Clustered damage in DNA: implicat ions, simulation, and detection.

 DNA damage is caused by the reactive sp ecies produced by the passage of a charged particle.

 Modeling is the only way to com p letely characteri z e the co mpl e xity of the dam a ge including the stochastic nature of the track and its effects on damage distri buti o ns.

 Particle tracks superim posed on a m o de l of linear DNA indicate that damage may occur in “clusters”.

 Such clusters of “locally m u ltiply da m a ged sites” will vary depending on the statistical nature of the track, the random orientation of the track with respect to the DNA, and the “co m pact ness” of the DNA.

 The biological consequences of cl ustered dam a ge ma y be serious. Additional damaged sites near a DS B may interfere with DNA repair enzy mes .

 Monte Carlo techniques are available to address the statistical nature of:

o the location of energy depositi on sites on the DNA,

o the am ount of energy deposition,

o diffusion of water radicals.

Descriptions of Energy Deposition

[Image removed due to copyright considerations]

[Goodhead, 1987]

As LET i n creases , dsbs become “more lethal”

[Image removed due to copyright considerations]

“Simple” Double Strand Break

[Image removed due to copyright considerations]

[Goodhead, 1994]

“Complex” Damage

[Image removed due to copyright considerations]

[Goodhead, 1994]

Track st ructure an alysis: implications of clustered damage .

 High-LET effects are qualitatively and quantitatively different than low- LET.

 All differences bet w een high- and low-LET r a diations arise from track structures.

 Radiation produces a spectrum of damage of varying com p lexity.

 The com p lexity depends on the clustering of a track’s ionizations in or very near to the DNA.

 Clustered damage occurs in DNA at sufficiently high frequency to be biologically signi ficant.

 Som e of the damage m a y be totally unrepairable.

 The spectrum of damage after repair, or attempted repair , is responsi b l e for the permanent cel lular effect s.

 The spectrum of permanent damage may be sk ew e d toward s the m o re severe com ponents of the initi al damage.

 Assays that m e asure initial damage may be m i sleading.

So w h at kills cells?

The role o f single and double strand break s in cell k illing.

How is sensitivity to killing by radiation related to DNA damage?

 Ionizing ra diation produces a plethor a of damages: including 1000 SSBs, 25- 40 DSBs per cell per Gy.

 Variation in intrinsic radiosensitivity relates to the different repair capability of various cell lines.

Do assays to detect DNA damage measure the “important“ damage?

 Sensitive methods now exist to me asure SSBs, DSBs, initial yields, and their rejoining in individual cells, chromatin regions, or individual genes.

Misrejoining

 Unrejoined or m i srejoined DSBs caus e chrom o some breaks or m i cronucleus form ation.

 These lesions correlate with cell killing.

[Image removed due to copyright considerations]

 Measurement m e tho dology is critical: su fficient time for the lesion to form , not too m u ch tim e for the cell to die (apoptosis, or necrosis).

 Methodology has been developed to m e as ure the initial yields of DSB, and the fractions rejoined correctly and incorrectly.

Complex lesions: a hallmark of ionizing radiat ion damage?

 Shown to be an im portant hall mark of radiation effects.

 More difficult to repair.

 More likely to lead to cell deat h, m u tation or transform a tion.

 Co m p lex dam a ge is related to LET and the density of ion pairs produced at track ends.

 Explains observation that DSBs are more l e thal for high-LET radiation.

 Illust rates that m e thods to measure DSBs cannot dist ingui sh the “severity” of the DSB.

How does a cell de fine DNA damage?

 Radiosensitive cells may see b o th clos ely opposed and m o re widely spaced SSBs as DSBs.

 Radioresistant cells may rep a ir the wide l y spaced l e sions as two independent SSBs.

 The rep a ir processes will be different, maybe the more co m p lex repair o f a DSB is more error-prone.

Does an assay define a DSB differently than a cell?

 An assay may “interpret” a DS B differently than a cell.

 Closely opposed versus widely spaced l e sions handle d differently in different assays.

DSB rejoining kineti cs may hold clues to the relevant DNA damage.

 Rejoining of DSB shows both fast and slow com ponents.

 The fast com ponent may be : sim p le lesions that ar e m o re readily rejoined, m o re accessible lesions, or SSBs.

 Data suggest that initial rate is correct rejoining, the slower com ponent is m i srejoining.

Is residual damage the key?

 If unrejoined DSBs are responsible fo r cell killing, then residual damage shoul d correlate with radiosensi tivity.

[Image removed due to copyright considerations]

One turn of the 30-nm chrom a tin fiber m odel is shown.

The 30-nm chromatin model

 A right-ha nded “super- solenoid” structure with 20 turns.

 Each turn is 11 nm long; the total length of the solenoid is 220 nm .

 One turn contains 6 helical nucleosomes.

 Each nucleosome consists of 171 base pairs of DNA wound in a left handed helix two full turns around a histone protein core: 146 bp wound around the histone and 25 bp as linker to the next nucleosome.

 The coordi nates of all atom s in the DNA sugars and bases are known.

 The com p lete m o del contains ~ 20.7 kbp.

 Two hydration layers of water are included;

o Tightly bound: 15 water m o lecules pe r nucleotide; participates in a charge transfer to the DNA. ( c harge transfer cross sections not known).

o Loosely bound: 18 water mo lecules per nucleotide; serves as a source of  OH.

 Model includes only the second hydrati on layer in the water radiation chem istry.

Calculations:

The chromatin is situated in an aqueous environm ent. (It is impossi ble to accurately m odel the co m p lete intracellular environment.)

A hydroxl radical scavenger is added to approxim a te the scavenging potential of the intracellular environm ent. The scavenger concentrati on is adjusted to give a mean  OH diffusi on length of about 3 nm .

Charged particle tracks are positioned at random orientation to the chromatin m odel and at random distances from the chrom a tin.

 The energy deposition events are si mulated,

 the generation of water radicals are followed,

 the wat e r r a dicals are allowed t o diffuse,

 each radical is followed until it either re acts with an other radical, a protein, or the DNA.

Direct effects are scored when DNA sites on th e chrom a tin fiber are close enough to be targets of direct energy depositi on events.

Indirect effects are scored when a hydroxyl radical origi n ating in a water mol ecule r e aches the DNA.

 Model the creation of the radicals by en ergy deposit ion events in the track and in the delta rays.

 The mean am ount of energy required to produce a hydroxyl radical is 17 eV.

 Up to 6 hydroxyl radicals can be creat ed in glancing collisions (100 eV cutoff for the glancing collisions).

 Each radical is track ed through m i grati ons in a series of random - walk jum p s.

 Each  OH radical is tracked until it meets one of 4 fates:

o Recom b ination with another radical

o Scavenged by another m o lecule in the nucleus

o Diffusion i n to a region of hist one protein

o Reaction with a DNA sugar or base

 Spherical reaction radii are defined for the DNA bases and the sugars DNA damage is scored from di rect and indirect effects.

DSBs are scored when SSB s occur on opposi t e strands withi n 10 bp of each other.

Bet ween 10 4 and 10 7 tracks are scored for each particular energy and Z.

Track St ructure

Two types of particle interactions are used: glancing collisions and “ k n ock - on”

or close co llisions.

 Glancing collisions generate a core to the track; the radius of which is dependent on the particle velocity , but not the charge.

 100 eV is selected as the cutoff betwee n the glancing collisions and the close collisions.

 The close collisions are further divide d as those involving energy transfe r events of 100 ev – 2 keV, and those > 2 keV.

 Each delta ray is m o deled explic itly as low-LET radiation acting independently of t h e origi n al particle track.

 All interactions are tracked and DNA damage is scored.

Approximate scale: 0.01 µ m

[Image removed due to copyright considerations]

Clustered Damage

Damage from three p a rticles was sim u lated:

 10 MeV/n helium : LET = 18.7 keV/  m

 30 MeV/n nitrogen: LET = 93 keV/  m

 10 MeV/n iron: LET = 2,470 keV/  m.

The iron particles show clear signs of si gnificant damage clustering. The intermediate LET nitrogen par ticle shows pot en tial clustering. The helium ion track shows a rela tively sparse damage pattern.

Damage clustering depends to a large exte nt on the track ionization densi t y, but also on the orientation of the track with respect to the chro matin.

[Image removed due to copyright considerations]

Mapping of the damage

 The entire linear sequence is known.

 Any sequence with a DSB (two SSBs w ithin 10 bp) is exam ined for the presence of additiona l breaks, or base damage sites, nearby (i.e., clustering of the damage).

 The degree of damage clustering is ev ident when projected onto linear maps of the damage sites.

[Image removed due to copyright considerations]

Inspection of linear sequences on a larger scale to look for patt erns of damage.

The relative proporti on of direct vs. indirect damage events is relatively constant over a wid e range of LET. All of these ar e considered high-LET.

[Image removed due to copyright considerations]

Prediction of fragment length distribution patterns

 The entire sequence is known.

 The locations of all DSBs are known.

 Fragment lengths can be calculated and fragment len g th distributions predicted.

The prediction of fragments of ~80 bp and ~1000 bp was new and related to the structure of the chromatin mod e l.

Interpreted as one turn around t h e hi stone and one turn around the 30-nm chromatin solenoid, respectively.

Extension of the fragment prediction to genomic DNA

[Image removed due to copyright considerations]

Prediction of fragm ent yields i n the 100 bp to 2 kbp range. Actual measurem ents show go od agreement (Rydberg, 1996).

Technically difficult : m u st go to larg e num bers of cells, and a large dose.

Summary:

The si mul a tion of track effects in the 30-nm chromatin m odel produced 3 previously unknown results.

 Prediction of small fragm ents with non-random fragment size distri buti o n that reflect s the structural feat ures of the chrom a tin m odel.

o Standard DSB assays do not measure s m all frag m ents.

o If the smaller fragm ents are unacc ounted for, RBEs for initial DSB will be underestimated .

 The model agrees wit h the general c oncept of clustered damage (“locally m u ltiply damaged sites”) and shows va rious degrees of clustering up to 40 bp.

 Multiple local clusters are pred icted over extended regions of up to several kbp. New term coined to de scribe this observation: “regionally multiply damaged sites” .

o Fragments in the intermediate size range were not m odeled but could result from higher order chromatin struct ure, i.e., loops attached to the nuclear matrix.

Biological im plications

 If small frag m ents are lost, DNA inform ation will be co m p ro m i sed.

 Clustered dam a ge is probabl y harder to repair, and a m o re re levant m e tric of biological effects.

Objective: development of methods to t e st th e Holley and Chatterjee predictions of non-random generation of small DSB fragments.

Approach:

 Label cells (GM38 primary human fibrobla s ts) with [ 3 H] thy m idine 4 days before the experiment. Ch ange to fresh, unlabeled me dium 24 hrs before the experiment.

 Harvest cells when cl ose to conf luency, i.e., > 90% are in the G 0 stage of the cell cycle.

 Em bed labeled cells in agarose plugs.

 Expose plugs to x rays or high-LET radiation at 0º (or 4ºC).

 Lyse the cells in the agarose plugs.

 Gel electrophoresis under co nditions designed to sepa rate fragments in the 0.1-2 kb si ze range.

 Electrophoresis under neutral cond itions to m e asure DSBs, and under alkaline co nditions to measure SSBs.

 Correlate the counts with fragmen t size, express as fragment size distri buti o n.

X rays

DNA size markers included in the gel are indicated by the arrows.

The gel was sliced and counted.

Increase in short fragments as a function of dose was linear, and greater than expected from random breakage.

[Image removed due to copyright considerations]

Sensitivity of the m e thod lim ited by counts per 3 x 10 6 cells in these small f r a g m e n t s . C o u n t r a t e s a r e about 2 x background.

Total counts in small fragments ~ 1 x 10 -4 of the total radioactivity in the DNA.

The peak at 0.5 kbp i s of unknown origi n and was present in the unirradiated cells.

High-LET particles

The total y i eld of fragments is greater than for x rays.

[Image removed due to copyright considerations]

[Image removed due to copyright considerations]

(a) Primary data is fragm ent weight distribu tion. This is proportional to both the num ber of fragments and the span of sizes in the gel slice.

Weight distributions were converted to num ber distri butions.

(b) Theoretical distribution of frag ments from 30 MeV/n nitrogen (from Holley and Chatterjee. 1996).

The measured and theoretical d i stribution patterns appear different .

[Image removed due to copyright considerations]

Total yield of fragments shows good agre em ent between measurements and theory.

[Image removed due to copyright considerations]

[Image removed due to copyright considerations]

[Image removed due to copyright considerations]

“Zigzag” model of DNA winding on the nucleosomes in the chromatin fiber produces a m u ch better fit between theory and experiment.

Heavy charged particle track interacting with the 30 nm solenoid fiber.

Damaged sugars and bases represented by pink and yellow spheres, respectivel y.

[Image removed due to copyright considerations]

[Image removed due to copyright considerations]