Increasing the Depth of Mass-Spectrometry-Based Structural Analysis

Subscriber access provided by ORTA DOGU TEKNIK UNIVERSITESI KUTUPHANESI

Article

Increasing the Depth of Mass Spectrometry-based Structural Analysis of Protein Complexes through the Use of Multiple Cross-linkers Yuehe Ding, Shengbo Fan, Shuang Li, Boya Feng, Ning Gao, Keqiong Ye, Si-Min He, and Meng-Qiu Dong Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b00281 • Publication Date (Web): 24 Mar 2016 Downloaded from http://pubs.acs.org on March 26, 2016

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 free 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 accessible to all readers and 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.

Analytical Chemistry 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 22

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

Increasing the Depth of Mass Spectrometry-based Structural Analysis of Protein Complexes through the Use of Multiple Cross-linkers Yue-He Ding 1,2, Sheng-Bo Fan 3, Shuang Li 1, Bo-Ya Feng 4, Ning Gao4, Keqiong Ye 1

, Si-Min He 3, Meng-Qiu Dong 1,2

1

National Institute of Biological Sciences, Beijing, Beijing 102206, China

2

Graduate Program in Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China

3

Key Lab of Intelligent Information Processing of Chinese Academy of Sciences (CAS); Institute of Computing

Technology of CAS; University of CAS, China. 4

Ministry of Education Protein Science Laboratory, Center for Structural Biology, School of Life Sciences, and

Department of Chemistry, Tsinghua University, Beijing 100084, China

ABSTRACT Chemical cross-linking of proteins coupled with mass spectrometry (CXMS) is a powerful tool to study protein folding and to map the interfaces between interacting proteins. The most commonly used cross-linkers in CXMS are BS3 and DSS, which have similar structures and generate the same linkages between pairs of lysine residues in spatial proximity. However, there are cases where no cross-linkable lysine pairs are present at certain regions of a protein or at the interface of two interacting proteins. In order to find the cross-linkers that can best complement the performance of BS3 and DSS, we tested seven additional cross-linkers that either have different spacer arm structures or that target different amino acids (BS2G, EGS, AMAS, GMBS, Sulfo-GMBS, EDC, and TFCS). Using BSA, Aldolase, the yeast H/ACA protein complex, and E. coli 70S ribosomes, we showed that, in terms of providing structural information not obtained through the use of BS3 and DSS, EGS and Sulfo-GMBS worked better than the other cross-linkers that we tested. EGS generated a large number of cross-links not seen with the other amine-specific cross-linkers, possibly due to its hydrophilic spacer arm. We demonstrate that incorporating the cross-links contributed by the EGS and amine-sulfhydryl cross-linkers greatly increased the accuracy of Rosetta in docking the structure of the yeast H/ACA protein complex. Given the improved depth of useful information it can provide, we suggest that the multi-linker CXMS approach should be used routinely when the amount of a sample permits.

ACS Paragon Plus Environment


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

INTRODUCTION Chemical cross-linking of proteins coupled with mass spectrometry (CXMS) is a recently developed methodology for the study of protein-protein interactions1-8, protein folding9-11, and protein dynamics12. Generally, two amino acids that are close in space have the potential to be covalently connected through a cross-linker molecule. After protease digestion, cross-linked peptides and the cross-linking sites can be identified by MS analysis, yielding low-resolution information about protein folding and protein-protein interactions. For structural analysis of protein complexes, CXMS has three major applications. First, CXMS results alone can be used to locate the regions of interaction between proteins, without having to generate serial truncations of a particular protein6,13,14. Second, provided with subunit structures, intermolecular CXMS distance restraint data can help model the assembly of the subunits within a protein complex15,16. Third, when the resolution of a single-particle cryoEM structure is not high enough, CXMS can be used to help determine the positions of certain subunits or regions of a protein1,5,16,17, especially when homologous proteins or regions are in question. Presently, CXMS is playing an increasingly important role in investigating the spatial arrangements of subunits within a protein complex, especially for large protein complexes that are difficult to crystalize1-5,7,17. To date, the amine-specific cross-linkers BS3 and DSS have been used almost exclusively in CXMS analysis. However, they are not always effective because there are often not enough lysine pairs, or because two target sites and their local environment may disfavor cross-linking (the distance may be too short or too long, the pocket too small, too negatively or positively charged, or too hydrophobic). According to our theoretical calculations based on a total of 1808 complexes selected from the PDB database, about 20% (352) of complexes cannot be expected to have any inter-molecular BS3 or DSS cross-links, and another 19% (342) have no more than five (Supplemental Figure 1). Therefore, other homo- and hetero-bifunctional cross-linkers must be incorporated into workflows to enable comprehensive CXMS analysis; we call this strategy ‘multi-linker CXMS’. A big step in this direction was taken by Leitner et al.18, who developed new cross-linkers that target acidic amino acid residues and demonstrated that they were much improved over a previous strategy19 and were complementary to BS3 and DSS. Other exciting developments include photo-activated chemical cross-linkers20 and photo-activated amino acids that can be incorporated into proteins21. To explore the properties of cross-linkers that best complement BS3 and DSS, we considered targeting non-lysine residues and altering the length and hydrophilicity of cross-linker spacer arms. We tested seven commercially available cross-linkers in addition to BS3 and DSS (Supplemental Figure 2). Among these, four are amine-amine homobifunctional cross-linkers, three are amine-sulfhydryl heterobifunctional cross-linkers, and two are amine-carboxyl heterobifunctional cross-linkers. Of the four amine-amine homobifunctional cross-linkers, BS2G has the shortest spacer arm and EGS has the longest spacer arm. Among the three amine-sulfhydryl heterobifunctional cross-linkers, the spacer arm of AMAS is shorter than those of both GMBS and Sulfo-GMBS. Of the amine-carboxyl heterobifunctional cross-linkers, EDC is a zero-length


Page 2 of 22

Page 3 of 22

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


cross-linker and TFCS has a spacer arm. These cross-linkers also differ in membrane permeability, water solubility, and/or spacer arm structure. Therefore, these cross-linkers should theoretically enable complementary access to various amino acid residues in differing local spatial and energetic environments. This study sought to systematically evaluate the performance of these potentially complementary cross-linkers in CXMS analysis and to assess their respective contributions and utility in the structural modeling of protein complexes. We applied the multi-linker CXMS approach to a four-subunit yeast H/ACA protein complex consisting of Cbf5, Nop10, Gar1, and Nhp222 in an attempt to model the position of Nhp2, whose interactions with the other subunits have evaded characterization using traditional methods. We also applied the multi-linker CXMS approach to a crude E. coli 70S ribosome sample in order to locate peripheral ribosomal proteins and ribosome-associated proteins. Our findings expand the chemical diversity of the cross-linker toolkit of CXMS, and thus extend the utility of CXMS to yet more researchers interested in studying protein-protein interactions and protein folding.

EXPERIMENTAL SECTION Peptide cross-linking The synthesized peptide HR-13 (sequence: HIIGIASYGTDCR) was dissolved, at a final concentration of 5 mM, in 20 mM HEPES, pH 7.5. The cross-linkers were added at a 1:1 molar ratio (peptide to cross-linker). Following cross-linking, the peptide was treated with either TCEP (5 mM), or IAA (10 mM), or FA (5%) alone, or TCEP (5 mM) followed by IAA (10 mM). About 50 pmol of the treated peptide was analyzed by LC-MS (Supplemental Text). Cross-linking of BSA, Aldolase, and the CNGP complex For experimental optimization, several pH values and linker concentrations were used (Supplemental Figure 3 and 4). Proteins were diluted to 0.6 mg/ml with HEPES buffer (20 mM HEPES and 150 mM NaCl). BSA and Aldolase were cross-linked with BS2G, BS3, DSS, EGS, AMAS, GMBS, and Sulfo-GMBS at pH 7.5 for 1 hour, or cross-linked with EDC for 2 hours at pH 7.0. For the TFCS reaction, proteins were first incubated with TFCS at pH 7.2 for 1 hour, then the pH was adjusted to 8.0 and maintained for one additional hour to remove the amine protection group. Following this, the buffer pH was exchanged to 7.0 using an Amicon filter, followed by EDC cross-linking for 2 hours. Reactions for the CNGP samples were all at pH 8.0. The following cross-linker concentrations were used: 0.5 mM for BS2G, BS3, DSS, EGS, GMBS, and Sulfo-GMBS, 1 mM for AMAS and TFCS, and 2 mM for EDC with 3 mM Sulfo-NHS (Supplemental Figure 3 and Supplemental Figure 4). Cross-linking reactions with BS2G, BS3, DSS, and EGS were quenched by the addition of 20 mM NH4HCO3; the AMAS, GMBS, and Sulfo-GMBS reactions were quenched by the addition of 20 mM NH4HCO3 and 20 mM βME; the EDC and TFCS reactions were quenched by the addition of 20 mM Hydroxylamine•HCl.



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

Sulfo-NHS was used in the EDC reaction at a final concentration that was 1.5-fold greater than that of EDC at the molar level, unless noted otherwise. CXMS and data analysis See Supporting Text for details about sample preparation, mass spectrometry analysis, and pLink search parameters. The pLink search results were filtered by requiring FDR < 5%, E-Value < 0.01, and spectrum count ≥ 2. Rosetta docking See Supporting Text for details. The cluster parameters were the same as described by Kahraman et al. 15,16.


Page 4 of 22

Page 5 of 22

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


RESULTS AND DISCUSSION Theoretical calculations To theoretically assess the contribution of the amine-amine cross-linkers for studying protein-protein interactions, we selected 1808 non-redundant, high-resolution (< 3 Å) crystal structures of protein complexes that contained at least two polypeptides with a combined molecular weight of 12 KD or greater from the PDB database (http://www.rcsb.org/). For the purpose of determining the spatial relationships between subunits within a protein complex, we focused on inter-molecular cross-links. Using Xwalk23, we calculated the number of theoretically possible amine-amine, amine-sulfhydryl, and amine-carboxyl cross-links by requiring a Cβ-Cβ Euclidean Distance (ED) of 22 Å or less. This distance cutoff was chosen to match the maximum Cα-Cα distance (24 Å) allowed by DSS and BS3, the two most commonly used cross-linkers in CXMS. Of the 1808 complexes examined, about 20% (352) did not have any inter-molecular BS3 or DSS cross-links, and another 19% (342) had no more than five (Supplemental Figure 1). Clearly, these findings indicate that amine-amine cross-linking with DSS and BS3 is not sufficient for comprehensive CXMS analysis of protein complexes. By adding amine-sulfhydryl or amine-carboxyl cross-links, or both, the percentage of protein complexes with more than 5 virtual inter-molecular cross-links increased from 38% to 69%, 88%, or 89%, respectively, and the percentage of structures harboring more than 25 virtual cross-links increased from 27.8% to 36.4%, 72.0%, or 73.6% (Supplemental Figure 1). These calculations indicate that that the use of heterobifunctional cross-linkers should, in theory, dramatically enhance the scope of CXMS for the analysis of protein complexes. Thus, to expand the cross-linking toolkit available to researchers, we evaluated seven additional, commercially-available cross-linkers with different reaction specificity, arm length, or degree of hydrophilicity.

Optimization of cross-linking reactions Prior to systematically testing the performance of these cross-linkers, we initially optimized the reaction conditions. We used GST, BSA, and a synthetic peptide to test different cross-linking conditions (buffer pH, reaction time, and concentration) for each cross-linker, and chose conditions that yielded the highest number of cross-links and yet were least likely to perturb protein conformations or cause sample loss. EDC reactions are typically involved a complicated a pH change from 6.0 to 7.2. However, we found that a constant pH at 7.0 is just as good, and more importantly, it avoids disturbing the protein conformations (Supplemental Figure 3). The TFCS cross-linking reaction starts at pH 7.2 to allow the NHS ester moiety of TFCS to react with protein primary amines; the pH is then changed to 8.0 to uncage the amine group of TFCS. This amine group then reacts with the COOH group of proteins with the help of a buffer change to pH 7.0 and the addition of EDC. This process caused significant sample loss, and, in the case of the CNGP protein complex, severe perturbation



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

of protein conformation (to the extent of precipitation). Given these issues, we conclude that TFCS has only limited use for CXMS analysis. For cross-linking using amine-sulfhydryl cross-linkers, we recommend the use of 0.5 mM for GMBS and Sulfo-GMBS, 1 mM for AMAS, with pH values ranging between 7.0-7.5 (Supplemental Figure 4). TCEP and IAA treatments have adverse effects on the cross-linking products (Supplemental Figure 5a-c). Thus, TCEP and IAA treatments are not recommended; β-mercaptoethanol (βME) can be used to quench reactions (Supplemental Figure 5d). As an amine-amine cross-linker, the two ester bonds in the linker region of EGS will hydrolyze under either acidic or basic conditions. We encountered a severe hydrolysis problem when EGS cross-linking products were digested at pH 8.5, which is a common pH used in trypsin digestion protocols. For primary mono-links—peptides that are modified by EGS but not linked to a second peptide—the mass addition contributed by EGS is 244.06 Da. However, hydrolysis of the EGS spacer arm can generate two additional forms of mono-linked products with mass additions of 100.01 or 144.04 Da (Supplemental Figure 6). Lowering the digestion pH from 8.5 to 7.0 significantly reduced the spectral counts of the additional forms of mono-links, while the spectral counts of the primary mono-links increased significantly, as did the spectral counts for the loop-linked and inter-linked peptides. In particular, the spectra counts of the inter-linked peptides increased 40-fold following this change in pH. Serving as an internal standard, the spectral counts of the regular, non-cross-linked peptides under these two pH conditions were about the same (Supplemental Figure 6). Therefore, we conclude that EGS cross-linking products should be digested at pH 7.0, and pH 7.0 should be maintained at until immediately before LC-MS/MS analysis, at which point formic acid can be added to a 5% final concentration. The optimized conditions for the above cross-linkers are described in the methods and in Supplemental Figures 3 and 4.

Experimental evaluation of multi-linker CXMS on BSA, Aldolase, and the CNGP complex With the goal of identifying the combination of particular cross-linkers that would enable us to obtain as much structural information as possible with a minimal number of cross-linkers, we tested all nine cross-linkers. BSA (homo-dimer), Aldolase (homo-tetramer), and the yeast H/ACA protein complex consisting of Cbf5, Nop10, Gar1, and Nhp2 at 1:1:1:1 stoichiometry (also referred to as the CNGP protein complex) were used as model systems. Two biological replicate experiments, each with two technical replicates, were conducted using the optimized reaction conditions for each cross-linker (Figure 1a, Supplemental table 1). The TFCS cross-linking result of CNGP was missing, because the proteins aggregated upon the pH change to 7.2. The crystal structure of the Cbf5-Nop10-Gar1 (CNG) sub-complex has been solved22, but the structure of the four-protein CNGP complex remains unknown. Thus, for performance evaluation concerning structural compatibility, only the cross-links of the CNG sub-complex were used at this stage of our study.


Page 6 of 22

Page 7 of 22

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


As shown in Figure1b, these linkers each yielded more than five pairs of cross-linked amino acid residues (site-pairs) for each of the model proteins. DSS and BS3 had the most non-redundant cross-links, followed by EGS and Sulfo-GMBS (Figure 1c). As classified based on the number of non-redundant cross-links observed, the amine-amine cross-linkers tended to perform better than the amine-sulfhydryl or the amine-carboxyl cross-linkers. This is not surprising because the frequency of cysteine in proteins (1.4% in the UniProtKB/Swiss-Prot data bank) is lower than that of lysine (5.8%). Only a small number of cross-links were identified with TFCS for BSA and Aldolase, and none were identified for CNGP, owing to protein precipitation. Next, using the solved crystal structures of these proteins, we evaluated the compatibility of the cross-links obtained in these experiments with the nine cross-linkers (Figure 1b-c). The percentages of structurally compatible cross-links detected with Aldolase and the CNG complex were lower than that detected with BSA, suggesting that the conformations of Aldolase and the CNG complex may be more dynamic than that of BSA22,24. The amine-amine cross-linkers tended to yield more cross-links and higher structure compatibility rates (about 70% on average) than did the other types of cross-linkers, which is consistent with previous CXMS reports6,16,25,26. The amine-sulfhydryl cross-links displayed a wider Cα-Cα distance distribution as compared with the amine-amine cross-links (Supplemental Figure 7), so the distance restraints may need to be relaxed to 30 Å or more for the amine-sulfhydryl cross-links. The EDC cross-links consistently had the lowest structure compatibility rates if EDC was treated strictly as a zero-length cross-linker (Figure 1b-c). However, for 85% of the EDC cross-links, the Cα-Cα distances between the cross-linked lysine and aspartate or glutamate residues were no more than 25 Å. Therefore, we propose that EDC should not be taken as a zero-length cross-linker in structure modeling—a distance restraint of 25 Å may be appropriate for EDC cross-links. Among the amine-amine cross-linkers, DSS seems to be the best performer, as it led to the identification of the largest number of cross-links. Although most of the BS2G or BS3 cross-links were a subset of the DSS cross-links, there were some that were uniquely identified with BS2G or BS3 (Figure 1d and Supplemental Figure 8a). The difference between the BS3 and DSS cross-linking results are probably due to the sulfo group, which is present in BS3 but not DSS and thus may repel BS3 off negatively charged surface on a protein. EGS generated fewer cross-links than DSS, but 37 out of the 119 EGS-identified cross-links were unique. Therefore, EGS significantly increased the number of non-redundant cross-links identified when combined with BS2G, BS3, or DSS (Figure 1e and Supplemental Figure 8b). We speculate that the hydrophilic nature of the spacer arm of EGS is possibly a more important property than the length of its spacer arm. The spacer arm of EGS in its most extended state is longer than those of BS3 or DSS, but EGS rarely adopts this extended conformation and most often it is actually shorter than DSS and BS3 according to MD simulation27 results (Supplemental Figure9); this idea is also supported by our experimental data (Supplemental Figure 7). Among the amine-sulfhydryl cross-linkers, Sulfo-GMBS generated the largest number of cross-links and covered most of the cross-links that were identified with AMAS and GMBS



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

(Figure 1f and Supplemental Figure 8a). However, as with DSS and BS3, cross-links obtained with GMBS and Sulfo-GMBS do not overlap completely. As shown in Supplemental Figure 10, the use of different cross-linkers in the CXMS analyses provided rich information about the 3-D structures of BSA and Aldolase. When mapped to the crystal structures of BSA and Aldolase, it was clear that the great number of intra-protein cross-links (Supplemental Figure 10a-b) identified with the use of the nine cross-linkers covered all regions of the proteins. The inter-protein cross-links obtained from any single cross-linker were relatively sparsely distributed, but there was good complementarity among the different cross-linkers (Supplemental Figure 10c), highlighting the necessity of using multiple cross-linkers to characterize the interface between interacting proteins. Hence, we suggest that a multi-linker CXMS approach be taken in order to obtain relatively more comprehensive structural information for proteins or protein complexes. To do this economically and efficiently, we suggest the use of DSS, EGS, Sulfo-GMBS, and EDC in combination; note that the distance constraints for both Sulfo-GMBS and EDC should be relaxed during structure modeling.

Multi-linker CXMS provides complementary structural information that facilitates structure modeling Box H/ACA ribonucleoprotein particles (RNP) mediate pseudouridine post-transcriptional modification at specific sites of ribosomal RNAs and small nuclear RNAs28-30. The crystal structure of an archaea H/ACA RNP composed of Cbf5, Nop10, Gar1, L7Ae (homologous with Nhp2 in yeast), and the guide RNA has been solved31, but the atomic structure of the eukaryotic H/ACA RNA complex is not available. In 2011, Nhp2 structure has been determined by NMR32 and Li et al.22 reported the structure of the yeast CNG sub-complex (PDB ID:3U28); we used this model in the present study to develop a workflow for multi-linker CXMS-guided structure modeling. Further, because the structure of the eukaryotic CNGP complex remains unknown, we then applied our workflow to model the position of the fourth subunit, Nhp2, relative to the CNG sub-complex. In both of these efforts, the distance restraints that we used came from the CXMS experiments of the CNGP complex using all the cross-linkers described above, with the exception that TFCS was not used because it caused precipitation of CNGP. We used only high-quality cross-links defined as having more than eight spectra identified in the four MS analyses and best E-values < 1.0E-05. We obtained five inter-molecular cross-links (Figure 2a) to model the position of Gar1 relative to Cbf5 and Nop10 in the CNG sub-complex. The Cbf5(134)-Gar1(104) cross-link was uniquely identified with EGS, the Cbf5(190)-Gar1(59) cross-link was identified with GMBS and Sulfo-GMBS, and the other three were identified redundantly with BS2G, BS3,and DSS. Using the Rosetta docking protocol15,33,34, with Cbf5 and Nop10 fixed, we failed to obtain native-like poses for Gar1; all of the conformational clusters predicted with Rosetta docking were at least 20 Å away from the native pose seen in the crystal structure (shown in gray in Figure 2b). When more restraints were included, we obtained more conformational clusters resembling the native structure, and the size of these clusters increased (Figure 2b). Additional filtering and local refinement


Page 8 of 22

Page 9 of 22

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


processes were employed to increase docking efficiency (Supplemental Figure 11a-d). The use of CXMS restraints from BS3 and DSS generated fewer, larger clusters (Supplemental Figure 11e-f), and these clusters more closely resembled the native structure (Figure 2c-d). Incorporating the cross-links from EGS and GMBS/Sulfo-GMBS facilitated Rosetta docking, as the clusters with native-structure-like conformations ranked at the top by cluster size (4 out of the top 6, Figure 2b). Among the six most dominant poses obtained using all five distance restrains, only one deviated significantly from the native conformation (depicted in the graphic table of content), and the representative model from the largest cluster had an ligand-RMSD as low as 2.0 Å compared to the native structure (Figure 2e-f). These findings demonstrated that the CXMS constraints are extremely helpful in protein-protein docking studies. Encouraged by the success with the CNG sub-complex, we then modeled the structure of the yeast CNGP complex using the same workflow. Based on the archaea box H/ACA RNP structure31, Li et al.22 proposed a homology model for the yeast CNGP complex. We first evaluated the Nhp2 model using intra-molecular cross-links from amine-amine cross-linkers; 25 of the 27 total cross-links had Cα-Cα distances less than 24 Å based on the homology model, suggesting that Nhp2 in solution adopts a conformation similar to that in the homology model. Having thusly verified the Nhp2 homology model, we proceeded to model the position of Nhp2 with respect to the CNG sub-complex. No cross-link was observed between Nhp2 and Gar1. Two cross-links were identified between Cbf5 and the N-terminal tail of Nhp2 (1-26) with EGS. Nine cross-links were identified between Nhp2 and Nop10 using BS2G, BS3, DSS, EGS, and EDC (Supplemental Table 4). Of the eleven inter-molecular cross-links identified, four occurred at the N-terminal tail of Nhp2 (Cbf5(180)-Nhp2(6), Cbf5(180)-Nhp2(12), Nhp2(15)-Nop10(19), and Nhp2(9)-Nop10(22)) and thus could not be used for modeling because the first 26 amino acids are missing in the homology model of Nhp2. The remaining seven cross-links, including a unique EGS cross-link and six DSS cross-links, were translated into docking restraints. Similar to the Gar1 analysis, Rosetta docking without CXMS restraints distributed Nhp2 in a spherical shell space enclosing CNG (Figure 3a). Docking with all seven restraints yielded a total of 32 conformational clusters of Nhp2, and they all located atop Nop10 in a small region encompassing what was prescribed to Nhp2 in a previously proposed homology model22 (Figure 3b-c and Supplemental Figure 12). The conformational clusters 2, 6, 11, and 15 satisfied the Solvent Accessible Surface Distances (SASDs)23 restraints imposed by all seven cross-links. The L-RMSD values of the four clusters to the homology model of the yeast CNGP complex22 are 16.7, 10.6, 10.3, and 15.3 Å, respectively (Figure 3d-h). These different poses of Nhp2 likely resulted from the loose and sparse distant restraints provided by cross-links. Of note, docking without the EGS restraint yielded conformational clusters that were smaller in size and less similar to the CNGP homology model (Figure 3c). This again demonstrates that multi-linker CXMS affords more structural information and enhances modeling.

Application of multi-linker CXMS to a crude 70S ribosome



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

Finally, we applied multi-linker CXMS to the analysis of crude 70S ribosomes isolated from E. coli and identified a total of 392 non-redundant cross-links. DSS, EGS and EDC yielded 190, 155, and 48 cross-link identifications, respectively (Supplemental table 2). Less than ten cross-links were identified with GMBS and Sulfo-GMBS because cysteine is rare in ribosomal proteins (0.5% in PDB 3OFA and 3OFC35). BS2G and BS3 together contributed 69 unique cross-links, and EGS alone contributed 58 (Supplemental Figure 13). Of the cross-links that can be mapped to the crystal structure of the 70S ribosome (PDB 3OFA and 3OFC), 153 (80%) are structurally compatible (Figure 4a). As shown, cross-links obtained from different cross-linkers complemented each other well and successfully probed distinct regions of the 70S ribosome (Figure 4). Likewise, the remaining 47 cross-links that were not compatible with the crystal structure came from different cross-linkers and involved yet other regions of the 70S ribosome (Figure 4b). There are multiple explanations for the incompatible cross-links, including the dynamic nature of 70S ribosomes in solution, false identifications, and possible errors in the crystal structure. For example, many incompatible cross-links involve RL9. It has been reported that, as compared to the open conformation of RL9 seen in the crystal structures (3OFA and 3OFC), RL9 takes a closed conformation in the cryoEM structure (5AFI)36. Also, the highly dynamic peripheral ribosome proteins RL1, RL7/RL12, RL10, RL31, and RS1 are missing in the crystal structures (3OFA and 3OFC)35-39. These proteins are known to have important functions in protein translation40. Our use of a series of complimentary cross-linkers in multi-linker CXMS enabled the successfully mapping of the positions of these dynamic peripheral ribosome proteins to the core of the 70S ribosome (Supplemental Figure 14). These results demonstrate that multi-linker CXMS is a powerful tool to investigate the assembly of large protein complexes. CONCLUSIONS In this study, we demonstrate that amine-sulfhydryl, amine-carboxyl, and amine-amine cross-linkers provide useful complementary structural information in the CXMS analysis of protein complexes. Rosetta docking without inter-molecular CXMS distance restraints failed to make the native conformation of the yeast CNG sub-complex stand out. Docking with the CXMS restrains obtained using a single cross-linker improved structure modeling but only to a degree that was insufficient for identification of the native conformation (Figure 2b). When only the inter-molecular distance restraints generated from multiple cross-linkers were all included, a small number of conformational clusters became dominant as they each gained a large cluster size, and the native conformation was easily obtained from the most highly congregated cluster. In fact, four out of the six largest conformational clusters approached the native conformation after local refinement (Figure 2b and Figure 2e). Using the same multi-linker CXMS approach, we also succeeded in modeling the yeast CNGP complex structure by docking Nhp2 to the CNG sub-complex (Figure 3). For the more complicated 70S ribosomes, inter-molecular cross-links from multiple cross-linkers helped to define the position of four peripheral ribosomal proteins (RL1, RL7/RL12, RL31, and RS1). Each cross-linker has unique structural and performance characteristics and cannot be completely replaced by any other cross-linker (Figure 1 and Supplemental Figure 8). For example, most of the


Page 10 of 22

Page 11 of 22

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


lysine pairs cross-linked by BS2G can be cross-linked by BS3 or DSS, but there are BS2G-specific cross-links that are likely facilitated by the fact that the smaller BS2G molecule can reach narrower pockets (Supplemental Figure 15). However, TFCS is not recommended for use because its cross-linking reactions are multi-stepped and require pH shifts, which results in sample loss and protein conformational changes. Considering the number of the cross-links detected with each cross-linker, we conclude that, among the three amine-sulfhydryl cross-linkers, Sulfo-GMBS is the best choice (Supplemental Figure 8a). Among the amine-amine cross-linkers, we conclude that DSS is the best choice, although BS3 is a close second. EGS is the best cross-linker to complement DSS or BS3 because EGS generates many unique cross-links not seen with either DSS or BS3 (Figure 1d-e). However, the two ester bonds in the EGS linker region can be hydrolyzed easily under acidic or basic conditions; this can be a problem when, for example, trypsin digestion is conducted at pH 8.5 or after a digestion reaction is quenched by acidification with formic acid. It might be desirable to replace these ester bonds with ketone or ether bonds to increase the stability of the cross-linked products. Developing new cross-linkers targeting acidic amino acids or arginine will further increase the depth of multi-linker CXMS analysis because they are highly abundant in proteins.



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

ASSOCIATED CONTENT Supporting Information Additional materials as described in the text. This material is available free of charge via the Internet at http://pubs.acs.org

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]

Notes The authors declare no competing financial interests

ACKNOWLEDGEMENTS This work was supported by the National Scientific Instrumentation Grant Program (2011YQ09000506 to M.-Q.D.), the National Natural Science Foundation of China (Grant No. 21375010 to M.-Q.D., 21475141 to S.-M.H., 31422016 to N.G., and 31325007 to K.Y), and Chinese Academy of Sciences (CAS Interdisciplinary Innovation Team grant to S.-M.H.). We thank John Hugh Snyder for language editing.

REFERENCES (1) Lasker, K.; Förster, F.; Bohn, S.; Walzthoeni, T.; Villa, E.; Unverdorben, P.; Beck, F.; Aebersold, R.; Sali, A.; Baumeister, W. Proceedings of the National Academy of Sciences 2012, 109, 1380-1387. (2) Kao, A.; Randall, A.; Yang, Y.; Patel, V. R.; Kandur, W.; Guan, S.; Rychnovsky, S. D.; Baldi, P.; Huang, L. Mol. Cell. Proteomics 2012, 11, 1566-1577. (3) Kalisman, N.; Adams, C. M.; Levitt, M. Proc Natl Acad Sci USA 2012, 109, 2884-2889. (4) Herzog, F.; Kahraman, A.; Boehringer, D.; Mak, R.; Bracher, A.; Walzthoeni, T.; Leitner, A.; Beck, M.; Hartl, F.-U.; Ban, N. Science 2012, 337, 1348-1352. (5) Ciferri, C.; Lander, G. C.; Maiolica, A.; Herzog, F.; Aebersold, R.; Nogales, E. eLife 2012, 1.


Page 12 of 22

Page 13 of 22

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


(6) Yang, B.; Wu, Y. J.; Zhu, M.; Fan, S. B.; Lin, J.; Zhang, K.; Li, S.; Chi, H.; Li, Y. X.; Chen, H. F.; Luo, S. K.; Ding, Y. H.; Wang, L. H.; Hao, Z.; Xiu, L. Y.; Chen, S.; Ye, K.; He, S. M.; Dong, M. Q. Nat. Methods 2012, 9, 904-906. (7) Soares, D. C.; Bradshaw, N. J.; Zou, J.; Kennaway, C. K.; Hamilton, R. S.; Chen, Z. A.; Wear, M. A.; Blackburn, E. A.; Bramham, J.; Böttcher, B. J. Biol. Chem. 2012, 287, 32381-32393. (8) Gong, Z.; Ding, Y.-H.; Dong, X.; Liu, N.; Zhang, E. E.; Dong, M.-Q.; Tang, C. Biophysics Reports 2015, 1-12. (9) Fioramonte, M.; dos Santos, A. M.; McIIwain, S.; Noble, W. S.; Franchini, K. G.; Gozzo, F. C. Proteomics 2012, 12, 2746-2752. (10) Rozbesky, D.; Man, P.; Kavan, D.; Chmelik, J.; Cerny, J.; Bezouska, K.; Novak, P. Anal. Chem. 2011, 84, 867-870. (11) Petrotchenko, E. V.; Serpa, J. J.; Hardie, D. B.; Berjanskii, M.; Suriyamongkol, B. P.; Wishart, D. S.; Borchers, C. H. Mol. Cell. Proteomics 2012, 11. (12) Huang, B. X.; Kim, H.-Y. ACS Chem. Biol. 2011, 7, 387-394. (13) Zhang, J.-M.; Liu, X.-M.; Ding, Y.-H.; Xiong, L.-Y.; Ren, J.-Y.; Zhou, Z.-X.; Wang, H.-T.; Zhang, M.-J.; Yu, Y.; Dong, M.-Q. PLoS Biol. 2014, 12, e1001946. (14) Liu, X.-M.; Sun, L.-L.; Hu, W.; Ding, Y.-H.; Dong, M.-Q.; Du, L.-L. Mol. Cell 2015, 59, 1035-1042. (15) Kahraman, A.; Herzog, F.; Leitner, A.; Rosenberger, G.; Aebersold, R.; Malmström, L. PLoS ONE 2013, 8, e73411. (16) Herzog, F.; Kahraman, A.; Boehringer, D.; Mak, R.; Bracher, A.; Walzthoeni, T.; Leitner, A.; Beck, M.; Hartl, F. U.; Ban, N.; Malmstrom, L.; Aebersold, R. Science 2012, 337, 1348-1352. (17) Plaschka, C.; Larivière, L.; Wenzeck, L.; Seizl, M.; Hemann, M.; Tegunov, D.; Petrotchenko, E.; Borchers, C.; Baumeister, W.; Herzog, F. Nature 2015. (18) Leitner, A.; Joachimiak, L. A.; Unverdorben, P.; Walzthoeni, T.; Frydman, J.; Förster, F.; Aebersold, R. Proceedings of the National Academy of Sciences 2014, 111, 9455-9460. (19) Novak, P.; Kruppa, G. H. European Journal of Mass Spectrometry 2008, 14, 355. (20) Brodie, N. I.; Makepeace, K. A.; Petrotchenko, E. V.; Borchers, C. H. Journal of proteomics 2015, 118, 12-20. (21) Piotrowski, C.; Ihling, C. H.; Sinz, A. Methods 2015, 89, 121-127. (22) Li, S.; Duan, J.; Li, D.; Yang, B.; Dong, M.; Ye, K. Genes Dev. 2011, 25, 2409-2421. (23) Kahraman, A.; Malmström, L.; Aebersold, R. Bioinformatics 2011, 27, 2163-2164.



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

(24) Clifford-Nunn, B.; Showalter, H. H.; Andrews, P. C. J. Am. Soc. Mass Spectrom. 2012, 23, 201-212. (25) Merkley, E. D.; Rysavy, S.; Kahraman, A.; Hafen, R. P.; Daggett, V.; Adkins, J. N. Protein Sci. 2014, 23, 747-759. (26) Rappsilber, J. J. Struct. Biol. 2011, 173, 530-540. (27) Case, D.; Berryman, J.; Betz, R.; Cerutti, D.; Cheatham III, T.; Darden, T.; Duke, R.; Giese, T.; Gohlke, H.; Goetz, A. University of California, San Francisco 2015. (28) Meier, U. T. Chromosoma 2005, 114, 1-14. (29) Matera, A. G.; Terns, R. M.; Terns, M. P. Nature reviews Molecular cell biology 2007, 8, 209-220. (30) Kiss, T.; Fayet-Lebaron, E.; Jády, B. E. Mol. Cell 2010, 37, 597-606. (31) Li, L.; Ye, K. Nature 2006, 443, 302-307. (32) Koo, B.-K.; Park, C.-J.; Fernandez, C. F.; Chim, N.; Ding, Y.; Chanfreau, G.; Feigon, J. Journal of molecular biology 2011, 411, 927-942. (33) Gray, J. J.; Moughon, S.; Wang, C.; Schueler-Furman, O.; Kuhlman, B.; Rohl, C. A.; Baker, D. J. Mol. Biol. 2003, 331, 281-299. (34) Wang, C.; Bradley, P.; Baker, D. J. Mol. Biol. 2007, 373, 503-519. (35) Dunkle, J. A.; Xiong, L.; Mankin, A. S.; Cate, J. H. Proceedings of the National Academy of Sciences 2010, 107, 17152-17157. (36) Fischer, N.; Neumann, P.; Konevega, A. L.; Bock, L. V.; Ficner, R.; Rodnina, M. V.; Stark, H. Nature 2015. (37) Unge, J.; Al-Karadaghi, S.; Liljas, A.; Jonsson, B.-H.; Eliseikina, I.; Ossina, N.; Nevskaya, N.; Fomenkova, N.; Garber, M.; Nikonov, S. FEBS Lett. 1997, 411, 53-59. (38) Diaconu, M.; Kothe, U.; Schlünzen, F.; Fischer, N.; Harms, J. M.; Tonevitsky, A. G.; Stark, H.; Rodnina, M. V.; Wahl, M. C. Cell 2005, 121, 991-1004. (39) Lauber, M. A.; Rappsilber, J.; Reilly, J. P. Mol. Cell. Proteomics 2012, 11, 1965-1976. (40) Stelzl, U.; Connell, S.; Nierhaus, K. H.; Wittmann‐Liebold, B. eLS 2001.


Page 14 of 22

Page 15 of 22

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


Figure legends Figure 1. Comparison of nine cross-linkers using purified proteins. (a) Workflow of cross-linking experiments using BSA, Aldolase, and the CNGP complex. (b) Cross-links identified with each cross-linker for each test sample, expressed as the number of amino acid pairs (site pairs) that were cross-linked together. The percentages of structurally compatible cross-links are labeled on the top. Inter- and intra-molecular cross-links are colored red and blue, respectively. Cross-links involving Nhp2 are not included, as there is no available structure for this protein in the complex. (c) Tabulated summary of the cross-links identified with each cross-linker, ranked by the total number of site pairs. Com% indicates the percentage of cross-links that are structurally compatible. (d) Venn diagram showing the overlap of site pairs identified with BS2G, BS3, DSS, and EGS. (e) Accumulation curves of identified non-redundant cross-links showing distinct contributions from EGS. (f) Venn diagram showing the overlap of site pairs identified with AMAS, GMBS, and Sulfo-GMBS.



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 ACS Paragon Plus Environment

Page 16 of 22

Page 17 of 22

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


Figure 2. Rosetta docking of Gar1 to Cbf5-Nop10 with or without CXMS distance restraints. (a) Cross-links identified between Gar1 and Cbf5 are mapped to the yeast CNG sub-complex

structure (PDB code: 3U2822). Cbf5(134)-Gar1(104) is a unique EGS cross-link and is shown in green. Cbf5(190)-Gar1(59) was identified with GMBS and Sulfo-GMBS and is shown in yellow. Three cross-link identified with BS3 and DSS are shown in red. (b) A summary of Rosetta docking results with or without CXMS restraints. Each dot represents a cluster of conformations obtained, and the x- and y-axis values indicate cluster size (number of poses in a cluster) and ligand-RMSD (the distance between a representative pose of a cluster and the native structure22), respectively. (c) Rosetta docking in the absence of CXMS restraints resulted in 32 conformational clusters, represented here by the best models (gray), one for each cluster, after local refinement. The native structure is presented with Cbf5 in green, Nop10 in pale cyan, and Gar1 in pink. (d) Rosetta docking with five inter-molecular distance restraints resulted in 12 conformational clusters, represented here in the same way as in (c). (e) Hierarchical clustering of 12 clusters from Rosetta docking of Gar1 to Cbf5-Nop10 with all five restraints considered. The clusters are numbered by their rank in cluster size—cluster 1 has the largest cluster size, cluster 2 the second largest, and so on. (f) Clusters 1, 3, 4, and 6 differ from the native structure (3U28) by less than 3.0 Å in L-RMSD.




Page 18 of 22

Page 19 of 22

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


Figure 3. Rosetta docking of Nhp2 to Cbf5-Nop10-Gar1 with seven CXMS distance restraints.

Conformations obtained without (a) or with the CXMS restraints (b). (c) A summary of Rosetta docking results with no CXMS restraints (gray), with six DSS cross-links (green), or with six DSS plus one EGS cross-links taken into account (red). Each dot represents a cluster of conformations, and the x- and y-axis values indicate cluster size (number of poses in a cluster) and ligand-RMSD (the distance between a representative pose of a cluster and the CNGP homology model), respectively. The homology model (d) is shown with representative models of clusters 2, 6, 11, and 15 (e-h). The seven cross-linked site pairs are indicated as dashed lines (blue for DSS and red for EGS), and the cross-linked residues are colored red.




Page 20 of 22

Page 21 of 22

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


Figure 4. Multi-linker CXMS analysis of E. coli 70S ribosomes Cross-links obtained with different cross-linkers are colored as follows: green for BS2G, lavender for BS3, dark blue for DSS, orange for EGS, light blue for GMBS or Sulfo-GMBS, and yellow for EDC. Cross-links that are compatible with the crystal structure of the 70S ribosome (3OFA and 3OFC) are shown in (a) and those that are not compatible are shown in (b). For overlapping amine-amine cross-links, the color labeling follows the hierarchy: BS3, DSS, BS2G, EGS. For example, if a cross-link is identified with both BS3 and DSS, it is colored as a BS3 cross-link.



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

For TOC only


Page 22 of 22

Increasing the Depth of Mass-Spectrometry-Based Structural Analysis

Recommend Documents