Time-resolved synchrotron X-ray “footprinting”, a new approach to the study of nucleic acid structure and function: application to protein-DNA interactions and RNA folding 1

JMB MS 1694 [24/1/97] J. Mol. Biol. (1997) 266, 144±159

Time-resolved Synchrotron X-ray ``Footprinting'', a New Approach to the Study of Nucleic Acid Structure and Function: Application to Protein ±DNA Interactions and RNA Folding Bianca Sclavi1, Sarah Woodson2*, Michael Sullivan1, Mark R. Chance1,3* and Michael Brenowitz3* 1

Department of Physiology and Biophysics, Albert Einstein College of Medicine of Yeshiva University, 1300 Morris Park Avenue, Bronx, New York 10461, USA 2 Department of Chemistry and Biochemistry, University of Maryland, College Park MD 20742-2021, USA 3

Department of Biochemistry Albert Einstein College of Medicine of Yeshiva University 1300 Morris Park Avenue Bronx, New York 10461, USA

Hydroxyl radicals (
OH) can cleave the phosphodiester backbone of nucleic acids and are valuable reagents in the study of nucleic acid structure and protein± nucleic acid interactions. Irradiation of solutions by high ¯ux ``white light'' X-ray beams based on bending magnet beamlines at the National Synchrotron Light Source (NSLS) yields suf®cient concentrations of
OH so that quantitative nuclease protection (``footprinting'') studies of DNA and RNA can be conducted with a duration of exposure in the range of 50 to 100 ms. The sensitivity of DNA and RNA to X-ray mediated
OH cleavage is equivalent. Both nucleic acids are completely protected from synchrotron X-ray induced cleavage by the presence of thiourea in the sample solution, demonstrating that cleavage is suppressed by a free radical scavenger. The utility of this time-dependent approach to footprinting is demonstrated with a synchrotron X-ray footprint of a protein ±DNA complex and by a time-resolved footprinting analysis of the Mg2‡-dependent folding of the Tetrahymena thermophilia L21 ScaI ribozyme RNA. Equilibrium titrations reveal differences among the ribozyme domains in the cooperativity of Mg2‡-dependent
OH protection. RNA
OH protection progress curves were obtained for several regions of the ribozyme over timescales of 30 seconds to several minutes. Progress curves ranging from 53.5 to 0.4 minÿ1 were obtained for the P4-P6 and P5 sub-domains and the P3-P7 domain, respectively. The
OH protection progress curves have been correlated with the available biochemical, structural and modeling data to generate a model of the ribozyme folding pathway. Rate differences observed for speci®c regions within domains provide evidence for steps in the folding pathway not previously observed. Synchrotron X-ray footprinting is a new approach of general applicability for the study of time-resolved structural changes of nucleic acid conformation and protein± nucleic acid complexes. # 1997 Academic Press Limited

*Corresponding authors

Keywords: RNA folding; synchrotron; X-rays; footprinting; kinetics

Internet addresses: [email protected]; [email protected]; [email protected] Author to contact with editorial questions: Dr. Michael Brenowitz [email protected] 718 430-3179 (Telephone) 718 430-8565 (FAX) Abbreviations used:
OH, hydroxyl radical; LacI, Lac repressor; NSLS, National Synchrotron Light Source, Brookhaven National Laboratories; OD, optical density. 0022±2836/97/060144±16 $25.00/0/mb960775

Introduction Nuclease protection or ``footprinting'' refers to assays in which the cleavage of the backbone of a nucleic acid polymer by an enzymatic or chemical nuclease is inhibited by the binding of a ligand to speci®c sequences of bases or by the conformation of the nucleic acid. When the reaction products are separated by gel electrophoresis, the decrease in radioactivity of bands corresponding to bases # 1997 Academic Press Limited

JMB MS 1694 [24/1/97] 145

Time Resolved X-ray Footprinting

within a protected sequence is a footprint. Footprinting of protein± DNA complexes was developed using the endonuclease DNase I (Galas & Schmitz, 1978; Schmitz & Galas, 1979) and has been subsequently extended to a variety of enzymatic and chemical nucleases (cf Revzin, 1993) including the hydroxyl radical (
OH) generated by the Fenton reaction using Fe-EDTA as a catalyst (Tullius & Dombroski, 1986; Tullius et al., 1987). Nucleic acid cleavage by
OH is predominantly dependent upon the solvent accessibility of the phosphodiester backbone (Dixon et al., 1991; Latham & Cech, 1989) and is relatively insensitive to base sequence and whether the nucleic acid is single or double stranded (Celander & Cech, 1990). Therefore,
OH cleavage is a sensitive probe of ligand association and the conformation of nucleic acids (e.g. Burkhoff & Tullius, 1987; Celander & Cech, 1991; Price & Tullius, 1993; Strahs & Brenowitz, 1994). The potential of
OH as a footprinting probe for time-dependent studies of protein± DNA interactions has been explored by Anderson and coworkers using peroxonitrous acid to generate
OH (King et al., 1992, 1993). An advantage of this reagent is that stable solutions of peroxonitrite can be readily prepared and used in the laboratory. However, the 2 seconds half-life of the disproportionation of peroxonitrite at neutral pH limits the time scales accessible using this reagent. Fe-EDTA cleavage of DNA on timescales on the order of seconds have not been obtainable in our laboratories (our unpublished results) and tens of minutes have been required in published studies of FeEDTA mediated cleavage of RNA (e.g. Celander & Cech, 1991; Murphy & Cech, 1993). It is for these reasons that a new approach to time-resolved
OH footprinting has been developed. The radiolysis of water by X-rays with energies from 100 eV up to the MeV range produces free electrons and
OH according to the overall reaction illustrated in equation (1) (Klassen, 1987): hn

H2O

‡ ÿ H2O ÿ! H2O‡ ‡ eÿ d ry ÿ! H3O ‡
OH ‡ eaq …1†

The
OH generated by this reaction can abstract a hydrogen from the C40 carbon of the ribose sugar of DNA and RNA leading to breakage of the phosphodiester backbone of the polymer (cf Stelter et al., 1976; von Sonntag & Dizdaroglu, 1977; von Sonntag & Schulte-Frohlinde, 1978; Beesk et al., 1979). Irradiation of aqueous solutions with various forms of radiation, including gamma-rays, beta particles, and fast neutrons has been successfully used to obtain structural data for sequencespeci®c protein± DNA complexes comparable to those obtained with chemically generated
OH footprinting methods (Hayes et al., 1990; Price & Tullius, 1992; Franchet-Beuzit et al., 1993). Bending magnet beamlines without intervening optics at the National Synchrotron Light Source (NSLS) deliver  1015 photons sÿ1 of a continuous spectrum of X-rays of energies of 1 to 100 keV in a

beam characterized by relatively little divergence (Shenoy et al., 1988). The high brightness of synchrotron X-ray sources, with a substantial proportion of the ¯ux at hard Xray energies, has been exploited in order to conduct time-resolved
OH footprinting studies. Studies of the cleavage of DNA and RNA demonstrate the feasibility of conducting footprinting experiments on timescales of milliseconds. An analysis of the Mg2‡-dependent folding of the Tetrahymena thermophila L-21 ScaI ribozyme (Grosshans & Cech, 1989; Celander & Cech, 1991; Banerjee & Turner, 1995; Zarrinkar & Williamson, 1994, 1996a,b) is presented that demonstrates the value of this approach to the study of RNA folding. Abstracts describing the development of this work have been published (Sclavi et al., 1994, 1995, 1996).

Results Synchrotron-mediated radiolytic cleavage of nucleic acids The ability of X-ray radiation to generate
OH radicals, resulting in the cleavage of the phosphodiester backbones of nucleic acids, is well documented (cf von Sonntag, 1987; Farhatazis & Rodgers, 1987). The synchrotron X-ray beam ¯ux of 75 mW impacting on the 1 mm deep aqueous samples yields a steady state [
OH] estimated to be 0.5 mM (see Discussion). The heating of the samples was measured to be <1 C per second beam exposure and thus has a negligible in¯uence on the cleavage reactions. Accurate ms exposures to the X-ray beam were obtained using a shutter consisting of a steel plate containing an adjustable slit (Figure 1; see Materials and Methods). The reaction products resulting from the exposure of 32 P labelled DNA and RNA to the synchrotron Xray beam are shown in Figures 2 and 3, respectively. The DNA ladder obtained using the synchrotron beam is comparable to that obtained when FeEDTA is used to generate
OH. The absence of electrophoretic bands on the DNA ladder with mobilities intermediate to those of the major reaction products is consistent with
OH cleavage via nucleophilic attack of the 40 carbon of the ribose sugar to generate a free 30 hydroxyl as the primary reaction product (Figure 2, insert). The reaction products obtained by synchrotron beam mediated cleavage of RNA are visually comparable to those obtained using Fe-EDTA to generate
OH (Figure 3; Celander & Cech, 1991). The variations in band intensity observed for RNA in the absence of Mg2‡ most likely re¯ects the sensitivity of
OH to the primary sequence of the RNA (Celander & Cech, 1990) or the presence of secondary or tertiary structure. Dose response curves relating 32P labelled DNA and 32P labelled RNA cleavage to exposure to the X-19C synchrotron beam are shown in Figure 4. The ef®ciencies of X-ray cleavage of DNA and

JMB MS 1694 [24/1/97] 146

Time Resolved X-ray Footprinting

Figure 1. Schematic representation of the experimental design used to achieve ms exposure to the X-ray beam. A microfuge tube is held in a holder on its side. The 10 ml sample is held by surface tension at the bottom of the tube. A steel shutter is released from outside the beamline hutch. The rate of travel and the length of the slit determines the length of X-ray exposure.

RNA are identical. The linearity of the semi-log plots is consistent with the expected Poisson distribution of the cleavage products (Brenowitz et al., 1986). Under ideal footprinting experiment conditions each nucleic acid molecule is nicked, on average only once (Brenowitz et al., 1986). Beam exposures in the range of 50 to 100 ms ful®ll this criteria by nicking 30% of the nucleic acids (Figure 4). Thus, millisecond beam exposures are suf®cient for footprinting both nucleic acids. In the samples exposed to the synchrotron X-ray beam, H2O is at 1109 M excess compared to the nucleic acid. The dilute aqueous solutions result in virtually all the X-rays incident on the sample being absorbed by the water. Thus, ribose oxidation via
OH produced by H2O radiolysis is expected to be the predominant mechanism of DNA and RNA cleavage; direct X-ray absorption contributes little to the cleavage reaction under such experimental conditions. Thiols such as thiourea are effective radical scavangers that can quench both
OH and the carbon based radicals that result from
OH attack of nucleic acids (Biagolow, 1987). The presence of thiourea equivalently protects the DNA and RNA from cleavage by the X-ray beam and completely quenches the cleavage reactions at a concentration of 10 mM (Figure 5). These results demonstrate that the synchrotron X-ray mediated cleavage of the nucleic acids is suppressed by a free radical scavenger. X-ray "footprinting" of a protein ±DNA complex Figure 6 shows densitometric traces along the direction of electrophoresis for the DNA bases that constitute a high af®nity binding sequence for Lac repressor (LacI). Thermodynamic studies have characterized the sequence-speci®c binding of LacI to the 16 bp operator used in these studies

Figure 2. Digital images obtained by phosphor storage imaging of a 20% polyacrylamide gel of the reaction products resulting from a 300 ms exposure to the synchrotron X-ray beam of a 185 bp DNA restriction fragment from plasmid pDW001 32P-labeled at the 30 termini of the HindIII restriction site. The lane designated Fe-EDTA was treated with the Fenton reaction (10 mM Fe, 20 mM EDTA, 1 mM ascorbate, 0.03% (v/v) H2O2) for ten minutes. The lane designated Control was treated identically with the experimental samples except that it was not exposed to the X-ray beam or to the chemical reagents. The images are unprocessed except for the compression of the 16 bit PhosphorImager2 ®le to an 8 bit (256 gray level) image suitable for printing using the ImageQuant2 software (Molecular Dynamics) and the default compression parameters. The insert is an enlargement of the indicated area.

(Brenowitz et al., 1991, 1993) and titration experiments were used to demonstrate saturation of the operator under these experimental conditions (data not shown). The
OH footprints of this 16 bp symmetric sequence are generally comparable whether radiolysis or Fe-EDTA catalysis is used to produce
OH although subtle differences in the protection patterns are apparent. Both the number of bases protected and the overall extent of their protection (i.e. the magnitude of the decrease in band density) are comparable for the two methods of
OH generation. The synchrotron X-ray results obtained for LacI are also very similar to radiolytic footprints of this protein obtained through radiolytic cleavage with several sources (Franchet-Beuzit et al., 1993).

JMB MS 1694 [24/1/97] 147

Time Resolved X-ray Footprinting

Figure 4. Dose response curves relating the amount of DNA and RNA that remains uncleaved to the time of synchrotron X-ray beam exposure. The top band on the gels, representing the full length nucleic acid molecules, was quantitated as described in Materials and Methods. The data sets shown represent multiple independent determinations. The dotted lines indicate the time of X-ray beam exposure required to nick 30% of the nucleic acids as required for conducting footprinting experiments.

Figure 3. Representative digital image obtained by phosphor storage imaging of a titration with Mg2‡ of the 32 P-50 labelled L-21 ScaI ribozyme. For each set of samples, several gels were run for different times in order to best resolve the bands in different regions of the RNA. Lanes denoted A and G correspond to RNA cut with RNases U2 and T1, respectively. Lane C is a control that was not exposed to the X-ray beam. Lane 0 has no Mg2‡ present. The values at the top of each lane denote [Mg2‡] in mM. The annotation along the left side of the image denotes the bases and regions of ribozyme secondary structure.

X-ray and Fe-EDTA cleavage of RNA The bases protected from X-ray mediated
OH cleavage (Figure 3) in the samples in which Mg2‡ was present were visually comparable to those observed using Fe-EDTA (Latham & Cech, 1989; Celander & Cech, 1991; Heuer et al., 1991; Murphy & Cech, 1993; Laggerbauer et al., 1994). Since

X-ray footprinting of an RNA folding reaction The secondary structure of the 388 base L-21 ScaI Tetrahymena thermophila ribozyme is shown in Figure 7; this representation of the ribozyme also represents its proposed tertiary structure based on molecular modeling, structural and biochemical studies (Cech et al., 1994). Upon addition of Mg2‡ certain bases became protected or hypersensitive to
OH relative to the sample exposed to the X-ray beam in the absence of Mg2‡. However, only
OH protections have been analyzed in the present studies. Groups of bases within local regions whose [Mg2‡] or time dependence were identical were analyzed as a ``block'' in order to increase the signal-to-noise of the quantitative analysis. This analysis is appropriate for groups of bases with comparable time dependencies, in a manner analogous to DNase I footprints of protein ±DNA interactions (Brenowitz et al., 1986).

Figure 5. Quenching of synchrotron dependent DNA cleavage by the free radical scavenger, thiourea. To each sample of pM concentrations of 32P-labelled DNA and 32 P-labelled RNA was added the indicated concentration of thiourea. The samples were then exposed to the X-ray beam for 100 ms. The abscissa of the graph shows the fraction of the nucleic acids that were cleaved normalized to the 20% cleavage determined in the absence of thiourea.

JMB MS 1694 [24/1/97] 148

Figure 6. Densitometric scans through the pDW001 DNA comprising the Lac repressor operator sequence (bold text) in the absence (black continuous lines) and presence of Lac repressor (red continuous lines). The top and bottom panels show
OH cleavage mediated by synchrotron X-rays and Fe-EDTA, respectively.

Time Resolved X-ray Footprinting

A striking feature of the
OH protection isotherms are the variations in their steepness as a function of [Mg2‡] (Figure 8, left columns). The Hill coef®cient, nH, is dependent upon this characteristic of isotherms and provides a measure of the minimum stoichiometry and cooperativity of a binding reaction (Hill, 1913). Within the catalytic core of the ribozyme, bases protected from
OH cleavage within the P7 helix (containing the guanosine-binding site), the P3 helix, and the single-stranded region, J8/7, are all characterized by values of nH > 4.0 (Figures 7 and 8; Table 1). In contrast, the Mg2‡ folding isotherms determined for the P4-P6 and P2 domains have signi®cantly lower nH values, from 1.5 (0.5) to 2.4 (0.4). While the precision of the values of nH are not suf®cient to differentiate variations in cooperativity within these domains, the values obtained for the P4-P6 and P2 domains are all signi®cantly different from nH 5 4.0. Kinetic analysis of Mg2‡-dependent RNA folding

regions of
OH protection are not represented by a step function, the ends of the regions enclosed by open or gray shaded boxes shown in Figure 7 should be considered to have a precision no greater than 1 base. The bases within the shaded boxes were protected to a lesser extent than the others. A subset of these protected bases were chosen for further quantitative thermodynamic and kinetic analysis; the color coding of Figure 7 indicates regions with comparable rates of reaction. Thermodynamic analysis of Mg2‡-dependent RNA folding The binding isotherms and progress curves shown in Figure 8 are transformed to re¯ect the ``fractional saturation'' (Y) of the protected sites by numerical ®tting of the transition upper and lower endpoints as described in Materials and Methods (equations (4) and (5)). Information regarding variations in the ``extent'' of
OH protection (i.e. the
A of the bands within the protected region) among the regions of the ribozyme is not present in the transformed data sets. The relative absence of protection in the Mg2‡ free samples (Figure 3) and the well de®ned lower and upper plateaus in all of the Mg2‡-dependent
OH protection isotherms demonstrates that complete Mg2‡-dependent transitions are observed in the synchrotron Xray experiments (Figure 8, left columns). The Kd values that describe the RNA folding isotherms (Table 1; Figure 8) are lower than those determined using Fe-EDTA to generate
OH (Celander & Cech, 1991) and a hybridization-competition assay (Zarrinkar & Williamson, 1994, 1996a,b). Minor differences in the experimental conditions of the three studies, such as temperature and buffer may account for this variation.

The differences in the thermodynamic cooperativity of Mg2‡ binding among the domains of the ribozyme are also re¯ected by the time dependent changes in the
OH protection pattern (hereafter referred to as ``
OH kinetics'') that occur upon the addition of Mg2‡. All of the progress curves are adequately described by a single exponential; additional kinetic phases are not discernible in any of the progress curves within the resolution of the these experiments (Figure 8). The rate constants determined for each region quantitated are summarized in Figure 7 and presented with their con®dence limits in Table 1. The
OH kinetics of the P7 and P3 helices and the single stranded region J8/7 comprising the ribozyme core are the slowest of the domains of the ribozyme (Figure 7, ®lled yellow). At the opposite end of the spectrum of rate constants, the
OH kinetics measured for the regions that constitute the ``core'' of the P4-P6 domain (Cate et al., 1996) are at the limit of time resolution (3.5 minÿ1) of the manual mixing methods employed in these studies (Figure 7, ®lled orange). Intradomain differences in the
OH kinetics are discernible within the P4-P6 domain. Speci®cally, regions of P5, P5a and J5/4 (Figure 7, ®lled purple) that are located on the outside of the folded domain (Cate et al., 1996) and the single-stranded region (bases 105± 106 and 259 ± 260; Figure 7, cross-hatched yellow) predicted to be involved in a triple helical scaffold in the core of the ribozyme (Michel & Westhof, 1990; Michel et al., 1990; Doudna & Cech, 1995), display rate constants signi®cantly slower than 3.5 minÿ1. The rate constants determined for the GAAA tetraloop in L5b and the bulge in P6a (Figure 7, hatched orange), regions that form a speci®c tertiary contact (Cate et al., 1996), appear slightly slower than 3.5 minÿ1. However, the time resolution of the

JMB MS 1694 [24/1/97] Time Resolved X-ray Footprinting

present experiments precludes the differentiation of processes on this timescale.

Discussion Radiolysis,
OH chemistry and nucleic acid degradation The radiolysis of water is a well understood process that produces primarily
OH and eÿ (cf von Sonntag, 1986; Farhatazis & Rodgers, 1987).
OH

149 abstraction of the C(4) hydrogen results in cleavage of the phosphodiester backbone (von Sonntag, 1987; Dizdaroglu & Bergtold, 1986). The nucleosides are also sites of attack by
OH and free electrons. Base damage infrequently results in direct cleavage of the phosphodiester backbone but can result in alkali-labile sites (Duplaa & Teoule, 1985). Base modi®cations are not detected in the present work but are a source of additional information that may be obtained from synchrotron X-ray footprinting studies.

Figure 7. Secondary structure of the L-21 ScaI Tetrahymena thermophila ribozyme (adapted from Cech et al., 1994). The outlined bases are protected from, or become hypersensitive to,
OH cleavage upon the addition of Mg2‡. The values of k (minÿ1) and nH determined for selected regions are shown in the colored boxes. The con®dence limits for these values are found in Table 1. The color coding of the Figure highlights sequences that have comparable rates of change of
OH reactivity.

JMB MS 1694 [24/1/97] 150

Time Resolved X-ray Footprinting

Figure 8.
OH protection isotherms (left hand panels) and progress curves at 10 mM MgCl2 (right hand panels) for the Mg2‡-dependent folding of the Tetrahymena ribozyme. The continuous lines represent the best-®t to equations (1) and (2a) and (1) and (2b) to the isotherms and progress curves, respectively.

JMB MS 1694 [24/1/97] 151

Time Resolved X-ray Footprinting Table 1. Equilibrium and kinetic constants for the Mg2 + -dependent folding of the L-21 ScaI ribozyme Base number P2 P2.1 P2.1 P2.1 P4 P4 P4 P4 J5/4 P5 P5a P5a P5a P5b P5b P5c P5c P5c P5c P6a P6a P6a P7 P3 J8/7

45±46 59±62 82±85 93±97 105±106 109±112 209±211 212±215 203±207 118±121 126±127 139±140 183±187 (`A' bulge) 151±152 (tetraloop) 153±154 167±169 171±173 174±176 177±178 221±222 224±225 (bulge) 259±260 263±268 272±277 299±306

k (minÿ1) nd 1.9 (0.4) 1.4 (ÿ0.4, ‡0.7) 2.6 (0.6) 2.0 (0.4) 53.5 53.2 53.4 1.6 (0.3) 1.4 (0.5) 2.2 (ÿ0.4, ‡0.7) 53.5 53.9 2.8 (ÿ0.6, ‡0.9) 53.7 53.5 53.2 2.5 (ÿ0.7, ‡1.1) 1.7 (0.4) 1.3 (0.4) 2.7 (ÿ0.7, ‡1.5) 1.8 (0.5) 1.0 (0.3) 0.5 (0.1) 0.4 (0.1)

nH 2.1 1.5 2.0 1.9 2.0 1.9

(0.5) (0.5) (0.5) (0.3) (0.3) (0.3) n.d. n.d. n.d. n.d. 1.6 (0.5) 2.3 (0.4) n.d. 2.4  0.4 n.d. n.d. 2.3 (0.4) 1.8 (0.3) 1.9 (0.2) n.d. n.d. n.d. 54 54 54

Kd (mM) 0.15 0.07 0.12 0.14 0.15 0.13

(0.02) (0.01) (0.02) (0.01) (0.02) (0.02) n.d. n.d. n.d. n.d. 0.15 (0.02) 0.15 (0.02) n.d. 0.20 (0.02) n.d. n.d. 0.18 (0.01) 0.15 (0.01) 0.14 (0.01) n.d. n.d. n.d. 0.2 0.2 0.2

n.d., not determined.

A key feature of X-ray footprinting is that rapid reactions can be studied without the addition of high concentrations of nucleases or nucleic acid modifying reagents. The chemical precursor for radiolytic footprinting, H2O, is always present at concentrations 1109 times greater than the nucleic acid present in the sample mixtures. Thus, the ratio of
OH-mediated and direct X-ray absorptionmediated cleavage events is proportional to the ratio of H2O and nucleic acid concentrations since their X-ray cross sections are similar. The overall correspondence of the DNA and RNA
OH footprints obtained using X-ray and Fe-EDTA argues for the dominance of solvent accessibility as the determining factor in nuclease protection (Figures 2 and 3). Direct X-ray cleavage would not be inhibited by protein-binding or tertiary structure due to the high penetration of X-rays at the energies present in the synchrotron beam. The result is that ms exposure of DNA and RNA samples to the synchrotron light source produces suf®cient breaks in the phosphodiester bond for footprinting (Figures 2 to 4). The equivalent cleavage of DNA and RNA by
OH generated by Xrays was initially surprising given the inef®ciency of
OH cleavage of RNA mediated by Fe-EDTA driven Fenton chemistry reported in the literature (for example, Latham & Cech, 1989). However, the use of Tris buffer (an effective
OH scavenger) and DTT instead of ascorbate/H2O2 in the published Fe-EDTA studies of RNA most likely accounts for the inef®cient cleavage, as comparable rates of DNA and RNA cleavage are obtained using Fe-EDTA when the reactions are conducted in the same buffer (unpublished data).

An advantage of a synchrotron light source for radiolytic footprinting is that a high dose of radicals is delivered to a small solution volume. The

Figure 9. A model of the folding pathway of the Tetrahymena ribozyme based upon the
OH kinetics data. The broken lines indicate reactions for which no data was directly collected.

JMB MS 1694 [24/1/97] 152

Time Resolved X-ray Footprinting

critical energy typical for a NSLS bending magnet beamline is 5 ±6 keV with a total ¯ux of 51  1015 photons secondÿ1 milliradianÿ1 delivered into a sample area of several mm2. Although alternative sources of radiation have comparable total ¯ux, the con®guration of the beam at the synchrotron results in the delivery of a much greater fraction of the ¯ux to the sample (see below). The small solution volumes that can be utilized also permit the parsimonious use of biological macromolecules. The >50 MHz frequency of the NSLS synchrotron results in radical generating pulses every 20 ns. This frequency provides an instantaneous concentration of radicals low enough so that bimolecular recombination of
OH does not dominate the chemistry. In effect, the synchrotron operates close to a continuous-wave mode. Ultra-fast pulse radiolysis sources, such as Van de Graaff generators, linear (electron) accelerators (LINAC) and Febetrons, create a much higher instantaneous concentration of radicals, but their ``cutting power'' is predicted to be much lower due to the short lifetime of the radicals (Hoshi et al., 1992; Kobayashi, 1994). A rotating anode source can have a total power output much greater than a synchrotron although the ¯ux is delivered over 4p steradians. Thus lower concentrations of radicals, compared with the synchrotron source, are produced in small sample volumes even when the samples are placed extremely close to this source. Timescales potentially accessible by synchrotron footprinting Consideration of radiolysis chemistry allows a prediction of the timescales accessible using the synchrotron source to be made. In the keV energy range, the interaction of X-rays with matter involves the photoelectric effect and Compton scattering with the former dominating X-ray± water interactions at and below energies of 20 keV. The absorption of a photon promotes an electron to an unbound state with kinetic energy equal to the incident photon energy minus the ionization energy. (Compton scattering also results in the production of secondary electrons.) The electrons produced by radiolysis are thermalized and deposit their energy in discrete ionizations of other water molecules. The ionized H2O molecules react with other H2O molecules to produce
OH as shown in equation (1). For every 100 eV of energy absorbed, the stoichiometry is obtained under anaerobic conditions (Buxton, 1987): 4:14 H2O 2:7 eÿ 7 H‡ ‡ 0 :6 1H
aq ‡ 2: ‡ 2:87
OH ‡ 0 :43H2 ‡ 0 :6 1H2O2 ‡ 0 :0 3HO2

…2†

(Additional products are likely to be present under the aerobic conditions used in the present studies.) The stoichiometric coef®cients on the right hand side are called G values, classically they represent

the number of molecules or ions produced per 100 eV thermalized. G values are converted to standard units by multiplying by 0.104, this yields 0.3 mmol of hydroxyl radical produced per Joule of energy thermalized in the solution (Hoshi et al., 1992). Based on these values, the [
OH] generated by exposure to the X-19C beamline can be estimated. During the ®rst hundreds of ms of exposure of H2O to the X-ray beam, [
OH] increases. However, as [
OH] increases, the lifetime of the radicals decreases as bimolecular recombination balances the production of new radicals and a steady state is achieved. Using the appropriate absorption coef®cients, the G value described above and the average power accepted by the samples on beamline X19C, 7.5  10ÿ2 J sÿ1 (75 mW), approximately 10 nmol sÿ1
OH is generated in a 10 ml solution yielding an effective dose rate (D
OH) of 1 mM sÿ1. At the steady state concentration of
OH, the dose rate equals the loss due to bimolecular recombination. This relationship can be approximated by: D
OH ˆ 2krec ‰
OHŠ2

…3†

For a dose rate of 1 mM sÿ1 and krec ˆ 5  109 Mÿ1 sÿ1, we estimate [
OH] ˆ  0.45 mM. Since [
OH]steady state  time is proportional to the amount of nucleic acid cleavage at constant ¯ux (Figure 4), the prediction of equation (3) is that [
OH] varies as the square root of the dose rate. Thus, if the entire ¯ux of a white beamline such as X-19C were focused on a sample using a toroidal mirror, the ¯ux would increase 25-fold and the footprinting timescale is predicted to decrease to 10 to 20 ms. Sequence-specific DNA-binding by Lac repressor Lac repressor (LacI) is a member of an extensive family of structurally and functionally homologous bacterial regulatory proteins (Weickert & Adhya, 1992). Common features of this family of proteins include the presence of a ``helix ± turn ±helix'' motif, their binding 2-fold symmetric DNA sequences as dimers and highly selective allosteric regulation. The speci®city of DNA binding is as great as 1  107 under some solution conditions (Ha et al., 1992). Our understanding of sequencespeci®c binding increased dramatically with the solution of the DNA ±protein co-crystal structures of the homologous Purine repressor (Schumacher et al.,1994) and of LacI repressor (Lewis et al., 1996). These complexes demonstrate a commonality of structure of the protein ± DNA interface for this family of proteins. Both proteins ``insert'' an ahelix from each monomer into adjacent major grooves on the same face of the double helix. An unexpected feature of the repressor ±DNA complexes is the presence of two hydrophobic amino acids intercalating into the minor groove in the middle of the binding site resulting in a bend in the DNA away from the protein.

JMB MS 1694 [24/1/97] 153

Time Resolved X-ray Footprinting

Studies conducted with cI-repressor demonstrated that the pattern of
OH protection characteristic of the helix ± turn ±helix DNA-binding proteins results from the reduced solvent accessibility of the 40 carbon of the ribose sugar in the minor groove (Tullius & Dombroski, 1986). The LacI
OH footprint is consistent with this interpretation; the bases that are protected from
OH are those that are in close proximity to the protein in the co-crystal structures. No perturbations in the
OH protection pattern due to the deformation of the DNA are readily apparent in either the X-ray or Fe-EDTA mediated footprints. In all aspects, the
OH footprints of LacI obtained using X-rays and Fe-EDTA are quite similar (Figure 6) and are consistent with
OH protection patterns being predominantly determined by the solvent accessibility of the bases. Despite the overall similarity of the footprints, there are subtle differences in the protection patterns obtained using X-rays and Fe-EDTA. Such differences may prove important since
OH reactivities not readily interpretable in this context of solvent accessibility have been observed in other systems (e.g. Strahs & Brenowitz, 1994). Quantitative analyses of protection patterns obtained by synchrotron X-ray footprinting and kinetic analysis of several protein ±DNA interaction reactions are in progress. Thermodynamic analysis of ribozyme folding The Tetrahymena ribozyme adopts a compact structure in the presence of divalent cations and Mg2‡ is required for formation of the catalytically active species (Cech, 1993 and references cited therein). Under the experimental conditions chosen for these studies, much of the ribozyme secondary structure is stable in the absence of Mg2‡; the addition of Mg2‡ predominantly induces formation of tertiary contacts (Latham & Cech, 1989). The ribozyme is fully active in the Na-cacodylate/ MgCl2 buffer used in these studies (J. Pan & S. W., unpublished results). Protection from
OH cleavage is believed to predominantly result from the decreased solvent accessibility of the phosphodiester backbone within the interior of the folded ribozyme (Latham & Cech, 1989; Celander & Cech, 1991; Heuer et al., 1991; Cate et al., 1996). Thus, the isotherms and progress curves determined from the
OH protection data are interpreted to re¯ect the folding of the ribozyme. It is assumed in this analysis that the extent of
OH cleavage of RNA is directly related to the solvent accessibility of the phosphodiester backbone. Quanti®cation of individual sites of protection as a function of [Mg2‡] reveals that the domains of the ribozyme bind Mg2‡ with varying degrees of cooperativity (Table 1). The three sites analyzed within the P3-P7 domain are all characterized by nH 5 4 (Table 1), consistent with the high Mg2‡ cooperativity observed in other studies (Wang & Cech, 1994; Zarrinkar & Williamson, 1994). The equivalence of nH values for the three sites within the

P3-P7 domain is indicative of the highly concerted nature of Mg2‡ binding by the ribozyme core. The high cooperativity of the Mg2‡-dependent folding of the P3-P7 domain contrasts with the more moderate values (nH values clustering around 2) determined for the other domains of the ribozyme (Table 1; Figure 8); these values are comparable with the average value of nH  3, determined previously (Celander & Cech, 1991). The conclusion that can be drawn from these results is that the effects of Mg2‡ binding are not uniformly distributed throughout the ribozyme. It is tempting to speculate that the binding of an additional Mg2‡ is linked to the folding of the P3P7 domain since nH cannot exceed the stoichiometry of the binding reaction (Hill, 1913). (However, identical Mg2‡ stoichiometries with the cooperativity of Mg2‡-dependent folding being more highly concerted in the P3-P7 domain is an interpretation not excluded by the data.) This notion is consistent with the observation that Mg2‡ binding follows the rate-limiting folding of P3 to form the active ribozyme (Zarrinkar & Williamson, 1994). It is probable that the additional Mg2‡ ion(s) stabilize the ribozyme core. Phosphorothioate substitution and rescue by Mn2‡ has identi®ed several divalent cation binding sites on the ribozyme, for example, along the phosphodiester backbone of the J8/7 single stranded region (Christian & Yarus, 1992, 1993). In the three-dimensional model of the Tetrahymena ribozyme core (Michel & Westhof, 1990) this highly conserved region is buried deep in the core of the domain and forms the interface between the P4-P6, P3-P7 and P1 domains. Kinetic analysis of ribozyme folding The results presented in this paper demonstrate that time-resolved synchrotron X-ray footprinting reveals time-dependent changes in local regions of
OH protection of RNA. In this assay the rate of formation of tertiary interactions is followed by measuring changes in the accessibility of local regions of the RNA to the
OH probe. The hybridization-competition assay of Zarrinkar & Williamson (1994, 1996a,b) provides a different view of the folding reaction by assaying the accessibility of single-stranded regions of the RNA to hybridization with an oligonucleotide probe. Timedependent reaction of unpaired guanosines with kethoxal has also been used detect slow steps in ribozyme folding (Banerjee & Turner, 1995). The internal consistency of the
OH kinetics data, as will be shown below, and the consistency of the
OH kinetics with other methods provides compelling evidence that these approaches accurately monitor the RNA folding reaction. Moreover, X-ray footprinting offers the advantage of being less invasive than other methods. In order to simplify the interpretation of the progress curves, we assume a single dominant folding pathway. Although a critical test of this assumption is needed (Thirumalai & Woodson, 1996), the

JMB MS 1694 [24/1/97] 154
OH kinetics data do allow a general picture of the relative rates of formation of tertiary contacts to be developed. Within the context of this assumption, these data (Figure 7) suggest a model for the Tetrahymena ribozyme folding pathway that is consistent with models developed by several laboratories using a variety of experimental approaches (Celander & Cech, 1991; Murphy & Cech, 1993; Zarrinkar & Williamson, 1994, 1996a,b; Banerjee & Turner, 1995). The fact that detectable lags in the slower progress curves were not observed (Figure 8; Zarrinkar & Williamson, 1994, 1996a,b) is evidence that folding is not strictly sequential. Rather, it is likely that each domain folds independently at its own intrinsic rate and that interdomain interactions modify the relative rates of subsequent steps in the folding pathway.

The folding of the P4-P6 domain The observation that the
OH kinetics of several regions (Figure 7, orange boxes) within the P4-P6 domain proceeds at rates 53.5 minÿ1 suggests that the initial step in the folding pathway is the concerted ``collapse'' of this domain into a stable folded structure (Flor et al., 1989). This conclusion is consistent with demonstrations that this domain can fold independently into a structure that can stabilize the ribozyme's core (Joyce et al., 1989; van der Horst et al., 1991; Murphy & Cech, 1993). The recently solved structure of this domain (Cate et al., 1996) directly demonstrates the tertiary structure interactions critical to the structure of the domain. Comparable rate constants were observed in those cases where ``both sides'' of intradomain tertiary contacts were analyzed. The ``adenine-rich bulge'' of P5 makes a tertiary contact with P4 and is required for the folding of the P4-P6 domain (Flor et al., 1989; Murphy & Cech, 1993, 1994; Cate et al., 1996); the bulge and P4 show identical
OH kinetics. Both strands facing the A-rich bulge become protected at the same fast rate (Table 1; Figure 7). The protection of these bases results from the intradomain interaction since they persist in the footprint of the isolated domain (Murphy & Cech, 1993; Murphy et al., 1994). In addition, the protection pattern of these bases agrees with their solvent accessibility as calculated from the crystal structure (Cate et al., 1996). The self-consistency of these results supports the conclusion that the folding of the P4-P6 domain is concerted and is the initial step in the folding pathway. Cech and co-workers proposed such a model based upon the lower equilibrium binding constant for Mg2‡ in this region compared to the remainder of the ribozyme (Celander & Cech, 1991; Murphy & Cech, 1993). Differences in equilibrium binding constants cannot be necessarily interpreted in the context of folding kinetics. The present study provides direct evidence that the P4-P6 structure forms more rapidly than the remainder of the intron core.

Time Resolved X-ray Footprinting

Of interest is the proposed tertiary contact between the GAAA tetraloop in L5b and the bulge in P6a that connects the two arms of the ``U-shaped'' domain (Murphy & Cech, 1994; Cate et al., 1996). The data hint at the possibility that this tertiary contact forms more slowly than the core of the P4P6 domain, such as the adjacent P5b helix, although the con®dence limits on the determinations of the rate constants do not exclude the faster rates (Table 1). It is tempting to speculate that the tetraloop-bulge contact may serve to ``lock'' the folded tertiary structure. A critical test of this hypothesis will require experimental data on time scales much shorter than the 3.5 minÿ1 limit of resolution of the manual mixing methods used in these experiments. Experiments using a stopped¯ow apparatus will provide the time resolution required to discern these differences in rates (see below). The folding of the P3-P7 domain The
OH kinetics of P3 and J8/7 are the slowest measured in these studies, 0.5 and 0.4 minÿ1, respectively (Table 1; Figure 7). The J8/7 region is critically required for catalysis and is one of the most conserved sequences in group I introns. It has been proposed that J8/7 coordinates interactions in the active site of the ribozyme and directly contacts the P1 substrate helix (Michel & Westhof, 1990; Pyle et al., 1992). Thus, the slow
OH kinetics observed for the ribozyme core are consistent with the slow onset of catalytic function upon folding of the ribozyme. This result is consistent with the studies of Zarrinkar & Williamson (1996a) based on an analysis of ribozyme mutants using a hybridization-competition assay. The fact that these two assays give very similar results provides additional evidence that the stabilization of the double helices of P3 and P7 occurs concurrently with their interaction with the other domains. Interactions between the P4-P6 and P3-P7 domains The
OH kinetics of the two single-stranded regions connecting the P4-P6 domain (105 ± 106 in P4 and 259 ±260 in P6b) are slower (2.0 minÿ1) than the rates measured for their neighboring bases (53.5 minÿ1; Table 1; Figure 7). These bases are proposed to participate in an interaction with the P3-P7 domain via a triple helical structure (Michel & Westhof, 1990; Michel et al., 1990; Doudna & Cech, 1995). Interestingly, the
OH kinetics of these single-stranded regions are faster than those determined individually for the P3 and P7 helices and for the recovery of catalysis (Table 1; Figure 7). Furthermore, a mutation within these regions affects the folding rate of P3-P7 (Zarrinkar & Williamson, 1996a). Taken together, these results suggest interdomain tertiary contacts form subsequent to the folding of the P4-P6 domain and prior to the folding of the P3-P7 domain. This

JMB MS 1694 [24/1/97] 155

Time Resolved X-ray Footprinting

interpretation is consistent with P3-P7 being stabilized by its contacts with P4-P6. Proposed interactions of the P9 domain Several lines of evidence indicate that bases within P9 participate in tertiary interactions with the P4P6 and P3-P7 domains. Bases 126 ± 127 in P5a are solvent exposed in P4-P6 in the absence of the rest of the ribozyme (Murphy & Cech, 1993; Murphy et al., 1994). Deletion of the P9 domain (P9, 9.1 and 9.2) results in the loss of
OH protection in P5 and J5/4 (Laggerbauer et al., 1994). Thus, it is reasonable to speculate that the slower
OH kinetics of bases quantitated in P5, P5a and J5/4, which face the same side of the P4-P6 domain (Cate et al., 1996), represent interactions with the P9 domain (Table 1; Figure 7). An additional site that is solvent exposed in the absence of P9 includes bases 263 to 268 of P7 (Laggerbauer et al., 1994). The
OH kinetics of these bases are clearly ``slow'', but are slightly faster than those of the P3 and J8/7 (Table 1; Figure 7). The phosphodiester backbone in this region faces away from the ribozyme core in the Michel & Westhof (1990) model. Thus, the
OH kinetics are consistent with the interaction of P9 and P7 proposed by Michel & Westhof (1990) and Wang & Cech (1992). Additional evidence that P9 interacts with P7 is that the rate of folding of the P3-P7 domain is greatly reduced in the NheI mutant ribozyme (lacking the P9 domain) compared to the wild-type (Zarrinkar & Williamson, 1996b). A model for the folding of the ribozyme A model of the folding pathway of the Tetrahymena ribozyme based upon the
OH kinetics data is shown in Figure 9. This model is consistent with previously proposed models (Celander & Cech, 1991; Zarrinkar & Williamson, 1994, 1996a,b; Banerjee & Turner, 1995). An assumption of this model is that some, but not all, elements of secondary structure are present at 42 C in the absence of Mg2‡. The striking changes in the
OH protection pattern that occur upon the addition of Mg2‡, and the generally uniform cleavage in the absence of Mg2‡ (Figure 3), show that little stable tertiary structure is present in its absence. The
OH kinetics data do not determine whether there is an obligatory sequence of folding intermediates. Rather, the rate constants are averages that re¯ect the population distributions of RNA molecules. The fastest reaction we are able to observe is the highly concerted folding of P5 and P4-P6 resulting in the formation of the ``P4-P6 domain''. While most of this reaction occurs at a rate greater than the time resolution of the manual mixing methods employed, there is the suggestion of an intermediate step, the interaction of the tetraloop with the P6 bulge. The P9 domain presumably interacts on comparable timescales with the folded P4-P6 domain and P7. These interactions,

together with the formation of the triple helix, guide P3 and J8/7 into proximity with P6 to form the compact core of the ribozyme. P2 likely participates in this last step, as well as in bringing the P1 substrate helix to the active site. Critical testing of the predictions made in the model described above, and the identi®cation of additional intermediate steps, will require the analysis of more rapid timescales. In the case of RNA this is particularly important, as early studies of tRNA unfolding lead us to expect that some tertiary conformational transitions will occur over 50 to 100 ms timescales (Crothers et al., 1974). The studies presented in this paper demonstrate that quantitative
OH footprinting binding isotherms and kinetic progress curves can be obtained by ms exposure to a synchrotron X-ray beam and establish the foundation for a technique capable of obtaining kinetic progress curves with ms time resolution. Stopped-¯ow DNase I footprinting, which has been developed for the study of protein ±DNA interactions, allows the determination of kinetic constants on tens of ms timescales over a wide range of physiologically relevant conditions (Petri et al., 1995; Hsieh & Brenowitz, 1996). This stopped-¯ow approach has proven to be readily extendible to
OH footprinting utilizing the synchrotron X-ray beam (unpublished observations). Studies of RNA folding and protein-DNA interactions utilizing stopped-¯ow synchrotron X-ray footprinting are in progress. Materials and Methods Nucleic Acids An EcoR1/HindIII DNA restriction fragment of 185 bp was excised from the plasmid pDW001 (DalmaWeiszhausz, 1995). This fragment was 30 end-labeled at the HindIII restriction site with a single 32P labelled nucleotide and puri®ed as has been described (Strahs & Brenowitz, 1994). The Tetrahymena thermophila L-21 ribozyme was prepared as previously described (Zaug et al., 1988), except that the RNA was puri®ed by G-50 chromatography after transcription with T7 RNA polymerase. The RNA was precipitated, resuspended in TE buffer (10 mM Tris-HCl (pH 7.5), 1 mM Na-EDTA) and treated with phosphatase. The 50 end was radiolabeled by kinase and [g-32P]ATP. The 32P labelled RNA was puri®ed using a ChromaSpin TE-100 spin column (Clontech) and stored in TE buffer (pH 7.5) at 4 C. Protein The Escherichia coli Lac repressor (LacI) used in these studies was obtained by expression in E. coli of plasmid pMB1 (E. Jamison & M. B., unpublished results) in which the LacI gene was cloned into the pET14a vector. Ef®cient expression was obtained by changing the naturally occurring initiation codon GTG (Beyreuther et al., 1973; Farabaugh, 1978) to ATG. The LacI was puri®ed and stored as described for the E. coli Gal repressor (Hsieh et al., 1994) except for the following changes: the initial column used was a Pharmacia Mono S column equilibrated with buffer A (25 mM bis-Tris (pH 6.0),

JMB MS 1694 [24/1/97] 156 1 mM MgCl2, 1 mM DTT, 15% (v/v) glycerol) plus 230 mM KCl. The LacI is eluted with a 230 to 540 mM KCl gradient of buffer A. An 0.5 ml aliquot of buffer B (25 mM Tris (pH 8.0), 1 mM EDTA, 1 mM DTT, 15% glycerol, 600 mM KCl) was added to the collection tubes prior to elution in order to minimize the exposure of the protein to acid pH. The LacI containing fractions were dialyzed against buffer B, concentrated and applied to a 30 ml Sephacryl S-200 column equilibrated with buffer B. Synchrotron exposure The experiments utilizing 32P-labelled DNA were conducted in a buffer containing 25 mM Na-cacodylate, 100 mM KCl, 2 mM MgCl2 and 1 mM CaCl2 at pH 7.5. The experiments utilizing 32P-labelled RNA were conducted in a buffer containing 10 mM Na-cacodylate and 0.1 mM EDTA at pH 7.4 to which was added the indicated concentration of MgCl2. Samples of 32P-labelled DNA or 32P-labelled RNA (10 ml) were exposed to the X-ray beam while held by surface tension at the bottom of an open microcentrifuge tube placed on its side (Figure 1). The samples were 3 mm in diameter and 1 mm deep. Irradiation of the samples with X-rays was carried out using NSLS white light beamline X-19C. The X-19C beamline has 0.5 mm beryllium and 0.05 mm aluminum windows with no additional optics. The placement of the samples 21 meters from the source results in their accepting 0.12 milliradian of the 2 milliradian present in the beam. The estimated total power incident on the sample is 75 mW (Shenoy, 1988). Exposure of the 32P-labelled DNA and 32P-labelled RNA samples to the X-ray beam for times of 100 ms or less was accomplished using a gravity-driven shutter (Figure 1). A solenoid allows the remote dropping of a stainless steel plate containing an adjustable slit from outside the experimental hutch. The sample is protected from the X-ray beam by the steel plate when the shutter is either cocked or at rest. Upon release of the solenoid, the velocity of the steel plate approaches 1 meters/second-after a drop of several centimeters. The sample is exposed to the X-ray beam during passage of the slit; exposure time is determined by the width of the slit yielding exposure times of 1.0 to 100.0 ms. While longer exposure times could be achieved by repetitive drops of the shutter, exposure times of 100 ms to one second were achieved by implementing a gas piston into the design of the shutter that reproducibly controlled the rate of travel of the shutter. The exposure times of the shutter were calibrated using a visible light source and photomultiplier tube connected to a digital storage oscilloscope (data not shown). X-ray exposure times >one second were achieved using the beamline safety shutter and a stopwatch. Protein ± DNA footprinting The LacI footprint was obtained by mixing 37 nM LacI with an equal volume of 20 pM 32P-labelled DNA (30,000 cpm) and allowing the mixture to equilibrate (Brenowitz & Jamison, 1993). The samples were then exposed to the X-ray beam. Immediately following exposure to the X-ray beam, 5 ml of a 0.1 M thiourea, 0.2 M EDTA solution was added to quench any long lived radicals. An amount (30 ml) of cold ethanol was then added to the sample to precipitate the 32P-labelled DNA. The

Time Resolved X-ray Footprinting samples were maintained on wet ice until just prior to their separation by electrophoresis. At that time, the samples were precipitated in a dry ice bath, washed with cold 80% (v/v) ethanol, dried, resuspended in the formamide loading buffer and heat denatured at 95 C (King et al., 1993; Strahs & Brenowitz, 1994). Electrophoresis was conducted using either 10% or 20% (w/v) polyacrylamide denaturing gels. The gels were dried and exposed to a phosphor storage screen. Digital image ®les were obtained by scanning the phosphor storage screens with a PhosphorImager2 (Molecular Dynamics).

RNA footprinting The RNA folding kinetics experiments were conducted by mixing an 8 ml aliquot of 32P-labelled RNA at 5 nM (100,000 cpm), equilibrated at 42 C, with 2 ml of a MgCl2 stock solution to a ®nal concentration of 10 mM. Immediately after placing the sample in the holder of the shutter, the experimentalist exited the hutch and initiated the sequence of operations required to enable the X-ray beam. Once the X-ray beam was enabled, the shutter was dropped thereby exposing the sample. The ®rst data points were collected 30 seconds after addition of the Mg2‡ to the RNA solution. The RNA samples were resuspended in the urea loading buffer, heat denatured at 75 C and processed for electrophoresis as described above for DNA.

Densitometric analysis Quantitation of the digital images was conducted, where indicated, using the ImageQuant2 software following published protocols (Brenowitz et al., 1986, 1993; Strahs & Brenowitz, 1994). The fraction of uncleaved nucleic acid is determined by quantifying the intensity of the band on the electrophoretogram that is the full length single-stranded nucleic acid (Brenowitz et al., 1986a). Fractional saturation (Y) of individual sites was determined from the fractional protection of bands visualized on the PhosphorImager2 by non-linear least squares ®tting of the data against the equation: f ˆ m  Y‡ b

…4†

where Yˆ

KnH ‰Mg ŠnH 1 ‡ KnH ‰Mg ŠnH

…5a†

or Yˆ eÿkt

…5b †

for the equilibrium titration (5a) or kinetic progress (5b) curves (Brenowitz & Senear, 1989; Hsieh & Brenowitz, 1996) and mˆ1/(UU ± LL) and bˆLL/(LL ± UU) where LL and UU are the lower and upper transition endpoints, respectively, K is the equilibrium association constant, nH is the Hill coef®cient (Hill, 1913), k is the ®rst-order kinetic constant and t is time. The assumption is made in the kinetic analysis that the reverse reaction is negligible under these experimental conditions. Each data set was analyzed and scaled to Y using the best-®t transition endpoints. When multiple data sets were globally analyzed, each data set was weighted by the inverse of the square root of the variance of its individual ®t. Extensive discussions of the issues that must be considered in con-

JMB MS 1694 [24/1/97] Time Resolved X-ray Footprinting ducting and analyzing quantitative nuclease protection experiments have been published (Brenowitz et al., 1986, 1993; Brenowitz & Senear, 1989; Hayes et al., 1990; Koblan et al., 1992; Senear & Bolen, 1992; Baskin & Tullius, 1993).

Acknowledgements We thank Elizabeth Jamison for preparing the DNA and Lac repressor used in these studies. This work was supported by grants from the NIH (GM39929, GM51506 and GM52348), the NSF (MCB-9410748) and the Hirshl Weill-Caulier Trust. The synchrotron footprinting technology is being developed by the Regional Center for Time-Resolved Synchrotron Spectroscopy, a Biotechnology Research Resource supported by a grant (RR01633) from the Division of Research Resources of the National Institutes of Health. M.R.C. is the recipient of the Joseph & Anne Wunsch fellowship in Biophysical Engineering from the Albert Einstein College of Medicine. S.W. acknowledges the Pew Scholars Program and a Camille Dreyfus Teacher Scholar Award. The National Synchrotron Light Source of Brookhaven National Laboratories is supported by the Department of Energy, Division of Materials Sciences. The data in this paper are from a thesis to be submitted by Bianca Sclavi in partial ful®llment of the requirements for the degree of Doctor of Philosophy in the Sue Golding Graduate Division of Medical Sciences, Albert Einstein College of Medicine.

References Banerjee, A. R. & Turner, D. H. (1995). The time dependence of chemical modi®cation reveals slow steps in the folding of a group I ribozyme. Biochemistry, 34, 6504± 6512. Banerjee, A. R., Jaeger, J. A. & Turner, D. H. (1993). Thermal unfolding of a group I ribozyme: the lowtemperature transition is primarily disruption of tertiary structure. Biochemistry, 32, 153 ±163. Baskin, J. S. & Tullius, T. D. (1993) . Hydroxyl radical footprinting. Footprinting Techniques for Studying Nucleic Acid± Protein Complexe: A Volume of Separation, Detection, and Characterization of Biological Macromolecules (Revzin, A, ed.), pp. 75 ±106, Academic Press, New York. Beesk, F., Dizdargoglu, M., Schulte-Frohlinde, D. & von Sonntag, C. (1979). Radiation-induced DNA strand breaks in deoxygenated aqueous solutions. The formation of altered sugars as end groups. Int. J. Radiat. Biol. Relat. Stud. Phys. Chem. Med. 36, 565± 576. Beyreuther, K., Adler, K., Geisler, N. & Klemm, A. (1973). The amino-acid sequence of lac repressor. Proc. Natl Acad. Sci. USA, 70, 3576± 3580. Biagolow, J. E. (1987). Ionizing radiation in mammalian cells. Radiation Chemistry Principles & Applications (Farhatazis, I. & Rodgers, M. A., eds), pp. 321± 348, VCH Pubs, TX. Brenowitz, M. & Jamison, E. (1993). Modulation of the stability of a lac repressor mediated looped complex by temperature and ions: allosteric regulation by chloride. Biochemistry, 32, 8693± 8701.

157 Brenowitz, M. & Senear, D. F. (1989). DNase I footprint analysis of protein-DNA binding. Current Protocols in Molecular Biology (Ausubel, F. M., Brent, R, Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A. & Struhl, K., eds), Unit 12.4 edit. , John Wiley and Sons, New York. Brenowitz, M., Senear, D. F., Shea, M. A. & Ackers, G. K. (1986). Quantitative DNase footprint titration: a method for studying protein-DNA interactions. Methods. Enzymol. 130, 132± 181. Brenowitz, M., Pickar, A. & Jamison, E. (1991). The stability of a lac repressor mediated looped complex. Biochemistry, 30, 5986± 5998. Brenowitz, M., Senear, D. F., Jamison, L. & DalmaWeiszhausz, D. D. (1993) . Quantitative DNase I footprinting. Footprinting Techniques for Studying Nucleic Acid ± Protein Complexes A Volume of Separation, Detection, and Characterization of Biological Macromolecules (Revzin, A., ed.), pp. 1 ± 43, Academic Press, New York. Burkhoff, A. M. & Tullius, T. D. (1987). The unusual conformation adopted by the adenine tracts in kinetoplast DNA. Cell, 48, 935± 943. Buxton, G. V. (1987) . Radiation chemistry of the liquid state: (1) water and homogeneous aqueous solutions. Radiation Chemistry Principles & Applications (Farhatazis, I. & Rodgers, M. A., eds), pp. 321± 348, VCH Pubs, TX. Cate, J. H., Gooding, A. R., Podell, E., Zhou, K., Golden, B. L., Kundrot, C. E., Cech, T. R. & Doudna, J. A. (1996). Crystal structure of a group I ribozyme domain: principles of RNA packing. Science, 273, 1678± 1685. Cech, T. R. (1993). Structure and mechanism of the large catalytic RNAs: group I and group II introns and ribonuclease. The RNA World (Gesteland, R. F. & Atkins, J. F, eds) , pp. 239± 269, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Cech, T. R., Damberger, S. H. & Gutell, R. R. (1994). Representation of the secondary and tertiary structure of group I introns. Nature. Struct. Biol. 1, 273 ± 280. Celander, D. C. & Cech, T. R. (1990). Iron(II)-ethylenediaminetraacetic acid catalyzed cleavage of RNA and DNA oligonucleotides: similar reactivity toward single- and double-stranded forms. Biochemistry, 29, 1355± 1361. Celander, D. C. & Cech, T. R. (1991). Visualizing the higher order folding of a catalytic RNA molecule. Science, 251, 401± 407. Christian, E. L. & Yarus, M. (1992). Analysis of the role of phosphate oxygens in the group I intron from Tetrahymena. J. Mol. Biol. 228, 743± 758. Christian, E. L. & Yarus, M. (1993). Metal coordination sites that contribute to structure and catalysis in the group I intron from Tetrahymena. Biochemistry, 32, 4475± 4480. Crothers, D. M., Cole, P. E., Hilbers, C. W. & Shulman, R. G. (1974). The molecular mechanism of thermal unfolding of Escherichia coli formylmethionine transfer RNA. J. Mol. Biol. 87, 63± 88. Dalma-Weiszhausz, D. D. (1995). The Escherichia coli gal operon: cross-talk between positive and negative regulation of transcription PhD. thesis, Albert Einstein College of Medicine, Bronx, NY. Dixon, W. J., Hayes, J. J., Levin, J. R., Weidner, M. F., Dombroski, B. A. & Tullius, T. D. (1991). Hydroxyl radical footprinting. Methods Enzymol. 208, 380± 413.

JMB MS 1694 [24/1/97] 158 Dizdaroglu, M. & Bergtold, D. S. (1986). Characterization of free radical-induced base damage in DNA at biologically relevant levels. Anal. Biochem. 156, 182 ± 188. Doudna, J. A. & Cech, T. R. (1995). Self-assembly of a group I intron active site from its component tertiary structural domains. RNA, 1, 36 ±45. Duplaa, A. M. & Teoule, R. (1985). Sites of gamma radiation-induced DNA strand breaks after alkali treatment. Int. J. Radiat. Biol. 48, 19 ± 32. Farabaugh, P. J. (1978). Sequence of the lacI gene. Nature, 274, 765 ± 769. Farhatazis, I. & Rodgers, M. A. (1987). Radiation Chemistry Principles & Applications, VCH Pubs., TX. Flor, P. J., Flanegan, J. B. & Cech, T. R. (1989). A conserved base-pair within the helix P4 of the Tetrahymena ribozyme helps to form the tertiary structure required for self-splicing. EMBO J. 8, 3391± 3399. Franchet-Beuzit, J., Spotheim-Maurizot, M., Sabattier, R., Blazy-Baudras, B. & Charlier, M. (1993). Radiolytic footprinting. B rays, g photons and fast neutrons probe protein ± DNA interactions. Biochemistry, 32, 2104± 2110. Galas, D. & Schmitz, A. (1978). DNase footprinting: a simple method for the detection of protein-DNA binding speci®city. Nucl. Acids Res. 5, 3157± 3170. Grosshans, C. A. & Cech, T. R. (1989). Metal-ion requirements for sequence-speci®c endoribonuclease activity of the Tetrahymena ribozyme. Biochemistry, 28, 6888± 6894. Ha, J. H., Capp, M. W., Hohenwalter, M. D., Baskerville, M. & Record, M. T. Jr (1992). Thermodynamic stoichiometries of participation of water, cations and anions in speci®c and non-speci®c binding of lac repressor to DNA. Possible thermodynamic origins of the ``glutamate effect'' on protein ± DNA interactions. J. Mol. Biol. 228, 252 ± 264. Hayes, J. J., Kam, L. & Tullius, T. D. (1990). Footprinting protein-DNA complexes with gamma-rays. Methods Enzymol. 186, 545 ± 549. Heuer, T. S., Chandry, P. S., Belfort, M., Celander, D. W. & Cech, T. R. (1991). Folding of group I introns from bacteriophage T4 involves internalization of the catalytic core. Proc. Natl Acad. Sci. USA, 88, 11105± 11109. Hill, A. V. (1913). The combination of hemoglobin with oxygen and with carbon monoxide. Biochem. J. 7, 471 ± 480. Hoshi, M., Uehara, S., Yamamotos, O., Sawada, S., Asao, T., Kobayashi, K., Maezawa, H., Furusawa, Y., Hieda, K. & Yamada, T. (1992). Iron (II) sulphate (Fricke Solution) oxidation yields for 8.9 and 13.6 keV X-Rays from synchrotron radiation. Int. J. Radiat. Biol. 61, 21 ±27. Hsieh, M. & Brenowitz, M. (1996). Quantitative kinetics footprinting of protein-DNA association reactions. Methods Enzymol. 274, 478 ±492. Hsieh, M., Hensley, P., Brenowitz, M. & Fetrow, J. S. (1994). A molecular model of the inducer binding domain of the galactose repressor of Escherichia coli. J. Biol. Chem. 269, 13825± 13835. Joyce, G. F., van der Horst, G. & Inoue, T. (1989). Catalytic activity is retained in the Tetrahymena group I intron despite removal of the large extension of element P5. Nucl. Acids. Res. 17, 7879± 7889. King, P. A., Anderson, V. E., Edwards, J. O., Gustafson, G., Plumb, R. C. & Suggs, J. W. (1992). A stable solid that generates hydroxyl radical upon dissol-

Time Resolved X-ray Footprinting ution in aqueous solutions: reaction with proteins and nucleic acid. J. Am. Chem. Soc. 114, 5430± 5432. King, P. A., Jamison, E., Strahs, D., Anderson, V. E. & Brenowitz, M. (1993). Footprinting of protein ± DNA interactions using potassium peroxonitrite. Nucl. Acids Res. 21, 2473± 2478. Klassen, N. V. (1987) . Primary products in radiation chemistry. Radiation Chemistry Principles & Applications (Farhatazis, I. & Rodgers, M. A., eds), pp. 29±61, VCH Pubs., TX. Kobayashi, K. (1994). Radiobiology using synchrotron radiation. Proc. Enrico Fermi Inst. Phys. In the press. Koblan, K. S., Bain, D. L., Beckett, D., Shea, M. A. & Ackers, G. K. (1982). Analysis of site-speci®c interaction parameters in protein ±DNA complexes. Methods Enzymol. 210, 405±425. Laggerbauer, B., Murphy, F. L. & Cech, T. R. (1994). Two major tertiary folding transitions of the Tetrahymena catalytic RNA. EMBO J. 13, 2699± 2676. Latham, J. A. & Cech, T. R. (1989). De®ning the inside and outside of a catalytic RNA molecule. Science, 245, 276± 282. Lewis, M., Chang, G., Horton, N. C., Kercher, M. A., Pace, H. C., Schumacher, M. A., Brennen, R. G. & Lu, P. (1996). Crystal structure of the lactose operon repressor and its complexes with DNA and inducer. Science, 271, 1247± 1254. Michel, F. & Westhof, E. (1990). Modeling of the threedimensional architecture of group I catalytic introns based on comparative sequence analysis. J. Mol. Biol. 216, 585± 610. Michel, F., Ellington, A. D., Couture, S. & Szostak, J. W. (1990). Phylogenetic and genetic evidence for base triples in the catalytic domain of group I introns. Nature, 347, 578± 580. Murphy, F. L. & Cech, T. R. (1993). An independently folding domain of RNA tertiary structure within the Tetrahymena ribozyme. Biochemistry, 32, 5291± 5300. Murphy, F. L. & Cech, T. R. (1994). GAAA tetraloop and conserved bulge stabilize tertiary structure of a group I intron domain. J. Mol. Biol. 236, 49± 63. Murphy, F. L., Wang, Y.-H., Grif®th, J. D. & Cech, T. R. (1994). Coaxially stacked RNA helices in the catalytic center of the Tetrahymena ribozyme. Science, 265, 1709± 1712. Petri, V., Hsieh, M. & Brenowitz, M. (1995). Thermodynamic and kinetic characterization of the binding of the TATA binding protein to the adenovirus E4 promoter. Biochemistry, 34, 9977± 9984. Price, M. A. & Tullius, T. D. (1992). Using hydroxyl radical to probe DNA structure. Methods Enzymol. 212, 194± 219. Price, M. A. & Tullius, T. D. (1993). How the structure of an adenine tract depends on sequence context: a new model for the structure of TnAn DNA sequences. Biochemistry, 32, 127± 136. Pyle, A. M., Murphy, F. L. & Cech, T. R. (1992). RNA substrate binding site in the catalytic core of the Tetrahymena ribozyme. Nature, 358, 123± 128. Revzin, A. (1993). Footprinting Techniques for Studying Nucleic Acid ± Protein Complexes: (A Volume of Separation, Detection, and Characterization of Biological Macromolecules), Academic Press, New York. Schmitz, A. & Galas, D. (1979). The interaction of RNA polymerase and lac repressor with the lac control region. Nucl. Acids Res. 6, 111± 137. Schumacher, M. A., Choi, K. Y., Zalkin, H. & Brennen, R. G. (1994). Crystal structure of LacI member

JMB MS 1694 [24/1/97] Time Resolved X-ray Footprinting PurR, bound to DNA: minor groove binding by a helices. Science, 266, 763 ± 769. Sclavi, B., Newman, J., Chance, M. R. & Brenowitz, M. (1994). A new method for DNA footprinting by photolysis of methyl- and adenosyl-cobalamin bound to DNA. Biophys. J. 66, A155. Sclavi, B., Chance, M. & Brenowitz, M. (1995). New methods for time-resolved footprinting. Biophys. J. 68, A296. Sclavi, B., Woodson, S., Sullivan, M., Chance, M. & Brenowitz, M. (1996). The application of synchrotron X-ray radiation to the development of time resolved methods for the study of RNA folding. Biophys. J. 70, A443. Senear, D. F. & Bolen, D. W. (1992). Simultaneous analysis for testing of models and parameter estimation. Methods Enzymol. 210, 463 ±481. Shenoy, G. K., Viccaro, P. J. & Mills, D. M. (1988). Characteristics of the 7-GeV Advanced Photon Source : A Guide for Users, Argonne National Laboratory publication ANL-88-9. Stelter, L., Von Sonntag, C. & Schulte-Frohlinde, D. (1976). Phosphate ester cleavage in ribose-5-phosphate induced by OH radicals in deoxygenated aqueous solution. The effect of Fe(II) and Fe(III) ions. Int. J. Radiat. Biol. Relat. Stud. Phys. Chem. Med. 29, 255 ± 269. Strahs, D. & Brenowitz, M. (1994). Correlation of DNA conformational changes with the cooperative binding of cI-repressor of bacteriophage l. J. Mol. Biol. 244, 494 ± 510. Thirumalai, D. & Woodson, S. A. (1996). Kinetics of folding of proteins and RNA. Acc. Chem. Res. In the press. Tullius, T. D. & Dombroski, B. A. (1986). Hydroxyl radical footprinting: high-resolution information about DNA± protein contacts and application to l repressor and Cro protein. Proc. Natl Acad. Sci. USA, 83, 5469± 5473. Tullius, T. D., Dombroski, B. A., Churchill, M. E. & Kam, L. (1987). Hydroxyl radical footprinting: a

159 high resolution method for mapping protein ± DNA contacts. Methods Enzymol. 155, 537± 558. van der Horst, G., Christian, A. & Inoue, T. (1991). Reconstitution of a group I intron self-splicing reaction with an activator RNA. Proc. Natl Acad. Sci. USA, 88, 184± 188. von Sonntag, C. (1987). The Chemical Basis of Radiation Biology, Taylor and Francis, London. von Sonntag, C. & Dizdaroglu, M. (1977). The reaction of OH radicals with D-ribose in deoxygenated and oxygenated aqueous solution. Carbohydr. Res. 58, 21±30. von Sonntag, C. & Schulte-Frohlinde, D. (1978). Radiation-induced degradation of the sugar in model compounds and in DNA. Mol. Biol. Biochem. Biophys. 27, 204± 226. Wang, J.-F. & Cech, T. R. (1992). Tertiary structure around the guanosine-binding site of the Tetrahymena ribozyme. Science, 256, 526±529. Wang, J.-F. & Cech, T. R. (1994). Metal ion dependence of active-site structure of the Tetrahymena ribozyme revealed by site-speci®c photo-cross-linking. J. Am. Chem. Soc. 116, 4178± 4182. Weickert, M. J. & Adhya, S. (1992). A family of bacterial regulators homologous to gal and lac repressors. J. Biol. Chem. 267, 15869± 15874. Zarrinkar, P. P. & Williamson, J. R. (1994). Kinetic intermediates in RNA folding. Science, 265, 918± 924. Zarrinkar, P. P. & Williamson, J. R. (1996a). The kinetic folding pathway of the Tetrahymena ribozyme reveals possible similarities between RNA and protein folding. Nature Struct. Biol. 3, 432± 438. Zarrinkar, P. P. & Williamson, J. R. (1996b). The P9.1P9.2 peripheral extension helps guide folding of the Tetrahymena ribozyme. Nucl. Acids Res. 24, 854 ± 858. Zaug, A. J., Grosshans, C. A. & Cech, T. R. (1988). Sequence-speci®c endoribonuclease activity of the Tetrahymena ribozyme: enhanced cleavage of certain oligonucleotide substrates that form mismatched ribozyme-substrate complexes. Biochemistry, 27, 8924± 8931.

Edited by D. E. Draper (Received 19 August 1996; received in revised form 5 November 1996; accepted 5 November 1996)