A. Anal. Chem. 2010, 82, 9476. (34) Heemskerk, A. A. M.; Busnel, J.-M.; Schoenmaker, B.; Derks, R. J. E.; Klychnikov, O.; Hensbergen, P. J.; Deelder, A. M.; Mayboroda, O. A. Anal. Chem. 2012, 84, 4552. (35) Haselberg, R.; Ratnayake, C. K.; de Jong, G. J.; Somsen, G. W. Journal of Chromatography A 2010, 1217, 7605. (36) Thakur, D.; Rejtar, T.; Karger, B. L.; Washburn, N. J.; Bosques, C. J.; Gunay, N. S.; Shriver, Z.; Venkataraman, G. Anal. Chem. 2009, 81, 8900. (37) Taichrib, A.; Pioch, M.; Neususs, C. Electrophoresis 2012, 33, 1356. (38) Han, X.; Wang, Y.; Aslanian, A.; Bern, M.; Lavallée-Adam, M.; Yates, J. R. Anal. Chem. 2014, 86, 11006. (39) Li, Y.; Compton, P. D.; Tran, J. C.; Ntai, I.; Kelleher, N. L. Proteomics 2014, 14, 1158. (40) Zhao, Y.; Sun, L.; Champion, M. M.; Knierman, M. D.; Dovichi, N. J. Anal. Chem. 2014, 86, 4873. (41) Rudick, R. A.; Goodkin, D. E.; Jacobs, L. D.; Cookfair, D. L.; Herndon, R. M.; Richert, J. R.; Salazar, A. M.; Fischer, J. S.; Granger, C. V.; Simon, J. H.; Alam, J. J.; Simonian, N. A.; Campion, M. K.; Bartoszak, D. M.; Bourdette, D. N.; Braiman, J.; Brownscheidle, C. M.; Coats, M. E.; Cohan, S. L.; Dougherty, D. S.; Kinkel, R. P.; Mass, M. K.; Munschauer, F. E.; Priore, R. L.; Whitham, R. H.; et al. Neurology 1997, 49, 358. (42) Orru, S.; Amoresano, A.; Siciliano, R.; Napoleoni, R.; Finocchiaro, O.; Datola, A.; De Luca, E.; Sirna, A.; Pucci, P. Biol. Chem. 2000, 381, 7. (43) Conradt, H. S.; Egge, H.; Peter-Katalinic, J.; Reiser, W.; Siklosi, T.; Schaper, K. J. Biol. Chem. 1987, 262, 14600. (44) Córdova, E.; Gao, J.; Whitesides, G. M. Anal. Chem. 1997, 69, 1370. (45) Haselberg, R.; de Jong, G. J.; Somsen, G. W. J. Sep. Sci. 2009, 32, 2408. (46) Haselberg, R.; de Jong, G. J.; Somsen, G. W. Anal. Chim. Acta 2010, 678, 128. (47) Horvath, J.; Dolník, V. Electrophoresis 2001, 22, 644. I
DOI: 10.1021/acs.analchem.5b03218 Anal. Chem. XXXX, XXX, XXX−XXX
