Introduction

Antibodies are most relevant and indispensable tools for analytical laboratory assays, such as enzyme linked immunosorbent assay (ELISA), Western blot, and immuno-histochemistry [1–3], all of which are applied routinely in numerous laboratories around the world. Antibodies contribute a great share to disease diagnostics [4] with huge market values [5]. In addition, antibodies have become of immense clinical importance as diagnostic biomarkers [4], and with “Personalized Medicine” concepts gaining momentum, antibody-based therapeutics constitute the fastest growing class of medication with increased sales volumes of billions of US$ [6–9]. Obviously, reliance on the functionality of an antibody either as a therapeutic agent or as a bioanalytical reagent is huge and the pitfalls that one might step into when an antibody’s functionality has not been understood in detail have been extensively discussed in high impact journals [5, 10, 11]. To increase reliability of such precious reagents, there is a tremendous demand for antibody characterization, both structurally and functionally.

Since the most specific property of an antibody is its capability to bind to its antigen in a unique fashion via precise paratope-epitope recognition, experimental determination of epitopes (i.e., partial surfaces on the antigen to which an antibody binds) is of utmost importance for antibody characterization. The two most important strategies for epitope mapping either apply methods for precise structural determinations of antigen partial surfaces (X-ray diffraction, NMR) or make use of functional methods that include competition assays (ELISA, biosensors), antigen modification (H/D exchange, chemical modification of side chains), proteolytic or chemical antigen fragmentation, and synthetic peptides [12, 13]. The bottleneck of all available experimental epitope mapping procedures lies in the rather sophisticated, but up to now unavoidable, multi-step in-solution handling procedures [14], leaving an unmet need for rapid and reliable epitope mapping methods [15, 16]. Facilitating in-solution handling is expected to generate a real breakthrough in routine epitope mapping.

Upon its introduction, ion mobility mass spectrometry [17, 18] enjoys vastly growing interest and finds many new applications in studies on biomolecular structures and dynamics thanks to an additional separation dimension in the gas phase according to the ions’ mobilities in a cell filled with a neutral gas [19–22]. Commercial systems have now become available in which such ion mobility drift cells have been flanked by collision cells. The latter enable gas-phase ion reactions and thereby give access to deeper insights into protein structures and into protein–protein complex properties [23–25].

In this paper, we present the development and application of a fast and easy to apply epitope mapping method that identifies the epitope peptide of an antibody of interest in a single experiment. The intact transition epitope mapping (ITEM) procedure makes use of (1) the determining property of an antibody (i.e., its ability to strongly bind to its antigen), (2) the survival of the intact immune-complex when transitioned into the gas phase, (3) ion separation by ion mobility and/or quadrupole filtering, (4) dissociation of the immune complex by collision induced dissociation, and (5) time-of-flight analysis of the complex constituents, all within the mass spectrometer.

Experimental

Preparation of Antibody and Antigen Solutions

Anti-hnRNP-A2/B1 mouse IgG2a (antiRA33; product no. R4653, lot no. 044K4766; Sigma, St. Louis, MO, USA), anti-His-tag mouse IgG1 (product no. MCA1396, batch no. 0309, AbD Serotec, Oxford, UK), and anti-FLAG M2 mouse IgG1 (product no. F3165, lot no. 128H9200; Sigma, St. Louis, MO, USA) monoclonal antibodies and recombinant human TNF protein (rhTNFα) were subjected to buffer exchange using Amicon Ultra centrifugal filters with cutoff 50 K (Millipore Corporation, Ireland). The respective volumes of antibody and antigen stock solutions (as delivered by suppliers) containing 50 μg of antibodies and antigen, were each loaded onto one filter unit. The volumes on the filter units were filled up to 500 μL with 200 mM ammonium acetate buffer (pH 7.1). Then, the units were centrifuged for 10 min at 13,000 rpm. After centrifugation, 430 μL of 200 mM ammonium acetate buffer (pH 7.1) was added on top of the residual volumes above the filters (ca. 70 μL) and centrifugation (10 min at 13,000 rpm) was repeated. Refilling and centrifugation were repeated eight times. After that, filter units were placed upside down into a new vial and the retentates (ca. 50 μL volumes) were collected by centrifugation for 2 min at 4500 rpm. Such re-buffered antibody solutions were directly used for preparation of antigen/peptide-antibody mixtures. Aliquots (ca. 2 μg) were subjected to protein concentration determinations with the fluorescence-based Qubit assay (Invitrogen, Carlsbad, CA, USA). To prepare the Qubit working solution, 1990 μL of Qubit buffer was mixed with 10 μL Qubit reagent. Next, 190 μL of the Qubit working solution was mixed with 10 μL of the three calibration standards (0, 200, and 400 ng/μL). The mixtures were vortexed and incubated for 15 min and after that they were used to calibrate the Qubit 2.0 fluorometer. An antibody solution (ca. 2 μg of antibody) was mixed with the Qubit working solution to reach a final volume of 200 μL, and the mixture was incubated for 15 min. Then, raw fluorescence values were measured and the concentration of the protein in the assay tube was automatically calculated. Antibody solutions were stored at –20 °C for future use.

In-Solution Peptide Mixture Preparation

A peptide mixture (solution 1) was generated by combining 10 μL of each of the following six peptide solutions (peptide concentrations: 0.1 μg/μL, dissolved in 200 mM ammonium acetate, pH 7.1): FLAG peptide, [M + H]+ 1013.39; angiotensin II, [M + H]+ 1046.54; GPI tryptic peptide, [M + H]+ 1142.59; TRIM21 tryptic peptide, [M + H]+ 2098.11 and [M + 2H]2+ 1050.5; substance P, [M + H]+ 1347.74; and RA33 tryptic peptide, [M + H]+ 1633.87 and [M + 2H]2+ 817.44. To 3 μL of an antiFLAG M2 monoclonal antibody solution with a concentration of 1 μg/μL (6.7 μM) in 200 mM ammonium acetate buffer, pH 7.1 (solution 2), was added 1.2 μL of the peptide mixture (solution 1) to yield a molar ratio of the antiFLAG M2 antibody to the FLAG peptide of 1:1. The antibody-peptide mixture (solution 3) was kept at room temperature and was directly used for nanoESI-MS/MS and nanoESI-IMS-MS/MS analysis, respectively.

Proteolysis of Antigen Proteins

The His-RA33 protein (Euroimmun, Luebeck, Germany; 50 μL, 0.48 μg/μL, dissolved in 8 M urea, 1 M sodium chloride, 50 mM sodium phosphate, pH 7.4), was subjected to in-solution digestion with LysC (Roche Diagnostics GmbH, Mannheim, Germany; reconstituted according to the manufacturer’s protocol) using an enzyme to substrate ratio of 1:50 (w/w). Digestion was performed at room temperature overnight and subsequently at 37 °C for 6 h. The proteolytic peptide-containing digestion mixture was desalted with RP-packed tips (ZipTip C18 tips; Millipore, Billerica, MA, USA) loading 5 μL portions onto one tip, which was reconstituted using 50% ACN, and equilibrated using 0.1% TFA solution (pH 1.7). Washing was performed twice using 10 μL of 0.1% TFA solution (pH 1.7) each time and peptides were eluted with 5 μL 80% ACN/0.1% TFA solution (pH 1.7), each. Ten desalted peptide portions (total volume 50 μL) were pooled and 10 μL of 5 M BrCN solution in ACN was added and incubated at 25 °C in the dark for 20 h [26]. Protein G′e (Sigma, St. Louis, MO, USA; 50 μL, 1 μg/μL, dissolved in 100 mM ammonium bicarbonate, pH 8), was subjected to in-solution digestion with trypsin (Promega, Madison, WI, USA, reconstituted according to the manufacturer’s protocol) using an enzyme to substrate ratio of 1:20 (w/w). Digestion was performed at 37 °C for 48 h. The proteolytic peptide-containing digestion mixture (5 μL) was loaded onto an RP-packed tip (ZipTip C18 tips, Millipore, Billerica, MA, USA), which was reconstituted using 50% ACN, and equilibrated using 0.1% TFA solution (pH 1.7). Washing was performed twice using 10 μL of 0.1% TFA solution (pH 1.7) each time, and peptides were eluted with 5 μL of 80% ACN/0.1% TFA solution (pH 1.7) [27]. Peptide mixtures were lyophilized using a SpeedVac concentrator (Martin Christ GmbH, Osterode, Germany), re-solubilized in 10 μL of  200 mM ammonium acetate buffer, pH 7.1, and stored at –20 °C for future use.

Preparation of Peptide-Antibody Mixtures for Epitope Extraction

Synthetic RA33 epitope peptide (MAARPHSIDGRVVEP-NH2; Peptides&Elephants, Potsdam, Germany) and synthetic FLAG peptide (DYKDDDDK; Thermo Fisher Scientific GmbH, Ulm, Germany) were each dissolved in 200 mM ammonium acetate buffer (pH 7.1) to obtain concentrations of 0.01 μg/μL (6.1 μM and 9.8 μM, respectively). AntiRA33 and anti-FLAG M2 monoclonal antibodies with concentrations of 1 μg/μL (6.7 μM each, in 200 mM ammonium acetate buffer, pH 7.1) were mixed with each of the peptide solutions to yield molar ratios of 1:1. To the peptide mixture (10 μL) that derived from LysC/BrCN digestion of His-tag-containing RA33 protein was added 1 μL of synthetic RA33 epitope peptide (0.1 μg/μL) solution. To 1 μL of this peptide mixture were added 3 μL of antiRA33 antibody solution (1 μg/μL). To the peptide solution derived from tryptic digestion of protein G´e (5 μL) were added 5 μL of antiHis-tag antibody solution (1 μg/μL) yielding in a molar ratio of ca. 1:14 between antiHis-tag antibody and His-tag carrying peptide. All antibody-peptide mixtures were prepared at room temperature and directly used for nano-ESI-IMS-MS/MS analysis. Excesses of the prepared mixtures were stored at +4 °C for a maximum 1 wk.

Preparation of Antigen-Antibody Mixtures and Proteolysis of Immune Complexes for Epitope Excision

Five μL of rhTNFα (0. 36 μg/μL; 92 pmol) in 200 mM ammonium acetate buffer (pH 7.1) was mixed with 10 μL of anti-His-tag antibody (0.68 μg/μL; 46 pmol) in 200 mM ammonium acetate buffer (pH 7.1), and the immune complex mixture was incubated overnight at room temperature. Trypsin (Promega, Madison, WI, USA) was first reconstituted in 3 mM HCl with a concentration of 1 μg/μL (stock solution). From this, a working solution with a trypsin concentration of 2 ng/μL was prepared with 200 mM ammonium acetate buffer (pH 7.0). Next, 1 μL of trypsin working solution was added to the immune complex mixture (generating a ratio of 100:1 between rhTNFα and trypsin). After 10 min incubation, this mixture was directly applied for nano-ESI-IMS-MS/MS analysis.

NanoESI-IMS-MS/MS Acquisition Conditions

Nano-ESI capillaries were prepared in-house from borosilicate glass tubes of 1 mm outer diameter and 0.5 mm inner diameters (Sutter Instrument, Novato, CA, USA) using a P-1000 Flaming/Brown Micropipette Puller System (Sutter Instrument). Capillaries were gold-coated using a sputter coater BalTec SCD 004 (Bal-Tech, Balzers, Liechtenstein). For each measurement, 3 μL of antibody-antigen/peptide mixtures was loaded into nano-ESI capillaries using a microloader pipette tip (Eppendorf, Hamburg, Germany). Nano-ESI-IMS-MS/MS measurements were performed on a Synapt G2-S mass spectrometer (Waters MS-Technologies, Manchester, UK) equipped with a traveling-wave ion mobility cell (TW-IMS). Stability of arrival times and gas pressures of the instrument were checked by performing 10 ESI-IMS-MS experiments, five of which were performed on day 1 and 5 others after 2 days’ time. We used for the 10 measurements the RA33 peptide (exp. molecular mass: 1632.879 ± 0.021 Da) and the FLAG peptide (exp. molecular mass: 1012.402 ± 0.012 Da). The mean arrival time of the doubly protonated RA33 peptide was 8.202 ± 0.115 ms and that of the singly protonated FLAG peptide was 13.922 ± 0.090 ms. The IMS gas pressures during the measurement series were 3.256 ± 0.006 mbar and 3.254 ± 0.002 mbar, respectively. The instrumental parameters were optimized as follows: source temperature, 50 °C; source offset, 80 V; trap collision energy, 4 V; trap gas flow, 10 mL/min; helium cell gas flow, 180 mL/min; IMS gas flow, 102 mL/min; wave velocity, 650 m/s; wave amplitude, 40 V. Purge gas was set to 600 L/h. EDC delay coefficient of the instrument was 1.41. Capillary and sample cone voltages were optimized for each measurement and were varied between 1.3–2 kV and 60–150 V, respectively. Transfer collision energy (TCE) was raised from 2 to 220 V in a stepwise manner (20–30 V/step). Mass spectra were acquired in positive-ion mode applying a mass window of m/z 200–10,000. External mass calibration was performed with 1 mg/mL sodium iodide dissolved in an isopropanol/water solution (50:50, v/v). Data acquisition and processing were performed with the MassLynx software ver. 4.1 (Waters MS-Technologies, Manchester, UK) and the DriftScope software ver. 2.4. CorelDraw X4 was used for data visualization [28].

NanoESI-MS/MS Acquisition Conditions for ITEM with Quadrupole Ion Filtering

The instrumental parameters were optimized as follows: source temperature, 50 °C; source offset, 150 V; capillary voltage, 1.8 V; cone voltage, 150 V; trap collision energy, 4 V; and purge gas, 600 L/h. Transfer collision energy (TCE) was set to either 2 or 220 V for low and high fragmenting conditions, respectively. To record peptide ion signals at low TCE, the quadrupole profile was set to “auto” which corresponds to dwelling at 250 m/z for 25% of the scan time, and then using 75% of the scan time to ramp up to 6640 m/z. For suppressing the ions in the low m/z range the quadrupole profile was manually set to: M1 = 4000 with dwell time of 25% and ramp time of 25%; M2 = 5000 with dwell time of 25% and ramp time of 25%; M3 = 6000. All times are given in % of the mass window scan time. Mass spectra were acquired in positive-ion mode applying a mass window of m/z 200–8000. External mass calibration was performed with 1 mg/mL sodium iodide dissolved in an isopropanol/water solution (50:50, v/v). Data acquisition and processing was performed with the MassLynx software ver. 4.1 (Waters MS-Technologies, Manchester, UK). CorelDraw X4 was used for data visualization [28].

Results

Method Development

The ITEM method was developed using synthetic peptides (Table 1) in conjunction with commercially available monoclonal antibodies. In each case, three solutions were prepared and subsequently analyzed. Solution 1 contained just the peptide(s), solution 2 contained the antibody, and solution 3 was a mixture of solutions 1 and 2.

Table 1 Antibody-Binding Peptides
Full size table

Key operation for determining a peptide as an epitope was to compare the electro-sprayed peptides’ and antibodies’ ion abundances and their arrival-times from the mixtures (solution 3) after passing the ions through an ion filtering device, such as an ion mobility drift cell or a quadrupole, and upon exposing all those ions that passed the ion filter (i.e., the unbound peptides, the free antibodies, and the immune complexes) to different collision energies in the subsequently aligned collision cell (transfer cell energy; TCE). Two transfer cell energy conditions were chosen for comparisons: low collision induced dissociation (CID) conditions (2 V TCE), and high CID conditions (220 V or 120 V TCE). With ion mobility separation ion signals were recorded and displayed as arrival-time versus mass-over-charge plots (AToMZ plots), i.e., digital images in which the ions’ different intensities are represented by grey-scaled pixels. Analysis of peptides (solutions 1) and antibodies (solutions 2) alone is not needed for epitope identification but was performed for comparisons and controls.

In the first experiment, the FLAG peptide was dissolved in 200 mM ammonium acetate buffer (solution 1) from which it was electro-sprayed. The protonated peptide ion was allowed to travel through the ion mobility drift cell, and was exposed to 2 V TCE (i.e., conditions that did not cause excessive fragmentation). The resulting mass spectrum showed a singly charged ion signal of the peptide, referred to as unbound peptide (UBP), at m/z 1013.4 and its sodium adduct at m/z 1035.4 (Figure 1a). These FLAG peptide ion signals were observed with arrival times of 13.3 ms in the AToMZ plot (solid line circle in Figure 1b). They were accompanied by a few rather low-intensity fragment ions, forming a diagonal trend line in the AToMZ plot on which all singly charged ions were lined up. When the collision energy in the transfer cell of the mass spectrometer was elevated to 220 V TCE, the ion signals of the intact FLAG peptide completely disappeared because under these conditions the peptide was entirely fragmented. As a result of harsh fragmentation conditions, no resolved ion fragments were observed in the AToMZ plot; instead, fragment ion signals were found in a “cloud” with low m/z values (at the very left of the m/z scale), all with arrival times of their precursors, i.e., ca. 13.3 ms (Figure 1c).

Figure 1
figure 1

NanoESI mass spectra and AToMZ plots from the FLAG epitope peptide. (a), (b) Ion signals from solution 1 at low TCE (2 V), and (c), (d) at high TCE (220 V). (e), (f) Ion signals from FLAG epitope peptide in the presence of antiFLAG M2 antibody (solution 3) at low TCE (2 V), and (g), (h) at high TCE (220 V). Solvent: 200 mM ammonium acetate, pH 7.1. UBP positions are marked with orange solid line circles; CoRP positions with red dotted line circles. Charge states of selected UBPs and CoRPs are given. Trend lines of ions are emphasized by orange and light red lines. For m/z values of selected ion signals see Table 1

Full size image

When the mixture consisting of the anti-FLAG M2 antibody and the synthetic FLAG peptide (solution 3) was electro-sprayed and ions passed through the ion mobility drift cell and were subjected to 2 V TCE, the resulting mass spectrum and AToMZ plot (low mass range) resembled the one from solution 1 at low TCE (Figure 1f). By contrast, when solution 3 was exposed to high collision energy (220 V TCE), a singly charged ion signal at m/z 1013.4 for the protonated FLAG peptide was observed in the AToMZ plot (dotted circle in Figure 1g). Strikingly, the arrival time of the FLAG peptide ion was now 18.8 ms, which corresponds to the arrival time of the multiply charged antibody or its fragments (Figure 2). Of note, the other singly charged ion in this spectrum at m/z 810.40 could not be assigned to any of the ion signals from solution 1 and, therefore, was assumed as being an antibody-derived fragment.

Figure 2
figure 2

NanoESI mass spectrum and AToMZ plot of antiFLAG antibody. (a), (b) Ion signals from solution 2 at high TCE (220 V). Solvent: 200 mM ammonium acetate, pH 7.1. Selected multiply charged ion signals from the intact antibody and its fragments and/or contaminants are labeled (cf. Supplementary Table S1)

Full size image

This later arrival time of the FLAG peptide ion is explained by the formation of the noncovalent immune complex in-solution and, upon transition into the gas phase, its migration through the ion mobility cell as “cargo” of the antibody followed by release of the intact FLAG peptide from the antibody, referred to as complex-released peptide (CoRP). Dissociation of the complex is caused by the high collision energy in the transfer cell. For comparison, electro-spraying the antibody solution (solution 2), multiply charged ion signals were recorded with arrival times of 18.6 ms (Supplementary Table S1). From this result, two main features of CoRPs are deduced that allow distinguishing them from UBPs in the peptide-antibody mixture:

  1. 1.

    CoRPs survive the peptide fragmenting conditions in the transfer cell.

  2. 2.

    CoRPs possess arrival times that match the drift times of the antibody ions.

As the ion mobility arrival-time shift of CoRPs in comparison to UBPs is one of the two important features that ITEM takes into account, the epitope peptide ion to be identified by this method must be fast enough to provide an observable arrival-time difference when comparing UBP and CoRP ion signals in the AToMZ plots. This has been true for the fairly small FLAG peptide (8 amino acid residues in length).

By contrast, the singly charged ion signal of the 15 amino acid residue long RA33 epitope peptide that is recognized by the anti-RA33 antibody [26] migrates with similar arrival times as do the multiply charged ions from the antibody. Yet, the doubly charged peptide ion signal provides the telltale distinctive shift in arrival times in the AToMZ plots. More details on our studies with longer peptides as well as with quadrupole ion filtering or with non-epitope peptide–antibody interactions can be found in sections I–III of the Supplementary Results. In general, in our experiments we observed that the chance that a doubly protonated ion was produced as the most abundant ion by the ESI process – as opposed to the singly protonated ion – increased with the length of the peptide. As only longer peptides with single charges would cause the problem of arival time overlapping with the antibodies’ arrival times, the double-charging effect of “longer” peptides automatically resolves the problem of potential ambiguity (see Supplementary Results, section I).

To test whether the experimental settings for ITEM are effective enough to clearly identify epitopes from samples with high complexity, we generated a peptide mixture by proteolytic digestion of full-length proteins. The peptide mixture (solution 1) was obtained by digesting the His-RA33 protein with LysC through which rather long peptides were created. These were further cleaved chemically by BrCN into smaller peptides. Although this procedure produced many peptides with ion signals in the mass range between m/z 500 and 1200 (ca. 40 ion signals were recorded with adequate intensities; cf. Supplementary Figure S1a), the “native” epitope peptide was not among them. This observation is consistent with previous findings [26] that place the His-RA33 protein into the group of difficult to digest proteins. Therefore, the synthetic RA33 epitope peptide, resembling the partial amino acid sequence aa78-92 of the His-RA33 protein (underlined partial sequence in Supplementary Figure S1b), was spiked into the mixture (solution 1).

Upon electro-spraying solution 1 and after passing all ions through the ion mobility cell and when 2 V TCE was applied, the RA33 epitope peptide was observed as doubly charged UBP ion with m/z 817.4 in the ESI-MS spectrum of the peptide mixture (Figure 3 and Table 1). In the AToMZ plots of solutions 1 and/or 3, the doubly charged UBP ion signal of the RA33 epitope peptide was observed along the drift-time trend line of the doubly charged peptides with an arrival time of 8.2 ms (Figure 3b).

Figure 3
figure 3

NanoESI mass spectra and AToMZ plots from the His-RA33 peptides from in-solution digestion in the presence of antiRA33 antibody. (a), (b) Ion signals from solution 3 at low TCE (2 V), and (c), (d) at high TCE (220 V). Solvent: 200 mM ammonium acetate, pH 7.1. UBP positions are marked with orange solid line circles, CoRP positions with red dotted line circles. Charge states of selected UBPs and CoRPs are given. Contrast enhanced region is boxed; magnification factor is given. For m/z values of selected ion signals see Table 1

Full size image

By contrast, when ion mobility-separated ions from solution 3 were exposed to 220 V TCE, the only singly and doubly charged ion signals in the AToMZ plot were CoRPs from the RA33 epitope peptide (dotted circles in Figure 3c). Consistent with previous observations, the arrival time of the doubly charged RA33 CoRP (i.e., the epitope peptide) was now shifted to 18.7 ms and corresponded with the arrival time of the ions from the antibody (Figure 4).

Figure 4
figure 4

NanoESI mass spectrum and AToMZ plot of antiRA33 antibody. (a), (b) Ion signals from solution 2 at high TCE (220 V). Solvent: 200 mM ammonium acetate, pH 7.1. Selected multiply charged ion signals from the intact antibody and its fragments and/or contaminants are labeled (cf. Supplementary Table S1)

Full size image

Of note, the accuracy of determination of the RA33 epitope peptide in this experiment was 18 ppm (the experimentally determined mass of the epitope peptide was 1633.8654, which agrees well with the theoretical mass of 1633.8359). This finding highlights the ITEM feature of determining the epitope peptide mass with isotopic resolution and high mass accuracy because at low m/z range monoisotopic masses of CoRPs can be determined very precisely. Conversely, in the high m/z range of the mass spectrum of solution 3, three ion series were observed for the antibody and the immune complexes with molecular masses of 150,019.3 ± 72.5 Da, 151,640.3 ± 42.3 Da, and 153,273.1 ± 48.7 Da with average mass differences of 1621.0 Da and 1632.9 Da, respectively (Figure 5).

Figure 5
figure 5

NanoESI mass spectrum of the mixture of antiRA33 antibody and RA33 epitope peptide at 2 V TCE. Solvent: 200 mM ammonium acetate, pH 7.1. The 1:2 (antibody:epitope) mixture shows ion signals belonging to the antibody (series “a”) and to the immune complexes in which one epitope peptide (series “b”) and two epitope peptides (series “c”) were bound, respectively. Selected charge states and m/z values are indicated

Full size image

Hence, when using the experimentally determined masses of the immune complexes to calculate the epitope peptide mass directly from the ESI mass spectra, the accuracy with which the RA33 epitope peptide (Mr 1633.8359) was determined was 7871 ppm and 573 ppm, respectively. Obviously, subtracting experimentally determined masses of the antibody from the masses of the immune complexes leads to rather imprecise determination of the epitope peptide mass, independent of the mass spectrometer performance. The primary reason for the poor resolution of around 160 (FWHM) at this mass range is due to the mean peak widths of the antibody/immune complex ion signals, which were between 25 and 30 Th, caused mostly by antibody heterogeneity.

The observations on AToMZ plot resolutions and inspections of m/z traces prompted us to investigate alternative quadrupole ion filtering as an approach within the ITEM method (see Supplementary Results, section II). In contrast to quadrupole ion filtering, which functions like an off-switch with respect to transmission of low m/z ions from the ESI source, ion mobility filtering spreads out the ions that are produced in the ESI source on a traveling time scale and allows adding an extra dimension to the mass spectrum. This feature enables to display the data as two-dimensional plots, termed AToMZ plots, and affords instant identification of an ion signal as deriving from a UBP or a CoRP without ambiguity.

Application Examples

Epitope extraction

We first applied the ITEM method to identify the His-tag epitope from recombinant protein G´e by epitope extraction. Protein G´e is a commercial protein [27] from which 25 assignable peptide ion signals with adequate intensities were produced by tryptic digestion (cf. Supplementary Figure S2a), yielding 100% sequence coverage. The complex peptide mixture containing the His-tag carrying peptide served as solution 1. Solution 3 consisted of solution 1 to which an antiHis-tag antibody (solution 2) was added. When solution 3 was electrosprayed and ions were ion mobility separated and exposed to 2 V TCE, the His-tag peptide was recorded in the AToMZ plot as doubly and triply charged ion signals at m/z 884.9 and m/z 590.3 with arrival times of 8.2 ms and 5.8 ms, respectively (solid line circles in Figure 6; Table 1).

Figure 6
figure 6

NanoESI mass spectra and AToMZ plots from the protein G´e peptides derived from in-solution digestion. (a), (b) Ion signals from digest peptide mixture in the presence of antiHis-tag antibody (solution 3) at low TCE (2 V), and (c), (d) at high TCE (120 V). Solvent: 200 mM ammonium acetate, pH 7.1. UBP positions are marked with orange solid line circles and yellow squares; CoRP positions with red dotted line circles and dark yellow dotted squares. The blue arrow points to a non-fragmenting peptide. Charge states of selected UBPs and CoRPs are given. Trend lines of ions are emphasized by orange and light red lines. Contrast-enhanced region is boxed; magnification factor is given. For m/z values of selected ion signals see Table 1

Full size image

Electrospraying solution 3, separating all ions in the ion mobility cell, and exposing the ions to 120 V TCE afforded in the AToMZ plot singly charged ion signals for the His-tag peptide at m/z 1768.9 (Table 1) but now with 21.5 ms arrival time (dashed line circle in Figure 6c), which matched the arrival time of the antibody ions and their respective fragment ions (Supplementary Figure S3). This characteristic change in position in the AToMZ plots qualified the His-tag peptide ions as CoRPs, i.e., identified the peptide as epitope (underlined partial sequence in Supplementary Figure S2b).

Interestingly, another peptide ion signal that survived fragmentation when 120 V TCE was applied was found at m/z 1535.6. It was assigned to the partial sequence aa17-31 from protein G´e (cf. Supplementary Figure S2). However, the position of this ion signal in the AToMZ plots did not match the arrival time of the antibody, indicating that it was not a CoRP but survived fragmentation, at least partially, because of its stability. Of note, in all spectra from the antiHis-tag antibody (solution 2), rather strong ion signals within the mass range of m/z 2000 and 4000 at arrival times above 12 ms were observed from which molecular masses of ca. 38 kDa were calculated (Supplementary Table S1). As these ion signals were present in the AToMZ plots at both low TCE (data not shown) and high TCE (Supplementary Figure S3), they were assigned as unknown contaminants. Owing to their multiple charge states, they did not interfere with assignment of epitope peptide ions as the latter typically were doubly or singly protonated.

Even more intriguing was that in addition to the His-tag carrying epitope peptide, two more closely spaced, singly charged peptide ion signals with m/z 2144.9 and 2161.9 and with arrival times of 24.2 ms each, were recorded in the AToMZ plots when solution 3 was investigated with high CID conditions (120 V TCE). As the arrival times of these two ions matched those of the antibody (dashed squares in Figure 6c), their positions in the AToMZ plots marked them as CoRPs as well. The ion signal at m/z 2161.9 was assigned to the identical partial amino acid sequences aa77-95 and/or aa147-165 of protein G´e (dashed lines in Supplementary Figure S2b), and the ion signal at m/z 2144.9 was regarded as a deamination product therefrom (Table 1). The latter was most likely produced by CID from the N-terminal Q residue [29] of this peptide. Reexamination of the AToMZ plot from solution 3 with 2 V TCE revealed the presence of the respective doubly charged peptide ion at m/z 1081.4 with 9.8 ms arrival time (solid line rectangle in Figure 6b; the ion signal is marked with “#” in Supplementary Figure S2a). From X-ray crystallography data it is known that protein G´e binds strongly to Fc parts of antibodies [30] and the region of protein G´e that makes contact with the antibodies encompasses the partial amino acid sequences aa77-95 and/or aa147-165 [31]. Hence, our ITEM result stands in full agreement with crystal structure analyses. More specific investigations on antibody-protein G binding can be found in the Supplementary Results (section III).

Epitope Excision

For epitope excision, we first prepared solution 3 by mixing rhTNFα (as a full-length antigen) with the antiHis-tag antibody and subjected this immune complex-containing mixture to enzymatic digestion with trypsin. All solvent mixing was done using 200 mM ammonium acetate buffer (pH 7). Then, following the standard ITEM procedure, we electrosprayed the resulting peptide and protein mixture as a whole and exposed the ion mobility-separated ions first to low (2 V TCE) and second to high (200 V TCE) fragmentation conditions. When solution 3 was subjected to 2 V TCE (Figure 7b, c, and Supplementary Figure S4), a few peptide ion signals, which were formed upon digestion of the antigen with trypsin, were observed in the low m/z range of the mass spectrum, all with less than 15 ms arrival times. The His-tag-carrying peptide was found as a doubly charged ion signal at m/z 884.89 with an arrival time of 7.95 ms. The presence of the unbound His-tag peptide in this mixture can be explained by the fact that rhTNFα is a trimer, potentially leaving some of the His-tag peptides inaccessible to the antibody.

Figure 7
figure 7

NanoESI mass spectra and AToMZ plots from the rhTNFα peptides derived from epitope excision. (a), (b) Ion signals from digest peptide mixture in the presence of the antiHis-tag antibody (solution 3) at low TCE (2 V), and (c), (d) at high TCE (120 V). Solvent: 200 mM ammonium acetate, pH 7.1. UBP positions are marked with orange solid line circles; CoRP positions with red dotted line circles. Charge states of selected UBPs and CoRPs are given. Trend line of ions is emphasized by an orange line. Contrast enhanced region is boxed; magnification factor is given. For m/z values of selected ion signals see Table 1

Full size image

When the ion mobility-separated ions from solution 3 were exposed to high TCE voltage, unbound peptides were fragmented and the only nonfragmented ion signal that could be attributed to the antigen was that of the doubly charged His-tag-carrying peptide at m/z 884.88. But now its arrival time of 21.4 ms matched that of the antibody (Figure 7c). Again, the antiHis-Tag antibody was prone to degradation when exposed to high TCE voltage, producing many multiply charged and some singly charged fragment ions (cf. Supplementary Table S1). The fact that the peptide with m/z 884.88 survived high TCE conditions and possessed an arrival time like that of the anti-His-tag antibody proves that this peptide is a CoRP and, consequently, the epitope peptide.

Discussion

We have developed the ITEM method to provide a facile and routinely applicable procedure to rapidly determine antigen-derived epitopes of an antibody of interest using both epitope extraction and epitope excision. These two epitope mapping methods have proven to be well applicable to identify assembled (conformational, discontinuous) as well as sequential (linear, continuous) epitopes [14, 32] by mass spectrometry, and typically afford peptide masses as read-outs by which the epitopes are defined. It has been proven that higher-order structured peptides are able to bind to antibodies. Peptides of a certain length (ca. 6–10 amino acids and more) are known to be able to adopt higher order structures such as alpha-helices in solution [19, 33, 34], and peptides are known to be able to bind to antibodies (or other binding partners) via a mechanism that is called “induced fit” [35]. For instance, with Western blotting as well as with so-called peptide chips, one is able to identify epitopes [26, 36] independent of the fact that by applying these methods the structures that are bound by the antibodies are taken out of the context of their highly structured “natively folded” antigen proteins. Despite not knowing the precise structure of the bound amino acids in an immune complex, the experimentally determined epitope is sufficiently encoded by the peptide that contains the partial structure which is recognized by the antibody. Hence, the identified CoRPs reflect the epitope peptides with high accuracy.

Our ITEM approach, in fact, follows the same step-wise experimental sequence as was developed for in-solution epitope mapping methods, yet without immobilization of the antibody and/or the immune complex. Instead, after in-solution formation of the specific antibody–peptide complex, a transition of this complex into the gas phase is induced together with all other constituents in that mixture. Sample preparation for ITEM has, thus, been minimized to the generation of antigen/peptide-antibody mixtures using volatile buffers. The demand on purity for both the antigen/peptide solution (solution 1) and the antibody solution (solution 2) is rather moderate. With nanospray capillaries, the consumed volume in one experiment is ca. 3–5 μL and the amount of required peptide and antibody is in the low pmol range for each. In our hands, as long as a suitable electrospray was obtained from the mixture (solution 3) the epitope mapping experiment was successful.

Already during the desolvation step, electrospray conditions can be selected such that weakly bound molecules and nonspecifically bound “stickier” peptides are efficiently removed from the specific immune complex while antibody–epitope interactions are not broken; they are typically very strong with KD values of around 10 nM [32]. Therefore, the traditional “washing step” that is implemented to remove nonspecific or unbound mixture components occurs predominantly in the transition step from solution to the gas phase in the source and to a lesser extent within the ion mobility drift cell. “Elution” of the epitope peptide from the immune complex is achieved by efficient collision induced dissociation in the transfer cell (or the collision cell). Simultaneously, peptides that are not bound to the antibody (UBPs) but that passed the ion filter are efficiently fragmented under the applied CID conditions. After these simultaneous gas-phase processes, which are equivalent to “washing” and “elution”, only a few peptides survived, resulting in just a few ion signals (i.e., spots in an AToMZ plot) which, therefore, are easy to interpret.

It turned out that the harshness of both the electrospray and the gas-phase dissociation conditions were to be fine-tuned to match the individual stabilities of all constituents of the sprayed mixture, including those of the antibodies [37], the peptides, and the contaminants in order to obtain good signal intensities (signal to noise values) and to simultaneously avoid generation of too many Ab-derived fragment ion signals. In cases when milder collision energy regimens in the transfer cell were selected for successful ITEM analysis (e.g., 120 V as opposed to 220 V), some fairly stable UBPs may survive unfragmented. However, they can be easily identified by comparing their arrival times with the arrival times of the antibody ion signals (from solution 3) and by the absence of arrival time shifts.

Of note, since CoRP ions are not produced by the ESI process in the source region of the mass spectrometer, they receive/retain their protons (charges) from the antibody–epitope complexes during CID in the transfer cell. Thus, in principle it may be possible that neutral CoRPs are obtained by CID, which would not be detectable in the mass spectrometer and, therefore, ITEM might be limited. Yet a number of studies have shown that via CID an asymmetric distribution of charges on the dissociated components occurs [38] in which the smaller of the two complex partners takes the relatively larger numbers of protons upon dissociation. This stands in agreement with our observations that showed that the CoRPs were either singly or doubly protonated upon dissociation of the immune complex. Also, the fact that ionization of CoRPs is not taking place in the ESI source makes them free from so-called “matrix effects” where peptides from a more or less complex mixture are competing for the available protons under the respective solution ionization conditions. Therefore, ion yields which are observed for peptides that are ionized from complex peptide mixtures are not decisive for the abundance by which the epitope peptide ion signals will be observed by the ITEM method. For comparison, high resolution structure analysis techniques, such as X-ray diffraction [39, 40] or NMR [41] of immune complexes, also suffer from limitations like high material demands, time-consuming preparations, and molecular size restrictions [42].

In addition to alternative ion filtering methods like ion mobility and quadrupole separation that can be applied with ITEM, there are several mass spectrometric methods that have found application for gas-phase fragmentation of protein–protein complexes. Black-body infrared dissociation (BIRD) seems one potential alternative for fragmentation of noncovalent peptide–protein bonds. Yet BIRD is so far not routinely available with commercial mass spectrometers [43, 44]. The closest alternative to CID breakage of noncovalent bonds in the gas phase seems to be surface induced dissociation (SID) [45, 46]. However, it was reported that in SID experiments, charge distribution is more symmetric and charges are distributed proportionally to the masses of dissociated constituents [47].

In summary, ITEM is very powerful and allows the direct identification of an epitope as in-solution handling is reduced to mixing of antigen/epitope peptide and antibody solutions. Since suitable mass spectrometry equipment has become available, our ITEM method seems to be easily adaptable by mass spectrometry laboratories around the world.