Надмолекулярные комплексы липидов и ДНК: соотношение структуры, морфологии и активности тема автореферата и диссертации по физике, 01.04.07 ВАК РФ

>/ / Й- //313

г

Московский государственный университет им. М.В. Ломоносов, Физический факультет Кафедра физики полимеров и кристаллов

На правах рукописи

ЭЛКАДИ Ашраф Сайед Ибрахим

НАДМОЛЕКУЛЯРНЫЕ КОМПЛЕКСЫ ЛИПИДОВ И ДНК: СООТНОШЕНИЕ СТРУКТУРЫ, МОРФОЛОГИИ И АКТИВНОСТИ

Специальности 01.04.07 - физика конденсированного состояния и

03.00.02 - биофизика

Диссертации на соискание ученой степени кандидата физико-математических наук

Научные руководители: - Доктор физико-математических наук,

академик РАН, профессор ХОХЛОВ А.Р. - Доктор химических наук, профессор ЖДАНОВ Р.И. Консультант: - Доктор физико-математических наук

ЯМИНСКИЙ И.В.

Москва - 2003

Moscow State University Faculty of Physics Dept.Physics of Polymers and Crystals

STRUCTURAL AND FUNCTIONAL PECULIARITIES OF DNA-LIPID SUPRAMOLECULAR COMPLEXES

Thesis submitted by

as a requirement for obtaing Ph.D. degree in Specialities: Condensed Matter Physics (01.04.07) &

Molecular Biophysics (03.00.02)

Supervisors: Prof.Dr. A.R.Khokhlov Prof.Dr. R.I Zhdanov

Consultant: Dr.I.V.Yaminsky

Ashraf Sayed Ibrahim Elkady

2003

DEDICATION

To my parents

Contents

Introduction 7 Chapter I. Literature Review

1.1 Theoretical description of DNA 11

1.1.1 Statistical Thermodynamic Models and Molecular Dynamics 11

1.1.2 Polyelectrolyte Models 13

1.2 DNA condensation 22

1.3 Examples of DNA condensing agents 24

1.4 Morphology of condensed particles 28

1.5 Mechanism of condensation 30

1.6 Structure of condensed DNA 32

1.7 Intermolecular forces involved in condensation 34

1.8 Biological significance of condensation 37

1.9 DNA-lipid complexes as a non-viral gene delivery systems 38

CHAPTER II

MATERIALS AND METHODS

n.l. Materials and samples preparation 56

II.2. Physical and biophysical approaches 60

CHAPTER III

RESULTS AND DISCUSSIONS

III. 1 - Structure of DNA/Dicationic Lipid Complex as revealed by

Coherent Phase and Atomic Force Microscopy 72 III.2. The Structural-Activity Relationships for Lipoplexes

Prepared from a Newly Synthesized Dicationic Lipid and plasmid DNA 85

m.3. Tailoring DNA condenstaes using a dicationic lipid with different spacers 94

III.4. Structural and Morphological Peculiarities of DNA-Vitamin

D complexes: A Fluorescence and Atomic Force Microscopy Study 107

DI.5 A Spectroscopic and Atomic Force Microscopy Study for Oleic 115 Acid/DNAComplexes

Concluding remarks 129

References 132

Abbreviations

AFM atomic force microscope

CPM coherent phase microscope

CTAB cetyl trimethyl ammonium bromide

CL cationic lipid

DNA deoxyribonucleic acid

DLVO Dejaguin, Landau, Verwey, Overbeek

DCL dicationic lipid

DEGA novel dicationic lipid based on glutamic acid derivatives-(N'-

dimethyl-N'- dimethyl)-bis-(dihexadecylglutamate) butan or hexan

DOTAP 1, 2-diacyl-3trimethylammonium propane

DOPC dioleoyl phosphatidylcholine

DOPE dioleoyl phosphatidylethanolamine

FM Fluorescence microscope

GFP green fluorescence protein

SANS small angle neutron scattering

pDNA plasmid DNA

YD vitamin D

ACKNOWLEGEMENTS

I'd like to thank and acknowledge those whose encouragements and support were indispensable in the completion of this dissertation. With a feeling of accomplishment and honor, that I have had the privilege of working with a distinguished scientist over the last three years: Prof. Dr. Acad. A. R. Khokhlov, my supervisor who has given me wide latitude and freedom in the research problems that are presented here. I'd like also to express my gratitude to my co-supervisor Prof. Dr. R. I Zhdanov who was not only my co-supervisor but a kind friend as well. Their approach to science and research, I assume, will stay with me throughout my career. I thank Dr. I. V. Yaminsky who introduced me to the field of scanning force microscopy and gave me access to his labs. I also want to thank Dr. E. E. Makhaeva for guesting me in her laboratory and for her valiable advices and useful comments during the work; Dr. Tatiana Laptinskaya, for her patience and guidance during preparation of formal documents. I also benifitted from cooperation with Dr.Y.Livovich, Dr.S.Abramtchuk, Prof V. Tychinski, Dr. T. Vyshenskaya. Special thanks to Dr. Gerlinde Bischoff for the spectroscopic measurements and Alexei Moskovtsev for his assistance in experiments with cultured cells. I also acknowlege Dr.Y.Yermakov for surface potential measurements and for his thoughtful comments and explanations.

Introduction

1. Motivation for the wçrk

The last decade has witnessed a rapid and continuing development in the field of of polymer science which resulted in applications well beyond the scope of physics and chemistry, and in order to solve complex problems, this science has had to combine its efforts with many other sciences, e.g. molecular biology, biotechnology and nanomedicine.

Molecular self-assembly presents a 'bottom-up' approach to the fabrication of objects specified with nanometre precision. DNA molecular structures and intermolecular interactions are particularly amenable to the design and synthesis of complex molecular objects (1). Particularly, the self-assembly of amphiphilic molecules constitute one of the most fundamental mechanisms for the construction of soft condensed matter biomaterials (2). It is, however, well known that lipids, as well as mixtures of anionic and cationic single chain surfactants, can readily form bilayers (3, 4) that can adopt a variety of distinct geometric forms: they can fold into soft vesicles or random bilayers (the so-called sponge phase) or form ordered stacks of flat or undulating membranes

2. Aims and scope of the study

In the present work we investigated nucleic acids-lipid complexes, and their applications in vitro transfection and gene expression in human cancerous cells. We emphasized on the physico-chemical and biophysical characterization of the DNA-lipid complexes, and structure-activity relationships. The main goals of the work are:

1. The physico-chemical characterization of the complexes, and structure-activity relationships.

2. Studying the self-assembly of DNA with a newly synthesized dicationic lipid with different analogues (DEGA), as well as neutral lipids (vitamin D and oleic acid).

3. To bring these increasingly complex molecular systems within the scope of detailed structural studies and identify the effect of lipid structure on the interaction with DNA and morphology of the resulting self-assebled complexes.

4. Identification of the key parameters that are crucial for optimising gene transfer by correlating the structural features to transfection efficiencies.

3. Scientific findings

1. A novel approach for controlling the size and morphology of DNA condensates, using a dicationic lipid (DCL) with several different analogues, is introduced.

2. The potential of liposomes, prepared from the DCL, for delivering pEGFP-N1 DNA into eukaryotic cells was examined by monitoring gene expression in different cultured cell lines, e.g. MCF-7, SKOV-3 and 239 cells. It was found that lipoplexes composed from DCL with short spacers showed high transfection efficiency compared to those with long spacers.

3. To the first time, the self-assembly of phage T4 and plasmid DNA with vitamin D group, was examined in diluted aqueous solutions, in the absence and presence of Ca2+ cations, using Fluorescence and Atomic Force Microscopy. A simple method for preparing DNA-Vitamin D complex in the existence of divalent cations is introduced. The complex structure ranged from amorphous aggregates; beads-on-strings to compact single and multiple globules (0.1-1 (im), depending on vitamin nature, Ca2+ concentration and incubation time. A nucleation mechanism and flower-like aggregates are proposed as an initial state for complex formation. A new approach for targeted gene delivery (vitafection), using lipophilic vitamins (A, D, E, and K), is presented.

4. It was found that oleic acid shows molecular recognition to AT b.p. motifs by groove binding. GC tracts - in particular alternating d[G-C] motifs - are strongly influenced by ligand interaction up to a ratio of one molecule per two base pairs. The estimated amount of tightly bound oleic acid molecules is one molecule per 2-3 base pairs. As consequence, a new mechanism of

regulation of gene expression at nuclear membrane or by lipids inside DNA double helix is introduced.

Chapter I Literature Review

1.1 THEORETICAL DESCRIPTION OF DNA

Several different theories have been proposed to explain the behavior of DNA. They range from polymer and poly electrolyte theories to statistical mechanics and computer simulations. A good review is introduced in ref. (6). 1.1.1 Statistical Thermodynamic Models and Molecular Dynamics

Currently, molecular dynamics simulations are frequently used to describe polymer behavior. The movements of atoms are described by Newton dynamics in a potential energy surface determined by a particular model. Typical approarh is a grand canonical Monte Carlo simulation (7). Due to long computational times, oligonucleotides are normally modeled. The dodecane duplex ((dCCGCGAATTCGCG)2) is the most studied oligonucleotide, both theoretically and experimentally (NMR, X-ray scattering, other spectroscopic and diffraction techniques). Initial structure is often taken from crystallographic data, and its energy is minimized (relaxed) in a given potential field. In addition to spatial and temporal profile of the distribution of DNA and solvent atoms, interactions with ions, molecules or drugs are studied in such conditions. Additionally, molecular deformations, and from them bending and torsion

moduli can be determined (8). A typical result are stereoviews of a section of DNA with water molecules and counterions. These images of structure and ion and solvent distribution can be studied in time to elucidate dynamical properties of the system can be studied in time to elucidate dynamical properties of the system, such as torsional and angular flexibility, motion of solvent molecules and similar. Additionally, parameters such as translational and rotational diffusion coefficients, persistance length, or bending modulus can be modeled and compared to experimental and theoretical results. One has to be aware, however, that despite very long calculation times, for 5 bp oligonucleotide with several hundred water molecules and counterions almost a day of Cray time is needed (what corresponds to 6 months of calculation on a VAX machine), many assumptions have to be made in describing potentials and interactions (9). Topological properties of DNA, especially its supercoiling and knotting, are also very interesting. Theoretical models from biomolecular folding were used to investigate supercoiling of DNA and its knotting. Computer simulations find minimal energy conformations and can give insight into kinetics of their transformations. The knotting seems to be related to, collective bending and twisting modes of the molecule (10).

These methods, which treat the system as an ensemble of DNA, small ions and water molecules and all its complexity by Monte Carlo or molecular dynamics methods, therefore require extremely powerful computers and can be

useful to investigate fine structural details of DNA and its surroundings under fixed conditions. For calculating free energy of the system, however, such methods are not realistic at present. For instance, a conformational study showed that electrostatic effects which influence the B-Z form transition are of the order of hundredths of kT per base pair and obviously such accuracy cannot be obtained by first principles (11). For macroscopic description of these systems, therefore, thermodynamic models based on electrostatic interactions, which can be described by Poisson equation, are used and if proper approximations are used, give surprisingly, good agreements with experimental data.

1.1.2 Polyelectrolyte Models

Polyelectrolyte theories, one of which will be briefly presented below, were very popular to describe the structure and behavior of DNA in solution. The aim of such theories is to present a model of DNA based on molecular level assumptions and a priori to predict results of certaiain measurements, such as activity and osmotic coefficients of counterions, enthalpies of mixing and other thermodynamic parameters. It turned out that a rather simple concept, namely phase and conformational transitions of a linear, rigid (persistence length » Debye length) array of charges and surrounded by a cloud of counterions which relieve part of the electrostatic stress on the chain, as introduced by Manning (12), could describe properties with 30 to 50% accuracy, a rather good

agreement for its simplicity. More complex models, such as a cell model of poly-electrolytes, did not add further improvements, possibly due to the fact that some errors cancelled in the simple model. In contrast to these models which assume that electrostatic interactions are predominant and are the only ones taken into account, it is becoming clear that other interactions are also important. Realistic solution theory should encompass interactions of all ions with the solvent as well as van der Waals interactions and hydrogen bonding and hydrophobic effect. This brings about very complex models and because each system would have to be modeled separately we are not aware of general models that could be useful for describing DNA in solution more precisely.

As in other polyelectrolyte solutions, counterions that shield the linear array of negative charges with a valency z separated by d can bind site specifically or form an ionic atmosphere. None of the simple models include site specific binding and they treat counterion condensation as a free movement of counterions in a bound volume. Thermodynamic properties and interactions depend on this phenomenon and a dimensionless parameter ^ defined as

^ = z2e2/ekTd (1)

where z is the valence of counterion and e is solvent dielectric constant can describe counterion capture into ionic atmosphere. Manning showed that there is counterion condensation only for £ > 1. Interestingly, the amount of

condensation does not depend on the ionic strength. From DNA at 25 °C =0.714 nm and the fraction of condensed counterions can be calculated from

0=l-l/4/z

(2)

DNA in B conformation has 2 charges per 0.34 nm; hence d = 0.17 nm, % = 4.2 and 0 = 0.76 which means that 76% of the counterions form an ionic cloud in the case of a univalent counterion. Furthermore, the net charge, 1- 0, equals d/7.14 indicating that there is one net charge per 0.714 nm. We can substitute this net charge with e in the Poisson-Boltzmann equation and calculate electrostatic free energy and use it for further evaluation of the solution behavior of DNA. This is the so-called weakly charged polyion approach and Debye-

Huckel approximation can be used because \y/\ << 1 (in kT units). Potential at a

distance r from the polyion with radius can be calculated from

\|/"(r) + (l/r)lj/'(r) = K2 y/ (r)

(3)

j

where k~ is Debye length, rD, which can be expressed by Bjerrum length lB = e

2

/ s kT as

K 1 = rD = (871 lB Co)"1/2 = (6 kT/ e2 871 cQ)

-1/2

(4)

and the boundary conditions are

\\J ' (a) = 2 £, /a and y/ (R) = 0

(5)

The solution of the equation is

i|/ (r) = -2 £ (Ko (kt)/K, (^a) (6)

where Km(x) is a modified Bessel function (11).

From the known potential we can calculate electrostatic free energy and a relatively simple result can be obtained (12):

Gei = -Rg T/ 5 . ln[ 1 - exp(-Kd)] (7)

and for diluted solutions, assuming that Kd <S1

Gei = -Rg T/ . In (-Kd) (8)

Debye length is defined in univalent electrolyte solution as

k2 = 8ti 10"3 LA(e2/e kT) I (9)

here I is ionic strength and L Avogadro's number and can be approximated by

k= 3.3 (I)l/2 nm"1 (10)

Which gives 0.95 nm in biological fluids and Kd = 0.88.

This model assumes only atmospheric binding of counterions. Site specific binding is not taken into account. Electrostatic free energy would to compensate all the charges and that would lead to complete neutralization which would cause spontaneous bending, folding and condensation (self-collapse, "collapsing upon itself') of DNA. That would the hypothetical case at absolute zero temperature. A term which opposes this is entropy and free energy of such system is described

a sum of electrostatic and mixing entropy terms, which describes counterions and water. The first term therefore drives to complete binding and the last one to complete dissociation; the balance between the two determines the actual fraction of counterion condensation. Entropic conatributions to free energy include mixing of free cations, bound cations, solvent molecules and can be expressed as

Gmix = ein(ioooe/cV) (ii)

Where G is a fraction of ions associated with each charge, c concentration ounterions, and V the volume of the region surrounding the DNA in which cations are bound (12).

According to the condensation model, the driving force for DNA behavior in solution is the counterion diffusion potential which, due to electrostatic attraction, cannot diffuse freely. DNA tends to relieve its stress originating in the repulsion of phosphate groups and this is accomplished by the counterion atmosphere. From Equation 2 we can see higher valence ions are much more effective in relieving this stress, to stronger interaction and smaller loss in entropy upon neutralizing 1 charge. When more than 90% of the negative charge is compensated, can spontaneously condense. This model therefore does not distinguish between various polyvalent cations, polycations or positively charged polyelectrolytes. It does not assume dehydration upon binding and therefore

DNA condensates produced upon any of the above agents should be similar and redissociable upon increasing salt concentration above 1.2 M NaCI what was indeed observed experimentally.

Using this theory, we can get more insight in the local geometry, melting and other properties of DNA. Recent developments in the theoretical treatments of electrostatic interactions in macromolecular solutions added to charge-charge interactions also terms due to charge-solvent interactions. However, due to complexity and nonuniversality of such models, it seems that there is not much interest in their development. This theory, however, cannot account for the differential effects of counterions. For instance, the affinity of DNA for divalent cations decreases in the order Ca > Mg > Co > Mn and obviously other effects important. These include geometric terms in the Poisson-Boltzmann equation as well as non-electrostatic effects, including steric interactions, hydrophobic effect and other entropic contributions.

It was shown however, that Debye-Huckel approximation is not good enough for DNA and that better agreement with experiments and numerical solutions can be obtained by using more rigorous electrostatic approach (11).

A more sophisticated theory for the description of polyelectrolyte behavior was described by the cell model (13). However, it was found to be applicable mostly in dilute solutions of rigid polymers and not for the solutions of DNA. However, one can use the electrostatic potential around the rigid rod to describe

electrostatic behavior of short rodlike segments of DNA, by using the Poisson-Boltzmann equation, which describes electric potential around charge(s) at a distance r

A if/ = -(471 e0/e) X n z exp [-z e0 \\J (r)/kT] (12)

where A is Laplace operator. While for flat surfaces one can use A i|/ = d2 V|//dx2 and for spherical symmetry one can express the Laplace operator in

spherical coordinates, for rodlike symmetry the introduction of cylindrical coordinates yields

(l/r)d/dr (r d \\f /dr) = -(4 n eQ /e) I n z exp [-z e0 \\i (r)/kT] (13)

and using boundary conditions (potential is zero at r = and equals surface charge at r = a, where a is the radius of the polymer rod) one gets the following solution (14, 15)

\\f = -kT/ e0 In [2(32 /k cos2 ( (3 In (r/a) - C)] (14)

where the integration constants (3 and C are given as

^ = z2 e2 / e kTd = 1+ (32/(l+ (3ctg (3 y) (15)

C=arctg(4-l)/p, (16)

y = ln(R/a) (17)

where R is the distance where the potential from the polyion fades away. This boundary condition is not well denned at higher concentrations neighboring molecule is closer than R) and therefore the deviations of experimental data and theoretical predictions may be above 20 to 30% t higher concentrations and at lower ionic strengths.

For low ionic strengths Poisson—Boltzmann equation can be written as: (r) + (1/r) v|/' (r) = k2 sh i|/ (r) (18)

and the boundary conditions aTe \|; ' (a) = 2 £/a and \\j (R) = 0. The solution can be approximated for low ionic strengths as

xy(a) = 2^1n(^a), 5<1 (19)

v|/(a) - 2 ln($ a) - ln[4(^-1 )2 a] , 1 (20)

and it was shown that at C0 = 0.1 M the potential is correct within 5% as compared to the numerical calculation of Equation 15.

A geometric parameter appears in these equations through the parameter % which is the dimensionless charge per unit length. Note that = /B /d, where d is the separation between charges and that Debye length contains /B. For instance, in studies of B-Z configuration transition one has 2d = 0.34 nm and correspondingly at room temperature = 4.2 for the B conformation and 2d =

0.37 nm and = 3.9 for the Z form. For the absence of salt Katchalsky et al. (1976) (13) derived equation

\j/"(r) + (1/r) v|/'(r) =[(l/r)d/dr (r d v|/ /dr)] = -0.5 p2 e v(r) (21)

where parameter p is proportional to the outer radius R„ of the cell where the potential is zero (11). This equation can be solved analytically and potential decays as a logarithm of distance

\Hr) = ln[C'(lBc+)2r2 cos2(C"lnr)] (22)

where C and C" are integration constants which can be obtained from the boundary conditions (13).

Currently it is accepted that Debye Huckel approximation, despite its attractive simplification, is too simplified and that the distinction between condensed and noncondensed counterions does not reflect physical reality. Because Poisson-Boltzmann equation also contains ad hoc assumptions, a more rigorous theory was developed, which showed that Poisson-Boltzmann equation is a much more accurate approach than Debye Huckel approximation and condensation approach used by Manning. While estimates can still be made by using the condensation theory, Poisson Boltzmann formalism is recommended for better precision. Detailed analysis of more delicate processes, such as helix-coil or B-Z transition show that the condensation theory yields even erroneous

predictions (13). In conclusion, one can use condensation model for rough estimates while for more detailed analyses more rigorous treatments are required.

Empirical mechanical models would, in contrast to statistical mechanical models, define certain deformations, such as bending and torsion and define measurable constants, such as bending, elasticity, and torsion modulus to characterize minimal energy states of DNA in diluted solution. Experimentally, the elasticity of a single DNA molecule was measured in molecules attached with one end onto a cover slip by photodigoxigenin bond and at the other end had attached a magnetic bead via biotin streptavidin bond. A magnetic field was used to rotate and pull the beads and forces around piconewtons were used. It was shown that the elastic behavior is important in transcription and replication as well as in theoretical description of twisting and bending contributions which may not be harmonic and isotropic (16). Similarly, nowadays DNA is being manipulated with optical tweezers, lasers, attached to optical fibers and its mechanical properties, such as bending, stretching or unwinding can be studied.

1.2 DNA condensation

DNA condensation has become a lively area for research in diverse areas of science. In biochemistry, biophysics, and molecular biology it represents a process by which the genetic information is packaged and protected. In polymer physics and condensed matter physics, it presents intriguing problems of phase

transitions, liquid crystal behavior, and polyelectrolytes. Besides, in biotechnology and medicine, it provides a promising means whereby DNA containing genes of therapeutic interest can be prepared for transfer from solution to target cells for gene therapy applications.

DNA condensation could be defined as the collapse of extended DNA chains into compact, orderly particles containing only one or a few molecules. The decrease in size of the DNA domain is striking, as is the characteristic toroidal morphology of the condensed particle, so the phenomenon of DNA condensation has drawn considerable attention.

Genomic DNA is a very long molecule, which must fit into a very small space inside a cell or virus particle. Fully extended, the 160,000 base pairs of T4 phage DNA span 54 jim. Yet the T4 DNA molecule has to fit in a virus capsid about 100 nm in diameter, a 540-fold linear compression. The 4.2 million base pairs of the E. coli chromosome extend 1.4 mm (half this as a fully-stretched circular chromosome), yet must fit into a nucleolar region about 1 (im across, a linear compression of about 1400. Thinking about the issue in a different way, the radius of gyration of the T4 wormlike coil is about 950 nm, so its volume compression ratio is about 6900. The molecular volume of the T4 DNA molecule, considered as a cylinder 2 nm in diameter phosphate-to-phosphate, is 1.7 x 105 nm3; the internal volume of the capsid is about 5 x 105 nm3, so the DNA occupies about 1/3 of the volume within the capsid. If a 0.3 nm shell of

water is added to the DNA surface, the fractional volume occupancy rises to about 1/2 (17)

Considering the obvious energetic barriers to such tight packaging - the loss of configurational entropy of the long DNA molecule, the tight bending of the stiff double helix, the electrostatic repulsion of the negatively charged DNA phosphates - it is no surprise that organisms expend considerable metabolic energy to accomplish the task. Recent estimates run about 1/2 ATP hydrolyzed per base pair packaged (18, 19).

Thus it is a considerable surprise that DNA collapse, or condensation, can occur spontaneously in the test tube, upon adding a low concentration of multivalent cation to low ionic strength aqueous buffer (20). Even more surprising, the morphology of the condensed DNA particles is most commonly that of a compact, orderly toroid, strongly reminiscent of the structure of intraphage DNA gently lysed from virus capsids (21). X-ray scattering shows that the surface-to-surface spacing between DNA helices is only about 1-2 water diameters (22), so that the packing density of DNA condensed in vitro is entirely comparable to that of intraphage DNA.

1.3 Examples of DNA condensing agents

Chemical agents cause condensation by modifying electrostatic interactions

between DNA segments, by modifying DNA-solvent interactions, by excluding

volume to the worm-like coil, by causing localized bending or distortion of helical structure, or by some combination of these effects.

1.3.1 Multivalent cations

Early studies indicated that in aqueous solutions at room temperature, a cation valence of +3 or greater is necessary to cause condensation (e.g. the naturally occurring polyamines spermidine and spermine and the inorganic cation Co(NH3)6 3+ , polylysine, and basic histones). However, recent results (23) show that Mn can produce toroidal condensates of supercoiled plasmid DNA, but not of linearized plasmid. Supercolling appears to aid Mn2+ in stabilizing helix distortions, and also provides a 'pressure' that enhances the side-by-side association of DNA segments, an effect also observed with Mg2+ (24, 25). High concentrations of divalent transition metals cause the aggregation of linear DNA, but not into ordered condensates (26, 27). In the diaminoalkane series NH3+ (CH2)n NH3+ (n=l-6), compounds with n=3 and n = 5 cause compaction of single T4 DNA molecules, but those with n=2, n=4, and n=6 do not (32), indicating that linker length and hydrophobicity, as well as charge play a role.

1.3.2 Alcohol

Ethanol (80%) is commonly used to precipitate DNA, but as little as 15-20% ethanol will cause condensation to toroids or rods if Co(NH3)63+ is also added to a solution at low ionic strength (29). Methanol and isopropanol behave similarly.

o 0 0 0 0 0

Figure 1. Condensation of DNA. Typical condensed structures are

rodlike and toroidal particles. They are formed upon the action of various

i

different condensing agents, such as Co ions, spermine, spermidine, and others. In some other cases, DNA may not form regular condensates. Such condensates may structurally change in time while toroids and rods can precipitate. (Drawing courtesy of Stan Hansen.)

1.3.3 Basic proteins

A variety of basic proteins can produce toroidal or rod-like condensates. Four different proteins (sea urchin historic HI, sea cucumber histone ft), chicken erythrocyte histone H5, and clupeine) have little effect on the size or morphology of condensates with DNAs of various lengths (30). Transition protein TP2, which is involved in chromatin condensation, shows GC-rich sequence preference and is zinc dependent [31].

1.3.4 Neutral polymers

Even neutral polymers such as PEG, at high concentrations and in the presence of adequate concentrations of salt, can provoke DNA condensation through an excluded