Journal of Proteomics 212 (2020) 103572
Contents lists available at ScienceDirect
Journal of Proteomics
journal homepage: www.elsevier.com/locate/jprot
Intact Transition Epitope Mapping - Thermodynamic Weak-force Order T
(ITEM - TWO)
Bright D. Danquaha, Yelena Yefremovaa, Kwabena F.M. Opunib, Claudia Röwera, Cornelia Koya,
Michael O. Glockera,⁎
a Proteome Center Rostock, University Medicine Rostock, Rostock, Germany
b School of Pharmacy, University of Ghana, Legon, Ghana
A R T I C L E I N F O A B S T R A C T
Keywords: We have developed an electrospray mass spectrometry method which is capable to determine antibody affinity
Mass spectrometric epitope mapping in a gas phase experiment. A solution with the immune complex is electrosprayed and multiply charged ions are
Antibody specificity determination translated into the gas phase. Then, the intact immune-complex ions are separated from unbound peptide ions.
Antibody affinity determination Increasing the voltage difference in a collision cell results in collision induced dissociation of the immune-
Nano-electrospray mass spectrometry complex by which bound peptide ions are released. When analyzing a peptide mixture, measuring the mass of
Native MS
Q-ToF analyzer the complex-released peptide ions identifies which of the peptides contains the epitope. A step-wise increase in
the collision cell voltage difference changes the intensity ratios of the surviving immune complex ions, the
released peptide ions, and the antibody ions. From all the ions´ normalized intensity ratios are deduced the
thermodynamic quasi equilibrium dissociation constants (K #Dm0g) from which are calculated the apparent gas
G#
phase Gibbs energies of activation over temperature ( m0g ). The order of the apparent gas phase dissociation
T
constants of four antibody – epitope peptide pairs matched well with those obtained from in-solution mea-
surements. The determined gas phase values for antibody affinities are independent from the source of the
investigated peptides and from the applied instrument. Data are available via ProteomeXchange with identifier
PXD016024.
Significance: ITEM - TWO enables rapid epitope mapping and determination of apparent dissociation energies of
immune complexes with minimal in-solution handling. Mixing of antibody and antigen peptide solutions in-
itiates immune complex formation in solution. Epitope binding strengths are determined in the gas phase after
electrospraying the antibody / antigen peptide mixtures and mass spectrometric analysis of immune complexes
under different collision induced dissociation conditions. Since the order of binding strengths in the gas phase is
the same as that in solution, ITEM – TWO characterizes two most important antibody properties, specificity and
affinity.
1. Introduction reagents have demonstrated their various applications in disease diag-
nostics to improve early disease detection [4,5], e.g. as part of a per-
Antibodies are widely used as powerful tools in immunologic assays, sonalized medicine concept [6]. Also, the past few decades of antibody
such as enzyme linked immunosorbent assays (ELISA), Western blot research have provided enormous evidence of the versatility of anti-
analysis, and immuno-histochemistry [1–3]. Additionally, bio- bodies as therapeutic agents to fight cancer, autoimmune diseases, and
technology development and testing of new generations of antibody infections [7–11]. Antibodies are, thus, an extremely important class of
Abbreviations: Nano-ESI, nano-electrospray ionization; ToF, time of flight; CID, collision induced dissociation; KD, dissociation constant; ΔG0, standard Gibbs free
energy change; SAW, surface acoustic wave; G#Dm0g , apparent Gibbs energy of activation of the abundance weighted mean charge state of multiply charged and
accelerated antibody-epitope peptide complex ions in the gas phase, without external energy contributions; K #Dm0g , apparent thermodynamic quasi equilibrium
dissociation constant of mean charge state of multiply charged and accelerated antibody-epitope peptide complex ions in the gas phase, without external energy
contributions; #, apparent; m, mean of charge states; O, without external energy contributions; g, gas phase
⁎ Corresponding author at: Proteome Center Rostock, University Rostock Medical Center, and Natural Science Faculty, University of Rostock, Schillingallee 69,
18057 Rostock, Germany.
E-mail address: michael.glocker@med.uni-rostock.de (M.O. Glocker).
https://doi.org/10.1016/j.jprot.2019.103572
Received 5 July 2019; Received in revised form 25 October 2019; Accepted 27 October 2019
Available online 01 November 2019
1874-3919/ © 2019 Elsevier B.V. All rights reserved.
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
biomolecules and continue to play an increasingly important role in 2. Materials and methods
both, life-sciences and medicine [12–14]. It is imperative, therefore, to
both structurally and functionally characterize antibodies in order to 2.1. Antibodies, peptides, and chemicals
gain in-depth understanding of their functions; a task that although self-
explanatory is not easily fulfilled [15,16]. Rabbit antiTRIM21 antibody (polyclonal 52 kDa Ro/SSA antibody; sc-
First of all, characterizing antibodies requires the precise determina- 20960 lot: F0503), raised against amino acids 141–280 of TRIM21 (52 kDa
tion of epitopes on the antigens they bind to. So far, among all developed Ro/SSA) of human origin, was obtained from Santa Cruz Biotechnology,
methods for epitope identification, mass spectrometry provides one of the Inc. (Heidelberg, Germany). Mouse monoclonal antiFLAG M2 antibody
greatest potentials to become the method of choice [17]. Mass spectro- (product code F 1804) was obtained from Sigma-Aldrich (Steinheim,
metry has proven to be appropriate for studying non-covalent complexes, Germany). Mouse monoclonal antiHis-tag antibody (MCA 1396; Batch no.
such as antibody-peptide complexes [18], since it allows the transfer of 0309) was supplied by AbD Serotec (Oxford, United Kingdom). Rituximab
intact non-covalent complexes into the gas phase through gentle deso- (Batch number: H0013) was produced by Roche Ltd. (Welwyn Garden
lvation [19–21]. With our earlier developed epitope mapping method, City, UK). The RA33 epitope peptide (MAARPHSIDGRVVEP-NH2,
“Intact Transition Epitope Mapping (ITEM)” [22], now renamed “Intact 1632.86Da), GPI epitope peptide (ALKPYSPGGPR, 1141.62Da),
Transition Epitope Mapping – One-step Non-covalent force Exploitation TRIM21A epitope peptide (LQELEKDEREQLRILGE, 2097.11Da),
(ITEM-ONE)”, epitopes are identified by the accurate masses of the com- TRIM21B epitope peptide (LQPLEKDEREQLRILGE, 2065.12Da),
plex released peptides (CoRPs). And with “Intact Transition Epitope TRIM21C epitope peptide (LQELEKDEPEQLRILGE, 2038.06Da), and His-
Mapping – Targeted High-Energy Rupture of Extracted Epitopes (ITEM- tag epitope peptide (GPSIVHRKSFHHHHHH, 1948.98Da) were synthe-
THREE)” we subsequently introduced a method which identifies epitope sized by Peptides & Elephants GmbH (Potsdam, Germany). The synthetic
peptides by partial amino acid sequencing of the CoRPs [23] in a rapid and FLAG epitope peptide (DYKDDDDK, 1012.40Da; article no. 020015) was
accurate fashion. In both methods neither antibodies nor immune com- obtained from ThermoFisher Scientific GmbH (Ulm, Germany).
plexes are immobilized. Instead, both methods make use of ion mobility Angiotensin II (DRVYIHPF, 1045.53Da) was purchased from Sigma-
mass spectrometry for ion separation coupled with gas phase dissociation Aldrich (Steinheim, Germany). Actin, cytoplasmic 1 recombinant human
of immune complexes in the mass spectrometer's collision cell. The speed protein (rhßactin) was purchased from GenWay Biotech (Catalog no. 10-
by which ITEM-ONE [22] or ITEM-THREE [23] can be conducted, as well 288-23014F, San Diego, California, USA). Bovine Serum Albumin (BSA)
as the low required sample amounts, outperforms many of the other epi- was from Carl Roth GmbH + Co. KG (Karlsruhe, Germany). Trypsin (Lot
tope mapping methods [17,24]. number 070289) was obtained from Promega Corp. (Madison, Wisconsin,
Unlike for epitope mapping, up to now there exists no mass spec- USA). Ammonium acetate was from Fluka Chemika (Buchs, Switzerland).
trometric method that has reached equal acceptance for determining 16-Mercaptohexadecanoic acid, phosphate buffered saline powder, etha-
thermodynamic properties of antibody - antigen interactions as com- nolamine, N-Hydroxysuccinimide, 2- [N-Morpholino] ethanesulfonic acid,
pared to in-solution investigations with e.g. calorimetric or spectro- 1-ethyl-3-[3-dimethylaminopropyl] carbodiimide (EDC), and sodium
scopic techniques [25]. In some cases, mass spectrometry-based acetate were obtained from Sigma-Aldrich (Steinheim, Germany). DTT
methods have been applied as read-outs for determining in-solution KD was from Serva Electrophoresis GmbH (Heidelberg, Germany), and io-
values of protein - protein complexes [26–28]. Introducing correction doacetamide from Bio-Rad Laboratories GmbH (Munich, Germany).
factors for differences of surface activities of analytes in the electro-
spray droplet as well as for additional gas phase ion suppression effects 2.2. Preparation of peptide solutions (solutions 1)
[29] yielded satisfactory correlation with results from conventional
methods. Only very recently have we reported the development of a The synthetic His-tag peptide (120 μg) was dissolved in 200mM
mass spectrometry-based method by which thermodynamic properties ammonium acetate buffer (240 μl), pH 7. A final concentration of His-
of protein - protein interactions could be determined in the gas phase tag peptide of 15 μM was then generated after measuring the actual
[30]. With this method, the same order of stabilities was found as that peptide concentration of the stock solution by diluting with 200mM
which was inferred from in-solution analyses. ammonium acetate buffer (solution 1). To prepare solutions 1 to con-
Here, we report on the combination of our gas-phase epitope tain TRIM21A peptide, TRIM21B peptide, or TRIM21C peptide, re-
mapping methods with our mass spectrometry-based procedure by spective amounts of lyophilized peptide powder were first dissolved in
which quasi-thermodynamic information was obtained on desolvated 200mM ammonium acetate buffer, pH 7, and then diluted further with
and multiply charged and accelerated protein - protein complex ions in 200mM ammonium acetate buffer, pH 7, to obtain concentrations of
the gas phase. This mass spectrometry-based method expands the above 9.5 μM, 14.5 μM, and 5 μM respectively. A mixture of seven synthetic
mentioned analytical concept to antibody - peptide complex analyses peptides (GPI peptide, FLAG peptide, Angiotensin II, TRIM21A peptide,
and is named “Intact Transition Epitope Mapping - Thermodynamic TRIM21B peptide, TRIM21C peptide, and RA33 peptide) was prepared
Weak-force Order (ITEM-TWO)” since it enables the determination of (solution 1) by firstly, dissolving lyophilized peptides individually in
an important antibody feature, affinity, i.e. binding strength, in a 200mM ammonium acetate buffer, pH 7, to obtain peptide concentra-
straightforward gas phase experiment. Upon formation of specific im- tions of ca. 0.30 μg/μl, each. Next, respective volumes of each peptide
mune complexes in solution, transfer of all components into the gas solution were mixed and diluted with 200mM ammonium acetate
phase, and separation of unbound peptide ions (UBPs) from immune buffer to obtain final concentrations of 10 μM of each peptide in the
complex ions using the mass spectrometer's ion filtering capability, we mixture. For tryptic digestion and peptide mixture preparation from rhβ
dissociated the antibody – peptide immune complex in the gas phase, actin to generate a His-tag peptide see Supplement (section V.1). For
through which CoRPs are generated, which in effect are the epitope peptide concentration determinations see Supplement (section V.2).
peptides. Using three different antibodies we identified their respective
epitope peptides (CoRPs) from epitope peptide-containing solutions. In 2.3. Preparation of antibody solutions (solutions 2)
the same experiment we simultaneously obtained the thermodynamic
quasi equilibrium dissociation constants (K #Dm0g) from which were cal- The antiTRIM 21, the antiFLAG M2, and the antiHis-tag antibodies
culated the respective apparent activation energies over temperature were all obtained dissolved in PBS buffer, pH 7.4, with protein con-
G#
( m0g ) of the gas phase dissociation processes of each of the immune centrations of 200 μg/ml, 1 μg/μl and 1 μg/μl respectively. They each
T
complexes. The order of the apparent gas phase dissociation constants were rebuffered into 200mM ammonium acetate buffer, pH 7, by
matched well with those obtained from in-solution measurements. loading the respective volumes of the antibody stock solutions, con-
taining ca. 20 μg of the respective antibodies into centrifugal filters
2
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
(Microcon with cutoff 50 K; Merck Millipore Ltd., Tullagreen, repeated for peptide solutions, for peptide mixture solutions (solutions
Carrigtwohill Co CORK, Ireland) and addition to 200mM ammonium 1) and each antibody solution (solutions 2). Negative control experi-
acetate buffer, pH 7, to reach a volume of 500 μl, each. The solutions ments were performed using a mixture (solution 3) of Rituximab anti-
were then centrifuged at 13,000 rpm for 10min. To the retentates on body as non-specific antibody (solution 2) and the seven peptide-con-
the filters (ca. 30 μl, each) were added respective volumes of 200mM taining mixture (solution 1). All the mass spectra were acquired in
ammonium acetate buffer, pH 7, to reach again a total volume of 500 μl, positive-ion mode applying a mass window of m/z 500–8000. Data
each. After discarding the filtrates, the solutions were centrifuged were acquired and processed using MassLynx software version 4.0
again. This centrifugation / re-filling procedure was repeated eight (Waters MS-Technologies, Manchester, UK). The Savitzky-Golay
times for each antibody. All the filtrates were discarded. After the last method was used for smoothing in 10 cycles with a window of 20 for
spinning, the filter units were inverted into new vials and centrifuged at the high mass range and in 3 cycles with a window of 5 for the low mass
4500 rpm for 5min to collect the retentates, ca. 20–30 μl, each. range. For detailed description of mass spectrometric data analysis see
Antibody solutions (solutions 2) were either directly used for prepara- Supplement (section I). All mass spectrometry raw data (and because of
tion of antibody-epitope peptide complexes after their concentrations formal necessity a “search engine results file”) have been deposited in
had been determined (see Supplement, section V.2), or were stored at the PRIDE database [31].
4 °C and consumed within a maximum period of one week. Typical
concentrations of the antibodies were between 0.3 μg/μl and 0.9 μg/μl. 2.6. Determinations of in-solution KDs and ΔGs values
2.4. Preparation of antibody - peptide mixtures (solutions 3) Determination of KD and ΔGs values was performed using the K5 S-
Sens® SAW biosensor (SAW Instruments, Bonn, Germany) as described
Solution 3 which contained His-tag epitope peptide and antiHis-tag in the Supplement (section V.6) following the rules of in-solution KD
antibody in a molar ratio of 2.2 : 1 was prepared by adding 2 μl of a value determinations (see Supplement, section II) [30].
15 μM His-tag epitope peptide solution (solution 1) to 10 μl of 0.2 μg/μl
antiHis-tag antibody solution (solution 2). To prepare 15 μl of a solution 3. Results
that contained the TRIM21A peptide and the antiTRIM21 antibody in a
molar ratio of 2.2 : 1, 2.4 μl of TRIM21A peptide (9.5 μM; solution 1) 3.1. Characterization of starting materials and antibody - peptide complex
was added to 5 μl of antiTRIM21 antibody (0.3 μg/μl; solution 2) and formation in-solution
the mixture (solution 3) was filled up with 200mM ammonium acetate
buffer, pH 7. For obtaining solution 3 to consist of TRIM21B peptide Before generating immune complexes in solution, the starting ma-
and antiTRIM21 antibody, 1.55 μl of a TRIM21B peptide solution terials, i.e. the epitope peptides and the antibodies, were separately
(14.5 μM; solution 1) was added to 5 μl of antiTRIM21 antibody solu- characterized by mass spectrometry. The mass spectrum of the His-tag
tion (0.3 μg/μl; solution 2) and was filled up to 15 μl with 200mM peptide (Supplement Fig. S1A) or those of either the TRIM21A,
ammonium acetate buffer, pH 7. Similarly, 2.9 μl of a TRIM21C peptide TRIM21B, and TRIM21C peptides (data not shown) in 200mM am-
solution (5 μM; solution 1) was added to 3.2 μl of antiTRIM21 antibody monium acetate buffer, pH 7, showed doubly and triply protonated
solution (0.3 μg/μl; solution 2) and filled up to 9.6 μl with 200mM peptide ions with good intensities (solutions 1) in the low m/z range of
ammonium acetate buffer, pH 7. AntiFLAG antibody solution (5 μl; the mass spectrum, i.e. below m/z 2000. The experimental masses
0.2 μg/μl; 1.33 μM; solution 2) was mixed with 1.5 μl of the peptide matched well with the calculated masses (Supplement Table S1).
mixture of seven peptides (solution 1) to obtain solution 3 with molar Likewise, the mass spectrum of the mixture of the seven synthetic
ratios of 2.2 : 1 of peptide to antibody. For the application example, peptides (solution 1) displayed well resolved ion signals with good in-
solution 3 was prepared by mixing 1 μl of peptide mixture solution from tensities for all the peptides in the mixture (Supplement Fig. S1B). The
tryptic digest of rhβ actin (peptide concentration was 0.05 μg/μl; so- ion signals´ abundance differences are caused by different ionization
lution 1) and 10 μl of antiHis-tag antibody (antibody concentration was efficiencies of the peptides and by suppression effects during the ESI
0.2 μg/μl; solution 2). All the complex-containing solutions (solutions process despite their equimolar concentrations (10 μM, each).
3) were left to stand at room temperature for at least 1 h after pre- Upon electrospraying an antiHis-tag antibody containing solution
paration, and were used directly for nanoESI-MS analysis. (solution 2; 200mM ammonium acetate buffer, pH 7), we observed in
the high m/z range, i.e. above m/z 5000, a series of multiply charged
2.5. Data acquisition by NanoESI-MS ions of the antibody with charge states between 21+ and 26+
(Fig. 1A). The multiply charged ions with highest intensities were found
Nano-ESI-MS measurements were performed using a Q-TOF 2 in- at the 24+ charge state. Similarly, an ion series between 22+ to 27+
strument and a Synapt G2S instrument, respectively. Both instruments ions with highest intensities at 24+/25+ was observed for the anti-
are from Waters MS-Technologies (Manchester, UK). For calibration TRIM21 antibody (solutions 2; Supplement Figs. S2A and S3A). Ion
and instrument settings see Supplement (sections V.3 and V.4). For each signals were broader in case of the polyclonal antibody as compared to
measurement, 3–4 μl of antibody – peptide complex-containing solution those obtained for a monoclonal antibody. Likewise, in the mass spec-
(solution 3) was loaded into nano-ESI capillaries using a microloader trum of the antiFLAG antibody ion signals were recorded in the high
pipette tip (Eppendorf, Hamburg, Germany). The pulling and gold mass region for the 20+ up to the 26+ ion (solution 2, Fig. 2A). And
sputter coating of nano-ESI capillaries is described in the Supplement for Rituximab, which served as a negative control in this study, ion
(section V.5). The pressure conditions within the mass spectrometer signals of the 22+ ion up to the 27+ ion were recorded when elec-
were adjusted to preserve non-covalent interactions as previously de- trosprayed from aqueous ammonium acetate solution (solution 2,
scribed [18,30]. Next, the transmission of unbound peptides was Supplement Fig. S4A). From the multiply charged antibody ion series
blocked by setting the quadrupole profile to only allow transfer of ions the average molecular masses of all the antibodies were determined
with high m/z values (above m/z 2000 in the Q-ToF 2 instrument and (Supplement Table S2).
above m/z 5000 in the Synapt G2S instrument) and the collision gas The respective volumes of solutions 1 and 2 were combined to
was switched on and collision voltage difference in the collision cell generate mixtures of antibodies and epitope peptides in molar ratios of
was increased in a stepwise manner (20–30 V/step) to cause dissocia- 1: 2.2 (solutions 3) and to initiate immune complex formation in so-
tion of the antibody – epitope peptide complexes. To identify the ions lution. Accordingly, when the mixture of antiHis-tag antibody in pre-
which were due to background noise and probable contaminants from sence of the His-tag peptide was electrosprayed, the mass spectrum in
the antibody and peptide solutions, the above procedure was also the high mass region showed three narrowly spaced multiply charged
3
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
Fig. 1. Nano-ESI mass spectra of antiHis-tag anti-
body (solution 2) and antiHis-tag antibody - epitope
peptide complex (solution 3). A: AntiHis-tag anti-
body. B: AntiHis-tag antibody – His-tag peptide
complex. Selected ion signals are labeled with m/z
values and charge states of ion signals are given.
Open squares, open circles and filled circles re-
present ion series belonging to antibody: epitope
peptide ratios 1 : 0, 1 : 1, and 1 : 2, respectively. For
molecular masses of antibodies and antibody: epi-
tope peptide complexes see Supplement Table S2.
Solvent: 200mM ammonium acetate, pH 7.
ions at each charge state causing series of triplet ion signals (Fig. 1B). respectively (Supplement Table S2). From these masses the respective
The molecular masses which were determined therefrom were ion series were identified as representing antiHis-tag antibody, antiHis-
151,037 ± 103 Da, 152,992 ± 74 Da, and 154,985 ± 76 Da, tag antibody with one bound His-tag peptide, and antiHis-tag antibody
Fig. 2. Nano-ESI mass spectra of antiFLAG antibody
(solution 2) and antiFLAG antibody - epitope peptide
complex (solution 3). A: AntiFLAG antibody. B:
AntiFLAG antibody – FLAG peptide complex.
Selected ion signals are labeled with m/z values and
charge states of ion signals are given. Open squares,
open circles and filled circles represent ion series
belonging to antibody: epitope peptide ratios 1 : 0, 1
: 1, and 1 : 2, respectively. For molecular masses of
antibodies and antibody: epitope peptide complexes
see Supplement Table S2. Solvent: 200mM ammo-
nium acetate, pH 7.
4
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
Fig. 3. Nano-ESI mass spectra of antiHis-tag anti-
body – His-tag epitope peptide complex (solution 3)
after blocking of transmission of ions below m/z
2000. Different collision cell voltage differences
(ΔCV) were applied to dissociate the complex. A: 4 V.
B: 40 V. C: 80 V. D: 120 V. Charge states and m/z
values for selected ion signals are given for the
complex ions (right ion series) and for the released
His-tag epitope peptide ions (left ion series). The
insert zoom shows the CoRP ion signal with isotopic
resolution. For normalized ion signal intensities see
Supplement Table S3. Solvent: 200mM ammonium
acetate, pH 7.0.
with two bound His-tag peptides. The ion signals of free antiHis-tag molecular masses of Rituximab remained the same whether determined
antibody were very little and barely visible above the noise. from solution 2 or from solution 3 (Supplement Table S2), which was
By contrast, the analysis of the antiTRIM21 antibody - TRIM21A peptide expected.
complex or that of the antiTRIM21 antibody - TRIM21B peptide complex in Since upon successful immune complex formation in solution the
the high mass region showed only multiply charged ion series of broad ion respective antibody – peptide complex ions were detectable in high
signals in which antibody-peptide complex ion signals dominated but yields by nano-electrospray mass spectrometry, consistent with our
overlap with free antibody ion signals could not be excluded (Supplement previous reports [22,23], we conclude that the antibodies maintained
Figs. S2B and S3B). The respective molecular masses which were de- their in-solution binding properties while transitioned into the gas
termined from the ion signal series showed large mass increases upon phase.
complex formation. However, the mass differences were not in
agreement with the calculated masses of the immune complexes 3.2. Gas phase dissociation of antibody – epitope peptide complexes and
(Supplement Table S2). Evidently, the Q-TOF 2 instrument was not able to determination of activation energies and dissociation constants
resolve the ion series and therefore, an attempt to determine the epitope
peptide masses by using the experimental mass differences between im- Albeit analyses of the peptide mixtures (solutions 1) and antibody
mune complex ions and antibody ions in the high m/z range was rather solutions (solutions 2) alone are not required for epitope identification
inaccurate. The rather broad peak widths of the antibody-peptide complex or binding strength determination, they were investigated here as
ions may be due to heterogeneity in the antibody composition and, in ad- controls. Having successfully prepared four different antibody – epitope
dition, to e.g. ammoniation, sodiation, and/or potassiation. Of note, the peptide complexes in solution (solutions 3) and after having success-
analysis in the high m/z range of the antiTRIM21 antibody - TRIM21C fully transitioned the respective immune complexes into the gas phase,
peptide mixture presented only multiply charged ion series of broad ion we studied their gas phase dissociation behaviors one after the other. To
signals like the ones of the antiTRIM21 antibody alone (from solution 2). No do so, we first blocked transmission of low mass components, i.e. below
increase in mass could be found, and hence, no immune complex formation m/z 2000, by switching the settings of the quadrupole ion filter ac-
had taken place (data not shown), which is consistent with previous results cordingly. As a result, the full range mass spectrum (between m/z 500
[23,24]. and m/z 8000) of solution 3, which consisted of the antiHis-tag anti-
Interestingly, the mixture of antiFLAG antibody and the mixture of body and the His-tag peptide, showed no ion signals of any unbound
seven peptides produced a spectrum in the high mass range in which peptides but displayed in the high mass range only the multiply charged
again three narrowly spaced multiply charged ions at each charge state ion signals of the remaining free antibody together with those of the
were found as triplets (Fig. 2B). Their corresponding molecular masses respective immune complexes (Fig. 3A); the latter were dominating in
were determined to be 148,730 ± 92 Da, 149,799 ± 45 Da, and intensities. Upon switching on the collision gas and by increasing the
150,785 ± 61 Da, respectively (Supplement Table S2). The respective collision cell voltage difference (ΔCV) in a step-wise manner (20–30 V /
ion series were identified as representing antiFLAG antibody, antiFLAG step), dissociation of the immune complexes caused appearances of
antibody with one bound FLAG peptide, and antiFLAG antibody with complex-released peptides (CoRPs) in the low mass ranges of the
two bound FLAG peptides. Finally, when solution 3 with Rituximab as spectra with increasing yields (Fig. 3B–D). Increasing the collision cell
control antibody in the presence of seven peptides was electrosprayed, voltage differences from 4 V to 200 V, we observed appearances and
the spectrum in the high mass range showed again an ion series of incremental rises of doubly and triply charged ion signals in the lower
multiply charged ions with charge states between 22+ and 27+ and m/z range. Their m/z values corresponded precisely to the calculated
the 24+ protonated ion with highest intensity (Supplement Fig. S4B). values of the respective epitope peptides. At the same time the ion
Since neither the m/z values nor the intensities of the multiply charged signal intensities in the high m/z region diminished.
ions nor the ion signal widths were significantly altered as compared to The same behaviors under comparable experimental conditions
those from the mass spectrum of Rituximab from solution 2, we con- were observed for the immune complexes consisting of antiTRIM21
cluded that none of the peptides had bound to Rituximab. The average antibody and TRIM21A peptide (Supplement Fig. S5), for the immune
5
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
Fig. 4. Nano-ESI mass spectra of antiFLAG antibody -
FLAG peptide complex (solution 3) after blocking of
transmission of ions below m/z 2000. Different col-
lision cell voltage differences (ΔCV) were applied to
dissociate the complex. A: 20 V. B: 70 V. C: 120 V. D:
150 V. Charge states and m/z values for selected ion
signals are given for the complexes (right ion series)
and for the released FLAG peptide (left ion series).
The insert zoom shows the CoRP ion signal with
isotopic resolution. For normalized ion signal in-
tensities see Supplement Table S4. Solvent: 200mM
ammonium acetate, pH 7.
complexes consisting of antiTRIM21 antibody and TRIM21B peptide The ΔCV50 value for the antiHis-tag antibody – His-tag peptide
(Supplement Fig. S6), and for the immune complexes consisting of complex dissociation was 42 V, that for the antiTRIM21 antibody -
antiFLAG antibody and FLAG peptide (Fig. 4). Note that the antiFLAG TRIM21A peptide was 98 V, that for the antiTRIM21 antibody -
antibody complex only released the FLAG peptide as CoRP despite the TRIM21B peptide was 86 V, and that for the antiFLAG antibody - FLAG
presence of six other peptides in solution. Mass accuracies of 31 ppm peptide was 89 V. Apart from determining the ΔCV50 values, the norm
were obtained for the His-tag peptide, 20 ppm for the FLAG peptide, products vs ΔCV plots also enabled to identify the “steep parts” of the
66 ppm for the TRIM21A peptide, and 68 ppm for the TRIM21B peptide. dissociation reaction dependences, i.e. those “energy regimes” in which
Confirmation of specific binding between antibody and its re- the greatest effects on dissociation yields were observed. After calcu-
spective epitope peptide was concluded by the results summarized lating the equations of the lines that followed the points on the steep
above in conjunction with the control experiments in which no binding parts of the sigmoidal shaped curves, the linear rising values around
between antibody and non-epitope peptides was first predicted and ΔCV50 plus or minus the respective dx (see Supplement, section I) were
then observed under the respective experimental conditions. No release used to generate the respective y values which were then used to cal-
of CoRPs was observed with the antiTRIM21 antibody / TRIM21C #culate ( Gmg) values.
peptide pair (data not shown) as well as with the mixture of Rituximab T G#mg
and seven peptides (Supplement Fig. S7), since the peptide mixture did Plotting vs ΔCV showed the linear dependencies of the apparentT
not contain a peptide that resembled the Rituximab epitope. When Gibbs energies of activation over temperature on collision cell voltage
solution 3 of Rituximab plus seven peptides was investigated at ele- differences (ΔCV) for the investigated immune complex ions within the
vated collision cell voltage differences, the only ion signals which were above determined ranges (Fig. 5B). And, by linear extrapolations of the
observed in the low mass range (below m/z 2000) were multiply lines of these plots to ΔCV=0V, the respective intercepts with the y-
charged fragment ions which originated from ruptured antibody. axis could be determined mathematically. The intercepts provide the
To determine apparent Gibbs energies of activation for the dis- apparent Gibbs energies of activation for the dissociation of immune
sociation of immune complexes in the gas phase, G#g , we utilize the complexes in the gas phase over temperature with negligible additional#
apexes of Gaussian fits of multiply charged and accelerated antibody - Genergy influences, termed mOg . Additional energy influences that
epitope peptide complex ions and apexes of Gaussian fits of CoRP ions, Tneeded to be “subtracted” from the experimentally obtained values
leading to K #Dm0g as well as to G#mg. More precisely (see Supplement, arise from charging and acceleration of the immune complex ions.
section I), our method uses averaged intensities of the immune complex Subsequently, we calculated apparent gas phase dissociation constants
ions, of the free antibody, and of the respective CoRPs (Supplement (K #Dm0g) for “neutral and resting” immune complexes in the gas phase
Tables S3–S6) from triplicate measurements. From them, after nor- G#
malization, ion intensity ratios are determined and used to reflect the from the experimentally obtained
mOg values using the van't Hoff
T
relative amounts of the various species in the gas phase at each applied G#equation, as is outlined in the Supplement (section I). Both, the mOg
collision cell voltage difference. Plotting normalized ion intensities of Tvalues and the K #Dm0g values showed that the antiTRIM21 antibody –products (norm products), i.e. CoRPs and free antibodies, as a function TRIM21A peptide complex was the strongest of the four antibody –
of stepwise elevated collision cell voltage differences (ΔCV) yielded peptide complexes, followed by the antiTRIM21 antibody – TRIM21B
sigmoidal shaped courses to which Boltzmann fitting provided the best peptide complex. Next was the antiFLAG antibody – FLAG peptide
matching curves (Fig. 5A). complex, followed by the weakest of the investigated immune
6
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
complexes, the antiHis-tag antibody – His-tag epitope peptide complex
(Table 1). The TRIM21C peptide did not form a stable immune complex
with the antiTRIM21 antibody.
3.3. Determination of in-solution free energies and dissociation constants of
antibody – epitope peptide complexes
To ascertain the resemblance or otherwise of gas phase activation
energies and pseudo dissociation constants with respective in-solution
values, we performed real time in-solution bioaffinity analyses of the
studied antibody - peptide complexes. To this purpose, peptides
TRIM21A, TRIM21B, TRIM21C, FLAG, and His-tag were one after the
other immobilized onto the gold surface of a biosensor chip. The bio-
sensor determinations of binding affinities were performed in triplicate
or in pentaplicate independent measurements using the respective an-
tibody. Again, Rituximab solution was applied as non-binding antibody
control. The interaction between the antiTRIM21 antibody and the
TRIM21A peptide showed the strongest binding (Supplement Table S7),
providing the lowest KD value (36.0 ± 10.4 nM). The antiTRIM21
antibody had a stronger affinity to the TRIM21A peptide than to the
TRIM21B peptide. Next in the order were the interactions between the
antiFLAG antibody with the FLAG peptide, and the immune complex
with the weakest interaction was that from the antiHis-tag antibody
with the His-tag peptide. At last, the antiTRIM21 antibody showed ra-
ther weak in-solution binding to the TRIM21C peptide, with a KD value
of 113.6 ± 7.0 nM. This rather weak binding may be caused by a
secondary epitope which was not present in the gas phase experiments
because of the peptide's helical structure in the gas phase, whereas a
linearized peptide structure may be generated only during the surface-
based experiment [23,24,32]. Unspecific adsorption onto the self-as-
sembled monolayer (SAM) was rather little, i.e. below 10% in all in-
vestigated cases (Supplement Table S8). It should be noted that the
antibody – peptide complexes´ binding strength order that was de-
Fig. 5. Courses of normalized ion intensities and of apparent dissociation ac- termined by the in-solution measurements was the same as that which
tivation energies over temperature as functions of collision cell voltage differ- was obtained by our gas phase analyses of apparent dissociation acti-
ences (ΔCV). A: Normalized intensities of complex released product ions (norm vation energies (Table 1).
products). Each curve was fitted using a Boltzmann function. Each data point is
the mean of three independent measurements and standard deviations are 3.4. Simultaneous gas phase epitope mapping and binding strength
shown by vertical bars (cf. Supplement Tables S3 to S6). B: Apparent dis-
# determination: an application example
sociation activation energies over temperature ( Gmg ) and linear extrapolations
T
to ΔCV=0. FLAG peptide (green filled triangles and lines), TRIM21A peptide As an application example for the ITEM-TWO procedure, we sub-
(red filled squares and lines), TRIM21B peptide (orange filled circles and lines), jected recombinant human β actin (rhβ actin) which contained a C-
and His-tag peptide (black filled squares and lines). The intercepts with the y- terminal His-tag to simultaneous epitope mapping and binding strength
G#
axis give m0g values (see Table 1). determination; including control measurements. First, starting mate-
T
rials were characterized. The nanoESI mass spectrum of the peptides
obtained from a tryptic digest (solution 1) was recorded using a
WATERS Synapt G2S instrument and showed the presence of mainly
Table 1 doubly protonated peptide ions appearing below m/z 2000
Free energies and dissociation constants of antibody – peptide complexes. (Supplement Fig. S8). In all, 46 peptide ions were identified by both,
peptide antibody gas phase in-solution nanoESI MS and MALDI MS analyses (Supplement Table S9). Amino
acid sequence coverage was 96%. Upon electrospraying the antiHis-tag
G# # −3m0g KDm0g x 10 ∆Gs KD s ± stdv antibody alone (solution 2), multiply charged ions with charge states
T [Ø]a [kJ/mol] [nM] between 22+ and 27+ were observed above m/z of 5000 with highest
[J/mol⁎K] intensity for the 25+ charge state (Supplement Fig. S9A). The mass
TRIM21 A antiTRIM21 46.1 3.9 42.04 36.0 ± 10.4 spectrum of the peptide – antibody mixture (solution 3) showed the
TRIM21 B antiTRIM21 42.9 5.7 41.52 44.5 ± 6.7 presence of unbound peptides in the low mass range (below m/z 2000)
FLAG antiFLAG 39.3 8.9 40.39 70.6 ± 8.0 together with immune complex ions above m/z 5000 (Supplement Fig.
His-tagb antiHis-tag 28.0 34.5 39.35 107.7 ± 9.2
His-tagc antiHis-tag 26.0 43.8 n.d.e n.d.e S9B).
TRIM21 C antiTRIM21 n.b.d n.b.d 11.61 113.6 ± 7.0 Second, ITEM-TWO analyses were carried out on a WATERS Synapt
G2S instrument. When the quadrupole was set to block transmission of
a dimensionless. ions below m/z 5000 and a low collision cell voltage difference of 4 V
b synthetic peptide. was applied, the unresolved antibody and immune complex signals
c from digest of rhβ actin. remained at high m/z values (Fig. 6A). Even before increasing the
d n.b. no binding.
e collision cell voltage difference, only one ion signal appeared at m/zn.d. not determined. 695.82 with low intensity. This ion signal's m/z value matched to that
of the monoisotopic mass of the isotopically resolved doubly charged
7
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
Fig. 6. Nano-ESI mass spectra of antiHis-tag anti-
body – His-tag epitope peptide from tryptic digest of
rhβ actin (solution 3) after blocking of transmission
of ions below m/z 5000. Different collision cell vol-
tage differences (ΔCV) were applied to dissociate the
complex. A: 4 V. B: 20 V. C: 50 V. D: 150 V. E: 200 V.
Charge states and m/z values for selected ion signals
are given for the complex ions (right ion series) and
for the multiply charged fragment ions of the
antiHis-tag antibody as well as the released His-tag
epitope peptide ions (left ion series). The insert zoom
shows the CoRP ion signal with isotopic resolution.
For normalized ion signal intensities see Supplement
Table S10. Solvent: 200mM ammonium acetate,
pH 7.0.
peptide of the C-terminal tryptic His-tag-containing peptide of rhβ actin difference between the transition and the ground states' internal en-
(Supplement Table S9); mass accuracy was 7 ppm. As was observed ergies (“activation energy”) forms the activation barrier which needs to
with the other antibody - peptide complexes during ITEM-TWO method be overcome for each respective reaction. Obviously, the reaction rate
development, upon increasing the collision cell voltage difference fur- and, hence, the activation energy, becomes the limiting factor for ir-
ther in a stepwise manner, the intensities of the released singly charged reversible dissociation of immune complexes in the gas phase. Non-
(m/z 1390.63) and doubly charged (m/z 695.82) His-tag epitope pep- covalent complex dissociation under CID conditions requires an energy
tide ions increased and the multiply charged ion signals of unresolved input above a critical threshold and proceeds irreversibly but com-
antibody and immune complex reduced in intensities (Fig. 6B–E). Ap- paratively slowly. An ‘effective reaction rate’ results in energy regimes
plying a collision cell voltage difference of above 80 V and higher did slightly above the threshold that was insufficient to reach completeness
not only result in the release of the His-tag epitope peptide ions from within the timeframe of each single measurement. So, despite the de
the immune complex but also in subsequent fragmentation of the an- facto irreversible character of the dissociation reaction, an apparent
tibody. This became evident by the appearance of multiply charged ion equilibrium exists; the similarity of which to solution behavior of the
signals at around m/z 2000. Nevertheless, from the plot of norm h1 vs respective antibody – epitope peptide complexes was exploited.
ΔCV, a rise of the epitope peptide ions was observed to again follow a Our ITEM-TWOmethod in fact combines gas phase epitope mapping
sigmoidal shape curve (Fig. 7A) when looking at collision cell voltage [22,23] with binding strength assessments of antibody – epitope pep-
differences below 80 V. The ΔCV value at which 50% of the antiHis-tag tide interactions in the gas phase [17,30]. Stability of non-covalent
epitope peptide were released from the immune complex (ΔCV50) was forces in the gas phase as well as structural properties of desolvated
reached at a collision cell voltage difference of 22 V. This ΔCV50 value proteins have been investigated using so-called “native MS” conditions
was ca. two times lower than that which was measured when analyzing [33–35]. A growing body of evidence has been accumulated over time
the release of the synthetic His-tag epitope peptide from the respective which demonstrates that higher order protein structures may be
immune complex on the Q-ToF 2 instrument (see above). Despite these maintained nearly intact in the gas phase – at least for a certain period
instrument-related differences, the apparent gas phase dissociation ac- of time [36–38], directing to the conclusion that intact non-covalent
G#
tivation energy over temperature ( m0g ) with negligible external en- complexes can exist in the gas phase as well, despite absence of sol-
T
ergy influences (at ΔCV=0V) was 26.0 Jmol−1 K−1, as was de- vation spheres [19,39,40].
termined by extrapolation of the experimentally accessible data, Also of importance, once the temperature of the immune complex in
(Fig. 7B; Table 1). The corresponding apparent gas phase dissociation the collision cell has been determined, the apparent activation energies
constant calculated thereof was 43.8× 10−3. Each of the values were and dissociation constants can be determined directly from the gas
rather close to the ones which were obtained from the interaction phase experiments. The temperature of ions in the ESI source is grossly
analyses of the synthetic His-tag epitope peptide with the antiHis-tag estimated to adopt about room temperature [41], but might increase
antibody which were determined using a Q-ToF 2 instrument. Those during CID. According to estimations from peptide fragmentation re-
G#−1 −1 m0g −3 K # actions [42,43], and considering the greater numbers of degrees ofvalues were 28.0 Jmol K for and 34.5×10 for
T Dm0g freedom in larger protein molecules, temperatures of immune complex
(Table 1). ions in a collision cell of a typical Q-ToF instrument are assumed to be
just marginally rising [44]. Since all ITEM-TWO experiments have been
4. Discussion performed under identical conditions and instrument settings, all
pressures have been kept constant for all investigations in the respective
For in-solution equilibria, the nominal stability of a protein complex compartments of the mass spectrometers. So, the influence of the
is described by the internal energy difference between educts and pressure should be considered unchanged; the same holds true for the
products, with the activation energy determining the reaction rate. temperature. We assume that due to collisional cooling in the source,
With the exception of tunneling effects or barrier-less (“downhill”) the temperature that protein complexes have adopted in-solution is not
processes, any reaction goes through at least one transition state - ir- exactly the same as the one they possess in the gas phase. Of note, with
respective of whether proceeding forwards or backwards. The
8
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
binding partners of antibody – epitope peptide complexes, the energy
regimes that are needed to release the epitope peptides are easily met
with commercial instruments [52].
To compare calculated gas phase dissociation constants (K #Dm0g)
between different antibody – epitope peptide complexes with each
other entropy contributions should be negligible, which is achieved
when (i) a similar transition state and (ii) the dissociation process fol-
lows a hard spheres model [30]. Such prerequisites are in fact provided
by antibody – antigen complexes in solution where it was experimen-
tally proven that their interactions are enthalpy-driven [53,54]. This
quality of antibody – antigen complexes seems preserved in the gas
phase, at least to some extent. Hence, from all above considerations it
appears well possible to qualitatively compare apparent gas phase and
in-solution dissociation constants of antibody – antigen complexes,
after correcting the energy terms, i.e. subtracting external energy con-
tributions by linear extrapolation.
The fact that gas phase binding strengths can be ordered and result
in the same ranking as that which was obtained by in-solution binding
strength determinations has been observed as well in a study that had
investigated vancomycin – peptide interactions [55]. This does, how-
ever, not exclude that there are cases where entropy-dependences be-
come of more importance for protein – protein complex formation and/
or dissociation.
Using surface plasmon resonance measurements the dissociation
constants of a PentaHis antibody were determined to be 10 nM and for a
3D5 antibody to be 340 nM, respectively [56]. Fairly large variability in
detecting His-tagged Epo proteins using four different anti-His-tag an-
tibodies were observed in Western blot analyses [57]. For the mono-
clonal “RGS-His” antibody which recognizes the RGSHHHH motif were
given values for KD s of approx. 10 nM to 50 nM. The KD s value which
was determined by us upon immobilization of the synthetic His-tag
epitope peptide (107.7 ± 9.2 nM) falls within the range of the reported
values. Differences in KD s values have also been reported with mono-
Fig. 7. Courses of normalized ion intensities and of apparent dissociation ac- clonal anti-FLAG M2 antibody binding to immobilized FLAG epitope
tivation energies over temperature as functions of collision cell voltage differ- peptides as opposed to immobilizing the antibody and providing a
ences (ΔCV). A: Normalized intensities of immune complex (norm h0), complex FLAG-GFP fusion protein as binding partner. When monitoring by SPR
released peptide ions (norm h1) and antibody fragments (norm h2). Each curve imaging the antibody-FLAG binding using a peptide array KD s was
was fitted using a Boltzmann function. Each data point is the mean of three determined to be 15 nM [58]. By contrast, when immobilizing the an-
independent measurements and standard deviations are shown by vertical bars tiFLAG M1 antibody and using a FLAG-GFP fusion protein as analyte,
(cf. Supplement Table S10). B: Apparent dissociation activation energies over
G Km# g D s values of 324 nM to 412 nM have been reported, depending ontemperature ( ) and linear extrapolations to ΔCV=0. Synthetic His-tag
T Ca2+ presence or absence [59]. Since the in-solution determined KD s
peptide (black filled squares and lines) and His-tag peptide from tryptic digest value (70.6 ± 8.0 nM) which we had determined for an antiFLAG M2
of rhβ actin (gray filled triangles and lines). The vertical dashed line marks antibody is within the published values, we feel confident that in-so-
lution values obtained in this study can be used for comparisons with
the ITEM-TWO-determined gas phase binding strengths.
# G#
extrapolation of experimentally determined Gmg values to m0g , in- Also of importance, the binding strength of antibody – antigen in-
T T
strument-related parameters have become of lesser importance as can teraction in solution has been correlated with snug binding of just three
be seen when comparing the gas phase binding strengths of two dif- to five amino acid residues from an epitope with their counterparts of
ferent His-Tag epitope peptides, one from a synthetic peptide (28.0 the antibody's paratope [60,61]. The assumption of just a few amino
Jmol−1 K−1) and the other from a tryptic digest (26.0 Jmol−1 K−1), acid residues being of importance for molecular interactions on large
which both had been complexed to one and the same antiHis-tag an- protein partial surfaces stands in agreement with observations from
tibody. The error of the linear extrapolation procedure has been esti- chemical modifications of protein surfaces [62–64] which also point to
mated to approx. 10% [45]. “key-residues” which determine binding specificity and binding
It should be noted that the process that generates epitope peptide strength. It seems that existence of such minimal binding conditions by
ions during CID is completely different to that by which ionized pep- “specificity-determining residues” are maintained by antibody – an-
tides are produced during ESI [20,22]. Upon CID one observes asym- tigen complexes in the gas phase even when devoid of solvent [65].
metric charge distribution along with dissociation of the complex ions With our ITEM-TWO method, we have demonstrated that it is possible
[46–50]. Because this process takes place in the gas phase, the “matrix to determine apparent dissociation activation energies over tempera-#
effect”, i.e. unpredictable ion suppression effects during ESI, is avoided. Gture ( m0g ) that are required to dissociate antibody – epitope peptide
T
Therefore, the time the immune complex ions spend in the collision cell complexes in the gas phase, thereby providing semi-quantitative quasi
becomes crucial with respect to whether or not their non-covalent thermodynamic properties of the investigated immune complexes.
bonds break, since with increasing time the number of collisions in- Consequently, it becomes possible with ITEM-TWO to determine the
creases. The time that it takes for an immune complex to traverse the role of each amino acid at any position in the epitope sequence with
collision cell is estimated to last more than 100 μs [42,51] if not close to respect to its energetic contribution to the binding strength by which
1ms [44]. Since CID is efficiently breaking the bonds between the the epitope peptide binds to an antibody's paratope. In sum, ITEM-TWO
9
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
is very powerful for direct identification of an epitope and its binding generation of therapeutic antibodies, Nat. Rev. Immunol. 10 (5) (2010) 345–352
strength, since in-solution handling is reduced to mixing of antigen or (Epub 2010/04/24).
epitope peptide and antibody solutions. Because suitable mass spec- [13] F. Mack, M. Ritchie, P. Sapra, The next generation of antibody drug conjugates,Semin. Oncol. 41 (5) (2014) 637–652 (Epub 2014/12/03).
trometry equipment has become amply available, ITEM-TWO should be [14] D.A. Richards, A. Maruani, V. Chudasama, Antibody fragments as nanoparticle
easily adaptable by mass spectrometry laboratories all around the targeting ligands: a step in the right direction, Chem. Sci. 8 (1) (2017) 63–77 (Epub
world. 2017/04/30).[15] M.G. Weller, Quality issues of research antibodies, Anal. Chem. Insights. 11 (2016)
21–27 (Epub 2016/03/26).
Authorship statement [16] M. Baker, Reproducibility crisis: blame it on the antibodies, Nature. 521 (7552)
(2015) 274–276 (Epub 2015/05/23).
[17] K.F.M. Opuni, M. Al-Majdoub, Y. Yefremova, R.F. El-Kased, C. Koy, M.O. Glocker,
Significant contributions to the conception of the reported study Mass spectrometric epitope mapping, Mass Spectrom. Rev. 37 (2) (2018) 229–241
came from MOG, BDD, YY, KFMO. The design of the study originated (Epub 2016/07/13).
from MOG, BDD, CK. Execution of experiments was done by BDD, YY, [18] M. Al-Majdoub, K.F. Opuni, Y. Yefremova, C. Koy, P. Lorenz, R.F. El-Kased, et al., A
novel strategy for the rapid preparation and isolation of intact immune complexes
CR, CK. Interpretation of results was done by MOG, BDD, YY, KFMO, from peptide mixtures, J. Mol. Recognit. 27 (9) (2014) 566–574 Epub 2014/07/22.
CR. [19] M. Przybylski, M.O. Glocker, Electrospray mass spectrometry of biomacromolecular
complexes with noncovalent interactions - new analytical perspectives for supra-
molecular chemistry and molecular recognition processes, Angew. Chem. Int. Ed.
Declaration of Competing Interest Eng. 35 (8) (1996) 807–826.
[20] Y. Yefremova, B.D. Danquah, K.F. Opuni, R. El-Kased, C. Koy, M.O. Glocker, Mass
The authors declare no conflict of interest. spectrometric characterization of protein structures and protein complexes in
condensed and gas phase, Eur. J. Mass Spectrom (Chichester). 23 (6) (2017)
445–459 (Epub 2017/12/01).
Acknowledgements [21] K. Ishii, M. Noda, S. Uchiyama, Mass spectrometric analysis of protein-ligand in-
teractions, Biophys. Physicobiol. 13 (2016) 87–95 (Epub 2016/12/08).
We express our thanks to Mr. Michael Kreutzer for providing his [22] Y. Yefremova, K.F.M. Opuni, B.D. Danquah, H.J. Thiesen, M.O. Glocker, IntactTransition Epitope Mapping (ITEM), J. Am. Soc. Mass Spectrom. 28 (8) (2017)
expertise on bioinformatics and data handling. We acknowledge the 1612–1622 (Epub 2017/06/16).
German Academic Exchange Service (DAAD) for providing scholarships [23] B.D. Danquah, C. Röwer, K.F.M. Opuni, R.F. El-Kased, D. Frommholz, H. Illges,
for BD (No. 91566064), YY (No. 91523785), and KO (No. 91548123). et al., Intact Transition Epitope Mapping - Targeted High-Energy Rupture ofExtracted Epitopes (ITEM - THREE), Mol. Cell. Proteomics 18 (8) (2019)
The WATERS Synapt G2S mass spectrometer has been bought through 1543–1555.
an EU grant [EFRE-UHROM 9] made available to MOG. The mass [24] M. Al-Majdoub, K.F. Opuni, C. Koy, M.O. Glocker, Facile fabrication and instant
spectrometry data have been deposited to the ProteomeXchange application of miniaturized antibody-decorated affinity columns for higher-order
structure and functional characterization of TRIM21 epitope peptides, Anal. Chem.
Consortium via the PRIDE partner repository with the dataset identifier 85 (21) (2013) 10479–10487 (Epub 2013/10/08).
PXD016024. [25] A.D. Robertson, K.P. Murphy, Protein structure and the energetics of protein sta-
bility, Chem. Rev. 97 (5) (1997) 1251–1268 (Epub 1997/08/05).
[26] E.B. Erba, K. Barylyuk, Y. Yang, R. Zenobi, Quantifying protein-protein interactions
Appendix A. Supplementary data within noncovalent complexes using electrospray ionization mass spectrometry,
Anal. Chem. 83 (24) (2011) 9251–9259.
Supplementary data to this article can be found online at https:// [27] T.J.D. Jorgensen, P. Roepstorff, A.J.R. Heck, Direct determination of solution
doi.org/10.1016/j.jprot.2019.103572. binding constants for noncovalent complexes between bacterial cell wall peptideanalogues and vancomycin group antibiotics by electrospray ionization mass
spectrometry, Anal. Chem. 70 (20) (1998) 4427–4432.
References [28] S.R. Krishnaswamy, E.R. Williams, J.F. Kirsch, Free energies of protein-protein as-
sociation determined by electrospray ionization mass spectrometry correlate ac-
curately with values obtained by solution methods, Protein Sci. 15 (6) (2006)
[1] E. Bjorling, M. Uhlen, Antibodypedia, a portal for sharing antibody and antigen 1465–1475.
validation data, Mol. Cell. Proteomics 7 (10) (2008) 2028–2037 (Epub 2008/ [29] N.B. Cech, C.G. Enke, Practical implications of some recent studies in electrospray
08/01). ionization fundamentals, Mass Spectrom. Rev. 20 (6) (2001) 362–387 (Epub 2002/
[2] S. Dubel, O. Stoevesandt, M.J. Taussig, M. Hust, Generating recombinant antibodies 05/09).
to the complete human proteome, Trends Biotechnol. 28 (7) (2010) 333–339 (Epub [30] Y. Yefremova, F.T.I. Melder, B.D. Danquah, K.F.M. Opuni, C. Koy, A. Ehrens, et al.,
2010/06/12). Apparent activation energies of protein-protein complex dissociation in the gas-
[3] M. Uhlen, F. Ponten, Antibody-based proteomics for human tissue profiling, Mol. phase determined by electrospray mass spectrometry, Anal. Bioanal. Chem. 409
Cell. Proteomics 4 (4) (2005) 384–393 (Epub 2005/02/08). (28) (2017) 6549–6558 Epub 2017/09/14.
[4] R.F. El-Kased, C. Koy, P. Lorenz, S. Drynda, R. Guthke, Z. Qian, et al., Mass spec- [31] Y. Perez-Riverol, A. Csordas, J. Bai, M. Bernal-Llinares, S. Hewapathirana,
trometric and peptide chip epitope analysis on the RA33 autoantigen with sera from D.J. Kundu, et al., The PRIDE database and related tools and resources in 2019:
rheumatoid arthritis patients, Eur. J. Mass. Spectrom. (Chichester) 16 (3) (2010) improving support for quantification data, Nucleic Acids Res. 47 (D1) (2019) D442-
443–451 Epub 2008/01/01. D50.
[5] C. Stafylis, J.D. Klausner, Evaluation of two 4th generation point-of-care assays for [32] M. Al-Majdoub, C. Koy, P. Lorenz, H.-P. Thiesen, M.O. Glocker, Mass spectrometric
the detection of human immunodeficiency virus infection, PLoS One 12 (8) (2017) and peptide chip characterization of an assembled epitope: analysis of a polyclonal
e0183944(Epub 2017/08/29). antibody model serum directed against the Sjogren autoantigen TRIM21, J. Mass
[6] M. Linnebacher, P. Lorenz, C. Koy, A. Jahnke, N. Born, F. Steinbeck, et al., Clonality Spectrom. 48 (6) (2013) 651–659.
characterization of natural epitope-specific antibodies against the tumor-related [33] C.S. Hoaglund, S.J. Valentine, C.R. Sporleder, J.P. Reilly, D.E. Clemmer, Three-di-
antigen topoisomerase IIa by peptide chip and proteome analysis: a pilot study with mensional ion mobility/TOFMS analysis of electrosprayed biomolecules, Anal.
colorectal carcinoma patient samples, Anal. Bioanal. Chem. 403 (1) (2012) Chem. 70 (11) (1998) 2236–2242 (Epub 1998/06/13).
227–238 Epub 2012/02/22. [34] A. Konijnenberg, A. Butterer, F. Sobott, Native ion mobility-mass spectrometry and
[7] O.H. Brekke, G.A. Loset, New technologies in therapeutic antibody development, related methods in structural biology, Biochim. Biophys. Acta 1834 (6) (2013)
Curr. Opin. Pharmacol. 3 (5) (2003) 544–550 (Epub 2003/10/16). 1239–1256 (Epub 2012/12/19).
[8] P. Chames, M. Van Regenmortel, E. Weiss, D. Baty, Therapeutic antibodies: suc- [35] J. Seo, W. Hoffmann, S. Warnke, M.T. Bowers, K. Pagel, G. von Helden, Retention of
cesses, limitations and hopes for the future, Br. J. Pharmacol. 157 (2) (2009) native protein structures in the absence of solvent: a coupled ion mobility and
220–233 (Epub 2009/05/23). spectroscopic study, Angew. Chem. Int. Ed. Eng. 55 (45) (2016) 14173–14176
[9] A.C. Chan, P.J. Carter, Therapeutic antibodies for autoimmunity and inflammation, (Epub 2016/10/30).
Nat. Rev. Immunol. 10 (5) (2010) 301–316 (Epub 2010/04/24). [36] J.T.S. Hopper, N.J. Oldham, Collision induced unfolding of protein ions in the gas
[10] M. Zhang, G. Yu, B. Chan, J.T. Pearson, P. Rathanaswami, J. Delaney, et al., phase studied by ion mobility-mass spectrometry: the effect of ligand binding on
Interleukin-21 receptor blockade inhibits secondary humoral responses and halts conformational stability, J. Am. Soc. Mass Spectrom. 20 (10) (2009) 1851–1858.
the progression of preestablished disease in the (NZB x NZW)F1 systemic lupus [37] Y. Zhong, S.J. Hyung, B.T. Ruotolo, Ion mobility-mass spectrometry for structural
erythematosus model, Arthritis Rheum. 67 (10) (2015) 2723–2731 Epub 2015/ proteomics, Expert Rev. Proteom. 9 (1) (2012) 47–58 (Epub 2012/02/02).
06/23. [38] K. Breuker, F.W. McLafferty, Stepwise evolution of protein native structure with
[11] R. Thakare, A. Dasgupta, S. Chopra, Bezlotoxumab for the treatment of Clostridium electrospray into the gas phase, 10(−12) to 10(2) S, Proc. Natl. Acad. Sci. U. S. A.
difficile-associated diarrhea, Drugs Today (Barc). 53 (7) (2017) 385–392 (Epub 105 (47) (2008) 18145–18152.
2017/08/25). [39] A.J.R. Heck, Native mass spectrometry: a bridge between interactomics and struc-
[12] A. Beck, T. Wurch, C. Bailly, N. Corvaia, Strategies and challenges for the next tural biology, Nat. Methods 5 (11) (2008) 927–933.
10
B.D. Danquah, et al. Journal of Proteomics 212 (2020) 103572
[40] J. Loo, Studying noncovalent protein complexes by electrospray ionization mass valency using an Fv fragment fused to a fragment of protein-A, J. Biol. Chem. 268
spectrometry, Mass Spectrom. Rev. 16 (1997) 1–23. (27) (1993) 20668–20675.
[41] J.E. Carpenter, C.P. McNary, A. Furin, A.F. Sweeney, P.B. Armentrout, How hot are [54] W. Ito, Y. Iba, Y. Kurosawa, Effects of substitutions of closely-related amino-acids at
your ions really? A threshold collision-induced dissociation study of substituted the contact surface in an antigen-antibody complex on thermodynamic parameters,
benzylpyridinium "thermometer" ions, J. Am. Soc. Mass Spectrom. 28 (9) (2017) J. Biol. Chem. 268 (22) (1993) 16639–16647.
1876–1888. [55] T.J.D. Jorgensen, D. Delforge, J. Remacle, G. Bojesen, P. Roepstorff, Collision-in-
[42] F. Ichou, D. Lesage, X. Machuron-Mandard, C. Junot, R.B. Cole, J.C. Tabet, Collision duced dissociation of noncovalent complexes between vancomycin antibiotics and
cell pressure effect on CID spectra pattern using triple quadrupole instruments: a peptide ligand stereoisomers: evidence for molecular recognition in the gas phase,
RRKM modeling, J. Mass Spectrom. 48 (2) (2013) 179–186. Int. J. Mass Spectrom. 188 (1–2) (1999) 63–85.
[43] F. Ichou, A. Schwarzenberg, D. Lesage, S. Alves, C. Junot, X. Machuron-Mandard, [56] K.M. Muller, K.M. Arndt, K. Bauer, A. Pluckthun, Tandem immobilized metal-ion
et al., Comparison of the activation time effects and the internal energy distribu- affinity chromatography/immunoaffinity purification of His-tagged proteins–eva-
tions for the CID, PQD and HCD excitation modes, J. Mass Spectrom. 49 (6) (2014) luation of two anti-His-tag monoclonal antibodies, Anal. Biochem. 259 (1) (1998)
498–508 Epub 2014/06/11. 54–61 (Epub 1998/05/30).
[44] J.M. Wells, S.A. McLuckey, Collision-induced dissociation (CID) of peptides and [57] N. Debeljak, L. Feldman, K.L. Davis, R. Komel, A.J. Sytkowski, Variability in the
proteins, Methods Enzymol. 402 (2005) 148–185 (Epub 2006/01/13). immunodetection of His-tagged recombinant proteins, Anal. Biochem. 359 (2)
[45] M.M. Santoro, D.W. Bolen, A test of the linear extrapolation of unfolding free en- (2006) 216–223 (Epub 2006/11/04).
ergy changes over an extended denaturant concentration range, Biochemistry. 31 [58] G.J. Wegner, H.J. Lee, R.M. Corn, Characterization and optimization of peptide
(20) (1992) 4901–4907 (Epub 1992/05/26). arrays for the study of epitope-antibody interactions using surface plasmon re-
[46] J.L. Benesch, B.T. Ruotolo, D.A. Simmons, C.V. Robinson, Protein complexes in the sonance imaging, Anal. Chem. 74 (20) (2002) 5161–5168 (Epub 2002/10/31).
gas phase: technology for structural genomics and proteomics, Chem. Rev. 107 (8) [59] A. Einhauer, A. Jungbauer, Affinity of the monoclonal antibody M1 directed against
(2007) 3544–3567 (Epub 2007/07/26). the FLAG peptide, J. Chromatogr. A 921 (1) (2001) 25–30 (Epub 2001/07/20).
[47] J.C. Jurchen, E.R. Williams, Origin of asymmetric charge partitioning in the dis- [60] M.H.V. Van Regenmortel, Mapping epitope structure and activity: from one-di-
sociation of gas-phase protein homodimers, J. Am. Chem. Soc. 125 (9) (2003) mensional prediction to four-dimensional description of antigenic specificity,
2817–2826 (Epub 2003/02/27). Methods 9 (3) (1996) 465–472 Epub 1996/06/01.
[48] S.V. Sciuto, J. Liu, L. Konermann, An electrostatic charge partitioning model for the [61] E.A. Padlan, C. Abergel, J.P. Tipper, Identification of specificity-determining re-
dissociation of protein complexes in the gas phase, J. Am. Soc. Mass Spectrom. 22 sidues in antibodies, FASEB J. 9 (1995) 133–139.
(10) (2011) 1679–1689 (Epub 2011/09/29). [62] M.O. Glocker, S. Nock, M. Sprinzl, M. Przybylski, Characterization of surface to-
[49] F. Sobott, C.V. Robinson, Characterising electrosprayed biomolecules using tandem- pology and binding area in complexes of the elongation factor proteins EF-Ts and
MS - the noncovalent GroEL chaperonin assembly, Int. J. Mass Spectrom. 236 (1–3) EF-Tu center dot GDP from Thermus thermophilus: a study by protein chemical
(2004) 25–32. modification and mass spectrometry, Chem. Eur. J. 4 (4) (1998) 707–715.
[50] R.R. Ogorzalek Loo, J.A. Loo, Salt Bridge Rearrangement (SaBRe) explains the [63] H.P. Happersberger, M. Przybylski, M.O. Glocker, Selective bridging of Bis-
dissociation behavior of noncovalent complexes, J. Am. Soc. Mass Spectrom. 27 (6) Cys¬teinyl residues by arsonous acid derivatives as an approach to the character-
(2016) 975–990. ization of protein tertiary structures and folding reactions by mass spectrometry,
[51] A. Kuki, L. Nagy, M. Zsuga, S. Keki, An approach to estimate the activation energies Anal. Biochem. 264 (1998) 237–250.
of fragmentation occurring in quadrupole collision cell of the mass spectrometer, J. [64] M.O. Glocker, M. Kalkum, R. Yamamoto, Y. Schreurs, Selective biochemical mod-
Mass Spectrom. 52 (10) (2017) 672–680 (Epub 2017/07/22). ification of functional residues in recombinant human macrophage Colony-
[52] J.L. Benesch, B.T. Ruotolo, F. Sobott, J. Wildgoose, A. Gilbert, R. Bateman, et al., Stimulating Factor β (rhM-CSF β); identification by mass spectrometry,
Quadrupole-time-of-flight mass spectrometer modified for higher-energy dissocia- Biochemistry. 35 (1996) 14625–14633.
tion reduces protein assemblies to peptide fragments, Anal. Chem. 81 (3) (2009) [65] F. Lermyte, F. Sobott, Electron transfer dissociation provides higher-order structural
1270–1274 Epub 2008/12/25. information of native and partially unfolded protein complexes, Proteomics. 15 (16)
[53] W. Ito, Y. Kurosawa, Development of an artificial antibody system with multiple (2015) 2813–2822 (Epub 2015/06/18).
11