- Email: [email protected]
Directed Cleavage of RNA with Protein-Tethered EDTA–Fe
METHODS: A Companion to Methods in Enzymology 18, 78 – 84 (1999) Article ID meth.1999.0759, available online at http://www.idealibrary.com on
Directed Cleavage of RNA with Protein-Tethered EDTA–Fe Kathleen B. Hall* and Robert O. Fox† *Department of Biochemistry and Molecular Biophysics, Washington University School of Medicine, St. Louis, Missouri 63110; and †Department of Human Biological Chemistry and Genetics and The Sealy Center for Structural Biology, University of Texas Medical Branch, Galveston, Texas 77555
There are several methods for locating the RNA site where a protein binds. One of the less common methods is directed cleavage of the RNA by an EDTA–Fe reagent tethered to the protein. The reaction of the EDTA–Fe(III) with ascorbate or hydrogen peroxide produces reactive oxygen species, such as ˚ radius of the iron hydroxyl radicals, localized within a 10-A center. The reactive oxygen species will attack the ribose or deoxyribose of nucleic acids as well as proximal polypeptide backbones. One EDTA–Fe reagent, (EDTA-2-aminoethyl)-2pyridyl disulfide complexed to iron (EPD–Fe), has been tethered to several proteins through a disulfide linkage to engineered cysteine thiols and used to cleave DNA, proteins, and RNA. A second tethered EDTA–Fe reagent, 1-(pbromoacetamidobenzyl)–EDTA–Fe, or BABE, has also been used to cleave RNA. Here we describe the issues involved in using these reagents with any RNA binding protein. © 1999 Academic Press
THE PRINCIPLE One of the first goals in the elucidation of a functional RNA:protein association is to identify the binding site(s) on the RNA. This binding site could be a single RNA hairpin specifically bound by a single protein (e.g., the human U1A protein bound to U1 stem loop II (1)), a linear array of RNA sequences that are bound with different relative affinities by several proteins (e.g., hnRNP proteins on mRNA (2)), an RNA sequence bound by one protein in a larger protein complex (e.g., the Sm proteins bound to U-snRNA (3)), a highly structured stable RNA that is bound at specific sites by specific proteins 78
(the ribosome), or an RNA that undergoes dynamic reorganization by many factors over its functional lifetime (pre-mRNA in the spliceosome). Using the example of the spliceosome, some protein:RNA associations exist only transiently, yet are required for activity, and it is critical to be able to detect formation of these associations. While the ribosome and the spliceosome are two prominent examples of assemblies of proteins on an active RNA, there is also a collection of polyadenylation proteins on the 39 UTR of mRNA that controls cleavage and addition of the poly(A) tail. Other RNA:protein complexes that serve housekeeping as well as regulatory functions include the arrangement of hnRNP proteins on nuclear pre-mRNA and mRNP proteins on cytoplasmic mRNA. The spatial and temporal organization of proteins on these RNAs is not well characterized, but is certainly critical for assembly and regulation. RNA:protein interactions constitute a major component of RNA function and require a variety of methods to probe the different types of complexes. There are several methods to physically map RNA:protein interactions, including photochemical crosslinking, footprinting, nuclease and chemical modification, modification interference, and, more recently, site-directed cleavage. Here, we will examine the use of EDTA–metal complexes incorporated at selected sites within a protein to generate directed cleavage of the bound RNA. Following is the protocol for attaching one such EDTA–iron reagent, (EDTA-2-aminoethyl)-2-pyridyl disulfide complexed with iron (EPD–Fe) (4, 5) to an RNA-binding protein. EDTA–metal complexes have been used to probe 1046-2023/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
DIRECTED CLEAVAGE OF RNA
the three-dimensional arrangements of proteins (5, 6) and protein–DNA associations (4, 7). Of particular interest to the RNA community, RNA:protein interactions on the 16S rRNA and within the 30S ribosomal subunit were mapped using the EDTA–Fe(III) complex [1-(p-bromoacetamidobenzyl)– EDTA–Fe, or BABE] tethered to ribosomal proteins S4 and S13 (8, 9); the position of EF-G on a posttranslation ribosome was mapped with BABE attached to EF-G at several sites, to look for cleavage in 16S and 23S rRNA (10); and the orientations of two human U1A RBD1 domains on its mRNA 39 UTR were mapped using EPD–Fe(III) tethered to RBD1 (11). Since for many applications of directed EDTA–Fe cleavage, it will not be known a priori how and where the RNA is bound on the protein, the placement of the probe on the body of the protein must be rationally selected. The mechanism of attachment of most EDTA–Fe reagents is through a cysteine sulfhydryl group; this necessitates first removing existing cysteines and then engineering a single cysteine on what is (sometimes) known to be the surface of the protein. Each mutant protein in the above examples was assayed for function, since it was likely that some mutations would affect both binding affinity and specificity, as well as protein folding. Functional mutants were then reacted with the EDTA–Fe complex, which is covalently attached to the cysteine through its reduced sulfhydryl group. It is expected that not all EDTA–Fe probes will show reactivity with the RNA, since the probe and RNA ˚ of the riboses must be within approximately 10 A Fe(II) reactive center and correctly oriented for cleavage to occur. The cleavage patterns allow construction of a map of the regions of RNA that are in proximity to specific regions of the protein; however,
79
these regions will not usually correspond to the RNA sequence that is specifically recognized by the protein, since cysteine substitutions within the proteinbinding site may not be functional for RNA binding. Cleavage sites typically are those regions of the RNA that are near the surface of the protein in close proximity to the EDTA–Fe probe. The tether length of EPD–Fe is approximately 14 ˚ (Fig. 1), and the tether for BABE is about 12 A ˚ A ˚ (12). Together with a 10 A radius from the reactive Fe(II) center, BABE–Fe and EPD–Fe have an effec˚ from the site of the cysteine sulfur tive reach of 22 A (9), although because EPD–Fe is more flexible and hydrophilic than BABE–Fe, they may not give identical cleavage patterns. The EDTA–Fe complexes are used to generate reactive oxygen species (hydroxyl radicals in the case of BABE–Fe) by the Fenton reaction (13); these species are responsible for cleavage. The site of cleavage in nucleic acids is in the sugar; in DNA, cleavage occurs most frequently at the C49 carbon (14), but in RNAs where the ribose is more accessible, the cleavage preference may not be so pronounced. EDTA–Fe will also cleave proteins (6, 12, 15) including its host molecule; this feature might be useful to map the relative orientations of proteins in RNA:protein complexes.
DESCRIPTION OF THE METHOD In this protocol, the assumption is made that the starting material is purified protein and that its sequence is known. It is further assumed that this is a mutated protein that now contains a single cysteine at one specific position in the structure, which will be used as the site of attachment of the EDTA–
FIG. 1. Reaction of EPD–Fe with a protein having an accessible cysteine residue yields a protein with EDTA–Fe covalently attached through the cysteine thiol.
80
HALL AND FOX
Fe. Since substitution of a cysteine could alter the stability or function of the protein, each mutant protein must first be successfully used in a functional assay. The third assumption is that the EDTA–Fe compound is in hand. The Fox group has synthesized the (EDTA-2-aminoethyl)-2-pyridyl disulfide– Fe31 (EPD–Fe) compound (5), and the Ebright group has synthesized the related iodoacetylphenylenediamine–EDTA (IPD) (4). Limited quantities of EPD–Fe are available from the Fox group. BABE was synthesized by the Meares group (12). At this time, no compound is available commercially. The activated protein:EDTA–iron derivative is used to cleave the proximal RNA sugar–phosphate backbone in the RNA:protein complexes. It will probably be necessary to move the cysteine: EDTA–Fe probe around the protein surface to juxtapose it to bound RNA; this will necessitate making several mutants, each of which needs to be checked for biological function. Cleavage of the RNA target can be observed directly if the RNA is either 59- or 39-end-labeled with 32 P. For other RNAs, the sites of cleavage can be mapped using primer-extension (8). Since this will be a function of the target RNA, no attempt is made here to describe these methods.
PROTEIN PREPARATION The cysteine mutant of the protein should be stored in DTT to keep its cysteine in reduced form. It needs to be purified away from DTT, in order to assay for S–H reactivity with Ellman’s reagent (5,59dithio-bis-(2-nitrobenzoic acid), DTNB). Depending on the properties of the protein, several methods are available for removing the DTT. Procedure Method 1. Pack a 1-5 mL spin-column with BioGel P2 (100–200 mesh, MWCO 1600; Bio-Rad), and extensively wash it with Chelex-treated buffer. In order to monitor the protein flowthrough and separation from DTT (MW 154), load a duplicate column with an equal volume of a solution of blue dextran (MW 8 3 106) and orange G (MW 452); this column can be conveniently used as a balance. The protein should come off in the void volume, as does the blue dextran, while the DTT will elute similar to the orange G.
A typical loading is 300/400 mL of concentrated protein. Spin at 200 rpm in a table-top centrifuge; collect the effluent. Calculate the protein concentration as accurately as possible (spectroscopically is most convenient) since the subsequent amount of reagent added will be calculated to be in excess. This method has the disadvantage that many proteins precipitate upon concentration on the spin column. Method 2. Dialyze the protein against buffer with no DTT. A microdialyzer is the most convenient, since the protein concentration should be kept high in a small volume. To ensure complete removal of the DTT in a timely fashion, keep the surface to volume ratio high in the dialyzer. Determine the protein concentration after dialysis. The pH should be kept below 7.0 and the buffers purged with argon to minimize air oxidation of the sulfhydryl group. Method 3. For gel-filtration chromatography, pour a long, thin column with a matrix appropriate to the protein molecular weight so that it comes off in the void volume while the DTT is retained on the column. Run the column in a buffer with a low pH (to keep the S–H groups reduced). The disadvantage of this method is that the protein will be diluted when it comes off the column, and it will be necessary to concentrate it for subsequent reactions. A Centricon or Microcon system (Amicon) is most convenient for this purpose. Comments 1. Buffers must be free of trace metal ions (and nucleases). A reliable method to remove both is to filter the buffers through a Nalgene disposable filter unit which has a 0.2-mm nitrocellulose membrane (to bind nucleases), upon which is layered a 5-mm bed of Chelex-100 (to remove trace metal ions). 2. Most free S–H groups will remain reduced for several weeks at 4°C at pH 5.3 in the absence of DTT, so the buffers used for storage after removal of DTT should have a lower pH than is optimal for the reaction conditions.
DTNB ASSAY Determine the reactivity of the free thiols in the protein using DTNB (Ellman’s reagent (16)). The
DIRECTED CLEAVAGE OF RNA
reactivity of EPD–Fe with free S–H groups is similar to that of DTNB, so this assay provides an estimate of the amount of free sulfhydryl in the protein and its accessibility to reagent. This is a spectrophotometric assay, based on the absorbance at 412 nm of the covalently bound colored chromophore formed upon reaction of DTNB with the sulfhydryl. Procedure The reagents used are DTNB (or Ellman’s reagent), 1 M Na2HPO4, pH 8.0, and buffer without DTT, which should be pH 4 –7.5, although the rate of reaction is reduced below pH 7. 1. Make up a DTNB solution to 4 mg/mL in 0.1 M Na2HPO4, pH 8.0. 2. A convenient method to measure the extent of reaction is to add 850 mL buffer to a 1-mL cuvette (1 cm pathlength), add 100 mL of DTNB solution, and add 50 mL protein stock. Record the absorbance at 412 nm after 15 min (or record continuously for 15 min, to obtain an estimate of the rate of the reaction). Calculate the concentration of reactive sulfhydryl groups using e412 5 1.36 3 104 M21 cm21, the extinction coefficient of the chromophore, in the expression (A412 5 e 3 c 3 l). [S-H] 5
A 412 (1 cm 3 1.36 3 104 M21 cm21 3
1000 mL (dilution factor) 50 mL
Calculate the ratio of reacted sulfhydryl groups to total molar S–H concentration (moles reacted)/ (moles total) to determine the percent reactive groups. 3. As a control, do the identical determination with 50 mL buffer in place of the protein solution. This gives a background absorbance of the buffer and also reveals residual DTT in the protein solution.
81
the concentration of protein in the cuvette should be 3.68 3 1025 M [A412 5 0.5 5 e 3 l 3 c 5 (1.36 3 104 M21 cm21)(1 cm)(3.68 3 1025 M)]. If a portion of the free thiols is oxidized or inaccessible to reagent and therefore does not react, then the absorbance after reaction will be less than 0.5. If there are other thiols from other sources (such as residual DTT), then the observed A412 will be greater than 0.5. 2. If there is little or no reactivity with the DTNB, then either the sulfhydryl group is not reduced (not free SH) or it is not accessible to the reagent. If it is oxidized, incubating the protein in 10 mM DTT for a day or two may reduce it; often DTT reduction is rapid at pH 7.5. If it is inaccessible, it is possible to denature the protein with urea and do the modification under denaturing conditions. The modification proceeds more slowly in urea, but will distinguish between inaccessible and oxidized sulfhydryls. Obviously, if the thiols are inaccessible as determined by this assay, then these are not appropriate sites for EPD–Fe attachment. 3. It may be necessary to experiment with the time and temperature of the DTNB reaction, to increase the reactivity of a free sulfhydryl that is conformationally occluded. The conditions to vary depend on the properties of the protein as, for example, its tolerance to temperature, pH, and salt. The DTNB addition reaction is pH dependent, with an optimal pH range of 6–8; lower pH reduces the reaction rate and changes e412. Two examples of reaction conditions are those for the human U1A RBD1, where the optimal reaction time for EPD addition was 1 h in 100 mM Pipes, pH 7, at 22°C, while for mutants of gd resolvase, the reaction conditions were 100 mM Tris, pH 7.3, 1 M NaCl, 20 h, 4°C, where the salt and temperature were dictated by the protein solubility. The conditions determined for optimal reaction with DTNB will be those used for reaction with the EDTA–Fe reagent, so investing the effort into defining these conditions is worthwhile. 4. The extent of reaction varies. For the U1A RBD1, between 50 and 60% of the single cysteine was consistently modified in the DTNB assay (11), while for other systems, the modification approaches 100%.
Comments 1. An absorbance after complete reaction of A412 5 0.5 is convenient, since this value provides a clear signal and does not exceed the linear response of most spectrophotometers. Assuming 1 mol of protein sulfhydryls that react to completion with the DTNB,
THE PROTEIN:EPD–Fe COMPLEX Based on the results of the DTNB assays, use the optimal protocol for addition of the EPD–Fe to the protein.
82
HALL AND FOX
Procedure 1. Prepare the protein as before to remove DTT, and determine its concentration. 2. Add a twofold molar excess of EPD–Fe. 3. React under the previously determined optimal conditions. As examples, for the U1A RBD1(Cys20) reaction, a twofold excess of [3 mM] EPD–Fe was added to [1.5 mM] protein in chelexed 50 mM Pipes, pH 7.0, 100 mM NaCl, and allowed to react for 1 h at 22°C (11). Ribosomal protein S4(Cys31) was incubated with 10 mM BABE–Fe in high-salt buffer at 37°C for 15 min (8). Again, these conditions are specific for each protein. 4. Separate the modified protein from free EPD–Fe and the reaction by-product thiopyridone. Methods include the Bio-Gel or similar spin column, separation in a Microcon3 vial (Amicon) with several buffer washes, reverse-phase HPLC, or a gelfiltration column. The final protein concentration cannot be accurately determined spectroscopically because EPD–Fe absorbs strongly at 290 nm, so assume that recovery of protein was complete. 5. To stabilize the protein, add an equal volume of 100% glycerol. Store at 220°C and use within 2 to 3 weeks. Slight protein degradation has been observed if modified proteins are stored for an extended period of time, probably due to cleavage by the attached EPD–Fe. Comments 1. Store the EPD–Fe(III) as a frozen dry powder. A 3 mM stock solution in distilled water will be stable at 220°C for at least 18 months. 2. To determine the extent of modification that has occurred in the protein, Noller’s group has used mass spectrometric analysis of the protein–BABE complex, as well as a competition assay to look for incorporation of iodo[14C]acetate (8). Amino acid analysis of oxidized and hydrolyzed gd resolvase cysteine mutants coupled to EPD–Fe showed between 40 and 90% derivatization, depending on the mutant (7). Other groups have assumed that the reactivity of DTNB is an adequate indicator of what is expected from the EDTA–Fe addition. 3. It is possible to leave the unreacted reagent together with the protein, but controls must be done to show that the excess free EPD–Fe does not cleave the protein. The integrity of the protein should be monitored by gel electrophoresis, using standard
SDS/PAGE gels. In general, protein cleavage yields are low, about 2–15%.
CLEAVAGE REACTIONS Cleavage reactions should use chelexed buffer at a pH from 4 to 7.5 that has been filtered though a nitrocellulose filter to remove nucleases. The buffer conditions are determined by the specific RNA:protein system. Procedure 1. Prepare the RNA:protein complex with the (EPD–Fe)protein. 2. Prepare 300 mM sodium ascorbate in chelexed buffer and adjust the pH. 3. Add sodium ascorbate to the protein:RNA solution to a final concentration of 30 mM. Reaction times and temperatures vary with the system and need to be experimentally determined. As an example, for RBD1(C20)–EPD–Fe, all cleavage reactions (final volume 10 ml) were carried out in chelexed 50 mM Pipes, pH 6.5, 100 mM NaCl, 30 mM sodium ascorbate for 50 min at 37°C (11). Lower temperatures did not show reaction, and shorter times at 37°C did not reproducibly produce strong cleavages. However, the [BABE–Fe–Cys31]S4 complex with 16S rRNA was incubated for 10 min at 4°C (8). The differences may reflect the proximity of the reagent to the RNA, as well as the flexibility of the tethered reagent in the complex. 4. For control reactions, incubate RNA (a) without protein in the presence of ascorbate, (b) with protein–(EPD–Fe) in the absence of ascorbate, and (c) with unreacted protein in the presence and absence of ascorbate. 5. Terminate the reactions by phenol extraction, freezing, addition of b-mercaptoethanol to 4% (v/v), or ethanol precipitation, as convenient. 6. Analyze the RNA as appropriate (either directly or by primer extension, followed by denaturing gel electrophoresis and autoradiography). Comments 1. The chemistry of EPD–Fe(III) appears to involve reduction of the coordinated Fe(III) (equiva-
DIRECTED CLEAVAGE OF RNA
lent to Fe31) by ascorbate, which leads to Fe(II) (Fe21); Fe(II) reacts with O2 or H2O2 to produce a reactive oxygen species which attacks the sugar of nucleic acids. The chemistry of BABE–Fe(III) is different, for it requires H2O2 and ascorbate to reduce the bound Fe(III) and produce hydroxyl radicals (OH z ). Addition of H2O2 to the EPD-Fe complex results in the production of different reactive species and it is not needed in EPD–Fe:RNA cleavage reactions. 2. The major product of the reaction of EDP–Fe is the oxidative release of the reagent from the protein. The cleavage of the RNA (and the protein) occurs with a low yield (5–15%), in part because the reactive species has a very short lifetime in aqueous solution. The dominant chemistry is unproductive reactions of the reagent. When the reagent falls off the protein, its concentration is too low to detectably cleave either protein or RNA. 3. Both double- and single-stranded RNAs are cleaved equally efficiently (17), since the target is the ribose. Regions of RNA in tight tertiary structures will be less accessible to EDTA–Fe cleavage in general (18) and are expected to be resistant to cleavage from a tethered reagent.
83
4. Since the tether of the reagent is flexible, the reactive iron will describe a volume that is likely to allow more than one cleavage event on the RNA. In the case of the end-labeled RNA bound to RBD1, cleavage was observed at four or five riboses on one side of the RNA duplex (Fig. 2), although notably only on one side of the duplex (11); one site is cleaved more efficiently than others. This pattern could reflect mobility or conformational heterogeneity of the loop in which the cysteine resided or multiple contact sites of the cysteine–EPD–Fe along the RNA backbone due to reagent flexibility. Materials Materials used included Chelex-100 (biotechnology grade, Bio-Rad) for removing metal ions from buffers; gel-filtration matrix, such as Bio-Gel P2 or Sephadex G-50 (depends on size of protein); spin column; Nalgene filter units (nitrocellulose membranes); Blue dextran (MW 8 3 106); Orange G (MW 452); DTNB (or Ellman’s reagent) (Pierce); 1 M Na2HPO4, pH 8.0; buffer (pH 6 – 8); and 300 mM sodium ascorbate (fresh).
ACKNOWLEDGMENTS We thank Dr. David Ledman and Dr. Mario Ermacora for helpful discussions. Development of the cleavage techniques was supported by the Howard Hughes Medical Institute (R.O.F.), the Welch Foundation (R.O.F.), and the NIH (GM51332 to R.O.F.; GM46318 to K.B.H.).
REFERENCES 1. Scherly, D., Boelens, W., van Venrooij, W. J., Dathan, N. A., Hamm, J., and Mattaj, I. W. (1989) EMBO J. 8, 4163– 4170.
FIG. 2. Cleavage sites of RBD1(Cys20 –-EPD–Fe) bound to its 39 UTR and U1 snRNA stem loop II. The RNA recognition site includes the loop-closing base pair and the 59 seven nucleotides on the internal loops and the hairpin loop. The EPD–Fe is located on an engineered cysteine in a flexible loop of the protein (11), which is not involved in any sequence-specific interactions with the RNA (19). This result gives the orientation of the protein on the RNA molecules, but note that it does not unambiguously define the binding site. Longer arrows indicate dominant cleavages.
2. Dreyfuss, G., Matunis, M. J., Pinol-Roma, S., and Burd, C. G. (1993) Annu. Rev. Biochem. 62, 289 –321. 3. Will, C. L., Kastner, B., and Luhrmann, R. (1995) in RNA Processing: A Practical Approach (Rickwood, D., and Hames, B. D., Eds), pp. 141–177, IRL Press, Oxford. 4. Ebright, Y. W., Chen, Y., Pendergast, P. S., and Ebright, R. H. (1992) Biochemistry 31, 10664 –10670. 5. Erma´cora, M. R., Delfino, J. M., Cuenoud, B., Schepartz, A., and Fox, R. O. (1992) Proc. Natl. Acad. Sci. USA 89, 6383– 6387.
84
HALL AND FOX
6. Erma´cora, M. R., Ledman, D. W., Hellinga, H. W., Hsu, G. W., and Fox, R. O. (1994) Biochemistry 33, 13625–13641. 7. Mazzarelli, J. M., Erma´cora, M. R., Fox, R. O., and Grindley, N. D. F. (1993) Biochemistry 32, 2979 –2986. 8. Heilek, G. M., Marusak, R., Meares, C. F., and Noller, H. F. (1995) Proc. Natl. Acad. Sci. USA 92, 1113–1116. 9. Heilek, G. M., and Noller, H. F. (1996) RNA 2, 597– 602. 10. Wilson, K. S., and Noller, H. F. (1998) Cell 92, 131–139. 11. Beck, D. L., Stump, W. T., and Hall, K. B. (1998) RNA 4, 331–339. 12. Rana, T. M., and Meares, C. F. (1991) Proc. Natl. Acad. Sci. USA 88, 10578 –10582.
13. Imlay, J. A., Chin, S. M., and Linn, S. (1988) Science 240, 640 – 642. 14. Tullius, T. D. (1989) Annu. Rev. Biophys. Biophys. Chem. 18, 213–237. 15. Erma´cora, M. R., Ledman, D. W., and Fox, R. O. (1996) Nature Struct. Biol. 3, 59 – 66. 16. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70 –77. 17. Celander, D. W., and Cech, T. R. (1990) Biochemistry 29, 1355–1361. 18. Latham, J. A., and Cech, T. R. (1989) Science 245, 276 –282. 19. Oubridge, C., Ito, H., Evans, P. R., Teo, C. H., and Nagai, K. (1994) Nature 372, 432– 438.
Directed Cleavage of RNA with Protein-Tethered EDTA–Fe Kathleen B. Hall* and Robert O. Fox† *Department of Biochemistry and Molecular Biophysics, Washington University School of Medicine, St. Louis, Missouri 63110; and †Department of Human Biological Chemistry and Genetics and The Sealy Center for Structural Biology, University of Texas Medical Branch, Galveston, Texas 77555
There are several methods for locating the RNA site where a protein binds. One of the less common methods is directed cleavage of the RNA by an EDTA–Fe reagent tethered to the protein. The reaction of the EDTA–Fe(III) with ascorbate or hydrogen peroxide produces reactive oxygen species, such as ˚ radius of the iron hydroxyl radicals, localized within a 10-A center. The reactive oxygen species will attack the ribose or deoxyribose of nucleic acids as well as proximal polypeptide backbones. One EDTA–Fe reagent, (EDTA-2-aminoethyl)-2pyridyl disulfide complexed to iron (EPD–Fe), has been tethered to several proteins through a disulfide linkage to engineered cysteine thiols and used to cleave DNA, proteins, and RNA. A second tethered EDTA–Fe reagent, 1-(pbromoacetamidobenzyl)–EDTA–Fe, or BABE, has also been used to cleave RNA. Here we describe the issues involved in using these reagents with any RNA binding protein. © 1999 Academic Press
THE PRINCIPLE One of the first goals in the elucidation of a functional RNA:protein association is to identify the binding site(s) on the RNA. This binding site could be a single RNA hairpin specifically bound by a single protein (e.g., the human U1A protein bound to U1 stem loop II (1)), a linear array of RNA sequences that are bound with different relative affinities by several proteins (e.g., hnRNP proteins on mRNA (2)), an RNA sequence bound by one protein in a larger protein complex (e.g., the Sm proteins bound to U-snRNA (3)), a highly structured stable RNA that is bound at specific sites by specific proteins 78
(the ribosome), or an RNA that undergoes dynamic reorganization by many factors over its functional lifetime (pre-mRNA in the spliceosome). Using the example of the spliceosome, some protein:RNA associations exist only transiently, yet are required for activity, and it is critical to be able to detect formation of these associations. While the ribosome and the spliceosome are two prominent examples of assemblies of proteins on an active RNA, there is also a collection of polyadenylation proteins on the 39 UTR of mRNA that controls cleavage and addition of the poly(A) tail. Other RNA:protein complexes that serve housekeeping as well as regulatory functions include the arrangement of hnRNP proteins on nuclear pre-mRNA and mRNP proteins on cytoplasmic mRNA. The spatial and temporal organization of proteins on these RNAs is not well characterized, but is certainly critical for assembly and regulation. RNA:protein interactions constitute a major component of RNA function and require a variety of methods to probe the different types of complexes. There are several methods to physically map RNA:protein interactions, including photochemical crosslinking, footprinting, nuclease and chemical modification, modification interference, and, more recently, site-directed cleavage. Here, we will examine the use of EDTA–metal complexes incorporated at selected sites within a protein to generate directed cleavage of the bound RNA. Following is the protocol for attaching one such EDTA–iron reagent, (EDTA-2-aminoethyl)-2-pyridyl disulfide complexed with iron (EPD–Fe) (4, 5) to an RNA-binding protein. EDTA–metal complexes have been used to probe 1046-2023/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.
DIRECTED CLEAVAGE OF RNA
the three-dimensional arrangements of proteins (5, 6) and protein–DNA associations (4, 7). Of particular interest to the RNA community, RNA:protein interactions on the 16S rRNA and within the 30S ribosomal subunit were mapped using the EDTA–Fe(III) complex [1-(p-bromoacetamidobenzyl)– EDTA–Fe, or BABE] tethered to ribosomal proteins S4 and S13 (8, 9); the position of EF-G on a posttranslation ribosome was mapped with BABE attached to EF-G at several sites, to look for cleavage in 16S and 23S rRNA (10); and the orientations of two human U1A RBD1 domains on its mRNA 39 UTR were mapped using EPD–Fe(III) tethered to RBD1 (11). Since for many applications of directed EDTA–Fe cleavage, it will not be known a priori how and where the RNA is bound on the protein, the placement of the probe on the body of the protein must be rationally selected. The mechanism of attachment of most EDTA–Fe reagents is through a cysteine sulfhydryl group; this necessitates first removing existing cysteines and then engineering a single cysteine on what is (sometimes) known to be the surface of the protein. Each mutant protein in the above examples was assayed for function, since it was likely that some mutations would affect both binding affinity and specificity, as well as protein folding. Functional mutants were then reacted with the EDTA–Fe complex, which is covalently attached to the cysteine through its reduced sulfhydryl group. It is expected that not all EDTA–Fe probes will show reactivity with the RNA, since the probe and RNA ˚ of the riboses must be within approximately 10 A Fe(II) reactive center and correctly oriented for cleavage to occur. The cleavage patterns allow construction of a map of the regions of RNA that are in proximity to specific regions of the protein; however,
79
these regions will not usually correspond to the RNA sequence that is specifically recognized by the protein, since cysteine substitutions within the proteinbinding site may not be functional for RNA binding. Cleavage sites typically are those regions of the RNA that are near the surface of the protein in close proximity to the EDTA–Fe probe. The tether length of EPD–Fe is approximately 14 ˚ (Fig. 1), and the tether for BABE is about 12 A ˚ A ˚ (12). Together with a 10 A radius from the reactive Fe(II) center, BABE–Fe and EPD–Fe have an effec˚ from the site of the cysteine sulfur tive reach of 22 A (9), although because EPD–Fe is more flexible and hydrophilic than BABE–Fe, they may not give identical cleavage patterns. The EDTA–Fe complexes are used to generate reactive oxygen species (hydroxyl radicals in the case of BABE–Fe) by the Fenton reaction (13); these species are responsible for cleavage. The site of cleavage in nucleic acids is in the sugar; in DNA, cleavage occurs most frequently at the C49 carbon (14), but in RNAs where the ribose is more accessible, the cleavage preference may not be so pronounced. EDTA–Fe will also cleave proteins (6, 12, 15) including its host molecule; this feature might be useful to map the relative orientations of proteins in RNA:protein complexes.
DESCRIPTION OF THE METHOD In this protocol, the assumption is made that the starting material is purified protein and that its sequence is known. It is further assumed that this is a mutated protein that now contains a single cysteine at one specific position in the structure, which will be used as the site of attachment of the EDTA–
FIG. 1. Reaction of EPD–Fe with a protein having an accessible cysteine residue yields a protein with EDTA–Fe covalently attached through the cysteine thiol.
80
HALL AND FOX
Fe. Since substitution of a cysteine could alter the stability or function of the protein, each mutant protein must first be successfully used in a functional assay. The third assumption is that the EDTA–Fe compound is in hand. The Fox group has synthesized the (EDTA-2-aminoethyl)-2-pyridyl disulfide– Fe31 (EPD–Fe) compound (5), and the Ebright group has synthesized the related iodoacetylphenylenediamine–EDTA (IPD) (4). Limited quantities of EPD–Fe are available from the Fox group. BABE was synthesized by the Meares group (12). At this time, no compound is available commercially. The activated protein:EDTA–iron derivative is used to cleave the proximal RNA sugar–phosphate backbone in the RNA:protein complexes. It will probably be necessary to move the cysteine: EDTA–Fe probe around the protein surface to juxtapose it to bound RNA; this will necessitate making several mutants, each of which needs to be checked for biological function. Cleavage of the RNA target can be observed directly if the RNA is either 59- or 39-end-labeled with 32 P. For other RNAs, the sites of cleavage can be mapped using primer-extension (8). Since this will be a function of the target RNA, no attempt is made here to describe these methods.
PROTEIN PREPARATION The cysteine mutant of the protein should be stored in DTT to keep its cysteine in reduced form. It needs to be purified away from DTT, in order to assay for S–H reactivity with Ellman’s reagent (5,59dithio-bis-(2-nitrobenzoic acid), DTNB). Depending on the properties of the protein, several methods are available for removing the DTT. Procedure Method 1. Pack a 1-5 mL spin-column with BioGel P2 (100–200 mesh, MWCO 1600; Bio-Rad), and extensively wash it with Chelex-treated buffer. In order to monitor the protein flowthrough and separation from DTT (MW 154), load a duplicate column with an equal volume of a solution of blue dextran (MW 8 3 106) and orange G (MW 452); this column can be conveniently used as a balance. The protein should come off in the void volume, as does the blue dextran, while the DTT will elute similar to the orange G.
A typical loading is 300/400 mL of concentrated protein. Spin at 200 rpm in a table-top centrifuge; collect the effluent. Calculate the protein concentration as accurately as possible (spectroscopically is most convenient) since the subsequent amount of reagent added will be calculated to be in excess. This method has the disadvantage that many proteins precipitate upon concentration on the spin column. Method 2. Dialyze the protein against buffer with no DTT. A microdialyzer is the most convenient, since the protein concentration should be kept high in a small volume. To ensure complete removal of the DTT in a timely fashion, keep the surface to volume ratio high in the dialyzer. Determine the protein concentration after dialysis. The pH should be kept below 7.0 and the buffers purged with argon to minimize air oxidation of the sulfhydryl group. Method 3. For gel-filtration chromatography, pour a long, thin column with a matrix appropriate to the protein molecular weight so that it comes off in the void volume while the DTT is retained on the column. Run the column in a buffer with a low pH (to keep the S–H groups reduced). The disadvantage of this method is that the protein will be diluted when it comes off the column, and it will be necessary to concentrate it for subsequent reactions. A Centricon or Microcon system (Amicon) is most convenient for this purpose. Comments 1. Buffers must be free of trace metal ions (and nucleases). A reliable method to remove both is to filter the buffers through a Nalgene disposable filter unit which has a 0.2-mm nitrocellulose membrane (to bind nucleases), upon which is layered a 5-mm bed of Chelex-100 (to remove trace metal ions). 2. Most free S–H groups will remain reduced for several weeks at 4°C at pH 5.3 in the absence of DTT, so the buffers used for storage after removal of DTT should have a lower pH than is optimal for the reaction conditions.
DTNB ASSAY Determine the reactivity of the free thiols in the protein using DTNB (Ellman’s reagent (16)). The
DIRECTED CLEAVAGE OF RNA
reactivity of EPD–Fe with free S–H groups is similar to that of DTNB, so this assay provides an estimate of the amount of free sulfhydryl in the protein and its accessibility to reagent. This is a spectrophotometric assay, based on the absorbance at 412 nm of the covalently bound colored chromophore formed upon reaction of DTNB with the sulfhydryl. Procedure The reagents used are DTNB (or Ellman’s reagent), 1 M Na2HPO4, pH 8.0, and buffer without DTT, which should be pH 4 –7.5, although the rate of reaction is reduced below pH 7. 1. Make up a DTNB solution to 4 mg/mL in 0.1 M Na2HPO4, pH 8.0. 2. A convenient method to measure the extent of reaction is to add 850 mL buffer to a 1-mL cuvette (1 cm pathlength), add 100 mL of DTNB solution, and add 50 mL protein stock. Record the absorbance at 412 nm after 15 min (or record continuously for 15 min, to obtain an estimate of the rate of the reaction). Calculate the concentration of reactive sulfhydryl groups using e412 5 1.36 3 104 M21 cm21, the extinction coefficient of the chromophore, in the expression (A412 5 e 3 c 3 l). [S-H] 5
A 412 (1 cm 3 1.36 3 104 M21 cm21 3
1000 mL (dilution factor) 50 mL
Calculate the ratio of reacted sulfhydryl groups to total molar S–H concentration (moles reacted)/ (moles total) to determine the percent reactive groups. 3. As a control, do the identical determination with 50 mL buffer in place of the protein solution. This gives a background absorbance of the buffer and also reveals residual DTT in the protein solution.
81
the concentration of protein in the cuvette should be 3.68 3 1025 M [A412 5 0.5 5 e 3 l 3 c 5 (1.36 3 104 M21 cm21)(1 cm)(3.68 3 1025 M)]. If a portion of the free thiols is oxidized or inaccessible to reagent and therefore does not react, then the absorbance after reaction will be less than 0.5. If there are other thiols from other sources (such as residual DTT), then the observed A412 will be greater than 0.5. 2. If there is little or no reactivity with the DTNB, then either the sulfhydryl group is not reduced (not free SH) or it is not accessible to the reagent. If it is oxidized, incubating the protein in 10 mM DTT for a day or two may reduce it; often DTT reduction is rapid at pH 7.5. If it is inaccessible, it is possible to denature the protein with urea and do the modification under denaturing conditions. The modification proceeds more slowly in urea, but will distinguish between inaccessible and oxidized sulfhydryls. Obviously, if the thiols are inaccessible as determined by this assay, then these are not appropriate sites for EPD–Fe attachment. 3. It may be necessary to experiment with the time and temperature of the DTNB reaction, to increase the reactivity of a free sulfhydryl that is conformationally occluded. The conditions to vary depend on the properties of the protein as, for example, its tolerance to temperature, pH, and salt. The DTNB addition reaction is pH dependent, with an optimal pH range of 6–8; lower pH reduces the reaction rate and changes e412. Two examples of reaction conditions are those for the human U1A RBD1, where the optimal reaction time for EPD addition was 1 h in 100 mM Pipes, pH 7, at 22°C, while for mutants of gd resolvase, the reaction conditions were 100 mM Tris, pH 7.3, 1 M NaCl, 20 h, 4°C, where the salt and temperature were dictated by the protein solubility. The conditions determined for optimal reaction with DTNB will be those used for reaction with the EDTA–Fe reagent, so investing the effort into defining these conditions is worthwhile. 4. The extent of reaction varies. For the U1A RBD1, between 50 and 60% of the single cysteine was consistently modified in the DTNB assay (11), while for other systems, the modification approaches 100%.
Comments 1. An absorbance after complete reaction of A412 5 0.5 is convenient, since this value provides a clear signal and does not exceed the linear response of most spectrophotometers. Assuming 1 mol of protein sulfhydryls that react to completion with the DTNB,
THE PROTEIN:EPD–Fe COMPLEX Based on the results of the DTNB assays, use the optimal protocol for addition of the EPD–Fe to the protein.
82
HALL AND FOX
Procedure 1. Prepare the protein as before to remove DTT, and determine its concentration. 2. Add a twofold molar excess of EPD–Fe. 3. React under the previously determined optimal conditions. As examples, for the U1A RBD1(Cys20) reaction, a twofold excess of [3 mM] EPD–Fe was added to [1.5 mM] protein in chelexed 50 mM Pipes, pH 7.0, 100 mM NaCl, and allowed to react for 1 h at 22°C (11). Ribosomal protein S4(Cys31) was incubated with 10 mM BABE–Fe in high-salt buffer at 37°C for 15 min (8). Again, these conditions are specific for each protein. 4. Separate the modified protein from free EPD–Fe and the reaction by-product thiopyridone. Methods include the Bio-Gel or similar spin column, separation in a Microcon3 vial (Amicon) with several buffer washes, reverse-phase HPLC, or a gelfiltration column. The final protein concentration cannot be accurately determined spectroscopically because EPD–Fe absorbs strongly at 290 nm, so assume that recovery of protein was complete. 5. To stabilize the protein, add an equal volume of 100% glycerol. Store at 220°C and use within 2 to 3 weeks. Slight protein degradation has been observed if modified proteins are stored for an extended period of time, probably due to cleavage by the attached EPD–Fe. Comments 1. Store the EPD–Fe(III) as a frozen dry powder. A 3 mM stock solution in distilled water will be stable at 220°C for at least 18 months. 2. To determine the extent of modification that has occurred in the protein, Noller’s group has used mass spectrometric analysis of the protein–BABE complex, as well as a competition assay to look for incorporation of iodo[14C]acetate (8). Amino acid analysis of oxidized and hydrolyzed gd resolvase cysteine mutants coupled to EPD–Fe showed between 40 and 90% derivatization, depending on the mutant (7). Other groups have assumed that the reactivity of DTNB is an adequate indicator of what is expected from the EDTA–Fe addition. 3. It is possible to leave the unreacted reagent together with the protein, but controls must be done to show that the excess free EPD–Fe does not cleave the protein. The integrity of the protein should be monitored by gel electrophoresis, using standard
SDS/PAGE gels. In general, protein cleavage yields are low, about 2–15%.
CLEAVAGE REACTIONS Cleavage reactions should use chelexed buffer at a pH from 4 to 7.5 that has been filtered though a nitrocellulose filter to remove nucleases. The buffer conditions are determined by the specific RNA:protein system. Procedure 1. Prepare the RNA:protein complex with the (EPD–Fe)protein. 2. Prepare 300 mM sodium ascorbate in chelexed buffer and adjust the pH. 3. Add sodium ascorbate to the protein:RNA solution to a final concentration of 30 mM. Reaction times and temperatures vary with the system and need to be experimentally determined. As an example, for RBD1(C20)–EPD–Fe, all cleavage reactions (final volume 10 ml) were carried out in chelexed 50 mM Pipes, pH 6.5, 100 mM NaCl, 30 mM sodium ascorbate for 50 min at 37°C (11). Lower temperatures did not show reaction, and shorter times at 37°C did not reproducibly produce strong cleavages. However, the [BABE–Fe–Cys31]S4 complex with 16S rRNA was incubated for 10 min at 4°C (8). The differences may reflect the proximity of the reagent to the RNA, as well as the flexibility of the tethered reagent in the complex. 4. For control reactions, incubate RNA (a) without protein in the presence of ascorbate, (b) with protein–(EPD–Fe) in the absence of ascorbate, and (c) with unreacted protein in the presence and absence of ascorbate. 5. Terminate the reactions by phenol extraction, freezing, addition of b-mercaptoethanol to 4% (v/v), or ethanol precipitation, as convenient. 6. Analyze the RNA as appropriate (either directly or by primer extension, followed by denaturing gel electrophoresis and autoradiography). Comments 1. The chemistry of EPD–Fe(III) appears to involve reduction of the coordinated Fe(III) (equiva-
DIRECTED CLEAVAGE OF RNA
lent to Fe31) by ascorbate, which leads to Fe(II) (Fe21); Fe(II) reacts with O2 or H2O2 to produce a reactive oxygen species which attacks the sugar of nucleic acids. The chemistry of BABE–Fe(III) is different, for it requires H2O2 and ascorbate to reduce the bound Fe(III) and produce hydroxyl radicals (OH z ). Addition of H2O2 to the EPD-Fe complex results in the production of different reactive species and it is not needed in EPD–Fe:RNA cleavage reactions. 2. The major product of the reaction of EDP–Fe is the oxidative release of the reagent from the protein. The cleavage of the RNA (and the protein) occurs with a low yield (5–15%), in part because the reactive species has a very short lifetime in aqueous solution. The dominant chemistry is unproductive reactions of the reagent. When the reagent falls off the protein, its concentration is too low to detectably cleave either protein or RNA. 3. Both double- and single-stranded RNAs are cleaved equally efficiently (17), since the target is the ribose. Regions of RNA in tight tertiary structures will be less accessible to EDTA–Fe cleavage in general (18) and are expected to be resistant to cleavage from a tethered reagent.
83
4. Since the tether of the reagent is flexible, the reactive iron will describe a volume that is likely to allow more than one cleavage event on the RNA. In the case of the end-labeled RNA bound to RBD1, cleavage was observed at four or five riboses on one side of the RNA duplex (Fig. 2), although notably only on one side of the duplex (11); one site is cleaved more efficiently than others. This pattern could reflect mobility or conformational heterogeneity of the loop in which the cysteine resided or multiple contact sites of the cysteine–EPD–Fe along the RNA backbone due to reagent flexibility. Materials Materials used included Chelex-100 (biotechnology grade, Bio-Rad) for removing metal ions from buffers; gel-filtration matrix, such as Bio-Gel P2 or Sephadex G-50 (depends on size of protein); spin column; Nalgene filter units (nitrocellulose membranes); Blue dextran (MW 8 3 106); Orange G (MW 452); DTNB (or Ellman’s reagent) (Pierce); 1 M Na2HPO4, pH 8.0; buffer (pH 6 – 8); and 300 mM sodium ascorbate (fresh).
ACKNOWLEDGMENTS We thank Dr. David Ledman and Dr. Mario Ermacora for helpful discussions. Development of the cleavage techniques was supported by the Howard Hughes Medical Institute (R.O.F.), the Welch Foundation (R.O.F.), and the NIH (GM51332 to R.O.F.; GM46318 to K.B.H.).
REFERENCES 1. Scherly, D., Boelens, W., van Venrooij, W. J., Dathan, N. A., Hamm, J., and Mattaj, I. W. (1989) EMBO J. 8, 4163– 4170.
FIG. 2. Cleavage sites of RBD1(Cys20 –-EPD–Fe) bound to its 39 UTR and U1 snRNA stem loop II. The RNA recognition site includes the loop-closing base pair and the 59 seven nucleotides on the internal loops and the hairpin loop. The EPD–Fe is located on an engineered cysteine in a flexible loop of the protein (11), which is not involved in any sequence-specific interactions with the RNA (19). This result gives the orientation of the protein on the RNA molecules, but note that it does not unambiguously define the binding site. Longer arrows indicate dominant cleavages.
2. Dreyfuss, G., Matunis, M. J., Pinol-Roma, S., and Burd, C. G. (1993) Annu. Rev. Biochem. 62, 289 –321. 3. Will, C. L., Kastner, B., and Luhrmann, R. (1995) in RNA Processing: A Practical Approach (Rickwood, D., and Hames, B. D., Eds), pp. 141–177, IRL Press, Oxford. 4. Ebright, Y. W., Chen, Y., Pendergast, P. S., and Ebright, R. H. (1992) Biochemistry 31, 10664 –10670. 5. Erma´cora, M. R., Delfino, J. M., Cuenoud, B., Schepartz, A., and Fox, R. O. (1992) Proc. Natl. Acad. Sci. USA 89, 6383– 6387.
84
HALL AND FOX
6. Erma´cora, M. R., Ledman, D. W., Hellinga, H. W., Hsu, G. W., and Fox, R. O. (1994) Biochemistry 33, 13625–13641. 7. Mazzarelli, J. M., Erma´cora, M. R., Fox, R. O., and Grindley, N. D. F. (1993) Biochemistry 32, 2979 –2986. 8. Heilek, G. M., Marusak, R., Meares, C. F., and Noller, H. F. (1995) Proc. Natl. Acad. Sci. USA 92, 1113–1116. 9. Heilek, G. M., and Noller, H. F. (1996) RNA 2, 597– 602. 10. Wilson, K. S., and Noller, H. F. (1998) Cell 92, 131–139. 11. Beck, D. L., Stump, W. T., and Hall, K. B. (1998) RNA 4, 331–339. 12. Rana, T. M., and Meares, C. F. (1991) Proc. Natl. Acad. Sci. USA 88, 10578 –10582.
13. Imlay, J. A., Chin, S. M., and Linn, S. (1988) Science 240, 640 – 642. 14. Tullius, T. D. (1989) Annu. Rev. Biophys. Biophys. Chem. 18, 213–237. 15. Erma´cora, M. R., Ledman, D. W., and Fox, R. O. (1996) Nature Struct. Biol. 3, 59 – 66. 16. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70 –77. 17. Celander, D. W., and Cech, T. R. (1990) Biochemistry 29, 1355–1361. 18. Latham, J. A., and Cech, T. R. (1989) Science 245, 276 –282. 19. Oubridge, C., Ito, H., Evans, P. R., Teo, C. H., and Nagai, K. (1994) Nature 372, 432– 438.