Use of proteinase K nonspecific digestion for selective and comprehensive identification of interpeptide cross-links: application to prion proteins

Chemical cross-linking combined with mass spectrometry is a rapidly developing technique for structural proteomics. Cross-linked proteins are usually digested with trypsin to generate cross-linked peptides, which are then analyzed by mass spectrometry. The most informative cross-links, the interpeptide cross-links, are often large in size, because they consist of two peptides that are connected by a cross-linker. In addition, trypsin targets the same residues as amino-reactive cross-linkers, and cleavage will not occur at these cross-linker-modified residues. This produces high molecular weight cross-linked peptides, which complicates their mass spectrometric analysis and identification. In this paper, we examine a nonspecific protease, proteinase K, as an alternative to trypsin for cross-linking studies. Initial tests on a model peptide that was digested by proteinase K resulted in a "family" of related cross-linked peptides, all of which contained the same cross-linking sites, thus providing additional verification of the cross-linking results, as was previously noted for other post-translational modification studies. The procedure was next applied to the native (PrP(C)) and oligomeric form of prion protein (PrPβ). Using proteinase K, the affinity-purifiable CID-cleavable and isotopically coded cross-linker cyanurbiotindipropionylsuccinimide and MALDI-MS cross-links were found for all of the possible cross-linking sites. After digestion with proteinase K, we obtained a mass distribution of the cross-linked peptides that is very suitable for MALDI-MS analysis. Using this new method, we were able to detect over 60 interpeptide cross-links in the native PrP(C) and PrPβ prion protein. The set of cross-links for the native form was used as distance constraints in developing a model of the native prion protein structure, which includes the 90-124-amino acid N-terminal portion of the protein. Several cross-links were unique to each form of the prion protein, including a Lys(185)-Lys(220) cross-link, which is unique to the PrPβ and thus may be indicative of the conformational change involved in the formation of prion protein oligomers.     

Isotopically coded cleavable cross-linker for studying protein-protein interaction and protein complexes

An emerging approach for studying protein-protein interaction in complexes is the combination of chemical cross-linking and mass spectrometric analysis of the cross-linked peptides (cross-links) obtained after proteolysis of the complex. This approach, however, has several challenges and limitations, including the difficulty of detecting the cross-links, the potential interference from non-informative "cross-linked peptides" (dead end and intrapeptide cross-links), and unambiguous identification of the cross-links by mass spectrometry. Thus, we have synthesized an isotopically coded ethylene glycol bis(succinimidylsuccinate) derivate (D12-EGS), which contains 12 deuterium atoms for easy detection of cross-links when applied in a 1:1 mixture with its H12 counterpart and is also cleavable for releasing the cross-linked peptides allowing unambiguous identification by MS sequencing. Moreover, hydrolytic cleavage permits rapid distinguishing between different types of cross-links. Cleavage of a dead end cross-link produces a doublet with peaks 4.03 Da apart, with the lower peak appearing at a molecular mass 162 Da lower than the mass of the H12 form of the original cross-linked peptide. Cleavage of an intrapeptide cross-link leads to a doublet 8.05 Da apart and 62 Da lower than the molecular mass of the H12 form of the original cross-linked peptide. Cleavage of an interpeptide cross-link forms a pair of 4.03-Da doublets, with the lower mass member of each pair each shifted up from its unmodified molecular weight by 82 Da because of the attached portion of the cross-linker. All of this information has been incorporated into a software algorithm allowing automatic screening and detection of cross-links and cross-link types in matrix-assisted laser desorption/ionization mass spectra. In summary, the ease of detection of these species through the use of an isotopically coded cleavable cross-linker and our software algorithm, followed by mass spectrometric sequencing of the cross-linked peptides after cleavage, has been shown to be a powerful tool for studies of multi-component protein complexes.     

Protein cross-linking analysis using mass spectrometry, isotope-coded cross-linkers, and integrated computational data processing

Distance constraints in proteins and protein complexes provide invaluable information for calculation of 3D structures, identification of protein binding partners and localization of protein-protein contact sites. We have developed an integrative approach to identify and characterize such sites through the analysis of proteolytic products derived from proteins chemically cross-linked by isotopically coded cross-linkers using LC-MALDI tandem mass spectrometry and computer software. This method is specifically tailored toward the rapid analysis of low microgram amounts of proteins or multimeric protein complexes cross-linked with nonlabeled and deuterium-labeled bis-NHS ester cross-linking reagents (both commercially available and readily synthesized). Through labeling with [18O]water solvent and LC-MALDI analysis, the method further allows the possible distinction between Type 0 and Type 1 or Type 2 modified peptides (monolinks and looplinks or cross-links), although such a distinction is more readily made from analysis of tandem mass spectrometry data. When applied to the bacterial Colicin E7 DNAse/Im7 heterodimeric protein complex, 23 cross-links were identified including six intersubunit cross-links, all between residues that are close in space when examined in the context of the X-ray structure of the heterodimer. In addition, cross-links were successfully identified in five single subunit proteins, beta-lactoglobulin, cytochrome c, lysozyme, myoglobin, and ribonuclease A, establishing the generality of the approach.     

Analysis of site-specific protein-RNA cross-links in isolated RNP complexes, combining affinity selection and mass spectrometry

An important aspect of the assembly of RNPs, and in particular of spliceosomes, is the succession of proteins bound to any given site on the RNA. Protein-RNA cross-linking is a well-established technique for investigating this, but the identification of a cross-linked protein has so far relied upon the availability of antibodies for immunoprecipitation or Western blot studies. To facilitate identification of proteins independent of these techniques, site-specific protein-RNA cross-links were purified in a large scale, which were then used for mass spectrometry (MS). This approach was carried out by the use of a minimal pre-mRNA construct containing a single photoactivatable azidophenacyl group and an adjacent biotin-dT tag for affinity purification of the cross-linked product. To test the feasibility of the method, we purified cross-links to nucleotide 9 downstream of the 5' splice site of pre-mRNA in the spliceosomal complexes A ("pre-spliceosome") and H. By this method, we were able to identify several proteins by MS; the hnRNP proteins A2/B1 were cross-linked to the pre-mRNA in complex A, and FUSE 2/FBP (a homolog of the intronic splicing enhancer KSRP) was cross-linked in complex H.     

Quantification of protein-protein interactions with chemical cross-linking and mass spectrometry

Chemical cross-linking in combination with mass spectrometry has largely been used to study protein structures and protein-protein interactions. Typically, it is used in a qualitative manner to identify cross-linked sites and provide a low-resolution topological map of the interacting regions of proteins. Here, we investigate the capability of chemical cross-linking to quantify protein-protein interactions using a model system of calmodulin and substrates melittin and mastoparan. Calmodulin is a well-characterized protein which has many substrates. Melittin and mastoparan are two such substrates which bind to calmodulin in 1:1 ratios in the presence of calcium. Both the calmodulin-melittin and calmodulin-mastoparan complexes have had chemical cross-linking strategies successfully applied in the past to investigate topological properties. We utilized an excess of immobilized calmodulin on agarose beads and formed complexes with varying quantities of mastoparan and melittin. Then, we applied disuccinimidyl suberate (DSS) chemical cross-linker, digested and detected cross-links through an LC-MS analytical method. We identified five interpeptide cross-links for calmodulin-melittin and three interpeptide cross-links for calmodulin-mastoparan. Using cross-linking sites of calmodulin-mastoparan, we demonstrated that mastoparan also binds in two orientations to calmodulin. We quantitatively demonstrated that both melittin and mastoparan preferentially bind to calmodulin in a parallel fashion, which is opposite to the preferred binding mode of the majority of known calmodulin binding peptides. We also demonstrated that the relative abundances of cross-linked peptide products quantitatively reflected the abundances of the calmodulin peptide complexes formed.     

Covalent cross-linking of proteins without chemical reagents

A facile method for the formation of zero-length covalent cross-links between protein molecules in the lyophilized state without the use of chemical reagents has been developed. The cross-linking process is performed by simply sealing lyophilized protein under vacuum in a glass vessel and heating at 85 degrees C for 24 h. Under these conditions, approximately one-third of the total protein present becomes cross-linked, and dimer is the major product. Chemical and mass spectroscopic evidence obtained shows that zero-length cross-links are formed as a result of the condensation of interacting ammonium and carboxylate groups to form amide bonds between adjacent molecules. For the protein examined in the most detail, RNase A, the cross-linked dimer has only one amide cross-link and retains the enzymatic activity of the monomer. The in vacuo cross-linking procedure appears to be general in its applicability because five different proteins tested gave substantial cross-linking, and co-lyophilization of lysozyme and RNase A also gave a heterogeneous covalently cross-linked dimer.     

Identification of the interactions between cytochrome P450 2E1 and cytochrome b5 by mass spectrometry and site-directed mutagenesis

The reaction cycles of cytochrome P450s (P450) require input of two electrons. Electrostatic interactions are considered important driving forces in the association of P450s with their redox partners, which in turn facilitates the transfer of the two electrons. In this study, the cross-linking reagent, 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC), was used to covalently link cytochrome P450 2E1 (CYP2E1) with cytochrome b(5) (b(5)) through the formation of specific amide bonds between complementary charged residue pairs. Cross-linked peptides in the resulting protein complex were distinguished from non-cross-linked peptides using an (18)O-labeling method on the basis that cross-linked peptides incorporate twice as many (18)O atoms as non-cross-linked peptides during proteolysis conducted in (18)O-water. Subsequent tandem mass spectrometric (MS/MS) analysis of the selected cross-linked peptide candidates led to the identification of two intermolecular cross-links, Lys(428)(CYP2E1)-Asp(53)(b(5)) and Lys(434)(CYP2E1)-Glu(56)(b(5)), which provides the first direct experimental evidence for the interacting orientations of a microsomal P450 and its redox partner. The biological importance of the two ion pairs for the CYP2E1-b(5) interaction, and the stimulatory effect of b(5), was confirmed by site-directed mutagenesis. Based on the characterized cross-links, a CYP2E1-b(5) complex model was constructed, leading to improved insights into the protein interaction. The described method is potentially useful for mapping the interactions of various P450 isoforms and their redox partners, because the method is relatively rapid and sensitive, and is capable of suggesting not only protein interacting regions, but also interacting orientations.     

Expanding the chemical cross-linking toolbox by the use of multiple proteases and enrichment by size exclusion chromatography

Chemical cross-linking in combination with mass spectrometric analysis offers the potential to obtain low-resolution structural information from proteins and protein complexes. Identification of peptides connected by a cross-link provides direct evidence for the physical interaction of amino acid side chains, information that can be used for computational modeling purposes. Despite impressive advances that were made in recent years, the number of experimentally observed cross-links still falls below the number of possible contacts of cross-linkable side chains within the span of the cross-linker. Here, we propose two complementary experimental strategies to expand cross-linking data sets. First, enrichment of cross-linked peptides by size exclusion chromatography selects cross-linked peptides based on their higher molecular mass, thereby depleting the majority of unmodified peptides present in proteolytic digests of cross-linked samples. Second, we demonstrate that the use of proteases in addition to trypsin, such as Asp-N, can additionally boost the number of observable cross-linking sites. The benefits of both SEC enrichment and multiprotease digests are demonstrated on a set of model proteins and the improved workflow is applied to the characterization of the 20S proteasome from rabbit and Schizosaccharomyces pombe.     

Mapping spatial proximities of sulfhydryl groups in proteins using a fluorogenic cross-linker and mass spectrometry

Chemical cross-linking of proteins in combination with mass spectrometric analysis of the reaction products has gained renewed interest as a method of obtaining distance constraints within a protein and determining a low-resolution three-dimensional structure. We present a method for identifying spatially close sulfhydryl groups in proteins employing chemical cross-linking with the fluorogenic, homobifunctional cross-linker dibromobimane, which cross-links thiol pairs within approximately 3-6A. The applicability of our strategy was demonstrated by cross-linking the sulfhydryl groups of Cys-18 and Cys-78 in gamma-crystallin F, which are within a distance of 3.57A according to the X-ray structure. Intramolecularly cross-linked gamma-crystallin was first separated from reaction side products by reversed-phase chromatography on a C-4 column. Subsequently, the fraction containing the reacted protein was enzymatically digested with trypsin, and the resulting peptide mixture was separated by a second reversed-phase chromatographic step on a C-18 column, in which the cross-linked peptides were tracked by their fluorescence. The cross-linking product between Cys-18 and Cys-78 in gamma-crystallin F was identified by matrix-assisted laser desorption/ionization-time of flight mass spectrometry. This strategy presents a rapid method for mapping sulfhydryl groups separated by a distance of approximately 3-6A within a protein.     

BiPS, a photocleavable, isotopically coded, fluorescent cross-linker for structural proteomics

Cross-linking combined with mass spectrometry is an emerging approach for studying protein structure and protein-protein interactions. However, unambiguous mass spectrometric identification of cross-linked peptides derived from proteolytically digested cross-linked proteins is still challenging. Here we describe the use of a novel cross-linker, bimane bisthiopropionic acid N-succinimidyl ester (BiPS), that overcomes many of the challenges associated with other cross-linking reagents. BiPS is distinguished from other cross-linkers by a unique combination of properties: it is photocleavable, fluorescent, homobifunctional, amine-reactive, and isotopically coded. As demonstrated with a model protein complex, RNase S, the fluorescent moiety of BiPS allows for sensitive and specific monitoring of the different cross-linking steps, including detection and isolation of cross-linked proteins by gel electrophoresis, determination of in-gel digestion completion, and fluorescence-based separation of cross-linked peptides by HPLC. The isotopic coding of BiPS results in characteristic ion signal "doublets" in mass spectra, thereby permitting ready detection of cross-linker-containing peptides. Under MALDI-MS conditions, partial photocleavage of the cross-linker occurs, releasing the cross-linked peptides. This allows differentiation between dead-end, intra-, and interpeptide cross-links based on losses of specific mass fragments. It also allows the use of the isotope doublets as mass spectrometric "signatures." A software program was developed that permits automatic cross-link identification and assignment of the cross-link type. Furthermore photocleavage of BiPS assists in cross-link identification by allowing separate tandem mass spectrometry sequencing of each peptide comprising the original cross-link. By combining the use of BiPS with MS, we have provided the first direct evidence for the docking site of a phosphorylated G-protein-coupled receptor C terminus on the multifunctional adaptor protein beta-arrestin, clearly demonstrating the broad potential and application of this novel cross-linker in structural and cellular biology.     

The structure of apolipoprotein A-II in discoidal high density lipoproteins

It is well accepted that high levels of high density lipoproteins (HDL) reduce the risk of atherosclerosis in humans. Apolipoprotein A-I (apoA-I) and apoA-II are the first and second most common protein constituents of HDL. Unlike apoA-I, detailed structural models for apoA-II in HDL are not available. Here, we present a structural model of apoA-II in reconstituted HDL (rHDL) based on two well established experimental approaches: chemical cross-linking/mass spectrometry (MS) and internal reflection infrared spectroscopy. Homogeneous apoA-II rHDL were reacted with a cross-linking agent to link proximal lysine residues. Upon tryptic digestion, cross-linked peptides were identified by electrospray mass spectrometry. 14 cross-links were identified and confirmed by tandem mass spectrometry (MS/MS). Infrared spectroscopy indicated a beltlike molecular arrangement for apoA-II in which the protein helices wrap around the lipid bilayer rHDL disc. The cross-links were then evaluated on three potential belt arrangements. The data clearly refute a parallel model but support two antiparallel models, especially a "double hairpin" form. These models form the basis for understanding apoA-II structure in more complex HDL particles.     

Shotgun cross-linking analysis for studying quaternary and tertiary protein structures

We developed a new approach that employs a novel computer algorithm for the sensitive and high-throughput analysis of tertiary and quaternary interaction sites from chemically cross-linked proteins or multi-protein complexes. First, we directly analyze the digests of the chemically cross-linked proteins using only high-accuracy LC-MS/MS data. We analyze these data using a computer algorithm, we term X!Link, to find cross-links between two peptides. Our algorithm is rapid, taking only a few seconds to analyze approximately 5000 MS/MS spectra. We applied this algorithm to analyze cross-linked sites generated chemically using the amino specific reagent, BS3, in both cytochrome c and the mitochondrial division dynamin mutant, Dnm1G385D, which exists as a stable homodimer. From cytochrome c, a well-established test protein, we identified a total of 31 cross-links, 21 interpeptide and 10 intrapeptide cross-links, in 257 MS/MS spectra from a single LC-MS/MS data set. The high sensitivity of this technique is indicated by the fact that all 19 lysines in cytochrome c were detected as a cross-link product and 33% of all the Lys pairs within 20 A were also observed as a cross-link. Analysis of the cross-linked dimeric form of Dnm1G385D identified a total of 46 cross-links, 38 interpeptide and 8 intrapeptide cross-links, in 98 MS/MS spectra in a single LC-MS/MS data set. These results represent the most abundant cross-links identified in a single protein or protein dimer to date. Statistical analysis suggests a 1% false discovery rate after optimization of filtering parameters. Further analysis of the cross-links identified using our approach indicates that careful manual inspection is important for the correct assignment of cross-linking sites when multiple cross-linkable sites or several similar sequences exist. In summary, we have developed a sensitive MS-based approach to identify peptide-peptide cross-links that does not require isotopic labeling or comparison with non-cross-linked controls, making it faster and simpler than current methodologies.     

Chemical cross-linking and mass spectrometry to determine the subunit interaction network in a recombinant human SAGA HAT subcomplex

Understanding the way how proteins interact with each other to form transient or stable protein complexes is a key aspect in structural biology. In this study, we combined chemical cross-linking with mass spectrometry to determine the binding stoichiometry and map the protein–protein interaction network of a human SAGA HAT subcomplex. MALDI-MS equipped with high mass detection was used to follow the cross-linking reaction using bis[sulfosuccinimidyl] suberate (BS3) and confirm the heterotetrameric stoichiometry of the specific stabilized subcomplex. Cross-linking with isotopically labeled BS3 d0-d4 followed by trypsin digestion allowed the identification of intra- and intercross-linked peptides using two dedicated search engines: pLink and xQuest. The identified interlinked peptides suggest a strong network of interaction between GCN5, ADA2B and ADA3 subunits; SGF29 is interacting with GCN5 and ADA3 but not with ADA2B. These restraint data were combined to molecular modeling and a low-resolution interacting model for the human SAGA HAT subcomplex could be proposed, illustrating the potential of an integrative strategy using cross-linking and mass spectrometry for addressing the structural architecture of multiprotein complexes.     

In vitro cross-linking of elastin peptides and molecular characterization of the resultant biomaterials

Background

Elastin is a vital protein and the major component of elastic fibers which provides resilience to many vertebrate tissues. Elastin's structure and function are influenced by extensive cross-linking, however, the cross-linking pattern is still unknown.

Methods

Small peptides containing reactive allysine residues based on sequences of cross-linking domains of human elastin were incubated in vitro to form cross-links characteristic of mature elastin. The resultant insoluble polymeric biomaterials were studied by scanning electron microscopy. Both, the supernatants of the samples and the insoluble polymers, after digestion with pancreatic elastase or trypsin, were furthermore comprehensively characterized on the molecular level using MALDI-TOF/TOF mass spectrometry.

Results

MS² data was used to develop the software PolyLinX, which is able to sequence not only linear and bifunctionally cross-linked peptides, but for the first time also tri- and tetrafunctionally cross-linked species. Thus, it was possible to identify intra- and intermolecular cross-links including allysine aldols, dehydrolysinonorleucines and dehydromerodesmosines. The formation of the tetrafunctional cross-link desmosine or isodesmosine was unexpected, however, could be confirmed by tandem mass spectrometry and molecular dynamics simulations.

Conclusions

The study demonstrated that it is possible to produce biopolymers containing polyfunctional cross-links characteristic of mature elastin from small elastin peptides. MALDI-TOF/TOF mass spectrometry and the newly developed software PolyLinX proved suitable for sequencing of native cross-links in proteolytic digests of elastin-like biomaterials.

General significance

The study provides important insight into the formation of native elastin cross-links and represents a considerable step towards the characterization of the complex cross-linking pattern of mature elastin.     

Identification of protein-protein interactions using in vivo cross-linking and mass spectrometry

The purification of protein complexes can be accomplished by different types of affinity chromatography. In a typical immunoaffinity experiment, protein complexes are captured from a cell lysate by an immobilized antibody that recognizes an epitope on one of the known components of the complex. After extensive washing to remove unspecifically bound proteins, the complexes are eluted and analyzed by mass spectrometry (MS). Transient complexes, which are characterized by high dissociation constants, are typically lost by this approach. In the present study, we describe a novel method for identifying transient protein-protein interactions using in vivo cross-linking and MS-based protein identification. Live cells are treated with formaldehyde, which rapidly permeates the cell membrane and generates protein-protein cross-links. Proteins cross-linked to a Myc-tagged protein of interest are copurified by immunoaffinity chromatography and subjected to a procedure which dissociates the cross-linked complexes. After separation by SDS-PAGE, proteins are identified by tandem mass spectrometry. Application of this method enabled the identification of numerous proteins that copurified with a constitutively active form of M-Ras (M-Ras(Q71L)). Among these, we identified the RasGAP-related protein IQGAP1 to be a novel interaction partner of M-Ras(Q71L). This method is applicable to many proteins and will aid in the study of protein-protein interactions.     

Improved identification of enriched peptide RNA cross-links from ribonucleoprotein particles (RNPs) by mass spectrometry

Direct UV cross-linking combined with mass spectrometry (MS) is a powerful tool to identify hitherto non-characterized protein-RNA contact sites in native ribonucleoprotein particles (RNPs) such as the spliceosome. Identification of contact sites after cross-linking is restricted by: (i) the relatively low cross-linking yield and (ii) the amount of starting material available for cross-linking studies. Therefore, the most critical step in such analyses is the extensive purification of the cross-linked peptide-RNA heteroconjugates from the excess of non-crosslinked material before MS analysis. Here, we describe a strategy that combines small-scale reversed-phase liquid chromatography (RP-HPLC) of UV-irradiated and hydrolyzed RNPs, immobilized metal-ion affinity chromatography (IMAC) to enrich cross-linked species and their analysis by matrix-assisted laser desorption/ionisation (MALDI) MS(/MS). In cases where no MS/MS analysis can be performed, treatment of the enriched fractions with alkaline phosphatase leads to unambiguous identification of the cross-linked species. We demonstrate the feasibility of this strategy by MS analysis of enriched peptide-RNA cross-links from UV-irradiated reconstituted [15.5K-61K-U4atac snRNA] snRNPs and native U1 snRNPs. Applying our approach to a partial complex of U2 snRNP allowed us to identify the contact site between the U2 snRNP-specific protein p14/SF3b14a and the branch-site interacting region (BSiR) of U2 snRNA.     

Oxidative modification of nucleoside diphosphate kinase and its identification by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry

Nucleoside diphosphate kinase (NDPK, Nm23) has been implicated as a multifunctional protein. However, the regulatory mechanism of NDPK is poorly understood. We have examined the modification of NDPK in oxidative stresses. We found that oxidative stresses including diamide and H(2)O(2) treatment cause disulfide cross-linking of NDPK inside cells. This cross-linking was reversible in response to mild oxidative stress, and irreversible to strong stress. This suggests that disulfide cross-linked NDPK may be a possible mechanism in the modification of cellular regulation. To confirm this idea, oxidative modification of NDPK has been performed in vitro using purified human NDPK H(2)O(2) inactivated the nucleoside diphosphate (NDP) kinase activity of NDPK by producing intermolecular disulfide bonds. Disulfide cross-linking of NDPK also dissociated the native hexameric structure into a dimeric form. The oxidation sites were identified by the analysis of tryptic peptides of oxidized NDPK, using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS). Intermolecular cross-linking between Cys109-Cys109, which is highly possible based on the X-ray crystal structure of NDPK-A, and oxidations of four methionine residues were identified in H(2)O(2)-treated NDPK. This cross-linkng was confirmed using mutant C109A (NDPK-A(C109A)) which had similar enzymatic activity as a wild NDPK-A. Mutant NDPK-A(C109A) was not cross-linked and was not easily denatured by the oxidant. Therefore, enzymatic activity and the quaternary structure of NDPK appear to be regulated by cross-linking with oxidant. These findings suggest one of the regulatory mechanisms of NDPK in various cellular processes.     

Integrating Cross-Linking Experiments with Ab Initio Protein–Protein Docking

Ab initio protein–protein docking algorithms often rely on experimental data to identify the most likely complex structure. We integrated protein–protein docking with the experimental data of chemical cross-linking followed by mass spectrometry. We tested our approach using 19 cases that resulted from an exhaustive search of the Protein Data Bank for protein complexes with cross-links identified in our experiments. We implemented cross-links as constraints based on Euclidean distance or void-volume distance. For most test cases, the rank of the top-scoring near-native prediction was improved by at least twofold compared with docking without the cross-link information, and the success rate for the top 5 predictions nearly tripled. Our results demonstrate the delicate balance between retaining correct predictions and eliminating false positives. Several test cases had multiple components with distinct interfaces, and we present an approach for assigning cross-links to the interfaces. Employing the symmetry information for these cases further improved the performance of complex structure prediction.