&IAPTER 1 DNase I Footprinting Ben& Lehlanc and Ibm Moss 1. Introduction DNase I footprinting was developed by Galas and Schmitz in 1978 as a method to study the sequence-specific binding of proteins to DNA (I). In this technique a suitable uniquely end-labeled DNA fragment is allowed to interact with a given DNA-binding protein and then the complex is partially digested with DNase 1. The bound protein pro- tects the region of the DNA with which it interacts from attack by the DNase. Subsequent molecular weight analysis of the degraded DNA by electrophoresis and autoradiography identifies the region of pro- tection as a gap in the otherwise continuous background of digestion products (for examples, see Fig. 1). The technique can be used to determine the site of interaction of most sequence-specific DNA- binding proteins but has been most extensively applied to the study of transcription factors. Since the DNase I molecule is relatively large compared to other footprinting agents (see Chapters 3 and 7 in this volume), its attack on the DNA is more readily prevented by steric hindrance. Thus DNase I footprinting is the most likely of all the footprinting techniques to detect a specific DNA-protein interaction. This is clearly demonstrated by our studies on the transcription factor xUBF (seeF ig. 1B). The xUBF interaction with thexenopus ribosomal DNA enhancer can be easily detected by DNase I footprinting but has not yet been detected by other footprinting techniques. DNase I footprinting can not only be used to study the DNA inter- actions of purified proteins but also as an assay to identify proteins of From- Methods m Molecular Biology, Vol. 30: DNA-Protein Interactions: Principles and Protocols Edlted by G. G Kneale CopyrIght 01994 Humana Press Inc , Totowa, NJ 1 2 Leblanc and Moss 60 60 81 60 81 Fig. 1. Examples of DNase I footprints. A. Footprint (open box) of a chicken erythrocyte DNA binding factor on the promoter of the H5 gene (2) (figure kindly donated by A. Ruiz-Carrillo). B. Interaction of the RNA polymerase I transcrip- tion factor xUBF with the tandemly repeated 60 and 81 bp. Xenopus ribosomal gene enhancers. Both A and B used 5’ end-labeled fragments. (-) and (+) refer to naked and complexed DNA fragments and (G + A) to the chemical sequence ladder. DNase I Footprinting 3 interest within a crude cellular or nuclear extract (2). Thus it can serve much the same function as a gel-shift analysis in following a specific DNA-binding activity through a series of purification steps. Since DNase I footprinting can often be used for proteins that do not “gel- shift,” it has more general applicability. However, because of the need for a protein excess and the visualization of the footprint by a partial DNA digestion ladder, the technique requires considerably more material than would a gel-shift. DNase I (EC 3.1.4.5) is a protein of roughly 40 8, diameter. It binds in the minor groove of the DNA and cuts the phosphodiester back- bone of both strands independently (3). Its bulk helps to prevent it from cutting the DNA under and around a bound protein. However, a bound protein will usually have other effects on the normal cleavage by DNase I, resulting in some sites becoming hypersensitive to DNase I (see Figs. 1 and 2). It is also not uncommon to observe a change in the pattern of DNase cleavage without any obvious extended protec- tion (Fig. 2). Unfortunately, DNase I does not cleave the DNA indiscriminately, some sequences being very rapidly attacked whereas others remain unscathed even after extensive digestion (4). This results in a rather uneven “ladder” of digestion products after electrophoresis, some- thing that limits the resolution of the technique (see naked DNA tracks in Figs. 1 and 2). However, when the protein-protected and naked DNA ladders are run alongside each other, the footprints are nor- mally quite apparent. To localize the position of the footprints, G + A and/or C + T chemical sequencing ladders of the same end-labeled DNA probe (5) should accompany the naked and protected tracks (see Note 9). Since a single end-labeled fragment allows one to visu- alize interactions on one strand only of the DNA, it is usual to repeat the experiment with the same fragment labeled on the other strand. DNA fragments can be conveniently 5’ labeled with T4 kinase and 3’ labeled using Klenow, T4 polymerase (fill out), or terminal transferase (6). A combination of the 5’ and 3’ end-labeling allows both DNA strands to be analyzed side by side from the same end of the DNA duplex. DNase I footprinting requires an excess of DNA-binding protein over the DNA fragment used. The higher the percent occupancy of a site on the DNA, the clearer a footprint will be observed. It is there- 4 Leblanc and Moss Fig. 2. Course of digestion with increasing amounts of DNase I. Here xUBF was footprinted on the Xenopus ribosomal promoter using a 5’ end-labeled fragment. The numbers above the tracks refer to the DNase I dilution, in U&L employed, and (-) and (+) refer to the naked and complexed DNAs respectively. The predominant footprints are indicated by open boxes. fore important not to titrate the available proteins with too much DNA. This limitation can in part be overcome when a protein also gener- ates a gel-shift. It is then feasible to fractionate the partially DNase digested protein-DNA complex by nondenaturing gel electrophore- DNase I Footprinting 5 sis and to excise the shifted band (which is then a homogeneous pro- tein-DNA complex) before analyzing the DNA by denaturing gel elec- trophoresis as in the standard footprint analysis (see Chapters 4, 6, and 21 in this volume). Footprinting crude or impure protein fractions usually requires that an excess of a nonspecific competitor DNA be added, The competi- tor binds nonspecific DNA-binding proteins as effectively as the spe- cific labeled target DNA fragment and hence, when present in sufficient excess, leaves the main part of the labeled DNA available for the sequence-specific protein. Homogeneous and highly enriched protein fractions usually do not require the presence of a nonspecific competitor during footprinting. When planning a footprinting experi- ment, it is a prerequisite to start by determining the optimal concen- tration of DNase I to be used. This will be a linear function of the amount of nonspecific DNA competitor but more importantly and less reproducibly, this will be a function of the amount and purity of the protein fraction added. As a general rule, more DNase is required if more protein is present in the binding reaction, whether or not this protein binds specifically. Thus, very different DNase concentrations may be required to produce the required degree of digestion on naked and protein-bound DNA. A careful titration of the DNase concentra- tion is therefore essential to optimize the detection of a footprint and can even make the difference between the detection or lack of detec- tion of a given interaction. The following protocol was developed to study the footprinting of the Xenopus ribosomal transcription factor xUBF, which is a rather weak DNA-binding protein, with a rather broad sequence specificity. The protocol is not original, being derived from several articles (I, 7). It does, however, represent a very practical approach that can be broadly applied. We recommend that the reader also refers to the avail- able literature for more information on the quantitative analysis of protein-DNA interactions by footprinting (8). 2. Materials 1. 2X Binding buffer: 20% glycerol, 0.2 mM EDTA, 1 rnM DTT, 20 mM HEPES, pH 7.9, and 4% polyvinyl alcohol (see Note 1). 2. Poly d(AT): 1 mg/mL in TE (10 mM Tris-HCl, pH 8.0, 1 rnM EDTA). Keep at -20°C (see Note 2). 6 Leblanc and Moss 3. End-labeled DNA fragment of high-specific activity (see Note 3). 4. Cofactor solution: 10 n&f MgCl*, 5 mM CaClz. 5. DNase 1 stock solution: A standardized vial of DNase I (D-4263, Sigma, St. Louis, MO) is dissolved in 50% glycerol, 135 n&f NaCl, 15 mM sodium acetate, pH 6.5, at 10 Kunitz U&L. This stock solution can be kept at -20°C for many months (seeN ote 4). 6. 1M KCl. 7. Reaction stop buffer: 1% SDS, 200 n&f NaCl, 20 mM EDTA, pH 8.0, 40 pg/mL tRNA (see Note 5). 8. 10X TBE buffer: 900 mM Tris-borate, pH 8.3, 20 mM EDTA. 9. Loading buffer: 7M urea, 0.1X TBE, 0.05% of xylene cyanol, and bromo- phenol blue. 10. Sequencing gel: 6% acrylamide, 7M urea, 1X TBE. 11. Phenol-chloroform (1: 1) saturated with 0.3M TNE (10 mM Tris-HCl, pH 8.3, 1 mM EDTA, 0.3M NaCl). 12. Ethanol 99% and ethanol 80%. Keep at -20°C. 13. 1M pyridine formate, pH 2.0. Keep at 4°C. 14. 1OM piperidine. 3. Methods The footprinting reaction is done in three stages: binding of the protein to the DNA, partial digestion of the protein-DNA complex with DNase I, and separation of the digestion fragments on a DNA sequencing gel. 1, The binding reaction IS performed in a total volume of 50 pL contain- ing 25 p.L of 2X binding buffer, 0.5 pL of 1 mg/mL poly d(AT), 2-3 ng of end-labeled DNA fragment (-15,000 cpm) (see Note 6), the protein fraction and 1M KC1 to bring the final KC1 concentration to 60 mM. The maximum volume of the protein fraction that can be used will depend on the salt concentration of this solution. The reaction is per- formed in a 1.5~mL Eppendorf tube. 2. Incubate on ice for 20 min. 3. During the binding reaction, dilute the DNase I stock solution in dis- tilled water at 0°C. We suggest working concentrations of about 0.0005- 0.1 Kunitz U&L, depending on the level of protein present (see Note 7 and step 5). A good range is the following: 0.0005; 0.001; 0.005; 0.002; 0.02; 0.08 Kunitz U&L. 4. After the incubation, transfer the reaction tubes in batches of eight to a rack at room temperature and add 50 pL of the cofactor solution to each. 5. Add 5 pL of the appropriate DNase I dilution to a tube every 15 s (from DNase I Footprinting the 0.0005-0.005 Kunitz U&L stocks for naked DNA; from the 0.002- 0.08 ones for DNA + proteins). 6. After 2 min digestion, each reaction is stopped by the addition of 100 PL of the stop solution (see Note 8). 7. After all the reactions have been processed, extract each reaction once with phenol-chloroform as follows: add 1 vol phenol-chloroform (1: 1) saturated with 0.3M TNE, vortex briefly, and centrifuge in a desktop microcentrifuge for about 10 min. Recover the top phase and transfer to a new mrcrocentrrfuge tube. 8. Add 2 vol(400 pL) of ethanol 99% (-20°C) and allow nucleic acids to precipitate at -80°C for 20 min. 9. Microcentrifuge for 15 min, -lO,OOOg, and remove the supernatant with a Pasteur pipet. Check the presence of a radioactive pellet with a Geiger counter before discarding the ethanol. 10. Add 200 pL of 80% ethanol (-20°C) to the pellet and microcentrifuge again for 5 min. After removing the supernatant, dry the pellets in a vacuum dessicator. 11. Resuspende ach pellet in 4.5 pL loading buffer, vortex, and centrifuge briefly. 12. A G + A ladder and a molecular weight marker should be run in parallel with the samples on a sequencing gel (see Note 9). The G + A ladder can be prepared as follows (5): -200,000 cpm of end-labeled DNA are diluted into 30 J.~LH z0 (no EDTA). 2 p.L of 1M pyridine formate, pH 2.0, are added and the solution incubated at 37°C for 15 min. One hun- dred fifty microliters of 1M piperidine are added directly and the solu- tion incubated at 90°C for 30 min in a well sealed tube (we use a 500~pL microcentrifuge tube in a thermal cycler). Add 20 J.~Lo f 3M sodium acetate and 500 pL of ethanol and precipitate at -80°C for lo-20 min. Microcentnfuge (lO,OOOg,1 0 min) and repeat the precipitation. Finally, redissolve the pellet in 200 pL of Hz0 and lyophilize. Resuspend in loading buffer and apply about 5,000 cpm/track. 13. Prerun a standard 6% acrylamide, sequencing gel (43 x 38 cm, 0.4 mm thick, 85 W) for 30 min before loading each of the aliquots from the DNase I digestion, plus the markers. Running buffer is 1X TBE. Wash the wells thoroughly with a syringe, denature the DNA for 2 min at 9O”C, and load with thin-ended micropipet tips. Run the gel hot to keep the DNA denatured (see Note 10). After the run, cover the gel in plastic wrap and expose it overnight at -7OOC with an intensifying screen. We use either a Cronex Lightning Plus (DuPont, Wilmington, DE) or Kyokko Special (Fuji, Japan) screens, the latter being about 30% less sensitive but also less expensive. Several different exposures will prob- ably be required to obtain suitable band densities. Leblanc and Moss 4. Notes 1. This binding buffer has been shown to work well for the transcription factor NF-1 (6), and in our laboratory for both the hUBF and xUBF factors and thus should work for many factors. Glycerol and PVA (an agent used to reduce the available water volume and hence concentrate the binding activity) are not mandatory. The original footprinting con- ditions of Galas and Schmitz (I) for the binding of the lac repressor on the Zuc operator were 10 mM cacodylate buffer, pH 8.0, 10 mM MgC12, 5 mM CaCl,, and 0.1 mM DTI’. Particular conditions of pH, cofactors, and ionic strength may need to be determined for an optimal binding of different factors to DNA. 2. Since poly d(IC), another nonspecific general competitor, has been shown to compete quite efficiently with G-C rich DNA sequences, poly d(AT) is prefered here. The choice of an appropriate nonspecific com- petitor (whether it is synthetic, as in this case, or natural, e.g., pBR322 or calf thymus DNA) may have to be determined empirically for the protein studied. When working with a pure or highly enriched protein, no competitor is usually needed. The DNase I concentration must then be reduced accordingly (to about naked DNA values). 3. Single-stranded breaks in the end-labeled DNA fragment must be avoided because they give false signals indistinguishable from genuine DNase I cleavage and hence can mask an otherwise good footprint. It is therefore advisable to check the fragment on a denaturing gel before use. Always use a freshly labeled fragment (3-4 d at the most) because radiochemical nicking will degrade it. 4. These standardized vials allow for very reproducible results. Glycerol will keep the enzyme from freezing, as repeated freeze-thaw cycles will greatly reduce its activity. 5. Do not be tempted to use too much RNA since it causes a very annoying fuzziness of the gel bands that prevents resolution of the individual bands. 6. The use of S’end-labeling with kinase in the presence of crude protein extracts can sometimes lead to a severe loss of signal because of the presence of phosphatases. In these cases 3’end-labeling by “fill out” with Klenow or T4 polymerase is to be prefered. 7. For naked DNA and very low amounts of protein, working stocks diluted to 0.0005-0.005 Kunitz U&L give a good range of digestion. 8. It is convenient to work with groups of eight samples during the DNase I digestion. Cofactor solution is added to eight samples at a time and then the DNase I digestions begun at 15 s intervals: 15 s after adding DNase to the eighth sample, stop solution is added to sample 1 and then to the other samples at 15 s intervals. DNase I Footprinting 9. In comparing a chemical sequencing ladder with the products of DNase I digestion, one must bear in mind that each band in the sequencing ladder corresponds to a fragment ending in the base preceding the one read because chemical modification and cleavage destroys the target base. For example, if a DNase I gel band corresponds in mobility to the sequence ladder band read as G in the sequence ACGT, then the DNase I cleavage occured between the bases C and G. DNase I cleaves the phosphodiester bond, leaving a 3’-OH, whereas the G + A and C + T sequencing reactions leave a 3’-P04, causing a mobility shift between the two types of cleavage ladders. This is a further potential source of error. However, in our experience the shift is less than half a base and hence cannot lead to an error in the deduced cleavage site. 10. Sequencing gels are not denaturing unless run hot (7M urea produces only a small reduction in the T,,, of the DNA). A double-stranded form of the DNA fragment is therefore often seen on the autoradiogram, espe- cially at low levels of DNase I digestion (see Fig. 2) and can sometimes be misinterpreted as a hypersensitive cleavage. By running a small quan- tity of undigested DNA fragment in parallel with the footprint this error can be avoided. Acknowledgments The authors wish to thank A. Ruiz-Carrillo for providing the autoradiogram in Fig. 1A. The work was supported by the Medical Research Council of Canada (MRC). T. Moss is presently an F.R.S.Q. “Chercheur-boursier” and B. Leblanc was until recently supported by a grant from the F.C.A.R. of Qudbec. References 1. Schmitz, A. and Galas, D. J. (1978) DNase I footprinting: a simple method for the detection of protein-DNA binding specificity Nucleic Acids Res. 5,3 157- 3170. 2. Rousseau, S., Renaud, J., and Ruiz-Carrillo, A. (1989) Basal expression of the histone H5 gene is controlled by positive and negative c&-acting sequences. Nucleic Acids Res. 17,7495-75 I 1. 3. Suck, D., Lahm, A., and Oefner, C. (1988) Structure refined to 2 8, of a nicked DNA octanucleotide complex with DNase I. Nature 332,464-468. 4. Drew, H. R. (1984) Structural specificities of five commonly used DNA nucle- ases. J. Mol. Biol. 176,535-557. 5. Maxam, A. M. and Gilbert, W. (1980) Sequencing end-labeled DNA with base- specific chemical cleavages, in Methods in Enzymology, vol. 65 (Grossman, L. and Moldave, K., eds), Academic, New York, pp. 499-560. 6 Current Protocols in Molecular Biology, Chapter 3 (1991) (Ausubel, F. M., 10 Leblanc and Moss Brent, R., Kingston, R. E., Moore, D. E., Smith, S. A., and Struhl, K., eds.), Greene and Wiley-Interscience, New York. 7. Walker, P. and Reeder, R. H. (1988) The Xenopus luevis ribosomal gene promoter contains a binding site for nuclear factor-l. Nucleic Acids Res. 16, 10,657-10,668. 8. Brenowitz, M., Senear, D. F., and Kingston, R. E. (1991) DNase footprint analysis of protein-DNA binding, in Current Protocols in Molecular Biology (Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. E., Smith, S. A., and Struhl, K., eds.), Greene and Wiley-Interscience, New York, pp. 12.4.1-12.4.11.