nihms91589.pdf

Recognition of Abasic Sites and Single Base Bulges in DNA by a

Metalloinsertor

†

Brian M. Zeglis

Jennifer A. Boland

, and

Jacqueline K. Barton

Division of Chemistry and Chemical Engineering, California Institute of Technology, Pasadena, CA

91125

Abstract

Abasic sites and single base bulges are thermodynamically destabilizing DNA defects that can lead

to cancerous transformations if left unrepaired by the cell. Here we discuss the binding properties

with abasic sites and single base bulges of Rh(bpy)

(chrysi)

, a complex previously shown to bind

thermodynamically destabilized mismatch sites via metalloinsertion. Photocleavage experiments

show that Rh(bpy)

(chrysi)

selectively binds abasic sites with affinities of 1-4 × 10

-1

; specific

binding is independent of unpaired base identity but is somewhat contingent on sequence context.

Single base bulges are also selectively bound and cleaved, but in this case, the association constants

are significantly lower (~10

-1

), and the binding is dependent both on unpaired base identity and

bulge sequence context. A wide variety of evidence, including strand scission asymmetry, binding

enantiospecificity, and MALDI-TOF cleavage fragment analysis, suggests that Rh(bpy)

(chrysi)

binds abasic sites, like mismatches, through insertion of the bulky chrysi ligand into the base pair

stack from the minor groove side and ejection of the unpaired base. At single base bulge sites, a

similar, though not identical, metalloinsertion mode is suggested. The recognition of abasic sites and

single base bulges with bulky metalloinsertors holds promise for diagnostic and therapeutic

applications.

Genomic integrity is of paramount importance to cellular survival and replication. However,

a wide variety of agents, ranging from genotoxic chemicals to error-prone cellular processes,

render DNA dangerously susceptible to damage and mutation.

The types of DNA defects are

as varied as their causative agents, yet the most common forms are single base mismatches,

abasic sites, single base bulges, and oxidized bases. Left unrepaired, all of these defects can

lead to deleterious mutations, often in the form of single nucleotide polymorphisms. To counter

these threats, the cell has evolved complex DNA repair machineries, most notably the mismatch

repair (MMR) and base excision repair (BER) pathways.

Under normal conditions, the

MMR (mismatches and single base bulges) and BER (abasic sites and oxidized bases)

machineries will quickly and efficiently repair their target defects, thereby preventing any

lasting damage to the cell or its genome. However, the suppression or disabling of these

pathways is often met with dire consequences: mismatch repair deficiency, for example, has

been implicated in 80% of hereditary non-polyposis colon cancers in addition to significant

percentages of breast, ovarian, and skin cancers.

It thus becomes clear that the synthesis and

study of molecules able to specifically target these defects may aid in the development of new

cancer diagnostics and therapeutics.

The design and application of metal complexes capable of specifically targeting one such

defect, single base mismatches, have been focuses of our laboratory for over a decade.

These

†

Financial support for this work from the National Institutes of Health (GM33309 to J.K.B.) is gratefully acknowledged.

*To whom correspondence should be addressed. Email: jkbarton@caltech.edu. Telephone: (626) 395-6075. Fax: (626) 577-4976.

NIH Public Access

Author Manuscript

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

Published in final edited form as:

Biochemistry

. 2009 February 10; 48(5): 839–849. doi:10.1021/bi801885w.

NIH-PA Author Manuscript

metal complexes, most notably Rh(bpy)

(chrysi)

(chrysi = chrysene-5,6-quinone diimine)

and Rh(bpy)

(phzi)

(phzi = benzo[a]phenazine-5,6-quinone diimine) (Figure 1), bear

sterically bulky ligands that are too wide to fit between matched base pairs and thus instead

preferentially target thermodynamically destabilized mismatched sites.

The compounds are

highly specific (

≥

1000-fold) for mispaired sites over matched base pairs and recognize over

80% of mismatches in all possible sequence contexts, with only thermodynamically stable, G-

containing mismatches escaping binding altogether.

Furthermore, the complexes can, upon

irradiation with ultraviolet light, promote direct cleavage of the DNA backbone at the binding

site. More recently, crystallography and NMR studies have revealed that these complexes do

not bind via classical intercalation, in which the complex binds from the major groove and

increases the base pair rise by stacking an aromatic ligand between intact base pairs. Rather,

they employ a unique binding mode that we have termed metalloinsertion: the complex binds

the DNA from the minor groove, ejecting the mismatched bases into the major groove and

replacing them in the base stack with the sterically expansive aromatic ligand.

These

structural data make quite clear the origin of the correlation between recognition and

thermodynamic destabilization: the less stable the mismatch, the easier the ejection of the

mismatched bases.

Yet mismatches are not the only destabilizing DNA defect. Indeed, the relationship between

thermodynamic instability and specific metalloinsertor binding has led our laboratory to

investigate the recognition of two different DNA defects: abasic sites and single base bulges.

Abasic sites arise from the cleavage of the glycosidic bond between the ribose and the

nucleobase; this can occur spontaneously, as a result of exogenous agents, or as an intermediate

in the BER pathway (Figure 2).

In the cell, abasic sites exist primarily in equilibrium between

two hemiacetal anomers; just as important to the structure of the site, however, is the unpaired

base complementary to the abasic site. Numerous structural studies have shown that the

conformation of this unpaired base can be extra- or intrahelical depending upon its identity and

the surrounding bases.

Unpaired purines are almost always intrahelical, whereas unpaired

pyrimidines likely exist in equilibrium between extrahelical and intrahelical forms, with the

extrahelical form favored when the base is flanked by other pyrimidines. Relative to intact

duplex DNA, abasic sites are thermodynamically destabilized by 3-11 kcal/mol.

Both the

sequence context and the identity of the unpaired base play roles in the magnitude of the

destabilization: sites in which the abasic ribose is flanked by purines are more stable than those

flanked by pyrimidines, and, to a lesser degree, sites with unpaired purines are more stable

than those with unpaired pyrimidines.

Single base bulges are defects in which a base is inserted in one strand of an otherwise well

matched duplex. Caused by errors in recombination and replication, these sites are more

thermodynamically stable than abasic sites, with destabilizations ranging from 0-3 kcal/mol.

Recent computational and spectroscopic studies have shown that while bulged base identity

and sequence context certainly influence the destabilization of the site, reliable patterns such

as those for abasic sites do not exist.

Several structural studies have shown that the unpaired

base may be intra- or extrahelical.

Similar to the case for abasic sites, unpaired purines

are almost always intrahelical, whereas an equilibrium between intra- and extrahelical

conformations is likely for unpaired pyrimidines. Unpaired bases flanked by purines are more

likely to remain intrahelical than those surrounded by pyrimidines. Regardless of unpaired base

helicity, all duplexes with single base bulges are bent relative to well-matched DNA.

Under normal conditions, abasic sites and single base bulges are repaired through the BER and

MMR pathways, respectively. However, if left unrepaired, both lesions represent significant

threats to cell viability: abasic sites can lead to single nucleotide polymorphisms, block

transcription and replication, and act as a potent topoisomerase poison,

while single base

Zeglis et al.

Page 2

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

bulges are a common source of frame-shift mutations.

Indeed, deficiency in the repair of

both types of lesions has been associated with several different cancers.

Given the links to cancer, it is not surprising that agents that recognize these lesions have

already been designed and studied. Methodologies for the targeting of abasic sites include

organic bis-intercalators

and nucleophilic amines

that react with the minor aldehylic form

of the natural abasic site. Organic agents have also been designed for single base bulge

recognition.

Other recognition agents resemble metalloinsertors; dinuclear ruthenium

and octahedral cobalt

compounds have been shown to bind multiple base bulges along with

DNA hairpins. Yet despite some successes, almost all of these recognition agents exhibit

affinities, specificities, or reactivities that are less than ideal for diagnostic or therapeutic

applications.

Our investigation of metalloinsertors for abasic site and single base bulge recognition is thus

motivated both by the desire to augment our understanding of DNA lesion recognition by metal

complexes and by the opportunity to create a useful diagnostic reagent for the detection of

these deleterious defects. We have previously communicated our initial finding that bulky

metalloinsertors specifically bind and photocleave abasic sites and single base bulges.

Here,

we present a more comprehensive study geared at elucidating the scope and means of

recognition at both types of defects.

EXPERIMENTAL

Reagents, instrumentation, and general methods

All reagents were the highest purity commercially available and, unless otherwise noted, were

used as obtained without further purification. Rh(bpy)

(chrysi)

and Rh(bpy)

(phzi)

were

synthesized and purified as previously reported.

The enantiomers of Rh(bpy)

(chrysi)

were

likewise resolved as described earlier. Standard oligonucleotides were synthesized from

phosphoramidites on an ABI 3400 DNA synthesizer (reagents from Glen Research). Given the

instability of the natural hemiacetal abasic lesion, the often employed tetrahydrofuranyl abasic

site analogue was used instead.

In all text, the symbol

denotes the abasic site. Abasic site-

containing oligonucleotides were ordered from Integrated DNA Technologies. Following

synthesis or delivery, the oligonucleotides were purified both with and without dimethoxytrityl

(DMT) protecting groups via reverse phase HPLC [HP1100 HPLC system with Varian

DynaMax™ C18 semi-preparative column, gradient of 5:95 to 45:55 MeCN:50 mM NH

OAc

(aq) over 30 min for DMT-on purification and 2:98 to 17:83 MeCN:50 mM NH

OAc (aq) over

30 min for DMT-off purification]. UV-Vis absorption spectra were taken on a Beckman DU

7400 spectrophotometer.

Metal complex concentrations were determined using UV-visible spectrophotometry with

extinction coefficients of

302

= 57,000 cm

-1

and

315

= 52,200 cm

-1

for Rh

(bpy)

(chrysi)

and

304

= 65,800 cm

-1

and

314

= 67,300 cm

-1

for Rh

(bpy)

(phzi)

. DNA strand concentrations were also determined spectrophotometrically using

base extinction coefficients of

260

= 15,400 cm

-1

(A),

260

= 7,400 cm

-1

(C),

260

11,500 cm

-1

(G), and

260

= 8,700 cm

-1

(T). DNA concentrations are presented per

strand. Duplex melting temperatures were determined by following hypochromicity at 260 nm

for 1

M duplex in a buffer of 50 mM NaCl, 10 mM NaPi, pH 7.1, via variable temperature

UV-Vis.

All oligonucleotid es were 5’-radioactively labeled with

P using [

P]ATP (MP

Biomedicals) and polynucleotide kinase (Roche) employing standard methodologies and

purified via 20% polyacrylamide gel electrophoresis (National Diagnostics).

All

photocleavage experiments were performed using end-labeled DNA with identical sequence,

Zeglis et al.

Page 3

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

unlabeled carrier DNA in a buffer of 50 mM NaCl, 10 mM NaPi, pH 7.1. Duplexes were

annealed by incubation at 90 °C for 15 min followed by slow cooling to room temperature.

Recognition and photocleavage experiments

Solutions of Rh(bpy)

(chrysi)

or Rh(bpy)

(phzi)

were incubated with 5’-

P-labeled

oligonucleotides either containing or lacking a central DNA lesion (see

Results

section for

sequence details). Unless otherwise noted, final solutions were prepared 20 minutes prior to

irradiation, contained 1

M duplex and 1

M metalloinsertor, and were 20

L in volume. Dark

and light control samples, of course, lacked the appropriate solution components. Because

metalloinsertor photocleavage is single-stranded, each duplex was interrogated twice, once

with each of the two strands radioactively labeled. Samples were irradiated with an Oriel

Instruments 1000 W solar simulator (320-440 nm). Irradiations were performed in open,

horizontally oriented 1.7 mL microcentrifuge tubes. After irradiation, samples were incubated

at 60 °C for 30 min and then dried under vacuum. Dried samples were redissolved in denaturing

formamide loading dye and electrophoresed on 20% denaturing polyacrylamide gels. Images

of the gels were obtained via phosphorimagery (Molecular Dynamics) and quantified using

ImageQuant software.

Determination of defect-specific binding constants

Photocleavage titrations were performed to determine the thermodynamic binding constants

for Rh(bpy)

(chrysi)

with lesion sites of interest. Solutions of DNA (1

M) were incubated

with variable concentrations of Rh(bpy)

(chrysi)

(0-20

M) and subsequently irradiated on

an Oriel Instruments solar simulator for 10 min. After irradiation, the samples were incubated

at 60 °C for 30 min and then dried under vacuum. Dried samples were redissolved in denaturing

formamide loading dye and electrophoresed on 20% denaturing polyacrylamide gels. Images

of the gels were obtained via phosphorimagery (Molecular Dynamics). The fraction cleaved

at the lesion site was quantitated using ImageQuant software, expressed as a fraction of the

total parent DNA, and fit to a single site, one parameter binding model.

MALDI-TOF cleavage product analysis

For mass spectrometry analysis of photocleavage products, 2

M solutions of duplex were

incubated with 2

M Rh(bpy)

(chrysi)

and irradiated as described above. After irradiation

and incubation, the samples were dried under vacuum, resuspended in 10

L water, and

desalted using 10

L OMIX C18 tips (Varian). Light and dark controls were also prepared.

Mass spectrometry was performed using a Voyager DE-PRO MALDI-TOF instrument with a

337 nm nitrogen laser source (Applied Biosystems). A 4-hydroxypicolinic acid matrix was

employed. All mass spectra were internally calibrated using the mass of the parent

oligonucleotide.

RESULTS

Sequence design and melting temperature analysis

A series of oligonucleotides was synthesized and purified to allow for the interrogation of

abasic sites and single base bulges in variable sequence contexts and with all possible unpaired

bases. The 27-mer single strands are identical except for a central six base region in which the

sequence variation occurs. Four different oligonucleotides containing abasic sites were

designed, each placing the abasic site in a different sequence context: 5’-G

T-3’ (AB1), 5’-

A-3’ (AB2), 5’-A

G-3’ (AB3), and 5’-T

C-3’ (AB4) (Table 1). For each abasic strand,

four complements were prepared. Each positions a different base complementary to the abasic

site: for example, 3’-C

A-5’ (AB1-A), 3’-C

T-5’ (AB2-C), 3’-T

C-5’ (AB3-G), and 3’-

G-5’ (AB4-T). These oligonucleotides, taken together, allow us to examine the recognition

Zeglis et al.

Page 4

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

of abasic sites in the three major sequence context types (Pur/Pur, Pyr/Pur, Pyr/Pyr) with all

possible opposing unpaired bases. For purposes of comparison, matched and mismatched

strands were also created for each sequence context; complementary in each case to the AB#-

C strand, these oligonucleotides create either a fully matched duplex or one containing a central

CC mismatch.

Four additional oligonucleotides were synthesized to facilitate the study of single base bulges

(Table 2). These, termed B1-B4, are identical to the AB# strands in all respects except that

they lack the tetrahydrofuranyl abasic site. Thus, when these 26mers are annealed to the

complements of the abasic oligonucleotides, duplexes with single base bulges are formed. In

each case, the nucleotide formerly complementary to the abasic site is now the bulged base:

for example, 3’-C

A-5’ (B1-T), 3’-C

T-5’ (B2-A), 3’-T

C-5’ (B3-C), and 3’-A

G-5’ (B4-

G). The same sets of matched and mismatched duplexes were employed as controls. In all, 32

oligonucleotides forming 28 unique duplexes were created.

Melting temperature analysis of the DNA allows us to determine the relative thermodynamic

destabilization created by the lesions. All four matched duplexes have melting temperatures

around 64 °C. Relative to these, the mismatched duplexes are destabilized by 7-8 °C. Duplexes

containing single base bulges are similarly destabilized, if not slightly more stable, with melting

temperatures 6-8 °C lower than that of the corresponding matched duplex. In contrast, abasic

site duplexes are even less stable than their mismatched counterparts with melting temperatures

reduced by 8-11 °C. Taken together, these

values are in agreement with the published

literature.

It is somewhat surprising, however, that in the family of abasic duplexes we do

not see significant variation in

based upon sequence context or unpaired base identity.

This result is more likely a product of instrument sensitivity rather than the absence of such

influences on site stability. Nonetheless, these measurements plainly illustrate the relative

stabilities of the duplexes at hand: abasic site < mismatched base pair < single base bulge

≪

matched base pair.

Recognition and photocleavage of abasic sites by Rh(bpy)

(chrysi)

Polyacrylamide gel assays clearly indicate that Rh(bpy)

(chrysi)

specifically recognizes and

photocleaves abasic sites in DNA (Figure 3). Indeed, the metalloinsertor binds and promotes

strand scission at lesion sites in all sequence context types (5’-Pur

Pur-3’, 5’-Pur

Pyr-3’, and

5’-Pyr

Pyr-3’) and with all possible unpaired bases. No photocleavage is observed in the

absence of metalloinsertor or with well matched DNA. In total, twelve of the sixteen abasic

sites are bound and cleaved. Specifically, duplexes AB1, AB2, and AB4 are recognized and

cleaved regardless of unpaired base identity; surprisingly, however, no photocleavage is

observed for the AB3 duplexes. This pattern corresponds precisely to that observed for the

strands bearing a central CC mismatch: AB1-MM, AB2-MM and AB4-MM are all recognized

and cleaved, while AB3-MM escapes binding and scission. That the AB3 duplexes are not

bound and cleaved is certainly not a consequence of the sequence context type, for AB2, like

AB3, places the abasic site in a 5’-Pur

Pur-3’ sequence context and is, in fact, cleaved quite

readily. The answer likely lies in the sensitivity of metalloinsertor mismatch recognition to

specific sequence contexts. Similar effects of sequence context have been seen previously for

the family of mismatched duplexes.

Indeed, experiments employing higher rhodium

concentrations and longer irradiation times suggest that Rh(bpy)

(chrysi)

does bind and

cleave the AB3 abasic sites, just not nearly as strongly or efficiently as those in the other

sequence contexts.

Photocleavage experiments also reveal interesting patterns in the strand asymmetry of scission.

Regardless of unpaired base identity, duplexes AB1 and AB2 are cleaved on the strand

containing the unpaired nucleotide. Interestingly, however, duplex AB4 is cleaved instead on

the strand containing the tetrahydrofuranyl abasic site, again irrespective of unpaired base. This

Zeglis et al.

Page 5

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

behavior exactly mirrors mismatch photocleavage. While, of course, the mismatched duplexes

contain no unpaired bases or abasic sites, the AB1-MM and AB2-MM assemblies are cleaved

on the strand corresponding to that containing an unpaired base in the abasic duplexes, and the

AB4-MM assembly is cleaved on the strand corresponding to that containing the abasic site in

the abasic duplex. This observation must reflect the binding architecture of the complex in the

abasic site (see

Discussion

Another important similarity between mismatch and abasic photocleavage is the length of the

scission products. Regardless of unpaired base identity, AB1 cleavage fragments are 14 base

pairs long, AB2 fragments 15 base pairs long, and AB4 fragments 13 base pairs long. These

fragments correspond to cleavage at the ribose 3’ to the unpaired base in duplexes AB1 and

AB2 and at the ribose 3’ to the abasic site in the AB4 duplexes. Importantly, photocleavage at

the CC mismatch in each duplex produces fragments of identical length.

Recognition and photocleavage of abasic sites by Rh(bpy)

(phzi)

In order to probe the generality of metalloinsertor recognition of abasic sites, photocleavage

experiments were also performed using Rh(bpy)

(phzi)

, a second generation complex with

a heterocyclic bulky ligand (Figure 4).

Rh(bpy)

(phzi)

is clearly able both to recognize and,

upon irradiation, to cleave the representative abasic sites. Again, no recognition or

photocleavage is observed in the absence of metalloinsertor or DNA lesion. Significantly,

photocleavage with Rh(bpy)

(phzi)

is observed at much lower concentrations (100 nM) than

with Rh(bpy)

(chrysi)

, a characteristic also observed for mismatch photocleavage and

attributed to the added

-stacking capabilities of the heterocyclic inserting ligand.

Binding affinities of Rh(bpy)

(chrysi)

for abasic sites

Photocleavage titration experiments were employed to determine site-specific binding

constants for the twelve abasic sites and three mismatches for which photocleavage was

observed (Figure 5 shows a representative titration, Table 1). The binding constants for the

mismatched sites, 2.2(2) × 10

-1

(AB1-MM), 1.7(2) × 10

-1

(AB2-MM), 2.5(3) × 10

-1

(AB4-MM), are comparable to those previously reported for CC mismatches.

Since

metalloinsertor binding affinity correlates directly to site destabilization, it is not surprising

that the binding constants of Rh(bpy)

(chrysi)

for abasic sites are similar to if not somewhat

greater than those for the most destabilizing (e.g., CC) mismatches.

Despite likely differences in site destabilization, little variation is observed between the values

for the three different sequence contexts, a result that suggests a threshold behavior in the

relationship between destabilization and binding affinity. Such behavior has previously been

suggested for mismatch binding.

Small differences, however, do appear based on unpaired

base identity within a single sequence context. For example, the values for AB2 are 1.4(5) ×

-1

(G), 2.1(1) × 10

-1

(A), 2.6(5) × 10

-1

(C), and 3.5(3) × 10

-1

(T). The

metalloinsertor seems to bind abasic sites with unpaired pyrimidines slightly tighter than sites

with unpaired purines. These differences are admittedly minor; however, the trend is consistent

among the three sequence contexts. An explanation based on the kinetics and helicity of the

unpaired base in each case is perhaps most likely.

Enantiospecificity of Rh(bpy)

(chrysi)

for abasic sites

Photocleavage assays employing

-Rh(bpy)

(chrysi)

and

-Rh(bpy)

(chrysi)

clearly

indicate that abasic recognition is enantiospecific for the right-handed isomer of the

metalloinsertor (Figure 6). PAGE experiments reveal that concentrations of 1

-Rh

(bpy)

(chrysi)

bind and cleave all abasic sites interrogated while incubation and irradiation

with 1

-Rh(bpy)

(chrysi)

produce no photocleavage bands. This chiral specificity has

been well-documented for metalloinsertor recognition of mismatched sites.

Recent structural

Zeglis et al.

Page 6

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

studies of Rh(bpy)

(chrysi)

bound to a CA mismatch have shed light on the question; because

the metalloinsertor binds the mismatch site from the narrow, sterically constrictive minor

groove, the chirality of complex must match that of the helix to prevent steric clash between

the ancillary ligands and the DNA backbone. In short, the right-handed helix can only

accommodate the right-handed enantiomer. The observation that metalloinsertor recognition

of abasic sites is also enantiospecific argues strongly for site binding via the minor groove.

MALDI-TOF analysis of abasic site photocleavage products

While we have predominantly employed gel electrophoresis in our study of abasic site

recognition, MALDI-TOF mass spectrometry affords a unique opportunity to investigate not

only the site specificity of recognition but also the identity of the individual photocleavage

products. A similar investigation has been previously reported for metalloinsertor mismatch

recognition.

Here the photocleavage of all 12 cleaved abasic duplexes and their mismatched

analogues was investigated. The MALDI-TOF analysis of AB1-C photocleavage provides a

suitable example (Figure 7). In light (no Rh, with irradiation) and dark (Rh, no irradiation)

controls, only peaks corresponding to the singly (DNA

) and doubly (DNA

) charged parent

single strands are observed, m/z = 8198.7 and 4100.3 for AB1 and 8213.2 and 4106.9 for AB1-

C (Supp. Info.). Photocleavage samples reveal three new masses in addition to the parent

strands at m/z = 3733.7, 4286.8, and 4475.9. These fragments are consistent with the DNA

only being cleaved on the AB1-C strand. We assign the cleavage fragment at m/z = 3733.7 as

a 12-mer with a 5’-phosphate group and the product at m/z = 4286.8 as a 14-mer with a 3’-

phosphate group. These fragments correspond to common DNA oxidation products and clearly

indicate scission on the 3’-side of the unpaired base. The final cleavage fragment, appearing

at m/z = 4475.9, corresponds to the above 14-mer with a 3’-2,3-dehydronucleotide rather than

a phosphate. Upon sample incubation for 24 h at 20 °C, however, complete conversion of the

dehydronucleotide product to the 3’-phosphate fragment is observed, suggesting that the

former is a metastable intermediate.

Analogous results are obtained for all abasic sites that are cleaved on the strand containing the

unpaired base. The situation changes only slightly for the AB4 sequence context, in which

scission occurs on the strand containing the abasic site; for these duplexes, all of the same

cleavage products are observed, but strand scission occurs on the 3’ side of the abasic site.

Importantly, analogous products are also seen for photocleavage of the mismatched strands.

Indeed, exactly the same products are seen for AB1-C and AB1-MM: strand scission occurs

on the 3’-sides of the unpaired cytosine in AB1-C and the corresponding mispaired cytosine

in AB1-MM, resulting in identical fragments (Supp. Info.).

Taken together, these mass spectrometry experiments provide a number of key insights. First,

the data confirm observations made via gel electrophoresis regarding site specificity, strand

asymmetry of scission, and cleavage product length. More important, however, is light shed

on the relationship between abasic site and mismatch recognition and photocleavage. As stated

above, analogous, and in some cases indistinguishable, products are observed for mismatch

and abasic site photocleavage. This result strongly suggests a similar, if not identical, binding

mode for metalloinsertors at abasic sites. Furthermore, based on cleavage product analysis and

structural information, it has been posited that mismatch photocleavage proceeds via an H1-

abstraction mechanism;

based on these results, it is almost certain that abasic site strand

scission occurs via the same pathway.

Recognition and photocleavage of single base bulges by Rh(bpy)

(chrysi)

Compared to abasic sites, single base bulges are recognized less effectively and, when bound,

cleaved less efficiently. In fact, out of the sixteen possible single base bulges in this

investigation, only seven were recognized and cleaved: B1-C, B1-G, B1-T, B2-A, B2-C, B2-

Zeglis et al.

Page 7

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

G, and B2-T (Figure 8). Furthermore, in some cases, even faint bulge photocleavage bands

required longer irradiation times (20-30 min, compared to 10 min for substantial abasic site

cleavage). Based on comparison to shorter labeled oligonucleotides, the bulge photocleavage

fragments appear to be 14 bases long for the B1 duplexes and 15 bases long for the B2 duplexes,

indicating strand scission on the 3’-side of the bulged base. However, the low photocleavage

efficiency at single base bulges precludes the accurate determination of binding affinities.

Based on photocleavage titrations and qualitative observations, however, it is evident that in

each case the metalloinsertor binding affinity is ~10

-1

Both sequence context and bulged base identity appear to play a role in recognition. Single

base bulges in the B3 and B4 sequence contexts escape binding and photocleavage

in toto

whereas all of the bulges in the B2 sequence context are recognized and cleaved to some extent.

The recognition of single base bulges in the B1 sequence context seems to be dependent upon

bulged base identity; the bulged cytosine, guanine, and thymine are cleaved, whereas the bulged

adenine is not. Proffering an explanation for this behavior proves difficult, especially without

the aid of simple bulge site destabilization trends (see

Discussion

Despite the lack of generality in single base bulge recognition, the initial photocleavage assay

and subsequent experimentation do provide some insight into how the metalloinsertor may

bind single base bulges. First, the strand asymmetry and cleavage product length of single base

bulge scission match those of mismatch photocleavage. Second, photocleavage assays

employing

- and

-Rh(bpy)

(chrysi)

clearly indicate that bulge recognition is

enantiospecific for the right-handed isomer of the metalloinsertors. Third, MALDI-TOF

analysis of bulge photocleavage products reveal fragments analogous to those produced in

mismatch and abasic site recognition and scission (Supp. Info.).

For example, Rh

(bpy)

(chrysi)

photocleavage of the B2 -A duplex produces fragments of m/z = 7999.9,

8251.1, 3442.7, 4614.8, and 4802.3. The first two values correspond to the parent single strands

of the duplex. The peak at m/z = 3442.7 corresponds to an 11-mer fragment with a 5’-phosphate,

that fragment at m/z = 4614.8 to a 15-mer with a 3’-phosphate, and that at m/z = 4798.7 to the

same 15-mer fragment but with a 3’-2,3-dehydronucleotide instead of a 3’-phosphate. These

products are, in fact, almost identical to those produced via cleavage of the AB2-A abasic site.

Thus the data clearly suggest that even though Rh(bpy)

(chrysi)

only recognizes single base

bulges in a minority of cases, lesion binding, when it does happen, likely occurs in a mode

analogous to that of the metalloinsertor at mismatches and abasic sites.

DISCUSSION

Based upon the data described, Rh(bpy)

(chrysi)

recognizes abasic sites with high affinity

and specificity with little regard for sequence context or the opposing unpaired base. The

targeting of single base bulges, however, appears to be more complicated, with only seven of

sixteen possible abasic sites bound and cleaved by the metal complex. Yet, now that we have

shown that Rh(bpy)

(chrysi)

can, indeed, bind both lesions, two simple questions follow:

(1) how does the complex bind each lesion and (2) what are the constraints upon the recognition

of these defects?

Rh(bpy)

(chrysi)

binds abasic sites in DNA via metalloinsertion

NMR and X-ray crystallographic evidence has revealed that Rh(bpy)

(chrysi)

binds

mismatched sites not by classical major groove intercalation but rather via a previously unseen

binding mode: insertion. The complex approaches the DNA from the minor groove, ejects the

mismatched bases into the major groove, and replaces the extruded bases in the

-stack with

its own aromatic ligand.

Zeglis et al.

Page 8

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

In the absence of concrete structural information for the abasic site binding, we must rely on

comparisons to mismatch recognition when considering how Rh(bpy)

(chrysi)

targets abasic

sites. The similarities are striking. First, mismatch and abasic site photocleavage exhibit

identical strand asymmetry. In the AB1 and AB2 duplexes, the metal complex cleaves the

strand containing the unpaired bases; in the AB4 duplexes, the strand containing the abasic site

is cut. Mismatch photocleavage mirrors this behavior, with the corresponding strands of the

mismatched duplexes being photocleaved. Second, the enantiospecificity of recognition is

revealing. While bis(bpy) complexes intercalate into B-DNA with very little enantioselectivity,

- Rh(bpy)

(chrysi)

is able to target and cleave mismatched sites enantiospecifically, a

consequence of metalloinsertion occurring from the sterically constrictive minor groove. The

same high specificity is observed for abasic site recognition: only the right-handed enantiomer

targets and cleaves the abasic lesion. This clearly argues strongly for involvement of the minor

groove. Third, mass spectrometry photocleavage product analysis provides still more evidence

for similarity. This technique reveals that both abasic sites and mismatches are cleaved on the

3’-side of the lesions, producing three products: (1) a fragment containing a 5’-phosphate, (2)

a fragment containing a 3’-phosphate, and (3) a metastable fragment identical to (2) but with

a 3’-2,3-dehydronucleotide. Indeed, when the unpaired base in the abasic assembly is a cytosine

and thus contains the same sequence as the mismatched assembly, identical photocleavage

fragments are formed. These products are consistent with H1’-hydrogen abstraction by the

photoactivated ligand, a mechanistic pathway accessible only via the minor groove. Finally, a

variety of other, more minor similarities between abasic site and mismatch recognition exist,

including the failure of Rh(bpy)

(chrysi)

to target either defect in the AB3 sequence context

and the similarity of metalloinsertor binding affinity for both types of lesion, and these

observations also argue for similar binding modes. In sum, this study clearly indicates that

abasic site recognition and photocleavage by metalloinsertors occur in a manner almost, if not

precisely, identical to mismatch targeting. Thus, these data are fully consistent with Rh

(bpy)

(chrysi)

targeting abasic sites via metalloinsertion from the minor groove. It should

be noted that this conclusion fits well with an intuitive, and teleological, approach to the system:

to a metalloinsertor, an abasic site looks like a mismatch with half the extrusion work already

accomplished.

Rh(bpy)

(chrysi)

likely binds single base bulges in DNA via metalloinsertion

Single base bulge recognition presents a somewhat more difficult task for the metalloinsertor.

Of the sixteen different single base bulge sites interrogated in the study, only seven were bound

and cleaved by Rh(bpy)

(chrysi)

. Again, a comparison to mismatch recognition is useful in

exploring the recognition of single base bulges.

Several observations point to a binding mode for single base bulges similar to that for

mismatches and abasic sites. First, photocleavage strand asymmetry for single base bulges

mirrors that for both other defects. Furthermore, bulge recognition, like that of mismatches and

abasic sites, is enantiospecific for the

-isomer of the metalloinsertor. Lastly, the DNA

fragments produced by bulge photocleavage are analogous to those produced by scission

neighboring the other two lesions. Yet two significant differences indicate that the binding

mode must be at least somewhat different. Both the binding affinity and photocleavage

efficiency at single base bulges are substantially reduced compared to that at the corresponding

mismatches and abasic sites. While this certainly does not preclude minor groove insertion, it

does suggest that the orientation of the complex within the binding site differs somewhat from

that at the other two lesions.

Intuitively, this is not surprising. Like binding at an abasic site, metalloinsertion at a single

base bulge requires only the ejection of a single base. However, a key structural difference

exists between single base bulges and the other two lesions: in a bulged duplex, the ribose of

Zeglis et al.

Page 9

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript

the abasic site or the second mismatched base is not there. It is thus not possible for Rh

(bpy)

(chrysi)

to target a bulged site in the same manner it does a mismatch or an abasic site.

One half of the ligand may bind in a manner similar to insertion, extruding the bulged base and

replacing it in the DNA base stack, but the other half of the ligand, without an abasic ribose or

a mismatched base to eject, must bind stacked between adjacent bases in a mode far closer to

metallointercalation than metalloinsertion.

Factors affecting metalloinsertor recognition of abasic sites and single base bulges

Certainly the most puzzling aspect of the abasic site recognition investigation is the absence

of photocleavage for the abasic AB3 duplexes. Neither sequence context nor thermodynamic

stabilization provide explanations; the AB2 duplexes, which also house the abasic site in a 5’-

Pur

Pur-3’ sequence context, are cleaved, and melting temperature measurements suggest that

the AB3 duplexes are as destabilized as the other abasic assemblies. The failure of Rh

(bpy)

(chrysi)

to cleave the AB3 duplex containing a central CC mismatch is equally, if not

more, surprising. Cytosine-cytosine mismatches are among the most destabilizing mispairs

and are readily recognized and cleaved by metalloinsertors in almost any sequence context. It

follows that the most likely, if slightly unsatisfying, explanation is purely based on sequence:

the particular 5’-A

G-3’ sequence context in the AB3 duplexes simply does not allow for

efficient lesion binding and photocleavage. Such anomalies, though poorly understood at

present, have been reported for mismatch targeting and constitute only a very small percentage

of cases.

The somewhat sporadic single base bulge cleavage of Rh(bpy)

(chrysi)

also merits some

attention. As we have noted, only seven of sixteen possible bulges were recognized and cleaved.

A thermodynamic rationale is not available, principally due to the lack of reliable, reported

patterns between bulge sequence and destabilization. Sequence context surely plays a role, but

it cannot be the sole determining factor; both the B2 and B3 assemblies place the bulged base

in a 5’-PyrXPyr-3’ context, but one set of duplexes (B2) exhibits cleavage regardless of bulged

base identity while the other (B3) escapes recognition entirely. The selective cleavage of three

bulged bases in the B1 assemblies suggests bulged base identity as a determining factor, but

the recognition of the B2 sequence bulges regardless of base identity suggests a slightly more

complicated rationale.

One possible explanation may be found in the likely conformation of the bulged base. In the

B2 duplexes, all of which are photocleaved by Rh(bpy)

(chrysi)

, each bulged base is in a 5’-

PyrXPyr-3’ sequence context and is therefore likely to spend at least some time in a extrahelical

conformation. In contrast, the B4 duplexes house the bulged base in a 5’-PurXPur-3’

conformation, with the better stacking purines shifting the likely position of the bulged base

from extra- to intrahelical; in this case, none of the single base bulges is bound and cleaved.

The B1 duplexes provide an intermediate case. Here, the bulged bases are in a 5’-PyrXPur-3’

sequence context. In this case, the bulged bases, likely in an extrahelical conformation, the

pyrimidines C and T, are bound and cleaved, while one of those more likely to prefer an

intrahelical orientation, the purine A, escapes recognition. In sum, the data suggest that the

more likely a base is to exist in an extrahelical conformation, the more easily it will be targeted

by our metalloinsertors. It should be noted, however, that this hypothesis fails to explain the

targeting of the bulged guanine in the B1-G assembly.

CONCLUSIONS

This investigation clearly illustrates that both abasic sites and single base bulges are targeted

by Rh(bpy)

(chrysi)

, a sterically bulky metalloinsertor. Abasic sites are targeted with high

specificity and affinity in all sequence contexts and with all unpaired bases, and a wide variety

of evidence points to minor groove metalloinsertion as the binding mode of the complex at

Zeglis et al.

Page 10

Biochemistry

. Author manuscript; available in PMC 2010 February 10.

NIH-PA Author Manuscript