Bridging PNAs can bind preferentially to a deleted mitochondrial DNA template but replication by mitochondrial DNA polymerase γ in vitro is not impaired

Biochimica et Biophysica Acta 1629 (2003) 73 – 83 www.bba-direct.com

Bridging PNAs can bind preferentially to a deleted mitochondrial DNA template but replication by mitochondrial DNA polymerase g in vitro is not impaired Alistair McGregor, Paul M. Smith, Gu¨nther F. Ross, Robert W. Taylor, Douglass M. Turnbull, Robert N. Lightowlers * School of Neurology, Neurobiology and Psychiatry, The Medical School, University of Newcastle upon Tyne, Framlington Place, Newcastle upon Tyne, NE2 4HH, UK Received 20 February 2003; received in revised form 24 July 2003; accepted 12 August 2003

Abstract Mutations in mitochondrial DNA (mtDNA) are an important cause of neurological and other human pathologies. In the vast majority of cases, supportive care only is available. Mutated and wild-type mtDNAs often coexist in the same cell. A strategy for treatment has been proposed whereby replication of mutated mtDNA is inhibited by selective hybridisation of a nucleic acid derivative, allowing propagation of the wild-type genome and correction of the associated respiratory chain defect. Peptide nucleic acid molecules (PNAs) can be designed to selectively target pathogenic mtDNA with single point mutations. Molecules harbouring deletions present a complex problem. Deletions often occur between two short repeat sequences (4 – 13 residues), one of which is retained in the deleted molecule. With the more common large repeats, it is therefore difficult to design an antigenomic molecule that will bind selectively under physiological conditions. Following limited success with antigenomic oligodeoxynucleotides (ODNs), we have repeated these studies with a series of bridging PNAs. Molecules complementary to the sequence flanking either side of the 13 bp ‘common deletion’ were synthesised. The PNAs demonstrated markedly greater affinity for the delete than to the wild-type template. In runoff assays using Klenow fragment, these PNAs selectively inhibited replication of the delete template. However, no selective inhibition was observed when a polymerase g-containing mitochondrial fraction was used. D 2003 Elsevier B.V. All rights reserved. Keywords: Mitochondrial disease; Heteroplasmy; Mitochondrial gene expression; Peptide nucleic acid; Surface plasmon resonance

1. Introduction All nucleated human cells contain mitochondrial DNA (mtDNA), housed in the mitochondrial matrix. This extranuclear, autonomously replicating circular duplex of 16,569 bp encodes 13 polypeptides, which are essential components of the mitochondrial respiratory chain. In addition, the genome encodes 22 mt-tRNA and 2 mt-rRNAs, the complete set required for intramitochondrial protein synthesis. Defects of this genome are an important cause of progressive muscle and neurological disease that often result in severe disability and death [1,2]. In a population study of the * Corresponding author. Tel.: +44-191-222-8028; fax: +44-191-2228553. E-mail address: [email protected] (R.N. Lightowlers). 0167-4781/$ - see front matter D 2003 Elsevier B.V. All rights reserved. doi:10.1016/j.bbaexp.2003.08.005

northeast of England, the minimum incidence of pathogenic mtDNA mutations has been calculated at approximately 1:8000 individuals [3]. The genetic defect involves either a mtDNA rearrangement (deletion or duplication) or a point mutation, resulting in the impairment of cellular oxidative phosphorylation. Patients often harbour subpopulations of mutated and wild-type mtDNA molecules within the same cell and tissue—a phenomenon known as mtDNA heteroplasmy. The amount of wild-type mtDNA is possibly the critical determinant of the expression of the defect, and hence the clinical phenotype. Although this threshold level is dependent upon both the nature of the mutation and the tissue affected, many pathological mtDNA mutations are highly recessive with biochemical dysfunction only becoming apparent with low levels of wild-type mtDNA [4,5]. Despite widespread recognition of their clinical importance,

74

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

(PNAs) and derivatives. These molecules comprise a backbone of repeating N-(2-aminoethyl)-glycine units connected to standard purine or pyrimidine bases which are able to base pair in both standard Watson – Crick or Hoogsteen forms [9]. As PNAs are uncharged, they show increased binding affinity when pairing with complementary oligonucleotides. DNA – PNA complexes are more sensitive to single base mismatches than the DNA duplex counterpart and their unusual chemistry makes them resistant to degradation under physiological conditions [10,11]. We have been able to show that by using a PNA targeted to a characterised pathogenic mutation (PNA-MERRF, targeting the A8344G mutation in mt-tRNALys), the PNA was able to selectively bind and inhibit runoff replication of the MERRF single-stranded template by a DNA polymerase g-containing fraction [12]. Identical concentrations of this PNA did not inhibit the replication of a wild-type template that differed by just a single substitution. Furthermore, by conjugating PNA-MERRF to a mitochondrial preprotein or to a lipophilic cation, we have been able to target this antigenomic molecule to mitochondria in human cultured cells [13]. More recently, we have attempted to adapt this antigenomic approach to inhibit the replication of pathogenic mtDNA carrying large deletions [14]. Large-scale mtDNA rearrangements represent a major group of reported mtDNA gene defects, of which 30– 50% manifest the 13 bp ‘common deletion’ [15]. Delete genomes have been observed at high levels in patients with several forms of mtDNA disease and may accumulate during normal human ageing [16]. Deletions typically occur at the site of repeat sequences of 4 – 13 bp, only one of which is lost in the deleted molecule [15,17]. As such, the repeat sequence and either the 3V or 5V flanking sequence is common to both the wild-type and the deleted

Fig. 1. Schematic representation to describe how bridging PNAs can selectively bind deleted mtDNA. The wild-type mitochondrial genome contains two 13-base pair repeats, indicated by the solid blocks. These are bounded by four separate sequences, A, B, C, and D. The delete genome has the sequences between the repeats removed, along with one of the repeats. The remaining repeat is bound by sequences A and D. We have designed PNAs that bind to sequences A and D (shown here as A – X – X – D).

there is no effective curative treatment for the vast majority of these disorders [6]. On account of the heteroplasmic and recessive nature of many pathogenic mtDNA mutations, we have proposed a strategy for therapy – inhibition of mutated mtDNA replication by the selective binding of antigenomic nucleic acid derivatives. By allowing the selective replication of wildtype genomes, the relative amount of wild-type mtDNA would increase over time, correcting the biochemical lesion and preventing disease progression [7,8]. As there will be a period during replication when the mitochondrial genome is single-stranded, we have focused our attention on trying to inhibit replication of this intermediate. Our search for an ideal antigenomic agent has centred on peptide nucleic acids

Table 1 Sequence of the wild-type and delete DNA templates and the antigenomic PNAs used in this study

8421 Wild Type DNA PNA 5.5 PNA 6.6 PNA 7.7 PNA 8.8 PNA 9.9 PNA 8xx8 PNA 9xx9 PNA 8.8* PNA 8.0 PNA 9.0 Delete DNA

8469

8483

8509

3′ gtg......tgtttgatggtggatggagggagtggtttcgggtatttttattttttaatattg-5′ cacct ttggc ccacct ttggca ttggcag accacct ttggcagc taccacct ttggcagcc ctaccacct X X ttggcagc taccacct ctaccacct X ttggcagcc X taccacctacctccct taccacct ctaccacct 3′ gtg......tgtttgatggtggatggagggagtggtaaccgtcggatcgtaatc - 5′ 8421

8469

13460

13477

The sequences of the wild-type and delete DNA templates shown are antisense to the published L strand sequence [26]. The underlined bases denotes the repeat sequence. The italicised sequence is that of the templates used in the thermal melt and SPR assays. The line between the two halves of the PNAs shows that the halves are linked with no intervening moieties, whereas the X denotes an inert spacer. The PNAs are shown right to left as N-terminal, analogous to the 5Vend of DNA, to C-terminal, analogous to the 3Vend of DNA.

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

molecule (Fig. 1). Thus, dependent on the size of the repeat sequence, the deleted genome is refractory to our standard antigenomic approach as there is no unique sequence of suitable length that can be targeted to give selective hybridisation at physiological temperatures. However, if the repeat sequence could be bridged, the unique proximity of the 3V and 5V flanking regions in the deleted genome could be exploited for selective hybridisation (Fig. 1). Due to the prevalence of the common deletion, our

75

research has focused on identifying an antigenomic agent that could efficiently target the unique sequence flanking the 13 bp repeat. We have previously shown that bridging oligodeoxynucleotides (ODNs) could be constructed that bind to either side of the deletion sequence on a single-stranded template, with the repeat sequence being successfully bridged by the addition of two inert polyethylene glycol spacer molecules [14]. Although these bridging molecules showed greater

Fig. 2. Thermal denaturation analysis of PNAs 9.9 and 9xx9. The PNA and template ODN were prepared and annealed as described in Materials and methods. As the temperature is gradually increased, the complex dissociates and the absorbance at 260 nm (A260) is collected. Data is shown for two separate experiments (shown in grey or black), represented as primary data (a, c, e, g) or transformed to show the first derivative dA/d jC (b, d, f, h). Wild-type (a, b, e, f) or delete (c, d, g, h) template is shown in tandem with PNA 9.9 (a – d) or the spacer derivative 9xx9 (e – h).

76

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

affinity for the delete over the wild-type template, the greater affinity was reduced significantly when a larger (9 nt) repeat was bridged in comparison to the initial 5 nt repeat. Further, the bridging ODNs were extremely inefficient at inhibiting replication by mtDNA polymerase g. We have now substituted these oligomers with bridging PNAs and have performed similar experiments. PNAs can be synthesised that show dramatically different binding affinities for the delete and wild-type templates and inhibit runoff replication by E. coli DNA polymerase Klenow fragment or Taq DNA polymerase. However, the bound molecules are inefficient at inhibiting replication by T7 DNA polymerase or a partially purified fraction containing mtDNA polymerase g, both highly processive polymerases. We conclude that for successful inhibition, antigenomic bridging PNAs may require essentially irreversible attachment to the DNA template.

2. Materials and methods 2.1. PNA synthesis PNAs were synthesised in-house on an Applied Biosystems Expedite 8909 system using the standard synthesis protocol. Reagents were obtained from Applied Biosystems except where stated otherwise. For the insertion of spacer molecules, the Fmoc 8-amino-3,6-dioxaoctanoic acid (AEEA) was used. Following synthesis, a final 20% piperidine deblock was used to remove the remaining Fmoc group. Cleavage from the support and Bhoc deprotection was carried out using 80% TFA/20% m-cresol (v/v) and the PNAs precipitated/washed with diethyl ether. Correct synthesis was confirmed by MALDI-TOF analysis (Nathan Harris, Moredun Research Institute, Penicuik, Scotland), and by HPLC. PNA sequences are shown in Table 1. 2.2. Polymerase c-containing fraction preparation The preparation of the mitochondrial fraction enriched in mt-polymerase g was based on that of Wong and Clayton [18] and has been detailed in Ref. [12].

absorbance against temperature was used to obtain the Tm. The template DNAs (TAGN, Newcastle) used, with the repeat sequence underlined, in the thermal melt analysis were: Wild-type (8470) melt Delete (common) melt

5V-TAT GGG CTT TGG TGA GGG AGG TAG GTG GTA G-3V 5V-GGC TGC CAA TGG TGA GGG AGG TAG GTG GTA G-3V

2.4. Surface plasmon resonance (SPR) experiments The binding kinetics of the PNAs with immobilised DNA templates were studied using a BIAcore 2000 SPR detector and the streptavidin-coated SA5 (research grade) sensor chip. Approximately 600 Resonance Units (RU) of each DNA were immobilised in separate flow cells on the same sensor chip with the two remaining cells left as blanks. PNAs of various concentrations in HBS (10 mM HEPES, pH 7.4, 150 mM KCl, 3.4 mM EDTA) were passed through all the flow cells simultaneously and the change in RU monitored with time. The RU was further monitored whilst HBS alone was then passed through the cells so that dissociation of the PNA could be observed. All experiments were carried out at 25 jC. Flow cells were regenerated by washing with 50 mM NaOH in HBS. The template DNAs (TAGN) used (with the repeat sequence underlined) were:

WT (8470) biotin DEL (common) biotin

5V-biotin-TAT GGG CTT TGG TGA GGG AGG TAG GTG GTA G-3V 5V-biotin-GGC TGC CAA TGG TGA GGG AGG TAG GTG GTA G-3V

The sensorgram for a blank flow cell, showing any nonspecific binding of PNA to the chip, was subtracted from the sensorgrams obtained from flow cells carrying immobilised template DNA. The association (ka) and dissociation (kd) rate constants of the corrected sensorgrams were determined by nonlinear fitting to a 1:1 (Langmuir) isotherm. The equilibrium dissociation constant (KD) was calculated from the ratio of kd and ka. Local fittings were determined using the BIAevaluation 3.0 software package (Biosensor, Stevenage).

2.3. Thermal melt analysis 2.5. Polymerase runoff assays Analysis was performed on a Lambda Bio 20 UV/ visible spectrophotometer (Perkin Elmer, Cheshire) using the Perkin Elmer UV Winlab software and a PTP-1 peltier system. Melts were carried out in 20 mM Tris/HCl, pH 7.5, 10 mM MgCl2, 150 mM KCl, with each PNA – DNA at 1.7 AM. Samples were initially annealed by heating to 90 jC for 5 min and cooling to room temperature over 30 min. The absorbance at 260 nm of the samples was monitored for three successive up/down temperature ramps with a ramping rate of 0.5 jC/min. The first derivative of

Reactions were performed in 100 Al of 20 mM Tris/HCl, pH 7.0, 150 mM KCl, 10 mM MgCl2, 14 mM h-mercaptoethanol, 1 mM ATP and 100 AM of each dNTP. Template DNA concentrations of 0.34 and 0.034 AM were used with a third the amount of Cy5-labeled primer and various amounts of PNA. These were annealed by heating at 90 jC for 10 min then cooling to 37 jC over 40 min. Polymerase was added and the samples incubated at 37 jC for 15 min. Either 5 Al of mitochondrial polymerase g-containing fraction (1.9

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

77

Table 2 Biophysical properties of all PNA derivatives binding to either the delete or wild-type template are shown PNA

Mean Tm F S.E. jC (n) Wild-type

Delete

5.5 6.6 7.7 8.8 9.9 8xx8 9xx9 8.8* 8.0 9.0

ND ND ND ndm ndm ndm 43.3 F 2.1 67.7 F 0.4 38.9 F 0.9 39.3 F 1.8

44.7 F 1.7 47.4 F 2.1 56.7 F 1.4 62.0 F 0.3 66.6 F 0.2 54.5 F 1.9 60 F 0.3 67.6 F 0.6 43.6 F 1.5 43.9 F 1.7

(3)a (6) (3)a (6)a

KD F S.E. (M) (5) (5) (6) (6) (3) (4) (5) (3) (6)a (6)a

Wild-type

Delete

>10 6 >10 6 >10 6 >10 6 >10 6 >10 6 >10 6 3.7 F 1.0  10 >10 6 >10 6

3.7 F 0.2  10 7.7 F 0.3  10 4.4 F 1.0  10 1.3 F 0.2  10 1.1 F 0.1  10 2.7 F 0.3  10 1.6 F 0.2  10 6.1 F 1.0  10 >10 6 >10 6

10

7 8 9 9 9 9 9 10

The melting temperatures (Tm) and equilibrium dissociation constants (KD) of the various PNA – DNA pairs were calculated as detailed in materials and methods. ND, not determined; ndm, no distinct melt transition. In many cases, the equilibrium dissociation constant was greater than 10 6 M and could not be determined accurately by SPR. a The first derivative of absorbance against temperature showed a broad peak.

Ag/Al), 2 U of Klenow fragment (Roche Diagnostics Limited, Lewes, UK), 2.5 U Taq polymerase (ABgene, Epsom, UK), or 5 U Sequenase T7 DNA polymerase (Amersham USB, Little Chalfont, UK) was used. Reactions were stopped by the addition of 180 Al of phenol, the aqueous phase was chloroform-extracted and the DNA precipitated. DNA pellets were briefly dried before dissolving in 6 M urea in TBE (90 mM Tris/borate, 1 mM EDTA) and heating at 90 jC for 5 min. Samples were electrophoresed through a 12% polyacrylamide gel/8 M urea in TBE. The gels were dried and directly scanned using a Molecular Dynamics Storm 860 in fluorescence mode at 950 V. The wild-type and delete templates and the primer (TAGN, HPLC purified) used (with the repeat sequence underlined) were:

WT (8470)

DEL (common)

Cy5-primer

5V-GTT ATA ATT TTT TAT TTT TAT GGG CTT TGG TGA GGG AGG TAG GTG GTA GTT TGT GTT TAA TAT TTT TAG TTG GGT GAT GAG GAA TAG TG-3V 5V-CTA ATG CTA GGC TGC CAA TGG TGA GGG AGG TAG GTG GTA GTT TGT GTT TAA TAT TTT TAG TTG GGT GAT GAG GAA TAG TG-3V 5V-Cy5-TGT AAA ACG ACG GCC AGT CAC TAT TCC TCA TCA CCC-3V

In all cases, sizes of full-length and truncated products were calculated from the position of the unextended labeled primer and from the ladder generated from partially extended primers visible in all gels analysed. 2.6. Gel mobility shift analysis of PNA Reactions were performed under identical conditions to the runoff assays using various concentrations of PNA. A complementary ODN was end-labeled with g-ATP (3000 Ci/mmol, Amersham) by polynucleotide kinase following

manufacturers recommendations. After incubation, the mitochondrial fraction and PNA were heat-inactivated at 85 jC and approximately 1 pmol of oligonucleotide added. The mixture was heated again to 95 jC for 10 min and cooled to room temperature over 20 min to allow for annealing of the PNA – DNA duplex. The mixture was then separated through a 16% (19:1) polyacrylamide nondenaturing gel by electrophoresis, transferred to Whatman 3MM paper, dried and exposed to a PhosphorImager cassette.

3. Results The aim of these experiments was to determine if a PNA could be designed that would bind with greater affinity to a template carrying the ‘common’ deletion than to the wildtype counterpart under physiological conditions and to assess whether this molecule could inhibit runoff replication. Generation of the ‘common’ deletion is nucleated by a 13-bp repeat sequence separated by almost 5 kb on the mitochondrial genome. As illustrated in Fig. 1, one copy of this repeat sequence is retained in the delete genome. Considering the sequences immediately prior to and following the first repeat (8470 –8482) as A and B, respectively, and those prior to and following the second repeat (13,447 – 13,459) as C and D, respectively, the delete molecule retains a 13-bp repeat bounded by sequences A and D (see Fig. 1). We have designed and synthesised a series of PNAs that are complementary to sequences A and D or to A alone. Molecules were standard PNAs with equal numbers of bases complementary to each side of the repeat. They contained no spacers (5.5, 6.6, 7.7, 8.8, and 9.9) or had two inert spacers between the two sides (8xx8 and 9xx9). Standard PNAs with complementarity to a single side of the repeat (8.0, 9.0) were also made. A control PNA, 8.8*, was synthesised that was complementary to the eight bases prior to the repeat and the next eight contiguous bases within the

78

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

repeat (Table 1). To determine the binding kinetics of these molecules to single-stranded 31-mer ODNs representing either the wild-type or delete sequence, two biophysical techniques were used; thermal melt and SPR techniques. 3.1. Thermal melt analysis of PNAs with wild-type and delete templates The thermal denaturation temperatures (Tm) of the various PNAs bound to wild-type and delete templates were measured. Fig. 2 shows the first derivative and untransformed data (UV absorbance against temperature) from two typical repeat thermal denaturation experiments, in this case

for the 9.9 and 9xx9-mer PNAs. As is shown in Table 2, all complexes showed predictable relative stability. PNA 8.8*, which binds to a contiguous 16 nucleotide stretch present in both templates, showed similar melting temperatures with the wild-type and delete templates of 67.7 and 67.6 jC, respectively. PNA 8.8, complementary to eight bases on each side of the delete template repeat, has a melt temperature of 62.0 jC, whereas the longer PNA 9.9 melted at 66.6 jC, comparable to PNA 8.8*. Interestingly, PNAs 8xx8 and 9xx9, which have two inert spacers to bridge the repeat sequence between the two regions of complementarity, demonstrated lower melt temperatures than their respective PNAs lacking the spacers. PNAs that interact with only one

Fig. 3. The bridging PNAs have a greater affinity for the delete than the wild-type template. Details of the kinetic experiments are given in Materials and methods and the results are summarised in Table 2. Representative sensorgrams are shown. (a) Sensorgrams for the binding and dissociation of PNAs 9.9 and 9xx9 to the wild-type and delete templates. (b) Sensorgrams for the binding and dissociation of the contiguous 16mer, PNA 8.8*, to the wild-type and delete templates. (c) Sensorgrams for the binding and dissociation of PNA 9.0 to the wild-type and delete templates. Note the different response unit scales and that the PNA concentrations were (a, b) 125 nM and (c) 3.75 AM.

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

of the two binding regions, such as PNA 8.8 to the wild-type template or PNA 8.0 to either template, showed a low Tm derived from a broad first derivative, or the first derivative was too broad for an accurate Tm value to be calculated. 3.2. SPR analysis of PNAs with wild-type and delete templates PNAs with complementary bases on either side of the repeat all bound the delete template with measurable equilibrium dissociation constants. These ranged from 3.7  10 7 M for PNA 5.5 to 1.1  10 9 M for PNA 9.9 (see Fig. 3 and Table 2), with the differences largely being

79

due to the dissociation rate constants of the PNAs decreasing with their increasing length. Interestingly, the data suggests that it is not necessary to artificially bridge the repeat sequence as the bridging PNAs containing the polyethylene glycol linkers bound with lower affinity than their counterparts lacking the spacer moieties (i.e. 2.7  10 9 M for 8xx8 cf. 1.3  10 9 M for 8.8). This trend was also noted with the thermal melt experiments. PNAs 8.0 and 9.0, which bind to one side of the delete repeat, generated small changes in the RU of the sensorgrams such that the data could not be fitted to a simple 1:1 Langmuir isotherm. Similarly, only the fully complementary control PNA 8.8* demonstrated any measurable binding to the wild-type

Fig. 4. Bridging PNAs inhibit Klenow fragment-mediated runoff from the delete and not wild-type templates. The polymerase assay was performed using wildtype and delete templates (34 nM) as indicated. The effect of (a) PNAs 9.9 and 9xx9, (b) PNA 8.0, (c) PNA 8.8, and (d) PNA 8xx8 on replication termination were examined. The PNA concentrations used were (a) 1, 0.25, and 0.063 AM, (b) 10, 1, 0.5, 0.125 AM with wild-type template; 10, 1, 0.5, 0.125 and 0.063 AM with delete template, and (c, d) 10, 1, 0.5, 0.25, 0.125 and 0.063 AM. 0, no PNA added; in, 1 AM PNA 8.8* added for maximal inhibition. All products were accurately sized against free dye-labeled primer (not shown) and the ladder of partially extended products that can be seen in each gel. The various runoff products are wt, full-length product from wild-type template; del, full-length product from delete template; it, intermediate truncation product; ft, expected truncation product; pr, indicates unextended primer.

80

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

template with a KD of 3.7  10 10 M, comparable to 6.1  10 10 M with the delete template, as expected. All other PNAs examined had such poor binding affinities for their wild-type templates that an accurate determination of the equilibrium dissociation constant could not be made (problems with the solubility of the PNAs above 5– 10 AM prevented attempts to accurately quantify weak binding affinities below 10 6 M). Thus, all PNAs containing at least eight bases of complementarity to both sides of the repeat demonstrated greater than 1000-fold affinity for the delete over the wild-type template. This is considerably greater than previous experiments using standard ODNs, where linking two 9mers to span a 5 nt repeat sequence only increased the affinity for the delete over the wild-type template by approximately 25-fold (11). 3.3. Klenow-mediated polymerisation is inhibited by PNA binding to delete and not the wild-type template Results from these previous experiments showed that the bridging PNAs bound with far greater affinity to the delete as compared to the wild-type template. Thus, it would be predicted that the bridging PNAs might selectively inhibit extension of the delete and not the wild-type template. We therefore examined whether these PNAs could discriminate between the two templates in a polymerase runoff assay. Initially, we examined primer extension driven by the large fragment of E. coli polymerase I (Klenow fragment) in the presence of our PNA molecules. PNAs 8.0 and 9.0 showed no discrimination between the wild-type and delete templates, and high concentrations (10 AM) of PNA were required to inhibit runoff. Discrimination between the delete and wild-type templates, however, could be seen with PNAs

9.9, 9xx9, 8.8, 8xx8 (Fig. 4), 7.7, and 6.6 (data not shown), with the synthesis of full-length products being progressively inhibited by increasing concentrations of PNA. All products were sized against free dye-labeled primer and the ladder of partially extended products that can be seen in each gel. With the bridging PNAs, the predominant truncation product extended to the nucleotide immediately prior to the 5V(amino) base of the predicted PNA binding site. Interestingly the PNAs that bound to either side of the repeat sequence on the delete template but that lack the spacers (8.8 and 9.9) produced two truncated runoff products from the delete template. These corresponded to the expected truncated species and a slightly longer product terminating towards the end of the repeat sequence and prior to the second half of the bound PNA molecule. PNAs containing the inert spacers did not generate the intermediate truncation product. Although affinities and thermal denaturation profiles were similar for 8.8* and 9.9 (Table 2), the unbroken complementarity of 8.8* was more efficient at blocking extension by all polymerases (Fig. 4, in), confirming that inhibition efficiencies cannot merely be inferred by the biophysical data alone. 3.4. Bridging PNAs do not selectively inhibit pol c-mediated runoff replication Extension of the two templates by Klenow fragment is differentially inhibited by the bridging PNAs, consistent with the thermal melt and binding data. However, as the long-term goal of this project is to selectively inhibit the replication of pathogenic mtDNA in vivo, it was important to confirm that runoff assays programmed with a relatively

Fig. 5. Bridging PNAs do not discriminate between runoff replication of wild-type and delete templates when using highly processive DNA polymerases. The assay was performed using (a) the polymerase g-containing mitochondrial fraction at 37 jC, (b) Taq polymerase at 50 jC, and (c) Sequenase at 37 jC. In (a) PNA 9.9 concentrations of 10, 5, 0.5, 0.1 and 0.01 AM were used with the wild-type template and 10, 5, 1, 0.5, 0.1 and 0.01 AM with the delete template, in (b) PNA 7.7 and PNA 9.9 concentrations of 1, 0.25, and 0.063 AM were used, and in (c) PNA 9.9 was used at 1 AM. Abbreviations and template concentrations are identical to those given for Fig. 4.

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

crude preparation of the mtDNA polymerase g behaved in a similar fashion. A fraction was prepared from bovine liver mitochondria as previously described [12] and used to program a runoff replication assay. Surprisingly, this preparation generated full-length products from both the wild-type and delete templates when the bridging PNAs were added, although a large proportion of the primer remained unextended (Fig. 5a). Varying the amount of polymerase fraction or extending the incubation time had negligible effect (data not shown). Although a large molar excess of the bridging PNAs could inhibit the synthesis of full-length delete molecules, there was little discrimination and production of the full-length wild-type product was inhibited to a similar degree. To determine whether the PNA had been degraded by treatment with the polymerase g-containing mitochondrial fraction, a gel mobility shift assay was designed (Fig. 6). PNA was visualised by binding to a radiolabeled complementary ODN, with the complex showing a slower migration through a nondenaturing gel. Various concentrations of PNA were exposed to lysate under identical conditions to the runoff assays, the lysate was heat-inactivated and radiolabeled antisense ODN added. A similar shift in mobility was detected with the PNA after exposure to the pol g fraction, confirming that the PNA remained intact during the assay. A major difference between the two polymerases is that Klenow has a low processivity, whereas mtDNA polymerase g is highly processive due to its accessory subunit. Two additional polymerases were therefore examined, the standard thermostable enzyme, Taq, which has relatively poor processivity, and the bacteriophage T7 enzyme, Sequenase, which is highly processive. When runoff assays were performed with Taq polymerase at 50 jC, PNA 9.9 inhibited the production of the full-length delete product and gener-

Fig. 6. PNAs are not degraded during incubation with mitochondrial protein. Increasing amounts of PNA (0.3, 1.25 and 2.5 AM) were incubated with mitochondrial fraction prior (Fract III), or after heat inactivation (Heat Inact Fract III) or without the addition of the mitochondrial fraction (0). A radiolabeled oligonucleotide complementary to the PNA was added to the mixture and the complex was then resolved using nondenaturing polyacrylamide electrophoresis. The shift of PNA – DNA complex was similar with all PNA concentrations irrespective of exposure to mitochondrial protein.

81

ated the intermediate truncation product (Fig. 5b). Production of full-length wild-type product was not affected. In contrast, PNA 9.9 failed to cause truncation from either template when Sequenase polymerase was used in similar assays to those performed with the mitochondrial pol g fraction (Fig. 5c), consistent with the hypothesis that highly processive polymerases were able to remove the bound bridging PNAs.

4. Discussion The thermal denaturation and SPR data show that the affinity of the bridging PNAs is far greater for the delete than the wild-type template. It was interesting that the PNAs that bind either side of the repeat but lack the AEEA spacer groups between the two binding regions of the PNA (i.e. 8.8 and 9.9) bound more strongly than the equivalent PNAs that had spacers (8xx8 and 9xx9). Presumably, the nucleotides of the template around the repeat can assume a more stable conformation that either permits the close apposition of the bases on each side of the repeat, or places them in such an orientation that the spacer containing PNAs have difficulty in binding both parts simultaneously. However, the affinity of PNA 9xx9 to the delete template was more than 600-fold greater than its binding to the wild-type template under conditions used in these experiments. In contrast, the 9xx9 ODN employed by Taylor et al. [14] to bridge only nine residues of the common deletion, showed a much smaller variation in binding affinity for the two templates, with a difference of only 20-fold. The 9xx9 ODN also had a lower Tm than PNA 9xx9 (51 jC cf. 60 jC) as would be predicted. Both biophysical techniques provided data to suggest that the bridging PNAs rather than ODNs would be far more effective as antigenomic agents in targeting repeat sequences around deletion sites. In replication runoff assays using Klenow or Taq polymerase, the bridging PNAs inhibited replication of the delete template to a greater degree than the wild-type. However, a similar discrimination was not observed when using mtDNA polymerase g or Sequenase polymerase. In contrast to Klenow [19] and Taq [20], where the average polymerisation per productive binding event is just a few nucleotides or 60 –80 nt, respectively, up to several thousand nucleotides can be added by the intact mtDNA polymerase g [21,23] and Sequenase [22], both highly processive enzymes. It is therefore tempting to conclude that the selective inhibition can be correlated to differences in processivity. Previously we demonstrated that an 11 base antisense PNA could discriminate between a mutant template with a single base change and the wild-type template in a runoff assay using a mtDNA polymerase g-containing fraction [12]. This initially appears to conflict with the data presented here, but the two experiments are in fact significantly different. The 11 base PNA binds a contiguous stretch of nucleotides, whereas the bridging PNAs interact

82

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83

with two smaller binding regions that act cooperatively in increasing their overall binding affinity to the DNA. In the runoff assay, it is plausible that the highly processive polymerases are either able to displace each half of the bridging PNA in turn, or rapidly proceed from a paused position adjacent to the bound PNA when that PNA dissociates. We also previously reported that there was a slight degree of discrimination between a wild-type and delete template with the mtDNA polymerase-g fraction when bridging ODNs were used to truncate replication [14]. This could possibly be due to these ODNs only having to bridge 9 nucleotides rather than 13. Slight differences in the mitochondrial polymerase fraction preparation such as partial loss of the p55 accessory factor could also provide an explanation. It has been demonstrated that the p55 subunit of polymerase g confers processivity, and in addition promotes the extension of mismatched termini [23]. The PNAs examined demonstrate a markedly greater affinity for the delete than for the wild-type templates. However, binding of these bridging PNAs did not cause a selective truncation of polymerase g-mediated replication of the delete template, suggesting that they would not be suitable as therapeutic agents in their present form. The absence of discrimination with the 9.9-mer PNA suggests that with processive polymerases, it may be necessary to use longer bridging PNAs. To address this possibility, we attempted to synthesise the relevant 10.10, 11.11 and 12.12-mers, but found repeated problems with solubility (the addition of lysine residues to these molecules is likely to have resolved this problem, but would have complicated direct comparisons with the shorter molecules). However, although it is disappointing that these simple bridging PNAs alone are not sufficient to inhibit replication when bound to a single-stranded template, we believe it will still be possible to design antigenomic agents around a PNA derivative to target the delete species. Linking the specificity of these PNAs to a moiety able to cross-link to DNA could potentially overcome this difficulty in inhibiting extension by some polymerases. It is also well known that bis-PNAs bind in a virtually irreversible fashion [24], but it is not easy to see how such a molecule could easily be designed to bridge sequences as would be necessary in this situation. Finally, although we intend to target single-stranded replication intermediates, the new pseudocomplementary PNAs that show sequence-selective strand invasion of duplex DNA could provide an effective template to design a bridging agent [25]. To date, however, the import of any of these PNA derivatives into the mitochondrial matrix remains unproven.

Acknowledgements We would like to thank the Muscular Dystrophy Campaign (RWT, RNL and DMT) and The Wellcome Trust

(DMT and RNL, programme grant no. 056605) for financial support.

References [1] P.F. Chinnery, D.M. Turnbull, Mitochondrial DNA and disease, Lancet 354 (Suppl. 1) (1999) 117 – 121. [2] S. DiMauro, E. Bonilla, M. Davidson, M. Hirano, E. Schon, Mitochondria in neuromuscular disorders, Biochim. Biophys. Acta 1366 (1998) 199 – 210. [3] P.F. Chinnery, M.A. Johnson, T.M. Wardell, R. Singh-Kler, C. Hayes, D.T. Brown, R.W. Taylor, L.A. Bindoff, D.M. Turnbull, The epidemiology of pathogenic mitochondrial DNA mutations, Ann. Neurol. 48 (2000) 188 – 193. [4] A. Chomyn, A. Martinuzzi, M. Yoneda, A. Daga, O. Hurko, D. Johns, S.T. Lai, I. Nonaka, C. Angelini, G. Attardi, MELAS mutation in mtDNA binding site for transcription termination factor causes defects in protein synthesis and in respiration but no change in levels of upstream and downstream mature transcripts, Proc. Natl. Acad. Sci. U. S. A. 89 (1992) 4221 – 4225. [5] L. Boulet, G. Karpati, E.A. Shoubridge, Distribution and threshold expression of the tRNAlys mutation in skeletal muscle of patients with Myoclonic Epilepsy and Ragged Red Fibres (MERRF), Am. J. Hum. Genet. 51 (1992) 1187 – 1200. [6] Z.M. Chrzanowska-Lightowlers, R.N. Lightowlers, D.M. Turnbull, Gene therapy for mitochondrial DNA defects: is it possible? Gene Ther. 2 (1995) 311 – 316. [7] R.W. Taylor, P.F. Chinnery, K.M. Clark, R.N. Lightowlers, D.M. Turnbull, Treatment of mitochondrial disease, J. Biomembranes Bioenerg. 29 (1997) 195 – 205. [8] R.W. Taylor, T.M. Wardell, P.M. Smith, A. Muratovska, M.P. Murphy, D.M. Turnbull, R.N. Lightowlers, An antigenomic strategy for treating heteroplasmic mtDNA disorders, Adv. Drug Deliv. Rev. 49 (2001) 121 – 125. [9] M. Egholm, O. Buchart, L. Christensen, C. Behrens, S.M. Freier, D.A. Driver, R.H. Berg, S.K. Kim, B. Norden, P.E. Nielsen, PNA hybridizes to complementary oligonucleotides obeying the Watson – Crick hydrogen-bonding rules, Nature 365 (1993) 566 – 568. [10] H. Orum, P.E. Nielsen, M. Egholm, R.H. Berg, O. Buchardt, C. Stanley, Single base pair mutation analysis by PNA directed PCR clamping, Nucleic Acids Res. 21 (1993) 5332 – 5336. [11] V.V. Demidov, V.N. Potaman, M.D. Frank-Kamenetskii, M. Egholm, O. Buchardt, S.H. Sonnichsen, P.E. Nielsen, Stability of peptide nucleic acids in human serum and cellular extracts, Biochem. Pharmacol. 48 (1994) 1310 – 1313. [12] R.W. Taylor, P.F. Chinnery, D.M. Turnbull, R.N. Lightowlers, Selective inhibition of mutant human mitochondrial DNA replication in vitro by peptide nucleic acids, Nat. Genet. 15 (1997) 212 – 215. [13] P.F. Chinnery, R.W. Taylor, K. Diekert, R. Lill, D.M. Turnbull, R.N. Lightowlers, Peptide nucleic acid delivery to human mitochondria, Gene Ther. 6 (1999) 1919 – 1928. [14] R.W. Taylor, T.M. Wardell, B.A. Connolly, D.M. Turnbull, R.N. Lightowlers, Linked oligodeoxynucleotides show binding cooperativity and can selectively impair replication of deleted mitochondrial DNA templates, Nucleic Acids Res. 29 (2001) 3404 – 3412. [15] E.A. Schon, R. Rizzuto, C.T. Moraes, H. Nakase, M. Zeviani, S. DiMauro, A direct repeat is a hotspot for large-scale deletion of human mitochondrial DNA, Science 244 (1989) 346 – 349. [16] E.J. Brierley, M. Johnson, R.N. Lightowlers, O. James, D.M. Turnbull, Role of mitochondrial DNA mutations in human ageing: implications for the central nervous system and muscle, Ann. Neurol. 43 (1998) 217 – 223. [17] A. Baumer, C. Zhang, A. Linnane, P. Nagley, Age-related human mtDNA deletions: a heterogeneous set of deletions arising at directly repeated sequences, Am. J. Hum. Genet. 54 (1994) 618 – 630.

A. McGregor et al. / Biochimica et Biophysica Acta 1629 (2003) 73–83 [18] T.W. Wong, D.A. Clayton, Isolation and characterization of a DNA primase from human mitochondria, J. Biol. Chem. 260 (1985) 11530 – 11536. [19] R. Kuchta, V. Mizrahi, P. Benkovic, K. Johnson, S. Benkovic, Kinetic mechanism of DNA polymerase I (Klenow), Biochemistry 26 (1987) 8410 – 8417. [20] R. Abramson, S. Stoffel, H. Gelfand, Extension rate and processivity of Thermus aquaticus DNA polymerase, FASEB J. 4 (1990) A2293. [21] M. Fry, L. Loeb, Animal Cell DNA Polymerases, CRC Press, Florida, 1986. [22] S. Patel, I. Wong, K. Johnson, Pre-steady-state kinetic analysis of processive DNA replication including complete characterisation of an exonuclease-deficient mutant, Biochemistry 30 (1991) 511 – 525. [23] M. Longley, D. Nguyen, T. Kunkel, W. Copeland, The fidelity of human DNA polymerase g with and without exonucleolytic proof-

83

reading and the p55 accessory subunit, J. Biol. Chem. 276 (2001) 38555 – 38562. [24] Y. Kosaganov, D. Stetsenko, E. Lubyako, N. Kvitko, Y. Lazurkin, P.E. Nielsen, Effect of temperature and ionic strength on the dissociation kinetics and lifetime of PNA – DNA triplexes, Biochemistry 39 (2000) 11742 – 11747. [25] J. Lohse, O. Dahl, P.E. Nielsen, Double duplex invasion by peptide nucleic acid: a general principle for sequence-specific targeting of double-stranded DNA, Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 11804 – 11808. [26] S. Anderson, A.T. Bankier, B.G. Barrell, M.H.L. De Bruijn, A.R. Coulson, J. Drouin, I.C. Eperon, D.P. Nierlich, B.A. Roe, F. Sanger, P.H. Schreier, A.J.H. Smith, R. Staden, Y.I. J, Sequence and organization of the human mitochondrial genome, Nature 290 (1981) 457 – 465.