A novel strategy for the identification of protein–DNA contacts by photocrosslinking and mass spectrometry

Photochemical crosslinking is a method for studying the molecular details of protein–nucleic acid interactions. In this study, we describe a novel strategy to localize crosslinked amino acid residues that combines laser-induced photocrosslinking, proteolytic digestion, Fe³⁺-IMAC (immobilized metal affinity chromatography) purification of peptide–oligodeoxynucleotide heteroconjugates and hydrolysis of oligodeoxynucleotides by hydrogen fluoride (HF), with efficient matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF-MS). The new method is illustrated by the identification of the DNA-binding site of the restriction endonuclease MboI. Photoactivatable 5-iododeoxyuridine was incorporated into a single site within the DNA recognition sequence (GATC) of MboI. Ultraviolet irradiation of the protein–DNA complex with a helium/cadmium laser at 325 nm resulted in 15% crosslinking yield. Proteolytic digestion with different proteases produced various peptide–oligodeoxynucleotide adducts that were purified together with free oligodeoxynucleotide by Fe³⁺-IMAC. A combination of MS analysis of the peptide–nucleosides obtained after hydrolysis by HF and their fragmentation by MS/MS revealed that Lys²⁰⁹ of MboI was crosslinked to the MboI recognition site at the position of the adenine, demonstrating that the region around Lys²⁰⁹ is involved in specific binding of MboI to its DNA substrate. This method is suitable for the fast identification of the site of contact between proteins and nucleic acids starting from picomole quantities of crosslinked complexes.     

Isotopically labeled crosslinking reagents: resolution of mass degeneracy in the identification of crosslinked peptides

Mass spectrometry in three dimensions (MS3D) is a newly developed method for the determination of protein structures involving intramolecular chemical crosslinking of proteins, proteolytic digestion of the resulting adducts, identification of crosslinks by mass spectrometry (MS), peak assignment using theoretical mass lists, and computational reduction of crosslinks to a structure by distance geometry methods. To facilitate the unambiguous identification of crosslinked peptides from proteolytic digestion mixtures of crosslinked proteins by MS, we introduced double 18O isotopic labels into the crosslinking reagent to provide the crosslinked peptides with a characteristic isotope pattern. The presence of doublets separated by 4 Da in the mass spectra of these materials allowed ready discrimination between crosslinked and modified peptides, and uncrosslinked peptides using automated intelligent data acquisition (IDA) of MS/MS data. This should allow ready automation of the method for application to whole expressible proteomes.     

Identification of base-specific contacts in protein-DNA complexes by photocrosslinking and mass spectrometry: a case study using the restriction endonuclease SsoII

Specific protein-nucleic acid interactions are of paramount importance for the propagation, maintenance and expression of genetic information. Restriction endonucleases serve as model systems to study the mechanisms of DNA recognition by proteins. SsoII is a Type II restriction endonuclease that recognizes the double stranded sequence downward arrow CCNGG and cleaves it in the presence of Mg(2+)-ions, as indicated. SsoII shows sequence similarity over a stretch of approximately 70 amino acid residues with several other restriction endonucleases that recognize a similar sequence as SsoII (Cfr10I, EcoRII, NgoMIV, PspGI). In NgoMIV this stretch is involved in DNA recognition and cleavage, as shown by the crystal structure analysis of an enzyme-product complex. To find out whether the presumptive DNA recognition region in SsoII is indeed in contact with DNA we have photocrosslinked SsoII with an oligodeoxyribonucleotide in which the first guanine of the recognition sequence was replaced by 5-iodouracil. Following digestion by trypsin, the peptide-oligodeoxyribonucleotide conjugate was purified by Fe(3+)-IMAC and then incubated with hydrogen fluoride, which hydrolyzes the oligodeoxyribonucleotide to yield the peptide-deoxyuridine conjugate. The site of photocrosslinking was identified by MALDI-TOF-MS and MALDI-TOF-MS/MS to be Trp189, adjacent to Arg188, which aligns with Arg194 in NgoMIV, involved in recognition of the second guanine in the NgoMIV recognition sequence G downward arrow CCGGC. This result confirms previously published conclusions drawn on the basis of a mutational analysis of SsoII. The methodology that was employed here can be used in principle to identify the DNA binding site of any protein.     

Characterization of the long-terminal repeat single-strand tail-binding site of Moloney-MuLV integrase by crosslinking

Processing of viral DNA by retroviral integrase leaves a dinucleotide single-strand overhang in the unprocessed strand. Previous studies have stressed the importance of the 5' single-stranded (ss) tail in the integration process. To characterize the ss-tail binding site on M-MuLV integrase, we carried out crosslinking studies utilizing a disintegration substrate that mimics the covalent intermediate formed during integration. This substrate carried reactive groups at the 5' ss tail. A bromoacetyl derivative with a side chain of 6 A was crosslinked to the mutant IN 106-404, which lacks the N-terminal domain, yielding a crosslinked complex of 50 kDa. Treatment of IN 106-404 with N-ethylmaleimide (NEM) prevented crosslinking, suggesting that Cys209 was involved in the reaction. The reactivity of Cys209 was confirmed by crosslinking of a more specific derivative carrying maleimide groups that spans 8A approximately. In contrast, WT IN was not reactive, suggesting that the N-terminal domain modifies the reactivity of the Cys209 or the positioning of the crosslinker side chain. A similar oligonucleotide-carrying iodouridine at the 5'ss tail reacted with both IN 106-404 and WT IN upon UV irradiation. This reaction was also prevented by NEM, suggesting that the ss-tail positions near a peptide region that includes Cys209.     

Identification of the oligonucleotide and oligopeptide involved in an RNA--protein crosslink induced by ultraviolet irradiation of Escherichia coli 30 S ribosomal subunits.

When 30 S ribosomal subunits are irradiated with ultraviolet light, we have found that an RNA-protein crosslinking reaction occurs whose primary target is protein S7. This paper describes the identification of the oligopeptide and oligonucleotide at the crosslinking point, by direct analysis (a) of the peptide remaining attached to an oligonucleotide (after total digestion of the RNA in the crosslinked complex with ribonucleases A and T₁, followed by digestion with trypsin), and (b) of the nucleotides remaining attached to the crosslinked protein (after digestion of the RNA in the complex with ribonuclease T₁ alone).The crosslinking site was found to lie within a single short peptide, Ser-Met-Ala-Leu-Arg (positions 113 to 117 in the S7 sequence), with methionine as the probable amino acid concerned. The principal RNA site was found to lie within an oligonucleotide three to six bases long, the underlined portion of the partially ordered sequence $\text{C-U-A-C-}\text{A-A-U-G.G.C}\text{-G}$ in section P of the 16 S RNA. The methodology involved has been designed with a view to being generally applicable in future RNA-protein crosslinking studies, where several proteins are simultaneously attached to the RNA.     

Structure at restriction endonuclease MboI cleavage sites protected by actinomycin D or distamycin A

Restriction endonuclease MboI cleavage of DNA was inhibited by actinomycin D and distamycin A. The two inhibitors protected different subsets of the 8 cleavage sites in polyoma DNA. The cleavage reactions were analyzed both in the presence of minimal inhibitory concentrations of the compounds and at higher concentrations, allowing cleavage at only 1 site/DNA molecule. The experiments showed that cleavage sites most efficiently protected by actinomycin D had putative inhibitor binding sites at a distance of 1-2 base pairs from the MboI recognition sequence. Distamycin A, in contrast, apparently has to bind immediately adjacent to the MboI recognition sequence to protect from cleavage.     

Purification of tomato sucrose synthase phosphorylated isoforms by Fe(III)-immobilized metal affinity chromatography

The major phosphorylation site of maize sucrose synthase (SuSy) is well conserved among plant species but absent in the deduced peptide sequence of the tomato SuSy cDNA (TOMSSF). In this study, we report the in vitro phosphorylation of 25-day-old tomato fruits SuSy on seryl residue(s) by an endogenous Ca2+-dependent protein kinase activity. Two distinct 32P-labeled peptides detected in the tryptic peptide map of in vitro 32P-radiolabeled tomato fruit SuSy were purified. Amino acid sequencing and phosphoamino acid analysis of the major 32P-labeled peptide revealed the presence of a SuSy isozyme in young tomato fruit having the N-terminus phosphorylation site present in other plant species. By using Fe(III)-immobilized metal affinity chromatography [Fe(III)-IMAC] as a final purification step of tomato fruit SuSy, two 32P-labeled tomato SuSy isoforms were separated from a nonradiolabeled SuSy fraction by using a pH gradient. The major 32P-SuSy isoform was phosphorylated exclusively at the seryl residue related to the phosphorylation site of maize SuSy. The multiphosphorylated state of the second radiolabeled SuSy fraction was indicated by a higher retention during Fe(III)-IMAC and by tryptic peptide mapping analysis. Kinetic analyses of SuSy isoforms purified by Fe(III)-IMAC have revealed that phosphorylation of the major phosphorylation site of tomato fruit SuSy was not sufficient by itself to modulate tomato SuSy activity, whereas the affinity for UDP increased about threefold for the multiphosphorylated SuSy isoform.     

Identification of a site of psoralen crosslinking in E. coli 16S ribosomal RNA

We have developed a 2-dimensional gel method for identification of RNA sequences crosslinked by the intercalative drug 4'-hydroxymethyl-4,5',8-trimethylpsoralen (HMT). This method is being used to localize such sites in E. coli ribosomal RNA. We report here the identification of a site for HMT crosslinking within positions 434 and 497 of 16S rRNA. We suggest a likely site for HMT intercalation, in which residues U548 and U473 become crosslinked via the drug.     

A method for the isolation of covalent DNA-protein crosslinks suitable for proteomics analysis

The covalent crosslinking of protein to DNA is a form of DNA damage induced by a number of commonly encountered agents, including metals, aldehydes, and radiation as well as chemotherapeutic drugs. DNA-protein crosslinks (DPCs) are potentially bulky and helix distorting and have the potential to block the progression of translocating protein complexes. To fully understand the induction and repair of these lesions, it will be important to identify the crosslinked proteins involved. To take advantage of dramatic improvements in instrument sensitivity that have facilitated the identification of proteins by proteomic approaches, improved methods are required for isolation of DPCs. This article describes a novel method for the isolation of DPCs from mammalian cells that uses chaotropic agents to isolate genomic DNA and stringently remove noncrosslinked proteins followed by DNase I digestion to release covalently crosslinked proteins. This method generates high-quality protein samples in sufficient quantities for analysis by mass spectrometry. In addition, the article presents a modified form of this method that also makes use of chaotropic agents for promoting the adsorption of DNA (with crosslinked proteins) to silica fines, markedly reducing the DPC isolation time and cost. These approaches were applied to radiation- and camptothecin-induced DPCs.     

Thermal stability of hemoglobin crosslinked in the T-state by bis(3,5-dibromosalicyl) fumarate

Bis(3,5-dibromosalicyl) fumarate was used to crosslink oxyhemoglobin between Lys 82 beta 1 and Lys 82 beta 2 (Walder, J. A., et al. (1979) Biochemistry 18, 4265) and deoxyhemoglobin between Lys 99 alpha 1 and Lys 99 alpha 2 (Chatterjee R.Y., et al. (1986) J. Biol. Chem. 261, 9929). Thermal denaturations demonstrated that alpha crosslinked hemoglobin (alpha 99XLHb A) has the same stability as the beta crosslinked one (beta 82XLHb A). Both alpha and beta crosslinked methemoglobins have a denaturation temperature in 0.9 M guanidine of 57 degrees C compared to 41 degrees C of Hb A. The second product from the T-state crosslinking reaction was found to be crosslinked between the beta chains by chain separation and amino acid analysis. The possible positions for this crosslink are limited to the bisphosphoglycerate binding site in the three-dimensional structure. Its stability is comparable to that of the alpha 99XLHb A or beta 82XLHb A. These modified hemoglobins are potential blood substitutes.     

Crosslinking of cytochrome c and cytochrome b5 with a water-soluble carbodiimide. Reaction conditions, product analysis and critique of the technique

A water soluble carbodiimide, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC), has been used to crosslink horse heart cytochrome c and trypsin-solubilized bovine liver microsomal cytochrome b5. The reaction was conducted under a variety of solution conditions, and the products were purified by a combination of gel filtration and ion-exchange chromatography. Under all conditions of pH, ionic strength, EDC/protein ratio and reaction time that were studied, multiple 1:1 crosslinked complexes were observed with no evidence of a single, dominant species. Acetate, which is often used as a quencher of such reactions, was found to increase the complexity of the reaction products, presumably through EDC-promoted coupling to cytochrome c. Hydroxylamine treatment of the crosslinked complexes, a procedure frequently used to reverse EDC modification of tyrosyl residues, did not reduce the number of crosslinked components observed. The cytochrome b5 heme group was readily extracted from each of the 1:1 crosslinked complexes by standard techniques, so the crosslinking of heme propionate 7 with Lys79 of cytochrome c that might have been anticipated on the basis of molecular graphics modeling [Salemme, F.R. (1976) J. Mol. Biol. 102, 563-568] was not evident from this analysis. Analysis of HPLC tryptic peptide maps produced from crosslinked complexes revealed reduced specificity of trypsin in hydrolysis of EDC-crosslinked protein-protein complexes and unsatisfactory resolution of crosslinked or branched peptides. Nevertheless, it was possible to demonstrate that residues 52-72 of cytochrome b5, a region predicted to be critical to interaction with cytochrome b5 [Salemme, F.R. (1976) J. Mol. Biol. 102, 563-568] was absent from all peptide maps of 1:1 cytochrome c.cytochrome b5 complexes. Based on these results and a review of the literature involving EDC crosslinking of electron transfer proteins, we conclude that the techniques available for specific protein hydrolysis and separation of crosslinked peptides are not adequate to permit routine unambiguous identification of crosslinking sites in carbodiimide-crosslinked complexes.     

Identifying an interaction site between MutH and the C-terminal domain of MutL by crosslinking, affinity purification, chemical coding and mass spectrometry

To investigate protein–protein interaction sites in the DNA mismatch repair system we developed a crosslinking/mass spectrometry technique employing a commercially available trifunctional crosslinker with a thiol-specific methanethiosulfonate group, a photoactivatable benzophenone moiety and a biotin affinity tag. The XACM approach combines photocrosslinking (X), in-solution digestion of the crosslinked mixtures, affinity purification via the biotin handle (A), chemical coding of the crosslinked products (C) followed by MALDI-TOF mass spectrometry (M). We illustrate the feasibility of the method using a single-cysteine variant of the homodimeric DNA mismatch repair protein MutL. Moreover, we successfully applied this method to identify the photocrosslink formed between the single-cysteine MutH variant A223C, labeled with the trifunctional crosslinker in the C-terminal helix and its activator protein MutL. The identified crosslinked MutL-peptide maps to a conserved surface patch of the MutL C-terminal dimerization domain. These observations are substantiated by additional mutational and chemical crosslinking studies. Our results shed light on the potential structures of the MutL holoenzyme and the MutH–MutL–DNA complex.     

An enhanced protein crosslink identification strategy using CID-cleavable chemical crosslinkers and LC/MS(n) analysis

We describe a novel two-step LC/MS(n) strategy to effectively and confidently identify numerous crosslinked peptides from complex mixtures. This method incorporates the use of our gas-phase cleavable crosslinking reagent, disuccinimidyl-succinamyl-aspartyl-proline (SuDP), and a new data-processing algorithm CXLinkS (Cleavable Crosslink Selection), which enables unequivocal crosslink peptide selection and identification on the basis of mass measurement accuracy, high resolving power, and the unique fragmentation pattern of each crosslinked peptide. We demonstrate our approach with well-characterized monomeric and multimeric protein systems with and without database searching restrictions where inter-peptide crosslink identification is increased 8-fold over our previously published data-dependent LC/MS3 method and discuss its applicability to other CID-cleavable crosslinkers and more complex protein systems.     

Photocrosslinking locates a binding site for the large subunit of human replication protein A to the damaged strand of cisplatin-modified DNA.

The repair proteins XPA, XPC and replication protein A (RPA) have been implicated in the primary recognition of damaged DNA sites during nucleotide excision repair. Detailed structural information on the binding of these proteins to DNA lesions is however lacking. We have studied the binding of human RPA (hRPA) and hRPA-XPA-complexes to model oligonucleo-tides containing a single 1, 3-d(GTG)-cisplatin-modification by photocrosslinking and electrophoretic mobility shift experiments. The 70 kDa subunit of hRPA can be crosslinked with high efficiency to cisplatin-modified DNA probes carrying 5-iodo-2"-deoxyuridin (5-IdU) as crosslinking chromophore. High efficiency crosslinking is dependent on the presence of the DNA lesion and occurs preferentially at its 5"-side. Examination of the crosslinking efficiency in dependence on the position of the 5-IdU chromophore indicates a specific positioning of hRPA with respect to the platination site. When hRPA and XPA are both present mainly hRPA is crosslinked to the DNA. Our mobility shift experiments directly show the formation of a stable ternary complex of hRPA, XPA and the damaged DNA. The affinity of the XPA-hRPA complex to the damaged DNA is increased by more than one order of magnitude as compared to hRPA alone.     

Thermal stability of DNA with interstrand crosslinks

Although many anticancer drugs exert their biological activity by forming DNA interstrand crosslinks (ICLs), the thermodynamics of biologically relevant long crosslinked DNAs has not been intensively studied in contrast to short duplexes. Here, we carry out computer modeling of the shift of melting temperature of long DNAs caused by ICLs taking into account crosslinking effect in itself and concomitant local alterations in the free energy (δG) of the helix-coil transition at sites of ICLs. Depending on δG, DNA interstrand crosslinks at per nucleotide concentration r = 0.05 can change the melting temperature by value from -17 to +47°C, and the influence weakly depends on DNA sequence and GC content. A change in melting temperature caused by introduction of interstrand crosslinking in modified DNA at sites of modifications also depends on δG and varies from 0 to +12°C. Comparison with experiment for the three platinum crosslinking compounds demonstrates utility of the theoretical method for understanding how crosslinking compounds can influence the melting behavior. On the basis of the method, interdependence of local distortions and crosslinking in itself was studied for thermal effect of ICLs. A method for evaluating the nature of the structural alteration that produces a change in thermal stability for short crosslinked DNA is also proposed. The methods can be used for comparative thermodynamic characterization of various DNA crosslinking agents.     

Cleavage map of BK virus DNA with restriction endonucleases MboI and HaeIII.

Specific cleavage of BK virus (MM) DNA with restriction endonuclease MboI gives rise to 10 fragments. Two techniques were used to determine the location of these fragments on the viral genome with respect to the three known sites for HindIII cleavage. In the first method, reciprocal digestion, individual MboI fragments were digested with HindIII and individual HindIII fragments were digested with MboI. In the second method, single-end 32P-labeled HindIII subfragments were partially digested with MboI, and then the sizes of the radioactive partial products were used to deduce the nearest neighboring fragment. Information from these two methods is more than adequate to map all the MboI enzyme sites. Cleavage of BK virus (MM) DNA with restriction enzyme HaeIII produces 21 fragments. With the aid of the same two methods, these fragments have also been ordered with respect to the known map locations of the HindIII and MboI sites.     

Cross-Reactions of Human Lupus Autoantibodies with 8-Methoxypsoralen Photomodified DNA Fragments

DNA fragments of around 200 base pair (average size) have been covalently crosslinked with 8-methoxypsoralen under 365 nm UV light. The photoadduct, induced antibodies in rabbits with a titer of > 1:12,800 by direct bindng ELISA. Binding data showed that the induced antibodies are conformation-specific recognizing restricted conformational change at site of crosslinking. Human autoantibodies against DNA, bound not only to native DNA but to the photomodified DNA fragment as well. In addition, binding patterns of SLE sera obtained from different patients were remarkably similar, indicating the recognition of altered conformation of the modified polymer by naturally occurring SLE anti-DNA autoantibodies.     

A two-tracked approach to analyze RNA-protein crosslinking sites in native,nonlabeled small nuclear ribonucleoprotein particles

Much attention is currently being devoted to questions of protein and RNA tertiary structures and to the quaternary arrangement of the individual macromolecules in ribonucleoprotein (RNP) particles. In this article we describe two complementary strategies that allow the identification of RNA-protein contact sites in assembled, nonlabeled RNP particles after UV crosslinking. The first combines immunoprecipitation of UV-irradiated RNP particles under mildly denaturing conditions followed by primer-extension analysis of the crosslinked (and thus coprecipitated) RNA. The second involves the purification of crosslinked peptide-oligonucleotide from RNP particles and the subsequent analysis of the crosslinked peptide and RNA by Edman degradation and matrix-assisted laser desorption/ionization (MALDI)-mass spectrometry (MS), respectively. Although the first approach provides a rapid method for the exact identification of RNA-protein contact sites in purified nonlabeled RNP particles, the latter adds valuable information about potential RNA binding domains within proteins and, thus, about the arrangement of these proteins within the quaternary structures of complex RNP assemblies. Recently, we applied both these strategies successfully to native purified spliceosomal RNP. These methods may be generally applicable to the analysis of RNP complexes, especially as they avoid labeling and reconstitution, both of which risk introducing artifacts.     

Enhanced phosphopeptide isolation by Fe(III)-IMAC using 1,1,1,3,3,3-hexafluoroisopropanol

IMAC can be used to selectively enrich phosphopeptides from complex peptide mixtures, but co-retention of acidic peptides together with the failure to retain some phosphopeptides restricts the general utility of the method. In this study Fe(III)-IMAC was qualitatively and quantitatively assessed using a panel of phosphopeptides, both synthetic and derived from proteolysis of known phosphoproteins, to identify the causes of success and failure in the application of this technique. Here we demonstrate that, as expected, peptides with a more acidic amino acid content are generally more efficiently purified and detected by MALDI-MS after Fe(III)-IMAC than those with a more basic content. Modulating the loading buffer used for Fe(III)-IMAC significantly affects phosphopeptide binding and suggests that conformational factors that lead to steric hindrance and reduced accessibility to the phosphate are important. The use of 1,1,1,3,3,3-hexafluoroisopropanol is shown here to significantly improve Fe(III)-IMAC enrichment and subsequent detection of phosphopeptides by MALDI-MS.