This Glen Report is Part One of 2; Part 2 will be released in a subsequent report.
In a recent Glen Report, we described the ultrafast photoreactive nucleosides CNVK and CNVD for DNA or RNA interstrand cross-linking. Alternatively, the photo-cross-linking reactions between modified nucleosides in DNA and RNA oligonucleotides with electron-rich amino-acid side-chains of proteins have proven broadly useful in elucidating the interaction of nucleic acids with specific binding proteins, and has seen an expanded interest in studying photo-cross-linking of nucleic acids to ribonucleoproteins. In an earlier Glen Research literature review of oligonucleotide crosslinking, the halogenated nucleosides received brief mention compared with other approaches; yet 5-halogenated pyrimidine nucleosides are especially useful in site-specific cross-linking of oligonucleotides to specific protein binding sites.
In this report we will describe improved methods for the photo-cross-linking of single- and double-stranded oligonucleotides containing 5-halogenated pyrimidines with corresponding oligo binding proteins. DNA and RNA oligonucleotides containing these modified nucleosides are readily synthesized via standard machine synthesis using phosphoramidite monomers (Figure 1). Through examples, we hope to improve recognition of the special utility of 5-bromo- and 5-Iodo-pyrimidine nucleosides in synthetic DNA and RNA oligonucleotides by providing background on a) the photolysis reaction mechanisms and their dependence on UV wavelength, b) the steric impact of 5-halogenated uridine containing oligos on binding affinity, and c) conditions used to obtain high efficiency photo-cross-linking to oligo binding protein. Hopefully data presented here can dispel a misplaced notion that low cross-linking yields are obtained when using these compounds, and that their highly specific photo-cross-linking reactions will lead to additional research utilization.
|5-Br-dU-CE Phosphoramidite||5-I-dU-CE Phosphoramidite|
|5-I-dC-CE Phosphoramidite||5-I-U-CE Phosphoramidite|
Long-time readers may recall that 5-halogen substituted deoxyuridine (dU) has been useful in solving oligonucleotide protein complex structures using X-Ray crystallography. In addition, 5-Bromo-dU (as triphosphate) incorporated into nascent transcripts has found application in cell cycle investigations where antibodies to 5-Br-dU are used for immunostaining of nuclear structures. Similarly, synthetic oligonucleotides containing 5-Br-dU are also useful as in situ hybridization probes, providing a route towards highly specific immunostaining of mRNA.
The use of short wavelength UV254nm cross-linking (UVC) of native nucleic acids has utility in discovery research, but in some applications is limited by an associated propensity for non-specific covalent cross-links between double stranded nucleic acids and non-specific coupling to numerous nearby amino acid residues in nucleic acid-protein complexes. The concurrent use of sensitizers in photochemical cross-linking expanded the utility of photoreactions by providing higher cross-linking efficiency and has been shown to increase specificity by facilitating use of longer wavelength UV radiation (UVB), affording fewer non-specific nucleic acid interstrand cross-links and strand breaks. For example, in the case of psoralens as photosensitizers, interstrand photo-cross-linking and photo-reversal of crosslinks facilitate precise mapping of folded RNA substructures.
Early photo-cross-linking studies with Br-dU modified oligos used short wavelength UV254nm, and results were confounded due to photo-activated side reactions (see below). However, the Tad Koch lab at the University of Colorado, Boulder demonstrated that lower energy UV irradiation (UVB or UVA) could be used to minimize protein and nucleic acid side reactions and promote high-yield photo-cross-linking of halogen-nucleoside containing oligo with protein when complexed to the target protein. In model studies of the photo-cross-linking reactions, irradiation wavelength dependence and its relationship to the formation of different products were studied, and two major routes to these differing products were elucidated.
While UV254nm results in homolytic C-Br bond cleavage and generation of a highly reactive uridyl radical, the formation of highly specific cross-linking was proposed to occur using lower energy (308nm) excitation initially yielding a n,π* singlet state. Through intersystem crossing, the molecule converts to a lower energy triplet state that can oxidize a nearby electron-rich peptide moiety such that the resulting uridyl radical ion will couple to an adjacent electron-rich peptide functional group, substituting it for the bromine (Figure 2. Jablonski diagram for photoactivation of the 5-Bromo-Uracil chromophore. Reprinted from “Photochemical coupling of 5-bromouracil (BU) to a peptide linkage. A model for BU-DNA protein photocrosslinking,” T.M. Dietz, R.J. von Trebra, B.J. Swanson, and T.H. Koch, 1987, Journal of the American Chemical Society, 109 (6), 1793-1797. Copyright 1987 American Chemical Society.).
Much has been learned since that early work. In designing photo-cross-linking studies, one should note that the photocoupling reaction can be compromised by using excess laser power since this can cause secondary cross-link breakage as well as protein photodamage as was shown using phage R17 coat protein.8 In addition, intrastrand crosslinks can form between 5-Br-dU or 5-Br-dC and adjacent 5’- or 3’-dG and -dA residues.9 Yields of intrastrand crosslinks are much higher when the neighboring nucleoside is a dG versus when it is a dA, and Br-dC forms these cross-links much more efficiently than 5-Br-dU. Recently 300 nm UV (UVB) was used to study single strand breaks and intrastrand crosslinks in double stranded oligo containing 5-Br-dC or 5-Br-dU.10 These studies indicate that Br-dC is not the better choice when designing oligo to protein photo-cross-linking studies. In addition, minimization of undesirable strand cleavage and intrastrand cross-linking side reactions can be impacted by the position of the halogen containing nucleoside in the oligo sequence (e.g. adjacent to dG), and by the wavelength of UV used. Despite the potential limitations, high efficiency photo-cross-linking using these molecules continues to be a valuable technique.
Today, we increasingly understand that folded RNA structures in concert with linear sequence in RNA-protein complexes are involved in the control of gene expression. As a result, researchers continue to seek better understanding of the molecular details of these regulatory ribonucleoprotein complexes, including the specific location of the interacting oligo and protein partners in the complex.11 In this report we highlight a few discovery applications that include DNA- and RNA-protein cross-linking.
Studying NF BA1, a nuclear factor that binds to the promoter region of human apoB gene, Kardassis prepared a synthetic DNA oligonucleotide modified with five 5-Br-dU residues replacing T and corresponding to the coding sequence (-80 to -63) of the apoB gene (Figure 3. apoB Promoter binding motif oligo, modified with 5-Br-dU). The 5-Br-dU oligo was radiolabeled and annealed to its complementary oligo forming the double stranded sequence that binds NF BA112 protein. The double stranded DNA was incubated with purified NF BA1 protein alone or in the presence of single stranded oligo sequences either fully homologous, with mismatches, or fully non-homologous to the coding sequence.
The specificity of the photo-cross-linking reaction was demonstrated by PAGE analysis (Figure 4. Cross-linking of ApoB Gene to NF BA1). The presence of photo-cross-linked 67 Kd radiolabeled protein adduct was abolished by including in the photo-cross-linking reaction either BA1 or CIII-B oligonucleotides that bind NF BA1 (see corresponding gel lanes). Photo-cross-linking was highly sequence specific, as shown by including BM2 or CIII-C oligos in the cross-linking reaction mixture. BM2 is a 3-base mismatched analog of BM1 oligo, while CIII-C is a non-homologous promoter sequence just downstream of the BA1 promoter sequence. As the gel lanes demonstrate, the presence of either the partially complementary oligo BM2 or non-homologous CIII-C oligo do not interfere with cross-linking of brominated double stranded BA1 DNA oligo to NF-BA1 protein to form the 67Kd covalent adduct.
Similar utility has been shown for other 5-Br-dU modified DNA oligonucleotides where point contacts have been established by photo-cross-linking to associated proteins. 13,14
Similar photo-cross-linking of a 5-Br-dU containing duplex DNA to associated proteins included the mapping of point contacts at the oligo-protein interface. The E. coli lac Repressor protein, a 150-Kda tetramer of identical subunits having two operator binding sites and two inducer sites, has been extensively studied. Investigations in several labs revealed the double-stranded operator sequence region through which lac Repressor protein recognizes its specific operator sequence and, when bound, RNA polymerase transcription of the lactose metabolic genes is inhibited.15 The inducer molecule allolactose binds to sites on lac Repressor which serve to induce transcription of the genes regulated by lac Repressor; in the study below the allolactose mimic isopropyl-β-D-thiogalactoside (IPTG) is used as inducer.
Thus, Wick and Matthews annealed two synthetic complementary 40-mer oligos to create the double stranded DNA operator sequence that binds to the lac Repressor protein (Figure 5). To identify points of interaction of the operator sequence in its protein complex, a homologous series of synthetic double stranded bromodeoxyuridine-substituted Operator DNA mimics were prepared with 5-Br-dU replacing single thymine residues and their binding affinity and photo-cross-linking with lac Repressor protein was studied.
Thus, a series of double stranded operator DNA mimics were prepared, nine in which each singly 5-Br-dU substituted top strand oligo was annealed to its unmodified complementary bottom strand oligo, and similarly another nine in which each singly 5-Br-dU substituted bottom strand oligo was annealed with its unmodified top strand. In addition to these singly 5-Br-dU substituted operator mimics, two heavily 5-Br-dU modified control operator oligos were prepared, one, Tper/B, has 9 central thymidine positions substituted with 5-Br-dU and an unmodified bottom strand (see Figure 5). The other, denoted T/Bper, contains a bottom strand with all T residues replaced with 5-Br-dU and an unmodified upper strand. To detect reactions using denaturing PAGE, oligos were end-labeled using 32P-ATP with T4 polynucleotide kinase and the reaction products were visualized by autoradiography.
In this early study a UV254nm source was used for cross-linking and so intrastrand bond scission is a significant side reaction. PAGE analysis of control experiments in the absence of lac repressor protein showed that 90 second UV irradiation of the highly 5-Br-dU substituted operators yields an array of shorter-mer fragments corresponding to cleavage of labeled strands at each substituted position. Thus, Tper/B yielded T strand fragments at -5, -3, +3, +4, +6, +14, +20, +21, +22 and Bper/T yielded strand +1, +2, +8, +13, +15, +16, +18, +19, +24.
The eighteen singly 5-Br-dU modified Operator duplexes were then used to study duplex binding affinity to lac Repressor. The Tx modified mimics (top strand modified) had dissociation constants, Kd, in the range 4.1 to 8.0 x 10-10 M and the nine Bx mimics Kd were in the range 4.0 to 8.3 x 10-10 M. Essentially in all singly modified analogs affinity was little changed from the completely unmodified operator sequence (Kd = 6.2 x 10-10 M). These results confirm that substituting bromine for a methyl group on the pyrimidine ring has little steric influence on binding affinity, in accord with expectation, since the van der Waals radius of bromine (1.95 Å) is quite similar to a methyl group (1.99 Å).
Impact of Transcription activator isopropyl-β-D-thiogalactoside (IPTG): Each of these 18 5-Br-dU operator mimics was photo-cross-linked in the presence and absence of IPTG; as a transcription activator it was thought that IPTG binding to the tetrameric protein may affect concurrent binding of the double stranded operator sequence and, as a result, influence the specificity of any cross-linking reactions. Indeed, it was observed that cross-linking specificity was profoundly impacted by the presence of IPTG. In its absence some amount of cross-linking was observed using all the duplexes with their varied 5-Br-dU substitution positions. However, with IPTG present, photo-cross-linking was much more specific, occurring essentially at predominantly 5 positions of substitution, revealing the presence of five specific points of close contact between the operator and lac Repressor (Figure 6).
Thus, singly 5-Br-dU containing DNA oligonucleotides confirmed that halogen substitution for a methyl group has a minimal impact on duplex affinity for the lac repressor binding site. In addition, these modified operators revealed the identity of the specific points of contact between the operator duplexes and the lac Repressor, and also showed the impact of transcription activator IPTG in focusing crosslinking to a small subset of sites within the binding pocket.
Photo-cross-linking of a short hairpin RNA fragment to bacteriophage R17 coat protein was studied and the efficiency and specificity benefits of using lower energy UV irradiation of the 5-I-U nucleoside were demonstrated. Bacteriophage R17 replicase gene translation is repressed when a small hairpin RNA binds to the bacteriophage R17 coat protein. Earlier workers used homologous synthetic RNA variants to establish the binding site size and critical RNA sequence parameters and found a 21-nucleotide fragment of R17 RNA binds to phage coat protein with similar affinity to the natural sequence.16 In photo-cross-linking studies in Tad Koch’s lab,17 Gott and coworkers prepared R17 RNA hairpins by in vitro transcription from synthetic DNA templates to contain 5-bromo-U substituted for various uridine residues.18 From studying photo-cross-linking of the varied 5-Br-U containing analogs to coat protein, it was clear that bromouridine in the hairpin loop was implicated, and while some of the other analogs were capable of cross-linking, these cross-linkings were strongly dependent on the position of 5-Br-U in the fragment.
Willis and co-workers19 synthesized by in vitro transcription (IVT) native and singly substituted 5-Br-U and 5-I-U RNA variants of the close-analog 19-nucleotide R17 RNA hairpin and studied their binding affinity and photo-cross-linking to purified coat protein, using a nitrocellulose filter method and alpha 32P-C labeled R17 RNA (Figure 7).
Interestingly, the dissociation constants, Kd, for the iodo- and bromo-modified RNAs bound to coat protein revealed 3- and 5-fold stronger binding affinity to the coat protein, respectively, than the native sequence. Thus, Willis et al. found 5-Br-U and 5-I-U labeled R17 RNA hairpins were well-tolerated within the binding site of R17 coat protein; this strong binding occurs despite that the van der Waals radii of bromine (1.95 Å) and iodine (2.15 Å) are considerably larger than the hydrogen at ring position 5 of uracil (1.2 Å).20
It was noted that 5-Br-U cross-linking yields were low, and the researchers observed that photoactivation using a low-pressure mercury lamp UV254nm versus monochromatic UV308nm from a xenon chloride excimer laser resulted in different cross-linking results; using longer wavelength UV higher photo-cross-linking yields were obtained, with less protein damage and less RNA strand scission. High energy photo-cross-linking appears to result in C-Br bond homolysis forming a highly reactive uridyl and Br radical pair, while the lower energy irradiation generates an alternative excited state6.
Thus, using the 308 nm XeCl excimer laser, the 5-Br-U containing R17 RNA hairpin 1 (5-Br-U-R17 RNA1) achieved a maximal 40% coupling yield with R17 coat protein, while the 5-I-U-R17 RNA2 achieved 80% coupling yield in 5 minutes of irradiation.
The researcher’s interest in further reducing photo-generated side products prompted the use of longer wavelength light to activate and cross-link the 5-I-U nucleoside modified R17 hairpin analog. The time course of photo-cross-linking using 325 nm helium cadmium (HeCd) laser irradiation of 5-I-U-RNA 2 to the R17 coat protein is shown in Figure 8. SDS PAGE analysis revealed up to 94% coupling yields with negligible amounts of side products even upon extended irradiation.
To establish the identity of the cross-linked amino acid in the RNA-coat protein adduct, a 308 nm XeCl laser irradiated cross-linking reaction was carried out using the radiolabeled 5-Br-U RNA and coat protein. The mixture was ethanol precipitated and the redissolved pellet was trypsin digested. The mixture was purified by DEAE adsorption to remove free RNA and was subjected to a salt step gradient to remove peptides. Then RNA and tryptic adducts were eluted in 0.6 M NaCl, ethanol precipitated and purified by denaturing 20% PAGE followed by electroblotting onto a PVDF membrane. The RNA-peptide adduct was then sequenced via automated Edman degradation directly off the membrane, revealing that a single tyrosine residue, Tyr85, had been covalently cross-linked to the uridine residue.
In summary, this work showed that even the replacement of uridine by the larger 5-Br-U and 5-I-U in R17 RNA did not significantly perturb binding affinity for phage coat protein; surprisingly the already strong RNA-R17 coat protein binding was enhanced in the halogen modified R17 RNA hairpins. In addition, both the bromine and iodine singly halogen substituted RNAs were rapidly photo-cross-linked to coat protein with high efficiency using monochromatic laser irradiation at 308 nm or 325 nm, respectively; a single tyrosine residue in R17 coat protein was found to be cross-linked. Photo-cross-linking studies using longer wavelength UV irradiation resulted in reduced side-reactions and are more specific for 5-halogen-uridine containing RNA and DNA cross-links.
Low energy UV photo-cross-linking can be achieved using quite simple equipment. Dietz et al.21. demonstrated similar specific cross-linking results to those achieved with the laser by excitation instead using the 313 nm emission from a high-pressure mercury lamp selected with a monochromator at 310 nm with a band pass of 20 nm. Gott and coworkers used a medium wavelength transilluminator for some experiments. Even simpler methods to obtain useful photo-cross-linking conditions are available, as described by Xue and Nicholson.22
Several labs have recognized the utility of photochemical cross-linking methods to better understand aptamer interaction with targets. An early application of photo-cross-linking to aptamer development by Jensen and coworkers used in vitro selection to direct the covalent attachment of high-affinity RNA ligands, aptamers containing 5-I-U, to human immunodeficiency virus type 1 Rev protein.23
A feature of the Jensen lab study was the application of SELEX methodology to include 5-iodouridine triphosphate (5-I-UTP) instead of UTP in the phage T7 RNA polymerase transcription reaction mixtures during evolution of the aptamers. The resulting high affinity RNA aptamers were used for efficient photo-cross-linking of the aptamer to HIV-1 Rev protein and, although this work used transcription of 5-iodouridine triphosphates (as opposed to synthetic oligos from phosphoramidite synthesis), the value of photo-cross-linking methodology in elucidating aptamer interaction with target protein was clearly shown.
The utility of directly using 5-Br-dU or 5-I-dU as a randomization monomer amidite in SELEX is complicated by the base pairing ambiguity of these two monomers. Still, there is another use for these nucleoside phosphoramidites later in aptamer studies by substituting for T residues in an already optimized aptamer. For example, Mallikaratchy et al. used SELEX to evolve aptamers from live cells.24 Using Ramos cells, a Burkitt’s lymphoma cell line, an aptamer, TD05, was discovered that recognizes with high affinity (Kd 0.75 nM) a membrane bound heavy mu chain of IgM5 (IGHM), this latter a protein component of a B-cell receptor complex in these cells.25 Then, using the fluorescent FITC-labeled TD05 aptamer, a cell binding assay was established by fluorescence activated cell sorting (FACS). Using this assay, specific binding of the aptamer to a yet unidentified target within cells could be demonstrated by competition with unlabeled TD05.
The researchers then prepared a series of 5-I-dU modified TD05 aptamers (5-iodo-2’-deoxyUridine is called 5dUI in the paper) using machine DNA synthesis where 5-I-dU was substituted for thymine, including fully substituted (all T replaced) and various partially substituted TD05 analogs. Fully substituted 5-I-dU TD05 did not bind to cells, however less-substituted variants did bind strongly; an optimally labeled 5-I-dU TD05 aptamer analog was identified that contained four 5-I-dU residues (Figure 9). An analogous variant was then synthesized by solid phase synthesis that also incorporated a 3’-cleavable disulfide linked biotin tail to capture and identify the protein target.
To identify the aptamer target, the 5’-32P radiolabeled modified aptamer was photo-cross-linked to its target within cells using nanopulsed XeCl excimer laser (308 nm) irradiation. The resulting cells were lysed/homogenized, and solids were pelleted to yield crude membrane protein extract; this was detergent solubilized and separated from insoluble cell debris. The soluble fraction was incubated with streptavidin magnetic beads, washed, and the cross-linked protein mixture was released from beads by disulfide bond cleavage; mass spectrometry was used to identify four candidate protein targets among what remained a complex mixture also containing nuclear proteins. Further experiments demonstrated IGHM was a TD05 target. This effort led to subsequent studies in which the TD05 aptamer was prepared covalently coupled to light‐activated photosensitizer molecule chlorin e6 (Figure 10) and, using the conjugate, these workers demonstrated specific killing of cancer model Ramos cells.26
In summary, SELEX optimized DNA and RNA aptamers have been modified with 5-iodouracil nucleoside residues substituted at one or more T or U residues, respectively, resulting in 5-iodouracil-modified aptamers with retention of strong target binding affinity. The specific, efficient photo-cross-linking within a complex milieu of live cells has facilitated identification of a membrane bound receptor specific to these Burkitt’s lymphoma cells and led to development of a cancer cell-specific aptamer-drug conjugate.
The photo-cross-linking regiospecificity of these methods as applied to many DNA and RNA-protein complexes has been used to reveal molecular details of complex formation. Today, an arsenal of halogen modified deoxyribonucleosides and ribonucleosides are activated by long-wavelength UV to yield photoadducts, as shown in Figure 11 below, written generally to include deoxyribo- and ribonucleotides. Glen Research carries many of the halogen containing phosphoramidites needed for chemical synthesis of photo-cross-linkable oligonucleotides. Because these nucleosides interact mainly with the electron rich amino acids phenylalanine, tyrosine, tryptophan and histidine, researchers may encounter nucleic acid-protein complexes in which a photoactivatable probe’s protein binding site may not have a suitably located nearby amino acid side chain with which to efficiently photo-cross-link. Even so, taking advantage of oligo design flexibility a great many examples have proven successful and instructive.
Although not the main focus of this report, it should be mentioned that thio-analogs of nucleosides have also been used as photo-cross-linking probes of nucleic acid interaction with proteins.27 This includes the DNA analogs 4-thio-dT, 4-thio-dU, 6-thio-dG and the RNA analog 6-thio-G. Possible advantages of using the thio derivatives for photo-cross-linking is their similarity to the natural structures and the availability of additional protein interfaces for cross-linking due to the additional photoactive analogs thioG and thiodG. Typically, thio-analogs are photoactivated at 340 nm wavelength and so, similar to irradiation of 5-Br-uracil and 5-I-uracil containing nucleosides, will result in low non-specific side reactions.
A long history of photo-cross-linking using 5-bromo and 5-iodopyrimidine nucleoside analogs in synthetic DNA and RNA continues aiding researchers to understand the intimate interactions of nucleic acids with partner proteins involved in cellular control and signaling pathways. Researchers have found new applications of these molecules in aptamer design and characterization, including development of drug candidates through cell-based SELEX methods in the discovery of membrane bound receptors for potential use in targeting cancers. The use of these efficient photo-cross-linkers seems well positioned to lead to other research insights among the growing network of ribonucleoproteins and RNP complexes.
The size change resulting from bromine and iodine substitution on position 5 of the uracil ring has a minimal impact on DNA oligo binding to its receptor protein binding site. Surprisingly, halogen substitution in RNA, despite the increased size of bromine and iodine over hydrogen, did not reduce binding affinity in the examples shown. Rapid and highly specific photo-cross-linking of 5-Br-dU/5-Br-U and 5-I-dU/5-I-U residues in single and double stranded DNA or RNA oligonucleotides in association with cognate binding proteins can be achieved using longer wavelength UV irradiation (laser or bandpass filtered) centered at 308 nm and 325 nm, respectively.
Glen Research’s line of molecules includes the halogenated-deoxycytidine and -deoxyuridine phosphoramidites and bromodeoxyuridine CPG support. For RNA constructs, the bromo- and iodo-uridines are available as TBDMS protected phosphoramidites and also as the bromo-uridine, 2’-O-methyl phosphoramidite. In general, mild room temperature deprotection is required. See use instructions for dissolution, coupling and deprotection conditions for each monomer.
In Part 2 of this series, we will describe elegant studies of the RNA interference pathway demonstrating how photo-cross-linking of 5-halogen-Uracil containing oligos have been used to unravel the mechanistic details of the multiple step, multiple component ribonucleoprotein RISC complex.