|Home | About | Journals | Submit | Contact Us | Français|
Although autosomal dominant polycystic kidney disease is transmitted in an autosomal dominant fashion, there is evidence that the pathophysiology of cystogenesis involves a second hit somatic mutation superimposed upon the inherited germline mutation within the renal tubule cells. The polypurine·polypyrimidine (Pu·Py) tract of PKD1 intron 21 may play a role in promoting somatic mutations. To better characterize this tract and to evaluate its potential to participate in mutagenesis, we investigated the thermodynamics of intramolecular triplex formation by 15 Pu·Py mirror repeat tracts from PKD1 intron 21 by 2D gel electrophoresis. We demonstrate that intramolecular triplexes form with modest superhelical tensions for all the tracts examined. Primer extension studies demonstrated significant polymerase arrest within the Pu·Py tracts in one direction of replication only. We found correlation between polymerization arrest and both the potential length of the triplex and superhelical tension of intramolecular triplex formation. The presence of a Pu·Py tract also led to a replication blockade and double-strand breakage using an SV40 in vitro replication assay with HeLa cell extracts. During DNA replication, the G-rich template of the PKD1 Pu·Py tracts may form a triplex structure with the nascent strand, thereby blocking replication and potentially leading to recombination and mutation.
Autosomal dominant polycystic kidney disease (ADPKD) is one of the most common genetic diseases worldwide (1) and is responsible for up to 10% of cases of end-stage renal disease in the United States (2). However, the expressivity of disease is remarkably variable both within the affected population as a whole and within individual families who share the same germline mutation. In a monozygotic twin study, Levy et al. studied 20 pairs of monozygotic twins and concluded that the disease course in the twin pairs was as different as if they were not related (3). Therefore, additional factors beyond the inheritance of a germline mutation must modify the phenotypic expression of the disease.
More than 85% of the cases of ADPKD are associated with a mutation in the PKD1 gene. Despite sharing the same genetic information, only a minority of the tubules develop into cysts (4). One explanation for this observation could be that cystogenesis requires an initiating event, such as a somatic mutation in the PKD1 allele from the non-affected parent (5,6). Having lost the normal copy of the PKD1 gene in the renal tubular epithelial cell, cysts develop and progress. This concept of a somatic mutation superimposed on a heterozygous germline mutation has been referred to as the ‘two-hit hypothesis’ with the first ‘hit’ being the inherited germline mutation and the second being the acquired somatic mutation, resulting in loss of heterozygosity (LOH) (7,8). Such a model, initially proposed by Knudson to explain carcinogenesis (9), has been proposed for other autosomal dominant diseases (10–14). The two-hit model in ADPKD is supported by studies demonstrating LOH at PKD1 within a subset of cysts from diseased individuals (5,15–17).
Based on the two-hit hypothesis, the number of cysts found in most patients with ADPKD would imply a higher spontaneous mutation rate within PKD1 than would be expected. While previous studies have suggested that renal cells have a greater mutation rate than cells elsewhere (18,19), another explanation would be the presence of sequence motifs within the PKD1 gene that facilitate somatic mutation. One candidate as a mutation-promoting sequence is the polypurine·polypyrimidine (Pu·Py) tract located in PKD1 intron 21. Pu·Py mirror repeat sequences are capable of forming intramolecular DNA triplexes and are often found at or near sites of increased recombination frequency (20–22). The PKD1 intron 21 Pu·Py tract is 2.5 kb in length, contains 65% cytosine and 32% thymidine (97% pyrimidine) on the coding strand, and is one of the largest intragenic Pu·Py tracts in the human genome. There are 23 mirror repeat sequences within the entire Pu·Py tract (23).
We have previously used primer extension reactions following chloroacetaldehyde and potassium permanganate modification to establish that a mirror repeat Pu·Py tract from the PKD1 gene can form triplex DNA (24). Using this same tract, Tiner et al. imaged the intramolecular triplex structure using atomic force microscopy (25). Triplex formation has been shown to present a potent block for DNA replication (26–28). However, the features of Pu·Py tracts which favor the inhibition of replication are less clear.
To further evaluate the potential of the PKD1 intron 21 to participate in DNA-directed mutagenesis, we present the thermodynamic characterization of the Pu·Py sequences within the P1 nuclease-sensitive regions that engage in triplex DNA formation. To begin to understand the characteristics of Pu·Py tracts which may be used to predict interference with DNA replication we evaluated the effect of those sequences on DNA polymerization. We demonstrate evidence that PKD1 Pu·Py tracts significantly interfere with DNA replication and that there is correlation between replication blockade and both the potential length of triplex and the superhelical tension of intramolecular triplex formation.
There are ten individual Pu·Py mirror repeat tracts within the four P1 nuclease-sensitive regions of PKD1 intron 21 which make up five overlapping pairs. We cloned the ten individual and five overlapping sequences (Fig. (Fig.1)1) into the EcoRI site of pUC19. The constructs were prepared and verified as previously described (24). All plasmids were prepared by alkaline lysis and cesium chloride buoyant density centrifugation (29). The correct size of the insert after cloning was confirmed by restriction endonuclease digestion of the plasmid by EcoRI, followed by T4 polynucleotide kinase labeling with [γ-32P]ATP (Perkin Elmer Life Sciences Inc, Boston, MA, USA) and electrophoretic separation in 5% acrylamide (see Supplementary Fig. S1). pBluescript (a pUC-based cloning vector) was used as a control. Like other pUC-based vectors, it does not exhibit significant conformational changes at low pH. Plasmid topoisomers were generated as previously described (24). All reagents and enzymes were supplied by Gibco (Life Technologies, Rockville, MD, USA) unless otherwise noted.
Topoisomer mixtures were incubated in electrophoresis buffer (13.5 mM Tris pH 5.0 with 1 mM magnesium acetate) for 60 min at 37°C prior to electrophoresis in the first dimension. All two-dimensional (2D) electrophoreses were performed in 1.25% agarose (20 cm × 20 cm) at 4°C using 5 V/cm for 17 h. The buffer was circulated to avoid the development of a pH gradient. The gel was equilibrated in a second electrophoresis buffer (40 mM Tris, 25 mM sodium acetate, 1 mM EDTA pH 8.3 and 40 mM chloroquine phosphate) for 4 h prior to the second electrophoresis perpendicular to the first dimension. At the completion of electrophoresis the gels were rinsed with distilled water and stained with ethidium bromide (0.5 µg/ml). The amount of superhelical tension required for, and absorbed by, triplex formation can be determined by counting the writhe (number of superhelical turns) at which structural transition occurs using 2D gel analysis. The superhelical tension σ required for structural transition of each plasmid was calculated using the formula:
where ω = the writhe at which structural transition occurred and N = the total number of base pairs in the plasmid. The superhelical tension can be converted to Gibbs free energies of formation and absorption by the formulae:
where R = gas constant (8.31441 J K–1 mol–1), T = temperature (Kelvin) and Δω = number of writhes absorbed by triplex formation.
Template/primer preparation. One microgram of each of the 15 plasmids was digested with 10 units PvuII in a 10 µl reaction for 2 h at 37°C. pUC19 was used as the control for the primer extensions. pUC19 has two PvuII sites, one on either side of the insert. Oligonucleotide primers flanking the EcoRI site of pUC19 within the PvuII sites were synthesized at the University of Cincinnati DNA core facility. One microgram of each primer forward (5′-TGGCGAAAGG GGGATGTG-3′) and reverse (5′-AGCGGATAACAATTT CACACAGGA-3′) were end-labeled with [γ-32P]ATP (Perkin Elmer Life Sciences Inc, Boston, MA, USA) using 10 units T4 polynucleotide kinase in a 25 µl reaction.
Primer extension reaction. A pair of primer extension reactions was run for each plasmid, one for the cytosine-rich template and one for the guanine-rich template. One microliter of labeled primer was mixed with 1 µl of the plasmid digestion product along with 5 units of Taq polymerase, 0.5 mM dNTP, extension buffer (20 mM Tris–HCl pH 8.4, 50 mM KCl) and 3 mM MgCl2. The ratio of primer to template was 100:1. Primer extension conditions were identical for all samples: initial template denaturation at 94°C for 3 min; 30 cycles of 94°C for 1 min, 55°C for 2 min and 72°C for 2 min; final extension at 72°C for 10 min. The primer extension reaction products were mixed with formamide as a denaturant and bromophenol blue as a marker, heated to 94°C for 3 min, snap cooled and resolved on an 8% denaturing polyacrylamide gel. For size determination, a 25 bp molecular weight marker was used. The gel was dried at 80°C for 75 min, exposed to a PhosphorScreen (Molecular Dynamics, Sunnyvale, CA, USA) overnight, and imaged using the Storm 860 (Molecular Dynamics) PhosphorImager. Densitometric analysis using ImageQuant (Molecular Dynamics) software was used to normalize the amount of polymerase blockade by dividing the area of arrested products by the total area of both arrested and full-length products for each of the plasmid/primer combinations. Comparisons done in this manner minimize differences caused by subtle changes in the amount of template. The primer extension reactions were performed three times and the average amount of polymerase arrest as a percentage of total product was used for comparisons.
Correlation between polymerase arrest and the following factors was examined: superhelical tension required for intramolecular triplex formation, number of superhelical turns unwound by triplex formation, potential number of base pairs participating in triplex formation, percentage of mirror symmetry, amount of G·C base pairs within the mirror segment of the tract, amount of A·T base pairs within the center of the tract, and length separating the mirror repeat halves. After examining the scatter plots to rule out a nonlinear relationship, the correlation between the above factors and amount of polymerase blockade was calculated using the Spearman rank correlation coefficient.
We used the clone pJB4 containing two overlapping mirror repeats, including the longest mirror repeat in the entire PKD1 Pu·Py tract that also has been shown to form triplex structures at physiologic superhelical tensions (24), to generate single-stranded templates. Using the very high fidelity polymerase system described by Barnes (30), the radiolabeled primer was annealed 5′ to the cloning site and this annealed product was incubated with dNTPs and replication-competent extract containing human SSB called replication protein A for 60 min at 37°C. The DNA was phenol/chloroform purified and resolved on a sequencing gel.
Mammalian cells (HeLa) were grown in Dulbecco’s minimal essential medium at 37°C and 5% CO2. Cell-free extracts for DNA replication in vitro were prepared by the method of Li and Kelly (31).
DNA replication reactions were carried out (in 30 µl) using 50–60 µg of HeLa cell extract protein (32–34); 20 ng of DNA template; 30 mM HEPES pH 7.5; 7 mM MgCl2; 0.5 mM dithiothreitol; 100 µM each dATP, dGTP and dTTP; 50 µM dCTP; 10 µCi of [α-32P]dCTP; 200 µM each GTP, UTP, and CTP; 4 mM ATP; 40 mM creatine phosphate; 10 µg of creatine kinase; and 1 µg of SV40 large T antigen (Molecular Biology Resources, Madison, WI). Reactions were incubated for 4 h at 37°C, and the reaction products were isolated and purified as described previously (35). Replication products were separated by electrophoresis on a 1% agarose gel containing 5 µg/ml ethidium bromide. Replicated DNA was visualized by autoradiography of the dried gel.
This assay uses the viral large T antigen protein to initiate eukaryotic DNA replication. The cloning vector pZ189 contains the SV40 virus replication origin. Large T antigen initiates replication by targeting the origin and melting the duplex DNA so that cell extract proteins can load onto the plasmid and replicate the DNA. We cloned the Pu·Py tract used in the primer extension reactions above into the EcoRI site of pZ189. These plasmids were purified by isopycnic banding and used in replication assays as previously described (33,36,37).
To better understand how the Pu·Py tracts in PKD1 and more centromeric homologous genes could facilitate mutagenesis, we thermodynamically characterized the structures formed by the mirror repeat sequences. Furthermore, because replication-blocking lesions can lead to recombinational repair, we examined the effects of the Pu·Py tracts on DNA replication.
Two dimensional gel electrophoresis was used to thermodynamically characterize the triplex structures that were formed by the PKD1 Pu·Py tract sequences (Fig. (Fig.2).2). Each plasmid exhibited a structural transition. Supplementary Tables S1 and S2 list the thermodynamic data from these experiments. The superhelical tension at which the first structural transition occurred ranged from –0.015 to –0.042 (mean –0.030, median –0.031). Using equation 2, this corresponds to the energy of formation ranging from 16.5 to 125.8 kJ (mean 64.5 kJ, median 66.5 kJ). Plasmids containing overlapping Pu·Py mirror repeats required less superhelical tension and energy to form intramolecular triplexes than plasmids containing an individual mirror repeat sequence (–0.021 versus –0.034 and 30.7 versus 81.5 kJ, P < 0.001 using Mann-Whitney test for both parameters). The number of superhelical turns unwound by the first alternative DNA structure formation ranged from 0.5 to 5.5 (mean 3.1, median 3.5), corresponding to the energy of absorption ranging from 0.26 to 31.19 kJ (mean 12.31 kJ, median 12.62 kJ) using equation 3. A number of factors may influence the formation of intramolecular triplexes (38). The correlation coefficients between these factors and superhelical tension (σ) are listed in Table Table1.1. Potential length of the triplex was the only factor with a statistically significant correlation.
To begin to understand the effects of the PKD1 mirror repeat sequences on DNA replication, we used a primer extension assay. In order to compare the DNA structural blockade to an internal control, we cut the plasmid with PvuII (Fig. (Fig.3A).3A). If the DNA polymerase were to completely replicate the template using the forward primer, a full-length band of 323 bp plus the insert size (ranging from 34–88 bp) would be predicted (Fig. (Fig.3B).3B). If the polymerase were to completely replicate the template using the reverse primer, the full-length band would be predicted to be 192 bp plus the insert size (Fig. (Fig.3C).3C). Bands corresponding to the full-length product were observed in all lanes in variable amounts. If the DNA polymerase were to arrest at the center of the Pu·Py insert site using the forward primer, a band of ~110–136 bp would be expected, depending on the insert size. With the reverse primer, the expected arrested bands would be ~120–150 bp. As seen in Figure Figure3D,3D, bands corresponding to polymerase arrest within the Pu·Py tract were observed in either one or the other lane for each insert, but never in both lanes for any specific insert. This is consistent with polymerase being blocked only on the polyguanine template (26,27). Depending on the orientation with which the sequence was cloned into pUC19, the primer extension template could be G-rich using either the forward or reverse primer, explaining why premature polymerase arrest bands were observed in either the ‘A’ or ‘B’ lane. The pausing is not seen on the Py-rich template strand because the pH of the extension reaction buffer is nonpermissive. A summary of the densitometric analyses listing the average amount of replication arrest as a percent of total product of the three primer extension reactions is presented in Figure Figure3E,3E, indicating the amount and range of premature polymerase arrest.
To determine whether intramolecular triplex formation could be used to predict replication blockade, correlation coefficients between specific factors influencing intramolecular triplex formation (38) and observed replication blockade were calculated and the results are listed in Table Table2.2. After removing the outlying value associated with pJB3 (which also had the greatest relative variability in amount of polymerase arrest), the Spearman rank correlation coefficient (rs) between these superhelical tension and percent replication blockade was –0.53 (P < 0.05) (see Supplementary Fig. S2). The correlation between the potential number of base pairs involved in triplex formation of the individual mirror repeats and the percent polymerase arrest was also statistically significant (rs = 0.59, P < 0.05) (see Fig. S3). The percentage of mirror symmetry (i.e. the percentage of nucleotide pairs +n and –n bases around the center that are identical) and the percentage of G·C pairs within the mirror repeat segment involved in triplex formation both approached but did not achieve statistical significance at the 0.05 level. The correlation between replication blockade and number of superhelical turns unwound by triplex formation, center length and amount of A·T base pairs in the center were not significant.
While purified polymerase systems are helpful, they do not adequately reflect the in vivo replication process. To better understand the effects of the PKD1 mirror repeat sequences on mammalian replication, we used replication-competent extracts in primer extension experiments (31). The PKD1 Pu·Py tract demonstrated significant replication fork blockade (Fig. (Fig.4A).4A). The magnitude of the replication blockade was similar to that induced by DNA damage from chemical damage (data not shown). The Pu·Py tract from the PKD1 gene revealed an intense arrest site on the gel precisely halfway through a mirror repeat in the tract. Like the purified polymerase studies, the replication blockade occurred only on the G-rich template.
To detect possible effects of triplex forming sequences on replication, the 88 bp fragment from pJB4 was cloned into the shuttle vector pZ189 to create pZ4.1 (Fig. (Fig.4B).4B). These plasmids were then used in an in vitro replication assay (Fig. (Fig.4C)4C) (33,36,37). As predicted, very little replication took place without the addition of large T antigen in either plasmid. With the addition of this protein, intense bands for both supercoiled and open circular (form II) plasmids were visualized. The addition of the 88 bp fragment from pJB4 resulted in significant changes in the replication reaction products. Linearized molecules (form III) and replication intermediates, felt to be late Cairns structures, were also identified and accounted for ~25% of the resolved replication products. The presence of linearized molecules supports a role for double-strand breakage during the replication process, and the additional replication intermediate band support replication blockade. Given that the only difference in the two plasmids was the presence of the 88 bp Pu·Py tract, it is likely that the replication blockade and double-strand breakage can be attributed to this tract. These findings support the replication blocking ability of this sequence identified in the purified polymerase studies.
The biological effects of Pu·Py tracts remain to be fully elucidated. The studies presented here demonstrate that mirror repeat Pu·Py tracts from PKD1 intron 21 form DNA intramolecular triplex structures under modest energies and inhibit replication. Bacolla et al. found that nonB-DNA structures formed by the PKD1 Pu·Py tract inhibited replication in Escherichia coli (39). We demonstrate that these tracts inhibit replication in eukaryotic replication systems as well. Such interference with DNA replication can lead to mutation (40).
We thermodynamically characterized intramolecular triplex formation by the Pu·Py mirror repeats found at the P1 nuclease-sensitive regions of PKD1 intron 21. Blaszak (24) found that pJB4b formed a triplex at σ = –0.0375. pJB4b is the longest individual Pu·Py mirror repeat from the region of intron 21 examined. We found that shorter Pu·Py mirror repeats from this region also form triplexes at modest superhelical tension. Sequences containing overlapping Pu·Py tracts required significantly less energy to form triplexes than individual Pu·Py tracts. This observation may indicate that the longer repetitive tracts undergo slipped mispairing more frequently, lowering the energy required to melt the sequence and making transition to a triplex structure more likely.
Although the physiologic relevance of the ability of PKD1 Pu·Py sequences to adopt an alternative secondary structure in vitro under acidic conditions is unclear, it does appear to predict replication arrest. The replication fork blockade demonstrated in Figures Figures3D3D and and4A4A most likely represent replication arrest due to a Pu·Pu·Py triplex formation between the nascent and template strands during replication with the polypurine template strand folding back over the polypyrimidine nascent strand. This interstrand effect on replication would not be dependent on superhelical tension given the single-stranded template. While independent of superhelical tension, however, the foldback-triplex formation would be expected to share common features with the intramolecular triplex formation. For example, the energy required for foldback-triplex formation may correlate with the superhelical energy required for intramolecular triplex formation. Additionally, a longer mirror repeat sequence would be more likely to form either kind of triplex and would probably be more difficult for the polymerase to bypass.
We looked for other features of the Pu·Py tracts that correlate with polymerization arrest in an effort to identify sequence characteristics that can be used to predict which Pu·Py tracts in the human genome are more likely to interfere with replication. Besides superhelical tension of intramolecular triplex formation, we found that only the potential length of the triplex had a statistically significant correlation with polymerization arrest. The relationship between the length of the Pu·Py tracts and polymerase pausing has been suggested previously (28,41). As shown in Figure Figure3E,3E, the primer extension studies demonstrated a considerable range in the amount of polymerase blockade. Given the correlation coefficients obtained, this variability is only partially explained by the factors examined in this study, reflecting the role of additional factors.
A limitation of this study is that Taq polymerase was used, which is not as processive as other polymerases. This limitation in processivity may explain the multiple bands observed in the primer extension studies that were not consistent with either a full-length product or polymerase arrest within the Pu·Py region. Because repeat sequences are prone to slipped mispairing mutagenesis, digestion of the plasmids with EcoRI was performed and showed a single length insert size, indicating that the starting templates for the primer extensions were homogenous in size (see Fig. S1).
Another limitation is that replication blockade using purified Taq polymerase does not model eukaryotic replication systems. We therefore examined the effect of the PKD1 Pu·Py tracts on eukaryotic replication systems using both a primer extension assay (Fig. (Fig.4A)4A) and and an SV40 system (Fig. (Fig.4C).4C). Both assays demonstrated significant replication blockade in the presence of a Pu·Py tract from PKD1 (Fig. (Fig.4A4A and C). In addition, the SV40 replication assay shows the presence of double-strand breaks which is only observed in the presence of the Pu·Py tract (Fig. (Fig.4C).4C). Both findings further support the hypothesis that these sequences play a role in the development of human ADPKD.
A disease model is proposed in which the G-rich template of Pu·Py mirror repeat tracts of PKD1 or its homologous genes engage in a foldback-triplex structure with the nascent strand during DNA replication resulting in a replication fork blockade (Fig. (Fig.5).5). In an effort to reactivate the blocked replication fork, recombination or double-strand breaks may be required which can compromise replication fidelity, leading to a second hit mutation in the normal PKD1 allele and LOH. Such a model is consistent with the gene conversion mutagenic mechanism proposed by Watnick et al. (42). Following replication blockade, gene conversion can occur using the homologous tracts of the PKD1 gene at chromosome 16p13.1 (42–44).
This model supports the potential role of the Pu·Py tract in PKD1 or homologous genes as a mechanism leading to somatic mutation and cystogenesis in ADPKD. Variation in the rate at which this second hit occurs may explain the variable expressivity seen with ADPKD. The development of the ability to predict which Pu·Py tracts are more likely to lead to mutagenesis and the development of methods to prevent or slow down the rate of triplex formation during replication may offer an opportunity to alter the disease course in these patients for whom only supportive treatment is currently available.
This work was supported in part by the NIH (DK61458) (J.J.B.), the Polycystic Kidney Disease Foundation (J.J.B.) and the Kidney Foundation of Greater Cincinnati (H.P.P.).