Analysis of the Structural Changes Caused by Positive DNA Supercoiling

by-

Marita Christine Barth

B.S. Bioresource Research Oregon State University, 1998

Submitted to the Division of Biological Engineering in partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Macromolecular Biochemistry and Biophysics at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY

February 2007

C 2007 Massachusetts Institute of Technology. All rights reserved.

Signature of Author: Division of Biological Engineering January 12, 2007

Certified by: (I Peter C. Dedon Professor of Toxicology and Biological Engineering Thesis Supervisor

Accepted by: L". % SLr/ .,: Alan Grodzinsky Irof's r of Biological Engineering rMASSACK·KIEMEWIJf nNte OF TECHNOLOGY /1 Chairman, Co ittee for Graduate Students AUG 0 2 2007I

LIBRARIES ARQHNi

This doctoral thesis has been examined by a committee of the Division of Biological Engineering as follows:

Associate Professor Bevin P. Engelward Chairman

Professor Peter C. Dedon Supervisor

Professor John M. Essigmann fl,(\

Dr. Richard J. Roberts ---

Analysis of the Structural Changes Caused by Positive DNA Supercoiling

by

Marita Christine Barth

Submitted to the Division of Biological Engineering on January 12, 2007 in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Macromolecular Biochemistry and Biophysics

ABSTRACT

The procession of helix-tracking enzymes along a DNA molecule results in the formation of supercoils in the DNA, with positive supercoiling (overwinding) generated ahead of the enzyme, and negative supercoiling (underwinding) in its wake. While the structural and physiological consequences of negative supercoiling have been well studied, technical challenges have prevented extensive examination of positively supercoiled DNA. Studies suggest that at sufficiently high levels of overwinding, DNA relieves strain by adopting an elongated structure, where the bases are positioned extrahelically and the backbones occupy the center of the helix. This transition has only been identified, however, at a degree of supercoiling substantially higher than is generated physiologically. To examine the structural changes resulting from physiological levels of positive DNA supercoiling, I have developed a method for preparing highly purified positively supercoiled plasmid substrates. Based on a method previously developed in this laboratory, this allows for preparation of large quantities of very pure, highly positively supercoiled plasmid. It also expands on earlier methods by exploiting ionic strength to modulate the direction of supercoiling introduced, allowing preparation of either positively or negatively supercoiled substrates. A combination of approaches has been used to elucidate changes to DNA structure that result from physiological levels of positive supercoiling. Enzymatic probes for regions of single-stranded character are not reactive with positively supercoiled plasmid, indicating that stably unpaired regions are not present. Additionally, the effect of supercoiling on the activity of restriction enzymes has been examined. With the enzymes tested, no substantial differences in cleavage rates were observed with either positively or negatively supercoiled substrates. To examine structural changes at a wider range of superhelical densities, design and preparation was undertaken on 2-aminopurine- containing DNA substrates for use in fluorescence studies with a magnetic micromanipulator. Technical limitations rendered these experiments infeasible with current instrumentation, but important insights were gained for future fluorescence-based micromanipulation experiments. A destabilizing effect on the base pairs, however, can be seen using Raman difference spectroscopy, suggesting a subtle shift toward the more extreme extrahelical state. The Raman data suggest that structural adjustments due to positive supercoiling are small but significant, and in addition to the base-pairing effects, alterations are observed in phosphodiester torsion and the minor groove environment, as well as a slight shift in sugar pucker conformation to accommodate lengthening of the DNA backbone. These results point to subtle changes in DNA structure caused by biologically relevant levels of positive superhelical tension and positive supercoiling. All of the changes are consistent with the mechanical effects of helical overwinding and suggest a model in which base pair destabilization in overwound DNA could affect the search mechanisms used by DNA repair enzymes and the binding of other proteins to DNA.

Thesis Supervisor: Peter C. Dedon Title: Professor of Toxicology and Biological Engineering Acknowledgements

I would like to start by thanking my family, particularly my parents, Merritt and Jenny, my grandmother Mary, my brother Stephen and my aunt Marita Jo Broadus for their support. They never failed to offer encouragement when I needed it, and I couldn't have done this without them.

The research experience I gained as an undergraduate was vital to my success in graduate school, and for that I wish to thank my undergraduate thesis advisor, Professor George S. Bailey, as well as Kate Mathews and Dr. Ulrich Harttig from his laboratory. They were all amazingly kind and encouraging, and I appreciate their willingness to work with and trust an undergraduate in a research environment.

I have gained a lot from my interactions with all current and former members of the Dedon Lab, and I would like to extend my thanks to all of them. I would particularly like to thank Debra Dederich, who has been invaluable in preparation of substrates for this research, and whose dedication and hard work in the face of many technical challenges is tremendously appreciated. I would also like to thank Dr. C. Eric Elmquist, who performed the MS analyses of chloroacetaldehyde-treated plasmid samples, and Dr. Michael DeMott who assisted greatly with the editing of this thesis. Dr. Koli Taghizadeh, Yelena Margolin, Dr. Min Dong, and Dr. Christiane Collins all gave important input into my research, and have been a pleasure to work with as well.

The support staff for the Dedon Lab has been absolutely amazing, and I would like to thank them all. Marcia Weir, Kristine Marzilli and Dawn Erickson have all done much to make my life easier, and have my gratitude. Olga Parkin is in a class all her own, with an uncanny knack for solving any problem that might come up. Her assistance in all matters administrative, as well as her friendship, have meant a great deal to me.

I would also like to thank several of the grad students and post-docs whose friendship I have enjoyed through the years: Dr. Maxine Jonas, David Appleyard, Vasileios Dendroulakis, Dr. Can Ozbal, Dr. Teresa Wright, Dr. Joe Newman, Dr. Maryann Timins, Dr. Elaine Chin, Dr. Janice Lansita, Dr. Jane Sohn, and J.P. Cosgrove.

This thesis project has involved multiple collaborations, which have significantly enhanced both my education and the quality of the information obtained. I would like to thank Professor George Thomas at the University of Missouri at Kansas City for opening up his lab to me for the Raman spectroscopic studies of supercoiling. Within the Thomas lab, I wish to thank Professor James Benevides for sharing his expertise with me, and allowing me so much instrument time on my visits to Kansas City. Additionally, he and his family were wonderful hosts, and truly made me feel welcome during my stays there. For their collaboration on the 2-aminopurine studies, I wish to thank Professor Peter So and Dr. Serge Pelet for their input and experimental assistance. I would like to thank my thesis committee, Professor Bevin Engelward, Professor John Essigmann and Dr. Richard Roberts for all of their support and input. They've been incredibly encouraging, and have been a wonderful source of challenging questions and intriguing ideas.

For help in more ways than I can list, I would like to thank Gavin McNett. His keen editorial eye and assistance with graphics have dramatically improved the quality of this thesis. Beyond that, his kindness, good nature, wit, phenomenal cooking skill, and (above all) patience have been vital to seeing me through the thesis writing process. I find it difficult to adequately express my gratitude and respect.

Finally, I would like to thank my thesis advisor, Professor Peter Dedon, for all his support and encouragement through the years. He has been incredibly generous and supportive, and I truly appreciate him sticking with me through what proved to be a very challenging and difficult project. His openness to new ideas and the freedom he has given me to explore different approaches to the problems I have faced have greatly enriched my education. I am truly grateful. Biographical Note

Marita Christine Barth was born in 1975 in Dallas, Oregon, and graduated from Dallas High School in 1993. After a summer internship at the Laboratory Services Division of the Oregon Department of Agriculture, she enrolled at Oregon State University in Corvallis, OR. Following a junior year abroad at Lincoln University in Canterbury, New Zealand, she returned to complete a thesis project, entitled "In vitro Mechanisms of Chlorophyllin Anticarcinogenesis Against Dibenzo[a, l]pyrene," in the laboratory of Professor George S. Bailey. She was awarded a Bachelor of Science in Bioresource Research with an option in Toxicology and a minor in Chemistry, in 1998. She subsequently enrolled at MIT, and during the course of her graduate studies has been awarded a National Defense Science and Engineering Graduate Fellowship, and a Poitras Pre-Doctoral Fellowship.

Table of Contents

Abstract ...... 5

Acknowledgements ...... 7

Biographical Note ...... 9

List of Abbreviations ...... 15

List of Figures ...... 17

Chapter 1 - Introduction

Packaging of DNA in Eukaryotic Cells ...... 19

Topological Domains and DNA Supercoiling ...... 20

The Twin-Domain Model of DNA Supercoiling ...... 21

Mathematical Descriptions of Supercoiling ...... 23

Energetics of Supercoiling ...... 26

Topoisomerases ...... 27

Physiological Roles for Negative Supercoiling ...... 30

Physiological Effects of Positive Supercoiling ...... 34

Structural changes in negatively supercoiled DNA ...... 35

Structural Changes in Positively Supercoiled DNA ...... 37

Figures ...... 41

Literature Citations ...... 47

Chapter 2 - Preparation of Positively and Negatively Supercoiled Substrates

Introduction ...... 55

Materials and Methods ...... 56

Results and Discussion ...... 60

11 Conclusions ...... 63

Figures ...... 65

Literature Citations ...... 68

Chapter 3 - Reactivity of Supercoiled DNA with Chemical and Enzymatic Probes of DNA Structure

Introduction ...... 71

Materials and Methods ...... 73

Results ...... 76

D iscussion ...... 81

Figures ...... 84

Literature Citations ...... 89

Chapter 4 - Raman Spectroscopic Studies of Positive Supercoiling

Introduction ...... 93

Materials and Methods ...... 95

Results ...... 96

D iscussion ...... 100

C onclusions ...... 105

Figures ...... 106

Literature Citations ...... 112

Chapter 5 - Design and Optimization of Fluorescent DNA Substrates for Micromanipulator Studies

Introduction ...... 115

Materials and Methods ...... 118

Results ...... 120

D iscussion ...... 125

Figures ...... 130 Literature Citations ...... 135

Chapter 6 - Effects of Supercoiling on Cleavage by Type II Restriction Enzymes

Introduction ...... 139

Materials and Methods ...... 142

R esults ...... 143

D iscussion ...... 146

Figures ...... 149

Literature Citations ...... 153

Chapter 7 - Conclusions

Conclusions ...... 157

Literature Citations ...... 164

Abbreviations

2AP 2-Aminopurine 6MI 6-Methylisoxanthopterin A Adenine BCIP 5-Bromo-4-chloro-3-indolyl phosphate bp Base pair(s) C Cytosine CBE Chicken blood extract dA 2'-Deoxyadenosine dAPTP 2-Amino-2'-deoxyadenosine-5'-triphosphate dC 2'-Deoxycytidine ALk Linking number difference dG 2'-Deoxyguanosine DIG Digoxygenin DMS Dimethyl sulfate dNTP 2'-Deoxynucleotide triphosphate dT 2'-Deoxythymidine DTT Dithiothreitol EDTA Ethylenediaminetetraacetic acid EA 1,N6-Ethenoadenine EC 3,N4-Ethenocytosine G Guanine HPLC High pressure liquid chromatography IPTG Isopropyl P-D-1 -thiogalactopyranoside Lk Linking number Lko Linking number of a completely relaxed molecule NBT p-Nitrotetrazolium blue NMR Nuclear magnetic resonance PCR Polymerase chain reaction SDS-PAGE Sodium dodecyl sulfate-polyacrylamide gel electrophoresis Ua Superhelical density T Thymine TBE Tris-borate-EDTA TIR Total internal reflection Tris 2-Amino-2-(hydroxymethyl)- 1,3-propanediol Tw Twist Wr Writhe

List of Figures

Figure 1.1: Packaging of eukaryotic DNA into chromatin

Figure 1.2: The twin-domain model of DNA supercoiling

Figure 1.3: Twist and writhe parameters of DNA supercoiling

Figure 1.4: Toroidal and plectonemic DNA writhe

Figure 1.5: Formation of a "chicken foot" structure at a stalled replication fork promoted by accumulated positive supercoiling

Figure 1.6: Two general types of cruciform extrusion at inverted repeat sequences of DNA

Figure 2.1: Reaction scheme for preparation of supercoiled plasmid substrates

Figure 2.2: Optimization and analysis of prepared supercoiled substrates

Figure 2.3: Purification of closed-circular plasmid preparation by extraction with acidic phenol

Figure 3.1: Kinetics of reaction of S1 nuclease with supercoiled and relaxed substrates

Figure 3.2: Reaction of S I nuclease with supercoiled and relaxed substrates

Figure 3.3: Kinetics of reaction of nuclease BAL-31 with supercoiled and relaxed substrates

Figure 3.4: Reaction of chloroacetaldehyde with DNA bases

Figure 3.5: Reaction of chloroacetaldehyde with supercoiled and relaxed substrates

Figure 4.1: Raman spectra of negatively supercoiled and relaxed pUC 18 plasmid DNA at 752 nm excitation

Figure 4.2: Raman spectra of negatively supercoiled and relaxed pUC 18 plasmid DNA at 532 nm excitation

Figure 4.3: Raman spectra of positively supercoiled and relaxed pUC 18 plasmid DNA at 752 nm excitation

Figure 4.4: Raman spectra of positively supercoiled and relaxed pUC 18 plasmid DNA at 532 nm excitation

Figure 4.5: Raman spectra of negatively supercoiled and relaxed pUC 18 plasmid DNA in D20 Figure 4.6: Raman spectra of positively supercoiled and relaxed pUC 18 plasmid DNA in D20

Figure 5.1: Hydrogen bonding schemes of the canonical Watson-Crick adenine- thymine base pair, the fluorescent base analog 2-aminopurine paired to thymine, and a proposed protonated form of 2AP that could pair with cytosine

Figure 5.2: Preparation of components of micromanipulator substrate

Figure 5.3: Assembly scheme for micromanipulator substrate

Figure 5.4: Fluorescence lifetime of free and incorporated 2-aminopurine

Figure 5.5: Comparison of original design of micromanipulator substrate with optimized design

Figure 6.1: Possible reaction modes of closed circular DNA with restriction enzymes

Figure 6.2: Kinetics of reaction of NarI with supercoiled and relaxed substrates

Figure 6.3: Kinetics of reaction of Ehel with supercoiled and relaxed substrates

Figure 6.4: Kinetics of reaction of EcoRI with supercoiled and relaxed substrates Chapter 1 - Introduction

Packaging of DNA in Eukaryotic Cells

The extensive length of genomes makes it necessary for both prokaryotic and

eukaryotic cells to compact their DNA. The human genome, for example, consists of

-3xl09 bp, and codes for an estimated 20,000-25,000 genes (1). For a single human diploid cell, this amount of DNA laid end-to-end would result in a length of more than two meters. Since the nucleus that must contain this DNA is a mere -10 -5 m in diameter,

a significant degree of compaction is necessary.

While complete compaction of the genome might be ideal for storage, as well as for cellular events like chromosomal segregation, other cellular processes periodically require at least some of the DNA to be unpackaged. RNA polymerases need to process along the DNA to function, so any compaction would need to be removed from transcriptionally active genes before transcription ensues. This means that in certain

stages of the cell cycle the DNA needs to be partially compacted and partially unpacked.

To further complicate the situation, cells need the ability to rapidly up- and down- regulate genes in response to stimuli, so compaction must be both dynamic and tightly regulated. All of this must be accomplished in a fashion that minimizes tangling and knotting of the DNA molecules.

The eukaryotic solution to the problem is to store the DNA as chromatin, which utilizes protein-DNA and protein-protein interactions to package the DNA in increasing levels of compaction (reviewed in Ref. 2). First, the naked DNA is wrapped into nucleosomes, in which a 147 bp segment of DNA is wrapped 1.8 times around a core of positively charged histone proteins in a left-handed toroid. The nucleosomes are separated from one another by 50-100 bp segments of linker DNA, giving extended nucleosomal arrays the appearance of "beads on a string" when viewed in electron micrographs. This nucleosomal array, also known as the 10 nm fiber, is further compacted into a secondary structure known as the 30 nm fiber, although debate persists as to whether this takes a solenoidal (3,4), or irregularly zig-zagging "slinky-like" form

(5,6). Further compaction of chromatin into a tertiary structure requires long-range interactions of the chromatin fiber, and it is affected by a number of different proteins in a process that is still not well understood (Fig. 1.1).

Topological Domains and DNA Supercoiling

One important consequence of the compaction into chromatin is the division of

DNA molecules into topological domains (7). The DNA in chromatin is anchored to the nuclear matrix at intervals, in such a fashion that it cannot freely rotate. The DNA between any two of these points is known as a topological domain. In an unconstrained linear DNA molecule, any change in twist that is introduced will diffuse through the molecule and will gradually dissipate, as one strand of the DNA rotates relative to the other. When such free rotation is limited, as happens within a topological domain, any added twist will be retained by the DNA, since it cannot diffuse past the boundaries of the domain. Topological domains vary dramatically in size. In human cells, the most recent estimate ranges from <95,000 to >150,000 bp (8), although other estimates offer a broader distribution from 1,000 to 300,000 bp for eukaryotic genomes (9-11). If an average of 100,000 bp/domain is assumed, this would suggest that a diploid human cell has -60,000 topological domains.

Supercoiling in DNA occurs when the DNA is either underwound (negative supercoiling) or overwound (positive supercoiling) relative to the B-form helix.

Constrained toroidal supercoiling, in the form of DNA wrapped around histone cores, is present in cells independent of any topological domains. This supercoiling is trapped by the binding of the proteins to the DNA, and as such cannot diffuse out. Because DNA wraps around histones in a left-handed fashion, the supercoiling that is constrained is negative.

The partitioning of cellular DNA into topological domains does allow the retention of unconstrained supercoiling. This type of supercoiling is free to diffuse through the molecule, within the limits set by the boundaries of the topological domain.

This normally occurs when the strands of DNA are separated during various cellular processes. The earliest detailed description of this phenomenon is the twin-domain model of transcription proposed by Liu and Wang (12).

The Twin-Domain Model of DNA Supercoiling

In the twin-domain model, Liu and Wang account for the consequences of the requirement for the RNA polymerase and the DNA template to rotate relative to each other during the transcription process (12). In theory, either molecule could rotate, however evidence indicates that polymerases are anchored to the nuclear matrix (13,14).

Therefore, for transcription to occur, the DNA must rotate relative to the polymerase. As this happens, negative supercoiling is generated behind the complex, while positive supercoiling is generated ahead of it. (Fig. 1.2) This supercoiling can diffuse rotationally along the helix, but cannot diffuse past the limits of the topological domain.

According to this model, the degree of supercoiling present in a topological domain is determined by the orientation of transcribed genes, by how transcriptionally active these genes are, and by the activity of different topoisomerases within the cell. If two transcribed genes are oriented in the same direction, the supercoils will eventually cancel each other in the region between the genes, as the positive supercoils generated ahead of the one gene are met by the negative supercoils generated behind the other. If the genes are oriented in opposite directions, significant supercoiling of either sign can be

generated between the two genes, as it will be accumulating from both directions.

Topoisomerases can affect the degree of supercoiling within a topological domain by

selectively adding or removing supercoils. If a topoisomerase that is removing

supercoiling of one sign is more active than the topoisomerase removing the other, an

imbalance will develop in the net amount of supercoiling. Topoisomerases will be

discussed in greater detail later in this chapter.

Evidence in support of the twin-domain model has been found in vitro (15,16), in

prokaryotes (17,18), and in eukaryotes (19,20), illustrating its importance as a widespread

biological phenomenon. The phenomenon also isn't limited to transcription. While the

original model described transcriptionally-induced supercoiling, any protein that forces

rotation of the DNA template can generate supercoiling in a similar fashion. This

includes helicases such as the E. coli nucleotide excision repair complex UvrAB (21) and

the SV40 large tumor antigen (22), and translocating enzymes such as the type I

EcoAI (23). Mathematical Descriptions of Supercoiling

Before delving into the structural and physiological consequences of supercoiling, it is useful to consider the mathematical conventions used to describe it. In most respects, a topological domain is analogous to a closed-circular DNA molecule, so several of the same concepts apply.

DNA has a double helical structure comprised of two interwound strands. The linking number of the DNA, Lk, is the number of times the strands are intertwined for a given molecule. For B-form DNA, the linking number of a completely relaxed molecule,

Lko, can be obtained by dividing the number of base pairs, n, by the helical repeat. In physiological solution conditions, the helical repeat is -10.4 bp/turn (24), so:

Lko = n/10.5

After torsion has been introduced into the molecule, Lk will change by 1 for every complete turn of the helix added (or removed). The linking number difference, ALk, is equal to the new linking number minus the relaxed linking number:

ALk= Lk- Lko

In a closed-circular molecule, both strands of the DNA are unbroken, which means that

Lk must be an integer value. From the definition of Lko, however, we can see that it is not necessarily an integer value. Thus, depending on the precise length of the DNA, some closed-circular molecules of DNA cannot reach a completely relaxed state. It also means that ALk is not necessarily an integer value. Because changes in ALk for an intact molecule can only occur by breaking one or more strands of the DNA and then introducing full turns before religation, changes in ALk must be of an integer value. It is also important to note that under consistent solution conditions, localized supercoiling can occur in an intact topological domain, but the net Lk for that domain cannot change without breakage of at least one strand of the DNA followed by re- ligation. This means that if one portion of an intact molecule is subjected to a localized positive linking number change, a compensatory negative supercoiling of equivalent Lk value must occur elsewhere in the domain.

Two closed circular molecules differing only in ALk are referred to as topoisomers. Populations of topoisomers can be resolved using agarose gel electrophoresis, as an increase in I ALkJ will result in increased mobility of a plasmid on a gel (25).

Changes in the linking number describe differences between topoisomers of the

same molecule, but give no information about the length of DNA that is accommodating the introduced torsion. A ALk of +1 would put considerably more strain on a 100 bp molecule than it would on a 1,000 bp molecule. To account for the length of a molecule when describing supercoiling, the superhelical density, o, is used:

a = ALLk/Lko

Superhelical density values allow for comparison of the degree of supercoiling among molecules of different sizes. DNA isolated from most cells is negatively supercoiled; in

E. coli the superhelical density will generally vary from o Z -0.03 to -0.09, depending on

growth conditions (26-28).

Since the linking number represents the number of times the two strands of the

DNA are intertwined, it is tempting to think of changes in Lk as simple changes in the twist (or helical repeat) of the molecule. However, this is oversimplified. By way of analogy, consider an electrical cord that has one end held stationary while the other end is turned. As the number of turns increases and twists along the length accumulate, the cord will eventually begin to fold upon itself. DNA behaves in a similar fashion; Lk is partitioned between two components - twist and writhe:

Lk= Tw + Wr

The twist value represents the number of times the strands wrap around the helical axis, while writhe is the crossing of the helical axis over itself (Fig. 1.3). The double helical structure of DNA means that twist is always present, while writhe will only occur when

Lk • Lko. Therefore:

ALk = ATw + Wr

Because of this relationship between twist and writhe, any change in one will necessarily affect the other. Partitioning between the two can therefore be changed by any factor that affects either component. A number of factors have been identified that can shift this distribution.

A change in twist effectively means a change in the helical repeat of the DNA.

With underwinding, the helical repeat will be >10.4 bp/turn, while with overwinding it will be <10.4 bp/turn. Intercalating agents are one example of a factor that will change the twist component of DNA. They act by introducing localized unwinding of the DNA at the site of intercalation, which requires a compensatory change in the writhe component of the DNA. Depending on the degree of intercalator binding, the writhe may actually have to be of the opposite sign, with a positive writhe offsetting an excess of negative twist. Writhe exists in two forms, toroidal and plectonemic. Toroids form when the

DNA is wrapped around a central axis such as a histone complex. Unconstrained supercoiling, such as occurs in the absence of toroidal wrapping in nucleosomes, takes a plectonemic form (29,30) (Fig. 1.4). Salt conditions have a significant effect on plectonemic writhing in DNA. The highly negatively charged backbone of DNA makes it difficult for different segments of the molecule to come into close proximity during the crossing-over that occurs in plectonemic writhing. Cations in the salt solution help to neutralize this charge repulsion, reducing the effective diameter of the DNA and allowing it to interwind more tightly (31). At physiological salt conditions, writhe has been estimated to account for approximately two-thirds to three-quarters of the linking number difference (29,30,32).

Energetics of Supercoiling

A DNA molecule is considered to be supercoiled if it has a ALk # 0 and there is a considerable free energy cost associated with supercoiling. The free energy of supercoiling (AG,,) has been empirically determined by measuring the binding affinities of intercalating dyes (33), and examining distribution of topoisomers (34,35). It has also been estimated computationally, using Monte Carlo simulations (32). The results are in general agreement (for a comparison, see ref. (32); the dependence of free energy on the degree of supercoiling is quadratic:

AGsc=KRT(ALk) 2/N

Where R is the gas constant, T is the absolute temperature, N is the number of base pairs, and K is a length- and ionic strength-dependent constant. As ionic strength increases, K decreases, resulting in a reduction in the free energy of supercoiling (35). The length dependence of K is more significant for shorter segments of DNA; above -2000 bp K is inversely proportional to N, and the value of NK becomes length independent (36).

The enthalpy (AH) and entropy (AS) of supercoiling have been estimated computationally by Monte Carle simulations (37), and also measured experimentally using microcalorimetry (38), Gibbs-Helmholtz (39), and van't Hoff (40) methods.

Experimental methods show the enthalpy of supercoiling to be positive, and substantially larger than the free energy. Given this, and the relationship:

AG = AH - (TAS)

We can conclude that the entropy of supercoiling must also be large and positive. This is in direct contradiction to the computationally predicted values, which calculate a negative entropy for supercoiling. Intuitively, a negative entropy makes sense: The limitations in the conformations which the DNA can adopt upon supercoiling would result in a negative entropy for supercoiling. This is not experimentally borne out, however, so the small decrease in configurational entropy must be more than compensated for by a much larger positive entropy component. This has been hypothesized to be due to the disruption of bound ions and water molecules resulting from the twisting and bending motions of the

DNA upon supercoiling (32), as well as changes in local interactions between base pairs and between base pairs and solvent (39).

Topoisomerases

Topoisomerases are the enzymes responsible for regulating and maintaining levels of supercoiling within a cell. They can be classified into two major types, I and II. Type I enzymes function via a swivel mechanism, cleaving one strand of the DNA and either allowing or driving rotation prior to re-sealing the cleaved strand (41,42). Type II topoisomerases act via a strand passage mechanism: Both strands of the DNA are cleaved and held open by the enzyme, like a gate, while another segment of the molecule is passed through the opening (43). The mechanisms are directly related to the degree to which the topoisomerases change the linking number of the molecule. Type I enzymes can alter ALk in single integer increments, while Type II always change the linking number in multiples of two.

Both major types of topoisomerase are further split into two sub-families, A and

B. Type lA enzymes pass the intact strand through the gap created by the single strand cleavage prior to religation, changing the linking number in steps of __l (44). While most lA enzymes require only a metal as a cofactor, reverse gyrase is a unique type lA that can introduce positive supercoils in an ATP-dependent fashion (45,46). Type 1B enzymes, while also cleaving a single strand, are thought to allow rotation around the intact strand, changing the linking number by an average of as much as ±5 in a single step

(47).

In most known cases, Type I topoisomerases act to remove superhelical tension from the molecule. In large molecules, however, thermal fluctuations in structure can occur that result in a molecule being apparently relaxed at the site of topoisomerase action while there is still some supercoiling present in the molecule. Because of this, whenever a Type I topoisomerase acts to relax a population of supercoiled molecules, a distribution of topoisomers, rather than an entirely relaxed population, will result. The pattern of topoisomer distribution is predictable and follows a Boltzmann distribution, with the median at the most relaxed possible state for the given conditions (34).

Type II topoisomerases utilize ATP to drive their reaction, and depending on the specific enzyme, can either relax DNA or introduce supercoils. Types IIA and IIB both exhibit similar mechanisms, with the primary difference between the two sub-families being global structural features. These enzymes change the superhelicity of a molecule using a sign inversion mechanism: At the site of a DNA crossover, when both strands of a molecule are cleaved and another segment of the molecule is passed through, it reverses the handedness of the crossover (e.g changing a positive writhe to a negative writhe), resulting in a change in ALk of ±2 (48). Because of the double-stranded nature of the break induced by Type II topoisomerases, under certain conditions they can catenate/decatenate and knot/unknot DNA in addition to their function in modulating superhelicity (49,50).

While the mechanism and degree of supercoiling change is defined by the subfamily to which a topoisomerase belongs, the directionality of the supercoiling that is removed (or introduced) varies with the enzyme. Some topoisomerases will only act on supercoils of a specific handedness, while others will remove supercoils of either sign.

The balance of activities of these different topoisomerase species controls the overall level of unconstrained supercoiling within a cell.

One well-studied example of this topoisomerase balance is in E. coli, where the level of supercoiling is tightly homeostatically regulated (51,52). It is primarily controlled by two enzymes: The type IA enzyme topoisomerase I is responsible for removal of negative supercoils (53), and the type II enzyme gyrase selectively removes positive supercoils and introduces negative ones (54,55). Increasing the level of negative supercoiling results in an up-regulation of topoisomerase I expression and a down- regulation of gyrase (56); reducing the level of supercoiling has the opposite effect (51).

Perturbation of the activity of either of these enzymes through gene alteration (53) or use of chemical inhibitors (57) will change the overall level of supercoiling in the cells. Until recently, a deletion of the topA gene coding for topoisomerase I was thought to be lethal without compensatory gyrase mutations (58). Strains have now been created where this is not the case, although deletion of topoisomerase III in addition to topA will result in non-viable cells (59).

Topological controls are also present in eukaryotic cells (19,60,61), suggesting that this phenomenon is of broad biological relevance. Taken together, this evidence suggests that tight control of supercoiling is important for cellular survival, and as such, superhelicity must play a vital physiological role.

Physiological Roles for Negative Supercoiling

The importance of supercoiling implied by these tight controls bears out, with many cellular processes are affected by supercoiling. Physiological effects can arise from the alteration in either the twist or the writhe of the molecule. Effects due to twist can result from the change in the helical repeat, transitioning to alternative secondary structures, and torsional tension on the helix. Writhe is required for compaction, facilitates looping and bending, and can bring distant elements on a single molecule into close proximity. One area that has been well studied in regards to supercoiling is gene expression; the supercoiling state of the molecule can affect gene expression via a number of different mechanisms.

One way in which negative supercoiling can affect gene expression is by assisting in open complex formation in the DNA (62). When the RNA polymerase complex initiates transcription, it first binds to the -35 and -10 elements in the promoter, forming a closed complex (63). This is followed by nucleation, where DNA in the -10 region is unwound. The open region of the DNA is then extended to the start site of transcription, allowing the polymerase complex to initiate transcription (reviewed in (64). Denaturation of DNA is energetically unfavorable, and creation of an open region of DNA by a protein carries a significant energetic cost. When the DNA is negatively supercoiled, this energetic cost can be "paid" by the supercoiling-associated free energy trapped in the molecule, concentrating the negative supercoiling into the denatured region and as a result relieving the superhelical strain on the remainder of the molecule (65).

While a high degree of negative supercoiling is useful for stabilizing denatured regions of DNA, other mechanisms that control gene expression are more precise in the degree of superhelical density required to exert an effect. Increasing the degree of negative supercoiling will not always increase the level of expression of a particular gene.

Early evidence of this was found by Stirdivant and co-workers, who examined two adjacent genes from the maize chloroplast chromosome and discovered that they were optimally expressed at different levels of supercoiling in vitro (66). Interestingly, while one gene exhibited an increase in expression followed by a plateau as negative supercoiling increased, the other showed an increase, peaked, and then decreased under the same conditions. This suggests that multiple mechanisms are responsible for the supercoiling-based changes in expression levels.

Steck and co-workers did a broader study in vivo and discovered that the optimal level of supercoiling for gene expression in E. coli could generally be correlated to the length of the spacer DNA between the -10 and -35 hexameric elements in the promoter region of the genes (67). Genes with a 17 bp spacer were optimally expressed at normal levels of supercoiling. For genes with a shorter spacer, a reduction in negative supercoiling increased expression, and for those with a long spacer, increased levels of negative supercoiling increased expression.

The spacer phenomenon can be explained in terms of the relative rotational orientation of the -10 and -35 elements on the molecule (68). The RNA polymerase complex binds to both elements during initiation, and as such the rotational orientation of the elements is an important determinant of binding. If 17 bp is an optimal spacer for the elements to be in a given orientation at physiological levels of supercoiling, then a longer spacer would need an increased level of negative supercoiling to have the same number of helical turns (and hence the same orientation) in between the two elements. The reverse is true for a shorter spacer.

The writhe component of DNA supercoiling can also play a role in gene expression. One notable example of how it may do so involves the ability of prokaryotic enhancers to affect transcription when they are located at a substantial distance from the promoter. This works by increasing the probability that the enhancer will be brought close enough to the promoter for the two to interact. In a molecule with no writhe, any effort to bring the two elements together is essentially a search in three-dimensional space. Normal fluctuations in the molecule may bring the two elements into proximity, but since the chances of the two elements meeting in three-dimensional space are low, these events are infrequent. When the molecule is crossed over itself, as in writhe, the search becomes two-dimensional; the molecule can move along itself in a slithering fashion (reptation), and in this manner dramatically increase the odds of bringing two distant sites into close proximity. A Monte Carlo simulation estimated that for sites separated by 3 kb, the probability of the two sites coming into contact is two orders of magnitude higher when the DNA is subjected to physiological levels of supercoiling than when it is relaxed (37). Experiments in vitro with the glnAP2 promoter and the NtrC- dependent enhancer from E. coli confirm this prediction: in a molecule where the promoter and enhancer are separated by 2500 bp, effective communication between the two elements is increased 50-fold when the molecule is negatively supercoiled relative to when it is relaxed (69). This observed difference was significantly reduced when the promoter and enhancer were located closer together on the molecule.

Control of gene expression by negative supercoiling is not limited to these mechanisms, nor are the physiological effects of negative supercoiling limited to control of gene expression. In addition to affecting the initiation of transcription, negative supercoiling has been demonstrated to play a role in transcription elongation, as well

(70). In other cellular processes, the twist and the writhe components of supercoiling have been implicated in the binding of histone proteins to DNA (71), replication (72), site-specific recombination (73), and transposition (74). Physiological Effects of Positive Supercoiling

The discussion thus far has been limited to the effects of negative supercoiling.

However, evidence suggests that positive supercoiling has important physiological consequences as well. These have not been as well elucidated; plasmid DNA isolated from eubacteria under normal conditions is negatively supercoiled, making it a much more easily obtained substrate than positively supercoiled plasmid, which must be prepared artificially. Additionally, because DNA isolated from most organisms is negatively supercoiled, it is a more obvious candidate for study. Still, the twin-domain model predicts the existence of at least transient positive supercoiling in cellular DNA, and this presence has been verified experimentally (17,19,75). Also, several proteins have been identified that can introduce either free or constrained positive supercoiling into DNA independent of a helix-tracking process, including the archaebacterial topoisomerase reverse gyrase (46), and the ubiquitous human chromatin-associated protein DEK (76). This suggests an important role for positive supercoiling in the functioning of the cell.

With the enhancing effect of negative supercoiling on transcription and replication, an obvious function of positive supercoiling would be to act as a negative regulator of these processes. Any process that requires unwinding of the helix will face a growing energy barrier to activation as the degree of positive supercoiling increases.

Positive supercoiling has been demonstrated to inhibit transcription in vitro at levels as low as a = +0.03, potentially acting by interfering with initiation and elongation steps

(77). Histone complexes bind with less affinity to positively supercoiled DNA (71,78), leading to the theory that waves of positive supercoiling generated ahead of helix- tracking enzymes can help to drive nucleosomes off the DNA, making it more accessible

(71,78). Accumulated positive supercoiling can force regression of a stalled replication fork to form a "chickenfoot" structure analogous to a Holliday junction (Fig. 1.5) (79).

Most recently, evidence has been found that positive supercoiling significantly enhances the rate of telomere resolution in the linear chromosomes of the Lyme disease spirochete

Borrelia burgdorferi (80). The diversity of processes that appear to be affected by positive supercoiling suggests a broad biological role, but considerably more information is necessary to get a more complete picture of its effects within a cell.

Structural changes in negatively supercoiled DNA

Since plasmid DNA isolated from eubacteria under standard growth conditions is negatively supercoiled, a ready substrate is available for researchers to use in examining the effects of helical unwinding on the structure of DNA, and the phenomenon is well- studied. Negative supercoiling increases the susceptibility of DNA to single-strand specific nucleases (81-83), a result that has been attributed to stabilization of regions of denatured base pairs within the DNA. The nature of negative supercoiling lends itself to denaturation; the two strands of the DNA molecule become less intertwined, making disruption of base pairing more likely as the level of underwinding increases. This phenomenon has been treated theoretically by Benham, who predicted that stably denatured regions would occur at physiologically expected levels of untwisting

(equivalent to c• -0.05) (65). As the level of negative supercoiling increases, so too does the likelihood of forming a denatured region. In some cases, certain sequences of DNA can undergo more substantial transitions to alternative secondary structures as a result of negative supercoiling. These alternative structures occur in much the same way as stably denatured regions, with the free energy that is stored in the molecule as supercoiling acting to stabilize otherwise unstable conformations. This concentration of free energy into a relatively small region of the

DNA molecule reduces the superhelical strain in the rest of the topological domain. One example of this phenomenon is cruciform extrusion, which occurs at sites of inverted repeats in the DNA.

Cruciforms are branched DNA structures in which an inverted repeat, the halves of which are usually separated by a short segment of non-palindromic sequence, extrudes from the helix to form two opposing stem-loop structures (84). Cruciform extrusion can happen by one of two processes. In C-type extrusions, the formation of the cruciform is preceded by the stable melting of the palindromic region of DNA, while S-type extrusions are more complex and involve the formation of partially extruded intermediates that develop into cruciforms through a branch migration process (Fig. 1.6).

The mechanism of extrusion is dictated by the sequence context of the inverted repeat; the sequences flanking the ColE 1 inverted repeat, a prototype for C-type extrusion, are

A-T rich and more likely to be prone to denaturation. Consistent with the requirement for large-scale melting, C-type cruciforms extrude preferentially under low salt, and have a high activation energy. S-type cruciforms, which are much more common, require salt of at least 50-60 mM concentration and have a lower activation energy for extrusion

(85,86). There is still some debate over the physiological role of cruciforms. Some studies have suggested that such structures can affect gene expression, forming above a certain negative superhelical density and acting as a block to the progression of the polymerase complex (77). The structure has also been demonstrated to be a specific binding site for regulatory proteins. In one instance, a combination of cruciform extrusion and the interaction of the cruciform with the RuvA protein from E. coli appears to act as a topological switch, regulating processes that require interaction of cis-acting DNA elements by controlling the reptation of supercoiled molecules (87).

Besides denatured regions and cruciforms, negative supercoiling can assist with the formation of other alternative structures as well. These include left-handed Z-DNA

(88) and triplex H-DNA (89), both of which have their own unique sequence and supercoiling requirements, and their own predicted biological function(s).

Structural Changes in Positively Supercoiled DNA

As is the case with physiological effects, much is unknown about the structural consequences of positive supercoiling in DNA. Again, since positively supercoiled substrates are much more difficult to obtain than the negatively supercoiled form, they have not been studied in as much depth. Most studies are limited by a low level of positive superhelical density in the substrates, or by a reliance on the presence of small molecules bound to the DNA helix to generate positive supercoiling. Much of the current information about structural consequences of positive supercoiling comes from computational studies, and suffers from a lack of supporting experimental evidence.

Understanding the structural changes that are induced by positive supercoiling is vital to further understanding the mechanisms of the physiological effects already hypothesized, as well as predicting effects that have not yet been identified.

An important insight into structural changes caused by positive supercoiling came from Allemand et al. (90). This group used single-molecule biomechanical experiments to look at global structural changes in DNA at varying levels of introduced torsion. A long (17,000 bp) piece of DNA was attached at one end to a solid support, and on the other to a paramagnetic bead. Magnets were used to rotate the bead, introducing precise amounts of either positive or negative torsion into the molecule. The magnets were also used to stretch the DNA, preventing writhe, and therefore forcing the entire ALk to be partitioned into twist. Both the length of the DNA and the force being used to pull on the bead can be measured using this system. To monitor changes in DNA structure resulting from introduced torsion, both length vs. force at a constant o, or length vs. oy at a constant force were measured.

In the case of underwound DNA, the molecule underwent denaturation as a means of relieving the overall torque (90). This phenomenon has been previously established through modeling (65), experiments with negatively supercoiled plasmids (39,91), and micromanipulator studies (92). With overwinding of the DNA, a different structural transition occurs, albeit at a higher level of force (3 pN vs. 0.3 pN for underwinding).

Modeling of the data led to a proposed structure with a helical repeat of 2.6 bp/turn where the DNA is significantly (-75%) longer than B-DNA, and the phosphate backbones occupy the center of the helix, with extruded bases along the outside. Similarities to the

DNA structure originally proposed by Linus Pauling (93) led to the structure being termed "P-DNA" (90). The solvent-exposed positioning of the bases was confirmed by reaction of the twisted molecules with glyoxal, a chemical agent that reacts selectively with unpaired bases (94).

In both under- and overwinding, the structural change occurs as a phase transition, where above a certain critical force (and associated critical torque) writhing becomes energetically unfavorable (90). At this point, the molecule adopts regions of the alternative structure to alleviate strain. Just as in the case of cruciform structure discussed previously, a small region of alternative structure can reduce the strain on the rest of the molecule considerably; in the case of overwinding, adopting the new structure reduces the twist in the rest of the molecule by three turns for every 10.4 bp converted.

While evidence for a non-base paired structure is compelling, much of the study was completed under conditions that would not be expected physiologically.

Experiments were carried out at low salt (10 mM phosphate), at which the helix is less stable than at physiological ionic strength; experiments in higher salt showed that greater thresholds were necessary to observe the same structural transitions. Additionally, the entire linking number difference was confined to twist, a partitioning that would not be expected in unrestrained supercoiling. Conditions resulting in this phase transition may be transiently generated immediately ahead of a processing helix tracking enzyme, but are less likely to be sustained in other circumstances.

At the same time, it may be possible for some degree of base unpairing to be present at lower levels of torsional strain. In negatively supercoiled plasmid, single strand specific nuclease sensitivity can be detected at values as low as c = -0.02 (82), despite the denaturation phase transition requiring significant pulling force to occur at a c = -0.015. Could some regions of P-DNA-like structure develop in DNA that is overtwisted to a lesser extent than the critical force that was identified in the micromanipulator experiments?

There is some experimental evidence to suggest this may be possible. Nuclease

BAL-31 reacts with DNA when enough of the intercalating agent ethidium bromide has bound to give an estimated superhelical density of Ya +0.15 (82). A long AT(,n) tract becomes sensitive to the unpaired-base specific agent osmium tetraoxide when positive supercoiling is induced by actinomycin D, a CpG binding agent (95). While these results are intriguing, both studies suffer from a major flaw: reliance on a bound molecule to introduce positive supercoiling prevents a realistic assessment of how positively supercoiled DNA behaves in the absence of a bound ligand.

The ability to prepare positively supercoiled substrates with a high superhelical density in the absence of extraneous ligands is crucial for further understanding of the structural consequences of positive supercoiling in DNA. In experiments presented in the succeeding chapters, I describe modifications to such a method that was previously developed in this laboratory that improve the consistency and purity of prepared positively supercoiled substrates. Using these substrates, I have performed chemical, enzymatic, biomechanical, and spectroscopic studies to characterize structural changes that occur with positive supercoiling. Taken together, these studies should provide an important guide for future research into the structural and physiological consequences of positive supercoiling in DNA. ~0~0~

DNA is wraps around histone complexes to form nucleosomes

Nucleosomes are compacted into the 30 nm fiber Ail

Chromatin loops attach to protein matrix to form topological domains

Further compaction leads to fully condensed chromosome

Figure 1.1. Packaging of eukaryotic DNA into chromatin. DNA is packaged in successive stages of compaction, going from naked DNA through nucleosomes and the 30 nm fiber to chromatin loops, and finally a completely condensed chromosome. Negative Supercoiling Positive Supercoiling (underwinding) (overwinding) > 10.4 bp/turn < 10.4 bp/turn X%^A#4%9

je helix-tracking enzyme as RNA polymerase)

Figure 1.2. The twin-domain model of DNA supercoiling. As a helix-tracking enzyme moves along a DNA molecule, the two must rotate relative to each other. If the enzyme is prevented from rotating, the DNA molecule itself must rotate, causing a change in the helical repeat of the DNA (supercoiling). Positive supercoiling (overwinding) will be generated ahead of the enzyme, while negative supercoling (underwinding) will be generated behind it. If the ends of the DNA are constrained in such a fashion that they cannot freely rotate, this supercoiling will be retained in the DNA. underwinding overwinding

B

Figure 1.3. Twist and writhe parameters of DNA supercoiling. A) Twist refers to the number of times the strands wrap around the helical axis. Supercoiled DNA can be either under- or over-twisted relative to relaxed B-DNA B) Writhe is the crossing of the DNA helical axis over itself. The direction of cross-over is determined by the direction of suDercoiline. A B

K~)

Figure 1.4. Toroidal and plectonemic DNA writhe. A) Toroidal writhe occurs when the DNA wraps around a central axis, such as a protein. B) Plectonemic writhe is unconstrained and occurs in the absence of toroidal wrapping. A

Figure 1.5: Formation of a "chicken foot" structure at a stalled replication fork promoted by accumulated positive supercoiling. A: A lesion on the leading strand causes replication to stall. Arrowheads indicate the 3' ends of the strands; parent strands are depicted in black while daughter strands are in red. B: A regression of the stalled replication fork promoted by accumulated positive supercoils ahead of the fork. The daughter strands anneal to each other due to this regression, forming a "toe" in the "chicken foot". Figure adapted from reference 74. 1111111 lllilllllIl1illnllilnlrir

A

!1111:!111

Figure 1.6: Two general types of cruciform extrusion at inverted repeat sequences in DNA. A. S-type extrusion: Extrusion is initiated by formation of a small denatured bubble that develops into opposing hairpin loops. Branch migration leads to a fully extruded cruciform structure. B. C-type extrusion: under low salt conditions a large denatured bubble forms. This denatured region form the full cruciform directly. Negative supercoiling is necessary for either type of extrusion; the type a particular sequence undergoes is sequence dependent. LITERATURE CITATIONS

1. (2004) Finishing the euchromatic sequence of the human genome. Nature, 431, 931-945.

2. Hansen, J.C. (2002) Conformational dynamics of the chromatin fiber in solution: determinants, mechanisms, and functions. Annu Rev Biophys Biomol Struct, 31, 361-392.

3. Finch, J.T. and Klug, A. (1976) Solenoidal model for superstructure in chromatin. Proc ANatl Acad Sci US A, 73, 1897-1901.

4. Robinson, P.J., Fairall, L., Huynh, V.A. and Rhodes, D. (2006) EM measurements define the dimensions of the "30-nm" chromatin fiber: evidence for a compact, interdigitated structure. Proc Natl Acad Sci USA, 103, 6506-6511.

5. Woodcock, C.L., Grigoryev, S.A., Horowitz, R.A. and Whitaker, N. (1993) A chromatin folding model that incorporates linker variability generates fibers resembling the native structures. Proc Natl Acad Sci U S A, 90, 9021-9025.

6. Dorigo, B., Schalch, T., Kulangara, A., Duda, S., Schroeder, R.R. and Richmond, T.J. (2004) Nucleosome arrays reveal the two-start organization of the chromatin fiber. Science, 306, 1571-1573.

7. Cook, P.R. and Brazell, I.A. (1975) Supercoils in human DNA. J Cell Sci, 19, 261-279.

8. Kramer, P.R. and Sinden, R.R. (1997) Measurement of unrestrained negative supercoiling and topological domain size in living human cells. Biochemistry, 36, 3151-3158.

9. Benyajati, C. and Worcel, A. (1976) Isolation, characterization, and structure of the folded interphase genome of Drosophila melanogaster. Cell, 9, 393-407.

10. Hofmann, J.F., Laroche, T., Brand, A.H. and Gasser, S.M. (1989) RAP-1 factor is necessary for DNA loop formation in vitro at the silent mating type locus HML. Cell, 57, 725-737.

11. Jackson, D.A., Dickinson, P. and Cook, P.R. (1990) The size of chromatin loops in HeLa cells. Embo J, 9, 567-571.

12. Liu, L.F. and Wang, J.C. (1987) Supercoiling of the DNA template during transcription. Proc Natl A cad Sci USA, 84, 7024-7027.

13. Jackson, D.A., Hassan, A.B., Errington, R.J. and Cook, P.R. (1993) Visualization of focal sites of transcription within human nuclei. Embo J, 12, 1059-1065. 14. Hozak, P., Hassan, A.B., Jackson, D.A. and Cook, P.R. (1993) Visualization of replication factories attached to nucleoskeleton. Cell, 73, 361-373.

15. Tsao, Y.P., Wu, H.Y. and Liu, L.F. (1989) Transcription-driven supercoiling of DNA: direct biochemical evidence from in vitro studies. Cell, 56, 111-118.

16. Leng, F. and McMacken, R. (2002) Potent stimulation of transcription-coupled DNA supercoiling by sequence-specific DNA-binding proteins. Proc Natl Acad Sci USA, 99, 9139-9144.

17. Wu, H.Y., Shyy, S.H., Wang, J.C. and Liu, L.F. (1988) Transcription generates positively and negatively supercoiled domains in the template. Cell, 53, 433-440.

18. Figueroa, N. and Bossi, L. (1988) Transcription induces gyration of the DNA template in Escherichia coli. Proc Natl Acad Sci USA, 85, 9416-9420.

19. Giaever, G.N. and Wang, J.C. (1988) Supercoiling of intracellular DNA can occur in eukaryotic cells. Cell, 55, 849-856.

20. Brill, S.J. and Sternglanz, R. (1988) Transcription-dependent DNA supercoiling in yeast DNA topoisomerase mutants. Cell, 54, 403-411.

21. Koo, H.S., Claassen, L., Grossman, L. and Liu, L.F. (1991) ATP-dependent partitioning of the DNA template into supercoiled domains by Escherichia coli UvrAB. Proc Natl Acad Sci US A, 88, 1212-1216.

22. Yang, L., Jessee, C.B., Lau, K., Zhang, H. and Liu, L.F. (1989) Template supercoiling during ATP-dependent DNA helix tracking: studies with simian virus 40 large tumor antigen. Proc Natl Acad Sci US A, 86, 6121-6125.

23. Janscak, P. and Bickle, T.A. (2000) DNA supercoiling during ATP-dependent DNA translocation by the type I restriction enzyme EcoAI. JMol Biol, 295, 1089- 1099.

24. Wang, J.C. (1979) Helical repeat of DNA in solution. Proc Natl Acad Sci USA, 76, 200-203.

25. Keller, W. and Wendel, I. (1975) Stepwise relaxation of supercoiled SV40 DNA. Cold Spring Harb Symp Quant Biol, 39 Pt 1, 199-208.

26. McClellan, J.A., Boublikova, P., Palecek, E. and Lilley, D.M. (1990) Superhelical torsion in cellular DNA responds directly to environmental and genetic factors. Proc Natl Acad Sci US A, 87, 8373-8377.

27. Kusano, S., Ding, Q., Fujita, N. and Ishihama, A. (1996) Promoter selectivity of Escherichia coli RNA polymerase E sigma 70 and E sigma 38 holoenzymes. Effect of DNA supercoiling. JBiol Chem, 271, 1998-2004. 28. Higgins, C.F., Dorman, C.J., Stirling, D.A., Waddell, L., Booth, I.R., May, G. and Bremer, E. (1988) A physiological role for DNA supercoiling in the osmotic regulation of gene expression in S. typhimurium and E. coli. Cell, 52, 569-584.

29. Adrian, M., ten Heggeler-Bordier, B., Wahli, W., Stasiak, A.Z., Stasiak, A. and Dubochet, J. (1990) Direct visualization of supercoiled DNA molecules in solution. Embo J, 9, 4551-4554.

30. Boles, T.C., White, J.H. and Cozzarelli, N.R. (1990) Structure of plectonemically supercoiled DNA. JMol Biol, 213, 931-951.

31. Bednar, J., Furrer, P., Stasiak, A., Dubochet, J., Egelman, E.H. and Bates, A.D. (1994) The twist, writhe and overall shape of supercoiled DNA change during counterion-induced transition from a loosely to a tightly interwound superhelix. Possible implications for DNA structure in vivo. JMol Biol, 235, 825-847.

32. Vologodskii, A.V. and Cozzarelli, N.R. (1994) Conformational and thermodynamic properties of supercoiled DNA. Annu Rev Biophys Biomol Struct, 23, 609-643.

33. Bauer, W. and Vinograd, J. (1970) Interaction of closed circular DNA with intercalative dyes. II. The free energy of superhelix formation in SV40 DNA. J Mol Biol, 47, 419-435.

34. Pulleyblank, D.E., Shure, M., Tang, D., Vinograd, J. and Vosberg, H.P. (1975) Action of nicking-closing enzyme on supercoiled and nonsupercoiled closed circular DNA: formation of a Boltzmann distribution of topological isomers. Proc Natl Acad Sci USA, 72, 4280-4284.

35. Rybenkov, V.V., Vologodskii, A.V. and Cozzarelli, N.R. (1997) The effect of ionic conditions on DNA helical repeat, effective diameter and free energy of supercoiling. Nucleic Acids Res, 25, 1412-1418.

36. Horowitz, D.S. and Wang, J.C. (1984) Torsional rigidity of DNA and length dependence of the free energy of DNA supercoiling. JMol Biol, 173, 75-91.

37. Vologodskii, A.V., Levene, S.D., Klenin, K.V., Frank-Kamenetskii, M. and Cozzarelli, N.R. (1992) Conformational and thermodynamic properties of supercoiled DNA. J Mol Biol, 227, 1224-1243.

38. Seidl, A. and Hinz, H.J. (1984) The free energy of DNA supercoiling is enthalpy- determined. Proc Natl Acad Sci US A, 81, 1312-1316.

39. Bauer, W.R. and Benham, C.J. (1993) The free energy, enthalpy and entropy of native and of partially denatured closed circular DNA. JMol Biol, 234, 1184- 1196. 40. Lee, C.H., Mizusawa, H. and Kakefuda, T. (1981) Unwinding of double-stranded DNA helix by dehydration. Proc Natl Acad Sci USA, 78, 2838-2842.

41. Champoux, J.J. and Dulbecco, R. (1972) An activity from mammalian cells that untwists superhelical DNA--a possible swivel for DNA replication (polyoma- ethidium bromide-mouse-embryo cells-dye binding assay). Proc Natl Acad Sci U SA, 69, 143-146.

42. Champoux, J.J. (1976) Evidence for an intermediate with a single-strand break in the reaction catalyzed by the DNA untwisting enzyme. Proc Natl Acad Sci US A, 73, 3488-3491.

43. Roca, J., Berger, J.M., Harrison, S.C. and Wang, J.C. (1996) DNA transport by a type II topoisomerase: direct evidence for a two-gate mechanism. Proc Natl Acad Sci USA, 93, 4057-4062.

44. Brown, P.O. and Cozzarelli, N.R. (1981) Catenation and knotting of duplex DNA by type 1 topoisomerases: a mechanistic parallel with type 2 topoisomerases. Proc Natl Acad Sci US A, 78, 843-847.

45. Kikuchi, A. and Asai, K. (1984) Reverse gyrase--a topoisomerase which introduces positive superhelical turns into DNA. Nature, 309, 677-681.

46. Forterre, P., Mirambeau, G., Jaxel, C., Nadal, M. and Duguet, M. (1985) High positive supercoiling in vitro catalyzed by an ATP and polyethylene glycol- stimulated topoisomerase from Sulfolobus acidocaldarius. Embo J, 4, 2123-2128.

47. Stivers, J.T., Harris, T.K. and Mildvan, A.S. (1997) Vaccinia DNA topoisomerase I: evidence supporting a free rotation mechanism for DNA supercoil relaxation. Biochemistry, 36, 5212-5222.

48. Brown, P.O. and Cozzarelli, N.R. (1979) A sign inversion mechanism for enzymatic supercoiling of DNA. Science, 206, 1081-1083.

49. Kreuzer, K.N. and Cozzarelli, N.R. (1980) Formation and resolution of DNA catenanes by DNA gyrase. Cell, 20, 245-254.

50. Liu, L.F., Liu, C.C. and Alberts, B.M. (1980) Type II DNA topoisomerases: enzymes that can unknot a topologically knotted DNA molecule via a reversible double-strand break. Cell, 19, 697-707.

51. Menzel, R. and Gellert, M. (1983) Regulation of the genes for E. coli DNA gyrase: homeostatic control of DNA supercoiling. Cell, 34, 105-113.

52. Snoep, J.L., van der Weijden, C.C., Andersen, H.W., Westerhoff, H.V. and Jensen, P.R. (2002) DNA supercoiling in Escherichia coli is under tight and subtle homeostatic control, involving gene-expression and metabolic regulation of both topoisomerase I and DNA gyrase. Eur J Biochem, 269, 1662-1669. 53. Pruss, G.J., Manes, S.H. and Drlica, K. (1982) Escherichia coli DNA topoisomerase I mutants: increased supercoiling is corrected by mutations near gyrase genes. Cell, 31, 35-42.

54. Gellert, M., Mizuuchi, K., O'Dea, M.H. and Nash, H.A. (1976) DNA gyrase: an enzyme that introduces superhelical turns into DNA. Proc Natl Acad Sci US A, 73, 3872-3876.

55. Ruthenburg, A.J., Graybosch, D.M., Huetsch, J.C. and Verdine, G.L. (2005) A superhelical spiral in the Escherichia coli DNA gyrase A C-terminal domain imparts unidirectional supercoiling bias. JBiol Chem, 280, 26177-26184.

56. Tse-Dinh, Y.C. (1985) Regulation of the Escherichia coli DNA topoisomerase I gene by DNA supercoiling. Nucleic Acids Res, 13, 4751-4763.

57. Lockshon, D. and Morris, D.R. (1983) Positively supercoiled plasmid DNA is produced by treatment of Escherichia coli with DNA gyrase inhibitors. Nucleic Acids Res, 11, 2999-3017.

58. DiNardo, S., Voelkel, K.A., Sternglanz, R., Reynolds, A.E. and Wright, A. (1982) Escherichia coli DNA topoisomerase I mutants have compensatory mutations in DNA gyrase genes. Cell, 31, 43-51.

59. Stupina, V.A. and Wang, J.C. (2005) Viability of Escherichia coli topA mutants lacking DNA topoisomerase I. J Biol Chem, 280, 355-360.

60. Esposito, F. and Sinden, R.R. (1987) Supercoiling in prokaryotic and eukaryotic DNA: changes in response to topological perturbation of plasmids in E. coli and SV40 in vitro, in nuclei and in CV-1 cells. Nucleic Acids Res, 15, 5105-5124.

61. Wang, Z. and Droge, P. (1996) Differential control of transcription-induced and overall DNA supercoiling by eukaryotic topoisomerases in vitro. Embo J, 15, 581-589.

62. Ehrlich, R., Larousse, A., Jacquet, M.A., Marin, M. and Reiss, C. (1985) In vitro transcription initiation from three different Escherichia coli promoters. Effect of supercoiling. Eur JBiochem, 148, 293-298.

63. Chamberlin, M.J. (1974) The selectivity of transcription. Annu Rev Biochem, 43, 721-775.

64. Travers, A. and Muskhelishvili, G. (2005) DNA supercoiling - a global transcriptional regulator for enterobacterial growth? Nat Rev Microbiol, 3, 157- 169.

65. Benham, C.J. (1979) Torsional stress and local denaturation in supercoiled DNA. Proc Natl Acad Sci U S A, 76, 3870-3874. 66. Stirdivant, S.M., Crossland, L.D. and Bogorad, L. (1985) DNA supercoiling affects in vitro transcription of two maize chloroplast genes differently. Proc Natl AcadSci USA, 82, 4886-4890.

67. Steck, T.R., Franco, R.J., Wang, J.Y. and Drlica, K. (1993) Topoisomerase mutations affect the relative abundance of many Escherichia coli proteins. Mol Microbiol, 10, 473-481.

68. Wang, J.Y. and Syvanen, M. (1992) DNA twist as a transcriptional sensor for environmental changes. Mol Microbiol, 6, 1861-1866.

69. Liu, Y., Bondarenko, V., Ninfa, A. and Studitsky, V.M. (2001) DNA supercoiling allows enhancer action over a large distance. ProcNatl Acad Sci US A, 98, 14883-14888.

70. Schultz, M.C., Brill, S.J., Ju, Q., Sternglanz, R. and Reeder, R.H. (1992) Topoisomerases and yeast rRNA transcription: negative supercoiling stimulates initiation and topoisomerase activity is required for elongation. Genes Dev, 6, 1332-1341.

71. Pfaffle, P., Gerlach, V., Bunzel, L. and Jackson, V. (1990) In vitro evidence that transcription-induced stress causes nucleosome dissolution and regeneration. J Biol Chem, 265, 16830-16840.

72. Funnell, B.E., Baker, T.A. and Kornberg, A. (1986) Complete enzymatic replication of plasmids containing the origin of the Escherichia coli chromosome. JBiol Chem, 261, 5616-5624.

73. Droge, P. (1993) Transcription-driven site-specific DNA recombination in vitro. Proc Natl Acad Sci US A, 90, 2759-2763.

74. Craigie, R., Arndt-Jovin, D.J. and Mizuuchi, K. (1985) A defined system for the DNA strand-transfer reaction at the initiation of bacteriophage Mu transposition: protein and DNA substrate requirements. Proc Natl Acad Sci US A, 82, 7570- 7574.

75. Ljungman, M. and Hanawalt, P.C. (1992) Localized torsional tension in the DNA of human cells. Proc Natl Acad Sci US A, 89, 6055-6059.

76. Waldmann, T., Eckerich, C., Baack, M. and Gruss, C. (2002) The ubiquitous chromatin protein DEK alters the structure of DNA by introducing positive supercoils. JBiol Chem, 277, 24988-24994.

77. Brahms, J.G., Dargouge, O., Brahms, S., Ohara, Y. and Vagner, V. (1985) Activation and inhibition of transcription by supercoiling. JMol Biol, 181, 455- 465. 78. Levchenko, V., Jackson, B. and Jackson, V. (2005) Histone release during transcription: displacement of the two H2A-H2B dimers in the nucleosome is dependent on different levels of transcription-induced positive stress. Biochemistry, 44, 5357-5372.

79. Postow, L., Ullsperger, C., Keller, R.W., Bustamante, C., Vologodskii, A.V. and Cozzarelli, N.R. (2001) Positive torsional strain causes the formation of a four- way junction at replication forks. JBiol Chem, 276, 2790-2796.

80. Bankldead, T., Kobryn, K. and Chaconas, G. (2006) Unexpected twist: harnessing the energy in positive supercoils to control telomere resolution. Mol Microbiol, 62, 895-905.

81. Beard, P., Morrow, J.F. and Berg, P. (1973) Cleavage of circular, superhelical simian virus 40 DNA to a linear duplex by Si nuclease. J Virol, 12, 1303-1313.

82. Lau, P.P. and Gray, H.B., Jr. (1979) Extracellular nucleases of Alteromonas espejiana BAL 31.IV. The single strand-specific deoxyriboendonuclease activity as a probe for regions of altered secondary structure in negatively and positively supercoiled closed circular DNA. Nucleic Acids Res, 6, 33 1-357.

83. LeBon, J.M., Kado, C.I., Rosenthal, L.J. and Chirikjian, J.G. (1978) DNA modifying enzymes of Agrobacterium tumefaciens: effect of DNA topoisomerase, restriction endonuclease, and unique DNA endonuclease on plasmid and plant DNA. Proc Natl Acad Sci US A, 75, 4097-4101.

84. Lilley, D.M. (1981) Hairpin-loop formation by inverted repeats in supercoiled DNA is a local and transmissible property. Nucleic Acids Res, 9, 1271-1289.

85. Lilley, D.M. (1985) The kinetic properties of cruciform extrusion are determined by DNA base-sequence. Nucleic Acids Res, 13, 1443-1465.

86. Sullivan, K.M. and Lilley, D.M. (1988) Helix stability and the mechanism of cruciform extrusion in supercoiled DNA molecules. Nucleic Acids Res, 16, 1079- 1093.

87. Shlyakhtenko, L.S., Hsieh, P., Grigoriev, M., Potaman, V.N., Sinden, R.R. and Lyubchenko, Y.L. (2000) A cruciform structural transition provides a molecular switch for chromosome structure and dynamics. JMol Biol, 296, 1169-1173.

88. Peck, L.J., Nordheim, A., Rich, A. and Wang, J.C. (1982) Flipping of cloned d(pCpG)n.d(pCpG)n DNA sequences from right- to left-handed helical structure by salt, Co(III), or negative supercoiling. Proc Natl Acad Sci USA, 79, 4560- 4564.

89. Beltran, R., Martinez-Balbas, A., Bernues, J., Bowater, R. and Azorin, F. (1993) Characterization of the zinc-induced structural transition to *H-DNA at a d(GA.CT)22 sequence. JMol Biol, 230, 966-978. 90. Allemand, J.F., Bensimon, D., Lavery, R. and Croquette, V. (1998) Stretched and overwound DNA forms a Pauling-like structure with exposed bases. Proc Natl AcadSci USA, 95, 14152-14157.

91. Kowalski, D., Natale, D.A. and Eddy, M.J. (1988) Stable DNA unwinding, not "breathing," accounts for single-strand-specific nuclease hypersensitivity of specific A+T-rich sequences. Proc Natl Acad Sci US A, 85, 9464-9468.

92. Strick, T.R., Allemand, J.F., Bensimon, D. and Croquette, V. (1998) Behavior of supercoiled DNA. Biophys J, 74, 2016-2028.

93. Pauling, L. and Corey, R.B. (1953) A Proposed Structure For The Nucleic Acids. Proc Natl Acad Sci US A, 39, 84-97.

94. Nakaya, K., Takenaka, O., Horinishi, H. and Shibata, K. (1968) Reactions of glyoxal with nucleic acids. Nucleotides and their component bases. Biochim Biophys Acta, 161, 23-31.

95. McClellan, J.A. and Lilley, D.M. (1991) Structural alteration in alternating adenine-thymine sequences in positively supercoiled DNA. JMol Biol, 219, 145- 149. Chapter 2 - Preparation of Positively and Negatively Supercoiled

Substrates

INTRODUCTION

The presence of superhelical tension in DNA is ubiquitous in the genomes of both prokaryotes and eukaryotes (1-4), and is critical to of the physiology of cellular DNA (5-

10). However, comparative studies of the physical and biological chemistry of different supercoiling states of DNA are hampered by the lack of methods to produce the negative, relaxed and positive states of DNA supercoiling in large quantity. While negatively supercoiled plasmid is readily available for study (plasmid isolated from E. coli typically has a superhelical density of -0.06), positively supercoiled DNA is not as readily obtained. Furthermore, modified bases and base analogs cannot readily be incorporated into supercoiled DNA, much less into DNA with different states of supercoiling. In order to obtain substrates for my studies, I have developed a method for the preparation of

DNA in the three states of supercoiling, thus providing biochemically well-controlled

DNA substrates for comparison of different states of superhelical tension.

Several strategies exist for obtaining supercoiled molecules, including interfering with the topoisomerase balance within growing cells (11), treatment of closed-circular molecules with enzymes that introduce supercoils, such as gyrase or reverse gyrase

(12,13), and binding of the DNA with an agent that will constrain supercoils followed by removal of compensatory free supercoils (14,15). While some (though by no means all) of these methods fall short with regards to ease of preparation, cost of production, and/or level of introduced superhelical density, they are all lacking in versatility; none allow for introduction of either positive or negative supercoils using a single method.

Using the method for preparing positively supercoiled plasmid developed previously by this lab (15) as a basis, I have developed a modified protocol to prepare either negatively or positively supercoiled plasmid with a high superhelical density from relaxed, closed circular DNA. This method exploits an observation that ionic strength determines the direction of DNA binding to the archaeal histone rHMfB (16), using salt conditions to control the sign of the supercoils induced in the plasmid substrates.

MATERIALS AND METHODS

Preparationofplasmid DNA. Plasmid pUC18 DNA was isolated from transformed E. coli strain DH5a using a standard alkaline lysis procedure. Prior to use,

10 mg quantities of DNA in 5 ml of TE (10 mM Tris. 1 mM EDTA; pH 8.0) were dialyzed in 100,000 MWCO dialysis tubing against 1 change of 4 L of 3 M NaCl in TE, followed by 4 changes of 4 L of TE alone.

Preparationof topoisomerase-containingextract. A topoisomerase-containing extract of chicken erythrocyte nuclei (herein referred to as chicken blood extract or CBE) was prepared from sterile, citrate-treated chicken blood (Rockland Immunochemical,

Gilbertsville, PA) according to the method of Camerini-Otero and Felsenfeld (17). We routinely process 200 mL of chicken blood in a single batch; the resulting extract is aliquoted and stored at -800 C. One unit of CBE was defined as the amount of the extract required to completely relax 1 pig of pUC18 under standard relaxation conditions (see below). The nicking activity of the CBE (i.e., endonuclease activity as opposed to topoisomerase activity) was assayed by resolving portions of the relaxed plasmid on 1% agarose gels containing 0.25 [tg/mL ethidium bromide such that relaxed closed-circular plasmid DNA was separated from single-strand nicked plasmid DNA. Batches of CBE that showed significant (>25%) nicking at a ratio of 10 units of CBE per ýpg of plasmid were discarded. Most batches of CBE caused nicking of less than 20% of the supercoiled pUC18 plasmid (data not shown). Inactivated CBE for use in preparing sham-treated substrates was prepared by heating CBE to 95 OC for 20 min, which eliminated all nicking and toposiomerase activity.

Isolation ofrHMfB. An rHMfB-expressing strain of E. coli was created by transforming electrocompetent JM105 with plasmid pKS323 (generously provided by K.

Sandman and J. Reeve, Ohio State University). rHMfB was isolated as described in (18), with several minor changes. Cells from IPTG-induced cultures were lysed by two passages through an Emulsiflex-C5 homogenizer (Avestin, Ottawa, ON) at 20,000 psi.

Affinity chromatographic purification of rHMfB was performed on a Hi-Prep 16/10

Heparin FF column (GE Healthcare, Piscataway, NJ) with buffer and gradient conditions as described elsewhere (18). Fractions containing rHMfB were identified by SDS polyacrylamide gel electrophoresis (SDS-PAGE), and protein in the pooled fractions was dialyzed against 4x rHMfB binding buffer (40 mM Tris, 4 mM EDTA, 8 mM K2HPO 4,

200 mM NaCl; pH 8.0), and stored at 4' C. Relaxation ofplasmid DNA. Negatively supercoiled pUC18 plasmid at a concentration of 50 ýpg/mL was relaxed by treatment with 1 unit of CBE per [Ig plasmid in 200 mM NaC1, 20 mM Tris, 0.25 mM EDTA and 5% glycerol (pH 8.0) for 1.5 h. at 37

TC. Reactions were stopped by addition of SDS to 1% and proteinase K to 150 [tg/mL with further incubation at 37 TC for 1.5 h. DNA was further purified by phenol/chloroform/isoamyl alcohol extraction followed by extraction with chloroform alone, and then ethanol precipitation. After resuspension in TE buffer, the DNA was desalted by passage over a NAP-25 column (GE Healthcare), and quantified spectrophotometrically at 260 nm.

Preparationofsupercoiledplasmid DNA. Prior to large-scale supercoiling reactions, the optimal ratio of rHMfB to DNA was determined using small-scale test reactions. For these, 1 ýtg purified relaxed pUC18 at 25 [tg/mL was bound to rHMfB at ratios between 0.5 and 2.5 [tg protein per [tg plasmid. Binding was carried out for 45 minutes at 370 C in lx rHMfB reaction buffer. After binding, unrestrained compensatory supercoiling was removed by addition of 10 units of CBE and adjustment of the buffer with 1/10 volume 590 mM Tris, 8 mM K2HPO 4, 22 mM EDTA, 100 mM NaCl; pH 8.0 and 1/10 volume H20. Reactions were incubated for a further 1.5 h. at 37 TC. DNA was then purified using the proteinase K/SDS treatment, phenol/chloroform/isoamyl alcohol extraction and ethanol precipitation procedures described above. After resuspension, samples were resolved on a 1% agarose in lx TBE gel. The ratio of protein to DNA that produced optimal supercoiling was identified, and used for the large-scale reaction.

Conditions in the large-scale reaction were identical to those in the test reactions, with adjustment of the binding buffer to a final concentration of 350 mM potassium glutamate for induction of negative supercoils. Unrestrained compensatory supercoiling was removed by addition of 10 units of CBE and adjustment of the buffer with 1/10 volume of the adjustment buffer described above, and 1/10 volume of either H20 (for positive supercoils) or 1/10 volume of 1.4 M potassium glutamate (for negative supercoils).

Reactions were purified as before.

Acid phenol extraction to remove nickedplasmid DNA. To remove nicked and linear plasmid species from the supercoiling preparations, a modified version of the method of Zasloff et al. (19) was employed. After resuspending the newly supercoiled plasmid in 1 mM Tris, 1 mM EDTA; pH 8.0, the preparation was desalted by passage over a NAP-25 column. The DNA solution (<0.5 ýpg/p.L) was adjusted to 75 mM NaCl and 50 mM sodium acetate (pH 4.0). Molecular biology grade phenol, equilibrated with

50 mM sodium acetate (pH 4.0), was added in equal volume to the salt- and pH-adjusted

DNA solution, and the resulting extraction mixtures were vortexed for 5 min. The extractions were centrifuged at 6,000 x g for 5 min, and the aqueous layer was immediately removed and pH-neutralized by adjustment to 200 mM Tris (pH 8.0).

Residual phenol was removed by choloroform extraction, and plasmid was precipitated with one volume of isopropanol, followed by resuspension in TE buffer.

Analysis ofplasmid topology and topoisomer content in supercoiled substrates.

After purification, topoisomer content and direction of supercoiling in the substrates was determined by 1- and 2-dimensional gel electrophoresis as described elsewhere (20). The fraction of nicked molecules was determined by gel electrophoresis in the presence of ethidium bromide.

RESULTS AND DISCUSSION

This method for introducing defined states of supercoiling into covalently closed circular DNA exploits two phenomena. The first is the constraint of toroidal supercoils by tetramers of the archaeal histone rHMfB bound to DNA (20). The addition of rHMfB to closed circular DNA molecule introduces constrained supercoiling of one sign that must be balanced by induction of an equivalent number of supercoils of the opposite sign in the unconstrained DNA. The unconstrained compensatory supercoils can be removed using a DNA topoisomerase, so that the toroidal supercoils constrained by rHMfB are released as active supercoiling in the closed circular DNA molecules upon removal of the rHMfB protein molecules (Fig. 2.1). This lab previously made use of this behavior to develop a method for introducing positive supercoiling into plasmid DNA (15).

I have now exploited a second phenomenon to develop a method to introduce either positive or negative supercoiling into any closed circular DNA molecule. With rHMfB, the handedness of the toroids bound by the histone complex is modulated by the salt composition of the buffer. At low salt concentrations, rHMfB constrains positive toroids, while higher concentrations (>300 mM K) cause the toroids to reverse in direction around the rHMfB clusters to constrain negative supercoiling (16). This change has been attributed to alterations in tetramerization of the protein at the histone fold that can occur under conditions of high ionic strength (21). Practically speaking, the dependence of the direction of supercoiling on ionic strength provides a convenient way to control the sign of supercoiling introduced when generating supercoiled molecules.

The ability of rHMfB to bind DNA with high affinity in either circumstance means that highly supercoiled plasmid of either sign can be readily produced.

A central feature of this method for preparing supercoiled DNA substrates is the strict dependence on the use of potassium glutamate as the salt species for controlling the sign of the supercoiling. We tested several different salt species for efficacy in this protocol. Of those tested, only potassium glutamate resulted in consistently highly negatively supercoiled molecules. It is thus recommended for use in preparing negatively supercoiled substrates. The lack of production of high degrees of negative supercoiling with other salt species was assumed to be primarily due to inhibition of topoisomerase activity by the salts in question. To test this hypothesis, we reacted negatively supercoiled plasmid with increasing amounts of CBE in the absence of rHMfB, using buffer conditions identical to those present during the relaxation of compensatory supercoils. Some degree of topoisomerase inhibition exists even without a salt supplement; this is expected, due to the change of buffer conditions from those which are ideal for topoisomerase activity to those required for optimal rHMfB binding. Without exception, the salts that did not produce negatively supercoiled plasmid showed a near complete inhibition of relaxation of supercoiled plasmid by CBE, even at 10 U CBE / 1

[rg DNA (data not shown).

There are several parameters critical to achieving optimal DNA supercoiling.

One important consideration is that there is an optimal ratio of a protein to DNA that provides maximal supercoiling of either sign (Fig. 2.2a). At low relative protein concentrations (<0.8 mass units protein per mass unit of DNA), there are fewer supercoils constrained by the protein and, as a result, the final product will have a lower average superhelical density. This importance of this issue is illustrated by the observation at low salt concentrations that favor positive supercoiling that low levels of negative supercoiling are actually introduced, rather than the expected positive (20). In the case of excessive quantities of protein relative to DNA, there is again a reduction in the average superhelical density introduced into the DNA, presumably as a result of interference with access of topoisomerases to DNA and thus the ability of the topoisomerase to relax the unconstrained compensatory supercoiling (20,22). When this optimal ratio is used, large

(>500 [tg) amounts of supercoiled substrate can be produced with a high degree of superhelical density, with >50% of the DNA molecules in having o-values of 10.03-0.061

(Fig. 2.2b-c).

To achieve this optimal ratio, which is not a threshold ratio, we routinely perform small-scale (1 [tg of DNA) test reactions by varying the input quantity of rHMfB.

Occasionally, we have encountered circumstances in which the optimal rHMfB:DNA ratio does not produce maximal supercoiling. In these cases, we have generally found that further purification of the DNA or an increase in the amount of CBE used is sufficient to overcome the problem.

A second important parameter involves the amount of CBE for optimal supercoiling. While a single unit of CBE is sufficient to completely relax 1 [tg of negatively supercoiled plasmid DNA under standard reaction conditions, significantly more (-10-fold) is necessary to fully relax the compensatory supercoiling resulting from rHMfB binding. We attribute this to a combination of factors, including a partial inhibition of topoisomerase activity under rHMfB binding conditions and a limitation on the accessibility of the unconstrained DNA to topoisomerase molecules. It is especially important to prepare the CBE within weeks of the supercoiling procedure to avoid loss of activity with prolonged storage (months) at -80 oC. Under no circumstances should the

CBE be subjected to freeze-thaw conditions.

While this method for preparing DNA with positive and negative supercoiling generally results in a population of plasmid molecules with less than 10% nicked form, we have found that the application of an acid phenol extraction step (19) allows the preparation of plasmid samples with less than 5% nicked species. We have modified the acid phenol method of Zasloff et al. by limiting the length of time that the DNA is exposed to the acidic conditions that can induce depurination. As shown in Figure 2.3, 5 minutes of acid phenol extraction reduce depurination to undetectable levels. This technique is effective regardless of the superhelical state of the closed-circular molecules

(relaxed, positively or negatively supercoiled), and one extraction step is generally sufficient to reduce the total fraction of nicked plasmid molecules in the preparation to

<5%.

CONCLUSIONS

We have described a new method for the preparation of large quantities of DNA substrates containing high levels of either positive or negative supercoiling. The use of the archaeal histone rHMfB in such a protocol is advantageous, due to its relatively easy preparation, and ability to reverse binding orientations of DNA in different salt conditions. Substrates can be prepared via this method that are both highly supercoiled and free from non-closed circular DNA contaminants, making it ideal for preparation of substrates for biochemical or biophysical studies. A B C +

+ 1+ I/

~ WJ

Figure 2.1. Reaction scheme for preparation of supercoiled plasmid substrates. A) Relaxed, closed-circular plasmid is bound with rHMfB in the absence (top) or presence (bottom) of 350 mM potassium glutamate. Toroidal supercoils of one direction are constrained, while compensatory supercoiling of the opposite sign is unconstrained. B) Topoisomerase- containing CBE is added without (top) or with (bottom) an additional salt supplement, removing unconstrained supercoils, while leaving the protein bound supercoils are intact. C) Bound proteins are removed from the DNA, allowing induced positive (top) or negative (bottom) supercoils to diffuse throughout the molecules. rHMfB

B C 2in~d omns-ro - - 2d2 cnseorEW -*

Figure 2.2. Optimization and analysis of prepared supercoiled substrates. A) Small-scale optimization of rHMfB:DNA ratio. At lower levels of rHMfB, low levels of binding occur, resulting in products with low levels of supercoiling. Above the optimum ratio (denoted with an *) rHMfB binding interferes with topoisomerase relaxation of compensatory supercoils, and supercoiling in the product decreases. B) 1- and 2-dimensional analysis of a positively supercoiled plasmid. In the first dimension, no intercalator is present, and plasmids resolve on the basis of superhelical density; more supercoiled plasmids travel farther on the gel. In the second dimension, a low level of intercalator is added, which induces a low level of positive writhe in the plasmids. While this allows positively supercoiled molecules to travel farther, negatively supercoiled molecules will have supercoils cancelled out, and thus have movement retarded. C) 1- and 2- dimensional analysis of negatively supercoiled plasmid, generated in the presence of potassium glutamate. pre- post- A extraction extraction

OC

CC

untreated putrescine D

OC

CC

Figure 2.3. Purification of closed-circular plasmid preparation by extraction with acidic phenol. A) Damaged plasmid sample before and after acidic phenol extraction (OC = open circular, CC = closed circular). Extraction reduces the open circular population to less than 5% of the total plasmid. B) Treatment of acidic phenol-extracted plasmid with putrescine. Incubating extracted plasmid with 0.1 M putrescine (pH 7.0) for one hour at 37 'C does not introduce any appreciable nicks to the plasmid. Since putrescine converts abasic sites to strand breaks, this confirms that no significant depurination is occurring during our acidic phenol extraction. LITERATURE CITATIONS

1. Wu, H.Y., Shyy, S.H., Wang, J.C. and Liu, L.F. (1988) Transcription generates positively and negatively supercoiled domains in the template. Cell, 53, 433-440.

2. Sinden, R.R., Carlson, J.O. and Pettijohn, D.E. (1980) Torsional tension in the DNA double helix measured with trimethylpsoralen in living E. coli cells: analogous measurements in insect and human cells. Cell, 21, 773-783.

3. Ljungman, M. and Hanawalt, P.C. (1992) Localized torsional tension in the DNA of human cells. Proc Natl Acad Sci US A, 89, 6055-6059.

4. Kramer, P.R. and Sinden, R.R. (1997) Measurement of unrestrained negative supercoiling and topological domain size in living human cells. Biochemistry, 36, 3151-3158.

5. Lim, H.M., Lewis, D.E., Lee, H.J., Liu, M. and Adhya, S. (2003) Effect of varying the supercoiling of DNA on transcription and its regulation. Biochemistry, 42, 10718-10725.

6. Funnell, B.E., Baker, T.A. and Kornberg, A. (1986) Complete enzymatic replication of plasmids containing the origin of the Escherichia coli chromosome. JBiol Chem, 261, 5616-5624.

7. Wang, J.C. (1974) Interactions between twisted DNAs and enzymes: the effects of superhelical turns. JMol Biol, 87, 797-816.

8. Bae, S.H., Yun, S.H., Sun, D., Lim, H.M. and Choi, B.S. (2006) Structural and dynamic basis of a supercoiling-responsive DNA element. Nucleic Acids Res, 34, 254-261.

9. Bernstein, L.B., Mount, S.M. and Weiner, A.M. (1983) Pseudogenes for human small nuclear RNA U3 appear to arise by integration of self-primed reverse transcripts of the RNA into new chromosomal sites. Cell, 32, 461-472.

10. Levchenko, V., Jackson, B. and Jackson, V. (2005) Histone release during transcription: displacement of the two H2A-H2B dimers in the nucleosome is dependent on different levels of transcription-induced positive stress. Biochemistry, 44, 5357-5372.

11. Lockshon, D. and Morris, D.R. (1983) Positively supercoiled plasmid DNA is produced by treatment of Escherichia coli with DNA gyrase inhibitors. Nucleic Acids Res, 11, 2999-3017.

12. Gellert, M., Mizuuchi, K., O'Dea, M.H. and Nash, H.A. (1976) DNA gyrase: an enzyme that introduces superhelical turns into DNA. Proc Natl Acad Sci USA, 73, 3872-3876. 13. Forterre, P., Mirambeau, G., Jaxel, C., Nadal, M. and Duguet, M. (1985) High positive supercoiling in vitro catalyzed by an ATP and polyethylene glycol- stimulated topoisomerase from Sulfolobus acidocaldarius. Embo J, 4, 2123-2128.

14. Richardson, S.M., Boles, T.C. and Cozzarelli, N.R. (1988) The helical repeat of underwound DNA in solution. Nucleic Acids Res, 16, 6607-6616.

15. LaMarr, W.A., Sandman, K.M., Reeve, J.N. and Dedon, P.C. (1997) Large scale preparation of positively supercoiled DNA using the archaeal histone HMf. Nucleic Acids Res, 25, 1660-1661.

16. Musgrave, D., Forterre, P. and Slesarev, A. (2000) Negative constrained DNA supercoiling in archaeal nucleosomes. Mol Microbiol, 35, 341-349.

17. Camerini-Otero, R.D. and Felsenfeld, G. (1977) Supercoiling energy and nucleosome formation: the role of the arginine-rich histone kernel. Nucleic Acids Res, 4, 1159-1181.

18. Starich, M.R., Sandman, K., Reeve, J.N. and Summers, M.F. (1996) NMR structure of HMfB from the hyperthermophile, Methanothermus fervidus, confirms that this archaeal protein is a histone. J Mol Biol, 255, 187-203.

19. Zasloff, M., Ginder, G.D. and Felsenfeld, G. (1978) A new method for the purification and identification of covalently closed circular DNA molcules. Nucleic Acids Res, 5, 1139-1152.

20. Musgrave, D.R., Sandman, K.M. and Reeve, J.N. (1991) DNA binding by the archaeal histone HMf results in positive supercoiling. Proc Natl Acad Sci US A, 88, 10397-10401.

21. Marc, F., Sandman, K., Lurz, R. and Reeve, J.N. (2002) Archaeal histone tetramerization determines DNA affinity and the direction of DNA supercoiling. J Biol Chem, 277, 30879-30886.

22. Broyles, S.S. and Pettijohn, D.E. (1986) Interaction of the Escherichia coli HU protein with DNA. Evidence for formation of nucleosome-like structures with altered DNA helical pitch. JMol Biol, 187, 47-60.

Chapter 3 - Reactivity of Supercoiled DNA with Chemical and

Enzymatic Probes of DNA Structure

INTRODUCTION

The studies presented in this chapter address biochemical approaches to defining the structure of positively supercoiled DNA. As discussed in detail in Chapter 1, DNA supercoiling is ubiquitous in both prokaryotic and eukaryotic cells (1-4), and has been implicated in a number of important physiological processes, including replication, transcription, and recombination (5-7). Understanding the structural changes that occur with the introduction of supercoiling is key to elucidating the mechanisms by which these physiological effects are exerted. The twin domain model of supercoiling predicts the presence of both positively and negatively supercoiled DNA in cells (8), but the ready availability of negatively supercoiled substrate has allowed for a more thorough analysis than for the positive form.

Negative supercoiling has been shown to produce a number of changes in the structure of DNA, including localized denaturation and sequence-dependent formation of alternative secondary structures such as cruciforms, Z-DNA, and triplex DNA (9-12).

However, structural changes that result from positive supercoiling have not been well defined. Previous studies have suggested that, under certain positive supercoiling conditions, the bases have an increased solvent exposure (13,14). It has been shown, for example, that introduction of very high levels of twist using a micromanipulator causes the DNA to adopt a structure in which a portion of the bases become extrahelical (15). Findings are limited, however, both in scope and by the methods used to generate positive supercoiling. Additional studies, with conditions more closely resembling those in cells, are necessary to better understand the effects of positive supercoiling on the structure of DNA.

Preliminary studies from this laboratory indicate that the methylating agent dimethyl sulfate (DMS) is more reactive with positively than negatively supercoiled plasmid at both the N 7 position of guanine and the N3 position of adenine (16,17). The 3- methyladenine result is particularly revealing, since in B-form DNA the N3 position is buried in the minor groove. When the strands of DNA are more tightly intertwined, as they are in positive supercoiling, the N3 position could be protected from dimethyl sulfate alkylation. An increase in methylation at this site could indicate an extrahelical positioning of at least some of the bases in positively supercoiled DNA.

A number of agents have been identified as probes to elucidate non-canonical

DNA structures. These include single-strand specific nucleases and chemical agents that react specifically with unpaired bases. S1 nuclease and nuclease BAL-31 are both capable of cleaving duplex DNA at sites of local melting, including perturbations in base pairing occurring at a single site (18-20). In addition, the alkylating agent chloroacetaldehyde has been used to identify non-canonical DNA structures such as the loop region of cruciforms (21) and the junctions between B- and Z-form DNA (22).

Using these agents along with carefully prepared supercoiled substrates, we have investigated the effects of positive supercoiling on the structure of DNA. Most noteworthy, positively supercoiled plasmid is not sensitive to cleavage by single stranded nucleases, in contrast to the very sensitive nature of negatively supercoiled DNA. Unfortunately, chloroacetaldehyde results were inconclusive, suggesting that this agent

may be better suited for identifying specific structural phenomena occurring in the DNA.

MATERIALS AND METHODS

Preparationofsupercoiledplasmid substrates. Positively supercoiled pUC18

substrates were prepared as described in Chapter 2. Relaxed and negatively supercoiled

substrates were prepared in the same manner as the positive, substituting heat-inactivated

chicken blood extract (CBE) for live enzyme at the appropriate steps to obtain substrates

with the desired presence and direction of supercoiling. All substrates were dialyzed

once in 2 L of TE buffer with 3 M NaC1, using 100,000 MWCO dialysis tubing, followed

by four changes of 2 L of TE buffer, prior to use in chemical or enzymatic experiments.

Correctionfactors for quantifying supercoiledsubstrates using ethidium fluorescence. To establish correction factors to account for supercoiling-dependent

differences in the binding of ethidium bromide to the plasmid DNA, a gel assay utilizing

both linearized and supercoiled plasmid was performed. For these, a solution of 25 [M

DNA in lx NEbuffer #2 (50 mM NaCl, 10 mM Tris-HC1, 10 mM MgCl 2, 1 mM DTT,

pH 7.9, 100 tig/mL BSA) was prepared, and split in half. To one half, 50 U/mL of

BamHI (New England Biolabs, Ipswich, MA) was added, to the other, an equal volume

of water; this was followed by reaction for 1 hr at 37 °C. Aliquots of 10 [L were

removed from both the linearized and untreated samples, and mixed with 5 tL of

nuclease stop solution. Samples of each intact form were carefully and completely

loaded in adjacent lanes to each of the linearized samples on a 1% agarose in lx TBE gel containing 0.5 Vig/mL ethidium bromide. Electrophoresis was then performed at 4.5 V/cm for 90 min. Fluorescence from each band was quantified on an Alpha Innotech

Fluorchem 8900 Imager (San Leandro, CA), and the amount of fluorescence of the intact form of each plasmid relative to the linearized form was determined.

Nuclease cleavage assays. Reactivity of plasmid DNA with S1 nuclease

(Fermentas, Hanover, MD) was assayed both kinetically and as a dose response. For the kinetics experiments, 25 [M DNA was reacted with 100 U/mL S nuclease at 37 'C in lx S1 buffer (40 mM sodium acetate, pH 4.5, 300 mM NaCI, 2mM ZnSO 4). At defined time points, 10 [tL aliquots were removed and mixed with 10 [tL of nuclease stop solution (100 mM EDTA, 100 mM Tris-HC1, pH 8.0, 40% sucrose, 0.1% bromophenol blue) to quench the reaction. Samples were then resolved and quantified on an ethidium bromide-containing agarose gel. For the dose response, 175 ng of plasmid was reacted with the nuclease in amounts ranging from 0 to 200 U for 1 hr at 37 "C in lx S1 buffer in a total volume of 20 [tL. Cleavage was stopped and DNA was quantified as in the kinetics experiments. To examine the effects of low salt on the reactivity with supercoiled plasmid, the dose response experiments were repeated with 0-5 U S1 nuclease in lx low salt S1 buffer (40 mM sodium acetate, pH 4.5, 50 mM NaCi, 2 mM

ZnSO 4).

To measure reactivity of supercoiled plasmid with nuclease BAL-31 (New

England Biolabs, Ipswich, MA), the enzyme was reacted at a concentration of 12.5 U/mL with 25 [tM plasmid at 30 OC in lx BAL-31 reaction buffer (600 mM NaCI, 12 mM

CaC12, 12 mM MgC12, 20 mM Tris-HC1, 1 mM EDTA, pH 8.0). Reaction was initiated by addition of the nuclease, and at defined time points 10 [tL aliquots were removed and the cleavage reaction stopped by addition of 10 1tL of phenol/chloroform/isoamyl alcohol. After extraction of the aqueous phase, samples were precipitated with an equal volume isopropyl alcohol, resuspended in TE, and resolved on an ethidium bromide- containing agarose gel. Relative populations of the different forms of DNA were

quantified using ethidium bromide fluorescence as described above.

Assaying reactivity of chloroacetaldehyde with plasmid substrates. Formation of

etheno adducts on the DNA bases was assayed by reacting 2 ýtg plasmid DNA with

chloroacetaldehyde (Aldrich, St. Louis, MO) at concentrations ranging from 0 to 4.5 mM

in a total volume of 40 [tL of TE buffer for 16 hr at 370 C. Reactions were quenched and

chloroacetaldehyde was removed by precipitation with 2.5 volumes of 100% ethanol,

followed by a 70% ethanol rinse. DNA was resuspended in ddH20, and an aliquot was

removed for spectrophotometric quantitation.

The remainder of the DNA was spiked with internal standards consisting of 100

fmol each of 15N-labeled 1,N6-ethenoadenine (EA), and 3,N4-ethenocytosine (EC)

nucleoside, and digested to deoxynucleoside monophosphates by adjustment to a volume

of 200 tL with sodium acetate buffer (30 mM, pH 5.6, 0.2 mM zinc chloride) and

treatment with 4 units of nuclease P1 at 37 TC for 3 hr. Further digestion to nucleosides

was accomplished by addition of 200 [tL of 30 mM sodium acetate (pH 8.1), 20 units of

alkaline phosphatase, and 1 unit of phosphodiesterase, followed by incubation at 370 C for

6 hr. Enzymes were removed by passage of the reaction mixture over a YM- 10

centrifugal concentrator (Millipore, Billerica, MA), and adducts were isolated and

concentrated using an Agilent 1100 HPLC (Santa Clara, CA). Analysis was performed on an Agilent model 1100 HPLC equipped with a 1040A diode array detector coupled to an API 3000 tandem mass spectrometer (Applied

Biosystems, Foster City, CA). All reversed-phase HPLC was carried out on a

Phenomenex C18 column that was eluted isocratically with 98:2 permanganate-treated

H20 (0.1% acetic acid):acetonitrile (0.1% acetic acid). The mass spectrometer was operated in positive ion mode; the first quadropole was set to transmit the precursor ions

MH+ at m/z 255 and 281 for the 15N-EC and eA internal standards, and at m/z 252 and

276 for the EC and EA analytes. The product ions were monitored in the third quadropole at m/z 139 and 165 for the 15N-labeled internal standards, and 136 and 160 for the analytes.

RESULTS

Correctionfactors for supercoiledplasmid. The binding of ethidium to DNA causes a localized unwinding at the site of intercalation. For linear DNA, the compensatory overwinding in the remaining DNA that occurs as a result of ethidium binding can diffuse out of the ends of the molecule, so the overall binding is limited only by the number of available binding sites on the DNA. A covalently closed-circular molecule presents a different situation. In this case, the global overwinding that compensates for the localized unwinding must remain in the molecule, the energetics of which effectively translate into a positive writhe. There is a maximal amount of positive writhe that can be introduced into a closed-circular molecule, which limits the binding of ethidium to the plasmid molecule (23). As a result, not only will closed-circular DNA bind ethidium differently than an equivalent linear molecule, but the resultant supercoiled

states will also vary. Positively supercoiled plasmid will be able to bind the least amount

of ethidium before reaching a point where the molecule cannot accommodate any more

introduced supercoils. Negatively supercoiled plasmid will bind considerably more

ethidium, although not as much as an equivalent linear molecule; relaxed will bind an

intermediate amount.

Since measurements based on ethidium fluorescence are dependent on the amount

of bound ethidium, it is crucial to account for this differential binding to obtain accurate

quantification. To ensure that measurements were accurate for quantifying the differently

supercoiled molecules, ethidium fluorescence of equal amounts (determined

spectrophotometrically) of the three linearized forms of plasmid on a gel were compared

at saturating ethidium concentrations. The quantified bands were virtually identical, with

variations small enough to be accounted for by pipetting errors. Having established that

A260 readings were sufficient to quantify DNA and that the same amount of plasmid was

present in each sample, fluorescence of the relaxed, positively and negatively supercoiled

plasmid were compared. As expected, negatively supercoiled plasmid was the most

fluorescent, followed by relaxed, and then positively supercoiled DNA. The relative

fluorescence (compared to negatively supercoiled plasmid) of relaxed and positively

supercoiled samples was 0.67 and 0.5, respectively. Previous studies determined that

linearized pUC 19 is 1.4 times more fluorescent than the natively negatively supercoiled molecule (24). This is in general agreement with our results, defining the correction

factors for closed circular molecules as follows: negatively supercoiled plasmid: 1.40; relaxed plasmid: 2.09; positively supercoiled plasmid 2.54. Positively supercoiledplasmidis not cleaved at an increasedrate by single strand specific nucleases. S nuclease is a glycoprotein from Aspergillus oryzae that has both exo- and endonuclease activity. It exhibits endonuclease activity at regions of single- stranded DNA, to the level of a single unpaired base (25). In all conditions tested, negatively supercoiled plasmid was highly reactive with S1 nuclease, presumably due to the underwinding and associated transient denaturation of the DNA strands (Figs. 3.1-3).

This is consistent with published reports (20,25,26). In the kinetics experiment, both relaxed and positively supercoiled plasmid exhibited significantly less reactivity than negative, with relaxed plasmid reacting slightly more efficiently than the positively supercoiled form (Fig. 3.1). In addition, the reactivity of relaxed and positive is similar in the dose response experiments (Fig. 3.2), which may possibly result from cooperative enzyme activity at higher concentrations rather than any single-stranded character of the

DNA. Reducing the ionic strength increased the reactivity of all three substrates considerably (data not shown), consistent with previous studies that found higher levels of cleavage at low ionic strength (20). While the overall reactivity of all topological forms of DNA increased with lower ionic strength, no differences between positively supercoiled and relaxed plasmid emerged. Lowering the ionic strength reduces the shielding of the highly charged phosphate backbone by counterions, which prevents the segments in the helix from coming into as close proximity as occurs at higher ionic strength. As a result, the writhe component in the molecule is reduced and as the twist must increase to compensate (27). The lack of a difference in reactivity between relaxed and positive substrates at lower ionic strength suggests that even though the writhe is decreasing at lower salt, the resulting increase in twist is not responsible for the increased reactivity, as the relaxed molecules show an increase in cleavage equivalent to that seen in positive.

Nuclease BAL-31 from Alteromonas espejianaalso possesses both exo- and endonuclease activity, but it is active at a pH that is much closer to neutrality than S (pH

8.0 vs. pH 4.5) (19). Again, the negatively supercoiled plasmid was the most reactive, with relaxed and positively supercoiled plasmid being cleaved at significantly slower rates (Fig. 3.3). At this low enzyme concentration, the relaxed plasmid was cleaved slightly more efficiently than positive, much as it was in the S1 nuclease kinetics experiments.

Neither S1 nor BAL-31 showed an increase in reactivity with positively supercoiled plasmid, relative to the relaxed form. In fact, at low concentrations of either enzyme, such as were used in the time course studies, positively supercoiled substrates were cleaved at a slower rate than the relaxed form. While these findings are not consistent with extrahelical bases in positively supercoiled substrates, it is possible that structural changes induced by positive supercoiling interfere with the binding or activity of these enzymes. Although both enzymes have been shown to cleave at sites of unpaired bases (13,25), it is not clear exactly what the enzymes recognize in terms of DNA structure. A structure in which extrahelical bases are present in positively supercoiled

DNA may be sufficiently different from the structure being recognized in negatively supercoiled plasmid that it would not be cleaved by either enzyme, particularly if the backbone was shielded at the center of the helix.

Chloroacetaldehyde is not an effective probe for transient single-strandedregions in DNA. Chloroacetaldehyde is a vinyl chloride metabolite that reacts with the hydrogen- bonding region of bases to form four different exocyclic etheno adducts (Fig. 3.4)

(28,29). The formation of both 1,N6-ethenodeoxyadenosine and 3,N4-ethenocytosine were analyzed using LC-MS (Fig. 3.5). While there appears to be a slight increase in the quantities of both adducts in the negatively supercoiled plasmid, this effect is not statistically significant and it is not as pronounced as would be expected based on the single-strand nuclease data. This may be due to the multi-step reaction mechanism proposed for the formation of the etheno adducts. The details of the reaction are still in dispute, but a generally accepted scenario involves attack of the chloroacetaldehyde at the exocyclic amino group on the base to form an unstable intermediate, which then reacts with a heterocyclic nitrogen to form a cyclic adduct that finally undergoes a dehydration step to form an etheno adduct (30,31). Even if the molecule were able to undergo an initial attack, the nature of the solvent exposure of the base could be incompatible with subsequent cyclization and dehydration steps. Because this entire process is necessary to form the analytes detected here, chloroacetaldehyde may not be an ideal agent to measure base solvent exposure.

Chloroacetaldehyde has been used to identify sites of single-stranded character in several instances (21,22), but this is generally as a qualitative agent. Typically, chloroacetaldehyde-reacted DNA is linearized, and then treated with S1 nuclease to express the adducted base as a strand break (21), although Maxam-Gilbert sequencing methods are used, as well (22). Such methods are useful for identifying specific sites that are susceptible to damage, such as the loop region of cruciforms, but the presence of multiple adduction sites throughout the DNA would complicate interpretation and prevent a good quantitative assessment of base unpairing. Interestingly, the most informative aspect of chloroacetaldehyde reactivity may be in the partitioning between the two guanine adducts. After the initial attack on the exocyclic nitrogen, the adduct can

cyclize with either of two ring nitrogens, the N' and the N3 . Increased hydrogen bonding

will interfere with cyclization at the N' position, shifting the reaction to favor cyclization

at the N3 (32). Measuring changes in the ratios between these two adducts may provide

key information about the hydrogen bonding state of guanine in supercoiled DNA.

It is worth noting that a higher dose of chloroacetaldehyde was tested than is

shown in Figure 3.5 (data not shown). In all cases, the dose-response deviated from

linearity at this higher dose, showing a substantial increase in the rate of formation of

adducts. This could be due to a "seeding" effect due to the inability of an etheno-

adducted base to return to a normal base-pairing geometry. This perturbation may serve

to destabilize neighboring base pairs, making those bases more solvent exposed and thus

more susceptible to chloroacetaldehyde attack. Mismatches involving canonical base

pairs are known to destabilize the helix (33,34); so it is reasonable to expect bulky etheno

adducts to have a similar, if not more pronounced, effect.

DISCUSSION

Biomechanical and biochemical experiments have suggested that sufficient levels

of positive supercoiling can result in DNA structures in which the bases are positioned

extrahelically. We have investigated the possibility that such structures may occur at physiologically relevant levels of positively supercoiling by reacting supercoiled and relaxed plasmid substrates with enzymatic and chemical agents that are selective for unpaired bases and single stranded structure.

No evidence was found in these studies for extensive or obvious solvent exposure of bases induced by the levels of positive DNA supercoiling that are introduced with our method (c = +0.04). There are several possible explanations for this result. The simplest among these is the absence of solvent exposed bases under physiologically relevant conditions of torsional tension, at least at the superhelical density of the substrates utilized in these assays (c = +0.04). It could be that the threshold for transition to an extrahelical base structure occurs at superhelical density values that are too high for any extrahelical bases to be detected in these substrates. It is also possible that the methods of analysis were insufficient to detect extrahelical bases that were nonetheless present transiently, perhaps as a result of fast kinetics of base flipping. Recognition of overwinding-induced structural changes by single-strand specific nucleases has not been established, so the possibility of such changes being present cannot be ruled out.

Chloroacetaldehyde was not effective in identifying increased reactivity with negative supercoiling, where solvent exposed bases are known to occur, so the lack of increased reactivity of bases in positively supercoiled plasmid is not conclusive in these studies.

Further study with additional chemical probes may clarify what structural changes occur when DNA is subjected to overwinding. Repeating the DMS studies that were initially performed in this laboratory (16,17) with our improved substrate, and expanding them to include relaxed as well as supercoiled plasmid could be informative. A number of other chemical agents are well established as probes of stably unpaired structure, including potassium permanganate (35) and diethyl pyrocarbonate (36). Comparing reactivity of these agents with relaxed and supercoiled plasmids should help broaden the picture of structural changes introduced with positive supercoiling.

Still, the structural changes that occur when DNA is positively supercoiled may be too subtle at the level of supercoiling present in these studies to be detected using chemical or enzymatic means. For this reason, complementary approaches should be considered that would allow for both detection of subtle structural changes, as well as introduction of precise and relatively high levels of twist into the DNA. In the following chapters, these approaches are addressed. Raman spectroscopy is utilized to examine the subtle structural changes occurring in supercoiled plasmid substrates, giving detailed information about precisely which bonds are being affected. Preliminary substrate design and considerations for fluorescence-based micromanipulation experiments are then discussed, providing a foundation for experiments that will allow for elucidation of structural changes as a function of introduced twist. 75 * positive * negative A relaxed

50

25

ýW ' .'•I1m, 7

I

20 40 60 Time (minutes)

Figure 3.1. Kinetics of reaction of S 1 nuclease with supercoiled and relaxed substrates. Negatively supercoiled plasmid is cleaved by the single-strand specific S nuclease at a much faster rate than either relaxed or positively supercoiled plasmid. Plasmid substrates were reacted with S nuclease as described in Materials and Methods. Data points shown are the average of four replicates; error bars represent the standard error. 100 A-~~-~- -- I'1~1- M

75

* positive * negative A relaxed Eb- 50 100 150 200 U Si nuclease

Figure 3.2. Reaction of S1 nuclease with supercoiled and relaxed substrates. Negatively supercoiled plasmid is cleaved by the single-strand specific S1 nuclease considerably more than either relaxed or positively supercoiled plasmid. Plasmid substrates were reacted with S1 nuclease as described in Materials and Methods. Data points shown are the average of four replicates; error bars represent the standard error. 4 tf% IUV - C

, 75 L,.

Uc 50 0U Spositive Snegative Arelaxed C" 25

04 1 0 10 20 30 Time (minutes)

Figure 3.3. Kinetics of reaction of nuclease BAL-31 with supercoiled and relaxed substrates. Negatively supercoiled plasmid is cleaved by the single-strand specific nuclease BAL-31 at a much faster rate than either relaxed or positively supercoiled plasmid. Plasmid substrates were reacted with nuclease BAL-31 as described in Materials and Methods. Data points shown are the average of at least four replicates; error bars represent the standard error of the data. NH2 N NNI N

NH2

N N 'N"- N O

0 CI

Chloroacetaldehyde

O

N NH O N NH2 N NH N N N

Figure 3.4. Reaction of chloroacetaldehyde with DNA bases. Chloroacetaldehyde reacts with adenine, cytosine, and guanine to form 1, N6-ethenoadenine, 3,N4-ethenocytosine, 1,N2- ethenoguanine and N2,3-ethenoguanine. 12 * positive negative A relaxed

ii 4

ol k0 0 0.5 mM chloroacetaldehyde

.1 ·-- * positive 9 negative A relaxed

Al rIlL ~------i I · 9 I 0 0.5 1 mM chloroacetaidehyde Figure 3.5. Reaction of chloroacetaldehyde with supercoiled and relaxed substrates. No significant effect is seen on the formation of etheno adducts as a function of supercoiling. Plasmid substrates were treated with chloroacetaldehyde as described in Materials and Methods. Data points shown are the average of two replicates; error bars represent the standard error. LITERATURE CITATIONS

1. Wu, H.Y., Shyy, S.H., Wang, J.C. and Liu, L.F. (1988) Transcription generates positively and negatively supercoiled domains in the template. Cell, 53, 433-440.

2. Figueroa, N. and Bossi, L. (1988) Transcription induces gyration of the DNA template in Escherichia coli. Proc Natl Acad Sci US A, 85, 9416-9420.

3. Brill, S.J. and Sternglanz, R. (1988) Transcription-dependent DNA supercoiling in yeast DNA topoisomerase mutants. Cell, 54, 403-411.

4. Giaever, G.N. and Wang, J.C. (1988) Supercoiling of intracellular DNA can occur in eukaryotic cells. Cell, 55, 849-856.

5. Funnell, B.E., Baker, T.A. and Kornberg, A. (1986) Complete enzymatic replication of plasmids containing the origin of the Escherichia coli chromosome. JBiol Chem, 261, 5616-5624.

6. Steck, T.R., Franco, R.J., Wang, J.Y. and Drlica, K. (1993) Topoisomerase mutations affect the relative abundance of many Escherichia coli proteins. Mol Microbiol, 10, 473-481.

7. Droge, P. (1993) Transcription-driven site-specific DNA recombination in vitro. Proc Natl Acad Sci US A, 90, 2759-2763.

8. Liu, L.F. and Wang, J.C. (1987) Supercoiling of the DNA template during transcription. Proc Natl Acad Sci USA, 84, 7024-7027.

9. Kowalski, D., Natale, D.A. and Eddy, M.J. (1988) Stable DNA unwinding, not "breathing," accounts for single-strand-specific nuclease hypersensitivity of specific A+T-rich sequences. Proc Natl Acad Sci US A, 85, 9464-9468.

10. Lilley, D.M. (1981) Hairpin-loop formation by inverted repeats in supercoiled DNA is a local and transmissible property. Nucleic Acids Res, 9, 1271-1289.

11. Peck, L.J., Nordheim, A., Rich, A. and Wang, J.C. (1982) Flipping of cloned d(pCpG)n.d(pCpG)n DNA sequences from right- to left-handed helical structure by salt, Co(III), or negative supercoiling. Proc Natl Acad Sci US A, 79, 4560- 4564.

12. Beltran, R., Martinez-Balbas, A., Bernues, J., Bowater, R. and Azorin, F. (1993) Characterization of the zinc-induced structural transition to *H-DNA at a d(GA.CT)22 sequence. JMol Biol, 230, 966-978.

13. Lau, P.P. and Gray, H.B., Jr. (1979) Extracellular nucleases of Alteromonas espejiana BAL 31.IV. The single strand-specific deoxyriboendonuclease activity as a probe for regions of altered secondary structure in negatively and positively supercoiled closed circular DNA. Nucleic Acids Res, 6, 331-357.

14. McClellan, J.A. and Lilley, D.M. (1991) Structural alteration in alternating adenine-thymine sequences in positively supercoiled DNA. J Mol Biol, 219, 145- 149.

15. Allemand, J.F., Bensimon, D., Lavery, R. and Croquette, V. (1998) Stretched and overwound DNA forms a Pauling-like structure with exposed bases. Proc Natl AcadSci USA, 95, 14152-14157.

16. LaMarr, W.A. (1998) Thesis Ph. D. --Massachusetts Institute of Technology Division of Toxicology 1998.

17. LaMarr, W.A. Unpublished observations.

18. Shenk, T.E., Rhodes, C., Rigby, P.W. and Berg, P. (1975) Biochemical method for mapping mutational alterations in DNA with S 1 nuclease: the location of deletions and temperature-sensitive mutations in simian virus 40. Proc Natl Acad Sci USA, 72, 989-993.

19. Gray, H.B., Jr., Ostrander, D.A., Hodnett, J.L., Legerski, R.J. and Robberson, D.L. (1975) Extracellular nucleases of Pseudomonas BAL 31. I. Characterization of single strand-specific deoxyriboendonuclease and double-strand deoxyriboexonuclease activities. Nucleic Acids Res, 2, 1459-1492.

20. Wiegand, R.C., Godson, G.N. and Radding, C.M. (1975) Specificity of the S1 nuclease from Aspergillus oryzae. JBiol Chem, 250, 8848-8855.

21. Dayn, A., Malkhosyan, S., Duzhy, D., Lyamichev, V., Panchenko, Y. and Mirkin, S. (1991) Formation of (dA-dT)n cruciforms in Escherichia coli cells under different environmental conditions. JBacteriol,173, 2658-2664.

22. Kohwi-Shigematsu, T., Manes, T. and Kohwi, Y. (1987) Unusual conformational effect exerted by Z-DNA upon its neighboring sequences. Proc Natl Acad Sci US A, 84, 2223-2227.

23. Radloff, R., Bauer, W. and Vinograd, J. (1967) A dye-buoyant-density method for the detection and isolation of closed circular duplex DNA: the closed circular DNA in HeLa cells. Proc Natl Acad Sci US A, 57, 1514-1521.

24. Milligan, J.R., Aguilera, J.A. and Ward, J.F. (1993) Variation of single-strand break yield with scavenger concentration for plasmid DNA irradiated in aqueous solution. Radiat Res, 133, 151-157.

25. Beard, P., Morrow, J.F. and Berg, P. (1973) Cleavage of circular, superhelical simian virus 40 DNA to a linear duplex by S 1 nuclease. J Virol, 12, 1303-1313. 26. Dasgupta, S., Allison, D.P., Snyder, C.E. and Mitra, S. (1977) Base-unpaired regions in supercoiled replicative form DNA of coliphage M13. JBiol Chem, 252, 5916-5923.

27. Bednar, J., Furrer, P., Stasiak, A., Dubochet, J., Egelman, E.H. and Bates, A.D. (1994) The twist, writhe and overall shape of supercoiled DNA change during counterion-induced transition from a loosely to a tightly interwound superhelix. Possible implications for DNA structure in vivo. JMol Biol, 235, 825-847.

28. Kochetkov, N.K., Shibaev, V.N. and Kost, A.A. (1971) New reaction of adenine and cytosine derivatives, potentially useful for nucleic acids modification. TetrahedronLetters, 12, 1993-1996.

29. Sattsangi, P.D., Leonard, N.J. and Frihart, C.R. (1977) 1,N2-ethenoguanine and N2,3-ethenoguanine. Synthesis and comparison of the electronic spectral properties of these linear and angular triheterocycles related to the Y bases. J Org Chem, 42, 3292-3296.

30. Sattsangi, P.D., Barrio, J.R. and Leonard, N.J. (1980) 1,N6-Etheno-bridged adenines and adenosines. Alkyl substitution, fluorescence properties, and synthetic applications. Journalof the American Chemical Society, 102, 770-774.

31. Kusmierek, J.T. and Singer, B. (1982) Chloroacetaldehyde-treated ribo- and deoxyribopolynucleotides. 1. Reaction products. Biochemistry, 21, 5717-5722.

32. Guengerich, F.P. and Persmark, M. (1994) Mechanism of formation of ethenoguanine adducts from 2-haloacetaldehydes: 13C-labeling patterns with 2- bromoacetaldehyde. Chem Res Toxicol, 7, 205-208.

33. Patel, D.J., Kozlowski, S.A., Marky, L.A., Rice, J.A., Broka, C., Dallas, J., Itakura, K. and Breslauer, K.J. (1982) Structure, dynamics, and energetics of deoxyguanosine. thymidine wobble base pair formation in the self- complementary d(CGTGAATTCGCG) duplex in solution. Biochemistry, 21, 437- 444.

34. Aboul-ela, F., Koh, D., Tinoco, I., Jr. and Martin, F.H. (1985) Base-base mismatches. Thermodynamics of double helix formation for dCA3XA3G + dCT3YT3G (X, Y = A,C,G,T). Nucleic Acids Res, 13, 4811-4824.

35. Sasse-Dwight, S. and Gralla, J.D. (1989) KMnO4 as a probe for lac promoter DNA melting and mechanism in vivo. J Biol Chem, 264, 8074-8081.

36. Leonard, N.J., McDonald, J.J., Henderson, R.E. and Reichmann, M.E. (1971) Reaction of diethyl pyrocarbonate with nucleic acid components. Adenosine. Biochemistry, 10, 3335-3342.

Chapter 4 - Raman Spectroscopic Studies of Positive Supercoiling

INTRODUCTION

Unconstrained supercoiling in cellular DNA is ubiquitous (1-3), resulting from the forced rotation of chromosomal DNA by anchored, helix-tracking polymerases and other enzymes (4). Both positive and negative supercoiling are generated during these processes (5), and both can effect vital cellular processes, including transcription, replication and recombination, among others (6-9). Establishing the effects of DNA supercoiling on the structure of the molecule is critical to gaining a more complete understanding of the resultant physiological effects, since the activity of DNA is so intimately tied to its structure.

In the previous chapter, chemical and enzymatic probes that are sensitive to altered DNA secondary structure were used to study structural changes in positively and negatively supercoiled DNA. While negatively supercoiled DNA showed increased reactivity to single strand-specific nucleases, no changes were seen with positively supercoiled DNA. Studies with chemical probes were largely inconclusive. Taken together, these studies suggest that structural changes that occur with positive supercoiling in DNA are subtle and may not be identifiable with these relatively crude structural probes that are usually employed to identify more overt structural changes, like cruciform and Z-DNA formation.

Because of the potentially subtle nature of the structural changes occurring in positively supercoiled DNA, a more sensitive method of detection was required. To this end, spectroscopic and crystallographic methods could provide more detailed information, but methods such as NMR and X-ray crystallography that are often used to determine macromolecular structure are not applicable to supercoiled DNA because of the size and flexibility of supercoiled plasmid molecules. Our current DNA supercoiling method is limited to fairly large DNA molecules (>500 base pairs), which tumble too slowly on the NMR time scale and are too complicated for crystallographic methods.

Fortunately, Raman spectroscopy does not suffer from these same limitations

(10). This method has been successfully applied to investigate changes in DNA structure caused by a multitude of factors, including sequence effects, formation of non-canonical

DNA structures, and the binding of proteins and small molecules to DNA (11-18). DNA has a well-studied Raman signature: there is a significant body of work on the contribution of the different vibrational modes to the overall spectrum, which provides insight concerning specific and subtle changes that occur when a structural effector is introduced. Since the intensity of a given Raman band is proportional to the population of that specific molecular species present in the solution being probed, both the identity and the degree of structural change can be determined by carefully monitoring peak frequencies and intensities.

In this study, highly purified supercoiled plasmid substrates were prepared and, in collaboration with colleagues at the University of Missouri at Kansas City, the Raman spectra of positively and negatively supercoiled plasmids were compared to the spectrum of the relaxed form. Resulting data show that all supercoiling states of the plasmid are predominately B-form DNA, with subtle but reproducible changes induced by

supercoiling. Changes are observed primarily in the DNA sugar-phosphate backbone. MATERIALS AND METHODS

Preparationofsupercoiled and relaxed samples. Positively supercoiled pUC18 plasmid substrates were prepared as described in Chapter 2. Negatively supercoiled and relaxed plasmids were prepared in the same fashion, substituting heat-inactivated CBE for active enzyme at the appropriate relaxation steps to arrive at the final desired supercoiling state. Following purification, all plasmid preparations were dialyzed once against 2 L of 3 M NaCl in TE (10 mM Tris, pH 8.0, 1 mM EDTA) using 100,000 molecular weight cut-off tubing, followed by 4 buffer changes of 2 L 200 mM NaCi adjusted to p-I 7.4.

Samples were concentrated to a final concentration of 25-40 mg/mL using

Microcon YM-30 centrifugal concentrators (Millipore, Billerica, MA). Prior to use, concentrators were washed with 0.1 N NaOH, followed by several rinses with H20 to removed residual glycerin from the membranes. For deuterated samples, concentrated plasmids were lyophilized to dryness, and redissolved in D20 (D, 99.9%) (Cambridge

Isotope Laboratories, Inc., Andover, MA) to the pre-lyophilization volume. In both cases, pH or pD was adjusted prior to preparation of sample cells: pH 7.4 in the case of hydrated samples, and pD 7.0 in the case of deuterated samples. For the sample cells, 8-

12 tiL of concentrated sample was sealed in a glass capillary tube.

Collection of Raman spectra. Raman spectra were collected on two different instruments utilizing different excitation wavelengths. Spectra were collected on a Spex

500-M single spectrograph (Metuchen, NJ) equipped with a liquid nitrogen-cooled, back- thinned, charge-coupled-device detector (Spectrum One, Instruments S.A., Edison, NJ).

Spectra were excited at 532 nm with a solid state Nd:YVO4 laser (Verdi, Coherent, Santa

Clara, CA). Spectra were also collected using near-infrared excitation on a Kaiser Optics

HoloSpec VPT spectrometer (Ann Arbor, MI) equipped with a liquid nitrogen-cooled, charge-coupled-device detector optimized for the near-infrared region (Roper, model

1024EHRB, Trenton, NJ). Spectra were excited at 752 nm with a Ti:sapphire laser

(Coherent, Model 890), which was pumped at 532 nm (Spectra-Physics, model Millennia

X, Mountain View, CA). Raman wavenumbers were calibrated using the emission lines

1 1 from a neon lamp and the 459 cm- band of liquid CC14 and are accurate to ± 1 cm' .

Samples were thermostatted at 17 OC during data collection. Data presented represent an average of approximately 200 exposures of 30 s each.

Processingof Raman spectral data. Following spectral averaging, weak scattering by the aqueous solvent (200 mM NaCl in either H20 or D20) was removed and baselines redrawn using established methods (19). The Raman marker of the DNA phosphate group at 1092 cm-1 was employed for spectral intensity normalizations, as also described previously (13).

RESULTS

Considerationsfor sample preparation. Laser Raman spectroscopy is a powerful technique that is highly sensitive to contamination by non-analyte molecules possessing covalent bonds. Because of this, the utmost care must be taken in the preparation and purification of molecules for Raman studies. Two species of particular concern are Tris cation and glycerol. Tris is commonly used as a buffer in the preparation and manipulation of nucleic acids. It has a distinctive Raman signature with prominent bands at 757, 894, 1064 and 1468 cm 1', all of which could potentially interfere with interpretation of the Raman spectra of the DNA (20). Careful removal of Tris by dialysis against a high salt buffer followed by repeated buffer changes of the final analysis buffer

(200 mM NaC1) is sufficient to remove the cation and prevent it from interfering with the

Raman signal from the supercoiled target.

Another potential source of contamination is glycerin, which is present in small amounts in the centrifugal concentrator membranes to allow for dry storage. Washing

concentrator membranes with 0.1 N NaOH followed by repeated rinses with H2 0 is effective in removing the glycerin and preventing this species from interfering with the

subsequent Raman analyses.

Finally, great care should be taken to avoid contamination with even small amounts of fluorescent species. The nature of Raman scattering, in which -1 in 107 photons is scattered inelastically, requires an intense excitation source to get a relatively

small signal. Because of this, even slight contamination with fluorescent molecules can lead to significant background fluorescence in Raman spectra. Such fluorescence can dramatically reduce the signal to noise ratio. Potential sources of fluorescent contaminants include oxidation products from phenol used in DNA purification procedures, and plasticizers present in H20 from some sources. Using fresh reagents for substrate purification and ultra-pure H20 can significantly reduce fluorescent contamination. The use of near-IR (752 nm) excitation can also dramatically reduce the background from fluorescent contaminants, as these contaminants are less likely to be excited at the longer wavelength.

Raman spectra of negatively supercoiled and relaxedplasmid. Raman spectra of negatively supercoiled and relaxed pUC 18 plasmid and their resultant difference spectra, obtained at 752 and 532 nm, are displayed in Figures 4.1 and 4.2, respectively. Parent spectra had minimal fluorescence and high signal-to-noise ratios that permitted generation of high quality difference spectra.

Raman spectra obtained at both excitation wavelengths reveal that negatively supercoiled and relaxed plasmids exist predominantly in the B-form. The Raman band at

680 cm-1 , indicative of the C2'endo/anti deoxyribose conformation of guanosine nucleosides, the 790 and 833 cm' bands that are indicative t/t phosphodiester torsions

1 in the gauche-/gauche-range, and the C2'H 2 band at 1421 cm- are all reliable markers of

B-form DNA (21), and are clearly present in both negatively supercoiled and relaxed plasmids. The absence of a C3' endo/syn Raman marker band near 625 cm-1 confirms the absence of Z-DNA in both cases (11), as would be expected based upon the sequence of the pUC18 plasmid.

The difference spectra indicate that there are subtle but distinct changes in conformation between negatively supercoiled and relaxed plasmids. These conformational differences can be attributed primarily to modest changes in DNA backbone conformation that accompany the introduction of negative torsion, although some disruption in hydrogen bonding is also apparent. Specific features of the difference spectra will be discussed in detail in the Discussion. Raman spectra ofpositively supercoiledplasmid. In Figures 4.3 and 4.4, the

Raman spectrum obtained on the positively supercoiled plasmid is compared to the

spectrum of the relaxed plasmid, at 752 and 532 nm excitation wavelengths, respectively.

As with negatively supercoiled and relaxed samples, the Raman spectrum of the positively supercoiled plasmid indicates that the B form predominates, with Raman

marker bands evident at 680, 790, 833 and 1421 cm-'. The difference spectra again show

subtle differences between positively supercoiled and relaxed plasmids, although the

differences are not identical to those seen with negative supercoiling. Again, specific

features of the difference spectra will be detailed in the Discussion.

Raman spectra ofsupercoiledand relaxed substrates in D20. To confirm the data

obtained with hydrated samples, Raman scattering measurements were repeated with

deuterated samples (Figs 4.5 and 4.6). In all cases the results obtained on hydrated

samples were reinforced, although spectral baselines for the deuterated samples were not

nearly as good as for the hydrated samples. The reason for this is partial hydration of the

deuterated samples due to humidity in the air. Even with minimization of exposure of the

deutrated samples to the environment, humidity in the air can rapidly exchange with the

solvent. In hydrated samples, the contribution of H20 to the spectra can be removed by

subtracting the spectra of 200 mM NaCl in H20 from the sample spectra. Ideally, the

deuterated samples could be corrected for solvent contributions by subtracting a spectrum

of 200 mM NaCl in D20. However exchange of solvent deuterons with protons from the

environment results in the solvent contribution being a combination of D20 and HDO,

which have distinctly different scattering characteristics. The proportion of D20 to HDO is not identical from sample to sample, leading to the difficulty in solvent correction and the uneven spectral baselines seen in Figures 4.5 and 4.6.

DISCUSSION

Generation ofRaman difference spectra. Raman difference spectroscopy is a highly sensitive technique that can be used to identify subtle and specific differences in

DNA structure with the introduction of an effector of DNA structure. Because of the sensitivity of this technique, it is crucial to eliminate any extraneous differences in the substrates being compared to ensure that all differences observed are a result of the introduction of the parameter being studied. By carefully preparing and purifying the substrates, and comparing supercoiled to relaxed (rather than linear) plasmid of the same sequence, we have hopefully eliminated all causes for the observed difference bands, other than the supercoiled state of the molecule. Since the same difference peaks are arising at both excitation wavelengths, the appearance of these peaks is a real phenomenon, and not merely noise. Further confirmation comes from the deuterated spectra. Negatively supercoiled and relaxed spectra have been obtained in duplicate at both excitation wavelengths; positively supercoiled plasmid awaits another replicate.

Changes in DNA structure with the introduction ofsupercoiling: Base stacking.

Several bands corresponding to base moieties exhibit hypochromism upon base unstacking, the most sensitive of which are the 727 cm' band of adenine and the 1236 cm-1 band of thymine (22,23). No significant intensity increase is observed at either of

100 these positions in difference spectra of negatively or positively supercoiled plasmids

(Figures 4.1 and 4.3, respectively), indicating that base stacking is not significantly

disrupted in either case.

Changes in DNA structure with the introduction ofsupercoiling: Base pairing.

The peak shift from -1483 to 1495 cm-' in both supercoiled states suggests a decrease in the hydrogen bonding strength at the guanine N7 position (16,17). Since the GN7 position

is located in the major groove, a change in the hydrogen bonding status at this site may reflect an increase in exposure of the groove to solvent molecules. Such an increase in major groove exposure might be expected with the increase in DNA bending and twisting that accompany supercoiling. Recently, it has been hypothesized that this shift may also

involve a contribution from the hydrogen bonding state of the 06 position in guanine, so this shift may also suggest a decrease in the stability of G:C base pairing in both supercoiled states.

Positively supercoiled plasmid also shows a slight band shift from 1574 to 1581

n -1 6 cm' . The 1577 cm band in DNA results from the contributions of two modes, the N H2

6 scissor vibration of adenine, and the adenine ring (24). The scattering from the N H2 scissor is very sensitive to deuteration, shifting to 1190 cm-', which is not seen here.

Since the bandshift remains at the 1574 to 1581 cm-' location in the deuterated sample, the shift is due to the contribution from the purine ring (25). Such a shift is associated with a disruption of the hydrogen bonding state of the moiety, so this may represent a decrease in stability of A:T base pairing in positively supercoiled DNA.

Changes in DNA structure with the introduction of supercoiling: Effects on DNA backbone conformation. Both positively and negatively supercoiled plasmids show an

101 increase in intensity at 792 cm -1', a band associated with ca/ý phosphodiester torsions in the gauche-lgauche-range (21,25). In pUC18, which has -50% GC content, the phosphodiester torsion is overlapped by contributions from both cytosine (784 cm ')1 and thymine (790 cm' 1) (26). The intensity change, however, is likely due to perturbation of the phosphodiester torsion, since the shift is not noticeably sensitive to deuteration, as it would be if it were due to either of the base moieties.

The peak at 854 cm-1 that appears with negative but not positive supercoiling indicates a shift in the a/13/y backbone torsions from gauche-/trans/gauchet to a trans/trans/transconformation for some phosphodiester moieties (27). This band has been shown previously to be influenced by helical unwinding, which is consistent with its appearance in negatively but not positively supercoiled DNA.

Positive supercoiling results in an increase in intensity at 680 cm " 1 that is not

observed with negative supercoiling. This peak is diagnostic of C2' endo/anti

deoxyguanosine nucleosides that are characteristic of B-form DNA (11). A

corresponding trough at 811 cm -~ induced by positive supercoiling is also consistent with

a small reduction in the number of sugar moieties in a C3' endo/anti conformation, a

conformation that is generally associated with A-form DNA (28). Additionally, both

positive and negative supercoiling result in an increase in intensity in the 1340-1345 cm- 1

range, with the peak exhibiting greater intensity with positive supercoiling. The 1345

cmr1 band is associated with C2'endo/anti sugar pucker in deoxyadenosine (dA)

nucleosides (25,28). Taken together, the data suggest that a small number of nucleosides

with A-form character are converting to the B-form with the introduction of positive

supercoiling. The presence of C3' endo/anti deoxynucleosides in relaxed mixed-

102 sequence B DNA is not unprecedented. Although the predominant nucleoside conformation is expected to be C2' endo/anti for B DNA, a fraction of the furanose rings may exist in a. C3' end/anti conformation (29). The conversion of these moieties to a C2' endo/anti conformation as seen here is consistent with the anticipated lengthening of the backbone that is required when DNA accommodates positive supercoiling. When DNA is overwound, the sugar-phosphate backbone must lengthen in order to circumscribe the helical axis an increased number of times. A C3' endo pucker shortens the sugar- phosphate backbone of DNA, bringing adjacent phosphates -1 A closer together than they would be in the C2' endo form. Conversion of C3' endo moieties to the C2' endo conformation allows the backbone to lengthen slightly in order to accommodate the additional helix twist.

Positive and negative supercoiling produce a peak at 930 cm' and a trough at

1063 cm l'. Both are diagnostic of additional perturbations in DNA backbone structure.

The 930 cm-1 band has been assigned to a stretching vibration localized in the deoxyribose ring and has been shown to decrease significantly in intensity upon binding of the protein SRY to DNA (17). The intensity decrease was attributed to hydrophobic contacts between SRY side chains and the deoxyribose rings of the DNA backbone. The increased intensity at 930 cm-' observed in this study may be attributed to a change in deoxyribose ring environments that occurs as a consequence of the significant bending

DNA undergoes when plectonemically supercoiled.

Comparison of Raman spectra of Supercoiled DNA with published values.

Several attempts have been made to ascertain the effects of supercoiling on DNA structure by comparing the Raman spectrum of a supercoiled plasmid to either its relaxed

103 or linearized form (30-32). Such attempts have seen limited success, primarily due to signal-to-noise ratios that are insufficient to reveal the subtle differences in structure induced by supercoiling. Both instrumentation and spectral processing software have improved considerably over time allowing for detection of more subtle changes in the

Raman spectrum of supercoiled DNA (33).

Most recently, the Raman spectrum of a negatively supercoiled pUC 19 plasmid was compared to its linearized form (20). Interestingly, results from that study are not consistent with the data presented here. The primary reason may be the difference in the forms of DNA being compared. Comparing supercoiled to relaxed DNA will not necessarily reveal the same spectral differences as comparing the supercoiled to the linear form. Circularizing DNA necessarily imposes some restrictions on DNA geometry that will dictate what conformations the molecule is free to sample. These restrictions may manifest themselves in slight differences in the Raman spectrum of DNA, particularly in the DNA backbone region. A recent study that compared Raman spectra of short (222 bp) predominately supercoiled circularized molecules with their linear pre-cursor found a

nearly null difference spectrum with modest bands at positions similar to those seen in

the supercoiled versus linear pUC19 study (34). Based on this finding, it seems likely

that the differences observed in the previous pUC 19 study better reflect the differences

between circular and linear DNA than the differences between supercoiled and relaxed

DNA. Comparison of relaxed pUC18 plasmid with the linear form may help clarify the

contributions of circularization versus supercoiling to the Raman spectrum.

104 CONCLUSIONS

In collaboration with colleagues at the University of Missouri at Kansas City, I have generated Raman difference spectra to analyze the structural changes introduced to the DNA by positive and negative supercoiling. When supercoiling of either direction is introduced, the molecules remain in a predominately B-form conformation, with only slight changes in structure occurring to compensate for the change in helix winding of the

DNA. These are primarily in the backbone of the molecule, with phosphodiester torsions changing in both positively and negatively supercoiled DNA. Perhaps the most striking observation was that of sugar pucker in positively supercoiled DNA converting to a more homogenously C2'endo form to allow for the extra length in an overwound backbone.

Base stacking, was not significantly perturbed, but the environment of the exchangeable protons of the bases was to some extent, suggesting a slight change in the accessibility of the major groove as well as some destabilization of base pairing.

105 A

3 3. B

C

D

400: 600 800 1000 1200 1400 1600

Figure 4.1. Raman spectra (300-1750 cm1 752 nm excitation) of solutions of A) negatively supercoiled and B) relaxed pUC18 plasmid DNA in 200 mM NaCl, pH 7.4 in H20 at 17 OC. C) Difference spectrum between the two. D) Difference spectrum amplified x3.

106 532 nrm exc4tabo

Nega-lwvey 2:rded I

A

1 ili 'PRead

:I 'I 2

rlierence

- -~ -,r . .-. . Differnce8 X3 .. C a wamNOV

i.. 0 .... 600ý ,...... 400 600 800C 1000 20.0 1400 600 cm Figure 4.2. Raman spectra (300-1750 cm -1 532 nm excitation) of solutions of A) negatively supercoiled and B) relaxed pUC18 plasmid DNA in 200 mM NaC1, pH 7.4 in H20 at 17 'C. C) Difference spectrum between the two. D) Difference spectrum amplified x3.

107 :~~~ ~-~~~~---~--~I---I~-1~---I------1------I ------1 ----- ·----- 1----1-~1~ '*:" - z rim excitauon

..... •4o rc

kts tr% ~~ce j] `II ct'1A4 .:if K

Iz kl i ci_I:' A it k9. :jL r't K

Rezaxed I' I K1 K- )D erence C

e D

Ch. F21 :I :i IN ----...... ------...~...... -- I...... 400 600 800B 1000 1200 1400 1:600 cm-' Figure 4.3. Raman spectra (300-1750 cm 1 752 nm excitation) of solutions of A) positively supercoiled and B) relaxed pUC18 plasmid DNA in 200 mM NaCl, pH 7.4 in H20 at 17 'C. C) Difference spectrum between the two. D) Difference spectrum amplified x3.

108 532 nm Exc-:acen

ii 'I J ;,C6s live y S4Derc•Z', ii ii' I `i --i

INJ V I ~ :~I~ f (idi1 j1ai 1A i:I A~ 1- i·1r i, A 'i ": * i : ii i 1 i; ~I ii· it i ii·

i ii:

Reiaxed ii: ii· · ,i iii B '" I 1 :· ii NJ i: i h ilj ·i\i. i: dliv ji *J·t/ Ic~B

V-A.C

bfferernce X 3 ,7 -

· a). 1 8 '-A I4 L - -- · -0)11

400 600 800 1000 1200 1400 1600 cm- Figure 4.4. Raman spectra (300-1750 cm-l 532 nm excitation) of solutions of A) positively supercoiled and B) relaxed pUC18 plasmid DNA in 200 mM NaC1, pH 7.4 in H20 at 17 oC. C) Difference spectrum between the two. D) Difference spectrum amplified x3.

109 A

B

C 3-r

D

400 600 800 1000 12:00 1400 1600

Figure 4.5. Raman spectra (350-1800 cm 1' 752 nm excitation) of solutions of A) negatively supercoiled and B) relaxed pUC18 plasmid DNA in 200 mM NaC1, pH 7.4 in D2 0 at 17 'C. C) Difference spectrum between the two. D) Difference spectrum amplified x3.

110 A

B ý3

¸ 2 C

D

400 600 800 1000 1200 1400 1600

Figure 4.6. Raman spectra (350-1750 cm 1 752 nm excitation) of solutions of A) positively supercoiled and B) relaxed pUC18 plasmid DNA in 200 mM NaC1, pH 7.4 in D20 at 17 'C. C) Difference spectrum between the two. D) Difference spectrum amplified x3.

111 LITERATURE CITATIONS

1. Figueroa, N. and Bossi, L. (1988) Transcription induces gyration of the DNA template in Escherichia coli. Proc Natl Acad Sci U S A, 85, 9416-9420.

2. Giaever, G.N. and Wang, J.C. (1988) Supercoiling of intracellular DNA can occur in eukaryotic cells. Cell, 55, 849-856.

3. Brill, S.J. and Sternglanz, R. (1988) Transcription-dependent DNA supercoiling in yeast DNA topoisomerase mutants. Cell, 54, 403-411.

4. Liu, L.F. and Wang, J.C. (1987) Supercoiling of the DNA template during transcription. Proc Natl Acad Sci USA, 84, 7024-7027.

5. Wu, H.Y., Shyy, S.H., Wang, J.C. and Liu, L.F. (1988) Transcription generates positively and negatively supercoiled domains in the template. Cell, 53, 433-440.

6. Stirdivant, S.M., Crossland, L.D. and Bogorad, L. (1985) DNA supercoiling affects in vitro transcription of two maize chloroplast genes differently. Proc Natl Acad Sci USA, 82, 4886-4890.

7. Funnell, B.E., Baker, T.A. and Kornberg, A. (1986) Complete enzymatic replication of plasmids containing the origin of the Escherichia coli chromosome. JBiol Chem, 261, 5616-5624.

8. Droge, P. (1993) Transcription-driven site-specific DNA recombination in vitro. Proc Natl Acad Sci US A, 90, 2759-2763.

9. Pfaffle, P., Gerlach, V., Bunzel, L. and Jackson, V. (1990) In vitro evidence that transcription-induced stress causes nucleosome dissolution and regeneration. J Biol Chem, 265, 16830-16840.

10. Thomas, G.J. (1999) Raman spectroscopy of protein and nucleic acid assemblies. Annu Rev Biophys Biomol Struct, 28, 1-27.

11. Benevides, J.M. and Thomas, G.J., Jr. (1983) Characterization of DNA structures by Raman spectroscopy: high-salt and low-salt forms of double helical poly(dG- dC) in H20 and D20 solutions and application to B, Z and A-DNA. Nucleic Acids Res, 11, 5747-5761.

12. Benevides, J.M., Weiss, M.A. and Thomas, G.J., Jr. (1991) Design of the helix- turn-helix motif: nonlocal effects of quaternary structure in DNA recognition investigated by laser Raman spectroscopy. Biochemistry, 30, 4381-4388.

13. Benevides, J.M., Weiss, M.A. and Thomas, G.J., Jr. (1991) DNA recognition by the helix-turn-helix motif: investigation by laser Raman spectroscopy of the phage

112 lambda repressor and its interaction with operator sites OL 1 and OR3. Biochemistry, 30, 5955-5963.

14. Benevides, J.M., Wang, A.H., van der Marel, G.A., van Boom, J.H. and Thomas, G.J., Jr. (1989) Effect of the G.T mismatch on backbone and sugar conformations of Z-DNA and B-DNA: analysis by Raman spectroscopy of crystal and solution structures of d(CGCGTG) and d(CGCGCG). Biochemistry, 28, 304-310.

15. Benevides, J.M. and Thomas, G.J., Jr. (2005) Local conformational changes induced in B-DNA by ethidium intercalation. Biochemistry, 44, 2993-2999.

16. Benevides, J.M., Li, T., Lu, X.J., Srinivasan, A.R., Olson, W.K., Weiss, M.A. and Thomas, G.J., Jr. (2000) Protein-directed DNA structure II. Raman spectroscopy of a leucine zipper bZIP complex. Biochemistry, 39, 548-556.

17. Benevides, J.M., Chan, G., Lu, X.J., Olson, W.K., Weiss, M.A. and Thomas, G.J., Jr. (2000) Protein-directed DNA structure. I. Raman spectroscopy of a high- mobility-group box with application to human sex reversal. Biochemistry, 39, 537-547.

18. Benevides, J.M., Kang, C. and Thomas, G.J., Jr. (1996) Raman signature of the four-stranded intercalated cytosine motif in crystal and solution structures of DNA deoxycytidylates d(CCCT) and d(C8). Biochemistry, 35, 5747-5755.

19. Benevides, J.M., Wang, A.H., van der Marel, G.A., van Boom, J.H., Rich, A. and Thomas, G.J., Jr. (1984) The Raman spectra of left-handed DNA oligomers incorporating adenine-thymine base pairs+. Nucleic Acids Res, 12, 5913-5925.

20. Serban, D., Benevides, J.M. and Thomas, G.J., Jr. (2002) DNA secondary structure and Raman markers of supercoiling in Escherichia coli plasmid pUC 19. Biochemistry, 41, 847-853.

21. Prescott, B., Steinmetz, W. and Thomas, G.J., Jr. (1984) Characterization of DNA structures by laser Raman spectroscopy. Biopolymers, 23, 235-256.

22. Erfurth, S.C. and Peticolas, W.L. (1975) Melting and premelting phenomenon in DNA by laser Raman scattering. Biopolymers, 14, 247-264.

23. Movileanu, L., Benevides, J.M. and Thomas, G.J., Jr. (2002) Determination of base and backbone contributions to the thermodynamics of premelting and melting transitions in B DNA. Nucleic Acids Res, 30, 3767-3777.

24. Toyama, A., Takeuchi, H. and Harada, I. (1991) Ultraviolet resonance Raman spectra of adenine, uracil and thymine derivatives in several solvents. Correlation between band frequencies and hydrogen-bonding states of the nucleic acid bases. Journalof Molecular Structure, 242, 87-98.

113 25. Movileanu, L., Benevides, J.M. and Thomas, G.J. (1999) Temperature dependence of the Raman spectrum of DNA. Part I - Raman signatures of premelting and melting transitions of Poly(dA-dT).Poly(dA-dT). Journalof Raman Spectroscopy, 30, 637-649.

26. Deng, H., Bloomfield, V.A., Benevides, J.M. and Thomas, G.J., Jr. (1999) Dependence of the Raman signature of genomic B-DNA on nucleotide base sequence. Biopolymers, 50, 656-666.

27. Benevides, J.M., Kukolj, G., Autexier, C., Aubrey, K.L., DuBow, M.S. and Thomas, G.J. (1994) Secondary Structure and Interaction of Phage D108 Ner Repressor with a 61-Base-Pair Operator: Evidence for Altered Protein and DNA Structures in the Complex. Biochemistry, 33, 10701-10710.

28. Thomas, G.J., Jr. and Benevides, J.M. (1985) An A-helix structure for poly(dA- dT) X poly(dA-dT). Biopolymers, 24, 1101-1105.

29. Westhof, E. (1987) Re-refinement of the B-dodecamer d(CGCGAATTCGCG) with a comparative analysis of the solvent in it and in the Z-hexamer d(5BrCG5BrCG5BrCG). JBiomol Struct Dyn, 5, 581-600.

30. Christens-Barry, W.A., Martin, J.C. and Lebowitz, J. (1989) Raman spectroscopy of supercoiled and nicked ColEl plasmid. Biopolymers, 28, 1515-1526.

31. Brahms, S., Nakasu, S., Kikuchi, A. and Brahms, J.G. (1989) Structural changes in positively and negatively supercoiled DNA. Eur JBiochem, 184, 297-303.

32. Vasmel, H. (1985) Influence of supercoiling on DNA structure: laser Raman spectroscopy of the plasmid pBR322. Biopolymers, 24, 1001-1008.

33. Benevides, J.M., Overman, S.A. and Thomas, G.J. (2005) Raman, polarized Raman and ultraviolet resonance Raman spectroscopy of nucleic acids and their complexes. JRaman Spectroscopy, 36, 279-299.

34. Benevides, J.M., Serban, D. and Thomas, G.J., Jr. (2006) Structural perturbations induced in linear and circular DNA by the architectural protein HU from Bacillus stearothermophilus. Biochemistry, 45, 5359-5366.

114 Chapter 5 - Design and Optimization of Fluorescent DNA Substrates

for Micromanipulator Studies

INTRODUCTION

In previous chapters, I have demonstrated that physiological levels of positive

supercoiling have subtle effects on the structure of DNA. Both positive and negative

supercoiling are accommodated by small changes in sugar-phosphate backbone and a low

level disruption of base pairing. While crude chemical and enzymatic probes showed no

evidence of base unpairing in positively supercoiled DNA, Raman spectral data indicate

that disruption of pairing does indeed occur in a small fraction of the bases in positively

supercoiled molecules, accompanied by slight changes in phosphodiester torsion and

sugar pucker. These studies, however, were limited to plasmid substrates with a fixed

level of superhelical density. Just as DNA itself is a dynamic molecule, supercoiling is a

dynamic parameter, and subject to a number of determining factors. Topoisomerase

activity, level of transcription, and relative orientation of neighboring genes in the same topological domain will all affect the degree of supercoiling present (1,2). Additionally,

supercoiling is not always evenly distributed throughout a given domain; supercoiling generated by an actively processing enzyme will exist in a gradient, with the highest degree of supercoiling immediately adjacent to the enzyme (3). Because the degree of supercoiling can vary so greatly, it would be useful to examine its consequences at a wide range of superhelical densities.

115 Since plasmids are usually examined at a fixed superhelical state, they only provide information at one point along the continuum. While it is possible to isolate individual topoisomers for experimentation, obtaining the requisite amounts for detailed analysis is difficult, and technical limitations prevent the overall level of positive supercoiling from being above a certain superhelical density. Therefore, a potentially more informative line of experimentation would involve a substrate in which the superhelical density could be easily manipulated and precisely controlled over a wide range of superhelical densities.

One approach to this level of precise control involves the use of a magnetic micromanipulator. This entails tethering one end of a surface-anchored DNA molecule to a magnetic bead that is manipulated in a controlled magnetic field, which allows for incremental rotation of the DNA helix (4-6). Such studies thus far, however, have only

looked at global changes in the molecular structure after phase transition, rather than elucidating the fine structural alterations that could occur with lower levels of introduced torsion (4-6). The nature of the data obtained in these earlier studies precludes such fine detail, as the only parameters measured are the force applied to the bead, the length of the

DNA, and the superhelical density (4-6).

Considerably more information could be obtained by including a sensitive reporter of DNA structure. One such reporter is 2-aminopurine (2AP), which is a

fluorescent base analog widely used in oligonucleotide-based studies of DNA structure

and DNA-protein interactions (7-11). 2AP forms a stable base pair with thymine, in a

Watson-Crick geometry (12) that is only slightly less stable than its A:T counterpart (Fig

5.1) (13). It has an excitation maximum at 303 nm, allowing it to be selectively excited

116 at a wavelength that will reduce autofluorescence from canonical bases in the helix (14).

Perhaps most importantly, fluorescence properties of 2AP are extremely sensitive to the

environment of the base: Both the fluorescence yield and lifetime are significantly

reduced when the base analog is stacked in a normal B-form helix, and the effects on the

fluorescence properties for various perturbations in the environment of the base have

been studied in some detail (7,8,14,15). Thus, 2AP has been used as a sensitive probe to

investigate changes in the base environment resulting any number of factors, ranging

from the effects of abasic sites on neighboring bases (7,8), to the action of "base-

flipping" enzymes that act by everting a base from the DNA helix into an active site

pocket of the protein (9,16).

If positive supercoiling leads to solvent exposure of DNA bases, then 2AP should

help reveal the nature and extent of this exposure. Combining 2AP-containing DNA with

a magnetic micromanipulator would allow for structural changes to be determined as a

function of twist, providing a detailed picture of the base environment as precisely

controlled amounts of twist are introduced. This would significantly broaden our

understanding of overwinding and its effect on the solvent environment of DNA bases.

In this chapter, I describe preliminary efforts to prepare 2-aminopurine-containing

micromanipulator substrates. Current technical limitations have rendered these studies unfeasible, but the information obtained in these preliminary efforts should prove valuable for future efforts at studying changes to the base environment using a micromanipullation approach.

117 MATERIALS AND METHODS

Design of a 2-aminopurine-containingDNA substratefor use in a magnetic micromanipulator. A detailed description of the design for the micromanipulator substrate is given in Figures 5.2 and 5.3. Briefly, a 2.1 kb segment of 2-aminopurine- containing DNA is generated using PCR, and digested with restriction enzymes to produce unique overhangs on each end. These fragments are dimerized, and ligated to a large X-DNA restriction fragment to produce a circular molecule. A further restriction digest removes a small fragment of the circle opposite the 2AP-labeled portion, resulting in a long linear molecule with a region of incorporated 2AP in the middle. Short linker fragments, generated by restriction digestion of plasmid pUC 18, followed by labeling with either biotin or digoxygenin, are then ligated on to the ends of the long molecule, and the resultant substrate is ready for use in the micromanipulator.

Introduction of2-aminopurine into DNA substrates using modified PCR. A 2135 bp PCR product was amplified from plasmid pBR322 using the primers 5'-

CCTGCTCGCTTCGCTACTTG-3' and 5'-GGATAACCGTATTACCGCCTTTG-3' with Taq polymerase (New England Biolabs, Ipswich, MA). Reactions were carried out

in lx ThermoPol Buffer (20 mM Tris-HC1, 10 mM (NH4)2 SO4, 10 mM KC1, 2 mM

MgSO 4, 0.1% Triton X- 100, pH 8.8). 2-Aminopurine-2' deoxyribose-5'-triphosphate

(dAPTP) (TriLink Biotechnologies, San Diego, CA) was added to the reaction mixture to a concentration of 150 tiM; the concentration of all other dNTPs was 200 rtM. After an initial denaturation step at 95 'C for 5 min, 30 amplification cycles were performed,

118 consisting of the following segments: 95 TC for 30 sec, 61 TC for 60 s and 72 TC for 105

s. Following the amplification cycles, a final elongation step at 72 'C for 7 min was

performed. 2-AP incorporation was assessed by fluorescence emission yield (ex: 303

nm, em: 370 nm) after purification and depurination.

Digestion and purification of components for the micromanipulatorsubstrate.

Restriction enzyme digests to obtain the requisite DNA fragments are summarized in

Table 5.1. Following digestion, the desired fragments were isolated by resolution on a

Gen-Pak FAX anion exchange column (Waters, Milford, MA), using an Agilent 1100

HPLC (Santa Clara, CA). Fragments were eluted using a gradient of 0 to 1 M NaCl in

TE buffer (10 mM Tris, pH 8.0, 1 mM EDTA) at a flow rate of 0.75 mL/min for 30 min.

Absorbance at 254 nm was used to identify peaks for collection, which were then

precipitated by addition of an equal volume of ice-cold isopropanol, followed by

centrifugation at 16,000 x g for 30 min. The resulting pellets were air dried and

resuspended in TE prior to further manipulation.

Labeling of linkerfragments with biotin or digoxygenin. Following HPLC

purification, linker segments of DNA were labeled with either biotin or digoxygenin

using Chem-Link labeling kits (Roche Applied Sciences, Indianapolis, IN) according to the instructions provided, with the exception of a reduced reaction temperature of 65 'C.

Degree of labeling was assessed colorimetrically using streptavidin- or anti-DIG-linked

alkaline phosphatase in conjunction with NBT/BCIP (Roche Applied Sciences).

Fluorescence lifetime studies of 2-aminopurine substrates in solution. For fluorescence lifetime measurements of both free 2APTP and 2AP-containing DNA, samples were excited with femtosecond pulses from a Ti:sapphire laser (Mira, Coherent,

119 Santa Clara, CA) tuned to 909 nm and frequency tripled to 303 nm with an Inrad model

5-050 frequency harmonic generator (Northvale, NJ). Emission was detected at 370 nm using a custom-made microscope based on an Axiovert 110 inverted microscope (Zeiss,

G6ttingen, Germany) equipped with a Becker-Hickl 730 time-correlated single photon counting module (Berlin, Germany). Steady state total internal reflection (TIR) fluorescence was measured using the same instrumentation; the beam was reflected in a quartz prism to produce the evanescence required for excitation, and the emission was measured using the inverted microscope described above equipped with a photomultiplier tube (R5600-P, Hamamatsu, Bridgewater, NJ).

RESULTS

2-Aminopurine can be efficiently incorporatedinto DNA using PCR. 2-

Aminopurine incorporation has frequently been used to study polymerase fidelity (17-

19), however, a PCR-based method to generate densely-labeled 2-aminopurine- containing substrates has previously been lacking. Given the stable base pairing of 2AP with thymine (12), and the high level of incorporation by some polymerases (20,21), it is reasonable to expect that such a method could be established. In fact, adding 2AP deoxynucleoside triphosphates to the PCR reaction mixture leads to significant levels of

incorporation, under appropriate conditions.

In recent years, considerable effort has been directed toward engineering thermostable polymerases with improved fidelity. In contrast, our needs demand a polymerase with relatively low fidelity and also limited 3' -5' exonuclease activity (21).

120 By using a polymerase with a higher error rate and limited proofreading capabilities, the amount of incorporated 2AP can be maximized.

While increasing the concentration of 2AP-containing dNTP (dAPTP) increases the level of incorporation, the trade-off is that the overall product yield is diminished.

Therefore, a 150 [tM concentration of dAPTP was selected for preparation of 2AP- containing substrates, as it provides a reasonable level of incorporation (-1 in every 50 bp), while still resulting in acceptable product yield. Having enough 2AP bases incorporated is crucial for adequate detection, but excessive 2AP labeling may partially destabilize the helix, and would thus be undesirable.

Buffer conditions are critical for efficient incorporation of dAPTP by the polymerase. While 2AP was well-incorporated when using New England Biolabs

ThermoPol buffer (20 mM Tris-HC1, 10 mM (NH4)2 SO 4, 10 mM KC1, 2 mM MgSO 4,

0.1% Triton X- 100; pH 8.8), the same company's Standard Taq Reaction buffer (10 mM

Tris-C1, 50 mM KC1, 1.5 mM MgCl 2 ; pH 8.3) resulted in no detectable 2AP incorporation. The reasons for this were not fully investigated, although an increase in

Mg2+ concentration and a increase in pH have both been demonstrated to reduce the fidelity of Taq polymerase (22).

Functionalizationof linker DNA with Chem-Link labeling kits. To retain introduced torsion, both strands of the DNA substrate must be securely anchored, thus multiple biotins distributed on each strand of the linker are desirable. Two methods of biotinylation of the linker DNA were tested: EZ-Link photoactivatable biotin (Pierce

Biotechnology, Rockford, IL), and Chem-Link biotin (Roche Applied Science). The latter was determined to be considerably more effective, with a greater than ten-fold

121 increase in labeling, therefore Chem-Link was also used to functionalize linker DNA with digoxygenin.

PurificationofDNA components using HPLC The Gen-Pak Fax column was effective in purifying desired components ranging in size from -800 to -18000 (23).

Restriction digests were optimized so as to maximize the size differential between the desired component and the remaining fragments, so that the separations were clean and effective in all cases. Theoretically, fragments that are closer in size than those purified in this study could be resolved by slightly adjusting the NaCl gradient, but this was not necessary in any of these separations. Elution of fragments with NaCl in TE buffer facilitates further manipulation, as eluted fragments can be easily purified with an alcohol precipitation step.

Fluorescence lifetime offree and incorporated2-aminopurine. As an initial test of both the instrumentation and the DNA substrate, the fluorescence lifetimes of free dAPTP and 2AP incorporated into purified PCR product were measured in solution (Fig.

5.4). The fluorescence lifetime of free dAPTP was 10.9 ns, consistent with previously published values (7,24). The lifetime for the incorporated 2AP had both a fast and a slow decay component: 55% 0.13 ns, and 45% 3.28 ns. Other values for quenched 2AP

lifetimes were in line with these values, although most studies use more lifetime

components to fit the data (7,24-26). Any variance from previously published values is

likely because those values reflect 2AP incorporation at a single location within an

oligonucleotide, while the PCR product has 2AP incorporation into multiple contexts

opposite a thymine base; sequence context is known to affect fluorescence properties of

2AP in DNA (15). Despite these small differences, this result clearly shows that 2-

122 aminopurine is incorporated sufficiently into DNA using PCR, and that fluorescence

signal quenched to a degree that will allow for easy discernment of bases that are

normally stacked and paired relative to extrahelical bases.

Preliminary tests of2-aminopurine in the TIR instrumentation. In order to

measure sensitivity of the micromanipulator's optical instrumentation prior to assembly

of the substrate within the system, dilutions of dAPTP were tested for signal levels.

Despite an early calculation suggesting that a 20 nM effective concentration would be

sufficient for all proposed experiments, testing showed that a 1 [tM solution would be

necessary for a 2:1 signal-to-noise ratio of fluorescence emission.

In order to calculate the effect of this difference on the feasibility of our system,

we first had to consider the volume of excitation. The laser beam exciting the TIR has a

diameter of 3x10-3 m, for a total area of 7.1x10 -6 m. The evanescent waves that excite the

substrate in TIR are strongest at the interface between the reflective surface and the

solvent medium, and decrease exponentially as the distance from the interface increases.

For the quartz prism used, we will assume an effective excitation depth of 120 nm,

keeping in mind that the level of excitation will not be homogenous throughout. This

gives us a solution volume of 8.5x10-10 L. Since a 1 tM solution is necessary for

adequate signal, a total of 8.5x10-16 mol of unquenched 2AP is necessary to have a

detectable fluorescence signal.

We can now compare this value to the number of moles of 2AP that we could

reasonably expect within our proposed DNA substrates. The magnetic beads used in these experiments are 1 x 10-6m in diameter, so the maximum expected labeling of the

surface with DNA strands would be approximately one strand every 2 x 10-6 m 2. In the

123 excitation area of the solvent medium, we could then expect to have roughly 3.5 x 106

DNA molecules.

Assuming optimal 2AP labeling, we could expect approximately one 2AP base for every 50 bp of DNA. With a B-form DNA rise of 0.34 nm per bp, roughly the first

350 bp of every strand would be excited. This gives an approximate total of 7 2AP residues per molecule, for a total of 4.2x1 017 molecules of 2AP within the excitation volume.

In the case of unquenched fluorophore, this is roughly 20-fold fewer 2AP molecules than would be necessary for adequate signal, but our experiments require that we measure signal from paired and stacked fluorophores as well as from unquenched molecules. 2AP is quenched significantly when it is incorporated into B-form DNA, between 25- to 125-fold, depending on the context (10). If the 2AP is only quenched 25- fold within the context of the helix, the described system would fail to detect the full range of 2AP, from completely stacked to completely extrahelical, by approximately 500-

fold.

Even assuming ideal conditions, the fluorescence yield is significantly less than that required for viable experiments. Incorporation of 2AP and the density of DNA

strands on the surface are near to their maximal values, therefore other changes must be

made in order to realize the proposed studies.

124 DISCUSSION

In this study, I have made initial efforts to develop a system that combines an

incorporated fluorescent base analog with a magnetic micromanipulator in order to study

changes in the DNA base environment that result from supercoiling. These experiments

are not yet technically feasible; however, important considerations were revealed which

will aid any future efforts. Most notable among these are the development of a method to

incorporate fluorescent bases into DNA substrates using PCR, and the optimization of

substrate design for use in a fluorescence micromanipulator system.

The fluorescent base analog 2-aminopurine can be readily incorporated into PCR

substrates using appropriate solvent conditions and a suitably low-fidelity polymerase.

However, despite the utility of 2AP in oligonucleotide-based studies of DNA structure, it

may not be ideal for use in a TIR excitation system. 2AP excites in the low-UV range,

leading to considerable autofluorescence from any impurities in the solvent and in the

quartz prism, as well as a low fluorescence efficiency and a high rate of photobleaching.

Choosing a fluorophore with a red-shifted excitation maximum could significantly reduce

these problems.

One such fluorophore is 6-methylisoxanthopterin (6MI), a bicyclic pteridine

analog that pairs with cytosine and has an excitation maximum at 340 nm (27). The

quantum yield is similar to 2AP: 0.70 (27) compared to 0.68 for 2AP (14). Both the quantum yield and fluorescence lifetime of 6MI are significantly reduced upon incorporation into a DNA helix, although this effect is less pronounced than it is for 2AP

(10,27). Base: pairing is stable with 6MI; only slight melting point depression relative to

125 the unsubstituted form was observed when 6MI was incorporated into an oligonucleotide opposite a cytosine (27). Additionally, a recent study has shown 6MI to be amenable to two-photon excitation, while three-photon excitation of 2AP was significantly less viable (28).

The fluorescence properties of 6MI, particularly with an excitation wavelength that is red-shifted relative to that of 2AP, make it an intriguing candidate for use in micromanipulator studies. However, two major obstacles exist to testing 6MI for these purposes: incorporation and availability. While both 2AP and 6MI form Watson-Crick base pairs, 2AP is a purine rather than a pteridine, and as such may be more readily incorporated by a polymerase. It is possible that a PCR-based method could be developed to incorporate 6MI into DNA, however this may prove difficult, and would require further investigation. Additionally, 6MI has thus far seen limited use, and is not now commercially available in base, nucleoside, or nucleotide form, so obtaining sufficient quantities to construct a substrate is likely to be either expensive or labor- intensive.

In addition to the possible use of a fluorophore, this study has also provided insights into effective substrate design for use in a fluorescence-based micromanipulation

system. The most apparent change that arises in substrate design is the location of the

fluorophore-containing DNA segment. The original substrate had the 2AP-containing portion in the middle, flanked by two pieces of unlabeled extender DNA, which are then

attached to the linkers (Fig. 5.5a). Because TIR excitation is effective only within a

limited distance from the surface of the reflective prism, the fluorophore-containing

region should be as close to that surface as possible. This means that the 2AP-containing

126 fragment should be connected directly to surface linker segment, with a single long piece

of unlabeled extender DNA connected to the bead-linker region (Fig. 5.5b). In addition

to providing better excitation of the fluorophore, such a substrate would be considerably

simpler to construct, as it would require fewer ligation and purification steps.

Another important consideration arising from our initial experiments is the

distance between the fluorophore-containing portion of the DNA molecule and the

excitation source, which is determined by the length of the linker region. If TIR is to be

used as an excitation source, the fluorophores in the DNA molecule must be located as

close to the solid support as possible for maximal fluorescence yield. In the original

substrate, there were two significant problems that must be rectified.

The first involves is the length of the linker, since the longer it is, the further from

the excitation source any attached fluorophore-containing DNA will be. With the current

linker length of -800 bp, the linker DNA would be approximately 275 nm in length.

While the linker will be flattened against the surface to some extent due to contacts

between the functional groups on the linker and their conjugates on the surface, some of

the linker DNA is expected to be free from these linkages and will therefore contribute to

the overall distance between the excitation source and the DNA containing the

fluorophores. Minimizing the length of this linker DNA is therefore desirable.

The second problem with the initial design was the use of digoxygenin (DIG)

bound to anti-DIG antibodies as a method for attaching the linker to the quartz prism. In

this scheme, the DNA was functionalized with DIG, and the prism surface was coated with anti-DIG antibodies. This linkage will both interfere with excitation of the

fluorophore, and add to the noise in the system. In terms of interference of excitation,

127 IgG antibodies are on the order of 15 nm in total length (29), meaning that the linker would not be immediately proximal to the surface, but instead would be offset by the equivalent of -44 bp in length. Since the evanescence in the current system is only effective up to -120 nm, and is strongest at the prism surface, a significant fraction of the excitation energy will not reach the DNA. Additionally, with antibodies coating the TIR interface, the strongest available evanescence will excite the weakly fluorescent amino acids in the protein, leading to a high background signal.

To eliminate these problems, the length of the linker used should be shortened, and a linkage scheme that does not employ antibodies should be used. The linkage used to connect the DNA substrate to the magnetic bead is not subject to such limiting requirements, so it may be possible to swap linkages and use the DIG-anti-DIG for connecting to the bead while using smaller molecules, biotin and streptavidin, to connect to the solid support. Alternatively, covalent linkages could be designed to directly anchor both strands to the surface. In either case, a short oligonucleotide with the appropriate functionalization on each strand should suffice. The main requirements for any linker would remain the same: it needs to bind both strands of the DNA to the support with

sufficient strength to allow retention of applied torsion, and it must not significantly perturb the structure of the DNA.

The realization of a magnetic micromanipulator system in which alterations to the base environment can be measured fluorometrically as a function of twist will likely

require significantly more optimization, both of the instrumentation and the DNA

substrates. Still, this would further our understanding of supercoiling-induced changes to

128 DNA structure tremendously. Information gained from this preliminary study will serve as a guide for any future studies.

129 H

-n ,, H 4N 3

N

N-/ n'

' H13 N /P/N ---- H- N ///ý-N / N-H ----0O

2N H N

,+N-H ---- N N N• _•H N--H ---- 0

H

Figure 5.1. Hydrogen bonding schemes of A) the canonical Watson-Crick adenine-thymine base pair B) the fluorescent base analog 2-aminopurine (2AP) paired to thymine, and C) a proposed protonated form of 2AP that could pair with cytosine.

130 *If

~*I

MtI,`IMLý

Figure 5.2. Preparation of components of micromanipulator substrate. A) A 2AP-containing PCR product is amplified from pBR322, and digested with BamHI and ApaLI to yield a fragment with different four base 5'-overhangs on each end. B) Linker fragments of -800 bp are prepared by double digestion of pUC18 with either XmnI/XbaI or XmnI/BamHI. Linkers were labeled with either biotin or digoxygenin using Roche Chem-Link labeling reagents. C) An 18,909 bp fragment is isolated from a BclI digestion of N6-methyladenine-free X-DNA, and treated with BamHI methyltransferase to protect the two BamHI recognition sites from cleavage in subsequent steps.

131 1 - ~+- I

I

M

.9,,

.4 qyly

Figure 5.3. Assembly scheme for micromanipulator substrate. A) 2-aminopurine-containing fragments are ligated in the presence of BamHI, generating dimers with identical four base 5'overhangs on each end. B) These dimers are ligated to the X-DNA fragment in the presence of BamHI and BclI to prevent self-ligation, in dilute conditions to promote circle formation. C) These circles are double digested with BglII and Xbal to form a linear molecule with unique four base 5'overhangs. D) Labeled linker DNA is ligated to the ends of the linear molecule, completing the substrate.

132 0.75

U C U U) o:3 9 0.5 V N E z0 ZCo 0.25

0 1 2 3 4 5 6 7 Time (ns)

Figure 5.4. Fluorescence lifetime of free (top curve) and incorporated (bottom curve) 2- aminopurine. Lifetime was significantly quenched by incorporation into the DNA helix. Fluorescence lifetimes were determined to be: free - 10.9 ns, incorporated - 55% 0.13 ns, 45% 3.28 ns.

133 B •1111 •111

Figure 5.5. Comparison of original design of micromanipulator substrate with optimized design. A) Substrate as originally designed. Fluorophore is located in the middle of the substrate, too distant from the quartz surface to be excited. Antibodies link a long linker to the surface, putting the fluorophore containing region at a further distance from the excitation. B) Substrate design optimized for TIR excitation. Fluorophore-containing region is as close to the quartz surface as possible, linked to the surface by a short, non-antibody-based linker.

134 LITERATURE CITATIONS

1. Wu, H.Y., Shyy, S.H., Wang, J.C. and Liu, L.F. (1988) Transcription generates positively and negatively supercoiled domains in the template. Cell, 53, 433-440.

2. Tsao, Y.P., Wu, H.Y. and Liu, L.F. (1989) Transcription-driven supercoiling of DNA: direct biochemical evidence from in vitro studies. Cell, 56, 111-118.

3. Wang, Z. and Droge, P. (1997) Long-range effects in a supercoiled DNA domain generated by transcription in vitro. J Mol Biol, 271, 499-510.

4. Allemand, J.F., Bensimon, D., Lavery, R. and Croquette, V. (1998) Stretched and overwound DNA forms a Pauling-like structure with exposed bases. Proc Natl Acad Sci USA, 95, 14152-14157.

5. Strick, T.R., Allemand, J.F., Bensimon, D., Bensimon, A. and Croquette, V. (1996) The elasticity of a single supercoiled DNA molecule. Science, 271, 1835- 1837.

6. Strick, T.R., Allemand, J.F., Bensimon, D. and Croquette, V. (1998) Behavior of supercoiled DNA. Biophys J, 74, 2016-2028.

7. Rachofsky, E.L., Seibert, E., Stivers, J.T., Osman, R. and Ross, J.B. (2001) Conformation and dynamics of abasic sites in DNA investigated by time-resolved fluorescence of 2-aminopurine. Biochemistry, 40, 957-967.

8. Stivers, J.T. (1998) 2-Aminopurine fluorescence studies of base stacking interactions at abasic sites in DNA: metal-ion and base sequence effects. Nucleic Acids Res, 26, 3837-3844.

9. McCullough, A.K., Dodson, M.L., Scharer, O.D. and Lloyd, R.S. (1997) The role of base flipping in damage recognition and catalysis by T4 endonuclease V. J Biol Chem, 272, 27210-27217.

10. Bloom, L.B., Otto, M.R., Eritja, R., Reha-Krantz, L.J., Goodman, M.F. and Beechem, J.M. (1994) Pre-steady-state kinetic analysis of sequence-dependent nucleotide excision by the 3'-exonuclease activity of bacteriophage T4 DNA polymerase. Biochemistry, 33, 7576-7586.

11. Raney, K.D., Sowers, L.C., Millar, D.P. and Benkovic, S.J. (1994) A fluorescence-based assay for monitoring helicase activity. Proc Natl Acad Sci US A,91, 6644-6648.

12. Sowers, L.C., Fazakerley, G.V., Eritja, R., Kaplan, B.E. and Goodman, M.F. (1986) Base pairing and mutagenesis: observation of a protonated base pair between 2-aminopurine and cytosine in an oligonucleotide by proton NMR. Proc Natl Acad Sci USA, 83, 5434-5438.

135 13. Eritja, R., Kaplan, B.E., Mhaskar, D., Sowers, L.C., Petruska, J. and Goodman, M.F. (1986) Synthesis and properties of defined DNA oligomers containing base mispairs involving 2-aminopurine. Nucleic Acids Res, 14, 5869-5884.

14. Ward, D.C., Reich, E. and Stryer, L. (1969) Fluorescence studies of nucleotides and polynucleotides. I. Formycin, 2-aminopurine riboside, 2,6-diaminopurine riboside, and their derivatives. JBiol Chem, 244, 1228-1237.

15. Jean, J.M. and Hall, K.B. (2002) 2-Aminopurine electronic structure and fluorescence properties in DNA. Biochemistry, 41, 13152-13161.

16. Allan, B.W., Beechem, J.M., Lindstrom, W.M. and Reich, N.O. (1998) Direct real time observation of base flipping by the EcoRI DNA methyltransferase. JBiol Chem, 273, 2368-2373.

17. Goodman, M.F., Hopkins, R. and Gore, W.C. (1977) 2-Aminopurine-induced mutagenesis in T4 bacteriophage: a model relating mutation frequency to 2- aminopurine incorporation in DNA. Proc Natl Acad Sci US A, 74, 4806-4810.

18. Clayton, L.K., Goodman, M.F., Branscomb, E.W. and Galas, D.J. (1979) Error induction and correction by mutant and wild type T4 DNA polymerases. Kinetic error discrimination mechanisms. JBiol Chem, 254, 1902-1912.

19. Watanabe, S.M. and Goodman, M.F. (1982) Kinetic measurement of 2- aminopurine X cytosine and 2-aminopurine X thymine base pairs as a test of DNA polymerase fidelity mechanisms. Proc Natl Acad Sci USA, 79, 6429-6433.

20. Grossberger, D. and Clough, W. (1981) Incorporation into DNA of the base analog 2-aminopurine by the Epstein-Barr virus-induced DNA polymerase in vivo and in vitro. Proc Natl Acad Sci US A, 78, 7271-7275.

21. Pless, R.C., Levitt, L.M. and Bessman, M.J. (1981) Nonrandom substitution of 2- aminopurine for adenine during deoxyribonucleic acid synthesis in vitro. Biochemistry, 20, 6235-6244.

22. Eckert, K.A. and Kunkel, T.A. (1990) High fidelity DNA synthesis by the Thermus aquaticus DNA polymerase. Nucleic Acids Res, 18, 3739-3744.

23. Strege, M.A. and Lagu, A.L. (1991) Anion-exchange chromatography of DNA restriction fragments. J Chromatogr, 555, 109-124.

24. Guest, C.R., Hochstrasser, R.A., Sowers, L.C. and Millar, D.P. (1991) Dynamics of mismatched base pairs in DNA. Biochemistry, 30, 3271-3279.

25. Rachofsky, E.L., Osman, R. and Ross, J.B. (2001) Probing structure and dynamics of DNA with 2-aminopurine: effects of local environment on fluorescence. Biochemistry, 40, 946-956.

136 26. Nordlund, T.M., Andersson, S., Nilsson, L., Rigler, R., Graslund, A. and McLaughlin, L.W. (1989) Structure and dynamics of a fluorescent DNA oligomer containing the EcoRI recognition sequence: fluorescence, molecular dynamics, and NMR studies. Biochemistry, 28, 9095-9103.

27. Hawkins, M.E., Pfleiderer, W., Balis, F.M., Porter, D. and Knutson, J.R. (1997) Fluorescence properties of pteridine nucleoside analogs as monomers and incorporated into oligonucleotides. Anal Biochem, 244, 86-95.

28. Katilius, E. and Woodbury, N.W. (2006) Multiphoton excitation of fluorescent DNA base analogs. JBiomed Opt, 11, 044004.

29. Werner, T.C., Bunting, J.R. and Cathou, R.E. (1972) The shape of immunoglobulin G molecules in solution. Proc Natl Acad Sci US A, 69, 795-799.

137

Chapter 6 - Effects of Supercoiling on Cleavage by Type II Restriction

Enzymes

INTRODUCTION

In previous chapters, I have addressed structural changes that occur with the introduction of positive supercoiling into DNA. Both positively and negatively supercoiled DNA accommodate introduced torsion with small conformational changes in base pairing and the sugar-phosphate backbone. Details from Raman difference spectroscopy suggest that with positively supercoiled DNA these changes include altered phosphodiester torsion, sugar pucker, and base pairing status. It is crucial to elucidate these structural transitions in order to understand the interaction of DNA with other cellular constituents, including proteins and small molecules. Since unconstrained supercoiling generated by processive enzymes is ubiquitous in the DNA of both prokaryotes and eukaryotes (1-3), proteins that interact with DNA are likely to encounter supercoiling in the course of normal cellular activity. Even with subtle changes in DNA structure, supercoiling has been shown to alter the way in which proteins are able to interact with DNA, through mechanisms stemming from both the twist and writhe components (4,5).

While: many proteins have been identified that have differential interactions with negatively supercoiled DNA (6-10), positively supercoiled DNA has not been as well- examined, though some examples of altered binding or reactivity are known (8,9,11).

139 Enzymes that may be active on both positively and negatively supercoiled substrates are likely to interact with each substrate in subtly different ways.

One example is human topoisomerase I, a Type IB topoisomerase that nicks one strand of DNA and allows elastic free energy to drive controlled rotation around the intact strand, thus relieving superhelical tension (12). The crystal structure of the substrate-bound protein shows the enzyme as a hinged structure with two "caps", which wrap around the substrate DNA, bringing opposable "lips" into close proximity and forming a cylinder around the duplex (13). Further crystal structures suggest that this cylinder must be flexible to allow rotation of the DNA (14). Interestingly, molecular dynamics simulations suggest that the protein accommodates this rotation differently when it is relaxing positive, rather than negative, supercoils. When the enzyme is releasing positive supercoils, the "lips" open slightly, while the "hinge" side opens up to release negative supercoils (15). Since relieving supercoils of opposite signs requires rotation of the nicked strand in opposite directions, these structural adjustments differ in order to accommodate the sign of supercoiling being relieved.

While human topoisomerase I is one of the better-characterized proteins that shows differential interaction, in theory any enzyme that interacts with DNA directly could be affected by supercoiling. One such class of enzymes is comprised of the Type II restriction endonucleases, a diverse group that cleaves DNA in a site-specific fashion.

Restriction enzymes are classified as Type II on the basis of their production of reproducible fragmentation patterns and cleavage at fixed sites at or near their recognition sequence (16). This broad definition allows for significant diversity within the type and several subdivisions have been created to classify enzymes based on such factors as

140 subunit composition, nature of recognition site, number of targets, and associated methylation activity (17). The utility of Type II enzymes as tools for molecular biology has led to a widespread search for new enzymes and to date more than 3,500 have been identified (18). Because so many enzymes have been identified in a relatively short period of time, however, the details of target site recognition and reaction mechanism for many of these enzymes remain unknown.

Evidence suggests that negative supercoiling may interfere with the ability of some Type II enzymes to cleave DNA; for some enzymes, supercoiled plasmid with a single recognition site is not cut as efficiently as linear , DNA (19). An exact role for supercoiling, however, isn't entirely clear, as other factors can affect cleavage as well.

These include requirements for an allosteric effector, or interaction with bases outside of the enzyme's recognition sequence.

In this study, I have tested three restriction enzymes (NarI, Ehel, and EcoRI) with relaxed and positively and negatively supercoiled substrates in order to investigate their sensitivity to supercoiling. Only NarI and EcoRI showed subtle changes in reactivity with the introduction of supercoiling. While none of the enzymes tested showed substantial changes, this class of enzymes is sufficiently diverse that it remains possible that some may be strongly affected. Identifying such enzymes for further study would provide valuable information about the ways in which supercoiling affects interactions between DNA and its modifying enzymes.

141 MATERIALS AND METHODS

Preparationofsupercoiled substrates. Positively supercoiled pUC18 substrates were prepared as described in Chapter 2. Relaxed and negatively supercoiled substrates were prepared in the same manner as the positive, substituting heat-inactivated chicken blood extract (CBE) for live enzyme at the appropriate steps to obtain substrates with the desired presence and direction of supercoiling. All substrates were dialyzed once in 2 L of TE buffer with 3 M NaC1, using 100,000 MWCO dialysis tubing, followed by four changes of 2 L of TE buffer, prior to use in enzymatic experiments.

Enzymes and enzyme reactions. Restriction enzymes NarI and EcoRI were obtained from New England Biolabs (Ipswich, MA). EheI was obtained from Fermentas

(Hanover, MD). All reactions took place in the appropriate lx buffer conditions, using the buffer provided by the manufacturer. These conditions were as follows. NarI: 10 mM bis-Tris-propane-HC1, 10 mM MgCl 2, 1 mM dithiothreitol, pH 7.0; EcoRI: 100 mM

Tris-HC1, 50 mM NaC1, 10 mM MgCl 2, 0.025% Triton X-100, pH 7.5; Ehel; 33 mM

Tris-acetate, 10 mM magnesium acetate, 66 mM potassium acetate, 0.1 mg/mL BSA, pH

7.9.

Enzymatic reactions were carried out at 37 TC in 200 [L total volume of the appropriate buffer, with 5 nM DNA concentration. Reactions were initiated by addition of 5 ýtL of the appropriate diluted enzyme, at concentrations of 40 U/mL for NarI, 10

U/mL for EcoRI, and 5 U/mL for Ehel. At pre-determined time points, 10 [tL aliquots were removed from the reaction and quenched by addition of 10 1iL of nuclease stop

142 solution (100 mM EDTA, 100 mM Tris, pH 8.0, 40% sucrose, 0.1% bromophenol blue).

Samples were resolved at 4.5 V/cm for 90 minutes on 1% agarose gels in lx TBE containing 0.5 [tg/mL ethidium bromide. Relative fractions of open circular, linear, and closed-circular plasmid were quantified based on ethidium fluorescence, as described in

Chapter 3.

RESULTS

The effects of supercoilingon the activity of restrictionenzymes. Several factors can affect cleavage of cognate DNA by restriction enzymes, including buffer conditions, sequence flanking the recognition site, and the number of recognition sites present on the molecule. By using plasmids that differ only in supercoiling status, and carefully controlling reaction conditions, other factors can be ruled out and any changes in reactivity can. be attributed to supercoiling.

Type II restriction enzymes cleave DNA at a fixed site at or near their recognition sequence. Cleavage of a molecule to linear form can result from one of two modes (Fig.

6.1) (20-22). In the first, intact DNA is bound by the enzyme, which cleaves both strands of the DNA in a single binding step. In the second, the enzyme cleaves a single strand before releasing a nicked molecule, and the intact strand is later cleaved in a second binding event. Since our relaxed and supercoiled substrates are all closed-circular molecules, the products of reaction with restriction enzymes offer insight into these cleavage modes. Conversion of supercoiled plasmid exclusively to linear form indicates the first, two-strand cleavage mode is occurring. Conversely, appearance of open circular

143 (nicked) plasmid in the reaction products is indicative of the second mode. In the second mode, linear products will also appear as the open circular molecules are cleaved. Any given enzyme does not necessarily follow a single reaction pathway; in fact, some enzymes have been shown to partition between the two under some circumstances (22).

NarI cleaves both strands slightly more efficiently in relaxedplasmid DNA. NarI is a type II restriction enzyme that cleaves at the palindromic sequence 5'-GG CGCC'3- to leave a two base 5' overhang (23). It has been reported to cleave supercoiled pUC19

20 times less efficiently than linear X DNA (19). The enzyme cleaves considerably more efficiently when a second recognition site is present (24), but this requirement for an allosteric effector likely would not explain the reported difference in cleavage between supercoiled pUC 19 and linear X, as both contain a single NarI recognition site. NarI cleavage may be also be affected by flanking sequence, cleaving more efficiently in certain contexts. Such sequence differences could contribute significantly to the observed cleavage variations, although supercoiling may also play a role.

To ascertain the effect of supercoiling on cleavage by NarI, the enzyme was reacted with relaxed, positively and negatively supercoiled DNA and the kinetics of reactions were monitored. As shown in Figure 6.2, the overall rate of cleavage was not significantly different among the different states of supercoiling. Overall, the plasmids were cleaved relatively slowly, which is expected for a single-site substrate with an enzyme that requires an allosteric effector. Closed-circular plasmid disappeared at approximately the same rate for all three forms. A difference arises, however, when the rate of conversion to linear and to open circular forms are analyzed separately. While all

144 three forms are predominately converted to open circular, a small fraction of relaxed plasmid is converted to linear.

NarI has previously been demonstrated to cleave negatively supercoiled plasmid via the two-binding-event pathway, cleaving one strand of the substrate during an initial binding and returning to convert open circular plasmid to linear at a much slower rate

(21). This pathway is followed regardless of whether or not an allosteric effector is present (21). The small amount of linear plasmid formed from relaxed plasmid, however, is likely due to a partitioning between the two modes of cleavage. Open circular plasmid is structurally identical, regardless of the original supercoiled state of the molecule, so it would not be converted to the linear form faster in one reaction than in the others. The formation of linear plasmid should therefore be the result of cleavage of both strands in a single binding event.

EheI is not sensitive to eitherpositive or negative supercoiling. Ehel is a neoschizomer of NarI that leaves a blunt end at 5'-GGC GCC-3' sites (25). The enzyme does not require an allosteric effector, and has been shown to cleave both strands of the recognition sequence in a single binding step (21). As can be seen in Figure 6.3, Ehel is largely insensitive to supercoiling, cleaving at approximately equal rates in relaxed, and positively and negatively supercoiled plasmid. Supercoiling also did not affect the cleavage mode of the enzyme, with direct conversion to linear seen in all cases.

Positive supercoilingslightly increases the rate of cleavage by EcoRI. EcoRI is perhaps the best-studied of all restriction enzymes, cleaving at G AATTC sequences to leave a four base 5' overhang (26). The crystal structure of EcoRI bound to its recognition site has been solved, and the protein is shown to distort the DNA upon

145 binding. EcoRI induces a kink in the DNA backbone that is accompanied by widening of both grooves, an increased rise between base pairs and a slight unwinding of the helix

(27). The enzyme can cut either one or both strands in a single binding event (20); the partitioning between single and double cleavage in the first binding event appears to be at least partially determined by the sequence flanking the recognition site (22). EcoRI is reported to cut X DNA somewhat more efficiently than it does supercoiled plasmids pUC19 and pBR322 (19), although the reason for this differential cleavage is not known.

In these studies, EcoRI predominately cleaved directly to linear for all three forms, producing only a small amount of open circular plasmid (Fig. 6.4). The enzyme was slightly affected by supercoiling, cleaving positively supercoiled plasmid at a faster rate than either relaxed or negatively supercoiled plasmid. It is not clear if the change in reactivity is accompanied by a change in partitioning between one- and two-strand cleavage in the first binding event; both the linear and open circular fractions are increased with positive supercoiling relative to the other two forms.

DISCUSSION

Type II restriction enzymes comprise a broad class that act to cleave DNA at a fixed site at or near their recognition sequence. Previous studies have suggested that several of these enzymes may be sensitive to DNA supercoiling, although many of these reports are in conflict (19,28-3 0).

In this study, I have shown that cleavage by restriction enzymes can be affected by supercoiling, although the differences observed were small. Supercoiling could

146 conceivably affect the interaction of these enzymes with DNA in a number of ways, including altering binding of the enzyme and product release.

Specific binding of several Type II enzymes is known to induce conformational changes in the DNA structure, including kinks or bends in the DNA backbone, and unwinding of the helix (27,31). Supercoiling-induced structural changes may affect the ease with which an enzyme can induce these changes. A negatively supercoiled molecule, for example, is more readily unwound than a positive one, since unwinding is already present in the negative molecule. Additionally, supercoiling-induced changes in the structure of the DNA grooves could alter how the enzyme is able to track along the helix during facilitated diffusion, potentially affecting location of the target site.

Supercoiling could also contribute to changes the rate at which the enzymes release cleaved product. Product release is the rate-limiting step for several Type II enzymes under certain conditions (30). When the backbone of a supercoiled molecule is cleaved, the free energy of supercoiling that is trapped in the molecule is held in place only by the bound enzyme. The force of this elastic torsion may be sufficient to drive the enzyme free from the target site at an accelerated rate, allowing for a faster overall cleavage rate by the enzyme. It also may manifest by changing the partitioning between one- and two-strand cleavage in the first binding event. Such a change in partitioning might explain the small fraction of relaxed plasmid that NarI is able to cleave directly to linear; the lack of supercoiling in the molecule may allow the enzyme to remain bound after the first cleavage event in some fraction of the molecules.

The interaction of restriction enzymes with DNA is a complex one, which could be affected by supercoiling in many different ways. Because of this complexity, it is

147 impossible to attribute any difference in reactivity to a given cause without more detailed study. The changes in reactivity I have shown here are small, and provide limited incentive for further investigation. Type II restriction enzymes as a class, however, are sufficiently diverse that it is reasonable to expect an enzyme to be identified with more pronounced supercoiling sensitivity. Once such an enzyme was located, more in depth studies could be employed to elucidate the specific changes supercoiling produces.

148 H H

B C

Figure 6.1. Possible reaction modes of closed circular DNA with restriction enzymes. A) Relaxed, negatively and positively supercoiled plasmid are all covalently closed circular molecules. B) A restriction enzyme binds at its recognition site, and cleaves both strands of the target DNA in a single binding event. C) Alternatively, the restriction enzyme might bind at its recognition site, and cleave one strand in the first binding event. This leaves an open circular molecule. The enzyme then returns to cleave the molecule to linear in a second binding step.

149 * Positive 4 Negative A Relaxed

~- I 14I ~----~-`---"-11-"11`I----~-"~

15

10

4 0 4 ...... 100

Time (minutes)

Figure 6.2. Kinetics of reaction of NarI with supercoiled and relaxed substrates. Relaxed plasmid is cleaved by NarI endonuclease directly to linear at a slightly faster rate than either positively or negatively supercoiled plasmid. Plasmid substrates were reacted with NarI as described in Materials and Methods. Data points shown are the average of four replicates; error bars represent the standard error.

150 S ~~~1 0 Negative L A. Positive A Relaxed ...... 0 0OlL ý- 0. o I I RI~irr 4 "iIts* a r. I I

7----1-- A; ----- 1 t dTh -t---- ...... ------·------;------

1 0

f 0______J i ioo100 I, £ 75 I

50

25

Time (minutes)

Figure 6.3. Kinetics of reaction of Ehel with supercoiled and relaxed substrates. Cleavage progresses directly to linear; rates are the same regardless of supercoiling state. Plasmid substrates were reacted with Ehel as described in Materials and Methods. Data points shown are the average of four replicates; error bars represent the standard error.

151 20

I 15 ( ------0 ~-LIIIII^·l~·--·-·-··---·--~ 100 75 I 50

25

20 40 60 Time (minutes)

Figure 6.4. Kinetics of reaction of EcoRI with supercoiled and relaxed substrates. Positively supercoiled plasmid is cleaved by EcoRI endonuclease at a slightly faster rate than either relaxed or negatively supercoiled plasmid. Plasmid substrates were reacted with EcoRI as described in Materials and Methods. Data points shown are the average of four replicates; error bars represent the standard error.

152 LITERATURE CITATIONS

1. Wu, H.Y., Shyy, S.H., Wang, J.C. and Liu, L.F. (1988) Transcription generates positively and negatively supercoiled domains in the template. Cell, 53, 433-440.

2. Giaever, G.N. and Wang, J.C. (1988) Supercoiling of intracellular DNA can occur in eukaryotic cells. Cell, 55, 849-856.

3. Ljungman, M. and Hanawalt, P.C. (1992) Localized torsional tension in the DNA of human cells. Proc Natl Acad Sci USA, 89, 6055-6059.

4. Steck, T.R., Franco, R.J., Wang, J.Y. and Drlica, K. (1993) Topoisomerase mutations affect the relative abundance of many Escherichia coli proteins. Mol Microbiol, 10, 473-481.

5. Vologodskii, A.V., Levene, S.D., Klenin, K.V., Frank-Kamenetskii, M. and Cozzarelli, N.R. (1992) Conformational and thermodynamic properties of supercoiled DNA. J Mol Biol, 227, 1224-1243.

6. Waldmann, T., Baack, M., Richter, N. and Gruss, C. (2003) Structure-specific binding of the proto-oncogene protein DEK to DNA. Nucleic Acids Res, 31, 7003-7010.

7. Peng, H. and Marians, K.J. (1995) The interaction of Escherichia coli topoisomerase IV with DNA. JBiol Chem, 270, 25286-25290.

8. Mazur, S.J., Sakaguchi, K., Appella, E., Wang, X.W., Harris, C.C. and Bohr, V.A. (1999) Preferential binding of tumor suppressor p53 to positively or negatively supercoiled DNA involves the C-terminal domain. J Mol Biol, 292, 241-249.

9. McClendon, A.K. and Osheroff, N. (2006) The geometry of DNA supercoils modulates topoisomerase-mediated DNA cleavage and enzyme response to anticancer drugs. Biochemistry, 45, 3040-3050.

10. Bae, S.H., Yun, S.H., Sun, D., Lim, H.M. and Choi, B.S. (2006) Structural and dynamic basis of a supercoiling-responsive DNA element. Nucleic Acids Res, 34, 254-261.

11. Levchenko, V., Jackson, B. and Jackson, V. (2005) Histone release during transcription: displacement of the two H2A-H2B dimers in the nucleosome is dependent on different levels of transcription-induced positive stress. Biochemistry, 44, 5357-5372.

12. Stewart, L., Redinbo, M.R., Qiu, X., Hol, W.G. and Champoux, J.J. (1998) A model for the mechanism of human topoisomerase I. Science, 279, 1534-1541.

153 13. Redinbo, M.R., Stewart, L., Kuhn, P., Champoux, J.J. and Hol, W.G. (1998) Crystal structures of human topoisomerase I in covalent and noncovalent complexes with DNA. Science, 279, 1504-1513.

14. Redinbo, M.R., Stewart, L., Champoux, J.J. and Hol, W.G. (1999) Structural flexibility in human topoisomerase I revealed in multiple non-isomorphous crystal structures. J Mol Biol, 292, 685-696.

15. Sari, L. and Andricioaei, I. (2005) Rotation of DNA around intact strand in human topoisomerase I implies distinct mechanisms for positive and negative supercoil relaxation. Nucleic Acids Res, 33, 6621-6634.

16. Pingoud, A., Fuxreiter, M., Pingoud, V. and Wende, W. (2005) Type II restriction endonucleases: structure and mechanism. Cell Mol Life Sci, 62, 685-707.

17. Roberts, R.J., Belfort, M., Bestor, T., Bhagwat, A.S., Bickle, T.A., Bitinaite, J., Blumenthal, R.M., Degtyarev, S., Dryden, D.T., Dybvig, K. et al. (2003) A nomenclature for restriction enzymes, DNA methyltransferases, homing endonucleases and their genes. Nucleic Acids Res, 31, 1805-1812.

18. Roberts, R.J., Vincze, T., Posfai, J. and Macelis, D. (2005) REBASE--restriction enzymes and DNA methyltransferases. Nucleic Acids Res, 33, D230-232.

19. (2006). Cleavage of Supercoiled DNA - New England Biolabs Technical Reference. http://wvvw.neb.com/nebecomm/tech reference/restriction enzymes/cleavage su percoiled dna.asp.

20. Modrich, P. and Zabel, D. (1976) EcoRI endonuclease. Physical and catalytic properties of the homogenous enzyme. JBiol Chem, 251, 5866-5874.

21. Gowers, D.M., Bellamy, S.R. and Halford, S.E. (2004) One recognition sequence, seven restriction enzymes, five reaction mechanisms. Nucleic Acids Res, 32, 3469-3479.

22. Rubin, R.A. and Modrich, P. (1978) Substrate dependence of the mechanism of EcoRI endonuclease. Nucleic Acids Res, 5, 2991-2997.

23. Roberts, R.J. (1982) Restriction and modification enzymes and their recognition sequences. Nucleic Acids Res, 10, r117-144.

24. Oller, A.R., Vanden Broek, W., Conrad, M. and Topal, M.D. (1991) Ability of DNA and spermidine to affect the activity of restriction endonucleases from several bacterial species. Biochemistry, 30, 2543-2549.

25. Kulba, A.M., Abdel-Sabur, M.S., Butkus, V.V., Janulaitis, A. and Fomichev, Y.K. (1987) New type-II restrictase from cells of Erwinia herbicola. Mol. Biol. (Mosk), 21, 250-254.

154 26. Hedgpeth, J., Goodman, H.M. and Boyer, H.W. (1972) DNA nucleotide sequence restricted by the RI endonuclease. Proc Natl Acad Sci USA, 69, 3448-3452.

27. Kim, Y., Choi, J., Grable, J.C., Greene, P., Hager, P. and Rosenberg, J.M. (1994) In Sarma, R. H. and Sarma, M. H. (eds.), StructuralBiology: The State of the Art. Adenine Press, New York, pp. 225-246.

28. Halford, S.E., Johnson, N.P. and Grinsted, J. (1979) The reactions of the EcoRi and other restriction endonucleases. Biochem J, 179, 353-365.

29. Halford, S.E. and Johnson, N.P. (1981) The EcoRI restriction endonuclease, covalently closed DNA and ethidium bromide. Biochem J, 199, 767-777.

30. Hinsch, B. and Kula, M.R. (1981) Reaction kinetics of some important site- specific endonucleases. Nucleic Acids Res, 9, 3159-3174.

31. Winkler, F.K., Banner, D.W., Oefner, C., Tsernoglou, D., Brown, R.S., Heathman, S.P., Bryan, R.K., Martin, P.D., Petratos, K. and Wilson, K.S. (1993) The crystal structure of EcoRV endonuclease and of its complexes with cognate and non-cognate DNA fragments. Embo J, 12, 1781-1795.

155

Chapter 7 - Conclusions

In the previous chapters, I described the preparation of highly purified supercoiled substrates and how I used these substrates to demonstrate that physiological levels of supercoiling induce subtle changes into the structure of DNA. This supercoiling can also affect the interaction of DNA with various DNA-modifying enzymes, as can be seen by the small changes that I have shown in cleavage of DNA by Type II restriction enzymes.

A rigorous method for the preparation of supercoiled DNA substrates was crucial to these studies with the inherent difficulty in preparing highly positively supercoiled plasmid serving as a hindrance in previous studies of the phenomenon. On the basis of a method developed earlier in this lab (1), I undertook the development of a new method for producing highly positively supercoiled substrates. This was necessary due to variable topoisomer distributions, high degree of nicking, and batch-to-batch variability in reactivity of substrates prepared using the original method

The new method employs more stringent quality controls in order to ensure high levels of positive supercoiling, to remove differences between substrates that are not due to supercoiling, and to reduce the population of open circular molecules that contaminate the preparations. Additionally, I have expanded this method to allow the use of a simple salt supplement to control the direction of supercoiling that is induced. This new method allows generation of either positively or negatively supercoiled substrates, and should facilitate studies of supercoiling in which the desired substrate cannot be propagated in a bacterial system, such as DNA circles with non-canonical bases or base adducts inserted.

157 To assess the solvent accessibility of supercoiled substrates, relaxed and positively and negatively supercoiled plasmids were reacted with both enzymatic and chemical probes of DNA structure. Whereas negatively supercoiled plasmid was very sensitive to cleavage by the single strand-specific nucleases S1 and BAL-31, no increased reactivity was seen with positively supercoiled substrates. Chloroacetaldehyde, which reacts with the hydrogen bonding region of the DNA bases to form cyclic etheno adducts, did not show an increase in either 1,N6-ethenoadenine or 3,N4-ethenocytosine with either negative or positive supercoiling. The result with negative supercoiling is somewhat surprising, since negative supercoiling is known to induce denaturation of the DNA helix with increased solvent exposure of the bases for reaction with chloroacetaldehyde. Still, the nature of the denaturation in negatively supercoiled plasmid may be transient, which might prevent etheno adduct formation from occurring at an increased rate.

Chloroacetaldehyde forms etheno adducts in a multi-step reaction process and only the end products are being analyzed in our system. It is possible that the rate-limiting step of the chloroacetaldehyde reaction is slower than the base pair opening rates in the various supercoiled substrates.

The reactivity of negatively supercoiled plasmid that is seen here is consistent with a molecule that has transient denaturation of the DNA strands, but without any stable supercoiling-induced secondary structures. The results with positively supercoiled plasmid are less informative. The lack of reactivity with single strand-specific nucleases could indicate a lack of increased solvent exposure of the bases, although alternative possibilities remain. The structure that is recognized as a target for cleavage by these enzymes is not clear. It may be possible that the structure of positively supercoiled

158 plasmid is changed in such a way that the enzyme is unable to identify it as a cleavage

target. Another possibility is that the rate at which the bases are solvent exposed with

positive supercoiling is kinetically too fast to be recognized by the enzymes. The lack of

reactivity with chloroacetaldehyde is relatively uninformative, since no increase is seen in

reactivity with negatively supercoiled plasmid with its known denaturation. The

strongest conclusion that can be drawn from this data is that an alternative secondary

structure with stably unpaired bases is not forming in the positively supercoiled

substrates.

Studies with other structural probes may prove helpful in furthering our

understanding of structural changes induced by positive supercoiling. While the rate of

formation of the EA and EC adducts may not be a useful measure of solvent exposure, it

has been suggested that the partitioning between the two ethenoguanine adducts (1,N2-

ethenoguanine and N2,3-ethenoguanine) can be affected by base-pairing (2). The N1

position is blocked by hydrogen bonding, so partitioning would shift to favor the N2 ,3EG

adduct in the case of base pairing. Comparing the relative levels of these two adducts in

our substrates may prove more informative than studying the absolute levels of EA and

EC. Dimethyl sulfate should also be investigated further. Preliminary studies in this lab

have suggested that positively supercoiled DNA is methylated at an increased rate at both

GN7 and AN3 by dimethyl sulfate, relative to negatively supercoiled plasmid.

Unfortunately, studies on relaxed substrate have not been performed. Since the reactivity

of relaxed substrates provides a crucial baseline for evaluating the level of reactivity that

is not caused by supercoiling, the lack of this data prevents detailed interpretation of these results. Expanding the dimethyl sulfate studies to include relaxed substrates should

159 provide useful information about the status of the GN 7 and AN3 sites in supercoiled substrates. Reactivity with other chemical probes of DNA structure, such as potassium permanganate and diethyl pyrocarbonate, should also be considered.

To examine the fine structural changes that occur as a result of positive supercoiling, we employed Raman difference spectroscopy. The information-dense and well-defined Raman signature of DNA is very sensitive to conformational changes in the molecule, and by digitally subtracting the spectrum of relaxed plasmid from that of a supercoiled substrate, we can identify supercoiling-induced changes in structure. All forms were shown to be in predominately B-form conformation, with no identifiable conversion to alternative structures such as Z-DNA occurring. This is consistent with our chemical and enzymatic data.

The structural changes that did occur with supercoiling were reproducible, but subtle, suggesting that at the levels of supercoiling tested, the molecules accommodate supercoiling by making small structural adjustments. These adjustments were largely in the sugar-phosphate backbone of the DNA, with a smaller contribution from a slight disruption of the base environment. Importantly, the difference spectra showed that while some changes occurred in both forms of supercoiling, others were unique to the direction of supercoiling induced. In negatively supercoiled plasmid, the most notable of these was a difference peak diagnostic of a change in the torsion angles of the phosphate backbone. For the positive, a series of difference peaks suggests that a small fraction of sugars in the C3' endo form is converting to a C2' endo form that is more consistent with

B-DNA. Since conversion from the C3'endo to the C2'endo sugar pucker allows adjacent phosphates to be farther apart by -1 A, this makes sense in the context of the

160 lengthening of the backbone that is required with positive supercoiling. It is worth noting

that the formation of "P-DNA" at high levels of introduced twist is thought to result from

the phosphate backbone having reached maximal bond lengths, and then adapting to

additional twist by moving to the center of the helix to reduce the helical radius (3). The

change in sugar conformation seen in my experiments may then be an early step in this

process.

Other notable changes that are seen in the positively supercoiled substrates

include an indication of altered hydrogen bonding at the GN 7 position, and a

destabilization of base pairing. Located in the major groove, the GN7 position may be

experiencing a disruption in solvent interactions as a result of the bending and twisting

motions of supercoiled DNA. Free energy studies have determined that the entropy of

supercoiling is large and positive. Since supercoiling limits the structures that a molecule

can sample, the configurational entropy would be expected to be negative and another

entropy component would have to be responsible for the positive change. This increased

entropy has been proposed to result from the disruption of the interaction of DNA with

solvent and other molecules, due to the motions of the supercoiled DNA (4). This would

be consistent with our observations at GN7. Further evidence of this phenomenon could

be obtained from the results of the chemical studies discussed previously, as the GN7 is a target site for dimethyl sulfate methylation. Two peak shifts in the positively supercoiled

difference spectra suggest destabilization of base pairing. Taken together with enzymatic

studies, this suggests that while stable eversion of the bases from the helix is not occurring, some degree of base pair disruption is caused by physiological levels of positive supercoiling.

161 In order to examine the effects of supercoiling at a wide range of superhelical densities, I designed and began preparation of 2-aminopurine-containing DNA substrates for use in a magnetic micromanipulator. In such a system, a long segment of DNA would be anchored on one end to a solid support, and on the other end to a magnetic bead which could be used to introduce precise amounts of twist into the molecule. By incorporating a base analog with known fluorescence properties, changes in the DNA structure could be monitored fluorimetrically. Unfortunately, due to instrument sensitivity, the experiments are not technically feasible. The design of the substrate has been optimized in such a way to significantly improve any future studies. The time and effort involved in preparing these substrates is substantial, however, so it is crucial to verify adequate instrument sensitivity with the proposed system before proceeding with substrate construction.

I then proceeded to investigate the effects of supercoiling on interaction between

DNA and the structurally-sensitive Type II restriction enzymes. In two of the three enzymes tested, a slight sensitivity to supercoiling was seen. The interaction between these enzymes and DNA is complex, however, and particularly with such small differences, the mechanisms behind the observed changes could not be attributed.

Screening of these enzymes is fairly easy, however, and with the diversity of Type II enzymes it seems reasonable that one might be identified with more substantial supercoiling-induced changes in reactivity. A more detailed study of such an enzyme should provide important new information about the effects of supercoiling on protein-

DNA interactions.

Overall, I have demonstrated that positive supercoiling has subtle, but distinct effects on the structure of DNA, which are consistent with the mechanical effects of

162 overwinding. With the ubiquity of supercoiling in cells, and the structural sensitivity of many classes of DNA-interactive enzymes, it is likely that even these small changes can have significant physiological effects. Further study of both the structural consequences of positive supercoiling, and the effects that this has on enzymes that interact with DNA, should help to clarify the biological role of this important phenomenon.

163 LITERATURE CITATIONS

1. LaMarr, W.A., Sandman, K.M., Reeve, J.N. and Dedon, P.C. (1997) Large scale preparation of positively supercoiled DNA using the archaeal histone HMf. Nucleic Acids Res, 25, 1660-1661.

2. Guengerich, F.P. and Persmark, M. (1994) Mechanism of formation of ethenoguanine adducts from 2-haloacetaldehydes: 13C-labeling patterns with 2- bromoacetaldehyde. Chem Res Toxicol, 7, 205-208.

3. Allemand, J.F., Bensimon, D., Lavery, R. and Croquette, V. (1998) Stretched and overwound DNA forms a Pauling-like structure with exposed bases. Proc Natl AcadSci USA, 95, 14152-14157.

4. Vologodskii, A.V. and Cozzarelli, N.R. (1994) Conformational and thermodynamic properties of supercoiled DNA. Annu Rev Biophys Biomol Struct, 23, 609-643.

164