Charge-State-Dependent Variation of Signal Intensity Ratio between

Apr 13, 2018 - Charge-State-Dependent Variation of Signal Intensity Ratio between ... This situation raises serious concerns regarding the reliability...
0 downloads 0 Views 999KB Size
Subscriber access provided by UNIV OF SCIENCES PHILADELPHIA

Charge-state dependent variation of signal intensity ratio between unbound protein and protein-ligand complex in electrospray ionization mass spectrometry: the role of solvent-accessible surface area Konstantin Chingin, and Konstantin Barylyuk Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.7b05349 • Publication Date (Web): 13 Apr 2018 Downloaded from http://pubs.acs.org on April 19, 2018

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Charge-state dependent variation of signal intensity ratio between unbound protein and protein-ligand complex in electrospray ionization mass spectrometry: the role of solvent-accessible surface area

Konstantin Chingin1 and Konstantin Barylyuk2 1

Jiangxi Key Laboratory for Mass Spectrometry and Instrumentation, East China University of Technology, Guanglan road 418, Nanchang, Jiangxi, China, 330013 2 Department of Biochemistry, University of Cambridge, Hopkins Building, Tennis Court Road, Cambridge, CB2 1QW, United Kingdom

Correspondence to: [email protected]; [email protected].

Keywords: native mass spectrometry, protein-ligand complexes; binding constant; charge state distribution; charged residue model; electrospray ionization.

ACS Paragon Plus Environment

1

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 26

Abstract Native electrospray ionization mass spectrometry (ESI-MS) is nowadays widely used for the direct and sensitive determination of protein complex stoichiometry and binding affinity constants (Ka). A common yet poorly understood phenomenon in native ESI-MS is the difference between the charge state distributions (CSDs) of the bound protein-ligand complex (PL) and unbound protein (P) signals. This phenomenon is typically attributed to experimental artefacts such as non-specific binding or in-source dissociation and is considered highly undesirable because the determined Ka values display strong variation with charge state. This situation raises serious concerns regarding the reliability of ESI-MS approach for the analysis of protein complexes. Here we demonstrate that, contrary to the common belief, the CSD difference between P and PL ions can occur without any loss of complex integrity, simply due to a change in the solvent-accessible surface area (∆SASA) of the protein upon ligand binding in solution. The experimental CSD shifts for PL and P ions in ESI-MS are explained in relation to the magnitude of ∆SASA for diverse protein-ligand systems using a simple model based on the charged residue mechanism. Our analysis shows that the revealed ∆SASA factor should be considered rather general and be given attention for the correct spectral interpretation of protein complexes.

ACS Paragon Plus Environment

2

Page 3 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Many studies indicate that the native structure of proteins and noncovalent protein-ligand complexes can be partially preserved during electrospray ionization (ESI) and that the properties of gas-phase protein ions, in particular, their charge states, can be used to characterize the protein structure in solution.1-6 For example, folded proteins typically yield gasphase ions with a lower degree of charging and narrower charge state distribution (CSD) than their unfolded counterparts. Therefore, the charge states of gas-phase ions are useful to characterize the conformation of proteins in solution.7-9 Native ESI-MS generally allows the observation of noncovalent protein complexes in the gas phase.10-16 Protein-ligand complexes (PL) are typically observed in native ESI-MS together with unbound proteins (P). Binding affinity constant (Ka) is directly derived from the intensity ratio of PL and P signals. A fairly good agreement has been demonstrated between the Ka values derived by native ESI-MS and by earlier established analytical methods for a multitude of protein complexes.17-22 Direct native ESI-MS titration data allow for adequate characterization of complex protein-ligand binding equilibria involving multiple binding steps and cooperative effects,23-28 which is often helpful in order to validate or refine the binding model as well as to increase the accuracy of Ka value determination.29 Native ESI-MS is also an attractive technique for drug discovery30-32 via screening of compound or fragment libraries.33-40 However, significant deviations in the results are also rather common, both between MS and other methods40-42 as well as between different research laboratories.43 Clearly, closer attention is needed toward the controversial phenomena encountered in native MS in order to better understand the capabilities of this approach for the correct spectral interpretation of noncovalent protein complexes.43,44

ACS Paragon Plus Environment

3

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 26

A frequently encountered controversial phenomenon in native ESI-MS of PL complexes is that the ratio between P and PL signal intensities – and hence the determined Ka values – vary with protein ion charge state.19,28,43 The difference in CSD between PL and P ions is commonly attributed to either nonspecific P-L binding in ESI droplets, conformational changes of protein ions in ESI droplets, in-source dissociation of PL ions, m/z-specific efficiency of vacuum ion transfer, different signal yields by P and PL ions or the interplay of these factors.43 All these factors affect the accuracy of Ka determination. Therefore, the occurrence of CSD difference between PL and P ions is commonly seen as undesirable manifestation, and hence is a rather common practice that the softness of ionization process during the experiment is evaluated based on the degree of PL/P ratio variation in the mass spectrum: the lower the variation the more ‘native’ are considered the experimental conditions. Accordingly, experimental parameters would be tuned such as to bring the PL/P ratio variation to the minimum and thus attain closer similarity between the Ka values derived from different charge states. However, aside from the above-mentioned factors, the CSD difference between PL and P ions might also occur without any distortion in the stoichiometry of ion ensemble simply as follows: a) ligand binding brings about a shift in the solvent-accessible surface area (SASA) of PL complex relative to the unbound P in solution; b) the shift in SASA between PL and P causes a shift in the degree of charging between the PL and P ions formed by ESI; c) the shift in the degree of charging results in the different shape of CSD for PL and P ions, i.e. the CSD difference. Surprisingly, this ‘∆SASA’ factor has not been paid serious attention to and remains largely overlooked in the literature. However, as we demonstrate in this study, the ∆SASA factor

ACS Paragon Plus Environment

4

Page 5 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

should play a key role in shaping the CSD of P and PL ions and should be thoroughly considered for the correct spectral interpretation of noncovalent protein-ligand complexes. Protein charging in ESI-MS The average charge (ZAV) of ideally spherical ions in ESI-MS is typically (however, not always45-48) fairly well predicted by the charged residue model (CRM)49-53:

 =

4 (4   )/ (1) 

Here e is the elementary charge, ε0 is the vacuum permittivity, γ is the surface tension of aqueous droplets. Earlier experimental research revealed that ZAV and SASA of real proteins in native ESI-MS correlate with each other in fairly close agreement with the CRM predictions:9,5457

 ∼ (  ) (2) Different authors report slightly different experimental values for the β-factor, in the range from ca. 0.6 to ca. 0.7.9,55-57 The correlation (2) holds true throughout the wide range of proteins including those with pronounced non-globularity as suggested by the fit of experimental data.57 Note that it is SASA rather than the molecular weight (M) that primarily determines the extent of protein charging. Indeed, the majority of proteins are tightly packed in their native state, and their SASA monotonically increases with M (SASA ∼ M2/3). For such proteins, the extent of charging can be re-expressed in terms of molecular weight instead of surface area ( ∼  / ). However, ligand-binding proteins commonly contain cavities, grooves and other surface features that serve as binding pockets. For such proteins as well as

ACS Paragon Plus Environment

5

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 26

for proteins with pronounced non-globularity there can occur significant deviations from the SASA ∼ M2/3 law, and the extent of charging is correctly expressed in terms of SASA rather than molecular weight.57 It is also important to note that protein charging is a solution-phase process occurring in late ESI droplets, and hence is the relation between protein ZAV and its solutionphase SASA rather than gas-phase surface area.58 The structure of desolvated protein ions can significantly differ from the native solution structure due to the loss of hydrophobic interactions, the increasing role of electrostatic interactions and other factors related to the absence of a solvent.59 Therefore, the gas-phase collision cross-section of a protein ion cannot reliably substitute the solution-phase SASA in relation (2) in order to predict the extent of protein charging.60 Predicted shift in ZAV between PL and P ions Using relation (2), the shift in ZAV between PL and P can be derived as: ∆   −   (   ) − (   ) ∆   = ≈ ≈  (3)   

  (   ) It can be seen that the relative magnitude of ∆ZAV is determined by the relative difference in SASA between PL and P. For example, for a 10+ charge state 1 % difference in SASA between PL and P eq. (3) yields ∆ZAV ≈ 0.06-0.07, depending on which value of β is chosen for calculation.9,55-57 Even though this shift may appear minor, it may cause a notable difference between the CSD of PL and P. This is because protein ions in native MS have a very narrow CSD. The major share of the total signal intensity is commonly accumulated in just one or two adjacent charge states. Figure 1a shows an experimental mass spectrum of human carbonic

ACS Paragon Plus Environment

6

Page 7 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

anhydrase I (hCA I) obtained under native ESI-MS conditions. The spectrum is dominated by the charge state 10+ (ZAV = 9.84). We fitted the plot of signal intensity vs. protein charge state by a Gaussian function (Figure 1b). Gaussian fit is generally considered to be a good approximation for the CSD in native ESI-MS.57 The Gaussian plot was then shifted by 0.07 charge units to simulate the CSD of charge-shifted PL mass spectrum upon binding of a hypothetical ligand (Figure 1b). It can be seen that the shift ∆ZAV = 0.07 causes significant alteration in the shape of CSD, the PL/P ratio varying from 0.8 for the 9+ charge state to 1.3 for the 11+ charge state. This variation in PL/P ratio translates into a twofold variation of the calculated Ka value (from 0.15 to 0.3 µM-1).43 Larger ∆ZAV should cause even stronger deviation in PL/P ratio and a more pronounced effect on Ka. Shift in SASA upon ligand binding A shift in SASA of PL complex relative to the unbound P can generally be caused by several possible factors: a) conformational change of P due to the binding of L; b) occupation of SASA at the binding site and creation of new SASA due to the binding of L; c) conformational change of L upon binding.61 For the simplicity of consideration in this work, we focus on the P-L systems for which a shift in SASA is caused mainly by just one of these factors rather than their interplay. Protein conformational change upon ligand binding Many proteins are known to undergo conformational changes upon ligand binding. Most studied systems include metal-binding proteins, such as calmodulin, calbindin and αlactalbumin binding Ca2+.58,62 Conformational change of protein upon ligand binding can cause a significant shift in SASA of PL relative to P. Thus, the analysis of crystal structures indicates that

ACS Paragon Plus Environment

7

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 26

the SASA of holo-α-lactalbumin exceeds the SASA of apo-α-lactalbumin by ∼ 5 %.62 Note that metal binding can cause both elongation and compaction of a protein structure, and therefore ∆SASA and ∆ZAV can be both positive (protein elongation) and negative (protein compaction).58 The high sensitivity of protein charging to its conformation in ESI-MS has been well recognized and documented in earlier studies.9 Rigid proteins While it is evident that the conformational change in P due to the binding of L can produce notable ∆SASA, it is however also important to consider the opposite scenario, viz. when the binding of L does not alter the conformation of P (‘rigid protein’; Figure S1a-d). ∆SASA in the complexes of rigid proteins is contributed merely by the ligand surface. With regard to its role in a protein complex, the ligand surface can be expressed as a sum of two terms: SASAL = buried SASA (bSASA) + added SASA (aSASA) (Scheme 1a). The bSASA is the part of SASAL that comes into contact with protein upon binding. This surface becomes inaccessible to solvent upon ligand binding, and hence bSASA has a negative contribution to ∆SASA. In contrast, the aSASA is the part of SASAL that forms a new surface area of PL upon binding, and hence is the positive contribution of aSASA to ∆SASA. It is thereby easily seen that ∆SASA is contributed by the two opposite-sign components: ∆SASA = – bSASA + aSASA. Accordingly, the value of ∆SASA can be negative, zero, or positive, depending on the relationship between bSASA and aSASA (Scheme 1b). If bSASA > aSASA then the ligand surface mainly covers the binding interface (∆SASA < 0). The decrease in SASA comes due to the ‘flattening’ of concave binding pocket by the ligand (Figure 2a). If bSASA ≈ aSASA then ∆SASA ≈ 0. This scenario may be expected when

ACS Paragon Plus Environment

8

Page 9 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

the ligand occupies the groove-shaped binding site but at the same time creates an additional surface on the sides (Figure 2b). Finally, if bSASA > aSASA then the ligand surface mainly creates a new surface (∆SASA > 0). The increase in SASA comes due to a part of the bound ligand surface that protrudes out to the exterior of protein surface thus creating an additional surface (Figure 2c). Indeed, a key question remains open whether or not the effects discussed above are sufficiently strong to induce notable shifts of ZAV (Scheme 1c) and SASA in real proteinligand systems. This question can only be addressed experimentally. Experimental shifts of SASA and ZAV for noncovalent protein-ligand complexes Table S1 summarizes the magnitudes of ∆SASA and ∆ZAV derived from the available data sets for several popular protein-ligand systems widely used in MS studies. MS datasets were mainly selected from the studies with sufficiently rich statistics for various ligands and ligand concentration (and raw data availability whenever possible) and in which particular care was taken to minimize non-specific binding and other factors causing distortion in the stoichiometry of the detected protein complex ions (Table S2).28,35,41,63-67 SASA values for the corresponding proteins and their complexes with various ligands in solution were calculated from the available X-ray diffraction and NMR data, and ∆SASA was estimated using paired analysis as described in the Methods section of Supporting Information (Table S3). Figure 2 shows three representative examples for protein-ligand systems with negative, zero, and positive ∆SASA and ∆ZAV accordingly. For each protein-ligand system, statistical analysis was performed using the MS data for different ligand concentrations (see SI Methods for details). All MS data were produced on a qTOF mass spectrometer. Overall, our results presented in Tables S1-S3 and Figure 2

ACS Paragon Plus Environment

9

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 26

strongly support the prediction that there should be a correlation between ∆ZAV of PL complexes in the gas phase and ∆SASA in solution. Below we discuss experimental behavior for different protein-ligand systems. Human carbonic anhydrase I - acetazolamide Figure 2a-c shows the results for the complex of human carbonic anhydrase I (hCA I; 28739 Da) with acetazolamide (AZM; 222 Da). Note that the binding of AZM ligand to hCA I causes only minor – if any – changes to the protein structure, as indicated by the structural analysis shown in Figure S1a,b and Figure S2a,d. Therefore, hCA I can be considered a rigid protein. The paired analysis of crystal structures for protein complexes with and without ligand attached reveals a statistically significant (p < 0.0001) negative ∆SASA for all the ligands tested (Figure 2b). The negative value of ∆SASA (≈ - 117 A2) is explained by the flattening of the deep cavity-shaped binding pocket in hCA I upon binding with AZM (bSASA > aSASA). In agreement with the statistically significant negative ∆SASA in solution is the statistically significant (p = 0.0021) negative ∆ZAV of hCA I–AZM complex in ESI-MS (Figure 2c). Experimental ∆ZAV of hCA I–AZM complex spans the range from ca. -0.06 to -0.3, which is in good qualitative agreement with the prediction ∆ZAV ≈ -0.07 by the eq. (3) using β = 0.7.57 The experimental evidence for the negative ∆ZAV of a protein-ligand complex also serves as a good illustration that it is SASA rather than M that determines the extent of protein charging. If M was the primary factor, then the magnitude of ∆ZAV should have been positive for any protein-ligand system because ∆M is positive for any protein-ligand system. Lysozyme – NAG3

ACS Paragon Plus Environment

10

Page 11 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 2d-f shows the binding of lysozyme (lysC; 14299 Da) with NAG3 (628 Da). In contrast to hCA I, the binding site in lysC has a groove shape. Like in hCA I–AZM complex, there is only a negligibly small change of lysC conformation upon binding of a NAG3 ligand to lysC (Figure S1c,d, Figure S2b,e). Therefore, lysC can be considered a rigid protein. The NAG3 ligand flattens the groove surface but creates an additional surface on the sides (Figure 2d). As a result, even though the surface of NAG3 is larger than the surface of AZM, the binding of NAG3 ligand yields no statistically significant shift in SASA (∆SASA ≈ 0), as indicated by the analysis of highresolution 3D structures from the RCSB Protein Data Bank (Figure 2e). In other words, this means that aSASA ≈ bSASA. Consistently, no notable ∆ZAV shift is observed for lysozyme-NAG3 in ESI-MS (Figure 2f). The considered examples show that the magnitude of ∆SASA – and therefore ∆ZAV – cannot be predicted merely based on the surface area of the ligand. It is rather the specific configuration of protein-ligand binding interface that serves the primary factor for ∆SASA. Human Bcl-xL – Bak BH3 Figure 2g-i shows the results for the complex of human Bcl-xL (1-209, ∆45-84; 20650 Da) with an alpha-helical BH3-domain peptide derived from human Bak (72-87; 1725 Da). The binding site in BcI-xL is a shallow grove (Figure 2g): binding of the Bak peptide yields lower bSASA relative to the aSASA and therefore positive ∆SASA (Figure 2h). The broad span of experimental ∆SASA in Bcl-xL–BH3-peptide complexes (∼ 100-600 Å2) is due to a strong heterogeneity of the protein complex structures deposited in the RCSB Protein Data Bank, which contain a range of peptide ligands of different amino acid sequence and size, as well as variable deletions and

ACS Paragon Plus Environment

11

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 26

truncations of Bcl-xL chain. In part, the broad span might also reflect the considerable conformational freedom of the bound Bak ligand and/or of the receptor protein. There is some conformational rearrangement around the binding site of Bcl-xL (the so-called BC groove) upon BH3-domain peptide binding that includes winding of additional turns in α2 and concomitant unwinding in α3 (the two alpha helices flanking the BC groove), along with major side-chain rearrangements on both sides of the BC groove (Figure S1e-f).68 Therefore, Bcl-xL cannot be strictly considered a rigid protein. Furthermore, apart from the protein rearrangement, Bak BH3 peptide rearrangement can also be expected upon binding. Peptidic ligands quite commonly undergo conformational changes upon binding. A dramatic example is given by intrinsically disordered proteins, which can develop large binding interfaces without filling cavities but, rather, wrapping around the interactor.61 In agreement with the statistically significant (p < 0.0001) positive ∆SASA in condensed phase is the statistically significant (p < 0.0001) positive ∆ZAV of Bcl-xL–Bak complex in ESI-MS (Figure 2i). Experimental ∆ZAV of Bcl-xL–Bak complex spans the range from ca. 0.3 to 0.7, which is lower than the prediction ∆ZAV ≈ 0.06 – 0.4 by the eq. (3) using β = 0.7.57 Note, however, that the predicted range of ∆ZAV values is based on a ∆SASA estimate ∼ 100-600 Å2 (Figure 2h) that does not account for the conformational rearrangement in Bcl-xL. Expansion of the BC groove upon binding of alphahelical peptides increases the protein surface area, which adds to the total ∆SASA and should result in the higher ∆ZAV values measured in ESI-MS experiments. Bovine cationic trypsin – CMA

ACS Paragon Plus Environment

12

Page 13 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

The paired analysis of condensed-phase bovine cationic trypsin (23299 Da) structures with and without

noncovalent

2-aminomethyl-5-chlor-benzylamide

(CMA)

inhibitors

indicates

statistically significant (p < 0.0001) negative ∆SASA (Figure S3a). The negative ∆SASA in trypsin complexes has a similar origin to the negative ∆SASA in hCA I, i.e. due to the flattening of the cavity-shaped binding pocket upon binding with a ligand (bSASA > aSASA). Seemingly contradictory, the systematic ESI-MS measurements of trypsin complex with a range of CMA inhibitors (479-575 Da) done at different inhibitor concentrations did not reveal any statistically significant shift in ZAV (Figure S3b). Interestingly, the statistically significant negative ∆ZAV did appear in the ESI-MS upon the addition of imidazole additive in the spraying solution (Figure S3c). Imidazole is known to stabilize noncovalent complexes of trypsin and other proteins during ESI process when added into spraying solution prior to the analysis.64,69 Therefore, the lack of a notable shift in ZAV of trypsin-CMA (and therefore the closely similar CSD of P and PL) in the absence of imidazole should probably be regarded as the manifestation of non-specific L binding and/or dissociation during ionization. In contrast, the negative shift in ZAV (and hence the unequal CSD of P and PL) with imidazole should be regarded as the sign of enhanced protein complex stability during ionization due to the presence of imidazole additive. Experimental relationship between ∆SASA and ∆ZAV An experimental relationship between ∆SASA and ∆ZAV of different PL systems further supports the model proposed in this study (Figure 3). Figure 3a shows the relationship between ∆SASA for the noncovalent complexes of the analyzed proteins with various ligands normalized to SASAP and the magnitudes of bSASA normalized to SASAL derived from the condensed phase

ACS Paragon Plus Environment

13

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 26

data. A clear trend is seen that the normalized ∆SASA of the protein complexes is increased as the relative share of bSASA is decreased, i.e. as the relative share of aSASA is increased. ∆SASA is positive for normalized bSASA < 0.5, i.e. when aSASA > bSASA. ∆SASA is negative for normalized bSASA > 0.5, i.e. when aSASA < bSASA. Interestingly, the trend observed for Bcl-xL has a greater slope (fitted line b in Figure 3a) compared to that for the other analyzed proteins (fitted line a in Figure 3a). This is probably due to a higher rate of surface area growth for αhelical BH3-peptide ligands, whose shape is best approximated by a cylinder, compared to small organic molecules or oligosaccharides, whose shape is better approximated by a sphere. As discussed above, the conformational change of both Bcl-xL and alpha-helical BH3-domain peptide upon binding might additionally contribute to the observed trend. Figure 3b shows the relationship between the experimental magnitudes of ∆ZAV for the noncovalent complexes of various proteins normalized to ZAV and the magnitudes of bSASA normalized to SASAL derived from the condensed phase data. The relationship is very similar to that for ∆SASA in Figure 3a: ∆ZAV is increased as the relative share of bSASA is decreased. ∆ZAV is positive for normalized bSASA < 0.5, i.e. when aSASA > bSASA. ∆ZAV is negative for normalized bSASA > 0.5, i.e. when aSASA < bSASA. The relationship between ∆ZAV and bSASA can be roughly approximated using a linear fit. Interestingly, ∆ZAV value for the trypsin–CMA-inhibitor complexes in the presence of imidazole (open circle in Figure 3b) displays notable quantitative deviation from the trend. This deviation is most probably related to the fact that imidazole has a high gas-phase basicity and carries away a large portion of available protons from charged droplets during ionization. More research is needed to definitively confirm this explanation. Note that, despite the quantitative deviation, the negative sign of ∆ZAV for the trypsin–CMA-inhibitor–imidazole system is still

ACS Paragon Plus Environment

14

Page 15 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

consistently explained by the fact that normalized bSASA > 0.5 (Figure 3b). Finally, Figure 3c presents normalized ∆ZAV for the same PL systems as displayed in Figure 3b, now as a function of normalized ∆SASA. Vividly, ∆ZAV grows with ∆SASA. ∆ZAV is positive for ∆SASA > 0, i.e. when aSASA > bSASA; ∆ZAV is negative for ∆SASA < 0, i.e. when aSASA < bSASA. Overall, the observed experimental trends are in qualitative agreement with the model illustrated in Figure 2 and Scheme 1. On the reliable determination of Ka values from ESI-MS data A typical approach to cope with the manifold of charge-state specific Ka values in native MS analysis is to integrate the abundance of all detected P and PL charge states and thus derive an “averaged” Ka value.43 Yet, it is important to understand that the averaging approach can only be considered optimal if the CSD shift was caused by the ∆SASA factor. In turn, if the CSD difference is caused by non-native factors then the averaging approach cannot be considered optimal. For example, let us assume that the charge-state dependence of PL/P signal intensity ratio was mainly caused by the in-source dissociation of PL ions. In this case, a more realistic Ka value would rather be obtained by only considering the lowest charge states, which are typically more stable than higher charge states, than by integrating the intensity of all the charge states. Nonspecific binding during ionization usually also displays pronounced chargestate dependence. For example, nonspecific metal ion binding has been demonstrated to be most significant for the protein ions with intermediate charge states.58 Note that ∆SASA factor alone, in the absence of other factors, should cause no drift in ∆ZAV with ligand concentration. Thus, the PL systems considered in the present study revealed no notable variation of ∆ZAV with

ACS Paragon Plus Environment

15

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 26

ligand concentration (Figure S4, Table S5). The occasional observation of a distinct shift in ∆ZAV with ligand concentration in experiments40 should be regarded as a direct manifestation for the significant additional contribution of other factors such as in-source fragmentation or nonspecific ligand binding, which usually proceed unevenly with charge state. Based on our findings, the following general approach can be proposed for more reliable quantitative analysis of binding affinities using native ESI-MS. First, when the high-resolution 3D structure of a protein of interest and/or its complex with a ligand is known the ligand-induced ∆SASA can be determined and used to predict the CSD shift. Second, the experimentally observed ∆ZAV should then be compared to the predicted value and guide the adjustment of experimental parameters in order to minimize potential ESI-borne artefacts. Third, titration experiments should be preferred over single-concentration measurements. The occurrence (or the lack) of a ligand concentration-dependent trend in the magnitude and sign of ∆ZAV should allow direct discrimination between the genuine CSD shift due to the ∆SASA factor and experimental artefacts, even when the information on protein 3D structure is absent rendering prediction of ∆SASA impossible. Only after the ∆SASA factor has been proven to be the major cause of CSD asymmetry shift could the spectral peak intensities be safely integrated/averaged over charge states and used for the reliable calculation of the PL/P ratio and determination of Ka value. Certainly, our analysis is not overarching, and further studies are required to confirm the validity and test the general applicability of the proposed hypothesis. Of particular interest is to compare the behavior of specific and nonspecific complexes.

ACS Paragon Plus Environment

16

Page 17 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

ACKNOWLEDGMENT This work was partially supported by the National Natural Science Foundation of China (No. 21765001), Program for Changjiang Scholars and Innovative Research Team in University (PCSIRT) (No. IRT_17R20) and 111 Project (No. D17006).

ASSOCIATED CONTENT Supporting Information The Supporting Information is available free of charge on the ACS Publications website at http://pubs.acs.org/. Methods, Table S1, Figures S1-S4 (.pdf). Tables S2-S5 (.xls).

AUTHOR INFORMATION Corresponding Authors *E-mail: [email protected]; [email protected]. Notes The authors declare no competing financial interest.

ACS Paragon Plus Environment

17

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 26

References (1) Leney, A. C.; Heck, A. J. R. J. Am. Soc. Mass Spectrom. 2017, 28, 5-13. (2) Hopper, J. T. S.; Robinson, C. V. Angew. Chem. Int. Ed. 2014, 53, 14002-14015. (3) Konijnenberg, A.; Butterer, A.; Sobott, F. BBA-Proteins Proteom. 2013, 1834, 1239-1256. (4) Marcoux, J.; Robinson, C. V. Structure 2013, 21, 1541-1550. (5) Konermann, L.; Vahidi, S.; Sowole, M. A. Anal. Chem. 2014, 86, 213-232. (6) Mehmood, S.; Allison, T. M.; Robinson, C. V. Annu. Rev. Phys. Chem. 2015, 66, 453-474. (7) Beveridge, R.; Covill, S.; Pacholarz, K. J.; Kalapothakis, J. M. D.; MacPhee, C. E.; Barran, P. E. Anal. Chem. 2014, 86, 10979-10991. (8) Natalello, A.; Santambrogio, C.; Grandori, R. J. Am. Soc. Mass Spectrom. 2017, 28, 21-28. (9) Li, J.; Santambrogio, C.; Brocca, S.; Rossetti, G.; Carloni, P.; Grandori, R. Mass Spectrom. Rev. 2016, 35, 111-122. (10) Loo, J. A. Mass Spectrom. Rev. 1997, 16, 1-23. (11) Daniel, J. M.; Friess, S. D.; Rajagopalan, S.; Wendt, S.; Zenobi, R. Int. J. Mass Spectrom. 2002, 216, 1– 27. (12) Schug, K. A. Combinatorial Chemistry & High Throughput Screening 2007, 10, 301-316(16). (13) Heck, A. J. R.; Heuvel, R. H. H. V. D. Mass Spectrom. Rev. 2004, 23, 368-89. (14) Ganem, B.; Henion, J. D. Biorg. Med. Chem. 2003, 11, 311-4. (15) Hopper, J. T. S.; Robinson, C. V. Angew. Chem. Int. Ed. 2014, 53, 14002-14015. (16) Wysocki, V.; Quintyn, R.; Harvey, S.; Song, Y.; Yan, J.; Ju, Y.; Tanimoto, A.; Sahasrabuddhe, A. FASEB J. 2015, 29. (17) Rademacher, C.; Shoemaker, G. K.; Kim, H.-S.; Zheng, R. B.; Taha, H.; Liu, C.; Nacario, R. C.; Schriemer, D. C.; Klassen, J. S.; Peters, T.; Lowary, T. L. J. Am. Chem. Soc. 2007, 129, 10489-10502. (18) Jørgensen, T. J. D.; Roepstorff, P.; Heck, A. J. R. Anal. Chem. 1998, 70, 4427-4432. (19) Jecklin, M. C.; Touboul, D.; Bovet, C.; Wortmann, A.; Zenobi, R. J. Am. Soc. Mass Spectrom. 2008, 19, 332-343. (20) Hofstadler, S. A.; Sannes-Lowery, K. A. Nature Reviews Drug Discovery 2006, 5, 585-595. (21) Amr, E. H.; Kitova, E. N.; Lan, L.; Klassen, J. S. J. Am. Soc. Mass Spectrom. 2010, 21, 1893-1899. (22) Erba, E. B.; Zenobi, R. Annu. Rep. Prog. Chem., Sect. C: Phys. Chem. 2011, 107, 199-228. (23) Rogniaux, H.; Sanglier, S.; Strupat, K.; Azza, S. d.; Roitel, O.; Ball, V.; Tritsch, D.; Branlant, G.; Van Dorsselaer, A. Anal. Biochem. 2001, 291, 48-61. (24) Edwards, M. J.; Williams, M. A.; Maxwell, A.; McKay, A. R. Biochemistry 2011, 50, 3432-3440. (25) Dyachenko, A.; Gruber, R.; Shimon, L.; Horovitz, A.; Sharon, M. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 7235-9. (26) Lin, H.; Kitova, E. N.; Klassen, J. S. J. Am. Soc. Mass Spectrom. 2014, 25, 104-110. (27) Cubrilovic, D.; Haap, W.; Barylyuk, K.; Ruf, A.; Badertscher, M.; Gubler, M.; Tetaz, T.; Joseph, C.; Benz, J.; Zenobi, R. Acs Chemical Biology 2014, 9, 218. (28) Cubrilovic, D.; Barylyuk, K.; Hofmann, D.; Walczak, M. J.; Graber, M.; Berg, T.; Wider, G.; Zenobi, R. Chemical Science 2014, 5, 2794-2803. (29) Wolff, P.; Veiga, C. D.; Ennifar, E.; Bec, G.; Guichard, G.; Burnouf, D.; Dumas, P. J. Am. Soc. Mass Spectrom. 2017, 28, 347-357. (30) Hannah, V. V.; Atmanene, C.; Zeyer, D.; Van Dorsselaer, A.; Sanglier-Cianferani, S. Future Med. Chem. 2010, 2, 35-50. (31) Pedro, L.; Quinn, R. J. Molecules 2016, 21, 984. (32) Eschweiler, J. D.; Kerr, R.; Rabuck-Gibbons, J.; Ruotolo, B. T. Annual Review of Analytical Chemistry 2017, 10, 25-44. (33) Kitova, E. N.; Elhawiet, A.; Klassen, J. S. J. Am. Soc. Mass Spectrom. 2014, 25, 1908-1916.

ACS Paragon Plus Environment

18

Page 19 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

(34) Sanhueza, C. A.; Cartmell, J.; Elhawiet, A.; Szpacenko, A.; Kitova, E. N.; Daneshfar, R.; Klassen, J. S.; Lang, D. E.; Eugenio, L.; Ng, K. K. Organic & Biomolecular Chemistry 2015, 13, 283. (35) Han, L.; Shamsuddoha, K.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2016, 88, 8224. (36) Drinkwater, N.; Vu, H.; Lovell, Kimberly M.; Criscione, Kevin R.; Collins, Brett M.; Prisinzano, Thomas E.; Poulsen, S.-A.; McLeish, Michael J.; Grunewald, Gary L.; Martin, Jennifer L. Biochem. J 2010, 431, 51-61. (37) Maple, H. J.; Garlish, R. A.; Rigau-Roca, L.; Porter, J.; Whitcombe, I.; Prosser, C. E.; Kennedy, J.; Henry, A. J.; Taylor, R. J.; Crump, M. P.; Crosby, J. J. Med. Chem. 2012, 55, 837-851. (38) Riccardi, S. F.; Caronni, D.; Colombo, M.; Dalvit, C.; Paolucci, M.; Regazzoni, L.; Visco, C.; Fogliatto, G. European Journal of Pharmaceutical Sciences 2015, 76, 83. (39) Woods, L. A.; Dolezal, O.; Ren, B.; Ryan, J. H.; Peat, T. S.; Poulsen, S. A. J. Med. Chem. 2016, 59, 2192. (40) Göth, M.; Badock, V.; Weiske, J.; Pagel, K.; Kuropka, B. ChemMedChem 2017, 12, 1201. (41) Jecklin, M. C.; Schauer, S.; Dumelin, C. E.; Zenobi, R. Journal of Molecular Recognition 2009, 22, 319329. (42) Schiebel, J.; Radeva, N.; Köster, H.; Metz, A.; Krotzky, T.; Kuhnert, M.; Diederich, W. E.; Heine, A.; Neumann, L.; Atmanene, C.; Roecklin, D.; Vivat-Hannah, V.; Renaud, J.-P.; Meinecke, R.; Schlinck, N.; Sitte, A.; Popp, F.; Zeeb, M.; Klebe, G. ChemMedChem 2015, 10, 1511-1521. (43) Kitova, E. N.; El-Hawiet, A.; Schnier, P. D.; Klassen, J. S. J. Am. Soc. Mass Spectrom. 2012, 23, 431441. (44) Chingin, K.; Barylyuk, K.; Chen, H. Philos. Trans. Royal Soc. A 2016, 374. (45) Ogorzalek Loo, R.; Lakshmanan, R.; Loo, J. J. Am. Soc. Mass Spectrom. 2014, 25, 1675-1693. (46) Hogan, C. J., Jr.; Carroll, J. A.; Rohrs, H. W.; Biswas, P.; Gross, M. L. J. Am. Chem. Soc. 2008, 130, 6926. (47) Hogan, C. J., Jr.; Carroll, J. A.; Rohrs, H. W.; Biswas, P.; Gross, M. L. Anal. Chem. 2009, 81, 369-77. (48) Hogan, C. J., Jr.; de la Mora, J. F. J. Am. Soc. Mass Spectrom. 2011, 22, 158. (49) de la Mora, J. F. Anal. Chim. Acta 2000, 406, 93-104. (50) Kebarle, P.; Verkerk, U. H. Mass Spectrom. Rev. 2009, 28, 898-917. (51) Susa, A. C.; Xia, Z.; Tang, H. Y. H.; Tainer, J. A.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2017, 28, 332-340. (52) Konermann, L.; Ahadi, E.; Rodriguez, A. D.; Vahidi, S. Anal. Chem. 2013, 85, 2-9. (53) McAllister, R. G.; Metwally, H.; Sun, Y.; Konermann, L. J. Am. Chem. Soc. 2015, 137, 12667-12676. (54) Hautreux, M.; Hue, N.; Du Fou de Kerdaniel, A.; Zahir, A.; Malec, V.; Laprévote, O. Int. J. Mass Spectrom. 2004, 231, 131-137. (55) Hall, Z.; Robinson, C. V. J. Am. Soc. Mass Spectrom. 2012, 23, 1161-1168. (56) Testa, L.; Brocca, S.; Grandori, R. Anal. Chem. 2011, 83, 6459-6463. (57) Kaltashov, I. A.; Mohimen, A. Anal. Chem. 2005, 77, 5370-5379. (58) Flick, T. G.; Merenbloom, S. I.; Williams, E. R. J. Am. Soc. Mass Spectrom. 2013, 24, 1654-1662. (59) Breuker, K.; McLafferty, F. W. Proc. Natl. Acad. Sci. U. S. A. 2008, 105, 18145-52. (60) Li, J.; Lyu, W.; Rossetti, G.; Konijnenberg, A.; Natalello, A.; Ippoliti, E.; Orozco, M.; Sobott, F.; Grandori, R.; Carloni, P. Journal of Physical Chemistry Letters 2017, 8, 1105. (61) Arai, M. Biophysical Reviews 2018. (62) Chrysina, E. D.; Brew, K.; Acharya, K. R. J. Biol. Chem. 2000, 275, 37021-37029. (63) Cubrilovic, D.; Zenobi, R. Anal. Chem. 2013, 85, 2724-2730. (64) Cubrilovic, D.; Biela, A.; Sielaff, F.; Steinmetzer, T.; Klebe, G.; Zenobi, R. J. Am. Soc. Mass Spectrom. 2012, 23, 1768-1777. (65) El-Hawiet, A.; Kitova, E. N.; Arutyunov, D.; Simpson, D. J.; Szymanski, C. M.; Klassen, J. S. Anal. Chem. 2012, 84, 3867-3870.

ACS Paragon Plus Environment

19

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 26

(66) Tziridis, A.; Rauh, D.; Neumann, P.; Kolenko, P.; Menzel, A.; Brauer, U.; Ursel, C.; Steinmetzer, P.; Sturzebecher, J.; Schweinitz, A.; Steinmetzer, T.; Stubbs, M. T. Biol Chem 2014, 395, 891-903. (67) Yao, Y.; Shams-Ud-Doha, K.; Daneshfar, R.; Kitova, E.; Klassen, J. J. Am. Soc. Mass Spectrom. 2015, 26, 98-106. (68) Chipuk, J. E.; Moldoveanu, T.; Llambi, F.; Parsons, M. J.; Green, D. R. Molecular cell 2010, 37, 299310. (69) Kitova, E. N.; Yao, Y.; Klassen, J. S. Int. J. Mass Spectrom. 2017, 420, 2-8. (70) Chakravarty, S.; Kannan, K. K. J. Mol. Biol. 1994, 243, 298-309. (71) Cheetham, J. C.; Artymiuk, P. J.; Phillips, D. C. J. Mol. Biol. 1992, 224, 613-28. (72) Sattler, M.; Liang, H.; Nettesheim, D.; Meadows, R. P.; Harlan, J. E.; Eberstadt, M.; Yoon, H. S.; Shuker, S. B.; Chang, B. S.; Minn, A. J.; Thompson, C. B.; Fesik, S. W. Science 1997, 275, 983-986.

ACS Paragon Plus Environment

20

Page 21 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figures and Schemes

Figure 1. The origin of CSD shift of PL relative to P ions in native ESI-MS due to the difference in the degree of charging. (A) Mass spectrum of human carbonic anhydrase I (hCA I) obtained under native ESIMS conditions (protein concentration 10 μM, 50 mM ammonium acetate solution, pH 7.3). (B) Experimental CSD of native hCA I (blue) was shifted by ∆ZAV = 0.07 (≈ 1 % of SASAP) to simulate the CSD of a hypothetical PL complex (orange), with the assumptions that the ligand is present at equimolar amount with protein and occupies 50 % of the available binding sites (Ka = 0.2 μM-1).

ACS Paragon Plus Environment

21

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 26

Scheme 1. The relationship between a shift of the solvent-accessible surface area (∆SASA) and the respective shift of the charge state distribution (∆z) induced upon the binding of a ligand (L) to a rigid protein (P). (A) Upon binding, the surface at the P-L interface becomes buried, i.e. inaccessible to the solvent (bSASA), while a fraction of the ligand surface that is exposed from the binding pocket creates an additional surface area (aSASA). (B) Depending on the ratio between the buried and added surface area, the surface area of PL (SASAPL) can decrease, remain the same, or increase relative to that of P (SASAP). (C) ∆SASA brings about corresponding shift in the charge state distribution of PL relative to P (∆z): ∆z < 0 when bSASA > aSASA (∆SASA < 0); ∆z = 0 when bSASA ≈ aSASA (∆SASA ≈ 0); ∆z > 0 when bSASA < aSASA (∆SASA > 0).

ACS Paragon Plus Environment

22

Page 23 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 2. Distributions of calculated solvent-accessible surface areas (SASA) and average charge states of PL and P ions (ZAV) in three model systems: (A-C) human carbonic anhydrase I (hCA I) binding sulfonamide inhibitors; (D-F) chicken egg lysozyme (LysC) binding N-acetyl-D-glucosamine-containing oligosaccharides; (G-I) human anti-apoptotic protein Bcl-xL binding BH3-domain peptides. The analysis is described in detail in online Supporting Information. Shown in A, D, and G, respectively, are molecular surfaces of the complexes of hCA I with acetazolamide (PDB ID 1AZM70), lysozyme with NAG3 oligosaccharide (PDB ID 1HEW71), and Bcl-xL with Bak572-587 BH3-peptide (PDB ID 1BXL72). SASA of complexes (orange circles) and unbound proteins (blue circles) was compared via a paired t-test (n = 17, 14, and 62 in B, E, and H, respectively; ns – no significant difference, **** - p < 0.0001) to evaluate the statistical significance of ∆SASA = SASAPL – SASAP (grey circles). Example native ESI mass spectra of 10 µM hCA I mixed with 10 µM AZM,41 4.5 µM lysozyme mixed with 10 µM NAG3,63 and 3 µM Bcl-xL mixed with 2 µM Bak BH3-peptide28 are shown in C, F, and I, respectively. The P and PL peak intensities were extracted from the spectra to analyze CSDs (ZAV; blue and orange circles) and measure ∆ZAV = ZAV(PL) – ZAV(P) (grey circles). Statistical significance of the observed ∆ZAV was evaluated by paired ttest (n = 9, 7, and 28 in C, F, and I, respectively; ** – p = 0.0021, ns – no significant difference, **** – p < 0.0001). Horizontal bars with whiskers in the plots indicate mean values with standard deviations.

ACS Paragon Plus Environment

23

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 26

Figure 3. Experimental relationships between the surface area and charging of PL ions in native ESI-MS. (A) A correlation between the relative shift in SASA and the fraction of buried SASA in the total SASA of the ligand for several model PL complexes (fit line a). The data for Bcl-xL – BH3-peptide complexes (grey squares) were analyzed separately from the other protein data (empty circles) and showed a different trend (fitted line b). (B) Correlations of the relative shift of the average charge ∆ZAV/ZAV(P) and the fraction of buried SASA in the total ligand SASA, (C) the relative shift in SASA. The protein complexes

ACS Paragon Plus Environment

24

Page 25 of 26 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

analyzed are listed in Table S1. The empty point in B and C corresponds to trypsin – D-Cha-CMAinhibitor complexes measured in the presence of 10 mM imidazole, which was treated as an outlier. Error bars reflect the heterogeneity of PDB data, as well as variation in ESI mass spectra due to combining data on binding of several ligands to each of the considered proteins, rather than the actual error of measurements and calculations.

ACS Paragon Plus Environment

25

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 26

For TOC only

ACS Paragon Plus Environment

26